Osteoarthritis and microRNAs: Do They Provide Novel Insights into the Pathophysiology of This Degenerative Disorder?

Osteoarthritis (OA) is one of the most prevalent degenerative joint diseases in older adults and a leading cause of disability. Recent research studies have evidenced the importance of mi-croRNAs (miRs) in the pathogenesis of OA. In the present review, we focused on current literature findings on dysregulated miRs involved in the pathophysiology of OA. From the 35 case-control studies including OA patients compared to healthy controls, a total of 54 human miRs were identified to be dysregulated in OA. In total, 41 miRs were involved in the pathophysiological processes of OA, including apoptosis, inflammation, and proliferation, having either a protective or a progressive role in OA. The discovery of altered miR levels in OA patients compared to healthy controls determines a better understanding of the molecular mechanisms involved in the pathophysiology of OA and could open novel horizons in the field of orthopedics.


Introduction
Osteoarthritis (OA) is the most prevalent degenerative joint disease in older adults and a leading cause of disability [1]. Progressive breakdown of articular cartilage, remodeling of subchondral bone, chondrocyte hypertrophy, and synovium inflammation are clinically translated into pain, joint deformity, and malfunction of the affected joint [2]. Multiple etiological factors such as age, obesity, joint injury, mechanical stress, inflammation, and genetic factors are related to OA [3,4]. As several studies have already demonstrated, primary OA has a strong genetic component [5]. Recent evidence has made it noteworthy that epigenetic changes, altered expression of regulatory RNA, and its resulting gene expression modifications could also be involved in the biology of OA [6].
Epigenetics is the study of heritable phenotype changes that do not involve alterations in the DNA sequence [7]. Besides DNA methylation and histone modifications, microRNAs (miRs) are also linked with epigenetic mechanisms. Since the first human miR discovery, interest in this research field has grown tremendously, leading to an intense genomic investigation. MiRs are a class of small double-stranded non-coding RNAs, with a length of <25 nucleotides. They are transcribed in the nucleus to generate primary precursor molecules (pri-miRs), that undergo nuclear cleavage to form a precursor miR (pre-miR). The pre-miR is exported in the cytoplasm to create a miR duplex (miR:miR*) containing the Life 2022, 12,1914 2 of 14 mature miR. The miR base-pairs target messenger RNA (mRNA) to direct gene silencing via mRNA cleavage or translation repression based on the level of complementarity between the miR and the mRNA target [8]. Therefore, miRs regulate gene expression at the posttranscriptional level through translational inhibition or degradation of mRNAs [9].
In the past years, miRs were identified as being involved in various processes of cellular metabolism such as differentiation, proliferation, survival, apoptosis, embryogenesis, inflammation, and immune response [10]. They also appear to have a key role in tissue homeostasis [11] and in the pathogenesis of diseases and are dysregulated in several types of cancers [12], allergies [13], kidney disorders [14], psoriasis [15], psychiatric disorders [16], nervous system [17], and inflammatory diseases [18]. In OA in particular, the main function of miRs is to reduce the stability and translation level of their target mRNA involved in the pathophysiological processes, which includes inflammation, apoptosis, matrix synthesis, and chondrogenesis. MiRs seem also to modulate extracellular matrix deposition, and suppress or accelerate chondrocyte apoptosis [19,20]. A large number of studies have demonstrated the link between miRs and the pathogenesis of OA through the regulation of chondrogenic processes [21]. An imbalance between the catabolic and anabolic factors may be caused by abnormal expression of miRs, leading to cartilage degradation when cartilage homeostasis is disturbed [22].
Nonetheless, the exact biochemical mechanism of miRs in OA remains controversial, and information on the effects of these miRs are limited; therefore, in the present review, we focused on the current status of literature research on miRs involved in OA and reviewed the dysregulated miRs in OA, their function, and effective roles for a better understanding of OA pathogenesis.

Search Strategy and Study Selection
To find relevant articles dated through December 2021, a systematic search was performed in the PubMed, Web of Science, and Google Scholar databases using the following combination of keywords: "microRNAs" or "miRs", "osteoarthritis", "target gene", "function", "roles", "pathophysiology" (all fields)" with the following enabled parameters: human, English language, articles published from 2010 to present. After searching the databases electronically, two independent reviewers performed an initial sort of the articles based on the titles. Duplicate references were excluded. In a second step, the abstracts and methodology of the studies were screened based on the inclusion and exclusion criteria.

