A class of γδ T cell receptors recognize the underside of the antigen-presenting molecule MR1

See allHide authors and affiliations

Science  20 Dec 2019:
Vol. 366, Issue 6472, pp. 1522-1527
DOI: 10.1126/science.aav3900

A different way for γδ T cells to bind

The ligands bound by γδ T cell receptors (TCRs) are less well characterized than those of their αβ TCR cousins, which are antigens presented by major histocompatibility complex (MHC) and related proteins. Le Nours et al. identified a phenotypically diverse γδ T cell subset in human tissues that reacts to MHC-related protein 1 (MR1), which presents vitamin B derivatives. A crystal structure of a γδ TCR–MR1–antigen complex revealed that some of these TCRs can bind underneath the MR1 antigen-binding cleft instead of recognizing the presented antigen. This work thus uncovers an additional ligand for γδ T cells and reconceptualizes the nature of T cell antigen recognition.

Science, this issue p. 1522


T cell receptors (TCRs) recognize antigens presented by major histocompatibility complex (MHC) and MHC class I–like molecules. We describe a diverse population of human γδ T cells isolated from peripheral blood and tissues that exhibit autoreactivity to the monomorphic MHC-related protein 1 (MR1). The crystal structure of a γδTCR–MR1–antigen complex starkly contrasts with all other TCR–MHC and TCR–MHC-I-like complex structures. Namely, the γδTCR binds underneath the MR1 antigen-binding cleft, where contacts are dominated by the MR1 α3 domain. A similar pattern of reactivity was observed for diverse MR1-restricted γδTCRs from multiple individuals. Accordingly, we simultaneously report MR1 as a ligand for human γδ T cells and redefine the parameters for TCR recognition.

Alpha beta T cell receptors (αβTCRs) recognize antigens presented by major histocompatibility complex (MHC) and MHC class I–like molecules, including CD1 and MHC-related protein 1 (MR1) (1). A paradigm in T cell–mediated immunity is TCR binding atop the antigen-binding cleft of these molecules. However, knowledge of the range of physiological ligands that γδ T cells recognize is very limited. MR1 presents vitamin B precursors and by-products to a highly abundant innate-like T cell subset in humans called mucosal-associated invariant T (MAIT) cells (2, 3). Whether the γδ T cell lineage also encompasses MR1-reactive T cells is unknown.

We detected MR1-tetramer+ γδ T cells among human peripheral blood mononuclear cells (PBMCs) (Fig. 1A). These cells ranged from <0.001 to 0.1% of CD3+ T cells (Fig. 1B), and from <0.1 to 5% of γδ T cells (Fig. 1C). MR1–5-OP-RU tetramer+ γδ T cells [5-OP-RU, 5-(2-oxopropylideneamino)-6-D-ribitylaminouracil] were mostly CD4CD8α or CD8α+ with variable CD161 expression. Thus, they resembled other cells of the γδ T cell lineage (Fig. 1, D to G). Next, we magnetically enriched total γδTCR+ cells from healthy human blood and screened these with MR1 tetramers (Fig. 1, H to J). The staining profiles were variable, with some donors exhibiting defined populations with high- or low-intensity staining (donors 5 and 6, respectively). However, the majority of donors displayed more-diffuse staining patterns (like donor 7). These were similar in frequency to CD1d-reactive γδ T cells (Fig. 1, H to J). We also assessed tissue samples from healthy subjects (fig. S1). Similar to blood, subpopulations of γδ T cells stained with MR1 tetramers in liver, stomach, lung, and duodenum (fig. S1). Notably, a newly diagnosed Celiac disease patient had an enrichment of Vδ1+ γδ T cells that stained with MR1–5-OP-RU tetramers but not MR1–6-FP tetramers (fig. S2). In a tumor infiltrate from a Merkel cell carcinoma patient, γδ T cells were highly enriched (fig. S3). TCR sequencing of this population revealed clonally expanded cells, including 20% with a Vδ1-Vγ3+ TCR, which, upon transduction into human embryonic kidney 293T (HEK293T) cells, exhibited antigen-independent MR1 reactivity (fig. S3). Thus, MR1-restricted γδ T cells are present in the blood and tissues of healthy subjects and may be enriched in association with some diseases.

Fig. 1 Identification and characterization of MR1-restricted γδ T cells.

