Antagonism of Antiviral and Allogeneic Activity of a Human Public CTL Clonotype by a Single Altered Peptide Ligand: Implications for Allograft Rejection 1

Alloreactive T lymphocytes are central mediators of graft-versus-host disease and allograft rejection. A public CTL clonotype with specificity for the alloantigens HLA-B*4402 and B*4405 is often expanded to large numbers in healthy HLA-B*0801+ individuals, driven by cross-reactive stimulation with the common, persistent herpesvirus EBV. Since such alloreactive memory CTL expansions have the potential to influence transplantation outcome, altered peptide ligands (APLs) of the target HLA-B*0801-binding EBV peptide, FLRGRAYGL, were screened as specific antagonists for this immunodominant clonotype. One APL, FLRGRFYGL, exerted powerful antagonism of a prototypic T cell clone expressing this immunodominant TCR when costimulated with target cells presenting HLA-B*0801FLRGRAYGL. Significantly, this APL also reduced the lysis of allogeneic target cells expressing HLA-B*4402 by up to 99%. The affinities of the agonist and antagonist complexes for the public TCR, measured using solution and solid-phase assays, were 8 and 138 μM, respectively. Surprisingly, the half-life of the agonist and antagonist complexes was similar, yet the association rate for the antagonist complex was significantly slower. These observations were further supported by structural studies that suggested a large conformational hurdle was required to ligate the immunodominant TCR to the HLA-B*0801 antagonist complex. By defining an antagonist APL against an immunodominant alloreactive TCR, these findings raise the prospect of exploiting such peptides to inhibit clinical alloreactivity, particularly against clonal T cell expansions that react with alloantigens.

T cell recognition of allogeneic HLA molecules is a major barrier to successful transplantation. Although the repertoire of T cells available for use in an alloresponse against a single allo-HLA molecule is diverse, the actual repertoire used may be highly selected (1)(2)(3)(4)(5). The basis for this limited diversity is unclear; however, it is possible that pre-existing expansions of alloreactive T cells, which have been demonstrated in healthy individuals, could be primarily involved (6). Recent studies have characterized alloreactive T cell expansions in healthy individuals that are driven by cross-reactive stimulation with EBV, a human pathogen that persistently infects ϳ90% of adults (7,8). For example, in unrelated HLA-B8 ϩ individuals, a highly dominant CTL response is generated against the nonameric EBV peptide, FLRGRAYGL, which is also alloreactive against distinct members of the HLA-B44 family. The public TCR expressed by this cross-reactive clonotype, termed LC13, is alloreactive against HLA-B*4402 and HLA-B*4405 but not HLA-B*4403. This is somewhat surprising given that there is only one amino acid difference between these alleles. These expanded alloreactive populations are often so large that such T cells dominated conventional MLRs from some HLA-mismatched individuals. Thus, a prior history of infection with an immunogenic virus such as EBV can influence an individual's level of responsiveness to an alloantigen and such mechanisms may underlie the observed clinical association between herpesvirus exposure and graft-versus-host disease (GVHD) 3 (9).
The limited use of the TCR repertoire in GVHD and allograft rejection (1)(2)(3)(4)(5) may provide the opportunity to therapeutically disrupt the alloresponse by targeting a selected T cell population for inactivation, as has been achieved in experimental animal models (10). Altered peptide ligands (APLs) represent one such strategy developed to modulate the immune response, whereby subtle alterations of the cognate peptide can potentially convert an agonist to a superagonist, weak agonist, or antagonist (for review, see Ref. 11). Despite the dramatically different biological outcomes an APL can elicit, only slight conformational readjustments are observed at *Department of Biochemistry and Molecular Biology, Protein Crystallography Unit, School of Biomedical Sciences, Monash University, Clayton, Australia; † Cellular Immunology Laboratory, Queensland Institute of Medical Research and Department of Molecular and Cellular Pathology, University of Queensland, Brisbane, Australia; and ‡ Department of Microbiology and Immunology, University of Melbourne, Parkville, Australia the TCR/MHC interface (12,13). With some exceptions, biological outcome is best correlated to the half-life of the TCR/MHC complexes, such that antagonists display a shorter half-life than their agonist counterparts (14,15).
