Frequency of Class I HLA-Restricted Anti-HIV CD8?T Cells
in Individuals Receiving Highly Active Antiretroviral Therapy
Clive M. Gray,2* Jody Lawrence,* Jonathan M. Schapiro,* John D. Altman,†‡
Mark A. Winters,* Meg Crompton,* Muoi Loi,* Smriti K. Kundu,* Mark M. Davis,†and
Thomas C. Merigan*
Peptide/MHC tetrameric complexes were used to enumerate the frequency of HLA class I-restricted epitope-specific CD8?T cells
in 18 HLA-A*0201 HIV type 1-infected asymptomatic patients. HLA-A*0201 molecules were complexed to HIV Gag p17 (amino
acids 77–85) and reverse transcriptase (amino acids 464–472) peptides, biotinylated, and bound to streptavidin-phycoerythrin to
form tetramers. We show in this study that 17 of 18 HIV-1-infected asymptomatic patients have circulating frequencies of
1/50–1/1000 CD8?T cells that recognize both Gag and Pol CTL epitopes or either epitope alone. The functional nature of these
cells is open to interpretation, as we show that despite relatively high frequencies of fresh epitope-specific CD8?T cells, variant
epitope sequences in viral plasma progeny were rare. In addition, the majority of tetramer-positive cells did not display discernible
fresh CTL activity; only after restimulation with specific peptide in culture was there an expansion of epitope-specific CD8?cells,
correlating with high CTL activity. These data suggest that fresh tetramer-stained cells probably represent memory precursors;
we demonstrate, with the application of highly active antiretroviral therapy, that the interruption of chronic antigenic stimulation
causes significant reductions in the frequency of these cells in five of six patients. In conclusion, this study provides evidence that
persistently replicating viral populations are probably required to maintain high frequencies of HIV-1 epitope-specific CD8?T
cells in asymptomatic chronically infected individuals The Journal of Immunology, 1999, 162: 1780–1788.
antiretroviral therapy (HAART)3can be improved. Previous re-
ports have suggested that improvement may be possible, and that
the immune system has the flexibility to respond positively to the
removal or reduction of high levels of replicating HIV-1 (1–3).
Data from different studies have shown, using V? repertoire anal-
ysis, that oligoclonally expanded CD4?(4) or CD8?(3) T cells
are significantly diminished during HAART. We have shown re-
cently that a skewed CD8?T cell repertoire, as measured by V?
and V? TCR expression on CD8?cells, was significantly reduced
after 8 wk of HAART (3). This finding provides indirect evidence
that lowering persistently high levels of viral replication reduces
the stimulus that maintains expanded CD8?clones. However, a
recent study has shown that a highly perturbed CD8?TCR rep-
ith the advent of highly active antiretroviral drugs,
one of the major questions is whether immune integ-
rity in HIV-1-infected patients receiving highly active
ertoire is not influenced by a reduction in viral burden in patients
receiving HAART (4). These data may reflect persistent CTL ex-
pansion despite the suppression of virus, although the Ag speci-
ficity of these cells was never examined. We have decided to ad-
dress this issue and to extend our earlier observation (3) by
investigating the frequency of epitope-specific CD8?cells using
peptide/MHC tetrameric complexes (5). We wished to explore the
hypothesis that high levels of replicating HIV-1 are required to
maintain anti-HIV CD8?cells. As it has been shown functionally
that a restricted anti-HIV CTL clonal repertoire exists in infected
individuals (6, 7), it is possible that a reduced clonal repertoire
during HAART could potentially result in broader CTL responses
to other Ags.
Since the use of peptide/MHC tetramers was first reported (5),
we have been investigating the relationship between the occur-
rence of these cells with CTL function in relation to HAART.
Recent reports have shown that there is a good relationship be-
tween IL-7-driven in vitro-expanded CTL and tetramer staining
(8), and that a significant relationship exists between freshly
stained tetramer and fresh CTL lysis (9). This latter study also
indicated that tetramer-positive cells had an inverse correlation
with viral load in natural infection and concluded that the main-
tenance of CTL was driven by virus.
In this study, we have used peptide/MHC complexes to focus
our investigation on CD8?T cells that recognize two HLA-
A*0201-restricted HIV-1 CTL epitopes. We have related the
frequency of epitope-specific CD8?cells with viral CTL
epitope sequence changes, CTL function, and the effect of
HAART; consequently, we were able to examine the effect of
removing persistent HIV-1 replication. Our data suggest that
the frequencies of CD8?T cells binding peptide/MHC tetram-
ers are likely to be memory CTLs rather than active effector
*Center for AIDS Research, Division of Infectious Diseases and Geographic Medi-
cine, Stanford University Medical Center, Stanford, CA 94305;†Department of Mi-
crobiology and Immunology and Howard Hughes Institute, Stanford University, Stan-
ford, CA 94305; and
University, Atlanta, GA 30322
‡Department of Microbiology and Immunology, Emory
Received for publication June 4, 1998. Accepted for publication October 15, 1998.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance
with 18 U.S.C. Section 1734 solely to indicate this fact.
1C.M.G. is a recipient of the James Gear Fellowship awarded by the Poliomyelitis
Research Foundation of South Africa. This paper was presented by C.M.G. at the
Clinical Immunology Society Annual Meeting during Experimental Biology 1998,
San Francisco, California, April 18–22, 1998.
2Address correspondence and reprint requests to Dr. Clive M. Gray, AIDS Unit,
National Institute for Virology, Private Bag x4, Sandringham 2131, Johannesburg,
South Africa. E-mail address: email@example.com
3Abbreviations used in this paper: HAART, highly active antiretroviral therapy; B-
LCL, B lymphoblastoid cell line; DC, dendritic cell; SQV, saquinavir.
