evidence that fgf10 contributes to the skeletal and...

SPECIAL ISSUE DISEASE CONNECTIONS

Evidence That Fgf10 Contributes to theSkeletal and Visceral Defects of an ApertSyndrome Mouse ModelMohammad K. Hajihosseini,1* Raquel Duarte,2† Jean Pegrum,2 Anne Donjacour,3 Eva Lana-Elola,4

David P. Rice,4‡ James Sharpe,5 and Clive Dickson2

Apert syndrome (AS) is a severe congenital disease caused by mutations in fibroblast growth factorreceptor-2 (FGFR2), and characterised by craniofacial, limb, visceral, and neural abnormalities. AS-typeFGFR2 molecules exert a gain-of-function effect in a ligand-dependent manner, but the causative FGFs andtheir relative contribution to each of the abnormalities observed in AS remains unknown. We havegenerated mice that harbour an AS mutation but are deficient in or heterozygous for Fgf10. The geneticknockdown of Fgf10 can rescue the skeletal as well as some of the visceral defects observed in this AS model,and restore a near normal level of FgfR2 signaling involving an apparent switch between ERK(p44/p42) andp38 phosphorylation. Surprisingly, it can also yield de novo cleft palate and blind colon in a subset of thecompound mutants. These findings strongly suggest that Fgf10 contributes to AS-like pathologies andhighlight a complexity of Fgf10 function in different tissues. Developmental Dynamics 238:376–385, 2009.© 2008 Wiley-Liss, Inc.

Key words: FGFR2 signaling; FGF10; Apert syndrome; craniosynostosis; sternum; lung

Accepted 9 June 2008

INTRODUCTION

Mutations that perturb the levelsof FGFR1, FGFR2, or FGFR3 signal-ing during embryonic developmentare the underlying causes of sev-eral congenital craniosynostosis syn-dromes, including: Jackson-Weiss,Beare-Stevenson, Muenke, Crou-zon, Pfeiffer, and Apert syndromes(Wilkie, 2005; Passos-Bueno et al.,2008). Apert syndrome is the most

severe, and in addition to craniosyn-ostosis—the premature fusion of cal-varial bones at their intervening su-tures—is characterised by a host ofvisceral and neural defects (Cohenand Kreiborg, 1993a; Cohen andKreiborg, 1993b; Ornitz and Marie,2002). In contrast to its related syn-dromes, AS is caused exclusively bymutant FGFR2 molecules that needto bind FGF ligands in order to exert

their gain-of-function effects (Ander-son et al., 1998; Yu et al., 2000; Ibra-himi et al., 2001, 2004). Identifyingthese ligands and their contributionto each of the abnormalities ob-served in AS is important as this willfurther our understanding of the mo-lecular etiology of AS and may helpdevise strategies to alleviate AS-re-lated pathologies.

The mammalian FGF signaling sys-

Additional Supporting Information may be found in the online version of this article.1School of Biological Sciences, University of East Anglia, Norwich, United Kingdom2Cancer Research UK, London, United Kingdom3Department of Anatomy, University of California, San Francisco, California4Department of Orthodontics and Craniofacial Development, King’s College, London, United Kingdom5ICREA, EMBL-CRG Systems Biology Unit, Centre for Genomic Regulation, UPF, Barcelona, Spain.Grant sponsor: CRUK; Grant sponsor: The Royal Society; Grant sponsor: John & Pamela Salter Charitable Trust.†Raquel Duarte’s present address is Department of Internal Medicine, University of the Witwatersrand, 7 York Road, Parktown 2193, SouthAfrica.‡David P. Rice’s present address is Department of Orthodontics, University of Helsinki, Helsinki, Finland.*Correspondence to: Mohammad K. Hajihosseini, School of Biological Sciences, University of East Anglia, Norwich, NR4 7TJ,UK. E-mail: [email protected]

DOI 10.1002/dvdy.21648Published online 4 September 2008 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 238:376–385, 2009

© 2008 Wiley-Liss, Inc.

tem is comprised of 18 extracellulary-acting ligands and four receptors (FG-FRs 1–4) (Itoh and Ornitz, 2008). Atypical FGFR molecule is composed ofan extracellular domain harbouringtwo or three immunoglobulin (Ig)-likeregions, a transmembrane element andan intracytoplasmic tyrosine kinase do-main. FGFR signaling is triggered bydimerization of FGFR molecules uponbinding to a complex of FGF ligandsand sulphated proteoglycans (McKee-han et al., 1998; Plotnikov et al., 1999),and is transduced cytoplasmically byMap kinases, PI3 kinase, and PLC-�pathways. FGFR signaling can operateboth in a morphogen- and a threshold-dependent manner and is fine-tuned inpart by the activity of Map kinase phos-phatase 3 (Mkp3/Pyst1) and Sproutys(Spry) 1-4 (Hajihosseini et al., 2004;Tsang and Dawid, 2004; Hajihosseini,2008).

FGFR genes encode numeroussplice variants with distinct FGF-binding specificities and biological ac-tivities (McKeehan et al., 1998; Zhanget al., 2006). For example, the so-called IIIb and IIIc isoforms of FGFR2arise from alternative splicing of ex-ons that encode the carboxyl half ofthe third Ig-like domain. FgfR2-IIIb isnormally expressed by epithelial cellsand is activated by Fgfs -3, -7, -10, and-22 derived predominantly from theadjacent mesenchymal cells. By con-trast, FgfR2-IIIc is predominant inmesenchymal and neural tissues, andis activated by epithelially-derived li-gands such as Fgfs -2, -4, -8, -9, -18,and -20 (Peters et al., 1992; Orr-Urtre-ger et al., 1993; Zhang et al., 2006).

Most AS cases arise from Ser252Trp(S252W) or Pro253Arg (P253R) sub-stitutions in the region linking the Ig-like domain II and III of FGFR2,which induce the mutant receptors tobind their cognate FGF ligands with ahigher affinity, or lose specificity andinteract with a broader set of FGFligands (Anderson et al., 1998; Yu etal., 2000; Ibrahimi et al., 2001, 2004).As heterozygous mutations, the mu-tant receptors act in the presence ofwild type receptors to cause a net in-crease in the length or strength of theFGFR2 signal within mesenchymaltissues. More rarely, AS is caused byheterozygous Alu-insertions withinFGFR2-IIIc exon, or a heterozygous1.9-kb deletion of FGFR2-IIIc exon

and its flanking intronic sequences(Oldridge et al., 1999; Bochukova etal. 2008). These rare mutations inducean illegitimate expression of FGFR2-IIIb isoform alongside FGFR2-IIIc inmesenchymal cells (Oldridge et al.,1999; Bochukova et al. 2008), whichrenders the cells responsive to bothIIIb- and IIIc-activating FGFs. Be-cause FGFR2-IIIb and IIIc isoformshave identical cytoplasmic domains,stimulation by a broader set of FGFligands is perceived as a net gain-of-FGFR2 signaling in the affected cells.

