Physiological evidence for diversification of IFNα- and IFNβ-mediated response programs in different autoimmune diseases

Type I interferons (IFNs) comprise a large family of cytokines with antiviral, immunomodulatory and anti-proliferative activities. The type I IFN family consists of 17 closely related members, including 13 IFNα subtypes and 4 unique members, i.e., IFNβ, IFNε, IFNκ and IFNω, of which IFNα and IFNβ are most commonly expressed and well-characterized.

Type I IFNs achieve their biological effects by binding to multi-subunit receptors, IFNAR1 and IFNAR2, on the cell surface. This leads to receptor dimerization and activation of the JAK-STAT pathway, a complex cascade of intracellular secondary messengers that emerge in transcriptional activation of genes containing IFN-stimulated response elements (ISRE) and/or IFN gamma-activated sequences (GAS) [1‐4]. Upregulation of type I IFN response genes (IRGs) is referred to as a type I IFN signature and is a reflection of type I IFN bioactivity.

Initially, type I IFNs were defined by their antiviral effects and as a consequence, they were used for the treatment of chronic viral infections such as hepatitis B and hepatitis C [5]. The antiviral activity involves suppression of viral replication, induction of apoptosis in virally infected cells, stimulation of T cell and B cell responses, natural killer cell-mediated and CD8+ T cell-mediated cytotoxicity and activation of dendritic cells [6].

Increasing insight in the activities of type I IFNs has revealed their role as pleiotropic cytokines with a critical role in modulating immune responses. Several observations indicate involvement of type I IFNs and the presence of a type I IFN signature in autoimmune diseases, including systemic lupus erythematosus (SLE), Sjögren’s syndrome, systemic sclerosis, multiple sclerosis (MS), idiopathic inflammatory myopathies (IIM) and rheumatoid arthritis (RA) [7‐9]. Compelling evidence from studies in SLE demonstrates that IFNα in particular is directly implicated in the pathogenesis of SLE [10, 11]. SLE is characterized by the presence of autoantibodies to nucleic acid and associated proteins, which are able to induce IFNα protein [12]. Serum levels of IFNα are increased in SLE and associated with disease severity and organ involvement [13, 14]. In support of a pathogenic role of IFNα in SLE was the observation that virally infected people and cancer patients treated with IFNα sometimes produce anti-nuclear antibodies and occasionally develop SLE-like symptoms [15, 16]. The mechanisms by which IFNα may contribute to autoimmunity are the induction of autoreactive lymphocytes, enhancement of long-term antibody responses and priming of myeloid cells.

In contrast to the pathogenic effects of prolonged IFNα signaling in SLE, IFNβ administration has notable therapeutic effects in MS, an autoimmune disease of the central nervous system characterized by progressive neurological dysfunction due to demyelination and axonal damage [17]. In patients with MS, treatment with IFNβ reduces clinical relapses and brain disease activity, and slows down progression of disability [18]. The anti-inflammatory and tissue-protective mechanism of IFNβ likely involves anti-proliferative and pro-apoptotic effects, as well as induction of anti-inflammatory mediators such as IL-10, IL-1R antagonist and soluble TNF receptors and reduction of pro-inflammatory mediators such as IL-1, IL-6 and TNFα [19].

From the above, the question emerges why type I IFNs can be pathogenic in SLE but therapeutic in MS. It is tempting to speculate that despite mechanistic similarities, IFNα and IFNβ have distinct roles in immune regulation that confer these opposing effects. Comparison of the primary amino acid sequences reveals that IFNα differs from IFNβ by approximately 70 % [20]. Receptor binding studies demonstrate that IFNα and IFNβ interact with their receptors in a different manner, suggesting that IFNα and IFNβ activate the IFNAR1/IFNAR2-mediated signal transduction pathway in a slightly different way [21‐23]. Accordingly, in vitro studies reveal that IFNβ appeared to be more potent at inhibiting cell proliferation and inducing apoptosis than IFNα [24]. However, it is as yet unknown whether there are differences in the downstream gene activation program of IFNα- and IFNβ-induced IFN signatures in vivo.

In the present study, we used transcript profiling to compare the IFN signature gene components regulated by IFNα in SLE patients to those of MS patients who were treated with IFNβ. Moreover, we exploited our findings to delineate the nature of the type I IFN signature in IIM, RA and patients with MS who were IFNβ-naïve.

Presence of a type I IFN signature is often discussed as a similarity among autoimmune diseases. In the present study we provide evidence that type I IFN signatures in autoimmune diseases appear less uniform than generally assumed.

IRG expression patterns were different between SLE and IFNβ-treated MS, two autoimmune diseases in which IFN activity has opposing effects on immune pathology and regulation, believed to be a consequence of differential effects of IFNα and IFNβ. Log-ratios of the differentially expressed type I IFN gene clusters, designated GC-A (SLE-IFNα-related) and GC-B (MS-IFNβ-related), revealed excellent separation of patients with SLE and IFNβ-treated MS patients. Moreover, use of the GC-A/GC-B log-ratios in RA, IIM and IFNβ-naïve MS patients provided insight into the origin of the type I IFN signatures in these diseases.

