Hedgehog-Gli Activators Direct Osteo-chondrogenic Function of Bone Morphogenetic Protein toward Osteogenesis in the Perichondrium*

Background: During endochondral ossification, cells in the perichondrium give rise to osteoblast precursors. Results: Bone morphogenetic protein (BMP) interacted with hedgehog (Hh) to enhance osteogenesis, whereas in the absence of Hh, BMP enhanced ectopic chondrogenesis in the perichondrium. Conclusion: Hh alters the function of BMP to specify perichondrial cells into osteoblasts. Significance: This provides an insight into signaling network in osteogenesis. Specification of progenitors into the osteoblast lineage is an essential event for skeletogenesis. During endochondral ossification, cells in the perichondrium give rise to osteoblast precursors. Hedgehog (Hh) and bone morphogenetic protein (BMP) are suggested to regulate the commitment of these cells. However, properties of perichondrial cells and regulatory mechanisms of the specification process are still poorly understood. Here, we investigated the machineries by combining a novel organ culture system and single-cell expression analysis with mouse genetics and biochemical analyses. In a metatarsal organ culture reproducing bone collar formation, activation of BMP signaling enhanced the bone collar formation cooperatively with Hh input, whereas the signaling induced ectopic chondrocyte formation in the perichondrium without Hh input. Similar phenotypes were also observed in compound mutant mice, where signaling activities of Hh and BMP were genetically manipulated. Single-cell quantitative RT-PCR analyses showed heterogeneity of perichondrial cells in terms of natural characteristics and responsiveness to Hh input. In vitro analyses revealed that Hh signaling suppressed BMP-induced chondrogenic differentiation; Gli1 inhibited the expression of Sox5, Sox6, and Sox9 (SRY box-containing gene 9) as well as transactivation by Sox9. Indeed, ectopic expression of chondrocyte maker genes were observed in the perichondrium of metatarsals in Gli1−/− fetuses, and the phenotype was more severe in Gli1−/−;Gli2−/− newborns. These data suggest that Hh-Gli activators alter the function of BMP to specify perichondrial cells into osteoblasts; the timing of Hh input and its target populations are critical for BMP function.


Specification of progenitors into the osteoblast lineage is
an essential event for skeletogenesis. During endochondral ossification, cells in the perichondrium give rise to osteoblast precursors. Hedgehog (Hh) and bone morphogenetic protein (BMP) are suggested to regulate the commitment of these cells. However, properties of perichondrial cells and regulatory mechanisms of the specification process are still poorly understood. Here, we investigated the machineries by combining a novel organ culture system and single-cell expression analysis with mouse genetics and biochemical analyses. In a metatarsal organ culture reproducing bone collar formation, activation of BMP signaling enhanced the bone collar formation cooperatively with Hh input, whereas the signaling induced ectopic chondrocyte formation in the perichondrium without Hh input. Similar phenotypes were also observed in compound mutant mice, where signaling activities of Hh and BMP were genetically manipulated. Single-cell quantitative RT-PCR analyses showed heterogeneity of perichondrial cells in terms of natural characteristics and responsiveness to Hh input. In vitro analyses revealed that Hh signaling suppressed BMP-induced chondrogenic differentiation; Gli1 inhibited the expression of Sox5, Sox6, and Sox9 (SRY box-containing gene 9) as well as transactivation by Sox9. Indeed, ectopic expression of chondrocyte maker genes were observed in the perichondrium of metatarsals in Gli1 ؊/؊ fetuses, and the phenotype was more severe in Gli1 ؊/؊ ;Gli2 ؊/؊ newborns. These data suggest that Hh-Gli activators alter the function of BMP to specify perichondrial cells into osteoblasts; the timing of Hh input and its target populations are critical for BMP function.
