Analysis of TCRαβ Combinations used by Simian Immunodeficiency Virus-Specific CD8⁺ T Cells in Rhesus Monkeys: Implications for CTL Immunodominance¹

A number of recent studies have demonstrated the importance of virus-specific CD8⁺ CTLs in controlling HIV-1 replication in humans and SIV replication in rhesus monkeys. Persistent, potent CTL responses are associated with low viral loads and delayed progression of clinical disease in infected individuals (1, 2, 3). Furthermore, CD8⁺ T lymphocyte-depleted rhesus monkeys were unable to contain SIV replication during primary or chronic infection (4, 5). Recent studies have also indicated that SIV replication following viral challenge is better contained in monkeys with vaccine-elicited, virus-specific CTL as compared with control animals (6, 7, 8). It is therefore widely believed that HIV-1 vaccine candidates should elicit efficient virus-specific CTL responses in addition to neutralizing Ab responses.

Immunodominance is a common feature of Ag-specific CTL responses generated as a result of infection or immunization (9, 10). The mechanisms that underlie immunodominance are complex: Ag processing, Ag presentation, and T cell stimulation and expansion can have implications for T cell responses. It has been suggested that proteasome specificity (11), transporter-associated Ag-processing selectivity (12), and differences in binding affinity of the epitopes to the MHC class I molecules may contribute to CTL immunodominance (11, 12, 13, 14, 15, 16). A detailed analysis of CTL immunodominance is complicated during chronic virus infection because prolonged in vivo Ag exposure may modulate the magnitude of the responses (15, 17, 18, 19, 20, 21, 22, 23, 24). The dynamics of these responses can be further complicated during persistent infection with pathogens such as HIV or SIV that undergo antigenic variation. A single point mutation can easily eliminate dominant T cell epitopes and, thereby, alter the hierarchy of T cell responses (25, 26, 27). A detailed knowledge of immunodominance is crucial for understanding cellular immunity and viral evasion mechanisms. Furthermore, this information will provide a rationale for improving vaccine design.

Differences in Ag processing and binding affinity of peptides to the MHC class I molecules could be factor(s) for immunodominance. In our previous studies, we showed that the relative peptide binding affinities of both peptides (p11C and p68A) were comparable by the standard iodinated peptide-binding assay (data not shown) as well as by the peptide-binding assay using the TCR tetramer (28, 29). Moreover, the immunodominance hierarchy was maintained when the animals were vaccinated with a DNA vaccine construct containing the minimal epitopes of both peptides separated from one another with triple alanine spacers (28, 29). This construct likely expresses the same copy number of both epitopes and serves to normalize Ag-processing steps. These data demonstrated that Ag processing and the binding affinities of dominant p11C and subdominant p68A epitopes to the Mamu-A*01 molecule may not be the key factors for immunodominance in this model.

In this study, we investigated the mechanisms of SIV-specific CTL immunodominance in a well-established model of Mamu-A*01-expressing rhesus macaques that consistently generate dominant (Gag p11C) and subdominant (Pol p68A) CTL responses following SIV Ag exposure (6, 30, 31). Moreover, vaccinated animals were used to avoid any issues of antigenic variation or dynamic changes in CTL responses caused by chronic Ag exposure.

Blood samples were collected from adult rhesus macaques (Macaca mulatta) vaccinated with multiepitope plasmid DNA that encoded a series of peptide fragments of simian HIV-89.6 previously shown to be presented by MHC class I molecule Mamu-A*01 (28). These monkeys were maintained at the New England Primate Research Center in accordance with the guidelines of the Committee on Animals for the Harvard Medical School and the Guide for the Care and Use of Laboratory Animals (44).

The peptides used in this study were SIVmac251 Gag p11C (CTPYDINQM) and Pol p68A (STPPLVRLV).