Inclusion and Exclusion Criteria
The research articles inclusion criteria were: (1) case-control studies of subjects diagnosed with OA compared to healthy controls; (2) studies profiling miR expression levels of patients with OA and controls in human cartilage collected by surgery; (3) studies reporting cut-off criteria for determining whether miRs were differentially expressed between the OA and normal control; (4) original articles, published in the English language.
The research articles exclusion criteria were: (1) studies not conducted in human subjects (animal studies, in-vitro studies); (2) studies assessing other miRs, such as circ-miRs, lncRNA etc.; (3) non-original research papers, such as reviews, meta-analysis, pilot studies, abstracts, letters to editors; (4) duplicate studies; (5) studies written in other languages than English.

Data Extraction
At the initial search, 1045 articles were found. Considering the preliminary review of the titles and abstracts and excluding the irrelevant articles, a total of 203 studies were considered relevant. The validation process was determined by the authors after reviewing each full-text article separately. After full-text reading, 168 articles with irrelevant data were removed. In the end, a total of 35 studies published between 2010 and 2021 were selected to be included in the present literature review. Figure 1 presents a flowchart of the Life 2022, 12,1914 3 of 14 review selection process. After applying the eligibility criteria, two reviewers individually extracted the following information from the resulting studies: author, name of the study, the studied miRs and their direction of dysregulation (up-or downregulated), gene target and miRs function in OA.

Data Extraction
At the initial search, 1045 articles were found. Considering the preliminary review of the titles and abstracts and excluding the irrelevant articles, a total of 203 studies were considered relevant. The validation process was determined by the authors after reviewing each full-text article separately. After full-text reading, 168 articles with irrelevant data were removed. In the end, a total of 35 studies published between 2010 and 2021 were selected to be included in the present literature review. Figure 1 presents a flowchart of the review selection process. After applying the eligibility criteria, two reviewers individually extracted the following information from the resulting studies: author, name of the study, the studied miRs and their direction of dysregulation (up-or downregulated), gene target and miRs function in OA.

Results
From the 35 included studies, a total of 54 human miRs were identified to be related to OA. Table 1 presents for each study the analyzed miRs and their dysregulation (up-regulated, down-regulated, or unchanged miRs) in patients suffering from OA compared to healthy controls. From the 54 differentially expressed miRs reported in the included studies, 25 were upregulated, 33 were downregulated, and 4 miRs had bidirectional dysregulation, both up-and downregulated (shown in bold).

Results
From the 35 included studies, a total of 54 human miRs were identified to be related to OA. Table 1 presents for each study the analyzed miRs and their dysregulation (upregulated, down-regulated, or unchanged miRs) in patients suffering from OA compared to healthy controls. From the 54 differentially expressed miRs reported in the included studies, 25 were upregulated, 33 were downregulated, and 4 miRs had bidirectional dysregulation, both up-and downregulated (shown in bold).
All miRs involved in all three processes (apoptosis, inflammation, and proliferation) can be seen in Figure 2.
From the total 19 miRs involved in apoptosis, only three of them were specific for apoptosis, 10 miRs were common with the proliferation process and 9 were also involved in inflammation. 16 miRs were involved in proliferation. Only 3 miRs were specific for proliferation and 6 miRs were common with inflammation. A total number of 23 miRs were related to inflammation, out of which only 11 were specific, and the other miRs were common for apoptosis and proliferation, as seen in Table 4. From the 19 miRs involved in apoptosis (Table 1), 9 were also found to be involved in OA-related inflammation (miR-195-5p, -335-5p, -9, -145, -27b-3p, -125b, -140-5p, -455-3p, -142-5p, respectively), for which the information regarding the induced effect and signaling pathway can be found in Table 1. The remaining 14 miRs are only listed in Table 2, unfortunately with no available information regarding their pathophysiological effects (inflammation) in OA.
All miRs involved in all three processes (apoptosis, inflammation, and proliferation) can be seen in Figure 2.
From the total 19 miRs involved in apoptosis, only three of them were specific for apoptosis, 10 miRs were common with the proliferation process and 9 were also involved in inflammation. 16 miRs were involved in proliferation. Only 3 miRs were specific for proliferation and 6 miRs were common with inflammation. A total number of 23 miRs were related to inflammation, out of which only 11 were specific, and the other miRs were common for apoptosis and proliferation, as seen in Table 4.

