Mechanosensitive miRNAs and Bone Formation

Mechanical stimuli are required for the maintenance of skeletal integrity and bone mass. An increasing amount of evidence indicates that multiple regulators (e.g., hormone, cytoskeleton proteins and signaling pathways) are involved in the mechanical stimuli modulating the activities of osteogenic cells and the process of bone formation. Significantly, recent studies have showed that several microRNAs (miRNAs) were sensitive to various mechanical stimuli and played a crucial role in osteogenic differentiation and bone formation. However, the functional roles and further mechanisms of mechanosensitive miRNAs in bone formation are not yet completely understood. This review highlights the roles of mechanosensitive miRNAs in osteogenic differentiation and bone formation and underlines their potential therapeutic application for bone loss induced by the altering of mechanical stimuli.


Mechanosensitive miRNAs and Osteogenic Differentiation
Mechanosensitive miRNAs are identified to be involved in osteogenic differentiation due to the expressions of these miRNAs changing with mechanical stimuli, altering osteogenic differentiation. Microarray and bioinformatics analysis are effective and prompt methods to search for mechanosensitive miRNAs altering with different mechanical stimuli in osteogenic cells (Table 1). Two microarray data of MC3T3-E1 cells subjected to a four-point bending stretch suggested that miR-3077-5p, -3090-5p, -3103-5p, -191*, and -3070a were significantly up-regulated, in contrast, miR-466i-3p, -466h-3p, -218 and -33 were down-regulated; accompanied by alkaline phosphatase (ALP) activity, mRNA levels of Alp, Ocn (osteocalcin), Col I (Collage I), and the protein levels of BMP-2 and BMP-4 increased in the mechanical stretch group [41,42]. Furthermore, analysis of putative target genes indicated that this mechanosensitive miRNAs might be involved in osteoblast differentiation [41,42]. Additionally, four-point bending stretch could have significantly promoted cementoblastogenesis, mineralized nodule formation, and miR-146b-5p expression in OCCM-30 cells. Smad4 (SMAD family member 4), a common mediator of both BMP (Bone morphogenetic protein) and TGF-β (Transforming growth factor-β) signaling, was a target gene of mechanosensitive miR-146b-5p; this was confirmed by a dual-luciferase reporter assay [43].

Mechanosensitive miRNAs Promoted Osteogenic Differentiation
Several mechanosensitive miRNAs were reported to promote osteogenic differentiation (Table 2) [39,40]. To deeply explore the function of mechanosensitive miR-21 on the osteogenic differentiation of stretch-induced PDLSCs, Wei's group tested the correlation analysis between stretch times and miR-21 [46,51]. The expression of miR-21 and osteogenic marker genes (Runx2 (runt related transcription factor 2) and Ocn) gradually rose with stretch times increasing by degrees. Further, miR-21's target gene Acvr2b (activin receptor type IIB) partially inhibited the stretch-induced ALP activity and Runx2 and Ocn gene expression, revealing that mechanosensitive miR-21 facilitated stretch-induced PDLSC osteogenic differentiation through suppressing Acvr2b. The miR-21 level was down-regulated approximately 70% under FSS in MC3T3-E1 cells [48]. Additionally, miR-21 was reported to promote the level of osteogenic differentiation and increase matrix mineralization by directly repressing Smad7 in MC3T3-E1 cells [52], while miR-21 expression gradually decreased during PDLC mineralization [53]. The results of these studies indicated that miR-21 might have a diametric role in different types of bone cell osteogenic differentiation and under various mechanical stimuli. This phenomenon also appeared in another miRNA (miR-33), which was reported to decrease in the mechanical stretch group in MC3T3-E1 cells [41]. However, one subtype of miR-33, miR-33-5p expression, was significantly up-regulated under increasing FSS, and the knockdown of miR-33-5p partially prevented FSS-induced osteoblast differentiation [40]. Not only was miR-33-5p sensitive to mechanical loading, but it was also responsive to mechanical unloading in MC3T3-E1 cells. In view of miR-33-5p positively regulating osteoblast differentiation, Wang investigated the roles of miR-33-5p in osteoblast activity under mechanical unloading. The overexpression of miR-33-5p partially attenuated the inhibition of MC3T3-E1 osteogenic differentiation induced by mechanical unloading and Hmga2 (high mobility group AT-hook 2), as miR-33-5p's target, was confirmed to negatively regulate osteoblast differentiation [40].