Specification of progenitors into a certain lineage is an essential event for organogenesis. In mammals, appendicular and axial skeletons develop through endochondral ossification, one of the two ossification processes in vertebrates (1). In an early phase of the process, two distinct cell populations arise shortly after the condensation of mesenchymal cells: cartilage-forming chondrocytes and perichondrial cells (1). Sox9 (SRY box-containing gene 9) has been shown to be essential for the mesenchymal condensations and subsequent cartilage formation (2,3). Perichondrial cells constitute the perichondrium, a thin layer of fibroblastic cells surrounding the cartilage mold. During normal development, a population of the perichondrial cells is specified into an osteoblast lineage; they differentiate into osteoblasts through several stages of precursors, contributing to formation of the bone collar, a predecessor of cortical bones (1,4). Runx2 (Runt-related transcription factor) and Osx (osterix), two essential transcription factors for osteoblast differentiation, are known to sequentially regulate the differentiation program (5)(6)(7).
Several lines of evidence have suggested that Hh, 3 Wnt, and BMP signaling pathways act as a switch for the program in perichondrial cells and that a bipotential cell population exists in the perichondrium (8 -10). We and others have clarified crucial roles of Hh signaling in an early commitment and differentiation of osteoblasts in the perichondrium (8,(11)(12)(13). Defects in Wnt/␤-catenin signaling as well as the absence of Hh input lead to a complete lack of osteoblasts, allowing perichondrial cells to adopt a chondrocyte lineage (8,9,14,15). On the other hand, forced activation of BMP signaling has been shown to cause ectopic chondrocytes in the perichondrium (16). Although Wnt/␤-catenin signaling acts downstream of Hh signaling in the specification of osteoblasts in the perichondrium (9), the hierarchy or interaction of BMP and Hh signaling pathways in the specification has not been clarified. In addition, there have been no data that advance our understanding of the characteristics of perichondrial cells at the single-cell level.
The present study was aimed to elucidate molecular mechanisms underlying the specification of an osteoblast lineage in perichondrial cells, with a particular focus on the interaction between Hh and BMP signaling pathways, through an organ culture system and mouse genetics as well as biochemical approaches. In addition, to gain insight into the specification process, we investigated the expression of osteoblast and chondrocyte marker genes in primary perichondrial cells at the single-cell level.
Cell Culture-C3H10T1/2 cells were obtained from the RIKEN Cell Bank. Perichondrial cells were isolated as described previously (11). Plasmid transfection was performed using FuGENE HD (Roche).
Single-cell Analyses-Primary perichondrial cells treated with DMSO or SAG (1 M) for 5 days were collected by a capillary tip after trypsin treatment. cDNA libraries from a singlecell were prepared by using oligo(dT) 30 -immobilized beads and gene expressions were quantified by qPCR as described previously (21). Hierarchical clustering was performed with Mathematica software (version 7.0, Wolfram Research, Inc.). The detailed procedures and primer sequences are described in supplemental "Experimental Procedures." qRT-PCR-Total RNA extraction, reverse-transcription, and qPCR were performed as described previously (11). The primer sequences are available upon request.
Histology-The procedures for von Kossa staining, safranin O staining, and in situ hybridization have been described previously (11,20). Images were taken using an Axio Imajor A1 microscope (Carl Zeiss).
Luciferase Assay-Cells were plated onto 24-well plates and transfected with 0.4 g of DNA mixture containing the test reporter plasmids, control reporter plasmids encoding Renilla luciferase, and effector plasmids. A Dual-Luciferase assay was performed as described previously (11).
ChIP-ChIP was performed with a One-Day ChIP kit (Diagenode). To shear genomic DNA, a Shearing ChIP kit (Diagenode) was used according to the manufacturer's instructions. The primer sequences are available upon request.
Statistical Analysis-The means of groups were compared by analysis of variance, and the significance of differences was determined by post hoc testing using Tukey's method.

Development of an Organ Culture System That Enables the Observation of Bone Collar Formation in a Near in Vivo Setting-
With the ultimate aim of defining the roles of signaling pathways in osteogenesis in the perichondrium, we set out to establish an organ culture system that enabled us to analyze bone collar formation in endochondral ossification ex vivo. We found that a metatarsal organ culture using an osteogenic medium reproduced bone collar formation in synchronization with cartilage mineralization, which had not yet been achieved by the culture methods reported previously (20). To verify that it reliably reproduced phenotypes that were observed in mice with mutations in Hh, BMP, or Wnt signaling, we histologically analyzed metatarsals cultured with activators and inhibitors of the Hh, BMP, or Wnt signaling pathway. We used SAG, an Hh signaling activator; cyclopamine, an Hh signaling inhibitor; rhBMP2; dorsomorphin, a BMP signaling inhibitor (22); lithium chloride (LiCl), a glycogen synthase kinase-3 (GSK3) inhibitor; and IWP2, an inhibitor of Wnt processing and secretion (23).
