Interferon regulatory factor signaling in autoimmune disease

Abstract

Interferon regulatory factors (IRFs) play critical roles in pathogen-induced innate immune responses and the subsequent induction of adaptive immune response. Dysregulation of IRF signaling is therefore thought to contribute to autoimmune disease pathogenesis. Indeed, numerous murine in vivo studies have documented protection from or enhanced susceptibility to particular autoimmune diseases in Irf-deficient mice. What has been lacking, however, is replication of these in vivo observations in primary immune cells from patients with autoimmune disease. These types of studies are essential as the majority of in vivo data support a protective role for IRFs in Irf-deficient mice, yet IRFs are often found to be overexpressed in patient immune cells. A significant body of work is beginning to emerge from both of these areas of study – mouse and human.

Interferon regulatory factors (IRFs) are a family of transcription factors that were first identified as regulators of virus-induced type I interferon (IFNA and B) gene expression [1]. With the later discovery of other IFN families (type II and III), most of the IRFs, with the exception of IRF6, have been implicated in their regulation [2]. Subsequent findings revealed that IRFs play important roles in the regulation of both innate and adaptive immune responses. In particular, IRFs have been shown to be involved in the activation and differentiation of distinct immune cell populations [3, 4]. To date, ten IRFs have been discovered (IRF1-10) in vertebrates; however, some are rendered inactive or eliminated in mice and humans, such as IRF10 [5]. A number of structural features within IRFs are well conserved. All IRF family members share a highly conserved N-terminal DNA-binding domain of approximately 120 amino acids, which binds to the core recognition sequence, GAAANNGAAAG/CT/C, termed the IFN stimulated response element (ISRE) [6]. With the exception of IRF1 and IRF2 that contain a PEST (proline-, glutamic acid-, serine-, and threonine-rich) domain for protein-protein interactions, all other family members contain a non-homologous C-terminal IRF association domain (IAD) for interaction with other family members and/or protein interacting partners [7, 8].

In order to initiate signaling through the IRFs, many of them need to be “activated” by a signal that leads to post-translational modification, followed by altered cellular localization. The types of modifications shown to alter IRF activation/function include phosphorylation, acetylation, ubiquitination, and sumoylation [9]. The signaling cascades that lead to IRF activation are generally downstream of pathogen recognition receptors (PRRs). Identification of PRRs that lead to IFN gene expression has been the subject of intense research in the past decade. Notably, many of the cellular sensors shown to be responsible for pathogen recognition require IRF activation and downstream signaling. Cellular sensors shown to be involved in IRF signaling are: membrane bound Toll-like receptors (TLR), cytoplasmic RIG-like receptors (RLR), nucleotide binding oligomerization domain (NOD)-like receptors, and a disparate family of cytoplasmic and possibly nuclear DNA sensors [1, 10-12]. Activation of IRFs through many of these pathways lead to the expression of IFNs, as well as other inflammation-associated cytokines [13].

The most studied pathway(s) that leads to IRF activation is through the TLRs. TLR signaling can be divided into two pathways: one that is dependent on the adapter protein MyD88 (myeloid differentiation primary response), and the other that is MYD88-independent and requires the adapter protein TRIF (Toll/IL-1 receptor domain-containing adaptor inducing IFN-B). Depending on the IRF, one or both pathways may lead to activation, resulting in its binding to the promoters of target genes and modulation of immune responses [1, 3, 10]. The steps leading to IRF activation and function can be broadly summarized into five major stages: 1) signal, 2) post-translational modification, 3) dimer formation, 4) nuclear translocation, 5) regulation of target gene expression.

With ongoing advances in the field of genome-wide association studies (GWAS), genetic variations in IRFs have been identified and associated with numerous autoimmune diseases. These genetic associations have opened up an expansive field of research focused on determining the role of IRFs in autoimmune disease pathogenesis. Autoimmune diseases are characterized by an attack on one’s “self”. The attack can be either specific to a particular organ, as is the case in inflammatory bowel disease (IBD), or systemic, as seen in systemic lupus erythematosus (SLE), rheumatoid arthritis (RA) and Sjögren’s Syndrome (SS). A number of genetic variants or single nucleotide polymorphisms (SNPs) within IRF genes have been detected at higher rates in patients with autoimmune diseases than matched healthy controls that enable an association with risk or protection from disease. Although this review will not cover the in-depth details from GWAS since a number of excellent reviews already exist [14-18], findings from GWAS are what initiated this new field of study - IRFs in autoimmunity. We instead provide a concise summary of in vivo and in vitro data that implicate dysregulated IRF signaling in autoimmune disease pathogenesis with a specific focus on the common and/or distinct pathways leading to and from the IRFs. Given the critical role of IRFs in IFN gene regulation, it was first postulated that genetic variants would contribute to elevated type I IFN expression that is now detected in multiple autoimmune diseases [19]. Based on emerging data, however, we propose that IRF function (or dysfunction) in autoimmune diseases is more robust than just as regulators of the IFNs.

IRF genetic association with human autoimmune diseases

Below we summarize findings from GWAS that identified genetic variation within the IRFs that associate with either susceptibility to or protection from a particular autoimmune disease. Only IRFs in which genetic variation has been associated with autoimmune diseases are detailed here. It is this body of work that initiated a new field of research focused on understanding the pathogenic role of IRFs in autoimmunity.

