The Mode of Ligand Recognition by Two Peptide:MHC Class I-Specific Monoclonal Antibodies

Mice

OT-1 TCR transgenic mice (22) were obtained from Drs. F. Carbone (Monash Institute, Melbourne, Australia) and W. Heath (Walter and Eliza Hall Institute, Melbourne, Australia) and were maintained in the Memorial Sloan-laettering Cancer Center vivarium.

Cells

TAP-deficient cell lines RMA-S (23) and SJ-2 (24), expressing K^b and K^bm8 class I molecules, respectively, were used for the thermodynamic stabilization assays. The S.bm cell lines (S.bm1, S.bm3, S.bm5, S.bm8, S.bm10, S.bm11, S.bm23), expressing the indicated K^bm mutant molecules (25), were used to map the footprints of D1 and C5 over the K^b molecule. A detailed map of the mutations and the solvent accessibility of the side chains of involved amino acids is shown in Fig. 1⇓. This provides a measure of the side chain availability for the TCR contact; buried residues have no solvent accessibility and chiefly participate in the contacts with the MHC.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Ribbon diagram of the H-2K^b molecule with the position and orientation of the residues changed in K^bm molecules. The highlighted amino acid residues denote the changes occurring in the K^bm mutants. The list below the figure details the characteristics and the predicted orientation of the side chain of each residue changed in individual K^bm mutants, as deduced from Refs. 18, 25, and 30, with an emphasis on functional studies in the OVA-8 model using K^bm molecules (25 ). TCR denotes solvent accessibility and the potential of the side chain to interact with the TCR. It does not implicate lack of interaction of the side chain with the MHC.

Peptides

A complete list of the peptides used in the experiments is provided in Table I⇓. All peptides were variants of the H-2K^b binding immunodominant OVA_257–264 (OVA-8, SIINFEKL) peptide except for GAG-10, the HIV gp120-derived decamer (RGPGRAFVTI), which does not bind K^b and was used as a negative control. All peptides were purchased from Research Genetics (Huntsville, AL). The peptides were named to identify the original amino acid in OVA-8, the position of the substituted amino acid, and the new amino acid. For example, N4R is a variant of OVA-8 in which the Asn residue (N) at position 4 was replaced with Arg (R). Otherwise, peptide positions in the text are referred to as P# from the N terminus (e.g. P2 is the second position from the N terminus).

Abs and flow cytofluorometric (FCM) analysis

Abs 25-D1.16 (D1) and 22-C5.9 (C5) (21) were used as tissue culture supernatants from hybridoma kindly donated by R. Germain (National Institutes of Health). The K^b-specific mAb Y3 (American Type Culture Collection, Manassas, VA) was used in the form of ascitic fluid, generated in our laboratory. PE-conjugated secondary Abs specific for IgG1 (to detect D1), IgG2a (C5), and IgG2b (Y3) were purchased from Fisher Biotech (Malvern, PA). Cells were stained on ice for 30 min using excess Ab, washed, and then incubated with an excess of fluorochrome-conjugated secondary Ab. Following another incubation and a wash, cells were resuspended in a paraformaldehyde fixative and analyzed on a FACScan instrument using the CellQuest 3.1 software (Beckton Dickinson, Mountain View, CA). A total of 5–10 × 10³ cells were analyzed per sample using unstained, isotype-stained, and secondary Ab controls to determine specific fluorescence. Results are reported as mean fluorescence intensity (MFI) corrected for the background, or as a derivative or normalized values thereof, as described for the assays below.

Assays for mAb recognition of pep:MHC complexes

Peptide binding to K^b and K^bm8 molecules was assessed by a FCM-based assay that quantifies the surface expression of thermodynamically stable class I molecules on TAP-deficient cell lines in the presence of different concentrations of peptides, exactly as described previously (24, 26). Briefly, TAP-deficient cells were incubated overnight at 29°C to allow maximal accumulation of empty class I molecules at the cell surface. Peptide was then added to triplicate samples in serial 10-fold dilutions (10⁻⁵–10⁻¹⁰ M) and the cells were incubated for 30 min at 29°C, following by another 3 h at 37°C. Excess peptide was then washed (three times), and each of the triplicate samples was stained with Y3, D1, or C5. Y3 recognizes only properly conformed MHC molecules, but it binds to all K^bm molecules (25), and this binding is not affected by the nature of the bound peptide. As this mAb also detects the “empty” MHC molecules (in fact, probably the ones occupied with suboptimally fitting peptides) that are stable at 29°C but not at 37°C, it is suitable for quantification of total expression of K^b-like molecules (24). Therefore, Y3 fluorescence was used to provide a measure of overall K^{b or bm} expression at 37°C in normal cells (S.bm series) or a measure of both the capacity for “maximal” K^{b or bm8} expression and of peptide binding in TAP-deficient cells.

