Mucosal-associated invariant T (MAIT) cells are characterized by a semi-invariant TCR that recognizes vitamin B metabolite Ags presented by the MHC-related molecule MR1. Their Ag restriction determines a unique developmental lineage, imbuing a tissue-homing, preprimed phenotype with antimicrobial function. A growing body of literature indicates that MR1-restricted T cells are more diverse than the MAIT term implies. Namely, it is increasingly clear that TCR α- and TCR β-chain diversity within the MR1-restricted repertoire provides a potential mechanism of Ag discrimination, and context-dependent functional variation suggests a role for MR1-restricted T cells in diverse physiological settings. In this paper, we summarize MR1-restricted T cell biology, with an emphasis on TCR diversity, Ag discrimination, and functional heterogeneity.
Much of what we know about T cell biology pertains to that of “conventional” adaptive CD4+ and CD8+ αβ T cells that recognize peptide Ags presented by MHC molecules via their TCR (1). However, many T cells, collectively termed “unconventional T cells,” recognize nonpeptidic Ags presented by monomorphic MHC class I (MHC I)–like Ag-presenting molecules or peptide Ags presented by the nonclassical MHC molecules (2). For example, the deep hydrophobic pockets of the CD1 family of MHC I–like molecules, which includes CD1a, CD1b, CD1c, and CD1d, permits the presentation of lipid-based Ags to a broad family of CD1-restricted T cells (2–4). An extensively studied group of cells within this family are the CD1d-restricted NKT cells that exhibit a restricted TCR repertoire and fast-responding innate-like phenotype (2). More recently, the MHC-related molecule (MR1) has emerged as another important MHC I–like Ag-presenting molecule with the ability to present nonpeptidic Ags to T cells (5). In the case of MR1, a region of the Ag-binding pocket (termed the A’-pocket) is lined with aromatic residues that facilitate the binding of microbial vitamin B2 metabolites and derivatives thereof (6–8). The vast majority of studies on MR1-restricted T cells have focused on the population commonly known as mucosal-associated invariant T (MAIT) cells. There is, however, an emerging body of literature demonstrating not only diversity within the MAIT cell compartment but also the existence of a broader family of MR1-restricted T cells beyond that of MAIT cells. In this review, we will discuss the defining features of the human MR1–MAIT cell axis and highlight emerging data pertaining to so-called “atypical” MR1-restricted T cells.
MAIT cells—a brief history
In 1993, Porcelli and colleagues (9) detected an enriched TCR α-chain rearrangement comprising TRAV1-2 rearranged to TRAJ33 in human CD4−CD8− double negative (DN) T cells. Lantz and colleagues subsequently detected this invariant TCR α-chain in both humans and mice and showed that it was paired with a constrained TCR β-chain repertoire (highly enriched for the TRBV6 family and TRBV20-1 in humans and TRBV13 and TRBV19 in mice). They also showed that this population of T cells was enriched in the gut mucosa. These characteristics served as the basis for the term MAIT cells (10). This group subsequently showed that MAIT cells are restricted to MR1 (11), highlighting the unique nature of these cells. For many years, the types of Ag(s) displayed by MR1 was unclear until in 2012, a breakthrough came when it was determined that MR1 captures and presents small molecule metabolites, including a photodegradation product of folic acid [6-formylpterin (6-FP)] and Ags derived from the microbial vitamin B2 (riboflavin) metabolic pathway (6, 8). Although mammals obtain their riboflavin from dietary sources, many strains of bacteria and yeast have the ability to synthesize riboflavin via the Rib genes (12, 13). A microbe-specific intermediate in this pathway, 5-amino-6-d-ribitylaminouracil (5-A-RU), can react with small chemical adducts endogenous to microbial and mammalian cells such as methylglyoxal to produce unstable pyrimidine neoantigens such as 5-(2-oxopropylideneamino)-6-d-ribitylaminouracil (5-OP-RU) (7). 5-OP-RU can be captured by endoplasmic reticulum–resident stores of nascent MR1, triggering MR1 egress to the cell surface for presentation of the captured neoantigen to MAIT cells via their TCR (14, 15). Thus, these ribityl Ags represent a signature of microbial infection that is monitored by MAIT cells via the MAIT TCR–MR1 axis (6, 7).
A unique developmental pathway instructs distinctive effector function.
