MicroRNAs in Osteoclastogenesis and Function: Potential Therapeutic Targets for Osteoporosis

Abnormal osteoclast formation and resorption play a fundamental role in osteoporosis pathogenesis. Over the past two decades, much progress has been made to target osteoclasts. The existing therapeutic drugs include bisphosphonates, hormone replacement therapy, selective estrogen receptor modulators, calcitonin and receptor activator of nuclear factor NF-κB ligand (RANKL) inhibitor (denosumab), etc. Among them, bisphosphonates are most widely used due to their low price and high efficiency in reducing the risk of fracture. However, bisphosphonates still have their limitations, such as the gastrointestinal side-effects, osteonecrosis of the jaw, and atypical subtrochanteric fracture. Based on the current situation, research for new drugs to regulate bone resorption remains relevant. MicroRNAs (miRNAs) are a new group of small, noncoding RNAs of 19–25 nucleotides, which negatively regulate gene expression after transcription. Recent studies discovered miRNAs play a considerable function in bone remodeling by regulating osteoblast and osteoclast differentiation and function. An increasing number of miRNAs have been identified to participate in osteoclast formation, differentiation, apoptosis, and resorption. miRNAs show great promise to serve as biomarkers and potential therapeutic targets for osteoporosis. In this review, we will summarize our current understanding of how miRNAs regulate osteoclastogenesis and function. We will further discuss the approach to develop drugs for osteoporosis based on these miRNA networks.


Introduction
Osteoporosis, the growing metabolic skeletal disorder worldwide, has been called a "silent killer" due to its high rate of incidence and disability. Currently, drug selection for osteoporosis is limited. Bisphosphonates, the first-line drugs for osteoporosis due to their low price and high efficiency in reducing the risk of fracture, have gastrointestinal side-effects and, more importantly, long-term application of bisphosphonates inhibit the osteoblast and osteoclast functions simultaneously [1]. Bone formation and bone resorption are elaborate and well-coupled processes. The inhibition of bone resorption will lead to inhibition of bone formation and ultimately affect the efficacy of anti-bone-resorption drugs. Previous studies suggested that, except for coupling factors, the maintenance of osteoclast numbers is very important for keeping normal osteoblast functions. For example, bisphosphonates promote osteoclast apoptosis and decrease osteoclast numbers, which lead to inhibition of osteoblast functions. On the contrary, the application of inhibitors of cathepsin K and chloride channel 7, which are all involved in bone resorption, have no effects on osteoblast functions, since these inhibitors only reduce bone resorption capacity but not osteoclast numbers. Abnormal osteoclast formation and resorption play a fundamental role in osteoporosis pathogenesis. Therefore, understanding osteoclast proliferation, differentiation, apoptosis, bone resorption, and the coupling mechanism between osteoclasts and osteoblasts, have a key role in the development of new drugs for osteoporosis.
MicroRNAs (miRNAs) are a class of small, noncoding RNAs of 19-25 nucleotides, which exist widely in eukaryotes and are highly conserved during biological evolution. After binding to 3'-untranslated regions (3'-UTR) within a target mRNA, miRNAs play a negative role in gene expression by regulating transcript localization, polyadenylation, and translation [2][3][4]. In 1993, lin-4RNAs were first discovered in Caenorhabditis elegans by Lee etc. [5]. In 2004, Chen et al. identified three miRNAs, which were not only specifically expressed in hematopoietic cells but the expression was dynamically modulated during early hematopoiesis and lineage commitment as well [6]. Since then, an increasing number of miRNAs have been identified to participate in osteoclast formation, differentiation, apoptosis, and resorption.
We have searched literature from PubMed and referred three other reviews [7][8][9]. This review aims to summarize our current understanding of how miRNAs regulate osteoclastogenesis and briefly refer to their potential clinical implications, such as biomarkers and the development of new drugs for osteoporosis based on these miRNA networks.

