Alterations in TCR-MHC Contacts Subsequent to Cross-Recognition of Class I MHC and Singly Substituted Peptide Variants

Mice

The derivation of TCRα^−/−TgVα2⁺ transgenic mice was described in detail previously (27). These mice are transgenic for the TCR α-chain of the Vα2⁺Vβ13⁺ anti-VSV CTL clone N30.7 (23). Because they also carry a targeted disruption of both alleles of the endogenous TCR Cα locus, they cannot express their endogenous TCR α-chain genes. C57BL/6 (B6) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All mice were maintained in the Animal Institute of the Albert Einstein College of Medicine.

Cell lines

R8 is a heterozygous (H-2^bxd) pre-B lymphoblastoid cell line transformed by Abelson leukemia virus. R8 cell lines containing point mutations of H-2K^b have been previously described (23, 28, 29, 30, 31). The mutant R8 lines used in this study and the location and identity of their H-2K^b mutations are as follows: R8.8 (G56E), R8.9 (E58K), R8.62 (R62Q), R8.110.43 (R75Q), R8.18 (R79K), R8.24 (T80I), R8.313 (L82P), R8.208 (L82F), R8.125 (G90D), R8.127 (M138K), R8.14 (L141R), R8.331 (A150P), R8.110.2 (A158T), R8.34 (G162D), R8.347 (E166K), R8.10 (W167R), and R8.353 (N174K). Wild-type R8 cells and all H-2K^b mutants were maintained in DMEM supplemented with 5% heat-inactivated FCS. The H-2K^b mutants expressed comparable levels of MHC molecules relative to the wild-type R8 cells as revealed by flow cytometric analysis (32).

Peptide synthesis and purification

VSV peptide (RGYVYQGL) and all singly substituted variant peptides were synthesized by standard solid phase methods using F-moc chemistry in a peptide synthesizer (model 433A, Applied Biosystems, Foster City, CA) at the Peptide Synthesis Facility of the Albert Einstein College of Medicine. The peptides were cleaved from the resin, and the side chain protecting groups were removed using 95% trifluoroacetic acid. They were purified to >98% homogeneity by reverse phase HPLC on a Vydac C₁₈ semipreparative column (218TP510, Vydac, Hesperia, CA). The identity of the purified peptides was confirmed using a tandem quadrupole mass spectrometer (TSQ700, Finnigan MAT, San Jose, CA).

Establishment and maintenance of CTL clones

TCRα^−/−TgVα2⁺ transgenic mice were immunized in their hind footpads with 15 μg of peptide emulsified in CFA. The mice were boosted once with 15 μg of peptide emulsified in IFA 1 wk after the primary immunization. One week after the booster, spleen tissues were removed, and spleen cells (5 × 10⁷) were cultured in 10 ml of culture medium with 1 × 10⁻⁶ M peptide in tissue culture flasks (Falcon 3082, Becton Dickinson, Franklin Lakes, NJ) for 7 days at 37°C under 9% CO₂ in air. The culture medium used was Iscove’s modified Dulbecco’s medium supplemented with 10% heat-inactivated FCS (HyClone, Logan, UT), 2 mM glutamine, 50 U/ml penicillin, 50 μg/ml streptomycin, and 50 μM 2-ME. The harvested cells (1 × 10⁵) were passaged and maintained by weekly restimulation with 2 × 10⁻⁷ M peptide and mitomycin C (Sigma, St. Louis, MO)-treated B6 spleen cells (5 × 10⁶) in complete medium supplemented with 50 U/ml human recombinant IL-2 (Life Technologies, Gaithersburg, MD) in 24-well culture plates (Falcon 3047, Becton Dickinson). CTL clones were derived from CTL lines by limiting dilution in the presence of 2 × 10⁻⁷ M peptide and mitomycin C-treated B6 spleen cells (1 × 10⁶) in 96-well culture plates (3799, Costar, Cambridge, MA). Clones were tested for cytotoxicity and maintained by weekly restimulation as described for CTL lines.