MAIT cells exhibit potent antimicrobial effector function (12, 13) underpinned by a preprimed transcriptional profile acquired during thymic development (16). Unlike conventional CD8+ T cells that are selected by MHC I–expressing epithelial cells, the rearrangement of a MAIT TCR by developing thymocytes instructs selection on MR1-expressing CD4+CD8+ double positive (DP) thymocytes, resulting in divergence from the conventional T cell lineage (17). MAIT cells subsequently follow a three-stage developmental pathway, including the upregulation of the master transcription factor PLZF that drives acquisition of a preprimed innate effector scheme. MAIT cells subsequently emerge from the thymus with a tissue-homing, proinflammatory phenotype poised for rapid effector function (16). Upon thymic emigration, presumably in response to microbial priming (16, 18), MAIT cells expand to high numbers, especially in humans, where they represent an average of 5% of all T cells and are even higher in some peripheral organs such as liver, where they represent up to 45% of T cells in humans (18, 19). For reasons that are not entirely clear, but likely reflect the reduced microbial burden in laboratory mice, MAIT cells do not expand to the same high frequency in mice, where they typically remain below 1% of T cells (20). In humans, resting blood MAIT cells coexpress low to intermediate levels of transcription factors RORγt, T-bet, and Helios and high levels of Eomes, and upon activation, produce large quantities of proinflammatory cytokines IFN-γ, TNF, and, under certain microenvironmental conditions, IL-17A (16, 21–24). MAIT cells also acquire cytotoxic activity, including upregulation of perforin and granzymes, allowing them to lyse infected cells (25, 26). Moreover, MAIT cells express high levels of innate cytokine receptors such as IL-1R, IL-12R, IL-18R, and IL-23R (21, 22) and can respond to these cytokines in the absence of TCR ligation (27), foregoing the requirement for detection of microbial Ags and allowing responses to other inflammatory stimuli such as viral infections (28–30). This offers a potential mechanism for MAIT cells to play a role in nonbacterial, inflammatory diseases such as viral infection, autoimmunity, and cancer (31). Thus, MAIT cells likely act as peripheral sentinels primed for an immediate response to infection and/or inflammation.
Accordingly, three defining features of the MAIT cell lineage emerge 1) expression of a unique, semi-invariant TCR 2), MR1 restriction and recognition of microbial riboflavin metabolite Ag, and 3) distinct effector function acquired during development and driven by PLZF expression. As human blood MAIT cells are often characterized by their invariant TCR α-chain, they are often identified using a combination of Abs directed against the TRAV1-2 TCR α-chain as well as high expression of the C-type lectin CD161 (18), the IL-18Rα-chain CD218a (12), or the ectopeptidase CD26 (21, 32) (Fig. 1). However, some of these markers are acquired late during MAIT cell development, their expression can be lost in some settings (33), and some non-MAIT cells can also express some of these markers (34). The recent development of MR1-Ag tetramers (7, 20, 35, 36) has provided a more definitive method of MAIT cell isolation because they identify MAIT cells based upon their MR1-restricted TCR specificity and are independent of surrogate phenotypic markers (37).
There are many parallels between the MAIT cell lineage and that of NKT cells (38). Type I NKT cells also express a semi-invariant TCR that drives restriction to lipid Ags presented by CD1d. They also follow a unique developmental pathway that involves selection on CD1d-expressing DP thymocytes and acquisition of an innate-like phenotype driven by PLZF expression. It is well established that there exists heterogeneity within the NKT cell family as well as a broader family of CD1d-restricted T cells beyond the type I NKT lineage (2). More recently, a building body of literature is emerging that suggests that similar heterogeneity exists within the MAIT cell lineage and extends beyond MAIT cells to a broader family of MR1-restricted T cells.
