The role of microRNAs in bone development

Abstract

Epigenetic regulation is critical for proper bone development. Evidence from a large body of published literature informs us that microRNAs (miRNAs) are important epigenetic factors that control many aspects of bone development, homeostasis, and repair processes. These small non-coding RNAs function at the post-transcriptional level to suppress expression of specific target genes. Many target genes may be affected by one miRNA resulting in alteration in cellular pathways and networks. Therefore, changes in levels or activity of a specific miRNA (e.g. via genetic mutations, disease scenarios, or by over-expression or inhibition strategies in vitro or in vivo) can lead to substantial changes in cell processes including proliferation, metabolism, apoptosis and differentiation. In this review, Section 1 briefly covers general background information on processes that control bone development as well as the biogenesis and function of miRNAs. In Section 2, we discuss the importance of miRNAs in skeletal development based on findings from in vivo mouse models and human clinical reports. Section 3 focuses on describing more recent data from the last three years related to miRNA regulation of osteoblast differentiation in vitro. Some of these studies also involve utilization of an in vivo rodent model to study the effects of miRNA modulation in scenarios of osteoporosis, bone repair or ectopic bone formation. In Section 4, we provide some recent information from studies analyzing the potential of miRNA-mediated crosstalk in bone and how exosomes containing miRNAs from one bone cell may affect the differentiation or function of another bone cell type. We then conclude by summarizing where the field currently stands with respect to miRNA-mediated regulation of osteogenesis and how information gained from developmental processes can be instructive in identifying potential therapeutic miRNA targets for the treatment of certain bone conditions.

1. Introduction: bone development and microRNAs

Bone development begins by the establishment of mesenchymal stem cell condensations that prefigure the shape, size and location of mature bone elements [1,2]. In addition to these skeletal patterning events, mesenchymal cells within these condensations receive signals to differentiate toward either cartilage-forming chondrocytes or bone-forming osteoblasts. The end result of a complex, tightly-controlled progenitor cell differentiation program is the generation of bone tissue consisting of a unique mineralized extracellular matrix (ECM).

Bone formation that transitions via a cartilaginous template is referred to as endochondral ossification and occurs during limb development [3-5]. Cells within this cartilage template terminally differentiate to form large hypertrophic chondrocytes that regulate mineralization of the surrounding ECM and induce vessel invasion. These events lead to the formation of a primary ossification center, and eventually cancellous bone following replacement of cartilage tissue by bone ECM components. Until recently, it was generally accepted that endochondral bone-forming osteoblasts are derived from progenitor cells lining or within blood vessels as well as from progenitor cells of the adjacent perichondrium [6]. However, it has now been established in recent years that hypertrophic chondrocytes, or a subset of progenitor cells within hypertrophic cartilage, are a significant source of osteoblasts during endochondral ossification [7-10]. Coupling of chondrogenesis and osteogenesis is also apparent in the formation of cortical bone of the limbs. Mature cortical bone is derived from the bone collar region in the perichondrium of developing limbs. While this process generally involves differentiation of progenitor cells directly to osteoblasts, hypertrophic chondrocytes have been proposed to play an important role in regulating “perichondrial osteogenesis” [3,4,6,11]. In addition to endochondral bone formation, other bones in the body are generated without the requirement of a cartilage template and involve cells within mesenchymal condensations differentiating directly to osteoblasts. This process, called intramembranous ossification, occurs in some bones of the cranium, and parts of the mandible and clavicle [12].

Toward the end phases of endochondral or intramembranous bone development, some osteoblasts will give rise to osteocytes that are found embedded deep within the mineralized bone ECM. Osteocytes can communicate with adjacent cells (including surrounding osteocytes, osteoblasts, osteoclasts, endothelial cells) via cytoplasmic extensions that occupy tiny canals called canaliculi [13]. These cells play critical mechano-sensing roles to control bone formation and homeostasis. Specifically, there is evidence that osteocytes can regulate the differentiation and function of bone-forming osteoblasts as well as the bone-resorbing osteoclasts [14]. In addition, crosstalk between osteoblasts and osteoclasts also occurs thereby increasing the complexity controlling bone development and turnover [15].

With respect to molecular regulation of osteoblast differentiation during endochondral or intramembranous bone formation, many important players have been identified. These include key transcription factors that are critical for chondrocyte or osteoblast formation (i.e. SOX9, RUNX2, respectively), homeodomain proteins that control various stages of osteoblast differentiation [16-19], growth factors (including FGFs, IGFs, VEGF, BMPs and other TGF-β superfamily members) as well as other signaling pathways (Wnt/β-catenin, Hedgehog, PTHrP, etc) [3-6]. It is also apparent that bone formation is regulated by epigenetic factors that can function at the level of transcription or translation to alter gene or protein expression. Examples of epigenetic regulators include histone modifying enzymes (HDACs, HACs), enzymes that control DNA methylation (DNMTs, TETs), long non-coding RNAs and microRNAs (miRNAs) [20-22].

