1. Introduction
Rheumatoid arthritis (RA) is an autoimmune disease characterized by sustained chronic inflammation resulting in joint damage and severe disability. New and more effective therapies developed over the past two decades targeting the inflammatory mediators and immune cells [
1,
2] have revolutionized the management of RA [
3]. However, disease remission remains uncommon [
4]. Furthermore, therapies targeting different aspects of the immune response render patients more susceptible to infection [
5]. In light of these challenges, approaches that focus on other components of joint inflammation have been proposed as potential solutions [
4].
In RA, inflammatory cytokines, such as interleukin (IL)-6 and tumor necrosis factor-alpha (TNF-α), cause dysregulated proliferation and drive a migratory and invasive phenotype of synovial fibroblast [
6,
7] (also known as fibroblast-like synoviocytes (FLS)) [
8], resulting in pannus formation. Synovial fibroblasts are directly involved in cartilage and bone destruction by production of matrix metalloproteinases (MMPs) and activation of osteoclasts through receptor activator of nuclear factor kappa-B ligand (RANKL) [
8]. Synovial fibroblasts also contribute to inflammatory amplification via IL-6 production [
9]. Although immune cells and inflammatory mediators have been feasible targets for RA treatment [
1,
2,
10], there has never been a treatment strategy targeting specifically the aggressiveness of synovial fibroblasts. Compared with inflammatory mediators and immune cells, little is known about the causes of the particularly aggressive nature of RA synovial fibroblasts. Therefore, a more comprehensive understanding of synovial fibroblasts in RA may yield promising novel therapeutic targets.
A substantial number of epigenetic mechanisms are known to regulate dynamic changes in gene expression of various cells. MicroRNA (miRNA) is one best-known mechanism which plays a central role in post-transcriptional modification of gene expression via antisense binding to messenger RNA (mRNA). It was well-established that miRNA participated in pathophysiological processes of various autoimmune diseases, such as ankylosing spondylitis and uveitis [
11,
12]. However, the roles of miRNA in RA synovial fibroblasts were less clear.
In the past, several studies explored transcriptome of RA synovial fibroblasts utilizing RNA-seq [
13,
14]. However, they compared transcriptome changes before and after inflammatory mediator stimulation rather than transcriptome differences between RA and normal donors. Moreover, whether and how miRNA contributed to transcriptome changes in RA was not fully understood. To explore the full mRNA transcriptome and miRNA interactome of RA synovial fibroblasts in genome-wide scale, we performed transcriptome-wide RNA-seq and small RNA-seq in synovial fibroblasts from RA patient and normal donor to reveal changes to the synovial fibroblasts transcriptome and to elucidate the contributing miRNA to transcription signatures in RA synovial fibroblasts. We further applied interaction analysis of transcription factors, miRNA, and target genes, functional annotation and regulatory network mapping to elucidate involved molecular programs. Experimentally validated targets of miRNA and data from publicly available databases were integrated to validate our results. These results offer a map to the synovial fibroblasts transcriptome and miRNA interactome and shed light on the pathophysiology of RA.
4. Discussion
RA synovial fibroblasts constitute a unique cell type that distinguishes RA from other arthritic conditions and contribute significantly to the initiation and perpetuation of the disease [
39,
40]. Thus a detailed understanding of the internal state of synovial fibroblasts in RA pathogenesis is critical. By combining RNA-seq and small RNA-seq data, this study provides a global view of the transcriptome and miRNA interactome profile and discloses the significant contribution to altered transcriptome by miRNA in RA synovial fibroblasts. Additionally, we identified and validated one transcription factor (FOXO1), which contributed to altered transcriptome in RA synovial fibroblasts utilizing iRegulon and past microarray results [
25]. Moreover, three pairs of miRNA–target gene interaction (hsa-miR-31-5p:WASF3, hsa-miR-132-3p:RB1, hsa-miR-29c-3p:COL1A1) were validated through combining small RNA-seq with RNA-seq results, miRTarBase, and previous mRNA/miRNA profiling studies [
25,
27]. These results highlight the particular roles played by these transcription factors and miRNA in RA synovial fibroblasts.
Through interaction analysis of miRNA and target genes, a significant contribution to transcriptome alteration by miRNA was revealed. This exemplifies the complexity of interaction between miRNA and mRNA in RA synovial fibroblasts. This miRNA interactome provides a comprehensive molecular basis for additional information on the pathogenetic mechanisms of rheumatoid arthritis and offer a roadmap to directly probe miRNAs of interest with their likely downstream signaling pathways and functional roles of target proteins. Furthermore, miRNAs target multiple mRNAs in a network and, via dysregulation, implicated in numerous autoimmune diseases [
41]. Although targets of several miRNA (hsa-miR-17-3p, hsa-miR-125a-3p, hsa-miR-23a-5p, and hsa-miR-652-3p) (
Table S6) failed to be validated with published data, significant evidence of interaction between these targets and corresponding miRNA was found from miRTarBase, and these targets should be followed up in future studies. Moreover, considering the successful application of antisense oligonucleotide strategies in human diseases [
42] and therapeutic potential of miRNA in preclinical studies of RA [
43], targeting such a dysfunctional miRNA–mRNA interaction may hold promise for RA.
