Objectives Non-selective histone deacetylase (HDAC) inhibitors (HDACi) have demonstrated anti-inflammatory properties in both in vitro and in vivo models of rheumatoid arthritis (RA). Here, we investigated the potential contribution of specific class I and class IIb HDACs to inflammatory gene expression in RA fibroblast-like synoviocytes (FLS).
Methods RA FLS were incubated with pan-HDACi (ITF2357, givinostat) or selective HDAC1/2i, HDAC3/6i, HDAC6i and HDAC8i. Alternatively, FLS were transfected with HDAC3, HDAC6 or interferon (IFN)-α/β receptor alpha chain (IFNAR1) siRNA. mRNA expression of interleukin (IL)-1β-inducible genes was measured by quantitative PCR (qPCR) array and signalling pathway activation by immunoblotting and DNA-binding assays.
Results HDAC3/6i, but not HDAC1/2i and HDAC8i, significantly suppressed the majority of IL-1β-inducible genes targeted by pan-HDACi in RA FLS. Silencing of HDAC3 expression reproduced the effects of HDAC3/6i on gene regulation, contrary to HDAC6-specific inhibition and HDAC6 silencing. Screening of the candidate signal transducers and activators of transcription (STAT)1 transcription factor revealed that HDAC3/6i abrogated STAT1 Tyr701 phosphorylation and DNA binding, but did not affect STAT1 acetylation. HDAC3 activity was required for type I IFN production and subsequent STAT1 activation in FLS. Suppression of type I IFN release by HDAC3/6i resulted in reduced expression of a subset of IFN-dependent genes, including the chemokines CXCL9 and CXCL11.
Conclusions Inhibition of HDAC3 in RA FLS largely recapitulates the effects of pan-HDACi in suppressing inflammatory gene expression, including type I IFN production in RA FLS. Our results identify HDAC3 as a potential therapeutic target in the treatment of RA and type I IFN-driven autoimmune diseases.
- Rheumatoid Arthritis
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
Histone-modifying enzymes are epigenetic regulators implicated in the control of inflammatory processes, including immune and stromal cell activation, survival and proliferation.1 Histone acetyltransferases (HATs) acetylate lysine residues on histone tails, while histone deacetylases (HDACs) counterbalance HAT activity by deacetylating histone proteins. The delicate equilibrium between the acetylated state and the deacetylated state of chromatin orchestrates gene transcription.2 Furthermore, HATs and HDACs can also affect the acetylation status of non-histone proteins, thereby regulating signalling proteins and transcription factors to influence gene expression and cellular function.3 As a consequence, HDAC function could be essential to the development and perpetuation of chronic inflammatory diseases, such as rheumatoid arthritis (RA).4 In fact, HDAC activity and expression were shown to be altered in total peripheral blood mononuclear cells (PBMCs), synovial tissue and fibroblast-like synoviocytes (FLS) from patients with RA.5 Despite inflammatory mediators such as tumour necrosis factor (TNF) were found to positively associate with HDAC expression in synovial tissue and rapidly induce HDAC activity in FLS6 ,7 single targeting of TNF may not be sufficient to restore the HDAC balance in immune cells of patients with RA.8 This suggests that multiple factors contribute to the altered HAT/HDAC balance and that inhibition of HDACs could have a therapeutic contribution to RA treatment.
We and others have shown that pan-HDAC inhibitors (pan-HDACi) reduce cytokine production in FLS and in immune cells from patients with RA,8–10 display antiarthritic properties in vivo11 ,12 and demonstrated primary clinical efficacy in the treatment of rheumatic diseases.13 Which HDAC or combination of HDACs is specifically involved in RA pathology, however, remains unknown. The HDAC family includes 18 members divided into class I HDACs (HDACs 1–3 and 8), class IIa HDACs (HDACs 4–5, 7 and 9), class IIb HDACs (HDACs 6 and 10), class III sirtuins (Sirt1–7) and class IV HDAC11.14 Accumulating evidence suggests that some of the class I and class IIb HDAC family members could contribute to RA pathology, as their synovial activity is elevated compared with disease controls and further increased by inflammatory stimuli,6 ,7 ,15 and inhibition of their activity is protective in animal models of arthritis.11 ,16 Class IIa HDAC9 deficiency was found to enhance regulatory T cell function and was protective in disease models of systemic lupus erythematosus and colitis, but there is little indication for a direct involvement of HDAC9 activity in regulating cytokine expression.17–19 Furthermore, our previous data indicated that synovial expression of class IIa HDACs does not positively correlate with RA disease parameters nor with mediators of inflammation, and that class IIa HDAC5 is a negative regulator of chemokine expression in RA FLS.6
In this study, we investigated the potential differential contribution of class I and class IIb HDAC family members to the inflammatory status in RA FLS using the combination of selective HDACi and genetic silencing of individual HDAC expression.
