RAG deficiency leads to expansion and activation of ILC2s during inflammation and at steady state.

A) Experimental schematic of AD-like disease. WT (Control) B6 mice or Rag1−/− mice treated topically to the inner surface of each ear with 2 nmol MC903 in 10 μL ethanol vehicle daily for 7 days develop AD-like inflammation. B) Ear thickness measured daily in AD-like inflammation. Data representative of at least 2 independent experiments, 5 mice/group. ** P < 0.01 by 2-way ANOVA with Sidak’s multiple comparisons test, day 7. C) Proportion of CD90+, Lin− cells (Lin− defined as CD3−, CD5−, CD11b−, CD11c−, CD19−, NK1.1−, and FcεR1−) determined to be ILC2s (IL-33R+) in skin-draining lymph nodes (sdLN) from WT or Rag1−/− mice with AD-like ear inflammation. Percent ILC2 from sdLN in AD-like disease following PMA/iono stimulation positive for D) IL-5 or E) IL-13 staining. F) Schematic of steady state analysis of sdLN from WT (Control) or Rag1−/− mice. G) Proportion of steady state sdLN CD90+, Lin− cells determined to be ILC2s as in (C). Percent ILC2 from sdLN in steady state following PMA/iono stimulation positive for H) IL-5 or I) IL-13 staining. C-E; G-I) Data representative of at least 2 independent experiments, 4-5 mice/group * P < 0.05, ** P < 0.01, *** P < 0.001 by two-tailed Welch’s t test. All data represented as mean with standard deviation.

Homeostatic expansion and activation of RAG-deficient ILC2s is cell intrinsic

A) Experimental schematic of AD-like disease in splenocyte chimera experiment. WT B6 or Rag1−/−mice received WT splenocytes and developed AD-like inflammation after subsequent topical treatment with 2 nmol MC903 in 10 μL ethanol vehicle to each ear daily for 10 days. B) Ear thickness measured daily in AD-like inflammation. Data representative of 2 independent experiments, 4-5 mice per group. **** P <0.0001 by 2-way ANOVA with Sidak’s multiple comparisons test, day 10. C) Proportion of CD90+, Lin− cells (Lin− defined as CD3−, CD5−, CD11b−, CD11c−, CD19−, NK1.1−, and FcεR1−) determined to be ILC2s (IL-33R+). Percent ILC2 from sdLN in splenocyte chimera mice with AD-like disease after PMA/iono stimulation positive for D) IL-5 or E) IL-13 staining. F) Schematic of bone marrow chimera experiment. Equal quantities of bone marrow cells from Rag1−/− (CD45.2, CD90.2 – orange) and WT (CD45.2, CD90.1 – blue) C57Bl/6J donor mice were used to reconstitute the immune systems of irradiated recipient WT (CD45.1 – black) C57Bl/6J mice. G) Proportion of donor (CD45.2+) ILC2 defined as in (C) in sdLN by donor source (CD90.1+ – WT, CD90.2− – Rag1−/−). Proportion of Lin− ILCs by donor source positive for H) IL-5 and I) IL-13 following PMA/iono stimulation and cytokine staining. C-E) Data representative of at least 2 independent experiments, 4-5 mice per group. ** P < 0.01, **** P < 0.0001 by two-tailed Welch’s t test. G-H) Data representative of at least 2 independent experiments, 4-5 mice per group. ** P < 0.01 by two-tailed ratio means paired t test. All data represented as mean with standard deviation.

A history of RAG expression marks a population of ILC2s in the sdLN.

A) Schematic of RAG fate mapping in the lymphoid cell compartment using reporter mice expressing Cre-inducible tandem dimer red fluorescent protein (tdRFP) from the Rosa26 locus crossed to mice expressing Cre recombinase from the Rag1 locus. B-F) Histograms of tdRFP signal in CD45+ sdLN cells by cell type for B) CD4+ T cells (B220−, CD3+, CD4+), C) CD8+ T cells (B220−, CD3+, CD8+), D) B cells (MHCII+, B220+), E) NK cells (B220−, CD3−, CD4−, CD8−, CD49b+, NK1.1+), F) ILC2s (B220−, CD3−, CD4−, CD8−, CD49b−, NK1.1−, CD11b−, CD11c−, SiglecF−, CD90+, KLRG1+ or ICOS+ or IL-33R+), F) quantification of tdRFP+ proportion of each cell type.

