829

Biochem. J. (1984) 224, 829-837

Printed in Great Britain

Cyanide inhibition of cytochrome c oxidase

A rapid-freeze e.p.r. investigation

Peter JENSEN, Michael T. WILSON,* Roland AASA and Bo G. MALMSTROM

Department of Biochemistry and Biophysics, University of Gothenburg and Chalmers University of
Technology, S-412 96 Gioteborg, Sweden

(Received 14 May 1984/Accepted 29 August 1984)

The inhibition of cytochrome c oxidase by cyanide, starting either with the resting or
the pulsed enzyme, was studied by rapid-freeze quenching followed by quantitative
e.p.r. It is found that a partial reduction of cytochrome oxidase by transfer of 2 electron equivalents from ferrocytochrome c to cytochrome a and CuA will induce a transition from a closed to an open enzyme conformation, rendering the cytochrome a3CUB site accessible for cyanide binding, possibly as a bridging ligand. A heterogeneity
in the enzyme is observed in that an e.p.r. signal from the cytochrome a33+-HCN
complex is only found in 20% of the molecules, whereas the remaining cyanide-bound
a3-CuB sites are e.p.r.-silent.
The inhibition of cytochrome c oxidase (EC
1.9.3.1) by cyanide remains an interesting problem
half a century after its discovery. The present
phase of study of this phenomenon was initiated
by Van Buuren (1972), who in his comprehensive
thesis summarized the earlier findings and also reinvestigated the action of cyanide on the oxidase in
order to establish the site of inhibition and its
redox state. Among the four redox centres in the
enzyme (cytochrome a and a3, CUA and CUB), cytochrome a3 has long been considered the binding
site for cyanide as well as for many other
inhibitors. A dilemma, however, has been the
observation that the rate of cyanide binding to
either the fully oxidized or the fully reduced
enzyme is too slow to account for the rate of onset
of inhibition in turnover experiments or with
systems in vivo; recent work (Jones et al., 1984) has
established that the rates differ by as much as a
factor larger than 105. From his experiments, Van
Buuren (1972) concluded that the species susceptible to rapid inhibition is the partially reduced
oxidase. He also suggested that partial reduction
causes a conformational change that makes oxidized cytochrome a3 more available to cyanide.
In stopped-flow experiments with reduced cytochrome c and 02 (Antonini et al., 1977) it was
*

Visiting scientist of the Nobel Institute for Chemis-

try, 1983. Permanent address: Department of Chemistry, University of Essex, Wivenhoe Park, Colchester,

Essex C04 3SQ, U.K.

Vol. 224

found that cytochrome oxidase can appear in two

forms, which are optically similar but kinetically
different: the 'resting' and the (kinetically faster)
'pulsed' form. The term 'pulsed' refers to the
enzyme that has been freshly oxidized by the
reaction of the reduced enzyme with 02- In the
absence of reducing substrates it slowly decays
back to the resting form (Brunori et al., 1981). The
pulsed oxidase is related to another highly active
enzyme form found by Sekuzu et al. (1959) and
named 'oxygenated' oxidase by them. These

findings opened the possibility that the enzyme

forms with high activity are also the conformational states that bind cyanide rapidly, as had in
fact been suggested by Brittain & Greenwood
(1976) for the 'oxygenated' species.
This question has recently been addressed in
detail in a re-investigation by Jones et al. (1984) of
the kinetics of cyanide binding. They showed that
a model with only two conformational states,
resting and pulsed, is inadequate. An extended
model was suggested, in which the enzyme in the
pulsed as well as in the resting state can exist in a
'closed' or an 'open' conformation, of which only
the latter reacts rapidly with cyanide. It was
further proposed that the transition from the closed
to the open conformation is triggered by the
reduction of cytochrome a and CuA (and perhaps
also CUB), and that the conformational change is
followed by a rapid binding of cyanide to oxidized
cytochrome a3.
Although the model proposed by Jones et al.

830

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstrom

(1984) can adequately account for their extensive

kinetic results, it would seem desirable to have
more direct physical evidence to test the validity of
its details. To this end we have followed the
reaction of reduced cytochrome c with resting and
with pulsed oxidase in the absence and in the
presence of cyanide by e.p.r. spectroscopy with the
aid of a rapid-freeze quenching technique. Our
data show that the majority of the enzyme
molecules in the resting oxidase react rapidly with
cyanide when cytochrome a and CUA have been
reduced, and this prevents entry of more than 2
electron equivalents from reduced cytochrome c.
Also, with the pulsed oxidase only 2 electron
equivalents can enter the enzyme in the presence
of cyanide. In agreement with observations by
Brudvig et al. (1981), we find, however, that the
resting oxidase is heterogeneous. About 20% of
the molecules are in a form in which the cyanideinhibited enzyme has accepted 3 electron equivalents, as evidenced from the appearance of a lowspin Fe3+ e.p.r. signal from cytochrome a3-HCN,
showing that CUB must be reduced; this fraction of
the molecules gives a high-spin Fe3+ signal in the
absence of cyanide. Our results also indicate that
the pulsed oxidase is heterogeneous, some molecules reacting rapidly with cyanide before both
cytochrome a and CUA have been re-oxidized.