The GC-A/GC-B log-ratios revealed that IIM is predominantly characterized by a GC-A, hence an SLE-like, IFN signature. This fits with the many similarities between SLE and IIM that have been reported previously, including those related to the IFN pathway [37]. With regard to RA, the GC-A/GC-B log-ratio was close to zero for some patients, indicating that both IFNα and IFNβ contribute to the IFN signature, as has been suggested before [38]. Interestingly, GC-A/GC-B log-ratios differed among IFNβ-naïve MS patients, as half of the patients had GC-A dominance, and others had GC-B dominance or ratios close to zero. This could implicate different mechanisms underlying the type I IFN pathway activation in these patients. We previously showed that presence of a baseline IFN signature in patients with MS is related to non-responsiveness to IFNβ treatment [27] and it is highly relevant to further investigate the role of GC-A and GC-B in this perspective, which is the objective of future studies. Altogether, these results suggest a differential role of type I IFNs in autoimmune diseases.

The considerable variance of type I IFNs in humans suggests that, although they bind to the same receptor, the effects they exert might be different. For example, it has been shown that IFNβ is more potent than IFNα in inhibiting proliferation, inducing apoptosis and cell differentiation [39, 40]. Differences were partly explained by the different affinities of type I IFNs for their receptors, resulting in different receptor trafficking, phosphorylation and signaling kinetics [21, 41]. More recently, it has been described that IFNβ can uniquely ligate to IFNAR1, independently of IFNAR2 [42]. Overall, IFNβ appears more potent in activating signal transduction than IFNα, as demonstrated by a more stable receptor complex formation [43], a lower concentration of drug giving the half-maximal response (EC₅₀) for ISGF3 phosphorylation [44] and induction of a larger amount of genes than IFNα, especially at long incubation times of 16–36 h [45]. Notably, these long incubation times conform to the chronic IFNα exposure in SLE and 3 months of IFNβ therapy in MS. Interestingly, Moraga et al. hypothesized that the short-term complex formation of IFNα with its receptors might cause a constant low level of apoptosis, whereas the long-term complex formation of IFNβ with its receptors could more potently induce high levels of apoptosis. As SLE pathology is characterized by impaired clearance of apoptotic cells, resulting in immune complex formation and consequent IFNα induction, the low levels of apoptosis as mediated by IFNα could be key to persistence of a vicious pro-inflammatory circle [46]. In MS, however, apoptosis of autoreactive T cells is considered to be one of the anti-inflammatory actions of IFNβ therapy [47].

The implication of IFNα-and IFNβ-specific signatures is supported by the experiments of Der et al., who performed an in vitro experiment in which the fibrosarcoma cell line HT1080 was stimulated with either IFNα or IFNβ, followed by gene expression measurements using oligonucleotide arrays for ±6,800 genes [48]. From these experiments, seven genes overlap with our gene clusters, one from the GC-A cluster and six from the GC-B cluster. IFITM1, a GC-A gene, had slightly higher expression in IFNα-stimulated cells compared to IFNβ, whereas the GC-B genes EIF2AK2, IFIT1, IFIT2, MX1, OAS2 and PLSCR1 were all induced more by IFNβ than by IFNα (1.2-fold to 7-fold higher induction) [48]. Despite the small overlap of genes, the consistency of these results is striking.

It has been suggested that the type I IFN response differs among immune cell types [49], implying that the observed differences between SLE and MS could be partly due to differences in immune cell composition. However, the agreement between our data and those of Der et al. suggests that the observed differences are due to consistent differential signaling in all cells rather than large differences in immune cell compositions or IFNAR expression. However, for replication and complete definition of IFN subtype-specific response programs, whole genome expression studies are required.

Analysis of transcriptional regulation of GC-A or GC-B genes showed enrichment of IRF8/ICSBP binding sites and ISRE in the GC-B cluster. Remarkably, none of the IRGs from the GC-A cluster contained an ISRE, the response element that binds the ISGF3 complex downstream of canonical type I IFN signaling. As they did contain binding sites for STAT1 and/or STAT2, they are probably induced via STAT1-STAT1 monomers or STAT1-STAT2 heterodimers, which both IFNα and IFNβ are able to activate [50, 51]. The observation that these genes are increased in SLE compared to IFNβ-treated MS might be explained by the indication that IFNβ, in contrast to IFNα, might more potently activate a broader range of signaling proteins, including ISGF3 and IRF8, resulting in relatively less activation of the GC-A genes by IFNβ.

Expression of IRF8 is restricted to immune cells and it has the ability to act as a repressor or activator of the IFN response, depending on its interaction partner. A study by Meraro et al. reported that the IRF1-mediated induction of the IFN response gene ISG15 was inhibited in the presence of IRF8, whereas interaction of IRF8 and PU.1 synergistically enhanced ISG15 induction [52]. This suggests an immunomodulatory role for IRF8, which might be key to the different effects of IFNα and IFNβ on the immune system.

Conclusively, this study demonstrated that the IFN signatures display distinct differences between autoimmune diseases. Considering the pro-inflammatory nature of IFNα in SLE and the anti-inflammatory role of IFNβ in MS, specification of the type I IFN response in autoimmune diseases might give new insights into its role in disease pathology and/or its therapeutic potential.