We have recently established the structural basis for T cell immunodominance in the CTL response to the FLRGRAYGL determinant (16). We determined the crystal structure of the prototypical immunodominant LC13 TCR in its nonliganded state (17) and in complex with HLA-B8 FLRGRAYGL (16,18). The immunodominant TCR was observed to make a number of specificity-driven contacts with HLA-B8 FLRGRAYGL , including P7-Tyr of the peptide, a pivotal residue that sat centrally within the pocket of LC13. This structure provides an important starting point from which to understand the biochemical and structural basis of APL-induced antagonism.
In the present report, we describe the inactivation via APL antagonism of the immunodominant LC13 T cell clone. The measured affinity for the LC13 TCR/HLA-B8-agonist complex, at 8 M, compares to a value of 138 M for the LC13 TCR/HLA-B8antagonist complex. However, surprisingly the half-lives of the respective complexes were similar. Instead, the association rate for the antagonist complex was much slower than that of the authentic agonist complex, suggesting a larger hurdle for conformational change is required upon binding to the antagonist complex, which was consistent with the observed structural data. The inhibitory activity of the antagonist peptide was not dependent on binding to the allogeneic target cell since pretreatment of the cross-reactive HLA-B*0801-expressing CTLs with the peptide was sufficient to block their effector function. This raises the prospect of exploiting selected antagonist peptides to inhibit host T cell reactivity toward allografts mediated by immunodominant TCRs.

Establishment and maintenance of cell lines
Lymphoblastoid cell lines (LCLs) and PHA blasts were generated as described previously (7). LCLs were generated by exogenous transformation of peripheral B cells with EBV derived from the QIMR-Wil cell line. The LC13 CTL clone used in this study has been described previously (19,20).

Cytotoxicity assay
Peptides were tested in duplicate for inhibition of cytotoxicity in the standard 5-h chromium release assay using an effector:hot target ratio of 2:1 (21). In most cases (but not the experiments presented in Fig. 2, E and F, see figure legend), 51 Cr-labeled targets, CTL effectors, and the peptide were plated out for the assay at the same time, and the peptide remained present throughout the assay. Peptides were synthesized by Mimotopes Pty. on a 1-mg scale.

Expression, purification, crystallization, and structure determination
The LC13 TCR was expressed, refolded, and purified as previously described (22). The HLA-B8 antagonist complex was expressed, purified essentially as previously described (22), except that the EBV epitope was replaced by the antagonist peptide FLRGRFYGL. Briefly, the truncated forms of the HLA-B8 H chain (hc) and full-length ␤ 2 -microglobulin (␤2m) were expressed in Escherichia coli and each protein was purified from inclusion bodies. The complex of hc/␤2m/peptide is refolded by diluting the hc and ␤2m inclusion body preparations into refolding buffer containing a molar excess of peptide ligand. The refolded complexes were concentrated and purified by anion exchange chromatography. The complexes were further purified by gel filtration chromatography to a high level of purity before crystallization.
Crystals of the HLA-B8-antagonist complex were obtained using the protocol previously described (22), using a reservoir buffer of 10 -15% PEG 4000, 0.1 M sodium citrate (pH 5.6), 0.2 M ammonium acetate, and 10 mM cadmium chloride. The crystals belong to space group P2 1 2 1 2 1 with unit cell dimensions a ϭ 85.41, b ϭ 90.10, and c ϭ 125.28 Å. A 2.6 Å data set was collected using inverse geometry and was processed and scaled using the HKL suite of programs (23). For a summary of statistics, see Table I.
The structure was refined using the dimeric HLA-B8 FLRGRAYGL complex (18) minus the peptide and the water molecules as the start model. The progress of refinement was monitored by the R free value (4% of the data), with neither a nor a low resolution cutoff being applied to the data. The structure was refined using rigid-body fitting of the individual domains followed by the simulated-annealing protocol implemented in CNS (version 1.0) (24) using a methodology previously used (18). The electron for all data except for 3% which was used for the c R free calculation. density for the bound antagonist peptide was very clear in both monomers. See Table I for summary of refinement statistics and model quality.

