Virus-Specific CD8 T-Cell Responses Better ... - Journal of Virology

2 downloads 0 Views 1MB Size Report
Nov 18, 2009 - der, E. S. Rosenberg, M. Altfeld, and B. D. Walker. 2003. .... Simons, B. C., S. E. VanCompernolle, R. M. Smith, J. Wei, L. Barnett, S. L.. Lorey, D.
JOURNAL OF VIROLOGY, May 2010, p. 4461–4468 0022-538X/10/$12.00 doi:10.1128/JVI.02438-09 Copyright © 2010, American Society for Microbiology. All Rights Reserved.

Vol. 84, No. 9

Virus-Specific CD8⫹ T-Cell Responses Better Define HIV Disease Progression than HLA Genotype䌤† Warren L. Dinges,1,2 Julia Richardt,1 David Friedrich,1 Emilie Jalbert,1 Yi Liu,3 Claire E. Stevens,2 Janine Maenza,2 Ann C. Collier,2 Daniel E. Geraghty,4 Jeremy Smith,1 Zoe Moodie,5 James I. Mullins,2,3,6 M. Juliana McElrath,1,2,6 and Helen Horton1,2*‡ Vaccine and Infectious Disease Institute,1 Clinical Research Division,4 and Statistical Center for HIV Research and Prevention,5 Fred Hutchinson Cancer Research Center, Seattle, Washington, and Departments of Medicine,2 Microbiology,3 and Laboratory Medicine,6 University of Washington School of Medicine, Seattle, Washington Received 18 November 2009/Accepted 1 February 2010

HLA alleles B57/58, B27, and B35 have the strongest genetic associations with HIV-1 disease progression. The mechanisms of these relationships may be host control of HIV-1 infection via CD8ⴙ T-cell responses. We examined these immune responses in subjects from the Seattle Primary Infection Cohort with these alleles. CD8ⴙ T-cell responses to conserved HIV epitopes within B57/58 alleles (TW10 and KF11) and B27 alleles (KK10 and FY10) delayed declines in CD4ⴙ T-cell counts (4 to 8 times longer), while responses to variable epitopes presented by B35 alleles (DL9 and IL9) resulted in more rapid progression. The plasma viral load was higher in B57/58ⴙ and B27ⴙ subjects lacking the conserved B57/58- and B27-restricted responses. The presence of certain B57/58-, B27-, and B35-restricted HIV-specific CD8ⴙ T-cell responses after primary HIV-1 infection better defined disease progression than the HLA genotype alone, suggesting that it is the HIV-specific CD8ⴙ T cells and not the presence of a particular HLA allele that determine disease progression. Further, the most effective host CD8ⴙ T-cell responses to HIV-1 were prevalent within an HLA allele, represented a high total allele fraction of the host CD8ⴙ T-cell response, and targeted conserved regions of HIV-1. These data suggest that vaccine immunogens should contain only conserved regions of HIV-1. sponses in early HIV-1 infection in subjects possessing at least one of the above-mentioned disease-altering HLA alleles. We examined the relationships between CD8⫹ T-cell responses restricted by these alleles and the decline in CD4⫹ T-cell counts and HIV-1 RNA viral loads. These data provide the first clear evidence for CD8⫹ T-cell responses in early infection directly altering HIV-1 disease progression. Furthermore, we show that the most effective CD8⫹ T-cell responses are those that (i) are prevalent within an HLA allele, (ii) represent a high total allele fraction of the host CD8⫹ T-cell response, and (iii) target conserved regions of HIV-1. These data suggest that next-generation T-cell-based HIV-1 vaccine candidates should contain only conserved regions of HIV-1.

The toll of the HIV-1 pandemic continues to climb, with a recent estimate of 33 million people living with HIV-1 (36). The time from HIV-1 infection to the onset of AIDS varies markedly among subjects, with the average being 8 to 10 years without antiretroviral therapy (28, 34). Some HLA class I alleles are associated with delayed progression to AIDS (e.g., B57/58 and B27), some are associated with rapid progression (e.g., B35Px alleles: B*3502, B*3503, B*3504, and B*5301) (10, 13), and others are associated with a more moderate course (e.g., B35PY alleles: B*3501 and B*3508). The mechanisms underlying these associations are unclear, although recent data have shown that, unlike CD8⫹ T-cell responses restricted by other alleles, responses restricted by HLA B57/58 and B27 appear to be resistant to peripheral tolerance (14), as they retain their proliferative capacity throughout chronic infection. The differential function of cytotoxic T-lymphocytes (CTL) is very likely to be a critical component of host control of HIV-1 infection, along with B-lymphocyte-derived antibodies, helper T-lymphocyte function, natural killer cell responses, and dendritic cell function (5). Better elucidation of the mechanisms of host control over HIV-1 disease progression would have enormous implications for future therapy and rational vaccine design. Therefore, we assessed the breadth and magnitude of HIV-1-specific gamma interferon (IFN-␥)-secreting T-cell re-

MATERIALS AND METHODS Subjects. HIV-1-infected individuals were selected from the Seattle Primary Infection Clinic (PIC) cohort based on HLA genotype and sample availability. Appropriate institutional review boards approved the studies, and volunteers provided written informed consent. Study design. This study was designed as a four-armed nested case-control study comparing subjects with selected HLA-B alleles (B57/58, B27, B35Px, and B35PY) from the PIC cohort. The primary outcome was disease progression defined by CD4⫹ T-cell counts declining below thresholds (500, 350, and 200 per mm3). The samples used in this study were drawn within 150 days of primary HIV-1 infection (PHI) (32). The PIC cohort’s estimated HIV-1 infection date was used as the date of infection; this was the symptom onset date for symptomatic subjects or the midpoint between the last negative and first positive HIV tests in those lacking symptoms. All subjects were antiretroviral therapy naïve at the time of sampling. Most subjects (n ⫽ 31) were studied at a single time point, while some (n ⫽ 14) were sampled at two time points. CD4ⴙ T-cell count and HIV-1 viral-load assays. CD4⫹ T-cell counts were determined by the University of Washington (UW) Laboratory Medicine clinical laboratory by combining percent CD4⫹ T cells from flow cytometry and lymphocyte counts from hemocytometry. Before July 2003, HIV-1 viral-load testing

* Corresponding author. Mailing address: Seattle Biomedical Research Institute, 307 Westlake Ave. N, Seattle. WA 98109. Phone: (206) 256-7239. Fax: (206) 256-7229. E-mail: [email protected]. † Supplemental material for this article may be found at http://jvi .asm.org/. ‡ Present address: Seattle Biomedical Research Institute, Seattle, WA. 䌤 Published ahead of print on 10 February 2010. 4461

4462

DINGES ET AL.

