In the joint, the surface of the connecting bone is covered with a layer of articular cartilage. The articular cartilage can absorb and buffer forces between the joints to the maximum extent. Cartilage damage, one of the major reasons for disability in the elderly [1], can occur in knees, hips, ankles, and elbows. Slight cartilage injuries may get better on their own in a few weeks, while severe damage may eventually require surgery. Because of the lack of self-regeneration for the damaged cartilage, it is essential to understand the molecular mechanisms in the progression of cartilage damage [2, 3]. Some progress has been made in understanding the mechanisms of cartilage matrix degradation, and has promoted the progressive remodeling of the affected joints [4, 5]. It is reported that the complex network of signaling molecules can fine-tune the cartilage differentiation [6, 7], so mining the molecular changes of damaged cartilage from a systematic level may shed light on the discovery of specific therapeutic targets.

Microarray technology has been extensively used in identifying disease-related biomarkers [8, 9]. Besides mining the cartilage damage and regeneration-related mRNAs, emerging evidence indicates that miRNAs may also play an indispensable role in the network of regulating cartilage development [10, 11]. MiRNAs are found to comprehensively modulate cartilage development by establishing an interaction network with target genes, transcription factors, and cytokines [12]. In this study, we performed a series of bioinformatics analyses to a public microarray data containing the cartilage-damaged samples and their corresponding control samples for the sake of revealing the mechanism of cartilage damage and mining cartilage damage and repair-related mRNA, miRNA biomarkers and their regulatory network.

We compared the gene expression profiles between damaged and non-damaged knee cartilage, and revealed the differences between them from a system biology level. We screened out 422 DEGs that were involved in extracellular matrix, ossification, osteoblast differentiation, and mesenchyme development, suggesting that these DEGs played important roles in the occurrence of cartilage damage. The PPI network was constructed using the STRING database, and the topological analysis showed that the node degree of the network was subject to the power-law distribution, which is a typical characteristic for a biological network [21]. Thirty DEGs with high degrees identified in the PPI network may play a pivotal role and modulate the functions of this network. Three modules were identified in the PPI network, and the 30 hub genes were highlighted in the modules.

Genes in module1 were involved in cell cycles, and there were eight hub genes including CCNB1, CDK1, CDKN3, EZH2, KIF11, KIF20A, TOP2A, and TYMS. Among them, CDK1 (cyclin-dependent kinase 1) plays a key role in controlling the cell-cycle process for the eukaryotes. It can interact with multiple interphase cyclins to promote G2-M transition, and it regulates the G1 progress and G1-S transition [22, 23]. KIF11 (kinesin family member 11) is an essential molecular motor protein for mitosis. It can support cell proliferation by mediating centrosome separation and formation of the bipolar mitotic spindle [24, 25, 26].

Genes in module2 were associated with immune response related categories. The immune system is involved in the process of tissue injury, which indicates that tissue, organ, or appendage regeneration might be influenced by cartilage damage [27]. There were five hub genes including C1QB, C3AR1, CXCR4, FCGR3A and TYROBP in this module. Among the five hub genes, CXCR4 is an important molecular in cartilage degradation and the CXCL12/CXCR4 axis plays a pivotal role in injury and cartilage repair by acting as a chemo attractant of cells involved in inflammation and stem cell migration [28, 29, 30].

Module 3 enriched genes involved in ECM-related functions. Four hub genes including COL1A1, COL5A1, POSTN and PTGS2 were highlighted in this module. COL1A1 and COL5A1 are collagen genes. The expression of COL1A1 encoding a pro-alpha1 chain of type I collagen is abundant in bone. The mutations on COL1A1 are the major cause of osteogenesis imperfecta [31]. COL5A1 is the minor component of connective tissue and encodes the alpha chain of type V collagen. An animal study showed that the dysfunction of COL5A1 can generate an abnormal joint phenotype, such as joint laxity and early-onset osteoarthritis [32]. The extracellular matrix loss can lead to destruction of cartilage through uncontrolled production of matrix-degrading enzymes [33]. POSTN encodes a secreted extracellular matrix protein, playing a part in tissue development and regeneration. It can solidify connective tissues by crosslinking to other ECM proteins [31]. POSTN-null mice showed defective collagen crosslinks and decreases resistance to mechanical stress [34]. POSTN is re-expressed in fibrous tissues formed after injury and recruits mesenchymal cells by interacting with integrin, which is followed by tissue repair [35]. Therefore, POSTN has a crucial role in tissue repair. POSTN mRNA level was significantly higher in the OA cartilage than that in the controls [36]. In our analysis, POSTN is an up-regulated DEG (logFC = 4.02; p = 9.05 $\times$ 10${}^{-4}$), indicating that the POSTN may start the function of tissue repair after the cartilage damage. PTGS2 is an important gene for osteoarthritis. Dranitsina et al. [37] reported that the expression of TGFB1 and PTGS2 genes changed in rats’ knee joint cartilage under osteoarthrosis, suggesting the development of pathological processes in cartilage tissue, especially in degeneration and inflammation tissues.

Furthermore, upstream regulatory miRNAs were predicted for the DEGs. Mir-335-5p can regulate bone development to promote osteogenic differentiation [38]. In our study, the target mRNAs for miR-335 were mainly up-regulated, which suggests that the expression level of miR-335 may be down-regulated. It was reported that tissue damage or proinflammatory signals may cause downregulation of miR-335 and then activate the proliferative, migratory, and differentiation capacities of MSCs [39]. MiR-92a-3p plays a key role in chondrogenesis and cartilage degradation, and Mao et al. [40] revealed that miR-92a-3p enhances cartilage development and prevents degradation by targeting WNT5A. In this study, WNT5A was significantly up-regulated with log2 fold change 1.60 and p value 4.47 $\times$ 10${}^{-4}$, which was in accordance with the results that when the cartilage was damaged, miR-92a-3p repressed the chondrogenic differentiation and reduced cartilage matrix synthesis by enhancing the expression of WNT5A [41]. MiR-98-5p is involved in the regulation of osteoblast differentiation and promotes its differentiation through targeting and regulating CKIP-1 expression [42].

Although several important clusters and genes were identified in the cartilage damage patients, there were still some limitations for the current work. It is not available for additional microarray/RNA-seq data to validate the results in GSE129147. The mechanisms underlying the findings have not been thoroughly clarified and wet experiments such as qPCR and western blot in the mouse model or in clinical patients need to be carried out.

In this work, we analyzed a microarray data for cartilage-damaged patients, compared it with the controls, and identified three important modules, key genes and miRNAs related to the cartilage damage. Most of them were validated previously and related to the cartilage development or similar functions. All of these results need to be validated by more experiments.

BL conceived and designed the study, collected and analyzed the data, wrote and revised the manuscript. XX revised the manuscript.

Not applicable.

The author thanks the researchers from Ankara Yildirim Beyazit University submitting the dataset to GEO.

This work was supported by the Innovative Special Project of Agricultural Sci-Tech (Grant No. CAASASTIP-2014-LVRI-09).

The authors declare no conflict of interest.

The GSE129147 was downloaded from GEO website of https://www.ncbi.nlm.nih.gov/gds/?term=GSE129147.