Smoc1 and Smoc2 regulate bone formation as novel downstream molecules of Runx2

Runx2 is an essential transcription factor for bone formation. Although osteocalcin, osteopontin, and bone sialoprotein are well-known Runx2-regulated bone-specic genes, the skeletal phenotypes of knockout (KO) mice for these genes are marginal compared with those of Runx2 KO mice. These inconsistencies suggest that unknown Runx2-regulated genes play important roles in bone formation. To address this, we attempted to identify the Runx2 targets by performing RNA-sequencing and found Smoc1 and Smoc2 upregulation by Runx2. Smoc1 or Smoc2 knockdown inhibited osteoblastogenesis. Smoc1 KO mice displayed no bula formation, while Smoc2 KO mice had mild craniofacial phenotypes. Surprisingly, Smoc1 and Smoc2 double KO (DKO) mice manifested no skull, shortened tibiae, and no bulae. Endochondral bone formation was also impaired at the late stage in the DKO mice. Collectively, these results suggest that Smoc1 and Smoc2 function as novel targets for Runx2, and play important roles in intramembranous and endochondral bone formation.


Introduction
In vertebrates two different types of bone formation processes, intramembranous ossi cation and endochondral ossi cation, are known to occur during embryonic and postnatal skeletogenesis. In intramembranous ossi cation, mesenchymal stem cells directly differentiate into osteoblasts that subsequently form bone tissues (1,2). For endochondral ossi cation, chondrocytes differentiate from mesenchymal stem cells to form cartilage tissues that are eventually replaced by bone tissues containing osteoblasts and osteoclasts (3,4).
Bone morphogenetic protein (Bmp) family members are powerful cytokines that exhibit bone and cartilage formation activities by inducing osteoblast and chondrocyte differentiation. Among the Bmp family members, Bmp2 regulates the expressions and functions of runt related transcription factor 2 (Runx2), Sp7 transcription factor 7 (Osterix), and Sex determining region Y-box 9 (Sox9), as critical transcription factors for bone and cartilage development (5)(6)(7).
In particular, Runx2, a member of the Runt family of transcription factors, plays an indispensable role in bone formation and osteoblastogenesis. Runx2-de cient mice manifest no bone formation (8). Mutations in the RUNX2 gene cause cleidocranial dysplasia, characterized by impaired bone formation in the calvaria and clavicles (9). Runx2 was also identi ed as a transcription factor that binds to the osteoblastspeci c element 2 present in the osteocalcin (Ocn) gene promoter (10). Additionally, Runx2 was su cient to promote mesenchymal cell differentiation into osteoblasts (11). During osteoblast differentiation, Runx2 speci cally regulated the expressions of osteoblast-speci c and osteogenic genes, including Ocn, osteopontin, and bone sialoprotein (12)(13)(14). However, the skeletal phenotypes in knockout (KO) mice for these genes are very marginal or absent compared with those in Runx2 KO mice (14)(15)(16)(17)(18). Therefore, it is predicted that currently unknown Runx2-target molecules play critical roles in bone formation.
In the present study, we aimed to identify novel downstream molecules of Runx2 and investigate their functional roles in skeletal development. RNA-sequencing and reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analyses indicated that Smoc1

