Priming programed Mucosal-Associated Invariant T cells protect against systemic or local bacterial infection Authors

Mucosal-Associated Invariant T (MAIT) cells have potent antibacterial functions. Their protective capacity, in vivo , has been demonstrated in mouse models, particularly of respiratory infections. We now show that during systemic infection of mice with Francisella tularensis Live Vaccine Strain (LVS), MAIT cell expansion was evident in the liver, lungs, kidney, spleen and blood. MAIT cells manifested a polarised T h 1-like (termed “MAIT-1”) phenotype and cytokine profile that conferred a critical role in controlling bacterial load. After resolution of the primary infection, the expanded MAIT cells developed to a stable memory-like MAIT-1 cell population, suggesting a basis for vaccination and protection against subsequent challenge. Indeed, a systemic vaccination with synthetic ligand (5-OP-RU) in combination with CpG adjuvant boosted MAIT-1 cells and resulted in enhanced protection against systemic and local infections with F. tularensis and Legionella longbeachae . Our study highlights the potential utility of targeting MAIT cells to combat multiple bacterial pathogens. cells due to their expression, upon thymic egress, of a “memory-like” phenotype (CD62L low , CD44 high ) 43 . We show here, that upon infection their initial response (accumulation and cytokine secretion) elicited a critical bridging protection role between innate and adaptive immunity. After the resolution of infection, a memory-like MAIT cell pool was formed and maintained at significantly higher frequency and with an infection-imprinted functional phenotype (MAIT-1) even after 100 days. A single vaccination reconstituted a similar memory-like MAIT cell pool post an infection (i.e. boosted number of MAIT-1 cells). The vaccinated mice controlled bacterial burdens faster and more efficiently than naïve, non-boosted mice. Thus, MAIT-1 cells show both an “innate-like” rapid response and adaptive memory-like features, consistent with our previous observations on MAIT-17/1 cells from respiratory infection models with S.


INTRODUCTION
Microbial infections in patients with compromised immunity (e.g. organ transplant recipients), particularly with multi-drug resistant pathogens, cost several billion dollars each year 1, 2, 3 . Bacterial sepsis is associated with a very high mortality of 30% to 50% 4,5 and is estimated to cause 6 million deaths among 30 million people affected worldwide every year 6 , making it a leading cause of death in hospitals. Thus, there is great interest in the development of new preventive and therapeutic approaches, including T cell-based vaccination 7 . Although some progress has been made in this area, complete protection by T cell vaccines has not yet been achieved and there remains a great need to understand the generation of effective T cell memory and responses 7 .
Mucosal-Associated Invariant T (MAIT) cells are an abundant ab T cell subset (averaging 3% of T cells in human blood 8 ) that express semi-invariant T cell antigen receptors (TCRs), comprising TRAV1-2-TRAJ33/12/20 in humans 9, 10, 11 or TRAV1-TRAJ33 (Vα19-Jα33) in mice, paired with a limited TCR b chain repertoire. MAIT cells recognise conserved vitamin B2 metabolite-based antigens (Ag) from microbes, presented by the Major histocompatibility complex (MHC) Class I related protein 1 (MR1) molecule. Riboflavin biosynthesis, the source of MAIT cell Ag, is an essential and highly conserved metabolic pathway present in a diverse range of bacteria and yeasts.
We and others have shown in mouse models, that MAIT cells afford a protective function against several bacterial pathogens 12,13,14,15,16,17 , including respiratory infections with Legionella longbeachae 12 and Francisella tularensis 13 . Human in vitro studies demonstrated reactivity of MAIT cells to an even broader range of microbial species, reviewed in 14 , including Streptococcus spp. 15,16 , Shigella flexneri 17 and fungal pathogens Aspergillus fumigatus 18 and Candida albicans 19 . Alterations in MAIT cell frequency and function have been reported in patients with tuberculosis 20,21,22 and other infectious diseases 23,24,25,26,27 . Sandburg and colleagues reported a patient with cystic fibrosis affected by recurrent, and eventually fatal, bacterial infections which correlated with a striking lack of detectable circulating MAIT cells 24 .