IRF1

IRF1 regulates the transcription of genes that play essential roles in viral infection, tumor immune surveillance, pro-inflammatory injury, and immunity system development [20]. With regard to GWAS, little is currently known of IRF1 polymorphisms and risk of autoimmune diseases. Initial data found differences in IRF1 polymorphisms between healthy controls and patients with juvenile idiopathic arthritis (JIA) [21] and Behcet’s disease (BD) [22]. In BD, IRF1 polymorphisms associated with risk in women and patients with thrombosis [22]. In JIA, however, a subsequent study examining a larger cohort of patients with different control subjects contradicted the early associations [23]. IRF1 polymorphisms have been identified in multiple sclerosis (MS), yet more replicative studies are needed [24]. In RA, SNPs within the introns and untranslated regions (UTRs) of IRF1 were found to associate with risk in black South African patients [25]. Copy number variations were also identified in IRF1 that associated with RA [26]. Continued replication of GWAS in multiple ethnicities and cohorts will be required to confirm associations of IRF1 genetic variants in JIA, BD, MS and RA.

IRF2

IRF2 negatively regulates type I IFN responses and plays a role in the induction of Th1 differentiation [8, 27]. While not well replicated, association of IRF2 genetic variants with susceptibility to SLE has been shown, and the risk haplotype was suggested to be associated with transcriptional activation of IRF2 [28, 29].

IRF3

IRF3 is constitutively expressed in all cell types and is primarily known for its regulation of type I IFNs in response to pathogens (reviewed in [2]). Similar to IRF1, few replicative association studies have been performed to link IRF3 genetic variants to autoimmune diseases. The most studied has been IRF3 polymorphisms in SLE, yet little agreement has been reached [8, 9, 30]. In a Mexican cohort, the IRF3 SNP rs2304206 associated with increased susceptibility to SLE and elevated serum type I IFN levels in patients positive for dsDNA autoantibodies [31]; however, this finding has yet to be replicated. Surprisingly, the IRF3 rs2304204 GG and AG genotypes were found to confer decreased risk of SLE [31]. SNPs −925A/G and −776C/T in the promoter region of IRF3 showed significant association with SLE risk while the GG genotype of −925A/G was protective [30]. Additional replicative studies will be required to confirm associations.

IRF4

Distinct from other IRF family members, IRF4 expression is restricted to immune cells, and can either function as a transcriptional activator or repressor depending on its interacting partner(s) [32]. IRF4 is induced upon the activation of T and B cells. In CD4+ T helper cells (T_H), IRF4 plays an essential role in the regulation of IL21 production, whereas in B cells it controls class switch recombination and plasma cell differentiation [33]. The IRF4 SNP rs9378815 has been reported to associate with risk of RA in European and Asian ancestries [34, 35]. IRF4 was also identified as a common susceptibility locus for systemic sclerosis (SSc) [35]. No association was found between IRF4 SNP rs872071 and SLE susceptibility a Han Chinese [36]. IRF4 continues to be examined for genetic variation associated with autoimmune diseases.

IRF5

IRF5 has been demonstrated to play a key role in modulating innate and adaptive immune responses in numerous cell types, including dendritic cells, monocytes/macrophages, and B cells. IRF5 is regarded as the “master regulator of proinflammatory cytokines” [11]. In addition to type I IFN production, IRF5 is responsible for upregulating expression of IL6, IL12b, IL17, IL23, IFNβ, IP10, MCP1, and RANTES [37, 38]. Findings from GWAS have shown that IRF5 genetic variants impact a broad range of autoimmune diseases. In the case of SLE, numerous GWAS have supported associations between IRF5 SNPs and functional risk variants in the IRF5 gene across multiple ancestral backgrounds [39-45]. IRF5 risk variants have also been identified in other autoimmune diseases such as RA [46, 47], SSc [40, 48], MS [49, 50], and IBD [51] suggesting a common pathogenic role for IRF5 in autoimmune diseases. Of note, two SNPs - rs10954213 in the 3’UTR and rs2004640 in the 5’-UTR - are shared amongst the most commonly associated with risk of any autoimmune diseases. Importantly, however, not all IRF5 SNPs are associated with risk. Some SNPs have been shown to be protective. In RA, the T allele of SNP rs2280714, C allele of rs729302 and G allele of rs2004640 associated with protection [52].

IRF7

IRF7 expression is not restricted to immune cells but nor is it ubiquitously expressed. It is best known as a master regulator of type I IFN gene expression and IFN-dependent innate immune responses [53]. Recent findings from GWAS identified the IRF7 risk SNP rs1131665 in SLE patients from Asian, European American, and African American populations [54, 55]. Similar to IRF5, genetic variants in IRF7 are associated with increased levels of IFNs in SLE patients [1]. The IRF7 variant rs1131665 is associated with SLE [54]. Genetic variants of IRF7 may act as risk factors for SLE due SNPs in the IRF7/PHRF1 locus [56]. Additionally, IRF7 has been implicated in the pathogenesis of metabolic autoimmune disorders such as diabetes [57].

IRF8

IRF8 is a key transcription factor for myeloid cell differentiation [58]. With regard to autoimmune diseases, two genetic variants in the IRF8 gene have been associated with BD in Han Chinese [59]. Increased IRF8 mRNA expression and IFNγ production, together with a concomitant decrease in IL10 production, is described in individuals carrying the rs11642873/AA and rs17445836/GG genotype [59]. Two independent studies identified additional SNPs in IRF8 that associated with risk of SLE [60, 61]. SNP rs17445836 was also found to associate with susceptibility to MS and SLE. In SLE, the risk SNP associated with anti-dsDNA antibodies and serum IFNα levels in patients of both African-American and European ancestry [62].

Dysregulation of IRF expression and/or signaling in human autoimmune diseases

In this section, we report on predominant findings of IRF function and signaling in some of the more common human autoimmune diseases. IRFs that have not yet been functionally characterized in these diseases will be excluded from the discussion. Figures 1 and 2 depict a summary of findings from this section.

