MicroRNA 874-3p Exerts Skeletal Anabolic Effects Epigenetically during Weaning by Suppressing Hdac1 Expression*

Embryonic skeletogenesis and postnatal bone development require the transfer of calcium from the mother to the offspring during pregnancy and lactation. Therefore, bone resorption in the mother becomes elevated during these periods, resulting in significant maternal skeletal loss. There follows an anabolic phase around weaning during which there is a remarkable recovery of the maternal skeleton. However, the mechanism(s) of this anabolic response remain(s) largely unknown. We identified eight differentially expressed miRNAs by array profiling, of which miR-874-3p was highly expressed at weaning, a time when bone loss was noted to recover. We report that this weaning-associated miRNA is an anabolic target. Therefore, an agomir of miR-874-3p induced osteoblast differentiation and mineralization. These actions were mediated through the inhibition of Hdac1 expression and enhanced Runx2 transcriptional activation. When injected in vivo, the agomir significantly increased osteoblastogenesis and mineralization, reversed bone loss caused by ovariectomy, and increased bone strength. We speculate that elevated miR-874-3p expression during weaning enhances bone formation and that this miRNA may become a therapeutic target for conditions of bone loss.

The role of microRNAs (miRNAs) 3 in regulating the skeleton is just beginning to emerge (1). Notably, the conditional deletion of the miRNA-processing endoribonuclease Dicer in cells of the osteoblast lineage demonstrates a clear requirement for miRNAs in embryonic skeletogenesis as well as in postnatal bone growth, modeling, and remodeling (2). miRNAs are small non-coding RNAs that mediate translational inhibition or degradation of a transcript by binding to complementary sites of the target mRNAs. This permits effective control of gene expression beyond genomic interactions between transcription factors and their cognate elements. Many miRNAs are evolutionary conserved with well defined developmental and cell type-specific expression patterns (2)(3)(4).
In this study, we aimed to identify miRNAs that are anabolic to the skeleton. Therefore, we examined miRNA expression during lactation and weaning, the two phases of reproduction characterized by marked fluctuations in bone resorption and formation. There is profound maternal bone loss during lactation to cope with the calcium requirement for postnatal skeletal growth and modeling. This bone loss is, however, reversed through rapid new bone synthesis that replenishes the maternal skeleton during weaning (5). If the processes become decoupled, then pregnancy-and lactation-associated osteoporosis ensue (5). There is evidence for an association between breast feeding and vertebral fractures, supporting the idea that lengthy lactation may, in fact, predispose to a greater risk of osteoporosis-related fractures in postmenopausal women (6).
Testifying to the efficiency and uniqueness of this rapid remodeling phase, rats, which typically nurse multiple offspring, can lose up to 35% of their bone mineral content over 21 days of lactation. This loss is reversed completely after weaning. For this reason, the weaning phase is considered the most anabolic phase in the reproductive cycle of a rodent and provides a vital window for the identification of new anabolic targets (7)(8)(9). For example, elevated levels of the neurohypophyseal hormone oxytocin during lactation and weaning led us to examine its action on bone (10,11). Mice deficient in oxytocin or its receptor have been found to be osteopenic, and this led us to speculate that oxytocin plays a role in enabling maternal skeletal replenishment during weaning (12).
Here we report the identification of eight miRNAs whose expression was modulated specifically during weaning. We chose to further explore the action of miR-874-3p on the osteoblast both in vitro and in vivo. We conclude that miR-874-3p stimulates osteoblast differentiation, mineralization, and bone
Calvarial Cell Culture-For experiments, 1-day-old mouse calvarial cells were used, harvested from nine to ten calvariae at room temperature. Briefly, individual calvariae were isolated surgically from the skull, sutures were segregated, and the pooled calvariae were kept for repeated digestion (15 min/digestion) with 0.05% trypsin and 0.1% collagenase P to release cells. The first digestion was discarded, and cells were collected from the next four digestions. Cells were cultured in ␣-minimum Eagle's medium containing 10% FCS and 1% penicillin/ streptomycin (complete growth medium). Calvariae were allowed to reach 70 -80% confluence for the experiments (16).
Alkaline Phosphatase (ALP) and Mineralization Assay of Bone Marrow Cells (BMCs)-At the end of the treatments, BMCs from the femora of mice were flushed out in osteoblast differentiation medium containing 10 Ϫ7 M dexamethasone (bone marrow differentiation medium). BMCs were cultured for 21 days with a change of medium every 48 h. At the end of the experiment, ALP activity was measured at 405 nm, and mineralized nodules were stained and quantified at 405 nm (16).
Analysis of miRNA Target Site Prediction-In silico putative targets were screened for each of the differentially expressed miRNAs using TargetScan and miRanda on the basis of specific base-pairing rules (17,18).
In Vivo Studies-All animal care and experimental procedures were approved by the Institutional Animal Ethics Committee and Institutional Animal Care and Use Committee.
