First, we asked if co-culture with activated MAIT cells could impact MR1 expression and function in airway epithelial cells (AEC). To address this, we examined MR1 mRNA expression of primary human AEC co-cultured with the human MAIT cell clone D426G11 alone or in the context of Streptococcus pneumoniae (Sp) infection.

We noticed significantly increased MR1 mRNA expression in AEC from healthy donors when infected with Sp and cultured with MAIT cells ( Figure 1A , Supplementary Figure 1A , Supplementary Table 1 ). Infection with Sp or co-culture with MAIT cells alone did not stimulate a significant response. We replicated this system using a model bronchial epithelial cell line (BEAS-2B cells) infected with Mycobacterium smegmatis (Ms). Similarly, we observed increased MR1 expression in the Ms-infected BEAS-2B cells co-cultured with MAIT cells, with no impact of either condition alone ( Figure 1B , Supplementary Figure 1B , Supplementary Table 1 ). Using flow cytometry to quantify surface MR1 protein expression, we likewise found increased MR1 expression with both Ms infection and MAIT cell co-culture compared to either condition alone ( Figure 1C , Supplementary Figure 1C ). These data suggest that MAIT cells, when activated by presentation of bacterial antigens, could lead to increased MR1 mRNA or surface protein expression in the infected cell.

We next asked whether microbial infection is required for this transcriptional increase or if the presence of MR1 ligand alone is sufficient. We treated BEAS-2B cells with either the stimulatory antigen 5-(2-oxopropylideneamino)-6-d-ribitylaminouracil (5-OP-RU) or the non-stimulatory ligand 6-formylpterin (6-FP) and measured expression of MR1. Neither 5-OP-RU nor 6-FP increased MR1 expression alone ( Figures 1D, E ). In co-culture with MAIT cells, however, MR1 mRNA expression increased when BEAS-2B cells were treated with 5-OP-RU, while treatment with 6-FP had no impact on MR1 expression ( Figures 1D, E , Supplementary Figures 1D, E ). These data demonstrate that the upregulation of MR1 mRNA expression requires activation of MAIT cells, and this may be stimulated by antigen presentation alone or bacterial infection.

Among the effector molecules produced by activated MAIT cells, IFNγ is well known to stimulate transcription of MHC Class I molecules (15–17). We hypothesized MR1 expression could be regulated through similar mechanisms, despite the differences in chromosomal location and gene arrangement from classical HLA genes. To test if IFNγ alone is sufficient to stimulate MR1 expression, we treated primary AEC with recombinant human IFNγ. IFNγ treatment significantly increased MR1 transcription ( Figure 2A , left). As expected, IFNγ also increased expression of positive control HLA-A ( Figure 2A , right). In BEAS-2B cells treated with IFNγ, we also observed significant increases in both MR1 and HLA-A mRNA expression ( Figure 2B ).

To quantify MR1 protein expression, we measured surface MR1 expression by flow cytometry. We found that IFNγ treatment also significantly increased surface MR1 expression and control MHC-I expression in primary AEC and BEAS-2B cells ( Figures 2C, D , Supplementary Figure 1F ). This approach does not distinguish between 1) increased surface expression of MR1 proteins due to increased MR1 transcription and translation or 2) increased translocation of existing MR1 molecules and stabilization on the cell surface. To determine if IFNγ signaling impacts post-transcriptional protein stability of MR1, we utilized BEAS-2B cells expressing MR1-GFP under a doxycycline-inducible promoter (11). IFNγ treatment did not increase expression of MR1 mRNA, total MR1-GFP gMFI, or surface MR1 in these doxycycline-treated cells ( Figure 3A-C ). As expected, IFNγ treatment increased MHC-Ia surface expression and 6-FP treatment induced significant stabilization of total MR1-GFP protein expression and surface translocation ( Figures 3C, D ). Together, these data indicate that increased surface MR1 in IFNγ-treated wildtype cells resulted from stimulation of MR1 transcription rather than protein-level impacts.