MR1 ligand diversity
A number of riboflavin-based Ags have been described that can activate MAIT cells with varying degrees of potency (6, 7, 39). The first of these Ags to be described were the bicyclic ribityllumazines 7-hydoxy-6-methyl-8-d-ribityllumazine (RL-6-Me-7-OH), 6,7-dimethyl-8-d-ribityllumazine (RL-6,7-diMe), and the highly potent reduced 6-hydroxymethyl-8-d-ribityllumazine (rRL-6-CH2OH) (6). Subsequently, rRL-6-CH2OH was shown to in fact take on a different chemical structure to that which was initially suggested and instead formed a single-ringed pyrimidine Ag, 5-OP-RU (7). 5-OP-RU and the related 5-(2-oxoethylideneamino)-6-d-ribitalaminouracil (5-OE-RU) pyrimidine Ags are universally recognized by MAIT cells (7), and 5-OP-RU is routinely used in the production of MR1-Ag tetramers that efficiently stain all MAIT cells, thus serving as the archetypal MAIT cell Ag (7, 20, 37). More recently, further lumazine Ags have been described, including acetylated RL-6-Me-7-OH, 6-(2-carboxyethyl)-7-hydroxy-8-ribityllumazine (photolumazine I; PLI), 6-(1H-indol-3-yl)-7-hydroxy-8-ribityllumazine (photolumazine III; PLIII), and the riboflavin analog 7,8-didemethyl-8-hydroxy-5-deazariboflavin (FO) (39). Reductionist genetic approaches have identified the microbial Rib enzyme that catalyzes the production of 5-A-RU as necessary for the generation of MAIT cell agonist Ags (6, 7, 40). Thus 5-A-RU, which can theoretically serve as a precursor for the production of each of these agonist Ags, has emerged as a central molecule for MAIT cell recognition. Indeed, the presence and activation of the riboflavin operon correlates with an ability for diverse bacterial species to activate MAIT cells (41, 42). The extent to which these different Ags exist in nature remains unclear, nonetheless their unifying feature is the conserved ribityl tail that is directly recognized by the MAIT TCR upon docking with MR1 [(43) and reviewed in Ref. 44)]. Beyond riboflavin-based Ags, vitamin B2 (folate) derivatives can also be captured and presented by MR1, including the folate photodegradation product 6-FP (6) and its synthetic derivative acetyl 6-FP (Ac–6-FP) (45). The physiological relevance of these folate-based MR1 ligands remains unclear, but due to the lack of a ribityl moiety they are generally not recognized by MAIT TCRs and are considered to be MAIT cell inhibitors (45). Structural studies in this axis highlight a degree of plasticity within the MR1-binding groove with a “cradle” of aromatic residues, allowing the capture and presentation of diverse small-ringed compounds (6, 7, 43, 45). In a recent publication, in silico docking studies were combined with in vitro cellular assays, in vivo studies, and structural studies to describe a broad series of structurally distinct compounds that can bind MR1, some of which can inhibit MAIT cells in vitro and in vivo (46). This includes some clinically indicated pharmacological agents, including metabolites of diclofenac, the active compound in Voltaren. Moreover, in another study, MR1-Ag complexes were produced using an insect cell expression system that enabled incorporation of MR1 ligands derived from two bacterial species: Mycobacterium smegmatis and Escherichia coli (39). Elution of MR1-furnishing ligands identified a large repertoire of ions, including riboflavin-based Ags mentioned above (PLI, PLIII, and FO) that were differentially recognized by MAIT cells and also suggested that riboflavin itself (and adducts thereof) can bind MR1 and inhibit MAIT cells. This study also identified nonriboflavin-based Ags such as hesperidin and many other ligands that could not be chemically assigned. However, given the heterogeneous nature of the expression system employed, coupled with the presence of MR1 ligands in the tissue culture supernatant, the physiological relevance of some of these candidate ligands needs to be established. Thus, beyond microbial riboflavin-derivative compounds, MR1 may play a role in sampling key components of the cellular metabolome for T cell–mediated immune surveillance.
MAIT cell diversity—variation in an “invariant” population
Within the MAIT cell lineage, there exists multiple levels of heterogeneity at both the level of TCR repertoire and Ag recognition as well as their phenotype and function.
Ag-specific MAIT cell subsets.
Extensive x-ray crystallography studies have shown that MAIT TCRs conform to a pattern recognition–like docking mode atop MR1, whereby the germline-encoded CDR1 and 2 loops of the TCR α- and β-chains straddle the α2 and α1 helices of MR1, respectively, placing the invariant CDR3α loop in a conserved position extending into the Ag-binding pocket of MR1 (43, 44, 47). This allows CDR3α to interact directly with the ribityl tail of riboflavin-based Ags, with the sole contact being mediated by a TRAJ gene–encoded tyrosine at position 95 of the TCR α-chain (Tyr95α; Fig. 2A–D).