Female BALB/c mice (18 Ϯ 5 g) and female adult Sprague-Dawley rats (220 Ϯ 30 g) were kept in a 12-h light-dark cycle with controlled temperature (22-24°C) and humidity (50 -60%) and ad libitum access to standard rodent food and water (19). Micro-computed tomography (CT) measurements were carried out on a SkyScan 1076 CT scanner (Aartselaar, Antwerp, Belgium). Briefly, femora, tibiae, and vertebrae were scanned at a nominal resolution of 18 mm. One hundred projections were acquired (angular range, 180°), and image slices were reconstructed using a modified Feldkamp algorithm (according to Sky Scan Nrecon software). Parameters, including bone volume-to-tissue volume (BV/TV), trabecular number (Tb.N), and trabecular separation were calculated (15). Agomir and miR-C were administered with an anionic liposome (Invivofectamine 2.0, Life Technologies), which allows for fusion of the liposome-nucleic acid complex with the cell membrane and subsequent uptake by endocytosis. Bone formation was assessed by injecting tetracycline (20 mg/kg) into all groups of mice 30 days apart. Using fluorescence imaging and Leica-Qwin software (Leica Microsystems Inc., Buffalo Grove, IL), we calculated mineralized surface (mineral apposition) rate and bone formation rate.
Statistical Analysis-For experiments requiring multiple comparisons testing, we used one-way ANOVA followed by Newman-Keuls test of significance (GraphPad Prism v.5). Student's t test was used for experiments with only two treatments.

Results
We first studied femur epiphyseal trabecular bone across the lactation-to-weaning transition, during which rats suffer maximum bone loss followed by complete recovery (Fig. 1A). Nulliparous rats served as controls. Micro-CT showed significant trabecular and cortical loss from day 9 onward. These changes were reversed by day 60. BV/TV, Tb.N, and cortical thickness were reduced maximally at around day 18 and normalized by day 60 (Fig. 1, B and C). Cultures of bone marrow cells isolated from rats during weaning (day 60) showed enhanced mineralized nodule formation as well as increased expression of the osteoblast genes Runx2, Bglap, and Col1a1 compared with rats on day 18 (Fig. 1, D and E).
We attempted to predict the target for miR-874-3p using TargetScan and miRBase. The putative targeted gene was confirmed as Hdac1, which showed an 8-nt seed sequence match with miR-874 in the 3Ј UTR and was highly conserved among vertebrates (Fig. 2F) (21). In concordance with our in silico data, we found that the agomir to miR-874-3p repressed Hdac1 expression, whereas the antagomir increased its expression (Fig. 2G). These changes were corroborated at the protein level by Western blotting (Fig. 2G).
We next examined whether Hdac1 repression was required for miR-874-3p-induced Runx2 activation during osteoblast differentiation. Knocking down Hdac1 using siRNA increased Alp activity, mineralization, and nodule formation compared with scrambled siRNA-transfected cells (Fig. 2H). Furthermore, we assessed the expression of a Runx2 and its target gene, Bmp2, in Hdac1-depleted cells. The cells displayed a markedly increased expression of Runx2 and Bmp2 at the mRNA and protein levels (Fig. 2I). These data suggest that chromatin acetylation triggered by Hdac1 down-regulation was critical for miR-874-3p-induced Runx2 expression and osteoblast differentiation (Fig. 2J).
To address the anabolic function of miR-874-3p in vivo, we injected ovariectomized or sham-operated mice with the agomir of miR-874-3p (7 mg kg Ϫ1 ) (8), scrambled miRNA (miR-C, 7 mg kg Ϫ1 ), or PBS (0.2 ml) for 6 weeks on days 1-3 of the first, third, and fifth weeks. We first measured the expression of miR-874-3p in bone extracts to demonstrate effective in vivo tissue delivery and uptake in sham-operated and ovariectomized mice (Fig. 3A). We then evaluated trabecular bone at three sites: the femoral and tibial metaphyses and the vertebral column. At all three sites, ovariectomy expectedly reduced BV/TV and Tb.N (and increased trabecular separation). This marked trabecular loss was partially reversed by the miR-874-3p agomir, whereas miR-C and vehicle were both ineffective (Fig. 3, B-E). These putative anabolic actions were confirmed by a profound effect on indices of bone formation; namely, significant increases in mineral apposition rate and bone formation rate (Fig. 3F). We also found that the agomir increased bone strength and stiffness not only in ovariectomized mice but also in sham-operated controls. The latter is consistent with an increase in Tb.N noted in this group (Fig. 3C).
To explore the mechanism of the increase in bone mass by miR-874-3p, we isolated bone marrow stromal cells at 6 weeks.
