• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Dev Cell. Author manuscript; available in PMC Mar 13, 2013.
Published in final edited form as:
PMCID: PMC3306603

Sox9 Directs Hypertrophic Maturation and Blocks Osteoblast Differentiation of Growth Plate Chondrocytes


The transcription factor Sox9 is necessary for early chondrogenesis, but its subsequent roles in the cartilage growth plate, a highly specialized structure that drives skeletal growth and endochondral ossification, remain unclear. Using a doxycycline-inducible Cre transgene and Sox9 conditional null alleles in the mouse, we show that Sox9 is required to maintain chondrocyte columnar proliferation and generate cell hypertrophy, two key features of functional growth plates. Sox9 keeps Runx2 expression and β-catenin signaling in check, and thereby inhibits not only progression from proliferation to prehypertrophy, but also subsequent acquisition of an osteoblastic phenotype. Sox9 protein outlives Sox9 RNA in upper hypertrophic chondrocytes, where it contributes with Mef2c to directly activate the major marker of these cells, Col10a1. These findings thus reveal that Sox9 remains a central determinant of the lineage fate and multi-step differentiation program of growth plate chondrocytes, and thereby illuminate our understanding of key molecular mechanisms underlying skeletogenesis.


The skeleton is critically important in vertebrates. It forms the body framework, assists in locomotion, and fulfills key physiological functions. It is subject to prevalent inherited and acquired diseases in humans, many of which remain poorly treatable (Woolf and Pfleger, 2003; Rimoin et al., 2007). To help find efficient cures, it is thus essential that we reach fuller understanding of the basis of the skeleton complex structure and regulation, and in particular the mechanisms underlying its development.

Building the vertebrate skeleton consists in generating two main tissues, cartilage and bone, at the right time, in over two hundred locations in the embryo, and subsequently ensuring their proper growth and maturation (Provot and Schipani, 2005). These hard connective tissues greatly differ in composition, function, and regulation, but develop through closely related, mutually interacting processes. Chondrocytes (cartilage-forming cells) and osteoblasts (bone-making cells) derive from osteochondroprogenitors, bipotent cells that arise from multipotent mesenchyme. Skull vault bones and other flat bones form intramembranously, through direct differentiation of osteochondroprogenitors into osteoblasts. Other bones form endochondrally, that is, through a cartilage intermediate. Osteochondroprogenitors first condense into precartilaginous masses. Inner cells differentiate into early chondrocytes and perichondrium cells remain uncommitted. Early chondrocytes then undergo further steps of differentiation in a staggered manner, establishing structures referred to as growth plates in reference to their prime contribution to skeletal elongation. Chondrocytes flatten, stack in longitudinal columns, and proliferate actively until reaching the prehypertrophic stage. This stage is a major step in skeletogenesis because the cells switch to a mature phenotypic program and induce osteogenesis in the perichondrium. Following subsequent hypertrophy, another key step in skeleton elongation, chondrocytes terminally mature and most if not all undergo apoptosis. Osteoblast precursors, endothelial cells, osteoclasts, and hematopoietic cells migrate from the perichondrium into the cartilage remnant to remodel the tissue and lay down bone and marrow (Maes et al., 2010). Each chondrocyte and osteoblast developmental stage is characterized by expression of specific genes (Lefebvre and Smits, 2005). Typical markers include Col2a1 (collagen 2) and Agc1 (aggrecan) for early chondrocytes; Fgfr3 (fibroblast growth factor receptor 3) for columnar cells; Ppr (parathyroid hormone-related protein receptor), Ihh (Indian hedgehog) and Col10a1 (collagen 10) for prehypertrophic cells; and Col10a1 only for hypertrophic cells. Terminal chondrocytes express Mmp13 (matrix metalloproteinase 13) and Bsp (bone sialoprotein), and mineralize the extracellular matrix, as do mature osteoblasts, whereas early osteoblasts express Osx (Osterix) and Col1a1 (collagen 1).

Like other developmental processes, skeletogenesis is spatially and temporally governed by intricate networks of regulatory molecules, among which lineage-specific transcription factors have key fate-determining roles (Karsenty et al., 2009). The Sry-related transcription factor Sox9 is one of them (Akiyama, 2008). Research on its functions started when SOX9 heterozygous mutations were found to cause campomelic dysplasia (CD), a severe form of dwarfism affecting all cartilage and endochondral structures (Foster et al., 1994; Wagner et al., 1994). Sox9 expression is turned on in mesenchymal precursors, maintained in developing chondrocytes until prehypertrophy, but turned off in other lineages. Sox9 is absolutely necessary for chondrocyte specification and early differentiation (Bi et al., 1999; Akiyama et al., 2002). It directly activates all major cartilage-specific extracellular matrix genes expressed by early chondrocytes, and is helped in this function by two distant relatives, Sox5 and Sox6 (Lefebvre and Smits, 2005). The three Sox proteins are needed and sufficient for early chondrogenesis, and thus referred to as the chondrogenic trio (Ikeda et al., 2004). Subsequent differentiation of chondrocytes is directed from the prehypertrophic stage by the Runt domain transcription factors Runx2 and Runx3, and by MADS box transcription factors, mainly Mef2c (Takeda et al., 2001; Yoshida and Komori, 2005; Arnold et al., 2007). Runx2 is also necessary for osteoblast specification and differentiation (Ducy et al., 1997; Komori et al., 1997; Otto et al., 1997), along with the zinc finger transcription factor Osx (Nakashima et al., 2002).

Strong expression of Sox9 in growth plate chondrocytes until prehypertrophy and marked shortening of campomelic dysplasia growth plates strongly suggest that Sox9 has important roles in growth plates. These roles, however, remain unclear. Sox9 was first proposed to inhibit chondrocyte proliferation and hypertrophy (Akiyama et al., 2002 and 2004), but was more recently proposed to be necessary for chondrocyte survival and hypertrophy, and to delay terminal maturation (Hattori et al., 2010; Ikegami et al., 2011). Some of the data in these previous studies were difficult to interpret because the mouse transgenes that were used to inactivate or overexpress Sox9 were active from the precursor or early chondrocyte stage, causing defects in cartilage primordia that precluded definitive identification of growth plate-specific roles for Sox9. To solve this problem and clarify the roles of Sox9 in the growth plate, we used in this study mice harboring Sox9 conditional null alleles and a Cre transgene inducible in differentiated growth plate chondrocytes. We show that Sox9 continues to fulfill essential roles at several stages of differentiation of these cells to ensure cartilage-mediated skeletal growth and coordinate this process with endochondral ossification.