PE- or allophycocyanin-labeled tetrameric p11C/ or p68A/Mamu-A*01 complexes (0.1 μg) were used in conjunction with FITC-labeled anti-rhesus monkey CD3 (FN18; Invitrogen Life Technologies) and energy-coupled dye-labeled anti-human CD8αβ (2ST8.5H7) for staining epitope-specific CD8⁺ T cells as described previously (32, 33). Either 100 μl of fresh whole blood or 5 × 10⁵ cells of cultured PBMCs were stained with these reagents. Whole blood samples were lysed using a Coulter Immunoprep reagent system and a Q-prep Workstation (Beckman Coulter) before washing and fixing the cells. Samples were analyzed on a FACSCalibur (BD Biosciences), and data analyses were performed using FlowJo version 6.0 (Tree Star).

Freshly isolated PBMCs (5 × 10⁵ cells/well) were cultured in the presence of various concentrations of peptide (10⁻¹⁰ M to 10⁻⁵ M) in 200 μl of RPMI 1640 medium containing 10% FCS (R10 medium) in a 96-well flat-bottom microplate. Recombinant human IL-2 was added at a final concentration of 20 U/ml at day 2, and half of the culture medium was replenished with fresh R10 medium with 20 U/ml recombinant human IL-2 at day 6, 9, and 12. The cells were collected at day 13 and stained with p11C/ or p68A/Mamu-A*01 tetramer as described above.

Fresh PBMCs (5 × 10⁶ cells/ml) isolated from four multiepitope DNA-vaccinated rhesus macaques boosted with recombinant modified vaccinia Ankara-expressing SIV Gag and Pol (rMVA-gag/pol) were labeled with 5 μM CFSE at 37°C for 30 min and then washed with ice-cold R10 medium three times. CFSE-labeled PBMCs were cultured at 5 × 10⁵ cells/well with or without p11C (10⁻⁷ M) or p68A (10⁻⁹ M) in 200 μl of R10 medium in a 96-well flat-bottom microplate. The cells were collected at day 2, 4, 6, and 8, and stained with p11C/ or p68A/Mamu-A*01 tetramer and anti-human CD8αβ-energy-coupled dye. CFSE-labeled PBMCs stimulated with 2 μg/ml Con A was used as a positive control. Cell proliferation was assessed by analysis of CFSE dilution of CD8⁺ peptide/Mamu-A*01 tetramer⁺ cells.

PBMCs (2 × 10⁶ cells) were stained for 30 min at room temperature with the corresponding PE-labeled Mamu-A*01 tetramers (5 nM). The cells were then washed three times to eliminate unbound tetramers, resuspended in staining buffer, and incubated at 15°C in the presence of excess (50 nM) unlabeled tetramers to avoid rebinding of the PE-labeled tetramers. An aliquot (2 × 10⁵ cells) was taken and analyzed for the indicated time points.

Samples were analyzed on a FACSCalibur, and data analyses were performed using FlowJo version 6.0.

Gag p11C/and Pol p68A/Mamu-A*01 tetramer-binding CD3⁺CD8⁺ T cells from two vaccinated rhesus monkeys were sorted by a FACSDiVa cell sorter (Invitrogen Life Technologies). Total RNA was extracted with the RNAeasy kit (Qiagen). Complementary DNA was synthesized using the SMART RACE cDNA amplification kit (Clontech Laboratories). The DNA fragments of the extracellular domain of TCR-α- and -β-chains were amplified by PCR using a 5′ universal primer provided by the kit and the 3′ primer specific for conserved region of TCR-α (TCRAR2, 5′-CCC GGC CAC TTT CAG GAG GAG G-3′) and TCR-β (MmB1-R, 5′-TGA TGG CTC AAA CAC AGC GAC CTT GGG TGG-3′), respectively. The amplified PCR products were inserted into pGEM-T Easy vector (Promega) and then sequenced using a T7 promoter primer or a SP6 promoter primer. The usage of TCR-α- and -β-chain from epitope-specific CD8⁺ T cells was classified according to ImMunoGeneTics (IMGT) nomenclature (34).