Tryptophan fluorescence measurements
Tryptophan fluorescence binding experiments were conducted at 25°C in a PerkinElmer LS50B spectrofluorometer using a stirred quartz cuvette with a 1-cm path length. Samples were excited at 295 nm and the change in the emission intensity at 340 nm was recorded. Increasing concentrations of LC13 were added to 10 mM Tris-HCl buffer (pH 8.0) in the absence and presence of 1 M HLA-B8 FLRGRAYGL . The change in fluorescence intensity was calculated by subtracting the response of a given LC13 concentration from the LC13/HLA-B8 FLRGRAYGL response. The mean of five independent experiments was corrected for the OD and the data were fitted using GraphPad Prism version 3.0 (GraphPad Software).

Surface plasmon resonance analysis
All surface plasmon resonance (SPR) experiments were conducted at 25 o C on a Biacore 3000 instrument using HBS buffer (10 mM HEPES-HCl (pH 7.4), 150 mM NaCl, and 0.005% surfactant P20 supplied by the manufacturer). The Ab W6/32 (25) was coupled in 10 mM citric acid (pH 5.0) to research grade CM5 chips using standard amine coupling at a level of 9000 -11000 resonance units (RU). For each experiment, 200 -400 RU of the HLA-B8-peptide was immobilized on the Ab. LC13 samples were injected over all flow cells at 60 l/min for 1 min for HLA-B8 FLRGRAYGL and at 15 l/min for 40 s for HLA-B8 FLRGRFYGL . The final response was calculated by subtracting the response of the Ab alone from the Ab-HLA-B8 complex. The Ab surface was regenerated between each analyte injection using 100 mM glycine-HCl (pH 2.8). BIAevaluation version 3.1 (Biacore AB) was used to fit the data to the 1:1 Langmuir binding model, allowing for local fitting of the binding maximum, to calculate the kinetic constants. The equilibrium data were analyzed using GraphPad Prism.

APL identification
To identify APL antagonists for this immunodominant CTL clonotype that cross-recognizes the HLA-B44 alloantigen and the HLA-B8-binding EBV epitope FLRGRAYGL, 171 monosubstituted peptide analogues of the FLRGRAYGL peptide were synthesized such that each residue was sequentially replaced with all other genetically coded amino acids. The CTL LC13, a representative clone expressing this well-characterized, cross-reactive TCR, was used as an effector in a standard 51 Cr release assay against target cells expressing syngeneic HLA-B*0801 molecules and presenting endogenously processed FLRGRAYGL. CTL lysis assays were conducted in the presence of each peptide analogue at two different concentrations (5 and 0.5 mol/L). Lysis of LCL targets expressing endogenous FLRGRAYGL in the absence of APLs was 31.2%. Data are presented in Fig. 1 as percent inhibition of lysis, relative to the lysis observed without exogenous peptide addition. One peptide analogue, FLRGRFYGL, reduced lysis of the LCLs by up to 50% in this screening assay, and this peptide was chosen for more detailed analysis.

The APL FLRGRFYGL is a potent antagonist of recognition of both cognate and allogeneic targets
The FLRGRFYGL peptide was first tested as an agonist for the LC13 CTL clone on HLA-B*0801 PHA blasts and was found to exhibit no agonist activity at peptide concentrations up to 1 mM ( Fig. 2A). This peptide was then tested as an antagonist for the LC13 CTL clone over a wide range of concentrations. Lysis of the HLA-B*0801 ϩ LCL from donor RM was inhibited very efficiently (up to 69%) when the peptide was added to target cells at concentrations above 16 mol/L (Fig. 2B). LCLs and PHA blasts coexpressing HLA-B*0801 and the allogeneic target Ag HLA-B*4402 were also tested for lysis by the LC13 CTL clone in the presence of this antagonist peptide. As shown in Fig. 2C, strong inhibition of cytotoxicity was observed when peptide FLRGRFYGL was present during the chromium release assay. As a specificity control for the experimental data shown in Fig. 2, A-C, a CTL clone that also recognizes the FLRGRAYGL epitope but that expresses a different (subdominant) TCR was also included in the assays (clone CF34, see Ref. 19). In contrast to the data shown for CTL clone LC13, addition of the FLRGRFYGL peptide had no significant effect on the lysis of the HLA-B*0801 ϩ target cells by this subdominant T cell clonotype (our unpublished data). The FLRGRFYGL antagonist peptide also failed to inhibit lysis by CTL clone LC13 of HLA-B8 ϩ LCLs that had been pretreated with high concentrations (Ͼ1 g/ml) of the FLRGRAYGL agonist peptide (our unpublished data). Interestingly, this peptide could also reduce lysis mediated exclusively toward the allogeneic target Ags, HLA-B*4402 or B*4405. Thus, LCLs and PHA blast lines expressing HLA-B*4402 or B*4405, but negative for HLA-B*0801, were normally lysed efficiently by CTL clone LC13, but when peptide FLRGRFYGL was added to this culture, this lysis was reduced by up to 72% (Fig. 2D).
