• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Dec 12, 2006; 103(50): 19004–19009.
Published online Dec 1, 2006. doi:  10.1073/pnas.0605170103
PMCID: PMC1748167
Developmental Biology

Dominance of SOX9 function over RUNX2 during skeletogenesis

Abstract

Mesenchymal stem cell-derived osteochondroprogenitors express two master transcription factors, SOX9 and RUNX2, during condensation of the skeletal anlagen. They are essential for chondrogenesis and osteogenesis, respectively, and their haploinsufficiency causes human skeletal dysplasias. We show that SOX9 directly interacts with RUNX2 and represses its activity via their evolutionarily conserved high-mobility-group and runt domains. Ectopic expression of full-length SOX9 or its RUNX2-interacting domain in mouse osteoblasts results in an osteodysplasia characterized by severe osteopenia and down-regulation of osteoblast differentiation markers. Thus, SOX9 can inhibit RUNX2 function in vivo even in established osteoblastic lineage. Finally, we demonstrate that this dominant inhibitory function of SOX9 is physiologically relevant in human campomelic dysplasia. In campomelic dysplasia, haploinsufficiency of SOX9 results in up-regulation of the RUNX2 transcriptional target COL10A1 as well as all three members of RUNX gene family. In summary, SOX9 is dominant over RUNX2 function in mesenchymal precursors that are destined for a chondrogenic lineage during endochondral ossification.

Keywords: differentiation, mesenchymal, skeletal dysplasias, osteoblasts, transcriptional repressor

During embryogenesis, the majority of bones are formed via endochondral ossification; mesenchymal progenitor cells differentiate into chondrocytes that are eventually replaced by osteoblasts (1, 2). It is a well coordinated process regulated by a complex transcriptional network in which the transcription factors Runx2 and Sox9 play essential roles. Runx2 is required for osteoblast differentiation and chondrocyte maturation both in vivo and in vitro (3). We and others have shown that mutations in RUNX2 cause cleidocranial dysplasia, a dominantly inherited skeletal dysplasia characterized by hypoplastic clavicles, large fontanels, dental anomalies, and delayed skeletal development (4, 5). Sox9 is a potent transcriptional activator for chondrocyte-specific genes such as Col2a1 and Col11a1, and mouse genetic studies demonstrate that it is required for the successive steps of chondrocyte differentiation and cartilage formation (68). Mutations in human SOX9 result in campomelic dysplasia (CMD1), a disorder characterized by generalized hypoplasia of endochondral bones (9, 10).

Although Runx2 is a strong transcriptional activator for osteoblast-specific and hypertrophic chondrocyte-specific genes, its embryonic expression is present in osteochondroprogenitor cells during mesenchymal condensations as early as embryonic day 10 (E10), before overt chondrocyte differentiation or osteoblast differentiation (11, 12). Hence, a strong context-dependent inhibition of Runx2 must occur before cell fate commitment to the chondrogenic lineage. Because Sox9 is also highly expressed in all osteochondroprogenitor cells and in proliferating (prehypertrophic) chondrocytes (6), we hypothesize that, in addition to its well established role as transcriptional activator for chondrogenesis, Sox9 also acts as a transcriptional repressor for osteoblast differentiation and chondrocyte hypertrophy in part via inhibition of Runx2 transactivation of its target genes.

Results

To test whether SOX9 had any effect on RUNX2 transcriptional activity, SOX9 and RUNX2 were cotransfected with a RUNX2-responsive Osteocalcin reporter plasmid 6xOSE2-luc into COS7 cells that do not express any Runx proteins (13, 14). Whereas RUNX2 alone activated this reporter >150-fold, cotransfection with SOX9 decreased its activity almost to basal levels in a dose-dependent manner (Fig. 1a). In a well differentiated osteoblast cell line that expresses high levels of endogenous Runx2, ROS 17/2.8, transfection of SOX9 led to a >5-fold decrease of an osteoblast-specific Osteocalcin reporter, OSC-114x2-luc, in a dose-dependent manner (Fig. 1b). To determine whether this inhibition extended to RUNX2 targets during chondrocyte maturation, SOX9 and RUNX2 were also cotransfected with a RUNX2-responsive type X collagen reporter, 8xA-Min-Col10a1-luc (15). Whereas RUNX2 alone transactivated this reporter by almost 150-fold, cotransfection with SOX9 resulted in repression of its activity back to basal levels (Fig. 1c). This effect was specific for RUNX2 because SOX9 had no effect on the activity of a chimeric transactivating construct pGal4-BD-NFκB (Fig. 1d). Therefore, these results demonstrate that SOX9 can specifically inhibit RUNX2 transactivation in vitro. Importantly, the converse is not true, and RUNX2 had no effect on SOX9's transactivation of its chondrocyte-specific type II collagen enhancer reporter in transient transfection studies (data not shown). This finding is consistent with the observation of normal initiation of chondrogenesis in Runx2-null mice (16, 17).