MR1 antigen presentation and MAIT cell responses are increased in MR1 over-expression systems (10, 11, 25). We next investigated if the IFNγ-dependent increase in MR1 expression similarly enhanced MAIT cell responses to wildtype cells. BEAS-2B cells were treated with IFNγ for 12 hours, thoroughly washed to remove excess soluble IFNγ, then used as antigen-presenting cells in an IFNγ ELISPOT assay to quantify MAIT cell activation. Filtered M. smegmatis supernatant was used as the antigen source to avoid potential confounding impacts of IFNγ treatment on bacterial infection. MAIT cell responses to IFNγ pre-treated BEAS-2B cells were significantly increased compared to UT controls ( Figure 3E ). Therefore, IFNγ treatment is sufficient to stimulate MR1 transcription, leading to increased protein expression and antigen presentation to MAIT cells.

Using MHC-Ia transcription pathways as a starting point, we queried the JASPAR CORE 2018 Vertebrates database to determine if the MR1 promoter contained binding motifs for known IFNγ-induced transcription factors (26–28). We highlighted notable predicted elements on this region, including those common with IFNγ-mediated HLA transcription factor sites ( Figure 3F ). For example, we found putative binding motifs for IRF1 and members of the NLRC5 enhanceosome complex (18, 19, 29–31). This targeted search suggested that IRF1 and NLRC5 could be of interest in IFNγ-mediated MR1 transcription.

We first validated that IFNγ signaling induces IRF1 and NLRC5 mRNA expression in our cells. Transcripts of both these genes were significantly increased by IFNγ treatment of both primary human AEC and BEAS-2B cells ( Figures 3G, H ).

To determine if IRF1 or NLRC5 are required for the IFNγ-mediated increase in MR1 transcription, we first used siRNA to knock down the genes separately or together in BEAS-2B cells. IRF1 KD alone significantly decreased IFNγ-stimulated MR1 mRNA expression compared to missense controls ( Figure 4A , Supplementary Table 2 ). Although IRF1 siRNA significantly reduced IRF1 expression, we were unable to sufficiently knock down NLRC5 expression by siRNA ( Supplementary Figures 2A, B ). Therefore, we generated monoclonal NLRC5^-/- BEAS-2B cell lines by CRISPR/Cas9. Loss of NLRC5 did not lead to any significant impact to IFNγ-induced MR1 mRNA expression ( Figure 4B ). We did observe a decrease in HLA-A expression with NLRC5 knockout, although not significant ( Figure 4C ).

We used siRNA to silence IRF1 expression in the Cas9⁺ and NLRC5^-/- cells ( Supplementary Figure 2C ). MR1 expression was significantly impacted by IRF1 knockdown in Cas9⁺ control cells ( Figure 4B ), agreeing with the previous siRNA results in wildtype BEAS-2b cells. A trend in reduced MR1 expression was also observed in two clones of NLRC5^-/- cells treated with IRF1 siRNA ( Figure 4B , Supplementary Figure 2D ). Expression of surface MR1 proteins in IFNγ-treated cells was similarly decreased with IRF1 silencing and unaffected by NLRC5 knockout ( Supplementary Figure 2E ). Finally, we used ELISPOT assays to quantify if loss of NLRC5 and/or IRF1 would impact the IFNγ-stimulated boost in MR1 antigen presentation to MAIT cells. IRF1 siRNA knockdown significantly reduced MAIT cell responses to both Cas9⁺ cells and NLRC5^-/- cells ( Figure 4D ). The missense-treated NLRC5^-/- cells stimulated similar MAIT cell activity as the Cas9⁺ control cells in response to IFNγ ( Figure 4D , Supplementary Table 2 ). Together, these results indicate that IRF1 expression is required for IFNγ stimulation of MR1 expression and antigen presentation function, while NLRC5 does not appear to impact this pathway.