Effective Roles of miRs in OA Pathophysiology
MiRs involved in OA pathophysiology (apoptosis, inflammation, and/or proliferation) seem to have either a protective or destructive role, the latter leading to the progression of the disease. Table 5 shows the miRs and their positive or negative role in the pathophysiological process of OA. Inconsistencies across studies reveal that three miRs (miR-140-5p, miR-140-3p, and miR-146a, respectively) could have both protective and destructive roles in OA (shown in bold).  Figure 3 shows that 23 miRs have only a protective role in OA and 25 miRs were involved only in the progression of OA. Three miRs can have both a protective and progressive role in OA: miR-140-3p, miR-140-5p, and miR-146a.

Discussion
In this study, we investigated the association between miRs and the pathophysiology of OA. Regarding miRs' expression level, we found that miRs are generally more downregulated than upregulated in OA patients' samples (articular cartilage, bone, synovial fluid and tissue, plasma, peripheral blood mononuclear cells (PBMCs), and whole blood) relative to healthy controls. MiRs seem to be involved in various pathophysiological processes of OA as well, such as apoptosis, inflammation, and proliferation; the great majority being involved in the inflammation process, and three miRs (miR-195-5p, miR-335-5p and miR-140-5p) being involved in all three processes. In addition, when studying their effector roles, miRs are displayed as having either protective roles, destructive roles, or (interestingly) both for miR-140-3p, miR-140-5p, and miR-146a.
It has previously been found that increased risk for OA development is associated with genetic and environmental factors, mechanical stress, fluid pressure, joint trauma, and inflammation. Several biological processes are involved in cartilage homeostasis, including proliferation, differentiation, inflammation, and apoptosis. Many of these processes appear to be regulated by different miRs. For example, miR-675-3p [23] and miR-203a [24] have been described as reducing inflammation and apoptosis. The proliferation of chondrocytes is also essential for maintaining the integrity of the articular cartilage [36]. In this context, recent studies have linked dysregulated expression of some miRs to several cellular processes, such as cartilage cell proliferation and differentiation. MiR-26a/b [37], miR-335-5p [25], miR-34a [38], miR-92a-3p [39], and miR-146a [26] seem to be involved in the pathogenesis of OA by regulating the expression of target genes related to chondrocyte proliferation. Namely, the downregulation of miR-197 in OA cartilage acts as a signal for chondrogenesis and promotes the proliferation and migration of cartilage cells by direct targeting of EIF4G2 [27]. Overexpression of miR-103 also promotes OA development by targeting and inhibiting the SOX6 gene. This miR has also been shown to inhibit chondrocyte proliferation and suppresses chondrocyte formation and maturation by reducing the expression of collagen II, glycosaminoglycan, and aggrecan [40]. In addition, up-regulation of miR-20 inhibits chondrocyte proliferation by directly targeting the ATG10 gene [28]. Duan et al. showed that miR-15a-5p promotes chondrocyte degradation by regulating the expression and function of PTHRP [29].
Furthermore, chondrocyte apoptosis has also been shown to be part of OA pathogenesis, with apoptotic chondrocytes being more abundant in damaged cartilage than in smooth cartilage [44,45]. In this regard, upregulation of miR-1236 promotes OA progression through induced apoptosis and inhibition of chondrocyte proliferation [33]. Overexpression of miR-337-3p and of miR-455-3-p were shown to promote proliferation and inhibit apoptosis of chondrocytes in OA [46,47].
In addition to articular cartilage damage, OA is also associated with other pathological changes, such as synovial inflammation. Synovitis often occurs at different stages of OA and is associated with pain and joint dysfunction [48]. In synovial fibroblasts, upregulated miR-261-5p increased apoptosis, reduced inflammation, and decreased injury in vivo and in vitro [49]. MiR-16 and miR-132 were reported to be less expressed in synovial fluid of OA than in controls [50]. Let-7e, miR-454, and mir-885-5p [34] have even been identified as potential serum biomarkers for severe OA detection associated with synovitis. Synovitisrelated, downregulation of miR-33b-3p, miR-140-3p, and miR-671-3p in serum OA has also been reported [35]. In contrast to previous studies, one study reported no differences in plasma miRs levels between OA patients and healthy controls [51]. The author reports some limitations to his study that may affect the final results, such as a prolonged freezing time period (up to 7 years) and the use of a lower amount of total RNA than is normally required by deep sequencing techniques.