Although the MAIT TCR is highly restricted, variation nonetheless exists within this repertoire, allowing MAIT cells, at a population level, to cater for antigenic variation. Although most MAIT TCR α-chains pair TRAV1-2 with TRAJ33 with limited N-nucleotide additions at the junction, a sizeable portion of MAIT TCRs instead use TRAJ12 or TRAJ20 (35, 37). These distinct TRAJ genes are unified by a germline-encoded tyrosine, such that when rearranged with TRAV1-2, a CDR3α sequence length of 10 aa positions the tyrosine in a structurally identical location (45). Thus, use of either of these three TRAJ genes results in an equivalent structural motif for recognition of ribityl Ags (45). Recent studies, however, suggest that TRAJ gene usage may influence the CDR3β repertoire via structural compatibility of CDR3α and CDR3β loops, which can alter Ag specificity as discussed below (37). Thus, TRAJ gene usage may indirectly modulate Ag specificity. Indeed, one study of human peripheral tissues suggested distinct tissue distributions between TRAJ33+, -12+, and -20+ MAIT TCRs, which may be suggestive of distinctive function (48). Moreover, rare populations of MAIT cells use other TRAJ genes to encode noncanonical CDR3α junctions that lack the Tyr95α (37, 49). How these other TCRs recognize riboflavin-based Ags is unclear.
Although different TRBV genes can be used by the MAIT TCR, the germline-encoded CDR1 and 2β loops of the TCR β allow the MAIT TCR to dock a conserved region of the α1 helix of MR1 and are not involved in Ag recognition (43, 45). The CDR3β loop, however, is hypervariable, encoding the majority of the variation within the MAIT TCR repertoire (10). Furthermore, the conserved docking mode positions the CDR3β loop directly adjacent to the CDR3α loop and within close proximity to the ribityl tail of the riboflavin-based Ags (Fig. 2C). Indeed, in some cases, the CDR3β loop can make direct contact with the Ag, and thus, CDR3β sequence variation can fine-tune responsiveness to MR1-restricted Ags (45, 46). For example, whereas all MAIT cells bind MR1-Ag tetramers loaded with 5-OP-RU, tetramers loaded with folate-based Ags 6-FP or Ac–6-FP identify discrete subpopulations of MAIT cells (24). Cellular and structural studies highlight that CDR3β variation permits this recognition and imbues a fine specificity, allowing different MAIT TCRs to distinguish between subtle differences in the chemical structure of Ags such as 6-FP and Ac–6-FP. In another example, diclofenac metabolite reactivity by a MAIT TCR was mediated by CDR3β that underwent major structural repositioning to accommodate the drug, highlighting a degree of flexibility in the CDR3β loop relative to the fixed “linchpin” CDR3α loop and germline-encoded CDR1 and 2 loops, which maintain the conserved docking mode (46) (Fig. 2E). Similarly, some MAIT cell clones differentially react to APCs infected with different strains of bacteria (50) and can differentiate between ribityllumazine Ags (39). How MAIT TCRs differentiate between ribityl Ags is unclear; however, CDR3β remains the defining difference between these distinct TCRs and is thus likely to be central to this recognition event. This mode of Ag discrimination is reminiscent to that of the type I NKT TCR recognition of CD1d–lipid Ag complexes, whereby all type I NKT TCRs dock onto CD1d with a conserved footprint and recognize the prototypical lipid Ag α-galactosylceramide (αGalCer) via direct contacts made by the invariant TCR α-chain. The fine specificity toward distinct lipid Ags, however, can be modulated via the TCR β-chain, although in the case of NKT TCRs, this is modulated via differential affinity to CD1d, and the CDR3β loop does not contact the Ag itself (51–53). Thus, CDR3β hypervariability provides a mechanism for semi-invariant TCRs to accommodate antigenic variation perhaps in an evolutionary effort to prevent microbial antigenic drift.