of re-oxidation. The amount of polymerized cytochrome c was less than 2%, determined from the
difference in absorbance between ascorbate and
dithionite-reduced cytochrome c at 550nm. Transferrin was kindly given by Dr. P. Aisen.
All experiments were performed in 50 mM-Hepes
[4-(2-hydroxyethyl)- 1 -piperazine-ethanesulphonic
acid]/NaOH buffer, pH 7.4, at room temperature.
Stock solutions of 0.48M-KCN were titrated with
HCl to neutrality before use.
All chemicals were of analytical grade. All
solutions were prepared fresh on the day of the
experiment.
Spectroscopy
Optical spectra were recorded in a Beckman
Acta MIV spectrophotometer. E.p.r. spectra at
15K and 9GHz were recorded with a Varian E-9
spectrometer equipped as described in Wilson et
al. (1982). Four to eight spectra were collected from
each sample and time-averaged to improve the
signal-to-noise ratio before integration.

Materials and methods

Enzyme and chemicals
Cytochrome c oxidase was prepared from ox
heart by the method of Van Buuren (1972). The
enzyme contained 10nmol of haem A and
0.1-0.15mg of phospholipids (measured as lipidbound P)/mg of protein. Absorbance ratios
(typical) were: A445(red.)/A42O(red.) > 2.3, A605(red.)/

was also replaced by a tangential rapid-flow mixer,

similar to the one described by Ballou & Palmer
(1974). Gas-tight 2.5 ml Hamilton syringes were
used throughout the experiments (equal mixing).
The experiments starting with the resting enzyme were performed as follows. Syringe A was
filled with 246pM-cytochrome oxidase solution in
buffer equilibrated with air, and syringe B was
filled with buffer only. An oxidized reference
sample was then collected by mixing the contents
of the two syringes in equal proportions. The
buffer in syringe B was then replaced with 2mM
aerated reduced cytochrome c. After collection of
one sample without delay the cell compartment
was attached and samples with different delay
times were collected by using the multiple-push
facility of the system. KCN was then injected in
syringe B from the concentrated stock solution to
give a final concentration of 20mM. The same
number of samples were collected with the same
delays as above.
Experiments with the pulsed enzyme were
performed by reducing the cytochrome oxidase in
syringe A (new enzyme preparation; 345 Mm) with
an equivalent amount of sodium ascorbate and
with a catalytic amount of cytochrome c present.
After a reference sample had been collected by
mixing cytochrome oxidase with the aerated
buffer, syringe B was filled with 02-bubbled

A554(red.)> 2.3, A445(red.)/A605(red.)> 5.0.

The mol-

ecular activity was 20s-1, measured essentially as

described by Smith & Conrad (1956). When the
purified enzyme was incorporated into asolectin
vesicles the activity rose to around 400s-1 (pH 6.2,
high ionic strength).
Enzyme concentrations were determined as
described in Jensen et al. (1981). Sodium dodecyl
sulphate/polyacrylamide-gel electrophoresis by the
procedure of Downer et al. (1976) gave a subunit
composition similar to that described by Wikstrom
et al. (1981) (seven major subunit bands).
Horse heart cytochrome c (type III) from Sigma
Chemical Co. was used without further purification. Stock solutions of 2mM and 2.5mM reduced
cytochrome c were prepared by titration with
sodium ascorbate until the absorption band of
cytochrome c at 695 nm was completely bleached.
The solution was then equilibrated with air or pure
02. Inspection of the 695nm band showed no signs

Rapid-freeze experiments
The rapid-freeze equipment and freezing procedures were identical with the description given in
Wilson et al. (1982) except that the optical cell was
used only as a container for the enzyme solution
after mixing when delays were used. The T-mixer

