Strong TCR Conservation and Altered T Cell Cross-Reactivity Characterize a B*57-Restricted Immune Response in HIV-1 Infection12

HLA-B*57 is associated with slower disease progression to AIDS, and CD8+ T cell responses to B*57-restricted epitopes are thought to contribute to this protective effect. In this study, we evaluate the B*57-restricted p24 KAFSPEVIPMF (KF11) immune response which is immunodominant during chronic infection. Previously, we observed that the KF11 clade variants KGFNPEVIPMF [A2G,S4N] and KAFNPEIIMPF [S4N,V7I], sharing a position 4 mutation, are differentially recognized by KF11-specific T cells. By combining structural and cellular studies, we now demonstrate that the KF11 and [A2G,S4N] epitopes induce distinct functional responses in [A2G,S4N] and KF11-specific T cells, respectively, despite minimal structural differences between the individual B*57-peptide complexes. Recently, we also elucidated the highly distinct structure of KF11 in complex with B*5703, and have now characterized the CD8+ T cell repertoire recognizing this epitope. We now report striking features of TCR conservation both in terms of TCR Vα and Vβ chain usage, and throughout the hypervariable region. Collectively, our findings highlight unusual features of the B*5701/B*5703-KF11-specific immune responses which could influence disease progression and that might be important to consider when designing future vaccine regimens.

T he MHC class I genes represent crucial host factors influencing disease outcome during HIV-1 infection. MHC class I molecules dictate the repertoire of epitopes presented to CD8 ϩ T cell and, therefore, the nature of the immune response against HIV. Maximum heterozygosity at the MHC class I loci is favorable, and individual MHC class I alleles also associate with prolonged AIDS-free survival, of which B*27 and B*57 are consistently described (1).
The influence of the MHC class I loci on clinical outcome warrants further investigation to further our understanding of disease pathogenesis and to direct the generation of effective vaccines. The heterozygous advantage at the MHC class I locus probably reflects the availability of distinct MHC class I molecules to present di-verse epitopes to T lymphocytes, resulting in broader immune responses, and delaying the incidence of viral immune evasion. The preference for individual MHC class I alleles might reflect their ability to present conserved viral epitopes to CD8 ϩ T cells; the conservation of a single immunodominant epitope, namely the gag-derived KRWIILGLNK (KK10) peptide, appears to account for the association of B*27 with prolonged AIDS-free survival in infection. KK10 requires a series of complex mutations, both internal and external to the epitope, to enable the emergence of a successfully mutated epitope that evades immune recognition with eventual onset of AIDS (2). The HLA class I subtype B*57 is consistently associated with slower disease progression in HIV-1 infection, particularly the B*5701 and B*5703 alleles which relate to better clinical outcome in Caucasian and African patients, respectively (3)(4)(5)(6). The mechanism(s) by which B*57 confers favorable outcome are not fully understood. In contrast to B*27, a large number of potent B*5701/03-restricted HIV-1 epitopes have been described, most notably the gag epitopes ISPRTLNAW (IW9), TSTLQEQIGW (TW10), and KAFSPEVIPMF (KF11), which are immunodominant, and map to the p24 capsid protein (7) (5,8). Previously, it was assumed these p24-derived epitopes were highly conserved, however, the loss of B*57 epitope recognition through escape mutation has recently been described. First, a single point mutation arising in the TW10 epitope (T242N) during the acute phase of HIV-1 infection results in escape from CD8 ϩ T cell recognition (9). Second, an epitope-flanking mutation adjacent to the IW9 epitope (A to P mutation) abrogates processing of the IW9 epitope in B and C clade infections. Although not associated with an increase in viral load, it has been suggested that the TW10 mutation impacts on viral fitness as it is lost upon transmission to B*57-negative individuals. The IW9 flanking mutation is associated with a 20-fold increase in viral load, although it is unclear whether this mutation precipitates clinical progression.
The focus of our study is the B*5701/03-restricted KF11 immune response which prevails during chronic HIV-1 infection (10,11). Importantly, we now show this to be the immunodominant B*57-restricted response in a cross-sectional study of chronically infected B*5701 ϩ slow progressors harboring both TW10-and IW9-mutated viruses.
KF11 is also interesting from a structural perspective. We have recently solved the crystal structure of B*5703 in complex with the KF11 epitope (12). KF11 adopts a distinctive conformation that differs significantly from previously published human peptide-MHC class I complexes. The central region of the epitope encompassing the position (p) 5 to p9 amino acids buckles high above the peptide-binding groove. Proline residues at p5 and p9 form the corner stones of the "bulge" structure, and stabilize the hairpin-like conformation. Together with peptide residues F3 and I8, they form a network of hydrophobic contacts at the core of the central bulge structure. As epitope residues E6 and V7 are located at the crest of the bulge, their side chains are solvent-exposed and accessible to responding TCRs.
Although KF11 is highly conserved in B clade viruses, diverse KF11 clade variants are documented in the Los Alamos Database. Previously, we assessed cross-reactivity to these variants, and observed broad but differential recognition of the epitopes (10). Interestingly, much of our functional data are readily explained by our structure analysis; mutations involving the p5 proline (P to Q mutation, (P5Q)) and the p3 phenylalanine (F to L mutation, (F3L)), for example, resulted in diminished T cell recognition, most likely due to destabilization of the central bulge structure. If the KF11 immune response is also associated with better immune control, it is important that we also consider the immune response to [A2G,S4N] particularly given the association of B*5703 with slow disease progression, and the prevalence of this viral clade in African cohorts. Determining the degree of KF11 and [A2G,S4N] cross-reactivity in B*5703 ϩ patients is also important to evaluate, particularly from the perspective of vaccine develop-ment. Consequently, in this study, we investigate the recognition of KF11 and [A2G,S4N] using a number of strategies. First, we performed functional studies of KF11 and [A2G,S4N] to determine whether CD8 ϩ T cell recognition patterns vary according to the prevailing virus sequence in B*5701 ϩ and B*5703 ϩ patients. Second, we performed structural studies to ascertain whether the peptide variants adopt different conformations when in complex with HLA-B*5703 and are differentially presented to CD8 ϩ T lymphocytes. Finally, we assessed TCR usage and CDR3 loop sequence analysis to determine whether recognition of these epitopes is a function of similar or distinct TCR clonotypes.

