INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Cytotoxic T lymphocytes (CTL) play an important role in the eradication of intracellular pathogens. CTL become activated when their T-cell receptors (TCR) specifically recognize foreign antigen in the form of a molecular complex between a major histocompatibility complex (MHC) class I molecule and a short antigenic peptide (31, 33). TCR molecules are heterodimers whose  and  chains are composed of a constant (C) and a variable (V) extracellular domain. Somatic recombination of a number of V, diversity (D), and junctional (J) gene segments, imprecise joining, and addition of N nucleotides are responsible for the TCR- and --chain diversity which is further increased by combinatorial - pairing (38). The specificity of the TCR is determined mainly by three hypervariable complementarity-determining regions (CDR) of the  and  chains. CDR1 and CDR2 are encoded by V segments, whereas CDR3 is encoded by J elements for the  chain and by D and J elements for the  chain. This exceptional potential diversity (estimated at more than 10¹⁵ for TCR) allows CD8 T cells to respond to a wide variety of antigens (14).

Recent studies have focused on the diversity of TCR in recognizing one given antigen (9). For example, TCR expressed by CTL clones specific for a single Plasmodium berghei circumsporozoite peptide were highly diverse in terms of V, J, and J segments and amino acid compositions of the junctional regions. However the V segment utilization was strongly conserved among the different clones (10). In contrast, analysis of the TCR repertoire of CTL directed against a peptide derived from the human class I MHC molecule HLA-CW3 presented by murine MHC class I (H-2K^d) molecules at the surface of P815 tumor cells revealed a very limited heterogeneity both in terms of V, J, V, and J segments and in terms of lengths and sequences of both CDR3 and - (8). This highly restricted TCR usage allowed the identification of antigen-specific T cells in individual immune DBA/2 mice either by staining with monoclonal antibodies (MAbs) to the V domain (25) or by single-cell PCR (26). In addition, V preferences have been demonstrated in CD8⁺ T-cell responses to acute infection by several viruses, including human immunodeficiency virus (30), simian immunodeficiency virus (12), Epstein-Barr virus (7) and lymphocytic choriomeningitis virus (24).

In a previous publication (5), we showed that the CD8⁺ T cells responsible for the rejection of Moloney murine leukemia virus (M-MuLV)-induced tumor cells had a very restricted usage of both V and V gene segments. Indeed, immunization of C57BL/6 (B6) mice with M-MuLV-induced tumor cells (MBL-2) led to an overwhelming expansion of CD8⁺ T cells that recognized exclusively a virally encoded immunodominant epitope. This epitope (CCLCLTVFL), presented by H-2D^b, is shared by leukemia and lymphoma cell lines infected by the Friend-Moloney-Rauscher (FMR) group of leukemia viruses and is encoded in the leader sequence of the gag polypeptide (11, 22). These M-MuLV gag-specific CD8⁺ T cells could be readily monitored ex vivo by flow cytometry since the majority coexpressed the V3.2 and V5.2 gene segments.

In the present study, we were interested in analyzing the CD8⁺-T-cell response to M-MuLV-induced tumor cells in mice unable to express the dominant V3.2⁺ V5.2⁺ TCR. For this purpose we took advantage of congenic B6.V^a mice (28) that have a large deletion at the TCR locus, including the V5.2 gene segment (2). Interestingly, despite the absence of V5.2⁺ cells, B6.V^a mice were able to reject M-MuLV-induced tumor cells. Moreover, analysis of the TCR repertoire in these immune B6.V^a mice indicated a dramatic expansion of CD8⁺ T cells coexpressing V3.2 together with either the V3 or the V17 gene segment and recognizing the same immunodominant gag epitope. Interestingly, these two V gene segments differ between the V^a and V^b haplotypes. Indeed, the V17 gene is not expressed in the V^b haplotype due to the presence of a stop codon whereas the V3 gene segment differs between the two haplotypes by a point mutation, resulting in a single amino acid substitution at position 31 (Phe in V^a versus Val in V^b) (32). In (B6 × B6.V^a)F₁ mice immunized with M-MuLV-induced tumor cells, we observed a clear hierarchy in V usage by CD8⁺ T cells (V17 > V3 > V5.2). The structural basis for the hierarchal recognition of a single peptide-MHC complex by TCR utilizing three distinct V domains was also investigated by sequencing the CDR3 regions of both the  and  chains.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Mice. B6 mice were obtained from HARLAN OLAC (Bicester, United Kingdom). Congenic B6.V^a mice were kindly provided by A. Livingstone (Basel Institute for Immunology, Basel, Switzerland). These mice were derived by transferring the V^a haplotype (which has an extensive deletion at the TCR locus, including the V5, -8, -9, -11, -12, and -13 gene segments [2]) from C57L mice (H-2^b, V^a) to B6 mice (H-2^b, V^b). However, two other V gene segments (V17 and V19) that are not expressed in V^b mice are expressed in the V^a haplotype. The B6.V^a mice used were backcrossed for 15 generations to B6 mice. (B6 × B6.V^a)F₁ mice were bred in our animal facilities.