Copyright © 1999 by The American Association of Immunologists0022-1767/99/$02.00
cells. In turn, the relatively high frequency of these cells ap-
pears to be maintained by the presence of HIV-1 replication,
because only when viral turnover is potently suppressed does
the frequency of epitope-specific CD8?T cells decline.
Materials and Methods
A total of 18 HLA-A*0201 patients were analyzed in a cross-sectional
manner to explore the relationship between the frequency of Gag- and
Pol-specific CD8?T cells and the variation in viral CTL epitope se-
quences. Table I shows the details of treatment history before the current
investigation, estimated time of HIV-1 diagnosis, Centers for Disease Con-
trol and Prevention (CDC) disease stage, CD4?counts, and RNA copies
per milliliter at the time of investigation. The estimated time of HIV-1
diagnosis was determined according to the first recorded p24 Ag. The
cohort was subdivided into three groups: six patients who never received
any form of treatment (treatment-naive); six patients who received either
glycoprotein 160 vaccine/placebo or allogeneic dendritic cell (DC) therapy
(10) that ended 12 mo before the current study (drug-naive), and six pa-
tients who received combinations of antiretroviral drug therapy with or
without adjunctive immunotherapy. A selection of these patients was fur-
ther assessed for fresh and in vitro-cultured CTL function and correlated to
peptide/MHC tetramer binding. Table II indicates which patients were se-
lected for this analysis.
Of the patients in the study, 9 of 18 were then selected for further study
in a longitudinal substudy to assess the effects of HAART on the frequency
of CD8?T cells binding peptide/MHC tetramers (Table II). Five of six
patients in the first group were selected (treatment-naive) to assess the
effect of this regimen on patients receiving drug treatment for the first time.
The effect of HAART on this group was compared with four other patients:
P12, who had received DC immunotherapy (drug naive), and P13, P15, and
P17, who had received protease inhibitor monotherapy followed by a triple
HAART rescue regimen (11).
The selection of patients for study was made on the basis of HLA-A*0201
type, which was identified using standard class I serological methods.
Blood was collected in tubes containing acid citrate dextrose to prevent
coagulation and PBMCs were isolated using Ficoll-Hypaque (Pharmacia,
Piscataway, NJ). After washing to remove excess platelets, cells were re-
suspended in PBS at a concentration of 1 ? 106cells/ml.
CD45RA (Becton Dickinson, San Jose, CA); anti-CD28, anti-CD38, and
anti-HLA-DR (Immunotech-Coulter, Hialeah, FL); and CyChrome-conju-
gated anti-CD8 (clone RPA-T8) (PharMingen, Cupertino, CA).
Tetrameric peptide/MHC complexes
The synthesis of the HLA-A*0201 tetrameric complexes used in this study
has been described elsewhere (5) and was folded to express one epitope in
measuring Gag and Pol peptide/MHC tetramer binding and viral epitope sequence analysisa
Treatment history and clinical parameters of 18 HIV-1-infected HLA-A*0201 individuals used for
Patient (P) Treatment History Prior to Study
gp160 vaccine or placebo
gp160 vaccine or placebo
gp160-pulsed allogeneic DC
gp 160 vaccine or placebo
allo- and auto-DC
IND?SQV?3TC?AZT; gp160-pulsed DC
IND?SQV?3TC and allogeneic CTL
aIND, indinavir; 3TC, lamuvidine; AZT, azuvidine.
effects of HAART
Patients used in a substudy for analysis of CTL function and
Patient (P)CTL FunctionEffect of HAART
1781The Journal of Immunology
the p17 region of Gag (SLYNTVATL) and the other in reverse transcrip-
Cell staining and flow cytometry
Either freshly isolated, thawed from frozen, or peptide-stimulated cells
(500,000) were resuspended in 15 ?l of PBS plus 2.5% FCS, supplemented
with 2 mM sodium azide, and incubated on ice for 45–60 min along with
25 ?l of phycoerythrin-labeled HLA-A*0201 tetramers (4 ?g). In addition,
5 ?l of anti-CD8-CyChrome stock and one of a panel of FITC-conjugated
Abs (5 ?l) were added along with the tetramer complexes. Consequently,
the total volume per stain was 50 ?l, making the final tetramer concentra-
tion 2 ?g/stain. Stained cells were washed twice in cold PBS/2.5% FCS
and once with cold PBS and fixed in PBS plus 2% formaldehyde. After
staining, cells were analyzed within 24 h using a FACScan flow cytometer.
A CD8?lymphocyte gate was made and 50,000 events were collected.
Subset analysis was performed using either CellQuest (Becton Dickinson)
or FlowJo software (TreeStar, Cupertino, CA). Color compensation set-
tings were made with each round of staining using patient cells labeled
singly with anti-CD8-FITC, -phycoerythrin, and -CyChrome.
Generation of EBV B lymphoblastoid cell line (B-LCL)
For each patient that was measured for CTL activity, EBV-B-LCLs were
generated ?6–8 wk before assay. These cells were used as autologous
targets in all CTL assays. EBV-B-LCLs were generated by infecting 2?
106PBMCs with EBV stock supernatants from B95-8 cell lines (American
Type Culture Collection, Manassas, VA) and were cultured in the presence
of cyclosporin A (14 ?g/ml) in media supplemented with 10% FCS. Trans-
formation usually occurred after 2–3 wk, and lines were established after
Fresh CTL activity was assessed using a standard 6-h51Cr release assay
employing radioactively labeled autologous EBV-B-LCL as targets pulsed
with Gag or Pol peptides. In vitro peptide-stimulated cells were assessed
for CTL activity in a 4-h assay using EBV-B-LCL pulsed with peptides (4
?M) (10). Background51Cr release was always ?25%. Specific CTL lysis
was calculated as follows: ([E ? M]/[T ? M]) ? 100%, where E is the
experimental release, M is the minimum release in the presence of media
alone, and T is the maximum release after targets were lysed with 10%
Triton X-100 detergent.