Subjects
Eleven B*57 ϩ subjects were enrolled in this study (Table I). Ethical approval was obtained from the relevant ethical committees, and all patients gave consent to donate blood. HLA class I typing was performed by the amplification refractory mutation systems PCR using sequence specific primers.

Isolation of PBMCs
PBMCs were isolated from heparinized venous blood by Ficoll-Hypaque (Nycomed) density gradient centrifugation. Cells were maintained in RPMI 1640 (Sigma-Aldrich) supplemented with 10% human serum 100 IU of penicillin/L, 100 mg of streptomycin/L (Invitrogen Life Technologies), referred to herein as H10.
Peptide synthesis B*57-restricted peptides were synthesized by F-moc chemistry using a Zinnser analytical synthesizer (Advanced Chemtech) and purity was determined by HPLC (Table II).

Generation of MHC class I tetramers
The generation of HLA-B*57 tetramers was performed according to previously described protocols (13,14). Avidin-PE (Sigma-Aldrich) and streptavidin allophycocyanin (BD Biosciences) were used as fluorochromes.

Establishment of CTL lines
A total of 5 ϫ 10 6 PBMCs were incubated with 50 M peptide at 37°C for 1 h, before resuspension in H10 supplemented with 25 ng/ml IL-7. Cells were fed on day 3 with rIL-2 at a concentration of 100 U/ml (10,15).

Peptide-based IFN-␥ ELISPOT assay
A standard ELISPOT assay was used to detect IFN-␥ release by PBMCs (16). Responses are reported as spot-forming units (SFUs) per million PB-MCs. SFUs double that observed with medium alone, and in excess of 20 per million PBMCs, were considered positive.

Intracellular cytokine staining.
A total of 200,000 T cell clones was incubated with 100,000 BCLs either pulsed with 10 M peptide or medium alone at 37°C. A total of 10 g/ml brefeldin A (Sigma-Aldrich) was added during the second hour of incubation. After 6 h, cells were washed in BD Perm/Wash buffer before fixation/permeabilization in Cytofix/Cytoperm buffer (BD Biosciences). Staining was performed for 30 min on ice using the following panel of mAbs: anti-human (h) IFN-␥ FITC, anti-hTNF-␣ PE (Dako), and anti-hCD8 PerCP (BD Biosciences). Cells were washed and stored in PBS-5% formaldehyde (17).

