IRF5 Is Required for Bacterial Clearance in Human M1-Polarized Macrophages, and IRF5 Immune-Mediated Disease Risk Variants Modulate This Outcome

Bacterial products stimulate pattern recognition receptors (PRRs) to secrete proinflammatory cytokines. These PRR responses must be tightly regulated during infection and on mucosal surfaces as excessive cytokine production can result in ongoing inflammation and tissue damage. As such, many gene variants in pathways modulating PRR-initiated responses are associated with immune-mediated diseases (1). Given the importance of balancing inflammation with effective bacterial clearance, variants regulating PRR-initiated outcomes resulting in decreased inflammation may be protective to immune-mediated diseases. Simultaneously, these gene variant–regulated reduced PRR outcomes may increase susceptibility to microbial-mediated diseases (2) as PRR-initiated pathways contribute to bacterial clearance. For example, SOCS1/LITAF/RMI2 region variants that confer decreased inflammatory bowel disease (IBD) risk (2) are associated with increased leprosy risk (3). Mouse models can demonstrate a similar dichotomy. For example, TNF-deficient mice are protected from LPS-induced lethality but are more susceptible to infection with Listeria monocytogenes or Mycobacterium tuberculosis (4, 5). Conversely, IRAKM^−/− mice show increased experimental systemic lupus erythematosus (6) but more effectively clear Pseudomonas aeruginosa relative to wild-type mice (7).

IFN regulatory factor 5 (IRF5) is critical for multiple PRR-induced functions, including cytokine secretion (8, 9), glycolysis (10), and myeloid cell–mediated T cell activation in a coculture system (9). IRF5 can regulate these outcomes both as a proximal signaling molecule (8, 10–12) and as a transcription factor (1, 8, 9, 13). Consistent with its critical role in PRR-initiated outcomes, noncoding IRF5 genetic variants resulting in increased IRF5 mRNA (14) and protein (10) expression are associated with multiple immune-mediated diseases characterized by increased inflammation, including systemic lupus erythematosus, IBD, rheumatoid arthritis, Sjögren's syndrome, primary biliary cirrhosis, systemic sclerosis, and multiple sclerosis (15). These IRF5 variants modulate PRR-induced cytokines in human myeloid-derived cells (16, 17). In part because of the common distribution of both the risk and nonrisk rs2004640 IRF5 variants and that IRF5 variants show one of the strongest expression quantitative trait loci in the genome (18), the rs2004640 polymorphism is a major determinant of the variance in interindividual PRR-induced cytokine secretion (16). PRR-initiated pathways can also regulate bacterial clearance in myeloid cells. However, limited studies have examined IRF5 in bacterial clearance. In mice, IRF5 upregulates NOS2 and Th1 responses following Leishmania donovani infection (19). Moreover, IRF5^−/− mice show decreased type I IFN induction upon Newcastle disease virus infection (20). Consistently, in vitro, IRF5 in mouse myeloid cells is required for optimal M. tuberculosis or Streptococcus pyogenes–mediated type I IFN induction (21–23). However, these studies did not examine the macrophage-intrinsic role and mechanisms for IRF5 in bacterial clearance. Furthermore, it is not known if IRF5 plays a role in bacterial clearance in human macrophages and if macrophages from disease-associated IRF5 genetic carriers show altered bacterial clearance. Mouse and human cell inflammatory outcomes can dramatically differ (24), such that it is critical to examine these questions in human cells. As such, many questions remain unanswered. Is there a human macrophage-intrinsic role for IRF5 in bacterial clearance? What are the mechanisms through which IRF5 mediates putative bacterial clearance in macrophages? How might this clearance be regulated by IRF5 in distinct macrophage subsets? Do immune-mediated disease-protective IRF5 genetic variants associated with decreased IRF5 expression and decreased PRR-induced cytokine secretion also show decreased bacterial clearance and mechanisms mediating this clearance? Can IRF5 expression modulation in IRF5 polymorphism carrier cells regulate these outcomes?