We next used CRISPR/Cas9 knockout to generate monoclonal IRF1^-/- BEAS-2B cell lines to confirm this finding. IFNγ-mediated stimulation of MR1 mRNA and MR1 surface expression was impaired in IRF1^-/- cells compared to Cas9⁺ cells ( Figures 4E,G , Supplementary Figures 2F, G ). Expression of HLA-A mRNA or MHC-Ia surface proteins were likewise impaired in the IRF1^-/- cells following IFNγ treatment ( Figures 4F,H ; Supplementary Figure 2G , right). Together the siRNA knockdown and CRISPR knockout results indicate that IRF1 mediates the IFNγ signaling pathway leading to MR1 transcription, surface expression, and antigen presentation, likely through mechanisms independent of the NLRC5 enhanceosome.

To validate this mechanism in a physiologically relevant system, we returned to our co-culture experiments. We first demonstrated that co-culture of infected AEC with MAIT cells led to upregulation of IFNγ-stimulated pathways by staining phosphorylated STAT1 (pSTAT1). Ligation of the IFNGR activates Janus kinase (JAK) dimer 1/2, which in turn phosphorylates STAT1 (17, 32). As expected, pSTAT1 staining was significantly increased in AEC infected with S. pneumoniae and co-cultured with MAIT cells, along with AEC treated with recombinant IFNγ ( Figures 5A, B , Supplementary Figure 3A , Supplementary Table 3 ).

To confirm that activation of MAIT cells in co-culture is sufficient to drive IFNγ signaling, we next assessed expression of IFNγ-stimulated genes. Co-culture of Sp-infected primary AEC with MAIT cells significantly induced expression of HLA-A, B2m, IRF1, and NLRC5 ( Figures 5C–F ; Supplementary Figures 3B–D, G ; Supplementary Table 3 ). We observed increased expression of these genes in AEC incubated with either Sp or MAIT cells alone, which may indicate the contribution of other inflammatory signaling pathways in AEC. However, the combined co-culture induced significantly greater expression for almost all genes, pointing to the role of Ag-induced MAIT cell activation in driving inflammatory gene expression.

BEAS-2B infected with Ms significantly induced expression of HLA-A and IRF1 only in combination with MAIT cell co-culture ( Figures 6A, C ; Supplementary Figures 3E, H ; Supplementary Table 3 ). We next used 5-OP-RU as the antigen source to confirm that MR1 antigen presentation alone is sufficient to stimulate MAIT cell IFNγ production and subsequent inflammatory gene expression, absent other microbial stimuli. As expected, HLA-A and IRF1 expression were significantly increased in 5-OP-RU-treated BEAS-2B cells when MAIT cells were present, but not stimulated by treatment with 5−OP−RU alone or MAIT cell co-culture with untreated BEAS-2B cells ( Figures 6B, D ; Supplementary Figures 3F, I ; Supplementary Table 3 ). Non-stimulatory presentation of 6-FP ligands failed to significantly induce either gene alone or in combination with MAIT cells. Together, these results indicate that activated MAIT cells produce sufficient IFNγ to stimulate expression of downstream genes.

Finally, we used our IRF1^-/- cells in this co-culture setting to demonstrate the role of IFNγ in mediating MR1 expression, antigen presentation, and MAIT cell activation. IRF1^-/- cells infected with Ms and co-cultured with MAIT cells failed to exhibit the increase in MR1 expression seen in the Cas9⁺ control cells ( Figure 6E , Supplementary Table 4 ). We then used exogenous 5-OP-RU treatment to directly test the role of IRF1 following MR1-dependent MAIT cell activation. MR1 transcription was enhanced in Cas9⁺ cells with 5-OP-RU and MAIT cell co-culture, confirming that TCR-stimulated MAIT cells directly led to increased MR1 transcription ( Figure 6F , Supplementary Table 4 ). In contrast, the IRF1^-/- cells treated with 5-OP-RU did not express greater MR1 transcripts in co-culture, confirming the importance of IRF1 in this pathway.