CD107a and CD107b expression
A total of 200,000 T cell clones was incubated either 100,000 autologous BCLs pulsed with 10 M peptide or with medium alone, plus 5 l of both CD107a-and CD107b-FITC-labeled mAbs at 37°C (BD Biosciences). Golgi-Stop (BD Biosciences) was added to each sample after the first hour of incubation (18). Following 6 h, cells stained with anti-hCD8 PerCP (BD Biosciences) and fixed in 5% formaldehyde-PBS before analysis.

Tetramer decay analysis
To test whether B*57 KF11 and B*57 A2G,S4N tetramers bound [A2G,S4N]specific T cell clones with similar affinities, a tetramer decay analysis was performed (19). A total of 2 ϫ 10 6 T cells from clone ML525 p5-10 was incubated with B*5703 KF11 and B*5703 A2G,S4N PE tetramers at 4°C for 45 min. Following extensive washing in PBA-0.1% BSA, cells were resuspended and aliquot t ϭ 0 was removed. A 5-fold excess of allophycocyanin-conjugated B*5703 A2G,S4N tetramer was added to each tube, and samples were incubated on ice. At 20-min intervals, aliquots corresponding to t ϭ 1, t ϭ 2, etc., were removed, washed, and fixed in PBS-5% formaldehyde. Tetramer binding was analyzed by flow cytometry, and the natural log of the mean geometric intensity of FL-2 immunofluorescence for each samples were plotted against sampling time.
Sequencing of p24 epitopes gDNA (proviral DNA) was isolated from PBMCs using the PureGene DNA Isolation kit (Gentra Systems) from which p24 was amplified by seminested PCR (10). The p24 5Ј and 3Ј outer primers include primer-GAGATA(A/C)(A/G)AGACACCAA(A/G)GAAGC and CATGCTGTC ATCATTTCTTCTA were used during the first round of PCR amplifications, and 5Ј primer CAGCCAAAATTACCCTATAGTGC plus 3Ј outer from the first-round amplifications were used during second-round PCRs. Nested p24 PCR products were cloned into TOPO TA vectors (Invitrogen Life Technologies). Ampicillin-resistant (Amp R ) colonies were expanded and plasmid DNA was isolated for sequence analysis.

Staining of tetramer-reactive T cells with V region-specific TCR mAbs
Staining of B*57-reactive T cells using V region-specific mAbs and tetramer was performed as described previously (20). A total of 5 ϫ 10 5 fresh PBMCs or 5 ϫ 10 4 CTL lines were stained with a panel of TCR-specific Abs and rabbit anti-mouse Ab conjugated to FITC before labeling with B*57 tetramers and anti-CD8 PerCP (BD Biosciences). Cells were fixed in PBS-5% formaldehyde solution.

Sequencing of TCR V␣ and V␤ segments
mRNA was isolated from 1 ϫ 10 6 CD8 ϩ T cell clones using TriReagent (Sigma-Aldrich) and cDNA was synthesized using the cDNA Cycle kit for RT-PCR (Invitrogen Life Technologies). A panel of V␤-and V␣-specific primers were used to screen for TCR usage by PCR analysis (21). PCR 5 Abbreviations used in this paper: BCL, autologous B cell; SFU, spot-forming unit; h, human; ICS, intracellular cytokine staining; FSC, forward scatter; SSC, side scatter.
a KF11 amino acid changes at positions 2, 4 and 7 are highlighted in boldface type.
amplification was conducted using Red Taq Ready Mix PCR Mix (Sigma-Aldrich). Control reactions included V region primers and H 2 O. TCR PCR products were cloned by TOPO TA cloning (Invitrogen Reaction), and plasmid DNA was isolated for sequencing analysis.

Protein crystallization and data collection
Crystallizations were done by the sitting drop vapor diffusion technique. After equilibration, and cross-seeding from wild-type HLA-B*5703-KF11 crystals, single crystals of HLA-B*5703-peptide complexes were obtained from a reservoir condition of 16% Peg 8000, 50 mM MES (pH 6.5). The largest crystals were soaked briefly and sequentially in mother liquor solutions supplemented with 10% and 20% glycerol, then flash-cooled and maintained at 100 K in a cryostream (Oxford Cryosystems  (22). Statistics are summarized in Table III.