The FLRGRFYGL antagonist peptide is structurally unsuited to binding HLA-B*4402 or B*4405 since it contains an HLA-B8 peptide-binding motif ( (26,27). Therefore, we reasoned that the APL specifically inhibited the anti-HLA-B44 alloreactivity through APL-mediated antagonism resulting from T-T presentation by the HLA-B8 ϩ effector cells. To investigate this possibility, the LC13 CTLs were pretreated with the antagonist peptide for 60 min and washed before use in the cytotoxicity assay. This pretreatment protocol resulted in greatly reduced alloreactive cytotoxicity against the B*4402 ϩ and B*4405 ϩ targets (Fig. 2E). Lysis of four different HLA-B*4402 ϩ B*0801 Ϫ PHA blast lines was reduced from a maximum of 31-38% using untreated CTLs down to 0.2-5% (87-99% inhibition) with CTL effectors that were pretreated with 100 mol/L of the antagonist peptide (Fig. 2E). These data indicate that the LC13 clone was interacting with the HLA-B*0801-binding FLRGRFYGL antagonist peptide presented on surrounding CTLs and not the allogeneic B*4402 target Ag expressed by the PHA blasts. To further examine the outcome of LC13 interaction with the antagonist peptide, we studied the phosphorylation of the T cell protein tyrosine kinase, Zap70, by agonist and antagonist ligands. Previous studies have shown that "classical antagonists" induce only partial phosphorylation of Zap70 (28,29), a kinase that is known to be critical for T cell signaling. At concentrations of FLRGRFYGL that antagonize recognition of HLA-B8 FLRGRAYGL , there was no detectable phosphorylation of LC13 Zap70 above background (Fig. 3). In contrast, the agonist HLA-B8 FLRGRAYGL ligand induced significant Zap70 phosphorylation under the same conditions. The findings provide evidence that antagonism by the FLRGRFYGL peptide is associated with unproductive TCR interaction and inadequate phosphorylation of Zap70.
To examine whether the alloreactivity of this clone could also be inhibited by other primary cells expressing HLA-B*0801 ϩ and presenting the antagonist peptide, cold-target inhibition assays were performed. Unlabeled HLA-B*0801 ϩ PBMCs were pretreated with the antagonist peptide (100 or 20 mol/L) or left untreated and tested for their ability to inhibit lysis of chromiumlabeled PHA blasts expressing HLA-B*4402. As shown in Fig. 2F, at a cold:hot target ratio of just 2:1 significant inhibition of anti-B*4402 alloreactivity was observed (between 54 and 85% inhibition).

