Antibody specific for the peptide.major histocompatibility complex. Is it T cell receptor-like?

Antibodies recognizing peptide bound to a major histocompatibility complex (MHC) protein usually have a higher affinity for the composite peptide.MHC (pMHC) ligand than T cell receptors (TCR) with the same specificity. Because the solvent-accessible peptide area constitutes only a small portion of the contacting pMHC surface, we hypothesized that the contribution of the MHC moiety to the TCR-pMHC complex stability is limited, ensuring a small increment of the binding energy delivered by the peptide to be distinguishable by the TCR or the peptide-specific antibody. This suggests that the gain in affinity of the antibody-pMHC interaction can be achieved through an increase in the on-rate without a significant change in the off-rate of the interaction. To test the hypothesis, we have analyzed the binding of an ovalbumin peptide (pOV8) and its variants associated with soluble H-2Kb protein to the 25-D1.16 monoclonal antibody and compared it with the binding of the same pMHC complexes to the OT-1 TCR. This comparison revealed a substantially higher on-rate of the antibody-pMHC interaction compared with the TCR-pMHC interaction. In contrast, both the antibody and the TCR-pMHC complexes exhibited comparably fast off-rates. Sequencing of the 25-D1.16 VH and VL genes showed that they have very few somatic mutations and those occur mainly in framework regions. We propose that the above features constitute a signature of the recognition of MHC-bound peptide antigens by TCR and TCR-like antibodies, which could explain why the latter are rarely produced in vivo.

T cell antigen-specific receptors (TCR) 1 must recognize major histocompatibility complex (MHC) protein and at the same time discriminate between many different peptides bound to that MHC. Approximately 80% of the solvent-accessible area of the peptide is buried in the MHC binding groove (1), limiting the amount of energy that a peptide can contribute to the TCR-pMHC interaction. To ensure TCR specificity for the MHC-bound peptide, the amount of binding energy coming from the MHC moiety must be restricted, allowing the peptide contribution of the composite ligand to be distinguishable by the TCR. In accord with this idea, it has been found that only two to three residues of an MHC class I protein mediate critical TCR-MHC contacts (2,3). In addition, experimentally measured values of the free energy for various TCR-pMHC interactions (4 -8) are significantly lower than the free energy of a typical protein-protein interaction (9,10). However, it has been shown that the TCR intrinsic affinity can be significantly increased without the loss of its specificity (11). The increase in the TCR affinity was mainly the result of the on-rate, not the off-rate, indicating that limited activation energy of the dissociation phase of the TCR-pMHC reaction is required to preserve the peptide specificity. This led us to suggest that, similar to the TCR with enhanced affinity, the apparent increase in the binding energy of pMHC-specific antibodies is achieved through a faster on-rate, without a significant change in the off-rate of the interaction.
To test this hypothesis we analyzed the binding of an ovalbumin peptide, SIINFEKL (pOV8), and its variants with known biological activities (12) in association with soluble K b protein to the pOV8⅐K b -specific monoclonal antibody (mAb) 25-D1.16 (13). Biosensor technology was used to compare the binding parameters with those measured previously (14,15) for the interaction of the OT-1 TCR with the same set of pMHC complexes. Comparison of the equilibrium and kinetic constants of OT-1 TCR and 25-D1. 16 antibody binding parameters revealed that an increase in intrinsic affinity of antibody interactions with peptide⅐K b complexes was mainly the result of changes in the on-rate but not the off-rate. We have also found very few somatic mutations in the variable domain of antibody heavy chain (V H ) and light chain (V L ) 25-D1. 16 genes encoding this mAb, which appeared to be very similar to the corresponding germ line genes. Based on these data, we suggest that the above features are essential characteristics of the recognition of short peptides bound to MHC proteins by TCR and TCR-like antibodies. This knowledge improves our understanding of the specificity of recognition of pMHC ligands and may be useful for designing pMHC-specific reagents. It may also help to understand why pMHC-specific antibodies are usually selected from combinatorial libraries in vitro but are very rarely elicited in vitro in response to immunization in vivo. NYEKL (V-OVA), EIINFEKL (E1), SAINFEKL (A2), SIIRFEKL (R4), and SIINFEDL (D7); RGYVYQGL, a vesicular stomatitis virus nucleocapsid protein (52-59) peptide (VSV); FAPGNYPAL, a Sendai virus nucleoprotein (324 -332) peptide (SV9); and KVVRFDKL, a chicken ovalbumin (55-62) peptide (KVDL).
