Review: The Role of MicroRNAs in Osteoarthritis and Chondrogenesis

Authors


Introduction

The etiology of osteoarthritis (OA) is complex, with genetic, developmental, biochemical, and biomechanical factors contributing to the disease process. Chondrocytes in articular cartilage must express appropriate genes to achieve tissue homeostasis, and this is altered in OA. One facet of the aberrant gene expression in OA is the replay of chondrocyte differentiation with the expression of genes associated with chondrocyte hypertrophy. The pattern of gene expression and the transcription factors that control chondrogenesis are known in some detail. Mechanisms that lead to altered gene expression in OA, however, are less well understood.

MicroRNAs (miRNAs) are small noncoding RNAs that have recently been recognized as important regulators of gene expression in human cells. A number of miRNAs are regulated across chondrogenesis, and their function is beginning to be delineated. Similarly, miRNAs are differentially expressed in OA cartilage compared to normal tissue. MicroRNA-140 (miR-140), which is highly and selectively expressed in cartilage, has been the focus of much work to date, though the full gamut of its actions is still to be defined. Many other regulated miRNAs likely act as a network to control cartilage homeostasis, catabolism, and repair.

Chondrocytes are the sole cell type in cartilage, and they produce and maintain the extracellular matrix (ECM) that gives the tissue its load-bearing function ([1]). Chondrocytes originate from mesenchymal stem cells (MSCs) through chondrogenic differentiation ([2]). Investigation of the mechanisms mediating chondrogenic differentiation of MSCs as well as regulation of their functions will contribute to a better understanding of skeletal development and new strategies for treating diseases such as OA.

Chondrogenesis and chondrocyte function are highly regulated by transcription and growth factors. The SOX family members SOX9, L-SOX5, and SOX6 are necessary for chondrogenic differentiation (see, for example, ref.[3]). Many family members of the bone morphogenetic protein (BMP) and transforming growth factor β (TGFβ) signaling pathways have also been shown to control chondrogenesis ([2]). An additional level of regulation mediated by miRNAs has been identified ([4]), and miRNAs may represent novel therapeutic targets for pharmacologic control of skeletal diseases.

OA, the most prevalent degenerative joint disease, causes pain, tenderness, limitation of movement, and a variable degree of inflammation ([5]). OA is characterized by articular cartilage destruction due to an imbalance between the synthesis and degradation of ECM components, mainly type II collagen and the proteoglycan aggrecan. Matrix-degrading enzymes, e.g., the matrix metalloproteinases (MMPs) and ADAMTS, play important roles ([1]). The pathogenesis of OA is complex and poorly understood but involves the interaction of multiple factors, ranging from genetic predisposition to mechanical and environmental components ([5]). Studies are in progress to define molecular mechanisms underlying OA, including the roles of specific miRNAs in, e.g., phenotype shift, apoptosis, and regulation of gene expression in chondrocytes ([6]). MicroRNA-140, the major miRNA implicated in OA to date, plays a role in chondrogenesis and cartilage development ([7-9]). The knockout or overexpression of miR-140 in vivo has profound effects on the development of OA ([10, 11]). Other miRNAs appear to follow this pattern (see, for example, ref.[12]), but it is likely that further miRNAs that contribute to OA play a role in, for example, mechanotransduction or inflammation. The utility of miRNAs may be in the diagnosis of OA, tissue and cell engineering approaches, or in direct treatment of disease. A complete understanding of all facets of miRNA function in the joint is therefore needed ([4, 6, 13]). This review summarizes the current knowledge in this area that will inform future research.

The biology of miRNAs

In 1981, the first miRNA, lin-4, was discovered in Caenorhabditis elegans ([14]). In the early 1990s, Ambros and Ruvkun demonstrated that lin-4 controlled a specific step in development by down-regulating lin-14 (a conventional protein-coding gene) ([14-16]). They found that the lin-14 3′-untranslated region (3′-UTR) harbored multiple sites of imperfect complementarity to lin-4 and proposed that lin-4 bound to these sites and blocked lin-14 translation. Investigation of the genetic basis of phenotypes led to the identification of a second miRNA in C elegans, known as let-7 ([17]), which targeted lin-41 and hbl-1. The concept of miRNAs then jumped from worms to higher species, since let-7 had well-known homologs in humans and flies. In 2001, the term “microRNA” was coined for this class of noncoding gene regulators ([18-20]).

MicroRNAs are an abundant class of evolutionarily conserved, short (∼22-nucleotide long), double-stranded RNA molecules that have emerged as important posttranscriptional regulators of gene expression that function by binding to specific sequences within target messenger RNAs (mRNAs) ([21]). To date, 2,042 mature miRNAs have been identified in human cells (miRBase version 19; http://mirbase.org/), and each is predicted to regulate several target genes ([22]). Computational predictions indicate that >50% of all human protein-coding genes are potentially regulated by miRNAs ([23]). The abundance of mature miRNAs varies from as few as 10 to more than 80,000 copies in a single cell, providing flexibility in the regulation of gene expression ([24]). The regulation exerted by miRNAs is reversible, as indicated by the existence of feedback/forward regulatory loops ([25]).

