MicroRNA-34a Inhibits Osteoblast Differentiation and In Vivo Bone Formation of Human Stromal Stem Cells


  • Li Chen,

    Corresponding author
    1. Molecular Endocrinology Laboratory (KMEB), Odense University Hospital, University of South Denmark, Odense C, Denmark
    • Correspondence: Li Chen, Ph.D., KMEB, Odense University Hospital, Kloevervaenget 25, first floor, DK-5000 Odense C, Denmark. Telephone: +45-65504081; Fax: +45-66503920; e-mail: lchen@health.sdu.dk; or Moustapha Kassem, M.D., Ph.D., KMEB, Odense University Hospital, Kloevervaenget 25, first floor, DK-5000 Odense C, Denmark. Telephone: +45-65411606; Fax: +45-66503920; e-mail: mkassem@health.sdu.dk

    Search for more papers by this author
  • Kim HolmstrØm,

    1. Bioneer A/S, Kogle Allé 2, Hørsholm, Denmark
    Search for more papers by this author
  • Weimin Qiu,

    1. Molecular Endocrinology Laboratory (KMEB), Odense University Hospital, University of South Denmark, Odense C, Denmark
    Search for more papers by this author
  • Nicholas Ditzel,

    1. Molecular Endocrinology Laboratory (KMEB), Odense University Hospital, University of South Denmark, Odense C, Denmark
    Search for more papers by this author
  • Kaikai Shi,

    1. Molecular Endocrinology Laboratory (KMEB), Odense University Hospital, University of South Denmark, Odense C, Denmark
    Search for more papers by this author
  • Lea Hokland,

    1. Molecular Endocrinology Laboratory (KMEB), Odense University Hospital, University of South Denmark, Odense C, Denmark
    Search for more papers by this author
  • Moustapha Kassem

    Corresponding author
    1. Molecular Endocrinology Laboratory (KMEB), Odense University Hospital, University of South Denmark, Odense C, Denmark
    2. Danish Stem Cell Center (DanStem), Panum Institute, University of Copenhagen, Copenhagen, Denmark
    • Correspondence: Li Chen, Ph.D., KMEB, Odense University Hospital, Kloevervaenget 25, first floor, DK-5000 Odense C, Denmark. Telephone: +45-65504081; Fax: +45-66503920; e-mail: lchen@health.sdu.dk; or Moustapha Kassem, M.D., Ph.D., KMEB, Odense University Hospital, Kloevervaenget 25, first floor, DK-5000 Odense C, Denmark. Telephone: +45-65411606; Fax: +45-66503920; e-mail: mkassem@health.sdu.dk

    Search for more papers by this author


Osteoblast differentiation and bone formation (osteogenesis) are regulated by transcriptional and post-transcriptional mechanisms. Recently, microRNAs (miRNAs) were identified as novel key regulators of human stromal (skeletal, mesenchymal) stem cells (hMSC) differentiation. Here, we identified miRNA-34a (miR-34a) and its target protein networks as modulator of osteoblastic (OB) differentiation of hMSC. miRNA array profiling and further validation by quantitative RT-PCR revealed that miR-34a was upregulated during OB differentiation of hMSC, and in situ hybridization confirmed its OB expression in vivo. Overexpression of miR-34a inhibited early commitment and late OB differentiation of hMSC in vitro, whereas inhibition of miR-34a by anti-miR-34a enhanced these processes. Target prediction analysis and experimental validation confirmed Jagged1 (JAG1), a ligand for Notch 1, as a bona fide target of miR-34a. siRNA-mediated reduction of JAG1 expression inhibited OB differentiation. Moreover, a number of known cell cycle regulator and cell proliferation proteins, such as cyclin D1, cyclin-dependent kinase 4 and 6 (CDK4 and CDK6), E2F transcription factor three, and cell division cycle 25 homolog A were among miR-34a targets. Furthermore, in a preclinical model of in vivo bone formation, overexpression of miR-34a in hMSC reduced heterotopic bone formation by 60%, and conversely, in vivo bone formation was increased by 200% in miR-34a-deficient hMSC. miRNA-34a exhibited unique dual regulatory effects controlling both hMSC proliferation and OB differentiation. Tissue-specific inhibition of miR-34a might be a potential novel therapeutic strategy for enhancing in vivo bone formation. Stem Cells 2014;32:902–912


Decreased bone formation due to impaired osteoblastic (OB) cell proliferation, differentiation, and function is the main pathophysiological mechanism underlying osteoporosis; a disease characterized by low bone mass, deteriorated bone structure, and an increased risk for fracture [1]. In postnatal adult organisms, bone formation is mediated through recruitment of stem cells known as stromal (skeletal or mesenchymal) stem cells (MSC) that are located within the bone marrow near the bone formation surfaces. Thus, impaired bone formation is the result of inefficient proliferation or differentiation of MSC into OB cells [2]. Identifying factors regulating MSC proliferation and OB differentiation is an area of intensive investigation with potential for identifying novel targets to enhance bone formation for osteoporosis treatment.

Small noncoding single stranded microRNAs (miRNAs) have been identified as regulators of various physiological and pathological processes [3-6]. miRNAs regulate gene expression by promoting mRNA degradation and/or inhibiting RNA translation through binding to a specific binding site in the 3′ untranslated region (UTR) of the target mRNA [7-10]. Each miRNA can target hundreds of genes in a particular cell, and most mammalian mRNAs are conserved targets of miRNAs. Each target transcript contains a number of different miRNA binding sites resulting in a complex miRNA-gene target interaction network that regulates different biological processes within the cell [11].