Amongst the 35 studies included in the present review, only 4 of them were performed in biological tissues other than articular cartilage or bone of patients with OA. In general, the analyzed compartments were plasma, peripheral mononuclear blood cells (PBMCs) and synovial fluid. Interestingly, some miRs have been shown to have a heterogenous dysregulation among different tissues. For example, miR-146a was upregulated in PBMCs [52,53] and downregulated in chondrocytes [54]. By contrast, miRs derived from plasma and synovial fluid have been shown to have constant dysregulations: miR-155 [52], miR-34a-5p [38,55] and miR-16-5p [51] were upregulated in the same direction as the miRs derived from osteoarthritic cartilage, as shown in the study performed by Fan et al. for miR-155 [56], by Endisha H et al. for miR-34a-5p [38] and by Li L et al. for miR-16-5p [57].
The literature contains little information regarding the role of miRs as biomarkers for OA and their identification in comparison with other evaluation methods, such as radiographic imaging or MRI. The lack of evidence relies on the fact that miR levels are usually altered by coexisting comorbidities. In order to act as a biomarker for diagnosis, a miR should have high specificity and sensitivity and its analysis must be done as minimally invasive as possible. Among the advantages of miRs as biomarkers are high reproducibility (using RT-qPCR), their identification in body fluids, and their binding to proteins or exosomes, thus having a high degree of stability [58]. Some of the comorbidities display the same miRs as OA, thus reducing their sensitivity and specificity. Due to this fact, the role of miRs as biomarkers, especially in the early stages of both OA and early stages of other comorbidities is diminished since the miR concentration can lead to a false interpretation of the analysis, leading to a wrong conclusion and thus overlooking some of the pathologies and delaying their diagnosis and treatment. For example, high levels of miR-146a can also be identified in Type I and II Diabetes Mellitus and their associated complications and it is related to tumoral involvement in regulating metastatic turnover [59]. Furthermore, most of the studies are focused on miR concentration in synovial fluid once the OA process is ongoing, but their reliability as screening markers is diminished at the moment. Rousseau et al. conducted a study in 2020 and showed that miR-146 and miR-186-5p can be used as sensitive biomarkers in OA. Moreover, miR-146 levels were higher in grades 3 and 4 of OA, Life 2022, 12,1914 10 of 14 while miR-186-5p was a consistent biomarker in preclinical OA and its presence showed that the patients can develop clinical OA in the following four years [60].
As presented, miR related to OA can be found in blood, plasma, and synovial fluid; therefore, these fluids become the target for miR sampling. Many miRs can be found in various combinations between fluids, but there is no miR that can be found in every fluid. MiR-146-a was found in plasma and peripheral blood, miR-27 in synovial fluid; miR-27a-3p, miR-101-5p, and miR-378-5p were detected in synovial fluid and their detection was correlated with the late stages of OA [61]. Furthermore, Tavallaee et al. concluded that high concentrations of 23a-3p, miR-24-3p, miR-186-5p, miR-29c-3p, miR-34a-5p, and miR-27b-3p can be identified in synovial fluid in late-stage OA. Moreover, high levels of miR-155, miR-146A, miR-181a, and miR-223 were found in PBMCs when compared to healthy individuals [62,63]. Despite the fact that many miRs can be found in both synovium and serum, their display in serum is recognized as having inferior sensitivity. Xie et al. concluded that the serum-derived exosomes contain less miR information than synovial fluid exosomes; thus serum samples might be insufficient for an accurate identification and stratification risk. However, some clinicians might be reluctant at taking synovium samples, since it is a more invasive technique and it can lead to serious infectious complications and the risk might overcome the benefit; so the first choice for miR sampling might be serum taken from the peripheral blood [64].