Smoc1 and Smoc2 regulation by Runx2 and Bmp2
To identify novel molecules involved in osteoblastogenesis and bone formation, we performed RNAsequence analyses using limb bud cells infected with control or Runx2 adenoviruses, or Bmp2 adenovirus, which regulates Runx2 function and expression (19) and induces bone formation. Searching for genes that displayed a log2 fold change value of > 1 expression level in cells infected with Runx2 or Bmp2, we determined the upregulation of 653 genes by Bmp2 and 497 genes by Runx2 ( Supplementary  Fig. 1a, b; Supplementary Tables 1 and 2). From the 180 common genes that were upregulated by both, Bmp2 and Runx2 ( Supplementary Fig. 1c), we focused on the matrix proteins Smoc1 and Smoc2, which contain an extracellular calcium-binding domain (20,21), and are members of the secreted acidic cysteine-rich glycoprotein (SPARC)-related family genes. We con rmed the induction of Smoc1 and Smoc2 expressions in limb bud cells by both, Runx2 and Bmp2 by RT-qPCR analyses (Fig. 1a, 1b). To understand the importance of Runx2 for Smoc1 and Smoc2 expressions, we determined the effects of DN-Runx2 on their expressions in primary osteoblasts isolated from mouse calvariae. As expected, DN-Runx2 overexpression markedly inhibited Osterix and Osteocalcin expressions, both of which are wellknown Runx2 transcriptional targets (19,22) in primary osteoblasts (Fig. 1c, Supplementary Fig. 2), and Bmp2 induced Osterix and osteocalcin expressions in WT limb bud cells, but not in Runx2 KO limb bud cells (Fig. 1g, h). DN-Runx2 treatment suppressed Bmp2-dependent Smoc1 and Smoc2 induction in WT limb bud cells (Fig. 1i). In addition, Bmp2-dependent Smoc1 induction was absent in Runx2 KO limb bud cells (Fig. 1i) Fig. 3a, b). Furthermore, we con rmed Runx2 binding on the Smoc1 and Smoc2 genes promoters as determined by chromatin immunoprecipitation assays ( Supplementary Fig. 4). These results suggest that Smoc1 and Smoc2 are downstream molecules of Runx2 and Bmp2 signaling.

Smoc1 and Smoc2 expressions in skeletal tissues
To clarify whether Smoc1 and Smoc2 expressions are regulated by Runx2 in vivo, we examined their expressions in Runx2 KO and WT littermate mice at the E12.5 stage as determined by whole-mount in situ hybridization. The analyses revealed that Smoc1 was strongly expressed in the skull (Fig. 2a), forelimbs, hindlimbs (Fig. 2b), and vertebrae (Fig. 2c). The Smoc1 signals in the skull and vertebrae of Runx2 KO mice were moderately less than those of WT mice (Fig. 2a, c). Strong Smoc2 expression was observed in the skull of WT mice (Fig. 2d), whereas only very weak signals were detected in other tissues (Fig. 2e, f). Notably, a drastic decrease of Smoc2 expression in the skull of Runx2 KO mice was seen compared with that in the skull of WT mice (Fig. 2d). Additionally, the localization of Runx2 expression was similar to Smoc2 expression in the skull. (Supplementary Fig. 5a (Fig. 3a, b). Runx2 and Osterix expressions were signi cantly suppressed in osteoblasts by shSmoc1 and/or shSmoc2 retrovirus infection (Fig. 3c, d). Furthermore, shSmoc1 and/or shSmoc2 retrovirus infection inhibited osteoblast alkaline phosphatase activity (Fig. 3e, f) and mineralization (Fig. 3g) Fig. 9a-f). However, a moderate craniofacial phenotype, with shorter nasal to eye and nasal to parietal bone distances, was observed in Smoc2  Mutations in the SMOC1 gene were shown to result in ophthalmo-acromelic syndrome (OAS), also known as Waardenburg anophthalmia syndrome, which manifests a distinctive pattern of distal limb anomalies (33,34 Fig. 8). Functional studies needed to understand the SMOC family still appear complex. Restricted distribution of Bmp signaling is regulated by a number of mechanisms, as shown in previous reports; ligand binding by extracellular Bmp antagonists (35); interactions with extracellular matrix proteins, including heparan sulfate proteoglycans (36). Smoc protein has also been shown to bind to heparan sulfate proteoglycans (37). Smoc1 has been shown to inhibit Bmp signaling and to be essential for postgastrulation development in Xenopus (38). Application of Bmp2 to NIH3T3 broblasts transfected with SMOC was shown to inhibit Smad1/5/8 phosphorylation (38). Pent, the orthologue pentagon of Smoc in Drosophila, was found to be expressed in developing wing imaginal discs and to inhibit BMP signaling (39). Meanwhile, a recent study showed that SMOC proteins have dual functions as BMP inhibitors and expanders of BMP signaling (40