In vitro CTL cultures
Freshly isolated patient PBMCs (3–5 ? 106) were incubated with 2
?mol/ml of either Gag or Pol peptide for 24 h, after which 50 U/ml
rIL-2 (Life Technologies, Gaithersburg, MD) was added. The peptides
SLYNTVATL and ILKEPVHGV were purchased from Quality Controlled
Biochemicals (Hopkinton, MA) and were purified once by HPLC. Cultured
cells were then fed with fresh media plus 50 U/ml rIL-2 every 3 days, and
CTL activity was measured after 14 days.
Plasma viral load was assessed using the Roche Amplicor kit (Burlington,
NC), according to the manufacturer’s instructions.
RNA was extracted from plasma using Qiagen Viral RNA Prep kits (Chats-
worth, CA) according to the manufacturer’s instructions. The purified RNA
was reverse-transcribed and amplified using Superscript One-Step reagent
(Life Technologies) according to the manufacturer’s instructions. The
primers for the Gag epitope were MAW-5 (GTG CGA GAG CGT CGG
TA) and SK39 (TTT GGT CCT TGT CTT ATG TCC AGA ATG C);
primers for the Pol epitope were B (GGA TGG AAA GGA TCA CC) and
MAW-19 (GCT GGC TAC TAT TTC TTT TGC). The cycling parameters
were 45°C for 30 min and 95°C for 2 min followed by 40 cycles at 94°C
for 15 s, 55°C for 20 s, and 72°C for 2 min. A second-round PCR was
performed using 5 ?l of the first PCR reaction with 1? PCR buffer, 2.5
mM MgCl2, 2.5 U of Taq DNA polymerase (Life Technologies), 150 ?M
of deoxynucleoside triphosphate (Pharmacia), and 10 pmol of primer
(Operon Technologies, Alameda, CA). The second-round primers for the
Gag epitope were SK431 (TGC TAT GTC AGT TCC CCT TGG TTC
TCT) and MAW-29 (AAC ATA TAG TAT GGG CAA G); second-round
primers for the Pol epitope were MAW-15 (TTC CTT TGG ATG GGT
TAT GA) and MAW-20 (TTC TTG GGC CTT ATC CTA TTC C). The
cycling parameters were 35 cycles at 94°C for 15 s, 55°C for 20 s, and
72°C for 2 min. PCR products were then diluted 1/2–1/4 with water, and
10 ?l was used in dichloro-rhodamine terminator reactions (Applied Bio-
systems, Foster City, CA) according to the manufacturer’s instructions.
The sequencing primers for the Gag epitope were 77seq (AGC CTT CTC
TTC TAC TAC TTT TAC) and MAW-7 (ACA ACC ATC CCT TCA
GAC); sequencing primers for the Pol epitope were RT21 (CTG TAT TTC
TGC TAT TAA GTC TTT TGA TGG G) and MAW-17 (TTG GGC AAG
assessed on gated CD8?lymphocytes after collecting 50,000 events within the gate. Viral epitope sequencing was performed from extracted plasma RNA
using dideoxyterminator sequencing (see Materials and Methods) and was compared with reference samples using Sequence Navigator software (Applied
Pol and Gag peptide/MHC tetramer stains in 18 HLA-A*0201 patients in relation to plasma HIV-1 epitope sequence. Tetramer staining was
1782 ANTI-HIV CD8?T CELLS AND HAART
TCA GAT TTA CG). Data were collected on a model 377 DNA sequencer
(Applied Biosystems) and edited manually. Mixtures were reported when
a minority peak was 30% of the majority peak.
The strength of association between variables was measured using Spear-
man rank order correlation. Significant differences between values were
measured using the paired or unpaired Student t test or the Wilcoxon test
for unpaired data sets.
Frequency of Gag- and Pol-specific CD8?cells
The frequency of peptide/MHC Gag and Pol tetramer-positive
CD8?cells in the 18 HLA-A*0201 patients is shown in Fig. 1. The
objective of this part of the study was to observe the frequency of
epitope-specific CD8?T cells in a spread of patients. The Spear-
man rank correlation between the frequency of either Gag or Pol
tetramer binding and plasma viral load showed a lack of associa-
tion (r ? ?0.12 for Gag; r ? ?0.17 for Pol); this finding is in
contrast to the observations of Ogg et al. (9), who showed a sig-
nificant negative correlation between viral load and pooled Gag
and Pol tetramer staining. It is likely that the heterogeneity of the
treatment regimens given to the patients before analysis in the
current study may have obscured any potential relationship be-
tween tetramer-positive CD8?cells and viral load.
A total of 13 of 18 patients possessed a greater proportion of
CD8?cells recognizing the Gag epitope, and 4 individuals pos-
sessed CD8?cells that showed higher recognition of the Pol
epitope. Of the 13 patients who displayed binding to both Gag and
Pol tetramers, 11 had higher frequencies of CD8?cells recogniz-
ing the Gag epitope (range: 0.26–1.98%).
