Calvarial osteoblasts were harvested from cell culture obtained from 1-2 day old neonatal mice pups in accordance with standard protocols (16). Calvariae were surgically excised and then exposed to five successive digestions in 0.1% dispase (Sigma, St. Louis, MO, USA) and 0.1% collagenase P (Sigma, St. Louis, MO, USA) solution at 37°C. Cells released from the second to fifth digestions were collected, centrifuged, resuspended and plated in T-25 cm² flasks in osteoblast growth medium containing α-MEM media (Sigma, St. Louis, MO, USA), 10% fetal bovine serum (FBS, Gibco, Life technologies, USA) and 1% penicillin/streptomycin (Sigma, St. Louis, MO, USA).

MCOs were seeded 1 day prior to the transfection in cell culture plates in osteoblast growth medium. Cells at 60-70% confluence were transfected with oligo miRNAs like miC (negative control), miR-539-3p (mimic) and anti-miR-539-3p (inhibitor) and siRNAs purchased from Thermo Fisher Scientific, USA by using lipofectamine 2000 (Thermo Fisher Scientific, USA) transfection reagent in Opti-MEM^®I reduced serum (Gibco, Life technologies, USA) medium as per manufacturer’s protocol. Sequences are given in Supplementary Table 1 . After 6 h, the medium was replaced with osteoblast differentiation medium. Further analysis was done by harvesting cells at different time intervals after transfection.

ALP activity was measured by trysinization of MCO at 70-80% confluence followed by seeding in 96 well (3000 cells/well) plates in osteoblast growth medium. Cells were then transfected with miR-539-3p, anti-miR-539-3p and negative control (miC) for 6h and changed with differentiation medium containing 10 mM beta-glycerophosphate (Sigma, St. Louis, MO, USA) and 50 μg/ml ascorbic acid (Sigma, St. Louis, MO, USA) for 48h (16, 17). Total ALP activity was evaluated by measuring absorbance at 405 nm using p-nitrophenylphosphate as a substrate (PNPP, Sigma, St. Louis, MO, USA).

For mineralization studies, murine osteoblasts were cultured according to our previously published lab protocol (18, 19). Cells were seeded in α‐MEM medium, supplemented with 10% fetal bovine serum, 50 μg/ml ascorbic acid, and 10 mM β-glycerophosphate) and transfected with miC, miR-539-3p and anti-miR-539-3p. Cells were cultured for 14 days at 37°C in a humidified CO₂ incubator and media was changed every 48 h. For ex-vivo after mineralization femoral bone marrows of autopsied Balb/c female mice of different groups were cultured in α‐MEM medium, supplemented with 10% fetal bovine serum, 10^-7 M dexamethasone, 50 μg/ml ascorbic acid, and 10 mM b-glycerophosphate) and seeded in six‐well plate with a density 2 x 10⁶ cells/well and cultured for 21 days (20). After completion of time, the cells were fixed with 4% para formaldehyde (PFA) for 30 min at room temperature and stained with 40 mM alizarin red S, which stains nascent calcium and quantified with 10% Cetylpyridinium chloride (CPC) solution. The absorbance was taken at 595 nm on the next day (21).

Quantitative RT-PCR was employed to examine the mRNA expression levels. mirVana miRNA isolation kit was used for miR-539-3p quantification along with TaqMan miRNA reverse transcription kit (Thermo Fisher Scientific, USA) and a TaqMan miRNA assay kit (Thermo Fisher Scientific, USA). Total RNA was isolated from the cultured cells as per manufacturer’s instructions (mirVana miRNA isolation kit (Ambion, Carlsbad, CA, USA). TaqMan miRNA reverse transcription kit was used to create cDNA. 100 ng of total RNA was added to the reaction mixture, which included 10 RT buffer, Multi Scribe reverse transcriptase, 5 RT primers, 100 mM dNTPs, RNase inhibitor and nuclease-free water. Total RNA was isolated from the confluent cultured cells utilizing TRIzol (Life Technologies) for mRNA, while 2 μg of total RNA was used to synthesize cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems™, Thermo Fisher Scientific, USA) in accordance with the manufacturer’s instructions. Expression analysis of several genes and miRNA by qRT PCR was carried out in accordance with previously reported protocol (22). Supplementary Table 1 depicts the list of sequences of sense and antisense oligonucleotide primers. For triplicate reactions, the mean Ct (cycle threshold) results of each sample were calculated. The relative expression levels of miRNA and mRNAs were calculated using log2|2—DCt|, where DCt was measured by subtracting the Ct value of the target miRNA and mRNA from the Ct value of the internal control U6 (Cat no. 4427975) and GAPDH.