J. VIROL.

employed the Amplicor HIV-1 Monitor Test (Roche Molecular Systems, Inc.) with a lower limit of detection of 50 copies per ml; thereafter, an in-house real-time reverse transcription (RT)-PCR method was used, with a lower limit of detection of 30 copies per ml (17). The HIV-1 plasma viral load was measured at time point 1 for viral-load analyses. IFN-␥ ELISPOT. Cryopreserved PBMC were thawed and rested overnight at 37°C before use at 65,000 to 200,000 PBMC per well in IFN-␥ enzyme-linked immunospot (ELISPOT) assays (Mabtech) following the manufacturer’s guidelines. Synthesized peptides corresponding to described HLA class I-restricted CTL epitopes were tested according to an individual’s HLA type. BioSyn Corp., New England Peptide, Inc., and Mimotopes synthesized the 8- to 11-mer peptides. Peptides were used at a final concentration of 2 ␮g ml⫺1. Phytohemagglutinin (PHA-P) (0.5 ␮g ml⫺1; Murex) served as a positive control, and wells with medium alone served as negative controls. An IFN-␥ ELISPOT result was designated positive when the number of spot-forming cells (SFC) was twice background (negative control) and at least 50 SFC per 106 peripheral blood mononuclear cells (PBMC). HLA class I typing. Most HLA typing was performed by clinical laboratories according to Clinical Laboratory Improvement Amendment (CLIA)-accredited techniques. Most typing was conducted at the Puget Sound Blood Center, initially using serology methods for lymphocyte lysis assays, later using sequencespecific primer (SSP) PCR at low resolution, and more recently with highresolution SSP-PCR reagents from One Lambda, Inc., and Olerup SSP from Genovision (now Qiagen Inc.). A subset was analyzed at the Oklahoma Health Sciences Laboratory using CLIA-accredited sequence-based typing (SBT) PCR (35). Currently, a high-throughput resequencing SBT-PCR is being performed by D. E. Geraghty at the Fred Hutchinson Cancer Research Center (31). A subset of subjects were previously typed only at low resolution, so high-resolution HLA-B35 typing was performed with a LabType SSO Class I B Locus Typing Test (RSSO1B_011_BPI REV 1; One Lambda, Inc.) on a Luminex instrument. Conservation score analysis. For each 8- to 11-mer peptide examined, the conservation score was represented by the frequency of occurrence of the peptide sequence in the HIV-1 M-group or B-subtype sequences in the Los Alamos HIV-1 sequence database of 2005 (18). Only one sequence per subject was analyzed. Statistics. Survival models were developed based on HLA genotype grouping alone and on the presence of good, moderate, or bad CTL responses using ESB1 and ESB2 categorizations (see below). Kaplan-Meier (KM) plots and Cox proportional-hazard (PH) models were used to analyze the time from the infection date to the first time below threshold CD4⫹ T-cell counts. Subjects were censored when they left the cohort or initiated antiretroviral therapy (ART). The Cox PH models assumed censoring was independent of the CD4 decline event. Much of the censoring was due to ART initiation. However, at the time the data were collected (1994 to 2007), treatment was often initiated soon after diagnosis and for reasons other than CD4⫹ T-cell counts or viral load. Figure S1 in the supplemental material illustrates the CD4 counts at the visit at least 3 days prior to ART initiation for those censored due to ART for the three thresholds; the majority of those beginning ART did not have CD4⫹ T-cell counts that would trigger ART initiation according to current guidelines (27); hence, the independent-censoring assumption appears reasonable. Kruskal-Wallis (KW) tests were used to compare total ELISPOT responses and viral loads across all groups, and Wilcoxon-Mann-Whitney (WMW) tests were used to compare two groups. All tests were two sided, with a threshold of P ⫽ 0.05. There were no adjustments for multiple testing. All analyses were conducted in R version 2.8.1.

RESULTS Most subjects had symptomatic infection and 3 HIV-1-specific responses. We analyzed epitope-specific HLA-B (B57/58, B27, B35Px, and B35PY alleles; collectively referred to as ESB)-restricted IFN-␥ ELISPOT responses in 45 subjects (B57/58, n ⫽ 9; B27, n ⫽ 7; B35Px, n ⫽ 12; and B35PY, n ⫽ 20) from the PIC cohort (Fig. 1). Like the parent cohort and the demographics of HIV-1 infection in the Pacific Northwest, these subjects were mostly Caucasian men who have sex with men (MSM) (Table S1); we included one woman, one male self-identified as mixed race, and two males self-identified as Hispanic. The subjects were evaluated within 150 days of PHI. Most (n ⫽ 40) had symptomatic PHI, while some (n ⫽ 5) were identified with asymptomatic infection by routine testing. The

FIG. 1. Study design. Primary infection subject (n ⫽ 45) selection is shown. Inclusion was limited by available HLA genotype data; presence of an ESB-restricted allele of interest, HLA-B57/58, B27, B35Px, or B35PY; sample availability; and cell viability of ⬎60%. Four subjects lacked ELISPOT responses, and four had poor cell viability. One subject possessed both B27 and B57, and two subjects possessed both B35Px and B35PY.

median time to evaluation was 48 days. A subset of 14 subjects were analyzed at a later time (time 2) a median of 135 days postinfection or after onset of clinical symptoms of PHI (collectively referred to as days postinfection [p.i.]). Due to limited PBMC availability, subjects were screened only for HIV-1-specific responses to previously reported epitopes based on HLA genotype (16). This screening depended on sample availability according to the following sequential hierarchy: ESB, other HLA-B, HLA-A, and then HLA-C (see Tables S2 and S3 in the supplemental material). The median number of epitopes detected at time 1 by screening was 3 (range, 1 to 10) (Fig. 2a), and the mean response magnitude was 560 SFC per 106 PBMC (median, 384; range, 62 to 2,060). The total magnitude of responses restricted by HLAB57/58, B27, and B35 was not significantly different (KW, P ⫽ 0.4) (Fig. 2b). Details of the 226 detected responses from 59 PBMC specimens are reported in Tables S4 and S5 in the supplemental material and summarized in Table 1. HIV-1-specific responses doubled in the first 4 months of infection. Fourteen subjects had a second assessment of CTL responses 4 months after infection. In these, the number of epitopes detected and the total magnitude of responses within a subject increased at the second time point. However, the median magnitudes of IFN-␥-secreting ELISPOT responses from time points 1 and 2 were no different (Wilcoxon signedrank [WSR], P ⫽ 0.8). The median number of responses increased from 2 to 5 (range, ⫺2 to 6) (WSR, P ⫽ 0.004) (Fig. 2c). The total magnitude of responses increased from a median of 821 to 1,986 SFC per 106 PBMC (range, ⫺203 to 3,325) (WSR, P ⫽ 0.02) (Fig. 2d). This underscores the fact that the

