Despite Biased TRBV Gene Usage against a Dominant HLA B57-Restricted Epitope, TCR Diversity Can Provide Recognition of Circulating Epitope Variants¹² Free

Chronic viral infections with extensive variation, such as hepatitis C and HIV, pose a great challenge for the cellular immune response (1, 2, 3, 4). Several studies have attempted to link the quantity, typically measured by the frequency of virus-specific cytokine-producing cells, and quality of host responses, assessed by either the ability to secrete a diverse array of cytokines or by the proliferative capacity of virus-specific T cells, to the level of control of viremia (5, 6, 7, 8). Our laboratory has recently demonstrated that the level of epitope-specific TCR repertoire diversity during acute hepatitis C infection may be critical for limiting immune escape and for subsequent control of viral replication (9). These findings have been replicated in the SIV acute infection model (9, 10, 11). However, the level of TCR diversity in subjects with long-term control of HIV-1 viremia, and its potential role in the control of viremia has not been completely defined.

Efforts to understand TCR repertoire diversity have focused on epitopes frequently recognized by subjects with a particular HLA allele. HLA B57 has shown the strongest association with control of viremia, and subjects with this allele typically have robust HLA B57-restricted CD8⁺ T cell responses (12, 13, 14, 15). Furthermore, several HLA B57 epitopes have been fine-mapped, including a dominantly recognized, highly conserved epitope located in p24 Gag (KAFSPEVIPMF; KF11) (15, 16). Patterns of TCR usage directed against this epitope may shed light on how TCR recruitment over the course of infection mediates control of viremia.

The TCR gene usage of KF11-specific immune responses has been assessed for shared motifs. In previous work, Gillespie et al. (17) found conserved TCR variable region β (TRBV)⁴ usage (TRBV19) and highly conserved CDR3 region motifs among isolated CTL clones from three out of five subjects recognizing the KF11 epitope. Yu et al. (18) more recently studied TCR usage of KF11-specific T cells and found that HLA B*5701 subjects had striking usage of TRBV19 with shared CDR3 motifs, very little variation of the circulating KF11 epitope, and these TCRs displayed cross reactivity to published KF11 variants. In contrast, HLA B*5703-restricted responses had more diverse TCR usage and more in vivo variation of the KF11 epitope, yet were unable to recognize in vivo HIV variants. These researchers concluded that a two amino acid difference between the HLA B*5701 and B*5703 alleles was likely responsible for the HIV epitope variation seen in HLA B*5703 subjects, and consequently these subjects had more diverse TCR repertoires (18). While it remains unclear whether TCR diversity is a prerequisite for control of HIV-1 viremia, the structure of the TCR repertoire is the driving force behind the immune system’s ability to recognize and respond to virus variants. Therefore, understanding how the selection of TCR repertoires influences the host’s ability to contend with HIV variation is important for understanding the correlates of control of viremia.

We evaluated the TCR repertoires of HLA B57-KF11-specific CD8⁺ T cells in subjects with either the HLA B*5701 or the HLA B*5703 allele. Our goal was to determine whether the diversity of the TCR repertoire specific for this immunodominant epitope was directly related to control of viremia. In addition to a detailed TCR repertoire analysis of directly sorted T cells, we sequenced autologous virus, performed detailed tetramer off-rate analyses, and functional avidity assays by ELISPOT. We identified common usage of the TCR β-chain TRBV7 (IMGT) within the KF11-specific repertoires of eight of the nine HLA B57 subjects examined, regardless of whether these subjects possessed the HLA B*5701 or HLA B*5703 allele. We found a wide range in the number of epitope-specific TCR clonotypes within each subject, and no apparent structural or functional advantage to this TRBV7 usage. However, analysis of the functional avidity for a KF11 variant we observed in our cohort, K162R, revealed a potential role for clonotype diversity in the recognition of viral variants during chronic infection. These data suggest some degree of selective recruitment of TCR for a conserved epitope and highlight convergent TCR usage in subjects with a favorable disease course.

