Article

HIV Controllers Exhibit Effective CD8+ T Cell

Recognition of HIV-1-Infected Non-activated CD4+ T
Cells
Graphical Abstract Authors
Blandine Monel, Annmarie McKeon,
Pedro Lamothe-Molina, ..., R. Brad Jones,
Sylvie Le Gall, Bruce D. Walker

Correspondence
[email protected]

In Brief
The cure for HIV is not achievable due to
HIV reservoirs, mostly in resting CD4+
T cells. Monel et al. show that CD8+ T cells
from HIV controllers are able to establish
immunological synapses with HIV+
resting CD4+ T cells, leading to IFN-g,
MIP1-b production, degranulation, and
the elimination of the target cells.

Highlights
d Non-activated CD4+ T cells are permissive for HIV entry but
not productively infected

d CD8+ T cells from HIV controllers recognize HIV+ non-

activated CD4+ T cells

d Antigens from incoming viral particles are presented to CD8+

T cells through HLA-I

d CD8+ T cell recognition leads to the death of the HIV+ non-

activated CD4+ T cells

Monel et al., 2019, Cell Reports 27, 142–153

April 2, 2019 ª 2019 The Authors.
https://doi.org/10.1016/j.celrep.2019.03.016
 Cell Reports

Article

HIV Controllers Exhibit Effective CD8+

T Cell Recognition of HIV-1-Infected
Non-activated CD4+ T Cells
Blandine Monel,1,2 Annmarie McKeon,1 Pedro Lamothe-Molina,1,2 Priya Jani,1 Julie Boucau,1 Yovana Pacheco,1,3
R. Brad Jones,1,4 Sylvie Le Gall,1 and Bruce D. Walker1,2,5,6,*
1Ragon Institute of Massachusetts General Hospital, Massachusetts Institute of Technology, and Harvard University, Cambridge, MA 02139, USA
2Howard Hughes Medical Institute, Chevy Chase, MD 20815, USA
3Center for Autoimmune Diseases Research (CREA), School of Medicine and Health Sciences, Universidad del Rosario, Bogotá, Colombia
4Division of Infectious Diseases, Weill Cornell Medicine, New York, NY 10065, USA
5Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
6Lead Contact

*Correspondence: [email protected]
https://doi.org/10.1016/j.celrep.2019.03.016

SUMMARY to abortive infection. The resulting cDNA fragments are sensed

by interferon-g (IFN-g)-inducible protein (IFI-16) and induce
Even with sustained antiretroviral therapy, resting pyroptosis, leading to the production of pro-inflammatory cyto-
CD4+ T cells remain a persistent reservoir of HIV infec- kines and subsequent CD4+ T cell depletion (Doitsh et al.,
tion, representing a critical barrier to curing HIV. Here, 2010, 2014; Monroe et al., 2014).
we demonstrate that CD8+ T cells recognize infected, Although HIV infection of resting CD4+ T cells is mostly abor-
non-activated CD4+ T cells in the absence of de novo tive (Doitsh et al., 2010; Tilton et al., 2014), reverse transcription
can occasionally be completed and the viral cDNA imported
protein production, as measured by immune synapse
into the nucleus, resulting in either pre- or post-integration la-
formation, degranulation, cytokine production, and
tency (Chavez et al., 2015; Pan et al., 2013; Zhou et al.,
killing of infected cells. Immune recognition is induced 2005) without an intermediate phase of productive infection
by HLA-I presentation of peptides derived from (Chavez et al., 2015; Vatakis et al., 2009). Pre-integration la-
incoming viral particles, and recognition occurred tency occurs when viral cDNA is blocked in the nucleus of a
either following cell-free virus infection or following resting cell without being able to integrate into the host chro-
cell-to-cell spread. CD8+ T cells from HIV controllers mosome (Petitjean et al., 2007; Pierson et al., 2002; Sloan
mediate more effective immune recognition than and Wainberg, 2011); subsequent integration can occur after
CD8+ T cells from progressors. These results indicate cellular activation, leading to productive infection (Thierry
that non-activated HIV-infected CD4+ T cells can be et al., 2015). Post-integration latency is the most stable form
targeted by CD8+ T cells directly after HIV entry, of latency in which viral cDNA is efficiently integrated but leads
to no or few viral transcripts (Dahabieh et al., 2015; Hakre et al.,
before reverse transcription, and thus before the
2012; Stevenson, 1997). These cells contain a viral genome that
establishment of latency, and suggest a mechanism
can be reactivated by different factors, leading to the produc-
whereby the immune response may reduce the size tion of new infectious particles (Bruner et al., 2015; Chun
of the HIV reservoir. et al., 1997, 1998; Davey et al., 1999; Ho et al., 2013; Laird
et al., 2015; Wong et al., 1997). Both forms of latency
INTRODUCTION contribute to a reservoir of resting infected cells that are pre-
sumed to be invisible to HIV-specific CD8+ T cell responses.
Despite the ability of combination antiretroviral therapy (cART) The persistence of this HIV reservoir is a major obstacle to cur-
to reduce plasma viremia to undetectable levels, treatment rent cure efforts (Archin and Margolis, 2014; Katlama et al.,
does not lead to viral eradication in HIV-infected persons, 2013; Massanella et al., 2013; Pitman et al., 2018; Siliciano
committing them to lifelong therapy. The inability to eradicate and Siliciano, 2013).
HIV is predominantly due to a reservoir of resting CD4+ T cells Killing HIV-infected resting CD4+ T cells early after viral entry
non-productively infected by HIV (Bukrinsky et al., 1991; Chun before the reverse transcription step would abrogate both abor-
et al., 1997; Finzi et al., 1997; Massanella and Richman, 2016; tive and latent infection and would thus help to decrease CD4+
Siliciano et al., 2003; Soriano-Sarabia et al., 2014). Resting T cell depletion and inflammation, and could affect the size of
CD4+ T cells are permissive for HIV entry (Tilton et al., 2014), the latent viral reservoir. It has been shown that HIV Gag and
but once inside, cytoplasmic host restriction factors such as Pol-specific CD8+ T cell lines recognize peptides from incoming
SAMHD1 can impede the reverse transcription of viral RNA particles after HIV entry into activated CD4+ T cells (Payne et al.,
into cDNA (Baldauf et al., 2012; Descours et al., 2012), leading 2010; Kløverpris et al., 2013), and at least one study indicates

142 Cell Reports 27, 142–153, April 2, 2019 ª 2019 The Authors.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
that resting cells may also be targeted (Buckheit et al., 2013), production was detected by means of HIV long terminal repeat
although the role of antigen processing, restricting human (LTR)-driven GFP expression (Cavrois et al., 2002; Tilton et al.,
leukocyte antigen (HLA) alleles, and synapse formation in the 2014).
observed elimination of infected cells was not evaluated. We Using this system, we assessed viral entry and levels of produc-
tested the hypothesis that HIV-specific CD8+ T cells from HIV tive infection, comparing activated to non-activated CD4+ T cells
controllers, a very small proportion of the HIV-infected popula- from healthy donors. The activation status of live CD3+CD4+
tion who manage to spontaneously control viral replication T cells in whole peripheral blood mononuclear cells (PBMCs)
and maintain stable CD4+ T cell counts without the need for an- was assessed ex vivo by flow cytometry by analyzing the expres-
tiretroviral therapy (Carrington and Walker, 2012; Deeks and sion of CD25 and CD69, inducible cell surface glycoproteins
Walker, 2007; Walker and Yu, 2013), could recognize and kill acquired during lymphocyte activation. In the absence of exoge-
non-activated, infected CD4+ T cells due to the recognition of nous stimulation, CD4+ T cells within the PBMCs were quiescent,
processed viral proteins following viral entry, without requiring but were readily activated by incubation with CD3/CD28 beads
productive infection. We used a combination reporter virus sys- for 2 days. A representative experiment is shown in Figure S1A.
tem that allowed us to sensitively and specifically track the ki- Of note, the activation status was similar when CD4+ T cells
netics of infection of resting CD4+ T cells after viral entry and were first isolated from PBMCs (data not shown).
before any de novo viral protein production. We show that Two hours following infection, activated and non-activated
CD8+ T cells from HIV controllers readily establish functional CD4+ T cells were assessed for viral entry, as evidenced by
synapses with non-activated infected CD4+ T cells, leading to b-lactamase-mediated cleavage and fluorescence of the cyto-
HLA class I-restricted degranulation, cytokine production, and plasmic substrate. Non-activated (CD25 , CD69 ) CD4+ T cells
target cell death, and does not require reverse transcription, were highly permissive to entry by X4-tropic HIV (Figure 1A),
indicating that de novo viral protein production is not needed. with viral entry detected in 65% ± 11% of resting CD4+ T cells
Moreover, we show that cell-cell transmission also sensitized at the multiplicity of infection used (Figure 1B, top). The entry
cells to HIV-specific CD8+ T cell recognition, before viral reverse of R5 tropic viruses was also detected, but to a lesser extent
transcription occurs. This response is significantly more potent (5% ± 1% of resting CD4+ T cells), consistent with lower C-C
in HIV controllers than in progressors, suggesting a mechanism chemokine receptor type 5 (CCR5) expression on the resting
whereby the immune response may influence the size of the HIV CD4+ T cells (Figures 1B, bottom, and S1B). Similar levels
reservoir. of infection were observed when non-activated CD4+ T cells
were first isolated from PBMCs (data not shown). To be certain
RESULTS that the cleaved substrate corresponded to viral entry, a virus
missing the envelope (HIV DEnv) and a fusion-defective virus
HIV Infection of Primary Non-activated CD4+ T Cells (HIV X4 Env-F522Y) were used as controls (Figure S2). Quan-
Direct HIV infection of non-activated CD4+ T cells leads predom- tification of GFP expression in CD4+ T cells 2 days later re-
inantly to abortive infection and to a lesser extent, latent infec- vealed that most of the non-activated HIV-exposed CD4+
tion, which renders cells largely invisible to HIV-specific CD8+ T cells remained non-productively infected, contrary to activated
T cells (Pan et al., 2013; Tilton et al., 2014). Since incoming CD4+ T cells (Figure 1C). These results are consistent with previ-
virions can sensitize activated CD4+ T cells for recognition by ous reports (Haqqani et al., 2015; Tilton et al., 2014) and further
CD8+ T cells (Buseyne et al., 2001; Kløverpris et al., 2013; Payne suggest that most of the directly infected non-activated CD4+
et al., 2010), we first sought to confirm whether resting CD4+ T cells remain non-productively infected during the period
T cells would likewise be permissive for HIV entry, as previously observed.
shown (Tilton et al., 2014), and to determine whether these cells
could be recognized by CD8+ T cells pre-integration and thus Recognition of HIV+ Non-activated CD4+ T Cells by HIV-
before possible abortive infection or establishment of latent Specific CD8+ T Cells
infection. We next tested whether an HIV Gag-specific CD8+ T cell line
To assess the ability of non-activated CD4+ T cells to become derived by stimulating PBMCs from an HIV controller with a
infected with HIV, we used a combination reporter virus system Gag peptide pool could recognize these non-productively in-
that allowed for discrimination between viral entry into the fected CD4+ T cells. We first examined the formation of func-
cytoplasm and subsequent de novo virion production in the tional immunological synapses by fluorescence microscopy. A
infected cell (Tilton et al., 2014). Resting CD4+ T cells were in- cytolytic immunological synapse requires recognition of the viral
fected with HIV containing b-lactamase fused to HIV Vpr (Vpr- peptide:major histocompatibility complex (MHC) by the T cell
blam). Viral entry was detected by pre-labeling cells with a receptor (TCR) and polarization of adhesion molecules including
fluorescence resonance energy transfer (FRET) cytoplasmic LFA-1 and actin toward the point of contact to form a ring with
substrate (coumarin cephalosporin fluorescein, a fluorescent central clearance, through which cytolytic granules are released
beta-lactamase substrate [CCF2-AM]) composed of a hydroxy- (Ritter et al., 2015; Stinchcombe and Griffiths, 2003). To deter-
coumarin donor conjugated to a fluorescein acceptor via a mine whether immune synapse formation was occurring, HIV-
b-lactam ring. Cleavage of the b-lactam ring is mediated via infected non-activated CD4+ T cells (defined by the expression
the b-lactamase protein carried by the incoming virus, inducing of the b-lactamase cleaved substrate) were co-cultured for
an emission shift that allows for the colorimetric detection of 30 min with an autologous HIV Gag-specific CD8+ T cell line
viral entry into the cell by flow cytometry. De novo HIV protein 3 h post-HIV entry, a time point shown above to lack viral protein

