Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Expression of osterix Is Regulated by FGF and Wnt/β-Catenin Signalling during Osteoblast Differentiation

  • Katharina Felber,

    Affiliation Bateson Centre and Department of Biomedical Science, University of Sheffield, Sheffield S10 2TN, United Kingdom

  • Philip M. Elks,

    Affiliation Bateson Centre and Department of Biomedical Science, University of Sheffield, Sheffield S10 2TN, United Kingdom

  • Maria Lecca,

    Affiliation Bateson Centre and Department of Biomedical Science, University of Sheffield, Sheffield S10 2TN, United Kingdom

  • Henry H. Roehl

    Affiliation Bateson Centre and Department of Biomedical Science, University of Sheffield, Sheffield S10 2TN, United Kingdom

Expression of osterix Is Regulated by FGF and Wnt/β-Catenin Signalling during Osteoblast Differentiation

  • Katharina Felber, 
  • Philip M. Elks, 
  • Maria Lecca, 
  • Henry H. Roehl


Osteoblast differentiation from mesenchymal cells is regulated by multiple signalling pathways. Here we have analysed the roles of Fibroblast Growth Factor (FGF) and canonical Wingless-type MMTV integration site (Wnt/β-Catenin) signalling pathways on zebrafish osteogenesis. We have used transgenic and chemical interference approaches to manipulate these pathways and have found that both pathways are required for osteoblast differentiation in vivo. Our analysis of bone markers suggests that these pathways act at the same stage of differentiation to initiate expression of the osteoblast master regulatory gene osterix (osx). We use two independent approaches that suggest that osx is a direct target of these pathways. Firstly, we manipulate signalling and show that osx gene expression responds with similar kinetics to that of known transcriptional targets of the FGF and Wnt pathways. Secondly, we have performed ChIP with transcription factors for both pathways and our data suggest that a genomic region in the first intron of osx mediates transcriptional activation. Based upon these data, we propose that FGF and Wnt/β-Catenin pathways act in part by directing transcription of osx to promote osteoblast differentiation at sites of bone formation.


The bony skeleton initially develops in one of two ways, either by ossification of a cartilage template (chondral ossification) or in the absence of a cartilage template (achondal ossification). Osteoblasts (specialised cells that synthesise bone) are derived from multipotent mesenchymal stem cells which are found in different tissues. During chondral bone development, osteoblasts initially differentiate in the perichondrium (a tissue which surrounds the cartilage), and in achondral bone development osteoblasts differentiate in mesenchymal cell condensations. Later on in development and in adults, osteoblast progenitors are found in the bone marrow as well as in the periosteum (a tissue which surrounds bone). Genetic analysis in mice has identified two transcription factors, Runx2 (Runt-related transcription factor 2) and Osx (also known as Sp7), that act in a transcriptional cascade during osteoblast differentiation. In mice lacking either gene osteoblasts throughout the body fail to differentiate while other cell types are largely unaffected[13]. In osteoblast progenitors, Runx2 expression precedes that of Osx and it is known that Runx2 is required to activate Osx transcription[4]. Both transcription factors have been shown to activate expression of many markers of mature osteoblasts including Collagen 1(Col1), Secreted protein acidic cysteine-rich (SPARC), Osteopontin (Opn), Bone Sialoprotein (BSP), Osteocalcin (Osc) and Alkaline phosphatase (ALP)[5,6]. Whether Osx also plays a pivotal role in osteoblastogenesis in humans is unclear as one study suggests a relatively mild skeletal phenotype occurs when OSX is mutated[7].

FGF signalling plays a crucial role during skeletal development. Mutations in human FGFR1, FGFR2, FGFR3, FGF10 and FGF23 all cause skeletal defects consistent with a role in osteoblast differentiation and/or function[8]. However, experiments to define the role of the FGF pathway in osteoblastogenesis have often generated conflicting results (reviewed in [9] and [5]). For example it has been shown that FGF signalling activates expression of Runx2 in MSCs to initiate the osteoblast lineage, and later activates Opn and BSP expression in maturing osteoblasts [1014]. This is supported by in vivo studies that have shown that mutations that impair FGF sigalling reduced bone density [1517]. On the other hand, activation of FGF signalling in vitro results in reduced expression of ALP and Col1 and induces osteoblast apoptosis [1820]. These results suggest that FGF signalling may play different roles during osteoblast differentiation and that timing and strength of the FGF signal is crucial in these outcomes.

Wnt signalling via the β-Catenin pathway has more recently been identified as a key regulator of osteoblastogenesis [2123]. As with FGF signalling, a consensus has not emerged regarding the precise role of the Wnt/β-Catenin pathway. Conditional inactivation of β-Catenin in the murine embryo has established that it is required for Osx, Osc and Col1 expression in the osteoblast lineage [2427]. However, β-Catenin knock-out also causes an increase in the expression of Runx2 and expression of a constitutively active form of β-Catenin blocks entry into the osteoblast lineage. Together these results suggest that Wnt/β-Catenin acts at two sequential stages, to inhibit differentiation initially, then to promote differentiation after commitment. Other studies using murine MSCs have found that Wnt3a treatment upregulates ALP, but does not affect Runx2, Osc or Col1 levels[28,29]. Further studies have shown that Wnt/β-Catenin signalling promotes early osteoblastogenesis in vivo and in mouse embryonic fibroblasts by direct activation of Runx2 expression [30,31]. Studies using human MSCs have found that Wnt/β-Catenin acts to suppress entry into the osteoblast lineage [3234] and analysis of Wnt10b-/- mice suggests that an osteopenic phenotype results from decreased maintenance of adult MSC in bone [35]. The finding that Osx and Wnts interract in positive and negative regulatory loops may explain why it has been very difficult to ascribe a simple role for Wnts in skeletal development [36,37]. Together these studies indicate that role of Wnt/β-Catenin signalling varies according to the precise timing and context of the signalling event.