Our data have shown thus far that IFNγ stimulates MR1 transcription; however other inflammatory cytokines can also induce transcription. Type I interferons like IFNβ stimulate IRF1 and NLRC5 to induce MHC-Ia transcription (15–19, 33, 34). Activated MAIT cells may also produce TNFα and IL-17, which have been demonstrated to stimulate transcription of inflammatory genes including IRF1 and HLA-A (35–37). To assess whether the IFNγ-induced increase in MR1 transcription is representative of general inflammatory signaling mechanisms or specific to IFNγ stimulus, we treated BEAS-2B cells with recombinant human inflammatory cytokines IFNβ, IFNγ, IFNλ, TNFα, and IL-17. Of these cytokines, only IFNβ and IFNγ elicited a significant increase in MR1 transcription compared to untreated controls ( Figures 7A, B , Supplementary Table 5 ).

Interestingly, the IFNγ-mediated increase in MR1 expression was significantly greater than the increase due to IFNβ treatment, while expression of HLA-A and B2m were similarly induced by both IFNγ and IFNβ treatment ( Figures 7A–C , Supplementary Figure 4 , Supplementary Table 5 ). We further investigated the role of type I and II IFNs in mediating expression of MR1 and MHC-Ia. While both IFNβ and IFNγ increased surface expression of MHC-Ia, only IFNγ led to a significant increase in surface MR1 protein expression ( Figure 7D ). BEAS-2B cells pre-treated with IFNγ induced significantly greater MAIT cell responses to Ms infection than control UT cells, while IFNβ pre-treatment did not generate a significantly different dose-response curve ( Figure 7E ). These data indicate that IFNγ plays the largest role in stimulating MR1 expression and function.

We quantified IRF1 and NLRC5 transcripts to further explore how MR1 and MHC-I expression are differentially stimulated by IFNβ and IFNγ. Although both IFNγ and IFNβ increased IRF1 expression, the relative fold change was significantly greater with IFNγ than IFNβ ( Figure 7F , Supplementary Figure 4 , Supplementary Table 5 ). Both interferons induced significant increases in NLRC5 and B2m expression ( Figures 7G, H , Supplementary Figure 4 , Supplementary Table 5 ). In light of our finding that IFNγ stimulates MR1 transcription through IRF1 and not NLRC5, this magnitude of IRF1 induction may relate to the specific induction of MR1 surface expression and antigen presentation by IFNγ and not IFNβ. Further, these results indicate that transcription of MR1 and HLAA occur via distinct IFNγ-stimulated mechanisms.

Together, our data support a feed-forward model of inflammatory signaling ( Figure 8 ). MR1 antigen presentation by an infected cell activates a MAIT cell to release IFNγ, which then acts on the airway epithelial cell to stimulate IRF1 expression through a pSTAT1 pathway. IRF1 then binds to the MR1 promoter to induce MR1 transcription, leading to more MR1 protein available for antigen presentation and subsequent MAIT cell activation.

This study was conducted according to the principles expressed in the Declaration of Helsinki. Study participants, protocols and consent forms were approved by Oregon Health & Science University Institutional Review Board (IRB00000186). Written and informed consent was obtained from all donors. Human participants are not directly involved in the study. Healthy adults were recruited from among employees at Oregon Health & Science University as previously described to obtain human serum (92).

Primary airway epithelial cells (AEC) were purchased from Lonza Biosciences or harvested from deceased human donor lung tissue through the Cascade Alliance (formerly Pacific Northwest Transplant Bank) as previously described (22, 93). The healthy donor AEC from (22) were likewise grown in Bronchial Epithelial Growth Media (“BEGM”, CC-3170) and harvested using ReagentPack Subculture reagents (CC-5034) per manufacturer’s protocols (Lonza).

The BEAS-2B bronchial epithelial cell line (CRL-9609, American Type Culture Collection) was grown in DMEM medium (Gibco) supplemented with L-glutamine (25030164, Life Technologies) and 10% heat-inactivated fetal bovine serum (“DMEM-FBS”). BEAS-2B cells overexpressing MR1-GFP under a tetracycline-inducible promoter (“BEAS-2B:doxMR1-GFP”) (11) were similarly cultured in DMEM-FBS. Expression of MR1-GFP was induced with doxycycline for 16 hours prior to harvest. BEAS-2B cells stably expressing Cas9 (94) were grown in DMEM-FBS and used to generate CRISPR knockouts.