Cell Reports 27, 142–153, April 2, 2019 143

A Figure 1. HIV Infection in Primary Non-acti-
vated CD4+ T Cells
(A) Non-activated CD4+ T cells were infected for
2 h with NL4.3 X4 (HIV X4, blue) or R5 virus (HIV R5,
yellow) carrying the fusion protein Vpr-b-lacta-
mase and an internal ribosome entry site (IRES)-
GFP cassette. HIV entry was determined 2 h later
by incubation with an FRET-b-lactamase sub-
strate, where cleaved substrate+ cells represent
HIV+ cells. A representative experiment is shown.
(B) Composite infection data as in (A) for non-acti-
vated (top) and activated (bottom) CD4+ T cells,
expressed as means ± SDs for three independent
experiments from three different individuals, each
B C performed in duplicate.
(C) The percentage of cells expressing GFP, rep-
resenting productive infection, was measured 48 h
post-infection in non-activated CD4+ T cells (top)
and activated CD4+ T cells (bottom). The results
are expressed as means ± SDs for three inde-
pendent experiments from three different in-
dividuals, each performed in duplicate. Statistical
significance was calculated using an unpaired
t test relative to the non-infected (NI) condition.
*p < 0.05, **p < 0.01, ***p < 0.001, and ****p <
0.0001.
See also Figures S1 and S2.

We next evaluated effector functions

elicited by the CD8+ T cells upon the
recognition of non-productively HIV-
infected CD4+ T cells using flow cytome-
try. Non-activated PBMCs from an HIV
production. Ten contacts between CD4+ T cells and CD8+ controller were infected with HIV (X4 or R5 Vpr-blam+) for 2 h,
T cells from three independent experiments were analyzed washed to remove free virus, and then incubated at 37 C for
by confocal microscopy and are represented with 3D recon- 5 h in the presence of anti-CD107a antibody. HIV entry was
structions (Figure 2A), showing the presence of perforin (upper documented with a parallel aliquot of cells using the Vpr-blam
images) or the degranulation marker CD107a (lower images) at assay and flow cytometry gating on the CD4+ T cells. At the
the point of contact. Quantitative image analysis revealed higher end of the 5-h incubation, the CD8+ T cell response in whole
perforin (Figure 2B) and CD107a (Figure 2C) expression at PBMCs was determined by flow cytometry by measuring
points of contact with HIV+ CD4+ T cells (average 66% ± 3% CD107a expression on live CD3+CD8+ T cells. Non-activated
and 80% ± 6% of points of contact were positive for perforin CD4+ T cells infected with either X4 or R5 virus using 500 ng of
or granzyme, respectively) compared to non-infected CD4+ viral p24 based on dose-response experiments (Figure S3) eli-
T cells (13% ± 3% and 17% ± 3%, respectively). The observed cited significant CD8+ T cell degranulation compared to the
background likely represents random localization of the cyto- non-infected condition (p < 0.001 and p < 0.05, respectively)
lytic granules near the point of contact. The presence of immune (Figures 2G and 2H), indicating that the immunological synapses
synapses was supported by the observation of an actin ring led to the functional engagement of CD8+ T cells. Since R5 viral
(blinded counting) for 55% ± 5% of the contacts between entry rates were relatively low (approximately 5% of non-acti-
HIV+ CD4+ T cells and the autologous CD8+ T cells compared vated CD4+ cells compared to 60% with X4 virus; Figure 1B),
to 0% for non-infected CD4+ T cells (Figures 2D and 2E), and these results suggest that only a few HIV+ CD4+ T cells are
higher actin polarization toward the point of contact (Figure 2F) required to induce a CD8+ T cell response in this assay, irrespec-
automatically determined by the ratio of actin fluorescence tive of co-receptor usage, since we observed only a 2-fold higher
signal at the point of contact versus the rest of the cell with CD8+ T cell response with X4 than R5 HIV infection despite a
Imaris software. Video S1 shows the rotation of a 3D synapse 12-fold higher HIV entry.
indicating the polarization and the actin ring at the synapse. We further evaluated intracellular expressions of IFN-g, tumor
These results show that an HIV-specific CD8+ T cell line can necrosis factor a (TNF-a), and perforin in CD8+ T cells and
establish immune synapses with autologous HIV+ non-activated observed similar polyfunctionality profiles in CD8+ T cells re-
CD4+ T cells following viral entry, in the absence of de novo viral sponding to either Gag peptide pool stimulation or to infected,
protein production. non-activated CD4+ T cells (Figure S4). We conclude that

144 Cell Reports 27, 142–153, April 2, 2019

 Figure 2. Recognition of HIV+ Non-activated
CD4+ T Cells by Autologous HIV-Specific
CD8+ T Cells
Two hours after a Vpr-blam infection assay, the
primary non-activated HIV+ CD4+ T cells (blue) from
an elite controller (EC) were co-cultured for 30 min
with an autologous Gag-specific CD8+ T cell line
(green), fixed, and stained for actin with phalloidin
(orange) and perforin (red) or CD107a (red), and cell
contacts analyzed by confocal microscopy. A
compilation of z stacks was acquired for each cell
contact and analyzed with IMARIS software. NI,
non-infected; HIV, HIV infected.
(A) 3D reconstructions of two representative ex-
periments with non-infected cells (left) or infected
cells (right) are shown with perforin staining (top) or
CD107a staining (bottom).
(B and C) The percentage of points of contact dis-
playing fluorescent signal for perforin (B) or CD107a
(C) is quantitated and shown. Means ± SDs for three
independent experiments using cells coming from
three different HIV controllers are shown; statistics
were calculated by unpaired t test. *p < 0.05.
(D) For each cell contact, the actin signal was
analyzed and quantified with IMARIS software in 3D.
Two representative experiments are shown, demon-
strating actin rings only in the HIV-infected condition.
(E) The percentage of points of contact displaying
actin rings is calculated for both conditions and
shown as means ± SDs for three independent ex-
periments with cells coming from three different HIV
controllers. Statistics were calculated by unpaired
t test.
(F) Actin polarization on CD8+ T cells toward the
point of contact was quantified by measuring the
ratio of actin fluorescent signal at the point of con-
tact versus the rest of the cell. The means ± SDs for
three independent experiments with 10 cell con-
tacts analyzed per experiment with cells coming
from three different HIV controllers; statistics were
calculated by unpaired t test.
(G) CD107a staining gated on live CD3+CD8+ cells
revealed the degranulation of CD8+ T cells in
response to autologous HIV X4 infected cells (blue)
or HIV R5 infected cells (yellow), compared to Gag
peptide-sensitized target cells (red). A representa-
tive experiment is shown.
(H) The means ± SDs for three independent experiments with cells coming from one HIV controller are represented. Statistics were calculated by an ANOVA
multiple comparison test relative to the non-infected condition. *p < 0.05, **p < 0.01, and ***p < 0.001.
(I) Antibodies to LFA-I, ICAM-2, and LFA-3 were used in an experiment similar to (G) with HIV X4. The means ± SDs for four independent experiments with cells
coming from four different HIV controllers are shown, and statistics were calculated with a one-way ANOVA multiple comparison test relative to the isotype
control. **p < 0.01; ns, not statistically significant.
See also Video S1 and Figures S3 and S4.

non-productively HIV-infected CD4+ T cells, before viral protein I, the main adhesion molecule involved in immunological synapse
production, elicit similar CD8+ T cell responses to autologous formation (Dustin and Shaw, 1999; Grakoui et al., 1999), resulted
cells loaded with pooled Gag peptides. in decreased degranulation of CD8+ T cells (Figure 2I). Blocking
The above results suggest that autologous CD8+ T cells can LFA-3 or ICAM-2 did not reduce the CD8+ T cell response, sug-
form immunological synapses and degranulate and release gesting that these molecules may be less important to the recog-
cytokines in response to non-productively HIV-infected resting nition of infected, non-activated CD4+ T cells.
CD4+ T cells. To assess the relation between synapse formation
and degranulation, we inhibited the formation of immunological CD8+ T Cell Recognition of Non-activated C4+ T Cells
synapses using antibodies against adhesion molecules involved following Cell-Cell Transmission
in this interaction (LFA-1, intercellular adhesion molecule 2 The above results were obtained using cell-free infection with
[ICAM-2], LFA-3). Consistent with our hypothesis, blocking LFA- a high viral inoculum. To further confirm our results under

Cell Reports 27, 142–153, April 2, 2019 145

 Figure 3. HIV Cell-to-Cell Transmission in
Non-activated CD4+ T Cells and Autologous
CD8+ T Cell Recognition
HeLa cells transfected with HIV Vpr-blam plasmids
were co-cultured for 2 h at 37 C with primary non-
activated CD4+ T cells from an HIV controller, and
then the CD4+ T cells were separated and put in co-
culture with autologous primary bulk CD8+ T cells
for 5 h.
(A) Schematic representation of the experiment.
(B) A Vpr-blam assay was conducted on the non-
activated CD4+ T cells after co-culture with trans-
fected HeLa cells to assess HIV entry by HIV X4
(blue), HIV R5 (yellow), or the control fusion defec-
tive virus HIV X4-F522Y (gray).
(C) CD107a staining of CD8+ T cells as analyzed
by flow cytometry. The means ± SDs for three in-
dependent experiments with cells from three
different HIV controllers are represented, and sta-
tistics were calculated with an ANOVA multiple
comparison test relative to the non-transfected (NT)
condition.
*p < 0.05.