To be able to compare data from different experiments, the Y3 MFI was converted into stabilization percentage, with MFI of cells incubated overnight at 29°C = 100% and MFI of cells shifted to 37°C in the absence of peptide = 0%. These stabilization percentage values were normalized at each peptide concentration to the percentages obtained using the index peptide I2S (unlike OVA-8, I2S binds in an equivalent fashion to K^bm8 and K^b). For D1 and C5, which do not recognize K^b in the absence of OVA-8, the MFI of I2S at 10⁻⁵ M was taken as 100%, and the other values were normalized to it.

To assess mAb recognition of the OVA-8:K^bm complexes, S.bm cells were incubated overnight with 10⁻⁶ M of OVA-8 or GAG-10 at 37°C. The cells were then washed and stained with Y3 to determine the H2-K levels, and with D1 and C5 to measure specific recognition of the various OVA-8:class I complexes. The MFI values were then converted into specific % of OVA-8:K^b binding using the following formula: [experimental MFI (OVA-8:K^bm) − control MFI (GAG-10:K^bm)]/maximal MFI (OVA-8:K^b) − control MFI (GAG-10:K^b)] × 100.

Ig chain sequencing

Sequencing of the Ig heavy and light chains was carried out as described (27). For sequencing of the Ig heavy and light chains of C5 and D1, total RNA was extracted from 5 million C5 and D1 hybridoma cells using the RNA-isolator kit (Genosys Biotechnologies, The Woodlands, TX) and quantitated by optical density. A total of 5 μg of total RNA was then reverse-transcribed in a 30 μl reaction using the Avian myeloblastosis virus (AMV)-reverse transcriptase kit (Boehringer Mannheim, Indianapolis, IN). After the reaction was completed, 30 μl of double-distilled water was added and 2 μl of the final mixture was used for PCR. PCR was performed using sense V-specific and antisense C-specific primers (Vκ consensus, GGRGTCCCATCAAGGTTCAGTGGC; Cκ, GGGAAGATGGATACAGTTGGTGCA; VhJ558, AAGGCCACACTGACTGTAGAC; IgG1, AAGGACACAGGGATCATTTACC; IgG2, GTGRGTGCTGAGCTCAATTTACC, where R is any nucleotide). The PCR products were visualized on a 1.5% agarose gel using ethidium bromide and purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA). One-fourth of the purified PCR products was then sequenced by using the same primers as for the PCR and the BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA). Nucleotides contributed by the V H/k, J H/k, or D genes were identified by alignment with germline gene sequences (28). Only those nucleotides that could not have been contributed by one of these genes were designated as VH-D or D-JH junctions for the heavy chains and as Vk-Jk junctions for the light chains.

OVA-8:K^b-specific mAbs D1 and C5 contact OVA-8:H-2K^b near the C terminus of the peptide

To compare the recognition patterns of D1 and C5 to that of OVA-8:K^b-specific TCRs, we mapped the footprint of the mAbs over the K^b molecule. To this end we used the K^bm mutants previously shown to bind OVA-8 and measured D1 and C5 binding to the various OVA-8:K^bm complexes (25). The K^bm mutants are spontaneous natural variants of the K^b molecule that differ from it by up to 5 aa changes located in the α1 and α2 domains (29). Fig. 1⇑ shows the location of substituted amino acids in a ribbon diagram of the K^b molecule. Some of these changes are predicted to dominantly affect TCR binding, some to influence peptide binding and conformation, whereas others are likely to influence both, as judged by the solvent accessibility of the side chains of the OVA-8:K^b complex (30) and the actual contacts of a different peptide, dEV-8, with K^b and the 2C TCR (18) (Fig. 1⇑).

