IL-10 targets IRF transcription factors to suppress IFN and inflammatory response genes by epigenetic mechanisms

Abstract

Interleukin-10 (IL-10) is pivotal in suppressing innate immune activation, in large part by suppressing induction of inflammatory genes. Despite decades of research, the molecular mechanisms underlying this inhibition have not been resolved. Here we utilized an integrated epigenomic analysis to investigate IL-10-mediated suppression of LPS and TNF responses in primary human monocytes. Instead of inhibiting core TLR4-activated pathways such as NF-κB, MAPK–AP-1 and TBK1–IRF3 signaling, IL-10 targeted IRF transcription factor activity and DNA binding, particularly IRF5 and an IRF1-mediated amplification loop. This resulted in suppression of inflammatory NF-κB target genes and near-complete suppression of interferon-stimulated genes. Mechanisms of gene inhibition included downregulation of chromatin accessibility, de novo enhancer formation and IRF1-associated H3K27ac activating histone marks. These results provide a mechanism by which IL-10 suppresses inflammatory NF-κB target genes, highlight the role of IRF1 in inflammatory gene expression and describe the suppression of IFN responses by epigenetic mechanisms.

IL-10 is a key suppressor of innate immune activation and inflammatory responses. A well-recognized mechanism of IL-10 action is the suppression of nuclear factor-κB (NF-κB) target genes, such as TNF, IL6, IL1B and IL12B, in monocytes, dendritic cells and macrophages^1,2. Loss of IL-10 results in hyperactivation of innate and inflammatory responses to environmental challenges, and the spontaneous development of inflammatory conditions, most notably colitis³. Myeloid-cell-specific deletion of the IL-10 receptor or of the IL-10-activated STAT3 transcription factor has illustrated the important role of IL-10 signaling in monocytes and macrophages in preventing excessive inflammation^4,5. Loss-of-function mutations in STAT3 or the IL-10 receptor result in inflammatory bowel diseases in humans^3–6.

Despite extensive efforts, the mechanisms by which IL-10–STAT3 signaling suppresses inflammatory gene induction remain unclear⁷. STAT3 does not bind to or directly suppress the transcription of inflammatory genes⁸, and a large body of data does not support the early hypothesis that IL-10–STAT3 signaling suppresses Toll-like receptor (TLR)-induced activation of NF-κB or MAPK signaling^1,2,7,9,10. Alternatively, various IL-10-induced transcriptional repressors have been identified, but they do not have a non-redundant role in the suppression of core inflammatory genes^1,11–13. IL-10-mediated regulation of cell metabolism¹⁴ underlies suppression of inflammasome activation, but the mechanisms by which IL-10 suppresses inflammatory gene transcription remain unclear. There is limited evidence that IL-10 suppresses inflammatory genes through epigenetic mechanisms^8,15.

The inhibitory effects of IL-10 on inflammatory gene induction are typically studied in mouse macrophages stimulated with prototypical activators, such as the TLR4 ligand lipopolysaccharide (LPS). TLR4 activates proinflammatory genes directly through NF-κB and MAPK signaling, and interferon-stimulated genes (ISGs) indirectly through inducing production of autocrine-acting interferon-β (IFN-β)¹⁶. TLR4 signaling also induces expression of, or post-translationally activates, various IRFs¹⁷. Of these, activated IRF3 drives IFNB1 (also called IFNB) expression, and IRF5 (ref. 18) contributes to and amplifies inflammatory gene expression, at least in part through cooperation with NF-κB^17,19. IRF1 is induced by TLR and IFN signaling and is thought to mainly augment the expression of ISGs²⁰. However, in the paper that first reported its molecular cloning, IRF1 was reported to induce IFNB expression²¹, and it also binds to inflammatory gene loci²², suggesting a potentially broader function beyond just ISG induction. The regulation of IRFs by IL-10 has not been investigated.

Here, we used an integrated epigenomic approach combining RNA sequencing (RNA-seq), assay for transposase accessible chromatin (ATAC-seq) and CUT&RUN to investigate how IL-10 regulates LPS- and tumor necrosis factor (TNF)-induced responses in primary human monocytes. We found that IRF1 and IRF5 had a key role in the induction of IFNB, ISGs and inflammatory genes, and that IL-10 suppressed TLR4- and TNFR-induced gene expression by suppressing expression and activity of IRFs, rather than inhibiting NF-κB, MAPK–AP-1 or TBK1–IRF3 signaling pathways. IL-10 used epigenetic mechanisms involving suppression of chromatin accessibility, enhancer formation and IRF1-associated histone modification. Our study highlights the inhibition of IFN responses as a biological activity of IL-10 and describes a mechanism of IL-10-mediated suppression of TLR4- and TNFR-induced gene expression.

Results

IL-10 suppresses IFN and inflammatory responses

CD14⁺ human monocytes were isolated from peripheral blood buffy coats (see Methods), cultured for 18 h with IL-10 (100 ng ml^–1) and then stimulated for 3 h with LPS (Extended Data Fig. 1a). Real-time quantitative PCR (RT–qPCR) analysis showed that IL-10 strongly and significantly inhibited LPS-induced expression of not only the canonical inflammatory genes IL6 and TNF, but also the ISGs CXCL10 and ISG15, in monocytes from seven independent donors (Fig. 1a). Bulk RNA-seq and Gene Set Enrichment Analysis (GSEA) revealed that IL-10 more consistently suppressed the LPS-induced Hallmark interferon response gene set than the Hallmark inflammatory gene set (Fig. 1b and Extended Data Fig. 1b,c). IL-10 also suppressed LPS-induced gene expression in monocyte-derived macrophages (Extended Data Fig. 1d). Gene clustering based on expression patterns and pathway analysis revealed that gene cluster 2 (C2), whose induction was strongly suppressed by IL-10, was most significantly enriched in IFN response genes (Fig. 1c,d). By contrast, the LPS-induced C3 genes, which were only moderately suppressed by IL-10 (Fig. 1c), were most significantly enriched in inflammatory NF-κB pathways (Fig. 1d). The remaining clusters contained IL-10-inducible genes (C1 and C6) or LPS-inducible genes that were not suppressed by IL-10 (C4 and C5). Additionally, 92.4% of the LPS-induced genes that IL-10 suppressed by at least 40% (436 out of 1,745 genes, Extended Data Fig. 1e,f) were found in C2 or C3 (Extended Data Fig. 1g), suggesting that these clusters represented the core of the IL-10-suppressed genes. The core set of the IL-10-suppressed genes (those suppressed by 40% or more) showed the most significant enrichment of IFN pathways, with less significant enrichment of inflammatory and NF-κB pathways (Fig. 1e). Heatmaps depicting expression of Hallmark pathway inflammatory and IFN response genes showed that ISGs were more strongly suppressed by IL-10 than were inflammatory genes (Fig. 1f,g). IL-10 did not inhibit the LPS-mediated activation of NF-κB signaling (Extended Data Fig. 1h), similar to previous reports^9,11.

Fig. 1 |. IL-10 preferentially suppresses IFN relative to inflammatory responses in LPS- and TNF-stimulated monocytes.

a, qPCR assay showing the expression levels of IL6, TNF, CXCL10 and ISG15 mRNA, normalized to GAPDH mRNA, in CD14⁺ human monocytes cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with LPS (10 ng ml^–1) for 3 h. Dots correspond to independent donors (n = 7). b, GSEA of RNA-seq data performed on differentially expressed genes (DEGs), ranked by log₂(counts per million), in human CD14⁺ monocytes treated with IL-10 and LPS as in a. Hallmark gene sets from the Broad Institute and LPS-regulated IFN and inflammatory pathways are displayed. NES, normalized expression score; ES, expression score; adj. P, adjusted P value. c, k-means clustering analysis (k = 6) of differentially upregulated genes in any pairwise comparison of CD14⁺ monocytes treated with IL-10 and LPS as in a, relative to resting unstimulated monocytes. d,e, Hallmark pathway enrichment analyses on clusters as in c (d) or on LPS-induced genes suppressed by IL-10 (e). f,g, Heatmaps of LPS-induced genes suppressed by IL-10 (n = 436), clustered by inflammatory and IFN response genes (f) and representative inflammatory and IFN response genes suppressed by IL-10 (g). h, Hallmark pathway enrichment analysis performed on TNF-induced genes suppressed by IL-10 in CD14⁺ monocytes cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with TNF (20 ng ml^–1) for 6 h. i, Heatmap of TNF-induced genes suppressed by IL-10 as in h (n = 268), clustered by inflammatory and IFN response genes. j, qPCR of CXCL10 and TNF mRNA normalized relative to GAPDH mRNA in CD14⁺ monocytes stimulated with IL-10 and TNF as in h (n = 4 independent donors). In b–i, data are from three independent donors. In c, f, g, and i, z-score-normalized data were used to make heatmaps. In a and j, data are depicted as mean ± s.e.m. ***P < 0.0005; ****P < 0.0001 by one-way analysis of variance (ANOVA) with Tukey’s multiple-comparisons test. Created with BioRender.com.