VOL. 84, 2010

CD8⫹ T-CELL RESPONSES DEFINE HIV DISEASE PROGRESSION

4463

FIG. 2. CD8⫹ T-cell IFN-␥ ELISPOT responses. (a) A median of 3 (range, 1 to 10) IFN-␥ ELISPOT responses were detected at a median of 48 days p.i. by HIV-1 with testing of HLA-predicted screening of 45 subjects, and the distribution is shown. (b) The total IFN-␥ responses were no different by HLA group (Kruskal-Wallis, P ⫽ 0.36). (c) In a subset of 14 subjects also evaluated at a later time point (median, 135 days p.i.), the median number of responses increased significantly (WSR, P ⫽ 0.004) from 2 to 5 responses, a change of median 3 (range, ⫺2 to 6). (d) There was also a significantly greater total response magnitude at the second time point (WSR, P ⫽ 0.02). Horizontal lines, median and interquartile range. x2, two subjects had 1 and then 6 responses; x3, three subjects had 2 and then 4 responses.

early CTL responses to HIV-1 detected by HLA-based screening increased about 2-fold from 1 to 4 months of infection. The HLA genotype was a predictor of HIV-1 disease progression. To better understand the impacts of epitope-specific CD8⫹ T-cell responses restricted by HLA-B57/58, B27, and B35, we conducted survival analyses of subjects grouped by their HLA genotypes: B57/58, B27, B35Px, and B35PY. KM survival curves were constructed, along with Cox PH models of the time until CD4⫹ T-cell counts first declined below threshold values of 500, 350, and 200 (Fig. 3). In the KM plots, there was a significant difference in at least one of the HLA groups from the time to CD4⫹ T-cell count below 500, 350, and 200 (Table 2). In a Cox PH estimate that included the three HLA groups, only the B35 group was significantly faster to decline below the 500 and 350 CD4⫹ T-cell count thresholds than the B57/58 group, with hazard ratios (HR) of 5.8 and 9.7. For the ⬍200 CD4⫹ T-cell count threshold, the B57/58 and B27 groups had 0/15 events, and the B35 (Px or PY)groups had 6/30 events; the HR was infinite. ESB-restricted responses also predicted HIV-1 disease progression. We also conducted survival analyses of subjects grouped by two different models of epitope-specific HLA-Brestricted responses (ESB1 and ESB2). The ESB models were based on the hypothesis that CTL responses to conserved regions of HIV-1 would result in delayed progression while CTL responses to variable regions would result in HIV-1 disease progression. We arbitrarily defined conserved epitopes as those with a frequency of identical sequences in the Los Alamos HIV-1 sequence database greater than 70% for Mgroup or B-subtype virus sequences. We also limited our ana-

lyses to responses present in at least half of each HLA group so that a sufficient number of subjects would be present in CTL response groupings. We selected the HLA-B57/B58-restricted TW10 and KF11 responses and the HLA-B27-restricted KK10 and FY10 responses as hypothetical key prevalent and conserved responses. Likewise, we selected the highly prevalent and variable B35-restricted DL9 and IL9 responses as the hypothetical variable responses. Subjects were grouped into strata of good, moderate, and bad based on the presence (⫹) or absence (⫺) of beneficial or detrimental ESB responses: for ESB1, good (B27KK10⫹ and B57TW10⫹), moderate (B27KK10⫺, B57TW10⫺, and B35IL9⫺/ DL9⫺), and bad (B35IL9⫹/DL9⫹) were used. For ESB2, good (B27KK10⫹/FY10⫹ and B57TW10⫹/KF11⫹), moderate (B27KK10⫺/FY10⫺, B57TW10⫺/KF11⫺, and B35IL9⫺/ DL9⫺), and bad (B35IL9⫹/DL9⫹) were used. The ESB1 and ESB2 models were developed to attempt to discriminate between the effects of the major versus minor prevalence responses compared to the HLA genotype alone. The ESB1 and ESB2 models differed only in the handling of the secondary B27FY10 and B57KF11 responses, and there was no difference in the B35 individuals. The B35 subjects, whether Px or PY, could be placed into the bad group only if they had a detrimental response to the highly variable IL9/DL9 epitopes; otherwise, they were in the moderate groups. The KM plots for the ESB1 and ESB2 groups showed significant differences between groups in the time to CD4⫹ T-cell count decline below 500, 350, and 200 (Fig. 3), with smaller P values than comparable KM plots grouped by HLA only (Table 2). In a Cox PH model of the three ESB1 groups, there was

4464

DINGES ET AL.

J. VIROL. TABLE 1. Summarized CD8⫹ T-cell IFN-␥ ELISpot responsesa

HLA allele

A03 A29 B08 B08 B08 B13 B27e B27e B27e B35Pxf B35Pxf B35PYg B51 B51 B57/58h B57/58h B57/58h B57/58h B57/58h

Response prevalenceb

5/10 (50) 3/3 (100) 3/7 (43) 3/7 (43) 3/7 (43) 3/3 (100) 4/7 (57) 5/7 (71) 3/7 (43) 7 ⫹ 2/12 (58–75) 6/12 (50) 9/20 (45) 3/4 (75) 3/4 (75) 4/9 (44) 4/9 (44) 7/9 (78) 6/9 (67) 6/9 (67)

29 67 17 21 13 58 26 67 19 36 37 11 47 55 13 20 27 34 42

Conservation score (%)d

Epitope

Response dominance (%)c HXB2 site

Peptide

Abbreviation

M

B

Gag p1720-28 Env gp160209-217 Gag p24128-135 Nef90-97 Env gp160848-856 Nef106-114 Pol Int185-194 Gag p24131-140 Vpr31-39 Env gp16078-86 Pol RT293-301 Pol RT175-183 Vpr29-37 Pol Int28-36 Nef116-124 Gag p2415-23 Pol RT244-252 Gag p2430-40 Gag p24108-117

RLRPGGKKK SFEPIPIHY EIYKRWII FLKEKGGL RQGLERALL RQDILDLWI FKRKGGIGGY KRWIILGLNK VRHFPRIWL DPNPQEVVL IPLTEEAEL NPDIVIYQY EAVRHFPRI LPPVVAKEI HTQGYFPDW ISPRTLNAW IVLPEKDSW KAFSPEVIPMF TSTLQEQIGW