Figure 1. Implications of IRF signaling in human autoimmune diseases.

Schematic summarizing the effect(s) of IRF signaling in specific cell types, in response to specific stimuli, and the specific autoimmune disease it has been implicated in. Gene expression affected by individual IRFs is shown by matched colored font for each IRF. For example, the IRF1 molecule is shown in blue and gene expression affected by IRF1 is shown in blue. Overlapping functions for the IRFs are written in black font. Diseases affected by the IRFs are shown at the bottom surrounded by green. MS, multiple sclerosis; SS, Sjögren’s syndrome; RA, rheumatoid arthritis; IBD, inflammatory bowel disease.

Figure 2. Cell type-specific IRF signaling in human SLE.

Schematic summarizing the role of individual IRFs in SLE pathogenesis. The primary downstream effect of all indicated IRFs is on type I IFN expression.

Multiple Sclerosis (MS)

MS is an inflammatory demyelinating disease of the central nervous system (CNS), which is characterized by an autoimmune inflammatory reaction against CNS myelin. IRF1 signaling in microglial cells has been shown to be involved in the pathogenesis of MS [24, 63]. IRF1 SNPs have been associated with progressive MS [24, 63] and IRF1-regulated genes, such as MHC class I (MHCI), TNF receptor (TNFR) and caspase 1, have been shown to be elevated in MS patients. In glial cells, IRF1 was found to cooperate with NFkB for the regulation of MHCI and inducible nitric oxide synthase (iNOS) expression leading to cytokine synergism [24, 63-65]. Glial cells were found to be the primary CNS cell with IRF1 expression in demyelinating lesions of MS [66]. IRF1 has been also shown to control the expression of two pro-apoptotic genes, caspase 1 and double-stranded RNA-dependent protein kinase (PKR) that play important roles in the process of inflammatory demyelination. Later studies implicated IRF1 and signal transducer and transcription activator 1 (STAT1) in IFNγ-induced oligodendrocyte progenitor cell death [66]. IRF8 is the only other family member that has been examined in the context of MS. Similar to IRF1, IRF8 was shown to mediate IFNγ signaling leading to oligodendroglial progenitor cell apoptosis [67].

Sjögren ’s syndrome (SS)

SS is a systemic autoimmune disease involving multiple organs. It is characterized by extensive dryness, extreme fatigue, chronic pain, neuropathies and lymphomas. It is called primary SS (pSS) when it occurs alone or secondary SS when it is in association with other autoimmune diseases like SLE and RA. SS is further characterized by infiltration of T_H cells and a T_H1/T_H2 imbalance has been shown to play a role in SS pathogenesis [67-74]. Like SLE, an IFN gene signature has been shown to be associated with SS and the strength of the IFN signature positively correlated with disease severity. IRF1 plays an important role in T_H differentiation and is known to regulate gene expression of factors involved in T_H1 differentiation. IRF1, along with IFNγ, is shown to induce and expand T_H1 responses [8, 75]. A higher expression of IRF1 was detected in peripheral blood of patients with primary SS along with more IRF1 positive cells in epithelial islands, lymphocytes, and ductal epithelial cells of the parotid glands compared to healthy control subjects [74]. Viral infections such as Epstein Barr Virus and chronic hepatitis infections have been known to be the initial inducers of pSS and initiate the production of IFNs. Plasmacytoid dendritic cells (PDCs) are considered to be a primary source of IFNα in pSS. IRF1 may also enhance and prolong the immune reaction leading to autoimmunity. A recent report associated B cell activating factor (BAFF) with pSS and suggested that the expression of BAFF is regulated by IRFs. BAFF is a cytokine required for the proper development and selection of B cells. In humans, BAFF is induced by type I IFNs [76]. In patients with pSS, BAFF expression in neutrophils positively correlated with type I IFN activity. BAFF is induced by type I IFN signaling via the transcription factors IRF1 and IRF2 whereas IRF4 and IRF8 were shown to negatively regulate BAFF expression [76]. Interestingly, SS patients carrying 4 repeats of the IRF5 CGGGG promoter insertion/deletion (indel) were found to have a high level of IRF5 mRNA in PBMCs and salivary gland epithelial cells after in vitro viral infection. SS patients with high levels of IRF5 mRNA also had high levels of mRNA for ISGs in PBMCs [77]. The mechanism(s) or pathway leading to upregulated IRF5 and ISG expression in SS is not known.

Rheumatoid Arthritis (RA)

RA is a systemic autoimmune disease that is characterized by expansion of synovial tissue, chronic inflammation and progressive destruction of a joint. Several immune cell subsets, including macrophages, monocytes, neutrophils, dendritic cells, natural killer cells, T cells, and B cells have been shown to play a role in RA pathogenesis [78]. TNFα, IL1β and IL6 are the major proinflammatory cytokines that play a prominent role in RA progression. Fibroblast-like synoviocytes (FLS) are the major constituent of synovial hyperplasia and the FLS-dependent effector molecules are implicated as important mediators of RA [78]. FLS play a critical role in invading cartilage and bone tissue [78]. IRF3 has been shown to play a major role in the regulation of ISGs in RA [79]. High levels of phosphorylated IRF3 were detected in RA synovial tissue. When FLS were transfected with siRNAs targeting IRF3 and then stimulated with poly (I-C) in culture, IRF3 was shown to regulate ISRE promoter activity as well as IFNβ, IRF5, IRF7, RANTES, IFN-inducible protein-10, MCP-1, and MIP1α gene expression [79]. Moreover, IRF3 knockdown blocked expression of metalloproteinase 3 (MMP3), MMP9, IL6, and IL8. A significant decrease in AP-1 binding of activated c-Jun in IRF3 deficient cells was found. This study further demonstrated that phosphorylation of IRF3 was regulated by IKK-related kinase IKKε [79].