Multiomic analysis of ILC2s through single nuclei sequencing of the sdLN.

A) Schematic of the gene expression (GEX) assay derived from snRNA-seq data. B) Schematic of the gene activity (GA) assay, representing estimated transcription scores derived from snATAC-seq data using Signac. C) UMAP visualizations of independent analyses of RNA-seq and ATAC-seq data for 2034 sdLN cells after dimensional reduction and clustering combined using weighted nearest neighbor (WNN) analysis in Seurat. Cluster identities are color coded consistently throughout the following panels. Heatmaps of D) top 25 GEX marker genes and E) top 25 GA marker genes identified for each cluster. See Table S1 for full lists of genes. F) Dotplots comparing selected marker genes for each cluster between the GEX and GA assays, with emphasis on known cell type-specific markers. G) Schematic of differentially accessible (DA) chromatin assay, which finds the nearest gene to any peak calculated to be differentially open in a particular cell cluster. See Table S2 for full lists of top 25 DA cluster markers. H) Overlap of top 100 markers for the ILC2 cluster from the GEX, GA, and DA assays. See Table S3 for top 100 DA peaks and distances to nearest genes and Table S4 for full list of top 100 ILC2 markers. I) Selected genes from the ILC2 gene set for each assay individually and for overlaps.

A history of RAG expression imprints transcriptomic and epigenomic modulation of ILC2 gene programs.

A) Schematic of transcriptional RAG fate mapping. Sequenced cells from the RAG fate map mouse (see Figure 3A) transcribe tdRFP only after Cre is expressed from the Rag1 locus. Cells were assigned as either having a history of RAG expression (RAGexp – tomato red) or not (RAGnaïve – dark gray) based on detection of tdRFP transcript in the RNA-seq data. B) Schematic of mapping gene to peak links (GPLs). The LinkPeaks function of Signac (see methods) calculates significant correlations between open chromatin at defined peaks (teal bars) and nearby gene expression. These links represent inferred epigenomic-transcriptomic regulation, or “regulomes” based on the correlated snRNA– and snATAC-sequencing data. After calculating GPLs separately for each population (gray for RAGnaïve and tomato red for RAGexp), GPLs found in only one group, but not the other, can be identified (teal boxes). The difference in GPLs based on RAG experience for any given gene (e.g. Gene X) can be visualized on a bar graph, with the number of GPLs for RAGnaïve (gray – left) and RAGexp (red – right) plotted, with the difference overlaid as a black bar. C) GPLs calculated as in (B) for the multiomic ILC2 gene set identified in Figure 4H and Table S4). All identified GPLs are listed in Table S7, while ILC2 GPLs are listed in Table S8. Genes are sorted from more links identified in the RAGnaïve population at top to more links identified in the RAGexp population at bottom. Select genes are labeled. Full ranked list by difference in GPLs is available in Table S9.

A history of RAG expression broadly influences ILC2 genes at steady state and in AD-like inflammation.

A) Schematic of the process to determine contribution of RAG fate map and disease states to GPLs for subsequent intersection analyses. GPLs were first calculated for all indictated cells, regardless of disease state or fate map. Cells were then split, first by RAG fate map (RAGexp and RAGnaïve), and again by disease state (SS – steady state, AD – AD-like inflammation). GPLs were recalculated for each split sample and matched back to the original set of total GPLs. B) UpSet plot visualizing intersections of peaks identified from ILC2 GPLs for split samples. Each row represents one of the four sets, and each column corresponds to an intersection of one or more sets (see methods). See Table S10 for full list of GPLs for all genes. Table S11 lists total and ILC2 peaks used for intersection analyses in each of the four sets. Columns identifying key intersections are color coded by the corresponding RAG fate map or treatment groups. The blue column indicates the intersection of peaks from RAGnaïve cells and peaks induced by AD-like disease in RAGexp cells. C) Top genes with the most AD-like disease-induced peaks. Peaks from the intersection between RAGnaïve cells and inflamed RAGexp cells were identified in corresponding GPLs, and genes were ranked by number of linked peaks identified. See Table S12 for full list of ranked genes and associated GPLs. Open chromatin in the ILC2 cell cluster split by disease state and then by RAG fate map for the genomic loci of C) Ccr6 and D) Rora.

