The Energetic Basis Underpinning T-cell Receptor Recognition of a Super-bulged Peptide Bound to a Major Histocompatibility Complex Class I Molecule

Background: T-cell receptor (TCR) recognition of a lengthy peptide-human leukocyte antigen (HLA) complex indicates a peptide-centric mode of interaction.

Results: The energetic landscape of this interaction is centered on the peptide and peptide-contacting residues.

Conclusion: Lengthy peptides drive a more peptide-focused mode of TCR recognition.

Significance: This is the first description of the energetic landscape underpinning TCR recognition of lengthy HLA-restricted peptides.

Abstract

Although the major histocompatibility complex class I (MHC-I) molecules typically bind short peptide (p) fragments (8–10 amino acids in length), longer, “bulged” peptides are often be presented by MHC-I. Such bulged pMHC-I complexes represent challenges for T-cell receptor (TCR) ligation, although the general principles underscoring the interaction between TCRs and bulged pMHC-I complexes are unclear. To address this, we have explored the energetic basis of how an immunodominant TCR (termed SB27) binds to a 13-amino acid viral peptide (LPEPLPQGQLTAY) complexed to human leukocyte antigen (HLA) B*3508. Using the crystal structure of the SB27 TCR-HLA B*3508^LPEP complex as a guide, we undertook a comprehensive alanine-scanning mutagenesis approach at the TCR-pMHC-I interface and examined the effect of the mutations by biophysical (affinity measurements) and cellular approaches (tetramer staining). Although the structural footprint on HLA B*3508 was small, the energetic footprint was even smaller in that only two HLA B*3508 residues were critical for the TCR interaction. Instead, the energetic basis of this TCR-pMHC-I interaction was attributed to peptide-mediated interactions in which the complementarity determining region 3α and germline-encoded complementarity determining region 1β loops of the SB27 TCR played the principal role. Our findings highlight the peptide-centricity of TCR ligation toward a bulged pMHC-I complex.

Introduction

The specificity of cytotoxic T-cells is characterized by the surface expression of the αβ T-cell receptor (TCR),⁶ which recognizes peptide fragments (typically between 8 and 10 amino acids in length) presented by class I major histocompatability complex (MHC-I) molecules on target cells. The antigen-recognition site of the TCR is comprised of six complementarity-determining regions (CDRs), three each from the TCR α and β chains, of which the CDR1 and -2 loops are germline encoded, whereas the CDR3 loops can be comprised of germline-encoded and non-germline-encoded regions. The rearrangement of TCR gene elements yields a potential 10¹⁵ different TCRs in humans, which after undergoing positive and negative selection represent a mature T-cell repertoire of ∼10⁸ different TCRs (1). Typically, the antigenic peptide is tethered within the antigen (Ag)-binding cleft of the MHC molecule, whereupon a few of the surface-exposed amino acids represent potential TCR contact points (2). Accordingly, relative to the short peptide, the polymorphic MHC-I heavy chain dominates the interactions with the TCR (3). Successful peptide-MHC (pMHC) engagement by the TCR is an obligate requirement for thymic selection and for the recognition of foreign Ags in the periphery, which subsequently leads to a series of effector functions. Accordingly, understanding the factors that dictate the nature of the TCR-pMHC engagement is an area of intense interest. However, generalizations regarding the nature of this interaction are compounded by the extensive human leukocyte antigen (HLA) polymorphism and the diverse T-cell repertoire.

Structural studies have shown that the TCR can engage the pMHC-I surface in a wide range of docking modes (4). Nevertheless, a conserved docking mode between the TCR-pMHC-I is observed, in which the Vα and Vβ domains are positioned over the MHC α2- and α1-helices, respectively, suggesting some hidden basis in TCR-pMHC engagement that may be attributable to signaling outcome (5, 6). Whether this hidden logic is underpinned by germline-encoded TCR-MHC recognition remains an area of intense investigation (7–11). The insight gleaned from TCR-pMHC-I structural studies have been bolstered significantly by associated biophysical, thermodynamic, and mutational studies, which have shown that the relative energetic contributions from the CDR loops can vary between different TCR-pMHC systems (7, 12–19). To date, most of the structural/mutagenesis studies have focused on TCR recognition of short peptides (8–10 amino acids), whereas it is known that ∼5–10% of the repertoire of peptides bound to MHC-I can be longer than 10 amino acids in length (20, 21).