IL18 is another cytokine that plays an important role in the pathogenesis of RA [80]. IL18 promotes T cells to produce T_H1 type cytokines and helps in the differentiation of T_H1 cells. IL18 also regulates the cytotoxic activity of natural killer cells by upregulating Fas ligand [81-87] and promotes angiogenesis [88-91], all of which contribute to RA pathogenesis. The biologic activity of IL18 is controlled by IL18 binding protein (IL-18BP). There is an IRF1 binding site in the IL18BP promoter [92]. Using fibroblasts isolated from synovium of RA patients who had undergone total joint replacement or synovectomy, an in vitro knockdown system was utilized to analyze the effect of TNFα-induced IL18BP expression. TNFα was shown to induce IRF1 nuclear translocation and regulate IL18BP expression; however, when NFκB or JNK2 signaling pathways were blocked, IRF1 nuclear translocation was reduced. Further, it was shown that IRF1 forms a complex with NFκB and cJun in the nucleus [93].

With regard to other family members, IRF4 and IRF5 are expressed in synovial fluid and synovium from RA patients. Anti-citrullinated protein antibodies (ACPAs) were found to induce IRF4 and IRF5 protein expression. IRF5 siRNA knockdown impaired ACPA activity significantly while ACPAs induced IRF5 activity and led to M1 macrophage polarization [94]. Additional studies will be required to determine the pathological contribution of elevated IRF4 or IRF5 expression in RA pathogenesis.

Behcet’s disease (BD)

BD is a systemic vasculitis disease with unknown etiology. It is characterized by recurrent attacks of oral aphthous ulcers, genital ulcers, and ocular uveitis. Altered T cell responses are associated with BD [22]. While IRF1 and IRF8 genetic variants are associated with BD, no functional studies have been performed yet.

Inflammatory Bowel Disease (IBD)

There are two main forms of IBD: Crohn’s disease and ulcerative colitis. Both serum and mucosal autoantibodies against intestinal cells and perinuclear anti-neutrophil cytoplasmic antibodies (pANCA) have been detected and found to associate with disease activity [95]. IRF1 expression was initially detected as increased in the nuclei of lamina propria mononuclear cells in colonic specimens from pediatric patients with active Crohn's disease [96]. Later studies, however, detected lower levels of IRF1 in patients who were undergoing standard medical treatment compared to untreated patients [96]. A study examining intestinal epithelial cells isolated from freshly obtained colonic mucosal biopsies revealed another role for IRF1. They found that intestinal epithelial cells expressed CARD4/NOD1 at both mRNA and protein levels [97]. CARD4/NOD1 transcription was induced by IFNγ through IRF1. An IRF1 binding motif was identified within CARD4/NOD1 and found to be necessary for the IFNγ effect [97].

IL7 plays a critical role in the regulation of mucosal immunity and has pleiotropic functions in the intestinal immune system. Intestinal epithelial cells produce IL7, which induces the proliferation of lamina propria lymphocytes and intraepithelial lymphocytes [98, 99]. IL7 also increases the production of cytokines by lamina propria lymphocytes in humans [100]. IRF1 and IRF2 are expressed in normal human colonic epithelial cells and upregulate IFNγ-induced IL7 [101].

Elevated TBK1, NAP1, IRF3 and IRF7 mRNA levels were found in biopsied mucosal specimens collected during colonoscopy and human colonic epithelial lines from inflamed colonic mucosa as compared to normal mucosa [102]. Rebamipide is used to treat gastric ulcers and has been used in a case study to treat IBD. Rebamipide reduced LPS (TLR4)/dsRNA (TLR3)-induced expression of IRF3, IRF7 and TBK1 in human colonic epithelial cells [102].

IRF4, IL17A and IL22 expression levels are increased in colonic biopsy specimens of IBD patients as compared to control patients. Moreover, their expression was upregulated in sections with inflamed mucosa compared to sections without inflammation. This is the first evidence to suggest that there might be a correlation between IRF4 levels and expression of Th17 cytokines [103].

Type 1 Diabetes

Type 1 diabetes is characterized by the autoimmune destruction of insulin producing β cells of the islets of Langerhans in the pancreas. Cytotoxic T lymphocytes are the main effector cells involved in the death of β cells [104]. A substantial amount of work has been done studying the importance of IRF signaling in mouse models of diabetes but no work has been published in humans.

Systemic lupus erythematosus (SLE)

