The evolutionary conservation of T lymphocyte subsets bearing αβ TCRs using invariant α-chains is indicative of unique and important functions. Among these T lymphocytes, NKT cells that express an invariant TCRα-chain and recognize lipid Ags presented by the nonclassical MHC class I molecule CD1d are probably the most studied. However, a new population of evolutionarily conserved T cells with another invariant TCRα rearrangement was recently characterized. These cells, which are very abundant in humans, tend to reside in mucosal tissues and, therefore, were named mucosal-associated invariant T (MAIT) cells. Until recently, little was known about MAIT cells; however, several recent advances in our understanding of MAIT cell characteristics and functions secure their upcoming rise to fame in the immunology field and in clinical practice.
T lymphocytes that express an αβ TCR are usually categorized into either CD4+ or CD8+ subsets that recognize peptide Ags presented by class II or class I molecules of the MHC, respectively. However, mature T cells that express αβ TCRs, yet lack expression of CD4 and CD8, also exist in mice and humans. Analysis of TCR expression by the peripheral human CD4− CD8− double-negative (DN) αβ T cells demonstrated, as early as 1993, that these cells preferentially expressed a very limited TCR repertoire (1). Two essentially invariant TCRα rearrangements were consistently identified in the αβ DN population of several individuals. These results suggested that αβ DN T cells might recognize a limited spectrum of Ags potentially presented by nonpolymorphic MHC molecules. The first TCRα rearrangement identified corresponded to an invariant rearrangement between TRAV10 (Vα24) and TRAJ18 (Jα18) gene segments with no N region diversity. Today, we know that this unique TCRα-chain, when combined with a limited number of Vβ-chains in mice or with TRBV25 (Vβ11) in humans, is expressed by invariant NKT (iNKT) cells, which recognize various lipid and glycolipid Ags presented by the non-MHC–encoded and nonpolymorphic molecule, CD1d (2).
The second TCRα rearrangement frequently identified in DN αβ T cells used the human TRAV1-2 (Vα7.2) and TRAJ33 (Jα33) gene segments (TRAV1 [Vα19] and TRAJ33 [Jα33] in mice), with two variable amino acids encoded in the V–J junction. In 1999, a seminal study by the Lantz laboratory (3) ascribed this unique TCRα rearrangement to a new subset of T cells. These cells were found to be conserved between mammalian species and were enriched within the gut lamina propria, an observation that led to their denomination as mucosal-associated invariant T (MAIT) cells. MAIT cells were found to be restricted by MR1 (4), a monomorphic class I–related MHC molecule encoded outside of the MHC region with extremely high sequence conservation among mammals in its ligand-binding groove (5).
Altogether, this stringent evolutionary conservation suggested an important and perhaps nonredundant function(s) fulfilled by MAIT cells. However, further detailed characterization of the MAIT–MR1 axis remained hampered by the lack of tools for the identification of MAIT cells ex vivo, and especially by the unknown nature of the MAIT Ag(s). In the past two years, these major hurdles have been overcome, revealing even more surprising discoveries. First, although MAIT cells appear to be relatively rare in mice, they constitute a very significant population of T cells in humans, accounting for between 1 and 10% of T cells in peripheral blood and representing a predominant population of T cells in the human liver and mucosal tissues (6–9). Second, MR1 molecules were shown to present a completely new and unexpected class of Ags to MAIT cells: vitamin B metabolites (10). These findings reveal a new facet of immunity that is largely unexplored and would predict that MAIT cells are key players in immunity.
MAIT cell identification and phenotype
Until recently, most work on MAIT cells had relied on the study of a few isolated MAIT clones and hybridomas, as well as TCR-transgenic mice expressing the invariant MAIT TCR. Although useful, these tools had obvious limitations. A new mAb directed against the human TRAV1-2 TCR chain, when used in combination with anti-CD161 or anti–IL-18Rα, was shown to be sufficient to identify MAIT cells within the CD3+ CD4− compartment (6). Furthermore, with the identification of at least one MAIT Ag (see below), MR1–Ag tetramers have been developed that enable the characterization and tracking of MAIT cells ex vivo, both in mice and humans (7).
In humans, MAIT subsets are mostly DN or CD8+ (expressing mostly the homodimeric form of CD8αα), with a minor subset being CD4+ cells. MAIT cells express an effector memory phenotype, as well as high levels of the NK receptor CD161, of ABCD1 (MDR, a membrane transporter that efflux xenobiotics), and of the receptors for IL-12, IL-18, and IL-23. Human MAIT cells also express high amounts of CCR6 and CXCR6, which are involved in cell trafficking to peripheral tissues, particularly the intestine and the liver. In addition, they express the transcription factors PLZF and RoRγt. Upon TCR stimulation by the MR1–Ag complex, human MAIT cells produce IFN-γ, TNF-α, and IL-2, as well as IL-17 after PMA + ionomycin stimulation (6–9). Furthermore, they were recently shown to be cytotoxic, killing bacteria-infected epithelial cells (11).