Cells and reagents
LentiX-293T cells were purchased from Takara (Shiga, Japan). Plat-E cells were a generous gift from Dr. Kitamura (The University of Tokyo, Tokyo, Japan). Osteoblasts and limb bud cells were cultured in alpha modi cation of Eagle's minimum essential media (α-MEM; Thermo Fisher, Waltham, MA, USA) containing 10% fetal bovine serum (FBS) at 37 °C in a humidi ed 5% CO2 incubator. Plat-E cells and LentiX-293T cells were cultured in Dulbecco's modi ed Eagle's medium (DMEM; Thermo Fisher) containing 10% FBS. Recombinant Bmp2 was obtained from conditioned medium of LentiX-293T cells transfected with a Bmp2 expression vector as described previously (41). Bmp2 activity was determined by comparison with human recombinant Bmp2 (Peprotech, Rocky Hill, NJ, USA).
Osteoblasts were isolated from calvariae of 3-5-day-old neonatal mice by a sequential enzymatic digestion method as described previously (42). Brie y, mouse calvariae were gently incubated with 5 mM ethylenediaminetetraacetic acid (EDTA) in phosphate-buffered saline (PBS) for 1 h at 37 °C, followed by three 20-min digestions with 0.25% collagenase in DMEM for 20 min at 37 °C. Cells obtained during the last two digestion processes were collected together in α-MEM containing 10% FBS. Throughout the subsequent experiments, α-MEM containing 10% FBS, 50 μg/mL ascorbic acid, and 5 mM sodium βglycerophosphate was used to induce osteoblastic differentiation. Limb bud cells were isolated from mouse embryos at E12-E13 and digested with 0.05% trypsin/0.53 mM EDTA in PBS for 10 min at 37 °C. Cells obtained during the digestion were collected in α-MEM containing 10% FBS. For monolayer culture, limb bud cells were seeded at 1.6 × 10 5 cells/cm 2 . All other chemicals used were of the highest purity commercially available.

RNA-sequence and data analysis
Limb bud cells were infected with Venus, Runx2, or Bmp2 adenovirus. After 4 days of incubation, total RNA was extracted using a Nucleospin RNA Plus Kit (Takara). Sequencing was performed on an Illumina HiSeq 2500 platform (Illumina, San Diego, CA, USA) in a 75-base single-end mode. Sequenced reads were mapped to the mouse reference genome sequences (mm10) using TopHat version 2.1.1 in combination with Bowtie 2 version 2.4.2. and SAMtools version 1.11. The fragments per kilobase of exon per million mapped fragments were calculated using Cu inks version 2.2.1. RNA-sequencing data was analyzed by iDEP 9.1. Raw reads from these samples were submitted to the National Center for Biotechnology Information Gene Expression Omnibus database (accession number: GSE166982).

Plasmids
Venus, Runx2, DN-Runx2, or Bmp2 cDNAs were ligated to the pAXcawt adenovirus vector (Takara) as described previously (43). Flag-tagged-DN-Runx2 used in this study contains amino acids 2-247 of Runx2 (11). This construct lacks the transcriptional activation domain at the C-terminal region. The generation of these adenoviruses was performed using an adenovirus generation kit (Takara). A Venus adenovirus was used as the control adenovirus (44).

Retrovirus infection
PLAT-E cells were seeded at 8 × 10 4 cells/cm 2 at 1 day before transfection. Polyethylenimine (PEI) was used for all transfections. The pSIREN-retroQ shRNA expression vectors for shSmoc1 and shSmoc2 were mixed with PEI, and the plasmid-PEI complexes were incubated in Opti-MEM (Thermo Fisher) for 15 min at room temperature, and added to PLAT-E cells. The virus supernatant was collected at 48 h after transfection, and used to infect osteoblasts for 48 h in the presence of 4 μg/mL polybrene.
Determination of ALP activity ALP activity was determined as described previously (42,45,46). In brief, cells were washed with PBS and solubilized with 0.1% Triton X-100, followed by determination of the ALP activity in lysates using pnitrophenol phosphate as a substrate. Protein contents of the lysates were determined using Bradford protein assay reagent (Bio-Rad, Hercules, CA, USA). For cytochemical analysis, cells were washed with PBS and xed with 4% paraformaldehyde in PBS. Subsequently, cells were stained with a mixture of 330 μg/mL nitro blue tetrazolium, 175 μg/mL bromochloroindolyl phosphate, 100 mM NaCl, 50 mM MgCl 2 , and 100 mM Tris (pH 9.5).