SLE is the prototype autoimmune disease where genes regulated by type I IFNs are overexpressed and shown to contribute to disease pathogenesis. SLE is characterized in part by elevated antinuclear antibodies (ANA). 95% of patients possess ANA that recognize “self” chromatin components including dsDNA, histones, and nucleosome. Each of these components contributes to a break in self-tolerance, resulting in autoimmune responses [105]. Circulating ANAs bind self-nucleic acids and associated proteins exposed on the surface of apoptotic cells, generating self-antigen-containing immune complexes (ICs). These ICs are internalized by PDCs in an FcgRIIB-dependent manner, resulting in TLR activation and type I IFN expression [106-108]. The most studied IRF to date in SLE is IRF5. IRF5, like IRF7, is an IFN-induced gene, thus its expression can be enhanced by IFN production via a positive feedback loop, and subsequently enhance type I IFN gene transcription [109]. We and others showed that IRF5 expression was significantly elevated in PBMCs from SLE patients as compared to healthy donors and in lymphoblastoid cell lines from SLE patients carrying the risk haplotype [39,43,45]. Subsequent studies found that elevated IRF5 expression associated with risk SNPs and enhanced type I IFN activity in SLE patients with high anti-RBP or anti-dsDNA autoantibodies [110]. Using imaging flow cytometry, we later showed that IRF5 was constitutively activated (i.e. nuclear localized) in monocytes from SLE patients and that activation could be replicated in healthy donor monocytes stimulated with SLE serum. Activated IRF5 resulted in elevated IL6, TNFα and IFNα secretion [111]. Results from this study suggested that SLE-induced IRF5 activation may be TLR-, FcγRIIB- and IFN-independent in monocytes [111]. Later studies suggested that IRF5 risk haplotypes may predict specific B cell functioning pathways [112]. Others described a correlation between the IRF5 rs2004640 risk SNP and elevated cytokine expression when monocyte-derived cells were treated with different Nod2 and TLR ligands [113]. Of significant interest, the rs2004640/rs2280714 TT/TT IRF5 disease-risk-carrier cells were shown to have increased IRF5 expression and increased PRR-induced Akt2 activation, glycolysis, pro-inflammatory cytokines, and M1 polarization relative to GG/CC carrier macrophages, supporting additional functions for IRF5 in autoimmune disease pathogenesis [114]. Much of the work done on IRF5 signaling in human SLE has been focused on TLR-mediated IRF5 activation. While relevant given its critical role downstream of MyD88-dependent TLR signaling [115], this may be too simplistic of a view since SLE serum contains a number of complex triggers [111].

IRF3 and IRF7 signaling has also been shown to be involved in SLE pathogenesis (Figure 2). Enhanced IRF3 expression was found in PDCs from SLE patients, which associated with higher IFNα serum levels [31]. Studies on IRF7 in human SLE have expanded in the last few years due to its role in IFN expression. Indeed, genetic variants of IRF7 were shown to be associated with increased serum IFNα levels in SLE patients with anti-Sm and anti-dsDNA autoantibodies [116]. Independent studies found that induction of IRF7 protein expression correlated with recurrent lupus disease activity. It was found that phosphorylation of IRF3 and activation of 4 E-BP1, a translational repressor of IRF7, preceded disease flare, while recurrent disease activity and induction of IRF7 protein expression correlated with induction of the IFN signature [117]. Using an ISRE-Luc reporter assay, SNP rs1131665 in IRF7 was shown to confer elevated IRF7 transactivation ability [54]. Data on IRF1, IRF2, and IRF8 are more limited but many of the studies relate these IRFs back to type I IFN signaling or B cell function. Interestingly, 63% of genes with H4 hyperacetylation in SLE had potential IRF1 binding sites [118, 119]. Since there exists a higher preponderance of SLE in women than men, IRF1 becomes a molecule of interest as it is one of the major downstream mediators of prolactin signaling. Prolactin is immune stimulatory and can break B cell tolerance. IRF1 provides a potential nexus of female hormones, inflammation and type I IFN signals [118]. IRF1 not only activates gene expression as a transcription factor but may perpetuate SLE disease by leading to a dysregulated epigenome [120]. Although there is limited data on IRF2 in SLE, elevated transcriptional activity was described using an IRF2 promoter reporter carrying risk polymorphisms [121]. Additionally, SNP rs17445836G was found to be associated with increased IRF8 expression in SLE patient B cells, decreased serum IFN activity in patients with anti-dsDNA antibodies, and decreased ISG expression in PBMCs from anti-dsDNA-negative patients [122].

Irf loss-of-function studies in murine models of autoimmune disease

Given the difficulties in obtaining appropriate samples sizes for functional studies in human primary immune cells, knockout mice offer one of the most powerful means for studying gene function in a living animal. Indeed, knockout mice have been widely utilized in the study of autoimmune disease pathogenesis. Here we summarize data from murine models of autoimmune diseases that lack one or more Irf(s) (Table 1).

Table 1.

Autoimmune disease outcome in Irf-deficient models

	Lupus	Arthritis	EAE	IBD	Autoimmune Diabetes
Irf1−/−	↓auto-Ab ↓kidney disease	↓incidence ↓severity Delayed onset	↓incidence No effect on severity	↑incidence ↑severity	Completely protected
Irf2−/−	Naturally develops CLE-like inflammation	NT	NT	NT	NT
Irf3−/−	Association of disease protective	No effect	↓incidence ↓severity	↑severity	NT
Irf4−/−	↓auto-Ab ↓kidney disease	Association of disease exacerbation	Completely protected	Completely protected	Completely protected
Irf5−/−	↓auto-Ab ↓kidney disease	No effect	NT	No effect	NT
Irf7−/−	↓auto-Ab No effect on kidney disease	↑suceptibility	NT	NT	NT
Irf8−/−	↓auto-Ab ↓kidney disease	NT	Completely protected	NT	NT
Irf9−/−	↓auto-Ab ↓kidney disease	NT	NT	NT	NT

Irf, interferon regulatory factor; EAE, experimental autoimmune encephalomyelitis; IBD, inflammatory bowel disease; CLE, cutaneous lupus erythematosus; NT, not tested.

