Analysis of a gain-of-function FGFR2 Crouzonmutation provides evidence of loss of functionactivity in the etiology of cleft palateAlison K. Snyder-Warwicka,b, Chad A. Perlyna,b,c, Jing Pand, Kai Yub, Lijuan Zhangd, and David M. Ornitzb,1

aDivision of Plastic Surgery and bDepartments of Developmental Biology and dPathology and Immunology, Washington University School of Medicine, St.Louis, MO 63110; and cDivision of Plastic Surgery, Miami Children’s Hospital, Miami, FL 33155

Communicated by Joseph Schlessinger, Yale University School of Medicine, New Haven, CT, December 29, 2009 (received for review August 18, 2009)

Cleft palate is a common birth defect in humans and is a commonphenotype associated with syndromic mutations in fibroblastgrowth factor receptor 2 (Fgfr2). Cleft palate occurred in nearly allmice homozygous for the Crouzon syndromemutation C342Y in themesenchymal splice formof Fgfr2.Mutantembryos showeddelayedpalate elevation, stage-specific biphasic changes in palate mesen-chymal proliferation, and reduced levels ofmesenchymal glycosami-noglycans (GAGs). Reduced levels of feedback regulators of FGFsignaling suggest that this gain-of-function mutation in FGFR2 ulti-mately resembles lossofFGFfunction inpalatemesenchyme.Knowl-edge of how mesenchymal FGF signaling regulates palatal shelfdevelopment may ultimately lead to pharmacological approachesto reduce cleft palate incidence in genetically predisposed humans.

Crouzon syndrome | fibroblast growth factor receptor 2 | cellproliferation | cell surface receptor | glycosaminoglycan

Craniosynostosis syndromes, as well as syndromic and non-syndromic cleft palate, have been associated with fibroblast

growth factor receptor (FGFR) mutations. Of the four highlyconserved FGFRs, FGFR2 is the most commonly mutated FGFreceptor (1–4). The FGF-FGFR family is involved in multipleintracellular signaling mechanisms in embryonic development,cell growth, wound healing, and tumorigenesis. The FGFRs arereceptor tyrosine kinases that signal via a ternary complex ofFGFR, FGF, and glycosaminoglycans (GAGs) (5). The ternarycomplex formation leads to receptor dimerization, autophos-phorylation, and activation of downstream signaling cascades (6).FGF-FGFR signaling is active during palatogenesis, and

genetic FGF mutations may contribute to 5% of cases of non-syndromic cleft lip and palate (7). In children with isolated cleftpalate, mutations in Fgfr1, Fgfr2, Fgfr3, and Fgf8 have beenidentified. Additionally, single nucleotide polymorphisms inchildren with nonsyndromic cleft lip and palate are associatedwith Fgf3, Fgf7, Fgf10, Fgf18, and Fgfr1, providing confirmation ofthe importance of the FGF-FGFR system in palate development(7). Knockout mouse models for Fgf10 and Fgfr2b develop cleftpalate (8, 9), establishing the necessity of epithelial FGF sig-naling in normal palatogenesis. Cleft palate in Fgf18−/− mice (10)suggests that mesenchymal FGF signaling may also be importantfor palate development.Cleft palate has been associated with both gain-of-function and

loss-of-function FGFRmutations (7, 11–15). Why gain-of-functionor loss-of-function mutations result in the same palatal phenotyperemains unclear, and understanding this phenomenonmay provideadditional insight into the pathogenesis of cleft palate. To addressthis question, we have focused on Crouzon syndrome (CS), a dis-order resulting from a missense Fgfr2 mutation that displays anincreased incidence of cleft palate in humans (14).CS occurs with an incidence of 12.5 permillion births and results

from genetic gain-of-function mutations in Fgfr2 (1). Cranio-facial anomalies are its hallmark feature, including cranio-synostosis, midface hypoplasia, proptosis, and oral anomalies suchas an increased incidence of cleft palate, class III malocclusion,

and a constricted dental arch.Anarrow, high-arched palate ismostcommonly seen.Over 30 different FGFR2mutationsmay result inCS, most of which localize to the FGFR2c isoform (16, 17), andtherefore are mesenchymally expressed. All patients with CS areheterozygous for the mutation; homozygous mutations are pre-sumed to be lethal.A murine model of CS has been constructed using the most

commonly observed mutation in human patients: a missensemutation of cysteine 342 (TGC→TAC) in exon 9 of the Fgfr2gene (Fgfr2C342Y) (18). This missense mutation results inunpaired cysteine residues, which cause FGFR2 dimerization viaformation of intermolecular disulfide bonds, resulting in ligand-independent activation. The spectrum of palatal phenotypes inthe mouse model closely parallels that of human patients with CS(19). In addition, the murine model allows evaluation ofhomozygous (Fgfr2C342Y/C342Y) embryos, which have a high inci-dence of cleft palate, and thus provides an excellent model forthe investigation of palate development.Murine palatogenesis closely resembles that of humans, con-

sisting of well-delineated stages of palatal shelf outgrowth, ele-vation, and fusion between embryonic days (E) 13.5–15.5.Abnormalities in any of these events may result in cleft palate. Inthis study, we show that the C342Y substitution in Fgfr2 results inaltered signaling, abnormal cellular proliferation, and alteredproduction of the components of the extracellular matrix. Wepropose that these defects reduce palatal shelf elongation anddelay palatal shelf elevation, the combination of which results incleft palate.