Cell-mediated cytotoxicity assay

R8 and its H-2K^b point mutants were labeled with 1.85 MBq of Na₂⁵¹CrO₄ (Amersham, Arlington Heights, IL) for 1 h at 37°C under 9% CO₂ in air. They were washed and used as target cells. In sensitization assays, 10 μl of peptide solution was added to 5 × 10^{3 51}Cr-labeled target cells (100 μl) and incubated for 1 h at room temperature before adding effector cells (100 μl). Final peptide concentrations are given in the figure legends. After incubating target and effector cells for 4 h at 37°C under 9% CO₂ in air, the supernatants (100 μl) were removed, and their radioactivities were measured with a gamma counter. The percent specific lysis was calculated as 100 × (experimental release − spontaneous release)/(maximum release − spontaneous release), where experimental release is the radioactivity in the supernatant of target cells mixed with effector cells, spontaneous release is that in the supernatant of target cells incubated alone, and maximum release is that in the supernatant after lysis of target cells with 2% Triton X-100.

RT-PCR and TCR β-chain sequence determination

mRNA was purified from approximately 1 × 10⁶ Ficoll-purified CTLs using the QuickPrep Micro mRNA Purification Kit (Pharmacia, Piscataway, NJ). The mRNA was reverse transcribed into single strand cDNA using Moloney murine leukemia virus reverse transcriptase and oligo(dT)₁₅ as primer. Vβ7 or Vβ13 TCR cDNAs were amplified by PCR using a Vβ7-specific primer (5′-TACAGGGTCTCACGGAAGAAGC-3′) or a Vβ13-specific primer (5′-AGGCCTAAAGGAGCTAACTCCAC-3′), paired in each case with a Cβ primer (5′-CACTGATGTTCTGTGTGACAG-3′). PCR products were purified using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA), and the CDR3 regions were sequenced from both directions at the DNA Sequencing Facility of the Albert Einstein College of Medicine.

Anti-VSV clones are Vβ13⁺ and express TCR β-chains with highly homologous CDR3β loops

We previously generated TCR α-chain transgenic mice in a TCR α-deficient background using as the transgene the α-chain from the VSV peptide-specific CTL clone N30.7. We showed that immunization of these mice with the VSV peptide enabled us to generate CTL lines that could recognize and lyse target cells pulsed with the VSV peptide (27). From such lines, we have now raised four CTL clones, all of which are CD8⁺ T cells as revealed by flow cytometry (data not shown). All the CTL clones recognize VSV peptide-pulsed R8 target cells that express H-2K^b (Fig. 1⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Specificity of anti-VSV CTL clones for position 6 variant peptides. CTL clones were derived from TCRα^−/−TgVα2⁺ transgenic mice immunized with the VSV peptide. ⁵¹Cr-labeled R8 target cells were incubated with various concentrations of the VSV peptide or its position 6 variants for 1 h at room temperature before addition of effector CTL clones. Cytotoxicity was measured using a 4-h chromium release assay and an E:T cell ratio of 4.

Determination of the TCR β-chain sequences of the four CTL clones revealed remarkable sequence similarity (Fig. 2⇓). All the CTL clones generated from VSV immunization used the Vβ13 gene segment and had Gly and Val at positions β97 and β98 of the CDR3 loop, respectively. These results are consistent with our previous study in which we showed that the Vβ13 gene segment is used by >80% of primary CTLs generated by VSV immunization of our transgenic mice, and that the CDR3β Gly-Val motif is highly conserved in such CTLs (27). Two of the four clones had TCR β-chain sequences identical with those identified by RT-PCR analysis of primary CTLs reported in this previous study. Three of the four clones had a CDR3β loop of nine amino acids, the preferred length for Vβ13⁺ VSV-specific CTLs from the transgenic mice, although one (Tg.W2B) had 10 amino acids. Note that the TCR β-chain of Tg.W3A was almost identical with that of the N30.7 CTL clone, which was the source of the Vα2 transgene, except for the two amino acids at the N-terminal end of the CDR3β loop.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Amino acid sequences of TCR β-chain V-D-J junctional regions from the anti-VSV CTL clone N30.7 and anti-VSV clones derived from TCRα^−/−TgVα2⁺ transgenic mice. TCR sequences were determined as described in Materials and Methods. The N30.7 sequence was previously reported (23). The asterisks in A indicate amino acid sequences identical with those encoded by cDNA clones described in our previous study (27). In B, the CTL clones are grouped by similar sequences, with conserved residues shown in capitals and in shaded boxes. The amino acid residue at position 98 of each β-chain is shown in boldface.

