Polyclonal Mucosa-Associated Invariant T Cells Have Unique Innate Functions in Bacterial Infection

INTRODUCTION

Innate and adaptive components of the cellular immune system function sequentially to provide protection against microbial infection. Unlike conventional CD4 and CD8 T cells, which must undergo clonal expansion in response to infection, NK cells expressing germ line-encoded receptors respond rapidly to kill infected cells or release cytokines and provide the initial defense during infection. To kinetically bridge the innate and adaptive responses, innate T cells, such as invariant natural killer T (iNKT) cells and M3-restricted T cells, respond more quickly than conventional T cells to infection. Innate T cells are less dependent upon clonal expansion than conventional CD4 and CD8 T cells, because they appear to detect molecular patterns or danger signals rather than specific peptides bound to polymorphic major histocompatibility complex (MHC) molecules. For example, iNKT cells use an invariant T-cell receptor alpha (TCRα) chain (Vα24-Jα18 in humans and the homologous Vα14-Jα18 in mice) combined with a limited set of TCRβ chains to detect the nonclassical MHC class I (class Ib) molecule CD1d, which binds an assortment of self and microbial lipid antigens. However, iNKT cells display little antigen discrimination, since the invariant TCRα chain primarily engages the CD1d-lipid complex while the TCRβ chain makes limited contact (6). Additionally, iNKT cell activation can occur by various pathways, some of which do not require the iNKT detection of CD1d (8, 74). Therefore, with a limited TCR repertoire, iNKT cells can rapidly secrete cytokines in response to a wide range of infections by bacterial, viral, protozoan, or fungal pathogens. The immunomodulatory role of iNKT cells is well established, but the physiological role of iNKT cells in host immunity remains incompletely understood (4, 69). Mucosa-associated invariant T (MAIT) cells, a recently identified T-cell subset in mammals, share striking similarities and interesting differences with iNKT cells (73). MAIT cells have been implicated recently in detection of diverse microbial infections (33, 50), raising important questions of whether and how they might function in pathogen protection as an addition to the innate T-cell population.

Unlike iNKT cells, MAIT cells preferentially reside in mucosal tissues, and their homotypic expansion is dependent upon commensal microbes (72). In addition, MAIT cells are more abundant in humans than mice, whereas iNKT cells are more abundant in mice than humans (73). Similar to iNKT cells, MAIT cells express an invariant TCR that recognizes a novel nonpolymorphic class Ib molecule. More specifically, MAIT cells express an invariant TCRα chain (Vα7.2-Jα 33 in humans and the homologous Vα19-Jα33 in mice) that is preferentially associated with certain TCRβ chains (Vβ2, Vβ13, and Vβ22 in humans and Vβ6 and Vβ8 in mice) (45, 71). Most mouse MAIT cells are CD4/CD8 double negative (DN), but some are also singly positive for CD8 or CD4 coreceptors (45, 55, 72). Human MAIT cells have been reported in either the DN or CD8⁺ subsets (33, 55, 71). The developmental selection and peripheral activation of MAIT cells is restricted by the nonpolymorphic class Ib molecule MHC-related protein 1 (MR1) (72). The selection of MAIT cells in the thymus requires the expression of MR1 on hematopoietic cells, likely including double-positive (DP) thymocytes, macrophages (MΦ), and dendritic cells (DCs) (13, 55, 72). The Mr1 gene is highly conserved among mammals (64, 78, 79). The predicted amino acid sequences of mouse and human MR1 are 89%/90% identical in their α1/α2 domains. In contrast, mouse and human CD1 α1/α2 domains are only 60%/62% identical (38). Interestingly, the sequence analysis of Mr1 gene orthologs has indicated their strong purifying selection during evolution, suggesting their physiological importance (41). In functional support of this conclusion, mouse and human MAIT cells, like iNKT cells, have the ability to be activated in a cross-species manner (41).