APL activity of FLRGRFYGL does not correlate with the halflife of the TCR/MHC-peptide complex
Having characterized the antagonism of the LC13 TCR, we then compared the affinities and the kinetic constants for the agonist and antagonist complexes using initially SPR experiments. The HLA-B8 FLRGRAYGL complex was captured onto the chip using the MHC class I-specific mAb W6/32 (25) and increasing concentrations of the LC13 TCR were passed over the surface (Fig. 4). We have previously determined that W6/32 does not interfere with the LC13 TCR interaction (data not shown). As depicted in Fig. 4A, a concentration-dependent increase in binding response was observed. The affinity constant was calculated to be 8.2 M by fitting the binding response at equilibrium to a 1:1 binding equation (Fig.  4B). The Scatchard analysis also confirms the 1:1 binding relationship and calculates the dissociation equilibrium constant to be 8.8 M. The association and dissociation rate constants were calculated to be 3.58 ϫ 10 4 M Ϫ1 s Ϫ1 and 0.42 s Ϫ1 , respectively, and the calculated dissociation equilibrium constant was 12.5 M. These values are in close agreement with those reported recently (30). The affinity of the LC13/HLA-B8 FLRGRFYGL interaction was also examined by SPR. The dissociation equilibrium constant was determined to be significantly weaker at 138 M. Because the interaction between LC13 and HLA-B8 presenting the antagonist ligand is much weaker, the flow rate and surface density was op-timized to measure the kinetic constants (Fig. 4, C and D). The LC13/ HLA-B8 FLRGRAYGL interaction was used as a control and under the optimized flow rate conditions the affinity and kinetic constants of the LC13/HLA-B8 FLRGRAYGL interaction were not affected (data not shown). The antagonist kinetic rate constants were calculated to be 2.62 ϫ 10 3 M Ϫ1 s Ϫ1 and 0.35 s Ϫ1 for the association and dissociation, respectively. In comparison to the authentic agonist epitope, the antagonist ligand had a significantly slower on rate (Table II); however, these ligands displayed similar dissociation kinetics with half-lives of 1.65 and 1.98 s, respectively.
The binding affinity of HLA-B8 FLRGRAYGL for the LC13 TCR was also verified in solution by measuring the change in Trp fluorescence upon complex formation (Fig. 5A). The use of Trp fluorescence to measure conformational change and binding affinities is a well-established technique (31). The majority of the Trp residues in HLA-B8 lie remote from the TCR contact site; however, Trp 147 and Trp 167 reside within the Ag-binding groove. Moreover, having determined the crystal structure of the LC13/HLA-B8 FLRGRAYGL complex, we were able to ascertain that Trp 147 is only 3.5 Å away from the incoming CDR3␤ loop of LC13, and we reasoned that this Trp may act as a reporter for TCR binding. The concentration of HLA-B8 FLRGRAYGL was kept constant in the presence of increasing concentrations of LC13. These data were fitted to a 1:1 binding model and the affinity constant was calculated to be 5.5 M (Fig. 5B). The Scatchard analysis of this data calculated the dissociation equilibrium constant to be 5.4 M (Ϫ1/ slope) with a linear relationship supporting the 1:1 binding model  ( Fig. 5C). This affinity compares favorably to the 8.8 M calculated in solution using SPR, indicating that capturing the HLA-B8 molecule to the chip does not compromise the affinity of the TCRbinding interaction.

APL activity of FLRGRFYGL is associated with minor structural alteration of the HLA-B8-peptide complex
Next, we aimed to elucidate the structural basis of the observed antagonism. The crystal structure of the HLA-B8-antagonist complex was very similar to that of the HLA-B8 FLRGRAYGL complex that was determined previously (18). Accordingly, the structure description will be limited to the impact the antagonist peptide has on the Ag-binding groove. The electron density for the bound antagonist peptide was unambiguous. The site of substitution (P6) represents a surface-exposed position in the FLRGRAYGL peptide that is located within the central bulge of the peptide, adjacent to the P7-Tyr residue essential for T cell activation. Accordingly, the Ala to Phe substitution, albeit a nonconservative one, neither impacts on the conformation of the bound peptide nor on the HLA-B8 conformation (18). The P6-Phe adopts a conformation such that it folds back toward the N terminus of the peptide, making van der Waals (vdw) contacts with the peptide backbone of Gly 4 (Fig. 6A). Accordingly, the conformation of the P7-Tyr is not affected by the conformation adopted by Phe 6 . The conformation of the P6-Phe side chain, despite being fully solvent exposed, does not protrude prominently from the Ag-binding groove, as distinct from the prominent P-7 Tyr.
