Analytical and Preparative Separation Methods of Biomacromolecules PDF

0.0 10.0 20.
0
0.0 10.0 20.0
Time [min] Timc [min]
edited by
Hassan Y. Aboul-Enein
Analytical and
Preparative
Separation Methods
of Biomacromolecules
Analytical and
Preparative
Separation Methods
of BiomacromoIecuIes
edited by
King Faisal Specialist Hospital and Research Centre
Riyadh, Saudi Arabia
M A R C E L
MARCELDEKKER,
INC.
D E K K E R
Library of Congress Cataloging-in-Publication Data
Analytical and preparative separation methods of biomacromolecules / edited by

Hassan Y. Aboul-Enein.
p. cm.
ISBN 0-8247-1996-4 (alk. paper)
I . Macromolecules-Separation. 2. Biomolecules-Separation. 3
Chromatographic analysis. 4. Electrophoresis. I. Aboul-Enein, Hassan Y.
QP5 19.9.S45A53 1999
5 7 2 . 8 4 ~ 12 99-26367
CIP
This book is printed on acid-free paper.
Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 100 I6
tel: 2 12-696-9000; fax: 2 12-685-4540
Eastern Hemisphere Distribution

Marcel Dekker AG
Hutgasse 4, Postfach 8 12, CH-4001 Basel, Switzerland
tel: 4 1-61-26 1-8482; fax: 4 1-61-26 1-8896
World Wide Web

http://www.dekker.com
The publisher offers discounts on this book when ordered in bulk quantities. For more infor-
mation, write to Special Sales/Professional Marketing at the headquarters address above.
Copyright O 1999 by Marcel Dekker, Inc. All Rights Reserved.
Neither this book nor any part may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopyng, microfilming, and recording, or
by any information storage and retrieval system, without permission in writing from the
publisher.
Current printing (last digit):

I 0 9 8 7 6 5 4 3 2 1
PRlNTED IN THE UNITED STATES OF AMERICA

To my beloved Partent and the most special people in my life,
Nagla, Youssef, Faisal, and Basil
Preface
It is my pleasure to be an editor of this book, which gave me a chance to be

exposed to this fascinating field of separation technology of biomacromolecules
which represent, medically and biochemically, an important class of compounds.
This book consists of sixteen chapters representing the latest research re-
sults and recent developments in the area of high-performance liquid chromatog-
raphy and capillary electrophoresis. Also, the book includes chapters discussing
in detail: biochromatography, displacement chromatography, affinity capillary
electrophoresis, and the new technologies of slalom chromatography. The scope
of the topics presented are broad and therefore give a comprehensive view of
the current status of the established and newly developed separation techniques
that are useful for biomacromolecule isolation and purification.
I would like to thank the contributors, internationally renowned experts in
their respective research fields, who have made this book possible. Also, my
thanks are extended to Marcel Dekker, Inc., and, in particular, Mr. Russell Dekker
for his encouragement to publish this book.
As we approach the next millennium and reflect on the rapid developments
that have been achieved in the area of biotechnology, I certainly hope that this
book will be a useful reference for scientists and researchers dealing with isola-
tion, purification, and analysis of biomolecules.
Contents
Preface iii
Contributors xi
Analysis of HbAIc and Some Hb Variants by HPLC 1

Ursula Turpeinen and Ulf-Hikan Stenman
Analysis of Posttranslational Modifications in Recombinant Proteins

by HPLC and Mass Spectrometry 13
Ruiner Bischoff and Bernadette Bouchon
Analytical and Preparative Separations of Peptides by Capillary and

Free-Flow Zone Electrophoresis 39
Vaclav Kaiicka
Bioaffinity Chromatography 99
Jaroslava Turkovd
Characterization and Partial Purification of Steroidogenic Factors

from Thymic Epithelial Cell-Conditioned Medium 167
Mehmet Uzumcu, David R. Brigstock, and Young C. Lin
viii Contents
Determination of Affinity Constants of Lectins for Sugars by Affinity

Capillary Electrophoresis 187
Kiyohito Shimura and Ken-ichi Kasai
Displacement Chromatography of Biomolecules 203

Ruth Freitag
High-Performance Membrane Chromatography of Proteins 255

Tatiana Tennikova and Ruth Freitag
HPLC Purification of Recombinant Proteins 301

Carr J. Smith, Patricia Martin, Sandra M. Scott, and Thomas H. Fischer
Isolation, Purification, and Characterizaton of Human Seminal

Plasma Proteins and Their Immunological Behavior In Vitro . 331
Afrozul Hag, Nona Remo Rama, and Sultan T. Al-Sedairy
Isolation, Purification, and Structural Study of Allergenic

Proteins 353
Jean-Pierre Dandeu
Multiangle Light Scattering Combined with HPLC with Examples

for Biopolymers 369
Philip J. Wyatt
Purification and Characterization of Connective Tissue Growth

Factor Using Heparin Affinity Chromatography 397
David R. Brigstock
Slalom Chromatography: A New Hydrodynamic-Based

Chromatographic Mode Applicable to Size-Dependent Separation
and Physicochemical Analysis of Large DNA Molecules 415
Jun Hirabayashi and Ken-ichi Kasai
Simultaneous Chromatographic Separation of Ceruloplasmin and

Serum Amine Oxidase 431
Mircea-Alexandru Mateescu, Xin-Tao Wang, Olivia Befani, Marie-Josee'
Dumoulin, and Bruno Mondovi
Contents
16. Fast, Single-Step Affinity Chromatography Purification of

Hemoglobin 445
Mircea-Alexandru Mateescu and Wiljrid Jacques
Index 457
Contributors
Sultan T. Al-Sedairy Research Center Administration, King Faisal Specialist

Hospital and Research Center, Riyadh, Saudi Arabia
Olivia Befani Department of Biochemical Sciences and CNR Center of Molec-

ular Biology, Rome University "La Sapienza," Rome, Italy
Rainer Bischoff, Ph.D.* Protein Analytical Group, Transgene S.A., Stras-

bourg, France
Bernadette Bouchon Protein Analytical Group, Transgene S.A., Strasbourg,

France
David R. Brigstock, Ph.D. Associate Professor, Surgery and Medical Bio-

chemistry, Ohio State University and Children's Hospital, Columbus, Ohio
Jean-Pierre Dandeu Unit6 d'Immuno-Allergie, Institut Pasteur, Paris, France
Marie-Josee Dumoulin Department of Chemistry and Biochemistry, Uni-

versit6 du Qukbec i Montreal, Quebec, Canada
* Current affiliarion: Senior Research Scientist, Biochemistry and Bioanalytical Chemistry Depart-
ment. Astra-Draco AB, l a n d , Sweden
xi
xii Contributors
Thomas H. Fischer, Ph.D. Research Assistant Professor, The School of Medi-

cine, Center for Thrombosis and Hemostasis, University of North Carolina at
Chapel Hill, Chapel Hill, North Carolina
Ruth Freitag, Ph.D. Professor, Chemistry Department, ETH Lausanne, Lau-

sanne. Switzerland
Afrozul Haq, Ph.D. Associate Scientist, Biological and Medical Research,

King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia
Jun Hirabayashi Department of Biological Chemistry, Faculty of Pharmaceu-

tical Sciences, Teikyo University, Sagamiko, Kanagawa, Japan
Wilfrid Jacques Professor, Department of Chemistry and Biochemistry, Uni-

versitC du QuCbec i Montrbal, QuCbec, Canada
Ken-ichi Kasai Faculty of Pharmaceutical Sciences, Teikyo University, Sa-

gamiko, Kanagawa, Japan
Viclav KaiiEka, Ph.D. Senior Research Scientist, Department of Biochemistry

of Peptides, Institute of Organic Chemistry and Biochemistry, Academy of Sci-
ences of the Czech Republic, Prague, Czech Republic
Young C. Lin, Ph.D., D.V.M. Laboratory of Reproductive and Molecular En-

docrinology, Department of Veterinary Biosciences, College of Veterinary Medi-
cine, Ohio State University, Columbus, Ohio
Patricia Martin, Ph.D., D.A.B.T. Senior Research and Development Chemist,

Analytical Chemistry Division, R. J. Reynolds Tobacco Company, Winston-
Salem. North Carolina
Mircea-Alexandru Mateescu, Ph.D. Professor, Department of Chemistry and

Biochemistry, UniversitC du QuCbec i MontrCal, Quibec, Canada
Bruno Mondovi, M.D. Professor, Department of Biochemical Sciences and

CNR Center of Molecular Biology, Rome University "La Sapienza," Rome,
Italy
Nona Remo Rama, B.Sc. Research Technician, Biological and Medical Re-
search, King Faisal Specialist Hospital and Research Center, Riyadh, Saudi
Arabia
Contributors xiii
Sandra M. Scott, B.S. Research Assistant, Research and Development, Envi-

ronmental and Molecular Toxicology, R. J. Reynolds Tobacco Company, Win-
ston-Salem. North Carolina
Kiyohito Shimura, Ph.D. Associate Professor, Department of Biological

Chemistry, Faculty of Pharmaceutical Sciences, Teikyo University, Sagamiko,
Kanagawa, Japan
Carr J. Smith, Ph.D., D.A.B.T. Master Scientist, Research and Development,

Environmental and Molecular Toxicology, R. J. Reynolds Tobacco Company,
Winston-Salem, North Carolina
Ulf-Hikan Stenman, M.D., Ph.D. Laboratory, Helsinki University Central

Hospital, Helsinki, Finland
Tatiana Tennikova Professor, Institute of Macromolecular Compounds, Rus-

sian Academy of Sciences, St. Petersburg, Russia
Jaroslava Turkovi, Ph.D., Dr.%. Senior Research Scientist, Institute of Or-

ganic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic,
Prague, and University of Pardubice, Pardubice, Czech Republic
Ursula Turpeinen, Ph.D. Chemist, Laboratory, Helsinki University Central

Hospital, Helsinki, Finland
Mehmet Uzumcu, Ph.D., D.V.M. Assistant Scientist, Biological and Medical

Research, King Faisal Specialist Hospital and Research Center, Riyadh, Saudi
Arabia
Xin-Tao Wang, Ph.D. Department of Chemistry and Biochemistry, Universitk

du Quebec h Montreal, Qukbec, Canada, and Rome University "La Sapienza,"
Rome, Italy
Philip J. Wyatt, Ph.D. President, Wyatt Technology Corporation, Santa Bar-

bara, California
Analysis of HbA,c and Some Hb
Variants by HPLC
Ursula Turpeinen and Ulf-Hakan Stenman
Helsinki University Central Hospital, Helsinki, Finland
I. INTRODUCTION
This chapter deals with selected aspects of high-performance liquid chromatogra-

phy (HPLC) methods for quantitating of glycohemoglobin (HbAlc) and charac-
terizing some Hb variants. Determination of HbAlc by ion exchange HPLC is
used for monitoring of glycemic control in diabetic patients. The percentage of
HbAlc reflects blood glucose concentrations of the previous 2-3 months. It is
therefore considered a valuable indicator of long-term diabetic control [I]. How-
ever, the methods currently used for its measurement in clinical chemistry labora-
tories show large differences between reported values, and comparison of results
from different laboratories is difficult [2]. Lack of standardization of glycohemo-
globin measurements remains the major source of interlaboratory variation. At
present there is no acknowledged reference method or an accepted standard. This
fact is well recognized and an IFCC (International Federation of Clinical Chemis-
try) working group is developing a reference method based on HbA,, as the
biochemically well-defined major glycohemoglobin component. Recently, the
use of calibration based on a cation exchange HPLC method has been shown to
increase the comparability between various analytical methods [3,4].
The classical method for the rapid determination of HbA ,c by chromatogra-
phy on the weak cation exchange resin Bio-Rex 70 was introduced by Trivelli
et al. in 1971 [5] based on the method of Allen et al. on Amberlite IRC-50 [6].
Several approaches, including minicolumns [7] and HPLC [8], were used to make
ion exchange chromatography acceptable for use in clinical laboratories. Though
acceptable resolution can be obtained with Bio-Rex 70, it has lower resolution
2 Turpeinen and Stenman
and is more time consuming than HPLC methods. Flow rate is limited by the
compressibility of the resin and its nonuniform particle size. These drawbacks
stimulated development of several methods using ion exchange resins for HPLC,
such as Synchropak CM 300 [9], silica-based carboxymethylpolyamide [lo],
Mono S [8,1 I], and the Glycopak resin of the Diamat analyzer (Bio-Rad). These
methods can usually also be used to measure fetal hemoglobin (HbF). They differ
in their sensitivity to interferences such as adducts formed between Hb and
urea, aldehydes, or acetylsalicylic acid and other hemoglobin (Hb) variants
[12,13].
Glycohemoglobin can also be assayed by affinity chromatography using
boronate agarose columns [14]. This method measures HbAlc together with gly-
cohemoglobins, which elute in the HbA, fraction in ion exchange chromatog-
raphy.
In chromatographic methods, abnormal Hb variants may cause erroneous
HbA,, results [15,16], but at the same time they permit identification of individu-
als with such variants. This information is of importance for the interpretation
of the HbA,, results. Therefore it is an advantage if an HPLC method used for
clinical purposes separates not only HbA,, but also other hemoglobin compo-
nents to permit identification of abnormal hemoglobins. The good resolution
of more than 35 frequently encountered human hemoglobins and HbAlc on
PolyCAT A [I71 makes it a useful method for preliminary characterization of
Hb variants. The Diamat method has been shown to separate seven variants [18],
but many overlap partially with the HbA,, peak.
II. CURRENT HPLC METHODS FOR THE DETERMINATION

OF HbAIc BY CATION EXCHANGE CHROMATOGRAPHY
A. Mono S
The Mono S system was introduced by Stenman et al. and Jeppsson et al. [8,11].
The Mono S method has been widely used in clinical laboratories. It is precise,
accurate, and easily operated [ l l ] . The monodisperse cation resin, Mono S, does
not shrink, swell, or leak functional groups with a LiCl gradient in malonate
buffers at pH 5.7. The acetaldehyde adduct of Hb in alcoholics may interfere
with HbAlc determinations since it coelutes with HbAlc. In uremic patients, car-
bamylated Hb may increase the HbA,, values by as much as 1%. The separation
has been further optimized by using a smaller column load, decreased flow rate,
and a steeper LiCl gradient [19]. We have also used Mono S extensively for
HbAlc analysis by slightly modifying the original method [8]. Phosphate buffers
with 10% acetonitrile and a higher pH of 6.85 were used at 30°C. These changes
Analysis of HbA,, by HPLC 3
improved the separation and the column life increased up to several thousand
samples with constant resolution.
B. PolyCAT A
PolyCAT A is a weak cation exchanger with polyaspartic acid linked to silica
[20,2]]. This resin has been used for the determination of HbAlC [22,23], of
several minor molecular forms of hemoglobin [24], and for identification of he-
moglobin variants 1171. The separation of HbALC from other Hb components
such as acetylated Hb [13] and HbF depends critically on buffer composition,
particularly on the pH and salt gradient. It is often necessary to make minor
variations in the steepness of the salt gradient in order to get a good separation.
The assay is precise [22] and it depends more on the quality of the chromato-
graphic separation, the level of noise in the spectrophotometer, and the mode of
integration than on volumes injected.
PolyCAT A with bis-tris buffers and sodium acetate gradients has been
used to separate and quantitate many normal and abnormal Hb components
[23,24]. Minor fetal Hb forms can be separated and HbA,c can simultaneously
be quantitated. We have compared the performance of PolyCAT A chromatogra-
phy with a boronate affinity binding assay and the automated Diamat system for
the determination of HbA,c [23]. Elution was achieved with a linear gradient
consisting of buffers A and B at a flow rate of 1.1 mllmin at room temperature.
Buffer A (pH 6.87) contained 10 mmol of bis-tris and 1.0 mmol of KCN per
liter. Buffer B (pH 6.57) contained buffer A plus 200 mmol of NaCl per liter.
The gradient was: time 0-2 min, 21% B; time 16 min, 47% B; time 22 min, 100%
B; time 24 min, 100% B; and time 26 min, 21% B. Integration was performed by
the valley-to-valley method.
When we compared the two cation exchange methods, PolyCAT A and
Diamat, we obtained an acceptable correlation (Fig. 1). The correlation coeffi-
cient was 0.90 but the regression equation (PolyCAT A = 1.06 X Diamat - 3.06)
showed that much lower results were obtained by PolyCAT A. This suggests that
the HbAlc value measured by the Diamat assay includes a background of
2-3% of the total Hb. This might be due to the fact that the Diamat method also
measures carbamylated and acetylated forms of Hb 1121 and possibly
some other derivatives formed in blood during storage, which can be separated
from the HbAlc peak by the use of methods with higher resolution. The Poly-
CAT A assay has been optimized to separate different Hb variants from HbAlc.
However, the difference between the Diamat method and the PolyCAT A assay
cannot be explained only by carbamylated and acetylated Hbs, for which levels
below 0.4% have been reported [12]. The slope of the regression line, 1.06, shows
that the PolyCAT A method reflects differences in glycated Hb in the same way
Turpeinen and Stenman
0 2 4 6 8 1 0 1 2 1 4
Diamat (%)
Figure 1 Correlation between the HbA,, values obtained by PolyCAT A (y) and Diamat
(x).Corresponding regression equation calculated by the standardized principal compo-
nent is y = 1 . 0 6 ~-3.06, r = 0.90.
as the Diamat method. The negative bias of about 2-3% of total Hb is present
at all levels of HbAlc when PolyCAT A is compared to Diamat.
C. Diamat and Variant Analyzers

The Diamat analyzer of Bio-Rad Laboratories is an automated HPLC instrument
using a weak cation exchange column. The support material is silica with car-
boxymethyl functional groups. The analyzer forms a stepwise gradient of three
phosphate buffers of increasing ionic strength to separate the HbAlc from other
Hbs. The system has been optimized for the quantitation of HbAlc, and HbA2
coelutes with HbA,,. Quantitation is based on the absorbance at 415 and 690 nm.
Before analysis with the Diamat analyzer 5 yl of the whole-blood sample is
diluted with 1.25 ml of hemolysis reagent containing 0.1 % (v/v) polyoxyethylene
ether in a borate buffer. The tubes are incubated at 37°C for 30 min to remove
the labile HbAlc fraction.
The Variant analyzer of Bio-Rad Laboratories offers a broader test selection
of methods than the Diamat system. Although the separation principle is the
same, with various programs it is possible to separate and quantitate, in addition
to HbAlc, also HbF and HbA,. The method has been calibrated against the Diamat
method. In practice, however, there seems to be a slight difference between these
systems, with the Variant giving slightly higher results than the Diamat [25].
Ill. DETERMINATION OF HbAlc BY AFFINITY

CHROMATOGRAPHY
The boronate affinity chromatography method involving minicolumns has gained

wide acceptance because it is not affected by Hb variants, slight temperature
changes, carbamylated Hb, and the labile glycated fraction [26]. The application
of HPLC to affinity methods has the same advantages as the minicolumns. An
automatic affinity HPLC system is now available and has been evaluated for the
monitoring of glycohemoglobin [27].
A considerable bias has been observed between cation exchange and affin-
ity chromatography. In two studies the correlations between the methods were:
affinity chromatography = 1.40 X ion exchange - 2.19 [28] and affinity chroma-
tography = 1.45 X ion exchange + 0.04 [24]. The difference in slope is explained
by the fact that affinity methods measure glycated Hbs other than HbA,c. The
negative y intercept may be explained by the fact that nonglycated components
of HbA lc are measured by the ion exchange chromatography [23]. Methods based
on boronate affinity detection of the glycated amino terminus of the P chains of
Hb together with glycosylated Hbs eluting in the HbA, fraction in ion exchange
chromatography may be expected to reflect blood glucose control more accurately
than ion exchange chromatography methods, which also measure coeluting non-
glycated Hb. However, this has to be determined in a true clinical setting.
IV. REFERENCE VALUES FOR HbAlc
Determination of a method-specific reference range is important in glycohemo-

globin analysis, due to differences between methods. We have estimated the refer-
ence values of HbA,c which are lower with the PolyCAT A method than with
the Diamat method, apparently because of better separation of HbAIC from non-
glycated coeluting forms of Hb with PolyCAT A. The reference range for HbAlc
is 4.5-5.8% for the Diamat and 2.5-4.4% for the PolyCAT A method [23]. The
difference in mean values between the PolyCAT A (3.4%) and Diamat (5.1%)
methods was 1.7%.
The reference range with PolyCAT A is lower than that observed with our
original HPLC method, 4.2-6.6%, using the Mono S column [8] and that reported
by Jeppsson et al., 3.9-5.3%, using the same column but a different gradient
[I I]. With PolyCAT A and other gradients, reference ranges of 2.8-5.5% [24]
and 3.9-5.7% [22] have been reported. Slightly higher reference ranges of 5.0-
7.6% [28] and 4.8-6.4% [29] have been reported for boronate affinity chromatog-
raphy methods. Of the glycated Hb adsorbed to boronate affinity columns, about
50% elutes in ion exchange chromatography as HbA,,, about 40% in the HbAo
peak and 10% as HbA,,,, [28]. The values for the PolyCAT A method can be
estimated to correspond to about 60-70% of those obtained by boronate affinity
chromatography. This suggests that the PolyCAT A method measures the gly-
cated component of the HbAlCpeak fairly specifically. A certain change in blood
glucose causes a larger change in glycated Hb measured by affinity chromatogra-
phy than in HbAlc. This is demonstrated by the slope of the correlation, about
1.4-1.5 [24,28] in spite of similar reference values.
V. HPLC CATION EXCHANGE CHROMATOGRAPHY

OF Hb VARIANTS
HPLC provides a rapid and sensitive means for the examination of,abnormal
hemoglobins. Methods with high sensitivity and specificity have been developed
for screening and confirmation of hemoglobinopathies in newborns [30,31]. Most
methods have used Synchropak CM 300, a silica support with carboxylic acid
residues or PolyCAT A. The screening procedures are designed to rapidly detect
the major variants. However, the resolution of these methods does not permit
differentiation of some commonly encountered hemoglobin variants such as HbE
17.9 H b b ,
-
r
20.4HbA2 ,
-21.1 HbD
19.7 HbE
7- 22.3 HbS
26.7-
25.1 HbC
Figure 2 Cation exchange chromatography of HbF, HbA,, HbE, HbA,, HbD, HbS, and
HbC on PolyCAT A. Time of elution since sample injection is indicated at each peak.
Detection at 415 nm.
Analysis of HbA,, by HPLC
10 - -
5 9.2 Hb Helsinki
Figure 3 Cation exchange chromatography on PolyCAT A of a red cell hemolyzate

with a P-chain variant of (a) Hb-Helsinki and (b) Hb-Vaasa. Time of elution since sample
injection is indicated at each peak. Detection at 415 nm.
from HbA,. The sensitivity to detect hemoglobin variants occurring at low con-
centrations is also poor.
Recently, Ou and Rognemd [17] described a method for Hb variant analysis
on the PolyCAT A column. The good resolution obtained with more than 35
frequently encountered human Hbs also makes it useful for HbAlcdeterminations
and for preliminary identification of Hb variants. Earlier, PolyCAT A was used
for screening of cord blood for hemoglobinopathies [32]. The column has the
resolution necessary for both screening and confirmatory purposes. We have in-
vestigated the use of a 4.6 X 200 mm PolyCAT A column with 5-pm particles
for separation of Hb variants. A typical chromatographic separation of an Hb
mixture containing Ale, F, A,,, E, A,, D, S, and C is shown in Fig. 2. We have
also separated other Hb variants by using bis-tris buffers and an NaCl gradient
(Fig. 3). The baseline resolution permits accurate quantitation of these hemoglo-
bins, which is important for the correct diagnosis of hemoglobinopathies. In addi-
tion, HbA, can be accurately measured. A slight adjustment of the gradient pro-
gram may be necessary for optimal separation of all variants.
Recently, a new automatic HPLC system using a Bio-Rad MA7 polymeric
cation exchange material has been used to screen newborns for sickle cell anemia
and other hemoglobinopathies [33]. The method is reported to resolve hemoglo-

bins F, A, S, C, E, and D.
The Diamat system is also suitable for hemoglobinopathy screening. Al-
though it was originally designed mainly for determinations of HbAlc, it can be
used for detection of some Hb variants. It has been shown to separate seven
variants, but many overlap partially with the HbA,c peak [18]. Detection of aber-
rant peaks with the Diamat method, which interfere with the determination of
HbA ,c, led us to the detection of two Hb variants, Hb Broussais and Hb Ceme-
nelum [34]. Hb Broussais caused underestimation and Hb Cemenelum overesti-
10 HbAi c
- 15.1 Hb Broussais
17.7 HbAO
20.2 HbA2
15.7 Hb Cemenelum
'
d
1 18.6 HbAO
21.5 HbA2
Figure 4 Cation exchange chromatography on PolyCAT A of a red cell hemolyzate

with an a-chain variant of (a) Hb-Rroussais and (b) Hb-Cemenelum. Time of elution since
sample injection is indicated at each peak. Detection at 415 nm.
mation of HbAlc concentration in the Diamat method. With PolyCAT A, both

variants were totally separated (Fig. 4). The elution pattern can be used for pre-
liminary identification of these Hb a variants, especially if standards are available.
VI. CONCLUSIONS
HPLC is a widely used technique for the determination of HbAlC although re-
cently developed immunoassays have started competing with HPLC. Since there
is no general agreement on standardization of the methods for determination of
glycated Hbs, HPLC ion exchange methods measuring HbAlc are used to cali-
brate both boronate affinity methods and some new immunochemical methods
[35].Because of its good precision, ion exchange HPLC has been recommended
as the gold standard for measurement of glycohemoglobin. However, it should
be recognized that the presently used reference method is far from ideal because
it measures some unglycated other Hb variants but not all of the glycated ones.
The ongoing standardization of the HbAlcassay methods initiated by the Interna-
tional Federation of Clinical Chemistry aims at developing a new reference
method.
REFERENCES
Goldstein DE, Parker KM, England JD, England JE Jr, Wiedmeyer H-M, Rawlings
SS, et al. Clinical application of glycosylated hemoglobin measurements. Diabetes
1982;3 1 (suppl. 3):70-78.
John WG. Glycated haemoglobin analyses-assessment of within- and between-
laboratory performance in a large UK region. Ann Clin Biochem 1987;24:453-460.
Bodor GS, Little RR, Garrett N, Brown W, Goldstein DE, Nahm MH. Standardiza-
tion of glycohemoglobin determinations in the clinical laboratory: three years of
experience. Clin Chem 1992;38:2414-2418.
Weykamp CW, Penders TJ, Muskiet FAJ, van der Slik W. Effect of calibration
on dispersion of glycohemoglobin values determined by 11 1 laboratories using 21
methods. Clin Chem 1994;40: 138- 144.
Trivelli LA, Ranney HM, Lai H-T. Hemoglobin components in patients with diabe-
tes mellitus. N Engl J Med 1971;284:353-357.
Allen DW, Schroeder WA, Balog J. Observations on the chromatographic heteroge-
neity of normal adult and fetal human hemoglobin: a study of the effects of crystalli-
zation and chromatography on the heterogeneity and isoleucine content. J Am Chem
SOC1958;80:1628-1634.
B i s i E, Abraham EC. New less temperature-sensitive microchromatographic
method for the separation and quantitation of glycosylated hemoglobins using a non-
cyanide buffer system. J Chromatogr 1985;344:8 1-91.
Stenman U-H, Pesonen K, Ylinen K, Huhtala M-L, Teramo K. Rapid chromato-

graphic quantitation of glycosylated haemoglobins. J Chromatogr 1984;297:327-
332.
Toren EC, Vacik DN, Mockridge PB. Cation-exchange, high-performance liquid
chromatographic determinationof hemoglobin A ,. JChromatogr 1983;266:207-212.
Gupta SP, Hanash SM. Separation of hemoglobin types by cation-exchange high-
performance liquid chromatography. Anal Biochem 1983;134: 117-121.
Jeppsson J-0, Jerntorp P, Sundkvist G, Englund H, Nylund V. Measurement of he-
moglobin A,, by a new liquid-chromatographic assay: methodology, clinical utility,
and relation to glucose tolerance evaluated. Clin Chem 1986;32: 1867- 1872.
Weykamp CW, Penders TJ, Siebelder CWM, Muskiet FAJ, van der Slik W. Interfer-
ence of carbamylated and acetylated hemoglobins in assays of glycohemoglobin by
HPLC, electrophoresis, affinity chromatography, and enzyme immunoassay. Clin
Chem 1993;39:138-142.
Turpeinen U, Stenman U-H, Roine R. Liquid-chromatographic determination of
acetylated hemoglobin. Clin Chem 1989;35:33-36.
Abraham EC, Peny RE, Stallings M. Application of affinity chromatography for
separation and quantitation of glycosylated hemoglobins. J Lab Clin Med 1983;102:
187-197.
Allen KR, Hamilton AD, Bodansky HJ, Poon P. Prevalence of haemoglobin variants
in a diabetic population and their effect on glycated haemoglobin measurement. Ann
Clin Biochem 1992;29:426-429.
Weykamp CW, Penders TJ, Musket FAJ, van der Slik W. Influence of hemoglobin
variants and derivatives on glycohemoglobin determinations, as investigated by 102
laboratories using 16 methods. Clin Chem 1993;39:1717-1723.
Ou C-N, Rognerud CL. Rapid analysis of hemoglobin variants by cation-exchange
HPLC. Clin Chem 1993;39:820-824.
Delahunty T. Convenient screening for hemoglobin variants by using the Diamat
HPLC system. Clin Chem 1990;36:903-905.
Philcox JC, Haywood MR, Rofe AM. Hemoglobin A,, by HPLC with the Pharmacia
Mono S HR 515 cation-exchange column: influence of sample protein load on opti-
mal chromatographic conditions. Clin Chem 1992;38: 1488- 1490.
Alpert AJ. Cation-exchange high-performance liquid chromatography of proteins on
poly(aspartic acid)-silica. J Chromatogr 1983;266:23-37.
Ou C-N, Buffone GJ, Reimer GL, Alpert AJ. High-performance liquid chromatogra-
phy of human hemoglobins on a new cation exchanger. J Chromatogr 1983;266:
197-205.
Ellis G, Diamandis EP, Giesbrecht EE, Daneman D, Allen LC. An automated "high-
pressure" liquid-chromatographic assay for hemoglobin A,,. Clin Chem 1984;30:
1746- 1752.
Turpeinen U, Karjalainen U, Stenman U-H. Three assays for glycohemoglobin com-
pared. Clin Chem 1995;41: 190- 1%.
Bisse E, Wieland H. High-performance liquid chromatographic separation of human
haemoglobins. Simultaneous quantitation of foetal and glycated haemoglobins. J
Chromatogr 1988;434:95- 110.
Weykamp CW, Penders TJ, Miedema K, Muskiet FAJ, van der Slik W. Standardiza-
Analysis of HbA,c by HPLC 11
tion of glycohemoglobin results and reference values in whole blood studied in 103
laboratories using 20 methods. Clin Chem 1995;41:82-86.
Fliickiger R and Mortensen HB. Glycated Haemoglobins. J Chromatogr 1988;429:
279-292.
Cefalu WT, Wang ZQ, Bell-Farrow A, Kiger FD, Izlar C. Glycohemoglobin mea-
sured by automated affinity HPLC correlates with both short-term and long-term
antecedent glycemia. Clin Chem 1994;40: 1317- 132 1.
Abraham EC, Perry RE, Stallings M. Application of affinity chromatography for
separation and quantitation of glycosylated hemoglobins. J Lab Clin Med 1983;102:
187-197.
Herold DA, Boyd JC, Bruns DE, Emerson JC, Bums KG, Bray RE, et al. Measure-
ment of glycosylated hemoglobins using boronate affinity chromatography. Ann Clin
Lab Sci 1983;13:482-488.
Wilson JB, Headlee ME, Huisman THJ. A new high-performance liquid chromato-
graphic procedure for the separation and quantitation of various hemoglobin variants
in adults and newborn babies. J Lab Clin Med 1983;102:174-186.
Rogers BB, Wessels RA, Ou C-N, Buffone GJ. High-performance liquid chromatog-
raphy in the diagnosis of hemoglobinopathies and thalassemias. Am J Clin Path01
l985;84:67 1-674.
van der Dijs FPL, van den Berg GA, Schemer JG, Muskiet FD, Landman H, Mus-
kiet FAJ. Screening cord blood for hemoglobinopathies and thalassemia by HPLC.
Clin Chem 1992;38: 1864-1 869.
Eastman JW, Wong R, Liao CL, Morales DR. Automated HPLC screening of new-
borns for sickle cell anemia and other hemoglobinopathies. Clin Chem 1996;42:
704-710.
Turpeinen U, Sipila I, Anttila P, Karjalainen U, Kuronen B, Kalkkinen N, et al. Two
a-chain hemoglobin variants, Hb Broussais and Hb Cemenelum, characterized by
cation-exchange HPLC, isoelectric focusing, and peptide sequencing. Clin Chem
1995;41:532-536.
Karl J, Bums G, Engel WD, Finke A, Kratzer M, Rollinger W, et al. Development
and standardization of a new immunoturbidimetric HbA,c assay. Klin Lab 1993;39:
991 -996.
Analysis of Posttranslational
Modifications in Recombinant
Proteins by HPLC and Mass
Spectrometry
Rainer Bischoff* and Bernadette Bouchon
Transgene S.A., Strasbourg, France
I. INTRODUCTION
A number of recombinant proteins are being produced as pharmaceuticals with

annual sales of U.S. $9 billion in 1994 [1,2]. Therapeutic applications of these
biopharmaceuticals include stimulation of cell growth, regulation of the immune
system and fibrinolysis, to name a few. These advances of recombinant proteins
toward the marketplace was made possible by the development of efficient and
reproducible production and purification systems that permit manufacture of these
complex molecules in large amounts with consistent quality suitable for human
use. Control of the quality of biopharmaceuticals is crucial not only during large-
scale manufacturing but also during the research and development phase, as im-
purities and contaminants have to be characterized and reduced to acceptable
levels.
A complete analysis of a recombinant protein necessitates the combined
use of separation techniques such as electrophoresis and high-performance liquid
chromatography (HPLC) with methods which allow the characterization of the
separated molecular species such as protein sequencing and, in particular, mass
spectrometry (MS). It is probably fair to say that no instrumental analytical
method has had a greater impact on protein analysis in the last 5-10 years than
MS [3]. The advent of ionization techniques that allow us to generate gas phase
ions of high molecular weight, nonvolatile biomolecules such as electrospray
* Current aflliation:
Senior Research Scientist, Biochemistry and Bioanalytical Chemistry Depart-
ment, Astra-Draco AB, Lund, Sweden.
13
14 Bischoff and Bouchon
ionization (ESI) [4], and matrix-assisted laser desorption ionization (MALDI) [S]
has opened possibilities for obtaining precise molecular mass measurements on
proteins, peptides, oligonucleotides, and oligosaccharides. This development has
enabled especially the growing biotechnology industry to integrate MS into strat-
egies to characterize recombinant proteins in detail and to assure consistent qual-
ity [6-81:
The' combination of efficient separation techniques such as HPLC and capil-
lary electrophoresis (CE) with MS is rapidly becoming an indispensable analyti-
cal tool in many academic and industrial laboratories [9-141. The power of ana-
lyzing separated components by MS results from the fact that the molecular mass
is an inherent property of a given compound. In the case of recombinant proteins,
where the amino acid sequence is generally known, it is possible to compare the
calculated mass with the measured mass and thus to confirm the expected struc-
ture or to detect modifications. More detailed information can be obtained when
analyzing defined proteolytic digests of a protein by either off-line (fraction col-
lection) or on-line HPLC-mass spectrometry (LC-MS). Such analyses permit lo-
calization of the modified site(s) and collection of precise data about the molecu-
lar mass of the modification. Thus it is not surprising that LC-MS studies have
focused in many cases on the characterization of posttranslational modifications
such as glycosylations and phosphorylations as well as on the identification of
modifications that were due to culture, purification, or storage conditions [IS-
221.
The following studies of two recombinant proteins that are being developed
as vaccine candidates against schistosomiasis or Lyme disease illustrate the char-
acterization of posttranslational modifications by combining MS with chromato-
graphic separation techniques. rSmp28, an antigen that was originally isolated
from the parasite Schistosoma mansoni [23], was expressed to high levels in
the cytoplasm of Saccharomyces cerevisiae, whereas rOspA, an outer membrane
protein of Borrelia burgdolferi, the Lyme disease-causing agent, was produced
in Escherichia coli [24]. Both proteins were shown to be heterogeneous due to
posttranslational modifications as determined by isoelectric focusing in the case
of rSmp28 [25] or by MS for rOspA [26,27]. In this chapter we describe the
analytical strategy that led to the characterization of these modifications, stressing
the importance of integrating efficient separation methods such as HPLC with
MS.
II. RECOMBINANT SCHISTOSOMA MANSONI

GLUTATHIONE-STRANSFERASE rSmp28
Schistosomiasis is a chronic debilitating parasitic disease affecting approximately

200 million people worldwide and causing 500,000 deaths per year. Although
Recombinant Proteins and MS 15
effective chemotherapy exists, reinfection remains a serious problem. It is thus

of major interest to develop novel approaches to vaccines based on recombinant
proteins such as the recently identified glutathione-S-transferase Smp28 [23].
The study on rSmp28, a 210-amino-acid protein expressed intracellularly
in Succharomyces cerevisiae, illustrates that highly efficient expression systems
in cultures grown to elevated biomass in a bioreactor may lead to unexpected
posttranslational modifications. Characterization of such modifications necessi-
tates their detection in samples taken during the cultivation process, isolation of
the modified protein, and structural analysis.
A. Modification of rSmp28 During High-Biomass Culturing

of S. cerevisiae and Isolation of the Major Modified
Form rSmp2812 by Anion Exchange HPLC
rSmp28 was produced as a cytoplasmic protein in S. cerevisine. Overexpression
of rSmp28 to more than 6 g / L of culture was accompanied by the appearance
of a second form (rSmp28i2) with a measured isoelectric point of 7.9, indicating
an increase in net negative surface charge (Fig. I). Analysis of cellular extracts
throughout the process by isoelectric focusing showed that the amount of
rSmp2812 increased with time (Fig. 1: lane 1, preculture; lane 2, after 19.5 h;
lane 3, after 5 1 h). During culture biomass increased from 8.8 g/L cell dry weight
at 19.5 h to 78.7 g / L after 51 h and the ethanol concentration in the culture
supernatant went through a maximum of 12 g / L at 45 h, indicating some anaero-
bic glucose metabolism.
Analyses of the protein bands by N-terminal amino acid sequencing using
Edman degradation after electroblotting to a polyvinylidene difluoride (PVDF)
Figure 1 Isoelectric focusing of rSmp28 expressed in S. ctrevisine at different stages

of the culture. Lane I: preculture used to inoculate the bioreactor. Lane 2: after 19.5 h
of culture (biomass: 8.8 g/L cell dry weight). Lane 3: after 51 h of culture (biomass: 78.7
g/L cell dry weight). The pH gradient in the gel, as indicated on the right margin was
determined with marker proteins of known isoelectric points. (From Ref. 39.)
membrane confirmed that both of them contained the expected N-terminal se-
quence excluding N-terminal blocking by, for example, formylation or acetyla-
tion [25] in contrast to the presence of an N-acetylalanine in the natural protein
1281. To further characterize the modification, rSmp2817 and rSmp2812 were
separated by anion exchange HPLC using a shallow gradient of potassium chlo-
ride (KCI) at pH 8.55 (0.5 m M KClImin) (Fig. 2). Anion exchange HPLC allowed
separation of three major forms of the protein (rSmp2811, 2, and 3). The nature
of the modification in rSmp2813, which appeared only after purification, has not
been identified in this case but may be related to the binding of glutathione to
the single free cysteine residue (CyslS9)as previously shown [13]. rSmp2811 and
rSmp2812 were baseline-separated and appeared to be homogeneous when ana-
lyzed by isoelectric focusing in an immobilized pH gradient (Fig. 2, insert). It
is noteworthy that rSmp2812 was slowly transformed into rSmp2811 upon incu-
bation at 37°C in water, underscoring the labile nature of the N-terminal modifi-
cation.
T m e (min)
Figure 2 Separation of rSmp2811 and rSmp2812 by anion exchange HPLC. A third

form (rSmp2813) with increased affinity for the stationary phase is visible. This form
occurred after purification and may correspond to the previously described glutathiony-
lated rSmp28 1131. Insert: lsoelectric focusing of purified rSmp2811 and rSmp2812 in an
immobilized pH gradient gel covering the pH range from 7.1 to 8.1. (From Ref. 39.)
Recombinant Proteins and MS
B. Peptide Mapping by LC-MS

LC-MS of the tryptic peptide maps of rSmp2811 and rSmp2812 resulted in very
similar chromatograms (Fig. 3). However, comparison of the traces revealed that
the map of rSmp2812 contained an additional peak at 12.5 min (marked with a
circle in Fig. 3B) with a measured mass of 679.8 + 0.2 Da. Amino acid composi-
tion analysis of the corresponding peptide was in good agreement with the ex-
pected composition of the N-terminal tryptic fragment AGEHIK, whereas the
measured mass deviated by 26.1 Da from the expected value of 653.7 Da. Further-
more, the peptide fragment proved to be inaccessible to automated Edman degra-
dation in contrast to rSmp2812 from which it was derived. Analysis of the broad
peak preceding the modified peptide fragment showed that the peptide map of
rSmp2812 also contained some nonmodified N-terminal peptide (retention time
11.2-1 1.5 min; labeled with an asterisk in Fig. 3B) with a measured mass of
653.9 + 0.2 Da. The peptide map of rSmp2811 contained only the nonmodified
N-terminal peptide fragment eluting between 11.2 and 11.5 min (labeled with an
asterisk in Fig. 3A; measured mass 653.8 -t 0.4 Da). These results indicated that
the labile N-terminal modification of rSmp2812 had partially rearranged to result
in a permanently blocked N-terminal peptide fragment and a smaller amount of
the unmodified N-terminal peptide. Complete analysis of the residual peptide
map by mass matching and Edman degradation confirmed that all other peptide
fragments were unmodified and identical for both rSmp2811 and rSmp2812, indi-
cating strongly that the shift in isoelectric point was due to modification of the
amino terminal region.
C. Analysis of the Modified N-Terminal Peptide from

rSmp2812 by Reversed-Phase HPLC and Deuterium
Exchange Mass Spectrometry
In order to characterize the modification within the N-terminal peptide in more
detail, reversed-phase HPLC with a shallow gradient of acetonitrile (0.27%lmin)
was employed to confirm its homogeneity. Surprisingly, the modified peptide
was separated into two components (Fig. 4), which had the same amino acid
compositions as well as the same molecular mass of approximately 680 Da. Isom-
erization of the N-terminal peptide fragment obtained after tryptic digestion of
rSmp2812 can be explained by the fact that the peptide contains a cyclic linkage
that formed during digestion. Since this phenomenon was only observed in the
case of rSmp2812, it is likely to be related to the original N-terminal modification
that led to the shift in isoelectric point.
To test the hypothesis of a cyclic linkage, both the free and blocked N-
terminal peptides were subjected to deuterium exchange followed by MS. In this
Retention Time (mini
Figure 3 Peptide mapping of rSmp2811 (A) and rSmp28t2 (B) after tryptic digestion by on-line liquid chromatography
electrospray ionization mass spectrometry (LC-MS). Traces show the base-peak intensity corrected total ion current. The
modified N-terminal peptide fragment in rSmp2812 is marked with a circle (B) and the nonmodified N-terminal peptide
fragments are labeled with asterisks (A and B). (From Ref. 39.)
7
Recombinant Proteins and MS
15 20 25 30 35 40
Time (min)
Figure 4 Separation of the modified rSmp28 N-terminal peptide fragment into two iso-
meric components by reversed-phase HPLC. The separated components proved to have
identical amino acid compositions and molecular masses. Both were inaccessible to N-
terminal Edman degradation. The ratio between the earlier and later eluting peptides was
1 :2 as determined by quantitative amino acid composition analysis. (From Ref. 39.)
Bischoff and Bouchon
Ala-Gly-Glu-His-Ile-Lys
Figure 5 Schematic representation of the structure of the unmodified rSmp28 N-termi-
nal peptide fragment showing the expected number of exchangeable hydrogen atoms and
fragmentation sites.
analysis, MS will detect an increase of 1 Da for each exchanged hydrogen atom.

Since only acidic hydrogen atoms bound to nitrogen (amines, amides, imidazoles)
or to oxygen (carboxylic acids, alcohols) will exchange, one would expect a mass
increase of 12 Da for the nonmodified N-terminal peptide (Fig. 5). While the
free peptide derived from rSmp2811 showed the expected mass increase of 12.1
Da (Fig. 6A), its blocked counterpart showed only a mass increase of 10.0 Da,
which is consistent with a cyclic modification removing two exchangeable hydro-
gen atoms (Fig. 6B).
D. Tandem Mass Spectrometry of the Modified

N-Terminal Peptide
While the above results indicated that the modification of the N-terminal peptide
fragment was due to intramolecular cyclization, it did not allow localization of
its position within the peptide or determination of its structure. Tandem MS with
collision-induced dissociation (CID) was thus applied to obtain characteristic
fragment ions. Due to the limited amount of modified peptide available (approxi-
mately 15 pmol), tandem MS was performed using the nanoelectrospray tech-
nique, which permits introduction of the sample into the electrospray ion source
at a flow rate of approximately 20 nllmin [29]. It was thus possible to extend
the measuring time to 2.5 h on a sample volume of 3 p1, which allowed thorough
'1" N-TERMINAL PEPTIDE: rSmp28/1
Measured Mass:665.8 Da
Expected Mass @on-deutemted): 653.7 Da
Mass Difference: 12.1 Da
3
-
P)
0 - I -
691.8
I I l . - - l . ~ ~ ~ l ~ l
N-TERMMAL PEPTIDE: rSmp2OR w+D]1+

Measured Mass: 689.8 Da
Measured Mass (non-deuterated): 679.8 Da
Mass Difference: 10.0 Da
Figure 6 Deuterium-hydrogen exchange of the nonmodified (A) and modified (B) N-terminal peptide fragments of rSmp28 in DzO
(expected number of exchangeable protons in the nonmodified peptide = 12; see Fig. 5 ) . The measured mass of the nonmodified peptide h,
A
(665.8 Da) differed by 12.1 Da from the expected mass of the nondeuterated form in agreement with its structure, whereas the measured
mass of the modified peptide (689.8 Da) differed by only 10.0 Da from the measured mass of its nondeuterated form. indicating the loss
of two exchangeable hydrogen atoms. (From Ref. 39.)
Figure 7 Tandem mass spectrometry of the unmodified (A and C; doubly charged parent ion: 328 amu) and modified N-terminal peptide
(B and D; doubly charged parent ion: 341 amu) of rSmp28 using the nanoelectrospray technique [29]. (For exact mass values, see Table
1; the region between 180 and 355 amu is not shown.) Panels A and B show a comparison of the fragments obtained in the range from
40 to 180 Da. It is noteworthy that the a , fragment of the modified peptide appeared at the expected mass (mlz = 44.0) as well as at mlz
= 70.0 (Am: +26.0 Da) confirming that Alal was one of the modified amino acid residues. Furthermore, the y", fragment of the modified
peptide occurred at the expected mass (mlz = 147.0) as well as at mlz = 173.2 (Am: +26.2 Da) (Fig. 7B) identifying Lys6 as another
modified site. The abundant immonium ion i,, which is derived from HisJ [30], occurred at the expected value (nzlz = 110.0) and at
mlz = 136.0 (Am: +26.0 Da) In the modified peptide (see Fig. 7B). These results show that cyclization occurred between Ala' and Lys6
as well as with Hisvsee Fig. 8). These results were corroborated by other a, b, and y" ions in the region between 355 and 650 amu, which
showed a +26 Da series in the case of the modified peptide (Fig. 7D) while the unmodified peptide gave the expected fragment ions (Fig.
7C). Fragment ions at 404.5 and 517.6 Da could not be assigned but are related by a Am of 113.1 Da, which corresponds to Ile. (From
Ref. 39.)
Table 1 Comparison of the Measured and Calculated Mass Values for Fragment Ions Obtained by Nanoelectrospray Tandem Mass
Spectrometry of rSmp28 N-Terminal Peptides (see Fig. 7).
Measured mass (Da) Measured mass (Da) Expected
Fragment ion" Nonmodified peptide Modified peptide mass (Da) Fragment
-
147.0 and 173.2 (Am: +26.2) L Y ~

260.2 and 286.4 (Am: +26.2) Ile-Lys
397.6 and 423.6 (Am: +26.0) His-Ile-Lys
526.6 and 552.7 (Am: +26.1) Glu-His-Ile-Lys
583.6 and 609.7 (Am: +26.1) Gly-Glu-His-Ile-Ly s
44.0 and 70.0 (Am: +26.0) Ala
101.8 and 127.0 (Am: +25.2) Ala-Gly
n.d. Ala-Gly-Glu
393.3 (Am: +25.9) Ala-Gly-Glu-His
n.d. Ala-Gly -Glu-His-Ile
129.2 and 155.1 (Am: +25.9) Ala-Gly
42 1.4 (Am: +26.0) Ala-Gly-Glu-His
n.d. Ala-Gly-Glu-His-Ile
110.0 and 136.0 (Am: +26.0) His-derived immonium ion
"ccording to Ref. 38
Not detected.
optimization of the fragmentation conditions as well as adjustment of the resolu-

tion of the mass analyzers.
Edman degradation had already established that the N termini of the modi-
fied peptides were blocked, implicating N-terminal alanine as one of the modified
sites. Comparative analyses of fragments generated by CID from the doubly
charged molecular ions of both the modified and the unmodified N-terminal pep-
tides (parent ions: ml,: = 328 and 341, respectively) confirmed that the N-termi-
nal alanine residue is one of the modified sites, since the a, fragment appeared
at the expected 44.0 Da in the case of the unmodified peptide whereas a second
fragment at 70.0 Da (Am = 26.0 Da) was generated for the modified peptide
(Fig. 7A and B). The fragmentation pattern of the free peptide allowed assignment
of the complete yo-ion series and most of the a-ion series (Table 1, Fig. 7A
and C), which provided a comparative spectrum to assign the modified sites.
Furthermore, three of the five expected b-ions and a histidine-related immonium
fragment ion (i,: mlz = 110.0) were detected [30]. The pattern was in complete
agreement with the expected sequence AGEHIK. In comparison, the fragmenta-
tion pattern of the modified peptide differed from its nonmodified counterpart
with respect to they"-, b-, and a-ion series as well as the histidine-related immon-
ium ion, which occurred at the expected mass values as well as at mass values
that were 26 Da higher (Table 1, Fig. 7B and D). A mass difference of 26 Da
was indicative that the respective fragment ion contained the unknown modifica-
tion. This mass difference was found in all of the five y"-ions and in particular
in the y",-ion which corresponded to the C-terminal lysine residue (Fig. 7A),
indicating that an intramolecular crosslink had formed between Ala' and
Lysh. Furthermore, the histidine-related internal immonium fragment ion i4
(mlz = 110.0) was modified giving rise to a second fragment ion at 136.0 Da.
Taken together these results showed that cyclization of the N-terminal peptide
derived by tryptic digestion from rSmp2812 occurred between the N-terminal
amino group and the C-terminal lysine residue as well as with His4 in agreement
with the obserkation that two isomers were formed upon cyclization.
E. rSmp2812 Is Modified b y A c e t a l d e h y d e During

Culturing of S. cerevisiae
The obtained results are consistent with the initial formation of an aldimine at
the N terminus of rSmp28 due to the addition of acetaldehyde followed by intra-
molecular cyclization with either His4 or Lys6 during tryptic digestion to produce
two cyclic peptide derivatives as shown in Fig. 8. This interpretation is based on
the following data. (1) rSmp2812 is modified in a reversible manner, since
rSmp2811 was regenerated upon incubation for 3 days at 37OC in water. Such
chemical instability has also been observed for a peptide that was modified at
its N terminus with acetaldehyde in the case of the hemoglobin P chain [31].
Furthermore, rSmp2812 is not N-terminally "blocked" when analyzed by Edman
degradation excluding modifications such as N-terminal formylation or acetyla-
tion and again underlining the labile nature of the modification. (2) Tryptic diges-
tion of rSmp2812 produced a modified form of the N-terminal hexapeptide
AGEHIK that was blocked to Edman degradation. The observed mass difference
Ala-Gly-Glu-His-lle-Lys-PROTEIN
/
Tryptic Digestion
pH 8.3
cyclization via His
cyclization via Lys
Figure 8 Schematic drawing of the mechanism leading to the generation of two isomeric
cyclic peptides after tryptic digestion of the aldimine-modifiedprotein rSmp2812. (From
Ref. 39.)
of 26 Da, measured with a precision of +0.2 Da, eliminated N-terminal formyla-

tion as a possibility and is consistent with the cyclic modification shown in Fig. 8.
(3) The isolated peptide was composed of two isomers that lacked two deuterium-
exchangeable hydrogen atoms as compared to the nonmodified peptide, which
is consistent with the cyclic structures shown in Fig. 8. (4) Fragments obtained
by tandem MS localized the 26-Da mass difference to the N-terminal alanine as
well as to the C-terminal lysine and the internal histidine residue identifying these
amino acids as part of the cycle. This example of an unexpected posttranslational
modification shows that its successful identification was due to highly resolving
chromatographic methods to purify the modified protein and to obtain the modi-
fied peptide after tryptic digestion as well as to detailed mass spectrometric mea-
surements including tandem MS.
Ill. ANALYSIS OF THE POSlTRANSLATIONAL LIPIDATION

OF RECOMBINANT OUTER SURFACE PROTEIN A
(rOspA) FROM BORRELIA BURGDORFERI
Lyme disease is a tick-borne infection caused by the spirochete Borrelia burg-

dorferi. Infected people develop antibodies in response to several B. burgdorferi
antigens. The outer surface protein A, OspA, a 257-amino-acid protein, is one
of them [32] and is thus considered as a potential vaccine candidate. Cloning of
the gene suggested that OspA is likely to be a lipoprotein due to the lipopeptide
consensus signal sequence [33]. This was confirmed by incorporation of [3H]pal-
mitate in OspA lipoprotein in vivo [24] and by electrospray mass spectrometry
(ESMS) analysis of the purified recombinant protein [26,27].
Bacterial lipoproteins constitute a class of strongly hydrophobic membrane
proteins, with an N-terminal cysteine posttranslationally modified to an S[2,3-
bis(acy1oxy)-(2RS)-propyll-N-acylcysteine [34] (Fig. 9), according to the model
established by Hantke and Braun [35]. Very few structural studies have been
performed on such lipoproteins due to their amphipathic nature [26,36] which
renders purification of even analytical amounts rather difficult. In order to analyze
the lipidation of rOspA produced in E. coli, a combination of GC-MS and normal
phase HPLC was developed.
A. Fatty Acid Analysis

Analyses of the intact protein indicated that it was posttranslationally modified
through lipidation of the N-terminal cysteine [26,27] (see Fig. 9). After detergent
removal by acetone precipitation, intact rOspA was submitted to an acidic metha-
nol treatment in order to generate fatty acid methyl esters via transmethylation.
Fatty acid methyl esters were extracted into hexane, which permitted their analy-
S-[2,3-Bis(palmitoyloxy)-(2RS)-propyl]-N-palmitoyl-(R)-cysteinyl ("Pam3Cys")configuration
(A2, A3, A4, ... : amino acids)
Figure 9 Structure of the lipidated N-terminal tryptic peptide CK of rOspA containing

three palmitoyl residues.
sis by GC-MS. The chromatogram (Fig. 10) shows the fatty acid composition of
rOspA, with fatty acids ranging from C14 to C18. The individual mass spectra
of each of these fatty acids show the molecular ion and the typical fragmentation
pattern permitting the identification of saturated fatty acids (C 14: 0, C 16 :0; A
and C) and monounsaturated fatty acids (C 16 : I, C 18 : 1; B and D) [37]. C 16 :0
(65%) and C16: 1 (21%) were the major components of the total fatty acid con-
tent. To ascertain that the fatty acids were attached to the N terminus as expected,
complementary analyses were performed focusing on the lipidated N-terminal
tryptic peptide.
B. Purification and Analysis of the N-Terminal Peptide

A tryptic digestion of rOsp A was performed on 1 nmol (25 pg) and the peptides
were analyzed by HPLC. While all attempts at using short-chain reversed-phase
HPLC supports (Cl, C4) failed to give reproducible results, normal phase HPLC
on a silica column (Iatrobeads) reproducibly separated the N-terminal lipopeptide
from the more hydrophilic peptide fragments. The lipidated peptide eluted first
Figure 10 GC-MS analysis of rOspA fatty acids. The trace is the summation of m/z 50-550 ion intensities. Time scale,
1 scan = 1 s. The components labeled A, B, C, and D have been identified by their mass spectrum to be (A) C14:0,
myristic acid (expected mass, 242 Da); (B) C16: 1 (expected mass, 268 Da); (C) C16:0, palmitic acid (expected mass,
270 Da); (D) C18:l (expected mass, 296 Da). Due to isomerization that occurs during the acidic treatment, the precise a
double-bond position could not be assigned in the unsaturated fatty acids. (From Ref. 27.)
Figure 11 Continued (B) ESMS spectrum of fraction 3 corresponding to rOspA N-terminal peptide (see Table 2). I
corresponds to the expected complete form (Pam3CysLys); IV to an incomplete form (Pam2CysLys); I1 and 111 to fully 5
lipidated forms but with two palmitates and one myristate or two palmitates and one octadecenoate residue, respectively.
Oxidized forms (cysteine sulfoxide) of the different forms were also observed (+ 16 Da). (From Ref. 27.)
as expected for the most hydrophobic component of the mixture (Fig. 11A). Anal-
ysis by ESMS of the collected fraction permitted localization of the N-terminal
peptide based on its mass and ionization profile (Fig. 11B). Due to the very short
peptidic chain (a single peptide bond, C'K2, Fig. 9), absorption at 205 nm was
very low, and the peptide appeared as a shoulder on the major chromatographic
peak. The purified peptide ( < I pg) was collected for further analysis.
Tryptic peptides are generally characterized during ESMS analysis by the
presence of two ionization positions (The N terminus and the basic C-terminal
residue, except for the C-terminal peptide of any digest). In the case of rOspA,
however, the N-terminal peptide presented only one ionization site due to N-
terminal blockage. The most abundant peak in this spectrum (Fig. 11B) corre-
sponded to a molecular mass of 1036.2 Da (expected mass of the N-terminal
peptide containing three palmitates: 1038.6 Da; Am = -2.4 Da), indicating that
one fatty acid was unsaturated. In addition, several minor components were de-
tectable. Peaks corresponding to plus and minus 28Da (I1 and I11 in Fig. 11B)
were present, indicating different forms of the lipidic moiety, the major one being
a peptide with three C16 fatty acids, whereas the minor forms contain two C16s
plus one C14 (I1 in Fig. 118) and two C16s plus one C18 (111 in Fig. 11B) in
agreement with the GC-MS data. Another peak corresponded to the loss of one
palmitate residue (IV in Fig. 1IB, Am = -238 Da). This could be attributable
to a fragmentation taking place in the source of the mass spectrometer, or most
likely to an incomplete lipidation, as previously described on the same molecule
[26]. The presence of a population of Pam2Cys-modified rOspA molecules that
had been detected in the intact protein was thus confirmed by the analysis of the
N-terminal peptide mixture that contained some Pam2CyrLys (IV in Fig. 1 IS;
expected mass, 800.2 Da; measured mass, 797.8 Da). The measured mass of the
N-terminal tryptic peptide appeared to be 2 Da lower than expected for two
C16:O indicating one unsaturated fatty acid in agreement with the GC-MS data
which showed the presence of about 20% of C16: 1.
To confirm these results, a fast atom bombardment (FAB) mass spectrum
using a magnetic sector mass analyzer was run on approximately 100 ng of the
remaining N-terminal peptides. A complex spectrum was obtained (Fig. 12). As
the sample had remained in solution for several days prior to FABMS analysis
it is likely that the increased complexity was due at least in part to oxidation of
the peptide at the cysteine-lipid thioether moiety. Additional peaks (I1 and I11
in Fig. 12 and Table 2) corresponding to plus and minus 28 Da on [M+H]+ were
present, confirming and extending the ESMS data on the lipidic composition. In
particular, the high mass resolution allowed discrimination of two components
differing in one unsaturation (Ia and Ib in Fig. 12 and Table 2). Fragmentation
(b, fragment) data indicated the presence of lysine in agreement with the expected
structure of the N-terminal peptide containing one cysteine and one lysine resi-
due. Masses corresponding to the loss of one palmitate residue (IV in Fig. 12.
Table 2 Molecular Species Detected in the ES and FAB Mass Spectra of the Lipidated N-Terminal Tryptic Peptide of rOspA
Average mass (Da) Monoisotopic mass (Da)
Theoretical fatty Measured Measured Label in
acid composition Expecteda (ESMS) Am Expecteda (FABMS) Am Proposed structures Figs. 3 and 4
Three C16:O 1038.6 1036.2 -2.4 1037.8 1035.7 -2.1 C16:O-C16:O-C16: 1 Ia
1037.7 -0.1 C16:O-C16:O-C16:O Ib
Two C16:O + One C14:O 1010.6 1008.9 -1.7 1009.8 1007.5 -2.3 C16:O-C16: 1-C14:O I1
Two C16:O + One C18:O 1066.7 1063.7 -3.0 1065.9 1063.6 -2.3 C16:O-C16:O-C18:l 111
Two C16:O 800.2 797.8 -2.4 799.6 797.3 -2.3 C16:O-C16:l IV
"ssuming completely saturated structures.
Am = -238 Da) confirmed that the major fatty acid residue was C16:O. These
results indicated the fatty acid composition of the major components to be C16:
0-C16: 0-C16: 1 and C16: 0-C16: 0-C16: 0 in accordance with results obtained
by GC-MS. Of the two minor components, one had a lipidic composition of C14:
0-C16: 1-C16:0, and the other component had a lipidic composition of C18 : 1-
C16 :0-C 16 :0, according to the GC-MS data.
In conclusion, the use of a combination of GC, normal phase HPLC, and
MS allowed a detailed characterization of the lipidation of rOspA. The heteroge-
neity of the molecule related to its lipidic composition was analyzed using fatty
acid analysis by GC-MS. It indicated that C16:O is the major component as ex-
pected [35] but that other fatty acids including C 16 : 1, C 14:0, and C 18 : 1 were
also present. Purification of the N-terminal peptide and its analysis by ESMS and
FABMS indicated that a minor fraction of rOspA molecules contained a Pam2-
Cys modification and that the major rOspA form corresponded to a peptide having
three C16 residues, two of them being saturated and one being unsaturated. This
structure is consistent with the proposed mode1 for bacterial lipoproteins [35] in
that the major fatty acid component is palmitate.
IV. DISCUSSION
Posttranslational modifications of proteins are often related to biological function

as is strikingly demonstrated by the importance of phosphorylations and dephos-
phorylations. In this chapter we describe two types of posttranslational modifica-
tions: lipidation, which was expected and assures localization of the protein to
the outer membrane of B. burgdorferi, and addition of acetaldehyde to the amino
terminus, which was not expected and results from anaerobic glucose metabolism
of S.cerevisiae during high-cell-density culturing.
In both cases a combined strategy using efficient separation techniques
based on HPLC or GC methods and MS led to a detailed structural understanding
of these modifications. A key step to the successful analysis in the case of rOspA
was the development of a normal phase chromatographic method that allowed
isolation of the amphipathic N-terminal lipopeptide after tryptic digestion of the
protein. Final structural proof of the addition of acetaldehyde to the N terminus
of rSmp28 was obtained by tandem MS of the modified N-terminal peptide in
comparison with its nonmodified counterpart. Methodological advances as de-
scribed in this chapter will provide a more detailed insight into the structure of
recombinant proteins ultimately allowing to define them in physicochemical
terms similar to chemically synthesized pharmaceuticals. Such an approach will
help to assure the consistent quality of biopharmaceuticals with the goal of pro-
ducing safe and efficient medicines.
ACKNOWLEDGMENT
The contributions from colleagues in the Yeast and Pilot Development Depart-
ments at Transgene and the Laboratoire de Spectromktrie de Masse Bioorganique
(Universitk Louis Pasteur, Strasbourg) are gratefully acknowledged.
REFERENCE
Koths K. Curr Biol l995;6:68 1-687.

Buckel P. Trends Pharm Sci 1996;17:450-456.
Andersen JS, Svensson B, Roepstorff P. Nature Biotechnol 1996;14:449-457.
Fenn JB, Mann M, Meng CK, Wong SF, Whitehouse CM. Science 1989;246:64-
71.
Karas M, Hillenkamp F. Anal Chem 1988;60:2299-2301.
Carr SA, Hemling ME, Bean MF, Roberts GD. Anal Chem 1991;63:2802-2824.
Chait BT, Kent SB. Science 1992;257: 1885-1894.
Aebersold R, Hess D, Morrison HD, Yungwirth T, Chow DT, Affolter M, Amankwa
LN. J Protein Chem 1994;13:465-466.
Hopfgartner G, Wachs T, Bean K, Henion J. Anal Chem 1993;65:439-446.
Tomer KB, Moseley MA, Deterding LJ. Parker CE. Mass Spectrom Rev 1994;13:
431 -457.
Coulot M, Domon B, Grossenbacher H, Guenat C, Maerki W, Miiller DR, Richter
WJ. J Mol Struct 1993;292:89-104.
Ling V, Guzzetta AW. Canova-Davis E, Stults JT, Hancock WS, Covey TR, Shushan
BI. Anal Chem 1991;63:2909-2915.
Klarskov K, Roecklin D, Bouchon B, Sabatie J, Van Dorsselaer A, Bischoff R. Anal
Biochem 1994;216:127-134.
Smith RD, Udseth HR, Wahl JH, Goodlett DR, Hofstadler SA. Meth Enzymol 1996;
271 Part B:448-486.
Carr SA, Huddleston MJ, Bean MF. Protein Sci 1993;2:183-196.
Conboy JJ, Henion JD. J Am Soc Mass Spectrom 19923304-814.
Huddleston MJ, Bean MF, Carr SA. Anal Chem 1993;65:877-884.
Huddleston MJ, Annan RS, Bean MF, Carr SA. J Am Soc Mass Spectrom 1993;4:
710-717.
Amankwa LN, Harder K, Jirik F, Aebersold R. Protein Sci l99S;4: 1 13-1 25.
Sutton CW, O'Neill JA, Cottrell JS. Anal Biochem 1994;218:34-46.
Huberty MC, Vath JE, Yu W. Martin SA. Anal Chem 1993;65:2791-2800.
Bischoff R, Lepage P, Jaquinod M, Cauet G, Acker-Klein M. Clesse D, Laporte M,
Bayol A, Van Dorsselaer A. Roitsch C. Biochemistry 1993;32:725-734.
Balloul JM, Sondermeyer P, Dreyer D, Capron M, Grzych JM, Pierce RJ, Carvallo
D, Lecocq JP, Capron A. Nature 1987;326:149-153.
Erdile LF, Brandt MA, Warakomski DJ, Westrack GJ, Sadziene A, Barbour AG.
Mays JP. Infect lmmun 1993;61:8l-90.
Bischoff R, Roecklin D, Roitsch C. Electrophoresis 1992;13:214-219.

Bouchon B. Van Dorsselaer A, Roitsch C. Biol Mass Spectrom 1993;22:358-360.
Bouchon B, Klein M, Bischoff R, Van Dorsselaer A, Roitsch C. Anal Biochem
1997;246:52-61.
Bouchon B, Jaquinod M, Klarskov K, Trottein F, Klein M, Van Dorsselaer A,
Bischoff R, Roitsch C. J Chromatogr 1994;662:279-290.
Wilm M, Mann M. Anal Chern 1996;68:1-8.
Falick AM, Hines WM, Medzihradsky KF, Baldwin MA, Gibson BW. J Am Soc
Mass Spectrom 1993;4:882-893.
Sillanaukee P, Hurme L, Tuominen J, Ranta E, Nikkari S, Seppa K. Eur J Biochem
1996;240:30-36.
Barbour AG, Heiland RA, Howe TR. J Infect Dis 1985;152:478-484.
Bergstrom S, Bundoc VG, Barbour AG. Mol Microbial 1989;3:479-486.
Hayashi S, Wu HC. J Bioenerg Biomemb 1990;22:451-471.
Hantke K. Braun V. Eur J. Biochem 1973;34:248-269.
Metzger J, Beck W, Jung G. Angew Chem Int 1992;31:226-228.
Murphy RC. Mass Spectrometry of Lipids: Handbook of Lipid Research. Vol. 7.
New York: Plenum Publishing, 1993.
Roepstorff P, Fohlmann J. Biomed Mass Spectrom 1984;11:601.
Roecklin D, Klarskov K, Cavallini B, Sabatie J, Bouchon B, Loew D, Van Dorssel-
aer A, Bischoff R. Eur J Biochem 1997;245:589-599.
3
Analytical and Preparative
Separations of Peptides
by Capillary and Free-Flow
Zone Electrophoresis
Vaclav KaSiCka
Academy of Sciences of the Czech Republic, Prague, Czech Republic
I. INTRODUCTION
Peptides represent a large and complex group of biologically active substances.

Their occurrence and function in nature are very variable and of vital importance.
Peptides act for example, as hormones, neurotransmitters, immunomodulators,
coenzymes or enzyme inhibitors, drugs, toxins, and antibiotics. The demands of
chemists, biochemists, and all other specialists dealing with peptide research and
applications for the methods that should be able to separate and characterize
peptide preparations both qualitatively and quantitatively are continuously in-
creasing. consequently, the application of capillary and free-flow zone electro-
phoresis for peptide analysis, preparation, and physicochemical characterization
is an area of great importance.
Natural peptides are composed of about 20 amino acid species linked by
peptide bonds and sometimes also crosslinked by disulfide bridges. In the syn-
thetic preparations of peptides (analoges, derivatives, and fragments of biopep-
tides), noncoded amino acid residues and various types of crosslinking might
also occur. In practice, an inexhaustible number of variations of amino acid se-
quences in a polypeptide chain is the cause of an extraordinary diversity of pep-
tides. They differ in their size (depending on the number of the linked amino
acid residues, their relative molecular mass can range from few hundreds for
oligopeptides containing 2-10 amino acid residues up to several thousands for
polypeptides containing tens of amino acid residues), electric charge, and steric
arrangement. Charge and shape of a peptide molecules depend not only on the
type and number of amino acid residues but also on their sequence in the peptide
chain. All of these differences result in different electrophoretic mobilities of
peptides, which allows their separability by electromigration methods. In this
chapter some applications of two free-solution (camerless) electromigration sep-
aration methods, high-performance capillary zone electrophoresis (CZE) and
free-flow zone electrophoresis (FFZE), to peptide analysis and preparation will
be described. Based on the correlation between these two techniques a procedure
for conversion of analytical peptide separation into preparative scale will be dem-
onstrated.
II. STRATEGY FOR SELECTION

OF SEPARATION CONDITIONS
A. General
Strategy for the rational selection of the experimental conditions of analytical
and preparative separations of peptides by CZE and FFZE results from the elec-
tromigration properties and other specific characteristics of peptides and from
general rules for the development and optimization of CZE and FFZE separation
procedures. The basic aspects of this strategy, including detection of peptides in
CZE separations, will be described in this section.
B. Electromigration Properties of Peptides

1. Effective Mobility
As electromigration properties of peptides we indicate the properties connected
with their electrophoretic mobility, which is the central magnitude of all electro-
migration separation methods [l-31. Electrophoretic mobility, m, is defined as
a velocity, v, of the movement of a charged component in a liquid medium in
the dc electric field of unit intensity:
where E is the intensity of electric field. The mobility of a peptide, similar to

mobility of any other substance, is a complex function of
1. Properties of the peptide itself (charge, size, shape)
2. Properties of a medium (background electrolyte) in which peptide is
moving (composition, pH, ionic strength, viscosity, permitivity, and
temperature)
Separations of Peptides by CZE and FFZE 41
3. Interactions of peptide with components of the medium (solvation, dis-

sociation, complex formation)
Depending on so many parameters, peptide mobility has to be referred to
given experimental conditions under which it was determined, i.e., it must be
related to the given composition, pH, and temperature of the background electro-
lyte. In this way characterized mobility is called effective mobility and only the
effective mobilities of peptides obtained under the same conditions are compara-
ble electromigration characteristics of peptides.
The relationship between effective mobility of peptide P, m,,,, and its ef-
fective charge, z,,,, and relative molecular mass, Mp, has been investigated from
the beginning of electrophoretic separations of peptides and proteins. First it was
quantitatively described by Offord [4] who derived empirically from peptide sep-
arations by paper zone electrophoresis the following equation:
where k is an empirical constant.

The validity of offord's relation was confirmed also for free-solution CZE
by Rickard et al. [5], whereas Compton [6] prefers the form:
Other forms were suggested by Grossman et al. [7]:

mp.=r= A log ( z ~ , +
, ~ l)/nR (4)
where n is the number of amino acid residues, and by Cifuentes and Poppe [8]:
In all of these equations A, B, and C are empirical constants that are depen-
dent on the electrolyte system used and on the size and structure of peptides
investigated. Small peptides may behave as simple organic ions of a compact
structure (tightly would coil); longer peptides may occur in the form of random
coil or linear chain.
In spite of different forms of the above given relation it can be concluded
that the mobility of peptide is directly proportional to its charge and indirectly
proportional to its relative molecular mass with exponent in the range 113-213.
The influence of charge, size, and shape of peptides and proteins on their
mobilities is discussed in detail in some special articles, book chapters, and re-
views [9- 171.
2. Effective Charge
From the physicochemical point of view, peptides are amphoteric (po1y)electro-
lytes or (po1y)ampholytes. They contain in their molecules different types of iono-
genic groups, e.g., carboxyl groups of C-terminal amino acids and of side chains
of aspartic and glutamic acids, amino groups of N-terminal amino acids and of
side chain of lysine, etc. The survey of ionogenic groups present in peptides and
the approximate range of their acid dissociation constants excerpted from the
literature [18-201 are given in Table 1. The effective (net) charge of peptides
(sum of all charges including their signs) is given by the sum of charges of all
ionogenic groups present in the polypeptide chain. If amino acid sequence or at
least amino acid composition of a peptide is known, then from the present iono-
genic groups and from their dissociation constants it is possible to estimate which
groups are dissociated at a given pH and how they conribute to the effective
charge. Generally, a given group is dissociated to 50% in a solution the pH of
which is equal to the pK value of that group. For pH = pK + 1, the group is
90% dissociated and for pH = pK - 1 the group is 10% dissociated.
The amino acid residues of aspartic and glutamic acids are the main contrib-
utors of negative charge, whereas arginine, lysine, and partly histidine are camers
of the positive charge in peptide chain.
In addition to this rough estimation, a more precise, computer program-
based calculation has been developed that allows one to calculate the effective
and specific charges of any peptide whose amino acid sequence is known and
for which the pK values of present ionogenic groups are known or can be esti-
mated [21]. The program is based on the mathematical model of the acid-base
equilibria of a general ampholyte. It is described in detail elsewhere [22]; here
only the main points will be mentioned.
Let us consider a peptide P that can have maximal (positive) charge M and
minimal (negative) charge N. M and N express numbers of elementary charges
Table 1 Approximate Range of pK Values of Ionogenic Groups

of Amino Acid Residues in Polypeptide Chain
Ionogenic group Amino acid residue PK
- - -
-S03H Cysteic acid

a-COOH C terminal of peptide chain
p-COOH Aspartic acid
-S-CH2-COOH S-carboxymethylcysteine
y-COOH Glutamic acid
Imidazolium Histidine
a-NH3+ N terminal of peptide chain
-SH Cysteine
&-NH3+ Lysine
Phenol Tyrosine
Guanidium Arginine
including sign, i.e., maximum and minimum are meant in the mathematical sense.
Let P(J) be an ionic form of peptide P with charge J and K(J) is the apparent
dissociation constant of the equilibria between the ionic forms P(J + 1) and P(J):
where cp,, and c ~ ( ~ are

+ , , the equilibrium concentrations of peptide ionic forms
P(J) and P(J + 1) and cHis the equilibrium concentration of hydrogen ions. The
molar fraction of component P(J), Dp(j),referred to the total concentration, cp, of
peptide P is defined as
It was derived that Dp(j,can be expressed as a function of cH with parameters

K(J), N, M, i.e.,
where J E (N, M), J f 0, H = CH. If molar fractions of all ionic forms P(J) are
calculated, then the effective charge of peptide P, zPzef,
can be determined:
From the effective charge, Z P , the

~ ~ ,specific charge, z,,~,,i.e., the charge referred
to unit relative molecular mass, Mp, of peptide P can be calculated:
or the ratio of effective charge and relative molecular mass with a general expo-
nent d, called corrected specific charge, z ~ , ~ ~ , ~ ~ ~ :
From the known molar fractions Dp(J,of the individual ionic forms of peptide P
and from their known or estimated ionic mobilities, mp(,,,the effective mobility
of peptide P can be calculated as the weighted sum of ionic mobilities:
where sign(J) = + 1 for J > 0, sign(J) = -1 for J < 0, and sign(J) = 0 for
J = 0.
Equation (12) represents a mathematical definition of effective mobility of
a general ampholytic electrolyte. Mobilities of cations are positive values, mobili-
ties of anions are negative values.
The calculated dependencies of effective, specific, and corrected speci-
fic charges [effective charges divided by relative molecular mass M, with expo-
nent d = 213; see Eq. (1 I)] of simple peptides of diglycine (N = - 1, M = 1,
M, = 132.2, pK(0) = 3.12, pK(- 1) = 8.17) and triglycine [N = - 1, M = 1,
M, = 189.3, pK(0) = 3.26, pK(-1) = 7.911 on pH are shown in Fig. 1.
It is obvious that from these dependencies the useful information for
the selection of pH and composition of background electrolyte (BGE) can
be obtained, e.g., the regions of minimal (negative) and maximal (positive)
charges, regions of strong or weak dependence of effective and specific charges
on pH, and the pH range in which peptide charge is close to zero can be deter-
mined.
An important electromigration characteristic of peptides is their isoelectric
point (PI), i.e., the pH value of a solution in which the effective charge and
consequently the effective mobility is equal to zero. For small oligopeptides con-
taining only few amino acid residues with distant pK values of their ionogenic
groups, the pH range in which effective charge is close to zero is relatively broad
(see Fig. 1) and instead of isoelectric point we are speaking about isoelectric
zone. Isoelectric point differs from the similar characteristic, isoionic point, in
that it is related to a given buffer coniposition in which the measurement of
charge (mobility) is performed. This includes electrostatic interactions with all
the ions present in the solution whereas the isoionic point takes into account only
interactions with protons. The isoelectric and isoionic points are mostly close
values, but generally not identical. Peptide moves in a dc electric field as cation
at pH < pI and as an anion at pH > PI. The isoionic point can be read from the
above given dependence of effective charge on pH (see Fig. 1).
The calculated pH dependence of effective charges and of corrected specific
charges, i.e., effective charges divided by relative molecular mass with exponent
213, for longer peptides with higher number of ionogenic groups. growth hor-
mone-releasing peptides (GHRPs), insulin and desoctapeptide insulin, are pre-
sented in Figs. 2 and 3. Figure 2 shows these dependencies for His'-GHRP (hexa-
peptide with the sequence His-D-Trp-Ah-Trp-~-phe-Lys.NH?and with the
following parameters of calculation: N = 0, M = 3, M, = 873.1. pK(2) = 5.8,
pK(1) = 7.8, pK(0) = 10.4) and for Tyrl-GHRP (hexapeptide with the sequence
Tyr-D-Trp-D-Ala-Trp-~-phe-Lys.NH~ and with the following parameters, N = - 1,
M = 2, M, = 899.1, pK(1) = 7.6, pK(0) = 10.0, pK(-I) = 10.4). Figure 3
shows the pH dependencies of effective and specific corrected charge of hu-
man insulin (two-chain polypeptide consisting of 51-amino-acid residues,
M, = 5750.0, N = - 10, M = 6) and des-B23-30-octapeptide insulin (43-amino-
acid residues, M, = 4837.8, N = -9, M = 5). The accuracy of these calculated
Separations of Peptides by CZE and FFZE
Effective charge
Gly2 -
Gly3 - - - - .
SpeclBc charge
0.0072t '1' ' ' ' ' ' ' ' ' ' ' 4
Corrected specik charqe
Figure 1 Dependence of (A) effective charge, (B) specific charge (effective charge di-
vjded by relative molecular mass), and (C) corrected specific charge (effective charge
divided by relative molecular mass with exponent 213). of diglycine, c ~ w Gly2, e and
triglycine, curve Gly3, on pH. Parameters of the calculation, pK values, and relative rnolec-
ular masses are given in the text.
Effective charge
3.0
Hisl-GHRP -
Tyrl-GHRP - - - - -
1.0 -
0.0
!
I
A
,'.
Corrected specific charge
Figure 2 Calculated dependence of (A) effective charge and (B) corrected specific
charge (effective charge divided by relative molecular mass with exponent 213) of growth
hormone-releasing peptides (GHRPs), His'-GHKP and Tyr'-GHRP, on pH. Peptide se-
quences and parameters of the calculation are given in the text.
dependencies is negatively influenced by the fact that the average (estimated) pK

values of the present ionogenic groups (see Table 1) were used instead of individ-
ual constants of these groups in the given peptides. In reality, pK values depend
not only on the amino acid residue itself but on the neighboring amino acid
residues and on the steric arrangement of a polypeptide chain. For that reason
Effective charge
Insulin -
DO1 ---- -
-4 -
-8 -
A
0 2 4 6 8 10
pH
14
Corrected specific charge

0.0156 Insulin -
DO1 ----
0.0078
Figure 3 Calculated dependence of (A) effective charge and (B) corrected specific
charge (effective charge divided by relative molecular mass with exponent 213) of pig
insulin and desoctapeptide-B23-B30-insulin (DOI) on pH. Parameters of the calculation
are given in the text.
the conclusions derived from the calculated dependencies can be used only as
the first approximation for the selection of the suitable pH of BGE, which can
be further optimized by experimental tests. Similar strategy for the prediction of
peptide charge and mobility from the peptide sequence has been published by
Cifuentes and Poppe [8] who try to predict more precise values of individual
pK values of ionogenic groups present in peptide chain taking into account the
environment of given ionogenic group.
The pH dependence of the effective charge and/or effective mobility of
peptides can also be obtained by other methods, e.g., by acid-base titration curves
[23] or electrophoretic titration curves [24,25]. A large set of mobilities of small
oligopeptides in the pH range 6-10 was determined by capillary isotachophoresis
[26]. Isoelectric point of peptides can be determined by capillary or slab-gel
isoelectric focusing 127-291 or by CZE [30], or it can be calculated theoreti-
cally on the similar bases as the above-described approach for acid-base equi-
libria [3 1,321.
C. Experimental Conditions of Capillary and Free-Flow

1. General
When selecting the suitable experimental conditions for separation of peptides
by CZE and by FFZE, the general rules for the selection of the experimental
conditions of CZE [33-371 and FFZE [38] should be followed. In addition, the
following specific properties of peptides should be taken into account.
1 . Amphoteric character: the effective charge (mobility) of peptide is
strongly dependent on pH: peptides can be separated as cations or
anions (see Section 1I.B)
2. Diversity of amino acid composition and sequences resulting in a wide
range of solubilities and mobilities, including relatively low mobilities
and low solubilities
3. Biological activity, chemical and temperature lability
4. Complexity of peptide mixtures isolated from biological matrices (se-
rum, cerebrospinal fluid, tissue extracts) and interactions of peptides
with the components of these matrices
5. Tendency of the long peptides to be adsorbed to the inner walls of the
separation compartments
The influence of these factors on the selection of parameters of CZE and FFZE
separation is described below.
2. Composition and pH of Background Electrolyte

As follows from the amphoteric character of peptides, for the rational selection of
pH of the background electrolyte in which electromigration separation of peptides
should be performed it is advantageous to know the dependencies of the effective
and specitic charges and/or effective mobilities of these peptides on pH, or at
least their isoelectric points should be available (see Section 1I.B).
The pH of BGE should be at least 1-2 units distant from p l since close
to p l both electrophoretic mobility and solubility of peptide are relatively low.
The pH of BGE should be taken from the region where the specific charge is
greater than about 2 X e (e is elementary charge unit), i.e., a polypeptide
with relative molecular mass 5000 should possess at least one elementary charge
per molecule. With respect to the fact that effective mobility is directly propor-
tional to the effective charge and indirectly proportional to the relative molecular
mass [see Eqs. (2)-(5)], it is obvious that the difference of peptide mobilities
will be maximal in the same region where the difference of specific charges or
corrected specific charges of these peptides is maximal. Consequently, from this
region of maximal difference of charges the suitable pH of BGE should be
chosen.
For the given example of diglycine and triglycine it is evident that the
suitable pH for their separation is either at low acid region (pH < 3) or at alkaline
pH (pH > 9), since in these regions the differences of specific and corrected
specific charges are maximal. On the basis of these calculations and with respect
to other aspects (see below, 0.5 mol/L acetic acid, pH 2.5, was chosen as BGE
for the separation of these two peptides, which are often used as test mixture
in our laboratory. The record of CZE separation of these two peptides and the
electroosmotic flow marker, phenol, is given in Fig. 4.
Similarly, from the course of pH dependence of corrected specific charge
of GHRP derivatives (see Fig. 2), it follows that the suitable region for their
separation is in the acid pH region and that for pH < 4.5 the difference of cor-
Figure 4 CZE separation of test mixture of diglycine (0.3 mglml), peak 1, triglycine
(0.3 mglml), peak 2, and phenol (0.1 mglml), peak 3, using 0.5 mol1L acetic acid, pH
2.5 as BGE in a home-made CZE device described in Section 1II.B.I. Capillary: i.d. 0.056
mm, total (effective) length 310 (200) mm. Voltage 9.0 kV. current 10.1 FA; AZObr ab-
sorbance at 206 nm; t. migration time.
rected specific charges (effective mobilities) is independent of pH. Separation at
high pH (pH > 11) should be from the point of view of different specific charges
also possible but it cannot be recommended because of hydrolysis of amide
groups at high pH. As follows from the calculated pH dependence of effective
and corrected specific charges of insulin and desoctapeptide insulin (see Fig. 3),
the suitable pH of BGE for CZE separation of these two peptides is between 7
and 9 and from this range the pH of the BGE was chosen for monitoring of
tryptic conversion of pig insulin to desoctapeptide insulin by CZE (see Fig. 9 in
Section 1II.B.1).
According to the predicted suitable pH for peptide separation the composi-
tion of BGE is selected. The main constituent(s) of BGE should possess buffering
capacity at a given pH, i.e., their pK should fulfill the conditions pHBGE= pK
+ 0.5. The ionic strength of the buffer is typically in the range of 10-200 mM
concentrations of the main buffer constituents. At a given pH/ionic strength com-
bination, several buffers can be used in principle. Preference should be given to
the amphoteric, so-called Good-type buffers, which are available for the pH re-
gion 5.5-1 1 and which fulfill the condition of sufficient buffering capacity at
relatively low electric conductivity [39]. The classical buffers such as phosphate,
citrate, and acetate can be used in moderately low pH regions, and borate and
phosphate buffers can be used in weakly alkaline pH range. Common buffers
utilized for peptide separations by CZE and FFZE are given in Table 2.
Of course, the pH of BGE must be selected not only according to pH depen-
dence of effective and specific charge of peptides to be separated, but also with
respect to their solubility, chemical and thermal lability, and biological activity.
These aspects are discussed in more details in the following sections.
3. Some Factors Influencing Selection of Background

Electrolyte
The optimal pH region and generally the optimal composition of BGE should
fulfill the following conditions:
1. There should be sufficient solubility of analyzed peptides (at least
about 1 mmol/L).
2. There should be chemical stability of peptides including some their
labile groups (amides, disulfides, sugar moieties, etc.).
3. There should be preservation of biological activity (if peptides are used
for biological tests after their CZE or FFZE separation).
4. Effective mobilities of peptides and their relative differences should
be sufficiently high for their separation (approx. mef> l.10-9 m2 V-'
s-I, Amef/mef> 0.02, i.e., the relative difference in mobilities should
be higher than around 2%).
5. Adsorption of peptides to capillary wall is suppressed.
Table 2 Suitable Constituents of Background

Electrolyte for Capillary and Free-Flow Zone
Electrophoresis of Peptides
BGE constituent PK Mobility
Anions:
Phosphoric acid
Citric acid
Formic acid
Glutamic acid
Acetic acid
MESa
ACES a
HEPESa
Tricinea
Boric acid
Alanine
BALAa
CAPSa
Cations:
Gly-Gly
BALAa
EACAa
Creatinine
Histidine
Imidazole
Trisa
Arnmediola
Ethanolamine
WES, 2-(N-morpholino)ethanesulfonic acid; ACES, N-2-acet-
amido-2-aminoethanesulfonic acid; HEPES, N-2-hydroxy-
ethylpiperazine-Nt-2-ethanesulfonic acid; Tricine, [N-(tri-
sh~d&ymethyl)methy~]glycine; BALA, p-alanine; CAPS,
3-(cyclohexylamino)propanesulfonic acid; EACA, €-amino-
caproic acid; Tris, tris(hydroxymethy1)aminomethane; a m m e -
diol, 2-amino-2-methyl-1.3-propanediol.
Some brief rules for the selection of suitable composition and p H of B G E

with respect to these demands are given below.
a. Solubility Peptide solubility can be affected by a number of parameters:
1. pH. T h e pH of B G E should b e at least 1-2 p H units distant from the
isoelectric point since the solubility of peptide is relatively lower at p I
and its close vicinity.
2. Ionic strength and the buffer constituents. The composition of BGE is
relatively free, and it is possible to use relatively concentrated solutions
of acids and bases (e.g., 0. I M phosphoric acid, 0.5 M acetic acid, 0.1
M Tris), which is favorable for peptide solubilization. Higher ionic
strength can contribute to the better solubility of peptides but overly
conductive buffers should be avoided since great Joule heat would be
generated in these buffers at high voltages generally applied in CZE
and because of problems with Joule heat removal from the flow-
through electrophoretic chamber.
3. Solubilizing additives. Different types of additives can be added to the
BGE to improve the peptide solubilization, i.e., chaotropic agents, such
as urea and its derivatives [401. nonionogenic and zwitterionic deter-
gents [4 1-43]. and cyclodextrins and micelle-forming compounds such
as anionic detergent sodium dodecyl sulfate, (SDS) or cationic deter-
gent cetyltrimethylammonium bromide (CTAB) [44]. These agents
should be added to BGE very carefully because they can also change
or reverse the electroosmotic flow, can form complexes or ion pairs
with peptides [45], or can change the separation mechanism from
chargelsize-based separation to hydrophobicity-based separation, e.g.,
by micellar electrokinetic chromatography (MEKC) [46-481. This
technique is suitable for separation of noncharged peptides or peptides
with the same or very similar specific charge, i.e., peptides containing
the same charged amino acid residues and similar noncharged amino
acid residues. Complete separation of porcine, equine, bovine, and
sheep insulins differing in the presence of neutral amino acid residues,
thrconine. alanine, serine, glycine, valinc, and isoleucine in positions
8-10 of A chain of insulins was achieved by capillary MEKC with
SDS and CTAB micellar pseudophase with the addition of acetonitrile
to the BGE 1491. Capillary MEKC with mixed fluorocarbon-hydrocar-
bon anionic surfactants (lithium perfluorooctane sulfonate and lithium
dodecyl suleate) allowed separation of nine small tryptophan-con-
taining charged peptides with nearly identical electrophoretic mobility
[SO]. The above example of insulins separations also demonstrates that,
although mostly performed in watcr buffers, CZE separations of pep-
tides can also be performed in the buffers to which organic solvents
such as acetonitrile, methanol, and other alcohols are added [49,51-
531. The organic solvent may not only improve peptide solubility, but
due to changed selectivity i t can also increase resolution of peptide
separations [54,55].
The solubilizing additives also help to reducc the adsorption of separated

peptides to the inner capillary wall.
b. Biological Activity, Chemical and Thermal Lability Biological activity

(e.g., hormonal or immunochemical function) of peptides is connected with cer-
tain conformations of polypeptide molecules which are stable only under the
defined conditions. Consequently, if the biological activity of peptide should be
preserved, e.g., after preparative separation by FFZE the peptides are used in the
tests of biological activity on animals, then this FFZE separation must be per-
formed under conditions whereby irreversible denaturation of peptide does not
occur, i.e., composition, pH, solvents, and additives of BGE have to be chosen
with respect to this demand. For preserving biological activity and for preventing
the losses of separated peptides it is necessary to prevent adsorption of analyzed
peptides to the inner walls of the separation compartment by dynamic or covalent
inner surface coating [56].
Chemical lability of some groups in the peptide molecule has to be taken
into account, e.g., peptides containing amide, disulfide, and sulfhydryl groups
should not be analyzed at high pH because of hydrolytic and transsulfidation
reactions of these groups at this pH.
In addition, the electric input power should be sufficiently low (lower than
2 WIM of capillary length for capillaries with i.d. 0.05-0.1 mm), and active
removal of Joule heat by flowing air or circulating liquid is recommended in
order to prevent thermal denaturation of separated polypeptides.
D. Detection
1. Optical Detection
Similar to the other classes of compounds, optical detection is the most frequently
used detection mode in CZE separation of peptides. Of the several optical detec-
tion modes used in CZE [57], only (spectro)photometric UV absorption and fluo-
rescence detection will be mentioned here in more detail. Descriptions of the
other modes, used only rarely for peptide detection, say for refractive index and
thermooptical absorbance detection, can be found in the above given review [57].
a. UVAbsorption Detection Thanks to the relatively strong absorption of
peptide bond CO-NH in the short-wavelength UV region, absorption detection
at these wavelengths (200-220 nm) can be used as a universal detection principle.
Special UV absorption detectors have been designed utilizing, for example, high-
frequency excited iodine discharge lamp with emission at 206 nm [58,59] or Zn
or Cd lamps with emission at 214 nm [60], in addition to those utilizing classical
deuterium lamp with continuous light emission in the whole UV region [61].
With the fused silica capillaries detection below 200 nm. up to 185 nm, is also
possible [62]. Due to the higher molar absorption coefficient of peptides at this
region the detection sensitivity is increased, e.g., twofold gain in the signal-to-
noise ratio was obtained by changing the detection wavelength from 2 14 to 200
nm [63], but the choice of BGE is limited to those buffers that do not absorb at
this wavelength. Borate and phosphate buffers are useful in this respect, but many
biological Good's buffers are inappropriate for use below 215 nm. For polypep-
tides separated in the presence of low-UV-absorbing BGE constituents and addi-
tives, 230-nm wavelength provides a good compromise between detectability and
peak-area analysis accuracy on one side and stronger light absorption of the
background electrolyte on the other side.
The UV absorption of the peptide zone in the spectral region 200-220 nm
characterizes the peptide bond quantity, i.e., a longer peptide provides a higher
absorbance signal response at equal molar concentration [64]. Micromolar con-
centrations of peptides are usually detectable by capillary electrophoresis (CE)
with UV absorption detection. Increased sensitivity can be achieved at some spe-
cial designs, as a Z-shaped cell, a bubble cell, or a sleeve cell [61,65,66].
Peptides containing aromatic amino acid residues can be detected more
specifically by UV absorption at 275-280 nm or with a lower sensitivity at 254
nm. The highest absorption coefficient is characteristic for tryptophan residue and
also for tyrosine residue, where significant changes of the absorption coefficient
influenced by pH occur. Less sensitive is the detection of phenylalanine residue.
Although the absorption coefficients and hence the resulting detectability of aro-
matic residue containing peptides at 280 nm is lower than at 206 nm, more free-
dom in selection of BGE composition partially compensates for this drawback.
More information about the quality and quantity of peptides separated and
confirmation of peak homogeneity and identity can be obtained by multiple-
wavelength detection or by full spectrophotometric detection realized by fast
scanning spectrophotometric detection or by diode array detection [61]. Spectral
analysis and the use of spectral libraries and automated library searches can
distinguish closely related peptides, e.g., cyclic and linear forms of synthetic
oligopeptides, and allows identification of individual peaks of peptide maps of
proteins [67].
b. Fluorescence Detection
Laser-Induced Fluorescence Detection of Native Peptides By its inherent
character (measurements are performed with a negligible background signal)
fluorescence detection is more sensitive than absorption detection. In the first
generation of CE devices the fluorescence detectors with classical light sources
(xenon lamps) used in HPLC were adapted for capillary columns [68]. They
suffered from the same drawback as the adapted HPLC UV absorption detectors,
i.e., the amount of light coupled to the analyte zone in the capillary was rather low.
More successful versions of fluorescence detectors use lasers as sources of
excitation light [69-721. Laser-induced fluorescence (LIF) is the most sensitive
detection mode in CZE. With special designs of liquid sheath-flow cuvettes, LIF
detection is approaching the absolute limit-detection of a single molecule [73].
The native fluorescence of peptides excited by UV light in the region 200-

300 nm is exclusively dependent on the presence of aromatic amino acid residues
in the peptide molecule. The optimal excitation wavelength was found at 275
nm [74]. Tryptophan residue obviously plays the dominating role. The quantum
yield of the fluorescence of tryptophan residue is positively influenced by a hydro-
philic microenvironment. The fluorescence intensity of tyrosine is about two or-
ders of magnitude weaker. Almost negligible is the fluorescence of phenylalanine
residue.
In the required UV excitation range 200-300 nm only few lasers can oper-
ate. Frequency-doubled Ar laser operating at 257 nm and Kr laser operating at
284 nm were used for detection and spectral differentiation of peptides containing
tryptophan and tyrosine [75]. The detection concentration limits related to the
above amino acids were 2.10E-10 (Trp) and 2.10E-8 (Tyr). Acquisition of the
fluorescence emission spectrum allowed distinction of three classes of peptides,
those containing Trp or Tyr or both Trp and Tyr.
Pulsed air-cooled KrF UV laser operating at 248 nm was used for the
analysis of tyrosine-and tryptophan-containing dipeptides and proteins [76-781.
The detection limits for these peptides were at least two orders of magnitude
lower when compared with UV absorption at 214 nm.
Availability of the above-mentioned deep UV lasers meant a great progress
for CE analysis of peptides since it makes possible very efficient excitation of
native fluorescence of tryptophan-and tyrosine-containing peptides. This LIF ap-
proach greatly simplifies the preseparation sample handling and is useful espe-
cially for analysis of peptides occurring in biological fluids and tissue extracts
at low concentration levels. High sensitivity and no need for derivatization of
peptides and proteins, which is a problem especially at low concentrations of
analytes, was utilized for quantitative determination of native peptides and pro-
teins in single cells, as, for example, insulin in single pancreatic cells [79] and
hemoglobin in individual human erythrocytes [80].
Laser-Indxced Fluorescence Detection of Labelled Peptides A more general
and more common way of fluorescence detection of analytes in CE is based on
both precolumn and postcolumn labeling of these analytes with a fluorescent
marker [81]. Unfortunately, this approach has a number of problems in detection
of peptides and proteins. Peptides and proteins usually contain more than one
derivatization site, which leads to multiple labeling. More derivatives with differ-
ent electrophoretic mobilities and consequently multiple peaks are obtained for
originally single peptide species. To avoid this problem some special procedures
utilizing Edman degradation chemistry [82] or fluorescein isothiocyanate (FITC)
labeling at lower than normal derivatization buffer pH [83] have been developed,
but some differences in the extent of the precolumn incorporation of the tags
causing a rise of multiple peaks after CZE must be taken into account. Often
precolumn derivatization reagents reacting with amino group were used as, say,
5-dimethy laminonaphthalene- 1-sulfonyl (DANSYL) chloride, naphthalene-2,3-
dicarboxaldehyde (NDA), fluorescein isothiocyanate (FITC) and 3-(4-carboxy-
benzoy1)-2-quinolinocarboxaldehyde (CBQCA) [84]. Subzeptomole detection
limits for CZE of FITC derivatives of peptides were reported [72]. CZE peptide
mapping of 360 attomole of tryptic human serum albumin (HSA) digest involving
selective derivatization of arginine-containing peptides with benzoin was per-
formed by Cobb and Novotny [85]. Selective determination of arginine or tyro-
sine-containing peptides was achieved by using two different amino acid-selec-
tive fluorogenic reagents (benzoin with guanidine moiety for arginine residues
and 4-methoxy-l,2-phenylenediaminereacting with formylated tyrosine resi-
dues) and utilizing He-Cd laser operating at 325 nm for LIF detection [86]. LIF
detection of selectively labeled phosphoserine residue (fluorescein attached to
1,2-ethanedithiol-derivatizedphosphoserine) in peptides and proteins allowed di-
rect quantitative evaluation of peptide and protein phosphorylation at the attomole
level [87]. Trace levels of peptides (derivatized by fluorescamine) from neuronal
and subneuronal samples were analyzed by CE with LIF detection using He-Cd
laser at 354 nm [88]. Detection of attomolar concentrations of alkaline phospha-
tase (corresponding to nine molecules of this enzyme in 1 p1 sample volume) was
achieved by CZE monitoring of the enzymatic conversion of a fluorogenic sub-
strate into the highly fluorescent product using Ar ion laser with excitation wave-
length 458 nm [89]. LIF-CZE was used also for the determination of in vivo
neuropetide release from the ewe median eminence [90,9] ] and for monobromo-
bimane-derivatized glutathione and hemoglobin in single erythrocytes [92,93].
Nonfluorescent regents such as fluorescamine and o-phthalaldehyde (OPA)
may be used both for precolumn and postcolumn derivatization under mild condi-
tions with subsequent chemiluminescent detection [94,95]. Lower resolution of
the CZE zones as a result of postcolumn detection usually occurs.
In spite of the above-mentioned problems LIF detection of labeled peptides
in general seems to be the most sensitive method for peptide detection in CZE.
2. Mass Spectrometric Detection

Mass spectrometry (MS) represents an ideal detection principle for CZE because
of its universality, sensitivity, and selectivity [96-991. On-line coupling of CZE
with MS was made possible thanks to the introduction of two ionization tech-
niques: continuous flow fast-atom bombardment (CF-FAB) [loo] and electro-
spray ionization (ESI) [ l o l l , which solved the problem of how to desolvate,
ionize, and transfer into gas phase and into vacuum conditions required for MS
analysis the analytes separated in liquid phase separation methods such as HPLC
and HPCE. Both of these ionization techniques opened the door of MS to the
analysis of polar, nonvolatile compounds, including peptides and proteins. The
MS detection principle combined with high resolution of CZE is becoming im-
portant for peptide mapping and for separation of peptides, i.e., fragments of
proteins originated from specific enzymatic hydrolysis of proteins [20,85]. These
peptide maps are frequently used for the confirmation of known protein sequences
[I021 and for the investigation of posttranslation modification, e.g., phosphoryla-
tion of proteins [103]. ESI is the preferred mode of coupling CE apparatus with
mass analyzer, namely, because of its capability to generate the ions with multiple
charges so that the ion mlz (masslcharge) values for even very large species
(such as polypeptides and proteins) may fall within the limited mlz detection
range of most mass spectrometers. Equally important is the fact that a population
of multiply charged ions having different numbers of charges is created for these
larger ions so that the mlz data can be mathematically deconvoluted to give the
original molecular masses of the species. CE-ESI-MS is now a widely used tech-
nique for separation and identification of closely related peptides, e.g., neuroten-
sin and angiotensin analogs [104], model mixtures of synthetic peptides
[105,106], and naturally occurring biologically active peptides, e.g., enkephalins,
oxytocin, bradykinin, angiotensin, and bombesin at femtomole to attomole levels
[107]. An example of ESI-MS detection of CZE-separated peptide and protein
mixtures is shown in Fig. 5. Combination of CZE with tandem MS, CZE-ESII
MSIMS [log-1101, allowed determination not only of molecular mass of the
whole molecules of proteins and peptides but also of their fragments. From the
collision-induced production-ion spectra the amino acid sequences of CZE-sepa-
rated peptides and proteins were obtained, which allowed their identification by
comparison of these sequences with those in a protein database [109,110]. For
on-line CZE-MSIMS coupling the best signal-to-noise ratio was obtained when
acidic buffers of low ionic strength for CZE peptide separations were applied
[log]. Extremely large number of components of complex peptide mixtures origi-
nating, for example, from protein hydrolysis (peptide mapping) can be resolved
and identified when the MS detector is coupled to an on-line combination of
HPLC and CZE, and in such a way a three-dimensional separation. HPLC-CZE-
MS, is realized [ I l l ] .
The newest version of MS, matrix-assisted laser desorption-ionization mass
spectrometry with time-of-flight mass analyzer (MALDI-TOF-MS), is combined
with CZE separations of peptides and proteins mostly in an off-line mode
[112,113]. Its advantage, i.e., soft ionization and generation of predominantly
singly charged molecular ions of even polypeptide and protein macromolecules
as antibodies [I 141, is used for exact determination of relative molecular mass
with an accuracy of 20.1% and for identification of peptides and proteins isolated
after their CZE separation. In some experimental arrangements the CZE fractions
are deposited directly on the MALDI probes so that individual peaks from electro-
phoreogram are associated with a single sample spot on the probe [112]. MALDI
matrices with high acid concentrations afford enhanced tolerance of CZE buffers
to be used for introducing peptides to the mass analyzer. Keough et al. [115]
mi. 531 A Bradykinln
1 1 (MW 1060)
mlz 749
1
1 S p e n Whale Myoglobln
(MW 17.199)
mtz 1157
1
1
m l z 808
1
Porcine Insulin
I
Horse Myoglobin
(MW 5 7 7 8 ) (MW 16.950)
(M.5H)"
0 10 20 0 10 20 30 40
Time (mln.1 Time (min.)
Figure 5 Single-ion electrophoreograms and reconstructed ion electrophoreogram

(RIE) for separation of peptide and protein mixture by CZE-ESI-MS. (A) Peptide mixture
consisting of approximately 5 pmollcomponent separated in 0.6 m X 0.1 m m i.d. fused
silica capillary at 15 kV using pH I I sodium phosphate buffer as BGE. (B) Peptide and
protein mixture consisting of approximately 2 pmollcomponent separated in a pH 8.4 Trjs-
HCI-buffered BGE in an untreated 1.1 m X 0.1 mm i.d. fused silica capillary at 25 kV.
(From Ref. 101.)
reported a 250-fmol detection limit for smaller peptides and a 100-fmol off-line
detection limit for a-lactalbumin. The MS detection also enables one to detect
and characterjze absolutely the nonpeptidic part, e.g., the glyco-component,
attached to the peptide chain [ I 161.
3. Electrochemical and Conductivity Detection

Electrochemical (EC) methods offer high selectivity and sensitivity for analytes
that are easily electrooxidized or electroreduced with the detectability comparable
with that of fluorescence detection methods. The relative simplicity and low cost
of EC instrumentation, including the ease with which the small electrodes can
be constructed and the associated small currents measured, make the EC methods
particularly attractive for CZE applications [117-1191. Most of EC detection in
CZE are performed in an amperometric mode. The problem of "decoupling" of
small EC potentials (about 1 V) and currents (in the pA range) from the much
larger driving CE voltage (10-30 kV) and current (tens of FA) was solved by
using porous glass joint or porous membrane covered fracture decoupler or by
end-column application of the working electrode [I 18,1201. The major drawback
is the small number of analytically important species that are electroactive at
modest potentials at the carbon electrodes conventionally used, e.g., catechols,
phenols, aromatic amines, and thiols. This is also the reason why EC techniques
have not been applied to the detection of peptides very frequently up to now.
Thiol-containing peptides and amino acids, e.g., glutathione, cysteine, and cys-
tine, were determined by amperometric detection at copper microelectrode [121]
or at gold/mercury amalgam microelectrode [122]. With the latter the detection
limits for glutathione were about 0.5 fmol or 20 nmol/L. Pulsed amperometric
detection with gold fiber microelectrode was used for characterization of glyco-
peptides from recombinant coagulation factor VIIa separated by CZE [123].
An end-column detection cell that can be used for both electrochemical
and conductivity detection in CZE was designed by Huang et al. [I241 and by
Muller et al. [125]. A few other types of conductivity detectors for CZE have
been developed 1126-1281, but mostly they are not used for direct detection of
peptides in CZE. However, conductivity detection can be recommended for CZE
and CITP determination of small organic and inorganic ions, e.g., anionic coun-
tenons (acetates, trifluoroacetates) of basic peptide preparations [21,129,130].
Ill. ANALYSIS OF PEPTIDE PRODUCTS

A. General
Qualitative and quantitative analysis of synthetic peptides and peptides isolated
from natural material is the most common application of CZE in peptide chemis-
try. In particular, the need for a purity test of synthetic peptides has risen in many
areas in the recent years. In biochemistry synthetic peptides are used as substrates
and inhibitors of enzymes in the elucidation of the mechanism of their catalytic
effect; in the investigation and modeling of the interactions of antigens with anti-
bodies, hormones with receptors, and peptides and proteins with nucleic acids;
in the mapping of antigenic determinants (epitopes) of proteins; and in the study
of the dependence of secondary and tertiary structure of a peptide chain on amino
acid sequence. The use of synthetic peptides in pharmaceutical research and in
human and veterinary medicine is also widespread. In the food industry peptides
are used as sweeteners and additives.
In a majority of these applications of synthetic, natural, or biotechnologi-
cally prepared peptides CZE can be used as a sensitive control method for the
determination of their purity, or as a control method of the efficiency of the other,
mainly chromatographic separation methods used for their purification. CZE pro-
vides rapid and accurate qualitative and quantitative data about the peptide prepa-
rations.
Thanks to the high parameters of CZE separations, efficiency achieving
hundreds of thousands or millions of theoretical plates, sensitivity in the range
of femtomole-attomole of an analyte in the nanoliter-picoliter sample volume,
high-speed of analysis with the average time of analysis in the order of few
minutes, and its compatibility with different detection modes, CZE is capable of
providing not only the data on peptide purity but also on the identity, structural
changes, and physicochemical characteristics of the peptides and proteins they
constitute. CZE is capable of revealing subtle differences between peptides origi-
nating from minimal changes in amino acid sequences of synthetic oligopeptides
of the same amino acid composition [131] or in the disulfide bridge position
[132]. The latter case can be demonstrated by CZE separation of 70-amino-acid
residues containing insulin-like growth factor (IGF-1) (disulfide bridges in posi-
tions 6-48 and 47-52) and recombinat IGF-I byproduct (disulfide bridges in
positions 6-47, 48-52) [133]. Reduction of intrachain disulfide bridge in 27-
amino-acid residues containing peptide was also monitored by CZE [134]. CZE
was capable of separating the deamidation products of human insulin (HI) and
human growth hormone (HGH) from their natural molecules, i.e., polypeptides
differing in single elementary charge unit per 51-amino-acid residues (HI) or
per 191-amino-acid residues (HGH). respectively [135]. CZE was used also for
separation of closely related stereoisomers and isoforms of tryptic heptapeptide
fragment of HGH [136], for enantiomeric and diastereomeric separations of di-,
tri-, and tetrapeptides [137-1391 using cyclodextrins as chiral selectors and for
resolution of cis and trans isomers of proline-containing peptides [140,141].
In addition to analytical applications, CZE can also be used for the determina-
tion of physicochemical characteristics of peptides, as effective mobilities [7.
10,12,142,143], effective charges [16,144], relative molecular masses [16,145],
diffusion coefficients [146], and for monitoring of peptide interactions with other
biomolecules and for determination of association and/or dissociation constants
of peptide complexes with chiral selectors and other ligands [147-1491.
B. Evaluation of Peptide Purity