TNF induces the expression of inflammatory and IFN response genes through distinct upstream signaling pathways^23–25. RNA-seq on primary human monocytes using the same experimental design (Extended Data Fig. 1a), but with stimulation for 6 h with TNF (20 ng ml^–1) instead of LPS, indicated that IL-10 strongly suppressed the TNF-induced IFN genes, but only moderately suppressed TNF-induced inflammatory NF-κB target genes (Fig. 1h–j and Extended Data Fig. 1i–m). Collectively, these results show that IL-10 strongly inhibits IFN response genes in human monocytes.

IL-10 suppresses H3K27ac near IRF1-binding motifs

To test whether IL-10 inhibited gene expression through an epigenetic mechanism, we performed CUT&RUN to obtain the genome-wide profile of histone H3 acetylated at K27 (H3K27ac), which marks active gene-regulatory elements (enhancers and promoters), in monocytes treated with IL-10 and LPS (IL-10 + LPS), using the same experimental design as that for RNA-seq (Extended Data Fig. 1a). LPS significantly increased H3K27ac levels at 5,749 genomic regions (false discovery rate (FDR) ≤ 0.05 and fold change of ≥2; Extended Data Fig. 2a,b). Of these 5,749 H3K27ac peaks, 3,450 (60%) were not called as significant in IL-10 + LPS-treated monocytes (Fig. 2a). We classified these IL-10-suppressed peaks as group 1. By contrast, 2,299 of the 5,749 LPS-inducible H3K27ac peaks were resistant to IL-10 inhibition and were classified as group 2. Quantitation of peak intensity using deepTools and visualization on heat maps (Fig. 2b) and violin plots (Fig. 2c) showed that IL-10 suppressed group 1 H3K27ac peaks very significantly and strongly, bringing them close to baseline levels (P < 0.0001, effect size (d) = 1.27; see Methods). IL-10 also significantly suppressed H3K27ac at group 2 peaks, but this suppression was partial, and the mean peak intensity remained elevated relative to the baseline observed in unstimulated monocytes (Fig. 2b,c, d = 0.78). LPS-induced peaks were distributed in intergenic, intronic and promoter regions (Fig. 2d), suggesting that IL-10 targeted both enhancer and promoter regulatory elements to suppress associated genes. IL-10 differentially suppressed LPS-induced H3K27ac at the group 1 IFIT1 locus and group 2 IL6 locus (Fig. 2e, representative gene tracks). In addition, CUT&RUN analysis of H3 trimethylated at K4 (H3K4me3), a positive promoter mark that enhances transcription, in monocytes showed that genes with decreased transcription and decreased H3K27ac levels also had diminished levels of H3K4me3, whereas H3K4me3 levels remained unchanged at gene loci at which H3K27ac was not suppressed (Fig. 2e,f). Genes associated with group 1 H3K27ac peaks were strongly enriched in Hallmark IFN pathways (Extended Data Fig. 2c), and a large proportion of IL-10-suppressed genes were associated with group 1 IL-10-suppressed peaks (Extended Data Fig. 2d). This indicates that IL-10 suppressed the LPS-induced activation of chromatin and suggests that this suppression was associated with decreased expression of LPS-induced genes. To gain insight into whether group 1 and group 2 peaks could be distinguished on the basis of their regulation by different transcription factors, we used HOMER to identify transcription-factor-binding sequences enriched under H3K27ac peaks. Both group 1 and group 2 peaks were enriched for AP-1 and NF-κB motifs; however, only group 1 peaks, which were strongly suppressed by IL-10, showed enrichment of IRF1 motifs (Fig. 2g and Extended Data Fig. 2e). These results suggest that genes regulated by IRFs could be more susceptible to inhibition by IL-10 than genes solely induced by NF-κB and AP-1.

Fig. 2 |. IL-10 suppresses H3K27ac at genomic regions enriched in IRF1-binding motifs.

a, Venn diagram of CUT&RUN data, showing the numbers of differentially upregulated H3K27ac peaks in CD14⁺ human monocytes from two independent donors. Cells were cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with LPS (10 ng ml^–1) for 3 h (group 1, LPS unique, n = 3,450; group 2, common, n = 2,299; group 3, IL-10 + LPS unique, n = 515). b, Heatmap of CUT&RUN data showing the normalized signal density of H3K27ac binding in CD14⁺ monocytes treated with IL-10 and LPS as in a. The results show reads per kilobase per million mapped reads (RPKM) values within a range of ±3.0 kb around peak centers. c, Violin plots showing normalized average signal density of H3K27ac binding in CD14⁺ monocytes treated with IL-10 and LPS as in a. Data are plotted as log₂(RPKM + 1) counts of H3K27ac reads (n = 3,450 for group 1; 2,299 for group 2; and 515 for group 3). In the embedded box plots, the midline represents the median, the box spans the interquartile range (IQR) from the 25th (Q1) to the 75th (Q3) percentile, the whiskers extend to the minimum and maximum values within 1.5 × IQR, and any points beyond this range are plotted as outliers (black dots). *P < 0.05; ****P < 0.0001 by two-way ANOVA with Tukey’s multiple-comparisons test. d was calculated using Cohen’s d method. d, Feature distribution plot showing localization of H3K27ac peak genomic coordinates in CD14⁺ monocytes treated with IL-10 and LPS as in a. UTR, untranslatable region. e,f, Representative Interactive Genome Viewer (IGV) gene tracks of CUT&RUN data showing H3K27ac (e) or H3K4me3 (f) binding at IFIT and IL6 loci in CD14⁺ monocytes treated with IL-10 and LPS as in a. H3K4Me3 CUT&RUN data were obtained using monocytes from two independent donors. g, De novo motif analysis results using HOMER showing enriched motifs under H3K27ac peaks in CD14⁺ monocytes treated with IL-10 and LPS as in a. Created with BioRender.com.

IL-10 decreases chromatin accessibility and IRF activity

We next used ATAC-seq to test whether inhibition of H3K27ac by IL-10 was associated with reduced chromatin accessibility in IL-10 + LPS-treated monocytes, as described above (Extended Data Fig. 1a). IL-10 suppressed 1,112 of the 8,813 (12.6%) LPS-induced ATAC-seq peaks (Fig. 3a–d), whereas 60% of LPS-induced H3K27ac peaks were suppressed. IL-10 suppressed LPS-induced ATAC-seq peaks at the IFIT, but not the IL1B, locus (Fig. 3e). Pathway enrichment analysis of genes linked to LPS-induced open chromatin regions showed that genes associated with peaks closed in monocytes treated with LPS and IL-10, compared with those treated with just LPS, showed more significant enrichment in IFN pathways. By contrast, genes associated with ATAC-seq peaks that were not suppressed by IL-10 were comparably enriched in NF-κB and inflammatory pathway genes (Extended Data Fig. 3a). LPS-induced ATAC-seq peaks that were suppressed by IL-10 showed a similar reduction of H3K27ac levels (Extended Data Fig. 3b), indicating that a subset of genomic elements targeted by IL-10 showed a decrease in both activating histone marks and chromatin accessibility. Chromatin regions that lost accessibility showed specific enrichment of ISRE or IRF motifs (Fig. 3f), which are associated with canonical ISGs. Accordingly, IL-10-suppressed chromatin regions were associated with canonical ISGs, such as IFIT1 and ISG15 (Fig. 3e,g). Thus, IL-10 preferentially suppressed chromatin accessibility at ISG loci, and this was associated with decreased ISG expression.