Selected TCR-α and TCR-β clones were used to screen Ag-specific TCR-αβ pairs using the Drosophila expression system as described before (29). In brief, each 5′ primer specific for sequences of selected clones was designed with addition of a KpnI restriction site, and 3′ primers specific for TCR-α (3′ BamHTCRA, 5′-CCC CAG CCC AGA AAG TGT CTG TGG ATC CGC G-3′) and β-chain (3′BamHTCRB, 5′-GAG GCC TGG GGT AGA GCA GAC TGT GGA TCC GCG-3′) were designed with a BamHI restriction site at the 3′ end to amplify the coding sequence upstream of the transmembrane region by PCR. The amplified products were digested with KpnI and BamHI, gel-purified, and then subcloned into the expression plasmid pMT/MmDRαTR or βTR, which carries the transmembrane sequence of the rhesus macaque MHC class II DR molecule α and β-chain, respectively, at the C terminus. Drosophila melanogaster Schneider 2 (S2) cells were cotransfected with all possible combinations of TCR-α- and -β-chains using the calcium phosphate transfection kit (Invitrogen Life Technologies). Five days after transfection, protein was induced by the addition of 1 mM copper sulfate for 24 h and then the cells were stained with p11C/ or p68A/Mamu-A*01 tetramer.

A mathematical approach was taken to understand the expression of these TCR-αβ complexes and the apparent dominance and subdominance of certain responses. Similarities in sequences for suspected dominant and subdominant epitopes were compared, and probabilities of the needed coupling of such sequences and chains were determined through the manipulation of the equations below.

Equation 1 defines the probability of coupling the needed α- and β-chain sequences to be equal to the sum of the probability of generating the needed α-chain sequence and probability of generating the needed β-chain sequence.

Equation 2 defines the probabilities of generating the needed α sequences to be equal to the ratio of Vα-chain usage, n(α, V), to the total possible Vα sequences, N(α, V), multiplied by the ratio of Jα-region usage, n(α, J), to the total possible Jα-regions, N(α, J).

Equation 3 defines the probabilities of generating the needed β sequences to be equal to the ratio of Vβ-chain usage, n(β, V), to the total possible Vβ sequences, N(β, V), multiplied by the ratio of D-region usage, n(β, D), to the total possible D-regions, N(β, D), multiplied by the ratio of Jβ-region usage, n(β, J), to the total possible Jβ-regions, N(β, J).

It has been reported that immunodominant CTL populations require significantly lower peptide concentrations to induce maximal in vitro Ag-specific T cell expansion than subdominant CTL populations (35). Therefore, we examined the in vitro expansion of both the dominant p11C- and the subdominant p68A-specific CTL population to different peptide concentrations. PBMCs from four vaccinated animals were stimulated with decreasing concentrations of the minimum peptide epitopes. The magnitude of the Ag-specific T cell expansion was monitored at 13 days poststimulation with the corresponding peptide/Mamu-A*01 tetramers. Tetramer⁺ T cells were only expanded in the presence of the appropriate peptides in all animals tested (data not shown). It is also important to note that the maximum percentage of tetramer⁺ T cells following peptide stimulation directly reflects the percentage of tetramer⁺ T cells in fresh PBMCs isolated from all four animals. Specifically, the higher percentage of in vitro expansion of p11C-specific T cells compared with p68A-specific T cells directly reflects the higher percentage of p11C-specific T cells compared with p68A-specific T cells in freshly isolated PBMCs (Fig. 1). Because the standardized minimum detection limit of tetramer staining by flow cytometry in our laboratory is 0.02%, p68A-tetramer staining of freshly isolated PBMCs from animals nos. 90-98 and 128-97 is considered to be undetectable (<0.02%). Therefore, the expansion rate of these cells was comparable to that from other animals (nos. 95-98 and 135-97), although the maximum in vitro expansion of these cells was the lowest (4.55 and 2.90%). The subdominant p68A-specific CTL population maximally expanded even at lower peptide concentrations compared with the dominant p11C-specific CTL expansion (Fig. 1). Interestingly, we noted an inhibitory effect on the subdominant p68A-specific T cell expansion at higher peptide concentrations, which has been previously described as a characteristic of dominant epitope-specific CTL responses (35).

