A dosage-dependent role for Spry2 in growth and patterning during palate development

Abstract

The formation of the palate involves the coordinated outgrowth, elevation and midline fusion of bilateral shelves leading to the separation of the oral and nasal cavities. Reciprocal signaling between adjacent fields of epithelial and mesenchymal cells directs palatal shelf growth and morphogenesis. Loss of function mutations in genes encoding FGF ligands and receptors have demonstrated a critical role for FGF signaling in mediating these epithelial-mesenchymal interactions. The Sprouty family of genes encode modulators of FGF signaling. We have established that mice carrying a deletion that removes the FGF signaling antagonist Spry2 have cleft palate. We show that excessive cell proliferation in the Spry2-deficient palate is accompanied by the abnormal progression of shape changes and movements required for medially-directed shelf outgrowth and midline contact. Expression of the FGF responsive transcription factors Etv5, Msx1, and Barx1, as well as the morphogen Shh, is restricted to specific regions of the developing palate. We detected elevated and ectopic expression of these transcription factors and disorganized Shh expression in the Spry2-deficient palate. Mice carrying a targeted disruption of Spry2 fail to complement the craniofacial phenotype characterized in Spry2 deletion mice. Furthermore, a Spry2-BAC transgene rescues the palate defect. However, the BAC transgenic mouse lines express reduced levels of Spry2. The resulting hypomorphic phenotype demonstrates that palate development is Spry2 dosage sensitive. Our results demonstrate the importance of proper FGF signaling thresholds in regulation of epithelial-mesenchymal interactions and cellular responses necessary for coordinated morphogenesis of the face and palate.

1. Introduction

Facial morphogenesis requires the coordinated outgrowth, patterning and fusion of multiple structures. Paired facial primordia, derived from the first branchial arch, give rise to the mandibular and maxillary arches that extend and join at the midline with the frontonasal process to frame the lower jaw, upper jaw and face. A more recent evolutionary advance is the secondary palate, formed by the outgrowth and midline fusion of bilateral maxillary shelves to separate the oral and nasal cavities. The palatal shelves are an elongated bud-like outgrowth of maxillary mesenchyme covered by an epithelial sheet. During mouse development proliferative expansion of the cranial neural crest derived mesenchyme results in the initial vertical extension of the palatal shelves to a position lateral to the tongue. Following this initial phase of vertical growth, midline contact of the bilateral shelves is achieved by a combination of rotation and elevation movements as well as medially directed growth. Contact between the bilateral shelves is typically initiated in the mid palate region first where the vertical shelves swing dorsally into close apposition above the tongue. In the anterior and more posterior palatal shelf, contact requires more extensive tissue reshaping to orient outgrowth towards the midline (Chou et al., 2004; Sakamoto et al., 1989). Upon midline contact from E14.5-E15.5 remodeling and fusion removes the medial epithelial seam in order to form a continuous palate (Ferguson, 1988). Concurrent with these morphogenic processes, fields of cells must also be directed to exit the cell cycle in order to initiate skeletogenic differentiation to form the bony processes of the hard palate that extend from the maxilla and palatine bones. These bones arise from mesenchymal condensations from which cells directly differentiate into bone forming osteoblasts, a process known as intramembranous ossification. Mutations that alter regulation at any stage of this complex developmental process can lead to prominent birth defects including clefting of the lip and secondary palate.

The outgrowth of bud-like structures such as the palatal shelves is regulated via interactions between the overlying epithelium and underlying mesenchyme. These epithelial-mesenchymal interactions are mediated by growth-factor dependent signaling pathways in order to provide instructive cues necessary to coordinate a range of cellular behaviors. For example, the fibroblast growth factor (FGF) and bone morphogenetic protein (BMP) signaling pathways control cellular responses including changes in gene expression, cell adhesion, proliferation, and survival. Furthermore, regional domains in the facial primordia and palate are defined by differences in gene expression in response to these signaling pathways with consequences for the growth and patterning along the anterior-posterior axis of the palate (Hilliard et al., 2005). Msx1 and Barx1 are two homeobox transcription factors involved in the morphogenesis of several craniofacial elements (Barlow et al., 1999; Bei and Maas, 1998; Satokata and Maas, 1994; Tucker et al., 1998a). Distal Msx1 and proximal Barx1 expression in the mesenchyme of the first branchial arch is maintained in the developing maxilla to form complementary anterior Msx1 and posterior Barx1 expression domains in the palatal shelves (Barlow et al., 1999). Msx1 expression in the anterior palate has been shown to depend on Bmp4 (Zhang et al., 2002). However, Msx1 is known to be a downstream target of both BMP and FGF signaling in various developmental contexts. It has also been demonstrated that Barx1 expression is regulated by FGF and BMP signaling. For example, local inhibition of BMP4 signaling results in FGF associated ectopic expression of Barx1 in the distal mandible and leads to transformation of the distal incisor to a proximal molar tooth phenotype (Tucker et al., 1998b). Thus, rather than pathway specific input, the patterning of the proximal-distal axis of the branchial arch and subsequent anterior-posterior axis of the palatal shelf into domains of Msx1 and Barx1 expression is likely dependent on the relative local strength of BMP and FGF signaling.