Fig. 3 |. IL-10 decreases chromatin accessibility and preferentially suppresses IRF activity.

a, Principal component (PC) analysis plot of ATAC-seq data in CD14⁺ human monocytes from four independent donors. Cells were cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with LPS (10 ng ml^–1) for 3 h. b, Venn diagram showing the numbers of ATAC-seq peaks in CD14⁺ monocytes treated with IL-10 and LPS as in a. c, Heatmap showing normalized signal density of ATACseq peaks in CD14⁺ monocytes treated with IL-10 and LPS as in a. The results are presented as RPKM values within a range of ± 3.0 kb around peak centers. d, Violin plots showing normalized average signal density of the ATAC-seq peaks in c. Data are plotted as log₂(RPKM + 1) counts of ATAC-seq reads (n = 7,701 for non-suppressed ATAC peaks, and 1,112 for suppressed ATAC-seq peaks). In the embedded box plots, the midline represents the median, the box spans the interquartile range (IQR) from the 25th (Q1) to the 75th (Q3) percentile, the whiskers extend to the minimum and maximum values within 1.5 × IQR, and any points beyond this range are plotted as outliers (black dots). ****P < 0.0001 by two-way ANOVA with Tukey’s multiple-comparisons test. n.s., not significant. e, Representative IGV gene tracks showing ATAC-seq peaks at the IL1B and IFIT loci in CD14⁺ monocytes treated with IL-10 and LPS as in a. f, De novo motif analysis using HOMER of LPS-induced ATAC-seq peaks that were suppressed (left) or were not suppressed (right) by IL-10 in CD14⁺ monocytes treated with IL-10 and LPS as in a. g, Heatmap depicting expression of LPS-induced ISGs associated with ATAC-seq peaks that were suppressed by IL-10 in CD14⁺ monocytes treated with IL-10 and LPS as in a. h, Volcano plot of differential transcription-factor-binding analysis of ATAC-seq peaks using TOBIAS in CD14⁺ monocytes treated with IL-10 and LPS as in a. i, Heatmap of the differential transcription factor activity scores derived from ChromVAR analysis of ATACseq peaks in CD14⁺ monocytes treated with IL-10 and LPS as in a. ATAC-seq data (a–i) were obtained using CD14⁺ monocytes from four independent donors. Created with BioRender.com.

We next used TOBIAS to measure transcription-factor occupancy at motifs within ATAC-seq peaks in monocytes²⁶. LPS alone induced occupancy at NF-κB and AP-1 motifs, and less so at IRF motifs, whereas IL-10 alone enhanced occupancy of STAT3 motifs (Extended Data Fig. 3c). In monocytes stimulated with IL-10 + LPS, the occupancy of NF-κB and AP-1 motifs was mostly preserved and IRF motif occupancy was diminished, compared with that in monocytes stimulated with LPS alone (Extended Data Fig. 3c). A direct comparison of occupancy in LPS-treated monocytes versus IL-10 + LPS-treated monocytes indicated that IL-10 most strongly and significantly suppressed IRF binding (Fig. 3h). Motif footprinting analysis in monocytes stimulated with nothing, IL-10, TNF or IL-10 + TNF indicated that transcription factor occupancy at NF-κB and AP-1 motifs was preserved, whereas occupancy of IRF motifs was distinctly suppressed in cells treated with IL-10 + TNF compared with those treated with TNF alone (Extended Data Fig. 3d,e). A complementary computational analysis using ChromVAR, which computationally generates a transcription-factor activity score, indicated that IL-10 predominantly suppressed IRF activity, whereas the LPS-induced NF-κB and AP-1 activity was mostly preserved in monocytes treated with IL-10 + LPS compared with those treated with only LPS (Fig. 3i). The lack of suppression of LPS + TNF-induced NF-κB and AP-1 activity by IL-10 is in agreement with reports that IL-10 does not suppress NF-κB or MAPK signaling⁷. In addition, ChromVAR indicated that IL-10 suppressed the activity of STAT1–STAT2 as part of ISGF3 (Fig. 3i), which is induced by autocrine signaling of IFN-β (ref. 16). Thus, IL-10 suppressed the LPS-induced transcriptional response mainly by targeting IRF motif occupancy and also the TLR4-induced IFN-β-mediated autocrine loop.

IL-10 suppresses LPS-induced IRF1 binding and H3K27ac

Because IRF1 is induced by LPS and activates the transcription of ISGs and inflammatory genes¹⁷, we next investigated the genomic profile of IRF1 binding in monocytes stimulated with IL-10 and/or LPS, as described above (Extended Data Fig. 1a). In agreement with low-level tonic IFN signaling in myeloid cells²⁷, a small number of IRF1 peaks were detected in unstimulated monocytes (Extended Data Fig. 4a and Fig. 4a,b). IRF1 binding peaks were greatly induced in number and amplitude in LPS-stimulated monocytes relative to control unstimulated monocytes (Fig. 4a,b and Extended Data Fig. 4a,b). IL-10 strongly and significantly suppressed LPS-induced IRF1 binding across the genome: 3,931 of the 4,395 LPS-induced peaks (89.5%) were lost in monocytes treated with both LPS and IL-10 relative to monocytes treated with only LPS (Fig. 4a,b). IL-10 did not suppress basal IRF1 peaks that were detected in resting monocytes (Fig. 4b). Pathway enrichment analysis of genes associated with LPS-induced IRF1 binding showed significant enrichment of IFN pathways and inflammatory response pathways (Extended Data Fig. 4c). Motif analysis showed that LPS-induced IRF1-binding peaks that were suppressed by IL-10 were enriched for IRF, AP-1 and NF-κB motifs (Fig. 4c). The IRF1-binding peaks that were partially preserved in LPS- and IL-10-treated monocytes were enriched for IRF1 and AP-1, but not NF-κB, motifs (Fig. 4c), suggesting that genomic elements that bind both IRF1 and NF-κB were preferentially suppressed.

Fig. 4 |. IL-10 broadly suppresses LPS-induced IRF1 binding and associated histone acetylation.

a, UpSet plot of IRF1 CUT&RUN data showing the numbers of IRF1 binding peaks (log₂(FC) ≥ 1 and FDR ≤ 0.05) in CD14⁺ human monocytes from three independent donors. Cells were cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with LPS (10 ng ml^–1) for 3 h. b, Heatmap showing the normalized signal density of LPS-induced IRF1 peaks (top) and violin plot showing normalized average signal density of IRF1 peaks (bottom, n = 4,395) in CD14⁺ monocytes treated with IL-10 and LPS as in a. c, De novo motif analysis using HOMER, showing motifs enriched under IRF1-binding peaks in CD14⁺ monocytes treated with IL-10 and LPS as in a. d, Heatmap showing ATAC-seq normalized signal density surrounding LPS-induced IRF1 peaks (top), and violin plot showing ATAC-seq normalized average signal density (bottom, n = 4,395) under IRF1 peaks in CD14⁺ monocytes treated with IL-10 and LPS as in a. e, Heatmap showing H3K27ac normalized signal density surrounding LPS-induced IRF1 peaks (top), and violin plot showing H3K27ac normalized average signal density (bottom, n = 4,395) in CD14⁺ monocytes treated with IL-10 and LPS as in a. f, Representative IGV gene tracks showing IRF1 binding, H3K4me3 and H3K27ac peaks and ATAC-seq peaks at the CXCL10 locus in CD14⁺ monocytes treated with IL-10 and LPS as in a. In b, d and e, heatmaps were generated using RPKM values within a range of ±3.0 kb around peak centers, and violin plot data are plotted as log₂(RPKM + 1) counts. In the embedded box plots, the midline represents the median, the box spans the interquartile range (IQR) from the 25th (Q1) to the 75th (Q3) percentile, the whiskers extend to the minimum and maximum values within 1.5 × IQR, and any points beyond this range are plotted as outliers (black dots). ****P < 0.0001 by two-way ANOVA with Tukey’s multiple-comparisons test. d was calculated using Cohen’s d method. IRF1 CUT&RUN analysis (a–f) was done using monocytes from three independent donors. Created with BioRender.com.

We next investigated whether LPS-induced IRF1 binding was associated with increased chromatin accessibility, as measured by ATAC-seq, or increased H3K27ac levels, a measure of chromatin activation. Chromatin regions bound by IRF1 were already accessible in resting monocytes, and accessibility changed minimally in monocytes stimulated with IL-10 and/or LPS (Fig. 4d). By contrast, increased H3K27ac deposition at IRF1 binding peaks was observed in LPS-treated monocytes and was strongly decreased in IL-10 + LPS-treated monocytes, in which IL-10 suppressed IRF1 binding (Fig. 4e). IL-10 suppressed IRF1 binding and H3K4me3 and H3K27ac deposition while maintaining open chromatin at the ART3, CXCL10, CXCL11 and OAS3 loci (Fig. 4f and Extended Data Fig. 4d, representative gene tracks). These results suggest that IRF1 does not regulate chromatin accessibility, but instead is associated with activation of chromatin through histone acetylation, and that the latter is suppressed by IL-10.