Many known miRNAs are located in introns of protein-coding genes; a lower percentage of miRNAs originate from exons or noncoding mRNA-like regions ([26]). A significant number of miRNAs are found in polycistronic units encoding more than one miRNA, and these are often functionally related ([18, 19]).

Despite obvious differences between the biology of miRNAs and mRNAs, available evidence suggests that these transcripts share common mechanisms of transcriptional regulation ([27]). MicroRNAs are transcribed by RNA polymerase II ([28]) and are subsequently capped and polyadenylated ([29]). Transcription results in a primary transcript miRNA (pri-miRNA) harboring a hairpin structure ([27]). Within the nucleus, the RNase II–type molecule Drosha and its cofactor DGCR8 process the pri-miRNAs into 70–100-nucleotide-long precursor miRNA (pre-miRNA) structures ([30]), which in turn, are exported to the cytoplasm through the nuclear pores by exportin 5 ([31]). Subsequently, the RNase III–type protein Dicer generates a double-stranded short RNA in the cytoplasm ([32]). This duplex miRNA is unwound by a helicase into single-stranded short RNA in the cytoplasm. One strand of the miRNA duplex is selected as the guide miRNA and remains stably associated with the miRNA-induced silencing complex (mRISC), while the other strand, known as the passenger strand, can be rapidly degraded ([33]). The strand with less-stable base pairing at its 5′ end is usually destined to become the active strand ([34]). However, some miRNA passenger strands are themselves active and can negatively regulate gene expression. One hypothesis is that both strands could be used differently in response to extracellular or intracellular cues to regulate a more diverse set of protein-coding genes as needed, or strand selection could be tissue specific ([35]). The mature miRNA guides the mRISCs mainly to the 3′-UTR of its target miRNA ([36]). The seed sequence, comprising nucleotides 2–8 at the 5′-end of the mature miRNA, is important for binding of the miRNA to its target site in the mRNA ([36]).

In an alternative pathway for miRNA biogenesis, short hairpin introns, termed mirtrons, are spliced and debranched to generate pre-miRNA hairpin mimics ([37]). These are then cleaved by Dicer in the cytoplasm and incorporated into typical mRISCs. The presence of mirtrons may be an evolutionary strategy to diversify miRNA-based gene silencing ([38]).

Regulation is mainly exerted by the binding of the miRNA to the 3′-UTR of the target mRNA, but binding to other positions on the target mRNA, e.g., in the 5′-UTR or coding sequence, has also been reported ([39]). Interestingly, miRNA binding sites within the coding region of a transcript are reported to be less effective at mediating translational repression. Aside from the location of miRNA binding sites, factors including the sequence context of the miRNA binding site, the number of target sites within the mRNA, the focal RNA structure, and the distance between target sites all contribute to the efficacy of repression mediated by miRNAs ([40]).

The degree of base pairing between the miRNA and its target in the mRISC determines the fate of the transcript. If there is perfect complementarity between the miRNA and its target, the mRNA target is cleaved by Ago2 at the annealing site, with subsequent degradation of the mRNA. In contrast, when the miRNA is only partially complementary to its corresponding 3′-UTR (in almost all cases in mammals), inhibition of target mRNA translation occurs via Ago1. Translation inhibition may be exerted at many levels. The exact mechanism of translation remains a subject of controversy and may be, for example, assembly of the initiation complex, recruitment or dissociation of ribosome, inhibition of elongation, destabilization of target mRNA, sequestration in processing bodies, degradation of nascent protein, or reorganization of chromatin following gene silencing ([41]) (Figure 1).

Figure 1.

Biogenesis of microRNAs (miRNAs). The miRNAs are transcribed as primary transcript miRNAs (pri-miRNAs), which are processed by Drosha and its cofactor DGCR8 into short hairpin structures (precursor miRNAs [pre-miRNAs]). These are exported into the cytoplasm by exportin 5, where they are further processed by Dicer and TRBP (Dicer-TAR RNA binding protein) into an miRNA duplex. The duplex is unwound by a helicase and the “guide” strand is incorporated into the RNA-induced silencing complex (RISC) while the “passenger” strand (asterisk) undergoes degradation. This miRNA–RISC complex acts by two possible mechanisms, degradation of the target messenger RNA (mRNA) or translation inhibition. A, Degradation of the target mRNA occurs when the miRNA is nearly perfectly complementary to the 3′-untranslated region of the target mRNA. B, Translation inhibition occurs when the miRNA is only partially complementary to its target mRNA. Protein translation is a sequence of initiation, ribosome assembly, elongation, and termination, involving circularization of mRNA. MicroRNAs can regulate translation through several mechanisms, including upon the initiation process by preventing assembling of the initiation complex or recruiting the 40S ribosomal subunit (1), upon ribosome assembly by blocking the joining of the 60S subunit (2), upon the translation process (3), or upon the degradation of mRNA (4). eIF4f = eukaryotic initiation factor 4f.