Viral epitope sequences
To determine whether frequencies of epitope-specific CD8?T
cells were exerting anti-HIV pressure, plasma virus was isolated
from each patient and sequenced in regions of Gag and Pol span-
ning the epitope sequences corresponding to the peptides folded
into each tetramer. Sequence analysis of the two epitopes showed
that 50% (9 of 18) of infected individuals had variant Gag epitopes
and 17% (3 of 18) had variant Pol epitopes when compared with
the peptides used in the construct (consensus sequence for subtype
B) Fig. 1. Two patients (P2 and P8) showed a tyrosine for phe-
nylalanine amino acid substitution at position 3. Another two pa-
tients (P12 and P13) had mixed viral populations shown by thre-
onine ? valine and leucine ? valine and at positions 8 and 2,
Relationship between tetramer binding and CTL function
To evaluate whether the frequency of tetramer binding related to
CTL function, we performed tetramer staining and measured CTL
activity from fresh PBMCs and after in vitro culture with peptide.
The relationship between the frequency of Gag and Pol tetramer-
positive CD8?cells with fresh CTL activity failed to show a sig-
nificant correlation (Fig. 2, A and B). Despite positive tetramer
binding to either Gag or Pol (ranging from 0 to 1.81% for Gag and
from 0 to 0.72% for Pol), the corresponding fresh peptide-specific
lysis was often below the 5%51Cr release cut-off; lysis was not
detectable in three individuals. Two of seven individuals displayed
fresh lysis above 5%51Cr release against the Gag epitope (Fig. 2A)
and a further two against the Pol epitope (Fig. 2B). Stimulation of
cells in vitro with specific peptide and rIL-2 showed that tetramer-
positive CD8?cells could be expanded, corresponding to peptide-
specific CTL activity. Fig. 2C shows that a significant correlation
(r ? 0.76, p ? 0.006) existed between the percentage of Pol CTL
lysis (25:1 E:T ratio) after in vitro culture and fresh Pol tetramer
stain before culture. Likewise, there was a highly significant as-
sociation between the percentage of Pol-specific CTL lysis and
corresponding tetramer staining after in vitro culture (r ? 0.96,
p ? 0.001; Fig. 2D). Correlation between fresh and in vitro-cul-
tured Pol tetramer staining (Fig. 2E) revealed a significant linear
relationship (r ? 0.83, p ? 0.005). Similar restimulation experi-
ments with the Gag epitope revealed that only one of nine patients
(P4) showed an expansion of Gag tetramer-positive CD8?cells
that yielded high peptide-specific lysis (Fig. 2F). In those patients
in which Gag tetramer-positive cells could not be expanded, alter-
native stimulation strategies were employed. The addition of rIL-7
and a different source of peptide (Bachem California, Torrance,
CA) yielded similar results (data not shown), and indicated that
Gag-specific cells could not always be propagated in culture de-
spite being detectable in fresh PBMCs.
Phenotypes of resting vs activated tetramer-binding cells
Fresh CD8?cells binding peptide/MHC tetramers were shown
previously to display surface phenotypes characteristic of memory
ramer binding. Correlations between fresh Gag and Pol CTL lysis and
tetramer staining (A and B), between Pol CTL lysis after culture and fresh
Pol stain (C) and Pol stain after culture (D), and between Pol stain after
culture and Pol stain before culture (E) are shown. The correlation between
Gag CTL lysis and the percentage of CD8?Gag?T cells after in vitro
culture is shown in F. The strength of association between parameters for
all correlations was measured using Spearman rank order correlation.
Correlation between functional CTL and peptide/MHC tet-
1783The Journal of Immunology
cells (5). We have extended these observations to include the co-
expression of activation markers. Table III shows median values of
the percent expression of CD45RA, CD28, CD38, HLA-DR, and
CD69 Ags on gated CD8?cells coexpressing either Gag or Pol
tetramers. We selected treatment- and drug-naive patients for anal-
ysis to avoid the influence of antiretroviral drug therapy, which has
been associated with lower activation marker expression on CD8?
cells (1, 3). Of 12 patients (Table I), 11 showed Gag tetramer
binding and 8 showed Pol binding (Fig. 1) and thus represented
patient numbers for phenotype analysis (Table III). Large propor-
tions of Gag- or Pol-expressing CD8?cells were devoid of the
CD45RA naive cell marker, confirming that these cells were pre-
dominantly memory cells (5). Coexpression of CD28 and
HLA-DR on fresh cells appeared to be more variable between
patients, with CD28 coexpression found on both Gag?and Gag?
CD8?populations. A greater proportion of CD8?Pol?cells co-
expressed CD28 and HLA-DR (not significant); a representative
example is shown in Fig. 3. Coexpression of activation markers
CD38 and CD69 on either CD8?Gag?or CD8?Pol?cells re-
vealed a lack of expression, indicating that fresh tetramer-positive
cells were not acutely activated in these patients.
Representative tetramer staining in P12 is shown in Fig. 3,
where fresh CD8?Gag?and CD8?Pol?cells expressing costimu-
latory and activation markers are shown. In this patient, there was
differential expression of CD28 between Gag and Pol CD8?cells;
discernible fresh CTL activity was apparent only against autolo-
gous targets pulsed with the Pol peptide. After stimulation with
peptide in vitro, the frequency of CD8?Pol?-binding cells in-
creased 24-fold and appeared to lose CD28 expression and gain
expression of CD38 and intermediate CD69 expression. Increased
expression of these markers corresponded to an enhanced Pol-spe-
cific CTL activity of 84% at a 25:1 E:T ratio.
