A Shared TCR Bias toward an Immunogenic EBV Epitope Dominates in HLA-B*07:02–Expressing Individuals

Greater than 90% of the global human population have been exposed to EBV (1), with infection typically occurring early in life and manifesting as a high viral load lytic phase followed by a life-long asymptomatic latent phase. During the initial phase of EBV infection, naive CD8⁺ T cells significantly expand with priming events dependent on the presentation of virus-derived peptides by HLA class I molecules. These activated CD8⁺ T cells then use their armory of effector functions, including release of cytotoxic granules and production of Th1 cytokines, to eliminate virus-infected cells and abrogate further infection. This antiviral T cell immunity is of particular importance in the clinical syndromes of primary infectious mononucleosis and EBV-associated lymphoproliferative diseases (i.e., Burkitt lymphoma, posttransplant lymphoproliferative disease) (2, 3). Interestingly, a significant portion of the circulating CD8⁺ T cell repertoire is devoted to controlling asymptomatic EBV infection, with individual EBV epitope specificities accounting for 0.05–2% of total CD8⁺ T cells (4–6). Preferential recognition of lytic phase CD8⁺ T cell epitopes have been documented during early EBV infection, and during latter stages of infection, both lytic- and latent-specific memory CD8⁺ T cells are reported (7–9). Furthermore, latent-specific CD8⁺ T cell responses, across a wide range of HLA class I allotypes, have been shown to be particularly focused toward epitopes derived from the EBNA-3 protein family (2).

HLA-B*07:02 (HLA-B7) is one of the most common class I alleles within white populations, 12 and 13% of individuals in Europe and North America, respectively (10). However, relatively little is known about HLA-B7–restricted EBV epitopes compared with other common HLA, such as the HLA-A*02:01 (HLA-A2)–restricted BMLF1-derived GLCTLVAML (EBV_GLC) epitope (11–15) or the HLA-B*08:01 (HLA-B8)–restricted EBNA-3A–derived FLRGRAYGL (EBV_FLR) epitope (16–20). Interestingly, in herpes virus-exposed individuals [i.e., CMV (21) and EBV (22)], the coexpression of HLA-A2 and -B7 allotypes induced an immunodominance hierarchy, in which the HLA-B7–restricted epitope-specific responses were significantly greater in magnitude than HLA-A2–restricted epitope-specific responses. Further, examination of a panel of HLA-B7–restricted immunogenic EBV peptides revealed that CD8⁺ T cell responses were highly focused toward the EBNA-3A–derived epitope (residues 379–387) RPPIFIRRL (EBV_RPP) in >80% of healthy individuals (22–24). In comparison, either no or weak CD8⁺ T cell responses were observed toward other HLA-B7–restricted EBV peptides including QPRAPIRPI (EBNA-3C), LPCVLWPVL (BZLF1), VPAPAGPIV (EBNA-3A), IPQCRLTPL (EBNA-1), RPQKRPSCI (EBNA-1), and LPRAWLQRL (BBLF2/3) (22–28).

TCR repertoire diversity is associated with the effective control of a myriad of pathogenic infections and has been shown to vary across epitopes and individuals (29–31). The net benefit of broad TCR diversity can include reduced disease severity (29), enhanced CD8⁺ T cell function (32), peptide variant cross-reactivity (33, 34), and the prevention of viral escape mechanisms (35, 36). Conversely, the narrowing of the TCR repertoire specific for an individual epitope, promoting TCR bias, can result from preferential usage of certain TCR gene segments. Such TCR bias is observed in various disease settings and is influenced by factors such as thymic selection, initial T cell activation and proliferation fitness, persistent infection and is characterized by the selection of “private” versus “public” TCRs (13). Extensive thymic TCR diversity dictates that substantial TCR sharing between individuals is unlikely; however, identical TCR sequences specific to defined Ags in different individuals have been well documented and are referred to as public T cell responses (37). For example, the HLA-B8–restricted EBV_FLR epitope selects for CD8⁺ T cells expressing a highly biased and public TCR typified by the LC13-clonotype (38, 39). Conversely, other TCRs specific for HLA-B8 viral epitopes, including the EBV BZLF-1–derived epitope RAKFKQLL, show minimal sequence similarities and are essentially private TCRs that are unique to individuals (40).

