Wdpcp Regulates Proliferation and Differentiation in the Developing Limb via Hedgehog Signaling

Background: Mice with a loss of function mutation in Wdpcp were described previously to display severe birth defects in the developing heart, neural tube, and limb buds. Further characterization of the skeletal phenotype of Wdpcp null mice was limited by perinatal lethality. Results: We utilized Prx1-Cre mice to generate limb bud mesenchyme specic deletion of Wdpcp. These mice recapitulated the appendicular skeletal phenotype of the Wdpcp null mice including polydactyl and limb bud signaling defects. Examination of later stages of limb development demonstrated decreased size of cartilage anlagen, delayed calcication, and abnormal growth plates. Utilizing in vitro assays, we demonstrated that loss of Wdpcp in skeletal progenitors lead to loss of hedgehog signaling responsiveness and associated proliferative response. In vitro chondrogenesis assays showed this loss of hedgehog and proliferative response was associated with decreased expression of early chondrogenic marker N-Cadherin. E14.5 forelimbs demonstrated delayed ossication and expression of osteoblast markers Runx2 and Sp7. P0 growth plates demonstrated loss of hedgehog signaling markers and expansion of the hypertrophic zones of the growth plate. In vitro osteogenesis assays demonstrated decreased osteogenic differentiation of Wdpcp null mesenchymal progenitors in response to hedgehog stimulation. Conclusions: These ndings demonstrate how Wdpcp and associated regulation of the hedgehog signaling pathway plays an important role at multiple stages of skeletal development. Wdpcp is necessary for positive regulation of hedgehog signaling and associated proliferation is key to the initiation of chondrogenesis. At later stages, Wdpcp facilitates the robust hedgehog response necessary for chondrocyte hypertrophy and osteogenic differentiation. 1 room HRP conjugated anti-mouse (Abcam) (1:1000) or poly-HRP conjugated anti-goat antibody (Pierce) (1:2000) in 1% milk/TBS-T. Immunoreactive bands were visualized with SuperSignal West Dura Extended Duration Substrate (Thermo Scientic) on a Fotodyne imaging system. Each blot was repeated in duplicate, and representative blot is presented.


Background
The primary cilium is a transiently formed microtubule-based organelle that coordinates many signaling processes within vertebrate development and has been found in almost all vertebrate cell types [1]. There are several hundred proteins and associated complexes controlling formation and maintenance of the cilium [2]. Mutations in the genes encoding these proteins are associated with diseases known as ciliopathies with a wide array of associated phenotypes affecting almost all organ systems [3].
An unbiased N-ethyl-N-nitrosurea (ENU) induced mutagenesis screen in mice for homozygous recessive mutations recovered the Cys40 mouse bearing a null mutation in Wdpcp [17]. Wdpcp was rst identi ed in D. melanogaster as Fritz [21], and the vertebrate homolog was later demonstrated to play a role in hedgehog signaling and ciliogenesis [17,22]. Mutations in Wdpcp in humans are associated with Bardet-Biedl Syndrome and Meckel Gruber Syndrome, ciliopathy disorders with features that include polydactyly, congenital heart defects, renal abnormalities, blindness, truncal obesity, and genitourinary malformations [23].
The Cys40 mice had disrupted ciliogenesis and displayed a gross appendicular skeletal phenotype of polydactyly and shortened, dysmorphic long bones. Analysis of the limb buds showed disrupted formation of truncated Gli3 repressor and altered hedgehog signaling. In addition to the skeletal phenotype, the Cys40 mutant had a number of developmental defects including congenital heart malformations, cystic kidneys, tracheoesophageal stula, and cloacal septation defects that lead to embryonic lethality. A conditional deletion mouse model of Wdpcp bearing LoxP sites on either side of exon 5 was con rmed to recapitulate the phenotype of the Cys40 mutant when crossed with a constitutively expressed CMV-Cre [17].
We generated Prx1-Cre;Wdpcp Cys40/Flox mice to bypass the high perinatal lethality of the Wdpcp constituitive knockout mice and facilitate study of the role of Wdpcp in later stages of skeletal development including the organization of the growth plate and ossi cation of appendicular skeletal elements. Histomorphometric characterization of the developing limbs demonstrated diminished size of the skeletal elements and delayed ossi cation. These ndings corresponded with delayed expression of chondrocyte hypertrophy/early osteoblast markers. Consistent with prior studies of genes necessary for ciliogenesis [6,12,13,[24][25][26], we noted alteration of hedgehog pathway activation in the developing limb bud as well as the growth plate. These ndings were explored further with in vitro assays that demonstrated decreased proliferation, and delayed early chondrogenic and osteogenic differentiation of mesenchymal progenitor cells lacking Wdpcp in response to hedgehog stimulation. In summary, we demonstrate a critical role for Wdpcp in modulating hedgehog signaling to facilitate multiple processes in mammalian appendicular skeletal development including proliferation, growth plate organization, and differentiation of skeletal precursors.