Calvarial bones form by direct dif-ferentiation of mesenchymal cells intoosteoblasts (Ornitz and Marie, 2002).Within the calvarial sutures, a broadrange of FgfR2-IIIb and FgfR2-IIIc ac-tivating ligands are present (Hajihos-seini and Heath, 2002), but a distinctlevel of FgfR2-IIIc signaling per se isrequired to maintain mesenchymalosteoprogenitor stem cells and suturepatency (Iseki et al., 1997, 1999; Riceet al., 2003). In AS, perturbed levels ofFGFR2 signaling results in the pre-mature differentiation of sutural os-teoprogenitors into bone and the fu-sion of calvarial plates.

We have previously described micethat harbour a heterozygous 1.3-kb de-letion of FgfR2-IIIc exon and its flank-ing introns (i.e., FgfR2-IIIc�/� ; hereonreferred to as IIIc�/�) and show illegiti-mate expression of FgfR2-IIIb in mes-enchymal and neural tissues (Hajihos-seini et al., 2001). These mice exhibitApert syndrome-like phenotypes in-cluding fusion of the coronal and facialsutures, premature ossification of ster-nal bones, and branching-morphogene-sis defects in several visceral organs(Hajihosseini et al., 2001; Jaskoll et al.,2002; De Langhe et al., 2006). More re-cent analysis has also revealed a per-turbation of neurogenesis in the brainthat will be reported elsewhere. IIIc�/�

mice show postnatal growth retarda-tion and die within a few days of birth(Hajihosseini et al., 2001).

We now show that a genetic knock-down of Fgf10 expression, a ligand thatactivates FgfR2-IIIb, can rescue notonly the craniofacial but also the ster-nal and some of the visceral defects ob-served in IIIc�/� mice. Using the lungsfor detailed analysis, we show that thephenotypic rescue is reflected in a nearnormal restoration of FGFR2 signalinglevels that involves an antagonism be-

tween ERK (p44/p42) and p38 phos-phorylation. Surprisingly, Fgf10 knock-down also results in de novo cleft palateand blind colon in a subset of IIIc�/�

mice. These findings strongly implicateFgf10 in craniosynostosis and other AS-related pathologies, and highlight a di-verse and complex function of Fgf10 indifferent tissues.

RESULTS

Abrogation of Fgf10Function in FgfR2-IIIc�/�

Mice RescuesCraniosynostosis

It has been shown that a heterozygousloss of FgfR2-IIIc-function does not re-sult in craniosynostosis (Eswaraku-mar et al., 2002). In IIIc�/� mice,FgfR2-IIIb is upregulated in mesen-chymal tissues including calvarial su-tures (Hajihosseini et al., 2001; DeLanghe et al., 2006; and see below)and so the IIIc�/� craniosynostosislikely results from a gain-of-FgfR2function, requiring the combined ac-tions of FgfR2-IIIc and -IIIb signalingand their activating ligands. We havepreviously shown that a host of Fgfligands are indeed expressed in thedeveloping sutures (Hajihosseini andHeath, 2002) and so it be would ex-pected that a genetic knockdown ofthe key activating ligand/s in IIIc�/�

mice would rescue craniosynostosis.To test this hypothesis and towardsidentifying such ligands, we gener-ated double transgenic mice that ab-errantly express FgfR2-IIIb but lackits major ligand, Fgf10. i.e., FgfR2-IIIc�/�; Fgf10�/� mice (hereon abbre-viated to IIIc�/�; Fgf10�/�; Fig. 1A;see Supplemental Table 1, which canbe viewed online). Noteworthy is thatloss of Fgf10 per se also does not ap-pear to affect normal calvarial devel-opment (Fig. 2A; see also Discussionsection), even though Fgf10 null micedie perinatally from the agenesis ordysgenesis of multiple organs includ-ing lungs and limbs (Min et al., 1998;Ohuchi et al., 2000).

As expected, IIIc�/�; Fgf10�/� dou-ble mutants died at birth and werereadily recognisable by the absence oflimbs (Fig. 2F,F�), although all geno-types were confirmed by PCR analysis(Fig. 1B; Suppl. Table 2). Remarkably,these double mutants (n � 4/4) pre-

INVOLVEMENT OF FGF10 IN APERT-LIKE PATHOLOGY 377

sented patent coronal sutures similarto wild type (WT) and uncharacteristicof the IIIc�/� phenotype (Fig. 2D,E).Moreover, fusion of zygomatic archand maxilla-premaxillary joints, nor-mally observed in IIIc�/� mice (Fig.2G,H), was also reverted in these dou-ble mutants. These observations sug-gest that deletion of Fgf10 neutralizesthe ectopic FgfR2-IIIb signaling inIIIc�/� animals and reverts thecraniofacial defects associated withthis gain-of-function allele.

A Heterozygous Knockdownof Fgf10 Can Also Rescuethe Skeletal and SomeVisceral Defects in IIIc�/�

Mice

By selective breeding, we also ob-tained mice that are heterozygous or

wild type for Fgf10 on the IIIc�/�

background (i.e., IIIc�/�; Fgf10�/�

and, IIIc�/�; Fgf10�/� (or simplyIIIc�/�); Fig. 1A). Like IIIc�/� mice,all IIIc�/�; Fgf10�/� mice died a fewdays after birth with no difference insurvival rates. However, we were sur-prised to find that nearly half of these(n � 13/28) lacked the dome-shapedhead, truncated mid-face, and pro-truding eyes that are characteristic ofIIIc�/� mice (Hajihosseini et al., 2001)and instead resembled WT. Skeletalstaining and sectioning confirmed thepresence of patent coronal and facialsutures/joints in these mice (Figs.2J,K, 3B).

The gain-of-FgfR2 function inIIIc�/� mice also induces precociousossification of inter-sternebral carti-lage (Hajihosseini et al., 2001) (Fig.2I, I�), and as in calvaria, Fgf10 ap-

pears not to be required for normalsternal development (Fig. 2C). In over80% (n � 16/19) of IIIc�/�; Fgf10�/�

mice, the sternal abnormality was re-versed, with 58% (n � 11/19) showingfull rescue (Fig. 2L; Suppl. Table 3).These findings show that a reductionin Fgf10 gene dosage can revert notonly the membranous (craniofacial)but also the endochondral bone pheno-types of the IIIc�/� mice.

