IL-23 tunes inflammatory functions of human mucosal-associated invariant T cells

Abstract

IL-23 signaling plays a key role in the pathogenesis of chronic inflammatory and infectious diseases, yet the cellular targets and signaling pathways affected by this cytokine remain poorly understood. We show that IL-23 receptors are expressed on the large majority of human mucosal-associated invariant T (MAIT), but not of conventional T cells. Protein and transcriptional profiling at the population and single cell level demonstrates that stimulation with IL-23 or the structurally related cytokine IL-12 drives distinct functional profiles, revealing a high level of plasticity of MAIT cells. IL-23, in particular, affects key molecules and pathways related to autoimmunity and cytotoxic functions. Integrated analysis of transcriptomes and chromatin accessibility, supported by CRISPR-Cas9 mediated deletion, shows that AP-1 transcription factors constitute a key regulatory node of the IL-23 pathway in MAIT cells. In conclusion, our findings indicate that MAIT cells are key mediators of IL-23 functions in immunity to infections and chronic inflammatory diseases.

Keywords: Immunology; Transcriptomics.

Graphical abstract

Figure 1

IL-23 enhances production of cytotoxic molecules and IL-10 by MAIT cells (A) Percentage of IL-23R positive cells in different immune cell populations was assessed by flow cytometry on freshly isolated PBMCs (n = 11). (B and C) PBMC were activated with Tet for 6 days or left unstimulated (black). Sorted MAIT cells (CD3⁺CD161⁺Va7.2⁺) were re-stimulated with Tet in the absence (gray) or presence of IL-23 (red) or IL-12 (blue) for 24 h. Cytokine secretion in the supernatants was measured using Luminex technology (n = 9). (B) Heatmap of measured cytokines. Dendograms on the top and left sides correspond to hierarchical clustering. (C) Boxplots (lines correspond to the first quartile, the median and the third quartile) representing the concentration of the indicated cytokines (adjusted p values, paired Wilcoxon test, Benjamini-Hochberg correction). (D) Single cell secretome of MAIT cells activated for 6 days with Tet in the presence or absence of IL-23 or IL-12 was assessed using Isolight technology. The heatmap represents the single-cell co-secretion patterns and their frequencies (n = 4). The values for granzyme B are ∼15% for the Tet+IL-23 condition, and ∼17% for Tet+IL-12. See also Figure S1 and Tables S1 and S2.

Figure 2

Combined transcriptomic and phenotypic analysis of MAIT cells by single-cell profiling PBMCs were activated with Tet in the presence or absence of IL-23 or IL-12 for 6 days, MAIT cells were sorted (CD3⁺MR1/5-OP-RU⁺) and CITE-seq was performed (n = 3). (A) UMAP of stimulated MAIT cells colored by the 16 identified clusters. Numbers and frequencies of cells in each cluster are indicated. (B) Heatmap showing row-scaled log-transformed normalized expression of the top marker genes for each MAIT cell cluster. (C) UMAP of stimulated MAIT cells colored by the stimulation condition. (D) Proportion of cells from the different stimulation groups in each cluster. Gray: Tet, red: Tet+IL-23, blue: Tet+IL-12. (E) Violin plots of the scores of the indicated signatures by cluster. (F) UMAP of stimulated MAIT cells colored by cytotoxicity genes score. (G) Expression of cytotoxicity genes in MAIT cells from the different stimulation groups. See also Figures S2–S5, Tables S3 and S4.

Figure 3

Mowgli analysis of CITE-seq data identifies factors associated to IL-23 stimulation (A–C) UMAPs of the integration of protein and RNA modalities in CITE-Seq Data for donors 1 (A), 2 (B) and 3 (C). Gray: cells treated with Tet; red: Tet+IL-23; blue: Tet+IL-12. (D–F) Distribution of the weights associated to each factor according to treatment. D: Donor 1, Factor 95; E: Donor 2, Factor 90; F: Donor 3, Factor 98. (G) Spearman correlation across the factors associated to Tet+IL-23 for each donor (black rectangles) in the RNA (left) or antibody space (right). Correlations are shown only between Donor 1 and Donor 2 or 3. For RNA, only genes in common across the highly variable features of each sample were used to compute the correlation. (H and I) Top ten genes (H) and antibodies (I) associated to Tet+IL-23 response in the relevant factor identified by Mowgli in each donor. Colored bars denote the presence of a gene or antibody in the other top ten genes of another donor. Gray bars represent genes or antibody specific for that donor. See also Figure S6.