Results
Wdpcp is expressed in skeletal tissues, limb mesenchyme speci c deletion of Wdpcp recapitulated the Cys40 appendicular skeleton phenotype Quanti cation of RNA transcripts of Wdpcp in multiple tissues con rmed that Wdpcp is widely expressed and is enriched in musculoskeletal tissues (Fig. 1A). We chose to focus our studies speci cally on the appendicular skeleton and wanted to determine the spatiotemporal requirement for Wdpcp for normal appendicular skeletogenesis. Previous conditional studies in mice had demonstrated proteins necessary for normal primary cilia formation functioned within the limb bud mesenchyme to facilitate skeletogenesis [12]. Based on these studies, we hypothesized that Wdpcp deletion from the limb bud mesenchyme would recapitulate the Cys40 phenotype. We utilized a conditional allele of Wdpcp in combination with the Prx1-Cre mouse to con rm that Wdpcp functions within the limb bud mesenchyme. The Prx1-Cre mouse expresses Cre recombinase in the forelimb bud mesenchyme at E9.5 and in the hindlimb bud mesenchyme at E10.5 [27]. Mice with a limb bud mesenchyme speci c deletion of Wdpcp were generated by crossing the Prx1-Cre;Wdpcp Cys40/+ mice to the Wdpcp Flox/Flox mice. Quanti cation of RNA transcripts as well as Western Blot for Wdpcp protein con rmed deletion of Wdpcp in E11.5 limb buds (Fig 1B,C). Alizarin Red/Alcian Blue staining of E17.5 Prx1-Cre;Wdpcp Cys40/+ mice (Fig. 1D) and Prx1-Cre;Wdpcp Cys40/Flox mice (Fig. 1E) revealed a similar phenotype to that observed in the Cys40 mutant including severe polydactyly with shortened, dysmorphic long bones [17].
Prx1-Cre;Wdpcp Cys40/Flox mice have smaller cartilage anlagen, delayed ossi cation associated with delayed expression of chondrocyte hypertrophic marker, and expanded zone of hypertrophy in growth plates.
Histomorphometric characterization of E14.5 fore limb buds was notable for signi cantly decreased size of cartilage anlagen in Prx1-Cre;Wdpcp Cys40/Flox (referred to as Wdpcp-cKO) mice versus Prx1-Cre;Wdpcp Cys40/+ (referred to as control) mice. There was a trend toward progressive shortening with more distal elements (Fig 2A-C). Additionally, it was noted that at E14.5 there was delayed ossi cation of the Wdpcp-cKO as indicated by lack of alizarin red staining ( Fig. 2A-B). To explore this nding further, we performed immunohistochemical staining for early chondrocyte hypertrophy markers Runx2 (Cbfa1) and Sp7 (Osterix) expression of which immediately precedes calci cation and vascular invasion [28]. While the controls demonstrated expected staining at the humerus primary center of ossi cation, there was no signi cant staining in the Wdpcp-cKO humerus (Fig. 2D,E). The P0 growth plates of the Wdpcp-cKO mice demonstrated an expanded zone of hypertrophic chondrocytes with loss of the clear zonal demarcation of the prehypertrophic and hypertrophic zones ( Fig. 2F-H).
Hedgehog signaling is disrupted at multiple stages of skeletal development.
Given that prior studies of Wdpcp reported hedgehog signaling dysfunction in the limb bud [17] and the central role this pathway plays in several aspects of skeletal development including chondrogenesis, osteogenesis, and growth plate patterning [29], we chose to investigate the effects of Wdpcp-cKO on hedgehog signaling. We utilized a Gli1 expression as marker of hedgehog pathway activity [30].
Gli1-LacZ;Prx1-Cre;Wdpcp Cys40/Flxo E11.5 limb buds demonstrated signi cantly diminished staining versus control consistent with lack of full activation of the hedgehog pathway (Fig. 3A,B). Similarly, while Gli1 expression in P0 growth plate is typically enriched in the prehypertrophic and early proliferating chondrocytes (Fig. 3C), the Wdpcp-cKO growth plates demonstrated a distinct lack of staining or patterning (Fig. 3D). Previously published data showed failure of Gli3 repressor processing in the E10.5 limb bud [17]. This would theoretically lead to a lack of Gli3 repressor and abnormal activation of hedgehog pathway in contrast to what we observed at E11.5. The discordance of these results highlights the importance of the Wdpcp in not only repression, but also full activation of the hedgehog pathway. The importance of nely tuned hedgehog pathway sensitivity to limb development has been highlighted by recent studies [31]. Our ndings suggest that in the developing limb Wdpcp is necessary not only for full repression of hedgehog signaling but also for full activation. To investigate this further, we examined the responsiveness of mesenchymal progenitor cells to Smo stimulation with SAG treatment, and saw a signi cantly diminished responsiveness in the Wdpcp-cKO cells in expression of the two hedgehog pathway responsive genes Gli1 and Ptch1 [32] (Fig. 3E,F). Several reports have suggested that one of the primary functions of hedgehog signaling in mesenchymal progenitors is control of proliferation necessary to achieve cell density required for differentiation of skeletal tissue [33][34][35]. We examined proliferation of mesenchymal progenitors in response to Smo agonist (SAG) and saw a dramatic