Soluble pOV8⅐K b Complexes-The pRMHa-3 plasmids coding the H-2K b extracellular domain of the H-2K b with His 6 tag at the Cterminal end and mouse ␤ 2 -microglobulin were kindly provided by Anders Brunmark. These plasmids and a plasmid containing a neomycin resistance gene (Invitrogen) were co-transfected into Schneider cells (S2) by calcium phosphate precipitation, and stable transfectants were selected as described previously (16). The cells were expanded in serumfree medium Sf-900 II (Invitrogen) and grown to a density of 1.4 -2.0 ϫ 10 7 cells/ml. The expression H-2K b had been induced by 1 M cupric sulfate for 72 h, and soluble H-2K b molecules were isolated from the culture supernatant essentially as described previously (8,17).
Soluble, "empty" H-2Kb protein was loaded with peptides of interest. Typically, 50 g of peptide in 5 l of Me 2 SO was added to 1.2 mg of H2-K b in 250 l of phosphate-buffered saline, pH 7.4, and the reaction mixture was incubated at room temperature (22-24°C) overnight. The peptide⅐MHC complexes were stored in the presence of peptide excess at 4°C. Gel filtration of the peptide⅐K b complexes on a Sephacryl S200 HiPrep 16/60 column (Amersham Biosciences) did not reveal the presence of aggregates in the samples.
The monoclonal antibody 25-D1.16 was purified from culture supernatant by affinity chromatography on protein G-agarose. Fab fragment was produced by papain digestion and purified on a MonoQ anion exchange column (Amersham Biosciences). The identity of purified Fab fragment was confirmed by SDS-PAGE and enzyme-linked immunosorbent assay with soluble pOV8⅐K b ligand.
N-terminal Sequencing of mAb 25-D1. 16 Heavy and Light Chains-Partial N-terminal amino acid sequences of mAb 25-D1.16 heavy and light chains were determined by Edman degradation on a PROCISE-cLC (Applied Biosystems) sequencer. The heavy chain of the mAb contains glutamine at the N terminus that cyclizes to pyroglutamic acid precluding sequencing. To remove pyroglutamic acid, 25-D1.16 mAb was treated with pyroglutamate aminopeptidase from Pyrococcus furiosus (Takara Biotechnology) essentially as described previously (19).
8.5 mg of dithiothreitol and 300 mg of guanidine chloride (final concentration approximately 3 M) were added to 0.5 ml of the antibody (2 mg/ml) in 0.35 M Tris-HCl, pH 8.5, and incubated under argon at 60°C for 90 min. After cooling to room temperature, solid sodium iodoacetate (28 mg) was added to the reaction, and the mixture was incubated in the dark for 45 min at room temperature. The reaction was terminated by the addition of 5 mg of solid dithiothreitol. The reduced/ carboxymethylated mAb 25-D1.16 was dialyzed against 100 mM sodium phosphate buffer, pH 8.0, containing 2 mM EDTA overnight at 4°C, and the protein concentration was adjusted to 1 mg/ml. 5 l of glycerol, 5 l of pyroglutamate aminopeptidase from P. furiosus stock solution (0.2 units/ml) in 1ϫ reaction buffer (50 mM sodium phosphate buffer, pH 7.0, containing 10 mM dithiothreitol and 1 mM EDTA), and 25 l of 5ϫ reaction buffer were added to 0.1 mg of reduced/carboxymethylated IgG, and the mixture was incubated at 37°C for 24 h. The heavy and light chains of deblocked 25-D1.16 mAb were separated by SDS-12% PAGE at reducing conditions, blotted onto polyvinylidene difluoride membrane, and stained by Coomassie Blue G-250. The protein bands corresponding to heavy (50 kDa) and light (25 kDa) chains were excised, and 10 and 30 cycles, respectively, of Edman sequencing were performed on each sample.