Materials and methods
Patient material and FLS isolation
FLS were derived from synovial tissue specimens obtained from patients with RA by needle arthroscopy, as previously described,20 cultured in medium containing 10% fetal bovine serum (FBS, Invitrogen), and used between passages 4 and 10. All patients fulfilled the criteria for the classification of RA and had active disease, including clinical arthritis of the joint from which the synovial biopsies were obtained.21 Clinical characteristics of patients are shown in table 1. Informed written consent was obtained from patients prior to inclusion in the study.
FLS treatment and stimulation
FLS were cultured overnight in medium containing 1% FBS prior to incubation with cytokines. FLS were stimulated with 1 ng/mL interleukin (IL)-1β (R&D Systems), 1000 U/mL interferon (IFN)-β (Peprotech) or IFN-α (Bio-Connect Life Sciences). The pan-HDACi ITF2357 and inhibitors specific for HDAC1/2, HDAC3/6, HDAC6 and HDAC8 (Italfarmaco) were used at concentrations ranging from 20 nM to 2 µM. Information about the specificity of the HDACi has been previously published.22 ,23
Data are presented as mean±SEM, unless otherwise indicated. Friedman test followed by Dunns' post hoc test and repeated measures analysis of variance (ANOVA) followed by Bonferroni correction were used for analysing sets of data requiring multiple comparisons. The ratio t test was used for all other comparisons. Data were analysed using GraphPad software with p values <0.05 considered statistically significant.
Detailed descriptions of immunoblotting, HDAC activity, thiazolyl blue tetrazolium bromide (MTT) assay, mRNA expression analysis, ELISA, invasion assay, siRNA transfection, signal transducers and activators of transcription (STAT)1 DNA binding and immunoprecipitation are provided in online supplementary materials and methods.
Selective class I HDACi differentially regulate global protein acetylation in RA FLS
Pan-HDACi are broad-acting anti-inflammatory agents that are beneficial in several disease models.24 As primary evidence from in vitro and animal studies of arthritis pointed to class I HDACs as important contributors in the pathogenesis of RA,6 ,8 ,16 we attempted to dissect the potential roles of individual class I HDACs in mediating the inflammatory activation of RA FLS, using both pan-HDACi and inhibitors selective for HDAC1/2, HDAC3/6 and HDAC8. Treatment of RA FLS with each inhibitor resulted in distinct effects on global protein lysine acetylation. Both pan-HDACi and HDAC3/6i dose-dependently induced hyperacetylation of tubulin, a known HDAC6 substrate (figure 1A, top panel, 52 kDa band), as well as histone 3 (H3) and histone 4 (H4) (figure 1A, top panel, 18 and 14 kDa, respectively). In contrast, HDAC1/2i and HDAC8i displayed minimal to negligible effects on acetylation of these substrates. To confirm the pharmacological activity of the compounds, we measured the enzymatic activity of class I (figure 1B, upper panel), class IIb (figure 1B, lower panel and data not shown) and class IIa HDACs (data not shown) in lysates of FLS treated with the inhibitors. Pan-HDACi (p<0.0001), HDAC1/2i (p<0.05) and HDAC3/6i (p<0.05) significantly reduced class I HDAC activity, while a trend towards reduction in class I and class IIb HDAC activities was observed with HDAC8i and HDAC3/6i, respectively. Together, these data suggest that while each of the inhibitors displays pharmacological activity in RA FLS, HDAC3 and/or 6 are primarily responsible for mediating tubulin, H3 and H4 lysine acetylation.