The Vanderbilt-Meharry CFAR Cohort was comprised of subjects recruited through the Comprehensive Care Center (Nashville, TN) and all subjects were HLA class I typed (four-digit resolution) (DCI Tissue Typing Laboratory; Nashville, TN). Based on HLA typing results, 17 HLA B57 subjects were selected for further study. All subjects were antiretroviral therapy naive at the time of study with a range of CD4⁺ T cell numbers from 144 to 1260/mm³ and log viral load measurements from 1.7 to 4.25 copies/ml. This study was approved by the Vanderbilt University Medical Institutional Review Board, and all subjects provided informed consent.

Autologous virus was population sequenced from plasma RNA. Viral RNA was isolated from plasma and reverse transcribed as described (19). Gag DNA was amplified by PCR with the following primers: 5gag7–28 5′-GCG AGA GCG TCA GTA TTA AGC G-3′ and 3gag1668–1693 5′ TCT GAG GGA AGC TAA AGG ATA CAG TT-3′. PCR fragments were then gel purified and sequenced bi-directionally on an ABI 3100 PRISM automated sequencer. Sequencher (Gene Codes) was used to edit and align sequences.

Fresh or cryopreserved PBMC samples were first CD8⁺ T cell enriched by magnetic separation (Robosep, Stem Cell Technologies), and then stained with PE-labeled B*5701/KF11 tetramer, which has been previously shown to bind equally to B5701- and B5703-restricted KF11-specific CD8⁺ T cells (20). Tetramer-positive CD8⁺ T cells, as well as an equal number of tetramer-negative CD8⁺ T cells (as a negative control) were sorted by a FACSAria instrument (BD Biosciences) under BSL3 conditions directly into STAT 60. Electronic compensation was performed with PBMC from the same subject stained separately with individual Abs used in the test samples. The purity of sorted cell populations was consistently >95%.

RNA was extracted from purified T cells using STAT-60 (Tel-Test B). A modified anchored RT-PCR was performed with Powerscript Reverse transcriptase (Clontech) from total RNA as previously described (9) using a gene-specific primer for the β constant region with a modified cDNA anchor primer (Clontech). Negative controls were included at all amplification steps. Amplification of cDNA by PCR was performed using TCR constant region based primers and an anchor-specific primer 5′-AAT CCT TTC TCT TGA CCA TG-3′. PCR products of 600 to 700 base pairs were gel purified and cloned using the TOPO TA cloning kit (Invitrogen). Selected colonies were sequenced using Taq DyeDeoxy Terminator cycle sequencing kit (PE Applied Biosystems) and capillary electrophoresis on an ABI 3700 PRISM automated sequencer (PE Applied Biosystems). Sequences were edited and aligned using Sequencher (Gene Codes) and compared with the human TRBV genes database (http://imgt.cines.fr). To accommodate for different CDR3 regions length, an alignment using clustalW was performed. Positions with >50% gaps were excluded from analysis to prevent any substantial bias introduced by minor populations. The TRBV classification system is that of the international ImMunoGeneTics database.

Amino acid variability in TCR CDR3 regions was determined using the Shannon entropy (H) calculation for protein sites as described previously (19) by the formula H = −∑π log2 π where π is the fraction of residues at a site that is amino acid type i. For the 20 amino acids, H can range from 0 (site contains only one amino acid in all sequences) to 4.32 (all amino acids are represented equally at this site). Positions that contained >50% gaps were excluded from analysis. In the comparative analysis of TRBV population off rates, the Mann-Whitney U test was used to calculate the difference in means and the Fligner-Kileen test was used to compare the difference in variance between the TRBV population off rates.

Cryopreserved PBMC were thawed and resuspended at a concentration of 10⁷/ml in R10 medium and stained with PE-KF11 Tetramer (Beckman Coulter) at a concentration of 1:100 in FACS buffer plus sodium azide (1%). Remaining surface Abs, including the appropriate Vβ Ab, were added 10 min after the initial tetramer stain. Vβ Ab (BD Biosciences) selection was determined by sequencing data obtained for KF11-specific CD8⁺ T cells (see Figs. 2 and 4A). An individual well was set up for each Vβ population to determine Vβ off rates. To determine the TRBV7 off rate, all Abs corresponding to the identified Vβ populations by sequencing were including in one condition, and the “non-Vβ” population was gated to measure loss of fluorescence (Supplemental Table I).⁵ PBMC were then washed and resuspended in 100 μl of a 1:4 concentration of APC KF11 Tetramer (Beckman Coulter) in FACS buffer (PBS/2% FCS/0.1% sodium azide). PBMC were then incubated at 37°C and at each time point (0, 10, 20, 40, and 80 min), 20 μl of the mixture was removed and placed into 180 μl of 1% paraformaldehyde. Samples were analyzed on a FACSAria using FACSDiva software. CD8⁺/tetramer⁺/Vβ⁺ cells were gated and this gate extended over the entire range of PE expression levels. The geometric mean of PE fluorescence was measured over time and normalized to time 0 as previously described (21, 22). Tetramer off-rates were calculated with Graph Prism software, first rate kinetics equations. Individual data sets were normalized to compare off rates among subjects and individual cell populations.

