Genetic and structural basis for selection of a ubiquitous T cell receptor deployed in Epstein-Barr virus infection

Abstract

Despite the ∼10(18) αβ T cell receptor (TCR) structures that can be randomly manufactured by the human thymus, some surface more frequently than others. The pinnacles of this distortion are public TCRs, which exhibit amino acid-identical structures across different individuals. Public TCRs are thought to result from both recombinatorial bias and antigen-driven selection, but the mechanisms that underlie inter-individual TCR sharing are still largely theoretical. To examine this phenomenon at the atomic level, we solved the co-complex structure of one of the most widespread and numerically frequent public TCRs in the human population. The archetypal AS01 public TCR recognizes an immunodominant BMLF1 peptide, derived from the ubiquitous Epstein-Barr virus, bound to HLA-A*0201. The AS01 TCR was observed to dock in a diagonal fashion, grasping the solvent exposed peptide crest with two sets of complementarity-determining region (CDR) loops, and was fastened to the peptide and HLA-A*0201 platform with residue sets found only within TCR genes biased in the public response. Computer simulations of a random V(D)J recombination process demonstrated that both TCRα and TCRβ amino acid sequences could be manufactured easily, thereby explaining the prevalence of this receptor across different individuals. Interestingly, the AS01 TCR was encoded largely by germline DNA, indicating that the TCR loci already comprise gene segments that specifically recognize this ancient pathogen. Such pattern recognition receptor-like traits within the αβ TCR system further blur the boundaries between the adaptive and innate immune systems.

Figure 1. Overview of the AS01-GLC-A2 complex structure.

(A) Ribbon representation of the AS01-GLC-A2 complex. The TCR α-chain and TCR β-chain are depicted in light grey; CDR1α, CDR2α, CDR3α, CDR1β, CDR2β and CDR3β are depicted in green, red, blue, lime green, orange and aqua, respectively. The HLA-A*0201 platform is depicted in light pink, with the stalk and β2-microglobulin in silver. The GLCTLVAML peptide is colored orange and represented in ball-and-stick format. (B) Magnified view of the AS01-GLC-A2 complex interface from the same angle as in panel A. The position of the AS01 TCR CDR loops over the central peptide bulge can be observed. For clarity, the MHC α-2 helix is omitted. (C) Magnified view of the AS01-GLC-A2 complex interface at 90° clockwise rotation from panel B along the horizontal axis. The overall position of the AS01 TCR CDR loops over both the MHC helices and the peptide can be observed.

Figure 2. Interactions of the AS01-GLC-A2 complex.

(A) Contact footprint of the AS01 TCR on the GLC-A2 surface. The HLA-A*0201 molecule is surface represented and colored grey; the GLCTLVAML peptide is shown in ball-and-stick format and colored orange. The binding footprint over the GLC-A2 surface of the AS01 TCR CDR1α, CDR2α, CDR3α, CDR1β, CDR2β and CDR3β loops are depicted in green, red, blue, lime green, orange and aqua, respectively. The TCR/pMHC crossing angle is depicted in red. (B) Interactions governing anchoring of Leu5 to the MHC α-2 helix. The MHC cleft is represented in ribbon format and the GLCTLVAML peptide, as well as the corresponding α-2 helix contacts, are represented in ball-and-stick format. (C) Antigen-specific interactions between the AS01 TCR CDR1α loop (green sticks), the CDR3α loop (blue sticks) and the N-terminus of the GLCTLVAML peptide. (D) Antigen-specific interactions between the AS01 TCR CDR1β loop (lime green sticks), the CDR3β loop (aqua sticks) and the C-terminus of the GLCTLVAML peptide. (E) MHC (light pink sticks) interactions with the genetically unique components of the TCR α-chain framework (FW) region (grey sticks) and CDR1 loop (green sticks). (F) MHC (light pink sticks) interactions with the genetically unique components of the TCR β-chain FW region (grey sticks) and CDR2 loop (orange sticks). (G) Schematic representation of TCR and contacts (hydrogen bond, salt bridge or van der Waals interactions) with peptide and the germline origins of the TCR contact residues. Only wholly non-germline encoded residues are depicted in light grey. Asp93 is partially encoded by non-germline DNA.