Generation of a Cre transgene inducible in differentiated chondrocytes

We previously showed that an Agc1 (aggrecan) upstream enhancer was sufficient to activate the Col2a1 promoter in differentiated chondrocytes in transgenic mice (Han and Lefebvre, 2008). Here, we cloned these regulatory elements into a bigenic template (Utomo et al., 1999) to generate a mouse line expressing an Agc1 enhancer-driven, tetracycline-inducible Cre (ATC) transgene (Fig. S1A). We characterized transgene activity using a dual R26TG Cre reporter (Muzumdar et al., 2007). This reporter expresses Tomato ubiquitously before Cre recombination and GFP following recombination. R26TGATC fetuses at gestation day 17.5 (E17.5) showed Cre activity in few cells in the end of growth plates, nuclei pulposi and bone in absence of tetracycline (Fig. S1B–D). When their mothers drank water containing the tetracycline compound doxycycline (Dox) from E15.5, they showed Cre-mediated recombination within two days in all differentiated chondrocytes (except in epiphyseal lateral sides) and nucleus pulposus cells, and in some myoblasts and bone cells, but none in perichondrium cells and other cell types (Fig. S1B–E). We concluded that ATC should be an excellent model to study gene functions in growth plate chondrocytes independently of functions in precursors and most other cell types.

Sox9 is necessary to maintain functional growth plates

To determine if Sox9 has specific roles in the growth plate, we bred females carrying Sox9 conditional null alleles (Sox9fl/fl; Kist et al., 2002) with Sox9fl/+ATC males, treated them with Dox from E15.5, and analyzed fetuses daily afterwards. Sox9 RNA remained abundant in non-hypertrophic chondrocytes of Sox9fl/fl and Sox9fl/+ control fetuses through E18.5, but it was lost in most growth plate cells of Sox9fl/flATC mutants by E17.5 (Fig. 1A). Control epiphyses and growth plates kept the same length over time while primary ossification centers elongated, reflecting balanced turnover of cartilage, and Sox9fl/+ATC growth plates remained virtually normal (Fig. 1A–B). In contrast, columnar zones shortened in Sox9fl/flATC growth plates, chondrocytes lost the ability to enlarge, and endochondral bones stopped elongating by E17.5, resulting in severe dwarfism (Fig. 1A–C). These data thus revealed that Sox9 is essential to maintain functional growth plates.

Figure 1
Sox9 is required to maintain functional growth plates

Sox9 is needed to maintain growth plate chondrocyte proliferation and viability, delay prehypertrophy, and allow hypertrophy

Shortening of Sox9 mutant columnar zones could result from chondrocyte slow proliferation, untimely death, or precocious maturation. In BrdU incorporation assays performed after two days of fetus treatment with Dox, growth plate chondrocytes from Sox9fl/flATC fetuses showed normal proliferation rates in epiphyses, but columnar cells growth arrested about three times as fast as control cells (Fig. 2A). This result was confirmed by immunostaining for Ki-67 (Fig. S2A). In TUNEL assay, Sox9fl/fl growth plates never showed a significant rate of chondrocyte death until the cells reached the ossification front (Fig. 2B). In contrast, Sox9fl/flATC growth plates showed 0.1% of dying cells in the columnar zone and 4.5% in the prehypertrophic zone after two days on Dox, and 2.8 and 24.8% in the same respective zones by the next day. Immunoreactivity for cleaved caspase 3 indicated that mutant cell death occurred through apoptosis (Fig. S2B). Thus, Sox9 deletion led to premature growth arrest of columnar cells, followed by apoptosis. Consistent with precocious prehypertrophy, Sox9fl/flATC chondrocytes turned on Ppr and Ihh concomitantly with growth-arrest (Fig. 2C). Surprisingly, however, they failed to activate the hypertrophic markers Col10a1, Bmp6 and Has2, but ultimately expressed the terminal chondrocyte and osteoblast marker Mmp13 (Fig. 2D and Fig. S2C). They expressed the chondrocyte proliferation inhibitor Fgfr3 at a normal level, and chondrocyte maturation inhibitors Pthrp (parathyroid hormone related peptide) and Ptc (Patched) at a high level (Fig. 2E). They strongly expressed Runx2 and Mef2c, which encode transcription factors required for chondrocyte maturation, and Hdac4, encoding a histone deacetylase inhibitor of these factors (Vega et al., 2004), indicating that Sox9 may delay prehypertrophy by downregulating Runx2 and Mef2c, but must control hypertrophy differently. Overall, these data thus showed that Sox9 is required to maintain growth plate chondrocytes proliferating, delay prehypertrophy, and allow hypertrophy before terminal maturation and apoptosis.

Figure 2
Sox9 is required for chondrocyte proliferation, survival, and hypertrophy, but delays prehypertrophy

Sox9 contributes with Mef2c to directly activate Col10a1

Based on the facts that Col10a1 is expressed exclusively in prehypertrophic and hypertrophic chondrocytes and that Sox9 is required for hypertrophy, we investigated whether Sox9 directly controls Col10a1. Prerequisite to this function, Sox9 protein must be present in Col10a1-expressing chondrocytes. By RNA in situ hybridization, we confirmed (Lefebvre et al., 1998) that Sox9 and Col10a1 RNAs overlap each other in prehypertrophic cells, and by immunostaining we discovered that the Sox9 protein outlives its RNA and remains nuclear in upper hypertrophic chondrocytes (Fig. 3A and B). Concomitant loss of Sox9 RNA and protein in Dox-treated Sox9fl/flATC growth plates ascertained antibody specificity. Thus, Sox9 protein is present in cells activating Col10a1.