Since the first discovery of a miRNA in Caenorhabditis elegans over 25 years ago [23], many more have now been identified in cells of humans, rodents, flies, viruses, plants and other species. In the current miRBase website (http://www.mirbase.org), 1917 mature miRNAs have been identified in humans and 1234 in mice. Mature non-coding miRNAs are commonly 19–24 nucleotides (nt) in length and are derived from larger precursor RNAs. Genes encoding miRNAs (predominantly located in intergenic regions or within introns of protein-coding genes) are first transcribed as large primary precursors (pri-miRNAs). In some cases, miRNA-encoding genes may be clustered (i.e. adjacently located within 10 Kb of each other as per miRBase definition) and transcribed in a polycistronic manner. Primary miRNA transcripts are processed in the nucleus by a Drosha-containing complex and the resulting precursor miRNAs (pre-miRNA) are transported to the cytoplasm and processed further by a Dicer-containing complex to form a short, mature miRNA duplex containing a 5p and 3p strand [24]. Commonly, one of these strands is functional whereby it will bind via its seed sequence (positions 2–8 of the mature miRNA strand) to a complementary region within the 3′UTR of a target mRNA. This interaction occurs within the RNA-induced silencing complex (RISC), the end result being either degradation of the target mRNA or inhibition of mRNA translation [25,26]. Fig. 1 depicts the stages of miRNA transcription, processing, and interaction with a target mRNA. Compared to short interfering RNAs (siR-NAs) that are generated exogenously, the level of miRNA-induced target suppression is quite modest. This is because the entire sequence of siR-NAs bind with high specificity (100% complementarity) to one mRNA target resulting in robust suppression via mRNA cleavage, while miRNAs interact with target mRNAs via imperfect pairing, with the exception of the seed sequence interaction [27]. However, unlike siRNAs, miRNAs have the ability to target tens to even hundreds of mRNAs within a given cell type [28], thereby resulting in modulation of many cellular pathways and networks. Complexity is enhanced by the fact that multiple miRNAs may compete to bind to a specific target mRNA.

Fig. 1.

Biogenesis, processing and function of microRNAs in the cell. In the nucleus, RNA polymerase II (RNA pol II)-mediated transcription results in generation of primary miRNA transcripts which are processed by a complex containing Drosha and DiGeorge Critical Region 8 (DGCR8) to form stem-loop precursor miRNAs (pre-miRNA). Pre-miRNAs are exported out of the nucleus by Exportin 5 and, in the cytoplasm, are processed further by a complex containing Dicer and TAR RNA Binding Protein (TRBP). Following unwinding of mature miRNA duplexes, one functional strand (here shown as the 5p strand) enters the RNA-induced Silencing Complex (RISC) where it binds to a specific region of the 3′UTR of its target mRNA. Complementary binding via the miRNA seed sequence is shown with black dotted lines. The outcome of miRNA-mediated binding is either mRNA degradation or suppression of mRNA translation.

Although miRNAs account for only 1–5% of the human genome [29], up to 60% of protein-coding genes may be modulated by miRNAs [30] thereby rendering these non-coding RNAs as major epigenetic regulators in the cell. While the majority of miRNAs carry out their function in the cytoplasm, many miRNAs have also been localized to other organelles including the nucleoli, processing bodies and mitochondria [31]. It has also been demonstrated that miRNAs can exist inside extracellular vesicles such as exosomes and that cells can communicate with each other via exosomal delivery of miRNAs [32,33]. Overall, a massive body of published research indicates that miRNAs play important roles in development, homeostasis, turnover, disease and repair of many different tissue types. Specifically, miRNAs have been shown to regulate a wide range of cellular processes including proliferation, cytoskeleton formation, apoptosis, growth factor signaling, metabolism, cell differentiation, and many others.

In this review, we will discuss some in vivo findings that highlight the importance of miRNAs in regulating skeletal development with an emphasis on osteogenesis. In Section 2, some information describing the importance of miRNAs in skeletal development via in vivo mouse models and human clinical reports is similar to that included in our previously published review on miRNAs in orthopaedic research [34]. We believe that such details are also important in the context of this review as well. However, new information is described in Section 2 that was not included in our previous review including details on Prx1-Cre and CD11b-Cre deletion of Dicer in mice, the effects of deleting miR-181 and miR-182 in vivo, a recent report describing a new human neomorphic mutation in miR-140 [35], and a newly published manuscript describing the smallest human deletion mutation in 1q24 containing a microRNA cluster that is associated with skeletal phenotypes [36]. Given some recently published reviews on miRNAs regulating bone formation [34,37-39] we have focused Section 3 of this review on findings published predominantly within the last three years with respect to miRNA regulation of osteoblast differentiation. Preference was given to discussing studies where a miRNA target was validated and/or a specific pathway or cellular process was identified to be regulated by the miRNA of interest. While bone-resorbing osteoclasts are also important in the regulation of bone development, information on how miRNAs affect these cells will be covered by another review in this Special Edition. However, this review will include details of some studies showing that miRNAs in exosomes derived from either osteoclasts or osteocytes can regulate osteoblast differentiation/function, thus highlighting the complex miRNA-mediated crosstalk that likely occurs between different cell types in bone.

2. The importance of miRNAs in skeletal development: lessons from in vivo findings

a). Modulation of miRNA processing via Cre-driver lines in skeletal cells

Conditional transgenic mice devoid of proteins involved in miRNA processing have been generated. While the function of these proteins is not completely restricted to regulating miRNAs, findings from these mice (which contain substantially lower levels of functional miRNAs in specific cell types) suggest an important role for these non-coding RNAs in controlling cellular processes involved in proper skeletal development.

