miR-203 and miR-320 Regulate Bone Morphogenetic Protein-2-Induced Osteoblast Differentiation by Targeting Distal-Less Homeobox 5 (Dlx5)

MicroRNAs (miRNAs) are a family of small, non-coding RNAs (17–24 nucleotides), which regulate gene expression either by the degradation of the target mRNAs or inhibiting the translation of genes. Recent studies have indicated that miRNA plays an important role in regulating osteoblast differentiation. In this study, we identified miR-203 and miR-320b as important miRNAs modulating osteoblast differentiation. We identified Dlx5 as potential common target by prediction algorithms and confirmed this by knock-down and over expression of the miRNAs and assessing Dlx5 at mRNA and protein levels and specificity was verified by luciferase reporter assays. We examined the effect of miR-203 and miR-320b on osteoblast differentiation by transfecting with pre- and anti-miRs. Over-expression of miR-203 and miR-320b inhibited osteoblast differentiation, whereas inhibition of miR-203 and miR-320b stimulated alkaline phosphatase activity and matrix mineralization. We show that miR-203 and miR-320b negatively regulate BMP-2-induced osteoblast differentiation by suppressing Dlx5, which in turn suppresses the downstream osteogenic master transcription factor Runx2 and Osx and together they suppress osteoblast differentiation. Taken together, we propose a role for miR-203 and miR-320b in modulating bone metabolism.

The expression of the bone-inducing transcription factor, Dlx5, is associated with osteoblast differentiation [11] and exhibits its highest expression when extracellular matrix mineralizes at the final stage of osteoblast differentiation. Studies show that in BMP-2-induced osteoblast differentiation, Dlx5 plays an essential role in upregulation of the downstream osteogenic master transcription factor Runx2 and Osx, whose expression in turn sequentially regulates the expression of osteoblast specific genes to induce osteoblast differentiation [8,12]. Concomitantly, Dlx5 inhibits adipogenic differentiation by inhibiting PPARγ (Peroxisome proliferator-activated receptor gamma) expression in bone marrow mesenchymal stem cells (MSCs) [13].
Runx2 is the master transcription factor for osteoblast differentiation. Runx2 expression is substantially upregulated by BMP2 through the activation of Smad signaling [9,14]. The physical interaction of Runx2 with Smad1 and Smad5 is necessary to enhance its transcriptional and osteogenic activity. Studies have shown that the BMP2-Smad-Runx2 axis is important for osteoblast differentiation. This complex also induces alkaline phosphatase (ALP) activity and the expression of other osteoblast-specific genes.
Osterix is a novel zinc finger-containing transcription factor, which acts downstream of Runx2, and its expression is essential for osteoblast differentiation. During the differentiation of mesenchymal cells into osteoblasts, the expression of Osterix is considerably upregulated by BMP-2, suggesting the BMP2 acts upstream of Osterix during osteoblast differentiation [10].
MicroRNAs (miRNAs) are an abundant class of small, single-stranded non-coding RNAs (~22 nucleotides long). They have emerged as important post-transcriptional regulators in diverse processes, e.g., cell proliferation and differentiation [15,16]. MiRNAs anneal to the 3 UTR of their target genes and regulate protein translation and/or mRNA stability [17]. There is an increasing number of miRNAs identified recently that contribute to the regulation of osteoblast differentiation and bone formation. MiRNAs may target negative regulators of osteogenesis and as a result operate as positive regulators, or they may act as negative regulators by targeting important osteogenic factors. As a result, miRNAs exert control over skeletal gene expression [18][19][20][21]. Studies show that several bone-inducing transcription factors and signaling molecules that are involved in the function and differentiation of MSCs to osteoblasts are targets of miRNAs. Recent studies reported many bone-regulating miRNAs ("osteomiRs") that orchestrate BMP2-induced osteogenesis, such as miR-141 and miR-200a by downregulating Dlx5 represses BMP2-induced pre-osteoblast differentiation [22]. miR-133 directly targets Runx2, and miR-135 attenuates Smad5 pathway and inhibits osteoblast differentiation [23]. Studies also show that mir-93 attenuates osteoblast mineralization by directly targeting Osx [24], whereas mir-206 targets connexin 43 and inhibits osteogenesis [25].
In the present study, we aimed to identify miRNAs involved in BMP2-induced osteogenesis using primary human osteoblasts (HOBs). Our results show that miR-203 and miR-320b target Dlx5, a bone-inducing transcription factor, and together they suppress osteoblast differentiation. These results were confirmed by Western blot analysis, real-time PCR, and luciferase reporter assays. The activity of miR-203 and miR-320b were modulated using an antimiR oligonucleotide, which markedly increased osteogenic differentiation in vitro, whereas miR-203 and miR-320b overexpression reversed these effects. This study confirmed earlier reports that Dlx5 is a common upstream regulator of Runx2 and Osx, and both genes are regulated independently by Dlx5. These results strongly suggest that miR-203 and miR-320b suppresses BMP-induced osteogenic differentiation by suppressing Dlx5 and its downstream signaling.