ResultsFgfr2 Is Expressed in the Posterior Palate of Wild-Type Mice. Duringmammalian development, palatal shelves extend along the lat-eral walls of the developing oropharynx and initially grow verti-cally on either side of the tongue. In the mouse (and rat), palatalshelves elevate at approximately E14.5 (20, 21). To determinewhere altered activity of the C342Y Fgfr2 mutation may regulatepalate development, we examined Fgfr2 expression before shelfelevation. At E13.5 (Fig. S1) and pre-elevation E14.5 (Fig. 1),Fgfr2 expression was strongest in the posterior palatal shelfmesenchyme with strong focal expression on the superior nasalhalf of the shelf (Fig. 1B′). Little expression was observed in theanterior, middle, and soft palate regions (Fig. 1 D′, C′, and A′,respectively). We therefore focused on the posterior palate incomparisons of CS and WT mice.

Author contributions: A.K.S.-W., C.A.P., L.Z., and D.M.O. designed research; A.K.S.-W.,C.A.P., J.P., and K.Y. performed research; A.K.S.-W., C.A.P., K.Y., and L.Z. contributednew reagents/analytic tools; A.K.S.-W., C.A.P., K.Y., L.Z., and D.M.O. analyzed data; andA.K.S.-W., C.A.P., K.Y., L.Z. and D.M.O. wrote the paper.

The authors declare no conflict of interest.1To whom correspondence should be addressed. E-mail: dornitz@wustl.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0913985107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0913985107 PNAS | February 9, 2010 | vol. 107 | no. 6 | 2515–2520

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Apr

il 3,

202

0

Palatal Shelf Morphogenesis Is Altered in CS Embryos.Normal palatedevelopment consists of three well-delineated stages: vertical pala-tal shelf outgrowth from the maxillary prominences beginning atE12.5 and continuing through E13.5; palatal shelf elevation, orreorientation, to a horizontal position above the tongue at E14.5;and dissolution of the medial epithelial seam and fusion of thepalatal shelves by E15.5. Fgfr2C342Y/+ and Fgfr2C342Y/C342Y miceexhibited increased incidence of cleft palate. A total of 115 of117 (98%) Fgfr2C342Y/C342Y embryos displayed a cleft at the com-pletion of palate development, whereas 9 of 234 (∼4%)Fgfr2C342Y/+

embryos examined displayed a palatal cleft. Qualitatively,Fgfr2C342Y/+ embryos displayed a larger spectrum of abnormalitiesin severity and timing compared with WT.At E13.5, the palate shelves of both Fgfr2C342Y/+ and

Fgfr2C342Y/C342Y embryos were similar to those of littermatecontrols (Fig. 2). E14.5 embryos, even from the same litter,displayed a variety of shelf morphologies that ranged from ver-tically oriented (Fig. 2B) to bilaterally elevated shelves. All E14.5Fgfr2C342Y/C342Y embryos examined, however, displayed verticallyoriented shelves (23 of 23). In contrast, 30 of 61 Fgfr2C342Y/+

embryos displayed vertically oriented shelves at E14.5, whereasonly 3 of 14 WT embryos had vertically oriented shelves at this

developmental stage. When compared with the vertically ori-ented shelves of Fgfr2C342Y/+ and WT embryos at E14.5,Fgfr2C342Y/C342Y shelves were relatively small, with an 11% (P <0.03) reduction in height. At E15.5, Fgfr2C342Y/C342Y shelves wereelevated and contained a large gap that eventually manifests asthe cleft palate phenotype.

Palatal Shelf Outgrowth: Altered Proliferation, but Not Cell Death, inthe Fgfr2C342Y/C342Y and Fgfr2C342Y/+ Palate. To assess both thegrowth of the palatal shelf and how differential growth couldregulate morphogenesis, we divided the palatal shelf into fourregions (SI Materials and Methods and Fig. 3A). Proliferation wasquantified by counting BrdU-stained and total nuclei in eachregion. Compared with WT, the posterior palatal shelves fromFgfr2C342Y/C342Y mice demonstrated significantly increased mes-enchymal cellular proliferation at E13.5 in the oral half of theshelf (regions III and IV; Fig. 3 B–D). By E14.5, however, pro-liferation was significantly decreased throughout shelf mesen-chyme (Fig. 3E). There were no differences in mesenchymalproliferation rates among genotypes at E15.5.To determine whether rates of cell death were affected by the

CS mutation, TUNEL assays were carried out on E13.5 andE14.5 tissues. Low levels of apoptosis were noted in the mes-enchyme and epithelium of all three genotypes at both timepoints. No differences in rates of cell death were noted amonggenotypes at E13.5 or E14.5 (Fig. S2).

