©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunology  aDVaNCE ONLINE PUBLICaTION �

A rt i c l e s

A subset of people infected with human immunodeficiency virus type 1 (HIV-1), called ‘elite controllers’ here, are distinguished by their abil-ity to maintain a state of apparently durable control of HIV-1 replica-tion without the need for antiviral therapy1,2. Viral control is linked to the expression of certain alleles encoding human leukocyte antigen (HLA) class I molecules3–5, particularly HLA-B*57, HLA-B*27 and HLA-B*5801, which suggests an immunological basis related to the function of CD8+ T cells. A published genome-wide association study has indicated that the nature of the HLA–viral peptide interaction is the main factor that modulates durable control of infection with HIV-1 in the absence of antiretroviral therapy6. However, the mechanistic basis for the association remains unclear, and it is also unclear why most people with so-called ‘protective HLA alleles’ actually develop progressive disease.

Many studies have attempted to define quantitative and qualita-tive differences in CD8+ T cell responses that may be associated with different outcomes in terms of control of viremia by the immune response. Simple quantitative measures have shown little correla-tion with viral control7,8, which suggests that qualitative features of CD8+ T cells may modulate efficacy. Factors that potentially modu-late protective HLA-associated CD8+ T cell responses include, among others, polyfunctionality9, antigen sensitivity or functional avidity10,11, proliferative capacity12, loading of lytic granules13, ex vivo expression

of perforin14, specific targeting of conserved viral regions15,16, immu-noregulatory mechanisms17–19, concurrent responses to multiple viral epitopes restricted by different HLA alleles20, CD8+ T cell–associated mutations that impair viral fitness21,22 and escape from the immune response23. Many studies have also suggested that properties of the interaction among the T cell antigen receptor (TCR), viral peptide and major histocompatibility complex may be involved24,25. However, the extent to which any of these factors influences the antiviral effi-cacy of the human immune response, as reflected by in vivo viral load, remains unclear, in part because of a lack of direct comparison of viral epitope–specific CD8+ T cell responses in people able to control infec-tion with HIV-1 (controllers) and those who progress to disease after infection with HIV-1 (progressors), sequence diversity in epitopes of HIV-1 targeted by the immune system that leads to escape from the immune response, and the potential confounding effect of the target-ing of multiple epitopes via diverse HLA molecules.

To address those limitations, we focused on HIV-1-infected people who express HLA-B*2705, which represents a situation in which the immune response is mostly if not exclusively mediated by targeting of a single epitope, KK10, of the group-associated antigen (Gag) pro-tein p24 (sequence, KRWIILGLNK; amino acids 263–272)23. From a large, well-pedigreed cohort26, we specifically selected five control-lers and five progressors who express HLA-B*2705 for whom the

1Ragon Institute of Massachusetts General Hospital, Massachusetts Institute of Technology, and Harvard University, Boston, Massachusetts, USA. 2Howard Hughes Medical Institute, Chevy Chase, Maryland, USA. 3Human Immunology Section, Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institute of Health, Bethesda, Maryland, USA. 4Faculty of Health Sciences, Simon Fraser University, Burnaby, Canada. 5Institute of Medical Science, University of Tokyo, Tokyo, Japan. 6Department of Internal Medicine, University Hospital Regensburg, Regensburg, Germany. 7These authors contributed equally to this work. Correspondence should be addressed to B.D.W. (bwalker@partners.org).

Received 15 March; accepted 11 May; published online 10 June 2012; doi:10.1038/ni.2342

TCR clonotypes modulate the protective effect of HLA class I molecules in HIV-1 infectionHuabiao Chen1,2,7, Zaza M Ndhlovu1,2,7, Dongfang Liu1, Lindsay C Porter1, Justin W Fang1, Sam Darko3, Mark A Brockman1,4, Toshiyuki Miura1,5, Zabrina L Brumme1,4, Arne Schneidewind1,6, Alicja Piechocka-Trocha1,2, Kevin T Cesa1, Jennifer Sela1, Thai D Cung1, Ildiko Toth1, Florencia Pereyra1, Xu G Yu1, Daniel C Douek3, Daniel E Kaufmann1, Todd M Allen1 & Bruce D Walker1,2

The human leukocyte antigens HLA-B*27 and HLA-B*57 are associated with protection against progression of disease that results from infection with human immunodeficiency virus type 1 (HIV-1), yet most people with alleles encoding HLA-B*27 and HLA-B*57 are unable to control HIV-1. Here we found that HLA-B*27-restricted CD8+ T cells in people able to control infection with HIV-1 (controllers) and those who progress to disease after infection with HIV-1 (progressors) differed in their ability to inhibit viral replication through targeting of the immunodominant epitope of group-associated antigen (Gag) of HIV-1. This was associated with distinct T cell antigen receptor (TCR) clonotypes, characterized by superior control of HIV-1 replication in vitro, greater cross-reactivity to epitope variants and enhanced loading and delivery of perforin. We also observed clonotype-specific differences in antiviral efficacy for an immunodominant HLA-B*57-restricted response in controllers and progressors. Thus, the efficacy of such so-called ‘protective alleles’ is modulated by specific TCR clonotypes selected during natural infection, which provides a functional explanation for divergent HIV-1 outcomes.

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�  aDVaNCE ONLINE PUBLICaTION nature immunology

A rt i c l e s

circulating viruses and cellular proviruses all had wild-type sequences in the dominant KK10 epitope targeted by this molecule at the time of analysis. This allowed comparative assessment of adaptive CD8+ T cell responses in people in whom the dominant CD8+ T cell response is to a single epitope of Gag in a setting in which the viral load and, by inference, the degree of CD8+ T cell–mediated control and not HLA allele or sequence variation in the targeted viral epitope, were the main variables. We made a detailed analysis of the epitope-specific CD8+ T cell responses and then extended that to include the dominant HLA-B*57-restricted epitope TW10 of Gag (sequence, TSTLQEQIGW; amino acids 240–49). Our data indicated that HLA-B*27- and HLA-B*57-restricted CD8+ T cells targeting the same epitopes in elite controllers and progressors were distinctly different, on the basis of potency and cross-reactivity of TCR recognition of HIV-1 and viral variants, which was in turn related to specific TCR clonotypes selected during natural infection.

RESULTSQuantitative measures of KK10-specific T cellsPublished studies have shown that people who express HLA-B*2705 generate an immunodominant response to the KK10 epitope of Gag p24; targeting of this epitope is critical to long-term control in these people23,27. Although escape from this response leads to accel-erated disease progression, variable viral loads and rates of CD4+ T cell decrease are already observed before escape occurs23. Given the importance of the KK10-specific CD8+ T cell response to dis-ease control, we reasoned that the identification of both controllers and progressors with the wild-type KK10 sequence would afford the opportunity to define the characteristics of effective and ineffective CD8+ T cell responses independently of any confounding effects of escape from the immune response.

We recruited five elite controllers and five progressors for detailed studies; all expressed HLA-B*2705 and had autologous virus with the wild-type KK10 epitope in HIV-1 RNA obtained from plasma and HIV-1 DNA obtained from cells (Table 1). We selected these subjects from a larger population that included people in whom vari-ants in this epitope were present in vivo. Viral loads in the controllers were all <50 RNA copies per milliliter of plasma and ranged as low as 0.2 RNA copies per milliliter of plasma in those for whom we did more sensitive testing. Progressors had viral loads that ranged from 4,073 to 22,094 RNA copies per milliliter of plasma (Table 1). We confirmed the dominance of the KK10-specific responses in these subjects by fine mapping with HLA-restricted optimal peptides in enzyme-linked immunospot (ELISPOT) assays of interferon-γ (IFN-γ; data not shown). We detected strong KK10-specific CD8+ T cell responses, as defined by tetramer analysis, in the peripheral blood of both controllers and progressors (Fig. 1a). Comparison of controllers and progressors showed there was no significant difference between these two groups in the proportion of KK10-specific cells in peripheral blood, as quantified by staining with a HLA-B*27–KK10 tetramer (Fig. 1b), or ELISPOT assay of IFN-γ (Fig. 1c), despite their substantial differences in plasma viremia.

As the TCR is a key structure that defines antigen recognition, we next evaluated whether differences in TCR use might be associated with different abilities to control viremia, due to TCR clonotypes with heterogeneous antiviral potential. We sorted cells positive for the HLA-B*27–KK10 tetramer and sequenced their genes encoding the TCR β-chain variable (V) region (‘TCRBV’) and TCR β-chain joining (J) segment (’TCRBJ’). Consistent with the findings of other studies28,29, we found considerable diversity of clonotype recruit-ment in all KK10-specific CD8+ T cell populations, and despite the

dominance of single clonotypes in each person, we did not observe ‘preferential’ use of a particular complementarity-determining region 3 (CDR3) motif among the various KK10-specific CD8+ T cell clonotypes (Table 1). Indeed, of the clonotypes identified, only two clonotypes were the same in two different people (subjects CTR22 and CR420 had the same clonotype (TCRBV27TCRBJ1-1) and CDR3 sequence (CASSGGRRAF); and subjects CTR22 and CR540 had the same clonotype (TCRBV21TCRBJ2-7) and CDR3 sequence (CASTNRGSEQY)), and only one subject (CR420) had a TCRBV4-3TCRBJ1-3 clonotype similar to that reported before in people who express HLA-B*27 in whom viral loads vary between 1,880 and 202,590 RNA copies per milliliter of plasma28,30,31. These data indicated that KK10-specific CD8+ T cells were quantitatively similar but demonstrated considerable heterogeneity in TCR use among people who targeted a genetically identical epitope through a genetically identical HLA allele, in whom we observed considerable differences in viral load.