Structure determination and analysis
Crystal structures were determined by molecular replacement using the program EPMR (23). The H chain and ␤2m domains from the HLA B*5703-KF11 structure (12) were used as the search probe (peptide and bound water coordinates omitted). This model yielded unambiguous solutions for both variant B*5703-peptide complexes with correlation coefficients of ϳ0.7 and R factors of ϳ35% for data between 30 and 4Å. Using CNS (24), the models were subjected to several rounds of rigid body refinement of individual domains (a1/a2, a3, ␤2m) after which Fo Ϫ Fc difference maps calculated for both complexes revealed the presence of the 11-mer peptides when displayed in the graphics program O (25). The peptides were modeled into the Fo Ϫ Fc difference density maps and further rounds of refinement were conducted using standard CNS protocols for bulk solvent correction and overall anisotropic B-factor scaling, positional refinement, simulated annealing and individual B-factor refinement. Manual refitting of the models was conducted using O. Conjugate gradient minimization refinement was performed at the final stages of crystallographic refinements after water picking with ARPw/ARP (26), using a restrained refinement algorithm in REFMAC5 (27). Structural superpositions were conducted using program SHP (28) and the figures were generated in Bobscript (29) Results B*57-specific CD8 ϩ T cell responses in B*5701 ϩ patients.
First, we assessed which of the previously mapped B*57 epitopes were immunodominant during chronic HIV-1 infection. We used the standard ELISPOT assay to evaluate IFN-␥ secretion in response to B*57-restricted epitopes including gag, nef, integrase, rev, vpr, vpu, and vif (Table II, "Previously defined B*57-restricted epitopes" section) in 9 of the 11 patients. Our results demonstrate that the p24 gag epitopes KAFSPEVIPMF (KF11) and ISPRTL NAW (IW9) were immunodominant both in terms of frequency and magnitude of recognition (Table IV).

Cellular analysis
Previously, we observed that the common A/C clade variant [A2G,S4N], and a second harboring a position 4 (p4) S to N mutation ( (10). Therefore, we questioned whether the amino acid change common to both variants, namely, the p4 residue, induced differential recognition. We generated KF11 variants with p4 mutations including A, D, I, Q, and T (Table II, "KF11 p4 variants" section), and used the standard ELISPOT assay to quantify IFN-␥ secretion in response to the p4 variant epitopes in patients' PBMCs where either KF11 or [A2G,S4N] represented the dominant viral species. Our results demonstrate the differential recognition of the p4 KF11 variants, which varied according to the dominant viral sequence in individual patients (Fig. 1) S4N]-specific T cell in individual cultures using tetramers. Interestingly, both cell lines grew at similar rates, and were highly cross-reactive as determined by double staining analysis using B*57 KF11 -and B*57 A2G,S4N -specific tetramers (Fig. 3A). One hundred percent of [A2G,S4N]-specific T cell cross-reacted with B*57-KF11-specific T cells, but a small population of non-crossreactive KF11-specific T lymphocytes were detected in both CTL   lines. However, both KF11-and [A2G,S4N]-derived CTL lines preferentially secreted IFN-␥ in response to [A2G,S4N], with limited recognition of KF11 (Fig. 3b). We also generated T cell clones from the individual CTL lines, and as demonstrated by tetramer staining, these were exclusively 100% cross-reactive irrespective of the epitope used to propagate them (Fig. 4A). Interestingly, however, tetramer decay analysis revealed that the B*5703 KF11 tetramer displayed weaker binding kinetics as compared with the B*5703 A2G,S4N tetramer when tested on KF11-propaged clone p5-10 (Fig. 4B). We also assessed the ability of KF11-and [A2G,S4N]-derived clones to secrete IFN-␥, and to up-regulate the degranulation markers CD107a and b in response to KF11 and [A2G,S4N] epitopes. All KF11-and [A2G,S4N]-derived clones behaved identically in that [A2N,S4N] preferentially induced IFN-␥ and CD107ab up-regulation with minimal recognition of KF11 (Fig. 5).