Bone Remodeling and Osteoclasts
Bone remodeling is a dynamic process throughout the whole lifetime of an individual, by which the skeleton maintains its structural integrity and exerts its metabolic functions as a repository of calcium and phosphorus [10]. Bone remodeling is regulated by the subtle equilibrium between osteoblastic bone formation and osteoclastic bone resorption. Firstly, there is an "activation" phase and a "resorption" phase. Cytokines are released at the site of remodeling to recruit osteoclasts to the bone surface. These osteoclasts form a ruffled border allowing them to adhere to the bone surface tightly. Between the osteoclast and the underlying bone, there exists a tiny isolated microenvironment into which the osteoclast's proton pump releases ions that create an acidic environment, making the mineralized component of the bone matrix dissolve. The organic matrix is exposed and degraded by cathepsin K [11]. Subsequently, the "reversal" phase begins. Mononuclear cells prepare the bone surface for osteoblasts and provide signals to recruit them. Along with proliferating, early osteoblasts secrete an extracellular matrix, which contains type I collagen abundantly. This matrix matures and is mineralized, and osteoblasts continue to differentiate. Finally, the bone surface is repaired. Those mature osteoblasts either undergo apoptosis, or eventually differentiate into osteocytes or bone surface lining cells [12].
Originating from mononuclear hematopoietic myeloid lineage cells, osteoclast precursors (OCPs) are formed in the bone marrow and subsequently attracted to the bloodstream by chemokines. Attracted by a variety of factors released from bone remodeling units (BRUs), OCPs are attracted back into bones and then they differentiate into osteoclasts [13]. During normal physiological conditions, osteoclastogenesis is regulated by osteoblasts and stromal cells, both of which provide two essential factors, macrophage colony-stimulating factor (M-CSF) and receptor activator of nuclear factor NF-κB ligand (RANKL). M-CSF plays an essential role in the survival, proliferation and the expression of RANK in early OCPs (monocyte/macrophage lineage). Therefore, the primary role of M-CSF is to provide survival signals during osteoclastogenesis [14]. In contrast, RANKL provides osteoclast differentiation signals and activates multiple signal transduction pathways, which turn on transcription factors NF-κB, c-Fos, transcription factor nuclear factor of activated T cells (NFATc1), and microphthalmia-induced transcription factor (MITF) [15].

Regulation of OCPs Formation via M-CSF Signaling
M-CSF is pivotal for the survival and proliferation of OCPs. M-CSF, through its receptor c-Fms, transmits signals to the cell and then activates extracellular signal-regulated kinase (ERK) through Grb2 and phosphoinositide 3-kinase (PI-3K)/Akt [16,17]. The differentiation process of hematopoietic stem cells to OCPs is induced by transcription factors, such as purine-rich binding protein 1 (gene symbol: SPI1, PU.1) and Mitf [18]. PU.1 is a hematopoietic-specific member of the ETS family [19]. Deletion of PU.1 in mice results in a complete lack of OCPs leading to osteopetrosis [20]. There are PU.1 binding sites within promoters of many genes involved in osteoclast formation and function [21]. During the process of hematopoietic stem cells differentiating into the monocyte/macrophage lineage, PU.1 stimulates the expression of CSF1R which is the receptor of CSF1 (i.e., M-CSF) [22]. By upregulating the transcription factor c-FOS, CSF1R induces expression of receptor activator of NFκB (RANK; TNFRSF11A). In cooperation with other transcription factors, PU.1 modulates the RANK gene transcription [23]. Activator protein 1 (AP-1) plays a critical role in osteoclastogenesis. The AP-1 transcription factor complexes comprise of the Fra, Fos, Jun, and activating transcription factor (ATF) families.
MITF is another critical transcription factor participating in the late stages of osteoclastogenesis. Through a conserved mitogen-activated protein kinase (MAPK) consensus site, M-CSF induces phosphorylation of MITF [24]. Then MITF induces the expression of BCL-2 and promotes macrophage survival. Both Mitf mi/mi and Bcl2 −/− mice suffer severe osteopetrosis [25]. Moreover, through binding to the sites within the RANK promoter, MITF and PU.1 increase RANK promoter activity three-fold and two-fold, respectively, and six-fold synergistically [26]. Conversely, Mitf-E levels are significantly upregulated by RANKL [27].
Both PU.1 and MITF not only play an important role in the survival of OCPs, but also participate in osteoclast-specific gene induction at the terminal stage of differentiation [21,28].