An increasing number of miRNAs has been implicated as regulators of different aspects of bone development, osteoblast differentiation, and osteoporois pathophysiology [12]. Some studies have reported a number of miRNAs, for example, miR-133, miR-204/211 as inhibitors of osteoblast differentiation by targeting the osteoblast master transcriptional factor: Runx2 [13, 14]. Our group has identified miR-138 as a regulator of OB differentiation of human MSC through targeting PTK2, a gene coding for focal adhesion kinase and subsequently decreasing signaling through ERK pathway leading to impaired phosphorylation of Runx2 [15]. In addition, miR-204/211 and miR-637 have been suggested to function as “molecular switch” controlling the balance between MSC differentiation to OB versus adipocytes [14, 16]. Due to the complex biology of osteoblast differentiation and bone formation that require coordination of a large number of genes controlling cell proliferation, cell differentiation, bone matrix production as well as matrix mineralization, the number of miRNAs regulating these processes are expected to be large and until now has not been identified in its entirety.

In this study, we identified miR-34a as a regulator of OB differentiation and in vivo bone formation of hMSC. We demonstrated that inhibition of miR-34a by an anti-miR oligonucleotide markedly increased OB differentiation in vitro and enhanced heterotopic bone formation in vivo, whereas miR-34a overexpression reversed these effects. In addition, we identified Jagged one (JAG1), cyclin D1, cyclin-dependent kinase 4 and 6 (CDK4 and CDK6), E2F transcription factor three (E2F3), and cell division cycle 25 homolog A (CDC25A) as miR34a targets regulating both hMSC proliferation and OB differentiation.

Materials and Methods

Cell Culturing and Osteoblastic Differentiation

Both primary human MSC isolated from human bone marrow aspirate and hMSC-TERT cell line were used in the study. hMSC-TERT cell line (referred to in the text as hMSC) was created by overexpression of human telomerase reverse transcriptase (hTERT) gene and exhibits all characteristics of primary MSC in vitro and form normal heterotopic bone in vivo [17]. Cells were grown in normal medium (NM) containing minimal essential media without Phenol red and l-glutamine, supplemented with 10% (v/v) fetal bovine serum (FBS), 2 mM Glutamax, 100 units/mL penicillin, and 100 µg/mL streptomycin (Gibco-Invitrogen, Naerum, Denmark, http://www.lifetechnologies.com/dk/en/home/brands/invitrogen.html). In vitro OB differentiation of hMSC was carried out using OB induction medium (OIM) containing standard growth medium supplemented with 10−8 M dexamethasone, 0.2 mM l-ascorbic acid, 10 mM β-glycerophosphate, and 10 mM 1.25-vitamin-D3 (Sigma-Aldrich, Brøndby, Denmark, http://www.sigmaaldrich.com).

miRNA Microarray Analysis

hMSC cells were seeded as a monolayer in culture dishes or embedded into 3D-spheroid hydroxyapatite/tricalcium phosphate (HA/TCP) that promote a more synchronized OB differentiation [18]. Cells were cultured in NM or OIM for 14 days. Total RNA was harvested by Trizol reagent (Invitrogen) followed by DNase I treatment (Sigma-Aldrich) at day 0, 1, 3, 7, 14. RNA quality was checked by Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA, http://www.agilent.com). One microgram of the total RNA from each sample was labeled with Hy3 by the miRCURY Array Power labeling kit according to the manufacturer's instructions (Exiqon, Vedbaek, Denmark, http://www.exiqon.com) and hybridized to miRCURY LNA microRNA array (Exiqon, Vedbaek, Denmark, http://www.exiqon.com) representing version 10.0 of the Sanger miRBase. The hybridization and washes were performed in HS-400-Pro microarray hybridization station (Tecan, Grödig, Austria) according to the manufacturer's instructions. The dried slides were scanned in an ArrayWoRx white-light CCD-based scanner (Applied Precision, Issaquah, WA, http://www.api.com/) at a 10 µm resolution. The resulting images were imported into ImaGene 8.0 (BioDiscovery, El Segundo, CA, http://www.biodiscovery.com) where spot intensities and background measurements were calculated after flagging of bad quality spots. To identify differentially expressed miRNAs during the time course of osteoblast differentiation, spot intensity data of all relevant samples were analyzed using GeneSight-Lite 4.1.6 (BioDiscovery). Normalization of expression profiles was done by division of the mean signal of each array representing a single sample at the different time points. The resulting data were visualized using the Multiple Experiment Viewer [19] and hierarchical clustering using average linking clustering and a Pearson correlation distance metric. The resulting heat map and the intensity data were inspected for differentially expressed miRNAs across the time course corresponding to changes that occurred during OB differentiation.

