Rheumatoid arthritis (RA) is a chronic systemic autoimmune disease characterized by hyperplastic synovium, invasive synovitis, and progressive joint damage (Scott et al., 2010). The incidence of RA is approximately 1% in the adult population, mainly affecting middle-aged to elderly women (Stephenson et al., 2018). RA is the leading cause of disability worldwide and imposes a significant public health and economic burden (Schmidt et al., 2020). The etiology of RA is complex and multifactorial (McInnes and Schett, 2011). Despite there have been major advances in the understanding of RA development, the pathogenesis of RA remains incompletely elucidated.

Microarray analyses has improved the ability of human to study the pathogenesis of diseases, increasingly being used to explore disease epigenetics and to screen for effective biomarkers for disease diagnosis and treatment (Kulasingam and Diamandis, 2008). Although several genes, such as HLA-DR, PTPTN22, CTLA-4, and PADI4, have been identified as genetic factors that contribute to RA susceptibility, we still don’t know the exact mechanisms by which these genes trigger RA (Moran-Moguel et al., 2018). Therefore, we can elucidate the pathogenesis of RA by identifying potential biomarkers and potential therapeutic targets. In addition, the competing endogenous RNA (ceRNA) hypothesis could facilitate the elucidation of molecular mechanisms underlying disease progression (Salmena et al., 2011). Tissue-specific expressed hub genes and regulatory networks of disease may be screened through transcriptome microarray and bioinformatic analysis (Cheng et al., 2021).

Del Rey et al. conducted transcriptional analysis of normal and pathological synoviocytes using microarray expression profiling (Del Rey et al., 2012). Georgel P et al. used high throughput expression analysis to evaluate the overall miRNA expression level in fibroblast-like synoviocytes (FLS) isolated from RA patients in comparison with FLS from osteoarthritic (OA) patients (Liu and Ren, 2020). Bi et al. assessed the differential expression profiles of lncRNA in FLSs between the RA group and the healthy control groups (Bi et al., 2019). In the present study, we downloaded the expression profiles of the above three different types of RNA from GEO database to screen the differentially expressed mRNAs (DEGs), miRNAs (DEMs), and lncRNAs (DELs) between normal and RA synovial samples. Hub genes, tissue-specific expressed genes, RNA regulatory networks, and novel anti-RA potential drugs were obtained by comprehensive analysis. This work provides a deeper understanding of the molecular mechanism underlying RA development and reveals novel biomarkers for disease diagnosis and potential drug for disease therapy.

RA is one of the most common chronic autoimmune disease and is heterogeneous with complex pathogeneses. Although there is an accumulating evidence suggesting the involvement of novel transcripts in the development of RA, detailed mechanisms still remain elusive. In this study, we obtained the 418 DEGs in FLS from RA patients and OA patients. We further perform a tissue-specific expression analysis of DEGs, the result reveals that the hematologic/immune system is the greatest distribution system, which may explain the disease occurrence of RA.

Module analysis of the PPI network using ClueGO uncovered that NF-κB pathway may perform a significant role in RA pathogenesis. This has been confirmed by a number of studies. B cell activating factor of TNF family (BAFF) promoted synovial inflammation by activating of B cells in RA through NF-κB signaling pathway (Zhang et al., 2021). Growth differentiation factor 11 (GDF11) was able to significantly restrain the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling pathway and prevent the development of RA (Li et al., 2019). Overexpression of miR-496 in RA-FLS may inhibit cell proliferation by inactivating the NF-κB pathway (Xing et al., 2021). CS-semi5, a semisynthetic chondroitin sulfate, may effectively improve synovial inflammation, and cartilage erosion in RA through NF-κB deactivation (Li et al., 2021). Hence, the intrinsic relationship between NF-κB pathway and RA needs to be further studied in the future.

Module analysis of the PPI network using MCODE identified 3 tissue-specific expressed hub genes involving in the hematologic/immune system. SPAG5 is a Protein Coding gene involved in the functional and dynamic regulation of mitotic spindles. There is growing evidence that it is upregulated in many human cancers, acting as an oncogene and promoting cell proliferation (Liu et al., 2018; He et al., 2020; Canu et al., 2021). CUX2 is a Protein Coding gene encoding a transcription factor which contains three CUT domains and a homeodomain. CUX2 functions as an accessory factor in the repair of oxidative DNA damage (Pal et al., 2015) and plays an important role in the neuron specification and spine development (Cubelos et al., 2010). THEMIS2, also known as ICB1 (Induced on contact with basement membrane 1), is a protein expressed in B cells and is associated with signaling proteins Grb2 and Vav1 (Hartweger et al., 2014). Themis2 is not required for B cell development and activation, but it lowers the threshold for activation of B cells in positive selection (Cheng et al., 2017). Themis2 may constitute a control point in macrophage inflammatory response (Peirce et al., 2010). There have been no reports of three tissue-specific expressed hub genes in RA-related research. Nonetheless, SPAG5 and THEMIS2 were overexpressed in the synovial samples of RA in our study. In addition, the results of ROC analysis revealed that these genes had an important diagnostic value for RA. Therefore, these findings suggest that SPAG5 and THEMIS2 may play an essential in progression of synovial hyperplasia.