RANKL-RANK Signaling
RANKL commits the OCPs to osteoclast fate. The activation of RANKL-RANK signaling leads to the expression of genes involving the fusion of mononuclear osteoclast precursors, like dendritic cell-specific transmembrane protein (DC-STAMP), as well as of genes regulating resorption capacity of multinucleated osteoclasts, including cathepsin K, chloride channel 7, matrix metalloprotein 9, and calcitonin receptor.

Regulation of OCPs Formation via M-CSF Signaling
M-CSF is pivotal for the survival and proliferation of OCPs. M-CSF, through its receptor c-Fms, transmits signals to the cell and then activates extracellular signal-regulated kinase (ERK) through Grb2 and phosphoinositide 3-kinase (PI-3K)/Akt [16,17]. The differentiation process of hematopoietic stem cells to OCPs is induced by transcription factors, such as purine-rich binding protein 1 (gene symbol: SPI1, PU.1) and Mitf [18]. PU.1 is a hematopoietic-specific member of the ETS family [19]. Deletion of PU.1 in mice results in a complete lack of OCPs leading to osteopetrosis [20]. There are PU.1 binding sites within promoters of many genes involved in osteoclast formation and function [21]. During the process of hematopoietic stem cells differentiating into the monocyte/macrophage lineage, PU.1 stimulates the expression of CSF1R which is the receptor of CSF1 (i.e., M-CSF) [22]. By upregulating the transcription factor c-FOS, CSF1R induces expression of receptor activator of NFκB (RANK; TNFRSF11A). In cooperation with other transcription factors, PU.1 modulates the RANK gene transcription [23]. Activator protein 1 (AP-1) plays a critical role in osteoclastogenesis. The AP-1 transcription factor complexes comprise of the Fra, Fos, Jun, and activating transcription factor (ATF) families.
MITF is another critical transcription factor participating in the late stages of osteoclastogenesis. Through a conserved mitogen-activated protein kinase (MAPK) consensus site, M-CSF induces phosphorylation of MITF [24]. Then MITF induces the expression of BCL-2 and promotes macrophage survival. Both Mitf mi/mi and Bcl2´{´mice suffer severe osteopetrosis [25]. Moreover, through binding to the sites within the RANK promoter, MITF and PU.1 increase RANK promoter activity three-fold and two-fold, respectively, and six-fold synergistically [26]. Conversely, Mitf-E levels are significantly upregulated by RANKL [27].
Both PU.1 and MITF not only play an important role in the survival of OCPs, but also participate in osteoclast-specific gene induction at the terminal stage of differentiation [21,28].

RANKL-RANK Signaling
RANKL commits the OCPs to osteoclast fate. The activation of RANKL-RANK signaling leads to the expression of genes involving the fusion of mononuclear osteoclast precursors, like dendritic cell-specific transmembrane protein (DC-STAMP), as well as of genes regulating resorption capacity of multinucleated osteoclasts, including cathepsin K, chloride channel 7, matrix metalloprotein 9, and calcitonin receptor.
RANKL-RANK binding recruits TRAF-6 to activate PI-3K, NF-κB family of transcription factors and all three MAPK pathways, including ERK, JNK (Janus N-terminal kinase), and p38. NF-κB is required for the expression of a variety of cytokines, including IL-6, IL-1, TNF-α, GM-CSF, RANKL, and other growth factors. Protein kinase p38 is activated via phosphorylation of MAPK kinase (MKK) 6. The activation of p38 results in the downstream activation of MITF [29]. Hence, MITF exists downstream of the M-CSF and RANKL signaling pathways. Treatment with the p38 inhibitors increases phosphorylation of ERK, showing a balance between ERK and p38 phosphorylation.