Peptide purity and peptide content in the sample can be quantified by several ways
depending on whether the standard of given peptide is available or unavailable. If
the standard of a peptide P is available, then the quantity of this peptide in the
analyzed sample can be determined absolutely by the calibration curve method
or by the method of internal standard addition, i.e., by comparison of migration

times, peak heights, and/or peak areas of standard peptide and of the peptide in
the given sample. If the standard is not available, which is the case of peptides
synthesized or isolated for the first time, the relative evaluation of peptide purity
based on the peak height or peak area ratio has to be used.
1. Relative Degree of Purity

The relative CZE degree of purity of peptide P, P ~ , c ~ (E P. ~~, , ~ is~ defined
~ , . ~ )as
the ratio of the peak height (area) of peptide P itself to the sum of heights
(areas) of all (n) peaks present on electrophoreogram of the given preparation
of peptide P:
It must be emphasized that both ways of expressing the CZE purity degree
can be used only as approximate and relative criteria of peptide purity since
the molar absorption coefficients of the individual sample components may be
generally different. However, in the case of synthetic peptide preparations most
of the admixtures will have the similar structure to that of the main synthetic
product and consequently also the similar molar absorption coefficients.
Generally, peak height is less suitable for quantitative analysis than peak
area, since peak height is more dependent on the experimental conditions. Stack-
ing or destacking of the sample zone can occur due to concentrating effect of
the BGE constituents or the sample matrix components [150]. Differences in
mobilities of analytes and BGE constituents cause electromigration dispersion of
analyte zones, which also influences their heights.
The peak height-based CZE purity degree is recommended to be used for
CZE-grams with great difference in migration times and with similar peak shape
and width and when the signal-to-noise ratio is too low for precise integration
of peak area. Because of limited linearity between sample amount (concentration)
and peak height the difference in the analyte amounts (concentrations) should be
within one order.
The advantage of peak area-based quantitative evaluation is that the above-
mentioned peak shape distortions do not affect peak area. However, if peak area
is used for quantitative evaluation, then a correction for different migration veloc-
ities of the analytes has to be taken into account. Due to these differences, analytes
with lower migration velocities (mobilities) remain in the detection window for
a longer time than those with higher migration velocities (mobilities). Conse-
quently, they exhibit broader peaks with a larger peak area than the faster moving
analytes, although the physical length of their zones may be the same as that of
the faster moving analytes. The deviation of peak area caused by this effect can
be corrected by dividing the peak area of an analyte by the migration time of
this analyte 11511.
The migration velocity noncorrected peak area-based CZE purity degree
can be used for evaluation of CZE grams where the peaks of some components
are broader not because of slow migration through the detector window but due
to the adsorption of analytes to the capillary wall or due to electromigration dis-
persion.
A further few applications of CZE to peptide analysis are demonstrated.
All of the presented examples were performed on the home-made CZE device
developed in our laboratory [152]. It consists of the untreated fused silica capil-
lary with outer polyimide coating (i.d. 0.056 mm, 0.d. 0.200 mrn, total length
3 10 mm, effective length 200 mm), UV photometric detector at 206 nm, and a
high-voltage power supply. Peptide samples and electroosmotic flow markers
were dissolved in the BGE in the concentration range 0.1 - 1.5 mglml. The sample
was introduced into the capillary manually forming a hydrostatic pressure (50
mm of water column) for the time period 5-20 s. High-voltage power. supply
was utilized in constant voltage mode. Experiments were performed at ambient
temperature 22-24°C without active cooling of the separation compartment.
Application of CZE to the determination of peptide purity is demonstrated
in Fig. 6. It shows CZE analysis of standard preparation of dalargin, a synthetic
hexapeptide with the sequence H-Tyr-D-Ala-Gly-Phe-Leu-Arg-OH, and a deriv-
ative of dalargin, a synthetic hexapeptide with the sequence of H-Tyr-D-Ala-
Gly-D-Phe-Leu-Arg-NH2.Whereas no admixtures were found in the standard pre-
paration, i.e., its CZE purity degree approaches loo%, in the case of dalargin
derivative in addition to the peak of the main synthetic product two admixtures
with lower mobility are present. Peak height-based purity degree of the dalargin
derivative preparation is 86.1% and peak area-based purity is 89.6%.
CZE analysis of insect oostatic hormone fragment, a synthetic octapeptide
with the sequence H-Tyr-Asp-Pro-Ala-Pro-Pro-Pro-Pro-OH (see Fig. 7), shows
the utilization of CZE as a control technique for evaluation of the suitability of
other separation technique, namely reversed-phase HPLC (RP-HPLC), for pre-
parative separation (purification) of a given peptide preparation. Comparison of
CZE analysis of crude synthetic product of insect oostatic hormone and of the
same product purified by RP-HPLC (see Fig. 7) shows that RP-HPLC was a
very efficient method in the purification procedure; it was capable of removing
completely four admixtures (al, a2, a4, a5) of the five admixtures (al-a5) of the
crude synthetic product and to decrease substantially the content of the admixture
a3, but for the complete purification of the preparation further technique has to
be employed or RP-HPLC has to be repetitively used.
The high separation power of CZE is demonstrated by the separation of
IW1
lW1
I , . . . . . . . . . I . . . . . . . . .
Figure 6 CZE analysis of standard preparation of dalargin, hexapeptide with sequence

H-Tyr-D-Ala-Gly-Phe-Leu-Arg-OH (a) and dalargin derivative, hexapeptide with the se-
quence H-Tyr-D-Ala-Gly-D-Phe-Leu-Arg-NH2 (b). Sample concentration 0.5 mg/ml.
BGE: 0.5 mol/L acetic acid, pH 2.5; capillary: i.d. 0.056 mm, effective length 200 mm,
total length 310 mm; voltage 9.0 kV, current 10.1 PA, temperature ambient 23'C; AU,
absorbance at 206 nm.
human insulin (HI) and its closely related derivatives and fragments, such as HI
with D-phenylalanine in positions 2 4 and 2 5 of B chain, HI with N-phenylacetyl-
protected amino group of lysine in position B29, L-Phe- and D-Phe-B-24,B25-
octapeptide-B23-B30-HI (see Fig. 8).
CZE can b e used not only for qualitative and quantitative microanalysis of
0 20 40 60 80 100 120 la M 180 a00
l i n e Isecl
IW1
0.sssoC , , , , , , , , , , , , , , , , , , , ,d
0 20 40 60 80 100 120 la M 180 ~KI
line lsecl
Figure 7 Monitoring of HPLC purification of insect oostatic hormone, synthetic octa-

peptide with the sequence H-Tyr-Asp-Pro-Ala-Pro-Pro-Pro-Pro-OH by CZE. (a) Crude
synthetic product (sample concentration 1.1 mglml). (b) Product purified by HPLC (sam-
ple concentration 0.6 mglml). BGE: 0.04 M Tris, 0.04 M tricine, pH 8.1, constant voltage
10.0 kV, current 14.5 FA.Capillary: same as in Fig. 6. 1, main synthetic product of the
insect oosatic hormone; al-a5, nonidentified admixtures; AU, absorbance at 206 nm.
peptide preparations but also for monitoring of chemical and enzymatic conver-
sions of peptides. Figure 9 shows CZE monitoring of enzymatic cleavage of pig
insulin (PI) by trypsin 150 pg of pig insulin was dissolved in 200 p1 of the
background electrolyte (10 mM tricine, 5.8 mM morpholine, 10 mM tricine, 20
mM NaCl, adjusted by 0.1 M NaOH to pH 8.0) developed earlier for the separa-
tion of PI and desoctapeptide insulin. Composition and pH of the BGE was cho-
sen on the basis of the calculated pH dependencies of effective, specific, and
Figure 8 CZE separation of human insulin (HI) and its derivatives and fragments. Sam-
ple components dissolved in ,BGE in concentration range 0.5-1.5 mglml. BGE: 10 mM
tricine, 5.8 mM morpholine, 20 mM NaCl, pH 8.6 adj. by 0.1 M NaOH. constant current
25 FA, voltage 8.6 kV. Capillary: same as in Fig. 6. 25, Human insulin (HI); 34, [D-Phel-
B24,B25-HI; 39, [~-Phe]-B24,B25-[Lys-N-PhAc]-B29-HI, 40, [Lys-NPhAcI-B29-HI;35,
[~-Phe]-B24,B25-0ctapeptide-B23-B30-H1; 36, [L-Phel-B24,B25-octapeptide-B23-B30-
HI; NA, nicotinamide (electroosmotic flow marker). A:, absorbance at 206 nm: t, migra-
tion time.
corrected specific charges of insulin and desoctapeptide insulin (see Fig 3 in

Section 1I.B). Analysis of uncleaved PI is shown in Fig. 9a. Solution of PI was
then mixed with 50 pg of trypsin preparation dissolved in 100 p1 of BGE. Nicotin-
arnide (NA) was added to BGE as an electroosmotic flow marker. The reaction
mixture was analyzed on-line by CZE in the given time periods; only a few
nanoliters of the solution was applied to the capillary without any disturbance
of the continuing enzymatic reaction. The reaction was stopped by the addition
of benzamidine to the reaction mixture after 60 min. The analyses of the reaction
mixture after 7,20,40, and 60 min are shown in Fig. 9b-e. The presented electro-
phoreograms indicate a decreasing concentration (amount) of insulin and an in-
creasing concentration (amount) of desoctapeptide insulin in the reaction mixture
in the increasing time. From the CZE analyses of the reaction mixture the kinetics
of the tryptic cleavage of PI can be evaluated.
Only illustrative applications of CZE to analysis of peptide products were
shown here. More examples can be found in several reviews [ l 1,l5,153- 1571,
book chapters [132,158,159], and books [35,160].
D nu
0 . m
D RH
omm
IV. PREPARATIVE SEPARATIONS OF PEPTIDES

A. Micropreparative Peptide Separations by Capillary
The advantages of CZE in the analysis of peptides were demonstrated in the
previous section. In this field CZE is already accepted as a recognized counterpart
and/or complement to what have been the most widely spread separation tech-
niques, i.e., different modes of HPLC. However, in the field of preparative peptide
separations, the application potential of CZE is much smaller. This is the case
for two main reasons:
1. More complicated adaptation of CE systems from analytical to prepara-
tive scale due to the fact that both ends of the capillary are dipped into
the buffer in the electrode compartments and electric field is applied
during the whole time of experiment.
2. Low preparative capacity of capillary systems.
The first problem was overcome by special adaptations of CZE devices
and several procedures for fraction collections from the capillary have been devel-
oped. One of the first approaches was based on the short interruption of an electric
field and during this time the electrode vessel at the outlet end of the capillary
was replaced by a microvial with a small volume (few microliters) of (diluted)
BGE or water. Then the electric field is switched on again and the zone of interest
is let to be eluted by its own electromigration and/or by electroosmotic flow
into this microvial [161]. This procedure, called electroelution, was later used in
commercial CZE instrumentation, where autosamplers are also used as fraction
collectors [162-1641. Sometimes the electroelution is accelerated or completely
replaced by hydrodynamic flow introduced at the inlet end of the capillary
[165,166]. The disadvantage of this approach is that the eluted sample component
is diluted many times in comparison with concentration in the capillary, the elec-
tric field has to be interrupted each time the fraction is collected, and it is difficult
if not impossible to preserve the spatial resolution of closely neighboring sample
zones even when the capillary outlet is moved from one fraction microvial to
another using an automated programmable procedure. Nevertheless, after evapo-
Figure 9 CZE analysis of pig insulin (a) and monitoring of the cleavage of pig insulin
by trypsin after 7, 20, 40, and 60 min (b-e). BGE: 10 mM tricine, 5.8 mM morpholine,
20 mM NaC1, pH 8.0 adj. by 0.1 M NaOH, constant current 20.0 PA, voltage 7.3 kV.
Capillary: same as in Fig. 6. For more details, see the text. PI, pig insulin; DOI, desoctapep-
tide-B23-30-insulin; NA, nicotinamide.
apillary Electrophoresis
PVDF Membrane
3MM Filter
papers
Ground Electrode
MEMBRANE
ASSEMBLY
Figure 10 Schematic diagram of the membrane fraction collection for capillary klectro-
phoresis. (From Ref. 174.)
ration of some of the solvent the collected sample component can be reconcen-
trated to the level detectable by other off-line detection methods. For this post-
column off-line detection and/or characterization of the separated peptides,
mostly different modes of MS detection or amino acid and sequence analysis
are used [167-1701.
For the continuous fraction collection in CE it is necessary to use special
designs of the separation capillary in which the electrical circuit is completed
prior to its outlet. This was achieved by the use of a porous glass joint [171], an
on-column frit [172], or on-column microfractures of the capillary [173]. Then
the sample components can be collected at an electrically isolated exit, while
maintaining the electrical connection between high-potential and grounded elec-
trode.
Another adaptation of the CZE system for micropreparative purposes
consists of connection of the outlet end of the capillary with the membrane
assembly [174], as indicated in Fig. 10. This membrane assembly, consisting of
Immobilon transfer membrane (used for protein blotting) and of two
layers of filter paper serving as electrolyte reservoir and stainless steel ground
electrode, rotates and the sample components eluted by electromigration and by
electroosmotic flow from the outlet end of the capillary are adsorbed to the slowly
rotating membrane assembly. Proteins and peptides caught on the Immobilon
membrane are then detected by staining, e.g., with Coomassie blue, and can be
subjected to further characterization, e.g., Edman degradation in sequence ana-

lyzers.
A special high-precision multicapillary fraction collection for CE has been
developed by Muller et al. [175]. They utilize the approach of coaxial sheath
liquid interface at the outlet end of the capillary [176]. Through this sheath buffer
flow an electric circuit in the separation capillary is completed and simultaneously
this sheath buffer flow transports the sample components leaving the exit of the
capillary to appropriate collection vials. Up to sixty fractions of microliter or
smaller volumes could be automatically collected in capillaries used as collection
vials. The collection capillaries were placed on a cylinder and a computer-con-
trolled stepping motor aligned the appropriate capillary with the column exit.
Precise collection was achieved by fiberoptic UV detection close to (approx. 1
cm) the end of the capillary.
The second problem of preparative applications of CZE, i.e., limited prepar-
ative capacity, is fundamental. Because of small dimensions of capillary separa-
tion compartment (typical i.d. is 0.050-0.1 mm), only small amounts of peptides,
usually less than 1 pg, can be isolated from capillary systems. These quantities
are sufficient only for a limited number of applications, e.g., for further character-
ization of isolated peptides by the above-mentioned sequence and amino acid
analysis, spectroscopic or MS analysis, or for very sensitive enzymatic and immu-
nochemical tests in biochemistry, molecular biology, and medicine [177].
The preparative capacity of capillary systems can be partially enlarged by
increasing the inner diameter of the capillary. In the wider bore capillaries
(greater than 0.2-0.3 mm) the Joule heat is much worse removed from the separa-
tion compartment, which leads to radial temperature gradient and broadening of
sample zones, as well as dramatic loss of separation efficiency. Partial increasing
of preparative capacity can be achieved in rectangular cross-sectional capillary
columns with the dimensions 0.05 X 1 mm [178,179], by microconcentric capil-
lary column [180], or by a relatively higher concentration of peptides to be sepa-
rated [181], which leads, however, to a lower separation resolution. Higher Sam-
ple load can be used if an isotachophoretic concentrating effect is employed for
the separation of peptides occurring at low concentrations [182]. Sometimes re-
petitive fraction collection is necessary to get a sufficient amount of peptide
needed for further characterization. A procedure for optimization of experimental
conditions of preparative CZE was suggested by Cifuentes et al. [183].
B. Preparative Free-Flow Zone Electrophoresis

A principal solution of the problem of enlarging preparative capacity of zone
electrophoresis-based separation of peptides is to perform this separation in a
free-flow electrophoresis mode, i.e., to realize the zone electrophoretic separation
principle in a continuous free-flow arrangement in the flow-through electropho-
Figure 11 Cross-section of the flow-through electrophoretic chamber and the principle
of free-flow zone electrophoresis. A2+,B C , Coyand D are sample components that are
continuously introduced to the injection point of the chamber and collected at the outlet
side of the chamber; E, background (carrier) electrolyte laminary continuously flowing
through the chamber; E', background electrolyte circulating in the electrode compartment
(EC); M, ion exchange membrane between the separation chamber and the electrode com-
partment.
retic chamber [184-1861. In this instrumental format the preparative capacity

can achieve values up to hundreds of milligrams per hour, which is several order
enlargement in comparison with preparative capacity of CZE.
The principle of FFZE [38,187,188] is as follows (see Fig. 11). The separa-
tion compartment is formed by two planparallel glass plates. The background
electrolyte (camer buffer) is continuously laminary flowing in the narrow gap
(0.5 mm) between these two plates. Sample solution is also continuously intro-
duced into the background electrolyte as a narrow zone. Electrode compartments,
separated by semipermeable membranes from the separation part of the chamber,
are situated on both sides of the chamber. BGE is turbulently circulating through
the electrode compartment. Electric field is applied perpendicularly to the direc-
tion of laminar flow of background electrolyte and sample, and causes the differ-
ent deflection of the sample components movement depending on their effective
mobilities. Cations are deflected to cathode, anions to anode, and noncharged
components will move in the straight direction if no electroosmotic flow occurs
in the chamber. At the outlet side of the chamber the separated sample compo-
nents are collected in the fraction collector.
The advantage of FFZE is that it works continuously, in a free solution
under mild conditions in which biological activity of separated substances is pre-
served. FFZE can be applied to preparative separation of a wide range of both
low molecular mass and high molecular mass species and also for the separation
of cells, organelles, and other bioparticles [187,189- 1921.
Because of some problems with thermoconvection and sedimentation in
FFZE processes on the earth, some attempts to perform FFZE experiments in
microgravity conditions inside orbiting space craft have been performed [193-
1951. Successful separation of macromolecules (DNA, proteins) and bioparticles
was achieved in these experiments, but the improvement of the separation effi-
ciency and capacity was less than originally expected.
Miniaturized FFZE device integrated onto a silicon chip has been devel-
oped for continuous microanalysis of biomolecules [196].
C. Correlation of Capillary and Free-Flow

CZE and FFZE represent two modes of the same separation principle. Both of
them are performed in the carrierless medium with the same background electro-
lyte. Consequently, a direct correlation exists between these two methods, which
can be utilized for conversion of capillary microscale separation into preparative
ones [152,197]. The bases of this correlation are described below.
The diagram of the electrophoretic, electroosmotic, and hydrodynamic
movements in CZE and FFZE and the vector sum of these migration velocities
in the capillary and in the flow-through electrophoretic chamber are shown in
Fig. 12.
The resulting migration velocity of a charged component in the dc electric
field in the capillary, v,, is given by the sum of electrophoretic velocity, v,,, and
electroosmotic flow velocity, v,, (see Fig. 12a):
The velocities in Eq. (15) can be expressed as the ratio of effective length of the
capillary, l,, and the corresponding migration times, resulting migration time, t,,
electrophoretic migration time, t,,, and electroosmotic migration time, t,,, respec-
tively:
'ef
-
Jr
Veo Vep
CHAMBER
I \ \
\
1 1 \
1 1 \
I \ \
I \ \
\
I \
Figure 12 Superposition of the migration velocities in capillary zone electrophoresis (a)

and in free-flow zone electrophoresis (b). v,,. Electrophoretic velocity; v,,, electroosmotic
velocity; v,, resulting migration velocity; I,,. effective length of the capillary; D, detection
position on the capillary; vl, hydrodynamic flow velocity; d,,, electrophoretic migration
distance; d,,, electroosmotic migration distance; d,, resulting migration distance. (From
Ref. 201 .)
From the combination of Eqs. (15) and (16), the following relation can be ob-
tained for the electrophoretic velocity in the capillary:
Equation (17) allows one to obtain the electrophoretic velocity from the
data experimentally available from CZE analysis: t, is the resulting migration
time of charged analyte, which is moved both by electrophoretic movement and

by electroosmotic flow, t,, is the migration time of electroosmotic flow marker
(electroneutral compound at given experimental condition), and lCfis the effective
length of the capillary (from the injection end to the detector).
The velocity of the electroosrnotic flow (EOF) in the capillary, v,,,,, can
be calculated from the effective length of the capillary, lefrand the migration time
of the EOF marker, t,,:
The resulting migration velocity of the charged component in the flow-through

electrophoretic chamber is given similarly as in the capillary by the sum of elec-
trophoretic velocity, v,,, and EOF velocity, v,,, and in addition to it also by the
vector sum of these two velocities with the velocity of hydrodynamic flow, vn
which is perpendicular to the direction of v,, and v,, (see Fig. 12b). The resulting
migration distance in FFZE can be expressed as a sum of the electrophoretically
migrated distance, d,,, and the electroosmotically moved distance, d,,:
The electrophoretically migrated distance, d,,, is given by the product of electro-

phoretic velocity in the flow-through chamber, v , , ~and
, the mean flow-through
time of the BGE in the chamber, t,.
The electroosmotically migrated distance is given by the product of EOF velocity

in the flow-through chamber, v,,,~,and mean flow-through time of BGE, t,:
From Eqs. (19) and (20) the following relation for electrophoretic velocity in the
chamber, v , , ~ can
, be derived:
Using relation (22) the electrophoretic velocity of the charged analyte in the
chamber, v,,,, can be calculated from the experimentally available data, namely,
from the resulting migration distance of the charged analyte. d,, from the mi-
gration distance of the EOF marker, d,,, and from the flow-through time of
BGE, tf.
The EOF velocity in the chamber, v,,,~,can be calculated from the migration
distance of EOF marker, d,,, and from the flow-through time of BGE, t,.
For the description of the correlation between CZE and FFZE it is advantageous
to express the ratio of electrophoretic velocities in the chamber and in the capil-
lary, qe,, and the ratio of EOF velocities in the chamber and in the capillary, q,,.
From Eqs. (17) and (22) the ratio q,, can be expressed as
and from Eqs. (18) and (23) the ratio q,, can be expressed as
Provided that the adsorption of the sample components to the walls of the
separation compartments (both capillary and flow-through chamber) can be ne-
glected it is reasonable to assume that q,, is approximately constant for different
charged components separated by CZE and FFZE under the same separation
conditions. This is a realistic assumption because it means that if the electropho-
retic velocity of component A in FFZE is q times higher than in CZE, then the
electrophoretic velocity of component B, analyzed under the same conditions as
A, will be also q times higher in FFZE than in CZE. Consequently, as follows
from Eq. (24), coefficient q,, (determined for standard component S) can be used
to predict the electrophoretic velocities of sample components (A, B, C) in FFZE,
if their electrophoretic velocities in CZE were determined under the same condi-
tions as the electrophoretic velocity of standard S. A similar conclusion can be
applied for the ratio of EOF velocities in CZE and FFZE, i.e., this ratio can be
considered as a constant if the conditions of CZE and FFZE are the same as they
were in the experiment when EOF velocity was determined.
Knowing the values of ratios, qepand q,, allows us to predict the migration
velocities and migration distances of analytes in FFZE from the data obtained
by their CZE analysis. This fact alone is the core of the procedure for conversion
of analytical CZE separations to preparative FFZE separations.
D. Conversion of Analytical Capillary Electrophoretic

Separations into Preparative Free-Flow
Electrophoretic Processes
Based on the above given relations and assumptions, a procedure has been devel-
oped for the conversion of analytical microscale CZE separations to preparative
continuous separation processes realized by FFZE. The procedure consists of the
following steps:
(1) Let us have a sample containing charged analyte A (analyte of our

interest from both analytical and preparative point of view) and some charged
admixtures (Al, A2, . . . ) and noncharged component(s) N (EOF marker). First,
suitable conditions for CZE analysis of the given sample of analyte A have to
be developed under which a good separation of the analyte A from the charged
admixtures A l , A2 and noncharged component(s) N is achieved. From this exper-
iment the electrophoretic velocity of the analyte A in the capillary, v,,,,, is calcu-
lated according to Eq. (17), where t, = t,,Aand t,, = tN;t,., and t Nare the resulting
migration times of analyte A and EOF marker N, respectively.
(2) CZE separation of standard component(s) S (S 1, S2, . . . ) and of EOF
marker N is performed under the same condition as CZE analysis of the sample
of analyte A and electrophoretic velocity of the standard components S in the
capillary, v,,~,,, is calculated according to Eq. (17), where t, = t,,, and t,, = tN;
tIXsand tNare the resulting migration times of components S and N, respectively.
From the migration time of EOF marker N, tN,the EOF velocity in the capillary,
v,,~, is calculated according to Eq. (18), where t, = tN.
(3) Standard component(s) S (Sl, S2, . . . ) and EOF marker N are sepa-
rated in the "standard" (empirically developed) FFZE regimen with the same
BGE as that used in CZE. From this experiment the electrophoretic velocity of
standard component S in the flow-through chamber, v,,,~,,,is obtained according
to Eq. (22), where d, = d,.s and deo = dN(d,, and dN are resulting migration
distances of components S and N, respectively) and EOF velocity in the chamber,
v,,,, is calculated according to Eq. (23), where deo= dN.
(4) From the results obtained in steps 2 and 3, the ratio, q,,, of electropho-
retic velocities of standard component S in the flow-through chamber and in the
capillary is determined:
Similarly, the ratio of EOF velocities in the chamber and in the capillary, q,,, is
obtained:
(5) From the electrophoretic velocity of analyte A (obtained in step 1)

and from the coefficients qep and q,, (obtained in step 4), the electrophoretic
velocity of analyte A in FFZE, v,,,~,,, and EOF velocity in FFZE chamber, v,,~,
are calculated:
Then the predicted resulting migration distance of analyte A in the FFZE cham-
ber, dr,A,pre,
can be obtained as the sum of electrophoretically moved distance,
dep,A,and electroosmotically moved distance, d,,:
(6) The resulting migration distances can also be calculated for the other
components of the sample (admixtures A l , A2, . . . and neutral component(s)
N) and their separability in FFZE can be estimated. If the distances of the compo-
nents of interest at the outlet side of the chamber are sufficient for their separation,
then the FFZE separation can be performed under the same conditions as those
used for separation of standard components.
If the predicted distances are not sufficient for the separation of sample
components of interest, then the separation conditions of FFZE, namely, clamp
voltage and/or flow-through time, have to be further optimized. If the predicted
distances are too small and the separation of sample components is not achieved,
then the voltage and/or flow-through time should be increased. If the predicted
distances for the sample components are too long and there is a danger ,that the
fastest component will reach the close vicinity of the ion exchange membrane
separating the separation chamber from the electrode compartment (see Fig. 1 I),
where this component can be damaged or lost because of concentration, pH, and
conductivity nonhomogeneities occurring in this region, then the clamp voltage
and/or flow-through time must be decreased. The migrated distance is approxi-
mately directly proportional to the voltage and to the flow-through time, i.e., p%
prolongation of migration time and r% increasing of voltage will result in (p +
r)% prolongation of migrated distance. Following this rule suitable separation
conditions can be selected.
The above-described procedure allows one to develop suitable separation
conditions in the more economical and faster microscale by CZE. Only then
can the optimized conditions be converted to the preparative scale realized by
continuous FFZE separation.
E. Combined Application of Capillary and Free-Flow Zone

Electrophoresis to Peptide Analysis and Preparation
In this section some examples of combined CZE and FFZE applications to analyt-
ical and preparative separations of synthetic biopeptides will be demonstrated.
CZE was performed on the experimental device developed in our institute
[152], described in more detail in Section 1II.B above. Acetic acid (0.5 mol/L,
pH 2.6) was used as the BGE. Peptide samples and EOF marker (phenol) were
dissolved in this BGE in the concentration range 0.1-0.8 mglml. The sample
was introduced into the capillary manually forming a hydrostatic pressure (50
mm H 2 0 column) for the time period 5-20 s. Experiments were performed at
ambient temperature 22-24°C without active cooling of the separation compart-

ment.
FFZE experiments were performed in the apparatus developed in our insti-
tute [184]. The core of this system is a flow-through electrophoretic chamber
consisting of two planparallel glass plates (500 X 500 mm) with a 0.5-mm gap
between them. The background electrolyte is introduced through six inlets by a
six-piston pump with a flow-through time of 31 min. Sample solution was intro-
duced by a peristaltic pump with a flow rate of 1.5 mllh. The effective length
of the separation trajectory (from the sample inlet to the chamber outlet) was
440 mm. Both sides of the chamber were cooled by the air to -3°C. The separa-
tions were performed in the constant voltage regimen (3000 V, 122- 125 mA).
At the outlet side of the chamber the carrier electrolyte and sample components
were collected in 48 fractions and periodically sucked into the fraction collector.
The fractions were evaluated by off-line UV absorption measurement at 230 or
280 nm.
The developed procedure for the conversion of analytical CZE separation
to preparative FFZE process will be demonstrated by analysis and preparation
of [D-Tle2%S]dalargin, synthetic hexapeptide with the sequence H-Tyr-o-Tle-Gly-
Phe-D-Tle-Arg-OH. ~ - T l eindicates amino acid residue of tertiary leucine in the
D configuration, the other amino acid residues are in L configuration. [D-
Tle2,']Dalargin is an analog of dalargin, a hexapeptide with the sequence H-Tyr-
D-Ala-Gly-Phe-Leu-Arg-OH, an enkephalin-type peptide with opiate activity.
CZE analysis of the crude synthetic product of [D-Tle2,5]dalargin(see Fig.
13) shows that in addition to the main synthetic dalargin product (peak D), there
are some other admixtures, apparently side reaction products of the solid phase
synthesis. For the better orientation in the CZE-gram some of the admixtures are
letter-indicated, the fastest one by F, the slowest charged admixture by S, and
the noncharged admixture by N. Since the main dalargin product D is relatively
well separated from the other admixtures by CZE, it was decided to use FFZE
for preparative purification of the main synthetic product.
However, from the above given theory of the correlation of CZE and FFZE
it follow? that for conversion of analytical CZE separation into a preparative one
realized by FFZE it is necessary to have not only the data from CZE analysis of
the given sample, but also the ratios, i.e., q,,, q,, of electrophoretic and electroos-
motic velocities in the FFZE chamber and in the capillary, respectively. The
values of these coefficients were obtained from the separation of the standard
mixture of charged components diglycine and triglycine and a noncharged com-
ponent phenol by CZE and FFZE using the same background electrolyte, 0.5
mol1L acetic acid, in which a good separation of main dalargin product was
achieved by CZE. The record of CZE separation of the standard mixture has
already been presented in Fig. 4; the record of FFZE separation of this mixture
is given in Fig. 14. FFZE separation of diglycine, triglycine, and phenol was
Figure 13 CZE analysis of crude synthetic product of [~-Tle',~] dalargin. Sample con-
centration 1.5 mglml. BGE: 0.5 rnol1L acetic acid. Capillary: i.d. 0.056 mrn, effective
length 200 mm, total length 310 mm. Voltage 9.0 kV, current 10.1 yA. F, fastest,compo-
nent; D, main synthetic product-[D-Tle2,5]dalargin; S, slow component; N, noncharged
component. A206,absorbance at 206 nm; t, time.
evaluated by CZE analysis of FFZE fractions. Although the FFZE separation of

diglycine and triglycine was only partial due to the lower separation power of
FFZE than that of CZE, it was possible to use the experimental data obtained
from this separation (resulting migration distances) for calculation of ratios q,,
and q,,. The data of CZE and FFZE separations of the standard mixture and the
calculated values of electrophoretic and electroosmotic velocities according to
Eqs. (17) and (18) for CZE and according to Eqs. (22) and (23) for FFZE and
the values of coefficients q,,, q,, calculated according to Eqs. (24) and (25), re-
spectively, are presented in Table 3. From the migration times of the charged
components F, D, S and from the migration time of electroneutral component N
(see Fig. 13 and CZE columns in Table 4) the electrophoretic and electroosmotic
velocities of these sample components in the capillary, v,,, and v,,,, were calcu-
lated according to Eqs. (17) and (18), respectively (see the CZE columns in Table
4). From these velocities of the sample components in the capillary and using
the values of coefficients q,,, q,, (see Table 3), the electrophoretic velocities, v , , ~ ,
and the predicted migration distances, d,,,, of the sample components F, D, S,
and N in FFZE chamber were calculated (see FFZE columns in Table 4). These
predicted migration distances (15-221 mm) and the differences of these distances
between individual sample components at the outlet side of the chamber indicated
that a sufficient separation will be achieved in the same FFZE regimen that was
used for the separation of standard mixture.
For this reason, the conditions of FFZE, under which the standard mixture
fraction number
Figure 14 FFZE separation of standard mixture of diglycine (15 mglml), peak 1, tri-
glycine (15 mglml), peak 2, and phenol (5 mglml), peak 3. BGE: 0.5 mol1L acetic acid,
flow-through time 31 min, sample flow rate 1.5 mllh, voltage 3000 V, current 120 mA,
temperature -3°C; Peak height = peak height of the analytes obtained by CZE analyses
of the aliquots of the FFZE fractions; 1',sample inlet position.
of diglycine, triglycine, and phenol was separated (clamp voltage 3000 V, flow-
through time 31 min), were applied also to the separation of the components of
the crude product of [~-Tle~.~]dalargin. The lyophilyzate of the crude synthetic
product (190 mg) was dissolved in 5 ml of BGE (0.5 mol/L acetic acid), centri-
fuged, and applied to FFZE separation. The record of FFZE separation (off-line
W absorption measurement of FFZE fractions) is shown in Fig. 15. Comparing
Fig. 13 and 15, the ''qualitative" similarity of CZE and FFZE separation profiles
can be observed. In addition to the main dalargin product (peak D), the fastest
component F, slowest component S, and noncharged component N can be found
on both records. The differences in relative peak heights are caused by the differ-
ent detection wavelengths in CZE (206 nm) and in FFZE (280 nm). Obviously,
a better separation of sample components is achieved in CZE than in FFZE. This
Table 3 Migration Times, Migration Distances, Electrophoretic and Electroosmotic
Velocities of Standard Components Separated by CZE and FFZE
CZE FFZE FFZEICZE
Standard tr Vep,~ V,O,C dr V~p.f Veo,f

component (s) (mmls) (mmls) (mm) (mmls) (mm/s) q,,, 4 eo
Diglycine 168 0.805 - 145 0.0699 - 0.087 -
Triglycine 184 0.707 - 135 0.0645 - 0.091 -

Phenol 510" 0 0.396 15" 0 0.0081 - 0.0205
-
;' Resulting migration time, t, (migration distance, d,) of phenol is equal to migration time, t,,, (migra-
tion distance, d,,,) of electroneutral EOF marker.
t,, resulting migration time; v,,,,, electrophoretic velocity in CZE; v,,,,, electroosmotic velocity in
CZE; d,, resulting migration distance; v,,,~, electrophoretic velocity in FFZE; v,,,,, electroosmotic
velocity in FFZE; q,, (q,,), ratio of electrophoretic (electroosmotic) velocities in FFZE and CZE.
is quite understandable when taking into account the differences in the experi-
mental conditions of these two methods: more efficient anticonvective stabiliza-
tion and Joule heat transfer in the capillary (i.d. 0.056 mm, wall thickness 0.07
mm) than in the flow-through chamber (rectangular gap 0.5 mm between two
glass plates of the thickness 4 mm), about one order lower separation time and
sample concent~ationin CZE than in FFZE, absence of hydrodynamic flow in
CZE, and a relatively large width of the collected f~action(10.4 mm) in FFZE.
Table 4 Migration Times, Electrophoretic Velocities, and Predicted and

Experimental Migration Distances of Some Components of CZE and FFZE Separation
of the Crude Product of [~-Tle~,']dalargin
CZE FFZE
tr vcp,c vat Vep~ ~ 6 0 . i dr.pre dr,cxp

Samplecomponent (s) (mmls) (mmls) (mm/s) (mmls) (mm) (mm)
F (fast comp.) 123 1.234 - 0.1111 - 221 235

D (dalargin comp.) 175 0.747 - 0.0665 - 139 155
S (slow comp.) 310 0.250 - 0.022 - 56 55
N (noncharged) 510" 0 0.396 0 0.0110 15 15
;'Resulting migration time, t,, of electroneutral component N is equal to migration time, t,,, of EOF
marker.
t,, resulting migration time; v,,,, electrophoretic velocity in CZE; v,,, electroosmotic velocity in
CZE; v,,,, electrophoretic velocity in FFZE; v , , ~ electrophoretic
, velocity in FFZE; d,,,,,, predicted
migration distance for FFZE; d,.,,,, experimental migration distance in FFZE.
fraction number
Figure 15 FFZE separation of crude synthetic product of [D-Tle2.5]dalargin.Sample

concentration 38 mglml. BGE: 0.5 molll, acetic acid, flow-through time 31 min, sample
flow rate 1.5 mllh, voltage 3000 V, current 120 mA, temperature -3°C. F, fastest compo-
nent; D, main synthetic product-[~-Tle2~5]dalargin;S, slow component; N, noncharged
component; T, sample inlet position.
The UV record of FFZE separation in Fig. 15 indicates that a good separa-

tion of the main dalargin product, peak D, from the component with higher mobil-
ity was achieved, but the separation from the component with lower mobility
was not quite complete. CZE analysis of individual fractions of peak D showed
that four of these fractions 124-271 contained pure dalargin, as demonstrated by
a single peak of CZE analysis of fraction 26 in Fig. 16a. On the other hand, CZE
analyses of some other fractions, as, for example, that of fraction 36 (see Fig.
16b), showed that some sample components remained unresolved after FFZE.
Due to the lower separation power of FFZE in comparison with CZE, it would
be unrealistic to expect the same degree of separation of all sample components,
but it is important that it is possible to develop such conditions of FFZE under
which the product of interest is well separated from the admixtures, and it is less
important that some byproducts are not completely separated.
The high-purity degree of the main synthetic product was confirmed also
by other methods, particularly HPLC and amino acid analysis. The advantage of
utilization of acetic acid as BGE is that peptide is obtained in acetate, i.e., a
physiologically tolerable form, that can be directly applied to biological tests.
Figure 16 CZE analysis of fraction 26 (a) and 36 (b) of FFZE separation of [~-Tle'.~]da-
largin preparation presented in Fig. 15. Aliquot of FFZE fraction was directly applied to
CZE analysis. Experimental conditions the same as in Fig. 13. A,,,, absorption at 206 nm;
t, migration time
Comparison of the experimentally determined migration distances of se-

lected components with their predicted migration distances (see Fig. 15 and FFZE
columns in Table 4) shows a good predictive power of the theoretical model of
the correlation between CZE and FFZE. The relatively small discrepancy be-
tween the predicted and experimental distance of the fast sample component F
(approx. 6%)confirms quantitative correlation between CZE and FFZE. The rela-
tively larger discrepancy (approx. 10%) between the predicted and experimental
migration distance of the main dalargin product is probably caused by the differ-
ences in partial adsorption of this sample component to the fused silica capillary
wall and to the glass wall of the flow-through electrophoretic chamber andlor
by more than one order higher peptide concentration in FFZE than in CZE.
Another example of the application of CZE and FFZE to synthetic peptide
analysis and preparation is shown in Fig. 17. Disulfide fragment of human cathep-
sin D, eicosapeptide with the following structure (indicated as VI),
was synthesized by the solid phase method and purified by RP-HPLC 11981. CZE
analysis of the HPLC-purified preparation revealed three admixtures of the main
synthetic product (see Fig. 17a). The admixture indicated in this figure as I was
later on identified as Cys-acetamidomethyl-derivative of octapeptide monomer
of the dimer VI. In HPLC this octapeptide derivative was co-eluted with the
heterodimer VI but in CZE these two peptide species were well separated. For
that reason it was decided to use preparative FFZE for purification of heterodimer
VI. The CZE analysis of FFZE-purified preparation of heterodimer VI confirmed
the high purity of this preparation (see Fig. 17b) and the suitability of FFZE for
separation of the admixtures not removed by HPLC. This is a good example of
complementarity of CZE (FFZE) and HPLC for peptide analysis and preparation
resulting from different separation principles of these methods.
A combination of CZE and FFZE has been applied to analysis and prepara-
tion of some other biopeptides synthesized at our institute, such as growth
hormone-releasing peptide (GHRP) and its analogs and fragments 1152,1991,
derivatives of luteinizing hormone-releasing hormone (LHRH) [200], and B23-
30-octapeptide fragments of insulin [201].
Combined application of CZE and FFZE represents a new systematic ap-
proach to analysis and preparation of synthetic biopeptides. First, CZE is used
for analysis of the peptide preparation and suitable conditions for its analysis
are developed in a microscale minimizing time and material expenses for the
development of the optimized separation conditions. Then, based on the theory
of the correlation between CZE and FFZE, the optimized conditions of CZE
separation are converted to FFZE separation. FFZE separation of peptide prepara-
Figure 17 CZE analysis of eicosapeptide disulfide fragment of human cathepsin D prep-
aration purified by HPLC (a) and by FFZE (b). Experimental conditions the same as in
Fig. 13. VI, main synthetic product-eicosapeptide disulfide fragment of human cathepsin
D. I, octapeptide fragment of eicosapeptide VI. For peptide sequences and other details,
see the text.
tion provides an efficient tool for preparative purification of peptide sample with
the preparative capacity of tens to hundreds of milligrams per hour. The advan-
tage of this method is that it works continuously, in a carrierless medium, under
mild conditions in which the biological activity of separated peptides is retained
and the loss of material minimized. The purity of peptide fractions separated by
FFZE can then be checked by CZE and/or another method. A combination of
more methods based on different separation principles should be preferred be-
cause that way more complete information about peptide purity and identity can
be obtained.
ACKNOWLEDGMENTS
The results presented in this chapter were obtained with the financial support of
the Grant Agency of the Academy of Sciences of the Czech Republic, grant no.
4551 1, and of the Grant Agency of the Czech Republic, grant nos. 20319310718,
20319410698, and 203196lK128, which is greatfully acknowledged. The author
thanks his coworkers from the Laboratory of Electromigration Methods of the
Institute of Organic Chemistry and Biochemistry-Dr. Z.Pmsik, Dr. P. S5zelovi,
Mrs. V. LiBkov6, Mr. J. &gp6nek, and Mgr. M. Machov6-for their cooperation
and assistance in the experimental work and for their help in the preparation of
this chapter.
ABBREVIATIONS
BGE background electrolyte

CBQCA 3-(4-carboxybenzoy1)-2-quinolinocarboxaldehyde
CF-FAB continuous flow fast-atom bombardment
CE capillary electrophoresis
CITP capillary isotachophoresis
CTAB cetyltrimethylammonium bromide
CZE capillary zone electrophoresis
LIANSYL 5-dimethylaminonaphthalene-I-sulfonyl
EC electrochemical
EOF electroosmotic flow
ESI electrospray ionization
FFZE free-flow zone electrophoresis
FITC fluorescein isothiocyanate
GHRP growth hormone-releasing peptide
HGH human growth hormone
HI human insulin
HPCE high-performance capillary electrophoresis
HPLC high-performance liquid chromatography
HSA human serum albumin
IEF isoelectric focusing
IGF insulin-like growth factor
LHRH luteinizing hormone-releasing hormone
LIF laser-induced fluorescence
MALDI matrix-assisted laser desorption ionozation
MEKC micellar electrokinetic chromatography
MS mass spectrometry
NA nicotinamide
NDA naphthalene-2,3-dicarboxaldehyde
PI pig insulin
RP-HPLC reversed-phase high-performance liquid chromatography
SDS sodium dodecyl sulfate
Tris tris(hydroxymethy1)aminomethane
Tricine N-[tris-hydroxymethyl]methylglycine
UV ultraviolet
REFERENCES
Vacik J. Theory of electromigration processes. In: Deyl Z, ed. Electrophoresis: A

Survey of Techniques and Applications. Pt. A: Techniques. Amsterdam: Elsevier,
1979:l-21.
Vacik J. Evaluation of the results of electrophoretic separations. In: Deyl Z, ed.
Electrophoresis: A Survey of Techniques and Applications. Pt. A: Techniques. Am-
sterdam: Elsevier, 1979:39-43.
Mosher RA, Saville DA, Thormann W. The Dynamics of Electrophoresis. Wein-
heim: VCH, 1992.
Offord RE. Electrophoretic mobilities of peptides on paper and their use in the
determination of amide groups. Nature 1966;211:591-593.
Rickard EC, Strohl MM, Nielsen RG. Correlation of electrophoretic mobilities from
capillary electrophoresis with physicochemical properties of proteins and peptides.
Anal Biochem 1991;197:197-207.
Compton BJ. Electrophoretic mobility modeling of proteins in free zone capillary
electrophoresis and its application to monoclonal antibody microheterogeneity
analysis. J Chromatogr 1991;559:357-366.
Grossman PD, Colburn JC, Lauer HH. A semiempirical model for the electropho-
retic mobilities of peptides in free-solution capillary electrophoresis. Anal Biochem
1989;179:28-33.
Cifuentes A, Poppe H. Simulation and optimization of peptide separation by capil-
lary electrophoresis. J Chromatogr A 1994;680:32 1-340.
Hiker VJ, Worosila GD, Rudnick SE. Protein and peptide mobility in capillary
zone electrophoresis. A comparison of existing models and further analysis. J Chro-
matogr 1993;630:329-336.
Basak SK, Ladisch MR. Correlation of electrophoretic mobilities of proteins
and peptides with their physicochemical properties. Anal Biochem 1995;226:
51-58.
Messana 1, Rossetti DV, Cassiano L, Misiti F, Giardina B, Castagnola M.
Peptide analysis by capillary (zone) electrophoresis. J Chromatogr B 1997;699:
149-171.
Survay MA, Goodall DM, Wren SAC, Rowe RC. Self-consistent framework for
standardising mobilities in free solution capillary electrophoresis: applications to
oligoglycines and oligoalanines. J Chromatogr A 1996;741:99-113.
Micinski S, Gronvald M, Compton BJ. Structure-mobility relationships in free solu-
tion zone electrophoresis. Meth Enzymol 1996;270:342-358.
Adamson NJ, Reynolds EC. Rules relating electrophoretic mobility, charge and
molecular size of peptides and proteins. J Chromatogr B 1997;699:133-147.
Cifuentes A, Poppe H. Behavior of peptides in capillary electrophoresis: effect of
peptide charge, mass and structure. Electrophoresis 1997;18:2362-2376.
Gao JM, Whitesides GM. Using protein charge ladders to estimate the effective
charges and molecular weights of proteins in solution. Anal Chem 1997;69:575-
580.
Chae KS, Lenhoff AM. Computation of the electrophoretic mobility of proteins.
Biophys J 1995;68:ll20-1127.
Stauff J, Jaenicke R. Physikalische chemie der losungen. In: Rauen HM, ed. Bio-
chemisches Taschenbuch. Berlin: Springer-Verlag, 1964:37- 121.
Prusik Z. Peptides and structural analysis of proteins. In: Deyl Z, ed. Electrophore-
sis: a survey of techniques and applications. Pt. B: Applications. Amsterdam: Elsev-
ier, 1983:81-107.
Rickard EC, Towns JK. Applications of capillary zone electrophoresis to peptide
mapping. Meth Enzymol 1996;271 :237-264.
KaSiEka V, Prusik Z. Isotachophoretic analysis of peptides. Selection of electrolyte
systems and determination of purity. J Chromatogr 1989;470:209-22 1.
KaSiEka V. Analysis and preparation of peptides and proteins by isotachophoresis.
PhD thesis (in Czech). Prague: Czechoslovak Academy of Science, 1985.
Mosher RA, Gebauer P, Thormann W. Computer simulation and experimental vali-
dation of the electrophoretic behavior of proteins 3. Use of titration data predicted
by the protein's amino acid composition. J Chromatogr l993;638: 15.5- 164.
Westermeier R. Electrophoresis in Practice: A Guide to Methods and Applications
of DNA and Protein Separations. Weinheim: VCH, 1997.
Nath S, Schutte H, Hustedt H, Deckwer WD. Correlation of migration behavior in
free-flow zone electrophoresis and electrophoretic titration curve. Electrophoresis
1990;11:612-616.
Hirokawa T, Kiso Y, GaS B, Zuskova I, Vacik J. Simulated quantitative and qualita-
tive isotachophoretic indices of 73 amino acids and peptides in the pH range 6.4-
10. J Chromatogr 1993;628:283-308.
Wehr T, Zhu M, Rodriguez-Diaz R. Capillary isoelectric focusing. Meth Enzymol
1996;270:358-374.
Righetti PG, Gelfi C, Chiari M. Isoelectric focusing in capillaries. In: Righetti PC,
ed. Capillary Electrophoresis in Analytical Biotechnology. Boca Raton: CRC Press,
1996:509-539.
Righetti PG, Gelfi C. Isoelectric focusing in capillaries and slab gels: a comparison.
J Cap Elec 1994;1:27-35.
Yao YJ, Khoo KS, Chung MCM, Li SFY. Determination of isoelectric points of
acidic and basic proteins by capillary electrophroesis. Chromatogr A 1994;680:
431-435.
Sillero A, Ribeiro JM. Isoelectric points of proteins: theoretical determination. Anal
Biochem 1989;179:319-325.
Henriksson G, Englund AN, Johansson G, Lundahl P. Calculation of the isoelectric
points of native proteins with spreading of pKa values. Electrophoresis 1995;16:
1377-1380.
Grossman PD. Background concepts. In: Grossman PD, Colburn JC, eds. Capillary
Electrophoresis. San Diego: Academic Press, 1992:3-43.
Botek P. Analytical capillary electrophoresis. In: Churitek J, ed. Advanced Instru-
mental Methods of Chemical Analysis. Praha: Academia, 1993:97- 14 1.
Foret F, Kfivinkovi L, BoEek P. Capillary Zone Electrophoresis. Weinheim: Ver-
lag Chemie, 1993.
Moring SE. Buffers, electrolytes, and additives for capillary electrophoresis. In:
Righetti PC, ed. Capillary Electrophoresis in Analytical Biotechnology. Boca Ra-
ton: CRC Press, 1996:37-60.
Issaq HJ, Janini GM, Chan KC, Elrassi Z. Approaches for the optimization of exper-
imental parameters in capillary zone electrophoresis. Adv Chromatogr 1995;35:
101-169.
Hannig K, Heidrich HK. Free-Flow Electrophoresis. Darmstadt: GIT, 1989.
Reijenga JC, Verheggen TPEM, Martens JHPA, Everaerts FM. Buffer capacity,
ionic strength and heat dissipation in capillary electrophoresis. J Chromatogr A
l996;744: 147- 153.
Nielsen RG, Rickard EC. Method optimization in CZE analysis of hGH tryptic
digest fragments. J Chromatogr 1990;5 16:99-114.
Swedberg SA. Use of non-ionic and zwitterionic surfacants to entrance selectivity
in HPCE. J Chromatogr 1990;503:449-452.
Greve KF, Nashabeh W, Karger BL. Use of zwitterionic detergents for the separa-
tion of closely related peptides by capillary electrophoresis. J Chromatogr A 1994;
680: 15-24.
Matsubara N, Koezuka K, Terabe S. Separation of eleven angiotensin I1 analogs by
capillary electrophoresis with a nonionic surfactant in acidic media. Electrophoresis
1995;16:580-583.
Liu J, Cobb KA, Novotny M. Capillary electrophoretic separations of peptides us-
ing micelle-forming compounds and cyclodextrins as additives. J Chromatogr 1990;
519:189-197.
Kornfelt T, Vinther A, Okafo GN, Camilleri P. Improved peptide mapping using
phytic acid as ion-pairing buffer additive in capillary electrophoresis. J Chromatogr
A 1996;726:223-228.
Matsubara N, Terabe S. Micellar electrokinetic chromatography. Meth Enzymol
l996;270:3 19-34 1.
Matsubara N, Terabe S. Micellar electrokinetic chromatography in the analysis of
amino acids and peptides. In: Righetti PC, ed. Capillary Electrophoresis in Analyti-
cal Biotechnology. Boca Raton: CRC Press, 1996:1.55- 182.
Vindevogel J, Sandra P. Introduction to Micellar Electrokinetic Chromatography.
Heidelberg: Huthig, 1992.
Yashima T, Tsuchiya A, Morita 0. Separation of closely related large peptides by
micellar electrokinetic chromatography with organic modifiers. Anal Chem 1992;
64:298 1-2984.
Ye B, Hadjmohammadi M, Khaledi MG. Selectivity control in micellar electroki-
netic chromatography of small peptides using mixed fluorocarbon hydrocarbon an-
ionic surfactants. J Chromatogr A 1995;692:29 1-300.
Castagnola M. Cassiano L, Messana I, Paci M, Rossetti DV, Giardina B. Effect
7
of 2,2,2-trifluoroethanol on capillary zone electrophoretic peptide separations. J

Chromatogr A 1996;735:271-281.
Oda RP, Madden BJ, Morris JC, Spelsberg TC, Landers JP. Multiple-buffer-addi-
tive strategies for enhanced capillary electrophoretic separation of peptides. J Chro-
matogr A 1994;680:341-35 1.
Johansson IM, Huang EC, Henion JD, Zweigenbaum J. Capillary electrophoresis
atmospheric pressure ionization mass spectrometry for the characterization of pep-
tides: instrumental considerations for mass spectrometric detection. J Chromatogr
1991;554:311-327.
Hansen SH, Tjornelund J, Bjornsdottir I. Selectivity enhancement in capillary elec-
trophoresis using non-aqueous media. Trends Anal Chem 1996; 15: 175- 180.
Sahota RS, Khaledi MG. Nonaqueous capillary electrophoresis. Anal Chem 1994;
66:1141-1146.
Chiari M, Nesi M, Righetti PG. Surface modifications of silica walls: a review of
different methodologies. In: Righetti PG, ed. Capillary Electrophoresis in Analyti-
cal Biotechnology. Boca Raton: CRC Press, 1996:l-36.
Yeung ES. Optical detectors for capillary electrophoresis. Adv Chromatogr 1995;
35:l-51.
Flint CD, G r o c h o w i c z ' ~Simpson
~, CF. Design, construction and evaluation of an
ultraviolet absorbance detector for dapillary electrophoresis. Anal Proc Anal Comm
l994;3l:ll7-l2l.
Prusik Z, KaSitka V, StanEk S, Kuncova G, Hayer M, VrkoE J. Experimental device
for electrokinetic micellar chromatography exploiting some components of capil-
lary isotachophoresis instrumentation. J Chromatogr 1987;390:87-96.
Boring CB, Dasgupta PK. An affordable high-performance optical absorbance de-
tector for capillary systems. Anal Chim Acta 1997;342: 123- 132.
Heiger DN, Kaltenbach P, Sievert HJP. Diode array detection in capillary electro-
phoresis. Electrophoresis 1994;15:1234- 1247.
Quanta 4000E high performance capillary electrophoresis system. A quantum
leap in protein, peptide, and DNA analysis. Technical note, Millipore Corporation,
1993.
Heiger D, Grimm R, Herold M. Peptide mapping and analysis using capillary elec-
trophoresis. Hewlett-Packard Application Note, Publication number 1993;12-
5091-9062E.
Becklin RR, Desiderio DM. The amount of ultraviolet absorbance in a synthetic
peptide is directly proportional to its number of peptide bonds. Anal Lett 1995;28:
2 175-2 190.
Heiger DN. High Performance Capillary Electrophoresis: An Introduction. Hew-
lett-Packard Primer, Publication 12-509 1 -7913E, 1993.
Djordjevic NM, Widder M, Kuhn R. Signal enhancement in capillary electrophore-
sis by using a sleeve cell arrangement for optical detection. J High Res Chromatogr
1997;20: 189-192.
Ross G. Applications of the HP-3D Capillary Electrophoresis System. Hewlett-
Packard Primer, Publication 12-5963-7140E. 1995.
Green JS, Jorgenson IC. Variable wavelength on-column fluorescence detector for
open tubular zone electrophoresis. J Chromatogr 1986;352:337-343.
Lee 'IT, Yeung ES. Capillary electrophoresis detectors: Lasers. Meth Enzymol
1996;270:419-449.
Nouadje G, Amsellem J, Couderc B, Verdeguer P, Couderc F. Capillary electropho-
resis with laser-induced fluorescence detection: optical designs and applications.
In: Parvez H, Caudy P, Parvez S, Roland-Gosselin P, eds. Capillary Electrophoresis
in Biotechnology and Environmental Analysis. Utrecht: VSP, 1997:49-72.
Schwartz HE, Ulfelder KJ, Chen lTA, Pentoney SL. The utility of laser-induced
fluorescencedetection in applications of capillary electrophoresis. J Cap Elec 1994;
1:36-54.
MacTaylor CE, Ewing AG. Critical review of recent developments in fluorescence
detection for capillary electrophoresis. Electrophoresis 1997;18:2279-2290.
Chen DY, Dovichi NJ. Single-molecule detection in capillary electrophoresis: Mo-
lecular shot noise as a fundamental limit to chemical analysis. Anal Chem 1996;
68:690-696.
Lee TT, Yeung ES. High-sensitivity laser-induced fluorescence detection of native
proteins in capillary electrophoresis. J Chromatogr 1992;595:319-325.
Timperman AT, Oldenburg KE, Sweedler JV. Native fluorescence detection and
spectral differentiation of peptides containing tryptophan and tyrosine in capillary
electrophoresis. Anal Chem 1995;67:3421-3426.
Chan KC, Janini GM, Muschik GM, Issaq HJ. Pulsed UV laser-induced fluores-
cence detection of native peptides and proteins in capillary electrophoresis. J Liq
Chromatogr 1993;16:1877- 1890.
Chan KC, Muschik GM, Issaq HJ. Separation of tryptophan and related indoles by
micellar electrokinetic chromatography with KrF laser-induced fluorescence detec-
tion. J Chromatogr A 1995;718:203-210.
Chang HT, Yeung ES. Determination of catecholaminesin single adrenal medullary
cells by capillary electrophoresis and laser-induced native fluorescence. Anal Chem
l995;67: 1079- 1083.
Tong W, Yeung ES. Determination of insulin in single pancreatic cells by capillary elec-
trophoresis and laser-induced native fluorescence. J Chromatogr B 1996;685:35-40.
Lee TT, Yeung ES. Quantitative determination of native proteins in individual hu-
man erythrocytes by capillary zone electrophoresis with laser-induced fluorescence
detection. Anal Chem 1992;64:3045-305 1.
Bardelmeijer HA, Waterval JCM, Lingeman H, Vanthof R, Bult A, Underberg
WJM. Pre-, on- and post-column derivatization in capillary electrophoresis. Elec-
trophoresis 1997;18:2214-2227.
Zhao JY, Waldron KC, Miller J, Zhang JZ, Harke H, Dovichi NJ. Attachment of
a single fluorescent label to peptides for determination by capillary zone electropho-
resis. J Chromatogr 1992;608:239-242.
Little MJ, Paquette DM, Harvey MD, Banks PR. Single-label fluorescent derivati-
zation of peptides. Anal Chim Acta 1997;339:279-288.
Liu JP, Hsieh YZ, Wiesler D, Novotny M. Design of 3-(4-carboxybenzoy1)-2-quin-
oline-carboxaldehyde as a reagent for ultrasensitive determination of primary
amines by capillary electrophoresis using laser fluorescence detection. Anal Chem
1991;63:408-412.
Cobb KA, Novotny MV. Peptide mapping of complex proteins at the low-picomole
level with capillary electrophoretic separations. Anal Chem 1992;64:879-886.
Cobb KA, Novotny MV. Selective determination of arginine-containing and tyro-

sine-containing peptides using capillary electrophoresis and laser-induced fluores-
cence detection. Anal Biochem 1992;200:149-1 55.
Fadden P, HaysteadTAJ. Quantitative and selective fluorophore labeling of phospho-
serine on peptides and proteins: characterization at the attomole level by capillary
electrophoresis and laser-induced fluorescence. Anal Biochem 1995;225:81-88.
Shippy SA, Jankowski JA, Sweedler JV. Analysis of trace level peptides using
capillary electrophoresis with UV laser-induced fluorescence. Anal Chim Acta
1995;307:163-171.
Craig DB, Wong JCY, Dovichi NJ. Detection of attomolar concentrations of alka-
line phosphatase by capillary electrophoresis using laser-induced fluorescence de-
tection. Anal Chem 1996;68:697-700.
Advis JP, Guzman NA. Capillary electrophoresis coupled to fluorescence detection
for the determination of in-vivo releaseof multiple neuropeptides from the ewe
median eminence. J Liq Chromatogr 19%; l6:2 129-2 148.
Park SS, Hung WL, Schaufelberger DE, Guzman NA, Advis JP. Determination
of neuropeptides by capillary electrophoresis. In: Irvine GB, Williams CH, eds.
Neuropeptide Protocols. Totowa: Humana Press, 1997:101-1 1 1.
Hogan BL, Yeung ES. Determination of intracellular species at the level of a single
erythrocyte via capillary electrophoresis with direct and indirect fluorescence detec-
tion. Anal Chem 1992;64:2841-2845.
Wong KS, Yeung ES. Simultaneous monitoring of glutathione and major proteins
in single erythrocytes. Mikrochim Acta 1995;120:321-327.
Tsukagoshi K, Tanaka A, Nakajima R, Hara T. On-line capillary zone electropho-
retic separation-chemiluminescence detection of protein labeled with fluores-
camine. Anal Sci 1996; 12525-528,
Staller TD, Sepaniak MJ. Chemiluminescence detection in capillary electrophore-
sis. Electrophoresis l997;18:229 1-2296.
Tomer KB, Deterding LJ, Parker CE. Capillary electrophoresis coupled with mass
spectrometry. Adv Chromatogr 1995;35:53-99.
Smith RD, Udseth HR. Wahl JH, Goodlett DR, Hofstadler SA. Capillary electro-
phoresis mass spectrometry. Meth Enzymol 1996;271:448-486.
Banks JF. Recent advances in capillary electrophoresis electrospray mass spectrom-
etry. Electrophoresis 1997;18:2255-2266.
Tomer KB, Parker CE, Deterding LJ. Capillary electrophoresis interfaced with
mass spectrometry: electrospray ionization and continuous flow fast atom bombard-
ment. In: Righetti PG, ed. Capillary Electrophoresis in Analytical Biotechnology.
Boca Raton: CRC Press, 1996:123-153.
Suter MJF, Caprioli RM. An integral probe for capillary zone electrophoresis/con-
tinuous-flow fast atom bombardment mass spectrometry. J Am Soc Mass Spectrom
l992;3: 198-206.
Loo JA, Udseth HR, Smith RD. Peptide and protein analysis by electrospray ioniza-
tion-mass spectrometry and capillary electrophoresis-mass spectrometry. Anal
Biochem 1989;179:404-412.
Licklider L, Kuhr WG, Lacey MP, Keough T, Purdon MP, Takigiku R. On-line
microreactors capillary electrophoresis mass spectrometry for the analysis of pro-
teins and peptides. Anal Chem 1995;67:4 170-41 77.
Hynek R, KaSirka V, Kurerovi Z, KiS. J. Fast detection of phosphorylation of
human pepsinogen A, human pepsinogen C and swine pepsinogen using a combina-
tion of reversed-phase high-performance liquid chromatography and capillary zone
electrophoresis for peptide mapping. J Chromatogr B 1997;688:213-220.
Koezuka K, Ozaki H, Matsubara N, Terabe S. Separation and detection of closely
related peptides by micellar electrokinetic chromatography coupled with electro-
spray ionization mass spectrometry using the partial filling technique. J Chromatogr
B 1997;689:3-11.
Banks JF, Dresch T. Detection of fast capillary electrophoresis peptide and protein
separations using electrospray ionization with a time-of-flight mass spectrometer.
Anal Chem 1996;68: 1480- 1485.
Fang LL, Zhang R, Williams ER, Zare RN. On-line time-of-flight mass spectromet-
ric analysis of peptides separated by capillary electrophoresis. Anal Chem 1994;
66:3696-3701.
Major HJ, Ashcroft AE. Analysis of a standard peptide mixture by capillary electro-
phoresis with mass spectrometry and with tandem mass spectrometry. Rapid Com-
mun Mass Spectrom 1996; 10:1421- 1426.
Moseley MA, Jorgenson JW, Shabanowitz J, Hunt DF, Tomer KB. Optimization
of capillary zone ~lectrophoresis/electrosprayionization parameters for the mass
spectrometry and tandem mass spectrometry analysis of peptides. J Am Soc Mass
Spectrom 1992;3:289-300.
Figeys D, Vanoostveen 1, Ducret A, Aebersold R. Protein identification by capillary
zone electrophoresis/microelectrosprayionization-tandem mass spectrometry at
the subfemtomole level. Anal Chem 1996;68:1822-1828.
Valaskovic GA, Kelleher NL, Mclafferty FW. Attomole protein characterization
by capillary electrophoresis mass spectrometry. Science 1996;273: 1199-1202.
Lewis KC, Opiteck GJ, Jorgenson JW, Sheeley DM. Comprehensive on-line RPLC-
CZE-MS of peptides. J Am Soc Mass Spectrom 1997;8:495-500.
Walker KL, Chiu RW, Monnig CA, Wilkins CL. Off-line coupling of capillary
electrophoresis and matrix-assisted laser desorption ionization time-of-flight mass
spectrometry. Anal Chem 1995;67:4197-4204.
Yoo YS, Han YS, Suh MJ, Park J. Analysis of phosphopeptides by capillary electro-
phoresis and matrix-assisted laser-desorption ionization time-of-flight mass spec-
trometry. J Chromatogr A 1997;763:285-293.
Alexander AJ, Hughes DE. Monitoring of IgG antibody thermal stability by micel-
lar electrokinetic capillary chromatography and matrix-assisted laser desorptiont
ionization mass spectrometry. Anal Chem 1995;67:3626-3632.
Keough T, Takigiku R, Lacey MP, Purdon M. Matrix-assisted laser desorption mass
spectrometry of proteins isolated by capillary zone electrophoresis. Anal Chem
l992;64: 1594- 1600.
Kelly JF, Locke SJ, Ramaley L, Thibault P. Development of electrophoretic condi-
tions for the characterization of protein glycoforms by capillaryelectrophoresis
electrospray mass spectrometry. J Chromatogr A 1996;720:409-427.
Ewing AG, Mesaros JM, Gavin PF. Electrochemical detection in microcolumn sep-
arations. Anal Chem 1994;66:A527-A537.
Voegel PD, Baldwin RP. Electrochemical detection in capillary electrophoresis.
Electrophoresis 1997;18:2267-2278.
Voegel PD, Baldwin RP. Electrochemical detection with copper electrodes in liquid
chromatography and capillary electrophoresis. Int Lab 1996;26: 16A- 16M.
O'Shea TJ, Greenhagen RD, Lunte SM, Lunte CE, Smyth MR, Radzik DM, Wata-
nabe N. Capillary electrophoresis with electrochemical detection employing an on-
column Nafion joint. J Chrornatogr l992;593:305-3 12.
Ye JN, Baldwin RP. Determination of amino acids and peptides by capillary elec-
trophoresis and electrochemical detection at a copper electrode. Anal Chem 1994;
66:2669-2674.
O'Shea TJ, Lunte SM. Selective detection of free thiols by capillary electrophoresis
electrochemistry using a gold mercury amalgam microelectrode. Anal Chem 1993;
65:247-250.
Weber PL, Kornfelt T, Klausen NK, Lunte SM. Characterization of glycopeptides
from recombinant coagulation factor VIIa by high-performance liquid chromatog-
raphy and capillary zone electrophoresis using ultraviolet and pulsed electrochemi-
cal detection. Anal Biochem 1995;225: 135- 142.
Huang XH, Zare RN, Sloss S, Ewing AG. End-column detection for capillary zone
electrophoresis. Anal Chem 1991;63:189- 192.
Muller D, Jelinek I, Opekar F, Stulik K. A conductometric detector for capillary
separations. ~lectroanal~sis 1996:8:722-725.
Huang X, Pang TKJ, Gordon MJ, Zare RN. On-column conductivity detector for
capillary zone electrophoresis. Anal Chem 1987;59:2747-2749.
Dasgupta PK, Bao LY. Suppressed conductometric capillary electrophoresis sepa-
ration systems. Anal Chem 1993;65:1003- 101l.
Kar S, Dasgupta PK, Liu HH, Hwang H. Computer-interfaced bipolar pulse con-
ductivity detector for capillary systems. Anal Chem 1994;66:2537-2543.
Jandik P, Bonn G. Capillary Electrophoresis of Small Molecules and Ions. Cam-
bridge: VCH, 1993.
Huang XH, Luckey JA, Gordon MJ, Zare RN. Quantitative analysis of low molecu-
lar weight carboxylic acids by capillary zone electrophoresislconductivity detec-
tion. Anal Chem 1989;61:766-770.
Grossman PD, Wilson KJ, Petrie G, Lauer HH. Effect of buffer pH and peptide
composition on the selectivity of peptide separations by capillary zone electropho-
resis. Anal Biochem 1988;l73:265-270.
Colburn JC. Capillary electrophoresis separations of peptides: practical aspects and
applications. In: Grossman PD, Colburn JC, eds. Capillary Electrophoresis: Theory
and Practice. San Diego: Academic Press, 1992:237-27 1.
Nashabeh W, Greve KF, Kirby D, Foret F, Karger BL, Reifsnyder DH,
Builder SE. Incorporation of hydrophobic selectivity in capillary electrophoresis:
analysis of recombinant insulin-like growth factor I variants. Anal Chem 1994;66:
2148-2154.
Grossman PD, Colburn JC, Lauer HH, Nielsen RG, Riggin RM, Sittampalam GS,
Rickard EC. Application of free-solution capillary electrophoresis to the
analytical scale separation of proteins and peptides. Anal Chem 1989;61: 1186- 1194.
Nielsen RG, Sittampalam GS, Rickard EC. Capillary zone electrophoresis of insulin
and growth hormone. Anal Biochem 1989;177:20-26.
Vinther A, Holm A, Hoeejensen T, Jespersen AM, Klausen NK, Christensen T,
Sorensen HH. Synthesis of stereoisomers and isoforms of a tryptic heptapeptide
fragment of human growth hormone and analysis by reverse-phase HPLC and capil-
lary electrophoresis. Eur J Biochem 1996;235:304-309.
Wan H, Blomberg LC. Enantiomeric separation by capillary electrophoresis of di-
and tripeptides derivatized with 9-fluorenylmethyl chloroformate using vancomycin
as chiral selector. J Microcolumn Sep 1996;8:339-344.
Wan H, Blomberg LC. Enantiomeric and diastereomeric separation of di- and tri-
peptides by capillary electrophoresis. J Chromatogr A 1997;758:303-311.
Sanger-van de Griend CE, Groningsson K, Arvidsson T. Enantiomeric separation
of a tetrapeptide with cyclodextrin: extension of the model for chiral capillary elec-
trophoresis by complex formation of one enantiomer molecule with more than one
chiral selector molecules. J Chromatogr A 1997;782:271-279.
Moore AW, Jorgenson JW. Resolution of cis and trans isomers of peptides con-
taining proline using capillary zone electrophoresis. Anal Chem 1995;67:3464-3475.
Ma S, Kalman F, Kalman A, Thunecke F, Howath C. Capillary zone electrophore-
sis at subzero temperatures: separation of the cis and trans conformers of small
peptides. J Chromatogr A 1995;716: 167- 182.
Survay MA, Goodall DM, Wren SAC, Rowe RC. Oligoglycines and oligoalanines
as tests for modelling mobility of peptides in capillary electrophoresis. J Chro-
matogr 1993;636:81-86.
KaSiEka V, Prusik Z, Mudra P, StEtpinek J. Capillary electrophoresis device with
double UV detection and its application to the determination of effective mobilities
of peptides. J Chromatogr A 1995;709:3 1-38.
Gao JM, Gomez FA, Harter R, Whitesides GM. Determination of the effective
charge of a protein in solution by capillary electrophoresis. Proc Natl Acad Sci
USA l994;9l: 12027- 12030.
Zhang Y, Lee HK, Li SFY. Separation of myoglobin molecular mass markers using
non-gel sieving capillary electrophoresis. J Chromatogr A 1996;744:249-257.
Walbroehl Y, Jorgenson JW. Capillary zone electrophoresis for the determination
of electrophoretic mobilities and diffusion coefficients of proteins. J Microcolumn
Sep 1989;1:41-45.
Rao JH, Whitesides GM. Tight binding of a dimeric derivative of vancomycin with
dimeric L-Lys-D-Ala-D-Ala. J Am Chem Soc 1997;119:10286-10290.
Rao JH, Colton IJ, Whitesides GM. Using capillary electrophoresis to study the
electrostatic interactions involved in the association of ~ - A l a - ~ - Awith
l a vancomy-
cin. J Am Chem Soc 1997;119:9336-9340.
Letourneau DL, Allen NE. Use of capillary electrophoresis to measure dimerization
of glycopeptide antibiotics. Anal Biochem 1997;246:62-66.
Kiivinkovi L, BoCek P. Synergism of capillary isotachophoresis and capillary zone
electrophoresis. J Chromatogr B 1997;689: 13-34.
Goodall DM, Williams SJ, Lloyd DK. Quantitative aspects of capillary electropho-
resis. Trends Anal Chem 1991;10:272-279.
Prusik Z, KaSitka V, Mudra P, StEtpinek J, SmCkal 0 , Hlavieek J. Correlation
of capillary zone electrophoresis with continuous free-flow zone electrophoresis:
application to the analysis and purification of synthetic growth hormone releasing
peptide. Electrophoresis 1990;11:932-936.
Novotny MV, Cobb KA, Liu JP. Recent advances in capillary electrophoresis of
proteins, peptides and amino acids. Electrophoresis 1990;11:735-749.
Righetti PG, Gelfi C, Perego M, Stoyanov AV, Bossi A. Capillary zone electropho-
resis of oligonucleotides and peptides in isoelectric buffers: theory and methodol-
ogy. Electrophoresis 1997;18:2145-2153.
Kuhr WG, Monnig CA. Capillary electrophoresis. Anal Chem 1992;64:R389-R407.
Monnig CA, Kennedy RT. Capillary electrophoresis. Anal Chem 1994;66:R280-R3 14.
Stclaire RL. Capillary electrophoresis. Anal Chem 1996;68:R569-R586.
KaSiEka V, Prusik Z. Capillary zone electrophoresis and isotachophoresis of biolog-
ically active peptides. In: Parvez H, Caudy P, Parvez S, Roland-Gosselin P, eds.
Capillary Electrophoresis in Biotechnology and Environmental Analysis. Utrecht:
VSP, 1997:173-197.
Castagnola M, Messana I, Rossetti DV. Capillary zone electrophoresis for the anal-
ysis of peptides. In: Righetti PG, ed. Capillary Electrophoresis in Analytical Bio-
technology. Boca Raton: CRC Press, 1996:239-275.
Li SFY. Capillary Electrophoresis. Amsterdam: Elsevier, 1992.
Rose DJ, Jorgenson JW. Fraction collector for capillary zone electrophoresis. J
Chromatogr 1988;438:23-34.
Lee HG, Desiderio DM. Preparative capillary zone electrophoresis of synthetic pep-
tides: conversion of an autosarnpler into a fraction collector. J Chromatogr A 1994;
686:309-3 17.
Smith AJ. Analytical and micropreparative capillary electrophoresis of peptides.
In: Smith BJ, ed. Protein Sequencing Protocols. Totowa: Humana Press, 1997:91-99.
Boss HJ, Rohde MF, Rush RS. Multiple peptide fraction collection by capillary
electrophoresis with reinjection analysis. Peptide Res 1996;9:203-209.
Herold M, Wu SL. Automated peptide fraction collection in CE. LC GC-Mag Sepa-
ration Sci 1994;12:531-533.
Grimm R, Herold M. Micropreparative single run fraction collection of peptides
separated by CZE for protein sequencing. J Cap Elec 1994;1:79-82.
Lee HG, Tseng JL, Becklin RR, Desiderio DM. Preparative and analytical capillary
zone electrophoresis analysis of native endorphins and enkephalins extracted from
the bovine pituitary: mass spectrometric confirmation of the molecular mass of
leucine enkephalin. Anal Biochem 1995;229:188-197.
Boss HJ, Rohde MF, Rush RS. Multiple sequential fraction collection of peptides
and glycopeptides by high-performance capillary electrophoresis. Anal Biochem
1995;230:123-129.
Bergman AC, Bergman T. Micropreparation of peptides by capillary electrophoresis
for matrix assisted laser desorption mass spectrometry. FEBS Lett 1996:397:45-49.
Bergman T, Agerberth B, Jornvall H. Direct analysis of peptides and amino acids
from capillary electrophoresis. FEBS Lett 1991;283:100- 103.
Wallingford RA, Ewing AG. Amperometric detection of catechols in capillary zone
electrophoresis with normal and micellar solutions. Anal Chem 1988;60:258-263.
Huang X, Zare RN. Continuous sample collection in capillary zone electrophoresis
by coupling the outlet of a capillary to a moving surface. J Chromatogr 19905 16:
185-189.
Fujimoto C, Fujikawa T, Jinno K. Sample collection by a capillary zone electropho-
retic system with an on-column fracture. J High Res Chromatogr 1992;15:201-203.
Cheng YF, Fuchs M, Andrews D, Carson W. Membrane fraction collection for
capillary electrophoresis. J Chromatogr 1992;608:109-116.
96 Kasicka
Muller 0 , Foret F, Karger BL. Design of a high-precision fraction collector for

capillary electrophoresis. Anal Chem 1995;67:2974-2980.
Chiu RW, Walker KL, Hagen JJ, Monnig CA, Wilkins CL. Coaxial capillary and
conductive capillary interfaces for collection of fractions isolated by capillary elec-
trophoresis. Anal Chem 1995;67:4190-4196.
Banke N, Hansen K, Diers I. Detection of enzyme activity in fractions collected
from free solution capillary electrophoresis of complex samples. J Chromatogr
1991;559:325-335.
Tsuda T, Sweedler JV, Zare RN. Rectangular capillaries for capillary zone electro-
phoresis. Anal Chem 1990;62:2149-2152.
Cifuentes A, Rodriguez MA, Garciamontelongo FJ. Rectangular capillary electro-
phoresis: study of some dispersive effects. J Chromatogr A 1996;737:243-253.
Fujimoto C, Matsui H. Sawada H, Jinno K. The use of a microconcentric column
in capillary electrophoresis. J Chromatogr A 1994;680:33-42.
Camilleri P, Okafo GN, Southan C, Brown R. Analytical and micropreparative
capillary electrophoresis of the peptides from calcitonin. Anal Biochem 1991;198:
36-42.
Schwer C, Lottspeich F. Analytical and micropreparative separation of peptides by
capillary zone electrophoresis using discontinuous buffer systems. J Chromatogr
1992;623:345-355.
Cifuentes A, Xu X, Kok WT, Poppe H. Optimum conditions for preparative opera-
tion of capillary zone electrophoresis. J Chromatogr A 1995;716: 141- 156.
musik Z. Free-flow electromigration separations. J Chromatogr 1974;91:867-872.
Prusik Z. Continuous flow-through electrophoresis. In: Deyl Z, ed. Electrophoresis,
Part A: Techniques. Amsterdam: Elsevier. 1979:229-25 1.
Wagner H, Mang V, Kessler R, Speer W. A free-flow system for preparative separa-
tions. In: Holloway CJ, ed. Analytical and Preparative Isotachophoresis. Berlin:
Walter de Gruyter, 1984347-356.
Wagner H, Heinrich J. Free-flow electrophoresis for the separation and purification
of biopolymers. In: Tschesche H, ed. Modern Methods in Protein and Nucleic Acid
Research. Berlin: Walter de Gruyter, 1990:69-97.
Roman MC, Brown PR. Free-flow electrophoresis. Anal Chem 1994;66:A86-A94.
Nath S, Schutte H, Hustedt H, Deckwer WD. Application of continuous zone elec-
trophoresis to preparative separation of proteins. Biotechnol Bioeng 1993;42:829-835.
Nath S, Schutte H, Hustedt H, Deckwer WD, Weber G. Separation of enzymes
from microorganism crude extracts by free-flow zone electrophoresis. Biotechnol
Bioeng 1996;5 1: 15-22.
Baier TG, Weber G, Hartmann K, Heinrich U, Schonberg D. Preparative separation
of human B and T lymphocytes by free flow electrophoresis. Anal Biochem 1988;
171:91-95.
Weber G, BoZek P. Optimized continuous flow electrophoresis. Electrophoresis
1996;17:1906-1910.
Clifton MJ, Roux-de Balmann H, Sanchez V. Protein separation by continuous-
flow electrophoresis in microgravity. AIChE J 1996;42:2069-2079.
Kobayashi H, Ishii N, Nagaoka S. Bioprocessing in microgravity: applications of
continuous flow electrophoresis to rat anterior pituitary particles-free flow elec-
trophoresis of C-elegans DNA. J Biotechnol 1996;47:367-376.
Hymer WC, Salada T, Cenci R, Krishnan K, Seaman GVF, Snyder R, Matsumiya

H, Nagaoka S. Bioprocessing in microgravity: applications of continuous flow elec-
trophoresis to rat anterior pituitary particles. J Biotechnol 1996;47:353-365.
Raymond DE, Manz A, Widmer HM. Continuous sample pretreatment using a free-
flow electrophoresis device integrated onto a silicon chip. Anal Chem 1994;66:
2858-2865.
Kaiitka V, Prusik Z, PospiSek J. Conversion of capillary zone electrophoresis to
free-flow zone electrophoresis using a simple model of their correlation: application
to synthetic enkephalin-type peptide analysis and preparation. J Chromatogr 1992;
608: 13-22.
Jeiek J, Velkovi V, KaSiEka V, Prusik Z, Ubik K, Bartova K, Mare: M. Side
reaction during the deprotection of Cys(Acm)-containing peptides with iodine. Syn-
thesis of disulfide fragments from cathepsin D structure. Collect Czech Chem Com-
mun 1995;60: 1042- 1049.
KaSic'ka V, Prusik Z, Smkkal 0,HlaviEek J, Barth T, Weber G, Wagner H. Applica-
tion of capillary and free-flow zone electrophoresis and isotachophoresis to the
analysis and preparation of the synthetic tetrapeptide fragment of growth hormone-
releasing peptide. J Chromatogr B l994;656:9Y - 106.
Prusik Z, KaSiEka V, ~ e b e G,r Pospiiek J. Correlation of capillary electrophoresis
with continuous free-flow electrophoresis systems and their application in peptide
separations: analysis and preparation of LHRH derivative. In: Radola BJ, ed. Elek-
trophorese Forum '91. Technische Universitat Munchen, 1991:201-206.
KaSiEka V, Prusik Z, Sizelova P, JiriEek J, Barth T. Theory of the correlation
between capillary and free-flow zone electrophoresis and its use for the conversion
of analytical capillary separations to continuous free-flow preparative processes.
Application to analysis and preparation of fragments of insulin. J Chromatogr A
1998;796:211-220.
Bioaffinity Chromatography
Jaroslava Turkova
Academy of Sciences of the Czech Republic, Prague, and University
of Pardubice, Pardubice, Czech Republic
I. INTRODUCTION
Macromolecules such as proteins, polysaccharides, and nucleic acids often differ

only in their physicochemical properties within the individual groups, and their
isolation on the basis of these differences, e.g., by ion exchange chromatography,
gel filtration, or electrophoresis, is therefore difficult and time consuming. Conse-
quently, their activity decreases considerably during the isolation procedure ow-
ing to denaturation, cleavage, enzymatic hydrolysis, and so forth. One of the
most characteristic properties of these biologically active macromolecules is their
ability to bind other molecules reversibly. For example, active and regulatory
sites of enzymes form complexes with substrates, inhibitors, cofactors, or ef-
fectors; antibodies bind antigens against which they were prepared, etc. The for-
mation of these biospecific, dissociable complexes of biological macromolecules
serves as a basis for bioaffinity chromatography.
II. THE PRINCIPLE AND USE OF BlOAFFlNlTY

CHROMATOGRAPHY
Bioaffinity chromatography is a form of adsorption chromatography, which is

based on the exceptional ability of biologically active substances to bind specifi-
cally and reversibly complementary substances. These are generally called li-
gands, affinity ligands, or affinants. If one of the components of the complex is
immobilized, a specific sorbent is formed for the second component of the corn-
99
100 Turkova
plex assuming that the conditions necessary for the formation of this complex
exist. The binding sites of the immobilized substances must be sterically accessi-
ble even after their coupling to the solid support, and they must not be deformed
by immobilization.
A solid support with a covalently bound affinant is used as the stationary
phase in a chromatographic column. When a crude mixture containing the biolog-
ically active products to be isolated is passed through the column of affinity
adsorbent, then all of the compounds that under given experimental conditions
have no complementary binding site for the immobilized affinity ligand will pass
through unretarded. By contrast, products showing affinity for the insoluble affi-
nant are adsorbed onto the column. They can be released by elution with a solu-
tion of a soluble affinity ligand or by changing of the solvent composition by
so-called deforming buffers (e.g., by a change in pH, ionic strength, temperature,
etc.).
Assuming that a single enzyme of the crude proteins has an affinity for the
specific adsorbent, the equilibrium between the attached affinity ligand L and the
isolated enzyme E is given by the equation:
Affinity constant (= equilibrium or association constant):
Dissociation constant:
Dissociation constants of some affinity pairs were described by Bayer and

Wilchek in their review of avidin-biotin technology [I]. The unprecedented inter-
action between binding proteins' egg-white avidin, or its bacterial counterpart
streptavidin, and their target molecule vitamin biotin is based in the exceptionally
high affinity constants lOI5 M-I. Lower affinity constants were described for bio-
specific complexes: receptors-hormones, toxins, etc. (109-10'2 M-I); antibod-
ies-antigens (10'-10" M-I ); transport proteins-vitamins, sugars, etc. (lo6-10'
M ' ) ; lectins-carbohydrates (10'-lo6 M I ) ; and enzymes-substrates (10'-lo5
M - I).
Formation of the biologically functioning complexes involves the participa-
tion of common molecular forces and interactions systematized under the terms
of ionic, hydrogen and hydrophobic bonds, London's dispersion forces, dipole-
dipole interactions, charge-transfer interaction, and so on. The simultaneous and
concentrated action of several of these forces in the complementary binding site

constitutes the basis of the high specificity and efficiency of the biospecific bond.
The number of noncovalent interactions in complementary binding sites of
biospecific complexes can best be shown by x-ray diffraction analysis. The affin-
ity ligand most commonly used in the bioaffinity chromatography of aspartate
proteinases is the naturally occurring peptide inhibitor pepstatin. The mode of
binding of the first part of the pepstatin molecule (the fragment isovaleryl-valyl-
valyl-statine) to the binding site of penicillopepsin was elucidated by James and
coworkers [2].They determined the noncovalent interactions between the pep-
statin fragment and penicillopepsin which occur at a distance of 0.4 nm or less.
The total number of 87 includes a variety of interactions. The high number of
noncovalent interactions between the binding sites of microbial proteinases and
pepstatin is why the Mucor miehei proteinase cannot be eluted after adsorption
to an isovaleryl-pepstatin-Sepharose. However, after Kobayashi and coworkers
[3]had replaced immobilized N-isovalerylpepstatin by N-isobutyryl-, N-propio-
nyl-, or N-acetylpepstatin they were able to increase the yield of the enzyme
eluted from 0 to 12%, 46%, and 90%, respectively. The affinity constants of these
complexes were not reported.
The affinity constants of the biospecific complexes of modified pepstatin
were determined by Kay and coworkers [4].The value of the affinity constant
of the complex of cathepsin D with isovalerylpepstatin was equal to 10"' M - ' .
After the isovaleryl group had been replaced by the lactoyl group, the affinity
constant dropped three orders of magnitude.
This example shows that formation of complexes of lower affinity should
be used for biospecific affinity chromatography. However, in connection with
the use of biospecific complex formation in bioaffinity chromatography we must
always take into consideration the fact that the affinity constant of the biospecific
complex is influenced not only by the microenvironment of both solid and liquid
phases but by steric accessibility, the conformation and concentration of the af-
finity ligand, etc. This adsorbent can be useful for the isolation, determination,
or removal of biologically active substances. Complexes of higher affinity con-
stants can be used for oriented immobilization, e.g., of enzymes by simple adsorp-
tion to their suitable immunosorbent [ S ] .
Ill. CHOICE OF AFFINITY LIGANDS (AFFINANTS)

A. Highly Specific and Group-Specific Matrices
All of those compounds are suitable as affinity ligands for the isolation of biologi-
cally active products that bind such products specifically and reversibly. Hence,
given the varied nature of biologically active materials, affinants represent a very
102 Turkova
wide range of chemical compounds. Their classification can therefore be based

on biochemical function rather than on chemical' structure.
Two main criteria determine the selection of an affinity ligand:
1. The affinant to be immobilized must possess a functional group that
can be modified for attachment to the solid support without impairing
or abolishing its recognition by the complementary molecule. Not all
affinants that are suitable for a complementary binding of molecules
also have suitable functional groups for their attachment to a solid sup-
port. These groups must first be introduced into the affinants, as well
as suitably long spacing arms, which are usually indispensable in case
of low molecular weight affinity ligands because such arms are neces-
sary to enable a bonding interaction.
2. The affinity ligand should have an adequate affinity for the molecule
to be purified. Satisfactory results have usually been obtained with spe-
cific complexes having an affinity constant KA in the region of lO4-
lo8M-'. However, it is important to mention here that the K, values
under consideration are those determined for the complex of the sub-
stance to be isolated with the immobilized affinity ligand and that there-
fore they need not correspond to the affinity constant values determined
in solution. For the determination of KA the most commonly used
method is so-called quantitative affinity chromatography, which is
based on the elution of biological macromolecular substances from an
affinity matrix with soluble affinant solutions of various concentrations
[61.
Many examples of affinants used for the isolation of antibodies, antigens
and haptens, cells and cell organelles, cofactors and vitamins, enzymes, enzyme
subunits and modified derivatives, glycoproteins and saccharides, hormones, in-
hibitors, lectins, lipids, nucleic acids, nucleotides and nucleosides, transfer recep-
tors and binding proteins, proteins and peptides containing -SH group, specific
peptides, viruses, and other substances by use of low-pressure, high-performance,
and large-scale bioaffinity chromatography are shown on pages 372-578 of the
book Bioaflnity Chromatography [7].
According to the source we can distinguish (1) naturally occurring and
(2) synthetic affinity ligands. For the isolation of chymotrypsin and trypsin from
a crude pancreatic extract the specific adsorbents were prepared by coupling of
two naturally occurring proteinase inhibitors (ovomucoid for trypsin and polyva-
lent trypsin inhibitor, antilysine, for chymotrypsin) and also synthetic low molec-
ular weight proteinase inhibitors (N-benzyloxycarbonylglycyl-D-phenylalanine
for chymotrypsin andp-aminobenzamidine for trypsin) to a hydroxyalkyl methac-
rylate copolymer [8]. In Figs. 1 and 2 it is shown that identical results were
obtained with specific adsorbents prepared with both high molecular weight and
(1) --
1.5 -
C
a --- -
I
A 280 I
I
Figure 1 Chromatography of crude pancreatic extract on ovomucoid-Spheron (I) and

antilysine-Spheron (11). (I) A sample of active pancreatic extract (100 ml) was placed on
a column of ovomucoid-Spheron (10 X 2 cm), which was subsequently eluted with an
aqueous solution of ammonium formate (0.05 M formic acid adjusted to pH 9.0 with 5%
aqueous ammonia). Fractions (6 ml) were collected at 20-min intervals. The arrow desig-
nates the change in pH from 8.0 to 3.5 (0.1 M formic acid adjusted to pH 3.5 with ammo-
nia). (11) The fraction (a) of material not adsorbed (180 ml) was placed directly on the
antilysine-Spheron column (10 X 2 cm). The course of the chromatography was analogous
to that described for (I). (a) Contaminants and chymotrypsin; (b) trypsin; (c) chymotryp-
sin; - , absorbance at 280 nm; ---,pH. (Data from Ref. 8.)
low molecular weight inhibitors. Unlike the naturally occumng inhibitors, which
could undergo denaturation because of their protein character, the synthetic low
molecular weight inhibitors are completely stable. The capacity of specific adsor-
bents prepared with these inhibitors can be regenerated virtually without limit,
if a stable solid support and stable bonds between the support and the amino
groups of peptide inhibitors are used.
According to the specificity we distinguish (1) highly specific and (2)
group-specific matrices. The practical utility of specific sorbents increases if, in-
stead of the narrowly specific ligands, a so-called general ligand is used for their
preparation. As is implied by the name, a group-specific matrix prepared in this
Turkova
Figure 2 Chromatography of crude pancreatic extract on N-benzyloxycarbonylglycyl-

D-phenylalanine-NH,-Spheron (1) and NH2-benzamidine-NH2-Spheron (11). (I) A sample
of active pancreatic extract (100 ml) was applied to the column of N-benzyloxycarbonyl-
glycyl-D-phenylalanine-NH2-Spheron(6.0 X 1.5 cm). The course of the chromatography
was identical with that shown in Fig. 1. (11) Fraction (A), filtered through a column of
N-benzyloxycarbonylglycyl-D-phenylalanine, was placed directly on a column of NH2-
benzamidine-NH2-Spheron (25 X 1 cm). The course of chromatography is identical with
that shown in Fig. 1. ( A )Contaminants and trypsin; (B) chymotrypsin; (C) trypsin; -,
absorbance at 280 nm; ---,pH. (Data from Ref. 8.)
manner displays an affinity for a more or less large group of biological macromol-
ecules. For example, the enzymes related to the metabolism of aspartic acid show
group-specific adsorption affinity to N-(w-aminohexy1)-L-aspartic acid-Sepha-
rose. Asparaginase, aspartase, aspartate-P-decarboxylase, and asparaginase modi-
fied with tetranitromethane [9] could all be sorbed onto this immobilized affinant.
B. Enzymes
The combination of several complementary binding sites on the surface of the
enzyme molecule permits the formation of a relatively large number of biospec-
ific complexes to be achieved; these complexes can be utilized for efficient isola-
tion as well as for oriented immobilization. Figure 3 shows the surface of an
enzyme molecule covered with several complementary binding sites [lo]. Such
an enzyme could be, for instance, carboxypeptidase Y (CPY) containing the com-
plementary binding site for glycylglycyl-p-aminobenzylsuccinicacid, a specific
inhibitor, which after immobilization was used for the isolation of CPY from
autolyzates of various kinds of yeast. The active sile of the enzyme also contains
a free SH- group and therei'ore can be adsorbed to mercury-Spheron. CPY is a
glycoprotein whose carbohydrate moiety specifically interacts with concanavalin
A, a lectin. If CPY was adsorbed on immobilized concanavalin A followed by
crosslinking with glutaraldehyde, the bound enzyme retained 96% of the native
catalytic activity. The antigenic sites of the enzyme can be determined by investi-
gation of the antigenic structures of the peptide chain in experiments with specific
antibodies to this enzyme. Enzymes whose coenzymes are nucleotides can form
biospecific complexes with nucleotides. Enzymes form complexes with substrates
and their analogs, allosteric effectors, metal ions, etc. The type of complex formed
determines the mode of their action in chemical processes that take place in the
living cell.
The isolation of enzymes by use of immobilized inhibitors is shown in Fig.
I and 2. The catalytic activity of many enzymes depends on the presence of
coenzymes or cofactors. The coenzymes nicotinamide adenine dinucleotide and
Figure 3 Poss~bleways in which an enzyme can form biospecific coniplexes

106 Turkova
other nucleotides of adenine, uridine, guanine, and flavine, coenzymes of pyri-

doxal, folate and its analogs, biotin, lipoic acid, cobalamines, and porphine deriv-
atives are frequently employed as affinants. The function of immobilized cofac-
tors as bioaffinity ligands depends on the spacer length and on the position of
covalent binding. Figure 4 demonstrates four positions on the adenine nucleotide
that were derivatized before their attachment to CNBr-activated Sepharose 4B
[l 11. When comparing the binding of various dehydrogenases and kinases on N6-
(6-aminohexy1)-5'-AMP-Sepharose and P1-(6-aminohexyl)-P2-(5'-adenosine)py-
rophosphate-Sepharose 1121, the strength of the interaction between the enzyme
and the immobilized nucleotide was expressed by the so-called binding (0). This
term represents the concentration of potassium chloride (mM) in the center of
the enzyme peak when the enzymes are eluted with a linear potassium chloride
gradient (Fig. 5 , Table 1). The results reflect the nature of the enzyme-nucleotide
interactions.
The Corey-Pauling-Koltin structural model of NAD and the reactive textile
dye Cibacron Blue F3GA (Fig. 6) was published by Thompson et al. [13]. From x-
ray studies of the binding of Cibacron Blue F3GA to liver alcohol dehydrogenase
performed by Biellman et al. [14] it became evident that the dye binds at the
nucleotide binding site of the enzyme with correspondences of the adenine and
ribose rings, but not with the nicotinamide. The ever-increasing use of dye-based
affinity techniques in many fields of biomedical research and biotechnology is
Figure 4 The structures of several immobilized AMP adsorbents. (a) N6-(6-

aminohexy1)-AMP-agarose. (b) 8-(6-Aminohexy1)-AMP-agarose.(c) P1-(6-Aminohexyl)-
P2-(5'-adenosine)-pyrophosphate-agarose. (d) Ribosyl-linked AMP.
0 8 16 24 30
ELUTION VOLUME, m l
Figure 5 Chromatography of a crude yeast extract on N"(6-aminohexy1)-5'-AMP-

Sepharose. A sample (100 ~ 1of) a crude yeast extract was applied to a column (50 X 5
mm) containing 0.5 g of N6-(6:aminohexyl)-5'-AMP-Sepharose equilibrated with 10 mM
KH,POI-KOH buffer, pH 7.5. After washing through nonadsorbed proteins, enzymes were
eluted with a linear salt gradient (0.1 M KCl; 20 ml total volume) at a flow rate of 8 mllh.
lnert protein (--), glucose 6-phosphate dehydrogenase (I]), glutathione reductase (a),
malate dehydrogenase (O), and yeast alcohol dehydrogenase (A) were assayed in the
effluent (From Ref. 12.)
reflected in many contributions from a worldwide gathering of experts at the

International Conference on Modem Aspects of Protein-Dye Interaction: Role in
Downstream Processing [15]. Many advantages of using dye ligands, particularly
for large-scale applications, were described [15]. Compared to biological ligands,
dye ligands are inexpensive materials available in tonnage quantities worldwide.
There is a wide range of chromophors available that are dye-based biologically,
chemically, and photochemically stable adsorbents, potentially sterilizable in situ,
with no degradation of the ligand itself. Because they are reactive materials, they
are very easily immobilized to hydroxyl polymers, generally by a single-step
process. Such adsorbents have a high capacity, a very broad binding capability
in terms of the complementary proteins, and are easily reusable. Several thousand
different types of proteins would interact with an immobilized textile dye. Exam-
ples are oxidoreductases, phosphokinases, and nearly all coenzyme-dependent
enzymes, hydrolases, various transferases, a number of proteins that interact with
mono- and polynucleotides, synthetases, hydroxylases, nearly all of the glycolytic
enzymes, phosphatases, and a variety of blood proteins and other nonenzyme
proteins. A number of studies using classic enzyme kinetics, circular dichroism,
108 Turkova
Table 1 Comparison of B~ndingof Various Enzymes to N6-(6-Aminohexyl)-5'-AMP-

Sepharose (I) and P'-(6-Aminohexyl)-P2-(5-adenosine)-pyrophosphate-Seph~rose(11).
Binding (P)"
Enzyme
Code numbel I 11
Lactate dehydrogenase
Glucose 6-phosphate dehyclrogenase
Malate dehydrogenase
Alcohol dchydrogenase
p-Glyceraldehyde 3-phosphate dehydrogenase
3-Phosphoglycerate kinase
Pyruvate kinase
Hexokinase
Myokinase
Glycerokinase
'Binding (P) is he KC1 concenmtion (mM) at the cenler of the enzyme peak whcn the enzyme IS
eluted with a linear gradient of KCI.
Elution was erfected by a 200-yl pulse o f 5 mM NADH.
NEUTRAL 0
SULFONATE OR
PHOSPHATE 0
Figure 6 Corey-Pauling-Koltin structural model of NAD and Cibacron Blue F3GA.

NAD is shown on the left in the conformation in which it binds to dehydrogenases. The
Blue A model is shown on the right in a conformation similar to the NAD. Note the close
reseniblance in size, orientation of 7c electrons, and position of negatively charged sulfo-
nate or phosphate groups. (From Ref. 13.)
affinity labeling, x-ray diffraction, and other techniques have been utilized to
demonstrate the specificity of dye binding to the active sites of proteins.
C. Antibodies and Antigens

A refinement of the technique has been the development of immunoaffinity chro-
matography (IAC), which adds the selectivity and specificity of immunological
reactions to the separation process. Antigens or antibodies are increasingly used
as tools to separate, determine, or isolate complementary immunosubstances.
By binding of antigen to a solid support, a specific immunoadsorbent is
formed, which should possess the following properties:
1. It should be capable of adsorbing the complementary antibody from a
mixture of components.
2 . The liberation of the adsorbed antibody from the specific adsorbent
should be quantitative and should be carried out under conditions that
are harmless for the specific antibody activity.
3. It should possess a high capacity for the adsorption of the specific
antibody.
4. It should retain its biological activity after repeated use and storage.
5 . It should possess adequate mechanical properties, permitting centrifu-
gation, filtration, and use in a column.
The fulfillment of these requirements is not dependent on the quality and amount
of bound antigen only, but also on the nature of the solid support and the nature
of the bond.
Protein A, which is a coat protein extracted from the bacterium Staphylo-
coccus aureus, has the unique capacity to bind mammalian immunoglobulins,
especially IgG. Protein A-coated glass beads were developed as a universal sup-
port medium for HPIAC by Phillips et al. [16]. The preparation and mechanism
of IgG antibody binding is demonstrated in Fig. 7. Protein A has the ability to
bind to the Fc, or tail, portion of IgG antibodies. Protein A is composed of five
subunits, each with its own Fc receptor, but three of these receptors become
inactive when the molecule is immobilized. If Protein A is applied as a coating
to either solid or controlled pore glass beads, its Fc receptors become points of
attachment on which the antibody can be immobilized. The bound antibodies are
then covalently immobilized on the Protein A coat by crosslinking them with
carbodiimide. In turn this attachment helps to orient the antigen receptors of the
antibody toward the mobile phase of the column. In addition, the linear nature
of the Protein A molecule becomes a spacer arm for the immobilized antibodies,
thus preventing charges, which can rise from the bead surface, inhibiting the
formation of the antibody-antigen complex. Lee and Chuang [17] used Protein
A immobilized on nonporous silica for the isolation of human immunoglobulin
Turkova
IgC Fc RECEPTORS
B ANTIGEN RECEPTORS
PROTEIN A
RABBIT IgG ANTIBODIES
\
Fc PORTION
Figure 7 Diagrammatic representation of the process for binding rabbit antibodies to

Protein A-coated glass beads. (A) Illustration of the chemistry used to couple Protein A
to the carbonyl diimidazole-derivatized glass beads. (B) Mechanism of IgG antibody bind-
ing to the Fc receptors on the Protein A-coated glass beads. (From Ref. 16.)
G and for both theoretical and experimental investigation of elution behavior of

protein.
D. Lectins, Glycoproteins, and Saccharides

Lectins are proteins or glycoproteins from plants, invertebrates, and bacteria.
They form complexes with sugar residues of glucosides, oligo- and polysaccha-
rides, glycoproteins, glycolipids, membrane proteins, enzyme-antibody and gly-
coprotein conjugates, as well as viruses, cell fragments, and cells. Hydrogen
bonding and electrostatic interactions are fundamentally involved in complex for-

mation. Furthermore, the metal ions Ca2+and Mn2+are essential constituents of
the sugar binding sites and are also essential for the tertiary structure. Most known
lectins are multimeric, consisting of noncovalently associated subunits, which
are either identical (e.g., concanavalin A) or different (e.g., Ulex europeus agglu-
tinin). This multimeric structure is the reason for their ability to agglutinate cells
or form precipitates with glycoconjugates.
Bioaffinity chromatography for the purification of lectins was reviewed by
Lis and Sharon [18]. They described three major types of biospecific adsorbents
for the purification of lectins: (1) polysaccharides, either native or modified;
(2) matrix-bound glycoproteins and glycopeptides; (3) matrix-bound monosac-
charides and disaccharides. Hatakeyama et al. [19] described an assay for lectin
activity using microtiter plate with chemically immobilized carbohydrates: lac-
tose, galactose, mannose, glucose, and N-acetylglucosamine. After incubation
of the lectins (Ricinus communis agglutinin, concanavalin A, and wheat germ
agglutinin) bound proteins were measured by the protein assay using the colloidal
solution.
The isolation of glycoproteins by means of immobilized lectins makes use
of their differing affinities for terminal carbohydrate residues characteristic of
single glycoproteins. For the elaboration of a suitable procedure for the purifica-
tion of the given glycoproteins or glycopeptides by means of lectins, Kristiansen
[20] recommended the following stages:
1. Identification of the terminal sugar or sugars in the carbohydrate part
of the substance under consideration
2. Selection of a lectin with a corresponding specificity
3. Preparation of the selected lectin
4. Immobilization of the lectin by a covalent bond to a solid support
5. Choice of optimum conditions for the adsorption of the isolated sub-
stance on the immobilized lectin
6 . Choice of conditions for desorption
It is possible to choose either nonspecific elution, consisting mainly in a
change in pH or salt concentration; or a specific method can be used, i.e., displace-
ment of the adsorbed glycoprotein by competing carbohydrates. Assuming that
the terminal sugar or sugars of glycoproteins have been determined, the choice
of a suitable lectin can follow. In most instances lectins are not specific for one
sugar only, although great differences exist in the degree of specificity. For exam-
ple, wheat germ agglutinin has much lower affinity interactions than concanavalin
A [19]. Wheat germ agglutinin affinity HPLC (SigmaChrom AF-WGA) was used
by Pergami et al. [21] in semipreparative chromatographic method to purify the
normal cellular isoform of the prion protein in nondenatured form. It can be the
basis for study of the conversion of the cellular isoform in the pathogenesis of
112 Turkova
spongiform encephalopathies (prion diseases). Additional ligands in bioaffinity

chromatography of glycoproteins can be antibodies against carbohydrate antigens
and boronic acid ligands.
E. Nucleotides and Nucleic Acids

The ever-increasing use of genetic engineering is reflected in the rapidly increas-
ing number of publications devoted to the isolation of nucleic acids, genes, oligo-
nucleotides, and nucleic acid fragments, to the purification of special proteins
and enzymes, as well as to investigations of protein-nucleic acid interactions.
Schott [22] summarized this interesting field in his book AfJinity Chromatogra-
phy-Template Chromatography of Nucleic Acids and Proteins. When oligonu-
cleotides of a defined sequence are immobilized on cellulose, base pairing can
take place with complementary oligonucleotides having either a homologous se-
quence or an alternating one according to the Watson-Crick theory. Those nucleo-
tides undergoing such a base pairing are adsorbed onto the cellulose and can thus
be separated from any noncomplementary partners present in the mixture. Schott
described selective adsorptions of complementary oligonucleotides in the mobile
phase on the immobilized template if chromatography takes place under the con-
ditions necessary for base pairing. Desorption was then carried out with a temper-
ature gradient. The application of the principles of molecular biology for the selec-
tive separations of nucleotides and peptides was called template chromatography.
The chromatography of peptides on polyvinyl alcohol substituted with oli-
godeoxythymidylic acid and bound irreversibly on DEAE-cellulose by ionic
bonds has been employed for the study of the interactions of peptides with nucleo-
tides. The quantitative measure of the peptide-nucleotide interaction is the in-
crease in the retention of a peptide on oligonucleotide-DEAE-cellulose in com-
parison with that on unmodified DEAE-cellulose. In order to eliminate possible ;
effects of various column parameters, Schott [22] expressed all elution volumes
relative to the elution volume of alanine, which displays no measurable retention
on these columns. The relative elution ratio (V,) is thus obtained as the ratio of !
the elution volume found for the investigated peptide to that for alanine (V, = j
V,,,/V,,,). The peptide-oligonucleotide interaction is then evaluated on the basis 1
1
of the difference in the relative elution volumes obtained by chromatography on
both columns. i
A solid phase triple-helix-mediated affinity capture method was described 1
i.
for the purification of single-stranded M13 DNA for use as template in fluores-
cence-based DNA sequencing reactions by Johnson et al. [23]. In this method,
a biotinylated polypyrimidine oligonucleotide "loop" bound to streptavidin-
1
coated magnetic beads was used to selectively capture single-stranded M13 DNA
from high-titer phage supernatant through the formation of a cooperative and a
/1
polypurine site previously cloned into the M13 vector. I
I
Eisenberg et al. [24] reviewed the purification of DNA-binding proteins by

site-specific DNA affinity chromatography. They described the numerous advan-
tages that they experienced with the purification of DNA-binding protein (OBFI)
by use of DNA-cellulose. (1) The DNA affinity matrix has a high protein capac-
ity, which is determined by the number of copies of the protein recognition se-
quence inserted into the plasmid. It is possible to construct a plasmid in which
the protein-binding sequences conlprise at least 50% of the total DNA content.
Once inserted into the plasmid, followed by DNA amplification in E. coli, very
large quantities of these sequences could be obtained. Since about 1 mg of DNA
can be coupled to 1 g of cellulose powder, it is relatively easy to prepare sufficient
quantities of the DNA affinity matrix for a large-scale purification. (2) The protein
appears to bind more tightly to the DNA on the column than in solution. They
found that in solution the OBF1-DNA complex could not be formed in the pres-
ence of NaCl concentrations above 0.2 M, while the binding of OBFl to the
multimeric site-specific DNA-cellulose matrix occured at 0.4 M NaCl. The tight
binding to the site-specific DNA-cellulose column ensures effective separation
of OBFl from other DNA-binding proteins by simple salt elutions and, when
combined with standard ion exchange chromatography, a highly purified protein
may be obtained.
F. Receptors, Binding and Transport Proteins, Hormones,

Vitamins, Toxins, Growth Factors, Lipids, and Other
Substances
The last decades saw a dramatic increase in the number of analytical problems
demanding the determination of trace levels of analytes in complex samples.
This can be accomplished by selective sample preparation steps, which eliminates
interfering matrix and enriches the analyte to a concentration surpassing the deter-
mination limit of the detection method. The introduction of monoclonal antibod-
ies against receptors and their subunits as affinity ligands rather than receptor
substrates greatly advanced the application of immunology to membrane bio-
chemistry. A discussion of monoclonal antibodies against acetylcholine, insulin,
adrenergic, transfenin, thyrotropin, neurological, viral. and complement re-
ceptors was presented by Phillips [25]. Immunoaffinity chromatography of
polycyclic aromatic hydrocarbons in columns prepared by sol-gel method was
described by Cichna et al. [26]. The paper describes the synthesis of pyrene-
selective immunoaffinity columns and their application for the determination of
pyrene in water. The immunoaffinity columns prepared had a pyrene break-
through capacity of 78 nglg and 38 nglmg IgG. Corticosteroids dexamethasone
and betamethasone were extracted from bovine urine by use of immunoaffinity
chromatography by Bagnati et al. [27]. The described extraction and purification
114 Turkova
system for the purification of corticosteroids is the combination of immunoaffin-

ity and HPLC columns.
Receptor affinity chromatography based on the specificity and reversibility
of the receptor-ligand interaction for the purification of biomolecules by use of
matrix-bound receptor was described by Weber and Bailon [28]. As a model
system they described the purification of recombinant human interleukin-2 from
microbial and mammalian sources using the soluble subunit of the human in-
terleukin-2 receptor attached to silica-based NuGel P-AF polyaldehyde po1y-N-
hydroxysuccinimide. A comparison of receptor and immunoaffinity purification
methods showed that the binding capacity of the immobilized receptor is higher
than that of the immunosorbent and the receptor affinity-purified interleukin-2
was more homogeneous than the immunoafhity-purified material.
The primary effect of some hormones is aimed at the plasma membrane
of the target cells. The amount of their receptors in the tissues is very small in
comparison with other material present. The interaction of such a small amount
with the immobilized hormones must be very effective in order to permit a strong
binding of large membrane fragments. The isolation of peptide or protein recep-
tors posed a number of difficult problems owing to their scarcity, their location
in a complex lipid-rich environment, and the difficulty with which even a starting
plasma membrane is prepared. The coupling of peptide hormones to activated
Sepharose results in nonspecific attachment that can generate a number of differ-
ent species with varying affinities for the desired receptor. For these reasons Finn
and Hofmann [29] have devoted their efforts to developing biospecific sorbent
based on avidin-biotin technology where the hormone can be attached in a tar-
geted fashion. The application of the avidin-biotin systems to the isolation of
hormone receptors is shown in Section 1II.G.
G. Biotin and Avidin or Streptavidin
The binding of water-soluble vitamin biotin to the egg-white protein, avidin, or
to its bacterial counterpart, streptavidin, is accompanied by a vast decrease in
free energy compared to that observed for other noncovalent interactions. The
change of enthalpy, AH, for biotin bound to avidin and streptavidin has been
determined by Green [30] as being -86 and -98.9 kJ/mol biotin, respectively.
Each avidin or streptavidin molecule can bind four molecules of biotin. The
change in entropy for this reaction is essentialy zero. The affinity constant for
biotin-avidin determined by Green is approximately l O I 5 M-' .
In the past decade the interaction between biotin and avidin or streptavidin
has provided the basis for establishing a new avidin-biotin technology. The basic
concept is that biotin, coupled to either low or high molecular weight molecules,
is recognized by avidin. Methods for the biotinylation of membranes, nucleic
acids, antibodies, and other proteins have been developed in many laboratories.
Figure 8 shows biotin and various derivatives, suitable for various types of bonds.
-co-NH-NH, - Sugars. w l e i c acids
-CO-NN-CO-WBr ---,Mxleophiles
Figure 8 Biotin and derivatives.
However, a spacer arm should be used when proteins are modified by biotin.
Detailed information on biotin-binding proteins, the preparation of biotin, avidin,
and streptavidin derivatives, assays for avidin and biotin, and applications is
given in Volume 184 of Methods in Enzymology titled Avidin-Biotin Technology,
edited by Wilchek and Bayer [31].
The use of an avidin-biotin complex in bioaffinity chromatography was
the isolation of avidin from egg white on a biocytin-Sepharose column [32]. The
conditions required for its dissociation were extremely drastic requiring elution
by 6 M guanidinium hydrochloride, pH 1.5. In order to decrease the strong affinity
of avidin for biotin, Finn and Hofmann [29] used immobilized succinylated avidin
for the isolation of hormone receptor. Soluble receptor (R) is percolated through
a bioaffinity column to form the complex shown in the center of Fig. 9. The
column is then exhaustively washed to remove contaminating materials. Alterna-
tively, the biotinylated hormone ligand (B-H) can be added to a solution of solubi-
lized receptor to form a soluble complex BHR. Applying a solution containing
this complex to a column of immobilized succinylated avidin (SA) will result in
the formation of the same complex (Fig. 9B). Both of these schemes have been
used for the isolation of insulin receptors from human placenta. Removal of active
insulin receptor can be achieved by eluting the column with acetate buffer, pH
5 containing either 1 M NaCl or biotin.
116 Turkova
Figure 9 Application of the avidin-biotin system to the isolation of hormone receptors.