IIIc�/� mice also carry secondarybranching defects in the lungs, lacri-mal glands (LG), kidneys, and sub-mandibular glands (SMG) (Hajihos-seini et al., 2001; Jaskoll et al., 2002).Complete loss of Fgf10 results in theagenesis or dysgenesis of these organs(Min et al., 1998; Ohuchi et al., 2000)and so their potential rescue could notbe investigated in IIIc�/�; Fgf10�/�

double mutants. However, next weasked whether a reduction in theFgf10 gene dosage can rescue thesevisceral defects in IIIc�/�; Fgf10�/�

mice.In IIIc�/� mice, the cranial, middle,

and caudal lobes of the right lung failto separate while the accessory lobe iscompletely absent (Fig. 4D) (Hajihos-seini et al., 2001). Also, both the leftand right lobes of IIIc�/� lungs carryan abnormally compact mesenchymewith reduced alveologenesis (Fig.4E,E�,F; Suppl. Fig. 1), which appearsto result from aberrant mesenchymalcell differentiation, elevated Spry4 ex-pression, and perturbed Wnt signal-ing (De Langhe et al., 2006).

In 90% (n � 39/43) of IIIc�/�;Fgf10�/� lungs, the cranial, middle,and caudal lobes were found to be dis-tinct, as in WT (Fig. 4G; Suppl. Table3). Examination of conventional his-tology and optical projection tomo-graphic sections (OPT) (Sharpe et al.,2002) through these lungs further re-vealed a well-branched pattern of al-veologenesis throughout the left andright lobes (Fig. 4H,H�,I; Suppl. Fig.1). The remaining 10% showed a par-tial rescue.

Surprisingly, the accessory lobefailed to develop in all IIIc�/�;Fgf10�/� lungs (Fig. 4G; data notshown), as in IIIc�/� single mutants.We investigated the cause usingwhole mount in situ hybridization(WMISH) with the epithelial cellmarker Foxa2 (Hnf�3) (Motoyama etal., 1998) and neutral red staining to

Fig. 1. Abrogation or genetic knock down of Fgf10 in FgfR2-IIIc�/� mice. A: Mating schemes used forthe generation of IIIc�/�; Fgf10�/� and IIIc�/�; Fgf10�/� mice, denoted as (i) and (ii), respectively. IIIc�/�

mutants are generated through loxP-CRE excision, with female Fgf10�/� mice providing the Crerecombinase, and males the “floxed” FgfR2-IIIc allele/s. Use of males that are additionally heterozygous(Scheme I) or wild type (Scheme II) for Fgf10, yields a series of allelic combinations, including IIIc�/�

single mutant (iii) and wild type (v) mice. B: Identification of alleles: (i), (ii), and (iii) by PCR. Top set, IIIc�/�

allele (Hajihosseini et al., 2001); middle and lower sets, Fgf10 null and wild type alleles using primers, P1,P2, and P3, whose positions are indicated on a partial genomic map of mouse Fgf10. C: RNAseprotection assays on E17.5 lungs. Top left: Lower levels of protected Fgf10 message is detected ingenotype (ii) compared to (iii) (n � 2/2). Bottom left: Gapdh protected-products as control for RNAloading. Top and Bottom right: Fgf10, Gapdh, and control tRNA probes.

378 HAJIHOSSEINI ET AL.

Fig. 2. Rescue of skeletal defects following genetic knockdown of Fgf10. Alizarin-red stained skeletons of IIIc�/� mutant mice carrying a null (D–F�),WT (G–I�), or heterozygous (J–L) Fgf10 allele, as compared with Fgf10�/� (A–C) and WT (M–O) mice of comparable ages. The mouse shown in D lackslimbs (F, F�) confirming its Fgf10 null genotype. Patency of the coronal sutures (white arrows) is evident as a gap separating the frontal (f) and parietalbones (p) in A,D,J, and M, but not in G. B,E,H,K,N: Lateral views of the left face, showing the patency of joints that separate the maxilla-premaxilla(single arrows) and zygomatic arch bones (double arrows) in all but H. C,I,I�,L,O: Frontal and right-lateral views of the sternum. The IIIc�/� sternum (Iand I�) shows precocious intersternebral ossification at its posterior margin (arrows) as well as retarded longitudinal growth, when compared to IIIc�/�;Fgf10�/� (L), or WT (O). Scale bars � 1 mm in C,I,I,L,O; 3 mm in F and F�; and 2 mm in the rest of the panels.

Fig. 3. Close-up analysis of patent and fused sutures, and distribution of FgfR2-IIIb expression byin situ hybridization. A,B,D: Vibratome sections through E18.5 IIIc�/� (A), IIIc�/�; Fgf10�/� (B), andwild type (D) skulls shown in Figure 2G,J, and M, respectively. Coronal sutures are patent in B andD (white arrows) but not in A (black arrow). C: Paraffin section through an E17.5 non-rescued IIIc�/�;Fgf10�/� suture recapitulating the IIIc�/� phenotype (not shown) compared to WT (E). Comparisonof C with E shows that expression of FgfR2-IIIb is stronger in C and localised to the edges of thebones as well as sutural mesenchymal cells (black arrowhead), displaced to the sides of the fusingparietal and frontal bones. Fig. 6.


detect apoptotic cells (Hajihosseini etal., 2004). We found that at E11.5, anaccessory lobe mesenchymal primordiais present in both IIIc�/� and IIIc�/�;Fgf10�/� embryos, but subsequentlythe mesenchyme appears to degenerateas it fails to become epithelialized (n �3 of each genotype; Fig. 4J,K).

The LG and SMG defects were notrescued by Fgf10 knockdown (Jaskollet al., 2005); i.e., as in IIIc�/� mice, theIIIc�/�; Fgf10�/� LG carried a rudi-mentary mesenchyme void of any ep-ithelial branching (Fig. 4M). The kid-neys showed only a partial andinconsistent rescue (Fig. 4N,O; Suppl.Table 3; data not shown).

These findings show that of thebranching organs examined, the res-cue is most robust and consistent inthe lungs.

Modulation of Fgf10 andFgfR2 Isoform Expression

We wanted to know whether the phe-notypic rescue in IIIc�/�; Fgf10�/�

mice is accompanied by a reduction inFgf10 transcript levels and/or involvesa downregulation of aberrant FgfR2-IIIb expression.

We performed RNAse protection as-says on E17.5 lungs, a tissue in whichthe phenotypic rescue is most consis-tent and readily discernable, and, asexpected, found that Fgf10 transcriptlevels are reduced in IIIc�/�; Fgf10�/�

when compared to IIIc�/� or WT (Fig.1C and data not shown). A similaranalysis on calvarial suture RNAproved difficult for quantitative rea-sons and also because a priori wecould not judge whether a IIIc�/�;Fgf10�/� suture goes on to remainpatent or become fused.