When we annotated differentially expressed transcripts using KEGG database, metabolic pathway was the top-ranking functional category. Several studies demonstrated that the dysregulated synovial cellular bioenergetics switched RA synovial fibroblast profiles and promoted inflammatory natures of RA synovial fibroblasts [
44,
45]. Furthermore, rewiring of synovial fibroblasts metabolism facilitated resolution of arthritis in the animal model [
44,
45]. Combined with beneficial effects of metabolic reprogramming in other autoimmune diseases clinically [
46], our study draws attention to the therapeutic potential of metabolic reprogramming for RA.
In the step of regulatory network mapping, downregulated FOXO1 was identified as one master regulator. Past studies showed FOXO1 induced apoptosis in RA synovial fibroblasts [
47]. Furthermore, differentially expressed genes regulated by FOXO1 (
Table S7) also participated in the proliferation, migration, and invasion. For example, CCND1, PAI-1, NOTCH1, ACADM, MEF2C, CDK1, TNNT1, CCNB1, DIO2 facilitated proliferation, migration, and invasion [
48,
49,
50,
51,
52,
53,
54,
55,
56,
57,
58], while RAB7, RUNX2, ICAM1, NFKB1, TXNIP, IL23A, NEP, catalase, GCK, PPARG, PUMA, AXIN2 suppressed proliferation, migration, and invasion [
59,
60,
61,
62,
63,
64,
65,
66,
67,
68,
69,
70,
71]. These results together suggest a contributing role of FOXO1 in RA synovial fibroblast proliferation (
Figure 7).
With regard to miRNA, three pairs of verified miRNA–target interaction with validated differential expression of both miRNA and target genes (hsa-miR-31-5p:WASF3, hsa-miR-132-3p:RB1, hsa-miR-29c-3p:COL1A1) were revealed. Hsa-miR-31-5p inhibited proliferation, migration and invasion of malignant cells [
72,
73]. Furthermore, WASF3 contributed to cancer cells proliferation, migration and invasion [
36]. These findings suggest a regulatory role of hsa-miR-31-5p:WASF3 in ameliorating proliferation, migration, and invasion of RA synovial fibroblast (
Figure 7). Concerning hsa-miR-132-3p:RB1, it was well-established that hsa-miR-132-3p increased cell proliferation in pancreatic cells [
37], and RB1 inhibited malignant cell proliferation [
74]. It is possible that downregulated hsa-miR-132-3p acted as a feedback loop to suppress proliferation of RA synovial fibroblasts (
Figure 7). Regarding hsa-miR-29c-3p:COL1A1, it appears hsa-miR-29c-3p inhibited proliferation, migration, and invasion of malignant cells [
75]. Moreover, COL1A1 promoted proliferation, migration, and invasion of malignancy [
76,
77]. As a result, downregulation of hsa-miR-29c-3p and upregulation of COL1A1 potentially enhanced proliferation, migration, and invasion of RA synovial fibroblasts (
Figure 7).
Conceptually, some of the present data may seem paradoxical, as some differentially expressed miRNA and mRNA resulted in ameliorated disease. However, it was compatible with current knowledge that upregulated negative feedback loop accompanied disease-promoting processes in disease pathogenesis [
78]. Similar phenomena involving miRNA and mRNA have been demonstrated in numerous inflammatory diseases and carcinogenesis. For example, miR-10b-5p, which inhibited production of IL-17, the central cytokine of ankylosing spondylitis, was upregulated in ankylosing spondylitis [
79]. IGF-I, which rendered cells susceptible to transformation and thereby contributeed to tumor progression, was decreased in breast cancer [
80].
In this study, synovial fibroblasts from normal donor rather than osteoarthritis were used for comparison. Most studies of synovial fibroblasts utilized osteoarthritis synovial fibroblasts as controls owing to limited accessibility of normal donor synovial fibroblasts [
7,
47]. Considering altered transcriptome of osteoarthritis synovial fibroblasts [
81], results of this study might be more biologically relevant for RA synovial fibroblasts.
Recent miRNA studies utilized various bioinformatics approaches as screening tools to identify miRNA–mRNA interactome without biological replicates [
82,
83]. This raised questions about validity of findings from these studies. The identification and validation of miRNA–mRNA target interactions are critical for our understanding of the regulatory networks governing biological processes. In our study, restricting miRNA–mRNA to experimentally validated interactions makes findings of our results more convincing and more suitable to serve as a starting point for investigations of miRNA-based therapies.