MAIT cells are very rare in conventional naive mouse strains. As such, their phenotypical and functional characterization in unmanipulated animals has yet to be fully described. However, a recent study showed that MAIT cells could proliferate to make up ∼25% of all αβ T cells in the lungs postinfection with the intracellular bacteria Francisella tularensis (12), arguing that, like in humans, MAIT cells can constitute a very significant proportion of T cells in mice.
To circumvent the problem of low MAIT cell numbers in unchallenged mice, three independent lines of TRAV1-TRAJ33 TCR-transgenic mice have been generated (6, 13, 14). Although the same TRAV1-TRAJ33 TCR α-chain was used in each case, the genetic elements used to control transgene expression were different. The MAIT cells generated in these transgenic animals differ greatly in terms of cell surface phenotype and function (6, 13, 14). Therefore, it remains unclear which, if any, of these models actually recapitulates MAIT cell development in wild-type mice. It is also unclear to what extent MAIT cell development in mice reflects MAIT cell development in humans and whether mouse MAIT cells are functionally and phenotypically equivalent to MAIT cells in humans. The strong phylogenic conservation of the MAIT–MR1 axis would argue that it should be the case, but it remains to be formally demonstrated.
MAIT reactivity and Ags
In 2010, two publications (15, 16) described the antimicrobial activity of human and mouse MAIT cells. MAIT cells were shown to react to APCs infected with a wide variety of, although not all, bacteria and yeasts but not viruses (15, 16). This reactivity required MR1 expression on APCs and did not seem to involve the main innate immune pathways (15). These results suggested that MAIT cells might be responsive to a conserved microbe-derived product presented by MR1 molecules. This product was resistant to protease digestion and did not copurify with conventional lipids (17, 18). By manipulating the refolding of the MR1 protein produced under various culture conditions, followed by mass spectrometry analysis of MR1-complexed ligands, the groups of McCluskey and Rossjohn (10) revealed that MR1 could present vitamin B metabolites. They first identified the presence of the folic acid (vitamin B9) metabolite, 6-formyl pterin (6-FP), bound to MR1. However, 6-FP did not stimulate MAIT cells. They further identified several metabolites derived from the riboflavin (vitamin B2)–biosynthesis pathway when MR1 was refolded in the presence of culture supernatant from Salmonella typhimurium, a bacterial strain known to stimulate MAIT cells. These riboflavin metabolites were able to activate human MAIT cells, as well as Jurkat cells engineered to express three human MAIT TCRs. Strikingly, all of the microorganisms that had been previously shown to activate MAIT cells have a functional riboflavin-biosynthesis pathway and, therefore, produce these metabolites, whereas bacteria lacking this biochemical pathway are unable to stimulate MAIT cells. Among the three stimulatory compounds characterized, the two less active ones are generated by a known biochemical pathway from the immediate riboflavin precursor. However, the biochemical origin of the most stimulatory compound (1000 times more active) is still unclear (19). Altogether, these results demonstrate that a new class of Ag(s), in the form of vitamin metabolites, can be presented to, and be recognized by, the immune system. Because mammals do not synthesize riboflavin, this might represent a new mechanism of self/nonself discrimination that alerts the host to microbial infection. However, the ubiquitous nature of these small molecules that are found in many bacteria, including commensals, as well as their capacity to freely diffuse across epithelial layers, raises many questions regarding how the activation of MAIT cells might be regulated in vivo.
The identification of MAIT Ags, combined with the production of soluble MR1 molecules, led to the development of MR1–Ag tetramers (7). This new tool, which stains all MAIT cells in TCR-transgenic mice, as well as in humans, revealed some unexpected heterogeneity in TCR usage by MAIT cells. In addition to the typical TRAV1-2–TRAJ33 invariant α-chain, 5–15% of human MAIT cells were found to use two other TRAJ segments (TRAJ12 and TRAJ20) (7). The functional relevance of these “alternate” MAIT cells remains unknown.
There is little doubt that these new reagents, once available to the general research community, will be invaluable in the further characterization and tracking of MAIT cells in health and disease states.
MAIT TCR recognition
The molecular basis of MR1 Ag presentation and its recognition by the MAIT TCR also was recently solved (20–22). MR1 adopts the overall binary structure of a class I MHC fold associated with the β2m subunit. Its ligand-binding cavity is smaller than that of classical MHC class I molecules and is mostly closed due to the presence of aromatic and basic residues, with the exception of a small opening on the top of the groove (Fig. 1). Plasticity in ligand conformation within the groove of MR1 molecules has been found for very similar ring-based ligands (20, 21) (Fig. 1), suggesting that the binding cavity might be able to accommodate a range of structures. In turn, this suggests that other, currently unknown, small molecules presented by MR1 might operate as Ags for MAIT cells, as well.