Experimental autoimmune encephalomyelitis (EAE)

EAE is the most commonly used experimental model for MS and mice deficient of Irf1, 3, 4 or 8 are protected from EAE. Levels of protection were distinct and may be due in part to differences in background strains [123-126]. For instance, Irf1^−/− L/J mice challenged with myelin basic protein (MBP) were found to have a significantly lower incidence of EAE as compared to wild-type (WT) littermates but disease severity was not altered[124]. Decreased iNOS induction and subsequent NO production in macrophages was indicated as a possible mechanism. Two other EAE models in Irf1^−/−mice of different backgrounds, 129/Sv and C57Bl/6J, were almost completely resistant to myelin oligodendrocyte glycoprotein (MOG)-induced EAE [127, 128]. Further mechanistic studies showed that autoreactive T cells isolated from Irf1^−/−129/Sv mice could not induce EAE in WT mice [127, 128]. Bone marrow (BM) chimeric C57Bl/6J mice were generated to differentially express Irf1 in the CNS and the immune system to dissect the role of Irf1 in either compartment. In this model, it was found that the susceptibility to EAE was associated with Irf1-mediated immune reconstruction in the periphery, whereas differences in disease severity (clinical score and disease duration) was regulated by Irf1 expression in the CNS [128]. The cause of disparity in phenotypes between Irf1^−/− strains is likely due to differences in genetic backgrounds, as well as the mechanism of EAE induction.

Irf3^−/− C57Bl/6 mice have significantly reduced incidence and severity of MOG-induced EAE[125], while Irf4^−/− C57Bl/6 mice [126] and Irf8^−/−C57Bl/6 mice [123] are completely resistant. The underlying mechanism(s) of protection in Irf3^−/− and Irf4^−/− mice seem to be similar and T cell-related. In both models, in vivo T_H17 differentiation was impaired and in vitro IL17 production was not induced. Notably, transfer of myelin-reactive Irf3^−/− CD4+ cells to WT mice failed to induce EAE in T_H17-polarised models as did WT cells transferred into Irf3^−/− recipients [125]. However, transfer of WT CD4+ T_H cells into Irf4^−/− mice resulted in the development of EAE in these previously resistant Irf4^−/−mice, and contrary to expectations, injection of Irf4^−/−CD4+ T_H cells into Irf4^−/− and WT mice had no substantial effect on disease course in either recipient mice. A decrease in IL6-induced RORγt expression and lack of IL6-mediated downregulation of Foxp3 were likely the reasons for failure of T_H17 differentiation in Irf4^−/− T_H cells [126]. IRF8 in T cells is also capable of regulating T_H17 differentiation and Irf8^−/−mice fail to generate T_H1, T_H17 and T_reg cells after MOG injection. However, unlike Irf3 and Irf4, Irf8^−/− T cells do not have a consequential role in EAE. Mice with monocyte- and macrophage-specific Irf8 deletion, but not T cell-specific Irf8 deletion, remained resistant to EAE, indicating that Irf8 in the monocyte and macrophage is primarily responsible for causing EAE [123]. Last, Irf7 expression was elevated in the CNS of MOG-induced EAE; yet, knocking out Irf7 in C57Bl/6 mice resulted in more severe EAE. Irf7^−/− C57Bl/6 mice had significantly higher numbers of T cells in the CNS, and both IL17 and IFNγ were expressed at higher levels in the CNS [129].

Arthritis

A commonly used model of arthritis in mice is the collagen-induced arthritis (CIA) model. In CIA, IRF1 appears to be involved in both the induction and effector phases of disease. Irf1^−/− DBA/1 mice challenged with collagen type II (dCII) had reduced incidence, severity and delayed onset of disease [124]. IFNγ production by lymph nodes cells and proinflammatory cytokine production by peritoneal macrophages in response to dCII were defective in Irf1^−/− mice. The role of IRF1 in the effector phase was studied by adoptive transfer. Injection of a CII-specific T cell line together with sera from arthritic mice into naive DBA/1 mice showed that disease was suppressed in mice lacking Irf1. The suppression of CIA in Irf1^−/−mice was even more dramatic than that reported in CD4- or CD8-deficient mice or in TNFreceptor p55-deficient mice [124, 130].

Chikungunya virus (CHIKV) infection is another model of murine arthritis that exhibits similar disease manifestations as those seen in humans [131]. While deficiency of Irf3 alone had no significant effect on disease progression, Irf7^−/− or double-deficient Irf3^−/−Irf7^−/− mice significantly increased susceptibility to CHIKV that resembled human disease [131, 132]. Thus, in the CHICKV infection model, Irf3 and Irf7 appear functionally redundant. Irf7 was studied in another model of murine arthritis, passive K/BxN serum transfer arthritis [133, 134]. Similar to findings from the CHIKV model, Irf7^−/− C57Bl/6 mice with K/BxN serum transfer had increased clinical severity of arthritis, with augmented systemic and local proinflammatory cytokine profiles, indicating an overall counter-regulatory role of Irf7 in the later stages of disease [135]. IFNβ and poly (I-C) treatment have been also shown to inhibit inflammatory arthritis in mice [136-138]. Both treatments resulted in decreased K/BxN arthritis in Irf7^−/− mice suggesting that Irf7 plays a counter-regulatory role in this model, possibly through its regulation of IFN production and cytokine gene expression.

In the B6-me^v/me^v(viable mothaeten) model of murine arthritis, Irf5+IL23+macrophages were significantly increased, suggesting a possible link between Irf5 and arthritis [139]. However, Irf5^−/−C57Bl6 mice with CIA displayed similar severity of paw swelling, as well as serum IgM, total IgG, IgG2a and anti-CII autoantibody levels as WT mice [27]. Thus, at least under these experimental conditions, Irf5 is not required for clinical manifestation of CIA-induced arthritis.

While the effects of Irf4 deficiency in mouse models of inflammatory arthritis have not yet been investigated, mice deficient in Irf4-binding protein (IBP^−/−) rapidly develop rheumatoid arthritis-like joint disease and large-vessel vasculitis [140]. IBP deficiency leads to the aberrant production of IL17 and IL21, which are regulated by Irf4. Absence of IBP leads to the enhanced targeting of the IL17 and IL21 promoters by Irf4, and the presence of IBP prevents Irf4 from binding to and transactivating the IL21 promoter. Irf4 is likely a pro-disease factor.