We also investigated potentialchanges in Fgf10 and FgfR2-IIIb and-IIIc isoform expression by radioactiveISH on E17.5 coronal suture and lung

sections. E17.5 represents the onset ofcraniosynostosis in IIIc�/� sutures,and at this age branching is still on-going at the periphery of WT lungs.We also conducted semi-quantitativeRT-PCR on E14.5 brain, where FgfR2-IIIc is highly expressed (Hasegawa etal., 2004), and in a tissue in which wehave found neurogenesis defects (datanot shown).

While the IIIc-specific riboprobe didnot show a sufficiently high signal togive meaningful results, the IIIb-spe-cific probe confirmed that the level ofFgfR2-IIIb expression is higher in cal-varial sutures of IIIc�/� (n � 3) andnon-rescued IIIc�/�; Fgf10�/� whencompared to WT (n � 3) (Fig. 3C,E;and data not shown). The analysis forrescued IIIc�/�; Fgf10�/� suturesproved inconclusive as we could notobtain sufficient numbers or high-quality calvarial sections of this geno-type. Nonetheless, we found thatFgfR2-IIIb does not appear to bedown-regulated in the lung mesen-chyme or brains of IIIc�/�; Fgf10�/�

mutants (Figs. 4H,H and 5A). Finally,and in accordance with the RNAseprotection assays (Fig. 1C), we foundthat the level of Fgf10 is indeed re-duced in IIIc�/�; Fgf10�/� lungs (Fig.4I) when compared to IIIc�/� (Fig. 4F)or WT (Fig. 4C).

Restoration of Near WildType Levels of FgfR2Signaling in IIIc�/�; Fgf10�/�

Lungs

We sought evidence that the rescue ofphenotypes in IIIc�/�; Fgf10�/� micereflects changes in FgfR2 signalingdynamics, by measuring the amountof ERK (p44/p42) and p38 phosphory-lation, two key components of the Mapkinase signal transducer pathways(Tsang and Dawid, 2004). For the rea-sons explained above, we used lungtissue for this analysis. Immunoblot-ting (n � 4) showed that levels of p44and p42 phosphorylation are raised inIIIc�/� by at least fourfold above WTlevels, but become significantly re-duced in IIIc�/�; Fgf10�/� (Fig. 5Bmiddle lane; and data not shown). Bycomparison, p38 phosphorylation wasevident in WT and IIIc�/�; Fgf10�/�

samples (80% of WT levels), but barelydetectable in IIIc�/� samples (8% ofWT levels) (Fig. 5B right-hand lane).

These differences show that withinthe IIIc�/�; Fgf10�/� lungs, FgfR2 sig-naling dynamics move towards WTlevels and this correlates with an ap-parent partial switch between phos-phorylation of p38 and p44/p42.

De Novo Phenotypes in aSubset of IIIc�/�; Fgf10�/�

Mice

In the course of these studies, we weresurprised to encounter two de novophenotypes: a cleft palate and blindcolon, in approximately 25% of new-born IIIc�/�; Fgf10�/� mice (Fig. 6;Suppl. Table 3). These abnormalitiesare not observed in IIIc�/� or Fgf10�/�

single mutant mice and do not resultfrom a complete loss-of-FgfR2-IIIc sig-naling (Eswarakumar et al., 2002),but, invariably they do occur inFgf10�/� and FgfR2-IIIb�/� mice (DeMoerlooze et al., 2000; Ohuchi et al.,2000). We also noted that the IIIc�/�

SMG phenotype is exacerbated inIIIc�/�, Fgf10�/� mice (Jaskoll et al.,2005).

A likely explanation is that inIIIc�/� mice, the ectopic mesenchy-mally expressed FgfR2-IIIb competeswith epithelially expressed FgfR2-IIIbfor Fgf10, and when this competitionis accentuated by reducing the level ofFgf10 ligand in IIIc�/�; Fgf10�/� dou-ble-mutants, occasionally epithelialFgfR2-IIIb signaling may fail, result-ing in phenotypes that recapitulate aloss of Fgf10/ FgfR2-IIIb function.

Some Apert syndrome patientsshow cleft palate and more rarely,ano-rectal anomalies (Cohen andKreiborg, 1993a; Slaney et al., 1996)and our observations therefore raisethe interesting possibility that thesespecific defects may be the result ofloss-of-FGFR2-IIIb function in epithe-lial cells, secondary to alleles thatcause a gain-of-FGFR2 function inmesenchymal cells. Clearly, furtherinvestigations are required to verifythis hypothesis.

DISCUSSION

Apert syndrome is caused by a ligand-dependent perturbation of FGFR2 sig-naling levels during embryonic devel-opment, resulting in a host of skeletal,visceral, and neural defects (Cohenand Kreiborg, 1993a,b; Ornitz and

Fig. 6. Occurrence of de novo phenotypes in asubset of FgfR2-IIIc�/�; Fgf10�/� mice. A–C:Ventral views of newborn palates. Arrows in Cpoint to a clear cleft in a IIIc�/�; Fgf10�/� sam-ple. D: Trans-illuminated lateral views of dis-sected lower gastrointestinal tract and bladdershowing a patent rectum in a IIIc�/� sample(parallel to the dashed line), but a blind colon ina IIIc�/�; Fgf10�/� mouse (arrow). Scale bars �2 mm in A–C; 1 mm in D.


Fig. 4. Impact of Fgf10 knock-down on several branching organs in FgfR2-IIIc�/� mice. Comparison of lungs (A–K), lacrimal glands (L,M), and kidneys(N,O) of IIIc�/�; Fgf10�/� mice versus WT or IIIc�/�. A–I: Lobe separations and alveolar branching pattern in IIIc�/�; Fgf10�/� lungs (G–I) resembles WT(A–C) but differs from IIIc�/� (D–F). Dashed lines highlight the absence of the accessory lobe. J,K: Foxa2 expression (arrowed in WT) is absent fromthe rudimentary accessory lobes of E11.5 IIIc�/� and IIIc�/�; Fgf10�/� (dashed lines), where specks of neutral red stain (arrows in K) indicative of celldeath is evident. B,B�,E,E�,H,H: Clustered expression of FgfR2-IIIb (black arrows) associated with epithelial cells at the tips of developing alveolae isevident in E17.5 WT and IIIc�/�; Fgf10�/� lungs, but much reduced in IIIc�/�. Diffuse mesenchymal expression of FgfR2-IIIb is evident in IIIc�/� andIIIc�/�; Fgf10�/� lungs (E�,H�). C,F,I: Mesenchymal expression of Fgf10 is reduced in IIIc�/�; Fgf10�/� (I) when compared to WT or IIIc�/�. L,M: IIIc�/�;Fgf10�/� lacrimal gland contains a rudimentary/degenerating mesenchymal sac (dashed border) that lacks carmine-stained epithelial branching. N,O:H&E-stained sections of newborn kidneys show fewer “S and comma-shaped” bodies (white arrows) as well as sparse and dilated epithelial tubules(black arrows) in IIIc�/�; Fgf10�/� when compared to WT. Cr, cranial; Mi, middle; Ca, caudal; Ac, accessory lobes. Scale bars � 2 mm (A,D,G); 100�m (B,B�,E,E�,H,H); 250 �m (C,F,I); 300 �m (J–M); and 200 �m (N,O).