SAG induced ectopic bone collar formation in the perichondrium and inhibited cartilage mineralization; cyclopamine blocked bone collar formation and accelerated cartilage mineralization (Fig. 1A). rhBMP2 did not induce ectopic bone collar formation but enhanced chondrogenesis in the growth plate; dorsomorphin did not affect osteogenesis but inhibited chondrogenesis (Fig. 1, A and B). LiCl promoted bone collar forma-tion in synchronization with cartilage mineralization; IWP2 inhibited bone collar formation (supplemental Fig. S1A). These observations were in line with those obtained from gain-and loss-of-function mouse models for the Hh, BMP, and Wnt signaling pathways (8, 9, 12, 14 -16, 22, 24, 25).
The integrity of the organ culture system was further verified and investigated in terms of the signaling hierarchy of Hh and Wnt and the interaction of Hh and TGF␤ during the commitment of the osteoblast lineage in the perichondrium. Hh signaling is known to be required prior to the requirement of Wnt signaling in osteoblast differentiation (9). In the cultured metatarsals, Osx was expressed in the SAG-induced ectopic bone collar. Although IWP2 inhibited SAG-induced formation of ectopic bone collar, ectopic Osx expression was maintained (supplemental Fig. S1B). As for the loss-of-function of Hh signaling, cyclopamine alone suppressed both bone collar formation and Osx expression in the perichondrium. LiCl did not reverse the cyclopamine-induced suppression of Osx expression and bone collar formation in the perichondrium (supplemental Fig. S1B). TGF␤ signaling is reported to be involved in the proliferation of perichondrial cells and osteoblast differentiation (26,27). In the cultured metatarsals in the present study, rhTGF␤1 increased the thickness of the perichondrium and inhibited the SAG-induced ectopic bone collar formation (supplemental. Fig. S2). These data not only confirmed the previous reports but also suggest that TGF␤ signaling inhibits Hh-induced osteogenesis. Thus, pharmacological manipulation of this organ culture system appeared to enable easier and less time-consuming investigation of the signaling network and hierarchy in the perichondrium than genetic approaches.
Hh Signaling-mediated Alteration of the BMP Function in the Perichondrium from Chondrogenic to Osteogenic-By taking advantage of the present organ culture system described above, we next attempted to clarify the signaling network between Hh and BMP in the perichondrium; we examined the effects of combined treatments of activators and inhibitors of Hh and BMP signaling on osteogenesis. Combined treatment of SAG and rhBMP2 increased the thickness but not the region of the bone collar compared with SAG alone (Fig. 1A). On the other hand, dorsomorphin, a BMP inhibitor, suppressed SAG-induced ectopic bone collar formation (Fig. 1A). Regarding the loss of function of Hh signaling, cyclopamine inhibited bone collar formation in organ culture (Fig. 1A), whereas it did not alter the expressions of BMP ligands including Bmp2, Bmp4, Bmp6, and Bmp7 in primary perichondrial cells (supplemental Fig. S3). Interestingly, rhBMP2 did not reverse the cyclopamine-induced suppression of bone collar formation (Fig.  1A). Instead, it induced safranin O-positive chondrocyte-like cells in the perichondrium when combined with cyclopamine, suggesting that ectopic chondrocytes were formed upon the treatment (Fig. 1B). In situ hybridization for Col2a1 (type II collagen ␣1 chain) revealed that cyclopamine alone induced ectopic Col2a1 expression in the perichondrium, and combined treatment with cyclopamine and rhBMP2 expanded the Col2a1-expressing region (Fig. 1C).
