Introduction
Osteoarthritis (OA), one of the most important causes leading to joint disability, is associated with increased social and medical burden [
1,
2]. Regardless of wonderful advancements in diagnosis and treatment of OA these years, the prevalence of OA still increases from 6.6% to 14.3% between 1999 and 2014 in the USA [
3,
4]. Currently, the main management for early-stage OA includes lifestyle modification and pharmaceutical drugs, such as regular physical activity, Tai Chi, non-steroidal anti-inflammatory drugs, intra-articular hyaluronic acid, and corticosteroid injections [
5‐
9]. Irrespective of their potential effectiveness in increasing the time from diagnosis of OA to joint arthroplasty, these non-operative treatments can hardly block or reverse OA progression [
10]. Eventually, total joint replacement was recommended by orthopedic surgeons for patients with advanced OA owing to serious radiographic grade, pain, and functional impairment of involved joints [
11]. A crucial reason for these lies in that the key candidate genes and relevant signaling pathways associated with OA remains largely unknown. As a result, it is critical to further elucidate the pathogenesis of OA onset and progression.
Accumulative evidence suggested that many differentially expressed genes (DEGs) may participate in OA development. Kuttapitiya and colleagues demonstrated that 218 DEGs in bone marrow lesions of OA patients were related to OA-induced pain [
12]. Ramos et al. suggested that 694 DEGs were identified in blood of OA patients and these DEGs mainly enriched in the apoptosis pathways, which may be associated with the onset of OA [
13]. Recently, bioinformatics analysis were widely used to identify DEGs and perform subsequent enrichment analyses, such as Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, which may largely promote the understanding of OA pathogenesis [
14,
15]. Surgical-induced OA rat model which mainly involves anterior cruciate ligament transaction and destabilizing medial meniscus, is frequently used to explore the pathogenesis of OA in vivo. Previous studies also indicated that several ectopically expressed genes, such as AQP-1, GDF5, and TAK1, participated in the development of surgically induced OA in rat models [
16‐
18]. However, most of researchers merely attached importance to individual OA-related gene, which can hardly have a comprehensive understanding of corresponding molecule mechanisms, which were usually complicated and networked. Understandably, bioinformatics analysis may be a powerful way to explore these complicated regulatory networks and molecule mechanisms in a surgical-induced OA rat model.
In the current study, we identified several DEGs in a surgical-induced OA rat model after re-analyzing the raw microarray data (GEO Series: GSE8077). Enrichment analyses of DEGs were performed using Metascape. Construction of protein–protein interaction (PPI) network and identification of key genes were conducted using STRING, Cytoscape, and Centiscape2.2. Furthermore, miRDB and Cytoscape v3.6.0 was used for visualization of miRNA-mRNA regulatory network. We also performed KEGG pathway analysis for predicted miRNAs based on DIANA-miRPath v3.0.
Discussion
The surgical-induced OA rat model is one of the most common animal models in vivo. Therefore, it is essential to unveil potential molecular mechanisms in a surgical-induced OA rat model, which will contribute to clarifying the pathogenesis of OA. In the current study, several key genes and pathways were identified through reanalyzing GSE8077 dataset using integrated bioinformatics analysis.
Three steps were followed to perform enrichment analysis of DEGs, which revealed that DEGs mainly enriched in vasculature development, response to growth factor, positive regulation of cell migration, and ECM-receptor interaction. Many studies found that vasculature development at the osteochondral junction and synovium was associated with the onset and development of OA [
29,
30]. Meanwhile, some growth factors, such as VEGF and NGF, were obviously upregulated in subchondral spaces, vascular channels and chondrocytes of OA patients [
31]. Vasculature development was usually activated by some growth factors, such as VEGF [
32]. Further studies also indicated that repression for angiogenesis in osteochondral junction and synovium may had a potential inhibitory influence on OA progression [
33]. Accordingly, these growth factors may act as potential therapeutic targets for OA. Generally, loss of cartilage homeostasis and the dysfunction of chondrocytes phenotypes including cell apoptosis, cell migration, and cell proliferation are the critical pathological process of OA [
34,
35]. The current study also revealed that DEGs enriched in positive regulation of cell migration and negative regulation of cell proliferation. Meanwhile, ECM-receptor interaction was found to be the most significantly enriched pathway for DEGs, which was further verified by many previous studies [
36,
37]. Therefore, dysfunction of these cell phenotypes and molecules may play important roles in OA development and can act as promising pathological signatures for OA in vitro and in vivo.