Dashed lines represent noricovalent bonds. (From Ref. 28.)
Despite the fact Lhal streptavidin is currently about 100 times more expen-
sive than avidin, its use is sometimes justified since immobilized streplavidin
exhibits less nonspecific binding. Avidin is highly positively charged at neutral
pH, with an isoelectric point above 10. Consequently, it binds negatively charged
molecules such as nucleic acids, acid proteins, or phospholipids in a nonspecific
manner. This can result i n nonspecific staining of, for example, the nucleus and
cell membranes. Avidjn is also a glycoprotein and therefore interacts with other
biological molecules such as lectins or other sugar-binding materials via its carbo-
hydrate moiety. The advantage of using streptavidin lies in the fact that it is a
neutral, nonglycosylated protein (pl lower than 7). To decrease its positive
charge, the lysines of avidin can be derivatized by succinylation, acetylation, etc.
A variety of avidin derivatives with average pl values of 7 or lower are now
commercially available. However, the ren~ovalof the carbohydrate residue from
avidin is much more difficult.
One method that can be generalized to link virtually any DNA substrate
by its 3' or 5' end to a chromatography matrix via streptavidin-biotin linkage has
been described by Fishel et al. [33]. Biotin-slreptavidin affinity selection as a
valuable tool permitting the analysis of the RNA components of splicing com-
plexes assembled on a wide variety of pre-mRNA substrates has been reviewed
by Grabowski [34]. A review of the isolation of cell surface glycoproteins by
the use of biotinylated lectin was published by Cook and Buckie [35]. They
demonstrated the use of immobilized streptavidin to obviate dissociation of com-
plexes formed between biotinylated concanavalin A and membrane glycopro-

teins.
The advantage of the biotinylation of antibody carbohydrate moieties using
biotin hydrazine for use in immunoaffinity chromatography has been described
by Phillips [25]. HPIAC (or HPIC) of specific phosphorylcholine receptors can
be employed using biotinylated mouse monoclonal antiidiotypic antibody ad-
sorbed to streptavidin-coated glass beads. Using this technique HPIAC analysis
was performed on isolated membranes from primed T cells obtained from mice
immunized with phosphorylcholine (PC). These membranes were prepared by
hypotonic lysis of the cells, followed by detergent NP40 solubilization of the
membrane-rich fraction obtained by ultracentrifugation. The preparation was
passed through a Sephadex G25 column to remove excess detergent prior to injec-
tion into the HPIAC column. Using monoclonal antiidiotype antibody as the im-
mobilized ligand, a single peak was eluted by sodium thiocyanate gradient at 24
min. Rucklidge et al. [36] used antibody against denatured collagen after adsorp-
tion to membrane with attached collagen for biotinylation. Biotinylation in situ
may protect the variable region of the IgG compared to antibodies biotinylated
in solution, thus increasing their antigen recognition.
H. Cells and Viruses

The bioaffinity chromatography of cells, cell organelles and membranes, phages
and viruses was reviewed by Sharma and Mahendroo [37]. Affinity ligands, in
their review, are divided into two types: ( I ) lectins and nonlectin ligands, which
contain hormones, neurotransmitters, and related ligands with affinity for cell
surface receptors, and (2) further inhibitors of membrane-bound enzymes. A re-
view on the use of immobilized glycoconjugates for cell recognition studies was
published by Schnaar [38]. In this review the difference in carbohydrate recogni-
tion by chicken and rat hepatocytes was demonstrated. Hepatocytes dissociated
from chicken livers bound readily to immobilized N-acetylglucosamine, but not
to other carbohydrate-derivatized surfaces, whereas rat hepatocy tes bound to ga-
lactose-derivatized surfaces. The rapid isolation of human erythrocyte plasma
membrane was achieved by Kaplan et al. [39] on an affinity matrix consisting
of wheat germ agglutinin covalently bound to Sepharose 6MB. After binding the
washed cells to the bioaffinity matrix, they were washed extensively and lysed.
The resulting ghosts were washed and then eluted from the matrix with N-acetyl-
P-D-glucosamine.
Bioaffinity chromatographic separation of T cells by the use of monoclonal
antibodies was discussed by Braun and Kiimel [40] as a valuable tool in studies
of the function of T cells and their subsets. According to these authors a broad
array of procedures has been developed, of which an indirect use of matrix-bound
second antibody appears to be the most practical and advantageous with respect
118 Turkova
to purity, functional activity, and viability of separated cells. The most frequently
employed affinants for the isolations of viruses are bound antibodies. An example
is the isolation of Alleutian disease virus from chronically infected mink [41] by
use of Sepharose with coupled IgG. In some cases rather than immobilized lec-
tins, antibodies or Protein A, other specific and reversible cell surface receptor-
ligand interactions can be successfully utilized. As an example one may mention
the isolation of acetylcholine receptor-bearing neuronal cells from chick embryo
sympathetic ganglia by the use of snake venom a-bungarotoxin immobilized on
Sepharose 6MB macrobeads [42]. Information on the use of mentioned and many
other affinants is given in Ref. 7.
IV. SOLID MATRIX SLIPPORTS

A. Required Characteristics
One of the most important factors in the development of bioaffinity chromatogra-
phy and immobilized enzymes is the development of solid supports. A correct
choice of solid support and the covalent coupling between the matrix and the
affinity ligand may be essential for the success of the desired bioaffinity chro-
matographic separation. Solid matrix supports can also have a considerable effect
on the stability of the immobilized affinity ligand and the adsorbed material. A
solid support may even be an affinity ligand itself, e.g., polysaccharides for some
lectins. An increasing number of different types of supports, their activated forms
and biospecific matrices prepared from them are now becoming commercially
available as ready-to-use adsorbents. In the scope of this chapter only basic solid
supports are described and briefly discussed.
An ideal matrix for successful application in bioaffinity chromatography
and for the immobilization of enzymes should possess the following properties
[43]:
1. Insolubility
2. Sufficient permeability and a large specific area
3. High rigidity and a suitable form of particles
4. Zero adsorption capacity
5. Chemical reactivity permitting the introduction of affinity ligands
6. Chemical stability under the conditions required for the attachment,
adsorption, desorption, and regeneration
7. Resistance to microbial and enzymatic attack
8. Hydrophilic character
Complete insolubility is essential not only for the prevention of losses of
affinity adsorbent but for prevention of contamination of the substance being
isolated by dissolved camer. However, a universal solid support that fulfills all
of the described requirements does not exist. When a certain affinity ligand is
immobilized an individual solid support and method of coupling must be selected
with respect to its future use.
Application of bioaffinity chromatography resulting from a changeover
from soft gel supports to small rigid particles used in high-performance liquid
bioaffinity chromatography (HPLBAC) has been reviewed by Ohlson et al. [44].
A comparison of HPLBAC with soft gel bioaffinity chromatography (BAC) is
shown in Fig. 10. Mechanically stable, rigid particles, with small and uniform
sizes, provide high flow rates with good mass transfer characteristics, giving over-
all high operational adsorption capacity. Much more favorable mass transport
and adsorption/desorption kinetic behavior with nonporous support has been
demonstrated by Anspach et al. [45]. The elution behavior of proteins on a nonpo-
rous silica-based adsorbent (with an average diameter of 1.4 pm) was investigated
both theoretically and experimentally using human immunoglobulin G and immo-
bilized protein A as the affinity pair by Lee and Chuang [17]. The desorption
rate constant and equilibrium association constant under elution conditions were
found to decrease elution time and improve the shape of the elution peak. How-
ever, the adsorption rate in column chromatography is limited by either slow
intraparticle diffusion for large beads or low axial velocities and high-pressure
Soft-gel BAC HPLBAC

Particle
0.1 m m soft 0.01 m m h a r d - oorous
0
0.001 m m h a r d -nonporous
Speed and resolution
hours m i nut es
Figure 10 A comparison of bioaffinity chromatographies using soft gel or porous and

nonporous small hard particles.
120 Turkova
drops for small beads. To overcome these limitations, a membrane-based bioad-

sorbent has been used for the isolation of human pepsin by Kucerovi and Turkovi
[461.
B. Biopolymers
Water molecules are essential for the structure and function of biologically active
compounds. BAC is therefore generally performed in the aqueous phase and hy-
drophilic biopolymers are usually employed as solid supports, chiefly natural
polysaccharides, such as agarose, dextran, cellulose, and, to a lesser extent, starch.
The modification necessary for the BAC can be carried out in a relatively simply
via their OH groups.
1. Agarose and Its Derivatives

Agarose is a linear polysaccharide consisting of alternating 1,3-linked P-D-galac-
topyranose and 1,4-linked 3,6-anhydro-a-L-galactopyranoseresidues. Arnott et
al. [47] postulated on the basis of x-ray studies that the polysaccharide chains
form a double helix, then aggregate via hydrogen bridges and hydrophobic inter-
actions into fibers or bundles with ordered structures. This network phase in a
gel that may contain up to 100 times more water than agarose means that the
structure contains relatively large voids through which large macromolecules can
diffuse. In contradistinction, a gel network comprising a comparable concentra-
tion of crosslinked soluble polymer, such as the crosslinked dextrans, would lead
to a lattice in which the mean pore size would be considerably smaller. These
relationships are shown diagrammatically in Fig. 11 and suggest that agarose has
particularly useful special properties as a chromatographic medium. The exclu-
sion limit may be varied within wide ranges because the pore size is inversely
proportional to the agarose concentration.
The main producers of agarose are Pharmacia Biotechnology (Uppsala,
Sweden), under the trade name Sepharose; Bio-Rad Laboratories (Richmond,
CA, USA), under the trade name Bio-Gel A; Reactifs IBF (Villeneuve la Gare-
nne, France), under the trade name Ultrogel A; and agarose gels under the name
SAG (Ago-Gel) supplied by Seravac Labs. (Maidenhead, Great Britain) and
Mann Labs. (New York).
Supports with large beads are advantageous for the affinity chromatography
of cells, so that they can pass through such columns without being physically
trapped. For this purpose an agarose gel has been developed with the trade name
Sepharose 6MB. The rnacrobeads have a large diameter (250-350 pm), uniform
shape, and low nonspecific adsorption of cells. The stability of an agarose matrix
can be considerably increased by crosslinking with epichlorohydrin, 2,3-dibro-
mopropanol, or divinylsulfone. In 1975, Sepharose CL (2B, 4B, 6B) was intro-
Figure 11 A comparison between an agarose gel (Sepharose) matrix (right) with a

crosslinked dextran (Sephadex) matrix (left) at equivalent polymer concentration. The ag-
gregates in agarose gels may contain 10-10' bundles of polysaccharide helices. (From
Ref. 47.)
duced, prepared from appropriate types of Sepharoses by crosslinking with

2,3-dibromopropanol in strongly alkaline medium, and further desulfurization by
alkaline hydrolysis under reducing conditions. The crosslinking of Sepharose
CL does not decrease the effective pore size, thus suggesting that crosslinks take
place mainly between chains within a single gel fiber. The structure of agarose
makes it inadvisable to dry and reswell the gels. When agarose is not in use it
should be stored in the wet or moist state and protected from microbial growth
by means of a suitable bacteriostat. The antimicrobial agent that is commonly
used is 0.02% sodium azide. In general, when agarose gels are stored for long
periods, the temperature should be below 8°C in the presence of a suitable bacteri-
ostat but without freezing. Freezing results in irreversible structural disruption
of the gel beads. Freeze-drying can be carried out only after the addition of protec-
tive substances, e.g., 15% lactose.
Gustavsson and Larsson [48] described the preparation of superporous
agarose, which combines the desirable propertes of traditional agarose supports
and a chromatography support for high-performance separations. Pharmacia Bio-
technology produces a highly crosslinked agarose matrix, resulting in a very rigid
gel, under the trade name Superose 6B. Both the crosslinking and the narrow
122 Turkova
particle size (20-40 pm) contribute to its performance as a chromatography sup-

port for HPLC separations. The new name for Sepharose high-performance bioaf-
finity columns are Hi-Trap Column, which have a wide range of life science
applications.
Reactifs IBF produces Magnogel A4R, which is a support composed of
agarose (4%, wlv) crosslinked with epichlorohydrin. Its magnetic nature results
from the incorporation of 7% (wlv) Fe30, in the interior of the gel beads. It has
applications in situations where column operation is not favored, e.g., in viscous
solutions or in the presence of insoluble particles such as cell debris.
The polysaccharide backbone of agarose can readily undergo substitution
reactions to yield products with a moderately high capacity for further derivatiza-
tion. Many types of agarose activated for the attachment of biologically active
compounds and a variety of affinity sorbents are available from many firms, e.g.,
Sigma (St. Louis, MO, USA). Derivatives of crosslinked agarose modified for
use in bioaffinity chromatography are also produced by Bio-Rad Laboratories un-
der the trade name Affi-Gel.
2. Dextran Gels
Dextran is a branched-chain glucose polysaccharide produced in solutions con-
taining sugar by various strains of Leuconostoc mesenteroides. Soluble dextran,
prepared by fractional precipitation with ethanol of partially hydrolyzed crude
dextran, contains more than 90% of a-1,6-glucosidic linkages with 1,2-, 1,3-,
and 1,4-glucoside branching. When crosslinked with 1-chloro-2,3-epoxypropane
in alkaline solution, dextran becomes suitable for chromatography.
The most important producer of dextran gels, supplied under the trade name
Sephadex, is Pharmacia Biotechnology. The gels are very stable to chemical at-
tack. The glucosidic bond is sensitive to hydrolysis at low pH, although it is stable
for 6 months in 0.02 M hydrochloric acid, or for 1-2 h in 0.1 M hydrochloric acid
or 88% formic acid. Aldehyde or carboxyl groups are formed under the effect
of oxidizing agents. Dextran gels can withstand heating in an autoclave at 110°C
(in solution) for 40 min, or at 120°C when dry. Drying and swelling is reversible.
The gels swell to some extent even in ethanol, ethylene glycol, formamide, N,N-
dimethylformamide, and dimethylsulfoxide.
A molecular sieve produced by covalent crosslinking of allyldextran with
N,Nf-methylenebisacrylamidehas been developed by Pharmacia under the trade
name Sephacryl. The advantage of matrix is the good flow rate that results be-
cause the support is exceptionally rigid. Unfortunately, the nonspecific adsorption
is increased over that of other dextran gels. The use of dextran gels is partly
restricted by their rather low porosity. They are widely used without any modifi-
cation as specific sorbents for the isolation of a series of lectins. One example
is the isolation of concanavalin A from jack bean seeds [49].
3. Cellulose and Its Derivatives

Cellulose is fornled by linear polymers of 0-1 ,Clinked D-glucose units with an
occasional 1,6 bond. Commercially available celluloses are generally crosslinked
with bifunctional reagents, such as 1-chloro-2,3-epoxypropane,and they are very
stable to chemical attack. Glycosidic bonds are sensitive to acid hydrolysis, and
under extreme conditions an almost quantitative decomposition to pure crystal-
line D-glucose may take place. On interaction with oxidative reagents, such as
sodium periodate, aldehyde, and carboxyl groups are formed. Cellulose can be
attacked, e.g., by microbial cellulases.
Macroporous-regenerated cellulose in regular beaded form is produced un-
der the trademark Perloza by Lovochemie A.S. (Lovosice, Czech Republic).
Beaded cellulose and its derivatives, in comparison with other biopolymer spheri-
cal materials, show good mechanical strength and resistance to shape deformation
and therefore provide better through-flow layers in columns. Good mechanical
strength is preserved in spite of the high porosity. Due to its chemical composition
it is highly hydrophilic and is well tolerated by biosystems. Its insolubility in
water and a number of other solvents allows it to be used in aqueous media.
Biospecific adsorbents prepared from Perloza was used by Bilkova et al. [5].
Medium-pressure Matrex Cellufine gels, composed of spherical beaded cellulose,
are supplied by Amicon Co. (Danvers, MA, USA). They offer low nonspecific
adsorption, outstanding physical strength, and high-pressure operating capabili-
ties-all at an economic cost. Adsorption of lysozyme on dye affinity sorbent
prepared by use of Cellufine was described by Anspach et al. [50]. The use of
fibrous cellulose particles as matrices for DNA-bearing supports is shown by
Hermanson et al. [51] in their work on immobilized nucleic acids. The isolation
of poly-mRNA from tumor cells by use of an oligo(dT)-cellulose slurry in an
Eppendorf Event 4160 vacuum filtration unit published was described by Noppen
et al. [52]. This method can be used for mRNA determination or purification in
diagnostics as well as biological and medical research. Cellulose and its deriva-
tives are produced by a number of firms (e.g., Whatman, Maidstone, Great Brit-
ain; Schleicher and Schull, Zurich, Switzerland; Serva, Heidelberg, Germany;
Bio-Rad, Richmond, CA, USA).
C. Synthetic Copolymers
The advantages of HPLBAC, shown in Fig. 10, and the usefulness of the prepara-
tion of biologically active compounds on a pilot or industrial scale are the impetus
behind the continual development of synthetic polymers. The main appeal of
solid supports for these applications is their inherent mechanical stability, which
provides good flow characteristics even under high pressures. They can be oper-
ated at pressures up to 100 psi and usually tolerate a wide pH range. They are
124 Turkova
suitable for affinity ligand immobilization and provide bioaffinity supports with
high capacities. They are biologically inert and because of their inert structure
are not subject to enzymatic or microbial degradation. The chemical Structure of
these supports can be characterized by their polyethylene backbone, which creates
excellent chemical and physical stability. They also contain modifiable side
chains R , , R2, R3, R4:
Only a limited number of synthetic polymers will be characterized. However,

many synthetic copolymers exist, and their intermediates for bioaffinity chroma-
tography and biospecific sorbents prepared from them are produced by many
firms.
1. Acrylamide Derivatives
Polyacrylamide gels are produced by copolymerization of acrylamide with the
bifunctional crosslinking agent N,N1-methylenebisacrylamide.The monomers
used in this synthesis are highly toxic and thus should be handled with care. The
main producer of polyacrylamide gels is Bio-Rad under the trade name Bio-Gel
P. This gel is produced with a range of pore sizes, from Bio-Gel P-2 with a
molecular weight exclusion limit of 1800, to Bio-Gel P-300 with a molecular
weight exclusion limit of 400,000. Commercial polyacrylamide beads are pur-
chased in the dry state and are swollen by mixing with water or aqueous solutions
for period of 4-48 h depending on the porosity. Bio-Gel P products are stable
to most eluants used in biochemical studies, including dilute solutions of salts,
detergents, urea, and guanidine hydrochloride, although high concentrations of
these reagents may alter the exclusion limits by up to 10%. The use of media
with pH values outside the range 2- 10 is to be avoided since some hydrolysis of
the amide side groups may occur with the consequent appearance of ion exchange
groups. Polyacrylamide gels are biologically inert and are not attacked by micro-
organisms.
Nonionic synthetic supports are obtained by copolymerization of N-acry-
loyl-2-amino-2-hydroxymethyl-1,3-propane diol with hydroxylated acrylic bi-
functional monomer. Hydrophilic supports under the trade name Trisacryl G F
are manufactured by IBF Reactifs. By strictly controlling the polymerization pro-
cess, it is possible to synthesize a complete line of products with covering a
wide range of molecular weights from 3000 (Trisacryl GF05) to about 20 million
(Trisacryl GF 2000). Trisacryl is characterized by a high degree of hydrophilicity,
due both to the presence of primary alcohol groups and also the secondary amide
function. It can be used under pressures up to 2-3 bar and is not affected by
organic solvents such as alcohols, ketones. dioxane, or chlorinated solvents. It
is stable to low (-20°C) and high (121 "C) temperatures. Denaturing agents have
no effect on the gel because its structure involves no hydrogen bonds. It is also
stable to acidic pH, but less stable to high pH because of the slow hydrolysis of the
amide linkage. Adsorption of lysozyme on the Trisacryl GF 2000 with attached
Cibacron Blue F3GA in comparison with dye affinity sorbents prepared by use
of other solid supports was studied by Anspach et al. [50].
2. Methacrylate Supports
The copolymerization of hydroxyalkyl methacrylate with alkylene dimethacry-
lates gives rise to heavily crosslinked xerogel microparticles which subsequently
aggregate and yield macroporous spheroids. Their structure is shown in Fig. 12.
These gels have some chemical properties in common with agarose [53]. Hydroxy-
alkyl methacrylate supports have been marketed under the trade name Separon
HEMA and is produced by Tessek S.R.O. (Prague, Czech Republic), or under the
trade name Spheron was produced by Lachema (Brno, Czech Republic). These
supports have good chemical and mechanical stability and are stable to heating
for 8 h in 1 M sodium glycolate at 150°C or boiling in 20% hydrochloric acid
for 24 h. They are biologically inert and are not attacked by microorganisms.
They can be employed in organic solvents. HPLBAC of porcine pepsin on Sep-
aron HlOOO modified with E-aminocaproyl-L-phenylalanyl-D-phenylalanine
methyl ester has been described by Turkovi et al. [54].
Hydroxyethyl methacrylate support is also produced under the trade name
Dynospheres by Dyno Particles. Their monodisperse microparticles with a size
range of 0.3-5 pm are promising synthetic polymers for HPLBAC application.
The Japanese firm Toyo Soda Manufacturing Co. (Tokyo, Japan) has devel-
oped a semirigid gel, a copolymer of oligoethylene glycol, glycidyl methacrylate,
and pentaerythrol dimethacrylate under the trade name Toyopearl. The identical
copolymer under the name Fractogel TSK HW type is sold by E. Merck (Darm
stadt, Germany). Optimal conditions for the activation of free hydroxyl groups
on the gel matrix by epichlorohydrin and subsequent immobilization of ligands
were investigated by Matsumoto et al. 1551. They successfully prepared affinity
adsorbents for bioaffinity chromatography of lectins and trypsin. The advantage
of this gel is its pressure stability up to 7 bar. Swelling of dry gels in water is
3-4 mllg for Fractogel TSK HW-65. Its molecular exclusion limit is 5 X 10"
for proteins and lo6 for polyethylene glycols or dextrans. The negligible change
126 Turkova
I
C0
I
C C H ~ C H ~ O ~ IE H 2 C H 2 0 H
-
MACROPORE
MACROSPHEROID
Figure 12 Structure of hydroxyalkyl methacrylate copolymer (Separon HEMA,

Spheron).
in swelling volume with changing eluents, results i n a very constant gel bed
volume. It may be used from pH 1 to 14. The high chemical stability has led to
applications at high temperature and therefore it may be autoclaved. Its properties
render Fractogel TSK particularly suitable for large-scale industrial application.
Oxirane acrylic beads are obtained by copolymerization of methacrylam-
ide, methylenebismethacrylamide, and allylglycidylether. Due to the nature of
monomers the copolymer has neutral and mostly hydrophilic matrix, with a slight
hydrophobic component due to the methyl groups along the polymer backbone.
The oxirane group content is 1000 pmollg dry beads. The beads are macroporous
and they show a water regain of 2.5 mllg of dry beads. This water regain is
independent of pH (0.5- 12.5) and ionic strength. The beads are morphologically
and chemically stable under these conditions, even if exposed to them for several
weeks. The mechanical stability upon stirring is very good. At present oxirane
acrylic beads are produced under the trademark Epergit by Rohm Pharma GMBH
(Darmstadt, Germany). HPIAC using Eupergit C beads was developed and opti-
mized by Fleminger et al. [56].
D. Polystyrene and Its Derivatives

Polymer-coated polystyrene/divinylbenzene beads under the name Poros matrix
have been developed by PerSeptive Biosystems (Cambridge, MA, USA). The
backbone of the matrix is unique in that it contains a network of large and small
pores within each spherical bead [51]. The Poros matrix structure consist of large
"through-pores" that allow rapid flow into the interieor of the bead and a network
of "diffusive pores" that gives the support good capacity for affinity application.
The polymer coating provides active sites for further modification and also blocks
the harsh hydrophobic character of the styrene core. Depending on the particular
application, PerSeptive has coated beads with crosslinked polyethyleneimine as
well as a proprietary polyhydroxylic polymer. Since the base matrix is a stable
crosslinked polystyrene, the chemical and physical nature of the support is excep-
tionally robust. The media is resistant to extremes in pH (1 - 14) and is compatible
with all common buffers and solvents used in HPLC. It can also withstand 0.5
N NaOH and 1.0 N HCI for cleaning and sterilizing purposes. The physical stabil-
ity of the matrix is reflected by its maximum pressure limit of 3000 psi.
Superparamagnetic polystyrene Dynabeads, consisting of Fe203-containing
core covered with a polymer, are produced by the firm Dynal A S . (Oslo, Nor-
way). They have a smooth surface that is easily coated with antibodies or other
selecting molecules. Combined with a magnet, Dynabeads make a unique tool
in positive or negative separation. Monosize magnetic particles in selective cell
separation was used by Ugelstad et al. [57] for the successful clinical application
of immunomagnetic beads for depletion of tumor cells or T lymphocytes from
bone marrow.
E. Combination of Biopolymers with Synthetic Polymers

In order to minimize the reduction of polyacrylamide porosity during activation,
copolymers of agarose and polyacrylamide have been produced. This support
combines the advantages of each individual polymer, while extending the poten-
tial range of derivatization procedures by virtue of the availability of both amide
and hydroxyl groups for activation. Polyacrylamide-agarose gels with varying
porosities are produced under the name Ultrogels AcA by IBF Reactifs. A sche-
matic representation of the Ultrogel AcA matrix is shown in Fig. 13. Ultrogel
AcA is available in four types, each composed of a three-dimensional polyacryl-
amide lattice enclosing an interstitial agarose gel. The gels are preswollen and
calibrated within a narrow size range of 60-140 km.
Turkova
Figure 13 Schematic representation of the Ultrogel AcA matrix.
During chemical reactions involving one ol' the polymers, certain precau-
tions must be taken in view of the other polymer properties. Thus the gel must
not be exposed to strongly alkaline media, since the amide groups will be hy-
drolyzed to carboxylic acids. The limited heat resistance of the agarose must
equally be respected: Ultrogel AcA must not be exposed to temperatures greater
than a b o ~ 40°C.
~t The preparation of po.lyacrylhydrazidoagarose based on perio-
date oxidation of Sepharose followed by reaction with polyacrylhydrazide has
been described by Miron and Wilchek [58]. These yielded matrices which were
colorless and stable after reduction with sodium borohydride. Polyacrylhydrazi-
doagarose could be used either directly or after further modification with various
additional reactive groups.
F. Inorganic Support
Inorganic supports have been reviewed by Weetall and Lee 1591. They classif ed
these supports into a few major categories: metals, metal oxides, ceramics, and
glasses. They described the preparations, properties, and applications of porous
glass, porous silica, titania, alumina, and zirconia bodies, and iron and nickel
oxides. These supports have an inherent advantage in their rigidity. Since inor-
ganic particles are friable materials, one should never use a stirring bar when
working with these materials. Particles can be separated for washing or assay
by several convenient methods. These include filtration, centrifugation, settling,
aspiration, or magnetic separation. Several of these methods can be utilized for
sizing inorganic particles, particularly when clumping has occurred or when one
wants particles of only a specific size range. Particles may be sized on the basis
of settling times, centrifugation speed, or filter porosity. The best method usually
determined by considering the size range that is desired and choosing the method
most likely to yield the desired particle range.
1. Controlled Pore Glass

Controlled pore glass (CPG) is synthesized by heating certain borosilicate glasses
to 500-800°C for prolonged periods of time. These glass mixtures separate on
such heat treatment into borate- and silicate-rich phases. The borate phase can
be dissolved by treatment with acid, leaving a network of extremely small tunnels
with pore diameters of 3-6 nm. Subsequent treatment with mild caustic soda
removes silica material from the pore interiors and thus enlarges the pore diame-
ter. Careful control of the various treatments can lead to a porous glass in the
range 4.5-2.50 nm. Glass derivatives have outstanding mechanical stability. The
rigidity of the beads permits high flow rates and facilitates fast and efficient sepa-
rations. Glass beads are resistant to microbial attack and may be readily sterilized
by disinfectants or autoclaving. The latter is a prime consideration in the purifica-
tion of pyrogen-free enzymes destined for in vivo or clinical studies. Commer-
cially available porous glass packing materials are produced, among others, by
Bio-Rad Labs. (Richmond, CA, USA) under the trade name Bio-Glass, by Corn-
ing Glass Works (Corning, NY, USA), Waters Assoc. (Milford, MA, USA). Elec-
tro-Nucleonics, Inc. (Fairfield, NY, USA), and Pierce Chemical Company (Rock-
ford, IL, USA) under the name CPG.
Glyceryl-CPG is a controlled pore glass whose surface has been chemically
modified to produce a hydrophilic, nonionic coating that shares most of the same
operating characteristics as conventional CPG. Glycerolpropyl glass was the
weakest adsorber of the protein. Glyceryl-CPG is distributed by Electro-Nucleon-
ics Inc. (Fairfield, NY, USA). Another supplier of glycophase-coated supports
under the name glycophase G/CPG is the Pierce Chemical Company (Rockford.
IL, USA). This company uses triethoxypropyl glycidosilane as the alkylsilane.
Nonporous glass beads, porous silicas, and soft gels coupled to Cibacron
Blue F3GA were employed by Anspach et al. [50] to investigate equilibration
times for the adsorption of lysozyme on the different dye affinity sorbents. In
the batch mode equilibration times varied from 20 s for nonporous glass beads
(20-30 pm) to more than 60 min in the case of a porous sorbent with a particle
diameter of 100-300 pm and 60 nm pore size.
130 Turkova
2. Porous and Nonporous Silica

The structure of silica is amorphous and its composition can be expressed as
Si02 . H20. Its basic unit is tetrahedral (SiO,) and its porosity depends on the
mode of preparation. Silica gel is formed from silicic acid sols by polycondensa-
tion of orthosilicic acid. Submicroscopic elemental particles are formed that re-
tain micelles of the starting acid and in the interior. The silica is bound by siloxane
Si-0-Si bonds. Every elemental particle touches the surfaces of several neigh-
boring particles and in this way a conglomerate is formed containing pores of
various diameters. The density of these inner spaces is very high and after drying
they contain large specific inner surface areas of hundreds of m2/g. The main
advantage of silica is its inherent mechanical stability which provides good flow
characteristics even under high pressure. The use of silica with pore sizes ranging
from 6 to 400 nm, or nonporous silica in small particles 1.5-10 pm, is advisable
to provide good mass transfer when performing protein separations. Silica is a
commonly used support for high-performance bioaffinity chromatography.
Slightly acidic silanol groups on the silica surface act as centers for nonspecific
adsorption. Under alkaline working conditions (pH > 8) the surface of silica gel
exhibits a high solubility that could be the reason of the contamination of the
purified product. The problems of nonspecific adsorption and the solubility of
silica have been largely overcome by derivatization of the silanol groups with
silanes, yielding silica derivatives coated with a hydrophilic layer. The most com-
mon reagent for blocking silanol groups is 3-glycidoxypropyltrimethoxysilane.
The resulting epoxide silica can be used for the direct attachment of affinity
ligands or can become the starting material for the synthesis of other products.
An example is the hydrazide-activated silica supports for HPLBAC developed
by Ruhn et al. [60]. They prepared diol-bonded silica from Nucleosil Si-300 or
1000 coated with 3-glycidoxypropyltrimethoxysilane by hydrolysis with sulfuric
acid. After periodate oxidation of the diol-silica, an aldehyde silica was produced
and used for the optimization of hydrazide-activated silica synthesis by use of
oxalic or adipic dihydrazide.
The improvement of affinity chromatographic performance using adsor-
bents prepared from nonporous monodisperse silicas in contrast to that obtained
subjected to porous silica supports with identical activation and immobilization
procedures was described by Anspach et al. [50]. The elution of proteins on
nonporous silica-based adsorbents was investigated both theoretically and experi-
mentally using human immunoglobulin G and immobilized protein A as the af-
finity pair by Lee and Chung [17]. Nonporous silica (average diameter 1.4 pm)
was silanized with y-aminopropyltriethoxysilane and activated with glutaralde-
hyde. A comparison was made between predicted and experimental elution peaks
from chromatography of IgG on immobilized protein A. The desorption rate con-
stant and equilibrium association constant under elution conditions were found
to have a substantial effect on elution time and peak shape. Many silica gel sup-
ports are available commercially in both irregular or spherical shapes, either un-
coated or glycophase-coated. Among them are, for example, silica-based supports
under the trademark LiChrospher Si, LiChrospher Diol, etc. (E. Merck, Darm-
stadt, Germany), Porasil A-F (Waters Associates, Milford, MA, USA), Progel-
TSK columns (Supelco, Bellefonte, PA, USA). Ultraaffinity-EP is one of the
commercially available epoxide silicas available in prepacked columns (Beck-
mann Instruments, Berkeley, CA, USA). Immobilization of the ligand onto such
a matrix can be performed by passing the material through the column slowly,
over a predetermined time period. The use of antichaotropic anions such as phos-
phates and sulfates is recommended because they stabilize the protein during
derivatization and increase their interaction with the activated support. Columns,
cartridges, or kits of a silica activated with epoxide functional bonded phases are
sold under the name Durasphere AS by Alltech Associates (Deerfield, IL, USA).
3. Iron and Nickel Oxides

Iron oxide can be prepared by several methods. One method involves the use of
commercially available iron oxide powders. These can be silanized directly and
used for the attachment of affinity ligands [59]. An even dispersion of magnetic
material (y-Fe203and Fe304)throughout the bead is in Dynabeads, which are
produced by Dynal A.S. This firm supplies uncoated Dynabeads, activated Dyna-
beads, and Dynabeads precoated with specific ligands. Their characterization is
that they are the only uniform, superparamagnetic, monodisperse polymer parti-
cles. Information about them is given in section 1V.C. Commercially available
particles consisting of an iron oxide core coated with a silane polymer terminating
in amine (particle size 1-2 pm), carboxyl (particle size 0.5-1 pm), or sulfydryl
(particle size 1-3 pm) functional groups are supplied by Advanced Magnetics
Corp. (Cambridge, MA, USA). Latex particles impregnated with iron oxide are
also commercially available from Seradyn Inc. (Indianapolis, IN, USA). These
materials show paramagnetic properties and have been successfully used in im-
munoassays. Nickel oxides produced by precipitation or directly supplied by a
chemical manufacturer have been used in a manner similar to the iron oxides [59].
G. Membranes and Tubes

Membranes are in many ways similar in structure to synthetic beaded polymer
supports. Membrane formation, characterization, and applications are described
in a book by Klein [61]. The chemical composition of membranes varies greatly.
Sheets can be constructed from any of a number of primary polymers including
cellulose, polyamide (nylon), polyacrylonitrile, and many other materials. The
most widely used materials for protein purification include cellulose and nylon
membranes.
Hydrophilic affinity microporous membranes composed of reactive rein-
132 Turkova
forced cellulosic polymers has been introduced by Memtec Corporation [51].

This membrane possesses reactive aldehydes for covalent coupling of amino
groups of proteins and other ligands. Reinforced cellulosic polymers are suitable
for pleated and specially cut membranes. Similar cellulose membranes produced
by Sartorius (Gottingen, Germany) have been used for selective removal of hu-
man serum amyloid P component from rat blood by immunoaffinity in an extra-
corporeal circulation system [62].
Synthetic polyamides, known as nylons, are a family of condensation poly-
mers of dicarboxylic acids and a,o-diamines. Several types of nylon, differing
only in the number of methylene groups in the repeating alkane segments, are
available in a variety of physical forms, such as fibers, hollow fibers, foils, mem-
branes, powders, and tubes. The purification of phosphofructokinase from yeast
cell homogenate using of the selective adsorption-desorption with Immunodyne
nylon membrane has been described by Huse et al. [63]. Immunodyne, Biodyne
A, and Loprodyne nylon membranes are supplied by Pall Filtrationstechnik (Drei-
lich, Germany). Immunodyne membranes are preactivated nylon membranes de-
veloped for the covalent fixation of molecules via hydroxyl, carboxyl, or amino
groups. The reactive group of Immunodyne membrane has the ability to selec-
tively bind phosphofructokinase and phosphoglycerate kinase from yeast cell ex-
tract.
A noninteractive polymer (hydrophilic polyvinylidene difluoride) is the
base material of Immobilon AV Affinity Membrane (IAV) produced by Millipore
(Bedford, MA, USA). A variety of ligands containing amines or thiols can be
covalently immobilized to chemically activated hydrophilic microporous mem-
branes under a range of pH conditions (pH 4-10), ionic strengths (0.01-1.0 M),
and temperatures (0-37°C). By varying these parameters, the user can immobi-
lize nanogram to milligram amounts of proteins (150 yg/cm2 is approximately
a protein monolayer). KuCerovL and Turkovi [46] coupled 3,5-diiodo-L-tyrosine
to this membrane and used it successfully for the isolation of pepsin from a crude
extract of human gastric mucosa. The advantage of membrane-based immobilized
ligand was that the time for sorption and desorption of pepsin was very short
when batch-wise isolation was used.
The Affinity-15 Membrane Chromatography System is also produced by
Sepracor Inc. (Marlborough, MA, USA). Affinity ligands are covalently bound
to the membrane using a stable, secondary amine linkage, and therefore ligand
leaching is virtually eliminated. Membrane technology for protein purification
increases the speed, lowers the costs, and simplifies the protein purification.
H. Commercial Availability of Activated Solid Supports

and Biospecific Adsorbents
The rapid evolution of many new unmodified or activated supports, biospecific
sorbents in beads, columns, cartridges, etc., is seen in the catalogues of many
companies and in the International Product Magazine.for Biotechnology. A well-

equipped separations laboratory should have a variety of commercially available
matrices at hand, as well as a full range of vendor catalogs and technical data
sheets on file. Fortunately, a large number of reliable suppliers offers a wide
range of practical and efficient matrices. An appendix of the book of Hermanson
et al. [51] lists vendors of various useful support materials. Tables 6.4 and 6.5
in the book of TurkovL [7] also contain examples of commercially available sup-
ports with full names and addresses of suppliers. However, many suppliers or
their commercially available supports have been rapidly changing and therefore
some of the information herein may be out of date.
V. SURVEY OF THE MOST COMMON COUPLING

PROCEDURES
When selecting the method of attachment, the primary consideration is which

groups of the affinity ligand can be used to form a linkage to a solid support
without affecting the binding site. The attachment also should not introduce non-
specifically adsorbing groups. From this point of view, it is often best to first
couple the spacer to an affinant and, only after it has been modified in this manner,
attach it to a solid support. The linkage between the surface of a solid support
and an affinant should be stable during adsorption, desorption, and regeneration.
When choosing an appropriate method, one should bear in mind the dependence
of the stability of the affinant on reaction conditions. It should also be noted that
the coupling procedure may be influenced by both the nature of the solid matrix
and the substance to be attached. When bifunctional compounds are used for the
coupling, complications arising from the crosslinking of both the carrier and the
proteins with one another can be expected. Therefore reaction conditions such
as pH, temperature, and time should be carefully chosen and monitored during
coupling.
When chemically inert solid supports are available, the preparation of bio-
specific adsorbents consists of two steps: (1) activation or functionalization of
the chemically inert support and (2) coupling of the ligand to this modified matrix.
The activation is dictated chiefly by the nature and the stability of the matrix it-
self. Thus, for example, conditions applicable to the derivatization of glass would
totally destroy agarose.
When the attachment of the affinity ligand onto the solid support is com-
pleted, it is necessary that any remaining reactive groups be eliminated. The prob-
lem of excess reactive groups is encountered in all coupling methods. This prob-
lem can be solved by one of two methods: (1) by coupling a highly penetrable
substance with no effect on the adsorption-desorption procedure; or (2) by solvol-
ysis, which may be used when the affinants are stable to the conditions needed
for the hydrolytic removal of the activated group. The last step in the preparation
134 Turkova
of specific sorbents before their use in bioaffinity chromatography consists of

thoroughly washing out all substances that are not covalently bound to the surface
of the solid matrix.
A. Effect of the Nature of Proteins and Solid Supports

The amount of protein to be coupled as well as the stability and biological proper-
ties of immobilized biomolecules can be affected by the choice of solid support
and the method of coupling. In order to determine the effect of the nature of
the support and the character of the proteins coupled, serum albumin, trypsin,
chymotrypsin, papain, and trypsin inhibitor antilysine were attached to glycidyl
methacrylate copolymer, oxirane-acrylic beads, epoxy-activated Sepharose, and
2,3-epoxypropoxypropy1 derivatives of glass and silica [641. Figure 14 shows the
difference in the coupling of proteins (in mglg of dry support) to methacrylate
copolymer and agarose. It is evident from the figures that the amount of protein
attached as a function of pH is affected by both the character of the proteins and
the nature of the solid support.
The effect of the solid support was much less pronounced when various
proteins were immobilized using benzoquinone. The smaller effect of the solid
support was seen when the effect of pH was investigated on the benzoquione-
mediated immobilization of trypsin, chymotrypsin, and serum albumin, either to
hydroxyalkyl methacrylate copolymer [65] or to agarose [66].
Figure 1 4 Effect of pH on the coupling of (0)serum albumin, (.)papain, (W)antilysin,

(O)trypsin, and (A)chymotrypsin (A) to the glycidylmethacrylate copolymer and (B) to
the epoxy-activated Sepharose 6B in dependence on pH. (Data from Ref. 69.)
B. Support Modification and Affinity Ligand

Immobilization
1. Epoxide-Containing Supports
Camer epoxides may react with amino, carboxy, hydroxy, and sulfydryl groups,
with some aromatic nuclei, such as indole, imidazole, etc., saccharides poly-
nucleotides, adipic acid dihydrazide, etc. T h e S-C, N-C, and 0 - C bonds formed
are extremely stable [7]. Smalla et al. [67] used epoxidated hydroxyethyl methac-
rylate copolymer Separon H E M A for the study of the influence of salts on the
coupling of different enzymes. Figure 1 5 shows the dependence of the activity
and protein binding of aminoacylase-Separon H E M A E conjugate on the salt
used for the immobilization reaction. Aminoacylase-Separon with the highest
activity was achieved using ammonium sulfate. However, as Fig. 16 shows, the
optimum concentration of immobilized protein depends o n the concentration of
this salt. The study of the effect of different concentrations of (NH4)?S04on the
coupling of thermitase, trypsin, chymotrypsin, pepsin, elastase, subtilisin, and
carboxypeptidase A showed a varying effect of added salt on the immobilization
efficiency. Salt-induced immobilization of D N A oligonucleotides on an epoxide-
Figure 15 Dependence of the activity and protein binding of aminoacylase-Separon

HEMA E conjugate on the salt added during the immobilization reaction. Aminoacylase
(5 mg) in 2.5 m10.2 M phosphate buffer (pH 8.0); 50 mg Separon HEMA E, 20 h at 5OC.
Salts added: (A) (NH4),S04, (W) Na2S04,( 0 )(NH,)?HPO,, (0) NaC1. a,, activity of the
enzyme-carrier conjugate in the absence of salts; a, activity of the enzyme-carrier conju-
gate in the presence of salts as indicated. (A) (NH4),S04 addition and proteins binding
(right axis). Aminoacylase (10 mg) in 2.5 ml phosphate buffer (pH 8.0); 50 mg Separon
HEMA E, 30 h at 5°C. (Data from Ref. 67.)
Turkova
Figure 16 Dependence of the activity of proteinase-Separon HEMA E conjugate on

the ammonium sulfate concentration in the medium during immobilization. Maximum
bound activity (%) of immobilized proteinases: (V) pepsin (5 mg) in 2.5 mlO.l M acetate
buffer (pH 8); (A) elastase (2.5 rng) in 2.5 ml 0.1 M phosphate buffer (pH 8); (U) subtilisin
(0.5 mg) in 2.5 ml 0.2 M phosphate buffer (pH 8); (0)chymotrypsin (5 mg) in 2.5 nd
0.1 M phosphate buffer (pH 8); ( 0 )trypsin (5 mg) in 2.5 ml 0.1 M phosphate buffer (pH
8). Conditions: 50 rng Separon HEMA E, 15 h at 20°C for pepsin, chymotrypsin, and
trypsin; 20 h at 5°C for subtilisin and elastase. (Data from Ref. 67.)
activated high-performance liquid chromatographic affinity support was studied

by Wheatley et al. [68].
Epoxide-containing supports can be obtained directly by copolymerization,
e.g., glycidyl methacrylate copolymer (Eupergit C). The disadvantage of biospec-
ific sorbents prepared from glycidyl methacrylate copolymer [69]is the formation
of new epoxide groups during prolonged storage, as determined by the coupling
of different dipeptides.
Bioxyranes (e.g., 1,4-butanediol diglycidyl ether) can be also used for the
introduction of reactive oxirane groups suitable for coupling of sugars via ether
linkages with their hydroxyl groups. Proteins and peptides form alkylarnine link-
ages through their primary amino groups. Thioether linkages are formed with
substances that contain thiol groups. Coupling of the antibiotic novobiocin via
its phenolic hydroxy group to epoxy-activated Sepharose was performed and the
product used for the bioaffinity chromatography of DNA gyrase by Staudenbauer
and Orr [70]. As a suitable blocker for residual oxirane groups of Eupergit C P-
mercaptoethanol was used by Fleminger et al. [56]. Blocking is performed at
nearly neutral pH (pH 8.0), and as it does not form charged groups with the
matrix (in contrast to ethanolamine), nonspecific ionic adsorption of proteins is
eliminated. Remaining reactive groups can also be eliminated by solvolysis. This

method may be used when affinants are stable to the conditions needed for the
hydrolytic removal of the activated group. Excess epoxide groups may be hy-
drolyzed to diols by treatment with 0.1 M perchloric acid or with 10 mM HCI.
2. Hydrazide-Derivatized Solid Supports

A review of hydrazide-derivatized supports in affinity chromatography has been
published by O'Shannessy [71]. Under suitably mild conditions, oxidation of
glycoproteins with periodate results in the generation of reactive aldehydes that
can subsequently be bound with nucleophiles such as primary amines or hydra-
zides. Both primary amines and hydrazides react with aldehydes only when they
are in the unprotonated form. The very low pK of hydrazides, usually around
2.6, in contrast to primary amines, which have pK values in the range of 9-10,
allows one to significantly reduce the formation of Schiff bases between protein
and oligosaccharide by performing the coupling reaction under mildly acidic con-
ditions at pH 4.5-5.5. IJnder such conditions the majority of the primary amines
of the protein moiety are protonated and unreactive. Also, the product of coupling
between an aldehyde and a hydrazide is a hydrazone, which is much more stable
than a Schiff base and, according to many authors, does not require reduction.
However, van Sommeren et a]. [72] observed ligand leakage during both storage
and chromatography in immunosorbent prepared by the attachment of antibody
to hydrazide-activated agarose. When such a condition of poor stability occurs,
the hydrazone produced can be stabilized by performing the reaction in the pres-
ence of sodium cyanoborohydride (hydrazine derivative: conjugate-CH,NHNH
support).
Lamed et al. [73] described the coupling of several nucleotide di- and tri-
phosphates after periodate oxidation to adipic acid dihydrazide coupled to CNBr-
activated Sepharose. In the case of nucleosides, nucleotides, and RNA, immobili-
zation onto hydrazide-derivatized matrices would appear to be the method of
choice. Enzymes with carbohydrate moieties and antibodies contain their active
site in the polypeptide part of the molecules. The presence of carbohydrate raises
the stability of glycoproteins. The carbohydrate moiety of the protein-after ei-
ther periodate or enzymatic oxidation-can readily be immobilized to supports
containing hydrazide groups, giving conjugates with very good steric accessibil-
ity of their active sites and with increased stability [74].
Chemical oxidation of the oligosaccharides may result in the oxidation of
some amino acid residues (such as serine, threonine, proline, and methionine),
thereby leading to a decrease in enzymatic activity. Therefore. periodate oxida-
tion of sugar residues containing galactose can be replaced by enzymatic oxida-
tion by galactose oxidase [75] and carbohydrate moieties in the Fc portion of
immunoglobulin G containing N-acetylneuraminic acid and galactose as terminal
138 Turkova
sugars by enzymatic oxidation utilizing a mixture of neuraminidase-galactose

oxidase. Solomon et al. [76] showed that the coimmobilization of neuraminidase
and galactose oxidase on Eupergit C-ADH beads provides an economical, effi-
cient, and selective system for the enzymic oxidation of monoclonal antibodies
without impairing their immunological activity. Both enzymes are glycoproteins
and therefore immobilization through their carbohydrate moieties after periodate
oxidation was used. After coupling of the enzymes to the carrier, the preparations
were treated with acetaldehyde to block the residual reactive hydrazide groups.
Coimmobilized galactose oxidase and neuraminidase exhibit the well-known ad-
vantages of immobilized enzymes, such as repeated use, good recovery of prod-
ucts, and the possibility of continuous use. However, Kelleher et al. [77] sug-
gested the possibility that purified galactose oxidase not only converted the C-6
hydroxyrnethyl group of galactose to aldehyde but also catalyzed further oxida-
tion to a carboxyl group.
3. Periodate Oxidation
A simple method for the binding of proteins to insoluble polysaccharides after
their periodate oxidation has been described by Sanderson and Wilson [78]. The
aldehyde formed reacts with the protein. Subsequent reduction with sodium boro-
hydride led to the stabilization of the bonds between the protein and the polysac-
Figure 17 pH dependence of the amount of G l y - D , L - P(pmollg

~~ of dry support) bound
to Separon H 1000 GLC oxidized by NaIO, (a)and Separon H 1000 E-NHZactivated
by glutaraldehyde (0).
(Data from Ref. 79.)
charide, and to the reduction of the residual aldehyde groups. In order to compare
the coupling of ligands to solid supports via presumed aldehyde groups, the
amount of G l y - D , L - Pwas
~ ~ coupled to periodate-oxidized glucose-Separon and
to Separon with attached hexamethylenediamine and activated by glutaraldehyde
in relation to pH [79]. From Fig. 17 it is evident that the two activated matrices
are clearly different. Another big difference between these two types of aldehyde
groups was also found in the stability of bonds with glycyl-D,L-phenylalanine.
Unlike the situation with oxidized glucose residues, in which case the reduction
with NaBH, is necessary, the bond between glutaraldehyde and dipeptide is very
stable. There is no difference between reduced and untreated preparation. In both
variants no release of nitrogen was found after 11 weeks at pH 7.0 and 24°C.
4. Glutaraldehyde Activation Technique

The glutaraldehyde activation technique can be used with solid supports having
carboxamide or primary m i n e groups. In the latter case, the activated matrix is
often colored. Weston and Avrameas [80] developed a method for the direct
binding of affinants to polyacrylamide gels using glutaraldehyde which, if present
in excess, reacts via one of its two aldehyde groups with the free amide group
present in the polyacrylamide gel. The remaining free active group then reacts
with the amino group of the affinant added during the subsequent binding reac-
tion. A firm bond is thus formed between the support and the affinant. 6-
Aminohexyl-Sepharose 4B was activated with glutaraldehyde by Cambiaso et al.
(811 and used for immobilization of immunoglobulins G and A, ferritin, and
albumin. A support based on nonporous silicon dioxide of particle size 0.01 -0.1
l m , modified by 3-(amino)propyltriethoxysilane and activated by glutaraldehyde,
was employed for the immobilization of concanavalin A, immunoglobulins, basic
pancreatic trypsin inhibitor, and trypsin by Fusek et al. [82]. Activation of sup-
ports by glutaraldehyde for cell immobilization by covalent linkage was used by
JirkG and Turkova [83].
A number of competing mechanisms have been proposed to describe the
reaction between the aldehyde and the amino function of the ligand. The most
plausible one is that reported by Monsan et al. [84] who presented evidence
showing that glutaraldehyde polymerizes to an unsaturated aldehyde. The latter
subsequently reacts with m i n e s to form a,bunsaturated imines, which undergo
resonance stabilization with the neighboring ethylene function. However, not all
of the data on the coupling products of glutaraldehyde and polypeptides are fully
explained. The products actually formed strongly depend on the pH of the reac-
tion and on the initial ratio and character of the reactants. In addition, competing
reaction mechanisms lead to different products. But in all cases the resulting
bonds can be expected to be chemically very stable. A number of other bifunc-
tional derivatives are described by Turkova [7]. However, when using bifunc-
140 Turkova
tional derivatives it should be born in mind that side reactions, such as crosslink-
ing of the support, may occur, and the permeability may decrease drastically.
5. Cyanogen Bromide Activation

One of the first methods introduced to activate agarose, dextran, and, less com-
monly, cellulose and hydroxyalkyl methacrylate gel was cyanogen bromide acti-
vation [85]. The mechanism of activation by CNBr and the subsequent coupling
of the affinant was elucidated by Kohn and Wilchek [86]. The coupling of an
affinant occurs predominantly via the free amines in the unprotonated state. The
following pH values are appropriate for coupling: aliphatic amines, pH -10;
amino acids, pH -9; aromatic amines, pH 7-8. Ribonucleotides are coupled
through their phosphate groups at pH 6. The attachment of amines to cyanogen
bromide-activated Sepharose produces an N-substituted isourea that is capable
of protonation at neutral and alkaline pH values. This linkage is unstable in the
presence of primary amines and ammonia, which is the primary flaw of this
method. Discussions of the mechanism for CNBr activation and the difference
between a freshly activated and commercially available CNBr-activated Sepha-
rose (which is treated with acid in order to achieve better stabilization of activated
resin) have been published by Wilchek and Miron [87]. CNBr-activated Sepha-
rose treated with 1 N hydrochloric acid for 1 h in order to form carbonate can
be used to couple amino-containing ligands to the resin, yielding columns that
consist of stable and uncharged carbonates. These activated resins will be useful
for coupling of low molecular weight ligands such as diaminohexane or aminoca-
proic acid, as the coupling to the resin has to be performed at high pH (-9.5)
because of the low reactivity of the carbonate formed. However, carbonate Sepha-
rose also can be used to couple proteins under mild conditions and in high yields.
6. Coupling with Condensation Agents

One of the most frequent combinations of gel and spacer, used for the binding
of low molecular weight affinity ligands, is Sepharose with attached hexamethy-
lenediamine (trade name AH-Sepharose) or E-aminocaproic acid (trade name CH-
Sepharose). In order to couple them to affinants carrying primary aliphatic or
aromatic amine or carboxyl groups, condensation via carbodiimide intermediater
is used. Dicyclohexylcarbodiimide in pyridine can be used for coupling of nucleo-
tides through their phosphate groups to supports containing hydroxyl groups [ 2 2 ] .
The binding reaction between the carboxyl group and the nucleophile can be
almost quantitative in the presence of excess of carbodiimide and the nucleophilic
reagent such as 1-ethyl-3-(3-dimethylaminopropyl)carbodiimidehydrochloride
(EDC) and 1 -cyclohexyl-3-(2-morpholinoethyl)carbodiimidemetho-p-toluene-
sulfonate (CMC). Their main advantage is that their corresponding urea deriva-
tives are soluble in water and they can therefore be easily eliminated from the
gel by washing with water. The pH range used for the carbodiimide condensation
is 4.7-6.5 and the reaction time is 1.5-72 h at a carbodiimide concentration
of 2-100 mglml. The disadvantage is that carbodiimides are relatively unstable
compounds and must be handled with care because of their toxicity. After the
coupling of the affinant, the reaction ought to be continued to block reactive
groups by carrying out further carbodiimide reactions with glucosamine or 2-
aminoethanol in the case of CH-Sepharose, or with acetic acid as blocking agent
for the amino groups of AH-Sepharose.
Boschetti et al. [88] published a comparative study of soluble carbodiimide
CMC with a condensation agent, N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquino-
line (EEDQ). The latter catalyzes the reaction between spacer arm and ligand by
forming a mixed anhydride on the carboxyl group of the spacer arm. The mixed
anhydride then reacts with the complementary amine to form a stable amide bond.
The mixed anhydride can also react with other nucleophilic groups such as sulfhy-
dry1 and hydroxyl groups. EEDQ must be used in a water-ethanol mixture, which
is an advantage when working with ligands that are only slightly soluble in water.
EEDQ is a suitable condensation agent, being stable, nontoxic, and cheap.
7. Active Esters
Cuatrecasas and Parikh [891 described the preparation of N-hydroxysuccinimide
(NHS) esters of succinylated aminoalkyl agarose derivatives. These active ester
derivatives of agarose, when stored in dioxane, are stable for several months.
These derivatives very rapidly form stable amide bonds (at 4°C) with nonproton-
ated forms of primary aliphatic or aromatic amino groups at pH 6-9. Among the
functional groups of amino acids tested, only sulfhydryl groups compete effec-
tively with the amino groups during the binding reaction.
Using the esterification of the carboxyl groups of CH-Sepharose 4B with
the application of N-hydroxysuccinimide, Pharmacia produces activated CH-
Sepharose 4B. The pH range suitable for binding on this derivative is indicated
by Pharmacia to be 5-10, with an optimum of pH 8. The advantage of lower
pH values is in the decreased ester hydrolysis but with a slower reaction rate.
Buffers that contain amino acids cannot be used (Tris or glycine buffers) in the
coupling reaction. Agarose derivatives containing N-hydroxysuccinimide ester
have been introduced by Bio-Rad under the names Affi-Gel 10 and Affi-Gel 15.
As shown in Fig. 18, Affi-Gel 10 couples proteins best at a pH near or below
their isoelectric point, and Affi-Gel 15 couples proteins best near or above their
isoelectric point. Therefore, when coupling at neutral pH (6.5-7.5), Affi-Gel 10
is recommended for proteins with isoelectric points of 6.5-1 1 (neutral or basic
proteins) and Affi-Gel 15 is recommended for proteins with isoelectric points
below 6.5 (acidic proteins). The difference in coupling efficiency of Affi-Gel 10
and Affi-Gel 15 for acidic and basic proteins can be attributed to interactions
142 Turkova
2.5 4.5 6.5 8.5 10.5

PROTEIN ISOELECTRIC POlNT
Figure 18 Immobilization of proteins to Affi-Gel 10 and Affi-Gel 15. Protein solutions
were gently mixed at 0-4OC with 2 ml Affi-Gel 10 or Affi-Gel 15 for 2 h. The reaction
was terminated by addition of ethanolamine, pH 8.0, to a concentration of 0.10 M, and
after 30 min transferred to a 1 X 10 cm chromatography column. Unreacted protein was
eluted with 7 M urea containing 1 M NaCl. Published values for the isoelectric points
used to construct this figure were: fetuin (a,k), pH 3.3; human a-antitrypsin (b,l), pH 4.0;
human y-globulin (f,p), pH 5.8-7.3; human transferrin (e,o), pH 5.9; ovalbumin (c,m),
pH 4.7; bovine serum albumin (d,n), pH 4.9; bovine hemoglobin (g,q), pH 6.8; equine
myoglobin (h,r), pH 6.8-7.8; cytochrome c (i,s), pH 9.0-9.4; and lysozyme (j,z), pH 11.O.
(0) Affi-Gel 10; ( 0 )Affi-Gel 15. (From Frost RG, et al. Biochim Biophys Acta 1981;
670:163-169.)
between the charge on the protein and charge on the gel. Hydrolysis of some of
the active esters during aqueous coupling will impart a slight negative charge to
Affi-Gel 10. This negative charge will attract positively charged proteins (proteins
buffered at a pH below their isoelectric point) and enhance their coupling effi-
ciency. Conversely, the negative charge will repel negatively charged proteins
(proteins buffered at a pH above their isoelectric point) and lower their coupling
efficiency. Affi-Gel 15, due to the tertiary amine incorporated into its arm, has
a slight overall positive charge, and the effects are reversed. Coupling under
anhydrous conditions is the preferred method when this is suitable for the ligand.
Since there is no hydrolysis of active esters in the absence of water, the only
reaction will be that of the ligand with the gel. After the coupling of affinity
ligands the unreacted groups can be eliminated by addition of 0.1 M Tris buffer,
pH 8.
8. Activation with Carbonylating Reagents
Activation of crosslinked agarose with 1,l'-carbonyldiimidazole (CDI) has been
described by Bethell et al. [90]. The activated imidazolylcarbamate supports react
with primary amino groups at pH 8.5-10.0 and are more stable to hydrolysis
compared with N-hydroxysuccinimide ester-activated supports. CDI-activated
agarose has a half-life of more than 14 weeks when stored in dioxane. A further
advantage of using CDI-activated agarose is that the resulting N-alkylcarbamates
are uncharged in normal pH ranges. Optimization of protein immobilization on
CDI-activated diol-bonded silica has been described by Crowley et al. [9]]. CDI-
activated glass matrices are commercially available, e.g., from Pierce Chemical.
9. Triazine Method
The covalent linkage of ligands to hydroxyl-containing supports activated with
cyanuric chloride (2,4,6-trichloro-s-triazine = TCT) was developed by Kay and
Crook [92]. An easily controllable variant for use with macroporous cellulose
bead with TCT was developed by BeneS et al. [93]. Bilkovi et al. [5] used this
method of coupling for the immobilization of the natural chymotrypsin inhibitor
antilysine. The disadvantage of this method is that the activating reagents are
highly toxic.
10. Reversible Covalent Immobilization of Proteins by

Thiol-Disulfide Interaction
Carlsson et al. [94] employed epoxide-activated agarose as the basis for the prepa-
ration of the mercaptohydroxypropylether of agarose gel, which they used for
covalent immobilization of a-amylase and chymotrypsin by thiol-disulfide inter-
change. This technique consists of two steps: (1) thiolation of enzymes with
methyl 3-mercaptopropioimidate; (2) binding of thiolated enzymes to a mixed
disulfide derivative of agarose obtained by reaction of the mercaptohydroxypro-
pylether of agarose with 2,2-dipyridyldisulfide. When the preparation had lost
its enzymatic activity, the inactive protein was reduced off and the gel used for
the binding of a new active thiolated a-amylase.
11. Benzoquinone Activation

The mechanism of the activation of hydroxyl-containing solid supports by means
of benzoquinone and coupling of NH2-containing compounds was described by
Brandt et al. [66]. Benzoquinone is a very active reagent and, presumably due
to the secondary reaction, the immobilized compounds are usually strongly col-
ored. In Section V. A it was already noted that in contrast to coupling of serum
albumin and chymotrypsin to epoxy-activated solid supports, the pH optimum
for their coupling to benzoquinone-activated Sepharose 4B and methacrylate co-
polymer is the same.
12. Diazotization
The first attachment of an affinant to cellulose was carried out by means of diazo-
nium groups by Campbell et al. [95]. The affinants are bound by their aromatic
144 Turkova
residues (mainly tyrosine and histidine), but also slowly through their amino
groups. Although this method is not frequently used at present, it offers two
essential advantages: (1) the bound ligand can easily be split off by reduction
with sodium dithionite; (2) this reductive cleavage allows protein-inhibitor conju-
gates to be isolated intact under mild conditions.
13. Other Methods

A simple method for the activation of supports carrying hydroxyl groups, such
as agarose, cellulose, diol-silica, glycophase-glass, or hydroxyalkyl methacrylate
copolymer, with 2,2,2-trifluorethanesulfonylchloride (tresyl chloride), was intro-
duced by Nilsson and Mosbach [96]. The advantage of tresyl chloride is that it
is very reactive and therefore allows efficient coupling under very mild condi-
tions, and to a number of different supports, including those used for HPLC. It
is suitable for coupling ligands containing -NH2 or -SH.
The use of divinylsulfone in the production of adsorbents was described
by Porath and Sundberg [97]. The activation takes place rapidly under fairly mild
conditions. The coupling can be performed not only with amino group-con-
taining molecules but with carbohydrates, phenols, and alcohols at a higher pH.
As a consequence, divinylsulfone crosslinking considerably improved the flow-
through properties of agarose. The product coupled via divinylsulfone is very
stable at acidic and neutral pH, but it is labile under alkaline conditions.
The activation of hydroxylic matrices by use of Zfluoro-l-methylpyridin-
ium toluene-4-sulfone (FMP) has been published by Ngo [98]. FMP-activated
gels can be used to couple ligands containing either amine or sulfhydryl groups in
slightly alkaline (pH 8-9) aqueous solutions or in organic solvents. The resulting
linkages are stable and nonionic. For more detailed information concerning ma-
trix activation and ligand coupling to solid supports, the reader is referred to
Hermanson et al. [5 1 ] and Turkovi [7].
.C. Molecular Imprinting Technology (MIT)

Molecular imprinting is a way of chemically preparing polymer materials for
roles in molecular separation. Through molecular imprinting technology (MIT),
it is possible to tailor-make polymers that are selective for different compounds
[99]. The principle of molecular imprinting is described in Fig. 19. The first
step is prearranging the print molecule and the monomers. After development of
complementary interaction between the print molecule (template) and the mono-
mers (a) is the polymerization around the print molecule-monomer complex (b).
The 1ast.step is the removal of the print molecule from the polymer. Polymeriza-
tion thus preserves the complementarity to the print molecule and subsequently
the polymer selectively adsorbs the print molecule. The print molecule binds
Figure 19 The principle of molecular imprinting. Development of complementary inter-

actions between the print molecule and the monomers (a); polymerization (b): removal
of the print molecule from the polymer (c). M, monomers; PM, print molecule; CR, cross-
linker.
more favorably to the extracted polymer than do structural analogs. Wulff and
Vesper [loo] preparecl chromatographic sorbents with chiral cavities for racemic
resolution. For this purpose, with the aid of a chiral template molecule, functional
groups were placed in a highly crosslinked polymer in such a way that they were
present in a chiral cavity in a given stereochemistry. For example, Cnitrophenyl-
a-~-mannoside-2,3;4,6-di-O-(4-~inylphenylboronate) (A) was copolymerized to
a macroporous polymer, from which the template 4-nitrophenyl-a-D-mannopyra-
noside (B) could be split off. These polymers were used for the chromatographic
resolution of the racemate of the template molecule B. Chromatographic resolu-
tion of racemic mixtures of amino acid derivatives by use of molecular imprinting
of amino acid derivatives in macroporous polymers was described by Sellergren
et al. [ l o l l .
Synthesis and characterization of polymeric receptors for cholesterol by
use of MIT was published by Whitcornbe et al. [102]. The polymers obtained
by this method were shown to bind cholesterol with a single dissociation constant,
thus displaying characteristics similar to the true biological receptor or synthetic
host. Mosbach [lo31 informed of an increasing number of applications of MIT.
These include (1) the use of molecularly imprinted polymers as tailor-made sepa-
ration materials; (2) antibody and receptor binding site mimics in recognition and
assay systems; (3) enzyme mimics for catalytic applications; and (4) recognition
elements in biosensors. The stability and low cost of molecularly imprinted poly-
146 Turkova
mers make them advantageous for use in analysis as well as in industrial scale
production and application.
D. General Considerations in the Choice of Sorbents,

Coupling and Blocking Procedures
When choosing the carrier and the coupling procedure one should take account
not only of the properties associated with the nature of affinity chromatography
itself but also of their field of use. Kukongviriyapan et al. [I041 described the
maximum binding of antibodies against Naja naja siamensis toxin 3 (T3) and
operational half-life of T3 immobilized at various ligand densities (Table 2). The
maximum binding capacity of antibody was obtained on immobilized T3 with
the lowest half-life (19 days). The lowest antibody binding capacity was obtained
on an adsorbent containing T3 coupled to albumin-Sepharose, where the half-
life was 108 days. The investigation described was undertaken to study various
parameters in the use of bioaffinity chromatography to purify antibodies against
cobra postsynaptic toxin from horse refined globulin for therapeutic purposes.
The choice of coupling procedures and solid supports depends on the use
to which the material will be put: analytical, semipreparative, or preparative pur-
poses. It is also necessary to take into consideration whether low or high molecu-
lar weight affinity ligands are being used. It should also be borne in mind that
the toxicity of such reagents as hydrazine, cyanogen bromide, trichloro-s-triazine,
and divinylsulfone is high; that of epichlorohydrin, bisepoxiranes, glutaralde-
hyde, carbonyldiimidazole, benzoquinone, diazonium, and tresyl chloride is mod-
erate; only periodate is nontoxic.
An important factor is a rapid increase in the production of commercially
available activated solid supports and many biospecific adsorbents. Hermanson
Table 2 Ligand Density, Maximum Antibody Binding Capacity, and Operational

Half-life of Various Adsorbents
Toxin immobilized Maximum binding capacityb Half-life

Affinity sorbent (mglml packed ge1)"mglmg immobilized toxin) (days)
T o l u m e of gel was measured in 0.15 N NaCI.

Protein eluted at pH 2.05.
' Determined by using '"I-labeled N. n. siamensis toxin 3 .
Seph, Sepharose 4B; Ae, aminoethyl; Suc, succinyl; T3, N. n. siamensis toxin 3 .
et al. [5 I ] list 34 suppliers, their addresses, and the products including affinity
matrices, activated supports, activation reagents, immobilized ligands, prepacked
columns and accessories, and affinity purification kits. Turkov6 [7] lists addresses
of 90 companies providing activated supports and biospecific adsorbents, along
with examples of their products. However, it was already mentioned that many
suppliers or their commercially available supports have been changing rapidly.
VI. GENERAL CONSIDERATION ON AFFINANT-SORBENT

BONDING
In order for the immobilized affinity ligands to be readily accessible to the binding
sites of biological macromolecules, it is necessary to have more than a solid
support with high porosity. The chemical groups of the affinant that participate
in the interaction with the macromolecular substance must also be sufficiently
remote from the surface of the solid matrix to avoid steric hindrance. The impor-
tance of spacing between the low molecular weight ligand and the surface of the
matrix in bioaffinity chromatography was illustrated by Cuatrecases et al. [I051
in one of the first successful applications of bioaffinity chromatography in the
isolation of enzymes. Figure 20 represents the bioaffinity chromatography of a -
chymotrypsin, both on Sepharose coupled with &-aminocaproyl-D-tryptophan
methyl ester (A) and on Sepharose coupled with D-tryptophan methyl ester (B),
in comparison with chromatography on unsubstituted Sepharose. In the first in-
stance (A), the bound inhibitor has high affinity for a-chymotrypsin and the en-
zyme can be released from the complex only by decreasing the pH of the eluting
buffer. By using 0.1 M acetic acid, pH 3.0, chymotrypsin is eluted in a sharp
peak and the volume of the eluted chymotrypsin does not depend on the volume
of the sample applied to the column. In the second instance (B), the inhibitor
coupled directly on Sepharose has a much lower affinity for the a-chymotrypsin
owing to steric hindrance. In this instance a change of buffer is not necessary
for enzyme elution and, as can be seen from the graph, the enzyme is eluted in
a much larger volume closely after the inactive material. In order to verify that
nonspecific adsorption on the carrier did not take place under the given experi-
mental conditions, chromatography of a-chymotrypsin on an unsubstituted car-
rier was carried out (C).
Studies on the conformation of model peptides in membrane-mimetic envi-
ronments [I061 lead the authors to expect that steric accessibility of the reactive
groups of the affinity ligand not only will be determined by their distance from
the surface of the solid support but will also depend on the interfacial water
region, which does not stabilize the same conformation as does bulk water. The
arrangement of the solvent layers on the surface of the solid support will also
obviously be one of the main factors affecting the penetration of the bound mole-
Turkova
ELUTION VOLUME , ml
Figure 20 Bioaffinity chromatography of a-chymotrypsin on inhibitor Sepharose col-
umns. The columns (50 X 5 mm) were equilibrated and run with 0.05 M Tris-hydrochloric
acid buffer of pH 8.0. Each sample (2.5 mg) was applied in 0.5 ml of the same buffer.
Columns were run at room temperature with a flow rate about 40 ml/h and fractions
containing I ml were collected. The arrows indicate a change of elution buffer (0.1 M
acetic acid, pH 3.0). (A) Sepharose coupled with E-aminocaproyl-D-tryptophanmethyl
ester. (B) Sepharose coupled with D-tryptophan methyl ester. (C) Unsubstituted Sepharose.
The first peaks in A and B were devoid of enzyme activity. (From R Cuatrecasas P, et
al. Proc Natl Acad Sci USA 1968; 639-643.)
cules to the surface of the support. As a logical result, the mode of immobilization
has a greater effect on the interaction of the substances isolated with immobilized
low molecular weight ligands than with high molecular ligands.
Angal and Dean [I 071 studied the effect of matrix on the binding of human
serum albumin to the sulfonated aromatic dye Cibacron Blue 3G-A immobilized
on 10 solid supports. The results are summarized in Table 3. A 16-fold range in
ligand concentration was observed in the amount of dye immobilized to the vari-
ous matrices despite the similarity of the reaction conditions used. The adsorption
was shown to be selective because rabbit, chicken, and bovine albumin did not
bind to the same immobilized dye.
High molecular weight affinity ligands usually offer more possibilities for
Table 3 Comparison of Properties of Different Support Media"

Ligand HSA HSA eluted 10-'X Apparent
Support concn. adsorbed by desorption association
material (plml of gel) (mglml of gel) (mglml of gel) constant (M-')
Cellulose 3.2 0.8 0.2 I. I
Sephacryl 1.9 10.7 8.4 25.0
Ultrogel AcA54 0.3 0.2 0.2 3.1
IJltrogel AcA44 0.4 0.3 0.2 4.8
Sepharose 6B 2.3 32.4 20.7 45.3
Sepharose 4B 1.5 15.7 14.1 42.0
Sepharose 2B 0.9 9.1 8.4 35.0
Sepharose CL6B 0.7 5.4 4.2 30.0
Sepharose CL4B 0.4 5.0 4.6 46.8
Sepharose CL2B 0.2 1.7 24.6 24.6
"The value for human serum albumin adsorbed is calculated by difference and after elution with
thiocyanate).
HSA, human serum albumin.
the preparation of affinity adsorbents. However, a very important condition in

this instance is that the attachment to the solid support should not cause a change
in the native conformation of the ligand. To protect the binding site of the lectin,
Clemetson et al. [I081 coupled the lectins to Sepharose 4B after CNBr activation
in the presence of 2% of the appropriate sugar. A solution that may overcome
both a low stability and the capacity of biospecific sorbents for some glycopro-
teins, first of all antibodies and enzymes, may lie in their immobilization through
their carbohydrate moieties [74]. The advantage of oriented immobilization by
use of biospecific complex formation was shown in Sections 111. B and 111. G.
In order to minimize the nonspecific sorption of inert compounds to affinity
sorbents and for the greatest possible exploitation of the attached affinity ligand
it should be used at the lowest possible content in the specific sorbent [7]. The
importance of the low concentration of the affinity ligand and the effect of the
uneven surface of the gel are illustrated in Fig. 21. This figure illustrates schemati-
cally the surface of the macroreticular hydroxyalkyl methacrylate polymer in the
form of aggregated beads. After the binding of the affinity ligand via the spacer,
well-accessible, less accessible, and sterically inaccessible molecules of the af-
finity ligand can be recognized. This steric hindrance may in many cases explain
the low saturation of molecules of the immobilized ligand with the isolated com-
pound and the heterogeneity in the affinity of immobilized ligands. For the prepa-
ration of a homogeneous bioaffinity sorbent it is hence necessary to select condi-
tions for the ligand-carrier binding under which the density of the affinity ligand
Turkova
-PECIFIC MULTI-POINT BONDING OF INERT PROTEIN
SPECIFIC COMPLEMENTARY "ONE -TO-ONE " BONDING OF ISOLATED

ENZYME
NONSPECIF~C MULTI- POINT BONDING OF ISOLATED ENZYME

IN INCORRECT ORIENTATION (LEFT ),
IN SPECIFIC MULTI- POINT BONDING ( MIDOLE )
AND IN STERIC HIFOERED BOWING ( RIGHT )
Figure 21 Schematic illustration of the effect of concentration of an immobilized aflin-

ity ligand and uneven surface of a solid carrier on specific and nonspecific sorption.
is low and the ligand is preferentially bound only to readily accessible sites (Fig.
21, middle). The low density of the affinity ligand is also required to prevent
nonspecific binding. The top part of the figure illustrates the sorption of macro-
molecules, e.g., enzymes, that d o not have a complementary binding site for the
immobilized affinity ligand. This binding is caused by the high density of the
immobilized affinity ligand, permitting the formation of multiple nonspecific
bonds between the macromolecules in solution and the solid phase. These nonspe-
cific bonds allow the compounds present in the mobile phase to bind to the affinity
ligand, the spacer, and the surface of the solid matrix. These multiple nonspecific
bonds may be stronger than a single complementary bond between the isolated
enzyme and immobilized complementary affinity ligand, e.g., an inhibitor. When
the nonspecific multiple bonds are involved in addition to the specific comple-
mentary bond (bottom part of the figure) they increase the strength of the binding
in the specific complex. This results in the elution of an enzyme in several frac-
tions and also gives rise to difficulties in enzyme elution. The multiple nonspecific
bonds may then lead to binding of the enzyme to the immobilized ligand in an
incorrect orientation (bottom part of the figure left). Thus, affinity chromatogra-
phy may yield good results only on a sorbent containing a low-affinity ligand
concentration (middle part of the figure) where the enzyme can bind only via the
complementary bond to the immobilized affinity ligand at a ratio of 1 : 1.
In order to determine experimentally the effect of the concentration of the
immobilized inhibitor on the course of affinity chromatography of proteolytic
enzymes, specific sorbents for carboxylic proteinases containing different con-
centrations of &-aminocaproyl-L-Phe-D-Phe-OMewere prepared by Turkovi et
al. [109]. The inhibitor was coupled to epoxide-containing Separon H 1000; the
resulting concentrations of E-aminocaproyl-L-Phe-D-Phe-OMein pmollg of dry
gel were 0.85, 1.2, 2.5, 4.5, and 155. Solutions of porcine, chicken, or human
pepsin were applied continuously to columns of these affinity sorbents until the
effluent showed the same activity as the solution applied (cf. Fig. 22). On columns
of affinity sorbents containing the inhibitor attached in the concentration range
0.85-4.5 pmollg, in all cases one sharp peak of very active pepsin was achieved.
The amount of desorbed pepsin was calculated from the adsorbance at 278 nm
and proteolytic activity measurement. In contrast, using the affinity sorbent con-
taining the inhibitor at a concentration of 155 pmollg, several pepsin peaks were
seen. This different behavior of the enzyme on affinity sorbents having low and
high amounts of immobilized inhibitor may be due to multiple bonding of the
enzyme molecule and inert proteins, as illustrated in Fig. 21. The amounts of
porcine, chicken, and human pepsins eluted depending on the concentration of
E-aminocaproyl-L-Phe-D-Phe-OMeof the individual affinity sorbents were deter-
mined. From the comparison of individual pepsins it was evident that &-aminoca-
proyl-L-Phe-D-Phe-OMe-Separonis a very good sorbent only for porcine pepsin.
The specific sorbent containing 0.85 pmol of inhibitor per g of dry support sorbed
29.4 mg of porcine pepsin per g of dry sorbent. Using the molecular weight of
pepsin (35,000) it can be calculated that 99% of the inhibitor molecules attached
were involved in the specific complex. As the amount of the affinity ligand
attached increased, the portion of the inhibitor molecules involved in the specific
complex with pepsin decreased sharply. With specific sorbents containing 4.5
pmol inhibitor per g only 26% of the total number of inhibitor molecules attached
FRACTION NUMBER
Figure 22 Bioaffinity chromatography of porcine pepsin on E-aminocaproyl-L-P~~-D-

Phe-OCH,-Sepharon columns with (A) low and (B) high concentrations of the immobi-
lized inhibitor. The solution of crude porcine pepsin was applied continuously onto affinity
columns (5 ml) equilibrated with 0.1 M sodium acetate (pH 4.5). At the position marked
by the first arrow equilibrated buffer was applied to the columns to remove unbound pepsin
and nonspecifically adsorbed proteins. The second arrow indicates the application of 0.1
M sodium acetate containing 1 M sodium chloride (pH 4.5). Fractions (5 ml) were taken
at 4-min intervals. The inhibitor concentration of affinity sorbents were (A) 0.85 and (B)
155 ~ m o l l gof dry support. Solid line, protein; broken line, proteolytic activity. a, b, and
c, fractions of pepsin of the same specific proteolytic activity. (From Ref. 109.)
take part in the sorption of pepsin. In an affinity sorbent with the lowest concen-
tration of the affinity ligand only, all of the molecules of the affinity ligand are
fully available for the formation of the complex with the isolated enzyme.
Liu and Stellwagen [I101 used Cibacron Blue F3GA immobilized at several
dye densities to study the different adsorptions of monomeric octopine dehydro-
genase and tetrameric lactate dehydrogenase. They determined that the half-time
for desorption of lactate dehydrogenase from Cibacron Blue F3GA-Sepharose
CL-6B was nearly identical (27 s) to the mass transfer half-time of the protein
in the matrix. The change in the visible adsorbance of Cibacron F3GA accompa-
nying its complexation with lactate dehydrogenase was used to observe the kinet-
ics of complexation. The results of their experiments indicated that it is chromato-
graphic mass transfer and not the chemistry of complexation that limits zonal
chromatography. This is why the effect of immobilized dye concentration on
protein complexation is usually studied using zonal chromatography.
VII. APPLICATION OF BlOAFFlNlTY CHROMATOGRAPHY

A. Isolation, Determination, and Removal of Biologically
Active Compounds
Bioaffinity chromatography has increasingly become the method of choice for
the isolation, determination, and removal of biologically active substances. The
use of biospecific adsorbents is described in an ever greater number of publica-
tions. Bibliographic review of the use of bioaffinity chromatography on 200 pages
was selected from more than 5000 papers [7]. Examples of classical and high-
performance bioaffinity chromatography are shown in preceding sections. The
expansion of this method, which has been strongly pursued for potential applica-
tions, has been largely due to developments in biotechnology.
The various conditions used in bioaffinity chromatography depend on the
nature of the substances to be isolated. Even if in many instances a homogeneous
compound could be isolated from the starting material by a single chromato-
graphic step, combinations of affinity chromatography with other purification
processes also occur. In high-performance liquid bioaffinity chromatography
(HPLBAC) the specificity of bioaffinity chromatography is combined with a
high-performance technology based on the use of rigid particles of uniform, small
size (1 -50 pm in diameter). The difference of affinity chromatography methods
using soft gel or porous and nonporous small hard particles has already been
shown in Fig. 10. Trends in the application of HPLBAC [44] are concentrated
on improvement in analytical and preparative biotechnology. Processes such as
the production of monoclonal antibodies or of recombinant DNA-specified pro-
teins and peptides have created a need for the rapid determination of specific
154 Turkova
biomolecules in complex mixtures during production. HPLBAC can be used for

monitoring biomolecules in complex mixtures, possibly providing on-line analy-
sis in bioreactors and downstream processing. Figure 23 shows the monitoring
of Protein A in fermentation broth of Staphylococus aureus. The authors used
HPLBAC routinely to optimize the culture time and consumption of media in the
fermentation. HPLBAC has an important role in the quantitative and qualitative
analysis of biologically active molecules, which leads to an assessment of their
purity, potency, and safety.
Large-scale affinity chromatography is usually performed in industrial lab-
oratories and most information is thus proprietary. Large-scale protein purifica-
tion was reviewed by Narayanan [I 1 1 1 . According to him, the requirements of
a large-scale purification protocol are largely determined by the nature and quality
of the desired final product and its intended use. For example, proteins for thera-
peutic use need to be extremely pure to minimize the risk of unwanted immuno-
genic responses. Four factors dictate at what stage affinity chromatography can
PROTEIN A
,
TIME min
Figure 23 Monitoring of Protein A in a fermentation broth using an IgG-HPLC column

(10 pm, 10 X 0.5 cm). Conditions: mobile phase, initially 0.1 M NaH,PO, pH 7.0, 0.15
M NaCI, changed to 0.1 M glycine-HCI, pH 2.2 (at arrow) to elute Protein A. Flow rate,
4 mllmin; sample, 0.96 ml fermentation broth; temperature, 22°C. (From Ref. 44.)
best be exploited: (1) the concentration of the desired product in the starting
material; (2) the composition of other components in the starting material, along
with its physical and chemical properties; (3) the desired product purity; and (4)
the volume of material to be processed. Each stage of an affinity chromatography
process-adsorption, washing, elution, and regeneration-needs to be optimized
before the process is scaled up. Moreover, each step must be consistent and repro-
ducible. Bioadsorbent stability is an important criterion because the ligands are
often labile biological molecules. Affinity media can lose its effectiveness be-
cause of unstable ligands, microbial contamination, and column clogging due to
the presence of insoluble matter in the sample and in the eluting buffers. Accumu-
lation of denatured protein, lipids, nucleic acids, etc., that are not eluted during
the regeneration process can also limit the lifetime of the column. A preparative
column is exposed to more protein in three or four preparative cycles than an
analytical column is exposed to in two or three hundred cycles. Under these
conditions, maintenance of the affinity media becomes very important. Stringent
clean-in-place procedures are recommended by the media manufactures to pro-
long its lifetime. The feasibility of such measures should be taken into consider-
ation before a purification method is scaled up. Suitable affinity ligands for large-
scale isolations are polyclonal or monoclonal antibodies because they can be
produced against any compound, even if the latter is only partially purified. Fur-
thermore, monoclonal antibodies can be selected with any desired affinity,
thereby making the use of biospecific columns, prepared by their immobilization,
very attractive. Such antibodies show absolute specifity for only one single epi-
tope: the smallest immunologically submolecular group on an antigen. Mono-
clonal antibodies can be produced in large quantities by the hybridoma technol-
ogy developed by Kohler and Milstein [112]. Tarnowski and Liptak [113] used
this antibody for the automated immunosorbent purification of interferon. Large-
scale purification of monoclonal antibody, which recognized a human mela-
noma-associated 250-kDa glycoprotein/proteoglycan,was isolated by Lee et al.
[114] by use of staphylococcal Protein A-Sepharose. As improvements in the
technology of chromatographic support materials are developed, the combination
of the unique selectivity of an affinity interaction along with the improved perfor-
mance of modern supports assures bioaffinity chromatography a commanding
position in the future of large-scale purification. At present, affinity chromatogra-
phy is already being increasingly used in large-scale purification of therapeutic
products.
B. Immobilization of Enzymes by Use of Their

Suitable lmmunosorbent
The high efficiency of most natural processes and their low-energy demands de-
pend on highly active and specific catalysts, i.e., enzymes. In nature, however,
156 Turkova
enzymes are produced by organisms for their own use: in regulated metabolic
processes their low stability, narrow specificity, and strictly defined requirements
are inherently connected with their function. However, it is important to take into
account that after completion of their function in living cells, their denaturation
and hydrolysis by proteinases occurs. For their stabilization in vivo as well as
their function, hydrophobicity plays an important role. However, the contact of
nonpolar amino acids with water is enthalpically disadvantageous and results in
ice-like water structure. Such contact of water with hydrophobic surface clusters
of proteins in vitro decreases protein stability. Hence, reduction of the nonpolar
surface area should stabilize proteins. Shami et al. [I 151 showed the dramatically
increased activity of amylase complexed with their antibodies. Sheriff and co-
workers [I 161 used x-ray crystallography to determine the three-dimensional
structure of antibody-antigen complex by use of antilysozyme Fab and lysozyme.
They demonstrated that more than 80 van der Waals bonds are formed during
the antibody-lysozyme interaction. Formation of the complex resulted in exclu-
sion of all molecules of water. The conclusion of cited data was that oriented
immobilization of enzymes by use of antigen-antibody interaction can result both
in good steric accessibility of the enzyme active site toward high molecular mass
substrates and an increased stability. Moreover, biospecific adsorption of an en-
zyme to a suitable immunosorbent combines the isolation of molecules with their
oriented immobilization.
To confirm this hypothesis, polyclonal antibodies suitable for the oriented
immobilization of chymotrypsin were prepared by chromatography on a bioaffin-
ity matrix that had the enzyme immobilized through its active site. This was
prepared by the attachment of chymotrypsin to pancreatic trypsin inhibitor antily-
sine, covalently linked to bead cellulose (Fig. 24). After periodate oxidation of
their carbohydrate moieties, the isolated antibodies were coupled to a hydrazide
derivative of bead cellulose and used for the sorption of chymotrypsin [5]. The
failure to detect any chymotryptic activity toward N-succinyl-I.-phenylalanyl-p-
nitroanilide showed that the enzyme was immobilized via its active site. No de-
crease in proteolytic activity resulted from incubation of chymotrypsin with spe-
cific antichymotrypsin IgG in a 1 : 1 molar ratio. These isolated antibodies are
thus suited for the preparation of a biospecific affinity matrix bearing immobilized
chymotrypsin oriented such that substrate proteins are accessible to its active
site. The antigenic affinity of the antichymotrypsin antibodies was essentially
unchanged by periodate oxidation of their carbohydrate moieties. Therefore, this
procedure was used to covalently attach isolated antichymotrypsin antibodies to
a hydrazide derivative of beaded cellulose. The molar ratio of biospecifically
adsorbed chymotrypsin molecules to immobilized antibody was 2 : 1, which is
the value predicted for enzyme occupancy of each of the two Fab binding sites
in IgG. The immobilized chymotrypsin retained practically 100% of the native
catalytic activity determined by use of the high molecular substrate, i.e., dena-
tured hemoglobin. Its proteolytic activity has not changed after more than a year.
G LTA
Figure 24 Schematic drawing (I) of oriented immobilization of chymotrypsin (CHT)

by use of covalent crosslinking of its active site to immobilized natural polyvalent trypsin
inhibitor antilysine (AL) with glutaraldehyde (GLTA) and (11) of use of its column for
isolation of immunoglobulin G (IgG) against chymotrypsin (with antigenic sites outside
the active site of CHT). Covalent bonds are shown as full lines. (Reproduced from Turkovi
1, et al. Macromol Chem Macromol Symp 1988;17:241-256)
In analogy to nonenzymatic heterogeneous catalysis, in which an important

role is played by the rate of diffusion of the reactants to the active surface of the
catalyst, the rate of diffusion of the substrates to the binding site of the enzyme
significantly affects the kinetic parameters of catalysis by an immobilized enzyme
system. Very efficient columns packed with I -pm nonporous spherical silica par-
ticles were described by Venema et a1.[117]. In order to eliminate the kinetic
limitation of chymotryptic hydrolysis of protein, porous bead cellulose was re-
placed by 1.2-pm nonporous hydroxyethyl methacrylate beads [ 1 181. Nonporous
carrier was prepared by radical dispersion copolymerization of 2-hydroxyethyl
methacrylate (HEMA) and ethylene dimethacrylate (EDMA) in the mixture of
alcohol-toluene. The polymerization was initiated by dibenzoyl periodate and
stabilized by cellulose butyrate. Isolated antichymotrypsin antibodies, attached
to nonporous beads containing 1.9 pmol of dihydrazide groups per 1 g of dry
product, were used for the coupling of antichymotrypsin antibodies. After the
adsorption of chymotrypsin on the prepared immunosorbent, 166.7 pg of chymo-
trypsin per gram of dry product was obtained (Fig. 25). Immobilized chymotryp-
sin retained practically 100% of the native proteolytic activity.
Monoclonal antibody (mAB) against horseradish peroxidase does not inter-
fere with its enzyme activity. It also possesses a high affinity toward the enzyme
158 Turkova
Figure 25 Schematic drawing of chyrnotrypsin immobilized by use of suitable antibod-

ies coupled through their carbohydrate moieties on bead nonporous poly(HEMA-co-
EDMG) beads with adipic acid dihydrazide.
and therefore was used for the preparation of a highly active immobilized enzyme
by Solomon et al. [ I 191. They prepared the complex of peroxidase and mAB in
solution and attached this imrnunocomplex to immobilized anti-Fc antibodies.
REFERENCES
1. Bayer EA, Wilchek M. Application of avidin-biotin technology to affinity-based

separations. J Chromatogr 1990;5 10:3- 11.
2. James MNG, Sielecki AR, Moult J. Crystallographic analysis of a pepstatin ana-
logue binding to the aspartyl proteinase penicillopepsin at 1.8 A resolution. In:
Hruby VJ, Rich DH, eds. Proceeding of the 8th American Peptide Symposium.
Rockford, IL: Pierce Chem 1983521-530.
3. Kobayashi J , Kusakabe I, Murakami K. Rapid isolation of microbial milk-clotting
enzymes by N-acetyl (or N-isobutyry1)-pepstatin-arninohexylagarose Anal Bio-
chem 1982;122:308-3 12.
Kay J, Afting EG, Aoyagi T, Dunn M. The effects of lactoyl-pepstatin and the
pepsin inhibitor peptide on pig cathepsin D. Biochem J 1982;203:795-797.
Bilkovi Z, ChuriEek J, Kuzerovi Z, Turkovi J. Purification of anti-chymotrypsin
antibodies for the preparation of a bioaffinity matrix with oriented chymotrypsin
as immobilized ligand. J Chromatogr 1997;689:273-279.
Winzor DJ. Review: recent developments in quantitative affinity chromatography.
J Chromatogr 1992;597:67-82.
Turkovi J. Bioaffinity Chromatography. 2nd ed. Amsterdam: Elsevier, 1993.
Turkovi J, Seifertovi A. Affinity chromatography of proteases on hydroxyalkyl meth-
acrylate gels with covalently attached inhibitors. J Chromatogr 1978;148:293-297.
Tosa T, Sato T, Sano R, Yamamoto K, Matuo Y, Chibata I. Characteristics and
applications of N-(w-aminohexy1)-L-asparticacid-Sepharose as an affinity adsor-
bent. Biochim Biophys Acta 1974;334:1-1 1.
Turkovi J, Fusek M, Maksimov JJ, Alakhov YuB. Reversible and irreversible im-
mobilization of carboxypeptidase Y using biospecific adsorption. J Chromatogr
1986:315-321.
Trayer IP, Trayer HR. Affinity chromatography of nicotinamide nucleotide-depen-
dent dehydrogenases on immobilized nucleotide derivatives. Biochem J 1974;141:
775-787.
Craven DB, Harvey MJ, Lowe CR, Dean PDG. Affinity chromatography on immo-
bilized adenosine 5'-monophosphate. Eur J Biochem 1974;41:329-340.
Thompson ST, Cass KH, Stellwagen E. Blue dextran-Sepharose: an affinity column
for the dinucleotide fold in proteins. Proc Natl Acad Sci USA 1975;72:669-672.
Biellmann JF, Samama JP, Briinden CI, Eklund H. X-Ray studies of the binding of Ciba-
cron Blue F3GA to liver alcohol dehydrogenase. Ew J Biochem 1979;102:107-1 10.
Vijayalakshrni MA, Bertrand 0, eds. Protein-dye interactions: developments and
applications. New York: Elsevier, 1989.
Phillips TM. Queen WD, More NS, Thompson AM. Protein A-coated glass beads:
a universal support media for high performance immunoaffinity chromatography.
J Chromatogr 1985;327:213-219.
Lee WC, Chuang CY. Performance of pH elution in high-performance affinity chro-
matography of proteins using non-porous silica. J Chromatogr A 1996;721:31-39.
Lis H, Sharon N. Affinity chromatography for the purification of lectins. J Chro-
matogr 1981;215:361-372.
Hatakeyama T, Murakami K, Miyamoto Y, Yamasaki N. An assay for lectin activ-
ity using microtiter plate with chemically immobilized carbohydrates. Anal Bio-
chem 1996;237: 188-1 92.
Kristiansen T. Group-specific separation of glycoproteins. Meth Enzymol 1974;
184:629-641.
Pergami P, Jaffe H, Safar J. Semipreparative chromatographic method to purify
the normal cellular isoform of the prion protein in nondenatured form. Anal Bio-
chem 1996;236:63-73.
Schott H. Affinity Chromatography: Template Chromatography of Nucleic Acids
and Proteins. Chromatographic Science Series. New York: Marcel Dekker, 1984.
Johnson AF, Wang R, Ji H, Chen D, Guilfoyle RA, Smith LM. Purification of
single-stranded M13 DNA by cooperative triple-helix-mediated affinity capture.
Anal Biochem 1996;234:83-95.
160 Turkova
Eisenberg S, Francesconi SC, Civalier C, Walker SS. Purification of DNA-binding pro-

teins by site-specific DNA affinity chromatography.Meth Enzymol 1990;182:521-529.
Phillips TM. Isolation and recovery of biologically active proteins by high perfor-
mance immunoaffinity chromatography. In: Kerlavage AR, ed. The Use of HPLC
in Receptor Biochemistry. New York: Alan R. Liss, 1989:129-154.
Cichna M, Knopp D, Niessner R. Immunoaffinity chromatography of polycyclic
aromatic hydrocarbons in columns prepared by the sol-gel method. Anal Chim Acta
1997; 339:241-250.
Bagnati R, Ramazza V, Zucchi M, Simonella A, Leone F, Bellini A, Fanelli R.
Analysis of dexamethasone and betamethasone in bovine urine by purification with
an "on-line" immunoaffinity chromatography-high-performance liquid chroma-
tography system and determination by gas chromatography-mass spectrometry.
Anal Biochem 1996;235: 119- 126.
Weber DV, Bailon P. Application of receptor-affinity chromatography to bioaffinity
purification. J Chromatogr 1990;510:59-69.
Finn FM, Hofmann K. Isolation and characterization of hormone receptors. Meth
Enzymol 1990;184:244-274.
Green NM. Thermodynamics of the binding of biotin and some analogues by avi-
din. Biochem J 1966:101 :774-780.
Wilchek K, Bayer EA. Avidin-biotin technology. Meth Enzyn~ol1990;184: 1-746.
Cuatrecasas P, Wilchek M. Single-step purification of avidin from egg white by
affinity chromatography on biocytin-Sepharose columns. Biochem Biophys Res
Commun 1968;33:235-239.
Fishel R, Anziano P, Rich A. Z-DNA affinity chromatography. Meth Enzymol
1990;184:328-340.
Grabowski PJ. Isolation and analysis of splicing complexes. Meth Enzymol 1990;
184:319-327.
Cook GMW, Buckie JW. Lectin-mediated isolation of cell surface glycoproteins.
Meth Enzymol 1990;184:304-3 14.
Rucklidge GJ, Milne G, Chaudhry M, Robins SP. Preparation of biotinylated, affin-
ity-purified antibodies for enzyme-linked immunoassays using blotting membrane
as an antigen support. Anal Biochem 1996;243:158-164.
Sharma SK, Mahendroo PP. Affinity chromatography of cells and cell membranes.
J Chromatogr 1980; 184:471-499.
Schnaar RL. Immobili~edglycoconjugates for cell recognition studies. Anal Bio-
chem 1984;143:1-13.
Kaplan LJ, Barr FG, Daims M, Nelson D, Tanner TB. The use of lectin affinity
chromatography for the selective isolation of plasma membranes. Prep Biochem
1984;14:149-161.
Braun RW. Kiimel G. Separation of T cell subpopulations by monoclonal antibod-
ies and affinity chromatography. Meth Enzymol 1986;121:737-748.
Yoon JW, Kenyon AJ, Good RA. Demonstration of Alatian mink disease virus
in cell culture. Nature New Biol 1973;245:205-207.
Dvorak DJ, Gipps E, Kidson C. Isolation of specific neurones by affinity methods.
Nature 1978;271:564-566.
Porath J, Sundberg L. High capacity chemisorbents for protein immobilization. Na-
ture New Biol l972;238:26 1-262.
Ohlson S, HanssonL, GladM, Mosbach K, LarsonPO. High performance liquidaffinity

chromatography: a new tool in biotechnology. Trends Biotechnol 1989;7:179-1 86.
Anspach FB, Wirth HJ, Unger KK, Stanton P, Davis JR, Hearn MTW. High-perfor-
mance liquid affinity chromatography with phenylboronic acid, benzaminidine, tri-
L-alanine and concanavalin A immobilized on 3-isothiocyanatopropylethoxysiIane-
activated nonporous monodisperse silicas. Anal Biochem 1989; 179: 17 1 - 18 1.
KuEerovi Z, Turkovi J. Isolation of human pepsin by use of membrane based bio-
adsorbent. Internatl J Bio-Chromatogr 1997;2: 145-15 1.
Amott S, Fulmer A, Scott WE, Dea ICM, Moorhouse R, Rees DA. The agarose
double helix and its function in agarose gel structure. J Mol Biol 1974;90:269-
284.
Gustavsson PE, Larsson PO. Superporous agarose, a new material for chromatogra-
phy. J Chromatogr 1996;734:23 1-240.
Nandedkar UN, Sawhney SY, Bhile SV. Kale NR. Sephacryl S-300-an affinity
matrix which distinguished concanavalin A from other D-mannose1~-glucose-spe-
cific lectins. J Chromatogr 1987;396:363-368.
Anspach FB, Johnston A, Winh HJ, Unger KK, Hearn MTW. High-performance
liquid chromatography of amino acids, peptides and proteins XCV. Thermody-
namic and kinetic investigation on rigid and soft affinity gels with varying particle
and pore sizes: comparison of thermodynamic parameters and the adsorption behav-
ior of proteins evaluated from bath and frontal analysis experiments. J Chromatogr
l99O;499:103- 124.
Hermanson GT, Mallia AK, Smith PK. Immobilized affinity ligand techniques. San
Diego: Academic Press, 1992.
Noppen C, Spagnoli GC, Schaefer C. Isolation of Multiple mRNAs from a few
eukaryotic cells: a fast method to obtain templates for RT-PCR. Bio techniques
1996;21:394-396.
Turkovi J, Hubilkovi 0 , Givikovi M, eoupek J. Affinity chromatography on
hydroxyalkyl methacrylate gels. I. Preparation of immobilized chymotrypsin and
its use in the isolation of proteolytic inhibitors. Biochim Biophys Acta 1973;322:1-9.
Turkovi J, Bliha K, HoriEek J, VajEner J, Frydrychovi A, eoupek J. Hydroxyalkyl
methacrylate gels derivatized with epichlorohydrin as suppons for large-scale and
high-performance affinity chromatography. J Chromatogr 198 1 ;215: 165- 179.
Matsumoto I, Ito Y, Seno N. Preparation of affinity adsorbents with Toyopearl gels.
J Chromntogr 1982;239:747-754.
Fleminger G, Wolf T, Hadas E, Solomon B. Eupergit C as a carrier for high-perfor-
mance liquid chromatographic-based immunopurification of antigens and antibod-
ies. J Chromatogr 1990;51 0 3 1 1-3 19.
Ugelstad J, Berge A, Ellingsten T, Aune 0 , Kilaas L, Nilsen TN, Schmid R, Stens-
tad P, Funderud S. Kvalheim G, Nustad K, Lea T, Vartal, F, Danielsen H. Mono-
sized magnetic particles and their use in selective cell separation. Makromol Chem
Macromol Symp 1988;17:177-211.
Miron T, Wilchek M. Polyacrylhydrazido-agarose: preparation via periodate oxida-
tion and use for enzyme immobilization and affinity chromatography. J Chromatogr
198 1 ;215:55-63.
Weetall HH, Lee MJ. Antibodies immobilized on inorganic supports. Appl Bio-
chem Biotech 1989:22:3 11-330.
162 Turkova
Ruhn P, Garver S, Hage DS. Development of dihydrazide-activated silica supports

for high-performance affinity chromatography. J Chromatogr 1994;669:9-19.
Klein E. Affinity Membranes: Their Chemistry and Performance in Adsorption
Separation Processes. New York: John Wiley and Sons, 1991.
Adachi T, Mogi M, Harada M, Kojima K. Selective removal of human serum amy-
loid P component from rat blood by use of an immunoaffinity membrane in an
extracorporeal circulation system. J Chromatogr 1996;682:47-54.
Huse K, Himmel M. Gartner G, Kopperschlager G, Hofmann E. Use of an activated
nylon membrane (Immunodyne) as an affinity adsorbent for the purification of
phosphofructokinase and phosphoglycerate kinase from yeast. J Chromatogr 1990;
502:171-177.
Zemanovi I, Turkovi J, kapka M, Nakhapetyan LA, Svec F, Kalal J. Effect of the
nature of proteins on their coupling to different epoxide-containing supports. En-
zyme Microb Techno1 1981;3:229-232.
Stambolieva N, Turkovi J. Covalent attachment of proteins to Spheron by means
of benzoquinone. Collect Czech Chem Commun 1980; 45: 1137-1 143.
Brandt J, Andersson LO, Porath J. Covalent attachment of proteins to polysac-
charide carriers by means of benzoquinone. Biochim Biophys Acta 1975;386: 196-
202.
Smalla K, Turkovi J, koupek J, Hermann P. Influence of salts on the covalent
immobilization of proteins to modified copolymers of 2-hydroxyethyl methacrylate
with ethylene dimethacrylate. Biotechnol Appl Biochem 1988;10:21-31.
Wheatley JB, Lyttle MH, Hocker MD, Schmidt DE Jr. Salt-induced immobilization
of DNA oligonucleotides on an epoxide-activated high-performance liquid chro-
matographic affinity support. J Chromatogr A 1996;726:77-90.
i Vanzurovi D, Svec F, Kilal J. Methacrylate
Turkovi J, Bliha K, ~ a l a n i k o v M,
gels with epoxide groups as supports for immobilization of enzymes in pH range
3-12. Biochim Biophys Acta 1978;524: 162-169.
Staudenbauer WL, Orr E. DNA gyrase: affinity chromatography on novobiocin-
Sepharose and catalyic properties. Nucleic Acid Res 1981;9:3589-3603.
O'Shannessy DJ. Hydrazido-derivatized supports in affinity chromatography. J
Chromatogr l99O;5 10: 13-2 1.
Van Sommeren APG, Machilsen PAGM, Gribnau TCJ. Comparison of three acti-
vated agaroses for use in affinity chromatography: effects on coupling performance
and ligand leakage. J Chromatogr 1993;639:23-31.
Lamed R, Levin Y, Wilchek M. Covalent coupling of nucleotides to agarose for
affinity chromatography. Biochim Biophys Acta 1973;304:231-235.
Turkova J, Kuzerova Z, Vaiikovi H, BeneS MJ. Stabilization and oriented immobi-
lization of glycoproteins. Internatl J Bio-Chromatogr 1997;3:45-55.
Petkov L. Sajdok J, Rae K, h c h o v i M, KG J, Turkov9-J. Activation of galactose-
containing glycoprotein and solid supports by galactoe oxidase in presence of cata-
lase for immobilization purposes. Biotechnol Techniques 1990;4:25-30.
Solomon B, Koppel R, Schwartz F, Fleminger G. Enzymatic oxidation of mono-
clonal antibodies by soluble and immobilized bifunctional enzyme complexes. J
Chromatogr 19905 10:32 1-329.
Kelleher FM, Dubbs SB, Bhavanandan VP. Purification of galactose oxidase from
r
Dacryliurn dendroides by affinity chromatography on melibiose-polyacrylamide.