Interestingly, miRNA binding to the 3′-UTR can also induce translation of some target mRNAs. MicroRNAs have been identified that activate translation upon cell cycle arrest by directing Ago-containing protein complexes to AU-rich elements in the 3′-UTR ([42, 43]).

MicroRNAs in skeletal development and chondrogenesis

Experimentally removing the majority of miRNAs through mutating or deleting Dicer reveals that the miRNA pathway plays a prominent role in skeletal development. Conditional knockout of Dicer in limb mesenchyme at the early stages of embryonic development leads to the formation of a smaller limb ([44]). Dicer-null growth plates display a lack of chondrocyte proliferation in conjunction with enhanced differentiation to postmitotic hypertrophic chondrocytes; the latter may be explained by Dicer having distinct functional effects at different stages of chondrocyte development ([44]). Recently, Kobayashi et al reported that Dicer-null chondrocytes in mice resulted in skeletal growth defects and premature death ([45]).

Wienholds et al showed that miR-140 was specifically expressed in the cartilage of the jaw, head, and fins in zebrafish during embryonic development ([46]). Later, Tuddenham et al found that miR-140 is specifically expressed in cartilage tissue during mouse embryonic development ([9]). Importantly, universal knockout of miR-140 led to mild dwarfism, probably as a result of impaired chondrocyte proliferation ([10, 11]). Recently, we found that miR-455-3p expression was restricted to the cartilage and perichondrium of the developing long bones in chicks and to the long bones and joints in mouse embryos ([12]).

Differential disruption of the Dicer gene in mice suggests that miRNAs play a significant role in chondrogenic differentiation. Many studies have profiled the expression of miRNAs to investigate their function in differentiating MSCs and have shown altered expression upon differentiation into chondrocytes ([7, 8, 47-52]) (Table 1). There was no consensus among those studies regarding the expression signature of miRNAs, and this is likely due to differences in experimental design, including different inducers of differentiation, cell types, numbers of miRNAs detected, probes, and organisms (Table 1).

Table 1. Studies examining miRNA profiles in MSCs and chondrocytesa
 Sorrentino et al, 2008 ([47])Suomi et al, 2008 ([48])Miyaki et al, 2009 ([8])Lin et al, 2009 ([49])Lin et al, 2011 ([7])Yang et al, 2011 ([50])Yan et al, 2011 ([51])Karlsen et al, 2011 ([52])
  1. aMSCs = mesenchymal stem cells; TGFβ3 = transforming growth factor β3; BMP-2 = bone morphogenetic protein 2; BM-MSCs = bone marrow–derived MSCs; qPCR = quantitative polymerase chain reaction.
  2. bPassenger strand of the microRNA (miRNA).
StimulatorsTGFβ3BMP-2; TGFβ3BMP-2TGFβ3
CellsBM-MSCsBM-MSCsBM-MSCsC2C12 cellsDedifferentiated articular chondrocytesBM-MSCsBM-MSCs; articular chondrocytesArticular chondrocytes
OrganismHumanMouseHumanMouseHumanMouseMouseHuman
Probe number22635380875
Cutoff, fold1.31.542
PlatformMicroarrayqPCRMicroarrayMicroarrayMicroarrayMicroarrayMicroarrayMicroarray
Up-regulated miRNAs31, 32, 136, 146, 149, 185, pre-miR-192, 199a-2-5, 204, 212, pre-miR-212, pre-miR-21424, 101, 124a, 199b, 199a15b, 16, 23b, 27b, 140, 148, 197, 222, 328, 505199b, 221, 298, 374, let-7e26a, 140b, 140, 222, 320a, 320d, 491b, 547-5p, 720, 1308, let-7d, let-7f, let-7a30a, 81a-1, 99a, 125b, 127, 140, 140b, let-7f21, 22, 27b, 27a, 140, 140b, 152, 291bb, 330, 431, 433, 455, let-7b, let-7d, let-7l30d, 140b, 210, 451, 563
Down-regulated miRNAs10a, 10b, 21, 23a, 24-1-3p, 24-2, 26b, 29b, 30a-5p, 34, 100, 103-2, 107, 130a, 138-1, pre-miR-143, 145, 181a-1, 191-5p, let-7a-1, let-7a-2, let-7a-3, let-7c, let-7d18, 9621, 125a, 125b, 143, 145, 21018a, 27a, 146a, 193b, 220b, 342-5p, 335, 365, 519e, 548e, 1248, 1284125bb, 132, 143, 145, 2121, 23a, 23b, 24, 26b, 99a, 99b, 99bb, 125a-5p, 143, 144, 145, 146a, 181a, 181d, 191, 199a, 199ab, 210, 320, 355-5p, 431, 503, 652, let-7a, let-7c, let-7g, let-7f15b, 31, 132, 138, 143, 145, 221, 222, 379, 382, 432, 494, 654b, 1308, let-7e