The notion that MR1 presents antigen to MAIT cells is supported by extensive mutagenesis studies (39, 56) as well as by the fact that all αβ T cells characterized thus far detect MHC or MHC-like molecules bound by antigen. What is known about the MR1 presentation pathway is that MAIT cell development and activation is independent of transporters associated with antigen processing (TAP) (40, 71). Moreover, the activation of mouse MAIT cell hybridomas by MR1-overexpressing cell lines was found to be proteasome independent and enhanced by endocytic trafficking (40). Thus, MR1 clearly does not use the same antigen presentation pathway as classical MHC class I (class Ia) molecules. More interestingly, unlike class Ia or CD1 molecules, which are constitutively expressed on the plasma membranes of antigen-presenting cells (APCs), MR1 proteins reside mostly in the endoplasmic reticulum (ER), and only low levels have been detected at the plasma membrane (13, 35, 40). However, surface expression of mouse MR1 can be stabilized at the cell surface using selective monoclonal antibodies (MAbs) (13), whereas human MR1 was shown to be inducible on a human lung epithelial cell line after in vitro infection with Mycobacterium tuberculosis (33). Collectively, these findings demonstrate that MR1 activation of MAIT cells represents a novel and unique mechanism of αβ T-cell activation that has been conserved throughout mammalian evolution, suggesting physiologic importance in host immunity.

In support of their physiologic importance, two recent studies showed that human and mouse MAIT cells can detect microbial infection in an MR1-restricted manner (33, 50). In a human study, Gold et al. cloned M. tuberculosis-reactive CD8⁺ T cells from infected and uninfected individuals and found in both populations that a high frequency of the M. tuberculosis-reactive T cells are human MR1-restricted MAIT cells (33). Furthermore, these M. tuberculosis-reactive human MAIT cells are also activated by APCs infected with other bacterial strains, including Escherichia coli, Salmonella enterica serovar Typhimurium, and Staphylococcus aureus but not Listeria monocytogenes. A concomitant study by Le Bourhis et al. showed that both mouse and human MAIT cells display activation markers after in vitro coculture with APCs infected with various strains of bacteria and yeasts but not viral pathogens (50). This highly cross-reactive recognition of microbial-infected APCs by MAIT cells suggests that their activation may be more similar to pattern recognition than ligand discrimination. However, the upregulation of activation markers on MAIT cells was found not to be dependent upon known innate receptors or adaptors, including Toll-like receptors (TLRs), Nod-like receptors, MyD88, and TRIF (50). In contrast, innate receptor pathways have been shown to be important in iNKT cell activation both in vivo and in vitro (8, 9, 15, 59, 74, 75). Regarding the clinical relevance of MAIT cells, the two studies mentioned above reported that patients with ongoing M. tuberculosis infection have fewer MAIT cells in the blood and higher levels in the lung than healthy individuals (33, 50). This observation suggested that MAIT cells could migrate from the periphery into infected tissues to provide protection during infection.

To probe the role of MAIT cells in infection in vivo, antibacterial immune responses have been compared in MR1-deficient (MR1^−/−) and wild-type (WT) mice; however, findings using this approach thus far have been ambiguous, complicating mechanistic interpretations. For example, Le Bourhis et al. reported that there was no significant difference in bacterial burden between MR1^−/− and WT mice infected with Mycobacterium abscessus (50). This finding was discordant with their observation that M. abscessus activated MAIT cells in vitro. In a subsequent study, Georgel et al. reported that MR1 was not required for bacterial clearance after infection with E. coli, Shigella dysenteriae, or Yersinia enterocolitica, but MAIT cells were required for clearance of Klebsiella pneumoniae (32). It is unclear why in vitro activation experiments showed that mouse MAIT cells respond to APCs infected with diverse microbes, yet most in vivo experiments showed that MR1^−/− mice which lack MAIT cells are not impaired in their ability to control infection with most bacterial strains tested. A possible explanation of this conundrum is that WT mice lack a sufficient number of MAIT cells to show an antibacterial phenotype. Consistent with this explanation, Vα19-Jα33 TCR-transgenic mice that overexpress MAIT cells had lower bacterial loads than control mice after infection with E. coli or M. abscessus (50). However, because the TCR-transgenic mice used in these comparisons have few if any conventional CD4⁺ and CD8⁺ T cells, this model system lacks physiologic context or the ability to address the question of whether MAIT cells have a nonredundant role in controlling bacterial infection (12). Another possible explanation to the inconsistency between in vivo and in vitro studies is that perhaps an infection model in mucosal tissues, where MAIT cells mainly reside, needs to be employed. To our knowledge, in vivo infections that have been reported so far to compare MR1^−/− and WT mice are conducted through intraperitoneal injection (32, 50). The in vivo role of MAIT cells may be overlooked in the systemic challenge and therefore did not provide results that correlated with the in vitro reactivity to diverse microbes.