Loss of Wdpcp disrupts early chondrogenesis.
We examined the effect of Wdpcp-cKO on chondrogenic differentiation with limb bud micromass cultures with SAG. We noted similar Alcian Blue staining between Wdpcp-cKO and control ( Fig. 4A-B). Similar to results with mesenchymal progenitor cells, we observed higher expression of hedgehog reporter genes Gli1 and Ptch1 (Fig. 4C,D). Prior studies have shown that Ccnd1 expression is closely correlated with proliferation in mesenchymal progenitors [36]. We found that increased expression of Gli1 and Ptch1 corresponded with increased expression of the proliferation marker Ccnd1 ( [37], and Cdh2 upregulation is a robust marker of early chondrogenic differentiation [38]. We noted signi cantly decreased expression of Cdh2 in Wdpcp-cKO versus control (Fig. 4G). Early chondrogenic differentiation marker Sdc3 (Fig. 4F) as well as later chondrogenic differentiation markers Sox9 (Fig. 4H) and Col2a1 ( Fig. 4I) had lower expression in the Wdpcp-cKO versus control, but did not reach signi cance.

Mesenchymal progenitor cells lacking Wdpcp have impaired hedgehog mediated osteogenic differentiation
Given the observation of delayed calci cation ( Fig. 2A,B) and expression of chondrocyte hypertrophy markers at the primary center of ossi cation (Fig. 2C,D), we wanted to investigate the effect loss of Wdpcp in mesenchymal progenitor differentiation into osteoblasts. Mesenchymal progenitor cells were cultured in osteogenic medium with SAG. Alizarin red S staining at 20 days of culture showed a dramatic difference between control and Wdpcp-cKO cells (Fig. 5A,B). Alkaline phosphatase activity at 4 days of culture showed signi cantly decreased activity in Wdpcp-cKO cells (Fig. 5C). Expression of hedgehog activity reporter genes Gli1 and Ptch1 at 20 days of culture was signi cantly decreased in Wdpcp-cKO ( Fig. 5D,E). We quanti ed the expression of several markers of chondrocyte hypertrophy/osteoblast differentiation. Runx2 is one of the key chondrocyte hypertrophy markers [39]. Expression of Runx2 is necessary for expression of the key osteoblast transcription factor Osx (Sp7) [40]. Ocn and Col1a1 are key components of osteoid matrix and characteristic markers of mature osteoblasts [28]. Expression of Runx2 is not signi cantly different from control in Wdpcp-cKO (Fig. 5F). Expression of Sp7, Col1a1, and Ocn (Bglap2) is signi cantly decreased in Wdpcp-cKO versus control ( Fig. 5G-I).