Fig. 1.
Potent inhibition of RUNX2 activity by SOX9. (a) COS7 cells were transfected with the RUNX2-responsive Osteocalcin reporter 6xOSE2-luc and expression plasmids for RUNX2 (100 ng) and SOX9 in increasing amounts (10 ng, 25 ng, 50 ng, and 100 ng). (b) ROS17/2.8 ...

To identify the domain within RUNX2 that is required for SOX9 inhibition of its activity, we performed cotransfection of SOX9 and two different RUNX2 deletion mutants, delQ/A and R391X (Fig. 2a). RUNX2 contains several domains: an N-terminal stretch of 23 consecutive polyglutamines followed by 17 polyalanines (Q/A domain); the runt domain; and a C-terminal proline/serine/threonine-rich activation domain (Fig. 2a). The delQ/A mutant has most of the Q/A domain deleted at its N terminus, whereas the R391X mutant is a previously characterized nonsense mutation in cleidocranial dysplasia. It contains a C-terminal deletion of the proline/serine/threonine-rich domain where numerous transcription factors have been reported to interact with RUNX2 (3, 18). As predicted, both deletions resulted in reduction of RUNX2 transactivation compared with full-length RUNX2. However, addition of SOX9 resulted in a further decrease of their activities to basal levels (Fig. 2a), suggesting that the Q/A domain and C terminus of RUNX2 do not mediate its repression by SOX9. Instead, the evolutionarily conserved runt domain of RUNX2 has the unique ability of mediating both DNA binding and protein–protein interaction (3). To investigate whether it is also responsible for SOX9 inhibition on RUNX2 transactivation, a chimera construct, RUNT-NFκB, containing the runt domain fused to the NF-κB transactivation domain was cotransfected with SOX9 into COS7 cells. Whereas RUNT-NFκB transactivated the RUNX2-responsive reporter 6xOSE2-luc >8-fold, addition of SOX9 reduced reporter activity back to its basal level (Fig. 2b). This strongly suggests that SOX9 inhibition of RUNX2 is mediated through the runt domain.

Fig. 2.
Direct interaction of SOX9 and RUNX2 through their DNA binding domains. (a) COS7 cells were transfected with 6xOSE2-luc reporter and expression plasmids for SOX9 and various RUNX2 deletion mutants. (b) COS7 cells were transfected with 6xOSE2-luc reporter ...

Next we performed a GST pull-down experiment to test for a physical interaction between SOX9 and the runt domain of RUNX2 (Fig. 2c). SOX9 contains a high-mobility-group (HMG)-type DNA binding domain that often also mediates protein–protein interaction, a proline/glutamine/alanine rich (P/Q/A) domain, and a potent transactivation domain (8). GST pull-down experiments demonstrated that SOX9-N containing its DNA binding HMG domain bound the runt domain almost as strongly as full-length SOX9, whereas SOX9-C containing its transactivation domain did not bind (Fig. 2c). This result showed that N-terminal half of SOX9 containing primarily the HMG domain directly interacted with the runt domain of RUNX2.

To test whether SOX9 had any effect on RUNX2 binding to its target sequence, we performed EMSA with SOX9 and RUNX2 proteins. Indeed SOX9 decreased RUNX2 binding to a well characterized consensus RUNX2 binding site located in the type X collagen promoter whereas control β-galactosidase had minimal effect (Fig. 2d) (15). Thus, SOX9 represses RUNX2 transactivation by directly reducing its DNA binding activity to target cis elements. Consistent with the GST pull-down data, it is SOX9-N but not SOX9-C that decreased RUNX2 binding to its target sequence (Fig. 2d).