Regulation of chondrogenesis from murine MSCs in response to stimulation by TGFβ3 has been investigated ([48, 50]) (Table 1). Suomi et al compared the expression of 35 miRNAs in chondroblasts derived from MSCs, and found that miR-199a and miR-124a were strongly up-regulated, while miR-96 was substantially suppressed ([48]). They demonstrated that miRNAs and transcription factors could fine-tune cellular differentiation by showing that miR-199a, miR-124a, and miR-96 can target hypoxia-inducible factor 1α, regulatory factor X1, and SOX5, respectively ([48]). Similarly, Yang et al revealed 8 significantly up-regulated and 5 down-regulated miRNAs in this process ([50]). The miRNA clusters miR-143/145 and miR-132/212 were down-regulated, while miR-140-3p was the most up-regulated ([50]). Similar expression patterns for miR-145 and miR-143 were also described in other studies ([7, 49, 51, 52]). Corresponding targets of these differentially expressed miRNAs were predicted, including ADAMTS-5 (miR-140-3p), activin receptor 1B (ACVR1B) (miR143/145), Smad family members Smad2 (miR-132/212), Smad3, and Smad5 (miR-145), SOX family members SOX9 (miR-145) and SOX6 (miR-143, miR-132/212), and runt-related transcription factor 2 (RUNX-2) (miR-140-3p) ([50]).

Further study confirmed that miR-145 is a key mediator that antagonizes early chondrogenic differentiation in mice via attenuating SOX9 at the posttranscriptional level ([53]). In cells overexpressing miR-145, the expression of chondrogenic markers, including COL2A1, Agc1, cartilage oligomeric matrix protein (COMP), COL9A2, and COL11A1, was significantly decreased at the mRNA level ([53]). Consistent with this finding, Martinez-Sanchez et al reported that miR-145 directly regulated SOX9 in normal human articular chondrocytes through binding to a nonconserved specific site in its human 3′-UTR ([54]). Overexpression of miR-145 in articular chondrocytes reduced the levels of SOX9 and the cartilage matrix components COL2A1 and Agc1, and significantly increased the hypertrophic markers RUNX-2 and MMP-13 ([54]) (Figure 2).

Figure 2.

Regulatory networks of microRNAs (miRNAs) involved in osteoarthritis (OA) and chondrogenesis. MicroRNAs regulate components of different pathways in which they directly (solid lines) or indirectly (broken lines) repress target expression. MicroRNAs can thus regulate enzyme expression (e.g., ADAMTS-5 or matrix metalloproteinase [MMP-13]) directly or can affect signaling pathways (e.g., transforming growth factor β [TGFβ]) that are altered in OA. Factors involved in chondrocyte development or phenotype (e.g., SOX9 or histone deacetylase 4 [HDAC-4]) can regulate or be regulated by miRNAs. Inflammatory pathways and miRNAs (e.g., miR-155 or miR-146a) can affect all of these functions. Gray boxes indicate cytokines, brown boxes indicate membrane proteins, yellow boxes indicate transcription factors, and blue borders indicate other components. The asterisk indicates the passenger strand of an miRNA. BMP = bone morphogenetic protein; CHRDL1 = chordin-like protein 1; IL-1β = interleukin-1β; IL-1R = IL-1 receptor; TLR-2 = Toll-like receptor 2; MyD88 = myeloid differentiation factor 88; IRAK = IL-1 receptor–associated kinase; TRAF6 = tumor necrosis factor receptor–associated factor 6; TAB2 = TGFβ-activated kinase 1 binding protein 2; HIF-2α = hypoxia-inducible factor 2α; MEF2C = myocyte enhancer factor 2C; TNFα = tumor necrosis factor α; PPARA = peroxisome proliferator–activated receptor α; SIRT-1 = sirtuin 1; VEGF = vascular endothelial growth factor; COMP = cartilage oligomeric matrix protein; IGFBP-5 = insulin-like growth factor binding protein 5.

The same group also reported that miR-675, processed from H19, a noncoding RNA, was tightly regulated by SOX9 during chondrocyte differentiation. MicroRNA-675 could up-regulate the expression of COL2A1, albeit indirectly, indicating its potential importance in maintaining cartilage integrity and homeostasis. Forced overexpression of miR-675 rescued COL2A1 mRNA levels in either SOX9- or H19-depleted primary human articular chondrocytes ([55]). Although its target mRNAs remain unknown, these data suggest that miR-675 may modulate cartilage homeostasis by suppressing a COL2A1 transcriptional repressor ([55]) (Figure 2).