Negative feedback or inhibitory interactions play a critical role in setting the relative strength of various signaling pathways. Elucidating the role of specific inhibitors in defining signaling domains and thresholds will provide insights into the cellular signaling mechanisms that direct orofacial development. FGF signaling is antagonized by sprouty (spry), an intracellular inhibitor of the receptor tyrosine kinase (RTK) pathway, identified as a negative regulator of airway branching in Drosophila (Hacohen et al., 1998). SPRY antagonistic activity is evolutionarily conserved and the expression of Spry1, Spry2 and Spry4 overlaps with FGF signaling domains during mouse development including developing craniofacial structures (Minowada et al., 1999). The FGF signaling pathway plays a central role in reciprocal interactions between adjacent tissues during diverse morphogenic processes including palate and facial development (Thisse and Thisse, 2005). Several members of the FGF family of ligands and their receptors including Fgf8, Fgf9, Fgf10, Fgf18, Fgfr1, and Fgfr2 are specifically expressed in either the oral epithelium or mesenchyme during craniofacial morphogenesis and when mutated lead to cleft palate (Alappat et al., 2005; Colvin et al., 2001; Liu et al., 2002; Rice et al., 2004; Trokovic et al., 2003; Trumpp et al., 1999). The craniofacial defects that result from mutations in these genes are often associated with haploinsufficient or variably penetrant phenotypes. This further highlights the importance of establishing the proper FGF signaling dose in the regulation of these processes.

Phenotypic analysis of targeted mutations has demonstrated a requirement for vertebrate Sprouty genes during kidney morphogenesis, differentiation of the auditory epithelium, regulation of odontogenesis, limb patterning, and enteric neuron development (Basson et al., 2005; Hansen et al., 2003; Klein et al., 2006; Shim et al., 2005; Taketomi et al., 2005). In addition, disruption of both Spry2 and Spry4 results in severe defects in craniofacial, limb, and lung morphogenesis (Taniguchi et al., 2007). In this study we examine the connection between Spry2 function and FGF signaling interactions that are involved in palate and facial development. We show that palate morphogenesis is disrupted in mice homozygous for a Mb-scale deletion that removes the Spry2 locus. Abnormal palate development is marked by excessive cell proliferation and the misexpression of FGF-responsive genes. We also establish that a targeted disruption of Spry2 fails to complement the cleft palate phenotype uncovered by the Spry2-deletion allele. In addition, Spry2-BAC transgenic mouse lines expressing reduced levels of Spry2 show complete rescue of facial clefting and dosage-dependent rescue of the palatal defect. The sensitivity of palate development to Spry2 dose highlights the importance of proper regulation of FGF signaling thresholds necessary to coordinate the timing of craniofacial development.

2. Results

2.1. Spry2 deficient mice exhibit palate and facial clefting

We identified Spry2 as a candidate gene for craniofacial defects in mice homozygous for the Ednrb^s-36Pub (hereafter, 36Pub) deficiency of the piebald deletion complex (Peterson et al., 2002; Roix et al., 2001). The piebald deletion complex is a collection of overlapping Mb-scale chromosomal deficiencies centered around the endothelin receptor B (Ednrb) coat color spotting locus on distal mouse chromosome 14 (Mmu14). We have used the piebald deletions to uncover several loci essential for mammalian development. 36Pub deletion homozygotes (36Pub−/−) have a high incidence (83%) of cleft secondary palate and a lower incidence (27%) of cleft lip (Table 1). The genetically defined ∼1 Mb 36Pub critical interval completely removes the FGF signaling antagonist Spry2 as well as two other genes, Rbm26 and Ndfip2, ∼600 Kb proximal to Spry2 (Peterson et al., 2002). Based upon the nature of the 36Pub palate phenotype and the established role of FGF signaling in the regulation of cellular proliferation of both the epithelium and mesenchyme during palate development we prioritized Spry2 as a candidate gene within the 36Pub critical interval (Rice et al., 2004).

Table 1.