(A) Flow cytometry dot plots showing MR1 tetramer+ γδ T cells in PBMCs (gated on CD3+CD19CD14 cells). (B) Percentage of MAIT cells (red), MR1 tetramer γδ T cells (green), and MR1 tetramer+ γδ T cells (orange) of total T cells or (C) percentage of γδ T cells in PBMC (n = 12 PBMC samples). Points on x axis are <0.001% (B) or <0.1% (C) (n = 12 donors). Whiskers denote minimum and maximum points. (D) Flow cytometry dot plots showing CD4, CD8, and CD161 expression on conventional T cells, MAIT cells, MR1 tetramer, and MR1 tetramer+ γδ T cells from a representative donor. (E to G) Cumulative data showing CD4 and CD8 distribution (E), CD161 expression (F), and CD161 mean fluorescence intensity (MFI) on CD161+ cells (G) for T cell subsets (n = 12). Error bars denote SEM. DN, double negative (CD4CD8); Conv., conventional. (H) Flow cytometry dot plots showing MR1–5-OP-RU and CD1d-αGalCer-tetramer staining of Vδ1+ γδ T cells enriched from three representative PBMC samples. (I and J) Percentage of MR1–5-OP-RU or CD1d–αGalCer-tetramer+ cells of total γδ T cells (I) or Vδ1+ T cells (J) (n = 7 donors). Whiskers denote minimum and maximum points. (K) Flow cytometry dot plots showing MR1-tetramer staining of enriched and expanded TRAV1-2 MR1–Ac-6-FP tetramer+ cells from four PBMC samples. SAv-PE, streptavidin-phycoerythrin.

Next, total Vδ2 γδ T cells were magnetically enriched from healthy donor PBMC samples (fig. S4) (4). Although there were some antigen-related modulations in MR1-tetramer staining intensity of some populations, most of the MR1 tetramer+ γδ T cells were stained regardless of the type of antigen loaded. Thus, there appears to be an inherent autoreactivity toward MR1 by most PBMC-derived MR1-restricted γδ T cells (Fig. 1K). We then determined which γδTCR genes were used by these cells (table S1). Analysis of 76 TCR δ-chains revealed that most (72%) used TRDV1. The remainder, aside from one TRDV5+ clone, expressed TRDV3 (Fig. 2A). MR1-restricted γδ T cells used all functional TRGV genes including TRGV2, 3, 4, 5, 8, and 9 (Fig. 2A). The TRDV1+ and TRDV3+ TCRs predominantly paired with TRGV8. The unusual “clustered” staining profile was due to different clonal populations exhibiting variable staining intensities (fig. S5). Thus, MR1-restricted γδ T cells can use diverse γδTCR genes to bind to MR1.

Fig. 2 Characterization of MR1-restricted TCRs.

(A) Distribution of TRDV and TRGV genes used by MR1-restricted γδ T cells (n = 76 and 57 unique sequences for TCR-δ and TCR-γ, respectively). (B and C) Flow cytometry histograms showing tetramer staining and MFI of HEK293T cells transiently transfected with MAIT TCRs (MBV28 and M33-64), a CD1d-restricted γδTCR (9C2), or MR1-restricted γδTCRs (G7, G19, G21, G82.C7.1, G82.C8.2, G83.C4, and G83.C5) (hu, human; mu, mouse), using MR1-Ag or CD1d-Ag tetramers (B) or empty MR1 tetramers (C). Experiments were performed three times with similar results. (D) CD69 expression (mean of MFI) on Jurkat76 cells transduced with TCRs as in (B), after coculture with wild-type, MR1HI, or MR1KO C1R APC cell lines in the presence of absence of Ac-6-FP or Ac-6-FP plus anti-MR1. Basal indicates Jurkat76 cells alone. Error bars denote mean ± SEM. Each data point is from an independent experiment and represents mean of two technical replicate wells. Statistical analyses performed were paired Student’s t tests with Bonferroni correction for multiple hypothesis testing. (E) Detection of ERK1/2 phosphorylation (pERK1/2) in Jurkat76 cell lines described in (C), after coculture with MR1HI or MR1KO C1R cell lines analyzed by flow cytometry. Data shows means ± SEM for (n = 7 separate experiments for G7 and G21; n = 4 for remaining cell lines). Statistical analysis was performed using the Mann-Whitney U test.