Functional characteristics of KK10-specific T cellsMany reports have suggested qualitative features of CD8+ T cells associated with viral control. One such measure is antigen sensitivity (often called ‘functional avidity’)10,11,31, and HLA-B*27 is character-ized by T cell responses of high sensitivity28. However, whether anti-gen sensitivity varies with viral load in people who express HLA-B*27 and contain virus with the wild-type epitope has not been determined, to our knowledge. We next assessed antigen sensitivity in each of our subjects by examining (by ELISPOT) IFN-γ responses at limiting con-centrations of KK10. There was no difference between controllers and progressors in the sensitizing dose of peptides needed to yield 50% maximal effector-cell triggering of IFN-γ production by peripheral blood mononuclear cells (PBMCs; Fig. 2a). These results, in which all responses were detected in the presence of wild-type KK10 and therefore were not confounded by potentially cross-reactive responses induced by substitutions in the epitope sequence, were consistent with published reports showing that most clones specific for KK10 have similar antigen sensitivity10.

We next examined polyfunctionality ex vivo, including the ability of HLA-B*2705-restricted, KK10-specific CD8+ T cells to simultaneously produce the effector cytokines and chemokines IFN-γ, interleukin 2 (IL-2), tumor necrosis factor and CCL4 (MIP-1β) and to release cyto-toxic factors by monitoring expression of the degranulation marker CD107a after stimulation with KK10. Published population studies have shown that the ability of CD8+ T cells to produce four or five cytokines and chemokines concurrently is associated with HIV-1 controllers9, and although some epitope-specific responses have been evaluated in this manner10,28,32, to our knowledge this has not been examined for HLA-B*27-restricted responses in people known to have the wild-type virus, in whom the inducing antigen is thus the same. When we did this analysis for the KK10 epitope (Fig. 2b), cells that secreted CCL4 (MIP-1β) dominated the response in both progressors and controllers, consistent with published findings9. Pairwise comparisons showed several subsets in the two groups with significantly different functional profiles. For example, cells with dual expression of both IFN-γ and CCL4 (MIP-1β) were of significantly greater frequency in controllers, whereas cells with dual expression of both IFN-γ and tumor necrosis factor were of significantly greater frequency in progressors (Fig. 2b). However, although controllers showed enrichment for cells with expression of more than three cytokines and chemokines, this did not reach statistical significance (P = 0.1) and these cells made up only a small subset of the total KK10-specific CD8+ T cell response.

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunology  aDVaNCE ONLINE PUBLICaTION �

A rt i c l e s

Published studies have shown that chronic infection with HIV-1 skews the maturation of HIV-1-specific CD8+ T cells toward devel-opment into preterminally differentiated cells with poor cytotoxic activity33. We stained peripheral blood cells for CD27 and CD45RA as described34 to phenotypically distinguish four distinct subpopulations of KK10-specific cells. We detected similar proportions of KK10- specific cells with a central memory phenotype (KK10+CD27+CD45RA−), effector memory phenotype (KK10+CD27−CD45RA−) or terminally differentiated effector memory phenotype (KK10+CD27−CD45RA+)

in the controllers and progressors (Supplementary Fig. 1). Moreover, we observed that HIV-1-infected GXR cells (a green fluorescent protein (GFP) reporter T cell line, derived from the human lympho-blastoid CD4+ T cell line CEM, that fluoresces green after infection with HIV-1) expressing HLA-B*2705 (refs. 35,36) stimulated the pro-liferation of KK10-specific cells to a similar degree in both groups (Fig. 2c,d). Together these data indicated that the superior control of wild-type viremia in these elite controllers relative to that in chronic progressors expressing HLA-B*27, all of whom were treatment naive,

Table 1 Clonal analysis of HLA-B*2705-restricted KK10-specific CD8+ T cell populationsSubject and HLA KK10 Tet+ cells (%) pVL (copies/ml) CD4 count TCR β-chain and CDR3 Frequency

EC CTR22 KRWIILGLNK 3.53 <50 337 Vβ20-CSARDRTRANYGYT-J1.2 22/25HLA-A*0201/0201 KRWIILGLNK Vβ27-CASSGGRRAF-J1.1 1/25HLA-B*2705/5101 Vβ5.1-CASSSPDRTYGYT-J1.2 1/25HLA-C*0102/1402 Vβ21-CASTNRGSEQY-J2.7 1/25

EC CTR40 KRWIILGLNK 1.56 <50 1,415 Vβ7.2-CASSLSGRWSTDTQY-J2.3 11/21HLA-A*0201/0201 KRWIILGLNK Vβ7.2-CASSLEGRYSNQPQH-J1.5 4/21HLA-B*2705/5701 Vβ7.6-CASSLGTGKVEGYT-J1.2 1/21HLA-C*0102/0602 Vβ7.9-CASSPEGPRAIEQF-J2.1 2/21

Vβ15-CATSRELTGGPSYEQY-J2.7 1/21Vβ7.8-CASSQARASHLI-J1.4 1/21Vβ9-CASSEDRDTEAF-J1.1 1/21

EC CTR203 KRWIILGLNK 9.99 1.1 411 Vβ25.1-CASSEADFEAF-J1.1 13/37HLA-A*2601/6801 KRWIILGLNK Vβ18-CASSPGQFSHEQY-J2.7 11/37HLA-B*0702/2705 Vβ27-CASSARTGELF-J2.2 2/37HLA-C*0702/0202 Vβ20.1-CSARDGGEQY-J2.7 6/37

Vβ9-CASSPLGNSGNTIY-J1.3 2/37Vβ7.9-CASSLDRLEQF-J2.1 3/37

EC FW56 KRWIILGLNK 7.08 0.2 937 Vβ4.3-CASRPGLASNEQF-J2.1 16/31HLA-A*0201/0301 KRWIILGLNK Vβ6.5-CASRPGQGATEAF-J1.1 10/31HLA-B*1501/2705 Vβ20.1-CSARDRGTREVADNYGYT-J1.2 3/31HLA-C*0304/0102 Vβ15-CATSETGTTLEQY-J2.7 2/31

EC 13587 KRWIILGLNK 1.35 <50 994 Vβ7.9-CASSRDSNEQF-J2.1 15/27HLA-A*0206/3201 NA Vβ20.1-CSAREGLAGVLYEQY-J2.7 1/27HLA-B*1524/2705 Vβ28-CASSSSGGAGDTQY-J2.3 1/27HLA-C*0202/0304 Vβ20.1-CSAPTTEVAGSTDTQY-J2.3 1/27

Vβ5.4-CASSLTNLGEQY-J2.7 3/27Vβ27-CASSRTTGELF-J2.2 6/27

CP CR540 KRWIILGLNK 6.85 17,300 505 Vβ2-CASSAGPGQYGNTIY-J1.3 16/34HLA-A*0201/0201 KRWIILGLNK Vβ5.6-CASGGGTVYEQY-J2.7 8/34HLA-B*2705/4402 Vβ21-CASTNRGSEQY-J2.7 6/34HLA-C*0102/0501 Vβ4.3-CASSPGTNAYEQY-J2.7 4/34

CP CR420 KRWIILGLNK 2.93 13,900 582 Vβ20.1-CSAREGVEGYT-J1.2 21/30HLA-A*0201/1101 KRWIILGLNK Vβ27-CASSGGRRAF-J1.1 6/30HLA-B*2705/4402 Vβ4.3-CASSQGSGSGNTIY-J1.3 1/30HLA-C*0102/0501 Vβ4.3-CASSQVLRGVYGYT-J1.2 1/30

Vβ5.4-CASSLLAGGTDTQY-J2.3 1/30

CP CR338 KRWIILGLNK 2.2 6,800 870 Vβ27-CASSPRTGELF-J2.2 16/29HLA-A*0101/0201 KRWIILGLNK Vβ27-CASSQRTGELF-J2.2 3/29HLA-B*0702/2705 Vβ27-CASSRATGELF-J2.2 4/29HLA-C*0102/0702 Vβ5.1-CASSLEGGANL-J1.2 1/29

Vβ6.4-CASSVVRGNPNEQF-J2.1 5/29

CP 8222 KRWIILGLNK 1.9 4,073 269 Vβ27-CASSGSNLEAF-J1.1 21/30HLA-A*0201/2902 KRWIILGLNK Vβ7.9-CASSPLGVRAYEQY-J2.7 3/30HLA-B*2705/4403 Vβ4.3-CASSPGTSTYEQY-J2.7 6/30HLA-C*0102/1601

CP FEN33 KRWIILGLNK 6.4 22,094 603 Vβ7.9-CASSLAGGDSYEQY-J2.7 21/23HLA-A*0201/1101 KRWIILGLNK Vβ27-CASSGGVFYGYT-J1.2 1/23HLA-B*2705/5101 Vβ19-CATLGGFPDGYT-J1.2 1/23HLA-C*0102/0303

Clonal analysis includes HLA subtypes, presented below subject identifier (far left); KK10 epitope sequence of the virus in plasma (top) and provirus in PBMCs (bottom); frequency of tetramer-positive (Tet+) cells among total CD8+ T cells; plasma viral load (pVL), in RNA copies per milliliter of plasma; CD4+ T cells (CD4 count) per microliter of plasma (cells/µl); TCR β-chain variable region (Vβ), CDR3 sequence and β-chain joining region (J); and frequency of clonotype among KK10-specific CD8+ T cell populations (far right). P = 0.4206, CD4 count, controllers versus progressors (Mann-Whitney test). Data are from one experiment (time point) per subject.