Delayed Palatal Shelf Elevation: Altered GAG Levels in the Fgfr2C342Y/C342Y

and Fgfr2C342Y/+ Palate. We assessed differences in GAG contentwithin the palate, because GAG accumulation and hydration havebeen hypothesized to regulate elevation. Palate GAG quantifica-tion by high-performance liquid chromatography (HPLC) (22)showed an increase in total GAG content, standardized to glycine,in the whole palate of all genotypes just before shelf elevationcompared with pre- and postelevation time points (Fig. 4A).Fgfr2C342Y/C342Y embryos displayed decreased (P< 0.02) total GAGin the whole palate at E14.25 (Fig. 4A), and this finding was con-sistent within both the anterior (P < 0.05) and posterior (P < 0.02)palate before elevation (Fig. 4B). Consistent with the regionalGAG HPLC analysis, Alcian blue staining, used to qualitativelyassess total GAG levels in histologic sections, showed decreasedstaining in posterior palatal mesenchyme of Fgfr2C342Y/C342Y

embryos compared with WT and Fgfr2C342Y/+ littermates justbefore shelf elevation (early E14.5; Fig. 4 C–E).In addition to decreased GAG quantities, HPLC data also

showed altered GAG composition in the CS palates. Comparedwith WT, palates from Fgfr2C342Y/C342Y embryos contained sig-nificantly (P < 0.05) decreased amounts of galactosamine (GalN)in both the anterior and posterior regions of the palate at E14.25(Fig. 4F). Examination of the entire palatal shelf at E14.5

Soft palate Posterior hard palate Middle hard palate Anterior hard palate

pt

A

A’

B

B’

C

C’

D

D’

Fig. 1. Fgfr2 expression in the wild-type murine palate. (A–D) H&E stained coronal sections through the soft palate, posterior, middle and anterior regions ofthe pre-elevation E14.5 WT mouse palate. (A′–D′) Fgfr2 in situ hybridization (red) on adjacent sections. The greatest Fgfr2 expression is seen in the super-omedial portion of the posterior palate (arrow). p, palate; t, tongue. (Scale bar: 200 μM.)

D E F

G H I

E13.5 E14.5 E15.5

A B C

p t pt

p

t

no

n

o

WT

Fgf

r2C

3 42Y

/+F

gfr 2

C3 4

2Y/C

342Y

Fig. 2. Palate development in WT, Fgfr2C342Y/+, and Fgfr2C342Y/C342Y mice.(A, D, and G) E13.5, (B, E, and H) E14.5, and (C, F, and I) E15.5 H&E stainedcoronal sections of developing mouse palates. Normal palate developmentconsists of palate shelf outgrowth (A), elevation (B), and fusion (C), whereasdevelopment of the Fgfr2C342Y/C342Y palate (G–I) is notable for narrowershelves, delayed elevation, and a cleft palate. Fgfr2C342Y/+ palate develop-ment (D–F) includes a spectrum of phenotypes, such as delayed elevationand normal fusion, shown here. n, nasal cavity; o, oral cavity; p, palate; t,tongue. (Scale bar: 200 μM.)

2516 | www.pnas.org/cgi/doi/10.1073/pnas.0913985107 Snyder-Warwick et al.

Dow

nloa

ded

by g

uest

on

Apr

il 3,

202

0

showed an increased ratio of galactosamine to glucosamine(GlcN) in the CS palates compared with WT (P < 0.03; Fig. 4G).A statistically significant (P < 0.01 for all genotypes) decrease inthe ratio of galactosamine to glucosamine was apparent atE14.25 compared with E13.5, and to E14.5 for all genotypes aswell (Fig. 4G). There were no differences in GAG quantity orcontent among genotypes earlier in development (E13.5) usingeither Alcian blue staining or HPLC; all differences in GAGsoccurred near the time of palatal shelf elevation.To test if hyaluronic acid (HA) biosynthetic enzyme expres-

sion is altered, quantitative RT-PCR was used to assess hyalur-onic acid synthase 2 (Has2) expression. At E13.5, there were nodifferences in Has2 expression between Fgfr2C342Y/C342Y embryosand WT (Fig. 4H). However, at E14.25, just before normalpalate elevation, Has2 expression was significantly (P < 0.05)down-regulated in the posterior palate of Fgfr2C342Y/C342Y

embryos (Fig. 4H).

Palatal Shelf Fusion: Fgfr2C342Y/C342Y and Fgfr2C342Y/+ Palates Fuse. Toassess whether the CS palate was capable of fusion, dissectedpalatal shelves placed in direct contact with one another weregrown in culture for 96 h. All palates (5/5) from each of the threegenotypes (WT, Fgfr2C342Y/+, and Fgfr2C342Y/C342Y) were capableof fusion in vitro in two separate experiments (Fig. S3).

Manipulation of FGF Signaling Affects in Vitro WT Palatal ShelfDevelopment. To better understand the FGF-FGFR signalingtrend responsible for altered palate development in CS, weassessed the effects of FGF gain-of-function and loss-of-function

signaling on normal palate development in vitro (Fig. 5A). E12.5whole maxilla, including the early palatal shelves, were grown inorganotypic culture for 72 h in the presence of FGF2, FGF18, orthe FGFR tyrosine kinase inhibitor PD173074. Untreated andcontrol WT palates showed normal elevation and fusion in vitro(24 of 24). Augmentation of FGF-FGFR signaling with FGF2 (1μg/mL) resulted in cleft palate in 9 of 10 WT samples performedin four different experiments. Similarly, FGF18 (1 μg/mL) sup-plementation resulted in clefts in 6 of 6 WT palates over threedifferent experiments. Interestingly, inhibition of the FGF-FGFR signaling pathway with PD173074 in WT palates alsoresulted in a 100% incidence (12 of 12) of clefting (Fig. 5A).