Inhibition of HDAC3/6 displays similar effects to pan-HDACi in suppressing inflammatory gene expression in RA FLS
To exclude the possible effects of compound toxicity in our analysis, we exposed FLS to increasing concentrations of HDACi and verified cell viability by MTT assay. Treatment for 24 h had no discernible effect on overall FLS metabolic activity (figure 2A). To further assess the contributions of the different HDACi to the inflammatory activation of FLS, we analysed the expression of a panel of 83 IL-1β-inducible genes in the presence or absence of the HDACi by quantitative PCR (qPCR) array (figure 2B and data not shown). Eighty per cent of the genes downregulated by more than twofold with pan-HDACi in each of the three RA FLS lines subjected to this analysis were similarly affected by HDAC3/6i. In contrast, only the lymphotoxin (LTA) gene was downregulated by HDAC8i and none by HDAC1/2i. To confirm the effects of HDAC3/6i on gene expression, we performed independent qPCRs on a selected subset of targets using mRNA from additional RA FLS lines treated with the inhibitor (figure 2C). In agreement with qPCR array data, HDAC3/6i significantly suppressed IL-1β-induced expression of interferon-β1 (IFNB1), CXCL9, CXCL10, CXCL11, CCL2, CCL3, IL6, IL8, matrix metalloproteinase (MMP)1 and MMP3. In contrast, HDAC1/2i and HDAC8i failed to inhibit IL6 and IL8 induction, even at concentrations as high as 2 µM (see online supplementary figure S1). Consistent with its effects on mRNA expression, HDAC3/6i significantly suppressed IL-6 and IL-8 protein production following IL-1β stimulation (figure 2D, left panel and right panel, respectively) and reduced RA FLS invasive capacities (figure 2E), an effect possibly associated with decreased levels of MMPs, and induced expression of TIMP1 (figure 2B, C).
HDAC3, but not HDAC6, mediates IL-1β-induced gene expression in RA FLS
The HDAC3/6i used in these studies effectively targets both HDAC3 and HDAC6.23 HDAC3 has previously been identified as a key epigenetic modulator of inflammatory activation of murine macrophages and human PBMCs.8 ,25 ,26 Also, HDAC6 inhibition was shown to inhibit proinflammatory TNF-α and IL-6 cytokines in lipopolysaccharide (LPS)-stimulated THP-1 cells.27 To determine whether HDAC3 or HDAC6 might be responsible for the transcriptional changes observed with HDAC3/6i, we made use of an additional inhibitor specific for HDAC6.23 In initial experiments, we assessed the concentration of HDAC6i (1 µM) which induced a similar degree of tubulin acetylation as to HDAC3/6i (figure 3A) and had no effect on FLS viability (see online supplementary figure S2). Under these conditions, we observed no significant effect of HDAC6i on the expression of genes induced by IL-1β and suppressed by HDAC3/6i (figure 3B). While this suggested a primary role for HDAC3 in the effects of HDAC3/6i in suppressing FLS inflammatory activation, we sought to confirm this independently by knocking down HDAC3 and HDAC6 in FLS. Silencing efficiency at the level of mRNA and protein was confirmed for both HDAC3 (figure 3C, D, respectively) and HDAC6 (figure 3F, G). Silencing of HDAC3 significantly suppressed the IL-1β-mediated induction of genes targeted by HDAC3/6i (figure 3E and data not shown), contrary to HDAC6 silencing (figure 3H). Together, these results suggest that the HDAC3/6i prevents inflammatory gene expression primarily through its effects on HDAC3.
HDAC3 regulates STAT1 phosphorylation independently of STAT1 acetylation
Macrophages deficient in HDAC3 display an impaired inflammatory gene expression programme upon LPS stimulation, partially dependent on altered transcriptional activation of STAT1.26 Sustained levels of STAT1 protein and its activated phosphorylated forms are elevated in RA FLS and synovium, likely contributing to the maintenance of the active inflammatory process.28 ,29 We therefore investigated whether HDAC3 might regulate gene expression in a STAT1-dependent manner in RA FLS. Treatment of FLS with HDAC3/6i had no effect on total STAT1 protein expression, but prevented STAT1 Tyr701 phosphorylation in response to IL-1β stimulation. In contrast, HDAC6i had no effect on STAT1 phosphorylation (figure 4A). Consistent with HDAC3/6i effects on STAT1 phosphorylation, which is required for its transcriptional activation, HDAC3/6i also completely blocked the induction of STAT1 DNA-binding activity in response to IL-1β (figure 4B). Silencing of HDAC3 expression similarly prevented IL-1β-induced STAT1 phosphorylation (figure 4C). It has been previously reported that STAT1 hyperacetylation is a prerequisite for STAT1 dephosphorylation and inactivation,30 so we examined if HDAC3/6i regulates the acetylation status of STAT1. We found that HDAC3/6i had no effect on STAT1 acetylation after 4 h of treatment (figure 4D) or at earlier time points (figure 4E). We conclude that while HDAC3 strongly regulates STAT1 activity in RA FLS, this does not occur via a direct acetylation event.
