As the most frequent degenerative condition in the joints, osteoarthritis (OA) has been associated with adults’ pain and disability. Joint damage, overweight, aging, and heredity factors are regarded as an etiologic factor for OA (1). Several abnormalities in the articular cartilage, subchondral bone, synovial tissue, and meniscus have been detected in the course of OA. Destruction of articular cartilage, the formation of osteophytes, subchondral sclerosis, and hyperplasia of synovial tissue are hallmarks of OA (1). Several molecules and pathways such as TGF-β, Wnt3a, Hedgehog, Smad3, β-catenin, and HIF-2α have been identified to contribute to the pathologic event during the OA course (1). In addition, systemic inflammation and the secreted cytokines in this process, particularly IL-1β and TNF-α can activate the NF-κB pathway in synovial cells and chondrocytes, participating in the pathogenesis of OA (1, 2). More recently, several investigations have underscored the regulatory roles of non-coding RNAs (ncRNAs) in OA development. Different classes of non-coding RNAs, including long ncRNAs (lncRNAs), microRNAs (miRNAs), and circular RNAs (circRNAs), have been reported to affect the development of OA. In the present article, we aimed at reporting the role of these transcripts in this process.

LncRNAs are transcripts with sizes of more than 200 nucleotides. Although they are not translated into functional polypeptides, they have crucial functions in regulating protein-coding genes’ expression. As a novel epigenetic control level, they affect several human disorders’ pathogenic course (3). Despite poor evolutionary conservation across different species (4) and a low level of expression in many tissues (5, 6), their functionality in the regulation of gene expression in cis- and trans- modes has been verified (3). These transcripts can serve as molecular sponges for miRNAs to release miRNA targets from inhibitory effects of miRNAs. Moreover, they can induce the H3K27 trimethylation, as a repressive epigenetic mark in the promoter of certain genes, thus inactivating target genes (7).

In the course of OA, several lncRNAs have been reported to be dysregulated. Function of some of lncRNAs in the pathogenesis of osteoarthritis has been described with more details in research papers. In this section, we selected some of them with clearer mechanistical information. For instance, expression of H19 has been reported to be up-regulated in samples from OA patients and chondrocytes cultured in the presence of IL-1β (8). H19 up-regulation has suppressed proliferation and stimulated apoptosis in these chondrocytes, whereas H19 silencing has exerted the opposite impact. These effects are mediated through sponging miR-106a-5p (8). Notably, expression of this lncRNA has also been shown to be elevated in peripheral blood of patients with OA in correlation with the Kellgren and Lawrence (K-L) grading system. Besides, its expression has been inversely correlated with bone metabolism parameters, namely PINP, N-MID, BGP, BALP, and Lysholm score, while being positively correlated with β-CTX parameter and VAS and WOMAC scores (9). In addition to H19, expression of HOTAIR has been up-regulated in cartilage samples of the femoral condyles or tibial plateaus of patients affected with OA compared with control samples. Up-regulation of this lncRNA has led to a severe upsurge of apoptotic rate and decreased chondrocyte viability. Mechanistically, HOTAIR increases Bax expression and the proteolytic cleavage of caspase 3 and decreases survivin and Bcl-2 levels. In chondrocytes, functional studies have shown that HOTAIR acts as a sponge for miR-130a-3p, i.e. sequesters this miRNA and releases its targets from inhibitory effects of this miRNA (10). Another functional route for the participation of HOTAIR in the development of OA is through enhancement of expression of genes related to cartilage destruction. HOTAIR directly represses the expression of Wnt inhibitory factor 1 (WIF-1) through induction of H3K27 trimethylation in its promoter, thus activating the Wnt/β-catenin pathway (7). DANCR is another up-regulated lncRNA in human OA cartilage and lipopolysaccharide (LPS)-induced chondrocyte cells. DANCR silencing has attenuated LPS-associated apoptosis and inflammation, enhanced cell survival, abridged apoptosis, and reduced IL-1β, IL-6, IL-8, and TNF-α levels. DANCR functions are mediated via sponging miR-19a (11). In addition, DANCR has been recently found to regulate expression of SOX9 (12). DANCR has also been shown to enhance the proliferation of synovial fluid-derived mesenchymal stem cells and increase chondrogenesis through sponging miR-1275, a miRNA that regulates the expression of MMP13 and regulate its expression. Through similar route, DANCR regulates expression of SOX9 (13). Expression of MALAT1 has also been elevated in the synovial tissues of obese OA patients compared with normal-weight OA cases or non-OA controls. Its expression has been sharply activated following the induction of OA synovial fibroblasts with pro-inflammatory cytokines. MALAT1 silencing has reduced levels of CXCL8 in OA synovial fibroblasts while increasing TRIM6, IL7R, HIST1H1C, and MAML3 levels. Moreover, MALAT1 silencing has suppressed the proliferation of synovial fibroblasts of obese OA patients (14). NEAT1 and XIST are among other lncRNAs whose contribution to the pathogenesis of OA has been vastly investigated. Nevertheless, the results of the studies of their expression patterns are not consistent. In this section, we summarize studies that reported their up-regulation in OA. NEAT1 was described as a regulator of the OA development through sponging miR‐193a‐3p, miR-543, and miR-377-3p, thus increasing expressions of SOX5 and PLA2G4A. Subsequently, it affects chondrocyte proliferation and apoptosis and extracellular matrix (ECM) degradation (15–17). XIST increases OPN levels while decreasing miR-1277-5p, miR‐142-5p and TIMP-3 levels. Changes in these genes’ expression result in ECM destruction, induction of inflammatory responses, and abnormal proliferation/apoptosis of chondrocytes (18–21). Table 1 shows the results of studies that demonstrated up-regulation of lncRNAs in OA tissues.