Anti-VSV clones with highly homologous TCR β-chains differ in the ability to recognize position 6 and position 4 variant peptides

In our previous structural (25) and functional studies (24), positions 1, 4, and 6 of the VSV peptide were shown to be exposed for potential TCR interaction. Gln at position 6 is an important TCR contact residue that interacts with the TCR β-chain (27). We investigated the ability of our CTL clones to cross-react with position 6 variant peptides (Fig. 1⇑). Three of the four clones (Tg.W2B, Tg.W3A, and Tg.W7A) were specific for the VSV peptide and did not recognize any position 6 variant peptides, including those containing negatively charged amino acid substitutions (D6, Gln to Asp; E6, Gln to Glu), positively charged amino acid substitutions (K6, Gln to Lys; R6, Gln to Arg), or a hydrophobic amino acid substitution (I6, Gln to Ile). Only clone Tg.W4A was cross-reactive, being able to recognize the D6 and E6 peptides bound to H-2K^b, although its reactivity to D6 was weak. Clones Tg.W4A, Tg.W7A, and Tg.W2B all expressed the transgenic TCR α-chain and had TCR β-chains that were identical from the amino terminus up to and including β98 (Fig. 2⇑). However, only Tg.W4A showed cross-reactivity to the variant peptides D6 and E6. These results indicate that the amino acid residues C-terminal to position 98 of the TCR β-chain are important in determining the capability for cross-reaction to position 6 variant peptides.

To evaluate the ability of the four anti-VSV clones to cross-react with peptides substituted at the two other potential TCR contact sites, peptide variants with substitutions at position 4 or position 1 were analyzed for the ability to sensitize target cells to lysis by the CTL clones. While most substitutions at position 4 were not well tolerated, peptides with conservative substitutions were recognized by some of the clones (Fig. 3⇓). A4 (Val to Ala) was permissive for recognition by Tg.W3A, Tg.W4A and Tg.W7A, while I4, containing a conservative methylene group addition (Val to Ile), was permissive for recognition by Tg.W2B and Tg.W3A. Tg.W7A and Tg.W2B have nearly identical TCR β-chains (Fig. 2⇑), except that Tg.W7A has Ser-Tyr at β99 and β100, while Tg.W2B has Gly-Gly-Arg at β99 to β101. Despite this extensive sequence similarity, they reacted differently to position 4 variant peptides. These results suggest that the CDR3β amino acid residues C-terminal to β98 are important in determining the capability for cross-reaction to position 4 variant peptides.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Sensitization activity of VSV peptide variants singly substituted at TCR-contacting residues. ⁵¹Cr-labeled R8 target cells were incubated with 1 × 10⁻⁷ M peptide for 1 h at room temperature before addition of the indicated CTL clones. Cytotoxicity was measured using a 4-h chromium release assay and an E:T cell ratio of 4. The activities of peptides singly substituted at positions 1 (left), 4 (center), or 6 (right) of the VSV peptide are shown. The results are the means of triplicate assays.

In contrast to the results obtained with position 4 variant peptides, most of the position 1 variant peptides were recognized by all four clones (Fig. 3⇑). Overall, no striking differences were seen among the four clones in the ability to recognize position 1 variant peptides, although Tg.W7A recognized the M1/Y1 and N1/H1 peptide mixtures very weakly.

Clones with very similar TCR β-chains show different recognition patterns on VSV/H-2K^b