With the exception of K^bm1 and K^bm8, it was previously demonstrated that all K^bm mutants can present OVA-8 to OVA-8-specific CTLs. Furthermore, mice expressing various K^bm mutants are able to mount a CTL response against the OVA-8 epitope (25, 29). As a consequence of an altered Ag binding B pocket, the K^bm8 molecule binds OVA-8 only at high peptide concentrations (10⁻⁵–10⁻⁶ M) whereas K^b can still bind it at 10⁻⁸–10⁻⁹ M. Furthermore, bm8 mice do not mount a CTL response against OVA-8 due to a lack of positive selection of OVA-8-specific CTLs (29).

The effects of the K^bm mutations on the recognition pattern of D1 and C5 are summarized in Figs. 2⇓ and 3. An example of actual FCM profiles is provided in Fig. 2⇓, illustrating binding of mAbs to K^b and K^bm3. Recognition by D1 was abrogated by the K^bm1, K^bm3, and K^bm11 mutations. K^bm1 is located on the α2 helix near the C terminus of the peptide and affects both TCR and peptide binding (Refs. 29 and 32; Fig. 1⇑). K^bm11 and K^bm3 mutations are located at the C-terminal part of the α1 helix and share a substitution at amino acid residue 77. However, as the mutation at position 77 is also shared with K^bm23, and the OVA-8:K^bm23 complex is recognized by D1, it is unlikely that the change at residue 77 affects recognition. The most plausible possibility is that D1 is sensitive to independent changes in K^bm3 (residue 89) and K^bm11 (residue 80), none of which are present in K^bm23. Alternatively, and much less likely, the mutation at residue 77 could directly affect recognition by D1 of the OVA-8:K^b complex, but the second mutation at residue 75 of K^bm23 would compensate for its effect.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

25-D1.16 (A) and 22-C5.9 (B) binding to the OVA-8:K^b and OVA-8:K^bm3. FCM histograms are shown for the fluorescence obtained for S.B6 (K^b, solid line) and S.bm3 (K^bm3, dotted line), following staining with no Ab (A), Y3 (B), D1 (C), or C5 (D). All cell lines were incubated with 10⁻⁶ M peptide for 1 h prior to staining, as described in Materials and Methods. Data are representative of four experiments.

Although C5 recognition was affected by the K^bm1 and K^bm3 mutations, it was not affected by the K^bm11 one (Figs. 2⇑ and 3⇓). This result can only be explained by concluding that D77 → S mutation plays no role in Ab binding, and that it is the change at loop position K89 → A that results in the loss of C5 recognition. Indeed, using K^bm3 point mutants, it was shown that this mutation affects binding of five mAbs, whereas neither of them was affected by D77 → S (33). These results show that D1 and C5 must contact the MHC residues at the C-terminal part of the α1 helix, close to the C terminus of the peptide, suggesting that both Abs interact with the pep:MHC area contacted by the β-chain of the TCR and with the α1 helix β-strand 5 loop outside of the peptide binding site.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

25-D1.16 (A) and 22-C5.9 (B) binding to the OVA-8:K^bm complexes. The results are represented as percentage recognition standardized to the MFI of the parental complex, OVA-8:K^b, as detailed in Materials and Methods. The black and gray bars represent the percentage recognition by D1 or C5 of the different OVA-8:K^bm and GAG-10:K^bm complexes, respectively. GAG-10 is the HIV gp120 decamer that does not bind to any of the molecules tested. OVA-8 and GAG-10 were used at 10⁻⁶ M. Results are shown as an average of triplicates ± SD, representative of four independent experiments. For the position of amino acid changes, please refer to Fig. 1⇑.

This hypothesis was further tested by comparing the patterns of recognition of D1 and C5 to that of the OVA-8-specific T cell line OT-1.2. Splenocytes obtained from an OT-1 transgenic mouse were stimulated in vitro with irradiated syngeneic splenocytes coated with OVA-8. They were then tested for their recognition of the OVA-8:K^bm complexes (K^bm8 was excluded from the analysis owing to the well-known inability to present OVA-8 to K^b-derived CTLs). Target cells expressing K^bm1, K^bm3, and K^bm11 complexed to OVA-8 were not lysed by the OT-1.2 T cells. The same pattern of reactivity was found in the OVA-8-specific CTL clone GA4 (data not shown), confirming the previously published results (25). The biological data obtained from the OT-1 and GA4 OVA-8-specific CTL cells correlate with the pattern of recognition of D1 and C5 obtained through FCM analysis and further support the idea that the footprints of D1 and C5 at the C terminus of OVA-8:K^b overlap with that of the TCR β-chain.