This study explores whether the prominent CD8⁺ T cell response toward the immunodominant EBV_RPP epitope is driven by a biased TCR usage in HLA-B7–expressing individuals. Both healthy individuals, as well as lung transplant recipients (LTR), which are exposed to a larger antigenic load because of their immunocompromised state, were examined with similar EBV_RPP–specific TCR repertories being observed across both groups. Our findings demonstrate a biased V region usage of the TCRβ-chain (TRBV4-1), with this profile representing a shared TCR in our cohort. The molecular basis for the repeated selection of these TRBV4-1 TCRs in multiple individuals was investigated by interrogating the interaction of a representative TRBV4-1⁺ TCR with HLA-B7/EBV_RPP using a combination of biochemical and structural analyses.

HLA-B7–positive healthy individuals (n = 5) and LTR (n = 5) were recruited to the study (Table I). All LTR received standard triple-therapy immunosuppression (calcineurin inhibitor, steroid, and antiproliferative drugs), with blood samples collected at three time intervals in the first year after transplantation (Tx) (pre-Tx or 1, 6, 11–13 mo). All LTR were EBV IgG seropositive at the time of transplant, except LTR133. Additionally, no EBV viremia was detected in the first 12 mo post-Tx for any LTR. All study participants provided written consent, with ethics approval granted by The Alfred Hospital (Melbourne, VIC, Australia; Ethics numbers 175/02 and 110/12), Monash University (Clayton, VIC, Australia; Ethics number CF12/1636–2012000880), and the Australian Bone Marrow Donor Registry (Alexandria, NSW, Australia; Ethics number 2012-05). Blood samples were collected in heparinized vacutainer tubes and PBMC were isolated by standard Ficoll-Paque (GE Healthcare, Uppsala, Sweden) density gradient centrifugation and cryopreserved at −196°C until required.

HLA-B*07:02–expressing APCs derived from either HLA class I–reduced (C1R; C1R.B7) (41) or K562 (42) (K562.B7 kindly provided by Prof. Frans Claas, Leiden University Medical Centre, Leiden, the Netherlands) cells were maintained in RF10 (composed of RPMI 1640 [Life Technologies, Grand Island, NY] supplemented with 2 mM MEM nonessential amino acid solution [Life Technologies], 100 mM HEPES [Life Technologies], 2 mM l-glutamine [Life Technologies], penicillin/streptomycin [Life Technologies], 50 mM 2-ME [Sigma-Aldrich, St. Louis, MO], and 10% heat-inactivated FCS [Sigma-Aldrich]) as previously described (43). HLA-B7 cell surface expression on these cell lines, compared with parental cells, was confirmed via flow cytometry after staining with anti-human HLA-B7/27 ME-1 hybridoma Ab (produced in-house) and goat anti-mouse IgG PE (Southern Biotech, Birmingham, AL).

In vitro expansion of EBV_RPP–specific memory CD8⁺ T cells was achieved following stimulation of PBMC with γ-irradiated EBV_RPP–pulsed autologous cells (1 μM peptide, 3000 Rads) at a 2:1 ratio in RF10 supplemented with 20 U/ml IL-2 (Cetus, Emeryville, CA or PeproTech, Rocky Hill, NJ) for 13 d at 37°C, 5% CO₂, essentially as previously described (19, 21, 44).

The specificity and activation of outgrown EBV_RPP–specific CD8⁺ T cells were assessed by anti-CD8 and EBV_RPP tetramer costaining, followed by measurement of intracellular Th1 cytokine production for functionality using flow cytometry. HLA-B7/EBV_RPP protein for tetramer staining was generated as previously described (43). Briefly, 2 × 10⁵ day 13 T cell cultures were stimulated with 1 × 10⁵ APCs (±1 μM peptide) or peptide alone for a total of 6 h with 10 μg/ml brefeldin A (Sigma-Aldrich) added for the last 4 h. T cells were phenotyped with anti-CD8 PerCP Cy5.5 (clone SK1; BD Biosciences, San Jose, CA), HLA-B7/EBV_RPP tetramer (conjugated to either PE or allophycocyanin), and live/dead fixable aqua stain (Thermo Fisher Scientific, Waltham, MA). T cells were then fixed in 1% paraformaldehyde (ProSciTech, Kirwan, QLD, Australia), permeabilized in 0.3% saponin (Sigma-Aldrich) containing anti–IFN-γ PE-Cy7 (clone B27; BD Biosciences) and anti-TNF-α V450 (clone Mab11; BD Biosciences), then acquired on an LSRII Flow Cytometer (Becton Dickinson, San Jose, CA). All flow cytometry data were analyzed using FlowJo software (TreeStar, Ashland, OR).