HIV-infected CD4+ T cells, in this case

following cell-cell transmission.

Incoming Viral Particles Present

Viral Peptides to CD8+ T Cells
Previous studies have shown that acti-
vated simian immunodeficiency virus-
negative (SIV ) and HIV-infected CD4+
T cells can be recognized by CD8+
more physiologic conditions, we established a cell-to-cell T cells in response to the presentation of viral peptides derived
transmission assay to simulate what would be expected to from incoming particles and presented by HLA class I mole-
occur in tissues in vivo (Monel et al., 2012; Sourisseau et al., cules (Kløverpris et al., 2013; Payne et al., 2010; Buseyne
2007). HeLa cells were transfected with viral plasmids encod- et al., 2001). We next investigated whether this was also the
ing Vpr-blam viruses: HIV X4, HIV R5, or an X4 virus containing case for non-activated HIV+ CD4+ T cells in which activities
a mutation in the envelope (F522Y) preventing viral fusion with of antigen-processing enzymes are lower than in activated
the cellular membrane, but not the interaction with the recep- CD4+ T cells (J.B., unpublished data). To address this ques-
tor and co-receptor (Bergeron et al., 1992; Clavel et al., 1989). tion, cells were infected with a wild-type (WT) HIV X4 virus, a
HeLa cells were subsequently co-cultured with non-activated, fusion-defective mutant (HIV F522Y), or a virus rendered
CCF2-AM labeled uninfected CD4+ T cells for 2 h. A Vpr-blam non-infectious by the lack of the envelope protein (HIV
assay was performed on an aliquot of the CD4+ T cells, con- DEnv), and the experiment was conducted in the presence or
firming cell-to-cell transmission and viral entry. The remaining absence of the reverse transcriptase inhibitor efavirenz. As
CD4+ T cells were co-cultured with primary autologous shown in Figure 4A, degranulation in response to fusion-defec-
bulk CD8+ T cells for 5 h in the presence of CD107a antibody tive virus was reduced to background levels. Thus, virus entry
to analyze CD8+ T cell degranulation (Figure 3A). In this sys- into the cytoplasm, and not potential stress signaling caused
tem, primary, non-activated CD4+ T cells were permissive to by Env-receptor and co-receptor interactions, was absolutely
HIV X4 and HIV R5 entry by cell-to-cell transmission from necessary to induce an HIV-specific CD8+ T cell response.
transfected HeLa cells, but were not infected by the fusion- We also determined that viral reverse transcription was not
defective virus F522Y (Figure 3B). Furthermore, viral entry required to trigger the CD8+ T cell degranulation, as shown
into resting CD4+ T cells in this assay induced CD8+ T cell by the same level of CD107+ CD8+ T cells with or without efa-
degranulation directly ex vivo using unstimulated PBMCs virenz (Figure 4A).
from HIV-infected donors (Figure 3C). The relatively low levels Based on these data, we hypothesized that CD8+ T cell re-
of specific degranulation are consistent with expectations, sponses are triggered shortly after viral entry into the cytoplasm
given the frequency of HIV-specific effector CD8+ T cells in of the target cell, suggesting a role for incoming, as opposed to
PBMCs (data not shown). These data confirm the previous de novo transcribed, viral particles. Anti-HLA class I antibody
observations of CD8+ T cell recognition of non-productively blockade during the following 5-h incubation period post-infection

146 Cell Reports 27, 142–153, April 2, 2019

 Figure 4. CD8+ T Cell Response Induced by
the Presentation of Viral Peptides from
Incoming HIV Particles in Non-activated
CD4+ T Cells through the HLA-I Molecule
(A) The CD8+ T cell response to HIV+ non-activated
CD4+ T cells was evaluated by flow cytometry with
CD107a staining without HIV Env-CD4 and co-re-
ceptor interactions (HIV DEnv), without HIV entry into
CD4+ T cells (HIV F522Y), or in the presence of a
reverse transcription inhibitor (efavirenz) and
compared to HIV infection alone (HIV).
(B) The importance of the HLA-I molecule in the
CD8+ T cell response was evaluated by blocking the
HLA-I molecule with the anti-class I antibody (clone
W6/32), and measuring degranulation.
(C–E) The role of the antigen-processing pathway in
CD8+ T cell recognition was evaluated by measuring
degranulation by flow cytometry. This was done by
blocking the aminopeptidases with bestatin (C), the
cysteine proteases with E64 (D), and the protea-
some with epoxomicin (E).
(F) Schematic representation of the required steps
from viral entry to antigen presentation in a non-
activated CD4+ T cell to induce a CD8+ T cell
response. Means ± SDs for three independent ex-
periments with cells coming from three different HIV
controllers are represented.
Statistics were calculated with ANOVA multiple
comparison tests relative to the HIV condition for
(A), each column relative to every other column
for (B), and relative to the untreated condition for
(C)–(E).
*p < 0.05, **p < 0.01, and ***p < 0.001. Regarding
epoxomicin, the drug induced CD107a surface
expression on CD8+ T cells by itself through an un-
known mechanism, but in the absence of HIV pep-
tide stimulation or HIV exposure (data not shown).
For this reason, IFN-g staining was used.

iments support that antigen processing is a

prerequisite for CD8+ T cell recognition of
non-productively infected CD4+ T cells.
These data suggest that HIV fusion into
the cytoplasm, HIV degradation by ami-
nopeptidases and proteasomes, but not
reverse transcription, are required for anti-
resulted in a marked decrease in CD8+ T cell responses (Fig- gen processing of incoming viral particles and presentation by
ure 4B), indicating that the CD8+ T cell degranulation observed HLA class I (Figure 4F).
in response to HIV+ non-activated CD4+ T cells is driven by HLA
class I presentation of viral peptides entering the cytoplasm after Killing of HIV+ Non-activated CD4+ T Cells by HIV-
viral fusion. Specific CD8+ T Cells
To more precisely define the involvement of antigen pro- The above studies show that HIV-infected, non-activated CD4+
cessing in the recognition of non-activated cells, we blocked T cells trigger CD107a expression, but do not directly demon-
enzymatic aminopeptidase activities with bestatin, cysteine strate that these cells are killed. To address this, we performed
cathepsin activities with E64, and the proteasome with epoxomi- a chromium release assay using HIV+ non-activated CD4+
cin (Bogyo and Wang, 2002; Groll and Huber, 2004; Mathé, 1991; T cells (Chen et al., 2012) shortly after viral entry (Figure 5A).
Vaithilingam et al., 2013). Inhibition of aminopeptidases with Release of 51Cr into the supernatant was quantified, revealing
bestatin or blocking of proteasome activities, but not blocking direct killing by autologous HIV-specific CD8+ T cell lines at
of cysteine cathepsin activities, resulted in a decrease in CD8+ multiple effector-to-target ratios (Figures 5B and 5C). Further-
T cell responses to HIV+ non-activated CD4+ T cells (Figures more, killing was dramatically decreased in the presence of
4C–4E). The results from these pharmacological inhibition exper- an anti-HLA-I antibody compared to isotype antibody controls

Cell Reports 27, 142–153, April 2, 2019 147

A Figure 5. Killing of HIV+ Non-activated CD4+
T Cells by Autologous HIV-Specific CD8+ T Cells
(A) Schematic illustration of the experiment.
(B and C) Non-activated CD4+ T cells from HIV con-
trollers were infected with HIV-Vpr-blam+ for 2 h and
then incubated with 51Cr for 1 h. The cells were then
co-cultured with autologous KK10-specific (B) or TW10-
specific CD8+ T cell lines (C) at the indicated effector:
target cell ratios for 6 h. Blocking antibodies against
HLA-I or LFA-I were added before co-culture, as indi-
cated. The release of 51Cr in the supernatant was
measured, and the percentage of specific killing was
calculated as [(mean experimental cpm mean spon-
taneous cpm)/(mean maximum cpm mean sponta-
neous cpm)] 3 100.
Means ± SDs for two independent experiments in
duplicate with cells coming from two different HIV con-
trollers are represented, and statistics were calculated
B C with a two-way ANOVA multiple comparison test relative
NI
to 1:10 isotype condition.
50 **p < 0.01 and ***p < 0.001.
HIV
40
% specific killing