Figure 3
Sox9 protein outlives its RNA in upper hypertrophic chondrocytes and binds next to Mef2c on the Col10a1 promoter

To test whether Sox9 could transactivate Col10a1, we searched for putative Sox9 binding sites in the Col10a1 promoter and hypertrophic enhancer (Zheng et al., 2009). We found only one sequence that was evolutionarily conserved and matched a Sox9 binding site, i.e., a pair of sites resembling the Sox consensus CA/TTTGA/TA/T, oriented head to head, and separated by 3 to 5 nucleotides (Sock et al., 2003). This sequence was located at −186/−169 (Fig. 3C and D), near a Mef2c binding site (Arnold et al., 2007). Electrophoretic mobility shift assays showed the ability of each protein to bind its site (Fig. 3E), but no evidence of cooperative binding (Fig. 3F). Chromatin immunoprecipitation assays revealed efficient binding of both proteins to the Col10a1 promoter in primary growth plate chondrocytes forced to upregulate Col10a1 by treatment with okadaic acid, a potent inhibitor of Pp2a phosphatase (Kozhemyakina et al., 2009) (Fig. 3G and H).

To test whether Sox9 binding to the Col10a1 promoter resulted in transactivation, we constructed Col10a1/βgeo/EGFP reporters (Fig. S3A). As expected, the activity of the Col10a1 promoter was upregulated in primary chondrocytes when the reporters contained the Col10a1 hypertrophic enhancer and the cells were treated with okadaic acid (Fig. 4A). Forced expression either human SOX9 or mouse Mef2c enhanced transactivation, and forced expression of both proteins resulted in additive or slightly stronger effects. The proteins acted similarly in nonchondrocytic 293T cells, even though the enhancer was virtually inactive in these cells (Fig. S3B). In agreement with the protein binding locations, promoter mutagenesis experiments revealed that SOX9 and Mef2c largely mediated their activity through the Col10a1 −210/−94 sequence (Fig. S3C). The mutation of either protein site reduced reporter activity in chondrocytes in absence and presence of exogenous protein, and the mutation of both sites abolished reporter activity (Fig. 4B). Taken together, these results support the concept that Sox9 and Mef2c transactivate Col10a1, and act additively.

Figure 4
Sox9 and Mef2c act cooperatively to activate Col10a1

To test this concept in vivo, we generated Sox9/Mef2c compound mutants. Single and double heterozygotes obtained using Prx1Cre, a transgene active in limb bud mesenchyme, including chondrocyte precursors (Logan et al., 2002), displayed no major histological defects at E18.5 (Fig. 4C). Col10a1 RNA level was slightly but not significantly decreased in single heterozygotes, but was significantly reduced by 2.4-fold in double heterozygotes. Further, and consistent with the reported roles of Mef2c (Arnold et al., 2007), Mef2cfl/flATC fetuses on Dox since E15.5 showed greatly expanded growth plates by E17.5 and severely delayed Col10a1 activation (Fig. 4D). Sox9fl/+Mef2cfl/flATC littermates had similar histological defects, but virtually failed to express Col10a1. Thus Sox9 and Mef2c also act together to activate Col10a1 in vivo. Interestingly, Mef2cfl/flATC mutants initiated Ppr and Runx2 expression in a timely manner and delayed Ppr deactivation and Ihh activation, whereas Sox9fl/+Mef2cfl/flATC littermates exhibited stronger and earlier activation of Ppr, and a partial rescue of the Ihh activation delay. Thus, Sox9 inhibits chondrocyte prehypertrophy from the Ppr expression stage, Mef2c activates it from the Ihh expression stage, and both proteins are then required for chondrocyte hypertrophy and Col10a1 expression.

Sox9 prevents osteoblastic differentiation of prehypertrophic chondrocytes

Besides Col10a1 transactivation, we considered the possibility that Sox9 could use additional, indirect mechanisms to account for its absolute requirement for chondrocyte hypertrophy. Since Runx2 is necessary for osteoblast differentiation and chondrocyte maturation, its inability to cause hypertrophy of Sox9-deficient chondrocytes, despite being strongly expressed, made us wonder whether Sox9fl/flATC prehypertrophic chondrocytes were switching lineage. After two days on Dox, control and mutant prehypertrophic cells, as identified by expression of their exclusive marker, Ihh, still contained RNA for Col2a1, but none for Sox9, Sox5 and Sox6 (Fig. 5A and data not shown). Since the Sox trio is required for Col2a1 transcription, the presence of Col2a1 RNA in these cells was unlikely due to recent production, but more likely due to earlier production of this slowly decaying RNA (Murakami et al., 2000). Importantly, this result attested of the chondrocyte origin of the cells. Strikingly, mutant prehypertrophic cells upregulated not only Runx2, but also Osx, required for osteoblast differentiation. Osx RNA is normally present at an extremely low level in prehypertrophic chondrocytes compared to osteoblasts, but Sox9-deficient chondrocytes contained as much Osx RNA as periosteum and endochondral bone osteoblasts. Further, they contained low levels of Col1a1 RNA, which is abundant in osteoblasts but absent in chondrocytes, and Bsp RNA (bone sialoprotein), a marker of osteoblasts and terminal chondrocytes. Sox9fl/flATC prehypertrophic chondrocytes were thus transitioning into early osteoblasts within two days on Dox. One day later, these mutant cells still contained Col2a1 RNA, but no longer contained Ihh RNA, and they were expressing Col1a1 and Bsp RNA as strongly as wild-type osteoblasts (Fig. 5B). Furthermore, they were strongly expressing genes involved in matrix mineralization, such as Alpl (alkaline phosphatase) and Ank (ankylosis), and were surrounded by mineralized matrix (Fig. 5C). They had thus completed differentiation into osteoblasts and reached the mature stage in this lineage. Confirming conversion into a single lineage, Sox9fl/flATC growth plate cells treated for 3 days with Dox were negative for markers of other mesenchyme-derived cell types, including Gdf5 (joint precursors), MyoD (myoblasts), Pparg (adipocytes), and Scx (tenocytes) (Fig. S4). We concluded that Sox9 is necessary to maintain the lineage fate of prehypertrophic cells and prevent acquisition of an osteoblastic phenotype.