In addition, miRs also represent a potential target for future therapies. At the moment their therapeutic role is very limited due to several factors: the first is finding the appropriate therapeutic agent to exert its effect upon a specific miR; second, an efficient delivery system that is minimally invasive and is highly efficient; currently the preferred delivery systems are parenteral or local route, miRs having a reduced absorption via oral administration. Lipid-based vesicles are used for delivery since they can easily cross the plasmatic membrane of the cells and have great tolerability and low cytotoxicity, but there are other available delivery systems such as nanoparticles, but with less efficiency and they present a certain degree of antigenicity; third, a therapeutic agent that has as few interactions as possible with other medications, has an acceptable level of toxicity, and has high bioavailability. Since one miR targets several genes, adverse effects must be taken into consideration when a specific miR is selected as a therapeutic target. In this case, local delivery (intraarticular) might be beneficial since the systemic effects should be reduced, and also a higher quantity of medication might reach the injury site without diluting it in the bloodstream. As presented, the levels and the type of miRs involved in OA differ as the pathology evolves. Specific therapeutic agents might not be as effective in the late stages as in the early stages because the levels of the targeted miRs change and the therapeutic effect might be diminished. Consequently, the therapy must be adapted accordingly to the present levels of miR, not only to have maximum efficiency, but also to avoid unnecessary side effects. The adjustment of the therapy requires continuous evaluation of miR levels [65].
Furthermore, other authors have reported similar reviews to our own, such as the study conducted by , where they mainly focused on the role of miRs in relationship with the pathogenesis of OA (via the regulation of articular chondrocytes' gene expression), and on the potential of miRs as future therapeutic targets and diagnostic biomarkers for this degenerative disorder [66]. Mihanfar et al. (2020) conducted a review on the roles of exosomal miRs in OA, therefore focusing only on one type of sample specimen, possibly to reduce potential heterogeneity, and presenting only those miRs which were found to be encapsulated in microvesicles such as exosomes [67].
Nonetheless, there is one report assessing the roles of non-coding RNAs in different diseases with a focus on OA, but the authors merged together different molecules (such as miRs, long non-coding RNAs, circular RNAs, piwi-interacting RNAs, and small nucleolar RNAs) in a larger, more extended review, but less specific than our work [68]. On the other hand, Wang et al. (2022) reported a review on the role of long non-coding RNAs in OA, without mentioning any miR in their study [69]. Similarly, there is another report conducted by Kim et al. (2022) assessing the role of circular RNAs in knee OA, in an attempt to predict potential future therapeutic targets for OA [70]. Table 6 shows comparative data of common miRs found in both our research and different studies across the literature. Some of the miRs display different regulations, suggesting that there are many individual factors that contribute to their regulation. ↓ ↑↓ [67] * designates those miRs which are consistent in our review compared to other review reports; "↑" symbol means upregulated and "↓" means downregulated (referring to miR gene expression).
Taken together, our review has several limitations worth acknowledging. Firstly, given the heterogeneity of the sample specimens that were used across studies, the direction of the expression level of the reported miRs was taken with respect to the highest trend. Secondly, due to a large degree of inconsistency across reports, we only listed the miRs with potential roles in the pathophysiology of OA, excluding from the table exact mechanisms in general, which were generally conflicting among reports. Thirdly, in vitro and animal model studies were excluded because of the biological differences between different animal species (mice, rats, rabbits, etc.) and human samples. Thus, differences in study designs, methodologies, and choice of statistics introduce a large degree of heterogeneity and bias across research studies, which makes the design of a review report a laborious task, and therefore the implementation of consensus regarding methodology guidelines and protocols is highly warranted.

Conclusions
In conclusion, understanding the exact roles and mechanisms of non-coding RNA species (such as miRs) in the pathogenesis of OA could bring new insights into disease management and future therapeutical approaches and improve the overall quality of life for patients suffering from this degenerative disorder.