Deletion of the pre-miRNA processing enzyme, Dicer, in osteochon-droprogenitor cells via Prx-1-Cre mice resulted in formation of smaller limbs due, in part, to increased cell death during early limb bud development. Interestingly, there did not appear to be defects in basic patterning or in overall cartilage and bone differentiation within these smaller limbs [40]. However, when Dicer was eliminated specifically in chondrocytes by crossing Dicer floxed mice with Col2a1-Cre driver mice, growth plate analysis revealed modest defects in chondrocyte differentiation which subsequently resulted in reduced length and width of the developing long bones [41]. Given the cross-talk between hypertrophic cartilage and bone collar/cortical bone formation, it would have been interesting to determine if there were any defects in matrix mineralization and bone production in the limbs of these mice. Thorough analysis of post-natal bone formation was not possible given that these mice normally die around the time of weaning. Following Dicer deletion in osteoprogenitor cells via Col1a1-Cre mice, skeletal examination of E14.5 embryos revealed compromised ECM mineralization in cartilage and osseous tissues as well as an overall significant reduction in bone tissue [42]. Whether this phenotype represents delayed rather than inhibition of ossification was not concluded from these studies given that embryo survival was compromised after E14.5, which may have been partly due to the lack of marrow cavity formation which support hematopoiesis. This study also showed that when Dicer was deleted specifically in mature osteocalcin-producing osteoblasts (Ocn-Cre), a post-natal increase in long bone and vertebral bone mass (but not cranial bone mass) was found [42]. These findings suggest an overall positive role for Dicer-generated miRNAs in regulating osteoblast differentiation during embryonic bone development, particularly bones that are generated by endochondral ossification. On the other hand, the Ocn-Cre findings suggest that miRNAs may generally function to suppress bone formation during post-natal bone development and turnover. However, conditional knock-out of Dicer via Runx2-Cre revealed growth retardation, low bone mass and impaired bone formation rate in post-natal mice [43]. A recent study also revealed decreased cortical bone mass in young and adult mice following inducible post-natal ablation of Dicer in osterix (Sp7)-producing cells (Sp7-CreER^T2; Dicer^flox/flox) [44]. It was also shown that Dgcr8 (DiGeorge syndrome critical region 8) deletion in osteopro-genitors (Col1a1-Cre) resulted in enhanced bone formation due, in part, to decreased miR-22 and enhanced osteocalcin transcripts [45]. Dgcr8 is a critical component of the pri-miRNA processing complex via its interaction with Drosha. Taken together, it is apparent that the timing and the choice of Cre-driver model to induce deletion of miRNA processing proteins are critical factors in determining the outcome of miRNA deficiency on bone formation.

Dicer has also been deleted predominantly (but not specifically) in murine osteoclasts via CD11b-Cre [46] or Cathepsin K-Cre [47] approaches. In both cases, increased post-natal bone mass was found due to impaired osteoclastogenesis. Whether bone mass or mineralization was affected at the embryonic level was not investigated in these studies. The latter study by Mizoguchi et al [47] also reported no changes in post-natal bone mass when Dicer was removed by Col1a1-Cre mediated deletion. This result is intriguing given that Gaur et al reported embryonic lethality when this enzyme was knocked-out using the same 2.3 kb collagen type I promoter Cre-mice [42]. Similar defects in osteoclast differentiation and function were noted when Dgcr8 was deleted in osteoclasts using Cathepsin-Cre mice [48]. These mice also displayed growth retardation as well as increased bone mass.

b). miRNAs regulating human skeletal development

Data from clinical reports have demonstrated the importance of miRNAs in regulating human skeletal development. One study found that a mutation within the 3′UTR of HDAC6 disrupts the miR-433 binding site resulting in a dominant X-linked chondrodysplasia [49]. Deletion or duplication of the miR-17~92 cluster was reported to cause short stature, digit abnormalities, microcephaly and abnormal facial features [50,51]. Additional findings show that deletion of miR-17~92 specifically in murine Col1α2-expressing cells resulted in smaller bones as well as a reduction in periosteal bone formation following mechanical loading [52].

A recent report identified a mutation within the gene encoding miR-140, one of the most studied miRNAs in cartilage. The resulting production of neomorphic mutant miR-140-5p caused a number of skeletal defects in family members including short stature, brachydactyly, premature degeneration of intervertebral discs and delayed epiphyseal ossification of the hip and knee [35]. Previous findings from miR-140 knock-out mice [53,54] revealed shortened limbs as a result of defects in endochondral ossification due to accelerated hypertrophic chondrocyte differentiation.

Deletions within specific regions of chromosome 1q24 appear to cause a range of skeletal phenotypes and cognitive disabilities [55-58]. Skeletal issues include short stature, microcephaly, brachydactyly and, in some cases, a marked delay in bone age. Interestingly, the various deletions reported contain the clustered miRNAs, miR-199a and miR-214, which are located within a long non-coding RNA (lncRNA) transcript called DNM3OS. Further evidence that heterozygous deletion of these non-coding RNAs may be responsible for the skeletal phenotype comes from a newly published study reporting the smallest 1q24 microdeletion to date (94 Kb) in the genome of two patients [36]. In addition, transgenic mice devoid of Dnm3os, and hence the miR-199a~214 cluster, presented with similar skeletal phenotypes to that reported in human patients [59]. While one cannot rule out the possibility that Dnm3os may function as a lncRNA independent of its role in serving as a miRNA precursor, it is likely that the miRNAs themselves play a role in regulating the skeletal phenotype. In fact, a number of published studies have reported a functional role for miR-199a or miR-214 in regulating osteogenesis in vitro or in vivo.

c). miRNAs with functional roles in regulating bone formation in vivo

Altered expression of other miRNAs in mice has revealed functional roles in regulating bone formation in vivo. For example, expression of members of the miR-34 family (miRs-34a, b, c) were found to increase during osteoblast differentiation of murine calvarial cells [60]. This study also showed that when miR-34b and miR-34c were deleted in Col1a1-producing cells in mice, increased bone mass was observed during embryonic development and increased bone mass accrual was observed post-natally. The opposite was found when miR-34c was over-expressed in murine Col1a1-producing cells. One of the mechanisms proposed for the negative effects of miR-34 on bone development was via targeting and suppression of SATB2 (special AT-rich sequence-binding protein 2). In agreement with this work, miR-34a was found to suppress osteoblast differentiation of human MSCs in vitro (in part via targeting Jagged 1) and reduce formation of bone following subcutaneous transfer of hMSC-loaded ceramic beads in SCID mice [61]. Interestingly, when miR-34a was over-expressed predominantly in osteoclasts, osteoclastogenesis and bone resorption was suppressed and provided some protection against ovariectomy (OVX)-induced bone loss [62]. These findings demonstrate the functional divergence between miR-34 family members but also highlight the fact that modulation of miRNAs in different cell types within the same tissue can result in different biological outcomes.