Figure 5
Sox9 prevents osteoblastic differentiation of chondrocytes

Sox9 secures the fate of growth plate chondrocytes by reducing β-catenin activity

Sox9 and β-catenin oppose each other in osteochondroprogenitors to specify the chondrocyte and osteoblast lineages, respectively (Day et al., 2005; Hill et al., 2005). We thus considered the possibility that Sox9 blocks expression of an osteoblastic phenotype by growth plate chondrocytes by opposing β-catenin signaling. We observed that Sox9fl/flATC prehypertrophic chondrocytes on Dox for two days were avidly expressing Tcf1 (T cell-specific transcription factor 1) and Ccnd1 (cyclin D1), two targets of β-catenin signaling very active in osteoblasts (Fig. 6). Inactivation of β-catenin along with Sox9 in Sox9fl/flCtnnbfl/flATC fetuses resulted in repression of these targets and partial downregulation of Osx. Importantly, it prevented Col1a1 and Bsp activation, thus osteoblastic differentiation, but did not prevent precocious activation of Ihh and upregulation of Runx2, and did not rescue cell hypertrophy and Col10a1 expression (Fig. 6 and Fig. S5A). Thus, Sox9 blocks expression of an osteoblast phenotype by chondrocytes at least in part by suppressing β-catenin signaling, but does not use this mechanism to keep Runx2 expression low and permit hypertrophy. In a complementary experiment we asked whether forced upregulation of β-catenin signaling in growth plate chondrocytes would suffice to change the fate of the cells. For this purpose, we generated ATC fetuses harboring CtnnbflEx3, an allele that allows constitutive activation of β-catenin (Harada et al., 1999). The growth plates of these fetuses exhibited expanded hypertrophic zones, but looked otherwise normal after two days on Dox (Fig. S5B). Strong expression of Tcf1 and Ccnd1 in columnar and prehypertrophic chondrocytes proved upregulation of β-catenin signaling (Fig. S5C). Ihh and Runx2 were upregulated and their expression domain was broadened, as was that of Col10a1. Osx was slightly expressed, but Col1a1 remained silent. Not surprisingly, therefore, mutant chondrocytes still contained high levels of Sox9 RNA and protein (Fig. S5D). Thus, upregulation of β-catenin signaling in a Sox9 wild-type context promotes chondrocyte maturation, as previously shown (Tamamura et al., 2005; Guo et al., 2009), but is insufficient to induce osteoblastic differentiation.

Figure 6
Sox9 inhibits β-catenin signaling, but β-catenin signaling does not inhibit Sox9 in the growth plate


This study clarified and provided novel insights on the roles of Sox9 in the growth plate (Fig. 7). It showed that Sox9 maintains columnar proliferation, delays prehypertrophy and then prevents osteoblastic differentiation of chondrocytes by lowering β-catenin signaling and Runx2 expression. Further, Sox9 is required for chondrocyte hypertrophy, both indirectly by keeping the lineage fate of chondrocytes, and directly by remaining present in upper hypertrophic cells and transactivating Col10a1 along with Mef2c. It is thus essential to form and maintain functional growth plates.

Figure 7
Proposed roles of Sox9 in the growth plate

Most of our findings were made using an ATC conditional deletion system, which allows for precise temporal and spatial deletion in differentiated chondrocytes upon induction by doxycycline. The drug was delivered for several days without adverse effects, whereas tamoxifen, used for many inducible transgenes, is often abortive in pregnancy. ATC was instrumental to uncover growth plate-specific roles of Sox9, independently of roles in precursor, perichondrium, and joint cells, and should be equally useful to study the roles of many other genes in differentiated chondrocytes.

In contrast to inhibition of chondrocyte proliferation, inferred from the analysis of fetuses overexpressing SOX9 from the Col2a1 locus (Akiyama et al., 2004), we found that Sox9 sustains columnar cell proliferation. Akiyama and colleagues compared cell proliferation in control and mutant fetuses using an average value for entire cartilage elements, without considering differences in cell developmental stages that existed between controls and mutants. We measured cell proliferation in discrete areas succeeding each other from the top to the end of the growth plate to account for cell differentiation changes, and we did this in established growth plates promptly after Sox9 inactivation. This precise assay pinpointed to premature growth arrest of columnar cells, coincident with prehypertrophy, as reported in Sox5/6 mutants (Smits et al., 2004). Since the Sox trio transactivates cartilage matrix genes, it may influence chondrocyte destiny via regulatory events taking place in this matrix. In addition, it may interact with Ihh or Pthrp signaling, which control transition from proliferation to prehypertrophy (Karsenty et al., 2009). Supporting this idea, Sox9 is a target of Pthrp signaling (Huang et al., 2001) and pathway components are normal or upregulated in Sox9 and Sox5/6 mutants.

We observed extensive apoptosis in Sox9-deficient growth plates, but only three days after Sox9 inactivation when the cells had converted into mature osteoblasts. This cell fate change and time lapse argue against a direct role for Sox9 in chondrocyte survival. This conclusion contrasts with a recent report by Ikegami and colleagues (2011) that Sox9 sustains chondrocyte survival through Pik3ca-Akt pathways. These authors showed that control chondrocytes were positive for pAkt and Sox9 until prehypertrophy. The cells then underwent hypertrophy and terminal maturation before apoptosis. Similarly, mutant chondrocytes lost pAkt upon losing Sox9, and as in our study, went on to undergo prehypertrophy and mineralize their matrix before dying. It is thus unclear that Sox9 and pAkt directly control chondrocyte viability.

Concurring with Ikegami and colleagues (2011), we found that Sox9 is required for chondrocyte hypertrophy, including Col10a1 expression. The presence of Sox9 protein in prehypertrophic and upper hypertrophic chondrocyte nuclei, Sox9 binding to the Col10a1 promoter in hypertrophic cells, and upregulation of Col10a1 reporters by Sox9 through this site, point to a direct role for Sox9 in this process. The late onset of Col10a1 expression in chondrocytes implies that early and hypertrophic chondrocytes use distinct mechanisms to specify Sox9 activity. Candidate partners of Sox9 in prehypertrophy and hypertrophy include Runx2/3 and Mef2c. Several Runx2 and Mef2c binding sites in the Col10a1 enhancer, promoter, and intervening region are needed for reporter activity, and while Mef2c is sufficient to activate a full-length Col10a1 reporter, Runx2 is not (Zheng et al., 2003; Arnold et al., 2007; Li et al., 2011). Runx2 was insufficient in our experiments too, even in combination with Sox9 and Mef2c (data not shown). In contrast, we found that Sox9 and Mef2c act cooperatively to activate Col10a1 and Col10a1 reporters. They bind to adjacent sites in the promoter, and the simultaneous mutation of their sites inactivated reporters. The mutation of either site, however, only partially repressed reporter activation by either or both proteins. This suggests that a functional transactivation complex can be assembled on the Col10a1 promoter as long as one of the two proteins is present and allowed to bind DNA. This possibility would explain that Col10a1 might still be transcribed in lower hypertrophic chondrocytes, where Sox9 protein is no longer detected. However, since Sox9 inactivation abolishes Col10a1 expression, despite upregulation of Mef2c, we proposed and provided evidence that Sox9 contributes to chondrocyte hypertrophy through both direct and indirect mechanisms.