Immunizations. M-MuLV-infected MBL-2 (H-2^b) tumor cells were maintained by weekly passage in syngeneic B6 mice (6). For primary immunization, 40 × 10⁶ irradiated (10,000 rads) tumor cells were injected intraperitoneally into syngeneic mice. After 3 to 4 weeks, secondary responses were elicited by intraperitoneal injection of 10 × 10⁶ viable syngeneic tumor cells.

MLTC and CTL clones. Virus-specific CTL were generated in vitro in a 5-day mixed lymphocyte-tumor cell culture (MLTC) (3). Responder spleen cells (25 × 10⁶) from M-MuLV immune mice and irradiated MBL-2 cells (1 × 10⁶) were cocultured in 15 ml of Dulbecco modified Eagle medium (Gibco, Paisley, United Kingdom) supplemented with 2 × 10³ M L-glutamine, 2 × 10² M HEPES, 3 × 10⁵ M 2-mercaptoethanol, antibiotics, and 5% heat-inactivated fetal calf serum (Irvine Scientific, Santa Ana, Calif.). Cells recovered from MLTC were washed and restimulated with irradiated MBL-2 and syngeneic feeder cells for a further seven days in complete medium supplemented with 30 U of interleukin 2 (EL-4 cell supernatant) per ml. CTL clones were established by plating cells from an MLTC at limiting dilution as described previously (5).

Cytotoxic assays. CTL clones derived from M-MuLV-immune mice were used as effector cells. Target cells were either MBL-2 lymphoma (M-MuLV infected, H-2^b), RMA lymphoma (Rauscher virus infected, H-2^b), or EL-4 lymphoma (FMR uninfected, H-2^b) cells. The FMR gag-encoded epitope CCLCLTVFL (11) was synthesized and purified by standard procedures and dissolved in dimethyl sulfoxide supplemented with -mercaptoethanol. For cytotoxic assays, effector cells and ⁵¹Cr-labeled target cells were mixed at the ratios indicated below in the presence or absence of various concentrations of peptide. Supernatants were harvested after 4 h, and specific ⁵¹Cr release was calculated as described previously (5).

Flow microfluorometry. At various times after primary or secondary immunization with syngeneic M-MuLV-infected tumor cells, mice were bled by the tail vein and peripheral blood lymphocytes (PBL) were isolated by Ficoll-Hypaque gradient centrifugation (Pharmacia Biotech, Uppsala, Sweden).