1984

Cyanide inhibition of cytochrome c oxidase

2.5mM reduced cytochrome c solution, and
samples were taken as above. The cyanide addition
was also made as with the resting enzyme. As
internal standard 100 uM-transferrin (see below)
was added to the cytochrome oxidase solution in
syringe A before the pulsed experiments.
A second type of experiment with the pulsed
enzyme was carried out by just mixing the reduced
cytochrome oxidase with aerated cyanide-containing buffer with the same concentrations and delays
as above.
Quantitative measurement of the rapid-freeze
samples
It is for many reasons inevitable that difficulties
arise when making quantitative e.p.r. measurements of rapid-freeze samples. The main problems
occur during the packing of the powdered samples,
where the sample concentration after packing
can vary. This difference is due to variations in
the packing force applied on the sample, and
its packing characteristics. From studies of
CuSO4/EDTA solutions Ballou (1971) found a
variation in packing factor of 6% with the same
person packing the samples (packing factor defined as the ratio between the signal from a slowly
frozen sample and a rapid-freeze one of the same
material). It should be noted that this study was
carried out with aqueous solutions of small ions,
which pack easily. Uniform protein solutions poses
no problems in packing, but, if chemical reactions
that change the shape of the protein can take place,
it can have large effects on packing ability. During
the course of the experiment it was noted that
samples captured during turnover were more
difficult to pack compared with the resting
oxidized cytochrome oxidase. Thus introducing
delays before the reaction mixture was freezequenched led to such packing difficulties that we
were forced to decrease the jet speed by 30% to be
able to pack the samples at all. An interesting
observation was made in that these difficulties
were less pronounced after cyanide addition,
possibly indicating an altered protein conformation after cyanide binding to the turnover enzyme.
Isopentane adhering to the outside of the e.p.r.
tubes after freeze-quenching also causes variations, as it can be difficult to wipe off without
heating the sample by the resultant friction.
Remaining isopentane can lead to increased frost
attraction on the tubes from humid air, thus
lowering the sensitivity of the e.p.r. measurement.
To overcome these difficulties a small amount of
transferrin was added to the pulsed cytochrome
oxidase as an internal standard. Transferrin is an
iron-containing protein (Aasa et al., 1963) displaying a strong e.p.r. signal at g = 4.2 and a weaker
one at g = 9, lines that are well spaced and do not
Vol. 224

831

interfere too much with the cytochrome oxidase

e.p.r. signals. However, when cyanide was added
to the cytochrome oxidase solution, the transferrin
signal at g = 4.2 overlayed the g = 3.55 cyanide
signal, which made quantification difficult. Fortunately, though, the transferrin signal could be
subtracted digitally without disturbing the rest of
the e.p.r. spectrum. The amplitude of the g = 4.2
signal of transferrin added to the samples was
measured, and this value was used as an internal
standard to correct the ensuing integrations of the
e.p.r. signals of the samples for small variations in
sample temperature, instrumental sensitivity and
small differences in packing between samples.
These integrations were made by the procedure of
Aasa & Vanngard (1975), except for the cytochrome a signal, for which instead the amplitude
of the gx = 1.45 signal was measured relative to the
oxidized control sample. The g values for the lowspin a3-HCN signal was calculated by the method
of De Vries & Albracht (1979).
Results and discussion
Rapid-freeze experiments
Figs. 1 and 2 show e.p.r. spectra of samples
obtained from experiments in which cytochrome c
oxidase, as either the resting (Fig. 1) or the pulsed
(Fig. 2) form, was mixed with its substrates, ferrocytochrome c and 2, in the presence or in the
absence of KCN, and then quenched by the rapidfreeze technique. As a comparison and for quantitative analysis each Figure shows a control
experiment in which the enzyme, again in either
the resting or the pulsed form, was quenched in the
absence of cytochrome c. These Figures also show
spectra of samples captured either directly following mixing (no delay) or 5s after mixing.
The signals seen in these spectra are typical of
the enzyme and have been reported elsewhere
(Aasa et al., 1976; Johnson et al., 1981). Thus the
g = 3 (low-spin ferric cytochrome a) and g = 2
(cupric CUA) signals are seen in the controls and, to
various extents, in the enzyme after mixing with
substrates. In the experimental samples ferricytochrome c, produced as a consequence of electron
transfer to the enzyme, makes the dominant
contribution to the g = 3 region. In addition, the
g =6 signal of high-spin ferric cytochrome a3
appears in some of the spectra of the partially
reduced enzyme.
One of the important results of the experiments
may be readily appreciated by a qualitative
inspection of Figs. 1 and 2. It is seen that for both
resting and pulsed enzyme the addition of cyanide
makes very little difference to the e.p.r. signals
displayed by samples frozen at very short times
after mixing, i.e. the general features of spectra B

832

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstr6m

g-value
6

11111 I I

1'

12

1.5

-x5

t-

t~~0
C

0.1

0.2

0.3

0.4

0.5

Magnetic flux density (T)

Fig. 1. E.p.r. spectra from rapid-freeze experiments with resting oxidized cytochrome oxidase versus reduced cytochrome c
Spectrum A: resting control sample, 123 pM-cytochrome oxidase (after mixing with Hepes buffer). Spectrum B:
cytochrome oxidase mixed with cytochrome C2+; no delay after mixing; cytochrome oxidase concentration, 123 gm;
cytochrome c concentration, 1 mm. Spectrum C: as for spectrum B but with lOmM-KCN (added in the cytochrome c
syringe). Spectrum D: as for spectrum B but with 5 s delay after mixing. Spectrum E: as for spectrum C but with 5 s
delay after mixing. All concentrations are after mixing. The arrow indicates the signal due to low-spin ferricytochrome a3-cyanide complex at g = 3.55. The magnified signals at g,, 1.45 is due to cytochrome a. E.p.r. conditions:
microwave power, 2 mW; microwave frequency, 9.222 GHz; microwave modulation, 2 mT; temperature, 15 K. The
gain was the same for all spectra.