A number of R8 cell lines containing point mutations of H-2K^b were previously selected using mAbs (29, 31) or CTLs (28, 30), and nearly all carry nonconservative substitutions in the α1 or α2 helix of the K^b molecule. For each of the anti-VSV clones, these R8 cell lines were used to identify contacts made between the TCR and its ligand during the process of recognition. The R8 cell lines were assessed for recognition by each CTL clone, and loss of recognition of a cell line expressing K^b with a specific mutation was assumed to reflect a TCR-K^b contact point at the mutated residue. Responses of the anti-VSV CTL clones to the H-2K^b mutants are shown in Fig. 4⇓. The recognition pattern of N30.7 is also shown (23). Most of the clones were only affected by mutations located on the α2 helix of the K^b molecule, and common contact points for all the TCRs were identified at residues 166, 167, and 174. Tg.W3A, whose entire TCR β-chain differs from that of N30.7 by only two amino acids in CDR3β (positions 95 and 96), showed a similar, but not identical, recognition pattern on VSV/H-2K^b as did N30.7. Likewise, Tg.W2B and Tg.W7A, which also have highly homologous TCR β-chains (Fig. 2⇑), nonetheless showed different TCR recognition patterns. These results show that CTL clones with highly homologous TCR β-chains can have different recognition patterns on the same peptide/MHC class I complex. Conversely, clones with more dissimilar TCR β-chains can show identical recognition patterns. For example, Tg.W3A and Tg.W4A, which use different Jβ segments and share only two amino acids in CDR3β, showed identical recognition patterns on the H-2K^b mutants.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Recognition of the VSV peptide bound to the R8 H-2K^b point mutants. ⁵¹Cr-labeled R8 H-2K^b mutants were incubated with 1 × 10⁻⁸ M VSV peptide for 1 h at room temperature before addition of the indicated CTL clones. The position and identity of each K^b mutation are shown at the left. Cytotoxicity was measured using a 4-h chromium release assay and an E:T cell ratio of 3 (Tg.W2B) or 4 (Tg.W3A, Tg.W4A, Tg.W7A). The results for N30.7 (E:T cell ratio of 6) were previously reported (23). After comparison with the percent specific lysis of R8 wild-type target cells incubated with the same peptide, the results for each H-2K^b mutant are represented as follows: black square, lysis of the mutant was <30% of that of R8 wild-type cells (the mutation abolishes recognition); hatched square, lysis between 30–60% of that of wild-type R8 (the mutation partially affects recognition); and white square, lysis >60% of that of wild-type R8 (the mutation does not affect recognition). The results shown are the means of two to four independent assays.

The cross-reactive anti-VSV clone Tg.W4A demonstrates that single amino acid changes in the peptide can alter the recognition pattern of a cross-reactive TCR on its peptide/MHC ligand

We measured the responses of the cross-reactive Tg.W4A clone to the VSV peptide and several variant peptides presented by each of the H-2K^b mutants to assess the TCR-MHC interactions that occur during cross-recognition events. The recognition patterns diagrammed in Fig. 5⇓A show that certain mutations on the α1 helix of H-2K^b differentially affected the recognition of these peptides by Tg.W4A, demonstrating that these MHC residues are differentially contacted by the TCR depending on the peptide present in the peptide binding groove. (Quantitative relative lysis data for relevant α1 mutants are provided in Fig. 5⇓B.) The differences in the recognition patterns of VSV/H-2K^b and E6/H-2K^b shown in Fig. 5⇓A are particularly noteworthy, as we have shown that these ligands are recognized equally well by Tg.W4A in cytotoxicity assays over a broad range of peptide concentrations (Fig. 1⇑).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Single amino acid changes in the peptide alter the recognition pattern of Tg.W4A on the H-2K^b molecule. A, ⁵¹Cr-labeled R8 H-2K^b mutants were incubated with VSV, D6, E6, K1, or A4 peptides at a concentration of 1 × 10⁻⁷ M (D6, K1, and A4) or 1 × 10⁻⁸ M (VSV and E6) for 1 h at room temperature before addition of Tg.W4A. Cytotoxicity was measured using a 4-h chromium release assay and an E:T cell ratio of 4. The location of each point mutation is represented as a circle on the H-2K^b structure. (Two mutations were evaluated at position 82: L82P and L82F.) After comparison with the percent specific lysis of R8 wild-type target cells incubated with the same peptide, the results for each H-2K^b mutant are represented as follows: black circle, lysis of the mutant was <30% of the lysis of wild-type R8 (the mutation abolishes recognition); hatched circle, lysis between 30–60% of that of wild-type R8 (the mutation partially affects recognition); and white circle, lysis >60% of lysis of wild-type R8 (the mutation does not affect recognition). The results shown are the means of two to four independent assays. B, ⁵¹Cr-labeled R8 cells carrying point mutations on the α1 helix of H-2K^b were incubated with VSV (1 × 10⁻⁸ M), D6 (1 × 10⁻⁷ M), or E6 (1 × 10⁻⁸ M) peptides for 1 h at room temperature before addition of Tg.W4A. Cytotoxicity was measured as described in A. For each peptide and K^b mutant combination, the percent relative lysis was calculated as (% specific lysis of R8 mutant/% specific lysis of wild-type R8) × 100. Results shown are the means of two to four independent assays ± SEM. Competition assays indicated that all three peptides bound equally well to the H-2K^b mutants (data not shown).