Discussion
The skeletal phenotype of the WDPCP-cKO mice is similar to several other mutants with de cient ciliogenesis. Although mutations in proteins necessary for normal ciliogenesis typically result in perinatal lethality, some mouse models survive to later stages of development. A hypomorphic allele of the intra agellar transport protein Ift88, the Tg737 orpk mouse, has shortened long bones and abnormal growth plate morphology with delayed expression of chondrocyte hypertrophy marker Col10a1 [7]. Similarly, a hypomorphic allele of Intu demonstrates shortened long bones, delayed ossi cation, and delayed expression of hypertrophic markers with decreased hedgehog activation [20]. Mice with a hypomorphic mutation in Poc1a, a centriolar protein necessary for ciliogenesis, also display shortened long bones and disorganized growth plates [41]. Mice with an inactivating point mutation in Ick demonstrate shortened long bones and polydactyly with delayed hypertrophy and loss of hedgehog activation in the growth plate [42]. Talpid3 -/mice survive to birth and have disrupted ciliogenesis and hedgehog signaling [43] with associated polydactyly, shortened long bones, and delayed chondrocyte hypertrophy with expanded zone of hypertrophy in the growth plate [44]. Evc -/and Evc2 -/mice survive to birth and display no polydactyly, but do have shortened long bones. Evc localizes to the basal body of the chondrocyte primary cilium, and Evc -/mice display disorganized growth plate structure with decreased Gli1 and Ptch1 expression [26]. Unlike the Wdpcp Cys40/Cys40 mutant [17], Evc -/broblasts do not have a dramatic impairment in Gli3 full length to repressor processing, but do have decreased Gli1 expression in response to SAG [45]. Common to all of these models is a failure to activate hedgehog signaling and shortened appendicular skeletal elements.
The Wdpcp-cKO skeletal phenotype also has similarities with conditional knockout models of cilia related proteins. Prx1-Cre mice with conditional limb bud mesenchyme deletion of intra agellar transport proteins [12,46], kinesins [47], or basal body proteins [25] display polydactyly and shortened long bones with delayed ossi cation that correlates with disrupted cilia formation and loss of Gli1 expression and hedgehog signaling abnormalities similar to the Wdpcp-cKO mouse. The Wdpcp-cKO mouse represents the rst conditional limb bud mesenchyme knockout of a CPLANE protein. Col2a-Cre mice with conditional deletion of Kif3a or Ift88 at a later time point in skeletal development after chondrogenesis do not have polydactyly and have mildly shortened long bones, but do have abnormal growth plates with decreased growth plate proliferation and decreased hedgehog activation similar to the Wdpcp-cKO mice [6]. These studies suggest that proliferation and growth prior to onset of Col2a1 expression is an important contributor to the skeletal phenotype of the Wdpcp-cKO mice and other cilia de cient mice.
Proliferation is essential for initiation of chondrogenesis to achieve the cell density required [34].