Ninety-six-well MultiScreen filtration plates (Millipore) were coated with 0.1 μg/ml of an anti-human IFN-γ mAb (Mabtech). CD8⁺-depleted PBMC were added at a concentration 100,000 cells per well in a volume of 100 μl of RPMI 1640 medium supplemented with FCS (10%), HEPES buffer (10 mM), l-glutamine (2 mM), and penicillin-streptomycin (50 U/ml) (R10 medium). Cryopreserved PBMC were CD8⁺ enriched (Stem Cell Technologies) by magnetic separation, sorted on a FACSAria to >98% purity by Vβ specificity, and added back to the ELISPOT plate in the appropriate wells. Vβ specificity and selection were based on the individual Vβ populations identified by direct sequencing of the KF11- tetramer specific populations (Supplemental Table I). To ensure a positive and interpretable response, the number of selected CD8⁺ T cells added back was individually calculated based on tetramer percentage as well as clonotype frequency within that tetramer population. A positive response was defined as a minimum of 50 spot forming cells (SFC)/10⁶ cells at all concentrations and three times above background. Peptides were serially diluted in R10 medium and added to each well in a volume of 10 μl. Plates were incubated over night at 37°C in 5% CO₂ and developed the following day (23). Wells containing PBMC and medium with SEB or without any peptide were used as positive and negative controls respectively, and run in duplicate on each plate. Peptide-stimulated, CD8⁺-depleted PBMC were also included as a negative control. To calculate the number of specific T cells, the number of spots in the negative control wells was subtracted from the counted number of spots in each well. All negative controls were <30 SFC/10⁶ cells.

To examine the KF11 specific response, we identified 17 subjects with expression of the HLA B57 allele from a cohort of 146 therapy-naive subjects and evaluated their response to HIV-1 HLA class I-restricted epitopes. We assessed these subjects’ abilities to recognize HIV-1 peptides by IFN-γ ELISPOT, and compared the sum of all HLA B57 restricted responses to the total IFN-γ response directed against all HLA restricted peptides. The total B57-restricted responses typically made up the majority of the overall HLA-restricted responses (average 74%) (Fig. 1,A). This confirms recent findings describing the immunodominance of HLA B57-restricted responses (14). We were able to longitudinally follow eight of these subjects. In seven of the eight subjects, this HLA B57 dominance remained stable over a 10–21 mo follow-up period (Fig. 1,B). At the epitope level, these subjects had dominant recognition of the HLA B57-restricted KF11 peptide. Fourteen of the 17 subjects tested recognized this epitope, and in 10 of these subjects, it was the highest magnitude response. Of the eight longitudinally followed subjects, six subjects that recognized KF11 maintained dominant KF11 responses over time (Fig. 1 C).