To investigate the inhibitory effect of SOX9 on RUNX2 in vivo, we studied a model of established RUNX2 function during osteoblast differentiation. We used a well characterized osteoblast-specific 2.3-kb Col1a1 promoter to drive full-length SOX9 expression in osteoblasts in transgenic mice (19). The transgenic construct also included a tyrosinase minigene and a chicken β-globin HS4 insulator to enable rapid genotyping of embryonic founders by eye color selection and to enhance transgene expression by decreasing chromatin-mediated silencing effect (Fig. 3a) (20). A woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) was also included in the 3′ untranslated sequence to increase the expression level of the transgene (21). To confirm the specificity of the 2.3-kb Col1a1 promoter we first generated β-galactosidase transgenic mice using the same Col1a1-wpre-hGHpA transgenic vector. As previously reported, this promoter directed β-galactosidase expression specifically in osteoblasts (Fig. 3 b and c) (19). We then isolated 16 Col1a1-SOX9-FL transgenic founders that expressed the transgene and all of these mice displayed significant dwarfism (Fig. 3f). Transgene expression was confirmed by eye color selection, RT-PCR with RNA extracted from long bones and calvaria, and immunohistochemistry on hindlimb sections using a SOX9 antibody (Fig. 3 df). Most of the transgenic mice died within 2 postnatal weeks probably because of small rib cages, pulmonary hypoplasia and clefts of the secondary palate. von Kossa staining on long bone sections showed a much thinner and shorter mineralization zone in transgenic mice compared with wild-type mice, although growth plates in the transgenic mice appeared to be grossly normal (Fig. 3h). To assess the function of osteoblasts derived from transgenic mice, we established osteoblast cultures from E18.5 calvaria of wild-type and transgenic mice. Both alkaline phosphatase and von Kossa assays showed significantly decreased staining in transgene-expressing cells compared with wild type cells (Fig. 3g) demonstrating a defect in osteoblast function and mineralization. Runx2 has been shown to regulate all of the major genes expressed by osteoblasts in culture (11). Quantitative RT-PCR with RNA extracted from E18.5 long bones showed significantly decreased expression of osteoblast-related genes, such as Osteocalcin and type I collagen (Col1a1). As expected, no effect on the hypertrophic chondrocyte-specific marker Col10a1 was found because the transgene is not expressed in these cells where endogenous Runx2 is also active (Fig. 4a). These results indicate that transgene-expressing osteoblasts showed decreased functional activity as compared with wild-type osteoblasts.

Fig. 3.
SOX9 represses Runx2 function in established osteoblastic lineage in transgenic mice. (a) Schematic representation of Col1a1-SOX9-FL transgenic construct containing full-length SOX9 under the control of an osteoblast-specific 2.3-kb Col1a1 promoter in ...
Fig. 4.
Osteoblastic gene expression profile in Col1a1-SOX9-FL transgenic mice. (a) Real-time RT-PCR assay for osteoblast markers in hindlimbs from E18.5 wild-type (WT) and Col1a1-SOX9-FL transgenic (TG) mice. The data represent the mean ± SD from three ...

Runx2 is not only required for osteoblast differentiation during embryonic development, it also controls bone growth after birth (14). Therefore to further understand the inhibitory effect of SOX9 on RUNX2 in postnatal bone formation, we established two independent lines for Col1a1-SOX9-FL. Both lines exhibited very similar phenotypes. Although transgenic mice appeared grossly normal at birth, they quickly demonstrated growth disturbance by radiographic analysis as compared with wild type littermates (Fig. 5a). Histomorphometric analysis showed that the trabecular bone volume (BV/TV) was significantly reduced in transgenic mice (Fig. 5b). Furthermore, mineralization surface, mineral apposition rate, and bone formation rate were all significantly reduced in transgenic mice, consistent with its significantly decreased calcein labeling (Fig. 5 c and eg). At a cellular level, the numbers of characteristic cuboidal osteoblasts and their surface area were greatly reduced (Fig. 5 d, h, and i). As would be expected for an inhibitor of Runx2 function, osteocyte numbers were actually increased. The terminal differentiation of the osteoblast into the osteocyte is inhibited by Runx2 and inhibition of its function resulted in accelerated differentiation of osteoblasts (3) (Fig. 5j). RT-PCR assay showed that the osteoblast-specific markers downstream of Runx2 such as Col1a1, Osteopontin and Osteocalcin were all down-regulated, indicating that osteoblasts from Col1a1-SOX9-FL mice had severe functional defects (Fig. 4b). Interestingly, the phenotypes in Col1a1-SOX9-FL mice were very similar to those observed in Col1a1-Sox8 mice in which the expression of a highly related HMG protein Sox8 was driven by the same Col1a1 promoter element. Those mice also displayed impaired bone formation characterized by low bone mass, reduced number of osteoblasts, and decreased expressions of Runx2 and its downstream targets such as Col1a1 and Osteocalcin in osteoblasts (22).