The possible target genes for miR-539-3p were identified by employing two different target prediction tools, i.e., Target Scan and mirdb. 200 ng of empty pEZX-MT06 control clone (Genecopoeia Medical Center Dr. Suite 101 Rockville, MD 20850 USA) and 3′ UTR of LRP-6 clone were transfected for 6 h in Opti-MEM^®I reduced serum medium with lipofectamine 2000 reagent. Cells were co-transfected with miR-539-3p or miC at concentrations of 50 nM. Renilla luciferase activity was used for normalisation and as an internal control for transfection efficiency. A dual-luciferase Reporter Assay System (Promega, Madison, WI 53711 USA) was used on a FLUOstar galaxy to quantify firefly and renilla luciferase in cell lysates. The activity of Renilla luciferase was utilised as a normalizer and an internal check for transfection efficiency.

Confluent MCOs were transfected with or without miR-539-3p and total RNA was isolated. The integrity and quality of the isolated RNAs were tested by Agilent 2100 bio analyzer and the samples with RNA integrity score >7 were used for library preparation. NEBNext poly (A) mRNA magnetic isolation module (E7490S) was used for enrichment of mRNA using oligodT based beads. 100 ng of RNA was taken as input for the library preparation. RNA libraries were prepared as instructed in the manual of NEBNext Ultra II RNA library prep kit (E7775). Sequencing was performed on Illumina Hiseq 2500 next generation sequencing platform.

For data analysis, RNA-seq data were mapped to the mouse genome constructs using TopHat (v2.0.8b, http://tophat.cbcb.umd.edu/) (UCSC). The low-quality reads were eliminated from the raw sequencing data. Tophat (R software) was used to map the reads, and HTSeq (http://www-huber.embl.de/users/anders/HTSeq/doc/overview.html) was used to count the reads. The DESeq R software package was used to examine differentially expressed genes. The differentially expressed genes were discovered using Benjamini-Hochberg multiple testing adjustments.

For transient shRNA transductions, 3T3 cells were seeded in a 6 well plate at a density of 0.2×10⁶ and grown overnight in a CO₂ incubator. The cells were then transduced with concentrated lentiviral particles diluted in fresh Dulbecco’s Modified Eagle Medium-high glucose (DMEM-HG), using cationic polymer polybrene at a final concentration of 10μg/ml. At 72h post-transduction, cells were harvested and processed for immunoblotting to check the knockdown efficacy of the target gene.

For MCO study, cells treated with concentrated lentiviral particles with polybrene for 24h. After 24h, medium was changed with osteoblast differentiation medium. At 48h post-transduction, cells were harvested and processed for further analysis. Details of lentiviral vector and production of lentiviral particles has been given in supplementary section.

Immunoblotting was done to check the expression levels of osteogenic proteins as well components that are involved in the Wnt/beta catenin pathway. For this purpose, cultured MCO cells were transfected with different oligo miRNAs and si-RNAs and incubated for 48h in the differentiation medium. Whole cell lysate was extracted with mammalian cell lysis buffer comprising protease inhibitor cocktail and phosphatase inhibitor (Sigma, St. Louis, MO, USA). Protein concentration was estimated by BCA assay, and then separated on different percentage of SDS-PAGE, which were then electroblotted onto PVDF membranes (Immobilon-P, Millipore, Billerica, MA, USA). The membrane was then probed with Runx-2, Type I collagen, Akap-3, Wnt-3a, LRP-6, β-Catenin, P-β-catenin, Lef-1 and β-actin antibodies as primary antibodies and incubated with secondary antibodies conjugated with Horseradish Peroxide (HRP) at 4°C overnight. Antibody details were given in Supplementary Table 2 . On the Image Quant LAS 4000 (GEHealthcare) gel doc imaging system, Immunodetection was performed with an enhanced chemiluminescence kit (Immobilon-P, Millipore, Billerica, MA, USA). ImageJ software was used for the quantification of blots.

