Research ArticleAUTOIMMUNITY

The shared susceptibility epitope of HLA-DR4 binds citrullinated self-antigens and the TCR

See allHide authors and affiliations

Science Immunology  16 Apr 2021:
Vol. 6, Issue 58, eabe0896
DOI: 10.1126/sciimmunol.abe0896

TCR detection of citrullinated epitopes

The autoreactive T cell repertoire driving disease activity in rheumatoid arthritis (RA) includes CD4+ T cells that recognize MHC-bound peptides with arginine residues posttranslationally modified to citrulline. Some HLA-DRB1 alleles have a shared susceptibility epitope associated with increased RA incidence. Lim et al. analyzed the repertoire, binding properties, and structure of multiple T cell receptors (TCRs) derived from humanized mice reactive with citrullinated peptides presented by HLA-DR4. TCR repertoire analysis revealed a citrullinated antigen-specific motif, conserved in both mice and humans. Crystal structures revealed direct contact of the TCR with the shared epitope on HLA-DR4. Alleles of HLA-DRB1 with the shared epitope contribute to the development of RA through both binding of stimulatory peptide epitopes and direct contact with a biased set of TCRs.

Abstract

Individuals expressing HLA-DR4 bearing the shared susceptibility epitope (SE) have an increased risk of developing rheumatoid arthritis (RA). Posttranslational modification of self-proteins via citrullination leads to the formation of neoantigens that can be presented by HLA-DR4 SE allomorphs. However, in T cell–mediated autoimmunity, the interplay between the HLA molecule, posttranslationally modified epitope(s), and the responding T cell repertoire remains unclear. In HLA-DR4 transgenic mice, we show that immunization with a Fibβ-74cit69-81 peptide led to a population of HLA-DR4Fibβ-74cit69-81 tetramer+ T cells that exhibited biased T cell receptor (TCR) β chain usage, which was attributable to selective clonal expansion from the preimmune repertoire. Crystal structures of pre- and postimmune TCRs showed that the SE of HLA-DR4 represented a main TCR contact zone. Immunization with a double citrullinated epitope (Fibβ-72,74cit69-81) altered the responding HLA-DR4 tetramer+ T cell repertoire, which was due to the P2-citrulline residue interacting with the TCR itself. We show that the SE of HLA-DR4 has dual functionality, namely, presentation and a direct TCR recognition determinant. Analogous biased TCR β chain usage toward the Fibβ-74cit69-81 peptide was observed in healthy HLA-DR4+ individuals and patients with HLA-DR4+ RA, thereby suggesting a link to human RA.

INTRODUCTION

The human leukocyte antigen (HLA) locus is the most polymorphic region in the human genome, engendering HLA molecules to present a broad array of peptides (1). Together with a diverse T cell repertoire, this enables the T cell–mediated adaptive immune response to discriminate the myriad of self-peptides from non–self-peptides (2). Despite the key role of the HLA in protective immunity, certain HLA alleles, particularly those encoded by the HLA class II locus, are frequently correlated with aberrant T cell immunity, including autoimmunity and other immune-mediated inflammatory diseases (3). For example, HLA-DR4 allomorphs are associated with increased susceptibility to rheumatoid arthritis (RA) (4). However, the relationship between the HLA, the self-antigens, and the responding T cell repertoire that ultimately causes T cell autoimmunity remains unclear in many instances (5).

The breaking of T cell tolerance can be attributed to a number of mechanisms, including unusual T cell receptor (TCR)–HLA docking topologies, altered peptide binding to HLA molecules, molecular mimicry, and neoepitope generation (5). For example, certain posttranslational modifications (PTMs), such as the deamidation of glutamine (6), peptide trans-splicing (7), and citrullination (8), generate neoepitopes with improved binding to a given HLA molecule (5). However, whether such PTMs are directly recognized by TCRs has not been demonstrated.

The HLA-DR4 alleles are the strongest genetic association with the development of RA (9) and, in particular, HLA-DR4 allomorphs having a shared susceptibility epitope (SE) (residues 70 to 74) within the HLA-DR β chain (1012). This polymorphic SE motif defines the nature of the amino acid that can bind into the P4 pocket of the antigen-binding cleft. In the context of RA, the presence of anticitrullinated protein antibody (ACPA)–positive RA (70% of patients) is indicative of the presentation of citrullinated self-antigens by SE-bearing HLA allomorphs (1316). Namely, as positively charged residues are not accommodated within the P4 pocket, the PTM of arginine to citrulline enables HLA-DR4 binding (8, 17, 18). Mass spectrometry studies have established that citrullinated self-peptides are evident in the joint synovium of patients with RA (1923). Multiple citrullinated peptides have been shown to be B cell antigens in patients with RA, as is evidenced by the high level of ACPA positivity in patients with RA (1316). Numerous citrullinated peptides have also been shown to be T cell antigens in RA including α-enolase (2426), vimentin (24, 25), cartilage intermediate layer protein (24, 25), and fibrinogen (24, 25). Consistent with their role in disease, CD4+ T cells produce proinflammatory cytokines in response to citrullinated self-epitopes (27), citrulline-reactive CD4+ T cells are present in HLA-DR4 transgenic mice immunized with citrullinated self-antigens (28) and in the RA joint (26), and citrulline-reactive T helper 1 cells are elevated in patients with HLA-DR4+ RA (11, 24). Nevertheless, the molecular bases underpinning the relative interplay between HLA-DR4, citrullination, and the responding T cell repertoire remain unknown.

We have previously shown the structural basis for the association of the SE allomorphs, citrullination, and RA (8, 17). Here, we focus on how the TCR repertoire is shaped toward an epitope implicated in RA and associated comprehensive TCR-HLA-peptide structural information. Using HLA-DR4 tetramers bound to a citrullinated epitope of human fibrinogen β (Fibβ-74cit69-81), we examined the responding T cell repertoire in HLA-DRB1*04:01 transgenic mice. We show that biased TCR usage predominates in response to a citrullinated epitope, which is attributable to antigen-specific clonal expansion from the preimmune repertoire. Structural studies reveal that the TCR β chain plays a prominent role in interacting with the HLA-DR4 molecule, including the polymorphic SE motif, thereby revealing dual functionality for this motif. Whereas the P4-citrulline residue within the P4 pocket is not recognized by the TCR, the P2-citrulline residue of a double citrullinated epitope (Fibβ-72,74cit69-81) makes TCR-mediated contacts. This results in changes in the T cell repertoire induced by single and double citrullinated peptides and altered patterns of TCR reactivity. Hence, we demonstrate how a PTM not only affects HLA binding but also can play a direct role in shaping the nature of the T cell response. We also show remarkable convergence of the TCR repertoire toward the Fibβ-74cit69-81 epitope observed in the HLA-DR4 transgenic mouse and in human patients with RA. Accordingly, we provide important molecular insight into the T cell–mediated immune recognition and response to a citrullinated self-antigen implicated in RA, a prevalent autoimmune disease that affects about 1% of the human population.

RESULTS

Biased TCR repertoire in mice immunized with citrullinated fibrinogen epitope

We used the well-established HLA-DRB1*04:01 (HLA-DR4) transgenic mouse as an approach to investigate the immune response to the HLA-DR4–restricted citrullinated epitope, Fibβ-74cit69-81 (GGYRAXPAKAAAT, where X = Cit and the underline represents core binding region from P1 to P9 pockets of HLA-DR4) (24, 29). The Fibβ-74cit69-81 epitope was chosen because it has been shown to be a T cell (24) and B cell epitope (3033) associated with RA. The human and mouse peptides only differ at anchor position P9, with threonine in the mouse sequence. This minor difference has no effect on either peptide binding to HLA-DR4 or, as it is buried in the P9 pocket, TCR recognition. To characterize immune HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells, we immunized HLA-DR4 transgenic mice with the Fibβ-74cit69-81 peptide and, at 8 days after immunization, analyzed CD4+ T cells from draining lymph nodes (dLNs) for binding to the HLA-DRB1*04:01Fibβ-74cit69-81 tetramer. Whereas a small number of HLA-DR4Fibβ-74cit69-81–specific T cells could be identified in control-immunized mice, indicating the presence of HLA-DR4Fibβ-74cit69-81–specific cells in the naïve repertoire, there was a clear expansion of CD4+ tetramer+ T cells in response to peptide immunization, as evidenced by the increased number and frequency of CD44hi HLA-DR4Fibβ-74cit69-81–specific T cells relative to the control-immunized mice and lack of staining with control HLA-DR4 tetramers (Fig. 1, A and B, and fig. S1A). The immune response to the Fibβ-74cit69-81 peptide was also evident in the spleen and peripheral LNs (pLNs) at this time point (Fig. 1, A and B, and fig. S1A). This suggests that HLA-DRB1*04:01Fibβ-74cit69-81 tetramers reliably and specifically can detect an antigen-specific response in HLA-DR4 transgenic mice.