Cloning and Sequencing of V H and V L Genes-Total RNA was extracted from 5 ϫ 10 6 hybridoma 25-D1.16 cells using the RNeasy Mini kit (Qiagen). Approximately 5 g of total RNA was reverse transcribed in a reaction volume of 20 l using 200 units of SuperScript II RNase H Ϫ reverse transcriptase (Invitrogen) and 13 pmol of oligo(dT) 18 primer according to the manufacturer's (Invitrogen) protocol. PCR amplification was carried out using Taq DNA/polymerase (Eppendorf). The V H gene was amplified with 5Ј-ATG GGA TGG AGC TGG ATC TTT CTC-3Ј (HFOR) and 5Ј-CTC AAT TTT CTT GTC CAC CTT GGT GC-3Ј (HBACK) primers, whereas the light chain variable region (V L ) gene was amplified with 5Ј-ATG AAG TTT CCT TCT CAA CTT CTG CTC-3Ј (LFOR) and 5Ј-CTA ACA CTC ATT CCT GTT GAA GCT CTT GAC-3Ј (LBACK) primers. The amplified fragments were inserted into the pGEM-T Easy vector (Promega) and transformed into competent Escherichia coli JM109 cells. Transformants were selected by ␣-complementation screening, and plasmid DNA from several positive clones was purified. Sequencing of cDNA was performed by BigDye terminator reaction on an ABI PRISM 377 DNA sequencer (Applied Biosystems).
Binding of Peptide⅐K b Complexes to 25-D1. 16 Antibody-Analysis of the interaction between 25-D1.16 mAb and soluble peptide⅐K b complexes was performed on a model 3000 Biosensor (Biacore, AB). The antibody 25-D1.16 in 10 mM acetic buffer, pH 5.0, was covalently immobilized to the surface of sensor chip CM5 by amine coupling to obtain a surface density of 100 -400 resonance units. In the control flow cell, polyclonal mouse antibodies were immobilized at the same level and were used as a negative control.
Before experiments were performed, peptide⅐K b complexes were transferred to HBS-EP buffer (10 mM HEPES, pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.005% surfactant P20 (Biacore, AB)) using Micro Bio-Spin 6 columns (Bio-Rad). Soluble peptide⅐K b complexes were injected into the flow cells at concentrations ranging from 50 to 3000 nM and a flow rate of 30 l/min at 25°C. Six different analyte concentrations were used. Binding of analyte (peptide⅐K b complexes) was measured by surface plasmon resonance with association and dissociation times of 1 and 2 min, respectively. Injection mode KINJECT, optimized for kinetic measurements, was used in all experiments. To regenerate the surface, 0.1 M glycine-HCl, pH 2.2, was applied for a contact time of 0.5 min at a flow rate of 90 l/min. The response curve of each analyte sample on the control surface was subtracted from the corresponding experimental curve. Analysis of the binding kinetics and calculations of the rate constants of association (k a ) and dissociation (k d ) were performed using BIAevaluation version 4.1 software. No differences in calculated values of k a and k d were observed in control experiments, in which reduced surface densities of immobilized mAb 25-D1. 16 were used, showing that the interactions were not limited by mass transfer. Values of k a and k d were determined as averages of at least two independent measurements.
Thermodynamic Analysis-The free energy of binding (Gibbs energy) was calculated from the value of the equilibrium binding constant K eq , where R is the universal gas constant (R ϭ 1.987 cal K Ϫ1 mol Ϫ1 ) and T is the temperature (in degrees K). The difference in free energy between OT-1 and 25-D1. 16 binding to pOV8 and the other peptide⅐K b complexes was determined as shown in Equation 2.
The contribution of the association (⌬G a n ) and dissociation (⌬G d n ) phases of the interactions was determined from the Eyring equation (20), where k is the reaction rate constant of association or dissociation, k B ϭ 3.3 ϫ 10 Ϫ24 cal K Ϫ1 and h/2 ϭ 1.586 ϫ 10 Ϫ34 cal s are the Boltzmann and the Planck constants, respectively, and is the transmission coefficient. Because the value is not known, actual values of ⌬G a n and ⌬G d n are difficult to determine; here we assume that the value is similar for both reactions. Differences in energy contributions of ⌬G a n and ⌬G d n to the binding energy between OT-1 and 25-D1. 16 interactions with various peptide⅐K b complexes were calculated as shown in Equations 4 and 5.