The proliferative capacity of p11C- and p68A-specific T cells could also be a factor involved in immunodominance. To examine this theory, we combined CFSE labeling with tetramer staining after in vitro stimulation with the optimum Ag concentration. Each time a cell divides, CFSE (a fluorescent dye that stably binds to cytoskeletal actin) is apportioned equally among daughter cells, resulting in a 50% reduction in CFSE fluorescence. Therefore, the number of cell divisions can be determined by comparing the CFSE fluorescence intensity to that of nondividing cells. By measuring tetramer⁺ CD8⁺ T lymphocytes and CFSE intensity at different time points poststimulation by flow cytometry, the relative differences in proliferation between p11C- and p68A-specific T cells could be compared. Flow cytometry data from animal 135-97 are shown as representative of all four animals in Fig. 2. CFSE mean fluorescence intensity (MFI)⁵ changes of both tetramer⁺ cells from all animals tested are shown in Table I. Proliferation of both Ag-specific T cells became evident at day 4, as indicated by the decreased intensity of CFSE labeling of the tetramer⁺ T cells. Both dominant p11C- and subdominant p68A-specific T cell populations were shown to have comparable proliferative capacity at all time points analyzed after in vitro Ag stimulation.

To demonstrate the similarity of the TCR-peptide/MHC interaction of the p11C- and p68A-specific CTL populations, the off rate of their interaction was measured. First, we determined the minimum concentration of tetramers required to achieve the maximum staining intensity for both CTL populations. We selected vaccinated Mamu-A*01⁺ animals that showed detectable tetramer staining for the subdominant p68A- as well as the dominant p11C-specific CTL populations. Whole blood from these animals was incubated with increasing concentrations of the corresponding tetramers and analyzed by flow cytometry. Both tetramers required an equivalent amount (5 nM) of monomer to reach their maximum staining intensity (data not shown). Briefly, 5 nM of the corresponding tetramers that showed maximum staining for both CTL populations was used to stain. The cells were then washed and incubated at 15°C in the presence of an excess of unlabeled tetramers to avoid rebinding of PE-labeled tetramers to the TCR. The dissociation rate of the TCR-peptide/MHC interaction was compared by measuring the MFI changes of the corresponding tetramers over time. Fig. 3 shows that the MFI of both tetramer⁺ cells decreased similarly over time, indicating that the off rates of the TCR-peptide/MHC interaction of the dominant (p11C) and the subdominant (p68A) may also be comparable. It is well known that tetramer⁺ cells are oligoclonal, if not polyclonal, in SIV-infected animals (36, 37). If some clones have different binding affinities, we might expect to see a different tetramer staining intensity on a subpopulation of tetramer⁺ cells when increasing concentrations of tetramers were used. However, a very tight cluster of tetramer⁺ cells was seen, and these cells increased their intensity homogeneously with a higher concentration of tetramers (data not shown). These data suggest that not only do the TCRs for the dominant and the subdominant CTL epitopes have comparable binding affinities to their peptide/MHC complex, but the oligoclonal TCRs have comparable binding affinity for the same peptide/MHC complex as well.