Altered FGFR Expression and Signaling in the Fgfr2C342Y/C342Y Palate.FGFR signaling within the posterior palatal shelves was inves-tigated in WT and Fgfr2C342Y/C342Y embryos early in palatogenesis

0

0.1

0.2

0.3

0.4

I II III IV

Brd

UIn

co

rpo

ratio

n(%

)

Region

a

b

I

IIIII

IV

A

E

D

0

0.1

0.2

0.3

0.4

I II III IVRegion

Brd

UIn

co

rpo

ratio

n(%

)

a

b

c

d

B C

Fig. 3. Cellular proliferation in developing WT, Fgfr2C342Y/+, andFgfr2C342Y/C342Y palates. (A) Palatal shelf division into four regions toquantify BrdU staining. (B–D) At E13.5, Fgfr2C342Y/C342Y palates (C) haveincreased proliferation in regions III and IV compared with WT (B) palates (D;a, P < 0.004; b, P < 0.02, respectively; n = 4 or more for all genotypes). (E) AtE14.5, Fgfr2C342Y/C342Y show decreased BrdU incorporation compared to WTpalates throughout the palate mesenchyme (a, P < 0.02 region I; b, P < 0.002region II; c, P < 0.0007 region III; d, P < 0.02 region IV; n = 4 for all geno-types). Black bars, WT; gray bars, Fgfr2C342Y/+; white bars, Fgfr2C342Y/C342Y.

Relativeexpression

0

0.4

0.8

1.2

0.4

0

0.2

0.6

0.8

0

0.4

0.8

1.2

E13.5 E14.25 E14.5

TotalGAG

(pmol) a

bc

b

ca b

%GalN

Anterior Posterior 0

10

20

30

0

0.2

0.4

0.6

%GalN/%Gl cN

E13.5 E14.25 E14.5

da

Anterior Posterior

TotalGAG

(pmol)

EDC

F G

A Ba b

E13.5 E14.25

Has2H

a

Fig. 4. Glycosaminoglycan accumulation, composition, and synthesis indevelopingWT, Fgfr2C342Y/+, and Fgfr2C342Y/C342Ypalates. (A) GAGaccumulationduring palate development (a, P< 0.001; b, P< 0.02; c, P < 0.001). (B) Comparedwith WT, Fgfr2C342Y/C342Y palates have less total GAG accumulation in theanterior (a, P < 0.05) and posterior (b, P < 0.02) regions at E14.25. (C–E) WT (C)and Fgfr2C342Y/+ (D) palates show increased Alcian blue staining comparedwithFgfr2C342Y/C342Y (E) in the posterior region at E14.5 (representative of threeembryos for each genotype). (F) Decreased GalN composition is seen in bothregions of the Fgfr2C342Y/C342Y palate at E14.25 compared with WT (a, P < 0.04;b, P < 0.02). (G) GalN/GlcN ratio in whole palatal shelves. All genotypes have adecreased ratio of GalN/GlcN in the whole palate just before elevation (E14.25;a, P < 0.001), and the Fgfr2C342Y/C342Y GalN/GlcN ratio is further decreasedcompared with WT at E14.25 (b, P < 0.01). At E14.5, all genotypes show anincreasedGalN/GlcN ratio (c, P< 0.001), and the Fgfr2C342Y/C342YGalN/GlcN ratiois further increased compared with WT (d, P < 0.02). (H) Has2 expression wasdecreased in the posterior region of the Fgfr2C342Y/C342Y palate compared withWT at E14.25 (a, P < 0.05). Black bars, WT; gray bars, Fgfr2C342Y/+; white bars,Fgfr2C342Y/C342Y.

Snyder-Warwick et al. PNAS | February 9, 2010 | vol. 107 | no. 6 | 2517

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Apr

il 3,

202

0

(E13.5), just before shelf elevation (E14.25) and after elevation(E15.5). Expression of the Fgfr2c isoform showed a significant(P < 0.02) 2-fold increase in the Fgfr2C342Y/C342Y palates at E15.5but no differences in expression at earlier time points (Fig. 5B).No differences in expression of the FGF ligands Fgf9, Fgf10, orFgf18; the FGF pathway antagonist Dusp6; or the FGFR/MAPKadaptor Frs2 were noted throughout palate development.Examination of FGF-responsive transcription factors Pea3 andErm showed a trend (P= 0.06) toward increased Pea3 expressionin Fgfr2C342Y/C342Y palate at E13.5 and no differences in Ermexpression throughout palate development. Barx1 is an FGF-regulated homeobox transcription factor expressed in craniofacialstructures, including palate mesenchyme (23). Just before thetime of palatal shelf elevation, Barx1 expression was signifi-cantly increased by greater than 2.5-fold (P < 0.05; Fig. 5C) inFgfr2C342Y/C342Y palates. Sprouty (Spry) 2 and 4 are negative reg-ulators of FGF signaling. Spry2, which is expressed in palateepithelium, showed a trend toward decreased expressionthroughout palate development (significant at E15.5; P < 0.04;Fig. 5D) and Spry4, which is expressed in palate mesenchyme,showed significantly decreased expression at all three time points(P < 0.01 for E13.5 and E14.25, P < 0.03 for E15.5; Fig. 5E).

