Genetic Versus Non-genetic Drivers of SLE: Implications of IRF5 Dysregulation in Both Roads Leading to SLE

Introduction

Members of the interferon (IFN) regulatory factor (IRF) family of transcription factors are critical regulators of immune cell development, differentiation, and response, with abnormalities in IRF expression and function increasingly linked to autoimmune diseases [1•, 2•]. Genetic variants within or near IRF5 have been robustly associated with the autoimmune disease systemic lupus erythematosus (SLE) across every major ancestral group tested [3–7]. To date, four IRF5 homozygous risk variants have been identified that strongly associate with SLE risk and make up the major IRF5-SLE risk haplotype in European Caucasians. These four variants occur at regions of transcriptional control—two are in the 5′ untranslated region (UTR) of IRF5 (rs2004640 and a 5 bp CGGGG insertion/deletion (indel)) and one is in the 3′ UTR (rs10954213) [1•, 2•, 3–5]. The fourth one is 5 kb downstream of IRF5 at rs10488631 [4–6] (Fig. 1). Subsequent studies identified an additional single nucleotide polymorphism (SNP) in the 5′ UTR of IRF5 (rs4728142) that was suggested to explain all genetic risk from the four original variants; however, limited analysis has been performed in genotyped healthy donors and SLE patients that explain the functionality of this SNP [8]. Kottyan et al. reported that the risk allele of rs4728142 provided a site for increased binding of the transcription factor ZBTB3 in immortalized lymphoblastoid (B) cell lines [8]. Whether this SNP or ZBTB3 is directly functional leading to increased IRF5 expression is not yet known. Nonetheless, SLE patients carrying the major IRF5-SLE homozygous risk haplotype (Fig. 1) were found to have elevated IRF5 expression, at both the transcript and protein level [3–7, 9, 10]. Further, Niewold et al. reported that SLE patients carrying an IRF5 risk haplotype have elevated IRF5 expression that correlates with elevated serum IFNα in patients positive for anti-dsDNA or -RBP autoantibodies [10, 11]. In efforts to identify the functional contribution of SNPs to the regulation of IRF5 gene expression, many labs, including ours, have generated IRF5 mini-genes that contain individual genetic variants, or combinations of each [4, 6, 7, 9, 12]. To our great surprise, very few of these studies identified a significant functional role for individual variants in driving IRF5 expression. One possible explanation for this is that we have yet to identify the causal variant(s) that drives IRF5 expression. However, this is unlikely given the extensive replication of IRF5 genetic risk in multiple ethnic cohorts. A second possibility is that no single genetic variant is sufficient to drive risk and/or changes in IRF5 expression/function. Instead, alterations in IRF5 expression and function may require the full-risk haplotype [9, 10]. In support of this last thesis are findings from studies on SNP rs2004640. The risk allele of rs2004640 was originally identified (and functionally characterized) as supplying an alternative splice site that preferentially drives IRF5 transcription from non-coding exon 1B [4]. Unfortunately, however, this finding has not been well-replicated by others [9, 12], and thus, its functionality with regard to regulating IRF5 expression levels remains in question. However, numerous labs have reported changes in IRF5 expression in immortalized lymphoblastoid cell lines and primary blood cells from SLE patients carrying the rs2004640 risk allele along with other IRF5 risk variants [5–7, 9–11, 13, 14].

An additional alternate explanation, albeit more complex and intriguing, is the possibility that IRF5 genetic risk does not actually confer changes in IRF5 expression levels and instead confers changes in IRF5 activity. We began considering this possibility when we detected a very significant increase in IRF5 transcript levels in peripheral blood mononuclear cells (PBMCs) from SLE patients as compared to healthy donors that could not be explained by genotype alone (Fig. 2a) [9]. Stratification of IRF5 expression by SLE patients carrying either the homozygous risk or non-risk haplotype only provided a small association of elevated IRF5 expression with the IRF5-SLE risk haplotype [9]. These data were the first to suggest that there might be little contribution of the IRF5-SLE risk haplotype to elevated IRF5 expression. Instead, we postulated that other disease-associated factors, such as elevated levels of circulating immune complexes and/or type I IFNs, may be upregulating IRF5 expression [15]. Indeed, Niewold et al. reported that elevated IRF5 expression associated with elevated serum IFN activity in SLE patients carrying an IRF5 risk haplotype [10]. However, since this was an association study rather than a functional study, it is not currently clear whether the IRF5 risk haplotype is driving elevated IRF5 expression that leads to increased IFN activity, or the converse that increased IFN activity is inducing IRF5 expression [15].