DR, and CD69 surface Agsa
Phenotype of CD8?Gag?or CD8?Po1?tetramer-positive cells that coexpress (?) or do not coexpress (?) CD45RA, CD28, CD38, HLA-
% CD45RA % CD28% HLA-DR % CD38% CD69
% CD8?Gag?(n ? 11)
% CD8?Po1?(n ? 8)
aData are shown as % median values and interquartile ranges from the analysis of treatment- and drug-naive patients.
lymphocytes indicating the coexpression of CD28, CD38, HLA-DR, and CD69 in fresh Gag- and Pol-specific cells and subsequently after stimulation with
the Pol peptide in vitro are shown. The corresponding CTL activity also is shown. FACS data were acquired using a FACScan (Becton Dickinson) flow
cytometer and analyzed using FlowJo software set at 2% probability contour levels and displaying outlier events.
Surface Ag expression of CD8?cells binding peptide/MHC tetrameric complexes. Representative FACS contour plots of gated CD8?
1784 ANTI-HIV CD8?T CELLS AND HAART
Changes in the frequency of tetramer-binding T cells during
Fig. 4A shows changes in the frequency of CD8?Gag?or
CD8?Pol?cells in six patients receiving HAART for the first time
over variable periods of ?28 wk. Consistent for all patients was
the sharp decline in plasma RNA copy numbers to below detect-
able limits, irrespective of the initial starting viral load at baseline.
After 8 wk of HAART, five of six patients showed a significant
decline in the frequency of CD8?Gag?cells. P12 did not show
this trend, and there was a persistent and significant increase in the
frequency of CD8?Gag?cells lasting for ?28 wk (representative
FACS data is shown in Fig. 3 at 24 wk). The frequency of
CD8?Pol?cells showed a similar trend in this patient, but never
to the same magnitude, indicating a significant increase from
0.55 ? 0.06% to 0.95 ? 0.09% after 8 wk. P1 displayed an in-
creased frequency of CD8?Gag?cells in the first 2 wk of
HAART; by 4 wk the frequency was significantly reduced. Like-
wise, in P2 there was a significant increase of CD8?Gag?(0.66 ?
0.05% to 1.22 ? 0.14%) in the first 4 wk; by 8 wk the frequency
of CD8?cells was significantly reduced (0.28 ? 0.02%) (Fig. 4A).
Fig. 4B shows changes in the numbers of CD4?and CD8?cells
during HAART for these patients. All patients showed increased
CD4?counts from a mean of 322 ? 206 cells/?l at baseline (n ?
6) to 677 ? 219 cells/?l at wk 16 (n ? 5). Likewise, CD8?counts
increased from a mean of 841 ? 429 cells/?l at baseline to 943 ?
328 cells/?l at wk 16.
Changes in CD8?Gag?subsets during HAART
Closer analysis of the increased frequency of CD8?Gag?cells in
P2 over the first 4 wk of HAART is shown in Fig. 5. There was an
increase in the frequency of CD8?Gag?cells coexpressing CD38,
HLA-DR, and CD69 (Fig. 5A). By 8 wk, coexpression of these
markers was lost. Fig. 5B shows the emergence of these activated
Gag-specific CD8?cells as contour FACS plots at 2 and 4 wk,
with their disappearance at 16 wk.
CD4?and CD8?cell changes in these patients is shown (B). Viral load was measured in plasma using the Roche Amplicor kit. Each datapoint is the
mean ? SD of five replicate staining experiments; a comparison between timepoints was made using Wilcoxon’s signed rank test.
(A) Changes in CD8?T cells binding peptide/MHC tetrameric complexes in treatment-naive patients receiving HAART. The course of
binding peptide/MHC tetramers after commencement of HAART. A, Co-
expression of CD38, HLA-DR, and CD69 on CD8?Gag?cells in the first
16 wk of HAART in P2. B, FACS contour plots showing distinct popula-
tions of activated CD8?Gag?cells during the first 4 wk after the com-
mencement of therapy.
Expression of surface activation markers on CD8?cells
1785 The Journal of Immunology
Long-term follow-up of CD8?tetramer-positive cells
To assess changes of CD8?tetramer-positive frequencies over
longer-term viral load changes, we analyzed three patients who
initially failed saquinavir (SQV) monotherapy (11) and were pro-
vided with HAART as a rescue regimen. The frequencies of
CD8?Gag?cells (P13 and P15) and CD8?Pol?cells (P17) are
shown in Fig. 6, where each patient displayed individualistic re-
sponses to the switch in treatment to triple-drug HAART (repre-
sented by solid bars). Two patients showed a significant reduction
in the frequency of CD8?Gag?cells (P13) and CD8?Pol?cells
(P17) only when HAART had commenced. One of these patients
(P13) showed a significant elevation of CD8?Gag?cell frequen-
cies that oscillated for 24 wk during SQV monotherapy despite an
initial reduction in viral load (Fig. 6). After the commencement of
HAART, there was a further significant increase in CD8?Gag?
frequencies after 2 wk, which then fell to a nadir of 0.12% after 20
wk. There appeared to be more of a consistent decline in the fre-
quency CD8?Pol?cells in P17, although only limited timepoints
were available for analysis. Contrary to these observations, P15
displayed an initial fall in the frequency of CD8?Gag?cells after
4 wk, but values were significantly elevated above baseline during
the HAART regimen.