To confirm the interactions between Hh and BMP signaling in vivo, we analyzed E18.5 metatarsals of Ptch1 ϩ/Ϫ ;Prx1-cre; CAG-LoxP-caBmpr1a mice, in which both Hh and BMP signaling were up-regulated in limb mesenchymal cells. The thickness of the bone collar was increased in Ptch1 ϩ/Ϫ ;Prx1-cre; CAG-LoxP-caBmpr1a mice ( Fig. 2A), which was consistent with the findings in the metatarsals cultured with the combination of SAG and rhBMP2 (Fig. 1A). In Ptch1 ϩ/Ϫ or Prx1-cre; CAG-LoxP-caBmpr1a mice, as well as CAG-LoxP-caBmpr1a (control) mice, bone collar formation remained normal. We also analyzed E17.5 metatarsals of Ihh ϩ/Ϫ ;Prx1-cre;CAG-LoxP-caBmpr1a mice, in which Hh signaling was down-regulated and BMP signaling was up-regulated in limb mesenchymal cells. One out of three mutants of Ihh ϩ/Ϫ ;Prx1-cre;CAG-LoxP-caBmpr1a showed ectopic Col2a1 expression in the perichondrium (Fig. 2B), which was consistent with the findings in the metatarsals cultured with a combined treatment of cyclopamine and rhBMP2 (Fig. 1B). The other two mutants as well as Ihh ϩ/Ϫ and Prx1-cre;CAG-LoxP-caBmpr1a mice did not show such phenotypes. Overall, the findings on the in vivo genetic approach as well as pharmacological manipulation in the ex vivo organ culture suggested that BMP signaling acts as an accelerator for both osteogenesis and chondrogenesis in the perichondrium, after Hh-dependent lineage specification into osteoblasts or chondrocytes has taken place.
Heterogeneity of Perichondrial Cells-The different phenotypes in the perichondrium upon different combined stimuli in the organ culture suggest the heterogeneity of perichondrial cells. To gain insight into the heterogeneity, we analyzed the expression of osteoblast and chondrocyte marker genes in primary perichondrial cells at the single-cell level by quantitative RT-PCR (qRT-PCR) using a magnetic dT primer (21), focusing on their characteristics in nature as well as their properties acquired in response to Hh input.
The expressions of five genes were examined in 90 cells treated with SAG (activation of Hh signaling, 44 cells) or DMSO (control, 46 cells): Alpl, an early marker of osteoblast differentiation; Sp7 (osterix), a key transcription factor for osteoblast differentiation; Col2a1, a chondrocytic marker; Gli1, a readout of Hh signaling activation; and Eef1g, a housekeeping gene as a control. Clustering analysis on the qRT-PCR data classified the 90 cells into several clusters based on their expression patterns for these five genes (Fig.  3). Most of the DMSO-treated cells exhibited either of two features, chondrocyte-like phenotypes characterized by Col2a1 expression or non-osteochondrogenic phenotypes characterized by the lack of osteoblast and chondrocyte marker gene expression (NOC state). SAG-treated cells showed more heterogeneity in terms of their differentiation states, although the activation of Hh signaling was evidenced by up-regulation of Gli1 in these cells. Some of the cells still maintained an NOC state, but others expressed osteoblast marker genes, Alpl and Sp7, or both osteoblast and chondrocyte marker genes. Thus, the perichondrium likely consists of a heterogenous cell population with different properties and different levels of responsiveness to differentiation stimuli.
Hh Signaling Suppresses BMP-induced in Vitro Chondrogenic Differentiation, the Expression of Sox Trio, and Transactivation by Sox9-Given that both Hh and BMP signaling are activated in the perichondrium during endochondral ossification (28), the data shown in Figs. 1 and 2 led us to investigate whether Hh signaling had a negative impact on the chondrogenic activity of BMP signaling in the perichondrium. We considered that such an impact, if present, would arise via either of two mechanisms: a direct interaction between Hh and BMP signaling, or an indirect one, by which Hh signaling regulated chondrogenesis-related factors acting downstream of BMP signaling. We initially investigated the former possibility using the C3H10T1/2 cell line, which has been reported to differentiate into both chondrogenic and osteogeneic cells (29). SAG suppressed rhBMP2induced mRNA expression of chondrocyte marker genes including Col2a1, Col9a1 (type IX collagen ␣1 chain), and chondromodulin 1 (Chm1) (Fig. 4A). Time course analysis of gene expression revealed that the suppression was synchronized with the up-regulation of Gli1, a readout of Hh signaling (Fig. 4B). However, the following observations suggested that there was no direct interaction or signal cross-talk between Hh and BMP signaling in the regulation of chondrocyte phenotypes (Fig. 4B). First, there was a time lag between expression of Id1, a readout of BMP signaling, and Col2a1 expression in response to rhBMP2 treatment, which indicated that activation of BMP signaling indirectly up-regulated Col2a1 as suggested in previous reports (30,31). Second, the expression of target genes of either signaling was not altered by the addition of their counterparts (SAG versus SAGϩrhBMP2 in Gli1 expression; rhBMP2 versus SAGϩrhBMP2 in Id1 expression).