Arch Biochem Biophys 1988;263:349-354.
Sanderson CJ, Wilson DV. A simple method for coupling proteins to insoluble
polysaccharides. Immunology 197 1 ;20: 106 1- 1065.
Koelsch R, Fusek M, Hostomski Z, Lasch J, Turkovi J. Coupling of ligands to
insoluble matrices via aldehyde groups. Biotechnol Lett 1986;8:283-286.
Weston PD, Avrameas S. Proteins coupled to polyacrylamide beads using glutaral-
dehyde. Biochem Biophys Res Commun 1971;45:1574-1580.
Cambiaso CL, Goffinet A, Vaerman JP, Heremans JF. Glutaraldehyde-activated
aminohexyl-derivative of Sepharose 4B as a new versatile immunosorbent. Immu-
nochemistry 1975;12:273-278.
Fusek M, b p k a M, Turkovi J. Immobilization and nonspecific adsorption of pro-
teins to pyrogeneous highly dispersed silicon dioxide. Biotechnol Lett 1987;9:561-
566.
Jirki V, Turkovi J. Cell immobilization by covalent linkage. Meth Enzymol 1987;
135:341-357.
Monsan P, Puzo G, Mazarquil H. ~ t u d due mkcanisme d'ktablissement des liaisons
glutaraldkhyde-protCines. Biochimie 1975;57: 1281- 1292.
AxCn R, Porath J, Emback S. Chemical coupling of peptides and proteins to poly-
saccharides by means of cyanogen halides. Nature 1967;2 14: 1302- 1304.
Kohn J, Wilchek M. Procedures for the analysis of cyanogen bromide-activated
Sepharose or Sephadex by quantitative determination of cyanate esters and imido-
carbonates. Anal Biochem 1981;1 15:375-382.
Wilchek M, Miron T. Origin of the carbamate functional groups in cyanogen bro-
mide-activated, alkylamine-substituted Sepharose. J Chromatogr 1986;357:315-
317.
Boschetti E, Corgier M, Garelle R. Immobilization of ligands for affinity chroma-
tography. A comparative study of two condensation agents: 1 -cyclohexyl-3-(2-mor-
pholinoethy1)-carbodiimide-metho-p-toluene sulfonate (CMC) and N-ethoxycarbo-
nyl-l,2-dihydroquinoline(EEDQ). Biochimie 1978;60:425-427.
Cuatrecasas P, Parikh I. Adsorbents for affinity chromatography. Use of N-Hydro-
xysuccinimide ester of agarose. Biochemistry 1972; 11:229 1-2299.
Bethell GS, Ayers JS, Hancock WS, Hearn M W . A novel method of activation
of cross-linked agaroses with 1.1'-carbonyldiimidazole which gives a matrix for
affinity chromatography devoid of additional charged groups. J Biol Chem 1979;
25412572-2574.
Crowley SC, Chan KC, Walters RR. Optimization of protein immobilization on
1,l'-carbonyldiimidazole-activated diol-bonded silica. J Chromatogr. 1986;359:
359-368.
Kay G, Crook EM. Coupling of enzymes to cellulose using chloro-s-triazines. Na-
ture 1967;216:514-515.
BeneS MJ, Adimkovi K, Turkova J. Activation of beaded cellulose with 2,4,6-
trichlortriazine. Bioactiv Compat Polym 1991;6:406-4 13.
Carlsson J, AxCn R, Unge T. Reversible, covalent immobilization of enzymes by
thiol-disulphide interaction. Eur J Biochem 1975;59:567-572.
Campbell DH, Luescher EL, Lerman LS. Immunologic adsorbents I. Isolation of
Turkova
antibody by means of a cellulose-protein antigen. Proc Natl Acad Sci USA 1951;
37575-578.
Nilsson K, Mosbach K. Immobilization of enzymes and affinity ligands to various
hydroxyl group carrying supports using highly reactive sulfonyl chlorides. Biochem
Biophys Res Commun 1981 ;1O2:449-457.
Porath J, Sundberg L. High capacity chemisorbents for protein immobilization. Na-
ture New Biol 1972;238:26 1-262.
Ngo TT. Facile activation of Sepharose hydroxyl groups by 2-fluoro-l methylpyri-
dinium toluene-4-sulfonate: preparation of affinity and covalent chromatographic
matrices. Bio/Technology 1986;4:134- 137.
Ekberg B, Mosbach K. Molecular imprinting: a technique for producing specific
separation materials. Trends Biotechnol 1989;7:92-96.
Wulff G, Vesper W. Preparation of chromatographic sorbents with chiral cavities
for racemic resolution. J Chromatogr 1978; 167: 17 1 - 186.
Sellergren B, Ekberg B, Mosbach K. Demostration of substrate-and enantio-selec-
tivity by chromatographic resolution of racemic mixtures of amino acid derivatives.
J Chromatogr 1985;347:1 - 10.
Whitcombe MJ, Rodriguez ME. Villar P, Vulfson EN. A new method for the intro-
duction of recognition site functionality into polymers prepared by molecular im-
printing: synthesis and characterization of polymeric receptors for cholesterol. J
Am Chem Soc l995;l l7:7l05-7lll.
Mosbach K. The emerging technique of molecular imprinting and its future impact
on biotechnology. 12th International Symposium on Affinity Interaction, Kalmar.
Sweden, June 15-19, 1997, L15.
Kukongviriyapan V, Poopyruchpong N, Ratanabanangkoon K. Some parameters
of affinity chromatography in the purification of antibody against Naja naja sia-
mensis toxin 3. J Immunol Meth 1982;49:97-104.
Cuatrecasas P, Wilchek M. Single-step purification of avidin from egg white by
affinity chromatography on biocytin-Sepharose columns. Biochem Biophys Res
Commun 1968;33:235-239.
Gierasch LM, Lacy JE, Thompson KF, Rockwell AL, Watnick PI. Conformation
of model peptides in membrane-mimetic environments. Biophys J 1982;37:275-
284.
Angal S. Dean PDG. The effect of matrix on the binding of albumin to immobilized
Cibacron blue. Biochem J 1977;167:301-303.
Clemetson KJ, Pfueller SL, Luscher EF, Jenkins CSP. Isolation of the membrane
glycoproteins of human blood platelets by lectin affinity chromatography. Biochim
Biophys Acta 1977;464:493-508.
Turkovi J, Bliha K, Adamovi K. Effect of concentration of immobilized inhibitor
on the biospecific chromatography of pepsins. J Chromatogr 1982;236:375-383.
Liu YC, Stellwagen E. Accessibility and multivalency of immobilized Cibacron
Blue F3GA. J Biol Chem 1987;262:583-588.
Narayanan SR. Preparative affinity chromatography of proteins. J Chromatogr A
1994;658:237-258.
Kohler G, Milstein C. Continuous cultures of fused cells secreting antibody of pre-
defined specificity. Nature 1975;256:495-497.
r
Tarnowski SJ, Liptak RA. Automated immunosorbent purification of interferon.

In: Mizrahi A, van Wezel AL, eds. Advances in Biotechnical Processes. New York:
Alan R. Liss, l983:27 1-287.
Lee SM, Gustafson ME, Pickle DJ, Flickinger MC, Muschik GM. Morgan AC.
Large-scale purification of a murine antimelanoma monoclonal antibody. J Biotech-
no1 1986;4: 189-204.
Shami EY, Rothstein A, Ranljeesingh M. Stabilization of biologically active pro-
teins. Trends Biotechnol l989;7: 186- 190.
Sheriff S, Silverton EA, Padlan EA, Cohen GH, Smith-Gill SJ. Finzel BC, Davis
DR. Three-dimensional structure of an antibody-antigen complex. Proc Natl Acad
Sci USA 1987;84:8075-8079.
Venama E, Kraak JC, Poppe H, Tijssen R. Packed-column hydrodynamic using 1-
p n non-porous silica particles. J Chromatogr 1996;74O:159- 167.
Turkovi J, Bilkovi Z, Mazurova J, Horak D. Oriented immobilization of chymo-
trypsin by use of suitable antibodies coupled to porous or non-porous solid support.
12th International Symposium on Affinity Interaction: Fundamentals and Applica-
tions of Biomolecular Recognition, Kalmar, Sweden, June 15- 17, 1997, PA34.
Solomon B, Hadas E, Koppel R, Schwartz F, and Fleminger G. Highly active en-
zyme preparations immobilized via matrix-conjugated anti-Fc antibodies. J Chro-
matogr 1991;539:335-341.
Characterization and Partial
Purification of Steroidogenic
Factors from Thymic Epithelial
Cell-Conditioned Medium
Mehmet Uzumcu
King Faisal Specialist Hospital and Research Center, Riyadh,
Saudi Arabia
David R. Brigstock
Ohio State University and Children's Hospital, Columbus, Ohio
Young C. Lin
Ohio State University, Columbus, Ohio
I. INTRODUCTION
The thymus gland, besides its well-recognized role in the immune system,
plays a role in reproductive biology, particularly in regulating ovarian function.
Studies using neonatally thymectomized mice or congenitally athymic nude mice
( n u h u ) have clearly shown that thymus gland influences ovarian function, al-
though the mechanism for this influence is not completely known [I]. In this
chapter, we shall summarize the experimental evidence that supports the regula-
tion of the ovaries by the thymus gland and the mechanisms that might be in-
volved. We further describe our characterization and partial purification of two
potentially novel thymic factors that directly influence ovarian steroidogenesis
in vitro.
Uzumcu et al.
II. THYMUS AND OVARIES

A. Influence of the Thymus on the Ovaries
Establishment of a normal reproductive system depends on presence of an intact
thymus gland since absence of the thymus causes various reproductive disorders.
Neonatal thymectomy in certain strains of mice causes ovarian dysgenesis and
sterility in females [2]. Similarly, congenitally athymic nude mice ( n u h u ) have
an accelerated follicular atresia and premature ovarian failure [3]. Both congeni-
tally athymic and neonatally thymectomized mice show delays in vaginal opening
and onset of the first estrus [3]. The time of the neonatal thymectomy appears
to be critical based on the fact that while thymectomy on day 3 of life causes
ovarian dysgenesis, thymectomy on or after day 7 does not cause any apparent
ovarian anomaly [2]. Studies have indicated that ovarian histology of neonatally
thymectomized mice is similar to that of their littermates at birth. A progressive
ovarian dysgenesis starts at day 20 whereby ovary is gradually depleted from
follicular structures and invaded by lymphocytes. Ovarian follicles and corpora
lutea are then gradually replaced with interstitial tissue and the ovary is atrophied
by day 120 [4]. Ovarian dysgenesis following neonatal thymectomy is also ob-
served in other species including rats and primates [5-71. Endocrine disturbances
along with histological disorders are also observed in neonatally thymectomized
mice. These disturbances include but are not limited to high levels of androgens
[81 and low levels of progesterone and estradiol [9] which are attributed to the
destruction of follicular structures and their replacement by interstitial tissue.
B. Proposed Mechanisms for Influence of Thymus

on the Ovaries
An explanation suggested for the ovarian dysgenesis in neonatally thymecto-
mized mice is based on the premise that self-reactive CD4+ T cells are continu-
ously produced from thymus and released to the periphery. In euthymic animals,
ovarian dysgenesis is not observed because these autoreactive CD4+ T cells were
controlled by another CD4+ subset with regulatory or suppressive activity [lo].
The suppressive CD4+ cells are also generated from thymus but only after age
3 or 4 days. Thymectomy on day 3 of life causes the shift in the balance between
self-reactive and regulatory CD4+ populations to favor the former and explains
ovarian dysgenesis as a spontaneously occurring autoimmune disease [lo]. Ovar-
ian dysgenesis in thymectomized animal is often associated with other autoim-
mune disorders such as thyroiditis, gastritis, and parotitis [I 1 - 131.
That ovarian dysgenesis is directly due to autoimmunity was shown by
transfer experiments by which the disease was transferred by CD4+ T cells from
thymectomized animals to young recipients [14]. In addition, the transfer of ovar-
Steroidogenic Factors and TCM 169
ian dysgenesis was prevented by transfer of CD4+ CD5 + T cells [15]. Further-
more, autoreactive antibodies to the ovary have been shown in neonatally thymec-
tomized mice [16]. However, these antibodies were detected only after the age
of 30 days, which fails to explain the initial destruction at the ovary prior to day
30. The possible explanation of this failure is that disturbances of the ovary prior
to day 30 are due to the effect of thymectomy at the level of the hypothalamic-
pituitary axis. Specifically, the development of the hypothalamus remains incom-
plete if the thymus is absent or removed at an early age. Therefore, the underde-
veloped hypothalamus cannot release sufficient luteinizing hormone-releasing
hormone (LHRH), which in turn causes gonadotropin deficiency and ovarian dys-
genesis later in the life. Experimental findings using congenitally athymic mice
support this hypothesis. This strain of mice exhibits decreased levels of luteiniz-
ing hormone (LH) and follicle-stimulating hormone (FSH) in the pituitary gland
and in the circulation prior to vaginal opening (i.e., puberty), along with reduced
serum estrogen [17]. In addition, a reduced level of LHRH in hypothalami of
athymic nude mice was observed while the LH response of these animals to
exogenously administered LHRH in vivo was identical to that of their heterozy-
gote (nu/+) littermates [18]. Previously, Rebar and coworkers demonstrated that
reduced LH and FSH levels in athymic mice could be restored to normal by
neonatal thymic implantation 1191. Finally, it has been demonstrated that thy-
mosin fraction-5 (TF-5) and thymosin P4 stimulates LHRH from medial basal
hypothalami from normal cycling female rats superfused in vitro [20]. Thus, the
evidence from neonatally thymectomized and congenitally athymic mice suggests
that the influence of thymus on the reproductive functions is due to at least two
mechanisms:
1. Autoimmune. Removal of the thymus at an early age causes develop-
ment of autoimmune antibodies against functional structures at the
ovary due to the disturbance of the delicate balance at the immune
system.
2. Endocrine. Hormonal interactions between the thymus and hypothala-
mic-pituitary-ovarian axis exist. Since this interaction is disrupted in
neonatally thymectomized or athymic mice, ovarian dysgenesis devel-
ops later in the life.
In addition to these two separate but interrelated mechanisms, interaction
between the thymus and the ovaries may be partly through the direct effect of
secretory products of the thymus (i.e. thymic hormones) on the ovary without
any involvement of hypothalamus-pituitary axis or immune system. A consider-
able body of experimental evidence shows that there is a direct influence of vari-
ous immune products (e.g., interleukins and/or other cytokines) on ovarian func-
tion [21].
Uzumcu et al.
Ill. STUDIES WITH THYMIC EPITHELIAL

CELL-CONDITIONED MEDIUM
When we initially hypothesized that thymic factor(s) may directly influence ovar-
ian function, there were very few previously published studies in this area [22,23].
Since it was known that thymic epithelial cells are the main source of the thymic
hormones [24] and thymic hormones are produced by the epithelial cells in cul-
ture [25], we began our investigation in this area by examining the influence of
thymic epithelial cell-conditioned medium (TCM) on rat ovarian granulosa cell
(RGC) steroidogenesis in vitro [261.
A. Preparation of TCM
TCM was prepared as previously described [27]. Briefly, thymi from 21- to 23-
day-old rats were removed and minced into small pieces. The tissue fragments
were rinsed and centrifuged (50 g for 1 min) 5 times in 40 ml Hank's Calcium-
Magnesium Free Balanced Salt Solution (HCMF) to remove most of the lympho-
cytes. The remaining tissue fragments were digested with a mixture of trypsin
(0.1 25%) and ethylenediaminetetraacetic acid (EDTA) (0.01 %) in HCMF at 37OC
with stirring for 45 min. The released thymic cells were washed, counted, and
plated in 25 cm2 culture flasks at 2 X lo6cells/ml density. The culture medium
used was Minimum Essential Medium D-Valine modification (MEM) supple-
mented with 10% FBS. After 24 h in culture, the medium was replaced and
unattached cells (mostly lymphocytes) were removed. The attached cells were
allowed to grow for an additional 3 days and the medium was replaced with fresh
medium. The cells were cultured for two additional 5 day periods and the media
from these cultures were collected, centrifuged, and stored at -20°C until used
for evaluation. Heart cell-conditioned medium (HCM) and mock extract (ME)
were used as control media throughout these studies. HCM was prepared from
hearts of 21 - to 23-day-old rats similar to TCM. To alleviate the potential nonspe-
cific effects of FBS components, ME was prepared in an identical fashion as
TCM or HCM, but in cell-free culture dishes.
B. Preparation of RGC
Granulosa cells were prepared from 21- to 23-day-old diethylstilbestrol (DES)-
treated Sprague-Dawley female rats as previously described [27]. Briefly, granu-
losa cells were washed and plated at a density of 4 X l o 5 viable cells per well
in 24-well tissue culture plates containing 1 ml serum-free medium. The cells
were first cultured for 24 h at 37°C before TCM treatment.
C. Effect of TCM on Steroidogenesis from RGC
Steroidogenic activity of TCM was tested in time-course and dose-response
studies. TCM stimulated both progesterone (P4) and estradiol (E2) secretion from
Steroidogenic Factors and TCM
E+Zm No FSH A
FSH (100 nglml)
Q
m
Control --
ME HCM TCM Control -ME HCM TCM
10% conditioned media 10% conditioned media
Figure 1 Effect of TCM on basal and FSH-induced P, (A) and E2 (B) secretions from
cultured RGC as determined by radioimmunoassay (RIA). Data points are shown as the
mean 5 SD of picogram quantity of the steroids per microgram of cell protein from qua-
druplicate-culture wells in a representative experiment that was repeated 3 times with
similar results. The control represents the granulosa cells incubated with culture medium
containing no conditioned medium. (Modified from Ref. 28.)
RGC in a time- and dose-dependent manner [27]. Later, we observed 80-and 17-
fold increases in basal and FSH-induced P, secretion from rat granulosa cells
treated with 10% TCM for 48 h. (The treatment length was 48 h throughout the
entire study.) In addition, TCM stimulated basal and FSH-induced E2 secretion
about four- and threefold, respectively [28] (Fig. 1). Similarly, 10% TCM aug-
mented basal and FSH-induced 20a-hydroxyprogesterone (OH-P,) approxi-
mately 40- and 10-fold. Basal and FSH-stimulated aromatase enzyme activity
was also upregulated by TCM. The stirnulatory effect of TCM on steroidogenesis
from RGC was not due to cellular proliferation or growth since the TCM Ireat-
ment increased neither total cellular protein content nor [%]thymidine incorpora-
tion to the RGC [28].
D. Physicochemical Characterization of TCM

Boiling at 100°C for 20 min, acidification of pH to 3 by 1 N HCI followed by
neutralization, and acetone treatment completely eliminated the P4 stirnulatory
Uzumcu et al.
Physicochemicaltreatments
Figure 2 Effect of 25% TCM before or after various physicochemical treatments

on P, secretion in cultured RGC as determined by RIA. Data points are shown as the
mean t SD of nanogram quantity of P, per milliliter of medium from quadruplicate-
culture wells (n = 2). The conditions for various physicochemical treatments are as de-
scribed in Section 1I.D. (Adapted from Ref. 27.)
action of TCM, while charcoal sedimentation or heating at 56OC for 30 min

caused minimal loss of activity [27] (Fig. 2). In addition, HPLC-purified active
fractions of TCM lost about 80% of their original activity after one freeze-thaw
cycle [29], which is sometimes characteristic of peptide samples. These results
showed that stimulatory activity of TCM was heat-and acid-labile and could be
destroyed by acetone treatment or freeze-thaw cycles, but could not be sedi-
mented by charcoal. It was tentatively concluded that a peptide or protein fac-
t o r ( ~ )was responsible for the biological activity in TCM.
IV. STUDIES WITH GEL FILTRATION CHROMATOGRAPHY

FRACTIONS OF THE TCM
TCM was fractionated by gel filtration chromatography and the column fractions
were tested for their effects on cultured RGC steroidogenesis and morphology
[29]. For gel filtration chromatography, TCM was lyophilized using a freeze-
dryer (FTS Systems, Inc., Stone Ridge, NY) and reconstituted in 25% of its origi-
nal volume with distilled water. The samples were clarified by centrifugation,
filtered using 0.45-ym membrane, and 1 ml of TCM concentrate was applied to
TSK-Gel G2000 SW gel filtration column (0.8 X 30 cm; TosoHaas, Philadelphia,
PA, USA) equipped with a TSK-GSW guard column (0.8 X 4 cm; TosoHaas).
A fast protein liquid chromatography (FPLC; Pharmacia, Piscataway, NJ, USA)
system was used to maintain the flow rate at 0.5 mllmin during elution of the
proteins with PBS containing 0.5 M NaCI. The column eluate was monitored at
280 nm using an in-line UV detector and 200-yl fractions were collected. Frac-
tions were stored at -70°C prior to assay. The molecular weight of the fractions
was calculated by reference to the elution positions of ovalbumin (45 kDa), tryp-
sin inhibitor (20.1 kDa), lactalbumin (14.2 kDa), and epidermal growth factor (6
kDa). Similarly, HCM and ME concentrates were also subjected to the gel filtra-
tion FPLC, and the collected fractions were also used as controls. Gel filtration
FPLC of TCM resulted in the elution of several major protein peaks as detected
at 280 nm [29] (Fig. 3).
A. Effect of TCM Gel Filtration Fractions

on Steroidogenesis in RGC
When individual fractions were tested for their ability to stimulate E2 and P4
production by RGC, two distinct peaks of activity were observed [29] (Fig. 3).
The first peak of activity was eluted in fractions 50-60 that corresponded to M,
- 22,000 ("TCM-22") and stimulated P4 production. The peak fraction (no. 53)
stimulated P4 production approximately 10-fold but had no effect on E2 produc-
tion. The second peak of steroidogenic activity was eluted in fractions 84-94,
which corresponded to M, - <I000 ("TCM-I "), and stimulated both the pro-
duction of P, and E2. The peak fraction of TCM-1 stimulated P, and E2 approxi-
mately 15- and 10-fold, respectively. Production of 20a-OH-P4 from RGC was
also stitnulated by both TCM- I and TCM-22. In addition. TCM-1 but not TCM-
22 stimulated aromatase enzyme activity in cultured RGC. However. fractions
collected from gel filtration FPLC and HCM and ME were shown to be inactive
~91.
B. Morphological Changes Caused by TCM-1 in RGC

Phase contrast microscopic examination revealed that TCM-I but not TCM-22
caused drastic morphological alterations in cultured RGC that resembled morpho-
logical alterations observed following FSH treatment of RGC [29] (Fig. 4), or
human chorionic gonadotropin (hCG)/cAMP treatment of human granulosa cells
[30]. Following their treatment with FSH or TCM-1, RGC became rounded and
left finger-like processes attached to the substratum. In contrast, the cells treated
with TCM-22 or control medium appeared well spread and polygonal [29] (Fig.
Uzumcu et al.
30 40 50 60 70 80 90 100 110
Fraction numbers
Figure 3 Effect of TCM gel filtration fractions on P4 (A) and E, (e) secretion from
cultured RGC as determined by RIA. The figure also shows the absorbance of the column
eluate at 280 nm (a).In this experiment, control group is the granulosa cells that were
treated with culture medium with no gel filtration fraction. The elution positions of molecu-
lar weight marker proteins (a) ovalbumin (45 m a ) , (b) trypsin inhibitor (20.1 m a ) , (c)
lactalbumin (14.2), and (d) epidermal growth factor (6 kDa) are indicated at the top
(arrows). Data points for the steroids are expressed as picograms per microgram of cell
protein from single well. The experiment was repeated for at least 5 times with similar
results. (Modified from Ref. 29.)
Figure 4 Morphology of the RGC following a I-h treatment with FSH (500 nglml) (A),
TCM-1 (B), and control medium or TCM-22 (C). The cells that were treated with FSH or
TCM-I became rounded (mows) and left finger-like processes (arrowheads) attaching sub-
stratum. The cells that were treated with control medium or TCM-22 were well spread and
polygonal in shape. Magnifications of the micrographs are 720X. (Adapted from Ref. 29.)
Uzumcu et al.
Fraction Numbers
Figure 5 Effect of TCM-I and TCM-22 alone or in combination on steroidogenesis
from RGC. Gel filtration fractions containing 90 yl TCM-I (fractim 83-94; triangles)
and TCM-22 (fraction 50-61; squares) were tested individually or 45 yl of successive
fractions from each activity region were combined (i.e., no. 50 with no. 83, no. 5 1 with
no. 84, etc.; circles) and tested for their combined effects on P4 (A) and E, (B) secretion
from RGC. In this experiment, control represents the granulosa cell that were treated with
culture medium containing PBSIO.5 M NaCI. The steroids were measured by RIA in cul-
ture medium and expressed as the mean + SD of picogram quantity of the steroids per
microgram of cell protein in single-culture well (n = 3). (Modified from Ref. 29.)
4). The morphological changes that were induced by either FSH or TCM-1 started
approximately 15 min after treatment and were most prominent by 1 h post treat-
ment. The cells started gaining their normal appearance by 12 h and returned to
normal by 48 h post treatment.
C. Interaction Among TCM-1, TCM-22, and FSH

in Stimulating Steroidogenesis in RGC
To gain a better understanding of the actions of TCM-l and TCM-22, the com-
bined actions of these two activity regions were tested on P4 and E2 production
from RGC. In this experiment, granulosa cells were treated with 90 p1 of each
factor alone or 45 p1 of 12 successive TCM-1 fractions (83-94) that were com-
bined with 45 p1 of 12 successive TCM-22 fractions (50-61). In combination,
TCM-1 and TCM-22 caused significantly more P4 production than that caused
by each factor alone, suggesting a synergistic interaction between the two factors
[29] (Fig. 5). However, for Ez production the stimulatory action of TCM-I was
completely antagonized by TCM-22. Only TCM-1 was stimulatory, whereas
TCM-22 alone had no effect on E2 production. Collectively, these results sug-
gested that these two factors may be unrelated and have a different mechanism
of action. In addition, the lack of stimulation by TCM-22 on Ez production
and its antagonistic action on E2stimulated by TCM-I partially explains why un-
fractionated TCM had relatively minor action on basal E2 production as com-
pared to its highly significant action on P4 secretion (4-fold versus 80-fold stimul-
ation) [28].
To assess the interactions between TCM-1 or TCM-22 and FSH, dose-
response curves were generated for TCM-1 and TCM-22 in the presence or ab-
sence of 100 nglml FSH. Both TCM-1 and TCM-22 stimulated basal and FSH-
induced P4 secretion in a dose-dependent manner. While TCM- 1 stimulated basal
and FSH-induced Ez secretion, TCM-22 inhibited FSH-induced E2 secretion and
had no effect on basal E2 [29] (Fig. 6).
V. STUDIES WITH REVERSED-PHASE HPLC-PURIFIED

TCM-1 AND TCM-22
To further purify TCM-1 and TCM-22, reversed-phase HPLC was performed on

a Hitachi (San Jose, CA, USA) HPLC system using a Microsorb-MV C Rcolumn
(0.46 X 25 cm, 5 pm particle size; Rainin Instrument Co., Inc., Woburn, MA,
USA) that was equilibrated with either 0% or 10% (v/v) acetonitrile, both con-
taining 0.1% (vlv) trifluoroacetic acid (TFA) [29]. The eluate from the gel filtra-
tion column was divided into two pools comprising fractions 51-59 for TCM-
22 or fractions 85-93 for TCM-1. For each HPLC analysis, the samples were
7
Uzurncu et al.
0 10 100 0 10 100
TCM-1 (pllml) TCM-22 (pllml)
Figure 6 Effect of TCM-I or TCM-22 on basal and FSH-induced Pq (upper panels)

and E2(lower panels) production by RGC. The granulosa cells were cultured with different
amounts of TCM-1 (A and C) or TCM-22 (B and D) in the absence (A) or presence (0)
of 100 nglml FSH. After 48 h of treatment. the steroid concentrations in the medium were
determined by RIA. Data points represent mean 2 SD of the mean picogram quantity of
the steroids per microgram of cell protein from triplicate culture wells (n = 3). (Modified
from Ref. 29.)
adjusted s o as to contain 0.1% T F A and the same acetonitrile concentration as

the equilibration buffer.
For TCM-22, conditions for the elution of column-bound proteins were
10% acetonitrile from 0 to 10 min after sample injection, 10-20% acetonitrile
from 10 to 4 0 min, 20-85% acetonitrile from 4 0 to 125 rnin, 85-95% acetonitrile
from 125 to 126 min, and 95% acetonitrile from 126 to 135 min. Acetoni-
trile concentrations for HPLC of TCM-1 were 0 % from 0 to 50 min after
sample injection, 0-90% from 50 to 120 min, and 90% from 120 to 130 min. The
flow rate was 1 mllmin throughout, and the chromatogram (Azl4)was archived as
previously described [311. The column eluate was collected into tubes containing
50 p1 0.12510.25 N NaOH to immediately neutralize the TFA. For TCM- 1, frac-
tions of 500 p1 were collected 50 min after sample injection and 10-p1 aliquots
were tested for their stimulation of steroidogenesis or morphological changes in
cultured RGC. For TCM-22, fractions of 1 ml were collected immediately after
sample injection, and 10 p1 was tested for stimulation of P4 and E2secretion from
RGC. The remaining portion of the fractions was stored at -80°C until further
analyzed for peptide sequence of TCM- 1 or re-HPLC of the TCM-22.
A. Effect of Reversed-Phase HPLC-Purified TCM-1

on RGC
TCM-1 activity (i.e., stimulation of P,, E2, 20a-OH-P4, and aromatase activity,
and morphological changes) was eluted from the C8 column with a retention
time of 92-93 min in fractions 85-87 [29]. The peak HPLC fraction (no. 86)
stimulated P4 and E2 production about 59- and 31-fold, respectively [29] (Fig.
7). The remaining fractions did not significantly affect steroidogenesis or mor-
phology of the RGC.
B. Effect of Reversed-Phase HPLC-Purified TCM-22

on RGC
TCM-22 activity was eluted from the C8 column as two peaks of activity that
had a retention time of 55-56 min (fractions 55-56) and of 58-59 min (fractions
58-59). Peak fractions 55 and 59 stimulated P4 secretion approximately three-
to fourfold over the control and had no effect on E2 (Fig. 8). The remaining
fractions did not significantly stimulate steroidogenesis by RGC. In view of major
contaminating proteins in the activity regions, the peak fractions were subjected
to a second round of reversed-phase HPLC under identical conditions to those
described above. Unfortunately, it was not possible to further to purify the TCM-
22 samples and therefore their complete isolation was not possible (data not
shown). In the future, alternative chromatographic strategies will be required for
further purification and identification of TCM-22.
C. Peptide Sequence Analysis of TCM-1

Since elution of TCM-1 activity from the C8 column was correlated with an A214-
absorbing entity the peak fraction was concentrated using a Speed-Vac con-
centrator (Savant Instruments, Farmingdale, NY) and submitted for N-terminal
amino acid sequencing. However, three separate attempts at N-terminal sequenc-
Uzumcu et al.
80 85 90
2 Fraction numbers
a
-
0.00
0 20 40 60 80 100 120 140 160
l " " l " " l " " l " " l " " l " " l " " ~ ' ' ' ~ ~
50 60 70 80 90 100 110 120 130
Fraction numberslRetention time (min.)
Figure 7 Effect of reversed-phase HPLC-purified TCM-I on P, and E2production from

RGC. Gel filtration fractions 85-93 from 13 runs were pooled and subjected to C8 re-
versed-phase HPLC. The eluate was collected into 0.5-1111 fractions 50 min after injection
of sample. The figure shows the absorbance (214 nm) of the eluate collected between 50
and 115 min after sample injection and the effect of fraction 80-90 on P4 (A) and E2 (+)
production from cultured RGC as determined by RIA (inset). A 10+1 aliquot was tested
from fraction 46-1 33. In this experiment, the control group is the granulosa cells that were
treated with culture medium with no addition of HPLC fraction. Data points presented in
the steroidogenic activity curves are picogram quantity of the steroids per microgram of
cell protein in a single-culture well. The experiment was repeated for three times with
similar results. The arrow indicates the peak that contains TCM-I activity. (Modified from
Ref. 29.)
ing of HPLC-purified TCM- I failed to produce a readable sequence, suggesting

either that insufficient peptide was analyzed or that it was N-terminally blocked.
In the future, alternative strategies such as enzymatic or chemical cleavage to
produce peptide fragments will be required for the successful identification of
TCM- I.
VI. CONCLUSIONS
Here we have described the characterization and partial purification of steroido-

genic factors in TCM. It has been long recognized that the thymus has an influ-
r
-'-
2 45 50 55 60 65
3 Fraction numbers
-a
20 30 40 ' 50 60 70 80 90 100
Fraction numbersIRetention time (min.)
Figure 8 Effect of reversed-phase HPLC-purified TCM-22 on P, and E, production

from RGC. Gel filtration fraction 51-59 from three runs were pooled and subjected to
C8 reversed-phase HPLC. The eluate was collected into 1.0-ml fractions starting with the
injection of sample. The figure shows the absorbance (214 nm) of the eluate collected
between 20 and 100 rnin after sample injection and the effect of fraction 45-65 on P,
(A) and E, (*)production from cultured RGC as determined by RIA (inset). A 10-p1
aliquot was tested from fraction 1-90. In this experiment the control group is the granulosa
cells that were treated culture medium with no addition of HPLC fractions. Data points
presented in the steroidogenic activity curves are picogram quantity of the steroids per
microgram of cell protein in duplicate culture wells. The arrows indicate the peaks that
contain TCM-22 activity.
ence on the ovaries. When we started our work, the available information sug-
gested that this influence may be through the autoimmune and endocrine
mechanisms described in the Introduction. However, our data and those of others
have highlighted the possibility of a direct influence of the thymus on the ovary
through thymic factors. Aguilera and Romano [22] reported the presence of a
28-kDa factor in thymic reticuloepithelial cell-or thymus-conditioned media that
inhibited hCG-induced P, and E2 production from rat ovarian cell but had no
effect on P4 and E2 production in the absence of hCG. Interestingly, the same
laboratory has recently reported that rat thymus contains a heparin-binding factor
that modulates steroidogenesis in the rat testis 1321. Ledwitz-Rigby and Scheid
[23,33] reported that thymulin, a nanopeptide produced by the thymus gland,
182 Uzumcu et al.
could enhance gonadotropin-induced P, secretion from cultured porcine granu-

losa cells. It has been shown that thymulin enhanced both FSH- and LH-stimu-
lated P, secretion over 2-4 days of incubation, even though basal secretion was
not altered and suggested that thymulin could play an important role in granulosa
cell maturation. A variety of additional data have also accumulated showing a
direct influence of previously characterized thymic factors on ovarian function
in mammals. For example: (1) A higher number of germ cells were observed in
fetal ovaries cultured in medium supplemented with thymulin or cocultured with
fragments of fetal thymus as compared to the control ovaries [34]. A similar
effect of thymus fragments and thymulin has recently been observed on fetal rat
testes [35,36]. (2) TF-5 was shown to be stimulatory on basal but inhibitory on
FSH-induced P, and E2 secretion in RGC [37]. However, it is noteworthy that
we failed to show any effect of 5-250 glml thymulin and or thymosin a , on
basal steroidogenesis by RGC in vitro [27]. (3) It was demonstrated that the
thymus gland modulates ovarian responsiveness to gonadotropin in vivo [38],
and thymulin was shown to be the mediator for the modulatory action [39].
(4) Prothymosin-a mRNA, thymosin-P, and P,, mRNA, and protein are detected
in rat ovaries [40-421. In addition, gonadotropins and prostaglandin FZamodu-
lated the differential expression of thymosin-P4 and Dl0 in rat ovary [41,42].
Collectively, these reports support the hypothesis that known "thymic" factors,
produced either in the thymus gland itself or locally in the ovary, are capable of
affecting ovarian function in mammals. Based on our studies and studies of others
mentioned above, we now suggest that thymus directly influences ovarian func-
tion through a variety of thymus-derived signaling molecules. Further studies to
establish the identity of factors such as TCM-1 and TCM-22 will likely give
further impetus to this newly emerging field of reproductive biology.
ACKNOWLEDGMENTS
The studies described from our laboratory in this chapter were supported in part
by NIH grant DK 45916, and NATO Scientific Training ProgramIScientific and
Technical Research Council of Turkey
REFERENCES
1. Chapman JC, Michael SD. Hormonal and immunologic interactions between thymus
and ovary. In: Grossman CJ, ed. Endocrinology and Metabolism. Bilateral Commu-
nication Between the Endocrine and Immune Systems. New York: Springer-Verlag,
1994: 12-35.
2. Nishizuka Y, Sakakura T. Thymus and reproduction: sex-linked dysgenesia of the
gonad after neonatal thymectomy in mice. Science 1969;166:753-755.
r
Besedovsky HO, Sorkin E. Thymus involvement in female sexual maturation. Na-

ture 1974;249:357-359.
Michael SD, Taguchi 0 , Nishizuka Y. Effect of neonatal thymectomy on ovarian
development and plasma LH, FSH, GH and PRL in the mouse. Biol Reprod 1980;
22:343-350.
Lintern-Moore S. Effect of athymia on the initiation of follicular growth in the rat
ovary. Biol Reprod 1977; 17: 155- 161.
Hattori M, Brandon MR. Thymus and the endocrine system: ovarian dysgenesis in
neonatally thymectomized rats. J Endocrinol 1979;83: 10 1- 111.
Healy DL, Bacher J, Hodgen GD. Thymic regulation of primate fetal ovarian-adrenal
differentiation. Biol Reprod l985;32: 1127- 1133.
Nishizuka Y, Sakakura T, Tsujimura T, Matsumoto K. Steroid biosynthesis in vitro
by dysgenetic ovaries induced by neonatal thymectomy in mice. Endocrinology
1973;93:786-792.
Michael SD, Taguchi 0 , Nishizuka Y, McClure JE, Goldstein AL, Barkley MS.
The effect of neonatal thymectomy on early follicular loss and circulating levels of
corticosterone, progesterone, estradiol, and thymosin a,. In: Schwartz NB, Hun-
zicker-Dunn M, eds. Dynamics of Ovarian Function. New York: Raven Press, 1981:
279-284.
Hoek A, Schoemaker J, Drexhage HA. Premature ovarian failure and ovarian auto-
immunity. Endocr Rev 1997;18:107- 134.
Taguchi 0 , Nishizuka Y, Sakakura T, Kojima A. Autoimmune oophoritis in thymec-
tomized mice: detection of circulating antibodies against oocytes. Clin Exp Immunol
1980;40:450-453.
Miyake T, Taguchi 0 , Ikeda H, Sato Y, Takeuchi S, Nishizuka Y. Acute oocyte
loss in experimental autoimmune oophoritis as a possible model of premature ovar-
ian failure. Am J Obstet Gynecol 1988;158:186-192.
Tung KS, Smith S, Teuscher C, Cook C, Anderson RE. Murine autoimmune oophor-
itis, epididymo-orchitis, and gastritis induced by day 3 thymectomy. Am J Path01
1987;l26:293-302.
Sakaguchi S, Takahashi T, Nishizuka Y. Study on the cellular events in post-
thymectomy autoimmune oophoritis in mice. I. Requirement of Lyt-1 effector cells
for oocytes damage after adoptive transfer. J Exp Med 1982;156:1565- 1576.
Sakaguchi S, Takahashi T, Nishizuka Y. Study on the cellular events in post-
thymectomy autoimmune oophoritis in mice. 11. Requirement of Lyt-I in normal
female mice for the prevention of oophoritis. J Exp Med 1982;156: 1577- 1586.
Kosiewicz MM, Michael SD. Neonatal thymectomy affects follicle populations be-
fore the onset of autoimmune oophoritis in B6A mice. J Reprod lmmunol 1990;88:
427-440.
Rebar RW, Morandini IC, Erickson GF, Petze JE. Hormonal basis of reproductive
defects in athymic mice: diminished gonadotropin concentrations in prepubertal fe-
males. Endocrinology 1981;108:120- 126.
Rebar RW, Morandini IC, Silva de Sa MF, Erickson GF, Petze JE. The importance
of the thymus gland for normal reproductive function in mice. In: Schwartz NB,
Hunzicker-Dunn M, eds. Dynamics of Ovarian Function. New York: Raven Press,
1981:285-290.
184 Uzurncu et al.
Rebar RW, Morandini IC, Benirschke K, Petze JE. Reduced gonadotropin in athymic
mice: prevention by thymic transplantation. Endocrinology 1980;107:2130-2132.
Rebar RW, Miyake A, Low TLK, Goldstein AL. Thymosin stimulates secretion of
luteinizing hormone-releasing factor. Science 198 1 ;2 14:669-67 1.
Terranova PF, Rice VM. Review: cytokine involvement in ovarian processes. Am
J Reprod Immunol 1997;37:50-63.
Aguilera G, Romano MC. Influence of the thymus on steroidogenesis by rat ovarian
cells in vitro. J Endocrinol 1989; 123:367-373.
Ledwitz-Rigby F, Schneid PC. Serum thymic factor enhances porcine granulosa
cell responsiveness to gonadotropins in vitro. The Eighth Ovarian Workshop on
Regulatory Processes and Gene Expression in the Ovary. Maryville, Tennessee,
USA, July 12-14, 1990.
Lewis VM, Twomey JJ, Bealmear P, Goldstein G, Good RA. Age, thymic involu-
tion, and circulating thymic hormone activity. J Clin Endocrinol Metab 1978;47:
145- 150.
Dardenne M, Savino W, Gagnerault M-C, Itoh T, Bach J-F. Neuroendocrine control
of thymic hormonal production. I. Prolactin stimulates in vivo and in vitro the pro-
duction of thymulin by human and murine thymic epithelial cells. Endocrinology
1989;125:3-12.
Uzumcu M. Lin YC. Factor(s) from thymic cell culture medium conditioned medium
(TCM) stimulate(s) progesterone (P) secretion in cultured rat granulosa cells (abstr).
Biol Reprod 1990;42 (Suppl 1):163.
Uzumcu M, Akira S, Lin YC. Stimulatory effect of thymic factor(s) on steroidogene-
sis in cultured rat granulosa cells. Life Sci 1992;Sl :l2l7- 12.28.
Uzumcu M, Lin YC. Characterization of the stimulatory actions of thymic factor(s)
on basal and gonadotropin-induced steroidogenesis in cultured rat granulosa cells.
Mol Cell Endocrinol 1994;105:209-216.
Uzumcu M, Brigstock DR, Lin YC. Partial purification and characterization of two
non-FSH steroid-modulating factors in rat thymic epithelial cell-conditioned me-
dium (TCM). Dom Anim Endocrinol 1998;15:155- 168.
Soto EA, Kliman HJ, Strauss JF 111, Paavola LG. Gonadotropins and cyclic adeno-
sine 3',Sf-monophosphate (CAMP)alter the morphology of cultures human granulosa
cells. Biol Reprod 1986:34:559-569.
Bray W, Brigstock DR. Raw integrator data storage and retrieval using data acquisi-
tion software. Amer Lab 1994;26:38.
Porras MG, Reyes J, Romano MC. The rat thymus contains a heparin-binding factor
that modulates steridogenesis in the testis. Acta Physiol Pharmacol Ther Latin Am
1996;46:286-293.
Ledwitz-Rigby F, Schneid PC. Serum thymic factor enhances porcine granulosa cell
responsiveness to gonadotropins in vitro. In: Gibori G, ed. Signaling Mechanisms
and Gene Expression in the Ovary. New York: Springer-Verlag, 1991:473-478.
Prepin J. Thymus and thymulin increase the proliferation of oogonia in fetal rat
ovary in vitro. C R Acad Sci Paris Ser. 111 199 1 :3 13:407-4 11.
Prepin J. Fetal thymus and thymulin stimulate in vitro gonocytes in fetal testis in
rats. C R Acad Sci Paris Ser. I11 1993;3l6:4S 1-454.
Prepin J, Le Vigouroux P. Inhibition by TGF-beta I of the in vitro thymulin-stimu-
r
lated proliferation of gonocytes from fetal rat testes. Reprod Fertil Dev 1997;37:
203-206.
Gorospe WC, Fong Y-Y, Spangelo BL. Thymosin fraction-5 modulates steroidogen-
esis and interleukin-6 release by rat granulosa cells in vitro. Endocr J 1993;1:35-
39.
Rosas P, Hinjosa L, Garcia L. Evidence for the participation of the thymus on the
regulation of ovarian development in the prepubertal mouse (abstr). Biol Reprod
1995;52 (Suppl 1):130.
Hinojosa L, Rosa P. Effects of the serum thymic factor (FTS) on the gonaodropin-
induced ovulation in normal and hypothymic mice (abstr). Biol Reprod 1996;54
(Suppl 1):85.
Oikawa M, Dargan C, Ny T, Hsueh AJW. Expression of gonadotropin-releasing
hormone and prothymosin-a messenger ribonucleic acid in the ovary. Endocrinol-
ogy 1990;l27:2350-2356.
Hall AK, Aten R1, Behrman HR. Differential modulation of thymosin genes in the
immature rat ovary by gonadotropins. Mol Cell Endocrinol 1991:79:37-43.
Hall AK, Aten Rl,Behrman IIR. Thymosin gene expression is modulated by preg-
nant mare's serum gonadotropin, human chorionic gonadotropin, and prostaglandin
F2, in the immature rat ovary. Endocrinology 1991;128:951-957.
Determination of Affinity Constants
of Lectins for Sugars by Affinity
Capillary Electrophoresis
Kiyohito Shimura and Ken-ichi Kasai
Teikyo University, Sagamiko, Kanagawa, Japan
I. INTRODUCTION
A variety of life processes are dependent on the specific biological recognition

of macromolecules. Modem separation techniques, such as chromatography El],
electrophoresis [2-41, and centrifugation [5], are widely used for the analysis of
specific affinities of biological molecules including proteins, nucleic acids, and
sugars. These applications are based on the change in the physicochemical prop-
erties of interacting molecules as the result of complex formation. The complex
may have properties that are different from those of their component molecules
with respect to molecular weight, electrical charge, or hydrodynamic properties,
thus permitting its separation from the component molecules.
The recently developed microscale separation technique, capillary electro-
phoresis (CE), has many favorable features that are useful in the analysis of
molecular recognition as well as the separation of biological macromolecules [6].
The scale of the separation space in CE is almost one-thousandth of the other
separation techniques and the amount of sample required for analysis is quite
small: at the nanogram level for proteins with ultraviolet (UV) absorption detec-
tion. The temperature of the separation medium can be precisely controlled and
molecular interaction can be observed without any insoluble support or gel, which
might affect the interaction involved. The application of CE in the analysis of
biospecific interactions has been referred to as affinity CE [6-91.
For the estimation of affinity constants, two formats of affinity CE are appli-
188 Shirnura and Kasai
cable. Equilibrium mixture analysis, which is suitable for a system that shows
slow dissociation kinetics, measures the concentration of the free species in equi-
librium mixtures at different concentrations by zone electrophoresis. This allows
a binding isotherm to be determined and, thus, an estimation of the affinity con-
stant. Mobility change analysis, which is applicable to a system that shows rapid
dissociation kinetics, measures the mobility change of a zone of one component
in a solution of the other under a rapid binding equilibrium. The amount of the
mobility change can be correlated to the mole fraction of the bound species and
also leads to determination of the affinity constant. This latter format of affinity
CE is very useful as a microscale technique for the determination of affinity
constants for systems with relatively weak affinity, e.g., for systems with a disso-
ciation constant larger than the micromolar level.
The principle of the determination of affinity constants by mobility change
analysis has been established through the application of affinity electrophoresis
in gel slabs or rods [2.3]. This type of analysis has several advantages. First, it
is not necessary to determine precisely the concentration of an interacting compo-
nent applied as a zone. Considering the difficulty in the determination of precise
concentration of small quantities of a protein sample, this feature is particularly
valuable. Second, the sample that is applied as a zone need not necessarily be
pure, so long as the peak of the component can be traced. This represents a
big advantage, especially in cases where the protein is unstable and is prone to
denaturation. The simultaneous determination of the affinity constants of a group
of molecules sharing the same binding specificities is also possible, provided all
components can be separated. Third, the amount of an interacting component
applied as a zone can be very small.
In this chapter. we will describe the principle and applications of mobility
change analysis with two examples of the analysis of lectin-sugar interactions
by affinophoresis and affinity probe capillary electrophoresis (APCE). Before we
move to the body of this chapter, it would be useful to briefly review the basics
of CE.
II. CAPILLARY ELECTROPHORESIS
CE uses a narrow electrophoresis channel (usually from 10 to 100 pm in inner

diameter) that enables the swift dissipation of heat and the application of a high
field strength. This shortens the time required for separation, reduces the longitu-
dinal diffusion of analytes, and raises the efficiency of separation. The narrow
channel also minimizes convection, allowing electrophoresis in free solution. The
quantity of a sample required for analysis becomes significantly smaller as the
size of the separation space is reduced. Analytes are detected and quantified by
Determination of Affinity Constants by CE 189
means of an on-line UV or fluorescence detector. On the other hand, the ratio

of the inner surface area to the volume of the inner space where separation takes
place is very large and this emphasizes the need for strict control of the adsorption
of analytes on the wall.
The basic setup of free solution CE is as follows. Fused silica capillaries
coated with polyimide on their outer surface are commonly employed. The coat-
ing provides mechanical strength to the capillary that otherwise easily breaks. A
small segment of the polyimide coating is removed and the migrating analytes
in the capillary are optically detected through the silica wall. The capillary is
cooled by a temperature-controlled coolant. The capillary is tilled with an electro-
phoresis buffer and its ends are immersed in the buffer in electrode vessels that
are connected to an electrical power supply that can generate an electrical poten-
tial up to 30 kV.
The surface of fused silica is negatively charged at neutral pH range due
to the ionization of silanol groups. This negatively charged surface is compatible
with the separation of neutral and anionic molecules. Separation of cationic mole-
cules, however, might be compromised by ionic adsorption to the surface. The
electrophoretic migration of the counterions for these immobile charges on the
inner surface produces water flow toward the negative electrode. This flow, called
electroosmotic flow (EOF), is characterized by a constant velocity irrespective
of the lateral position inside the capillary in contrast to a pressure-driven flow,
in which the velocity is highest at the center. EOF can be very large in CE and
often larger than the electrophoretic mobility of the analytes. Even small anions
can be transferred to the negative end by the overwhelming electroosmosis over
their electrophoretic migration. Neutral molecules are transferred to the detection
point solely by the influence of EOF and are detected earlier than anionic species
migrating toward the positive end (Fig. 1).
An important consequence of the presence of EOF is that the detection
time (t) of a particular analyte is a function of its electrophoretic mobility (p,,),
and the electroosmotic mobility (p,,) of the capillary as well.
- -
EOF EOF EP
Figure 1 Zonal electrophoresis in a capillary with a negatively charged inner surface.

EOF, electroosmotic flow; EP, electrophoretic migration. Negatively charged ions (S) are
detected after neutral molecules (N) in the presence of a high electroosmotic flow.
D, detector; 0, electrophoresis origin.
190 Shimura and Kasai
where L (cm) is the distance between the injection end and the detection point,
and E (V cm-') represents the field strength. The polarity of the mobility is de-
fined as negative for movement toward the positive electrode. Thus, in the case
of the negatively charged ion (S) in Fig. 1, p,, is negative and pe, is positive.
Note that pep(cm2 V-' s-' ), a value of fundamental importance in the mobility
change analysis, is inversely related to the experimentally measurable value, t.
Coelectrophoresis of a neutral marker molecule allows the determination of p,
in the capillary and, therefore, pepof charged substances using Eq. (1). This point
will be discussed further in Section 111.
The EOF of a bare silica capillary is not very stable and is greatly affected
by the adsorption of ions, especially proteins. Experimentally, a sudden decrease
in EOF was also experienced for unknown reasons. We used fused silica capillar-
ies coated with succinylpolylysine in experiments described in this chapter. Such
capillaries show very stable EOF.
Ill. EQUATIONS FOR MOBII-ITY CHANGE ANALYSIS
When a charged particle is placed in an electric field, movement of the particle

is induced and reaches a steady state where the electric force is just balanced by
the frictional drag, i.e., qE = kv, where q is the number of charges on the particle
migrating at a steady-state velocity of v under a field of E with a translational
frictional constant k. The velocity under unit field is designated as the electropho-
retic mobility, p, with the relation p = q/k.
When two molecules, A and B, having different electrophoretic mobility
(pA and pB, respectively) form a complex, AB, it could have a mobility (pAB)
different from those of either of the component molecules.
The mobility of the complex is usually intermediate between those of its compo-
nents. In the case of protein-ligand interactions, the charge on the ligand is a
major source of the mobility change of the protein.
When electrophoresis of A is camed out in a solution of B, the microscopic
electrophoretic mobility of A should discontinuously change between pAand pAB
at the moments of dissociation and association with B. As a result, the observed
macroscopic mobility of A, p, should be an average of the two mobilities
weighted for the time fractions in which it migrates as the free molecule and as
the complex with B . Using aAB

as the time fraction for the complex AB, p is
given by Eq. ( 3 ) :
Equation ( 3 ) can be rewritten as A p = &,,,,aAB, where A p = p - PA is the

macroscopic mobility change of A in the presence of B , and Ap,, = pAB - pA
is the maximum mobility change of A in the presence of B at infinite concentra-
tion or, in other words, the mobility difference between free A and the complex
AB. Since aAB is the mole fraction as aAB = [ A B ] / ( [ A ]+ LAB]), it is related
to the equilibrium dissociation constant, K d = ( [ A : I [ B ] ) / [ A B ]by
, the relation,
aAB= [ B ] / ( K d+ [ B ] ) ,and the following equation can be derived [ 2 , 3 ] .
This relation is identical to Langmuir's adsorption isotherm. The two parameters

of affinity electrophoresis, K d and Ap,,,, can be determined by using a nonlinear
regression analysis by fitting them to measured A p values at different concentra-
tions of B. Its analogy to the Henry-Michaelis-Menten equation for enzyme kinet-
ics allows the use of several forms of linearized plots for the determination of
the two parameters. The plot of Ap against A p / [ B ] according to Eq. ( 5 ) . which
is identical to the Woolf-Hofstee plot in enzyme kinetics, should generally be
appropriate. The slope of the line gives the negative value of the dissociation
constant and the intercept with the Ap axis gives the Ap,,, [lo].
Inhibition experiments in the affinity electrophoresis system with neutral ligands

allow the determination of the affinity constants for the neutral ligands [ 2 ] .Such
competition experiments will be the main focus in the examples described below.
In CE, the directly measurable value related to the electrophoretic mobility
is the detection time of peaks. Since the electrophoretic mobility is inversely
related to detection time [Eq. ( I ) ] , the mobility change ( A p ) must be calculated
by use of the following relation.
where t and to are the detection time of component A in the presence and the
absence of B , respectively, and t, and t,' are the detection time of an electrophore-
sis marker in the presence and the absence of B , respectively. The prime symbol
is attached to show the possible variation of electroosmosis between two runs.
The marker should not interact with B and need not necessarily be electrically
192 Shirnura and Kasai
neutral. Fluctuations in electroosmosis are canceled out by this equation and the
net mobility change of A caused by the interaction with B can be calculated.
IV. ANALYSIS OF PEA LECTIN-SUGAR INTERACTIONS

BY AFFINOPHORESIS
When A is a protein and B is a charged ligand, the protein-ligand interaction

can be analyzed by mobility change analysis. The ligand, however, might not be
8 9 10 12
Time (min)
Figure 2 Capillary affinophoresis of pea lectin using a monoliganded affinophore bear-

ing mannoside. (A) The structure of the monoliganded affinophore bearing a-D-mannoside
as an affinity ligand on a matrix of succinylated glutathione. (B) Affinophoresis of pea
lectin with different concentrations of mannoside affinophore. A sample solution (0.7 nl)
containing pea lectin (**,0.2 mglml) and cytidine (*,0.2 mM, as an electrophoresis
marker) was injected by means of pressure at the positive end of a capillary (25 pm i.d.,
375 pm o.d., 57 cm long, internally coated with succinylpolylysine) filled with a solution
of the affinophore at the concentrations shown on the right. Electrophoresis was camed
out at a field strength of 350 V cm-' (7 pA) at 2 5 T , with detection in terms of A,,, at
50 cm from the injection end. Tris-acetate buffer (pH 7.9) was used throughout the system.
(From Ref. 10.)
able to produce a sufficient mobility change of the protein for analysis. We have
introduced the use of charged soluble polymers, such as succinylpolylysine, as
mobile matrices for affinity electrophoresis. We call the charged matrix bearing
ligands the "affinophore" and electrophoresis in its presence "affinophoresis"
[3,4]. The distinctive feature of affinophores is that the portion responsible for
specific binding and that for migration in an electric field are structurally sepa-
rated, i.e., an affinity ligand and a charged affinophore matrix. Therefore, once
a suitable ionic molecule is obtained as an affinophore matrix, the method is
applicable to many binding molecules by changing the affinity ligand.
The principle of affinophoresis was successfully applied to the mobility
change analysis of lectin-sugar interactions. For the determination of affinity
constants of pea lectin, a divalent sugar-binding protein, the use of a monoli-
ganded affinophore is straightforward. The polyliganded affinophores, used in
the previous applications, should interact with such a divalent protein in a more
complicated manner [ I l l . N-Succinylated glutathione was used as an anionic
matrix for a monoliganded affinophore [lo]. The affinity ligand p-aminophenyl
a-D-mannoside was iodoacetylated at its amino group and coupled to the thiol
of the matrix (Fig. 2A). The intermediates and the affinophore were easily purified
by using reversed-phase HPLC. The affinophore caused a mobility change of pea
lectin in a concentration-dependent manner (Fig. 2B). Pea lectin barely migrated
under the present electrophoresis conditions. The interaction with the negatively
Figure 3 A linear plot for the affinophoresis of pea lectin. The plot was made using
the data in Fig. 2B according to Eq. [5]. The slope is the negative value of K, (0.199
mM) and the intercept with the ordinate gives Ap,,,, (-5.35 X cm2 V - ' s-' ). (From
Ref. 10.)
charged affinophore increased its mobility toward the positive end, thus lengthen-
ing its detection time. The result of the affinophoresis was analyzed according
to Eq. (5) based on several assumptions, i.e., equivalency and independency of
the two binding sites of the dimeric lectin, and the same mobility change for the
successive binding of the two affinophores (Fig. 3). The determination of the two
parameters, Kd = 0.199 mM and Ap,,, = -5.35 x cm2 V-I s-' , for the
interaction between the lectin and the affinophore served the basis for the determi-
nation of the affinity constants of the lectin to neutral sugars by competition
experiments, using affinophoresis. It should be noted that Kdrepresents the reac-
tion of each binding site, i.e., a microscopic equilibrium constant.
A neutral sugar to be investigated was added only in the positive electrode
buffer and the affinophore only in the capillary. Since the neutral sugar does not
migrate electrophoretically, it is transferred to the negative end only by EOF
faster than the lectin under the affinophoresis conditions, where the lectin mi-
grates electrophoretically toward the positive end. With the affinophore (B) at a
Time (rnin)
Figure 4 Competition experiment of the affinophoresis by mannose. Affinophoresis of

pea lectin was carried out with the affinophore at a concentration of 0.5 mM in the presence
of mannose at various concentrations, as shown on the right. Mannose was added only
in the positive electrode buffer (100 pl). Other conditions were the same as in the experi-
ments in Fig. 2. **, *,
pea lectin; cytidine. (From Ref. 10.)
constant concentration, the macroscopic mobility change (Ap,) of the lectin be-
came smaller as the concentration of the neutral sugar (I) increased (Fig. 4). The
competition analysis is basically the same as that in enzyme kinetics. It can be
considered that the K, value in Eq. (4) increased by a factor of 1 +
[I]/Ki than
that without I in an identical relation to Eq. (4), i.e.,
[Bl
Api = W m a x
Kd app + [Bl
The K, ,,,,
an apparent dissociation constant, is in a relation with the K, as
Kd = Kd(l + []]/K,), where K, is the dissociation constant of pea lectin for
I. Since Apmaxhas already been previously obtained, K,,,, can be determined by
the measurement of Ap, according to Eq. (7). The plot of Kdapp/Kd versus [I]
gave lines with a slope of 1/K, (Fig. 5). The K , values obtained by this method
Figure 5 Determination of the dissociation constants of pea lectin for neutral sugars
(I) through competition experiments in the affinophoresis. The affinophoresis of the lectin
was carried out as described for Fig. 2 except for the addition of various concentrations
of neutral sugars to the positive electrode buffer (100 yl). Apparent K, values (K,,,,) were
determined from the observed mobility change (AyJ, the concentration of the affinophore
used ([B]), and the Ap,,,, value using Eq. [7]. The results are plotted according to Kd,I
Kd = [I]/Ki + 1.@, p-aminophenyl a-D-mannoside (0.25 mM); 0, methyl a-D-mannoside
(0.65 mM); A, D-mannose (1.6 mM); A,maltose (2.1 mM); W, methyl a-D-glucoside
(2.4 mM); 0, D-glucose (3.6 mM), where the Ki values determined are given in parenthe-
ses. (From Ref. 10.)
agreed well with those obtained by calorimetry. For example, Ki for methyl a -
mann no side was found to be 0.65 mM and was reported to be 0.61 mM by
calorimetric measurement. For D-mannose, it was found to be 1.6 rnM and was
reported to be 1.3 mM, calorimetrically.
V. ANALYSIS OF CONCANAVALIN A-SUGAR

INTERACTIONS BY AFFINITY PROBE CAPILLARY
ELECTROPHORESIS
In affinophoresis, we directed our attention to the mobility change of a protein.

Instead, that of a ligand produced by the interaction with a protein can also be
analyzed. A larger mobility change can be expected for a ligand than a pro-
tein during their interaction, and this approach should thus be useful when the
mobility change of a protein can be barely observed or is negligible. Using
a fluorescence-labeled affinity ligand (Fig. 6A), which is called an affinity
probe. the affinity constants for concanavalin A (Con A) for monosaccharides
were determined by competition experiments [12]. For the preparation of the
affinity probe, an identical approach was taken to that for the monoliganded
affinophore. The amino group of glutathione was labeled with a fluorescent dye
instead of succinylation in the case of the affinophore. Glutathione (oxidized
form) was labeled with Lissamine rhodamine B sulfonyl chloride at its amino
group and the disulfide was cleaved by dithiothreitol after purification. The affin-
ity ligand, p-aminophenyl a-D-mannoside, was iodoacetylated and coupled to the
free thiol of the dye-labeled glutathione. The affinity probe having two net nega-
tive charges was subjected to electrophoresis along with a hydrolysis product of
the reactive dye, a sulfonic acid derivative having one net negative charge, which
served as an electrophoresis marker.
The mobility of the affinity probe was originally larger than the marker
and was detected later, but its electrophoretic mobility was reduced by interaction
with Con A and the order of the two peaks was inverted (Fig. 6B). The analysis
using Eq. (5) allowed the determination of the two parameters for the affinity
probe-Con A interaction as Kd = 15.5 pM and Ap,,, = 1.20 X IW4 cm2 V-'
s-I. Once these two parameters are determined, the interaction between Con A
and neutral sugars (I) can be analyzed by competitive experiments of APCE.
When a neutral sugar having affinity for Con A is added to the Con A solution,
the concentration of the free binding site of Con A ([B] in Eq. (4)) will decrease
and the suppression of the electrophoresis of the affinity probe is released. Since
the values of Kd and Ap,,,, have already been obtained, [Con A] in the mixture
can be calculated by measuring Ap of the affinity probe. The determination of
[Con A] in the mixture allows the calculation of [Con A-I] and [I], since the
total concentration of Con A and I is known. The dissociation constant for the
Determination of Affinity Constants by CE
A B
Time (min)
Figure 6 Determination of the dissociation constants of Con A for neutral sugars by

affinity probe capillary electrophoresis. (A) The structure of the affinity probe bearing two
net negative charges. (B) Affinity probe capillary electrophoresis in the presence of Con
A and methyl a-D-mannoside. A mixture (2 nl) of the affinity probe and the electrophoresis
marker (lo-' M each) was injected at the positive end of the capillary (succinylpolylysine-
coated fused silica, 50 ym id., 375 ym o.d., 30 cm long) filled with 0.1 M Tris-acetic
acid buffer (pH 7.9). Con A and neutral sugars were dissolved in the buffer, introduced
into an open-ended polypropylene tube (0.5 mm id., 4.7 cm long, 10 p1 volume) and
placed between the positive end of the capillary and the positive electrode buffer as a
bridge. Electrophoresis was carried out at a field strength of 150 V cm-' (10 yA) at 2S°C,
and the fluorescence (590 nm) was detected 10 cm from the injection end by excitation
with a He-Ne laser (543.5 nm, 1 mW). Concentration of Con A: a, 0 yM; b-d, 21 FM.
Concentration of methyl mannoside: a and b, 0 yM; c. 120 yM; and d, 300 pM. AP,
affinity probe; M, electrophoresis marker. (From Ref. 12.)
Con A-neutral sugar complex, K, = ([Con AI[I])/[Con A-I], can be calculated

(Fig. 7). Note the difference in the manner of competition between the two experi-
ments. In the affinophoresis, a neutral sugar binds to molecule A applied as a
zone; however, in the case of APCE, it binds to molecule B added in the electro-
phoresis buffer.
Shimura and Kasai
0 0.1 0.2 0.3

[Me-Man](mM)
Figure 7 Plots for the determination of the dissociation constants of Con A for neutral
sugars (I). Prior to the analysis of neutral sugar-Con A interaction, the affinity probe-Con
A interaction was characterized by analysis according to Eq. [5] and K, and AF,, were
determined to be 15.5 FM and 1.20 X cm2 V-' s-'. When a neutral sugar was added
to the Con A solution, the mobility change of the affinity probe was reduced, as the result
of the reduction in the number of the free binding sites of Con A available for the interac-
tion with the affinity probe, as shown for methyl mannoside in Fig. 6B. From the reduced
mobility change, the equilibrium concentration of free binding site of Con A ([Con A],
corresponding to [B] in Eq. [4]) in a mixture with the neutral sugar (I) was calculated
according to Eq. [4] using the K, and AF,,, values determined previously. The determina-
tion of [Con A] allowed the calculation of [Con A-I] and [I], since the total concentrations
of Con A and I are known. The plot of [Con A-]]/[Con A] vs. [I] gave lines with slopes
of 1 IK,, where K, = ([Con A] X [I])/[Con A-I]. The plots and the determined values of
K,'s are: (a) methyl a- mann no side (0.104 mM); (b) mann nose (0.89 mM); (c) D-glucose
(4.5 mM); and (d) D-galactose (1.1 M). (From Ref. 12.)
Determination of Affinity Constants by CE
VI. EXPERIMENTAL SElTINGS
A typical experimental setup for affinophoresis of a protein with a negatively

charged affinophore (B) having higher mobility than protein (A) is as follows
[lo]: Fused silica capillaries coated with succinylpolylysine [12] are filled with
a buffer solution containing the affinophore. The same buffer solution is used in
the electrode vessels, but without the ligand. A solution of a binding protein is
pressure-injected as a small segment at the positive end of the capillary. When
an electrical field is applied, the protein is transferred to the negative end by
the overwhelming electroosmotic flow of the capillary and the relatively slow
electrophoretic migration of the protein. Soon after the initiation of electrophore-
sis, the protein is surrounded by the affinophore due to its lower electrophoretic
mobility toward the positive electrode, as compared to the affinophore. After
a transitional period, corresponding to the time required to establish a binding
equilibrium in the protein zone, the protein begins to migrate at a higher electro-
phoretic mobility toward the positive electrode than that observed in the absence
of B and its detection time increases. Note that the concentration of the free
affinophore in the protein zone has become identical to that in the solution origi-
nally used to fill the capillary at this time, since the protein zone is continuously
exposed to the unused portion of the affinophore solution due to the mobility
differences between the protein zone and the affinophore. The mobility change
is calculated from the detection time for the protein and the marker in the presence
and absence of the affinophore by Eq. (6), and the results of a set of experiments
over a range of concentration of the affinophore can be plotted according to Eq.
(5). The plot gives the K, value and the Ay,,,.
The range of the concentration of the affinophore should be chosen so that
it overlaps with the value of the dissociation constant to be measured and should
be sufficiently wide for precise determination of the constant. A protein concen-
tration of 0.1-0.5 yglyl would be sufficient to measure the detection time of the
protein by monitoring the absorption at 214 nm. The volume of the protein sample
applied for each run is less than several nanoliters and thus the amount of the
protein actually consumed for each run is quite small. The amount of protein
injected in the capillary has no effect on the result unless it exceeds that of the
affinophore. Since the protein sample does not contain the affinophore, a short
time is required before the injected plug of a protein solution is equilibrated
with the affinophore and reaches a steady-state velocity as described above. This
transitional period will be longer when the affinophore concentration is lower
relative to that of the protein solution. This effect is observed as an increasing
deviation of Ay versus Ay/[B] plot according to Eq. (5) in the lower concentration
range of the affinophore and this can be improved by reducing an amount of the
protein sample. Regarding the reproducibility of the determination of the K, val-
ues by this technique, the relative standard deviation was about 10% (n = 5)
with a manually operated instrument 1121 and 6% (n= 5) with a fully automated
instrument ( 101.
VII. PROSPECTS
Analysis of the interactions of biological macromolecules by electrophoresis has

a long history and such applications have been collectively referred to as affinity
electrophoresis [2]. Capillary electrophoresis, due to its versatility, shows great
promise for a variety of applications of this type. The basic experimental equip-
ment and theory for this technique has been elaborated, thus promising consider-
able progress for this type of application. In addition, the miniaturized separation
space in capillary closely matches the size of cells, i.e., the elementary units of
life. The microchannel, constructed in a planar chip [13], will further reinforce
the trend of miniaturization of electrophoresis initiated by the use of capillary.
The efficiency of the analysis of molecular recognition by affinity electrophoresis
suggests that this technique will be of increasing importance in the future.
REFERENCES
Kasai K. Trypsin and affinity chromatography. J Chromatogr 1992;597:3-18.

Takeo K. Advances in affinity electrophoresis. J Chromatogr 1995;698:89-105.
Shimura K. Progress in aftinophoresis. J Chromatogr l99O;j 10% 1-270.
Shimura K, Kasai K. Affinophoresis: selective electrophoretic separation of proteins
by using a specific carrier. In: Karger BL, Hancock WS eds. High Resolution Separa-
tion and Analysis of Biological Macromolecules. Part R , Methods Enzymol Vol.
27 1. New York: Academic Press, l996:203-2 18.
Shire SJ. Analytical ultracentrifugation and its use in biotechnology. In: Schuster
TM, Laue TM eds. Modern Analytical Ultracentrifugation. Boston: Birkhauser,
1994:261-297.
Shimura K, Kasai K. Affinity capillary electrophoresis: a sensitive tool for the study
of molecular interactions and its use in microscale analyses. Anal Biochem 1997;
251:l-16.
Chu Y-H, Avila ZL, Gao J, Whitesides GM. Affinity capillary electrophoresis. Acc
Chem Res 1995;28:461-468.
Pritchett T, Evangelista RA, Chen F-TA. Capillary electrophoresis-based immunoas-
says. Bioltechnology 1995; 13: 1449-1450.
Heegaard NHH. Characterization of biomolecules by electrophoretic analysis of re-
versible interactions. Appl Theor Electrophor 1994;4:43-63.
Shimura K, Kasai K. Determination of the affinity constants of pea lectin for neutral
sugars by capillary affinophoresis with a monoligand affinophore. J Biochem 1996;
120:1146-1152.
11. Shirnura K, Kasai K. Capillary affinophoresis of pea lectin with polyliganded af-
finophores: A model study of divalent-polyvalent interactions. Electrophoresis 1998;
19:397-402.
12. Shimura K. Kasai K. Determination of the affinity constants of concanavalin A for
monosaccharides by fluorescence affinity probe capillary electrophoresis. Anal Bio-
chern 1995;227: 186- 194.
13. Harrison DJ, Fluri K, Seiler K, Fan Z, Effenhauser CS, Manz A. Micromachining
a miniaturized capillary electrophoresis-based chemical analysis system on a chip.
Science 1993;261:895-897.
Displacement Chromatography
of Biomolecules
Ruth Freitag
ETH Lausanne, Lausanne, Switzerland
I. INTRODUCTION
Chromatography is based on subtle differences in substance distribution between

a fluid, usually mobile, and a solid, usually stationary, phase. Few laboratories
in the life sciences do not depend on this extremely versatile, high-resolution
separation technique, albeit mostly for analytical purposes. However, the recent
growth of the biopharmaceutical and biotechnical industry has also increased the
pressure to enrich and isolate a host of compounds from complex mixtures and
hence rekindled the interest in chromatography as a preparative rather than an
analytical tool. Given the particular needs of the bioindustries, the importance
of preparative chromatography is expected to grow in the foreseeable future.
Once before, in the 1940s and 1950s, chromatography was the only avail-
able option to separate closely related compounds such as rare earth oxides and
hydrocarbons from crude oil on an industrial basis within the technical means
of those days. Soon, however, the expensive chromatography was replaced by
unit operations such as distillation, extraction, fractionation, etc. A similar devel-
opment seems unlikely in the case of biopolymers, which are often incompatible
with these traditional separation procedures. However, in spite of the obvious
advantages, chromatographic operations are still somewhat awkward to imple-
ment at the industrial scale. Questions of scalability, continuous operation (rather
than the traditional batch approach), and costs obviously need to be addressed.
In order to keep up with the growing demands, existent procedures will have to
be improved and new and more convenient variants to be developed.
Freitag
Time
Time
Figure 1 Separation of a multicomponent mixture by linear elution (A) and overloaded

elution (B) chromatography.
Preparative chromatographic methods are often directly scaled-up versions

of the methods used at the laboratory scale. Thus, the (overloaded) elution lnode
predominates, and separations as illustrated in Fig. 1 are the goal. The substances
are resolved into individual peaks, which are kept apart by substance-free mobile
phase volumes. Usjng this approach, chromatographic units capable of producing
tons of material per year have been built. A further increase in scale within a
reasonable financial and technical framework seems at present unlikely, hence
the growing interest i n truly large-scale systems such as simulated moving beds.
Time
Figure 2 Separation of a rnult~componentmixture by displacement chromatography.

In the meantime, the question of what constitutes the most versatile approach to
biotechnical downstream processing in the kilogram to ton range awaits its final
answer. Especially at larger scale displacement chromatography, a method
whereby the components are resolved into consecutive zones of the pure and
highly concentrated substances (Fig. 2), may become a serious competitor to
overloaded elution chromatography in the establishment of high-resolution, effi-
cient biopolymer isolation schemes.
II. DISPLACEMENT CHROMATOGRAPHY
Displacement chromatography was first recognized as an individual chromato-

graphic mode besides elution and frontal chromatography by Tiselius [I]. To
illustrate the basic principle of displacement chromatography it is useful to first
imagine a substance A distributed at equilibrium in a two-phase system consisting
of a solid adsorbent and a second fluid phase. Under these conditions, a dynamic
equilibrium establishes itself and the relative amounts of bound and free A are
determined by the corresponding equilibrium isotherm. The equilibrium isotherm
can be linear or nonlinear; a nonlinear isotherm can in turn be favorable or unfa-
vorable (Fig. 3). A host of models have been developed to describe the complex
isotherm forms encountered for realistic systems. By necessity we will restrict
ourselves at this point to the most basic patterns.
Mobile phase concentration, c
Figure 3 Cases of equilibrium isotherms: (A) linear isotherm, (B) favorable nonlinear
isotherm, (C) unfavorable nonlinear isotherm.
206 Freitag
When the isotherm is linear, the bound amount is directly proportional to

the concentration adjusted for the fluid phase. When the concentration in the fluid
phase is increased, the number of adsorbed molecules increases linearly. Should
a second substance B be present in the fluid phase, also characterized by a linear
equilibrium isotherm, the interaction of the two substances with the solid phase
can be considered independently. Not so in case of a nonlinear isotherm. When
the isotherm is favorable, the increase in the amount bound decreases with in-
creasing fluid phase concentration, up to a point where it becomes independent
of the fluid phase concentration. The adsorbent surface has been saturated and the
isotherm runs in parallel with the x axis of the isotherm plot (strongly nonlinear
conditions). The opposite behavior is observed in the case of an unfavorable
isotherm. However, since displacement chromatography requires favorable iso-
therms of some kind, we will restrict ourselves to the case of favorable isotherms.
Under nonlinear conditions, the adsorption of a substance A can no longer
be considered independently of the concentration of an also present substance
B. When both substances are at equilibrium with the two phases, a direct competi-
tion for the binding sites ensues and less A will be bound than predicted by the
favorable single-component equilibrium isotherm. The equilibrium isotherm of
substance A has been suppressed by substance B under nonlinear conditions. In
displacement chromatography, the enforced competition for the solid phase bind-
ing sites drives the separation. The first step of a conventional, or batch, displace-
ment separation is the adsorption of the substance mixture on the column (Fig.
4). It is sometimes suggested that conditions more favorable to binding in dis-
placement than in elution chromatography be used, although this is not really
necessary and may even be disadvantageous. A considerable portion of the sta-
tionary phase capacity can and must be exploited during loading, since displace-
ment chromatography is only possible under nonlinear conditions (for definitions,
see Appendix). While the feed is introduced, some separation occurs already,
due to a frontal chromatographic effect.
In the second phase of the experiment, the actual separation, a solution
containing a so-called displacer, is pumped through the column. By definition
the displacer should bind more strongly to the solid phase than any of the relevant
feed components and therefore be able to compete successfully for the binding
sites with all of them. As the displacer front advances, the number of binding
sites available to the sample compounds decreases and the more strongly bound
substances begin to push the less strongly bound ones ahead. Ideally, all sample
components are finally focused into consecutive zones of the pure substances
lined up according to increasing adsorption energy. Following the breakthrough
of the displacer front, the column needs to be regenerated and conditioned for
further use.
The displacement mode was used repeatedly in the early preparative chro-
matographic separation, and biomolecules, such as amino acids, were among the
Displacement Chromatography of Biomolecules
Phase l Phase ll Phase Ill
Feed Displacer Regenerant Carrier
I I
elution volume
Figure 4 Displacement chromatography.
first applications (for an excellent review of the earlier applications of displace-

ment chromatography, see Ref. 2). The success of the earlier separations was
limited, however, since the resolving power in the displacement mode is just as
much dependent on column efficiency as in the elution mode. Only in the 1980s,
when highly efficient high performance liquid chromatography (HPLC) columns
and computers had become ubiquitous and the mathematical tools for dealing
with the problems of nonlinear chromatography had been considerably improved,
was the displacement mode rediscovered and applied to biomolecules, thanks
largely to the dedicated work of Csaba Horvath and his coworkers at Yale.
Mathematically the process can be described by the mass balance together
with the appropriate initial and boundary conditions. The mass balance, the initial
condition, and the exit boundary condition apply to the chromatographic situation
in general.
Mass balance:
(1 1
ac,/at + @ a q i / a t+ u o a ~ i / a=z D ~ ~ c ~ / & ~i = 1,2, . . . , n
where c, is mobile phase concentration: qi is stationary phase concentration;
@ is the phase ratio; u, is the linear flow velocity; D is the diffusion coefficient:
t is time; and z is dimensionless column length.
Freitag
Initial condition:
(2)
ci (0, z) = 0 0 5 z 5 L i = 1,2, . . . , n
Exit boundary condition:

( ~ c ~ I ~ z=) ,0- ~ L = column length
In case of displacement chronlatography the inlet boundary condition is given

by:
c,(t,O) = c ~ , ~0 < t 5 T, i = 1, 2 , . . . , n - I (Sample) (44
and
where T is duration of feed introduction and H(t) is step function.

Since in nonlinear chromatography the adsorption behavior of all,compo-
nents is coupled, another equation is needed to create this link. Usually the change
of q, with time is linked to the mobile phase concentration, c,, of all other com-
pounds by a suitable equation such as the multicomponent Langmuir isotherm
equation:
with a and b as substance-specific constants that can be calculated from the single-
component Langmuir isotherms.
This approach is often a good approximation, since many experimentally
recorded isotherms can be fitted to the Langmuir equation. However, the multi-
component Langmuir isotherm model is less useful in biopolymer chromatogra-
phy for reasons given in Section 111.
The effective use of the stationary and mobile phase capacity is among
the most obvious advantages of displacement over elution chromatography in
preparative separations. Contrary to affinity chromatography, several substances
can be purified simultaneously, which is clearly an advantage whenever the feed
contains more than one substance of value. Another major advantage of displace-
ment chromatography, especially in the context of large-scale biopolymer chro-
matography, stems from the fact that the feed, the displacer, and the regenerant
are introduced as simple step functions. Displacement chromatography is there-
fore much easier to operate in a continuous manner than overloaded gradient
elution chromatography. For example, de Carli et al. succeeded in operating a
continuous annular chromatograph in the displacement mode [3].
Ill. THEORY OF DISPLACEMENT CHROMATOGRAPHY
During a chromatographic separation various processes take place in and out of

the column besides (hopefully) the actual separation (Fig. 5). All of them influ-
ence the eventual result to some extent. There may be:
1. Extracolumn band broadening effects (mixing in valves, detector flow
cells, laminarlturbulent flow in tubings, etc.)
2. Nonideal flow in the column (axial dispersion)
3. External and internal mass transfer resistance (film and pore diffusion
effects)
4. Slow reaction kinetics of the adsorptionldesorption reaction
5. Secondary equilibria (chemical reactions in the fluid or solid phase.
denaturing, etc.)
The modeling of chromatographic separations is more difficult for nonlin-
ear than for linear conditions, due to the fact that the behavior of the various
sample components can no longer be considered independently in this case. The
first detailed analysis of nonlinear multicomponent systems was given by
Gliickauf 141. Subsequently, more general treatments were offered by Helfferich
and Klein I S ] as well as Rhee et al. [6]. All of these earlier approaches as well
as many current ones assume that chromatography is a predominantly thermody-
namically controlled sorption phenomenon, whereas all other (kinetic) effects
play a secondary, modifying role. According to this viewpoint, valuable informa-
tion can be gained from an ideal model. More accurate predictions become possi-
ble by considering some of the above-mentioned band broadening effects; how-
ever, this is feasible only at the price of increased complexity.
In biopolymer chromatography the kinetics of mass transfer and surface
reaction can rarely be neglected completely. These effects can, for example, be
incorporated into the model by the introduction of a dispersion coefficient (axial
dispersion) andlor pore and overall column efficiency parameters (pore and film
diffusion) into the mass balance equation [7,8]. The resulting numerical algo-
rithms have promoted an understanding of nonlinear chromatography to a consid-
erable extent [9,10]. They allow one to take (qualitatively) into account prior to
the design of a preparative separation parameters such as column dimensions,
particle diameter and porosity, mobile phase flow rate, composition and concen-
tration of the feed (and in our case also of the concentration and heterogeneity
of the displacer). Given below is an introduction to the modeling of displacement
chromatography with increasing degrees of complexity.
A. The Ideal Model of Displacement Chromatography

The ideal model of chromatography assumes a separation exclusively controlled
by the sorption equilibrium thermodynamics. The two phases are constantly at
Freitag
Extra column effects
Non-Ideal flow in tubings

(Lamlnar, turbulent)
Injector
Column
Turbulencies in
'mixing chambers"
/- (injector, detector)
T' Detector
Extraparticular effects
B
Molecular diffus~on Eddy diffus~on
(dlfferent pathways)
transfer (fllm dlffuslon)
Figure 5 Sources of dispersion in chromatography.

Displacement Chromatography of Biomolecules 21 1
Intraparticular effects
Pore diffusion Surface reaction
Figure 5 Continued.
equilibrium; there is no axial dispersion; the column efficiency (plate number)

is indefinite. As a consequence, the mass balance [Eq. ( I ) ] simplifies to:
The ideal model has a long tradition going back to Gliickauf and Tiselius in
the theoretical description of displacement effects. In spite of its acknowledged
limitations, it describes correctly the most important aspects of displacement
chromatography and the influence of parameters like the sample and displacer
concentration on a given separation. This is especially the case under highly
nonlinear conditions and for small molecules.
As outlined above, the goal of a displacement experiment is the separation
of the feed components into consecutive zones of the pure substances, i.e., estab-
lishment of the so-called isotachic state or displacement train. For this to happen
it is necessary that the component isotherms be favorable (convex upward). Most
authors assume Langmuir-type isotherm shapes with good results. According to
the ideal model, under these conditions the displacement train will always evolve
after a certain column length.
Under favorable isotherm conditions the front of a substance zone is self-
sharpening until, in the absence of modifying dispersive effects, a triangular
"peak" is formed, where the concentration jumps suddenly to the maximum
value (shock transition) (Fig. 6). This can be observed in overloaded elution chro-
matography for substances with favorable isotherms. In displacement chromatog-
Freitag
Figure 6 Change in the shape of a given substance's zone under increasingly overloaded
conditions in the case of a favorable isotherm.
raphy, the rear boundary of each zone will also be sharpened, since the normally
diffuse rear end will be pushed up by the sharp front of the next zone. Every
molecule that lags behind will immediately be displaced. The opposite is true
for a molecule that for some reason finds itself ahead of its own bulk zone. It
will be among molecules with a lesser stationary phase affinity than itself and
most likely will be retained until overtaken by its own zone.
Once the displacement train has been established, all substance zones move
at the speed of the displacer front, uD.According to the material balance argument
of DeVault 11 11, the velocity of the latter is given by:
UD =
uo
1 + @(A~DJAcD)
or
uD = uo
1 + @(~D/CD)
Since all zones move at the same speed as the displacer front,
,J, = ,J2 = ,Jg = ... = u D
Equation (6b) applies also to the various compounds, i.e., the ratio q/c for each
compound must be equal to qn/cn. An operating line with steepness qD/cDcan
be drawn to determine the concentration in the individual zones (Fig. 7).
The exact concentration within a given zone of the displacement train can
be calculated by equaling Eq. (6b) for the displacer and the compound and resolv-
ing for the compound's concentration. A "water-shed'' point can be determined
Operatlng line
Displacer
6 4 isotherm
Gill C l ~ C~
Mobile phase concentration, c
rlr Displacer
Elution volume
Figure 7 Treatment of displacement chromatography within the hermeneutics of the

ideal model. The set of single-component isotherms and their intersection with the op-
erating line defined by qD/cDis shown together with the corresponding chromatogram.
Substance A elutes ahead of the displacement train, since its isotherm is not intersected
by the operating line. The isotherm of substance B just touches the operating line (water-
shed point), hence the elution of substance B (overloaded conditions) immediately In front
of the displacement train. Substances C and D are displaced. Their concentration in the
pure zones is defined by the intersection point of their isotherms with the operating line.
214 Freitag
where the operating line becomes tangential to the substance isotherm. In this
case, the rear end of the eluting peak will just be touched by the front of the
displacement train. Substances whose isotherms are not intersected by the op-
erating line elute ahead of the displacement train. The only experimental parame-
ter needed for the treatment of displacement chromatography within the herme-
neutics of the ideal model thus are the equilibrium isotherms of the relevant
substances and the isotherm of the displacer.
Concentration in the individual zones is determined by the component's
isotherm and the displacer concentration. The length of each zone depends on
the original amount of the substance in the feed. By changing the displacer con-
centration, one changes both the speed of the separation [uD = f(cD)] and the
concentration of all substance zones. The concentration in the original feed, on
the other hand, is of little importance. According to the ideal model, this is also
the case for trace components and highly diluted feeds. In practice, dispersive
effects will prevent trace components from reaching their theoretically predicted
concentration maximum; however, the application of displacement chromatogra-
phy to isolate and enrich trace components prior to detailed analysis has been
demonstrated in a couple of real-life applications (see Section V1.F below).
Diluted and complex feeds, which are constantly encountered in the case
of high-value bioproducts such as recombinant proteins from mammalian cell
cultures, are both enriched and concentrated by displacement chromatography.
The fact that a concentration plateau rather than a peak is observed can also be
advantageous. While concentration is often a major goal in the bioseparations,
sometimes unwanted effects such as aggregation, denaturation, or maybe only a
troublesome increase in viscosity are observed as a certain critical concentration
is surpassed. In the elution mode the peak maximum would have to stay below
this critical value. Consequently, the concentration of the pooled fraction would
be considerably lower. In displacement chromatography the entire substance zone
could be kept just below this value resulting in a much higher average concentra-
tion in the pool. In at least two published applications, displacement chromatogra-
phy was used in the downstream process with the specific aim of keeping the
product concentration below the critical level [ 12,131.
Under the ideal conditions defined above, the displacement train will even-
tually form. The parameters of this train can easily be calculated given the equilib-
rium isotherms of the involved substances. Nothing has so far been said about
the developing train or the determination of the actual length required for the
development. Both can be calculated based on the ideal model provided the sys-
tem is characterized by Langmuir-type isotherms. Based on these models,
Gliickauf analyzed displacement phenomena and discussed effects of solute and
displacer concentrations as early as 1935. In the late 1960s, Helfferich et al.
presented the first algorithm for a mathematical description of displacement chro-
rnatography based on the theory of interference originally developed for stoichio-

metric ion exchange systems [14]. Rhee et al. later developed a similar theory
based on the theory of systems of quasi-linear partial equations and the methods
of characteristics [I 51.
According to these theories, the system of partial differential mass balance
equations for the various compounds coupled via the multicomponent Langmuir
isotherms is transformed using the so-called h transformation (Helfferich et al.),
or o transformation (Rhee et al.), into a set of simple (and at that time already
solvable) algebraic functions. Although the characteristic parameters cannot be
determined explicitly for systems of more than two substances, their calculation
by simple numerical methods is possible. The development of the displacement
train is usually shown in a normalized distance-time diagram (Fig. 8). The col-
umn length necessary for the development of the displacement train can be taken
directly from this diagram.
In spite of the simplifications the model of coherence proved to be surpris-
ingly suited for the description of displacement separations of small molecules
under highly nonlinear conditions. In 1985 it was extended by Frenz and Horvath
Column Length, cm
0 10 20 30 40 50 60 70
Figure 8 Displacement development graph of a binary mixture. The cross-hatched areas

represent mixed regions, the line-shaded areas pure component regtons. (From Ref. 16.)
216 Freitag
for the separation of proteins by high-performance displacement chromatography

(HPDC) [ 161.
1. The Steric Mass Action Model

The steric mass action (SMA) model of ion exchange displacement chromatogra-
phy also assumes ideal chromatography conditions [17]. The model has been
developed by Brooks and Cramer and describes nonlinear ion exchange chroma-
tography of large molecules (proteins) on the basis of simple mass action. Other
than the models proposed by Velayudhan and Horvath [I 81 or Regnier et al. [19],
the SMA model takes into account that large molecules will not only interact
with certain adsorptive sites on the stationary phase surface, but will also cover
other interaction sites simply due to their bulk (Fig. 9). Most importantly, the
model takes into account the fact that a salt gradient is induced in front of the
displacer front. This gradient causes changes in the displacement train that are
otherwise difficult to account for. Only a few characteristic parameters (the char-
acteristic charge, the steric factor, and the adsorption equilibrium constant) are
needed by the model, all of which can easily be determined experimentally.
The model is only able to simulate the fully developed displacement train
under ideal conditions. Simulations of the developing displacement train or dis-
persive effects are beyond its scope. Within its limitations, however, the reported
:,I masked 1 hindered counter ion
Figure 9 Idealized protein binding to an ion exchanger. (From Ref. 17.)

agreement with the experimental results is good. The model has since been ex-
tended to the description of immobilized metal affinity chromatography (IMAC)
WI.
B. The Equilibrium-Dispersive Model of Displacement

Chromatography
The ideal model of chromatography enjoys continuing popularity in theoretical
displacement chromatography. Good qualitative agreement between the theoreti-
cal predictions and the experimental results have been reported. This is not sur-
prising considering the conditions of the connected experiments, which employed
high-efficiency columns (several thousand plates per meter), strongly nonlinear
conditions (high sample concentration. large amounts), and were aimed for the
separation of small molecules (fast mass transfer and reaction kinetics). Such
experimental conditions approach ideal conditions to a high degree.
However, truly discontinuous concentration changes (shocks) as predicted
by the ideal model are never observed during the experiments, no matter how
high the column efficiency. Good displacement separations are instead character-
ized by steep but continuous changes in concentration, since dispersive effects
(molecular and eddy diffusion, mass transfer and reaction kinetics, etc.) counter-
act the equilibrium thermodynamics. The implementation of dispersive effects
into the simulations requires a model that somehow takes the limited column
efficiency into account. This is more difficult in nonlinear chromatography, where
the behavior of each compound depends on that of all others, than in linear chro-
matography, where the bands of the various compounds can be calculated inde-
pendently.
The simplest approach to do this is the equilibrium-dispersive or semideal
model. This model mirrors the situation of modern chromatography by assuming
the dispersive effects to have a modifying but not a fundamental influence on
the final band profile. In many cases this model is fully sufficient to predict the
influence of the dispersive effects on the separation correctly.
To accommodate for high but not infinite column efficiency, a constant
bulk axial dispersion coefficient, Dhull,is introduced into the ideal mass balance
equation. Equation (Ib) thus becomes:
In most cases the assumption of a constant bulk dispersion coefficient is

reasonable at sufficiently high column efficiencies. Large biopolymers (DNA,
proteins), however, tend to show a high and concentration-dependent viscosity
in their solutions. For such molecules Dhullbecomes concentration-dependent and
the unmodified equilibrium-dispersive model no longer applies.
An experimental value of the bulk dispersion coefficient can be derived
218 Freitag
by applying the plate model of chromatography (tanks in series model). The

characteristic parameter of the plate model, the height equivalent of a theoretical
plate (HETP), takes axial dispersion and mass transfer kinetics into account
[21,22]. For sufficiently high plate numbers (>100/m) the two approaches yield
similar results and the dispersion coefficient becomes proportional to the plate
height, H, and the plate number, N, of the column:
DbUn= Huo/2 = Luo/2N, with N = L/H (8)
The plate height can be calculated from the peak width and the retention time
of a given tracer under linear conditions as shown in Fig. 10.
In addition to the equilibrium isotherms, the equilibrium-dispersive model
requires the experimental determination of the column's plate height as a function
of the carrier flow rate (Knox or van Deemter curve). It is assumed that the plate
height of a given column is identical under linear and nonlinear conditions.
Currently, there are no closed-form analytical solutions to the equilibrium-
dispersive model. The comparative simplicity of the model facilitates the calcula-
Time
Figure 10 Calculation of the column plate height from experimental results (linear chro-
matography).
tion of numerical solutions to the relevant equations, using computation methods

such as finite differences, finite elements, or collocation. For an excellent intro-
duction to the numerical modeling of displacement chromatography, see Guio-
chon et al. [23].
Numerical solutions to differential equations always involve the introduc-
tion of a truncation error. A very elegant way of turning this into a positive feature
is to replace the real axial dispersion by a numerical dispersion artifact, i.e.,
choosing, e.g., the finite differences in such a way that the numerical error caused
by the truncation equals the dispersive term on the right-hand side of Eq. (lc),
which is in turn dropped from the equation.
Using numerical calculations, the band profiles can be calculated for any
specified parameters and initial conditions. As an example, the development of
a displacement train as predicted by the equilibrium-dispersive model is shown
in Fig. 11. The predictions show good agreement with the experimental results.
The basic features of a displacement train are determined by the equilibrium
thermodynamics. The dispersive effects do not introduce new phenomena. In-
stead their main effect is a smoothing out of the sharp edges predicted by the
equilibrium theory. The overall features, such as the length and height of the
zones, stay the same. The only difference is that instead of an abrupt change a
small overlapping zone containing both substances develops between two consec-
utive substances in the train. This layer is also called the shock layer. The shock
layer has similar properties (velocity, etc.) as the shock: its thickness depends
on the dispersive effects.
The equilibrium-dispersive model yields good results as long as the ob-
served band profiles are indeed determined primarily by the nonlinear thermody-
namics of equilibrium as described by the ideal model. However, all contributions
to the band broadening are considered to be simple additions to the axial disper-
sion in this model. The fact that the various contributions to the bulk coefficient
differ in their dependency on the concentration is ignored. This approximation
is no longer valid once the column efficiency becomes low and mass transfer
and reaction kinetics begin to determine the band profiles to a similar extent
as the equilibrium thermodynamics. In addition, the bulk dispersion coefficient
depends on the retention factor, k'. The concentration dependence of k' is also
ignored in the model, which assumes k' to equal k;, i.e., the value determined
for linear conditions. This approximation also generates significant error at low
column efficiencies.
C. Kinetic Models for Displacement Chromatography

The equilibrium-dispersive model lumps all dispersive effects into a single coef-
ficient. As we have seen, simulations obtained with this model agree well with
the experimental results as long as the assumption holds that the kinetic effects
C mg/ml
200 , a
I
100 I/
1
1
I, min
0 -
0 5 10 1 25 30 1. min 35 .
Figure 11 Development of a displacement train as predicted by the equilibrium-disper-

sive model. (From Ref. 23.)
are merely additions to the dominating axial dispersive effects. As we have also
discussed, this assumption is not always allowed for biopolymers, since the mass
transfer resistance is often several orders of magnitude higher in the case of these
large molecules (small diffusion coefficients). In contrast, certain types of chro-
matography popular in biopolymer separation are characterized by slow sorption
kinetics. Affinity chromatography, for example, quickly becomes surface reaction
limited, which may become a problem even in elution chromatography at elevated
flow rates. In cases like this, the equilibrium-dispersive model is not capable of
providing helpful simulations. Instead a kinetic model should be used.
In the kinetic models the mass balance equation [Eq. (I)] is combined with
a kinetic equation relating the rate of variation of the concentration of each com-
ponent in the stationary phase to its concentration in both phases and to the equi-
librium concentration in the stationary phase. Solutions to the equations of the
kinetic models are usually obtained numerically as for the equilibrium-dispersive
model. Various models have been proposed, which vary mainly in the choice of
the kinetic rate expressions (see table below).
An important decision concerns which degree of complexity is necessary
in a given situation. As long as the kinetics are not very slow, the various kinetic
models yield similar results, which often resemble the results of the equilibrium-
dispersive model [24]. As the influence of the kinetic terms increases, the various
models yield results of differing accuracy.
Mass transfer resistance Ref.
Negligible Lapidus and Amudson [25]

Langmuir isotherm with quasi-chemical Thomas [26], Goldstein [27], Wade et al.
mass transfer rate [28l
Linear rate equation
first-order kinetics Lapidus and Amudson [25]
solid film linear driving force Gliickauf and Coates [29], Hiester and
Vermeulen [30], Lin et al. [3 I], Phil-
lips [8]
liquid film linear driving force Guiochon et al. [23]
Intraparticle diffusion Rasmuson and Neretnieks [32]
Intraparticle diffusion/external film diffu- Rasmuson and Neretnieks [32]
sion
Dual intraparticle diffusion/external film Rasmuson [33]
diffusion
Theoretically, a general rate model that explicitly considers all effects will
be the most correct. However, it is also the most complex one and requires the
experimental determination of all involved rate constants. In most cases, how-
ever, a kinetic model that lumps all kinetic effects into a single expression will
suffice.
For increasing mass transfer resistance, such a lumped model predicts a
decrease in the length of the zones in the displacement train and a concomitant
increase in the shock layer thickness. For very small mass transfer coefficients
the plateaus disappear completely and the resolution between successive bands
222 Freitag
is poor. An asymptotic solution can be obtained for the concentration profile in

the shock layer once the displacement train has been established (Fig. 12). It is
usually assumed that both the diffusion coefficients and the rate constants are
equal for all compounds; otherwise the actual shock layer thickness lies between
the values calculated for the two values.
The shock layer thickness, AqL,between two successive zones in the iso-
tachic train according to Guiochon is given by:
with Kd = kA/(l + bdcd'),kl is the film mass transfer coefficient, and 9 is the
characteristic parameter (see Fig. 12 for details).
The shock layer thickness thus depends on the axial dispersion coefficient
and the mass transfer coefficient of the two components, on their separation fac-
tor, and on the concentration and retention factor of the displacer [34]. In a given
train the shock layer thickness depends only on the value of a. It does not depend
on the retention factor or the feed concentration of the components.
C',
Figure 12 Schematic presentation of the shock layer concept. (From Ref. 23.)
A differentiation Eq. (9) shows that the shock layer thickness would be at
minimum for Kd = 1, i.e., for
k: = 1 + bdcd (104
Consequently, the shock layer is wide whenever k: << 1 + bdcd,i.e., at low

displacer retention or at high displacer concentration (so-called overdisplacement
phenomenon). For k: >> 1 + bdcdthe shock layer thickness increases linearly
with increasing k:. As shown by Zhu and Guiochon, there is no optimum dis-
placer retention factor and optimum displacer concentration in displacement chro-
matography, but a combined optimum given by the above Eqs. (10a) and (lob).
Just as in elution chromatography, however, there is an optimum mobile
phase velocity, given by [35]:
At low flow velocities, diffusional effects caused by the steep concentration

gradients cause a widening of the shock layer. At high flow velocities, the limited
mass transfer and reaction kinetics again cause a shock layer broadening. Con-
trary to the situation in elution chromatography, however, in displacement chro-
matography the optimum mobile phase velocity depends not only on the axial
dispersion and mass transfer resistance, but also on displacer properties such as
retention factor and concentration [34]. Differentiation of Eq. (1 1) shows that an
optimal u,, is obtained in the case of Kd = 1. In practical terms this means
that the optimum flow rate will usually be lower in displacement than in elution
chromatography.
Application of the shock layer theory to the displacement situation permits
prediction of the influence of parameters such as column length and plate height,
particle diameter, carrier flow rate, or sample size and concentration. Briefly, one
can state the following rules. These rules apply to the established displacement
train, i.e., under conditions whereby a dynamic equilibrium between the band
sharpening effects of the thermodynamics and the eroding effects of the finite
column efficiency has been fully established (constant pattern behavior). As a
consequence, the described limitations correspond to the best achievable condi-
tions for a given experimental setup.
Plate height. The shock layer thickness and concomitantly the intermixing
of the substance zones increases with increasing plate height (decreasing
column efficiency); consequently, the recovery yield decreases. As a con-
sequence, high-efficiency columns are just as important in nonlinear
Freitag
(preparative) chromatography as in linear (analytical) chromatography.