The functional significance of recognition of nonribityl Ags by MAIT cells remains unclear. Their recognition is generally of low affinity relative to the ribityl Ags, and the threshold of TCR-mediated activation for primary MAIT cells is unknown (24, 45). However, given that MAIT cells can be activated in a TCR-independent manner (27), it is possible that in the context of an inflammatory microenvironment with cytokines such as IL-12 and IL-18 and other costimulatory factors, nonribityl Ags may contribute to the activation of MAIT cells as is the case where self-lipid Ags play an important role in NKT cell activation in association with inflammatory stimuli (54). Moreover, most MAIT cells express CD8 either as a CD8αβ heterodimer or as CD8αα homodimers, where CD8αβ+ MAIT cell homodimers predominate (37). CD8 has been shown to play a role in MR1-mediated MAIT cell activation (55, 56), and because of high sequence homology between the α3 domains of MR1 and MHC I (57), it is likely that CD8 binds MR1, increasing the overall avidity of the interaction. How this affects Ag responsiveness is unclear, but it is possible that CD8 is necessary for MAIT cell responses to low-affinity Ags. Indeed, MAIT cells that bind MR1–6-FP and -Ac–6-FP tetramers are exclusively CD8+ (24). Finally, how Ags of differing affinity influence MAIT cell effector function is also unclear. A recent study, however, showed that activation of MAIT cells with Ags of weak, intermediate, or high potency induce differential changes in surface marker expression (58). This is reminiscent of what has been described for NKT cells in that TCR-mediated signaling events of different strength are capable of activating distinct transcriptional pathways and effector functions (54). It seems likely that a similar scenario may govern MAIT cell responses to different Ags.
Phenotypic and functional diversity.
MAIT cells can be delineated into several phenotypic subsets based on CD4 and CD8 coreceptor usage (37, 56). Although the majority of human blood MAIT cells express either CD8αα homodimers or CD8αβ heterodimers, 10–20% are CD4− CD8− DN and minor populations (∼1%) are CD4+CD8− or CD4+CD8+ (37). CD4+ MAIT cells are difficult to identify without the use of MR1-Ag tetramers because surrogate markers such as CD161 and TRAV1-2 can stain many CD4+ T cells that are not MAIT cells; hence, caution should be taken interpreting data that use surrogate Ab-based identification techniques to study these cells (37). Although the CD8+ and DN subsets appear to be functionally similar, the CD8+ fraction has been shown to have moderately increased cytotoxic potential (59). CD4+CD8− MAIT cells appear to be more distinct from the others, expressing less surface markers typically associated with cytotoxicity such as CD56, NKG2D, and NKG2A and producing more IL-2 in vitro compared with their CD8+ and DN counterparts (37, 56). There are some discrepancies in the literature as to whether CD4+ MAIT cells exhibit cytokine profiles distinct to other MAIT cells, in particular expression of type 2 cytokines such as IL-4 and -13; however, this may be explained by the use of MR1 tetramers (31) versus surrogate Ab techniques to isolate the cells (56). Nonetheless, further investigation into the functional potential of CD4+ MAIT cells is warranted.
Although blood MAIT cells are relatively homogenous for many surface markers, some, such as the costimulatory receptor CD27 and the adhesion protein CD56 exhibit clear bimodal expression patterns (37, 60). Although the functional significance of these subpopulations remains to be fully investigated, one study showed that CD56+, CD84+, and CD94+ MAIT cells were more sensitive to IL-12– and IL-18–mediated IFN-γ production. These cells also had higher cytotoxic potential than their CD56−, CD84−, and CD94− counterparts as well as increased PLZF, eomesodermin, and T-bet transcription factor expression and innate cytokine receptor expression (60).