⌬⌬G a ϭ ⌬G a n (25-D1. 16 1C) (GenBank TM accession no. AY704179) and is joined to the J 5 segment (Fig. 1D). Five base pair differences were found with the germ line sequence, of which two are silent mutations. The others resulted in amino acid replacements in FR-L1 (Met-4 to Val) and FR-L3 (Gly-68 to Arg and Ser-74 to Ile). No mutations were present in the sequence contributed by J 5.

Analysis of V H and V L Sequences-The
CDRs of the 25-D1.16 mAb and the OT-1 TCR-Based on x-ray crystallographic analysis, CDRs of heavy and light chains of Ig molecules can be described in term of known canonical structures (21,22). The conformation of a particular canonical structure is determined by the length of the loop and the nature of amino acid residues at key positions. This conformation is conserved despite different positioning of the canonical loops relative to the framework regions. The structure of the hypervariable regions -H1, -H2, -L1, and -L2 of mAb 25-D1.16 (Fig. 2) suggests that they belong to canonical classes 1, 2A, 2B, and 1, respectively (21). The CDR-H3 that is expected to contact MHC-bound peptide is much more variable and is not included in the canonical structure description (22), whereas CDR-L3 belongs to canonical class 1.
The primary structures of the CDRs of the OT-1 and 25-D1.16 mAb are shown in Fig. 2. They appear to be very different, indicating that CDRs with divergent sequences may determine very similar specificities for pMHC binding (see below).
Binding Interactions of the 25-D1.16 mAb to Various Peptide⅐K b Complexes- Fig. 3 shows specific binding of the soluble pOV8⅐K b complex to the 25-D1.16 mAb immobilized on a biosensor surface. The binding of two irrelevant peptide⅐K b complexes containing either Sendai virus peptide (SV9) or peptide from vesicular stomatitis virus was undetectable (not shown).
The values of kinetic and equilibrium (affinity) constants for the 25-D1.16 mAb binding to pOV8⅐K b and variant peptide⅐K b complexes with defined biological activities measured at 25°C are summarized in Table I. Because the mAb was immobilized on the biosensor surface, soluble monovalent pOV8⅐K b complexes bound to the antibody-binding sites independently, allowing measurements of intrinsic parameters of the interaction. This was confirmed by measuring binding of the antibody monovalent Fab fragment to pOV8⅐K b immobilized on the biosensor surface (data not shown).
Derivation of the k a and k d values using a 1:1 Langmuir binding model demonstrated a good correlation between the experimental and globally fitted data ( 2 ϭ 0.05). The equilibrium binding constant K eq was calculated from the ratio k a /k d . For a two-step reaction, K eq measured from the maximum (plateau) binding response is expected to be different from the K eq value measured by the k a /k d ratio, suggesting a complex mechanism of the reaction (23, 24). Very often, however, the intermediate complex of the reaction is short lived, and the difference between the two K eq values is not significant (25). We did not investigate this issue in detail here and used the K eq value determined from the ratio k a /k d to compare it with corresponding data measured for the binding of the same pMHC complexes to the OT-1 TCR (14).
Comparison of the Specificity of the OT-1 TCR with the 25-D1.16 mAb-To compare the specificity of 25-D1.16 mAb with OT-1 TCR, we analyzed binding to peptide⅐K b complexes in which biological activity (12,26) and the binding to the OT-1 TCR (14,15) were determined previously. The k a value of mAb 25-D1.16 binding to pOV8⅐K b was 2.9 ϫ 10 4 M Ϫ1 s Ϫ1 , 1 order of magnitude higher than the k a value for the OT-1 TCR-pOV8⅐K b interactions, although the difference in rate constant of dissociation for the two reactions was only approximately 3-fold (Table I). Thus, the gain in the free energy of the 25-D1.16-pOV8⅐K b interaction is primarily determined by an increase in on-rate but not by a decrease in off-rate. This suggests that the activation energy of the dissociation phase of the antibody-pMHC interaction is only slightly higher than that of the TCR-pMHC interaction.