Despite the presence of MAIT cells at many sites 28 , including high numbers in the liver 29 , female reproductive tract 30 and intestine 31 , previous studies have mostly examined their protective role in local infections 13,32 . Thus far, a protective role has been demonstrated most convincingly in respiratory infections 12,13 , with little definitive in vivo evidence to date of their capacity to contribute to protective immunity against systemic infection. Francisella tularensis causes both respiratory disease and tularemia, a rare but often fatal systemic infection. Infection can occur via the respiratory route, by ingestion of contaminated water or via tick bites 33 . F. tularensis LVS was developed as a vaccine strain with reduced pathogenicity 34 . Although its use in humans was discontinued due to the risks associated with administration of a live organism with undetermined genetics, F. tularensis LVS provides a useful model for pathogenicity and immune response research 35 , and was previously used to demonstrate a protective role for MAIT cells during respiratory infection 13,36 .
Here, we examined the role of MAIT-1 cells in a systemic infection model of F. tularensis LVS infection, where the kinetics of the MAIT-1 cell response was investigated in multiple organs. MAIT-1 cells were highly responsive and were not only critical to controlling bacterial load, but also formed a "memory-like" population after infection. Vaccination with synthetic 5-OP-RU antigen in combination with CpG adjuvant recapitulated this expanded memory MAIT-1 cell population and afforded better control of subsequent systemic and local infections by both F. tularensis and L. longbeachae.

F. tularensis LVS activates MAIT reporter cells in vitro in an MR1-dependent manner and induces a systemic MAIT cell response in vivo.
To confirm that F. tularensis LVS, riboflavin biosynthesis proficient bacteria as shown in the Kyoto Encyclopedia of Genes and Genomes database, can stimulate MAIT cells, we used an in vitro activation assay with Jurkat.MAIT reporter cells expressing a TRAV1-2/TRBV6-1 TCR (Jurkat.MAIT-AF7) 11 co-cultured with C1R.hMR1 cells as APCs, as previously described 11 . F. tularensis bacterial lysate was able to stimulate Jurkat.MAIT-AF7 cells in this assay as assessed by up-regulation of cell surface CD69 expression (Fig. 1, A). Anti-MR1 monoclonal antibody (mAb) (26.5), but not an isotype control mAb, completely blocked this response, consistent with MR1dependent activation. Like previous studies using other bacteria 22,37 and consistent with MAIT cell expansion in the lungs in respiratory F. tularensis infection 13 these data demonstrated that F. tularensis LVS is capable of producing Ags that stimulate MAIT cells in an MR1-dependent manner.
To examine the MAIT cell response in vivo, C57BL/6 mice were infected with F. tularensis LVS intravenously (i.v) with a sublethal dose of 10 4 colony forming units (CFU). MAIT cell frequencies and numbers were assessed in a range of organs (liver, lungs, spleen, kidneys) and in the blood by MR1:5-OP-RU tetramer staining (Supplementary Fig. 1). Significant MAIT cell expansion was observed in all organs tested and in the blood at 14 days post infection (dpi), compared to uninfected mice (Fig. 1, B-D). Non-MAIT ab T cells were also increased in response to infection, as expected ( Fig. 1, E), but their increase was less pronounced as compared to that seen for MAIT cells. Since F. tularensis infection can occur naturally through inhalation, wounds or tick bites, we next tested various infection routes: intranasal, intratracheal, intraperitoneal (i.p.) and subcutaneous (s.c.), using a range of doses for each. Regardless of the infection route, systemic dissemination occurred resulting in significant MAIT cell accumulation in all tested organs in mice infected with sufficient dose (Supplementary Fig. 2). Due to the ease of consistent and accurate delivery of inoculum, and rapid dissemination of bacteria, we chose the i.v. route for all following infection experiments.