RNA extraction, cDNA synthesis, and PCR. Total RNA was extracted from 5 × 10⁶ cells from MLTCs or CTL clones with QIAshredder columns as the cell lyzate homogenizer and an RNeasy Mini Kit as the RNA extraction system (both from Qiagen AG, Basel, Switzerland). Single-stranded cDNA synthesis was carried out on total RNA with oligo(dT)15 and avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim Biochemicals, Indianapolis, Ind.). PCR was carried out in 50 µl on 2/50 of the cDNA with 5 U of Taq polymerase (Eurobio) according to the manufacturer's instructions. Oligonucleotides for the PCR amplification were the following: (V3) 5'-AAGTACTATTCCGGAGACCC-3', (Ca) 5'-TGGCGTTGGTCTCTTTGAAG-3', (Cb) 5'-ACACAGCAGGTTCTGGGTTC-3', (V3) 5'-CCTTGCAGCCTAGAAATTCAGTCC-3', (V5.2) 5'-AAGGTGGAGAGAGACAAAGGATTC-3', (V17) 5'-GAACAAACAGACTTGGTCAAG-3', (Ca) 5'-CCAGAAGGTAGCAGAGACCC-3', and (Cb) 5'-CTTGGGTGGAGTCACATTTCTC-3' (10). Forty cycles, each of 94°C for 15 s, 58°C for 45 s, and 72°C for 60 s, were completed in a thermocycler. PCR products were purified with a QIAquick PCR Purification Kit (Qiagen AG).

Sequencing reactions. Sequencing reactions of purified PCR products were done by fluorescent cycle sequencing with a Thermo Sequenase fluorescence-labeled primer cycle sequencing kit (Amersham Life Science Ltd.) according to the manufacturer's instructions and analyzed in a LI-COR DNA sequencer (MWG-biotech, Munchenstein, Switzerland).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

TCR V and V repertoires of M-MuLV-immune B6.V^a mice. B6.V^a mice were injected with irradiated syngeneic M-MuLV-infected (MBL-2) tumor cells and boosted 2 weeks later with viable cells. Immune mice were able to reject the tumor. These mice were then bled at day 7 after the second injection, and PBL were pooled and stained with a panel of anti-V or anti-V MAbs together with anti-CD4 or anti-CD8 MAb. MAbs against CD62L (Mel-14) were included in the fourth color to increase the sensitivity of detection of responding CD8⁺ or CD4⁺ cells and to be able to differentiate between activated (CD62L) and nonactivated (CD62L⁺) T cells (36). The TCR V and V repertoires of M-MuLV-immune B6.V^a mice are shown in Fig. 1. PBL from M-MuLV-immune B6.V^a mice were highly enriched for V3.2⁺, V3⁺, and V17⁺ cells in the activated (CD62L) subset of CD8⁺ cells following secondary immunization with MBL-2 cells. The other V and V domains tested showed lower levels in the CD62L compartment than in the CD62L⁺ subset. As expected, no preferential TCR V or V usage was observed among the nonactivated (CD62L⁺) subset of CD8⁺ cells or among activated (CD62L) CD4⁺ PBL.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   TCR V and V repertoires of M-MuLV-immune PBL in B6.Va mice. PBL from a pool of 15 B6.Va mice immunized twice with syngeneic M-MuLV-infected MBL-2 tumor cells were stained in four colors with MAbs to CD8, CD4, and CD62L and a panel of anti-V or anti-V MAbs. Open and filled bars represent percentages of CD8+ or CD4+ cells expressing the indicated V or V domain in the CD62L+ or CD62L subsets, respectively.

M-MuLV-immune CD8⁺ cells in B6.V^a mice preferentially express V3.2 in association with either V3 or V17. PBL from 15 M-MuLV immune B6.V^a mice were pooled, and the expression of V3.2 versus V3 or V17 was analyzed either among activated (CD62L) or nonactivated (CD62L⁺) CD8⁺ cells (Fig. 2). CD62L⁺ CD8⁺ PBL from immune and naive mice showed the same small percentage of V3.2⁺, V3⁺, or V17⁺ cells (data not shown). In contrast, a dramatic expansion of activated CD8⁺ cells expressing V3.2 in exclusive association with either V3 or V17 was observed in CD62L CD8⁺ PBL from M-MuLV-immune mice. The majority of the cells (55%) in this subset were V3.2⁺ V17⁺, whereas 11% of the cells were V3.2⁺ V3⁺ (Fig. 2B). Further analysis of 26 individual M-MuLV-immune B6.V^a mice revealed a good correlation between the percentage of V3.2⁺ and the percentage of V3⁺ and/or V17⁺ cells in the CD62L CD8⁺ subpopulation (Fig. 3A). Nevertheless the percentages of V3.2⁺ and V3⁺ and/or V17⁺ cells in the CD62L CD8⁺ subset were quite variable among individual immune mice and in a few cases did not exceed backgrounds levels found in normal mice.

