Rheumatoid arthritis (RA) is the most common chronic systemic autoimmune inflammatory disease. RA affects around 1% of people worldwide, and is more prevalent in women (Bartok and Firestein, 2010; Burmester and Pope, 2017). The quality of life in RA patients is critically compromised, which usually causes progressive articular destruction, early unemployment, and considerable disability implying a huge socioeconomic burden (Albers et al., 1999; Russell, 2008). RA is characterized by synovitis associated with the formation of a hyperplastic synovial membrane, which contains large number of immunocellular components including T and B cells, plasma cells, mast cells, macrophages, and activated RA synovial fibroblasts (RASFs) (Bartok and Firestein, 2010). The cellular interplay between these cells transforms the synovium into an invasive pannus and promotes angiogenesis, as well as the release of cytokines, chemokines, and matrix-degrading enzymes which facilitates the development of joint tissue damage. In particular, the RASFs, which actively proliferate and develop resistance to apoptosis are the major cell population in the synovial lesion (Perlman et al., 2001; Firestein, 2003; Muller-Ladner et al., 2007; Bartok and Firestein, 2010). These factors, coupled with activated RASFs' capability of secreting the aforementioned mediators of RA, establishes a paracrine/autocrine cycle that perpetuates RA synovitis, recruitment of new cells to the affected joint and induces joint destruction (Bartok and Firestein, 2010), which represents the potential therapeutic target of RASFs for achieving long-term remission of RA.

Early diagnosis and treatment is critical to the diminution of the seriousness of RA. Currently, non-steroidal anti-inflammatory drugs (NSAIDs), glucocorticoids (steroid hormones), and disease-modifying antirheumatic drugs (DMARDs) are standard pharmaceutical interventions for the inflammatory disorder. Of note, DMARDs medications suppress immune system functions to prevent further structural damage of bone and cartilage in the affected joints (Burmester and Pope, 2017; McInnes and Schett, 2017). Amongst which methotrexate (MTX) is the first-line therapy commonly prescribed, and the addition of targeted DMARD, such as tumor necrosis factor (TNF)-inhibitors, interleukin-6-inhibitors, B-cell depleting drugs, and Janus kinase (JAK)-inhibitors, are adjuvant interventions when poor prognostic factors are present. These new treat-to-target medications have improved the course of RA and most of them modulate specific mechanistic steps in the inflammatory process. However, a considerable numbers of individuals do not respond adequately to or are suffering from side-effects toward such treatments (Burmester and Pope, 2017; Tarp et al., 2017). Therefore, a better understanding of the molecular machinery associated with RA pathogenesis and the complex crosstalk between the constituting signaling pathways might help to facilitate the development or improvement of novel and conventional therapeutic strategies.

The Chinese medicinal herb Tripterygium wilfordii Hook f (TwHF) is a traditional remedy for RA treatment (Tao et al., 2002; Tang and Zuo, 2012). In fact, the recent randomized clinical trial studies further confirmed the therapeutic efficacy of TwHF in patients with active RA (Lv et al., 2015). Celastrol is the bioactive ingredient constituting TwHF which has demonstrated anti-proliferative and anti-inflammatory properties in both in vitro and in vivo models (Brinker et al., 2007; Kim et al., 2009; Venkatesha et al., 2011; Cascão et al., 2012; Nanjundaiah et al., 2012). It has been shown that celastrol promotes endoplasmic reticulum (ER) stress mediated by intracellular calcium (Ca²⁺) mobilization (Yoon et al., 2014). Ca²⁺ as a second messenger is required for the regulation of many cellular processes, including gene transcription, cell shape, motility, proliferation, mitochondrial function, apoptosis, and immune responses (Clapham, 2007). As early as two decades ago, the role of cellular Ca²⁺ in diseases was being noticed, from cardiovascular diseases to strokes, diabetes, the immune response (including inflammation) and cancer (Mooren and Kinne, 1998). More recently, studies have shown that intracellular Ca²⁺ signaling has been implicated in the pathogenesis of autoimmune disorders, such as RA (Izquierdo et al., 2014) and that altered cellular Ca²⁺ homeostasis is related to the control of various hallmarks of cancer. (Marchi and Pinton, 2016). Modulation of Ca²⁺ signaling has even been proposed as a possible therapy option for the treatment of cancer, though this remains as yet relatively unexplored (Rooke, 2014). Accordingly, celastrol is a suitable tool for investigating the role of Ca²⁺ signaling in pathomechanisms of RA. In this report, we hypothesized that compounds affecting cytosolic Ca²⁺ mobilization might serve to combat RA disease. The inflammatory gene expression profile which are regulated by Ca²⁺ dynamic changes in response to celastrol treatment were identified in patient-derived RASF. In addition, the possible transcriptional and molecular circuitry associated with the cytosolic Ca²⁺ mobilization was unraveled through literature review and bioinformatics analysis. Our findings provide novel insight into the application of ion channel modulators for RA intervention.