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�  aDVaNCE ONLINE PUBLICaTION nature immunology

A rt i c l e s

was not reflected in the quantitative measurement of KK10-specific CD8+ T cells or qualitative assessment of the functional avidity, cytokine secretion, proliferative capacity or differentiation pheno-types of these cells.

Neutralization of virus by KK10-specific T cellsHaving shown that the measures of responses to the immunodomi-nant KK10 Gag epitope noted above did not differentiate control-lers from progressors, we next evaluated the functional ability of responses to this epitope to inhibit HIV-1 replication in vitro37. We limited our initial analysis to the elite controllers, in whom outgrowth of autologous virus in CD4+ T cells is delayed considerably38, which allowed us to use controlled inocula of exogenous HIV-1 isolates to infect these cells and measure the ability of defined numbers of CD8+ T cells to inhibit viral replication. We included in this analysis all five HLA-B*2705+ elite controllers demonstrated to have wild-type KK-10 epitope sequences in virus in their plasma and provirus in their PBMCs (Table 1) and assessed the antiviral ability of bulk CD8+ T cells and CD8+ T cell samples depleted of KK10-specific cells to inhibit viral replication in autologous CD4+ T cells by measuring the production of p24 antigen in the supernatant over 7 d (ref. 35). The addition of bulk CD8+ T cells to HIV-1-infected CD4+ T cells

resulted in three to four logs less production of p24 antigen at day 7 in culture, whereas viral inhibition was over 90% less after the addi-tion of cell samples from which KK10-specific CD8+ T cells were depleted (Fig. 3a). This confirmed that the main immune control was mediated by the KK10-specific response in each of these subjects and showed that all of the HLA-B*27+ controllers were able to limit viral replication in vitro.

We next sought to assess the antiviral function of CD8+ T cells from the progressors. However, outgrowth of autologous virus in HIV-1 progressors complicated the virus-inhibition assay when we used autologous CD4+ T cells as target cells (data not shown). We therefore sought to confirm the utility of an assay based on the use of GXR cells encoding HLA-B*2705 as target cells35,36. When we used those target cells together with bulk CD8+ T cells from controllers, we noted con-siderable inhibition of replication; moreover, there was nearly complete loss of viral inhibition after depletion of KK10-specific cells from bulk CD8+ T cell populations (Fig. 3b). In contrast, we observed no viral inhibition by bulk CD8+ T cells from HIV-1− people in the presence of HIV-1-infected autologous CD4+ T cells or HLA-B*2705-expressing GXR cells. In addition, we observed no inhibition by bulk CD8+ T cells from controllers in the presence of HLA-B*2705− GXR cells after infection with the same virus (data not shown).

a

105

105

KK10 tet

104

104

103

103

CD

8

7.08

EC FW56(VL = 0.2)

CP CR540(VL = 17,300)

102

102

0

0

c5 *

EC CP

4

3

2

SF

Cs

103 (

per

106 P

BM

Cs)

1

0

b1110987654

Tet

+C

D8+

cel

ls (

%)

3210

*

EC CP

6.85

Figure 1 Quantification of KK10-specific CD8+ T cell responses. (a) Flow cytometry of CD8+ T cells from elite controller (EC) FW56 and chronic progressor (CP) CR540, assessed by staining with HLA class I tetramers. Numbers adjacent to outlined areas indicate percent CD8+ T cells specific for the HLA-B*27–KK10 tetramer (KK10 tet). (b) Frequency of cells positive for the HLA-B*27–KK10 tetramer among bulk CD8+ T cells from controllers (n = 5) and progressors (n = 5). *P = 0.7531 (Mann-Whitney test). (c) Responses of KK10-specific CD8+ T cells among PBMCs from controllers (n = 5) and progressors (n = 5), analyzed ex vivo by ELISPOT assay of IFN-γ after stimulation with KK10 peptide and calculated as spot-forming cells (SFC) per 1 × 106 PBMCs. *P = 0.7383 (Mann-Whitney test). Each symbol represents an individual subject; small horizontal lines indicate the mean. Data are representative of two experiments.

b ca

100 CTR22

ECCP

CTR40CTR203FW5613587CR338CR420CR540FEN338222IF

N-γ

pro

duct

ion

(%)

Cyt

okin

e-se

cret

ing

KK

10+ c

ells

(%

)

75

50

25

Peptide concentration (pM)101 102 103 104 105 106 107

0

403020

10

5

5 4 3 2 1

50

0IFN-γTNFIL-2

CD107aCCL4

d 50*

Pro

lifer

atio

n (%

)

40

30

20

EC CP10

Tet

105

CFSE

38.4EC FW56 CP CR540

61.6

0 0

31.9 68.1

0 0

104

103

102

0

1051041031020*

*

**

Figure 2 Functional characteristics of KK10-specific CD8+ T cells. (a) Functional avidity of KK10-specific CD8+ T cells from various subjects (identifiers in key), assessed by peptide ‘titration’ of PBMCs in an ELISPOT assay of IFN-γ. P = 0.4678 (Mann-Whitney test). (b) Expression of IFN-γ, tumor necrosis factor (TNF), IL-2, CD107a and CCL4 (MIP-1β) by KK10-specific CD8+ T cells from HLA-B*2705+ elite controllers (n = 5) and chronic progressors (n = 5) after stimulation of PBMCs with KK10, presented as the proportion (bar height) of subpopulations of KK10-specific cells expressing various combinations (below graph) of effector molecules (numbers in colored rectangles along horizontal axis indicate number of effector molecules assessed in group of bars encompassed by rectangle).*P < 0.05 (Mann-Whitney test). (c) Proliferation of KK10-specific CD8+ T cells at day 7 after stimulation of bulk CD8+ T cells from a controller (FW56) and a progressor (CR540) with HIV-1-infected, HLA-B*2705-expressing GXR cells, assessed by staining with the cytosolic dye CFSE (proliferation) and with the HLA-B*27–KK10 tetramer (KK10-specific cells). Numbers in quadrants indicate percent CFSElo (proliferating) cells (top left) and CFSEhi (nonproliferating) cells (top right). (d) Proliferative capacity of KK10-specific CD8+ T cells from controllers (n = 5) and progressors (n = 5), measured by flow cytometry of CFSE intensity. Each symbol represents an individual subject; small horizontal lines indicate the mean. *P = 0.2222 (Mann-Whitney test). Data are representative of two experiments (error bars (b), s.d.).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunology  aDVaNCE ONLINE PUBLICaTION �

A rt i c l e s

Having shown that this assay provided evidence of active viral neu-tralization by ex vivo KK10-specific CD8+ T cells in HLA-B*2705+ elite controllers and that this assay was sensitive to KK10 epitope specificity and HLA-B*2705 expression, we next evaluated CD8+ T cells from the progressors. Infected GXR cells expressing HLA-B*2705 were inhi-bited by the addition of CD8+ T cells from progressors (Fig. 3c), but to a much smaller degree than had been noted with cells from con-trollers. We obtained this result despite the finding of no quantitative differences in KK10-specific cells in controllers and progressors, as shown by tetramer staining and ELISPOT assay for IFN-γ (Fig. 1).

We next extended the studies reported above to determine the abil-ity of KK10-specific CD8+ T cell responses in controllers and pro-gressors to recognize viral variants known to arise in vivo that are able to escape the immune response36. Computational studies have suggested that protective HLA alleles are associated with enhanced cross-reactivity25, but this has not been evaluated, to our knowledge, in the context of a single allele encoding HLA class I and single viral epitope in a comparison of controllers and progressors. For this, we analyzed by flow cytometry the recognition of wild-type HIV-1 and viral variants by KK10-specific CD8+ T cells among bulk CD8+ T cells from controllers and progressors. We evaluated the proportion of GFP+ cells after infecting HLA-B*2705-expressing GXR cells. We again used the GXR cell assay system rather than autologous CD4+ T cells to overcome the problem of outgrowth of autologous virus from the chronic progressors and avoid potential variability of CD4+ T cell responses among study subjects.

We assessed the recognition of HIV-1 strain NL4-3 with the wild-type KK10 epitope, as well as HIV-1 NL4-3 variants with mutant KK10 sequences (Table 2), by KK10-specific CD8+ T cells from elite controllers and chronic progressors (results for HLA-B*2705+ elite controller FW56 and chronic progressor CR540, Fig. 4a). We found essentially complete inhibition of the replication of wild-type HIV-1 NL4-3 and broad recognition of viral variants by KK10- specific CD8+ T cells from controller FW56, as well as considerable inhibition of the typical early L6M mutant virus, which we did not detect in this subject. In contrast, inhibition of both of these targets by CD8+ T cells from progressor CR540 was present but minimal. Although peak infection with the other mutant viruses was less than that noted with wild-type virus or the L6M mutant virus, CD8+ T cells from controller FW56 inhibited all variants, whereas CD8+ T cells from progressor CR540 were ineffective, even though both subjects had similar proportions of KK10-specific effector cells, as quantified by tetramer staining (Table 1), and proliferative

capacity in response to HIV-1-infected, HLA-B*2705-expressing GXR cells (Fig. 2c). We consistently observed the superiority of HIV-1-specific CD8+ T cells from elite controllers in antiviral efficacy against wild-type HIV-1 NL4-3 and viral variants when we extended these detailed studies to all the elite controllers and chronic progressors (Fig. 4b). In addition, bulk CD8+ T cells from HIV-1− people (which included HLA-B*2705+ donors) did not inhibit viral replication in HLA-B*2705-expressing GXR cells (Fig. 4b). We obtained similar results in terms of killing efficacy against the same HIV-1-infected, HLA-B*2705-expressing GXR cells by standard chromium-release assays of cells from the control-lers and progressors (Fig. 4c).