Indeed, there is a growing understanding of the role longer MHC-I-restricted peptides play in immunity, where these longer determinants can form targets for MHC-restricted cytotoxic T-cells in tumor immunity and viral immunity (8, 22–24). The size limitation of peptides binding to MHC-I is, in part, defined by the hydrogen bonding network at the N- and C-terminal ends of the Ag binding cleft. As such, longer peptides generally bulge centrally from the Ag-binding cleft. Some of these long bulging MHC-I peptides display substantial mobility, whereas others assume a more rigid conformation, and thus they may be considered to potentially represent challenges for TCR ligation. Indeed, biased (immunodominant) TCR usage has been found as a mechanism for the adaptive immune system to deal with such challenging MHC-restricted viral peptides (11, 25). For example, biased TCR usage was observed against a flexible 11-mer peptide (EPLPQGQLTAY) (26) and a rigid 13-mer peptide (LPEPLPQGQLTAY, termed “LPEP”) (27) restricted to HLA B*3501 and HLA B*3508, respectively. The flexible 11-mer determinant was flattened upon TCR ligation, enabling the TCR to make substantial contacts with HLA B*3501 (28). In contrast, in recognizing the rigid 13-mer peptide, the prototypical biased TCR (termed SB27) made extensive peptide contacts but limited contacts with HLA B*3508 (29). As such, the SB27 TCR made noticeably fewer contacts with the MHC in comparison to TCRs interacting with canonical length (8–10-mer) epitopes. Given this observation, we aimed to explore the energetic basis of the SB27 TCR-HLA B*3508^LPEP interaction. How important are the TCR-MHC contacts in comparison to the TCR-peptide interactions? Our data show that longer peptides drive peptide-focused TCR recognition.

EXPERIMENTAL PROCEDURES

Generation of SB27 TCR and HLA B*3508 Mutants

Collectively, 28 single alanine mutants and 2 glycine substitutions (if the Cβ atoms only were involved in the contact) were made on either the SB27 TCR or HLA B*3508. These mutants were generated via site-directed mutagenesis techniques (QuikChange; Stratagene) with DNA templates that were previously described (8).

Protein Expression and Purification

Individual chains of the SB27 TCR, HLA B*3508 and the β₂-microglobulin were expressed as inclusion bodies in BL21(DE3) Escherichia coli cells, followed by refolding procedures as previously described (29). Post-refolding samples were dialyzed against 10 mm Tris-HCl, pH 8, twice daily and further purified through anion exchange chromatography and size-exclusion chromatography. The purity and the molecular weight of the samples were assessed with SDS-PAGE.

Surface Plasmon Resonance

Using a BIAcore 3000 instrument, all surface plasmon resonance (SPR) experiments were measured at 25 °C with a buffer containing 10 mm HEPES, pH 7.4, 150 mm NaCl, 0.005% surfactant P20, and 1% bovine serum albumin to avoid nonspecific binding. With a research grade CM5 sensorchip, a conformational specific monoclonal antibody (12H8) (12) was amine-coupled onto the surface to capture properly folded wild type or mutated SB27 TCR. Typically, one of the four flow cells was left empty without coupling any TCR (the control), whereas the other three flow cells were used to capture SB27 TCR (and mutants) to ∼200–400 response units. Serial dilutions of HLA B*3508^LPEP (or variants thereof) ranging between 0.78 and 200 μm were prepared and injected into the flow cells individually and the antibody surface was regenerated with Actisep (Sterogene) between each injection. All experiments were performed in duplicates and the kinetic constants were determined by the BIAevalution program (version 3.1) using 1:1 Langmuir binding models with the addition of a drifting baseline parameter.

Thermal Stability Assay

Thermal stability assays of HLA B*3508^LPEP variants were performed with the Real Time Detection instrument (Corbett RotorGene 300). Briefly, all pHLA samples were prepared at two concentrations (5 and 10 μm) in 10 mm Tris-HCl, pH 8, and 150 mm NaCl buffer. The fluorescent dye Sypro orange were added to the sample to enable monitoring of the protein unfolding process. Samples were heated from 30 to 95 °C at the rate of 1 °C/min and the changes in fluorescent intensity were recorded accordingly with the excitation wavelength of 530 nm and the emission wavelength at 555 nm. The results are summarized in supplemental Table S1.

Crystallization of pHLA and Structure Determination

Crystals of selected HLA B*3508^LPEP variants were obtained by the hanging-drop, vapor-diffusion method at 20 °C. Protein samples at 5 mg/ml concentration were mixed with the reservoir solution (0.1 m sodium citrate, pH 5.6, 16% polyethylene glycol (PEG) 4000, and 0.2 m ammonium acetate) at a 1:1 ratio. Microseeding techniques were applied to allow the formation of crystals with optimal size. Crystals of pHLA variants were soaked with reservoir solution containing an increased percentage of PEG 4000 (30%) followed by flash freezing in liquid nitrogen. All data were collected at the Australian synchrotron, Clayton, with an ADSC-Quantum 210 CCD on MX1 and an ADSC-Quantum 315r CCD on MX2 detector (at 100 K). The datasets were processed with XDS software (30), scaled with XSCALE and the structures were solved by molecular replacement using PHASER (31) and the published HLA B*3508^LPEP structure (PDB code 1ZHK (27)) minus the peptide as a search model. Manual building of all pHLA variants were performed with COOT (32) and further refined using maximum-likelihood refinement with PHENIX (33) and the structural representation showed in all figures were generated using PyMOL (30). The final models have been validated and submitted using the Protein Data Base validation web site and the final refinement statistics and PDB accession codes are summarized in supplemental Table S2.

