Mechanisms of epigenomic and functional convergence between glucocorticoid- and IL4-driven macrophage programming

Abstract

Macrophages adopt distinct phenotypes in response to environmental cues, with type-2 cytokine interleukin-4 promoting a tissue-repair homeostatic state (M2_IL4). Glucocorticoids (GC), widely used anti-inflammatory therapeutics, reportedly impart a similar phenotype (M2_GC), but how such disparate pathways may functionally converge is unknown. We show using integrative functional genomics that M2_IL4 and M2_GC transcriptomes share a striking overlap mirrored by a shift in chromatin landscape in both common and signal-specific gene subsets. This core homeostatic program is enacted by transcriptional effectors KLF4 and the glucocorticoid receptor, whose genome-wide occupancy and actions are integrated in a stimulus-specific manner by the nuclear receptor cofactor GRIP1. Indeed, many of the M2_IL4:M2_GC-shared transcriptomic changes were GRIP1-dependent. Consistently, GRIP1 loss attenuated phagocytic activity of both populations in vitro and macrophage tissue-repair properties in the murine colitis model in vivo. These findings provide a mechanistic framework for homeostatic macrophage programming by distinct signals, to better inform anti-inflammatory drug design.

Fig. 1. Transcriptomic analysis of the M2_IL4 and M2_GC populations.

a Transcriptomic changes in M0 macrophages polarized with IL4 or GCs corticosterone (Cort) or dexamethasone (Dex) for 24 h were determined by RNAseq (n = 3 biological replicates [mice]). Venn diagrams show DEGs in M2_IL4 and M2_GC relative to M0 (FC ≥ 2; FDR < 0.05); 92 of 133 genes are regulated in both M2_IL4 and M2_Dex in the same direction. b Volcano plots show DEGs in M2_IL4 (left) and M2_Dex (right) relative to M0 (LogFC ≥ 1; FDR < 0.05). Selected upregulated genes are highlighted in orange (M2_IL4) and green (M2_Dex). Downregulated genes are highlighted in red in both populations. Examples of the M2_IL4:M2_Dex shared DEGs are underlined. c RT-qPCR validation of genes upregulated selectively in M2_IL4 (top), M2_GC (middle), or both (bottom). One-tailed unpaired t test; p-values are shown; ns, non-significant (p ≥ 0.05). n = 4 biological replicates (mice), error bars are SEM. d Differentially regulated pathways (unadjusted p < 0.01; see ref. . equations 14-15) identified using QuSAGE and MsigDB canonical pathways (c2.cp.v7.3, Broad Institute) are shown for M2_IL4 (top) and M2_Dex (bottom). Underlined are the shared pathways between M2_IL4 (upregulated, orange; downregulated, red) and M2_Dex (upregulated, green; downregulated, red). Circle size is proportional to the number of genes in the pathway, and color signifies the p-value. e Genes from indicated pathways induced or repressed in M2_IL4, M2_Dex, or both are plotted as LogFC ± SD. Source data are provided as a Source data file.

Fig. 2. M2_IL4 and M2_GC macrophage populations share epigenetic landscape.

a Venn diagram shows the number of differential ATACseq peaks in M2_IL4 and M2_Dex relative to M0 (n = 4 biological replicates, FC ≥ 2; FDR < 0.05). 4913 of 8213 peaks differential in both populations were regulated in the same direction. b Heatmaps and numbers of the induced (red) and suppressed (blue) M2_IL4 and M2_Dex signal-specific and shared ATACseq peaks. c, d Enriched transcription factor binding motifs with associated p-values were identified by HOMER known motif analysis with JASPAR 2022 motif database in signal-specific and shared up- (c) and downregulated (d) ATACseq peaks in M2_IL4 and M2_Dex. Binomial distribution was used for p-value calculation, and no adjustments were made for multiple comparisons. e Polarization-induced increases in the 5’ Tn5 cut site counts within indicated binding motifs in the M2_IL4 and M2_Dex populations. The error band is a 95% confidence interval for LOESS smoothing; span = 0.025. f Venn diagram shows the differential M2_IL4 and M2_Dex H3K27ac ChIPseq peaks relative to those in M0 (n = 3 biological replicates, FC ≥ 2; FDR < 0.05) with 1304 of shared 1667 peaks regulated in the same direction in both populations. g Heatmaps of induced (red) and suppressed (blue) signal-specific and shared filtered H3K27ac peaks (FC ≥ 2; FDR < 0.05) in M2_IL4 and M2_Dex. h Volcano plots show differential H3K27ac ChIPseq peaks for M2_IL4 (left) and M2_Dex (right) relative to those in M0 (LogFC ≥ 1; FDR < 0.05); the location of selected hyperacetylated sites relative to the TSS of the closest gene are shown in orange (M2_IL4) or green (M2_Dex). Hypoacetylated sites are shown in red for each population. Sites overlapping in M2_IL4 and M2_Dex are underlined. i Non-promoter 12,211 H3K27ac peaks after excluding 1650 promoter-proximal peaks (− 300/ + 200 relative to TSS) are shown as a Venn diagram with the number of differential, relative to M0, M2 population-specific, and shared peaks (FC ≥ 2; FDR < 0.05) indicated. j Average H3K27ac ChIPseq signals at M2_IL4 (top), M2_Dex (middle)_, and shared (bottom) differential non-promoter peaks from (I) represented as violin plots. Box-plot insets show median and interquartile range (IQR), and whiskers are extended to 1.5*IQR.