Amino acid substitutions in MHC-bound peptide had an effect on the recognition of pOV8⅐K b by the OT-1 TCR that was similar to the 25-D1.16 mAb. For instance, the R4⅐K b complex, which induces positive selection of immature OT-1 ϩ thymocytes (26) and antagonizes cytolytic activity of mature OT-1 cytotoxic T lymphocytes (12), bound to the 25-D1.16 mAb with lower affinity as it did to the OT-1 TCR (Table I). In both cases, the loss of the free binding energy was caused by a faster off-rate, whereas changes in the on-rate were marginal. Similarly, the K b complex with V-OVA peptide, which shares solvent-accessible amino acid residues with pOV8 and has biological activity similar to R4 (12, 26), had a faster rate of dissociation from the antibody and the TCR (Table I). Interestingly, the free energy of V-OVA⅐K b and pOV8⅐K b binding to the 25-D1.16 mAb were very similar because of a faster association rate of the V-OVA⅐K b -25-D1.16 interaction. The most profound effect resulted from the substitution of a positively charged Lys for a negatively charged Asp at P7 of the peptide (Table I). The binding of the antibody to the D7⅐K b complex was not detectable. Although the OT-1-D7⅐K b interaction has not been analyzed, the D7 peptide failed to elicit a cytolytic response (12). Provided that D7 binds to K b with an affinity similar to pOV8 and antagonizes responses of pOV8-specific OT-1 cytotoxic T lymphocytes (12), the D7⅐K b complex is apparently recognized by the TCR with a low affinity. Most likely, the peptide contribution to the interaction is marginal, resulting in a very low free energy of binding. The Ala substitution at P2 did not have any effect on binding of the A2⅐K b complex either to the 25-D1.16 mAb or to the OT-1 TCR (Table I). Consistent with this, A2 peptide is an agonist for mature pOV8-specific cytotoxic T lymphocytes and causes negative selection of immature T cells (27).
Substituting the first Ser with negatively charged Glu in the E1⅐K b complex, which positively selects immature T cells (26) and functions as an agonist/antagonist (12), did not affect binding of the complex to the 25-D1.16 mAb but resulted in a faster off-rate of the interaction with the OT-1 TCR (Table I). Substituting Ser with positively charged Arg (V-OVA) or Lys (KVDL) led to a slightly higher constant of association rate of the interaction with 25-D.1.16 mAb, although the interaction of the OT-1 TCR with V-OVA⅐K b was characterized by a faster offrate and a slower on-rate. Although binding of the KVDL⅐K b complex to the OT-1 TCR was not measured, this complex did not induce any detectable activity of pOV8-specific cytotoxic T lymphocytes (data not shown), indicating that the KVDL⅐K b complex is not recognized by the TCR, but it is recognized by the antibody. Thus, changing the first peptide amino acid influences the recognition of pOV8⅐K b by the antibody and the TCR in different ways.
The magnitude of the free energy changes (⌬⌬G) resulting from various peptide substitutions was usually higher for OT-1 TCR binding than for 25-D1.16 mAb binding, suggesting that the peptide energetic contribution to the TCR-pMHC interaction was more significant than its contribution to the antibody-pMHC interaction (Table I). Thus, in the latter case, the MHC moiety probably contributes more significantly, reflecting the 3-fold slower off-rate for the antibody-pOV8⅐K b reaction. However, this increase in MHC energy contribution is still small enough to ensure sufficient peptide contribution and the ability to distinguish the MHC-bound peptide.
To further dissect the mechanism used by 25-D1.16 mAb to recognize K b -bound peptides, we compared the energy contributed by the association and dissociation phases of OT-1-peptide⅐K b with those of 25-D1.16-peptide⅐K b interactions. Comparison was made using the Eyring transition state theory, which allows determination of quasi-thermodynamic parameters of the reaction association and dissociation phases based on respective rate constants (20). The ⌬G a n and ⌬G d n values reflect the amounts of energy contributed by the association and dissociation phases to the free energy of interaction. The difference in these contributions (Table II) to the antibody and the TCR binding reflects changes in the amount of energy coming from the association and the dissociation phases of 25-D1.16-peptide⅐K b interactions compared with corresponding OT-1-peptide⅐K b interactions. With only one exception, the changes in the energy contributions of the dissociation phase (⌬⌬G d ) were smaller than those of the association phase (⌬⌬G a ). Mutation of Ser to Glu (E1 peptide) had a different effect on the E1⅐K b interaction with the TCR and the antibody; it did not affect antibody binding but did decrease the stability of the OT-1-E1⅐K b complex (Table I). This explains an unusually high ⌬⌬G d value for antibody binding to E1⅐K b (Table II). These data provide evidence that the affinity gain for 25-D1. 16 to its natural ligand and other peptide⅐K b complexes was caused mainly by the lower energy activation of the association phase, whereas changes in the energy of the dissociation phase were limited to preserve the antibody specificity for the MHCbound peptide.

