Mucosal-Associated Invariant T Cells Augment Immunopathology and Gastritis in Chronic Helicobacter pylori Infection

Mucosal-associated invariant T (MAIT) cells are a subset of T cells, which are abundant in humans (1–8% in blood), but relatively rare (∼0.1% in blood) in specific pathogen-free (SPF)-housed laboratory mice (1–3). MAIT cells typically express a semi-invariant αβ TCR, which in humans comprises TRAV1-TRAJ33 (Vα7.2-Jα33/20/12) TCR α-chain preferentially assembled with TRBV20⁺ (Vβ2) or TRBV6⁺ (Vβ13) TCR β-chains. In mice TRAV1-TRAJ33 (Vα19-Jα33) is mostly paired with TRBV19 (Vβ6) or TRBV13 (Vβ8) (3, 4). MAIT TCRs are restricted by the monomorphic MHC-related protein 1 (MR1) (5).

MR1 presents vitamin B–related Ags, including 5-(2-oxopropylideneamino)-6-d-ribitylaminouracil (5-OP-RU) and 5-(2-oxoethylideneamino)-6-d-ribitylaminouracil (5-OE-RU), which are derived from a precursor of microbial riboflavin (vitamin B2) biosynthesis and activate MAIT cells in vitro (6–8). Indeed, a correlation between the presence of the riboflavin synthesis pathway in diverse microbes, including bacteria and yeast, with MAIT cell activation, led to the discovery of this pathway as furnishing this new class of Ag (6, 9). In contrast, microbes that do not synthesize riboflavin, such as Listeria monocytogenes and Enterococcus faecalis, do not activate MAIT cells in a TCR-dependent manner (10). Thus, MAIT cells, which are enriched in tissues including lungs, intestine, and liver, are ideally placed to detect metabolically active microbes producing these Ags at mucosal sites. MAIT cells have been shown to respond, in vivo, to bacteria that produce the riboflavin-derived Ags, or to synthetic 5-OP-RU, in the presence of TLR agonists (1).

MAIT cells rapidly secrete cytokines (including IFN-γ, IL-17, and TNF) upon stimulation and play a protective role in some model bacterial infections, including Klebsiella pneumoniae, Mycobacterium tuberculosis, M. bovis bacillus Calmette–Guérin, and Francisella tularensis (11–14). They can also be activated in a TCR-independent manner by cytokines or viral infections (15, 16). Correlations between MAIT cell number and disease severity have been demonstrated for a range of chronic inflammatory conditions, including multiple sclerosis, rheumatoid arthritis, inflammatory bowel diseases, celiac disease, and ankylosing spondylitis (reviewed in Ref. 17), suggesting that MAIT cells may also contribute to immunopathology. The basis of these correlative associations could involve MAIT cell activation, but might also be attributable to patient treatment (18), which has not been considered in all studies. Few studies to date have attempted to understand the potential role of MAIT cells in immunopathology. Notably, Helicobacter pylori, which synthesizes riboflavin, and thus likely produces the known MAIT cell stimulatory Ags, has been shown to stimulate human MAIT cells in vitro (19). Indeed, one study showed a reduction in MAIT cell numbers in H. pylori–infected patients’ blood versus healthy controls, and human gastric MAIT cells can produce cytokines in response to H. pylori–infected macrophages in vitro (19). Hence, we reasoned such a role might be demonstrated in unresolved (i.e., chronic) bacterial infection and tested this idea in a model of H. pylori gastritis.

H. pylori are Gram-negative extracellular bacteria that colonize the stomach. A large proportion of the world’s population is infected with H. pylori and in most people, unless treated with antibiotics, the infection persists for decades. Most infected individuals remain asymptomatic, but 15–20% develop severe gastritis and overt disease such as peptic ulcers. A smaller proportion (1–3% of cases) develops gastric adenocarcinoma and fewer again (0.1%) MALT lymphoma (20). Infection with H. pylori induces both innate and adaptive immune responses by the host, generally Th1-biased, and tempered by regulatory T cells (reviewed in Ref. 21).

Mouse models of H. pylori infection have been widely used to understand disease development, and show the importance of CD4⁺ T cells including Th17 cells (22, 23). However, to date, no analysis of MAIT cells in mouse models of H. pylori has been reported, perhaps in part because of the low abundance of MAIT cells in mice and low numbers of lymphocytes in uninfected stomachs, and because MR1-tetramers that precisely identify murine MAIT cells have only recently been developed (2, 3, 6). The production of proinflammatory cytokines including IL-17 and IFN-γ by MAIT cells (2, 12, 24) suggests a potential role in the chronic inflammation (gastritis) that is characteristic of H. pylori infection, and possibly in the subsequent development of more severe MALT lymphoma and adenocarcinoma that occur in some patients. Because the role of MAIT cells in the immune response to H. pylori remains unclear, this circumstantial evidence prompted us to address this question experimentally, using Ag-specific MR1-5-OP-RU tetramers (MR1-tetramers) to detect MAIT cells (1–3, 6) thereby removing ambiguity inherent in surrogate markers of these cells. Here we use both Vα19iCα^−/− MAIT TCR transgenic mice and wild-type (wt) C57BL/6 mice with preboosted MAIT cells to investigate the role of MAIT cells in H. pylori infection, revealing for the first time, to our knowledge, their pathogenic potential in chronic bacterial infection.