Fig. 3. A tight correlation between epigenomic and transcriptomic macrophage programming in both M2_IL4 and M2_GC.

a Correlation between RNAseq, ATACseq and H3K27ac ChIPseq signals in IL4-specific Chil4, GC-specific Hif3a and shared Klf9 in each population. b Genome-wide changes in up- or downregulated differential ATACseq (left) and H3K27ac ChIPseq (right) peaks located within – 20 K + 20 K window centered on the TSS of RNAseq-differential or -non-differential genes in M2_IL4 and M2_Dex, as indicated. c, d Comparison of TSS window-associated ATACseq (c) or H3K27ac ChIPseq (d) peaks in differential (teal) vs. non-differential (gray) genes in M2_IL4 (left panels) and M2_Dex (right panels), as indicated, stratified by the number of gene-associated peaks. Only peaks with the largest positive (max) or negative (min) changes relative to M0 are shown. A subset of 92 shared up- (red) and downregulated (blue) DEGs identified by RNAseq (Fig. 1) are labeled in each of the 4 panels. e Correlation between normalized H3K27ac ChIPseq and ATACseq signals in the TSS-proximal – 1 K + 1 K windows for differential (teal) and non-differential (gray) genes. A subset of 92 shared upregulated (red) and downregulated (blue) DEGs identified by RNAseq (Fig. 1) are labeled.

Fig. 4. GR, KLF4 and GRIP1 genome-wide binding in differentially polarized macrophage populations.

a Venn diagram shows the numbers and overlap of GR ChIPseq peaks in M2_Dex and M2_Cort macrophages relative to M0 baseline (n = 4 biological replicates [mice]; FC ≥ 2, FDR < 0.05). b Venn diagram shows the numbers and overlap of (left) GRIP1 CUT&RUN peaks across populations (n = 3 biological replicates; FC ≥ 1.5, FDR < 0.05) or GRIP1 peaks in M2_IL4 (middle) or M2_Dex (right) with all KLF4 peaks (n = 2 biological replicates), as indicated. c GR (ChIPseq), KLF4 (CUT&RUN) and GRIP1 (CUT&RUN) read density distribution over Chil4, Hif3a and Klf9 loci in M0, M2_IL4 and M2_Dex. d Average profiles of GR, KLF4, and GRIP1 signal from each macrophage population centered on GRIP1 CUT&RUN M2_IL4- (left), M2_Dex- (middle) specific or invariant ‘static’ (right) peaks (n = 3 biological replicates; FC ≥ 2, FDR < 0.05). e, f Transcription factor binding motif enrichment Z-scores in KLF4 (e) and GRIP1 (f) peak subsets in M2_IL4, M2_Dex, and M0 macrophage populations, as indicated, were determined in chromVAR from the HOMER list of motifs. Motifs with high variability among the subsets (p_adj ≤ 1*10⁻⁶) were plotted, and indicated motif families are highlighted in boxes.

