Tissue signals imprint Aiolos expression in ILC2s to modulate type 2 immunity

Abstract

Group 2 innate lymphoid cells (ILC2s) manifest tissue heterogeneity and are crucial modulators of regional immune responses. The molecular mechanisms regulating tissue ILC2 properties remain elusive. Here, we interrogate the signatures of ILC2s from five tissues at the transcriptome and epigenetic level. We have found that tissue microenvironment strongly shapes ILC2 identities. The intestine induces Aiolos⁺ILC2s, whereas lung and pancreas enhance Galectin-1⁺ILC2s. Though being a faithful gut ILC2 feature under the steady state, Aiolos is induced in non-intestinal ILC2s by pro-inflammatory cytokines. Specifically, IL-33 stimulates Aiolos expression in both human and mouse non-intestinal ILC2s. Functionally, Aiolos facilitates eosinophil recruitment by supporting IL-5 production and proliferation of ST2⁺ILC2s through inhibiting PD-1. At the epigenetic level, ILC2 tissue characters are imprinted by open chromatin regions (OCRs) at non-promoters. Intestinal-specific transcription factor aryl hydrocarbon receptor (Ahr) binds to Ikzf3 (encoding Aiolos) locus, increases the accessibility of an intestinal ILC2-specific OCR, and promotes the Ikzf3 transcription by enhancing H3K27ac. Consequently, Ahr prevents ILC2s entering an "exhausted-like" state through sustaining Aiolos expression. Our work elucidates mechanism of ILC2 tissue adaptation and highlights Aiolos as a potential target of type 2 inflammation.

Fig. 1. Tissue ILC2s have distinct signatures.

a, b, e, g, k, l Analysis of aggregated and downsampled single-cell data from tissue ILC2s was performed using Seurat. a t-SNE plot shows the distinct clusters of tissue ILC2s. b Heatmap shows the top 30 differentially expressed genes from tissue ILC2s performed using “1 versus all” comparison. e, g, k Violin plots show differentially expressed genes by tissue ILC2s. c, d, f, h, i, j Expression of ILC2 tissue-specific genes was verified in purified ILC2s from female mice (c, d, f, and j) or both genders (h and i) by real-time RT-PCR. c, d, f, h, i, j Heatmaps show relative fold of gene expression of representative tissue signatures relative to the expression by BM ILC2s. Black blocks indicate the gene was not detected in the tissue ILC2s by real-time RT-PCR. Scale bar indicates z-score of gene expression relative to Actb. l Gene ontology analysis was performed on LI and SI ILC2 common signature genes. BM bone marrow, LI large intestine, SI small intestine, P pancreas, L lung.

Fig. 2. Tissue environment strongly shapes ILC2 identities.

a The expression of representative ILC2 tissue signatures gated on ILC2s was analyzed by flow cytometry. Intracellular 4-1BB was analyzed when cells were stimulated with PMA and ionomycin. Myeloid cell gate based on FSC and SSC was used as negative control. b Histogram shows Aiolos expression gated on human ILC2s from different tissues analyzed by flow cytometry. Negative control was Lin⁺ cells from bone marrow sample. c Schematic strategy of ILC2 transfer. d–g ILC2s were sorted from different tissues and transferred to Rag2^–/–Il2rg^–/– mice. Expression of Aiolos (d and e) and Gal-1 (f and g) gated on ILC2s (Lin^–GATA3⁺) from different tissues in Rag2^–/–Il2rg^–/– hosts was analyzed by flow cytometry 4 weeks after transfer. Mean fluorescence intensity (MFI) of Aiolos and Gal-1 is shown. h, i Mononuclear cells isolated from different tissues of the CD45.1 and CD45.2 mice were mixed at a 1:1 ratio and cultured for 48 h. The expression of Aiolos and Gal-1 gated on ILC2s (CD45.1⁺Lin^–GATA3⁺ cells) of CD45.1 origin was analyzed by flow cytometry. e–i Error bars are mean + SEM. a–i Data are representative of two to four independent experiments. BM bone marrow, LI large intestine, SI small intestine, P pancreas, L lung.