In this study, we found that IRF5 was required for optimal bacterial clearance in primary human monocyte-derived macrophages (MDMs) and particularly in PRR-stimulated M1-differentiated macrophages. Mechanisms mediating IRF5-dependent bacterial clearance in M1-differentiated macrophages included induction of members of the NADPH oxidase complex, NOS2, and autophagy pathways. Importantly, M1-polarized macrophages from rs2004640/rs2280714 GG/CC carriers protective for immune-mediated diseases and expressing lower IRF5 levels showed decreased bacterial clearance and decreased induction of reactive oxygen species (ROS), reactive nitrogen species (RNS), and autophagy pathways relative to TT/TT immune-mediated disease risk carriers. Therefore, we define how IRF5 and the immune disease–associated IRF5 variants regulate macrophage-intrinsic bacterial clearance and identify mechanisms leading to these outcomes. These findings highlight that the immune-disease risk variant expressing higher IRF5 shows a benefit with respect to macrophage-mediated clearance of bacteria and, conversely, that IRF5 variant carriers protected from immune-mediated diseases might be at increased risk for microbial diseases.

Informed consent was obtained per protocol approved by the institutional review board at Yale University. Healthy individuals were genotyped by TaqMan (Life Technologies, Grand Island, NY) or Sequenom Platform (Sequenom, San Diego, CA).

Human PBMCs were isolated from peripheral blood using Ficoll-Paque (Dharmacon, Lafayette, CO). Monocytes were purified from PBMCs by adhesion and tested for purity (>98% by CD11c expression). Monocytes were differentiated in M-CSF (10 ng/ml) (Shenandoah Biotechnology Systems, Warwick, PA) and culture conditions as in (10) for 7 d to generate MDMs. MDMs were treated for 24 h with 100 ng/ml Escherichia coli LPS (Sigma-Aldrich, St. Louis, MO) and 20 ng/ml IFN-γ (R&D Systems, Minneapolis, MI) for M1 polarization.

Unless otherwise indicated, 100 nM scrambled or siGENOME SMARTpool small interfering RNA (siRNA) against IRF5, ERK, p38, JNK, NEMO, Akt2, p40phox, p47phox, p67phox, NOS2, or ATG5 (Dharmacon, Lafayette, CO) (four pooled siRNAs for each gene) or 2 μg vectors expressing ATG5 [Addgene plasmid no. 24922; kindly deposited by T. Finkel (25)], NOS2 [generous gift of T. Eissa (26)], p47phox, p67phox [generous gifts of C. DerMardirossian (27)], IRF5 (GeneCopoeia, Rockville, MD), or empty vector was transfected into macrophages using Amaxa nucleofector technology (Amaxa, San Diego, CA). In the experiments that examined IRF5-dependent outcomes in M1-polarized macrophages, MDMs were first polarized into M1-differentiated macrophages for 24 h, transfected with 100 nM IRF5 siRNA for 48 h, and then treated with LPS for the indicated times (Fig. 1A), unless otherwise stated.

Proteins were detected in permeabilized cells by flow cytometry (gated on live cells) with Alexa Fluor 647, PE, Alexa Fluor 488, or biotin-labeled Abs to phospho-ERK (E10), phospho-p38 (3D7), phospho-JNK (G9), phospho-IκBα (14D4) (Cell Signaling Technology, Danvers, MA), or Abs to IRF5 (1H6) (Novus Biologicals, Littleton, CO), p40phox (B-8), p47phox (A-7), p67phox (D-6) (Santa Cruz Biotechnology, Santa Cruz, CA), NOS2, LC3II (Cell Signaling Technology), phospho-Akt2, or ATG5 (EPR1755) (Abcam, Cambridge, MA). Western blot was performed as per (10). Blots were incubated with IRF5 (E1N9G), GAPDH (6C5; MilliporeSigma), or the Abs above.

Total RNA was isolated using TRIzol reagent (Thermo Fisher Scientific). Quantitative PCR was performed using All-In-One qPCR Mix (GeneCopoeia) on the ABI PRISM 7000 (Applied Biosystems). Samples were normalized to GAPDH. Primer sequences are available upon request.

ROS was measured by flow cytometry using 10 μM 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA) (Invitrogen) following the manufacturer’s instructions or by NBT (Sigma-Aldrich) as in Ref. 28.

Macrophages were cultured in triplicate with Enterococcus faecalis, adherent-invasive E. coli (AIEC) strain LF82 (a generous gift from Dr. E. Mizoguchi) or Salmonella enterica serovar Typhimurium at 10:1 multiplicity of infection, or S. aureus at 1:1 multiplicity of infection for 1 h, washed with PBS, and incubated in HBSS with 50 μg/ml gentamicin for an additional hour. Cells were washed, lysed with 1% Triton X-100 (Sigma-Aldrich), and plated on MacConkey or Luria–Bertani agar.