DiscussionDefects in any of the stages that define normal palatogenesis mayresult in cleft palate. Examination of each of these devel-opmental stages in CS mice showed abnormalities in palatal shelfoutgrowth and elevation, but no defects in fusion. Examinationof Fgfr2 expression localized the highest levels of expression tothe posterior portions of the WT palatal shelf. Because the

C342Y mutation in Fgfr2 is localized in the mesenchymal “c”splice form of Fgfr2, we focused our analysis on the posteriorpalate mesenchyme. Cellular proliferation in the posterior palatewas initially increased in the oral half of the palatal shelf in theCS mice in a genetic dose-dependent fashion, whereas pro-liferation was decreased in the CS palate throughout the mes-enchyme at E14.5. As signaling through FGFR2 is integral tocellular proliferation (24), these data suggest temporal alter-ations in levels of Fgfr2 signaling. Despite the genetic gain-of-function Fgfr2 mutation, the resulting effects on proliferation arenot directly coupled to receptor activation. Multiple downstreammediators of FGF signaling may be activated or inhibited andmay feedback to modulate overall activity. Importantly, pro-liferation is diminished in the CS palate just before shelf ele-vation. The overall consequence of expression of the mutantFgfr2 is decreased growth, which results in palatal shelves thatare both narrower and incapable of contacting one another aftershelf elevation is complete, resulting in a palatal cleft.Increased total GAG quantities were noted in the whole pal-

ates of all three genotypes just before the timing of normalpalatal shelf elevation. This finding supports the concept thatGAGs are integral to the elevation process. The initial hypoth-esis that hydration of GAGs produces expansion of the extra-cellular matrix (ECM), and therefore provides an intrinsic forcewithin the palatal shelf tissue that allows palatal shelf elevationto occur, was postulated by Lazzaro in 1940 (25) and others (26–35). During palatogenesis, hyaluronic acid and sulfated GAGsare present (26, 33). HA is one GAG component and makes up60–65% of the ECM in the murine palate around the time ofpalatal shelf reorientation (26, 31). Levels of HA are elevatedbefore shelf reorientation, and then decrease once elevation iscomplete (34, 35). Our data showing increased GAGs beforeshelf reorientation support those of Pratt et al. (26), who dem-onstrated HA synthesis in the 24 h before shelf closure. Exper-imental models have also shown suppressed GAG synthesis interatogen-induced cleft palate (35, 36). Our data and previousmodels together support the necessity of GAG accumulationbefore palatal shelf reorientation.The Fgfr2C342Y/C342Y mouse displayed a 24-h delay in palatal

shelf elevation. Interestingly, administration of chlorcyclizine, asubstance that enhances HA and chondroitin sulfate degradationwithout affecting GAG synthesis (27), to pregnant WT miceresulted in a 100% incidence of cleft palate in the offspring. Thepalates of these treated embryos demonstrated a 24-h delay inshelf size and shape in the palate (29). Similarly, delayed palatalshelf reorientation and decreased GAG synthesis were noted inthe hamster after cleft palate induction with cyclophosphamide(36). The delay in palatal shelf elevation seen in the CS mouse isconsistent with other experimental models that have showndecreased GAG accumulation.Fgfr2 is localized to the posterior palate in regions shown to

contain the greatest amounts of HA, a component integral toshelf elevation (28). Consistent with a role for mutant FGFR2affecting palate development through regulation of GAG con-tent, decreased GAG quantities were noted in the posterior CSpalates compared with WT. FGF2 has been shown to modulateexpression of GAGs and proteoglycans. In vitro treatment ofhuman periodontal ligament cells with FGF2 resulted inincreased heparin sulfate in the supernatant, and may alsostimulate HAS1 and HAS2 (37). Has2 is one of three genesencoding HA synthesis and is present during palatogenesis (38).Decreased Has2 expression in the posterior CS palate may havea role in decreased HA accumulation and, interestingly, may alsohave a role in cellular proliferation as intracellular HA has beensuggested as a regulator of proliferation (39). Taken together,decreased Has2 expression and decreased GAG quantities mayresult in a deficiency of GAG accumulation, which may result in

a a

a cba

00.511.522.53

1.6

1.2

0.8

0.4

0

E13.5 E14.25 E15.5

B

D

A

E13.5 E14.25 E15.5

Spry2

E13.5 E14.25 E15.5

Barx1C

E

E13.5 E14.25 E15.5

Spry4

1.2

0.8

0.4

0

00.511.522.533.5

Relativeexpression

Rela tiveexpression

Relativeexpression

Relati veexpression

Fgfr2c

control FGF2 DMSO PD173074

Fig. 5. FGF signaling and palate development: In vitro FGF signalingmanipulation in WT palate and qPCR studies in WT and Fgfr2C342Y/C342Y

posterior palates. (A) WT palate explant cultures treated with FGF2 (1 μg/mL)or PD173074 (2 μM) exhibited cleft palate (arrows) compared with the fusedpalates of untreated and DMSO-treated control palates. (B) Compared withWT, the posterior palates of Fgfr2C342Y/C342Y mice exhibit a 2-fold increase inrelative expression of Fgfr2c at E15.5 (a, P < 0.02; n = 4 or more), but nodifferences at earlier time points. (C) Barx1 expression in the Fgfr2C342Y/C342Y

posterior palate increases greater than 2.5-fold (a, P < 0.05) at E14.25 (n = 7or more). (D) Expression of Spry2 is decreased in Fgfr2C342Y/C342Y mousecompared with WT at E15.5 (a, P < 0.04, n = 4 or more). (E) Expression ofSpry4 is decreased throughout palate development (a, P < 0.01; b, P < 0.01; c,P < 0.03; n = 4 or more). Black bars, WT; white bars, Fgfr2C342Y/C342Y.

2518 | www.pnas.org/cgi/doi/10.1073/pnas.0913985107 Snyder-Warwick et al.