FSS was also used to study the effects of mechanical stimuli on PDLCs. It is reported that increasing FSS in fixed increments regulated not only the PDLC proliferation and differentiation, but also miR-132 expression. Further, phosphorylated levels of the PI3K (phosphatidylinositol-3 kinase), AKT (serine/threonine kinase), mTOR (mammalian target of rapamycin), and p70S6K (ribosomal protein S6 kinase) proteins were significantly greater with FSS-treatment, however they were nonetheless blocked by the miR-132 inhibitor. The subsequent results suggested that up-regulation of miR-132 induced by FSS could activate the mTOR signaling pathway and PDLC osteogenic differentiation [39]. Bio-stretch was another type of stretch stress device, which exerted a uniaxial stretch with square wave patterns and was used to simulate stretch in chondrocytes. miR-365 was identified as the first mechanosensitive miRNA to stimulate chondrocyte differentiation via directly targeting Hadc4 (hystone deacetylase 4) [38].

Mechanosensitive miRNAs Inhibited Osteogenic Differentiation
Despite several mechanosensitive miRNAs promoting osteogenic differentiation (Table 2), there are more mechanosensitive miRNAs reported to serve as inhibitors in osteogenic differentiation (Table 3). Li et al. have demonstrated miR-154-5p was remarkably down-regulated in adipose-derived mesenchymal stem cells (ADSCs) subjected to four-point bending stretch. ALP activity and matrix mineralization were altered treatments with lentivirus-miR-154-5p or inhibitor of miR-154-5p under mechanical stretch, which indicated that mechanical-stretch-sensitive miR-154-5p prevented osteogenic differentiation of ADSCs through the Wnt/PCP (planar-cell-polarity) pathway by directly targeting Wnt11 (Wnt family member 11) [37]. Another miR-195-5p also targeted a Wnt family member (WNT 3A) to inhibit PDLC differentiation under Flexcell stretch. In addition, miR-195-5p was involved in FGF (fibroblast growth factor) and BMP signaling through targeting FGF2 and BMPR1A (bone morphogenetic protein receptor-1A) during osteogenic differentiation of PDLCs [54]. Although compression force was a type of mechanical loading condition, the effects of compressive force on osteogenic differentiation were opposite to other loading methods [55]. It is reported that the miR-29 family, including miR-29a, miR-29b and miR-29c, significantly decreased approximately 0.5-fold and suppressed the target extracellular matrix gene (Col1a1, Col3a1 and Col5a1) expression in PDLCs treated with Flexcell stretch. On the contrary, the significantly increased (1.8-4-folds) expression of the miR-29 family were detected when PDLCs were subjected to compressive force [55].
Extracorporeal shockwave (ESW), another type of mechanical stress, transduced the mechanical signals into biological signals and promoted osteogenic differentiation [56]. The miR-138 level was decreased with post-ESW treatment and the overexpression of miR-138 abrogated osteogenic differentiation induced by ESW through down-regulating FAK (Focal adhesion kinase), Runx2 expression and mineralization in MSCs (mesenchymal stem cells) [57]. In BMSCs, miR-138 was reported to inhibit osteogenic differentiation by targeting FAK, and antagomir-138 (miR-138 inhibitor) obviously enhanced ectopic bone formation [58,59]. In a preceding study, miR-132 positively regulated FSS-induced PDLC osteogenic differentiation through the mTOR signaling pathway [39]. However, miR-132-3p, one subtype of miR-132, was up-regulated (about 2.5-fold) in a mechanical unloading group and negatively correlated with osteoblast differentiation in primary rat osteoblasts. Moreover, Ep300 (E1A binding protein p300), miR-132-3p's target, significantly decreased under mechanical unloading and suppressed the activity and acetylation of Runx2, a key regulatory factor of osteoblast differentiation [60]. This is further evidence that the same miRNA played opposite roles in the differentiation of different osteogenic cells subjected to various mechanical stimuli. Mechanosensitive miRNAs temporally changed with mechanical use/disuse, and played important roles in osteogenic differentiation under different mechanical conditions. Without mechanical stimuli, the gain or loss of function of the mechanosensitive microRNAs still affects many bio-active molecules indispensable to the regulation of osteogenic differentiation in nature [21,[61][62][63].