Human gastric tissue, normally discarded following sleeve gastrectomy bariatric surgery, was obtained through The Avenue Hospital, Melbourne, with written consent from patients, after approval from The University of Melbourne and The Avenue Human Research Ethics Committees. All patients were female, between 27 and 57 y of age. Gastric mucosa was dissected from underlying tissue, rinsed with ice-cold HBSS, snap frozen in OCT and sectioned (8 μm) using a Leica CM3050 S cryostat. Acetone-fixed tissue was blocked with serum-free protein block (DAKO) for 30 min and/or 10% goat serum (Sigma) at room temperature (RT), prior to detection with primary Abs targeting CD3 (CD3-12; Bio-Rad), TRAV1-2 (3C10; BioLegend), and H. pylori rabbit polyclonal (DAKO) diluted in 2.5% normal goat serum, followed by Alexa Fluor 647 goat anti-rat, Alexa Fluor 488 goat anti-mouse, and Alexa Fluor 555 goat anti-rabbit secondary Abs, respectively (Life Technologies). Nuclei were counterstained with Hoechst 33342 (Life Technologies) prior to mounting with ProlongGold (Life Technologies). Slides were analyzed using Carl Zeiss LSM700 confocal microscope with a 20× objective and Carl Zeiss Zen software at the Biological Microscopy Platform at The University of Melbourne. Images were prepared using FIJI/ImageJ software.

Following dissection of mucosa, tissue was chopped into 2–5 mm pieces, then incubated in EDTA/BSS, 5% FCS, 1 mM DTT, and shaken at 150 rpm for 15 min at 37°C. Released cells were passed through a 70-μm strainer and washed in EDTA/BSS + 5% FCS. PBMCs were prepared by Ficoll-Paque density gradient separation from ∼20 ml fresh blood. Cells were frozen in FCS + 10% DMSO and stored in liquid N₂ until use. Upon thawing, cells were washed, rested for 1–2 h and filtered (70 μm) before staining. Cells (from 3 × 10⁵ to 2 × 10⁶) were stained with Ab to CD3 (OKT3 AF700 or PE-CF594; eBioscience), CD161 (HP-3G10, PE-Cy7; BioLegend), TCR Vα7.2 (3C10, BV875; BioLegend), human MR1-tetramer (PE, prepared in-house) and 7AAD (7-aminoactinomycin D, 1:500) or Live/Dead Aqua (ThermoFisher 1:500), and were fixed with 1% paraformaldehyde before analysis on a BD LSR Fortessa Flow Cytometer (BD Biosciences). Data were analyzed with FlowJo software (v10; Tree Star). Cells were gated sequentially for lymphocytes (forward scatter [height] versus side scatter [height]), single cells (forward scatter [height] versus forward scatter [area]), live cells (7AAD⁻), T cells (CD3⁺), and MAIT cells (CD161⁺MR1-tetramer⁺).

MAIT cell reporter activation and MR1 expression upregulation assays were performed essentially as previously reported (9). Jurkat cells overexpressing the MAIT TCR clone AF-7 (Jurkat.MAIT) were tested for activation by coincubation with compounds and C1R cells overexpressing MR1 (C1R.MR1) for 16 h. Cells were subsequently stained with PE-Cy7–conjugated anti-CD3 (UCHT1, 1:300; eBioscience), and APC-conjugated anti-CD69 (1:25; BD Biosciences) before analysis on a FACSCanto II (BD Biosciences) flow cytometer. Activation of Jurkat.MAIT was measured by an increase in surface CD69 expression. In some experiments, blocking Ab (prepared in-house from hybridoma anti-MR1 mAb 26.5, a gift from Dr. T. Hansen, Washington University School of Medicine, St. Louis, MO) (25) or isotype control 8A5 (prepared in-house) were added (final 10 μg/ml) prior to addition of stimuli. For MR1 expression, cells were additionally stained with biotinylated anti-MR1 mAb 26.5, followed by PE-conjugated streptavidin. Control Jurkat.LC13 cells (9) were activated by C1R cells expressing HLA-B8 in the presence of the EBV peptide FLRGRAYGL (FLR).

