Research ArticleMACROPHAGES

IRF5 guides monocytes toward an inflammatory CD11c+ macrophage phenotype and promotes intestinal inflammation

See allHide authors and affiliations

Science Immunology  22 May 2020:
Vol. 5, Issue 47, eaax6085
DOI: 10.1126/sciimmunol.aax6085

Targeting overzealous macrophages

Intestinal homeostasis relies on maintenance of a complex set of interactions between intestinal microbiota and the intestinal immune system. Pathogens that colonize the gut invariably disrupt these interactions and promote intestinal inflammation. Here, Corbin et al. have used a mouse pathobiont, Helicobacter hepaticus, that causes inflammation akin to human inflammatory bowel disease (IBD) to study the role of intestinal macrophages in driving inflammation. Using this model, they found the transcription factor IRF5 to be a critical regulator of macrophage inflammatory potential and that deletion of IRF5 rendered mice resistant to H. hepaticus–driven intestinal inflammation. Their studies propose IRF5 and molecules upstream of IRF5 to be potential drug targets in the treatment of human IBD.


Mononuclear phagocytes (MNPs) are vital for maintaining intestinal homeostasis but, in response to acute microbial stimulation, can also trigger immunopathology, accelerating recruitment of Ly6Chi monocytes to the gut. The regulators that control monocyte tissue adaptation in the gut remain poorly understood. Interferon regulatory factor 5 (IRF5) is a transcription factor previously shown to play a key role in maintaining the inflammatory phenotype of macrophages. Here, we investigate the impact of IRF5 on the MNP system and physiology of the gut at homeostasis and during inflammation. We demonstrate that IRF5 deficiency has a limited impact on colon physiology at steady state but ameliorates immunopathology during Helicobacter hepaticus–induced colitis. Inhibition of IRF5 activity in MNPs phenocopies global IRF5 deficiency. Using a combination of bone marrow chimera and single-cell RNA-sequencing approaches, we examined the intrinsic role of IRF5 in controlling colonic MNP development. We demonstrate that IRF5 promotes differentiation of Ly6Chi monocytes into CD11c+ macrophages and controls the production of antimicrobial and inflammatory mediators by these cells. Thus, we identify IRF5 as a key transcriptional regulator of the colonic MNP system during intestinal inflammation.


The term inflammatory bowel disease (IBD) encompasses a group of debilitating inflammatory conditions of the gastrointestinal tract that affects ~0.5 to 1% of westernized populations (1). The IBDs are associated with high morbidity and burden health care systems (2, 3). Conventional IBD therapies are limited by moderate-high rates of adverse events or patient unresponsiveness, whereas about 40% of patients successfully treated with anti–tumor necrosis factor–α (TNFα) become refractory to therapy (2). Therefore, there is unmet clinical need for IBD therapies. The etiology of IBD is unknown, but interplay between host genetics, environmental factors, and the microbiota contributes to disease pathogenesis (1).

Mononuclear phagocytes (MNPs), including monocytes, macrophages, and dendritic cells (DCs), are present in large numbers in the colonic lamina propria (cLP) and carry out diverse overlapping functions critical to the maintenance of intestinal homeostasis. The dysregulation of the intestinal MNP system leads to infection and inflammation (411).

The origin of the intestinal MNP systems has been the topic of considerable debate in recent years, clouded by inconsistent nomenclature and shared surface markers between macrophages and DCs. Intestinal lamina propria DCs at steady state are largely derived from pre-DC precursors, generated in the bone marrow, which are understood to differentiate into three major intestinal DC subsets. These subsets comprise an XCR1-positive (Xcr1+SIRPαCD103+Cd11bCX3CR1) population that is analogous to classical DC 1 (cDC1) cells, and two cDC2-like signal regulatory protein α (SIRPα)–positive (SIRPα+Xcr1Cd11b+CX3XR1+) subsets that can be further discriminated by CD103 expression (12, 13). In addition, the existence of a discrete population of hybrid macrophage/DC cells within the cDC2 intestinal compartment has been described (14). The ontogeny of intestinal DCs during inflammation is more complicated since some monocyte-derived cells may acquire phenotypic and functional DC hallmarks (1517).

Intestinal lamina propria macrophages have dual origins—from embryonically derived macrophages (CD4+ Tim4+) that self-renew and monocytes—but in the adult mouse, most of the macrophage turnover is of monocytic origin (18, 19). In mice, the differentiation of monocytes to macrophages in the cLP has been termed the “monocyte waterfall” (19). After entering the cLP, naïve Ly6Chi, major histocompatibility complex II–negative (MHCII) (P1) monocytes begin maturing by acquiring expression of MHCII (P2) before down-regulating Ly6C expression. The pool of MHCII+ cells comprises Ly6C+/−CX3CR1int monocyte/macrophage intermediates (P3) and fully mature Ly6CCX3CR1hiF4/80hiCD64hiMHCIIhi macrophages (P4) (19, 20). During infection, de novo recruited monocytes give rise to CD11c+ intestinal macrophages that are phenotypically proinflammatory (21). The blood origin of intestinal macrophage subsets was also confirmed in human studies where two monocyte-derived macrophage populations, CD11c+ with high turnover and CD11c with slow turnover, were identified at steady state (22). It was suggested that CD11c+ macrophages might be an intermediate between blood monocytes and tissue-resident CD11c macrophages (22).

The regulators that control the transition of monocytes through a number of intermediate differentiation states are largely unknown, but the cytokines, transforming growth factor–β (TGFβ) and interleukin-10 (IL-10), have been linked to the development of cLP tissue-resident macrophages (23, 24). CX3CR1IL10R− mice exhibited heightened inflammation, which maintained a proinflammatory mono-macrophage state, preventing their full differentiation and initiating spontaneous colitis (24). Loss of TGFβ receptor on macrophages resulted in a minor impairment of macrophage differentiation, defined by transcriptional profiling of monocyte to macrophage transition in the cLP (23).

One candidate intrinsic regulator of the intestinal macrophage signature is interferon regulatory factor 5 (IRF5), which was described to promote an inflammatory macrophage phenotype (25) and has variants that are genetic risk factors for ulcerative colitis and Crohn’s disease (2628). IRF5 is activated by phosphorylation and ubiquitination events downstream of pattern recognition receptors (PRRs), e.g., nucleotide-binding oligomerization domain-containing protein 2 (NOD2), Toll-like receptor 2 (TLR2), and TLR4, and directly regulates many cytokines associated with IBD (IL-1β, IL-6, IL-10, IL-12, IL-23, and TNF), placing IRF5 as a nexus for the regulation of inflammatory responses (1, 25, 29). To formally examine the role of IRF5 in the establishment of intestinal MNP system, we compared the continuum of cell states of wild type (WT) and IRF5-deficient (Irf5−/−) MNPs at steady state and during Helicobacter hepaticus (Hh)–induced intestinal inflammation using a combination of competitive mixed bone marrow chimera (MBMC), single-cell gene expression analysis [single-cell RNA-sequencing (scRNA-seq)], and functional validation approaches. Hh infection concomitant with the administration of anti–IL10 receptor (αIL10R) antibodies triggers IL-23–dependent intestinal inflammation with robust T helper cell 1 (TH1)/TH17 response, which carries many features of human IBD (9, 3032). In this model, CX3CRint and CD11c+ monocyte/macrophages intermediates drive immunopathology by producing proinflammatory cytokines such as IL-23, IL-1β, and TNFα (9, 33). We show that IRF5 promotes the differentiation of monocytes into a bactericidal and inflammatory CD11c+ macrophage phenotype during Hh + αIL10R–induced colitis and is essential for the development of immunopathology in this model.