Analysis of zebrafish bone development suggests that the regulation of osteoblastogenesis is conserved between fish and mammals. As in mammals, the retinoic acid, BMP and Hedgehog pathways regulate recruitment and/or anabolic activity of osteoblasts in zebrafish [3841]. Expression of runx2 (runx2a and runx2b), osx, col1a2, sparc, osc and opn mark progression of osteoblastogenesis [40,4246]. Reporter transgenes based upon osx genomic sequence have been generated in both zebrafish and medaka[46,47]. Both achondral and chondral skeletal development takes place in zebrafish, and although bone is cellular, bone marrow does not form [4850]. Here we analyse the roles that FGF and Wnt/β-Catenin pathways play during achondral ossification in the head. We find that both pathways promote ossification and act at the level of osx expression. Besides acting in parallel to regulate osx expression, we also find evidence that Wnt/β-Catenin signalling regulates the activity of the FGF pathway during skeletogenesis.

Materials and Methods

All methods were performed using standardised protocols[51]. All animal husbandry and experimentation was carried out under the supervision and approval of the Home Office (UK) and the University of Sheffield Ethics Board. Adult zebrafish were maintained with a 14 h light/10 h dark cycle at 28°C according to standard protocols and were mated using pair mating in individual cross tanks. To kill fish, larvae were anesthetized with 0.2 mg/ml of Tricane at 4°C. Unless otherwise stated, 10 larvae were analysed for each sample after chemical treatment or heat shock and representative images were chosen for figure panels.

Alizarin Red staining

Larvae were fixed 2 hours (hrs) in 4% formaldehyde/PBS, stored in 80%MeOH/H2O for 1 hour. The larvae were rinsed briefly in H2O with 0.1% Tween-20 (H2Otw). Then larvae were bleached in 1.5% H2O2 in 1% KOH for 30 minutes at 37C. Larvae were stained in 1% KOH with 0.04 mg/ml Alizarin red for 2 hrs. Larvae were passed through a glycerol series (25%, 50%, 80% glycerol in H2Otw) 10 minutes each and then photographed in 80% glycerol/H2O.

Alcian Blue staining

Larvae were fixed as above then stained in 0.1% Alcian Blue in 0.1N HCl overnight and bleached as above. The staining was fixed for 10 minutes in borate buffer (30% saturated-sodium tetraborate in H2Otw), and then cleared with trypsin (0.5mg/ml trypsin in borate buffer) until the tissue was completely digested away from the skeleton (2–3 hrs at 37°C). Larvae were then photographed in glycerol as above. Larvae stained this way can be stored indefinitely.

von Kossa staining

Larvae fixed at 120hpf for 2 hrs in 4%PFA were rinsed 3X 5min in H2Otw. Then larvae were incubated for 45–60 minutes in 1% aqueous silver nitrate under incandescent lamp (60W) and staining was monitored at regular time intervals. Following 3X 5min H2Otw rinses larvae were then fixed in 2.5% sodium thiosulfate for 10 minutes and post-fixed in 4%PFA for 30 minutes. Larvae were then photographed in glycerol as above. Larvae stained this way can be stored indefinitely.

Heat shock procedure

20–40 larvae were placed in 25ml E3 buffer in a 50ml falcon tube and put into a water bath at 38°C for one hr. Heat shock time refers to the start of the heat shock. Continuous transgenic expression was accomplished by heat shocking for one hour every 12 hours for the duration of the treatment. All experiments were done with hemizygous carriers and included the wild type siblings as controls, larvae were sorted after heat shock based upon fluorescence of the transgene.

Chemical treatments

SU5402 (Merck) was made up in DMSO to 10mM and stored at -20°C. GSK-3 Inhibitor XV (Merck) was made up in DMSO to 5mM and stored at -20°C. All treatments were done at a dilution of 10uM in E3 in the dark.

Time courses and qPCR

For FGF signalling, three batches of 15 wild type embryos were treated with SU5402 at 50hpf(2hrs), 50.5hpf(1.5hrs), 51hpf(1hr) and 51.5hpf(0.5hr). All treatments and an untreated batch were terminated at 52hpf and processed using Trizol and Superscript II following the manufacturer's protocols (Invitrogen). qPCR was performed using iQ SYBR Green Supermix (Bio-Rad) with a Bio-Rad MyiQ system. Primer pairs for pea3 and osx performed equivalently in a dilution series so the comparative CT method was used to calculate relative expression levels. For Wnt/β-Catenin signalling, hs:dkk1 hemizygous fish were crossed to Tg(TOP:GFP) homozygous carriers. The offspring were heat shocked at 48hpf and sorted based upon fluorescence. Three batches of fifteen hs:dkk1and fifteen sibling fish were fixed every two hours until 54hpf. RNA was processed using Ultraspec (AMS Biotechnology) and the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) using manufacturers' protocols. Input levels were quantified using TaqMan probes, and Universal Mastermix and ABI 7200 (Applied Biosystems). The housekeeping gene phosphoglycerate kinase 1 (pgk1) was used to normalise. Non-heatshocked fish at 48hpf (unsorted) were also processed as a control. Significance was tested using the Students T-test using three technical replicates and comparison of the individual dCT values. Primer sequences are listed in S1 Table.

Chromatin Immunoprecipitation (ChIP)

100 embryos were first deyolked using 1ml deyolking buffer and vortexed at 1000rpm for 5 minutes, cross-linked in 1ml fresh 1% formaldehyde for 15 minutes at room temperature and washed 3 times with 0.125M Glycine and 2 times with cold PBS. DNA was extracted by adding 600μl DNA extraction buffer (containing protease inhibitors), mixed 20 times with a 200μl tip and incubated on a shaker for 1 to 2 hours at 4°C. After centrifugation for 5 minutes at 3500rpm (4°C), the pellet was dissolved in 400μl IP dilution buffer and the DNA was sonicated. A sample of the DNA was analysed by gel electrophoresis to ensure a fragment size of approximately 200–1000 base pairs. 40μl of washed Protein A bead slurry was added and rotated at 4°C for 1 hour. The beads were removed and the antibodies (and input containing no antibody) were added and the sample was rotated over night at 4°C: β-catenin (Sigma C2206) 10μl; ETS 1/2 (Santa Cruz C-275) 25μl. 40μl of beads were added to each sample and the mix was incubated for 1 hour at room temperature. The beads were washed and elution buffer was added (150μl) to the beads and rotated for 15 minutes at room temperature. The elution step was repeated once more. Thereafter the beads were discarded and 22.5μl 4M NaCl and 1μl RNase A was added to the eluate and this was incubated 5 hours at 65°C. The sample was then precipitated, the pellet was dissolved in and incubated with proteinase K at 45°C for 3 hours. After Phenol/Chloroform extraction, the sample was precipitated again, and cleaned with the PCR purification kit (Qiagen). It was eluted in 15μl TE buffer and stored at -20°C. For a detailed protocol please contact the authors. Primer sequences are listed in S1 Table.