Significance was assessed using two-tailed t test. A p < 0.05 was considered significant.

To examine the contribution of IRF5 in intracellular bacterial clearance in human macrophages, we first corroborated the role of IRF5 in the differentiation of select macrophage subsets and then examined how it regulates the enhanced bacterial clearance under select conditions. Macrophages consist of multiple subtypes depending on environmental conditions. In particular, proinflammatory M1 macrophages more effectively clear bacteria compared with nonpolarized macrophages (29). We confirmed that upon culturing MDMs under M1 differentiation conditions (generated by IFN-γ and LPS), M1 markers were increased relative to M0 macrophages (MDMs) (Supplemental Fig. 1A). M1 macrophages secrete more TLR4-induced proinflammatory and less anti-inflammatory cytokines relative to undifferentiated MDMs, and we confirmed this was the case (Supplemental Fig. 1B). We (10) and others (9, 30) have found that IRF5 is required for optimal M1 differentiation and that IRF5 expression is increased upon M1 differentiation. We confirmed this increased IRF5 expression by flow cytometry (Supplemental Fig. 1C) and used Western blot as an independent approach (Supplemental Fig. 1D). We also confirmed our previous finding (10) that IRF5 expression is lower in M2-differentiated macrophages relative to undifferentiated MDMs (Supplemental Fig. 1C). Consistent with the IRF5 expression regulation, we observed that LPS-induced IRF5 phosphorylation, which contributes to IRF5 downstream outcomes (13), was higher in M1-differentiated macrophages compared with MDMs and M2-differentiated macrophages (Supplemental Fig. 1E). To confirm that IRF5 regulates M1 differentiation, we used siRNA to reduce IRF5 expression and established effective IRF5 knockdown through both Western blot and flow cytometry in undifferentiated (M0) macrophages (Supplemental Fig. 1F, 1G) and in macrophages in M1-polarizing conditions (Supplemental Fig. 1H, 1I). We will examine IRF5 protein expression by flow cytometry in the studies that follow, given the accurate per cell normalization and quantitation advantages (31) and the consistent results between flow cytometry and Western blot in the preceding studies. We also confirmed that cell viability was unaffected following IRF5 knockdown (Supplemental Fig. 1J). As expected, IRF5 was required for optimal induction of M1 polarization markers (Supplemental Fig. 1A) and TLR4-induced M1-associated proinflammatory cytokines (Supplemental Fig. 1B). As a comparison, we examined M2 macrophages as IRF5 has not generally been associated with M2 polarization (9, 30). IRF5 was required for the upregulation of M2-associated chemokine (e.g., CCL18, CCL22, CCL23) and cytokine (e.g., IL10 and TGFB1) transcripts and TLR4-induced IL-10, IL-1Ra, and TGF-β protein secretion (Supplemental Fig. 1A, 1B) but not for M2-associated CD36, MAF, MRC, SLC38A6, or TGM2 transcripts (Supplemental Fig. 1A). This indicates that IRF5 deficiency does not universally impair macrophage outcomes. To further ensure that IRF5 knockdown does not result in global macrophage dysfunction, we measured LPS-induced Akt1 phosphorylation, an IRF5-independent event (10), and confirmed that it remained intact in IRF5-deficient MDMs (Supplemental Fig. 1K).