Mechanosensitive miRNAs and Osteoblast Proliferation
Not only did osteoblast differentiation change, but osteoblast proliferation also altered with the transformation of mechanical stimuli (Table 4). Experimental evidence showed that osteoblasts under compressive force at 294 Pa had slower cell growth and higher expression of miR-494-3p compared with no-compressive group cells. Moreover, overexpression of miR-494-3p repressed osteoblast proliferation assay through targeting Fgfr2 (fibroblast growth factor receptor 2) and Rock1 (Rho associated coiled-coil containing protein kinase 1) [36]. Zhang's group paid massive attention to the effects of mechanical unloading on osteoblast activity, and they found that a mechanical unloading of sensitive miRNA, miR-103 [64,65], was involved in osteoblast proliferation. Increasing miR-103 expression and a decreasing number of EdU (5-Ethynyl-2 -deoxyuridine) positive cells induced by mechanical unloading demonstrated that miR-103 was negatively correlated to MC3T3-E1 cell proliferation. The miR-103' target Cav1.2 (calcium voltage-gated channel), an L-type voltage-sensitive calcium channel (LTCC), decreased under the mechanical unloading environment, which was moderated by the miR-103 inhibitor [64,65]. Taken together, the roles of mechanosensitive miRNAs in bone cells were concentrated on osteogenic differentiation and osteoblast proliferation.

Mechanosensitive miRNAs and Bone Formation
It is reported that some mechanosensitive miRNAs are involved in bone formation in vivo (Table 5). Accumulating evidence suggests that almost 40% of reports of exercise-sensitive miRNAs were in relation to skeletal muscle [66], but less to bone in vivo. Exercise, a natural method to investigate the influence of mechanical loading on bone formation in vivo, is well known to strongly benefit trabecular bone parameters, bone mineral density (BMD) and bone mechanical strength, including treadmill walking, swimming and artificial cyclic strain (four-point bending on tibia) [67][68][69]. Daily bouts of exercise [70] or high-frequency motions [71] were seen as the primary interventions to reduce the risk of bone loss. Conditional knockout (cKO) miR17-92 cluster in collagen type I-producing osteoblasts resulted in a 13-34% reduction of total body bone mineral content, a 10% reduction in maximum load, a 28% reduction in the periosteal bone formation rate, but no change in bone toughness and the resorbing surface [72]. Interestingly, miR17-92 cluster cKO mice subjected to two weeks of mechanical exercise by four-point bending, revealed no obvious change in tibia periosteal bone formation, compared to wild-type mice [72]. Besides, Sengul reported that miR-92 inhibitor (antagomir-92) increased tibia TV (Tissue volume), BV/TV (Bone volume/tissue volume) and density by 7-16% after two weeks mechanical exercise, but a similar increase in the above parameters (6-15%) was seen in the antagomir-NC (inhibitor negative control) treated group with two weeks mechanical exercise [73]. Both studies revealed that the mechanosensitive miR17-92 cluster blocked bone formation in response to mechanical exercise loading. miR-222, a potential regulator of an articular cartilage mechanotransduction pathway, increased in the anterior weight-bearing area of articular cartilage as compared with the posterior non-weight-bearing area with minimum contact pressure [74].
Apart from weight-bearing bone formation, alveolar bone formation was also influenced by mechanosensitive miRNAs under mechanical loading. As a matter of fact, alveolar bone formation and incisor tooth mechanical properties also decreased in miR17-92 cluster cKO mice [75]. The role of mechanosensitive miR-21 in PDLSCs has been investigated in vitro [51]. Similarly, miR-21 was dose-and time-responsive to orthodontic force, which induced alveolar bone formation in vivo [76]. During orthodontic tooth movement (OTM), miR-21 deficiency prevented force-induced alveolar bone formation just in the tensile side, while blocking bone resorption in both the compressive and tensile sides in miR-21 knockout mice [76]. Meanwhile, Liu, et al. indicated that miR-503-5p was involved in OTM and the overexpression of miR-503-5p reduced Runx2, Alp mRNA and protein expression, and decreased osteoblast numbers and osteoblastic bone formation in the OTM tension sides [22].