Quantification of in situ stainings

To quantify the staining, three independent sets (>10 fish each) were treated with DMSO, SU5402, hsdnFGFR1 or hsFGF3 and processed identically for each experiment. These were then scored for the presence of staining associated in the opercle, and the data was combined to generate an average with associated standard deviation. As loss of staining was generally all or nothing, animals were scored for the presence of any staining (scored as one) or the complete loss of staining (scored as 0). The researcher was blind to which sample was control or treated during this analysis.


FGF signalling is required for ossification and the development of mature osteoblasts

To determine whether FGF signalling acts during zebrafish bone development we took advantage of a zebrafish transgenic line called hs:dnfgfr1 that expresses a dominant negative form of the FGF under control of a heat shock promoter (Tg(hsp70l:dnfgfr1-EGFP); [52]). Hemizygous fish were continuously exposed to dnFGFR1 from 48 hours post fertilisation (hpf) until 120hpf (heat shock for one hr every eleven hours). This treatment completely abolished ossification except for a small part of the cleithrum which begins to ossify before the start of the treatment (Fig 1A and 1B). Chondrogenesis, which also takes place during this time period was not noticeably affected (Fig 1C and 1D). It is possible that treatment starting at earlier timepoints would affect the cartilagenous skeleton but such treatments result in disruption of morphogenesis of the embryo (data not shown). The loss of bone could reflect defects in calcium homeostasis or in osteoblast activity. To determine whether osteoblasts were present and secreting bone matrix proteins, we performed in situ analysis with opn, col10a1 and col1a2 (Fig 1E–1H and data not shown). Although expression of col10a1 is typically described as being associated with hypertrophic chondrocytes in mammals, it is expressed in mature osteoblasts in zebrafish[43,53]. Both markers were drastically reduced or absent in treated fish suggesting that FGF signalling is required for the differentiation or the anabolic activity of osteoblasts.

Fig 1. FGF signalling is required for ossification and osteoblast maturation.

(A-D) Continuous inhibition of FGF signalling from 48hpf to 120hpf in hs:dnfgfr1 larvae results in loss of ossification (B) but cartilage formation is relatively unaffected (D). (E-H) Expression of mature osteoblast markers col1a2 and col10a1 is severely reduced in hs:dnfgfr1 larvae (F,H). High magnification images of the opercle are shown to the right of panels E-H. Abbreviations: bs = branchiostegal ray, cl = cleithrum, op = opercle. Scale bar = 200μM.

To test whether osteoblast differentiation requires FGF signalling we decided to analyse earlier stages, focusing primarily on development of the opercle bone. runx2a and runx2b are early markers of osteoblasts that first appear in the region of the opercle at around 48hpf followed by osx at around 51hpf. We used the hs:dnfgfr1 line and the pharmacological inhibitor SU5402 to block FGF signalling and a hs:fgf3 line to activate FGF signalling. To determine when these treatments have their strongest effect, we used expression of pea3 (polyomavirus enhancer activator 3) and erm (ets related molecule) as transcriptional read-outs of the FGF pathway [54]. We found that three hours after the onset of treatment has the optimal effect on FGF signalling throughout the embryo (S1 Fig). Treatment starting at 48hpf lasting for three hours did not affect runx2a expression (Fig 2A–2D), and a mild reduction in runx2b was seen only after treatment with SU5402 (Fig 2E–2H). In contrast, osx expression was strongly downregulated by both hs:dnfgfr1 and SU5402 treated fish and upregulated by hs:fgf3 (Fig 2I–2L). To quantify these results we counted the number of fish which had staining in the opercle and did qPCR on whole larvae and both methods established that perturbation of FGF signalling has a significant effect on osx gene expression. To determine whether the expression of late osteoblast markers (i.e. bone matrix genes) is affected immediately following perturbation of FGF signalling, we treated fish for three hours starting at 60hpf (S2 Fig). Neither SU5402 treatment or hs:fgf3 had a an affect on the expression of col1a2 (and sparc, data not shown) indicating that FGF signalling is not likely to directly regulate the expression of these bone matrix genes.

Fig 2. FGF signalling regulates expression of osx.

(A-H) Larvae were treated at 48hpf, fixed at 51hpf and expression of several bone markers was analysed by in situ hybridisation. Arrowheads point to the opercle in the high magnification images to the right of each panel. Expression of runx2a (A-D) and runx2b (E-H) is unchanged after inhibition of FGF signalling using hs:dnfgfr1 larvae as well as over activation in hs:fgf3 transgenic fish. Expression of runx2b is slightly reduced after SU5402 treatment (F). (I-L) Expression of osx is decreased after inhibition of FGF signalling and increased after over activation of FGF signalling as shown by in situ hybridization. Abbreviations: cl = cleithrum, op = opercle. Scale bar = 200μM. (M) Quantification for presence or absence of osx staining in the opercle from I-L. Expression was normalised to the control. (N) qPCR performed in parallel to I-L confirm effects of FGF signalling on osx expression. Treatments were normalised to gapdh. Treatment with SU5402 showed a significant reduction (p<0.05 Student's t-test).

Wnt/β-Catenin signalling is required for ossification and the development of mature osteoblasts

To determine the role of Wnt/β-Catenin signalling during osteoblast differentiation we used two transgenic lines: hs:dkk1 (Tg(hsp70l:dkk1-GFP); [55]) to downregulate signalling and hs:wnt8a (Tg(hsp70l:wnt8a-GFP); [56]) to upregulate signalling. We found that continuous suppression of Wnt signalling starting at 48hpf resulted in reduced ossification levels at 120hpf (Fig 3B). Similar treatment using hs:wnt8a resulted in an increase in ossification (Fig 3C). Neither treatment resulted in a strong change to the cartilagenous skeleton (Fig 3D–3F). It is possible that treatment starting at earlier timepoints would affect the cartilagenous skeleton but such treatments resulted in disruption of morphogenesis of the embryo (data not shown). Treatment from 72hpf until 108hpf resulted in reduced col10a1 and col1a2 in hs:dkk1fish and increased expression in hs:wnt8a fish (Fig 3G–3L). These data are comparable to the results obtained by manipulating FGF signalling and suggest a role for Wnt/β-Catenin signalling in osteoblast differentiation or bone matrix secretion.