The MR1-restricted T cell clone (D426G11) was generated and expanded in RPMI medium (Gibco) supplemented with L-glutamine and 10% heat-inactivated human serum (“RPMI-HuS”) as previously described (2, 92).

Streptococcus pneumoniae ( 95) and Mycobacterium smegmatis Mc (2)155 (ATCC) were grown as described in the supplement of (22) and used from frozen stocks. At late log phase, M. smegmatis were pelleted and the supernatant was passed through a syringe-driven 0.22 μm filter and frozen for use as antigen in ELISPOT assays.

We generated IRF1^-/- and NLRC5^-/- CRISPR knockout BEAS-2B cells as previously described (94). Early passage Cas9⁺ BEAS-2B cells were transduced with sgRNA constructs targeting IRF1 (CRISPR845545_LV, ThermoFisher) or NLRC5 (CRISPR1120312_LV, ThermoFisher) in the presence of 200 μg Polybrene (Sigma). Following puromycin selection, monoclonal populations were produced by limiting dilution and screened by Western blot or ELISPOT. We validated genomic editing by Sanger sequencing. DNA was isolated from control Cas9⁺, IRF1^-/-, and NLRC5^-/- BEAS-2B clones using the QIAamp DNA Micro Kit (Qiagen) and amplified by PCR. The OHSU Vollum Institute DNA Sequencing Core performed Sanger sequencing and the resulting sequences were analyzed by TIDE (96) and ICE (97).

6-formylpterin (6-FP, Schirck’s Laboratories) was suspended in 0.01 M NaOH and used at a final concentration of 100 μM. 5-(2-oxopropylideneamino)-6-d-ribitylaminouracil (5-OP-RU) was freshly prepared from equal volumes of 32 mM 5-amino-6-d-ribitylaminouracil (5-A-RU)*HCl (OHSU Medicinal Chemistry Core) (98) and 650 mM methylglyoxal (Sigma) exactly following the second method described in (94) and used at a final concentration of 500 pM. Phytohemagglutinin PHA-L (L4144 Sigma) was suspended in RPMI−HuS and used at 1 μg/well. Doxycycline (Sigma) was suspended in sterile water and used at 2 μg/ml.

Recombinant human cytokines were reconstituted in sterile water and supplemented with bovine serum albumin as per manufacturer recommendations. Final concentrations used were: 66 ng/ml IFNγ (R&D Systems 285-IF-100), 66 ng/ml IFNβ (R&D Systems 8199-IF-010), 132 ng/ml IFNλ (PeproTech 300-02K), 66 ng/ml TNFα (R&D Systems 10291-TA-050), and 66 ng/ml IL-17 (PeproTech 200-17). Cells were treated with cytokines for 12 hours unless otherwise noted.

Antibodies used for ELISPOT assays: α-IFNγ (1-D1K, Mabtech) and alkaline phosphatase-conjugated secondary antibody (7-B6-1-ALP, Mabtech). Antibodies used for Western blot: α-IRF1 (D5E4, Cell Signaling Technology), α-Vinculin (V284, Bio-Rad). Antibodies used for flow cytometry: α-MR1 (26.5, conjugated to APC, Biolegend), α-HLA-A,B,C (W6/32, conjugated to APC, Biolegend), IgG2a isotype (MOPC-173, conjugated to APC, Biolegend), α-phospho-STAT1 (KIKSI0803, conjugated to PE, eBioscience).

Primary AEC were infected with S. pneumoniae (20 MOI) in antibiotic-free BEGM. After 1 hour, AEC were washed with PBS to remove non-adhered bacteria, then MAIT cells were added at a 1:1 ratio in BEGM complete with gentamycin-amphotericin (GA-1000, Lonza). BEAS-2B cells in antibiotic-free DMEM-FBS were infected with M. smegmatis or treated with 6-FP or 5-OP-RU for 1 hour, washed with PBS, then MAIT cells were added at a 1:1 ratio in DMEM-FBS with gentamycin. Following overnight co-culture, wells were extensively washed with PBS to remove MAIT cells before harvesting AEC or BEAS-2B cells.