and C of Fig. 1 are similar, as are those of spectra B

and C of Fig. 2 (see also the quantification below).
In contrast, however, the spectra of samples frozen
5s after mixing are significantly different in the
absence and in the presence of cyanide (compare
spectra D and E of each Figure). The g = 3 signal,
due to the production of ferricytochrome c by
aerobic oxidation catalysed by the enzyme, is much
smaller in the presence of cyanide. It is apparent
therefore that, during the first 5s after mixing,
cyanide inhibition is fully established. Indeed,
examinaton of the e.p.r. spectra (not shown) of
samples captured I s after mixing shows that in the
presence of cyanide the g = 3 signal due to ferricytochrome c is developed to its maximum extent
and does not increase thereafter. Thus the influ-

ence of cyanide is developed within the first second

of turnover.
A quantitative analysis of these spectra may be
obtained by integrating the signals as described in
the Materials and methods section. The results of
this quantification are shown in Table 1 and are
now discussed below.

Resting enzyme
For the resting enzyme and within the dead time
of the apparatus approx. 1.4 electron equivalents of
ferricytochrome c are produced both in the absence
and in the presence of 10mM-cyanide. This is to be
expected, as it is known that a rapid burst phase of
the reaction between ferrocytochrome c and cytochrome oxidase transfers 1-2 electron equivalents

1984

Cyanide inhibition of cytochrome c oxidase

833
g-value

12

11111

1.5

l-

('7

-/s-ViN~~~~~

x5

0.1

0.2

0.5

0.4

0.3

Magnetic flux density (T)

Fig. 2. E.p.r. spectra from rapid-freeze experiments with the pulsed cytochrome oxidase versus reduced cytochrome c plus 02
Spectrum A: pulsed control sample, reduced cytochrome oxidase mixed with 02-bubbled Hepes buffer and frozen
without delay; cytochrome oxidase concentration, 173 pM after mixing. Spectrum B: reduced cytochrome oxidase
mixed with cytochrome c2+ and 02 ; no delay after mixing; cytochrome oxidase concentration, 173 pM; cytochrome c
concentration, 1.25 mm. Spectrum C: as for spectrum B but with lOmM-KCN (added in the cytochrome c syringe).
Spectrum D: as for spectrum B but with 5s delay after mixing. Spectrum E: as for spectrum C but with 5 s delay after
mixing. Spectrum F: transferrin, 100pM (internal standard); the signal shown was digitally subtracted from spectra
A-E. Conditions were as described for Fig. 1. Freeze-dried transferrin was added to the cytochrome oxidasecontaining syringe just before the start of the experiment.

Table 1. Electron balance sheet for the reaction between ferrocytochrome c and cytochrome c oxidase
The cytochrome C3+ values have been corrected for remaining cytochrome a3+, i.e. cytochrome C3+ = total g = 3
signal -oxidized cytochrome a. * indicates that the electron distribution was unbalanced owing to transfer to 02.
- indicates that the signal was too small to integrate.
Electron equivalents
With cyanide

No cyanide
,

Delay
(s)