The fact that displacement chromatography makes better use of the col-
umn capacity often allows the use of high-efficiency analytical columns
at a semipreparative scale. Parameters influencing the plate height (flow
rate, particle diameter) have a direct influence of the shock layer thick-
ness and thus of the chromatographic result.
Displacer concentration. As already seen for the ideal model, an increase
in the displacer concentration causes an increase in the substance zone
concentration with a concomitant decrease in zone length. While this
was not a problem under ideal conditions, the introduction of the shock
layer concept shows that concomitant to the narrowing of the substance
zone the relative amount found in the shock layer increases. The term
"overdisplacement" has been coined for such a situation where bands
are so narrow that no concentration plateau of the pure substance is ob-
served while the two shock layers in front and at the end of the zone
touch.
Column length/.sample size. For a given column diameter, the column
length required to achieve the displacement train increases with increas-
ing sample amount. In contrast, the concentration hardly matters. Thus
even highly diluted feeds can be processed by displacement chromatog-
raphy. The loading factor is inversely proportional to the column length
and the loading factor for which the isotachic train is formed remains
constant.
Separation factor. The separation factor a , as defined as the quotient be-
tween the capacity factors, k', of any two compounds, is another impor-
tant chromatographic parameter, whose influence on the displacement
train cannot be fully accounted for by the ideal model. Even with the
equilibrium-dispersive model there is no simple equation linking the sep-
aration factor to the development of the displacement train. However, it
can be shown that the column length required to achieve the displace-
ment train increases rapidly as the separation factor approaches 1. The
shock layer thickens in proportion to ( a + l ) l ( a - 1) and increases
dramatically as a decreases toward unity. Displacement chromatography
is therefore not superior to elution chromatography in resolving binary
mixtures with a very small value of a . The shock layers would encom-
pass the entire displacement train, while the time required to achieve
isotachic conditions would result in a very low throughput. The fact that
displacement chromatography is equally suited to the separation of
closely related substances has been demonstrated by separation of the
genetic variants of fLlactoglobulin B [36], or structural and geometric
isomers (see Section V1.D).
Trace compounds. According to the ideal model the concentration in a

substance zone is independent of the original sample amount. Thus a
zone may become infinitesimally narrow in the case of a trace compound.
According to the equilibrium-dispersive model, a trace component will
be considerably enriched in displacement chromatography, often much
more so than in elution chromatography. However, when the zone width
reaches the same order of magnitude as the shock layer thickness, no
further enrichment takes place. Instead the zone width stabilizes. This
behavior is also observed in real-life applications of displacement chro-
matography.
Displacer Inzpurities [37]. Any displacer impurity acts as an additional
(trace) compound in the displacement train. Impurities with an equilib-
rium isotherm below that of certain compounds of the separation mixture
will contaminate the displacement train. If the separation factors between
the bulk displacer and the impurities are close to 1, formation of the fully
developed displacement train may be difficult. Therefore homogeneous
displacers are preferable.
These predictions are qualitatively correct for any isotherm type provided
its shape is convex and no intersection occurs. The latter, however, is quite a
common phenomenon in biopolymer chromatography, since the adsorption en-
ergy tends to increase with molecular mass, while the saturation capacity de-
creases in the same direction. A large molecule will therefore be characterized
by an isotherm with a steeper initial slope but a lower saturation plateau than a
similar but smaller molecule. The result may be an isotherm crossing leading at
worst to "elution azeotropes," which no column despite its length will be able
to dissolve [38]. Obviously this cannot be accounted for by the competitive
Langmuir-isotherm model, which assumes constant separation factors [7].
Other models such as the LeVan and Vermeulen isotherm derived from
the theory of the ideal adsorbed solution (JAS) [39] and modified Langmuir mod-
els [40] have been suggested to describe the multicomponent adsorption behavior
of complex molecules under linear and nonlinear chromatographic conditions.
Antia and Horvath have used this theory as a basis for their investigation of the
selectivity reversal phenomenon [38]. Langmuir-type single-component iso-
therms were also assumed in their case. They were able to show that the result
of an isotherm crossing is the development of a separation gap in the system
(Fig. 13). The position of the gap depends on the saturation capacities of the two
compounds. If the operating line crosses the isotherms in this gap, no separation
takes place. If it intersects on the right- or the left-hand side of the gap, a separa-
tion of the two compounds by displacement is possible. However, the order of
the substances in the displacement train is opposite in the two cases. If the initial
Freitag
slope of the isotherm A is higher than that of substance B, whereas the opposite
is true for the saturation capacities, A appears before B in the displacement train
if the operating line intersects on the right-hand side of the separation gap and
B before A if the points of intersection are placed to the left of the gap. The
occurrence of an adsorption azeotrope has been observed experimentally, e.g.,
by Carta and Dinerman for the separation of a-aminobutyric acid and isoleucin
on Dowex 50W-X8 resin [41] and by Kim and Crarner for protein separations
in the immobilized metal affinity chromatography (IMAC) mode 1421. A similar
reason was proposed by Kasper et al. for their inability to resolve a mixture of
antithrombin I11 and bovine serum albumin (BSA) on hydroxyapatite columns
under certain experimental conditions [43].
IV. PRACTICAL METHOD DEVELOPMENT IN

DISPLACEMENT CHROMATOGRAPHY OF
BIOMOLECULES
The theoretical basis of biomolecule and especially biopolymer displacement

chromatography is currently less well developed than that of smaller molecules.
Simulations can in most cases aide but not replace the experimental development
of the method. Bio(po1y)mer displacement chromatography by necessity often
takes place under exactly those conditions that were explicitly excluded in the
theoretical treatment of the displacement phenomenon, i.e.:
At comparatively low concentration due to the limited solubility of the

biopolymers and to the problems with viscosity observed for highly con-
centrated DNA and protein solutions
In the presence of secondary equilibria (denaturation, aggregation, reac-
tions)
In non-Langmuirian systems characterized by complex, sometimes cross-
ing or unfavorable isotherms
At comparatively low column efficiency
Under conditions whereby mass transfer and reaction kinetics are dominant
and cannot be treated simply as additions to the axial dispersion
Under conditions whereby the relevant parameters are noticeably concen-
tration-dependent
The growing number of examples of successful biodisplacement chroma-

tography shows, however, that displacement chromatography is less limited by
these circumstances than one would assume from the theoretical considerations.
Development of a biodisplacement chromatography will usually involve
the following steps:
Freitag
Choosing the stationary phase

Optimizing the mobile phase
Adjusting the column lengthlsample size
Adjusting the flow rate
Finding a displacer and, perhaps,
Adjusting the temperature (an increase in temperature may result in benefi-
cial effects such as a decrease in viscosity or improved mass transfer
and reaction kinetics. Many biopolymers are sensitive to elevated tem-
peratures, however, so rarely is this an option in preparative chromatog-
raphy .)
Since in displacement chromatography the substances are separated into

consecutive zones, the monitoring of a displacement separation can be more dif-
ficult than in elution chromatography, especially in the semipreparative scale
where an unspecific (UV) detector can usually still be used to monitor an elution
separation. In displacement chromatography fraction collection is inevitable. The
use of on-line analytical HPLC has been suggested to monitor the displacement
train and control the fraction collection [44]. Such an approach may become
necessary to allow the full automation of the system and to reduce the process
time, which up to now is largely increased by the amount of time needed to carry
out the off-line fraction analysis.
A. Stationary Phase
Choosing the stationary phase also involves choosing the interaction mode for
the separation. The vast majority of the published protein displacement separa-
tions has been done on an ion exchange column. Displacement separations of
peptides and other smaller biomolecules are mostly carried out in the reversed-
phase mode. In these standard cases a host of guiding examples can be found in
the pertinent literature; a number of the more recent ones will be discussed below
in Section VI. Occasionally other stationary phases have been used, including
immobilized metal affinity chromatography (IMAC), antibody exchange (Abx),
hydroxyapatite, or hydrophobic interaction chromatography (HIC) phases (see
Section VI for details).
Today's high-performance stationary phases are in general not designed
for displacement chromatography. If anything, the differences between stationary
phase materials and columns from different suppliers is even more pronounced
in displacement than in elution chromatography. It is thus highly advisable to
investigate a number of columns before coming to a final decision. Some attention
should be paid to the availability of the bulk material. The column length is an
important parameter in displacement chron~atography(see below) and the length

of the available prepacked colun~nsmay not be optimal.
The particle diameter of the stationary phase is another important parame-
ter, since mass transfer effects have a substantial effect in biopolymer displace-
ment chromatography. Smaller particles stand for higher efficiencies, i.e., lower
theoretical plate heights. However, for a given column length the pressure drop
increases considerably with decreasing particle diameter. Most authors advise to
use particles of less than 20 pm, although particles of 80 pm and more have
occasionally been used with success in displacement chromatography [45]. Theo-
retically, a minimum in d, exists, below which a further decrease in the particle
diameter will not improve the separation any further since the reaction kinetics
become rate limiting [46]. The particle size distribution should be narrow.
Since mass transfer can be even more limiting in displacement than in elu-
tion chromatography, the recent arrival of stationary phases for biopolymer chro-
matography that improve this particular feature may become interesting for dls-
placement chromatography. Examples include the perfusion chromatography beads
(PerSeptive Biosystems Inc., Framingham, MA, USA), the HyperD columns (Ri-
osepra Inc.. Marlborough, MA, USA), and the continuous bed column (UNO
column) recently introduced by Bio-Rad Inc. (Hercules, CA, USA) (Fig. 14).
In perfusion particles the mass transfer is supposedly improved due to the
presence of large throughpores in the particles. Convective flow is possible in
the throughpores; hence an improvement of the mass transfer. However, the mo-
bile phase flow velocity needs to be quite high-much higher than usual in tradi-
tional displacement chromatography-for the perfusion phenomenon to occur.
Even then only some 5% of the flow passes through the particle. Perfusion dis-
placement chromatography has been used to separate the genetic variants of p-
lactoglobulin at a flow rate of 4 mllmin [47]. Eighteen milligrams was thus pre-
pared within 90 seconds using heparin as displacer. The authors claimed a resolu-
tion similar to if not better than that achievable in conventional displacement
chromatography at much lower flow rates.
HyperD particles consists of a ceramic support filled with a gel. Mass trans-
fer is highly efficient in such beads, since the intraparticular diffusion of the feed
molecules takes place on the surface of the stationary phase (hyperdiffusion).
A continuous bed column is formed by direct radical polymerization of
monomers in a tube. The polymer molecules form small nodules (<0.1 ym),
which aggregate into a highly porous polymer rod. No extraparticular dispersive
effects are possible in a UNO column. Their van Deemter curve usually shows
an unusually low optimum, since the nodule size seems to determine the A term.
The column efficiencies remains more or less constant even at elevated flow rates
in elution chromatography. A 3.5-cm-long anion exchange UNO column (column
volume 1 ml) was recently used for displacement chromatography of whey pro-
Conventional porous bead HyperD bead
B
Flux = (c,- O)/d,) Flux = (q, - O)/d,)
Figure 14 Types of stationary phase materials with supposedly improved mass transfer
features: (A) perfusion material (Perfusion chromatography uses a bead with 2 types of
pores; throughpores: 6000-8000 A; diffusive-pores; 500-1500 A; throughpores enable
the eluent to pass the beads; diffusive pores are short and allow faster separation times.)
(9) hyperdiffusion material, (C) continuous bed column. Continuous Bed matrix unifor-
mity minimizes the band broadening seen in conventional packed beds. The non-porous
surface allows extremely fast mass transfer, minimizing band broadening even at high
flow-rates. The fimbriated surface structure of the nodules provides a large surface area
for good binding capacity.
7
Figure 14 Continued ,
teins. Compared to similar experiments on traditional beaded supports the shock

layer between the two protein zones is small in spite of the shorter column length
(3.5 cm for the UNO column versus 5.2 cm for the traditional column). However,
the possibility of using elevated flow rates with this column was not investigated.
6. Mobile Phase
In displacement chromatography the mobile phase is often merely considered as
an inert carrier. Concomitantly the use of conditions that aid strong binding of
the substances and the displacer to the stationary phase are advised. However,
this has not been shown to improve the separation. Instead, low retention factors
may actually improve displacement separations. The theoretically predicted opti-
mum values lay between 1.2 and 2.0 [46]. The influence of the separation factor,
a,on the shock layer thickness between two consecutive substances in the dis-
placement train should be kept in mind. A high value of a helps to achieve good
resolution between consecutive zones. Given the tendency of biopolymers to
show all-or-nothing binding behavior, the deciding factors in choosing the mobile
phase in biodisplacement chromatography are often less chromatographic and
more biological in nature, i.e., prevention of denaturation, interactionlaggrega-
tion, or solubility.
C. Column LengthEample Size

The optimization of column length and sample size should be interactive. Some
intuition has to be used. The displacement train needs a certain distance to de-
232 Freitag
velop and conditions of nonlinearity should prevail during the separation. For
any given sample size there is an optimum column length and vice versa. Dis-
placement chromatography of proteins has been shown on columns as short as
5 cm. Extending the column length beyond the optimum will not improve the
separation any further. Often it is easier to adjust the sample size to the dimen-
sions of a given column rather than optimize the column length for a given sam-
ple. For modern high-performance columns the utilization of up to 80% of the
column's capacity should be possible.
Finally the question to be addressed is whether the full development of the
displacement train is always necessary. While this does maximize the recovery
yield, it does not necessarily correspond to the highest productivity. Throughputs
may, for example, be higher when the column is not long enough to allow the
full development of the displacement train and instead the mixed zones are recy-
cled. The decision will depend on the position of the target molecule in the dis-
placement train and the stability of the molecule.
D. Flow Rate
The optimum flow rate in displacement chromatography will be up to one order
of magnitude lower than in elution chromatography. In almost all cases of bio-
polymer displacement chromatography, this optimun~flow rate will therefore be
much too low to be of practical relevance. Most authors use flow rates between
0. I and 0.5 mllmin in biopolymer displacement chromatography in the semiprep-
arative scale (typical column dimensions cm X mm). At higher flow rates, the
quality of the separation tends to decrease rapidly, although flow rates of several
n~illilitersper minute have occasionally been used with success even with analyti-
cal scale columns [48].
The lower flow rates in displacement chromatography do not necessarily
result in lower throughputs or in an economic disadvantage. In our laboratory,
for example, identical columns were used in the displacement and the overloaded
elution mode for the separation of whey proteins. The preparation of I g of a-
laktalbumin of comparable purity took 5 times as long in the elution than in the
displacement mode in spite of the one order of magnitude higher flow rate in the
elution column (1 mllmin versus 0. I mllmin). The fact that the sample volume
was much smaller in the elution case was largely responsible for this. In addition,
the two whey proteins were concentrated by a factor of 3 in the displacement
case, whereas dilution was observed for the pooled fractions collected from the
elution column. A similar observation was made by Gerstner 1493 for the separa-
tion of oligonucleotides by displacement and elution chromatography. Although
the flow rate was twice as high in the elution than i n the displacement mode and
the feed per run similar (with 15 L the elution column was 3 times as large
as the displacement column), 31 1 runs were required to produce 5 kg of product
in the displacement mode versus 467 runs in case of the elution column. The
number of production days was 26 in the displacement mode compared to 39

days in the elution mode. Due to the higher productivity, the lower consumption
of solvent, chemicals, and labor, as well as the much better recovery, the produc-
tion costs for this amount of product would amount to $3.657 million in the
elution case compared to only $2.677 million in the displacement case.
E. Displacer
The preceding subsections dealt with aspects of displacement chromatography,
which are not very different from similar considerations in elution chromatogra-
phy. This is not so in the case of the displacer, which is a unique feature of
displacement chromatography. At the same time, the choice of the displacer has
consequences not only for the success but for the economic soundness of the
final method.
The ideal displacer should have the following features:
It should be nontoxic, stable, detectable, and cheap.
It should combine high solubility in the carrier with a high binding tendency
toward the stationary phase.
Regeneration of the column should nevertheless be possible.
In addition, the displacer should be uniform and its removal from the product
zone possible. For pharmaceutical applications it might even be necessary to
sterilize the material.
Mixtures of small biologicals such as amino acids, peptides, and small pro-
teins (antibiotics, insulin) are usually processed by reversed-phase displacement
chromatography. Hydrophobic substances such as 2-(2-butoxyethoxy)ethanol
(BEE), decyltrimethylammonium bromide, cetyltrimethylammonium bromide
(cetramide), benzyldimethyldodecylammonium bromide, dodecyloctyldimethyl-
ammonium chloride, and palmitic acid are standard displacers for these applica-
tions [2]. Nucleotide and nucleoside mixtures have been separated using a similar
combination between reversed-phase stationary phases and hydrophobic dis-
placers. The separation of oligonucleotides is also possible in the anion exchange
mode using dextran sulfate as displacer [49].
Ion exchange chromatography is very popular in preparative protein chro-
matography because it is known to preserve biological activity to a high degree.
To this day the majority of protein purification schemes contains one or several
ion exchange steps. While hydrophobic interaction chromatography is gaining
ground in preparative elution chromatography, the ion exchange mode is still
almost exclusively used in protein displacement chromatography. The lack of
suitably hydrophobic protein displacers for the hydrophobic interaction mode
may be among the reasons for this, since it is easier to find a highly aqueous
mobile phase-soluble polyion than similar hydrophobic polymer. For ion ex-
change applications, a number of (semi)synthetic polyions such as chondroitin
234 Freitag
sulfate, dextran sulfate, carboxymethyl starch, alginate, Eudragit, Nacolyte 7105,

and polyethyleneimine (PEI) have been suggested as protein displacers [SO].
Since 1978, Torres and Peterson have promoted the use of (modified) carboxy-
methyldextrans (CM-D) for that purpose [5 11.
In almost any case a protein that is known to bind exceptionally well to a
given stationary phase material may be used as a displacer of less well-bound
proteins. This approach has also been used repeatedly, last but not least in the
few attempts to do hydrophobic interaction displacement chromatography [52].
For example, protarnine was suggested as a protein displacer in cation exchange
displacement chromatography [53], whereas heparin may act as a nontoxic dis-
placer in anion exchange methods [54]. While proteins as protein displacers yield
important data, proteins will hardly become industrial-type biopolymer dis-
placers. Most other substances named above are cheaper and thus more suited
to an industrial application. Unfortunately, they are usually highly heterogeneous,
difficult to detect, and almost impossible to recycle.
While most applicants clearly prefer to use large-protein displacers, certain
low molecular weight substances are also discussed as putative protein displacers
in the ion exchange mode, since it was possible to show that even smaller mole-
cules in the range of several hundred to several thousand grams per mole can be
effective displacing agents for much larger molecules [55-591. The removal of
any displacer contaminant from the product zone should be much easier in the
case of the small displacer molecules, since the large difference in size can be
exploited.
Examples of low molecular weight protein displacers include low molecu-
lar weight dextran sulfates [58] and pentosan polysulfate (M,: 3000 glmol),
which were used for the separation of the genetic variants of p-lactogobulin [56].
Polyvinylsulfonic acid (M,: 2000 glmol) [59] and pentaerythritol-based dendritic
polymers (M, 480-5 100 glmol) were used as displacers of basic proteins such
as a-chymotrypsinogen and cytochrome c on cation exchanger materials [57].
Modified ethylenglycols (M,: 1000-10,000) and small chelating agents such as
EGTA (ethylenglycolbis(~-aminoethylether)-N,N',-tetraacetic acid, M,:
380.4 glmol) and IDA (imminodiacetic acid, M,: 133.4 glmol) were suggested
as displacers of recombinant proteins and whey proteins from apatite columns
[55,60,61].
A comparison of the results obtained with these small displacers to those
obtained for similar but larger molecules shows that it is charge density and hence
adsorption energy rather than size and absolute number of interaction points that
determines the quality of a displacer. It was also seen, however, that the behavior
of small displacers depends to a much higher degree on the chromatographic
conditions, e.g., the salt content of the mobile phase. As a consequence, the switch
from displacer to elution promoter is more likely in the case of these small sub-
stances.
Among the host of synthetic and semisynthetic substances that have been
used for protein displacement, few if any have been synthesized with that explicit
goal in mind. Torres and Peterson were among the first to chemically modify
their (high molecular mass CM-D) displacers in order to gain control over the
stationary phase affinity [62]. An interesting approach to displacer design was
recently suggested by two groups. These synthetic displacer molecules mimic
certain features of biopolymers (proteins) such as their solubility behavior in
aqueous solutions. The triblock polymethacrylate-based displacers prepared by
Patrickios et al. [63] using group transfer polymerization carry a sequence of
positively charged groups at one end of the molecule and a sequence of negatively
charged ones at the other. In the middle a neutral, hydrophobic block is created
to separate the two. The resulting polymers show an isoelectric point much like
that of proteins and can be precipitated at the corresponding pH. The authors
claim good results with these substances as displacers for anion exchange dis-
placement chromatography.
A similar approach to the problem of displacer recovery, either to rid prod-
uct zones from contaminating displacer traces or to establish a recycling scheme,
was used by Vogt and Freitag [64]. In this case a copolymer was synthesized by
radical polymerization that showed a lower critical solution temperature (LCST),
a phenomenon that is based on a similar physicochemical basis as the salting-
out effect observed for many proteins. Due to a finely tuned balance between
hydrophobic and hydrophilic groups, LCST polymers are water-soluble at low
temperature, but precipitate rapidly when the temperature is increased and a cer-
tain critical temperature, the LCST, is passed. The exact value of the LCST can
be adjusted from between 10°C and 90°C, whereas the displacing character of
the molecules stays roughly the same. The polymers can be redissolved simply
by lowering the temperature again. The cycle can be run through several hundred
times; no unspecified protein coprecipitation was observed in the investigated
cases. The displacers were used in combination with anion exchange and hy-
droxyapatite columns.
V. SPECIAL FORMS OF DISPLACEMENT

CHROMATOGRAPHY
Most applications of displacement chromatography found in the literature deal

with a separation as outline above. The mixture components are focused into
consecutive individual zones by means of a displacer. The displacer should ide-
ally be a homogeneous substance. However, two variants of this schema have
been developed that use a somewhat different approach. So-called spacer dis-
placement chromatography does not employ a single displacer substance to de-
velop the displacement train but uses a mixture of related compounds as spacer1
236 Freitag
displacers [62]. The components of the spacerldisplacer mixture vary in their

adsorption energy within the range of the feed components. As a result, the more
strongly bound molecules of the mixture act as displacer in a manner similar to
that of any ordinary displacer, whereas the less strongly bound ones act as spacers
between the target molecule zones. Since the spacers are usually chosen to be
non-UV-active, the monitoring of the displacement train is facilitated. The spacer
displacement chromatography approach is, for example, necessary in thin-layer
displacement chromatography (TL-DC), where the analysis of the displacement
train would otherwise be difficult. In ordinary column liquid displacement chro-
matography such an approach has also been used (see Section VI). However,
this always entails the contamination of each recovered substance by at least two
spacer substances or a loss in yield corresponding to the rejection of the shock
layer fraction.
Complex displacement chromatography is related to ordinary displacement
chromatography less through its mechanism and more through the chemicals
used. A typical application of complex displacement chromatography would be
the isolation of a cationic protein (e.g., an mAb) using a cation exchange. column
and substances such as the carboxymethyldextrans (CM-D) ordinarily used as
displacers in anion exchange displacement chromatography [65].In complex dis-
placement chromatography, the "displacer" forms a complex with the adsorbed
target molecules. Once the net charge of the complex becomes sufficiently low,
the entire complex desorbs from the surface.
VI. APPLICATION OF DISPLACEMENT

CHROMATOGRAPHY
The number of applications of displacement chromatography for bioseparations

is still small compared to the elution mode. It is, however, already much too
large to be discussed in detail on a case-by-case basis. The following necessarily
incomplete list attempts to give an overview of what has already been done. The
listed conditions apply to the actual displacement steps. In some cases, different
flow rates or temperatures were used during sample loading.
Most authors working in the field of biodisplacement chromatography tend
to use standard mixtures to investigate the fundamental parameters andlor to
demonstrate the validity of simulations. While these exemplary separations are
interesting to anybody intending to develop a biomolecule displacement separa-
tion, the results are usually much more straightforward than anything achievable
for a "real" sample; they were not included in the following list for obvious
reasons. Instead we tried to compile published examples of displacement separa-
tions of crude biomolecule mixtures with practical relevance.
A. Protein Separation
Target molecule Conditions Comments Ref.
Crude P-galactosidase Method: weak anion Displacement mode su- [36]

(industrial enzyme) IE-DC perior in throughput,
from Aspergillus Column: two TSK less waste
oryzae DEAE-SPW column
in series, 75 X 7.5
mm
Flow rate: 0.2 mllniin
displacer: chon-
droitin sulfate
Human serum proteins Method: anion IE-DC General method for [66]
Column: DEAE fractionation of com-
Bio-Gel A, DEAE plex protein mix-
Selectagel, 140 X tures
5.5 mm Fractions are low in
Displacer: gradient of salt and can be di-
CM-D with increas- rectl y analyzed by
ingly higher content electophoresis
of carboxyl groups Methods for displacer
modification are dis-
cussed
Human Gc-2 globulin Method: two-anion IE- The protein was iso- [67]
DC at different pH lated from the blood
followed by elution of psoriasis patients
chromatography on and only known as
hydroxyapatite col- one of more than
umn: DEAE Sepha- 100 spots in 2D elec-
cel, 7 ml trophoresis, 6 ml of
Flow rate: 5 mllh serum (400 mg pro-
Spacerldisplacer: dif- tein) gave 0.5 mg
ferent CM-D pure substance
After the second dis-
placement step the
protein was only con-
taminated A-l lipo-
protein
By comparison, 4 mg
of the protein was re-
covered from 34 L
of serum using 13
elution chromatog-
raphy and electro-
phoresis steps
238 Freitag

Alkaline phosphatase Column: HPLC
from E. coli DEAE-5PW
periplasm Displacer: CM-D
mAb from ascites Method: complex DC Complex DC was nec-
on anion exchanger essary because the
Complexer/displacer: mAB was cationic
CM-D Scale-up from 1 to 450
ml, final purity at
largest scale: 79%
Guinea pig serum Method: spacer dis- Displacement on me-
placement chroma- dium resolution ad-
tography sorbent
Column: Fractogel
DEAE 650s
Flow rate: 10 ml/h
Spacerldisplacer: het-
erogeneous mixture
of CM-D
Mouse liver cytosol Method: spacer dis- Displacement on low-
proteins placement chroma- resolution micro-
tography granular cellulose
Column: DEAE cellu-
lose flow rate:
5 ml/h
Spacerldisplacer: mix-
ture of CM-D
Lactate dehydrogenase Method I: weak anion Investigation of scale-up
from beef heart IE-DC parameters (column
Column: Tris acyl dimensions, protein
DEAE, 91 X 10 mm load)
and 250 X 10 mm Comparison to EC
Displacer: Carboxy-
methyl starch
Method II: weak anion Comparison of dis-
IE-DC in connection placers
to affinity chromatog- Eudragits: readily avail-
raphy (Cibacron Blue) able, cheap, nontoxic
Column: Tris acyl
DEAE, 250 X 10 mm
Flow rate: 0.5 mllmin
Displacers: chondroitin
sulfate C, alginate,
and Eudragit L
and S
Industrial recombinant Method: anion IE-DC Application for final

human growth hor- polishing step
mone
Thrombolytic protein Method I: strong cation
from fermentation IE-DC
broth, containing al- Column: 88 X 5.0
bumin, insulin, mm, 8 pm
transperrin, aproti- Flow rate: 0.1 mllmin
nin, methotrexate, Displacer: DEAE-
and BSA dextran
Method II: ABx-DC ABx phases are
Column: 100 X 4.6 multimode antibody
and 50 X 4.6 mm, exchangers
5 Fm (Bakerbond) com-
Flow rate: 0.1 mllmin bining cation ex-
Displacer: DEAE- change, mild anion
dextran (10 KDa) exchange, and mild
hydrophobic interac-
tion sites on the sta-
tionary phase surface
The capacity of the ABx
phase is higher; better
resolution is achieved
with the cation ex-
changer
Recombinant human Method: hydroxyapa- Crossing isotherms
antithrombin 111 from tite-DC Final purity: 90%. recov-
CHO cell culture su- Column: 250 X 4 rnm, ery: 84%
pernatant hydroxyapatite, 2 pm, By comparison, a quanti-
100 nm tative separation was
Flow rate: 0.1 mllmin not possible in the elu-
Displacer: EGTA tion mode
Technical dairy whey Method: strong anion I ml whey containing all
IE-DC milk proteins save the
Column: BioScale-Q2, caseins was processed
52 X 7mm, 1 0 p m directly
Flow rate: 0.1 mllmin Feed: 3.45 gIL a-lactal-
Displacer: PAA (M, bumin, 12.65 glL P-
6000 glmol) lactoglobulin, other
UV-active com-
pounds
Yield: a-lac 78%, P-lg
92%
Concentration factor: 3
Freitag
B. Amino Acids and Peptides

Modem peptide displacement chromatography started in the early 1980s and is
closely connected to the group of Horvath and coworkers. Reversed-phase chro-
matography dominates that particular area of displacement separations. The sepa-
ration of the product of a peptide synthesis from its closely related byproducts
remains one of the typical applications.
Synthetic peptides Displacer: The peptides were both [73]

N-Benzoy1-L-arginyl-L- BEE ridded of impuri-
methioninamide Decyltrimethylammo- ties and concen-
N-Benzoyl-L-arginyl-L- nium bromide trated
methionyl-L- Decyltrimethylammo- Amino acids eluted
leucinamide nium bromide ahead of the dis-
L-Methionyl-L-leucyl-L- Cetryltrimethylammo- placement train
phenylalanineamide nium bromide In one case 5.2 g was
N-Benzoyl-L-arginyl-L- Method: direct RP-DC recovered from a ,
methionyl-L-leucyl- Column: analytical size, 500-ml feed in a
L-phenylalanin- various C-I 8 mate- single run
amideamide rials
Enzyme reactor: immo-
bilized carboxy-
peptidase Y
Synthetic peptide frag- Method: multidimen- Large-scale application 1741
ment correspond- sional, RP-DC and from 100 mg to 35
ing to fragment IE-DC g of Merrifield syn-
163-171 of h- Column, RP-DC: thesis-type product
interleucin-P LiChrosorb RP- 18, (purities >go%)
250 X 4 mm and No sample pretreatment
LiChroprep necessary
RP-18, 20, 40, 80
mm
Column, IE-DC: Mono-
Q 50 X 5 mm, I0
Fm and Q Sepha-
rose, 350 X 10
mm
(two-column
system)
Displacer, RP-DC: ben-
zyltrihutylammo-
nium chloride
Displacer, IE-DC: am-
monium citrate
Displacement Chromatography of Biomolecules 24 1
Synthetic peptide con- Method: RP-DC, 23°C Up to 50 mg was puri- 1741

taining two epi- Column: Aquapore RP- fied in a single ex-
topes of Plasmo- 18 250 X 4.6 mm, periment (purity
dium fakiparum 7 pm, 30 nm >95%)
circumporozoite Flow rate: 0.1 mllmin
protein (malaria Displacer: benzyldi-
vaccine) methy ldodecylam-
monium bromide
Synthetic peptides in- Method: RP-DC after Peptides produced by [75]
tended for Plasmo- l yophilization, solid phase syn-
dium vivax malaria gel filtration, and thesis
seroepidemiology IE-EC Advantages over elution
Column: Vytac 2 18TP5 chromatography
C-18, 250 X 4mm, demonstrated
5 pm, 30 nm Up to 107 mg of the
Flow rate: 0.1 mllmin crude mixture was
~ i s ~ l a c eBEE
r: processed (final
product purity:
85%)
Mellitin (honeybee Method: RP-DC, 40°C No intraparticular mass 1761
venom peptide)/ Column: Hy-Tach C- 18. transfer possible
synthetic variants 105 X 4.6 mm, 5 mg mellitin was iso-
2 pm, nonporous lated in 20 min
Flow rate: 0.2 mllmin from a 10-mg mix-
Displacer: benzyldi- ture
methylhexydecyl-
ammonium chlo-
ride
Crude a- and p-melano- Method: RP-DC, 22°C 30 mg was separated in [77]
cyte-stimulating Column: C-18 based on a single experiment
hormone mixtures Hypersil silica, 250
X 4.6 mm, 5 pm
Displacer: benzyldi-
methy ldodecylam-
monium bromide
Insulin (bovine, por- Method: RP-DC Semipreparative proto- [78]
cine) Column: Nucleosil C-8, col up to 500 mg
5 vm of raw insulin
Flow rate: 0.1 and 0.2 could be purified
mllmin (proinsulin level
Displacer: cetramide < 100 ppm)
242 Freitag

Synthetic luteinizing Method: RP-DC A displacement kit con- [12]
hormone-releas- Column: RPI DisKit taining column, car-
ing hormone column, 250 X 4.6 rier, displacer, and
(LHRH) mm, 5 pm regenerant was em-
Flow rate: 0.5 mllmin ployed (BioWest re-
Displacer: DisKit dis- search)
placer Goal was to keep the
peptide concen-
tration below the
critical value to pre-
vent aggregation
and precipitation
Na-9-fluorenoxy- Method: normal phase Scale up from 100 mg [79]
carbonyl-S- DC to 38 g
trity lcysteine Columns: LiChrocart
derivative (Fmoc- RP-18, 250 X 4
Cys-Tert-OH) mm, 10 pm (mg
scale) compressed
20-,40-,SO-mm
columns, LiChro-
prep Si-60 silica,
25-40 pm
(g scale)
Flow rates: 0.1 mllmin
(mg scale)
2 mllmin (g scale)
Displacer: benzyltri-
buty lammonium
chloride
C. Antibiotics

Commercial polymyxin Method: RP-DC Separation in the constit- [SO]
B sulfate Column: LiChrosorb uents
RP-8 C-8, 250 X 4.6 Product concentrations
mm, 5 pm between 10 and 20
Flow rate: 0.1 mllmin mglml were reached
Displacer: dodecyloctyl
ammonium chloride
Oligomyxins A, B, Method: RP-DC Highly hydrophobic sub- [81]

and C Column: Lichrosorb stances (carrier
RP- 18, 250 X 4.6 75% methanol in
mm, 5 ~m water)
Displacer: plamitic acid
Cephalosporin C from Method: RP-DC, 35OC 5 ml of culture supema- [82]
fermentation broth Column: Zorbax BP tant was processed
(2-18, 350 X 4.6 in 20 min
mm, 5 pm
Displacer: BEE
D. Isomers
Displacement chromatography has been repeatedly used by Vigh et al. to separate
optical and structural isomers at high throughputs and concentrations [83]. A
series of homologous displacers of varied affinity for Cyclobond I1 columns (a-
cyclodextrin silica) was introduced by the same group in 1995 [84].
Isobufen Displacer: 4-tert-butyl- Quantitative separation 1851

cyclohexanone even for a values of
Flow rate: 0.2 mllmin I .08
Puritieslyields were sim-
ilar to elution for the
less retained isomer,
better for the more
retained one
5,lO-Dideazatetra- Displacer: cetramide [861
hydrofolic acid Flow rate: 0.3 mllmin
Method: normal phase
and RP-DC, 4OC
Column: P-cyclodextrin-
silica (Cyclobond I),
250 X 2 mm, 2 in
series
244 Freitag
Enantiomers of 1,2-0- Method: normal phase Separation of a 20-mg [87]

dihexadecyl- DC, 15°C sample
rac-glycerol-3-0- Column: Pirkle-type
(3,5-dinitrophenyl) naphthylalanine sil-
carbamate ica, 250 X 4 mm,
5 pm
Displacer: 3,5-dinitro-
benzoyl ester of
n-heptanol
D.I.-Methinonine P- Method: enantioselec-
naphthylamide tive DC
Column: poly-L-valyl
groups as chiral se-
lectors on modified
silica. 250 X 4.6
mm
Displacer: D,I-mandelic
acid
E. Displacement Chromatography for On-Line Product

Removal Schemes
Reversible reactions require the removal of the product for high conversions. At
least two cases can be found in the literature whereby displacement chromatogra-
phy was used for this purpose. One concerns the preparation of a dipeptide from
the respective amino acids [89]; the other the synthesis of a nucleotide [90].
N-Benzoyl-L-arginyl-L- Method: tandem RP- The enzyme reactor was [89]

methionin amide DC, 50°C, 2 columns operated in the recir-
were alternated culation mode (recy-
Column: C-18, 250 X cling of L-methi-
4.6 mm, 10 pm oninamide)
Displacer: BEE 460 mg of product (pu-
Flow rate: 0.1 mllmin rity >99%) in 24 h
Packed bed enzyme re-
actor (immobilized
carboxypeptidase Y)
in tandem with dis-
placement column

Nucleic acid fragments Method: RP-frontal The enzyme reactor was [90]
(GPU) chromatography, RP- operated in the recir-
Separation from the DC and enzyme reac- culation mode
educt (cyclic GMP) tor in series 100 mg of GpU (purity
and large excess uri- Column: Zorbax C-18, 99.7%) in 2.4 h
dine 250 X 4.6 mm, 5 pn
Displacer: n-butanol
Packed bed enzyme re-
actor (immobilized ri-
bonuclease T I )
F. Displacement Chromatography for Sample Preparation

in Analytical Chemistry
Many analytical procedures in biochemistry, molecular biology, and related fields
involve the separation of a complex mixture, e.g., a peptide digest, prior to a
closer analysis of the individual components of the mixture or at least the re-
sulting less complex mixtures. The displacement process, which will focus even
trace components into highly concentrated zones while enriching all mixture
components to a high extent, is a prime choice for such a sample pretreatment
step. Trace components, which may be difficult to isolate by conventional chro-
matographic methods, can be obtained in sufficient amounts to allow chemical
characterization by established techniques. Frenz et al. used such a hyphenated
system to analyze the components of a tryptic digest of a recombinant growth
hormone [91,92]. Several of the collected fractions showed a completely different
spectrum from those seen in the preceding or following fractions of the major
peptides of the digest. Presumably, these fractions represent peptides fragments
of incorrectly expressed growth hormone molecules, whose detection would 0th-
envise have been difficult. A microsystem compatible to flow rates in the micro-
liter per minute range has been suggested for direct liquid chromatography-mass
spectrometry coupling [93].
246 Freitag

P-Naphthylamine con- Method: RP-DC Due to the focusing ef-
taining an impurity Column: Adsorbo- fect and the possibil-
at the ppm level sphere HSC-18, 25 ity of using a larger
X 4.6 mm sample size, the de-
Flow rate: 1 mllmin tection limit for the
Displacer: diethyl- trace compound
phthalate could be lowered by
three orders of mag-
nitude in DC com-
pared to EC
Recombinant growth Trace components
hormone (industrial (>0.1%) could be
product, BioWest) characterized by MS
containing low-level
impurities
Tryptic digest of recom- Method: RP-DC Displacement used in
binant growth hor- Column: Nucleosil connection to FAB-
mone C-18, 150 X 4.6 MS and ESI-MS for
mm (2 in series) the analysis of tryp-
Flow rate: 1 mllmin tic digests
Displacer: cetramide The exploitable column
capacity was up to
two orders of magni-
tude higher in DC
than in EC
Genetic variants of Method: anion IE-DC Displacement used in
P-lactoglobulin Column: Protein-Pak connection with low-
Q-8HR, 100 X 5 angle laser light scat-
mm tering photometer
Flow rate: 0.1 mllmin for on-line determi-
Displacer: 75-kDa dex- nation of molecular
tran sulfate mass of proteins
G. Thin-Layer Displacement Chromatography

The displacement mode has also been used in thin-layer and forced-flow thin-
layer displacement chromatography (TL-DC and FF-TL-DC, respectively) by the
group of Kalasz and coworkers [96,97]. The use of spacers is necessary in this
case to separate the substance spots.

Parent drug and radioly- Method: TL-DC Spacer TL-DC and car- [98]
sis products from rat Plates: silica gel 6OF2,,, rier spacer TL-DC
urine PSC silica, TLC a h - were compared
mina coated on 200 Various spacer1
X 200 mm glass displacer mixtures
plates were used
Deprenyl and metabo- Method 2D TLC (con- Analytical application [99]
lites from rat urine ventional TLC fol- for the identification
lowed by TL-DC) of metabolites
Displacer: triethanol-
amine
Spacer: Sudan black
dye mixture
Plant ecdysteroids Method: TL-DC and Pilot system for de- [ 1001
2D TLC termining optimum
Plates: silica gel conditions for
60GFm TL-DC
Displacer: triethanol-
amine, dimethylami-
nopropylamine
H. Miscellaneous
Nucleotides Method: RP-DC [loll

Displacer: butanol
Nucleosides Method: RP-DC Same
Displacer: benzyltribu-
tylammonium chlo-
ride
Methyl esters of polyun- Method: RP-DC, 30°C Difficult carrier selec- [lo21
saturated fatty acids Column: C-18 silica tion: low solubility
(EPA, DHA) Flow rate: 0.5 mllmin of polar compounds
Displacer: oleic acid desired during bind-
Carrier 1: acetonitrilel ing, high solubility
water 80: 20 for high displacer
Camer 2: acetonitrilel concentration during
water 90: I0 separation, solution:
two-carrier system
Final purities >90%,
concentration fac-
tors: 4- 13
248 Freitag

Oligonucleotides Method: anion IE-DC Large-scale approach [49]
Column: 10 in. X 10
cm
Flow rate: 2.1 L/min
(250 cm/h)
Displacer: dextran sul-
fate
VII. CONCLUSIONS
Preparative chromatography remains the method of choice for the separation of

many typical complex mixtures encountered in the life sciences. In many situa-
tions (trace analysis, diluted feeds, and high-throughputs among them), the dis-
placement approach may in fact be the most practical, most economical, and
easiest to realize. The goal of this chapter was to serve as an introductio'n to the
theoretical and practical aspects of displacement chromatography from theory
through method development and applications.
VIII. APPENDIX
A. Definitions
ldeal chromatography ldeal chromatography assumes infinite column effi-
ciency, i.e., all possible band broadening effects (mass transfer, reaction kinetics,
axial dispersion) are neglected. The two phases are constantly at equilibrium.
Models based on ideal chromatography allow, for example, study of the influence
of equilibrium thermodynamics alone on a chromatographic separation.
Linear chromatography Linear chromatography is characterized by a lin-
ear isotherm, i.e., the equilibrium concentration of a component in the mobile
phase and the amount adsorbed onto the stationary phase are directly proportional
at all times. The mass transfer processes are also linear. No heat or sorption
effects occur. Parameters such as the viscosity of the fluid phase are independent
of the composition. The mass flow rate is also constant. The sample concentration
has no influence on the peak shape (Gaussian) or the retention time. Both the peak
area and the peak height are proportional to the sample concentration. Analytical
chromatography usually approximates linear conditions.
Nonideal chromatography Nonideal chromatography allows for band
broadening effects, such as axial dispersion, mass transfer and reaction kinetics,
molecular and eddy diffusion. Increasingly complex models have been developed
that take into account some or all of these effects.
Nonlinear chromatography Nonlinear chromatography takes place in the

nonlinear range of the equilibrium isotherm. The ratio of adsorbed to free mole-
cules of a given type thus depends on the mobile phase concentration of the
substance itself, but also on that of all other compounds present at the time (cou-
pling of the respective mass balances via the multicomponent equilibrium iso-
therms). Peak shape and retention time depend on the sample composition.
B. Symbols
substance-specific constant (Langmuir isotherm)
substance-specific constant (Langmuir isotherm)
mobile phase concentration, displacer
mobile phase concentration
diffusion coefficient
bulk axial dispersion coefficient (equilibrium-dispersive model)
particle diameter
height equivalent to a theoretical plate (HETP)
Step function
equilibrium constant (KD = kb/(l + bucD)
capacity factor (t, - to)/to
film mass transfer coefficient
column length
apparent plate number of a given column (N = HIL)
stationary phase concentration, displacer
stationary phase concentration
velocity of the displacer front (shock layer)
linear flow velocity (mobile phase)
time
retention time of a solute (usually approximated as the retention time
of the peak maximum)
retention time of an inert tracer (column holdup time)
dimensionless column length
shock layer thickness
separation factor (k;/kj)
phase ratio
characteristic shock layer parameter (see Fig. 12 for details)
duration of feed introduction
C. Abbreviations
Abx antibody exchange
BEE 2-(2-butoxyethoxy)ethanol
1
250 Freitag
BSA bovine serum albumin

CM-D carboxymethyldextran
DC displacement chromatography
EC elution chromatography
EGTA ethylenglycolbis(~-aminoethylether)-N,N,,-tetraaceticacid
ESI electrospray ionization
FAB fast atom bombardment
HETP height equivalent to a theoretical plate
HIC hydrophobic interaction chromatography
HPDC high-performance displacement chromatography
HPLC high-performance liquid chromatography
IDA imminodiacetic acid
IE-DC ion exchange displacement chromatography
IMAC immobilized metal affinity chromatography
LCST lower critical solution temperature
LHRH luteinizing hormone-releasing hormone
m Ab monoclonal antibody
MS mass spectrometry
PAA polyacrylic acid
PEI polyethyleneimine
RP-DC reversed-phase displacement chromatography
SMA steric mass action
TLC thin-layer chromatography
TL-DC thin-layer displacement chromatography
2D two-dimensional
uv ultraviolet
REFERENCES
Antia F, Horvath C. Ber Bunsenges Phys Chem 1989;93:961-968.

Cramer SM, Subramanian G. Sep Purif Meth 1990;l9(l):31-91.
de Carli JP, Carta G, Byers Ch H. AlChE J 1990;36(8):1220-1228.
Gliickauf E. Proc R Soc London A 1946;186:35.
Helfferich F, Klein G. Multicomponent Chromatography. New York: Marcel Dek-
ker, 1970.
Rhee H.-K., Ark R, Amudson NR. AlChE J. 1982;28:423.
Bellot JC, Condoret JS. J Chromatogr 1993;657:305-326.
Phillips MW, Subramanian G, Cramer SM. J Chromatogr l988;454: 1-21.
Gu T, Tsai G-J, Tsao GT. Adv Biochem Eng 1993;49:45-71.
Katti AM, Guiochon G. J Chromatogr 1988;449:25-40.
DeVault D. J Am Chem Soc 1943;65:532.
r
Jakobson J. In: Horvath C, Ettre LS, eds. Chromatography in Biotechnology, ACS

Symposium Series, Vol. 529, 1993:77-84.
Freitag R, Vogt S. Cytotechnology, accepted for publication.
Helfferich F, James DB. J Chromatogr 1970;46:1-28.
Rhee H-K, Aris R, Amundson NR. First Order Partial Differential Equations 11.
Theory and Applications of Hyperbolic Systems of Quasilinear Equations. Engle-
wood Cliffs, NJ: Prentice-Hall, 1989.
Frenz J, Horvath C. AlChE J 1985;31(3):400-409.
Brooks CA, Cramer SM. AlChE J 1992;38:1969-1978.
Velayudhan A, Horvath C. J Chromatogr 1988;443:13-29.
Kopaciewicz W, Rounds MA, Fausnaugh J, Regnier FE. J Chromatogr 1983;266:3.
Kim YJ. Bioseparation 1995;5:295.
van Deemter J, Zuiderweg F, Klinkenberg A. Chem Eng Sci 1956;5:271.
Knox H, Saleem M. J Chromatogr Sci 1972;10:80.
Guiochon G, Golshan-Shirazi S, Katti A. Fundamentals of Preparative and Nonlin-
ear Chromatography. San Diego: Academic Press, 1994.
Golshan-Shirazi S, Guiochon G. J Chromatogr 1992;603:1- 11.
Lapidus L, Amudson NR. J Phys Chem 1952;56:984.
Thomas HC. J Am hem Soc 1944;66:1664.
Goldstein S. Proc R Soc London l953;A219: 151.
Wade JL, Bergold AF, Cam PW. Anal Chem 1987;59:1286.
Gliickauf E, Coates JI. J Hem Soc. 1947;1315.
Hiester NK, Vermeulen T. Chem Eng Progr 1952;48:505.
Lin B, Golshan-Shirazi S, Gulochon G. J Phys Chem 1989;93:3363.
Rasmuson A, Neretnieks I. AlChE J 1980;26:686.
Rasmuson A. Chem Eng Sci 1982;37:787.
Zhu J, Guiochon G. J Chromatogr 1994;659:15-25.
Zhu J, Ma Z, Giochon G. Biotechnol Prog 1993;9:421.
Liao A, Horvath C. Ann NY Acad Sci 1988;589:182-191.
Zhu J, Katti A, Guiochon G. Anal Chem. 1991;63:2183-2188.
Antia FA, Horvath CG. J Chromatogr 1991;556:119-143.
Frey DD. J Chromatogr 1987;409:1- 13.
Lin B, Ma Z, Golshan-Shirazi S, Gulochon G. J Chromatogr 1989;475:1 - 1 1.
Carta G, Dinerman AA. AlChE J 1994;40(10):1618-1628.
Kim YJ, Cramer SM. J Chromatogr 1991;549:89-99.
Kasper C, Breier J, Vogt S, Freitag R. Bioseparation 1996;6:247-262.
Freitag R, Breier J. In: Pyle DL eds. Separations for Biotechnology, published by
the Royal Society of Chemistry, Thomas Graham House, Science Park, Cambridge,
UK, 1994:166-172.
Subramanian G, Phillips MW, Jayaraman G, Cramer SM. J Chromatogr 1989;484:
225-236.
Felinger A, Guiochon G. J Chromatogr. 1992;609:35-47.
Gerstner JA, Moms J, Hunt T, Hamilton R, Afeyan NB. J Chromatogr 1995;695:
195-204.
Subramanian G, Cramer SM. Biotechnol Prog 1989;5(3):92-97.
Gerstner JA. BioPharm 1996;9:30.
1
Freitag
Freitag R, Horvath C. Adv. Biochem Eng Biotechnol 1995;53:17-59.

Peterson EA. Anal Biochem 1978;90:767-784.
Antia F, Fellegvari I, Horvath C. Ind Eng Chem Res 1995;34:2796-2804.
Gerstner JA, Cramer S. Biotechnol Prog 1992;8:540-545.
Gerstner JA, Cramer S. BioPharm 1992;Nov/Dec:42-45.
Freitag R. Breler J. J Chromatogr 1995;691:101.
Gadam SU, Cramer SM. Chromatographia 1994;39(7/8):409-418.
Jayaraman G, Li Y-F, Moore JA, Cramer SM. J Chromatogr 1995;702: 143- 155.
Jen SCD, Pinto NG. J Chromatogr Sci 1991;29:478-484.
Jen SCD, Pinto NG. J Chromatogr 1990;519:87-98.
Kasper C. Breler J, Vogt S, Freitag R. Bioseparation 1996;6:247-262.
Vogt S, Freitag R. J Chromatogr l997;76O: 125-137.
Torres AR, Peterson EA. J Chromatogr 1992;604:39-46.
Patrickios CS, Gadam SD, Cramer SM, Hertler WR, Hatton TA. Biotechnol Prog
1995;l 1:33-38.
Vogt S, Freitag R. (submitted).
Tores AR, Peterson EA. J Chromatogr 1990;499:47-54.
Peterson EA. Anal Biochem 1978;90:767-784.
Torres AR, Krueger GG, Peterson EA. Anal Biochern 1985;144:469-476.
Ghose S, Mattiasson B. J Chromatogr 1991;547: 145- 153.
Ghose S, Mattiasson B. Biotechnol Tech 1993;7(8):615-620.
Frenz J. LC/GC Int l992;5(12): 18-21,
Kim, YJ. Biotechnol Tech 1995;9(6):417-422.
Kim, YJ. Biotechnol Tech 1994;8(7):457-462.
Cramer SM, Horvath C. Prep Chromatogr 1988;1(1):29-49.
Viscome GC, Cardinali C, Longobardi G, Verdini AS. J Chromatogr 1991549:
175-184.
Bianchi E, Del Guidice G, Verdini AS. Pessi A. Int J Peptide Protein Res 1991;
37:7-13.
Kalghatgi K, Fellegvari I, Horvath C. J Chromatogr 1992;604:47-53.
Viscomi GC, Lande S, Horvat C. J Chromatogr 1988;440:157-164.
Vigh G, Varga-Puchony Z, Szepesi G, Gadzag M. J Chromatogr 1987;386:353-
362.
Cardinalli F, Ziggiotti A, Viscorni GC. J Chromatogr 1990;499:37-45.
Kalasz H, Howath C. J Chrornatogr 1981;215:295-302.
Valko K. Slegel P, Bati J. J Chromatogr 1987;386:345-351.
Subramanian G, Phillips MW, Crarner SM. J Chromatogr 1988;439:341-351.
Vigh G, Quintero G. Farkas G. In: Horvath C, Nikelly JG eds. Washington, DC:
American Chemical Society (ACS) Symposium Series, Vol. 434, 1990:181-197.
Quintero G, Vo M. Farkas G. Vigh G. J Chromatogr 1995;693:1-5.
Farkas G, Irgens LH, Quintero G. Beeson M, Al-Saeed A, Vigh G. J Chromatogr '
1993;645:67-74.
Irgens LH, Farkas G, Vigh G. J Chrornatogr 1994;666:603-609.
Carnacho-Toralba PL, Beeson MD, Vigh G. J Chrornatogr 1993:646:259-266.
Sinibaldi M, Castellani L, Federici F, Messina A. J Liquid Chromatogr 1993;
16(14):2977-2992.
7
Cramer SM, el Rassi Z, Horvath C. J Chromatogr 1987;394:305-314.

El Rassi Z, Horvath C. J Chromatogr 1983;266:319-340.
Frenz J, Quan CP, Hancock WS, Bourell J. J Chromatogr 1991;557:289-305.
Frenz J, Bourell J. Hancock WS. J Chromatogr 19905 l2:299-314.
Vigh G, Irgens LH, Farkas G. J Chromatogr 1990;502: 1 1 - 19.
Ramsey R. Katti AM, Guiochon G. Anal Chem 1990;62:2557-2565.
Mhatre R, Qian R, Krull IS, Gadam S, Cramer S. Chromatographia 1994;38(5/6):
349-354.
Kalasz H, Nagy J, Bathori M. J Planar Chromatogr 1989;2:39-43.
Kalasz H, Kerecsen L, Nagy J. J Chromatogr 1984;316:95-104.
Kalarz H, Bathori M, Matkovics B. J Chromatogr 1990;520:287-293.
Kalasz H. J High Resol Chromatogr & Chromatogr Commun 1983:6:49-50.
Kalasz H, Bathori M, Kerecsen L, Toth L. J Planar Chromatogr 1993:6:38-42.
Horvath C. Frenz J, el Rassi Z. J Chromatogr 1983;255:273.
Huang S-Y, Jin J-D. Bioseparation 1994;4:343-35 1.
High-Performance Membrane
Chromatography of Proteins
Tatiana Tennikova
Russian Academy of Sciences, St. Petersburg, Russia
Ruth Freitag
ETH Lausanne, Lausanne, Switzerland
I. INTRODUCTION
According to the International Union of Pure and Applied Chemists (IUPAC)

recommendations [I], "chromatography is a physical method of separation in
which the components to be separated are distributed between two phases, one
of which is stationary (stationary phase) while the other (the mobile phase) moves
in a definite direction." Today we also have chromatographic methods whereby
the two phases are in countercurrent motion, e.g., the simulated moving bed, the
continuous annular chromatography, etc. All chromatographic methods have one
thing in common and that is the dynamic separation of a substance mixture in a
flow system. Usually the chromatographic separation process takes place in a
column packed with the particles of the sorbent (stationarylsolid phase). Modem
high-performance liquid chromatography (HPLC) uses an extensive arsenal of
stationary separative phases that can effectively separate a wide variety of mix-
tures both of small molecules and of polymeric substances, including most impor-
tantly biologicals and biopolymers. According to the modem HPLC theory, the
sorbent particles can be regarded as rigid spheres of micrometer size that differ
considerably in porosity, i.e., ranging from totally nonporous to flow-through
(perfusive) ones.
256 Tennikova and Freitag
The adsorptive or interactive forms of HPLC are clearly the most important
ones for users in biotechnology and related fields. In these cases an adsorption-
desorption respectively a partitioning reaction takes place between the substances
to be separated and the sorbent (stationary phase). These interactions can be of
an electrostatic, hydrophobic, but also highly specific (biological complementary
type) character, as indicated by the respective names, such as ion exchange, re-
versed-phase, hydrophobic, pseudoaffinity, and affinity chromatography.
As mentioned above, HPLC is a most important tool for the analytical and
preparative separations of biomolecules, in particular proteins. The complex and
peculiar physicochemical character of these biopolymers has been a constant in-
centive to the development and deeper understanding of the methods that are
used for their analysis and isolation. t
II. "ONE-STEP" DESORPTION AND "MULTIPLE-STEP"

INTERACTION PROCESS IN CHROMATOGRAPHIC
SEPARATIONS
The chromatographic column packed with small particles of porous sorbent is

the main locus and thus a most important aspect of the HPLC separation processes
under discussion. The molecules of the substances to be separated are dissolved
in the mobile phase. In the case of conventional porous particles these molecules
are flushed with the mobile phase through the interparticle space of the column
and migrate by molecular diffusion through the stagnant fluid layer around each
particle to the outer surface of the sorbent bead, followed by penetration into
the porous bead by the same mechanism and, finally, adsorption on the inner sur-
face of the pores, where more than 90% of the total surface is found (Fig. 1).
The act of desorption, when molecules adsorbed on the sorbent surface are
displaced again by competing molecules from mobile phase (the modifier), is
really decisive for the separation. as differences in the desorption reaction rate
constants are larger than those of the adsorption reaction rate. It must be well
understood that even in a fairly short column the adsorption-desorption process
takes place many times and the effect on the separation of the substances is
"accumulated" as the substance molecules move from the top to the bottom of
the column.
In the interactive types of chromatography the substances can be separated
using two modes of elution: isocratic elution whereby the modifier (a competitor
for the adsorptive sites on the sorbent surface) is added to the eluent at constant
concentration, and gradient elution whereby the modifier concentration in the
mobile phase is increased during the separation either gradually or stepwise. Gra-
HPMC of Proteins
Bulk mobile Fluid layer Particle

phase
0 : Protein molecule
Figure 1 Stages involved in the trensrer of an rtnalyte molecule from the bulk of thc
mobile phase to the sorbent surface, where the adsorptionldesorption reaction takes place,
and back into the bulk mobile phase. ( I ) Diffusion from the bulk mobile phase through
!he stagnant film around the particle to the outer particle surface; (2) molec~~lar
diffusion
into the pore; (3) adsorption; (4) desorption; (5) molecular diffusion Crorn thc inner pore
surface through the pore space to the outer surface (pore mouth); (6) diffusion through
the stagnant film into the bulk mobile phase.
dient elution is more common in PI-otein chromatography than isocratic elution

because proteins tend to vary widely in adsorption energy. Thus it is often impos-
sible to find a single mobile phase composition that will desorb them all.
Since proteins differ widely in adsorption energy due to pronounced differ-
ences in the desorptjon rate constant. Theoretically i t is heref fore possible to
separate different substances in a single desorption step. It is therefore reasonable
to suggest that the separation of some definite group of substances (e.g., proteins)
is possible without using the whole length of column. i s . , repeated adsorption-
desorption events. This theoretical possibility has recently found a practical appli-
cation in a new chromatographic method, i.e., membrane chromatography. This
type of separation technique employs what is sometimes called a "membrane

adsorber" as stationary phase. This term comes from membranelfiltration tech-
nology and describes the geometry of the stationary phase in question, an interac-
tive, i.e., adsorptive, "filter" membrane. Various types of filtration operation
have been adapted to separation processes simply by using an interactive instead
of an inert membrane, including hollow fibers, cross-flow units, and others. These
methods have led to some interesting applications, some of which are included
in Section VI of this chapter. However, especially in the case of protein chroma-
tography, the possibility of high-performance membrane chromatography
(HPMC) also exists. The theoretical model of this exciting new method has by
now been developed and will be discussed in the following two sections. HPMC
is carried out on a certain well-defined type of separation media, the so-called
(convective interactive media) CIM discs, i.e., monolithic macroporous discs with
a well-defined and homogeneous pore structure design. Such monolithic layers
allow the construction of a rigorous theoretical model of the dynamic interphase
distribution of the substances during separation.
Formally, HPMC still falls into the IUPAC definition (see above) of chro-
matography as a dynamic separation of substances, if we accept the exchange of
a flat macroporous adsorptive layer for a complicated and expensive system Like
a shallow packed column. The interaction mechanism can still be the same, i.e.,
electrostatic interaction between charged units on the adsorber surface and oppo-
sitely charged groups on the protein surface. From the geometrical design of the
support one speaks correctly of a "membrane" (or a disc). From a chromato-
graphic point of view, this is still only a sorbent acting in the same way as the
packed HPLC column of the conventional systems.
A major difference in the theoretical treatment of disc versus column chro-
matography remains and that is the small length of the "separation distance,"
i.e., the thickness of the membrane, which a priori does not make any "multiple"
or "repetitive" interaction likely. To gain a very clear picture of what is going
on, one should regard the separation as the result of a one-step desorption process.
This does not mean that the adsorption step can be neglected in membrane
chromatography. On the contrary, since membrane chromatography is a single-
step process, the initial adsorption and its energy distribution has much more
consequence in membrane than in conventional chromatography, where any
distribution will be averaged out over the multitude of repeated adsorption-
desorption steps taking place along the column length (along the path of molecu-
lar migration). Membrane chromatography represents the unique possibility to
see a "momentary" picture of the molecules' distribution after the initial ad-
sorption. It is obvious that the theoretical treatment, as well as the practical
applications of this separation process, should take this very important feature
into consideration.
HPMC of Proteins
Ill. THEORETICAL MODELS OF INTERACTIVE

CHROMATOGRAPHIC SEPARATION
Membrane chromatography is a method based on the selective separation of ad-

sorbed substances as a consequence of the desorption process [2]. Taking into
account the huge amount of practical and theoretical research concerning models
of the interactive types of HPLC, a brief analysis of the major points will be
useful to understand the new chromatographic process.
A. Gradient and lsocratic Elution Chromatography

Cited most often are the papers of Snyder et al. [3-51, which can be considered
as the most fundamental. These papers are the result of a detailed investigation
of the process of gradient and isocratic HPLC of both small and large molecules.
Since we are mainly interested in the chromatography of proteins, the most im-
portant understanding to be drawn from these papers is that of the role played by
the mobile phase modifier,during the desorption of the protein molecules during
gradient elution. Snyder introduces a definition of the capacity or retention factor
kt, which in this case will correspond to a value averaged over the entire column
length. Parameter k' is largely dependent on the act of desorption [5]. The model
is based on the idea of a "multiple plate" [6] (or "multiple-step") adsorption-
desorption process and applies to both the isocratic and the gradient mode of
elution.
From our point of view, the most interesting aspect of this theory is the
conclusion that the number of parameters responsible for the effective resolution
of a protein mixture does not depend on column length. In the past this has served
as a rationale for the development of thin adsorptive porous layers as a stationary
phase for protein chromatography.
B. Perfusion Chromatography and Other Modes

of Chromatographic Separations with Putatively
Improved Mass Transfer
Conventional packed columns require high pressure to drive the mobile phase
through the packed bed at a sufficiently high linear flow rate. The high linear
flow rates necessary to achieve separations within a practical period of time entail
also significant diffusional limitations, particularly when proteins, which have a
much lower diffusivity than small molecules, are concerned. Both problems were
addressed and partially solved with the development of the so-called perfusion
columns [7-91.
Perfusion columns are packed with sorbent particles having bidisperse po-
rous structure. The big pores with a size comparable to the size of the channels
formed by the interparticle space allow mass transport into the particle by convec-
tion rather than diffusion. The channels "divide" the particle into smaller seg-
ments, which is an important factor for the decrease of the distance to be covered
by diffusion by a molecule on its way to the adsorptive surface. Thus the mass
transfer is much more rapid in perfusion than in conventional porous supports.
The second type of pores, which also exists in the perfusion particles, consists of
non-flow-through channels of much smaller diameter and was called "diffusion"
pores by the inventors. It is implied that the majority of the adsorptive surface
is located in the diffusion pores and that the separation takes place there as well,
in accordance with the adsorption-desorption mechanism. Thus, the major ad-
vantage of perfusion chromatography is the improved mass transfer of the sub-
stance from the mobile phase to the stationary one [lo-141. The enhancement
is due to the fact that the solubilized molecules are transferred through the station-
ary phase particle (site of adsorption-desorption) by convection rather than by
molecular diffusion.
Membrane chromatography is often linked or compared to perfusion chro-
matography because some superficial resemblance exists. In membrane chroma-
tography the mobile phase flows through the sorbent layer or, more correctly,
through the single "particle," just as in perfusion chromatography. How-
ever, these two methods cannot be described with the same mathematical equa-
tions. The main difference is that in membrane chromatography the adsorption-
desorption process takes place on the walls of flow-through channels under the
conditions (pressure, shear stress, etc.) defined by the flow rate of the mobile
phase. There are no diffusive pores. Rather than as a column, the membrane can
be imagined as a bundle of capillaries of different diameters but of the same
extremely small length. In membrane chromatography the step of intraparticle
diffusion is absent. So there is no necessity of considering any migration into
the particle by molecular diffusion followed by the adsorption on the inner parti-
cle surface. From the many equations describing diffusional effects in conven-
tional HPLC, only a single one remains relevant in membrane chromatography.
This is the diffusional migration through the boundary layer of liquid close to
the walls of pores. Another important difference between the theory of membrane
chromatography and that of perfusion chromatography is that the desorption act
itself is not considered important in the latter and that perfusion chromatography,
just as conventional chromatography, is modeled by assuming repetition (and
accordingly, averaging of the parameters) of the adsorption-desorption process
as the separation evolves along the column [12-141.
A pronounced improvement of the mass transport kinetics was also ob-
served in another type of HPLC column where totally nonporous particles with
bonded adsorptive phase on their surface were used [15,16]. This type of station-
HPMC of Proteins 261
ary phase bears a strong resemblance to chromatographic discs with a controlled

porous structure [17-251. Adsorption occurs at the outer surface of the particles
under the direct influence of the mobile phase, which flows through micrometer-
sized interparticle channels that the particles form between them. However, in
terms of column geometry, the governing principle is again a multistep interac-
tion between the substance and the sorbent surface. Besides that, monosized
beads of a certain diameter are necessary to ensure that the size of the flow-
through channels is large enough for a practically maintainable pressure-flow
curve. This usually means a reduction of the surface area by several orders of
magnitude when nonporous rather than porous beds are used. In adsorption and
especially so in preparative adsorption chromatography this quickly becomes a
handicap.
The monolithic rod columns also show improved mass transfer kinetics
[26-381. Such sorbents are formed by the in situ polymerization of a continuous
macroporous polymer network inside a tube. As in the case of the columns packed
with nonporous particles (or indeed of chromatographic discs), the various types
of diffusional resistance to the mass transport are eliminated in the continuous
bed columns. The separation process is characterized by quick kinetics and high
resolution of peaks; it again remains a multistep process.
It is interesting to note that while the polymer structure of the monolithic
porous rod columns and the monolithic porous discs may be very similar, the
process of separation is different in these cases simply as a result of the difference
in geometry. The long "rod," just like any other column, allows for a repetition
of the adsorption-desorption act. As a result, the chromatographic peaks reflect
the average of the various band broadening effects and the peaks will always be
narrower than those obtained in an analogous separations using discs. Moreover,
the multiple character of the adsorption-desorption event aids high-speed separa-
tions of small molecules using the rod columns [32]. In counterdistinction, a
separation of the small molecules is much more difficult in disc chromatography,
simply because in this case the separation relys mainly on the initial differences
in the energy of interaction of the molecules with the surface ligands. For small
molecules this difference is normally much smaller.
C. The Stoichiometric Retention Model

In ion exchange protein chromatography it is common to express the experimen-
tally determined parameter k' (capacity or retention factor) as a function of the
charged modifier concentration in the mobile phase. Generally, the relationship
can be reduced to an equation of the following form:
where f(Q) can be presented as polynom. In the most cases, it is possible to

approximate f(0) by a linear relationship that yields an equation like
For practical reasons, the logarithmic form of this equation is normally used:
where M and S are constants defined by the nature of the protein and Q is a
variable value representing the modifier concentration in the eluent.
A very thorough experimental and theoretical investigation of this relation
was camed out by Regnier and coworkers [39-411. In these papers, a theoretical
retention model is constructed by applying the law of mass action to describe the
various interactions in the system, e.g., competition between protein and modifier,
interaction between protein and sorbent or sorbent and modifier. Unfortunately,
this model cannot be extended to the nonlinear range of the adsorption isotherm,
i.e., the region where most preparative chromatography occurs.
Upon analysis of the various equilibria and writing the corresponding equa-
tions, the authors come to the following expression which relates the retention
factor to the concentration of the desorbing agent (modifier):
where N is a constant, y~ is the concentration of the modifier [molIL], and Z is

a stoichiometric parameter related to the number of moles of modifier needed
for the solvation of the adsorptive sites on the protein, and also to the stationary
phase adsorption of the modifier. The parameter Z is a cornerstone in the theory
of the separation of proteins by interactive chromatography. The value of k' may
change from 0 (at v + -) to infinity (at y~ + 0) in this system.
Equation (4) bears a certain mathematical similarity to Eq. (2). They be-
come identical near the desorbing concentration of the modifier. It may be imtat-
ing to realize that the dimensionless factor k' is expressed as a function of two
values, N and y~ (Z is an numerical parameter), which bear dimensions. However,
the logarithmic form of (4) has been used successfully in praxis for some time
now and an excellent correlation between the experimental results and the pre-
dicted values are normally found for both isocratic reversed-phase and ion ex-
change chromatography of proteins:
D. Retention Model for "Short Separation Layer"

Chromatography
A theoretical model of protein retention (and separation) in disc chromatography
should without fail take into account equations for the discussed above equilibria.
Since the separation of the substances according to our idea occurs by a one-
step act of adsorption-desorption (more exactly, the single step of desorption),
the initial distribution of the protein molecules at the adsorptive surface located
at the inner walls of the disc pores (channels) is most important. This distribution
is governed by the energy of interaction, which in turn is characterized by the
Z parameter. The desorbed molecules instantaneously leave their "landing
grounds" and are flushed from the disc by the flow of the mobile phase. In this
situation the Z parameter is not averaged as in the case of multiple adsorption-
desorption acts in a column.
For the chromatography over short distances (discs) it is important to exam-
ine the parameter as a function of time and, equivalently, space. The former
corresponds to a separation on a very short column with the efficiency of "one
theoretical plate" [42], while the latter acknowledges the existing of some defi-
nite "length of separation layer" [43,44]-terns that take into account the non-
stationary nature of the initial part of the process of gradient elution in a column.
It is useful to keep these terms in mind when choosing the gradient shape and
comparing it with the real length of separation layer, i.e., the thickness of
the actual disc. Unfortunately, it is not possible to give a full summary of the
published data, which could be interesting for the theory of HPMC. The inter-
ested reader is referred to some of the, in our opinion, most significant papers
[45-491.
IV. PRINCIPLES OF PROTEIN SEPARATION

IN MEMBRANE CHROMATOGRAPHY
A. Peculiarities of Proteins as Objects of Separation
Processes Governed by Mass Transfer Effects
The topic of this chapter is the separation of soluble proteins by HPMC. It is
therefore useful to recall some of the main physical parameters governing the
behavior of these globular structures [50]. The features to be discussed differ
significantly between proteins and smaller molecules when chromatography is
concerned. They influence (1) the kinetics of the adsorption process, (2) its ther-
modynamics, and (3) the character of the interaction between the protein mole-
cules and a surface.
It is first necessary to attribute the diffusional mobility (or diffusivity) of
proteins to the kinetic peculiarities. The diffusivity of proteins is two to three
orders of magnitude less than that of small molecules (including the solvent and
the modifier molecules; in other words, the eluent). This means that the residence
time of a protein molecule in the mobile phase between consecutive adsorption-
desorption acts is 2-3 times as large as that of a small molecule also present in
the eluent, e.g., the modifier.
A thermodynamic peculiarity of proteins, which also differentiates them

from small molecules, is the so-called steric factor. It is obvious that a molecule
of the size of a protein will experience significant steric limitations when inside
the pore channels of a porous particle. In addition, it becomes increasingly diffi-
cult to enter the "mouth" of a pore channel. This phenomenon is most important
for sorbents with rigid pore structure.
The peculiarity of interaction is also linked with the size of the protein
molecule. Protein molecules, which can be seen as large globules, carry at their
outer surface the functional groups that provide an adsorptive interaction with
the sorbent. The distribution of these interactive groups on the protein surface is
not a statistical one; instead, one finds local groups or clusters. This leads to
isolated strong interaction sites, which are inhomogeneously distributed over the
surface of the protein globule. This distribution, which partially defines the Z
parameter, differs widely between molecules. This difference is exploited in their
separation by HPMC.
Besides the above-mentioned peculiarities, which influence the chromato-
graphic process, it is useful to remember some others aspects that can also become
decisive in the different modes of interaction chromatography. In particular, we
should remember the concentration effect, which becomes influential whenever
the protein concentration is high enough to make the presence of more than one
molecule in a given pore likely. Apart from the concentration effect, proteins
will affect the adsorption of each other. This synergistic effect should always be
considered together with the concentration effect. Last but not least, it is neces-
sary to acknowledge the effect of "hysteresis," i.e., the fact that for many pro-
teins adsorption and desorption isotherms do not coincide. This phenomenon can
be observed because of the multiple-point character of the interaction of the pro-
tein with the sorbent surface. The desorption of the protein needs a higher mod-
ifier concentration than the one that allows its adsorption.
6. One-Step Desorption Process

Originally, the reason for wanting to model adsorption chromatography on thin
layers ("membranes") of sorbent was the observation that often the separation
effect of a column seemed to be independent of the column length [5,42-551.
Protein mixtures can often be separated on columns 5-10 cm in length. The
models were to aid a better understanding of this phenomenon. A first conse-
quence was a change in the phraseology, e.g., a change from the term "column
length" to the concept of "operative thickness (length) of adsorption layer,"
which corresponds to the minimum distance necessary to achieve a specific sepa-
ration.
A simple experiment demonstrating this effect can be performed by any-
body involved in column chromatography [56]. 1n.ject a sample of a protein mix-
ture into your column, then turn it up-side down and elute in your standard linear
gradient. Quite often you will see no difference in the resolution compared to
that obtained when the entire "separation length" is used. In this experiment the
operative thickness of the separation layer is defined by an amount of protein in
the sample.
The traditional understanding of (column) HPLC is that the experimentally
observed separation effect is accumulated during multiple acts of adsorption-
desorption. Within this theoretical framework, the effect of the initial step is
considered as too insignificant to deserve or require an individual evaluation. In
this context the theory of HPMC, as outlined below, may well engender a change
of mind. Unless indicated otherwise, the model described below interprets separa-
tions by disc chromatography as a one-act desorption process even in such cases
where the theoretical separation layer thickness does not exactly correlate with
the physical thickness of the chromatographic disc.
In our development we propose that expression (4) can indeed be used to
describe correctly the experimentally obtained equilibrium adsorption. For conve-
nience's sake we change to a single variable and write the expression for the
retention factor in a reduced form:
which is the result of the following substitution:
where yo = N'IZis the value of w at k' = 1.

By affecting these changes we are no longer troubled by the dimensions
or the lack of them for certain of these numbers.
Regnier and coworkers [40] recently published a theoretical model capable
of describing the experimental data of the equilibrium separation of proteins ac-
cording to both the ion exchange and the reversed-phase mechanism for isocratic
elution. However, the separation of proteins by adsorption chromatography is
usually done by gradient elution, where k' becomes a function of w [often given
w
as = ~ ( t ) in ] dependence of the gradient shape. Under these circumstances,
the adsorption process cannot be considered at equilibrium.
If the adsorption process is to be described in a statistical mode [57,58]
the probability of residence of the desorbed protein in the mobile phase at equilib-
rium (Pb) must be introduced. This probability is defined via the equilibrium
retention factor, k':
Note that the probability is normalized because the sum of the probability of
residence of the desorbed protein in the mobile phase and the probability of resi-
dence of the protein in the stationary phase Pa = k'/(l + k') needs to be 1. One
can correctly estimate the probability of residence of the protein in the mobile
phase at equilibrium as the integral over the probability of a transition of the
protein from the stationary to the mobile phase under nonequilibrium conditions.
The latter, in its turn, is defined as the derivative of P, over Y:
In other words, p is a probability density and the value p ( y ~ ) d yis~the probability

of the transition of a protein molecule from the stationary to the mobile phase
in a certain mobile phase modifier concentration interval. The value y~ = vo
corresponds to the maximum of p(y~).
To make this clearer, one can draw an analogy to the analysis of the pore
size distribution in a chromatographic sorbent by mercury porosimetry. In this
method, mercury vapor is forced into the pores of the support at different pres-
sures. For each pressure value the corresponding equilibrium amount of mercury
in the pores is determined. In our case this is analogous to the determination of
the protein amount in a mobile (or stationary) phase as a function of the modifier
concentration. In mercury porosimetry, a curve is defined, which links the equilib-
rium amount of mercury to the pressure drop. This curve represents the integral
over the pore size distribution. In the case of a continuous increase of the pressure
from zero to a defined value, we obtain the point of the curve corresponding to
the adjusted pressure value. Moreover, the measured value will be totally inde-
pendent of the way the pressure was changed on the way. In the same manner
is the distribution of a protein between the mobile and the stationary phase inde-
pendent of past changes in the eluent composition.
Thus, the concept of a probability density allows to use expression (6) to
define the moments of the parameter 2. The first moment describes the average
modifier concentration at which the protein is desorbed whereas the second mo-
ment defines the dispersion of this desorption process.
C. Peak Parameters in Desorption Chromatography

The experimental values of the Z parameter introduced in Eqs. (4) and (5) fall
for all proteins within a narrow interval (10'-lo2). This is a direct consequence
of the peculiarities of protein adsorption discussed above. We will use this fact
in the calculations of the moments following below. By combining Eqs. (6), (8),
and (9) we obtain the following expression for the first moment:
HPMC of Proteins
By analogy, for the second moment we find:
(7) = [ 7 p ( ~ ) d=~
2nlZ
sin (2nlZ)
Taking into account that Z >> 1, we obtain for the first moment:
and for the second moment:
The dispersion is defined by:
From the last equation, an expression for the Z parameter can be derived:
Recently it was shown [57] that expressions (12) and (13) reflect a fundamental
property of the process described by Eq. (4).
Combining Eq. (12), for the first moment, with Eq. (7) shows that, given
the typical protein values of Z, the average modifier concentration necessary for
protein desorption is equal to that for which k' = 1. This concentration corre-
sponds to the x axis intercept of a plot of In k' as a function of y~ (Fig. 2). Whether
or not a protein separation according to Eq. (6) is possible depends on the differ-
ences in the x axis intercepts of the corresponding plots and thus only on the Z
parameters of the proteins [57]. Besides that, it follows that all proteins are sepa-
rated at the same k' value. This conclusion was reached before [43], but in another
context.
Since the dispersion can be calculated from Eq. (14) and the Z parameter
from Eq. (15), it follows that the elution peak of a given protein seen after a one-
step desorption process is mathematically described by the probability density, p,
. obtain a more accurate expression for Z would necessitate a
as a function of y ~ To
measurement of the peak shape of the same protein passed through the chromato-
graphic layer in the absence of adsorptive interaction.
It was shown [58] that for processes describable by Eq. (4) or the corre-
Tennikova and Freitag
0
log 11D
Figure 2 Graphic dependence of the fundamental expression of interactive chromatog-
raphy of proteins. The lower "pole" of the crossed straight lines correspond tu the mem-
brane separation [59], whereas the upper ones reject the RP-HPLC results using a differ-
ent kind of displacer 1401.
sponding, linearized Eq. (3,so-called poles can be found, e.g., interceptions

between plots of Eq. (5) for different Z values. Without going into detail, we
would like to postulate here that similar poles exists for plots of Eq. (14) and
that a similar behavior is displayed in all modes of chromatography that described
k' as a function of the molecule and system parameters. In the case of one-step
desorption chromatography, experimental proof is shown in Fig. 2, where the
plot of published experimental data (from Refs. 40 and 59) according to Eq. (14)
yields straight lines. Figure 2 also shows some basic differences between column
and disc protein separations. Obviously, the pole is located at a much lower y
value in disc chromatography, while the distance between the x axis intercepts
(a measure for the resolution) are wider. The latter fact indicates that HPMC may
well be superior in terms of separation power.
D. Resolution and Efficiency in HPMC

It follows from the proposed theoretical model of protein separation by HPMC
that the resolving power of this process depends not on the number of theoretical
plates (the number of adsorption-desorption acts is one) but on the initial ("mo-
mentary") distribution of the protein molecules according to the distribution of
their energy of interaction with the sorbent surface. Obviously, for an increasing
Z parameter (sharper logarithmic dependence of k' on the modifier concentration)
the distribution of the probability density [ p ( ~ ) of
] protein transition to the mo-
bile phase will become sharper. In addition, the width of the protein peak depends
on the distribution of the Z parameter. The latter becomes broader with increasing
Z and exceeds significantly the second moment of the probability density, o,thus
becoming decisive for the peak width actually seen in the chromatogram.
If Eq. (15) is to be used for the determination of Z via the dispersion o,
it is necessary to separate the contribution of Z only to the overall distribution
from all other contributions to the peak shape. A possible way of achieving this
consists of fractionation of the eluting zone of a protein into small portions using
a long-step gradient with very small changes of the modifier concentration in
each step (low and long steps). It is then possible to obtain a set of "components"
with practically identical Z values for a given modifier concentration range [47].
Rechromatographing of the individual fraction of identical Z values will result
in peaks of a width that corresponds to the dispersion of the probability density
of the protein's residence in the mobile phase.
The resolution obtainable by HPMC will decrease rapidly if the Z values
of the substances to be separated are close. Moreover, an existing small difference
between the (average values of the) Z parameter can be completely swallowed
up by the distribution of the Z values for the individual proteins. As a conse-
quence, proteins having closely similar log k' versus log y/ dependencies are most
difficult to separate by gradient HPMC.
Since HPMC is interpreted as a one-step desorption process whereby the
parameters of the separation are not averaged out during the migration of the
substance along the adsorptive layer, the topography of the surface of the chan-
nels (pores) significantly influences the width of the eluted peak. The presence of
so-called 'connecting drifts' (e.g., stagnant, non-flow-through pores) introduces a
diffusional motive into the otherwise very simple mechanism of HPMC. As a

result, the distribution of the adsorption energy will become so wide as to render
the probability of separation to practically zero. Not only that, but these pores
present a serious kinetic limitation for the leaving of desorbed molecules from
the sorbent surface. Wide and asymmetrical peaks with low resolution will be
found in the ensuing chromatogram.
The use of very steep modifier gradients allows one to separate components
with very close properties by HPMC [20]. It should be noted that in this case
small peaks visible in the chromatogram may equally be the result of a momen-
tary energetic inhomogeneity of the adsorptive interaction, which can be caused
by an inhomogeneity of the adsorptive surface, or by the presence of impurities in
the sample. The latter can be confirmed by repeating the separation. The energetic
inhomogeneity has a more or less statistical character; thus peaks caused by this
phenomenon will occur at different elution volumes in consecutive chromato-
grams, whereas peaks due to impurities should be completely reproducible.
E. Fluid Dynamics of Pore Channels

In HPMC the mobile phase streams through pore channels whose walls also form
the adsorptive surface. Understanding the related fluid dynamic phenomena is
an important first step in the control and optimization of membrane chromatogra-
phy [59-621.
The overall linear flow velocity U of the mobile phase through the disc
can be written as follows:
where F is the volumetric flow rate, A is the cross-section of the disc, and E is
the porosity.
Equation ( 1 6) does not reflect the fact that the flow velocity in the individual
pores may not necessarily be equal to the overall velocity U, since it will depend
on the size of the pores. From the Hagen-Poiseuille law, we know that the flow
through a tube is proportional to the square of its diameter:
where AP is the pressure drop along the liquid path and q is the dynamic viscosity
of the liquid.
Although Eq. (17) applies exactly only to the flow through a straight cylin-
drical tube of radius r and length L,, it was recently demonstrated that the flow
through a macroporous monolithic layer (disc) can at least be approximated by
this law [17].
HPMC of Proteins
Rearrangement of Eq. (17) yields:
which shows that the pressure drop per unit of linear flow velocity increases
exponentially with decreasing tube diameter. In other words, the larger the pore,
the lower the flow resistance, and the more liquid flows through this pore at a
given pressure.
As a result, the flow through the small pores is very slow and can even
approach zero because the overall back pressure within the cartridge, which
drives the liquid through the porous matrix, is lower than that required to enforce
a flow through the smaller pores. This uneven distribution of flow has the effect
of decreasing the surface area available for the adsorption-desorption process.
Therefore, membranes with a broad pore size distribution are not desirable be-
cause a considerable portion of the pores may not be open to the mobile phase
and thus not participate in the separation. The ideal matrix for disc chromatogra-
phy may thus be assumed to be characterized by large pores and a narrow pore
size distribution.
A very simple description of the effect of pores and pore size distributions
on chromatographic separations can be based on the residence time of the mobile
phase within the membrane tre,and the time scale tfilm of the mass transfer through
the film of stagnant liquid to the pore wall (adsorptive surface). In order to achieve
a good separation, t,, must be larger than t,,,.
The residence time is a simple function of the thickness of the separation
layer L and the flow velocity U:
while t,,, increases exponentially with the pore diameter d,:
where D is the diffusion coefficient of the protein.

The residence time can be controlled through the flow rate and the thickness
of the disc, whereas the transport time is controlled by the pore size. Very thin
adsorber layers in combination with very large pores are thus not well suited for
the separation of proteins. The best separation can be achieved on discs with an
optimum size, i.e., pores that are sufficiently small to allow the protein molecule
to reach the wall within a reasonable amount of time, while they are also suffi-
ciently large for a fast flow at a reasonable back pressure. The separation will
benefit from a narrow distribution of the pore sizes (see above).
Membrane and especially high-performance membrane chromatography

has often been met by a certain amount of skepticism from putative applicants.
The doubts are often expressed in such questions as "How can the separation
of complex protein mixtures be possible with these thin supports?" The presented
theoretical considerations hopefully succeeded in removing some of these reser-
vations. Concomitantly we tried to outline some basic differences between the
process of column and membrane chromatography, which must be kept in mind
if the application of membranes to chromatographic separations is to have the
maximal practical impact.
V. APPLICATION OF MEMBRANE CHROMATOGRAPHY

FOR PROTEIN SEPARATION
In the first part of this chapter we tried to formulate the theoretical basis of
HPMC. Our goal was to define this evolving technique within the general frame-
work of adsorption chromatography, but also to convince the newcomer to the
field that it is indeed possible to separate protein mixtures over very short (mi-
crometer to millimeter) separation distances. In the second part of the chapter
we will illustrate the advantages of membrane chromatography using examples
drawn from the day-to-day laboratory practice. The application of HPMC will
be stressed, but other types of membrane chromatography will also be considered.
A. Elevated Flow Rates

Conventional chromatographic columns, especially when used for protein separa-
tions. usually show a rapid decrease in column efficiency (plate height) with
increasing flow rate. This increase in band broadening is due to mass transfer
resistance and should therefore be considerably less pronounced in HPMC. In-
deed, as Fig. 3 shows, the "plate height," H, formally calculated using the fol-
lowing formula
where w , , is the peak width at half height, t, is the retention time of the tracer,
and 6 is the thickness of the membrane, does not increase for flow rates between
0.1 and 30 mllmin. In spite of the fact that the plate height concept does not
fully apply to membrane chromatography it must be noted that the peak shapes
stay the same in disc chromatography independent of the mobile phase flow rate.
Taking into account that short separation layers cause less back pressure than
conventional columns, very fast separations should become possible in HPMC.
This is indeed the case for ion exchange membrane chromatography as Fig. 4
demonstrates. A mixture of three standard proteins could be resolved equally
HPMC of Proteins
0
0 5 10 15 20 25 30
Flow Rate [mUmin]
Figure 3 Dependency of the plate height, H, calculated from Eq. (22) as a function of
the mobile phase flow rate. Disc, Sartobind Q (Sartorius AG); tracer, lysozyme (1 mg/
ml).
well using mobile phase flow rates of 1 , 3 , 7 , and 40 mllmin, whereas the duration
of the separation was concomitantly reduced from 30 min to 30 s. (Fig 4a). This
is also hue for more complex separations such as that of human serum at 1, 5,
and 10 mllmin (Fig. 4b) [63] and that of a nucleotide mixture at 1 and 10 mll
min (Fig. 4c). This flow rate independence of the separation is not necessarily
found in all varieties of HPMC. While ion exchange chromatography is character-
ized by very fast surface interaction kinetics, other types, e.g., affinity chromatog-
raphy, show much slower reaction kinetics. If the flow rate is increased indiscrim-
inately in these cases, a loss in resolution can sometimes be observed [64], simply
because the separation becomes reaction-limited.
A second question concerns the effect of the flow rate on the capacity of
the disc. Here both the pore size distribution and the construction of the disc
cartridge (mainly the inlet where the incoming flow is distributed equally over
the entire cross-sectional area) are of major importance. While some discs can
be used at flow rates of more than 30 mllmin without significant loss in capacity,
others already lose more than 50% of the low flow rate capacity when the flow
rate is increased to 10 mllmin [65] (Fig. 5). Affinity discs seem again to be more
prone to a loss in capacity than the ion exchange types.
Once the advantage of HPMC over conventional column chromatography
in terms of fast and efficient protein separation has been realized, a number of
-
p so
i
0
20 40
[Sek.]
flow rate 1 mllmin

\_ separation time 50 min
flow rate 5 mllmin

separation time 10 min
flow rate 10 mllrnin

separation time 5 min
Figure 4 Demonstration of the flow rate dependency of HPMC separations. (a) Separa-
tion of a standard protein mixture (cytochrome c, a-chymotrypsinogen, lysozyme, 1 mgl
ml each); disc, strong cation exchanger (prototype Sartorius AG). (b) Separation of human
serum proteins (total protein 3 mglml) on an anion exchange membrane (prototype Sarto-
rius AG). (c) Separation of a nucleotide mixture (NADINADH, 50 pglml; NADPl
NADPH, 150 pglml) at 1 mllmin and 10 mllmin on a strong cation exchange membrane
(prototype Sartorius AG). (From Ref. 63.)
r
HPMC of Proteins
Abrurprion (AUFS) [%I
0.0 10.0 20.0 rnin
Figure 4 Continued
obvious fields of application appear. Chromatographic discs are real high-perfor-

mance stationary phases capable of resolving even complex mixtures. In Fig. 6
the example is given of the monitoring of the mounting product concentration
in a biotechnological fermentation broth [63]. In this area HPMC could become
useful for fast, automatic (given modem chromatography/bioreactor software),
and cheap at-line product monitoring. The replacement of the old trouble-prone
and low-resolution cartridges used in heterogeneous flow injection analysis (FIA)
by the respective discs has recently been proposed by Tennikova et al. [66]. In
this context HPMC was shown to improve the detection limit, the resolution, the
reliability, and even the standing time of the FI analyzer.
Another important analytical application can be found in the area of hy-
phenated analytical techniques. Many of these techniques rely on a chromato-
graphic step to resolve a complex sample and then use a second analytical tech-
nique (e.g., MS) to gain further information on the individual components. Here
again the combination of speed, resolution, and cheapness is intriguing. In Fig.
7, a 1-ml sample of a cell culture supernatant containing a recombinant protein
product and 2% fetal calf serum (FCS) is chromatographed using an anion ex-
change disc (prototype Sartorius AG). The product-containing fraction is auto-
matically injected into a capillary electrophoresis system for quantification [67].
Under optimized conditions, the entire analysis can be carried out in 2 min. Since
the product is not only enriched but also concentrated by HPMC, the detec-
tion limit of the coupled technique is one order of magnitude lower than that of
capillary electrophoresis alone. The reproducibility and standard deviations are
similar.
Static Capacity
r Dynamic Capacity
Flow Rate [mYmin]
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Flow Rate [mllmin]
X Anion Exchanger Cation Exchanger -regression
Figure 5 Influence of flow rate on capacity. (top) Static and dynamic antithrombin I11
binding capacity of a heparin affinity membrane adsorber (prototype Sartorius AG). (From
Ref. 65.) (bottom) Static capacity, strong anionlstrong cation exchanger (both prototype
Sartorius AG). The disc was loaded with human serum albumin until saturation. Afterward
the amount retained was eluted and quantified. (From Ref. 67.)
HPMC of Proteins
n j3-Galaktosidase
200
150
100
50
25
10
Medium
Figure 6 Quantification of P-galactosidase in culture supernatants with an anion ex-

change membrane adsorber (prototype Sartorius AG). (From Ref. 63.)
0. Preparative HPMC
Given the particular advantages of HPMC, this mode should also be applicable
to preparative separations. To date the majority of application protocols is given
for small discs (mm x cm range) (see below). Most applicants still tend to use
a single-step interaction mode, much in the tradition of affinity filtration, rather
than exploiting the full advantages of the new stationary phase geometry in chro-
matographic terms. Multistage separation processes are therefore still rare, al-
though they do exist [65,68].
Antithrombin I11 is an anticoagulant with considerable therapeutic value.
It has thus become a target substance for biotechnical production and several
mammalian cell lines exist, which express human antithrombin I11 as a recombi-
nant protein. The recovery of the product from the cell culture supernatant is
made quite complex by the fact that the product is normally found only in minute
amounts (in the case considered here typically 1% of the total protein concentra-
Time [rnin]
Time [rnin]
Figure 7 (top) Chromatogram of a cell culture supernatant using a strong anion ex-
change membrane adsorber (prototype Sartorius AG). The product containing fraction is
indicated. (bottom) Electropherogram of the product containing fraction. (From Ref. 67.)
tion) and that large amounts of FCS are normally added to the production environ-
ment (here typically 10%). Lowering the serum concentration often yields a prod-
uct of inferior quality, e.g., low activity, low stability [67]. HPMC was used as
a cheap but high-performance preparative approach to process between 0.1 and
10 L of culture supernatant.
In a first attempt, heparin affinity HPMC was used in combination with an
ultrafiltrationldiafiltration step 1651. While the latter served mainly to concentrate
the protein fraction in the culture supernatant, we hoped to isolate pure antithrom-
bin I11 by the heparin affinity step. Column heparin affinity chromatography is
a standard procedure to isolate antithrombin 111, since the protein shows a pro-
nounced affinity for its cofactor.
The results of the first attempt are summarized in Table 1. Of the original
AT I11 activity, 81% was recovered by the heparin disc. In addition, the product
was concentrated by a factor of 101 during this step. When the protein was ana-
lyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immu-
nodiffusion its purity was established at 64%. Large amounts of both bovine
serum albumin and bovine transferrin as well as traces of bovine immunoglobulin
G and other unidentified proteins were still present. One hundred sixty minutes
was required for the processing of one batch of culture supernatant in this case,
corresponding to 0.7 mglh AT 111.
In order to improve the final purity, ion exchange HPMC steps were inte-
grated in the isolation scheme. It was possible to nearly remove the b-IgG by
a strong anion exchange disc, as was verified by immunoblotting. The final pur-
ity of the antithrombin was 85%, with bovine serum albumin as the major con-
taminant.
Since none of the procedures was capable of removing the bovine serum
albumin totally, a further improvement of the product quality was attempted by
Table 1 Purification of Recombinant Antithrombin I11 from Cell Culture

Supernatants by a Three-Step HPMC Process
Cell culture Ultrafiltration1

supernatant diafiltration Heparin MA
AT 111 (pglml) 7.5 6 608

AT 111 ( ~ g ) 1875 Not determined 1520
Specific activity (Ulmg) 700 661.97 571.97
Total protein concentration (mglml) 6.1 5.64 0.95
BSA (mglml) 3.64 3.36 0.167
P-Transferrin (mglml) 0.21 0.187 0.043
P-IgG (mglml) 0.77 0.68 0.007
Source: From Ref. 65.
introducing a step for the specific removal of the albumin. Cibacron blue ligands
are known to interact with bovine serum albumin but not with antithrombin. It
was found that the Cibacron blue disc is best used before both the anion exchange
and the heparin disc (Table 2). A cycle required 300 min in this case and yielded
approximately 2 mg of product (0.4 mglh). The final procedure yields a 94%
pure product with a recovery of 72%. Purities of >99.9% were reached only
when the serum concentration in the culture was lowered to 3%. Then, however,
the biological activity of the product suffered [65].
In order to increase productivity, the flow rate over the discs was increased
stepwise from the 2 mllmin (0.12 Llh) used during development to 4.8 Llh. The
maximum flow rate through the heparin MA had to be restricted to 0.18 Llh
during these experiments, since affinity discs show a pronounced flow rate depen-
dency of the dynamic capacity. Nevertheless the cycle time was reduced consider-
ably by this increase of flow rate and the antithrombin throughput concomitantly
increased from 0.55 to 13.7 mglh (Table 3).
The scale-up of HPMC to fully preparative dimensions is also rarely done.
The easiest way to increase the scale of a HPMC separation is to increase the
number of discs involved, either by inserting several discs into a cartridge (mod-
em disc cartridges can usually incorporate varied numbers of discs) or by short-
circuiting several modules. Figure 8 shows such a scale-up by an increase in the
number of discs. By increasing that number from 1 to 6, the total amount of
trypsin inhibitor, a-chymotrypsinogen, and lysozyme separated could be in-
Table 2 Purification of Recombinant Antithrombin I11 from Cell Culture

Supernatants by a Three-Step HPMC Process
Anion
Cell culture Ultrafiltration/ exchange Heparin
supernatant diafiltration CB-MA MA MA
AT I11 (pglml) 11 10.9 48.1 43.8 657

AT 111 (pg) 2750 2725 2405 2190 1971
Specific activity 790.9 798.2 684 630.2 577.8
(Ulmg)
Total protein 5.7 5.5 8.1 0.36 0.699
concentration
(mg/ml)
BSA (mg/ml) 3.4 3.3 0.22 0.19 0.015
P-Transferrin 0. 16 0.16 0.76 0.02 1 0.001
(mg/ml)
0-IgG (mglml) 0.79 0.8 3.6 0.003 0
Table 3 Influence of the Mobile Phase Flow Rate on the Separation of Recombinant
Antithrombin 111 by the Three-Step HPMC Process
Process AT 111 Yield

Flow rate time throughput AT 111 recovery Concentration Purity
(L/h) (h) (rnglh) (mg) (%I factor (%I
\ . . . . . . , . . . , . .
0 2.0 4.0 6.0
Time [min]
Figure 8 Separation of a standard protein mixture (lysozyme, a-chyrnotrypsinogen, and

1 trypsin inhibitor) using anion exchange membrane adsorberr (prototype Sartorius AG).
(a) One layer, 0.25 mglml of each protein. (b) Three layers, 0.6 rnglml of each protein.
(c) Six layers, 1.2 mglrnl of each protein. Flow rate was 12 mllmin throughout. (From
Ref. 63.)
creased from 0.25 to 1.2 mg [63]. By keeping the flow rate and the gradient
volume constant, the retention times were kept equal. The resolution does suffer
slightly, since the protein zones are somewhat broader at higher concentration.
An interesting effect was observed when coupling modules to increase the
maximum sample load. When working close to the separation capacity (i.e., the
amount of protein that could still be resolved with a resolution of 1 or better)
the resolution was found to improve whenever two modules were used in series
even though the loading, i.e., the amount of protein per cm2, was kept constant
[69]. In other words, we find a higher separation capacity in the case of two
modules than for a single one. No further improvement is observed when more
than two modules are stacked. The improvement in separation is not due to the
increase in the number of disc layers or simply to an increase in total area, since
two consecutive smaller modules [total of six layers (2 X 3) representing 60 cm']
resolves a given protein mixture better than a single large one (five layers equal
to a total of 100 cm2).
While such an increase in the number of layers can often suffice to carry
a separation from the analytical to the semipreparative scale, a truly large-scale
membrane chromatography requires a more significant increase in area. In this
context HPMC can profit considerably from filtration theory, to which it bears
a certain resemblance. In fact, various forms of filtration have been adapted to
membrane chromatography, including cross-flow and hollow fiber modules. Fig-
ure 9 shows the scale-up of a separation of three proteins (a-chymotrypsinogen,
cytochrome c, lysozyme) from a small module (0.5 cm2) to a large one (130
cm2).The flow (ml/cm2)and the protein load (mg/cm2)were kept constant, while
the volumetric flow rate was increased from 0.2 to 50 ml/min. Given the difficulty
of recording a UV trace for the larger modules, the separation stayed much the
same.
The largest module (fitted this time with 10 layers, each 130 cm2) was also
used in the processing of technical dairy whey, where throughput is an important
point [69]. Five liters of feed solution was passed through this module at a flow
rate of 50 mllmin. Using a two-step NaCl gradient (up to 1 M NaCl), 248 mg
of protein was eluted within 14 min: 116 mg in step 1 (0.1 M NaCl, mostly a -
lactalbumin) and 132 mg in step 2 (0.5 M NaCl, mostly P-lactoglobulin). The
recovery was >70% for both proteins.
C. Mixed-Mode HPMC
Save for a few exceptions, such as hydroxyapatite chromatography or the
Bakerbond ABX phase (Baker Chemical Co., USA), which were especially de-
veloped for antibody purification, mixed-mode interactions are avoided in liquid
chromatography. On the other hand, single-stage chromatography is seldomly
sufficient for preparative protein separation and multistage procedures, where
HPMC of Proteins
Module 1: 0.5 crn2, 0.2 rnllrnin (1 rnl/crn2)
Module 4: 130 cm2, 50 mllrnin (1 rnllcrn3

i\
Figure 9 Scale-up of a separation of three standard proteins (a-chymotrypsinogen, cyto-

chrome c, lysozyme, I mglml each) using cation exchange HPMC (membrane material
Sartorius AG). The modules were either self-assembled or standard dead-end filtration
modules fitted with the interactive membrane layers.
several types of interaction are exploited in series are the norm. Controllable
mixed-mode phases may thus become an interesting tool. However, with conven-
tional porous particle-based stationary phases, mixed-mode supports are difficult
to realize in a reproducible, homogeneous, but flexible manner. The number of
(usually prepacked) columns to be offered by the suppliers would increase dra-
matically with the number of possible or desirable combinations. This is radically
different in the case of HPMC where the use of stacks of discs has already become
routine (see above). In this case even the most complex mixed-mode approach
calls only for the insertion of different discs into the cartridge. In fact, an approach
using discs with different interactive groups in one housing is currently being
promoted by a major HPMC supplier (BIA, Slovenia) under the name of combi-
natorial liquid chromatography (CLC).
That the separation by one type of disc is not disturbed by any other type
can be demonstrated in a simple experiment. Interactive discs, e.g., ion exchange
discs, are alternated with inert filtration discs and the chromatograms compared
with those obtained for a cartridge filled only with the ion exchange material. In
our experience there will hardly ever be a difference unless the putatively inert
filter discs are not really inert (in the case of proteins, hydrophobic interactions
may be present).
Figure 10 shows the separation of five standard proteins [(P-lactoglobulin
(IEP: 5.1), conalbumin (IEP: 6.8), a-chymotrypsinogen (IEP: 9.6), lysozyme
(IEP: 10.5), and cytochrome c (IEP: 11.2)] by mixed-mode ion exchange chroma-
tography together with the chromatograms of the single substances on the same
stationary phase [70]. Five strong anion exchanger and five strong cation ex-
changer membrane adsorbers (prototypes Sartorius AG) were alternated in the
cartridge. Note that the separation becomes possible at physiological pH with
the mixed-mode approach, since it is no longer necessary to go to extreme buffer
pH in order to ensure that all proteins carry the same net charge.
Alternating the discs gives a better resolution than using two stacks (Fig.
1 1 ) [70]. As always in ion exchange HPMC, the separation seems to be nearly
independent of the flow rate. Figure 12 shows again the separation of the standard
protein mixture on the alternating discs using flow rates between 2 and 12 mll
min [70]. The gradient volume and shape were kept constant during these experi-
ments. While the separation required 30 min at a flow rate of 1 mllmin it could
be performed in 1.6 min with a flow rate of 12 mllmin.
The mixed-mode approach is not restricted to mixed-mode ion exchange
HPMC. Mixed mode ion exchangelaffinity HPMC is also promising-especially
so because it combines the advantages of affinity chromatography, i.e., the ability
to be highly selective for the target molecule which is often also considerably
concentrated, with that of ion exchange chromatography, e.g., the ability to re-
solve the entire substance mixture. In mixed-mode ion exchangelaffinity HPMC
a single-step separation again becomes possible, which would otherwise have
HPMC of Proteins
0.0 10.0 20.0
Time [min] Time [min]
Figure 10 Mixed-mode approach to the separation of a mixture of five standard proteins

(left). On the right-hand side the chromatograms of the individual proteins on the mixed-
mode phase are given. Five layers of strong anion and strong cation exchange membrane
adsorbers (prototype Sartorius AG) were alternated in the cartridge. (From Ref. 70.)
required at least two steps [70]. We have seen above that the isolation of the
recombinant anticoagulant antithrombin 111 from cell culture supernatant requires
several chromatographic steps. The removal of bovine serum albumin and bovine
transfemn is especially difficult. The separation of the antithrombin I11 from these
two contaminants requires at least two chromatographic steps based on different
interaction principles. The albumin can be removed by Cibacron blue affinity
chromatography (Fig. 13a), which doesn't help to lower the bovine transfemn
concentration. Bovine transferrin, on the other hand, can be removed-due to
the difference in IEP-on a strong anion exchange chromatography (Fig. 13b).
However, this step does not remove the albumin. Only one possibility has been
reported to at least partially resolve the three-protein mixture and that was immo-
bilized metal affinity disc chromatography (IMADC) using Cu2+ions (Fig. 13c).
In this case, however, antithrombin activity is found in the breakthrough as well
as in one of the later eluting fractions.
By the simple measure of using controlled mixed-mode anion exchange1
affinity interaction, on the other hand, the separation of recombinant antithrombin
Ill, bovine serum albumin, and bovine transfenin becomes possible in a single
1
CHY
CYT LYS
CON
LJ
Alternating MA
\
0.0 20.0
Time [min]
Figure 11 Comparison of the alternating versus the two-stack approach to mixed-mode

HPMC of the standard protein mixture. (From Ref. 70.)
mixed-mode HPMC separation at pH 7.0 (Fig. 13d). Since the bovine serum
albumin may be expected to interact mainly with the affinity membrane adsorber
(i.e., a Cibacron blue disc, prototype Sartorius AG) under these circumstances,
it is not surprising that a comparatively high NaCl step gradient (3 M NaCl) is
required for elution of that particular protein. The antithrombin I11 and the bovine
transferrin, on the other hand, which should adsorb mainly onto the anion ex-
changer MA, are eluted in a linear NaCl gradient up to 1 M NaCl.
VI. EXAMPLES
Of the various types of adsorption chromatography, the two most commonly used
types of HPMC in preparative protein separations are affinity chromatography
("affinity filtration") and ion exchange chromatography. In spite of the potential
HPMC of Proteins
2dmin
4 mVmin
6dmin
8dmin
10 mVmin
Figure 12 Influence of the mobile phase flow rate on mixed-mode protein HPMC.
(From Ref. 70.)
of HPMC in general, only a limited number of applications in biotechnology are

known, hardly any of them in a large scale. The following list of examples is
not intended to be an all-inclusive summary. However, it is intended as a compre-
hensive guide to membrane chromatography of proteins and other biopolymers.
Some of the papers applied to more than one category but were nevertheless
listed only once. The interested reader is also referred to the review by Thommes
and Kula 1711.
Albumin
_J
I I
0 10 20 30
Time [min.]
0 20 40
Time [min]
Figure 13 Separation recombinant antithrombin 111, bovine serum albumin, and bovine
transferrin. (a) Cibacron blue affinity membrane adsorber. (b) Strong anion exchange
membrane adsorber. (c) Immobilized metal affinity membrane adsorber (Cu2+ions). (d)
Mixed-mode HPMC, using the Cibacron blue membrane adsorber and the strong anion
exchange membrane adsorber. (From Ref. 70.)
HPMC of Proteins
Transferrin
Imidazol
lmMl
r 20
Volume [ml]
BSA
10.0 20.0
Time [min]
Figure 13 Continued
A. Affinity Filtration
Target molecule Key phrases Ref.
Antibodies Protein A ligands, preacti- Lee et al. 1992 [72]

vated discs, discussion of
advantages of disc, con-
ference proceedings
Human antithrombin 111 High-performance mem- Josic et al. 1993 [73]
from plasma brane affinity chromatog-
raphy (HPMAC) is used
to monitor the purifica-
tion by conventional col-
umn chromatography
Rabbit antibodies, mono- Investigation of epoxy-acti- Langlotz and Kroner
clonal antibodies from vated discs, immobiliza- 1992 [74]
cell culture tion of Protein A and Pro-
tein G, application to
antibody purification
Malate dehydrogenase (pig Comparison of blue Sepha- Krause et al. 1991 [75]
heart, E. coli) rose (Pharmacia) and
Cibacron blue modified
discs (Sartorius AG),
comparison of dead-end
and cross-flow filtration
Antithrombin, albumin, vari- Comparison of discs and Kasper et al. 1997 [76]
ous antibodies, recombi- columns, investigation of
nant gp 2201350 Epstein- different membrane mate-
Barr virus surface rials, different coupling
antigen chemistries for pre-acti-
vated discs
Recombinant fusion protein Immobilized metal affinity Nier et al. 1994 [64]
(EcoR V, HIS,-tag), disc chromatography, in-
cytochrom C from vari- vestigation of the influ-
ous sources varying in ence of metal ion type,
the number of surface the flow rate, the salt con-
histidines centration, and the elut-
ing agent, isolation of the
fusion protein from cell
ly sates
Carbonic anhydrase from Epoxy-activated discs, cou- Abou-Rebyeh et al.
human erythrocytes pling of p-aminometh- 1991 [21]
ylbenzolsulfonamide as
ligand, coupling of the
enzyme itself for kinetic
investigations
Various Review article discussing Brandt et al. 1988 [78]

the potential of mem-
brane-based affinity tech-
nologies for biotechnol-
ogy giving some
theoretical consider-
ations and examples
Bovine immunoglobulin Protein G affinity chroma- Kochan et al. 1996 [79]
tography, investigation
of the process parameters
(flow rate, feed concentra-
tion)
B. Hollow Fiber Systems

The larger part of this chapter has explicitly dealt with chromatographic discs as
stationary phase. HPMC can be regarded as a special-and comparatively well
understood-form of membrane chromatography in general. Especially in the
area of affinity filtration, which operates according to the yes-or-no principle of
adsorption, other process modes have been used with much success. Hollow fiber
systems are popular because they allow a rapid cross-flow adsorptive filtration
of the feed with little danger of pore blockage. A number of examples are listed
below.
Bovine immunoglobulin G Hydrophobic amino acids Kim et a]. 1991 [SO]

(phenylalanine or trypto-
phan) as ligands, radia-
tion-induced grafting of
glycidyl methacrylate
onto a porous polyethyl-
ene hollow fiber to allow
ligand immobilization
Interferon-a2a, interleukin- Investigation antibody- Nachman et al. 1991
2, interleukin-2 receptor antigenlreceptor-based [Sl], 1992 [45]
membrane chromatog-
raphy
Standard proteins Immobilized metal affinity Serafica et al. 1994 [82]

chromatography, glass
hollow fiber microfiltra-
tion system, coupling of
imminodiacetic acid
(IDA) to the wall, com-
parison to conventional
columns, theoretical treat-
ment of hollow fiber
membrane chromatog-
raphy
Antibodies Protein A, Protein G, pro- Garg et al. 1992 [83]
cess development, scale-
up, various forms of
application, conference
proceedings
P-Galactosidase Ion exchange hollow fiber Heng and Glantz
system, charged fusion 1993 [84]
tag
Bovine serum albumin Microporous nylon-6 hol- Kugel et al. 1992 [85]
low fibers were modified
by reaction with lysine,
polyclonal antibodies
were linked to the sur-
face via their oxidized
carbohydrate side chain
C. Therapeutic Proteins
Factor VIll from human Use of ion exchange discs Strancar et al. 1997 [86]
plasma (also for process monitor-
ing) and scale-up via the
use of a compact porous
tube made from the same
material as the discs
Monoclonal antibodies from Cation exchanger, scale-up Liitkemeyer et al.
cell cultures from disc to hollow fiber 1992 [87]
system, process up to har-
vest from 100 L bioreac-
tor, recoveries >96%
Human tumor necrosis Recombinant protein pro- Lukas et al. 1994 [24]
factor-a duced in E. coli, product
electrophoretically pure,
anion exchange HPMC
Recombinant antithrombin Ion exchange and heparin Liitkemeyer et al.
111, monoclonal antibod- affinity disc chromatogra- 1993 1881
ies from cell cultures phy, scale-up to hollow
fiber system (2400 cm2)
in case of ion exchanger,
coupling chemistry for he-
parin
Serum and plasma proteins Scale-up from 10- to 50- Josic et al. 1992 [22]
mm disc (weak anion ex-
changer), comparison to
column chromatography
Annexins from liver plasma Anion exchange disc, sepa- Josic et al. 1994 (231
membranes, monospecific ration of annexins from
polyclonal antibodies each other in 10 min,
comparison of disc and
cellulose fiber modules,
immobilization of an-
nexins to epoxy-acti-
vated discs for antibody
purification
Monoclonal antibodies from Thiophilic membranes, Finger et al. 1995
cell culture membrane stacks and
cross-flow module, inter-
pretation of the flux be-
havior according to filtra-
tion theory
Kistler and Nitschmann's Weak anion exchanger Lacoste-Bourgeacq et al.
fraction IV of blood 1991 [90]
plasma
Recombinant antithrombin Multistage disc chromatog- Reif and Freitag
111 from cell culture raphy process (affinity 1994 16.51
and ion exchange chroma-
tography), scale-up to 10
L scale
Whey proteins, permeate Comparison to column chro- Splitt et al. 1996 [69]
components matography, mixed-mode
approach, influence of
the process parameters
(flow rate, pH), scale-up
considerations

Monoclonal antibodies Radial streaming ion Jungbauer et al.
exchange chromatog- 1988 [91]
raphy, scale-up study
(pilot scale), calculation
of scale-up factors, com-
parison with experimen-
tal results
Human serum albumin Two-step process yields Gebauer et al. 1997 [92]
(plasma fractionation) 98% pure albumin from
clarified, microfiltrated,
and desalted plasma, com-
parison to column ap-
proach
D. Endotoxin Removal
The lipopolysaccharides, which are a major component of the cell membrane of
gram-negative bacteria, are powerful endotoxins that cause a violent reaction of
the immune system of mammals upon entering the bloodstream. Pharmaceuticals
and especially injectables need to be endotoxin-free. The removal of endotoxins
is not an easy task and chromatographic methods have been suggested. HPMC
may present itself as a fast and efficient detoxification (depyrogenation) method.
Target Comments Ref.

- - -
Depyrogenation of water, Application note by a mem- Sartorius 1997 [93]

buffer, protein solutions brane manufacturer (Sar-
torius AG) giving vari-
ous protocols
Endotoxin removal from Strong anion exchange Belanich et al.
protein mixtures discs, endotoxin removal 1994 [94]
from bacterial extracts,
scale-up to 15 L, reduc-
tion from >500,000 to
<I00 EUlml
HPMC of Proteins
E. Process Monitoring
Target molecule Comments Ref.
a,-Antitrypsin, clotting Development of HPMC Strancar et al. 1996 [25]

factor IX method explicitly for bio-
process and downstream
monitoring, anion ex-
change, cation exchange,
and hydrophobic interac-
tion mode, separation of
three standard proteins in
less than 1 min
F. DiscIMembrane Systems and General Developments
Investigation of the Mem- 30- and 15-mg of protein Gerstner et al. 1992 [95]
Sep system for analytical were purified within 3
and preparative applica- min when a cation ex-
tions change MemSep 1010
system was operated in
the step and linear gradi-
ent mode, respectively
Investigation of kinetics of Theoretical treatment, appli- Nachman 1992 [46]
immunoaffinity mem- cation of the results to
brane chromatography the isolation of biothera-
peutics
Model for membrane affin- Mathematical model for the Suen and Etzel
ity chromatography simulation and the design 1992 [96]
of dead-end affinity fil-
tration
Modeling of disc chroma- Model and experimental Briefs and Kula
tography verification for various 1992 [97]
limiting conditions, appli-
cation to protein purifica-
tion and scale-up
Investigation of mass trans- Development of a mathe- Sarfert and Etzel
fer limitations matical model to describe 1997 [62]
various processes during
protein separation by
HPMC, film mass trans-
fer resistance is found to
be the decisive factor
G. Miscellaneous
Membranes exhibiting Preparation of membranes Bamford and Al-Lamee

molecular recognition for affinity separation 1994 [98]
and immunodiagnostics,
optimization of the mem-
brane using biological
ligands, introduction of
fully synthetic functions
(e.g., triazine molecules)
that exhibit recognition
properties
Proteins by affinity adsorp- Affinity ligand is immobi- Niven and Scurlock
tion lized on nylon belt sys- 1993 [99]
tem, which is passed
through the various solu-
tions, continuous purifi-
cation system
Preparation of macroporous Synthesis protocol, influ- Jelinkova and Votavova
discs suitable for chro- ence of the synthetic con- 1991 [I001
matography ditions on the chromato-
graphic parameters
Formed-in-place anion ex- The membranes are formed Li and Spencer
change membranes by adsorption of poly- 1992 [ l o l l
mers (e.g., polyethylene-
imine) on formed-in-
place microfiltration (tita-
nium dioxide) substrates
Bacterial plasmid DNA DNA and impurities con- Huynh et al. 1993 [I021
tained in the cleared
bacterial lysates were
adsorbed to DEAE cellu-
lose membranes,
the impurities were selec-
tively removed and the
pure plasmid DNA recov-
ered afterward
HPMC of Proteins
VII. CONCLUSIONS
Membrane chromatography of proteins and other biopolymers combines speed,

resolution, and capacity in a unique fashion. In the beginning the method was
beset with certain problems. But now the theory of HPMC has been developed
and the principles underlying the separation have become clear. The decisive
factors for designing a true high-performance short-separation-layer phase are
known and membrane chromatography may well be on its way to becoming the
protein separation technology of the 21st century.
ACKNOWLEDGMENTS
The authors gratefully acknowledge Dr. Michael Tennikov for the fruitful discus-
sion,regarding the preparation of this chapter.
REFERENCES
Nomenclature for chromatography, IUPAC Recommendations, J Pure Appl Chem

1993;65(4):819.
Roper DK, Lightfoot EN. J Chromatogr 1995;702:3.
Snyder LR, Stadalius MA, Quarry MA. Anal Chem 1983;55:1412A.
Stadalius MA, Quarry MA, Snyder LR. J Chromatogr 1985;327:93.
Snyder LR, Stadalius MA. In: Horvath CS, ed., High-Performance Liquid Chroma-
tography-Advances and Perspectives, Vol 4. New York: Academic Press, 1987:
195.
Knox JH. J Chromatogr Sci 1977;15:352.
Afeyan NB, Fulton SP, Regnier FE. J Chromatogr 1992;544:267.
Fulton SP, Afeyan NB, Gordon NF, Regnier FE. J Chromatogr 1992;47:452.
McCoy M, Kalghatgi K, Regnier FE, Afeyan NB. J Chromatogr 1996;753:221.
Rodrigues AE, Lopez JC, Lu ZP, Loureiro JM, Diaz MM. J Chromatogr 1992;590:
9.
Rodrigues AE, Zuping LU, Loureiro JM. Chem Eng Sci 1991;46(11):2765.
Liapis AI, Xu Y, Grosser OK, Tongta A. J Chromatogr 1995;702:45.
Heeter GA, Liapis AI. J Chromatogr 1996;743:3.
Heeter GA, Liapis AI. J Chromatogr 1997;761:35.
Unger KK, Jilge G, Kinkel JN, Hearn MTW. J Chromatogr 1986;359:61.
Janzen R, Unger KK, Giesche H, Kinkel JN, Heam MTW. J Chromatogr 1987;
397:91.
Tennikova TB. Belenkii BG, Svec F. J Liquid Chromatogr 1990;l3:63.
Tennikova TB, Bleha M, Svec F, Almazova TV, Belenkii BG. J Chromatogr 199 1 ;
555:99.
Svec F, Tennikova TB. J Biocompat Polym 1991;6:393.
Tennikova TB, Svec F. J Chromatogr 1993;646:279.