A search for expression signatures of nearly 380 miRNAs in C2C12 cells induced by BMP-2 found that 5 miRNAs, including miR-199a-3p and miR-221, were positively regulated, while miR-125a, miR-210, miR-125b, miR-21, miR-145, and miR-143 were repressed ([49]). In C3H10T1/2 cells, a well-established in vitro cell model of chondrogenesis, miR-199a-3p expression was reduced within 5 hours following BMP-2 treatment and was then increased at 24 hours and remained elevated, indicating that miR-199a-3p may function as a suppressor of early chondrogenic differentiation ([49]). Forced miR-199a-3p expression in C3H10T1/2 cells or in the prechondrogenic cell line ATDC5 suppresses multiple markers of early chondrogenesis, including COL2A1 and COMP, whereas the antagomir that blocks miR-199a-3p function has the opposite effect ([49]). Consistent with these observations, Smad1, a positive downstream mediator of BMP-2 signaling, was shown to be a direct miR-199a-3p target. Moreover, miR-199a-3p, through its inhibition of the Smad pathway, is able to inhibit the expression of downstream genes such as p204 ([49]) (Figure 2).

The expression pattern of miRNAs across the dedifferentiation of chondrocytes has also been explored ([7, 52]). MicroRNA-451, miR-140-3p, miR-210, miR-30d, and miR-563 were reported to be highly expressed in human primary articular chondrocytes at early passages compared with their dedifferentiated counterparts ([7]). Of these miRNAs, miR-140-3p had the highest expression. Conversely, 16 miRNAs were significantly up-regulated in dedifferentiated articular chondrocytes. Notably, miR-143 and miR-145 also had expression patterns similar to those previously reported ([7]). A second study also reported a group of 5 miRNAs, miR-451, miR140-3p, miR-210, miR-30d, and miR-563, to be up-regulated on differentiation, which may inhibit molecules participating in the dedifferentiation process, while a further 16 miRNAs were down-regulated and may act conversely ([52]).

MicroRNA profiling across ATDC5 cell chondrogenesis identified 7 clusters of miRNAs that may function cooperatively ([12]). Among these, 39 miRNAs were found to be potentially coregulated with miR-140, with expression increasing during chondrogenesis ([12]). Especially interesting was miR-455, which is located in an intron of the protein-coding gene COL27A1, a cartilage-expressed collagen. Consistent with the notion that miR-140 has a role in modulating TGFβ signaling, miR-455-3p was also found to directly target Smad2, ACVR2B, and chordin-like protein 1, again potentially attenuating the TGFβ pathway ([12]) (Figure 2).

MicroRNA-140 showed the most consistent expression pattern between studies and has been a focus of cartilage miRNA research to date. Miyaki et al compared gene expression in human articular chondrocytes and human MSCs, demonstrating that miR-140 had the largest difference in expression ([8]); this is consistent with the findings of later studies of humans, rats, and mice ([7, 50-52]). MicroRNA-140 was first shown to target histone deacetylase 4 (HDAC-4), a known corepressor of RUNX-2 and myocyte enhancer factor 2C, transcription factors that are essential for chondrocyte hypertrophy and bone development ([9]). MicroRNA-140 also targets CXCL12 ([56]) and Smad3 ([57]), both of which are implicated in chondrocyte differentiation. Interestingly, miR-140 was reported to suppress Dnpep, an aspartyl aminopeptidase, which has been suggested to antagonize BMP signaling downstream of Smad activation ([11]). Moreover, SOX9, a major transcription factor in maintaining cellular phenotype and preventing hypertrophy, particularly with L-SOX5 and SOX6 ([58]), has been shown to control the expression of miR-140 ([59]).

MicroRNA-194 may mediate chondrogenic differentiation via suppression of the transcription factor SOX5 ([60]) (Figure 2). The expression of miR-194 was significantly decreased in chondrogenic differentiation of adipose-derived stem cells, and chondrogenic differentiation of adipose-derived stem cells could be achieved through controlling miR-194 expression ([60]).

Using 3 rat models, Zhong et al demonstrated that miR-337 was directly implicated in chondrogenesis, acting as a repressor for TGFβ receptor type II expression at the protein level ([61]). Moreover, aggrecan was differentially expressed in both gain- and loss-of-function of miR-337 experiments, providing evidence that miR-337 could influence cartilage-specific gene expression in chondrocytes ([61]) (Figure 2).