In a subsequent study, we found that IRF5 is constitutively activated, i.e., nuclear-localized, in immune cells from SLE patients (Fig. 2b, c) [16]. In unstimulated cells, or cells from healthy donors, IRF5 generally resides in the cytoplasm [15, 17–19]. Only after stimulation of cells with pathogenic stimuli (virus, Toll-like receptor ligands, etc.) or treatment with DNA damaging agents does IRF5 undergo post-translational modification leading to activation and nuclear translocation [17–23]. Currently, it is not known whether the observed increase in basal IRF5 activation in SLE immune cells is driven by the IRF5-SLE risk haplotype or simply due to the presence of elevated circulating “pathogenic” triggers since the IRF5 genotype in these patients is not unknown (Fig. 2b, c) [16]. Thus, others and we have not yet excluded the possibility that the IRF5-SLE risk haplotype is driving IRF5 activation rather than elevated expression [16].

Given that IRF5 exists as multiple alternatively spliced transcripts that encode for protein isoforms with distinct cellular localization, cell type-specific expression and function [15], it is entirely plausible that total IRF5 expression levels could remain unchanged between risk and non-risk donors, yet IRF5 isoform expression is dramatically changed leading to altered IRF5 activity and/or function [24]. In support of this thesis, we previously found that the IRF5-SLE risk haplotype drives the abundance ranking of specific IRF5 transcript variants that drive distinct functions [24]. While the total level of IRF5 transcript expression was similar between risk and non-risk donors, the abundance ranking of individual transcripts was distinct between risk and non-risk donors that may indeed drive different IRF5 functionalities [24].

To date, the most well-replicated IRF5 functional variant is the promoter CGGGG indel as four copies (4×) of it provide an additional binding site for the Sp1 transcription factor that may play a role in the regulation of IRF5 expression and/or function [6, 25•]. Similar to SNP rs2004640, the CGGGG indel has its own controversies as well since some reports indicated no effect on IRF5 expression [25•], while others found gender-dependent effects [26]. Others yet reported decreased IFN-α production in plasmacytoid dendritic cells (pDCs) from donors carrying the CGGGG risk indel [27]. One possible explanation for the observed differences could be cell type-specific effects, as some of the studies were performed in B cells, while others were done in monocytes or pDCs [6, 25•, 26, 27]. As stated above for the rs2004640 risk allele, another possible explanation could be that even though a particular study may focus on one specific genetic variant, in this case the CGGGG indel, donors are likely to carry other risk variants that may be contributing to the observed phenotype [8]. It is extremely rare for an individual to carry only one IRF5 risk variant. Interestingly, a recent study by Calise et al. found no change in IRF5 expression levels between healthy donors carrying the 4× (risk) or 3× (non-risk) CGGGG repeat, and instead detected an increased susceptibility of myeloid cells (monocytes) from risk carriers to undergo DNA-damage-induced apoptosis [25•]. To date, the most studied functions for IRF5 are its role in regulating pro-inflammatory cytokine expression and cellular apoptosis [17–23, 28–30]. Important, however, is the fact that increased IRF5 expression is insufficient on its own to induce either of these functions [21–23, 25•]. Instead, IRF5 must be activated, which may or may not be a result of genetic risk [17–19, 28]. Studies by Hedl et al. found that monocyte-derived macrophages from healthy donors carrying SNP rs2004640 and the CGGGG risk indel produced significantly more pro-inflammatory cytokines when stimulated with a variety of TLR ligands [13, 14]. Unfortunately, it was not reported whether there was a change in basal IRF5 expression and/or activation in donor samples before stimulation [13, 14]. Altogether, these data support the premise that IRF5 genetic risk may confer changes in IRF5 activity/function rather than expression.