RK9 SY9 EI8 FL8 RL9 RI9 FY9 KK10 VL9 DL9 IL9 NY9 EI9 LI9 HW9 ISW9 IVW9 KF11 TW10

45 35 56 91 33 5 86 87 12 12 9 24 13 37 36 59 14 84 40

65 75 86 78 53 20 80 82 33 38 34 66 33 72 74 70 62 97 73

a

Only responses prevalent in at least 40% of those with an allele and in at least 3 subjects are shown. Number/total (%). Prevalence is the fraction or percentage of subjects with a response within a given HLA allele. Dominance is the percentage of the subject’s total response magnitude derived from this specific response. d Conservation was calculated as the frequency of complete epitope matches in aligned M-group (n ⫽ 615 to 1,224) and B-subtype (n ⫽ 97 to 243) sequences. e B27 included one subject with combined B27 and B57 without B27-restricted responses. f B35Px included two subjects with combined B35Px/PY with both B35Px/PY responses. g B35PY included one subject with B35PY homozygosity. h B57/58 included two subjects with B58. b c

a significant difference between the moderate and good groups in the time to decline below 500, with an HR of 8.2 (Table 2). There was also some evidence of a difference between groups in time to the 350 threshold: a hazard ratio of 8.2 and a significant difference for ESB1 bad versus good strata for declines below thresholds of 500 and 350, with HR of 16 and 25. For the 200 threshold, the bad, moderate, and good groups had 3/12, 3/22, and 0/11 events; the HR for moderate versus good and bad versus good were infinite. Similar results were seen in the Cox PH of the three ESB2 groups. There was a significant difference between the moderate and good groups in the time to decline below 500 CD4⫹ T-cell counts, with an HR of 3.6, and a trend in the differences in the time to decline below 350 CD4⫹ T cells (HR ⫽ 4.5). There was also a significant difference for ESB2 bad versus good strata for declines below thresholds of 500 and 350, with hazard ratios of 7.1 and 13. For the 200 threshold, the bad, moderate, and good groups had 3/12, 3/19, and 0/14 events; the HR for moderate versus good and bad versus good were infinite. In both ESB1 and ESB2 models, the P values for bad versus moderate were less than 0.20 at all thresholds, with ESB1 HR of 2.0, 3.1, and 6.0 and ESB2 HR of 2.0, 2.9, and 4.8. This suggests a faster CD4⫹ T-cell decline in B35 individuals with detrimental responses to the variable DL9 and IL9 epitopes. ESB-restricted responses provided additional predictive gains. Next, to understand the additive predictive effects of the CTL-based models to the HLA model, a Cox PH model of the HLA groups (reduced model) was compared to an additional Cox PH model of the HLA groups and the ESB1 group (full model). For the time to CD4⫹ T-cell count decline below 500,

the full model significantly improved upon the reduced model, with a Nagelkerke R2 of 0.36 versus 0.22 (Table 2). The results did not reach significance for the 350 threshold, although there was improvement in the R2 with the full model (0.25 versus 0.17). Interestingly, the ESB2 categories (full model) did not significantly improve the prediction of time to CD4 decline for either the 500 or 350 thresholds. These survival analyses demonstrate two key points. First, CD8⫹ T-cell responses to key HLA-B-restricted epitopes determined disease progression in HIV-1-infected subjects better than the HLA genotype alone. This was demonstrated in smaller log-rank P values (KM plots) and larger Cox PH HR and, most convincingly, by significantly improved prediction of the time to CD4⫹ T-cell count decline when ESB1 was used in addition to HLA groupings. Second, the presence of ESB responses to B27KK10/FY10 and B57TW10/KF11 delayed declines in CD4⫹ T-cell counts to 3.5 to 8.2 times longer or more than 8 years in HLA-B57/58- and B27-positive subjects, respectively. The HIV-1 viral load differed by ESB group. To evaluate the impact of CD8⫹ T-cell ESB responses on the control of viral replication, we examined the plasma HIV-1 viral loads at the same time points. As also reported by others (7, 22, 25, 33), the median plasma viral loads were different by HLA grouping (KW, P ⫽ 0.001), with medians of the B57/58 group being lower than those of the B35Px and B35PY groups (WMW, P ⬍ 0.0001 and P ⫽ 0.0003, respectively) (Fig. 4a). Likewise, the median HIV-1 viral loads of subjects grouped by ESB1 and ESB2 were different (KW, P ⫽ 0.004 and 0.01, respectively) (Fig. 4b and data not shown). Interestingly, the four highest viral loads in the B57/58 subjects (WMW, P ⫽ 0.02) and the

VOL. 84, 2010

CD8⫹ T-CELL RESPONSES DEFINE HIV DISEASE PROGRESSION

4465

FIG. 3. Survival curves. KM plots of survival fraction models are shown for the time to decline in CD4⫹ T-cell counts below a threshold of 500, 350, or 200 CD4⫹ T cells per ␮l. Subjects were grouped according to HLA genotype (B57/58, B27, B35Px, and B35PY) (top row) or by possession of particularly good, moderate, or bad IFN-␥ ELISPOT responses. For ESB1 (middle row), good (B27KK10⫹ and B57TW10⫹), moderate (B27KK10⫺, B57TW10⫺, and B35IL9⫺/DL9⫺), and bad (B35IL9⫹/DL9⫹) were used. For ESB2 (bottom row), good (B27KK10⫹/FY10⫹ and B57TW10⫹/KF11⫹), moderate (B27KK10⫺/FY10⫺, B57TW10⫺/KF11⫺, and B35IL9⫺/DL9⫺), and bad (B35IL9⫹/DL9⫹) were used. Censoring was due to treatment initiation or dropout. For the KM plots, at least one HLA stratum was significantly different for time to CD4⫹ T-cell counts below 500, 350, and 200 (log rank, P ⫽ 0.01, 0.008, and 0.02). For ESB1 and ESB2, there was also a significant difference for time to CD4⫹ T-cell counts below 500, 350, and 200 in at least one of the groups (log rank, P ⫽ 0.0004, 0.005, 0.01 for ESB1 and P ⫽ 0.002, 0.008, 0.007 for ESB2).