Fig. 1 HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells in naïve and immune HLA-DR4 transgenic mice.

(A) Representative HLA-DRB1*04:01Fibβ-74cit69-81 or DRB1*04:01HA306-318 tetramer staining on CD44hi CD4+ T cells from dLNs or CD4+ T cells from tetramer-enriched spleen and pLNs from HLA-DR4 mice immunized 8 days previously with Fibβ-74cit69-81 peptide or PBS emulsified in FCA (left) or from HLA-DRB1*04:01Fibβ-74cit69-81 tetramer-enriched spleen and pLNs from naïve HLA-DR4 and C57BL/6 mice (right). Gating strategies are shown in fig. S1A. Numbers in dot plots represent total number of CD4+ tetramer+ cells. (B) Frequency of CD4+ DRB1*04:01Fibβ-74cit69-81 (Fib74cit) or DRB1*04:01HA (HA) tetramer–positive CD4+ cells in the indicated tissues of individual Fibβ-74cit69-81 peptide-immunized (●) or PBS-immunized (○) HLA-DR4 mice and naïve HLA-DR4 (▲) and C57BL/6 (∆) mice. Symbols represent data from individual mice obtained in one to two independent experiments. Horizontal bars indicate means ± SD for each group. (C) HLA-DRB1*04:01Fibβ-74cit69-81–specific TCRαβ repertoires isolated from the pooled spleen and pLN or the dLN of individual Fibβ-74cit69-81 peptide-immunized HLA-DR4 mice (n = 2 mice for each, with 92 and 112 sequences, respectively), showing the frequency of individual clones and the TCR gene segment usage and CDR3 sequence for each clone. Asterisks indicate TCR in table S1 (M141, M134, M158-4, and M158-5, respectively) selected for further analysis. An (L)DW(G) motif in the CDR3β loop is indicated in red type, with underlined amino acids indicating non–template-encoded N region nucleotides. nd, not determined. (D) TCRdist clustering tree depicting the distance between TCRs within the immune DRB1*04:01Fibβ-74cit69-81–specific TCRαβ repertoire (53 clones from five mice). The thickness of the branches is proportional to the number of TCRαβ clones represented by those branches. The TCR logo containing the CDR3 sequence is shown on the left with residue height scaled by frequency. The colored bars underneath represent the source of the nucleotides that encode the logo, as depicted in key. The branches are colored according to TCR generation probability, as depicted in key. (E) Nearest neighbor (N-N) distance between each TCRαβ clone in a repertoire with its nearest neighbors, averaged across the entire repertoire (top), and the TCRdiv diversity measure, a modified version of the Simpson’s Diversity Index, incorporating receptor similarity and identity (bottom), for the non–antigen-specific total CD4+ TCRαβ repertoire, the naïve DRB1*04:01Fibβ-74cit69-81–specific TCRαβ repertoire (from fig. S2A), and the immune DRB1*04:01Fibβ-74cit69-81–specific TCRαβ repertoire (from D). For (D and E), data are analyzed independently of clonal abundance.

We next characterized αβTCR gene usage within the HLA-DRB1*04:01Fibβ-74cit69-81 tetramer+ populations, which revealed substantial clonal expansions in all immune repertoires (Fig. 1C). Two repertoires isolated from the spleen and pLN of distinct mice showed a single dominant clonal expansion that comprised 80% and 55% of the immune repertoire, whereas another two repertoires isolated from the dLN also contained dominant clonal expansions. All four of the immune repertoires showed a preference for TRBV4, TRBV13-1, and TRAV6 usage relative to the background CD4+ T cell repertoire (fig. S2A), with preferential pairing of the TRBV4+ and TRAV6+ chains (Fig. 1D). All four repertoires also exhibited an “(L)DW(G)” motif within the CDR3β loop (Fig. 1, C and D). The L residue of the complementarity-determining region 3β (CDR3β) loop is encoded by nontemplated N region nucleotides, whereas the DW(G) sequence was encoded by rearranged TRBD2 gene, indicating that selection of this motif occurred independently of the preferred TRBV4.

Thus, the CD4+ T cell immune response to the Fibβ-74cit69-81 epitope in HLA-DRB1*04:01 transgenic mice preferentially comprise a subset of T cell clones with defined sequence characteristics that are shared between different mice.

The naïve epitope-specific T cell repertoire

To determine whether this observed TCR bias preexisted within the naïve epitope-specific T cell repertoire or was due to preferential expansion during the immune response, we isolated HLA-DRB1*04:01Fibβ-74cit69-81 tetramer+ T cells from the preimmune repertoire of naïve mice (Fig. 1, A and B, and fig. S1A). HLA-DRB1*04:01Fibβ-74cit69-81–specific T cells identified in the pooled spleen and major LNs of naïve HLA-DR4 mice corresponded to a frequency of about 1.1 per million CD4+ T cells (Fig. 1B), which is comparable with observed naïve epitope-specific frequencies (0.2 to 10 per million CD4+ T cells) (34). Subsequent TCRαβ sequence analysis revealed that individual TCRαβ clones were detected only once (fig. S2B), indicating that the large immune clones were a result of antigen-driven proliferation. Detailed analysis of the naïve and immune HLA-DR4Fibβ-74cit69-81–specific TCRαβ repertoires, compared with a non–antigen-specific CD4+ T cell repertoire from HLA-DR4 mice, was performed using the TCRdist algorithm (35). Although almost completely absent from the non–antigen-specific background repertoire, prominent usage of the TRBV4-CDR3β (L)DW(G) motif, previously found to be selectively expanded in the immune repertoire, was also observed in the naïve repertoire (fig. S2, B and C). However, the immune repertoire was distinct from the naïve repertoire in that it showed groupings of highly similar TCRs, whereas the TCRs in the naïve repertoire were comparatively distinct from one another (Fig. 1D and fig. S2C). Analysis of the average “distance” across TCRs within a repertoire showed a progressive decrease from the non–antigen-specific TCRαβ repertoire to the naïve and immune HLA-DR4Fibβ-74cit69-81–specific TCRαβ repertoires (Fig. 1E, top), demonstrating a progressive focusing of the repertoire toward TCRs with more similar characteristics. This was further reflected by the progressively reduced diversity across the TCR repertoires (Fig. 1E, bottom). Last, modal CDR3β lengths of immune HLA-DR4Fibβ-74cit69-81–specific T cells also showed a focusing from the naïve to the immune repertoire (fig. S2D), further demonstrating the selective usage of a subset of HLA-DRB1*04:01Fibβ-74cit69-81–specific T cells in the immune repertoire, suggestive of high-avidity HLA-DRB1*04:01Fibβ-74cit69-81 recognition by the TCR.

Analysis of the entire dataset, including clonal expansions, showed a further increased frequency of sequences using either TRBV4+ or TRAV6D-6 gene segments (fig. S3A), containing the DW motif (fig. S3B) and altered CDR3α and CDR3β loop length (fig. S3C) in the immune repertoire compared with the naïve repertoire, indicating that epitope-specific clones with these characteristics are preferentially expanded after antigen encounter.