Cryopreserved EBV_RPP–specific CD8⁺ T cells lines were thawed and rested overnight in RF10. T cell lines were incubated with 1 μM of either EBV_RPP peptide alone or EBV_RPP–pulsed C1R.B7 cells for 2 h before detection of cytokine secretion using an anti–IFN-γ Ab, as previously described (IFN-γ Secretion Assay Detection Kit allophycocyanin; Miltenyi Biotec, Auburn, CA) (45). CD8⁺ T cells were single-cell sorted directly into either 96-well twin.tec (Eppendorf, Hamburg, Germany) or Hard-Shell semiskirted (Bio-Rad Laboratories, Berkley, CA) PCR plates based on both tetramer specificity and ± IFN-γ production: CD8⁺EBV_RPP⁺IFN-γ⁻ (hereafter referred to as CD8⁺EBV_RPP⁺) or CD8⁺EBV_RPP⁺IFN-γ⁺ (FACSAria I, BD Biosciences; operated by Doherty Institute Flow Cytometry Facility, University of Melbourne or FlowCore, Monash University). Sorted plates were immediately centrifuged cold (pulse spin) and stored at −80°C until required. TCR analysis of paired CDR3α and β loops were carried out using multiplex-nested RT-PCR and sequencing of α and β gene products as previously described (34, 45, 46). The CDR3 lengths were calculated from position four excluding the final xF motifs (47). All circular graphical representations were created using the circlize visualization package in R and sequence logos were generated using the WebLogo application (48–50).

Full-length human TCRα and TCRβ cDNA were cloned into a self-cleaving 2A peptide-based pMIG vector as described previously (51). HEK293T packaging cells were incubated with 4 mg pEQ-pam3(-E) and 2 mg pVSV-G packaging vectors, in the presence of 4 mg pMIG vector each containing a specific TCR transgene (LTR54.1 LTR54.2, LTR117, or LTR119) using Lipofectamine 3000 (Life Technologies, Carlsbad, CA). HEK293T cell culture supernatant containing virus particles carrying the TCR transgene was then used to retrovirally transduce GFP-tagged SKW3.hCD8αβ cells, which are negative for endogenous TCRαβ but contain CD3 and signaling components, as previously described (45). SKW3.hCD8αβ.TCR (hereafter referred to as SKW3) cell lines were maintained in RF10: SKW3.LTR54.1, SKW3.LTR54.2, SKW3.LTR117 and SKW3.LTR119. The specificity and functionality of the SKW3.TCRs were assessed via cell surface staining with anti-CD69 allophycocyanin (clone L78; BD Biosciences) following 16–20 h incubation with a 1:1 ratio of 1 × 10⁵ APCs (±1 μM peptide) at 37°C, 5% CO₂. These cells were costained with anti-CD8 PerCP Cy5.5, anti-CD3 PECy7 (clone SK7; BD Biosciences) and live/dead fixable aqua stain (Thermo Fisher Scientific). Stimulation with Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher Scientific) was included as a positive control.

The HLA-B7/EBV_RPP complex and the HD14 TCR were produced using bacterial expression of inclusion bodies and refolding. Briefly, soluble HLA class I H chain and the β2m proteins were expressed separately in Escherichia coli inclusion bodies as previously reported (43). The HD14 TCR-α and β-chains were also expressed in separate E. coli inclusion bodies, then subsequently purified and solubilized (52). We have previously reported the structure of the binary HLA-B7/EBV_RPP [Protein Data Bank (PDB) accession number 5WMO (43)]. Crystals of the ternary HD14 TCR–HLA-B7/EBV_RPP complex were grown by the hanging-drop, vapor-diffusion method at 20°C with a protein/reservoir drop ratio of 1:1, at 8 mg/ml in 10 mM Tris-HCl (pH 8), 150 mM NaCl using 24% PEG 400, 0.1 M NaCl, 0.1 M sodium-citrate (pH 6). The crystals were soaked in a cryoprotectant solution containing mother liquor solution with the PEG concentration increased to 30% (w/v) and then flash frozen in liquid nitrogen. Data were collected on the MX1 beamline (53) at the Australian Synchrotron (Clayton, VIC, Australia) using the ADSC-Quantum 210 CCD detector (at 100K). Data were processed using XDS software (54) and scaled using SCALA software (55) from the CCP4 suite (56). The structures were determined by molecular replacement using the PHASER program (57) with LC13 TCR as the search model for the TCR (PDB accession number 1KGC) (17) and the previously solved HLA-B7/EBV_RPP structure minus the peptide as the search model for the HLA (PDB accession number 5WMO) (43). Manual model building was conducted using the Coot software (58) followed by maximum-likelihood refinement with the Buster program (59). The final model has been validated using the PDB validation Web site and the final refinement statistics are summarized in Table II. All molecular graphics representations were created using PyMol (60). Structural alignment is calculated on Cα atoms of each residue using the program Superpose (61) from the CCP4 suite (56). The structure has been validated by the PDB validation server (HD14 TCR-HLA-B*07:02/EBV_RPP; accession number 6VMX [https://www.rcsb.org/structure/6VMX]).