To test antigen reactivity, some MR1-restricted γδTCRs were transfected into HEK293T cells. All of these TCRs bound human MR1 tetramers loaded with 5-OP-RU or with Ac-6-FP, although some (G19 and G82.C8.2) stained more brightly with MR1–5-OP-RU tetramer (Fig. 2B). Only some of these TCRs bound to mouse MR1–Ac-6-FP tetramer (Fig. 2B). However, all of these TCRs recognized the empty form of human MR1 (MR1–K43A) (Fig. 2C). Thus, MR1 autoreactivity appears to be a driving force for γδTCR recognition of MR1, but with some potential for antigenic modulation.

We investigated MR1-dependent signaling through the γδTCRs on TCR-transduced Jurkat-76 cell lines. Four of these lines were activated (marked by the up-regulation of CD69) in response to MR1-transduced cell lines (Fig. 2D). The other four TCRs (G7, G19, G82.C7.1, and G83.C5) failed to trigger CD69 up-regulation despite clear binding to MR1 tetramers. We then determined whether these TCRs signaled in response to MR1, as measured by mitogen-activated protein kinase 1 and 2 (ERK1/2) phosphorylation (Fig. 2E). A significant response was observed for all but one of the γδTCR+ cell lines tested. Thus, MR1-restricted γδTCRs are capable of activating T cells upon MR1 recognition, with the magnitude of signaling and activation varying between different γδTCRs.

We purified recombinant soluble forms of the G7, G19, and G21 γδTCRs and measured the affinity of their interaction with MR1 bound to 6-FP, Ac-6-FP, and 5-OP-RU using surface plasmon resonance (SPR) (fig. S6). All three γδTCRs recognized MR1–5-OP-RU with moderate affinity (fig. S6, A to C). The G7 and G21 γδTCRs interacted with MR1–6-FP and Ac-6-FP with similar affinity to that of MR1–5-OP-RU (fig. S6, A and B), whereas G19 bound with weaker affinity to MR1–6-FP when compared with that of MR1–5-OP-RU (fig. S6C). Thus, γδTCRs may exhibit MR1 autoreactivity, but with some potential to be modulated by MR1-bound antigen.

We determined the crystal structure of the G7 γδTCRMR15-OP-RU ternary complex (table S2). Here, the G7 γδTCR adopted a highly distinct and unusual docking strategy in that it bound to the α3 domain of MR1 (Fig. 3A and fig. S7), sharply contrasting the docking mode adopted by the MAIT TCR above the MR1-antigen binding cleft (Fig. 3B). The G7 γδTCR sat underneath the MR1 antigen-binding cleft, where the Vγ chain was orientated toward the plane of the membrane and the Vδ chain pointed toward the cleft. Given the highly unusual G7 γδTCR–MR1 docking topology observed in the crystal lattice, we investigated the docking topology in solution, using solution phase hydrogendeuterium exchange mass spectrometry (figs. S8 to S10) and small-angle X-ray scattering (SAXs) (fig. S11). These complementary approaches (see supplementary text and tables S3 and S4) fully validated the observed docking mode.

Fig. 3 Overview of the G7 γδTCR–MR1–5-OP-RU complex and molecular interactions.

(A) Cartoon representation of the G7 γδTCR–MR1–5-OP-RU ternary complex: huMR1, gray; β2m, light orange; γ9 chain, light blue; δ1 chain, pale green; CDR1γ, magenta; CDR2γ, orange; CDR3γ, green; CDR1δ, brown; CDR2δ, blue; and CDR3δ, red. C termini of γδTCR are shown as spheres. (B) Cartoon representation of the MAIT αβTCR–MR1–5-OP-RU ternary complex (PDB ID: 4NQC): MR1, gray; β2m, light orange; α-chain, teal; β-chain, salmon; CDR1α, yellow; CDR2α, blue; CDR3α, red; CDR1β, magenta; CDR2β, orange; CDR3β, green. (C) Molecular interactions of MR1 with CDR1δ, framework δ chain, CDR3δ, and framework γ chain. MR1, light gray; CDR1δ, yellow; CDR3δ, red; framework δ chain, pale green; framework γ chain, light blue. For clarity, only hydrogen bonds are shown as red dashed lines. Footprint of the G7 γδTCR on the surface of MR1 (color coding as above).