Structural determination of the KF11, [A2G,S4N], and [S4N,V7I] variant pMHC complexes
Our crystal structure of HLA-B*5703-KF11 shows that the unusual length of the 11-mer peptide is accommodated in the peptide-binding groove by the formation of a central peptide bulge. Peptide residues E6 and V7 are solvent exposed, project high above the peptide-binding groove, and are readily accessible for TCR recognition. The side chain of the p4 residue is also solvent exposed and on the basis of our cellular data, we predict that this is sufficient to be involved in TCR recognition. However, there may be a larger scale structural impact of p4 amino acid variation on the main chain bulge conformation. Therefore, to access whether any such conformational differences in the exposure of peptide residues could be responsible for the altered pattern of TCR recognition, we determined the structures of the [A2G,S4N] and [S4N,V7I] epitopes. Both structures were determined from crystals of the orthorhombic space group P2 1 2 1 2 1 which shared almost identical unit cell dimensions, with one HLA-B*5703-peptide complex per crystallographic asymmetric unit (  S4N]-specific T cell clones with similar affinities. T cells were incubated separately with individual tetramers and delay was monitored at 20-min intervals over a total period of 80 min. Tetramer binding was analyzed by flow cytometry, and the natural log of the mean geometric intensity of FL-2 immunofluorescence for each samples (y-axis) were plotted against sampling time (x-axis).
allowed the atomic coordinates for the variant peptide residues (and associated bound water molecules) to be defined accurately permitting a detailed comparison with the KF11 peptide in HLA-B*5703 (12). Superimposition of the HLA-B*5703-KF11, -[A2G,S4N], and -[S4N,V7I] structures (based on their peptidebinding ␣ 1/␣ 2 domain) reveals that the overall bulge conformation found in the main chain of the KF11 peptide is conserved with only minor adjustments in the conformations of some-solvent exposed side chains (Fig. 6). Although the A2G mutation is buried at the B pocket, the S4N and V7I mutations are exposed within the region of the pMHC surface commonly observed to engage with TCRs (30) and thus may alter interactions with specific TCRs.

TCR usage of B*57KF11-restricted CD8 ϩ T lymphocytes
To elucidate the nature of the responding B*57-KF11 T cell repertoire, we evaluated the TCR selection in terms of the pattern of TCR V region usage and CDR3 sequence analysis. We focused on B*5701 ϩ and B*5703 ϩ patients infected with KF11-and [A2G,S4N] viruses. We used a combination of B*57-KF11/B*57-[A2G,S4N] tetramers and mAbs specific for TCR V region segments to determine TCR usage of B*KF11-and [A2G,S4N]-specific T cells, and this resulted in the identification of TCR V region usage for four of the patients studied. We also generated CTL lines, and cell clones specific for KF11/[A2G,S4N] from which the sequence of TCR V␣ and V␤ gene segments were identified by PCR using a panel of predefined TCR V␣ and V␤ primers. Using mAbs specific for TCR V region segments, we observed that three of the five B*5701 ϩ KF11-infected donors used a V␤17 TCR. This receptor was expressed on 57-70% of PBMC-derived B*57-KF11-specific T lymphocytes in two donors (not assessed in OXAG). PCR analysis of T cell clones generated from these donors (LON005, OXAG, and LON201) used a V␤17/V␣15 TCR pair (sequence nomenclature as defined by Arden et al. (31), which is equivalent to a TRBV19/TRAV5 TCR complex as defined by the international ImMunoGeneTics (IMGT) database (32)), and sequence analysis revealed almost complete conservation of the amino acid usage in both the V␣ and V␤ CDR3 regions in all 3 donors, despite variation in nucleotide sequences (Fig. 7). The CDR3 V␣ and V␤ regions were of average length and donors displayed conservation of glycine and tyrosine residues. B*5701 ϩ donor MS used a V␤22 TCR (IMGT-TRBV2) and PCR analysis revealed this to comprise a V␤22/V␣6 (IMGT-TRBV2/TRAV6) TCR pair; the V␤22 CDR3 was short and the amino acids comprising the CDR3b loop were distinct from the conserved V␤17 CDR3 sequence (data not shown). In B*5703 ϩ donor ML525 where [A2G,S4N] represented the dominant viral sequence, a V␤15/V␣15 TCR (IMGT-TRBV24 -1/TRAV5) combination was used, but the amino acid composition of this V␣15 CDR3 loop was highly distinct from the conserved V␣15/V␤17 TCR (data not shown).