Together these data indicated a clear distinction between elite controllers and chronic progressors in HLA-B*2705-restricted CD8+ T cells targeting the KK10 epitope based on the potency and

EC FW56

EC FW56

106

a

b

c

105

104

103

102p24

(pg/

ml)

101

106

105

104

103

102

101

20

10

5GF

P+ c

ells

(%

)

GF

P+ c

ells

(%

)

GF

P+ c

ells

(%

)

0

15

20

101.5

1.0

0.5

0

CTR22

CR540

CR420

CR338

FEN3382

22HIV

–

CTR40

CTR203

FW56

1358

7HIV

–

CTR22

CTR40

CTR203

FW56

1358

7HIV

–

15

20

25

10

5

0

15

GF

P+ c

ells

(%

) 20

25

10

5

0

15

p24

(pg/

ml)

2 3 4

Time after infection (d)

Time after infection (d)

5 6 7

2 3 4

Time after infection (d)

CP FEN33

5 6 7

2 3 4 5 6 7

+ bulk CD8+ + KK10-depleted CD8+GXRB27 + virusGXRB27 only

+ bulk CD8+GXRB27 + virusGXRB27 only

+ bulk CD8+ + KK10-depleted CD8+CD4+ + virusCD4+ only

Table 2 HLA-B*2705-restricted KK10 and HLA-B*57-restricted TW10 sequencesViral isolate Epitope sequence

KK10 wild-type KRWIILGLNKL6M KRWIIMGLNKR2T KTWIILGLNKR2TL6M KTWIIMGLNKR2Q KQWIILGLNKR2QL6M KQWIIMGLNKTW10 wild-type TSTLQEQIGWT242N TSNLQEQIGWG248A TSTLQEQIAWG248D TSTLQEQIDW

Sequences of the HLA-B*2705-restricted KK10 epitope and HLA-B*57-restricted TW10 epitope of wild-type HIV-1 NL4-3 and its variants (mutant residues bolded). Derived from the Los Alamos Immunology Database.

Figure 3 Neutralization of virus ex vivo by KK10-specific CD8+ T cells. (a) Production of p24 antigen by autologous CD4+ T cells left uninfected (CD4+ only; negative control) or infected with wild-type HIV-1 NL4-3 and assessed alone (CD4+ + virus; positive control) or in the presence of bulk CD8+ T cells (+ bulk CD8+) or CD8+ T cell populations depleted of KK10-specific cells (+ KK10-depleted CD8+), presented for controller FW56 over 7 d (left) and for all controllers (n = 5) and an HIV− subject at day 7 (right). (b) Viral replication in HLA-B*2705-expressing GXR cells left uninfected (GXRB27 only) or infected with HIV-1 and cultured alone (GXRB27 + virus) or in the presence of bulk or depleted CD8+ T cells (as in a), analyzed by flow cytometry and presented as GFP+ cells (infected GXR cells with viral replication) in controller FW56 over 7 d (left) and for all controllers (n = 5) and an HIV− subject at day 7 (right). (c) Viral replication in HLA-B*2705-expressing GXR cells left uninfected or infected and cultured alone (as in b) or in the presence of bulk CD8+ T cells (as in b), analyzed by flow cytometry and presented as GFP+ cells from progressor FEN33 over 7 d (left) and for all progressors (n = 5) and an HIV− subject at day 7 (right). P < 0.0001 (Mann-Whitney test). Data are representative of two experiments.

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�  aDVaNCE ONLINE PUBLICaTION nature immunology

A rt i c l e s

cross-reactivity of the recognition of cells infected with wild-type HIV-1 and HIV-1 containing naturally arising substitutions in the KK10 epitope. In addition, we found that the CDR3 sequences of KK10-specific clonotypes were significantly closer to germline sequences in controllers than in progressors (P = 0.001; Supplementary Fig. 2), which would confer a greater ability to recognize epitope variants39. These functional data were also consistent with computational mod-eling showing that thymic selection in the context of protective HLA alleles is more likely to generate a cross-reactive CD8+ T cell rep-ertoire that targets mutant viral epitopes, thereby contributing to improved control of a highly variable pathogen25.

Antiviral efficacy of KK10-specific clonotypesThe data reported above indicated that there were differences between controllers and progressors in the potency and cross-reactivity of recognition of wild-type HIV-1 and viral variants in the KK10-specific CD8+ T cell responses, which suggested that the fine specificity of the TCR might be modulating these effects. Given that we had observed different clonotypes in the tetramer-positive populations in these subjects (Table 1), we next sought to determine clonotypic antiviral efficacy. We therefore cloned KK10-specific CD8+ T cells by limiting dilution from the sorted KK10 tetramer–positive cells from three elite controllers (CTR203, FW56 and CTR40) and two chronic progressors (CR540 and CR420) and then determined clonotypes by TCR sequencing (Table 3). In each of

these people, we cloned the dominant clonotypes identified in vivo, and in three of the five, we were able to generate multiple clones.

We assessed the ability of the clonotypic KK10-specific CD8+ T cells to kill HIV-infected target cells in a standard chromium-release assay and assessed their ability to inhibit viral replication, again using the HLA-B*2705-expressing GXR cells. For controller CTR203, we were able to establish clones representing five of the six clonotypes detected in peripheral blood. We observed considerable variation in the ability of these clonotypes to recognize HIV-1 and viral variants (Fig. 5) that ranged from broad recognition of the variants by the two most dominant CTR203 clonotypes to weak and narrow recognition by all three subdominant CTR203 TCR vari-ants. Extending this to clones from five subjects, we found that the most effective clonotypes were the immunodominant in vivo clono-types from the controllers, including the two codominant responses in subject CTR203, the two codominant responses in subject FW56 and the one dominant response in subject CTR40. These were sig-nificantly more potent at viral recognition than were the immuno-dominant clones from the progressors. We observed codominant

25

GF

P+ c

ells

(%

)

a20

15

10

5

0No CD8+ cells FW56 CD8+ cells CR540 CD8+ cells

WTL6MR2TR2TL6MR2QR2QL6M

Inhi

bitio

n (%

)

b100 *

*

**

**

ECCPHIV–

75

50

25

0

Spe

ci�c

lysi

s (%

)

c WTL6MR2TR2TL6MR2QR2QL6M

70

60

50

40

30

20

10

0R2QL6MR2QR2TL6MR2TL6MWT FEN33CR540FW56CTR203

Figure 4 Recognition of viral variants by KK10-specific CD8+ T cells. (a) Replication of wild-type HIV-1 NL4-3 (WT) and various HIV-1 NL4-3 variants (Table 2 and key) ex vivo in HLA-B*2705-expressing GXR cells alone (No CD8+ cells) or in the presence of CD8+ T cells isolated from a controller (FW56 CD8+ cells) or a progressor (CR540 CD8+ cells) at an effector cell/target cell ratio of 1:1, presented (as in Fig. 3b) as GFP+ cells at day 7 of culture. (b) Inhibition of the replication of wild-type HIV-1 NL4-3 and various HIV-1 NL4-3 variants (horizontal axis) ex vivo by CD8+ T cells from controllers (n = 5), progressors (n = 5) and HIV-1− people (n = 12 total, including n = 4 HLA-B*2705+ donors). *P < 0.0001 (Mann-Whitney test). (c) Lysis of live, HIV-infected (GFP+), HLA-B*2705-expressing GXR cells (sorted as viable infected cells by flow cytometry after infection for 5 d by virus in key) ex vivo by CD8+ T cells from controllers (CTR203 and FW56) and progressors (CR540 and FEN33) at an effector cell/target cell ratio of 10:1, assessed by a 4-hour chromium-release assay. Data are representative of two experiments.

Table 3 Clonotypes of HLA-B*2705-restricted KK10-specific CD8+ T cell clonesClone Vβ CDR3 J region

CTR203 S-C003 25.1 CASSEADFEAF 1.1CTR203 S-T001 18 CASSPGQFSHEQY 2.7CTR203 L015 27 CASSARTGELF 2.2CTR203 S-C007 20.1 CSARDGGEQY 2.7CTR203 S-T002 7.9 CASSLDRLEQF 2.1FW56 B3 4.3 CASRPGLASNEQF 2.1FW56 B5 6.5 CASRPGQGATEAF 1.1FW56 B6 20.1 CSARDRGTREVADNYGYT 1.2CTR40 H08 7.2 CASSLSGRWSTDTQY 2.3CR540 002 5.6 CASGGGTVYEQY 2.7CR540 013 2 CASSAGPGQYGNTIY 1.3CR540 015 21 CASTNRGSEQY 2.7CR420 C05 20.1 CSAREGVEGYT 1.2

TCR β-chain variable region, CDR3 sequence and joining region (as in Table 1).