Expression of miR-206, regarded as more of a muscle-specific miRNA, was found to be one of a number of miRNAs downregulated during BMP-2-induced osteogenic differentiation of C2C12 cells [63]. This study also showed that conditional over-expression of miR-206 in Col1a1-positive cells resulted in low bone mass in mice due, in part, to targeting and suppression of connexin 43 (Cx43). It has also been reported that when members of the miR-181 family were globally deleted in mice, those that survived were smaller in size [64,65]. This suggests a potential role for these miRNA paralogs in controlling growth plate and bone development. As will be discussed more in this review, miR-181a/b has been shown to enhance osteogenesis in vitro.

A recent report showed that miR-182 inhibits osteoclastogenesis in vitro and that myeloid-specific (LysM-Cre) deletion of miR-182 in mice results in increased post-natal trabecular bone mass [66]. These mice were apparently protected from bone loss associated with OVX and inflammatory arthritis. Mechanistically, miR-182 was found to target protein kinase double stranded RNA dependent (PKR) and regulate IFN-β signaling. As expected, this study also showed that in vivo over-expression of miR-182 in osteoclasts resulted in low trabecular bone mass. Cortical bone mass was not affected by miR-182 modulation and it is not clear if trabecular bone mass was affected at earlier embryonic stages of bone development. In agreement with these negative effects on bone mass, previous studies have shown that miR-182 (in part via targeting FoxO1) suppresses osteoblast differentiation in vitro and impairs bone formation in vivo in zebrafish [67]. Another study showed that suppression of miR-182 enhanced osteoblast differentiation in vitro and apparently induced the effects of alendronate in combating osteoporosis in rats via regulating Rap1/MAPK signaling [68].

In vitro studies found that miR-21 appeared to enhance both osteogenesis and osteoclastogenesis [69,70]. Interestingly, global knock-out of miR-21 in mice [71] did not appear to affect bone development, which may be due to functional redundancy by other miRNAs expressed during development. However, miR-21 deficiency promoted trabecular bone mass with age and also prevented OVX-induced bone loss during aging due, in part, to suppressed osteoclast function. In this case, as has been reported for many transgenic mice, a post-natal phenotype exists following aging or a disease/injury challenge, even though a developmental/baseline phenotype is absent.

3. Recent research on miRNAs regulating osteoblast differentiation

Given the large body of published studies and some recent reviews on osteogenic regulation by miRNAs, this section will cover research findings reported within the last three years (2017 to present). PubMed search keywords included “miRNA or microRNA and osteoblastogenesis or osteogenesis”. Preference was given to studies that determined a miRNA target gene and/or cellular pathway modulated by the miRNA. Fig. 2 lists the osteo-enhancing and osteo-suppressing miRNAs from this PubMed search as well some functional miRNAs discussed in Section 2 of this review. Tables 1 and 2 provide more information on the findings from our literature search covering the last three years. Table 1 lists miRNAs reported to enhance osteoblast differentiation in vitro when over-expressed while Table 2 lists miRNAs that suppress in vitro osteogenesis. Additional information provided in the Tables include the cell types used to study osteogenesis in vitro and, where applicable, the endogenous expression patterns of the miRNA of interest during in vitro osteogenesis, as well as the in vivo animal model utilized. The majority of these studies involved altering miRNA activity using either mimics or antagomirs whereas those that utilized lentiviral approaches are highlighted. We believe it is important to emphasize the mode of miRNA modulation given a report suggesting that transient transfection of miRNA mimics often results in supraphysiological levels of over-expression that may lead to non-specific changes in gene expression [72].

Fig. 2.

MicroRNAs regulating osteoblast differentiation and function. Listed are the osteo-enhancing and osteo-suppressing miRNAs included in Section 3 in addition to other miRNAs discussed in Sections 2 and 4 of this review. Confirmed miRNA targets are shown in parenthesis. Note that while long bone formation involves endochondral ossification, via formation of hypertrophic chondrocytes, all studies identifying functional miRNAs utilized in vitro osteogenesis assays that mimic intramembranous ossification whereby stem/progenitor cells differentiate directly to osteoblasts. Also shown are miRNAs present in exosomes of osteoclasts or osteocytes that may function in regulating osteoblast differentiation or function. See Tables 1 and 2 for a list of the various stem/progenitor cell types used to determine the function of the majority of miRNAs listed in this figure. Note: not included in this figure is the reference to a report suggesting that let-7a-5p, present in osteoclast-derived exosomes, may enhance hypertrophic chondrocyte differentiation [164]. Note: while osteoblasts certainly secrete exosomes, we did not identify published studies reporting exosome-derived miRNAs from osteoblasts directly regulating other specific bone-related cell types. Therefore, we did not include osteoblast-derived exosomes in this figure given that we cannot support this depiction with a specific published study.

Table 1.

MicroRNAs that enhance osteoblast differentiation in vitro.