mAb recognition of OVA-8:K^b involves the TCR contact residues P4, P6, and P7 of OVA-8

To determine the amino acid residues of OVA-8 important for recognition by D1 and C5, single amino acid substitutions of OVA-8 were assayed for K^b binding and recognition by D1 and C5. A complete list of the OVA-8 variants analyzed (with the exception of the P2 variants I2D and I2E, which bound poorly to K^b and were therefore excluded from this analysis) is given in Table I⇑. Typical peptides presented by K^b are 8 aa long and contain anchor residues F/Y at P5 and L/I at P8 (34). These amino acids are buried in the MHC pockets C and F, respectively, and their alteration severely affects K^b binding. Side chains at positions 2 and 3 are also completely or partially buried and can serve as auxiliary MHC binding anchors (35). Furthermore, in certain peptides, including OVA-8, mutations at the P2 have the potential to induce a change in the peptide conformation that can be detected at the level of TCR recognition (30). Residues P1S, P4N, P6E, and P7K are solvent accessible and were functionally implicated in TCR contact because their substitution alters the recognition by OVA-8:K^b-specific CTLs (24, 26, 31). To compare the specificity of OVA-8:K^b recognition by T cells to that by D1 and C5, we mutated residues at positions 1, 2, 4, 6, and 7. No changes were made at P3, P5, and P8 to preserve OVA-8 binding to K^b (Table I⇑).

We used TAP-deficient RMA-S cells to evaluate OVA-8 variant recognition by D1 and C5. The RMA-S cells were stained with Y3 to obtain a measure of OVA-8 variant binding to K^b and with D1 and C5 to measure recognition of the OVA-8 variant:K^b complex (Figs. 4⇓ and 5). The percentages obtained at 10⁻⁶ M were chosen for comparison between the OVA-8 variants because they typically resided at the linear portion of the titration curves. The comparative results are qualitatively summarized in Table II⇓.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

mAb recognition of the OVA-8 and position 4 OVA-8 variant:K^{b or bm} complexes. MHC class I stabilization assay was performed as described in Materials and Methods. The level of class I expression was determined by Y3 staining and measured by flow cytometry, ascertaining that all peptides bound well to K^b (A) and K^bm8 (D). The level of specific D1 (B and E) and C5 (C and F) binding were also determined by FCM. Mean fluorescence values were converted to relative stabilization, and further standardized to the index peptide I2S, as described in Materials and Methods. The remaining of the data is summarized in Tables II and III based on the percentage recognition values at 10⁻⁶ M. Results are shown as mean ± SD (n = 3) and are representative of six experiments.

Both D1 and C5 recognized P1 charge and size variants indistinguishably from the original peptide, OVA-8. D1 also tolerated many, and C5 tolerated all, changes at P4. A notable exception for D1 were the substitutions N4R and N4K, which completely abrogated recognition by this mAb (Fig. 4⇑B). To test whether this effect was due to the charge or to the size of these residues, we assessed the effects of the large hydrophobic N4F substitution. In this case, the large hydrophobic side chain of F did not interfere with D1 binding, indicating that it was the positive charge at P4 that interfered with D1 recognition.

All single amino acid changes at P6 and P7 completely abolished recognition by D1 (Fig. 5⇓). The same pattern applied to C5, with the exception of the K7A peptide, that was weakly recognized by this mAb (Fig. 5⇓). These results indicated that P6 and P7 must be directly involved in mAb contact, consistent with the initial analysis of Porgador et al. (21). Interestingly, the double substituant E6AK7R was recognized by D1 (but not by C5), despite the fact that the mutations E6A and K7R were not individually tolerated by this mAb (Table II⇑). One explanation for this effect is that the two mutations compensate for each other when present in the same variant (currently under investigation).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Recognition of position 6 and 7 variants of OVA-8. MHC class I stabilization assay was performed as described in Materials and Methods. The level of class I expression was determined by Y3 staining and measured by FCM, ascertaining that all peptides bound well to K^b (A). The level of specific D1 (B) and C5 (C) binding were also determined by FCM. Mean fluorescence values were converted to relative stabilization and further standardized to the index peptide I2S, as described in Materials and Methods. The remaining data are summarized in Tables II and III based on the percentage recognition values at 10⁻⁶ M. Results are shown as mean ± SD (n = 3) and are representative of six experiments.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