The γδTCRMR1 total buried surface area (BSA) upon complexation was 2030 Å2. Within this γδTCRMR1 footprint, the BSA of the G7 TCR was 1050 Å2, to which the TCR δ and γ chains contributed 84 and 16% BSA to the γδTCR–MR1 interaction, respectively (Fig. 3C). The TCR γ chain mainly contacted the base of the α3 domain (BSA: ~190 Å2) (Fig. 3C and table S5). The more dominant Vδ-chain contacts encompassed a greater area of the α3 domain of MR1 (BSA: ~660 Å2) (Fig. 3C). Here, the framework region of Vδ1 and the CDR1δ loop contributed 19 and 25% BSA, respectively, whereas the CDR3δ loop contributed 40% BSA. The CDR3δ loop lies across the α3 domain of MR1 where hydrophobic residues dominated the interactions with MR1 (Fig. 3C). Additionally, the Trp-rich germline-encoded CDR1δ contributed to the interface (Fig. 3C). The α3 domain of human MHC-I binds to natural killer cell receptors (LILRB1 and LILRB2) (58) (fig. S11) and to the CD8 co-receptor (fig. S12) (9). The overall G7 γδTCR contact zone on the MR1 α3 domain was more reminiscent of the latter. Thus, interactions mediated primarily between the TCR δ chain and the α3 domain of MR1 underpin G7 γδTCR docking on MR1.

To gauge the generality of the recognition strategy among other MR1-restricted γδTCRs, we undertook an extensive MR1 mutagenesis study (fig. S13). The majority of the α3 domain mutants significantly reduced the staining of the transfected G7 and G83.C5 γδTCRs (Fig. 4A and fig. S14A). These observations were supported by SPR measurements (Fig. 4B and fig. S14, B and C), which confirms the unusual G7 docking topology and also suggests a similar recognition strategy adopted by G83.C5 γδTCR. SAXs analysis of these γδTCRs (Fig. 4C and fig. S11) indicated that the G21 γδTCR interacted with the antigen-binding cleft of MR1. In contrast, ab initio reconstructions derived from the G83.C5 γδTCRMR1 scattering data revealed an overall shape similar to that of the G7 γδTCRMR1 crystal structure (Fig. 4C and fig. S11). Thus, two γδTCRs with reduced signaling profiles (Fig. 2, D and E) docked beneath the antigen-binding cleft.

Fig. 4 Generality of α3-docking MR1-restricted γδTCRs.

(A) Intensity of tetramer staining of HEK293T cells transfected to express MR1-restricted γδTCRs and a MAIT TCR and stained with a panel of α3 and α1-α2 domain MR1 mutant tetramers. Error bars denote SEM of three independent experiments. (B) Histograms of the affinity measurements of the G7 and G21 γδTCRs for the α3-domain MR1 mutants depicting the equilibrium dissociation constant (Kd) fold difference between the wild-type MR1 Kd and the corresponding mutants Kd (yellow bars: less than threefold difference; orange bars: threefold to fivefold difference; red bars: greater than fivefold difference). Error bars represent SEM from two independent experiments, each measurement in duplicate. (C) SAXs envelopes for the AF-7 αβTCR, G7 γδTCR, G21 γδTCR, and G83.C5 γδTCR bound to MR1 (also see fig. S11). (D) Flow cytometry showing wild-type or Mutx5 MR1-tetramer staining (± CD8 blockade) of enriched and expanded TRAV1-2 MR1-tetramer+ cells from 6 donors (out of 20) where staining was affected by the Mutx5 tetramers (gated on CD3+ TRAV1-2 γδTCR+ cells). Experiment was performed twice with similar results. Red boxes indicate populations with distinct staining profiles for different tetramers.

We next generated an α3-domain quintuple-mutant MR1 tetramer (Mutx5) on the basis of residues important for G7 binding. This was used to screen for other MR1 α3 domain–reactive T cells from 20 additional donors. Previous studies have suggested a potential role for the CD8 co-receptor in modulating TCR-mediated MR1 responsiveness (10), which could also be affected by the α3-mutant tetramer. Accordingly, we preincubated PBMC with blocking anti-CD8α monoclonal antibody to remove this confounding factor (11). These cells were then stained with wild-type or α3-mutant tetramers loaded with Ac-6-FP (Fig. 4D and fig. S15). After eliminating the influence of CD8 coreceptor, a comparison of α3-mutant MR1 tetramer to wild-type MR1 tetramer staining identified a population of α3 domain–dependent γδ T cells in 6 out of 20 donors (Fig. 4D).