Fig. 3. Aiolos is induced in non-intestinal ILC2s by IL-33.

a–c Purified lung ILC2s were treated with indicated cytokines in the presence of IL-7 (10 ng/ml) for 24 h. Concentration for all cytokines were used at 10 ng/ml except for TGF-β (1 ng/ml), TL1A (100 ng/ml) and TSLP (20 ng/ml). a mRNA expression of Ikzf3 was analyzed by Q-PCR. b Expression of Aiolos was analyzed by flow cytometry. c Mean fluorescence intensity (MFI) of Aiolos is shown. d ILC2s from different tissues of wild-type mice were sorted and cultured with IL-7 (10 ng/mL) and IL-33 (10 ng/mL) for 14 days. The expression of Aiolos in IL-33-treated ILC2s compared to freshly isolated ILC2s from indicated organs was analyzed by flow cytometry. e Expression of Aiolos in PBS or IL-33 (intraperitoneal injection of IL-33, 500 ng/day/mouse) treated mice gated on ILC2s from indicated organs was analyzed by flow cytometry. Histogram of Aiolos expression is shown. f Human peripheral blood ILC2s were isolated and treated with indicated cytokines for 5 days. Histogram shows Aiolos expression analyzed by flow cytometry. a, c Error bars are mean ± SEM. d–e BM bone marrow, LI large intestine, SI small intestine, P pancreas, L lung. a–f Data are representative of two to four independent experiments.

Fig. 4. Aiolos supports IL-5 expression by suppressing PD-1.

a–c, i–q ILC2s were purified from the large intestine of wild-type (a–c) or Pdcd1^–/– mouse (o–q) and cultured in vitro in the presence of IL-7 and IL-33 for 5 days. ILC2s were infected with control retrovirus or retrovirus expressing Ikzf3-shRNA#1 or Ikzf3-shRNA#2 (all have GFP as reporter) for two rounds. a Expression of indicated genes in purified virus-infected ILC2s (GFP⁺ cells) was analyzed by real-time RT-PCR. b, c, i–q GFP⁺ virus-infected cells were purified and cultured starting with equal cell numbers for another 48 h before analysis (in the presence of hamster IgG or α-PD1 antibody for l–n). Expression of PD-1 (b and c), IL-5 (b, c, l, o), Ki67 (i, m, p) gated on GFP⁺ cells was analyzed by flow cytometry. b Isotype control was used to set up gates for IL-5⁺ or PD-1⁺ cells. j, n, q Absolute numbers of ILC2s were shown. k Concentration of IL-5 in culture supernatant was analyzed by ELISA. d–f ILC2s infected with control virus expressing GFP or virus expressing Ikzf3-shRNA#2 were transferred to Rag2^–/–Il2rg^–/– mice. d, e Four weeks after transfer, LI LPLs were isolated and analyzed. Expressions of IL-5 and PD-1 gated on Lin^–GFP⁺ cells from host mice were analyzed by flow cytometry. FACS plots and statistics are shown. d Gating control for IL-5 and PD-1 were gated on Lin⁺ cells and Lin^–GFP^– cells, respectively. f Mice were injected with IL-33 for 4 days and were then sacrificed for analysis. Percentages of eosinophils (CD45⁺CD11b⁺Siglec-F⁺) gated on live leukocytes (CD45⁺ cells) are shown. g, h ILC2s from human PBMC were purified and cultured in the presence with IL-2, IL-7 and IL-33 for 3 days. Cells were treated with control siRNA or siRNA targeting IKZF3 for 48 h before harvested for analysis. g IKZF3 and IL5 mRNA expression were analyzed by real-time RT-PCR with GAPDH as reference and normalized to the expression of control siRNA group as 1. h Brefeldin A was added 2 h before cells were harvested for analysis. Histogram shows expression of Aiolos and IL-5 gated on ILC2s analyzed by flow cytometry. a–q Data are representative of at least three independent experiments. Error bars are mean + SEM. c, i, j, k Connected lines indicate a data pair using cells from same mouse.

Fig. 5. ILC2 tissue-specific open chromatin regions (TS OCRs) correlate with ILC2 tissue signatures.

ATAC-seq analysis was performed using ILC2s isolated from different tissues. a PCA analysis based on normalized coverage of all combined peaks observed in tissue ILC2s. b Correlation analysis was performed on normalized coverage of ILC2 OCRs from indicated regions. Spearman correlation coefficients are shown. c Hierarchical clustering and heatmap was generated based on z-score of the normalized coverage of ILC2 TS OCRs. Representative ILC2 TS OCRs correlated ILC2 TS genes are shown. d Genome browser tracks with red boxes indicating representative ILC2 TS OCRs are shown. e Heatmap shows hierarchical clustering based on p value ILC2 TS OCRs-enriched motifs. White boxes indicate the motif is not found to be enriched in indicated ILC2 TS OCRs, or the p value was higher than the cutoff value (1 × 10⁻⁶). f mRNA expression of indicated genes from isolated tissue ILC2s was analyzed by real-time RT-PCR. Heatmap shows relative average expression of indicated genes normalized to the expression of BM ILC2s as 1. *indicates significant difference in increased expression of tissue-specific genes compared with each of the other tissue ILC2s analyzed from four biological samples.