Detection of miRNAs Using In Situ Hybridization

Archival anonymous paraffin-embedded human osteosarcoma tissue slides were obtained from Department of Pathology, Odense University Hospital. Cell pellets from the 14th day of OB induction cultured hMSCs with HA/TCP scaffold beads were fixed in 4% paraformaldehyde for 3 hours, followed by paraffin embedding. SuperFrost Plus slides were made of 4 µm paraffin sections. The expression of miR-34a was visualized using locked nucleic acid (LNA)-digoxigenin (DIG)-labeled probes (Exiqon, Vedbæk, Denmark). Briefly, sections were deparaffinized and treated with Proteinase K (5 µM, Sigma-Aldrich, Denmark, http://www.sigmaaldrich.com) for 40 minutes. The hybridization was performed at 60°C for 1 hour in a hybridization buffer (4 M urea, 5× saline sodium citrate stock solution (SSC), 1× Denhardt's) containing antisense DIG-labeled LNA-containing probes (Exiqon, Vedbæk, Denmark). Sections were blocked in a blocking buffer (0.1 M Tris pH 7.5 + 0.15 M NaCl + 10% FBS and 0.05% Tween 20) for 30 minutes and washed in PBS for three times. Chromogenic detection was performed by incubating sections with anti-DIG-POD (Roche Applied Science, Basel, Switzerland, http://www.roche-applied-science.com) diluted 1:400 in blocking buffer followed by washing in Phosphate buffered saline solution (PBS) 3× 5 minutes. Di-amino-benzidine reagent (Roche Applied Science) was applied for 1.5 hours, and the color reaction was stopped by washing in PBS. Sections were dried and mounted with Aqua-Mount (Lerner Laboratories, Pittsburgh, PA) reagent. Images were acquired using an Axio Imager Z1 microscope (Zeiss, Oberkochen, Germany, http://corporate.zeiss.com) using a Plan-APOCHROMAT 20×/0.8 objective under bright-field microscopy. U6 and scramble probes were used as positive and negative controls, respectively.

Alkaline Phosphatase and Alizarin Red S Staining

Alkaline phosphatase (ALP) cytochemical staining was performed on cultured cells by rinsing the cell layer with PBS and fixation in acetone/citrate (1.5:1, v:v) buffer (pH 4.2) for 5 minutes at room temperature. The cells were incubated with buffer containing 0.2 mg/mL naphthol AS-TR phosphate (Sigma-Aldrich). After incubation for 1 hour at 37°C, the cell layer was washed with deionized water. Alizarin red S histochemistry was used to assess the presence of in vitro formed mineralized matrix. The medium was removed, and the cell layer was rinsed with PBS and fixed in 70% ethanol for 1 hour at −20°C. The cell layer was washed with deionized water. The fixed cells were stained with 40 mM Alizarin red S pH 4.2 (Sigma-Aldrich) for 10 minutes at room temperature. The cell layer was washed with deionized water. For quantitative assessment of the degree of mineralization, the red color was eluted by 10% (w/v) cetylpyridinium chloride (Sigma-Aldrich) for 1 hour and quantified by spectrophotometric absorbance measurements at OD570 nm [20].

ALP Activity

ALP activity was performed in 96-well plate. First cell number (viable cells) was determined by adding the CellTiter-Blue solution (Promega, Madison, WI, http://www.promega.com) for 2 hours and estimated by measuring the fluorescence at wavelength of 579Ex/584Em. The cells were then rinsed with TBS (20 mM Trizma base, 150 mM NaCl at pH 7.5) and fixed in 3.7% formaldehyde-90% ethanol for 30 seconds at room temperature. A reaction mixture containing 100 µL 50 mM NaHCO3, 1 mM MgCl2 (Sigma-Aldrich), and 1 mg/mL of p-nitrophenyl phosphate (Sigma-Aldrich) was added into each well and incubated at 37°C for 20 minutes. The reaction was stopped by adding 50 µL of 3 M NaOH. Absorbance was determined at 405 nm in an ELISA microplate reader. ALP enzymatic activity was normalized to cell number.

Cell Transfection

The anti-miR miRNA inhibitor and pre-miR miRNA for miRNA-34a, nontargeting miRNA controls (pre-miR and anti-miR control), small interfering RNA (siRNA) duplex oligos targeting JAG1 mRNA, as well as nontargeting duplex oligo (siRNA negative controls) were purchased from Ambion (Austin, TX, http://www.ambion.com). These reagents were transfected into hMSC-TERT at a final concentration of 25 nM for miRNA, 1–12 nM for siRNA. For plasmid transfection, 2 µg DNA was used for each plasmid when cotransfected with miRNA samples into hMSC-TERT in six-well plate. All transfections were performed using Lipofectamine2000 Transfection Reagent (Invitrogen, http://www.lifetechnologies.com/dk/en/home/brands/invitrogen.html), according to the manufacturer's instructions.

Quantitative Real-Time Reverse Transcription Polymerase Chain Reaction

For gene expression, total RNA was extracted with TrizoL reagent (Invitrogen), cDNA was prepared using revert Aid H minus first strand cDNA synthesis kit (Fermentas, Slangerup, Denmark, http://www.thermoscientificbio.com/fermentas). Primers used are listed in Supporting Information Table S1. The PCR products were visualized in real-time using SYBR Green I Supermix (Bio-Rad, Hercules, CA, http://www.bio-rad.com) and an iCycle instrument (Bio-Rad) using standard curve protocols, normalized to β-2-microglobulin (B2m). The quantitative data presented is an average of duplicate or triplicate per independent experiment. For quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) of miRNA, miRNA expression was quantified using TaqMan microRNA assay (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) according to the manufacturer's protocol. Amplification and detection were performed using StepOnePlus Real Time PCR system (Applied Biosystems). The expression levels were normalized to RNU48, an internal control, and measured by comparative Ct (▵▵Ct) method.