While most of the studies which assessed the expression of NEAT1 or XIST in OA samples reported their up-regulation, few studies have demonstrated the opposite trend in their expressions. Lian et al. have reported down-regulation of XIST in OA chondrocytes. They have also shown protective effects of XIST in chondrocytes against IL-1β-induced damage through modulating the miR-653-5p/SIRT1 axis (42). Wang et al. have demonstrated down-regulation of NEAT1 in OA tissues, parallel with up-regulation of miR-181a. They have also shown that down-regulation of NEAT1 suppresses cell growth, elevates apoptosis, and increases the production of pro-inflammatory cytokines in OA chondrocytes (43). SNHG7 is another down-regulated lncRNA in OA. This lncRNA sponges miR-34a-5p and miR-214-5p (44, 45). Due to its sponging effects on miR-34a-5p, down-regulation of SNHG7 results in down-regulation of SYVN1, the direct target of this miRNA. Up-regulation of SNHG7 enhances cell proliferation and suppresses apoptosis and autophagy in OA cells (44). Also, SNHG7 can enhance cell viability and inhibit apoptosis and inflammatory responses in IL-1β-mediated chondrocytes through sponging miR-214-5p and up-regulating PPARGC1B expression. Thus, the protective effects of SNHG7 against OA are exerted through induction of the PPARγ pathway and combating the cytotoxic impact of miR-214-5p (45). The protective effects of other lncRNAs against OA are summarized in Table 2 .