Fig 3. Wnt/β-Catenin signalling regulates ossification and development of mature osteoblasts.

(A-F) Continuous inhibition (hs:dkk1) or over activation (hs:wnt8) of Wnt/β-Catenin signalling from 72-120hpf. Inhibition of Wnt/β-Catenin signalling results in reduced ossification as shown by Alizarin red staining (B) whereas cartilage formation is largely unaffected (E). Overactivation of Wnt/β-Catenin signalling results in increased ossification and precocious ossification of the hyomandibula (C). (G-L) Analysis of osteoblast markers after treatment from 72-108hpf. Expression of mature osteoblast markers col1a2 and col10a1 is slightly reduced when Wnt/β-Catenin signalling is inhibited (H,K) and enhanced by increased Wnt/β-Catenin signalling (I,L). Abbreviations: hm = hyomandibula, op = opercle, te = teeth. Scale bar = 50μM.

To further investigate Wnt activity, we decided to test whether early markers of osteoblastogenesis are regulated by Wnt/β-Catenin signalling. We first assayed how rapidly the lines have the predicted effect on known Wnt/β-Catenin targets. We found that 8 hours treatment starting at 48hpf has a strong effect on expression of both wif1 and axin2 (S3 Fig). Next we looked at expression of runx2a, runx2b and osx and found that only osx expression was altered by heatshock treatment (Fig 4A–4I). To validate this result we used qPCR to quantify the reduction in osx expression after 12 hours of dkk1 over expression and found that expression is reduced to 40% of wild type levels (Fig 4K). We also tested whether col1a2 expression responds rapidly to Wnt/β-Catenin perturbation and found that it was unchanged (S2 Fig). Together these results suggested that Wnt/β-Catenin signalling acts to promote osteoblast differentiation and does so by activating osx expression.

Fig 4. Wnt/β-Catenin signalling regulates osx expression.

(A-I) Larvae were treated at 48hpf and fixed 12hours later. Expression of runx2a (A-C) and runx2b (D-F) is unchanged after inhibition (hs:dkk1) or over activation (hs:wnt8) of Wnt/β-Catenin signalling. (G-L) Expression of osx is reduced in hs:dkk1 and increased in hs:wnt8 larvae 12 hours after the treatment. Arrowheads point to the opercle in the high magnification images to the right of each panel. Abbreviations: cl = cleithrum. Scale bar = 50μM. (J) Quantification for presence or absence of osx staining in the opercle from G and H. Expression was normalised to the control. (K) qPCR performed in parallel on g and h confirms the effects of Wnt/β-Catenin signalling on osx expression (p<0.01 using Student's t-test). Treatments were normalised to gapdh.

FGF and Wnt/β-Catenin signalling may activate osx expression via an intronic cis-regulatory module

The activation of osx expression by FGF and Wnt/β-Catenin signalling is relatively rapid, suggesting that it may be a direct target of these pathways. To further test this possibility, we performed time courses to compare the kinetics of osx regulation to a known target gene. For FGF signalling we treated embryos with SU5402 from 0 to 2 hours and tracked osx expression relative to that of pea3 (Fig 5A). For Wnt/β-Catenin signalling we compared osx to the TOP:GFP transgene after over expression of dkk1 (Fig 5B). The TOP:GFP transgene contains a β-Catenin responsive promoter that drives GFP expression [57]. In both situations osx down regulation matched that of the known target gene, suggesting that osx is a transcriptional target of both pathways.

Fig 5. osx is likely to be a direct target of FGF and Wnt/β-Catenin signalling.

(A) Down regulation of osx and pea3 by SU5402 show the same kinetics as monitored by qPCR over a two hour period. (B) Down regulation of osx and TOP:GFP in hs:dkk1 fish shows the same kinetics as monitored by qPCR over a six hour period. (C) A diagram of the osx gene centred on the first start codon in exon 1 (white box). Light blue shading shows conservation between zebrafish and medaka genomic sequence ( Putative Lef/Tcf, Ets1/2 and Runx2 binding sites are indicated above. The amplicons used for ChIP analysis are indicated below, the numbers represent the approximate centre of the amplicon in relation to the start. (D, E) β-Catenin (at 53hpf) and Ets1/2 (at 53hpf) preferentially bind to osx genomic sequences when compared to the unrelated gene her9. The input bar is the ratio of her9 amplicon to osx amplicon before antibody pull down and the ChIP bar is that ratio after pull down. For normalisation, the input ratio of her9 to osx amplicon is set to 1. Panel E also shows ChIP from fish treated from 51-53hpf with SU5402 which have a mild reduction in Ets1/2 binding activity. p<0.015 for β-CatChIP at osx1146 and p<0.001 for EtsChIP at osx1146.

We next used antibodies to β-Catenin to test whether conserved regions of the osx genomic sequence ( are enriched by ChIP (β-CatChIP). We tested three regions (approximate positions at -13600, -8250 and +97 in relation to the first start codon) and based upon semi-quantitative PCR results (data not shown) focused on the conserved region closest to the start codon. We designed 4 amplicons in this region and performed ChIP with DNA from the hs:dkk1 and hs:wnt8a lines (Fig 5C and 5D and S4A Fig). We found that at all locations, pull-down by β-Catenin was sensitive to dkk1 and wnt8a over expression (S4B Fig). The most strongly enriched amplicon is at position +1146 and intriguingly this region is close to a putative Runx2 binding site at +824 and is outside of the conserved region of intron 1(Fig 5C).

The mitogen-activated kinase pathway is primarily activated by the FGF receptor in the embryo, and Ets (v-ets erythroblastosis virus E26 oncogene homolog) transcription factors are activated in response to the FGF/Mapk pathway [5863]. Consistent with a role in osx regulation, the Ets factors pea3, erm, ets1a and ets2 are all expressed in the region of the developing skeleton at 54hpf (S1 and S5 Figs). To test whether Ets factors bind to osx intron 1, we took advantage of a cross-reactive Ets1/2 antibody to do ChIP (EtsChIP). We found that as with β-CatChIP, EtsChIP preferentially enriched the +1146 amplicon (Fig 5E). Furthermore enrichment was slightly reduced in the presence of SU5402 suggesting that the ability of Ets1/2 to bind this region is partly FGF dependent. The qPCR time course and ChIP experiments support the model that both FGF and Wnt/β-Catenin directly activate osx expression via a shared intronic cis-regulatory module.