Western Blot Analysis

hMSC-TERT cells were washed in PBS and lysed in RIPA buffer (Invitrogen) supplemented with protease inhibitors (Roche). After 1 hour incubation at 4°C, samples were centrifuged for 15 minutes at 12,000 rpm at 4°C. Protein concentration was determined with Pierce Coomassie Plus Bradford assay (Thermo Fisher Scientific, http://www.thermofisher.com/), and equal amounts of proteins were loaded on a 10% polyacrylamide gel (Invitrogen). Blotted nitrocellulose membranes were incubated overnight with antibodies against platelet-derived growth factor receptor alpha (PDGFRa) (Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com), Deleted in bladder cancer protein one (DBC1) (Cell Signaling), JAG1 (Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com), Cleaved Notch (NICD) (Cell Signaling), cyclin D1 (Santa Cruz Biotechnology), CDK4 (Santa Cruz Biotechnology), CKD6 (Santa Cruz Biotechnology), CDC25A (Cell Signaling), and anti-rabbit α-tubulin (Sigma), overnight at 4°C. Membranes were incubated with horseradish peroxidase (HRP) conjugated anti-mouse or anti-rabbit secondary antibody (Santa Cruz Biotechnology) for 45 minutes at room temperature, and protein bands were visualized with Amersham ECL chemiluminescence detection system (GE Healthcare Bio-Sciences Corp, http://www.gelifesciences.com). Each antibody used in Western blot analysis was diluted as 1:1,000 in PBS/T buffer with 5% milk.

Reporter Vectors, DNA Constructs, and Reporter Gene Assay

The 3′UTR of human CDC25A was customer ordered from CD BioSciences Inc., as plasmid 3′UTR-CDC25A-Luc-pEZX-MT01, with both firefly and Renilla luciferase codes. The fragment of 3′UTR containing the predicted binding site for hsa-miR-34a in JAG1 (NotI/NotI) was amplified from human genomic DNA using primers with a short extension containing cleavage sites for NotI (5′ end) and Not I (3′ end). Amplicons were cleaved and cloned into the PsiCHECK-2 vector (Promega, Denmark, http://dk.promega.com) downstream of the Renilla luciferase reporter gene. The mutations at the element of has-miR-34a target of the 3′UTRs (JAG1 and CDC25A) were constructed by proper primers by PCR with Phusion High Fidelity DNA polymerase (Thermo Scientific) according to the manufacturer's guidelines. All constructions were confirmed by sequence analysis. Primers are included in Supporting Information Table S1. All transient transfections were conducted using Lipofectamine 2000 (Invitrogen). For Notch reporter detection, 12× CSL-luc reporter plasmid, a kind gift from Prof. Serup (DanStem, Panum Institute, University of Copenhagen) was used. 5 × 105 hMSC-TERT cells were cotransfected in one six-well plate with 2 µg 12× CSL plasmid, 1 µg renilla luciferase control plasmid (Promega), and 100 µmol miRNA (or 50 µmol siRNA) by Nucleofector 2B electroporation (Lonza, http://www.lonza.com). Cells were harvested 48 hours after transfection in cell lysis buffer and were subsequently assayed for luciferase activity. Transfections were performed in duplicate, and all experiments were repeated several times. Firefly and Renilla luciferase were measured in cell lysates using a Dual-Luciferase Reporter Assay System (Promega) on LUMIstar high-performance microplate luminometer (BMG Labtech, Ortenberg, Germany, http://www.bmglabtech.com/). Reporter luciferase activity was normalized to proper internal controls for transfection efficiency. The experiments were carried out at least three times in triplicates.

Cell Cycle Analysis Using Flow Cytometry (FACS)

For cell cycle analysis, cells were transfected with miRNA-34a and anti-miR-34a, nontargeting miRNA controls (Pre-miR and anti-miR control) as described above and harvested at day 4. Cells were trypsinized and washed with PBS twice and then fixed with cold ethanol overnight. The cell cycle stages were analyzed by using the Cell Cycle Analysis Kit (GenScript, Piscataway, USA, http://www.genscript.com) as described in the manufacturer's protocol. Flow cytometry was performed on the CellLabQuanta SC Flow Cytometer (Beckman Coulter, Copenhagen, Denmark. https://www.beckmancoulter.com). Propidium iodide staining was detected by the phycoerythrin emission signal detector (Ex = 488 nm; Em = 620 nm). Beckman Coulter software was used to for analysis of cell cycle. A total of 20,000 events were acquired and the cells were properly gated for analyses. Three independent experiments were carried out for analysis.

In Vivo Heterotopic Bone Formation Assay

hMSC (5 × 105 cells per sample) were loaded on 40 mg wet hydroxyapatite(HA)/tricalcium phosphate based ceramics beads (HA/TCP) (TCP) (Zimmer; Warsaw, IN, USA, http://www.zimmer.com), incubated at 37°C overnight and implanted subcutaneously on the dorsal side of 8-week-old NOD.CB17-Prkdcscid/J mice, with non-obese diabetic (NOD) and severe combined immunodeficiency (SCID) as previously described [21, 22]. Implants were removed after 8 weeks and fixed in 4% paraformaldehyde (Bie & Berntsen A/S, Herlev, Denmark. https://dk.vwr.com/app), decalcified in formic acid, and embedded in paraffin. Sections (4 µm) were cut and stained with H&E Y (Bie & Berntsen A/S, Rødovre, Denmark). Bone volume per total volume was quantified as previously described [22].