The MAIT cell expansion following F. tularensis infection is rapid and long-lasting in all organs
Upon F. tularensis LVS infection, a typical valley shaped-weight change was recorded in infected mice ( Fig. 2, H) with mice losing weight until day 5, then recovering and reaching their starting weight by day 12. The bacteria clearance kinetics showed an initial decline in the bacterial load in the blood, but expansion at early stages in all other sites, which reached a plateau around 3 dpi, followed by clearance (undetectable level) within 12 days (Fig. 2, A).
We next investigated the kinetics of MAIT cell expansion in various organs. Both the frequency (relative to total ab-T cells, Fig. 2, B) and number (Fig. 2 of ab-T cells in the liver, 27% in the lungs, 26% in the kidneys, 19% in the blood and 16% in the spleen. The frequency of MAIT cells reached a plateau between 9-14 dpi in the liver, lungs and kidney, in contrast to the total number, which plateaued at about 6 dpi in all organs. Interestingly, high frequencies of MAIT cells (43% in the liver, 8% in the lungs, 21% in the kidneys 4% in the blood and 2% in the spleen) were maintained at 100 dpi, long after clearance of bacteria to undetectable numbers. This was in contrast to non-MAIT ab-T cells, which contracted as expected to their homeostatic level (in numbers) in all organs (Fig. 2, C-G).

F. tularensis infection polarizes MAIT cells to a MAIT-1 (Th1 like) functional phenotype.
We next examined the functional phenotype of the MAIT cell populations that were expanded during F. tularensis infection by assessing their cytokine profile during the acute infection phase (6 dpi) and the expression of transcription factors in MAIT cells from naïve mice and at representative time points of acute infection (6 dpi), recent bacterial clearance (14 dpi) and memory phase (100 dpi). Four cytokines, TNF, IFNγ, GM-CSF, and IL-17, previously shown to be produced by mouse and human MAIT cells 12,37,38,39 , were examined directly ex vivo. The majority of MAIT cells were found to be polarized to a Th1 like phenotype, with high proportions, and numbers, of cells expressing TNF, IFNg and GM-CSF, but few detectable IL-17-producing cells during the acute phase of infection (6 dpi) ( Fig. 3 and Supplementary Fig. 3, 4). Interestingly, MAIT cells appeared to produce these Th1 cytokines in an organ specific pattern, with kinetic differences in the proportions of cells producing each cytokine. This was particularly apparent late in infection, with consolidation and persistence of population of MAIT-1 cells in the liver and spleen, but a reduction of MAIT-1 proportion in the lung and kidney (Fig. 3, B-C). This was most evident for IFNg-producing MAIT cells (Fig. 3, B), with TNF-producing MAIT cells displaying a similar but less pronounced pattern (Fig. 3, C), and the abundance of GM-CSF-producing MAIT cells increasing throughout the infection in all organs except for the kidney (Fig. 3, C). In contrast, the proportion of IL-17-producing MAIT-17 cells declined in both liver and lungs (Fig. 3, C), but was later increased in the lungs. The frequencies of cytokine producing non-MAIT ab-T cells were generally lower compared to MAIT cells and there was no clear pattern of changes over time, although IFNg production was the most abundant, particularly in the early stages of infection ( Fig. 3A-C).
The production of Th1 and Th17 cytokines by MAIT cells were consistent with their expression of the closely correlated transcription factors T-bet and RORγt, respectively. We found that naïve pulmonary MAIT cells were predominantly (>80%) T-bet -RORγt + (MAIT-17) (Fig. 4), consistent with our previous study 12 , whereas in the liver >60% had a MAIT-1 phenotype (RORγt -T-bet + ) and there was a mixture of MAIT-1 and MAIT-17 cells in the kidney, spleen and blood (Fig. 4). Following infection (6 dpi) with F. tularensis, almost all MAIT cells were RORγt -T-bet + in all organs and in the blood (Fig. 4). Remarkably, at 100 dpi, long after bacteria were cleared to undetectable levels ( Fig.   2, B), the phenotype of MAIT cells remained skewed towards MAIT-1. However, there was some shift in phenotype by this time towards the naïve phenotype in each organ consistent with the results seen in with intracellular cytokine staining (Fig. 3, B-C). This was most evident in the lungs, and suggested an organ-specific milieu may exist, which drives differently skewed responses (Fig. 4).