miRNAs are the utmost investigated small ncRNAs, representing an additional level of post-transcriptional controllers of gene expression that warrant the robustness of coordination in biological processes (67). These transcripts typically bind with the 3’ UTR of their target transcripts to either repress their translation or degrade them (68). In this section, we selected some miRNAs with clearer mechanistical information. Cheng et al. have reported up-regulation of miR-455-3p in the OA cartilages and IL-1β-exposed chondrocyte cells. This miRNA has been shown to partake in IL-1β-associated apoptosis and inflammatory responses. COL2A1 has been verified as a target of miR-455-3p designating the miR-455-3p/COL2A1 axis as a molecular mediator of OA (69). While confirming the role of miR-455-3p in OA’s chondrogenesis and development, Wen et al. have demonstrated down-regulation of this miRNA in the IL-1β model of OA. Over-expression of miR-455-3p has led to a significant decrease in PTEN and MMP13 while increases the COL2A1 expression levels. Moreover, based on their observations, miR-455-3p can decrease chondrocytes’ apoptotic rate by affecting PTEN expression (70). Despite using similar OA models, these studies have reported conflicting results regarding the role of miR-455-3p in the development of OA. Wand et al. have demonstrated up-regulation of miR-1236 in OA-affected cartilages compared to normal cartilages. Such up-regulation has inhibited chondrocyte proliferation and induced apoptosis in these cells through targeting PIK3R3 (71). miR-411 is another up-regulated miRNA in OA models. This miRNA directly affects the expression of HIF-1α. LC3, ULK-1, P62, and Beclin-1 have been among genes whose expressions have been affected by miR-411. miR-411 has been shown to enhance chondrocyte autophagy through modulating HIF-1α (72). miR-203 is another miRNA whose expression has been promoted by IL-1β stimulation. This miRNA enhances cellular inflammatory responses and cell damage and reduced aggrecan and Col2A1 levels. miR-203 binds with ERα and exerts its effects in OA development through this axis (73). miR-140 and miR-199 are two down-regulated miRNAs in the synovial tissues of OA patients compared with healthy controls. Expressions of these miRNAs have been shown to decrease during the course of OA. Moreover, their expressions have been inversely correlated with the severity of OA (74). The course of OA has been found to be alleviated by exosomal miR-9-5p produced by mesenchymal stem cells originated from bone marrow. This miRNA has been shown to decrease syndecan-1 levels and diminish pro-inflammatory cytokines as well as CRP (75). Tables 3 and 4 show the up-regulated and down-regulated miRNAs in OA, respectively.

These ncRNAs have a circular conformation shaped by routine spliceosome-mediated or lariat kind of splicing (85). Exonic circRNAs, circular intronic RNAs, exonic-intronic circRNAs, and tRNA intronic circRNAs constitute the main classes of circRNAs (85). Circ_0136474 is a member of this group which can inhibit cell proliferation by enhancing MMP-13 expression and decreasing miR-127-5p levels in OA (86). Hsa_circ_0005105 is another up-regulated circRNA in IL-1β-induced chondrocytes. Hsa_circ_0005105 has been shown to suppress transcriptional activity of miR-26a, thus up-regulating expression of NAMPT, the direct target of this miRNA. Moreover, hsa_circ_0005105 can decrease the levels of type II collagen and aggrecan, enhance MMP-13 and ADAMTS-4 levels, and increase the production of PGE2, IL-6, and IL-8 (87). CircHIPK3 is another circRNA that regulates the apoptosis rate of chondrocytes through the miR-124/SOX8 axis (88). On the other hand, circRNA-UBE2G1 mainly regulates OA development through influencing inflammatory responses. This circRNA targets miR-373 and increases IL-1β, IL-6, and TNF-α levels in LPS-treated cells (89). Tables 5 and 6 show the list of up-regulated and down-regulated circRNAs in OA, respectively.

In addition to cell line studies and expression assays in human samples, the expression and function of ncRNAs have been appraised in OA animal models. Commonly, the observed dysregulation of ncRNAs in the animal models is consistent with findings in human-derived OA tissues and in vitro studies. Moreover, deregulation of these transcripts has similar outcomes in the animals and in in vitro studies. In fact, animal studies have provided strong evidence for functionality of these transcripts in the pathogenesis of OA. As an example, HOTAIR has been shown to be elevated in articular cartilage samples of OA mice in association with down-regulation of miR-20b and up-regulation of PTEN. HOTAIR knockdown has ameliorated cartilage tissue injury in animal models and enhanced collagen II and aggrecan levels in this tissue while decreasing MMP-13 and ADAMTS-5 levels (96). LOC101928134 and LINC00662 are two other lncRNAs whose functions in OA development have been investigated in animal models (97, 98). While LOC101928134 increases apoptosis and cartilage damage through activation of the JAK/STAT signaling pathway (97), LINC00662 reduces apoptosis and inflammatory factors such as IL-6 and IL-8 (98). miR-34a, miR−363−3p, miR‐101a‐3p, circRNA.33186, and circRNA_Atp9b are other ncRNAs whose roles in the development of OA have been appraised in animal models ( Table 7 ).