Statistical Analysis

The data are presented as mean ± SEM of independent experiments. Statistical testing was performed using Student's t test to detect differences between groups. In cases of multiple groups testing, ANOVA was conducted followed by a posteriori t testing. Differences were considered statistically significant at *, p < .05; **, p < .001.


MiR-34a Is Expressed in hMSC and Upregulated During OB Differentiation of hMSC

To identify novel miRNAs that are associated with OB differentiation of hMSC, we carried out miRNA array profiling of hMSC following 14 days in vitro induction into OB in 3D-spheroid culture that promotes a more synchronized OB differentiation [18]. Expression levels of miRNAs were compared between OB differentiated and control undifferentiated samples at day 0, 3, 7, and 14. As presented in Figure 1A, miRNAs that had previously been reported to change during OB differentiation, such as Let-7, miR-29, miR-20, miR-125, miR-138, miR-148, miR-199 were detected (Fig. 1A). Their expression patterns coincided with the previous reports [12, 23]. miR-34a was identified among the most significantly upregulated miRNAs (Fig. 1A). miR-34b, another member of miR-34 family, exhibited modest regulation (Fig. 1A), while miR-34c was not detectable in the array profiling. The increased expression of miR34a during OB differentiation was observed in both 3D- and 2D-monolayer culture evidenced by qRT-PCR (Fig. 1B). Furthermore, in situ hybridization confirmed the expression of miR-34a after 14 days induction of OB differentiation in hMSCs in 3D HA/TCP cultures (Fig. 1C), as well as in vivo in mature OB cells (Fig. 1D).

Figure 1.

miR-34a expression pattern during osteoblast (OB) differentiation of human mesenchymal stem cell (hMSC) and its expression in human osteoblasts and bone tissue. (A): Heat map of significantly regulated miRNAs during OB differentiation at day 0, 1, 7, 14, 21 in 3D-HA/TCP scaffold revealed by miRNA array (n = 3). (B): Real time quantitative reverse transcriptase polymerase chain reaction of miR-34a expression during OB differentiation of hMSC-TERT in 3D-HA/TCP scaffold and monolayer culturing. (C): Detection of miR-34a expression by in situ hybridization in differentiated OB cells following 14 days OB induction of hMSC in 3D HA/TCP scaffold cultures (left) and human bone obtained from surgical osteosarcoma tissues (right). Scale bar = 50 µM. Abbreviations: HA/TCP, hydroxyapatite/tricalcium phosphate; miRNA, microRNA; NM, control normal medium; OIM, osteoblast induction medium.

MiR-34a Inhibits In Vitro Osteoblast Differentiation of hMSC

To evaluate the biological effect of miR-34a on OB differentiation, hMSC were transfected by miR-34a pre-miRs and anti-miRs oligos. The success of transfection was demonstrated by the presence of either high or low levels of miR-34a in transfected cells, respectively (Supporting Information Fig. S1). Following OB induction, we observed significant inhibition of ALP activity and staining (Fig. 2A), as well as inhibition of in vitro mineralization in cells overexpressing miR-34a. The opposite effects were observed in miR-34a deficient hMSC (Fig. 2B). Moreover, osteoblast marker genes expression: ALP, collagen type I (Col I), osteopontin (OPN), and osteonectin (ON) were inhibited by miR-34a overexpression, while their levels were enhanced in presence of anti-miR-34a (Fig. 2C). Similar effects on OB differentiation in primary bone marrow derived hMSC obtained from normal donors were observed upon inhibition of miR-34a (Supporting Information Fig. S2).

Figure 2.

The role of miR-34a and its inhibitor during osteoblast differentiation in human mesenchymal stem cell (hMSC). hMSC-TERT cells were cultured to 70%–80% confluence, then changed to osteoblast (OB) induction medium for 13 days. (A): On day 7, ALP staining and ALP activity measurement were determined, (B) on day 13, Alizarin red staining for mineralized matrix was performed and the eluted stain was quantified by spectrophotometer (OD570nm), (C) OB marker genes were measured by real time quantitative reverse transcriptase polymerase chain reaction at day 10. Results are presented as mean ± SD from at least three independent experiments. *, p < .05; **, p < .001. Abbreviations: ALP, alkaline phosphatase; Col (I), collagen type I; miRNA, micro RNA; OPN, osteopontin; ON, osteocalcin.

Molecular Targets of MiR-34a

To understand the molecular mechanisms underlying miR-34a-mediated regulation of OB differentiation, we searched for potential targets of miR-34a using the miRNA target prediction algorithms: TargetScan and PicTar. Platelet derived growth factor alpha (PDGFRα), Deleted in bladder cancer protein one (DBC1), and JAG1 were predicted as potential target genes by both programs. Those genes were highly upregulated during OB differentiation of hMSC as detected by qRT-PCR (Fig. 3A) and Western blot analysis (Fig. 3B). However, only JAG1, a Notch ligand, was clearly downregulated by miR-34a overexpression and upregulated by its inhibition (Fig. 3B). In silico analysis of human JAG1 gene 3′ UTR revealed 8-nt complementary site to miR-34a seed sequence (Fig. 3C). To test for direct binding of miR-34a to 3′ JAG1 gene UTR, we cotransfected the 3′UTR of JAG1 luciferase reporter plasmid and miR-34a in a dual luciferase system [24]. Pre-miR34a dose dependently blocked the luciferase activity of JAG1 3′UTR, while exerted no effects on the mutated version of JAG1 3′UTR (Fig. 3C), demonstrating the direct and specific binding to miR-34a binding sites in the 3′UTR of JAG1 mRNA. Moreover, gene expression levels of JAG1 were also decreased by overexpressing miR-34a in hMSC and increased in miR-34a deficient hMSC (Fig. 3D), indicating that JAG1 is regulated by miR-34a at both transcriptional and translational levels.