Overall, our data suggest that tissue-specific signals foster different polarization of MAIT cells at different sites. During infection, these are overruled by pathogen-derived signals, and long term after infecting the resultant MAIT cells are shaped by the balance of these factors.

MAIT-1 cells are critical for controlling bacteria during F. tularensis systemic infection.
Next we examined the protective role of MAIT-1 cells in F. tularensis systemic infection by comparing WT (C57BL/6) mice with Mr1 -/mice, which lack MAIT cells. Following inoculation of mice i.v. with 10 4 CFU F. tularensis LVS, significantly higher bacterial loads were observed at 6 dpi in all organs (liver, lung, spleen kidney and blood) of Mr1 -/mice than in WT mice, and the difference was also significant at 5 dpi and 7 dpi in some organs (Fig. 5 Fig. 5, B). Here, adoptive transfer of MAIT-1 cells was conducted to WT mice, followed by bacterial challenge with 10 CFU F. tularensis LVS (Supplementary Fig. 5, A-B).
Transferred MAIT cells were sorted from the livers of mice infected with F. tularensis to boost MAIT cell numbers. When adoptively transferred, these cells provided dose-dependent protection with full protection achieved when 5 x 10 5 MAIT cells were transferred (Supplementary Fig. 5, B).
To further test the capacity for MAIT-1 cells to provide protection, we then moved to an immune compromised setting and utilized Rag2 -/-gC -/mice, which lack T, B, and NK cells 41 . These mice are severely immune compromised and as expected all mice succumbed to a low dose of F. tularensis LVS (20 CFU) i.v. within 20 dpi (Fig. 6C). For these experiments, we sourced MAIT-1 cells for adoptive transfer from the livers of WT mice primed with synthetic 5-OP-RU Ag and CpG adjuvant ( Fig. 6, A) to ensure no carry over of live bacteria to the recipient mice before challenge. This priming method recapitulated the predominant MAIT-1 phenotype in the liver as seen with F. tularensis infection (Fig. 6, A). Mice receiving adoptively transferred MAIT cells were able to significantly prolong survival after subsequent i.v. challenge with F. tularensis LVS, compared to mice without MAIT cell transfer (Fig. 6, A-C), suggesting that MAIT-1 cells can indeed provide protection in their own right, in the absence of several other immune elements. To further explore the role of individual cytokines produced by MAIT cells in protecting against systemic F. tularensis LVS infection, adoptive transfer of WT vs cytokine-deficient MAIT cells to Rag2 -/-gC -/mice was performed. WT or IL-17-deficient MAIT cells were able to significantly prolong survival of mice following infection, with some mice surviving up to 58 dpi (Fig. 6, C). In contrast, MAIT cells deficient in production of TNF, INFg or GM-CSF were unable to provide protection, with mice succumbing to infection similarly to the untreated Rag2 -/-gC -/mice. This was consistent with the susceptibility of mice genetically deficient in individual cytokines (Supplementary Table 1). Together, these data show that MAIT-1 cells have the capacity to contribute to systemic protection against bacterial infection and this can be the difference between life and death in the absence of other arms of the immune system.

Systemically boosted MAIT-1 cells manifest vaccination potential against local and systemic pathogens.
We previously showed that MAIT cells can be primed, providing more rapid control of Legionella longbeachae infection in the lungs 12,39 . In the Legionella infection model the majority of MAIT cells were MAIT-17 39 . To explore the potential of systemically boosting MAIT-1 cells as a component of vaccination, here an immunization scheme was developed to boost MAIT cells systemically prior to bacterial challenge. Modifying our previously reported methods 12, 38, 39 , we used synthetic MAIT antigen 5-OP-RU administered intravenously together with CpG adjuvant (Fig. 7, A). This resulted in a significantly expanded MAIT-1 cell pool systemically (in liver, lungs and spleen) without significantly changing the total numbers of non-MAIT ab T cells (Fig. 7, A, and Supplementary   Fig. 6). Furthermore, the majority of MAIT cells boosted with this vaccination scheme had a MAIT-1 phenotype (Fig. 7, B), similar to the MAIT cell phenotype driven by systemic F. tularensis LVS infection ( Fig. 3 and 4).