Kim et al focused on the initiation of chondrogenesis in the chick, reporting that miR-221 and miR-34a, induced by JNK signaling, played pivotal roles ([62, 63]). Treatment of chick wing bud MSCs with a JNK inhibitor led to the suppression of cell migration and stimulation of apoptosis with a concurrent significant increase in the expression of miR-221 and miR-34a ([62, 63]). Notably, miR-221 may be involved in apoptosis, since treatment of MSCs with a miR-221 inhibitor increased cell proliferation, and this could be rescued by the JNK inhibitor ([63]). MicroRNA-221 is reported to directly target Mdm2, which encodes for an oncoprotein with E3 ubiquitin ligase activity ([63]). Inhibition of Mdm2 expression via miR-221 suppresses ubiquitination, leading to accumulation of Slug protein, whose expression is associated with an increase in apoptosis ([63]). Conversely, miR-34a affects MSC migration, not proliferation ([62]). EphA5, a receptor in Eph/ephrin signaling which mediates cell-to-cell interaction, has been proven to be a miR-34a target ([62]). Moreover, via regulating RhoA/Rac1 cross-talk, miR-34a negatively modulated reorganization of the actin cytoskeleton ([64]), an essential process for establishing chondrocyte-specific morphology.

MicroRNA-488 expression is up-regulated at the precondensation stage and down-regulated at postcondensation in chick limb chondrogenesis, suggesting a key role in this process ([65]). Interestingly, miR-488 could regulate cell-to-ECM interaction via modulation of focal adhesion activity by indirectly targeting MMP-2 ([65]). More recently, this group reported that miR-142-3p was an important modulator in position-dependent chondrogenesis and able to regulate ADAM-9 ([66]) (Figure 2).

MicroRNAs in osteoarthritis

The effects of miRNA deregulation on OA are evident through studies comparing the expression of miRNAs between OA tissue specimens and their normal articular counterparts ([67-69]). Iliopoulos et al measured the expression of 365 miRNAs and identified 16 miRNAs, with 9 miRNAs significantly up-regulated and 7 miRNAs down-regulated in OA cartilage compared with normal controls. Some of these were postulated to be involved in obesity and inflammation ([67]). Functional experiments implicated miR-9 in the regulation of MMP-13 expression, as well as miR-9, miR-98, and miR-146 in the control of tumor necrosis factor α (TNFα) expression, suggesting that these miRNAs may play a protective role in OA. Moreover, miR-22, whose expression correlated with body mass index, directly targets peroxisome proliferator–activated receptor α (PPARα) and BMP-7 at the mRNA and protein levels, respectively. Enforced miR-22 overexpression or small interfering RNA–mediated suppression of either PPARα or BMP-7 resulted in increases in interleukin-1β (IL-1β) and MMP-13 protein levels, again suggesting that miRNA deregulation can affect OA ([67]) (Figure 2).

Jones et al investigated the expression of 157 human miRNAs and identified 17 miRNAs whose expression varied by 4-fold or more when comparing normal versus late-stage OA cartilage ([68]). Consistent with the findings of Iliopoulos et al ([67]), the altered expression of miR-9, miR-98, and miR-146 in OA cartilage were highlighted. The overexpression of these miRNAs also reduced IL-1β–induced TNFα production, while inhibition or overexpression of miR-9 modulated MMP-13 secretion ([68]) (Figure 2).

More recently, profiling of 723 miRNAs identified 7 miRNAs that were differentially expressed in OA and normal human chondrocytes, with miR-483-5p up-regulated in OA chondrocytes while other miRs, including miR-149-3p, miR-582-3p, miR-1227, miR-634, miR-576-5p, and miR-641 were up-regulated in normal chondrocytes. These miRNAs were predicted to function in articular cartilage via TGFβ, Wnt, Erb, and mammalian target of rapamycin signaling ([69]).

The miR-140 gene is evolutionarily conserved among vertebrates. MicroRNA-140 expression in the cartilage of patients with OA was significantly lower than that in normal cartilage ([8, 70]), and decreased miR-140 expression was also reported in OA chondrocytes ([70]). This may depend on the stage of disease since Swingler et al reported an increase in miR-140 in patients with hip OA as compared to controls with hip fracture, but with a concomitant decrease in ADAMTS-5, a key target gene ([12]).

Deletion of miR-140 predisposed mice to the development of age-related OA-like changes ([10, 11]) and increased cartilage destruction in surgically induced OA. Conversely, in an antigen-induced arthritis model, transgenic overexpression of miR-140 in chondrocytes protected against cartilage damage ([10]). The ADAMTS-5 gene was shown to be a direct target of miR-140, while reciprocal regulation of these two observed in the in vivo models described above suggests a functional relationship ([10]). Nakamura et al identified the aspartyl aminopeptidase Dnpep as a key target for miR-140 mediating skeletal defects in miR-140–null mice ([11]). Using functional interference, Tardif et al confirmed that insulin-like growth factor binding protein 5 (IGFBP-5), whose expression in human chondrocytes was significantly higher in OA, is a direct target of miR-140 ([70]). More recently, miR-140 was shown to directly mediate MMP-13 expression in vitro ([71]) (Figure 2).