Discussion
HLA-B*5701 and -B*5703 are associated with slower HIV-1 disease progression. Both molecules present a similarly broad array of conserved epitopes to CD8 ϩ T cells, a feature that might contribute to their favorable association with prolonged AIDS-free survival.
Recently, escape mutations influencing the recognition of the dominant p24-derived epitopes IW9 and TW10 have been described (9,33), and while both studies help define important correlates of B*57-mediated immunity it is likely that additional factors relate B*57 to prolonged survival. Previously, we focused on the B*57-KF11-specific CD8 ϩ T cell response which prevails during chronic HIV-1 infection. Here, we demonstrate that the HLA-B*57-KF11-specific immune response is consistently dominant during chronic disease in B*57 ϩ slow progressors. Formerly, we also demonstrated broad CD8 ϩ T cell-mediated recognition of diverse KF11 clade variants in B*57 ϩ slow progressors, and our recent crystallographic analysis of B*5703 in complex with the KF11 epitope (10,12) provides an explanation for the inability of certain KF11 clade variants to induce CD8 ϩ T cell responses. However, differential recognition of the [A2G,S4N] and [S4N,V7I] clade variant harboring the same p4 mutation was less    (10). Our preliminary tetramer-decay assay reveals that B*5703 KF11 tetramers form less stable interactions with [A2G,S4N]-specific T cell clones when compared with B*5703 A2G,S4N tetramers. Hence, the interaction of KF11 tetrameric complexes with [A2G,S4N]-specific T cells ex vivo may represent an extended picture of the actual physiological interaction in vivo. Although N and S are both amino acids with polar side chains, they differ in size and hydrogenbonding potential, which might alter the specificity of the p4-mediated interactions with TCRs in vivo.
In donor ML525, the KF11 epitope behaved like an altered peptide ligand and supported [A2G,S4N]-specific T cell division without inducing immune effector functions. In the context of chronic HIV-1 infection, this scenario might adversely influence the immune response: if, as in patient ML525, for example, KF11 represents the emerging virus, it is possible that in addition to priming KF11-specific T cells, this epitope would support the expansion of T cells specific for the index epitope, namely, [A2G  we were unable to establish whether [A2G,S4N]-specific T cells were present at higher frequencies due to sample limitation. We are also unaware whether our patient was coinfected with both viral stains simultaneously, singularly on separate occasions, or developed either variant endogenously. Nonetheless, given our findings and as suggested by recent studies in mice (35), it may be crucial that clade variant epitopes, however similar, are incorporated into vaccines and administered simultaneously to ensure that distinct CD8 ϩ T populations are optimally primed. This approach may limit the biased expansion of T cells with restricted crossreactive effector potential.
We performed a high-resolution structural analysis of B*5703 in complex with the [A2G,S4N] and [S4N,V7I] variant peptides to determine whether there was significant change in the structures of these epitopes compared with the B*5703-KF11 complex. Compared with KF11, the structures of the two variant peptides are broadly conserved by the formation of the bulged hairpin loop in the central portion of the peptide. The variant epitopes p4 and p7 remain solvent exposed within the standard TCR-binding footprint, making these side chains candidates for interaction with cognate TCRs. Given the large number of structural features, and in particular the extensive main chain surface area exposed by this pMHC complex, a given TCR may engage with multiple structural components; thus, the effects of the conservative alterations such as the S4N and V7I may not prevent TCR binding altogether. However, subtle differences in TCR-binding parameters have been shown to affect T cell activation/effector functions and variations in TCR-binding kinetics or thermodynamic parameters may result from minor alterations in peptide structure (36).