Osteo-enhancing miRNA	Endogenous expression during OG	miRNA target	Progenitor cell type	Animal model	Reference
miR-10b		SMAD2	Human ASC	Heterotopic ossification	Li et al. [73]
miR-19a-3p		HDAC4	Human BMSC	NA	Chen et al. [74]
miR-26b		GSK3β	Rat BMSC	NA	Hu et al. [75]*
miR-29a		HDAC4	Human BMSC	NA	Tan et al. [76]
miR-29a	NA	DKK1	Mouse primary Cal Pre-OB	Ovariectomy-induced osteoporosis	Lee et al. [77]
miR-92a	NA	SMAD6	Mouse BMSC	NA	Yan et al. [78]
miR-98		HMGA2	Human BMSC	NA	Gao et al. [79]
miR-130a		SMURF2	Mouse BMSC	Age-related osteoporosis	Lin et al. [80]
miR-135-5p		HIF1AN	MC3T3-E1	NA	Yin et al. [81]
miR-181a/b-1		PTEN	Human DDC	NA	Zheng et al. [82]*
miR-219a-5p		RORβ	Mouse Cal OB	NA	Aquino-Martinez et al. [83]
miR-342-3p		ATF3	Human BMSC, MC3T3-E1	NA	Han et al. [84]
miR-365	NA	HDAC4	MC3T3-E1	NA	Xu et al. [85]
miR-374b	NA	PTEN	Mouse BMSC	Tibial plateau fracture	Ge et al. [86]
miR-433-3p	NA	DKK1	Human FOB 1.19, Rat ROS17/2.8	Ovariectomy-induced osteoporosis	Tang et al. [87]*
miR-450b		BMP3	Human ASC	Heterotopic ossification	Fan et al. [88]*

Legend: this list was generated from a PubMed literature search (2017–2020) to determine miRNAs that play a functional role in regulating osteoblast differentiation. This list contains miRNAs reported to enhance osteogenesis following over-expression in vitro. The second column shows endogenous expression patterns of the miRNA during in vitro osteogenesis assays: pink arrows represent reduced expression, green arrows denote increased expression and yellow arrows translate to unchanged expression at specific time points of osteogenic differentiation. MiRNA target genes and progenitor cell types used to study in vitro osteogenesis are also shown. Wherever applicable, the in vivo rodent model used to further test the effects of miRNA modulation on bone formation is also shown. Those studies marked with an asterisk (*) used lentiviral technology to modulate miRNA expression. All other studies utilized miRNA mimics or antagomirs. Abbreviations: ASC = adipose-derived mesenchymal stem/stromal cells; BMSC = bone marrow-derived mesenchymal stem/stromal cells; Cal OB = calvarial-derived osteoblasts; Cal pre-OB = calvarial-derived pre-osteoblasts; DDC = dedifferentiated chondrocytes (or cartilage progenitor cells); FOB = fetal osteoblasts; NA = not analyzed; OG = osteogenesis; ROS = rat osteosarcoma cells.

Table 2.

MicroRNAs that suppress osteoblast differentiation in vitro.

Osteo-suppressing miRNA	Endogenous expression during OG	miRNA target	Progenitor cell type	Animal model	Reference
miR-9-5p		WNT3A	Rat BMSC	NA	Zhang et al. [99]
miR-16-2-3p		WNT5A	Human BMSC	NA	Duan et al. [100]
miR-23a	NA	BMPR1B	Human PDLSC	NA	Zhang et al. [101]*
miR-23a–27a–24-2		HOXA5/10/11	MC3T3-E1	Age-related osteoporosis	Godfrey et al. [102]*
miR-27b-3p		SP7/OSX	Human MSMSC	Heterotopic ossification	Peng et al. [103]
miR-93-5p		BMP2	Human BMSC	NA	Zhang et al. [104]
miR-98	NA	BMP2	Human BMSC	NA	Zhang et al. [105]
miR-106b		BMP2	Human PMSC	Glucocorticoid-induced osteoporosis	Liu et al. [106]
miR-125b		BMPR1B	Human BMSC	Bone defect	Wang et al. [107]*
miR-125b		NKIRAS2	Human PDLSC	NA	Xue et al. [108]*
miR-132		GDF5	Human PDLSC	NA	Xu et al. [109]
miR-137-3p	NA	RUNX2, CXCL12	Rat BMSC	Steroid-induced osteonecrosis of the femoral head	Kong et al. [110]*
miR-138		RHOC	Human DDC	NA	Zheng et al. [111]*
miR-139-3p	NA	ELK1	MC3T3-E1	NA	Wang et al. [112]
miR-145		FOXO1	Human ASC	NA	Hao et al. [113]
miR-150		GPNMB	Mouse BMSC	Age-related osteoporosis	Moussa et al. [114]
miR-155		BMPR2, RUNX2	C2C12, MEF	Heterotopic ossification	Liu et al. [115]
miR-181a-3p		BMP10	Human BMSC	NA	Tao et al. [116]
miR-185	NA	BGN	MC3T3-E1, mouse BMSC	Ovariectomy-induced osteoporosis	Cui et al. [117]
miR-185-5p		DLX2	MC3T3-E1	NA	Chang et al. [118]
miR-193a		HMGB1	Human BMSC	NA	Wang et al. [119]
miR-200a		GLS	Human BMSC	NA	Lv et al. [120]
miR-206		GLS	Human BMSC	NA	Chen et al. [121]
miR-214-3p	NA	RUNX3	Human MSMSC	Heterotopic ossification	Peng et al. [122]
miR-223		DHRS3	Human BMSC	NA	Zhang et al. [123]
miR-339		DLX5	Human BMSC	NA	Zhou et al. [124]
miR-451		YWHAZ	Cal Pre-OB	Ovariectomy-induced osteoporosis	Pan et al. [125]
miR-451a	NA	BMP6	Mouse BMSC, MC3T3-E1	Ovariectomy-induced osteoporosis	Lu et al. [126]
miR-485-5p		SP7/OSX	Rat BMSC	NA	Zhang et al. [127]
miR-495	NA	HMGA2	Mouse Cal OB	Bone defect	Tian et al. [128]
miR-665		IL-6	Human ASC	NA	Wu et al. [129]
miR-1187		BMPR2, ARHGEF-9	Mouse Cal OB	Ovariectomy-induced osteoporosis	John et al. [130]