The human genome contains 2 miR-27 genes (miR-27a and miR-27b), and their major products differ by only 1 nucleotide in the 3′ region. MicroRNA-27a expression was shown to be decreased in OA compared to normal chondrocytes ([70]). Down-regulation of miR-27a was proposed to be connected with adipose tissue dysregulation in obesity, a strong risk factor for OA ([72]). Tardif et al suggested that miR-27a may indirectly regulate the levels of both MMP-13 and IGFBP-5 by targeting upstream positive effectors of both genes ([70]). Conversely, the expression of miR-27b was found to be significantly lower in OA cartilage samples compared with their normal counterparts and was inversely correlated with the expression of MMP-13, a direct target, in OA cartilage ([73]) (Figure 2).

MicroRNA-146a was strongly expressed in chondrocytes residing in the superficial layer of cartilage and in low-grade OA cartilage ([74, 75]). Its expression level gradually decreased with progressive tissue degeneration. Interestingly, when miR-146a was highly expressed, the expression of MMP-13 was low, suggesting that miR-146a has target genes that play a role in OA cartilage pathogenesis ([74]). MicroRNA-146a has recently been implicated in the control of knee joint homeostasis and OA-associated algesia by balancing the inflammatory response in cartilage and synovium with pain-related factors in glial cells ([76]) (Figure 2).

Mechanotransduction

Articular cartilage has the unique capacity to resist significant mechanical loading during the lifetime of the organism. The surface, middle, and deep zones within articular cartilage are distinct domains, with differential gene expression and attendant functional roles ([77]).

Mechanoresponsive miRNAs have been identified in chondrocytes, and evidence that specific miRNAs may affect mechanotransduction has also been reported. MicroRNA-365 was the first mechanically responsive miRNA identified in chondrocytes, regulating differentiation through inhibiting HDAC-4 ([78]). MicroRNA-221 and miR-222 were postulated as potential regulators of this pathway, since their expression patterns in bovine articular cartilage are higher in the weight-bearing anterior medial condyle compared with the posterior non–weight-bearing medial condyle ([79]). Li et al reported that miR-146a was induced by joint instability resulting from medial collateral ligament transection and medial meniscal tear in the knee joints of an OA mouse model, suggesting that miR-146a might be a regulatory factor of the mechanotransduction process in articular cartilage ([75]).

Inflammation

It is now accepted that the pathogenesis of OA has an inflammatory component, with soluble mediators coming from local (e.g., synovitis) or systemic (e.g., metabolic syndrome) sources ([80]). Altered miRNA expression has been clearly described in the overtly inflammatory rheumatoid arthritis (RA) seen in cells and tissues of the joint as well as in the peripheral circulation ([81]) with function assigned to some miRNAs, e.g., miR-155 ([82]).

Endogenous damage-associated molecular patterns revealed by degradation of molecules such as hyaluronan, fibronectin, collagen, or the proteoglycan aggrecan have been identified in OA in response to joint injury ([83, 84]). These molecules are recognized by Toll-like receptors (TLRs) in a manner similar to that of exogenous pathogen-associated molecular patterns ([85]). TLR pathways are capable of activating NF-κB, triggering the expression of a number of genes in establishing the inflammatory response in OA ([86]). Similarly, the proinflammatory cytokine IL-1 may play a prominent role in the catabolism of articular cartilage, in part via activation of NF-κB ([87]).

Some miRNAs have emerged as important controllers of these pathways. MicroRNA-155 exerts negative effects on the TLR/IL-1 inflammatory pathway through targeting TGFβ-activated kinase 1 binding protein 2, inhibiting the activation of TGFβ-activated kinase 1 and hence NF-κB and MAPK ([88]). MicroRNA-146a dampens the activation of this pathway by targeting IL-1 receptor–associated kinase 1 and TNF receptor–associated factor 6 ([89]). Interestingly, the induction of transcription of miR-146a by TNFα and IL-1β is dependent on NF-κB ([74, 75]), thus suggesting a feedback loop involving miR-146a and TLR signaling. In particular, miR-146a expression increases in OA cartilage, with peripheral expression in mononuclear cells according to disease progression and potentially functionally linked to OA pathology, including pain ([74, 76, 90]).

Other miRNAs reported to be induced by IL-1β are miR-34a ([91]), miR-194 ([60]), and miR-27b ([73]). Expression of miR-34a was significantly induced by IL-1β, while antagonism of miR-34a prevented IL-1β–induced chondrocyte apoptosis ([91]), as well as IL-1β–induced down-regulation of type II collagen, in rat chondrocytes ([91]). MicroRNA-140 was reported to be suppressed by IL-1β signaling ([8]), though Liang et al reported elevated expression of miR-140 and MMP-13 in IL-1β–stimulated C28/I2 cells, and expression of miR-140 was shown to be NF-κB dependent ([71]).