IRF5 deficiency has limited impact on colon physiology at steady state

In steady state, we found that the colons (Fig. 1A) and ceca (fig.S1A) of WT and Irf5−/− were comparable in morphology. Sections were scored for epithelial hyperplasia, nucleated cell infiltrate, area affected, and submucosal oedema and displayed no obvious signs of inflammation (score < 3) and no morphological differences between WT and Irf5−/− (Fig. 1B and fig. S1B). The immune compartment of the cLP was evaluated by flow cytometry and revealed that the number of leukocytes in the colon (live CD45+) were comparable between WT and Irf5−/− (Fig. 1C).

Fig. 1 IRF5 deficiency has limited impact on colon physiology at steady state.

(A) Representative hematoxylin and eosin (H&E) sections of colons from WT (left) and Irf5−/− (right) mice at steady state. (B) Histopathology scoring of WT (n = 6) and Irf5−/− (n = 5) colons. (C) Number of cLP lymphocytes (cLPLs) retrieved from steady-state WT (n = 9) and Irf5−/− (n = 5) mice. (D) IRF5 expression in the steady-state cLP of WT mice (n = 3). MFI, mean fluorescence intensity. (E) The frequency of intestinal MNPs in the cLP of steady-state WT (n = 9) and Irf5−/− (n = 5) mice. (F) Quantification of early phase (annexin V+ Live/Dead) and late phase (annexin V+ Live/Dead+) cell death assessed by annexin V labeling combined with viability dye staining in WT (n = 3) and Irf5−/− (n = 3) cLP Ly6ChiMHCII+ (P2) monocytes and macrophages using flow cytometry immediately after cell isolation. (B, C, and E) Data are pooled from two independent experiments. (D and F) Data are representative of two independent experiments. (B and C) Mann-Whitney U test. (E and F) Two-way analysis of variance (ANOVA) with Sidak correction. Data presented are means ± SEM. ns, not significant. **P ≤ 0.01.

Next, we assessed the levels of IRF5 expression in the cells of the colon and demonstrated that nonmyeloid and nonleukocyte populations expressed low levels of IRF5 compared with CD11b+ myeloid cells (Fig. 1D). Among myeloid cells, MNPs, i.e., monocytes, macrophages, and DCs, expressed the highest levels of IRF5 (Fig. 1D). The composition of the cLP myeloid compartment in WT and Irf5−/− was profiled using the gating strategy that included definition of the stages of monocyte differentiation (fig.S1C) (9, 20). Frequencies and absolute numbers of Ly6ChiMHCII (P1) and Ly6ChiMHCII+ (P2) monocytes and CD11b+ DCs among the infiltrated leukocytes were similar in Irf5−/− animals, but a higher frequency of F4/80+ macrophages was observed in WT mice (5.1%) than in Irf5−/− (2.9%) (Fig. 1E and fig. S1E). IRF5-deficient and WT Ly6ChiMHCII+ monocytes and macrophages were no different in their levels of apoptosis (Fig. 1F). Thus, we hypothesized that IRF5 may promote differentiation of monocytes to macrophages in the cLP.

IRF5 deficiency protects against intestinal inflammation

Next, we evaluated the effect of IRF5 deficiency on the pathogenesis of intestinal inflammation. WT and Irf5−/− mice were subjected to Hh + αIL10R colitis for 21 days, and inflammatory indices were analyzed upon sacrifice. Morphological analysis (Fig. 2A and fig. S2A) and histological scoring indicated that both colons (Fig. 2B) and ceca (fig. S2B) were protected from colitis by IRF5 deficiency. The leukocyte infiltrate to the cLP was significantly reduced in Irf5−/− mice (Fig. 2C), consistent with the reduced levels of inflammation in the colon and cecum of Irf5−/− animals. Next, we profiled TH1 and TH17 lymphocyte responses that are involved in the pathogenesis of colitis (34). Irf5−/− mice displayed a significantly reduced TH1 effector response as quantified by emergence of interferon-γ–positive (IFNγ+) CD4+ T cells and a nonsignificant reduction in the number of IL-17a+ TH17 cells and double-positive IFNγ+/IL-17a+ cells (Fig. 2D).

Fig. 2 IRF5 deficiency protects against intestinal inflammation.

(A) Representative H&E sections of colons from WT (left) and Irf5−/− (right) mice at d21 Hh + αIL10R. (B) Histopathology scoring of WT and Irf5−/− colons. (C) Number of cLPLs retrieved from uninfected (u) and d21 Hh + αIL10R WT and Irf5−/− mice. (D) frequencies of IFNγ-, IL17A-, and IFNγ/IL17-producing CD4+ T cells in WT and Irf5−/− mice after 4 hours of culture with phorbol 12-myristate 13-acetate/ionomycin and brefeldin assessed by intracellular flow cytometry. (E) Spleen weights of WT and Irf5−/− mice at steady state and d21 Hh + αIL10R. (B and E) Data are representative of two independent experiments (u, n = 3; Hh + αIL10R n = 7). Two-way ANOVA with Tukey’s correction. (C and D) Data are representative of two independent experiments (WT u, n = 6; Irf5−/− u, n = 3; Hh + αIL10R, n = 7). Two-way ANOVA with Tukey’s correction. (F) The frequency of intestinal MNPs in the cLP at d21 Hh + αIL10R WT (n = 12) and Irf5−/− (n = 11) mice. (G) Representative H&E sections of colons from CX3CR1IRF5+ (left) and CX3CR1IRF5− (right) mice at d21 Hh + αIL10R. (H) Histopathology scoring of colons from CX3CR1IRF5+ (n = 11) and CX3CR1IRF5− (n = 11) mice at d21 Hh + αIL10R. (F and H) Data are pooled from two independent experiments. (F) Two-way ANOVA with Sidak correction and (H) unpaired t test. Data presented are means ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P < 0.0001.