We have described a phenotypically diverse population of human γδ T cells that recognizes MR1. These MR1-reactive γδ T cells were found in both healthy and diseased tissues, suggesting a role in physiology and pathology. The prevailing view is that TCRs sit atop the antigen-binding platform and simultaneously co-recognize the presented antigen. Surprisingly, we show that MR1-restricted γδTCRs can adopt diverse binding modes with MR1, including underneath the antigen-binding platform of MR1. Thus, we simultaneously identify a human self-ligand for γδ T cells and transform our understanding of TCR recognition determinants.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S15

Tables S1 to S6

References (1226)

References and Notes

Acknowledgments: We thank L. Wakim for assistance in provision of lung tissue; P. E. O’Brien and P. Burton (Centre for Obesity Research) for provision of human gastric tissue; K. Visvanathan (St. Vincent’s Hospital, Melbourne) for provision of liver samples; J. Furness and M. Nikfarjam (University of Melbourne) for provision of intestinal tissue; and R. Hicks (Peter MacCallum Cancer Centre) for provision of Merkel cell carcinoma sample. We thank M. Herold for the CRISPR constructs, B. Meehan for assistance with MR1 tetramer production, and K. Loh and S. Gras for technical contributions. The crystallographic research was undertaken on the MX2 beamline at the Australian Synchrotron, part of ANSTO. We thank the staff at the Australian Synchrotron for assistance with data collection, the staff at the Monash Macromolecular Crystallisation Facility, and the staff from the University of Melbourne flow cytometry facilities. Funding: This work was supported by program grants (1013667, 1016629, and 1113293) and project grants (APP1159932 and APP1100240) from the National Health and Medical Research Council of Australia (NHMRC) and the Australian Research Council (ARC) (CE140100011). N.A.G. was supported by a Leukaemia Foundation of Australia Postgraduate Scholarship and a Cancer Council Victoria postdoctoral fellowship; J.L.N. is supported by an ARC Future Fellowship (FT160100074); A.J.C. is supported by an ARC Future Fellowship (FT160100083); S.B.G.E. is supported by an ARC DECRA Fellowship; R.B. is supported by an NHMRC CDF Fellowship (1109901); D.G.P. is supported by an NHMRC CDF Fellowship (1144308); A.P.U. is supported by an ARC Future Fellowship (FT140100278); D.I.G. and D.P.F. are supported by NHMRC Senior Principal Research Fellowships (1117766, 1117017); A.W.P. is supported by an NHMRC Principal Research Fellowship; J.R. is supported by an Australian ARC Laureate Fellowship; D.I.G., J.R., and J.M. are supported by a program grant from the National Health and Medical Research Council of Australia (NHMRC) 1113293; A.P.U. is supported by the Cancer Council Victoria (1126866). Author contributions: J.L.N. and N.A.G. are joint first authors and contributed to data generation, data analysis, and paper writing; S.H.R., W.A., F.W., B.S.G., Y.K., T.P., J.M.W., J.J.S., A.I.W., A.v.B., M.T.R., S.J.R., R.S., M.L.S.-R., S.L., M.S.D., R.B., T.T., A.W.P., M.N.T.S., and A.P.U. contributed to data generation and analysis; and S.B.G.E., A.J.C., H.H.R., L.L., D.P.F., E.M.G., G.P.W., R.W.T., J.M., and D.G.P. provided key reagents and samples. D.I.G. and J.R. are joint senior authors. They conceived of the study, analyzed data, and cowrote the paper. Competing interests: A.J.C., S.B.G.E., D.P.F., L.L., J.R., and J.M. are inventors on patents describing MR1 tetramers and MR1-ligands. Data and materials availability: Atomic coordinates and structure factors of the G7 γδTCR–MR1–5-OP-RU ternary complex were deposited in the Protein Data Bank (PDB) under the ID 6MWR. All remaining data are available from the corresponding authors upon request.

Stay Connected to Science

Navigate This Article