Abou-Rebyeh H, Korber F, Schubert-Rehberg K, Reusch J, Josic D. J Chromatogr
1991;566:341.
Josic D, Reusch J, Loster K, Baum 0 , Reutter W. J Chromatogr 1992;590:59.
Josic D, Lim Y-P, Strancar A, Reutter W. J Chromatogr B 1994;662:217.
Luksa J, Menart V, Molicic S, Kus B, Gaberg-Porekar V, Josic D. J Chromatogr
1994;661:161.
Strancar A, Koselj P, Schwinn H, Josic D. Anal Biochem 1996;68:3483.
Svec F, Frechet JMJ. Anal Chem 1992;64:820.
Wang QC, Svec F, Frechet JMJ. Anal Chem 1993;65:2243.
Wang QC, Svec F, Frechet JMJ. J Chromatogr 1994;669:230.
Svec F, Frechet JMJ. Chem Mat 1995;7:707.
Svec F, Frechet JMJ. J Chromatogr 1995;702:89.
Svec F, Frechet JMJ. Science 1996;273:205.
Svec F, Frechet JMJ. Macromol Symp 1996;110:203.
Hjerten S, Liao, J-L, Zhang R. J Chromatogr 1989;473:273.
Hjerten S, Li Y-M, Liao JL, Mohammad J, Nakazato K, Petterson G. Nature 1992;
356310.
Liao J-L, Zhang R, Hjerten S. J ~ h r o m a t 6 ~1991;586:21.
r
Hjerten S, Mohammad J, Liao J-L. Biotechnol Appl Biochem 1992:15:247.
Hjerten S, Nakazato K, Mohammad J, Eaker D. Chromatographia 1993;37:287.
Minakuchi H, Nakanishi K, Soga N. Isizuka N, Tanaka N. Anal Chem 1996;68:
3498.
Kopaciewicz W, Rounds MA, Regnier FE. J Chromatogr 1983;266:3.
Geng X, Regnier FE. J Chromatogr 1984;296:15.
Geng X, Regnier FE. J Chromatogr 1985;332:147.
Coffman JL, Roper DK, Lightfoot EN. Bioseparations 1994;4: 183.
Belenkii BG, Podkladenko AM, Kurenbin 01, Maltsev VG, Nasledov DG, Trushin
SA. J Chromatogr 1993;646:1.
Dubinina NI, Kurenbin 01, Tennikova TB. J Chromatogr 1996;753:217.
Nachman M, Azad ARM, Bailon P. J Chromatogr 1992;597:155.
Nachman M. J Chromatogr 1992;597: 167.
Belenkii BG, Maltsev VG. Bio techniques 1995;18:288.
Reif 0 - W (1994). Ph.D. thesis, University of Hannover, Germany.
Roper DK, Lightfoot EN. J Chromatogr 1995;702:69.
Tanford C. The Hydrophobic Effect. 2nd ed. New York: Wiley-Interscience, 1980.
Pearson JD, Lin NT, Regnier FE. Anal Biochem 1982;124:217.
Hearn MTW, Regnier FE, Wehr CT. Proceedings of the First International Sympo-
sium on HPLC of Proteins and Peptides. New York: Academic Press, 1983.
Blanquet RS, Bui KH, Armstrong DW. J Liq Chromatogr 1986;9:1933.
Bulton WC, Nugent KD, Slattery TK, Summers BA, Snyder LR, J Chromatogr
1988;443:363.
Yarnamoto S, Naganishi K, Matsudo R. Ion-Exchange Chromatography of Pro-
teins. New York: Marcel Dekker, 1988:94.
Tennikov MB, Tennikova TB, J Chromatogr (submitted for publication).
Tennikov MB, Chem Physics (submitted for publication).
r
HPMC of Proteins
Tennikov MB, Tennikova TB. Anal Chem (submitted for publication).

Tennikov MB, Gazdina NV, Tennikova TB, Svec F. J Chromatogr (accepted for
publication).
Weibrenner WF, Etzel MR. J Chromatogr 1994;662:414.
Adisaputro IA, Wu Y-J, Etzel MR. J Liquid Chromatogr 1996;19:1437.
Sarfert FT, Etzel MR. J Chromatogr 1997;764:3.
Reif 0-W, Freitag R. J Chromatogr 1993;654:29-40.
Reif 0-W, Nier V, Bahr U Freitag R, J Chromatogr 1994;664:13-25.
Reif 0-W, Freitag R. Bioseparation 1994;4:369-381.
Kasper C, Hagedorn J, Freitag R, Tennikova T. (submitted for publication).
Freitag R, Reif 0-W, Weidemann R, Kretzmer G. Cytotechnology 1996;21:205.
Champluvier B Kula M-R. Bioseparation 1992;2:343.
Splitt H, Mackenstedt I, Freitag R, J Chromatogr 1996;729:87-97.
Freitag R, Splitt H, Reif 0-W. J Chromatogr 1996;728:129.
Thommes J and Kula M-R. Biotechnol Prog 1995;11(4):357.
Lee FH, Murphy K, Lin P, Vasan S, An D, DeMarco S, Forte V, Pietronigro D.
Frontiers in Bioprocessing 11. Am Chem Soc, Washington, DC, 1992:392.
Josic D, Bal F, Schwinn H. J Chromatogr 1993;632:1-10.
Langlotz P, Kroner KH. J Chromatogr 1992;591 : 107- 1 13.
Krause S, Kroner KH. Deckwer W-D. Biotechnol Tech 1991;5(3):199-204.
Kasper C, Reif 0-W, Freitag R. Bioseparation 1997;6:373-382.
Abou-Rebyeh H, Korber F, Schubert-Rehberg K, Reusch J, Josic D. J Chromatogr
1991;566:34 1-350.
Brandt S, Goffe RA, Kessler SB, O'Conner JL, Zale SE. Biotechnology 1988;6:
779-782.
Kochan JE, Wu Y-J, Etzel MR. Ind Eng Res l996;35: 1150- 1155.
Kim M, Saito K, Furusaki S, Sato T, Sugo T, Ishigaki I. J Chromatogr 1991;585:
45-51.
Nachman M, Azad ARM, Bailon P. Biotech Bioeng 1992;40:564-571.
Serafica GC, Pimbley J, Belfort G. Biotech Bioeng l994;43:2 1-36.
Garg VK, Zale SE, Azad ARM, Holton OD. Frontiers in Bioprocessing 11. Am
Chem Soc Washington, DC, 1992:321-330.
Heng MH, Glatz ChE. Biotech Bioeng 1993;42:333-338.
Kugel K, Moseley A, Harding GB, Klein E. J Membr Sci 1992;74:115-129.
Strancar A, Barut M, Podgornik A, Koselj P, Schwinn H, Raspor P, Josic D. J
Chromatogr 1997;760:117-123.
Lutkemeyer D, Siwiora S, and Biintemeyer H. Bioengineering 1992;8(2):34-
44.
Lutkemeyer D, Bretschneider M, Biintemeyer H, Lehmann J. J Chromatogr 1993;
639:57-66.
Finger UB, Thommes J, Kinzelt D, Kula M-R. J Chromatogr 1995;664:69-78.
Lacoste-Bourgearcq JF, Desneux Ch, Allary M. Chromatographia 199 1 ;32( 112):
27-32.
Jungbauer A, Unterluggauer F, Uhl K, Buchacher A, Steindl F. Botech Bioeng
1988;32:326-333.
Gebauer KH, Thommes J, Kula M-R. Biotech Bioeng 1997;54(2):181 - 188.
Clearance of endotoxin from solution by adsorption filtration by Sartobind mem-

brane adsorbers. Sartorius, Focus On: Developing Methods 7.
Belanich M, Cummings B, Grob D, Klein J, O'Conner A, Yarosh D, Pharm Techno1
1994; March: 142- 150.
Bamford CH, Al-Lamee KG, Adv Ma 1994;6(6):500-502.
Niven GW, Scurlock PG, J Biotechnol l993;3 1 :179- 190.
Svec F, Jelinkova M, Votavova E, Angew. Makromol Chern 1991:188:167- 176.
Li Y, Spencer HG. J Biotechnol 1992;26:203-211.
van Huynh N, Motte JC, Pilette JF, Decleire M, Colson C. Anal Biochem 1993;
21 1:61-65.
Gerstner JA, Hamilton St. M, Cramer St. M. J Chromatogr 1992;596:173-180.
Suen S-Y, Etzel MR. Chem Eng Sci 1992;47(6):1355-1364.
Briefs K G , Kula M-R. Chem Eng Sci 1992;47(1):141-149.
HPLC Purification of Recombinant
Proteins
Carr J. Smith,l Patricia Martin,l Sandra M. Scott,'
and Thomas H. Fischer2
' R. J. Reynolds Tobacco Company, Winston-Salem, North Carolina
University of North Carolina at Chapel Hill, Chapel Hill, North Carolina
I. INTRODUCTION
Recombinant proteins are novel proteins produced through DNA technology by

the joining of genetic sequences from heterologous sources. Since the first suc-
cessful DNA cloning experiments in 1973, research has exploded in molecular
biology, oligonucleotide synthesis, identification and isolation of enzymes that
cleave and join DNA, characterization of bacterial plasmids, and gene expression
[31,63]. These areas are now known collectively as the field of biotechnology. In
1976, this research lead to the production of the first recombinant protein, the neuro-
transmitter human somatostatin [23]. This great achievement was followed shortly
by the first commercially successful recombinant protein, human insulin [I 81.
The scientific, medical, and commercial implications of recombinant pro-
teins were obvious. Very important proteins could now be readily and rapidly
produced at a purity previously unavailable. The possibilities for medical diag-
nostics and pharmaceuticals provided an incredible commercial driving force.
Commercially successful recombinant proteins have already been produced for
treating cancer, allergies, human immunodeficiency virus (HIV), neurological
disease, heart attacks, blood disorders, infections, wounds, and genetic diseases
[63]. Commercially successful recombinant proteins include insulin, human
growth hormone, interferons, erythropoietin, and tissue plasminogen activator
(tPA) [14].
Several steps are involved in producing a recombinant protein. These steps
302 Smith et al.
can be summarized as follows: (1) preparation of the appropriate gene; (2) incorpo-
ration of the gene into a vector; (3) transfer of the vector carrying the gene into
a recipient (host); (4) selection and cloning of the host cells carrying the gene;
(5) expression of the gene; and (6) extraction of the protein [5 I]. Each recombi-
nant protein has its own distinct chemical, physical, and biochemical properties.
In selecting an expression system for the production of a recombinant pro-
tein, the potential use of the protein is a determining factor. The size, complexity,
posttranslational modifications, and ease of production and purity of that particu-
lar protein are considerations in selection. Research into phenotypical changes
occurring in host cells necessitates an expression system closely imitating a pro-
tein's natural source [17]. The four major cell types currently employed for pro-
tein production include bacterial cells, yeast, insect cells, and mammalian cells.
When choosing which expression system to use, conservation of the biological
activity (proper folding) of the expressed recombinant protein as well as isolation
(purification) of this protein should be considered [17,63].
Escherichia coli and Bacteroides subtilis are the two most widely used
bacterial cell systems. The advantages of bacterial expression systems .are (1)
their simplicity and (2) their rapid and large yields at low cost. Bacterial cells
can also secrete the recombinant protein into the culture medium facilitating puri-
fication. One disadvantage is that the proteins often fail to fold properly, forming
insoluble inclusion bodies of biologically inactive proteins. While some small
proteins extracted from these inclusion bodies can be successfully refolded, most
larger proteins cannot. Bacterial cells also lack necessary enzymes for posttransla-
tional modifications required for proper functioning of the protein [17,63]. E.
coli was the expression system used to produce the first commercial recombinant
protein drug, human insulin 1181.
E. coli expression systems offer the option of fusion protein expression
[1,32,59]. Fusion proteins are a combination of the gene product of interest (usu-
ally eukaryotic) and a second "carrier" gene product (usually prokaryotic)
[32,63]. The fusion helps to stabilize the protein in the bacteria. The fusion protein
expression strategy allows good translation initiation, overcomes instability, and
yields a high degree of expression. The fusion partner also facilitates generic
protein purification methods and helps prevent formation of inclusion bodies.
The four main gene fusion expression systems in E. coli include Staphylococcus
protein A (SPA), Schistosomajaponicum glutathione S-transferase (GST), E. coli
maltose-binding protein (MBP), and E. coli thioredoxin [32].
Yeast can be grown as rapidly and economically as bacteria. Yeast can
also achieve high cell densities and allow secretion into the culture medium,
thereby facilitating purification. An advantage of yeast over bacteria is ability
to perform many posttranslational modifications found on human proteins. One
potential disadvantage in using yeast cells is the presence of proteases that de-
grade foreign proteins. Proteolytic degradation can be prevented by constructing
yeast strains in which the genes for these proteases have been deleted 117,631.
A subunit vaccine for hepatitis B virus (HBV) using only a surface protein to
initiate the immune response has been successfully produced in yeast [61]. The
major advantage is lack of infection risk.
A third, relatively new expression system uses baculovirus to insert foreign
proteins into insect cells. Advantages are abundant expression, correct folding,
and posttranslational modifications similar to those in mammalian cells. One dis-
advantage is the higher cost of culturing insect cells over that of bacteria and yeast.
Cost is still less than that for culturing mammalian cells [63]. A vaccine for the
autoimmune deficiency syndrome (AIDS) virus produced from a recombinant HIV
envelope protein by this method has been approved for clinical evaluation [3].
Finally, mammalian cells have been found to be the best place to produce
mammalian proteins. Mammalian cell systems are used for both transient and
stable expression. Transient expression is used to verify the biological activity
of engineered proteins and rapidly evaluate different vectors for the stable expres-
sion systems. In contrast, stable mammalian cell expression is more time consum-
ing but is necessary for large-scale production of proteins in mammalian cells.
The versatility of protein production possible from mammalian cells is the reward
for time and effort spent on the processes of selection, amplification, and growth
of stable cell lines. Thus far, there are no examples of proteins that cannot be
made in mammalian cells [13,63]. Tissue plasminogen activator (tPA) was the
first commercially available drug produced from a stably integrated mammalian
cell culture [48]. Factor VIII, a protein required for normal clotting of blood, has
also been efficiently produced in a stably integrated mammalian cell culture [45].
If an expression system conserves the biological activity of the recombinant
protein, the next consideration is whether this biological activity will be further
conserved following purification of the protein [53]. Currently, purification of
recombinant proteins generally utilizes the same purification techniques em-
ployed for separating other proteins from cell culture media. The large number
of proteins produced by the biotechnology revolution have also driven advances
in analytical techniques. High-performance liquid chromatography (HPLC) and
other analytical scale chromatographic techniques are essential to the achieve-
ment of the objectives of this revolution.
The techniques classically used for the purification and characterization of
proteins are ultracentrifugation, size exclusion, gel filtration, ion exchange, affin-
ity chromatography, electrophoresis, and radioimmunoassay 16,461. However,
these methods are generally time consuming and difficult to automate. The advent
of HPLC brought a new dimension to the purification and analysis of proteins. The
major advance was the use of microparticulate packing materials, which permitted
the use of high pressure. The cost of moving from traditional liquid chromatography
(LC)methods to HPLC was more than offset by improvements in speed, resolution,
sensitivity of chromatographic separations, and ease of sample recovery [ I 11. Ad-
304 Smith et al.
vancements in biotechnology continue to be a major driving force for advances in

HPLC for purification and identification of proteins. HPLC can be used alone or in
conjunction with classical techniques for protein purification.
Many excellent reviews and monographs have appeared in the literature on
the general techniques used in the purification of proteins with many dedicated to
HPLC in particular [6,8,9,14,15,46,54,55,65].Most ofthe applications forrecombi-
nant proteins require high-purity separations (including large scale) and good ana-
lytical characterization of the resulting product. Traditionally, proteins and protein
pharmaceuticals were extracted from blood and tissue-difficult separations in
their own right. Recombinant protein technology has created a new mix from which
proteins need to be isolated and purified. The purification of the desired protein
from the cell broth requires separation from the host cell proteins and DNA and
from other components of the culture medium. Purity is of particular concern for
the pharmaceutical proteins because of the possibility of endotoxin contamination
from products expressed in E. coli or viral products from mammalian cell culture
[14,5 11. The complexity of the system created by the biological and biochemical
processes of recombinant DNA technology demands a robust purification strategy.
HPLC has a prominent role in the purification of recombinant proteins.
Final purification is generally achieved via a multistep strategy utilizing various
modes of HPLC. Since each recombinant protein has its own distinct chemical,
physical, and biochemical properties, the strategy for employing chromatographic
modes is unique for each protein. HPLC has great utility in the biotechnology in-
dustry because it offers high purity, is generally applicable for laboratory scale puri-
fication, and has the potential to be scaled to a preparative level [14,25,46,72].
HPLC is a dynamic tool as new pumps, column packings, and more sensi-
tive detectors continue to be produced [8,9]. The strength of HPLC also comes
from its varied physicochemical bases and the interactive way in which tech-
niques can be combined and manipulated to achieve separation and purification.
The most frequently used modes of chromatography are ion exchange, reversed-
phase (RP) or hydrophobic interaction, size exclusion or affinity chromatography.
Biological macromolecules differ physicochemically in size, shape, charge, hy-
drophobicity, and qualitative and quantitative organization of functional groups
within a three-dimensional structure. The various modes of HPLC utilize these
different properties as the basis for separation [46,54,55].
Hydrophobic interaction chromatography (HIC) separates on the basis of
surface hydrophobicity. Reversed-phase chromatography separates on the basis
of general hydrophobicity but generally uses harsher conditions than HIC. Size
and shape are the principles underlying size exclusion chromatography. Ion ex-
change chromatography separates based on charge discrimination. Specific inter-
actions facilitate separation by affinity chromatography. Affinity separations uti-
lize a wide variety of immobilized ligands including metals, enzyme substrate
analogs, and monoclonal antibodies for interaction with desired proteins.
Separation on columns by any HPLC mode can be manipulated by ad-

justing the appropriate mobile phase. Another dimension of the power of HPLC
comes from the various methods of detection. Optimized detection methods en-
hance the sensitivity of HPLC and provide on-line data for component identifica-
tion. Available detection methods include UV absorbance, scanning and diode
array, fluorescence, electrochemical, pulse amperometric detection (PAD), and
on-line mass spectrometry [8,9,46,54,55].
The purpose of this chapter is to provide an illustrative overview of HPLC
purification of recombinant proteins. An exhaustive summary is not feasible be-
cause of the large and ever growing number of recombinant proteins. The re-
viewed literature is classified by the type of protein purified.
II. PURIFICATION OF FUSION PROTEINS
A large number of different fusion proteins have been isolated using a variety
of HPLC columns and solvent systems (Table 1). The following are illustrative
as there are many more examples.
A. Extracellular Fusion Proteins

Recombinant human insulin has been isolated by a TSK G3000SW size exclusion
column [28]. In a study by Patrick and Lagu, oxidative sulfitolysis and two-
dimensional HPLC determined the recombinant human proinsulin fusion protein
produced in E. coli [47]. An anion exchange TSK DEAE-5PW column, a cation
exchange TSK SP-5PW column, and a size exclusion Zorbax GF-250 column
were used for the separation.
Several extracellular fusion proteins have been purified on the basis of hy-
drophobicity. Leukemia inhibitory factor (LIF) was initially expressed as a fusion
product with glutathione S-transferase and then purified on a glutathione-agarose
affinity matrix. After release from the matrix by cleavage with thrombin, LIF
was isolated using a 300-A Brownlee C8 RP column [16]. Monobromobimane
has been used to resolve two recombinant proteins by RP-HPLC based on their
cysteine content. O'Keefe et al. used a Hy-Tach nonporous C 18 column to isolate
transforming growth factor a Pseudomot~asaeruginosa exotoxin A 40 (TGFa-
PE40), a recombinant protein synthesized in E. coli, and PE40, the C-terminal
fragment of TGFa-PE40 [44]. PE40 has a molecular weight (M,) of approxi-
mately 40,000 and copurifies with TGFa-PE40. Finally, isoforms of Bet v I, the
major birch pollen allergen, have been analyzed by HPLC. mass spectrometry
(MS), and cDNA cloning. A Waters pBondapak C- 18 column was used to sepa-
rate proteolytic fragments of purified natural Bet v'l and recombinant nonfusion
Bet v l a [58].
Table 1 Purification of Fusion Proteins
Protein
classification Ref. Substance purified Column type Solvent system
Extracellular Klyushnichenko VE, Wulfson Insulin-containing proteins TSK G3000SW (300 X Reagents used = acetonitrile,
AN. Recombinant human insu- 7.5 mm) methanol, sodium phosphate,
lin-11. Size-exclusion HPLC phosphoric acid (all esp. pure),
of biotechnological precursors. water purified on Milli-Q, so-
Factqrs influencing retention dium dodecyl sulfate, guani-
and selectivity. Pure Appl dine hydrochloride.
Chem 1993;65(10):2265-2272.
Patrick JS, Lagu AL. Determina- Human proinsulin fusion protein TSK DEAE-5PW (anion Anion exchange-"Loading"
tion of recombinant human pro- (ChPI) expressed in recombi- exchange); TSK SP- mobile phase A = 20 mM
insulin fusion protein produced nant E. coli 5PW (cation ex- Tris in 7 M urea (pH 7.0).
in Escherichia coli using oxida- change); Zorbax GF- "Eluting" mobile phase B =
tive sulfitolysis and two-dimen- 250 (size exclusion) 20 mM Tris and 0.5 M sodium
sional HPLC. Anal Chem sulfate in 7 M urea (pH 7.0).
1992;64(5):507-5 11. Cation exchange-Mobile phase
A = 40 mM monobasic potas-
sium phosphate in 7 M urea
(pH 3.5). Mobile phase B =
0.5 M in sodium sulfate, 40
mM in monobasic potassium
phosphate, and 7 M in urea
with pH 3.5.
Size exclusion-Mobile A phases
used in ion exchange: (1) 20 V)
mM Tris in 7 M urea (pH 3
-
5
7.0) & (2) 40 mM monobasic s
potassium phosphate in 7 M (P
urea (pH 3.5). p
Gearing DP, et al. Production of Leukemia inhibitory factor (LIF) 300-A Brownlee C8 RP 5 min linear gradient to 40% ace-
I
leukemia inhibitory factor in [initially expressed as a fusion tonitrile in 0.1% TFA followed w
Escherichia coli by a novel
procedure and its use in main-
product with glutathione S-
transferase, purified on glutathi-
by a 60 min linear gradient to
60% acetonitrile in 0.1% TFA.
5
taining embryonic stem cells one-agarose affinity matrix and
released from the matrix by
2*
+
in culture. Biotechnology ii'
1989;7(11):1157-1161. cleavage with thrombin]
O'Keefe DO, et al. Use of mono- 1) transforming growth factor-a Hy-Tach nonporous CIS Linear gradient of 34-64% aceto-
3
bromobimane to resolve two re- Pseudomonas aeruginosa exo- nitrile in 0.1% TFA
combinant proteins by re- toxin A 40 (TGFa-PE4O). a re- 0,
versed-phase high-performance combinant protein synthesized
liquid chromatography based in E. coli
on their cysteine content. Chro- 2) PE40, an M, = 40,000 C-ter-
matography 1992;627(1-2): minal fragment of TGFa-
137-143. PE40, which copurifies with
-5'
TGFa-PE40.
Swoboda I, et al. Isofoms of Bet Proteolytic fragments of purified Waters pBondapak C18 Linear gradient of acetonitrile
v 1, the major birch pollen al- natural Bet v 1 (nBet v l ) and [solvent A, 0.1% (vlv) TFA in
lergen, analyzed by liquid chro- Recombinant nonfusion Bet v water; solvent B, 0.07% (vlv)
(n
matography, mass spectrome- l a (rBet v la) [Bet v 1 = ma- TFA in acetonitrile; 0-40% B
try, and cDNA cloning. J Biol jor allergen of birch pollen] in 120 min]
Chem 1995;270(6);2607-2613.
lntracellular Sharma SK, et al. Metal affinity Recombinant HIV-1 reverse tran- C4 (Vydac) RP. 300-A 0-70% gradient of 0.1% TFAI
chromatography of recombi- scriptase containing a human pore size, 4.6 mm X H,O to 0.1% TFAICAN in 40
nant HIV- 1 reverse tran- renin-cleavable metal binding 15 cm min
scriptase containing a human domain
renin cleavable metal binding
domain. Biotechnol Appl Bio-
chem 1991;14(1):69-81.
Iwakura M, et al. Dihydrofolate Chemically synthesized polypep- Inertsil-ODs 5-pm col- Equilibrated with 0.1% TFA in
reductase as a new "affinity tides of 5-44 amino acids. ex- umn (RP) 15% acetonitrile. Linear gradi-
handle." J Biochem (Tokyo) pressed in E. coli as fusion pro- ent of acetonitrile up to 50%.
1992;lll(I):37-45. teins which show dihydrofolate 0
4
reductase (DHFR) activity.
Table 1 Continued w
0
OD
Protein
Surface Welling GW, et al. Isolation of Integral membrane protein of Sen- Mono Q HR 5/5 24-min gradient from 0.15 M to
membrane Sendai virus F protein by dai virus (a paramyxovirus of 1.5 M sodium chloride in 0.02
anion-exchange high-perfor- mice): F (fusion protein. M, = M sodium phosphate (pH 7.2)
mance liquid chromatography 6500) containing 0.1% Triton X- 100.
in the presence of Triton X
100. J Chromatogr 1983:266:
629-632.
Welling GW, et al. Isolation of Integral membrane proteins of TSK 4000SW (gel filtra- TSK 4000 SW-isocratic elution
detergent-extracted Sendai vi- Sendai virus (a paramyxovirus tion); Mono Q HR 5/ with 50 mM sodium phos-
rus proteins by gel-filtration, of mice): HN (hemagglutinin- 5 (ion exchange); C1 phate, pH 6.5, containing 0.1%
ion-exchange and reversed- neuraminidase. M, = 68,000) (reversed-phase) SDS.
phase high-performance liquid and F (fusion protein. M, = Mono Q HR 5/5-24-min gradi-
chromatography and the effect 6500) ent from 20 mM Tris-HC I , pH
on immunological activity. J 7.8, containing 0. 1% Triton X-
Chromatogr 1984:297:101- 100 to 0.5 M sodium chloride
109. in the same buffer.
CI-25-min gradient of 25% ace-
tonitrile in water with 0.05%
TFA to 75% acetonitrile in wa-
ter with 0.05% TFA.
Welling GW. Purification strate- Integral membrane proteins of TSK 4OOOSW (size ex- TSK 4000SW-elution with 0.1 %
gies for Sendai virus mem- Sendai virus (a paramyxovirus clusion); Mono Q HR SDS in 50 mM sodium phos-
brane proteins. J Chromatogr of mice): HN (hemagglutinin- 5/5 (anion exchange); phate (pH 6.5).
1987:397:165-174. neuraminidase, M, = 68,000) TSK Chelate-5PW Mono Q HR 5/5-gradient from V)
and F (fusion protein, M, = (metal chelate affin- 0 to 0.5 M sodium chloride in 3
:i'
6500) ity); Phenyl 5PW-RP 20 mM Tris-HC1 (pH 7.8). con- s
(reversed-phase) taining 0.1% (w/w) decyl- ?!
PEG.
TSK Chelate-5PW--
FIRSTPROGRAM: gradient from 0
to 0.1 M glycine in 20 mM
Tris-HCI (pH 8.0), containing
0.5 M sodium chloride and
0.1 % decyl-PEG (buffer A).
[Equilibrarion-with huffer A.
Then loaded with 0.2 M zinc
chloride and equilibrated agam
with buffer A]
SECOND PROGRAM: Gradient
from buffer A to huffer B.
[Both buffers contained 0.2 M
sodium acetate (pH 7.0) and
0.1% (wlw) decyl-PEG. Also,
buffer A contained 0.5 M so-
dium chloride, and huffer B
contained 0.5 M ammonium
chloride.]
[Equilibration-with buffer A.
Then loaded with 0.2 M zinc
chloride in buffer A and 10 ml
of buffer B. Before sample ap-
plication, column reequili-
brated with buffer A].
Phenyl 5PW-RP-24-min gradi-
ent from 15% to 75% acetoni-
rile in water containing 0.05%
TFA.
Table 1 Continued
2
Protein 0
Welling-Wester S, et al. Effect of Integral membrane proteins of Two tandem-linked Su- Following eluents were used: 6
detergents on the structure of Sendai virus (a paramyxovirus perose 6HR 10130 M guanidine hydrochloride in
integral membrane proteins of of mice): HN (hemagglutinin- (300 X 10 mm id.) 50 mM sodium phosphate (pH
Sendai virus studied with size- neuraminidase, M, = 68,000) size exclusion col- 6.5); 0.25% deoxycholate in
exclusion high-performance liq- and F (fusion protein, M, = umns 10 mM sodium phosphate (pH
uid chromatography and mono- 6500) 8.1); 0.1% Brij 35 in 50 mM
clonal antibodies. J Chro- sodium phosphate (pH 6.5);
matogr 1988;443:255-266. 0.1% triethylamine (pH 3.0)
with 0.1% decyl polyethylene
glycol-300; 20% acetonitrile
in 50 mM sodium phosphate
(pH 6.5); 0.1% lauryldimethy-
lamine oxide in 50 mM so-
dium phosphate (pH 6.5);
0.25% taurocholate in 10 mM
sodium phosphate (pH 7.4);
0.03% Tween 20 in 50 mM so-
dium phosphate (pH 6.5).
0.1% decyl polyethylene gly-
col-300 in 50 mM sodium
phosphate (pH 6.5); 0.1% SDS
in 50 mM sodium phosphate
(pH 6.5); 0.25% CHAPS in
100 mM sodium phosphate
V,
(pH 6.5); 0.1 % octylglucoside
3
in 50 mM sodium phosphate =
(pH 6.5); and 0.05% sarkosyl
in 10 m~ T ~ ~ S - H( Cp~7.5)
~ %
supplemented with 0.6 M so-
diurn chloride.
Van Ede J, et al. Comparison of Integral membrane proteins of Mono Q HR 515 (anion Anion exchange-gradient from V
non-ionic detergents for extrac-
tion and ion-exchange high-per-
Sendai virus (a paramyxovirus
of mice): HN (hemagglutinin-
exchange); Zorbax Bi-
oseries GF 450 or
0 to 0.5 M sodium chloride in 5
20 mM Tris-HCI (pH 7.8) con-
formance liquid chromatogra- neuraminidase, M, = 68,000)
and F (fusion protein, M, =
TSK G4000SW (size taining 0.1% (wlw) of deter- 5.
6
phy of Sendai virus integral exclusion) gent. 5'
membrane proteins. J Chro- 6500) Size exclusion-mobile phase %
matogr l989;476:3 19-327. was 50 mM sodium phosphate
a
(pH 6.5) containing 0.1% SDS.
Welling-Wester S, et al. Compari- Integral membrane proteins of Ion erchange: Ion exchange-Linear 12-min 2.
son of ion-exchange high-per- Sendai virus (a paramyxovirus Mono Q HR 515 (80- gradient from 20 mM Tris-HC1
0
formance liquid chromatogra- of mice): HN (hernagglutinin- nm pores). TSK (pH 7.8), containing 0.1% de- 0
phy columns for purification neuraminidase, M, = 68,000) DEAE-NPR(nonpo- tergent, to 0.5 M sodium chlo- 3
of Sendai virus integral mem- and F (fusion protein, M, = rous); and Zorbaw ride in the same buffer. [Pre- 5
brane proteins. J Chromatogr
1989;1988;476:477-485.
6500) BioSeries SAX
(30-nm pores)
ceded by isocratic elution for 5
min]
5
*
V
Size exclusion: Size exclusion-elution with 50 7
Two Zorbax GF 450 mM sodium phosphate (pH 2

E.
(250 X 9.4 mm 6.5), containing 0.1% SDS. a
U)
i.d.) in tandem
Welling GW, et al. Comparison Integral membrane proteins of Ion exchange: Ion exchange-Linear gradient
of detergents for extraction Sendai virus (a paramyxovirus MA7Q (nonporous), from 20 mM Tris-HC1 (pH
and ion-exchange high-perfor- of mice): HN (hemagglutinin- Zorbax BioSeries 7.8) containing 0.1% C,&5-9
mance liquid chromatography neuraminidase, M, = 68,000) SAX (30-nm (decylPEG-300) to 0.5 M so-
of Sendai virus membrane pro- and F (fusion protein, M, = pores). MonoQ HR dium chloride in the same
teins. J Chromatogr 1992; 6500) 515 (80-nm pores), buffer. (Preceded by isocratic
599(1-2): 157-162. y d PL-SAX 4000 elution for 10 min).
A (400-nm pores) Size exclusion-elution with 50
Size exclusion: mM sodium phosphate (pH
Polyol Si-500 (100 6.5) containing 0.1 % SDS.
nm x 4.6 mm i.d.)
Table I Continued 2
Iu
Protein
Welling-Wester S, et al. Effect of Hemagglutinin-neuraminidase Ion exchange: Ion exchange-Linear gradient

different amounts of the non- protein HN (M, = 68.000) and Mono Q HR 5/5 from 20 mM Tris-HCI (pH
ionic detergents C I OE5 and fusion protein F (M, = ("classical sys- 7.8) containing different deter-
C 12E5 present in eluents for 65,000) [2 integral membrane tem"); Mono Q PC gent concentrations
ion-exchange high-perfor- proteins of Sendai virus] 1.6/5 (Smart (buffer A) to 0.5 M sodium
mance liquid chromatography system) chloride in the same buffer
of integral membrane proteins Size exclusion: (buffer B). (Preceded by iso-
of Sendai virus. J. Chromatogr Superose 6 HR 10/30 cratic elution for 8 min)
1993:646(1):37-44. Size exclusion--elution with 50
mM sodium phosphate (pH
6.5) containing 0. I % SDS.
Folena-Wasserman G. et al. Recombinant protein R32Leu- 1.D; Brownlee C4 300- Equilibrated with 0.05% aqueous
Assay, purification and charac- Arg [32-tetrapeptide sequence A, 7-pm precolumn TFA. Linear gradient of O-
terization of a recombinant ma- from the immunodominant re- coupled to a Vydac 40% acetonitrile containing
laria circumsporozoite fusion peat region of the malaria cir- C4 300-.& 5-pm 0.05% TFA over 12 min.
protein by high performance cumsporozoite protein of Plus- column
liquid chromatography. J Chro- modium jalciparum (R32)
matogr 1987;411:345-354, linked to the dipeptide Leu-
Argl
Mizuochi T, et al. Structural char- N-linked oligosaccharides of the Ricinus communis agglu- Phosphate-buffered saline pH 7.4
acterization by chromato- HIV recombinant envelope gly- tinin (RCA 120) col- (last fraction eluted with 0.05
graphic profiling of the oligo- coprotein gp 120 produced in umn (affinity) M lactose).
saccharides of human CHO cells
immunodeficiency virus (HIV)
recombinant envelope glyco-
protein gp120 produced in Chi-
nese hamster ovary cells. Bio-
med Chromatogr 1988;2(6):
260-270.
Nuclear DuBois GC. Rapid purification of Fusion products of the v-myb on- Waters C18 ~ B o n d a - Waters C18 ~Bondapackcolumn
bacterially expressed fusion cogene and 13-14 amino acids pack column; Beck- and Beckman C3 column-
proteins by high-performance of the X I 1 gene (CII-myb fu- man C3 Ultrapore Equilibrated with 0.1% TFA
liquid chromatography meth- sion protein); fusion proteins RPSC protein separa- acid in water. Linear gradient
ods. Gene Anal Tech 1986; from HTLV-I Px with LC11 tion; Waters Protein- of 0-80% acetonitrile, which
3(1):6-11. (CII-HTLV-Px fusion protein) PAK DEAE 5PW also contained 0.1% TFA. Wa-
ters Proetin-PAK DEAE 5PW.
Linear gradient of 0-0.3 M so-
dium acetate In dialysis buffer;
Linear gradient of 0-0.5 M
NaCl in a dialysis buffer.
Other Greve KF, et al. Liquid chromato- CTLA4lg (an immunoglobulin fu- Tosohaas TSK gel col- Tosohaas TSK gel column-Mo-
graphic and capillary electro- sion protein) umn: Bakerbond Abx bile phase was aqueous solu-
phoretic examination of intact "mixed-mode" ion tion of 0.1 M sodlum sulfate,
and degraded fusion protein exchange column 0.1 M potassium phosphate,
CTLA4Ig and kinetics of con- monobasic, and 0.05% sodium
formational transition. J Chro- azide (pH 6.7).
matogr A 1996;723(2):273- Bakerbond Abx "mixed-mode"
284. ion-exchange column-Gradi-
ent from 0% B [B contained
0.65 M sodium acetate, 20%
methanol (pH 7.0)] to 20% B
in 0.025 M MES (pH 5.5).
Smith et al.
B. lntracellular Fusion Proteins

Metal affinity chromatography has isolated a recombinant HIV-1 reverse tran-
scriptase containing a human renin-cleavable metal-binding domain. The separa-
tion used a C4 (Vydac) RP, 300 pore size, 4.6 mm X 15 cm column [57].
Chemically synthesized polypeptides of 5-44 amino acids in length were ex-
pressed in E. coli as fusion proteins that showed dihydrofolate reductase activity.
These polypeptides were separated using an Inertsil-ODs 5 - ~ m column [24].
C. Surface Membrane Fusion Proteins

Many surface membrane fusion proteins have been separated by HPLC. Eight
publications on the HPLC purification of two different Sendai virus integral mem-
brane proteins [hemagglutinin-neuraminidase (HN) and fusion protein (F)] are
reviewed in chronological order to illustrate the progression in the technology
over the 10-year period from 1983 to 1993. Sendai virus F protein was isolated
by a Mono Q HR 515 anion exchange column in the presence of Triton X-100
[67]. Sendai virus is a paramyxovirus in mice. These authors published a second
study in which they isolated detergent-extracted Sendai virus proteins by gel
filtration, ion exchange, and RP-HPLC [66]. Three years later, this group summa-
rized their findings on purification strategies for Sendai virus membrane proteins
[68]. Integral membrane proteins of Sendai virus, HN (M, = 68,000) and F
(M, = 6500), were separated by a multistep process using the following columns:
TSK 4000SW (size exclusion); Mono Q HR 515 (anion exchange); TSK Chelate-
5PW (metal chelate affinity); and Phenyl 5PW-RP. In 1988, Welling-Wester et
al. examined the effect of detergents on the structure of the integral membrane
proteins of Sendai virus [70]. Size exclusion HPLC and monoclonal antibodies
were used by employing two tandem-linked Superose 6HR 10130 size exclusion
columns. These authors published two more studies in 1989. The first compared
the use of nonionic detergents for extraction and ion exchange HPLC of Sendai
virus integral membrane proteins [62]. This study initially used anion exchange
HPLC on a Mono Q column to isolate the Sendai virus integral membrane pro-
teins. The gradient for this step was from 0 to 0.5 M sodium chloride in 20 mM
Tris-HC1 (pH 7.8) containing 0.1 % (wlw) sodium dodecyl sulfate (SDS). Protein
recoveries after the anion exchange HPLC step were determined by size exchange
HPLC on either a TSK 400SW or two tandem-linked Zorbax GF 450 columns.
The mobile phase for this step was 50 mM sodium phosphate (pH 6.5) containing
0.1 % SDS. The second study compared different ion exchange HPLC columns
for the purification of Sendai virus integral membrane proteins [69]. In this study,
the highest recovery of HN protein was obtained by using either a Mono Q col-
umn or a TSK DEAE-NPR column, whereas the highest recovery of F protein
was obtained after chromatography on the Mono Q column. In 1992, Welling et
al. compared detergents for the extraction and ion exchange HPLC of Sendai
virus integral membrane proteins [64]. Extracted proteins were subjected to ion
exchange HPLC using four different columns (MA7Q, Zorbax BioSeries SAX,
Mono Q, and PL-SAX). The amount of protein in the extracts and fractions after
anion exchange HPLC was determined by size exclusion HPLC on a Polyol Si-
500 column, and the relative recoveries of protein were similar for all four anion
exchange columns. A final study on the purification of Sendai virus proteins by
this group examined the effect of different amounts of the nonionic detergents
ClOE5 and C12E5 in the eluents from Mono Q HR 515 ("classical system")
and Mono Q PC 1.615 (Smart system) ion exchange columns [71]. They found
that a detergent concentration of less than 0.026-0.05% allows the integral mem-
brane proteins of Sendai virus to remain on the column. The authors used size
exclusion HPLC to determine the yield from both classical and Smart anion ex-
change HPLC systems and found that the Smart system had the advantage that
only one-tenth of the sample amount needed for classical HPLC is required to
obtain comparable results. A disadvantage of the Smart system was the relatively
long time required for equilibrium of the column with different detergents.
A tandem column procedure using an I.D. Brownlee C4 300-A 7-pm pre-
column coupled to a Vydac C4 300-A 5-pm column isolated a recombinant ma-
laria circumsporozoite fusion protein [13]. Specifically, the isolated recombinant
protein R32Leu-Arg was a 32-tetrapeptide sequence from the immunodominant
repeat region of the malaria circumsporozoite protein of Plasmodium falciparum
(R32) linked to the dipeptide Leu-kg. An affinity column containing Ricinus
cummunis agglutin (RCA 120) was used to purify N-linked oligosaccharides of
the human immunodeficiency virus recombinant envelope glycoprotein gp120
produced in Chinese hamster ovary (CHO) cells [39].
D. Nuclear Fusion and Other Fusion Proteins

Nuclear fusion proteins have also been purified using HPLC. DuBois has reported
the rapid purification of bacterially expressed fusion products of the following:
v-myb oncogene, 13-14 amino acids of the LC11 gene (CII-myb fusion protein),
and fusion proteins from HTLV-1 Px with hCII (CII-HTLV-Px fusion protein)
[lo]. In a recent study by Greve et al., an immunoglobulin fusion protein,
CTLA4Ig, was isolated with a Tosohaas TSK gel column and Bakerbond Abx
"mixed-mode" ion exchange column [19].
Ill. PURIFICATION OF MALTOSE-BINDING PROTEINS
A variety of maltose-binding proteins (MBPs) have been purified using HPLC

techniques (Table 2). Illustrative examples of extracellular, intracellular (cyto-
plasmic), and surface membrane MBPs purified by HPLC are discussed by pro-
tein classification.
Table 2 Purification of Maltose-Binding Proteins
Protein
Extracellular Fassina, Giorgio, et al. High human endothelin ( 1-2 1) re- ABI Aquapore 30 X 2.1 Linear acetonitrile gradient
yield expression and purifica- leased from human big endo- mm i.d. RP; ABI (0.1% TFA) from 3% to 60%
tion of human endothelin-l . thelin (1-37) Aquapore RP-8 HPLC
Protein Expression Purif 1994;
5(6):559-568.
Hiraoka. Osamu, et al. Forma- mCRH-G-CSF complex [a cyto- TSK gel G3000SW (gel Equilibrated with 20 rnM so-
tion of l : l complex of the cy- kine receptor homologous filtration) dium phosphate buffer, pH
tokine receptor homologous re- (CRH) region of the murine 7.0, containing 0.2 M NaC1.
gion of granulocyte colony- granulocyte colony-stimulating Eluted with the same solution.
stimulating factor receptor factor (G-CSF) receptor ex-
with ligand. Biosci Biotechnol pressed by an E. coli maltose-
Biochem 1995;59(12):2351- binding protein (MBP) fusion
2354. system.]
Intracellulnr Hoener zu Bentrup, Kerstin, et Fragments generated by cleaving Vydac C4 Acetonitrile gradient (0-70%) in
al. Maltose transport in Aero- a periplasmic maltose-binding 0- 1% TFA
rnonas hydrophila: purifica- protein from Aeromonas h!-
tion, biochemical characteriza- drophila with CNBr
tion and partial protein
sequence analysis of a peri-
plasmic maltose-binding pro-
tein. Microbiology (Reading,
U. K.) 1994;140(4):945-951.
I
Kim. Do Hyung, et al. Expres- HIV- I protease Mono S HR/5 Equilibrated with the Mono S 'CI
sion and purification of HIV-1 equilibration buffer (50 mM
protease utilizing a maltose MES, 1 mM EDTA, 1 mM D
binding protein. Mol Cells
1994;4(1):79-84.
Dm, 10% glycerol, pH 6.2).
NaCl gradient (0-0.1 M for 5
5.
min, and 0.1 -0.4 M for 30 nl
e.
min). 0
3
Martinez, Aurora, et al. Expres- Fusion protein MBP-(FxJhPAH HiLoad Superdex 200 HR Mobile phase consisted of 20
sion of recombinant human [a recombinam human phenyl- (size exclusion) mM Na-Hepes and 0.2 M 0,
phenylalanine hydroxylase as alanine hydroxylase (hPAH) NaC1, pH 7.0 a
fusion protein in Escherichia fused through the target se-
W
coli circumvents proteolytic quences of the restriction pro-
degradation by host cell prote- tease factor Xa to the C-termi-
ases. Isolation and character- nal end of E. coli MBP]
ization of the wild-type en-
zyme. Biochem J 1995;306(2):
589-597. ;c
Surface Quadri, Luis EN, et al. Charac- Camobacteriocin B2 immunity C8-Vydac (10 X 250 Gradient from 38.5% to 45.5%
!?.
3
membrane terization of the protein confer- protein (CbiB2) mm, 10-ym particle of acetonitrile in 0.1 % TFA cn
ring immunity to the antimicro- size, 3 0 0 . ~pore size)
bial peptide carnobacteriocin
B2 and expression of carno-
bacteriocins B2 and BMI. J
Bacterial 1995;177(5):1144-
1151.
Smith et al.
A. Extracellular Maltose-Binding Proteins

Human endothelin (1 -21) has been purified by first employing an ABI Aquapore
30 X 2.1 mm i.d. RP column using a linear acetonitrile gradient [O. 1% trifluoro-
acetyl acid (TFA)] from 3% to 60%. The second step utilized an ABI Aquapore
RP-8 column and the same solvent system and gradient 11 21. A second extracellu-
lar MBP, the mCRH-G-CSF complex, has been isolated [20]. This protein is a
cytokine receptor homologous (CRH) region of the murine granulocyte colony-
stimulating factor receptor expressed by an E. coli MBP fusion system. Gel filtra-
tion separation was performed using a TSK gel G3000SW column equilibrated
with 20 mM sodium phosphate buffer, pH 7.0, containing 0.2 M NaCl. The same
solution was used for elution.
B. lntracellular (Cytoplasmic) Maltose-Binding Proteins

Fragments generated by cleaving a periplasmic MBP from Aeromonas hydrophila
with CNBr have been separated using a Vydac C4 RP column and an acetonitrile
gradient (0-70%) in 0- 1% TFA [2 11. A Mono S HRl5 column separated HIV- 1
protease using an MBP [27]. The column was equilibrated with Mono S equilibra-
tion buffer and the protein was eluted with an NaCl gradient. The fusion protein
resulting from recombinant human phenylalanine hydroxylase (hPAH) fused to
the c-terminal end of E. coli MBP was purified using a HiLoad Superdex 200
HR size exclusion column [37]. The mobile phase consisted of 20 mM Na-Hepes
and 0.2 M NaCl, pH 7.0.
C. Surface Membrane Proteins

Quadri et al. characterized the protein that confers immunity to the antimicrobial
peptide carnobacteriocin B2 and expression of carnobacteriocins B2 and BM1
[50]. In this study, the surface membrane carnobacteriocin B2 immunity protein
was eluted from a C8 Vydac column using a gradient from 38.5% to 45.5% of
acetonitrile in 0.1 % TFA.
IV. PURIFICATION OF OTHER RECOMBINANT PROTEINS

A. Extracellular Proteins
The hyperproduction of polyhedrin insulin-like growth factor 2 (IGF-2) fusion
protein has been achieved in silkworm larvae infected with recombinant Bombyx
mori nuclear polyhedrosis virus [38]. A site for cleavage by CNBr was introduced
in the IGF-2 gene, allowing the IGF-2 protein to be released from the polyhedrin
fusion protein by CNBr treatment and purified by ion exchange chromatography
and HPLC (Table 3). A modified C8 RP-HPLC separation technique was used
I
w
Table 3 Purification of Other Recombinant Proteins E
Protein
w
classification Ref. Substance purified Column type Solvent system s.2
Extracellular Marumoto, Yasumasa, et al. Insulin-like growth factor 2 Not given Acetonitrile gradient in 0.1% %.
0
Hyperproduction of polyhe- (IGF-2) (released from polyhe- TFA a
drin-IGF I1 fusion protein in drin-IGF2 fusion protein pro-
0,
silkworm larvae infected with duced in silkworm larvae in- I
I
recombinant B0mb.y~mori nu- fected with recombinant !?
clear polyhedrosis virus. J Bombvx mori nuclear polyhe-
Gen Virol 1987;68(10):2599- drosis virus)
2606.
e
a
D
Hummel, Michael, et a]. Gene TrpEIIGF fusion protein (immu- C8 RP Gradient of 2-propanol in for- =I
rr
synthesis, expression in Esche- noreactive human insulin-like mic acid created within 30
richia coli and purification of growth factors 1 and 2 fused min 9
o
rr
immunoreactive human insu- to the 300 N-terminal amino 9.
-.
lin-like growth factors I and acids of the E. coli trpE gene) =I
U)
11. Application of a modified
HPLC separation technique
for hydrophobic proteins. Eur
J Biochem 1989;l80(3):555-
561.
Mueller, Sahine, et al. The for- Purified protein isolate (in C4-Dynamax 300 A; Hy- Mobile phase was 0.1% TFA in
mation of diselenide bridges which selenocysteine suhsti- tach peptide analytical water (A) and 0.75% TFA in
in proteins by incorporation tuted for cysteine residues) ex- acetonitrile (B).
of selenocysteine residues: pressed from recombinant E
biosynthesis and characteriza- coli grown on a medium con-
tion of (Se)2-thioredoxin. Bio- taining selenocysteine
chemistry 1994;33(11):3404-
34 12.
Table 3 Continued
0
Protein 0
h)
Vakharia, Vikram N.. et al. Syn- Pheromone biosynthesis-activat- 4.6-mm-diam. Aquapore 4.6-mm-dian1.-gradient of ace-
thetic pheromone biosynthesis ing neuropeptide (PBAN) RP-300 column; 2.1-mm tonitrile rising from 10% at
activating neuropeptide gene gene product derived from ex- diam. Aquapore RP-300 0.67%/min, in 0.05 M so-
expressed in a haculovirus extracellular fraction of 5B 1-4 columns dium phosphate, pH 6.0;
pression system. Insect Bio- (an insect cell line) infected 2.1-mm-diam.
chem Mol Biol 1995;25(5): with vINV-4 (a recombinant -gradient of isopropanol rising
583-589. baculovirus) from 10% at O.5%/min in
0.1% TFA;
g r a d i e n t of acetonitrile in
0. I % heptafluorobutyric acid;
-gradient of acetonitrile in
0.1% TFA.
lntracellular Xue, Hong, et al. Purification of Bacillus suhrilis tRNATQgene Vydac C4-derivatized silica Eluted with 60 min linear gradi-
hyperexpressed Bacillus subti- product cloned in E. coli ent from buffer A (10 mM so-
lis tRNATrp cloned in Esche- dium phosphate, pH 5.5, 1 M
richia coli. J Chromatogr, Bio- sodium formate, 8 mM
med Appl 1993;613(2):247- MgC1,) to buffer B (10 mM
255. sodium phosphate: pH 5.5,
10% methanol), followed by
isocratic elution with 100%
buffer B for 20 min.
Yamagata, S., et al. Overexpres- Saccharomyces cerevisiae DEAE-SPW (ion ex- Not given
sion of the Sac-charomyces MET 17lMET25 gene product change); G3000SW (gel
cerevisiae MET17lMET25 O-acetylserine-O-acetylhomo- filtration)
gene in Escherichia coli and serine sulfiydrylase (OAS-
comparative characterization OAH) expressed in E. coli;
of the product with O-ace- OAS-OAH from yeast
tylserine-0-acetylhomoserine
sulfhydrylase of the yeast.
Appl Microbial Biotechnol
1994;42(1):92-99.
Nuclear Lillehoj, Erik P., et al. Virion- Tax (transregulatory) protein of Linear gradient of 0-1 00% aque-
associated transregulatory pro- HTLV- 1) from HUT- 102 ous acetonitrile/O.l% TFA
tein of human T-cell leukemia cells
virus type I. AIDS Res Hum
Retroviruses 1992;8(2):237-
244.
Surface Bhown, Ajit S., et al. Purifica- Gag gene products of avian type I2'I gel pem~eationcolumns Mixture of acetic acid/propanoll
membrane tion and characterization of C retroviruses highly purified water (20: 15:
the gag gene products of 65)
avian-type C retroviruses by
high-pressure liquid chroma-
tography. Anal Biochem
1981;112(1):128-134.
Kolbe, Hanno V.J., et al. Isola- Recombinant partial gag gene Sulfoethyl aspartamide; Sulfoethyl asparramide-Equili-
tion of recombinant partial product p18 (HIV- I,,) [= Nucleosil C4; Vydac brated with a blank gradient;
gag gene product p 18 (HIV- membrane-associated struc- 218TP54 C18 nonlinear gradient from 0 to
1Bru) from Escherichia coli. tural protein of HIV- I ] 100% Cat Ex-B buffer [40
J Chromatogr 1989;476:99- mM sodium phosphate (pH
112. 7.0)-1 M sodium chloride] in
CatEx-A buffer [20 mM so-
dium phosphate (pH 7.0)-40
mM sodium chloride] fol-
lowed by return to 100% Cat
Ex-A buffer.
Nucleosil C4-Nonlinear gradi-
ent from 90% eluent R1 -A
(0.1% TFA in Milli-Q water)
to 90% eluent RP I -B [0. 1%
TFA in acetonitrile-Milli-Q
water (70: 30. vlv)] and then
back to 90% RPI-A.
Vydac 218TP54 Cl8-Nonlin-
ear gradient from 99% eluent
RPI-A to 100% eluent RPI -B
and then back to 99% RPI-A.
322 Smith et al.
by Hummel et al. to purify immunoreactive human insulin-like growth factors 1

and 2 fused to 300 N-terminal amino acids of the E. coli trpE gene [22]. Using
recombinant E. coli grown on a medium containing selenocysteine, Mueller et al.
biosynthesized and characterized (Se)2-thioredoxin by forming diselenide brid-
ges in proteins that had incorporated selenocysteine residues [40]. Protein isolates
in which selenocysteine substituted for cysteine residues were purified with a C4
Dynamax 300-A column. Vakharia et al. used a baculovirus expression system
to express a neuropeptide gene that activates synthetic pheromone biosynthesis
[35,60]. Aquapore RP-300 columns were used to isolate the neuropeptide gene
product.
B. lntracellular Proteins
Xue et al. hyperexpressed the gene product from Bacillus subtilis tRNATQcloned
in E. coli [73]. Vydac CCderivatized silica was used to isolate the gene product.
Overexpression of the Saccharomyces cerevisiae MET17lMET25 gene in,E. coli
was conducted by Yamagata et al. [74]. Following overexpression, the gene prod-
uct was characterized and compared with 0-acetylserine-0-acetylhomoserine
sulfhydrylase from yeast. Both ion exchange and gel filtration HPLC were per-
formed on the two proteins to determine behavioral differences. Results implied
that no differences in molecular size and electric charge exist. It should be noted
that in this case HPLC was used for characterization instead of purification.
C. Surface Membrane Proteins

The gag gene products of avian-type C retroviruses have been purified and char-
acterized using Iz5I gel permeation columns [4]. A 20: 15:65 mixture of acetic
acid, propanol, and highly purified water served as the solvent system. A recombi-
nant partial gag gene product p18 that is a membrane-associated structural protein
of HIV-1 was isolated from E. coli [30]. The complicated separation procedure
required the use of sulfoethyl aspartamide, a Nucleosil C4 column, and a Vydac
218TP54 C18 column.
D. Nuclear Proteins
HUT-102 cells were engineered to produce a virion-associated transregulatory
protein of human T-cell leukemia virus type 1 [33]. A C4 RP column and a linear
gradient of 0-100% aqueous acetonitrile with 0.1% TFA was used to isolate the
transregulatory protein.
HPLC Purification of Recombinant Proteins
V. DISCUSSION
Since the advent of recombinant DNA techniques in the 1970s [5 11, a great num-
ber of different proteins have been produced in a variety of expression systems
[63]. As biotechnology progresses, molecular biologists are producing an ever
greater variety and quantity of recombinant proteins. As the examples in Tables
1-3 illustrate, the purification of recombinant proteins is accomplished using
standard protein purification techniques that rely on the ability to separate pro-
teins on the basis of molecular size, electrical charge, hydrophobicity, or affinity
for a specific monoclonal antibody. Examination of purification strategies for
different protein classifications, i.e., extracellular, intracellular, surface mem-
brane, and nuclear, suggests that there is no overall preferred method for any
given protein type. The studies reviewed show that combinations of ion exchange,
size exclusion, and RP-HPLC techniques have been used to purify each of these
protein categories. Two studies report the use of affinity HPLC to purify surface
membrane proteins [39,68]. The affinity method may be particularly suited for
surface membrane protein purification. This technique can exploit the unique
biological specificity of the protein-ligand interactions [ I I], which frequently
occur at the cell surface in in vivo systems, e.g., the interaction between a carbo-
hydrate moiety and a lectin [39]. The majority of methods for purifying surface
membrane proteins required a detergent in the solvent system. Detergents are
amphipathic molecules whose hydrophilic head and hydrophobic tail allow them
to compete with the phospholipids in the membrane lipid bilayer [36], thereby
freeing the membrane protein of interest from the bilayer [65]. Extraction of
membrane proteins with nonionic detergents, such as Triton X- 100, decylpolye-
thylene glycol (DecylPEG), and octylglucoside, preserves the biological activity
of the protein. Ionic detergents, such as SDS and naturally occurring deoxycho-
late and taurodeoxycholate bile salts, tend to irreversibly denature proteins. How-
ever, transmembrane protein hydrophobic regions have a greater affinity for this
type of detergent [65].
In many biological applications, retention of conformational structure and
the attendant biological activity is important. For example, because the biological
activity of human growth hormone (hGH) is preserved upon production in a bac-
terial expression system, recombinant techniques can be used to provide this hor-
mone for hGH-deficient children. Previously, the purification method involved
painstaking extraction of the hormone from the pituitary of human cadavers.
Studies by Nishi et al. and Knudtzon show that the therapeutic effects of pituitary
hGH are successfully duplicated by recombinant hGH [29,41].
Retention of the structure-activity relationship in various recombinant cy-
tokines has allowed the large-scale production of these molecules for patient ther-
apy. The biological activity of recombinant interferon-y (IFNy) has been shown
to supplement the activity of endogenous IFNy following injection into patients
324 Smith et al.
with malignant tumors [75], and IFNy has been successfully administered for the
treatment of chronic myeloproliferative syndromes and the prevention of chronic
granulomatous disease [52]. Also, recombinant granulocyte colony-stimulating
factor (G-CSF) and granulocyte-macrophage CSF (GM-CSF) stimulate granulo-
poiesis after chemotherapy or bone marrow transplantation and mobilize marrow
stem cells for peripheral blood stem cell transplantation.
Numerous interleukins retain their biological activity and therapeutic po-
tential when produced as recombinant proteins. Biologically active interleukin-
12 (IL-12) has been produced in vitro and in vivo by recombinant techniques
[5] and has been shown to provide effective therapy against tumors and infections
[7]. Recombinant human (rh) IL-11 has been used clinically to preserve the integ-
rity of gastrointestinal mucosa during cancer treatment regimens [49]. The cyto-
kine synthesis-inhibiting action of IL-I0 is duplicated in rhIL-10 and is therapeu-
tic for patients with myelomonocytic leukemia. Likewise, the pleiotropic bio-
logical activities of IL-6 on B cells, T cells, and hematopoietic progenitors are
preserved in recombinant IL-6 [26] and serve to inhibit advanced renal cell can-
cer [56]. Lastly. the activity of rhIL-2 provides effective inhibition of growth of
breast cancer [34].
Purifying fusion proteins whose synthetic sequences retain their original
conformational structures and attendant biological activities is another technical
challenge that promises advances in patient therapy. For example, antibody-
cytokine fusion proteins combine the unique targeting ability of antibodies with
the multifunctional activities of cytokines. Becker et al. produced antibody-IL-
2 fusion proteins by fusing a sequence coding for human IL-2 to genes encoding
antibodies [2]. Their use of antibodies targeted the IL-2 to the tumor site effec-
tively and eradicated human hepatic and pulmonary melanoma metastases in Se-
vere Combined Immune Deficiency (SCID) mice.
Immunotoxins, fusion proteins made up of a toxin and a monoclonal anti-
body, also offer an attractive approach to cancer therapy [63]. O'Boyle et al.
purified an immunotoxin combining a type 1 ribosome inactivating protein with
a murine monoclonal antibody reactive with a polymorphic determinant of class
2 HLA-DR histocompatibility leukocyte antigen (HLA) on human lymphoma
cells [42]. Such immunotoxins are advantageous because the antibody delivers
the toxin specifically to the target cell.
The use of mammalian cells for protein production has been favored be-
cause it is often difficult to retain the desired biological activity when therapeutic
proteins are produced in bacteria and yeast. Accordingly, recent efforts have fo-
cused on the development of numerous mammalian cell lines to host proteins
produced by recombinant techniques [43].
The field of chromatography is vital to recombinant technology. In particu-
lar, HPLC has played an important role in the purification of recombinant pro-
teins. Technological advances, e.g., the development of new detergents, will con-
tinue to provide new avenues for the use of HPLC.
REFERENCES
Balbas P, Bolivar F. Design and construction of expression plasmid vectors in Esche-

richiu coli. In Goeddel DV, ed. Methods in Enzymology. Vol 185. Gene Expression
Technology. San Diego: Academic Press, 1990.
Becker JC, Pancook JD, Gillies SD, Mendelsohn J, Reisfeld RA. Eradication of
human hepatic and pulmonary melanoma metastases in SCID mice by antibody-
interleukin 2 fusion proteins. Proc Nat Acad Sci USA. 1996;93(7):2702-2707.
Berman PW, Gregory TJ, Riddle L, Nakamura GR, Champe MA, Porter JP. Wurm
FM, Hershberg RD, Cobb EK, Eichberg JW. Protection of chimpanzees from infec-
tion by HIV-1 after vaccination with recombinant glycoprotein gp 120 but not bp
160. Nature 1990;345:622-625.
Bhown AS, Bennett JC, Mole JE, Hunter E. Purification and characterization of the
gag gene products of avhn-type C retroviruses by high-pressure liquid chromatogra-
phy. Anal Biochem 1981;112(1):128-34.
Bramson J, Hitt M, Gallichan WS, Rosenthal KL, Gauldie J, Graham FL. Construc-
tion of a double recombinant adenovirus vector expressing a hcterodimeric cytokine:
in vitro and in vivo production of biologically active interleukin-1 2. Hum Gene Ther
3996;7(3):333-342.
Chicz RM, Regnier FE. High-performance liquid chromatography: effective protein
purification by various chromatographic modes. In: Deutscher MP, ed. Methods in
Enzymology. Vol 182. Guide to Protein Purification. San Diego: Academic Press,
1990.
Chouaib S, Chehimi J. Human interleukin-12: biological role and therapeutic poten-
tial. Medicine/Sciences 1996;12(4):45 1-457.
Dorsey JG, Cooper WT, Siles BA, Foley JP, Barth HG. Liquid chromatography:
theory and methodology. Anal Chem 1996;68:5 15R-568R.
Dorsey JG, Cooper WT, Wheeler JF. Barth HG, Foley JP. Liquid chromatography:
theory and methodology. Anal Chem 1994;66(12):500R-546R.
DuBois GC. Rapid purification of bacterially expressed fusion proteins by high-
performance liquid chromatography methods. Gene Anal Tech 1986;3( 1):6- 1 I .
Ettre LS. Evolution of liquid chromatography: a historical overview. In: Horvath C.
ed. High-Performance Liquid Chromatography: Advances and Perspectives, Vol 1.
New York: Academic Press, 1980.
Fassina G, Merli S, Germani S, Ciliberto G, Cassani G. High yield expression and
purification of human endothelin-1 . Protein Expression Purif 1994;5(6):559-568.
Folena-Wasserman G, lnacker R, Rosenbloom J. Assay, purification and character-
ization of a recombinant malaria circumsporozoite fusion protein by high perfor-
mance liquid chromatography. J Chromatogr 1987;411:345-354.
Frenz J. Chromatographic separations in biotechnology. In: Horvath C, Ettre LS,
Smith et el.
eds. Chromatograpy in Biotechnology. Washington, DC: American Chemical Soci-

ety, 1993.
Frenz J, Hancock WS, Henzel WJ, Horvath C. Reversed phase chromatography in
analytical biotechnology of proteins. In: Gooding KM, Regnier FE, eds. HPLC of
Biological Macromolecules: Methods and Applications. New York: Marcel Dekker,
1990.
Gearing DP, Nicola NA, Metcalf D. Foote S, Willson TA, Gough NM, Williams
RL. Production of leukemia inhibitory factor in Escherichia coli by a novel proce-
dure and its use in maintaining embryonic stem cells in culture. Biotechnology 1989;
7(ll):ll57-ll6l.
Goeddel DV. Systems for heterologous gene expression. In: Goeddel DV, ed. Meth-
ods in Enzymology. Vol 185(1). Gene Expression Technology. San Diego, Aca-
demic Press, 1990.
Goeddel DV, Kleid DG, Bolivar F, Heyneker HL, Yansura DG, Crea R, Hirose T,
Kraszewski A, Itakura K, Riggs AD. Expression of chemically synthesized genes
for human insulin. Proc Natl Acad Sci USA 1979;76:106-110.
Greve KF, Hughes DE, Richberg P, Kats M, Karger BL. Liquid chromatographic and
capillary electrophoretic examination of intact and degraded fusion protein CmA4Ig
and kinetics of conformational transition. J Chromatogr A 1996;723(2):273-284.
Hiraoka 0, Hiroyuki A, Yoshimi 0. Formation of I : 1 complex of the cytokine
receptor homologous region of granulocyte colony-stimuIating factor receptor with
ligand. Biosci Biotechnol Biochem 1995;59(12):2351-2354.
Hoener zu Bentrup K, Schmid R, Schneider E. Maltose transport in Aeromonas
hydrophila: purification, biochemical characterization and partial protein sequence
analysis of a periplasmic maltose-binding protein. Microbiology (Reading, U. K.)
1994;l40(4):945-95 1.
Hummel M, Herbst H, Stein H. Gene synthesis, exression in Escherichia coli and
purification of immunoreactive human insulin-like growth factors I and 11. Applica-
tion of a modified HPLC separation technique for hydrophobic proteins. Eur J Bio-
chem 1989; 180(3):555-561.
Itakura K, Hirose T, Crea R, Riggs A, Heyneker HL, Bolivar F, Boyer H. Expression
in E. coli of a chemically synthesized gene for the hormone somatostatin. Science
l977;l98: 1056-1063.
Iwakura M, Tsuda K. Dihydrofolate reductase as a new "affinity handle." J Bio-
chem (Tokyo) 1992;111(1):37-45.
Jungbauer A. Preparative chromatography of biomolecules. J Chromatogr 1993;639:
3-16.
Keever-Taylor CA, Witt PL, Truitt RL, Ramanujam S, Borden EC, Ritch PS. Hema-
tologic and immunologic evaluation of recombinant human interleukin-6 in patients
with advanced malignant disease: evidence for monocyte activation. J Immunother
1996;19(3):231-243.
Kim DH, Lee J, Sung YC, Choi KY. Expression and purification of HIV-I protease
utilizing a maltose binding protein. Mol Cells 1994;4(1):79-84.
Klyushnichenko VE, Wulfson AN. Recombinant human insulin-11. Size-exclusion
HPLC of biotechnological precursors. Factors influencing retention and selectivity.
Pure Appl Chem l993;65(l O):2265-2272.
7
HPLC Purification of Recombinant Proteins
Knudtzon J. Growth hormone therapy of short stature. Acta Paediatr Scand 1986;
75(3):353-361.
Kolbe HVJ, Jaeger F, Lepage P, Roitsch C, Lacaud G, Kieny MP, Sabatie J, Brown
SW, Lecocq JP, Girard M. Isolation of recombinant partial gag gene product p18
(HIV-1 Bru) from Escherichia coli. J Chromatogr l989;476:99- 112.
Kornberg A. Why purify enzymes? In: Deutscher MP, ed. Methods in Enzymology.
Vol 182. Guide to Protein Purification. San Diego: Academic Press, 1990.
LaVallie ER, McCoy JM. Gene fusion expression systems in Escherichia coli. Curr
Opin Biotech 1995;6:501-506.'
Lillehoj EP, Alexander SS. Virion-associated trans-regulatory protein of human T-
cell leukemia virus type I. AIDS Res Hum Retroviruses 1992;8(2):237-244.
Liu DL, Yang MQ, Eberhardt J, Persson B. Repeated immunotherapy using intratu-
moural injection with recombinant interleukin-2 and tumour-infiltrating lymphocytes
inhibits growth of breast cancer and induces apoptosis of tumour cells. Cancer Lett
1996;103(2):131-136.
Luckow VA, Summers MD. Trends in the development of baculovirus expression
vectors [review]. Bio Technology 1988;6:47-55.
Martin DW. Membranes: In: Martin DW, Mayes PA, Rodwell VW and Granner DK,
eds. Harper's Review of Biochemistry. 20th ed. Los Altos, California: Lange, 1985.
Martinez A, Knappskog PM, Olafsdottir S, Doskeland AP, Eiken HG, Svebak RM,
Bozzini M, Apold J, Flatmark T. Expression of recombinant human phenylalanine
hydroxylase as fusion protein in Escherichia coli circumvents proteolytic degrada-
tion by host cell proteases. Isolation and characterization of the wild-type enzyme.
Biochem J 1995;306(2):589-597.
Marumoto Y, Sato Y, Fujiwara H, Sakano K, Saeki Y, Agata M, Furusaw M, Maeda
S. Hyperproduction of polyhedrin-IGF I1 fusion protein in silkworm larvae infected
with recombinant Bombyx mori nuclear polyhedrosis virus. J Gen Virol 1987;68(10):
2599-2606.
Mizuochi T, Spellman MW, Larkin M, Solomon J, Basa LJ, Feizi T. Structural
characterization by chromatographic profiling of the oligosaccarides of human im-
munodeficiency virus (HIV) recombinant envelope glycoprotein gp120 produced in
Chinese hamster ovary cells. Biomed Chromatogr 1988;2(6):260-270.
Mueller S, Senn H, Gsell B, Vetter W, Baron C, Boeck A. The formation of diselen-
ide bridges in proteins by incorporation of selenocysteine residues: biosynthesis and
characterization of (Se)2-thioredoxin. Biochemistry 1994;33(11):3404-3412.
Nishi Y, Masuda H, Nishimura S, Kihara M, Suwa S, Tachibana K, Takeda M,
Okada Y, Matsuda I. Isolated human growth hormone deficiency due to the hGH-I
gene deletion with (type IA) and without (the Israeli-type) hGH antibody formation
during hGH therapy. Acta Endocrinol 1990;122(2):267-271.
O'Boyle KP, Colletti D, Mazurek C, Wang Y, Ray SK, Diamond B, Rosenblum
MG, Epstein AL, Shochat D, Dutcher JP, Wiernik PH, Klein RS. Potentiation of
antiproliferative effects of monoclonal antibody Lym-1 and immunoconjugate Lym-
1-gelonin on human Burkitt's lymphoma cells with gamma-interferon and tumor
necrosis factor. J Immunother 1995;18(4):221-230.
Ogez JR, Builder SE. Downstream processing of proteins from mammalian cells.
Bioprocess Techno1 1990; lO:393-416.
328 Smith et al.
O'Keefe DO, Lee AL, Yamazaki S. Use of monobromobimane to resolve two re-
combinant proteins by reversed-phase high-performance liquid chromatography
based on their cysteine content. Chromatography 1992;627(1-2):137-143.
Paborsky LR, Fendly BM. Fisher KL, Lawn RM, Marks BJ, McCray G, Tate KM,
Vehar GA and Gorman CM. Mammalian cell transient expression of tissue factor
for the production of antigen. Protein Eng 1990;3:547-553.
Papadoyannis IN, ed. HPLC in Clinical Chemistry. New York: Marcel Dekker, 1990,
Chap. 19.
Patrick JS, Lagu AL. Determination of recombinant human proinsulin fusion protein
produced in Escherichiu coli using oxidative sulfitolysis and two-dimensional
HPLC. Anal Chem 1992;64(5):507-5 11.
Pennica D, Holmes WE, Kohr WJ, Harkins RN, Vehar GA, Ward CA, Bennett WF,
Yelverton E, Seeburg PH, Heyneker HL, Goeddel DV, Collen D. Cloning and ex-
pression of human tissue-type plasminogen activator cDNA in E. coli. Nature 1983;
301:214-221.
Peterson RL, Bozza MM, Dorner AJ. Interleukin-l l induces intestinal epithelial cell
growth arrest through effects on retinoblastoma protein phosphorylation. Am J Pa-
tho1 1996;149(3):895-902.
Quadri LEN, Sailer M, Terebiznik MR, Roy KL, Vederas JC, Stiles ME. character-
ization of the protein conferring immunity to the antimicrobial peptide carnobacteri-
ocin B2 and expression of carnobacteriocins B2 and BM1. J Bacteriol 1995;177(5):
1144-1151.
Richmond S, publisher. Scrip Biotechnology Made Simple, 4th ed. Richmond, Sur-
rey, UK: PJB Publications, 1991.
Robak T. Cytokiny w leczeniu chorob krwi (Cytokines in the treatment of blood
diseases). Acta Haematol Pol 1995;26(2 Suppl 1):72-78.
Sadana A. Inactivation of proteins and other biological macromolecules during chro-
matographic methods of bioseparation. Bioseparation 1992;3:145-165.
Schoneich C, Hiihmer AFR, Rabel SR, Stobaugh JF, Bigelow DJ, Williams
TD. Separation and analysis of peptides and proteins. Anal Chem 1995;67(12):
155R.
Schoneich C, Kivok SK, Wilson GS. Rabel SR, Stobaugh JF, Williams TD, Vander-
velde DG. Separation and analysis of peptides and proteins. Anal Chem 1993;65(12):
67R.
Schuler M, Peschel C, Schneller F, Fichtner J, Farber L, Huber C, Aulitzky WE.
Immunomodulatory and hematopoietic effects of recombinant human interleuhn-6
in patients with advanced renal cell cancer. J Interferon Cytokine Res 1996;16(11):
903-91 0.
Sharma SK, Evans DB, Vosters AF, McQuade TJ, Tarpley WG. Metal affinity chro-
matography of recombinant HIV-I reverse transcriptase containing a human renin
cleavable metal binding domain. Biotechnol Appl Biochem 1991;14(1):69-8 1.
Swoboda I, Jilek A, Ferreira F, Engel E, Hoffmann-Sommergruber K, Scheiner 0 ,
Kraft D, Breiteneder H, Pittenauer E. Isoforms of Bet v I, the major birch pollen
allergen, analyzed by liquid chromatography, mass spectrometry, and cDNA clon-
ing. J Biol Chem l995;270(6):2607-2613.
Uhlin M, Moks T. Gene fusions for purpose of expression: an introduction. In:
Goeddel DV, ed. Methods in Enzymology. Vol 185. Gene Expression Technology.
San Diego: Academic Press, 1990.
Vakharia VN, Raina AK, Kingan TG, Kempe TG. Synthetic pheromone biosynthesis
activating neuropeptide gene expressed in a baculovirus expression system. Insect
Biochem Mol Biol 1995;25(5):583-589.
Valenzuela P, Medina A, Rutter WJ, Ammerer G. Hall BD. Synthesis and assembly
of hepatitis B virus surface antigen particles in yeast. Nature 1982;298:347-350.
Van Ede J, Nijmeijer JRJ, Welling-Wester S, Oervell C and Welling GW. Compari-
son of non-ionic detergents for extraction and ion-exchange high-performance liquid
chromatography of Sendai virus integral membrane proteins. J Chromatogr 1989;
476~319-327.
Watson JD, Gilman M, Witkowski J, Zoller M. Recombinant DNA, 2nd ed. New
York: Scientific American Books, W.H. Freeman 1992.
Welling GW, Hiemstra Y, Feijlbrief M, Oervell C, Van Ede J, Welling-Wester S.
Comparison of detergents for extraction and ion-exchange high-performance liquid
chromatography of Sendai virus membrane proteins. J Chromatogr 1992;599(1-2):
157-162.
Welling GW, Van der Zee R, Weeling-Wester S. HPLC of membrane proteins. In:
Gooding KM, Regnier FE, eds. HPLC of Biological Macromolecules. New York:
Marcel Dekker, 1990.
Welling GW, Nijmeijer JRJ, Van der Zee R, Groen G, Wilterdink JB, Welling-
Wester S. Isolation of detergent-extracted Sendai virus proteins by gel-filtration, ion-
exchange and reversed-phase high-performance liquid chromatography and the ef-
fect on immunological activity. J Chromatogr 1984;297: 10 1 - 109.
Welling GW, Groen G, Welling-Wester S. Isolation of Sendai virus F protein by
anion-exchange high-performance liquid chromatography in the presence of Triton
X 100. J Chromatogr 1983;266:629-632.
Welling GW, Slopsema K, Welling-Wester S. Purification strategies for Sendai virus
membrane proteins. J. Chromatogr 1987;397:165- 174.
Welling-Wester S, Haring RM, Laurens H, Oervell C, Welling GW. Comparison
of ion-exchange high-performance liquid chromatography columns for purification
of Sendai virus integral membrane proteins. J Chromatogr 1989;476:477-485.
Welling-Wester S, Kazemier B, Oervell C, Welling GW. Effect of detergents on the
structure of integral membrane proteins of Sendai virus studied with size-exclusion
high-performance liquid chromatography and monoclonal antibodies. J Chromatogr
1988;443:255-266.
Welling-Wester S, Feijlbrief M, Koedijk DGAM, Braaksma MA, Douma BRK,
Welling GW. Effect of different amounts of the nonionic detergents C10E5 and
C12E5 present in eluents for ion-exchange high-performance liquid chromatography
of integral membrane proteins of Sendai virus. J Chromatogr 1993;646(1):37-44.
Wisniewski R. Principles of the design and operational considerations of large scale
high performance liquid chromatography (HPLC) systems for proteins and peptides
purification. Bioseparation 1992;3:77-143.
Xue H, Shen W, Wong JTF. Purification of hyperexpressed Bacillus subtilis
tRNATrp cloned in Escherichia coli. J Chromatogr Biomed Appl 1993;613(2):247-
255.
330 Smith et al.
74. Yamagata S, Isaji M, Nakamura K, Fujisaki S, Doi K, Bawden S, D'andrea R. Over-

expression of the Saccharomyces cerevisiae MET17lMET25 gene in Escherichia
coli and comparative characterization of the product with 0-acetylserine.0-acetylho-
moserine sulfhydrylase of the yeast. Appl Microbiol Biotechnol 1994;42(1):92-99.
75. Yamasaki T, Kagawa T, Takamura M, Moritake K. Pharmacokinetics of intratu-
moral interferon-gamma activity following subcutaneous administration of recombi-
nant interferon-gamma in a patient with metastatic brain tumor derived from renal
cancer. Neurol Surg 1995:23(2):169-173.
Isolation, Purification, and
Characterization of Human Seminal
Plasma Proteins and Their
immunological Behavior In Vitro
Afrozul Haq, Nona Remo Rama, and Sultan T. Al-Sedairy
King Faisal Specialist Hospital and Research Center, Riyadh,
Saudi Arabia
I. INTRODUCTION
Over the last few years, significant data have accumulated concerning compo-
nents present in the seminal plasma such as hormones, enzymes, peptides, and
cytokines [I-41. This progress has led us to the discovery of other components
in seminal plasma, not only in humans but in animals. We have achieved a greater
understanding of the functional properties of these components, especially semi-
nal plasma proteins in the male reproductive organs, its role in male fertility,
and its immunological effects in vitro. Still, the physiochemical and functional
properties of most of these proteins in seminal plasma remain unclear. Therefore,
further investigation should be established to determine the complexities of these
components present in seminal plasma and whether it has a role in the immunopa-
thogenesis of diseases in the male and female reproductive systems. The protein
makeup of seminal plasma appears to be extremely susceptible to variation on
account of large number of factors, such as species, age, frequency of semen
collection, the nature of ejaculates, the early ejaculates which show a protein
profile different from that of the later ejaculates, the hormonal status of the indi-
vidual, the condition of storage of semen, and possibly other factors such as diet,
weather, environmental and emotional factors including stress. Unlike the other
332 Haq et al.
body Ruids that are readily accessible for sampling and study, generation of se-
men, and therefore of seminal plasma, cannot be ordered about. The seminal
plasma of all species is rich in proteins but some of the proteins are not indigenous
to seminal plasma in that they may originate from serum. The proteins may also
leak into the seminal plasma from dead spermatozoa or spermatozoa damaged
during centrifugation or handling of the semen. Furthermore, some of the proteins
of seminal fluid may be bound to the spermatozoa immediately after ejaculation
and thus no longer be available in seminal plasma. The proteins in seminal plasma
are contributed from secretions of various male reproductive glands, e.g., epididy-
mis, seminal vesicles, prostate, and Cowper's and coagulating glands. The
remaining proteins found in the seminal plasma have been divided into the fol-
lowing groups: proteolytic enzymes; glycolytic enzymes; nucleolytic enzymes;
other enzymes; hormones and growth factors; antifertility factors; immunosup-
pressive factors; androgen-binding proteins; inhibin; immunoglobulins; and other
nonenzymatic, nonhormonal proteins.
II. PROTEINS
Proteins are fundamental components of all living cells and include many sub-
stances, such as enzymes, hormones, and antibodies, that are necessary for the
proper functioning of an organism. Proteins serve as the structural pattern of the
protoplasms from enzymes, hormones, chromosomes, and cell components. They
constitute about 15% of the protoplasms, are colloidal in nature, and are formed
of large molecules of great complexity and variety. These proteins are built up
by peptide linkages of amino acids with the peptide bonds between the amino
acid group (NH,) of one amino acid and the acid group (COOH) of the adjacent
one. They are made up of 20 amino acids as tripeptide and of several amino
acids as polypeptide. The human body produces thousands of different proteins,
including those that act as catalysts, those that are localized where they regulate
the flow of material in and out of the cell, those that are soluble or membrane-
bound, those that can interconvert different types of energy, those that recognize
foreign materials and microbes, those that are implicated in the repair of injury,
those that respond to biological stress, and those with purely structural roles.
Proteins can be structural (intracellular and extracellular proteins); catalytic (en-
zymes); vectoreal (transport proteins); regulatory (determine the rate at which
other proteins are made).
Any deficiencies and excess of these proteins can contribute to human dis-
eases. Significant studies have been done to determine chemical properties that
can be derived from the composition and distribution of amino acids. But still,
the relationship between the chemical compositiion, structure, and the function
of seminal plasma proteins is obscure and intriguing.
Isolation of Human Seminal Plasma Proteins 333
A. How Are Proteins Synthesized?

During the development of an organism, cells differentiate to perform specialized
functions. It starts with the replication of DNA, when the individual strands sepa-
rate resulting in two daughter strands of DNA that are identical to the original
DNA. Each daughter strand is synthesized in the direction of 5' to 3' requiring
one strand to be synthesized in a discontinuous fashion. DNA is transcribed into
RNA. The structures of DNA and RNA are similar in most respects. The only
difference is that in RNA the base uracil (U) replaces T, and U therefore base-
pairs with A; the other difference is that the sugar moiety of RNA is ribose instead
of deoxyribose. There are three major classes of RNA: messenger RNA (mRNA),
transport RNA (tRNA), and ribosomal RNA (rRNA). It is the mRNA that con-
tains sequences that are translated into protein. After the mRNA is formed in the
nucleus, it is transported to the cytoplasm where it is translated into protein whose
amino acid sequence is determined by the codons of the mRNA according to the
genetic code.
Ill. HUMAN SEMINAL PLASMA
Human seminal plasma (HSP) is a complex fluid mixture of numerous secretions

derived from various glands associated with the male reproductive tract, includ-
ing the testis whose primary exocrine function is the production of sperm and
whose secondary function is the production of the secretions that accompany
sperm; the epididymis whose main function is the absorption of fluid and the
addition of substances to the seminal fluid to nourish the maturing sperms; the
prostate, which produces a thin, milky fluid containing citric acid and acid phos-
phatase added to the seminal fluid at the time of ejaculation; the seminal vesicles,
which consist of a coiled tube and produce a secretion that is added to the seminal
fluid. Human seminal plasma composes approximately 80-90% of the volume
of normal ejaculate. It contains distinct protein components that are important in
the survival of the spermatozoa.
The long length of the duct of the epididymis provides storage space for
the spermatozoa and allows them to mature. The smooth muscle in the capsule
and stroma contract and the secretion from the many glands is squeezed into the
prostatic urethra. The prostatic secretion from the many glands is squeezed into
the prostatic urethra. The prostatic secretion is alkaline and helps to neutralize
the acidity of the vagina. The secretions contain substances that are essential
for the nourishment of the spermatozoa. At this point, the sperm are mixed with
the fluid from the seminal vesicles, the prostate, and bulbourethral glands and
become motile. This combined fluid is called semen, which flows through the
remaining urethra during ejaculation. The walls of the seminal vesicles contract
334 Haq et al.
during ejaculation and expel their contents into the ejaculatory ducts, thus wash-
ing the spermatozoa from the urethra.
A. Human Seminal Plasma Proteins

Seminal plasma contains different components necessary for the survival of the
spermatozoa. Several proteins have been identified in human seminal plasma;
some of these proteins are identical, as judged by electrophoretic mobility and
immunological methods, to blood plasma proteins such as albumin, globulins,
glycoproteins, acrosin inhibitor, and transferrins [5-81. However, many proteins
are specific to seminal plasma, and such proteins are of particular interest since
they could be used as markers of seminal stains in forensic medicine. A study
of the protein profiles of the seminal plasma of normospermic and vasectomized
men would also reveal the origin of the proteins [9]. The proteins of human
seminal plasma have been resolved by polyacrylamide gel electrophoresis
(PAGE), sodium dodecyl sulfate (SDS)-PAGE, isoelectrocfocusing, and normal
gel chromatographic and microdisck-gel electrophoresis methods. Human semi-
nal plasma has been resolved into about 20 protein bands by using SDS-PAGE,
with most of the bands lying in the low molecular weight range (10-20 kDa)
and a few bands of MW 32,49,68, and 75 kDa. No bands were observed above
80 kDa. Human seminal plasma contains membrane cofactor protein (MCP:
CD46) of 60,000 MW after gel filtration, which was associated with prostasomes
[lo]. It has been reported that prostasomes may play a complementary role to
other immunosuppressive factors contained in the human semen [ l 11. These fac-
tors may protect the sperm cells from the deleterious effects of phagocytosing
cells, prolong their life, and consequently enhance the chance of conception, and
at the same time possibly have a permissive effect on sexually transmitted dis-
eases. Prostasomes are trimellar to multimellar vesicles produced by the human
prostate and are present in appreciable amounts in normal human semen. These
substances may contribute to successful fertilization. A monoclonal antibody
(SEM-12) specific for human sperms was isolated [12]. Sperm surface glycopro-
teins may be involved in sperm-zona pellucida recognition. Some of these pro-
teins are of seminal plasma origin and their expression may change in the process
of capacitation and acrosome reaction.
Human seminal plasma was fractionated by affinity chromatography on
lentil lectin Sepharose gel chromatography on Ultrogel ACA 34 and immunoaf-
finity on Mab 456-coupled CNBr-Sepharose 4B [13]. The purified 4E6 antigen
consisted of three subunits with molecular weights of 70, 64, and 60 kDa. This
antigen is present in high amount in sera of infertile patients and therefore may
be involved in the pathogenesis of immunological infertility. Transforming
growth factor P (TGFP) was detected by Nocera and Chu [I41 from human semi-
nal plasma protein fractions with molecular weights of 100-440 kDa after gel
filtration. These proteins exhibit immunosuppressive activity and inhibit DNA

synthesis. T6Fa was then purified from human seminal plasma on Sephadex G-
75 chromatography and high-performance size exclusion liquid chromatography
[I 51. This growth factor showed mitotic activity. Another component of human
seminal plasma was studied as seminal plasma motility inhibitor (SPMI) [16].
This factor has the capacity to inhibit the movement of demembranated and intact
spermatozoa. SPMI decreases its biological activity rapidly in semen. Human
seminal plasma P-microseminoprotein was purified using DEAE-Sephacel and
zinc chelate Sepharose CL-6B column chromatography [17]. P-Microseminoami-
noprotein is a nonglycoprotein with a molecular weight of 19 or 17 kDa. An
acrosome reaction (AR)-inhibiting glycoprotein (ARIG) from human seminal
plasma that was isolated by differential centrifugation, chromatofocusing, and
Sephacryl S300 gel filtration was reported [4]. ARIG has a molecular weight of
74 kDa and inhibits sperm exocytosis. Its interaction with spermatozoa may be
mediated by carbohydrate-bindingproteins on the sperm cell. Moreover, fraction-
ated 59-kDa and 72-kDa proteins in human seminal plasma showed specific im-
munoreactivity and are documented to be the most frequently involved sperm
antigens in the immune response in fertile subjects [I 81. After ion exchange and
immunoaffinity chromatography of seminal plasma proteins in humans, it has
been demonstrated that the presence of kallikrein hK2 and Protein C inhibitor
(PCI) regulates its activity in seminal plasma [19]. Several workers documented
that seminal plasma TGFP may have some role to play in sexual transmission
of human immune deficiency virus (HIV) [14,20]. The authors' data demon-
strated the overexpression of TGFP in HIV-infected patients, suggesting that
TGFP is an important mediator in HIV infection. The presence of PC1 was de-
tected throughout the male reproductive tract in high concentration, about 200
pglml, in seminal plasma [21]. Their results suggest that PC1 might function as
a scavenger of prematurely activated acrosin, thereby protecting intact sur-
rounding cells and seminal plasma proteins from possible proteolytic damage.
Receptors for the Fc region of the immunoglobulin G (IgG) (soluble Fc y RIII)
have been recognized as a link between humoral and cellular responses [22]. Its
soluble form found in seminal plasma may modulate the immunosuppression of
antisperm immune responses in the male and female reproductive tracts.
The suppressive activity of the seminal plasma influences different cells of
the immune system. It can be assumed that more than one component of seminal
plasma is responsible for its immunosuppressive effect. It may be attributed to
the presence of zinc peptide complex in seminal plasma [23]. However, some
workers stated that it might be due to uteroglobulin and transglutaminase [24].
Fc receptor-binding protein has a vital role in terms of the immunosuppressive
effects of seminal plasma [25]. It may exert specific effects, e.g., receptors Fc
fractions of gamma globulin which might bind to inflammatory agents [26]. It
has also been reported that prostaglandin E (PGE) in seminal plasma predomi-
336 Haq et al.
nates and raises intracellular CAMP in leukocytes, which may contribute to the
immunosuppressive effects of seminal plasma. Immunosuppression may attribute
to the presence of a protein similar to pregnancy-associated Protein A [27]. The
presence of polyamines like spermin and spermidine has been reported and that
could also contribute to immunosuppression of seminal plasma [28]. Immunosup-
pression might also be due to the seminal nucleases and proteases in seminal
plasma [5]. Following is a brief list of important proteins of human seminal
plasma that have been studied extensively:
P-Microseminoprotein [29]
Sperm-binding proteins [30,3 11
Protein kinase inhibitor [32]
P-Endorphin [33]
Calcium-binding protein (calmodulin) [34]
Zinc-binding proteins [35]
Ion-binding proteins [36]
Placental proteins [37]
6. Seminal Plasma Proteins of Other Species

Several studies have been done on the isolation, purification, and characterization
of proteins in human seminal plasma. These studies were not limited to humans
but also included animals. Four major proteins of bovine seminal plasma-BSP-
A l , BSP-A2, BSP-A3, and BSP-30 kDa (collectively named BSP proteins)-
have been purified by affinity chromatography using p-aminophenylphosphoryl-
choline-agarose (PPC-agarose) matrix [38]. These proteins appear to be ubiqui-
tous in mammals and may possibly be involved in the modification of the lipid
content of the sperm plasma membrane. Furthermore, BSP-30 kDa protein was
studied, and it was found that this protein plays a role in sperm capacitation [39].
These workers also determined its amino acid sequence, disulfide bonds and
O-glycosylation sites. The mosaic structure of BSP-30 kDa suggests that this
glycoprotein might be a factor contributing to the different sperm capacitating
effects exerted by heparin in different mammalian species. A phosphodiesterase
was purified from bull seminal plasma by column chromatography on DEAE-
Sephadex A-50, ConA-agarose. chromatofocusing, and AMP-agarose [40]. This
purified enzyme is constituted from a single polypeptide chain of about 125 kDa.
Two major proteins from boar seminal plasma designated as PSP-I and PSP-I1
were purified and characterized [41]. CM-cellulose chromatography, gel filtration
on Sephadex G-75, and reversed-phase high-performance liquid chromatography
were used for complete purification. Recently, boar and stallion seminal plasma
were fractionated using affinity chromatography on heparin-Sepharose [42]. In
both species, among other proteins, the heparin-binding (H+) and non-heparin-
binding (H-) fractions each contained glycoforms of either porcine PSP-I or

equine HSP-1 and HSP-2. However, porcine H+/PSP-I eluted as a monomeric
protein, whereas H-/PSP-I formed a heterodimer with PSP-11, another major
seminal plasma protein. On the other hand, stallion proteins H+/HSP-1 and H/
HSP-2 eluted together as an aggregate of relative molecular mass (90 kDa),
whereas H/HSP-I and H/HSP-2 eluted as monomers (15 m a ) . Altogether these
data show that glycosylation has an indirect effect on the heparin binding ability
of PSP-I, HSP-I, and HSP-2 through modulation of their aggregation state.
Recently, a fertility-associated protein has been identified in bull seminal
plasma as lipocalin-type prostaglandin D synthase. Immunoreactive bands at 26
kDa appeared in western blots of seminal plasma and cauda epididymal fluid
(CEF) [43]. A 29-kDa band appeared in blots of rat testis fluid (RTF). Prostaglan-
din D synthase activity was detected in seminal plasma, cauda epididymal fluid,
and RTF. The amino acid sequence was 63-8070 identical to that of the enzyme
of other mammals.
IV. ISOLATION AND FRACTIONATION PROCESSES
Since proteins are really large molecules, in order to investigate their structure
and function, isolation and fractionation processes are required. There is no single
or simple way to purify all proteins. Procedures useful in the purification of one
protein may result in the denaturation of another (Fig. 1).
We were able to fractionate human seminal plasma using DEAE Sephadex
A-50 ion exchange columns [44]. Semen was obtained from healthy donors and
seminal plasma was recovered by ultracentrifugation. Seminal plasma was dia-
lyzed against PBS at 4°C overnight and then applied onto a DEAE-Sephadex
A-50 ion exchange column using different salt concentrations in phosphate buffer
pH 6.0 (Fig. 2). Forty men either fertile or under investigation for infertility
donated semen, which was collected by masturbation into sterile plastic contain-
ers after 2-3 days of abstinence from ejaculation. The mean age of the test sub-
jects was 34.2 -C 8 (20-45) years. Semen analysis was performed by using Cell-
soft Automated Semen Analyzer (Cryo Resources Ltd., New York). All patients
(normospermic 20-100 million sperms/ml) included in this study attended the
IVF and Infertility Clinics of King Faisal Specialist Hospital and Research Cen-
tre, Riyadh (Saudi Arabia). All of the semen samples were centrifuged at 10,000g
at 4°C for 15 min to remove sperm and the clear supernatant was stored at -70°C
until used.
The exchanger chosen was DEAE-Sephadex A-50 because it was reported
to provide better results than DEAE-cellulose in the purification of proteins [45].
Two hundred fifty milligrams (25 ml) protein from normospermic human seminal
plasma was applied to a XK 50130 (Pharmacia, Piscataway, NJ, USA) column
Haq et al.
I Human Seminal Plasma I
I Ultracentrifugatian I
I Protein Concentration I
( Chromatography I I 6
. ~~
1 Centrifueation of Fractions II
L
6
Dialysis of Fraclions
C
store at -70°C
I I I I
Figure 1 How chart for human seminal plasma proteins extraction.
- 75
TUBE NUMBER
100
Figure 2 Chromatogram of human seminal plasma proteins on DEAE-Sephadex A-50

column. Experimental conditions: column: XK 50130 (Pharmacia); sample: 250 mg pro-
tein (25 ml in phosphate buffer); eluent: phosphate buffer (PBS; pH 6) with increasing
salt concentration (0.01-1 M NaCI) [44].
Isolation of Human Seminal Plasma Proteins
Figure 3 SDS-PAGE of marker proteins and human seminal plasma and its fractions
in 12.5% gel. Lane I , the marker proteins (from top): phosphorylase b (97 kDa), bovine
serum albumin (66 kDa), ovalburnin (43 kDa), carbonic anhydrase (3 1 kDa), trypsin inhib-
itor (20 kDa), and lysozyme ( I 4 kDa). Lanes PI -P7 represent seminal plasma obtained
from DEAE-Sephadex A-50 ion exchange chromatography. Lane 2, the marker proteins
(hom top): myosin (200 kDa), P-galactosidase ( I 16 kDa), phosphorylase b (97 kDa), bo-
vine serum albumin (66 kDa), and ovalbumin (43 kDa) [44].
(5 X 15 cm) of DEAE-Sephadex A-50. The column was equilibrated with 0.05

M phosphate buffer (pH 6) containing 0.01 M NaCI. Fractions of 10 ml each
(40 mllh) were collected with increasing concentration of NaCl (0.01-1 M NaCI)
and read at 280 nm. Concentration of fractionated proteins was carried out by
using Centriprep 10 MW cutoff membranes (Amicon, Grace Co., Danvers,
MA, USA). All steps of protein purification and concentration were performed
at 4OC.
The complete elution profile of human seminal plasma proteins is shown
in Fig. 2. Complete separation was achieved in seven peaks. The elution pattern
of protein peaks and recovery was reproducible in different sets of experiments.
A total protein of 10.4 mg for peak 1, 61.7 mg for peak 2, 17.35 mg for peak
3, 20.6 mg for peak 4, 38.8 mg for peak 5, 12.5 mg for peak 6, and 6.2 mg for
340 Haq et al.
peak 7 was recovered after ion exchange chromatography. Of 250 mg protein

from pooled normospermic seminal plasma, a total of 67% protein was recovered
after ion exchange chromatography SDS-PAGE was then performed using 12.5%
gels [46]. Protein samples were mixed with 2X sample buffer and denatured for
5 min at 95°C. High and low molecular weight calibration kits (Bio-Rad, Rich-
mond, CA, USA) containing phosphorylase b (97 kDa), bovine serum albumin
(66 m a ) , ovalbumin (43 m a ) , carbonic anhydrase (31 m a ) , trypsin inhibitor
(20 kDa), lysozyme (12 kDa), myosin (200 m a ) , and P-galactosidase (1 16 kDa)
were used as marker proteins. Staining was done using R250 Coomassie brilliant
blue (0.1%). The results of SDS-PAGE are shown in Fig. 3.
V. CHARACTERIZATION OF A 20-KDA PROTEIN

FROM HUMAN SEMINAL PLASMA
Characterization of seminal plasma protein was carried out by using a fast protein
liquid chromatography (FPLC) system. Superose HR I0130 column (Pharmacia)
was used for the elution of the protein. A total of 100 ~1 (0.2 pglml) of protein
from seminal plasma was applied to the column [0.5 M phosphate buffer (KHPO,
+ Na2HP04)with 0.14 M NaCI]. The pH was adjusted to 7.3 by use of orthophos-
phoric acid. The buffer was degassed, filtered through 0.2-pm disposable filters,
and used as mobile phase. The flow rate was 0.2 mllmin and the fraction size
fixed at 0.5 ml. Ovalbumin (43 kDa) was used as a reference protein to confirm
the performance of the Superose column. These experiments of characterization
by FPLC and SDS-PAGE were carried out in the laboratory of Professor J. P.
Bouvet in the Institute Pasteur in Paris. The elution profile of a 20-kDa protein
as detected by FPLC is shown in Fig. 4. The results of SDS-PAGE are shown
in Fig. 5.
VI. IN VlVO IMMUNOLOGICAL EFFECTS OF SEMINAL

PLASMA
Proteins of human seminal plasma are known to have immunosuppressive proper-

ties. Studies on the immunosuppressive effects of seminal plasma in vivo are
scarce. The ability of seminal plasma to exert its immunosuppressive effect was
found to be dose-dependent, with smaller doses of antigen being most effective
[47]. Mouse seminal plasma also suppressed the in vivo antibody response to
bovine serum albumin. Studies have indicated that rectal infusion of prostaglan-
din E, or D2 into male rats reduces the in vitro response of T lymphocytes to
phytohemagglutinin (PHA) [48]. However, the T-cell response of female rats
Isolation of Human Seminal Plasma Proteins
START
Figure 4 Elution profile of a 20-kDa protein of human seminal plasma by using FPLC
(Superose column HR 10130); 0.5 M phosphate buffer (pH 7.3) with 0.14 M NaCl was
used.
treated in the same manner remained unchanged. This is an interesting finding

considering that prostaglandins are known to be present in seminal plasma and
both prostaglandin E, and D, suppress immune functions under in vitro condi-
tions.
A. In Vitro immunological Effects of Seminal Plasma

The immunosuppressive effects of human seminal plasma have been extensively
studied employing in vitro tests for immune functions. These studies have clearly
indicated that human seminal plasma can, either directly or indirectly, inhibit the
activities of most cells that participate in immune responses, such as T cells, B
cells, natural killer (NK) cells, macrophages, and polymorphonuclear leukocytes
(PMNs) [49]. In addition, human seminal plasma also impairs the activity of
Haq et al.
Figure 5 SDS-PAGE (12.5% gel) of marker proteins, whole human seminal plasma,
and a 20-kDa protein. Lane M represents protein markers (from top); phosphorylase b
(97 m a ) , bovine albumin (66 m a ) , ovalbumin (43 kDa), carbonic anhydrase (31 ma),
trypsin inhibitor (20 kDa), and lysozyme (14 kDa). Lane WS represents the whole human
seminal plasma and lane P represents the 20-kDa protein of human seminal plasma.
the C1 and C3 components of compliment. Recently, we studied the effect of

fractionated human seminal plasma on mixed lymphocte reaction (MLR) and
found a highly immunosuppressive effect on MLR using various mitogens [44].
This effect was dose-dependent and greatest with 10 mg protein per well. Peak
4 was most potent followed by peaks 7, 3, and 5. Peaks 1, 2, and 6 were found
to be stimulatory rather than suppressive. The whole seminal plasma was not
suppressive and this effect was more or less the same with 1 p g and 10 pg protein
per well. On the other hand, the stimulatory/suppressive effect of seminal plasma
on PMA-induced chemiluminescence was noted [50,51], and this could possibly
be due to interference with PMA binding to membrane receptors on PMNs. Based
on these findings, it is stated that the biological role of seminal plasma might be
to protect sperm cells with its antioxidant property from extracellular oxygen
radicals in the vagina during intercourse and from side effects of the activation
of macrophages and PMNs, namely, during mild infections. In this study, we
found that suppression or stimulation of chemiluminescence by human seminal

plasma is dose-dependent and has no correlation with the sperm density. More-
over, seminal plasma stimulatedlsuppressed chemiluminescence selectively and
therefore it may contain molecules of different characteristics [50].All of these
immunomodulatory effects of seminal plasma are important for successful repro-
duction because they help to reduce the immunogenicity of spermatozoa. Simulta-
neously, however, the immunosuppressive effects also endanger the host's de-
fense against infection and malignancy [49,52].
Immunosuppressive fraction was isolated from boar seminal vesicle secre-
tion by gel filtration on a Sephadex G-75 column and purified by reversed-phase
(RP)-HPLC, and resulted into three major peaks [53]. Immunosuppressive frac-
tion 3 was found to have inhibitory activity, estimated by inhibition of mitogen-
lymphocyte proliferation on porcine lymphocytes. There have been numerous
studies on the suppressive effect of seminal plasma proteins on T-cell activation
and proliferation. Factors that have been identified as contributing to this activity
include the prostasomes [54], polyamines [%I, prostaglandins of the E series
[11,56], and TGFP [14]. At present, little is known about the role of immunosup-
pressive factors in male immunological infertility, which accounts for up to 6%
of male infertility. It has also been demonstrated that the inability of spermatozoa
collected from infertile men to inhibit the lymphocytes of their female partners
was associated with the presence of antisperm antibodies in the serum of female
partners and a similar link between antibody formation and lymphocyte inhibition
in the male has been postulated [49,52,57,58]. In a recent study, an absence of
immunosuppressive activity was measured by the inhibition of graft rejection
response in vivo in the seminal plasma of men exhibiting immunological infertil-
ity. However, it remains to be established whether reduction in immunosuppres-
sive activity measured in seminal plasma is a contributing factor or a consequence
of sperm autoimmunity in the male. The polyamine spermine, naturally present
at millimolar levels in seminal plasma [59], inhibits proliferation of NK cells and
T lymphocytes, directly binds to DNA, and alters cervical cell ploidy, indicating
its potential contribution to the etiology of cervical cancer and its possible role
in suppression of the destruction of dysplastic and neoplastic cervical epithelial
cells by cytotoxic lymphocytes. This study demonstrated that spermine sup-
presses the sensitivity of cervical carcinoma cells to lymphokine-activated killer
(LAK) lymphocytes from more than half of normal individuals. Cervical carcinoma
cells are inherently very sensitive to LAK lymphocytes and spermine may be an
important immunosuppressive agent in natural immunity against cervical cancer.
B. Luminol-Dependent Chemiluminescence and

Phagocytic Activity
Although seminal plasma affects several aspects of the immune system, it is
believed that many of the inhibitory properties result from its effects on PMNs,
Haq et al.
SP Fractions
WA + I O O U Q I ~ I SP WA + 40ug1m1 SP PMA + I O U Q I ~ I SP
Figure 6 Effect of Seminal plasma (SP) and its fractions (P1 to P7) on phagocytic
response to phorbol myristate acetate (PMA). The x axis represents the whole SP and
its different fractions (Pl-P7), whereas the y axis represents the relative phagocytic
index [44].
monocytes, and other accessory cells. The phagocytic system is considered to be

the first line of defense against a variety of microorganisms that invade the host.
The bactericidal capacity of phagocytes might be impaired if seminal plasma
inhibited the production of reactive oxygen species (ROS) [50]. Effects of this
kind become more important when the infective microorganisms such as cyto-
megalovirus and HIV are contained in the ejaculate itself. Ingestion of microor-
ganisms by neutrophils is an active process that requires energy production by the
phagocytic cells. Subsequent intracellular events like degranulation and killing
depend on the success of ingestion. Phagocytosis denotes the ingestion phase of
the process, whereas the phagocytic index refers to the average number of parti-
cles ingested, which is considered as the measurement of ingestion rather than
phagocytosis. Relative phagocytic index (RPI) was calculated by dividing the
counts in the presence of seminal plasma by counts in the absence of seminal
plasma and multiplying by 100.
The results of chemilurninescence are expressed in terms of relative phago-
cytic index using whole blood. When phorbol myristate acetate (PMA) was used,
the relative phagocytic index was found to be less than 100% (suppression of
chemiluminescence) with peaks 2, 3, 4, and 7, whereas the RPI with peaks I,
5, 6, and whole seminal plasma was always greater than 100% (stimulation of
chemiluminescence). The RPI was 150% with whole seminal plasma when 100
pglml protein was used in the presence of opsonized yeast. Peaks I and 5 and
Table 1 Effect of Human Seminal Plasma Protein and Its Peaks on Phagocytic
Activity of Monocytes and Granulocytes by Flow Cytometry.
Monocytes Granulocytes
Mean fluorescent Mean fluorescent
channel number channel number
% Total Phagocytic % Total Phagocytic
Sample Phagocytosis population cells Phagocytosis population cells
Control 69 38 1 548 81 686 836
Whole 69 409 586 69 510 736
Peak 3 59 333 560 53 314 584
Peak 4 60 349 582 60 397 656
Peak 7 67 408 609 56 455 800
whole seminal plasma were stimulatory (RPI > loo%), whereas peaks 2, 3, 4,
and 7 were suppressive (RPI < 100%). These results are shown in Fig. 6. The
percent phagocytosis effect of peaks 3, 4, and 7 on monocytes and granulocytes
was also studied by using flow cytometry (Table 1).
C. Mixed Lymphocyte Reaction