Crystal structure of a naïve TCR–HLA-DRB1*04:01Fibβ-74cit69-81 complex

Next, we expressed, refolded, and purified four TRBV4+ TCRs (G08, C06, E05, and F08; table S1) from the naïve T cell repertoire and determined, using surface plasmon resonance (SPR) (36), that their steady-state binding affinities (KD) for HLA-DRB1*04:01Fibβ-74cit69-81 were 156.4, 12.5, 127.7, and 81.3 μM, respectively (Fig. 2A). To understand how TRBV4+ TCRs engage HLA-DRB1*04:01Fibβ-74cit69-81, we determined the structure of the G08 TCR–HLA-DRB1*04:01Fibβ-74cit69-81 ternary complex (Fig. 3A and table S2). The G08 TCR docked at an angle of ~57° across the central region of the antigen-binding cleft (Fig. 3A). The total buried surface area (BSA) of G08 TCR was 1700Å2, comparable with BSA values observed for TCR–pMHC (peptide bound to major histocompatibility complex) II structures previously determined (table S3) (37). The TRBV4-encoded TCR β chain dominated the interactions with HLA-DRB1*04:01Fibβ-74cit69-81 (table S3), with 80% of the total BSA at this interface. Here, the CDR3β loop served as the major contributor in interacting with HLA-DR4Fibβ-74cit69-81, with 42% BSA compared with ~17% each for CDR1β and CDR2β loops, whereas the CDR3α loop contributed ~15% BSA (Fig. 3A and table S3). Leu27β, Gly28β, and His29β from the CDR1β loop made multiple contacts with Tyr60, Gln64, and Lys65 in the HLA-DR4 β chain (Fig. 3B), whereas Asn57β, Asn58β, and Gln60β from the CDR2β loop made polar-mediated interactions with Ala64, Val65, and Ala68 from the HLA-DR α chain (Fig. 3C and table S4). The CDR3β loop, which contained the (L)DW(G) motif, contacted the HLA-DR4 β chain residues, Gln64, Lys65, Asp66, Leu67, and Gln70, as well as the Gly58, Ala61, and Asn62 from the HLA-DR4 α chain (Fig. 3D and table S4). The (L)DW(G) motif contacted the shared epitope of HLA-DR4 (70QKRAA74), which is strongly associated with susceptibility to RA (odds ratio, 4.44) (Fig. 3E) (11). Whereas the P4-Cit residue did not contact the TCR, Gln70 made H-bonds with Asp109β and van der Waals interactions with Leu108β in the CDR3β loop (Fig. 3E and table S4). Together, the ternary structure provided a molecular basis for biased TCR β chain usage toward HLA-DRB1*04:01Fibβ-74cit69-81 and showed that the SE of HLA-DR4 not only represents a critical feature for P4-Cit binding to the MHC but also is a key determinant for TCR recognition.

Fig. 2 Affinity analysis of TCR–HLA-DRB1*04:01Fibβ-74cit69-81/Fibβ-72,74cit69-81 interactions.

For KD determination, all data were derived from two or more independent experiments and combined for each TCR, and a single ligand-binding model was used for curve fitting. (A) Binding of TCRs G08, C06, E05, F08, M134, M141, M158-4, and M158-5 to HLA-DRB1*04:01Fibβ-74cit69-81. (B) Binding of TCRs G08, C06, E05, F08, M134, M141, M158-4, and M158-5 to HLA-DRB1*04:01Fibβ-72,74cit69-81. To control for nonspecific binding, HLA-DRB1*04:01CLIP was immobilized in the reference flow cell.

Fig. 3 TRBV4+ TCR recognition of HLA-DRB1*04:01Fibβ-74cit69-81.

(A) Top: Overall cartoon representation of naïve TCR (G08) and immune TCR (M134 and M141) complexed to HLA-DRB1*04:01Fibβ-74cit69-81. Bottom: Surface representation of TCR footprints and TCR docking. The CDR loops 1α, 2α, and 3α are highlighted in cyan, violet, and light green color, whereas 1β, 2β, and 3β are colored in blue, purple, and dark green, respectively. The framework residues are colored in beige. TCR footprint colors are in accordance with the nearest TCR contact residue. The Vα and Vβ center-of-mass positions are shown in red and blue spheres, respectively, and connected via a black line. The HLA-DR04:01 α and β chains are colored in white and brown, respectively, whereas the peptide is shown in orange sticks. Detailed interactions between (B) CDR1β, (C) CDR2β, (D) CDR3β with HLA-DRB1*04:01, and (E) P4-shared epitope are shown.

Structures of the immune TCR–HLA-DRB1*04:01Fibβ-74cit69-81 complex

We then investigated HLA-DRB1*04:01Fibβ-74cit69-81 recognition by TCRs from the Fibβ-74cit69-81–immunized T cell repertoire. Two TRAV6+TRBV4+ TCRs (M134 and M141) and two TRBV4 TCRs (M158-5 and M158-4; table S1) from the immune repertoire were expressed, refolded, and purified. As judged by SPR analysis, the steady-state binding affinity (KD) of the M134, M141, M158-5, and M158-4 TCRs for HLA-DRB1*04:01Fibβ-74cit69-81 were 9.7, 18.6, 9.3, and 8.5 μM, respectively, suggesting that TRAV6+TRBV4+ TCRs from the immune repertoire tend to have a higher affinity compared with those TRAV6+TRBV4+ TCRs isolated from the naïve repertoire (Fig. 2A). These TCRs did not recognize the enolase epitope presented by HLA-DR4, demonstrating their specificity to HLA-DRB1*04:01Fibβ-74cit69-81. We then determined the structures of two immune TRBV4+ TCRs in complex with HLA-DR4Fibβ-74cit69-81 and compared it with the naïve G08 TRBV4+ TCR HLA-DR4Fibβ-74cit69-81 ternary complex (Fig. 3A and table S2). The overall docking topologies of the M134 and M141 TCRs were similar to that of the naïve G08 TCR (G08), albeit with slightly higher total BSA of 1950 and 2050 Å2, respectively (Fig. 3A and table S3). This was due to greater contacts mediated by the TCR α chain of the M134 and M141 TCRs, which, in turn, was reflective of the differences in TCR α chain usage across the preimmune and immune TCRs. Namely, for the M134 and M141 TCRs, Ile57α from the CDR2α loop and Ser111α and Phe112α from the CDR3α loop made multiple van der Waals interactions with Glu55, Gln57, Gly58, and Ala61 from the HLA-DR4 α chain, whereas these contacts were absent in the preimmune TCR ternary complex (fig. S4 and tables S4 to S6). Despite the different TCR α chain usage, the contacts mediated by the germ line–encoded CDR1β and CDR2β loops were conserved across the three TCRs (fig. S4 and tables S4 to S6). Moreover, the interactions mediated by the 108LDW110 motif in CDR3β loop were conserved across the naïve and immune TCR ternary complexes (fig. S4 and tables S4 to S6), thereby underpinning the importance of the selective expansion of the biased TCR β chain usage in interacting with HLA-DRB1*04:01Fibβ-74cit69-81.

Energetic basis underpinning TRBV4+ TCR recognition of HLA-DR4Fibβ-74cit69-81

Next, we established the principal energetic determinants governing TRBV4+ TCR recognition of HLA-DR4Fibβ-74cit69-81 by undertaking alanine-scanning mutagenesis of the M134 TCR and measuring the impact of the mutants on the affinity of the interaction with HLA-DR4Fibβ-74cit69-81 using SPR (Fig. 4A and fig. S5). M134 TCR residues selected for mutagenesis were chosen on the basis of TCR-pMHC interatomic contacts within the corresponding ternary structure. As expected, the control mutation had no effect on the affinity of the interaction (Ser112βAla) (Fig. 4A). Mutating the following residues within the TCR α chain (Ile57α, Ser111α, and Phe112α) and TCR β chain (Gly28β, Asn58β, Gln60β, Lys84β, Ala111β, and Q113β) had limited impact on the affinity of the interaction (Fig. 4A). In contrast, mutation of residues within the CDR1β (His29β and Asp30β) and CDR2β loops (Asn57β), as well as the (L)DW(G) motif in the CDR3β loop of the M134 TCR, showed deleterious effects to HLA-DR4Fibβ-74cit69-81 recognition (Fig. 4, A and B). Mutation of Ser109α, His29β, Asp30β, and Trp110β, residues which contacted the Fibβ-74cit69-81 epitope, almost completely disrupted HLA-DR4Fibβ-74cit69-81 recognition (Fig. 4, A and B). Mutation of Asp109β, which hydrogen-bonded to Gln70 within the SE, also abrogated the HLA-DR4Fibβ-74cit69-81 interaction. Mutation of other residues in the CDR3β loop (Asn114β and Tyr117β) showed more than 20-fold reduced affinity as compared with the wild-type M134 TCR (Fig. 4A). These energetically critical residues of the M134 TCR bearing the L(DW)G motif formed a core hotspot at the epitope-binding groove around the shared epitope (Fig. 4B).

Fig. 4 Specificity determinants of the interaction with HLA-DRB1*04:01Fibβ-74cit69-81.