NcRNA can affect pathoetiology of OA through different routes such as JAK/STAT, NF-κB, PI3K/AKT and Wnt/β-catenin signaling pathways as well as autophagy. It is worth mentioning that a single ncRNA might affect pathogenesis of OA through modulation of different pathways. For instance, HOTAIR has been found to affect both PI3K/AKT signaling and autophagy, thus it has a pleiotropic role in OA. In fact, these effects might be complementary to each other to worsen disease progression. Similarly, the same signaling pathway can be affected by many different ncRNAs at different points. One might deduce that these ncRNAs act in a timely-concerted manner, yet no study has assessed the effects of these ncRNAs at different regulatory points of signaling pathways or during the course of OA. Thus, there is no proof for this hypothesis based on the currently available literature.

Activation of JAK/STAT signaling pathway acts as a common connection linking pro-inflammatory cytokines to inflammation in the context of OA (104). In addition, expression of the NF-κB family of transcription factors can be induced by pro-inflammatory cytokines and chemokines as well as degradation products of extracellular matrix. Activation of NF-κB molecules can increase expression of several genes which increase damage to the articular joint, thus participating in the pathogenesis of osteoarthritis (105). A number of ncRNAs can affect pathogenesis of OA via modulation of these pathways. Figure 1 illustrates the role of various ncRNAs in regulating the JAK/STAT and NF-κB signaling pathways in OA.

The abnormal alterations in the course of osteoarthritis mostly are linked with dysfunction of chondrocytes and autophagy, an intracellular mechanism of degradation that preserves the stable condition of cellular metabolism. This process is also regarded as a mechanism for restoring activity of injured chondrocytes. Thus, it has a role in alleviation of OA (106). Figure 2 represents the role of several ncRNAs in OA through regulating the autophagy pathway.

Wnt signaling has an important role in osteoarthritis pathogenesis and is regarded as a potential target for treatment of this disorder. Wnt pathway contains 19 Wnt genes and several receptors regulating canonical and non-canonical pathways, the latter being dependent on β-catenin. Wnt signaling has important roles in the regulation of proliferation and differentiation of cells, as well as their polarization (107). PI3K/AKT/mTOR signaling pathway is another pathway which has important roles in the normal metabolic pathways in the joints and participates in the development of OA through induction of cartilage degradation, impairment of function of subchondral bones, and induction of inflammatory responses in the synovial tissues (108). Figure 3 illustrates the role of different ncRNAs in OA through modulating the PI3K/AKT and Wnt/β-catenin signaling pathways.

H19, MEG3, and PRNCR1 are three lncRNAs whose polymorphisms have been associated with OA’s risk. For instance, the A allele of the rs217727 within H19 can enhance the risk of OA. However, the rs3741219 within this lncRNA has not affected the risk. Notably, the rs217727 polymorphism has been associated with the levels of H19, hsa-miR-4804-5p, hsa-miR-8071, hsa-miR-8072, and hsa-miR-3960 in the circulation. Besides, the A allele of the rs7158663 within MEG3 can increase the risk of OA. rs7158663 has been associated with the plasma levels of its host genes, i.e. hsa-miR-4307 and hsa-miR-1265 (109). Finally, the G allele of rs1456315 within PRNCR1 can increase the risk of OA (110).

An integrative analysis of lncRNAs association with OA has shown that RegulomeDB scores of three SNPS within H19, MEG3 and HOTTIP have been 2b (111). Based on the ChIP-seq data, these SNPs can bind with EZH2, E2F6, REST and IKZF1 proteins (http://regulome.stanford.edu/) (111). Previous studies have shown the involvement of these proteins in the pathogenesis of OA or regulation of cellular functions. For instance, suppression of EZH2 can ameliorate development of OA via modulation of Wnt/β-catenin pathway (112). E2F6 is involved in the regulation of cell cycle (113). REST encodes a transcriptional repressor which suppresses neuronal genes in non-neuronal tissues (114). IKZF1 is involved in the chromatin remodeling (115). Its role in the regulation of inflammation implies its involvement in the pathogenesis of OA (116).

Although the mechanisms behind involvement of these SNPs in the pathogenesis of OA have not been completely understood, it is possible that these SNPs affect interaction with other target RNAs. Moreover, they can influence expression levels of ncRNAs, thus affecting their regulatory effects.

