Young (6-8 weeks) and aged (20 months) C57BL/6 mice of both sexes were bred in the animal research facility. All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee of Nanjing Drum Tower Hospital.

De-identified intestinal tissues were collected at Nanjing Drum Tower Hospital. Relative normal tissues from patients with colorectal cancers were obtained and stored in cold SPS-1 storage buffer before immediate processing. De-identified samples from 7 individuals of middle age (45-65) and both sexes were obtained. The study was approved by the Ethics Committee of Nanjing Drum Tower Hospital.

For isolation of lymphocytes from mouse and human intestines, intestine tissues were cut into small pieces, and epithelial cells were dissociated with HBSS containing 5 mM EDTA, 1 mM DTT, and 2% FBS, shaken for 20 minutes at 37°C. This process was performed twice. After each step, samples were vortexed, rinsed with PBS, and the epithelial cell layer was discarded. The remaining tissue was digested in HBSS with 1 mg/mL Liberase TM (Roche), and 200 μg/mL DNase I (Roche) for 30 minutes at 37°C. Percoll centrifugation was performed to remove dead cells and debris. The resulting single cells were strained through a 70 μM mesh.

For isolation of lymphocytes from mouse lungs, euthanized mice were perfused by injecting 20 ml cold PBS into the right ventricle of the heart. The lung tissues were minced by a scissor, and digested in HBSS with 1 mg/mL Liberase TM, and 200 μg/mL DNase I for 30 minutes at 37°C. The cells were then filtered through a 70 μM mesh.

For isolation of skin lymphocytes, the ear pinnae were obtained and separated into ventral and dorsal sheets. The sheets were digested in RPMI 1640 media with 0.5 mg/mL DNase I, and 0.25 mg/mL Liberase TL (Roche) for 1 hour and 45 minutes at 37°C. After digestion, the skin sheets were homogenized, followed by Percoll centrifugation to remove dead cells and debris. The single cell suspension was then passed through 70-μm filters.

Cells were stained with MR1-5-OP-RU tetramers together with surface antibodies at room temperature for 30 minutes. MR1 tetramers were generated by tetramerizing biotinylated monomers loaded with 5-OP-RU (Immundex) using streptavidin conjugated to APC (Thermo), as previously described (3, 29). Antibodies for human samples include anti-CD45 (HI30, cat#MHCD4528, Thermo), anti-CD3 (OKT3, 317338, Biolegend), anti-CD161 (130-113-591, cat#130-113-591, Miltenyi Biotec), anti-RORγt (AFKJS-9, cat # 12-6988-82, Thermo). Antibodies for mouse samples included anti-CD45 (104, cat# 109820, Biolegend), anti-CD3 (145-2C11, cat#100326, Biolegend), anti-IL18R (P3TUNYA, 12-5183-82, Thermo), anti-CD90 (53-2.1, cat#140310, Biolegend), anti-RORγt,(B2D, cat# 12-6981-82, Thermo). Intracellular staining of RORγt was performed using FoxP3 Fix/Perm Kit (Thermo) according to the manufacturer’s instructions. Ki67 staining was performed using the PE Mouse Ki67 Staining Kit (BD Biosciences) according to the manufacturer’s instructions. Annexin V staining was performed with the PE Annexin V Apoptosis Detection Kit (BD Biosciences) following the manufacturer’s instructions. For fluorescence activated cell sorting (FACS), human MAIT cells were sorted by fluorescence activated cell sorting (FACS) as CD45+ CD3 + 5-OP-RU tetramer+, with CD161 included to assist with gating and to exclude debris and nonspecifically stained cells. Mouse MAIT cells were identified as CD45+ CD3 + 5-OP-RU tetramer+, with IL18R included for gating assistance and to exclude debris and nonspecifically stained cells.

For single cell RNA sequencing with mouse cells, MAIT cells were pooled and sorted from samples from 5 young or 5 aged mice. For single cell RNA sequencing with human cells, MAIT cells were pooled and sorted from samples from 7 different human donors. scRNA-seq libraries were prepared using the chromium 5’ single cell gene expression kit from 10x genomics according to the manufacturer’s instructions. Sequencing was carried out using a NextSeq 500 (Illumina). Analysis was carried out using Cellranger. Normalization of the data was achieved using the SCTransform function within Seurat. Cell clustering was performed using Uniform Manifold Approximation and Projection (UMAP). Feature plots, and violin plots were generated using normalized data. A Wilcoxon rank sum test was used as a statistical significance test.