BMP signaling has been shown to regulate chondrogenesis through Sox5, Sox6, and Sox9 (Sox trio)-dependent mechanisms, and a BMP-responsive element in the Col2a1 region corresponded to a Sox9 binding site (ϩ2429 to ϩ2435) in the first intron of the Col2a1 gene (30 -34). Therefore, we tested the second possibility, i.e. the existence of an indirect interaction between Hh and BMP signaling, by examining the effects of Hh signaling on the expression or function of Sox transcription factors. We found that SAG suppressed rhBMP2-induced mRNA expression of Sox5, Sox6, and Sox9 (Fig. 4C). A luciferase assay using the reporter construct containing a Sox9-responsive element in the COL2A1 region (ϩ285 to ϩ2450; COL2A1-Luc) showed that SAG inhibited Sox9-mediated reporter activity as well as rhBMP2-mediated activity (Fig. 4D). These data suggest that Hh signaling suppresses BMP signaling-mediated chondrogenesis, and that this effect occurs at least in part through inhibition of both the expression of the Sox trio and the transactivation by Sox9.
Gli Activators Are Involved in the Hh-mediated Suppression of Chondrogenesis in the Perichondrium-We next attempted to elucidate which molecules contribute to the inhibitory effect of Hh signaling on chondrogenic activity of the BMP-Sox trio, focusing on the transcriptional machinery of Hh-Gli signaling. In C3H10T1/2 cells, Gli1 overexpression suppressed the rhBMP2-induced Col2a1 mRNA expression, whereas this sup-pression was not observed in response to the overexpressions of Gli2, the full-length form of Gli3, or the repressor form of Gli3 (Fig. 5A). The suppressive effects of Gli1 on chondrogenesis were confirmed by gain-of-function analyses of primary perichondrial cells (supplemental Fig. S2A) and loss-of-function analyses using short interfering RNA for Gli1 (shGli1) in C3H10T1/2 cells (Fig. 5B).
Finally, we investigated the molecular mechanism underlying the suppression of chondrogenic differentiation by the Hh-Gli1 pathways. Gli1 overexpression suppressed the mRNA expressions of the Sox trio in C3H10T1/2 cells (Fig. 5E). A luciferase assay using COL2A1-Luc revealed that Gli1 inhibited the transactivation by Sox9 (supplemental Fig. S3B). We also examined the effects of Gli1 on the Sox9-dependent induction of COL2A1 mRNA using the HEK293 cell line, since this cell line is known to express COL2A1 mRNA upon transfection of the Sox trio (35); we found that Gli1 overexpression inhibited the COL2A1 mRNA expression induced by the Sox trio in a dosedependent manner (supplemental Fig. S2C), without any changes in the protein localization of exogenous Sox9 (supplemental Fig. S2D). In addition, a chromatin immunoprecipitation (ChIP) assay revealed that Gli1 overexpression inhibited the recruitment of Sox9 onto the Sox9 binding site in the COL2A1 first intron in a dose-dependent manner in perichondrial cells (Fig. 5F). Thus, Gli1 suppresses the expressions of the Sox trio and the recruitment of Sox9 to the Col2a1 enhancer region, and this suppression may underlie the negative impact of Hh signaling on chondrogenic activity of the BMP-Sox trio.