In line with this idea and driving the differential premise that the contribution of IRF5 to SLE susceptibility may be genotype-independent is based in part on our original finding that IRF5 expression (transcript and protein) and activation (nuclear translocation) were significantly elevated in immune cells from SLE patients as compared to healthy donors, which could not be fully explained by the IRF5-SLE risk haplotype [9, 16] (Fig. 2b, c). The observed dysregulation of IRF5 in SLE immune cells could instead be a result of circulating antigenic triggers that include apoptotic debris, immune complexes, and type I IFNs, resulting in upregulated IRF5 expression and activation [15, 16]. This idea is further supported by findings in multiple lupus-prone strains of mice that have been backcrossed to Irf5^−/− mice and show significant attenuation of disease onset and severity [31–38]. Each of these mouse strains (FcγRIIB^−/−, FcγRIIB^−/−Yaa, MRL/lpr, pristane-induced (Balb/c and C57Bl/6)) has distinct genetic backgrounds with distinct mechanisms that drive murine lupus and thus replicate much of the heterogeneity seen in human SLE. To our knowledge, none of these mouse strains/models have polymorphisms in the murine Irf5 gene, yet loss of Irf5 expression and function protects mice from disease onset and severity [31–38]. Other than attenuating disease onset and severity, however, very few, if any, of these studies has informed us of the exact role that Irf5 is playing in driving disease development. What we do know from these studies, though, is that when Irf5^−/− mice are backcrossed to any strain of lupus-prone mice, the mice are protected from disease [31–38]. Importantly, protection was also seen in mice with reduced Irf5 expression (Irf5^+/−) supporting a gene-dosage effect [31, 33, 37]. These data support the direct targeting of IRF5 inhibition in SLE, as a therapeutic effect would be expected even if we are unable to inhibit all cellular IRF5.

Thus, based on findings from mouse studies thus far, we have learned that a global (whole-body) loss of Irf5 expression results in poor lymphocyte activation, a Th2 phenotype, reduced cytokine and chemokine expression, reduced monocyte trafficking, reduced immunoglobulin G (IgG) class switching, reduced levels of anti-nuclear antibodies (ANA), and reduced plasma cell maturation [31–38]. While findings from these studies are important for the clinical validation of targeting IRF5 inhibition therapeutically, they also support that IRF5 is playing a role in lupus disease development through a mechanism that is independent of IRF5 genetics. Further studies in murine models of lupus that characterize Irf5 expression, activation, and function in the context of disease development—before clinical onset (negative for ANA(−) and proteinuria), during early onset (ANA+), and late-stages of disease (ANA+ and proteinuria)—will significantly aid in our understanding of how Irf5 becomes dysregulated in lupus, as well as providing insight into what might be driving its dysregulation.

Conclusions

In conclusion, numerous questions relevant to IRF5 and SLE disease pathogenesis still remain unanswered. These include the following: (1) What exactly is the contribution of IRF5 genetic risk to disease susceptibility? (2) Is genetic risk cell lineage-specific and are we looking in the right cells? (3) If IRF5 genetic risk confers elevated IRF5 activity that is independent of transcription levels, what is driving the activation? (4) Or, is dysregulated IRF5 activity simply driven by elevated levels of circulating antigenic triggers? And (5) conversely, if IRF5 activation is driven by risk variants, is it only then driving disease onset? In order to address many of these questions, a comprehensive analysis of immune-phenotypes from multiple independent cohorts of healthy donors that carry the major homozygous IRF5-SLE risk haplotype, as well as donors that carry individual risk and combinations of different risk alleles, will be required to understand the contribution of IRF5 genetic risk to SLE susceptibility. Independent of the manner in which IRF5 dysregulation contributes to SLE susceptibility, whether genetic- or non-genetic-based, compelling data from multiple labs now support that therapeutic strategies targeting IRF5 inhibition may prevent disease onset, as well as protect patients from the mortality associated with SLE.