Crosstalk between the Wnt/β-Catenin and FGF pathway

Having established that both pathways are likely to act in parallel via intron 1, we wondered whether there is any additional crosstalk between the two pathways which may modify osx expression. We first performed epistasis to see whether FGF and Wnt/β-Catenin act sequentially. We combined SU5402 and hs:wnt8a treatments and found that wnt8a activation of osx expression is blocked by SU5402 and is therefore FGF dependent (Fig 6A–6D). This would suggest that FGF acts downstream of Wnt/β-Catenin signalling. However, when we combined hs:dkk1with hs:fgf3 we found that fgf3 activation of osx is dependent upon Wnt/β-Catenin signalling (Fig 6E–6H). Together these data indicates that neither pathway is sufficient to induce osx expression on its own. One model to explain these results is that ETS factors and β-Catenin interact in the nucleus to form a complex on the osx gene.

Fig 6. FGF and Wnt/β-Catenin signalling interact to regulate osteoblast differentiation.

(A-D) Larvae were heat shocked at 48hpf, treated with SU5402 at 50hpf and fixed at 52hpf. Expression of osx is reduced after SU5402 (C) treatment and increased in hs:wnt8 larvae (B). Combined SU5402 and hs:wnt8 treatment still results in reduced expression of osx (D). (E-H) Larvae were heat shocked at 49hpf and fixed at 52hpf. Expression of osx is reduced in hs:dkk1 larvae (F) and increased in hs:fgf3 larvae (G). Expression is reduced in the combined treatment (H). (I-T) fgf3, sprouty1, sprouty4 and sef expression is inhibited by dkk1 over expression (K, N, Q, T) and activated by over expression of wnt8 (J, M, P, S). Arrowheads point to the opercle in the high magnification images to the right of each panel. Scale bars = 200μM.

We next checked to see whether Wnt/β-Catenin affects activity of the FGF pathway and vice versa. We found that wnt8a over expression upregulates components of the FGF pathway while dkk1 downregulates the same components (Fig 6I–6T). As many components of the FGF pathway are themselves regulated by FGF signalling (through negative feed back), it is difficult to say conclusively that Wnt/β-Catenin is having a positive affect on the pathway as a whole. However, given the timing of the experiments it seems likely that that Wnt/β-Catenin signalling has the capacity to indirectly influence osx expression by increasing FGF activity. To test whether a reciprocal interaction takes place, we tested whether expression of components of the Wnt/β-Catenin pathway are affected by SU5402 treatment, hs:fgf3 or hs:dnfgfr1 treatment (S6 Fig). None of these treatments altered expression of wif, axin2, lef1 or tcf7 suggesting that FGF signalling does not modulate the activity of the Wnt/β-Catenin pathway in this context.


Specification of the osteoblast fate choice in mesenchymal stem cells is a multistep process which involves several developmental signalling pathways. Here we investigate this process in the developing facial skeleton and show that both the FGF and Wnt/β-Catenin pathways act on osteoblast precursors to promote bone formation. We have found that manipulation of both pathways has an immediate and strong effect on osx expression while having a lesser effect on runx2 and bone matrix genes. These finding suggest that these pathways both act predominantly at an intermediate level of osteoblast specification after recruitment to the osteoblast lineage. We show that FGF and Wnt/β-Catenin dependant expression of osx has the same kinetics as that of known read-outs suggesting that osx is a direct target of both pathways. To identify the precise mechanism of action, we have performed ChIP experiments and identified a region of the osx first intron that is bound by β-Catenin and ETS1/2 in pull down experiments. Thus we have demonstrated that both pathways are likely to interact at the level of osx gene transcription to modulate osteoblast differentiation in vivo (Fig 7). In addition to this interaction on the DNA, we have also found that there is a second potential level of regulation by which the Wnt/β-Catenin pathway acts to modulate activity of the FGF pathway during facial skeletal development.

Fig 7. A model for how interactions between FGF and Wnt/β-Catenin signalling pathways interact to regulate osx transcription.

Many studies on the roles of these pathways in skeletogenesis have not yet revealed a clear picture of how they interact to coordinate osteoblast differentiation. These conflicting results may be explained in part by complex feedback loops and redundancies that may obscure results. Our study aimed to identify the earliest responses to signalling perturbation to help to tease apart these interactions and give a clearer picture of the direct effects of these pathways. We have also utilised an in vivo system to avoid artefacts that are inherent in cell-based studies. One surprising outcome of our study is the relatively weak influence that the Wnt/β-Catenin and FGF pathways exert on runx2a/b expression. A consensus of studies in mammals have shown that both pathways activate runx2 both in vivo and in vitro (reviewed in [64]). This difference may be because our study focuses upon the head skeleton which is primarily neural crest derived, or may indicate a more fundamental difference between zebrafish and mammalian osteoblast differentiation.

One fundamental question in skeletal biology is how signalling pathways regulate the morphogenesis of bones in the early embryo. One recent study showed that indian hedgehog a (ihha) drives proliferation of cells surrounding the distal edge of the opercle and in ihha mutants morphogenesis of the opercle is disrupted[65]. Consistent with this role for Hedgehog signalling, another study has shown that ectopic bone forms on the opercle in patched1 mutants in which Hedgehog signalling is elevated[66]. Intriguingly, neither study found that Hedgehog signalling regulates expression of runx2a, runx2b or osx during opercle osteoblastogenesis suggesting that ihha may act downstream of the Wnt/β-Catenin and FGF pathways. In the future, it will be interesting to identify which FGF and Wnt ligands are expressed around the opercle during this time and to determine how their activity is coordinated with that of ihha to the shape the opercle.

Supporting Information

S1 Fig. Inhibition and over activation of FGF signalling have optimal effects 3 hours after the treatment.