Spry2-BAC rescue and Spry2^Δ_ORF failure to complement facial defects in 36Pub mutants

Incidence of cleft palate1:	36Pub/+	Spry2ΔORF/+	36Pub/36Pub	36Pub/ Spry2ΔORF
Spry2-BAC nontransgenic	0/58	0/10	10/12 (83%)	6/17 (35%)
Spry2-BAC line 69 hemizygotes	-	-	1/12 (8%)	-
Spry2-BAC line 2 hemizygotes	-	-	5/12 (42%)	-
Incidence of facial clefting1:	36Pub/+	Spry2ΔORF/+	36Pub/36Pub	36Pub/Spry2ΔORF
Spry2-BAC nontransgenic	0/58	0/10	19/71 (27%)	0/17
Spry2-BAC line 69 hemizygotes	-	-	0/23	-
Spry2-BAC line 2 hemizygotes	-	-	0/25	-

¹Cleft palate scored at E18.5; Facial clefting scored in E12.5-E18.5 embryos.

We have further characterized the developmental basis of the clefting defects associated with the 36Pub mutant mouse. We compared serial frontal sections from E13.5 to E15.5 embryos to identify differences in the developmental progression of the wildtype and mutant palatal shelves. Palatal shelf morphology between mutant and wildtype embryos was comparable at E13.5 (data not shown). However, by E14.5 clear differences in the morphology of the 36Pub mutant were apparent, particularly in the mid to posterior palate. The shelves of the E14.5 wildtype palate had elevated dorsal to the tongue and contacted in the mid palate (Fig.1A-D). In contrast, the mutant shelves remained positioned lateral to the tongue in the anterior and middle region (Fig. 1E-F). Additionally, shape changes associated with the directed growth of the wildtype palate were altered in the mutant. Specifically, the distinct medially projecting prominence seen along the posterior palate and soft palate (Fig. 1C-D) was absent in the mutant (Fig. 1G-H). Instead, it appeared that the posterior region of the E14.5 mutant palate had failed to transition from the vertically directed growth that is more typical of E13.5 embryos. By E15.5 the palatal shelves of the wildtype embryo were in contact along their entire length and had nearly completed fusion as evidenced by the loss of the medial edge epithelium (Fig. 1I-L). Although the posterior half of the E15.5 mutant palate did show limited evidence of redirected growth towards the midline the anterior shelves had not elevated, possibly due to obstruction by the tongue (Fig. 1M-P). We observed that the morphology of the posterior palate of mutants is markedly altered even when compared to E14.5 wildtype embryos whose palates have not yet elevated (see Fig. 5). These observations suggest that medially directed growth of the posterior palate over the tongue may play a more significant role in the mechanical elevation of the palate than has been previously appreciated.

Fig. 1.

Altered morphogenic progression of the 36Pub mutant palate. H&E stained serial frontal sections through the entire anterior-posterior length of E14.5 and E15.5 palates were grouped into four regions each representing one quarter of the total palatal length and the morphology of the palate within each region was compared between wildtype (E14.5 A-D; E15.5 I-L) and mutant (E14.5 E-H; E15.5 M-P). Images show the right side palatal shelf of the wildtype beside the contralateral mutant shelf from the same region for comparative purposes. By E14.5 the anterior palatal shelves (A) of the wildtype have elevated above the tongue and in the middle region (B) the shelves are in contact although still separated by the medial epithelial seam. The posterior palate (C) and region of the soft palate (D) of the E14.5 wildtype exhibit a medially projecting prominence indicative of directed growth towards the midline (black arrows). E14.5 36Pub mutant palates have not elevated in the anterior half (E, F) and the posterior palate (G) and region of the soft palate (H) have failed to acquire the shape changes associated with the transition from vertically- to medially-directed outgrowth. The E15.5 wildtype palate is closed along its entire A-P length (I-L) and fusion of the shelves by breakdown of the medial epithelial seam is nearly complete (asterisks). By E15.5 the 36Pub mutant palate has still not elevated dorsal to the tongue (M,N) although the posterior half of the palate does exhibit limited growth towards the midline (arrows in O&P). Arrowheads point to sites of mesodermal condensation and matrix deposition associated with incipient intramembranous ossification that is particularly impacted in the middle and posterior of the mutant palate. Abbreviations: m, Meckel's cartilage; mes, medial epithelial seam; ns, nasal septum; t, tongue.

Fig. 5.