Additionally, tight regulation of hedgehog signaling is also critical for early initiation of chondrogenesis [29]. Recent studies of limb bud development have implicated hedgehog signaling as a key regulator of both mesenchymal limb bud progenitor proliferation and transition to chondrogenesis [36], and have highlighted the sensitivity of this process to small perturbations in the level of cilia associated hedgehog signaling activation [31]. Our ndings of disrupted hedgehog signaling, decreased proliferative response, and decreased expression of early chondrogenic markers in the Wdpcp-cKO mesenchymal progenitor cells is in line with these prior ndings. Furthermore, they suggest that a signi cant contributing driver of the phenotype observed in the Wdpcp-cKO mice is related to perturbed hedgehog signaling in the later stages of limb bud mesenchyme proliferation and differentiation to early stages of chondrogenesis.
Hedgehog signaling represents the molecular pathogenic link between ciliary dysfunction and skeletal phenotype [48,49]. Cilia coordinate the processing of the Gli transcription factors to their shortened repressor forms [31] as well as their fully activated forms downstream of hedgehog [50]. Similar to the Wdpcp-cKO mice, Ihh -/mice have shortened long bones, and knockdown of Ihh induces inhibition of cell proliferation and mature chondrocyte markers [51]. Prrx1-CreERT deletion of Ptch1 in postnatal mice activates hedgehog signaling and promotes mesenchymal progenitor proliferation as well as differentiation into chondrocyte and osteoblast [52]. Conversely, failure to activate hedgehog signaling in Wdpcp-cKO mesenchymal progenitors slows proliferation and differentiation into chondrocyte and osteoblast. Similarly, knockdown of Ift80 and Ift88 in cultured mesenchymal progenitors disrupts ciliogenesis and has been shown to decrease chondrocyte and osteoblast differentiation [8][9][10]53].
Previous studies of Wdpcp null mice have demonstrated perturbed Wnt signaling in developing heart and PCP defects including disrupted cochlear hair cell alignment with loss of Vangl2 expression as well as abnormal cell migration and cytoskeleton organization [17]. In the developing limb however, the Wdpcp-cKO mouse does not display a loss of attening or columnar organization observed in other mutants of the PCP pathway [54][55][56]. Thus, while PCP may contribute in part to the observed skeletal phenotype of the Wdpcp-cKO mouse, the overlap in hedgehog signaling changes and phenotype with other cilia mutants suggests that these are signi cant components of the pathology. Additionally, prior studies have demonstrated that while Wnt signaling is perturbed in cilia de cient mutants, this is a downstream effect of the alteration of hedgehog signaling [57]. Interestingly, mice bearing a hypomorphic allele of the Wnt PCP receptor Prickle1 have been found to have defective ciliogenesis [58,59], underscoring the close interplay between cilia and PCP signaling. Future studies with the Wdpcp-cKO mouse will help to elucidate the speci c roles of cilia and PCP in the developing limb and to re ne our molecular model of limb development and growth plate organization.