(A-H) Inhibition of FGF signalling, using SU5402 treatment or hs:dnfgfr1 larvae resulted in strong down regulation of pea3 and erm expression 3 hours after the treatment (B, C, F, G). Over activation of FGF signalling in hs:fgf3 larvae resulted in strong upregulation of pea3 and erm after 3 hours (D, H). (I-P) 6 hours after the treatment, inhibition of FGF signalling still resulted in slight down regulation of pea3 and erm expression (J, K, N, O) whereas expression was unchanged in hs:fgf3 larvae (L, P). Scale bar = 200μM.


S2 Fig. Neither FGF nor Wnt signalling affect col1a2 expression.

(A-C) Larvae were treated at 60hpf and fixed 3 hours later. Expression of col1a2 is unaffected 3 hours after inhibition (SU5402) or over activation (hs:fgf3) of FGF signalling. (D-F) Larvae were treated at 60hpf and fixed 12 hours later. Inhibition (hs:dkk1) or over activation (hs:wnt8) of Wnt/β-Catenin signalling also does not affect expression of col1a2. Arrowheads point to the opercle in the high magnification images to the right of each panel. Abbreviations: cl = cleithrum. Scale bar in F = 200μM.


S3 Fig. Wnt/β-Catenin signalling regulates wif1 and axin2 expression.

(A-P) Larvae were treated from 48hpf to 56hpf. Expression of the known Wnt/β-Catenin targets wif1 and axin2 is strongly reduced 8 hours after inhibition (hs:dkk1; D, H, L, P) or increased after over activation (hs:wnt8; B, F, J, N) of Wnt/β-Catenin signalling.


S4 Fig. β-CatChIP preferentially enriches for the +1146 amplicon and is sensitive to Wnt/β-Catenin signalling.

β-CatChIP was performed in hs:dkk1 and hs:wnt8 lines heat shocked at 48hpf then fixed at 54hpf and 60hpf respectively. The ratio of the control amplicon her9 to the osx amplicon in the input was used to normalise the levels after ChIP. (A) Shows that -962, +896 and +1146 all show enrichment after ChIP with +1146 being the highest. (B) β-CatChIP is sensitive to Wnt/β-Catenin signalling. Transgenic ChIP is compared directly to sibling ChIP to show that there is on average a 20% reduction in pull down efficiency in hs:dkk1 fish and a 1.5 fold increase in efficiency in hs:wnt8 fish.


S5 Fig. Expression of the Ets factors elk1, ets1a and ets2.

(A-C) All three factors are expressed in regions around the developing bone at 54hpf, with ets1a and ets2 showing the highest expression.


S6 Fig. Expression of Wnt/β-Catenin target genes is unaffected by FGF signalling.

(A-P) Larvae were heat shocked or treated with SU5402 at 48hpf and fixed at 51hpf. Expression of wif, axin2, lef1 and tcf7 is unaffected by inhibition (hs:dnfgfr1) or over activation (hs:fgf3) of FGF signalling. Scale bar = 200μM.


S1 Table. Sequences of primers used in this study.


Author Contributions

Conceived and designed the experiments: KF PME ML HHR. Performed the experiments: KF PME ML HHR. Analyzed the data: KF PME HHR. Wrote the paper: HHR.