Resting
Pulsed

No
5
No
5

Vol. 224

a2+
c3+
1.4 > 0.9
4.5
0.7
1.85
0.6
0.7
4.8

~~A

Total
(e- in
CUB+
CUA+ (g= 6) oxidase)
0.75 0.2
1.85
0.7
0.45
0.4
0.45

0.05
0.13

*
1.1

Total

(e- in
CUB+ (g= 3.55) oxidase)
a33+-HCN

C3+
1.4
2.4
0.65
2.05

a2+
>0.9
>0.9
0.6
0.80

CUA+
0.7
0.75
0.7
0.65

0.15

0.05
0.2

1.8
1.85
1.3

0.3

1.75

834

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstrom

from the donor in the first few milliseconds of the

reaction and that this burst is insensitive to the
presence of cyanide (Wilson et al., 1975). The
electrons from cytochrome c are distributed among
the sites in the oxidase as indicated in Table 1. In
each case cytochrome a is almost completely
reduced and CUA is approximately 70% reduced. A
small axial g = 6 signal arises within this time, this
presumably being due to reduction of a portion of
the CUB sites and thus the appearance of uncoupled
ferricytochrome a3. In the presence of cyanide a
part of this latter signal is converted into g = 3.55,
a low-spin ferric haem-cyanide signal. The total
number of electron equivalents within the oxidase
is approx. 1.8, somewhat higher than the expected
value of 1.4 but probably equal to this value within
the error of all the integrations involved.
After a 5s delay before quenching, the intensities of the e.p.r. signals are now quite different in
the absence and in the presence of cyanide. In the
former case the enzyme has clearly 'turned over'
and catalysed the oxidation of cytochrome c by 02.
Thus the ferricytochrome c concentration is
approx. 4.5 electron equivalents and approaches
the maximum expected for this experimental
regime. Under these conditions (1 mM-cytochrome
c and 123 /M-oxidase) this maximum should be
achieved when the 02 is exhausted; assuming
approximately 200 kM-02, we should see at maximum approx. 6 haem equivalents of cytochrome c
oxidized (i.e. 200 x 4 123), and less if the 02 concentration is lower than 200 pm in the strong
protein solutions used before mixing. The electrons
cannot be accounted for by reduction of the sites
within the enzyme and have presumably been
passed on to 02. In the presence of cyanide, however, only 2.4 equivalents of ferricytochrome c are
seen: 1.9 of these are accounted for by the enzyme
signals, whereas the remainder presumably have
been passed to 02 before complete inhibition
ensued. The g = 6 signal has now vanished, to be
entirely replaced by the cyanide-bound signal at
g = 3.55, which is equivalent to 0.2 haem equivalent. This latter value is particularly interesting as
it indicates that, although the enzyme is apparently
fully inhibited, only 20% of cytochrome a3 is in the
low-spin ferric haem-cyanide complex and thus it
cannot be this species that is of primary importance for the inhibition. Whatever the inhibited
form is, it is seemingly e.p.r.-silent. A good
candidate for this form is thus one in which both
CUB and cytochrome a3 remain oxidized and
coupled, possibly by cyanide acting as bridging
ligand, as suggested by Hill et al. (1983).
Our results on the resting enzyme are thus
consistent with the following interpretations. On
mixing the oxidized enzyme with ferrocytochrome
c in the presence of cyanide (which does not bind to

the oxidized enzyme on this time scale) approx. 2

electron equivalents enter 80% of the enzyme
molecules. On achieving this state of partial
reduction a transition occurs within the enzyme
that allows cyanide to approach and bind to the
binuclear cytochrome a3-CuB site. In so doing,
cyanide strongly stabilizes the oxidized forms of
the metals to which it is tightly bound, thus
stopping reduction of these sites and hence
catalysis. In some 20% of the molecules, however,
electrons may transfer also on to CUB, and the
uncoupled ferricytochrome a3 now becomes e.p.r.visible and binds cyanide, again leading to
inhibition. This interpretation thus supports the
earlier kinetic studies which suggest that it is
partially-reduced turnover intermediates that
rapidly bind cyanide and also highlights the
known heterogeneity of the resting enzyme.

Pulsed enzyme
The pulsed enzyme behaved in an essentially
similar manner to the resting enzyme. In the
absence of cyanide approx. 2 electron equivalents
are delivered to the enzyme in the burst phase, as
seen by the appearance of the ferricytochrome c
e.p.r. signal. Half of these electron equivalents can
be seen in the enzyme, the remainder we assume
having passed into 02. This situation is slightly
different from that of the resting enzyme, probably
reflecting the enhanced activity of the enzyme that
the pulsed form is known to possess (Antonini et
al., 1977). After 5s the enzyme has catalysed the
oxidation of the remaining cytochrome c.
In the presence of cyanide the situation is a little
more complex. After 5s delay the pattern seen for
the resting enzyme is repeated. Two electron
equivalents have passed into the enzyme, which is
now fully inhibited by cyanide, although only a
fraction of the enzyme displays the g = 3.55 signal.
With no delay, however, the picture is somewhat
different. A significantly smaller amount of ferricytochrome c is formed (approx. 0.7 electron
equivalent) while almost twice the electron equivalents are observed in the oxidase. We rationalize
this situation as follows. Immediately after the
mixing the enzyme for the pulsed experiments is
still fully reduced in the presence of 02 02 will
now bind and accept 2 electron equivalents with
a rate in the order of 3 x 104 s-. The sites in the
enzyme donating these have until recently been
considered to be cytochrome a3 and CUB; however,
Hill & Greenwood (1984) have shown that in a
proportion of the molecules (40%) cytochromes a
and a3 may be the donors. In either case the
enzyme will now exist with cytochrome a3 and
possibly CUB oxidized. Further electron transfer
from CUA and cytochrome a to the 02-binding site
is slower, with a rate of approx. 5 x 103s- , and
1984