In this study, we found that celastrol significantly downregulated the expression of a number of genes related to the control of RASFs cellular pathophysiology. In fact, the suppressive effects of celastrol in immunoregulatory genes expression have been documented. By using multiplex analysis, Venkatesha et al. demonstrated that celastrol can reduce the expression of cytokines, such as TNF-α and IL-1β, and chemokines including MCP-1, MIP-1α, and RANTES, in different immunocellular components collected from a murine adjuvant-induced arthritis (AIA) model (Venkatesha et al., 2012). However, the detail molecular linkages between celastrol and such inflammatory-associated signaling are yet to be defined. Our PCR array and quantitative PCR data validated that the expression of a significant number (close to 20%) of these inflammatory genes was manipulated by celastrol in a Ca²⁺-dependent manner. Most importantly, the expression of these genes, critical to the perpetuation of the inflammatory phenotype of RA, were sensitive toward cytosolic Ca²⁺ accumulation as illustrated by the BAPTA/AM treatment suggesting the efficacy of targeting cellular Ca²⁺ level in RA therapy. Intriguingly, CD40 receptor was the only upregulated gene upon Ca²⁺ mobilization amongst the reported genes. In fact, the engagement of CD40 ligand and RASFs-expressed CD40 receptor can activate RANKL induction via ERK-1/2, p38 MAPK, and NF-κB activation and result in osteoclast hyperplasia which leads eventually to cartilage and bone destruction (Lee et al., 2006). Such observation further suggested the significant role of cellular Ca²⁺ in RA pathogenesis, which also aroused the concern of adverse effects associated with celastrol, since CD40 upregulation may activate the downstream inflammatory pathways worsening the RA progression. However, as demonstrated in our network map, such inflammatory pathways could, at the same time, be downregulated by other celastrol-regulated genes. On the other hand, celastrol induced downregulation of LEPR, cognate receptor for the satiety factor leptin, implying the possibility of abnormal weight gain of animals or patients upon celastrol treatment. Recently, Liu et al., found that prolonged exposure of mice with celastrol lead eventually to weight loss of the animals without inducing any toxicity and that mice with desensitized leptin signaling phenotype presented no significant responses to celastrol in terms of their weight (Liu et al., 2015). Therefore, it is reasonable to conclude that administration of celastrol in a reasonable dosage will not create significant toxicity which suggested the therapeutic value of our findings. In fact celastrol treatment with the use of AIA rat model has been reported to mitigate inflammatory RA with no toxicity, including the liver and kidney damages, demonstrated (Cascão et al., 2015).

Many receptors and transporter proteins are responsible for maintaining cellular Ca²⁺ homeostasis (Schwaller, 2012), for example, sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA), inositol trisphosphate receptor (InsP₃), and ryanodine receptor (RyR) which located on the ER are the key managers of intracellular Ca²⁺ storage. In addition, the sodium-calcium (Liu et al., 2015) exchanger NCX and uniporter on the mitochondria are another pair of intracellular Ca²⁺ receptors important for cytosolic Ca²⁺ regulation. On the plasma membrane, the plasma membrane Ca²⁺ ATPase (PMCA), together with, other Ca²⁺ channels such as voltage-operated calcium channel (VOCC), receptor-operated calcium channels (ROCC), and store-operated channels (SOCC), help to control the Ca²⁺ trafficking between cytoplasmic and extracellular environment. The observed Ca²⁺ flux in our RASFs is most probably due to the coupling of celastrol with SERCA (unpublished data) which further antagonizes the function of the receptor. The inactivation of SERCA stops the transportation of Ca²⁺ into the ER and results in cytosolic Ca²⁺ accumulation which mediates the expression of the validated genes. Upon celastrol treatment, CAST, LEPR, TOLLIP, and TRAP1, could be partially restored to the untreated level by BAPTA/AM implying that other unknown molecular mechanisms may underpin the regulatory machinery. After extensive literature review, the transcriptional regulations of CD40, TRAP1, NFATC3, and CAST, have been reported before which are related to cytosolic Ca²⁺ mobilization (Casanova et al., 2006; Landriscina et al., 2010), the remaining 19 genes appear to be newly discovered to have their expression machinery regulated by Ca²⁺ signaling.