Altered gene expression along the anterior-posterior axis of the 36Pub mutant palate. (A-D) Etv5 expression. (A) Etv5 expression in the E13.5 wildtype palate is restricted to segmented domains in the anterior two-thirds of the epithelium and in the mesenchyme to the medial aspect of the vertical shelf. Expression of Etv5 in E13.5 36Pub (B) mutant is upregulated and expanded in both the epithelial and mesenchymal domains (arrows). (C) Wildtype expression of Etv5 in the mesenchyme at E14.5 becomes restricted to the anterior two-thirds of the palatal shelf. In E14.5 36Pub mutants (D) Etv5 remains upregulated and expanded in the epithelium and ectopic mesenchymal expression extends into the posterior of the shelves (arrows in D). (E-H) Barx1 expression. (E) E13.5 wildtype Barx1 expression is restricted to segmented domains in the anterior epithelium and to the mid-posterior half of the palatal mesenchyme. In the E13.5 36Pub mutant (F) Barx1 expression is elevated in the anterior epithelium and the posterior mesenchymal domain is expanded. Note arrows marking elevated medial domain of Barx1 in the posterior of mutants compared to wildtype, also compare bars marking length of Barx1 negative domains in the anterior mesenchyme. (G) Barx1 in the E14.5 wildtype is expressed in anterior epithelium and posterior mesenchyme and elevated and expanded expression domains persist in the E14.5 36Pub mutant (arrows in H). (I-L) Msx1 expression. (I) Msx1 is expressed at low levels in the mesenchyme in the anterior third of the E13.5 wildtype palate. (J) The 36Pub mutant palate shows elevated Msx1 expression anteriorly (arrows) and ectopic expression in the mid-posterior palate (arrowheads). (K) Weak Msx1 expression in the E14.5 wildtype anterior palate compared with (L) continued elevated expression in the E14.5 36Pub mutant palate (arrows). (M-P) Tbx22 expression. (M) Tbx22 is strongly expressed in the mesenchyme of the mid-posterior of the E13.5 wildtype palate. (N) We detected no difference in Tbx22 expression in the E13.5 36Pub mutant. (O) At E14.5 as the posterior palate initiates medially directed growth the wildtype Tbx22 expression domain extends to the posterior end of the palate. (P) Failure to initiate posterior expansion of Tbx22 expression at E14.5 (white arrowheads) is associated with altered morphological progression of the posterior palate in 36Pub mutants. Dashed lines outline posterior palatal shelf to highlight altered morphology of E14.5 36Pub mutants.

2.2. Dynamic expression of Spry2 and components of the FGF signaling pathway in the facial primordia and palate

To further evaluate the role of Spry2 during craniofacial development we examined its expression pattern during early facial morphogenesis. In embryos from E9.5 to E11.5 Spry2 was expressed in known FGF signaling domains of the developing facial primordia including the frontonasal process and maxillary and mandibular components of the first branchial arch (Fig. 2A&B). Spry2 was also dynamically expressed in the epithelium and mesenchyme throughout palate development. At E12.5 Spry2 is expressed in the anterior palate in the mesenchyme and overlying epithelium. In the E14.5 pre-fusion palate Spry2 expression becomes more restricted to the epithelium on the oral surface of the mid-palate, and along the medial aspect of the approaching palatal shelves with notably intense expression in the posterior half of the palate (Fig. 2C-D).

Fig. 2.

Spry2 expression during orofacial development. (A, B) Spry2 is expressed in structures that will outgrow and fuse to form the primary palate and jaw. (A) At E9.5 Spry2 is expressed in the telencephalon and olfactory placode as well as in both the maxillary and mandibular processes of the first branchial arch (arrow). By E11.5 (B) Spry2 expression is seen in the area of fusion between the frontonasal process and the anterior maxillary process of the first branchial arch (white arrow). (C, D) Wholemount view of the oronasal cavity (rostrocaudal from top to bottom) showing dynamic expression of Spry2 during palate development between E12.5 and E14.5. (C) At E12.5, Spry2 is strongly expressed in the epithelium and mesenchyme of the anterior half of the oral cavity that is derived from the mid-maxilla of the first branchial arch. (D) Prior to contact of the shelves at E14.5 Spry2 expression is seen in a broad domain in the mid-posterior palatal epithelium (arrow) and the along the medial aspect of the approaching shelves with strongest expression in the posterior half of the palate (white arrowhead). (E-H) Wholemount view of oronasal cavity (rostrocaudal from top to bottom) and (I-T) frontal section views of gene expression at E13.5 showing Spry2 spatially referenced to the expression of components of the FGF signaling pathway (Fgf10: E, I, J, K; Fgfr2: F, L, M, N; Spry2: G, O, P, Q; Etv5: H, R, S, T). (E) Fgf10 is expressed in the mesenchyme of the anterior two-thirds of the palate and becomes progressively localized to the oral side of the palatal shelf in the posterior region of its domain (I, J, & K). (F) Fgfr2b and Fgfr2c expression is detected with a single probe recognizing both isoforms. Fgfr2b is broadly expressed throughout the palatal epithelium (L, M, & N) whereas mesenchyme specific Fgfr2c is localized to discrete domains of osteogenic mesenchyme in the anterior maxilla and underlying the nasal epithelium (black arrowheads). (G,P) Spry2 is expressed throughout the palatal epithelium with strongest levels on the oral side of the mid-palate. Spry2 in the mesenchyme is strongly expressed in the anterior (O) and at slightly lower levels in the posterior (Q) with expression in the mid-palate tightly restricted to the mesenchyme subjacent to the nasal epithelium (P). (H) Etv5 in the epithelium is restricted to the anterior two-thirds of the palate in the oral surface directly overlaying the Fgf10 domain. Etv5 in the mesenchyme is broadly expressed in the anterior palatal shelves (R), although more medially restricted than that of Spry2 (compare O and R), and becomes progressively restricted to the mesenchyme subjacent to the nasal epithelium (S, T).