Cell pellets washed with PBS were either used fresh or stored dry at -80 °C before thawing in 37 °C water bath. RNA was isolated using the RNEasy Plus kit (Qiagen) and cDNA was synthesized using the High Capacity cDNA Reverse Transcription Kit (Life Technologies) as per the manufacturers’ protocols. RT-qPCR was performed using TaqMan (Applied Biosystems) gene expression assays: HPRT1 (Hs02800695_m1), MR1 (Hs01042278_m1), HLA-A,H (Hs01058806_g1), IRF1 (Hs00971965_m1), NLRC5 (Hs01072123_m1), and β2m (Hs00187842_m1). Gene expression data were normalized to internal control HPRT1 and relative expression levels for each target gene were determined using the 2^-ΔΔCt method (99). Some uninfected AEC HPRT1 and MR1 data were used as controls in (22).

To quantify surface expression of MR1 and MHC-I, AEC and BEAS-2B cells were treated as indicated and harvested. Samples were blocked in FACS buffer containing 2% heat-inactivated human serum, 2% heat-inactivated goat serum, and 0.5% heat-inactivated FBS for 30 minutes on ice, then stained with APC-conjugated IgG2a, α-MR1, or α-HLA-A,B,C antibody for 40 minutes. For pSTAT1 staining, cells were permeabilized with 0.2% saponin during the blocking step. Cells were washed with PBS and fixed with 1% paraformaldehyde, then analyzed with a Beckman Coulter CytoflexS. All analyses were performed using FlowJo10 (TreeStar).

IFNγ ELISPOT assays were performed as previously described (100) with the following modifications: ELISPOT plates (MSHAS4510, MilliporeSigma) were coated overnight with α-IFNγ antibody, then washed and blocked for 1 hour in RPMI-HuS. BEAS-2B cells were seeded in duplicate (1x10⁵) cells/well) and infected with M. smegmatis, treated with a titration of M. smegmatis supernatant, or incubated with control PHA or RPMI-HuS medium for 1 hour at 37 C. D426G11 MAIT cell clones were added at a 1:1 ratio in RPMI-HuS with gentamycin for overnight incubation at 37 C. Following extensive washing with PBS−0.05% Tween 20, plates were incubated with ALP secondary antibody for 2 hours before additional washing and colorimetric development. IFNγ spot−forming units (SFU) were quantified by AID ELISPOT reader. For experiments with cytokine pre-treatment, BEAS-2B cells were seeded in 6-well plates and treated with cytokines for 12 hours, then washed 3 times with PBS to remove any excess cytokine before harvesting and seeding into ELISPOT plate.

Gene silencing in wildtype, Cas9⁺, or NLRC5^-/- BEAS-2B cells was performed through nucleofection as in (101) and following the Amaxa Cell Line Nucleofector Kit T (Lonza) protocols. In brief, 2 μg total of Missense (4390843, ThermoFisher), IRF1 (s7501, ThermoFisher), and/or NLRC5 (s38591, ThermoFisher) siRNA were added to 1x10⁶ cells and transfected by the Amaxa Nucleofector 2b machine (Lonza) using program G-016. Cells were incubated for 48 hours before use in assays. Efficiency of gene silencing was validated by RT-qPCR.

Putative transcription factor binding sites were acquired through the Eukaryotic Promoter Database browser using the Search Motif Tool to perform on-the-fly scanning for transcription factor motifs using the FindM tool from the Signal Search Analysis (SSA) Server toolkit (28, 102–104).

All data were analyzed using Prism (GraphPad) and plots were generated using R 4.4.0 and packages such as tidyverse, ggprism, and rstatix. Statistical significance was determined as indicated by two-tailed unpaired or pairwise t tests, using α=0.05.