5-OP-RU was prepared as described previously (26). 6-Formyl pterin (6-FP) was purchased from Schircks Laboratories (Jona, Switzerland). TLR2/6 agonist Pam2Cys (27) was chemically synthesized and functionally verified.

All C57BL/6 mice, MR1^−/− mice, Vα19iCα^−/−MR1⁺ mice, and Vα19iCα^−/−MR1^−/− mice (all on C57BL/6 genetic background) were bred and housed in the Biological Research Facility of the Doherty Institute. MR1^−/− mice were generated by breeding Vα19iCα^−/−MR1^−/− mice (28) obtained from S. Gilfillan (Washington University School of Medicine, St. Louis, MO) with C57BL/6 mice, and intercrossing of F1 mice, and verified as described (1). Female mice aged between 6 and 12 wk were used in all experiments following approval by The University of Melbourne Animal Ethics Committee.

H. pylori SS1 (29) was cultivated as described previously (30). Mice were infected orogastrically with H. pylori (10⁷ CFU) suspended in 100 μl brain heart infusion broth (BHI). Salmonella enterica serovar Typhimurium BRD509 (31) and the S. Typhimurium BRD509ΔribDH mutant (1) have been previously described. For S. Typhimurium ΔribDH growth, culture medium (Luria broth) was supplemented with 20 μg/ml riboflavin. Mice were inoculated under isofluorane anesthesia with S. Typhimurium BRD509 (10⁶ CFU) or ΔribDH (10⁷ CFU unless otherwise stated) or Ags (1.52 μM 5-OP-RU or 6-FP alone in 50 μl, or in 45 μl plus 5 μl of 20 nmol Pam2Cys) prepared in 50 μl PBS, via the intranasal route.

Single cell suspensions of lungs, blood, mesenteric lymph nodes, and spleen were prepared as described previously (1). Gastric immune cells were isolated by perfusion from longitudinally halved stomachs, as described previously (32).

Mouse and human MR1-5-OP-RU labeled with BV421 or PE and MR1-6-FP (unlabeled) tetramers were generated as described previously (6) and prepared simultaneously.

Abs against CD19 (1D3, PerCP-Cy 5.5), CD3 (UCHT1, PE or 145-2C11, PE-Cy7), CD4 (GK1.5, APC-Cy7), CD45.2 (104, FITC), CD69 (FN50, APC), CD8α (53-6.7, PE), and TCRβ (H57-597, APC or FITC) were purchased from BD Pharmingen. Abs against MHC II (M5, Alexa Fluor 700), PLZF (Mags.21F7, PE), RORγt (B2D, APC), and T-bet (4B10, PE-Cy7) were purchased from eBioscience. To block nonspecific staining, cells were incubated with MR1-6-FP tetramer (unlabeled, 1:100) and anti–Fc receptor (2.4G2) for 15 min at RT prior to Ab staining. Cells were then incubated at RT with Ab cocktails including MR1-tetramer in PBS/2% FCS. Cells were then further incubated with 7-aminoactinomycin D (5 μl per sample) for 10 min in the dark at RT and fixed with 1% paraformaldehyde prior to analysis on LSR Fortessa (BD Biosciences). Data were analyzed with FlowJo software (Version 10, Tree Star). Cell counts were obtained using blank calibration particles (BD Pharmingen) or by flow rate calibration (33) (for Fig. 4C). Transcription factor staining was performed using a commercial transcription buffer staining set (eBioscience) according to the manufacturer’s instructions.

Longitudinally dissected half stomachs were fixed in 10% neutral buffered formalin, embedded in paraffin; 5-μm sections were stained with H&E and scored by a blinded operator under light microscopy. A pathologist also scored sections independently and the grading system is based on a previously described system (34) with modifications. Briefly, inflammation was assessed in two separate tissue sections for each animal using three parameters: 1) neutrophil infiltration and 2) lymphocyte infiltration: scored 0–4 on the basis of number of cells and whether widespread or multifocal infiltrate was observed, and 3) atrophy: scored 0–4 on the basis of loss of chief and parietal cells from the gastric glands.

Single cells were sorted from S. Typhimurium BRD509-primed, H. pylori–infected C57BL/6 mouse stomachs on a FACSAria flow cytometer. PCR amplification and TCR sequencing was performed as previously described (3).

Statistical tests were performed using Prism software (version 6; GraphPad software, La Jolla, CA). Comparisons between groups were performed using one-way ANOVA, Student t test, Mann–Whitney, Spearman correlation, or Kruskal–Wallis test as indicated in the figure legends.