(A) Effect of TCR point mutations at pMHC II interface. The SPR experiments were performed in duplicate (n = 2). The y axis represents the fold of affinity of mutant TCRs as compared with wild-type TCR. The x axis shows the position of M134 TCR mutants, and the impact of each mutation was classified as no effect (<3-fold affinity decrease, colored blue), moderate (3- to 5-fold affinity reduction, yellow), severe (5- to 10-fold affinity reduction, orange), or critical (>10-fold affinity decrease or no binding, red). (B) Energetic footprint of M134 TCR on the HLA-DRB1*04:01Fibβ-74cit69-81 complex. Energetic contribution was determined by SPR, and the impact of each mutation was colored corresponding to (A). (C) Effect of HLA-DR4 mutants at TCR interface. The x axis shows the position of HLA-DR4 mutants; the blue bar represents naïve TCR (G08), and magenta and deep teal bars indicate immunized TCR (M134 and M141, respectively). (D) 293T cells transiently cotransfected with a HLA-DRB1*04:01Fibβ-74cit69-81–specific TCR (A08, E04, G08, M141, 182-1, 182-2, or 182-3) and CD3γδεζ were stained with either HLA-DRB1*04:01Fibβ-74cit69-81 tetramer (orange), HLA-DRB1*04:05Fibβ-74cit69-81 tetramer (red), or control HLA-DRB1*04:01α-enolase-15cit10-22 tetramer (charcoal). Detailed interactions of Fibβ-74cit peptide against (E) G08 TCR, (F) M134 TCR, and (G) M141 TCR, respectively. The CDR loops 1α, 2α, and 3α are highlighted in cyan, violet, and light green color, whereas 1β, 2β, and 3β are colored in blue, purple, and dark green, respectively. Black dashes denote hydrogen bond; beige dashes denote van der Waals interactions. All amino acids are indicated in single-letter abbreviations.

Impact of the SE motif on TCR recognition

To further explore the role of the SE in TCR recognition, we mutated the Gln70 residue to either aspartic acid or alanine. The Gln70Asp substitution completely abrogated the interaction with all three TCRs—G08, M134, and M141—whereas the Gln70Ala mutant caused a ~10-fold reduced affinity to M134 and M141 TCR as compared with wild type (Fig. 4C and fig. S6). This is consistent with our structural findings, where Gln70 interacts with Asp109β from the (L)DW(G) motif. Overall, the mutational analyses, together with the structural observations, highlight the importance of defined residues within the TCR β chain in interacting with the HLA-DR4Fibβ-74cit69-81 complex, which includes the SE itself as a key TCR recognition determinant.

Given that a number of HLA-DR allomorphs have the SE (9) and the observation that the SE directly interacted with the TCR, we next examined how SE polymorphisms affected TCR recognition. The HLA-DRB1*04:05 allomorph, which is significantly associated with RA susceptibility (odds ratio, 4.22) (9) and common in Southeast Asian populations, differs from DRB1*04:01 only at positions 57 (Asp➔Ser) and 71 (Lys➔Arg) in the P10 and P4 pocket, respectively (17). To determine whether HLA-DR4Fibβ-74cit69-81–specific TCRs could recognize the Fibβ-74cit69-81 peptide bound to the HLA-DRB1*04:05 allomorph, we stained TCR transfectants separately with HLA-DRB1*04:01Fibβ-74cit69-81 and HLA-DRB1*04:05Fibβ-74cit69-81tetramers (Fig. 4D). Of the seven TCRs studied, only one (182-2) showed substantial cross-reactivity with HLA-DRB1*04:05Fibβ-74cit69-81, with two others (A08 and E04) showing marginal cross-reactivity at the highest TCR expression levels (Fig. 4D). The capacity to cross-react on the HLA-DRB1*04:05 allomorph was independent of whether the TCRs were identified from the pre- or postimmune TCR repertoire, with two of three tested naïve TCRs (A08 and E04) showing minimal HLA-DRB1*04:05Fibβ-74cit69-81 reactivity and, of four immune TCRs, only 182-2 TCR showing high cross-reactivity (Fig. 4D). Of similarly high-affinity TCRs (M134, M158-4, and M158-5) at KD ~ 10 μM, M158-4 and M158-5 TCRs showed cross-reactivity, whereas M134 TCR did not (fig. S7). The TCRs that showed cross-reactivity lacked the (L)DW(G) motif in the CDR3β loop, and all used TRAV6-7, consistent with the affinity analysis using SPR, in which the M158-4 and M158-5 TCRs showed KD at 4.7 and 27.7 μM, respectively (table S1 and fig. S7). These data suggest limited cross-reactivity of HLA-DRB1*04:01Fibβ-74cit69-81–specific TCRs with the HLA-DRB1*04:05 allomorph bound to the same epitope. Thus, polymorphisms within the SE motif not only shape binding of citrullinated peptides but also affect TCR recognition.

Specificity determinants underpinning Fibβ-74cit69-81 recognition

To understand the specificity toward the Fibβ-74cit69-81 epitope, we compared the structures of all three TRAV6+TRBV4+ TCR ternary complexes (Fig. 4, E to G). The interactions that are conserved across these TCRs are largely attributable to the P2, P5, and P8 residues of the Fibβ-74cit69-81 epitope, where the CDR3α and CDR3β loops played a critical role in peptide-mediated interactions (Fig. 4, E to G, and tables S4 to S6). Here, Ser109α in the CDR3α loop hydrogen bonded to P2-Arg, whereas Asp109β and Trp110β in the CDR3β loop made extensive van der Waals interactions with P5-Pro, emphasizing the importance of the (L)DW(G) motif in both HLA-DR4 and peptide contacts (Fig. 4, E to G, and tables S4 to S6). The P8-Ala formed a main chain hydrogen bond and multiple van der Waals interactions with Asp30 and Asn58 in the CDR1β and CDR2β loops, respectively. Other framework residues, including Lys84β, also interacted with the C-terminal P8-Ala of the Fibβ-74cit69-81 peptide (Fig. 4, E to G, and tables S4 to S6). Together, these findings indicate highly conserved interactions between TRAV6+TRBV4+ TCRs and residues spanning the length of the Fibβ-74cit69-81 peptide.

Impact of double citrullination on TCR recognition

The Fibβ-74cit69-81 peptide contains two potential citrullination sites at positions Arg72 and Arg74. Two members of the peptidyl arginine deiminase (PAD) enzyme gene family found in monocytes and neutrophils, PAD2 and PAD4, are considered the main drivers of citrullination in the inflamed synovium of patients with RA (3739). These enzymes have different substrate specificities and have been demonstrated to differentially citrullinate positions 72 and 74 of fibrinogen β (40), suggesting that four different isoforms could coexist: Fibβ-69-81 (not citrullinated), Fibβ-72cit69-81, Fibβ-74cit69-81, and Fibβ-72,74cit69-81. ACPA-positive sera from patients with RA can recognize individually citrullinated Fibβ-74cit69-81 or Fibβ-72cit69-81 or doubly citrullinated Fibβ-72,74cit69-81 (41), suggesting the presence of all three modified epitopes in patients with RA, which is supported by analysis of the citrullinome in the joint synovium of patients with RA using mass spectrometry (21). Because CD4+ T cells from HLA-DR4 mice and patients with HLA-DR4+ RA can recognize a doubly citrullinated peptide (24), we investigated whether TCRs reactive to the single citrullinated HLA-DRB1*04:01Fibβ-74cit69-81 could cross-react to the double citrullinated HLA-DRB1*04:01Fibβ-72,74cit69-81. We selected four TCRs (A08, E04, G08, and M141), two of which have the CDR3β (L)DW(G) motif (table S1), for transfection into human embryonic kidney (HEK) 293T cells (Fig. 5A). The specificity of the TCRs was confirmed by binding of the HLA-DRB1*04:01Fibβ-74cit69-81 tetramer (Fig. 5A). Three of the four TCR transfectants also bound the HLA-DRB1*04:01Fibβ-72,74cit69-81 tetramer, albeit binding was reduced to varying extents compared with that observed in the single citrullinated epitope (Fig. 5A). We also examined the impact of double citrullination on the affinity of eight TCRs toward the HLA-DRB1*04:01Fibβ-72,74cit69-81 complex using SPR. Three TCRs (E05, KD = 104.9 μM; F08, KD = 101.9 μM; and M158-4, KD = 13.4 μM) did not display a preference for the single or double citrullinated epitopes. However, five of eight TCRs cross-reacted to the double citrullinated fibrinogen presented by HLA-DR4 (HLA-DRB1*04:01Fibβ-72,74cit69-81) with two to eightfold weaker affinity [G08 (KD = 654.7 μM), C06 (KD = 51.4 μM), M134 (KD = 22.3 μM), M141 (KD = 155.7 μM), and M158-5 (KD = 60.8 μM)] in comparison with the single citrullinated fibrinogen epitope (Fig. 2B). Accordingly, the presence of a second citrullination site affected TCR recognition.

Fig. 5 Impact of the double citrullination on TCR repertoire and binding.