Surface plasmon resonance experiments were conducted at 25°C on the BIAcore 3000 instrument (GE Healthcare, Buckinghamshire, U.K.) with 0.1 M Tris-HCl (pH 8), 0.15 M NaCl buffer supplemented with 1% BSA to prevent nonspecific binding. The HD14 TCR was coupled to the sensor chip via the mAb 12H8 (18). The experiment was conducted (n = 2, in duplicate) as previously described (52) using a concentration range for the HD14 TCR with a maximum of 20 μM. BIAevaluation Version 3.1 was used for data analysis with the 1:1 Langmuir-binding model.

All statistical analyses were undertaken using Prism 8 (GraphPad, La Jolla, CA). Significance of ex vivo– and in vitro–specific CD8⁺ T cells was assessed using a Wilcoxon matched-pairs signed rank test and CD8⁺ T cell activation was assessed using a Friedman test with Dunn multiple comparisons test.

HLA-B7⁺ PBMC from five healthy individuals and five LTR (Table I) were in vitro stimulated with γ-irradiated EBV_RPP–pulsed autologous cells. All EBV-seropositive individuals showed detectable EBV_RPP–specific CD8⁺ T cells, with in vitro–expanded T cell cultures having an increased magnitude of EBV_RPP–specific CD8⁺ T cells compared with ex vivo T cells, as measured by tetramer staining (Fig. 1Ai and ii). The capacity of in vitro–expanded EBV_RPP–specific CD8⁺ T cells to elicit effector functions, such as the production of Th1 cytokines (IFN-γ and/or TNF-α), was confirmed following restimulation with cognate peptide (EBV_RPP). As expected, background (media) or negative controls (K562 parental and K562.B7) did not elicit any T cell reactivity. However, restimulation with K562.B7 APCs pulsed with cognate EBV_RPP peptide (K562.B7/EBV_RPP) resulted in >40% of EBV_RPP–specific CD8⁺ T cells producing IFN-γ and/or TNF-α (Fig. 1Bi and ii), with the majority of activated T cells (>80%) producing both Th1 cytokines. As expected, for the EBV seronegative patient LTR133, no measurable EBV_RPP–specific CD8⁺ T cells were observed (Fig. 1A, 1B), thus serving as a specificity control in this study.

To determine T cell functional avidity, in vitro–expanded EBV_RPP–specific CD8⁺ T cells (healthy individuals only) were restimulated with decreasing concentrations of cognate EBV_RPP peptide ranging from 10⁻⁶ to 10⁻¹² M, either as EBV_RPP–pulsed C1R.B7 cells (C1R.B7/EBV_RPP) or free peptide alone. The data clearly showed that EBV_RPP–specific CD8⁺ T cells were of very high functional avidity with responses to the cognate peptide in the nanomolar ranges. For example, IFN-γ production was readily detectable at 10⁻¹⁰ M (Fig. 1C), with these cells displaying an EC₅₀ of 1.6 × 10⁻⁹ M in response to EBV_RPP peptide.

To determine the composition and diversity of the TCRαβ repertoire recognizing the HLA-B7/EBV_RPP complex, we profiled the EBV_RPP–specific TCRαβ clonotypes from the expanded T cell lines of the five healthy individuals and three LTR in our study cohort. The TCRαβ signatures were determined using single-cell multiplex-nested RT-PCR for simultaneous detection of paired CDR3α and CDR3β-chains (Supplemental Tables I, II). For most of our cohort, the CD8⁺EBV_RPP⁺ TCRαβ clonotypes (sorted based on specificity using tetramer) reflected those of the CD8⁺EBV_RPP⁺IFN-γ⁺ (sorted based on tetramer and functionality by IFN-γ production) TCRαβ repertories, suggesting that functional activation was not restricted by the TCR clonotype. Of note, for two healthy donors (NM009 and NM015) a paired TCRβ-chain was ND for the majority of cells analyzed (>75%) and were therefore excluded from further analysis. From 437 TCRs sequenced, we identified 19 unique TCRαβ pairs (grouped based on shared variable and junction regions) across six individuals (Fig. 2A), with most displaying restricted repertories (mean of 4.6 ± 1.4 TCRs per donor) often biased to a single dominant TCR.