We first tested whether H. pylori synthesize Ags that stimulate MAIT cells. Previously, we have shown that both human and mouse MAIT cells are stimulated by Ags, including 5-OP-RU and 5-OE-RU, formed from the nonenzymatic reaction of a microbial riboflavin biosynthetic precursor with small metabolites (6, 9). H. pylori and other Helicobacter species contain an intact riboflavin biosynthesis pathway (Kyoto Encyclopedia of Genes and Genomes database), and thus were predicted to produce MAIT cell–stimulating riboflavin biosynthesis–derived Ags. Indeed, riboflavin production appears to be an essential metabolic pathway for H. pylori (35). To verify MAIT cell stimulation by H. pylori, we tested culture supernatant from H. pylori SS1 cultures for its ability to stimulate a reporter MAIT cell line, Jurkat.MAIT TRBV6-1 (9, 36). Similar to S. Typhimurium, previously shown to stimulate Jurkat.MAIT cells in this assay (6, 9), H. pylori culture supernatant caused upregulation of CD69 on Jurkat.MAIT cells in the presence of MR1 expressed on C1R cells (C1R.MR1) (Fig. 1A). As with the activation previously demonstrated with S. Typhimurium supernatant (9) this could be specifically blocked with the anti-MR1 mAb 26.5, but not an isotype control mAb (Fig. 1A), indicating that there is minimal nonspecific MAIT cell activation. Importantly, Jurkat cells expressing an irrelevant TCR (Jurkat.LC13) were not activated under the same conditions (Fig. 1B). These data indicated that H. pylori are capable of producing a MAIT cell–stimulating Ag, which is presented by MR1. In addition, H. pylori supernatant caused upregulation of MR1 on C1R.MR1 cells (Fig. 1C), similar to that previously demonstrated with S. Typhimurium supernatant or synthetic MR1 ligands (9, 37), a further confirmation of the presence of MR1-binding Ag.

Booth et al. (19) recently reported the presence of MAIT cells in human gastric biopsies detected with CD161 and TRAV1-2 markers. To confirm their observation, we examined stomach tissue obtained from patients following sleeve gastrectomy bariatric procedures by immunofluorescence microscopy. In three patients we observed CD3⁺TRAV1-2⁺ cells within the gastric mucosa in close proximity to H. pylori (Fig. 2A). As we previously reported, a proportion of TRAV1-2⁺ cells are not MAIT cells, as judged by MR1-tetramer staining (3). Thus, we examined single cell preparations by flow cytometry using MR1-tetramers. MAIT cells could be detected in single cell preparations of gastric tissue where they represented ∼1% (0.4–2.39) of T cells (Fig. 2B, 2C). Costaining with TRAV1-2 Ab and MR1-tetramer confirmed that the majority of TRAV1-2⁺CD161⁺CD3⁺ cells were MAIT cells (Fig. 2D). Interestingly, there was a higher proportion of MR1-tetramer⁻,TRAV1-2⁺ cells in the gastric tissue (∼25%), whereas in PBMCs almost all TRAV1-2⁺ cells were also MR1-tetramer⁺, thus highlighting the importance of defining MAIT cells using MR1-tetramers, particularly in studies involving cells isolated from tissues. Notably, there was a trend toward higher MAIT cell percentage in both PBMCs and gastric mucosae in patients where H. pylori were observed (Fig. 2C).

The presence of MAIT cells in gastric tissue and the ability of H. pylori to activate Jurkat.MAIT cells in vitro suggest a potential role for MAIT cells in the immune response to H. pylori infection. To test this experimentally, in a model allowing controlled infection, we directly assessed the behavior and function of MAIT cells in the mouse model of H. pylori SS1 infection (29). MAIT cells are rare in naive, SPF-housed wt C57BL/6 mice compared with humans. In the stomach of naive mice they represent only ∼1% of αβ T lymphocytes and <∼100 MAIT cells are recovered from each stomach (Fig. 3A, 3B). In mice, there are no MAIT cell–specific Abs and thus MR1-tetramers are necessary for MAIT cell detection. To determine if MAIT cells increase in number in response to H. pylori infection in vivo, we infected C57BL/6 mice with 10⁷ CFU H. pylori SS1 orally, and examined the MAIT cells in stomachs by flow cytometry, with MR1-tetramers, 12 wk postinfection. To assess pathology, histological assessment by H&E staining was conducted. Sections were scored blindly for cellular infiltration (neutrophilic and lymphocytic) and atrophy, a classical characteristic of chronic gastritis where there is a loss of chief and parietal cells, and a critical step in the development of gastric cancer (38).