(A) 293T cells transiently cotransfected with a HLA-DRB1*04:01Fibβ-74cit69-81–specific TCR (A08, E04, G08, and M141) and CD3γδεζ were stained with either HLA-DRB1*04:01Fibβ-74cit69-81 tetramer (40), HLA-DRB1*04:01Fibβ-72,74cit69-81 tetramer (blue), or control HLA-DRB1*04:01α-enolase-15cit10-22 tetramer (charcoal). (B) Representative HLA-DRB1*04:01Fibβ-74cit69-81 and HLA-DRB1*04:01Fibβ-72,74cit69-81 tetramer costaining on CD4+ T cells from dLNs of HLA-DR4 mice immunized 8 days previously with either Fibβ-74cit69-81 or Fibβ-72,74cit69-81 peptide. Right plots represent expansions of individual clones, determined via index sorting, with TCR gene segment usage, CDR3 sequence, and frequency shown for each clone. Asterisks indicate TCR shown in (C) and table S1 (182-1, 182-2, 182-3, 180-1, and 180-2, respectively). An (L)DW(G) motif in the CDR3β loop is indicated in red type, with underlined amino acids indicating non–template-encoded N region nucleotides. (C) 293T cells transiently cotransfected with TCR from selected clones in (B) (identified by symbols) and CD3γδεζ were stained with either HLA-DRB1*04:01Fibβ-74cit69-81 tetramer (40), HLA-DRB1*04:01Fibβ-72,74cit69-81 tetramer (blue), or control HLA-DRB1*04:01α-enolase-15cit10-22 tetramer (charcoal). (D) TCRdist clustering tree depicting the distance between TCRs within the immune DRB1*04:01Fibβ-72,74cit69-81–specific TCRαβ repertoire (n = 4 with 51 unique clones). The thickness of the branches is proportional to the number of TCRαβ clones represented by those branches. The TCR logo containing the CDR3 sequence is shown on the left, with residue height scaled by frequency. The colored bars underneath represent the source of the nucleotides that encode the logo, as depicted in key. The branches are colored according to TCR generation probability, as depicted in key. (E) Comparison of the occurrence of the (L)DW(G) motif in the non–antigen-specific (n = 2 and 91 clones), naïve (n = 2 and 79 clones) and immune (n = 5 and 53 clones) HLA-DRB1*04:01Fibβ-74cit69-81–specific, and immune HLA-DRB1*04:01Fibβ-72,74cit69-81–specific (n = 4 and 51 clones) repertoires. Symbols represent individual mice. Horizontal bars indicate means ± SD for each repertoire.

Impact of double citrullination on TCR repertoire

To determine the extent to which polyclonal TCRs induced in response to the single citrullinated Fibβ-74cit69-81 peptide could cross-react with double citrullinated Fibβ-72,74cit69-81 peptide, we immunized mice with either the Fibβ-74cit69-81 or the Fibβ-72,74cit69-81 peptide and determined reactivity at day 8 by tetramer staining (Fig. 5B). Distinct populations of epitope-reactive CD4+ T cells were observed (Fig. 5B), namely, those that bound to both HLA-DRB1*04:01Fibβ-74cit69-81 and HLA-DRB1*04:01Fibβ-72,74cit69-81 tetramers to varying extents and those that bound only the tetramer presenting the immunizing peptide (Fig. 5, B and C). TCR repertoire usage induced against the single and double citrullinated peptides showed that although T cells induced by immunization with the HLA-DRB1*04:01Fibβ-74cit69-81 peptide preferentially used TRBV4+ TCRs with the LDW motif in the CDR3β loop, the HLA-DRB1*04:01Fibβ-72,74cit69-81–induced TCRs showed no such preferential TCR β chain usage (Fig. 5, B, D, and E). Accordingly, the presence of a second citrullination site affected TCR repertoire usage.

To understand the impact of the double citrullination on TCR recognition, we determined the structures of the naïve G08 TCR and immune M134 and M141 TCRs complexed to HLA-DRB1*04:01Fibβ-72,74cit69-81 (table S2). These ternary complexes exhibited a high degree of conservation in the modes of docking when compared with the corresponding single citrullinated epitope, indicating that the presence of the second citrullination site does not cause large-scale adjustments at the TCR–HLA-DR4 interface (fig. S8 and tables S7 to S9). Instead, subtle changes in the placement of some of the CDR loops were induced by the presence of P2-Cit. Within the single citrullinated Fibβ-74cit69-81 structure, P2-Arg pointed toward the N-terminal region of the epitope, forming hydrogen bonds with Ser109α in CDR3α and van der Waals interactions with residues within the CDR1β and CDR3β loops (Fig. 6, A and B, and tables S7 to S9). Within the TCR–HLA-DR4Fibβ-72,74cit69-81 ternary complex, the P2-Cit residue was reoriented 62.5° such that it formed contacts with the P4-Cit and reduced the extent of contacts with the TCR α chain, including loss of hydrogen bonding contacts with Ser109α from the CDR3α loop (Fig. 6, A and C, and tables S7 to S9). These reduced TCR-mediated contacts were consistent with the lower affinity for TCRs, recognizing the double citrullinated epitope presented by HLA-DRB1*04:01. Hence, citrullination not only permits self-antigens binding to HLA-DR allomorphs having the SE but also can directly affect TCR repertoire and TCR recognition.

Fig. 6 Structural changes induced by double citrullinated epitope of Fibβ-72,74cit69-81 presented by HLA-DRB1*04:01.

(A) Overlaid structure of single and double citrullinated DRB1*04:01 complexed with M134 TCR. Detailed interactions between CDR loops and P2 citrulline in Fibβ-74cit (orange sticks) and Fibβ-72,74cit (yellow sticks) presented by HLA-DRB1*04:01 (brown cartoon). The CDR loops 1α, 2α, and 3α are highlighted in cyan, violet, and light green color, whereas 1β, 2β, and 3β are colored in blue, purple, and dark green, respectively. Detailed interactions between M134 CDR loops and (B) P2 arginine in Fibβ-74cit peptide or (C) P2 citrulline in Fibβ-72,74cit peptide. Black dashes denote hydrogen bond; beige dashes denote van der Waals interactions. All amino acids are indicated in single-letter abbreviations. (D and E) Detailed interactions of G08 (D) and M141 (E) TCR with double citrullinated fibrinogen β epitope (Fibβ-72,74cit) presented by HLA-DRB1*04:01. Black dashes denote hydrogen bond; beige dashes denote van der Waals interactions. All amino acids are indicated in single-letter abbreviations.

Characterization of HLA-DRB1*04:01Fibβ-74cit69-81–specific TCRs from RA+ and healthy humans

To determine whether HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells could be identified in HLA-DR4+ humans, we single-cell sorted peripheral blood mononuclear cells (PBMCs), enriched using the HLA-DRB1*04:01Fibβ-74cit69-81 tetramer, from two ACPA+ RA+ individual and two healthy controls. Four and 5 epitope-specific cells were identified from the peripheral blood of the RA+ individuals, and 7 and 10 cells were identified from each of the healthy individuals (Fig. 7A and fig. S1B). TCR sequencing of the epitope-specific cells yielded two, three, four, and two paired TCRα and β sequences from each sample, respectively (Fig. 7, A and B). Three of the five clones identified from the RA+ individuals and three of the four clones identified from the healthy individuals exhibited the (L)D(W) motif that was selectively expanded in response to peptide immunization in the HLA-DR4 transgenic mice. Moreover, 6 of the 11 TCR sequences, including those from both the patients with RA and healthy HLA-DR4+ individuals, used the TRBV3-1 gene segment. The human TRBV3-1 gene segment is the human ortholog of the mouse TRBV4 gene segment that was found to be expanded in the HLA-DR4 transgenic mice after peptide immunization. The utilization of these TCR characteristics in human HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells is particularly intriguing considering that (i) they arise independently of one another [i.e., the (L)D(W) motif is not encoded by the TRBV3-1 gene element]; (ii) unlike the mouse D gene element, which encodes the DW(G) motif (with the preceding L being N region encoded), human Dβ genes cannot encode DWG, with both the LD and DW sequences here encoded by N region nucleotide additions; and (iii) sequence alignment of the mouse and human V gene segments shows that seven of the eight key TRBV4 encoded residues involved in contacting the HLA-DRB1*04:01Fibβ-74cit69-81 complex are conserved in the human TRBV3-1 gene segment (Fig. 7, C and D). The specificity of the tetramer staining for the human TCRs was confirmed using HEK293T cells transduced with HLA-DRB1*04:01Fibβ-74cit69-81 and HLA-DRB1*04:01Fibβ-72,74cit69-81 restricted TCRs as well as an HLA-DRB1*04:01α-enolase-15cit10-22–restricted TCR as a control (Fig. 7E). Collectively, these data clearly indicate antigen-driven selection of notably similar characteristics between mouse and human TCRs specific for HLA-DRB1*04:01Fibβ-74cit69-81, further strengthening the relevance of our findings to human disease.