The approximate footprints of 25-D1.16 and 22-C5.9 on the K^b:OVA-8 complex as suggested by the recognition patterns obtained in this study. Results shown in this Figure and Tables II and III, and discussed in the text, demonstrate that the footprint of the two Abs, although largely overlapping, is not identical. Most notably, C5 is less sensitive than D1 to the changes at P4 and to the K^bm11 mutation and thus may be positioned more towards the C terminus of the peptide than D1.

Overall, the recognition of the OVA-8:K^b complex by D1 and C5 was not affected by changes at P1 or by those at P2 that did not alter the binding capacity of the variant. The binding of both Abs was severely affected by changes at P6 or P7 and that of D1 by introduction of positive charge at P4. Thus, D1 and C5 directly contact residues P6 and P7. D1 also must have made contacts in the vicinity of the residue P4 because it was affected by positively charged but not other variants at this position, whereas C5 was not affected by any of the P4 variants, suggesting that its footprint could be shifted towards the C terminus of K^b relative to that of D1 (also see Fig. 4⇑ and the text below). The data also suggest that C5 does not contact P7 to the same extent as D1 because it could partially tolerate the K7 → A change. Previous studies in single chain transgenic mice using peptide immunizations have shown that the TCR CDR3β interacts with the OVA-8 residues P4, P6, and P7, with the TCR α-chain making contacts to the P1 side chain (36). Therefore, OVA-8 variant and K^bm recognition data concordantly indicate that D1 and C5 footprints over the OVA-8:K^b complex overlap with that of the TCR β-chain.

D1 and C5 detect the K^bm8-induced conformational change at the solvent-exposed central part of the OVA-8:K^bm8 complex

The above data indicate that the recognition by D1 and C5 of OVA-8:K^b overlaps the area contacted by TCR β-chains. Hence, they may be considered mimics of a soluble TCRβ and be used to probe conformational changes introduced into the bound peptide by class I mutations. We investigated whether D1 and/or C5 can detect subtle differences in peptide conformation (24, 37) introduced by the K^bm8 mutation. This mutation encompasses amino acid mutations at positions 22, 23, 24, and 30, all buried within the peptide binding groove (38). Hunt et al. (39) showed that changes at positions 22 and 24 carry the bulk of the biological effect of this K^b variant. Both residues are part of the B pocket of the MHC, which interacts with the side chain of the P2 residue of the peptide. Their mutation results in a disruption of the hydrogen bond network that exists within the B pocket of K^b (24), with a potential to affect peptide binding and conformation. According to the TCR crystal structures and the biological data, the N-terminal domain of the pep:MHC complex is contacted by the α-chain. However, bm8 mice use a different CTL Vβ repertoire in response to HSV than the wild-type B6 mice (R. Dyall, I. Messaoudi, J. LeMaoult, and J. Nikolić-Z̆ugić, unpublished observations), suggesting that the conformational differences introduced by the B pocket mutation may extend into the central peptide portion, including the residues contacted by the β-chain.

We investigated whether D1 or C5 can detect the conformational differences between the OVA-8:K^b and OVA-8:K^bm8 complexes. To this end, we used the thermodynamic stabilization assay with TAP-deficient cell lines expressing K^b or K^bm8 and the OVA-8 peptide variants described in Table I⇑. We then compared the recognition pattern of D1 and C5 of the peptide:K^bm8 complex to that of peptide:K^b complex. The recognition pattern of D1 remained unchanged whether OVA-8 variants were bound to K^b or K^bm8 (Fig. 4⇑, B and E). However, in contrast to what was observed on K^b, the recognition pattern of C5 of the K^bm8 complex was now abolished by the presence of positive charges at P4 (Fig. 4⇑, C and F; Table III⇓). This strongly suggests that the side chain of P4 has moved upon binding to K^bm8, compared to its position in complex with K^b, so that it now can affect C5 binding.