We (32) and others (33) have found that chronic PRR stimulation of macrophages increases antimicrobial mechanisms and bacterial clearance efficacy. Chronic PRR stimulation can be observed with ongoing infection or in environments with persistent bacterial exposure, such as at mucosal sites, including in intestinal tissues. Therefore, we examined clearance of the resident intestinal luminal bacteria E. faecalis in MDMs (M0), M1 macrophages, and M1 macrophages chronically treated with LPS (Fig. 1A). We confirmed that compared with M0 macrophages, M1 macrophages exhibited enhanced bacterial clearance, which was further enhanced by chronic LPS treatment (Fig. 1B). We then examined the role of IRF5 in bacterial clearance under these conditions using IRF5 knockdown. Given that IRF5 is required for optimal M1 polarization, we asked if IRF5 contributes to bacterial clearance independent of its role in M1 polarization. We therefore knocked down IRF5 following M1 polarization in the subsequent studies, unless otherwise indicated. IRF5 contributed to bacterial clearance in M0 macrophages, even more so in M1-differentiated macrophages, and particularly following chronic LPS treatment of M1-differentiated macrophages (Fig. 1B). We observed similar requirements with additional bacteria, including AIEC, which is enhanced in the ilea of Crohn's disease patients (34), the resident bacteria S. aureus, and the intestinal pathogen S. Typhimurium (Fig. 1B). We asked if the reduced LPS-enhanced bacterial clearance observed following IRF5 knockdown was due to decreased TLR4 (receptor recognizing LPS) surface expression and found that this was not the case (Supplemental Fig. 1L). Taken together, these results indicate that IRF5 is required for optimal intracellular bacterial clearance in unpolarized MDMs and in M1-differentiated macrophages, and its contribution is more pronounced in PRR-stimulated M1 macrophages.

We next assessed mechanisms through which IRF5 contributes to bacterial clearance in PRR-stimulated M1 macrophages. IRF5 can modulate myeloid cell outcomes both as a transcription factor (1, 8, 9, 13) and by directly regulating proximal signaling pathways (8, 10–12). We previously found that IRF5 is required for PRR-induced MAPK, NF-κB, and Akt2 signaling in nonpolarized macrophages as well as in M1-polarized macrophages when IRF5 is knocked down prior to M1 polarization (10), and that IRF5-dependent Akt2 activation is required for M1 polarization (10). To ensure that these signaling pathways were also IRF5-dependent in already differentiated M1 macrophages, we examined their activation when IRF5 is knocked down after M1 polarization. LPS-induced MAPK, NF-κB, and Akt2 activation was reduced under these conditions (Fig. 2A–C). Similar results were observed using an independent approach by Western blot (Fig. 2D).

We next asked if MAPK, NF-κB, and Akt2 pathways were required for the enhanced bacterial clearance observed upon LPS treatment of M1-polarized macrophages. We therefore first polarized MDMs to M1 macrophages and then knocked down expression of each of these signaling pathways. Knockdown of the MAPK, NF-κB, or Akt2 pathways following M1 polarization decreased LPS-enhanced bacterial clearance (Fig. 2E). The decrease was more pronounced when the pathways were knocked down in combination, indicating that they cooperate for this outcome (Fig. 2E). Taken together, these results indicate that IRF5 is required for LPS-induced activation of MAPK, NF-κB, and Akt2 pathways in M1-polarized macrophages, and these signaling pathways, in turn, are required for optimal bacterial clearance in LPS-treated, M1-differentiated macrophages.

We next investigated the mechanisms through which IRF5 mediates bacterial clearance in PRR-treated M1 macrophages. ROS and RNS can contribute to bacterial clearance, and they cooperate to maintain homeostasis in environments heavily populated by microbes, such as the intestine (35). Furthermore, ROS is induced with chronic PRR ligand exposure (36). To our knowledge, IRF5 regulation of ROS production has not been reported. IRF5 knockdown in LPS-treated M1 macrophages resulted in reduced ROS as assessed by two independent approaches (Fig. 3A, 3B). One mechanism for ROS production is through regulation of the NADPH oxidase complex. Chronic LPS treatment of M1 macrophages induced p40phox, p47phox, and p67phox mRNA transcripts and this induction was reduced upon IRF5 knockdown (Fig. 3C). In contrast, the p22phox was not regulated by IRF5, and LPS did not induce the gp91phox (Fig. 3C). IRF5 regulates NOS2, an RNS-producing enzyme, in the liver of L. donovani–infected mice, and IRF5 enhances NOS2 promoter activity in HEK293 cells (19). We found that LPS treatment of M1 human macrophages upregulated NOS2 but that IRF5-deficient M1 macrophages failed to induce NOS2 (Fig. 3D).