The molecular pathways which are mediated by the products of our validated genes may form a complex network in RASFs with the constituting pathways crosstalk with each other. We accordingly generated a blueprint outlining such potential molecular circuitry using literature information documenting the signaling cascades that connect the products of individual validated gene to Ca²⁺ signaling, and RA pathogenesis. Although such signaling network is created by integrating the molecular pathways described in different cell types, they could potentially be involved in the regulatory machinery of RASFs which provided crucial hints for supporting follow-up functional experiments. Generally, these pathways could be summarized into five different categories of signaling cascade which are: (1) the calpains/calmodulin-calcineurin-NFATc3 axis; (2) PI3K-Akt/PKC-NF-κB axis; (3) JAK-Stat axis; (4) ERK-MMP axis; and (5) MyD88-AP1 axis. The calpains/calmodulin-calcineurin-NFATc3 axis is per se a well-known Ca²⁺-associated pathway inevitably regulating the cartilage catabolism and osteoclast formation, angiogenesis, and inflammation during RA pathogenesis (Sitara and Aliprantis, 2010). Upon Ca²⁺ influx, the calcium sensor proteins, including calpains and calmodulin, bind to the cation and activate calcineurin which further dephosphorylates NFATc3 (Im and Rao, 2004). After nuclear translocation, NFATc3 complexes with other cell-type specific transcription factors (TF) for regulating the downstream expression of genes, like RANKL, which is the molecular culprit causing bone erosion (Wu et al., 2007). Because of the capacity of NFATc3 to partner with other TF in various tissues or cellular components, the calpains/calmodulin-calcineurin-NFATc3-mediated gene expression could lead to a widespread of pathogenic effects. Our validated genes also include calpastatin which is the upstream inhibitor for calpains suggesting that cellular Ca²⁺ is holistically involving in the regulation of the various signaling pathways. Therefore, targeting the calpains/calmodulin-calcineurin-NFATc3 axis via the regulation of cytosolic Ca²⁺ distribution with Ca²⁺-mobilizing agent like celastrol, is a potential strategy for intervening RA progression. It is worth noting that, calpains could be the central hub in our molecular circuitry diagram intertwining the various signaling cascades. For example, calpains can intervene with the JAK-STAT axis by inhibiting the activity of suppressor of cytokine signaling 3 (SOCS3) (Pothlichet et al., 2008; Miyazaki et al., 2015). For the PI3K-Akt/PKC-NF-κB axis, calpains can regulate the pathway by inhibiting at the level upstream of NF-κB (Li et al., 2011, 2014; Storr et al., 2011, 2015). Also, calpains can manipulate the ERK-MMP axis by direct activation of ERK1/2 (Moshal et al., 2006). Although cellular and animal test are needed for further validation of our findings, other studies using AIA rat showed that celastrol can modulate the NF-κB pathway, MAPK pathway, and the JAK/STAT pathway (Venkatesha et al., 2016) which may support the signaling circuitry as proposed in this report. Also, the PI3K-Akt/PKC-NF-κB axis may significantly be regulated by celastrol-induced Ca²⁺ flux in RASF, since more than one third of the validated genes, including histamine receptor H1 (HRH1), CD40, NFKB1, LEPR, IL4R, FGF10, and ERBB2, potentially target this signaling axis with most of them functioned molecularly via PI3K, the upstream kinase of Akt and PKC. NF-κB involved extensively in the development of chronic inflammation (Tak and Firestein, 2001), our findings point toward the efficacy of controlling RA progression by regulating the balance of cellular Ca²⁺. On the other hand, the pathway mediating the toll-like receptor and IL-1α/βsignaling are also the potential target of the genes TOLLIP, TLR6, and IL1R1 via the MyD88-AP1 axis (Dower et al., 1986; Burns et al., 1998, 2000; Bulut et al., 2001; Ahmad et al., 2007; de Almeida et al., 2013). The manipulation of such genes could, therefore, regulate the innate immune responses such as cellular migration and inflammation. There are several validated genes of which the corresponding signaling pathways cannot be well defined or lack of information acquired from the literature search. However, the products of these genes are known to participate in the development of inflammation like invasion, angiogenesis, cellular proliferation, and etc. The functions of these genes (ADGRE5, BMP1, CMTM1, IK, NFX1, and SCUBE1) and the associated molecular mechanisms underlying the corresponding signaling, therefore, deserve in-depth investigation for completing the signaling network as depicted in this study.

Taken together, we have discovered that cellular Ca²⁺ homeostasis is an important factor regulating the expression of a group of genes in RASFs which may significantly affect RA pathogenesis. As supported by the bioinformatics analysis, this Ca²⁺-dependent mechanism induced by celastrol is a new mode of action which has not been documented previously by using other conventional RA pharmaceutical interventions such as methotrexate, rituximab, and tocilizumab (Burmester and Pope, 2017; McInnes and Schett, 2017). However, the use of celastrol also demonstrated similarities to two out of three of the examined FDA-approved RA drugs (methotrexate and tocilizumab) in terms of the expression of some genes for example BMP1 and ERBB2 which support the therapeutic potency of celastrol in RA treatment. In addition, products of the celastrol induced Ca²⁺-mediated genes are participating in both upstream and downstream of molecular mechanisms leading to a comprehensive signaling control in a top-down basis not just targeting a single signaling messenger within a particular pathway. Therefore, our data highlighted the potential application of compounds, like celastrol, which are capable of regulating cytosolic Ca²⁺ level, as molecular tools for investigating the pathomechanisms of RA and other chronic inflammatory disorders. Our findings also provide forecasting platform for the establishment of functional experiments to validate our recent findings by using in vitro and in vivo models which facilitate the exploitation of novel pharmaceutical targets and therapeutic compounds.

IdSRD and SM: conducted the experiments and drafted the manuscript. FG-M: prepared the figures and, materials and methods. IK and WH: conducted the bioinformatics analysis. BL: prepared the discussion and revised the whole manuscript. VW and LL: conceived the idea and designed the experimental plan.