100

75

KK10 WT L6M R2T R2TL6M

CTR203

50

Spe

ci�c

lysi

s

25

S-C003 S-T001 S-C007 S-T002 B3

FW56 CTR40 CR540 CR420

B5 B6 H08 002 013 015 C05L015

BV25.1

/J1.1

BV18/J2

.7

BV27/J2

.2

BV20.1

/J2.7

BV7.9/

J2.1

BV4.3/

J2.1

BV6.5/

J1.1

BV20.1

/J1.2

BV7.2/

J2.3

BV5.6/

J2.7

BV2/J1

.3

BV21/J2

.7

BV20.1

/1.2

Clone

Clonotype

Freq 13/37 11/37 2/37 6/37 3/37 16/31 10/31 3/31 11/21 8/34 16/34 6/34 21/300

Figure 5 Differences in the antiviral efficacy of clonotypes specific for HLA-B*27–KK10. Lysis of HIV-1-infected, HLA-B*2705-expressing GXR cells (as in Fig. 4c) by KK10-specific clonotypes from controllers (CTR203, FW56 and CTR40) and progressors (CR540 and CR420), assessed by chromium-release assay at an effector cell/target cell ratio of 1:1. Freq, frequency of clonotype among KK10-specific CD8+ T cell populations. P = 0.005, dominant clonotypes from controllers versus those from progressors (Mann-Whitney test). Data are representative of two experiments (error bars, s.d.).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunology  aDVaNCE ONLINE PUBLICaTION �

A rt i c l e s

clonotype TCRBV4-3 in controller FW56, which showed efficient recognition of wild-type virus and had less robust activity than codominant FW56 clonotype TCRBV6-5 had against the L6M vari-ant and was unable to recognize any of the other variants, at a low frequency in progressors CR540, CR420 and 8222 in the context of different sequences of CDR3 or joining segments.

Subdominant clonotypes from the elite controllers included CRBV27TCRBJ2-2, TCRBV20-1TCRBJ2-7 and TCRBV20-1TCRBJ1-2, all of which were associated with inferior recognition of wild-type virus and the L6M variant and showed the least efficacy against the other viral variants. However, these less-effective clonotypes were dominantly selected by progressors CR338, 8222 and CR420. They were rearranged with variant CDR3 or joining segments but were likewise less efficient at recognizing HIV-1-infected cells. Moreover, none of the subdomi-nant clonotypes, whether from controllers or progressors, were able to efficiently recognize HIV-1-infected cells or to inhibit viral replication. We obtained consistent results in terms of the ability of individual clono-types to inhibit viral replication (Supplementary Fig. 3). Together these data indicated that differences in the antiviral efficacy of KK10-specific CD8+ T cells in HIV-1-infected people were defined by the dominance of clonotypes that conferred distinct antiviral potential on CD8+ T cells.

Antiviral efficacy of TW10-specific clonotypesWe further examined the effect of individual TCR clonotypes on anti-viral efficacy with HLA-B*5701-restricted CD8+ T cell clones spe-cific for the epitope TW10 of Gag in a chromium-release assay with

HIV-1-infected, HLA-B*5701-expressing GXR cells (Fig. 6). We gene-rated HLA-B*5701-restricted TW10-specific CD8+ T cell clones by limiting dilution of cells sorted from HLA-B*57+ elite controllers (CTR53 and CR462) and a chronic progressor (CR555) through the use of an HLA-B*57–TW10 tetramer and then clonotypically assessed these by TCR sequencing. Again, we observed that variant clonotypes had differences in antiviral efficacy and that, overall, clones from the elite controllers were more potent and cross-reactive in recognition of HIV-1 and viral variants than were clones from the progressor (Fig. 6). Together these data indicated that the difference between the controllers and progressors of the same epitope-specific CD8+ T cell responses in the recognition of HIV-1 and viral variants was related to distinct TCR clonotypes selected during natural infection.

Lytic granule loading and delivery by clonotypesThe data reported above indicated clonotype-specific differences in antiviral function, which offered the opportunity to define the

EC CTR53 EC CR462 CP CR555

001 004 005 001 003 010 012 021CloneClonotype

0

25

50

Spe

cific

lysi

s (%

)

75

100

BV5.1/

J1

BV28/J2

.1

BV9.1/

J2.5

BV15/J1

.3

BV19/J2

.2

BV9.1/

J2.1

BV10.3

/J1.1

BV20.1

/J2.3

TW10 WT T242N G248A G248DFigure 6 Differences in the antiviral efficacy of clonotypes specific for HLA-B*57–TW10. Lysis of HIV-1-infected, HLA-B*2705-expressing GXR cells (as in Fig. 4c) by CD8+ T cell clones specific for HLA-B*5701–TW10, generated from HLA-B*57+ elite controllers (CTR53 and CR462) and a chronic progressor (CR555), assessed by chromium-release assay at an effector cell/target cell ratio of 1:1. Data are representative of two experiments (error bars, s.d.).

a

S-C00

3

S-T00

1Clone

Freg

Per

forin

-exp

ress

ion

KK

10+ c

ells

(%

)

13/3

70

25

50

75

100 EC CTR203 CP CR450

11/3

72/

376/

373/

378/

3416

/346/

34

L015

S-C00

7

S-T00

200

201

301

5

bDIC + GFP

CTR203S-C003

CTR203S-C007

F-actin Perforin Merge

d

Inte

nsity

per

GX

R c

ell

(104 A

U)

20

15

10

5

S-C003 S-C007 013 015

CR540CTR203

T

T

GXR

GXR

cIn

tens

ity p

er T

cel

l

(105 A

U) 10

5

0S-C003 S-C007 013 015

CR540CTR203

**

*

Figure 7 Differences in the loading and delivery of perforin by clonotypes. (a) Perforin expression by KK10-specific (KK10+) cells after 3 d of culture with HIV-1-infected HLA-B*2705-expressing GXR cells, assessed by flow cytometry and presented as frequency after subtraction of background frequency obtained by incubation with uninfected target cells. P < 0.0001, dominant clonotypes from controllers versus subdominant clonotypes from the controllers and all clonotypes from progressors (Mann-Whitney test). (b) Microcopy of the release of perforin from effector cells of the dominant clonotype S-C003 (top) and subdominant clonotype S-C007 (bottom) from controller CTR203 after 30 min of incubation with HIV-1-infected HLA-B*2705-encoding GXR cells (green; target cells), presented as differential interference contrast images (DIC + GFP; left) and as z-series confocal microscopy with projected serial confocal sections through conjugations between cells, by staining of F-actin with phalloidin–Alexa Fluor 647 (red) and of perforin with primary antibody to perforin, followed by Alexa Fluor 568–conjugated secondary monoclonal antibody (purple). Far right, merged overlays. Scale bars, 3.0 µm. (c) Intensity of perforin staining in T cells of dominant clonotypes (S-C003 from controller CTR203, and 013 from progressor CR540) and subdominant clonotypes (S-C007 from CTR203, and 015 from CR540) after exposure to HIV-1-infected GXR cells. *P < 0.0001 (Mann-Whitney test). (d) Intensity of perforin staining in GXR cells after exposure to the clonotypes in c. Each symbol (c,d) represents an individual cell; small horizontal lines indicate the mean (and s.e.m.). Data are representative of two experiments.

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�  aDVaNCE ONLINE PUBLICaTION nature immunology

A rt i c l e s

mechanisms that account for these phenotypes. We next determined the effect of TCR clonotype on the loading of lytic granules after recognition of HIV-1-infected target cells and the ability of clonotype-specific T cells to deliver perforin to infected target cells. We mea-sured by flow cytometry the expression of perforin and granzyme B by various clonotypes12,13. After culture for 3 d with KK10-specific, HLA-B*2705-expressing GXR cells infected with wild-type HIV-1, dominant clonotypes from controllers had efficient expression of perforin. In contrast, subdominant clonotypes from controllers and all the clonotypes from progressors were significantly less efficient at expressing perforin than were dominant clonotypes from controllers (Fig. 7a). We obtained similar results for the expression of granzyme B (P < 0.0001; Supplementary Fig. 4).

We also examined by confocal microscopy effector cell–target cell conjugation and the loading and delivery of granules40. After incubation for 30 min with HIV-infected, HLA-B*2705-expressing GXR cells, more perforin was polarized to synapses and released into target cells for the inhibitory clonotypes than for the non-inhibitory clonotypes (Fig. 7b). Extended quantitative analysis of perforin loading and delivery showed that significantly less per-forin was delivered to synapses (Fig. 7c) and released into target cells (Fig. 7d) by subdominant clonotypes from controllers and all the clonotypes from progressors than by dominant clonotypes from controllers. As a control, we confirmed that endogenous perforin was undetectable in HIV-1-infected GXR cells (Supplementary Fig. 5). We observed similar basal amounts of perforin in inhibi-tory and noninhibitory clonotypes without stimulation of target cells (P = 0.96; Supplementary Fig. 6), which indicated that the func-tionally effective clones rapidly upregulated perforin loading after the recognition of cognate antigens. These data indicated that TCR clonotypes associated with enhanced ability to inhibit HIV-1 replica-tion did so by rapidly upregulating lytic granules at immunological synapses after engagement of a target cell and delivery of these granules into the infected cell.

DISCUSSIONAlleles encoding certain HLA-B types are associated with enhanced control of viremia in HIV-1-infected people3–5, and this effect maps to specific host amino acids in the HLA-B peptide-binding groove6. However, most people with such so-called ‘protective alleles’, such as those encoding HLA HLA-B*27 and HLA-B*57, experience pro-gressive infection26. To address the basis of these differences in out-come, we compared CD8+ T cell responses to immunodominant viral epitopes in treatment-naive elite controllers and chronic progressors. To limit the number of potential confounding variables, we stud-ied only subjects with wild-type sequences of the respective T cell epitopes in the virus in their plasma and provirus in their PBMCs and with the same HLA class I–restricting allele. In this setting, in which contemporaneous escape from the immune response is not a confounding issue, we found that CD8+ T cell responses by HIV-1 controllers were more potent at inhibiting HIV-1 replication than were those of progressors that targeted the same epitopes and were better able to ‘cross-recognize’ HIV-1 viral variants that typically arise in vivo. Moreover, these effects were associated with a unique ability of the dominant TCR clonotypes to upregulate perforin and granzyme B, which provides a mechanistic explanation for the divergent disease outcomes in people with protective HLA alleles.