Hh + αIL10R colitis in WT mice led to splenomegaly but not in Irf5−/− mice (Fig. 2E), indicating that they were also protected from systemic aspects of the disease. Despite the altered immune response, Hh presence in the cecal feces were unaffected in Irf5−/− compared with WT, quantified by detection of the Hh cytolethal distending toxin B gene (fig. S2C), ruling out differential bacterial colonization in WT and IRF5-deficient animals.

Myeloid cells make up a substantial part of the leukocyte pool at the peak of inflammation in the colon (33). Ly6Chi monocytes are rapidly recruited to the gut in response to inflammatory signals, with Ly6ChiMHCII+ inflammatory monocytes becoming the predominant cells that carry out inflammatory effector functions (9, 15, 16, 20, 24, 35, 36). We observed an increase in the frequency of Ly6ChiMHCII+ inflammatory monocytes (0.9 to 2.2%) at the peak of Hh + αIL10R–induced inflammation, whereas the frequency of F4/80+ macrophages diminished (Figs. 1E and 2F and fig. S2D). The frequencies of the DC populations remained unaffected by ongoing inflammation (Figs. 1E and 2F). IRF5 deficiency significantly attenuated the predominance of Ly6ChiMHCII+ inflammatory monocytes (Fig. 2F), approximating the monocyte-macrophage waterfall observed at the steady state (Fig. 1F). At the peak of inflammation, all MNP populations in Irf5−/− animals were smaller in absolute numbers than those of their WT counterparts (fig. S2E). Last, to confirm that IRF5 activity in MNPs is a major contributor into the immunopathology of intestinal inflammation, we subjected Cx3cr1crexIRF5flox/flox animals, which are deficient in IRF5 specifically in their MNP compartment (fig. S2E), to the Hh + αIL10R colitis model. Histological scoring indicated that both the colons (Fig. 2, G and H) and ceca (fig. S2, F and G) of these animals were protected from colitis by IRF5 deficiency in MNPs. These data demonstrate that IRF5 plays a critical role in the pathogenesis of intestinal inflammation via the MNP system.

IRF5 has limited effect on monocyte development in the bone marrow and blood

At homeostasis, a higher frequency of fully differentiated F4/80+ macrophages was observed in WT compared with Irf5−/− mice (Fig. 1E), suggesting that IRF5 may play role in monocyte differentiation. However, in Hh + aIL10R–induced colitis, the different inflammatory environments between WT and Irf5−/− animals obscured this effect (Fig. 2F). To compare the differentiation competence of WT and Irf5−/− monocytes in a shared environment, we performed MBMC experiments. The lethally irradiated mice were reconstituted with 50:50 WT:Irf5−/− bone marrow mix, and the efficiency of reconstitution in the bone marrow, of blood monocytes, and of the cLP MNP compartment was investigated. We observed no difference in reconstitution of long-term (LT) or short-term (ST) hematopoietic stem cells (HSCs), myeloid progenitors [common myeloid progenitors (CMPs), granulocyte-monocyte progenitors (GMPs), and megakaryocyte-erythrocyte progenitors (MEPs)], or Ly6Chi mature monocyte population in the bone marrow (fig. S3, A and B). IRF5 expression assessed by intracellular staining using flow cytometry was negligible in LT-HSCs, ST-HSCs, and MEPs but detectable in CMPs and GMPs. Ly6Chi monocytes express the highest levels of IRF5 among the tested progenitor and mature cell populations (fig. S3C). The reconstitution of Ly6Chi monocytes in the blood was not affected by IRF5 deficiency, but more Ly6Clo monocytes appeared to be derived from WT progenitors (fig. S3D).

To further investigate an impact of IRF5 deficiency on the composition and phenotype of monocytes in the blood in noninflammatory conditions, we conducted scRNA-seq analysis of CX3CR1+ WT and Irf5−/− cells from five MBMCs. We identified five subpopulations of cells (fig. S4A) that included discrete sets of Ly6c2hi, Cd36hi, Cd74hi, and Cd74hi/Cd209hi cells similar to those reported previously (fig. S4C) (37). As would be expected for immature monocytes, the set of Ly6c2hi cells (cluster I) also showed high expression of Sell and Ccr2 (fig. S4D). Two clusters of Ly6clo Cd36+ cells (clusters II and III) were also positive for the transcription factors Cebpb and Nr4a1, which are known to regulate the transition from Ly6chi to Ly6clo monocytes (fig. S4D) (37). Of these, the largest (cluster II) was distinguished by high expression of Itgal, whereas the smaller (cluster III) showed high expression of Apoe (fig. S4C). The remaining two clusters (IV and V) both expressed Cd74 and Ccr2, with the smallest cluster (V) also showing expression of Cd209a, Ciita, Batf3, and H2-Dmb1, suggestive of a monocyte-derived DC (moDC) precursor phenotype (fig. S4D) (38). Overall, we found broadly similar proportions of WT and Irf5−/− in the different clusters, although, consistent with the fluorescence-activated cell sorting (FACS) data (fig. S3D), the knockout did show a small decrease in Cd36hi (47.2% Irf5−/− versus 55.7% WT) together with a concomitant increase in Ly6c2hi (32.6% Irf5−/− versus 26.1% WT) cell frequency relative to WT (fig. S4B). The transcriptional phenotypes of the Irf5−/− and WT cells were highly similar within each of the clusters. In total, only eight genes (including Irf5) were found to be significantly differentially expressed [|fold change (fc)| > 1.5, Benjamini-Hochberg (BH)–adjusted P < 0.05, Wilcoxon tests], with nearly all of the differences being identified in the putative moDC precursor population (fig. S6A). Together, these data indicate that, although IRF5 is unlikely to have a global impact on monocyte development and phenotypes in the bone marrow and blood, it may help to promote the transition of blood monocytes from Ly6Chi to Ly6Clo/Cd36hi and play a role in shaping the development of Cd74hi/Cd209hi moDCs.