In vitro Osteoclastogenesis was performed as per standard protocol (19). In brief, bone marrow cells (BMCs) collected from the femurs and tibias of 8 to 9-week-old female BALB/c mice were allowed to proliferate by incubating in a T-25 cm² flask using complete cell culture mix including α-MEM, 10% FCS, 1% penicillin/streptomycin, 10 ng/ml MCSF. Following overnight incubation, the non-adherent BMCs were plated at a density of 2 x 10⁵ cells/well in 48-well plates and then cultured in osteoclast medium containing 50ng/ml RANKL and 30ng/ml MCSF for 6 days by replacing the media after every 48h. The cells were then washed with PBS and fixed in 4% paraformaldehyde for 20 min. After that, tartrate-resistant acid phosphatase cells were stained according to the standard procedure (23). For transfection, at 60–70% confluence, mouse osteoclast cells were transfected with miC and miR-539-3p at a 50 nM concentration using RNAi Max (Invitrogen). After 48h of transfection, cells were isolated in order to measure the expressions of TRAP and RANK by employing real-time PCR and stained for TRAP positive cells (24).

All animal care and experimental methodologies have been approved in advance by the Institutional Animal Ethical Committee (IAEC) at the CSIR-Central Drug Research Institute in Lucknow, India. Animal experiments were carried out in accordance with the guidelines provided by the Committee for the Control and Supervision of Experiments on Animals of the Indian Government (CPCSEA). This study’s animal research and laboratory techniques have been approved by the IAEC (approval no. CDRI/IAEC/2019/49/Dated-04/01/2019). In all the experiments, six-week-old female Balb/c mice were utilized. Animals were ovariectomized with general anesthesia employing dorsal approach and left for 1 month to develop osteopenic condition. Animals were randomly distributed in four groups, one with Sham+PBS (ovary intact), Ovx+miC, Ovx+miR-539-3p and Ovx+anti-miR-539-3p. The three ovariectomized groups were subcutaneously injected (one injection/week) for three weeks with miC, miR-539-3p and anti-miR-539-3p respectively. 100 μl of oligonucleotide mixture containing 5 μg of oligonucleotide with N/P ratio of 8, i.e., 0.16 μl of in vivo-jetPEI^® (Cat. No.201- 10G, polyplus-transfection, Illkirch, France) per ug of oligonucleotide was given to each mouse via subcutaneous injection behind the head region. On the fourth week, the mice were sacrificed and femur bones as well as serum samples were collected for analysis of trabecular microarchitecture using μCT and other biochemical parameters (16).

The Sky Scan 1076 CT scanner (Aartselaar, Antwerp, Belgium) was used to perform micro-computed tomography of the femur bones as previously described (17). Briefly, the trabecular regions of bone samples were scanned with an X-ray source of 70 kV and 100 mA at a nominal resolution (pixels) of 18 mm. The image slices captured with 100 projections at 180° angular range were then reconstructed by utilizing the Sky Scan Nrecon software, employing a modified Feldkamp method that aids in the reconstruction of a distributed network. Several trabecular bone parameters were calculated.

Muscles were removed from collected bones from various groups, and bone marrow was flushed. Femur and tibia that had been cleaned were combined and pulverized in liquid N2. Total RNA was extracted using the frozen powder that had been placed into a tube containing Trizol followed by collected and qPCR analysis as described above. Supplementary Table 1 represents the list of primer sequences.

In order to assess the biochemical parameters, blood serum was isolated from all the animal groups, and centrifuged at 4,000 rpm for 30 min. Serum levels of amino-terminal propeptide of type 1 procollagen (serum P1NP, Elabscience Biotechnology Inc., Donghu Hi-Tec Development area, Wuhan, Hubei,China), C-terminal telopeptide of type I collagen (serum CTx, Elabscience,Wuhan,China) and TRAP-5b (immunodiagnostic systems Holdings Ltd. UK) were determined by enzyme-linked immunosorbent assay (ELISA) kits following the manufacturer’s protocols.