Retroviral Transduction into SKW3 Cells

The transduction of the SKW3 cell lines was carried out as previously described (8). In brief, 4 μg of pMIG plasmid containing SB27 or LC13 TCR (used as control) was mixed with two retroviral packaging vectors DNA (2 μg of each pPAM-E and pVSV-g) and FuGENE-6. The mixture was used to co-transfect a total number of 10⁶ 293T epithelial cells. The transfected 293T cells were cultured further for 5 days and all virus-containing supernatants were harvested and replaced twice daily to transduce SKW3 cells with the addition of 6 μg/ml of Polybrene. After 3–5 days of transduction, the surface expression level of TCRs on the SKW3 cells was monitored using FACS. TCR positive cells were enriched, and cloned, with a BD FACSAria cell sorter.

Fluorescence-activated Cell Sorting (FACS) and Tetramer Binding Assay

10⁵ SKW3-SB27 cells and the positive control SKW3-LC13 cells were stained with the anti-CD3 antibody (OKT3) on ice for 30 min or with peptide-loaded HLA class I-multimers (HLA B*3508^LPEP-PE tetramer) at 26 °C for 45 min. After washing twice with FACS buffer (phosphate-buffered saline containing 2% fetal calf serum and 0.02% azide), the cells were run through a FACS-calibur or FACS-Aria with Cell-QuestPro and analyzed with FlowJo. For cell sorting experiments, the desired number of cells was stained similarly and all steps were carried out aseptically in a Bio-hazardous hood and filter-sterile reagents. Tetramer binding was measured by the increase of the mean fluorescence (MF) of tetramer staining. The TCR expression level was normalized with tight gating on the expression of green fluorescent protein (GFP), which correlated well with TCR expression levels in the transfected cells.

RESULTS

Experimental Rationale

To define the underlying energetic basis of the interaction between SB27 TCR and HLA B*3508^LPEP, we undertook an alanine scanning approach on both the SB27 TCR and HLA B*3508^LPEP. We analyzed the effect of the mutations on the affinity of the interaction using surface plasmon resonance (SPR), and furthermore, the effect of the HLA B*3508 mutants were investigated via cell surface staining using tetrameric versions of the mutants. Given the structure of the SB27 TCR-HLA B*3508^LPEP complex had been solved (29) (Fig. 1), we were able to rationalize the SB27 TCR and HLA B*3508^LPEP residues selected for mutational analysis. Solvent-exposed SB27 TCR residues whose side chains interacted with HLA B*3508^LPEP were selected for substitution to alanine. Similarly, HLA B*3508 residues, as well as those from the viral peptide, were selected for substitution if their side chains made contact with the SB27 TCR. In total, the effect of 34 mutations was investigated (18 from the TCR, 10 from HLA B*3508, and 6 from the LPEP peptide), thereby permitting a comprehensive understanding of the energetic basis of this interaction.

FIGURE 1.

Overview of the SB27 TCR interaction with the HLA B*3508^LPEP. a, the SB27 TCR is represented in schematic with the α-chain in pink, the β-chain in blue, the CDR1, -2, and -3α are purple, green, and red, and the CDR1, -2, and -3β colored in yellow, brown, and orange, respectively. The HLA B*3508 is represented as a white schematic and the β₂-microglobulin in gray, with the LPEP peptide in blue stick. b, close-up of the SB27 CDR loops interacting with the bulged LPEP peptide (blue stick), with the CDR loops colored accordingly to panel a. c, peptide centric structural footprint of the SB27 TCR on the HLA B*3508^LPEP surface (white), the atoms of the HLA B*3508 and LPEP peptide involved in the interaction are colored by the CDR loops contacted accordingly to panel a.

SB27 TCR Mutations

The cytotoxic T-cell response to the HLA B*3508^LPEP epitope exhibited biased α-chain usage, with the SB27 TCR representing an archetypal clonotype of this T-cell repertoire (34) (encoded by the TRAV19*01, TRAJ34*01, TRBV6-1*01, and TRBJ2-7*01 genes (35)). The SB27 TCR docked orthogonally across the Ag-binding cleft to accommodate the bulged epitope (Fig. 1a), in which the Vα chain leaned toward the α2-helix, whereas the Vβ chain formed few contacts with HLA B*3508, and interacted more extensively with the peptide (29). In brief, the CDR1α, CDR2α, and CDR3α loops provided the main contacts with HLA B*3508 (Fig. 1, b and c), whereas the CDR1β loop played a prominent role in interacting with the peptide, thereby underscoring the differing roles underpinning the biased Vα and Vβ usage in the SB27 TCR.