Cyanide inhibition of cytochrome c oxidase

Table 2. Electron equivalents remaining in cytochrome a
and CUA
Reduced cytochrome oxidase (246pM) was mixed
with aerated 5OmM-Hepes buffer, pH7.4, at room
temperature, and freeze-quenched in cold isopentane. The sample without cyanide (OmM) was used
as the reference and the e.p.r. signals from the
completely re-oxidized cytochrome a and CUA in
this sample were measured (see the Materials and
methods section). The e.p.r. signals obtained after
mixing the cytochrome oxidase solution with cyanide-containing aerated buffer (10mM- and 30mMcyanide) were smaller than in the reference, and this
difference was taken as a measure of the number of
electron equivalents remaining in those samples.
Electron equivalents
Concn. of
CUA+
a2+
CN- (mM)
0

10

30

0.2
0.4

0.15
0.4

which is therefore comparable with the rate of

cyanide binding. At 10mM-cyanide and taking a
rate constant of approx. 106 M-1 s1 for the secondorder combination rate constant of cyanide to the
partially reduced enzyme (Jones et al., 1984), the
cyanide 'on' rate is approx. 103-104 s-1. Thus there
may be competition between cyanide binding to
and reduction of the oxidized sites in cytochrome
oxidase. If this is so, then we may expect that in a
fraction of the molecules cyanide will bind to the
enzyme and stop electron transfer from cytochrome a and CUA. The electrons that we observe
on cytochrome a and CUA (Table 1, row 3: pulsed,
no delay, plus cyanide) are therefore only partially
derived from cytochrome c and are in fact electrons
that have remained in the enzyme from the
original reduction before the mixing. This also
explains why we see more electron equivalents in
the enzyme than we see cytochrome c oxidized.
Support for this explanation comes from control
rapid-freeze experiments in which the fully reduced enzyme was mixed with oxygenated buffer
either in the absence or in the presence of cyanide.
Table 2 shows that in the absence of cyanide all the
sites are oxidized within the dead time of the
apparatus. In the presence of cyanide, however,
the oxidation of cytochrome a and CUA is partially
blocked, and this effect is exaggerated at higher
cyanide concentrations. Such observations are
consistent with the kinetic competition for sites as
discussed above.
Reduction ofcytochrome c oxidase by dithionite in the
presence and in the absence of cyanide
Parallel experiments were carried out monitoring the reduction of cytochrome oxidase by

Vol. 224

835
g-value
12 8 6 5

,I

1111__ I II

D-v110~~~~~~~
0.1

0.2

0.3

Magnetic flux density (T)

Fig. 3. E.p.r. spectra of cytochrome oxidase after freezethaw experiments with cyanide present
Spectrum A: cytochrome oxidase, 220/AM, frozen
lOs after the addition of Na2S204 (6.5mM) by
immersing the sample in cold (140K) isopentane
('fast freeze'). Spectrum B: as for spectrum A but
with KCN (0.5 mM) added together with the
dithionite. Spectrum C: the sample used for spectrum B (above) thawed to room temperature for
1 min and re-frozen ('fast freeze'). Spectrum D: the
sample used for spectrum C after an additional
10min thawing. Conditions were as described for
Fig. 1.

dithionite either by optical or e.p.r. spectroscopic

methods.
At a wavelength of 445nm and on addition of
dithionite (25 mM) a rapid reduction of a portion of
the enzyme (cytochrome a) was followed by slow
reduction of cytochrome a3 (Jones et al., 1983) with
a t4 of approx. 150s. In the presence of 300pMcyanide the rapid phase of reduction remained
unchanged, but the rate of the slower phase was
greatly lowered, ti approx. 1500s. These results are
in agreement with those reported by Jones et al.
(1984) and show that cyanide binds rapidly to the
partially reduced enzyme, stopping its full reduction. The rate of reduction of cytochrome a3 is now
limited by the cyanide 'of rate.
Fig. 3 reports the complementary e.p.r. spectra.

836

P. Jensen, M. T. Wilson, R. Aasa and B. G. Malmstrom

Spectrum A is that obtained from an anaerobic

(N2-equilibrated) sample of resting oxidase into
which 6.5mM-Na2S204 was mixed and the same
frozen after 10s. By this time the optical spectrum
indicates that cytochrome a is largely reduced
whereas cytochrome a3 is oxidized. The e.p.r.
spectrum (A) is in agreement with the optical
measurements, the g = 3 signal (cytochrome a) is
equivalent to only approx. 0.1 haem equivalent
and the CUA is similarly small. The g= 6 signal
(ferric cytochrome a3) comprises about 0.2 haem
equivalent and the remainder of the cytochrome a3
is presumably still oxidized and coupled to CUB.
Spectrum B shows the same experiment carried out
in the presence of 0.5 mM-KCN. Although optical
(see above) and activity (Jones et al., 1984)
measurements suggest that the enzyme is now fully
complexed with cyanide, we observe a g = 3.55
signal (cytochrome a33+-HCN) comprising only
about 0.15 haem equivalent, and the g = 6 signal
has decreased by the same amount. On waiting for
a further period (spectrum C), no additional signals
appear, but the remaining g = 6 signal is converted
into g = 3.55. The elapse of further time leads to no
changes except that the character of the g = 3.55
signal is altered and the g = 12 signal disappears.
Although we are at present uncertain of the
significance of these changes, the observations are
in agreement with those made by Johnson et al.
(1981) and Brudvig et al. (1981), and we do not
believe they alter our overall conclusions. The
results of the optical and e.p.r. studies on dithionite
reduction are therefore in full agreement with the
rapid-freeze investigation with the native reductant, cytochrome c, namely that cyanide binding
and inhibition gives an e.p.r.-silent species possibly involving binding to coupled ferricytochrome
a3 and cupric CUB. The heterogeneity observed in
the rapid-freeze experiments is again evident in
Fig. 3, with about 20% of the enzyme rapidly
accepting electrons into CUB, rendering cytochrome a3 e.p.r.-visible. This latter then binds
cyanide to yield the g = 3.55 signal.