Fig. 7 HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ TCR sequences from patients with RA and healthy controls contain the (L)D(W) motif encoded by nongermline residues.

(A) HLA-DRB1*04:01Fibβ-74cit69-81 tetramer staining on CD4+ T cells from PBMC of two HLA-DRB1*04:01+ RA+ patients and two HLA-DRB1*04:01+ healthy controls. Dot plots show CD14CD19live CD3+ cells. Gating strategy is shown in fig. S1B. (B) TCR gene segment usage and CDR3 sequences are shown for individual tetramer+ cells indicated with arrows. The (L)D(W) motif in the CDR3β loop is indicated in bold type, with non–template-encoded N region nucleotides colored in red. (C) Alignment of mouse (M141) and human (RA1.B04) HLA-DRB1*04:01Fibβ-74cit69-81–specific TCRβ chain sequences. CDR regions are shown in boxes; red highlighting indicates key contacts between mouse TCRβ chain and HLA-DRB1*04:01Fibβ-74cit69-81, as observed from structure. Asterisks indicate conservation in key residues between mouse and human V gene sequences. The gray shaded box indicates V gene and J gene encoded segments. The green shading indicates D gene encoded segments. Underlined residues are N region nucleotides. (D) Detailed interactions of mouse-human conserved residues and HLA-DRB1*04:01Fibβ-74cit69-81. The CDR loops 1α, 2α, and 3α are highlighted in cyan, violet, and light green color, whereas 1β, 2β, and 3β are colored in blue, purple, and dark green, respectively. Black dashes denote hydrogen bond; beige dashes denote van der Waals interactions. All amino acids are indicated in single-letter abbreviations. (E) 293T cells transiently cotransfected with a human HLA-DRB1*04:01Fibβ-74cit69-81–specific TCR (RA1.B04, HC1.A01, HC1.A04, and HC2.D04), a mouse HLA-DRB1*04:01Fibβ-74cit69-81–specific TCR (M141), or a human HLA-DRB1*04:01α-enolase-15cit10-22–specific TCR (RA2.7) and human or mouse CD3γδεζ were stained with either HLA-DRB1*04:01Fibβ-74cit69-81 tetramer (orange), HLA-DRB1*04:01Fibβ-72,74cit69-81 tetramer (blue), or HLA-DRB1*04:01α-enolase-15cit10-22 tetramer (charcoal) and antibody to mouse TCR (mTCR) or human CD3 (hCD3).

DISCUSSION

Our study provides several key conceptual advances in the context of T cell–mediated autoimmunity. In the context of ACPA+ RA, it was established that citrullination of self-epitopes enables binding into the P4 pocket of HLA-DR allomorphs having the SE (8, 10, 11). Here, we show that the SE also represents a key TCR recognition determinant, and this duality of function substantially deepens our understanding of the role of this HLA-DR4 SE in RA. Given the central location of the SE motif within the HLA-DR4 antigen-binding cleft, we suggest that the dual functionality of the SE motif observed here will be a general feature of TCR recognition of HLA-DR4 presenting different citrullinated self-epitopes. However, the nature of the interatomic interactions between the TCR and the SE motif is likely to vary as a function of the citrullinated self-epitope presented because of specific structural features inherent in any given epitope.

Polymorphisms within the SE across distinct HLA-DR4 molecules also affected TCR reactivity, thereby highlighting the exquisite fine specificity that the SE motif imparts to this T cell–mediated response. The impact of the lack of cross-reactivity between HLA-DR*04:01/05 is likely attributable to the Lys71Arg polymorphism within the SE. Namely, Arg71 would alter the conformation of P7-Lys, which would lead to a loss of contact with Asp109β from the LDW motif within the CDR3β loop. Consistent with this, the M158-4 and M158-5 TCRs, which lack the LDW motif can cross-react to HLA-DR*04:05 when presenting the Fibβ-74cit69-81 epitope. The fine specificity of the T cell response to the citrullinated fibrinogen epitope was also evident. Namely, citrullination at the second site within the fibrinogen epitope (Fibβ-72,74cit69-81) affected the extent of TCR cross-reactivity, which was attributable to the P2-Cit residue directly affecting TCR contacts. These observations also underscore the duality of citrullination in that it not only imparts binding to the HLA-DR4 molecules with the SE motif but also can indirectly or directly shape TCR repertoire usage, TCR avidity, and TCR cross-reactivity. The differing extent of citrullination of fibrinogen will also likely affect protease susceptibility and the liberation of neo–self-epitopes (8).

Using an HLA-DR4 transgenic mouse model, we show that reproducible patterns of TCR gene usage specifically confer recognition of HLA-DR4 tetramers presenting a citrullinated epitope. Whereas the biased TRBV4+ TCR gene usage, together with a motif within the CDR3β loop, was observed in the preimmune repertoire, the selective clonal expansion of these T cells from the naïve T cell repertoire indicated high-avidity recognition by the TCR. We also identified several DRB1*04:01Fibβ-74cit69-81 tetramer+ CD4+ T cells in human patients with HLA-DR4+ RA and healthy controls. Given the chronic nature of RA, the autoreactive T cell population concentrates at the site of inflammation and source of autoantigen. In RA, local enrichment of candidate antigen-specific T cells in synovial fluid as compared with peripheral blood has been demonstrated for both uncitrullinated and citrullinated autoantigens (26, 42). Consequently, because we were using PBMCs as the source tissue, the frequency of autoreactive T cells was expected to be extremely low. Nevertheless, we were able to establish that a number of these tetramer+ TCRs used the human equivalent (TRBV3-1) of the mouse TRBV4, and most of TCR residues involved in contacting the HLA-DR4–FibCit complex in mice are conserved in the human V gene segment. Moreover, many of the human TCR sequences exhibited the same (L)D(W) motif in the CDR3β loop, as observed in the immune T cell repertoire in the HLA-DR4 transgenic mouse, which is a strong indication of the importance of this motif in driving the specific interactions with the HLA-DR4Fibβ-74cit69-81 complex. Accordingly, there are notable parallels between the TCR repertoire in the HLA-DR4 transgenic mice immunized with an epitope implicated in RA pathogenesis and that of the TCR repertoire in HLA-DR4+ individuals. That we see a distinct TCR repertoire for a given citrullinated epitope and that TCRs within this repertoire do not cross-react with other citrullinated epitopes are not unexpected. Distinct T cell reactivities critical in autoimmune and inflammatory pathogenesis are exemplified in celiac disease, in which the two immunodominant peptides, DQ2.5-glia-α1 and DQ2.5-glia-α2, bound to HLA DQ2.5 elicit distinct TCR usage patterns, with both populations involved in disease pathogenesis (43). We suggest that the same situation arises in RA, and the data presented here form the foundation for future analyses of TCR repertoires from distinct epitope specificities in RA.

In general, PTMs are considered to solely influence HLA-binding propensity (5, 44). Our previous work established the structural basis for the binding of citrullinated epitopes to HLA-DRB1 SE allomorphs (8, 17). Here, we show that PTMs can also directly modulate T cell recognition. This has important implications for understanding how neoepitopes can shape the T cell repertoire and lead to autoimmunity. Furthermore, it highlights that, to delineate how TCR gene usage can lend itself to predicting epitope specificity, one needs to factor in the potential impact of such PTMs.

MATERIALS AND METHODS

Study design

The aim of this study was to determine the T cell response of citrullinated antigens presented by HLA-DR4, one of a group of allomorphs containing the “shared epitope” that is associated with increased risk of developing RA. To undertake this project, we performed experiments focused on the immune response to citrullinated antigens in both HLA-DR4 transgenic mice and patients with RA and healthy controls. A combination of techniques was used to carry out these studies including cellular immunology, flow cytometry, single-cell sequencing for TCR repertoire analysis, protein chemistry, and TCR affinity analysis using SPR and protein x-ray crystallography. The number of independent experiments is outlined in the figure legends, where appropriate.