Fig. 5. GRIP1 is required for the establishment of M2_IL4 and M2_GC transcriptomes.

a, b Expression of (a) GRIP1 or (b) IL4- (top) GC- (middle) or shared (bottom) M2 target genes in WT and GRIP1 cKO M0, M2_IL4, M2_Cort, and M2_Dex populations as assessed by RT-qPCR. The average relative expression of each transcript in WT M0 is arbitrarily set to 1. n = 4 biological replicates [mice]; One-tailed unpaired t test; p-values are indicated; ns, non-significant (p ≥ 0.05). Error bars are SEM. c Stacked bar plot shows the number of DEGs (FC ≥ 1.5) in GRIP1 cKO as a fraction of WT M2_IL4, M2_Dex, or shared DEGs relative to M0 (RNAseq, WT: n = 3, cKO: n = 4 biological replicates). d GRIP1 peaks from Fig. 4b were annotated to their closest gene using ChIPseeker, and overlap with GRIP1-independent (n = 1516) or -dependent (n = 432) DEGs from (c) was determined for all unique peaks. A bar graph shows the % (and number) of genes in each subset associated with IL4- and/or Dex-induced peaks. The P-value for overrepresentation was determined by bootstrapping samples from the whole GRIP1 peak atlas to determine how often a set of N peaks would contain the observed overlap with each gene list by chance. e Volcano plot shows DEGs in GRIP1 cKO vs. WT in M2_IL4 (left) and M2_Dex (right) relative to M0 of each genotype. (FC ≥ 1.5, unadjusted p-value < 0.05). Selected inflammation-related genes expressed at higher levels in the cKO M2_IL4 and M2_Dex relative to WT are highlighted in red. Key M2 target genes downregulated in the cKO are highlighted in both M2_IL4 (orange) and M2_Dex (green). Examples of the shared M2_IL4:M2_Dex GRIP1 target genes are underlined. f Differentially regulated pathways (unadjusted p < 0.01 defined as in Fig. 1d) identified using QuSAGE with MsigDB canonical pathways subset (Broad Institute) are shown for GRIP1 cKO vs. WT in M2_IL4 (left) and M2_Dex (right). Select pathways are labeled (upregulated, red; downregulated, orange for IL4 and green for Dex); M2_IL4:M2_Dex-shared pathways are underlined. Circle size is proportional to the number of genes in the pathway, and color signifies the p-value. Source data are provided as a Source data file.

Fig. 6. Macrophage GRIP1 contributes to phagocytic activity of M2_IL4 and M2_GC in vitro and tissue healing in vivo.

a Relative fluorescence intensity of WT and GRIP1-cKO M2_IL4, M2_Cort, and M2_Dex relative to that in WT M0 (set to 100%) after exposure to pHrodo Green S. aureus BioParticles (see “Methods”). Shown are mean ± SEM (n = 3). b Histopathological changes assessed by H&E staining of colons from WT and cKO mice 2 wks after initiating DSS-induced colitis. Grades 1, 2, and 3 are defined in Methods. Scale bar = 20 μm. Plotted is the total area representing grade 3 for colon specimens of WT and GRIP1-cKO. Shown are mean ± SEM (n = 4). c The expression of indicated genes in total RNA from colons of WT and cKO mice was measured by RT-qPCR. The expression of each gene was normalized to β-actin, and then the fold difference to baseline, non-treated (calibrant) value was calculated for WT or cKO. Shown are mean ± SEM (WT n = 7 (Il23) 10 (Mrc1), 11 (Cd163), 12 (Il10, Tnf), 13 (Klf4, Arg1, Il1b), 14 (Pparg) or 15 (Grip1, Klf9); cKO n = 5 (Il23), 6 (Mrc1, Il10), 7 (Pparg, Arg1), 8 (the rest)). d, e Cells were isolated from the large intestine lamina propria at day 11, and single-cell suspensions incubated in FACS buffer with conjugated antibodies to CD45-PE, CD11b-PE-Cy7, Ly6C-PerCP-Cy5.5, Ly6G-APC, and CD206-AF700. Dead cells were excluded using DAPI. Cell sorting was performed on Symphony S6 (BD Biosciences; see Supplementary Fig. 6d for FACS strategy), and (d) data analyzed with FlowJo v10.10 software. Ly6C⁺, CD206⁺, and Ly6C⁺CD206⁺ cells were pooled, total RNA was isolated, and expression of indicated genes was assessed by RT-qPCR (e). Shown are mean ± SEM (n = 4). P-values are indicated; ns, non-significant (p ≥ 0.05). In a–e, n = the number of independent biological replicates (mice), and significance is determined by a one-tailed unpaired t test. Source data are provided as a Source data file.