Interestingly, a common feature of all riboflavin metabolites that stimulate MAIT cells is the presence of a ribityl group, which is formed by the removal of one of the terminal hydroxyl groups of ribitol. This moiety emerges from the opening in the MR1 groove and is the only part of the Ag recognized by the MAIT TCR (Fig. 1). The MAIT TCR shows a highly conserved docking orientation, even when different Ags are presented by MR1 (20–22). This suggests that the MAIT TCR might act as an innate-like recognition receptor similar to the TCR expressed by iNKT cells (2). However, in contrast with iNKT cells, the orientation of the MAIT TCR on MR1 is more akin to other classical αβ TCR–MHC–peptide complexes, with a perpendicular orientation that straddles the α1 and α2 helices of MR1, rather than the parallel binding of the iNKT TCR on CD1d. The nearly invariant TRAV1-2–TRAJ33 TCR chain largely dominates the recognition of the MR1 molecule and of the Ag(s), thereby providing an explanation for its selective usage by MAIT cells (17, 20–23). Furthermore, the TRAJ33-encoded tyrosine residue at position 95 of the CDR3α loop, which is conserved in all MAIT TCRs, appears to behave as a sensor of Ag presentation by MR1 molecules. This residue is ideally positioned directly over the opening of the MR1 binding cavity and makes extensive contacts with MR1 residues. Additionally, contacts between this residue and the Ag ribityl group directly correlate with increased TCR engagement and subsequent MAIT cell activation (20, 21).
Although the nearly invariant TCRα-chain is found paired with a limited number of TCRβ-chains (TRBV20 [Vβ2], TRBV6 [Vβ13], and TRBV2 [Vβ22] in humans and TRBV19 [Vβ6], TRBV13 [Vβ8] in mice), their CDR3β loops are highly variable, and the role of the TCRβ-chain in the recognition of the complex remains unclear. In some cases, the TCRβ-chain appears to play a passive role in the activation of MAIT cells (23), whereas in others it modulates the overall affinity of the MAIT TCR and/or plays an active role in directly recognizing the Ag (21). These results suggest that the CDR3β variability may directly impact the extent of Ag reactivity by MAIT cells (21).
MAIT cell development
MAIT cell development is incompletely understood. In both mouse and humans, it is generally accepted that MAIT cells develop in the thymus (3, 6, 24). However, it remains unclear whether the developmental process is identical between the two species.
In the mouse, the vast majority of MAIT cells are positively selected by MR1-expressing bone marrow derived cells and not by the thymic epithelial cells (4). Based on data using TCR-transgenic mice, it was shown that MAIT cells can be selected by MR1 expressed on double-positive (DP) thymocytes (25). The idea that DP thymocytes represent the main MAIT-selecting cell type in the thymus is supported by the finding that mouse DP thymocytes express significant levels of MR1 molecules intracellularly (26). Therefore, in mice, MAIT cell development is clearly distinct from “conventional” T lymphocytes.
Positive selection of T lymphocytes on DP thymocytes was shown to impart the acquisition of “innate-like” characteristics (27) through a process that involves homodimeric interactions between SLAM family members expressed on cortical thymocytes, which signals through the SAP/Fyn/NF-κB pathway (28). Although both mouse MAIT and mouse iNKT cells likely encounter similar signals during their development, these two populations clearly diverge into unique developmental pathways. For example, the strong engagement of the iNKT TCR during positive selection (29–31), potentiated by the engagement of the SLAM family members (32), induces the transcription factor PLZF, which, in turn, is responsible for the effector phenotype of iNKT cells (28). Yet, MAIT cells in one of the transgenic models do not express PLZF (6). It is unclear what developmental clues are involved in setting the MAIT transcriptional program in mice. The memory/effector phenotype of mouse MAIT cells is acquired after birth in the periphery in a B cell– and microbiota-dependent manner (4).
Human MR1 also has been detected primarily on CD3+ CD45+ hematopoietic T cells (33), suggesting that human MAIT cells might also be selected by DP thymocytes, similar to what was described for mouse MAIT cells (25). This would be expected to drive the “innate-like” characteristics of MAIT cells and perhaps explain why MAIT cells in the human thymus express CD161 and both CD161 and IL-18Rα in human cord blood (6, 9, 33, 34). However, apart from these markers, human MAIT cells in the fetal thymus, spleen, mesenteric lymph nodes, and cord blood are otherwise immature functionally and phenotypically (24). Their maturation, which is associated with the acquisition of the PLZF transcription factor, occurs gradually in the mucosal tissues during fetal life and before exposure to environmental microbes and the commensal microflora (24). This process is likely independent of the SLAM/SAP pathway, because the number of MAIT cells was reported to be normal in five SAP-deficient patients (6), and SAP-deficient MAIT cells normally expressed PLZF (6).