Many characteristics of CD8+ T cells have been reported to be associated with enhanced control of viremia, including differences in polyfunctionality, proliferative capacity and functional avidity of KK10-specific CD8+ T cells. In the carefully controlled comparative

studies done here of a small cohort of well-pedigreed control-lers and progressors expressing protective alleles, none of those reported associations reached statistical significance. In contrast, the ability of both bulk CD8+ T cells as well as epitope-specific TCR clonotypes to inhibit viral replication, cross-recognize viral variants and upregulate perforin and granzyme B, which are probably the most important in vivo functions of these cells, was highly signifi-cant. Overall, our study has linked the antiviral efficacy of the two most protective HLA class I molecules to CD8+ T cell clonotypes selected during natural infection with HIV-1 and has demonstrated that TCR rearrangement modulated the effect of protective alleles on disease outcome.

Our study is distinct from other reports of TCR clonotype use by KK10-specific CD8+ T cells in people who express HLA-B*27 in that we selected the subjects by ‘extremes’ of viral load. As noted in other published studies28,29, we observed considerable diversity in clonotype recruitment and CDR3 motifs in KK10-specific CD8+ T cell populations, as well as a dominance of clonotypes in progressors who were unable to cross-recognize the L6M mutant. The mutation that results in this mutant is known to occur early during the course of HIV-1 infection with little or no effect on peptide processing27, bind-ing of peptide to HLA-B*27 (ref. 27) recognition by the TCR41 or viral fitness36, but it is an important intermediate mutation on the path to complete escape of the immune response. Although we detected such ineffective responses at the clonal level in both controllers and progressors in our cohort, HLA-B*27+ elite controllers had domi-nant clonotypes that targeted not only the L6M mutant but also other mutants, including substitution at position 2 of KK10 alone or in combination with the L6M substitution, which diminishes the bind-ing of peptide to HLA-B*27 and impairs viral replication36. Thus, the TCR clonotypes in controllers comprised effective and ineffec-tive clonotypes, whereas those in progressors were limited to less- effective clonotypes.

In contrast to some other published reports42,43, we found no sig-nificant difference between the two groups in the use of ‘public’ clono-types (defined as those that expressed identical TCR β-chain amino acid sequences and recurred in many people), despite significant dif-ferences in plasma viremia and the fact that ‘public’ clonotypes do not seem to dominate among controllers29. However, controllers used TCR β-chain clonotypes with sequences encoding CDR3 that were significantly more ‘germline-like’ than those used by the progressors. The lower number of nucleotide additions in the germline-like genes encoding CDR3 is a hallmark of clonotypes found at high frequency in naive and memory T cell pools and also shared by many people44. The mechanism underlying this advantage bestowed by germline-like genes encoding CDR3 may be related to higher precursor fre-quency and/or greater ability to recognize mutational variants of the epitope39,42. Furthermore, the most effective clonotypes, in terms of viral inhibition and cytotoxic recognition of wild-type and vari-ant viruses, were dominantly selected in vivo by the controllers but were either absent or subdominantly selected in KK10-specific CD8+ T cell populations of the progressors. In contrast, the clonotypes associated with inferior recognition of wild-type virus and the least efficacy against the viral variants were dominantly selected by the progressors but subdominantly selected by the controllers.

Although our study has clearly shown that the TCR modulated the protective effect of HLA molecules, it has many limitations. Our HLA-B*2705 studies were limited to only five controllers and five progressors, and many published reported associations with viral control did not reach statistical significance in this small study group. Nevertheless, even with these small numbers, the results showing that

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunology  aDVaNCE ONLINE PUBLICaTION �

A rt i c l e s

TCR clonotypes modulated the protective effect of HLA-B were of high statistical significance in demonstrating greater cytotoxic killing and greater cross-reactivity by the dominant clonotypes in controllers than by those in progressors. We were not able to generate CD8+ T cell clones that represented all detectable TCR clonotypes in all people, but in one subject we were able to test five of six clonotypes in vivo that represented 90% of the detectable TCR diversity in that subject. Moreover, all controllers evaluated at the clonal level had dominant TCR clonotypes that were highly effective, whereas these were absent in both dominant and subdominant clones established in the progres-sors. Of note, the important role of TCR clonotypes in distinguishing viral control from lack of viral control was probably mediated by direct cytotoxicity of HIV-1-infected cells, with perforin upregulation noted within 30 min of recognition of the cognate epitope and not requiring proliferation of CD8+ T cells. The data presented here contrast with the well-documented differences between nonprogressors and progres-sors in the proliferative capacity of CD8+ T cells12, probably because of the way proliferation was measured. In our study here, we used HIV-1-infected GXR cells expressing HLA-B*2705 as stimulator cells in culture with bulk CD8+ T cells and observed similar proliferative capacity for HLA-B*27+ KK10-specific CD8+ T cells from controllers and progressors. This could be explained by the similar abundance of CD4+ T cells in the progressors and controllers, the properties of regu-latory T cells in people who express alleles encoding protective HLA molecules17 or compensation for in vivo impaired CD4+ helper T cell function in progressors by cytokines produced by the cell line used for stimulation. Finally, whether our results can be extrapolated to other protective alleles and other epitopes will require additional study.

Together our data have indicated that TCR use modulates virus-inhibitory capacity and recognition of naturally arising HIV-1 vari-ants and thus modulates the effect of protective HLA alleles. Our data have suggested that TCR clonotypes that inhibit viral replication and confer cross-recognition of viral epitope variants that can eventually arise in vivo may be critical to long-term control of viremia. Efforts to define the factors that contribute to junctional rearrangement of more effective TCRs may be of critical importance for the design of T cell vaccines and therapeutic strategies for highly variable patho-gens such as HIV-1.

METHODSMethods and any associated references are available in the online version of the paper.

Note: Supplementary information is available in the online version of the paper.

ACKNoWLEDGMENTSWe thank J. Wong (Massachusetts General Hospital) for monoclonal antibody 12F6 to CD3 and monoclonal antibody CD3:8 bispecific for CD3 and CD8; and all study participants for their contributions. Supported by the Harvard University Center for AIDS Research (5 P30 AI060354-04), the Bill and Melinda Gates Foundation (B.D.W. and D.C.D.), the Doris Duke Charitable Foundation (B.D.W.), the US National Institutes of Health (AI030914 to B.D.W. and AI074415 to T.M.A.), the Howard Hughes Medical Institute (B.D.W.), the Mark and Lisa Schwartz Foundation (B.D.W.), the Intramural Research Program and the Office of AIDS Research of the US National Institutes of Health (D.C.D. and S.D.), the Canadian Institutes for Health Research (New Investigator Award to Z.L.B.) and the Canada Research Chair in Viral Pathogenesis and Immunity (M.A.B.).

AUTHoR CoNTRIBUTIoNSH.C. was responsible for the overall conduct of the study, under the supervision of B.D.W.; H.C., Z.M.N. and B.D.W. contributed to the experimental design; H.C., Z.M.N., L.C.P. and J.W.F. did the experiments and analyzed the data; S.D. and D.C.D. did germline analyses; T.M., Z.L.B., K.T.C. and J.S. did virus sequencing; M.A.B., A.S. and T.M.A. constructed HIV-1 variants and GXR cell lines; D.L. and

D.E.K. did the imaging experiments; T.D.C. and X.G.Y. helped with TCR sequencing; F.P., A.P.-T. and I.T. provided clinical samples; and H.C., Z.M.N. and B.D.W. wrote the paper and all authors contributed to revisions.

CoMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

Published online at http://www.nature.com/doifinder/10.1038/ni.2342. reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Migueles, S.A. & Connors, M. Long-term nonprogressive disease among untreated HIV-infected individuals. J. Am. Med. Assoc. 304, 194–201 (2010).

2. Deeks, S.G. & Walker, B.D. Human immunodeficiency virus controllers: mechanisms of durable virus control in the absence of antiretroviral therapy. Immunity 27, 406–416 (2007).

3. Kaslow, R.A. et al. Influence of combinations of human major histocompatibility complex genes on the course of HIV-1 infection. Nat. Med. 2, 405–411 (1996).

4. Migueles, S.A. et al. HLA B*5701 is highly associated with restriction of virus replication in a subgroup of HIV-infected long term nonprogressors. Proc. Natl. Acad. Sci. USA 97, 2709–2714 (2000).

5. Carrington, M. & O’Brien, S.J. The influence of HLA genotype on AIDS. Annu. Rev. Med. 54, 535–551 (2003).

6. Pereyra, F. et al. The major genetic determinants of HIV-1 control affect HLA class I peptide presentation. Science 330, 1551–1557 (2010).

7. Betts, M.R. et al. Analysis of total human immunodeficiency virus (HIV)-specific CD4+ and CD8+ T-cell responses: relationship to viral load in untreated HIV infection. J. Virol. 75, 11983–11991 (2001).

8. Addo, M.M. et al. Comprehensive epitope analysis of human immunodeficiency virus type 1 (HIV-1)-specific T-cell responses directed against the entire expressed HIV-1 genome demonstrate broadly directed responses, but no correlation to viral load. J. Virol. 77, 2081–2092 (2003).

9. Betts, M.R. et al. HIV nonprogressors preferentially maintain highly functional HIV-specific CD8+ T cells. Blood 107, 4781–4789 (2006).

