Systemic lupus erythematosus (SLE) is a systemic autoimmune condition predominantly affecting young women. It is a clinically heterogeneous disease with the potential to affect any organ system, and ranges from mild rashes and arthritis through to severe, life threatening renal and central nervous system disease (1). Few treatment advances in SLE have occurred in recent decades, which has resulted in a heavy reliance on glucocorticoids (GC) for the management of SLE. Whilst utilized for their immunosuppressive and anti-inflammatory properties, GC are associated with a myriad of detrimental metabolic effects (2). High doses of GC are often required to gain disease control in severe SLE, potentiating adverse effects and indicating a degree of GC resistance.

Whilst the pathophysiology of SLE remains incompletely defined, a pivotal role for the type 1 interferon (IFN) family has emerged, as evidenced by the overexpression of interferon-stimulated genes (ISG) in 60-80% of SLE patients (3, 4), and more recently by the success of an IFN receptor (IFNAR) blocking agent in improving clinical disease activity in clinical trials (5). The IFN pathway is most likely activated in SLE patients by immune complexes containing host nucleic acids (6). These stimulate Toll-like receptors in plasmacytoid dendritic cells (pDCs) to produce IFN, which in turn interacts with IFNAR, activating janus kinase 1 (JAK1) and tyrosine kinase 2 (Tyk2). This subsequently leads to the phosphorylation of signal transducer and activator of transcription (STAT) proteins. STAT1, STAT2 and interferon-regulatory factor 9 (IRF9) complex to form interferon-stimulated gene factor 3 (ISGF-3) which binds to interferon-stimulated response elements (ISRE), leading the transcription of ISG (7). Serum IFN has been difficult to measure in SLE patients, and consequently ISG have become a widely accepted surrogate measurement (5).

Despite their widespread use, GC have been shown to have poor ability to suppress ISG in patients with SLE except at very high doses, raising the possibility that IFN may be involved in inducing a GC resistant state (8). Recently, we have shown that IFN inhibits the expression of many GC induced genes, further implicating IFN in GC resistance in SLE (9).

GC, in conjunction with the glucocorticoid receptor (GR), acts as a transcription factor, with the ability to both induce and repress gene transcription (10). Glucocorticoid induced leucine zipper (GILZ) is an anti-inflammatory protein which is potently induced by the ligand-bound GR (11). GILZ has been shown to have a multitude of immunosuppressive effects relevant to SLE pathogenesis, including B lymphocyte regulation (12), and GILZ-deficiency exacerbates a murine model of SLE (13). The effect of IFN on GILZ is previously unreported.

Here we provide evidence that IFN suppresses GILZ. We show that GILZ is inversely correlated with ISGs in SLE patients and that GILZ is rapidly down regulated on exposure to IFN in human leukocytes. We demonstrate that transcriptional inhibition of GILZ by IFN signaling is mediated by Jak1/Tyk2, and importantly, show that IFN impairs GC induction of GILZ by reducing GR binding at the GILZ locus. This work identifies a novel mechanism via which IFN may induce GC resistance in interferonopathies such as SLE.

First, we analysed the relationship between GILZ mRNA expression and ISGs in a large cohort of SLE patients (GSE88884, n=1,796). As expected, SLE patients had over-expression of ISGs compared to healthy control subjects (median [IQR] normalized expression level -0.19 [-0.65- 0.43] vs 4.70 [2.00-5.44], p<0.0001) ( Figure 1A ). GILZ mRNA expression was lower in IFN-high SLE patients (n=1317) compared to IFN-low SLE patients (n=439) (0.10 [-0.36 - 0.56] vs 0.26[-0.12 – 0.61], p=0.0003) ( Figure 1B ), despite higher GC use in the IFN high group (IFN high median [range] prednisolone dose 10 (0–125)mg/day vs IFN low 5 (0–40)mg/day, p<0.001, data not shown). There was a weak negative correlation between IFN score and GILZ expression in this cohort ( Figure 1C ). Patients with high disease activity (SLEDAI score >/=10) had modest suppression of GILZ expression compared to patients with lower disease activity (SLEDAI<10) ( Figure 1D ). Similarly, patients with low complement levels had lower GILZ expression ( Figures 1E, F ). We previously published mouse model data showing GILZ deletion exacerbates clinical severity of disease without affecting autoantibody production (13), and likewise here we observed no difference in GILZ expression between patients with and without anti-dsDNA antibodies ( Figure 1G ).