IL-10 induces limited suppressive epigenomic changes

Next, we tested whether IL-10 reprogrammed the monocyte epigenome such that a subsequent LPS-induced signal, even if intact, would encounter a closed and suppressive chromatin environment that would prevent transcription activation. RNA-seq analysis showed that IL-10 target genes were induced in IL-10-treated monocytes compared with unstimulated monocytes (Fig. 5a). Pathway analysis of genes suppressed in IL-10-treated monocytes relative to unstimulated monocytes showed that the most significant suppression occurred in IFN response genes (Fig. 5b), suggesting that IL-10 suppressed tonic IFN signaling that maintains basal expression of STAT1 and select ISGs²⁷. IL-10 suppressed basal STAT1 expression, but not basal IRF1 expression (Fig. 5a). IL-10 had a minimal effect on the low basal expression of inflammatory genes and minimally suppressed basal H3K27ac (Fig. 5c). IL-10 suppressed basal chromatin accessibility at only 477 of the 4,811 ATAC-seq peaks (Fig. 5d). De novo motif enrichment analysis under the IL-10-suppressed ATAC-seq peaks showed enrichment of IRF motifs (Fig. 5e), whereas TOBIAS footprinting analysis showed decreased occupancy of STAT1 and IRF9 sites (Extended Data Fig. 3c). IL-10-induced ATAC-seq and H3K27ac peaks were enriched for AP-1 and STAT3 motifs (Fig. 5e,f). The ATAC-seq peaks suppressed by IL-10 in resting monocytes were completely distinct from those suppressed in LPS-stimulated monocytes (Fig. 5g). Collectively, these results suggest that IL-10 interrupted tonic IFN signaling but did not broadly reprogram the epigenome in resting monocytes to induce resistance of gene transcription to LPS stimulation.

Fig. 5 |. IL-10 alone induces limited epigenomic reprogramming.

a, Volcano plot of RNA-seq data showing DEGs in CD14⁺ human monocytes treated with IL-10 (100 ng ml^–1) for 18 h. Rest, resting unstimulated control CD14⁺ monocytes (n = 3 independent donors). b, Hallmark pathway enrichment analysis of genes suppressed by IL-10 in CD14⁺ monocytes treated with IL-10 as in a. c, Volcano plot of H3K27ac CUT&RUN peaks in CD14⁺ monocytes (n = 2 independent donors). d, Volcano plot of ATAC-seq peaks in CD14⁺ monocytes treated with IL-10 (100 ng ml^–1) for 18 h (n = 4 independent donors). e,f, HOMER de novo motif analysis showing motifs enriched under ATAC-seq peaks suppressed (top) or induced (bottom) by IL-10 (e), or under H327ac-binding peaks induced by IL-10 (f) in CD14⁺ monocytes treated with IL-10 for 18 h. e, n = 4 donors; f, n = 2 donors. g, UpSet plot showing numbers of ATAC-seq peaks in CD14⁺ monocytes from four independent donors cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with LPS (10 ng ml^–1) for 3 h. Created with BioRender.com.

IL-10 suppresses IRF1 and IRF5, which mediate IFNB induction

We next investigated the mechanisms by which IRF1 binding was suppressed by IL-10 in monocytes stimulated with IL-10 and/or LPS, as described above (Extended Data Fig. 1a). IL-10 strongly suppressed the LPS-mediated induction of IRF1 at both the mRNA and protein levels (Fig. 6a–c). Inhibition of IRF1 protein expression became apparent 2 h post-LPS stimulation in IL-10 + LPS-stimulated relative to LPS-stimulated monocytes (Extended Data Fig. 5a), in accordance with a requirement for protein synthesis for IL-10 inhibitory activity to be initiated¹. IRF1 is induced directly by TLR signaling or indirectly through autocrine IFN-β signaling. The Jak inhibitor baricitinib almost completely suppressed the induction of IRF1 by exogenous IFN-β, but only weakly and partially affected induction of IRF1 by LPS (37%) (Fig. 6d), indicating that autocrine IFN-β signaling was only partially responsible for the TLR4-mediated induction of IRF1 and that IL-10 worked at least in part by a distinct and complementary mechanism. We found that IL-10-mediated suppression of IRF1- and TLR4-induced genes, such as IL6 and TNF, was maintained even after IL-10 was washed out and cells were cultured for an additional 24 h (Extended Data Fig. 5b), consistent with epigenetic regulation of gene expression. Accordingly, IL-10 suppressed H3K27ac and H3K4me3 at the IRF1 locus (Fig. 6e), and this was associated with recruitment of STAT3 (Extended Data Fig. 5c). The results suggested that IL-10 inhibited IRF1 through a combined suppression of its induction through IFNB autocrine signaling and through epigenetic mechanisms that target the IRF1 locus.

Fig. 6 |. IL-10 suppresses expression of IRF1 and IRF5 which mediate IFNB induction in human monocytes.

a, qPCR assay showing expression of IRF1 mRNA normalized relative to GAPDH mRNA in CD14⁺ human monocytes cultured with IL-10 (100 ng ml^–1) for 18 h and then stimulated with LPS (10 ng ml^–1) for 3 h (n = 7 independent donors; same donors as in Fig. 1a). Data are depicted as mean ± s.e.m. ****P < 0.0001 by one-way ANOVA with Tukey’s multiple-comparisons test. b, Flow cytometry assay showing IRF1 protein expression in CD14⁺ monocytes treated with IL-10 and LPS as in a. Representative histogram plot (left) and cumulative MFI data (right) from seven independent donors. c, Immunoblot of IRF1, p38, pSTAT1 and STAT1 in whole cell lysates of monocytes treated with IL-10 and LPS as in a (n = 3 independent donors, p38 used as loading control). d, qPCR assay showing expression of IRF1 mRNA normalized relative to GAPDH mRNA in CD14⁺ monocytes pre-treated with baricitinib (1 μM) for 30 min before stimulation with IFN-β (5 ng ml^–1) or LPS (10 ng ml^–1) for 3 h (n = 3 independent donors). e, IGV gene tracks of H3K27ac (top) and H3K4me3 (bottom) peaks at IRF1 locus in CD14⁺ monocytes stimulated with IL-10 and LPS as in a. f, qPCR assay showing expression of IFNB mRNA normalized relative to GAPDH mRNA in CD14⁺ monocytes stimulated with IL-10 and LPS as in a (n = 7 independent donors; same donors as in Fig. 1b). g, Immunoblot of phosphorylated TBK1 and TBK1 using whole-cell lysates of CD14⁺ monocytes stimulated with IL-10 and LPS as in a (representative blot from one out of three independent donors). h, IGV gene tracks of LPS-induced IRF1 binding at IFNB enhancer (n = 3). i, qPCR assay showing expression of IFNB mRNA normalized relative to GAPDH mRNA in CD14⁺ monocytes in which IRF1, IRF5, IRF1 + IRF5 or HPRT (control) genes had been edited using CRISPR–Cas9. Monocytes were stimulated with LPS for 3 h (n = 9 independent donors and data are depicted as mean ± s.e.m.). Gene expression for each donor is plotted relative to the resting condition, which is set at 100. ****P < 0.0001 by Mann–Whitney two-sided test. For a, b, d, and f, each dot represents an independent donor and data are depicted as mean ± s.e.m. *P < 0.05; **P < 0.001; ***P ≤ 0.0005; ****P < 0.0001 by one-way ANOVA with Tukey’s multiple-comparisons test. Created with BioRender.com.

IL-10 significantly suppressed LPS-induced expression of IFNB (Fig. 6f and Extended Data Fig. 5d) but had minimal effects on ISG induction by exogenous IFN-β (Extended Data Fig. 5e), suggesting the IL-10-dependent repression of ISGs was mediated through the inhibition of IFN-β production. Induction of IFN-β by TNF is dependent on IRF1 (refs. 24,25) whereas, on the basis of experiments in mouse myeloid cells, IFN-β induction by TLR4 is generally thought to mediated by TBK–IRF3 signaling^28–30. IL-10 suppressed the TNF-induced expression of IRF1 in monocytes (Extended Data Fig. 5f) but did not affect the activation of TBK1 by LPS (Fig. 6g). In monocytes stimulated with LPS, IRF1 bound a newly identified enhancer at the IFNB locus (Fig. 6h), which was suppressed in LPS + IL-10-stimulated monocytes (Fig. 6h), suggesting that IRF1 mediates IFNB induction in human monocytes. This was tested using CRISPR–Cas9 editing of the IRF1 gene locus in primary human monocytes, which resulted in an almost complete loss of IRF1 protein expression (Extended Data Fig. 5g). However, LPS-mediated IFNB induction was not significantly decreased in IRF1-deficient monocytes from multiple blood donors (Fig. 6i). IRF5 expression was suppressed by IL-10 at both the mRNA and protein levels (Extended Data Fig. 5a,f,h), and LPS-mediated induction of IFNB mRNA was markedly decreased in monocytes with CRISPR-mediated combined disruption of IRF1 and IRF5 (hereafter IRF1/5-KO monocytes) relative to control monocytes with disruption of the housekeeping gene HPRT (hereafter HPRT-KO monocytes) (mean reduction, 61%; Fig. 6i). These results showed that IRF1 and IRF5 mediated IFNB induction by LPS and that IL-10 suppressed IFNB induction in human monocytes at least in part by targeting IRF1 and IRF5.