Having identified subjects with robust recognition of the KF11 epitope, we next evaluated the diversity of these immune responses at the clonotype level. We directly sorted KF11-specific T cells from nine chronically infected HIV⁺ B57 subjects (Table I and Fig. 2). In parallel, an equivalent number of KF11 tetramer-depleted CD8⁺ T cells were sorted and also subjected to TCR sequence analysis. At least 45 sequences from each sorted population of cells were analyzed for each individual subject. Sequences were assessed for diversity by calculating the entropy of aligned Vβ CDR3 sequences and categorized as individual clonotypes by identification of CDR3 nucleotide sequences. The mean (and median) number of TCR clonotypes utilized was 5, with a range of 1 (subject 10067) to 10 clonotypes (10070 and 20018) within each KF11-specific repertoire (Fig. 3). Despite a previous report suggesting differences in TRBV usage between the KF11 specific TCR repertoires of B*5701 and B*5703 subjects (18), we found no such differences in our subject cohort. In eight of the nine subjects, regardless of the level of clonotype diversity as defined by distinct CDR3 gene usage, clonotypes tended to use the TRBV7 gene. Only one individual, subject 10027 did not use TRBV7 for this response. We could not identify common motifs within the CDR3 regions of the KF11-specific TCR repertoires in our subject group (Fig. 3). However, when compared with the B*5703 subject group published by Yu et al. (18), we found four CDR3 motif similarities, and shared TRBJ usage among our subjects. Only subjects 20018 and 10070 shared some of the following CDR3 motifs with the previously published B*5703 cohort (18): TRBV7–6-ASSSW-X-G-X-D-X-Q-X (TRBJ 2–1): TRBV7–9-ASS-XX-GGYT (TRBJ 1.2): TRBV24–1- ATSDL-XXX-QF (TRBJ 2.1): TRBV7–9-ASE-X-GNTIY (TRBJ 1.3) (18). Despite this occasional presence of shared motifs, these clonotypes did not make up the majority of sequences within each KF11-specific population.

Of subjects with TRBV7 clonotypes within KF11 specific repertoires, the frequency of TRBV7 usage among TCR sequences ranged from 25 (subject 10071) to 100% (subject 10067). We found no relationship between the diversity of KF11-specific TRBV7-clonotypes or the percentage of TRBV7-using tetramer⁺ cells and markers of HIV-1 disease progression such as concurrent CD4⁺ T cell number or viral load (Table I). To determine whether the high frequency of TRBV7 within KF11 specific repertoires was a true usage bias, 12 other MHC class I-restricted epitope specific repertoires were analyzed (three non-KF11-specific HLA B57-restricted responses, and nine non-HLA B57-restricted responses). The TCR sequences derived from these responses demonstrated TRBV7 usage in only two cases compared with eight of the nine KF11 specific repertoires (p = 0.001, Fisher’s exact test, data not shown), suggesting that TRBV7 usage bias is a feature of the KF11-specific immune response regardless of the HLA B57 subtype.

We next evaluated the entropy of TCR sequences specific for KF11. Prior studies from our laboratory and others (9, 24, 25) suggested diverse TCR usage during acute infection could limit epitope escape and potentially contribute to control of viremia, but there are limited data evaluating TCR repertoires during HIV-1 chronic infection (17, 26, 27). The range in mean entropy of KF11 specific sequences from the nine subjects was 0 to 1.04 (Fig. 4,A) and there was a strong correlation (p = 0.015) between the entropy values and the number of clonotypes within each individual repertoire (Fig. 4,B). KF11-specific TCR entropy values were substantially lower than those of equivalent numbers of tetramer-depleted CD8⁺ T cells (p < 0.001 Mann-Whitney) derived from the same sort, and ruled against PCR bias (Fig. 4 C).

To determine the definition of narrow and broad repertoires defined by mean entropy calculations, we compared the entropies of KF11-specific repertoires to the entropies of 12 other epitope specific TCR repertoires (three HLA B57-restricted and nine non-HLA B57-restricted responses), as well as to entropies we calculated based on recently published results by Yu et al. (18). The range in entropy was similar between the two B57 cohorts (Yu et al.; 0.00–1.37). We found no difference in mean entropy values between the KF11 specific repertoires and the other epitope specific repertoires (p = 0.28, Mann-Whitney), indicating that KF11 specific TCR repertoires have similar diversity to that of other epitope-specific repertoires. We also found no relationship between the degree of KF11-specific TCR repertoire diversity and markers for disease progression such as CD4 count or viral load in the nine subjects. Although the level of KF11 specific CDR3 diversity was not distinct from other epitope specific sequences, our observation of TRBV7 usage bias prompted a more detailed structural and functional analysis of KF11-specific cells at the clonotype level.