We next examined whether immunodominance was the result of a higher number of TCR clones capable of recognizing a single epitope. Fresh PBMCs were isolated from the multiepitope DNA-vaccinated animals described previously after boosting with rMVA-gag/pol. Ex vivo cell sorting was performed after labeling with p11C/ and p68A/Mamu-A*01 tetramers, and the TCR-α and -β sequences from both CTLs were analyzed. All sorted cells were confirmed to be >98% pure and normalized to 50,000 cells before RNA extraction. An anchored RT-PCR was used to amplify all expressed TCR-α and TCR-β gene products without bias. Two hundred and 160 clones were analyzed for both α- and β-chains, respectively, for each isolated epitope-specific T cell population from animal nos. 95-98 and 135-97. For the immunodominant p11C-specific CTL population, 17 and 23 different TCR-β clones were obtained from animal nos. 95-98 and 135-97, respectively. In contrast, only 9 and 6 clones, respectively, were isolated from the same two animals for the subdominant p68-specific CTL. These data indicate that there is more TCR-β usage for the dominant CTL population than for the subdominant population. As expected, a greater number of TCR-α clones were isolated for the dominant p11C CTL as compared with the subdominant p68A CTL (Figs. 4 and 5).

Notably, 81% of the TCR-Vα used by the dominant p11C-specific CTLs from both animals were in the TCR-Vα8 subgroup. Three different TCR-Vα8-related sequences were found from lymphocytes obtained from both animals (Fig. 6,A), and 57% of those TCR-α clones used the identical TCR-Jα5*01 sequences. Interestingly, four of the TCR-α clones found in one animal contained identical amino acid sequences to those in the other animal. Three additional clones had the same CDR3 sequence but on different genes (Fig. 4). In previous reports, SIV-infected Mamu-A*01⁺ animals preferentially use the TCR-Vβ6 (TCR-Vβ13 when Arden’s nomenclature is used) subfamily (36, 37, 38). This was also observed on p11C-specific CD8⁺ T cell populations in our two vaccinated animals (Fig. 5). As shown in Fig. 6 B, at least five different TCR-Vβ6-related sequences were identified from these two vaccinated rhesus monkeys.

Contrary to a preferential usage of a certain TCR-Vα subfamily for the dominant CTL population, >55% of all TCR-Vα clones from the subdominant p68A-specific CTL in both animals used exactly the same Jα40*01 sequence. However, these originated from six completely different TCR-Vα subfamilies (TCR-Vα3, -Vα5, -Vα8.4A, -Vα8.6, -Vα9.2, and -Vα13.2) (Fig. 4). Similar to the pattern observed for the dominant p11C-specific CTL population, >55% of the TCR-Vβ usage by the subdominant p68A-specific T cells used a single TCR-Vβ13 gene sequence (Fig. 5). The preferential usage of the TCR-Vβ13 was consistently observed in p68A-specific CTL isolated from at least six chronically SIV-infected monkeys and five vaccinated animals (data not shown).

We have previously developed an assay to identify functional TCR-αβ pairs by reconstituting various single TCR chains (α and β) on Drosophila cells (29). The screening of the specific TCRs was performed by staining with their corresponding peptide/Mamu-A*01 tetramers.

Due to the clear preferential usage of TCR-Vα8 and TCR-Vβ6 by p11C-specific CTL populations, it was expected that these two TCR variable chains would pair to recognize the p11C/Mamu-A*01 complex. The amino acid sequences of four independent p11C-specific TCR-α clones isolated from two different animals were identical. However, identical sequences were not seen in the corresponding TCR-β-chains. Therefore, we speculated that those separate TCR-α-chains could combine with at least two different TCR-β-chains to recognize the p11C/Mamu-A*01 complex. One of these four identical TCR-α clones from these two animals was also discovered in the other independent SIV-infected macaque (data not shown). This TCR-α clone was used for screening of the β-chain that would be its partner to recognize the p11C/Mamu-A*01 complex. Surprisingly, this single TCR-Vα8 clone could use at least 15 different TCR-Vβ6 clones with different CDR3 sequences to recognize the p11C/Mamu-A*01 complex. Moreover, all five different TCR-Vβ6 sequences described in Fig. 6,B could be paired with the TCR-Vα8.4A no. 11 clone to recognize the same peptide/Mamu-A*01 complex (Table II, top). To expand our study, 12 more TCR-Vα8 clones containing the same TCR-Jα5*01 sequence but with one or two amino acid differences in the N region were also tested for the p11C/Mamu-A*01 specificity by pairing with different TCR-Vβ clones. As shown for the TCR-Vα8.4A no. 11 clone, all 12 clones recognize the p11C/Mamu-A*01 complex by pairing with at least five different TCR-Vβ6 clones (data not shown). These interactions appeared specific because two other TCR-Vβ (TCR-Vβ10 and TCR-Vβ21) did not stain positively when paired with this TCR-Vα clone.