Concluding remarks
The behaviour of cytochrome a3 (classically
defined as the binding site of the oxidase) towards
ligands is clearly strongly dependent not only on its
valence state but also on the valence state of other
metal centres and on the conformational status of
the enzyme. When the fully reduced enzyme reacts
with ligands, e.g. CO, CN, it does so in reactions
that are second-order and whose equilibrium
association constants measured by static titrations
are in accord with those calculated from determinations of the 'on' and 'off' rates. In other words
the ligand-binding site is accessible as it is in open

and rapid communication with the bulk-phase

solution. The same is true also for 02, which binds
exceedingly rapidly (k.. 108 M-1 * s- I) to the cytochrome a3-CuB binuclear centre.
Ligand binding to the fully oxidized enzyme is,
in contrast, complex. For example, although
cyanide binds very tightly to cytochrome a3
(KD 10-6 M), it does so very slowly and with a rate
independent of cyanide concentration and with
a high temperature coefficient. Such behaviour
suggests that the binding site is shielded from
approaching ligands and that complex-formation
is rate-limited by transitions in the protein,
possibly conformational, as suggested by the high
enthalpy of activation (Van Buuren et al., 1972),
which allow ligands from the bulk phase to
approach the binding site.
The transition from a 'closed' to an 'open' site is
particularly important to the consideration of
cyanide inhibition. It has been shown (Jones et al.,
1984; Van Buuren, 1972) that it is binding of
cyanide to partially reduced forms of the enzyme,
populated during turnover, that is responsible for
the powerful inhibiting effects of cyanide. In these
forms of the enzyme it appears that cytochrome a3
remains oxidized yet is accessible to cyanide,
as indicated by the rapid onset of inhibition
(Antonini et al., 1971). In a recent study (Jones et
al., 1984) a model was proposed that incorporated
this transition from a 'closed' to an 'open' form on
partial reduction. However, it was impossible to
ascertain from the kinetic experiments the degree
of reduction necessary to initiate this transition.
The present study suggests that the entry of 2
electron equivalents into the enzyme from cytochrome c is sufficient to switch the cytochrome a3
site to the accessible form. These electrons reside
on cytochrome a and CUA, and thus it appears that
reduction of the electron-accepting site is communicated to the ligand-binding site, which is now
able to bind cyanide rapidly. The e.p.r. measurements also show that in the cyanide-bound form
(apart from in a small fraction of the enzyme)
cytochrome a33+ is still coupled to CUB2+, and it is
likely that the ligand is stabilizing both metals in
their oxidized states by bridging between them, as
suggested by Hill et al. (1983).
Although entry of 2 electron equivalents is
sufficient to trigger the 'opening' of the ligandbinding site, we cannot be certain that both are
necessary. The time-resolution of our method does
not allow us to discount the possibility that
reduction of one of the sites, cytochrome a or CuA,
is sufficient.
In contrast, a common view has been that an
input of 3 electron equivalents is necessary for
cyanide binding (see, e.g., Nicholls, 1983). This
divergence can, however, be explained by consid1984