There was a significant increase in the number of MAIT cells, detected with MR1-tetramers, in the stomachs of infected mice compared with naive controls (Fig. 3A, 3B, left panel). An increase in non-MAIT αβ T cells was also observed in both infected wt mice and in MR1^−/− mice, which lack MAIT cells, compared with uninfected mice (Fig. 3B, right panel). Although no significant differences in absolute T cell numbers or gastric pathology scores (Fig. 3D) were observed between wt and MR1^−/− mice (Fig. 3B, 3C), the absolute number of MAIT cells, but not non-MAIT αβ T cells, correlated with an increase in gastric atrophy at 12 wk postinfection (Fig. 3C), suggesting a role for MAIT cells in driving pathology during chronic H. pylori infection.

Given the low numbers of T cells that can be isolated from naive stomachs, the analysis of MAIT cells is difficult in wt mice. Hence, we reasoned that increasing MAIT cell numbers would facilitate investigation of the role of these cells in immune protection or pathology in a bacterial infection where a complex immune response exists. One way to examine this is with the use of MAIT TCR transgenic mice (Vα19iCα^−/−MR1⁺), which have been used previously because of the low numbers of MAIT cells in wt SPF mice (28). MAIT cells from Vα19iCα^−/−MR1⁺ mice are abundant, produce IFN-γ and TNF, and proliferate in vitro in response to specific MAIT cell Ag 5-OP-RU (3).

Accordingly, we infected Vα19iCα^−/− transgenic mice on MR1⁺ and MR1^−/− backgrounds with H. pylori and then isolated single cell suspensions from the stomachs for flow cytometric analysis of MAIT cells at several time points. Following infection with H. pylori, MR1-tetramer reactive cells significantly increased as a proportion of αβ T cells from ∼30% to ∼60%, and in absolute numbers, by 4 wk postinfection in the Vα19iCα^−/−MR1⁺ mice as compared with the control Vα19iCα^−/−MR1^−/− mice, and remained high at 12 wk postinfection (Fig. 4A–C). MR1-tetramers are highly specific for MAIT cells, but in Vα19iCα^−/− mice, where the TCR repertoire is abnormally skewed toward selection of the MAIT Vα, a small proportion of “MAIT-like” cells also bind. These are thought to be MHC-restricted (14). A small population of these cells was noted in Vα19iCα^−/−MR1^−/− mice (Fig. 4A–C). Overall, the increase in tetramer-positive cells in Vα19iCα^−/−MR1⁺ mice upon infection suggests that MAIT cells are capable of responding not only to H. pylori Ags in vitro, but also to infection in vivo.

We next determined whether MAIT cells accumulating in the stomach during chronic H. pylori infection contribute to pathology. For this, Vα19iCα^−/−MR1⁺ mice and Vα19iCα^−/−MR1^−/− mice were infected with H. pylori and stomachs processed for histological assessment by H&E staining at various time points postinfection. Sections were scored blindly for cellular infiltration (neutrophilic and lymphocytic) and atrophy. An increased cellular infiltrate was seen in the stomachs of Vα19iCα^−/−MR1⁺ mice relative to Vα19iCα^−/−MR1^−/− mice at 12 wk (Fig. 4A, 4D), consistent with the increase in MAIT cells identified by tetramer staining (Fig. 4B, 4C), and as early as 2 wk in some mice (data not shown). By 12 wk postinfection, most Vα19iCα^−/−MR1⁺ mice developed severe atrophy (Fig. 4D, 4E), suggesting that MAIT cells not only accumulate postinfection, but also contribute to pathology.

MAIT cell accumulation and concomitant enhanced pathology was observed in Vα19iCα^−/−MR1⁺ MAIT TCR transgenic mice suggesting a role for MAIT cells in this process. Given the vastly expanded number of MAIT cells in these mice, it was important to confirm this finding in wt mice. Inbred wt SPF-housed mice have fewer MAIT cells in the blood than humans, and it has been hypothesized that a history of natural infections in humans boost MAIT cell numbers (39). Less than 100 MAIT cells were recoverable in single cell suspensions from the stomachs of naive wt C57BL/6 mice (0.5–1% stomach T cells) (Fig. 3A, 3B), compared with ∼1–3% in the human gastric mucosa (19) (Fig. 2B). We reasoned that boosting MAIT cell numbers in wt C57BL/6 mice by prior bacterial infection might expand MAIT cells, recapitulating the case in humans, and hence reveal their role in gastric infection as observed in the Vα19iCα^−/−MR1⁺ mice.