Ovariectomy induced a reduction in Alp activity, mineralization, and Cfu-ob formation, consistent with low-turnover bone loss (Fig. 4, A-C). Although miR-C did not affect these ex vivo parameters, the agomir partially restored Alp, mineralization, and Cfu-ob (Fig. 4, A-C). Furthermore, consistent with in vitro transfection (Fig. 2), Bmp2 and Runx2 expression was increased and Hdac1 was repressed in bone extracts from agomir-treated mice compared with miR-C or vehicle-treated animals (Fig. 4, D and E). This suggests that the in vivo anabolic action of miR-874-3p is also mediated through Hdac1 repression and Runx2 activation.

Discussion
Hundreds of microRNAs regulate protein synthesis by blocking mRNA translation and, therefore, coordinate a wide spectrum of biological processes, such as organogenesis. There has been keen interest in understanding the potential roles of microRNAs in embryonic skeletal development, postnatal bone growth and modeling, and adult bone remodeling. For example, skeletal patterning has been found to be disrupted when miR-196 was depleted conditionally early during skeletal development (22). Likewise, growth plate abnormalities, including reduced numbers of proliferating chondrocytes and enhanced hypertrophic chondrocyte differentiation, have been noted to occur when the endoribonuclease Dicer was deleted specifically in chondrocyte lineage cells (23). Two miRNAs inhibitory to chondrogenesis have been identified as miR-199a and miR145 (24,25).
MicroRNAs have also been implicated in regulating postnatal bone modeling and remodeling. This is testified by a high bone mass phenotype noted in mice when Dicer is deleted in mature osteoblasts (using an osteocalcin promoter) or in osteoclasts (using a CD11b or cathepsin K promoter) (26 -29). However, even conditional Dicer deletion, which depletes all miRNAs in a given cell, does not inform us on how individual miRNAs regulate cell function. In fact, only a handful of miRNAs have been implicated directly in bone homeostasis. Although miR-223 and miR-155 control the osteoclast, notable regulators of osteoblastogenesis include two miRNA clusters, miR-23aϳ27aϳ24-2 and miR-2861ϳ3960, that are repressed and activated, respectively, by Runx2 (28, 30 -32). Specific miR-NAs, miRNA-125b, miR-138, and miR-637, are also regulated by osterix during osteoblastogenesis (33)(34)(35).
To identify osteoblast-regulatory miRNAs of physiologic relevance, we profiled bone miRNAs during weaning, a period when bone formation is grossly accelerated. MicroRNA arrays during this period, day 60 in the rat, showed the differential regulation of eight of a total of 4716 detected miRNAs. Of these, miR-212, miR-327, miR-451, miR-672, and miR-874-3p were up-regulated, whereas the expression of miR-204, miR-322-3p, and miR-664 was diminished. This suggested that hitherto unrecognized miRNAs were likely involved in skeletal anabolism during a physiologically critical window. We therefore tested directly the effect of modulating the expression one of these miRNAs, miR-874-3p, using specifically designed agomirs and antagomirs. Of note is that miR-874-3p has been characterized as a tumor-suppressive miRNA (21). We found that transfection of osteoblast precursors with the agomir stimulated cell differentiation into a mature, mineralizing phenotype and that this effect was associated with elevated Runx2 and Smad1 expression. Furthermore, this action was mediated epigenetically in that agomir-induced osteoblastogenesis and Runx2 expression were both dependent on Hdac1 repression (Fig. 2K). This is not unexpected because there is prior evidence for an interaction between miRNAs, Hdacs, and Runx2. For example, Runx2 can positively regulate miR-2861, which, in turn, inhibits Hdac5 to up-regulate Runx2 expression in a feed- forward loop (32). We likewise propose a pathway in which miR-874-3p down-regulates Hdac1 expression to enhance Runx2 expression and osteoblast differentiation.
Of note is our finding that Bmp2 expression is also regulated similarly, which is not inconsistent with miR-140-induced repression of Hdac4 and the subsequent stimulation of bone formation. It is known that Bmp2 stimulates Runx2 acetylation and prevents Smurf2-mediated degradation of Runx2 (15). In contrast, by deacetylating Runx2, Hdac1 allows the protein to undergo Smurfmediated degradation. Therefore, we hypothesize that the inhibition of Hdac1 by miR-874-3p increases Runx2 acetylation, potentiating Bmp2-induced osteoblast differentiation.
The potential anabolic action of miR-874-3p was explored in vivo in 6-week-old mice in which the skeleton underwent active modeling. Mice were ovariectomized and injected with agomir or scrambled miRNA intermittently over 6 weeks. Agomir treatment prevented trabecular bone loss and improved bone strength not only in ovariectomized mice but also in the shamoperated group. Importantly, ex vivo cultures of bone marrow stromal cells from agomir-treated mice showed evidence of increased osteoblastogenesis, mineralization, and Bmp2 and Runx2 expression as well as reduced Hdac1, essentially confirming our in vitro data. Therefore, these findings document miR-874-3p as a potential anabolic target. Considering that this miRNA is elevated 4-to 6-fold during weaning, it is possible that it plays a key role in restoring skeletal mass following pregnancy and lactation.