Mixed lymphocyte reaction (MLR) was performed for 5 days in the presence
and absence of different concentrations of seminal plasma. Peripheral blood lym-
phocytes were purified from 15 ml heparinized blood by density gradient centrifu-
gation on Ficoll-Hypaque (Pharmacia). Pooled human stimulating cells were pre-
pared by mixing equal parts of three different lymphocyte suspensions of
approximately I million cellslml in RPMI- 1640 medium supplemented with pen-
icillin (100 Ulml), streptomycin (100 yglml), fungizone (25 yglml), and 10%
autologous serum from three different normal donors, and were irradiated by
1500 rads (Gamma Cell 1000, Atomic Energy of Canada, Ottawa, Ontario). Cul-
tures were set up in triplicate using 96-well sterile round-bottom microtiter plates
(Linbro Chemical Co., New Haven, CT, USA) by mixing 100 yl of responding
cell suspension and 100 pl of irradiated pooled cells. Twenty-five microliters of
different concentrations (1, 2.5, 5, and 10 yg) of human seminal plasma and its
seven fractions were added to each well. Control wells were without seminal
plasma and its fractions. Microtiter plates were incubated at 37°C in a humidified
atmosphere with 5% C 0 2 . After 5 days, the culture was pulsed with [jH]thymi-
dine to label nucleic acid in the responder cells. After 16 h, cells were harvested
346 Haq et al.
Figure 7 Effect of human seminal plasma (SP) and its fractions on mixed lymphocyte
reaction (MLR). SP represents the whole seminal plasma while P1-P7 represent peaks
obtained from DEAE-Sephadex A-50 ion exchange chromatography. The x axis represents
protein concentration (pg) and the y axis the index [44].
and counted for internalized radioactivity by using a p-scintillation counter. Fig-

ure 7 shows the effect of seminal plasma or its fractions on MLR. A highly
immunosuppressive effect was observed when fractionated seminal plasma was
used. This effect was dose-dependent7andfound to be maximal with 10 pg protein
per well. Peak 4 was most potent followed by peaks 7,3, and 5 . Peaks 1 , 2 , and 6
were found to confer a stimulatory effect rather than immunosuppression. Whole
seminal plasma was not suppressive and the effect was more or less the same at
both 1 and 10 yglwell protein concentrations.
VII. CYTOKINES AND HUMAN REPRODUCTION
The relationship between cytokines and human reproduction represents a growing

area of interest and investigation because of their involvement in different aspects
of reproductive physiology and infertility regulation, including gonadal and
sperm functions. Cytokines are powerful mediators capable of regulating the
function, growth, and differentiation of the cells of the immune status. There are
studies done by Naz and Kaplan [60] on the detection of tumor necrosis factor-
a (TNFa), interleukin- 1P (IL- LP), interferon-y (IFNy), and interleukin-6 (IL-6)
in human seminal plasma. IL-6 was significantly higher in infertile men compared
to those of fertile. TNFa and IL-1 P were not detected, whereas IFNy was detected
but the difference between the levels of fertile and infertile were not significant.
IL-6 could be associated with infertility. Production of IFNy was determined and
showed to be present in higher amounts in the seminal plasma of infertile donors
[6 11. Recently, the concentrations of various cytokines (IL-1P, IL-2, IL-6, NIL-2,
srIL-6) in seminal plasma have been studied [62]. According to these researchers,
several lines of evidence indicate that cytokines are involved in male infertility.
These cytokines are secreted by different parts of the male genital tract and may
exert effects on steroidogenesis, spermatogenesis, and sperm functions.
VIII. POSSIBLE FUNCTIONS OF PROTEINS

OF SEMINAL PLASMA
The proteins of seminal plasma are involved in the maintenance of spermatozoa

in the female reproductive tract after ejaculation. Some of the proteins, such as
the protease inhibitors present in seminal plasma, may prevent damage to
spermatozoa/spermatozoal membrane. These proteins may be involved and play
a role in capacitation and acrosome reaction; The processes that the spermatozoa
of higher species undergo after ejaculation but prior to acquiring ability to fertilize
an egg. Proteins with no known enzyme or hormone function include those that
appear to be involved in (1) maturation of spermatozoa during passage through
the male reproductive tract, (2) agglutination of spermatozoa or coagulation of
semen after ejaculation (in some species) and subsequent liquification of semen,
(3) maintenance of the viability of spermatozoa, (4) capacitation, (5) acrosome
reaction, (6) penetration of spermatozoa through the cervical mucus, and (7)
sperm-egg interaction. In addition to these functions, there are certain other
classes of proteins present in seminal plasma such as (1) proteins involved in the
growth, regulation, and functions of cells, tissues, and organs of the reproductive
system (sertoli and prostate), (2) inhibitors of hormones (follicle-stimulating hor-
mone) and enzymes (myosin, ATPase, or proteases) present in seminal fluid that
could conceivably inflict damage if their action is not prevented in time, (3)
immunosuppressive agents that might prevent and immune response in both
males and females to sperm and or male-specific antigens, (4) antimicrobial
agents that might prevent infection of the male or female reproductive tract fol-
lowing sexual intercourse, and (5) antifertility agents. Many seminal plasma pro-
teins seem to represent opposite functions, and there are enzymes and their inhibi-
tors, unlike the case in a cell. The balance of these two activities (one promoting
fertility and the other promoting antifertility) might provide for a fine control
348 Haq et al.
over a vital function of semen, e.g., fertilization. Chemically and biochemically,

spermatozoa may well be the best protected cells in the male. And no one will
argue that they deserve to be so well protected. About 15% of a total of 350
proteins thus far shown to be present in seminal plasma and secretions of male
reproductive tract are bound to spermatozoa.
IX. CONCLUSION
Proteins of seminal plasma play an important role in protecting the spermatozoa

during spermatogenesis (when they express stage-specific antigens) in the male
reproductive tract and after coitus in the female reproductive tract. Although these
proteins (immunomodulatory factors) also associated with diverse pathological
states in both males and females, it could be concluded that they are a necessary
evil [9]. Some studies have already stated that fertility is suppressed in couples
were antispermatozoal antibodies are present in seminal plasma. The presence
of spermatozoa1 antibodies in the seminal plasma of infertile and vasectomized
men is well known. Such a case may represent an immune reaction to spermato-
zoa due to a lack of immunosuppressive factors. Our results of mixed lymphocte
reaction and chemiluminescence further confirmed that seminal plasma contains
molecules that are responsible for the suppression of specific and nonspecific
responses [44,50,5 I]. Therefore, identification, purification, and characterization
of the various immunomodulating factors, an understanding of the mechanism
of action of these factors on various components of the immune system, and the
regulation of synthesis of these proteins would ultimately help in understanding
the important role of immunosuppression in human reproduction in general and
fertilization in particular.
REFERENCES
Huang K, Takamura S, Kinochi T, Takayama M, Ishida T, Ueyama H. Nishi K,

Ohkubo I. Alanyl aminopeptidase from human seminal plasma: purification, charac-
terization and immunohistochemical localization in the male genital tract. J Biochem
1997;122:779-787.
Skidgel RA, Peddesh P, Dars R. Isolation and characterization of a basic carboxy-
peptidase from human seminal plasma. Arch Biochem Biophys 1988;267:660-667.
Comhaire F, Bosmans E, Ombelet W, Punjabim V, Schooonjans F. Cytokines in
semen of normal men and of patients with andrological diseases. Am J Reprod Im-
munol 1993;31:99-103.
Drisdel RC, Mack SR, Anderson RA, Zaneveld LJ. Purification and partial character-
ization of acrosome reaction inhibiting glycoprotein from human seminal plasma.
Biol Reprod 1995;63:201-205.
Stites DP, Erickson RP. Suppressive effect of seminal plasma on lymphocytes acti-
vation. Nature 1975;253:727-729.
Caldini AL, Orlando C, Barni T, Messeri G, Pazzegli M, Baldi E. Seris M. Measure-
ment of transfemn in human seminal plasma by chemiluminescent method. Clin
Chem l986;32: 153- 156.
Bryan MKO, Baker HWG, Saunders JR, Kirszbaum L, Walker ID. Hudson P, Liu
DY, Clew MD, Apice AFD, Murphy M. Human seminal clusterin (SP-40,40). Isola-
tion and characterization. J Clin Invest 1993;85:1477- 1486.
Choi NH, Tobe T, Hara K, Yoshida H, Tomita M. Sandwich ELISA assay for quanti-
tative measurement of SP-40.40 in seminal plasma and serum. J Immunol Meth
1990;131:159-163.
Shivaji S, Scheit KH, Bhargava PM. Proteins of Seminal Plasma. New York: John
Wiley and Sons, 1990.
Kitamura M, Namiki M, Matsumiya K, Tanaka K, Matsumoto M, Hara T, Kiyohara
H, Okabe M, Okuyama A, Seya T. Membrane cofactor protein (CD46) in seminal
plasma is a prostasome-bound with complement regulatory activity and measles vi-
rus neutralizing activity. Immunology 1995;84:626-632.
Skibinski G, RElly RW, Harrison CM, McMillan LA, James K. Relative immuno-
suppressive activity of human seminal prostaglandins. J Reprod Immunol 1992;22:
189-195.
Iborra A, Morte C, Fuentes P, Garcia-Framis V, Andolz P, Martinez P. Human sperm
coating antigen from seminal plasma origin. Am J Reprod lmmunol 1996;36: 1 18-
125.
Dimitrova D, Kehayov I, Kyukechiev S. Purification and characterization of sperm-
coating antigen identified by monoclonal antibody. Andrologia 1993;25:271-277.
Nocera M, Chu TM. Transforming growth factor beta as an immunosuppressive
protein in human seminal plasma. Am J Reprod Immunol 1993;30:1-8.
Yie SM, Lobb DK, Clark DA, Younglai EV. Identification of transforming growth
factor-alpha like molecule in human seminal plasma. Fertil Steril l994;6: 129- 135.
Robert M, Gagnon C. Sperm motility inhibitor from human seminal plasma: associa-
tion with semen coagulum. Human Reprod 1995; 10:2192-2 197.
Ohkubo I, Tada T, Ochiai Y, Ueyama H, Eirnoto T, Sasaki M. Human seminal
plasma beta-microseminoprotein: its purification, characterization, and immunohis-
tochemical localization. Int J Biochem Cell Biol 1995;27(6):603-611.
Paradisi R, Pession A, Bellavia E, Focacci M, Flamigni C. Characterization of hu-
man sperm antigens reacting with antisperm antibodies from autologous sera and
seminal plasma in a fertile population. J Reprod Immunol 1995;28:61-73.
Deperthes D, Chapdelaine P, Tremblay RR, Brunet C, Berton J, Hebert J, Lazure
C, Dube JY. Isolation of prostatic kallikrein hK2, also known as hGK- I, in human
seminal plasma. Biochim Biophys Acta 1995; 1245:311-31 6.
Kekow J, Wachsman W, McCutchan JA, Gross WL, Zacchariah M, Carson DA,
Lozt M. Transforming growth factor-p and suppression of humoral immune re-
sponses in HIV infection. J Clin Invest l991;87: 1010- 1016.
Zheng X, Geiger M, Ecke S, Bielek E, Donner P, Eberspacher U, Scheuning WD,
Binder BR. Inhibition of acrosin by protein C inhibitor and localization of protein
C inhibitor to spermatozoa. Am J Physiol 1994;267:C466-472.
350 Haq et al.
Sedor J, Callahan HJ, Perussia B, Laftime EC, Hirsch IH. Soluble Fc gamma RIII
(CD 16) and immunoglobulin infertility. J Androl 1993;14:187- 193.
Chvapil M, Stankova L, Bernhard DS, Zukoski CF, Drach GW. Effect of prostatic
fluid and its fraction on some function of peritoneal macrophages. Invest Urol 1977;
15:173-179.
Mukherjee DC, Agrawal AK, Manjunathn R, Mukherjee AB. Suppression of epidid-
ymal sperm antigenicity in the rabbit by uteroglobulin and transglutaminase in vitro.
Science l983;219:989-991.
Witkin SS, Richards JM, Bangiovanni AM, Zelikovsky G. An IgG-Fc binding pro-
tein in seminal fluid. Am J Reprod Immunol 1983;3:23-27.
Kelly RW. Immunosuppressive mechanisms in semen implications for contracep-
tion. Human Reprod 1995;10: 1686-1693.
Bischof P, Marin-Du-Pan R, Lauber K, Girard JP, Herrmann WL, Sizonenko PC.
Human seminal plasma contains a protein that shares physiochemical, immunochem-
ical, immunosuppressive properties with pregnancy-associated plasma protein-A. J
Clin Endocrinal Metab 1993;56:359-362.
Byrd WJ, Jacobs DM, Amoss MS. Synthetic polyamines added to cultures con-
taining bovine sera reversibly inhibit in vitro parameters of immunity. Nature 1977;
267:621-623.
Akiyama K, Yoshioka Y, Schimid K, Offner GD, Troxler RF, Tsuda R, Hara M.
The amino acid sequence of human beta microseminoprotein. Biochim Biophys Acta
1985;829:288-294.
Abrescia P, Lombardi G, De Rosa M, Quagliozzi L, Guardiola J, Metafora S. Identi-
fication and preliminary characterization of a sperm-binding protein in normal hu-
man semen. J Reprod Fertil 1985;73:71-77.
Metafora S, Lombardi G, De Rosa M, Quagliozzi L, Ravagnan G, Peluso G, Abres-
cia P. A protein family immunorelated to a sperm-binding protein and its regulation
in human semen. Gamete Res 1987;16:229-241.
Freedman MF, Kopf GS. Characterization of a seminal plasma associated inhibitor
of human seminal plasma protein. Biol Reprod 1985;32:322-332.
Chan SY, Tang LC. Effect of reserpine on fertilizing capacity of human spermato-
zoa. Contraception 1984;30:363-369.
Marcum J, Dedman MJR, Brinkley BR, Means AR. Control of microtubule assem-
bly-disassembly by calcium-dependent regulatory protein. Proc Natl Acad Sci USA
1978;75:3771-3775.
Kavanagh J. Zinc binding properties of human prostatic tissue, prostatic secretion
and seminal fluid. J Reprod Fertil 1983;68:359-363.
Robert TK, Boettcher B. Identification of human sperm coating antigen. J Reprod
Fertil 1969;18:347-350.
Sinosich M, Seppala JM, Saunders DM, Grudzinskas G. Pregnancy-associated
plasma protein-A and placental protein in human seminal plasma. Ann NY Acad
Sci 1985;442:287-292.
Leblond E, Desnoyers L, Manjunath P. Phosphorylcholine-binding proteins from
the seminal fluids of different specie share antigenic determinants with the major
proteins of bovine seminal plasma. Mol Reprod Dev 1993;34:443-449.
Calvete JJ, Mann K, Sanz L, Raida M, Topfer-Petersen E. The primary structure of
BSP-30K, a major lipid-, gelatin-, and heparin-binding glycoprotein of bovine semi-

nal plasma. FEBS Lett 1996;399:147- 1.52.
Codini M, Fini C, Paolotti P, Floridi A. Bull seminal plasma phosphodiesterase.
Purification and general properties. Biochem Int 1992;28(6):989-997.
Rutherford KJ, Swiderek KM, Green CB. Chen S, Shively JE, Kwok SC. Purification
and characterization of PSP-I and PSP-11, two major proteins from porcine seminal
plasma. Arch Biochem Biophs 1992;295:352-359.
Calvete JJ, Reinert M, Sanz L, Topfer-Petersen E. Effect of glycosylation on the
heparin-binding capability of boar and stallion seminal plasma proteins. J Chro-
matogr l99S;7l l:l67-173.
Gerena RL, Irikura D, Urade Y, Eguchi N, Chapman DA, Killion GJ. Identification
of a fertility-associated protein in bull seminal plasma as lipocalin-type prostaglandin
D synthase. Biol Reprod 1998;58:826-833.
Haq A, Al-Tufail M, Sheth K, Abdullatif M, Hamilton C, Al-Abdul Jabbar F, A l -
Sedairy ST. Immunosuppression by human seminal plasma fractionated by DEAE-
Sephadex A50 ion exchange chromatography. Andrologia 1992;24:87-94.
Heystek J, Bmmmelhuis HGJ, Krijnen HW. Contributions to the optimal use of
human blood 11. The large scale preparation of prothrombin. A comparison between
two methods using the anion exchanger DEAE-cellulose DE52 and DEAE-Sephadex
A50. Vox Sang l973;2S: 113-123.
Laemli UK. Cleavage of structural proteins during the assembly of the head bacterio-
phage T4. Nature 1970;227:680-685.
Anderson DJ. Tarter TH. Immunosuppressive effect of mouse seminal plasma com-
ponents in vivo and in vitro. J Immunol 1982;128:535-539.
Kuno S, Uneno R, Hayaishi 0.Prostaglandin E2 administered via anus causes immu-
nosuppression in male but not in female rats: a possible pathogenesis of acquired
immunodeficiency syndrome in homosexual males. Proc Natl Acad Sci USA 1986;
83:2682-2683.
Alexander NJ, Anderson DJ. Immunology of semen. Fertil Steril 1987;47: 192-205.
Haq A, Sheth K, Abdullatif M, Al-Abdul Jabbar F, Hamilton C, Al-Sedairy S.
Suppression/stimulation of chemiluminescence by human seminal plasma. Clin
Chim Acta 1991;200:67-70.
Haq A, Sheth K, Abdullatif M, Al-Tufail M, Al-Abdul Jabbar F, Hamilton C, Al-
Sedairy ST. Suppression of luminol-dependent chemiluminescence of phagocytic
cells by human seminal plasma. Mol Androl 1992;4:297-307.
James K, Hargreave TB. Immunosuppression by seminal plasma and its possible
clinical significance. Immunol Today 1984;5:357-363.
Dostal J, Veselsky L, Drahorad J, Jonakova V. Immunosuppressive effect induced
by intraperitoneal and rectal administration of boar seminal immunosuppressive fac-
tor. Biol Reprod l955;52: 1209-1 2 14.
Kelly RW, Holland P, Skibinski G. Extracellular organelles (prostasomes) are
immunosuppressive components of human semen. Clin Exp Immunol 1991;86:550-
556.
Curry MC, Hussain JI, Smith CJ, Allen JC. Identification of a macromolecular inhih-
itor of in vitro lymphocyte proliferation in human seminal plasma as bound sperm-
ine. Eur J Cell Biol l980;2l: 180-1 82.
352 Haq et al.
56. Templeton AA, Cooper I, Kelly RW. Prostaglandin concentrations in the semen of
fertile men. J Reprod Fertil l978;52: 147-150.
57. Witkins SS. Failure of sperm-induced immunosuppression: association with anti-
sperm antibodies in women. Am J Obstet Gynecol 1989;160:1166-1168.
58. Baker HWG, Clarke GN. Hudson B. Treatment of sperm autoimmunity in men. Clin
Reprod F e d l983;2:55-71.
59. Evans CH, Lee TS, Flugelman AA. Spermine-directed immunosuppression of cervi-
cal carcinoma cell sensitivity to a majority of lymphokine-activated killer lympho-
cyte cytotoxicity. Nat Immun 1995;14: 157-163.
60. Naz RK, Kaplan P. Increased levels of IL-6 in seminal plasma of infertile men. J
Androl 1994;15:220-227.
61. Paradisi R, Capelli M, Mandini M, Bellavia E, Flamigni C. Increased levels of inter-
feron-gamma in seminal plasma of infertile men. Andrologia 1996;28:157-161.
62. Dousset B, Hussenet F, Daudin M, Bujan L, Foliguet B, Nobet P. Seminal cytokine
concentrations (IL-lbeta, IL-2, IL-6, sR IL-6), semen parameters and blood hor-
monal status in male infertility. Hum Reprod 1997;12: 1476-1479.
Isolation, Purification, and
Structural Study of Allergenic
Proteins
Jean-Pierre Dandeu
lnstitut Pasteur, Paris, France
I. INTRODUCTION
Allergy, from the Greek "allos" (other) and "ergeon" (action), can be defined as a
separate behavior in comparison with the classical immune response. It i s a modifica-
tion of the human or animal organism reactivity to aparticular immunogen, the ''aller-
gen.'' The molecule responsible for the sensitization was clearly defined at the begin-
ning of the twentieth century by Clemens von Pirquet and then by Paul Portier and
Charles Richet in their works on anaphylaxis. Either for sensitization or desensitiza-
tion an allergen can act at a very low concentration. To become sensitized to an
allergen one must be genetically programmed and able to synthesize immunoglobu-
lins of the particular IgE class in response to an antigen. However, an antigen can
only be considered as an allergen when it is able not only to induce an IgE response
but also to subsequently provoke an anaphylactic orland inflammatory reaction. De-
sensitization is a peculiar immunotherapy that leads the patient to synthesize IgG
instead of IgE in response to repeated injections of very low doses of allergen that
arepractically unable to induce any anaphylactic reaction. In this case, the best mole-
cule to be used should be one whose structure would have been modified to be nonre-
active with the specific IgE present on the mast cells and thus to prevent the cascade
of events leading to the inflammation process by inducing an IgG response.
Allergens are biological molecules such as proteins, glycoproteins, or poly-
saccharides originating from insects, plants, or animals. A biological extract often
contains several allergens. The relative molecular mass of these molecules is gener-
354 Dandeu
ally low and ranges between 10,000 and 50,000. Allergens are frequently qualified by
the terms major or minor allergens. These are referred to their particular frequencies of
IgE reactivity in a group of patients clinically allergic to these allergens but are never
referred to their concentrationsin a biological extract.
Nowadays many allergens have been isolated and purified. The genes of great-
est importance have been cloned and the corresponding recombinant proteins synthe-
sized. Both natural allergens and recombinant allergens are rather useful to improve
our knowledge on IgE synthesis and regulation but also to advance the field of allergy
diagnosis and therapy. We have personally isolated two allergens from cat, one from
horse, and one from a house dust mite. In this chapter we would like to point out the
usefulness of chromatographic methods in allergen purification. One of the most
important techniques generally used to prepare allergenic extracts was first described
by Guibert et al. [I], which essentially consists in an aqueous extraction. From this
extract the allergenic material was first concentrated by acetone precipitations at 25%
and 75%, then by an additional precipitation at 75% of saturation in ammonium
sulfate.
We do not pretend to examine all of the work done on so large a topic and we
have voluntarily limited this review to a comprehensive summary of our personal
results. Mention should be made before any description of chromatographic processes
is given that a combination of crossed line immunoelectrophoresis and rocket line
immunoelectrophoresis, two techniques previously described by Axelsen et al. [2],
are particularly helpful not only for following the degree of purity of a given antigen1
allergen but also for demonstrating whether the different antigens are bearing com-
mon or different epitopes.In rabbit, the majority of natural proteins are good immuno-
gens and give rise to high IgG antibody titers. Thus, it is easy to identify and follow
the purification of a particular antigen using a rabbit antiserum against the crude
material from which the allergen of interest will be purified. Each step of the purifica-
tion process can be followed up by immunoelectrophoresis. The allergenicity of the
protein can be determined by testing the ability to bind specific human IgE antibodies
present in the sera of sensitized patients. To this end, either crossedradioimmunoelec-
trophoresis, western blot, or enzyme-linked immunosorbent assay (ELISA) can be
performed. If sodium dodecyl sulfate-polyacrylamide gel electrophoreses (SDS-
PAGE) remains the most classical tool to define homogeneity of a proteinic fraction
or to determine its relative molecular mass, it is sometimes necessary to perform
chromatofocusing or/and capillary electrophoresis for a better characterization of
the molecule of interest.
II. PURIFICATION OF Equ cl, A HORSE

MAJOR ALLERGEN [3]
Several authors have demonstrated the presence of allergens in horse hair and
dandruff and have isolated several proteins of different molecular weights [4-
Isolation of Allergenic Proteins 355
71. The raw material we used to isolate and purify horse allergens was principally
obtained during grooming of healthy animals. After an aqueous extraction active
material was concentrated by a classical salting-out process using ammonium
sulfate followed by an exhaustive dialysis against water not only to eliminate
ammonium sulfate but to select molecules of a relative molecular mass above
10,000. For efficient concentration and storage the resulting solution was lyophi-
lized. The most classical way to separate proteins or glycoproteins undoubtedly
remains size exclusion chromatography (SEC), which can sometimes lead to al-
most pure active fractions. This is particularly true when the relative molecular
mass of the protein of interest, i.e., Equ c l , is suspected to be around 25,000, as
found in the pic SII (Fig. 1). Since ion exchange chromatography, previously
used by other authors, was not totally efficient in spite of our several attempts
to improve the technique and no more success was obtained with immobilized
Figure 1 Preparative size exclusion chromatography (SEC) performed on a Superose

column HR 16/50. Equilibration and elution were performed with the same buffered salt
solution, 1 M NaCl and 0.02 M phosphate buffer (pH 8.00). and 200 mg of horse hair
dander extract were loaded onto the column (From Ref. 3.)
356 Dandeu
metal ions affinity chromatography (IMAC), we successfully tried hydrophobic

interaction chromatography (HIC) to prepare a pure allergen from the partially
purified fraction defined above.
HIC is a technique whereby proteins are separated according to their selec-
tive interaction with hydrophobic groups, such as phenyl groups, bonded to the
bed material of a column. Moreover, hydrophobic molecules in an aqueous sol-
vent will self-associate due to these hydrophobic interactions. Antichaotropic
salts, e.g., ammonium sulfate, increase the hydrophobic effect [8]. Ammonium
sulfate in solution promotes protein aggregation, leading to precipitation, and this
property was used to prepare the partially purified horse dander extract. In a
solid-liquid phase system, such as HIC, these salts emphasize the interaction
with the immobilized hydrophobic groups [9,10]. They have also been shown to
promote hydrophobic interactions in contributing to the surface tension of the
solution. The salt that increases the surface tension most gives the strongest hy-
drophobic interaction [ l 11. The first gels of practicable use for HIC were prepared
Figure 2 Chromatogram obtained by analytical hydrophobic interaction chromatogra-

phy (HIC) of fraction SII from SEC, on a phenyl-Superose column HR 515 equilibrated
with 2 M (NH,),SO, in 0.02 M phosphate buffer (pH 8.00) (solution A). A 10-mg amount
of SII was loaded onto the column. Elution was carried out in four steps. A first isocratic
run with 100% solution A was followed by a linear decreasing gradient. Two plateaus
were introduced at 40% and 60% of solution B. The last step was performed with 0.02
M phosphate buffer (pH 8.00) (solution B). Flow rate, 0.5 mllmin; UV detection wave-
length, 280 nm. (From Ref. 3.)
by Porath et al. [12] and Hjerten et al. 1131. If HIC has been shown useful in
purifying enzymes 114,151, hormones [ 161, and fragments from IgM 1171, until
now it has never been used to isolate and purify allergenic molecules. Using
a column packed with phenyl-Superose (Pharmacia Sweden) an almost 100%
purification of the horse major allergen Equ c l was achieved. Elution of the
adsorbed molecules according to the order of increasing hydrophobicity was per-
formed by decreasing the antichaotropic salt concentration. i.e., ammonium sul-
fate [18]. For analytical purpose this was done in a linear gradient while for
preparative ones a stepwise elution was carried out (Fig. 2). Physicochemical,
biochemical, and immunochemical analyses stated that Equ c l prepared in that
manner is a pure protein consisting of a single peptide chain with a relative molec-
ular mass of 20,000 and a pI = 3.9. Its partial microsequencing suggeqted that
it could belong to the lipocalin family. This was further confirmed by cDNA
cloning and sequencing of the Equ c l gene [19].
Ill. PURIFICATION OF Feld 1, A CAT MAJOR

ALLERGEN [20]
Several proteins derived from the domestic cat Felix doiwsricus causes allergic
disorders. The most important one, previously designated Cat I, was first de-
scribed by Ohman [21]. Isolation and purification of this tnajor feline allergen,
recently named Fel d l , was achieved by affinity chromatography using polyclonal
and monoclonal antibody 121,221. If this method is somewhat efficient it unfortu-
nately gives a poor yield. Fel d l , which is an acidic protein with a native molecu-
lar mass of 35,000, is composed of two chains that are not covalently bound and
can be easily dissociated. Each monomer, M, 17,000, has an equivalent antigenic
and allergenic potency. Fel d l is particular1y present in house dust [21,23] from
which it can be extracted by water. Since Fel d l is an acidic protein with pI 3.8,
an anion exchanger was used to fractionate the ammonium sulfate-precipitated
fraction previously prepared according to the method described above. An analyt-
ical chromatography started with an isocratic elution followed by a linear gradient
of molarity showed that three fractions can be obtained, each at a very specific
concentration of salt. After that result a stepwise elution was defined and an
enriched fraction of Fel d I, Q2, was eluted at 1 M NaCl (Fig. 3). This partially
purified allergen was then loaded onto a copper chelate column. The chromato-
gram (Fig. 4) shows that five fractions were obtained following a stepwise elution
at different salt concentrations. All fractions were itnmunologically tested, and
the one eluted during the isocratic step appeared as a pure Fel d l preparation.
A second fraction, Q2Cu2, which showed a higher affinity for the copper ions,
also appeared to contain Fel d 1 but rather contaminated. Thus, we concluded that
there were two species of this allergenic protein, which were the monomer and
Dandeu
NaCl
Molarity
Figure 3 Chromatogram of a partially purified house dust extract on a Mono Q HR

10110 column equilibrated with 0.125 M NaCl in 0.02 M Tris-HCI buffer (pH 8.6) (solu-
tion A). Elution is camed out with 1 M NaCl in 0.02 M Tris-HC1 buffer (pH 8.6) (solution
B). Flow rate. 1 mllmin; UV detection at 280 nm. (From Ref. 20.)
Figure 4 Chromatogram of fraction Q2 from the Mono Q, on a chelating Sepharose fast-

flow HR 10110 column, charged with Cu2+ions equilibrated with 1 M NaCl in 0.02 M sodium
phosphate buffer (pH 7.0) (solution A). Elution is carried out with 1 M ammonium chloride
in solution A (solution B). Flow rate, 0.5 mllrnin; UV detection at 280 nm. (From Ref. 20.)
Isolation of Allergenic Proteins
Figure 5 Gel filtration experiments performed on a Superose 12 HR 16/50 column,

equilibrated in 1 M NaCl, 0.02 M Tris-HCI buffer (pH 8.6). Flow rate, 1 mllmn;
UV detection at 280 nm. (a) 4 2 Cu2 analysis; (b) 4 2 Cu3 analysis. d = daltons. (From
Ref. 20.)
the dimer, respectively (Fig. 5). These results lead us to assume that the number
of exposed metal binding sites on the protein able to bind to the copper ions was
greater on the dimer than on the monomer.
This type of experiment underlines the usefulness of affinity chroma-
tography not only to purify allergenic proteins but also to study their struc-
tures.
IV. PURIFICATION OF ALBUMIN, A SECOND CAT

MAJOR ALLERGEN [24]
Previous studies showed that several allergens can be extracted from cat pelts
[25-271. The major one, Fel d l [21], was isolated and purified by us [20]. Some
Dandeu
l lmn
Figure 6 Copper chelate chromatography. Column, HR 10110 packed with chelating

Sepharose fast-flow (Pharmacia), charged with Cu2+,washed with 1 mM imidazole in 1
M NaCl solution buffered with 0.02 M sodium phosphate buffer (pH 7.0), saturated with
10 mM imidazole in the same buffered salt solution. A 1-ml volume of cat serum, previ-
ously dialyzed against 1 mM imidazole, was loaded onto the column. Stepwise elution
was performed at a flow rate of 1.0 mllmn. (From Ref. 24.)
Figure 7 Anion exchange chromatography on Mono Q HR 10110 column equilibrated

with 0.03 M NaCl solution buffered with 0.02 M Tris-HCI (pH 8.0) (solution A). Elution
was performed at a flow rate of 1.0 mllmin with a linear concentration gradient of NaCl
from 0.03 to 1 M (50% solution B). Solution B consisted of 2 M NaCl solution buffered
with 0.02 M Tris-HCI (pH 8.0). A 1-rnl volume of cat serum albumin (CSA), from copper
chelate chromatography (20 mg), was loaded onto the column. (From Ref. 24.)
other allergens, such as albumin [26-281 and in~munoglobulin[28,29], were also

described. We focused on cat albumin, which is a potent allergen present in house
dust and is responsible for numerous sensitizations and crossed sensitizations
with other potent allergens, e.g., dog albumin. A complete study of the cross-
reactions between various albumins of animal origin was recently published [30].
To obtain a highly purified cat albumin we defined a simple chromatographic
process that eliminates the time-consuming ammonium sulfate precipitation. A
slightly diluted cat sera pool was submitted to an IMAC using copper ions as a
ligand. During this process immunoglobulins are concomitantly isolated
in a relatively pure form (Fig. 6). An anion exchange chromatography (Fig.
7) was then used to obtain a pure cat albumin preparation. After chromato
focusing studies (Fig. 8), the fraction appeared to be somewhat homogen
eous. To our knowledge, at the time we published this study, copper chel
ate chromatography had not been used to isolate and purify cat albumin.
This protein was only purified from a Cohn fraction on immobilized nickel
[3 (1.
Figure 8 Chromatofocusing with a Mono P HR 5/20 column equilibrated with 0.025

M sodium acetate buffer (pH 5.58). A 2 0 0 - ~ aliquot
l (4 mg of CSA) was loaded onto the
column. Elution was performed at a flow rate of 1.0 ml/min with a Servalyte 2-4 solution
(Serva) at 0.2% adjusted to pH 2.05. The pH gradient was recorded using a pH monitor
(Pharmacia). Solid curve, CSA from copper chelate chromatography; dashed curve, CSA
Cohn fraction V. (From Ref. 24.)
-
Dandeu
V. PURIFICATION OF Der p l , A HOUSE DUST MITE

MAJOR ALLERGEN [32]
The major allergen from Dermatophagoides pteronyssinus mite was first isolated
and purified about 10 years ago [33-351. It was shown to be a single peptide
with some traces of carbohydrates. Its attachment to the cysteine proteases family
of enzymes appeared from its mRNA [36,37] and cDNA [38] sequences. A fusion
protein produced by a cDNA clone was effectively shown to react with specific
rabbit IgG raised against natural Der p l . As the yield of pure protein from bacte-
rial lysate was reported to be very low, we defined in our laboratory a convenient
process to prepare Der p l from a mite culture using a simple chromatographic
method. Ion exchange chromatography, already effective with a lot of proteins,
was shown to be really efficient to purify Der p l on a large scale. This process
essentially consists of applying an ammonium sulfate fraction obtained from a
partially purified mite culture extract to a Mono Q column (Pharmacia, Uppsala,
Sweden). The protein of interest was eluted during the first isocratic part of the
run, pic Q1 (Fig. 9). All of the controls applied to the so obtained fraction stated
that it was a highly purified Der p l preparation with a real homogeneity in SDS-
NaCl
Molarity
!
f1
Figure 9 Chromatogram of a partially purified extract from a Dermatophagoides

pteronyssinus mite culture on a Mono Q HR 10110 column equilibrated with 0.02 M Tris-
HCl buffer (pH 8.6) (solution A). Elution was carried out with 1 M NaCl in 0.02 M Tris-
HCl buffer (pH 8.6) (50% solution B), flow rate 1 mllmin, followed by 2 M NaCl in 0.02
M Tris-HC1 buffer (pH 8.6) (100% solution B). Flow rate, 3 mllmin; UV detection at 280
nm. (From Ref. 32.)
7
Isolation of Allergenic Proteins
1 2 3 4 5 6 7 8 9 1 0
Migration distance(crn)
-
Figure 10 Electropheretogram obtained by capillary electrophoresis. (From Ref. 32.)
PAGE and an M, of 25,500-a result very close to that reported by other workers
[33,37]. Our Der p l preparation also appeared to be highly homogeneous in capil-
lary electrophoresis (Fig. 10) and in chromatofocusing (Fig. 1l), the most critical
methods of analysis available at that time. At the level of the 1-25 N-terminal
peptide sequence no differences were shown when compared with the same pep-
tide of the cDNA-coded Der p l .
VI. CONCLUDING REMARKS
In this chapter, we have demonstrated that chromatographic techniques, when

judiciously combined and completed with convenient methods of detection and
analysis, can lead to pure preparations of any protein. This was clearly illustrated
above with allergenic proteins. It seems obvious that preliminary analytical exper-
iments are needed to define the most fruitful chromatographic process.
Regarding the study of allergens, there is no doubt that we have not treated
the subject extensively, but we tried to illustrate what can be learned from the
chromatographic behavior of these particular proteins. Information so obtained
Dandeu
Figure 11 Chromatofocusing with a Mono P HR 5/20 column equilibrated with 0.025

M Tris-HC1 buffer (pH 8.0). An 8.5-mg amount of Der pl (QI) was loaded onto the
column. Elution was performed at a flow rate of 1.0 ml/min with a Servalytes 2-4 solution
at 0.2% adjusted to pH 3.0. The pH gradient was recorded using a pH monitor. (From
Ref. 32.)
added to the physicochemical characteristic furnished by electrokinetic methods

can lead to a better understanding of their structure.
As noted for Equ c l , results obtained using chromatographic procedure can
be good preliminaries for the molecular approaches leading to expression of a
recombinant allergen in a bacterial system. Tryptic fragments of the natural aller-
gen after microsequencing are indeed useful for the design of degenerate primers.
Obviously, chromatographic research is still needed for isolation and purification
of the recombinant protein.
ACKNOWLEDGMENTS
All figures were reprinted with permission from Elsevier Science NL, Sara Burg-
erhartstraat 25, 1055 KV Amsterdam, The Netherlands.
REFERENCES
1. Guibert L, Causse-Combes R. Activiti allerge&niquedes fractions obtenues par chro-

matographie de l'extrait de poussiire de maison. Ann Inst Pasteur 1965;108:579-
601.
Axelsen NH, Kroll J, Weeke B. A Manual of Quantitative Immuno-electrophoresis,

Methods and Applications. Scand J Immunol I973;2 (Suppl 1):47-56.
Dandeu JP, Rabillon J, Divanovic A, Carmi-Leroy A, David B. Hydrophobic interac-
tion chromatography for isolation and purification of Equ c l , the horse major aller-
gen. J Chromatogr 1993;62 1 :23-3 1.
Stanworth DR. The isolation and identification of horse-dandruff allergen Biochem
J. 1957;65:582-598.
Ponterius G, Brandt R, Hulten E, Yman L. Comparative studies on the allergens of
horse dandruff and horse serum. Int Arch Allergy Appl Immunol 1973;44:679-69 I.
Lowenstein H, Markussen B, Weeke B. lsolation and partial characterization of three
major allergens of horse hair and dandruff. Int Arch Allergy Appl Immunol 1976;
51:48-67.
Franke D, Maasch HJ, Wahl R, Schultz-Werninghaus G, Bretting H. Allergens of
horse epithelium: I. Physicochernical and immunochemical characterization of five
different horse epithelium raw materials used for allergen preparation. Int Arch Al-
lergy Appl Immunol 1990;92:309-317.
Porath J. Salt-promoted adsorption: recent development. J Chromatogr 1986;376:
331-341.
Tanford C. The hydrop'hobic effect: Formation of Micelles and biological mem-
branes. New York: Wiley Interscience, 1973.
Creighton TE. Proteins, Structure and Molecular Properties. New York: W. H. Free-
man, 1984.
Melander W, Horvath C. Salt effect on hydrophobic interactions in precipitation and
chromatography of proteins: an interpretation of the lyotropic series. Arch Biochem
Biophys 1977;183:200-2 15.
Porath J, Sundberg L, Fornstedt N, Olsson NI. Salting-out in amphiphilic gels as a
new approach to hydrophobic adsorption. Nature 1973;245(5426):465-466.
Hjerten S, Rosengren J, Palhman S. Hydrophobic interaction chromatography: the
synthesis and use of some alkyl and Aryl derivatives of agarose. J Chromatogr 1974;
101:281-288.
Lau KH, Freeman TK, Baylink DJ. Purification and Characterization of an acid
phosphatase that displays phosphotyrosyl-protein phosphatase activity from bovine
cortical bone matrix. J Biol Chem 1987;262:1389-1397.
Roos P, Nyberg F, Wide L. Isolation of human pituitary prolactin. Biochem Biophys
Acta 1979;588:368-379.
Prescott M, Peek K, Veitch DP, Daniel RM. J Biochem Biophys Methods. The use
of phenyl-Sepharose for the affinity purification of proteinases. J Biochem Biophys
Meth 1993;26:5 1-60.
Inouye K, Morimoto K. Single-step purification of F(ab')2 mu fragments of mouse
monoclonal antibodies (immunoglobulins M) by hydrophobic interaction high-per-
formance liquid chromatography using TSK gel ether-SPW. J Biochem Biophys
Meth 1993;26:27-39.
Hjerten S, Yao K. Eriksson KO, Johansson B. Gradient and isocratic high-perfor-
mance hydrophobic interaction chromatography of proteins on agarose columns. J
Chromatogr 1986;359:99-109.
Gregoire C, Rosinski-Chupin 1, Rabillon J, Alzari PM, David B, Dandeu J-P. cDNA
Dandeu
Cloning and sequencing reveal the major horse allergen Equ cl to be a glycoprotein
member of the lipocalin superfamily. J Biol Chem 1996;271:32951-32959.
Dandeu J-P, Rabillon J, Beltrand M-J, Lux M, Duval R, David B. Immobilized metal
ion affinity chromatography for the purification of Fel d I, a cat major allergen, from
a house-dust extract. J Chromatogr 1990;512: 177-88.
Leitermann K, Ohman JL. Cat allergen I: Biochemical, antigenic and allergenic
properties. J Allergy Clin Immunol 1984;74:147-1 53.
Chapman M, Aalberse RC, Brown MJ, Platts-Mills TAE. Monoclonal antibodies to
the major feline allergen Fel d l . 11. Single step affinity purification of Fel d l , N-
terminal sequence analysis, and development of a sensitive two-site immunoassay
to assess Fel d 1 exposure. J Immunol 1988;140312-8 18.
Ohman JL, Lorusso JR, Lewis S, Cat allergen content of commercial house dust
extract: comparison with dust extracts from cat-containing environment. J Allergy
Clin Immunol 1987;79:955-959.
Dandeu J-P, Rabillon J, Guillaume J-L, Camoin L, Lux M, David B. Isolation and
purification of cat albumin from cat serum by copper ion affinity chromatography:
further analysis of its primary structure. J Chromatogr 1991;539:475-484.
Ohman JL Jr, Lowell FC, Bloch KJ. Allergens of mammalian origin: characteri-
zation of allergen extracted from cat pelts. J Allergy Clin Immunol 1973;*:231-
241.
Ohman JL Jr, Lowell FC, Bloch KJ. Allergens of mammalian origin: 111. Properties
of a major feline allergen. J Immunol 1974;113: 1668-1677.
Andersson MC, Baer H. Allergenically active components of cat allergen extracts.
J Immunol 1981; 127:972-975.
Lowenstern H, Lind P, Weeke B. ldentification and clinical significance of allergenic
molecules of cat origin. Allergy 1985;40:430-44 1.
Ohman JL, Sundin B. Standard~zedallergenic extracts derived from mammals. Clin
Rev Allergy 1987;5:37-47.
Goubran-Botros H, Gregoire C, Rabillon J, David B, Dandeu J-P. Cross-antigenicity
of horse serum albumin with dog and cat albumins: study of three short peptides
with significant inhibitory activity towards specific human IgE and IgG antibodies.
Immunology 1996;88:340-347.
Andersson L, Sulkowski E, Porath J. Purification of commercial human albumin on
immobilized IDA-NiZ+.J Chromatogr 1987;421: 141- 146.
Dandeu J-P, Rabillon J, Gullaume J-L, Camoin L, Lux M, David B. Isolation of
Der pl, the Dermatophagoides pteronyssinus major mite allergen, from a crude mite
culture extract, purification by ion-chromatography, and comparison between the
material obtained and a cDNA-coded Der pl. J Chromatogr 1992;599:105-111.
Chapman MD, Platts-Mills TAE. Purification and characterization of the major aller-
gen from Dermatophagoides pteronyssinus-antigen PI. J Immunol 1980;125:587-
592.
Krilis S, Baldo BA, Basten A. Antigens and allergens from the common house dust
mite Dermatophagoides pteronyssinus. Part 11. Identification of the major IgE-bind-
ing antigens by crossed radioimmunoelectrophoresis. J Allergy Clin Immunol 1984;
74:142-146.
Lind P, Lowenstein H. Identification of allergens in Denriatophagoides pteronyssi-
nus mite body extract by crossed radioimmunoelectrophoresis with two different

rabbit antibody pools. Scand J Irnmunol 1983;17:263-273.
36. Stewart GA, Thomas WR. In vitro translation of messenger RNA from the house
dust mite Dermatophagoides pteronyssinus. Int Arch Allergy Clin Immunol 1987;
83:384-389.
37. Chua KY, Stewart GA, Thomas WR, Simpson RJ, Dilworth RJ, Plozza TM, Turner
KJ. Sequence analysis of cDNA coding for a major house dust mite allergen, Der
p I . Homology with cysteine proteases. J Exp Med 1988;167: 175- 182.
38. Stewart GA, Thomson WR, Chua KY, Geysen HM. An allergen and antigenic map-
ping analysis of a major mite allergen, Der pl. Adv Biosci 1989;74:297-3 l I.
Multiangle Light Scattering
Combined with HPLC with
Examples for Biopolymers
Philip J. Wyatt
Wyatt Technology Corporation, Santa Barbara, California
I. INTRODUCTION
In 1972, Huglin's collection [I] on light scattering from polymer solutions was
published. This was a full 2 years before Ouono and Kaye's classic paper [2]
reporting the combination of gel permeation chromatography (GPC, often called
size exclusion chromatography or SEC) and light scattering (IS). Although
Moore [3] had developed the GPC concept a decade earlier and James Waters
had launched his company a few years after that, nowhere in Huglin's text are
to be found references to the GPC concept. Indeed, Beckman did not introduce its
low-angle laser light scattering (LALLS) instrument until 1972, so this powerful
technique would remain but a curiosity until Chromatix licensed the concept from
Beckman and began their "missionary" work. Nevertheless, even without the
ability to fractionate samples before making LS measurements, LS techniques
themselves had reached a high degree of sophistication. Perhaps this was best
summarized for the field of biopolymers by Burchard and Cowie [4] in their
chapter entitled "Selected topics in biopolymeric systems." They stated that "an
attempt has been made to present a wide variety of examples which illustrate
most effectively the breadth of application of light scattering in this field. The
versatility of the technique is probably unique in providing information in depth
on biopolymers, and while it is by no means the only method in use, one hopes
it will become obvious to the unconverted that to neglect light scattering would
be to proceed under a distinct disadvantage." Although the numbers of "uncon-
369
370 Wyatt
verted" are still large not only in the field of biopolymers but in the broad areas
of organic polymers as well, the introduction of the DAWN instruments (which
implemented the concept of multiangle light scattering, or MALS, combined with
GPC) in the mid 1980s has begun to change those numbers. Referring back to
the various chapters in Huglin, it is of interest to note that huge areas of applica-
tion have yet to be examined by the combined GPCIMALS approach. It is the
object of this chapter to continue this "conversion" process while presenting some
important new results for biopolymers.
II. THEORETICAL REVIEW
Despite the increased importance ascribed to LS methods during the past few
years, there remains a great deal of uncertainty and misunderstanding about the
subject. A few summary remarks are in order. We begin with the basic equation
used most frequently to describe the relation between what is measured and what
is derived. Equation (1) is the result popularized by Zimm [ 5 ] in the development
of his now classical technique by which he was able to generate key molecular
parameters, i.e., the weight average molar mass, Mw, mean square radius, (r:),
and second virial coefficient, A*, from measurements made on an unfractionated
sample comprised generally of an unknown distribution of molar masses and
sizes.
where K* = 4n2(dnldc)n~l(NAh& NA is Avogadro's number, dnldc the refrac-

tive index increment, ho the vacuum wavelength, and no the refractive index of
the solvent. Most importantly, the excess Rayleigh ratio, R(O), and form factor,
P(O), are defined, respectively, by
where
I, is the incident light intensity (ergs/cm2-s), f(O),,,, a geometrical calibration

constant that is a function of the solvent and scattering cell's refractive index
and geometry, I(0) and Is(€))the normalized intensities of light scattered into a
unit solid angle subtended by the detector at angle 0 with respect to the illumi-
MALS Combined with HPLC 371
nated solution and solvent, respectively. The mean square radius is given by Eq.
(5) below, where the distances r, are measured from the molecule's center of
mass to the mass element m,:
Equation (5) applies to a single molecule, but the quantity measured from an
unfractionated ensemble is actually a light scattering average of the mean square
radii present in the sample. See Section IV, below, for further details.
Equation (1) is derived, to order c2, by Zimm from the more general [5,6]
reciprocal form:
In Eq. (6), A, is the third virial coefficient and Q(8) is a special form factor, of
magnitude always less than unity for finite 8, that describes additional light-scat-
tering contributions from the interacting molecules. At 8 = 0, both P and Q = 1.
Equations (1) and (6) are based on several approximations. First is the
assumption that the solvated molecules are well described by the Rayleigh ap-
proximation [6], i.e., that the refractive index (RI) of the molecules differs imper-
ceptibly from that of the solvent and that the solvated molecules do not affect
the phase of the (plane) wave of (polarized) light incident on the solution. Then
there are the assumptions that the concentration of the solute molecules is "van-
ishingly small" and that the molecules interact with each other at a single point.
Although the concentrations of samples after GPC fractionation are generally
very small, the assumption that very large molecules may interact only in the
single contact point approximation seems somewhat unreasonable. Remarkably,
the validity of these assumptions spans a surprisingly broad range of molar
masses.
Refemng to Eq. (I), we see some other important results of measurements
at very low concentrations. When the term 2A2c may be dropped at very low
concentrations, the mean square radius may be derived independently of any
knowledge of the concentration c or molar mass M. (See also Section IV for a
discussion of further limitations.) In this limit, Eq. (1) reduces to
K*c - 1
- -[I + a , sin' 812 - a2sin4 812 + ...I
R(8) MP(8) M
where a , is defined by Eq. (4).
Wyatt
Letting
we have the limiting derivative of 5 as 0 4 0:
Since the intercept at 5(0°) = 1/M,, the mean square radius is readily calculated
from the ratio of the slope to intercept. And that ratio is independent of M, and
c! Remember that 1/M, is calculated from the intercept, i.e., using the extrapo-
lated excess Rayleigh ratio, R(OO). So it does not matter what M, is actually
calculated: Set c = any arbitrary constant and one will generate an arbitrary M,.
Both M, and c drop out of the calculation. The important term is R(OO)and that
is obtained by extrapolation using mensured data.
Ill. SIGNIFICANCE OF MALS, LIGHT SCATTERING

ANOMALIES, AND IMPLICATIONS FOR DATA FITTING:
STRAIGHT LINES AND RANDOM COILS
Since the earliest days of on-line light scattering [2] there has been a considerable
amount of confusion as to the correct means for extracting the molar mass. At
vanishingly small concentrations and in the limit as the scattering angle, at which
the measurement is made. approaches zero, the excess Rayleigh ratio divided by
the product K*c becomes exactly equal to the weight average molar mass. This
is clearly seen from Eq. (7). In this manner, were the scattering angle taken small
enough, one could extract the molar mass of each eluting slice without having
to extrapolate the measurements following the methods developed by Zimm [ 5 ] .
The Beckman instrument (eventually developed further and marketed as the
Chromatix KMX6) made measurements at angles as low as 3" which, for all
intents, could be equated to 0" with only minor error up to molar masses well
into the millions.
Measurements at a single angle, of course, are insufficient to generate the
corresponding mean square radii, so the LALLS technique can at best generate
molar mass with no subsequent details of conformation or branching. The low-
angle measurement has other problems as well. In particular, the presence of any
dust or debris from column shedding results in additional signals throughout the
elution profile that often swamps the molecular signals themselves. Figure 1
shows the strip chart recording by Berkowitz [7] of the familiar broad polystyrene
standard NBS 706 illustrating the type of spike noise characteristic of such mea-
surements. (The Berkowitz paper describes various methods to suppress this
noise and try to restore the peak.) Figure 2 presents a three-dimensional plot of
MALS Combined with HPLC
TIME - TllE
Figure 1 Strip chart recording from raw LALLS data from N B S 706 and NBS 705.
Figure 2 Multiangle chromatograms from NIST SRM 706 polystyrene.

374 Wyatt
the same standard showing the variation of the noise with scattering angle for a
separation in toluene. The data of Fig. 2, however, were obtained with a DAWN-
DSP system with a lowest angle collected around 24" in contrast to a 3" collec-
tion for the LALLS device. The point to emphasize here is that as the scattering
angle decreases, the associated noise generally increases. A scattering angle
closer to 0" does not necessarily mean that the extracted molar mass will be any
the more accurate. Indeed, there are some unexpectedly erroneous results that
can occur by ignoring the significance of the errors associated with the data col-
lected at any angle. Let us examine in more detail the extraction of molar mass
(and root mean square, or rms, radii, when attainable) from light scattering mea-
surements.
Figure 3 shows a set of MALS chromatograms at 16 angles (plus the Re-
fractive Index, RI, detector) for an aqueous GPC separation of a nominal 400
kglmol pullulan "standard." The actual scattering angles accessible (together
with their corresponding sin2 812 values) with a DAWN K5 glass cell are listed
in Table 1. The measurements of Fig. 3 were achieved without detector 1 (3.3").
The chromatography, as indicated by the "noise" at the smaller scattering angles,
is of rather poor quality, yet even the smallest scattering angle measured (14.5")
is considerably larger than the 3" data collected from the KMX6. The value of
Figure 3 Multiangle chromatograms from a nominal 400 Kglmol pullulan standard.

Table 1 DAWN Scattering Angles and

Values of sin2 012 for K5 and Aqueous Solvent
Detector
no. Angle (O) Sin2(012)
sin2 812, however, is very small relative to the range of angles measured. Figure
4 (top) shows a Zimm plot for the single slice indicated by the vertical line in
the overlay plot (bottom) of the RI and 90" LS signals. For chromatographic
separations, the concentration at a given slice is assumed to be so small that the
second virial coefficient may be set equal to zero with negligible error. Thus the
Zimm plot has only a single concentration. The second virial coefficient may be
entered manually if required; however, since chromatographic separations are
occurring at a single concentration for each slice, the second virial coefficient
cannot be derived directly. Note that for rms radii up to about 40 nm, the plot
of K*c/R(B) versus sin2 812 is linear. The weighted fit of the data points to a
straight line will then yield the molar mass (intercept at sin2 812 = 0) and the
mean square radius (slope divided by the intercept times a constant). The ques-
tions that now might be asked include: How will these two quantities vary with
the lowest angle measured? Will they be more (or less) precise as the position
of the lowest angle measured is varied? Indeed, does the low angle even add
additional information?
A least-squares fit of a set of data points to a straight line is a straightfor-
P
Wyatt
Peak 1, Slice 435, 15.250 mL

Zimm Method, Degree 1
c: (2.140*0.0005) x g/mL
M (4.182*0.027) x 1O5 g/mL
r: 24.8*0.8 nm
Figure 4 Zimm plot for a single collection slice (top) indicated by the vertical indicia1
mark on the elution curves (RI and 90°LS) on the bottom.
ward process. The minimization of the sum of the squares with respect to the
slope a and intercept b (= 1 IM) yields the pair of equations:
and
where y = K*c/R(sin2 812) = ax + b, xi = sin2 8'12, y(x,) = yi, and w, is the

weight or reciprocal standard deviation of the measured y,.
If only a single angle is measured (such as with a LALLS system), then
the precision of the intercept becomes the same as the precision of the LS signal
which, for the noisy chromatography shown in Fig. 3, will show a large standard
deviation. Again, the only quantity derived from a LALLS measurement will be
the molar mass. If only two angles are measured, then Eqs. (10) and (1 1) take
on the simpler forms:
Y2 - Y I
a =- and b =
x2y1 - yzxl
X2 - X, x2 - X I
As expected, these latter results are independent of the standard deviations of

the measured data but depend only on the two average values measured. The
results shown in Eq. (12) do not represent a least-squares fit since there is no
redundancy in the measurements; there are as many data as there are variables.
The number of degrees of freedom is exactly equal to the number of data. Each
angle measured is weighted the same.
Consider a two-angle measurement made at sinZ8 , / 2 and sin2 8J2 where
82 > 81. From Eq. (7), the molar mass determined from measurement of K*c/
R(sin2 812) at the two angles 8 , and 8 2 (using the straight line fit) is given by
M = l l b = (x2 - x,)/(x2y, - xly2).
We next evaluate the error in M as we select smaller angles for one of the
two angles chosen to fit the straight line.
1
Wyatt
In Eq. (13), the standard deviation in Ay is expressed in terms of the standard

deviation of the measured excess Rayleigh ratio, AR, since y = K*c/R. Thus
Ay = - K*c AR/R2 and
In the limit as x l = sin2 8 , / 2 + 0, Eq. (14) simplifies to
The choice of a very small angle as one of the two points through which
the straight line is to be constructed indicates, from Eqs. (14) and (15), that the
error in the derived molar mass, M, will increase as the lower angle goes to zero.
For noisy chromatography, the fluctuations of R(sin2 01/2)will increase as 0 +
0. Accordingly, if only two angles are chosen for a LS measurement, the selected
angles should be as far away from 0" as practical.
For the case of three angles, where the lowest angle measured is also ex-
pected to have a correspondingly less precise value than either of the other two,
replacing only the lowest angle measured with an even smaller angle in no way
implies that the final values will be more precise. On the contrary, keeping mea-
surements away from 0" will generally increase the precision of the determina-
tion. Furthermore, as should be evident from Eqs. (10) and ( l l ) , the fit to the
straight line will now be a true least-squares fit and the weighted data will result
in the "best" fit being selected.
As might be expected, as the number of angles included in the final calcula-
tions is increased, the precision of both molar mass and mean square radius is
increased. Table 2 presents the relative precision of these final values for the case
of two (very low angle and 90°), three (approximately 40°, 90°, and 140°), and
many (see Table 1) angles for the example of Figs. 2 and 3.
One of the possibilities associated with making LS measurements with too
few angles lies in the hope of extending the range accessible by assuming a model
of the molecules measured. For certain types of models, the form factor, P(0),
is known a priori so there is no need to fit the data to a straight line or higher
Table 2 Precision of Measurements Based on

Clean (200 Kglmol Polystyrene) Chromatography
of Fig. 2 and Noisier (400 Kglmol Pullulan)
Chromatography of Fig. 3 as a Function of Number
of Detectors Used
Molar mass Radius

Angles (%I (%I
Clean chromatography
Many 0.07 0.6
40" + 90" + 130" 0.2 2
15" + 90" 0.6 10
Dirty chromatography
Many 0.5 3.0
40" + 90" + 130" 1 5
+
15" 90" 14 80
order polynomial for the subsequent determination of molar mass and rms radius.
Of course, such an assumption flies somewhat in the face of the concept of an
"absolute" measurement. Nevertheless, we see frequent use of this assumption
either directly or implicit in some software programs. The latter assumption is
insidious in that data are processed and results calculated without the possibility
of user intervention. If a random coil model is built into the software without
the user's knowledge, then there is no basis left for the user to understand why,
for example, the results contradict the results of other measurements such as
thermal analysis. Let us examine the random coil model and its dual-angle imple-
mentation in greater detail.
For a random coil molecule, the form factor P(8) takes on the simple closed
[5] form:
where
and y = [47cno/ho]sin 012. P(8) may be expanded in u as

380 Wyatt
Consider now the case of a two-angle measurement. In this case (and a

vanishingly small concentration, whereby we may neglect the term 2A2c), Eq.
(1) becomes
where P(0) is given by Eq. (16). Equation (19) has two unknowns, M u and
(ri). and thus again like the straight line fit of Eqs. (12), the number of data are
equal to the number of degrees of freedom. The average values of R(0) measured
at 0 , and O2 are the only data available for this fit, i.e., R(0,) and R(02). The
nonlinear Eq. (19) in the two unknowns. M, and (ri), has (at best) a single soh-
tion. Let the two measured excess Rayleigh ratios be 5, = R(0,) and c2 = R(0?).
Equation (19) is best solved first by eliminating M u i.e.:
where
One would generally begin by solving Eq. (20) for (ri) which value is then substi-
tuted in Eq. (16) to obtain P(0) at the angle for which R(0) has the smallest
standard deviation. This value and the corresponding R(0) are then substituted
in Eq. (19) to solve for M,. The most "accurate" value of M, derived from Eq.
(19) would best be calculated at the angle with the least noise (the largest, 02),
once (ri) had been calculated from Eq. (20). But note that the nonlinear calcula-
tions generating (ri) from Eq. (20) will have a precision that depends on the
measured quantities 5, = R ( 0 , ) andc2 = R(02).To maintain the greatest preci-
sion in this derivation of (ri), we see again that if only data at two angles are to
be collected, the angles should be chosen as far as possible from the small, noisier,
forward angles. Once a model is chosen, the smaller angle fits to that model add
to the errors of the final results as the experimental uncertainty of those data.
The precision of such a determination will depend not only on how small the
noise is at the angles selected but on the number of different angles measured
as well. Once again, the best way to fit a model such as the random coil model
is to extract the key physical constants by fitting weighted data in a least-squares
sense. Thus one must use as many angles as possible and certainly at least three.
In closing this section, it is worth remarking that traditional methods for
determining functional parameters by comparing with experimentally derived
data require that the number of such data exceed substantially the number of
functional parameters or degrees of freedom. Only on this basis can one perfom1
a "chi-squared" goodness-of-fit analysis and thereby establish the quality of the
selected model (function) by which the data are governed. Any model must be
tested at points throughout its range of application. With only two data points
and two angles, the results generated will depend always on the model chosen.
IV. THE "MOMENT" OF THE ROOT MEAN SQUARE

RADIUS FROM MALS AND OTHER MOMENTS
One often reads that the rms radius measured by light scattering [8] is actually
a z-average radius. (The misnomer "radius of gyration" is frequently found in
the literature to describe this same quantity.) It is important to understand just
what these so-called moments of the distributions mean since they play an impor-
tant role for a variety of analyses, not the least of which is the determination of
a sample's polydispersity. In addition, concepts such as the measurement and
quantitation of branching require that the moments be identical within each suit-
ably fractionated slice of a GPC separation. We begin by considering the MALS
measurement of a polydisperse sample. We may rewrite Eq. (7), at vanishingly
small concentrations and small angle, to obtain the excess Rayleigh ratio mea-
sured from the suspension:
where M, is the weight-average [see Eq. (25)] molar mass, c = Cc,, and K =
2molh0,as before. In the limit as 0 + 0, the MALS measurement yields the
weight-average molar mass, i.e., R(OO)= K*M,c. Note that the constant K* has
been factored out from the summations over the species present based on the
further assumption that the polymer is a homopolymer, or at the very least that
the dnldc value is the same for all species present (including, for example, homo-
geneous copolymers). In the event that a copolymer or polymer blend has been
separated, the mean square radius is not so easily extracted. A more detailed
discussion of the modifications to the LS formulation may be found in the paper
by Benoit and Froelich [9].
Wyatt
Kratochvil [8] has pointed out that
by definition. In other words, the z-average mean square radius has been dejned
by means of a weighting involving both mass and concentration. Let us examine
this and related moments more carefully.
The mass moments of a distribution are given traditionally by
for the number-average molar mass, where ni is the number of molecules of mass
M, and the summations are over all of the i species present; the concentration c,
of the ith species therefore is equal to Mini;
for the weight-average molar mass; and
for the z-average molar mass. Of course, Eqs. (24) and (25) have an associated
physical meaning in terms of numbers and weights. Each corresponds also to an
absolute mass measurement technique. Light scattering is the absolute method
of choice to measure the weight-average molar mass while osmometry is the
absolute technique of choice to measure the number-average molar mass of a
sample directly. Equation (26) corresponds to the quantity measured by the other
absolute measurement technique, sedimentation equilibrium. The subscript "z"
has its origin in the German word for centrifuge, Zentrijuge. From Eq. (26) it is
a simple matter to write expressions for the z + 1, z + 2, etc., moments. Note
how these "moments" are related. In particular, each higher moment corresponds
to "the next higher weighting" of both numerator and denominator by the factor
M;.
The number- and weight-average values of the mean square radius, (ri),
are easily written in terms of their definitions i.e.:
for the number-average mean square radius, i.e., the mean square radius of each
species is weighted by the number, ni, of such members present. So far so good,
but to calculate the weight-average mean square radius we must weight the mean
square radius by the concentration, ci, of each species present, i.e.:
In terms of the physical meaning of the number and mass weightings, as

opposed to the concept of moments [lo], Eqs. (27) and (28) indeed represent the
correct corresponding weightings for the mean square radius. Defining the z-
average mean square radius by Eq. (23) is completely acceptable since this is
what MALS actually measures. One might equally well call Eq. (23) the LS
average in order to avoid trying to relate the result to a seemingly artificial mo-
ment of the distribution of mean square radii present in an ensemble. Indeed,
rewriting Eq. (23) as
and Eq. (27) as
produces three equations [(30), (28), and (29)] that display the same type of mo-
ment relationship shown in Eqs. (24), (25), and (26). However, note that once
again the formal moments [6] of the mean square radii are with respect to the
masses MI and not the mean square radii. Note that for the special case of a
random coil structure at the theta point, M I a (ri). Replacing (r;) by M, in Eqs.
(30), (28), and (29) yields Eqs. (24) through (26), respectively. In order to empha-
size that the size moment measured by light scattering is not a true z average,
384 Wyatt
except for the rare occasion that the molecules being measured are random coils
in a theta solvent, the term LS average should be used in the future, viz.
V. CONFORMATION AND DIFFERENTIAL WEIGHT

FRACTION DISTRIBUTIONS FOR "INCOMPLETE"
GPC SEPARATIONS
Sometimes a GPC separation appears to be incomplete, i.e., the mass elution

curve ("calibration" curve) is not a monotonically decreasing function of the
elution volume. Examples are shown in Figs. 5 and 6. The former corresponds
to a polyacrylamide while the latter was obtained from a (polysaccharide) heparin
12.0 16.0 20.0 24.0 28.0

Volume (mL)
Figure 5 Molar mass versus elution volume of a polyacrylamide sample.

I HEPARIN
x'
6.5 7.0 7.5 8.0 8.5 9.0 9.5

Volume (mL)
Figure 6 Molar mass versus elution volume of a heparin sample.
sample. Both show the mass elution curves reversing their downward trends and
turning upward at larger elution volumes. In both cases the mobile phase was
aqueous. Note that these results were obtained via a MALS measurement com-
bined with an RI detector to monitor the concentration. The RI elution curves
are shown in the background. The types of mass elution calculated from MALS
would not have been calculated from standard GPC plus column calibration. Only
an absolute measurement would have disclosed this unusual elution behavior
since calibration techniques invariably require a set of standards that produce
decreasing molar mass with increasing elution.
A variety of explanations have been presented for the non-GPC effects seen
here, though without MALS they are rarely detected. Among them are column
interactions and the presence of microgels. The former effects are akin to affinity
chromatographic interactions. The latter are based on the fact that highly compact
structures with higher molar mass would be expected to elute at later times be-
cause their hydrodynamic size, for their greater mass, could be considerably
smaller than their noncompact companions. But irrespective of the possible expla-
nation, the mere presence of these fractions requires that they be included in the
mass distributions reported as they often represent a considerable fraction of the
386 Wyatt
sample on column. The most powerful means of presenting the results of a GPC
separation is to produce a differential weight fraction distribution. From such
a result, the cumulative distribution in addition to the various moments of the
distribution may be calculated. The differential weight fraction distribution also
represents the most direct means by which samples may be differentiated and
column performance evaluated. Let us look at these distributions in greater detail,
especially when applied to the results of Figs. 5 and 6.
The differential mass fraction distributions are generally presented as a
distribution in log,, M rather than M. This permits the details of distributions
spanning several orders of magnitude to be clearly visible. Perhaps most impor-
tantly, this type of presentation arose because the calibration curves are generally
prepared with standards spanning a few orders of magnitude. An ideal GPC col-
umn set will separate linearly with respect to the logarithm of the hydrodynamic
size. Since the logarithm of the hydrodynamic radius is proportional to the loga-
rithm of the molar mass for linear molecules, the presentation of distributions in
these terms has become an accepted practice. Let us review the definitions of
these quantities.
Following Shortt 11I],we define W(M) as the cumulative weight fraction
of molecules in the selected peak whose mass is less than M. This is the total
mass of molecules of mass less than M divided by the total mass of the molecules
present. Thus:
The differential mass fraction, w(M)dM, is the weight fraction of molecules of

mass between M and M + dM. Thus, w(M) = dW(M)/dM and again
Letting x(M) = dW(M)/d(logloM), we have
x(M) = -- w(M) = Mw(M) log, 10

log 10 e
Referring now to Fig. 5, we must combine the mass elution as a function

of elution volume with the corresponding concentration values to generate the
differential weight fraction distribution function of Eq. (34). The traditional pro-
cedure is to fit the data of Fig. 5 (log,, M as a function of V) by a polynomial
(generally linear) and apply the chain rule as follows:
However, dV/dM = [dV/d(logl o M)] d(1og M)/dM = [dV/d(log M)]

[log,, e]/M. Therefore we have, after canceling the factor Mllog,, e in Eq. (35),
x(M) = dW/dV[dV/d(logloM)] = (dW/dV)/[d(logl0M)/dV] (36)
For normal separations by the GPC mechanism (size exclusion), the term
dW/dV may be replaced by the concentration detector's response. Thus if the
concentration detector's baseline subtracted response is h(V), then dW/dV =
-h(v)/I h(V)dV, the integral representing the sum over all contributing concen-
trations. The differential weight fraction has thus been replaced [I I] by the nor-
malized concentration fraction. Note the negative sign which corresponds to the
1.ox1oF 1.OXI o7
Molar Mass (glrnol)
Figure 7 Differential mass fraction distribution for the polyacrylamide sample of Fig.
5 formed by re-sorting the slice collected data into bins in log,, M.
388 Wyatt
fact that the variation of the slope of the "calibration curve," d(log,, M)/dV, is
negative since larger volumes correspond to smaller molar masses. The differen-
tial ("log base 10") mass fraction is then written as
If the term d(log,, M)/dV is linear, then the differential fraction, x(M) is just a
rescaled concentration curve. As long as d(logloM)/dV remains monotonically
decreasing, the formalism of Eq. (37) is reasonable. However, when the variation
of d(logloM)/dV becomes zero and even reverses its slope, such as shown in
Figs. 5 and 6, the results are both difficult to understand and could be in error.
For example, in Fig. 5, a simple fifth-order polynomial does not even fit the data,
so that the results based on Eq. (37) may be in considerable error because the
larger mass fractions that elute irregularly (non-GPC) are not accorded their
proper contributions to the distribution of masses present in the sample. Let us
examine these results in more detail.
-.-
*
+
+ * PAM91-1
+ Norm = Log
3rd order
i
,2.0-
.-0
4- -
0
F -
L
E -
.-m
-
3
.- 1.0-
3
4-
C
2 -
w
E -
n
0.0- I 1 1 1 1 1
1 .OXI o6 1.ox1o7
Molar Mass (g/rnol)
Figure 8 Differential mass fraction distribution for the polyacrylamide sample of Fig.
5 using a third-order fit to the mass elution data. Note the double values about the point
of reversal of molar mass versus elution volume, i.e., where d ( l ~ gM)/dV
,~ = 0.
We begin again with the formalism of Eq. (34). At each elution volume
V, there is a corresponding mcasurcd value of M as well as a baseline corrected
concentration response h(V). The procedure to generate the differential weighted
log base 10 mass fraction, x(M), begins weighting the calculated mass M, by
(log, 10) h(V,)/C[h(V,)AV,],where the sum is taken over all i slices present in
the selected peak being processed. These values are then re-sorted into mass bins
divided in terms of a log,, M scaling. In this manner, similar mass contributions
at different elution volumes would be added together. Applied to the data of Fig.
5, for example, this procedure yields the very discontinuous result of Fig. 7.
Because of the uncertainty associated with the calculation of molar mass (the
ASTRA program calculates the precision of each measurement reported), one
might well consider distributing the mass fraction value into a range of bins rather
than into a single mass bin. Thus if the standard deviation of the calculated mass
is AM, then the mass fraction might be distributed following a Gaussian distribu-
tion over a range of bins spanning at least +2AM. Even if this procedure were
successful, the resulting distributions would still appear strange, and for good
PAMY 1
Norm = Log
3rd order
PAM91-2
1.OXI o6 1.OXI o7
Molar Mass (glrnol)
Figure 9 Overlay of the resorted data of Fig. 7 with the analytically fit data of Fig. 8.
390 Wyatt
reason! Light scattering measurements on occasion show features of the pro-

cessed sample that are by no means obvious.
Let us return to Fig. 5 and fit the data (smooth) to the simple third-order
polynomial indicated by the overlaying curve on Fig. 5. Applying the formalism
of Eq. (37) immediately results in the multivalued plot of Fig. 8: a strange, yet
expected, differential mass fraction distribution. Overlaying the resorted data of
Fig. 7 with the continuous differential mass fraction of Fig. 8 yields Fig. 9.
Clearly the two representations are similar, yet Fig. 8 shows more explicitly the
two almost distinct distributions of molecules present in the polyacrylamide sam-
ple. Even more instructive is an examination of the molecular conformation plot
for this non-GPC separation. Figure 10 presents a plot of the rms radii as a func-
tion of elution volume. Note that the reversal of the expected decreasing variation
of this rms radius occurs at the same elution volume where d(log,, M)/dV = 0.
Note, however, that the rms radius is not the same as the hydrodynamic radius,
so that if the separation is still believed to be governed by GPC mechanisms,
the drainage properties of the polymer must undergo significant changes beyond
Volume (mL)
Figure 10 The root mean square radius as a function of elution volume for the sample
of Fig. 5.
I .ox106 I .ox1o7
Molar Mass (glrnol)
Figure 11 The conformation plot for the polyacrylamide data of Figs. 5 and 10.
the point of the rms radius reversal. Figure 11 presents the conformation plot for
this molecular sample, which appears quite normal for molecules eluting before
the reversal point, yet which appears very aberrant and distorted for those frac-
tions eluting later in the so-called non-GPC mode. Matsumoto [I21 was the first
to associate the unusual appearance of the conformation plot with the onset of
microgel formation. Whatever the source, the MALS confirms that there is pres-
ent a fraction of the sample whose conformation is distinctively different than
that of a linear or even slightly branched polymer. Conformation plots and differ-
ential mass fraction distributions tell much about samples whose elution from
GPC columns don't follow the rules.
VI. EXAMPLES OF MALS APPLIED TO

OTHER BIOPOLYMERS
Unlike synthetic polymers and most natural polymers such as polysaccharides

and complex starches, proteins are produced with polydispersities that are essen-
392 Wyatt
tially unity. In other words, a protein such as bovine serum albumin is produced
at a single molar mass only as determined by the corresponding genetic code.
Naturally such proteins can form aggregates whose presence in various biologi-
cals can be immunogenic. Thus the detection of dimers, trimers, and higher order
aggregates can become an important task for the quality control of many types
of commercially sold biologicals.
Let us consider a few examples of the power of MALS combined with
differential refractive index (DRI) detection. In a recent paper by Riggs's group
at the University of Texas [13], the use of MALS with a DRI detector produced
the values for the broad range of values shown in Table 3. The values calculated
from the MALS measurements (DAWN-DSP operating at the He-Ne laser wave-
length 632.8 nm) are contrasted to the values calculated based on the expected
base pair sequences. A single value of dnldc (0.19) was used in their measure-
ments using a phosphate buffer saline mobile phase and two Toya Soda GPC
columns described in the article.
Light scattering measurements are particularly useful for proteins whenever
a DRI detector can be used due to the near-constancy of dnldc. This is in contrast
to reversed-phase chromatography where a UV detector must be used because
of the varying refractive index of the mobile phase. This in turn requires the
measurement of the corresponding UV extinction coefficient for each protein
species. For certain closely related proteins, where a single extinction coefficient
may be chosen, the light scattering results for determining molar masses of sepa-
rated species is equally impressive. Figure 12 shows the UV chromatogram from
a set of isolated wheat proteins whose corresponding sequence values are shown
in Table 4. These proteins were separated by reversed-chromatography using a
gradient of acetyl nitrate in water over the range of 0 to 15%. What is particularly
significant about this type of separation is that the elution sequence is not gener-
Table 3 Measured Molar Masses of a Broad Range of Proteins

Mass from Light Apparent
structure scattering error
Protein Pa) (D4 (%)
Carbonic anhydrase 29,023 29,800 +2.7

Alcohol dehydrogenase 146,980 149,000 + 1.4
b-Amylase 224,340 228,000 +1.6
Apoferritin 476,3 16 484,400 + 1.7
Thyroglobulin 669,000 679,000 + 1.5
Omithine decarboxylase 990,684 978.000 - 1.3
Octopus hernocyanin 3,440,000 3,450,000 +0.3
0 5 10 15
Volume (mL)
Figure 12 UV chromatogram from a set of isolated wheat proteins whose molar mass
and size were calculated from MALS measurements. See Table 4 for details.
ally in any particular order. There is no means to calibrate the columns used for
reversed-phase separation as is often the case for GPC separations.
Figure 13 illustrates an important element of MALS combined with a con-
centration-sensitive detector. Shown here are three differential mass fraction dis-
tributions [cf. Eq. (37)] of specially processed recombinant DNA-produced hya-
loronic acid. Although all overlap broadly with each other, the direct comparison
of their differential mass fractions permits an easy discrimination between them.
Finally, the process of adding polyethylene glycol (PEG) to proteins (so-
called pegylation), which has become of considerable importance as a time re-
lease drug delivery process, produces a copolymer whose core is a pure, monodis-
perse protein molecule. By iteratively calculating the copolymer mass at each
elution volume (note that each component has a different d d d c value), staff
Table 4 Calculated Molar Mass and Size for Each of the

Four Wheat Protein Peaks Identified in Fig. 12 Contrasted
with Their DNA Calculated Values
RMS radius
Molar mass (kD) (nm)
Proteins cDNA DAWN DSP DAWN DSP
1 68.7 67.6 + 0.3 11.7+ 1.5
2 87.2 87.4 + 0.4 11.4 t 1.5
3 83. I 79.7 2 0.4 11.7+ 1.5
4 NIA 70.6 2 0.7 13.8 5 2.6
Figure 13 Differential weight fraction distributions for three hyaluronic acid samples.
- Refractometer
Signal overlaid
8.0 9.0 10.0 1 1 .0 12.0

Elution volume
Figure 14 Variation of molar mass with elution volume for a pegylated protein sample.
Used a self-consistent iterative method to calculate the number of PEGIprotein.
member Robert Paulson in an unpublished Application Note [14] has been able
to calculate not only the total molar mass of each eluting fraction but the number
of PEG molecules (molar mass about 15,000) attached to each protein molecule
(molar mass about 150,000), as illustrated in Fig. 14.
VII. CONCLUDING REMARKS
MALS measurements following fractionation by GPC, reversed-phase, or other

techniques represent one of the most powerful means for studying the solution
properties of polymers. The distinctions between MALS and single- or dual-angle
light scattering measurements have been discussed together with the effects of
the latter measurements on the errors of the derived quantities (molar mass, size,
conformation, etc.).
Discussion of the various "moments" of the molar mass distributions has
shown that for all but random coil molecules in a theta solvent, the light scattering
measurement yields an unusual type of moment associated with it. Other mass
and size moments derived from measurements of fractionated samples have been
explained. For light scattering measurements of homopolymers, MALS determi-
nation of rms radii may be made (at the low concentrations characteristic of
chromatographically separated samples) without a concentration detector.
The unusual properties of samples whose separation appears in conflict
with that expected by normal HPLC means have been studied and shown most
vividly through the so-called conformation plot as well as the differential mass
fraction distribution. The power of the MALS measurement for understanding
complex macromolecular structures has been illustrated by these examples.
ACKNOWLEDGMENTS
This chapter is based on a paper presented at the Waters GPC Symposium, Sep-
tember 1996, San Diego, CA. The exceptional laboratory results, especially the
data shown in Figs. 10 through 12, were made under the direction of Dr. Michelle
H. Chen. Other individuals who made contributions include Dr. Greg Cauchon,
Lena Nilsson, and David Villalpando.
REFERENCES
1. Huglin MB, ed. Light Scattering from Polymer Solutions. London: Academic Press,
1972.
2. Ouano AC. Kaye W. J Polym Sci. A-l 1974; 12:1151.
396 Wyatt
Moore JC. J Polym Sci A 1964;2:835.

Burchard W, Cowie JMG. In: Huglin MB, ed. Light Scattering from Polymer Solu-
tions. London: Academic Press, 1972.
Zimm BH. J Chem Phys 1948;16:1093; ibid., 1099.
Wyatt PJ. Anal Chim Acta l993;272: 1-40.
Berkowitz SA. Rejection of spike noise from SECILALLS experiments. Anal Chem
1986;58:2571.
Kratochvil P. In: Huglin MB, ed. Light Scattering from Polymer Solutions. London:
Academic Press, 1972.
Benoit H, Froelich D. In: Huglin MB, ed. Light Scattering from Polymer Solutions.
London: Academic Press, 1972.
Billingham NC. Molar Mass Measurements in Polymer Science. New York: John
Wiley and Sons, 1977.
Shortt DW. J Liquid Chromatogr 1993; 16:3371-3391.
Matsumoto A. American Chemical Society National Meeting, Sept. 1990, Washing-
ton, D.C.
Zhu H, Ownby DW, Riggs CK, Nolasco NJ, Stoops JK, Riggs AF. Assembly of
the gigantic hemoglobin of the earthworm lumbricus terrestris roles of subunit equi-
libria, non-globin linker chains, and valence of the heme iron globin. J Biol Chem
1996;271:30007-30021.
Cf. the Worldwide Web site www,wyatt.com.
Purification and Characterization
of Connective Tissue Growth
Factor Using Heparin Affinity
Chromatography
David R. Brigstock
Ohio State University and Children's Hospital, Columbus, Ohio
I. INTRODUCTION
The term connective tissue growth factor (CTGF) was first used in 1991 to de-
scribe a mitogenic and chemotactic factor for fibroblasts that was produced by
human umbilical vein endothelial cells (HUVECs) in vitro [ I 1. The primary trans-
lational products of both human CTGF (hCTGF) and porcine CTGF (pCTGF)
are predicted to comprise 349 residues, the first 26 of which are a presumptive
signal peptide [1,2]. Mouse CTGF (mCTGF; also termedjsp-12 or PIG-M2) is
predicted to comprise 348 residues, of which the first 25 are a hydrophobic signal
peptide [3,4]. Secreted forms of CTGF from all three species are thus predicted
to comprise 323 residues, 38 of which are conserved cysteine residues. Metabolic
labeling and immunoprecipitation studies of cell lysates and conditioned medium
have shown that CTGF proteins of approximately 38 kDa are produced and se-
creted in various cell cultures [2,5,6]. CTGF, which is encoded by a transforming
growth factor-P (TGFP)-inducible immediate early gene [3], has been strongly
implicated in a variety of fibrotic disorders [7-I I], wound healing [12-141, em-
bryonic development [ 5 ] , and uterine function [2,15]. Although the 38-kDa form
of CTGF was initially implicated in many of these processes, low-mass forms
of CTGF of M, 10,000-20,000 have since been discovered in uterine secretory
fluids [ 151 and fibroblast conditioned medium [6].
397
398 Brigstock
Partial purification of 38-kDa hCTGF was first achieved using immuno-

affinity chromatography with an antiserum to intact platelet-derived growth factor
(PDGF) [I]. Since PDGF and CTGF are not structurally related, the reactivity
of CTGF with anti-PDGF was serendipitous and attributed to the presence of
cross-reactive epitopes on the CTGF and PDGF proteins [I]. More recently, hepa-
rin affinity chromatography has been used as an essential step in the isolation of
pCTGF from complex body fluids such as uterine secretions 1151, as well as for
the large-scale purification of recombinant hCTGF produced in a buculovirus
expression system [16]. This chapter will demonstrate the use of heparin affinity
fast protein liquid chromatography (FPLC) and reversed-phase high-performance
liquid chromatography (RP-HPLC) for the isolation and characterization of a 10-
kDa form of CTGF in pig uterine fluids.
II. MATERIALS AND METHODS

A. Uterine Secretory Fluids
Uteri were collected at random from pigs that were sacrificed at a local slaughter-
house. After transportation to the laboratory, the lumen of each uterine horn was
flushed with 10-50 ml phosphate-buffered saline (PBS). A single collection of
uterine luminal flushings (ULF) consisted of approximately 600-1 100 ml ob-
tained from 16-26 uteri. ULF were clarified by centrifugation at 13,500g for
30 min at 4°C and the supernatant was passed through glass wool to remove
floating material. Comparative heparin affinity FPLC was performed on ULF
obtained from cycling or pregnant animals (n = 6 per group) that were sacrificed
on day 1 1 , the time of onset of blastocyst elongation and peak estrogen production
in the pig [17-201.
6. Chromatography
1. Cation Exchange Chromatography
ULF were applied at 3.5 mllmin to a Bio-Rex 70 cation exchange column
(5 X 6 cm; Bio-Rad Laboratories, Richmond, CA, USA) which was then washed
with 1000 ml of PBS containing 0.2 M NaCl and subsequently developed with
a 500-ml gradient of 0.2-2.0 M NaCl in PBS. These procedures were performed
at 4°C. Fractions (10-ml) were collected during salt gradient elution of bound
proteins and were assayed for their ability to stimulate [3H]thymidineincorpora-
tion in Balblc 3T3 cells. The 0.3-0.7 M NaCl elute was selected for further
analysis by heparin affinity FPLC.
Purification and Characterization of CTGF
2. First-Step Heparin Affinity FPLC

The 0.3-0.7 M NaCl eluate from the Bio-Rex column was diluted threefold with
20 mM Tris-HC1 (pH 7.4) containing 0.1 % (wlv) 3-[(3-cho1amidopropyI)dimeth-
ylammonio]-1-propanesulfonate(CHAPS), clarified using a low protein binding
0.45-pm HT Tuffryn Acrodisc filter (Gelman Sciences, Ann Arbor, MI, USA),
and applied at room temperature to an EconoPac heparin column (0.7 X 3.6 cm,
Bio-Rad) at 2 mllmin using a peristaltic pump. The column was then attached
to an FPLC system (Pharmacia LKB Biotech Inc., Piscataway, NJ, USA), washed
with 50 ml 20 mM Tris-HC110.2 M NaCl/O.l% CHAPS (pH 7.4), and developed
at 1 mllmin with a 40-ml 0.2-2.0 M NaCl gradient in 20 mM Tris-HC1/0.1%
CHAPS (pH 7.4). Eluted proteins were collected into 1-ml fractions and tested
for their mitogenic activity. The absorbance of the column eluate was monitored
at 280 nm using an in-line UV monitor.
3. Second-Step Heparin Affinity FPLC

The 0.8 M NaCl eluates from nine individual EconoPac heparin purifications
were pooled, diluted threefold with 20 mM Tris-HC1 (pH 7.4), clarified by pas-
sage through a 0.2-pm HT Tuffryn filter membrane (Gelman Sciences), and ap-
plied at room temperature at 1 mllmin to a TSK heparin 5PW column (0.8 X
7.5 cm; TosoHaas, Philadelphia, PA, USA) using a Pharmacia FPLC system.
The column was washed and developed as described above for the EconoPac
heparin column except that CHAPS was omitted from the buffers to prevent
interference in the subsequent HPLC step.
4. First-Step Reversed-Phase HPLC

The 0.8 M NaCl eluate from the TSK heparin column was adjusted to 10%
acetonitrile10.1% trifluoroacetic acid (TFA), clarified using a 0.2-pm HT Tuffryn
Acrodisc filter (Gelman Sciences), and applied to a preequilibrated C8 HPLC
column (0.46 X 25 cm, 5 pm; Rainin Instrument Co, Woburn, MA, USA) using
a Hitachi HPLC system (Hitachi Instruments Inc., Danbury, CT, USA). After
sample application, the column was washed with 10 ml of 10% acetonitrilelo. 1%
TFA prior to elution of bound proteins with a 10-90% acetonitrile gradient (in
waterlo. 1% TFA) over 136 min. This step was performed at room temperature
and the flow rate was 1 mllmin. The absorbance of the column eluate was moni-
tored at 214 nm using an in-line UV monitor. The eluate was collected as 0.5-ml
fractions in siliconized tubes containing 50 pl of 125 mM NaOH. Aliquots of
selected fractions were dried by evaporation using a SpeedVac concentrator
(Savant Instruments, Farmingdale, NY, USA) and reconstituted in 20 mM Tris-
HCI (pH 7.4) for SDS-PAGE and for analysis of their mitogenic activity.
Brigstock
5. Second-Step Reversed-Phase HPLC

Bioactive fractions from the first HPLC step were pooled, diluted fivefold with
waterlo. 1% TFA, and applied to a Rainin C 8 HPLC column (0.46 X 25 cm, 5
pm) that was washed and developed with acetonitrile in water/O.l% TFA as
described above. Selected fractions were analyzed by SDS-PAGE and mitogenic
activity as described above. A single 10-kDa protein that was present exclusively
in biologically active fractions was selected for amino acid sequencing.
C. Protein Sequencing
The HPLC-purified IOkDa protein was evaporated to dryness using a SpeedVac
concentrator, redissolved in SDS-PAGE sample buffer, subjected to SDS-PAGE,
transferred to nitrocellulose, and located by staining the blot with Coomassie
R250. The protein was then excised and subjected to N-terminal microsequencing
on a model 470A gas phase sequenator (Applied BioSystems, Foster City, CA,
USA). Phenylthiohydantoin derivatives in each cycle were separated using C18
RP-HPLC. The sequence was assigned using Applied BioSystems sequence anal-
ysis software, with independent verification by an experienced operator with no
knowledge of the likely identity of the sample.
D. SDS-PAGE and Silver Staining

Samples were electrophoresed on a 18% SDS polyacrylamide gel under reducing
conditions at 200 V for 1 h. For analytical work, proteins in the gels were stained
with silver as described [2 11. For amino acid sequencing, the 10-kDa protein was
transferred from the gels to nitrocellulose in 10 mM 3-(cyc1ohexylamino)pro-
panesulfonic acid (pH 11.0) at 300 mA for 2 h.
E. Bioassay
Column fractions were tested for their stimulation of [3H]thymidine incorpora-
tion into the DNA of quiescent Balblc 3T3 cells. Neutralization studies were
performed by coincubation of the 0.8 M NaCl heparin column eluate with
25 pglml anti-PDGF IgG (Upstate Biotechnology Inc., Lake Placid, NY, USA)
or 10 pllml basic fibroblast growth factor (bFGF) antiserum ("77R," kindly
donated by Dr. M. Klagsbrun, Children's Hospital, Boston, MA, USA). PDGF
and bFGF standards were from R&D Systems (Minneapolis, MN, USA) and Life
Technologies (Grand Island, NY).
F. CTGF Peptide Synthesis

A peptide corresponding to residues 247-260 of pCTGF (EENIKKGKKCIRTP)
was produced on a Synergy 432A peptide synthesizer (Applied BioSystems) and
purified by RP-HPLC using a C18 column as described [IS]. One milligram of
the peptide in 2 ml of 10 mM Tris-HC1 (pH 7.4) was then applied to a TSK
heparin column, which was washed and developed as described above. The elu-
tion position of the peptide was determined spectrophotometrically at 214 nm.
Ill. RESULTS
A. Ion Exchange Chromatography and Heparin
Affinity FPLC
When ULF were applied to a Bio-Rex 70 column, the flow-through fraction was
shown to contain high-mass forms of epidermal growth factor (EGF) that stimu-
lated DNA synthesis in 3T3 cells [22]. However, material which bound to the
column and was eluted by 0.3-0.7 M NaCl was also shown to be mitogenic for
3T3 cells (Fig. 1). Affinity chromatography of this eluate using EconoPac heparin
NaCl (M)
A 0.0
0 10 20 30 40 50
Fraction Number
Figure 1 Cation exchange chromatography of pig ULF. Approximately 1 100 ml of ULF

from 26 animals was applied to a Bio-Rex cation exchange column at 3.5 mllmin. The
column was then washed with 1000 ml PBS10.2 M NaCl and treated with a 500-rnl gradi-
ent of 0.2-2.0 M NaCl in PBS. Fractions (10 ml) were collected and assayed at 50 pll
ml for their ability to stimulate [3H]thymidine incorporation in Balblc 3T3 cells.
402 Brigstock
columns resulted in the elution of four peaks of growth factor activity over a
0.2-2.0 M NaCl gradient (Fig. 2). The first activity peak was eluted by 0.5 M
NaCl and shown to contain a PDGF-like factor [23]. The second activity peak
was eluted by 0.8 M NaCl and contained a novel factor that was termed HBGF-
0.8 prior to its definitive identification as a 10-kDa form of CTGF [15]. The third
activity peak was eluted by approximately 1 M NaCl and contained heparin-
binding EGF-like growth factor (HB-EGF) and pleiotrophin (PTN) [23,24]. The
fourth activity peak was eluted by 1.6 M NaCl and was due to the presence of
bFGF [I 5,23,25].
To further purify 10-kDa CTGF, fractions containing the 0.8 M NaCl eluate
from the EconoPac heparin column were pooled, diluted, and subjected to a sec-
ond step of heparin affinity FPLC using a TSK heparin column. A peak of growth
factor activity was subsequently eluted by 0.8 M NaCl (Fig. 3), which in some
separations could be resolved into two microheterogeneous forms of 10-kDa
CTGF [15]. To show that 10-kDa CTGF was immunologically distinct from
NaCl (M)
0.2 0.65 1.1 1.55 2.0
0 10 20 30 40
Fraction Number
Figure 2 First-step heparin affinity FPLC. Mitogenic fractions from cation exchange
chromatography of 900 ml of ULF from 22 animals were pooled, diluted to three-fold,
and applied to an EconoPac heparin column, which was then washed with 10 m120 mM
Tris-HClf0.2 M NaC1/0.1% CHAPS (pH 7.4) and developed with a 40-ml gradient of
0.2-2.0 M NaCl in 20 mM Tris-HC1/0.1% CHAPS (pH 7.4). Fractions (1-ml) were col-
lected during NaCl gradient elution of the bound proteins and tested at 15 yllrnl for their
stimulation of [3H]thyrnidineincorporation in Balblc 3T3 cells. The figure shows the iden-
tity of various heparin-binding growth factors that were separated from one another.
NaCl (M)
0.2 0.65 1.1 1.55 2.0
300 1 I I 1 0.60 1
Fraction Number
Figure 3 Second-step heparin affinity FPLC. Nine individual samples of ULF (7.15-L
total volume from 206 pigs) were individually processed by cation exchange and EconoPac
heparin affinity chromatography as shown in Figs. 1 and 2. The 0.8 M NaCl eluates from
each heparin affinity purification were pooled, diluted, and applied to a TSK heparin col-
umn which was washed with 10 ml of 20 mM Tris-HC110.2 M NaCl (pH 7.4) and devel-
oped with a 40-ml gradient of 0.2-2.0 M NaCl in 20 mM Tris-HCI (pH 7.4). Fractions
(I-ml) were collected during NaCl gradient elution of the bound proteins and tested at 5
pllml for their stimulation of ["Hlthymidine incorporation in Balblc 3T3 cells.
bFGF or PDGF and was not contaminated by either of these factors, it was incu-
bated with neutralizing antisera to PDGF or bFGF. As shown in Table 1, CTGF
mitogenic activity was unaffected by either of these antisera, which were nonethe-
less very effective in antagonizing the activity of their respective ligands.
8. Reversed-Phase HPLC
Isolation of 10-kDa CTGF to homogeneity was achieved by performing two steps
of C, RP-HPLC on heparin-purified samples. Results from the second HPLC step
are presented in Fig. 4 that show the elution of a single peak of growth factor
activity at 45 min (28% acetonitrile). The elution profile of the mitogenic activity
was directly correlated with that of a single peak of protein, which was of M,,
10,000 as assessed by SDS-PAGE. Based on silver staining of SDS polyacryl-
amide gels, no proteins other than the 10-kDa polypeptide were present in any
of the biologically active fractions (Fig. 4).
Brigstock
Table 1 Lack of Effect of Neutralizing Anti-PDGF or Anti-bFGF

on CTGF Mitogenic Activity
['HIThymidine incorporation (cpml
Growth factor well x 10-l) (mean + SD)
Alone +anti-PDGF IgG
No addition 899 + 99 956 ? 20
PDGF (10 nglml) 21,634 ? 6,664 +
938 20
HBGF-0.8 (15 kllml) 47,677 ? 1,316 +
62,045 6,645
Alone +bFGF antiserurri
No addition 825 5 143 7,575 + 369
bFGF (6 nglrnl) 62,365 ? 2,669 7,615 ? 1,519
HBGF-0.8 (15 kllml) 85,152 + 1,330 83,238 & 2.538
C. Amino Acid Sequencing

Amino acid sequencing of the HPLC-purified protein revealed the N-terminal
sequence Glu-Glu-Asn-Ile-Lys-Ly s-Gly-Lys-Ly s-Xaa-Ile-Arg-Thr-Pro-Lys-Ile
(Table 2). This sequence was identical to residues 247-262 of hCTGF or pCTGF,
and to residues 246-261 of mCTGF, with the unidentified residue in cycle 10
corresponding to a cysteine residue in the predicted protein sequences from all
three species (Table 2). Although CTGF is structurally related to cyr6llcefl0
and nov(26-32), the protein sequence obtained experimentally aligned less well
with the corresponding internal sequence of these proteins, demonstrating that the
10-kDa protein was a low-mass form of CTGF that was substantially truncated at
its N terminus (Table 2).
D. Biological Activity of 10-kDa CTGF

As well as stimulating DNA synthesis in mouse Balblc 3T3 cells, submaximal
stimulatory doses of 10-kDa CTGF were shown to potentiate the activity of sev-
eral other 3T3 cell mitogens including EGF, bFGF, PDGF, and insulin-like
growth factor-1 (IGF-1) (Table 3). The mitogenic activity of 10-kDa CTGF was
highly susceptible to inactivation by exposure to heat or acid (Fig. 5). The 10-
kDa CTGF protein was also found to stimulate [.'H]thymidine incorporation in
primary cultures of pig endometrial stromal cells, suggesting that as a product
of uterine tissues CTGF may act via autocrine or paracrine growth stimulatory
pathways within the uterus (Fig. 6). Finally, analysis of the amount of 3T3 cell
Retention Time (mind
Figure 4 Second-step RP-HPLC. Two peak active fractions from first-step HPLC puri-
fication (data not shown) were subjected to second-step HPLC purification as described
in "Materials and Methods." The figure shows the elution of protein as measured at 3-14
nm and the ability of 20 FI of selected fractions to stimulate 3T3 cell DNA synthesis after
drying and reconstitution in 10 pl PBSIO.I% BSA. The inset shows a silver-stained SDS
polyacrylamide gel of 50-yl aliquots of successive fractions containing the peak of rnito-
genic activity.
406 Brigstock
Table 2 N-Terminal Amino Acid Sequences of HBGF-0.8 and Members of the

CTGF Familya
Protein Residues Sequence Ref.
HBGF-0.8 1-16 EENIKKGKKXIRTPKI 15
Human CTGF 247-262 1
Pig CTGF 247-262 2,15
Mouse CTGF 246-261 3,4
Mouse cyr6 1 275-290 3,27
Human cyr61 277-292 28
Chick ceflO 272-29 1 29
Chick nov 249-264 30
Human nov 255-270 31
Xenopus nov 240-255 32
Washes indicate residues that are identical to those of HBGF-0.8.
mitogenic activity in the 0.8 M NaCl eluate after two rounds of heparin affinity
FPLC showed that ULFs from pigs on day 11 of pregnancy contained approxi-
mately 5 times the level of CTGF-like activity as ULF from pigs on day 11 of
the estrous cycle (Fig. 7).
E. Heparin Binding Properties of N Terminus

of 10-kDa CTGF
To test the heparin binding properties of the N-terminal region of 10-kDa CTGF,
a synthetic peptide corresponding to residues 247-260 of the pCTGF primary
Table 3 Potentiation of Growth Factor Activity by 10-kDa CTGFa

[3H]Thymidineincorporation (cpml
well X 10-3j (mean + s.d.)
Treatment No addition + 10-kDa CTGF
None 293 I+- 42 2,843 ? 280
IGF-I (IOngIml) 621 -C 324 7,699 2 996
EGF (3nglml) 5,418 + 1,031 49,910 2 5,457
PDGF (IOngIml) 36,509 + 7,407 104,340 + 23,742
bFGF (0.3nglml) 21,095 2 2,612 46,836 I+- 17,091
Calf serum (250~llml) 71,731 2 11,992 -
"eparin-purified 10-kDa CTGF was added at a submaximal dose to triplicate

wells of Balblc 3T3 cells in the presence of the indicated concentrations of
other 3T3 cell mitogens.
Figure 5 Heat and acid treatment of 10-kDa CTGF. Heparin-purified 10-kDa CTGF
was incubated at pH 7.4 for 30 min at room temperature (a),pH 7.4 for 30 min at 56°C
(U), pH 7.4 for 5 min at 100°C (m), or pH 1.5 for 3 min at room temperature (0).Sanlples
were assayed for their stimulation of 3T3 cell DNA synthesis at equivalent doses. Values
shown are mean +SD of triplicate determination of [%]thymidine incorporation.
No IOkDa CTGF
addition 20pllrnl
Figure 6 Stimulation of pig endometrial cell DNA synthesis by 10-kDa CTGF. Primary
cultures of pig endometrial cells were treated for 18 h with 20 pllml heparin-purified
10-kDa CTGF (a concentration that induced maximal stimulation of DNA synthesis in
Balblc 3T3 cells) after which they were incubated with I pCi/ml ['Hlthymidine for 6 h.
Brigstock
Fract~onNumber
Figure 7 Heparin affinity FPLC of ULF from day 11 pregnant or cycling pigs. ULF
from day I I pregnant (@) or cycling (0) pigs (n = 6 per group) were subjected to two
rounds of heparin affinity purification essentially as shown in Figs. 2 and 3 except that
no Bio-Rex purification was performed and the first and second heparin affinity steps were
performed using heparin-Sepharose and EconoPac heparin columns, respectively. The
figure shows the stimulation of 3T3 cell DNA synthesis by selected fractions from the
second heparin affinity step when assayed at 2.5 pllml on 3T3 cells. The levels of CTGF
in ULF were 45.1 units in pregnant animals versus 8.8 units in nonpregnant animals, where
one unit is the amount of each growth factor per ml that was required to elicit half the
['Hjthymidine incorporation as 20% calf serum.
translational product (i.e., the N terminus of 10-kDa CTGF; see Table 2) was
subjected to heparin affinity FPLC. Although the pI of this peptide is the same
as that of PDGF (9.6) which is eluted from heparin affinity columns by 0.5 M
NaCl 133,341, the CTGF peptide required 0.7 M NaCl for elution from a TSK
heparin column (Fig. 8). These results showed that the peptide bound more
strongly to heparin than could be explained by its isoelectric point alone and
were suggestive of the presence of specific heparin-binding determinants in the
peptide.
IV. DISCUSSION
A. Heparin Affinity Purification of CTGF
Over the last 15 years, heparin affinity chromatography has proven a particularly
powerful means by which a variety of polypeptide growth factors have been
characterized and purified. These include acidic FGF (aFGF), bFGF, HB-EGF,
Fraction Number
Figure 8 Heparin aftinity FPLC of CTGF[247-2601. One milligram of CTGF[247-

2601 peptide was applied to a TSK heparin column in 20 mM Tris-HC1 (pH 7.4) containing
0.2 M NaCI. The column was washed with 10 ml of 20 mM Tris-HC110.2 M NaCl (pH
7.4) and bound material was eluted from the column with a 0.2-2 M NaCl gradient in
20 mM Tris-HC1 (pH 7.4). Fractions of 1 ml were collected throughout and measured at
214 nm.
PTN, vascular endothelial growth factor, keratinocyte growth factor, and amphi-
regulin [35-411. Since many of these factors exhibit different affinities for hepa-
rin, the relative binding of these factors to heparin has been a particularly useful
criterion for characterizing growth factor of unknown identity, especially when
used in conjunction with immunological (e.g., western blot) and biological (e.g.,
activity, target cell specificity) assays. Indeed, CTGF was first recognized as a
fibroblast rnitogen in uterine secretory fluids in 1989 as a result of its atypical
chromatographic behavior on heparin-agarose [25]. Key features that were subse-
quently used to distinguish it from other well-characterized growth factors (in-
cluding FGFs with which it was initially thought to be related [25]) were its !I)
elution from heparin by 0.8 M NaCI, (2) molecular mass of 10,000 as determined
by nondenaturing gel filtration and SDS-PAGE, (3) target cell specificity (mito-
genic for fibroblasts and smooth muscle cells but not endothelial cells), (4) heat
and acid lability, (5) potentiation of bFGF-, EGF-, PDGF-, or IGF-induced mito-
genic activity, and (6) lack of reactivity with PDGF or bFGF antisem (these
results and [15]). However, it was not until the development of this purification
protocol that the factor was isolated and definitively identified as a 10-kDa form
of CTGF. Heparin affinity chromatography not only played an important role
in the initial purification and characterization of 10-kDa CTGF, but also in the
separation of two microheterogeneous forms of the protein that differed by the
410 Brigstock
presence or absence of a single glutarnic acid residue (Glu"') at the N terminus

(see below and [15]. Using a similar purification strategy, heparin-binding 10-
to 12-kDa forms of CTGF have also been isolated from serum-free conditioned
medium of mouse or human fibroblasts maintained in vitro [6].
Recently, heparin affinity chromatography was used to substantially purify
recombinant human CTGF in a single-step procedure [16]. In those experiments,
38-kDa hCTGF produced by insect cells following infection with recombinant
baculovirus was purified by heparin-Sepharose chromatography, yielding a prep-
aration that was >95% pure as assessed by Coornassie staining of SDS-PAGE
gels 1161. Metabolically labeled native 38-kDa mCTGF, produced by NIH 3T3
cells, has been shown to be eluted from heparin-agarose beads by 0.4 M NaCl
[5]. Recently 1 mg of highly purified recombinant 38-kDa mCTGF was purified
from 500 ml of serum-free conditioned medium from recombinant baculovirus-
infected insect cells using Sepharose S cation exchange chromatography [5].
Since heparin affinity columns and S ion exchange columns both contain cationic
methyl sulfonate groups, it is possible that similar mechanisms are responsible
for the binding of CTGF to each type of matrix.
B. Semiquantitative Assessment of CTGF Levels

by Heparin Affinity Chromatography
Peaks of growth factor activity that are eluted from heparin columns have been
previously used to assess the relative levels of individual HBGFs in various sam-
ples. For example, the levels of bioactive aFGF and bFGF, eluting from heparin
affinity columns by 1.0 and 1.5 M NaCI, respectively, have been compared in
postnatal rat brain on days 10 and 40 [42] and in lysates of cultured smooth
muscle cells or endothelial cells [43]. Similarly, levels of PDGF and HB-EGF
in, respectively, the 0.5 M NaCl and 1 M NaCl eluates from heparin-Sepharose
columns were compared in conditioned medium from mononuclear cells that had
been placed in culture for 1-2 days or 7-1 1 days [34]. For CTGF, this type of
semiquantitative approach suggested that there is approximately fivefold more
bioactive CTGF in uterine fluids from day 11 pregnant pigs than day 11 nonpreg-
nant pigs. These initial findings suggest either that the embryo contributes to the
pool of soluble CTGF in the uterine lumen or that there is greater uterine synthesis
andlor release of CTGF during pregnancy.
C. Heparin Affinity Mechanisms of CTGF