Dow

nloa

ded

by g

uest

on

Apr

il 3,

202

0

reduced hydrostatic forces necessary for palatal shelf reorien-tation, resulting in delayed or even absent shelf elevation.GAGs are divided into galactosamine (GalN)-containing GAGs

and glucosamine (GlcN)-containing GAGs. GalN-containingGAGs include chondroitin sulfate and dermatan sulfate. Chon-droitin sulfate proteoglycans interact with HA in assembling theECM.GlcN-containing GAGs, on the other hand, include heparansulfate, heparin, keratan sulfate, andHA. All genotypes displayed asignificant decrease in the ratio of percentage GalN to GlcN con-tent just before shelf elevation compared with pre-elevation andpost-elevation time points, which may indicate decreased GalN-containing GAG biosynthesis/degradation and increased GlcN-containing GAG biosynthesis before shelf elevation. CS mice dis-played significantly decreased percentages of GalN compared withWT in the posterior palate at E14.25, consistent with less overallGAG content. Examination of the palates at a slightly later timepoint showed increased amounts of galactosamine in the wholepalate at the time of normal palate shelf elevation. This samepattern of increased galactosamine was noted by Bosi et al. (40) infibroblasts from human patients with nonsyndromic cleft palate.This group also noted that treatment of normal fibroblasts with ateratogen that induces cleft palate, diphenylhydantoin, resulted inreduced GAG synthesis. In addition, similar patterns of decreasedGAG and increased GalN were noted in calvarial fibroblasts inhuman patients with CS (41, 42). Our data fit not only with those ofsimilar disease phenotypes, but also suggest possible similarities inthe etiopathology of CS in different organ systems. In the palate,these abnormalities in GAG quantity and composition likely con-tribute to the delay in palatal elevation noted in the CS mouse,which is one of the key components of the cleft palate phenotype.The delay in shelf reorientation causes a loss of a critical devel-opmental window in which shelf elevation allows for shelf fusion tooccur normally.The CS mouse displays abnormal cellular proliferation and

ECM quantities and composition in the palate. These abnor-malities affect two developmental stages, palatal shelf outgrowthand elevation, which may combine for a synergistic effect ofaltered palatal shelf volume. Adequate palatal shelf volume hasbeen hypothesized as a mandatory prerequisite to enable shelfreorientation, or elevation, to occur. Shelf volume may be gainedvia three mechanisms: cellular proliferation (43, 44), cellularhypertrophy, or ECM accumulation and hydration (32, 34). Inthe CS mouse, decreased cellular proliferation and decreasedECM accumulation near the time of elevation may contribute tothe delayed shelf reorientation phenomenon. The timing of theseabnormalities provides a mechanistic clue; both abnormalitiesoccur at or near the timing of palatal shelf elevation (there wereno alterations in GAGs noted at earlier or later time points). Thedelay in shelf elevation is likely the key component of the cleftpalate phenotype displayed in this model.Although palatal shelf outgrowth and elevation are disrupted

in the CS mouse model, the final phase of palatogenesis, palatalfusion, when isolated, remains intact. Palatal fusion involves thecontact and then dissolution of the medial edge epithelium(MEE) of the opposing palatal shelves. This process is mediatedby transforming growth factor beta (TGFβ) signaling (reviewedin ref. 45). Our data support the concept that the CS Fgfr2mutation does not affect the TGFβ-mediated fusion process.One of the most interesting and confounding findings of this

work is that both augmentation and inhibition of FGF-FGFRsignaling in vitro resulted in an identical phenotype, a cleft pal-ate. These findings, perhaps more than any other data, supportour hypothesis that the gain-of-function genetic mutation in CSmay indeed result in a loss-of-function phenotype. Interestingly,

in vivo evidence for this observation exists as well. Mice with adeletion encompassing Sprouty2 (Spry2), an antagonist of FGFsignaling, exhibit cleft palate (46). In addition, conditionalknockout of Fgfr1 and Fgfr2 in palate mesenchyme results in cleftpalate with 100% penetrance (Fig. S4). Clearly, the mechanisticexplanation for this syndrome is not simple.To better understand the signaling mechanisms responsible for

the cleft palate phenotype in CS, qRT-PCR studies were com-pleted on several genes of interest. FGF ligand (Fgf9, Fgf10, andFgf18) andFgfr2 expression showed no differences between theCSand WT genotypes throughout palate development, suggestingthat feedbackmechanisms resulting fromconstitutive activation ofFGFR2do not affect expression of FGF ligands known to functionin palate development or FGFR2 itself. It is possible that the gain-of-function mutation in FGFR2 results in an initial increase inFGF-FGFR2 signaling that eventually reaches a threshold, andthen becomes down-regulated. This could occur through complexinteractions of downstream signaling cascades and feedback loops.Barx1 expression was increased in the posterior palates of thehomozygous CS mice compared with WT at the time just beforepalatal shelf elevation, and both Spry2 and Spry4 expression weredecreased during the time of palate development. These findingsfit with the hypothesis of dynamic FGFR2 signaling levels. As FGFsignaling regulates Spry gene expression, decreased Spry expres-sion levels would indicate decreased FGFR2 signaling. Our find-ings of increased Barx1 expression in association with decreasedSpry2 fit with those of Welsh et al. (46) in their construction of amouse with a deletion encompassing Spry2. These mice exhibitedcleft palate, and, in the absence of Spry2, the palatal mesenchymedisplayed expanded and ectopic Barx1 expression. It is importantto note, however, that decreased Spry2 expression resulting froman Fgfr2mutation suggests a pathway in which Spry is downstreamof FGFR signaling, whereas a genetic deletion of Spry2 itself isobviously a separate inciting event and may not be directly linkedto FGFR signaling. The correlation of these studies, however,provides information about related signaling pathways and theidea that altered FGFR signaling results in cleft palate.These studies suggest that a genetic gain-of-function Fgfr2