To determine whether MAIT cells function as innate T cells and what unique role they might play in bacterial infection, we employed in vitro and in vivo assays and identified a robust phenotype. Using the Mycobacterium bovis bacillus Calmette-Guerin (BCG) infection in vitro, we showed that MAIT cells secrete gamma interferon (IFN-γ), which enhances the antibacterial activity of infected MΦ in a manner dependent upon IL-12/23p40. Unexpectedly, the antibacterial activity in these in vitro assays was found to be largely independent of MR1 expression by infected MΦ, even though MR1 expression is required during ontogeny for MAIT cells to acquire their innate function. Furthermore, our in vivo studies using low-dose aerosol challenge showed that MR1^−/− mice had significantly higher bacterial loads in the lung after mycobacterial infection than WT mice. These findings demonstrate that the antibacterial activity of MAIT cells is unique and not redundant with the activities of other innate T-cell populations in the immune system.

MATERIALS AND METHODS

RESULTS

DISCUSSION

In this study, we provide in vitro and in vivo evidence that MAIT cells exert unique innate functions to control mycobacterial growth. In a coculture assay of purified MAIT cells and mycobacterium-infected MΦ, MAIT cells were found to curtail growth of intracellular bacterial replication associated with the secretion of IFN-γ and IL-17A, known mediators of immunity against intracellular and extracellular bacterial infection (17, 19, 22, 42, 47, 63, 80). Furthermore, the “first-line” production of IFN-γ and IL-17A by innate lymphocytes is known to play an important role in bridging innate and adaptive immunity (3, 21, 36, 46, 51, 67). Although both IFN-γ and IL-17A of MAIT cells were induced during the coculture with mycobacterium-infected MΦ, the control of mycobacterial growth was dependent only upon IFN-γ. Thus, IL-17A secretion by MAIT cells likely has a role other than the direct impact on bacterial intracellular growth in the mycobacterial infection model used here.

Importantly, the antibacterial activity of MAIT cells in our in vitro assays was found to be critically dependent upon their in vivo development in MR1-positive mice. Although in WT mice MAIT cell development is MR1 dependent, in transgenic mice expressing the MAIT cell canonical Vα19-Jα33 TCRα chain, Vα19i T cells can develop in both Mr1^+/+ and Mr1^−/− mice. However, only MR1-educated Vα19i T cells have a MR1-dependent bias in TCRβ chain usage and exhibit effector/memory and cytokine secretion profiles like nontransgenic MAIT cells (45, 55). In complete agreement with this conclusion from previous studies, only Vα19i transgenic T cells developing in the MR1-positive mice were found to control intracellular bacterial growth in our in vitro assays. Surprisingly, however, these MAIT cells selected by MR1 in vivo were found not to rely on MR1 engagement to impair intracellular bacterial replication in vitro. Indeed, this conclusion was supported by both the failure of anti-MR1 MAbs to block the antibacterial activity of MAIT cells and the observation that MAIT cells controlled bacterial growth in MR1^−/− MΦ. Thus, the MAIT cell effector function in vitro did not require TCR recognition of MR1 on infected APCs. In contrast, our observations suggest a noncognate pathway of MAIT cell activation mediated by the innate signal IL-12 during mycobacterial infection. Abolishing the IL-12 signal from infected APCs (with the use of anti-IL-12/23p40 MAb or p40^−/− MΦ) diminished the effector function of MAIT cells, including IFN-γ and IL-17A production as well as the inhibition of mycobacterial growth. Of note, when the IL-12 signal was prevented, anti-MR1 MAbs partially inhibited the anti-bacterial activity of MAIT cells, suggesting that an MR1-dependent component may be masked by a strong innate IL-12 signal.