IL-10-suppressed genes are targets of IRF1 and IRF5

Next, we characterized the functional consequences of disruption of IRF1 and IRF5 for the LPS-induced gene response in human monocytes. The induction of the ISGs CXCL10 and ISG15 (Fig. 7a and Extended Data Fig. 6a) and inflammatory genes TNF and IL6 (Extended Data Figs. 6b and Fig. 7b) was significantly decreased in LPS-stimulated IRF1/5-KO monocytes compared to HPRT-KO monocytes, but not in IL-10 + LPS-stimulated IRF1/5-KO monocytes (Fig. 7a,b) compared with HPRT-KO monocytes. This finding provides strong genetic evidence for the role of IRF1 and IRF5 in the LPS-induced expression of CXCL10, ISG15, TNF and IL6. To test whether the effects of IRF1 + IRF5 disruption phenocopied the effect of IL-10 on LPS-induced gene expression, we used RNA-seq to identify IRF1 + IRF5-dependent LPS-induced genes (Extended Data Fig. 6c) and compared them with the IL-10-suppressed gene set defined above (Fig. 1). GSEA comparison of gene expression between LPS-stimulated IRF1/5-KO and HPRT-KO monocytes showed significant suppression of IFN response genes and Hallmark inflammatory genes (Fig. 7c,d). Differential gene expression analysis identified a core set of 530 LPS-induced genes that were dependent on IRF1 and IRF5 in LPS-stimulated monocytes (Extended Data Fig. 6d–h). Comparison of this gene group with the IL-10-suppressed genes identified above revealed that 151 of the 436 IL-10-suppressed genes were dependent on IRF1 and IRF5 (Fig. 7e). This indicates that approximately 35% of the gene-suppressive activity of IL-10 was mediated by IRF1 and IRF5. The IL-10-suppressed IRF1 + IRF5-dependent genes were highly significantly enriched in both IFN response and inflammatory genes (Fig. 7f,g). Clustering based on gene expression patterns indicated that IRF1 + IRF5-dependent genes segregated into genes whose expression was strongly or partially diminished in IRF1/5 KO relative to HPRT KO monocytes, with ISGs being more strongly suppressed (Fig. 7h,i). These results identified targets of IRF1 and IRF5 in human monocytes and indicated that IL-10 suppresses a subset of LPS-induced IFN and inflammatory responses by inhibiting IRF1and IRF5 expression and function (Extended Data Fig. 7).

Fig. 7 |. Inflammatory and IFN-stimulated genes suppressed by IL-10 are targets of IRF1 and IRF5.

a,b, qPCR assay showing expression of ISG15 and CXCL10 (a) or TNF and IL6 (b) mRNA normalized relative to GAPDH mRNA in CD14⁺ monocytes in which IRF1 and IRF5 (IRF1/5-KO) or HPRT (HPRT-KO) genes had been edited using CRISPR–Cas9, followed by treatment with IL-10 (100 ng ml^–1) for 18 h and stimulation with LPS (10 ng ml^–1) for 3 h (dots represent independent donors). Data are depicted as mean ± s.e.m. Gene expression for each donor is plotted relative to the resting condition, which is set at 100. *P < 0.05 by by Mann–Whitney two-sided test. c,d, Hallmark IFN-α response enrichment plot (c) and inflammatory response enrichment plot (d) of RNA-seq data DEGs ranked by log₂(counts per million) in IRF1/5 KO versus HPRT KO CD14⁺ monocytes stimulated with LPS for 3 h (n = 3 independent donors). e, UpSet plot showing overlap between LPS-induced genes suppressed by IL-10, as identified by RNA-seq in Fig. 1 (n = 3), and genes dependent upon IRF1 and IRF5, identified by RNA-seq as in c and d, in CD14⁺ monocytes as described in a (n = 3). f, Hallmark pathway enrichment analysis of the genes in e. g, Hallmark pathway enrichment analysis of genes showing decreased expression in IRF1/5 KO relative to control HPRT KO CD14⁺ monocytes stimulated with LPS (10 ng ml^–1) for 3 h. h,i, Heatmaps showing genes downregulated in IRF1/5-KO monocytes relative to HPRT-KO CD14⁺ monocytes stimulated with LPS (10 ng ml^–1, n = 530) for 3 h, clustered by inflammatory and IFN response genes (h) or showing representative ISGs (i). Rest, unstimulated CD14⁺ monocytes. Created with BioRender.com.

Discussion

Here we found that instead of inhibiting core TLR4-activated pathways, such as NF-κB and MAPK signaling, IL-10 targeted the expression, DNA binding and transcription factor activity of IRF proteins that are activated or induced by TLR4 signaling. This resulted in a highly significant inhibition of inflammatory NF-κB target genes, in whose activation IRFs play an amplifying role, and a near-complete suppression of ISGs that are IRF-dependent. IL-10 reduced chromatin accessibility and de novo enhancer formation at ISGs and decreased the induction of IRF1-associated H3K27ac activating histone marks at both ISGs and inflammatory gene loci. IL-10 suppressed TNF-induced and LPS-induced ISG and inflammatory gene expression in a similar manner, further supporting an epigenetic mechanism of action. These results provide a mechanism by which IL-10 suppresses the induction of inflammatory genes and point to an underappreciated suppression of IFN responses by epigenetic mechanisms.

Our current understanding of the role of IRFs in the TLR4-induced IFN response is based mostly on research in mouse myeloid cells^16,17. In these cells, activation of IRF3 downstream of the signaling adapter protein TRIF plays a key role in inducing Ifnb, which acts in an autocrine manner to activate ISGs. Our results in primary human monocytes instead highlight a prominent role for TLR4-induced IRF1, in cooperation with IRF5, in the direct induction of IFNB and ISGs. Our results are in line with previous work showing that IRF1 interacts with NF-κB, AP-1 and ATF in cooperative assembly of an enhanceosome complex and recruitment of chromatin-modifying enzymes at the IFNB locus^31,32, and a recent report that proposed a mechanistic role for IRF1 in enabling IRF3 function³³. Previous reports have also provided signaling and genetic evidence that human IFNB and mouse Ifnb gene induction by TNF are mediated by IRF1, and IRF1 deficiency abrogates TNF-induced IFN responses in vivo^24,25. Thus, distinct IRFs are utilized in different contexts to drive an effective TLR4- or TNF-induced IFN response. As TBK1–IRF3 signaling remained intact in IL-10-treated human monocytes, it is likely that IRF3 contributed to the residual induction of ISGs when IRF1 and IRF5 were deleted.

Our findings implicated cooperative action of IRF1 and IRF5 in TLR4-mediated induction of inflammatory genes, which is strongly driven by NF-κB and MAPK–AP1 signaling, with an auxiliary role for other transcription factors. Combined disruption of IRF1 and IRF5 did not completely abrogate gene induction, and it is possible that additional IRFs contribute to gene expression. Rapid induction of IRF1 expression by TLRs, IFNs, cytokines such as TNF and inflammatory factors suggests that monocytes are wired to rapidly amplify inflammatory and IFN responses. In this scenario, IRF1 performs a rheostat function to broadly determine the magnitude of TLR4-induced gene responses, and this rheostat is tuned by modulating IRF1 expression. Furthermore, as IRF1 contributed to production of LPS-induced IFN-β, which then augmented IRF1 expression, our results suggest that ISG expression is magnified by an IRF1–IFN-β amplification loop.

The key finding of our study is that IL-10 turned down the IRF1-mediated ‘rheostat,’ thereby turning off the IRF1-mediated amplification of TLR4- and TNF-induced IFN and inflammatory responses. Suppression of other IRF family members, in particular IRF5, contributed to IL-10 bioactivity. IL-10 exerted a stronger suppression of ISGs than inflammatory genes in human monocytes, likely because of the central role of IRFs, in cooperation with STATs, in driving ISG transcription¹⁷. Suppression of inflammatory genes was weaker because IRFs play an auxiliary, rather than central, role in their induction. These findings provide a long-sought answer to the question of how IL-10 suppresses inflammatory genes without inhibiting NF-κB or MAPK signaling. In line with stronger suppression of ISG expression, IL-10 induced loss of both chromatin accessibility and positive histone marks at ISG loci. The decrease in chromatin accessibility at ISGs is likely explained by combined loss of IRF binding and suppression of the IFN-β autocrine loop and thus ISGF3 binding. By contrast, IL-10 minimally altered chromatin accessibility while decreasing positive histone marks at inflammatory gene loci, which is in agreement with recruitment of histone acetyl-transferases by IRFs¹⁷. Chromatin accessibility at inflammatory gene loci was likely preserved by intact NF-κB and MAPK signaling.