MAIT cells and diseases
The role of MAIT cells is still enigmatic. The findings that MAIT cells recognize bacteria-infected cells in a MR1-dependent manner (15, 16) and their dependence on the microbiota for expansion in the periphery (4) pointed toward a potential role for MAIT cells in the immune response against bacteria. This was confirmed in several mouse models of infections using Escherichia, Mycobacteria, Klebsiella, and Francisella, in which a lack of MAIT cells correlated with increased bacterial loads by these species (12, 15, 35, 36). Furthermore, the number of MAIT cells in the blood of bacterially infected patients was shown to be severely decreased (16) and correlated with the increased susceptibility of hospitalized patients to secondary nosocomial infections (37).
MAIT cells are apparently also involved in noninfectious inflammatory disorders. For example, MAIT cells were found enriched within the livers of patients with chronic hepatitis C infection, autoimmune hepatitis, primary biliary cirrhosis, alcoholic liver disease, and nonalcoholic steatohepatitis (38, 39). Patients with inflammatory bowel diseases have a reduced frequency of MAIT cells in peripheral blood that is associated with an enrichment of MAIT cells in the inflamed tissue (40). MAIT cells also have been detected in the brain of patients with multiple sclerosis (41, 42) and in kidney and brain tumors (43, 44). A protective role for MAIT cells was reported in the mouse model of autoimmune encephalomyelitis (14), whereas MAIT cells were proposed as one of the effector cells contributing to inflammation in a mouse model of arthritis (45). Finally, MAIT cell numbers were found to decline upon HIV infection (46–48).
In all of these situations, it is not obvious whether bacteria-derived vitamin B metabolites might be implicated in the activation of MAIT cells through their TCR. Microbiota imbalance (49–51) and/or increased permeability of the mucosa (52) might indirectly influence MAIT cells in these instances. The underlying autoreactivity of some MAIT cells and/or their high expression levels of IL-12, IL-18, and IL-23 receptors might also play a role. Indeed, MR1-independent cytokine-driven activation of mice MAIT cells was reported for IL-12 (36) and IL-23 (45). The combination of IL-12 and IL-18 also was shown to stimulate human MAIT cells (53). Altogether, these results suggest that MAIT cells could potentially be involved in a much broader range of diseases.
During the past few years, a tremendous amount of knowledge, including some unexpected findings, has been acquired regarding the new population of T lymphocytes known as MAIT cells. First, MAIT cells represent a very sizable proportion of T lymphocytes in humans, where they sometimes are the majority of T cells found within a given organ. Second, the semi-invariant TCR expressed by MAIT cells recognizes the evolutionarily conserved monomorphic MHC-like molecule MR1 with a very conserved docking strategy. This strategy of conserved recognition is more reminiscent of pattern recognition receptor binding than the strategy of Ag discrimination usually carried out by conventional T cells. Third, MAIT cells have a memory-effector phenotype and can rapidly secrete the cytokines IFN-γ and IL-17, as well as the chemokine CCL20, upon Ag recognition. Fourth, they recognize vitamin B metabolites generated by certain bacteria and yeasts when presented by MR1 molecules.
Thus, the immune system has evolved a previously unrecognized means of host alert against microbial infections through the recognition of highly conserved ligands by a large population of innate T cells.
Although the role of MAIT cells in health and diseases is starting to be revealed, many questions regarding their functions and how they might be manipulated still remain. For example, we do not know how MAIT cell responses affect the rest of the immune system and/or the microbiota. We also do not know whether the MAIT cell response can be manipulated and/or whether memory MAIT cells exist. However, based on their cytokine-secretion profile, MAIT cells might be good candidate targets for new vaccines under development. Furthermore, because the proportion of MAIT cells in PBMCs appears to be inversely correlated with the infectious status of an individual, MAIT cells might represent a potentially attractive biomarker. Finally, the loss of MAIT cells following HIV infection, associated with their broad antibacterial activity, including against Mycobacterium tuberculosis, might lead to strategies aimed at restoring MAIT cell numbers. Because PBMC-derived MAIT cells are notoriously poor at proliferating in vitro, reprogramming of MAIT cells into induced pluripotent stem cells, followed by redifferentiation into MAIT cells, was recently performed (54). This might provide a new avenue for therapies using MAIT cells.
At any rate, it is clear that immunologists and clinicians should check the MAIT.
I thank current and past members of my laboratory for fruitful discussion and Jennifer Matsuda, Manfred Brigl, and Olivier Lantz for critical reading of the manuscript. I apologize to my colleagues whose work was not cited due to space constraints.
This work was supported by Grant AI092108 from the National Institutes of Health.
Abbreviations used in this article:
mucosal-associated invariant T.
The author has no financial conflicts of interest.