10. Almeida, J.R. et al. Antigen sensitivity is a major determinant of CD8+ T-cell polyfunctionality and HIV-suppressive activity. Blood 113, 6351–6360 (2009).

11. Bennett, M.S., Ng, H.L., Dagarag, M., Ali, A. & Yang, O.O. Epitope-dependent avidity thresholds for cytotoxic T-lymphocyte clearance of virus-infected cells. J. Virol. 81, 4973–4980 (2007).

12. Migueles, S.A. et al. HIV-specific CD8+ T cell proliferation is coupled to perforin expression and is maintained in nonprogressors. Nat. Immunol. 3, 1061–1068 (2002).

13. Migueles, S.A. et al. Lytic granule loading of CD8+ T cells is required for HIV-infected cell elimination associated with immune control. Immunity 29, 1009–1021 (2008).

14. Hersperger, A.R. et al. Perforin expression directly ex vivo by HIV-specific CD8 T-cells is a correlate of HIV elite control. PLoS Pathog. 6, e1000917 (2010).

15. Dahirel, V. et al. Coordinate linkage of HIV evolution reveals regions of immunological vulnerability. Proc. Natl. Acad. Sci. USA 108, 11530–11535 (2011).

16. Kiepiela, P. et al. CD8+ T-cell responses to different HIV proteins have discordant associations with viral load. Nat. Med. 13, 46–53 (2007).

17. Elahi, S. et al. Protective HIV-specific CD8+ T cells evade Treg cell suppression. Nat. Med. 17, 989–995 (2011).

18. Kaufmann, D.E. et al. Upregulation of CTLA-4 by HIV-specific CD4+ T cells correlates with disease progression and defines a reversible immune dysfunction. Nat. Immunol. 8, 1246–1254 (2007).

19. Day, C.L. et al. PD-1 expression on HIV-specific T cells is associated with T-cell exhaustion and disease progression. Nature 443, 350–354 (2006).

20. Pérez, C.L. et al. Broadly immunogenic HLA class I supertype-restricted elite CTL epitopes recognized in a diverse population infected with different HIV-1 subtypes. J. Immunol. 180, 5092–5100 (2008).

21. Miura, T. et al. HLA-B57/B*5801 human immunodeficiency virus type 1 elite controllers select for rare Gag variants associated with reduced viral replication capacity and strong cytotoxic T-lymphotye recognition. J. Virol. 83, 2743–2755 (2009).

22. Lassen, K.G. et al. Elite suppressor derived HIV-1 envelope glycoproteins exhibit reduced entry efficiency and kinetics. PLoS Pathog. 5, e1000377 (2009).

23. Goulder, P.J. et al. Late escape from an immunodominant cytotoxic T-lymphocyte response associated with progression to AIDS. Nat. Med. 3, 212–217 (1997).

24. Lee, K.-H. et al. The immunological synapse balances T cell receptor signaling and degradation. Science 302, 1218–1222 (2003).

25. Kosmrlj, A. et al. Effects of thymic selection of the T-cell repertoire on HLA class I-associated control of HIV infection. Nature 465, 350–354 (2010).

26. Pereyra, F. et al. Genetic and immunologic heterogeneity among persons who control HIV infection in the absence of therapy. J. Infect. Dis. 197, 563–571 (2008).

27. Goulder, P.J. et al. Evolution and transmission of stable CTL escape mutations in HIV infection. Nature 412, 334–338 (2001).

28. Almeida, J.R. et al. Superior control of HIV-1 replication by CD8+ T cells is reflected by their avidity, polyfunctionality, and clonal turnover. J. Exp. Med. 204, 2473–2485 (2007).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

�0  aDVaNCE ONLINE PUBLICaTION nature immunology

A rt i c l e s

29. Mendoza, D. et al. HLA B*5701+ long-term nonprogressors/elite controllers are not distinguished from progressors by the clonal composition of HIV-specific CD8+ T-cells. J. Virol. 86, 4014–4018 (2012).

30. van Bockel, D.J. et al. Persistent survival of prevalent clonotypes within an immunodominant HIV Gag-specific CD8+ T cell response. J. Immunol. 186, 359–371 (2011).

31. Iglesias, M.C. et al. Escape from highly effective public CD8+ T-cell clonotypes by HIV. Blood 118, 2138–2149 (2011).

32. Day, C.L. et al. Proliferative capacity of epitope-specific CD8 T-cell responses is inversely related to viral load in chronic human immunodeficiency virus type 1 infection. J. Virol. 81, 434–438 (2007).

33. Champagne, P. et al. Skewed maturation of memory HIV-specific CD8 T lymphocytes. Nature 410, 106–111 (2001).

34. Hamann, D.r. et al. Phenotypic and functional separation of memory and effector human CD8+ T cells. J. Exp. Med. 186, 1407–1418 (1997).

35. Brockman, M.A., Tanzi, G.O., Walker, B.D. & Allen, T.M. Use of a novel GFP reporter cell line to examine replication capacity of CXCR4- and CCR5-tropic HIV-1 by flow cytometry. J. Virol. Methods 131, 134–142 (2006).

36. Schneidewind, A. et al. Structural and functional constraints limit options for cytotoxic T-lymphocyte escape in the immunodominant HLA-B27-restricted epitope in human immunodeficiency virus type 1 capsid. J. Virol. 82, 5594–5605 (2008).

37. Chen, H. et al. Differential neutralization of human immunodeficiency virus (HIV) replication in autologous CD4 T cells by HIV-specific cytotoxic T lymphocytes. J. Virol. 83, 3138–3149 (2009).

38. Julg, B. et al. Infrequent recovery of HIV from but robust exogenous infection of activated CD4+ T cells in HIV elite controllers. Clin. Infect. Dis. 51, 233–238 (2010).

39. Miles, J.J., Douek, D.C. & Price, D.A. Bias in the αβ T-cell repertoire: implications for disease pathogenesis and vaccination. Immunol. Cell Biol. 89, 375–387 (2011).

40. Liu, D. et al. Integrin-dependent organization and bidirectional vesicular traffic at cytotoxic immune synapses. Immunity 31, 99–109 (2009).

41. Stewart-Jones, G.B.E. et al. Crystal structures and KIR3DL1 recognition of three immunodominant viral peptides complexed to HLA-B*2705. Eur. J. Immunol. 35, 341–351 (2005).

42. Price, D.A. et al. Public clonotype usage identifies protective Gag-specific CD8+ T cell responses in SIV infection. J. Exp. Med. 206, 923–936 (2009).

43. Dong, T. et al. HIV-specific cytotoxic T cells from long-term survivors select a unique T cell receptor. J. Exp. Med. 200, 1547–1557 (2004).

44. Venturi, V. et al. A mechanism for TCR sharing between T cell subsets and individuals revealed by pyrosequencing. J. Immunol. 186, 4285–4294 (2011).

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunologydoi:10.1038/ni.2342

ONLINE METHODSStudy subjects. PBMCs and plasma samples from HIV-1-infected people and HIV-1− people were used in this study according to protocols approved by the Institutional Review Board of the Massachusetts General Hospital. Elite controllers were defined as having a concentration of HIV-1 RNA below the limit of detection for the assay used (for example, <75 RNA copies per ml by branched DNA assay, or <50 copies by ultrasenstive PCR), without antiretro-viral therapy. Treatment-naive chronic progressors in the study had a median viral load of 12,833 copies per ml (range, 4,073–22,094 copies per ml). CD4+ T cell counts, viral loads and HLA types were determined as described26 (characteristics of study subjects, Table 1).

Viruses and synthetic peptides. The chemokine receptor CXCR4-tropic HIV-1 laboratory strain NL4-3 was obtained from the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, US National Institutes of Health. HIV-1 laboratory strain NL4-3 was also modified to express one or more mutations in the gene encoding Gag p24 as described36,45. Peptides corresponding to described opti-mal HIV-1 epitopes and their variants were synthesized at the Massachusetts General Hospital Peptide Core Facility on an automated peptide synthesizer by fluorenylmethoxycarbonyl technology.

Virus sequencing. Nested PCR for viral DNA or RNA was done as described46. PCR fragments were sequenced by population for the identification of regions of sequence variation. All fragments were sequenced bidirectionally on an ABI 3730xl automated sequencer (Applied Biosystems).

ELISPOT assay. IFN-γ was measured by ELISPOT assay as described, with optimally defined epitopes and designated concentrations of peptide8. The density of input cells ranged from 1 × 104 cells per well to 1 × 105 cells per well. For the quantification of specific spot-forming cells, the number of spots in the negative-control wells was subtracted from the number of spots in each experimental well. Responses were considered positive if they had at least three times the mean number of spot-forming cells in the three negative-control wells; positive responses also had to achieve a value of least 50 spot-forming cells per 1 × 106 PBMCs. The magnitude of the epitope-specific response is presented as spot-forming cells per 1 × 106 cells.

Generation of CD8+ T cell clones. PBMCs were stained with fluorophore-conjugated HLA tetramer refolded with epitopic HIV-1 peptides (Beckman Coulter) and fluorophore-labeled antibody to CD8 (anti-CD8; RPA-T8; BD) and anti-CD3 (UCHT1; BD). Tetramer-positive CD8+ cells were sorted on a FACSAria (BD Biosciences) at 70 p.s.i., and single cells were placed into each well of 96-well plates, with irradiated allogeneic PBMCs and monoclonal anti-body 12F6 to CD3 (a gift from J. Wong) as a stimulus for T cell proliferation47. Developing epitope-specific clones were further tested by ELISPOT assay of IFN-γ with optimal epitopes and with tetramer staining. Cloned CD8+ T cells were maintained by restimulation every 14–21 d with monoclonal anti-CD3 and irradiated allogeneic PBMCs in RPMI-1640 medium containing 50 U/ml of recombinant IL-2, as described47.