A major function of Sox9 revealed in this study is to maintain the lineage decision of chondrocytes during prehypertrophy, to prevent osteoblastic differentiation and allow hypertrophy. Osteoblastic differentiation of chondrocytes during wild-type endochondral ossification has been debated for decades. Using inducible CreER transgenes and an R26lacZ reporter, Maes and colleagues (2010) recently showed that chondrocytes accumulate at chondro-osseous junctions while precursor cells emerge from the perichondrium to contribute to the trabecular osteoblast population. The time frame of the experiments was unfortunately too short to definitively conclude on the ultimate fate of chondrocytes, i.e., death or contribution to the trabecular osteoblast population. Since it remains uncertain that chondrocytes can turn into osteoblasts in vivo, we used gold-standard markers for cell lineages and differentiation stages to unmistakably prove that Sox9 mutant chondrocytes acquired a fully differentiated osteoblastic phenotype. This included demonstration that cells containing residual amount of Col2a1 RNA, thus of chondrocyte origin, also contained a specific signature of osteoblasts, i.e., both Col1a1 RNA and high levels of Osx RNA. No markers for other mesenchyme-derived lineages were expressed in these cells. While osteoblastic differentiation was thus irrefutable, it occurred only in prehypertrophic cells, indicating that permissive conditions were met only at that stage. These conditions undoubtedly included Ihh expression, as Ihh induces osteoblastogenesis of adjacent perichondrial cells (St-Jacques et al., 1999). A key difference that normally allows prehypertrophic chondrocytes to maintain a distinct lineage fate from perichondrium neighbors is their expression of Sox9. Wild-type terminal chondrocytes resemble osteoblasts in that they express Mmp13 and induce matrix mineralization, but they do not expression RNA for Col1a1 or the osteoblast differentiation factor Osx. They may thus contribute to bone formation as mature osteoblast-like cells rather than genuine osteoblasts, while perichondrium-derived cells undergo full osteoblastogenesis.

Searching for mechanisms whereby Sox9 prevents osteoblastic conversion of chondrocytes, we focused on β-catenin, because the two factors specify the fate of osteochondroprogenitors in opposite ways. While Sox9 inactivation in neural crest led to bone formation in prospective cartilage sites (Mori-Akiyama et al., 2003), β-catenin inactivation in skull vault precursors and endochondral bone perichondrium led to ectopic chondrogenesis (Day et al., 2005; Hill et al., 2005). Moreover, endogenous and constitutively activated β-catenin were shown to silence Sox9 in undifferentiated synovial joint and limb bud mesenchyme, respectively (Hill et al., 2006; Später et al., 2006). In addition to β-catenin blocking Sox9 expression, the two proteins were shown to oppose each other's action by mutually inducing their proteasomal degradation (Akiyama et al., 2004; Topol et al., 2009). Our study shows that endogenous Sox9 inhibits β-catenin signaling in the growth plate and thereby blocks osteoblastic conversion of prehypertrophic cells. This result is consistent with Sox9 inducing β-catenin degradation. Interestingly, we showed that increasing β-catenin signaling through constitutive activation of the protein was insufficient to change the fate of Sox9 wild-type growth plate chondrocytes, and in contrast to findings in undifferentiated mesenchyme, these differentiated chondrocytes were protected from β-catenin signaling-mediated Sox9 gene repression and protein degradation. One possible explanation can be provided based on the facts that Sox9 induces degradation of endogenous β-catenin by recruiting glycogen synthase kinase 3 (Topol et al., 2009) and that the constitutively active form of β-catenin expressed in our mutants is insensitive to this enzyme. Thus, Sox9 protein may have escaped degradation because it would normally be co-degraded with β-catenin. This explanation, however, does not explain how β-catenin signaling leads to Sox9 repression unless, as previously proposed (Kumar and Lassar, 2009), Sox9 maintains its own expression in chondrocytes through positive feedback. Upregulation of Runx2 is certainly another factor that led Sox9-deficient chondrocytes towards osteoblastogenesis, since Runx2 is required for the latter process. However, Runx2 upregulation occurred to the same extent in Sox9/Ctnnb-deficient and Sox9-deficient chondrocytes, proving that it is not sufficient on its own to change the fate of Sox9-deficient chondrocytes. Supporting this conclusion, forced expression of a Runx2 transgene in chondrocytes was shown to lead to chondrocyte ectopic maturation, but not to osteoblastic differentiation (Takeda et al., 2001). Our constitutively activated β-catenin mutants showed slight upregulation of Runx2 expression, but increased chondrocyte maturation rather than osteoblastic differentiation, proving that the combination of both events was insufficient to change the fate Sox9 wild-type chondrocytes. It is possible, however, that it was sufficient in the absence of Sox9, as Sox9 may be critical to shift the activities of Runx2 and β-catenin from osteoblast differentiation to chondrocyte maturation.