RAG suppresses the Th2 locus.

A) Coverage plot of the Th2 genomic locus. Open chromatin in the ILC2 cluster for each Rag1 fate-mapped state is shown on top, and corresponding peaks (teal) and gene to peaks links (GPLs) are shown below for the RAGnaïve sample (gray) and the RAGexp sample (tomato red). Only GPLs that fit in the coverage window are shown. B) All GPLs identified in each fate map state for the Th2 locus genes Il4, Il13, Rad50, and Il5. See Table S13 for full list of Th2 GPLs. The number of GPLs for each gene is shown on the left in gray for RAGnaïve and on the right in tomato red for RAGexp. The difference is superimposed in black, and genes are sorted from more GPLs identified in RAGnaïve at top to more links identified in RAGexp at bottom. C) UpSet plot of intersections of peaks identified from Th2 locus GPLs. GPLs were recalculated, this time for samples separately by both RAG fate map status (RAGexp and RAGnaïve) and disease (SS – steady state, AD – AD-like inflammation). Each row represents one of the four sets of peaks, and each column corresponds to an intersection of one or more sets (see methods). See Table S11 for full list of peaks from GPLs for all genes, including Th2 genes, in each of the four sets. Columns identifying key intersections are color coded by the corresponding RAG fate map or disease groups. The blue column indicates the intersection of peaks from RAGnaïve cells and peaks induced in AD-like disease in RAGexp cells. (D) Th2 genes sorted by number of AD-like disease-induced peaks. Peaks induced by AD-like disease were identified in corresponding GPLs, and genes were ranked by frequency of links to induced peaks (representation in identified GPLs). See Table S14 for full list of ranked Th2 locus genes and associated GPLs. E) Open chromatin tracks, split by disease (beige box – steady state; maroon box – AD-like disease) and by RAG fate map (RAGnaïve – gray, RAGexp – red) for Il13.

ILC2 and IL-5/IL-13 gating.

Gating for A) CD45+, CD90+, Lin− cells (Lin− defined as CD3−, CD5−, CD11b−, CD11c−, CD19−, NK1.1−, and FcεR1−), then gating on B) ILC2 (IL-33R+ Lin−) corresponding to Fig. 1C, with subsequent gating of C) IL-5+ and IL-13+ ILC2, corresponding to Fig. 1D-E.

Expansion and activation of ILC2s in RAG2 deficiency compared to littermates.

A) Schematic of steady state analysis of WT B6 (Control) mice or Rag2−/−mice. B) Proportion of CD90+ Lin− cells (Lin-defined as CD3−, CD5−, CD11b−, CD11c−, CD19−, NK1.1−, and FcεR1−) determined to be ILC2s (IL-33R+) in sdLN at steady state from WT or Rag2−/− mice. Percent ILC2 from sdLN at steady state following PMA/iono stimulation positive for C) IL-5 or D) IL-13 staining. Data representative of 2 independent experiments, 2-3 mice per group. * P <0.05, ** P < 0.01 by two-tailed Welch’s t test. All data represented as mean with standard deviation.

Confirmation of splenocyte reconstitution in splenocyte chimera mice.

Proportion of CD45+ splenocytes from splenocyte chimera mice (related to Fig. 2A-E) determined to be A) CD4+ T cells (CD4+, CD8−, CD19−), B) CD8+ T cells (CD4−, CD8+, CD19−), C) B cells (CD4−, CD8−, CD19+), and D) Eosinophils (SiglecF+, CD4−, CD8−).