To determine whether shared TRBV7 usage corresponds to distinct structural or functional characteristics of subpopulations of tetramer positive cells, we conducted tetramer off-rate experiments that included TCR Vβ staining and analysis by flow cytometry. Although Abs to TRBV7 are not currently available, in each case TRBV Abs were available for the corresponding non-TRBV7 TCR sequence derived from directly sorted tetramer⁺ cells (Supplemental Table I). In brief, our gating scheme encompassed the entire breadth of PE fluorescence of tetramer⁺ Vβ⁺ populations, and the geometric mean fluorescence was used as a raw measurement of tetramer decay. Raw measurements were normalized to time 0 and off rates were calculated using first-rate kinetics (GraphPrism; GraphPad Software).

In general, the tetramer off-rates of KF11-specific CD8⁺ T cells with TRBV7 usage were faster than non-TRBV-using cells, suggesting TRBV7 clonotypes have a lower structural avidity for the HLA B57: KF11 complex, but this difference in off-rates was not statistically significant (TRBV7 half life: 20 min vs nonTRBV7 half life: 35 min, p = 0.17, Mann-Whitney) (Fig. 5). In addition, the variance of off-rates among TRBV7-using KF11-specific populations was low as compared with the non-TRBV7 populations (p = 0.015, Fligner-Killeen test of homogeneity of variances), indicating structural similarity among TRBV7 clonotypes. We found no relationship between TRBV7 structural avidity (half-life) and HIV viral load.

We next assessed whether T cell clonotype structural avidity was related to functional avidity, which was assessed by IFN-γ ELISPOT. HIV Gag was sequenced from the plasma from six of the nine study subjects. We were unable to obtain Gag sequences from subjects with the lowest viral loads in our study cohort (10004, 10071, and 10067), despite efforts that included large volume plasma concentration and nested PCR specific for proviral gag DNA. KF11 has been described as a highly conserved epitope, and we found the majority of subjects had the consensus HIV clade B sequence. However, two subjects (subjects 10024 and 10076) harbored a K162R mutation (Table I). Due to the detailed nature of the sorting experiments, and the relatively large cell requirements, we focused our subsequent experiments on these two peptide sequences.

Our initial studies included whole PBMC IFN-γ ELISPOT with both the consensus KF11 (KAFSPEVIPMF) and the K162R variant (RAFSPEVIPMF). Results from whole PBMC IFN-γ ELISPOT indicated universal recognition of both the consensus KF11 and the K162R variant in all subjects. The magnitude of the responses to KF11 consensus peptide in subjects tended to be equal or higher than that of responses directed against the K162R variant (Fig. 6). Of the two subjects with a dominant circulating K162R variant, subject 10024 had a higher magnitude of response to K162R (maximal response 1,450 SFC/10⁶ PBMC) as compared with KF11 consensus peptide (maximal response 495 SFC/10⁶ PBMC), but the functional avidity of the consensus KF11 peptide (SD50 20 ng/ml) was higher than that of the K162R variant (SD50 2500 ng/ml). Conversely, subject 10076 showed preferential recognition of the consensus KF11 peptide (Fig. 6), with a higher magnitude of response (maximal response 7750 SFC/10⁶ PBMC), however functional avidity (SD50 4 ng/ml) was similar when compared with K162R recognition (maximal response 2500 SFC/10⁶ PBMC, SD50 5 ng/ml). With such a high incidence of cross-recognition of K162R, we wanted to determine whether cross-reactivity to this variant was mediated by the TCR using the dominant TRBV7 gene.

To evaluate differences in peptide recognition by T cell clonotypes, we performed a series of flow sorting experiments in which CD8⁺ T cells were separated and evaluated based on their TCR Vβ expression. Once we determined the TCR gene usage of the tetramer⁺ cell populations in each individual, we used this information to either positively select or deplete CD8⁺ T cells of the corresponding TRBV populations. For this series of experiments, we did not simultaneously stain with the KF11 tetramer since tetramer binding led to activation and IFN-γ secretion from the tetramer positive cells (data not shown). Because a commercial TRBV7 (Vβ6) Ab is not available, for each subject we depleted CD8⁺ T cells of the entire TRBV population corresponding to the “non-TRBV7” population of tetramer positive cells. For example, total CD8⁺ T cells from subject 10071 were depleted of all TRBV5–1 using T cells, thereby leaving behind all KF11-specific TRBV7 expressing cells. Negative controls were set up to ensure that responses were only elicited from KF11 specific CD8⁺ T cells (described in Materials and Methods). These two populations of CD8⁺ T cells (in this case the TRBV5–1 CD8⁺ sorted population as well as the CD8⁺ TRBV5–1 depleted population) were added back to CD8⁻depleted PBMC pulsed with serially diluted peptides, and evaluated by IFN-γ ELISPOT assay.