Reconstitution of the functional TCR-αβ pairs specific for the subdominant p68A/Mamu-A*01 complex was also performed. After analyzing the pattern of both TCR-α and TCR-β usage of the p68A-specific CTL population isolated from the two vaccinated animals, two major features could be noted: there were preferential usages of TCR-Vβ13 and of TCR-α containing the same Jα40*01 sequence (Figs. 4 and 5). Three representative TCR-Vβ13 clones containing different CDR3 sequences were selected to find TCR-α capable of recognizing the p68A/Mamu-A*01 complex. Nine different randomly selected TCR-α clones from the p68A-specific CTL in both animals were assessed. Interestingly, six TCR-α clones could be paired with all three TCR-Vβ13 clones to specifically recognize the p68A/Mamu-A*01 complex. All six reactive TCR-α clones had the same Jα40*01 sequence with three different TCR-Vα sequences (Table II, bottom).

TCR sequence data of both the dominant and the subdominant CTL populations in Mamu-A*01⁺ rhesus monkeys suggested that the CTL specific for the dominant epitope use a wider array of TCR clones than CTL for the subdominant epitope. To understand the hierarchy of this response, theories of probability were implemented to compare the likelihood of the generation of TCR-αβ pairs specific for the dominant p11C and the subdominant p68A epitopes. Data from TCR-α and -β sequencing and in vitro functional pairing experiments suggest that TCR usage is not random, and, in fact, every CTL seems to use a rather specific pattern of TCR-αβ pairs for different epitopes. As such, CTLs specific for the dominant p11C and the subdominant p68A may use the same TCR-Vβ6 sequence but use a unique TCR-Vα sequence as a pairing partner. Moreover, due to the fact that >54% of the CTLs specific for p11C (dominant) and p68A (subdominant) epitopes use a unique pattern of TCR-αβ combinations, we believed that the probability of the generation of these specific pairs would likely reflect the trend in the global response.

Due to the limited number of rhesus TCR sequences available, the number of the total variables of the TCR sequences was obtained from a human TCR database (IMGT, http://imgt.cines.fr). The number of Vα and Vβ sequence patterns that could be combined to generate specific αβ pairs for both epitopes is shown in Table III, and the probabilities of coupling of these chains were determined through the manipulation of the equations as described in Materials and Methods. The results show that the probability of generating TCR-αβ pairs specific for the p11C epitope is four times higher than for the p68A epitope (Table III). This analysis provides support for the notion that the dominance of the p11C epitope is due to the greater number of potential TCRαβ pairs than for the p68A epitope. Further examination of this analysis suggests that the J-region usage also contributes to the dominance of the p11C epitope.

Understanding CTL immunodominance is key to rational development of vaccines capable of generating effective Ag-specific CD8⁺ T cell responses. Virtually every step in the generation of this cellular immune response including Ag processing, T cell priming, and T cell expansion could impact immunodominance. In the present report, the major steps contributing to CTL immunodominance were studied in an outbred nonhuman primate model. In this study, we used a well-established system that consistently generates immunodominance hierarchies after SIV Ag exposure in Mamu-A*01⁺ rhesus monkeys and a nonreplicating vaccine as the source of Ag to avoid chronic Ag exposure and antigenic variation. Normalization of the Ag copy number as well as the Ag-processing steps of both Ags was attempted by delivering a DNA construct encoding the minimum epitopes of the dominant p11C and the subdominant p68A separated with the same alanine spacer (28). The reproducible immunodominance hierarchy observed in Mamu-A*01⁺ animals vaccinated with this construct, in conjunction with comparable peptide-binding affinities of both peptides to the Mamu-A*01 molecule, strongly suggest that efficiency of Ag processing is not the basis for immunodominance in this model.