Alizarin red staining
Following induction of differentiation, cultured osteoblasts were washed with PBS twice, xed in 70% ethanol, and stained with 0.4% alizarin red solution for 10 min.

Skeletal preparation of mice
Following removal of the skin and viscera, mice were xed in 96% ethanol for 24 h. Cartilage was stained for 24 h with alcian blue solution containing 0.015% alcian blue 8GX, 20% acetic acid, and 80% ethanol. Following dehydration with 100% ethanol for 3 days, the whole bodies were digested with 1% KOH at room temperature until the skeleton became clearly visible. The specimens were subsequently stained with 0.002% alizarin red in H 2 O for 24 h. Finally, the specimens were maintained in 100% glycerol and observed and photographed under an S-APO microscope (Leica, Wetzlar, Germany).

Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
Cultured cells were washed twice with PBS and subjected to total RNA extraction with a Nucleo Spin RNA Plus Kit (Takara). cDNA was synthesized using ReverTra Ace ® qPCR RT Master Mix with gDNA Remover (TOYOBO, Osaka, Japan). The individual cDNAs were ampli ed with THUNDERBIRD ® SYBR qPCR Mix (TOYOBO) using a StepOnePlus™ Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The relative expression levels of the target genes was determined by the delta-delta Ct method using transcripts of Actb as the internal reference for each mouse RNA sample. The primer pairs and probes used for ampli cation were: coding sequence (position 460-1500 in the NM_022315.2 cDNA sequence). The Runx2 probe was a 639bp fragment of the coding sequence (position 3173-3807 in the NM_001146038.2 cDNA sequence).
Hematoxilin and eosin (HE) and von Kossa staining Tissue preparation was conducted as previously described (47). Brie y, mice tibiae at E15.5 were xed in 4% paraformaldehyde in PBS and used to prepare para n-embedded sections with 7-µm thickness. The sections were depara nized and stained with Mayer's hematoxylin and eosin solution. Bone mineralization was analyzed using the von Kossa staining method. In brief, the sections were xed in 4% paraformaldehyde in PBS and subsequently incubated with 5% silver nitrate for 60 min under UV light. Following rinsing with distilled water, bone nodules were photographed using a phase-contrast microscope.
In situ hybridization Mice tibiae at E15.5 were xed in 4% paraformaldehyde in PBS and used to prepare para n-embedded sections with 7-μm thickness. DIG-labeled single-stranded RNA probes were prepared using a DIG RNA

Statistical analysis
The statistical signi cances of differences in data were determined by two-tailed and unpaired Student's t-tests for two groups. Difference between three or more groups were compared by one-way analysis of variance (ANOVA) or two-way ANOVA followed by Tukey's multiple-comparisons test. Values of P < 0.05 were considered to indicate statistical signi cance.   Bmp2 adenovirus for 6 h, followed by incubation in differentiation medium containing 50 μg/mL ascorbic acid and 5 mM sodium β-glycerophosphate for 5 days. Total RNA was extracted from the cells and expressions of Smoc1 (a), Smoc2 (b), Runx2 (c), and Osterix (d) mRNAs were determined by RT-qPCR. (eg) Primary osteoblasts isolated from mouse calvariae were infected with shGFP, shSmoc1, shSmoc2, or both shSmoc1 and shSmoc2 retroviruses. The cells were infected with Bmp2 adenovirus and incubated in the presence of 50 μg/mL ascorbic acid and 5 mM sodium β-glycerophosphate. After incubation for 3 or 5 days, the cells were evaluated by alkaline phosphatase staining (e), alkaline phosphatase activity (f), and alizarin red staining (g). Figure 4