Fig. 5.
Severe osteopenia in Col1a1-SOX9-FL transgenic mice. (a) Radiographic analysis of skeletons from wild-type and Col1a1-SOX9-FL transgenic mice at 4 weeks old. (b) von Kossa staining of undecalcified vertebral sections of wild-type (WT) and Col1a1-SOX9-FL ...

To address the possibility that the osteopenia phenotype in Col1a1-SOX9-FL transgenic mice was due to the nonspecific effects of misexpressing a strong transcription factor such as SOX9 in vivo, we generated two transgenic mice lines expressing only the N-terminal half of SOX9 (Col1a1-SOX9-N) containing the RUNX2-interaction domain using the same transgenic coat color vector. Because SOX9-N directly interacted with runt domain and decreased RUNX2's binding to its target sequence (Fig. 2 c and d), we would expect it to inhibit RUNX2 function similar to the full-length construct. Moreover, this would help to exclude potential nonspecific dominant negative effects of the C-terminal domain. Both Col1a1-SOX9-N lines appeared grossly normal at birth but quickly exhibited progressive dwarfism compared with wild-type littermates. Col1a1-SOX9-N line A exhibited an extremely severe osteopenic phenotype by both radiographic and histological analysis and none of F1 survived beyond 2 postnatal weeks (Fig. 6a and data not shown). The Col1a1-SOX9-N line B had a relatively milder phenotype but by 6 weeks they also showed dwarfism and osteopenia (Fig. 6 b and c). Bone histology and histomorphometry demonstrated that similar to Col1a1-SOX9-FL transgenic mice, Col1a1-SOX9-N transgenic mice had significantly reduced trabecular bone volume, decreased mineralization surface, mineral apposition rate, and bone formation rate. They also had fewer osteoblasts and reduced osteoblast surface area (Fig. 6c and data not shown). Furthermore, the very similar osteopenia phenotypes in Col1a1-SOX9-N and Col1a1-Sox8 transgenic mice strongly suggest that the conserved DNA-binding HMG domain of SOX9 plays an essential role in the inhibition of RUNX2 function.

Fig. 6.
Severe osteopenia in Col1a1-SOX9-N transgenic mice and RUNX2 target gene expression in CMD1 cartilage. (a) Radiographic analysis of skeletons from wild-type and Col1a1-SOX9-N transgenic line A mice at 2 weeks old. (b) Smaller size of Col1a1-SOX9-N transgenic ...

To correlate our transgenic mice misexpression data with a physiological scenario, we analyzed CMD1 cartilage, a severe skeletal dysplasia syndrome caused by heterozygosity for SOX9 mutations (9, 10), by quantitative RT-PCR analysis. We postulated that decreased SOX9 activity would lead to derepression of RUNX2 activity. Strikingly, whereas the expression of the chondrocyte-specific markers and known targets of SOX9 such as COL9A1, were down-regulated, expression of the hypertrophic chondrocyte-specific marker COL10A1, a known downstream target for RUNX2 in vivo, was increased by >20-fold (Fig. 6d) (15). Moreover, the expressions of all three members of RUNX gene family RUNX1, RUNX2, and RUNX3 were similarly increased by at least 2-fold in CMD1 cartilage (Fig. 6d). RUNX3 has been shown to play positive role during chondrocyte hypertrophy along with RUNX2 whereas RUNX1 has been postulated to mediate early events during mesenchymal condensations (12, 23, 24). At least in the case of RUNX2, autoregulation of its expression has been reported (14). Therefore, these data strongly suggest that haploinsufficiency of SOX9 leads to loss of inhibition of chondrocyte maturation markers and RUNX genes. More importantly because both HMG and runt domains are evolutionarily conserved among SOX and RUNX proteins, respectively, it is possible that not only SOX9, but perhaps other highly related HMG family members may also act as repressors for runt domain proteins during other developmental processes.