We next compared Spry2 expression during palate development in relation to the expression of other components of the FGF signaling pathway (Fig. 2E-T). FGF10 signaling has been shown to be critical for coordinating epithelial-mesenchymal interactions during the morphogenesis of several embryonic structures including the palate, limb, and lung (Min et al., 1998; Revest et al., 2001; Rice et al., 2004). Cleft palate due to the loss of Fgf10 function is associated with reduced proliferation throughout the palatal epithelium and to a lesser extent in the mesenchyme as well as increased apoptosis in the medial edge epithelium (MEE) (Rice et al., 2004). From E11.5 to E14.5 Fgf10 is expressed in the mesenchyme of the anterior two-thirds of the palate in a domain that will become the future hard palate. As development progresses from E12.5 to E13.5 Fgf10 expression becomes ventrally restricted to the oral side of the palate while maintaining a posterior boundary at the level of the molar tooth buds (Fig. 2E, I, J, & K).

Bidirectional FGF signaling between the epithelium and mesenchyme of the palate is mediated by fibroblast growth factor receptor 2 (Fgfr2). Alternative splicing of FGFR2 results in expression of epithelial specific FGFR2IIIb and mesenchymal specific FGFR2IIIc isoforms. Cleft palate due to specific loss of function of the Fgfr2b isoform demonstrates its role as the major receptor for mesenchymal FGF10 signaling (De Moerlooze et al., 2000; Rice et al., 2004). Fgfr2c isoform expression is required in early mesenchymal condensations to regulate normal osteoprogenitor cell proliferation and differentiation (Eswarakumar et al., 2002). An engineered gain of function mutation in Fgfr2c impacts osteoprogenitor differentiation and also results in cleft palate (Eswarakumar et al., 2004). We performed in situ hybridization using an antisense probe that recognizes both the IIIb and IIIc isoforms of Fgfr2 (Fig. 2F, L, M, & N). At E13.5, Fgfr2b is strongly expressed in the epithelium along the anterior-posterior length of the palate on both the oral and nasal side. Consistent with its role in osteoblast specification and differentiation, Fgfr2c signal at E13.5 was found in osteogenic regions prior to overt mesenchymal condensation in the anterior maxilla and along the medial aspect of the vertical palatal shelf adjacent to the nasal epithelium.

The ETS family of signal dependent transcription factors are known to mediate FGF/RTK signaling and their activity is regulated by phosphorylation by MAPK (Sharrocks, 2001). Etv5 is a member of the Pea3 subfamily of ETS transcription factors and as a direct transcriptional target of FGF/RTK signaling the expression pattern of Etv5 provides a report of the spatial activity of this signaling pathway. In the E13.5 palate, Etv5 is expressed in a broad domain within the anterior mesenchyme that becomes medially restricted in the mid-posterior palate to a domain encompassing that of Fgfr2c (Fig. 2H, R, S, & T). In the palatal epithelium, Etv5 is robustly expressed on the oral surface of the palatal shelf with a sharp medial boundary. Etv5 expression is not found in the MEE or in the nasal surface epithelium of the palatal shelf (Fig. 2R&S).

Spry2 expression largely overlaps that of Etv5 in the E13.5 palate with some key differences (Fig. 2G, O, P, & Q). In the anterior and posterior mesenchyme, Spry2 is more broadly expressed along the mediolateral axis of the palatal shelf, whereas in the mid palate expression is more tightly restricted to the medial aspect of the shelf within the domain of Fgfr2c expression. Furthermore, although the most robust expression of Spry2 in the epithelium overlaps with Etv5 on the oral surface it is expressed throughout the MEE and nasal side epithelium as well, implying differential regulation of FGF signaling strength along the medial aspect of the vertical shelf. The complementary and overlapping expression domains of Spry2 with other components of the FGF pathway in the palate highlight the reciprocal nature of FGF signaling during palate development and show a strong correlation with the phenotype of the Spry2-deficient 36Pub mutants.