Bone Cell Culture
Primary human osteoblast (HOB) cells were isolated from human trabecular bone collected from 3 donors undergoing total hip replacement as published previously [26][27][28]. The bone chips were washed thoroughly and minced with PBS. The minced bone chips were cultured in medium containing α-MEM (Sigma-Aldrich, Haverhill, UK) supplemented with 2 mmol/L L-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin and 10% fetal bovine serum (Sigma-Aldrich) at 37 • C with 5% CO 2 until confluence was reached. The culture medium was changed twice weekly. The study was approved by the local ethics committee (Ethical approval # Ups 03-561).

Total RNA Extraction
The HOB cells were harvested at the seven different time points. The cell lysates were homogenized using QIAshredder (Qiagen, Hilden, Germany). RNA was extracted from the cell lysates using the RNeasy Mini Kit (Qiagen). Agilent 2100 BioAnalyzer (Agilent Technologies, Palo Alto, CA, USA) was used to confirm high RNA quality for all samples, RIN values in our study were between 8.2 and 9.5. The concentrations were determined with NanoDrop ND-1000 (NanoDrop Technologies, Wilmington, DE, USA) with an OD 260/280 between 1.95 and 2.03.

Quantification by Real-Time PCR
Twenty nanograms of total RNA from each time point was reverse transcribed using TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA), according to the manufacturer's instructions, enabling miRNA specific cDNA synthesis for the miR-203 and miR-320 human miRNAs and 1 miRNA control.
Following the RT step, TaqMan MicroRNA Assays (Applied Biosystems) were performed using specific TaqMan miRNA probes hsa-miR-320b (ID 002844) and hsa-miR-203 (ID 000507) (Applied Biosystems) according to the manufacturer's instructions. PCR cycling began with AmpliTaq Gold enzyme activation at 95 • C for 10 min, then 40 cycles of 95 • C for 15 s, and 60 • C for 60 s performed on a 7500 Fast Real-Time PCR System (Applied Biosystems).
For the target osteogenic genes cDNA synthesis was performed in triplicate using total RNA reverse transcribed using High Capacity cDNA reverse transcription kit (Applied Biosystems), according to the manufacturer s instructions, with no template control added to ensure a lack of signal in assay background. The real-time PCR reactions were carried out with 10 µL of 2x TaqMan ® Universal PCR Master Mix, no AmpErase ® UNG (Applied Biosystems), 9 µL diluted cDNA, and 1 µL of TaqMan gene specific assay mix in a 20 µL final reaction volume. Reference gene beta-actin (ACTB) and GAPDH (Applied Biosystems) was selected as control for normalization of TaqMan data. Probes specific for Runx2 (Hs00231692_m1), Dlx5 (Hs00193291_m1), Osx (Hs01866874_s1), ACTB (Hs01060665_g1) and GAPDH (Hs02758991_g1) were purchased from Applied Biosystems. The amplification was carried out using the 7500 Fast Real-Time PCR System (Applied Biosystems) using a 40-cycle program. The 7500 software automatically calculates raw Ct (cycle threshold) values.
The comparative quantitation 2 −∆∆CT method (also called the ∆∆CT method) [32] was used to compare differences in cycle number thresholds for samples normalized for endogenous controls.