Discussion

As a key transcription activator for osteoblast differentiation and chondrocyte maturation, it is likely that Runx2 function is regulated by diverse cofactors during different developmental stages for proper bone and cartilage formation. It has been previously reported that the basic helix–loop–helix transcription factor Twist and histone deacetylase HDAC4 can directly interact with Runx2 and inhibit its activity during intramembranous ossification and chondrocyte maturation, respectively, both in vivo and in vitro (25, 26). In this study we identify a transcriptional repressor function of SOX9 on RUNX2 by biochemical, tissue culture, mouse genetics, and human genetic approaches that may act both early during chondrogenic cell fate commitment and later during chondrogenesis. Because of their coexpression during mesenchymal condensation, these data support the dominance of SOX9 function over RUNX2 during this early first step in the progenitor cell fate decision between osteoblastic vs. chondrogenic lineages. It has previously been shown by retroviral injection experiments in chicken embryos that Sox9 misexpression repressed Runx2 function and diverted cell fate from bone to cartilage in the craniofacial region (27). In a Sox9 knockin mouse model in which Sox9 was overexpressed in proliferating chondrocytes during endochondral ossification, osteoblast differentiation was also delayed (28). Furthermore, in Sox9+/− mice the hypertrophic zone was enlarged as assayed by the expression of Col10a1 (7), consistent with our observation that COL10A1 level was greatly up-regulated in CMD1 cartilages. Interestingly, recent mouse genetic studies showed that Sox9-expressing precursor cells could eventually differentiate into tendons and osteoblasts (29). Therefore, the inhibitory effect of Sox9 on osteoblastic and chondrocyte maturation via repression Runx2 function is an essential mechanism for osteochondroprogenitor cell fate determination.

Materials and Methods

DNA Transfections.

Transfection, luciferase assays, and β-galactosidase assays were performed as previously described (18). All transfections were performed with pSV2βgal as an internal control for transfection efficiency. All data are presented as fold activation or percentage relative to the activity obtained with the pcDNA3.1 empty vector plasmid. Bars represent the average ratios of luciferase to β-galactosidase activity. The standard deviations obtained from three independent transfections of one representative experiment are represented by the error bars.

Recombinant Proteins, GST Pull-Down Assays, and EMSA.

GST-RUNT and GST proteins were expressed in BL21-CodonPlus cells (Stratagene) and purified by using bulk GST purification modules according to the manufacturer's protocol (Amersham Bioscience, Piscataway, NJ). 35S-labeled SOX9 proteins were generated by a T7 in vitro transcription/translation kit (Novagen, Madison, WI) and incubated with GST or GST-RUNT immobilized on glutathione-Sepharose beads at 4°C for 2 h. The beads were then washed three times with TNN buffer containing 1% Nonidet P-40, boiled in 2× SDS sample loading buffer, separated by SDS/PAGE, and autoradiographed. Labeling of oligonucleotide probes, incubation of in vitro translated proteins, and EMSA were performed as previously described (15).

DNA Constructs, Transgenic Constructs, Skeletal Preparation, and Immunohistochemistry.

The full-length SOX9, its N terminus (SOX9-N, amino acids 1–304), and its C terminus (SOX9-C, amino acids 304–509) were cloned into 5′UT-FLAG-pcDNA3.1(+) expression vector between BamHI and EcoRI sites. SOX9 full-length, SOX9-N, and β-galactosidase were cloned under control of a 2.3-kb osteoblast-specific promoter in a coat color vector containing tyrosinase minigene and the WPRE posttranscriptional sequences. Transgenic founders were generated by pronuclear injection according to standard techniques (15). All transgenic lines were maintained on a FVB/N background. The transgenic mice were identified at birth by eye color change and confirmed by PCR by using specific primers for the WPRE element. Immunohistochemistry was performed on paraffin sections of newborn hindlimbs with goat anti-SOX9 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), and ABC reagent (Vector Laboratories, Burlingame, CA) was used as secondary antibodies for detection.

Skeletal Analyses and Bone Histomorphometry.

Skeletons from 1-week-old mice were prepared as described and stained with alcian blue 8GX for cartilage and alizarin red S for bone (14). For radiographic and histological analysis, mice were killed, internal organs were removed, and the whole skeletons were fixed in 4% PBS-buffered paraformaldehyde for 18 h at 4°C. The skeletons were analyzed by contact radiography with a Faxitron x-ray cabinet (Faxitron X-Ray, Wheeling, IL). Five-millimeter and 7-μm undecalcified lumbar vertebrate plastic sections were stained by toluidine blue and von Kossa, respectively, per standard protocol (14). For fluorochrome measurement of bone formation rate, mice were injected with calcein at 9 and 2 days before they were killed, and nonstained 5-mm sections were used for analysis. All static histomorphometry analyses were performed according to standard protocols by using the OsteoMeasure histomorphometry system (OsteoMetrics, Decatur, GA).