highest viral load among the B27 subjects were in those lacking their respective ESB1 TW10 and KK10 responses. Together, these data demonstrate that altered disease progression in B57/58 and B27 subjects is at least partially attributable to the impact of ESB-restricted CTL responses on viral replication. Conservation and dominance better defined HIV-1 response efficacy. Our initial interest focused on the highly prevalent responses to conserved regions restricted by key HLA alleles (B57TW10/KF11 and B27KK10/FY9), which are known to be dominant in early HIV-1 infection (3), and to variable regions (B35DL9/IL9), also with relatively high dominance. Table 1 summarizes all responses present in at least 40% of the subjects possessing a given HLA class I allele and in at least three subjects. This sorting allowed the focus to be on common CD8⫹ T-cell responses based on both HLA allele prevalence and response prevalence within an allele. A total of 19 epitopes met these criteria, representing 107 of 226 or 47% of the detected CTL responses. The lead B27KK10 epitope was highly conserved (87% M

group and 82% B subtype) and dominant (67% of subjects’ total IFN-␥ ELISPOT responses to the HIV-1 peptides tested). The similarly conserved (86% M group and 80% B subtype) but subdominant (26%) B27FY9 response contributed to the B27-restricted response, together representing a median of 64% of the total response magnitude of the host CTL response to HIV-1. Likewise, the five B57/58 epitopes demonstrated the diversity of B57/58 in targeting multiple, but conserved, regions of HIV-1. The B57/58-restricted epitopes represented a median of 86% of the total magnitude of the CTL response to HIV-1. For HLA-B57/58, then, it appeared that subdominant CTL responses within an allele additively contributed to a cumulative dominant and effective response to conserved regions of HIV-1. Responses to variable HIV-1 regions lacked efficacy. In contrast, B35DL9/IL9, B13RI9, B27VL9, and B51EI9 represented responses to variable HIV-1 regions, all with low conservation scores (⬍13% M group and ⬍38% B subtype). In the case of the B35DL9/IL9 responses, they were associated with more rapid progression in the survival models described above, sug-

4466

DINGES ET AL.

J. VIROL.

TABLE 2. Survival model results Parameter

KM P value HLAa ESB1c ESB2d Cox PHb HR (P value) HLAa B35 vs. B57/58 ESB1c M vs. G B vs. G B vs. M ESB2c M vs. G B vs. G B vs. M Cox PHd R2 (P value) HLAa ⫹ ESB1c (full) HLAa ⫹ ESB2c (full) HLAa (reduced)



CD4 T-cell count threshold

500 0.01 0.0004 0.002

350 0.008 0.005 0.008

200 0.02 0.01 0.007

5.8 (0.006)

9.7 (0.03)

⬁ (0.005)e

8.2 (0.007) 16 (0.0007) 2.0 (0.13)

8.2 (0.050) 25 (0.007) 3.1 (0.10)

⬁ (0.03)f ⬁ (0.02)f 6.0 (0.12)

4.5 (0.066) 13 (0.006) 2.9 (0.14)

⬁ (0.01) ⬁ (0.01)g 4.8 (0.18)

studies may be able to determine the threshold for effective conservation and dominance in the context of prevalence-defined sample size. Low HLA allele prevalence and low response prevalence prevented such analyses of the less conserved A29SY9, B08EI8, and B51LI9 epitopes (35 to 56% M group and 72 to 86% B subtype). Interestingly, the highly conserved Nef B08FL8 response has been associated with long-term nonprogression, and escape mutations have been associated with lower CD4⫹ T-cell counts and higher viral loads (12, 23, 24). Together, these data characterized the effective host CD8⫹ T-cell responses to HIV-1: prevalent within an HLA allele, representing a high total allele fraction of the host CTL response, and targeting conserved regions of HIV-1. DISCUSSION

3.6 (0.03) 7.1 (0.001) 2.0 (0.15)

0.36 (0.01) 0.25 (0.4) 0.22

g

0.25 (0.081) 0.21 (0.3) 0.17

a

Human leukocyte antigen model grouping: B57/58, B27, B35(Px or PY). Cox PH model HR and Wald test P value. B, bad; M, moderate; G, good. d Cox PH model Nagelkerke R2 improvement of prediction with full over reduced model; P value from likelihood ratio test. e B57/58, 0/9 events; B27, 0/6 events; B35, 6/30 events; HR was infinite; likelihood ratio test P value. f B, 3/12 events; M, 3/22 events; G, 0/11 events; HR was infinite; likelihood ratio test P value. g B, 3/12 events; M, 3/19 events; G, 0/14 events; HR was infinite; likelihood ratio test P value. b c

gesting that these CTL responses to variable regions were ineffective at limiting disease progression. The A03RK9, B08RL9, and B35PYNY9 epitopes had moderate conservation scores (24 to 45% M group and 53 to 66% B subtype), moderate prevalence (43 to 50%), and low dominance (11 to 29%). While numerous, the A03RK9 and B35PYNY9 responses did not demonstrate a significant effect on HIV-1 disease progression in our data (not shown). Future

The HLA-predicted CD8⫹ T-cell IFN-␥ ELISPOT responses we detected after primary HIV-1 infection matched the number of responses found in similar PHI cohorts (medians, 2 to 4; range, 0 to 7) (1, 4, 8). The longitudinal changes in the CD8⫹ T-cell responses in the first few months after infection resulted in more responses of a similar magnitude and an increasing total magnitude. We provided data showing that targeting of conserved versus variable epitopes correlated better with HIV-1 disease progression in persons with specific HLA alleles than the HLA genotype alone. To our knowledge, these are the first data showing that possession of HIV-1specific CD8⫹ T-cell responses in early infection is directly correlated with HIV-1 disease progression in terms of CD4⫹ T-cell count decline below clinically relevant thresholds. Recently, Streeck et al. saw an association with CD8⫹ T-cell response and set-point viral load only after primary HIV infection (which we and many others have observed [see below]), but they saw no association of CD8⫹ T-cell responses with changes in CD4⫹ T-cell counts after PHI (P ⫽ 0.52) (33). The association that they did find with CD8⫹ T-cell responses and CD4⫹ T-cell counts was observed only in chronic infection, and it was not associated with a similar correlation with the set-point viral load (P ⬎ 0.22). The study by Altfeld et al. did look at the correlation of total HIV-specific CD8⫹ T-cell responses during early infection with progression, but it did not find a significant correlation with any outcome other than

FIG. 4. HIV-1 viral loads. (a) The median HIV-1 RNA plasma viral loads by HLA grouping were different (KW, P ⫽ 0.001), with medians of the B57/58 group being lower than those of B35Px and B35PY (Wilcoxon-Mann-Whitney, P ⬍ 0.0001 and P ⫽ 0.003, respectively). (b) Likewise, the median viral loads were different by ESB1 grouping (KW, P ⫽ 0.004). (a and b) Horizontal lines indicate median and interquartile range.