We next sought to confirm that the protein pathways encoded by the antimicrobial transcripts regulated by IRF5 (Fig. 3C, 3D) were required for optimal bacterial clearance in LPS-treated M1 macrophages. We confirmed silencing of p40phox, p47phox, p67phox, and NOS2 protein by two independent approaches: Western blot (Fig. 3E) and flow cytometry (Fig. 3F). We will use flow cytometry to examine these proteins in the experiments that follow. Each protein was required for optimal clearance of E. faecalis (Fig. 3G), AIEC, S. aureus, and S. Typhimurium (data not shown). To clearly establish that IRF5 contributes to bacterial clearance through upregulation of the implicated NADPH oxidase subunits and NOS2, we transfected p47phox and p67phox, the IRF5-dependent NADPH oxidase subunits that contributed most to bacterial clearance in Fig. 3G, into IRF5-deficient LPS-treated M1 macrophages to restore expression of these proteins to physiological levels (Fig. 3H). We confirmed that the complementation of p47phox and p67phox expression restored ROS production in LPS-treated M1 macrophages (Fig. 3I). We also complemented NOS2 expression to physiological levels in IRF5-deficient LPS-treated M1 macrophages (Fig. 3J). Restoration of each protein partially rescued bacterial clearance in IRF5-deficient LPS-treated M1 macrophages (Fig. 3K). Taken together, these results indicate that in LPS-treated M1 macrophages, IRF5 is required for optimal ROS production through the upregulation of NADPH oxidase subunits and for optimal induction of the RNS-producing enzyme NOS2, thereby resulting in more efficient bacterial clearance.

Autophagy is another key bacterial clearance mechanism regulated through PRR stimulation (1). One report showed that IRF5 impairs hepatitis C virus–induced autophagy in MH-14– and C-5B–transfected cell lines (37); impaired autophagy could reduce effective bacterial clearance. However, how IRF5 regulates autophagy following PRR stimulation in human macrophages has not been reported. We found that IRF5 was, in fact, required for optimal induction of the autophagy marker LC3II upon LPS treatment of M1 macrophages per both Western blot (Fig. 4A) and intracellular flow cytometry (Fig. 4B). To define mechanisms wherein IRF5 enhances autophagy, we examined the expression of the autophagy-related genes ATG5, ATG7, and IRGM in M1-polarized macrophages followed by IRF5 knockdown. ATG5 expression increased upon LPS treatment of M1 macrophages, and optimal ATG5 expression induction required IRF5 (Fig. 4C). In contrast, LPS-induced ATG7 and IRGM did not require IRF5 (Fig. 4C). To determine if ATG5 contributes to intracellular bacterial clearance in M1 macrophages, we silenced ATG5 and confirmed reduced ATG5 protein expression through two independent approaches, Western blot and flow cytometry (Fig. 4D, 4E). We ensured that ATG5 knockdown decreased the expression of the autophagy marker LC3II in LPS-treated M1 macrophages as assessed by both Western blot and flow cytometry (Fig. 4F, 4G). We next ensured that ATG5 was required for optimal E. faecalis clearance in human LPS-treated M1 macrophages (Fig. 4H). Finally, complementing ATG5 expression in IRF5-deficient, LPS-treated M1 macrophages to physiological levels (Fig. 4I) restored LC3II expression (Fig. 4J) and partially rescued E. faecalis clearance (Fig. 4K). Taken together, IRF5 is required for optimal upregulation of ATG5 expression, which in turn regulates autophagy pathways and bacterial clearance in human M1 macrophages.

Complementing each of the IRF5-dependent antimicrobial pathways alone only partially rescued bacterial clearance in IRF5-deficient macrophages (Figs. 3K, 4K). We therefore asked if IRF5-dependent p47phox, p67phox, NOS2, and ATG5 upregulation cooperate to mediate optimal bacterial clearance in LPS-treated M1 macrophages. Clearance of each E. faecalis, S. aureus, AIEC, and S. Typhimurium was most impaired when all the pathways were disrupted in combination (Fig. 5A). Consistently, restoring expression of the IRF5-dependent proteins in combination in LPS-treated M1 macrophages deficient in IRF5 more fully rescued the impaired bacterial clearance relative to that observed with restoration of each protein alone (Fig. 5B). Furthermore, and consistent with the requirement for the IRF5-dependent MAPK, NF-κB, and Akt2 pathways in optimal bacterial clearance in LPS-treated M1 macrophages, MAPK and NF-κB pathway knockdown or Akt2 knockdown in M1-differentiated macrophages reduced the LPS-induced ROS and p47phox, p67phox pathways, NOS2 expression, and LC3II and ATG5 pathways observed in M1 macrophages (Supplemental Fig. 2A–E). Therefore, IRF5 regulates intracellular bacterial clearance in human macrophages through at least three mechanisms: ROS, RNS, and autophagy pathway induction, which in turn are regulated by IRF5-dependent MAPK, NF-κB, and Akt2 signaling. Importantly, these IRF5-dependent antimicrobial mechanisms cooperate to mediate bacterial clearance in LPS-treated M1 macrophages.