IRF5 has subtle effect on CD11c+ intestinal macrophages at steady state

Under homeostatic conditions, Ly6Chi monocytes continuously extravasate into the colon where they give rise to heterogenous populations of macrophages (13). Because a change in this process may affect susceptibility to colitis, we investigated whether IRF5 can act in a cell-intrinsic fashion to regulate the composition of MNP pools in the steady-state cLP. Although the reconstitution of Ly6Chi monocytes in the blood was not affected by IRF5 deficiency, more Ly6Chi monocytes in the cLP were derived from WT progenitors (fig. S3E). This finding is consistent with the previously reported more efficient recruitment of donor WT than Irf5−/− monocytes into the tissue in MBMC (39). Next, we performed scRNA-seq analysis of Cx3cr1+ MNP extracted from the cLP of the steady-state WT/Irf5−/− MBMC. We identified 10 distinct subpopulations (Fig. 3A and fig. S5) that comprised Cd209a+ DCs (cluster 10), Clec4a+ and Ly6c2hi monocytes (clusters 8 and 9), and seven Adgre1+ (encoding for F4/80) macrophage populations (clusters 1 to 7) (Fig. 3A). As expected, the Mϕ clusters represented the majority of MNPs in uninfected MBMC, consistent with the FACS-based analysis (Fig. 1E and fig. S6C). The Mϕ broadly split into two compartments that were distinguished by expression of Itgax (Cd11c) and Mrc1 (Cd206) (orange and red ellipses in Fig. 3A). The Cd11c Mϕ also showed expression Cd9 and Acp5 and comprised separate populations of Hes1+ and Il7r+/Il10+ cells that may represent epithelial associated and resident tolerogenic Mϕ populations, respectively (Fig. 3D) (33, 40). Macrophages with high expression of Mrc1 also showed higher expression of Alox5ap and the anti-inflammatory molecule Ifitm3 (Fig. 3E) (41). Subpopulations of the Mrc1 Mϕ were characterized by high expression of key monocyte chemoattractants Ccl8 (42) and Ccl2 (which encodes the ligand for CCR2). Orthogonal to groupings by Cd11c versus Mrc1 status, the Mϕ showed differences in the expression of Runx3 and Cx3cr1 that are associated with mature macrophages (Fig. 3F). Although surface expression of CX3CR1 protein is known to be associated with maturity, Cx3cr1 gene expression was higher in Runx3- cells, suggesting that transcription of this gene is down-regulated as Mϕ mature. In addition, both the Cd11c and Mrc1 Mϕ populations exhibited apparent differences in activation state being split between expression of Klf2, which is known to inhibit the proinflammatory activation of immune cells (43) and expression of genes associated with Mϕ activation such as the key nuclear factor κB target gene Rel and the Nlrp3 inflammasome (Fig. 3F). No Timd4 (Tim-4) expression was detected in any of the macrophage clusters, indicating that, as expected, the monocyte-independent resident macrophage population was not represented among the donor-derived cells (18). Overall, the distribution of WT and Irf5−/− cells between the clusters was similar (Fig. 3B), although there were fewer Irf5−/− (25.2%) than WT (32.1%) cells in the activated Cd11c Mϕ (clusters 1 and 6). In contrast, there was an increase in the frequency of DCs (4.0% versus 1.6%) and Clec4e+ monocytes (cluster 8; 7.4% versus 4.3%) among the Irf5−/− cells. Across the clusters, only 34 genes were significantly affected (|fc| > 1.5, BH-adjusted P <0.05, Wilcoxon tests) by the lack of IRF5, with the majority of differences (n = 23) being observed in the DC cluster (fig. S6B). However, among the Cd11c+ macrophage clusters 1, 3, and 6, we did note a consistent down-regulation of Ccl4 (also known as macrophage inflammatory protein-1β) in the Irf5−/− cells (fig. S6B).

Fig. 3 IRF5 has subtle effect on CD11c+intestinal macrophages at steady state.

WT and Irf5−/− CD45+CD11b+SiglecFLy6GCX3CR1+ cells were sorted from the colons of five MBMC animals and subjected to scRNA-seq analysis. (A) Graph-based clustering (58) of equal numbers of WT and Irf5−/− cells (n = 4780 total) identified nine clusters of MNPs and one cluster of DCs. (B) The bar plots show the percentages of WT and Irf5−/− cells that were found in each cluster. KO, knockout. Panels (C) to (F) show the expression of DC cell type markers (C), genes expressed in Cd11c+ macrophages (D), genes expressed in Mrc1+ macrophages (E), and genes with associated with macrophage differentiation and activation (F).

Together, absence of the IRF5 had a subtle effect on the composition and phenotype of the steady-state cLP MNP compartment that was suggestive of a role for IRF5 in controlling CD11c+ MNP development.

IRF5 promotes generation of CD11c+ macrophages in inflamed colon

Next, we subjected CX3CR1+ WT and Irf5−/− MNPs isolated from the inflamed cLP of three Hh + αIL10R MBMCs to scRNA-seq analysis (fig. S7A). Examination of the top cluster markers genes (fig. S7B) revealed the existence of two groups of monocytes, two clusters of macrophages, and four clusters of DCs (Fig. 4A). The monocyte clusters comprised a set of Ly6c2 high immature monocytes (“Ly6c mono”) and a group of mature monocytes (“MHCII mono”) that expressed MHCII genes such as H2-Ab1 and the proinflammatory cytokine Il1b (Fig. 4B). The two macrophage clusters were demarcated by expression of Adgre1 (F4/80), Cd81, and Cx3cr1. The largest cluster of Cd11c Mϕ was characterized by the high expression level of Itgax (Cd11c) and known cLP macrophage markers, such as MHC glycoproteins (H2-M2), complement molecules (C1qa, C1qb, and C1qc), tetraspanins (Cd63, Cd72, and Cd81), oxidative stress response (Hebp1), and antimicrobial molecules (Acp5 and Dnase1l3) (Fig. 4B and fig. S7B). The second cluster of Cd206 Mϕ lacked Itgax expression but was defined by high expression of Mrc1 (Cd206), chemokines (e.g., Ccl2, Ccl3, Ccl4, Ccl7, Ccl8, Ccl12, and Cxcl2), scavenger, phagocytic and immunoactivating receptors (Cd36, Fcgr4, and Clec4b1), and antiviral molecules (Ch25h and Gbp2b) (Fig. 4B and fig. S7B). The remaining four clusters of cells lacked Cd64 expression and showed expression of established DC markers such as Flt3, Cd11c, and the Ciita-dependent DC-specific MHCII genes H2-DMb2 and H2-Oa (44). “Sirpa DC i” and “Sirpa DC ii” clusters displayed a DC2-like profile being marked by expression of Sirpa, Kmo, Cd209a, and Cd7 (Fig. 4A and fig. S7B). The cells in these clusters also strongly expressed Spi1 (PU.1) but were distinguished by low Flt3 expression, suggesting that they may be moDC (Fig. 4B) (38, 45). The remaining two DC clusters comprised a set of Xcr1highIrf8highSirpalow cells (“Xcr1 DC”) that are likely to correspond to conventional cDC1 cells and a small group of migratory Ccr7-positive DCs (“Ccr7 DC”). In comparison to that observed in uninfected animals, the MNP population structure at the peak of Hh + αIL10R–induced inflammation (Fig. 4, A and C) showed a marked increase in the numbers of Ly6c2hiMhcII+ inflammatory monocytes, a larger and more heterogeneous DC population along with a diminished frequency of macrophages, consistent with above analysis (Fig. 2F and fig. S7C).

Fig. 4 IRF5 promotes generation of CD11c+macrophages in inflamed colon.