mutation may result in a loss-of-function phenotype; however,the underlying mechanism may be complex. The gain-of-functionmutation does not simply translate to a gain-of-function phe-notype; negative regulation of FGFR signaling can occur fol-lowing hyperactivation of FGFR2 during palate development.We conclude that abnormal signaling resulting from the C342Ymutation in Fgfr2 is a critical regulator of cellular growth andproduction of components of the extracellular matrix (GAGquantities and composition) during palate development. Thesedefects foster abnormal palatal shelf outgrowth and elevation,resulting in cleft palate, and it is these developmental processesthat should be targeted for future prevention strategies andpharmacologic interventions.

Materials and MethodsFgfr2C342Y/+ mice were maintained on a CD1 genetic background and inter-crossed to generate litters withwild-type (WT), Fgfr2C342Y/+, and Fgfr2C342Y/C342Y

embryos. Analysis of cell proliferation, cell death, GAG content, qRT-PCR, andpalate cultures are described in detail in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank C. Smith and G. Schmid for technicalassistance, and G. Morriss-Kay and C. Babbs for advice. We thank Dr. S.Mackinnon for providing C.A.P. and A.S.W. the opportunity for research.This work was supported by the Plastic Surgery Educational Foundation(A.S.W.), National Institutes of Health Grant HD049808, and the VirginiaFriedhofer Charitable Trust.

1. Wilkie AO (2005) Bad bones, absent smell, selfish testes: The pleiotropic consequences

of human FGF receptor mutations. Cytokine Growth Factor Rev 16:187–203.

2. Hajihosseini MK (2008) Fibroblast growth factor signaling in cranial suture

development and pathogenesis. Front Oral Biol 12:160–177.

Snyder-Warwick et al. PNAS | February 9, 2010 | vol. 107 | no. 6 | 2519

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Apr

il 3,

202

0

3. Marie PJ, Coffin JD, Hurley MM (2005) FGF and FGFR signaling in chondrodysplasiasand craniosynostosis. J Cell Biochem 96:888–896.

4. Ornitz DM, Marie PJ (2002) FGF signaling pathways in endochondral andintramembranous bone development and human genetic disease. Genes Dev 16:1446–1465.

5. McDowell LM, et al. (2006) Inhibition or activation of Apert syndrome FGFR2 (S252W)signaling by specific glycosaminoglycans. J Biol Chem 281:6924–6930.

6. Eswarakumar VP, Lax I, Schlessinger J (2005) Cellular signaling by fibroblast growthfactor receptors. Cytokine Growth Factor Rev 16:139–149.

7. Riley BM, et al. (2007) Impaired FGF signaling contributes to cleft lip and palate. ProcNatl Acad Sci USA 104:4512–4517.

8. Rice R, et al. (2004) Disruption of Fgf10/Fgfr2b-coordinated epithelial-mesenchymalinteractions causes cleft palate. J Clin Invest 113:1692–1700.

9. Alappat SR, et al. (2005) The cellular and molecular etiology of the cleft secondarypalate in Fgf10 mutant mice. Dev Biol 277:102–113.

10. Liu Z, Xu J, Colvin JS, Ornitz DM (2002) Coordination of chondrogenesis andosteogenesis by fibroblast growth factor 18. Genes Dev 16:859–869.

11. Slaney SF, et al. (1996) Differential effects of FGFR2 mutations on syndactyly and cleftpalate in Apert syndrome. Am J Hum Genet 58:923–932.

12. Dodé C, et al. (2003) Loss-of-function mutations in FGFR1 cause autosomal dominantKallmann syndrome. Nat Genet 33:463–465.

13. Mølsted K, Kjaer I, Giwercman A, Vesterhauge S, Skakkebaek NE (1997) Craniofacialmorphology in patients with Kallmann’s syndrome with and without cleft lip andpalate. Cleft Palate Craniofac J 34:417–424.

14. Peterson SJ, Pruzansky S (1974) Palatal anomalies in the syndromes of Apert andCrouzon. Cleft Palate J 11:394–403.

15. Tompach PC, Zeitler DL (1995) Kallmann syndrome with associated cleft lip andpalate: Case report and review of the literature. J Oral Maxillofac Surg 53:85–87.

16. Kan SH, et al. (2002) Genomic screening of fibroblast growth-factor receptor 2 revealsa wide spectrum of mutations in patients with syndromic craniosynostosis. Am J HumGenet 70:472–486.

17. Cohen MMJ (2004) FGFs/FGFRs and associated disorders. Inborn Errors ofDevelopment: The Molecular Basis of Clinical Disorders on Morphogenesis, edsEpstein CJ, Erickson RP, Wynshaw-Boris A (Oxford Univ Press, New York), pp 380–400.

18. Eswarakumar VP, Horowitz MC, Locklin R, Morriss-Kay GM, Lonai P (2004) A gain-of-function mutation of Fgfr2c demonstrates the roles of this receptor variant inosteogenesis. Proc Natl Acad Sci USA 101:12555–12560.

19. Perlyn CA, et al. (2006) The craniofacial phenotype of the Crouzon mouse: Analysis ofa model for syndromic craniosynostosis using three-dimensional MicroCT. Cleft PalateCraniofac J 43:740–748.