2.3. Complementation testing and a Spry2-BAC transgenic rescue with a hypomorphic phenotype

A role for Spry2 in palate and facial development is indicated by its expression pattern, activity as an FGF inhibitor, and the craniofacial defects observed in the Spry2 deficient, 36Pub deletion mutant mice. A targeted mutation of Spry2 (Spry2^Δ_ORF) has recently been reported that demonstrates a critical role for Spry2 in postnatal viability, differentiation of the auditory epithelium, and restricting responsiveness to epithelial-mesenchymal FGF signaling during early tooth development (Klein et al., 2006; Shim et al., 2005). However, palate defects have not been reported in Spry2^Δ_ORF loss of function mice. In order to determine whether Spry2 contributes to the craniofacial defects associated with the 36Pub deletion we performed a complementation cross using 36Pub deletion and Spry2^Δ_ORF mice. We scored palate closure at E18.5 in the progeny of intercrosses of Spry2^Δ_ORF and 36Pub heterozygous animals. We observed clefts of the secondary palate in 6 of 17 (35%) 36Pub/ Spry2^Δ_ORF compound mutants (table 1), indicating a failure of the Spry2^Δ_ORF mutation to complement craniofacial defects associated with the 36Pub deletion.

We also used a BAC transgenic approach to further evaluate Spry2 function during palate development. We tested whether a Spry2-BAC transgene could complement the craniofacial defects associated with the 36Pub deletion. Computational gene prediction (e.g. GRAIL), similarity searches against gene, EST and protein databases and comparative analysis of mouse and human genomic sequences support that the centrally located Spry2 is the only gene contained on the 80 Kb BAC used in these experiments (Peterson et al., 2002). Seven Spry2-BAC transgenic founders were generated. Analysis using markers that are polymorphic between the 129-derived BAC sequences and the C57BL/6J host genome revealed that portions of the Spry2-BAC had been deleted in two of the transgenic lines. The remaining five lines were used to generate offspring that were hemizygous for the Spry2-BAC transgene and homozygous for the 36Pub deletion (36Pub−/−, Spry2-Tg). This permitted the Spry2 expression pattern and levels derived from the Spry2-BAC transgene to be analyzed in the absence of endogenous Spry2 expression.

The expression of Spry2 in each of the BAC transgenic lines was initially examined by in situ hybridization in E10.5 36Pub−/−, Spry2-Tg embryos. Two transgenic lines (2 and 69) were selected that expressed readily detectable levels of Spry2 in a pattern comparable with endogenous Spry2. E10.5 36Pub−/−, Spry2-Tg-2 and 36Pub−/−, Spry2-Tg-69 embryos showed expression of Spry2 in regions including the developing CNS, limb buds and facial primordia (Fig. 3A-C). We also examined embryos at later stages to document expression in the developing palate. Spry2 expression was detected in the palatal tissue of E13.5 to E15.5 36Pub−/−, Spry2-Tg embryos in a pattern comparable to that seen in normal embryos (Fig. 3D-F). Although the expression pattern from the Spry2-BAC-2 and 69 transgenes was similar to that of the endogenous locus, the level of expression was diminished. We performed quantitative RT-PCR analysis to examine the expression level of Spry2 in the BAC transgenic lines compared with the endogenous locus. Quantitative expression studies focused on the E14.5 palate and showed that BAC transgenic line 2 expressed Spry2 at ∼12.5% and line 69 at ∼25% of wildtype levels (Fig. 3G).

Fig. 3.

Spatial expression pattern from Spry2-BAC transgene is similar to endogenous Spry2 expression but at reduced levels. (A-F) The Spry2-BAC drives expression in a pattern comparable to the endogenous locus. Spry2 transcripts are detected in the developing branchial arches and facial primordia of wildtype embryos at E10.5 (A) and in a frontal view of E13.5 wildtype palatal shelves (D). The Spry2-BAC hemizygous, 36Pub mutant shows comparable expression in the developing E10.5 embryo (B) as well as in the palatal shelves at E13.5 (E), note reduced expression levels (arrowheads). Spry2 transcripts are absent in the non-transgenic 36Pub mutant E10.5 embryo (C) and E13.5 palatal shelves (F). (G) Quantitative RT-PCR analysis of Spry2 expression in Spry2-BAC transgenic line 2 and 69 using E14.5 palatal tissue shows Spry2 is expressed at ∼12.5% and ∼25% of endogenous levels, respectively.

Although the Spry2-BAC transgenic lines express reduced levels of Spry2, we tested the ability of each transgene to rescue defective facial and palate development in 36Pub deletion mutant mice. Both Spry2-BAC transgenic lines 2 and 69 completely rescued the 27% incidence of cleft lip in the 36Pub mutant mice (Fig. 4A,B, & C, and Table 1). Mice representing both of the Spry2-BAC transgenic lines examined at E18.5 showed that rescue of the cleft palate phenotype is dosage-sensitive. The incidence of cleft palate was reduced to 42% in 36Pub−/−, Spry2-BAC-2 transgenic mice, while in the 36Pub−/−, Spry2-BAC-69 transgenic mice the incidence of cleft palate was further reduced to 8% (Fig.4D, E, & F, and Table 1). Thus, the reduced levels of expression in the Spry2-BAC transgenic mice are associated with the hypomorphic phenotype of incompletely penetrant palate clefting.