The GSEA algorithm was utilized to access the enriched gene sets from differential genes. In brief, the differential gene expression of targeted two groups was calculated with the R package limma 3.56.2. Then the differential genes were preranked by log2FoldChange and subjected to the R package clusterProfiler 4.8.3. Results with p value<0.05, NES>1 and False discovery rate<0.25 were considered significant.

For RT–qPCR analysis, mRNA was isolated from FACS sorted MAIT cells using either the RNeasy Plus Mini Kit (Qiagen), following the manufacturer’s protocols. Complementary DNA (cDNA) was synthesized using SuperScript II Reverse Transcriptase (Thermo). qPCR was carried out using ABI pre-optimized TaqMan probes. Probes for mouse genes included those to detect Ifng (Mm01168134_m1), Tbx21 (Mm00450960_m1), Ccl5 (Mm01302427_m1), Ccl4 (Mm00443111_m1), Il17a (Mm00439618_m1), Il22(Mm01226722_g1), Ccr6(Mm99999114_s1), Il23r(Mm00519943_m1), Gzmb (Mm00442837_m1), Nkg7(Mm01205900_g1), Prf1(Mm00812512_m1), Gzma (Mm01304452_m1), and Gapdh (Mm99999915_g1). Gene expression levels were calculated relative to Gapdh for normalization.

Survival analysis was conducted using RNA-seq Data data and clinical information from 446 colon cancer (COAD) patients, obtained from The Cancer Genome Atlas (TCGA). The primary objective was to investigate the correlation between gene expression and overall survival outcomes. To assess the prognostic significance of specific gene expressions, an univariate Cox proportional-hazards regression model was applied to each gene with the “coxph” function of the R package “survival”. The hazard ratios (HR) and corresponding 95% confidence intervals (CI) were calculated for both genes, with p-values indicating the significance of their association with survival. To dichotomize the patients into high-expression and low-expression groups for IL17A and RORC, the optimal cut-off points were determined by the “surv_cutpoint” function of the R package “survminer”. The cut-off values were selected based on the maximization of the log-rank statistic. Kaplan-Meier survival curves were plotted for the survival rates of high-expression and low-expression groups. The differences in overall survival between the two groups were statistically evaluated using the log-rank test.

For culturing of mouse MAIT cells, 96-well culture plates were bound with 3μg/ml of anti-CD3 (clone 2C11) and anti-CD28 (clone 37.51) antibodies overnight, before MAIT cell culture. 5000 mouse MAIT cells were sorted by FACS and cultured with DMEM medium supplemented with 1% ITS-G supplement (Thermo), 1% of penicillin and streptomycin, 100ng/ml IL-7 (R&D), in the presence of 5μM RORγt inhibitor GSK805. Cells were cultured for 3 days, and cell-free culture supernatant was collected. Concentrations, cytokines and chemokines were measured by bead-based multiplex assays using LegendPlex kits (Biolegend) according to the manufacterer’s instructions. Granzyme B concentrations were measured by ELISA (Thermo).

For culturing of human MAIT cells, 96-well culture plates were bound with 1μg/ml of anti-CD3 (clone OKT3) and 5μg/ml of anti-CD28 (clone CD28.2) antibodies overnight, before cell culture. 50,000 human MAIT cells were sorted by FACS and cultured with DMEM medium supplemented with 1% ITS-G supplement (Thermo), 1% of penicillin and streptomycin, 100ng/ml IL-7, and 5μM RORγt inhibitor GSK805 or vehicle control. Cells were cultured for 2-3 days and supernatant were collected. Concentrations of cytokines, chemokines and cytotoxic molecules were measured by bead-based multiplex assays using the LegendPlex kits (Biolegend).

Wilcoxon rank sum test was used to determine the significance of scRNA-seq data. Student T cells or one-way ANOVA was used for data obtained from other experiments. p<0.05 was considered significant.