VOL. 84, 2010

CD8⫹ T-CELL RESPONSES DEFINE HIV DISEASE PROGRESSION

death (3). In addition, this study did not look at (i) specific epitope/allele associations with disease outcomes, only the total HIV-specific IFN-␥-secreting T-cell response, or (ii) whether possession of HIV-specific T-cell responses restricted by protective or nonprotective alleles correlated better with survival than having the protective alleles alone. Our data showed, for the first time, a direct correlation of CD8⫹ T-cell responses induced during PHI with disease progression (decline in the CD4⫹ T-cell count below threshold) and an associated correlation with the viral load. Our analysis of the HIV-1 plasma viral load confirmed our survival results and matched prior reported associations of CTL responses with the viral load (12, 23, 24, 33). It also demonstrated that the plasma viral load is a possible correlate of CTL control of HIV-1 disease progression. Despite a nested case-control design looking at a limited number of HLA alleles (about 17% total allele prevalence) (9), we were also able to identify other responses (e.g., B08FL8, A03RK9, A29SY9, B08EI8, B08RL9, B35PYNY9, and B51LI9) that might contribute to differential control of HIV-1 infection. Recently, viral escape has been confirmed in some of the epitopes described here (7, 37). Likewise, other groups have confirmed CTL-driven viral escape in long-term nonprogressors after primary HIV-1 infection and in chronic HIV-1 disease (15, 19, 26, 37). For instance, the B57TW10 response resulted in complete escape early in infection, while the B08FL8 response resulted in 50% escape by 2 years (6). The B57/58KF11 response, with its 97% conservation score and its subsequent limited escape, clearly persisted as a critical component of the host response in B57/58 subjects (11, 30). It remains to be determined whether it is more important to have persistent responses to conserved epitopes, like B57/58KF11, or to have responses that result in escape, like B57TW10, that produce potentially less fit viruses. While our study population was limited, we demonstrated a differential effect on HIV-1 disease progression based on possession of certain CTL responses. This was supported by statistical analyses that showed a significant additional predictive effect of CTL responses over genotype alone. Future efforts to expand these survival results to more HLA alleles and other epitopes will clearly require larger numbers to include both rare HLA genotypes and lower-prevalence responses, so that more dominant, conserved responses that alter HIV-1 disease progression can be identified. The most thorough mapping of CTL responses would use autologous viral sequences of a distribution of a subject’s viral pool (2). A step down from this would be to use potential T-cell epitopes and then consensus viral sequences (21). Our approach, using only HLA-matched optimal epitopes, was a limitation of this study. It would also have been beneficial to use other methods for determining the magnitude of HIVspecific responses, such as tetramer staining. In addition, an assessment of epitope sequences in each individual’s virus would have allowed us to determine if the B35-restricted responses were not protective due to the fact that the epitopes rapidly mutated when under T-cell pressure. However, sample availability and lack of available tetramers for every epitope specificity limited the assays that could be performed. Our calculation of conservation scores equally weighted all positions in the peptide epitope; a strategy giving greater weight to

4467

the first and last 2 amino acids could provide different results, as these positions are more important in peptide-major histocompatibility complex (MHC) binding affinities (20, 29). In our efforts to associate a phenotype with the highly reported HLAgenetic associations of HIV-1 disease progression, the definition of a conserved epitope (⬎70%) was arbitrary. An exploration of the survival effects in less conserved B57/58 and B27 epitopes might provide further insight. Also, extending this approach to highly conserved epitopes restricted by alleles not associated with delayed progression (e.g., B08FL8, A03RK9, A29SY9, B08EI8, B08RL9, B35PYNY9, and B51LI9) may further confirm the CTL effect on progression. Our measure of CD8⫹ T-cell function was limited to IFN-␥ production in an ELISPOT assay. Future efforts might examine more cytokines (e.g., tumor necrosis factor alpha), effector markers (e.g., perforin and granzymes), and CTL proliferative capacity (via carboxyfluorescein diacetate succinimidyl ester staining) through other methods (e.g., intracellular cytokine staining, viral suppression assays, and T-cell functional assays). These expanded efforts will almost certainly require new longitudinal cohort enrollment and follow-up, as the methods are not possible with the limited cell numbers in currently stored historical cohorts. In conclusion, we conducted a nested case-control study that uniquely demonstrated HIV-1 disease progression outcome differences among subjects with HLA B57/58, B27, and B35 alleles based on their possessing CTL responses to conserved or variable regions of HIV-1. CTL responses within B57/58and B27-restricted alleles to conserved epitopes resulted in delayed declines in CD4⫹ T-cell counts, while responses within B35-restricted alleles to variable epitopes resulted in more rapid disease progression. HIV-1 plasma RNA viral loads matched these CTL effects. These data provide the first phenotypic mechanisms of the well-established disease nonprogression genotypes of HLA-B57/58, B27, and B35 in a clinical cohort. Our data suggest that vaccine immunogens should be designed to remove variable and nonimmunogenic regions of the proteome so that the host immune response has only the option of responding to conserved, immunogenic regions of HIV-1. ACKNOWLEDGMENTS We thank the volunteer subjects who made this work possible. This work was supported by National Institutes of Health grants R01 AI65328, U01 AI4674, U01 AI 46725, P01 AI57005, P30 AI27757, and M01-RR-00037 (University of Washington General Clinical Research Center). REFERENCES 1. Addo, M. M., X. G. Yu, A. Rathod, D. Cohen, R. L. Eldridge, D. Strick, M. N. Johnston, C. Corcoran, A. G. Wurcel, C. A. Fitzpatrick, M. E. Feeney, W. R. Rodriguez, N. Basgoz, R. Draenert, D. R. Stone, C. Brander, P. J. R. Goulder, E. S. Rosenberg, M. Altfeld, and B. D. Walker. 2003. 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. 2. Altfeld, M., M. M. Addo, R. Shankarappa, P. K. Lee, T. M. Allen, X. G. Yu, A. Rathod, J. Harlow, K. O’Sullivan, M. N. Johnston, P. J. R. Goulder, J. I. Mullins, E. S. Rosenberg, C. Brander, B. Korber, and B. D. Walker. 2003. Enhanced detection of human immunodeficiency virus type 1-specific T-cell responses to highly variable regions by using peptides based on autologous virus sequences. J. Virol. 77:7330–7340. 3. Altfeld, M., E. T. Kalife, Y. Qi, H. Streeck, M. Lichterfeld, M. N. Johnston, N. Burgett, M. E. Swartz, A. Yang, G. Alter, X. G. Yu, A. Meier, J. K.

4468

4.

5.

6.

7.

8.

9.

10. 11.

12.

13.

14.

15.

16.

17.

18.

19.

DINGES ET AL.