TRBV4-1 usage was observed in five of the six individuals and accounted for 30–100% of an individual’s TCR repertoire, with no other TRBV observed in donors NM005 and LTR54 (Fig. 2A; TRBV4-1 depicted in various shades of blue). The TRBV4-1 was rearranged with five different junctional regions (TRBJ1-2, 1–3, 2-1, 2–3, and 2–7), with the TRBJ2-3 expressed by three individuals (HD14, LTR54, LTR117). Interestingly, the CDR3β loop from three individuals only differed by 2 aa (CASSQETGXYZQYF, where X denotes a variable amino acid of Q/S/I and Z denotes a variable amino acid of A/T/E; refer to Fig. 2B, Supplemental Tables I, II). With respect to the CDR3β length, the CDR3β loops from TRBV4-1⁺ TCRs were generally shorter than those encoded for by other TRBV (9–11 versus 10–13 aa residues in length) (Fig. 2C). The TRBV4-1 β-chain was paired with α-chains expressing four different variable regions (TRAV38-1, 38-2, 24, and 40) (Fig. 2A).

The TCR repertoire of five individuals (HD14, NM005, LTR54, LTR117, and LTR119) shared a common TRAV38-1/TRBV4-1 usage, with three (HD14, LTR117, and LTR119) also expressing a TRAV24/TRBV4-1 usage (Fig. 2A; TRAV38-1 depicted in shades of red). The α-chain variable and junction regions were conserved across all five individuals, with the CDR3α loop only differing by 2 aa for NM005, HD14 and LTR54 (CAXXYNNNDMRF, where X is any amino acid; refer to Fig. 2B, Supplemental Tables I, II). Furthermore, TRAV38-2 only differs from the TRAV38-1 by 2 aa residues in the CDR1 loop (position 30, Asn to Ser; position 31, Asn to Asp). The CDR3α loop length was generally short and conserved (7-9mer) with the exception of TRAV40 (13mer); the TRAV38⁺ TCRs were found to be 7 aa in length (Fig. 2C).

To determine the stability of the EBV_RPP–specific TCR repertoire over time, the repertoire of HD14 (4-y period; 2010–2013) and LTR117 (12 mo period; pre-Tx to 11 mo post-Tx) were examined longitudinally. Overall the TCR repertoires were found to be highly static over time, with little perturbation of TCR clonotypes (Supplemental Tables I, II). The EBV_RPP repertoire also remained stable, given that no EBV reactivation events occurred during this period, suggesting a continuity of TCR clonotypes with the same gene usage bias.

To confirm that the EBV_RPP–specific TCRs identified corroborate with preliminary T cell line data, we transduced three high frequency TCRs containing the common TRBV4-1 gene (LTR54.1: TRAV38-1_TRBV4-1; LTR117: TRAV38-1_TRBV4-1 and LTR119: TRAV40_TRBV4-1) into the TCRαβ-negative SKW3.hCD8αβ cell line. In addition, a second TCR (LTR54.2) with the same V region usage but different junction and CDR3 regions to the LTR54.1 was also selected to determine the specificity of the TCR/peptide/HLA interaction. The capacity of these SKW3.TCRs to be activated following engagement with cognate peptide (EBV_RPP) presented by HLA-B7⁺ APCs was measured by upregulation of cell surface CD69 using flow cytometry. Background (media) or negative controls (C1R parental and C1R.B7) did not induce T cell reactivity. However, restimulation with C1R.B7 in the presence of cognate EBV_RPP peptide (C1R.B7 + EBV_RPP) mediated cell surface CD69 upregulation (Fig. 3A, 3B). Both LTR54-derived TCRs showed the highest and similar activation levels (mean fluorescence intensity [MFI] CD69⁺: LTR54.1 = 30,316 and LTR54.2 = 30,443), followed by LTR117 (MFI CD69⁺: 22,922) and LTR119 (MFI CD69⁺: 19,477).