Anti-D6 and anti-E6 clones are Vβ7⁺ and express TCR β-chains with highly homologous CDR3β loops

In an attempt to raise a second panel of CTL clones expressing identical TCR α-chains and highly homologous β-chains, we immunized the TCRα^−/−TgVα2⁺ transgenic mice individually with D6 or E6 peptide and derived CTL clones capable of recognizing the immunogen bound to H-2K^b (Fig. 6⇓). The anti-D6 and anti-E6 clones expressed the Vβ7 gene segment, and their TCR β-chains had Ser-Leu at positions β95-β96 in the N-terminal portion of the CDR3β loop and a conserved Arg at position β98 (Fig. 7⇓). We previously showed the importance of the Arg at β98 in the interaction with the negatively charged residue at position 6 of the D6 and E6 peptides (27). Two of the three CTL clones had β-chain sequences identical with those identified in primary CTLs in this earlier study.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Specificity of anti-D6 and anti-E6 CTL clones to position 6 variant peptides. CTL clones were derived from immunization of TCRα^−/−TgVα2⁺ transgenic mice with the D6 peptide (Tg.D6.1F and Tg.D6.2A) or the E6 peptide (Tg.E6.2C). ⁵¹Cr-labeled R8 target cells were incubated with various concentrations of the VSV peptide or its position 6 variants for 1 h at room temperature before addition of the indicated CTL clones. Cytotoxicity was measured using a 4-h chromium release assay and an E:T cell ratio of 2 (Tg.D6.1F and Tg.D6.2A) or 3 (Tg.E6.2C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Amino acid sequences of TCR β-chain V-D-J junctional regions from anti-D6 and anti-E6 CTL clones derived from TCRα^−/−TgVα2⁺ transgenic mice. TCR sequences were determined as described in Materials and Methods. The asterisks in A indicate amino acid sequences identical with those encoded by cDNA clones described in our previous study (27). In B, the CTL clones are grouped by similar sequences, with conserved residues shown in capitals and in shaded boxes. The amino acid residue at position 98 of each β-chain is shown in boldface.

Anti-D6 and anti-E6 clones with highly homologous TCR β-chains differ in the ability to recognize position 6 variant peptides

The anti-D6 and anti-E6 CTL clones were all cross-reactive to some position 6 peptide variants, but their fine specificities differed (Fig. 6⇑). Tg.D6.1F and Tg.D6.2A showed different specificities to position 6 variant peptides even though they had identical TCR β-chain sequences up to and including β99 (Fig. 7⇑). Supporting the conclusion drawn from analysis of the anti-VSV peptide clones, these results again indicate that the amino acid residues C-terminal to β99 are important in determining the capability for cross-recognition of position 6 peptide variants.

The cross-reactive anti-D6 and anti-E6 clones demonstrate that single amino acid changes in the peptide can alter the recognition pattern of a cross-reactive TCR on its peptide/MHC ligand

For each of the anti-D6 and anti-E6 clones, comparison of the recognition patterns of peptide/MHC using the immunogen peptides and the cross-reactive peptides revealed marked differences depending on the peptide being recognized (Fig. 8⇓). For example, the Tg.D6.1F TCR contacted position 62 on the α1 helix of K^b when recognizing the E6 peptide but not when recognizing the D6 peptide. Conversely, position 80 was contacted during D6 recognition by this TCR, but not during E6 recognition. Confirming the results obtained for the anti-VSV clone Tg.W4A (Fig. 5⇑A), the recognition patterns diagrammed in Fig. 8⇓ indicate that a single TCR can interact differently with the class I MHC molecule depending on the peptide being recognized, and that its recognition pattern can be affected by even subtle changes in a single residue of the peptide.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