View larger version (51K): [in this window] [in a new window]   FIG. 2.   M-MuLV-immune CD62L CD8+ PBL in B6.Va mice preferentially express V3.2 in association with either V3 or V17. Four-color analyzes of isolated PBL from M-MuLV-immune B6.Va mice were performed with MAbs to CD8, CD62L, V3.2, and V3 or V17. (A) The cytogram represents the staining of CD62L versus CD8. Region 1 (R1) represents activated CD8+ (CD62L) cells, whereas region 2 (R2) represents nonactivated CD8+ (CD62L+) cells. (B) The four cytograms represent V3.2 staining versus V3 or V17 expression in the indicated subsets. (C) The three histograms represent V3.2, V3, and V17 staining gated on CD62L CD8+ (shaded area) or CD62L+ CD8+ (nonshaded area) cells.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   Expression of V3.2 in association with either V3 or V17 on CD8+ PBL from individual M-MuLV-immune B6.Va mice. Four-color analyzes of isolated PBL from 39 individual M-MuLV immune B6.Va mice were performed with MAbs to CD8, CD62L, V3.2, and V3 or V17. (A) Correlation curve between the percentages of CD8+ CD62L PBL expressing V3.2 in association with either V3 or V17. (B) Open and filled bars indicate, respectively, the percentages of V17+ and V3+ cells in the CD8+ CD62L subset.

Down-regulation of TCR and CD8 expression in M-MuLV-immune T cells. In accordance with the observations made for PBL from M-MuLV-immune B6 mice, between 30 and 50% of CD8⁺ cells from M-MuLV-immune B6.V^a mice were CD62L at the peak of the response whereas only 2 to 3% of total CD8⁺ cells from naive animals had this activated phenotype (data not shown). In addition, a down-regulation of both CD8 and TCR expression was observed in this activated CD62L subset of CD8 cells. Indeed, the level of expression of CD8, V3.2, V3, and V17 was down-regulated by two- to threefold in activated CD8⁺ cells (Fig. 2B and C). This result, which was previously observed in other model systems (5, 36), suggests that down-regulation of both the TCR and coreceptor is a common feature of antigen-specific activation of CD8⁺ cells in vivo.

Hierarchy of V3 and V17 gene usage among CD62L CD8⁺ cells from individual M-MuLV-immune B6.V^a mice. Representative data describing the percentages of V3⁺ or V17⁺ cells among the CD62L CD8⁺ subsets of 39 individual M-MuLV-immune B6.V^a mice are shown in Fig. 3B. The majority of the mice showed a strong expansion of specific CD8⁺ cells expressing V3⁺ and/or V17⁺ gene segments (between 60 and 80% of CD62L CD8⁺ cells were V3⁺ or V17⁺). The ratio of the percentages of V3⁺ and V17⁺ cells among activated CD8⁺ cells varied from mouse to mouse. In the majority of the responding mice, the V17⁺ response was predominant compared to the V3⁺ response. However, in some mice, V3 and V17 responses were equivalent and even in one mouse, the response was due mainly to V3⁺ cells.

Absolute magnitude of the M-MuLV-specific CD8⁺-T-cell response in immune B6.V^a mice. Since these experiments were performed by analyzing antigen-specific cells via four-color staining, it was possible to calculate the absolute magnitudes of the different subsets of M-MuLV-specific CD8⁺ cells at the peak of the response in immune B6.V^a mice. V3.2⁺ V3⁺ CD62L cells accounted on average for 0.5% of the CD8 subset and 0.05% of total PBL in these mice, whereas the proportions observed in naive animals were, respectively, <0.05% and <0.01%. The absolute number of V3.2⁺ V17⁺ CD62L cells was even higher, since 5% of the CD8 subset and 0.05% of PBL in immune mice had this phenotype compared to <0.1% and <0.01%, respectively, in naive mice. It is important to point out that significant variations from mouse to mouse were observed.