Increased IRF1 expression and function might play a role in autoimmune diseases such as rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE)^17,34. Activated macrophages in the inflamed joints of people with RA have elevated expression of IRF1 and genes associated with IRF1-binding enhancers³⁵. IRF1 expression is also elevated in monocytes from people with SLE, and its genome-wide binding profile is associated with histone acetylation and increased gene transcription^36–39. These findings and our observations suggest that investigation of the IRF–IL-10 axis in autoimmune diseases could be a fruitful area of research.

Methods

Primary human monocytes

Deidentified buffy coats were purchased from the New York Blood Center following a protocol approved by the Hospital for Special Surgery Institutional Review Board. Peripheral blood mononuclear cells (PBMCs) were isolated using density gradient centrifugation with Lymphoprep (Accurate Chemical), and monocytes were purified with anti-CD14 magnetic beads from PBMCs immediately after isolation as recommended by the manufacturer (Miltenyi Biotec), except that only 30% of the recommended number of magnetic beads was used. Monocytes were cultured overnight at 37 °C, 5% CO₂ in RPMI-1640 medium (Invitrogen) supplemented with 10% heat-inactivated defined FBS (HyClone Fisher), penicillin–streptomycin (Invitrogen), l-glutamine (Invitrogen) and 20 ng ml^–1 human M-CSF. Then, the cells were treated as described in the figure legends.

Sample selection

No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those reported in previous publications^34,35,40. Data distribution was assumed to be normal, but this was not formally tested; individual data points are shown. We used random blood donors but there was no randomization in experimental conditions or stimulus presentation. Data collection and analysis were not performed with blinding to the experimental conditions. Samples or donors exhibiting activation at baseline after purification of monocytes from PBMCs, as indicated by elevated basal levels of interferon response genes such as CXCL10 and ISG15, were excluded.

qPCR analysis of mRNA amounts

Total RNA was isolated using the RNeasy Mini Kit (Qiagen, cat. no. 74106), following the manufacturer’s instructions. Reverse transcription of RNA into complementary DNA (cDNA) was performed using the RevertAid RT Reverse Transcription Kit (Thermo Fisher Scientific, cat. no. K1691), according to the manufacturer’s protocol, and the resulting cDNA was used for downstream analysis. For qRT–PCR, Fast SYBR Green Master Mix (Applied Biosystems, cat. no. 4385618) and a QuantStudio5 Real-time PCR system (Applied Biosystems) were used. CT values obtained from qPCR were normalized to the housekeeping gene GAPDH. Expression of target genes relative to GAPDH was calculated using the ΔCt method, where ΔCt represents the difference in threshold cycle values between the target gene and GAPDH. The results are presented as a percentage of GAPDH expression (100 / 2^ΔCt). Primer sequences used for the quantitative RT–qPCR reactions are provided in Supplementary Table 1.

Immunoblotting

For protein analysis, we followed a previously reported method⁴⁰. In brief, 2 × 10⁶ human monocytes were washed with cold PBS after the indicated treatments and were collected in 50 μl cold lysis buffer containing Tris-HCl pH 7.4, NaCl, EDTA, Triton X-100, Na₃VO₄, phos STOP EASYPACK, Pefabloc and EDTA-free complete protease inhibitor cocktail. After a 10-min incubation on ice, cell debris was pelleted at 16,000g at 4 °C for 10 min. The soluble protein fraction was combined with 4× Laemmli Sample buffer (Bio-RAD cat. no. 1610747) containing 2-mercaptoethanol and subjected to SDS–PAGE (electrophoresis) on 4–12% Bis-Tris gels. Following transfer of gels to polyvinylidene difluoride membranes, membranes were blocked in 5% (wt/vol) BSA in TBS with Tween-20 (TBST) at room temperature for at least 1 h. Incubation with primary antibodies (diluted 1:1,000 in blocking buffer) occurred overnight at 4 °C. Membranes were washed three times with TBST and probed with anti-rabbit-IgG secondary antibodies conjugated to horseradish peroxidase (diluted 1:2,000 in blocking buffer) for 1 h at room temperature. Enhanced chemiluminescent substrate (ECL) western blotting reagents (PerkinElmer cat. no. NEL105001EA) or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific cat. no. 34095) were used for detection, followed by visualization on autoradiography film (Thomas Scientific cat. no. E3018). For multi-protein detection on the same experimental filter while minimizing stripping and reprobing, membranes were horizontally cut on the basis of molecular mass markers and the target protein sizes. Restore PLUS western blotting stripping buffer (Thermo Fisher Scientific cat. no. 46430) was applied to membranes requiring multiple primary antibody probes. The antibodies used in the study are listed in Supplementary Table 2.

RNA sequencing

Libraries for sequencing were prepared using mRNA that was enriched from total RNA using NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs (NEB) cat. no. E7490L), and enriched mRNA was used as an input for the NEBNext Ultra II RNA Library Prep Kit (NEB, cat. no. E7770L), following the manufacturer’s instructions. The quality of all RNA and library preparations was evaluated with BioAnalyser 2100 (Agilent). Libraries were sequenced by the Genomic Resources Core Facility at Weill Cornell Medicine using a Novaseq SP flow cell, 50-bp paired-end reads to a depth of ~20 million–40 million reads per sample. Read quality was assessed and adapters trimmed using FastQC and cutadapt. Reads were then mapped to the human genome (hg38) and reads in exons were counted against Gencode v38 with STAR Aligner. Differential gene expression analysis was performed in R using edgeR. Only genes with expression levels exceeding four counts per million reads in at least one group were used for downstream analysis. The Benjamini–Hochberg FDR procedure was used to correct for multiple testing. Genes were categorized as upregulated if log₂(FC) ≥ 1 and FDR ≤ 0.05 threshold was satisfied, downregulated if log₂(FC) ≤ −1 and FDR ≤ 0.05. A heatmap with k-mean clustering was done using Morpheus web application and replotted using the R package pheatmap. Bioinformatic tools used are listed in Supplementary Table 3.

Gene Set Enrichment Analysis

GSEA was conducted using the Broad Institute’s GSEA software (version 4.3.2). log₂-transformed counts per million (log₂(CPM)) values of all expressed genes (CPM exceeding four in at least one group) in our RNA-seq dataset were used for gene ranking. The analysis utilized the h.all.v2023.2.Hs.symbols.gmt Hallmark gene sets database. GSEA was performed with 1,000 permutations, following default settings for the specific comparison outlined in the figure legends.

Pathway analysis

The pathway analysis focused on investigating terms from Hallmark gene sets (https://www.gsea-msigdb.org/gsea/msigdb/human/collections.jsp#H) within a specific list of genes of interest. Gene sets representing pathways and biological processes were obtained using the R msigdbr package, which provided curated collections of genes associated with specific biological functions and pathways. Overrepresentation analysis (ORA) was performed using the hypergeometric test to quantify the degree of enrichment, comparing the proportion of genes associated with a particular term within the list to the proportion in the entire genome. Implementation was conducted in R using the clusterProfiler package, with subsequent visualization of results through dot plots or bar plots, effectively illustrating significantly enriched terms alongside their corresponding P values.

ATAC sequencing

The ATAC-seq library preparation followed a previously reported method⁴¹. One million cells were lysed using cold lysis buffer (10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3 mM MgCl₂, and 0.1% IGEPAL CA-630), and nuclei were immediately spun at 500g for 10 min in a refrigerated centrifuge. The pellet obtained after nuclei preparation was resuspended in a transposase reaction mix consisting of 25 μl 2× TD buffer, 2.5 μl transposase (Illumina cat. no. 20034198) and 22.5 μl nuclease-free water. The transposition reaction was carried out for 30 min at 37 °C. Following transposition, the sample was purified using a MinElute PCR Purification kit. Library fragments were amplified using 1× NEB next PCR master mix and 1.25 M custom Nextera PCR primers, as previously described³⁷, with subsequent purification using a Qiagen PCR cleanup kit, yielding a final library concentration of ~30 nM in 20 μl. Libraries were amplified for a total of 10–13 cycles and subjected to high-throughput sequencing at the Genomic Resources Core Facility at Weill Cornell Medicine using the Illumina NovaSeq S1 Sequencer with 50-bp paired-end reads. Data from ATAC-seq experiments were derived from four independent experiments with different blood donors.