30

20 We therefore tested whether the greater levels

** of recognition of non-productively infected
10 ***
cells in HIV controllers versus progressors (Fig-
0 ure 6A) were attributable to greater frequencies
Isotype anti-HLA-I anti-LFA-I
of HIV-Gag-specific CD8+ T cells. Most of the
HIV progressor patients (9/10) showed CD8+
T cell responses to Gag peptide pool stimula-
tion (final concentration of 1 mg/mL/peptide),
(Figure 5B) and in the presence of anti-LFA1 antibody, confirm- as represented in Figure 6E (top), which was not significantly
ing the requirement for functional immunological synapses for different from that of HIV controllers (Figure S5D). Following
efficient killing (Figure 5C). We conclude that HIV-specific normalization of the IFN-g response to non-productively in-
CD8+ T cells are able to not only degranulate upon recognition fected cells to the total magnitude of Gag-specific CD8+ T cell
of HIV+ non-activated CD4+ T cells within hours of viral fusion response in the same subjects, we still observed proportionally
but also kill these cells. greater responses in HIV controllers as compared to progressors
(50% ±12% versus 18% ± 5.4%, p = 0.0334; Figure 6F). Of note,
CD8+ T Cell Responses to Non-productively Infected expression of programmed cell death protein 1 (PD-1), an
CD4+ T Cells Are Preferentially Detected in HIV exhaustion marker, was comparable in both groups (Figure S5C),
Controllers and exhaustion blockade did not enhance cell killing (Figure S6).
We next assessed whether the recognition of HIV+ non-activated Thus, the increased responsiveness of CD8+ T cells from control-
CD4+ T cells by HIV-specific CD8+ T cells was enhanced in HIV lers to non-productively infected cells cannot be attributed to
controllers compared to HIV progressors by comparing CD8+ greater magnitude or differential exhaustion status of Gag-spe-
T cell responses in 10 HIV controllers and 10 progressors, each cific CD8+ T cell responses.
expressing at least one protective allele (HLA-B*27 or HLA- To further analyze the role of the CD8+ T cell response in HIV
B*57; Table 1). As described above, non-activated PBMCs from control, we evaluated whether there was a link between IFN-g
these individuals were infected with HIV X4, and the CD8+ production by CD8+ T cells and the size of the reservoir in these
T cells responses were analyzed by flow cytometry 5 h later patients, measured by digital droplet PCR ex vivo, as described
using unfractionated PBMCs. Only CD8+ T cells from HIV control- by Henrich et al. (2012). Persons for whom HIV DNA could be
lers recognized HIV+ non-activated CD4+ T cells, as shown by measured were included (n = 6 for HIV progressors and n = 7
degranulation (p = 0.004; Figure 6A), IFN-g production (p = for HIV controllers). We observed a significant Spearman inverse
0.007; Figure 6B), and macrophage inflammatory protein 1b correlation between the antiviral response and the size of the
(MIP-1b) production (p = 0.016; Figure 6C). Of note, there were reservoir for all patients (Figure 6G, left), but also when only the
no differences in the production of TNF-a (Figure 6D), perforin, HIV controllers were analyzed (Figure 6 G, right). Moreover,
or granzyme B (Figures S5A and S5B) by the Gag-specific CD8+ T cells from HIV progressors did not respond to HIV-
CD8+ cells from HIV controllers and HIV progressors. infected, non-activated CD4+ T cells, even with a blockade of
As HIV particles are mainly composed of Gag and Pol proteins, exhaustion (Figure S6), which is consistent with a role for HIV-
we hypothesized that viral peptides presented through HLA-I specific CD8+ T cells in containing the HIV reservoir in non-acti-
shortly after HIV entry would preferentially target these proteins. vated CD4+ T cells.

148 Cell Reports 27, 142–153, April 2, 2019

Table 1. Characteristics of HIV-1-Infected Study Participants
ID HLA-A1 HLA-A2 HLA-B1 HLA-B2 HLA-CW1 HLA-CW2 ARV Therapy Status CD4 (Number/mm3) HIV VL (Copies/mL)
CP1 01:01 31:01 40:01 57:02 03:04 06:02 on therapy 10 y 953 %50
interrupted
CP2 02:01 02:01 27:02 51:01 02:02 02:02 on therapy 1 y 313 %50
CP3 01:01 24:02 35:03 57:01 07:01 12:03 on therapy 1 y 575 95
CP4 02:01 24:02 27:05 27:05 01:02 02:02 on therapy 1 y 713 197
CP5 02:01 02:01 27:05 44:02 01:02 02:02 on therapy <1 y 587 17,583
CP6 01:01 68:02 14:02 57:01 06:02 08:02 on therapy 3 y 421 %50
CP7 01:01 32:02 08:01 27:05 01:02 07:01 on therapy 3 y 1,208 %50
CP8 01:02 11:01 27:06 81:01 03:04 08:04 on therapy 4 y 419 %50
CP9 01:01 23:01 44:03 57:01 04:01 06:02 on therapy 1 y 408 75
CP10 01:01 03:01 35:01 57:01 06:02 08:02 on therapy 2 y 521 %50
EC1 01:01 02:01 51:01 57:01 06:02 14:02 off therapy 797 %50
VC2 02:01 30:01 13:02 57:01 06:02 06:02 ARV naive 579 1,470
VC3 32:01 32:01 27:05 44:02 01:02 05:01 off therapy 1,239 189
VC4 02:06 25:01 27:05 37:01 03:03 06:02 off therapy 352 51
EC5 01:01 11:01 27:05 35:03 02:02 04:01 off therapy 719 %50
EC6 01:01 24:02 38:01 57:01 06:02 12:03 ARV naive 1,219 %50
EC7 01:01 34:02 08:01 57:01 06:02 07:01 off therapy 724 %50
VC8 01:01 26:01 27:05 57:01 02:02 06:02 off therapy 357 105
EC9 03:01 26:01 15:01 27:05 02:02 03:03 ARV naive 871 %50
EC10 02:01 03:01 15:01 27:05 01:02 04:01 ARV off 1,082 %50
The analysis includes subject identifier (ID far left), detailed HLA subtypes, ARV therapy status, CD4 count (number/mm3), and viral load (copies/mL).
CP, chronic progressor; EC, elite controller; VC, viremic controller; HLA, human leukocyte antigen; ARV, antiretroviral; VL, viral load.

DISCUSSION induce pyroptosis and the release of inflammatory molecules

(Doitsh et al., 2010, 2014). These data indicate that HIV-specific
HIV reservoirs represent a major barrier to HIV eradication (Ar- CD8+ T cell responses have the potential to reduce the establish-
chin et al., 2014; Kimata et al., 2016), and substantial research ment of an HIV reservoir in non-activated CD4+ T cells and
efforts are being directed toward the goal of identifying and elim- reduce HIV-induced inflammation, hence contributing to HIV
inating this obstacle to cure (Archin and Margolis, 2014; Pitman control.
et al., 2018). Latently infected resting CD4+ T cells represent the To further define the mechanism of recognition, we show that
majority of the HIV reservoir (Bruner et al., 2015; Siliciano and Sil- the recognition of non-productively infected resting CD4+ T cells
iciano, 2015), and it has been shown that HIV latency can be by HIV-specific CD8+ T cells occurs through synapse formation
established in resting CD4+ T cells directly after infection, without and requires the presentation of viral peptides derived from
intervening productive infection (Chavez et al., 2015). Therapeu- incoming viral particles on HLA-I molecules. Moreover, we
tic strategies aimed at eliminating HIV reservoirs have been show that non-activated cells can be targeted either following
dominated by the ‘‘kick and kill’’ paradigm, which involves reac- exogenous infection or cell-to-cell spread, which is likely a major
tivating latent CD4+ T cells to induce de novo production of viral mode of ongoing viral replication in vivo (Chen et al., 2007; Dimi-
proteins, and the subsequent elimination by the immune system, trov et al., 1993; Phillips, 1994). Our results indicate that
notably CD8+ T cells (Archin and Margolis, 2014; Pegu et al., incoming viral proteins are degraded in the cytosol of non-acti-
2015). Another strategy could be the elimination of infected cells vated cells by the proteasome and the aminopeptidases to pro-
before the establishment of the reservoir. Here, we show that duce antigenic peptides for presentation by HLA class I, as
ex vivo CD8+ T cells from HIV controllers (possessing at least blocking these enzymes or HLA class I reduced the CD8+
one HLA-B*27 or HLA-B*57 allele), but not HIV progressors T cell responses.
expressing the same alleles, are able to directly recognize and It is important to note that these results were obtained using
kill HIV-infected non-activated CD4+ T cells through the forma- PBMCs from HIV controllers expressing HLA-B*27 or HLA-
tion of immunologic synapses and class I restricted recognition B*57 and with the laboratory strain of HIV NL4.3 possessing a
of processed viral peptides. This occurs within a few hours wild-type sequence for most of the optimal epitopes. Further
following viral entry, allowing an antiviral response before HIV studies will be required to examine the recognition of non-acti-
reverse transcription and thus before the eventual establishment vated infected cells in the context of other restricting HLA-I al-
of HIV latency. Moreover, this recognition precedes the produc- leles. However, our results provide support for the hypothesis
tion of the incomplete reverse transcripts that are needed to that proteins from incoming particles are degraded in the cytosol

Cell Reports 27, 142–153, April 2, 2019 149

 (legend on next page)

150 Cell Reports 27, 142–153, April 2, 2019

and presented directly at the surface of the target cell through STAR+METHODS
HLA class I.
Previous studies have reported pre-integration presentation of Detailed methods are provided in the online version of this paper
the SIV-Gag peptides (Sacha et al., 2007) and the HIV-Gag and include the following:
KRWIILGLNK epitope on HLA-B*27 in activated CD4+ T cells
(Buseyne et al., 2001; Kløverpris et al., 2013). In addition, a d KEY RESOURCES TABLE
previous study (Buckheit et al., 2013) performed with cells from d CONTACT FOR REAGENT AND RESOURCE SHARING
HIV controllers possessing HLA-B*57 reported the elimination d EXPERIMENTAL MODEL AND SUBJECT DETAILS
of HIV+ resting CD4+ T cells by CD8+ T cells. The present study B Study subjects
B Cell lines
adds to these previous reports by using the Vpr-blam assay
and a GFP-expressing virus to precisely distinguish entry from d METHOD DETAILS
protein production, by demonstrating that the observed effect B Proviral DNA constructs
B Cells
is induced by the presentation of viral peptides on HLA-I mole-
B HIV fusion assay (Vpr-blam assay)
cules, by showing that functional immune synapses are required
for recognition, and by demonstrating that more physiologic cell B Generation of HIV-specific CD8+ T cell lines
B Confocal Microscopy and Analysis
to cell transmission results in the sensitization of cells for CD8+
B Flow cytometry analysis
T cell recognition before the process of reverse transcription.
Antigen presentation of incoming viral particles has been well B Blocking antibodies and drugs