Immunoreceptor Tyrosine-Based Activation Motif (ITAM)-Dependent Costimulatory Signals
M-CSF and RANKL are not sufficient to activate the signals required for osteoclastogenesis. Immunoreceptor tyrosine-based activation motif (ITAM)-dependent costimulatory signals, activated by multiple immunoreceptors, are essential for osteoclastogenesis. Both Fc receptor common γ subunit (FcRγ) and DNAX-activating protein 12 (DAP12) are ITAM-harboring adapters. In osteoclast precursor cells, FcR and DAP12 associated with multiple immunoreceptors activate calcium signaling through phospholipase C [34]. These receptors include OSCAR, triggering receptor expressed in myeloid cells-2 (TREM-2), signal-regulatory protein β1 (SIRPβ1), and paired Ig-like receptor-A (PIR-A). These receptor-mediated signals cannot substitute RANKL but act with RANKL cooperatively. Therefore, ITAM-mediated signals can be identified as co-stimulatory signals for RANK.

miRNAs in Osteoclasts
Osteoclast differentiation is regulated by transcriptional, post-transcriptional, and post-translational mechanisms. miRNAs are fundamental post-transcriptional regulators of gene expression. miRNAs play a key role in the normal bone development. Heterozygous microdeletions in the MIR17HG locus, encoding microRNA 17-92 cluster, lead to autosomal dominant Feingold syndrome in humans, characterized by short stature, microcephaly, and abnormal development of fingers and toes [35]. Further study on animal models carrying targeted deletions of individual components of miR-17~92 revealed that miR-17 seed family is critical in patterning of the axial skeleton [36]. MicroRNA-related single nucleotide polymorphisms (SNPs) also have a potential impact on the skeletal phenotype [37]. Furthermore, emerging evidence suggests that miRNAs are involved in the multiple biological and pathological processes in osteoclast proliferation, differentiation, apoptosis, cytoskeleton formation, and bone resorption. During the early, middle, and late stages of murine osteoclastogenesis, miRNA microarray analysis showed 49 miRNAs were upregulated and 44 were downregulated [38]. In the following section, we will discuss relevant miRNAs in osteoclasts and their potential targets and signaling pathways (Table 1 and Figure 2).  ↑ means miRNAs are upregulated; ↓ means miRNAs are down-regulate.

miR-124-3p
miR-124-3p has been suggested to have a putative tumor-suppressive role. Previous study indicated it also may be an intrinsic negative regulator of osteoclast differentiation by suppressing NFATc1 expression. NFATc1 is a key regulator of osteoclastogenesis. TargetScan, a web-based bioinformatics tool, predicted two conserved binding sequences of miR-124-3p in the 3 1 UTR region of mouse NFATc1 gene. During the osteoclastic differentiation from BMMs induced by RANKL, the expression of miR-124-3p rapidly decreased. Pre-miR-124 significantly inhibited the RANKL-induced NFATc1 induction and osteoclast differentiation. On the contrary, inhibition of miR-124-3p potently promoted NFATc1 expression and osteoclastogenesis. miR-124-3p also reduces the expression of RhoA and Rac1, through which it might inhibit the migration of osteoclast precursors [67].

miR-218-5p
miR-218-5p has been demonstrated to stimulate bone formation. miR-218-5p has also been suggested to be a negative regulator of osteoclastogenesis. The expression of miR-218-5p was decreased in CD14+ PBMCs from post-menopausal osteoporosis patients compared with healthy control. During osteoclastogenesis from BMMs and RAW264.7 induced by RANKL, miR-218-5p expression was significantly downregulated. Upregulation of miR-218-5p obviously inhibited the formation of multinuclear osteoclasts, the migration of osteoclast precursors, actin ring formation, and bone resorption along with the decreased TRAP and Cathepsin K expression. Mechanistically, miR-218-5p suppresses osteoclastogenesis by targeting the p38MAPK-c-Fos-NFATc1 pathway [72].

