BMSCs from C57BL/6 mice were purchased from the Cyagen Biosciences, Inc. (Guangzhou, China). The culture were maintained in α-MEM (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Invitrogen; Thermo Fisher Scientific, Inc.) and 1% penicillin-streptomycin (Invitrogen; Thermo Fisher Scientific, Inc.) at 37°C in a humidified atmosphere containing 5% CO₂. Then, the medium was changed every 3 days. For all experiments, BMSCs were seeded at the density of 2.5×10⁴ cells/cm² in the cell culture dishes and cultivated until they reached 80% confluence (0 day). Then the cells were induced with osteoinductive medium containing 100 nmol/l dexamethasone, 50 µg⁄ml ascorbic acid and 10 mmol/l β-glycerophosphate (Invitrogen; Thermo Fisher Scientific, Inc.) and the medium was changed every 3 or 4 days. Control cultures were incubated under the same conditions with normal medium (α-MEM supplemented with 10% FBS and 1% penicillin-streptomycin).

Following induction of BMSCs with osteoinductive medium, the BMSCs were lysed by brief sonication on ice in radioimmunoprecipitation lysis buffer (CW Biotech, Beijing, China), and the protein concentration of the cell lysates were measured using the Bichinchoninic Acid Protein Assay kit (CW Biotech). The activity of ALP in the lysates was measured with a ALP assay kit (Nanjing Jiancheng Biotechnology Co., Ltd., Nanjing, China) using a p-nitrophenyl phosphate method, according to manufacturer's instructions. A single unit of ALP activity represented 1 µmol p-nitrophenyl phosphate hydrolyzed to p-nitrophenol/min, therefore the ALP activity in the proteins was expressed in U/g protein.

Following cultured with osteoinductive medium, the cells were washed with PBS and fixed with 4% paraformaldehyde for 20 min at 37°C. The fixed cells were soaked with the ALP staining kit (Nanjing Jiancheng Biotechnology Co., Ltd.) and Alizarin red-S staining kit (Nanjing Jiancheng Biotechnology Co., Ltd.) according to the manufacturer's instructions, and were then observed under an optical microscope.

After the BMSCs were induced with osteoinductive medium for 21 days, extracts of the cells were harvested in lysis buffer, protein contents of the cell lysates were quantified by BCA Protein Assay kit (CW Biotech). Proteins were separated in 10% sodium dodecyl sulfate-polyacrylamide gels, transferred to nitrocellulose membranes blocked with 2.5% BSA in Tris-buffered saline with 0.1% Tween-20, and incubated with primary antibodies, respectively. Membranes were then incubated with secondary antibodies conjugated with horseradish peroxidase (Santa Cruz Biotechnology, Inc., Dallas, TX, USA) respectively, and immunoreactive bands were detected using an enhanced chemiluminescence detection reagent (Pierce; Thermo Fisher Scientific, Inc., Waltham, MA, USA). β-actin in cell lysates was used as a loading control, data were normalized against those of corresponding optical density of β-actin.

miRNA microarray (Agilent Technologies, Inc., Santa Clara, CA, USA) was used to detect the miRNA expression levels in the BMSCs. The miRNA expression profiles of the BMSCs which were induced with osteoinductive medium were compared with the control group (cultured without osteoinductive medium). miRNAs were enriched from total RNA extracted using a mirVana miRNA Isolation Kit and labeled with a mirVana Array Labeling Kit (Ambion; Thermo Fisher Scientific, Inc.). The labeled miRNAs were used for hybridization on each miRNA microarray to determine the differential expression of miRNAs. This procedure was repeated twice. Target labeling, hybridization, imaging and data processing were performed at Phalan Biotech (Beijing, China) according to the protocols described by the manufacturer using the Mouse miRNA OneArray Microarray (Agilent Technologies, Inc.) and Sanger miRBase 19. Data were acquired using Agilent Feature Extraction software version 10.7 (Agilent Technologies, Inc.). Further data analyses were performed using GeneSpring GX 10.0 (Agilent Technologies, Inc.).

The expression levels of miRNAs were detected using RT-qPCR. The primers for RT-qPCR of miRNAs were provided by Guangzhou FulenGen Co., Ltd. (Guangzhou, China), the information of the primers were listed in Table I. Following cDNA synthesis using All-in-One^™ miRNA First-Strand cDNA Synthesis Kit (Guangzhou FulenGen), qPCR was performed using All-in-One^™ miRNA qPCR kit (Guangzhou FulenGen) according to the manufacturer's instructions. The reactions were incubated in a 96-well optical plate at 95°C for 10 min, followed by 40 cycles of 10 sec at 95°C, 20 sec at 60°C and 10 sec at 70°C (annealing and extension). Expression analysis was performed in triplicate for each sample. The average value of six reference genes were used as the normalization control (U6, U68, U70, U49A, U65, U72). The miRNA expression levels were quantified using an ABI Prism 7300 Sequence Detection system (Applied Biosystems; Thermo Fisher Scientific, Inc.).