Cross-reactive anti-D6 and anti-E6 clones show peptide-dependent alterations in their recognition patterns of H-2K^b molecules. ⁵¹Cr-labeled R8 H-2K^b mutants were incubated with VSV, D6, or E6 peptides at a concentration of 1 × 10⁻⁷ M (D6) or 1 × 10⁻⁸ M (VSV and E6) for 1 h at room temperature before the addition of Tg.D6.1F (A), Tg.D6.2A (B), or Tg.E6.2C (C). Cytotoxicity was measured using a 4-h chromium release assay and an E:T cell ratio of 2 (Tg.D6.1F and Tg.D6.2A) or 3 (Tg.E6.2C). The relative recognition of each peptide/H-2K^b mutant is represented as described in Fig. 5⇑A. The results shown are the means of two to four independent assays.

Small sequence changes in CDR3β are sufficient to alter the recognition pattern of a TCR on the class I MHC molecule

We previously demonstrated that VSV peptide-specific T cell clones from wild-type mice with quite diverse TCRs showed different recognition patterns on VSV/H-2K^b (23). In contrast to CTL clones generated from immunization of wild-type mice, the anti-VSV clones that we raised in this study expressed identical TCR α-chains and highly homologous β-chains. Nonetheless, we again found that each anti-VSV clone showed an individual recognition pattern on its H-2K^b ligand (Fig. 4⇑). Further, our data show that small sequence alterations in either the N- or C-terminal portions of the CDR3β loop are sufficient to alter the recognition pattern of a TCR on its ligand. Additional examples of individual recognition patterns were provided by our anti-D6 clones (Fig. 8⇑). The observation that a small change in the sequence of the TCR β-chain can alter the recognition pattern of a TCR on its ligand has also been made in the class II MHC system (33).

Cross-reactivity to position 4 and position 6 variant peptides can be mapped to short segments of CDR3β

In the VSV/H-2K^b system, we have previously shown that a number of anti-VSV CTL clones derived from wild-type B6 mice can cross-react with certain VSV peptide variants substituted at TCR contact position 1, 4, or 6 (23, 24). Different clones show different sensitivities to such substitutions in terms of both the positions that can be altered and the particular substitutions that are tolerated. However, the lack of homology between the TCRs of these clones precludes identification of the TCR sequences that are responsible for this differential cross-reaction capability. The presence of two productively rearranged TCR α-chains in some of the clones further compounds this problem.

In contrast to CTL clones generated from wild-type mice, the anti-VSV clones described in this study had highly homologous TCR β-chains that were paired exclusively with the transgenic TCR α-chain. Two such clones (Tg.W7A and Tg.W2B) showed different patterns of cross-reaction to position 4 variant peptides (Fig. 3⇑), with Tg.W7A reacting only to A4 and Tg.W2B reacting only to I4. These two clones have identical TCR β-chains, except that Tg.W7A has Ser-Tyr at β99-β100 near the middle of the CDR3β loop, while Tg.W2B has an extra amino acid in this region and the sequence Gly-Gly-Arg (Fig. 2⇑). These findings allow us to conclude that changes in this short stretch of amino acids can alter the cross-recognition of a TCR to position 4 variant peptides, suggesting the possibility of a direct interaction between this region of CDR3β and position 4 of the peptide. Support for such an interaction is provided by the crystal structures of the A6 and B7 TCRs complexed with the Tax 9-mer peptide and HLA-A2, in which the central residue of the peptide is found in a pocket between the CDR3α and CDR3β loops, and both loops form contacts with the central residue of the peptide (7, 9).

Our results show that cross-reactivity to position 6 variant peptides can also be mapped to a short segment of CDR3β. Tg.W4A and Tg.W7A are CTL clones that exhibit different abilities to cross-react with position 6 variants of the VSV peptide (Fig. 1⇑), yet both of the clones are Vβ13⁺ and have TCRs with CDR3β loops that are identical from positions β95–β98 (Fig. 2⇑). These findings indicate that the amino acid residues C-terminal to β98 in the CDR3β loop are important in determining cross-reaction to peptides altered at position 6. Although the two TCRs use different Jβ gene segments, it is unlikely that J-encoded amino acids C-terminal to the CDR3β loop also contribute to this differential cross-reaction, as these amino acids are framework residues. Further, seven of these 10 J-encoded amino acids are identical in the two TCRs, and another is a conservative Ile to Leu change (34). Results obtained with our anti-D6 clones also indicate the importance of the amino acid residues C-terminal to position β99 in the CDR3β loop in influencing cross-recognition to position 6 variant peptides (Figs. 6⇑ and 7⇑). These results are consistent with our previous study in which we demonstrated an interaction between the CDR3β loop and position 6 of the peptide (27).