In conclusion, this study has demonstrated that Sox9 has more roles in promoting chondrogenesis than previously realized. Like other fate-determining transcription factors in their respective lineages (Tapscott, 2005; Karsenty et al., 2009), Sox9 is involved at many steps of the chondrocyte differentiation pathway. It is necessary both to specify and maintain the lineage choice of the cells and to activate stage-specific markers. Sox9 inactivation in the growth plate resulted in dwarfism due to shortening of columnar and hypertrophic zones and in advanced ossification due to premature prehypertrophy and matrix mineralization. These typical features of campomelic dysplasia imply that the disease is likely due to growth plate defects in addition to cartilage primordia defects, as previously proposed (Bi et al., 2001). They also suggest that changes in Sox9 gene expression or protein activity in growth plates may contribute to many other types of defective growth diseases. Further, while conversion of chondrocytes into osteoblastic cells may not to be a normal process, this study has revealed that differentiated chondrocytes maintain a high degree of lineage plasticity, and thus suggests that chondrocyte ectopic hypertrophy and osteophyte formation, typical features of osteoarthritis, could result from changes in Sox9 activity. This study thus illuminates understanding of the essential roles of Sox9 in chondrogenesis and provides new insights on mechanisms that may underlie both inherited and acquired skeletal diseases.



Mice were used according to federal guidelines and as approved by the Cleveland Clinic Institutional Animal Care and Use Committee. ATC mice were generated as described (Fig. S1). Other alleles and transgenes are described in Results. Data were reproduced with two or more pairs of control and mutant littermates. Dox (D9891, Sigma) was administered at 2mg/ml in drinking water with 5% sucrose.

Assays on tissue sections

Sections of paraformaldehyde-fixed paraffin-embedded tissues were analyzed histologically following staining with nuclear fast red and Alcian blue or the von Kossa reagent using standard protocols. RNA in situ hybridization was performed using 35S-labeled RNA probes (Smits et al., 2004; Nakashima et al., 2002; Arnold et al., 2007; Dy et al., 2010, Table S1). Sox9 immunostaining was performed after antigen retrieval in sodium citrate using rabbit polyclonal anti-Sox9 antibody (1/200; AB5535, Chemicon) and Alexa Fluor 594-conjugated goat anti-rabbit IgG antibody (1/500; Invitrogen). Mounting was with DAPI-containing Vectashield medium (Vector Laboratories). TUNEL staining was performed using alkaline phosphatase In Situ Cell Death Detection Kit (Roche). BrdU incorporation was measured as described (Smits et al., 2004). Data were visualized with Leica DM2500 microscope, captured with Qimaging Micropublisher 5.0 RTV digital camera, and processed with Adobe Photoshop CS4 software. RNA in situ signal density was quantified using Image-Pro Plus 6.1 software (Media Cybernetics).

Col10a1 sequence analysis and reporters

Col10a1 sequences and conservation plots were downloaded from the University of California in Santa Cruz (UCSC) genome browser (http://genome.ucsc.edu/). Col10a1 reporters were constructed as described (Fig. S3A).

Transfection and real-time RT-PCR

Primary chondrocytes were prepared from newborn mouse rib cages as described (Han and Lefebvre, 2008) and cultured in DMEM/F12 medium with 5% FCS and 0.5× ITS (Gibco). For reporter assay, 105 cells were plated per 24-well dish in the presence of 4 U/ml hyaluronidase (Sigma). Transfection mixtures contained 1.2 μl FuGENE 6 (Roche), 250 ng Col10a1 reporter, 40 ng pGL3 control plasmid (Promega), 100 ng expression plasmid encoding no protein, FLAG-SOX9 (30 ng) or Myc-Mef2c (90 ng). Cells were treated with 50 nM okadaic acid or 25 μM forskolin (Kozhemyakina et al., 2009) after 24 h and assayed for luciferase and β-galactosidase activities (Applied Biosystems) 24 h later. Reporter activities were normalized for transfection efficiency and are presented as average with standard deviation of biological triplicates in one representative experiment. Total RNA was prepared using TRIzol (Invitrogen) and assayed by real-time RT-PCR as described (Wang et al., 2007).

Electrophoretic mobility shift assay and chromatin immunoprecipitation

EMSA, Cos1 protein extracts and oligonucleotide probes were prepared as described (Dy et al., 2010). Chromatin immunoprecipitation was performed as described (Bhattaram et al., 2010) using newborn rat growth plate primary chondrocytes transiently transfected with FLAG-SOX9 and Myc-Mef2c expression plasmids. Antibodies were mouse anti-FLAG M2 (Sigma), Myc-Tag (Cell Signaling), non-immune IgG (Sigma), and anti-RNA polymerase II (Upstate). PCR primers were as described (Table S2; Han and Lefebvre, 2008).

Supplementary Material



We thank G. Scherer for providing Sox9fl/fl mice, E.N. Olson for Mef2cfl/fl mice and plasmids, M. Taketo for CnntbflEx3 mice, B. de Crombrugghe, G. Karsenty, H. Kronenberg, W. Lee, and A. McMahon for plasmids, C. Grundy, A. Silvester, and P. Mehta for technical help, the Case Transgenic Facility for ATC founders, and T. Mead for critical advice on the manuscript. This work was funded by NIH/NIAMS grants AR46249, AR54153 and AR60016 to V.L. and an Arthritis Foundation Postdoctoral Fellowship to P.B.


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

SUPPLEMENTAL INFORMATION Supplemental Information includes two tables and five figures and can be found with this article online at doi:….