Potential Clinical Implications of miRNAs and miRNAs-Based Therapeutic Strategy for Osteoporosis
The existing therapeutic drugs include bisphosphonates, hormone replacement therapy, selective estrogen receptor modulators, calcitonin, and RANK ligand inhibitor (denosumab), etc. Bisphosphonates are nowadays the first-line anti-resorptive medication. There exist several potential adverse clinical events concomitant with the medication of bisphosphonates, including osteonecrosis of the jaw [74], atrial fibrillation [75], acute inflammatory response [75], and oversuppression of bone turnover [76]. Bone formation and bone resorption are well-coupled processes. The inhibition of bone resorption will result in inhibition of bone formation [1]. miRNAs are involved in the osteoclast proliferation, differentiation, cell-fusion, apoptosis, cytoskeleton formation, and bone resorption, which were summarized in Table 1 and Figure 2. As stated above, most miRNAs are involved in promoting or inhibiting osteoclast formation and maturation. Only a few miRNAs are capable of affecting osteoclast function.

Potential Use of miRNAs as Biomarkers
The role of miRNAs as biomarkers for bone diseases has drawn much attention recently.

Biomarkers for Osteoclasts Activity
As a positive regulator in osteoclast formation, the expression of miR-29 (a/b/c) family was upregulated during osteoclastogenesis in vitro cell culture, which was consistent with the expression of osteoclast markers TRAP and cathepsin K [43]. In contrast, upregulation of miR-218-5p, a negative ocstoclastogenesis regulator, was consistent with the decreased TRAP and Cathepsin K expression [72]. The expression of plasma TRAP5b has been used to estimate the activity of osteoclasts; thus, miR-29 and miR-218-5p need further study to validate their potential as novel biomarkers for osteoclasts activity.
Taken together, the role of miRNAs as osteoporotic biomarkers mostly attributes to a causal relationship between the miRNA and the osteoblast differentiation [80]. Interestingly, miRNAs, like miR-148a-3p [47] and miR-21-5p [39][40][41][42], were found to be involved both in osteogenesis and the regulation of osteoclastogenesis. Further cross validation of cell free blood-based miRNAs with bone miRNAs is strongly relevant.

miRNA Delivery System
Through a novel miRNA delivery system based on bacteriophage MS2 virus-like particles (MS2 VLPs), Yao et al. successfully transported miR-146a into human PBMCs, and subsequently demonstrated the inhibitory function of miR-146a-5p in osteoclastogenesis [69]. miR-148a-3p exerts potent inhibitory effects on osteoclast differentiation. Cheng et al. validated the bone mass in mice would increase via a single tail vein injection of a specific antagomiR-148a [47]. Liu et al. established efficient delivery systems to facilitate antagomir-148a-3p to bone resorption surfaces to reduce bone resorption with minimal off-target effects [48]. miR-503-5p acts as a negative regulator of osteoclastogenesis through inhibiting RANK. Ovariectomy (OVX) mice exhibited increased RANK protein expression, promoted bone resorption, and decreased bone mass after using a specific antagomir to silence miR-503-5p expression, whereas agomir-503 exhibit opposite effects [73].
In developing miRNAs-based therapy, it may be helpful to maintain the normal function of osteoblasts if only inhibiting the bone resorption function of osteoclast without affecting the number of osteoclasts. Therefore, it seems that miR-31-5p will be an ideal target, since it just promotes osteoclast function through inhibiting RhoA and RhoA plays a key role in actin ring formation as a small GTPase [44].

Conclusions
Cumulating evidence suggested that miRNAs are involved in multiple physiological and pathological processes of osteoclast differentiation and function. In general, the expression levels of miRNAs that exert inhibitory effects on osteoclastogenesis tend to decrease during osteoclast formation, and vice versa. MiRNAs-based therapy has been considered as a promising strategy for the treatment of osteoporosis. Author Contributions: Xiao Ji and Xijie Yu designed this review; Xiao Ji and Xiang Chen wrote the manuscript; Xijie Yu revised this manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.