Having recently determined the structure of the LC13/HLA-B8agonist complex (16), we were able to address, in part, how this antagonist peptide impacts on the mode of LC13 binding. A conserved diagonal footprint on the agonist (Fig. 6B) and antagonist complexes (Fig. 6C) was considered most probable. Accordingly, the LC13/HLA-B8 FLRGRAYGL complex was superposed onto the HLA-B8 FLRGRFYGL complex. Visual inspection of the docked LC13/HLA-B8 FLRGRFYGL complex revealed that the Phe 6 side chain of the antagonist did not sterically clash with any residues from the CDR loops of LC13. Thus, it is viewed most likely that the conformation of the CDR loops in the docked TCR/HLA-B8 FLRGRFYGL complex will be very similar to the observed conformation of the CDR loops in the LC13/HLA-B8 FLRGRAYGL crystal structure. Definitive experimental proof of this hypothesis will require the crystal structure determination of the LC13/HLA-B8 FLRGRFYGL however. As can be seen from this docked model (Fig. 6C), the bulky Phe 6 side chain nestles within a groove formed by the CDR3␣ loop, making vdw contacts with Leu 94 and Gly 97 of the TCR, whereas in the crystal structure of the agonist complex, this pocket is occupied by a water molecule.

Discussion
Previous studies have shown that APLs of immunogenic peptides may profoundly reduce the magnitude of the response to the wild-type epitope (32)(33)(34)(35)(36)(37)(38)(39). In this report, we have identified and characterized an antagonist APL for an immunodominant antiviral and alloreactive human TCR. The CTL response to EBV in unrelated HLA-B8 ϩ individuals is dominated by reactivity toward the RAKFKQLL peptide from the lytic Ag BZLF1 and the FLRGRAYGL peptide from the latent EBNA-3A Ag. Moreover the CTL response to this latter determinant is characterized by highly restricted TCR usage, such that in unrelated EBV ϩ HLA-B8 ϩ individuals, the ␣␤ TCR sequence is virtually identical (20). Fine epitope mapping of the FLRGRAYGL epitope revealed that the prototypic immunodominant TCR, termed  LC13, was exquisitely sensitive to substitutions at the P7-Tyr position, as well as the small flanking residues of P6-Ala and P8-Gly (19). We recently provided a structural basis for this immunodominant LC13 TCR/HLA-B8 FLRGRAYGL interaction, where the P7-Tyr was observed to sit centrally within the TCR pocket, while the small P6-Ala and P8-Gly side chains enabled the tyrosine to protrude deeply into the pocket (16). Since this antiviral CTL response is cross-reactive with several alloantigens and could therefore influence allograft rejection and GVHD, it was of interest to ascertain whether antagonist APLs could be identified. From an exhaustive screen, whereby each amino acid of the nonamer was substituted with the remaining 19 aa acids, one candidate antagonist was found. This antagonist peptide, FLRGRFYGL, was shown to be very effective in reducing lysis, as judged by the Cr release assay. The FLRGRFYGL peptide displayed no agonist activity and moreover antagonist activity that could not be explained by simple competition for binding to HLA-B*0801. Thus, prepulsing the LC13 CTL clone with peptide, followed by washing away the free APL, led to inhibition of LC13 responses toward both the cognate HLA-B*0801 FLRGRAYGL agonist as well as the alloantigens HLA-B*4402 and HLA-B*4405. This observation is consistent with the current view that CTL antagonism is through competition by antagonist MHC peptide complexes for TCR binding at the expense of agonist MHC peptides. Importantly, antagonist activity was also observed when the FLRGRFYGL peptide was present throughout the culture with agonist-bearing APC, a desirable property of any potential therapeutic antagonist.
To further characterize the molecular basis for the antagonism of this public CTL clonotype, we then determined the respective affinities and kinetic constants of the immunodominant TCR for the HLA-B8 FLRGRAYGL and HLA-B8 FLRGRFYGL complexes. Using SPR, the measured affinity for the agonist complex, K d 8 M, was in good agreement with the solution-phase experiments, where a K d 5.5 M was obtained using a Trp fluorescence reporter assay. This suggests that Trp fluorescence measurements may provide an alternative for the cell-based assays, ligand-labeling studies, analytical ultracentrifugation, or calorimetry techniques (reviewed in Ref. 40) that have previously provided the main alternative to SPR measurements in defining the affinity of a TCR/MHC interaction. A K d of 8 M falls within the range of the measured affinities of previous TCR-class I interactions; however, by comparison the association and dissociation rates for the agonist complex were faster.