We also analyzed the nature of the TCR repertoire recognizing the highly unusual B*57KF11 complex and compared this to the TCR usage of B*57-[A2G,S4N]-specific T cells. Interestingly, three HLA-B*5701 ϩ donors infected with KF11 viruses, used a conserved V␣15 and V␤17 TCR (or TRAV5/TRBV19 TCR (IMGT)), with striking conservation of CDR3 ␣ and ␤ amino acids, and interestingly, these T cells display greatest cross-reactivity ex vivo. Although Dong et al. (37) recently described the usage of the V␤13.2 chain bearing a conserved CDR3 loop in response to a B*0801-restricted nef epitope, this is the first report of a bias in usage of both TCR␣ and ␤ chains in human HIV-1 infection. The best-documented examples of conserved TCR ␣␤ chain usage have come from the study of influenza infection in humans, where the CD8 ϩ T cell responses to A*0201 complexed with the matrix peptide epitope, GILGFVFTL is dominated by a V␤17/V␣10 TCR pair in most individuals (15). The V␤17 CDR3 region is highly conserved and comprises an IRSSY motif (aa 97-101). The crystallographic structure of V␣10/V␤17 in complex with A*0201-GILGFVFTL provides a structural explanation for the conserved TCR V␤ chain usage (38). In this TCR-pMHC complex residues D32 and Q52, unique to the V␤ 17 CDR1 and CDR2 loops, respectively, hydrogen bond the epitope and may define the specific orientation of the TCR onto the pMHC such that the conserved CDR3 ␤ residue R98 inserts into a notch between the epitope and the HLA-A*0201 a2 helix. The R98 and S99 side chains create a dense hydrogen-bonding network between the TCR and pMHC. Similarly, CD8 ϩ T cell responses to the EBV EBNA3 epitope, FLRGRAYGL, are dominated by a common V␤ 6.1/V␣4.1 TCR (LC13) with conserved CDR3 ␣ and ␤ regions in B*0801 ϩ individuals (39) and recent crystallographic analysis of LC13 in complex with B*0801-FLRGRAYGL offers an explanation for the selection of the V and CDR3 ␣ and ␤ sequences comprising LC13 (40). Another striking example of TCR␣, ␤, and CDR3 sequence conservation was recently documented in the study of TCR usage in SIV infection in macaques; here, the authors observed that the majority of macaques infected with SIVmac251 used conserved TCR V␤ chains when responding to the Mamu-A*01-restricted TL8 (Tat) and CM9 (gag) T cell epitopes and that many TL8-and CM9-specific TCR V␤ CDR3 regions were conserved at the amino acid level. Additionally, there was evidence for preferred TCR V␣ chain usage in a small number of macaques responding to the TL8 epitope (41).
Given that the main chain conformation of the peptide epitope in B*5703-K11 is highly unusual it seems plausible that the available TCR repertoire may be limited in comparison to that capable of interacting with more conventional peptide-MHC complexes, hence the conserved pattern of TCR usage in B*5701 ϩ donors. However, B*5701 and B*5703 differ at amino acid positions (p) 114 and 116 which are situated near the F pocket; in B*5701, the p114 N and p116 S residues are replace by a D and Y in the B*5703 subtype, respectively. Based solely on our B*5703 crystallographic complexes, we predict that as p114 nor p116 side chains contribute few contact to peptide binding, KF11 and In the absence of a B*5701-KF11 complex, we are unable to evaluate whether a similarly subtle difference between B*5701 and B*5703 profoundly impact upon the orientation of the KF11 clade variant epitope.
In summary, we have shown that although [A2G,S4N] and KF11 when in complex with B*5703 appear similar from a structural perspective, they behave differently in terms of the functional outcome they induce. In terms of TCR cross-reactivity, KF11 seems to support the outgrowth of CD8 ϩ T cells with a greater capacity for cross-reactivity in B*5701 ϩ donors, whereas in B*5703 ϩ patients the [A2G,S4N]-derived TCR repertoire are limited, particularly in terms amino acid recognition at p4. Interestingly, B*5701 ϩ patients infected with KF11 variants used a conserved TCR that cross-reacted with the majority of p4 variants; it is tempting to speculate that this receptor might select against the emergence of a p4 variant, although it displayed limited crossreactivity to the most common database variant, [A2G,S4N]. Finally, whether our overall findings relate to the subtle p4 differences between KF11 and [A2G,S4N] or to the minor differences in the orientation of these peptide when presented by HLA-B*5701 and B*5703 that, in turn, dictate differential TCR repertoire selection are unknown. The TCR repertoire availability in different ethnic groups might also impact on TCR repertoire selection and TCR cross-reactivity. In-depth studies of larger B*5701 ϩ /B*5703 ϩ cohorts, and if possible, B*5701 ϩ patients infected with [A2G,S4N], are necessary to evaluate these hypotheses.