  1. 1. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, et al. (1997) Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell 89: 755–764. pmid:9182763
  2. 2. Nakashima K, Zhou X, Kunkel G, Zhang Z, Deng JM, Behringer RR, et al. (2002) The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108: 17–29. pmid:11792318
  3. 3. Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC, Rosewell IR, et al. (1997) Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell 89: 765–771. pmid:9182764
  4. 4. Nishio Y, Dong Y, Paris M, O'Keefe RJ, Schwarz EM, Drissi H (2006) Runx2-mediated regulation of the zinc finger Osterix/Sp7 gene. Gene 372: 62–70. pmid:16574347
  5. 5. Marie PJ (2003) Fibroblast growth factor signaling controlling osteoblast differentiation. Gene 316: 23–32. pmid:14563548
  6. 6. Stein GS, Lian JB, Stein JL, Van Wijnen AJ, Montecino M (1996) Transcriptional control of osteoblast growth and differentiation. Physiol Rev 76: 593–629. pmid:8618964
  7. 7. Lapunzina P, Aglan M, Temtamy S, Caparros-Martin JA, Valencia M, Leton R, et al. (2010) Identification of a frameshift mutation in Osterix in a patient with recessive osteogenesis imperfecta. Am J Hum Genet 87: 110–114. pmid:20579626
  8. 8. Su N, Du X, Chen L (2008) FGF signaling: its role in bone development and human skeleton diseases. Front Biosci 13: 2842–2865. pmid:17981758
  9. 9. Jackson RA, Nurcombe V, Cool SM (2006) Coordinated fibroblast growth factor and heparan sulfate regulation of osteogenesis. Gene 379: 79–91. pmid:16797878
  10. 10. Zhang X, Sobue T, Hurley MM (2002) FGF-2 increases colony formation, PTH receptor, and IGF-1 mRNA in mouse marrow stromal cells. Biochem Biophys Res Commun 290: 526–531. pmid:11779203
  11. 11. Xiao G, Jiang D, Thomas P, Benson MD, Guan K, Karsenty G, et al. (2000) MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1. J Biol Chem 275: 4453–4459. pmid:10660618
  12. 12. Kim HJ, Kim JH, Bae SC, Choi JY, Ryoo HM (2003) The protein kinase C pathway plays a central role in the fibroblast growth factor-stimulated expression and transactivation activity of Runx2. J Biol Chem 278: 319–326. pmid:12403780
  13. 13. Shimizu-Sasaki E, Yamazaki M, Furuyama S, Sugiya H, Sodek J, Ogata Y (2001) Identification of a novel response element in the rat bone sialoprotein (BSP) gene promoter that mediates constitutive and fibroblast growth factor 2-induced expression of BSP. J Biol Chem 276: 5459–5466. pmid:11087753
  14. 14. Jeon E, Yun YR, Kang W, Lee S, Koh YH, Kim HW, et al. (2012) Investigating the role of FGF18 in the cultivation and osteogenic differentiation of mesenchymal stem cells. PLoS One 7: e43982. pmid:22937141
  15. 15. Yu K, Xu J, Liu Z, Sosic D, Shao J, Olson EN, et al. (2003) Conditional inactivation of FGF receptor 2 reveals an essential role for FGF signaling in the regulation of osteoblast function and bone growth. Development 130: 3063–3074. pmid:12756187
  16. 16. Ohbayashi N, Shibayama M, Kurotaki Y, Imanishi M, Fujimori T, Itoh N, et al. (2002) FGF18 is required for normal cell proliferation and differentiation during osteogenesis and chondrogenesis. Genes Dev 16: 870–879. pmid:11937494
  17. 17. Eswarakumar VP, Monsonego-Ornan E, Pines M, Antonopoulou I, Morriss-Kay GM, Lonai P (2002) The IIIc alternative of Fgfr2 is a positive regulator of bone formation. Development 129: 3783–3793. pmid:12135917
  18. 18. Rodan SB, Wesolowski G, Yoon K, Rodan GA (1989) Opposing effects of fibroblast growth factor and pertussis toxin on alkaline phosphatase, osteopontin, osteocalcin, and type I collagen mRNA levels in ROS 17/2.8 cells. J Biol Chem 264: 19934–19941. pmid:2479640
  19. 19. Mansukhani A, Bellosta P, Sahni M, Basilico C (2000) Signaling by fibroblast growth factors (FGF) and fibroblast growth factor receptor 2 (FGFR2)-activating mutations blocks mineralization and induces apoptosis in osteoblasts. J Cell Biol 149: 1297–1308. pmid:10851026
  20. 20. Ambrosetti D, Holmes G, Mansukhani A, Basilico C (2008) Fibroblast growth factor signaling uses multiple mechanisms to inhibit Wnt-induced transcription in osteoblasts. Mol Cell Biol 28: 4759–4771. pmid:18505824
  21. 21. Milat F, Ng KW (2009) Is Wnt signalling the final common pathway leading to bone formation? Mol Cell Endocrinol 310: 52–62. pmid:19524639
  22. 22. Hartmann C (2007) Skeletal development—Wnts are in control. Mol Cells 24: 177–184. pmid:17978569
  23. 23. Lerner UH, Ohlsson C (2015) The WNT system: background and its role in bone. J Intern Med 277: 630–649. pmid:25845559
  24. 24. Rodda SJ, McMahon AP (2006) Distinct roles for Hedgehog and canonical Wnt signaling in specification, differentiation and maintenance of osteoblast progenitors. Development 133: 3231–3244. pmid:16854976
  25. 25. Hill TP, Spater D, Taketo MM, Birchmeier W, Hartmann C (2005) Canonical Wnt/beta-catenin signaling prevents osteoblasts from differentiating into chondrocytes. Dev Cell 8: 727–738. pmid:15866163
  26. 26. Hu H, Hilton MJ, Tu X, Yu K, Ornitz DM, Long F (2005) Sequential roles of Hedgehog and Wnt signaling in osteoblast development. Development 132: 49–60. pmid:15576404
  27. 27. Day TF, Guo X, Garrett-Beal L, Yang Y (2005) Wnt/beta-catenin signaling in mesenchymal progenitors controls osteoblast and chondrocyte differentiation during vertebrate skeletogenesis. Dev Cell 8: 739–750. pmid:15866164
  28. 28. Gong Y, Slee RB, Fukai N, Rawadi G, Roman-Roman S, Reginato AM, et al. (2001) LDL receptor-related protein 5 (LRP5) affects bone accrual and eye development. Cell 107: 513–523. pmid:11719191
  29. 29. Rawadi G, Vayssiere B, Dunn F, Baron R, Roman-Roman S (2003) BMP-2 controls alkaline phosphatase expression and osteoblast mineralization by a Wnt autocrine loop. J Bone Miner Res 18: 1842–1853. pmid:14584895
  30. 30. Bennett CN, Longo KA, Wright WS, Suva LJ, Lane TF, Hankenson KD, et al. (2005) Regulation of osteoblastogenesis and bone mass by Wnt10b. Proc Natl Acad Sci U S A 102: 3324–3329. pmid:15728361
  31. 31. Gaur T, Lengner CJ, Hovhannisyan H, Bhat RA, Bodine PV, Komm BS, et al. (2005) Canonical WNT signaling promotes osteogenesis by directly stimulating Runx2 gene expression. J Biol Chem 280: 33132–33140. pmid:16043491
  32. 32. De Boer J, Wang HJ, Van Blitterswijk C (2004) Effects of Wnt signaling on proliferation and differentiation of human mesenchymal stem cells. Tissue Eng 10: 393–401. pmid:15165456
  33. 33. Boland GM, Perkins G, Hall DJ, Tuan RS (2004) Wnt 3a promotes proliferation and suppresses osteogenic differentiation of adult human mesenchymal stem cells. J Cell Biochem 93: 1210–1230. pmid:15486964