While a lack of mechanical loading or physical activity disturbed the delicate balance of the homeostasis of weight-bearing bones [77], and a reduction in bone formation is frequently observed in the aging and in paraplegics [78]. It was known that sense and response to mechanical stimuli in middle-aged and old mice was less than in young adults, which led to the aged skeleton diminishing in bone formation [14,79,80]. There is growing evidence that age-altered miRNAs modulated bone formation [81][82][83]. More recently, miR-188 was identified as a key regulator of the age-associated switch between osteoblast and adipocyte differentiation of BMSCs [84]. Notably, the miR-188 level was positively correlated with age in both mouse and human BMSCs, whereas it was negatively correlated with trabecular bone formation parameters (BV/TV, Tb.N (Trabecular number)) and tibia maximum mechanical load. With aging, miR-188 knockdown in animals reduced decrease rates of bone formation and tibia mechanical strength; conversely, miR-188 overexpression in transgenic mice revealed a substantial age-related reduction of bone formation and tibia maximum mechanical load. From the molecular mechanism point of view, miR-188 inhibited the osteogenic differentiation of BMSCs by targeting hystone deacetylase 9 (Hdac9). In addition, the miR-188 inhibitor-via a BMSC-specific aptamer-relieved age-induced osteoporosis and reduced femur maximum mechanical load [84]. It is reported that miR-214 [85] was up-regulated, while miR-142-5p [86], miR-31 [83], the miR 17-92 cluster [82], miR-let-7f, miR-125b, miR-199-3p and miR-222 [87] were down-regulated with age increasing in different osteogenic cells or bone tissue in vivo.
In addition, external mechanical unloading resulted in the deterioration of bone micro-structure, particularly the trabecular bone compartment and weak mechanical properties in vivo [88,89]. Although the mechanosensitive miR-103 family has been identified as negatively regulating osteoblast differentiation [23] and proliferation [64,65] in vitro, there is still conflicting evidence showing either stimulatory or inhibitory effects of the miR-103 family on bone formation under mechanical unloading in vivo [23,90]. Pretreatment with miR-103a inhibitor partly counteracted the decrease of bone formation caused by hind-limb unloading [23], nevertheless the expression of miR-103-3p (same with miR-30b-5p and miR-142-3p) was significantly declined in the serum of rhesus monkeys after long-duration bed-rest and positively correlated with BMD [90]. Bed-rest was a reliable model for the simulation of mechanical unloading and miRNAs, as putative circulating biomarkers of diseases, were stable in plasma and serum. Li's group also demonstrated the circulating miRNA profile of plasma from 16 individuals after 45 days of −6 • head-down tilt bed-rest. 11 miRNAs including miR-103, -130a, -1234, -1290, -151-5p, -151-3p, -148a, -199a-3p, -20a, -363, and -451a revealed dramatically decreased after 45 days of bed-rest, while the expressions of these miRNAs were recovered after 10 days of mechanical recovery, except miR-148a, -199a-3p and -151-3p. Additionally, several miRNAs, especially miR-1234, were identified in positive correlation with bone formation parameters [91]. Expressions of miR-214 [85], miR-139-3p, miR-339-3p, miR-132-3p, miR-487b, miR-2985 and miR-34b [60] have been demonstrated to alter in the bones of hind-limb unloading animals. Significantly, miR-214 level was reported to vary with natural age in human osteoporotic bone specimens and increase in Alp + cells isolated from bones after 28 days hind-limb unloading, which implied that miR-214 might be potentially sensitive to mechanical unloading and played a crucial role in bone formation [85]. Furthermore, miR-214 has been shown to repress osteogenic differentiation or bone formation in different osteogenic cells through binding to various protein coding genes, such as Osterix in C2C12 myoblast cells [92], activating transcription factor 4 (ATF4) in osteoblasts [85], and FGF in BMSCs [93]. The roles of mechanosensitive miRNAs in bone formation were showed in Table 5. Block tibia periosteal bone formation response to mechanical exercise loading in