To verify whether these correlations were directly linked to type I IFN, we tested the effect of IFN on GILZ expression in ex vivo healthy human PBMC. We found that recombinant human IFNα reduced GILZ expression in a time- and dose-dependent manner ( Figure 2A ). We also analysed the effects of IFN on GILZ expression in a variety of conditions from several publicly available datasets. Splenocytes from IFNα transgenic mice (GSE123549) had lower GILZ expression than control cells ( Figure 2B ). Patients with multiple sclerosis treated with therapeutic recombinant interferon-β had suppressed GILZ expression compared to patients who did not receive this treatment (GSE138064) ( Figure 2C ).

Since GILZ is an important target for the immunosuppressive actions of GC, we next analysed the effect of IFN on GC-induced GILZ induction in SLE patients from dataset GSE88884. Overall, we noted that IFN-high patients receive more GC (median [IQR] dose 9.5[2.5-15]mg vs 5.0 (0–10)mg, p<0.0001)( Figure 3A ), but had lower GILZ expression as discussed above ( Figure 1B ), suggesting impaired GC-induced GILZ expression in IFN-high patients. When matched for GC dose, IFN-high patients had lower GILZ expression that IFN-low patients ( Figure 3B ). We verified this effect ex vivo in human PBMC, showing that IFNα reduced GC induction of GILZ at both the mRNA and protein level ( Figures 3C, D ).

We next investigated the mechanisms behind IFN regulation of GILZ. In mice treated with tofacitinib, a pan JAK inhibitor, GILZ expression was elevated to a level comparable to dexamethasone (Dex) treatment. The effect of tofacitinib and Dex was not additive. ( Figure 4A ). Moreover, when healthy human PBMC were pre-treated with a JAK1/Tyk2 inhibitor (18), STAT1 phosphorylation by IFNα was inhibited in a dose-dependent manner confirming its efficiency at inhibiting downstream IFNAR signaling ( Figure 4B ). We confirmed the expression of the ISG RSAD2 was inhibited by TS treatment ( Figure 4C ). Importantly, the inhibitory effect of IFNα on GILZ and DEX-induced GILZ was released by treatment with TS, demonstrating that JAK1/Tyk2 mediates inhibitory IFN signaling acting on GILZ expression. ( Figure 4D ).

Finally, we examined the TSC22D3 locus, the gene encoding GILZ protein, for potential transcription factor binding sites regulating its expression ( Figures 5A, B ). We identified DHS linkages (19) predicting physical looping of the chromatin between sections of the TSC22D3 locus ( Figure 5A ). The presence of multiple loops indicated a putative enhancer region around 15kB downstream of the TSC22D3 gene, which is confirmed by the presence of H3K27Ac peaks in this region ( Figure 5B ). Interestingly, by analyzing public datasets GSE99887, GSE107584 and GSE43036, we found multiple overlapping binding sites for STAT1 and GR in this potential enhancer region and in the TSC22D3 promoter region ( Figure 5B ). Therefore, we conducted ChIP in the human B cell line L363 to demonstrate GR binding at several of these sites in response to GC treatment and determine how this was affected by the presence of IFN. Regions A and Z, which are located around the promoter of TSC22D3 isoform 1, showed minimal or no impact of IFN on GR binding ( Figures 5C, D ). However, Regions K and L, located in the novel enhancer region of the TSC22D3 locus, showed a consistent but not significant reduction in Dex-induced GR binding when IFN was present ( Figures 5E, F ). This supports the concept that IFN directly interferes with GC-induced GILZ expression by activating STAT competition for GR binding sites at the GILZ locus.

Glucocorticoids have formed a central part of SLE management for over seven decades. IFN has been shown to play a central role in SLE pathogenesis but GC are poor at suppressing the IFN pathway, and this is likely to play a role in GC resistance (8, 9). In this study we have shown that GILZ, an anti-inflammatory GC-induced protein, is suppressed by IFN. We provide evidence that downregulation of GILZ by IFN is JAK1/Tyk2 pathway dependent, and that the presence of IFN results in reduced GR binding in key regulatory regions of the TSC22D3 gene, blunting GILZ induction by GC. Paired with the existing knowledge that GILZ regulates multiple other components of the immune system relevant to SLE pathogenesis, this work provides evidence that IFN regulation of GILZ may be a key mechanism involved in GC resistance in SLE.