Conclusions
The generation of the Wdpcp-cKO mouse characterized in this study established Wdpcp as essential for mouse appendicular skeletal development. We demonstrate that there is a signi cant impairment in the activation of hedgehog signaling that is associated with compromised proliferation and differentiation of skeletal tissues. Additionally, this new mutant mouse strain will serve as an important model for studying the postnatal functions of Wdpcp.

Genotyping
For genotyping, DNA was extracted from livers of embryos and tail clips of newly weaned mice using the REDExtract-N-Amp Tissue PCR Kit (Sigma). Primers used for genotyping are included in Table 1. The Wdpcp Cys40 allele was genotyped using Sanger sequencing of a 150 base pair fragment to identify the relevant A>G base pair change (Supplemental Fig. 1A). The Wdpcp Flox allele was genotyped using polymerase chain reaction (PCR) with primers to amplify a region containing an inserted ox site (424 base pair fragment) or lacking an inserted ox site (262 base pair fragment) that could be differentiated with agarose gel electrophoresis [17]. Presence of the Prx1-Cre allele was identi ed by PCR with primers speci c to Cre that ampli ed a 102 base pair fragment (Supplemental Fig. 1B,C) [60].

Skeletal Preparations
Skeletal preparations were made by co-staining embryos with Alizarin Red S for ossi ed, calcium rich tissue and Alcian Blue for cartilage as described previously [61]. Brie y, embryos were scalded in hot tap water, skinned, and their abdominal and thoracic organs removed. Livers were saved for genotyping.
Embryos were transferred into ethanol before staining 24 hours with 40% glacial acetic acid, 60% ethanol, 0.0001% Alcian Blue (Sigma) solution at room temperature. Embryos were destained for 24 hours in ethanol before being transferred to a solution containing 2% KOH (Sigma) in distilled water with 0.0015% Alizarin Red S (Sigma) for 5 hours. Embryos were subsequently destained in 2%KOH in distilled water overnight, followed by 1% KOH in 50% glycerol/50% distilled water. Embryos were transferred to 50% glycerol/50% distilled water for imaging. All images are representative of at least 3 specimens. For limb length quanti cation, the longest vector along each cartilage element was used. Six limbs from six separate animals were used for both mutants and controls.
Quantitative Real time RT-PCR (qRT-PCR) Total RNA was extracted using TRIzol (Invitrogen) with RNeasy columns (Qiagen) for puri cation and removal of genomic DNA. RNA was quanti ed on nanodrop, and cDNA was synthesized from 200ng of total RNA using Superscript III (Invitrogen) with the Oligo(dT) primers per the manufacturer's protocol. Quantitative real-time PCR was performed on Applied BioSystems StepOnePlus with Applied Biosystems SYBR Green PCR mastermix in 96-well plates. Primers were all used at a concentration of 200 nM. Cycling variables were as follows: 95 °C for 10 min, then 40 cycles of 15 second denaturation at 95 °C and 1 minute at 60 °C. Primer sequences are included in Table 2. Expression was reported normalized to housekeeping gene HPRT due to reported stability of expression of this gene during skeletal development [62,63]. The mean and standard deviation represent 3-4 biological replicates each composed of RNA isolated from 3 experimental replicates. All statistics were performed on difference from HPRT cycle counts to avoid error propagation.

Western Blotting
The limb bud speci c knockdown of Wdpcp was con rmed by individually genotyping embryos after forelimb buds had harvested for protein isolation. Forelimb bud protein content was isolated with Total Protein Extraction Kit (Millipore) supplemented with 5 mM EDTA and 1X Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scienti c). Samples were sonicated to homogenize lysate and protein concentration was quanti ed using BCA assay kit (Pierce) and standardized to same concentration using isolation buffer dilution. Western blots were performed as previously described [64]. Protein samples were subjected to reducing SDS-PAGE and transferred to low-uorescence background polyvinyl uoride (PVDF) membranes (Millipore). Membranes were blocked in 3% milk in 0.25% Tween-20 in TBS (TBS-T) for 1 hour at room temperature and probed overnight at 4C with primary antibody in 1% milk/TBS-T. Primary antibodies used were goat anti-Wdpcp (1:1000) (Santa Cruz, Sc-245737 T-20) and mouse anti-βactin (Santa Cruz sc-47778) (1:1000). After washing with TBS-T, membranes were incubated for 1 hour at room temperature with HRP conjugated anti-mouse (Abcam) (1:1000) or poly-HRP conjugated anti-goat antibody (Pierce) (1:2000) in 1% milk/TBS-T. Immunoreactive bands were visualized with SuperSignal West Dura Extended Duration Substrate (Thermo Scienti c) on a Fotodyne imaging system. Each blot was repeated in duplicate, and representative blot is presented.