In other studies the functional outcome of TCR ligation with APLs has correlated best with TCR affinities, most notably the dissociation rate with MHC-APL complexes. However, a number of exceptions to this correlate have been noted (14, 40 -42). As determined using Biacore, the affinity for the antagonist complex (138 M) is 17-fold lower than the agonist (8 M). Surprisingly however, the association rate of the antagonist complex is significantly slower (2.6 vs 35.8 ϫ 10 3 M Ϫ1 s Ϫ1 ), yet the half-life is actually marginally slower than the agonist complex (1.98 vs 1.65 s). Accordingly, factors other than binding kinetics govern the biological outcome from engagement of this immunodominant TCR.
The antagonist activity induced by the Ala-Phe substitution at P6 was surprising, given our earlier observation that LC13 is highly sensitive to substitutions at the P6 position within the FL-RGRAYGL determinant, thereby suggesting the need for a small side chain at P6 to permit TCR ligation (16). Therefore, we sought the structural basis for this observed phenomenon. Previously APLs against HLA-B8-HIV epitopes were observed to cause small but significant C␣ backbone shifts of the helices of the Ag-binding cleft (43). In addition, the crystal structures and models of a number of TCR-ligated APL complexes have suggested that only subtle conformational changes are required within the hot spot of the TCR/MHC-peptide interface to accommodate the APLs (12, 13, 44). The P6-Phe residue is solvent exposed and was observed not to impact on the conformation of the peptide nor the conformation of the HLA-B8 when compared with the binary agonist complex. As yet, we have not been able to obtain crystal of the LC13/HLA-B8 FLRGRFYGL complex, presumably due to the low affinity of the interaction. Nevertheless, molecular modeling suggests that the bulky Phe 6 group is accommodated readily in the docked LC13/ HLA-B8-antagonist complex. This suggests that the conformation of the CDR loops in the docked antagonist complex will be virtually identical to the conformation of the CDR loops observed in the crystal structure of the LC13/HLA-B8-agonist complex (16). How do these structural observations relate to the observed slower on rate for the antagonist complex in comparison to the agonist counterpart? Upon LC13 binding to HLA-B8 FLRGRAYGL , large movements in a number of the CDR loops were observed (16). The small side chain of P6-Ala reduced the steric hindrance of the CDR3␣ conformational change, where this loop was observed to swing away from the FLRGRAYGL peptide to enhance the interactions with HLA-B8. As stated, with respect to the agonist complex, the Phe 6 substitution is perceived not to significantly alter the final structure of the TCR within the antagonist complex. However the Phe 6 group is likely to affect the conformational dynamics of the CDR3␣ loop in that it may provide a greater hurdle to CDR3␣ conformational change upon complexation, and thereby provide a structural basis for the experimentally determined slower association rate. Our data are consistent with the notion that the properties of antagonism relate to the energetic hurdle of conformational change that takes place upon complexation and it will be of further interest to ascertain whether this correlates with changes in heat capacity (⌬Cp) (41). In addition, the favorable vdw interactions between the Phe 6 and the TCR also suggests a basis for the surprisingly longer half-life of the complex.
The identification of an APL antagonist for this immunodominant TCR is of particular interest because this Ag receptor is alloreactive with HLA-B*4402 and HLA-B*4405, and such alloreactivity could exacerbate T cell-mediated transplant rejection (7,8). Furthermore, this public TCR is expressed by an expanded CTL population found in most EBV-exposed, HLA-B8 ϩ individuals. Of particular interest was that this antagonist peptide could effectively inhibit this alloreactivity. Recently, an APL that antagonizes syngeneic and allogeneic agonists in the mouse 2C system has been described (45). The present report reinforces the proposal that TCR antagonist peptides could find application as specific therapeutics for GVHD and allograft rejection in humans. The biggest obstacle to this objective is likely to be the potential diversity of the TCR repertoire in response to alloantigens. However, if pathogen-driven clonal expansions of alloreactive T cells are involved in initiating some cases of allograft rejection and GVHD, it is possible that antagonist peptides such as that described herein could be exploited in transplantation to inhibit harmful T cell clonotypes. This might be especially relevant where TCR immunodominance creates a homogeneous target for APL-mediated antagonism.