Figure 3. Binding affinity and thermodynamics of the AS01-GLC-A2 interaction.

SPR measurements were conducted at different temperatures as shown. Ten serial dilutions of the AS01 TCR were measured in triplicate at each temperature; the mean response for each concentration is plotted. (A-F) The equilibrium binding constant (K_D) values were calculated in each case using a nonlinear curve fit (y =  (P₁ x)/P₂ + x)); mean plus SD values are shown. (G) The thermodynamic parameters were calculated according to the Gibbs-Helmholtz equation (ΔG° = ΔH − TΔS°). The binding free energies, ΔG° (ΔG° = -RTlnK_D), were plotted against temperature (K) using nonlinear regression to fit the three-parameter equation, (y = dH+dCp*(x-298)-x*dS-x*dCp*ln(x/298)), as previously reported . (H) AS01 TCR (18.1 µM) was injected over GLC-A2 at 5, 12, 19, 25, 32 and 37°C. The responses observed with injections of AS01 TCR over a control sample were deducted. Off-rates (K_off) were calculated assuming 1∶1 Langmuir binding using a global fit algorithm (BIAevaluate 3.1). Data and fits are shown.

Figure 4. Binding affinities of the AS01 TCR with GLC-A2 variants.

Equilibrium binding analysis at 25°C for wildtype GLC peptide (A) and alanine mutants (B-F). Ten serial dilutions of the AS01 TCR were measured in triplicate for each equilibrium experiment; the mean response for each concentration is plotted. The equilibrium binding constant (K_D) values were calculated using a nonlinear curve fit (y =  (P₁ x)/P₂ + x)) as previously reported , , ; mean plus SD values are shown.

Figure 5. Convergent recombination analysis of the AS01 TCR.

Previously reported TRAV5/TRAJ31 (A) and TRBV20-1/TRBJ1-2 (C) amino acid clonotypes specific for the GLC-A2 epitope are listed together with the number of individuals in which each clonotype was observed. Sequences were mined from previous studies A , , B and C . To assess the role of convergent recombination and TCR production frequency in the inter-individual sharing of these TRAV5/TRAJ31 and TRBV20-1/TRBJ1-2 amino acid clonotypes, computer simulations of a random V(D)J recombination process involving either the TRAV5/TRAJ31 or TRBV20-1/TRBJ1-2 gene combinations were used to estimate their relative production frequencies. The number of times that each observed GLC-specific TCR amino acid sequence was generated in the simulations of a random V(D)J recombination process is shown for the TRAV5/TRAJ31 (B) and TRBV20-1/TRBJ1-2 (D) amino acid clonotypes versus the number of individuals in which each sequence was observed. The solid horizontal lines represent the medians of the number of times that TCR sequences observed in a particular number of individuals were simulated. The dashed horizontal lines extending the width of the plots represent the mean frequency of sequence generation, across all observed GLC-specific TRAV5/TRAJ31 or TRBV20-1/TRBJ1-2 amino acid clonotypes, regardless of the number of individuals in which they were found. The data shown for the TRBV20-1/TRBJ1-2 clonotypes have been published previously and are shown here for completeness. ^a Four TRB clonotypes could not be made with the parameters used in the simulation; these are not included in panel D. The most shared GLC-specific TCRα amino acid clonotype (TRAV5/CAEDNNARLMF/TRAJ31) was encoded by 18 different nucleotide sequences and produced by a total of 117 different recombination mechanisms (i.e. different splicings of the germline genes and nucleotide additions) in the simulations. The most shared GLC-specific TCRβ amino acid clonotype (TRBV20-1/CSARDGTGNGYTF/TRBJ1-2) was encoded by 16 different nucleotide sequences and produced by a total of 47 different recombination mechanisms. The CDR3 nucleotide sequences coding for the AS01 TCR (E) and the CDR3 nucleotide sequences coding for the most shared TCRα and TCRβ amino acid clonotypes in the GLC-specific response (F). The corresponding TRBV, TRAV, TRBD, TRBJ and TRAJ genes are listed. Germline-derived nucleotides are highlighted in grey and non-germline-derived nucleotides are bolded and underlined.