Interestingly, besides infection, MAIT cells have been reported to play a regulatory role in mouse models of autoimmunity (10, 20, 58, 68). For example, Croxford et al. showed that MAIT cells provide protection in murine autoimmune encephalomyelitis (EAE) by inducing the IL-10 production by B cells (20). However, this protective mechanism was shown to be MR1 independent and partly ICOS dependent. More recently, Chiba et al. showed that MAIT cells adoptively transferred into MR1^−/− mice augment collagen antibody-induced arthritis (10). To explain how MAIT cells modulate the immune response in the absence of MR1, their in vitro analyses showed that MAIT cells are stimulated to proliferate by IL-1β and to produce IL-17A by IL-23 in the absence of TCR signals (10). In the present study, MAIT cells were found to release IFN-γ by the stimulation of IL-12 in the absence of infection. In accordance with these examples of cytokine driven activation, human MAIT cells express high levels of IL-12Rβ2 and IL-23R (2, 5). These combined findings point toward a model in which, after selection by MR1 in thymus, MAIT cells are programmed to respond to innate signals, at least some of which can occur without concurrent MR1 recognition. Alternatively, after egress from the thymus, MAIT cells adapt by quickly responding to infection (34). During infection, MR1 on either APCs or MAIT cells themselves may be a sufficient signature to allow MAIT cell activation by an innate cytokine signal.

Despite the above evidence of the non-MR1-dependent functions of MAIT cells, previously published studies have provided evidence for cognate interaction between MR1 and MAIT cells using the anti-MR1 MAbs used in this study (13, 33, 39–41, 50). In earlier mouse studies, MAIT cell hybridomas were selected based on TCR usage, and a small fraction of these MAIT hybridomas were found to be activated by MR1-overexpressing cell lines in a manner blockable by MAbs to MR1 or TCR (13, 39, 56). More recently, mouse MAIT cells from Vα19 TCR-transgenic mice as well as serologically isolated polyclonal human MAIT cells were found to express activation markers after culture with bacterium-infected cells in a manner that was also blockable with anti-MR1 MAb (50). In addition, numerous CD8⁺ human MAIT cell clones with activity against cells infected with M. tuberculosis and other bacteria were found to be blocked by anti-MR1 MAb (33). Although the above-mentioned studies provide compelling evidence for the occurrence of the MR1-dependent activation of MAIT cells, its overall importance relative to noncognate activation is difficult to assess without a better definition of functional heterogeneity within polyclonal MAIT cells.

As a notable parallel, the studies of iNKT cells provide a precedent for how innate T cells bearing limited TCR repertoires employ multiple mechanisms to control microbial infections (6, 8, 74). More specifically, iNKT cells react to microbial infections through at least three activation pathways: microbial antigen-driven, cytokine- and self-antigen-driven, and cytokine-driven pathways (8, 74). While microbial and self lipid antigens detected by iNKT cells have been identified (7, 8, 48, 74), recent studies show that the dominant IL-12 signal from APCs during microbial infection occurs independently of whether microbes express CD1d-presented antigens (9, 15). The IL-12 signal is thought to be an essential factor that enables iNKT cells to respond to a wide variety of microbes with limited TCR repertoires (8, 74). In the case of self-ligand recognition, IL-12 can provide costimulation to amplify a weak TCR-mediated signal (8, 9, 74). Moreover, during viral infection, cytokines, including IL-12 and IL-18, dominate the iNKT cell activation, while CD1d presentation is not required (shown by the use of CD1d^−/− mice and/or anti-CD1d MAb) (75). The constitutive expression of high levels of IL-12 receptor on iNKT cells provides an explanation for the immediate response of iNKT cells to cytokine-driven activation (9). It is likely that, since they share many innate-type features with iNKT cells, MAIT cells are also equipped for rapid responses using cognate and noncognate pathways of activation.

By bridging innate and adaptive immune responses, innate-type T cells are critical for providing the first-line protection against infection (65). In this role, different innate lymphocyte populations share redundant functions. For example, NK cells, γδ T cells, and iNKT cells contribute to innate IFN-γ and/or IL-17A production, but mice lacking functional NK cells, γδ T cells, or iNKT cells are not more susceptible to mycobacterial infection (24, 27, 31, 43, 49, 52, 69). Although MR1^−/− mice have normal development of these other innate lymphocytes (72), we show here that MR1^−/− mice are defective in the early control of mycobacterial infection. The inability of MR1^−/− mice to elicit a protective immune response is likely due to the defective IFN-γ and IL-17A production in the absence of MAIT cells. These findings demonstrate that MAIT cells have a nonredundant role as innate T cells. The infection model of mycobacteria has several advantages to define innate immune responses, including a lengthy doubling time of mycobacteria after lung infection and a slowly developing adaptive immune response (17).