For this purpose, we used a recently developed model of MAIT cell enrichment in wt C57BL/6 mice (1) that involves either intranasal infection with S. Typhimurium or coinoculation of MAIT Ag and TLR agonists. Following boosting, MAIT cells rapidly expand in an MR1-dependent manner, comprising ∼50% of T cells in the lungs at day 7 postinfection. Moreover, MAIT cells are redistributed to other organs and tissues, and their frequencies remain elevated up to 10 wk postinfection (1). Thus, we consider that this model more closely resembles MAIT cell distribution in adult humans where a history of infection and bacterial colonization may lead to a greater number of MAIT cells than observed in SPF-housed mice. In order to determine the effect of boosting MAIT cell numbers on subsequent H. pylori infection, wt C57BL/6 mice were infected with S. Typhimurium BRD509 and after 6–10 wk, when the Salmonella bacteria are cleared to below detectable levels, mice were challenged by gastric infection with H. pylori SS1 (Fig. 5A). As expected, prior pulmonary infection with Salmonella not only expanded the proportion of MAIT cells in the lungs and blood of infected mice, but also in the stomachs (Fig. 5B). Notably, in the stomach, the proportion of MAIT cells increased to ∼6% of T cells (Fig. 5B). When preprimed mice were subsequently challenged with H. pylori, there was an increase in MAIT cell numbers as early as 2 wk postinfection with a significant increase over time (Fig. 5C).

We next examined the phenotype of the MAIT cells accumulating after H. pylori infection in mice that were preboosted with S. Typhimurium. MAIT cells isolated from both the lungs and the stomach retained an “effector memory” phenotype, as previously described (1, 24), being CD44⁺ and CD62L^lo (Fig. 6A). The majority of gastric MAIT cells expressed CD69 and CD103, consistent with a tissue-resident phenotype (19) (Fig. 6B); this was also true for non-MAIT αβ T cells accumulating in the stomach after 8 wk of infection with H. pylori. There was a preferential accumulation of CD8⁺ MAIT cells in the stomach (Fig. 6C) and multiplex PCR and sequencing of single cells showed a TCR repertoire consistent with MAIT cells (Table I). MAIT cells are known to express a transcriptional signature that includes the regulators PLZF, RORγt, and T-bet. Following H. pylori infection we found that stomach MAIT cells showed a mixture of PLZF⁺ and PLZF⁻, as well as RORγt⁺ and RORγt⁻ cells. In most MAIT cells there was coordinated expression of these two markers, with approximately half of all MAIT cells in mice infected with H. pylori lacking detectable expression of either transcription factor (Fig. 7). T-bet was expressed on almost all MAIT cells after H. pylori infection (87–98% in primed and H. pylori–infected, and 91–93% in H. pylori only) (Fig. 7), and the small number of cells lacking T-bet were either RORγt and PLZF positive or negative. Thus, MAIT cells accumulating in the stomach after H. pylori infection have a transcription factor profile consistent with Tc17/Tc1-type cells characteristic of proinflammatory capability.

In order to understand the implications of MAIT cell accumulation during gastric infection, we examined the effect of MAIT cell induction on the non-MAIT αβ T cell response. In preprimed mice the increase in MAIT cells was accompanied by an increase in non-MAIT αβ T cells, observed at 8 and 12 wk postinfection. The infiltration of non-MAIT T cells was delayed in mice lacking MR1 (C57BL/6.MR1^−/−), indicating an MR1-dependent and therefore likely MAIT cell–driven effect on T cell recruitment (Fig. 8A). An important consideration is that, in addition to MAIT cell activation, Salmonella infection is expected to prime a complex immune response. To address this, we infected mice with a mutant strain of S. Typhimurium BRD509 in which key genes involved in riboflavin synthesis, ribD and ribH, are deleted (Salm.ΔribDH) (1). These mice did not accumulate MAIT cells in their stomachs following subsequent H. pylori infection (Fig. 8B). However, the response could be rescued when BRD509ΔribDH was supplemented with synthetic 5-OP-RU Ag (Fig. 8B). This is consistent with our previous studies, which showed that MAIT cell responses are dependent on microbial riboflavin metabolites (1, 6, 9). The non-MAIT αβ T cell accumulation was also partially dependent on the presence of 5-OP-RU, supplied either by the riboflavin producing S. Typhimurium BRD509 bacteria or by adding exogenous synthetic ligand (Fig. 8C), suggesting a role of MAIT cells for recruiting other T cells. Similar to our previous observations in the lung model (1), this result could be reproduced by intranasal introduction of the specific MAIT cell Ag 5-OP-RU and a costimulus such as the TLR2/6 agonist S-[2,3-bis(palmitoyloxy)propyl] cysteine (Pam2Cys) (40). In contrast, MAIT and non-MAIT αβ T cell accumulation was much less apparent when Pam2Cys was combined intranasally with the MR1 ligand 6-formyl pterin (6-FP, a folate metabolite), which does not stimulate MAIT cells (9) (Fig. 8D, 8E). Thus, Ag-specific priming of MAIT cells leads to accumulation of non-MAIT T cells in the stomach upon H. pylori infection above the direct accumulation caused by infection or TLR ligation in the absence of MAIT cell Ag.