In total, 18 (17 alanine and βS31G) mutants were generated within the α and β chains of SB27, which included βAsn⁶⁶ as a negative control (this residue was not involved in binding to HLA B*3508^LPEP). The residues that were mutated within the various CDR loops were: CDR1α, αThr²⁹ and αTyr³¹; CDR2α, αAsn⁵⁰ and αPhe⁵²; CDR3α, αSer⁹³, αPhe⁹⁵, αTyr⁹⁶, and αAsn⁹⁷; CDR1β, βAsn²⁸, βHis²⁹, βAsn³⁰, βSer³¹, βTyr³³; CDR2β, βSer⁵¹; CDR3β, βPro⁹⁵, βLeu⁹⁸, and βGlu¹⁰¹ and the framework residue βAsn⁶⁶. We mutated, expressed, refolded, and purified all mutants and, to ensure the structural integrity of these mutants, we compared their biochemical characteristics and reactivity toward a conformationally specific TCR mAb (PDB 12H8) (12). All mutants behaved similarly to the wild type SB27 TCR (data not shown), thereby ensuring the conformational integrity of the targeted mutants.

First, we measured the affinity and kinetic rate constants of the wild-type (WT) SB27 TCR-HLA B*3508^LPEP interaction. The SB27 TCR bound HLA B*3508^LPEP with a K_d,eq of 18.7 ± 0.4 μm, an on rate (K_on) of 6.56 × 10⁴ m s⁻¹; and a dissociation constant (K_off) of 0.12 s⁻¹, values that were consistent with previous measurements (29) (Fig. 2a, Table 1). Although these WT K_d,eq measurements were reproducible on two separate occasions and using two separate batches of WT SB27 TCR, all the SB27 TCR mutants (example shown in Fig. 2b) were, nevertheless, analyzed in the same experiment, in duplicate and relative to the same batch of WT SB27 TCR. This was to ensure that all the K_d,eq values were comparable and relative to one another (see Table 1). SB27 TCR mutations that caused less than a 3-fold loss in the affinity of the interaction with HLA B*3508^LPEP compared with WT SB27 TCR were considered to have no major effect; a 3–5-fold effect as moderate, and anything greater than 5-fold as critical to the energetics of the interaction.

FIGURE 2.

Surface plasmon resonance experiment of SB27 TCR variants binding to HLA B*3508^LPEP. Kinetic rate constant measurements of the SB27 TCR (a) and SB27-βAla⁹⁸ TCR (b), binding to a concentration series of HLA B*3508^LPEP complex; inset, equilibrium concentration versus response relationship. The experiments have been conducted in duplicate and the error bars are shown for each data point (mean ± S.E.).

TABLE 1.

Surface plasmon resonance analysis on SB27 TCR mutations

t_1/2 = 0.693/K_d,eq. ΔΔG = RTln(K_d,mut/K_d,WT), where R is the gas constant, T represents temperature in Kelvin, mut is the mutant studied and WT for the wild type SB27 TCR. *, less than 3- fold decrease difference compared to wild type affinity; **, between 3-and 5-fold decrease; ***, greater than 5-fold decrease. NB for non-binding observed at a maximum concentration of 200 μm HLA B*3508^LPEP, and ND, not determined.

CDR1α Loop

The CDR1α loop is situated above the α2-helix of HLA B*3508, forming interactions with a single HLA residue as well as the peptide antigen. Two mutants were investigated within this loop; namely αThr²⁹, which contacted Ala¹⁵⁸, and αTyr³¹, which contacted the peptide P7-Gln. Neither the αT29A nor αY31A mutants impacted on the affinity of the interaction, indicating that the germline-encoded CDR1α loop played a marginal role toward Ag recognition.

CDR2α Loop

The CDR2α loop contains two residues (αAsn⁵⁰ and αPhe⁵²) that contact the α2-helix of HLA B*3508 (Fig. 3a). αAsn⁵⁰ hydrogen bonds (H-bonds) exclusively to Glu¹⁵⁴, and the corresponding αN50A mutant had a critical impact on the affinity of the interaction (Table 1). αPhe⁵² formed van der Waals (VDW) contacts with three HLA B*3508 residues (Glu¹⁵⁴, Arg¹⁵⁷, and Ala¹⁵⁸) and the αF52A mutation abrogated HLA B*3508^LPEP recognition. Accordingly, the mutational data showed that the CDR2α played a critical role in the interaction with the HLA B*3508^LPEP complex.

FIGURE 3.

Interaction of the SB27 TCR with the HLA B*3508^LPEP complex. The residues on the four panels are colored according to the mutagenesis study, red are critical, orange moderately important, gray, did not affect the interaction, and white were not mutated. a, the CDR2α loop interacts with the HLA B*3508 residues, Arg¹⁵⁷ (red), Glu¹⁵⁴ (gray), and Ala¹⁵⁸ (orange) and makes contacts with the CDR3α loop. b, the CDR3α spanning across both HLA helices, represented in white schematic, mediating critical contacts (red residues) with the peptide and the HLA residues. c, the CDR1β loop “walls” the bulged peptide, forming an extensive bonding network primarily with the antigen. d, the CDR3β contacts P7-Gln of the peptide and the Arg¹⁵¹ of the HLA B*3508 by VDW interaction and a salt bridge, respectively. H-bonds are represented with red dashed line, VDW in blue dashed lines. (PDB access code 2AK4).