Rockstroh, T. M. Allen, H. Jessen, E. S. Rosenberg, M. Carrington, and B. D. Walker. 2006. HLA alleles associated with delayed progression to AIDS contribute strongly to the initial CD8(⫹) T cell response against HIV-1. PLoS Med. 3:e403. Altfeld, M., E. S. Rosenberg, R. Shankarappa, J. S. Mukherjee, F. M. Hecht, R. L. Eldridge, M. M. Addo, S. H. Poon, M. N. Phillips, G. K. Robbins, P. E. Sax, S. Boswell, J. O. Kahn, C. Brander, P. J. R. Goulder, J. A. Levy, J. I. Mullins, and B. D. Walker. 2001. Cellular immune responses and viral diversity in individuals treated during acute and early HIV-1 infection. J. Exp. Med. 193:169–180. Broliden, K., A. T. Haase, S. K. Ahuja, G. M. Shearer, and J. Andersson. 2009. Introduction: back to basics: mucosal immunity and novel HIV vaccine concepts. J. Intern. Med. 265:5–17. Brumme, Z. L., C. J. Brumme, J. Carlson, H. Streeck, M. John, Q. Eichbaum, B. L. Block, B. Baker, C. Kadie, M. Markowitz, H. Jessen, A. D. Kelleher, E. Rosenberg, J. Kaldor, Y. Yuki, M. Carrington, T. M. Allen, S. Mallal, M. Altfeld, D. Heckerman, and B. D. Walker. 2008. Marked epitopeand allele-specific differences in rates of mutation in human immunodeficiency type 1 (HIV-1) Gag, Pol, and Nef cytotoxic T-lymphocyte epitopes in acute/early HIV-1 infection. J. Virol. 82:9216–9227. Brumme, Z. L., I. Tao, S. Szeto, C. J. Brumme, J. M. Carlson, D. Chan, C. Brander, B. Walker, D. Heckerman, P. R. Harrigan, C. Kadie, and N. Frahm. 2008. Human leukocyte antigen-specific polymorphisms in HIV-1 Gag and their association with viral load in chronic untreated infection. AIDS 22:1277–1286. Cao, J., J. McNevin, S. Holte, L. Fink, L. Corey, and M. J. McElrath. 2003. Comprehensive analysis of human immunodeficiency virus type 1 (HIV-1)specific gamma interferon-secreting CD8⫹ T cells in primary HIV-1 infection. J. Virol. 77:6867–6878. Cao, K., J. Hollenbach, X. Shi, W. Shi, M. Chopek, and M. A. Ferna ´ndezVin ˜ a. 2001. Analysis of the frequencies of HLA-A, B, and C alleles and haplotypes in the five major ethnic groups of the United States reveals high levels of diversity in these loci and contrasting distribution patterns in these populations. Hum. Immunol. 62:1009–1030. Carrington, M., and S. J. O’Brien. 2003. The influence of HLA genotype on AIDS. Annu. Rev. Med. 54:535–551. Crawford, H., J. G. Prado, A. Leslie, S. Hue, I. Honeyborne, S. Reddy, M. van der Stok, Z. Mncube, C. Brander, C. Rousseau, J. I. Mullins, R. Kaslow, P. Goepfert, S. Allen, E. Hunter, J. Mulenga, P. Kiepiela, B. D. Walker, and P. J. R. Goulder. 2007. Compensatory mutation partially restores fitness and delays reversion of escape mutation within the immunodominant HLAB*5703-restricted Gag epitope in chronic human immunodeficiency virus type 1 infection. J. Virol. 81:8346–8351. Dong, T., G. Stewart-Jones, N. Chen, P. Easterbrook, X. Xu, L. Papagno, V. Appay, M. Weekes, C. Conlon, C. Spina, S. Little, G. Screaton, A. van der Merwe, D. D. Richman, A. J. McMichael, E. Y. Jones, and S. L. RowlandJones. 2004. HIV-specific cytotoxic T cells from long-term survivors select a unique T cell receptor. J. Exp. Med. 200:1547–1557. Gao, X., G. W. Nelson, P. Karacki, M. P. Martin, J. Phair, R. Kaslow, J. J. Goedert, S. Buchbinder, K. Hoots, D. Vlahov, S. J. O’Brien, and M. Carrington. 2001. Effect of a single amino acid change in MHC class I molecules on the rate of progression to AIDS. N. Engl. J. Med. 344:1668–1675. Horton, H., I. Frank, R. Baydo, E. Jalbert, J. Penn, S. Wilson, J. P. McNevin, M. D. McSweyn, D. Lee, Y. Huang, S. C. De Rosa, and M. J. McElrath. 2006. Preservation of T cell proliferation restricted by protective HLA alleles is critical for immune control of HIV-1 infection. J. Immunol. 177:7406–7415. Kawashima, Y., K. Pfafferott, J. Frater, P. Matthews, R. Payne, M. Addo, H. Gatanaga, M. Fujiwara, A. Hachiya, H. Koizumi, N. Kuse, S. Oka, A. Duda, A. Prendergast, H. Crawford, A. Leslie, Z. Brumme, C. Brumme, T. Allen, C. Brander, R. Kaslow, J. Tang, E. Hunter, S. Allen, J. Mulenga, S. Branch, T. Roach, M. John, S. Mallal, A. Ogwu, R. Shapiro, J. G. Prado, S. Fidler, J. Weber, O. G. Pybus, P. Klenerman, T. Ndung’u, R. Phillips, D. Heckerman, P. R. Harrigan, B. D. Walker, M. Takiguchi, and P. Goulder. 2009. Adaptation of HIV-1 to human leukocyte antigen class I. Nature 458:641–645. Korber, B. T. M., C. Brander, B. F. Haynes, R. Koup, J. P. Moore, B. D. Walker, and D. I. Watkins (ed.). 2006. HIV molecular immunology 2006/ 2007. Los Alamos National Laboratory, Los Alamos, NM. Kristiansen, T. B., T. B. Knudsen, S. Ohlendorff, and J. Eugen-Olsen. 2001. A new multiplex PCR strategy for the simultaneous determination of four genetic polymorphisms affecting HIV-1 disease progression. J. Immunol. Methods 252:147–151. Leitner, T., B. Foley, B. Hahn, P. Marx, F. McCutchan, J. Mellors, S. Wolinsky, and B. Korber (ed.). 2005. HIV sequence compendium, vol. 060680. Theoretical Biology and Biophysics Group, Los Alamos National Laboratory, Los Alamos, NM. Leslie, A. J., K. J. Pfafferott, P. Chetty, R. Draenert, M. M. Addo, M. Feeney, Y. Tang, E. C. Holmes, T. Allen, J. G. Prado, M. Altfeld, C. Brander, C. Dixon, D. Ramduth, P. Jeena, S. A. Thomas, A. S. John, T. A. Roach, B. Kupfer, G. Luzzi, A. Edwards, G. Taylor, H. Lyall, G. Tudor-Williams, V.