The forelimb was disarticulated through the knee joint and then decalcified using a decalcifying buffer (decalcifying solution lite, Sigma, USA) for 72 h. Thereafter, the femur’s epiphyseal region was treated with 70% isopropanol before being preserved in paraffin for sectioning. The representative pictures were taken after the staining of transverse sections (5 μm) with hematoxylin and eosin (Sigma-Aldrich (St. Louis, MO) on EvosFL Auto (Life technologies) microscope. Mature osteoblasts were visible as a single layer of cuboidal or polygonal cells with eccentrically located nuclei on bone surface. The osteoblasts number per bone surface below the growth plate was calculated by tracing the section image onto a digitizing platform with the aid of a cameral lucida attachment and bioquant analysis software according to standard protocols (25).

The data is presented as mean ± S.E.M. One-way ANOVA was used to analyze the data acquired from experiments with multiple treatments followed by multiple comparison Turkey test of significance using Graph Pad Prism version 7.04 software. In trials with only two treatment conditions, the two-tailed t-test was utilized to investigate statistical significance. All experiments were replicated thrice and representative experiments as mentioned.

Analysis of ALP activity, a marked osteoblast differentiation marker, after transfection of mouse calvarial osteoblasts with different concentrations of phex silencer was performed in order to examine the role of phex gene in osteoblast differentiation. The activity was found to be significantly inhibited at 30 nM concentration.

( Figure 1A ) which implied that silencing of Phex gene resulted in suppression of osteoblast differentiation. This RNAi mediated silencing of phex as well as over-expression of osteopontin (OPN) in phex treated osteoblast cells were further confirmed by immunoblotting ( Figures 1B–D ). Moreover, RT-PCR data revealed a remarkably downregulated expression levels of osteogenic markers in phex silenced osteoblasts in comparison to non-target control (NTC, scrambled siRNA) treated cells ( Figures 1E, F ). This overall suggested that phex gene has an essential importance in contributing towards regulation of osteoblastogenesis.

As silencing of Phex resulted in suppressed osteoblast differentiation, calvaria-derived mouse osteoblast cells were subjected to next generation microRNA analysis post transfection with or without siRNA of phex gene. MicroRNA analysis revealed that phex silencing resulted in up regulation of several miRNAs, while; only a few were found to be downregulated. Among them, miR-539-3p, was shown to exhibit 8.5-fold increase in expression during phex-silenced osteoblast development ( Figure 2A ). Though, miR-539-3p is often recognized as a tumor suppressor, it has never been investigated for its functional role in osteoblast differentiation. Quantitative real-time PCR (qRT-PCR) in murine osteoblast cells was used to further confirm the miRNA data analysis which showed ~5-fold upregulated expression of miR-539-3p in siPhex treated osteoblast cells ( Figure 2B ), whose expression was found to be decreased during the course of osteoblast differentiation ( Figure 2C ).

The effect of miR-539-3p on osteoblast differentiation was evaluated, by transfecting osteoblast cells, with 30 nM of miR-C, 50 nM of mimic miR-539-3p and 30 nM of inhibitor anti-miR-539-3p, in growth medium with 10 mM b-glycerophosphate and 50μg/ml ascorbic acid followed by measurement of ALP activity after 48 h. miR-539-3p-transfected cells demonstrated significantly diminished ALP activity when compared to miR-C transfected cells, while, in anti-miR-539-3p-transfected cells, this effect was found to be reduced ( Figure 2D ). Following this, the effect of miR-539-3p was seen on mineral nodule formation. As shown in Figures 2E, F transfection of miR-539-3p decreased mineralized nodule formation in 14-day culture. However, anti-miR-539-3p treatment significantly increased the mineral nodule formation. In addition to this, inhibition in expression of certain osteogenic gene markers like ALP, RUNX-2, Type I col, and OCN was found to be mediated by mimic miR-539-3p ( Figure 2G ). Contrary to this, increased expressions of these markers were found in anti-miR-539-3p-transfected cells. These results thus suggested that osteoblast differentiation is negatively regulated by miR-539-3p.