Marie, 2002; Wilkie, 2005). Severalworkers have used other models of ASand Crouzon syndrome to elegantlydemonstrate that chemical inhibitorsof FGFR signaling pathway, uncou-pling of mutant receptors from FRS2or short hairpin-RNA-mediated tar-geting of mutant receptors can be usedto restore normal levels of FGFR sig-naling and rescue the skeletal defectsassociated with these syndromes (Es-warakumar et al., 2006; Shukla et al.,2007). Knockdown of ligands thatcause a gain-of-FGFR2 function in AScould be another approach. Here wehave used a genetic approach to pro-vide strong evidence that Fgf10 is akey contributing ligand not only tothe skeletal but also to some of thevisceral defects associated with AS. Indue course, we discovered an antago-nism between ERK (P44/p42) andp38 phosphorylation downstream ofFGFR2 signaling in the lungs, and en-countered phenotypes that are remi-niscent of loss-of-FgfR2-IIIb signaling.

Validity of IIIc�/� Mice forDissecting Apert-Like FgfR2Signaling Dynamics

Ectopic FGFR2-IIIb expression inmesenchymal cells is observed in rarecases of Apert syndrome (Oldridge et

al., 1999; Bochukova et al. 2008) andIIIc�/� mice, but functionally, thesemutations recapitulate the effects ofmore common Apert mutations, be-cause the action of FGFR2 S252W andP253R mutant receptors may also in-volve an interaction with IIIb- as wellas IIIc-activating FGFs (Anderson etal., 1998; Yu et al., 2000; Ibrahimi etal., 2001, 2004). Moreover, the pheno-type of FgfR2�/S252W mutant mice isstrikingly reminiscent of IIIc�/� mice(Wang et al., 2005). Therefore, ourfindings likely pertain to a majority ofApert cases.

Rescue of Skeletal Defects

We observe patent coronal and facialsutures, and unfused sternum in allIIIc�/�; Fgf10�/� (newborn) mice andapproximately half of E18.5 IIIc�/�;Fgf10�/�. Previous work had sug-gested that the development andpathogenesis of calvarial, facial andsternal bones is regulated by a mor-phogen-mode of FgfR signaling (Isekiet al., 1997, 1999; Carlton et al., 1998;Hajihosseini et al., 2004) with thelevel and bioavailability of Fgf ligandsplaying a critical role in the process.The present data reinforce this notion.

Several observations argue against

the possibility that our skeletal rescuerepresents a delay in calvarial, facial,and sternal growth, owing to loss orreduction of Fgf10 gene per se. Coro-nal and zygomatic arch sutures ofnewborn Fgf10�/� and IIIc�/�;Fgf10�/� mice are marginally widerthan WT, but this could be secondaryto a general growth retardation ormalformations at the skull base inthese mice; we note that invariablyFgf10�/� and IIIc�/�; Fgf10�/� miceare significantly smaller than WT lit-ter mates, show a narrower and down-wardly-slanted snout (Fig. 2F), andcarry a cleft palate defect (Min et al.,1998). These same mice certainly donot show the degree of calvarialgrowth retardation observed in Fgf18null mice (Liu et al., 2002). Moreover,at E18.5, the stage at which we haveconducted our detailed comparison ofsuture patency, the calvaria of IIIc�/�;Fgf10�/�, Fgf10�/�, and WT are indis-tinguishable (Fig 2, and not shown).Hence, Fgf10 does not appear to berequired for normal embryonic calvar-ial growth.

We anticipate that the gain-of-FgfR2 function in IIIc�/� tissues iscaused by the combined action ofFgfR2-IIIb- and FgfR2-IIIc-activatingligands, a host of which are expressedin the developing sutures (Hajihos-seini and Heath, 2002). Here we haveidentified a IIIb-activating compo-nent. The IIIc-activating Fgf/s thatcause craniosynostosis, for example,could be identified through a similarknockdown strategy and putative can-didates include Fgfs -1, -2, - 9, -18, and-20 (Hajihosseini and Heath, 2002).However, an investigation of Fgf18’sinvolvement may require a condition-al/stage-specific gene targeting ap-proach because Fgf18, unlike Fgf10, isrequired for embryonic calvarialgrowth (Liu et al., 2002).

Variation in the Range andLevel of Rescue inBranching Organs

Reducing the Fgf10 gene dosage issufficient to rescue the IIIc�/� lungbranching defects. However, this ro-bust rescue is not observed in LGs orSMGs (Jaskoll et al., 2005) and therescued lungs of IIIc�/�; Fgf10�/�

mice still lack an accessory lobe. How

Fig. 5. Evaluation of changes in FgfR2 and Fgf10 expression, and FgfR2 signalling dynamics. A:Semi-quantitative RT-PCR analyses of whole E14.5 brain. Upregulation of FgfR2-IIIb is observed inIIIc�/� (middle lane) (Hajihosseini et al., 2001) and maintained in IIIc�/�; Fgf10�/� brain. Faint bandin WT is contamination from meningeal tissue. Fgf10 expression is marginally reduced in IIIc�/�;Fgf10�/� tissue. B: Immunoblots of E18.5 lungs using antibodies against phosphorylated andnon-phosphorylated ERK (p44/42) and p38 reveals an increase in phospho-44/42 levels and adistinct absence of phospho-p38 in IIIc�/� tissue (far right) when compared to WT (far left). Inrescued lungs (middle lanes), phospho-p44/p42 level is reduced and p38 is phosphorylated.


can these variations and differentiallevels of rescue be explained?

At its simplest, a “full” rescue ofthese particular defects in IIIc�/� micemay require complete abrogation ofFgf10, and by implication, completeblocking of the aberrant FgfR2-IIIbactivity. It may also be possible that inIIIc�/�, Fgf10�/� mice, the level ofFgf10 downregulation varies acrossorgans, or that other FgfR2-IIIb acti-vating ligands (i.e., Fgfs 1, 3, 7, and22) compensate for Fgf10. A stage-spe-cific study of these ligands and theirinactivation in IIIc�/� would be re-quired to address these possibilities.