Our in vivo findings supporting a role of MAIT cells in mycobacterial infection are congruent with published observations that MAIT cells are activated in vitro by mycobacterium-infected cells (33, 50). Similar to our findings, Georgel et al. showed that MR1^−/− mice infected intraperitoneally with K. pneumoniae failed to control bacterial burden, and serum cytokine analysis of the infected MR1^−/− mice showed a decreased proinflammatory response compared to WT mice (32). Again, congruent with these finding, in vitro studies showed that MAIT cells displayed activation markers after coculture with cells infected with K. pneumoniae (50). Curiously, however, while MAIT cells also showed strong in vitro reactivity to E. coli, another member of the family Enterobacteriaceae (50), MR1^−/− mice were not defective in controlling E. coli infection via the intraperitoneal route (32). Similarly, cells infected with M. abscessus activated MAIT cells in vitro, but MR1^−/− mice were not defective in controlling bacterial growth in vivo (50). However, Mr1^+/+ mice, in which all T cells expressed the transgene of invariant TCRα chain of MAIT cells, had lower bacterial loads after infection with E. coli or M. abscessus than Mr1^−/− mice with T cells expressing the same transgenes (50). This finding suggested that overexpression of MAIT cells may be required to detect their in vivo functions. However, our report and the report by Georgel et al. (32) would suggest otherwise. Thus, although MAIT cells are activated by diverse microbes in vitro, we speculate that their in vivo roles are most prominent when infections are introduced into mucosal tissues, when bacterial replication in vivo allows a window to monitor innate effector function, and when appropriate innate effector functions are monitored.

Recent findings and speculations of the importance of MAIT cells raise the question of why they were not detected sooner. Indeed, MAIT cells were likely a factor in previously reported findings comparing β₂-microglobulin (β₂m)-deficient mice and CD1d-, H2-M3-, or MHC class Ia-deficient mice (14, 23, 30, 70). Surface expression of MR1 requires β₂m assembly (56), and β₂m^−/− mice have been shown to be more susceptible to infections with K. pneumoniae (14) and M. tuberculosis (23) than mice lacking Kb- and Db-restricted CD8⁺ T cells or CD1d-restricted iNKT cells. MAIT may also have been involved in the previous studies of nonclassical T cells (either CD8⁺ or CD4/CD8 DN population) in the murine models of bacterial infection. CD8⁺ T cells presumably restricted by an undefined class Ib molecule represented a high percentage of IFN-γ producing cells in the lung of Kb^−/− Db^−/− mice after aerosol infection with M. tuberculosis (76). A comparative study using Kb^−/− Db^−/− and Kb^−/− Db^−/− M3^−/− showed that non-H2-M3 MHC class Ib-restricted CD8⁺ T cells express activated surface markers, exhibit cytotoxicity, and produce cytokines, including IFN-γ and IL-17A, in response to L. monocytogenes infection (11). Although this response was not blocked by anti-MR1 MAb, based on our findings this could be due to non-MR1-dependent activation of MAIT cells. Moreover, it is possible that the role of the MAIT cells in the CD4/CD8 DN T-cell population is overlooked in the studies focusing on CD8⁺ T cells in Kb^−/− Db^−/− or Kb^−/− Db^−/− M3^−/− mice (11, 76). In fact, a rare population of CD4⁻ CD8⁻ NK1.1⁻ αβ T cells in WT mice was shown to inhibit the intracellular growth of M. tuberculosis and Francisella tularensis (LVS) (18, 19), and these CD4/CD8 DN T cells secreted high levels of IFN-γ and IL-17A in response to F. tularensis LVS infection (19). These parallels between previous studies and recently defined features of MAIT cells support the notion that MAIT cells are key innate effectors against infection by diverse microbes.

To our knowledge, this study is the first to investigate the in vivo role of MAIT cells in a mucosal infection model. Evidence is provided that MAIT cells have an innate effector function and play a nonredundant role in the immune system during bacterial infection. Mechanistically, we show that MAIT cells control bacterial growth and produce proinflammatory cytokines. Further characterization of the mechanisms by which MR1 and MAIT cells are involved in the induction of protective immunity may help in the future design of vaccines against bacterial infection.

Supplementary Material