Further investigation to understand the possible mechanism by which miR‐539‐3p affect osteoblast differentiation was done by employing target prediction programs such as Targetscan (http://www.targetscan.org/cgi‐bin/search.cgii) and mirdb (http://www.mirdb.org/cgi‐bin/search.cgi). This revealed several target genes. Amongst these, LRP‐6 was chosen due to its important role in Wnt signaling. Besides, it possessed two sequence similarity sites, having 7‐mer A1 and 8-mer associations with miR‐539‐3p in the seed sequence and conserved sequence in different species ( Figures 3A, B ). In order to confirm that LRP-6 is the specific target of miR-539-3p, mRNA levels of LRP-6 was assessed in murine osteoblast cells transfected with miC or mimic or inhibitor. LRP-6 mRNA expression was shown to be dramatically reduced in cells transfected with miR-539-3p whereas, the same was found to be reversed in anti-miR-539-3p-transfected osteoblasts ( Figure 3C ). Further, to confirm the specificity of target, two luciferase reporter constructs were prepared, one without 3′‐UTR (control/empty clone) and other including 3′‐UTR LRP‐6. In osteoblast cells, these clones were co-transfected with miR-539-3p or miC, and luciferase activity was monitored. Co-transfection of miR‐539‐3p with 3′‐UTR of LRP-6 suppressed luciferase activity of reporter gene, However, the activity was found to be unaffected upon co-transfection of miR‐539-3p with empty vector/control clone (lacking 3′‐UTR), thereby confirming the target specificity ( Figures 3D, E ).

To investigate the impact of miR539-3p on Wnt signalling elements, miR‐539-3p was over-expressed in osteoblast cells. In order to examine its effect on Wnt signaling components as it directly targets LRP-6 (a known activator of Wnt-β catenin pathway). Cells were also transfected with miC (negative control) and inhibitor (anti‐miR‐539‐3p). Antibodies specific for the Wnt signalling pathway were used to probe the protein lysate obtained from transfected cells. Protein expression of Wnt3a, LRP-6, beta catenin, Lef-1 and RunX-2 were downregulated, as revealed by western blot analysis in miR-539-3p treated cells as compared to miC. Contrary to this, opposite result was obtained in cells transfected with anti-miR-539-3p ( Figures 4A–C, E, G, H ). Lesser effect on the expression of GSK3β was found ( Figure 4D ). Importantly, Phospho-beta catenin level was also found to be upregulated in miR-539-3p treated cells which was decreased in cells transfected with the inhibitor ( Figure 4F ). These results clearly showed that miR‐539‐3p inhibits osteoblast differentiation by suppressing the Wnt/beta catenin signaling.

After confirming the negative regulation of osteoblastogenesis by miR-539-3p, we have determined its direct role in osteoclastogenesis. Figure 5A depicts the representative image of TRAP positive cells in BMC cultures treated with or without miR-539-3p. The quantitative analysis indicated that ~34% TRAP positive cells were present in miR-539-3p treated cells in comparison to miC negative control, which was ~16% ( Figure 5B ). Moreover, miR-539-3p elevated the mRNA expression levels of RANK (P<0.05) and TRAP (P<0.05) as compared to miC negative control ( Figure 5C ). These data clearly indicated that miR-539-3p directly affects osteoclastogenesis by stimulating it.

To further investigate the functional role of miR-539-3p, in vivo studies were conducted which involved the use of chemically modified sense and antisense oligonucleotides specific to miR-539-3p. For this, animals were ovariectomized to instigate osteopenic condition. Figure 6A depicts the overall study plan. Subcutaneous administration of mice with readymade miRNA of miC, miR-539-3p and anti-miR-539-3p at a dose of 5μg/animal followed by treatment with oligonucleotides for three weeks was done, wherein; Sham+PBS (0.1 ml) served as controls. Micro-CT analysis of euthanized animals on the fourth week was performed to quantify the BV/TV (bone volume/tissue volume ratio), (Tb.N) trabecular number, (Tb.Sp) trabecular separation, (Tb.Th) trabecular thickness, (Tb.Pf) trabecular bone pattern factor and BS/TV (Bone surface/tissue volume). The representative micro-CT images are shown in ( Figure 6B ). Quantification of micro-CT data revealed that miR-539-3p inhibitor-treated Ovx mice showed a significant increase in bone parameters, including BV/TV, BS/TV, Tb.N and Tb.Th ( Figures 6C–E ), with a concomitant decrease in Tb.Pf and Tb.Sp in femora bones. However, reverse effects were observed in miR-539-3p treated Ovx mice ( Figures 6F–H ).