IBD

To date, only dextran sulfate sodium (DSS)- and oxazolone-induced colitis has been studied in Irf-deficient mice. Contrary to findings in other autoimmune disease models, lack of Irf1 is not protective in DSS colitis [141]. Instead, Irf1^−/− C57Bl/6J mice had increased distortion of crypt architecture, higher incidence of colonic dysplasia, and decreased expression of caspases and genes involved in antigen presentation. Similar to Irf7^−/− mice, loss of Irf3 exacerbated DSS-induced colitis [142]. The weight of Irf3^−/− C57Bl/6 mice continued to decrease even after DSS was replaced with water, indicating a defect in recovery from colitis. Colonic expression of Tslp and Il33 genes, which are critical for recovery from colitis, was significantly decreased in Irf3^−/− mice before and after DSS treatment, suggesting that deficiency in their expression might account for disease pathogenesis [141]. Even though Irf5 is activated by TLR and RLR pathways, which contribute to the suppression of colitis [143, 144], Irf5 was not essential for DSS-induced colitis. Knocking out Irf5 neither exacerbated nor improved DSS-induced colitis [142].

IL6 has an important role in intestinal inflammation as neutralizing anti-IL6R antibodies lead to suppression of established intestinal inflammation in murine models of IBD [145-147]. Irf4^−/− C57Bl/6J mice were completely protected from oxazolone-induced colitis, and only developed mild colitis after TNBS (oxazolone, trinitrobenzene sulfonic acid) treatment, whereas WT mice had severe colitis [148]. The striking phenotype seen in Irf4^−/− mice with either model of induced colitis was decreased mucosal IL6 production and suppression of anti-apoptotic IL6 signaling in Irf4^−/− T cells. Administration of recombinant IL6 or hyper-IL6 abrogated the protective effect of Irf4 deficiency in both TNBS and oxazolone colitis. Furthermore, mucosal IL6 production is a key regulator of T cell resistance against apoptosis in Crohn's disease and experimental colitis [149]. Lastly, immunodeficient RAG mice reconstituted with CD45RB^hiCD4⁺ T cells from Irf4^−/− mice, but not from WT mice, had less IL6 production and were resistant to colitis. This suggests that the protective effect of Irf4 deficiency in experimental colitis is mediated via T cell-derived IL6 [148].

Autoimmune diabetes

Non-obese diabetic (NOD) mice are a widely used animal model of autoimmune diabetes. Nakazawa et al. bred Irf1^−/− C57BL/6J mice onto the NOD background to examine its role in autoimmune diabetes development [150]. Loss of Irf1 completely blocked the development of insulitis and diabetes. It was found that the percentage of CD4+ and Mac1+ spleen cells were significantly increased, whereas CD3+, CD8+ and B220+ cells decreased in Irf1^−/− NOD mice. Furthermore, spleen cell proliferation in response to Con A or murine GAD65 peptide, a major autoantigen of the pancreatic β cell, significantly increased, and the IFNγ/IL10 ratio in culture supernatants significantly decreased from Irf1^−/− mice. IFNγ/TNFα synergism has been suggested to be responsible for autoimmune diabetes in vivo as well as β cell apoptosis in vitro [151]. IRF1, activated by STAT1 in pancreatic β cells, plays a pivotal role in this synergism. Thus, it is plausible that lack of Irf1 in NOD mice blocks the Stat1/Irf1 pathway, hence pancreatic β cell apoptosis. Irf4^−/−NOD mice were completely protected from the development of diabetes. Interestingly, Irf4^−/− NOD mice were also protected. Insulin autoantibody (IAA) expression was absent in both lrf4^+/− and Irf4^−/−NOD mice [152]. Disease resistance in Irf4-deficient mice may be due to the attenuation of effector T cell function in NOD mice. In an adoptive transfer study, WT T cells, but not pre-diabetic Irf4^−/− T cells, were able to induce diabetes in NOD/severe combined immunodeficiency (SCID) mice, while lrf4^+/− T cells were only able to induce mild diabetes. Further stratification of CD4+ and CD8+ T cells revealed that diabetes is suppressed in both Irf4^−/− CD4/Irf4^+/+ CD8 recipient mice and Irf4^+/+ CD4/Irf4^−/− CD8 recipient mice [152]. These results indicate that Irf4 plays a crucial role in both CD4+ T cells and CD8+ T cells in autoimmune diabetes.

Lupus

One of the first murine datasets that implicated Irf5 and Irf7 in lupus disease pathogenesis was stimulation of PDCs from Irf5^−/− and Irf7^−/− mice with purified human SLE ICs and measurement of type I IFN secretion; cells from both knockouts showed significantly reduced IFNα and β [153]. From then on, multiple murine models of lupus have been utilized to study the in vivo role of Irfs in disease pathogenesis. The majority of these models are genetic models of lupus onset, however, pristane-induced lupus is a commonly used model to study IFN-associated pathologies [154]. A number of Irf-deficient mice have been bred onto lupus prone backgrounds. Even though different Irf knockouts have been bred onto mice with different genetic backgrounds, including B6lpr, NZB, Faslpr/lpr MRL/lpr FcγRIIB−/−Yaa (RII.Yaa) and FcγRIIB−/−, most show protection against lupus [155-160].