The data indicate that the effect of the B pocket mutation in K^bm8 does not remain localized to the P2 residue but rather affects the orientation of the solvent-accessible TCR contact residues in the middle of the peptide. This result confirms the conclusions of our previous study using CTL recognition (24) and extends it to provide the first direct measure of the conformational differences introduced by the K^bm8 mutation.

CDR3 sequence analysis of D1 and C5 chains reveals a conserved R between mAb heavy chains and OVA-8 specific TCR β-chains

Positively charged residues at P4 abrogate the binding of D1 and C5 to OVA-8:K^b and OVA-8:K^bm8, respectively. To investigate the molecular basis for this repulsion, the CDR3 regions of the heavy and light chain of both D1 and C5 were sequenced (Table IV⇓). The two Igκ-chain CDR3s were identical and contained an R in the middle of the region. The two IgH chains differed, but contained an R-K pair at the same positions in the CDR3s. It is tempting to speculate that one or more of these positively charged residues may have caused the inability of these mAbs to bind to OVA-8 variants which had a positive charge at P4, but so could have any of the other CDRs not sequenced in this study. Curiously, a conserved R residue was previously observed in many TCRβ CDR3 obtained from OVA-8:K^b specific CTL lines and clones (35, 40).

Concluding remarks

In this study, we investigated the molecular contacts that define ligand recognition of two peptide:class I-specific mAbs and compared them to that of the TCR. The data show that D1 and C5 footprints overlap that of the β-chain of the TCR. Mutations of the K^b molecule at the C terminus of the α1 and α2 helices disrupted the binding of D1 and C5, whereas changes at the N terminus did not affect binding of either Ab. Moreover, changes at the TCR contact residues P6, P7, and, to a lesser extent, P4, were detrimental to the binding of both D1 and C5, whereas changes at TCR contact residue P1 did not affect either Ab. TCR contact residues P4, P6, and P7 were previously shown to be contacted by the CDR3 of the β-chain, whereas P1 is contacted by the α-chain. Therefore, D1 and C5 partly mimic TCR β-chain interaction with OVA-8:K^b. One possible footprint of these Abs over the pep:MHC complex is shown in Fig. 5⇑. (Our data are consistent with this layout. However, this should be considered a coarse and a necessarily imprecise approximation, shown for illustrative purposes.) Additional preliminary data using the point mutants generated by Nathenson and colleagues (41, 42) support our conclusions. For example, neither D1 nor C5 could recognize alterations at positions 80 (α1 helix) and 141 (α2 helix), and both were oblivious to the changes at positions 158 and 173 (the peptide N-terminus-proximal part of the α2 helix) (data not shown). The influence of residue 141 on the mAb binding stands in contrast to its lack of effect on TCR binding (41), again stressing the overlapping, but not identical, contacts between TCR and anti-pep:MHC mAbs.

Based on the above findings, the anti-pep:MHC Abs studied here appear to make a compromise between a TCR-like and an Ab-like mode of interaction with Ag. Unlike the TCR, these Abs do not interact with the pep:MHC complex in a parallel fashion to the β-pleated sheets. Unlike the typical anti-MHC Abs, they also do not wrap around only one part of the MHC molecule (e.g., the loops behind the α1 helix, or parts of α helices; Refs. 33 and 42). Rather, the recognition mode appears to be a hybrid between the two. At least one loop is involved (the loop connecting the end of the α1 helix to β-strand 5, carrying the residue 89), but so is a larger and flatter area composed of the MHC residues 80 (α1 helix) and one or more residues in the 141–156 segment (α2 helix) and the peptide residues P4, P6, and P7. This area is contacted by the TCR β-chain during TCR recognition. The above findings are entirely consistent with the affinity of D1 and C5 (8 × 10⁻⁷ M; 21), which is significantly lower than that of most affinity-matured IgG Abs (10⁻⁹–10⁻¹² M; 15), but is also clearly higher than that of TCRs (10⁻⁵–10⁻⁶, exceptionally 1 × 10⁻⁷ M; 2).

D1 and C5 also detected conformational changes introduced by mutations of the K^b molecule that exclusively affected peptide binding. Therefore, they provided the first direct measure of the conformational difference introduced by the K^b allele K^bm8. Indeed, previous experiments all relied on CTL lysis, which involved multiple steps from TCR ligation by pep:MHC to CTL degranulation and target cell lysis. Abs against pep:MHC complexes can therefore be useful in studying the general topology of pep:MHC complexes.