studied for diverse viruses in antigen-presenting cells (APCs), B HIV cell-to-cell transmission
B Chromium release assay
mostly in dendritic cells and macrophages in the context of the
exogenous antigens’ pathway to prime the CD8+ T cells. Howev- B Quantification of HIV DNA reservoir by digital droplet

er, CD4+ T cells are not known as efficient APCs, and thus the PCR
presentation of viral antigens from incoming particles through d QUANTIFICATION AND STATISTICAL ANALYSIS
HLA-I that does not rely on de novo protein production is not
completely expected and is not well characterized in CD4+ SUPPLEMENTAL INFORMATION
T cells. Furthermore, activated CD4+ T cells and resting CD4+
Supplemental Information can be found online at https://doi.org/10.1016/j.
T cells possess different enzymatic activities regarding the anti-
celrep.2019.03.016.
gen-processing pathway (J.B., unpublished data), and due to
their activation status, resting CD4+ T cells are not expected to ACKNOWLEDGMENTS
efficiently present antigen from incoming particles. Therefore,
our study confirms that resting CD4+ T cells are able to present The authors thank Thomas Murooka and Thorsten Mempel for the HIV proviral
antigens from incoming particles through HLA-I and provides constructs, the Ragon Cell Processing Lab, the clinical staff at the Massachu-
some mechanistic insight regarding this antigen-presentation setts General Hospital, and all of the study participants, as well as the Ragon
Microscopy Core (particularly Thomas J. Diefenbach) and the Ragon Flow
pathway.
Cytometry Core (particularly Michael Waring). The authors thank also Thomas
Additional studies will be required to understand why CD8+
Obadia from Institut Pasteur (Paris) for help with the biostatistics analysis. This
T cells from HIV controllers are more efficient in recognizing study was supported in part by grants from the Bill and Melinda Gates
non-activated infected cells and to determine factors that modu- Foundation (OPP1066973 and OPP1146433), the NIAID (R37AI067073 and
late this recognition. It will be important also to determine R01AI118544), and the Harvard University Center for AIDS Research (CFAR;
whether this is unique to the protective alleles HLA-B*27 and P30 AI060354), which is supported by the following institutes and centers
B*57 or can be seen in the context of other alleles. Nevertheless, co-funded by and participating with the NIH: NIAID, NCI, NICHD, NHLBI,
NIDA, NIMH, NIA, FIC, and OAR.
these data indicate that non-productively infected cells can be
targeted by CD8+ T cells, suggesting a path forward to reducing
AUTHOR CONTRIBUTIONS
the viral reservoir. Furthermore, our study could have implica-
tions for T cell-based prophylactic vaccines as an adjunct to anti- B.M. designed the study, performed the experiments, analyzed the data, and
body-mediated vaccines, as the presence of these cells at the wrote the manuscript; P.L-M. provided intellectual input and helped to write
time of infection may limit the establishment of the reservoir. the manuscript; A.M. and P.J. performed the experiments; Y.P. performed

Figure 6. The CD8+ T Cell Response to HIV+ Non-activated CD4+ T Cells Participates in HIV Control
(A–D) CD8+ T cell responses from HIV controller (red) or progressor (blue) patients (all HLA-B*27 or B*57, and n = 10 patients for both groups) in the context of
whole non-activated PBMCs were measured by (A) CD107a expression, (B) intracellular IFN-g expression, (C) intracellular MIP-1b expression, and (D) intracellular
TNF-a expression after 2 h of HIV infection and 5 h of incubation. Results are shown following background subtraction. Statistical differences were assessed
using PRISM software and an unpaired t test. *p < 0.05, **p < 0.01, and ns, not statistically significant.
(E) CD8+ T cell response to Gag pool peptides (final concentration 1 mg/mL/peptide) was measured for each patient by intracellular IFN-g staining. A repre-
sentative experiment is shown for an HIV progressor (top) and for an HIV controller (bottom).
(F) The CD8+ T cell response to HIV+ non-activated CD4+ T cells expressed relative to the Gag pool peptide response.
(G) The HIV DNA reservoir was measured by droplet PCR, and the Spearman correlation between the log10 of Gag copies/106 non-activated CD4+ T cells and the
CD8+ T cell IFN-g response is shown for all of the patients (left, n = 12 patients) and by sub-groups (right, n = 6 patients for each group).
See also Table 1 and Figures S5 and S6.

Cell Reports 27, 142–153, April 2, 2019 151

the experiments, provided intellectual input, and helped to write the manu- Chun, T.W., Engel, D., Berrey, M.M., Shea, T., Corey, L., and Fauci, A.S. (1998).
script; J.B. and S.L.G. provided reagents, intellectual input, and helped to write Early establishment of a pool of latently infected, resting CD4(+) T cells during
the manuscript; R.B.J. performed the experiments and helped to write the primary HIV-1 infection. Proc. Natl. Acad. Sci. USA 95, 8869–8873.
manuscript; and B.D.W. was responsible for the overall design and conduct Clavel, F., Hoggan, M.D., Willey, R.L., Strebel, K., Martin, M.A., and Repaske,
of the study. R. (1989). Genetic recombination of human immunodeficiency virus. J. Virol.
63, 1455–1459.
DECLARATION OF INTERESTS Dahabieh, M.S., Battivelli, E., and Verdin, E. (2015). Understanding HIV la-
tency: the road to an HIV cure. Annu. Rev. Med. 66, 407–421.
The authors declare no competing financial interests. Davey, R.T., Jr., Bhat, N., Yoder, C., Chun, T.W., Metcalf, J.A., Dewar, R.,
Natarajan, V., Lempicki, R.A., Adelsberger, J.W., Miller, K.D., et al. (1999).
Received: July 27, 2018 HIV-1 and T cell dynamics after interruption of highly active antiretroviral ther-
Revised: November 25, 2018 apy (HAART) in patients with a history of sustained viral suppression. Proc.
Accepted: March 5, 2019 Natl. Acad. Sci. USA 96, 15109–15114.
Published: April 2, 2019
Deeks, S.G., and Walker, B.D. (2007). Human immunodeficiency virus control-
lers: mechanisms of durable virus control in the absence of antiretroviral
REFERENCES therapy. Immunity 27, 406–416.
Descours, B., Cribier, A., Chable-Bessia, C., Ayinde, D., Rice, G., Crow, Y., Ya-
Archin, N.M., and Margolis, D.M. (2014). Emerging strategies to deplete the
tim, A., Schwartz, O., Laguette, N., and Benkirane, M. (2012). SAMHD1 restricts
HIV reservoir. Curr. Opin. Infect. Dis. 27, 29–35.
HIV-1 reverse transcription in quiescent CD4(+) T-cells. Retrovirology 9, 87.
Archin, N.M., Sung, J.M., Garrido, C., Soriano-Sarabia, N., and Margolis, D.M. Dimitrov, D.S., Willey, R.L., Sato, H., Chang, L.J., Blumenthal, R., and Martin,
(2014). Eradicating HIV-1 infection: seeking to clear a persistent pathogen. M.A. (1993). Quantitation of human immunodeficiency virus type 1 infection
Nat. Rev. Microbiol. 12, 750–764. kinetics. J. Virol. 67, 2182–2190.
Baldauf, H.M., Pan, X., Erikson, E., Schmidt, S., Daddacha, W., Burggraf, M., Doitsh, G., Cavrois, M., Lassen, K.G., Zepeda, O., Yang, Z., Santiago, M.L.,
Schenkova, K., Ambiel, I., Wabnitz, G., Gramberg, T., et al. (2012). SAMHD1 Hebbeler, A.M., and Greene, W.C. (2010). Abortive HIV infection mediates
restricts HIV-1 infection in resting CD4(+) T cells. Nat. Med. 18, 1682–1687. CD4 T cell depletion and inflammation in human lymphoid tissue. Cell 143,
Bergeron, L., Sullivan, N., and Sodroski, J. (1992). Target cell-specific determi- 789–801.
nants of membrane fusion within the human immunodeficiency virus type 1 Doitsh, G., Galloway, N.L., Geng, X., Yang, Z., Monroe, K.M., Zepeda, O.,
gp120 third variable region and gp41 amino terminus. J. Virol. 66, 2389–2397. Hunt, P.W., Hatano, H., Sowinski, S., Muñoz-Arias, I., and Greene, W.C.
Bogyo, M., and Wang, E.W. (2002). Proteasome inhibitors: complex tools for a (2014). Cell death by pyroptosis drives CD4 T-cell depletion in HIV-1 infection.
complex enzyme. Curr. Top. Microbiol. Immunol. 268, 185–208. Nature 505, 509–514.
Bruner, K.M., Hosmane, N.N., and Siliciano, R.F. (2015). Towards an HIV-1 Dustin, M.L., and Shaw, A.S. (1999). Costimulation: building an immunological
cure: measuring the latent reservoir. Trends Microbiol. 23, 192–203. synapse. Science 283, 649–650.
Buckheit, R.W., 3rd, Siliciano, R.F., and Blankson, J.N. (2013). Primary CD8+ Finzi, D., Hermankova, M., Pierson, T., Carruth, L.M., Buck, C., Chaisson, R.E.,
T cells from elite suppressors effectively eliminate non-productively HIV-1 Quinn, T.C., Chadwick, K., Margolick, J., Brookmeyer, R., et al. (1997). Identi-
infected resting and activated CD4+ T cells. Retrovirology 10, 68. fication of a reservoir for HIV-1 in patients on highly active antiretroviral ther-
Bukrinsky, M.I., Stanwick, T.L., Dempsey, M.P., and Stevenson, M. (1991). apy. Science 278, 1295–1300.
Quiescent T lymphocytes as an inducible virus reservoir in HIV-1 infection. Sci- Grakoui, A., Bromley, S.K., Sumen, C., Davis, M.M., Shaw, A.S., Allen, P.M.,
ence 254, 423–427. and Dustin, M.L. (1999). The immunological synapse: a molecular machine
controlling T cell activation. Science 285, 221–227.
Buseyne, F., Le Gall, S., Boccaccio, C., Abastado, J.P., Lifson, J.D., Arthur,
L.O., Rivière, Y., Heard, J.M., and Schwartz, O. (2001). MHC-I-restricted pre- Groll, M., and Huber, R. (2004). Inhibitors of the eukaryotic 20S proteasome
sentation of HIV-1 virion antigens without viral replication. Nat. Med. 7, core particle: a structural approach. Biochim. Biophys. Acta 1695, 33–44.
344–349. Hakre, S., Chavez, L., Shirakawa, K., and Verdin, E. (2012). HIV latency: exper-
Carrington, M., and Walker, B.D. (2012). Immunogenetics of spontaneous imental systems and molecular models. FEMS Microbiol. Rev. 36, 706–716.
control of HIV. Annu. Rev. Med. 63, 131–145. Haqqani, A.A., Marek, S.L., Kumar, J., Davenport, M., Wang, H., and Tilton,
Cavrois, M., De Noronha, C., and Greene, W.C. (2002). A sensitive and specific J.C. (2015). Central memory CD4+ T cells are preferential targets of double
enzyme-based assay detecting HIV-1 virion fusion in primary T lymphocytes. infection by HIV-1. Virol. J. 12, 184.
Nat. Biotechnol. 20, 1151–1154. Henrich, T.J., Gallien, S., Li, J.Z., Pereyra, F., and Kuritzkes, D.R. (2012). Low-
Chavez, L., Calvanese, V., and Verdin, E. (2015). HIV Latency Is Established level detection and quantitation of cellular HIV-1 DNA and 2-LTR circles using
Directly and Early in Both Resting and Activated Primary CD4 T Cells. PLoS droplet digital PCR. J. Virol. Methods 186, 68–72.
Pathog. 11, e1004955. Ho, Y.C., Shan, L., Hosmane, N.N., Wang, J., Laskey, S.B., Rosenbloom, D.I.,
Chen, P., Hu €bner, W., Spinelli, M.A., and Chen, B.K. (2007). Predominant Lai, J., Blankson, J.N., Siliciano, J.D., and Siliciano, R.F. (2013). Replication-
mode of human immunodeficiency virus transfer between T cells is mediated competent noninduced proviruses in the latent reservoir increase barrier to
by sustained Env-dependent neutralization-resistant virological synapses. HIV-1 cure. Cell 155, 540–551.
J. Virol. 81, 12582–12595. Katlama, C., Deeks, S.G., Autran, B., Martinez-Picado, J., van Lunzen, J., Rou-
Chen, H., Ndhlovu, Z.M., Liu, D., Porter, L.C., Fang, J.W., Darko, S., Brock- zioux, C., Miller, M., Vella, S., Schmitz, J.E., Ahlers, J., et al. (2013). Barriers to a
man, M.A., Miura, T., Brumme, Z.L., Schneidewind, A., et al. (2012). TCR cure for HIV: new ways to target and eradicate HIV-1 reservoirs. Lancet 381,
clonotypes modulate the protective effect of HLA class I molecules in HIV-1 2109–2117.
infection. Nat. Immunol. 13, 691–700. Kimata, J.T., Rice, A.P., and Wang, J. (2016). Challenges and strategies for the
Chun, T.W., Stuyver, L., Mizell, S.B., Ehler, L.A., Mican, J.A., Baseler, M., eradication of the HIV reservoir. Curr. Opin. Immunol. 42, 65–70.
Lloyd, A.L., Nowak, M.A., and Fauci, A.S. (1997). Presence of an inducible Kløverpris, H.N., Payne, R.P., Sacha, J.B., Rasaiyaah, J.T., Chen, F., Takigu-
HIV-1 latent reservoir during highly active antiretroviral therapy. Proc. Natl. chi, M., Yang, O.O., Towers, G.J., Goulder, P., and Prado, J.G. (2013). Early
Acad. Sci. USA 94, 13193–13197. antigen presentation of protective HIV-1 KF11Gag and KK10Gag epitopes