To further understand TCR interactions with the HLA-B7/EBV_RPP complex, we examined one of these newly identified TCRs for further biophysical and structural characterization. The HD14 TCR (TRAV24/TRAJ37_TRBV4-1/TRJB1-2, CDR3α CAFGSSNTGKLIF, CDR3β CASSQDLFTGGYTF) was the highest frequency TCR observed in this healthy individual, contained the common TRBV4-1 gene and was produced recombinantly in sufficient yield to allow further biophysical and structural studies. Interestingly, the same variable and junction region usage was identified in LTR117 (TRAV24_TRBV4-1: CDR3α CAFASSNTGKLIF, CDR3β CASSQDIWTSGYTF) and LTR119 (TRAV24_TRBV4-1: CDR3α CAFSSSNTGKLIF, CDR3β CASSQDRFTDPYGYTF), with only a single amino acid mutation in the CDR3α (underlined) and a modified motif in the CDR3β loop (Supplemental Tables I, II).

To determine TCR affinity, we performed surface plasmon resonance with the HD14 TCR against HLA-B7/EBV_RPP complex. The HD14 TCR binds with a high affinity of ∼1 μM (Fig. 3C, Table II), which is at the upper end of affinity range for a HLA class I–restricted TCR and is well above the observed average CD8⁺ TCR binding affinity of 35 μM for solved peptide/HLA complexes (62, 63). In addition, the high affinity was associated with a very slow off rate observed for the HD14 TCR of 0.035 s⁻¹ compared with other CD8⁺ TCRs. Finally, the high affinity of the HD14 TCR for HLA-B7/EBV_RPP also supports the high avidity interactions (nanomolar levels) observed in the HD14 EBV_RPP–specific CD8⁺ T cell line (Fig. 1C).

To understand the molecular basis underpinning the TRBV4-1⁺ TCR bias we determined the crystal structure of the HD14 TCR in complex with HLA-B7/EBV_RPP at a resolution of 3.1 Å (Fig. 4A, Table III). As within the range of most αβTCR-peptide–MHC class I structures (62), the HD14 TCR docked 79° above the long axis of the HLA-B7/EBV_RPP–binding cleft (Fig. 4B, Supplemental Table III). The TCR α-chain and TCR β-chain contributed 57% and 43% of the buried surface area to the interaction, respectively. In this study, the CDR3β loop was the main contributor to the interaction (26% buried surface area), contacting both the HLA-B7 H chain and the peptide.

Previously, we reported the binary structure of the EBV_RPP peptide bound to HLA-B7 (PDB accession number 5WMO) (43). Upon binding by the HD14 TCR, the peptide–HLA-I adopted a very similar conformation (rmsd of 0.68Ų) to that observed in the binary structure (Fig. 4C). In relation to the peptide, P7-Arg and P8-Arg side chains showed slight movement, whereas the P6-Ile was pushed aside by the CDR3β loop reorientating to the plane of the Ag-binding cleft (Fig. 4C). The HD14 TCR interacted across the entire length of EBV_RPP peptide, with P1-Arg, P4-Ile, P5-Phe, P6-Ile, and P8-Arg forming contact points with the TCR (Fig. 4B).

All three HD14 TCR CDRβ loops interact with the peptide. Namely, Gln^108β, Asp^109β, and Thr^112β of the CDR3β loop interacted with P6-Ile and P8-Arg (Fig. 4D, 4E). Arg^37β from the CDR1β loop is positioned over the top of the P8-Arg side chain, whereas Tyr^58β from the CDR2β loop forms a hydrogen bond with P8-Arg (Fig. 4E).

In relation to the molecular basis underpinning the TRBV4-1 TCR bias, we examined the CDR1β and CDR2β loop interactions with the HLA-B7/EBV_RPP complex. First, the CDR1β Arg³⁷ interacted with both Glu⁷⁶ and Arg⁷⁵ residues, on the α1 helix of the HLA-B7 Ag-binding cleft, via a hydrogen bond and van der Waals interactions (Fig. 4F). Second, the Tyr^58β inserted itself between the EBV_RPP peptide and the α1 helix making hydrogen bonds with Thr⁷³ (Fig. 4G). Lastly, Ser^57β made a water-mediated hydrogen bond with another α1 helix residue Gln⁷² (Fig. 4F).