Mice and peptide immunization

Transgenic mice expressing HLA-DRB1*04:01 (HLA-DR4) on a mouse MHC class II knockout background were purchased from Taconic Biosciences (model 4149) (45). The HLA-DR4 molecules in these mice contain mouse I-Ed α2 and β2 domains to maintain interaction with the CD4 co-receptor. HLA-DR4 and C57BL/6J (H-2b) mice were housed in the animal facility at Monash University, and all animal experimentation were approved and conducted under guidelines set by the Monash University Animal Ethics Committee. Naïve 10- to 18-week-old mice were immunized subcutaneously in each hindfoot hock with either phosphate-buffered saline (PBS) or 100 μg of fibrinogen β 69-81 peptide citrullinated at either position 74 (69GGYRAXPAKAAAT81) or at positions 72 and 74 (69GGYXAXPAKAAAT81), emulsified 1:1 in Freund’s complete adjuvant (FCA) supplemented with Mycobacterium tuberculosis (1 mg/ml) in a total volume of 100 μl (50 μl per hock). dLNs (inguinal and popliteal) or spleen and pLNs were harvested for analysis 8 days later.

Tetramer-based magnetic enrichment of epitope-specific CD4+ T cells from mice

Tetramer-based magnetic enrichment was used for identification of epitope-specific CD4+ T cells in HLA-DR4 mice, as described in detail previously (46, 47). Briefly, single-cell suspensions of pooled spleen and major LNs (auxiliary, brachial, cervical, inguinal, and mesenteric) from naïve HLA-DR4 or C57BL/6 mice or from d8-immunized HLA-DR4 mice were stained with phycoerythrin (PE)– or allophycocyanin (APC)–labeled DR4 tetramers for 1 hour at room temperature, washed, and labeled with anti-PE– or anti-APC–conjugated magnetic microbeads, and tetramer-bound cells were enriched over a magnetic LS column (Miltenyi Biotec). Enriched cells were then stained with a cocktail of conjugated antibodies to identify epitope-specific cells from naïve CD4+ T cell populations (B220, CD11b, CD11c, F4/80, NK1.1, TCRβ, CD4, CD8, and CD62L) or immune CD4+ T cell populations (B220, F4/80, NK1.1, TCRβ, CD8, CD4, and CD62L) using the gating strategy in fig. S1A. Entire samples were acquired on a FACSAria III cell sorter with FACSDiva software (BD Immunocytometry Systems). Frequencies of epitope specific to total CD4+ T cells were determined by ratio of total epitope-specific CD4+ T cell numbers to total CD4+ T cells from unenriched samples.

Tetramer and antibody staining

Analysis of epitope-specific CD4+ T cells from the dLNs of immunized HLA-DR4 mice was performed at the acute phase of the response (d8) and was identified using PE- or APC-conjugated tetramer staining. Cells were incubated with designated tetramers for 1 hour at room temperature, washed, and stained with antibodies to mouse CD4, CD8, CD3, TCRβ, CD62L or CD44, B220, NK1.1, and F4/80. Cells were analyzed on a BD LSR II or LSRFortessa X20 or sorted on a BD FACSAria III with FACSDiva software (BD Immunocytometry Systems).

Analysis of epitope-specific T cell repertoires

Individual HLA-DRB1*04:01Fibβ-74cit69-81–specific, HLA-DRB1*04:01Fibβ-72,74cit69-81–specific, or total non-antigen–specific CD4+ T cells were sorted from naïve or immunized HLA-DR4 mice. For human analyses, individual HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells were sorted from patients with RA or DR4+ healthy controls. For both mice and human samples, mRNA was reverse-transcribed, and a multiplexed nested polymerase chain reaction (PCR) strategy was used to amplify cDNA encoding CDR3α and CDR3β regions using a panel of Vα- and Vβ-specific oligonucleotide primers (48, 49). PCR products were then purified and sequenced on an ABI 3730 DNA Sequencer, and TRBV and TRAV gene usage was determined using IMGT (www.imgt.org).

In vitro TCR expression

293T cells were transfected with a bicistronic mouse stem cell virus–based retroviral vector containing TCRαβ sequence with internal ribosomal entry site–green fluorescent protein (pMIGII), as well as pMIGII encoding the CD3γ, δ, ε, and ζ subunits (50). Transfected 293T cells were labeled with HLA-DRB1*04:01 or HLA-DRB1*04:05 tetramer 48 hours later, followed by fluorescently conjugated anti-TCRβ or anti-CD3 antibody and LIVE/DEAD Aqua Blue viability stain. Cells were analyzed on a BD LSRII or LSRFortessa X20 with FACSDiva software (BD Immunocytometry Systems).

Peptides

Fibrinogen β74-cit (12 mer) (69GGYRAcitPAKAAAT81; denoted as Fibβ-74cit69-81, where cit represented native arginine residue modified to citrulline) and fibrinogen β-72,74-cit peptides (69GGYcitAcitPAKAAAT81; denoted as Fibβ-72,74cit69-81) were synthesized by GL Biochem (China). The integrity of the peptides was verified by reverse-phase high-performance liquid chromatography and mass spectrometry.

Protein expression and purification

Human-mouse hybrid TCRs were designed with human constant region and mouse variable region. A disulfide linkage was engineered in the constant domain of TCRs to assist the protein stability (51). The TCR α and β chains were expressed separately as inclusion bodies in Escherichia coli BL21(DE3). The inclusion bodies were refolded in buffer containing 100 mM tris (pH 8.0), 5 M urea, 0.4 M l-arginine, 2 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 mM oxidized glutathione, and 5 mM reduced glutathione for 48 to 72 hours at 4°C. The refolded samples were dialyzed and further purified on a DEAE (GE Healthcare) anion exchange column in 10 mM tris (pH 8.0). Samples were then further purified using size exclusion (Superdex 200, 16/600; GE Healthcare), hydrophobic interaction (HiTrap SP HP; GE Healthcare), and anion exchange (HiTrap Q HP column; GE Healthcare) chromatography.

HLA-DR4 (DRA*01:01/DRB1*04:01 and *04:05) were cloned in pHLsec vector, expressed in HEK293 S (GnTI) cells, and purified as previously described (8). The constructs contained the Strep-tag II (WSHPQFEKGA; IBA, Gottingen, Germany) sequence followed by the class II–associated invariant chain peptide (CLIP) sequence (PVSKMRMATPLLMQA), designated Strep-CLIP (WSHPQFEKGAPVSKMRMATPLLMQA), covalently linked to the N terminus of the HLA-DR4 β chain via a Factor Xa cleavable flexible linker. The extracellular domains of the HLA-DR α and β chains had C-terminal Fos and Jun leucine zipper sequences, respectively. The HLA-DR β chain had a BirA biotin ligase biotinylation recognition sequence and polyhistidine tag following the Jun zipper. The CLIP sequence used in the DRB1*04:05 construct had a methionine-to-aspartate mutation at the P9 peptide binding register position (Met-to-Asp mutation italicized: WSHPQFEKGAPVSKMRMATPLLDQA) to increase peptide affinity for HLA-DRB1*04:05 (17). The expressed proteins were purified using immobilized metal ion affinity (Ni Sepharose 6 Fast Flow; GE Healthcare), size exclusion (Superdex 200, 16/600; GE Healthcare), and anion exchange (HiTrap Q HP column; GE Healthcare) chromatography.

Peptide loading of HLA-DR4

HLA-DR4 (DRB1*04:01_CLIP and DRB1*04:05_CLIP) was digested with Factor Xa (New England Biolabs, MA, USA) in the presence of 2 mM CaCl2 to cleave the covalently linked Strep-CLIP peptide. Fibβ-74cit69-81 or Fibβ-72,74cit69-81 peptide was loaded in 20 M excess to the Strep-CLIP–Factor Xa cleaved HLA-DR4 in buffer comprising 50 mM trisodium citrate (pH 5.4) and 5 mM EDTA at 37°C for 24 to 72 hours. The reaction was catalyzed by HLA-DM with a molar ratio of 1:5 of HLA-DM:HLA-DR4. Fibβ-74cit69-81 and Fibβ-72,74cit69-81 peptide–loaded HLA-DR4 was separated from unloaded HLA-DR4–Strep-CLIP using Strep-Tactin Sepharose (IBA, Gottingen, Germany). The monomeric HLA-DR4–peptide complexes were subjected to enterokinase (New England Biolabs, MA, USA) cleavage to remove the Fos/Jun leucine zippers and further purified using anion exchange (HiTrap Q HP column; GE Healthcare) chromatography. Target fractions were pooled and concentrated to 10 mg/ml.