To ensure that IRF5 was required for PRR-induced outcomes in macrophages differentiated to an M1-like phenotype through another approach, we polarized MDMs with IFN-γ (38) and then knocked down IRF5 and treated cells with LPS. IRF5 was required for optimal levels of bacterial clearance (Supplemental Fig. 2F), signaling (Supplemental Fig. 2G–I), and bacterial clearance mechanisms (Supplemental Fig. 2J) under these conditions. We also verified that IRF5-dependent MAPK, NF-κB, and Akt2 signaling pathways were required for bacterial clearance in these IFN-γ–polarized macrophages (Supplemental Fig. 2K). Taken together, IRF5 deficiency in IFN-γ–polarized macrophages results in decreased PRR-induced bacterial clearance–related outcomes.

Under physiological conditions, cells from IRF5 immune-mediated disease protective genetic variant carriers exhibit reduced IRF5 expression throughout the immune response, including before macrophage polarization. We therefore wanted to confirm that the most important outcomes we had defined also demonstrated dependency on IRF5 when IRF5 was knocked down prior to M1 polarization. We first assessed if the IRF5-dependent signaling pathways we had identified were required for M1 differentiation. We effectively knocked down intermediates essential for MAPK, NF-κB, and Akt2 pathway activation, as evidenced by lack of activation of the respective pathways (Supplemental Fig. 3A–C), and ensured that cell viability was intact (Supplemental Fig. 3D). MAPK and NF-κB pathways were essential for human M1 macrophage polarization (Supplemental Fig. 3E) and in fact, contributed to M1 polarization to a similar degree as did the Akt2 pathway (Supplemental Fig. 3E), which we had previously identified to be required for optimal M1 polarization (10). We then examined bacterial clearance pathways and observed that PRR-induced bacterial clearance (Supplemental Fig. 3F), activation of signaling pathways (Supplemental Fig. 3G–I), and bacterial clearance mechanisms (Supplemental Fig. 3J–O) were dependent on IRF5 under these conditions. Furthermore, restoration of ROS, RNS, and autophagy pathways in cells where IRF5 was knocked down prior to M1 polarization restored bacterial clearance (Supplemental Fig. 3P). Therefore, IRF5-dependent signaling pathways are required for M1 differentiation, and IRF5 deficiency prior to M1 differentiation results in defects in both M1 polarization and the induction of antibacterial clearance mechanisms observed in M1 macrophages.

Given our results that IRF5 is required for bacterial clearance and relevant mechanisms in human macrophages, we hypothesized that immune-mediated disease-protective rs2004640/rs2280714 GG/CC carrier macrophages that show reduced IRF5 expression (10, 14), which we confirmed in both undifferentiated MDMs (Fig. 6A) and in macrophages following M1 polarization and LPS stimulation (Fig. 6F), would show decreased bacterial clearance. LPS-treated undifferentiated MDMs and M1-differentiated macrophages from these carriers showed decreased E. faecalis and S. Typhimurium clearance relative to TT/TT carriers, with GT/CT heterozygotes showing an intermediate phenotype (Fig. 6B, 6G). Consistently, GG/CC carrier LPS-treated undifferentiated MDMs and M1 macrophages showed decreased induction of each of the IRF5-dependent mechanisms we had identified, including ROS and p47phox and p67phox (Fig. 6C, 6H), NOS2 (Fig. 6D, 6I), and LC3II and ATG5 (Fig. 6E, 6J) pathways. Taken together, these data indicate that rs2004640/rs2280714 GG/CC low IRF5-expressing undifferentiated MDMs and M1 macrophages show decreased bacterial clearance and bacterial clearance mechanisms relative to TT/TT immune-mediated disease-risk carrier cells.