WT and Irf5−/− CD45+CD11b+SiglecFLy6GCX3CR1+ cells were sorted from the colons of three MBMC animals at d21 of Hh + αIL10R colitis and subjected to droplet-based single-cell transcriptomic analysis. (A) Graph-based clustering (58) of equal numbers of WT and Irf5−/− cells (n = 1106 total) identified four clusters of MNPs and four clusters of DCs. (B) The violin plots show the expression levels (x axes) of selected known cLP MNP and DC subpopulation markers in each of the identified clusters (y axes). (C) The bar plots show the percentages of WT and Irf5−/− cells that were found in each cluster.

Fig. 5 IRF5 defines an inflammatory MNP signature during colitis.

WT and Irf5−/− P1 monocytes, P2 monocytes, and macrophages were sorted from three MBMC animals at d21 Hh + αIL10R. (A) The dot plot shows the expression [mean transcripts per million (TPM), n = 3 biological replicates] of selected genes found to be differentially expressed between WT and Irf5−/− cells at one or more stages of the monocyte waterfall. The significant changes (|fc| > 2, BH-adjusted P < 0.05) are indicated by the gray triangles. (B) Selected GO biological process categories that showed a significant enrichment (colored dots, gene set enrichment analysis, BH-adjusted P < 0.1) in at least one of the three Irf5 KO versus WT small-bulk RNA-seq comparisons. (C to E) Intracellular or extracellular flow cytometry labeling was used to quantify the expression of (C) TNFα, (D) pro–IL-1β, and (E) MHCII on WT versus Irf5−/− macrophages in MBMC uninfected (n = 3) and at d21 Hh + αIL10R colitis (n = 4). Two-way ANOVA with Sidak correction. Data from one representative experiment presented are means ± SEM. *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

When clusters were split by genotype, it was found that monocyte clusters had similar numbers of WT and Irf5−/− cells, whereas macrophage clusters, especially Cd11c Mϕ, had higher numbers of WT cells, while the Sirpa DC i and ii clusters contained higher numbers of Irf5−/− cells (Fig. 4C), findings that were confirmed by FACS-based analysis (fig. S7C). The decrease in the frequency of Irf5−/− Cd11c Mϕ relative to WT was more notable during the peak of inflammation (Fig. 4C) than at steady state (Fig. 3B), suggesting that the propensity of IRF5 to promote generation of CD11c+ macrophages is accentuated by the inflammatory environment. Together with the fact that CD11c+ monocyte/macrophages can drive immunopathology (9), this observation provided a possible explanation for the pathogenic role of IRF5 in intestinal inflammation (Fig. 2).

IRF5 defines an inflammatory MNP signature during colitis

To investigate the effect of IRF5 on the transcriptional phenotype of intestinal monocytes and macrophages, we conducted small-bulk RNA-seq analysis of WT and Irf5−/− Ly6ChiMHCII (P1) monocytes, Ly6ChiMHCII+ (P2) monocytes, and Ly6CMHCII+F4/80+ macrophages (n = 100 cells per sample) from each of the inflamed colons of three Hh + αIL10R MBMC animals. First, we identified the genes that showed significant variation between the WT monocyte and macrophage samples. Hierarchical clustering of these genes revealed that the transcriptome of P2 monocytes overlaps with the profiles of both P1 monocytes and macrophages in line with the concept that they represent a transitional state of monocyte to macrophage differentiation (fig. S8A). We also detected high expression levels for genes previously shown to be associated with mature intestinal macrophages, such as MHC molecules (H2-M2), tetraspanins (Cd72 and Cd81), complement molecules (C1qa, C1qb, and C1qc), chemokines (Ccl5 and Ccl8), and phagocytic and immunoactivating receptors (Fcgr4, Fcer1g, and Cd300e) in the macrophage populations (fig. S8C) (23). Next, we identified genes that were significantly (BH-adjusted P < 0.05, |fc| > 2) regulated by IRF5 in each of the P1 (n = 607 genes), P2 (n = 761 genes), and macrophage (n = 977 genes) compartments (fig. S8B). Among the differentially expressed genes, Ly6ChiMHCII Irf5−/− P1 monocytes showed significantly lower levels of Smad2 and Kdm3a that respectively transduce and positively regulate TGFβ and Jak2/Stat3 signaling, pathways of known importance for monocyte maturation (Fig. 5A). In line with this observation, IRF5-deficient macrophages failed to down-regulate genes highly expressed in P1 and P2 monocytes including Plac8, Cdkn2d (P19ink4d), and Irf1 and also showed significantly lower expression of the MHCII molecule H2-M2 (Fig. 5A). At the same time, in macrophages, loss of IRF5 reduced expression of the key proinflammatory cytokines (Il-12b, Ccl11, and Tnfsf13b/BAFF) and expression of the immunoactivating receptor Cd300e, tetraspanins (Cd81 and Cd72), and IL-10. A significant reduction in the expression of the key proinflammatory cytokine Il-12b was also observed in the P2 monocytes. These changes were accompanied by up- and down-regulation of the epigenetic regulators Hdac2 and Hdac9 in IRF5-deficient macrophages. At the pathway level, gene set enrichment analysis of Gene Ontology (GO) biological process categories revealed that IRF5 broadly modulated inflammatory pathways including “leukocyte activation,” “response to IFNγ,”, “response to bacterium,” and “regulation of T cell activation” in both monocytes (P1 and P2) and macrophages (BH-adjusted P < 0.1, Fig. 5B). Genes regulated by IRF5 in the P2 monocyte compartment displayed a significant enrichment of genes involved in “regulation of IL-12 production,” “IL-6 secretion,” and “IL-1 production,” whereas genes associated with “response to IL-1” and “TNF superfamily cytokine production” were also affected in macrophages (Fig. 5B). These pathways, and specifically production of IL-23, IL-1, and TNF, have been previously associated with colitis development and/or IBD (9, 29, 33). In independent experiments, using flow cytometry, we confirmed that colonic WT macrophages in the MBMCs produced higher levels of cytokines TNFα and IL-1β cytokines than Irf5−/− cells (Fig. 5, C and D). The surface expression of MHCII was higher on WT macrophages relative to Irf5−/− (Fig. 5E).

When small-bulk RNA-seq data were compared with scRNA-seq gene expression data, a good correspondence between the genes expressed in the Ly6c and MHCII monocyte clusters and P1 and P2 samples, respectively, was observed (fig. S8C). Both the Cd11c and Cd206 macrophage clusters showed similarities to the small-bulk macrophage sample (fig. S8C).