Luciferase Reporter Assay
After in silico target prediction as described above, the target region for miR-203 and miR-320 in Dlx5 was verified using Dual-Luciferase Reporter Assay. The psiCHECK-2 vector, a dual-luciferase plasmid, has both the synthetic Firefly Luciferase (Fluc) gene and the synthetic Renilla Luciferase (hRluc) gene incorporated, each possessing its own promoter and poly (A)-addition sites. Luciferase reporter plasmids were constructed by inserting a perfectly complementary (Wild type) 3 UTR fragment of Dlx5 between the XhoI-NotI restriction sites in the multiple cloning regions in the hRluc gene in the psiCHECK-2 vector (Promega, Madison, WI, USA) by Generay Company (Shanghai, China). We also constructed 3 mutant types by replacing 6-8 base pairs at the 3 -UTR of the seed sequence. DNA sequencing was employed confirming the nucleotide sequences of the constructed plasmids.

Statistical Analysis
The statistical difference between pairs of groups was determined by student s t-test. Two-way analysis of variance (ANOVA) was used to evaluate the statistical significance for comparisons within groups. p < 0.05 was considered as statistically significant. Data were analyzed using Statistica v12 software (Stat Soft Inc., Tulsa, OK, USA,) and GraphPad Prism software package (version 6.0, GraphPad Software, Inc., La Jolla, CA, USA).

miR-203 and miR-320b Targets Dlx5 as Shown by in Silico Analyses
In a BMP-2 induced osteoblast differentiation model, we aimed to determine target genes for miR-203 and miR-320b, that we previously had identified as miRNAs important in modulating osteoblast differentiation [33]. The putative binding sites for miR-203 and miR-320b were predicted using TargetScan, PicTar and miRanda, and Dlx5 was selected as one of the candidate target genes ( Figure 1A). The seed sequences of these miRNAs were conserved across species ( Figure 1B). In a BMP-2 induced osteoblast differentiation model, we aimed to determine target genes for miR-203 and miR-320b, that we previously had identified as miRNAs important in modulating osteoblast differentiation [33]. The putative binding sites for miR-203 and miR-320b were predicted using TargetScan, PicTar and miRanda, and Dlx5 was selected as one of the candidate target genes ( Figure 1A). The seed sequences of these miRNAs were conserved across species ( Figure 1B).

BMP-2 Stimulates Dlx5, Runx2 and Osx Expression
We examined the expression pattern of Dlx5, Runx2 and Osx in human osteoblast cells when stimulated with BMP-2 at different time intervals. In the absence of BMP-2, the expressions of these genes are low, but under BMP-2 stimulation, the HOBs displayed a gradual increase in expression of Dlx5, Runx2 and Osx mRNA up to 120 h (Figure 2A-C). The expression of miR-203 and miR-320b in BMP-2 stimulated cells showed a significant down regulation up to 120 h compared to cells untreated cells ( Figure 2D).

BMP-2 Stimulates Dlx5, Runx2 and Osx Expression
We examined the expression pattern of Dlx5, Runx2 and Osx in human osteoblast cells when stimulated with BMP-2 at different time intervals. In the absence of BMP-2, the expressions of these genes are low, but under BMP-2 stimulation, the HOBs displayed a gradual increase in expression of Dlx5, Runx2 and Osx mRNA up to 120 h (Figure 2A-C). The expression of miR-203 and miR-320b in BMP-2 stimulated cells showed a significant down regulation up to 120 h compared to cells untreated cells ( Figure 2D).