T cells and other cell types, including neutrophils and macrophages, have been implicated in pathology (gastritis) following H. pylori infection in humans and mouse models (41, 42). In mice preprimed with either S. Typhimurium BRD509 or Pam2Cys plus 5-OP-RU, and then infected with H. pylori, we observed, in addition to MAIT cell (Figs. 5C, 8) and non-MAIT αβ T cell (Fig. 8) accumulation, an increase in other immune cells including neutrophils, macrophages, eosinophils, and dendritic cells (Fig. 9A–D). This increase was significantly higher than in similarly primed and infected MR1^−/− mice, indicating a role for MAIT cells in the recruitment of other cell types. The MAIT cell accumulation and recruitment of other cell types was associated with gastritis, which was evident on H&E-stained stomach sections (Fig. 9E), and in significantly increased pathology scores in wt mice, compared with MR1^−/− mice, for lymphocytic infiltrate and atrophy. Strikingly, all MAIT preboosted wt mice infected with H. pylori developed atrophic gastritis by 8 wk postinfection (Fig. 9F); comparatively, only 2 of 5 MR1^−/− mice developed atrophic gastritis (score 1 out of 4). Gastritis was also observed following H. pylori infection in wt mice primed with riboflavin-deficient S. Typhimurium (Salm.ΔribDH) or with Pam2Cys, but this was significantly increased when the specific MAIT Ag 5-OP-RU was also added (Fig. 10), further demonstrating that MR1-Ag dependent MAIT cell activation drives gastric pathology.

We, and others (19), have observed the presence of MAIT cells in human gastric mucosae, prompting us to examine the role of MAIT cells in chronic H. pylori infection in mouse models. This study demonstrates, through a number of lines of evidence, that MAIT cells play an important role in gastric pathology following H. pylori infection. In wt C57BL/6 mice, following H. pylori SS1 infection, the absolute numbers of MAIT cells, as well as non-MAIT αβ T cells were increased in the stomach, and there was a correlation between MAIT cell numbers and pathology. Given that MAIT cells were virtually undetectable in stomachs of uninfected SPF-housed wt C57BL/6 mice, we next examined Vα19iCα^−/−MR1⁺ TCR transgenic mice. In these mice a rapid MR1-dependent expansion of MAIT cells, which are already in high percentage in these mice in homeostasis, was observed following gastric infection. The attendant early onset of severe atrophic gastritis again suggested the association of MAIT cells and pathology.

To verify the pathogenic role of MAIT cells observed in transgenic mice, we built on our established lung infection model (1), with two methods (Salmonella infection and synthetic 5-OP-RU Ag/TLR agonist) to boost MAIT cell numbers prior to infection with H. pylori, on the basis that prior expansion of these cells is integral to their impact, and may reflect the case in humans with a history of infections. In this model, MAIT cells are distributed to many different tissues and organs after resolution of a primary lung infection (1). We hypothesize that a previous history of infection of this kind can prime MAIT cells and increase their numbers. Indeed, we observed increased MAIT cells in the stomach of wt mice after a primary lung infection, possibly because of mucosal homing markers expressed on MAIT cells and a shared cytokine/chemokine milieu between the mucosae of the lungs and stomach (43, 44).

We show that when MAIT cells from wt C57BL/6 mice are primed first with either S. Typhimurium or with specific 5-OP-RU Ag and a TLR agonist, they subsequently respond rapidly upon challenge with H. pylori (∼10-fold increase in MAIT cell numbers at 2 wk postinfection) and these mice develop severe inflammation, recapitulating what was observed with the MAIT TCR transgenic mice. This was also the case in mice primed with TLR agonist plus synthetic 5-OP-RU Ag rather than Salmonella infection, but not TLR agonist alone. There was no change in MAIT cell proportions in the mesenteric lymph nodes following H. pylori infection of S. Typhimurium–primed mice (data not shown), indicating a gastric tissue–specific effect. These mice developed atrophic gastritis as early as 8 wk postinfection, which is normally not observed in wt C57BL/6 mice infected with H. pylori. The response was specific to MAIT cells and dependent on MR1. The accumulation of MAIT cells in the stomach showed a biased enrichment of CD8⁺ MAIT cells that was not seen in the blood. This could be due to upregulation of CD8, as the cytokine milieu has been shown to influence CD8 expression on CD4⁺ T intraepithelial lymphocytes (45), or it might reflect preferential expansion of CD8⁺ MAIT cells.