DISCUSSION
Although recognition of pMHC complexes on the cell surface is a normal function of the TCR, it has been shown that B cells can produce antibodies with TCR-like specificity in vivo as well (13,28,29). In addition, the technology of phage display librar-  c Reaction equilibrium binding constant, K eq , was determined as k a /k d ratio. d The free energy of binding (Gibbs energy), ⌬G, i.e. ⌬G ϭ ϪRT ln K eq e The difference in the free binding energy, i.e. ⌬⌬G, of receptor-pMHC interactions caused by replacement of K b -bound pOV8 to indicated peptides.
f NM, not measurable. g The binding of D7-K b to 25-D.1.16 immobilized on the biosensor surface was to too weak to measure; K eq of this interaction was estimated to be less than 10 4 M Ϫ1 .
h The binding parameters of OT-1 TCR to indicated peptide⅐K b complexes were determined previously by biosensor technology (14) and are presented for comparison.
i ND, not determined. j These values have been published previously (15).
ies (30,31) has been successfully used to generate pMHCspecific antibodies (32)(33)(34)(35)(36). Although these antibodies are of great utility and can be used to study antigen presentation as well as detection and targeting of virus-infected and transformed cells, the mechanism, which they use to specifically recognize an association with MHC peptide is not unclear. In one study, Biddison et al. (34) performed alanine scanning of MHC and peptide amino acid residues that make direct contact with the TCR and compared the effect of these substitutions on the recognition of these pMHC complexes by peptide-specific antibodies and the TCR. They found that some amino acid substitutions have very similar effects on the recognition of pMHC variants by the TCR and the antibody, whereas others produce different effects. This suggests that TCR and pMHCspecific antibodies use various contacts on the pMHC surface to achieve the same specificity. Such antibody, however, was produced in vitro using combinatorial libraries. In addition, the kinetics of the antibody binding to the pMHC complex have not been measured. Thus, it is important to investigate how the pMHC-specific antibodies that were derived in response to in vivo immunization achieve specificities similar to the TCR. In this study we determined the intrinsic equilibrium and kinetic constants for the binding of mAb 25-D1.16 to various peptide⅐K b complexes with known biological activity and compared these parameters with parameters for the interaction between the OT-1 TCR and the same set of peptide⅐K b complexes (Table I). The comparison yielded two major findings. The gain in affinity of the antibody for the peptide⅐K b ligands was mainly the result of an increase in the rate constant of association, and changes of the dissociation rate constant were significantly smaller. Differences between the binding of pOV8 and its variants associated with K b to the antibody and binding to the OT-1 TCR were mainly caused by significant changes in on-rate, whereas changes in off-rate were moderate. This pattern of changes suggests that the preservation of specificity for MHC-bound peptide limits variation in the off-rate but does not restrict changes in on-rate. In accord with these findings, quasi-thermodynamic parameters of the association and dissociation phases have shown that the gain in the free energy of 25-D1.16 binding is determined by a lower energy barrier of the association phase, whereas energy changes of the dissociation phase are similar for both the TCR and the antibody (Table II and Fig. 4).