Differences of proliferative capacity and sensitivity of Ag-specific T cells to the peptide/MHC complex could also have a strong impact on immunodominance. However, our data have shown that the subdominant p68A-specific CTL can be expanded as efficiently as the dominant p11C-specific CTL in response to the low peptide concentrations. Interestingly, the requirement of lower peptide concentration to achieve maximum CTL expansion observed for the subdominant p68A-specific CTL, was suggested to be a feature of the dominant CTL population (35).

The proliferative capacity of both CTL populations was also comparable when measured by the intensity of the CFSE staining of the tetramer⁺ cells after in vitro stimulation with the optimal peptide concentration. In addition, some TCR pairs specific for the same peptide/MHC complex have been shown to have higher binding affinities than others and, therefore, are suggested to be the mechanism of more efficient expansion (39). In this study, the TCR-peptide/MHC interaction of the dominant and the subdominant oligoclonal CTL populations was compared by measuring the dissociation rate of their interaction. However, no difference in TCR binding affinity of the oligoclonal CTL populations between p11C and p68A was seen.

CTLs specific for two additional peptides restricted to the Mamu-A*01 molecule were shown to require the same concentration of monomer to achieve the maximum fluorescence intensity (data not shown). Our data demonstrate that TCR binding affinities of CTL populations specific for four independent epitopes are similar, suggesting that this may be a common feature of Ag-specific CTL. There is a discrepancy between our data and previous studies that showed differences in the binding affinity of TCR clones recognizing the same peptide/MHC complex. This could be a result of differences in the technical approach. Our assay measured the binding affinity of TCR naturally expressed on CD8⁺ T cells with their complete coreceptor, including the CD8 molecule, whereas previous studies compared the binding affinity between purified TCR and peptide/MHC complexes (40). Thus, it is possible that coreceptor(s) may contribute to the binding of some TCR clones to the same peptide/MHC complex. These data suggest that, in this experimental system, the peptide binding affinity, efficiency of Ag presentation or processing, and T cell functional differences are not likely to be the mechanism responsible for immunodominance.

Analysis of TCR repertoire revealed the usage of higher numbers of TCR clones by the dominant p11C-specific CTL than the subdominant p68A-specific CTL population. It was surprising to find the preferential usage of the TCR-Vα8 subfamily and the TCR-Vβ6 subfamily by the dominant p11C-specific CTL population obtained from two outbred animals. Four of the TCR-α clones found in one animal had identical amino acid sequences as those from another animal. This phenomenon could be attributed to the analysis of Ag-specific CTL populations from vaccinated animals with short-term Ag exposure without antigenic variation, in contrast to the previous studies performed in infected animals that are usually under continuous Ag exposure and complex antigenic variation due to virus mutation under immunological pressure (36, 37, 38). The preferential usage of the same TCR-Jα5*01 sequence was as noteworthy as the usage of TCR Vα8 in the p11C-specific CTL population obtained from two independent animals. The preferential usage of the same TCR-Jα40*01 sequence was also evident for the subdominant p68A-specific CTL population. At least five different TCR-Vα sequences were found to contain the same TCR-Jα40*01 sequence, in contrast to the usage of the unique TCR-Jα5*01 sequence by a single TCR-Vα8 subfamily for the p11C-specific CTL.