Cyanide inhibition of cytochrome c oxidase

ering the different proportions of enzyme forms, by
using the criteria of Brudvig et al. (1981), in
oxidase prepared by the methods of Yonetani and
Van Buuren (Wilson et al., 1982) respectively.
These differences are also reflected in the size of
the g = 3.55 e.p.r. signal from the a33+-HCN
compex, which amounts to the equivalent of 1
haem with a Yonetani preparation (Johnson et al.,
1981) whereas with an enzyme preparation by
the method of Van Buuren only about 20% of
the molecules are in the form giving a g = 3.55
signal.
The nature of the transition leading to an open
configuration may involve gross conformational
changes [as suggested by the occluded nature of the
resting site (see above) and by the change in the
packing characteristics in the rapid-freeze experiments (see the Materials and methods section)], a
displacement of an intrinsic ligand from the sixth
co-ordination site of the haem of cytochrome a3, or
most probably both of these changes. For example
fully oxidized pulsed enzyme, though being much
more accessible to cyanide than the resting form
(Brittain & Greenwood, 1976; Jones et al., 1984), is
certainly not as open to attack by cyanide as are
partially reduced turnover forms.
The 'opening' and 'closing' of the site to ligands
presumably has a functional significance for the
enzyme in vivo. An obvious possibility is that it
constitutes a device whereby 02 may only approach the binding site when two metal centres are
reduced. Thus, once electrons have passed to both
CUB and cytochrome a3, the 2 may bind and the
first 2-electron reduction to form tightly bound
peroxide may proceed. A second possibility is that
such a conformational transition constitutes part
ofthe energy transduction system, e.g. the mechanical part of a proton pump.
This investigation was supported by the Swedish
Natural Science Research Council. A British Council
Travel Grant is also acknowledged (M. T. W.). M. T. W.
also acknowledges Science and Engineering Research
Council Support. We thank Professor Tore VanngArd for
valuable suggestions and discussions, and Mr. Per-Eric
Thornstr6m, Mrs. Kajsu Valimaa and Mrs. Elsy Lagerdahl for skilful technical assistance during the enzyme
preparations and the ensuing characterizations.

References
Aasa, R. & Vainngard, T. (1975) J. Magn. Reson. 19, 308315

Vol. 224

837
Aasa, R., Malmstrom, B. G., Saltman, P. & VanngArd,
T. (1963) Biochim. Biophys. Acta 75, 203-222
Aasa, R., Albracht, S. P. J., Falk, K.-E., Lanne, B. &
Vanngard, T. (1976) Biochim. Biophys. Acta 422, 260272
Antonini, E., Brunori, M., Greenwood, C., Malmstrom,
B. G. & Rotilio, G. C. (1971) Eur. J. Biochem. 23, 396400
Antonini, E., Brunori, M., Colosimo, A., Greenwood, C.
& Wilson, M. T. (1977) Proc. Natl. Acad. Sci. U.S.A.
74, 3128-3132
Ballou, D. P. (1971) Ph.D. Thesis, University of
Michigan
Ballou, D. P. & Palmer, G. (1974) Anal. Chem. 46, 12481253
Brittain, T. & Greenwood, C. (1976) Biochem. J. 155,
453-455
Brudvig, G. W., Stevens, T. H., Morse, R. H. & Chan,
S. I. (1981) Biochemistry 20, 3912-3921
Brunori, M., Antonini, E. & Wilson, M. T. (1981) in
Metal Ions in Biological Systems (Segel, H., ed.), vol.
13, pp. 187-228, Marcel Dekker, New York
De Vries, S. & Albracht, S. P. J. (1979) Biochim. Biophys.
Acta 546, 334-340
Downer, N. W., Robinson, N. C. & Capaldi, R. A.
(1976) Biochemistry 15, 2930-2936
Hill, B. & Greenwood, C. (1984) FEBS Lett. 166, 362366
Hill, B., Brittain, T., Eglinton, D. G., Gadsby, P. M. A.,
Greenwood, C., Nicholls, P., Peterson, J., Thomson,
A. J. & Woon, T. C. (1983) Biochem. J. 215, 57-66
Jensen, P., Aasa, R. & Malmstr6m, B. G. (1981) FEBS
Lett. 125, 161-164
Johnson, M. K., Eglinton, D. G., Gooding, P. E.,
Greenwood, C. & Thomson, A. J. (1981) Biochem. J.
193, 699-708
Jones, G. D., Jones, M. G., Wilson, M. T., Brunori, M.,
Colosimo, A. & Sarti, P. (1983) Biochem. J. 209, 175182
Jones, M. G., Bickar, D., Wilson, M. T., Brunori, M.,
Colosimo, A. & Sarti, P. (1984) Biochem. J. 220, 57-66
Nicholls, P. (1983) Trends Biochem. Sci. 8, 353-354
Sekuzu, I., Takemori, S., Yonetani, T. & Okunuki, K.
(1959) J. Biochem. (Tokyo) 46, 43
Smith, L. & Conrad, H. (1956) Arch. Biochem. Biophys.
63, 403-413
Van Buuren, K. J. H. (1972) Ph.D. Thesis, University of
Amsterdam
Van Buuren, K. J. H., Nicholls, P. & Van Gelder, B.
(1972) Biochim. Biophys. Acta 256, 258-276
Wikstrom, M., Krab, K. & Saraste, M. (1981) Cytochrome Oxidase: A Synthesis, pp. 14-54, Academic
Press, New York
Wilson, M. T., Greenwood, C., Brunori, M. & Antonini,
E. (1975) Biochem. J. 147, 145-153
Wilson, M. T., Jensen, P., Aasa, R., Malmstr6m, B. G. &
Vanngard, T. (1982) Biochem. J. 203, 483-492