Identification of risk variants for development of OA can help in development of novel OA therapeutic approaches such as gene editing or gene replacement therapies for OA. Table 8 summarizes these studies.

Recent investigations have tested the appropriateness of ncRNAs in diagnostic purposes in OA. Circulating ncRNAs (in blood/plasma/serum) are definitely one of the most interesting biomarkers for OA due to the easy accessibility of sample. Although synovial fluid samples have also been applicable for this purpose, blood/plasma/serum samples are superior since they are obtained through less invasive methods. For instance, expression levels of H19 in the blood samples could distinguish OA cases from normal subjects with AUC, critical, sensitivity, and specificity values of 0.89, 1.87, 96%, and 85.7%, respectively (9). In a study with limited numbers of cases and controls, GAS5 has been shown to predict the presence of OA with an accuracy of 0.86 (40). The highest diagnostic power among lncRNAs has been achieved by MIR4435-2HG (AUC=0.96) (50). Hsa_circ_0032131 is the only circRNA whose appropriateness for diagnostic strategies in OA has been appraised (117). Table 9 gives an overview of the diagnostic impact of ncRNAs in OA, based on the studies that assessed expression of these transcripts in the circulation of patients. These ncRNAs are merely exemplificative of the work being published in the field.

OA is a multifactorial disorder in which several classes of ncRNAs, including lncRNAs, circRNAs, and miRNAs participate. Notably, the two former classes of ncRNAs mainly exert their effects in this process through acting as molecular sponges for miRNAs. These ncRNAs collaborate to influence chondrocyte proliferation and apoptosis, inflammatory responses, and degradation of ECM. Studies that investigated ncRNAs’ role in OA can be classified according to their design to in vitro studies, expression assays in clinical samples, and functional studies in animal models. The latter type of studies has provided essential concepts about the role of ncRNAs in this process, as it could assess these transcripts’ functional roles in a natural context. Meanwhile, clinical studies, particularly those assessing expression levels of ncRNAs in the peripheral blood, have the advantage of discovering appropriate markers for the diagnosis of OA and prediction of its course.

NcRNAs can be involved in the fine tuning of the RUNX2 expression and through this rout, they can affect pathogenesis of OA (119, 120). NF-κB, Wnt/β-catenin, TGF-β and JAK/STAT pathways are the most critical pathways through which ncRNAs exert their effects in the pathogenesis of OA. Based on the functional relevance of these pathways with inflammatory responses, one can conclude that this process has a prominent role in the development of OA. A PPAR-α agonist has been found to inhibit LPS-associated inflammatory responses in synovial fibroblasts through modulation of NF-κB signaling (121). Therefore, ncRNAs associated with these pathways might also represent therapeutic targets for OA.

The cartilage tissue has been mainly studied for the assessment of the ncRNAs’ expression. However, limited numbers of studies have investigated the expression of these transcripts in patients’ synovial membrane or peripheral blood, evaluating their continuation as non-invasive markers for the detection of OA. These studies have reported diagnostic power values ranging from 0.80 to 0.96. Nevertheless, most of these studies have been conducted in limited numbers of cases and controls, precluding the generalization of their results.

The data presented above shows involvement of several ncRNAs in the pathoetiology of OA. This information can be used for design of novel therapeutic options for this disorder. Moreover, it can be used to find genetically susceptible people to OA. However, further assessment of applicability of ncRNAs-targeting treatment modalities in animal models is a prerequisite for translation of this filed of basic science into clinical application.

Despite valuable research, this field lacks a comprehensive assessment of different classes of ncRNAs in OA samples. Such study would increase our understanding of the functional relationship between circRNAs, lncRNAs, and miRNAs, thus expanding our knowledge about the pathobiology of OA.

Another gap in this field is the scarcity of assessment of the impact of functional polymorphisms within ncRNAs in conferring OA risk and modulating the disease course. Identification of genomic variants that affect the risk of OA would help in the modification of lifestyle in order to attenuate the course of the disorder.

SG-F wrote the draft and revised it. MT designed and supervised the study. CP and MM revised the draft. AT collected the data and designed the figures and tables. All the authors read and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.