The consensus KF11 peptide was recognized by all identified KF11-specific clonotypes. In subject 10071, both TRBV5–1 and TRBV7-utilizing CD8⁺ T cells recognized the KF11 peptide with similar functional avidities with SD50 values of 3 ng/ml (TRBV5–1 clonotype) and 6 ng/ml (TRBV7 clonotype) (Fig. 7,A). In contrast, the circulating clonotypes in subjects10024 (Fig. 7,B) and 10076 (Fig. 7 C) had different SD50 values. In subject 10024, the TRBV7 clonotype exhibited higher avidity (SD50 0.6 ng/ml) than TRBV3 (SD50 7.9 ng/ml). In subject 10076, the TRBV7 clonotype exhibited lower avidity (SD50 50 ng/ml) than TRBV2 (SD50 5 ng/ml). Because a prior study suggested that dominant TRBV usage may imply cross-reactivity against epitope variants (28), we next evaluated recognition of the only epitope variant present in our study subject population, K162R.

Although all clonotypes were able to recognize the KF11 consensus peptide, not all were able to recognize epitope variants. In subject 10071, both clonotypes were able to recognize the K162R variant peptide. The TRBV5–1 clonotype maintained higher functional avidity (SD50 0.6 ng/ml) than the TRBV7 clonotype (SD50 8 ng/ml) (Fig. 7,A). Analysis of subject 10024 demonstrated that the TRBV7 clonotype had a higher K162R avidity, with a SD50 of 1.25 ng/ml, as compared with the TRBV3 clonotype with a SD50 of 251 ng/ml (Fig. 7,B). Results from subject 10076 indicated that only the TRBV2 clonotype recognized the variant with an SD50 of 8 ng/ml (Fig. 7,C). The TRBV7 population was unable to recognize the variant, even at the highest concentration despite recognition of the consensus KF11 peptide (Fig. 7 C). This series of experiments demonstrates that functional avidity and cross-reactivity of an epitope-specific response can vary at the clonotype level.

Epitope-specific TCR diversity has been shown to influence the course of chronic viral infections. In the hepatitis C chimpanzee model, our laboratory found that narrow TCR diversity during the early phase of infection was associated with subsequent immune escape and the establishment of chronic viral infection, whereas broader epitope-specific TCR diversity was associated with a lack of escape or resolution of infection (9). In the SIV model, the level of epitope-specific TCR diversity during early infection is linked to mutations within dominant MHC class I-restricted epitopes (11, 29, 30). The TCR repertoires of two dominantly recognized Mamu A*01-restricted epitopes CM9 (Gag) and TL8 (Tat) differ in their level of TCR diversity, as the CM9 specific repertoire is diverse in comparison to the TL8 specific repertoire (11). In vitro studies suggested that TL8-specific CD8⁺ T cells were more effective suppressing viral replication; however, the TL8 epitope mutation is known to occur very early in the course of SIV infection, providing escape with little fitness cost to the virus (29). This emphasizes that a limited TCR repertoire may be quite effective at suppressing viral replication, but in the environment of extensive variation and high-level viral replication in vivo, the limited repertoire is more susceptible to immune escape. It is unknown whether broad TCR repertoires directed against conserved epitopes are maintained throughout the course of chronic infection in this model.

The HLA B57-restricted epitope KF11 is both dominantly recognized and highly conserved (15). In this study. we demonstrate a wide range of clonotypic diversity among subjects able to recognize this epitope. However, despite this clonotypic diversity, we found dominant TRBV7 usage among KF11-specific clonotypes. HLA B57- restricted responses, and responses to the KF11 epitope in particular, are dominant in HIV-1⁺ individuals with chronic infection (14). Migueles et al. (15) observed conservation of HLA B*5701-restricted epitopes and the responses directed against these epitopes during chronic infection in both HLA B*5701 long-term nonprogressors and progressors, suggesting that B*5701 epitope variation does not contribute to disease progression in these subjects. Even in the case of dominant circulating epitope variants, immune responses against a consensus peptide can remain dominant during chronic infection, as Koibuchi et al. (31) demonstrated in two non-HLA B57 subjects followed longitudinally over 6 years. We likewise observed stably dominant KF11 responses in all six subjects that recognized this epitope, despite mutations within KF11 in the circulating plasma viral populations of two of these subjects (Table I).