Our work shows that IFN is associated with suppressed GILZ in multiple conditions, including in vitro studies of human PBMC, as well as in vivo in SLE patients, in patients with MS taking therapeutic IFN, and in IFN-overexpressing mice. While ability for IFN to downregulate GILZ expression has not before been reported, our group has previously shown that patients with active SLE have reduced GILZ expression in B lymphocytes and plasmablasts compared with healthy controls (12). Given the known association of active SLE with high IFN (20), this data is consistent with the concept that IFN downregulates GILZ in SLE.

Previous studies have alluded to a possible role for IFN in causing GC resistance. Patients with SLE have a blunted response to GC, as measured by a reduction in cortisol suppression post GC treatment compared to control subjects (21). A similar phenomenon has also been reported in patients receiving therapeutic IFNα treatment for hepatitis C (22). Clinically, patients with SLE often require high doses of GC to control their disease compared to patients with autoimmune conditions such as rheumatoid arthritis, in which IFN does not play a central role. Here we have shown that IFN suppresses GILZ induction by GC, and that the presence of IFN leads to reduced GR binding in regulatory regions of the TSC22D3 locus, providing a mechanism via which IFN impairs a key GC immunoregulatory function. GILZ has been shown to mediate many of the anti-inflammatory effects of GC via inhibition of the key pro-inflammatory transcription factor NF-κB (23), as well as interactions with other mediators of inflammation such as transcription factor AP-1 (24) and inhibition of extracellular signal-related kinase (ERK) and PI3K-Akt signaling pathways (25, 26). GILZ has also been reported to have several cellular effects that are relevant to SLE pathogenesis, including a crucial role in B lymphocyte regulation (12) and effects on arthritis and leukocyte recruitment (27, 28). A predominance of Th17 cells and deficient or defective T regulatory cells have been reported in SLE (29, 30), and GILZ has been shown to both regulate the proliferation of Th17 cells (31), and favor T regulatory cell differentiation (32). Given GILZ has such key roles in mediating GC related anti-inflammatory effects, the reduced ability for GC to induce GILZ in high IFN states is likely to contribute to GC resistance in SLE.

Our findings suggest that the inhibition of GILZ by IFN occurs via the Jak/STAT pathway, as inhibition of this pathway ameliorated the effect of IFN on GC induction of GILZ. We also showed that inhibition of Tyk1/Jak2 with tofacitinib in lupus-prone mice produced an increase in GILZ expression, suggesting this pathway has an inhibitory effect on GILZ transcription. Consistent with this, anifrolumab, a therapeutic agent which inhibits the IFNAR receptor, and subsequently prevents the activation of Tyk1 and Jak2, was recently shown to have significant steroid-sparing effects in patients with active SLE in a successful phase 3 clinical trial (5). It will be valuable to assess potential restoration of GILZ expression by anifrolumab.

Previous studies have reported that the GR can directly interact with STAT proteins (33, 34). In our study, IFN seems to reduce binding of the GR at potential regulatory regions of the TSC22D3 locus, which indicates that the GC/GR transcription factor and transcription factors downstream of Tyk1/Jak2, such as the STAT proteins or the ISGF-3 complex, are competitively binding at the GILZ locus or directly interact with each other to regulate GILZ transcription. Further work, including determining the functional role of the potential regulatory regions, is needed to delineate the mechanisms at play.

In conclusion, our work shows that in the presence of IFN, GC have reduced ability to induce expression of the key anti-inflammatory protein GILZ, and that blocking the function of IFN by targeting the JAK-STAT pathway restores GILZ induction by GC. GC resistance in SLE is problematic as higher doses of GC are required to control disease activity but cause detrimental metabolic effects. This work provides a mechanism by which IFN may induce GC resistance in SLE and highlights pathways which may be targeted in the development of steroid-sparing agents.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Monash Health HREC. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Juntendo University Animal Ethics Committee.

WD, MN, PH, EM, and SJ contributed to conception and design of the study. WD, MN, TB, AD’C, RS, LG, BR, IM, and SS conducted experiments and analyzed data. MF, KH, and KI provided data from experiments. WD and MN wrote the manuscript. WD, MN, EM, and SJ revised the manuscript. All authors contributed to the article and approved the submitted version.

Work included in this manuscript was supported by grants from the Lupus Research Alliance (EM) and National Health and Medical Research Council of Australia (EM, MN, and SJ). Some data included in this manuscript were presented at the American College of Rheumatology congress, 2020 (abstract #0304).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.