However, a more plausible explana-tion may lie in differences in the uti-lization or functionality of Fgf10 indifferent branching organs. For exam-ple, treatment of mesenchyme-free ep-ithelial explants of lung and LGs withFGF10 induces branching morpho-genesis in the former (Mailleux et al.,2001) but only cell proliferation in thelatter (Makarenkova et al., 2000). More-over, a reduced level of Fgf10 expres-sion in Fgf10�/� mice and patients har-bouring hypomorphic FGF10 mutationsinduces aplasia of SMG and LG but notlungs (Entesarian et al., 2005; Jaskollet al., 2005; Milunsky et al., 2006).Hence, in mature LGs and SMGs, Fgf10may govern epithelial cell survival anddivision and this would be in addition toits earlier role in the induction/forma-tion of these glands (Makarenkova etal., 2000; Entesarian et al., 2005).

The accessory lobe-specific defectin lungs may reflect the requirementfor differential levels of Fgf10 signal-ing in different parts of the sameorgan. Supporting this notion is thedetection of a higher and more per-sistent level of Fgf10 expression inthe accessory lobe when compared tocranial, middle, or caudal lobes atE11.5–E12.0 (Bellusci et al., 1997).Strong parallels have been drawnbetween signaling interactions thatgovern lung and limb developmentand it has previously been reportedthat development of digit I (toes) isregulated by a level of FgfR signal-ing that is distinct from that of digitsII–V (Partanen et al., 1998; Lewan-doski et al., 2000; Hajihosseini et al.,2004). By analogy, the accessory lobemay be regulated by a level of Fgf10signaling that is distinct from otherlobes.

AntagonisticPhosphorylation ofComponents of the MapKinase Pathway

While comparing the ERK signalingdynamics across different genotypes,we noted that levels of p38 phosphor-ylation is highest in WT and rescued(IIIc�/�; Fgf10�/�) lungs, where phos-phorylated p44/p42 is low. By con-trast, higher levels of p44/p42 phos-phorylation in IIIc�/� lungs wereaccompanied by a dramatic reductionin the levels of p38 phosphorylation.Although the significance of this an-tagonism and its occurrence in othertissues remains to be determined, astudy of chondrocyte differentiationsuggests that such an antagonismmay be critical to cell differentiation(Stanton et al., 2003). Nonetheless,our study is the first to describe suchan antagonistic relationship down-stream of FgfR2 signaling.

In summary, through genetic ma-nipulation of an AS mouse model wehave highlighted some of the key con-tributory factors that mediate thegain-of-FgfR2 activity and providedfurther insights into the complexity ofFgf10 function and FgfR signalingsystem.

EXPERIMENTALPROCEDURES

Generation and Genotypingof Mutant Animals

FgfR2-IIIc�/� and Fgf10�/� alleles re-sult in neo/peri-natal lethality (Min etal., 1998; Hajihosseini et al., 2001), andso the former allele is generatedthrough a LoxP-Cre mediated excisionof a FgfR2-IIIc “floxed” (flanked by LoxPsequences) allele, while the latter aremaintained as heterozygotes(Fgf10�/�). Female PGK-CRE (Lalle-mand et al., 1998) mice were used toderive the FgfR2-IIIc�/� allele in across with either homozygous floxed(FgfR2-IIIcfloxed/floxed) or heterozy-gous floxed (FgfR2-IIIcfloxed/�) males(Fig. 1A). Performing these crosseswith parents into which we had pre-viously introduced an Fgf10�/� al-lele, resulted in the generation of aseries of alleles, numbered as (i) to(v) in Figure 1A.

Animals were maintained on a C57/

Bl6J background and in compliancewith local regulations governing trans-genic breeding. Genotyping was carriedout by PCR using a previously de-scribed protocol and primer sets to de-tect the FgfR2-IIIcfloxed/� and FgfR2-IIIc�/� alleles (Hajihosseini et al.,2001), or sets of primers to specificallydetect the PGK-CRE transgene orFgf10 wild type and targeted alleles(P1, P2, P3; Fig. 1B and Suppl. Table 2).

Staining Procedures

Staining of skeletons with Alizarinred; vibratome sectioning of calvarialbones; whole-mount staining of thelacrimal glands with the dye Carmineand detection of apoptotic cells by neu-tral red staining were as previouslydescribed (Hajihosseini et al., 2001,2004; Hajihosseini and Heath, 2002).Six-micrometer-thick sections throughparaffin-embedded lungs and kidneyswere stained with H&E. In situ hybrid-ization reactions were carried out usingeither digoxygenin-labelled riboprobesin Whole Mount assays (WMISH), or byapplying 35S[UTP]-labelled riboprobesto 7 �m-thick tissue sections, accordingto previously described protocols (Haji-hosseini et al., 2001, 2004; Hajihosseiniand Heath, 2002; Rice et al., 2003). Ri-boprobes were as follows: Foxa2 (900bp), prepared from a plasmid providedby Dr. S. Bellusci; FgfR2-IIIb (161 bp),FgfR2-IIIc (139 bp), and Fgf10 (584 bp),prepared from partial cDNA sequencescloned in to Bluescript KSII� (Rice etal., 2003). Sense probes were used asnegative control. Digital images werecaptured using a Zeiss Stemi-2000Cand AxioPlan 2ie microscopes.

Semi-Quantitative ReverseTranscriptase-PCR andQuantitative RNAseProtection Assays

Two-step RT-PCR reactions were con-ducted as previously described (Haji-hosseini et al., 2001), using the follow-ing: 1 �g of total RNA for each sample/genotype (isolated using the SigmaTri reagent); Amersham Pharmacia’sReady-to-Go beads, which containpre-formulated and pre-disposed ly-ophilized components; and 1.5 pmol ofeach of the relevant gene-specificprimer pairs (Hajihosseini et al., 2001;Suppl. Table 2). A total of 29 amplifi-


cation cycles was used and equalamounts of products for comparablegenes/genotypes was resolved on 1.5%agarose gels. RNAse protection assayswere performed as described byWerner et al. (1993), using 365-bpFgf10 RNA probes generated from acDNA template.

Cell Lysis, ImmunoblottingAssays, and DensitometricMeasurements

Conditions for cell lysis and immuno-blottings were as per recommenda-tions of Cell Signaling TechnologyInc., from where the pan and phospho-specific anti-p42, -p44, and -p38 anti-bodies were purchased. Cell lysateswere quantified using the 2-D QuantKit (Amersham Biosciences). HRP-coupled secondary antibodies (Dako)were used and detected by enhancedchemiluminescence (ECL, Amer-sham). The quantification of phos-phorylation levels was carried out inthe linear range using non-saturatedblots, scanning densitometry, and Im-ageQuant software. Values (densityunits) were expressed as observed vol-ume intensity after subtraction ofbackground.

Optical TomographicSectioning

Optical tomographic sections from 4%-paraformaldehyde-fixed lungs weregenerated according to a previously de-scribed protocol (Sharpe et al., 2002).