Little is known about the effects of aging on innate-like T cells. To investigate the effects of aging on MAIT cells, we performed flow cytometry analysis to examine MAIT cells in the large intestinal laminal propria (LILP) of young (4-6 weeks old) and aged (20-month-old) mice. MAIT cells were identified as T cells that stained positively with MR1 tetramers. Because MAIT cells in mouse tissues express IL18R (30, 31), we used IL18R to help gate MR1 tetramer-staining positive cells and to exclude debris and cells with nonspecific tetramer binding ( Figure 1A ). Notably, MAIT cell percentages and abundance greatly increased with aging ( Figures 1A–C ). Our quantification of MAIT cells was based on the total MAIT cells present in the entire large intestine tissue from each mouse ( Figure 1C ). Because isolating lymphocytes from the intestines is a highly time-sensitive procedure, we were unable to measure tissue weight before digesting the samples. The identity of MAIT cells was validated by their high expression of characteristic surface markers including CD90, IL18R and IL7R ( Figure 1D ). Ki67 expression was comparable between MAIT cells from young and aged mice ( Figures 1E, F ). Annexin V staining revealed minimal apoptosis in MAIT cells isolated from both young or aged mice ( Figures 1G, H ). These results suggest that aging does not significantly affect the proliferation or survival of MAIT cells. Furthermore, a similar accumulation of MAIT cells was observed in the lung and the skin of aged mice, indicating that age-related increase in MAIT cells with aging is not restricted to intestines ( Figures 1I–L ). Together, these data indicate that MAIT cells accumulate at mucosal barriers with aging in mice.

To examine the effects of aging on the gene expression profiles of MAIT cells, we performed single-cell RNA-sequencing (scRNA-seq) of MAIT cells sorted from LILP of young and aged mice ( Figure 2A ). The identity of MAIT cells was validated by their high expression of characteristic marker genes such as Rorc and Zbtb16 ( Figure 2B ). Mouse intestinal MAIT cells were initially separated into four subsets by UMAP ( Figure 2A ). Subset 3 expressed high levels of proliferating gene markers such as Mki67, and were therefore annotated “MAIT_proliferating” ( Figures 2C, D ). Subset 1 and 2 exhibited similar transcriptome profiles, showing low expression of various lymphocyte activation gene (e.g. Icos, Lcp2, Lck, Lat, Nfkb1, Cd40lg, Cd69, Nfil3) and effector molecule genes including Ifng and Il17a ( Figures 2A, D ). These cells were thus annotated as “MAIT_resting” ( Figure 2C ). Subset 0 expressed high levels of genes associated with lymphocyte activation, and were thus annotated as MAIT_activated ( Figure 2C ). While gene positively regulate T cell activation (Icos, Lcp2, Lck, Lat, Nfkb1, Cd40lg, Cd69, Nfil3) were expressed at the highest levels in MAIT cells, checkpoint genes negatively regulate T cell activation (Pdcd1, Ctla4) were expressed at the highest levels in MATI_proliferating cells. Notably, MAIT cells from aged mice had a significantly higher average frequency of MAIT_resting cells compared to those from young mice, indicating that aging is associated with an accumulation of MAIT cells exhibiting an inactive phenotype. ( Figures 2E, F ). The frequencies of MAIT_activated and MAIT_proliferating cells were lower in MAIT cells from aged mice ( Figures 2E, F ). However, due to the overall increase in the total number of MAIT cells, the abundance of MAIT_proliferating was similar between young and aged mice, while the abundance of MAIT_activated increased with age. The increased percentage of resting MAIT cells in aged mice was also associated with decreased expression of several T cell activation genes (Icos, Nfkb1, Cd40lg, Nfil3) and checkpoint genes (Pdcd1, Ctla4) in the overall MAIT cell population ( Supplementary Figure S1 ).

Interestingly, our scRNA-seq data revealed that MAIT cells from young mice primarily expressed Il17a, with minimal Ifng expression, while MAIT cells from aged mice showed a strong preference for Ifng ( Figure 2G ). Since the majority of cytokine-producing cells are found in the MAIT_activated subset, we focused our analysis on this group. GSEA indicated that gene sets associated with the interferon-gamma pathway and killer cell activity were significantly upregulated in MAIT_activated cells from aged mice compared to those from young mice ( Figure 2H ). MAIT_activated cells in aged mice exhibited markedly higher expression levels of Ifng, T helper type 1 (Th1)-related chemokines Ccl5 and Ccl4, and the Th1 transcription factor Tbx21 (encoding T-bet). In contrast, MAIT_activated cells from young mice expressed elevated levels of T helper type 17 (Th17)-related cytokines and receptors, including Il17a, Il22, Ccr6, and Il23r ( Figure 2I ). The expression of cytotoxic molecules varied with aging. Granzyme B (Gzmb) was more highly expressed in MAIT_activated cells from young mice, whereas the expression of other cytotoxic genes, Nkg7 and Prf1 (encoding perforin), was elevated in aged mice ( Figure 2I ). Thus, aging induced a shift from MAIT1 to MAIT17 profiles in MAIT cells, with variable effects on cytotoxic molecules.