Legend: this list was generated from a PubMed literature search (2017–2020) to determine miRNAs that play a functional role in regulating osteoblast differentiation. This list contains miRNAs reported to suppress osteogenesis following over-expression in vitro. The second column shows endogenous expression patterns of the miRNA during in vitro osteogenesis assays: pink arrows represent reduced expression, green arrows denote increased expression and yellow arrows translate to no significant changes in expression at specific time points of osteogenic differentiation. MiRNA target genes and progenitor cell types used to study in vitro osteogenesis are also shown. Wherever applicable, the in vivo rodent model used to further test the effects of miRNA modulation on bone formation is also shown. Those studies marked with an asterisk (*) used lentiviral technology to modulate miRNA expression. All other studies utilized miRNA mimics or antagomirs. Abbreviations: ASC = adipose-derived mesenchymal stem/stromal cells; BMSC = bone marrow-derived mesenchymal stem/stromal cells; Cal OB = calvarial-derived osteoblasts; Cal Pre-OB = calvarial-derived pre-osteoblasts; DDC = dedifferentiated chondrocytes (or cartilage progenitor cells); MEF = mouse embryonic fibroblasts; MSMSC = maxillary sinus membrane stem cells; NA = not analyzed; OG = osteogenesis; PDLSC = periodontal mesenchymal stem cells; PMSC = placenta-derived mesenchymal stem cells.

Table 1 shows that, where applicable, the endogenous expression of miRNAs reported to have enhancing activity appears to increase during in vitro osteogenesis, particularly at the early phases of differentiation induction. Additional over-expression of each miRNA listed in Table 1 [73-88] in a range of rodent or human progenitor cell lines enhanced the osteogenic program as shown by increased mineralized matrix formation (commonly detected by Alizarin red staining) and increased expression of osteoblast-related genes when compared to control cultures. A number of these osteo-enhancing miRNAs were found to target and suppress other epigenetic regulators including HDAC4 or HMGA2 [74,76,79,85]. Another common mechanism to enhance osteoblast differentiation is via miRNA-mediated suppression of Wnt inhibitors (i.e. DKK1 or GSK3β) [75,77,87] or via targeting negative regulators of RUNX2 (i.e. SMAD6 or SMURF2) [78,80]. Less commonly reported miRNA-mediated mechanisms to enhance bone formation include regulation of retinoic acid receptor-related orphan receptor beta (Rorβ). This transcription factor, which is a negative regulator of bone, was reported to be a direct target of miR-219-5p, thereby partly explaining the mechanism by which this miRNA enhances osteogenesis [83]. Two reports in Table 1 [82,86] show that miRNA-mediated targeting of PTEN (phosphatase and tensin homolog) results in enhanced osteogenesis. PTEN functions as the primary negative regulator of PI3K/AKT signaling in the cell [89] thereby affecting a number of cellular processes. Also, it was previously reported that mice lacking PTEN have higher bone mass and improved fracture healing [90,91], providing further evidence that mechanisms to suppress PTEN may have positive effects on bone formation. In addition to confirming PTEN protein suppression by over-expression of the miR-181a/b-1 cluster, research from our laboratory also showed that PI3K/AKT signaling was indeed increased during early phases of osteogenic differentiation following miR-181a/b-1 over-expression [82]. In this study, we also carried out RNA-Seq following over-expression of miR-181a/b-1 or a non-silencing control RNA during osteogenesis and found, via pathway analyses, that a number of cell processes related to mitochondrial metabolism were increased. We subsequently showed by Seahorse technology that mitochondrial respiration was elevated during osteogenesis by miR-181a/b-1 and that enhancing PI3K/AKT signaling may be partly responsible for these metabolic changes. Preliminary unpublished data from our laboratory has also shown enhanced endochondral ulnar fracture healing in mice following lentiviral delivery of miR-181a/b-1 to the fracture site. While it is known that glycolysis is a major metabolic process during osteoblast differentiation, a number of other studies clearly show that mitochondrial respiration is also critical for proper bone formation [92-97]. Research in our lab is currently focused on determining other miR-181a/b-1 target genes or pathways regulated by this miRNA cluster that may be responsible for regulating mitochondrial respiration. Interestingly, another in vitro study showed that miR-181a over-expression enhanced osteogenesis via repression of TGF-β signaling [98]. Undoubtedly, miR-181a/b paralogs will likely target a number of genes in the cell that will subsequently modulate multiple cellular pathways to regulate osteoblast differentiation.

Table 2 shows that endogenous expression of all miRNAs reported to have negative effects on osteoblast differentiation decreased during in vitro osteogenesis assays, particularly during the earlier phases of differentiation. The suppressive effect of these miRNAs on osteoblast differentiation was demonstrated following their over-expression by mimics or virus-mediated transduction [99-130]. Common target genes of a number of these miRNAs were found to be known activators of osteoblast differentiation including BMP2 [104-106], other BMPs [116,126], BMP receptors [101,107,115,130], specific WNTs [99,100] or transcription factors RUNX2 [110,115] and Osterix/Sp7 [103,127]. In agreement with the osteo-inhibitory function of miR-214-3p via targeting ATF4 [131], Table 2 lists a more recent study demonstrating that this miRNA can also target and suppress RUNX3 [122]. With respect to modulating cellular metabolism, two independent studies reported miRNA-mediated targeting and suppression of glutaminase (which regulates glutamine metabolism) was partly responsible for decreased osteogenesis [120,121]. This work is in agreement with a recent report showing a positive role for glutaminase in regulating osteoblast differentiation [132].