J. VIROL.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29. 30.

31.

32.

33.

34.

35.

36. 37.

Novelli, J. Martinez-Picado, P. Kiepiela, B. D. Walker, and P. J. R. Goulder. 2004. HIV evolution: CTL escape mutation and reversion after transmission. Nat. Med. 10:282–289. Lund, O., M. Nielsen, C. Kesmir, A. Petersen, C. Lundegaard, P. Worning, C. Sylvester-Hvid, K. Lamberth, G. Røder, S. Justesen, S. Buus, and S. Brunak. 2004. Definition of supertypes for HLA molecules using clustering of specificity matrices. Immunogenetics 55:797–810. Malhotra, U., F. Li, J. Nolin, M. Allison, H. Zhao, J. I. Mullins, S. Self, and M. J. McElrath. 2007. Enhanced detection of human immunodeficiency virus type 1 (HIV-1) Nef-specific T cells recognizing multiple variants in early HIV-1 infection. J. Virol. 81:5225–5237. Matthews, P. C., A. Prendergast, A. Leslie, H. Crawford, R. Payne, C. Rousseau, M. Rolland, I. Honeyborne, J. Carlson, C. Kadie, C. Brander, K. Bishop, N. Mlotshwa, J. I. Mullins, H. Coovadia, T. Ndung’u, B. D. Walker, D. Heckerman, and P. J. R. Goulder. 2008. Central role of reverting mutations in HLA associations with human immunodeficiency virus set point. J. Virol. 82:8548–8559. Maurer, K., E. G. Harrer, A. Goldwich, K. Eismann, S. Bergmann, M. Schmitt-Haendle, B. Spriewald, S. M. Mueller, and T. Harrer. 2008. Role of cytotoxic T-lymphocyte-mediated immune selection in a dominant human leukocyte antigen-B8-restricted cytotoxic T-lymphocyte epitope in Nef. J. Acquir. Immune Defic. Syndr. 48:133–141. Meyer-Olson, D., K. W. Brady, M. T. Bartman, K. M. O’Sullivan, B. C. Simons, J. A. Conrad, C. B. Duncan, S. Lorey, A. Siddique, R. Draenert, M. Addo, M. Altfeld, E. Rosenberg, T. M. Allen, B. D. Walker, and S. A. Kalams. 2006. Fluctuations of functionally distinct CD8⫹ T-cell clonotypes demonstrate flexibility of the HIV-specific TCR repertoire. Blood 107:2373–2383. Miura, T., M. A. Brockman, A. Schneidewind, M. Lobritz, F. Pereyra, A. Rathod, B. L. Block, Z. L. Brumme, C. J. Brumme, B. Baker, A. C. Rothchild, B. Li, A. Trocha, E. Cutrell, N. Frahm, C. Brander, I. Toth, E. J. Arts, T. M. Allen, and B. D. Walker. 2009. 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. Miura, T., C. J. Brumme, M. A. Brockman, Z. L. Brumme, F. Pereyra, B. L. Block, A. Trocha, M. John, S. Mallal, P. R. Harrigan, and B. D. Walker. 2009. HLA-associated viral mutations are common in human immunodeficiency virus type 1 elite controllers. J. Virol. 83:3407–3412. Panel on Antiretroviral Guidelines for Adults and Adolescents. 3 November 2008. Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. Department of Health and Human Services, Washington, DC. http://www.aidsinfo.nih.gov/Guidelines/GuidelineDetail.aspx?GuidelineID⫽7. Polk, B. F., R. Fox, R. Brookmeyer, S. Kanchanaraksa, R. Kaslow, B. Visscher, C. Rinaldo, and J. Phair. 1987. Predictors of the acquired immunodeficiency syndrome developing in a cohort of seropositive homosexual men. N. Engl. J. Med. 316:61–66. Sette, A., and J. Sidney. 1998. HLA supertypes and supermotifs: a functional perspective on HLA polymorphism. Curr. Opin. Immunol. 10:478–482. Simons, B. C., S. E. VanCompernolle, R. M. Smith, J. Wei, L. Barnett, S. L. Lorey, D. Meyer-Olson, and S. A. Kalams. 2008. Despite biased TRBV gene usage against a dominant HLA B57-restricted epitope, TCR diversity can provide recognition of circulating epitope variants. J. Immunol. 181:5137– 5146. Smith, W. P., Q. Vu, S. S. Li, J. A. Hansen, L. P. Zhao, and D. E. Geraghty. 2006. Toward understanding MHC disease associations: partial resequencing of 46 distinct HLA haplotypes. Genomics 87:561–571. Stekler, J., B. J. Sycks, S. Holte, J. Maenza, C. E. Stevens, J. Dragavon, A. C. Collier, and R. W. Coombs. 2008. HIV dynamics in seminal plasma during primary HIV infection. AIDS Res. Hum. Retroviruses 24:1269–1274. Streeck, H., Z. L. Brumme, M. Anastario, K. W. Cohen, J. S. Jolin, A. Meier, C. J. Brumme, E. S. Rosenberg, G. Alter, T. M. Allen, B. D. Walker, and M. Altfeld. 2008. Antigen load and viral sequence diversification determine the functional profile of HIV-1 specific CD8⫹ T cells. PLoS Med. 5:e100. Taylor, J. M., K. Schwartz, and R. Detels. 1986. The time from infection with human immunodeficiency virus (HIV) to the onset of AIDS. J. Infect. Dis. 154:694–697. Turner, S., M. E. Ellexson, H. D. Hickman, D. A. Sidebottom, M. FernandezVina, D. L. Confer, and W. H. Hildebrand. 1998. Sequence-based typing provides a new look at HLA-C diversity. J. Immunol. 161:1406–1413. UNAIDS. 2008. Report on the global AIDS epidemic. UNAIDS, Geneva, Switzerland. Wang, Y. E., B. Li, J. M. Carlson, H. Streeck, A. D. Gladden, R. Goodman, A. Schneidewind, K. A. Power, I. Toth, N. Frahm, G. Alter, C. Brander, M. Carrington, B. D. Walker, M. Altfeld, D. Heckerman, and T. M. Allen. 2009. Protective HLA class I alleles that restrict acute-phase CD8⫹ T-cell responses are associated with viral escape mutations located in highly conserved regions of human immunodeficiency virus type 1. J. Virol. 83:1845– 1855.