Fig. 4.

Hypomorphic rescue by Spry2-BAC transgene of craniofacial defects observed in 36Pub mutant mice. (A-C) Spry2-BAC transgene rescues defects of the primary palate. E15.5 wildtype (A) and E15.5 Spry2-BAC hemizygous, 36Pub mutant mice (B) show normal development of the primary palate and lip whereas E15.5 36Pub mutant mice (C) exhibit facial clefting. (D-F) Rescue of cleft palate at E18.5 is shown by complete closure in wildtype (D) and Spry2-BAC hemizygous, 36Pub mutant mice (E) compared with cleft palate in the 36Pub mutant (F).

2.4. Misexpression of FGF-responsive genes and altered palatal shelf patterning in the absence of Spry2

FGF signaling defines gene expression domains during facial development (Bachler and Neubuser, 2001; Francis-West et al., 1998; Shigetani et al., 2000; Trumpp et al., 1999). We examined FGF-responsive genes for misexpression resulting from the loss of Spry2. The ETS transcription factor family member Etv5 is regulated by FGF signaling during development of the face (Firnberg and Neubuser, 2002). Compared to control littermates, 36Pub mutants show a persistent upregulation and expansion of Etv5 throughout the anterior oral epithelium from E13.5 to E14.5 (Fig. 5A-D). At E13.5 Etv5 is also expressed in the mesenchyme along the A-P axis in the medial aspect of the vertical shelf with only the posterior most region of the future soft palate lacking Etv5 expression. By E14.5 the posterior limit of this domain is shifted anteriorly such that only the anterior two thirds of the palate show mesenchymal Etv5 expression (Fig. 5A&C). In E13.5 36Pub mutants this mesenchymal Etv5 expression appears elevated and by E14.5 extends ectopically into the posterior limit of the palatal shelf (compare Fig. 5A&C with 5B&D). These results are consistent with elevated and expanded FGF signaling activity due to the loss of Spry2 antagonism in the 36Pub mutant palate.

The homeobox transcription factor Barx1 is induced by FGF8 in the proximal mesenchyme of the E10.5 first branchial arch (Trumpp et al., 1999). The extent of the Barx1 expression domain in the maxillomandibular region and the resulting patterning of the facial primordia is determined by the balance of FGF/BMP signaling (Barlow et al., 1999; Shigetani et al., 2000). We therefore examined the impact of altered FGF signaling thresholds due to the loss of Spry2 on Barx1 expression. In E13.5 36Pub mutants Barx1 expression in the anterior epithelium is upregulated and expanded. The Barx1 expression domain in the posterior mesenchyme of E13.5 36Pub mutants is also expanded, both anteriorly and medio-laterally, into domains that normally show lower levels of Barx1 expression in wildtype embryos. Of particular note is the elevated expression in the medial aspect of the posterior palate where the subsequent altered morphology of the E14.5 mutant palate is most evident (arrows in Fig. 5E & F and dashed lines in E14.5 palates in Fig. 5). Persistent upregulated Barx1 expression is still evident at E14.5 to E15.5 (Fig. 5H and data not shown).

Msx1 in the anterior palate is a critical part of a BMP4-responsive pathway central to mesenchymal cell proliferation (Zhang et al., 2002). In certain contexts, Msx1 expression is also regulated by FGFs, for example in the developing tooth and cranial sutures (Alappat et al., 2003; Bei and Maas, 1998; Kettunen and Thesleff, 1998; Kim et al., 1998). Although bead implantation studies have demonstrated that FGF is able to induce Msx1 expression in palatal mesenchyme, Msx1 expression is unaltered in Fgf10−/− mutant palates (Alappat et al., 2005; Rice et al., 2005). Therefore, it remains unclear whether Msx1 expression in the palatal mesenchyme is also jointly regulated by BMP and FGF signaling. In the anterior palate the FGF antagonist Spry2 may act to restrict the role that FGF signaling plays in Msx1 expression. In support of this hypothesis, we found that Msx1 expression in the anterior palate is elevated in Spry2-deficient 36Pub mutants at E13.5 (Fig. 5 I&J). Persistent elevated Msx1 expression is still evident in E14.5 mutant palates when compared to wildtype (Fig. 5 K&L). Furthermore, the posterior boundary of the normally anteriorly restricted Msx1 is seen to ectopically extend into the mid-posterior palate of 36Pub mutants (Fig. 5 I-L).