IRF5 promotes monocyte to Cd11c macrophage differentiation during intestinal inflammation

Given the reduction in Irf5−/− macrophages numbers that we saw in the MBMC at the peak of Hh + αIL10R–induced colitis (Fig. 4C and fig. S7C), we examined an apparent role for IRF5 in controlling monocyte differentiation in more detail. First, we performed a global comparison of genes regulated by IRF5 in macrophages with those that were associated with differentiation from P1 monocytes to macrophages in the inflamed cLP. Overall, this analysis revealed a significant positive correlation between genes up-regulated during macrophage differentiation and those positively regulated by IRF5 in macrophages (Spearman’s rho: 0.45, P = 5.5 × 10−110) (fig. S9). Close examination of the scatter plot, however, also revealed a large number of genes that were up-regulated in macrophages independent of the presence of IRF5 including C1qc, Ptgs1 (Cox1), and Mmp13 (green dots, fig. S9). These data support a cell-intrinsic role for IRF5 in regulating myeloid cell differentiation and phenotype during intestinal inflammation.

To dissect the role of IRF5 in controlling MNP differentiation in the inflamed cLP in more detail, we applied the Slingshot pseudotime algorithm (46) to our scRNA-seq data. After exclusion of the DCs, higher-resolution analysis identified six clusters of monocytes and macrophages (fig. S10) that fell into three predicted lineages (with Ly6c2 monocytes assumed to represent the “root” state) (Fig. 6, A and B). These represented the differentiation of (i) Cd11c (Itgax) macrophages that also expressed Acp5; (ii) Cd206 (Mrc1) macrophages that were also positive for Cd36, Ccl2, and Ccl7; and (iii) a small population of Clec4e expressing mature (MHCIIhi) monocytes that resembled those found in the steady-state data (see Fig. 3). The Irf5−/− cells were underrepresented (4.3% versus 22% of WT cells) in the terminal cluster of the Cd11c lineage (Fig. 6C, cluster 3) with progression of Irf5−/− cells in pseudotime along this lineage being significantly different to that of the WT cells (Bonferroni-adjusted P = 2.2 × 10−6; Fig. 6D). The pseudotime distribution of Irf5−/− cells along the Cd206 lineage was also significantly altered (Bonferroni-adjusted P = 0.003; Fig. 6D), but this was associated with only a slight reduction in the number of Irf5−/− Cd206 macrophages (Fig. 6C; cluster 4, 8.5% versus 11.7% of WT cells). In contrast, a similar percentage of Irf5−/− and WT cells was found in the Clec4e monocyte cluster (Fig. 6C, cluster 5), and there was no difference between the progression of Irf5−/− and WT cells along this lineage (Fig. 6D).

Fig. 6 IRF5 promotes monocyte to Cd11c macrophage differentiation during intestinal inflammation.

WT and Irf5−/− monocytes and macrophages from the inflamed intestine (Fig. 4) were reclustered at higher resolution (fig. S10) and subjected to pseudotime analysis. (A) Embedding of the cells in the first three dimensions of a diffusion map shows the three differentiation trajectories (solid lines) identified by the Slingshot pseudotime algorithm (with the Ly6c2 monocytes assumed to represent the root state). (B) Expression of selected cell-type marker genes and genes associated with Cd11c (Itgax) and Mrc1 (Cd206) macrophages. (C) The bar plots show the percentages of WT and Irf5−/− cells that were found in each cluster. (D) The violin plots show the progression of the WT and Irf5−/− (KO) cells through pseudotime along the three identified trajectories [as shown in (A)]. Differences in the distribution of cells in pseudotime between the genotypes were assessed with a Kolmogorov-Smirnov tests (P values adjusted using the Bonferroni correction). The position of the cells in pseudotime is shown on top of the violin plots [cells colored by cluster as in (A)]. The position of the 50th quantiles is indicated by the bold vertical lines.

The reduction of Irf5−/− cells in the macrophage clusters was paralleled by an increase in the number of Irf5−/− cells in the immature Ly6c2 monocyte cluster (cluster 2, 23.0% versus 14.6% WT cells) and differentiating monocyte cluster (cluster 0, 36% versus 20.6% of WT cells) (Fig. 6C). These data suggest that IRF5 may promote the acquisition of CD11c expression by Ly6ChiMHCII+ monocytes and their differentiation to CD11c+ macrophages.

CD11c+ macrophages occupy a distinct colonic niche

We assessed the localization of CD206+ and CD11c+ macrophage subsets in the colon by performing labeling of colonic tissue sections with antibodies against CD11c, CD206, and F4/80 and subsequent analysis by confocal microscopy (Fig. 7A). This analysis revealed that the two macrophage subsets localized at distinct sites at steady state. CD206+/F4/80+ cells (CD206+ macrophages) dominated the colonic macrophage pool and were located within the lamina propria, with some found to be residing at the base of intestinal crypts (Fig. 7B). Macrophages at the base of crypts are believed to be involved in response to the mucosal barrier damage via secretion of CCL8 (42) and other chemokines and may transmit regenerative signals to neighboring colonic epithelial progenitors (47). Chemokines (Ccl2, Ccl7, and Cxcl2) were distinctively expressed in Cd206 Mϕs during Hh + αIL10R–induced colitis (fig. S11A).

Fig. 7 CD11c+macrophages occupy a distinct colonic niche.

(A) Representative images of immunofluorescent labeling of colonic sections at steady state. Individual channels visualize the distribution of F4/80+ (blue), CD206+ (green), and CD11c+ (red) cells within the structure of the colon. CD206+ macrophages (white arrow) and CD11c+ macrophages (*), as well as single-positive F4/80+ cells (#), can be detected in the merged image. Cell nuclei are labeled with Sytoxblue (gray). Scale bars, 50 μM. (B) Localization of double-positive CD11c+ F4/80+ (cyan) and CD206+ F4/80 (yellow) cells in steady state (n = 6) and colitic mouse colons (n = 6) by immunohistofluorescence. Separate channels based on overlap of staining were created. Cell nuclei were labeled with SYTOX blue (gray). A minimum of five sections per mouse were evaluated. Macrophages at the base of the crypts (white arrow), at the tips of the villi (*), interspersed within the villi (†), and at the muscularis mucosae membrane (#). Scale bars, 100 μM (in the overview images) and 50 μM (in the enlargement). (C) Schematic depiction of image quantification analysis. Localization of macrophages is assessed by their minimal distance (black arrow) to the tip of the villi (artificial luminal surface depicted in green), muscularis mycosae (red), and serous membrane (blue). (D) Quantification of minimal distance of CD206+ F4/80+ and CD11c+ F4/80+ cells in steady state and d21 Hh + αIL10R to the luminal surface. Minimal distance is presented as a percentage of the distance to the tip of the villi to the total distance between the luminal surface and serous membrane. Two-way ANOVA with Tukey’s multiple comparisons test. Data presented as means ± SEM. P > 0.05, *P < 0.05, **P < 0.01, and ***P < 0.001.