In mice, MAIT cells appear to be divided into two functionally distinct subsets based on their mutually exclusive expression of the transcription factors RORγt and T-bet (20). In human blood, MAIT cells coexpress RORγt and intermediate levels of T-bet, which are thought to drive the rapid proinflammatory potential of these cells (22, 23). It appears, however, that checkpoints are in place, requiring resting human blood MAIT cells to receive stimulatory signals beyond TCR ligation to elicit a full effector response (Fig. 3). Transcriptional studies have revealed that secondary signaling may be required to prime MAIT cells for efficient signaling via the TCR. For example, costimulation via CD28 or cytokines such as IL-1β, IL-12, IL-18, and IL-23 induces reorganization of key proteins in the TCR signaling pathway, enhancing MAIT cell responsiveness to TCR ligation (61, 62). Moreover, in vivo studies have demonstrated that in mice, intranasal administration of 5-OP-RU requires the coaddition of a TLR agonist or coinfection with riboflavin pathway–deficient bacteria to elicit a MAIT cell response (63). Furthermore, the context of these secondary signals appears to play an important role in dictating the nature of the ensuing effector response. For example, the majority of blood MAIT cells produce type I cytokines IFN-γ and TNF upon stimulation, but only minor subsets of blood MAIT cells produce IL-17A, and exogenous factors are required to license MAIT cell IL-17A production. In the context of in vitro TCR ligation, the addition of exogenous IL-7 and, to a lesser extent, IL-1β and IL-23 can induce MAIT cell IL-17A production (19). Indeed, even in the absence of TCR ligation, IL-7 can enhance cytolytic molecules perforin, granzyme A, and granzyme B and make MAIT cells more responsive to bacterial infection (23). Likewise, MAIT cell responses to E. coli infection are codependent upon MR1 and IL-12 and IL-18 signaling (26), and these cytokines are absolutely necessary for MAIT cell responses to infected Streptococcus pneumoniae–infected monocytes (41). Accordingly, the context of APC activation, for example, by TLR ligation, also plays a role in efficient MAIT cell activation, thereby indirectly influencing MAIT cell effector function (64). It seems logical that activation under different microenvironmental contexts, either by proinflammatory cytokine in the absence of TCR ligation or APC-derived cytokines together with TCR ligation and costimulation will shape the MAIT cell effector response. For example, exogenous IL-12/18 allows MAIT cells to produce IFN-γ and respond to viral infections in the absence of TCR ligation (28, 29), whereas MAIT cells in the liver (19) and female genital tract (65) are primed to produce IL-17 and also IL-22 in the latter. Likewise, MAIT cells in the colon differ substantially from their blood counterparts, expressing activation markers and inhibitory receptors and express higher levels of T-bet and RORγt, licensing stronger proinflammatory responses, in particular production of IFN-γ and TNF (58). These characteristics may also allow MAIT cells to promote autoimmunity (66) and cancer (31), where dysregulated expression of innate cytokines can elicit an undesired MAIT cell effector response even in the absence of defined MAIT cell Ag. There are many important questions that remain: is acquisition of cytotoxic, type I, or type 17 effector function transient or are polarized MAIT cells epigenetically maintained in that state, as is the case for their conventional CD4+ T cell counterparts (67)? Can MAIT cells be induced to produce type 2 cytokines such as IL-4 or IL-13 or acquire regulatory capacity? Do MAIT cells at different anatomical locations represent true tissue-resident MAIT cells, or are they circulating MAIT cells that transiently acquire a tissue-specific phenotype as they enter a new tissue? The resolution of these questions is the focus of current and future studies by many independent laboratories and will be integral to understanding the role of MAIT cells in health and disease.
Atypical MR1-restricted TRAV1-2neg αβ T cells
The use of MR1-Ag tetramers (35) has not only allowed for the unequivocal identification of MR1-Ag–restricted T cells but also, when combined with an Ab against TRAV1-2, has revealed the existence of discrete populations of TRAV1-2neg, MR1–5-OP-RU tetramer+ cells (24). During characterization of a cohort of patients with multiple myeloma, we observed one patient with an unusually high proportion of these T cells with a similar surface phenotype to MAIT cells (CD8α+, CD161HI, CD218HI) (24). TCR sequencing revealed a clonal population using a germline-encoded TRAV36/TRAJ34 TCR with no N-nucleotide additions at the CDR3α junction, paired with TRBV28, a TRBV gene not typically associated with MAIT TCRs (clone MAV36). This MR1-restricted TCR selectively bound MR1–5-OP-RU and lacked reactivity to folate-based Ags 6-FP and Ac–6-FP. Structural studies revealed a distinct TCR-MR1 docking mode compared with that of conventional MAIT TCRs, with 5-OP-RU Ag specificity mediated by the germline-encoded CDR1α loop (Fig. 4A–D). Moreover, the CDR3β loop of this TCR was positioned above the F′-pocket of MR1, distinct to that of a MAIT TCR, which sits more centrally (Fig. 4E).