miR-203 and miR-320b Regulate the Expression of Osteogenic Transcription Factor Dlx5
To verify experimentally if the putative binding sites of miR-203 and miR-320b were functional and to evaluate the role of miR-203 and miR-320b on osteoblast differentiation stimulated by BMP-2, HOBs were transfected with miR-203 and miR-320b mimic, anti-miR and NC. mRNA and protein levels were assessed at set time points using qPCR and Western blotting. miRNAs were overexpressed by transfecting HOBs with either miR-203 or miR-320b mimics or both, showing that endogenously expressed miRNA significantly reduced the expression of Dlx5 (Figure 3A,B). A concurrent decrease in the levels of Runx2 (Figure 3C,D) and Osx ( Figure 3E,F) at the mRNA and protein levels up to 120 h when compared to cells treated with mimic NC was also observed.

miR-203 and miR-320b Regulate the Expression of Osteogenic Transcription Factor Dlx5
To verify experimentally if the putative binding sites of miR-203 and miR-320b were functional and to evaluate the role of miR-203 and miR-320b on osteoblast differentiation stimulated by BMP-2, HOBs were transfected with miR-203 and miR-320b mimic, anti-miR and NC. mRNA and protein levels were assessed at set time points using qPCR and Western blotting. miRNAs were overexpressed by transfecting HOBs with either miR-203 or miR-320b mimics or both, showing that endogenously expressed miRNA significantly reduced the expression of Dlx5 (Figure 3A,B). A concurrent decrease in the levels of Runx2 (Figure 3C,D) and Osx ( Figure 3E,F) at the mRNA and protein levels up to 120 h when compared to cells treated with mimic NC was also observed.  Conversely, the expression of either miR-203 or miR-320b or both was repressed using anti-miR, to assess if miRNA repression regulated osteoblast differentiation. Dlx5 expression was significantly increased ( Figure 4A,B), and also observed was a concurrent increase in the levels of Runx2 (Figure Conversely, the expression of either miR-203 or miR-320b or both was repressed using anti-miR, to assess if miRNA repression regulated osteoblast differentiation. Dlx5 expression was significantly increased ( Figure 4A,B), and also observed was a concurrent increase in the levels of Runx2 ( Figure 4C,D) and Osx ( Figure 4E,F) at the mRNA and protein levels up to 120 h when compared to cells treated with anti-miR NC. GAPDH was used as an internal control for the Western blots. GAPDH for only one time point (120 h) is shown as the representative blot for all the time points. GAPDH levels for the rest of the time points are similar. All these results suggest that miR-203 and miR-320b negatively regulates Dlx5 expression at translational level in BMP-2 stimulated osteoblast differentiation, and in turn regulates Runx2 and Osx expression.

miR-203 and miR-320b Regulates BMP-2 Stimulated Human Osteoblast Differentiation
To study the impact of miR-203 and miR-320b on BMP-2 stimulated differentiation of human osteoblast, HOBs were transfected with either pre-miR or anti-miR. The osteoblast differentiation marker ALP expression was significantly enhanced with the inhibition of miR-203 and miR-320b after BMP-2 treatment compared to no BMP-2 treatment and NC, and in vitro matrix mineralization was visualized by Alizarin red staining. In contrast, HOBs transfected with pre-miR miR-203 and miR-320b, the ALP activity and matrix mineralization was reduced ( Figure 5A,B). Taken together, these results suggest that miR-203 and miR-320b act as negative regulators of osteoblast differentiation.

miR-203 and miR-320b Regulates BMP-2 Stimulated Human Osteoblast Differentiation
To study the impact of miR-203 and miR-320b on BMP-2 stimulated differentiation of human osteoblast, HOBs were transfected with either pre-miR or anti-miR. The osteoblast differentiation marker ALP expression was significantly enhanced with the inhibition of miR-203 and miR-320b after BMP-2 treatment compared to no BMP-2 treatment and NC, and in vitro matrix mineralization was visualized by Alizarin red staining. In contrast, HOBs transfected with pre-miR miR-203 and miR-320b, the ALP activity and matrix mineralization was reduced ( Figure 5A,B). Taken together, these results suggest that miR-203 and miR-320b act as negative regulators of osteoblast differentiation.