CD11c+ F4/80+ cells (CD11c+ macrophages) were more sparsely located and mainly found at the luminal surface (Fig. 7B). Thus, CD11c+ macrophages may represent a primed macrophage phenotype, ready to respond to microbial encroachment (21). Many genes specifically expressed in Cd11c Mϕs during Hh + αIL10R–induced colitis belonged to an antimicrobial defense program (Dnase1l3, Acp5, Mmp14, etc.) and protein recycling (Ctsa, Ctsh, and Ctsz) (fig. S11A). We quantified the distance of the macrophage subsets in relation to the luminal surface and the serous membrane (Fig. 7C). At steady state, CD11c+ macrophages were found to be significantly closer to the luminal surface than CD206+ macrophages, but this proximity was lost during inflammation (Fig. 7D), when CD206+ and CD11c+ macrophages started to intersperse throughout the cLP and at the muscularis mucosae (Fig. 7B).

IRF5 controls the phenotype of CD11c+ macrophages

Previously, CD11c+F4/80+ monocytes and macrophages were shown to be critical effector cells in the development of Hh + αIL10R–induced experimental colitis via the production of IL-23 (9). Here, we confirmed that CD11c+F4/80+ macrophages (fig. S11B) produced higher levels of IL-12p40, a subunit of IL-23 (Fig. 8A), and other inflammatory cytokines, such as TNF and IL-1β, than CD11c macrophages (Fig. 8, B and C). In the setting of the MBMC, we confirmed that IRF5 promoted the development of CD11c+ macrophages both at the peak of Hh + αIL10R–induced colitis (Fig. 8D) and in steady state (fig. S11C). In keeping with this observation, the expression of IRF5 protein was higher in CD11c+ macrophages and Ly6ChiMHCII+ monocytes compared with their CD11c counterparts (Fig. 8E and fig. S11D).

Fig. 8 IRF5 controls phenotype of CD11c+macrophages.

(A to C) Comparison of IL-12p40, TNF, and IL-1β inflammatory cytokine expression in CD11c+ versus CD11c cLP macrophages assessed by intracellular flow cytometry in uninfected MBMC (n = 3) and d21 Hh + αIL10R (n = 4). One experiment. Two-way ANOVA with Tukey’s correction. Data presented are means ± SEM. *P ≤ 0.05, **P ≤ 0.01, and ****P < 0.0001. (D) The frequency of parent WT and Irf5−/− macrophages expressing CD11c or CD206 at d21 Hh + αIL10R colitis. Two-way ANOVA with Sidak correction. Data presented are means ± SEM from two independent experiments. ****P < 0.0001. (E) IRF5 expression in CD11c+ versus CD11c macrophages in MBMC assessed by intracellular flow cytometry. One representative experiment, uninfected n = 3, Hh + αIL10R n = 4. (F) Heat map of expression of selected genes in WT and Irf5−/− Cd11c macrophages from the inflamed cLP of the MBMCs (see Fig. 5). All of the genes shown were found to be significantly differentially expressed between the WT and Irf5−/− cells of this cluster (Wilcoxon tests, BH-adjusted P < 0.05). Asterisks (*) denote significant differential expression between the genotypes in the macrophage small-bulk RNA-seq data.

At the peak of Hh + αIL10R–induced colitis, IRF5 positively regulated a cassette of genes that defined cLP macrophage phenotypes, such as MHC molecules (H2-M2), tetraspanins (Cd72 and Cd81), complement molecules (C1q), chemokines (Ccl4), acid phosphatase 5 (Acp5), and phagocytic and immunoactivating receptors (Fcgr4, Fcer1g, and Cd300e) in the Cd11c Mϕ population. A number of killer cell lectin-like receptor family members (Klrb1b, Klra2, and Klra17), not previously associated with macrophage function, were also affected by the lack of IRF5 in this compartment (Fig. 8F). Together, our data show that IRF5 promotes the differentiation and inflammatory phenotype of Cd11c+ macrophages during Hh + αIL10R–induced colitis.


Using a model of MNP development in the gut and a combination of MBMC approaches and single-cell analysis of gene expression, we have demonstrated the importance of IRF5 in promoting the generation of macrophages in the cLP. We found that it dictates an inflammatory CD11c+F4/80+ macrophage phenotype in inflammation and controls the immunopathology of Hh + αIL10R–induced colitis.

Our results revealed a cell-intrinsic role for IRF5 in the control of a wide range of genes and biological pathways related to monocyte differentiation, leukocyte activation, response to bacterium, PRR signaling pathway, and regulation of T cell activation in inflamed intestine (Fig. 5). Mice with a global or MNP-specific loss of IRF5 were protected from Hh + αIL10R colitis (Fig. 2). In comparison, IRF5 had a much more limited impact on gene expression at steady-state intestine (Figs. 1 and 3), and we observed no morphological differences in cLP between WT and Irf5−/− at steady state (Fig. 1), consistent with the recently published report (48).

The most consistent function of IRF5 identified in this study is its ability to promote a proinflammatory monocyte and macrophage state, which is positive for CD11c. Cd11c+ macrophages were found at the luminal surface at homeostasis and throughout the cLP in inflammation (Fig. 7). They transcribe antimicrobial molecules, such as cathepsins, and are efficient producers of inflammatory cytokines, such as TNF and IL-1β, that support pathogenic T cell responses in the intestine (Fig. 8 and fig. S8) (34, 49, 50). CD11c+ macrophages produce high quantities of IL-23 in the early stages of Hh-induced colitis and are essential for triggering intestinal immunopathology (9, 15). They are also essential producers of IL-1b and IL-23 in Citrobacter rodentium–induced colitis (21). CD11c+ intestinal macrophages were marked by high level of IRF5 (Fig. 8). IRF5 deficiency ameliorated the accumulation of CD11c+ macrophages in cLP (Fig. 8D and fig. S11B). With the recently established link between CD11c+ macrophages and IRF5 in the development of atherosclerotic lesions (51), our data here support the notion that IRF5 may guide monocyte differentiation toward inflammatory CD11c+ macrophages in a variety of tissues and pathologies. Although we only found a subtle effect for IRF5 on monocyte development in the bone marrow and subset conversion in the blood at steady state (fig. S4), it is possible that it has a larger effect in these compartments during inflammation.