Notably, populations of TRAV1-2neg cells with heterogeneous cell surface marker phenotypes were identified in most healthy donor blood samples tested (41). Some of these cells preferentially bound to MR1–5-OP-RU, whereas others reacted with MR1 plus folate-based Ags 6-FP or Ac–6-FP. TCR sequence analysis of in vitro–expanded clones of MR1-restricted TRAV1-2− cells revealed diverse TCR gene usage, and TCR gene transfer into reporter cell lines confirmed the MR1-Ag specificity of these clones, including the MAV21 TCR, which recognized MR1 with folate-based Ags but failed to recognize 5-OP-RU. This study (24) demonstrated that a diverse repertoire of TCR genes can permit MR1 restriction and that these TCRs can adopt distinct MR1-binding modes, permitting the recognition of distinct Ags presented by MR1, and provided the first demonstration of recognition of nonmicrobial Ags by MR1-restricted T cells.
Subsequently, Meermeier et al. (50) used a cytokine-based functional readout to demonstrate MR1-dependent reactivity to M. smegmatis–infected A549 cells by TRAV1-2neg MR1-restricted T cells in five healthy donors (50). One of these clones (clone D462-E4) using a TRAV12-2/TRAJ39+ TRBV29+ TCR was used to confirm MR1–5-OP-RU tetramer reactivity. Interestingly, this clone demonstrated a hierarchy of reactivity when tested against two other riboflavin-based agonists with strong reactivity to RL-6-Me-7-OH and weak reactivity to RL-6,7-diMe. Using a series of different bacterial strains, the authors then showed differential patterns of microbial reactivity by the D462-E4 TCR and a MAIT TCR, suggesting not only an ability to distinguish between riboflavin-based Ags but also differential expression of these Ags by different bacterial species (50). In a follow-up study, this same clone was shown to exhibit strong reactivity to PLI but weak reactivity to PLIII (39), whereas two TRAV1-2+ MAIT cell clones exhibited reciprocal patterns of Ag reactivity. These studies provided the first examples of a TCR that can distinguish between riboflavin metabolite–based Ags. Intriguingly, this TRAV1-2neg TCR reacted strongly to S. pyogenes, whereas the TRAV1-2+ MAIT cell clone failed to respond. Importantly, S. pyogenes genomes are not known to encode the rib operon required for riboflavin biosynthesis (41) and thus is unlikely produce 5-A-RU, the critical microbial component for the generation of ribityl-based MAIT cell Ags. This reactivity was blocked by addition of 6-FP, which presumably outcompeted the unidentified stimulatory Ag for MR1 binding. These data therefore suggest recognition by the D462-E4 TCR of a nonriboflavin-based Ag that is also distinct to 6-FP, supporting the concept that TRAV1-2neg MR1-restricted T cells can recognize diverse MR1-restricted Ags in antimicrobial immunity.
More recently, Lepore et al. (68) used limiting dilution to clone MR1-reactive MAIT cells from healthy blood. One of these (clone DGB129) exhibited MR1-restricted reactivity toward an MR1-overexpressing melanoma cell line (A375-MR1) in the absence of exogenous Ag. Indeed, this reactivity could be blocked by the addition of the ribityllumazine Ag RL-6,7-diMe. TCR sequence analysis revealed a TRAV29/TRAJ23+ TRBV12-4+ TCR whose MR1-Ag reactivity was recapitulated by transfer of TCR genes into T cell reporter lines. Additional MR1-reactive T cell clones were isolated from healthy donors using a functional readout of proliferation in response to A375-MR1 cells and/or CD137 upregulation upon rechallenge. These clones had a diverse TCR repertoire that excluded TRAV1-2. Reactivity of these clones could be inhibited by exogenous 6-FP and Ac–6-FP, whereas hydrophilic fractions of lysates obtained from the human THP-1 monocytic cell line as well as mouse breast tumors provided MR1-dependent stimulatory activity. Interestingly, two of these clones exhibited differential reactivity to two distinct hydrophilic fractions, suggesting presentation of multiple mammalian/tumor-derived Ags to distinct MR1-restricted T cells. The phenotypic features of these cells suggested heterogeneous expression of CD4, CD8, and CD161, and the clones provided diverse transcriptional and cytokine profiles, although this should be interpreted with caution, as these cells had been in vitro–stimulated for multiple weeks prior to analysis. Nonetheless, these data suggest potentially distinct specificities and wide-ranging functional potentials beyond microbial immunity by atypical MR1-restricted T cells.