V3.2⁺ V3⁺ and V3.2⁺ V17⁺ CTL clones predominantly recognize the dominant FMR gag-encoded epitope. In B6 mice, the protective CD8⁺ CTL response is restricted by the H-2D^b molecule and inhibited by anti-H-2D^b MAb (37). A recent study has shown that the protective CD8⁺ CTL response is directed against an immunodominant epitope (CCLCLTVFL) encoded in the leader sequence of the gag polypeptide of M-MuLV (11). This epitope is shared by leukemia and lymphoma cell lines infected by the FMR group of leukemia viruses.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   V3.2+ V3+ and V3.2+ V17+ T-cell clones recognize the dominant gag-encoded epitope. Representative individual M-MuLV-specific CTL clones 6 (V3.2+, V5.2+, H-2Db-restricted, derived from a B6 mouse), 471 (V3.2+, V17+, H-2Db restricted, derived from a B6.Va mouse), 487 (V3.2+, V3+, H-2Db restricted, derived from a B6.Va mouse), and 464 (V3.2, V5.2, H-2Kb restricted, derived from a B6.Va mouse) were tested for cytotoxicity at an effector cell/target cell ratio of 3:1 against EL-4 target cells (H-2b, FMR uninfected) in the presence or absence of various concentrations of the FMR gag-encoded peptide CCLCLTVFL.

Hierarchy of V gene usage in M-MuLV-immune (B6 × B6.V^a)F₁ mice. In order to further analyze hierarchy in V usage in the M-MuLV immune response, (B6 × B6.V^a)F₁ mice, which express V3, V5.2, and V17 gene segments, were bred in our animal facilities and immunized with MBL-2 tumor cells by the same immunization protocol. PBL from M-MuLV-immune (B6 × B6.V^a)F₁ mice were highly enriched for V3.2⁺ cells in the activated (CD62L) subset of CD8⁺ cells following secondary immunization with MBL-2 cells (data not shown). Regarding V usage, we found with 25 individual mice analyzed that activated CD8⁺ cells preferentially utilized the V3 and V17 gene segments and with some mice that they scarcely utilized the V5.2 gene segment (Fig. 5A). In addition, strong individual differences in V3/V17 ratios were observed in MuLV-immune (B6 × B6.V^a)F₁ mice, confirming the observations for B6.V^a mice. Of 25 immune F₁ mice analyzed, most preferentially utilized V17 while one mouse exclusively utilized V3 and others equally utilized the two dominant V chains. A minor expansion of V5.2⁺ cells was also seen in a few mice.

View larger version (45K): [in this window] [in a new window]   FIG. 5.   Hierarchy of V gene usage in individual M-MuLV-immune (B6 × B6.Va)F1 mice. Four-color analyzes of isolated PBL from 25 individual M-MuLV-immune (B6 × B6.Va)F1 mice were performed with MAbs to CD8, CD62L, V3.2, V3, V5.2, and V17. (A) Open, hatched, and filled bars indicate, respectively, the percentages of V17+, V5.2+, and V3+ cells in the CD8+ CD62L subset. (B) Kinetics of the V17+, V5.2+, and V3+ CD8+-T-cell response in three representative M-MuLV-immune (B6 × B6.Va)F1 mice.

TCR and - junctional sequences of M-MuLV-specific CTL clones derived from immune B6.V^a and B6 mice. The recognition of the gag-encoded immunodominant epitope (CCLCLTVFL) by TCR molecules having the same V chain (V3.2) but in association with at least three different V chains (V3, V5.2, and V17) prompted us to analyze at the molecular level TCR and - junctional regions of specific CTL clones derived from immune B6 and B6.V^a mice to see if there was any conserved sequence in these critical regions.