ACKNOWLEDGMENTSWe gratefully acknowledge the assis-tance of Gary Martin and Sam Hosk-ins for animal husbandry; ProfessorsDavid Ornitz and Peter Lonai forFgf10 heterozygous and PGK-Cre col-onies; Dr. Tina Jaskoll for criticalreading of the manuscript; and the fi-nancial support of CRUK, The RoyalSociety, and the John & PamelaSalter Charitable trust. We apologiseto colleagues whose work could not becited due to space limitations

REFERENCES

Anderson J, Burns HD, Enriquez-Harris P,Wilkie AO, Heath JK. 1998. Apert syn-drome mutations in fibroblast growthfactor receptor 2 exhibit increased affin-

ity for FGF ligand. Hum Mol Genet7:1475–1483.

Bellusci S, Grindley J, Emoto H, Itoh N,Hogan BL. 1997. Fibroblast growth fac-tor 10 (FGF10) and branching morpho-genesis in the embryonic mouse lung.Development 124:4867–4878.

Bochukova E, Roscioli T, Hedges DJ, Tay-lor IB, Johnson D, David DJ, DeiningerPL, Wilkie AOM. 2008 Rare Mutations ofFGFR2 causing Apert syndrome: identi-fication of the first partial gene deletion,and an Alu element insertion from a newfamily. Human Mutation (in press).

Carlton MB, Colledge WH, Evans MJ.1998. Crouzon-like craniofacial dysmor-phology in the mouse is caused by aninsertional mutation at the Fgf3/Fgf4 lo-cus. Dev Dyn 212:242–249.

Cohen MM Jr, Kreiborg S. 1993a. An up-dated pediatric perspective on the Apertsyndrome. Am J Dis Child 147:989–993.

Cohen MM Jr, Kreiborg S. 1993b. Visceralanomalies in the Apert syndrome. Am JMed Genet 45:758–760.

De Langhe SP, Carraro G, Warburton D,Hajihosseini MK, Bellusci S. 2006. Lev-els of mesenchymal FGFR2 signalingmodulate smooth muscle progenitor cellcommitment in the lung. Dev Biol 299:52–62.

De Moerlooze L, Spencer-Dene B, Revest J,Hajihosseini M, Rosewell I, Dickson C.2000. An important role for the IIIb iso-form of fibroblast growth factor receptor2 (FGFR2) in mesenchymal-epithelialsignalling during mouse organogenesis.Development 127:483–492.

Entesarian M, Matsson H, Klar J, Bergen-dal B, Olson L, Arakaki R, Hayashi Y,Ohuchi H, Falahat B, Bolstad AI, Jon-sson R, Wahren-Herlenius M, Dahl N.2005. Mutations in the gene encodingfibroblast growth factor 10 are associ-ated with aplasia of lacrimal and sali-vary glands. Nat Genet 37:125–127.

Eswarakumar VP, Monsonego-Ornan E,Pines M, Antonopoulou I, Morriss-KayGM, Lonai P. 2002. The IIIc alternativeof Fgfr2 is a positive regulator of boneformation. Development 129:3783–3793.

Eswarakumar VP, Ozcan F, Lew ED, BaeJH, Tome F, Booth CJ, Adams DJ, Lax I,Schlessinger J. 2006. Attenuation of sig-naling pathways stimulated by patholog-ically activated FGF-receptor 2 mutantsprevents craniosynostosis. Proc NatlAcad Sci USA 103:18603–18608.

Hajihosseini MK. 2008. Fibroblast growthfactor signaling in cranial suture devel-opment and pathogenesis. Front OralBiol 12:160–177.

Hajihosseini MK, Heath JK. 2002. Expres-sion patterns of fibroblast growth fac-tors-18 and -20 in mouse embryos is sug-gestive of novel roles in calvarial andlimb development. Mech Dev 113:79–83.

Hajihosseini MK, Wilson S, De MoerloozeL, Dickson C. 2001. A splicing switch andgain-of-function mutation in FgfR2-IIIchemizygotes causes Apert/Pfeiffer-syn-drome-like phenotypes. Proc Natl AcadSci USA 98:3855–3860.

Hajihosseini MK, Lalioti MD, Arthaud S,Burgar HR, Brown JM, Twigg SR,Wilkie AO, Heath JK. 2004. Skeletal de-velopment is regulated by fibroblastgrowth factor receptor 1 signalling dy-namics. Development 131:325–335.

Hasegawa H, Ashigaki S, Takamatsu M,Suzuki-Migishima R, Ohbayashi N, ItohN, Takada S, Tanabe Y. 2004. Laminarpatterning in the developing neocortexby temporally coordinated fibroblastgrowth factor signaling. J Neurosci 24:8711–8719.

Ibrahimi OA, Eliseenkova AV, PlotnikovAN, Yu K, Ornitz DM, Mohammadi M.2001. Structural basis for fibroblastgrowth factor receptor 2 activation inApert syndrome. Proc Natl Acad Sci USA98:7182–7187.

Ibrahimi OA, Zhang F, Eliseenkova AV,Itoh N, Linhardt RJ, Mohammadi M.2004. Biochemical analysis of pathogenicligand-dependent FGFR2 mutationssuggests distinct pathophysiologicalmechanisms for craniofacial and limbabnormalities. Hum Mol Genet 13:2313–2324.

Iseki S, Wilkie AO, Heath JK, Ishimaru T,Eto K, Morriss-Kay GM. 1997. Fgfr2 andosteopontin domains in the developingskull vault are mutually exclusive andcan be altered by locally applied FGF2.Development 124:3375–3384.

Iseki S, Wilkie AO, Morriss-Kay GM. 1999.Fgfr1 and Fgfr2 have distinct differenti-ation- and proliferation-related roles inthe developing mouse skull vault. Devel-opment 126:5611–5620.

Itoh N, Ornitz DM. 2008. Functional evo-lutionary history of the mouse Fgf genefamily. Dev Dyn 237:18–27.

Jaskoll T, Zhou YM, Chai Y, MakarenkovaHP, Collinson JM, West JD, HajihosseiniMK, Lee J, Melnick M. 2002. Embryonicsubmandibular gland morphogenesis:stage-specific protein localization ofFGFs, BMPs, Pax6 and Pax9 in normalmice and abnormal SMG phenotypes inFgfR2-IIIc(�/Delta), BMP7(�/�) andPax6(�/�) mice. Cells Tissues Organs170:83–98.

Jaskoll T, Abichaker G, Witcher D, SalaFG, Bellusci S, Hajihosseini MK,Melnick M. 2005. FGF10/FGFR2b sig-naling plays essential roles during invivo embryonic submandibular salivarygland morphogenesis. BMC Dev Biol5:11.