CDR3α Loop

The CDR3α loop of SB27 TCR contains four residues that bind to HLA B*3508^LPEP via side chain-mediated interactions (Fig. 3b), namely αSer⁹³ from the Vα gene, αPhe⁹⁵ from the non-germline-encoded region and αTyr⁹⁶ and αAsn⁹⁷ from the Jα gene. The CDR3α loop ran across the α1 and α2 helices of HLA B*3508, flanking the N-terminal end of peptide that bulges out from the Ag-binding cleft. αSer⁹³ is located directly above, and contacted P7-Gln; αPhe⁹⁵ and αTyr⁹⁶ contacted the α2-helix of HLA B*3508 (spanning between Gln¹⁵⁵ and Leu¹⁶³) as well as the peptide (between P4 to P7) via VDW and H-bond interactions; and αAsn⁹⁷ H-bonded to Gln⁶⁵ and interacted with P5 to P7 of the peptide. Given the extent of these contacts, the corresponding mutations, unsurprisingly, all had a critical effect on the affinity of the interaction (Table 1). This highlights the important role the CDR3α loop played in contacting the peptide and the HLA B*3508 molecule.

CDR1β Loop

Five HLA B*3508^LPEP residues within the CDR1β loop were mutated, where all but one residue (βAsn³⁰) exclusively contacted the peptide (between P5 and P10) (Fig. 3c, Table 1). The βH29A, βS31G, and βY33A mutants were judged to be critical for the interaction, whereas βN30A and βN28A had a moderate and negligible effect, respectively. Accordingly, the germline-encoded CDR1β loop was observed to play a critical role in the energetics of the interaction, principally by interacting with the bulged peptide.

CDR2β and CDR3β Loops

The CDR2β loop contributes minimally to the SB27 TCR-HLA B*3508^LPEP interface, and mutation of the sole CDR2β contacting residue, βSer⁵¹, did not impact on the affinity of the interaction (Table 1). The CDR3β loop contains three residues that contact HLA B*3508^LPEP, the βPro⁹⁵ sat above P7-Gln, whereas the βLeu⁹⁸ and βGlu¹⁰¹ contact HLA B*3508. The βE101A mutant did not impact on the affinity of the interaction, indicating that the salt bridge between βGlu¹⁰¹ and Arg¹⁵¹ was not energetically important for SB27 TCR binding (Fig. 3d, Table 1). Furthermore, whereas βLeu⁹⁸ formed VDW contacts with numerous HLA B*3508 residues, the βL98A mutation only resulted in a moderate impact on the binding affinity (Table 1). In contrast, the βP95A mutant significantly impacted on the affinity of the interaction, thereby highlighting the importance of this peptide-mediated contact.

Next, we mapped the impact of the mutated SB27 TCR residues onto its structure to evaluate the location and size of the energetically important residues (the hot spot). The substitutions that had a marked effect on the affinity of the interaction formed a central strip on the surface of the Ag-binding domain of the SB27 TCR (Fig. 4a). Collectively, of the six CDR loops mutated in the SB27 TCR, the sum energetic contributions indicated that CDR3α ≈ CDR1β > CDR2α > CDR3β ≫ CDR1α ≈ CDR2β.

FIGURE 4.

Energetic hot spots of the SB27-HLA B*3508^LPEP complex. a, surface representation of the SB27 TCR, with the α-chain in cyan; β-chain in pink; six CDR loops and the HLA B*3508^LPEP helices are shown as schematic representation; P4-P7 of the peptide are shown in green sticks. The critical residues are in red; the residues of moderate importance for the interaction are in orange; and the residue in dark gray were not important for the interaction. b, surface representation of the HLA B*3508 (white) bound to the LPEP peptide (light blue). SB27 CDR loops and the HLA B*3508 helices are shown as schematic representations. The CDR loops are colored in pink, green, and red for the CDR1, -2, and -3α, respectively, and in yellow, salmon, and orange for the CDR1, -2, and -3β, respectively. The critical residues are colored in red, the residues of moderate importance for the interaction are in orange, and the residues in dark gray were not important for the interaction. PDB access code 2AK4.

Energetic Landscape on HLA B*3508^LPEP Surface

To assess how HLA B*3508 contributes energetically toward SB27 TCR recognition, based on the crystal structure of the ternary complex, we substituted 10 HLA B*3508 residues to alanine, which included one control mutation (Gln⁷²) that was not involved in the interaction with the SB27 TCR. Additionally, one HLA B*3508 residue was mutated to glycine (A158G). HLA B*3508^LPEP mutations that caused less than a 3-fold loss in the affinity of the interaction with SB27 TCR compared with WT HLA B*3508^LPEP were considered to have no major effect; a 3–5-fold effect as moderate, and anything greater than 5-fold as critical to the energetic of the interaction.