HDAC3 controls the IL-1β-induced STAT1 phosphorylation via downregulation of IFN-β expression
STAT1 signalling is tightly regulated by the type I IFNs IFN-α and IFN-β and the type II IFN IFN-γ.31 In RA FLS, IFN-β is an essential mediator of TNF-dependent STAT1 activation and pharmacological inhibition of this pathway prevents T cell-attracting chemokine production in FLS.32 Here, we investigated whether regulation of IFN-β by HDAC3 might be responsible for the observed effects of HDAC3 inhibition on STAT1 activation and type I IFN gene responses in RA FLS. Exposure of FLS to IL-1β, IFN-α or IFN-β over time revealed differential kinetics of STAT1 phosphorylation (figure 5A). In particular, STAT1 phosphorylation in the presence of IL-1β was delayed compared with the responses induced by IFN-α or IFN-β. To verify whether late STAT1 activation by IL-1β would rely on type I IFN production, we silenced the expression of the IFN-α/β receptor alpha chain (IFNAR1) (figure 5B, C). IFNAR1 silencing potently blocked STAT1 phosphorylation in the presence of IL-1β or IFN-β (figure 5D), indicating that STAT1 activation by IL-1β is dependent on primary type I IFN signalling. Consistent with this, and in contrast to HDAC3/6i suppression of STAT1 phosphorylation in response to IL-1β, activation of STAT1 by exogenously added IFN-β was unaffected by HDAC3/6i (figure 5E). Silencing of IFNAR1 expression prevented the expression of a subset of genes regulated by HDAC3 (eg, CXCL9 and CXCL11), but left others (IL6 and IL8) unaffected (figure 5F). These data indicate that HDAC3 contributes to the activation of RA FLS in part through promoting type I IFN production and subsequent autocrine effects on STAT1-dependent gene expression and indicate that selective inhibition of HDAC3 could dampen the inflammatory activation of FLS.
In RA, substantial lack of responsiveness to available therapies is leading to the growing necessity to identify novel therapeutic targets which could suppress inflammatory cytokine production. Further on, relapse phenomena after reducing or stopping conventional and disease-modifying antirheumatic drug treatment may indicate a pathological epigenetically imprinted status of the immune and stromal cells that contributes to the perpetuation of inflammatory activation.33 Indeed, distinct DNA methylome signatures and elevated HDAC expression have been observed in long-term cultured RA FLS compared with osteoarthritis (OA) FLS,34 ,35 and pharmacological inhibition of proteins reading or modifying epigenetic marks was shown to prevent the inflammatory activation of RA FLS.5 ,36
Pan-HDACi were described as potent anti-inflammatory drugs in several immune-mediated diseases and in a variety of solid and haematologic tumours.24 Despite their safety and approved use for severe malignancies, including cutaneous T-cell lymphoma,37 evidence from cancer clinical trials has raised the potential of undesirable effects occurring upon HDACi treatment, such as thrombocytopenia, caused by defective megakaryocyte differentiation and platelet formation and possibly associated with tubulin hyperacetylation.38–40 Hence, selective HDAC inhibition may help to improve the therapeutic margin of safety. Emerging evidence indicates that specific class I HDAC family members (HDAC1–3, 8) could have a major role in the transcriptional regulation of inflammatory mediators, both in arthritis models and in other inflammatory diseases.16 ,22 ,26 Here, we compared the effects of ITF2357, a pan-HDACi shown to repress inflammation in in vitro and in vivo models of arthritis, with HDAC1/2, HDAC3/6 and HDAC8 inhibitors on gene expression in RA FLS. We found that inhibition of HDAC3/6, but not of HDAC1/2 nor HDAC8, highly resembled the effects of pan-HDACi, as it led to suppression of genes associated with RA pathogenesis, including cytokines, MMPs, as well as IFNB1 and IFN-related genes. Importantly, both HDAC1/2 and HDAC8 inhibitors that were used in this study are remarkably selective and retain selectivity at high doses,22 ,23 indicating that the lack of effect on the genes that we screened is unlikely to be associated with ineffective HDAC enzymatic inhibition. In line with this possibility, transcriptome analysis of HDAC1-knockout FLS revealed that HDAC1 is predominantly implicated in the control of cell migration and proliferation, rather than cytokine transcription.35 Additionally, the acetylation signature of PCI34051, a selective HDAC8 inhibitor, was found to be restricted to a limited set of targets, particularly SMC3.41 Taken together, HDAC1, HDAC2 and HDAC8 are likely to play roles in responses to other stimuli, target other genes not screened in our study, regulate the inflammatory response at later time points or have cell-specific roles.24 ,42 ,43
HDAC3 was previously shown to be an important epigenome modifier in the transcriptional regulation of inflammatory genes, as HDAC3 depletion prevents LPS-induced macrophage activation26 and its pharmacological disruption regulates atherogenic macrophage polarisation25 and cytokine production in RA patient PBMCs.8 On the other hand, HDAC6 plays a role in immunological tolerance in macrophages44 ,45 and its deficiency enhances regulatory T cells (Tregs) suppressive functions.46 Screening of a subset of genes affected by HDAC3/6i revealed that HDAC3 knockdown reproduced the effects of HDAC3/6i, though to a more moderate extent that reflected partial knockdown efficiency. In contrast, both HDAC6 silencing and inhibition showed null or mild effects on transcriptional regulation, indicating that HDAC3 has a primary role in mediating the IL-1β-induced activation of FLS.