We found no difference in TRBV usage between HLA B*5701 and HLA B*5703 subtypes. This observation is in contrast to recent findings suggesting the dominant usage of TRBV19 (17, 18) and structural differences in TCR repertoires among HLA B*5701 controllers when compared with subjects with the B*5703 allele (18). We only found TRBV19 usage in one HLA B*5701 subject, and in this case it made up a very small proportion of the KF11-specific TCR repertoire (2 out of 50 TCR sequences in subject 10004). Instead, we found dominant TRBV7 usage in our subjects regardless of HLA B57 subtype, results which are similar to the TCR usage previously described in B*5703 subjects (18). The reason for these discordant results is not clear, but may be due to the shorter duration of infection, or early initiation of anti-retroviral therapy during acute HIV infection in some cohorts (18, 32, 33, 34), which may influence the development of the HIV specific CD8⁺ TCR repertoire (35, 36). Our subjects were completely anti-retroviral therapy naive, and were infected a mean of 13 years (range 3–24). In this regard, the natural infection history of our study subjects may more closely resemble the B*5703 African subjects examined by Yu et al. (18). It is therefore possible that the duration of chronic infection in the absence of early anti-retroviral therapy has a greater influence on the make-up of the TCR repertoire than HLA subtype or the clade of HIV infection.

Common TRBV19 usage among KF11 specific T cell clones was demonstrated by Gillespie et al. (17). In that study, TRBV usage was first analyzed by Vβ staining. Based on the Vβ staining results, appropriate TRBV specific primers were subsequently used for TRBV sequencing (17). There is currently no commercially available TRBV7 (Vβ6) Ab, and the current panels of TRBV Abs only cover ∼70% of expressed TRBV chains. Thus, Ab screening would overlook TRBV7 usage. Until more Abs are available, direct sorting of epitope-specific cells is more reliable for TCR repertoire analysis.

The duration of HIV infection before study initiation may also determine the level of TCR diversity. We found no difference in TCR CDR3 variability (entropy analysis) between responses in subjects with HLA B*5701 and B*5703 subtypes. We performed a similar analysis on the data published by Yu et al. and likewise found no difference in entropies between responses restricted by the different HLA subtypes (p = 0.28, Mann-Whitney). Yu et al. also demonstrated that six of ten HLA B*5703 subjects used the TRBV7 genes without evidence of CDR3 motifs for this response. The non-TRBV7 sequences showed a wide variety of TRBV genes, likewise without evidence of CDR3 motifs, which is concordant with our results. However, when we compared cohorts, we found several shared CDR3 motifs as well as corresponding TRBJ regions among three TRBV7 clonotypes and one TRBV24–1 clonotype between subjects 10070 (B*5703) and 20018 (B*5701) from our study and subjects within the cohort published by Yu et al., despite differences in B57 subtypes. This provides evidence for convergent TCR usage beyond the TRBV portion of the epitope-specific TCR.

In our study, we found KF11-specific TCR diversity not to correspond with viral load, CD4 count, or duration of infection. Yet, because the HIV specific TCR repertoire shapes and directs the HIV-specific CTL immune response, we extended our analysis to the structural and functional aspects of the KF11-specific TCR repertoire at the clonotype level. TRBV7 structural avidity was assessed by a tetramer off-rate assay similar to recently published methods (37). Recent findings have indicated high structural avidity as a beneficial characteristic of HIV specific CTL (5, 37, 38, 39, 40), as well as convergent evolution of epitope-specific responses in murine models and human influenza studies (41, 42, 43). These data suggest a beneficial role of common TRBV usage against conserved epitopes, perhaps as a result of greater functional capacity. However, we found TRBV7 clonotypes generally had a lower structural avidity than other KF11 specific TRBV clonotypes. The variance in tetramer off-rates among TRBV7 clonotypes was significantly narrower than those of the non-TRBV7 clonotypes (p = 0.015, Fligner-Kileen) indicating a direct influence of TRBV structure on TCR avidity. This does not necessarily mean that strong structural avidity is not important for a particular immune response, but raises the possibility that an avidity threshold for optimal TCR activation exists, as has been described (44). HIV-1 infection has been associated with immune exhaustion, as measured by the expression of surface markers such as CD57 and PD-1 (45, 46, 47). When Ag is persistently present, it is possible that high structural avidity is disadvantageous, leading to constant activation and eventual exhaustion of the high avidity clonotypes (37).