152 Cell Reports 27, 142–153, April 2, 2019

from incoming viral particles facilitates rapid recognition of infected cells by Sacha, J.B., Chung, C., Rakasz, E.G., Spencer, S.P., Jonas, A.K., Bean, A.T.,
specific CD8+ T cells. J. Virol. 87, 2628–2638. Lee, W., Burwitz, B.J., Stephany, J.J., Loffredo, J.T., et al. (2007). Gag-specific
Laird, G.M., Bullen, C.K., Rosenbloom, D.I., Martin, A.R., Hill, A.L., Durand, CD8+ T lymphocytes recognize infected cells before AIDS-virus integration
C.M., Siliciano, J.D., and Siliciano, R.F. (2015). Ex vivo analysis identifies effec- and viral protein expression. J. Immunol. 178, 2746–2754.
tive HIV-1 latency-reversing drug combinations. J. Clin. Invest. 125, 1901– Siliciano, J.D., and Siliciano, R.F. (2013). HIV-1 eradication strategies: design
1912. and assessment. Curr. Opin. HIV AIDS 8, 318–325.
Massanella, M., and Richman, D.D. (2016). Measuring the latent reservoir Siliciano, J.M., and Siliciano, R.F. (2015). The Remarkable Stability of the
in vivo. J. Clin. Invest. 126, 464–472. Latent Reservoir for HIV-1 in Resting Memory CD4+ T Cells. J. Infect. Dis.
Massanella, M., Martinez-Picado, J., and Blanco, J. (2013). Attacking the HIV 212, 1345–1347.
reservoir from the immune and viral perspective. Curr. HIV/AIDS Rep. 10, Siliciano, J.D., Kajdas, J., Finzi, D., Quinn, T.C., Chadwick, K., Margolick, J.B.,
33–41. Kovacs, C., Gange, S.J., and Siliciano, R.F. (2003). Long-term follow-up
Mathé, G. (1991). Bestatin, an aminopeptidase inhibitor with a multi-pharma- studies confirm the stability of the latent reservoir for HIV-1 in resting CD4+
cological function. Biomed. Pharmacother. 45, 49–54. T cells. Nat. Med. 9, 727–728.
Monel, B., Beaumont, E., Vendrame, D., Schwartz, O., Brand, D., and Mam- Sloan, R.D., and Wainberg, M.A. (2011). The role of unintegrated DNA in HIV
mano, F. (2012). HIV cell-to-cell transmission requires the production of infec- infection. Retrovirology 8, 52.
tious virus particles and does not proceed through env-mediated fusion pores. Soriano-Sarabia, N., Bateson, R.E., Dahl, N.P., Crooks, A.M., Kuruc, J.D.,
J. Virol. 86, 3924–3933. Margolis, D.M., and Archin, N.M. (2014). Quantitation of replication-competent
Monroe, K.M., Yang, Z., Johnson, J.R., Geng, X., Doitsh, G., Krogan, N.J., and HIV-1 in populations of resting CD4+ T cells. J. Virol. 88, 14070–14077.
Greene, W.C. (2014). IFI16 DNA sensor is required for death of lymphoid CD4
Sourisseau, M., Sol-Foulon, N., Porrot, F., Blanchet, F., and Schwartz, O.
T cells abortively infected with HIV. Science 343, 428–432.
(2007). Inefficient human immunodeficiency virus replication in mobile lympho-
Murooka, T.T., Deruaz, M., Marangoni, F., Vrbanac, V.D., Seung, E., von cytes. J. Virol. 81, 1000–1012.
Andrian, U.H., Tager, A.M., Luster, A.D., and Mempel, T.R. (2012). HIV-in-
Stevenson, M. (1997). Molecular mechanisms for the regulation of HIV replica-
fected T cells are migratory vehicles for viral dissemination. Nature 490,
tion, persistence and latency. AIDS 11 (Suppl A), S25–S33.
283–287.
Stinchcombe, J.C., and Griffiths, G.M. (2003). The role of the secretory immu-
Pan, X., Baldauf, H.M., Keppler, O.T., and Fackler, O.T. (2013). Restrictions to
nological synapse in killing by CD8+ CTL. Semin. Immunol. 15, 301–305.
HIV-1 replication in resting CD4+ T lymphocytes. Cell Res. 23, 876–885.
Thierry, S., Munir, S., Thierry, E., Subra, F., Leh, H., Zamborlini, A., Saenz, D.,
Payne, R.P., Kløverpris, H., Sacha, J.B., Brumme, Z., Brumme, C., Buus, S.,
Levy, D.N., Lesbats, P., Saı̈b, A., et al. (2015). Integrase inhibitor reversal dy-
Sims, S., Hickling, S., Riddell, L., Chen, F., et al. (2010). Efficacious early anti-
namics indicate unintegrated HIV-1 dna initiate de novo integration. Retrovir-
viral activity of HIV Gag- and Pol-specific HLA-B 2705-restricted CD8+ T cells.
ology 12, 24.
J. Virol. 84, 10543–10557.
Tilton, C.A., Tabler, C.O., Lucera, M.B., Marek, S.L., Haqqani, A.A., and Tilton,
Pegu, A., Asokan, M., Wu, L., Wang, K., Hataye, J., Casazza, J.P., Guo, X., Shi,
J.C. (2014). A combination HIV reporter virus system for measuring post-entry
W., Georgiev, I., Zhou, T., et al. (2015). Activation and lysis of human CD4 cells
event efficiency and viral outcome in primary CD4+ T cell subsets. J. Virol.
latently infected with HIV-1. Nat. Commun. 6, 8447.
Methods 195, 164–169.
Pereyra, F., Addo, M.M., Kaufmann, D.E., Liu, Y., Miura, T., Rathod, A., Baker,
B., Trocha, A., Rosenberg, R., Mackey, E., et al. (2008). Genetic and immuno- Vaithilingam, A., Lai, N.Y., Duong, E., Boucau, J., Xu, Y., Shimada, M., Gandhi,
logic heterogeneity among persons who control HIV infection in the absence of M., and Le Gall, S. (2013). A simple methodology to assess endolysosomal
therapy. J. Infect. Dis. 197, 563–571. protease activity involved in antigen processing in human primary cells.
BMC Cell Biol. 14, 35.
Petitjean, G., Al Tabaa, Y., Tuaillon, E., Mettling, C., Baillat, V., Reynes, J., Se-
gondy, M., and Vendrell, J.P. (2007). Unintegrated HIV-1 provides an inducible Vatakis, D.N., Kim, S., Kim, N., Chow, S.A., and Zack, J.A. (2009). Human im-
and functional reservoir in untreated and highly active antiretroviral therapy- munodeficiency virus integration efficiency and site selection in quiescent
treated patients. Retrovirology 4, 60. CD4+ T cells. J. Virol. 83, 6222–6233.
Phillips, D.M. (1994). The role of cell-to-cell transmission in HIV infection. AIDS Walker, B.D., and Yu, X.G. (2013). Unravelling the mechanisms of durable con-
8, 719–731. trol of HIV-1. Nat. Rev. Immunol. 13, 487–498.
Pierson, T.C., Zhou, Y., Kieffer, T.L., Ruff, C.T., Buck, C., and Siliciano, R.F. €nthard, H.F., Havlir, D.V., Ignacio, C.C., Spina,
Wong, J.K., Hezareh, M., Gu
(2002). Molecular characterization of preintegration latency in human immuno- C.A., and Richman, D.D. (1997). Recovery of replication-competent HIV
deficiency virus type 1 infection. J. Virol. 76, 8518–8531. despite prolonged suppression of plasma viremia. Science 278, 1291–1295.
Pitman, M.C., Lau, J.S.Y., McMahon, J.H., and Lewin, S.R. (2018). Barriers Yang, O.O., Sarkis, P.T.N., Trocha, A., Kalams, S.A., Johnson, R.P., and
and strategies to achieve a cure for HIV. Lancet HIV 5, e317–e328. Walker, B.D. (2003). Impacts of avidity and specificity on the antiviral efficiency
Ritter, A.T., Asano, Y., Stinchcombe, J.C., Dieckmann, N.M., Chen, B.C., of HIV-1-specific CTL. J. Immunol. 171, 3718–3724.
Gawden-Bone, C., van Engelenburg, S., Legant, W., Gao, L., Davidson, Zhou, Y., Zhang, H., Siliciano, J.D., and Siliciano, R.F. (2005). Kinetics of
M.W., et al. (2015). Actin depletion initiates events leading to granule secretion human immunodeficiency virus type 1 decay following entry into resting
at the immunological synapse. Immunity 42, 864–876. CD4+ T cells. J. Virol. 79, 2199–2210.