Ex Vivo Osteoblast Culture, RNA Analysis, and Real-Time RT-PCR.

Osteoblasts from individual calvaria of E18.5 transgenic founders or wild-type littermates were isolated as previously described (14). Osteoblast differentiation was induced in αMEM containing 10% FBS, 50 μg/ml ascorbic acid, and 10 mM β-glycerophosphate. Three animals were analyzed independently per genotype. Total RNA was extracted by using TRIzol reagent (Invitrogen, Carlsbad, CA) from hindlimbs of E18.5 mouse embryos or long bones from 6-week-old mice that were dissected free of surrounding tissues with epiphyses and metaphyses removed. RNA extraction, cDNA synthesis, and PCR amplification were performed with specific primers for 6-week-old long bone samples by using standard protocols. Amplification of Hprt gene exon 2 was used as internal control for the quantity and quality of cDNAs. Real-time quantitative PCR amplifications were performed on LightCycler (Roche Applied Science, Indianapolis, IN) with cDNAs derived from newborn hindlimbs and analyzed as described previously (15). The Gapdh (Fig. 4a) and Hprt (data not shown) genes were used as internal controls for the quantity and quality of the cDNAs in all real-time PCR assays, and similar results were obtained. Analyses of real-time data were as previously described (15).

Analysis of CMD1 Cartilage.

SOX9 heterozygous deletion in CMD1 was confirmed by FISH on cells derived from samples with SOX9-specific probe. Growth plates RNA was extracted from CMD1 fetal cartilage and an age-matched control, and quantitative real-time RT-PCR analysis was performed on these samples as described previously (30).

Statistical Analyses.

Statistical analyses were carried out by using Student's paired t test. The transgenic mice were tested for difference from the age-matched wild-type group with n = 3–5. Values were considered statistically significant at P < 0.05.

Acknowledgments

We thank B. de Crombrugghe (University of Texas M. D. Anderson Cancer Center, Houston, TX) for Col1a1 promoter plasmid and 5′UT-FLAG-pcDNA3.1(+) expression vector; T. F. Tsai (National Yang-Ming University, Taipei, Taiwan) for Tyr-HS4 coat color vector; and P. Fonseca, M. M. Jiang, and T. Bertin for technical assistance. This work was supported by National Institutes of Health Grants DE15139 (to G.Z.), ES11253 (to B.L.), HD22657 (to B.L. and D.K.), and DE16990 (to B.L.) and the Baylor College of Medicine Mental Retardation and Developmental Disability Research Center (B.L.).

Abbreviations

CMD 1
campomelic dysplasia
HMG
high-mobility-group
En
embryonic day n

Footnotes

The authors declare no conflict of interest.

This article is a PNAS direct submission.