  34. 34. Liu G, Vijayakumar S, Grumolato L, Arroyave R, Qiao H, Akiri G, et al. (2009) Canonical Wnts function as potent regulators of osteogenesis by human mesenchymal stem cells. J Cell Biol 185: 67–75. pmid:19349579
  35. 35. Stevens JR, Miranda-Carboni GA, Singer MA, Brugger SM, Lyons KM, Lane TF (2010) Wnt10b deficiency results in age-dependent loss of bone mass and progressive reduction of mesenchymal progenitor cells. J Bone Miner Res 25: 2138–2147. pmid:20499361
  36. 36. Tan SH, Senarath-Yapa K, Chung MT, Longaker MT, Wu JY, Nusse R (2014) Wnts produced by Osterix-expressing osteolineage cells regulate their proliferation and differentiation. Proc Natl Acad Sci U S A 111: E5262–5271. pmid:25422448
  37. 37. Zhang C, Cho K, Huang Y, Lyons JP, Zhou X, Sinha K, et al. (2008) Inhibition of Wnt signaling by the osteoblast-specific transcription factor Osterix. Proc Natl Acad Sci U S A 105: 6936–6941. pmid:18458345
  38. 38. Hammond CL, Schulte-Merker S (2009) Two populations of endochondral osteoblasts with differential sensitivity to Hedgehog signalling. Development 136: 3991–4000. pmid:19906866
  39. 39. Spoorendonk KM, Peterson-Maduro J, Renn J, Trowe T, Kranenbarg S, Winkler C, et al. (2008) Retinoic acid and Cyp26b1 are critical regulators of osteogenesis in the axial skeleton. Development 135: 3765–3774. pmid:18927155
  40. 40. Laue K, Janicke M, Plaster N, Sonntag C, Hammerschmidt M (2008) Restriction of retinoic acid activity by Cyp26b1 is required for proper timing and patterning of osteogenesis during zebrafish development. Development 135: 3775–3787. pmid:18927157
  41. 41. Li N, Kelsh RN, Croucher P, Roehl HH (2010) Regulation of neural crest cell fate by the retinoic acid and Pparg signalling pathways. Development 137: 389–394. pmid:20081187
  42. 42. Pinto JP, Conceicao NM, Viegas CS, Leite RB, Hurst LD, Kelsh RN, et al. (2005) Identification of a new pebp2alphaA2 isoform from zebrafish runx2 capable of inducing osteocalcin gene expression in vitro. J Bone Miner Res 20: 1440–1453. pmid:16007341
  43. 43. Li N, Felber K, Elks P, Croucher P, Roehl HH (2009) Tracking gene expression during zebrafish osteoblast differentiation. Dev Dyn 238: 459–466. pmid:19161246
  44. 44. Flores MV, Tsang VW, Hu W, Kalev-Zylinska M, Postlethwait J, Crosier P, et al. (2004) Duplicate zebrafish runx2 orthologues are expressed in developing skeletal elements. Gene Expr Patterns 4: 573–581. pmid:15261836
  45. 45. van der Meulen T, Kranenbarg S, Schipper H, Samallo J, van Leeuwen JL, Franssen H (2005) Identification and characterisation of two runx2 homologues in zebrafish with different expression patterns. Biochim Biophys Acta 1729: 105–117. pmid:15894389
  46. 46. DeLaurier A, Eames BF, Blanco-Sanchez B, Peng G, He X, Swartz ME, et al. (2010) Zebrafish sp7:EGFP: a transgenic for studying otic vesicle formation, skeletogenesis, and bone regeneration. Genesis 48: 505–511. pmid:20506187
  47. 47. Renn J, Winkler C (2009) Osterix-mCherry transgenic medaka for in vivo imaging of bone formation. Dev Dyn 238: 241–248. pmid:19097055
  48. 48. Cubbage CC, Mabee PM (1999) Development of the cranium and paired fins in the zebrafish Danio rerio (Ostariophysi, Cyprinidae). Journal of Morphology 229: 121–160.
  49. 49. Witten PE, Huysseune A (2009) A comparative view on mechanisms and functions of skeletal remodelling in teleost fish, with special emphasis on osteoclasts and their function. Biol Rev Camb Philos Soc 84: 315–346. pmid:19382934
  50. 50. Huysseune A (2000) Skeletal System/Microscopic Functional Anatomy. In: Ostrander G, editor. The Laboratory Fish. New York: Academic Press. pp. 307–317.
  51. 51. Nüsslein-Volhard C, Dahm R (2002) Zebrafish: a practical approach. Oxford: Oxford University Press. xviii, 303 p. p.
  52. 52. Lee Y, Grill S, Sanchez A, Murphy-Ryan M, Poss KD (2005) Fgf signaling instructs position-dependent growth rate during zebrafish fin regeneration. Development 132: 5173–5183. pmid:16251209
  53. 53. Avaron F, Hoffman L, Guay D, Akimenko MA (2006) Characterization of two new zebrafish members of the hedgehog family: atypical expression of a zebrafish indian hedgehog gene in skeletal elements of both endochondral and dermal origins. Dev Dyn 235: 478–489. pmid:16292774
  54. 54. Roehl H, Nusslein-Volhard C (2001) Zebrafish pea3 and erm are general targets of FGF8 signaling. Curr Biol 11: 503–507. pmid:11413000
  55. 55. Weidinger G, Thorpe CJ, Wuennenberg-Stapleton K, Ngai J, Moon RT (2005) The Sp1-related transcription factors sp5 and sp5-like act downstream of Wnt/beta-catenin signaling in mesoderm and neuroectoderm patterning. Curr Biol 15: 489–500. pmid:15797017
  56. 56. Stoick-Cooper CL, Weidinger G, Riehle KJ, Hubbert C, Major MB, Fausto N, et al. (2007) Distinct Wnt signaling pathways have opposing roles in appendage regeneration. Development 134: 479–489. pmid:17185322
  57. 57. Dorsky RI, Sheldahl LC, Moon RT (2002) A transgenic Lef1/beta-catenin-dependent reporter is expressed in spatially restricted domains throughout zebrafish development. Dev Biol 241: 229–237. pmid:11784107
  58. 58. Wasylyk B, Hagman J, Gutierrez-Hartmann A (1998) Ets transcription factors: nuclear effectors of the Ras-MAP-kinase signaling pathway. Trends Biochem Sci 23: 213–216. pmid:9644975
  59. 59. Sawada A, Shinya M, Jiang YJ, Kawakami A, Kuroiwa A, Takeda H (2001) Fgf/MAPK signalling is a crucial positional cue in somite boundary formation. Development 128: 4873–4880. pmid:11731466
  60. 60. Christen B, Slack JM (1999) Spatial response to fibroblast growth factor signalling in Xenopus embryos. Development 126: 119–125. pmid:9834191
  61. 61. Simon MA (2000) Receptor tyrosine kinases: specific outcomes from general signals. Cell 103: 13–15. pmid:11051543
  62. 62. LaBonne C, Burke B, Whitman M (1995) Role of MAP kinase in mesoderm induction and axial patterning during Xenopus development. Development 121: 1475–1486. pmid:7789277
  63. 63. Umbhauer M, Marshall CJ, Mason CS, Old RW, Smith JC (1995) Mesoderm induction in Xenopus caused by activation of MAP kinase. Nature 376: 58–62. pmid:7541116
  64. 64. Komori T (2011) Signaling networks in RUNX2-dependent bone development. J Cell Biochem 112: 750–755. pmid:21328448
  65. 65. Huycke TR, Eames BF, Kimmel CB (2012) Hedgehog-dependent proliferation drives modular growth during morphogenesis of a dermal bone. Development 139: 2371–2380. pmid:22627283
  66. 66. Felber K, Croucher P, Roehl HH (2011) Hedgehog signalling is required for perichondral osteoblast differentiation in zebrafish. Mechanisms of development 128: 141–152. pmid:21126582