20. Ferguson MW (1988) Palate development. Development 103 (Suppl):41–60.21. Ferguson MW (1978) Palatal shelf elevation in the Wistar rat fetus. J Anat 125:

555–577.22. Studelska DR, Giljum K, McDowell LM, Zhang L (2006) Quantification of

glycosaminoglycans by reversed-phase HPLC separation of fluorescent isoindolederivatives. Glycobiology 16:65–72.

23. Tissier-Seta JP, et al. (1995) Barx1, a new mouse homeodomain transcription factorexpressed in cranio-facial ectomesenchyme and the stomach. Mech Dev 51:3–15.

24. Iseki S, Wilkie AO, Morriss-Kay GM (1999) Fgfr1 and Fgfr2 have distinctdifferentiation- and proliferation-related roles in the developing mouse skull vault.Development 126:5611–5620.

25. Lazzaro C (1940) The mechanism of closure of the secondary palate (Translated fromItalian). Monitore zoologico italiano 51:249–273.

26. Pratt RM, Goggins JF, Wilk AL, King CT (1973) Acid mucopolysaccharide synthesis inthe secondary palate of the developing rat at the time of rotation and fusion. DevBiol 32:230–237.

27. Wilk AL, King CT, Pratt RM (1978) Chlorcyclizine induction of cleft palate in the rat:Degradation of palatal glycosaminoglycans. Teratology 18:199–209.

28. Brinkley LL, Morris-Wiman J (1987) Computer-assisted analysis of hyaluronatedistribution during morphogenesis of the mouse secondary palate. Development 100:629–635.

29. Brinkley LL, Morris-Wiman J (1987) Effects of chlorcyclizine-induced glycosaminoglycanalterations on patterns of hyaluronate distribution during morphogenesis of the mousesecondary palate. Development 100:637–640.

30. Brinkley LL, Vickerman MM (1982) The effects of chlorcyclizine-induced alterations ofglycosaminoglycans on mouse palatal shelf elevation in vivo and in vitro. J EmbryolExp Morphol 69:193–213.

31. Knudsen TB, Bulleit RF, Zimmerman EF (1985) Histochemical localization ofglycosaminoglycans during morphogenesis of the secondary palate in mice. AnatEmbryol (Berl) 173:137–142.

32. Morris-Wiman J, Brinkley L (1992) An extracellular matrix infrastructure providessupport for murine secondary palatal shelf remodelling. Anat Rec 234:575–586.

33. Young AV, Hehn BM, Cheng KM, Shah RM (1994) A comparative study on the effectsof 5-fluorouracil on glycosaminoglycan synthesis during palate development in quailand hamster. Histol Histopathol 9:515–523.

34. Singh GD, Moxham BJ, Langley MS, Waddington RJ, Embery G (1994) Changes in thecomposition of glycosaminoglycans during normal palatogenesis in the rat. Arch OralBiol 39:401–407.

35. Singh GD, Moxham BJ, Langley MS, Embery G (1997) Glycosaminoglycan biosynthesisduring 5-fluoro-2-deoxyuridine-induced palatal clefts in the rat. Arch Oral Biol 42:355–363.

36. Shah RM, Izadnegahdar MF, Hehn BM, Young AV (1996) In vivo/in vitro studies on theeffects of cyclophosphamide on growth and differentiation of hamster palate.Anticancer Drugs 7:204–212.

37. Shimabukuro Y, et al. (2008) Basic fibroblast growth factor regulates expression ofheparan sulfate in human periodontal ligament cells. Matrix Biol 27:232–241.

38. Tien JY, Spicer AP (2005) Three vertebrate hyaluronan synthases are expressed duringmouse development in distinct spatial and temporal patterns. Dev Dyn 233:130–141.

39. Evanko SP, Wight TN (1999) Intracellular localization of hyaluronan in proliferatingcells. J Histochem Cytochem 47:1331–1342.

40. Bosi G, et al. (1998) Diphenylhydantoin affects glycosaminoglycans and collagenproduction by human fibroblasts from cleft palate patients. J Dent Res 77:1613–1621.

41. Bodo M, et al. (1996) Effects of interleukins on Crouzon fibroblast phenotype in vitro.Release of cytokines and IL-6 mRNA expression. Cytokine 8:772–783.

42. Baroni T, et al. (2002) Crouzon’s syndrome: Differential in vitro secretion of bFGF,TGFbeta I isoforms and extracellular matrix macromolecules in patients with FGFR2gene mutation. Cytokine 19:94–101.

43. Shah RM, Arcadi F, Suen R, Burdett DN (1989) Effects of cyclophosphamide on thesecondary palate development in golden Syrian hamster: Teratology, morphology,and morphometry. J Craniofac Genet Dev Biol 9:381–396.

44. Singh GD, Moxham BJ (1993) Cellular activity in the developing palate of the ratassessed by the staining of nucleolar organiser regions. J Anat 182:163–168.

45. Meng L, Bian Z, Torensma R, Von den Hoff JW (2009) Biological mechanisms inpalatogenesis and cleft palate. J Dent Res 88:22–33.

46. Welsh IC, Hagge-Greenberg A, O’Brien TP (2007) A dosage-dependent role for Spry2in growth and patterning during palate development. Mech Dev 124:746–761.

2520 | www.pnas.org/cgi/doi/10.1073/pnas.0913985107 Snyder-Warwick et al.

Dow

nloa

ded

by g

uest

on

Apr

il 3,

202

0