References

1. Karsenty G, Wagner EF. Dev Cell. 2002;2:389–406. [PubMed]
2. Nakashima K, de Crombrugghe B. Trends Genet. 2003;19:458–466. [PubMed]
3. Schroeder TM, Jensen ED, Westendorf JJ. Birth Defects Res C Embryo Today. 2005;75:213–225. [PubMed]
4. Mundlos S, Otto F, Mundlos C, Mulliken JB, Aylsworth AS, Albright S, Lindhout D, Cole WG, Henn W, Knoll JH, et al. Cell. 1997;89:773–779. [PubMed]
5. Lee B, Thirunavukkarasu K, Zhou L, Pastore L, Baldini A, Hecht J, Geoffroy V, Ducy P, Karsenty G. Nat Genet. 1997;16:307–310. [PubMed]
6. Akiyama H, Chaboissier MC, Martin JF, Schedl A, de Crombrugghe B. Genes Dev. 2002;16:2813–2828. [PMC free article] [PubMed]
7. Bi W, Huang W, Whitworth DJ, Deng JM, Zhang Z, Behringer RR, de Crombrugghe B. Proc Natl Acad Sci USA. 2001;98:6698–6703. [PMC free article] [PubMed]
8. Lefebvre V, Huang W, Harley VR, Goodfellow PN, de Crombrugghe B. Mol Cell Biol. 1997;17:2336–2346. [PMC free article] [PubMed]
9. Foster JW, Dominguez-Steglich MA, Guioli S, Kowk G, Weller PA, Stevanovic M, Weissenbach J, Mansour S, Young ID, Goodfellow PN, et al. Nature. 1994;372:525–530. [PubMed]
10. Wagner T, Wirth J, Meyer J, Zabel B, Held M, Zimmer J, Pasantes J, Bricarelli FD, Keutel J, Hustert E, et al. Cell. 1994;79:1111–1120. [PubMed]
11. Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G. Cell. 1997;89:747–754. [PubMed]
12. Smith N, Dong Y, Lian JB, Pratap J, Kingsley PD, van Wijnen AJ, Stein JL, Schwarz EM, O'Keefe RJ, Stein GS, Drissi MH. J Cell Physiol. 2005;203:133–143. [PubMed]
13. Ducy P, Karsenty G. Mol Cell Biol. 1995;15:1858–1869. [PMC free article] [PubMed]
14. Ducy P, Starbuck M, Priemel M, Shen J, Pinero G, Geoffroy V, Amling M, Karsenty G. Genes Dev. 1999;13:1025–1036. [PMC free article] [PubMed]
15. Zheng Q, Zhou G, Morello R, Chen Y, Garcia-Rojas X, Lee B. J Cell Biol. 2003;162:833–842. [PMC free article] [PubMed]
16. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, et al. Cell. 1997;89:755–764. [PubMed]
17. Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC, Rosewell IR, Stamp GW, Beddington RS, Mundlos S, Olsen BR, et al. Cell. 1997;89:765–771. [PubMed]
18. Zhou G, Chen Y, Zhou L, Thirunavukkarasu K, Hecht J, Chitayat D, Gelb BD, Pirinen S, Berry SA, Greenberg CR, et al. Hum Mol Genet. 1999;8:2311–2316. [PubMed]
19. Rossert J, Eberspaecher H, de Crombrugghe B. J Cell Biol. 1995;129:1421–1432. [PMC free article] [PubMed]
20. Hsiao YC, Chang HH, Tsai CY, Jong YJ, Horng LS, Lin SF, Tsai TF. Genesis. 2004;39:122–129. [PubMed]
21. Zufferey R, Donello JE, Trono D, Hope TJ. J Virol. 1999;73:2886–2892. [PMC free article] [PubMed]
22. Schmidt K, Schinke T, Haberland M, Priemel M, Schilling AF, Mueldner C, Rueger JM, Sock E, Wegner M, Amling M. J Cell Biol. 2005;168:899–910. [PMC free article] [PubMed]
23. Yoshida CA, Yamamoto H, Fujita T, Furuichi T, Ito K, Inoue K, Yamana K, Zanma A, Takada K, Ito Y, Komori T. Genes Dev. 2004;18:952–963. [PMC free article] [PubMed]
24. Yamashiro T, Wang XP, Li Z, Oya S, Aberg T, Fukunaga T, Kamioka H, Speck NA, Takano-Yamamoto T, Thesleff I. J Bone Miner Res. 2004;19:1671–1677. [PubMed]
25. Bialek P, Kern B, Yang X, Schrock M, Sosic D, Hong N, Wu H, Yu K, Ornitz DM, Olson EN, et al. Dev Cell. 2004;6:423–435. [PubMed]
26. Vega RB, Matsuda K, Oh J, Barbosa AC, Yang X, Meadows E, McAnally J, Pomajzl C, Shelton JM, Richardson JA, et al. Cell. 2004;119:555–566. [PubMed]
27. Eames BF, Sharpe PT, Helms JA. Dev Biol. 2004;274:188–200. [PubMed]
28. Akiyama H, Lyons JP, Mori-Akiyama Y, Yang X, Zhang R, Zhang Z, Deng JM, Taketo MM, Nakamura T, Behringer RR, et al. Genes Dev. 2004;18:1072–1087. [PMC free article] [PubMed]
29. Akiyama H, Kim JE, Nakashima K, Balmes G, Iwai N, Deng JM, Zhang Z, Martin JF, Behringer RR, Nakamura T, de Crombrugghe B. Proc Natl Acad Sci USA. 2005;102:14665–14670. [PMC free article] [PubMed]
30. Zheng Q, Sebald E, Zhou G, Chen Y, Wilcox W, Lee B, Krakow D. Am J Hum Genet. 2005;77:305–312. [PMC free article] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...