Diagnosis

Measurement of miRNAs in serum may become a powerful tool in the development of diagnostic biomarkers. MicroRNAs are relatively stable in freezing, thawing, or extreme pH conditions due to lipid or lipoprotein complexes. Extracellular miRNAs are detectable in almost all body fluids and excretions ([92, 93]). Interestingly, miRNAs in plasma have the capacity to interact with intact cells to modify recipient cell gene expression and protein production ([94]). This opens up the possibility of genetic exchange between cells and exogenous regulation of gene expression. MicroRNA-21 was the first serum miRNA biomarker to be discovered. Patients with diffuse large B cell lymphoma had high serum levels of miR-21, which were associated with increased relapse-free survival ([95]). Subsequently, the usefulness of serum/plasma miRNAs for determining diagnosis and prognosis has been reported for solid cancers and leukemia ([93]). In OA, Murata et al found a number of miRNAs in plasma, some of which were significantly different between OA patients and healthy controls as well as between RA patients and OA patients ([96]).

Besides the measurement of miRNAs in plasma, their measurement in peripheral blood mononuclear cells (PBMCs) could also be useful in developing a biomarker for OA. Circulating PBMCs may accumulate in the synovium of OA patients, producing proinflammatory cytokines and proteinases linked to disease progression. The high expression of miR-146a, miR-155, miR-181a, and miR-223 in PBMCs from OA patients versus normal controls may be related to the pathogenesis of OA ([90]). Interestingly, miR-146a is highly expressed in the cartilage of patients with early-stage OA ([74]), with expression gradually decreasing with OA progression and the possibility for diagnosis of early OA if specificity can be demonstrated.

Therapy

Currently, there are no disease-modifying therapies available for OA patients ([97]). There are over 100 clinical trials worldwide based on miRNA manipulation to treat a range of conditions, including cancers and cardiovascular disease; however, none of these are for arthritis (http://clinicaltrials.gov).

The targeting of miRNAs represents a novel therapeutic opportunity for OA treatment in which miRNA deficiencies could be corrected by either antagonizing or restoring miRNA function. In particular, it has been proven that systematic administration of anti-miRNAs modified with locked nucleic acids (LNA) could function without toxicity in non-human primates ([98]). Although this type of therapy has not been applied in OA, there is evidence for therapeutic potential, e.g., the silencing of miR-34a by LNA-modified antisense oligonucleotides could effectively reduce chondrocyte apoptosis induced by IL-1β ([99]).

Another approach is to combine miRNA technology with stem cell engineering, e.g., to enhance autologous MSCs for transplantation into articular cartilage defects. Deciphering the role of miRNA regulation in the chondrogenesis of MSCs may pave the way for the development of new treatments. OA patients often present with cartilage that already exhibits a damaged matrix and repair here is particularly difficult ([1]).

Conclusions and perspectives

There is growing evidence that indicates that miRNAs convey a novel and efficient means for the regulation of gene expression. Understanding their expression and dynamic regulation may be the key to enhancing chondrogenic differentiation or maintaining phenotype. This will affect tissue and cell engineering approaches, as well as direct treatments in OA aimed at preventing tissue destruction and driving repair.

It is clear that a single miRNA (as demonstrated for miR-140) can have profound effects on the development of OA in vivo. The role(s) of individual miRNAs may best be probed using knockout and transgenic mice in OA models, but other means of delivering miRNAs functionally to the joint or antagonize their function will be key to dissecting their function and to their eventual therapeutic use.

MicroRNA expression may also mediate risk factors for OA, such as aging. A number of studies have measured either circulating miRNAs or those expressed in peripheral blood cells across different ages in humans. Analyses demonstrate differential miRNA expression with age with specific miRNAs showing either an increase or decrease, including some of those implicated in OA (as described above) (100–102). Validation has shown, for example, that miR-21 levels were correlated with C-reactive protein, a marker of inflammation, and with TGFβ signaling ([102]). MicroRNA-221 expression, decreased with age, could mediate an increase in phosphatidylinositol 3-kinase signaling, a pathway shown to be implicated in OA ([101, 103]). These miRNAs may be diagnostic of OA, a disease where age is a major risk factor, where it has been difficult to find specific markers of either incidence or progression. Measurement of miRNAs represents another strategy that can be brought to bear on this problem.

Unraveling the role that miRNAs play in joint physiology and pathology will shed light upon the diagnosis, prevention, and treatment of OA. Numerous questions remain, however, and understanding the network of mRNAs expressed, the kinetics of miRNA expression, and the relationship with their cognate genes will not be trivial, but will be rewarding.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published.

Ancillary