We next assessed if modulation in IRF5 expression levels by rs2004640/rs2280714 accounts for the genotype-dependent regulation of bacterial clearance. We used a knockdown approach to decrease IRF5 levels in macrophages from high IRF5-expressing rs2004640/rs2280714 TT/TT carriers to the levels observed in GG/CC carriers (Fig. 7A). This reduced bacterial clearance (Fig. 7B) and bacterial clearance mechanisms (Fig. 7C–E) to levels comparable to GG/CC carrier LPS-treated M1 macrophages. We also used a complementary approach and transfected an IRF5-expressing vector in low IRF5-expressing GG/CC carrier macrophages to increase IRF5 expression to levels observed in TT/TT carriers (Fig. 7F). This increased bacterial clearance (Fig. 7G) and relevant mechanisms (Fig. 7H–J) to levels comparable to those observed in TT/TT carrier LPS-treated M1 macrophages. Taken together, the modulation in IRF5 expression levels by the rs2004640/rs2280714 variant accounts for the differences in IRF5-dependent bacterial clearance mechanisms across the rs2004640/rs2280714 genotypes in LPS-treated M1 macrophages.

The immune system must effectively balance handling microbial infections while simultaneously controlling the resulting proinflammatory responses. Consistently, polymorphisms resulting in less microbial-induced inflammation can reduce risk of immune-mediated diseases but simultaneously be detrimental to optimal clearance of infectious challenges. In this study, we find that IRF5, a protein required for PRR-induced cytokines (8, 9) and associated with multiple immune-mediated diseases (15), is also required for optimal bacterial clearance in human M1-differentiated macrophages (Supplemental Fig. 4). We identify mechanisms through which IRF5 contributes to bacterial clearance, including ROS production mediated by p40phox, p47phox, p67phox upregulation, induction of NOS2, and induction of autophagy pathways mediated through ATG5 upregulation. We find that these pathways cooperate for optimal bacterial clearance. Furthermore, the induction of antimicrobial pathways and bacterial clearance in LPS-treated M1 macrophages required IRF5-dependent MAPK, NF-κB, and Akt2 pathways. Importantly, bacterial clearance and the identified IRF5-dependent clearance mechanisms were modulated by IRF5 genotype, which regulates IRF5 expression, thereby highlighting that carriers of these low-expressing IRF5 polymorphisms might be at increased risk for bacterial infection.

Whereas IRF5 regulates M1 macrophage polarization (9, 10, 30) and has not generally been associated with M2 polarization (9, 30), we identify that IRF5 is required for certain M2-associated chemokines and cytokines in human macrophages, whereas other M2 markers are IRF5-independent (Supplemental Fig. 1A, 1B). A few studies suggest that IRF5 might be required for aspects of M2 polarization. They include a report that IRF5 is upregulated in M2 human macrophages (39) and another describing a requirement for IRF5 in secretion of IL-10, an M2-associated cytokine, in PRR-stimulated mouse bone marrow–derived macrophages (40), which we also observe in human MDMs (16). A reverse regulation of IL-10 by IRF5 was observed in another study of human macrophages (9). These discrepancies might be due to differences in conditions used for macrophage polarization, differences in concentrations, origins, and the duration of the PRR stimuli used, or other differences in experimental conditions. The differential IRF5 dependency in select M2 macrophage outcomes may reflect that M2 polarization consists of a spectrum of subtypes, arising as an outcome of different stimulation conditions activating distinct signaling pathways (41). For example, endotoxin tolerance upregulates some (e.g., CCL22) but not all (e.g., CD206) M2 markers in human myeloid cells (42). Importantly, we found that IRF5 was required for both LPS-induced M1- and M2-associated cytokines (Supplemental Fig. 1B), highlighting that IRF5 has a central role in PRR-induced cytokines regardless of macrophage polarization. With respect to the mechanisms through which IRF5 regulates M1 polarization, we had previously identified that IRF5-dependent Akt2 activation was required for the enhanced glycolysis in M1 macrophages (10). We now find that IRF5-dependent MAPK and NF-κB signaling is also required for optimal M1 polarization (Supplemental Fig. 3E). Consistent with its contribution to proximal signaling pathways in human macrophages, IRF5 is phosphorylated within 15 min of LPS treatment (Supplemental Fig. 1E). The NF-κB pathway is required for M1 polarization of mouse macrophages (43), but to our knowledge, its requirement for human M1 polarization has not yet been reported. The requirement for IRF5 in MAPK and NF-κB signaling has demonstrated variable results; IRF5 did not contribute to activation of these pathways in mouse B cells (8), yet it was required for these pathways in human macrophages (10) and interacted with intermediates activating these pathways in 293T cell lines and mouse and human macrophages (10–12). These differences might be partly dependent on cell type, species, and stimuli.