Among the genetically modified models, decreased autoantibody production and ameliorated kidney disease was seen in Irf1-, Irf4-, Irf5- and Irf8-deficient lupus models. One common mechanism among these models lies in their ability to regulate T cells. Multiple subpopulations of splenic T cells, such as CD4-CD8-CD44+ T cells (decreased) and CD4+CD25+ T cells (increased), are skewed in Irf1^−/− mice [155]. Lack of Irf4 reduced T_H1 and T_H17 effector T cells [157]; age-associated splenomegaly and expansion of T cells typical of NZB mice were largely abrogated in Irf8^−/−NZB mice, accompanied by a significant decrease in CD4+ T-cell activation [160]. In Irf5^−/− MRL/lpr mice, CD4+ T cells were significantly decreased [158] and CD69+CD44+ expressing splenic T cells were markedly reduced in Irf5^−/− RII. Yaa mice[159]. Importantly, the observed T cell regulatory properties of these Irfs, alone, do not account for all mechanisms of protection. Reduced plasma cell and increased serum cytokine levels by activating DCs and macrophages were also seen in Irf4^−/− (Fas)lpr mice [157]. Irf8^−/− NZB mice were almost completely absent of PDCs (CD11c^lowCD11b⁻B220⁺Siglec-H⁺PDCA1⁺) in the spleen; yet had intact B cell function [160]. Deficiency of Irf5 led to increasedCD19+ B cells in the MRL/lpr background, and reduced CD69+ B cells in the RII. Yaa background [159]. Together, these observations indicate that the mechanism(s) of protection in Irf-deficient mice was often distinct (Figure 3).

Figure 3. Cell-type specific roles for IRFs in murine lupus.

A schematic showing which cell types IRF functions have been implicated in from Irf-deficient models of murine lupus.

In the pristane model of induced lupus, Irf5-, Irf7- and Irf9-deficient mice were all protected from disease to varying levels [161-163]. IgG2a/c is the primary pathogenic autoantibody induced in this model and it was suppressed in Irf5^−/− C57Bl/6 and Irf9^−/− Balb/c mice [163, 164]. The mechanism(s) of selective autoantibody reduction differed between Irf-deficient mice. For instance, the protection seen in Irf5^−/− mice was at least partially due to a failure in early T cell activation and a skew towards Th2 polarization [164]. Irf5 has also been shown to directly regulate the IgG2a locus [27] and regulate monocyte responses, including cytokine/chemokine expression and trafficking [161]. In Irf7^−/− mice, the level of glomerular IgG deposits was similar to WT mice treated with pristane, yet autoantibodies to DNA- and RNA-containing antigens was impaired [162]. In Irf9^−/− mice, B cells were found to express significantly lower levels of TLR7 mRNA and this was suggested as an explanation for the observed protection [163]. Importantly, neither skewed T_H polarization nor impaired B cell function was found in Irf7^−/− mice [162]. Instead, Irf7^−/− mice were found to have decreased neutrophils and PDCs, suggesting that protection may come from loss of these cell types [162]. Interestingly, while Irf5^−/− C57Bl/6 and Irf9^−/− Balb/c mice were protected from proteinuria and kidney disease, Irf7^−/− C57Bl/6 mice developed glomerulonephritis as their WT littermates, but additional inhibition of the NFκB pathway provided protection from nephritis [162]. These data implicate NFκB in Irf7-mediated kidney disease. It is important to point out that in pristane-induced lupus, nephritis starts before detection of anti-dsDNA IgG [162, 165, 166]. This suggests that unlike human SLE, anti-dsDNA IgG does not play a role in pristane-induced nephritis and therefore caution should be taken when translating data from this model into human disease.

Although Irf3^−/− mice have not yet been studied in the context of murine lupus, the role of Irf3 in lupus pathogenesis has been implicated in an IL23R-deficient mouse model [156]. IL23R deficiency protects from the development of lupus in C57BL/6-lpr/lpr mice [167]. Irf3 physically interacts with the promoter of IL23p19 and bone marrow-derived macrophages from Irf3^−/− C57BL/6 mice failed to produce IL23p19 following TLR3 stimulation [156]. These findings suggest that Irf3 may be a ‘lupus prone factor'.

Unlike other IRFs, Irf2 deficiency made mice susceptible to spontaneous cutaneous lupus erythematosus (CLE)-like inflammation as early as 9 weeks-old [167]. Even though depletion of CD8+ cells prevented disease development, transfer of Irf2^−/− CD8+ failed to induce disease in irradiated mice. Therefore, it is likely that the disease develops as a result of the interaction of CD8+ T cells with another cell type(s) [167].

Conclusions

Although we have only focused on the more common autoimmune diseases in this review, it is important to point out that additional work is being done in humans and Irf-deficient mice that are not limited to the above discussed diseases. For instance, IRFs have been implicated in Hashimoto’s thyroiditis [168] and myasthenia gravis [169]. Irf1^−/− mice have been examined in the context of autoimmune heart disease and lymphocytic thyroiditis [170, 171]. We can expect that this area of research will continue to grow and provide clarification towards IRF-mediated pathways that should be targeted for therapeutic intervention. Thus far, data indicate that the signaling pathways leading to and from different IRFs in autoimmune diseases are not independent - they can be synergistic, complementary, overlapping, or contrary. Examples of this are highlighted in Figures 1, 2 and 3 revealing the complexities of IRF signaling dependent on cell type. To further elucidate distinct and overlapping signaling between the IRFs, it will be necessary to generate double and triple knockouts.

An astonishing fact that becomes apparent in this review is the significant lack in translation from in vivo murine studies to human disease. The majority of data on IRFs in human disease is from GWAS. Few studies have begun to interrogate IRF signaling in the pathogenic immune cells from patients with autoimmune disease. These types of studies will be essential to our understanding of dysregulated IRF signaling in human autoimmune diseases since in the majority of diseases studied thus far, IRF expression and signaling are upregulated rather than absent, which is what we have been studying in Irf-deficient mice. Last, data presented herein support that not all roads from the IRFs lead to IFNs.