Our novel in vitro functional TCR-α- and -β-chain pairing assay showed that TCR-Vα8 containing a unique TCR-Jα5*01 sequence could recognize the p11C/Mamu-A*01 complex by pairing with many clones of the TCR-Vβ6 subfamily but with none of the other TCR-β clones tested. Interestingly, although the length of all reactive TCR-β CDR3 regions appear to be important, the amino acid sequence was not crucial for their specific Ag recognition because at least 10 different TCR-Jβ* sequences using both D*01 or D*02 sequences were detected among the reactive clones. The importance of the TCR-α CDR3 region was also demonstrated in the in vitro functional TCR pairing experiment specific for the subdominant p68A epitope. In this case, TCR-α clones containing a specific TCR-Jα40*01 sequence could specifically recognize the p68A/Mamu-A*01 complex by pairing with TCR-Vβ13 clones that have different TCR-Jβ* and D* sequences. These data strongly suggest that the CDR3 region of the TCR-α-chain may be more important for the specific peptide recognition than the CDR3 region of the TCR-β-chain. A similar conclusion was reported in the mouse model studying the pairing pattern of a TCR specific for a dominant HIV envelope epitope (41).

Preferential usage of specific TCR has been previously described for many CTL specific for different Ags in various experimental models (36, 37, 38, 42, 43). Our study has shown a clear preference of certain TCR sequences for p11C and p68A epitopes. These data suggest that TCR usage is not random. In fact, we hypothesized that every peptide/MHC complex can only be recognized by a limited number of unique combinations of TCR-αβ pairs and that the CDR3 will ultimately determine their specificity. In other words, each peptide/MHC complex, with peptides containing few amino acid changes in the context of the same MHC molecule, may already be sufficiently structurally unique to be recognized by only selected TCR-αβ pairs. Because >54% of the CTLs specific for p11C (dominant) and p68A (subdominant) use a unique pattern of TCR-αβ combinations in two independent outbred animals, we believe that the probability of generating these specific pairs may reflect a global trend and explain why the dominant p11C-specific CTL population uses a wider array of TCR clones. The probability of generating TCR-αβ pairs specific for the p11C epitope was almost four times higher than for the p68A epitope. This process supports the hypothesis that immunodominance of the p11C epitope may be due to the higher number of naive CD8 T cells that could potentially recognize the p11C/Mamu-A*01 than the p68A/Mamu-A*01 complex.

The fact that the p11C-specific CTL response is consistently dominant in Mamu-A*01⁺ rhesus monkeys could be explained simply by the availability of a TCR-Vα8/TCR-Vβ6 combination to recognize the p11C/Mamu-A*01 complex. Rhesus monkeys express at least two independent TCR-Vα8 and five TCR-Vβ6 subfamilies containing different amino acid sequences that allow them to indistinguishably recognize the p11C/Mamu-A*01 complex. Thus, the likelihood of finding the correct TCR-Vα8/TCR-Vβ6 pairs specific for p11C/Mamu-A*01 complex among other TCR-αβ combinations, assuming that all TCR-α and -β are distributed equally, should be considerably higher. The preferential usage of TCR-Vβ13 by the subdominant p68A-specific CTL population may be explained by their unique TCR-αβ pairing pattern. At least five different TCR-Vα sequences containing the unique TCR-Jα40*01 region could recognize the p68A/Mamu-A*01 complex indistinguishably when combined with the TCR-Vβ13 with variable Jβ* and D* sequences. Thus, the likelihood of finding the p68A-specific TCR-Vβ13 increases at least 5-fold when compared with the rest of TCR-Vβ clones.

These data strongly suggest that one of the factors for CTL immunodominance may be a reflection of the size of the naive Ag-specific T cell population. Those epitopes that can be recognized by a larger population of Ag-specific T cells present before Ag exposure will dominate over other epitopes.

However, our data do not exclude any of the previously described factors that may modulate CTL stimulation or expansion. As described in this study, if CTL immunodominance is dictated by the naturally preexisting Ag-specific naive T cell population, development of HIV vaccines that can generate stronger immune responses toward the subdominant epitopes may require novel approaches.

We are grateful to P. Autissier, D. Gorgone, and M. Lifton for the flow cytometry expertise and H. Chhay and K. Martin for technical assistance. We thank Drs. Keith A. Reimann and Andrew A. Lackner for advice and help with data presentation.

The authors have no financial conflict of interest.