We identify that IRF5 and IRF5 polymorphisms regulate autophagy (Figs. 4–7). As autophagy mediates not only microbial clearance but numerous other pathways important for inflammatory and infectious diseases [e.g., cellular stress, apoptosis, and inflammasome regulation (44)], IRF5 may play a role in these pathways as well. Moreover, we observe that autophagy, ROS, and RNS pathways cooperate to mediate the IRF5-dependent role in bacterial clearance. These findings are consistent with the observation that mice with combined deficiency in NADPH oxidase and NOS2 show worse bacterial clearance in vivo compared with single knockout mice (35). Importantly, we found that complementing each pathway alone in IRF5-deficient M1 macrophages only partially restores bacterial clearance, whereas combined complementation optimizes clearance in IRF5-deficient cells (Fig. 5). Such regulation of broad processes by IRF5 may be in part due to the ability of IRF5 to affect multiple proximal signaling pathways, including MAPKs, NF-κB, and Akt2 pathways, and multiple transcriptional programs (8, 10–12). Importantly, IRF5 knockdown did not globally affect macrophage function as Akt1 signaling (Supplemental Fig. 1K), and select M2-associated markers (Supplemental Fig. 1A) were intact. Although macrophages can demonstrate a spectrum of differentiation conditions, we used common IFN-γ– and LPS-conditioned macrophages and then exposed the cells to additional LPS treatment to simulate conditions of ongoing microbial exposure. As such conditions may not be universally encountered, we also examined and observed the dependency on IRF5 for microbial clearance in undifferentiated macrophages and in macrophages differentiated with IFN-γ alone (Supplemental Fig. 2F–K). Given its ability to regulate multiple bacterial clearance pathways simultaneously, IRF5 might represent an effective therapeutic target when treating infectious diseases.

We previously found that IRF5 is a major genetic determinant of the interindividual variance in PRR-induced cytokine secretion (16). This is in part due to the similar frequencies of the IRF5 rs2004640 risk and nonrisk alleles, such that the three genotypes are commonly distributed across the population. In this study, we find that IRF5 immune-mediated disease-risk variants are associated with increased bacterial-induced ROS, RNS, and autophagy-associated proteins and increased macrophage-mediated clearance of multiple bacteria. Conversely, this would imply that the IRF5 variants protective for immune-mediated diseases may be disadvantageous in select infectious diseases. Whether common genetic variants confer a slight increase in susceptibility to transient bacterial infections can be difficult to assess and track (45) and is therefore not as well studied as how these variants affect chronic immune-mediated diseases or chronic infectious diseases (e.g., M. tuberculosis). However, that PRR-associated signaling intermediates affect acute bacterial clearance is evidenced by studies involving individuals harboring dramatic loss-of-function coding mutations, resulting in increased frequency of infections, as is observed in MYD88 and IRAK4 mutation carriers who show increased susceptibility to pyogenic infections (46). Furthermore, uncommon mutations significantly reducing ROS production and autophagy, pathways that we identify to be regulated by IRF5, are associated with increased risk for infectious diseases (2, 47). The dichotomy in regulation of inflammatory versus infectious disease susceptibility has been observed for SOCS1/LITAF/RMI2 and IL12B region polymorphisms associated with IBD (2) but protective for leprosy (3, 48). IL12B polymorphisms are associated with risk for M. leprae, M. tuberculosis, and S. enteritidis infections (49). IRF5 regulates multiple other cytokines in addition to IL-12 (Supplemental Fig. 1B), suggesting that variants in IRF5 might affect the balance between inflammatory and infectious diseases through additional pathways. Our findings identify critical functions for IRF5 and IRF5 immune-mediated disease variants in immunological processes critical for bacterial clearance. These findings advance our understanding of the emerging roles of IRF5 variants in the balance of immune- and microbial-mediated diseases and indicate that when considering targeting this pathway for immune-mediated diseases, caution might be needed because of potential risk for microbial infections.

We thank Tony Eissa, Celine DerMardirossian, and Emiko Mizoguchi for reagents and the blood donors for participation.

The authors have no financial conflicts of interest.