  • Akiyama H. Control of chondrogenesis by the transcription factor Sox9. Mod. Rheumatol. 2008;18:213–219. [PubMed]
  • Akiyama H, Chaboissier MC, Martin JF, Schedl A, de Crombrugghe B. The transcription factor Sox9 has essential roles in successive steps of the chondrocyte differentiation pathway and is required for expression of Sox5 and Sox6. Genes Dev. 2002;16:2813–2828. [PMC free article] [PubMed]
  • Akiyama H, Lyons JP, Mori-Akiyama Y, Yang X, Zhang R, Zhang Z, Deng JM, Taketo MM, Nakamura T, Behringer RR, McCrea PD, de Crombrugghe B. Interactions between Sox9 and beta-catenin control chondrocyte differentiation. Genes Dev. 2004;18:1072–1087. [PMC free article] [PubMed]
  • Arnold MA, Kim Y, Czubryt MP, Phan D, McAnally J, Qi X, Shelton JM, Richardson JA, Bassel-Duby R, Olson EN. MEF2C transcription factor controls chondrocyte hypertrophy and bone development. Dev. Cell. 2007;12:377–389. [PubMed]
  • Bhattaram P, Penzo-Méndez A, Sock E, Colmenares C, Kaneko KJ, Vassilev A, Depamphilis ML, Wegner M, Lefebvre V. Organogenesis relies on SoxC transcription factors for the survival of neural and mesenchymal progenitors. Nat. Commun. 2010;1:9. [PMC free article] [PubMed]
  • Bi W, Deng JM, Zhang Z, Behringer RR, de Crombrugghe B. Sox9 is required for cartilage formation. Nat. Genet. 1999;22:85–89. [PubMed]
  • Bi W, Huang W, Whitworth DJ, Deng JM, Zhang Z, Behringer RR, de Crombrugghe B. Haploinsufficiency of Sox9 results in defective cartilage primordia and premature skeletal mineralization. Proc. Natl. Acad. Sci. U.S.A. 2001;98:6698–6703. [PMC free article] [PubMed]
  • Day TF, Guo X, Garrett-Beal L, Yang Y. Wnt/beta-catenin signaling in mesenchymal progenitors controls osteoblast and chondrocyte differentiation during vertebrate skeletogenesis. Dev. Cell. 2005;8:739–750. [PubMed]
  • Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G. Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation. Cell. 1997;89:747–754. [PubMed]
  • Dy P, Smits P, Silvester A, Penzo-Méndez A, Dumitriu B, Han Y, de la Motte CA, Kingsley DM, Lefebvre V. Synovial joint morphogenesis requires the chondrogenic action of Sox5 and Sox6 in growth plate and articular cartilage. Dev. Biol. 2010;341:346–359. [PMC free article] [PubMed]
  • Foster JW, Dominguez-Steglich MA, Guioli S, Kwok C, Weller PA, Stevanović M, Weissenbach J, Mansour S, Young ID, Goodfellow PN, et al. Campomelic dysplasia and autosomal sex reversal caused by mutations in an SRY-related gene. Nature. 1994;372:525–530. [PubMed]
  • Guo X, Mak KK, Taketo MM, Yang Y. The Wnt/beta-catenin pathway interacts differentially with PTHrP signaling to control chondrocyte hypertrophy and final maturation. PLoS One. 2009;4:e6067. [PMC free article] [PubMed]
  • Han Y, Lefebvre V. L-Sox5/Sox6 drive expression of the aggrecan gene in cartilage by securing binding of Sox9 to a far-upstream enhancer. Mol. Cell. Biol. 2008;28:4999–5013. [PMC free article] [PubMed]
  • Harada N, Tamai Y, Ishikawa T, Sauer B, Takaku K, Oshima M, Taketo MM. Intestinal polyposis in mice with a dominant stable mutation of the beta-catenin gene. EMBO J. 1999;18:5931–5942. [PMC free article] [PubMed]
  • Hattori T, Müller C, Gebhard S, Bauer E, Pausch F, Schlund B, Bösl MR, Hess A, Surmann-Schmitt C, von der Mark H, et al. SOX9 is a major negative regulator of cartilage vascularization, bone marrow formation and endochondral ossification. Development. 2010;137:901–911. [PubMed]
  • Hill TP, Später D, Taketo MM, Birchmeier W, Hartmann C. Canonical Wnt/beta-catenin signaling prevents osteoblasts from differentiating into chondrocytes. Dev. Cell. 2005;8:727–738. [PubMed]
  • Hill TP, Taketo MM, Birchmeier W, Hartmann C. Multiple roles of mesenchymal beta-catenin during murine limb patterning. Development. 2006;133:1219–1229. [PubMed]
  • Huang W, Chung UI, Kronenberg HM, de Crombrugghe B. The chondrogenic transcription factor Sox9 is a target of signaling by the parathyroid hormone-related peptide in the growth plate of endochondral bones. Proc. Natl. Acad. Sci. U.S.A. 2001;98:160–165. [PMC free article] [PubMed]
  • Ikeda T, Kamekura S, Mabuchi A, Kou I, Seki S, Takato T, Nakamura K, Kawaguchi H, Ikegawa S, Chung UI. The combination of SOX5, SOX6, and SOX9 (the SOX trio) provides signals sufficient for induction of permanent cartilage. Arthritis Rheum. 2004;50:3561–3573. [PubMed]
  • Ikegami D, Akiyama H, Suzuki A, Nakamura T, Nakano T, Yoshikawa H, Tsumaki N. Sox9 sustains chondrocyte survival and hypertrophy in part through Pik3ca-Akt pathways. Development. 2011;138:1507–1519. [PubMed]
  • Karsenty G, Kronenberg HM, Settembre C. Genetic control of bone formation. Annu. Rev. Cell. Dev. Biol. 2009;25:629–648. [PubMed]
  • Kist R, Schrewe H, Balling R, Scherer G. Conditional inactivation of Sox9: a mouse model for campomelic dysplasia. Genesis. 2002;32:121–123. [PubMed]
  • Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, et al. Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell. 1997;89:755–764. [PubMed]
  • Kozhemyakina E, Cohen T, Yao TP, Lassar AB. Parathyroid hormone-related peptide represses chondrocyte hypertrophy through a protein phosphatase 2A/histone deacetylase 4/MEF2 pathway. Mol. Cell.Biol. 2009;29:5751–5762. [PMC free article] [PubMed]
  • Kumar D, Lassar AB. The transcriptional activity of Sox9 in chondrocytes is regulated by RhoA signaling and actin polymerization. Mol. Cell. Biol. 2009;29:4262–4273. [PMC free article] [PubMed]
  • Lefebvre V, Li P, de Crombrugghe B. A new long form of Sox5 (L-Sox5), Sox6 and Sox9 are co-expressed in chondrogenesis and cooperatively activate the type II collagen gene. EMBO J. 1998;17:5718–5733. [PMC free article] [PubMed]