Figure 3.

JAG1 is a target of miR-34a in human mesenchymal stem cell (hMSC) during osteoblast differentiation. mRNAs' expression patterns of three potential targets of miR-34a: PDGFRα, DBC1, JAG1 during osteoblast (OB) differentiation of hMSC-TERT were measured by real time quantitative reverse transcriptase polymerase chain reaction (RT-PCR) (A). Expression pattern at protein levels during OB differentiation, and the regulation by miR-34a or anti-miR34a as detected by Western Blot analysis (B). Regulation of JAG1 levels by miR-34a was analyzed by dual luciferase assay in JAG1 3′UTR reporter system. Mutation of miRNA target sites attenuated the inhibitory effects of miR-34a (C). Transcriptional regulation of JAG1 mRNA by miR-34a determined by real time quantitative RT-PCR (D). Results are derived from at least three independent experiments. *, p < .05; **, p < .001. Abbreviations: DBC1, Deleted in bladder cancer protein one; JAG1, Jagged one; miRNA, micro RNA; PDGFRa, platelet derived growth factor alpha; UTR, untranslated region.

To detect whether JAG1 is a regulator in OB differentiation, we carried out siRNA-mediated knockdown of JAG1 gene in hMSC. As shown in Figure 4A, low levels of JAG1 did not change hMSC cell proliferation but inhibited OB differentiation evidenced by reduction in ALP activity (Fig. 4B), the ability to form mineralized matrix (Fig. 4B) and gene expression of OB marker genes: ALP, Col1, OPN, and ON (Fig. 4C). We also observed that the enhancement of ALP and in vitro mineralization by anti-miR-34a was reversed by siRNA-dose-dependent knockdown of JAG1 gene (Fig. 4D).

Figure 4.

The role of JAG1 during osteoblast (OB) differentiation in human mesenchymal stem cell (hMSC). (A): siRNA knockdown of JAG1 in hMSC-TERT did not affect cell proliferation. Results presented as cell viability ratio in siR-JAG1 and siR-Ctrl. (B): siRNA knockdown of JAG1 in hMSC reduced ALP activity (day 7) and in vitro mineralized matrix formation (day 13). (C): Expression of OB marker genes was measured by real time quantitative reverse transcriptase polymerase chain reaction in siR-JAG1and siR-Ctrl. (D): siRNA knocking-down of JAG1 dose dependently reduced the enhancement of mineralized matrix formation by anti-miR-34a (AZR staining at day 13) and ALP (staining at day 7) of OB differentiation of hMSCs. *, p < .05. Abbreviations: ALP, alkaline phosphatase; AZR, alizarin red; Col I, collagen type I; JAG1, Jagged one; OPN, osteopontin; ON, osteocalcin; siR-JAG1, cells transfected by JAG1 siRNA; siR-Ctrl, nontarget siRNA control; siRNA, small interfering RNA.

Since JAG1 is one of Notch ligands, we tested whether miR-34a regulation of JAG1 leads to changes in Notch signaling pathway. We used a Notch canonical pathway reporter system (12×CSL-luc) that revealed decreased Notch signaling following siRNA-based JAG1 knockdown or miR-34a overexpression in hMSC. Conversely, inhibition of miR-34a by anti-miR34a led to enhanced Notch signaling (Fig. 5A). In association with these changes, the cleaved form of Notch (NICD), which occurs upon ligand-induced activation of Notch, was decreased by siRNA-based JAG1 knockdown or by miR34a overexpression (Fig. 5B). Moreover, adding N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester (DAPT), a γ-secretase inhibitor and indirect inhibitor of Notch, inhibited OB differentiation as shown by the reduction of ALP staining, matrix mineralization, and ALP enzymatic activity (Fig. 5C).

Figure 5.

miR-34a targeting JAG1. (A): Activation of Notch reporter system (12×CSL-Luc reporter system) was detected in hMSC-TERT cells after cotransfected with siRNA or miR-34a mimics for 2 days. (B): Western blot detections of JAG1 and NICD after transfecting with JAG1 siRNAs (siR-JAG1) and miR-34a mimics in hMSC for 3 days. (C): Notch inhibitor (DAPT) inhibited osteoblast differentiation of hMSC in a dose-dependent manner evidence by changes in ALP activity (day 7), ALP staining (day 7) and mineralized matrix formation (AZR staining, day 13). *, p < .05; **, p < .001. Abbreviations: ALP, alkaline phosphatase; AZR, Alizarin red; DAPT, N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester; JAG1, Jagged one; miRNA, micro RNA; NICD, Notch intracellular domain; siR-JAG1, cells transfected by JAG1 siRNA; siR-Ctrl, nontarget siRNA control; siRNA, small interfering RNA.