Interestingly, the TRBV4 TCR gene segment is the only one to carry both the Arg^37β and Tyr^58β germline-encoded residues within its CDR1/2 loops explaining the clonotypic bias of TRBV4-1. Moreover, analysis of CDR3β loop sequences of TRBV4-1⁺ HLA-B7/EBV_RPP–specific TCRs show that at position 108, the Gln residue is conserved, and at position 109, both Asp and Glu (another negatively charged residue) are prevalent (Fig. 2B). Accordingly, the structure of the ternary complex provides a molecular basis for understanding the observed TRBV4-1 bias toward the HLA-B7/EBV_RPP determinant; based on key germline-encoded residues from the CDR1β and CDR2β interacting with the peptide ligand and the α1 helix of HLA-B7, as well as the selection of prevalent negatively charged CDR3β loop residues that complement the positively charged residues of the peptide ligand.

Our study focused on the immunodominant and highly prevalent HLA-B7 EBV_RPP epitope, with all EBV-seropositive individuals showing detectable, yet variable, ex vivo and in vitro frequencies of EBV_RPP–specific CD8⁺ T cells. The HLA-B7 EBV_RPP–specific TCR repertoire in our cohort demonstrated remarkable bias toward the use of TRBV4-1 and TRAV38-1. Such biased TCR usage appears to be a general phenomenon of repertoires specific for immunodominant epitopes from persistent viral infections (64).

Dissecting the TCR repertoire specific for immunogenic epitopes is critical for understanding the molecular mechanisms of T cell–mediated responses toward combating viral infection. Although TCR repertoire studies focusing on immunodominant EBV epitopes restricted to HLA-A2 (EBV_GLC) (11–15) and HLA-B8 (EBV_FLR) (16, 18, 20, 65) have been extensively reported, there is less evidence regarding HLA-B7–restricted EBV_RPP–specific CD8⁺ T cell responses. Although studies have addressed CD8⁺ T cell EBV_RPP specificity and activation (22–24), only two studies have examined the TCRβ repertoire specific to HLA-B7/EBV_RPP (66, 67). In this study, EBV_RPP–specific CD8⁺ T cells were single-cell sorted to simultaneously determine paired α- and β-chain gene usage in both healthy and immunocompromised individuals. Despite immunosuppression following Tx, which likely results in exposure to a larger Ag load, we observed similar EBV_RPP–specific TCR repertories in our expanded CD8⁺ T cell lines from LTR to that of healthy individuals, which remained static over time in the absence of an EBV reactivation event. This is aligned with observations of immunodominant HLA-A2–restricted EBV_GLC–specific αβTCRs, in which common variable regions were found to be shared across individuals and the repertories were relatively stable under immunosuppressive conditions following organ Tx (15).

In our cohort, the EBV_RPP–specific TCR repertoires expressed by the majority of individuals were dominated by a single TCR clonotype (followed by a few smaller clonotypes) that shared identical α- and/or β-chain variable regions between donors, revealing strong bias in the TCR usage for recognition of the HLA-B7/EBV_RPP complex. In this study, we report that five individuals predominantly expressed TRBV4-1, with this V region accounting for 30–100% of the expanded HLA-B7/EBV_RPP repertoire. Our data are in concordance with a previous study of HLA-B7/EBV_RPP–specific TCR β-chains, in which TRBV4-1 represented >55% of the TCR repertoire (66). This clonotypic bias suggests that germline-encoded residues in the CDR1β and CDR2β loops (or the TCRβ framework region) may contribute specificity determining contacts that stabilize the TCR/peptide/HLA interface. Based on our structure, we can assert that two germline-encoded residues on CDR1β and 2β are making important interactions with the EBV_RPP peptide, residues that are only carried on the TRBV4-1 gene segment. Interestingly, TRBV4-1 expression bias has been observed for TCRs recognizing CD1 complexes (68–70), as well as TCR recognition of a HLA-B8–restricted hepatitis C virus peptide (HSKKKCDEL) (71). In addition, in this study, we report, to our knowledge, the first description of the paired αβTCR repertoire for HLA-B7/EBV_RPP. We frequently observed three different α-chains, TRAV38-1, TRAV38-2, and TRAV24, that were paired with TRBV4-1. As TRAV38-1 and TRAV38-2 only differ by 2 aa residues in the CDR1 loop, we hypothesize they might form the same interaction with HLA-B7/EBV_RPP. Finally, the TRAV38-1, TRAV38-2, and TRAV24 segments all share a tyrosine as the antepenultimate (i.e., third from the end) residue in their CDR1α loop, which based on our structural analysis of HD14 TCR (TRAV24) contacts the Gln¹⁵⁵ of the HLA-B7 α2-helix.