We also did ex-vivo mineral nodule formation efficacy in various groups. Less mineral nodule formation was seen in Ovx+miC (p < 0.01) group as compare to Sham+PBS. However, miR-539-3p treatment further decreased nodule formation (p < 0.001) in comparison to Sham+PBS. Opposite results were observed in anti-miR-539-3p treated group (p < 0.001) ( Figures 7A, B ). Further, osteogenic markers expressions were also assessed in bone samples of different groups. Transcript levels of T-1-col, RunX-2 and OCN were significantly decreased in miR-539-3p group. Opposite effect was found in anti-miR-539-3p group ( Supplementary Figure 2 ).

Serum P1NP and CTx are considered as crucial bone formation as well as bone resorption markers, respectively. Hence, the ability of miR-539-3p to enhance bone resorption and suppress bone formation was analyzed. Consequently, a remarkable dip in P1NP serum levels was noticed in mice treated with miR-539-3p. This effect was totally reversed in mice treated with anti-miR-539-3p ( Figure 7C ). On the other hand, CTx levels were much higher in mice treated with miR-539-3p which returned back to the sham levels in anti-miR-539-3p treated animals ( Figure 7D ). Further, effect of miR-539-3p on osteoclastogenesis was also checked by evaluating serum levels of Trap-5b. Trap-5b is a well-known osteoclast marker and was found to be increased in miR-539-3p treated group. Its levels were decreased in anti-miR-539-3p treated animals ( Figure 7E ).

Furthermore, histological analysis performed by H&E staining technique exhibited scanty and thinning trabeculae, deprived of connectivity in miR-539-3p treated femur bones, whereas treatment with anti-miR-539-3p considerably guarded the trabecular bone and manifested eminent connectivity ( Figure 7F ). Moreover, the osteoblasts number per trabecular bone surface were found to be less in miR-539-3p treated ovx mice while in anti-miR-539-3p administered group osteoblasts number per bone surface were comparable to sham ( Figure 7G ).

Taken together, miR‐539‐3p regulates bone metabolism and its inhibitor furnished defense against bone loss in ovariectomized mice that are estrogen depleted and susceptible to bone loss.

Once, the role of miR-539-3p was ascertained in osteoblasts, we went on to study a transcriptome signature of MCO transfected with or without miR-539-3p. Based on the fold change (log2fold), depicting differences between miR-539-3p and control samples, a heat map of most downregulated as well upregulated genes in miR-539-3p conditions was generated. Genes with the lowest log2fold was denoted by blue color and largest log2fold with yellow color, respectively, with significant P value ( Figure 8A ).

Based on the analysis, 5 most upregulated and 10 downregulated targets were selected and their mRNA expressions were checked by qPCR for validation of the transcriptome profiling data (data not shown). Among the downregulated targets, Akap-3 expression was decreased in miR-539-3p treated cells as compared to control ( Figure 8B ). Further, protein expression of Akap-3 as revealed by western blot analysis was downregulated in mimic condition and restored in inhibitor treated cells ( Figures 8C, D ). Thus, downregulation of Akap-3 in miR-539-3p treated osteoblast cells might be indicative of its potential role in osteoblast functions.

Akap-3 gene was knocked down using lentivirus shRNA, in 3T3 cells to investigate its role in osteoblast differentiation. Different shRNA clones of Akap-3 in 3T3 cells were confirmed, out of which Clone 6 and clone 7 were found to be downregulated as compared to scrambled-shRNA control ( Figures 9A, C ). These clones were then transfected in calvarial derived osteoblast cells. Clone 6 showed significant down regulated expression confirming the gene knock down as compared to scr-shRNA control ( Figures 9B, D ). Furthermore, Akap-3 expression was significantly decreased at both the transcriptional as well as translational level, in cells expressing Akap-3 shRNA as compared to control shRNA ( Figure 9E ).

To further characterize the effect of Akap-3 on osteoblast differentiation, mRNA and protein levels of osteogenic markers like RunX-2, Type-1-col and OCN were assessed which demonstrated a downregulated expression in sh-Akap-3 cells as compared to scr-shRNA control ( Figures 9F–H ). This again implies that loss of Akap-3 gene function inhibits osteoblast differentiation. The same observations were obtained when siAkap-3 was transfected in osteoblast cells ( Supplement Figure 1 )