Until very recently, it was not possible to directly measure and
visualize HIV epitope-specific CD8?cells. With the introduction
of peptide/MHC tetrameric complexes (5), it became apparent that
the frequency of Ag-specific cells was consistently higher than that
shown by limiting dilution analysis. Indeed, recent data have
shown that there is a 10- to 50-fold gap between the data obtained
using limiting dilution analysis and tetramer staining for LCMV-
specific CD8?T cells (12). Therefore, functional assays may well
underestimate CTL frequencies in response to viral infections, per-
haps due to cell death during culture (13). The present study has
shown that 72% of the HIV-infected individuals studied have
CD8?cells recognizing an epitope in Gag (SLYNTVATL) and to
a lesser extent Pol (ILKEPVHGV). The lower proportion of CD8?
cells recognizing the Pol epitope most likely reflects fewer reverse
transcriptase epitopes being expressed by infected cells (14–16)
and possibly with less efficiency than Gag (17). Despite the lower
Pol frequencies, half of the patients (n ? 9) possessed CD8?cells
recognizing both epitopes; this finding differs from the exclusivity
of epitope recognition that was reported recently (9). Despite the
relatively high frequencies of epitope-specific cells in at least three
patients (ranging from 1/50 to 1/70 CD8?cells that recognize
Gag), there was no evidence for epitope variation in replicating
viral populations. This would suggest that these cells were not
imparting any selection pressure (18), especially within the more
variable Gag region. However, sequencing did reveal that half of
the Gag epitopes were variant from the index sequence folded into
the tetrameric complex. Most of the amino acid substitutions, how-
ever, would not necessarily prevent epitope expression or TCR
engagement, except in two patients (P2 and P8) whose viral pop-
ulations showed phenylalanine to tyrosine substitution at position
3, which is a secondary anchor change that is known to prevent
CTL recognition (19). The change from lysine to arginine in two
Pol sequences may reduce binding efficiency (20) and may explain
why P5 has CD8?cells that do not recognize the Pol tetramer.
Overall, these data show that despite the existence of epitope-spe-
cific CD8?cells, there were no meaningful epitope variations in
replicating viral populations, suggesting that selection pressure
leading to CTL escape is not a common phenomena during chronic
infection. This may well be in agreement with a recent study by
Brander et al. (21); alternatively, these cells may not be active CTL
The lack of a significant correlation between tetramer-positive
cells and fresh discernible CTL activity suggests that the majority
of tetramer-binding T cells are more likely nonactive effectors.
Although the magnitude of fresh CTL lysis cannot be discounted
(9), the possibility that tetramer-binding cells are memory cells is
supported by restimulation experiments. Pol-specific CD8?cells
are expanded by ?20-fold when restimulated with peptide in cul-
ture, which correlates significantly with CTL activity (Fig. 2). Phe-
notypically, Pol-specific CD8?cells gained expression of CD38
tide/MHC tetramer staining in patients over 2–3 years. Initial treatment
consisted of SQV monotherapy (hatched bars) that was switched to a triple
HAART of indinavir plus lamuvidine (3TC) plus azuvidine (AZT) (solid
bars). Each datapoint is the mean ? SD of five replicate staining experi-
ments; a comparison between timepoints was made using Wilcoxon’s
signed rank test.
Longer-term follow-up of changes in the frequency of pep-
1786 ANTI-HIV CD8?T CELLS AND HAART
and CD69, with a maintenance of HLA-DR and a loss of CD28
expression (Fig. 3); this finding is compatible with activated CTL
effectors (22, 23). Consequently, we interpret the lack of substan-
tial CD38 and CD69 Ag expression on fresh circulating epitope-
specific CD8?cells (Table III) as being compatible with a memory
phenotype, thus providing further support that these cells are not
active effector CTLs.
When we measured the frequency of epitope-specific cells in
patients receiving potent antiretroviral therapy for the first time,
five of six showed a significant loss of tetramer-positive CD8?
cells in parallel with a suppression of replicating HIV (Fig. 4). This
observation suggests that the maintenance of HIV-1-specific
clones is dependent upon the persistence of replicating virus. A
reduction in cell frequency was independent of gross CD4?and
CD8?changes (Fig. 4B) and was also not likely due to mutations
occurring in Gag or Pol epitopes that were measured in the longer-
term patient group receiving triple HAART: P13, P15, and P17
(data not shown). As most epitope-specific CD8?T cells were not
acutely activated before drug therapy (Table III), a loss of circu-
lating tetramer-stained cells in response to HAART was unlikely
due to a decline in the gross activated CD8?T cell pool. There is
also a possibility that a migration or redistribution of Ag-specific
CD8?cells occurs from the periphery to reservoir sites of HIV
during HAART. It has been shown in natural HIV-1 infection that
skewed CD8?TCR repertoires are found within lymphoid sites
(24, 25), where intense viral replication and infection are taking
place. Conversely, it has also been shown that HIV-specific CTL
clones preferentially accumulate in the blood as opposed to the
lymph node (26). Therefore, it seems unlikely that such a redis-
tribution would take place, especially as the viral burden is also
reduced in lymphoid tissue during HAART (27). One patient in
our study (P12) was the only subject to show a continuous increase
in CD8?Gag?cells, and to a lesser extent CD8?Pol?cells, during
HAART. It is of interest that this patient received DC immuno-
therapy 12 mo before receiving a drug where some infusions of
DCs were pulsed with both the Gag and Pol epitopes (10). Intu-
itively, it would seem that expanded Gag-specific cells during
HAART may be potentially beneficial and that combined immu-
notherapy with HAART may prove a useful approach (28). How-
ever, these data may merely reflect an enrichment of a dominant
clone that may ultimately reduce the pool of functional T cells
available to control other infections and variations in HIV. Addi-
tional studies would need to be designed that would critically eval-
uate this possibility.
The observed increase in CD8?Gag?cells in the first few weeks
after HAART initiation in some patients (P1, P2, and P13) may
reflect a redistribution of cells from lymphoid tissue sites into the
periphery. A detailed analysis of one patient (P2) revealed that
these cells were acutely activated, expressed CD38 and CD69, and
were found in the circulation in the first 4 wk. It is likely that viral
load suppression is accompanied by reduced inflammatory signals
and more normal cytokine levels (2), allowing sequestered CD8?