Cell Reports 27, 142–153, April 2, 2019 153

STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER

Antibodies
AlexaFluor 488 Phalloidin Invitrogen A12379; RRID:AB_2315147
AlexaFluor 647 anti-CD107a clone H4A3 Biolegend 328612; RRID:AB_1227506
AlexaFluor 647 anti-perforin clone dG9 Biolegend 308109; RRID:AB_493255
PE/CY7 anti-CD107a antibody clone H4A3 Biolegend 328617; RRID:AB_11147761
PE anti-CD3 clone HIT3a Biolegend 300307; RRID:AB_314043
V500 anti-CD8 clone RPA-T8 BD Horizon 560775; RRID:AB_1937333
BV605 anti-CD4 clone OKT4 Biolegend 317437; RRID:AB_11204077
Pacific blue anti-Perforin clone B-D48 Biolegend 353305; RRID:AB_11124346
APC anti-IFN-g clone B27 Biolegend 506510; RRID:AB_315443
FITC anti-TNF-a clone MAb11 BD PharMingen 554512; RRID:AB_395443
PerCP-Cy5.5 anti-MIP1-b clone D21-13551 RUO BD PharMingen 560688; RRID:AB_1727567
AlexaFluor700 anti-GranzymeB clone GB11 BD PharMingen 561016; RRID:AB_2033973
Purified anti-human CD11a clone 38 Genetex GTX26132; RRID:AB_380799
Anti-human CD18 clone CBR LFA-1/2 Biolegend 366302; RRID:AB_2565276
Anti-human ICAM-2 clone CBRIC2/2 Genetex GTX42521; RRID:AB_11169501
Anti-human LFA3 clone TS2/9 Biolegend 330912; RRID:AB_2075983
Ultra-LEAFTM purified anti- HLA-A, B, C Biolegend 311428; RRID:AB_2561492
Chemicals, Peptides, and Recombinant Proteins
Probenecid Sigma-Aldrich P8761
CellTrackerTM Orange CMTMR dye ThermoFisher C2927
ProLongTM Gold Antifade mountant ThermoFisher P10144
GolgiStopTM BD Biosciences 554724
GolgiPlugTM BD Biosciences 555029
Fixable blue viability dye Life Technologies L-23105
BD Cytofix/CytopermTM BD Biosciences 554722
Efavirenz Sigma SML0536
Epoxomicin Enzo Life Science BML-PI127-0100
Bestatin Sigma Aldrich B8385
E64 Enzo Life Science ALX-260-007
BsaJI NEB R0536L
Critical Commercial Assays
CCF2-AM loading kit Invitrogen K1025
Dynabeads Human T-Activator CD3/CD28 ThermoFisher 111.31D
EasySep Human CD4+ T cell enrichment Stemcell 19052
Alliance HIV-1 P24 ANTIGEN ELISA Kit Perkin Elmer NEK050001KT
Gentra Puregene cell kit QIAGEN 158745
Experimental Models: Cell Lines
HeLa cells ATCC CCL-2
HEK293T cells ATCC CRL-3216
Recombinant DNA
Plasmid: NL4.3-GFP-X4 T.Murooka T. Mempel N/A
Plasmid: NL4.3-GFP-R5 T.Murooka T. Mempel N/A
Plasmid: pMM310 NIH 11444
(Continued on next page)

e1 Cell Reports 27, 142–153.e1–e4, April 2, 2019

Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Software and Algorithms
IMARIS Bitplane http://www.bitplane.com/imaris
GraphPad Prism GraphPad Software Inc. https://www.graphpad.com/scientific-software/prism/
FlowJo FlowJo LLC https://www.flowjo.com/
Quantalife ddPCR software QuantaLife, Inc. http://www.bio-rad.com/

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for reagents may be directed to and will be fulfilled by the Lead Contact, Bruce Walker (bwalker@
mgh.harvard.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Study subjects
PBMC from HIV-1-infected individuals were used for this study according to protocols approved by the Institutional Review Board of
the Massachusetts General Hospital. Peripheral blood mononuclear cells (PBMC) were isolated using Ficoll-Hypaque density
gradient centrifugation
HIV Controllers include Elite Controllers who were defined as having HIV-1 RNA below the level of detection for the respective
available ultrasensitive assay (e.g., < 75 RNA copies/ml by cDNA or < 50 copies by ultrasensitive PCR) without antiretroviral therapy
and Viremic Controllers as having HIV-1 RNA between 50 and 2000 RNA copies/mL. CD4+ T cell counts, viral loads and HLA types
were determined as described (Pereyra et al., 2008). HIV Progressors were treated with Anti-retroviral (ARV) therapy for 1 to 4 years.
Characteristics of the study subjects are shown in Table 1.

Cell lines
HeLa cells and HEK293T cells (both from ATCC) were cultured at 37 C in Dulbecco Modified Eagle Medium (DMEM) supplemented
with 10% of Fetal Bovine Serum and 1% of penicillin/streptomycin.

METHOD DETAILS

Proviral DNA constructs

HIV proviral constructs coding for replicative NL4.3-GFP-X4 and NL4.3-GFP-R5 were previously described (Murooka et al., 2012)
and were kindly provided by Thomas Murooka and Thorsten Mempel. Env-deleted (HIVDEnv) and fusion-defective (HIV F522Y)
HIV mutants were previously described (Bergeron et al., 1992; Clavel et al., 1989). F522Y carries a point mutation in Env, which
abrogates fusion but retains CD4 binding ability.

Cells
Peripheral blood mononuclear cells (PBMC) were isolated using Ficoll-Hypaque density gradient centrifugation, rested a few hours at
37 C before being infected, checked for activity status or activated with anti-CD3/anti-CD28 beads. CD4+ T cells were isolated from
PBMC by negative selection (StemCell Technologies, Vancouver, Canada). HeLa cells and HEK293T cells (both from ATCC)
were cultured at 37 C in Dulbecco Modified Eagle Medium (DMEM) supplemented with 10% of Fetal Bovine Serum and 1% of
penicillin/streptomycin.

HIV fusion assay (Vpr-blam assay)

We relied on the previously described Vpr–b-lactamase (Vpr-blam) assay (Cavrois et al., 2002; Tilton et al., 2014) to measure the
efficiency of HIV entry into the cytosol of target cells, using cell-free virus preparations. Virus stocks were produced by co-transfect-
ing 293T cells with HIV proviral clones and a plasmid encoding the Vpr gene fused to the b-lactamase gene (pMM310 NIH/AIDS
Reagent program catalog # 11444). Virus preparations were concentrated by ultracentrifugation on 20% sucrose (1 hr,
22,000 rpm, 4 C). Primary non-activated PBMC or PBMC activated with anti-CD3/anti-CD28 immunomagnetic beads (Dynal,
Invitrogen, ratio cell:beads 1:1) for 2 days, were then exposed to the virus preparation for 2 hours at 37 C. Cells were then washed
and loaded with the CCF2-AM loading kit (Invitrogen) in the presence of 1.8 mM Probenecid (Sigma). Cells were incubated for 1 hr at
room temperature and were then washed and fixed. Fluorescence following cleavage of CCF2-AM (excitation at 405 nm, emission at
450 nm) was measured by flow cytometry on a BD LSR-II system (Becton Dickinson) with FACSDiva8 software. These experimental
conditions were then adapted to measure virus infection in the context of cell-cell contacts, as described (Monel et al., 2012).
Briefly, HeLa cells were co-transfected with an HIV proviral clone and the Vpr–b-lactamase expressor plasmid. After 48 h, primary

Cell Reports 27, 142–153.e1–e4, April 2, 2019 e2

non-activated CD4+ T cells were added to the transfected and washed HeLa cells for 2 hours. The non-activated CD4+ T cells were
then harvested, washed, and processed as described above for the cell-free virus assay. Target cells were gated based on their sizes
to exclude doublets and syncytia.

Generation of HIV-specific CD8+ T cell lines

CTL lines specific were generated by stimulation with designated single peptides or a Gag peptide pool using cryopreserved
PBMC from HLA-B*27 or B*57 HIV Controllers. PBMC were thawed and incubated with the designated peptide (final concentration
1ug/mL/peptide) for 10 days in R10 medium (RPMI supplemented with 10% fetal bovine serum plus HEPES buffer, penicillin,
streptomycin, and L-glutamine). IL-2 (50 U/mL) was added on day 3. Specificity for cognate peptide was tested on day 10 by intra-
cellular cytokine staining (ICS) for IFN-g.