Lallemand Y, Luria V, Haffner-Krausz R,Lonai P. 1998. Maternally expressedPGK-Cre transgene as a tool for earlyand uniform activation of the Cre site-specific recombinase. Transgenic Res7:105–112.

Lewandoski M, Sun X, Martin GR. 2000.Fgf8 signalling from the AER is essentialfor normal limb development. Nat Genet26:460–463.

Liu Z, Xu J, Colvin JS, Ornitz DM. 2002.Coordination of chondrogenesis and os-teogenesis by fibroblast growth factor 18.Genes Dev 16:859–869.

Mailleux AA, Tefft D, Ndiaye D, Itoh N,Thiery JP, Warburton D, Bellusci S.


2001. Evidence that SPROUTY2 func-tions as an inhibitor of mouse embryoniclung growth and morphogenesis. MechDev 102:81–94.

Makarenkova HP, Ito M, Govindarajan V,Faber SC, Sun L, McMahon G, OverbeekPA, Lang RA. 2000. FGF10 is an inducerand Pax6 a competence factor for lacri-mal gland development. Development127:2563–2572.

McKeehan WL, Wang F, Kan M. 1998. Theheparan sulfate-fibroblast growth factorfamily: diversity of structure and func-tion. Prog Nucleic Acid Res Mol Biol 59:135–176.

Milunsky JM, Zhao G, Maher TA, Colby R,Everman DB. 2006. LADD syndrome iscaused by FGF10 mutations. Clin Genet69:349–354.

Min H, Danilenko DM, Scully SA, Bolon B,Ring BD, Tarpley JE, DeRose M, Simo-net WS. 1998. Fgf-10 is required for bothlimb and lung development and exhibitsstriking functional similarity to Dro-sophila branchless. Genes Dev 12:3156–3161.

Motoyama J, Liu J, Mo R, Ding Q, Post M,Hui CC. 1998. Essential function of Gli2and Gli3 in the formation of lung, tra-chea and oesophagus. Nat Genet 20:54–57.

Ohuchi H, Hori Y, Yamasaki M, Harada H,Sekine K, Kato S, Itoh N. 2000. FGF10acts as a major ligand for FGF receptor 2IIIb in mouse multi-organ development.Biochem Biophys Res Commun 277:643–649.

Oldridge M, Zackai EH, McDonald-McGinn DM, Iseki S, Morriss-Kay GM,Twigg SR, Johnson D, Wall SA, Jiang W,Theda C, Jabs EW, Wilkie AO. 1999. Denovo alu-element insertions in FGFR2identify a distinct pathological basis forApert syndrome. Am J Hum Genet 64:446–461.

Ornitz DM, Marie PJ. 2002. FGF signalingpathways in endochondral and in-

tramembranous bone development andhuman genetic disease. Genes Dev 16:1446–1465.

Orr-Urtreger A, Bedford MT, Burakova T,Arman E, Zimmer Y, Yayon A, Givol D,Lonai P. 1993. Developmental localiza-tion of the splicing alternatives of fibro-blast growth factor receptor-2 (FGFR2).Dev Biol 158:475–486.

Partanen J, Schwartz L, Rossant J. 1998.Opposite phenotypes of hypomorphicand Y766 phosphorylation site muta-tions reveal a function for Fgfr1 in an-teroposterior patterning of mouse em-bryos. Genes Dev 12:2332v2344.

Passos-Bueno MR, Serti Eacute AE, Je-hee FS, Fanganiello R, Yeh E. 2008.Genetics of craniosynostosis: genes,syndromes, mutations and genotype-phenotype correlations. Front Oral Biol12:107–143.

Peters KG, Werner S, Chen G, WilliamsLT. 1992. Two FGF receptor genes aredifferentially expressed in epithelial andmesenchymal tissues during limb forma-tion and organogenesis in the mouse. De-velopment 114:233–243.

Plotnikov AN, Schlessinger J, HubbardSR, Mohammadi M. 1999. Structural ba-sis for FGF receptor dimerization andactivation. Cell 98:641–650.

Rice DP, Rice R, Thesleff I. 2003. FgfrmRNA isoforms in craniofacial bone de-velopment. Bone 33:14–27.

Sharpe J, Ahlgren U, Perry P, Hill B, RossA, Hecksher-Sorensen J, Baldock R, Da-vidson D. 2002. Optical projection tomog-raphy as a tool for 3D microscopy andgene expression studies. Science296:541–545.

Shukla V, Coumoul X, Wang RH, Kim HS,Deng CX. 2007. RNA interference andinhibition of MEK-ERK signaling pre-vent abnormal skeletal phenotypes in amouse model of craniosynostosis. NatGenet 39:1145–1150.

Slaney SF, Oldridge M, Hurst JA, Moriss-Kay GM, Hall CM, Poole MD, Wilkie AO.1996. Differential effects of FGFR2 mu-tations on syndactyly and cleft palate inApert syndrome. Am J Hum Genet 58:923–932.

Stanton LA, Underhill TM, Beier F. 2003.MAP kinases in chondrocyte differentia-tion. Dev Biol 263:165–175.

Tsang M, Dawid IB. 2004. Promotion andattenuation of FGF signaling throughthe Ras-MAPK pathway. Sci STKE 2004:pe17.

Wang Y, Xiao R, Yang F, Karim BO, Iacov-elli AJ, Cai J, Lerner CP, RichtsmeierJT, Leszl JM, Hill CA, Yu K, Ornitz DM,Elisseeff J, Huso DL, Jabs EW. 2005.Abnormalities in cartilage and bone de-velopment in the Apert syndromeFGFR2�/S252W mouse. Development132:3537–3548.

Werner S, Weinberg W, Liao X, Peters KG,Blessing M, Yuspa SH, Weiner RL, Wil-liams LT. 1993. Targeted expression of adominant-negative FGF receptor mutantin the epidermis of transgenic mice re-veals a role of FGF in keratinocyte orga-nization and differentiation. Embo J 12:2635–2643.

Wilkie AO. 2005. Bad bones, absent smell,selfish testes: the pleiotropic conse-quences of human FGF receptor muta-tions. Cytokine Growth Factor Rev 16:187–203.

Yu K, Herr AB, Waksman G, Ornitz DM.2000. Loss of fibroblast growth factor re-ceptor 2 ligand-binding specificity in Ap-ert syndrome. Proc Natl Acad Sci USA97:14536–14541.

Zhang X, Ibrahimi OA, Olsen SK,Umemori H, Mohammadi M, Ornitz DM.2006. Receptor specificity of the fibro-blast growth factor family. The completemammalian FGF family. J Biol Chem281:15694–15700.


evidence that fgf10 contributes to the skeletal and...

Documents