Donor cell reconstitution and gating in sdLN of WT:Rag1−/− bone marrow chimera mice.

A) Gating of live host (CD45.1+) and donor (CD45.2+) cells and B) quantification in sdLN of WT:Rag1−/− bone marrow chimera mice. ** P < 0.01, **** P < 0.0001, by RM one-way ANOVA test with Geisser-Greenhouse correction. C) Gating, and D) quantification of CD45.2+ Lin− donor cells by genotype (WT – blue and Rag1−/− – orange) in sdLN of WT:Rag1−/−bone marrow chimera mice. * P < 0.05 by ratio means paired t test. All data represented as mean with standard deviation. Related to Fig. 2F-I.

sdLN multiome experiment.

A) Rag1Cre::Rosa26LSL-tdRFP reporter mice were given topical treatments with 2 nmol MC903 dissolved in ethanol vehicle or with ethanol vehicle alone to each ear daily for 7 days. Harvested sdLN processed using Magnetic Activated Cell Sorting (MACS) led to depletion of cells expressing the CD3, CD19, and CD11b lineage markers and remaining cells were further processed in the 10X Multiome pipeline, generating both single cell RNA-sequencing and single cell ATAC-sequencing data for each cell. B) Ear thickness measured daily in the AD-like disease multiome experiment. Data representative of one experiment, with 4 mice per group pooled for sequencing. **** P <0.0001 by 2-way ANOVA with Sidak’s multiple comparisons test, day 7. All data represented as mean with standard deviation.

ILC2 marker genes identified in the differentially accessible open chromatin assay.

Differentially accessible (DA) open chromatin peaks identified for the ILC2 cluster are highlighted in gray and shown next to the closest gene for A) Neuromedin U receptor 1 (Nmur1) and B) IL-5 (Il5). See Table S3 for top 100 DA peaks and distances to nearest genes for the ILC2 cluster.

Dotplots of selected ILC2 marker genes.

Dotplots comparing selected marker genes from the multiomic ILC2 gene set (Figure 4I, Table S4) for each cluster between the GEX and GA assays, with genes highlighted by color corresponding to individual assays or overlap of assays in which they were identified. Rora was not detected in the GA assay.

Gene set enrichment analysis of differentially expressed genes in ILC2s.

A) Volcano plot of differentially expressed genes (DEGs) by RAG fate map for the ILC2 cluster. A ranked list (Table S5) was constructed for all DEGs with log2(fold change) >0.1 for gene set enrichment analysis (GSEA). B) Dotplot of GSEA result calculated using ClusterProfiler and the gene ontology (GO) biological process (BP) database (see methods). Full results in Table S6. C) GSEA plot of the GO BP “positive regulation of immune system process” gene set. RAGnaïve, RAG fate map negative; RAGexp, RAG fate map positive.

Mapping gene to peak links in select ILC2 genes.

Gene to peak links (GPLs) mapped for the RAGnaïve and RAGexp states as depicted in Figure 5B for A) GATA binding protein 3 (Gata3) and B) Nedd4 family interacting protein 1 (Ndfip1). Only GPLs that fit in the coverage window are shown. Select peaks (teal bars) present in one state, but not the other, are highlighted in teal boxes. Full gene names not shown on figure in (A) are *9230102O04Rik and **4930412O13Rik and in (B) #Gm42690.

Gene to peak link analysis by RAG fate map and disease for all detected genes.

A) UpSet plot of overlaps in peaks identified from GPLs of all genes split by both RAG fate map (RAGexp and RAGnaïve) and disease state (SS – steady state, AD – AD-like inflammation). Each row corresponds to one of the four sets, and each column corresponds to an intersection of one or more sets (see methods). See Table S11 for full list of peaks from GPLs for all genes in each set. Columns identifying key intersections are color coded by the corresponding RAG fate map or disease groups. The blue column indicates the intersection of peaks from RAGnaïve cells and peaks induced by AD-like disease in RAGexp cells.