The second major population of macrophages detected in our analyses was marked by the expression of CD206 and predominantly located at the base of crypts at steady state (Fig. 7). These macrophages expressed phagocytic receptors, the scavenger receptor Cd36, which is critical for lysosomal lipolysis (52), the anti-inflammatory gene Ifitm3 (41), Alox5ap involved in leukotriene biosynthetic pathway, and a milieu of chemokines (Ccl2, Ccl7, Cxcl2, etc.) (Figs. 3 and 4 and fig. S11). These may represent resident macrophages involved in the clearance of senescent epithelial and apoptotic cells, sensing and regulating response to mucosal damage and possibly contributing to epithelial renewal. The Cd206 macrophages appeared largely unaffected by IRF5 deficiency (Fig. 3) and unlikely to be major contributors to the Hh-induced pathology (33). Pseudotime analysis of our scRNA-seq data indicates that the Cd206 and Cd11c macrophages broadly represent alternative macrophage differentiation trajectories during intestinal inflammation (Fig. 6). Together with the distinct distribution of these two macrophage populations in the cLP (Fig. 7) and their unequal dependence on IRF5 (Fig. 8), these data suggest that they may emerge independently in specific environmental niches. Our steady-state single-cell data (Fig. 3) suggest that there is substantial heterogeneity within both the Cd206 and Cd11c macrophage populations (the Cd206 populations can be readily split, for example, by Cd14 status) and further imaging and lineage tracing studies are needed to resolve the niches, origins, and functions of these subsets. In both our steady-state and inflamed single-cell datasets, we noted the presence of a small mature (MHCIIhi), activated (Rel+, Nlrp3+) Clec4ehi monocyte (F4/80-) population that appeared unaffected by absence of IRF5 (Figs. 3 and 6). Given the known roles of Clec4e (Mincle), a C-type lectin receptor, these cells may play important roles in host defense and tissue repair in the intestine (53).

The Cd206 macrophages in the cLP transcribed high levels of CCL2 (Fig. 4), a critical chemokine for accumulation of monocytes in the cLP (20). Consistent with previously published analysis (39), we observed more efficient recruitment of donor WT than donor Irf5−/− monocytes to cLP in the MBMC animals, highlighting another mechanism by which IRF5 could modulate inflammation, i.e., via controlling a pathogenic positive-feedback loop of inflammatory monocyte recruitment.

Last, our data suggest that in an inflammatory environment, IRF5 specifically promotes key aspects of macrophage differentiation while repressing DC transition (Fig. 4). The observed changes in expression of the histone deacetylases Hdac2 and Hdac9 (Fig. 5) may be consistent with a role for IRF5 in controlling of the epigenetic state of these cells. This process may be due to loss of competition for IRF binding sites and engagement of an IRF4-dependent differentiation program (54, 55). IRF4 and IRF5 were shown to compete for binding to myeloid differentiation primary response 88 and activation after TLR4 ligation (56). IRF4 is a key regulator of intestinal CD11b+ DC subsets and a critical transcription factor in the DC fate of monocytes in in vitro bone marrow cultures (55, 57). Thus, in the absence of IRF5, IRF4 may be able to dominate the fate choice of monocytes, explaining the increased predisposition to DC fate in Irf5−/−.

Although intestinal DCs are believed to be largely derived of FLT3L-dependent progenitors (13), several studies have provided evidence that Sirpa CD11b+ DCs are replenished by monocytes in the inflamed cLP (16, 17). It is intriguing that MHCII+ Cd209+ blood monocytes, previously identified as precursors of moDCs (38), showed the highest number of genes affected by the lack of IRF5, whereas all other monocyte populations remained largely unaffected (figs. S4 and S6A). This may reflect the more advanced differentiated state of MHCII+ Cd209+ monocytes, but more functional characterization of the populations is needed. In summary, the data presented here reveal that IRF5 controls the MNP system in the colon, is a critical driver of intestinal inflammation, and promotes monocyte differentiation toward bactericidal and inflammatory CD11c+ macrophages.


Study design

The purpose of this study was to understand the intrinsic role of IRF5 in directing macrophage polarization and intestinal inflammation. Flow cytometry, bulk- and scRNA-seq, and immunofluorescence labeling of intestinal tissue sections were used to analyze the leukocyte milieu in the colons of WT or Irf5−/− or mixed bone marrow chimeric mice. Mice were aged between 8 and 16 weeks at the commencement of experiments. Experimental sample sizes were not predetermined. Hh infections were ended upon mouse sacrifice at day 21 (d21) after infection. In general, experiments were performed at least twice unless indicated otherwise. Data were not excluded from analysis except for quality control failures in RNA-seq analysis detailed in Materials and Methods (Supplementary Materials). Histopathology assessment was conducted in a blinded manner independently by two researchers. Experimenters were not blinded to intervention groups for flow cytometry analysis.

Data availability

Next-generation sequencing datasets are available via the Gene Expression Omnibus under accession code GSE129258.


Materials and Methods

Fig. S1. IRF5 deficiency and cecum physiology at steady state.

Fig. S2. IRF5 deficiency and cecum in inflammation.

Fig. S3. MBMC: Monocyte development in the bone marrow and blood.

Fig. S4. scRNA-seq: Monocyte development in blood.

Fig. S5. scRNA-seq: MNP populations in cLP at steady state.

Fig. S6. scRNA-seq: IRF5 effect on blood monocytes and cLP MNPs at steady state.

Fig. S7. scRNA-seq: MNP populations in inflamed colon.

Fig. S8. scRNA-seq and bulk RNA-seq: Comparison of IRF5-dependent genes.

Fig. S9. IRF5 in monocyte to macrophage differentiation.

Fig. S10. scRNA-seq: Monocyte and macrophage populations in inflamed colon.

Fig. S11. scRNA-seq: Gene expression in Cd11c versus Cd206 macrophages.

Table S1. List of antibodies used for surface staining.

Table S2. List of antibodies used for intracellular staining.

Table S3. Raw data in Excel spreadsheet.


Acknowledgments: We are grateful to S. Teichmann (Wellcome Sanger Institute) for the help in establishing Smart-Seq2 protocol and generating preliminary data that inspired our subsequent single-cell analysis. We thank the High-Throughput Genomics Group (Wellcome Trust Centre for Human Genetics) for the generation of the sequencing data, C. Pearson for the assistance with the generation of MBMCs, J. Webber for the assistance with cell sorting, and the Kennedy Institute Histopathology Team for sectioning and staining of mouse colons. Funding: This work was supported by the Kennedy Trust for Rheumatology Research (A.L.C., M.G.-V., D.L.B., and S.N.S.), the MRC CGAT program (S.N.S.), the Novo Nordisk Foundation (Tripartite Immunometabolism Consortium - grant NNF15CC0018486 to I.A.U.), and the Wellcome Trust (Investigator Award 095688/Z/11/Z to F.M.P. and 209422/Z/17/Z to I.A.U.). Author contributions: A.L.C. performed all experiments, except as noted below. M.G.-V. and S.N.S. performed all computational analyses. D.L.B. conducted immunofluorescence microscopy. M.A. generated scRNA-seq libraries. I.C.A. provided advice and assisted with Hh infections and phenotype analysis. I.A.U., S.N.S., and F.M.P. devised and directed the study. A.L.C., I.A.U., and S.N.S. wrote the manuscript. Competing financial interests: F.M.P. received research funding or consultancy fees from GlaxoSmithKline (GSK), Genentech, Roche, and Union Chimique Belge (UCB). Other authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Stay Connected to Science Immunology

Navigate This Article