The observation made in our class I MHC system that changes in a limited region of a CDR3 loop can alter the cross-reaction ability of a TCR was also made for the class II system by Hsu et al. (35). They studied the cross-recognition capabilities of Hb_64–76/I-E^k-reactive T cell hybridomas from TCR β-chain transgenic mice and showed that a change of only two amino acids in the N-terminus of the CDR3 loop of the α-chain altered the fine specificity of the hybridomas for peptide variants at the central positions of the peptide.

Single conservative amino acid substitutions in the peptide are sufficient to alter the contacts between a cross-reactive TCR and the class I MHC molecule

The anti-VSV CTL clone Tg.W4A cross-reacts with certain singly substituted VSV peptide variants, including K1, A4, D6, and E6 (Figs. 1⇑ and 3⇑). Its cross-reactive TCR exhibits different recognition patterns on a class I MHC molecule depending on the peptide that is present in the peptide binding groove of the class I molecule (Fig. 5⇑A). We have found that recognition contacts can be altered by single substitutions at any of the three TCR contact residues of the VSV peptide, and that even relatively conservative amino acid changes (e.g., Arg to Lys in K1, Val to Ala in A4, Gln to Glu in E6, or Gln to Asp in D6) can lead to dramatically altered recognition patterns (Fig. 5⇑A). Results obtained with our three cross-reactive anti-D6 and anti-E6 CTL clones provide further evidence that single conservative amino acid substitutions in the peptide are sufficient to alter the recognition pattern of a cross-reactive TCR on its peptide/MHC ligand (Fig. 8⇑).

The idea that a single TCR can contact the same MHC in different ways was first proposed by Ehrich et al. in the class II MHC system (33). They described two CD4⁺ T cell hybridomas that recognized both a murine cytochrome C peptide (amino acids 88–103; 99K) and its 99E variant presented in the context of the class II MHC molecule I-E^k. For both hybridomas, the interaction with a panel of I-E^k mutants was found to be different depending on which peptide was being recognized. However, it was difficult to evaluate the significance of these findings, because the two hybridomas were approximately 100 times more sensitive to 99K than to 99E (containing Lys to Glu, which is a nonconservative change in terms of both size and charge). We were able to overcome this problem by identifying peptide variants that were well recognized by our cross-reactive CTL clones. This allowed us to show that even ligands that are recognized equally well by a CTL clone can be contacted differently by its TCR. For example, VSV and E6 are recognized to an equal extent by Tg.W4A (Fig. 1⇑), yet the recognition patterns of Tg.W4A for VSV/K^b and E6/K^b are quite different (Fig. 5⇑A).

In our experiments, for a given cross-reactive clone, peptide-dependent alterations in the recognition patterns were located primarily on the α1 helix (Figs. 5⇑A and 8). Only for Tg.D6.2A did a mutation on the α2 helix (at residue 166) abolish the recognition of one peptide while not affecting the recognition of another. These results are in contrast to those reported in the class II MHC system by Ehrich et al. (33). In their report MHC mutations that abolished the recognition of a given cross-reactive hybridoma for either 99E or 99K (but not both) were all found on the β1 helix, which is analogous to the α2 helix of the class I molecule. This difference suggests the intriguing possibility that a given TCR may adapt to slightly altered peptide ligands in different ways depending on whether the peptide is bound to a class I or a class II molecule.