Research from our laboratory showed that miR-138 inhibited osteogenesis, in part, via targeting and suppression of RHOC [111]. This was the first report implicating a role for this small GTPase in regulating osteoblast differentiation. We also found that a major effect of miR-138 over-expression and RHOC suppression was inhibition of actin polymerization. Another miRNA listed in Table 2, miR-1187, was also shown to suppress osteogenesis by inhibiting actin cytoskeletal rearrangement [130]. Together, these findings agree with previous studies showing that suppression of actin cytoskeleton formation has negative effects on osteoblast differentiation [133-135]. The inhibitory effect of miR-138 on osteogenesis was also shown previously and this study identified focal adhesion kinase (FAK) as a target of miR-138 [136]. Interestingly, previous studies have shown that cells defective in either RHOC or FAK are less invasive/metastatic [137,138]. A recent in vivo study showed that delivery of miR-138 antagomirs could enhance bone formation in a murine model of multiple myeloma [139]. While these findings are encouraging and support the in vitro data on miR-138 in regulating osteogenesis, long-term effects of inhibiting a miRNA with reported tumor suppressive activity should be considered. On the other hand, miR-138 over-expression may be a useful strategy to attempt to inhibit pathological bone formation. Indeed, preliminary, unpublished data from our laboratory showed a reduction in trauma-induced heterotopic bone formation in mice following lentiviral delivery of miR-138. Current studies are focused on investigating these findings further.

Determining the function and mechanism of a specific miRNA in regulating in vitro osteogenesis can be a useful first-step approach to justify pursuing in vivo studies in rodents. Such pre-clinical studies are strengthened by the fact that many miRNAs are conserved between rodent and human. Tables 1 and 2 contain information on studies that utilized models of bone loss, bone repair or ectopic bone formation to determine how increasing or inhibiting miRNA activity can affect bone formation in these scenarios. In addition, two of these studies also reported age-related changes in miRNAs whereby miR-130a levels were lower in BMSCs from older mice [80] while miR-219a-5p expression was decreased in bone samples from old mice or from aged humans when compared to young controls [83]. A number of other recent studies have also reported age-related changes in miRNAs in bone or bone marrow cells as a consequence of age [140-145]. Therefore, identifying miRNAs that are not only functional in regulating osteogenesis, but also appear to be regulated with age, or even in bone disease scenarios, would further improve the discovery of effective miRNA targets to treat low bone mass or enhance bone repair.

4. miRNA-mediated crosstalk in bone

It is now well-established that intercellular communication between different cell types via exosomes, microvesicles or matrix vesicles is important for proper regulation of bone development, turnover and repair [146-153]. Exosomes are extracellular vesicles (EVs) with an average diameter of ~100 nm and contain protein, DNA, RNA and other components depending on the cell type from which they are derived. They originate from the endosomal pathway via the formation of multivesicular bodies (MVBs). Exosome-containing MVBs can fuse with the plasma membrane resulting in exosome release from the cell [154]. The discovery that exosomes containing miRNAs are present in circulation has led to many studies aimed at identifying miRNA biomarkers associated with various bone diseases [155-157]. In addition, there are new research endeavors exploring the possibility of exosome-derived miRNAs in mediating intercellular signaling between bone cells (osteoblasts, osteoclasts, osteocytes) and other cell types involved in bone formation (e.g. BMSCs, periosteal progenitors, hypertrophic chondrocytes).

A recent study reported that expression of miR-31a-5p was higher in BMSC-derived exosomes from aged rats compared to young rats [140]. This miRNA negatively affects osteoblast differentiation but promotes osteoclast differentiation and bone resorption, in part, via targeting SATB2 and RhoA, respectively. Inhibition of miR-31a-5p apparently prevented bone loss and decreased osteoclast activity in vivo. Overall, their findings suggest that miR-31a-5p is a modulator of the bone marrow microenvironment to influence both osteoblast and osteoclast differentiation during aging and that BMSC-derived exosomes may be a significant source of this miRNA. Another study found that bone marrow-derived EVs from aged mice had higher levels of the miR-183 cluster (miR-96~183) compared to EVs isolated from young animals [142]. The negative effects of miR-183 on BMSC osteogenesis was also demonstrated in this work. These findings suggest that enriched miRNAs within BMSC-derived EVs may interact with other BMSCs in the bone marrow microenvironment to reduce their osteogenic potential. In general, it also very likely that the contents of BMSC-derived EVs may also influence the function of surrounding osteoblasts and osteoclasts to affect bone metabolism [158].

Emerging evidence also suggests that contents of exosomes derived from differentiated bone cells can regulate other cells types involved in bone formation/turnover. One study, showed that exosomes derived from a mineralizing osteoblast cell line (MC3T3) can promote osteoblast differentiation of the ST2 cell line [159]. While the transfer of miRNA cargo from exosomes was not examined in this work, exosome treatment was found to significantly alter miRNA profiles in the recipient cells. Other in-depth studies have demonstrated that exosomal miR-214 derived from osteoclasts could be transferred to osteoblasts to inhibit osteoblast activity and bone formation [160,161]. Recent work has also shown that miR-23a-5p from osteoclast-derived exosomes can suppress osteoblast differentiation, in part, by targeting Runx2 [162].