As miRNAs play their biological roles through regulating target genes expression at the post-transcriptional level, target genes of the miRNAs were predicted by online softwares, including miRDB (http://www.mirdb.org/), MicroRNASeq (http://cm.jefferson. edu/rna22v2/), TargetScan (http://wwwargetscan.org), and miRanda (http://www.microrna.org/). To identify the biological functions of these genes, Gene Ontology (GO) (http://www.geneontology.org/) and Kyoto encyclopedia of genes and genomes (KEGG) (http://www.genome.jp/) enrichment analysis were performed, respectively. Molecular Functions, Biological Process and Cellular Component of Gene Ontology were included. Search for gene annotation information using Ontology Gene database (http://www.geneontology.org/). According to the gene GO annotation, statistical methods were applied to calculate P-values, with P<0.01 as the significant threshold. All genes in this species were used as background genes. The high frequency notes with statistical significance were obtained and normalized to the background, information of distribution and significance of GO based gene sets were obtained. Based on the KEGG database (http://www.genome.jp/), gene sets were enriched in the biological pathway associated with osteoblastic differentiation were selected.

To identify differentially expressed miRNAs among the groups, Student's t-test was performed using SPSS 18.0 (SPSS, Inc., Chicago, IL, USA). The experiments were repeated in triplicate. Statistical significance between the groups was measured using Student's t-test; P<0.05 was considered to indicate a statistically significant difference.

The effect of osteoinductive medium on the differentiation of osteoblasts was investigated via measuring various markers of osteoblastic differentiation, including ALP, calcium deposits and OCN levels (Fig. 1). Following exposure of the BMSCs to osteoinductive medium, the ALP activities in BMSCs which were induced for 7 days and 14 days were notably higher than that in control group (Fig. 1A). Protein levels of OCN were also increased in BMSCs which were induced for 21 days (Fig. 1C). In addition, the images of ALP and Alizarin red-S staining of BMSCs indicated that the osteoinductive medium increased the levels of ALP and calcium deposits in the BMSCs at indicated time (Fig. 1B and D). ALP, OCN, and calcium deposits are all markers of osteoblastic differentiation (19–22). Therefore, the osteoinductive medium promoted osteoblastic differentiation of the BMSCs.

miRNA microarray was used to detect the miRNA expression levels in BMSCs. The results revealed that the miRNAs were deregulated in the induced group under the condition of ‘P<0.05 and fold-change >1.5’, compared with control group. Among them, the expression levels of 107 miRNAs were higher and the expression levels of 40 miRNAs were lower in the group induced with osteoinductive medium compared with the non-induced control group (Fig. 2, Table II).

21 miRNAs were selected from the chip results (the 21 miRNAs were all expressed at 14 days and 21 days, amomg the 21 miRNAs, 2 miRNAs were both expressed at 7 days), and used RT-qPCR for further validation. The results of RT-qPCR confirmed that the expression levels of mmu-let-7c-5p, mmu-miR-181c-3p, mmu-miR-3092-3p, and mmu-miR-5132-3p had the same expression trend as the microarray in the osteoinductive medium groups (Fig. 3). In the four miRNAs, mmu-let-7c-5p was differently expressed at 7, 14 and 21 days, which was testified with both miRNA microarray and RT-qPCR. The other three miRNAs were all differently expressed at 14 and 21 days (testified with both miRNA microarray and RT-qPCR). Therefore, these four miRNAs were considered to be responsive to the osteoinductive medium applied to the BMSCs.

Four online softwares (miRanda, Clip-Seq, TargetScan, and miRDB) were applied to predict target genes for the four differentially expressed miRNAs. Prediction results of the four software were further screened and sorted out in Table III (target genes predicted by any three softwares mentioned above at the same time were regarded as a candidate target gene of corresponding miRNA). As a result, 540 genes were predicted as putative target genes of the miRNAs respond to osteoinductive medium.

To identify the biological functions of these genes, GO and pathway enrichment analysis were performed, respectively. All the GO/KEGG annotation results in pathways and molecular functions related to osteogenic differentiation or bone formation were searched to excavate functions related genes. As illustrated in Fig. 4 and Table IV, the top 20 GOs sensitive to osteoinductive medium were showed in Fig. 4. In the result of KEGG, three items with minimum P-value were selected from the statistical results of the data, showed in Table IV.