Of the 10 selected and refolded single-site HLA B*3508 mutants, three and seven were located on the α1 helix (Gln⁶⁵, Thr⁶⁹, and Gln⁷²) and α2-helix (Arg¹⁵¹, Glu¹⁵⁴, Gln¹⁵⁵, Arg¹⁵⁷, Ala¹⁵⁸, Glu¹⁶¹, and Glu¹⁶³), respectively. We used SPR to measure the effect of the mutations, and established that the Q72A mutant did not affect the binding affinity. Of the two TCR-contacting residues located on the α1-helix (Q65A and T69A), neither of the corresponding mutations affected the binding affinity, thereby indicating that residues within the α1-helix were dispensable for the interaction with the SB27 TCR. Of the six α2-helix residues that were involved in contacts with the SB27 TCR, the mutational SPR analyses indicated that Glu¹⁵⁴ and Leu¹⁶³ were not required for the interaction. Interestingly αAsn⁵⁰, which contacted Glu¹⁵⁴, was shown to be critical for the interaction. However, the TCR residue αAsn⁵⁰ also made intra-TCR contacts with the CDR3α loop, and thus the role of the αAsn⁵⁰ side chain appears to be in maintaining the structural integrity of the SB27 TCR binding site. Indeed, the energetic hot spot on the HLA B*3508 surface was narrowed down to only two critically important residues (Arg¹⁵¹ and Arg¹⁵⁷) and two moderately important residues (Gln¹⁵⁵ and Ala¹⁵⁸), all located near the hinge region of the α2-helix (Fig. 4b, Table 2). Although Gln¹⁵⁵ H-bonded to the SB27 TCR and participated in stabilizing the peptide conformation, Arg¹⁵¹, Arg¹⁵⁷, and Ala¹⁵⁸ pointed away from the Ag-binding cleft, and contacted the TCR mainly via VDW contacts and one salt bridge (Arg¹⁵¹ and βGlu¹⁰¹ from the CDR3β loop), although the corresponding βE101A mutation indicated that the salt bridge was not energetically important.

TABLE 2.

Surface plasmon resonance analysis on HLA-B*3508^LPEP mutations

t_1/2 = 0.693/K_d,eq. ΔΔG = RTln (K_d,mut/K_d,WT), where R is the gas constant, T represents temperature in Kelvin, mut is the mutant studied and WT for the wild-type HLA or peptide. *, less than 3- fold decrease difference compared to wild type affinity; **, between 3-and 5-fold decrease; ***, greater than 5-fold decrease, ND, not determined.

To address if any of these four mutants (Arg¹⁵¹, Gln¹⁵⁵, Ala¹⁵⁸, and Arg¹⁵⁷) indirectly impacted on HLA B*3508^LPEP stability or peptide conformation, we undertook thermal stability assays, control SPR experiments, as well as structural studies. The thermal stability assay showed that all four HLA B*3508^LPEP mutants exhibit comparable melting temperature (T_m) to the wild type HLA B*3508 (supplemental Table S1). Arg¹⁵¹ and Arg¹⁵⁷ form intra-chain salt bridges (with HLA B*3508 Glu¹⁵⁴ and Glu¹⁶¹, respectively), and thus we considered that the Arg¹⁵¹ and Arg¹⁵⁷ mutants may have impacted on the local conformation of HLA B*3508, thereby indirectly impacting on SB27 TCR recognition. Thus we conducted the reciprocal mutagenesis experiments, and demonstrated that the E154A and E161A mutations did not impact the binding affinity (Table 2). Furthermore, to ascertain whether the mutations caused conformation shifts in the peptide itself, we determined the structures of the four HLA B*3508^LPEP mutants (supplemental Fig. 1a and supplemental Table S2). We showed that the peptide structure remains intact regardless of the HLA mutation introduced (root mean square deviation on the Ag-binding cleft range between 0.22 and 0.32 Å). As such, we concluded that the four HLA B*3508 hot spots were acting directly on SB27 TCR binding.