Consistent with MAIT cells in other tissues and previous studies (1, 19, 24), we observed that MAIT cells in the stomach have an effector memory phenotype, characterized by CD44^hiCD62L^lo expression. A large proportion (∼50–70%) expressed the tissue-resident memory markers CD69 and CD103, consistent with observations on cells isolated from H. pylori–infected stomachs (19), and reflecting the chronic nature of H. pylori infection. MAIT cells isolated from infected stomachs also showed high expression of T-bet, consistent with a Th1-biased immune response to H. pylori, and similar to that observed after acute bacterial lung infection (1). Interestingly, T-bet–deficient mice exhibit a higher bacterial load, but lower levels of gastritis after H. pylori infection, indicating that a Th1-biased response may be important for both protection and immunopathology (46). PLZF and RORγt were both expressed on ∼50% of stomach MAIT cells following H. pylori infection of primed mice. Comparison with primed uninfected mice, and mice infected with H. pylori alone revealed that this transcriptional profile was driven primarily by H. pylori infection, with minimum effect of the prior S. Typhimurium priming. Thus, H. pylori infection appears to drive a Tc1/Tc17 transcriptional profile in responding MAIT cells. Although MAIT cells in the stomachs of naive mice were too few to be included in this analysis, the lower expression of these transcription factors following infection differs from that seen previously in either naive or S. Typhimurium–infected lungs, where the majority of MAIT cells expressed PLZF and RORγt (1). This difference may result from reprogramming of MAIT cell cytokine profiles, the preferential expansion of MAIT cell subsets under different tissue-specific cytokine milieu during chronic stimulation, or other signals that differ between the two pathogens. Interestingly, NK T cells can downregulate PLZF in adipose tissue because of chronic activation (47). To our knowledge, this is the first description of such a marked shift in the expression pattern of these transcription factors by MAIT cells, and the first demonstration of enrichment of Tc1/Tc17 MAIT cells.

A number of studies using a range of mouse models have shown a Th1/Th17type T cell response in H. pylori (22, 23). The cytokines TNF, IFN-γ, and IL-17 are well established to have proinflammatory activity during infection (48–50). MAIT, and to a lesser extent non-MAIT αβ T cells in our study, showed mixed RORγt⁺ and RORγt⁻ populations and high T-bet expression. Although technical difficulties precluded detection of cytokine production by stomach MAIT cells, the high T-bet expression suggests MAIT cells likely express IFN-γ in H. pylori–infected stomachs, consistent with a role for this cytokine in driving pathology. Importantly, we showed that the presence of MAIT cells was accompanied by infiltration of innate immune cells, as well as increased numbers of non-MAIT αβ T cells in the earlier stages of infection. This was associated with the severe gastritis phenotype observed in wt C57BL/6 mice with preboosted MAIT cells, and MAIT TCR transgenic mice, infected with H. pylori. Previous studies have shown that T cells are major orchestrators of inflammation (22, 23, 51). Interestingly, IFN-γ– and IL-17A–deficient mice exhibited reduced gastritis in previous H. pylori infection studies, consistent with a role for MAIT or other cell types in orchestrating pathology (52). Given the low numbers of MAIT cells in stomachs, they likely mediate their pathogenic effect through other cell types rather than directly. The impact of MAIT cells on the function of other cells, and, in particular, their production of key mediators of inflammation in H. pylori infection, such as IL-8 (53), for example, requires further investigation. Their impact on IgA production would also be worthy of further investigation.

In summary, we show for the first time, to our knowledge, that in H. pylori–induced gastritis MAIT cells can play a pathogenic role in orchestrating immunopathology. This study clearly demonstrates, in vivo, that MAIT cells have the potential to drive inflammation and pathology during chronic bacterial infection, and this is likely associated with MAIT cells’ capacity to produce inflammatory cytokines, including IFN-γ, IL-17, and TNF. Notably, in humans only ∼20% of infected individuals develop overt disease, and the other factors at play are only partially understood. It is conceivable that the penetrance of severe gastritis in H. pylori–infected humans is influenced by a previous history of infections, and resultant MAIT cell priming, possibly at other mucosal sites. The recognition of a common MR1-restricted Ag, derived from bacterial riboflavin synthesis, suggests that MAIT cells may be an important factor in modulating inflammation in Helicobacter and other chronic bacterial infections. Our findings also indicate that previous infections or vaccination, such as with live attenuated S. Typhi (54), may activate MAIT cells and predispose an individual to subsequent immunopathology when other comorbidities are present.

We thank Dr. Wei-Jen Chua and Dr. Ted Hansen for kind provision of the 26.5 mAb, David Jackson for provision of Pam2Cys, and Paul E. O’Brien and Paul Burton (Centre for Obesity Research) for provision of human gastric tissue. We thank Prof. Paul Klenerman and Dr. Garrett Ng for critical review of the manuscript.

Z.C., S.B.G.E., D.P.F., L. Liu, J.Y.W.M., J.R., J.M., and A.J.C. are inventors on patents describing MR1-tetramers. The other authors have no financial conflicts of interest.