While our current knowledge of CTGF-heparin interactions is sparse, we and
others have shown that heparin is able to modulate CTGF mitogenic activity
[15,16]. Although analysis of the C-terminal 103 residues of CTGF revealed that
10 kDa has a net basic charge (PI 8.3), this property alone is unlikely to account
Purification and Characterization of CTGF 41 1
for the degree of heparin binding observed since 10-kDa CTGF has a lower iso-
electric point than PDGF (PI 9.6) yet requires a higher salt concentration than
PDGF for elution from heparin (0.8 M NaCl versus 0.5 M NaC1). As we have
previously reported, residues 250-255 resemble a heparin-binding consensus se-
quence and there is a very high proportion (50%) of basic amino acids (K, R)
in the 17-residue region CTGF[251-2671 1151. Evidence that the N terminus of
10-kDa CTGF may be involved in heparin binding is illustrated by ( I ) the require-
ment for 0.7 M NaCl to elute the peptide CTGF[247-2601 from a TSK heparin
column, which, by comparison to PDGF, is a higher salt concentration than would
be predicted from its pI alone; (2) the observation that CTGF[247-2601 binds
[3H]heparin more strongly than peptides that flank this sequence 1151; and (3)
the finding that a form of 10-kDa CTGF that commences at G ~ u ~exhibits
''~ greater
heparin binding than the form described here, which commenced at G ~ (and u ~ ~ ~
thus differs by the presence of a single acidic residue at its N terminus) [15].
While these data suggest that residues 247-260 are involved in heparin binding,
analysis of [3H]heparinbinding by a panel of peptides that spanned the C-terminal
103 residues of CTGF suggested that additional domains, including residues 305-
328, may also be involved in the interaction between 10-kDa CTGF and heparin
1151. While it has been proposed that residues 206-214 of CTGF resemble a
binding motif for sulfated glycoconjugates 1261, the involvement, if any, of this
domain in mediating interactions between larger forms of CTGF (e.g., the full-
length 38-kDa CTGF protein) and heparin has yet to be clarified.
D. Conclusions
Heparin affinity chromatography has played a key step in the isolation of native
CTGF from uterine secretions and recombinant CTGF from baculovirus expres-
sion systems. While the interaction of CTGF with heparin has been exploited for
the purification of 10-kDa and 38-kDa forms of CTGF, this property appears to
be the result of the presence of specific heparin binding motifs within the CTGF
molecule and may have physiological consequences in that heparin is a regulator
of CTGF biological activity. Future studies will undoubtedly focus on additional
structural and functional aspects of heparin binding by CTGF, including the pos-
sible role that cell surface heparan sulfate proteoglycans play in regulating the
bioavailability of CTGF and its binding and activation of cell surface receptors.
ACKNOWLEDGMENTS
I am grateful to Christy Steffen and Greg Kim for excellent technical help, Bill
Pope for providing timed cyclic and pregnant pigs, John Lowbridge for protein
412 Brigstock
sequencing, and Brad Baker for peptide synthesis. This work was supported by
NIH Grant HD30334 awarded to D.R.B.
REFERENCES
Bradham DM. Igarashi A, Potter RL, Grotendorst GR. Connective tissue growth
factor: a cysteine-rich mitogen secreted by human vascular endothelial cells is re-
lated to the SRC-induced immediate early gene product CEF-10. J Cell Biol 1991;
1 14:1285- 1294.
Harding PA, Surveyor GA, Brigstock DR. Characterization of pig connective tissue
growth factor (CTGF) cDNA, mRNA and protein from uterine tissue. DNA Se-
quence 1998;8:385-390.
Brunner A, Chinn J, Neubauer M, Purchio AF. Identification of a gene family regu-
lated by transfom~inggrowth factor-p. DNA Cell Biol 199 1 ;10:293-300.
Ryseck R-P, Macdonald-Bravo H, Mattei M-G, Bravo R. Structure, mapping and
expression of fisp- 12, a growth-factor-inducible gene encoding a secreted cysteine-
rich protein. Cell Growth Differ 1991;2:225-233.
Kireeva ML, Latinic BV, Kolesnikova TV, Chen C-C, Yang GP, Abler AS, Lau
LF. Cyr61 and Fispl2 are both ECM-associated signalling molecules: activities. me-
tabolism, and localization during development. Exp Cell Res 1997:233:63-77.
Steffen CL, Ball-Mirth DK, Harding PA, Bhattacharyya N, Pillai S, Brigstock DR.
Characterization of cell-associated and soluble forms of connective tissue growth
factor (CTGF) produced by fibroblast cells in vitro. Growth Factors 1998;15:199-
213.
Frazier KS, Grotendorst GR. Expression of connective tissue growth factor mRNA
in the fibrous stroma of mammary tumors. Int J Biochem Cell Biol 1997;29:153- 161.
Igarashi A, Nashiro K, Kikuchi K, Sato S, Ihn H, Grotendorst GR, Takehara K.
Significant correlation between connective tissue growth factor gene expression and
skin sclerosis in tissue sections from patients with systemic sclerosis. J Invest Der-
mat01 1995; 105:280-284.
Igarashi A, Nashiro K, Kikuchi K, Sato S, Ihn H, Fujimoto M, Grotendorst GR,
Takehara K. Connective tissue growth factor gene expression in tissue sections from
localized scleroderma, keloid, and other fibrotic skin disorders. J Invest Dermatol
1996;106:729-733.
Oemar BS, Werner A,, Gamier J-M, Do, D-D, Godoy N, Nauck M, Marz W, Rupp
J, Pech M, Luscher TF. Human connective tissue growth factor is expressed in ad-
vanced atherosclerotic lesions. Circulation 1997;95:83 1-839.
Shinozaki M, Kawara S, Hayashi N, Kakinuma T, Igarashi A, Takehara K. Induction
of subcutaneous tissue fibrosis in newborn mice by transforming growth factor P-
simultaneous application with basic fibroblast growth factor casues persistent fibro-
sis. Biochem Biophys Res Commun 1997;237:292-296.
Igarashi A, Okochi H, Bradham DM. Grotendorst GR. Regulation of connective
tissue growth factor gene expression in human skin fibroblasts and during wound
repair. Mol Biol Cell 1993;4:637-645.
13. ~ a m m e MS,
s Lieske JC, Pawar S, Spargo BH, Toback FG. Calcium oxalate mono-
Purification and Characterization of CTGF 413
hydrate crystals stimulate gene expression in renal epithelial cells. Kidney Int 1995;
48:501-509.
Pawar S, Kartha S, Toback FG. Differential gene expression in migrating renal epi-
thelial cells after wounding. J Cell Physiol 1995;165:556-565.
Brigstock DR, Steffen CL, Kim GY, Vegunta RK, Diehl JR, Harding PA. Purifica-
tion and characterization of novel heparin-binding growth factors in uterine secretory
fluids: identification as heparin-regulated 10,000-M,, forms of connective tissue
growth factor. J Biol Chem 1997;272:20275-20282.
Frazier K, Williams S, Kothapalli D, Klapper H, Grotendorst GR. Stimulation of
fibroblast cell growth, matrix production, and granulation tissue formation by con-
nective tissue growth factor. J Invest Dermatol 1996;107:404-411.
Anderson LL. Growth, protein content, and distribution of early pig embryos. Anat
Rec 1978;190:143-154.
Geisert RD, Brookbank JW, Roberts RM, Bazer FW. Establishment of pregnancy
in the pig: 11. Cellular remodelling of the porcine blastocyst during elongation on
day 12 of pregnancy. Biol Reprod l982;27:94 1-955.
Perry JS, Heap RB, Burton RD, Gadsby JE. Endocrinology of the blastocyst and
its role in the establishment of pregnancy. J Reprod Fert Suppl 1976;25:85-104.
Perry JS, Rowlands IW. Early pregnancy in the pig. J Reprod Fertil 1962;4: 175-
188.
Wray W, Boulikas T, Wray V. Hancock R. Silver staining of proteins in polyacryl-
amide gels. Anal Biochem I98 1;118:197-203.
Brigstock DR, Kim GY, Steffen CL, Liu A, Vegunta RK, Ismail NH. High molecular
weight forms of epidermal growth factor in pig uterine secretions. J Reprod Fertil
1996;lO8:3 13-320.
Kim GY, Besner GE, Steffen CL, McCarthy DW, Downing MT, Luquette MH,
Abad MS, Brigstock DR. Purification of heparin-binding EGF-like growth factor
from pig uterine luminal flushings and its production by endometrial tissues. Biol
Reprod l995;52:56 1-571.
Brigstock DR, Kim GY. Steffen CL. Pig uterine fluid contains the developmentally-
regulated neurotrophic factor, pleiotrophin. J Endocrinol 1996;148: 103- 1 1 I.
Brigstock DR, Heap RB. Brown KD. Polypeptide growth factors in uterine tissues
and secretions. J Reprod Fert 1989;85:747-758.
Bork P. The modular architecture of a new family of growth regulators related to
connective tissue growth factor. FEBS Lett 1993;327:125-1 30.
O'Brien TP, Yang GP, Sanders L, Lau LF. Expression of cyr61, a growth factor-
inducible immediate early gene. Mol Cell Biol 1990;10:3569-3577.
Jay P, Bege-Lefranc JL, Marsollier C. Mejean C, Taviaux S. Berta P. The human
growth factor-inducible immediate early gene, CYR61, maps to chromosome Ip.
Oncogene 1997;14:1753- 1757.
Simmons DL, Levy DB, Yannoni Y, Erickson RL. Identification of a phorbol ester-
repressive v-src-inducible gene. Proc Natl Acad Sci USA 1989;86: 1 178- 1 182.
Joliot V. Martinerie C, Dambrine G, Plassiart G, Brisac M, Crochet J, Perbal B.
Proviral rearrangements and overexpression of a new cellular gene (nov) in myelo-
blastosis-associated virus type 1-induced nephroblastomas. Mol Cell Biol 1992;12:
10-21.
Martinerie C, Huff V, Joubert I, Badzioch M, Saunders G, Strong L, Perbal B. Struc-
414 Brigstock
tural analysis of the human nov proto-oncogene and expression in Wilms tumors.
Oncogene 1994;9:2279-2732.
Ying Z. Ling ML. Isolation and characterization of xnov, a Xenopus laevis ortholog
of the chicken nov gene. Gene 1997;171:243-248.
Shing Y, Folkman J, Butterfield C, Murray J, Klagsbrun M. Heparin affinity: purifi-
cation of a tumor-derived capillary endothelial cell growth factor. Science 1984;233:
1296-1299.
Besner GE, Higashiyama S, Klagsbrun M. Isolation and characterization of a macro-
phage-derived heparin-binding growth factor. Cell Regu 1990;1:811-8 19.
Lobb RR, Harper JW, Fett JW. Purification of heparin-binding growth factors. Anal
Biochem 1986;154:1-4.
Rosenthall RA, Megyesi JF, Henzel WJ, Ferrara N, Folkman J. Conditioned medium
from mouse sarcoma 180 cells contains vascular endothelial growth factor. Growth
Factors 1990;4:53-59.
Cook PW, Mattox PA, Keeble WW, Pittelkow MR, Plowman GD, Shoyab M, Adel-
man JP, Shipley GD. A heparin sulfate-regulated human keratinocyte autocrine fac-
tor is similar or identical to amphiregulin. Mol Cell Biol 1991;11:2547-2557.
Shoyab M, Plowman GD. Purification of amphiregulin from serum-free conditioned
medium of TPA-treated cell lines. Meth Enzymol 1991 ;198:213-221.
Klagsbrun M. The fibroblast growth factor family: Structural and biological proper-
ties. Prog Growth Factor Res 1989;1:207-235.
Higashiyama S, Lau K, Besner G, Abraham JA, Klagsbrun M. Structure of heparin-
binding EGF-like growth factor: multiple forms, primary structure and glycosylation
of the mature protein. J Biol Chem 1992;267:6205-6212.
Milner PG, Li Y-S, Hoffman RM, Kodner CM, Siege1 NR, Deuel TF. A novel 17
kDa heparin-binding growth factor (HBGF-8) in bovine uterus: Purification and N-
terminal amino acid sequence. Biochem Biophys Res Commun 1989;165:1096-
1 103.
Caday CG, Klagsbrun M, Fanning PJ, Mirzabegian A, Finkelstein SP. Fibroblast
growth factor (FGF) levels in the developing brain. Dev Brain Res 1990;52:241-
246.
Weich HA, Iberg N, Klagsbrun M, Folkman J. Expression of acidic and basic fibro-
blast growth factors in human and bovine vascular smooth muscle cells. Growth
Factors 1990;2:3 13-320.
Slalom Chromatography
A New Hydrodynamic-Based
Chromatographic Mode Applicable to
Size-Dependent Separation and
Physicochemical Analysis of Large
DNA Molecules
Jun Hirabayashi and Ken-ichi Kasai

Teikyo University, Sagarniko, Kanagawa, Japan
I. INTRODUCTION
Chromatography is one of the best research tools for the biosciences. It makes
it possible to analyze biomolecules and to obtain the most valuable and important
information on them, such as their chemical, physicochemical, and biochemical
properties. At the same time, it affords complete separation of components.
Therefore, the more sophisticated the chromatographic mode, the higher the qual-
ity of the information obtained. For example, affinity chromatography reveals
the nature of the active site of enzymes; gel permeation chromatography discloses
the size of biomolecules. Moreover, development of high-performance liquid
chromatography (HPLC) has greatly enhanced the utility of chromatography by
improving speed, sensitivity, reproducibility, and accuracy.
Use of chromatography as an intelligent tool has not been adequately ap-
preciated in the field of nucleic acid research because of several inherent difficul-
ties [I]. Though size-dependent separation of large polynucleotides such as DNA
fragments must be the first step of not only the majority of basic analyses but also
416 Hirabayashi and Kasai
a variety of applications, the chromatographic method has rarely been applied for
such a purpose, in contrast to the protein research field where gel permeation
chromatography has always been indispensable. This is mainly because gel per-
meation chromatography is extremely inefficient for the separation of DNA frag-
ments in comparison with agarose gel electrophoresis, which has been much more
popular and efficient, as described below.
First, in gel permeation chromatography, separation of component mole-
cules must take place within a relatively small volume, i.e., the elution volume
of a salt subtracted by the flow-through volume. This corresponds to the volume
of the liquid retained in the stationary phase, in other words, the volume of the
liquid retained in the pores of the packing particles. It is difficult to expect high
resolution of many components in such a small volume. Second. target macro-
molecules must diffuse into the pores of packing particles and an equilibrium
state must be reached between the moving phase and the stationary phase. The
time required to reach an equilibrium state becomes considerably long for large
molecules due to their small diffusion constant, and this also reduces the resolu-
tion. Third, most DNA fragments are usually too large to be applied to conven-
tional gel permeation media. Even a small DNA fragment has a Stokes radius
considerably larger than that of proteins of similar molecular weight. Fourth, the
possible risk of cleavage of long DNA molecules due to shear force has also
made researchers reluctant to use chromatographic separation. However, in spite
of these difficulties, to make full use of chromatography in the field of nucleic
acids would seem to be meaningful. Chromatography is often complementary to
electrophoresis in terms of both preparative and analytical purposes, and some-
times it can provide unique results. In the field of nucleic acid research, analytical
scale chromatography serves both analytical and preparative purposes because if
only a trace amount of a DNA or RNA fragment is separated it can be easily multi-
plied by polymerase chain reaction (PCR) and recombinant DNA techniques.
We recently discovered a new chromatographic mode that enables the sepa-
ration of DNA fragments according to their size [2,3]. This method, firstly discov-
ered by chance, requires only an ordinary HPLC system and a commercially
available column packed with small, rigid, spherical beads, such as those used
for high-performance gel permeation. Strikingly, the order of elution was found
to be opposite to that expected for gel permeation chromatography, i.e., larger
fragments were eluted later than smaller ones. It is not surprising that longer
DNA fragments would be retarded more if they interacted with the column pack-
ing. The amazing thing, though, is that no evidence could be found of interaction
between them. Moreover, separation patterns significantly depended on the flow
rate and the size of the packing particles, but not on their pore size or chemical
nature. DNA fragments were retarded without any attractive force provided by
the column packings. Therefore. it is a quite unusual phenomenon and cannot
Slalom Chromatography 417
be explained by any mechanism based on the equilibrium between moving phase

and stationary phase. Therefore, we proposed a possible mechanism based on a
hydrodynamic phenomenon.
Here we describe only the essential point of the proposed mechanism (Fig.
1) and its details will be discussed later. When we apply a solution of DNA
fragments to a column for HPLC, DNA molecules in the moving phase are
stretched by laminar flow and forced to pass through the narrow and tortuous
channels created by the gap between tightly packed spherical beads; and thus,
they should turn very quickly and frequently. This should be more difficult for
longer DNA fragments in comparison with shorter ones. Therefore, the smallest
fragment is eluted first, the longest one is eluted last, and the others are eluted
in the order of size from smaller to larger. We named this new separation mode
"slalom chromatography" because the proposed model reminds us of a person
on skis going down a slope and turning quickly around flags [3-51. Similar obser-
vations were made independently by Boyes et al. [6]. In the following sections we
Figure 1 Illustration of DNA separation in slalom chromatography. When applied to

a column for HPLC, DNA fragments are stretched due to the occurrence of laminar flow.
They should turn rapidly and frequently in passing through the narrow and tortuous open-
ings between closely packed spherical particles. The longer the fragments. the more diffi-
culty for turning around the particles. If the DNA molecules are stretched to the maximum
extent, their length will be comparable to the diameter of the packing particles. The dis-
tance between particles is exaggerated.
discuss in detail various unusual phenomena that led us to the above hypothetical
mechanism.
II. UNIQUE CHARACTERISTICS OF SLALOM

CHROMATOGRAPHY
A. Particle Size of Column Packing Has a Great Influence
on Separation
Figure 2 shows one of the most important characteristics of slalom chromatogra-
phy [2,3]. Eight DNA fragments ranging from 10 to 38 kbp could be separated.
In these cases, separation media for high-performance gel permeation (Asahipak
series, synthetic polymer-based media manufactured by Asahi Chemical Industry,
Retention time (min)

Figure 2 Chromatogram of restnctlon fragment&of hDNA on Asahipak GFT-510 (5-
pm panicle; A) and GS-510 (9-pm particles; B). Column size were the same (7.6 X 250
pm). A set of DNA fragments was dissolved in 10 mM sodium phosphate buffer, pH 7.0,
containing 1 mM EDTA, and applied to the column, and DNA fragments were eluted
with the same buffer at room temperature. Flow rate was 0.3 mllmin. Fragment sizes
(kbp) are indicated in the figure.
Kawasaki, Japan) were used. Their exclusion limits (for protein, M, 5 X lo5)
were the same, but the particle sizes were different. One column was packed
with 5-pm particles (Fig. 2a, Asahipak GFT-5 10) and the other with 9-pm parti-
cles (Fig. 2b, Asahipak GS-5 10). Both columns separated eight fragments in the
order of smaller to larger. This order was completely opposite to that expected
for gel permeation chromatography. Moreover, though the pore sizes were the
same, the column packed with 5-pm particles was superior for the separation of
smaller fragments (less then 20 kbp), whereas larger fragments (more than 20
kbp) were better separated by the column packed with 9-pm particles. It is evident
that separation occurred by a mechanism other than the gel permeation mode
because the order of elution was opposite and the size range depended not on
the pore size but on the particle size.
Figure 3 shows the separation pattern of HindIII-digested h phage DNA
(hlHindII1) on columns packed with 5-pm and 9-pm particles. Again, the 5-pm
packing gave a better resolution for fragments of less than 10 kbp. Such experi-
ments were carried out more systematically using four columns (average particle
diameters of 5.0,9.0, 13.1, and 19.1 pm, Asahipak GS-3 10 series), and the results
are summarized in Fig. 4 [4]. The relative retention time, defined as the ratio of
Elution Volume ( ml
Figure 3 Chromatography of XIHindIII digests on Asahipak GS-310 columns packed

with 5-pm (A) and 9-pm packings (B). DNA fragments were eluted at a flow rate of 0.3
mllrnin. Fragments contained in peak 1 are 0.13, 2.03, and 4.36 kbp. Peak 2, 3, and 4
correspond to 6.56, 9.42, and 23.13 kbp, respectively.
Hirabayashi and Kasai
Figure 4 Dependence of relative retention times of DNA fragments on their size. Pack-
ings of different particle size (5, 9, 13, and 19 ~ m were
) used, and different flow rates
were applied. Relative retention time was plotted as a function of DNA size (kbp). Flow
rates were 0.3 (a),0.6 (A), and 1.2 mllmin (m).
the retention time of a particular fragment to that corresponding to flow-through

fraction, was plotted against the number of base pairs of DNA fragments. Appar-
ently, the four columns have different ranges of resolution. For example, if we
pay attention to the curve at a flow rate of 0.6 mllmin (middle curves in Fig.
3a-d), the 5,9-, 13-, and 19-ym columns could separate DNA fragments larger
than 7, 9, 13, and 17 kbp, respectively, from the fragments that appeared at the
flow-through volume (relative retention time = I .O). In addition, smaller pack-
i n g ~showed better resolution for smaller DNA fragments, whereas larger ones
were better for larger fragments. High-resolution zones could be assigned as 9-
17, 15-30,23-40, and 35-50 kbp for the 5-, 9-, 13-, and 19-pm packings, respec-
tively.
B. Higher Flow Rates are More Advantageous

for Separation
The results shown in Fig. 4 also revealed that the extent of retardation of DNA
fragments was dependent on the flow rate. It is unlikely for gel permeation chro-
matography that the flow rate would affect the size range of fractionation and
that faster flow rates are advantageous for resolution. A lower flow rate is usually
preferable to obtain good resolution, as it ensures the equilibrium between the
stationary and mobile phases. In slalom chromatography, however, at the lowest
flow rate applied (0.03 mllmin, corresponding to a linear flow rate of 0.067 cm/
min), all DNA fragments of different sizes were eluted together in the flow-
through fraction; no size-dependent separation occurred. When the flow rate was
increased, longer DNA fragments began to slow down, and the extent of retarda-
tion became greater as the flow rate was increased further. This means that better
resolution is attained at higher flow rates and that the range of separable fragment
size also increased. Such unusual flow rate dependency also suggested that the
mechanism is completely different from an equilibrium phenomenon. At a very
low flow rate, all DNA fragments would take a random coil form, which is more
compact and would have little difference in the Stokes radius, and therefore they
would be eluted together in the flow-through fraction. On the other hand, when
a higher flow rate is applied, DNA molecules would be stretched and the slalom
effect would thus appear.
C. Pore Size and Chemical Nature of Packings Have Little

Influence on Separation
Although the phenomenon of slalom chromatography was first observed with
packings for gel permeation, their pore size was found to have no relation to the
resolution [2-41. When columns packed with beads of the same particle size (9
pm) but differing in pore size (e.g., Asahipak GS-220, 310, and 510: exclusion
limit for proteins, M, 3 X lo3,4 X lo4,and 3 X 10" respectively) were compared,
almost the same relative retention time was obtained for each DNA fragment.
This indicates that the DNA fragments did not permeate the pores.
The chemical nature of column packing was also proved to have no influ-
ence on separation. Comparative experiments carried out using silica-based parti-
cles (TSK G2000SW and TSK G3000SW, 10-pm diameter, manufactured by
Toso, Tokyo, Japan) gave essentially the same results as those obtained with
synthetic polymer-based particles [3]. Although they have different exclusion
limits (for dextran: G2000SW, M, 1 X lo5; G3000SW, M, 5 X lo5),both separa-

tion patterns and flow rate dependency were almost the same, and also very simi-
lar to those obtained for the column packed with 9-pm particles.
D. Separation Does Not Depend on Interaction Between

DNA Fragments and Packing Particles
In slalom chromatography, size-dependent separation of DNA is not due to elec-
trostatic or hydrophobic interaction with the packing particles. Retention time of
DNA fragments was not affected by the addition of NaCl up to 0.5 M or 20%
(v/v) acetonitrile [5]. Moreover, separation by the slalom chromatography mode
was achieved even with cation exchange columns, which should strongly repel
polynucleotides. Two cation exchange columns bearing sulfopropyl groups, TSK
SP-5PW and TSK SP-NPR (a nonporous polymer with a particle size as small
as 2.5 pm), were examined. SP-5PW gave a chromatogram similar to that ob-
tained with Asahipak of corresponding particle size, i.e., a high-resolution zone
was found in the range 15-30 kbp [5]. SP-NPR seemed to resemble GS-310
packed with 5-pm particles to some degree. These results indicate again the ab-
sence of interaction between the chromatographic media and DNA fragments,
and suggest that the separation was achieved by a hydrodynamic phenomenon.
E. Temperature Affects the Separation

Temperature has a significant effect on the slalom mode separation [5]. The rela-
tive retention time of DNA fragments increased significantly when the tempera-
ture was lowered. This suggests that the viscosity of the eluent has an important
effect, probably because it determines the steepness of the velocity gradient of
laminar flow in the narrow channels in the column.
F. Separation Depends on Physical Length of DNA

Fragments and Not on Their Molecular Weight
The experimental results described above were obtained for linear DNA mole-
cules. The behavior of circular DNA molecules was also examined. When the
circular replicative form (double strand) of M13 phase DNA (7 kbp) and its
linearized form were compared, the former was eluted faster than the latter. This
showed that even DNA molecules of the same size can be distinguished, if their
conformations are different. Longer circular DNAs were also analyzed. Even at
the highest flow rate examined, 42-kbp circular DNA was eluted with a similar
retention time to that of 20-kbp linear DNA. This result is quite reasonable be-
cause the length of circular DNAs in a maximally stretched state will be equal
to that of the linear DNA of half size. Therefore, slalom chromatography sepa-
rates DNA fragments according to their physical length and not to their molecular
weight.
Ill. MECHANISM OF SLALOM CHROMATOGRAPHY

A. Characteristics of Slalom Chromatography
From the observations described above, the characteristics of slalom chromatog-
raphy can be summarized as follows:
1. DNA fragments do not interact with the matrix of the packing material
because synthetic polymer-based and silica-based packings gave es-
sentially similar separation patterns, and addition of a salt or organic
solvent had little effect. Separation has no relation to the presence of
pores in the packing particles. Even nonporous particles and anion ex-
change particles, which should repel DNA molecules, showed the sla-
lom mode separation.
2. Particle size of packings greatly affects separation. The size range of
DNA fragments being separable was found to depend largely on the
particle size. Smaller particles could resolve smaller DNA fragments
better, whereas larger particles could resolve larger DNA fragments
better.
3. Flow rate has a significant effect on the separation range. At a faster
flow rate, the relative retention time of DNA fragments increased and
resolution became better.
4. Temperature has a significant effect. At lower temperature, DNA frag-
ments were more retarded. This suggests that separation depends on
the viscosity of moving phase solvent.
B. A Possible Mechanism of Slalom Chromatography

When a solution of DNA is applied to a column for HPLC, DNA fragments are
stretched because of the laminar flow generated by the solvent passing through
the narrow channels created by gaps between packing particles. For example, the
end-to-end distance of the maximally stretched 23-kbp hlHindIII fragment is as
long as 7.9 pm, which is comparable to the diameter of the packing particle used
(3-10 pm).
Though a small double-stranded DNA molecule (e.g., smaller than 20 bp)
has the structure of a rigid rod, large double-stranded DNA fragments (larger
than 1 kbp) has a kink at every -150 bp and consequently forms a random coil,
which is entropically the most favorable shape. For example, bacteriophage T4
DNA, 166 kbp (extended length, 50 pm), was found to form a random coil with
a diameter of -2 pm in its most contracted state. The random coil is very flexible,
and its shape changes without a break between the contracted and slightly ex-
tended forms by an external force provided by Brownian motion of water mole-
cules. The speed of the intramolecular movement (expansion and contraction)
can be as fast as - 15 pm/s, though the migration velocity of the whole molecule
(center of gravity) is only 1 ym/s [7]. However, in the presence of laminar flow,
the DNA molecule is stretched. As the flow rate increases, its shape changes
gradually from an ellipsoid to a solenoid one, then to a thick filament one, and
finally to a thin filament one.
DNA molecules are too large to permeate into the pores of packing particles
and can only pass through the narrow channels between particles. Since these
channels are extremely tortuous, when the moving phase is forced to move, DNA
fragments have to turn very frequently. It will be more difficult for longer DNA
strands to turn quickly, and this will result in retardation of longer DNA strands.
A simple calculation shows that DNA fragments turn about 36,000 times if ap-
plied to a column of 300-mm length packed with 10-ym packing particles. Under
these conditions, a 23-kbp fragment turns as frequently as 70 times per second,
when a flow rate of 0.6 ml/min is applied. If we use smaller packing particles,
the channels become narrower and more tortuous; and also, the velocity gradient
of laminar flow becomes steeper, which makes shear force stronger. At a flow
rate of 1.2 ml/min, in a column of 5-ym particles, the fragment has to turn 72,000
times at a frequency of 280 times per second. On each turn, DNA strands would
be exposed to a significant frictional force against the solvent, and this force
would increase with the increase in DNA length, so that larger DNA molecules
are retarded.
As to the effect of flow rate, at a flow rate of close to zero all DNA frag-
ments are in a random coil state; as none of them are retarded, they are eluted
together in the flow-through fraction. Application of a faster flow rate has at least
dual effects. First, it increases the frequency of turn. Second, it increases the
velocity gradient of the laminar flow. This generates stronger shear force and
consequently raises average end-to-end distance of DNA strands, and resulting
in stronger frictional force between the DNA strands and the solvent (Fig. 5).
These considerations may explain why the use of smaller packing and a higher
flow rate is advantageous for the separation of smaller molecules.
This new chromatographic mode is attributed to the combination of several
hydrodynamic phenomena, i.e., stretch by laminar flow of DNA fragments that
otherwise assume a random coil state because of entropic force; frequent turning
of extremely long molecules along the curved surface of packing particles; and
an extremely complicated flow pattern due to the channels made between packing
particles, which may cause continuous perturbation of the three-dimensional
shape of DNA. Although the explanation presented here is undoubtedly oversim-
plified, it should be a useful model in elucidating the precise mechanism of slalom
chromatography.
Narrow
Flow Rate
l==s Wide
Figure 5 An explanation of the effect of the flow rate and particle size. For simplicity,
the channel between closely packed particles is considered as a capillary. As the flow rate
is increased, velocity gradient of the moving phase is generated due to laminar Row. In
a narrow capillary, the velocity gradient is steep, and then a DNA molecule is subjected
to a strong shear force and stretched to a greater extent. As a result, the DNA molecule is
retarded greatly due to the slalom effect. On the contrary, in a wide capillary, no significant
velocity gradient is generated especially in the central part of the cross-section. Therefore,
the DNA molecule is not subjected to a strong shear force, stays as a more compact form,
and goes through the capillary without significant retardation.
A mode of size-dependent separation of submicrometer to micrometer or-

der colloid particles, named hydrodynamic chromatography, was reported [a]. It
was performed by using a column packed with nonfunctionalized, nonporous
particles of around 15-20 ym in diameter. In the column, large colloid particles
are excluded from the interface, where the fluid velocity is lowest. The larger
the particle, the higher its mean velocity. Consequently, larger particles are eluted
faster than smaller ones, as in gel permeation chromatography. Therefore, this
mode is different from slalom chromatography.
IV. POTENTIAL APPLICATION TO STUDIES ON

PHYSICOCHEMICAL PROPERTIES OF DNA
The utility of this new procedure is not limited to size analysis. This method will
undoubtedly provide us valuable information on DNA molecules almost equiva-
lent to those gel filtration has provided on the properties of proteins. Moreover,
this hydrodynamics-based method has potential to analyze not only the sizes
of DNA molecules but also physicochemical properties, such as conformation,
topology, and rigidity. Preliminary experiments showed that it can distinguish
the difference in the topology of circular DNA molecules because different flow
rate dependencies were observed for supercoiled, relaxed, and single-stranded
DNAs. These results might be attributed to their rigidity or special three-dimen-
sional structure. Therefore, it will be possible to separate and analyze DNA mole-
cules according to their physicochemical properties. If the efficiency of slalom
chromatography becomes much higher, isolation of topoisomers of supercoiled
DNAs according to the extent of coiling might become possible.
V. MIXED-MODE SLALOM CHROMATOGRAPHY
Although the uniqueness of slalom chromatography is its independency from

interaction between solute molecules and packing particles, it might be possible
to add weak interaction with the matrix to improve chromatographic perfor-
mance. This appeared to be the case because column packings having weak hy-
drophobic nature proved to be effective in improving resolution of DNA frag-
ments smaller than 5 kbp, which had been difficult to separate by pure slalom
chromatography [9]. Figure 6 shows separation patterns of AlHindIII fragments
obtained by use of a column packed with a silica-based media derivatized with
67 5
Retention Time (min)
Figure 6 Separation profiles of hlHindIII fragments on a Phe-Hypersil-3 column. Flow

rates are indicated in the figure. Acetonitrile (20%) was added in the buffer to raise recov-
ery of DNA fragments. At the flow rate of 1.2 mllmin, the best resolution was attained,
and a 4.4-kbp fragment was separated from a 2.3-kbp fragment.
phenyl group (Phe-Hypersil-3,3-km particles, manufactured by Shandon, Chesh-

ire, UK). A fragment as small as 4.4 kbp was separated from the flow-through
fraction at the flow rate of 1.2 mltmin; this had not been achieved previously.
It was found that the addition of acetonitrile to the moving phase was necessary to
raise the recovery of DNA fragments. It seems necessary to reduce hydrophobic
interaction to some extent. If the hydrophobic interaction is too strong, DNA
fragments would be adsorbed, and the slalom effect would disappear. In the pres-
ence of 20% acetonitrile, the best resolution was obtained. Other weak hydropho-
bic media (e.g., Capcell-Pak C1, Capcell-Pak Phe, manufactured by Shiseido,
Tokyo, Japan) were also found to be effective in separating small fragments,
although in the case of Capcell-Pak C1 and Capcell-Pak Phe the addition of
NaCl was necessary to attain good resolution. For these media, reinforcement of
hydrophobic interaction seems necessary. Although chromatographic conditions
have to be appropriately set up for individual packings, several columns devel-
oped for reversed-phase chromatography were found to be useful for mixed-mode
slalom chromatography. It is advantageous from a practical viewpoint because
a number of packing materials have been industrially developed for reversed-
phase chromatography, some of which show excellent physicochemical perfor-
mance.
VI. PRESENT AND FUTURE SLALOM CHROMATOGRAPHY

A. Advantages and Limitations of Slalom Chromatography
At present, slalom chromatography is the only chromatographic method applica-
ble to the size-dependent separation of large DNA molecules. Its uniqueness is
the hydrodynamics-based mechanism and length-dependent separation. Column
packing particles serve only for the construction of channels. The merits of slalom
chromatography are as follows:
Both preparative and analytical applications are possible.
The experimental procedure is very simple, and results can be obtained
rapidly. Only an ordinary HPLC apparatus and easily available HPLC
columns are needed.
Only an isocratic elution program is needed. Therefore, a process for
column washing or reequilibration is not necessary between runs.
Separation and recovery of DNA fragments can be completed in a
much shorter period than in the case of gel electrophoresis. Recovered
DNA fragments are free from undesirable contaminants originating
from the agarose gel.
DNA can be detected without the use of harmful reagents, such as
ethidium bromide.
6. This procedure can be used as an effective tool for physicochemical

and hydrodynamic studies of DNA.
There are, of course, limitations that should be overcome in the future:
1. The range of separable size is rather narrow. Commercial columns
available can be applicable only to 5-50 kbp DNA fragments.
2. The resolution efficiency is still inferior to that of gel electrophoresis.
3. Chromatography should be performed under relatively high flow rates,
which might cause fragmentation of extremely large DNAs. However,
DNA fragments of less than 50 kbp proved to be generally very stable
under the conditions for most experiments (e.g., flow rate less than
1.2 mllmin), contrary to the expectations of most researchers handling
DNA. Therefore, the use of slalom chromatography is strongly encour-
aged.
B. Comparison with Other Size-Separation Methods

for DNA
Pulsed-field gel electrophoresis is widely used as a size separation procedure for
extremely large DNAs [10,1 11. This technique is based on the difference in the
ability of DNA molecules to change direction of migration in response to a fre-
quently changing electric field. Because larger DNA molecules take longer to
reorient, size-dependent separation is achieved. This method has features in com-
mon with slalom chromatography: both are based on the ability of DNA mole-
cules to adapt to a frequently changing environment (direction of electric field
or flow) that depends on size. Although the size range of DNA separable by
pulsed-field electrophoresis is much wider than that of slalom chromatography,
the latter has a variety of merits (already pointed out), and thus the two methods
are complementary. Capillary electrophoresis also seems to be very promising
as a procedure for the size-dependent separation of nucleic acids. If the principles
of slalom chromatography and capillary electrophoresis can be combined, a much
more effective procedure could possibly be realized.
C. How To Make the Best Use of Slalom Chromatography

Slalom chromatography is very efficient and interesting from the viewpoints of
both investigation and application. This principle will undoubtedly provide us
with a valuable tool for nucleic acid research, just as the invention of gel perme-
ation chromatography has contributed to the biological sciences, especially in
the field of protein research. Although only applications to DNA have been re-
ported so far, it should also be useful for RNA and other fibrous macromolecules
such as proteoglycans. Possible applications include the following:
1. Size-dependent separation of DNA

2. Estimation of size of nucleic acids
3. Monitoring and analysis of size changes of nucleic acids
4. Separation of nucleic acids based on conformation or topology
5. Analysis of interaction of nucleic acids with other molecules, such as
nucleic acid-binding proteins
6. Distinction of types of circular DNA, e.g., supercoiled, relaxed, and
single-stranded
7. Studies of the physicochemical properties of nucleic acids, e.g., rigid-
ity, elasticity, bendability, etc.
8. Hydrodynamic studies of nucleic acids
The combination of hard, spherical packings and the application of high
flow rates led to the discovery of this hydrodynamics-based chromatography. It
is very unlikely that slalom chromatography is the only technique that can be
based on hydrodynamic principles. Likewise, other interesting and effective chro-
matographic modes applicable to a variety of macromolecules will probably be
discovered if extensive studies are done by researchers in a variety of fields. Such
research will contribute greatly not only to expanding the horizon of chromato-
graphic science but to developing a number of effective and intelligent research
procedures.
REFERENCES
Kasai K. Size-dependent chromatographic separation of nucleic acids. J Chromatogr

1993;618:203-221.
Hirabayashi J, Kasai K. Slalom chromatography. A new size-dependent separation
method for DNA. Nucleic Acid Symp Ser 1988;20:67-68.
Hirabayashi J, Kasai K. Size-dependent, chromatographic separation of double-
stranded DNA which is not based on gel permeation mode. Anal Biochem 1989;
l78:336-341.
Hirabayashi J, Itoh N, Noguchi K, Kasai K. Slalom chromatography: size-dependent
separation of DNA molecules by a hydrodynamic phenomenon. Biochemistry 1990;
29~9515-9521.
Hirabayashi J, Kasai K. Slalom chromatography. A size-dependent separation
method for DNA molecule based on a hydrodynamic principle. In: Ngo TT, ed.
Molecular Interactions in Bioseparations. New York: Plenum Press, 1993:69-87.
Boyes BE, Walker DG, McGeer PL. Separation of large DNA restriction fragments
on a size-exclusion column by a nonideal mechanism. Anal Biochem 1988;170:127-
134.
Yanagida M, Hiraoka Y, Katsura I. Dynamic behavior of DNA molecules in solution
studied by fluorescence microscopy. Cold Spring Harbour Symp Quant Biol 1982;
47~177-187.
8. Small HJ. Hydrodynamic chromatography. Technique for size analysis of colloidal

particles. Colloid Interface Sci 1974;48:147- 161.
9. Hirabayashi J, Kasai K. Applied slalom chromatography. Improved DNA separation
by the use of columns developed for reversed-phase chromatography. J Chromatogr
1996;722:135- 142.
10. Schwarz DC, Cantor CR. Separation of yeast chromosome-sized DNA by pulse field
gradient gel electrophoresis. Cell 1984;37:67-75.
11. Carle GF, Olson MVG. Separation of chromosomal DNA molecules from yeast by
orthogonal-field-alteration gel electrophoresis. Nucleic Acids Res 1984;12:5647-
5664.
Simultaneous Chromatographic
Separation of Ceruloplasmin and
Serum Amine Oxidase
Mircea-Alexandru Mateescu,' Xin-Tao Wang,lz2 Olivia Befani,2
Marie-Josee Dum~ulin,~ and Bruno Mondovi2
' Universite du Quebec a Montreal, Quebec, Canada
Rome University "La Sapienza," Rome, Italy
I. INTRODUCTION
Ceruloplasmin (CP) is the multifunctional blue [I] copper-containing plasma

protein (a2-globulin), with an important role in copper transport [2]. Acting as
an oxidase, ceruloplasmin (EC 1.16.3.1) is involved in the regulation of the level
of several phenols and aromatic amines. Also known as ferroxidase I, CP pro-
motes the oxidation of ferrous ions (Fez++ Fe3+),thus preventing the accumula-
tion of highly toxic Fez+ion [3]. Ceruloplasmin is also a powerful plasma antioxi-
dant and oxygen free radical (OFR) scavenger [4-71; furthermore, ceruloplasrnin
is an antiinflammatory protein that acts as an acute phase serum reactant against
the oxygen metabolites released by the macrophages [8].
In recent years it has also been shown that CP acts as a neuromodulator
(depolarizes neuronal cell membrane by inhibiting K+ channels [9]) and cardio-
protector (with an antifibrillatory action on isolated hearts subjected to ischemia-
reperfusion) [lo]. The antiarrhythmic effects of CP appeared closely related to
its molecular and conformational integrity [lo].
Recently, CP was shown to be involved in angiogenesis, in relation to its
This chapter was adapted from X. T. Wang et a]., Prep Biochem 1994: 24: 237-250. Copyright
0 1994 by Marcel Dekker, Inc.
432 Mateescu et al.
function as copper carrier. There is a recently growing interest for genetic dis-
eases characterized by a C P deficiency or by a defective loading of copper into
CP. Wilson's, Menkes', Parkinson's, Alzheimer's, and, as recently found, hemo-
siderosis are diseases related to C P deficiency.
Despite the abundant literature linking C P to several pathological condi-
tions, little research has been done on its possible therapeutic applications, proba-
bly because of its high susceptibility to proteolysis and instability, even during the
purification steps. In fact, structural aspects of CP, mainly related to the molecular
characteristics and the copper content, have been controversy (mostly due from
its high susceptibility at proteolysis). Also controversial was its complex physio-
logical role (antioxidantlprooxidant) related to the molecular integrity of the mol-
ecule, as recently reported [ l l l.
The "blue copper" center of CP has a characteristic absorption band at
610 nm and is electronic paramagnetic resonance (EPR)-detectable. It is now
accepted that C P contains six copper atoms per molecule and that CP structure
consists of six domains [12]-surprisingly, in a configuration closed to that of
clotting Factor VIII. Three copper atoms are aggregated in a cluster which is the
blue copper center of ceruloplasrnin. Two others form a diamagnetic pair. The
last one is paramagnetic (EPR-detectable). An absorbancy ratio Ab10nmlA280nm =
0.040 was considered in the literature as characteristic for the homogeneous stan-
dard pure enzyme.
Previously reported was a new single-step chromatographic method for the
fast ceruloplasmin purification [13], starting directly from plasma, which leads
to a purified, electrophoretically homogeneous CP. The procedure is based on
the highly selective retention of C P on the aminoethyl (AE)-agarose, a new chro-
matographic material (not commercially available), realized in our laboratories
113- 151. This single-step purification procedure described by Calabrese et al.
[14] is very fast, requiring only plasma dilution (about 20 times), followed by
AE-agarose chromatography and a final concentration of the purified CP. Despite
the speed at the laboratory scale, for scaling-up purposes, handling of large vol-
umes (dozens of liters) of diluted plasma is somewhat cumbersome. Furthermore,
when large plasma volumes are used, some clotting phenomena can occur during
the chromatographic step, producing a decrease in flow rate, difficulties in regen-
eration, and partial loss of the chromatographic material. At the same time, han-
dling of large protein volumes for the final concentration is also cumbersome.
These limitations can be completely overcome by the recently reported purifica-
tion procedure [16], here described.
Bovine serum amine oxidase (SAO), also known as benzylamine oxidase
belongs to the semicarbazide-sensitive amine oxidase (SSAO), is a copper en-
zyme (EC 1.4.3.6) catalyzing the oxidative deamination of various biogenic pri-
mary amines, with release of aldehydes, ammonia, and hydrogen peroxide. Con-
sidering the toxicity of aldehyde and H 2 0 2 ,SAO was suggested as an antitumoral
agent [I71 with an important cytotoxicity-inducing activity [18]. Recent data
Chromatographic Separation of CP and SAO 433
showed that SAO has a multifunctional role as regulator of polyamine level [I 91,
as cardioprotective agent [20,21], and as modulator of ionic channel [22]. It was
also shown that SAO substrate specificity is not limited only to small molecular
biogenic amines such as spermine and spermidine, but is extended to polylysine
and some peptides or proteins [23]. This suggested the hypothesis that SAO can
be involved in the process of protein posttranslational modification [23]. The
enzyme was first isolated in crystalline form by Yamada and Yasunobu [24].
Presently, several purification procedures are available. An affinity separation of
SAO on AH-Sepharose was described by Svenson and Hynning [25] and a two-
step procedure (ammonium sulfate and affinity chromatography on AH-Sepha-
rose followed by elution with aniline SAO inhibitor) was reported by Vianello
et al. 1261. The method described by Mondovi et al. [27] (ammonium sulfate
precipitation followed by chromatography on AH-Sepharose and Con A-Sepha-
rose) leads to high SAO-specific activities (0.33 EUImg). Since the AH-Sepha-
rose is no longer commercially available, it has been substituted with Q-Sepha-
rose 1281. The procedure usually leads to a purified SAO [27] with a slightly
lower specific activity (0.28 EUImg). In some chromatographic runs on Q-Sepha-
rose, the peaks of ceruloplasmin and bovine serum amine oxidase were partly
overlapped, thus lowering the purification yield and the specific SAO activity.
This can be satisfactorily improved by elimination of C P from the plasma protein
preparation before passing it through a Q-Sepharose column, leading to higher
specific activities and yields.
SAO and ceruloplasmin exhibit several common features. Both are serum
copper proteins of a relative high molecular mass: 180,000 Da (two monomers
of 90 kDa) for SAO and 132,000 Da for CP. Both enzymes are involved in
oxidative processes and both are glycoproteins (6-12% carbohydrate). Both pro-
teins have antioxidant properties and act as cardioprotective [20,29] and neuro-
modulatory agents [9,22]. Furthermore, the two proteins exhibit, to a certain
extent, a similar behavior toward some chromatographic materials (e.g.,
Q-Sepharose and AH-Sepharose), often leading to interference during the purifi-
cation procedures. A joint purification procedure leading to higher purity parame-
ters or yields should therefore be of interest to laboratories working in the field
of copper proteins.
In this study, a joint chromatographic purification procedure of C P and
SAO is described and the results are compared with independent CP [14] and
SAO [27] purification procedures.
II. EXPERIMENTAL
A. Chromatographic Material
Aminoethyl agarose is a chromatographic material (not commercially available)
obtained in laboratory by the treatment of agarose beads (i.e., Superose 6B, Phar-
434 Mateescu et al.
macia, Uppsala, Sweden) with 1-chloro-2-ethylamine (chlorohydrate) (Aldrich,

Milwaukee, WI), under conditions previously described [13- 151. Practically, 300
rnL of Superose 6B was suspended to about 500 mL and heated to 70°C. Then
130 g of 2-chloroethylamine hydrochloride was added to the suspension with
stimng. A volume of about 25 mL of 10 N NaOH was added to the suspension
to adjust the pH to 10. The reaction mixture was incubated for 2 h with continuous
stimng. During the first 20 min of the incubation, small volumes (few milliliters)
of 10 N NaOH were added from time to time in order to maintain the pH at 9-
10. After the reaction, the gel was washed by filtration on a Biichner funnel until
pH = 7 in washing solution. The resulting chromatographic material contains
aminoethyl functional groups and will be named here as AE-agarose.
Ceruloplasmin enzyme activity was determined, with p-phenylenediamine
as substrate, following the Osaki et al. method [30]. One enzyme unit is consid-
ered the amount of enzyme that would generate an increase of one absorbancy
unitlmin at h = 540 nm.
SAO enzymatic activity was assayed spectrophotometrically, with benzyl-
amine as substrate, according to Tabor et al. [31]. One enzyme unit (EU) was
defined as the amount of enzyme able to catalyze the formation of 1 pmole of
benzaldehydelmin under the reaction conditions.
The protein concentrations were determined by the Bradford method [32].
B. Joint Purification Procedure of Ceruloplasrnin and SAO

The joint purification flow of CP and SAO as well as the reference methods for
the single-step purification of CP [14] and for the SAO [27] are schematically
presented in Fig. 1. If CP purification is required, especially when large amounts
are required, channel c should be followed and the procedure is ended with the
AE-agarose chromatographic step. In detail, the purification of CP andlor of SAO
can be carried out following the procedure here presented.
1. Collection of Plasma
A volume of 10 L of bovine blood was collected at the slaughterhouse, mixed
with 1 L of 2.5% sodium citrate solution and centrifuged at 3000g for 20 rnin,
retaining the plasma. The crude CP-specific activity is usually about 8.14 X
EUImg and SAO specific activity of about 2.4 X to 7.0 X EUImg.
2. Ammonium Sulfate Fractionated Precipitation

Ammonium sulfate was added to 4 L of plasma to 35% of saturation, stirred for
2 h at 4°C and centrifuged at 10,000 g for 20 min, retaining the supernatant to
which ammonium sulfate has been added up to 55% saturation, maintaining the
stirring for 30 min at 4°C and then centrifuging at 10,000g for 20 min. The
Chromatographic Separation of CP and SAO
+ Bo- blood
(centrifuged at 3000g)
t
Bovine plasma (.) Dilntion
(about 20 limes)
1 I.
Ib) Precipitation
with (NH+)2 SO+ AE-Agar~se
I Con A-Sephamse
(PH 7-21
1.A. = 0.33
SAO SIO
Figure 1 Separation flow of CP and SAO. Channel a: "Single-step" CP purification

[14]. Channel b: current method of SAO purification [27]. Channel c: joint SAO and CP
purification. (From Ref. 16.)
436 Mateescu et al.
precipitate was retained and dissolved in 200 mL of 0.1 M potassium phosphate

buffer, pH 7.2. The solution was dialyzed for 20 h against 20 L of a 10 mM
potassium phosphate buffer, pH 7.2 with two changes of buffer. The dialyzed
plasma proteins solution was centrifuged at 10,000g for 20 min and the precipitate
discarded.
3. AE-Agarose Chromatography and CP Purification

The AE-agarose column (30 X 2.5 cm) was equilibrated with a 10 mM potassium
phosphate buffer, pH 7.2, which was also used as first eluent. Half of the dialyzed
plasma protein solution (Fig. 1) was applied onto the AE-agarose column, at a
flow rate of 120 mL/h. The nonretained fraction was collected for further SAO
purification. The AE-agarose column was then washed with: 500 ml of starting
10 mM phosphate buffer, then with 200 mL of 20 mM buffer solution, and finally
with 100 mL of 30 mM phosphate buffer, all at pH 7.2. The retained CP was
eluted with 100 mL of a 0.2 M phosphate buffer, collecting (Frac-200, Pharrnacia,
LKB) fractions of 2 mL each. The fractions with the ratio A610/A2R0 higher than
0.045 were collected in a pool with a CP concentration of 8 mg/mL. Purified
CP was stored at -20°C.
Ultrafiltration. If CP concentrations higher than those obtained directly
from the column are required, a further ultrafiltration step (Amicon 8200, mem-
brane YM 100) will lead to a final purified CP at a concentration of about 20
mg/mL.
4. Q-Sepharose Chromatography (pH 7.2)

and SAO Purification
A Q-Sepharose column (40 X 2.5 cm) was equilibrated with 10 mM potassium
phosphate buffer at pH 7.2. The SAO-containing fraction not retained by the AE-
agarose was applied onto the Q-Sepharose column at a flow rate of 150 mL/h.
The column was first washed with 500 mL of starting 10 mM phosphate buffer
and then eluted with a linear concentration gradient from 0.05 M to 0.35 M NaCl
in 10 mM starting buffer. The gradient was generated by a double-chamber (300
mL each) device. The elution rate was 100 mL/h, collecting fractions of 8 mL
each. The fractions with SAO-specific activity higher than 0.05 EU/mg were
retained and pooled.
5. Q-Sepharose Chromatography (pH 8.0)

The Q-Sepharose column was regenerated and equilibrated with a 20 mM po-
tassium phosphate buffer at pH 8.0. The protein solution (step 4) was diluted
(I : 1) with the starting buffer and carried onto the column, washed with 300 mL
starting buffer, and eluted with a continuous gradient from 0.05 M to 0.35 M
NaCl in 20 mM buffer, as described above. The fractions with SAO-specific

activity higher than 0.14 EUImg were collected.
6. Con A-Sepharose Chromatography

A Con A-Sepharose column (30 X 2.0 cm) was equilibrated with 0.1 M potas-
sium phosphate buffer at pH 7.2. The SAO-containing solution (step 5 ) was di-
luted (1 : 1) with starting buffer and applied onto the column and washed first
with 500 mL of starting 0.1 M phosphate buffer and then with 300 mL of 0.02
M methyl-a-D-glucopyranoside in 0.1 M phosphate buffer at pH 7.2. The SAO
was then eluted with 200 mL of 0.5 M methyl-a-D-mannopyranoside in 0.1 M
phosphate buffer, at pH 7.2, collecting fractions of 6 mL. The samples with SAO-
specific activity higher than 0.30 Ulmg were pooled and dialyzed against 0.1 M
potassium phosphate buffer at pH 7.2, for 6 h, concentrated by ultrafiltration and
stored in fractions of 1 mL, in Eppendorf tubes at -20°C.
C. Single-Step CP Purification by AE-Agarose

Chromatography
The modified CP purification method here described was compared to the initial
C P purification done under the conditions described by Calabrese et al. [14], as
a fast single-step chromatographic method based on the specific CP retention on
AE-agarose from a whole plasma preparation, without previous ammonium sul-
fate precipitation. This method, easily applicable to small and medium laboratory
scale CP preparations, has been carried out in parallel with the joint purification
here described (Fig. 1) starting from the same bovine plasma prepared as de-
scribed above. Practically, a volume of 200 mL plasma was diluted with 3 mM
potassium phosphate pH 7.2 buffer about 20 times, such as to reach a conductivity
of 1.4 millimho, a value similar to that of 10 mM phosphate buffer. The diluted
plasma was loaded to AE-agarose column (30 X 2.0 cm), previously equilibrated
in 3 mM potassium phosphate buffer. pH 7.2. The column was washed first with
200 mL of 10 mM phosphate starting buffer, then with 100 mL 20 mM phosphate
buffer, and finally with 50 mL of 40 mM phosphate buffer, pH 7.2. Ceruloplasmin
was retained while the other proteins were eluted from the column. Ceruloplasmin
was then eluted with 100 mL of 0.1 M potassium phosphate buffer. With the
increase in ionic strength, the CP first concentrates at the bottom of the column
and then elutes, the blue fractions being collected.
The purified CP (specific activity of 0.23 EUImg) was electrophoretically
homogeneous and with the absorbancy ratio Ab10/A280 of 0.045 (value correspond-
ing to the standard criteria for the purified enzyme [4-81); UV-VS absorption
spectra were recorded on a Beckman DU-6 spectrophotometer.
Mateescu et al.
KDa Standard
Figure 2 PAGE electrophoretic patterns of the purified CP and SAO. Electrophoresis

was carried out by Pharmacia-LKB Phast System, in the presence of 5% P-mercaptoetha-
nol. (From Ref. 16.)
D. SAO Purification
The SAO separation method according to Mondovi et al. [27], slightly modified
[B. Mondovi and 0 . Befani, unpublished data, 19921 as is now currently applied
(without the AE-agarose column), was carried out in parallel with the joint puri-
fication here described (Fig. I), starting from the same bovine plasma prepared
as described above.
Electrophoresis
The degree of homogeneity as well as the molecular integrity and characteristics
of purified CP and SAO were evaluated by SDS-PAGE fast electrophoresis (Phast
System, Pharmacia-LKB), in the presence of 5% P-mercaptoethanol (Fig. 2).
Ill. RESULTS AND DISCUSSION
The characteristics of C P purification by the modified procedure here described,

and by the single-step procedure [14], are summarized in Table 1. It was found
that the modified method (including ammonium sulfate precipitation) yields
larger amounts of purified CP protein (123 mg versus 50.6 mg) for a separation
run and many more enzyme units (85.5 EU versus 11.5 EU) than the "single-
Table 1 Comparison of CP Purification by the "Single-Step" Method [I41 and by

the Modified Procedure
< 'Single-step" Modified
Purification characteristics method [14] method
Starting material volume (mL plasma) 200b 2000
Total CP activity (EU) in plasma 12.37 132.7
Total protein amount (mg) in plasma 15,200 152,000
Specific CP activity in plasma (EUImg protein) 8.14 X 8.14 x
Specific activity (EUImg) of purified CP 0.23 0.70
Purification factora 282 860
(A6dA280) 0.045 0.057
Total recovered activity (EU) of purified CP 11.5 85.5
Total purified CP protein (mg) 50.6 123.0
Yield (recovered EU) (%) 93 69
"efined as the ratio: specific activity of purified enzymelspecific activity in the starting material.
hLower starting volume than for the combined method, since the required 1 :20 dilution leads to
4 L (difficult to handle in the chromatographic run).
step" method [14]. However, the recovery appears higher for the initial single-
step procedure (50.6 mg CP yielded from 200 mL plasma versus 123 mg CP
from 2000 mL plasma). Considering the amount of CP in plasma (200-300 mg/
L), the CP recovery by the single-step procedure [14] is very high (250 mg/L).
On the other hand, the modified method, despite a lower recovery (60-65 mg
CPIL), seems more advantageous because dealing with smaller volumes yields
a higher CP amount (120-125 mg) per separation run.
The characteristics of CP purified by the modified method (including am-
monium sulfate precipitation) are better than those obtained by the single-step
procedure. For instance, the specific activity is three times higher (0.70 versus
0.23 EU/mg) while the purity expressed by the spectral parameters (A610/A280)
are improved by more than 25% (higher than 0.057 versus 0.045). The improve-
ment in CP purification can be related to the partial elimination of contaminants
as well as of clottable proteins (in fact, no clotting phenomena were observed
during the AE-agarose chromatographic run). Minimizing the sample story, the
risk of protein degradation is limited. In fact we suppose, following a reexamina-
tion of CP spectral properties (EPR [14]), that CP purified by chromatography
on AE-agarose is closer to its real native structure than commercial CP obtained
by other methods.
This method can also allow the CP immobilization directly on this specific
chromatographic support during its purification [33]. The conjugation of CP with
440 Mateescu et al.
biocompatible polymers is important because the immobilized enzyme conju-

gates show sought-after advantages such as higher stability, lower antigenicity,
and the possibility of continuous use in various devices.
The characteristics of the SAO purification following the previous method
[27] and by the combined purification here described are presented in Table 2.
As shown in Table 2, the combined purification, when compared to the
current method, leads to a 16% increase in purified enzyme (42.0 versus 36.2
mg) and a 24% increase in total SAO enzyme units (13.9 versus 11.2 EU) as
well as to a slightly higher SAO specific activity (0.33 versus 0.31 EUImg). The
increase in purification yield obtained by the combined procedure is about 25%.
The electrophoretic sodium dodecyl sulfate-polyacrylamide gel electro-
phoresis (SDS-PAGE) patterns of the purified CP and SAO are shown in Fig. 2.
The combined purification method leads, in the case of both C P and SAO,
to electrophoretic homogeneity, with preservation of their molecular integrity.
The C P appears homogeneous with an unique band corresponding to the mole-
cular mass of 132 kDa. The SAO monomers also appear homogeneous (a band
at 90 kDa).
Although an additional AE-agarose step is required, the depletion of CP
and other contaminants (eluted with 20 and 30 mM phosphate buffer) appears
to be an advantage, since these proteins can interfere with SAO. In the first Q-
Sepharose chromatographic step following the current SAO purification method
Table 2 Comparison of Characteristics of SAO Obtained by the Current Method of

Mondovi et al. [27] including a Q-Sepharose Step and by the Combined CP-SAO
Purification Method Here Described, with the Additional Step of AE-Agarose
Chromatography
Current Combined
Purification characteristics method method
Starting material volume (mL plasma)
Total SAO activity (EU) of starting plasma
Total protein amount (mg) in plasma
Specific SAO activity in plasma (EUImg protein)
Specific activity (EUIrng) of purified SAO
Purification factora
Purity (A481dA280)
Total purified SAO protein (mg)
Total activity (enzyme units) of purified SAO
Yield (recovered EU) (%)
"efined as the ratio: specific activity of purified enzymelspecific activity in the starting material.
[27], the elution profile (Fig. 3a) shows a slight overlap between the CP and SAO
peaks. As a result of the additional AE-agarose step provided by the combined
CP-SAO separation procedure, the chromatographic interference by CP is
avoided (Fig. 3b). With a previous AE-agarose column. the SAO specific activity
after the first Q-Sepharose purification step was increased by about 35% (the
result of four different separation runs) when compared with the current method
(without the AE-agarose step).
As shown in Fig. 1, differently from the previous separation procedure, the
Con A-Sepharose step was carried out after the second Q-Sepharose column. The
advantage of this inversion is that the use of a second Q-Sepharose column leads
to a major increase in specific activity, with a simultaneous decrease in the vol-
ume to be handled in the last Con A-Sepharose step.
We are not considering the joint CP and SAO purification method presented
here as a new method for either CP or SAO purification. However, it is worth
mentioning that the mutually improving effect generated by associating the
two methods of the independent C P and SAO purification procedures into a joint
method for C P and SAO purification leads to better characteristics for each puri-
fied protein. The improvement is more important for the CP purification, as it
leads to several fold increases in different characteristics (specific activity and
the total activity units obtained by a separation run) and to a better molecular
integrity as reflected by the A,,,,/A2,, ratio. The increased yield of SAO of more
than 25% by the joint purification technique is also useful.
The current procedures for the SAO purification [27] require dialysis and
concentrations between each chromatographic step. In the modified method, dial-
ysis and concentration between the chromatographic columns can be eliminated
and replaced by simple dilution (1 : 1) of the enzyme solution with one volume
of the buffer used to equilibrate the subsequent column. This improvement allows
a faster purification run, leading to higher yields, increased specific activities,
and ameliorated spectral characteristics. For instance, with these modifications,
the yield is increased (100 to 115 mg SAO from 10 L of blood). The specific
enzymatic activity is 0.32 to 0.35 EU/mg and the absorbancy ratio A4xonn,/A280nm
= 0.013 (denoting a high quality of the enzyme). At the same time, the eluent
volumes were reduced. For example, the eluent volume of the first Q-Sepharose
column was reduced from 1800 mL to 600 mL (while maintaining the initial and
final gradient concentrations) and thus the method is faster and easier to apply.
The modified method is therefore time saving (only 3 days for the whole separa-
tion instead of 4-5 days for the current procedure) and leads to improved yields
and quality of SAO. Apart from the mutual improvement effects, better yields,
and molecular characteristics, the combined purification procedure may be of
interest for large-scale or industrial preparations, since the same inexpensive bo-
vine plasma source and preliminary precipitation as well as the AE-agarose steps
are common steps for both C P and SAO purification.
Mateescu et al.
Figure 3 Typical elution profiles of plasma proteins fractions on Q-Sepharose column

without (a) and with (b) a previous chromatographic step on AE-agarose column. (From
Ref. 16.)
Chromatographic Separation of CP and SAO
ACKNOWLEDGMENTS
The financial supports provided by M R C (Canada)-CNR (Italy) international

collaboration, by MURST and by CNR grant-Biotechnology and Molecular Biol-
ogy, contract No. 97.04376.CT17, as well as a postdoctoral fellowship from
"Fondation UQAM" offered to X. T. Wang, are gratefully acknowledged.
REFERENCES
Gutteridge GM, Stocks J. Ceruloplasmin: physiological and pathological perspec-

tives. Crit Rev Clin Lab Sci 1981;14:257-329.
Paulik MD, Weiss ML. In: Putnam FW, ed. The plasma proteins. Vol. 2. New York:
Acad Press, 1975;51- 108.
Gutteridge JMC, Richmond R, Halliwell B. Oxygen free radicals and lipid peroxida-
tion: inhibition by the protein ceruloplasmin. FEBS Lett 1980;122:269-274.
Lovstad RA. The protective action of ceruloplasmin on Fez+stimulated lysin on rat
erythrocytes. Int J Biochem 1981;13:221-228.
Al-Timimi DJ, Dormandy TL. The inhibition of lipid auto-oxidation by human ceru-
loplasmin. Biochem J 1977;168:283-29 1.
Mateescu MA, Chahine R, Roger S, Atanasiu R, Yamaguchi N, Lalumiere G, Na-
deau R. Protection of myocardial tissue against deleterious effects of oxygen free
radicals by ceruloplasmin. Arzneim Forsch 1995;45:476-480.
Dumoulin MJ, Chahine R, Atanasiu R, Nadeau R, Mateescu MA. Comparative anti-
oxidant and cardioprotective effects of ceruloplasmin, superoxide dismutase and al-
bumin. Arzneim Forsch 1996;46:855-861.
Goldstein IM, Kaplan HB, Edelson SH, Weissman G. Ceruloplasmin. A scavenger
of superoxide anion radicals. J Biol Chem 1979;254:4040-4052.
Wang R, Zhang L, Mateescu MA, Nadeau R. Ceruloplasmin: an endogenous depo-
larizing factor in neurons? Biochem Biophys Res Commun 1995;207:599-605.
Atanasiu R, Dumoulin MJ, Chahine R, Mateescu MA, Nadeau R. Antiarrhythmic
effects of ceruloplasmin during reperfusion in the ischemic isolated rat heart. Can
J Physiol Pharmacol 1995;73:1253-1261.
Fox PL, Mukhopadhyay C, Ehrenwald E. Structure, oxidant activity, and cardiovas-
cular mechanisms of human ceruloplasmin. Life Sci 1995;56:1749- 1758.
Pemberton S, Lindley P, Zaitsev V, Card G, Tuddenham EG, Kemball-Cook G. A
molecular model for the triplicated A domains of human factor VIII based on the
crystal structure of human ceruloplasmin. Blood 1997;89:24 13-242 1.
Dimonie M, Schell HD, Hubca G, Mateescu MA, Todireanu S, Languri J, Maior 0 ,
Iosif M. Polyvinyalcohol synthesis by suspension methanolysis of polyvinylacetate.
Derivatization and some applications. J Macromol Sci 1985;22A:729-754.
Calabrese L, Mateescu MA, Carbonaro M, Mondovi B. Reexamination of spectro-
scopic properties of ceruloplasmin freshly isolated with a novel, very rapid single-
step procedure. Biochem Int 1988;16:199-208.
444 Mateescu et al.
Calabrese L, Mateescu MA, Riccio P-L. Carbonaro M, Natoli G, Mondovi B. Single step
chromatographic purification of plasma ceruloplasmin. Italian Patent 479 14A86 (1986).
Wang XT, Dumoulin MJ, Befani 0, Mondovi B, Mateescu MA. Joint chromato-
graphic purification of bovine serum ceruloplasmin and amineoxidase. Prep Bio-
chem 1994;24:237-250.
Mondovi B, Gerosa P, Cavaliere R. Studies on the effect of polyamines and their
products on Ehrlich ascites tumours. Agents Actions 1982; 1 :450-45 1.
Agostinelli E, Bates DA, Przybytkowski E, Mateescu MA, Mondovi B. Cytotoxicity
of polyamines in Chinese hamster ovarian (CHO) cells in the presence of serum
amine oxidase. Life Chemistry Reports 1991;Y:193-204.
Mondovi B, Riccio P, Agostinelli E. The biological functions of amine oxidases and
their reaction products: an overview. Adv Exp Med Biol 1988;250:147-161.
Mateescu MA, Dumoulin MJ, Wang XT, Nadeau R, Mondovi B. A new physiologi-
cal role of copper amine oxidases: cardioprotection against reactive oxygen interme-
diates. J Physiol Pharmacol 1997;48 Suppl 2: 110-121.
Mondovi B, Wang XT, Pietrangeli P, Wang R, Nadeau R, Mateescu MA. New as-
pects on the physiological role of copper amineoxidases. Curr Top Med Chem 1997;
2:3 1-43,
Wu L, Mateescu MA, Wang XT, Mondovi B. Wang R. Modulation of K ' channel
currents by serum amineoxidase in neurons. Biochem Biophys Res Commun 1996;
220:47--52.
Wang XT, Pietrangeli P, Mateescu MA, Mondovi B. Extended substrate specificity
of serum amine oxidase: possible involvement in protein posttranslational modifica-
tion. Biochem Biophys Res Commun 1996;223:91-97.
Yamada H, Yasunobu KT. Purification, crystallization and properties of plasma
monoamine oxidase. J Biol Chem l962;237: 151 1- 1516.
Svenson A. Hynning PA. Preparation of amine oxidase from bovine serum by affin-
ity chromatography on aminoheyxyl-Sepharose. Prep Biochem 1981;11:99-108.
Vianello F, Di Paolo ML, Zennaro L, Stevanato R, Rigo A. Isolation of amine oxi-
dase from bovine plasma by a two-step procedure. Protein Expression Puri 1992;3:
362-367.
Mondovi B, Turini P, Befani 0, Sabatini S. Purification of bovine plasma amine
oxidase. Meth Enzymol 1983;94:3 14-3 18.
Susan-Janes M, David MV, Wemmer D, Smith AJ, Kaur S, Maltby D, Burlingam
AL, Klinman JP. A new redox cofactor in eukaryotic enzymes: 6-hydroxydopa at
the active site of bovine serum amine oxidase. Science l990;248:98 1-987.
Atanasiu R, Gouin L, Mateescu MA, Cardinal R, Nadeau R. Class 111 antiarrhythmic
effects of ceruloplasmin on rat heart. Can J Physiol Pharmacol 1996;74:652-656.
Osaki S, Johnson DA, Frieden E. The possible significance of the ferrous oxidase activity
of ceruloplasmin in normal human serum. J Biol Chem 1966;241:2746-275 1.
Tabor CA, Tabor H, Rosenthal SM. Purification of amine oxidase from beef plasma.
J Biol Chem 1954;208:645-665.
Bradford M. A rapid and sensitive method for the quantitation of microgram of protein
utilizing the principle of protein-dye binding. Anal Biochem 1976;72:248-254.
Mateescu MA, Fortier G, Neidhart S, Roger S. Selective immobilization of cerulo-
plasmin on the chromatographic material used for its purification. Chromatographia
1988;26:110-121.
Fast, Single-Step Affinity
Chromatography Purification
of Hemoglobin
Mircea-Alexandru Mateescu and Wilfrid Jacques
Universite du Quebec a Montreal, Quebec, Canada
I. INTRODUCTION
Hemoglobin (Hb), the main oxygen carrier, is a tetramer in which each monomer
( a , , a,, p,; and p,) contains a porphyrin prosthetic group. Initiated in 1937, the
long path to understanding Hb biochemistry led to high-resolution X-ray Hb anal-
ysis (Perutz, 1968), which was a landmark achievement for the formulation of
the "Perutz mechanism" based on the stereochemistry of the cooperative effects
in Hb oxygenation [I]. One of the most complex proteins, Hb is also the most
studied over the last 30 years. Since oxygen exchange is controlled by the revers-
ible steric configuration of the protein, Hb has been nominated as an "honorary
enzyme" (Voet and Voet [2]).
Hemoglobin is not a glycoprotein, but it can be easily glycosylated with
free glucose, by a nonenzymatic mechanism. About 6% of the erythrocytary Hb
is glycosylated in normal subjects. In diabetes the glucose level can be three
times higher [3,4], leading, with different rates, to glycated forms HbA,,, HbAlh,
and HbAI,. More knowledge of interactions of Hb with sugars is of interest since
the oxygenation-deoxygenation capacity of the glycated Hb is reduced by 20-
25%. Furthermore, the effects of the alteration of oxygen transport function are
enhanced by the relatively high half-life of Hb, up to 120 days. It was also
Part of this work was adapted from W. Jacques and M. A. Mateescu, Anal. Lett. 1993;26:875-886.
Copyright O 1993 by Marcel Dekker, Inc.
446 Mateescu and Jacques
reported [5] that, unexpectedly, the velocity of glycosylation of the amine termi-
nal group of the P subunit located on the allosteric center of Hb is three times
higher than the glycosylation rate of other amine groups on the Hb surface. Since
the mechanism is nonenzymatic, the kinetics is in disagreement with the statistical
distribution of amine groups susceptible to glycosylation.
On the basis of the peculiar kinetics of glycosylation, we have advanced
the hypothesis of a selective recognition center for sugars. Previously [6,7], we
have described specific interactions of Hb with some polysaccharides as CL-
amylose (CLA) and CL-agarose, while practically no interaction has been ob-
served with CL-dextran (Sephadex). All of these chromatographic materials are
obtained by the polysaccharide chain treatment with epichlorohydrin [8,9], lead-
ing to swellable three-dimensional polymeric structures. The fact that Hb is re-
tained by CL-amylose (main chains based on glucose units linked by a-1,4-glyco-
sidic bonds) and not by Sephadex (main chains based on glucose units linked by
a-1,6-glycosidic bonds) suggests that Hb can selectively recognize and differenti-
ate between the two types of linkages characteristic for each polymer. At the
same time it was shown that Hb can recognize the CL-agarose (main chains
based on galactose and 3,6-anhydro-I--galactose), indicating that both the type
of linkage and the saccharidic units forming the carbohydrate polymer are impor-
tant [7]. Furthermore, the study described specific interactions of Hb with mono-
and disaccharides. The competitive elution of Hb retained by affinity on CL-
arnylose and on CL-agarose (Sepharose CL-4B) chromatographic materials, with
different sugars as competitive "affinants," showed that glucose, fructose, and
cellobiose are poor eluants, whereas lactose is a good competitive eluant, sug-
gesting that it better interacts with Hb and, more important, that the interaction
is selective [7].
Another related study reported Hb separation by specific affinity chroma-
tography retention on CL-arnylose, with a capacity of 4.5-10 mg Hblml gel bed,
followed by desorption with deforming eluants, as NaCl [6]. The CL-amylose
material was previously used in exclusion chromatography [9] and in affinity
chromatography for a-amylase separation [ 101.
This new aspect on the Hb properites can offer a better understanding of
Hb-carbohydrate and Hb-glycoprotein interactions and, at the same time, a pos-
sibiltiy to purify Hb directly from erythrocytary lysate, by a single-step chromato-
graphic procedure on crosslinked amylose columns [6]. These Hb separation ex-
periments were carried out using commercial bovine Hb (Sigma Chemical Co.)
solutions.
Hemoglobin and its derivatives are currently isolated by various ion ex-
change chromatographic techniques. However, several difficulties have been re-
ported, mainly related to the Hb contamination with various compounds, and
often a rechromatography is required. Thus, alternative procedures have been
elaborated based on the Hb elution from the starch gel after the electrophoretic
Purification of Hemoglobin 447
Hb separation [ll]. Several separation procedures were described for the Hb

purification by affinity chromatography on ATP-agarose columns [12,13]. This
method is efficient for separation of Hb from red cells derived from unconven-
tional sources (blood from surgery, placenta, etc.) [12], but is long enough, since
ultracentrifugation and dialysis steps are required after hemolysis and prior to
ATP-agarose column. The procedures that allow the preservation of Hb capacity
to retain and to exchange oxygen as oxyhemoglobin (i.e., the crosslinked Hb or
the ATP-Hb complex [13]) are of great interest for the transfusional medicine.
Fast methods, minimizing the sample story (preventing the accumulation of met-
hemoglobin) and large-scale procedures, are of great interest for laboratories in-
volved in Hb purification.
We are now presenting a fast, single-step, easy-to-apply method for the Hb
purification directly form erythrocytes, by affinity chromatography on polysac-
charidic chromatographic materials such as CL-agarose i.e., Superose 12 and
Sepharose CL.
II. EXPERIMENTAL
A. Materials and Reagents
Agarose-based chromatographic materials (Sepharose C L 4 B , Sepharose CL-6B,
Superose 12 Prep. Grade, Sephadex G-25, and Sephadex G-100) were from Phar-
macia LKB (Uppsala, Sweden). Commercial bovine hemoglobin (lyophilizedpow-
der), bovine serum albumin (BSA), and Coomassie brillant blue R-250 were from
Sigma Chem. Co (St. Louis, Missouri). CL-amylose (CLA) is a semisynthetic po-
lyglucose chromatographic material realized by amylose crosslinking with epichl-
orohydrin [6,7,9]. As the Sephadex gel, CLA was here used almost for comparison
in terms of Hb interactions with agarose-based materials. Human blood was ob-
tained from the H6pital Charles-Lemoyne, Longueuil (Qukbec), Canada.
B. Separation of Erythrocytary Hemoglobin by Single-Step

Affinity Chromatography on Agarose-Based Column
1. Bovine Erythrocytary Lysate
The crude bovine erythrocytary extract was prepared from fresh red blood cells
separated from plasma by centrifugation at lOOOg for 30 min at 4°C. The superna-
tant was discarded and the erythrocytes were washed with an equal volume of
isotonic (0.9%) NaCl solution, followed by centrifugation under the same condi-
tions as before. The washings were repeated three times. Then 1 volume of eryth-
rocyte sediment was suspended in 10 volumes of bidistilled water for hemolysis
and the suspension was centrifuged at 12,000g for 20 min at 4"C, retaining the
supernatant.
2. Human Erythrocytary Preparation

Human blood was subjected to the same procedures as bovine blood. Human
Hb extract (supernatant) was stored at -20°C. Since the Hb interaction with
polysaccharides is highly sensitive to ionic strength, the Hb extract conductivity
has been determined and adjusted to values less than 15 mmho, by dilution or
dialysis.
This chapter describes the fast Hb purification at a laboratory scale. The
recommended chromatographic material is Superose 12 Prep. Grade, which has
the best retention capacity. Alternatively, other crossliked materials, selected
from Sepharose C L 4 B , Sepharose CL-6B, or CL-amylose, can be used leading
to Hb with comparable homogeneity but with moderately lower yields.
3. Affinity Chromatography Hemoglobin Separation on

Agarose-Based Column
For separation, a volume of about 2-3 ml (or more) of sample (5-10 mglml) of
commercial Hb dissolved in bidistilled water or erythrocytary extract was applied
to the Superose 12 column (10 X 2.5 cm, 40-ml gel bed). The column was first
washed with bidistilled water for the elution of nonretained contaminants and
then eluted isocratically with 0.9% NaCl or with an increasing continuous gradi-
ent up to 5% NaCl solution, with an elution flow of 25 mllh. The gradient elution
was realized with a gradient mixer system (Pharmacia), over a volume of 60 ml.
The entire operation has been camed out at 20 mllh (with a Marlow peristaltic
pump), collecting fractions of 2 ml (Gilson fraction collector). Since Hb is eluted
with a relatively low concentration (0.1-1%) NaCl, for further separation runs
a 0.9% NaCl concentration was used for the Hb elution step.
The column hemoglobin retention capacities (expressed in mg Hblml gel
bed) of various chromatographic materials were determined under the same con-
ditions: microcolumns (5 X 0.6 cm) with a 3-ml gel bed, equilibrated in deminer-
alized water and a flow rate of 20 mllh). The columns were overloaded with
commercial Hb samples and washed to eliminate nonretained Hb. The capacity
was established by the integration of the Hb amounts determined in each collected
fraction (following elution with 0.9% NaCl solution), as previously described [6].
Retention capacities for the types of chromatographic material used (Superose
12, Sepharose CL-4B, Sepharose CL-6B, CL-amylose-40) were evaluated and
compared with those of Sephadex G-25 and Sephadex G-100 [14]. Retention
capacities were also evaluated for related hemoglobin proteins as globin (heme-
free hemoglobin monomer) and for horse heart myoglobin.
Hemoglobin concentration has been determined spectrophotometrically at
h = 406 nm (characteristic for the heme prosthetic group) and at h = 280 nm

(characteristic for the proteic part, the globin). Calibration curves have been ob-
tained from commercial Hb samples in concentration up to 250 pg/ml, as stan-
dard.
The Hb purity degree was determined from the A4n6,,,IA,8n,,,ratio, and also
checked by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-
PAGE) (10% polyacrylamide) as described by Laemmli [15] and staining with
Coomassie brilliant blue R-250. The molecular integrity of the Hb tetramer puri-
fied on Superose 12 column was confirmed by exclusion chromatography on
Sephadex G-100 (1 X 70 cm) column equilibrated in 0.045 M sodium phosphate
buffer, pH 7.2. The column parameters were evaluated with 0.5 ml calibrating
mixture consisting of blue dextran T-2000 (for the void volume, Vn), BSA, and
cytochrome c , in concentrations of 2 mglml each.
Ill. RESULTS AND DISCUSSION
The chromatographic profile of Hb purification by affinity retention on Superose

12 column and deforming elution with 0.9% NaCl (0.15 M) solution is presented
in Fig. 1A. Since hemoglobin is sensitive to ionic strength, the deforming saline
solution produces a fast and total Hb desorption in a sharp peak. Similar results
were obtained with methemoglobin and this, irrespective of the oxy- or deoxy-
Hb form (data not shown).
Since it was previously reported [7] that among monosaccharides only ga-
lactose was able to generate a desorption (with a diffuse peak) and among disac-
charides only lactose (glucose linked to galactose) generated a sharper elution
of Hb from the CL-amylose, the conclusion was that Hb preferentially interacts
with galactose-containing sugars. Thus, Hb retention capacities on crosslinked
agarose (containing galactose and 3,6-anhydro-L-galactose units), chromato-
graphic materials (Sepharose C L d B , Sepharose CL-6B, and Superose 12), and
on CL-amylose (a material exhibiting long chains of glucose repetitive units
linked by a-1,4-glucosidic bonds that can be selectively and reversibly recog-
nized by Hb [6]), were determined (Table I). Sepharose CL-4B has a retention
capacity (8.4 mg Hblml gel bed) comparable to that observed for the CL-amylose
(8.0 mg Hblml gel bed). The Hb retention capacities of various agarose chromato-
graphic materials (Table 1) depend directly on the concentration of agarose in
the gel bed.
Since the best retention capacity (23 mglml) was obtained with Superose
12 (Prep. Grade), the experimental details of the procedure for Hb purification
are based on this chromatographic material.
Commercial Hb in aqueous solution can be retained on crosslinked agarose
or amylose columns and recovered by elution with physiological (0.9%) NaCl
Table 1 Available Hemoglobin Retention Capacity of Different Carbohydrate

Chromatographic Materials
Hb retention capacity
Chromatographic material (mg Hblml gel bed) Carbohydrate matrix
CL-Amylose (CLA-40)" 8.0 2 0.6 Crosslinked amylose
Sepharose CL-4Bh 8.4 2 0.8 Crosslinked agarose
Sepharose CL-6B 16.5 2 0.9 Crosslinked agarose
Superose 12 Prep grade 22.9 2 1.1 Highly crosslinked agarose
" CLA-40: Crosslinked (CL)-arnylose;The number indicates the crosslinking degree expressed as the
amount in grams of epichlorohydrin used to crosslink 100 grams of amylose.
For agarose chromatographic materials, the number indicates the approximate percentage of carbo-
hydrate in the gel; B stands for bead form.
0 2 4 6 8 10 12 14 16 18 20 22
Elution Volume (mL)
Figure 1 Typical chromatographic (A) and electrophoretic (B) separation profiles of a
mixture of bovine serum albumin (BSA) (1 mglml) and commercial hemoglobin (Hb)(l
mglml) on a Superose 12 column (4 ml gel bed). The absorbances A,,, (-o-) and A280n,
(-) yielded for the first peak a ratio value A,,,,IA,s,,,,,,, = 0.21 and for the retained Hb
A40an,lA280,,= 5.2. The sample volume was 2 ml. SDS-PAGE was run for standard molec-
ular weight (Jane l), Hb and BSA mixture before separation (lane 2), nonretained (ascribed
to BSA) fraction (lane 3), and retained Hb eluted from the CLA column (lane 4).
solution. In the case of Hb associated with other proteins, e.g., with BSA in equal
concentrations in aqueous samples, a selective retention of Hb on the Superose
12 column was observed, while a total elution of the nonretained BSA was found
in the first peak (Figure 1).
No Hb retention was observed in identical chromatographic conditions on
Sephadex G-25 or Sephadex G- 100 (crosslinked dextran based on a-1,6-glucosi-
dic bonds). These data clearly indicated that Hb can specifically recognize and
be retained at the level of galactose and 3,6-anhydro-L-galactose sequences of
the Superose 12 chromatographic material. The Hb was eluted in a single step,
with a 0.9% NaCl solution (Figure 1A). It is worth noting that, being sensitive
to ionic strength, Hb can be desorbed at even lower NaCl concentrations. The
yield of Hb recovery was higher than 85%.
The BSA was chosen as a model contaminant because this protein, like
Hb, is not a glycoprotein, and presents a molecular weight of 68 kDa, close to
that of Hb (64 m a ) . The presence of small amounts of Hb in the first peak of
BSA or other contaminant proteins, even if the column is not overloaded, can
be ascribed to the presence of some modified forms in the commercial Hb (be-
yond the methemoglobin and the dimers, which can be associated with Hb in
commercial preparations). In fact, the presence of dimers (32 kDa) was confirmed
as a minor peak in the commercial Hb, by exclusion chromatoraphy of Sephadex
G-100, eluted after the major tetrameric Hb peak (64 m a ) .
The SDS-PAGE runs (Fig. 1B) clearly indicated the specific retention of
Hb on Superose 12, whereas BSA is not retained at all. In fact, only Hb recovered
from the column was visualized on lane 4 as a single band corresponding to the
Hb monomer (16 kDa) and another minor one for the dimer (32 kDa).
On the basis of this specific Hb retention onto the Superose 12 column,
the proposed affinity chromatography method allows the single-step purification
of Hb, directly from erythrocytary lysate. A selective chromatographic retention
of erythrocytary Hb (Figure 2) from a crude heterogeneous extract obtained by
red cell hemolysis in water (1 : 11 vlv), was found. As in the case of commercial
Hb, recovery of the erythrocytary Hb can be easily realized by direct elution with
physiological 0.9% NaCl solution (Figure 2A).
The absorbancy ratio A406nm/A280nm was considered (beyond SDS-PAGE) a
criterion to assess the purity of Hb eluted from the column. The purification
parameters are presented in Table 2. For all of the purified Hb recovered peaks,
the A,MnmIA2,0,m ratio values were 4.3 (as for the standard commercial Hb) or
even higher, up to 5.3, indicating an improvement of the degree of purity of
the commercial Hb. The first nonretained peaks (contaminants and eventually
overloaded Hb) present lower ratio values, due to various contaminants absorbing
at 280 nm. The retention capacity and the elution profiles were obtained with
small columns (1 X 5 cm), but the scaling-up process with bigger columns (50-
to 100-ml gel bed or more) allows the large-scale purification of Hb from blood
Mateescu and Jacques
Hb
dime r
24000
Hb
'8'0° monomer
B
0 6 12 18 24 30 36 42 48 54 60
Elution Volume (mL)
Figure 2 Typical chromatographic (A) and electrophoretic (B) profiles of the erythrocy-
tary separation on a Superose 12 column (3-ml gel bed). The absorbances (-0.) and
A,,,, (-) yielded for the erythrocyte lysate a ratio value Am,,lnlA2xonm= 3.7 and for the
retained Hb A41KII,,,IA2X0,,,II = 4.5. SDS-PAGE was run for standard molecular weight (lane
I), heterogeneous erythrocytary hemolysate before separation (lane 2). nonretained frac-
tion (lane 3), and rctained Hb eluted from the Superose 12 column (lane 4). The separation
conditions are presented In the "Materials and Methods" section.
Table 2 by Affinity Chromatography on Superose 12

Hernoglob~nPur~ficat~on
Column
Retention capacity
Purlfied H b (sources) Ari~,,,IAzuo,., (mg H b h l gel bed) Observations
Commercial Hb" 5.2 + 0.2 22.9 3- 1.1 A unique electrophoretic
Hb from erythrocytarylysate 4.9 + 0.3 23.5 + 1.8 band (64 kDa) was con-
firmed by exclusion
chromatography on
Sephadex G-100.
"For the commercial Hb (Sigma Chemical Co.) before the column, the A,IA,,,, ratio was 4.3. The
data represent a mean of four experimental values.
sources. The Superose 12 columns were repeatedly used for tens of cycles without
significant alteration of the retention capacity. A condition for the column to
work is that it be thoroughly washed first with 0. I M NaOH and then with HzO,
from one cycle to another (until conductivity in washing eluate decreases at less
than 15-30 mmho). However, after more than 50 cycles, and under certain condi-
tions, the mechanical properties could be partially affected by alkaline (0.1 M
NaOH) washings between the affinity cycles.
The erythrocytary Hb recovered from CLA-40 column (Figure 2B) was
electrophoretically (SDS-PAGE) homogeneous, obtaining a strong band charac-
teristic for Hb monomer (16 kDa) and another minor band for the Hb dimer (32
kDa) only, as for the purified commercial Hb (Figure 1 B, lanes 2 and 4). Dissocia-
tion into monomers is due to the presence of SDS in the electrophoretical gel
and has been reported by other groups [16].
The molecular integrity of purified Hb (64 kDa) was confirmed by exclu-
sion chromatography on Sephadex G-100, where a unique peak was observed at
the same elution volume (V,) generating a K, = 0.17, close to that found for
BSA (68 kDa).
The specific affinity interaction of bovine Hb with CL-agarose (Superose
12) and with CL-amylose was comparable with that found for human Hb; the
elution profiles (data not shown) were similar to those of bovine Hb. Furthermore,
the affinity retention on Superose 12 and on CL-amylose seems to be specific
for Hb and also for some related proteins. Globin, the heme-free hemoglobin
monomer, was shown to interact moderately with the CL-amylose, but to a lesser
extent than Hb [14], whereas heme is not retained at all. The similarities between
the affinity behavior of human and bovine Hb on agarose chromatographic mate-
rials, here mentioned, as well as previous data [14] showing affinity interaction
of horse heart myoglobin with the CL-agarose, suggest that the interaction with
these polysaccharides can be a general property of hemoglobins and related pro-
teins of various species. The selective sugar recognition of hemoglobins and re-
lated proteins appears to be irrespective of the origin (human or bovine hemoglo-
bin and horse heart myoglobin). The retention capacities of CL-amylose for
myoglobin (18.8 kDa) was of 20 mglg dry support [14], a value of about 25%
of the retention capacity of 86 mglg dry support for the tetrameric bovine Hb
(64 kDa). Since myoglobin's structure is very close to the Hb monomers (16
kDa), it is possible to suppose a unique type of sugar recognition center on the
Hb molecule. This supposition fits with the data of Haney and Bunn [5]. on the
higher rate and yield of nonenzymatic glycosylation of one P subunit, leading to
an excess of HbAl, glycosylated form when compared to the forms HbA,,, and
HbAlb.
The advantage of the method described here is that no other plasma or
erythrocytary proteins interact, under the experimental conditions presented,
with these CL-amylose and CL-agarose chromatographic materials that are not
derivatized with functional ionic exchange groups. In fact, affinity chromatogra-

phy on Superose 12 allows the single-step separation of Hb, electrophoretically
homogeneous, from the erythrocytary extracts. This is possible due to Hb specific
affinity for nonsubstituted agarose or amylose materials.
As previously mentioned, the elution of Hb retained by Superose 12 is very
sensitive to ionic strength; NaCl concentrations as low as 0.01 N can produce a
total Hb desorption by deforming elution [6]. It was also previously shown that,
as an alternative, Hb can be desorbed by competitive elution with a continuous
gradient of lactose [7]. Hemoglobin elution started at 0.05 M lactose, whereas
the other sugars such as glucose, fructose, and saccharose, even at a concentration
of 0.5 M, were unable to desorb Hb [7]. These elution data suggest that the Hb
interaction with (po1y)saccharides is specific and highly related to the folding
arrangement of the protein. Further works are in course for the quantification of
different Hb-sugar interactions, and for the localization and characterization of
the saccharide recognition center for normal and pathological hemoglobins from
different origins. Aspects of the Hb-sugar interactions will be dealt with in fur-
ther reports.
Since agarose chromatographic materials (Superose) resist at high compres-
sion effort, the fast, single-step method reported here can be extended for large-
scale Hb affinity purification in batch, continuous recycling, and fast protein liq-
uid chromatography (FPLC). The method is easy to apply, allowing purification
of electrophoretically homogeneous Hb within a few hours.
ACKNOWLEDGMENTS
A FCAR fellowship awarded to W.J. for graduate studies at UQAM is gratefully

acknowledged.
REFERENCES
1. Perutz MF. Stereochemistry of cooperative effects in haemoglobin. Nature 1970;

228:726-734.
2. Voet D, Voet JG. Biochemistry. New York: John Wiley and Sons, 1990:211-243.
3. Kokiloo N, Sjddiqui M. Interrelationship between blood glucose level, plasma pro-
teins and hemoglobin glycosylation. Ind J Clin Bjochem 1987;2:34-38.
4. Kennedy L, Lyons TJ. Non-enzymatic glycosylation. Br Med Bull 1989;45:174-
190.
5 . Haney DN, Bunn F. Glycosylation of hemoglobin in vitro: affinity labeling of hemo-
globin by glucose-6-phosphate. Proc Natl Acad Sci USA 1979;73:3534-3538.
Jacques W, Mateescu MA. Affinity chromatography of hemoglobin on cross-linked

amylose. Anal Lett 1993;26:875-886.
Jacques W, Mateescu MA. Specific hemoglobin (po1y)sacchariderecognition. J Mol
Recogn I995;8:106-llO.
Porath J, Flodin P. Gel filtration: a method for desalting and group separation. Nature
1959;183:1657-1659.
Serban M, Schell HD, Mateescu MA. Preparation and properties of new amylose
based chromatographic materials with application in exclusion chromatography (in
German). Rev Roum Biochim 1975;12:187-191.
Schell HD, Mateescu MA, Bentia T, Jifcu A. Alpha-amylase purification and separa-
tion from glucoamylase by affinity chromatography on cross-linked amylose. Anal
Lett (NY) 1981;14(B):1501-1514.
Luan Eng LI. Simple method for the isolation and purification of hemoglobin compo-
nents. J Chromatogr 1976;1 1753-58.
Carleton JCH, Er SS. Purification of stroma-free haemoglobin by ATP-agarose af-
finity chromatography. J Chromatogr 1986;374:143-148.
Carleton JCH, Er SS, Hronowski LJ, Persaud K, Ansari MR. ATP-hemoglobin puri-
fication by ATP-agarose affinity chromatography. J Chromatogr 1986;381 :153- 157.
Jacques W, Mateescu MA. Selective hemoglobin-polysaccharide interactions. 4th
Meeting on Biochromatography and Molecular Biology. La Grande Motte, France,
May 12-14, 1992.
Laemmli EK. Cleavage of structural proteins during the assembly of the head of
bacteriophage T4. Nature 1970;227:680-685.
Fantl WJ, Manning LR, Ueno H, Di Donato A, Manning JM. Properties of carboxy-
methylated cross-linked hemoglobin A. Biochemistry 1987;26:5755-5761.
Index
Abnormal hemoglobins, 2, 6 Analysis of pea-lectin sugar. 192
Acetylated hemoglobins, 3 Annexins, 293
Acrosin inhibitor, 334 Antibodies and antigens, 109, 290,
Acrylamide derivatives, 124 292
Active esters, 141 Aproferritin, 392
Adsorption chromatography, 99 Autoimmune deficiency syndrome
Adsorption-desorption process, 256 (AIDS) virus, 303
Affinity capillary electrophoresis, Avian-type C retroviruses, 322
187 Avidin-biotin technology, 1 14
Affinity chromatography, 2, 273,
286, 303, 4 15, 445 Bacillus subrifis tRNATT, 320
Affinity constant, 188 Bacterial lipoproteins, 35
Affinity filtration, 286, 290-291 Bacterial plasmid DNA, 296
Affinity ligands (affinants), 101 Baculovirus, 303
Affinity probe capillary electropho- Basic fibroblast growth factor, 402
resis (APCE), 188 Benzoquinone activation, 143
Affinophore, 193 Benzy lamine oxidase, 432
Affinophoresis, 188 b-amylase. 392
Agarose P-endorphin, 336
column, 447 P-galactosidase, 292
derivatives, 120 P-lactoglobulin b, 224
Albumin purification, 334, 359, P-microseminoprotein, 336
36 1 Bioaffinity chromatography, 99-
Alcohol dehydrogenase, 392 I64
Allergenic proteins, 353 Biotin-streptavidin affinity, 116
a,-Antitrypsin, clotting factor IX, Bombesin, 57
295 Borreliu burgdo$eri, 27
a-chymotrypsinogen, 282 Bovine immunoglobulin, 291
a-lactalbumin, 58 Bovine serum albumin, 292, 392
Amphiregulin, 409 Bovine serum amine oxidase, 432
Analysis of concanavalin a-sugar, Bradykinin, 5 7
196 Brownian motion, 424
Index
Bulk axial dispersion coefficient, Cyclodextrins, 52

Cytochrome c, 282
Cytokines and human reproduction,
Caicium-binding protein (calmodu- 346
lin), 336
Capillary electrophoresis, 354-428 Dalargin, 62, 77, 81
Capillary isotachophoresis, 48, 188, Deforming buffer, 100
Carbamylated hemoglobin, 3 Der pl, a house dust mite major al-
Carbonic anhydrase, 290, 392 lergen, 362
Carbonylating reagents, 142 Desoctapeptide insulin, 50
Carboxypeptidase Y, 105 Detection time, 189
Carnobacteriocin B2 immunity pro- Deuterium exchange mass spectrom-
tein (CbiB2), 317 etry, 17
Cation exchange chromatography, Dextran gels, 122
6, 398 Dihydrofolate reductase, 307, 3 14
Cationic detergent, 52 Disassociation constants
Cellulose and its derivatives, 123 Con A, I97
Ceruloplasmin purification, 43 1 - pea lectin, 195
432, 435, 437 Displacement chromatography,
Cervical carcinoma, 343 applications, 236-244
Chaotropic agents, 52 column length, 231
Chemically synthesized polypep- displacers, 233
tide, 307 equilibrium-dispersive model,
Chromatofocusing, 354 217
Circular DNA, 422, 426 flow rate, 232
Coelectrophoresis, 190 kinetic models, 219
Column hemoglobin retention ca- mobile phase, 23 1
pacities, 448 sample preparation, 245-246
Concanavalin A, 105 sample size, 23 1
Connective tissue growth factor, special forms, 235
397-4 14 stationary phase, 228
amino acid sequencing, 404 Displacer, 206
bioassay, 400
protein sequencing, 400 E. coli maltose-binding protein,
purification, 398, 409 302
Continuous annular chromatogra- E. coli thioredoxin, 302
phy, 255 Edman degradation, 25
Continuous bed column, 229 Effective charge, 41
Controlled pore glass, 129 Effective mobility, 40
Convective interactive media, 258 Electrochemical and conductivity
Coupling procedures, 133 detection, 58
Cyanogen bromide activation, 140 Electroosmotic flow, 72, 189
Index
Electrophoresis, 99, 303 Hemagglutinin-neurarninidase, 3 14

Electrophoretic mobility, 189 Hemoglobin
Elution azeotropes, 225 purification, 445
Endotoxin removal, 294 retention capacity, 450
Enkephalins, 57 Hemoglobin A,, determination
Enzyme-linked immunosorbent by affinity chromatography, 5
assay, 354 by cation exchange chromatogra-
Epoxide-containing supports, 135 phy, 2
Equ c 1, a horse major allergen, 354 Hemoglobinopathies in newborns,
Equine insulin, 52 6
Erythropoietin, 301 Heparin affinity
Extracellular fusion proteins, 305 chromatography, 397
Extracellular maltose-binding pro- columns, 408
teins, 31 8 membrane adsorber, 276
Heparin-sepharose chromatography,
410
Factor VIII, 292-294, 303, 432
Hepatitis B virus, 303
Fast atom bombardment mass spec-
High-performance capillary zone
trometry, 33
electrophoresis, 40
Fast protein liquid chromatography,
High-performance displacement
454
chromatography, 21 5
Fatty acid analysis, 27
High-performance liquid chroma-
Feld 1, a cat major allergen, 357
tography, 4 15
Ferroxidase i, 431
High-performance membrane chro-
Fetal hemoglobin, 2
matography, 255
Free-flow zone electrophoresis, 39
HIV- I protease, 3 17
Fusion proteins purification, 305
Hollow fiber systems, 29 1-292
Honorary enzyme, 445
Gel filtration, 99, 303 Human antithrombin 111, 290
Gel permeation chromatography, Human cathepsin D, 83
415 Human endothelin, 3 16
Globulins, 334 Human growth hormone, 60, 301,
Glutaraldehyde activation tech- 323
nique, 139 Human immunoglobulin G, 110
Glutathione-agarose, 305 Human insulin, 62, 302
Glutathione S-transferase, 302, 305 Human proinsulin fusion protein,
Glycohemoglobin, 1 306
Glycoproteins, 110, 334, 353, 433 Human seminal plasma, 33 1
Gradient elution, 259 immunological effects in vitro,
Growth factor-1, 113, 406 34 1-343
Growth hormone-releasing peptide, immunological effects in vivo,
83 340-34 1
Index
[Human seminal plasma] Ion exchange chromatography, 99,

isolation, 337-340 286
phagocytic activity, 343-345 Ionogenic groups, 42
proteins, 332-333 Iron and nickel oxides, 131
Human seminal plasma p-micro- Isocratic elution, 259
seminoprotein, 335 Isoelectric point, 48
Human serum albumin, 294 Isoionic point, 44
Human tumor necrosis factor-a,
293 Kallikrein hK2, 335
Hyaloronic acid, 393-394 Keratinocyte growth factor, 409
Hydrazide-derivatized solid sup- Kistler and Nitschmann's fraction
ports, 137 IV, 293
Hydrodynamic chromatography,
425 Langmuir's adsorption isotherm,
Hydrophobic interaction chromatog- 191
raphy, 228, 304, 356 Lectins. 1 10, I87
HyperD columns, 229 Lectin-sugar interactions, 188
Hysteresis, 264 Leukemia inhibitory factor, 305,307
Levan and Vermeulen isotherm,
225
Immobilized metal affinity chroma- Lipids. 1 13
tography, 217, 355 Lipocalin-type prostaglandin D syn-
Immunoaffinity chromatography, thase, 337
1 13, 398 London's dispersion forces, 100
Immunoglobulin, 36 1 Luminol-dependent chemilumines-
Immunosorbent, 155 cence, 343
Immunotoxins, 324 Luteinizing hormone-releasing hor-
Insulin, 50, 301 mone, 83, I69
Insulin-containing proteins, 306 Lyme disease, 14
Insulin-like growth factor, 60
Integral membrane proteins, 308 Maltose-binding proteins purifica-
Interferon-a2a, 29 1 tion, 315
Interferon-y, 323, 347 Mass spectrometric detection, 56
Interferons, 301 Membrane adsorber, 258
Interleukin- l -P, 346 Membrane chromatography, 272
Interleukin-2 receptor, 29 1 applications, 272-277
Interleukin-6, 347 efficiency, 269
Interleukins, 324 fluid dynamines, 270
Intracellular fusion proteins, 322 mixed mode, 282-286
Intracellular maltose-binding pro- preparative, 277-282
teins, 3 18 resolution, 269
Ion-binding proteins, 336 theoretical basis, 259-263
Index
Membrane cofactor protein Phagocytic index, 344

(MCP :CD46). 334 Physiochemical characterization of
Methacrylate supports, 125 TCM, 17 1
Micellar electrokinetic chromatogra- Pig insulin, 63
phy, 52 Placental proteins, 336
Molecular imprinting technology, Plasmodium falci~mrum,3 12
144 Plate height, 272
Monoclonal antibodies, 1 13, 153, Pleiotrophoin, 402
290, 292-294, 304 Polyamines, 336, 343, 433
Mucor miehei proteinase, 101 Polyclonal antibodies, 156
Multiangle light scattering, 369- Polyhedrin insulin-like growth fac-
396 tor 2 (IGF-2), 3 18
applications, 39 1-395 Polymerase chain reaction, 4 16
mechanism, 423 Polynucleotides, 4 15
theoretical review, 370-372 Polysaccharides, 353, 39 1
Polystyrene and its derivatives,
Nonlinear chromatography, 209 127
N-Succinylated glutathione, 193 Porous and nonporous silica, 130
Nucleotides and nucleic acids, 1 12 Posttranslational lipidation, 27
Nylon membranes, 132 Pregnancy-associated Protein A,
336
0-acetylserine-0-acetlyhomo-serine Preparation of TCM, 170
sulfhydrylase (OAS-OAH), Preparative free-flow zone electro-
320 phoresis, 69
Octopus hemocyanin, 392 Preparative separations of peptides,
Optical detection, 53 65
Ornithine decarboxylase, 392 Principle of bioaffinity chromatogra-
Overdisplacement phenomenon, phy, 99
223 Prostasomes, 343
Oxidative sulfitolysis, 305 Protein C inhibitor, 335
Oxytocin, 57 Protein kinase, 336
Protein separation
Peak parameters in desorption chro- by mass transfer effects, 263
matography, 266 one step desorption process, 264
Peptide mapping by LC-MS, 17 Protein sequencing, 400
Peptides, 39, 153 Proteinase inhibitors, 102
Peptide sequence analysis of Proteoglycans, 428
TCM-1, 179 Prothymosin-cx mRNA, 182
Perfusion chromatography, 259 Pseudomonas auruginosa exotoxin
Perfusion chromatography beads, A, 40, 305
229 Pulsed-field gel electrophoresis,
Periodate oxidation, 138 42 8
Index
Q-sepharose column, 442 [Slalom chromatography]

Quantitative affinity chromatogra- effect of pore size, 421
phy, 102 effect of temperature, 422
mechanism, 423-425
Radioimmunoassay, 303 mixed mode, 426
Random coil model, 379 Sperm-binding proteins, 336
Rayleigh ratio, 372 Spermin, 336
Recombinant DNA-specified pro- Steric mass action, 216
teins, 153 Steroidogenic factors, 167
Recombinant HIV-1 reverse tran- Stokes radius, 416, 421
scriptase, 307, 314 Surface membrane fusion proteins,
Recombinant human insulin, 305 314, 318, 322
Recombinant proteins, 30 1 Synthetic copolymers, 123
Recombinant proteins by HPLC Synthetic polyamides, 132
and mass spectrometry, 13
Recombinant Schistosoma mansoni Tandem Mass Spectrometry, 20
glutathione-S-transferase T cells, 1 17
rSmp28, 14 Therapeutic proteins, 292-294
Reference values for HbA,,, 5 Thin-layer displacement, 246
Resolution and efficiency in Thiol-disulfide interaction, 143
HPMC, 269 Thymic epithelial cell conditioned
Reversed-phase HPLC, 403 medium, 167
Thymosin a , , 182
Saccharomyces cerevisiae, 15 Thymosin P4, 169
Saccharomyces cerevisiae MEY 171 Thymulin, 181
MET 25 gene, 320, 322 Thyroglobulin, 392
Schistosoma japonicum, 302 Tissue plasminogen activator, 301
Schistosomiasis, 14 Topoisomers, 426
Semicarbazide-sensitive amine oxi- Toxins, 113
dase, 432 Transferrins, 334
Seminal plasma motility inhibitor Triazine method, 143
(SPMI), 335 Tumor necrosis factor-a, 347
Sendai virus, 308 Ultracentrifugation, 303
Serum amine oxidase, 43 1
Sheep insulins, 52 Vascular endothelial growth factor,
Single-stranded DNAs, 426 409
Size exclusion chromatography, 303 Vitamins, 113
Slalom chromatography, 415-430
Whey proteins, 293
advantages and limitations, 427
application, 425-426 Zimm plot, 375
characteristics, 4 18-424 Zinc-binding proteins, 336
effect of flow rate, 421 Zwitterionic detergents, 52

Analytical and Preparative Separation Methods of Biomacromolecules PDF

Uploaded by

Copyright:

Available Formats

Analytical and Preparative Separation Methods of Biomacromolecules PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Analytical and Preparative Separation Methods of Biomacromolecules PDF

Uploaded by

Copyright:

Available Formats

0.0 10.0 20.

Analytical and preparative separation methods of biomacromolecules / edited by

This book is printed on acid-free paper.

Eastern Hemisphere Distribution

World Wide Web

Copyright O 1999 by Marcel Dekker, Inc. All Rights Reserved.

Current printing (last digit):

PRlNTED IN THE UNITED STATES OF AMERICA

It is my pleasure to be an editor of this book, which gave me a chance to be

Analysis of HbAIc and Some Hb Variants by HPLC 1

Analysis of Posttranslational Modifications in Recombinant Proteins

Analytical and Preparative Separations of Peptides by Capillary and

Characterization and Partial Purification of Steroidogenic Factors

Determination of Affinity Constants of Lectins for Sugars by Affinity

Displacement Chromatography of Biomolecules 203

High-Performance Membrane Chromatography of Proteins 255

HPLC Purification of Recombinant Proteins 301

Isolation, Purification, and Characterizaton of Human Seminal

Isolation, Purification, and Structural Study of Allergenic

Multiangle Light Scattering Combined with HPLC with Examples

Purification and Characterization of Connective Tissue Growth

Slalom Chromatography: A New Hydrodynamic-Based

Simultaneous Chromatographic Separation of Ceruloplasmin and

16. Fast, Single-Step Affinity Chromatography Purification of

Sultan T. Al-Sedairy Research Center Administration, King Faisal Specialist

Olivia Befani Department of Biochemical Sciences and CNR Center of Molec-

Rainer Bischoff, Ph.D.* Protein Analytical Group, Transgene S.A., Stras-

Bernadette Bouchon Protein Analytical Group, Transgene S.A., Strasbourg,

David R. Brigstock, Ph.D. Associate Professor, Surgery and Medical Bio-

Jean-Pierre Dandeu Unit6 d'Immuno-Allergie, Institut Pasteur, Paris, France

Marie-Josee Dumoulin Department of Chemistry and Biochemistry, Uni-

Thomas H. Fischer, Ph.D. Research Assistant Professor, The School of Medi-

Ruth Freitag, Ph.D. Professor, Chemistry Department, ETH Lausanne, Lau-

Afrozul Haq, Ph.D. Associate Scientist, Biological and Medical Research,

Jun Hirabayashi Department of Biological Chemistry, Faculty of Pharmaceu-

Wilfrid Jacques Professor, Department of Chemistry and Biochemistry, Uni-

Ken-ichi Kasai Faculty of Pharmaceutical Sciences, Teikyo University, Sa-

Viclav KaiiEka, Ph.D. Senior Research Scientist, Department of Biochemistry

Young C. Lin, Ph.D., D.V.M. Laboratory of Reproductive and Molecular En-

Patricia Martin, Ph.D., D.A.B.T. Senior Research and Development Chemist,

Mircea-Alexandru Mateescu, Ph.D. Professor, Department of Chemistry and

Bruno Mondovi, M.D. Professor, Department of Biochemical Sciences and

Sandra M. Scott, B.S. Research Assistant, Research and Development, Envi-

Kiyohito Shimura, Ph.D. Associate Professor, Department of Biological

Carr J. Smith, Ph.D., D.A.B.T. Master Scientist, Research and Development,

Ulf-Hikan Stenman, M.D., Ph.D. Laboratory, Helsinki University Central

Tatiana Tennikova Professor, Institute of Macromolecular Compounds, Rus-

Jaroslava Turkovi, Ph.D., Dr.%. Senior Research Scientist, Institute of Or-

Ursula Turpeinen, Ph.D. Chemist, Laboratory, Helsinki University Central

Mehmet Uzumcu, Ph.D., D.V.M. Assistant Scientist, Biological and Medical

Xin-Tao Wang, Ph.D. Department of Chemistry and Biochemistry, Universitk

Philip J. Wyatt, Ph.D. President, Wyatt Technology Corporation, Santa Bar-

This chapter deals with selected aspects of high-performance liquid chromatogra-

II. CURRENT HPLC METHODS FOR THE DETERMINATION

C. Diamat and Variant Analyzers

Ill. DETERMINATION OF HbAlc BY AFFINITY

The boronate affinity chromatography method involving minicolumns has gained

IV. REFERENCE VALUES FOR HbAlc

Determination of a method-specific reference range is important in glycohemo-