The anti-inflammatory properties of HDAC3 depletion in macrophages were described to be dependent on altered STAT expression and function.26 This observation is in line with previous works showing that bulk HDAC activity and the expression of class I HDACs are required to regulate the Janus kinase (JAK)/STAT signalling.47 In RA FLS, we found that IL-1β-induced STAT1 Tyr701 phosphorylation, an indicator of STAT1 activation, was abrogated by HDAC3/6i. Notably, drugs interfering with the JAK/STAT signalling are beneficial in the treatment of patients with RA failing to respond to methotrexate, pointing to a relevant targetable pathway for the disease.48 As STAT signalling is regulated by HDACs, we wondered whether direct acetylation of STAT1 protein could interfere with normal STAT1 function. Evidence in literature indicates, in fact, that STAT1 acetylation is necessary for its subsequent dephosphorylation.30 ,49 However, we could not detect STAT1 acetylation after incubation with HDAC3/6i. Previous papers reporting STAT1 acetylation at late time points might indicate that this phenomenon is rather secondary30 and cannot, anyhow, explain HDAC3/6i early effects on gene regulation. On the contrary, we report that HDAC3/6i indirectly regulates STAT1 activation by primary suppression of type I IFN production. Given the variety of genes regulated by HDAC3/6i, we investigated to what extent selective suppression of IFN signalling could affect global gene expression in FLS. As expected, only the expression of a subset of genes involved in classical type I IFN response, such as CXCL9 and CXCL11, but not IL6 and IL8, was suppressed by IFNAR1 silencing. Thus, inhibition of the IFN signature is a relevant, but not exclusive, mechanism for the regulation of inflammatory gene expression by HDAC3/6i. In line with this possibility, previous findings from our group and by others identified control of mRNA decay as a distinct mechanism by which HDACi control gene expression.10 ,50 Specifically, IL6 transcript stability was significantly reduced after pan-HDACi treatment in RA FLS,10 suggesting that a similar regulation could occur upon selective HDAC3 inhibition.
While the process underlying HDAC3-mediated regulation of type I IFN signalling in FLS needs to be further characterised, and the action range of HDAC3/6i has yet to be investigated, data from this study and others provide strong evidence that HDAC3 can act as a crucial epigenetic regulator of inflammation. Our results suggest that the development of selective HDAC3 inhibitors could be beneficial in the therapy of inflammatory disorders, such as RA and other rheumatic diseases characterised by type I IFN signature,51 while limiting possible side effects of pan-HDACi.
The authors would like to thank Mr M W Tang and Ms T E Latuhihin (Academic Medical Center, University of Amsterdam) for providing patient clinical information.
Handling editor Tore K Kvien
Contributors CA, PAK and AMG contributed to research design, performed experiments, analysed data and contributed to writing the paper; IM vB, BSF, SG and BMF performed experiments and analysed data; TAM, PPT, GF, PM, DLB and KAR designed research, analysed and interpreted data and contributed to writing the manuscript. All authors read and approved the final version of the manuscript.
Funding This research was supported by a research grant from the Dutch Arthritis Association to KAR. BSF was funded by fellowships from the American Heart Association (12POST10680000) and The National Institutes of Health (NIH) (5F32HL124893-01). TAM was supported by NIH (HL116848 and AG043822) and the American Heart Association (Grant-in-Aid, 14510001). DLB is supported by a VICI grant from the Netherlands Scientific Organization (NWO) and a Consolidator grant from the European Research Council (ERC).
Competing interests None declared.
Ethics approval The study was approved by the medical ethics committee (METC) of the Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands (METC 2013_069).
Provenance and peer review Not commissioned; externally peer reviewed.