To test the protective capacity afforded by MAIT-1 cell vaccination, mice were then challenged with F. tularensis LVS i.v. as depicted in Fig. 7, C. Vaccinated mice were rested for a month to foster MAIT memory formation and to allow the immune system to return to homeostasis. Both the bacterial load and mouse survival were assessed independently. Vaccinated WT C57BL/6 mice showed a significant reduction in the bacterial load in the liver and lungs at 3 and 5 dpi compared to unvaccinated mice, CpG only treated mice or vaccinated Mr1 -/mice (Fig. 7, D, E). Thus, our data show that a systemic vaccination scheme that targets MAIT cells can result in earlier clearance of bacteria. Consistent with the stronger MAIT cell response, vaccinated mice also showed enhanced survival of otherwise lethal F. tularensis infection (Fig. 7, F). We next assessed whether systemic MAIT cells boosting could provide enhanced protection when challenged with a different bacterial pathogen. For this we used intranasal infection with L. longbeachae. The bacterial load in the lungs of mice at 5 and 7 post-challenge (intranasally), were significantly lower in vaccinated C57BL/6 mice than in the three control groups (Fig. 7, G) indicating that vaccination to boost MAIT cells can provide a level of protection against unrelated bacterial pathogens. (T-bet + /RORgt + double positive) 12,37 . The moderate skewing of liver MAIT cell population to MAIT-1 in naïve mice was further consolidated to >95% T-bet + after F. tularensis infection. In a previous study, elevated IL-17A gene transcription was detected in total lung tissue of WT compared to Mr1 -/mice following pulmonary F. tularensis LVS infection 13 . However, here, using intracellular cytokine staining, we did not observe an increase in IL-17 production in the lungs by MAIT cells.

DISCUSSION
Thus, in addition to tissue-specific cues, polarization signals from pathogens drive different MAIT cell responses, particularly during acute infection. It is unclear from the current study, whether polarised MAIT cell populations resulted from a preferential expansion, recruitment, or MAIT cell plasticity.
We previously showed that IFN-g production from MAIT cells was necessary for their protective Compounds, immunogens and tetramers. 5-OP-RU was prepared as described previously 48 . Ac-6-FP was purchased from Shircks laboratories. CpG combo (CpG B and CpG P together), sequence: T*C*G*T*C*G*T*T*T*T*G*T*C*G*T*T*T*T*G*T*C*G*T*TT*CG*T* CG*A*CG*A*T*CG*G*C*G*CG*C*G*C*C*G (*phosphorothioate linkage) non-methylated cytosine-guanosine oligonucleotides was purchased from Integrated DNA Technologies, Singapore).
Bacterial strains. Cultures of F. tularensis LVS were grown in 10 ml of brain heart infusion (BHI) broth for 16-18 h at 37 °C with shaking at 180 rpm, or on Cysteine Heart Agar (CYHA) plates containing 10 μg/ml ampicillin, 7.5 μg/ml colistin and 4 μg/ml trimethoprim for 4 days. For the infecting inoculum, with the estimation that 1 OD600=2.4x10 9 /ml, bacteria from overnight culture were washed and diluted in phosphate buffered saline (PBS) for instillation to mice. A sample of inoculum was plated onto cysteine heart agar plates with appropriate antibiotics (as above mentioned) for verification of bacterial concentration by counting colony-forming units (CFU). Legionella longbeachae NSW150 was grown in buffered yeast extract (BYE) broth or agar plates supplemented with streptomycin (50 µg/ml).   Data are presented as mean ± SEM (n=20 mice).