Hedgehog responsiveness experiments
Mesenchymal progenitor cells used in these experiments were isolated and expanded from E13.5 forelimbs. For all drug treatment experiments, cells were grown to con uence, trypsinized, and re-plated at full con uence (150,000 cells/cm 2 ) in appropriate size culture well in growth medium for 1 day before changing to starvation medium (DMEM supplemented with 0.25% MSC quali ed FBS). Previous studies had demonstrated that cell cycle arrest as induced by serum starvation and con uence is necessary to induce optimal ciliation and hedgehog responsiveness [65]. Following 2 days of culture with starvation medium, medium was changed to starvation medium supplemented with pharmacological agent dissolved in DMSO (Sigma) for 24 hours. RNA for RT-PCR and/or protein for Western Blots was isolated from cells at this time point. For expression studies, results presented are from 3 independent biological replicates. All drugs were diluted in dimethyl sulfoxide (DMSO) and added 1:1000 to culture medium. DMSO was added 1:1000 for carrier control. The Click-iT Plus Edu 594 assay was used to quantify cell proliferation over nal 24 hours of cell culture. Following serum starvation period, 10 mM EdU dissolved in PBS was added to culture medium at 1:1000 for a nal concentration of 10 uM. Following xation with 4% paraformaldehyde for 15 min at room temperature, EdU staining was performed per the manufacturer's protocol (Lifetechnologies). Hoechst staining was performed to identify all nuclei. Images were acquired on Zeiss inverted epi uorescent microscope with 10X objective. Three elds were captured for each treatment condition and presented images are representative. Three biologic replicates were performed per experimental condition.

Micromass Culture
Forelimbs were obtained from 11.5 dpc Prx1-cre;Wdpcp Cys40/Flox and control embryos. Micromass cultures were prepared as previously described [66][67][68]. Dissected limbs were dissociated in 1 unit/mL dispase (Stem Cell Technologies) for 1.5 h at 37 °C. Digested limbs were pipetted up and down with 200 uL pipette to break up cells before being diluted in 10 mL 2:3 DMEM/F-12 medium containing 10% MSC quali ed FBS (Invitrogen) to neutralize dispase. This suspension was passed through a 40 um lter and spun down to a pellet at 1200g. Cells were resuspended in 2:3 DMEM/F-12 medium and counted with hematocytometer. Cells were then divided into experimental groups, spun down and resuspended in 2:3 DMEM/F-12 medium supplemented with either DMSO or appropriate pharmacological agent dissolved in DMSO. Cells were diluted to 10 million cell/mL and spotted in 10 uL droplets on Nunc 24-well culture dishes. Cells were allowed to adhere for 1.5h in a cell incubator with humidi ed atmosphere containing 5% CO 2 . After two hours, wells were ood with growth medium containing appropriate pharmacological agent. Cultures were harvested after 3 days. For Alcian Blue staining, cultures were xed for 15 minutes at room temperature in 4% paraformaldehyde in 1X phosphate buffered saline before staining overnight at 4°C with 1% Alcian Blue stain solution pH 1.0 (EK Industries). Stained cultures were washed 3x with phosphate buffered saline before imaging. Alcian Blue staining shows a representative image of 3 to 4 independent biological replicates [60].

Histology and Immunohistochemistry
Specimens were xed in 4% paraformaldehyde at indicated time point for 24h overnight at 4C before being dehydrated and para n embedded. Forelimbs were sectioned at 7 micron thickness. For Safranin O/Fast Green staining, sections were depara nized and rehydrated before being stained in Weigert's Iron Hematoxylin (Sigma) and 0.02% aqueous Fast Green (Sigma) followed by a rinse in 1% acetic acid and 0.1% aqueous Safranin-O (Sigma). For immunohistochemistry, ImmPRESS Excel ampli ed polymer staining kit (Vector Labs) was used [60]. Primary antibodies used for immunohistochemistry were rabbit anti-Osx/Sp7 (

Mesenchymal progenitor isolation
Mesenchymal progenitor cells were isolated from the limbs of E13.5 embryos that were digested in 0.5% Trypsin EDTA and mechanically disrupted. Trypsin was neutralized after 5 minutes at 37C with DMEM supplemented with 10% MSC quali ed FBS (Gibco) with 1X Penicillin/Streptomycin/Fungizone (Gibco) growth medium. Cells were pipetted up and down to break up remaining tissue before spinning at 1000g to pellet in 15 mL Vulcan tube. Cells were then resuspended in growth medium and plated on T75 at what was considered passage 0. Cells were expanded two days with growth medium changed each day. Cells were passaged on day 2 and split (1:4). Cells were grown to 90% con uence, passaged, and frozen in cryoprotective freezing medium (Lonza) for storage in liquid nitrogen. All experiments using were conducted at passage 3 or 4 [60].