To validate our scRNA-seq findings, we performed extensive QPCR analysis on MAIT cells sorted from individual mice. Post sorting flow cytometry analysis verified high post sorting purity ( Supplementary Figure S2 ). QPCR results confirmed that MAIT cells from aged mice expressed higher levels of Th1-like cytokines and chemokines, including Ifng, Ccl5, and Ccl4 ( Figure 3A ). In contrast, MAIT cells from young mice showed elevated expression of Th17-like cytokines and associated receptors, such as Il17a, Il22, Ccr6, and Il23r ( Figure 3A ). Additionally, the Th1-related transcription factor Tbx21 was significantly upregulated in MAIT cells from aged mice compared to those from young mice ( Figure 3A ). While Gzmb was upregulated in MAIT cells from young mice, other cytotoxic genes, such as Nkg7 and Prf1, were more highly expressed in MAIT cells from aged mice ( Figure 3A ). Although Gzma mRNA was barely detectable in the scRNA-seq data, it was identified in MAIT cells from aged mice but not from young mice through QPCR analysis ( Figure 3A ). Together, our results indicate that intestinal MAIT cells shift from a MAIT17 profile to a MAIT1 profile with aging, accompanied by decreased expression of Gzmb and increased expression of other cytotoxic molecules.

To assess whether the effector functions of MAIT cells change with aging, we sorted MAIT cells by FACS and cultured them in vitro with plate-bound anti-CD3 and anti-CD28 antibodies, adding IL-7 to enhance cell survival. Both MAIT cells from young and aged mice proliferated at similar rates in vitro ( Figure 3B ). We then measured the concentrations of cytokines and chemokines in the culture supernatants using Legendplex, a bead-based multiplex kit. Th1-like cytokines and chemokines, including IFNγ, CCL3, and CCL4, were readily detectable in the culture supernatant of MAIT cells from aged mice, but not from young mice ( Figure 3C ). CCL5 concentrations were also significantly higher in aged mice compared to their young counterparts ( Figure 3C ). In contrast, IL17 was prominently detected in the supernatants of MAIT cells from young mice, but not in those from aged mice ( Figure 3C ). CCL20, a chemokine associated with Th17 cells, was similarly present in young mice’s culture supernatants but absent in those from aged mice ( Figure 3C ). Additionally, MAIT cells from young mice produced higher levels of Granzyme B compared to those from aged mice, as measured by ELISA ( Figure 3D ). Other cytotoxic molecules were not assessed due to the unavailability of reliable ELISA or multiplex assay kits for those targets. Collectively, these data suggest that aging induces a functional shift in MAIT cells from a MAIT17 profile to a MAIT1 profile.

RORγt is a transcription factor highly expressed in MAIT cells ( Figure 2B ). Using flow cytometry analysis, we verified positive expression of RORγt in MAIT cells from both young and age mice ( Figure 4A ). However, the expression level of RORγt showed a moderate decrease with aging ( Figure 4B ). To investigate the role of RORγt in regulating MAIT cell function, we sorted MAIT cells by FACS and cultured them with the RORγt inhibitor GSK805 and vehicle control. Post sorting analysis verified high post sorting purity ( Supplementary Figure S3 ). Inhibition of RORγt did not significantly affect MAIT cell growth in vitro ( Figure 4C ). However, we observed notable changes in cytokine and chemokine production ( Figure 4D ). Specifically, GSK805 significantly increased the production of Th1 cytokines and chemokines, including IFNγ, CCL3, CCL4, and CCL5, in MAIT cells from aged mice ( Figure 4D ). In contrast, the production of Th17-related cytokines and chemokines, such as IL17, IL22, and CCL20, in MAIT cells from young mice was diminished by GSK805 ( Figure 4D ). The production of Granzyme B remained unchanged with GSK805 treatment ( Figure 4E ). Together, these data indicate that RORγt promotes MAIT17 effector molecules while repressing MAIT1 effector molecules.