CTLs to appear in the circulation (29, 30).
Collectively, these data may possibly reflect a scenario whereby
a removal of Ag persistence reduces the requirement for protective
immune responses (31, 32). For the maintenance of T cell memory
and, hence, potential protective responses, a source of consistent
Ag stimulation is thought to be crucial (31). This view has been
upheld in studies in which adoptively transferred primed CD8?
CTLs appear to die unless rechallenged with specific Ag (33).
Although there are opposing views for the maintenance of T cell
memory (34), our findings are consistent with the notion that a
reduction of viral antigenic stimulus is associated with decreased
numbers of activated CD8?cells (1, 3); without constant stimu-
lation, Ag-specific CTLs disappear (35).
The data in this study may therefore support the view that high
frequencies of Ag-specific CD8?cells in asymptomatic HIV-1-
infected individuals may merely reflect viral turnover rather than
being determinants in the control of viral replication (35, 36).
We thank the nursing staff at the Stanford AIDS Clinical Trial Unit: Sylvia
Stout, Pat Cain, Jane Norris, and Sandra Valle whose cooperation was
invaluable. Thanks also go to Darcy Levee, Kristi Smith, and Yvette Girard
for excellent technical assistance, to Drs. Andrew Zalopa and Jose Mon-
taya of the Positive Care Clinic at Stanford for the recruitment of drug-
naive patients, and to Dr. Bradley Efron for statistical advice. Finally, we
extend our thanks to the patients for their participation in this study.
1. Autran, B., G. Carcelain, T. Li, C. Blanc, D. Mathez, R. Tubiana, C. Katlama,
P. Debre, and J. Leibowitch. 1997. Positive effects of combined antiretroviral
therapy on CD4?T cell homeostasis and function in advanced HIV disease.
2. Pakker, N. G., D. W. Notermans, R. J. D. Boer, M. T. Roos, F. D. Wolf, A. Hill,
J. M. Leonard, S. A. Danner, F. Miedema, and P. T. Schellekens. 1998. Biphasic
kinetics of peripheral blood T cells after triple combination therapy in HIV-1
infection: a composite of redistribution an proliferation. Nat. Med. 4:208.
3. Gray, C., J. Schapiro, M. Winters, and T. Merigan. 1998. Changes in CD4?and
CD8?T cell subsets in response to highly active antiretroviral therapy in HIV
type 1-infected patients with prior protease inhibitor experience. AIDS Res. Hum.
4. Gorochov, G., A. U. Neumann, A. Kereveur, C. Parizot, T. Li, C. Katlama,
M. Karmochkine, G. Raguin, B. Autran, and P. Debre. 1998. Perturbation of
CD4?and CD8?T cell repertoires during progression to AIDS and regulation of
the CD4?repertoire during antiretroviral therapy. Nat. Med. 4:215.
5. Altman, J., P. Moss, P. Goulder, D. Barouch, M. McHeyzer-Williams, J. Bell,
A. McMichael, and M. Davis. 1996. Phenotypic analysis of antigen-specific T
lymphocytes. Science 274:94.
6. Kalams, S., R. Johnson, and A. Trocha. 1994. Longitudinal analysis of T cell
receptor (TCR) gene usage by human immunodeficiency virus 1 envelope-spe-
cific cytotoxic T lymphocyte clones reveals a limited repertoire. J. Exp. Med.
7. Kalams, S., R. Johnson, M. Dynan, K. Hartman, T. Harrer, E. Harrer, A. Trocha,
W. Blattner, S. Buchbinder, and B. Walker. 1996. T cell receptor usage and fine
specificity of human immunodeficiency virus 1-specific cytotoxic T lymphocyte
clones: analysis of quasispecies recognition reveals a dominant response directed
against a minor in vivo variant. J. Exp. Med. 183:1669.
8. Lalvani, A., T. Dong, O. Graham, A. A. Patham, H. Newell, A. V. Hill,
A. McMichael, and S. Rowland-Jones. 1997. Optimization of a peptide-based
protocol employing IL-7 for in vitro stimulation of human cytotoxic T lympho-
cyte precursors. J. Immunol. Methods 210:65.
9. Ogg, G. S., X. Jin, S. Bonhoeffer, P. R. Dunbar, M. A. Nowak, S. Monard,
J. P. Segal, Y. Cao, S. L. Rowland-Jones, V. Cerundolo, et al. 1998. Quantitation
of HIV-1-specific cytotoxic T lymphocytes and plasma load of viral RNA. Sci-
10. Kundu, S., E. Engleman, C. Benike, M. Shapiro, M. Dupuis, W. V. Schooten,
M. Eibl, and T. Merigan. 1998. A pilot clinical trial of HIV antigen-pulsed al-
logeneic and autologous dendritic cell therapy in HIV-infected patients. AIDS
Res. Hum. Retrovirus 14:551.
11. Schapiro, J. M., M. A. Winters, F. Stewart, B. Efron, J. Norris, M. J. Kozal, and
T. C. Merigan. 1996. The effect of high-dose saquinavir on viral load and CD4?
T-cell counts in HIV-infected patients. Ann. Intern. Med. 124:1039.
12. Murali-Krishna, K., J. Altman, M. Suresh, D. Sourdive, A. Zajac, J. Miller,
J. Slansky, and R. Ahmed. 1998. Counting antigen-specific CD8?T cells: a
reevaluation of bystander activation during viral infection. Immunity 8:177.
13. Klenerman, P., and R. M. Zinkernagel. 1997. What can we learn about human
immunodeficiency virus infection from a study of lymphocytic choriomeningitis
virus? Immunol. Rev. 159:5.