An alternate study by Koning et al. (67) examined the TCR β-chain repertoires specific to 12 viral Ags, demonstrating that the ex vivo and in vitro repertoires were divergent with shifts in the clonotype hierarchy. However unsurprisingly, these changes were much more pronounced in polyclonal than monoclonal or oligoclonal TCR repertoires. In agreement, we have previously observed ex vivo–derived and in vitro–expanded analogous oligoclonal TCR repertoires specific to HLA-A*02:01–restricted CMV-specific epitope, NLVPMVATV (45). The ex vivo EBV_RPP–specific repertoires observed by Koning et al. (66, 67) were oligoclonal (average number of four clones, n = 6), with the ex vivo repertoire reasonably well reflected in the expanded repertoire across three individuals. Therefore, we believe our in vitro repertoire data captures the main TCR usage shared across donors specific for the HLA-B7/EBV_RPP complex.

Having profiled the HLA-B7/EBV_RPP TCR repertoires from eight individuals, we sought to determine the structural basis of the biased HLA-B7/EBV_RPP recognition. We selected the HD14 TCR for structural analysis, which shared sequence features with several of the other clones characterized in this study (Supplemental Tables I, II) and others (67). The HD14 TCRαβ clonotype was also observed for LTR117, consisting of TRAV24/TRAJ37_TRBV4-1/TRBJ1-2 genes that also comprised the same number of CDR3αβ amino acid residues (CDR3α CAFXSSNTGKLIF, CDR3β CASSQDXXTXGYTF, where X denotes a variable amino acid). In addition, the exact β-chain has already been reported by Koning et al. (67) and the α-chain (paired with an unknown β-chain) was identified in our donor NM009. Mapping of the TCR residue interactions with the HLA-B7/EBV_RPP complex revealed that TCR CDR3αβ and EBV_RPP peptide interactions (sequence underlined above) are conserved across these three α-chain and three β-chain sequences that share the TCR clonotype.

Our structural analysis suggests that the TRBV4-1 gene bias is likely because of germline-encoded residues, including the dual and unique presence of Arg³⁷ and Tyr⁵⁸ within the CDR1/2β loops. These two residues are pivotal in the bonding network of both the EBV_RPP peptide and HLA-B7 molecule and thus, foreseeably provide the basis for the high selection of this Vβ-chain within the HLA-B7-EBV_RPP–specific clonotypes. Interestingly, a recent study using alanine scanning of CD1c-restricted TCRs also exhibiting a TRBV4-1 bias showed that Arg^37β and Tyr^58β were both critical in conferring reactivity (70). In addition, we found the Gln^108β residue to be conserved across all TRBV4-1 sequences observed, whereas position 109β was either Asp or Glu, which share very similar chemical properties and likely maintain the interaction with the HLA/peptide complex. This conservation is most likely because of their interaction with the EBV_RPP peptide itself. Furthermore, the CDR3α loop was observed to act as a “lid” on top of the N-terminal part of the Ag-binding cleft, which might explain the conserved loop length observed in the HLA-B7-EBV_RPP TCR repertoire. The importance of the length instead of the sequence within the CDR3 loop has been observed before (72).

Skewing of the TCR repertoire is often observed in response to persistent viral infections including HIV/SIV, EBV, CMV, and influenza virus (64), with this feature suggested to result from the prior selection of T cells with high-affinity TCRs (37). In accordance, our study has unveiled a shared TRBV4-1 TCR bias that dominates the EBV_RPP–specific CD8⁺ T cell response among unrelated HLA-B7⁺ individuals. The highly prevalent pairing of TRBV4-1 with TRAV38 or TRAV24 focused toward the latent EBNA-3A protein further confirms the shared nature of HLA-B7/EBV_RPP–specific TCR repertoire in individuals with chronic EBV infection. We also highlight key residues within the CDR1/2β loops that drive molecular interactions between this TCR and HLA-B7/peptide complex. Together these findings provide a greater understanding of the antiviral CD8⁺ T cell response toward a latently expressed EBV epitope in HLA-B7⁺ individuals.

We thank the Doherty Institute Flow Cytometry Facility, Monash University FlowCore, Monash Macromolecular Crystallization Facility, and the Australian Synchrotron staff for technical assistance.

The authors have no financial conflicts of interest.