  • Lefebvre V, Smits P. Transcriptional control of chondrocyte fate and differentiation. Birth Defects Res. C Embryo Today. 2005;75:200–212. [PubMed]
  • Li F, Lu Y, Ding M, Napierala D, Abbassi S, Chen Y, Duan X, Wang S, Lee B, Zheng Q. RUNX2 contributes to murine COL10A1 gene regulation through direct interaction with its cis-enhancer. J. Bone Miner. Res. 2011 doi: 10.1002/jbmr.504. [PMC free article] [PubMed]
  • Logan M, Martin JF, Nagy A, Lobe C, Olson EN, Tabin CJ. Expression of Cre Recombinase in the developing mouse limb bud driven by a Prxl enhancer. Genesis. 2002;33:77–80. [PubMed]
  • Maes C, Kobayashi T, Selig MK, Torrekens S, Roth SI, Mackem S, Carmeliet G, Kronenberg HM. Osteoblast precursors, but not mature osteoblasts, move into developing and fractured bones along with invading blood vessels. Dev. Cell. 2010;19:329–344. [PMC free article] [PubMed]
  • Mori-Akiyama Y, Akiyama H, Rowitch DH, de Crombrugghe B. Sox9 is required for determination of the chondrogenic cell lineage in the cranial neural crest. Proc. Natl. Acad. Sci. U.S.A. 2003;100:9360–9365. [PMC free article] [PubMed]
  • Murakami S, Lefebvre V, de Crombrugghe B. Potent inhibition of the master chondrogenic factor Sox9 gene by interleukin-1 and tumor necrosis factor-alpha. J. Biol. Chem. 2000;275:3687–3692. [PubMed]
  • Muzumdar MD, Tasic B, Miyamichi K, Li L, Luo L. A global double-fluorescent Cre reporter mouse. Genesis. 2007;45:593–605. 2007. [PubMed]
  • Nakashima K, Zhou X, Kunkel G, Zhang Z, Deng JM, Behringer RR, de Crombrugghe B. The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell. 2002;108:17–29. [PubMed]
  • Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC, Rosewell IR, Stamp GW, Beddington RS, Mundlos S, Olsen BR, et al. Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell. 1997;89:765–771. [PubMed]
  • Provot S, Schipani E. Molecular mechanisms of endochondral bone development. Biochem. Biophys. Res. Commun. 2005;328:658–665. [PubMed]
  • Rimoin DL, Cohn D, Krakow D, Wilcox W, Lachman RS, Alanay Y. The skeletal dysplasias: clinical-molecular correlations. Ann. N. Y. Acad. Sci. 2007;1117:302–309. [PubMed]
  • Smits P, Dy P, Mitra S, Lefebvre V. Sox5 and Sox6 are needed to develop and maintain source, columnar and hypertrophic chondrocytes in the cartilage growth plate. J. Cell Biol. 2004;164:747–758. [PMC free article] [PubMed]
  • Sock E, Pagon RA, Keymolen K, Lissens W, Wegner M, Scherer G. Loss of DNA-dependent dimerization of the transcription factor SOX9 as a cause for campomelic dysplasia. Hum. Mol. Genet. 2003;12:1439–1447. [PubMed]
  • Später D, Hill TP, O'sullivan RJ, Gruber M, Conner DA, Hartmann C. Wnt9a signaling is required for joint integrity and regulation of Ihh during chondrogenesis. Development. 2006;133:3039–3049. [PubMed]
  • St-Jacques B, Hammerschmidt M, McMahon AP. Indian hedgehog signaling regulates proliferation and differentiation of chondrocytes and is essential for bone formation. Genes Dev. 1999;13:2072–2086. [PMC free article] [PubMed]
  • Takeda S, Bonnamy JP, Owen MJ, Ducy P, Karsenty G. Continuous expression of Cbfa1 in nonhypertrophic chondrocytes uncovers its ability to induce hypertrophic chondrocyte differentiation and partially rescues Cbfa1-deficient mice. Genes Dev. 2001;15:467–481. [PMC free article] [PubMed]
  • Tamamura Y, Otani T, Kanatani N, Koyama E, Kitagaki J, Komori T, Yamada Y, Costantini F, Wakisaka S, Pacifici M, Iwamoto M, Enomoto-Iwamoto M. Developmental regulation of Wnt/beta-catenin signals is required for growth plate assembly, cartilage integrity, and endochondral ossification. J. Biol. Chem. 2005;280:19185–19195. [PubMed]
  • Tapscott SJ. The circuitry of a master switch: Myod and the regulation of skeletal muscle gene transcription. Development. 2005;132:2685–2695. [PubMed]
  • Topol L, Chen W, Song H, Day TF, Yang Y. Sox9 inhibits Wnt signaling by promoting beta-catenin phosphorylation in the nucleus. J. Biol. Chem. 2009;284:3323–3333. [PMC free article] [PubMed]
  • Utomo ARH, Nikitin AY, Lee W-H. Temporal, spatial, and cell type–specific control of Cre-mediated DNA recombination in transgenic mice. Nat. Biotechnol. 1999;17:1091–1096. [PubMed]
  • Vega RB, Matsuda K, Oh J, Barbosa AC, Yang X, Meadows E, McAnally J, Pomajzl C, Shelton JM, Richardson JA, et al. Histone deacetylase 4 controls chondrocyte hypertrophy during skeletogenesis. Cell. 2004;119:555–566. [PubMed]
  • Wagner T, Wirth J, Meyer J, Zabel B, Held M, Zimmer J, Pasantes J, Bricarelli FD, Keutel J, Hustert E, et al. Autosomal sex reversal and campomelic dysplasia are caused by mutations in and around the SRY-related gene SOX9. Cell. 1994;79:1111–1120. [PubMed]
  • Wang L, Shao YY, Ballock RT. Thyroid hormone interacts with the Wnt/beta-catenin signaling pathway in the terminal differentiation of growth plate chondrocytes. J. Bone Miner. Res. 2007;22:1988–1995. [PubMed]
  • Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull. World Health Organ. 2003;81:646–656. [PMC free article] [PubMed]
  • Yoshida CA, Komori T. Role of Runx proteins in chondrogenesis. Crit. Rev. Eukaryot. Gene Expr. 2005;15:243–254. [PubMed]
  • Zheng Q, Zhou G, Morello R, Chen Y, Garcia-Rojas X, Lee B. Type X collagen gene regulation by Runx2 contributes directly to its hypertrophic chondrocyte-specific expression in vivo. J. Cell Biol. 2003;162:833–842. [PMC free article] [PubMed]
  • Zheng Q, Keller B, Zhou G, Napierala D, Chen Y, Zabel B, Parker AE, Lee B. Localization of the cis-enhancer element for mouse type X collagen expression in hypertrophic chondrocytes in vivo. J. Bone Miner. Res. 2009;24:1022–1032. [PMC free article] [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...