miR-203 and miR-320b Directly Target Dlx5
Target prediction algorithms TargetScan, PicTar and miRanda were utilized identifying Dlx5 as a potential target gene for miR-203 and miR-320b. To determine whether miR-203 and miR-320b

miR-203 and miR-320b Directly Target Dlx5
Target prediction algorithms TargetScan, PicTar and miRanda were utilized identifying Dlx5 as a potential target gene for miR-203 and miR-320b. To determine whether miR-203 and miR-320b directly target Dlx5, a Renilla luciferase reporter was constructed in the psiCHECK-2 vector containing the 3 UTR (Untranslated region) of Dlx5 (wild-type). In addition, three types of mutant luciferase reporter plasmids were constructed, containing mutation in the 3 UTR of Dlx5 ( Figure 6A). The wild-type and the three mutant plasmids were co-transfected with pre-miR-203 and pre-miR-320b in HOBs and the level of firefly and Renilla luciferase activity was measured consecutively. Luciferase activity was suppressed in cells transfected with the wild-type plasmid overexpressed with either miR-203 or miR-320b or both compared to NC. Mutations at the binding site of both the miRNA abolished this decrease in luciferase activity ( Figure 6B). In all experiments, the psiCHECK-2 vector constitutively expressing firefly luciferase activity served as a normalization control for transfection efficiency. These results strongly indicate that miR-203 and miR-320b specifically target Dlx5.
directly target Dlx5, a Renilla luciferase reporter was constructed in the psiCHECK-2 vector containing the 3' UTR (Untranslated region) of Dlx5 (wild-type). In addition, three types of mutant luciferase reporter plasmids were constructed, containing mutation in the 3' UTR of Dlx5 ( Figure 6A). The wild-type and the three mutant plasmids were co-transfected with pre-miR-203 and pre-miR-320b in HOBs and the level of firefly and Renilla luciferase activity was measured consecutively. Luciferase activity was suppressed in cells transfected with the wild-type plasmid overexpressed with either miR-203 or miR-320b or both compared to NC. Mutations at the binding site of both the miRNA abolished this decrease in luciferase activity ( Figure 6B). In all experiments, the psiCHECK-2 vector constitutively expressing firefly luciferase activity served as a normalization control for transfection efficiency. These results strongly indicate that miR-203 and miR-320b specifically target Dlx5.

Discussion
BMP-2 is a highly efficient inducer of osteogenesis, regulating osteoblast differentiation by binding to and thereby stimulating the BMP receptors (BMPR) BMPR-I and BMPR-II. The downstream regulators Smad proteins play an important role in relaying BMP signals from receptors to activate the expression of three osteogenic master transcription factors Dlx5, Runx2 and Osx in the nucleus [34]. Runx2 in turn regulates the expression of many osteoblastic genes, e.g., Col1A1 (Collagen type I AI), alkaline phosphatase (ALP), Osteopontin (OPN) and osteonectin (ON) [35,36] (Figure 7). Many bone regulating miRNAs have been recently identified exerting their effects at different stages of osteoblast differentiation when stimulated with BMP-2. Studies show that Smad proteins may bind to miRNA promoter genes and control their transcription and regulate miRNA biogenesis and processing [37][38][39][40]. Furthermore, during BMP-2 induced osteogenesis, the expressions of many osteomiRs are downregulated.