Confocal Microscopy and Analysis

Non-activated CD4+ T cells negatively isolated from PBMC were infected for 2 hours with HIV-X4 Vpr-blam viruses. After infection,
the cells were washed and incubated at 37 C. Two hours later the cells were loaded with the CCF2-AM loading kit for 1 hour at room
temperature. During this time, autologous Gag pool-stimulated CD8+ T cell lines were incubated with 5M CellTrackerTM Orange
CMTMR dye according to the manufacturer’s instructions. CD4+ T cells and CD8+ T cells were then mixed together at a 1:1 ratio
and were loaded onto polylysine-coated coverslips (1x106 cells in 160uL) in 24 well plates. After 30 min at 37 C, cells were fixed
with paraformaldehyde (PFA) 4% for 15min at room temperature, permeabilized with Triton 0.5% for 15min at room temperature,
blocked with PBS containing 3% bovine serum albumin (BSA) for 1 hour at room temperature and stained with AlexaFluor 488 Phal-
loidin (Molecular Probes) and AlexaFluor 647 anti-CD107a (clone H4A3 Biolegend #328612) or AlexaFluor 647 anti-perforin (clone
dG9 Biolegend #308109) overnight at 4 C. The coverslips were then mounted on slides using ProLong Gold Antifade Reagent (Life
Technology). Images were acquired with a Zeiss LSM 510 confocal microscope equipped with argon (488nm) and Diode (405nm,
561nm) lasers, 100X oil immersion objective and ZEN software. At least 10 contacts between CD4+ T cells and CD8+ T cells
were analyzed per condition with Z stacks spaced by 0.33 mm. The 3D images were then reconstituted using the IMARIS software
(Bitplane) and surfaces were created for each channel after background deduction.
Quantification of synaptic actin polarization
3D reconstituted images of synapses were analyzed with Imaris software. The fluorescent signal of the Phalloidin was quantified at
the site of contact between the effector and the target cell and compared to the Phalloidin signal on the rest of the CD8+ T cell with
Imaris software. Approximately 30 cells were analyzed per condition and the mean of the ratios (actin at point of contacts/actin in rest
of CD8+ T cell) was calculated.
Actin ring quantification
3D reconstituted images of synapses between HIV+ non-activated CD4+ T cells and CD8+ T cells were analyzed with Imaris
software. Surfaces were created for each channel after background deduction. En face views from each duplex were taken and
the blue pixels (target cells) were removed to allow an unobstructed view of the actin stain at the interface. The percentage of cells
showing a surface corresponding to actin in a ring-shape was blindly determined for each experiment.
Perforin and CD107a quantification at the site of contact
3D reconstituted images of synapses were analyzed with Imaris software. A 3D rectangle was drawn to delimit the interface between
the two cells. Spots objects identifying point-like objects were automatically created using the ‘‘Spots’’ tool of the Imaris software for
the channel corresponding to CD107a or perforin staining after background deduction. The percentage of interfaces presenting per-
forin or CD107a spots was determined for each experiment.

Flow cytometry analysis

Non-activated PBMC were infected or not with the indicated HIV strains for 2 hours at 37 C. Free virions were then washed with
PBS. The cells were resuspended in fresh media and incubated 5 hours at 37 C in the presence of PE/CY7 anti-CD107a
antibody (clone H4A3 Biolegend #328617) and in the presence of GolgiStopTM (BD Biosciences #554724) and GolgiPlugTM (BD
Biosciences #555029) for intracellular cytokine staining. The cells were then harvested and stained for viability (Fixable blue
viability dye, Life Technologies #L-23105) for 30 min, then stained with PE-CD3 (clone HIT3a Biolegend #300307), V500-CD8 (clone
RPA-T8 BD horizon #560775) and BV605-CD4 (clone OKT4 Biolegend #317437). After fixation and permeabilization (BD Cytofix/
CytopermTM #554722) the cells were stained for Pacific blue-Perforin (clone B-D48 Biolegend #353305), APC-IFN-g (clone B27
Biolegend #506510), FITC-TNF-a (clone MAb11 BD PharMingen #554512), PerCP-Cy5.5-MIP1-b (clone D21-13551 RUO BD
PharMingen #560688) and AlexaFluor 700-GranzymeB (clone GB11 BD PharMingen #561016). Fluorescence signals were then
quantified using a BD LSR-II flow cytometry system (Becton Dickinson) with FACSDiva8 software and results were analyzed with
FlowJo software.

Blocking antibodies and drugs

For blockade of adhesion molecules, purified anti-human CD11a antibody (Genetex cat: GTX26132 clone: 38), anti-human CD18
antibody (Biolegend cat: 366302 clone CBR LFA-1/2), anti-human ICAM-2 antibody (Genetex cat: GTX42521 clone: CBRIC2/2)
and anti-human LFA3 antibody (Biolegend cat: 330912 clone: TS2/9) were used.

e3 Cell Reports 27, 142–153.e1–e4, April 2, 2019

 For blockade of the HLA class I molecule, Ultra-LEAFTM purified anti-human HLA-A, B, C antibody (BioLegend clone W6/32) was
used. In order to block HIV reverse transcription, the retroviral drug Efavirenz (Sigma; SML0536) was used at a final concentration of
100nM. For the study of antigen processing pathways, the proteasome hydrolytic activities were blocked with epoxomicin (Enzo Life
sciences-BML-PI127-0100), the aminopeptidase activities with Bestatin (Sigma-Aldrich B8385) and the cysteine proteases (cathep-
sins B, H and L) with E64 (Enzo Life Sciences ALX-260-007) at different final concentrations as indicated on the figures.

HIV cell-to-cell transmission

HIV cell-to-cell transmission was performed as previously described (Monel et al., 2012). Donor HeLa cells were transfected with
different proviral constructs (HIV X4, HIV R5, HIV F522Y) and plated in 24-well plates (105 cells per well). Cells were then washed
to eliminate cell-free virions 48 h post-transfection, and 5x105 target cells (non-activated CD4+ T cells) were added in a final volume
of 500 ml. Target cells were collected 2 hours later; an aliquot was used for a Vpr-blam assay to check HIV entry while remaining cells
were co-cultured (ratio 1:1) with autologous CD8+ T cells in 24-well plates for 5 hours in the presence of anti-CD107a antibody. Live
CD8+ T cells were then analyzed by flow cytometry for degranulation.

Chromium release assay

Primary non-activated CD4+ T cells were isolated from PBMC by negative selection (EasySepTM Human CD4+ T cell enrichment kit
from StemCell) and infected with 500ng of p24 (determined by ELISA) of NL4.3 containing the fusion protein Vpr-blam for 2 hours at
37 C. After infection, the cells were washed, and an aliquot was removed to perform an HIV fusion assay as described ahead. The
remaining cells were labeled with chromium (around 50 mCi) for 1 hr at 37 C. HIV Gag pool- or Gag-KK10-stimulated CD8+ T cell lines
were then added at the indicated effector-target ratios, and a standard 6 hour chromium release assay was performed as previously
described (Yang et al., 2003). The plates were spin down to pellet the cells and the supernatants were transferred from each well to
the well of an absorbent plate. The plates were dried overnight and put in a plate reader to measure the 51Cr release. Each sample
were performed at least in duplicate. Percent specific lysis was calculated as [(mean experimental cpm – mean spontaneous cpm)/
(mean maximum cpm – mean spontaneous cpm)] 3 100. Spontaneous and maximum releases were determined by incubating the
labeled target cells with medium alone or 5% Triton X-100, respectively.

Quantification of HIV DNA reservoir by digital droplet PCR

Digital droplet PCR was performed as previously described (Henrich et al., 2012). Genomic DNA was extracted from primary
non-activated CD4+ T cells using the Gentra Puregene kit (Gentra) following the manufacturer’s instructions. For each PCR reaction,
5 units of restriction enzyme BsaJI (NEB) was directly mixed with 300ng of DNA, ddPCR Supermix for Probes (Bio-Rad), and final
concentrations of 900nM primers and 250nM probe. Primers/Probes were: RPP30 – forward primer GATTTGGACCTGCGAGCG,
reverse primer GCGGCTGTCTCCACAAGT, probe VIC-CTGAACTGAAGGCTCT-MGBNFQ; HIV-gag – forward primer TACTGAC
GCTCTCGCACC, reverse primer TCTCGACGCAGGACTCG, probe FAM-CTCTCTCCTTCTAGCCTC-MGBNFQ. Droplets were pre-
pared using the QX100 Droplet Generator (Bio-Rad) following the manufacturer’s instructions. Sealed plates were cycled using the
following program: 95 C for 10 min; 40 cycles of 94 C for 30 s, 60 C for 1 min; and 98 C for 10 min, with 2 C/sec ramping speed to
ensure even droplet heating. Reactions were analyzed using the QX100 Droplet Reader and number of template molecule per ml of
starting material was estimated using the Quantalife ddPCR software. We aimed to run 8 replicates of ddPCR per sample – depend-
ing on the amount of DNA availability. We consistently applied a pre-determined exclusion criterion to outliers that deviated from
mean values by > 2x the standard deviation.

QUANTIFICATION AND STATISTICAL ANALYSIS

Most of the data are represented as Mean ± standard deviation and statistical analyses were performed using GraphPad PRISM soft-
ware. t tests were used to compare two groups of samples, one-way ANOVA multiple comparison tests were used to compare each
column to a control column or to every other columns and two-way ANOVA multiple comparison tests were used when grouped data
were analyzed and compare to a control column (Figures 5B and 5C). Regarding Figure 6G we used a Spearman correlation to assess
monotonic relationships between the Log10 of Gag copies/10^6 non-activated CD4+ T cells and the CD8+ T cell IFN-g response. The
Spearman’s rank correlation coefficient (rs) is indicated in the Figure. The corresponding tests are indicated for each figure in the
legend. In Figure 6, the n indicated in the legend represents the number of patients studied for each group (n = 10 Controllers and
n = 10 Progressors for Figures 6A–D and 6F; n = 12 for Figure 6 left panel; n = 6 Controllers and n = 6 Progressors for Figure 6G right
panel. No methods were used to determine whether the data met assumptions of the statistical approach.

Cell Reports 27, 142–153.e1–e4, April 2, 2019 e4