ATAC-seq data analysis

For ATAC-seq data analysis, we utilized the TaRGET-II-ATACseq (https://github.com/Zhang-lab/TaRGET-II-ATACseq-pipeline) pipeline, available on a singularity image (ATAC_IAP_v1.1.simg), to process raw ATAC-seq data. The read alignments were performed against the GRCh38/hg38 reference human genome. Peak calling was conducted using MACS3 with the following parameters: ‘macs3 callpeak -f BAMPE -t replicate1 replicate2 replicate3 replicate4 -g hs -q 0.01 --keep-dup 1000 --nomodel --shift 0 --extsize 150’. A master consensus peak set was generated by merging the resulting peak files for each treatment condition, followed by merging peaks within 50 bp of each other. Quantification of peaks to compare global ATACseq signal changes in the BAM files was conducted using the NCBI/BAMscale program. Raw count matrices were obtained utilizing the BAMscale program. For subsequent analysis, we used the HSS Genomic Core’s reproducible ATACseq analysis pipeline for peak filtering, annotation relative to genomic features, differential peak analysis, and enrichment of signal around specific motifs using ChromVAR. Footprint analyses were performed using TOBIAS, following the user manual. In brief, we filtered the JASPAR2022-CORE_vertebrates transcription factor list on the basis of the transcription factors in our RNA-seq data and used this as transcription factor input for motif footprinting. Bioinformatics tools used in this study are listed in Supplementary Table 3.

Motif Enrichment analysis (HOMER)

De novo transcription factor motif analysis was carried out using the motif finder program findMotifsGenome in the HOMER⁴² package, focusing on the given peaks. Peak sequences were compared to random genomic fragments of the same size and normalized G+C content to identify enriched motifs in the targeted sequences.

Interactive Genome Viewer

To visualize the CUT&RUN and ATAC-seq data, bigwig files were generated for each condition by merging replicate BAM files and then creating normalized coverage bigwig files relative to the sequencing depth using BAMscale⁴³. The normalized bigwig files were then visualized using the IGV browser from the Broad Institute⁴⁴.

CUT&RUN

The Epicypher CUT&RUN kit (cat. no. 14–1048) was used, following the manufacturer’s instructions. To create CUT&RUN libraries, fragmented DNA obtained from the CUT&RUN assay was processed using the NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB, cat. no. E7645L), following the manufacturer’s instructions. Libraries were pooled and sequenced (50-bp paired-end reads) to obtain at least five million reads per sample on the Illumina Novaseq SP or Novaseq S4 at the Genomic Resources Core Facility at Weill Cornell Medicine.

CUT&RUN data analysis

A reproducible CUT&RUN analysis pipeline, CUT&RUNTools 2.0 (ref. 45), was used to process raw CUT&RUN data. In brief, the sequenced reads were aligned to the human genome (hg38) using bowtie2 (ref. 46). Peak calling was conducted using MACS2 with the following parameters: ‘macs2 callpeak -t replicate1 replicate2 replicate3 -g hs -f BAMPE -q 0.01 --scale-to small --nomodel --keep-dup all’. The same approach used for ATAC-seq was applied for downstream analysis. The bioinformatics tools used in this study are listed in Supplementary Table 3.

Flow cytometric analysis

Intracellular flow cytometry was utilized to detect IRF1 and phosphorylated p65 in human monocytes. Five minutes before cells were collected, 0.25% Pefablock was added to the cell culture medium. Cells were then centrifuged at 300g for 3 min, the supernatant was discarded, and the cell pellet was gently blotted with a paper towel to remove excess medium. Then, cells were washed once with FACS buffer. We used the Foxp3 Transcription Factor Fixation/Permeabilization kit (Invitrogen cat. no. 00-5521-00), following the manufacturer’s protocol, to fix cells for staining with antibodies to transcription factors and nuclear proteins. In brief, FOXP3 fixation buffer (100 μl per well) was added, and the plate was incubated for 20 min at room temperature. After centrifugation, cells were washed again with FACS buffer, followed by resuspension in FACS buffer. Next, cells were incubated with 0.05% digitonin for 10 min at room temperature, followed by centrifugation and a final wash. Fc blocker (BD Biolegend cat. no.422302) was added to reduce potential non-specific antibody staining mediated by IgG receptors and incubated for 10 min at room temperature. Intracellular antibody staining was performed using an antibody cocktail in 5% BSA-PBS for 60 min at room temperature, avoiding light exposure. Finally, cells were washed with FACS buffer and analyzed by flow cytometry. The antibodies and dilutions used in the study are listed in Supplementary Table 2

CRISPR

For CRISPR–Cas9 genome editing in primary human monocytes, the Lipofectamine CRISPRMAX Cas9 Transfection Reagent (Invitrogen cat. no. CMAX00008) was used. Initially, cells were resuspended at a concentration of 1.0 × 10⁶ cells ml^–1 in medium with M-CSF (20 ng ml^–1) and the Jak inhibitor baricitinib (1 μM, to prevent activation of Jak–STAT signaling in response to transfected guide RNAs). The cells were then plated at a density of 100,000 cells per well in a 96-well plate and incubated at 37 °C for 2–3 h. Cas9 ribonucleoproteins (Cas9-RNPs) were generated using CRISPRMAX Cas9 Transfection Reagent, following the manufacturer’s protocols. In brief, lyophilized single guide RNAs (sgRNAs) obtained from IDT-DNA (Supplementary Table 1) were reconstituted at a concentration of 50 μM in Nuclease-Free Duplex Buffer (IDT-DNA cat. no. 11-01-03-01). Subsequently, the two sgRNAs per gene were mixed with 61 μM Cas9 protein (volume adjusted to achieve equal molar ratio) and incubated with opti-MEM and Cas9 Reagents (part of Lipofectamine kit) for 5 min at room temperature to form Cas9-RNPs at a final concentration of 5 μM. For delivery of Cas9-RNP to cells, Lipofectamine CRIPSRMAX lipid nanoparticles were diluted with Opti-mem and incubated for 1 min at room temperature. Cas9-RNP and CRISPRMAX mix were incubated at a 1:1 ratio for 15 min at room temperature. These Cas9–gRNA–Lipofectamine complexes were added immediately to the plated cells. The cells were then incubated overnight at 37 °C. Following transfection, the medium was changed to remove residual components, and the cells were further incubated for 24 h before stimulation with LPS for 3 h.

Statistical analysis

Graphpad Prism for Macs was used for all statistical analysis. Information about the specific tests used, and number of independent experiments is provided in the figure legends. Two-way ANOVA with Tukey’s correction for multiple comparisons was used for grouped data; otherwise, one-way ANOVA with Tukey’s post hoc test for multiple comparisons was performed. The Mann–Whitney U test was performed to analyze non-parametric data. All statistical tests for P value calculations were performed using two-sided tests. Cohen’s d, a standardized measure of effect size used to quantify the difference between two groups in terms of their means, was calculated and expressed in s.d. units. It is calculated by dividing the difference between the means of two groups by the pooled s.d., which provides a dimensionless metric that facilitates comparison across studies. d values are interpreted as follows: a small effect size (d ≤ 0.2) indicates a subtle difference between groups, a medium effect size (d = 0.5) suggests a moderate difference and a large effect size (d ≥ 0.8) represents a substantial difference. Negative values indicate that the mean of the second group is larger than the mean of the first group, whereas positive values indicate the opposite. This effect size is particularly valuable in interpreting the practical significance of results, because it goes beyond P values to provide insight into the magnitude of observed differences.

Extended Data

Extended Data Fig. 1 |. IL-10 preferentially suppresses IFN relative to inflammatory responses in LPS- and TNF-stimulated monocytes.

a–f, Additional analysis of the RNAseq data shown in Fig. 1 obtained using monocytes from 3 independent donors. a, Experimental design. Human monocytes were cultured −/+ IL-10 (100 ng/mL) for 18 h and then stimulated with LPS (10 ng/ml) for 3 h. b Principal component analysis (PCA) of RNA-seq data for Resting, IL-10, LPS, and IL-10 + LPS treatments. c, Volcano plot of DEGs regulated by LPS (up=1745 and down=1738) relative to resting control. d, Monocytes or in vitro differentiated macrophages were pretreated with IL-10 and stimulated with LPS, and IL-6 mRNA was measured by RT-qPCR and normalized relative to GAPDH mRNA (dots correspond to independent donors; n = 3). ***p < 0.0005; ****p < 0.0001, 2-way ANOVA with Tukey’s multiple comparisons test. e, Venn diagram depicting LPS-induced genes that are suppressed vs those not suppressed by IL-10. f, Scatterplot of LPS-induced genes showing log2 fold change values for LPS and IL-10 + LPS relative to resting control. g, UpSet plot showing overlap between suppressed genes identified in e and genes in individual k-means clusters from Fig. 1c. h, FACS to analyze phosphorylation of NF-κB in −/+ IL-10 treated human monocytes challenged with LPS for the times (mins) indicted in the representative histogram plot. Left panels, representative histogram plots; right panels, box plot of MFI from FACS experiments (n = 4 independent donors). Dots correspond to independent donors, box plot showing the distribution of the data with markers for the minimum, 25th percentile, mean, 75th percentile, and maximum values. ***p < 0.0005; ****p < 0.0001, 2-way ANOVA with Tukey’s multiple comparisons test. i-m, analysis of RNAseq data obtained using monocytes from 3 independent donors and treated with −/+ IL-10 (100 ng/mL) for 18 h and then stimulated with TNF (20 ng/ml) for 6 h. i, PCA of RNA-seq data for Resting, IL10, TNF, and IL-10 + TNF treatments. j, GSEA plot of RNA-seq data performed on DEGs ranked by log2CPMs in TNF and IL-10 + TNF treated human monocytes. k, k-means (k = 6) clustering of differentially upregulated genes in primary human monocytes in any pairwise comparison to resting control. l, Hallmark pathway enrichment analyses of the gene clusters identified in k. m, Venn diagram depicting TNF-induced genes that are suppressed vs not suppressed by IL-10. Created in https://BioRender.com.