Crystallization, data collection, and processing

Crystallization of TCR–HLA-DR4Fibβ-74cit69-81/ Fibβ-72,74cit69-81 ternary complexes were carried out using the broad screen sitting-drop vapor diffusion method (Monash Molecular Crystallisation Facility, Monash University, Clayton, Australia). Crystal hits were obtained and optimized in solution containing 0.2 M trisodium citrate and 19 to 25% polyethylene glycol, molecular weight 3350, using the hanging-drop vapor diffusion method. Proteins were mixed at 1:1 ratio with mother liquor and equilibrated against 500 μl of mother liquor. Rod-like crystals typically grew within 2 to 6 days. Crystals were soaked in cryoprotectant containing 20 to 25% glycerol before flash freezing in liquid N2. Data were collected at the microfocus beamline (MX2) of the Australian Synchrotron (Victoria) and autoprocessed using program XDS (52) at the Australian Synchrotron.

Structure determination, refinement, and validation

Complex structures were solved by molecular replacement in PHASER (53) and CCP4 (54) suite using separate search model of TCR [Protein Data Bank (PDB) ID 4QRP for TCR α chain and PDB ID 5C08 for TCR β chain] and DR4 (PDB ID 6BIL). Iterative rounds of model building in Coot and restrained refinement using REFMAC (CCP4 suite) (54) and phenix.refine (PHENIX) (55) were carried out. The crystal structures were validated using PDB structure validation server. The TCR variable domains were numbered according to the IMGT unique numbering system. Data processing and refinement statistics were summarized in table S2. Ramachandran statistic of final models revealed 96 to 98% of residues in favored regions, with no outlier residue. BSA and contact analysis were performed using programs Areaimol and Contact in CCP4 Program Suite, respectively. All structural figures were generated by PyMOL (www.pymol.org/).

Tetramer preparation

Purified peptide-loaded HLA-DR4 was buffer-exchanged into 10 mM tris (pH 8.0) and biotinylated on the BirA peptide substrate tag in buffer containing 0.05 M bicine (pH 8.3), 0.01 mM adenosine 5′-triphosphate, 0.01 mM MgOAc, 50 μM d-biotin, and 2.5 μg of biotin protein ligase (BirA). Tetramerization, via streptavidin-fluorochrome conjugation, of HLA-DR4 was performed by the addition of streptavidin-PE conjugate (Invitrogen) or streptavidin-APC conjugate (Invitrogen) to biotinylated monomeric HLA-DR4 in a 1:4 molar ratio, with 10 consecutive additions of conjugate at 10 min intervals.

Surface plasmon resonance

Equilibrium affinity constants of TCR–HLA-DRB1*04:01Fibβ-74cit69-81/ Fibβ-72,74cit69-81 interactions were determined using SPR on a Biacore T200 instrument (GE Healthcare). Biotinylated HLA-DR4 molecules were surface-immobilized on a streptavidin sensor chip (GE Healthcare) up to a maximum of 3000 response units. To control for nonspecific binding, HLA-DRB1*04:01CLIP was immobilized in the reference flow cell. Decreasing concentrations of TCRs were passed over the flow cell surface in 20 mM Hepes (pH 7.5), 150 mM NaCl, 1 mM EDTA, and 0.005% Tween 20 at a flow rate of 10 μl/min. Data analysis of the sensograms and curves were plotted using GraphPad Prism version 7.0 (GraphPad).

Tetramer-based magnetic enrichment of epitope-specific CD4+ T cells in humans

Human experimental work was conducted according to the Declaration of Helsinki Principles and the Australian National Health and Medical Research Council (NHMRC) Code of Practice. Use of samples for analysis of epitope-specific cells from humans was approved by the Monash University Human Ethics Committee (HREC 23019). Tetramer-based magnetic enrichment was used for identification of epitope-specific CD4+ T cells in PBMC isolated from two patients with HLA-DRB1*04:01+ RA (RA1, female and 93 years old; RA2, female and 32 years old) or from two HLA-DRB1*04:01+ healthy donors (both male; HC1, 27 years old and HC2, 30 years old). Briefly, 36 million (RA1), 11 million (RA2), or 50 million (healthy donors) cryopreserved PBMC were thawed and rested overnight at 37°C, 5% CO2. After dasatinib treatment (50 nM) for 30 min at 37°C, cells were stained with PE-labeled HLA-DRB1*04:01Fibβ-74cit69-81 tetramer for 1 hour at room temperature, washed, and labeled with anti-PE–conjugated magnetic microbeads, and tetramer-bound cells were enriched over a magnetic LS column (Miltenyi Biotec). Enriched cells were then stained with a cocktail of conjugated antibodies to identify epitope-specific cells from naïve CD4+ T cell populations (CD14, CD19, CD3, CD4, and FVS700 viability stain) using the gating strategy in fig. S1B. Entire samples were acquired on a FACSAria III cell sorter with FACSDiva software (BD Immunocytometry Systems).

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/6/58/eabe0896/DC1

Fig. S1. Gating strategies for sorting HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells.

Fig. S2. Characterization of HLA-DRB1*04: 01Fibβ-74cit69-81–specific CD4+ TCR clonotype usage in naïve and immune HLA-DR4 transgenic mice.

Fig. S3. Characterization of HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ TCR repertoire in naïve and immune HLA-DR4 transgenic mice.

Fig. S4. Overlaid interactions of M134 and M141 TCR–HLA-DRB1*04:01Fibβ-74cit69-81 complexes.

Fig. S5. Affinity analysis of M134 TCR point mutants for HLA-DRB1*04:01Fibβ-74cit69-81.

Fig. S6. Representative TCR binding curves determined by SPR across HLA-DRB1*04:01 shared epitope mutant.

Fig. S7. Representative TCR binding curves determined by SPR across HLA-DRB1*04:05 allomorph.

Fig. S8. Overview structure of TCR-DR04:01Fibβ-72,74cit complexes.

Table S1. List of HLA-DRB1*04:01Fibβ-74cit69-81–specific CD4+ T cells in naïve and immune HLA-DR4 transgenic mice.

Table S2. Data collection and refinement statistics of TCR peptide–HLA-DR4 structures.

Table S3. TCR–pMHC II complex statistics.

Table S4. Contacts of G08 TCR–HLA-DRB1*04:01Fibβ-74cit69-81.

Table S5. Contacts of M134 TCR–HLA-DRB1*04:01Fibβ-74cit69-81.

Table S6. Contacts of M141 TCR–HLA-DRB1*04:01Fibβ-74cit69-81.

Table S7. Contacts of M134 TCR–HLA-DRB1*04:01Fibβ-72,74cit69-81.

Table S8. Contacts of M141 TCR–HLA-DRB1*04:01Fibβ-72,74cit69-81.

Table S9. Contacts of G08 TCR–HLA-DRB1*04:01Fibβ-72,74cit69-81.

Table S10. Raw data file (Excel spreadsheet).

REFERENCES AND NOTES

Acknowledgments: We thank the staff at the Australian Synchrotron for assistance with data collection, the staff at the Monash Macromolecular Crystallisation Facility, and the staff at Monash FlowCore. We thank P. Bradley (Fred Hutchinson Cancer Research Center, Seattle) for assistance with TCRdist analysis. Funding: This work was supported by Janssen Pty Ltd. and the Australian Research Council (ARC) (CE140100011). N.L.L.G. is supported by an ARC Future Fellowship. A.W.P. is supported by an NHMRC Principal Research Fellowship. J.R. is supported by an ARC Australian Laureate Fellowship. Author contributions: J.J.L., C.M.J., H.H.R., N.L.L.G., and J.R. designed the research. J.J.L., C.M.J., T.J.L., Y.T.T., P.Z., K.L.L., and R.K.S. performed the research. J.J.L., C.M.J., A.W.P., L.K., V.M., H.H.R., N.L.L.G., and J.R. analyzed the data. N.J.F., A.S., M.M., F.S., and D.G.B. provided reagents and intellectual input. J.J.L., C.M.J., H.H.R., N.L.L.G., and J.R. wrote the paper. Competing interests: This work has been supported, in part, by Janssen Research & Development LLC. N.J.F., A.S., M.M., F.S., and D.G.B. are employees of Janssen Research & Development. The other authors declare that they have no competing interests. Data and materials availability: The x-ray crystal structures were deposited in PDB with the following accession codes: G08 TCR–HLA-DRB1*04:01Fibβ-74cit69-81, 6V13; M134 TCR–HLA-DRB1*04:01Fibβ-74cit69-81, 6V1A; M141 TCR–HLA-DRB1*04:01Fibβ-74cit69-81, 6V18; G08 TCR–HLA-DRB1*04:01Fibβ-72,74cit69-81, 6V15; M134 TCR–HLA-DRB1*04:01Fibβ-72,74cit69-81, 6V19; and M141 TCR–HLA-DRB1*04:01 Fibβ-72,74cit69-81, 6V0Y. All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Immunology

Navigate This Article