One caveat of our off-rate experiments is the potential for Vβ Abs to influence the binding of the KF11 tetramer, and therefore influence the off-rate kinetics. Previous experiments in our laboratory in a subject that recognized a B*1501 tetramer, and where all the tetramer subpopulations could be labeled with available anti-TRBV Abs, indicated no effect of Vβ Abs on tetramer off rates (data not shown). However, to truly account for any influence of the Vβ Abs on tetramer off rate kinetics, one would need to analyze all Vβ Abs used. With only 70% of Vβ commercially available Abs, it would be difficult to conduct a thorough examination, especially in the case of TRBV7 where no Ab is currently available.

Despite the frequent TRBV7 usage of KF11-specific CD8⁺ T cells in our study cohort, TRBV7 usage does not necessarily correspond to recognition of epitope-variants. Prior studies have described narrow KF11-specific TCR repertoires in some individuals (18), as well as the ability of KF11-specific CD8⁺ T cells to recognize in vivo epitope variants (17, 18). It has therefore been suggested that the degree of epitope variant cross recognition is more important than the overall diversity of the KF11-specific TCR repertoire for control of viral replication. To further explore the epitope recognition of TRBV7 clonotypes, we assessed functional avidity of sorted subpopulations of tetramer⁺ cells to the epitope variants identified in our subject cohort. In the overall PBMC peptide titrations, all subjects recognized the KF11 consensus peptide as well as the K162R variant peptide, although KF11 recognition was generally higher than K162R in magnitude at maximal response (Fig. 6). However, at the clonotype level, differential recognition and functional avidity was observed. Subject 10071 recognized both KF11 and K162R, and both KF11-specific clonotypes (TRBV7 and TRBV5–1) demonstrated cross-recognition of the K162R variant. Results for subject 10024 were similar. In contrast, although both KF11-specific clonotypes from subject 10076 recognized the consensus KF11 peptide, the TRBV7 clonotype did not recognize the K162R variant even at the highest tested concentration. Both subjects 10076 and 10024 were identified to possess the K162R variant in circulating viral populations and subject 10076 has the highest viral load in our subject group (Table I). Due to sample availability and the extensive sorts that require large numbers of PBMC, we were only able to focus on the consensus KF11 and one KF11 variant we were able to identify in our cohort, thus our findings are limited to the two peptides analyzed. However, these data demonstrate that dominant clonotypes do not necessarily cross-react with epitope variants, which we have previously described at the clonal level (48), and suggest a potentially beneficial role for maintenance of TCR diversity even in the setting of restricted TRBV usage.

Although we could not find a relationship between TRBV7 usage within the KF11-specific repertoire and overall disease outcome, it appears to be a reproducible feature among chronically infected HLA B57 subjects (18). Our finding that TRBV7 clonotype avidity tended to be lower than that of the other KF11 specific clonotypes is consistent with the early deletion of high avidity T cell clonotypes after acute infection (37). Dominant TRBV7 usage may therefore represent convergent evolution toward populations of epitope-specific T cells with sufficient avidity to mediate control of viremia during chronic infection.

These studies highlight the importance of continuing analysis of the HIV-specific immune response at the clonotype level, which will help define how the development and maintenance of epitope-specific TCR repertoires influences HIV disease progression. As new assays are developed for the evaluation of vaccine-induced immune responses it will be important to maintain incorporation of clonotype analysis as a bridge between structure, phenotype, and the ability to control HIV-1 replication.

We thank Dr. Bryan Shepherd for statistical analysis and support (Department of Biostatistics, Vanderbilt University). We also thank Ruifeng Yang and the Aiken Laboratory, (Department of Microbiology and Immunology, Vanderbilt University) for technical support.

The authors have no financial conflict of interest.