Extended Data Fig. 2 |. IL-10 suppresses H3K27ac at genomic regions enriched in IRF1 binding motifs.

a–e, Additional analysis of H3K27ac CUT&RUN data shown in Fig. 2 obtained using monocytes from 2 independent donors and treated as depicted in Extended Data Fig. 1a. a, PCA plot of H3K27ac data for Resting, IL-10, LPS and IL-10 + LPS treatments of primary human monocytes. b, Bar plot depicting total number and the direction of regulation of the differentially regulated H3K27ac peaks relative to resting control. c, Hallmark pathway enrichment analyses of the genes associated with peaks in groups identified in Fig. 2a. d, UpSet plot showing interaction between suppressed genes identified in Extended Data Fig. 1e and individual groups of H3K27ac peaks from Fig. 2a. e, De novo motif analysis results using HOMER on Group 3 H3K27ac peaks. Created in https://BioRender.com.

Extended Data Fig. 3 |. IL-10 decreases chromatin accessibility and preferentially suppresses IRF activity.

a–e, Additional analysis of ATACseq data shown in Fig. 3 obtained using monocytes from 4 independent donors and treated as depicted in Extended Data Fig. 1a. a, Hallmark pathway enrichment analyses of genes associated with peaks identified as suppressed or non-suppressed in Fig. 3b. b, Heatmap of the H3K27ac normalized signal density surrounding LPS-induced ATAC peaks identified as suppressed or non-suppressed and plotted under the indicated conditions. Results are presented in RPKM values within a range of ± 3.0 kb around peak centers. c–e, Volcano plot of differential binding analysis of JASPAR motifs by TOBIAS using BINDetect algorithm c, for peaks associated with IL-10, LPS, and IL-10 + LPS compared to resting control, d, for TNF vs. IL-10 + TNF, e, for TNF, and IL-10 + TNF treated human monocytes compared to resting control. Created in https://BioRender.com.

Extended Data Fig. 4 |. IL-10 broadly suppresses LPS-induced IRF1 binding and associated histone acetylation.

a–d, Additional analysis of IRF1 CUT&RUN data shown in Fig. 4 obtained using monocytes from 3 independent donors and treated as depicted in Extended Data Fig. 1a. a, Bar plot depicting total number of IRF1 binding sites from CUT&RUN experiments. b, PCA plot of IRF1 CUT&RUN experiments for Resting, IL-10, LPS, and IL-10 + LPS treated primary human monocytes. c, Hallmark pathway enrichment analyses of genes associated with LPS-induced IRF1 binding peaks. d, Representative IGV gene tracks are shown, illustrating LPS-induced IRF1 binding, H3K4me3 and H3K27ac peaks, and ATACseq peaks at the OAS3 locus. Created in https://BioRender.com.

Extended Data Fig. 5 |. IL-10 suppresses expression of IRF1 and IRF5 which mediate IFNB induction in human monocytes.

a, Immunoblot of IRF1 (top panel) and IRF5 (bottom panel) using whole cell lysates of monocytes pretreated with IL-10 followed by the indicated time course of LPS stimulation. Representative blots from 3 independent donors are depicted, HPRT is used as loading control. b, mRNA of indicated genes was measured by RT-qPCR and normalized relative to GAPDH mRNA in cells stimulated with IL-10 for 18 h and then challenged with LPS for 3 h under two conditions. No washout of IL-10 (labeled 0): IL-10 remained in the culture during the LPS challenge; washout of IL-10 (labeled 24): IL-10 was removed and the cells were cultured for an additional 24 h before being challenged with LPS for 3 h. (n = 3 independent donors). Data are depicted as mean ± SEM. **p < 0.001; ***p ≤ 0.0005; ****p < 0.0001 by two-way ANOVA with Tukey’s multiple comparisons test. c, IGV gene tracks of IL-10-induced STAT3 binding at IRF1 regulatory elements. d, IFNB mRNA was measured by qPCR and normalized relative to GAPDH mRNA in cells stimulated IL-10 for 18 h and challenged with LPS at various time points (n = 3 independent donors). Data are depicted as mean ± SEM. ****p < 0.0001 by two-way ANOVA with Tukey’s multiple comparisons test. e, mRNA of indicated genes was measured by RT-qPCR and normalized relative to GAPDH. Cells were treated with IL-10 for 18 h and stimulated with exogenous IFN-β to assess whether IL-10 blocks induction of ISGs (n = 6 independent donors). Data are depicted as mean ± SEM. *p < 0.05; ***p ≤ 0.0005; ****p < 0.0001 by two-way ANOVA with Tukey’s multiple comparisons test. f, h Heatmaps of z-score normalized expression of f, TNF RNA-seq showing regulation of IRF and ISG genes by IL-10, h, of LPS RNA-seq showing regulation of IRFs and ISGs by IL-10. g, Immunoblot of IRF1 and p38 using whole cell lysates from IRF1 or HPRT-edited cells (CRIPSR-Cas9 mediated edits) and stimulated with LPS for 3 h (representative blot from one out of 3 independent donors). Created in https://BioRender.com.

Extended Data Fig. 6 |. Inflammatory and interferon-stimulated genes suppressed by IL-10 are IRF1/5 targets.

a, b, mRNA of indicated IFN-response genes (CXCL10 and ISG15) (a) and b, inflammatory genes (TNF and IL6) was measured by qPCR and normalized relative to GAPDH mRNA. This analysis was conducted after CRISPR-Cas9 gRNA targeting of IRF1 + IRF5 or HPRT-control for 48 h, followed by stimulation with LPS for 3 h (n = 13 independent donors, some donors are same as Fig. 7a, b). Data are depicted as mean ± SEM. ****p < 0.0001 by Mann-Whitney two-sided test. c–h, Additional analysis of RNA-seq data shown in Fig. 7 obtained using monocytes from 3 independent donors and treated as depicted in Extended Data Fig. 1a. c, PCA plot depicting the RNA-seq data from primary human monocytes subjected to CRISPR-Cas9-mediated deletion of IRF1/5 or HPRT, followed by stimulation with LPS for 3 h. d, k-means clustering analysis (k = 6) conducted on z-score normalized data of differentially upregulated genes in any pairwise comparison relative to resting control. e, Hallmark pathway enrichment analyses performed on clusters identified in d. f, Scatterplot of LPS-induced genes showing log2 fold change values for HPRT-edited and IRF1/5-edited monocytes relative to resting control. g, Venn diagram of LPS-induced genes in HPRT-edited controls compared to IRF1/5-edited monocytes. h, UpSet plot showing interaction between suppressed genes identified in g to individual k-means clusters from d. Created in https://BioRender.com.

Extended Data Fig. 7 |. IL-10 downregulates an IRF1-mediated rheostat to suppress TLR4-induced expression of inflammatory NF-κB target genes and ISGs.

TLR4 signaling induces expression of IRF1, which amplifies induction of inflammatory NF-κB target genes, IFNB, and ISGs in partnership with IRF5 and ISGF3. IFN-β further augments expression of IRF1 (lower right), which in turn will increase IFN-β production; this identifies an amplification loop for increasing expression of ISGs and inflammatory genes. IRF1 also auto-amplifies its own expression. IL-10 turns down this IRF1-mediated rheostat, interrupts the IFN-β-IRF1 amplification loop, and suppresses IRF transcription factor activity to broadly downregulate induction of inflammatory and interferon response genes. Created in BioRenders.com.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41590-025-02137-3.

Data availability

Sequencing data from this study have been deposited at the GEO and will be publicly available from the date of publication. the accession numbers are GSE261243, GSE261244 and GSE261245. Source data are provided with this paper.

Code availability

Any additional information required to reanalyze the data (including original code) reported in this paper is available from the first or last authors on request.