DISCUSSION
This study has three major findings. First, BMP signaling interacted with Hh signaling to enhance the bone collar formation, whereas in the absence of Hh input, BMP enhanced ectopic chondrocyte formation in the perichondrium. Second, perichondrial cells showed heterogeneity in terms of natural characteristics and responsiveness to differentiation stimuli. Third, Hh signaling suppressed BMP signaling-induced chondrogenic differentiation, possibly by the Gli1-mediated inhibition of expression of the Sox trio and transactivation by Sox9. The mRNA expression levels of the osteoblast marker genes and of Col2a1 and Gli1 were determined by qRT-PCR analysis. The cluster dendrogram indicates the four cellular subtypes: those expressing the osteoblast (OB) marker genes and the Col2a1, OB(ϩ);C2(ϩ) (shown in purple); those expressing osteoblast marker genes but not Col2a1, OB(ϩ);C2(Ϫ) (shown in red); those expressing Col2a1 but not the osteoblast marker genes, OB(Ϫ); C2(ϩ) (shown in blue); and those expressing neither osteoblast marker genes nor Col2a1, NOC (shown in black). Information on Gli1 expression is also presented for each of the above. Our organ culture system using osteogenic medium enabled us to observe the bone collar formation of endochondral ossification. The ex vivo cultures of fetal long bones reported previously had not fully reproduced endochondral ossification because they lacked bone collar formation (20). As indicated in this study, our system would offer clues to understanding the network of multiple signaling pathways that regulate the differentiation of perichondrial cells, in a near in vivo setting and in a simple manner, prior to proceeding to genetic approaches. Although it is important to keep in mind the potential off-target effects of stimuli in our organ culture system, the system would enable easy manipulation of different signaling pathways at different time points.
BMP signaling is known to function as an osteo-chondrogenic factor with positive impacts on both osteogenesis and chondrogenesis during endochondral ossification (36). Rodda et al. (9) raised the possibility that Hh signaling altered the responsiveness of osteoblast progenitors to a BMP input in the perichondrium. Our findings support this possibility and further demonstrate how BMP signaling interacts with Hh signaling to induce osteoblast differentiation in the perichondrium; BMP signaling is not likely to be involved in specification of the osteoblast lineage, but it clearly has a positive impact on osteogenesis in the committed osteoblast precursors. Given that BMP signaling enhanced chondrogenesis in the perichondrium in the absence of Hh input, and Hh signaling suppressed BMP- mediated chondrocyte differentiation in vitro, we infer that the osteo-chondrogenic function of BMP is directed by Hh signaling in the perichondrium.
We were not able to discern the reasons why only one of the three Ihh ϩ/Ϫ ;Prx1-cre;CAG-LoxP-caBmpr1a mutants showed ectopic Col2a1 expression in the perichondrium ( Fig. 2A). We infer that penetrance may underlie the phenotype (37). In addition, these data support the threshold concept for ectopic chondrocyte formation in the perichondrium, where the positive action of BMP and the inhibitory action of Hh may compete with each other in chondrogenesis. The balance between the two signals would determine a threshold for the ectopic chondrocyte formation; a low level of Hh signaling may decrease the threshold, resulting in chondrogenesis in the perichondrium upon the activation of BMP signaling. Moreover, single-cell analyses showed that among cells that expressed Gli1 upon Hh input, not all, but some populations expressed osteoblast marker genes. These data are consistent with the notion that Hh signaling is necessary but not sufficient to induce bone formation (8). Hence, the timing of Hh input and its target population is critical for the osteogenic function of BMP signaling.