There are reasons to believe that class I- and class II-restricted TCRs might bind their ligands in different ways. First, class I- and class II-restricted T cells use different MHC-binding coreceptors (CD8 or CD4, respectively). Second, although class I and class II MHC molecules have remarkably similar structures, there are some differences that are likely to be relevant to T cell recognition. For example, the class II molecule has a peptide-binding site that is open at both ends (36, 37). As a result, the N- and C-termini of the peptide can protrude out of the groove, allowing class II molecules to bind peptides having a variety of lengths (10–34 amino acids) (38, 39, 40). In contrast, the N- and C-termini of peptides bound to class I molecules form hydrogen bonds with conserved MHC residues and are buried at the ends of the binding groove (25, 26, 41, 42). For this reason, class I MHC molecules bind peptides of more defined length (8–10 amino acids) (22, 43, 44) than do class II molecules. Also relevant to the issue of T cell recognition, the total exposed peptide surface is larger for class II-bound peptides than for class I-bound peptides (45). Further, for peptides bound to class I, the exposed peptide surface is largely at the middle of the peptide, whereas it is distributed more uniformly across class II-bound peptides (45). In addition, Carson et al. (46) recently presented striking evidence that peptide-flanking residues of class II-bound peptides that lie outside the minimal epitope that is required to bind to the class II molecule can be directly recognized by certain TCRs and can even act as dominant TCR contact residues. Taken together, this information lends support to the idea that cross-reactive TCRs might adapt to variant peptide/MHC complexes in different ways depending on which class of MHC molecule is in the complex.

We have shown that single amino acid substitutions of TCR contact residues in an antigenic peptide can alter the recognition pattern of a cross-reactive TCR on its peptide/class I MHC ligand. These altered recognition patterns could potentially be due to peptide-induced changes in the conformation of the MHC molecule. Previous functional studies using peptide variants containing single substitutions of TCR-inaccessible residues suggested the occurrence of peptide-specific alterations in MHC conformation (47, 48). Structural evidence for such a phenomenon was provided by Fremont and colleagues (49), who compared the crystal structures of H-2K^b complexed with three unrelated peptides, OVA-8 (SIINFEKL), VSV-8 (RGYVYQGL), and Sendai virus-9 (FAPGNYPAL). They found that four MHC side chains took on peptide-specific conformations (Lys⁶⁶, Glu¹⁵², Arg¹⁵⁵, and Trp¹⁶⁷). Similarly, when Madden et al. (42) crystallized the human class I MHC molecule HLA-A2 with four different, unrelated 9-mer peptides or a 10-mer, they found three MHC residues (Arg⁹⁷, Tyr¹¹⁶, and Trp¹⁶⁷) that could exhibit peptide-specific conformations. For both H-2K^b and HLA-A2, structural comparisons in the presence of different peptides revealed a slight main chain shift at the N-terminal portion of the α2 helix (amino acids 144–151). Therefore, when a peptide is changed to an unrelated one, minor conformational changes can occur in the MHC molecule that could potentially be recognized by a TCR.

In contrast to the unrelated peptides studied by Fremont et al. (49) and Madden et al. (42), the peptide variants that we studied contained only single amino acid substitutions of TCR contact residues. Reid et al. (50) recently compared the crystal structures of HLA-B8 complexed with an HIV-1 P17 8-mer (GGKKKYKL) or two singly substituted variants (7Q, GGKKKYQL; 7R, GGKKKYRL) altered at position 7, a TCR contact point. When the three different structures were compared, the only changes observed were alterations in the mobilities of the position 7 side chain and two local MHC side chains on the α1 helix (Glu⁷⁶ and Asn⁸⁰). This result then suggests to us that the VSV peptide and its singly substituted variants are unlikely to form peptide/MHC complexes with dramatically different conformations, although crystal structures of H-2K^b with each of the variant peptides are needed to confirm this suggestion. Instead, depending on the particular substitution that is made in the peptide, a change in a single amino acid would alter the charge, size, and/or shape of the side chain available for contact with the TCR, and this change might also alter the orientations of neighboring side chains on the peptide or the MHC molecule (50). The altered recognition patterns that we observed for a given cross-reactive TCR on different peptide/H-2K^b complexes are likely to reflect an adaptive response by the TCR to such subtle local changes at or near the altered peptide residue. The ability of the CDR loops of a TCR to take on alternate conformations was recently demonstrated by Garcia et al. (8) when they compared the crystal structures of the 2C TCR alone or complexed with its ligand. It is therefore reasonable to hypothesize that a cross-reactive TCR adjusts the conformation of the CDR loops to accommodate several different, but closely related, ligands. Our finding that single conservative substitutions in the peptide are sufficient to alter the recognition contacts between a cross-reactive TCR and the class I MHC molecule is consistent with this hypothesis.