Sequencing of TCR a- and b-chains. Tetramer-positive CD8+ cells were sorted from PBMCs or cloned CD8+ T cells and mRNA was extracted with the RNeasy mini kit (Qiagen). Anchored RT-PCR was then done with a modi-fied version of the SMART (switching mechanism at 5′ end of RNA transcript) procedure and a 3′-primer for the TCR α- or β-chain constant region (Cα or Cβ) to obtain PCR products containing the Vα or Vβ chain in addition to the CDR3, the Jα or Jβ region and the beginning of the Cα or Cβ region. RT-PCR and sequencing and analysis of genes encoding TCR α- and β-chain were done as described48.

Tetramer staining. Cells were first stained for 15 min with blue viability dye (L-23105; Molecular Probes, Invitrogen), then were stained for 30 min at room temperature with allophycocyanin- or phycoerythrin-conjugated MHC class I tetramer folded with KK10 peptide (Beckman Coulter). Surfaces of cells were then stained with the following antibodies (all from BD Pharmingen): anti–human CD3 (UCHT1), anti–human CD8 (RPA-T8), anti–human CD14

(M5E2), anti–human CD19 (HIB19) and anti–human CD56 (B159). Samples were acquired on an LSRFortessa (BD Biosciences) and data were analyzed with FlowJo software (version 9.0.2).

Surface and intracellular staining. Cells (2 × 106) were stimulated for 1 h at 37 °C and 5% CO2 with KK10 at a final concentration of 20 ng/ml. Phycoerythrin-indodicarbocyanine–conjugated anti-CD107a (H4A3; BD) was also added at the beginning of the stimulation period for analysis of degranulation. After 1 h, 10 ng/ml of brefeldin A (B7651; Sigma) was added and the cells were incubated for another 5 h. After stimulation, surfaces of cells were stained with the following antibodies: V500–anti-CD8 (RPA-T8; BD), allophycocyanin-indotricarbocyanine–anti-CD27 (MT271; BD) and Qdot 605–anti-CD45RA (H100; eBioscience). Cells were then fixed and permeabilized with Cytofix/Cytoperm solution (554722; BD) and stained with phycoerythrin-indotricarbocyanine–anti-IFN-γ (B27; BD), fluorescein isothiocyanate–anti-IL-2 (5344.111; BD), Alexa Fluor 700–anti-TNF (Mab11; BD), phycoerythrin–anti-MIP-1β (D21-1351; BD) and Qdot 655–anti-CD3 (S4.1; Invitrogen). Cells were acquired on an LSRFortessa (BD Biosciences) and data were analyzed with FlowJo software (TreeStar).

Proliferation assay. Primary CD8+ T cells were isolated from PBMCs by nega-tive selection (Dynabeads; Invitrogen) with the proportion of CD3+ CD8+ T cells being >98%, as detected by flow cytometry. Cells were stained for 7 min at 37 °C with 0.35 µM CFSE (carboxyfluorescein diacetate succinimi-dyl ester; Molecular Probes) and then were cultured for 7 d with medium alone or with HIV-1-infected or uninfected HLA-B*27-encoding GXR cells in RPMI-1640 medium in the absence of IL-2. After being labeled with the appropriate tetramer (Beckman Coulter), anti-CD8 (RPA-T8; BD) and anti-CD3 (UCHT1; BD), cells were fixed in 1% paraformaldehyde and analyzed on an LSRII (BD Biosciences).

Chromium release assay. HLA-B*27- or HLA-B*57-expressing GXR cells (which contain a plasmid encoding GFP driven by the long terminal repeat of HIV-1) were constructed as described35,36 and were infected with wild-type HIV-1 or viral variants at the appropriate multiplicity of infection. On day 5 after infection, viable virus-infected cells were sorted on a FACSAria (BD Biosciences) and labeled for 1 h at 37 °C with chromium. Bulk CD8+ T cells isolated from PBMCs by negative selection (Dynabeads; Invitrogen) or CD8+ T cell clones were then added at the appropriate effector cell/target cell ratio, and a standard 4-hour chromium-release assay was done as described49. Percent specific lysis was calculated as follows: [(mean experi-mental c.p.m. − mean spontaneous c.p.m.) / (mean maximum c.p.m. − mean spontaneous c.p.m.)] × 100. Spontaneous release or maximum release was determined by incubation of labeled target cells with medium alone or 2% Triton X-100, respectively.

Virus-inhibition assay. HLA-B*27- or HLA-B*57-expressing GXR cells were infected for 4 h at 37 °C with the appropriate HIV-1 strain or viral vari-ant at the specified multiplicity of infection, then were washed and cultured together with bulk CD8+ T cells isolated from PBMCs by negative selection (Dynabeads; Invitrogen) or CD8+ T cell clones at the appropriate effector cell/target cell ratio. The ability to recognize HIV-1 and viral variants by CD8+ T cells was analyzed by flow cytometry as the proportion of GFP+ cells over 7 d in culture. To additionally address the relative antiviral efficacy of epitope-specific CD8+ T cell responses, the ability of bulk CD8+ T cells and CD8+ T cell populations depleted of epitope-specific cells to inhibit viral replica-tion in autologous primary CD4+ T cells was measured by analysis of p24 production as described37. Primary CD4+ T cells were isolated from PBMCs by negative selection (Dynabeads; Invitrogen). Over 98% of these primary cells coexpressed CD3 and CD4, as assessed by flow cytometry. Those CD4+ T cells were stimulated with monoclonal antibody bispecific for CD3 and CD8 (CD3:8; a gift from J. Wong)50 and were infected for 4 h at 37 °C at day 3 with the appropriate HIV-1 isolates at a multiplicity of infection of 0.1, except as otherwise specified. Virus-infected cells were then washed and incubated in the presence or absence of effector cells at an effector cell/target cell ratio of 1:1 in RPMI-1640 medium in addition of IL-2 at 50 U/ml. At regular inter-vals, cultures were ‘fed’ by removal and replacement of one-half of the culture

©20

12 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature immunology doi:10.1038/ni.2342

supernatant with fresh medium. The supernatant removed was cryopreserved for subsequent quantification of p24 antigen by ELISA (Dupont). Viral inhibi-tion was calculated as follows: % inhibition = 100 × [1 − ([% GFP+ cells with effector cells] / [% GFP+ cells without effector cells])].

Effector cell–target cell conjugation, granule loading and delivery. CD8+ T cell clones were cultured for 30 min with HIV-1-infected or uninfected HLA-B*27-expressing GXR cells in RPMI-1640 medium in the absence of IL-2. Perforin in CD8+ T cell clones was stained as described40. Cells were fixed for 15–30 min at room temperature with freshly prepared 4% para-formaldehyde and were washed three times with PBS. Cells were permeabi-lized for 30 min at room temperature in 0.5% Triton X-100 (Sigma) and 10% normal donkey serum (Jackson Immunoresearch) in PBS. Cells were stained for 60 min at room temperature with monoclonal mouse IgG2b antibody to perforin (δG9; Pierce). Anti-perforin was diluted (1:333) with 0.05% Triton X-100 and 3% normal donkey serum in PBS. After three washes in PBS, cells were incubated for 1 h at room temperature with anti-perforin (A-20186; Invitrogen) in 0.05% Triton X-100 and 3% normal donkey serum in PBS. Anti-perforin was conjugated to Alexa Fluor dyes (1:1,000 dilution). F-actin was stained with phalloidin–Alexa Fluor 647 (1:50 dilution; Invitrogen). Confocal images were collected on a Zeiss LSM510 Meta Confocal microscope with a plan apochromat 63× oil-immersion objective with a numerical aperture of 1.4. Differential interference contrast images were collected simultaneously

with the fluorescence images. Multitrack acquisition mode was used to avoid crosstalk between the different fluorophores. Images were analyzed with Imaris software (Bitplane).

Statistical analyses. An unpaired t-test with Welch’s correction and Mann-Whitney tests were done with GraphPad Prism version 4.0a. All tests were two tailed, and P values of less than 0.05 were considered significant.

45. Miura, T. et al. Genetic characterization of human immunodeficiency virus type 1 in elite controllers: lack of gross genetic defects or common amino acid changes. J. Virol. 82, 8422–8430 (2008).

46. Allen, T.M. et al. Selection, transmission, and reversion of an antigen-processing cytotoxic T-lymphocyte escape mutation in human immunodeficiency virus type 1 infection. J. Virol. 78, 7069–7078 (2004).

47. Walker, B.D. et al. Long-term culture and fine specificity of human cytotoxic T-lymphocyte clones reactive with human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. USA 86, 9514–9518 (1989).

48. Varadarajan, N. et al. A high-throughput single-cell analysis of human CD8+ T cell functions reveals discordance for cytokine secretion and cytolysis. J. Clin. Invest. 121, 4322–4331 (2011).

49. Yang, O.O. et al. Impacts of avidity and specificity on the antiviral efficiency of HIV-1-specific CTL. J. Immunol. 171, 3718–3724 (2003).

50. Wilson, C.C. et al. Ex vivo expansion of CD4 lymphocytes from human immunodeficiency virus type 1-infected persons in the presence of combination antiretroviral agents. J. Infect. Dis. 172, 88–96 (1995).