Further evidence of miRNA-mediated osteoblast-osteoclast communication comes from research showing accumulation of miR-125b within osteoblast-derived matrix vesicles in bone [163]. The authors of this work suggested that miR-125b could be released into the bone marrow microenvironment to suppress osteoclastogenesis and bone resorption. A recent report suggests that let-7a-5p from osteoclast-derived exosomes can enhance the expression of hypertrophic genes in the chondrocyte ATDC5 cell line [164]. These findings imply that miRNA-mediated intercellular communication between osteoclasts and chondrocytes may influence terminal hypertrophic chondrocyte differentiation. Another study demonstrated that exosomes derived from a common osteocyte cell line (Ocy454) could be taken up by osteoblastic MC3T3 cells resulting in a marked decrease in osteogenic potential of these cells [165]. This study also showed that myostatin treatment altered miRNA profiles in these osteocyte-derived exosomes and that, in particular, reduced expression of miR-218 was partly responsible for the negative effects on osteogenesis. A novel mechanism controlling muscle-bone communication was postulated from these findings. Further indication of possible miRNA-mediated crosstalk between osteocytes and osteoblasts was described in a recent study showing that miR-181b-5p in osteocyte-derived exosomes enhances osteogenesis of human periodontal ligament stem cells [166]. Interestingly, similar to our research, this miR-181 family member was also shown to target PTEN and enhance PI3K/AKT signaling [82].

However, it should be noted that the majority of these studies demonstrating exosomal miRNA-mediated cell-cell communication have been performed exclusively in vitro using cell lines rather than primary cells. While the concept of miRNA-mediated intercellular communication via EVs is intriguing and certainly possible given the proximity of the different cell types involved in regulating bone development and homeostasis, more research is needed to elucidate the mechanism of exosome transfer and miRNA uptake and to conclusively demonstrate this in vivo. One study suggests that an interaction between ephrinA2 and EphA2 facilitates the recognition of osteoclast exosomes by osteoblasts [161]. Most likely, other processes will be necessary to permit miRNA transfer via exosomes within the bone microenvironment and basic science research is currently ongoing to better understand this phenomenon in other systems [167,168]. In vitro studies using fluorescently-labeled exosomes or miRNAs has indeed shown that transfer to osteoblasts is possible [160,161,169]. Also, it was demonstrated that fluorescently-labeled prostate cancer cell-derived exosomes enriched in miR-141-3p could home to bone in mice to induce osteoblastic bone metastasis due to the pro-osteogenic function of miR-141a-3p [169]. While these studies are certainly rigorous and informative, more sophisticated imaging technologies [170] would be required to fully characterize exosome uptake and transfer of miRNAs to cells in the bone microenvironment in vivo.

5. Summary and perspectives

The importance of miRNA-mediated epigenetic regulation in controlling skeletal development and homeostasis is clearly evident. There is now a vast array of published reports demonstrating how osteoblast differentiation can be modulated in vitro by either over-expressing or inhibiting a specific miRNA. In many of these studies, the functional effect of targeting the miRNA of interest has been further confirmed in vivo utilizing rodent models of osteoporosis, bone fracture repair or heterotopic ossification, for example. It is therefore apparent that understanding how miRNAs regulate bone developmental processes will aid in the design of new therapeutic strategies to treat bone conditions.

These small non-coding RNAs are recognized as attractive therapeutic targets due to their size, known sequence and the fact that they can target multiple genes to subsequently alter cellular pathways and networks. This function is particularly relevant in the context of more complex diseases via the ability to target “interactomes”. In fact, a number of Phase I/II human clinical trials are underway toward testing the effects of miRNA mimics or antagomirs in vivo to treat specific diseases including cancers. While new research endeavors to improve the stability and cellular uptake of mimics/antagomirs are underway, additional efforts are needed to target cells specifically to the bone microenvironment. Indeed, a few studies have reported some success in the design of peptides/nanoparticles to target osteoclasts or osteoblasts in vivo [171-174]. Perhaps also attempts to better understand the biology behind why exosomes derived from certain cancer cells appear to home to bone [169] would also be advantageous toward designing strategies to predominantly target bone cells.

Many miRNAs known to regulate osteoblast differentiation have been discovered via expression profiling during in vitro osteogenesis using microarray or bulk RNA-Seq approaches. Other candidates have been identified from analysis of RNA extracted directly from bone or bone marrow. While useful information has been gained by such approaches, the ability to now determine expression of miRNAs at the single cell level in vitro or in vivo will aid in determining either new miRNA candidates to pursue and/or confirm those already shown to regulate bone cell function to then justify further studies. The introduction of miRNA-mRNA co-sequencing at the single cell level [175] will also aid in the development of regulatory networks that exist in different bone cells. These technologies may be useful in determining how miRNAs and regulatory networks change with age in specific bone cells, for example. Findings from such analyses could aid in determining miRNAs or miRNA-regulated pathways that could be targeted to combat age-related bone loss or to enhance bone repair that is often deficient as a consequence of aging.

However, what should be done with this ever-increasing list of miRNAs that have been reported to have similar functions in either enhancing or suppressing osteogenesis? In this review, we have covered only a fraction of the published literature on miRNAs regulating osteoblast differentiation. Also, additional miRNA candidates that have been shown to regulate terminal hypertrophic chondrocyte differentiation [176-180] may also turn out to be potential targets to enhance bone formation and repair given the requirement of hypertrophic cartilage in these endochondral processes. It may be that a large-scale project would be required to systematically test and directly compare the function of many of these miRNAs in vitro and in vivo to generate a consensus toward determining the most promising candidates to pursue with respect to modulating osteogenesis. From such studies, we may also learn that targeting multiple miRNAs may be better than modulating one in certain scenarios. Also, given that therapeutic strategies will likely result in miRNA delivery to multiple cell types within the bone microenvironment, the function of specific miRNAs regulating osteoblasts should also be determined in osteoclasts and other bone-associated cell types as well. This is particularly important given that miRNA-mediated cross-talk between bone cells may also represent a significant mode of regulation to control bone formation, turnover and repair.