After interacting 147 DEGs-related target miRNAs and 23 DEMs, 5 overlapping miRNAs were obtained as key miRNAs for further study. In addition to miR-20b and miR-30c, the roles of the other three miRNAs in RA have been reported. miR-496 can impair the proliferative ability and facilitate the apoptosis of IL-1β-treated MH7A through regulating MMP10 expression and NF-κB signaling pathway (Xing et al., 2021). miR-218-5p knockdown could modulate the proliferation, apoptosis and autophagy of RASFs by upregulating KLF9, and suppressing the activation of the JAK2/STAT3 signaling pathway (Chen et al., 2021). MiR-218 regulates the osteogenic differentiation of RA-FLS via the ROBO1/DKK-1 axis (Iwamoto et al., 2018). miR-26a could enhances cells proliferation and attenuate apoptosis of chondrocytes in RA rats (Jiang and Cao, 2020). In addition, miR-26a ameliorates the arthritis severity of the rats through directly targeting TLR3 in rat macrophages (Jiang et al., 2014). Given its prominent role in RA, plasma miR-26a has been proposed as the most promising non-invasive biomarkers for disease detection (Liu et al., 2019). Although there is no relevant research on miR-20b and miR-30c in RA, studies have shown that both of them have a definite effect on inflammation. miR-30c-5p adjusts macrophage-mediated inflammation through pro-apoptotic signals (Ceolotto et al., 2017). knockdown of lncRNA HOTAIR inhibits inflammatory cytokine secretion by increasing the expression of miR-20b and decreasing the expression of NLRP3, thereby relieving ankle swelling caused by gouty arthritis (Liu et al., 2021). miR-20b can restraint T Cell proliferation and activation by regulating NFAT Signaling Pathway in Thymoma-Associated Myasthenia Gravis (Xin et al., 2016). However, given that these results are based only on bioinformatics models and literature reports, further research is essential to validate the in-depth role of these 5 key miRNAs in RA.

We establish a ceRNA regulatory network was constructed based on five key miRNAs to investigate the role of lncRNA in RA. There are two regulatory axes involving hub genes among the network, including ARAP1-AS2/miR-20b-3p/TRIM3, ARAP1-AS2/miR-30c-3p/FRZB. Interestingly, the key miRNAs involved in these two regulatory networks are the same ones that have not been reported for RA. TRIM3 and FRZB are hub genes involved in the regulation of these two ceRNA axes, both are down regulated in our study. It is worth noting that our finding of FRZB differ from a prior study on circulating levels of FRZB in patients with early rheumatoid arthritis (Corallini et al., 2010). Experimental study has shown that TRIM3 expression in synovial tissue samples from patients with RA was lower than that of healthy controls, which is consistent with our findings, may play an anti-proliferative role in RA-FLS through the P38 signaling pathway (Wang et al., 2017). Our findings also offer insights into the roles of ARAP1-AS2. Unlike for rheumatoid arthritis, the study by Yang et al. showed that lncRNA-ARAP1-AS2 gradually up-regulated with the progression of diabetes (Yang et al., 2019). Mechanistically, lncRNA ARAP1-AS2 may participate in high glucose-induced proximal tubular cell injury by interacting with ARAP1 through persistent activation of EGFR/TGF-β/Smad3 pathway (Li X et al., 2020). ARAP1-AS2/ARAP1 may affect EMT processes and cytoskeleton rearrangement in human renal tubular epithelial cells via boosting of Cdc42-GTP levels (Li L et al., 2020). Although lncRNA ARAP1-AS2 has not been reported in RA, we assumed that ARAP1-AS2/miR-20b-3p/TRIM3 or ARAP1-AS2/miR-30c-3p/FRZB might be the key regulatory axes in the pathogenesis of RA based on the above analysis.

According to CMap analysis, five potential compounds for RA treatment were obtained, and their targets, and application scope were further searched. Troleandomycin (TAO) is one of cytochrome P-450 (CYP450) monooxygenases blockers, which is considered to be a novel drug in the treatment of human inflammatory bowel disease (Sasaki et al., 2003). This suggests that its therapeutic role in RA is also worth exploring. Antirheumatic agents are effective for Parkinson’s symptoms, but not for levodopa (Lee et al., 2020). Nonselective HDAC inhibitor trichostatin A (TSA) shared multiple molecular mechanism in suppressing inflammation and may represent a new principle in the treatment of RA (Chung et al., 2003; Gillespie et al., 2012; Grabiec et al., 2012). As a PI3K/AKT inhibitor, LY294002 is often used in RA-related studies (Brennan et al., 2002; Kim et al., 2005; Xing et al., 2016). This presents a number of possibilities for therapeutic strategies. Levamisole was previously used to treat rheumatoid arthritis due to its immunomodulatory properties. However, it had been withdrawn from the US market because of serious side-effects (Cascio and Jen, 2018).

This study indicates that screening for identify tissue-specific expressed hub genes and ceRNA network in RA using integrated bioinformatics analyses could help us understand the mechanism of development of RA. Besides, SPAG5 and THEMIS2 might be candidate biomarkers for diagnosis of RA. LY-294002, trichostatin A, and troleandomycin may be potential drugs for RA.