In this study, we performed the first examination of the populations of perichondrial cells at a single-cell level. Our data showed that cells stimulated by an osteogenic reagent showed heterogeneity in terms of their differentiation state. Most cells treated with DMSO were either cells expressing Col2a1 or NOC cells. In response to Hh input, the majority of cells expressed osteoblast marker genes. There is a possibility that the Col2a1-positive cells observed in the DMSO group were derived from cartilage due to contamination during the isolation procedure. If the contamination occurred, the chondrocytes might have kept their phenotypes or newly acquired osteoblast phenotypes upon Hh input. However, we did not observe a chondrocyte-like population in the SAG-treated group. Furthermore, chondrocytes are not likely to differentiate into osteoblasts upon Hh input because it was previously reported that no ectopic osteoblasts were observed in cartilageforming regions in chondrocytes overexpressing constitutive active Smoothened (38) or ablation of Patched1 (24). Therefore, we assume that little, if any, contamination of chondrocytes occurred during our isolation procedure and that the Col2a1positive cells or NOC cells residing in the perichondrium give rise to cells expressing osteoblast marker genes upon Hh input. Col2a1-expressing cells are known to exist in the perichondrium and contribute to the osteoblast linage (10), which supports this assumption. Indeed, several lines of study have shown that gene manipulation by Col2a1-Cre mice caused phenotypes characterized by osteogenesis in the perichondrium (8,15,24). Unlike the Col2a1-expressing cells mentioned above, the properties of NOC cells remain to be clarified, although it is possible that some of these cells are caused by the contamination of fibroblasts, or that they are a progenitor-like undifferentiated population contributing to the maintenance of a progenitor pool in the perichondrium and/or to osteoblast formation.
The Hh-Gli1 pathway inhibited chondrocyte differentiation, possibly by inhibition of the expression of the Sox trio and transactivation by Sox9. Given that Sox9 is required for the expressions of Sox5 and Sox6 (3) and Sox9 transcription is suggested to be controlled by Sox9 itself (39), the initial events in this context may be an inhibition of the recruitment of Sox9 into its own regulatory regions by Gli1. However, a recent paper showed that there were conserved Gli-binding sites near the Sox motif in the Sox5 and Sox6 regions (40), which raised the possibility that an inhibitory action by Gli1 might occur in these regions. This study, however, does not address how Gli1 inhibits the recruitment of Sox9 onto DNA. Further studies including genome-wide analyses may be needed to clarify whether competition between Gli1 and Sox9 occurs in association with DNA binding or through epigenetic regulation. On the other hand, Zhao et al. (41) reported that Hh-Gli2 signaling is a mediator of Bmp2 expression, which may not concur with our data. In our analyses, Hh agonist did not alter the expression of Id1, a target gene of BMP signaling (Fig. 4B), and the Hh inhibitor did not inhibit Bmp2 expression (supplemental Fig. S3). The discrepancy may be due to differences in the cells used between the Zhao et al. (41) study and our work; they used C2C12, a myoblast cell line, 2T3, a calvaria osteoblast cell line, and pri-mary calvaria osteoblasts to examine the effects of Hh-Gli2 signaling on endogenous Bmp2 mRNA expression, whereas we used C3H10T1/2 and primary perichondrial cells. We believe that clarifying the distinct roles of Hh and BMP signaling in each cell type will be important contributions to our understanding of bone development and clinical applications.
In conclusion, this study demonstrated that the osteo-chondrogenic function of BMP was directed toward osteogenesis by Hh-Gli activators in the perichondrium. Perichondrial cells showed various properties upon Hh input, which may underlie the Hh-mediated alteration of BMP function in perichondrial cells. Previous reports described interactions between Hh and BMP signaling toward chondrogenesis in limb bud mesenchymal cells (42,43), which appear to be inconsistent with our findings. We hypothesize that the discrepancy may be caused by the difference of cells used between these studies and our work. The target populations of Hh and BMP signaling, or other signaling pathways interacting with them, may be different between the limb bud and the perichondrium. In addition, cellular characteristics may be distinct between the limb bud and perichondrium, in terms of multipotency and responsiveness to stimuli for lineage specifications. Sox9-positive cells in the limb bud were reported to differentiate into various skeletal cells, including chondrocytes, osteoblasts, synovial cells, and tendon cells (44). However, the differentiation potential of perichondrial cells was restricted to an osteoblast or chondrocyte lineage in vivo (8). Although the distinct characteristics and signaling networks remain to be clarified, our study provides insights into the molecular basis of the signaling network between Hh and BMP in the perichondrium as well as the natural characteristics of perichondrial cells themselves. Studies including single-cell tracking and genome-wide analyses may lead to further understanding of the signaling network determining the fates of perichondrial cells and to clinical application of such knowledge to the treatment of bone defect-related diseases.