Furthermore, we carried out JAG1-miR-34a target protection assay, where JAG1 was protected from miR-34a targeting (Supporting Information Fig. S4). We observed inhibition of OB differentiation as evidenced by ALP activity, ALP staining, and AZR staining for mineralized matrix in presence of JAG1 target protector but the efficiency of inhibition was more apparent in the absence of JAG1 target protector (Supporting Information Fig. S4). The results suggest that JAG1 is a direct target of miR-34a in OB differentiation, but there should be other factors or pathways also involved in the regulatory effects of miR-34a on OB differentiation.

During the initial phase of in vitro OB differentiation of hMSC (day 1–day 9), overexpression of miR-34a led to a decrease in cell number, while cell transfection with anti-miR-34a led to an increase in cell number (Fig. 6A) suggesting that miR-34a regulated cell proliferation. Consistent with this observation, we found that miR-34a targeted a number of cell cycle regulating genes including cyclin D1, CDK4, CDK6, E2F3, CDC25A. miR-34a overexpression inhibited the expressions of the proteins, while anti-miR-34a enhanced their expressions (Fig. 6B). Previous studies have demonstrated that cyclin D1, CDK4, CDK6, E2F3 are direct targets of miR-34a at their 3′UTRs [25-30]. We identified that CDC25A is also a target of miR-34a (Fig. 6B) as evidenced by targeting at its 3′UTR mRNA (Supporting Information Fig. S5). All these targetings at cell cycle factors resulted in an arrest of cell cycle at G1 and G2 phases (Fig. 6C) and decreased cell proliferation (Fig. 6A).

Figure 6.

miR-34a reduced cell proliferation and targeted cyclin D1, CDK6, CDK4, E2F3, and CDC25A. (A): Cell viability ratio was measured in cells transfected by miR-34a, anti-miR-34a and controls, at day 0, 2, 3, 5, 7, 9, 11, and 13 during the induction of osteoblast (OB) differentiation of human mesenchymal stem cell (hMSC). (B): Expression pattern of cyclin D1, CDK6, CDK4, and CDC25A during OB differentiation (day 0 to day 13) and their regulation by miR-34a and its inhibitor as detected by Western blot analysis. (C): Cell cycle analysis using flow cytometry. The ratio of cells at G1, S, and G2 phases in cells cycle was compared. Data presented from three independent experiments. *, p < .05. (D): A working model for the role of miR-34a in hMSC differentiation and bone formation. Abbreviations: CDK, cyclin-dependent kinase; CDC25A, Cell division cycle 25 homolog A; E2F3, E2F transcription factor three; JAG1, Jagged one; miRNA, micro RNA.

MiR-34a Inhibited Bone Formation In Vivo

To investigate the function of miR-34a in vivo, we used a preclinical model of heterotopic bone formation in mice. Control hMSC, hMSC transfected with either pre-miR-34a or anti-miR-34a were loaded on HA/TCP scaffold and implanted subcutaneously in NOD/SCID immune deficient mice for 8 weeks. The heterotopic bone formed was quantified by quantitative histology. As seen in Figure 7, hMSC formed normal lamellar bone and the amount of bone formed was significantly increased (more than 3.5-fold) in implants containing cells transfected with anti-miR-34a compared with the anti-miR control (Fig. 7A, 7B). Furthermore, overexpression of miR-34a decreased bone formation by twofold compared with pre-miRNA controls (Fig. 7A, 7B).

Figure 7.

miR-34a regulates heterotopic bone formation in vivo. hMSC-TERT were transfected with miR-34a, its inhibitor or controls and implanted coupled to hydroxyapatite/tricalcium phosphate (HA/TCP) scaffold subcutaneously into immune deficient NOD/SCID mice. H&E staining was performed after 8 weeks of implantation to quantitate newly formed bone (A). Bone tissue (red); HA/TCP scaffold (brown). Scale bar = 250 µM. (B): Bone formation was expressed as bone area divided by total area (n = 7 per treatment). *, p < .05; **, p < .001. Abbreviation: miRNA, micro RNA;


In this study, we identified that miR-34a is a highly expressed miRNA during the course of in vitro OB differentiation as well as in vivo in mature osteoblastic cells. Furthermore, we demonstrated that miR34a is a negative regulator of OB differentiation since its silencing led to enhanced in vitro OB differentiation and overexpression led to the opposite effects. Finally, we demonstrated that targeting miR-34a in hMSC increased in vivo bone formation in a preclinical model of heterotopic bone formation in mice.

An increasing number of studies have identified several miRNAs that play an important regulatory role in bone biology [12]. The miR-34 family has been reported previously as a regulator of murine OB functions [30]. Interestingly, in this article the presence of miR-34 b/c, but not miR-34a, was detected in murine OB cells, which is at variance with our study in human MSC where we detected miR-34a, but not miR-34b/c, expressed at a high level during OB differentiation (Supporting Information Fig. S3). miR-34b/c are located in chromosome 11q23, while miR-34a is located in the region of chromosome 1p36 and has its own transcript [31], which regulated through both p53-depedent [31-34] and p53-indepdent [35, 36] mechanisms. miR-34b/c are mainly expressed in lung tissue but scared in other tissues, while miR-34a is more expressed in many tissues, including bone marrow and bone [31]. Thus, even though all members of miR-34 family act as negative regulators of bone formation, miR-34a might be more relevant and specific for normal human bone physiology.