To examine the transcriptomes of human intestinal MAIT cells, we performed scRNA-seq on human colon laminal propria MAIT cells, identified using MR1 tetramers and characterized by high levels of CD161 ( Figure 5A ). Due to the effects of aging on MAIT cells, we obtained de-identified samples from seven individuals of middle age (45-65 years), to ensure inclusion of both MAIT1 and MAIT17 profiles. The identities of the MAIT cell transcriptomes were validated by high expression levels of the characteristic transcription factors RORC and ZBTB16 ( Figures 5B, C ). UMAP analysis clearly separated MAIT1 and MAIT17 cells ( Figures 5D, E ). Subsets 0 and 2, which exhibited similar transcriptomes with elevated IFNG expression, were annotated as “MAIT1” ( Figures 5D, E ). Subsets 1 and 3, which were close at the transcriptomic level and showed high IL17A expression, were designated as “MAIT17” ( Figures 5D, E ). Proliferating MAIT cells did not form distinct clusters ( Figure 5E ). We then compared gene expression profiles between MAIT1 and MAIT17 cells. GSEA revealed that gene sets associated with lymphocyte cytotoxicity were upregulated in MAIT1 cells, while genes related to inflammatory cytokine responses were elevated in MAIT17 cells ( Figure 5F ). In addition to IFNG, MAIT1 cells exhibited higher levels of Th1-related chemokines, including CCL3, CCL4, and CCL5 ( Figure 5G ). Conversely, MAIT17 cells expressed higher levels of Th17-related cytokines and chemokines, such as IL17A and CCL20. MAIT17 cells also showed increased expression of GZMB, whereas MAIT1 cells expressed higher levels of other cytotoxic genes, including GZMA, NKG7, PFR1, and GNLY ( Figure 5G ). Furthermore, MAIT1 cells expressed higher levels of the Th1 characteristic transcription factor TBX21 and the cytotoxic lymphocyte transcription factor EOMES ( Figure 5G ). TCF7 expression was also higher in MAIT1 cells compared to MAIT17 cells ( Figure 5G ). These gene expression patterns closely resemble the aging-related changes observed in mouse MAIT cells. Together, these findings indicate that the transcriptomes of human intestinal MAIT cells separated into MAIT1 and MAIT17 subsets, exhibiting expression patterns of effector molecules similar to those seen in aging-related changes in mouse MAIT cells.

We used the RORγt inhibitor GSK805 to further investigate the role of RORγt in regulating human MAIT cell function. Flow cytometry analysis revealed that nearly all MAIT cells expressed RORγt ( Figure 6A ). Inhibition of RORγt by GSK805 led to an increase in the production of Th1-related cytokines and chemokines, including IFNγ, CCL3, CCL4, and CCL5, while the production of Th17-related cytokines and chemokines, such as IL17A and CCL20, was significantly reduced ( Figures 6B, C ). Additionally, the production of cytotoxic molecules by human MAIT cells remained unchanged following GSK805 treatment ( Figure 6D ). Together, these data indicate that RORγt inhibits the production of Th1-like cytokines and chemokines while promoting Th17-like cytokines and chemokines, without affecting the production of cytotoxic molecules in human MAIT cells.

To determine the association of MAIT effector function with disease prognosis, we examine the correlation between the expression of MAIT cell effector genes with patient survival using the gene expression data and clinical information of 446 patients from TCGA-COAD cohorts (32). We first applied a univariate Cox proportional-hazards regression model to assess the correlate of each MAIT cell effector molecule gene with patient mortality. Our results reveal that a lower level of IL17A and RORC in tumors correlates with worse survival in COAD patients ( Figure 7A ). We then separated patients into high-expression and low-expression groups for IL17A, RORC, and IFNG, using a best fit model to determine the cut-off points. Patients with lower expression of IL17A and RORC in tumors exhibited significantly poorer survival compared with those with higher expression of IL17A and RORC ( Figures 7B, C ). In contrast, the levels of IFNG were not associated with prognosis ( Figure 7D ). Together, these data indicate that expression levels of the MAIT17 signature gene IL17A and a positive MAIT17 regulator RORC positively correlate with prognosis of colon cancers.