View larger version (23K): [in this window] [in a new window]   FIG. 6.   TCR and - junctional amino acid sequences of M-MuLV-specific CTL clones derived from immune B6.Va and B6 mice. Fifteen H-2Db-restricted CTL clones derived from immune B6 (A) or B6.Va (B) mice plus bulk cultures (MLTC) derived from two different immune B6 mice are listed on the vertical axis. The V, V, J, and J segment usages are reported. Nomenclature and sequences for V and V segments are as described by Arden et al. (1). J sequences are from the work of Gascoigne et al. (20) and Chien et al. (13), and J sequences are from the work of Koop et al. (23). V and V usage were confirmed by surface staining with corresponding antibodies. The deduced amino acid sequences (in single-letter code) of the junctional, hypervariable, and putatively CDR3-like regions are indicated. The presumed immunoglobulin-like loop, designated CDR3 for convenience, is supported by two framework branches (FW).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The CD8⁺-T-cell response to M-MuLV (and, in general, FMR)-associated antigens in B6 mice is directed against an immunodominant gag-encoded epitope (CCLCLTVFL) and is restricted primarily to CTL expressing the V5.2 and V3.2 gene segments (5). The rationale of the experiments presented here was to examine the M-MuLV response in congenic B6.V^a mice which are unable to express the dominant V3.2⁺ V5.2⁺ TCR due to a large deletion at the TCR locus that includes the V5.2 gene segment. Interestingly, B6.V^a mice were still able to reject M-MuLV-infected tumor cells and CTL from these immune mice utilized the same V3.2 gene segment in association with two different V segments (V3 and V17). Surprisingly, these CTL recognized the same immunodominant M-MuLV gag epitope.

The fact that the same H-2D^b-restricted gag-encoded peptide CCLCLTVFL is recognized by V3⁺ and V17⁺, as well as V5.2⁺, TCR strengthens the argument that this is indeed a highly immunodominant epitope of M-MuLV (at least in H-2^b mice). This immunodominance is somewhat surprising in view of the fact that the M-MuLV gag peptide lacks a classical H-2D^b anchor residue at position 5. Indeed substitution of the normal H-2D^b anchor residue (Asp) for Leu at this position significantly increases the ability of the gag peptide to bind to H-2D^b. However, recognition of the modified peptide by CTL is greatly diminished (4a). The mechanism underlying the immunodominance of the gag epitope in H-2^b mice remains to be established. In this respect the hydrophobic nature of the peptide (which is encoded in the leader sequence of the gag polyprotein) or its putative ability to be efficiently processed may play a role.

Although CTL clones specific for the immunodominant FMR gag epitope were readily elicited in both B6 and B6.V^a strains, there was considerable variation among individual M-MuLV-immune mice in the frequencies of CD62L CD8⁺ PBL expressing the V3.2 gene segment in association with V3, V5.2, or V17. Nevertheless, all primed mice were able to reject M-MuLV-infected tumor cells. These data suggest that other CTL epitopes (in addition to the immunodominant one) are involved in protective immunity to M-MuLV-induced tumors. Moreover, helper-T-cell epitopes, which have been shown to be important for vaccination against FMR tumors (29), are likely to play an important role.

In interpreting the potential hierarchy of V usage in the M-MuLV response, it is important to note that the V3 and V17 gene segments utilized by most M-MuLV-specific CTL in V^a mice differ between the V^a and V^b haplotypes. In particular, the V17 gene is not expressed in the V^b haplotype due to the presence of a stop codon. Moreover, the V3 gene segment differs between the two haplotypes by a point mutation, resulting in a single amino acid substitution at position 31 (Phe in V^a versus Val in V^b). This polymorphic residue is located within the CDR1 domain of the V3 segment and thus might be expected to influence TCR recognition of the M-MuLV gag peptide by CD8⁺ cells since (i) the CDR1 region has been shown to contact the C-terminal residues of the antigenic peptide in crystallographic studies of TCR-MHC class I-peptide complexes (17-19), (ii) CDR1 polymorphism in the V10 gene segment has a dramatic influence on TCR recognition of the immunodominant H-2K^d-restricted HLA-CW3 epitope by CD8⁺ T cells (4), and (iii) CDR1 polymorphism of V3 has already been shown to dramatically affect TCR recognition of a dominant pigeon cytochrome c peptide by MHC class II-restricted CD4⁺ T cells (16). Thus, the failure of M-MuLV-immune CTL from V^b mice to use V3 or V17 is due to structural differences in these gene segments between the V^a and V^b alleles rather than an intrinsic preference for utilization of V5.2 within the V^b haplotype.