Many OA-related key genes were also identified in the surgical-induced rat model, including Ccl2, Col4a1, Col1a1, Aldh1a3, and Itga8. Integrin α8 (Itga8) was an important component of ECM-receptor interaction pathway. It was significantly upregulated in mesenchymal cells and played important roles in the expression of extracellular matrix components [
38]. Gong et al. revealed that Itga8 may participate in the degradation of extracellular matrix, including collagen type XI alpha 1, aggrecan, collagen type VI alpha 1 in periodontal ligament tissues [
39]. Considering that imbalance of extracellular matrix anabolism and catabolism was the critical pathological process in OA, it was worthwhile to explore the potential roles of Itga8 in OA. Several studies revealed that the abnormal expression of collagen-related genes (Col1a1, Col4a1 and MMP12) participated in the pathogenesis of OA onset and progression, which were consistent with our study [
40‐
43]. Our study also found that CCL2 was upregulated in surgical-induced rat OA model and may participate in OA pathogenesis, which was further supported by previous studies [
44,
45]. Recent studies revealed that CCL2 can be responsible for monocytes’ migration and cartilage degeneration, and the CCL2/CCR2 axis may play a critical role in OA-related pain [
46,
47]. As one of Aldehyde dehydrogenase isoforms, Aldh1a3 was obviously upregulated in human articular chondrocytes. Furthermore, the activation of Aldh1a3 may be responsible for the producing activity of collagen II in chondrocytes [
48]. Thus, whether the dysfunction of Aldh1a3 was associated with OA pathogenesis was worthwhile to be further explored. Collectively, considering that the important roles of these key genes in OA, they may be used as potential molecular biomarkers and therapeutic targets for OA.
Previous studies revealed that many differentially expressed miRNAs (for example, miR-145 and miR-140) were associated with OA development and progression [
49,
50]. Also, some researchers demonstrated that circulating miRNAs, such as miR-19b-3p, miR-122-5p, miR-486-5p, hsa-miR-140-3p, hsa-miR-671-3p, and hsa-miR-33b-3p, can be promising diagnostic biomarkers for knee OA [
36,
51]. In the current study, we identified that 99 predicted miRNAs were mainly enriched in MAPK signaling pathway and of these, miR-199a-3p and miR-539-59 may act as potential key miRNAs to regulate corresponding mRNAs. Actually, previous studies have verified that these predicted miRNAs and pathways played important roles in OA. For instance, Sun et.al. found that inhibition of P38-MAPK signaling pathway participated in repressing chondrocytes apoptosis and the release of proinflammatory cytokines in OA [
52]. Furthermore, Akhtar et.al. suggested that overexpression of miR-199a can inhibit MAPK signaling pathway, thus attenuating OA progression [
53]. Therefore, these important miRNAs and signaling pathways can be served as potential diagnostic biomarkers and therapeutic targets for OA, which may provide potential hallmarks for further experimental studies.
The strength of the current study was that we performed comprehensive enrichment analysis based on Metascape in a rat OA model. Apart from common DEGs between OA group and control groups, we also performed meta-enrichment analysis of all the DEGs in two comparison cohorts. Besides, pathway enrichment analyses were undertaken to explore the potential roles of predicted miRNAs. Regardless of aforementioned strengths, our studies also existed some limitations. Firstly, our findings were merely based on limited sample size (five in each group), so it was hard to exclude potential random error and false positive. Accordingly, further studies with large sample size should be warranted. Secondly, the results of the current study were totally based on bioinformatics prediction and lacked subsequent experimental verification, such as RT-qPCR, western blot, and immunohistochemistry. Actually, owing to limited available materials in the current study, it was hard for us to verify our findings with these experiments. Anyway, the current study may provide some potential useful orientation for future experimental studies.