Figure 4

Genes and pathways regulated by IL-23 or IL-12 in MAIT cells PBMCs were activated with Tet in the presence or absence of IL-23 or IL-12 for 6 days, CD3⁺CD161⁺Va7.2⁺ MAIT cells were sorted, and RNA-sequencing was performed. (n = 5–8) (A) Scatterplot of changes in gene expression (log2-fold change, x axis: Tet+IL-23 vs. Tet, y axis: Tet+IL-12 vs. Tet). Each dot corresponds to a differentially expressed gene (red: Tet+IL-23 vs. Tet, blue: Tet+IL-12 vs. Tet, purple: Tet+IL-23 vs. Tet and Tet+IL-12 vs. Tet). Gray shading indicates the 2D density plot of all the delta log-fold change. (B) Euler diagram representing the differentially express genes in the Tet+IL-23 vs. Tet, and in the Tet+IL-12 vs. Tet stimulation conditions. (C and D) Over-representation analysis was performed on the genes regulated by (C) IL-23 or (D) IL-12 using the GLAD4U database. Bubble plots of the top 20 associated pathways. Orange asterisks indicate pathways associated with chronic inflammatory and autoimmune diseases pathways. Blue asterisks indicate pathways associated with infectious diseases. See also Figure S7 and Tables S5, S6, S7, and S8.

Figure 5

IL-23 regulates MHC class II and AP-1 transcription factor genes in MAIT cells PBMCs were activated with Tet in the presence or absence of IL-23 or IL-12 for 6 days, CD3⁺CD161⁺Va7.2⁺ MAIT cells were sorted, and RNA-sequencing was performed. (n = 5–8). (A) Heatmap of the expression of 65 genes associated with autoimmune pathways. Gray: Tet, red: Tet+IL-23, blue: Tet+IL-12. (B) Boxplots of the expression of HLA genes. Differential gene expression analysis (limma, adjusted p values, TPM: transcripts per million). (C) PBMCs were left unstimulated or activated with Tet in the presence or absence of IL-23 or IL-12 for 6 days, then stained for HLA-II expression. Frequency of HLA-DR, HLA-DM, and HLA-DP positive cells in the MAIT population. (n = 6, paired t test). (D) Boxplots of the expression of AP-1 transcription factor genes.

Figure 6

IL-23 regulates the chromatin landscape in MAIT cells (A–E) Genome browser tracks of ATAC-seq at indicated loci for the Tet and Tet+IL-23 conditions. One representative replicate is shown for each condition. Significant gains or losses of accessibility are displayed as a separate track. (F) Scatterplot of changes in Tet+IL-23 vs. Tet conditions in gene expression (log2-fold change, y axis) as a function of changes in chromatin accessibility (log-fold change, x axis). Each dot corresponds to a differentially expressed and accessible peak-to-gene pair. Red: increase in both accessibility and expression, gray: decrease in both, yellow: increased accessibility, decreased expression, blue: decreased, accessibility, increased expression. Shape indicates position of the DA region relative to the gene’s TSS (triangle: overlapping, square: upstream, circle: downstream). Size is anticorrelated to the distance from the TSS. When several DA regions can be associated to a gene within 10kb, the closest to the TSS is represented. Golden shading depicts the 2D density plot of all log-fold change values. (G) 2D density plots of chromatin accessibility (log2BPM) and RNA expression (log2TPM) in Tet and Tet+IL-23 conditions, overlaid. Selected genes are displayed; arrows indicate direction and magnitude of changes in Tet+IL-23 vs. Tet. See also Figure S8 and Table S9.

Figure 7

IL-23 signals in MAIT cells are mediated by the AP-1 transcription factors BATF, JUNB, and BACH2 (A) Genome-wide in silico footprinting analysis was computed on ATAC-seq data from MAIT cells activated with Tet ± IL-23. Volcano plot of the genome-wide transcription factor binding changes. Each dot represents one motif, empty dots indicate non-significant events. (B) Scatterplot displaying log2-fold change of gene expression against differential binding score for each of the transcription factors used for footprinting. In the case of heterodimers, the gene expression log2-fold change corresponds to the mean of the two genes. Gray dashed lines link the different TF binding prediction for TF binding to several motifs. (C–E) PBMC were activated with Tet for 6 days. At day 5 of culture, PBMC were transfected with fluorescently labeled Cas9 RNP targeting BATF (Orange) or non-targeting (Control, gray). At day 6, transfected MAIT cells (CD3⁺CD161⁺Va7.2⁺ ATTO550⁺) were sorted, cultured for 4 additional days and gene expression and cytokine secretion in the supernatant were assessed (n = 3). (C) Schematic representation of the experimental design. (D) Boxplots representing gene expression (normalized RNA counts, nCounter technology) of BATF and HLA-DRA (n = 3, p-values, paired t test). (E) Concentration of IL-10 measured in the supernatant using MSD technology (n = 3). See also Table S11.