Kranz and colleagues (11) produced soluble TCR encoded by mutated genes that bound cognate pMHC ligand with an intrinsic affinity of 3.6 -6.0 ϫ 10 7 M Ϫ1 . The increase in the TCR affinity relative to natural TCR was caused mainly by an increase in the on-rate not the off-rate, suggesting again that limited activation energy of the dissociation phase of the TCR-pMHC reaction is required to preserve peptide specificity. Although this TCR with enhanced affinity clearly discriminates cognate pMHC on the cell surface, T cell hybridomas carrying this TCR respond not only to the cognate pMHC on target cells but also to target cells of the same haplotype that do not display the cognate pMHC complexes. This response is apparently mediated by a prohibitively high level of energy contribution of the MHC moiety, causing an autoimmune reaction. Thus, energy input that is limited by the MHC-TCR contact in recognition of syngeneic pMHC ligands is an important condition by which the MHC restriction and the peptide specificity are met and the reactivity against self-pMHC is avoided. The same group has also investigated the energy map of the interaction between the same TCR and its allogeneic pMHC ligand (37). Site-directed mutagenesis of the TCR showed that CDR1 and CDR2 residues contacting the MHC moiety collectively contribute more binding energy than the CDR3 residues that are responsible for contact with the peptide. This indicates that it is the MHC energy contribution of the allogeneic ligand that dominates, as opposed to that of the syngeneic pMHC. Other analyses (38) of the binding of some recombinant TCRs with enhanced affinity to a syngeneic or to an allogeneic pMHC revealed similar contributions of the MHC and the peptide moieties. However, in this particular system the recognition of allogeneic ligand is strongly peptide-dependent (39,40). In addition, the peptide induces conformational changes in the MHC protein that are detectable by antibodies (41,42). In these circumstances, the peptide may contribute significantly to the amount of binding energy delivered indirectly by the allogeneic MHC through conformational changes of the MHC moiety. This may also explain why the affinity of the TCR recognizing this allogeneic pMHC could be increased by significant changes in the off-rate of the interaction (43). Not surprisingly, T cells expressing this high affinity TCR did not respond to target cells bearing irrelevant pMHC complexes (43). In another example (44), the equilibrium binding constant for a TCR interaction with class II invariant chain-associated peptide (CLIP) bound to I-A b MHC class II protein was measured at 1.6 ϫ 10 7 M Ϫ1 . 2 Because CLIP is thought to cause conformational changes of MHC class II molecules (45), this TCR may recognize a unique conformation of the MHC bound to CLIP, and a substantial energy contribution from the MHC moiety may account for high TCR affinity in this case.
The increase of the k a value reflects more precise docking of interacting proteins and a higher probability of productive 2 L. Teyton and A. Y. Rudensky, personal communication. a The ⌬⌬G a values were determined from the difference between ⌬G a n (25-D1. 16) and ⌬G a n (OT-1), and values of ⌬⌬G d were calculated from the difference between ⌬G d n (25-D1. 16) and ⌬⌬G d n (OT-1). See "Experimental Procedures" for other details.
b Structure and biological activity of peptides are given in the footnotes of Table I. FIG. 4. Free energy changes in the interactions of the pOV8⅐K b complex with the 25-D1.16 mAb and OT-1 TCR. Activation free energy of the association phase (⌬G a n ) of the antibody binding is significantly higher than that of TCR binding, although activation free energy of the dissociation phase (⌬G d n ) is similar for both interactions.
protein-protein encounters. Usually, k a depends on long range electrostatic interactions (25,46). These interactions generate an "energy funnel" between the two molecules, facilitating proper orientation that increases the probability of specific complex formation (47). Higher stability and lower flexibility of the interacting surfaces (48) are thought to facilitate long range electrostatic interactions (49). The latter may not necessarily become a short term interaction contributing to the stability of the complex, measured by the reaction off-rate. Most likely, enhanced long range interactions lead to an increase of the on-rate of 25-D1.16 antibody binding to pOV8⅐K b but do not change the balance of energy contributions made by the peptide and the MHC, ensuring the peptide specificity of the antibody.
Analysis of the primary structure shows that the 25-D1.16 V genes encoding V H and V L domains have very few mutations, all of which are in the framework regions, whereas the CDRs appear to be in germ line configuration. This suggests that the V H and V L genes of 25-D1. 16 have not completed affinity maturation. It has been suggested (50) that the affinity maturation of antibodies is driven initially by an increased on-rate and that this process may be mediated by mutations in framework regions. Subsequent somatic mutations in CDRs lead to a lower off-rate of antibody binding and a further increase of the binding affinity (50). Davis and colleagues (51) found that further maturation of an antibody specific for moth cytochrome c peptide in association with I-E k MHC class II led to a higher affinity antibody that could no longer discriminate bound I-E k moth cytochrome c peptide from other self-I-E k complexes on the surface of live cells. Thus, it appears that the limited affinity of pMHC-specific antibodies and the absence of extensive somatic mutations in the CDRs of these antibodies may be linked and are similar to those properties of TCR in which genes are not a subject for somatic mutations. Normally, a B cell response to protein antigens is to produce IgG antibody with high affinity through extensive somatic hypermutation and clonal selection, which are manifested in the germinal center. The antibody studied here does not appear to fall into this conventional category, providing a possible explanation for the rare occurrence of pMHC-specific antibodies elicited in vivo.