Discussion
BMP-2 is a highly efficient inducer of osteogenesis, regulating osteoblast differentiation by binding to and thereby stimulating the BMP receptors (BMPR) BMPR-I and BMPR-II. The downstream regulators Smad proteins play an important role in relaying BMP signals from receptors to activate the expression of three osteogenic master transcription factors Dlx5, Runx2 and Osx in the nucleus [34]. Runx2 in turn regulates the expression of many osteoblastic genes, e.g., Col1A1 (Collagen type I AI), alkaline phosphatase (ALP), Osteopontin (OPN) and osteonectin (ON) [35,36] (Figure 7). Many bone regulating miRNAs have been recently identified exerting their effects at different stages of osteoblast differentiation when stimulated with BMP-2. Studies show that Smad proteins may bind to miRNA promoter genes and control their transcription and regulate miRNA biogenesis and processing [37][38][39][40]. Furthermore, during BMP-2 induced osteogenesis, the expressions of many osteomiRs are downregulated. To our best knowledge, this is the first study that relates miR-203 and miR-320b to Dlx5 in a BMP-2-induced human osteoblast differentiation. Previous studies have shown that miR-141 and To our best knowledge, this is the first study that relates miR-203 and miR-320b to Dlx5 in a BMP-2-induced human osteoblast differentiation. Previous studies have shown that miR-141 and miR-200a regulate BMP-2 mediated osteoblast differentiation by targeting Dlx5 [22], miR-208 target Ets1 and regulate BMP-2 induced mouse preosteoblast differentiation [41]. miR-203 has previously been shown to target Runx2 in osteolytic bone disease [42,43]. Our study was conducted in primary human osteoblast cells, and we used BMP-2 to induce osteoblast differentiation. Based on previous studies conducted in our group [33], we identified miR-203 and miR-320b as miRNAs important during osteoblast differentiation. Our results indicate that miR-203 and miR-320b, at least in part, regulate osteoblast differentiation by downregulating a common target, Dlx5. Studies have shown that several miRNAs may bind to their target mRNAs and cooperatively fine-tune the degree of protein translation of a single mRNA [44].The regulation of common target genes in which upstream regulators also have an impact on each other, is reminiscent of feed-forward loops established within a transcription factors-based hierarchy. Dlx5. We validated the Dlx5 as a target of miR-203 and miR-320b using Luciferase assays. BMP-2 downregulates miR-203 and miR-320b in a time dependent manner.
Studies by Ulsamer and Holleville et al [45,46] have shown that Dlx5 regulates the expression of Runx2 and Osx and their expression is upregulated when induced by BMP-2. To assess the effect of miR-203 and miR-320b on Dlx5 in regulating osteoblast differentiation, we overexpressed miR-203 and miR-320b and observed that osteoblast differentiation was suppressed by the decrease in the expression of Dlx5, Runx2 and Osx. Reciprocally, we also observed that when miR-203 and miR-320b are repressed, this results in an increased expression of the all three transcription factors. Although there was a clear effect by knockdown of miR-203 and miR-320b increasing the levels of Dlx5, Runx2 and Osx by 70%-80%, effects of other modulating factors including other osteogenic miRNAs cannot be ruled out. ALP activity was also diminished and mineral nodule formation was suppressed with miR-203 and miR-320b overexpression. In this paper, we demonstrate for the first time that miRNA 203 and miR-320b regulate Dlx5 in a human model of BMP-2 induced osteoblast differentiation.
In summary, we have identified two osteoblast related miRNAs that are crucial for regulating osteoblast differentiation. We indicate in our study that miR-203 and miR-320b actively inhibit osteogenesis and BMP-2 addition down regulated the miRNAs and promoted osteoblast differentiation. We observed that miR-203 and miR-320b together mediated repression of the transcription factor Dlx5, which is upstream of and activator of Runx2 and Osx, and they are important for the regulation of osteoblast differentiation. In view of these results, it can be suggested that downregulation of miR-203 and miR-320b by BMP-2 augmented the expression of Dlx5, hence bone formation. Taken altogether, we describe a novel function of miR-203 and miR-320b in negatively regulating BMP-2-induced osteoblast differentiation by suppressing Dlx5, which in turn suppresses the downstream osteogenic master transcription factor Runx2 and Osx. We also show that inhibition of miR-203 and miR-320b can increase osteoblast differentiation. Further studies are needed, investigating the role of miR-203 and miR-320b in modulating bone metabolism in vivo.

Conclusions
In summary, our study reveals a novel function of miR-203 and miR-320b in negatively regulating BMP-2-induced osteoblast differentiation by suppressing Dlx5, a bone-inducing transcription factor, which in turn suppresses the downstream osteogenic master transcription factor Runx2 and Osx and together they suppress osteoblast differentiation. Taken together, we propose that miR-203 and miR-320b suppresses BMP-induced osteogenic differentiation by suppressing Dlx5 and its downstream signaling.