Collectively, these studies provide strong evidence for the existence of MR1-restricted TRAV1-2neg T cells that exhibit both shared and distinct Ag specificities and functional niches relative to MAIT cells. This also suggests broader roles for these cells in human health and disease, including but not limited to antimicrobial immunity.
These studies raise many important questions. A major conundrum in the field of MR1-restricted T cell biology is the nature of the MR1-binding Ags associated with intrathymic selection of MR1-restricted T cells and their roles in cancer and autoimmunity (2, 31). Beyond microbial riboflavin-based Ags, what is the repertoire of MR1-presented Ags, which MR1-restricted T cell types recognize them, and what is the function of these cells? Given the nature of the MR1-binding groove and its apparent proclivity for small aromatic compounds, perhaps MR1 allows T cells to sample the cellular metabolome for immunosurveillence. Do metabolites of other aromatic molecules, whether microbial- or self-derived, act as MR1 ligands? Do cancer cells, which exhibit dysregulated cellular metabolism, present altered metabolites as signatures of cellular transformation? And can self-ligands associated with inflammation bind to MR1 and contribute to autoimmunity? A better understanding of these issues is important for design and interpretation of studies where MR1 is manipulated or blocked. Although it is usually assumed that any effects resulting from MR1 manipulation are due to MAIT cells, it is necessary to consider the potential role for members of the broader MR1-restricted T cell family.
It will also be important to determine whether atypical MR1-restricted T cells undergo a similar developmental program to that already defined for MAIT cells. Given that the three studies mentioned above suggest phenotypic heterogeneity by TRAV1-2neg MR1-restricted T cells, this suggests that atypical MR1-restricted T cells follow distinct developmental pathways. Of note, MR1-deficient Vα19 Tg mice have residual cells that bind MR1 tetramers (35, 36). Whether this is an artifact of the transgenic model or if these cells align to any observations made in human studies will be important to address. Finally, similar to the CD1d-restricted T cell family, does MR1 restriction extend to the γδ T cell lineage or is MR1 restriction exclusive to αβ T cells?
What’s in a name?
As the field of MR1-restricted T cell biology expands to encompass diverse T cell types and diverse Ags, it is important to consider how we define and classify this extended family of αβ T cells. It is already clear that the term MAIT cell should not be taken literally. Although these cells can be found at elevated frequencies in some mucosal surfaces, they are not always associated with the mucosa, and they are abundant in circulation and other tissues, often at very high frequency, for example, in the liver. Furthermore, although most MAIT cells express an invariant TRAV1-2+TRAJ33+ TCR α-chain, many human TRAV1-2+ MAIT cells express TRAJ12 or TRAJ20 instead of TRAJ33. And as discussed above, it is clear that some TRAV1-2neg cells can exhibit Ag restriction and phenotypic features similar to MAIT cells, whereas others are highly divergent from MAIT cells in terms of their Ag specificity in association with MR1 and their functional potential. We thus suggest that the term MAIT cell should refer to T cells that exhibit MR1 restriction, respond to the archetypal Ag 5-OP-RU, and exhibit an innate-like effector scheme generally associated with PLZF expression that is acquired during thymic selection on MR1-expressing DP thymocytes. This definition emphasizes the unique developmental lineage and Ag reactivity of these T cells, irrespective of TCR gene usage. MAIT cells could be further delineated as those with classical (TRAV1-2-TRAJ33/12/20+) or nonclassical (TRAV1-2neg) TCR gene usage. For MR1-restricted T cells that fail to meet these criteria, we suggest referring to these as atypical MR1-restricted T cells. This descriptive term allows for the likely high level of diversity in TCR usage, Ag specificity, cell surface phenotype, and biological function within the broader population of T cells that interact with MR1 as a restriction element (summarized in Fig. 5).
Further research in this field is clearly required to gain a comprehensive understanding of the role of MAIT cells and the broader family MR1-restricted T cells in health and disease and their potential for therapeutic manipulation.
This work was supported by National Health and Medical Research Council (NHMRC) Program Grant 1113293 and Australian Research Council (ARC) Centre of Excellence Grant CE140100011. D.I.G. is supported by an NHMRC Senior Principal Research Fellowship. J.R. is supported by an ARC Laureate Fellowship.
Abbreviations used in this article:
mucosal-associated invariant T
- MHC I
MHC class I
The authors have no financial conflicts of interest.