We observed that miR-34a overexpression reduced hMSCs proliferation, while silencing miR-34a increased the cell proliferation during OB differentiation. Both in silico target identification and previously published experimental data have demonstrated that the miR-34 family targets several cell cycle control factors: cyclinD1, CDK4 and CDK6 [25, 26, 29, 30] as well as E2F3 [27, 28], which can explain the negative effects on cell proliferation. Here, we observed that miR-34a regulates the expression of these genes in hMSC, too. Moreover, we also identified that miR-34a regulates an additional cell cycle gene, CDC25A, both by Western blot and in silico targeting analysis (Fig. 6B; Supporting Information Fig. S4). Cyclin D1, CDK4, CDK6, and E2F3 are factors regulating G1 phase of cell cycle, and CDC25A is a regulator of both G1 and G2 phases. Our findings suggest that miR-34a arrested hMSC cell cycle at G1 and G2 phases can be explained by its targeting of these cell cycle genes. Our current findings and findings from previous studies suggest a general role of miR-34a in regulation of cell proliferation and are consistent with the reported data that miR-34a may function as a tumor suppressor and as a part of the tumor suppressor of p53 gene network targets [32, 37].

We identified JAG1 as a target of miR-34a during OB differentiation. miR-34a downregulated JAG1 expression at both transcriptional and translational levels. Knocking down JAG1 by RNA interference blocked OB differentiation. JAG1 has been suggested to play an important role in normal human skeletal biology evidenced by human disease Alagille syndrome, a disease presenting with skeletal abnormalities and caused by microdeletion within JAG1 gene [38]. Also, in a genome-wide association study a polymorphic trait rs2273061 of the JAG1 gene has been associated with high bone mineral density and low risk of osteoporotic fracture [39].

Human JAG1 is a Notch one receptor ligand and its interaction with Notch one leads to release of the Notch intracellular domain (NICD), allowing it to translocate into the nucleus and activate Notch-responsive genes that are important for cell differentiation and morphogenesis in different biological systems [40-42]. There were several studies reported miR-34 family target Notch1 and Notch2 leading to inhibition of Notch signaling in different cells [43-45]. But we did not detect clear changes in Notch expressions by miR-34a in hMSC (data not shown). However, inhibiting JAG1 in hMSC leads to inhibition of Notch signaling, suggesting that JAG1 is one of the principal ligands of Notch in human OB cells. The relevance of Notch signaling for OB differentiation of hMSC was also demonstrated in our study by the inhibitory effects of DAPT, a known inhibitor of Notch signaling, on ALP activity and the mineralization ability of OB differentiated hMSC. Thus, miR-34a provides an example of a miRNA with a dual regulatory nature since miR-34a regulated both hMSC proliferation and differentiation, as seen in our working model (Fig. 6D). miR-34a regulated cell proliferation through targeting several cell cycle genes leading to cell cycle arrest and inhibited the OB differentiation through the targeting of Notch signaling.

There is an increasing interest in developing novel pharmacological agents for disease treatment based on targeting miRNA and some of the current miRNA inhibitors are undergoing clinical trials [46]. Previous studies in mice have shown that targeting miRNA led to the development of osteoporotic bone in mice [47] and contributed to primary osteoporosis in humans [48]. We observed that inhibiting miR-34a in hMSC enhanced the bone forming capacity in a preclinical model of heterotopic bone formation. Thus, inhibition of miR-34a in MSC in vivo may lead to enhanced bone formation. It is plausible that local implantation of hMSC deficient in miR-34a or hMSC cultured on a functionalized scaffold containing anti-miR-34a [49] will be an approach to enhance local bone formation needed for treatment of localized bone defects and non-healed fractures. These novel approaches for enhancing bone formation and bone mass need further investigation.

Summary and Conclusion

We identified miR-34a as a dual-effector miRNA with regulatory effects on hMSC differentiation and proliferation. miR-34a targeted JAG1 and downregulated Notch signaling, a mechanism responsible for the observed inhibitory effects of miR-34a on OB differentiation. miR-34a also inhibited cell proliferation through the targeting of a number of cell cycle genes: cyclin D, CDK4, CDK6, E3F3a, and CDC25A resulting in cell cycle arrest in G1 and G2 phases. In a preclinical model of ectopic bone formation, inhibition of miR-34a led to increased bone formed by hMSC. Pharmacological targeting of skeletal miR-34a expression represents a novel possible approach to enhance local bone formation.


This study was supported by the Novo Nordisk Foundation (L.C.), the A. P. Møller Foundation (L.C.), ECTS Postdoctoral Fellowship 2012 (L.C.), Danish Ministry of Science, Innovation and Higher Education (innovation consortium), and a grant from the local government of Southern Denmark (M.K.). We thank Dr. Charles Edward Frary for proof reading the manuscript.

Author Contributions

L.C.: conception and design, collection and/or assembly data, data analysis and interpretation, manuscript writing, and final approval of manuscript; K.H.: collection and/or assembly data, data analysis and interpretation, and manuscript writing; W.Q. and N.D.: collection and/or assembly data and data analysis and interpretation; K.S. and L.H.: collection and/or assembly data; M.K.: conception and design, manuscript writing, and final approval of manuscript.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.