the cKO mice; alveolar bone size and incisor tooth mechanical properties decreased in the cKO mice. [72,73,75] miR-222 anterior weight-bearing and posterior non-weight-bearing regions bovine articular cartilage A potential regulator of an articular cartilage mechanotransduction pathway. [74] miR-21 orthodontic force in periodontal tissue at 30 g of force for 7 d miR-21−/− mice miR-21 deficiency prevented force-induced alveolar bone formation just in the tensile side, while blocking bone resorption in both the compressive and tensile sides. [76] miR-503-5p the left maxillary first molars were mesially stretched at 50 g of force for 1, 3, 7, 14 d respectively alveolar bones Agomir-503-5p decreased Runx2, Alp mRNA and protein expression, decreased osteoblast numbers and osteoblastic bone formation in the orthodontic tooth movement (OTM) tension sides. [22] miR-188 3, 6, 12, 18-month mice 85 human females and 85 males at different ages BMSCs miR-188 level increased with age. Further, miR-188 deficiency revealed substantially reduced decrease of age-associated rates of bone formation and tibia maximum mechanical load; miR-188 transgenic mice determined opposite results; miR-188 inhibitor via a BMSC-specific aptamer relieved age-induced osteoporosis and reduction of femur maximum mechanical load. [84] miR-214 20 human females and 20 males at different ages Bone specimens Expression of miR-214 increased with age and was negatively correlated with the expressions of bone formation marker genes. [85] miR-142-5p Mice at 10 weeks and 18 months of age Bone callus miR-142-5p level was lower in the callus of aged mice than the young group, and Agomir-142-5p repaired the aged bone fractures. [86] miR-31 Mice at 2, 8, 25 months of age MSCs and bone tissue miR-31 level showed a consistent decreased trend with age increasing in MSCs and bone.

Conclusions and Perspectives
An increasing number of studies are showing that some miRNAs (e.g., miR 17-92 cluster, miR-103 family) have been identified in response to different mechanical stimuli and to act as crucial regulators in osteogenic differentiation and bone formation. However, controversial functions of mechanosensitive miRNAs existed in osteogenic differentiation under different mechanical stimuli: (1) Stretch stress and FSS showed the opposite effects of mechanosensitive miRNAs on osteogenic differentiation, for instance miR-21 and miR-33; (2) The expression of the mechanosensitive miR-132 family showed the same trend under opposite mechanical stimuli.
The future perspectives for the application of mechanosensitive miRNAs may include: (1) several mechanosensitive miRNAs mimics or inhibitors potentially served as therapeutic approaches for osteopenia due to a lack of mechanical loading; (2) the potential application of mechanosensitive miRNAs as biomarkers to assess the extent of osteopenia induced by the alteration of mechanical stimuli in patients' serum, plasma or exosomes; (3) exercise-sensitive miRNAs presumably accounted for bone-muscle interactions, owing to exercise changing many miRNAs expression in skeletal muscles; (4) mechano-sensitive miRNAs related to bone resorption is a compelling future research direction, since osteoclastogenesis and osteoclast activity altered with different mechanical stimuli [94][95][96]. Despite the functional roles of mechanosensitive miRNAs in osteogenic differentiation and bone formation being elucidated in the present review, the mechanisms of miRNA response to different mechanical stimuli during bone remodeling and clinical practice remain to be investigated in future.
Acknowledgments: This work was supported by funds from the Natural Science Foundation of China (No. 31570940), the Shenzhen Science and Technology Project (JCYJ20160229174320053), the Peak Experience Project of NPU.
Author Contributions: Zhihao Chen wrote the whole manuscript. Airong Qian and Ge Zhang made the outline of the manuscript. Zhihao Chen, Yan Zhang and Lei Chen searched for literatures cited in introduction and roles of mechanosensitive miRNAs in bone formation. Chao Liang made corrections for the manuscript. All of the authors proofread and approved the final manuscript.

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