Significant Peptide Contribution to SB27 TCR Binding

To compare the relative energetic contribution of the peptide as opposed to the HLA B*3508 molecule toward SB27 TCR recognition, we next mutated the six peptide positions that contacted SB27 TCR and undertook SPR-based affinity measurements. The peptide residues included P4-Pro, P5-Leu, P6-Pro, P7-Gln, P9-Gln, and P10-Leu-residues that spanned the bulged region of the peptide (Fig. 3c). First we assessed the thermal stability and structural impact of these introduced peptide mutations. The thermal stability assays demonstrated that only the P4-Pro to Ala mutant decreased the stability of the HLA B*3508^LPEP complex appreciably (by 9 °C) (supplemental Table S1). Furthermore, the crystal structure of the various HLA B*3508^LPEP peptide mutants showed that the conformation of the bulged peptide was not impacted significantly for five of them (supplemental Fig. S1b and Table S2). The P5-Leu mutation to alanine increased the mobility of the peptide, which leaned more toward the α1-helix of the HLA B*3508. This highlighted the importance of the P5-Leu residue for the peptide structure and so its recognition by the TCR. Thus, with the exception of P5-Leu mutant, any potential effect of peptide mutants could be attributable to directly impacting on SB27 TCR recognition. Although mutation of the two residues at the C-terminal end of the peptide (Pro⁹–Pro¹⁰) did not affect the affinity of the interaction, mutation of the Pro⁴–Pro⁷ positions resulted in marked reduction in binding affinity. Accordingly, the ascending and central region of the peptide bulge was critical for SB27 TCR recognition (Fig. 3c, Table 2). The ΔΔG analysis showed that, whereas the peptide only represents a small component of the solvent accessible HLA B*3508^LPEP interface, it contributes 56% to the overall energetics of the interaction with the SB27 TCR (Table 2).

The reciprocal mutagenesis on HLA B*3508^LPEP allowed us to determine the importance of the peptide and the HLA heavy chain on the TCR-contacting residues. The SB27 TCR utilized only one critical (αPhe⁵²) and three moderately (βLeu⁹⁸, αPhe⁹⁵, and αTyr⁹⁶) important residues to contact HLA B*3508. This is in marked contrast with the eight critical residues (αHis²⁹, αSer³¹, αTyr³³, αSer⁹³, αPhe⁹⁵, αTyr⁹⁶, αAsn⁹⁷, and βPro⁹⁵) and one moderately (αAsn³⁰) important residue used to contact the peptide.

Cellular Assays

Next we established whether the peptide-centric nature of the SB27 TCR, as judged by SPR analysis, was also observed at the cellular level. We transduced the SB27 TCR and LC13 TCR (as a negative control) into the αβTCR negative SKW3 cell line and titrated these cells with different concentrations of wild-type HLA B*3508^LPEP tetramer (Fig. 5a). This allowed us to determine the HLA B*3508^LPEP tetramer concentration at which half of the maximum staining level was reached, as well as establishing the background MF. Using this value, we assessed the multimer staining level achieved using the HLA B*3508^LPEP mutants and compared these values to wild type staining.

FIGURE 5.

Fluorescence-activated cell sorting experiments. a, titration of SKW3-SB27 and SKW3-LC13 cells with different concentrations of wild type HLA B*3508^LPEP multimer. The MF of HLA B*3508^LPEP multimer staining is shown on the y axis and the final concentration of the multimer is shown on the x axis. b, the impact of HLA B*3508^LPEP variants were measured by tetramer staining (single concentration representing 50% of the maximum binding for the WT tetramer) and the percentage of binding is calculated by comparing with the wild type tetramer staining. The experiments on both panels have been carried in out triplicate and the error bars show the mean ± S.E.

Of the nine HLA B*3508 mutants tested (Glu¹⁶¹ excluded), five mutant multimers (Q65A, T69A, R72A, E154A, and L163A) stain comparably to the wild-type tetramer (Fig. 5b). For the Q155A and A158G mutants, low levels of tetramer staining were observed with an MF of 18.2 and 15.2%, respectively, compared with the WT tetramer. Furthermore, when R151A and R157A multimers were tested, very low MF levels were observed, values that were similar the negative control (SKW3-LC13) background level. Last, in the context of various peptide mutants, the tetramer binding results showed that P4 to P7 of the peptide were critically important for T cell binding. Accordingly, there was a strong correlation between the SPR data and the cellular staining data, which collectively demonstrated that the SB27 TCR-HLA B*3508^LPEP interaction was dependent on a few HLA B*3508 residues and was largely focused on a number of peptide-mediated contacts.