In order to directly compare the levels of utilization of V3, V5.2, and V17 in the CD8⁺-T-cell response to the M-MuLV gag epitope, we used (B6 × B6.V^a)F₁ mice in which all three V domains are expressed. Analysis of the TCR repertoire of individual F₁ mice revealed a clear hierarchy in V utilization. Thus, V17 was used most frequently by responding CD8⁺ CD62L cells whereas V3 (and especially V5.2) were used to much lesser extents. Several possible explanations for this hierarchal V usage can be considered. First, it is possible that V17-bearing TCR have a higher affinity for the M-MuLV gag peptide than V3- or V5.2-bearing TCR. Although difficult to test directly, this explanation is, however, not supported by the comparable peptide dose-response curves of gag-specific CTL clones expressing V17, V3, or V5.2. Second, it is possible that CD8⁺ T cells expressing V17⁺ gag-specific TCR arise more frequently than V3⁺ or V5.2⁺ TCR with the same specificity, perhaps due to differential positive selection in the thymus. In this respect it is interesting that V17 and V3 chains of gag-specific TCR were quite heterogeneous in J usage and CDR3 sequence but that V5.2 chains were highly conserved (see below). These sequence data raise the possibility that V17⁺ and V3⁺ TCR recognizing the gag peptide are more frequent than gag-specific V5.2⁺ TCR in the naive CD8⁺-T-cell population. Clearly, more direct experiments using MHC class I gag peptide tetramers in association with limiting dilution experiments will be required to address this issue.

Sequence analysis of the TCR and - chains utilized by CTL clones specific for the M-MuLV gag epitope revealed several important differences in the V^a and V^b haplotypes. Strikingly, TCR and - chains were absolutely conserved in V^b mice, since all CTL clones analyzed utilized V3.2-J13 and V5.2-J1.4 with completely conserved CDR3 and CDR3 regions. These conserved TCR and - chains were representative of the M-MuLV-specific V^b CTL population as a whole, since identical V3.2-J13 and V5.2-J1.4 junctional sequences were obtained by PCR from two independent polyclonal MLTC populations. In contrast, the TCR repertoires of M-MuLV gag-specific V^a CTL clones were considerably more diverse. In particular, the TCR chain was again strikingly conserved, with all V^a CTL clones utilizing V3.2-J6 and a highly conserved (although not identical) CDR3 region. However, the TCR chains of V^a CTL clones were much more diverse, since the two V domains used (V3 and V17) were associated with three distinct J segments and diverse CDR3 sequences.

These TCR sequence data are of interest in the context of a model proposing that the diversity of the TCR repertoire in response to a given foreign epitope is dependent upon the extent of homology between that epitope and self-determinants (9). According to this model, a consequence of self-tolerance will be that foreign epitopes that are highly homologous to self-peptides will elicit a restricted TCR repertoire but that epitopes unrelated to self will elicit a diverse TCR repertoire. In contrast, the data presented here indicate that the M-MuLV gag epitope CCLCLTVFL, which is not highly homologous to any expressed protein sequence (including endogenous retroviral leader sequences) in current databases (data not shown), can elicit a diverse repertoire of TCR chains in V^a mice and a highly restricted -chain repertoire in V^b mice. Since these congenic mouse strains differ only at the V locus itself, it seems probable that the diversity of the -chain repertoire in this instance is determined by the availability of V segments rather than by the extent of tolerance imposed by self-homology of the peptide epitope.

Finally, the finding that several structurally distinct TCR are able to recognize the same M-MuLV-encoded gag peptide associated with H-2D^b is consistent with growing evidence that TCR specificity is intrinsically highly cross-reactive or degenerate (27). Indeed, a recent crystallographic study has demonstrated that the same peptide-MHC complex can be recognized by two distinct TCR in the same orientation despite the fact that almost all the individual peptide-MHC contact residues differ (15). Conversely, a single TCR can recognize a wide variety of distinct peptide-MHC complexes with apparent high affinity, as in the case of the widely studied 2C TCR (34, 35). Whether such TCR degeneracy reflects selection by a common self-peptide in the thymus during development (21) remains to be established.