14. Collins, K. L., B. K. Chen, S. A. Kalams, B. D. Walker, and D. Baltimore. 1998.
HIV-1 Nef protein protects infected primary cells against killing by cytotoxic T
lymphocytes. Nature 391:397.
15. Tsomides, T. 1994. Naturally processed viral peptides recognized by cytotoxic T
lymphocytes on cells chronically infected by human immunodeficiency virus type
1. J. Exp. Med. 180:1283.
16. Yang, O. 1996. Efficient lysis of human immunodeficiency virus type 1-infected
cells by cytotoxic T lymphocytes. J. Virol. 70:5799.
17. Brander, C., W. Pichler, and G. Corradin. 1995. Identification of HIV protein-
derived cytotoxic T lymphocyte (CTL) epitopes for their possible use as a syn-
thetic vaccine. Clin. Exp. Immunol. 101:107.
18. McMichael, A., and R. Phillips. 1997. Escape of human immunodeficiency virus
from immune control. Annu. Rev. Immunol. 15:271.
1787 The Journal of Immunology
19. Goulder, P., A. Sewell, D. Lallo, D. Price, J. Whelan, J. Evans, G. Taylor, Download full-text
G. Luzzi, P. Giangrande, R. Phillips, and A. McMichael. 1997. Patterns of im-
munodominance in HIV-1-specific cytotoxic T lymphocyte antigen (HLA)-iden-
tical siblings with HLA-A*0201 are influenced by epitope mutation. J. Exp. Med.
20. Ruppert, J., J. Sidney, E. Celis, R. Kubo, H. Grey, and A. Sette. 1993. Prominent
role of secondary anchor residues in peptide binding to HLA-A2.1 molecules.
21. Brander, C., K. E. Hartman, A. K. Trocha, N. G. Jones, R. P. Johnson, B. Korber,
P. Wentworth, S. P. Buchbinder, S. Wolinsky, B. D. Walker, and S. A. Kalams.
1998. Lack of strong immune selection pressure by the immunodominant, HLA-
A*0201-restricted cytotoxic T lymphocyte response in chronic human immuno-
deficiency virus-1 infection. J. Clin. Invest. 101:2559.
22. Ho, H., L. Hultin, and R. Mitsuyasu. 1993. Circulating HIV-specific CD8?cy-
totoxic T cells express CD38 and HLA-DR antigens. J. Immunol. 150:3070.
23. Fiorentino, S., M. Dalod, D. Olive, J.-G. Guillet, and E. Gombard. 1996. Pre-
dominant involvement of CD8?CD28?lymphocytes in human immunodefi-
ciency virus-specific cytotoxic activity. J. Virol. 70:2022.
24. Reich, K., I. Dahne, N. Endres, M. Classen, and K. Deusch. 1995. Biased T cell
receptor V? repertoire of intestinal CD8?T cells in HIV disease. Adv. Exp. Med.
25. Carbonari, M., M. Cibati, A. Pesce, L. Dell’Anna, G. D’Offizi, A. Angelici,
S. Uccini, and M. Fiorilli. 1996. Comparison of the V? repertoire in peripheral
blood and in lymph nodes of HIV-infected subjects reveals skewed usage pre-
dominantly in CD8?T cells. Clin. Immunol. Immunopathol. 81:200.
26. Pantaleo, G., H. Soudeyns, J. Demarest, M. Vaccarezza, C. Graziosi, S. Paolucci,
M. Daucher, O. Cohen, F. Denis, W. Biddison, R. Sekaly, and A. Fauci. 1997.
Accumulation of human immunodeficiency virus-specific cytotoxic T lympho-
cytes away from the predominant site of virus replication during primary infec-
tion. Eur. J. Immunol. 27:3166.
27. Wong, J., H. F. Gunthard, D. V. Havlir, Z.-Q. Zhang, A. T. Haase, C. C. Ignacio,
S. Kwok, E. Emino, and D. D. Richman. 1997. Reduction in HIV-1 in blood and
lymph nodes following potent antiretroviral therapy and the virologic correlates
of treatment failure. Proc. Natl. Acad. Sci. USA 94:12574.
28. Pantaleo, G. 1997. How immune-based interventions can change HIV therapy.
Nat. Med. 3:483.
29. Sprent, J., and D. Tough. 1995. HIV results in the frame: CD4?cell turnover.
30. Mosier, D. 1995. HIV results in the frame: CD4?cell turnover. Nature 375:193.
31. Zinkernagel, R. M., S. Ehl, P. Aichele, S. Oehen, T. Kundig, and H. Hengartner.
1997. Antigen localization regulates immune responses in a dose- and time-de-
pendent fashion: a geographical view of immune reactivity. Immunol. Rev. 156:
32. Kundig, T. 1996. On the role of antigen in maintaining cytotoxic T-cell memory.
Proc. Natl. Acad. Sci. USA 93:9716.
33. Bell, E. B., S. M. Sparshott, and C. Bunce. 1998. CD4?T-cell memory, CD45R
subsets, and the persistence of antigen: a unifying concept. Immunol. Today 19:
34. Lau, L., B. Jamieson, T. Somasundaram, and R. Ahmed. 1994. Cytotoxic T-cell
memory without antigen. Nature 369:648.
35. Bevan, M. J., and T. J. Braciale. 1995. Why can’t cytotoxic T cells handle HIV?
Proc. Natl. Acad. Sci. USA 92:5765.
36. Feinberg, M., and A. McLean. 1997. AIDS: decline and fall of immune surveil-
lance. Curr. Biol. 7:R136.
1788ANTI-HIV CD8?T CELLS AND HAART