DISCUSSION

Structural and associated biophysical investigations of TCR-pMHC-I interactions have primarily focused on peptides of canonical length (8–10 amino acids), in which the peptide sits flush with the Ag-binding cleft, thereby making it challenging, a priori, to ascertain the relative roles of the peptide versus the MHC molecule in “driving” the interaction with the TCR. Some studies have indicated that the conserved TCR-pMHC docking topology is dictated by TCR-MHC contacts (9), whereas other studies have shown the peptide-centricity of the interaction (37). Indeed, it has recently been shown that energetic importance of Vβ-MHC contacts can be modified by the Vα chain of the TCR (10, 11). Moreover, the absolute importance of TCR-MHC contact points can be dictated by the fine specificity that the TCR exhibits toward a viral epitope (14, 38, 39). However, as observed in viral immunity and autoimmunity, lengthy peptides can sit high above the Ag-binding cleft, thereby making their role in enabling the TCR-pMHC interaction intuitively clearer (29, 40). However, very little is known regarding how longer MHC-I restricted peptides determine the selection of the TCR repertoire as well as how TCRs recognize such featured landscapes. To address this, previously we investigated the biased T-cell repertoire selection against HLA B*3508^LPEP, and determined how the prototypical TCR (SB27) engaged HLA B*3508^LPEP. The SB27 TCR had a finger grip hold on the MHC while forming numerous contacts with the bulged viral determinant. However, it was unclear whether the limited TCR-MHC contact points represented the key energetic contributors to the interaction, or whether the TCR-peptide interactions played the prominent role. Guided by the SB27 TCR-HLA B*3508^LPEP crystal structure, using a mutagenesis, SPR, and cellular approach, we demonstrate that the SB27 TCR-MHC structural footprint is small, the corresponding energetic footprint is smaller still, with no residues on the α1-helix of HLA B*3508 being required for the interaction. The MHC residues that contribute to the interaction were limited to two key residues (Arg¹⁵¹ and Arg¹⁵⁷) and two residues that played a moderate role (Gln¹⁵⁵ and Ala¹⁵⁸) that were located at the “elbow” region of the α2-helix. Notably, the SB27 TCR, despite the limited involvement of MHC residues, is restricted to HLA B*3508, and does not cross-react to the closely related allomorph (HLA B*3501). HLA B*3501 is distinguished from HLA B*3508 by a single buried polymorphism at position 156 (Arg¹⁵⁶ in HLA B*3508; Leu¹⁵⁶ in HLA B*3501). The effect of this polymorphism is to cause a rigid body shift in the region of the α2-helix (27) (residues 145–158) that overlaps with the MHC energetic hot spot. This highlights how the SB27 TCR shows remarkable fine specificity toward HLA B*3508.

We show that of the six peptide residues contacted by the SB27 TCR, four are critical to the interaction, thereby showing, in relative and absolute terms, that the LPEP peptide contributes markedly more toward the energetic landscape of the interaction than the HLA B*3508 molecule. Thus, the peptide centricity observed in the crystal structure is mirrored by the peptide-centric energetic landscape. A previous study has suggested that a two-step binding mechanism underpins the TCR-pMHC interaction whereby the germline-encoded loops engage the MHC initially, followed by the molding of the CDR3 loops around the peptide Ag (41). This model is hard to reconcile in the recognition of lengthy MHC-I restricted peptides, where it is more likely that the peptide-centric focus of the TCR dictates the initial interaction, and TCR-MHC interactions are subsequently formed that collectively enable productive TCR-pMHC-I engagement.

Given that the super-bulged peptide plays a central role in contributing to the energetic landscape, the question arises whether the responding TCR repertoire has been shaped primarily based on its ability to primarily recognize the peptide. The TCR repertoire directed toward HLA B*3508^LPEP in unrelated individuals exhibits a public α-chain (TRAV19*01, TRAJ34*01) and a highly conserved β chain (TRBV6-1*01), which differs only in the CDR3β loop sequences. Our data demonstrated that CDR2α and CDR3α loops play an important energetic role in interacting with the HLA and pHLA, respectively, thereby explaining the strong selective pressure for TRAV19 and TRAJ34 gene usage. The energetic landscape of the SB27 β-chain rests heavily within the germline-encoded CDR1β loop that interacts predominantly with the peptide Ag. Accordingly, the biased TRBV6-1 gene usage was principally directed against the viral determinant. These observations have resonances with a germline-encoded TCR polymorphism, which determines the protective immune response to EBV, solely by contacting the viral determinant (8). The reduced energetic dependence of the CDR3β loop was consistent with a lack of sequence conservation within this loop in unrelated individuals, and consistent with recent observations regarding how CDR3β loop diversity can be tolerated in the anti-influenza response (36). Notably, whereas αPhe⁵² and βLeu⁹⁸ interact with the HLA B*3508 energetic hot spot (Arg¹⁵⁷ and Arg¹⁵¹, respectively), the remainder of the TCR residues that defined the energetic landscape coincided with TCR residues that contacted the peptide. It will be interesting to establish how other HLA B*3508^LPEP restricted TCRs, which vary in Vα and Vβ usage, interact with the same lengthy Ag.

The SB27 TCR-HLA B*3508^LPEP interaction was principally driven by peptide contacts. To illustrate, of the 10 mutants impacting pHLA binding, 50% of them contacted the bulged peptide exclusively, 30% of TCR residues contacted both the peptide and HLA hot spots, and only 20% solely bound to HLA hot spots. Moreover, of the TCR residues that contact both the peptide and HLA surface, they appeared to have a greater focus on the peptide, rather than the HLA, as judged by the reciprocal HLA-based mutagenesis experiments. Collectively our findings highlight the peptide-focused property of the SB27 TCR as well as highlighting the selective pressure of longer peptides in driving the restricted TCR repertoire. This has significant implication for T cell selection and the diversity of anti-viral responses when longer epitopes are presented.