Fig. 6. Ahr promotes Aiolos expression in ILC2s cell intrinsically.

a LI and SI ILC2 commonly specific OCRs and differentially decreased peaks correlated genes were overlapped with LI and SI ILC2 common signature genes. b, c Purified LI ILC2s were treated with DMSO or FICZ (1 nM) for 24 h and analyzed with flow cytometry. Mean fluorescence intensity (MFI) of Aiolos (b) and PD-1 (c) gated on live cells were shown. d, f Expression of Aiolos and PD-1 in tissue ILC2s (Lin^–GATA3⁺) from indicated organs of littermate wild-type or Ahr^–/– mice was analyzed by flow cytometry. e, g Mixed bone marrow cells at 1:1 ratio from CD45.2-Ahr-deficient and CD45.1-wild-type mice were transferred to half-lethally irradiated Rag2^–/–Il2rg^–/– mice. Six weeks later, expression of Aiolos (e) and PD-1 (g) in ILC2s from indicated tissues was analyzed by flow cytometry. Percentages of Aiolos⁺ cells or PD-1⁺ cells gated on ILC2s from indicated origins (CD45.1⁺Lin^–GATA3⁺ or CD45.2⁺Lin^–GATA3⁺) are shown. Error bars are mean + SEM. Data were pooled from three to five mice of two independent experiments. h–j ILC2s were purified from the large intestine of wild-type (WT, dashed line) or Ahr^–/– mouse and cultured in vitro in the presence of IL-7 and IL-33. Expression of IL-5 was analyzed by flow cytometry on day 3 (h) or day 10 (i) with brefeldin A added in the last 2 h. j Expression of Ki67 was analyzed by flow cytometry on day 10 after culture. k–o ILC2s were expanded for 5 days and infected with retrovirus expressing MIG or Ikzf3-MIG twice in the next 2 days. GFP⁺ virus-infected cells were purified and re-plated starting with equal cell numbers for another 48 h before cells were analyzed. k–n Expression of Aiolos (k), PD-1 (m), and Ki67 (n) gated on GFP⁺ cells was analyzed by flow cytometry. l MFI of Aiolos. o Absolute numbers of IL-5⁺ILC2s were total numbers of ILC2 time the percentages of IL-5⁺ cells. m–o Connected lines indicate cells from same mouse and same batch of experiment. b–o Data are from two to four independent experiments. d–g, l Error bars are mean + SEM. LI large intestine, SI small intestine.

Fig. 7. Ahr promotes the transcription of Ikzf3 epigenetically.

a mRNA expression of Ikzf3 in IL-33-treated ILC2s for 7 days was analyzed by real-time RT-PCR. b Genome browser tracks of tissue ILC2 ATAC-seq peaks with published ATAC-seq peaks and ChIP-seq peaks at the Ikzf3 locus. From top to bottom: ATAC-seq data on tissue ILC2s; published ATAC-seq data performed using Ahr-deficient and control ILC2s of Rag1^–/– mouse; published ChIP-seq data on small intestinal ILC2s; published ATAC-seq data on small intestinal ILC2s, ILC1s and ILC3s. Boxes highlight OCRs. Red boxes highlight Ahr-dependent OCRs. c Mixed bone marrow cells at 1:1 ratio from CD45.2-Ahr-deficient and CD45.1-wild-type mice were transferred to half-lethally irradiated Rag2^–/–Il2rg^–/– mice. Six weeks later, expression of Aiolos was analyzed by flow cytometry. Percentages of Aiolos⁺ cells gated on Lin^–non-ILC2s (CD45⁺Lin^–GATA3^{low and negative}) ILC2s from indicated origins are shown. d–g ChIP Q-PCR was performed on Ahr-deficient or wild-type (WT) ILC2s using IgG control, or α-Ahr, or α-H3K27ac antibodies. Relative enrichment of Ahr (d and e) or H3K27ac (f and g) at indicated locus marked in (b) normalized to IgG control group of WT-ILC2s is shown.