Mutations in the T-box containing transcription factor Tbx22 are frequently associated with X linked cleft palate and ankyloglossia, an abnormal thickening of the frenulum of the tongue that restricts its mobility (Marcano et al., 2004). During palate development, Tbx22 shows restricted expression to the middle and posterior mesenchyme (Fig. 5M), a region significantly altered in Spry2-deficient palates. As the posterior palate of the E14.5 wildtype initiates medially directed outgrowth, the domain of Tbx22 also expands to become expressed throughout the medial aspect of the posterior third of the palate (Fig. 5O). Expression of Tbx22 in E13.5 mutants appeared normal (Fig. 5N). However, at E14.5 the altered morphology in the posterior palate of 36Pub mutants was accompanied by a failure to posteriorly expand the Tbx22 expression domain (Fig. 5P). Consistent with perturbed epithelial-mesenchymal signaling, we observed elevated and ectopic expression of the FGF-responsive genes Etv5, Barx1, and Msx1 as well as the altered expression of Tbx22 in regions of the 36Pub mutant palate that normally express Spry2.

2.5. Loss of Spry2 alters Shh expression and signaling domains in the developing palate

SHH signaling plays a critical role in craniofacial development and disruption of this pathway can result in severe craniofacial defects including holoprosencephaly. During palate development the expression of Shh and that of its receptor Patched1 (Ptc1) is restricted to regions of thickened oral epithelium and the immediately adjacent mesenchyme, respectively (Rice et al., 2006). These regions of epithelial thickening on the oral surface of the palatal shelves correspond to the developing rugae. Typically, 7 to 9 rugae are found on the oral surface of the palate at birth (Fig. 6F). Based on the signaling center activity of regions of thickened epithelium in other developmental contexts such as the apical ectodermal ridge (AER) of the limb bud and the enamel knot of the tooth, it has been proposed that the developing rugae may similarly function to coordinate patterning and morphogenesis of the palate (Rice et al., 2004). It has been demonstrated that Shh expression in the developing palate is dependent upon input from both the FGF and BMP signaling pathways, highlighting signaling interactions between a network of genes that guide craniofacial morphogenesis (Rice et al., 2004; Zhang et al., 2002). We were therefore interested in determining whether the altered FGF signaling thresholds in the 36Pub mutant palate may also impact Shh expression.

Fig. 6.

Shh expression in the rugae highlights regional growth of the palate and Shh expression is altered by loss of Spry2. (A) Shh expression on the oral surface of the palatal shelves in wildtype embryos from E11.5 through E14.5. Expression is restricted to stripes corresponding to developing rugae (numbered 1-7) and a posterior domain (pd) corresponding to the presumptive soft palate. The number of Shh expressing rugae increases as palate development progresses. Using expression in the molar tooth bud (asterisks) and the first distinct stripe of Shh (1) as landmarks illustrates that the relative position of these two domains remains constant whereas the number of rugae increases anterior to this position. Because new rugae are formed just anterior to stripe #1, numbering the rugae in order of their formation places each subsequently formed rugae (#3-7) between the anterior most stripe #2 and the posterior stripe #1. This results in the apparent posterior regression of the rugae forming region and presumptive soft palate relative to the rostral extension of the anterior palate (compare relative A-P positions of red asterisks). (B&C) Well defined domains of Shh expression in the rugae of E14.5 wildtype (B), Shh expression in the E14.5 36Pub mutant palate is both diffuse and fragmented into isolated clumps of expressing cells (arrowheads in C). (D&E) Wildtype expression of Ptc1 is restricted to the mesenchyme immediately adjacent to the overlying Shh domain in the rugae and broadly expressed in a posterior domain (pd) of the mesenchyme of the presumptive soft palate (D). In the E14.5 36Pub mutant palate (E) Ptc1 expression is disorganized and ectopically expressed in expanded domains of the palate. Note that Ptc1 expression in stripe #1 appears less impacted than in more anterior domains (arrowheads in E). (F&G) Rugae morphology at E18.5. The wildtype palate typically has an orderly array of 7 to 9 rugae, the 3 anterior most of which fuse across the midline (F). Rugae morphology is disorganized and fragmented in the cleft palates of both 36Pub homozygous (not shown) and 36Pub/ Spry^2Δ_ORF compound mutants (arrows in G) at E18.5.

We first confirmed that Shh expression on the oral surface of the palatal shelves is restricted to the developing rugae. We also noted an additional posterior domain of expression in the presumptive soft palate that is oriented orthogonal to bands of expression in the rugae (Fig. 6A). We observed that as palate development progresses the number of Shh expression bands increases, presumably due to the formation of additional rugae. Remarkably, careful analysis of a group of incrementally staged wildtype embryos highlights a region in the mid palate medial to the molar tooth bud that appears to define the location of new rugae formation (Fig. 6A). Using the molar tooth bud and the first formed stripe of Shh expression as constant reference points it can be seen that the periodic formation of new rugae occurs just anterior to the first formed stripe. Shh expression in the more anterior rugae is robust and constant, however, expression medial to the tooth bud appears to gradually increase and then separate into two bands that then become distinct stripes of expression. The region of rugae formation is initially positioned in the anterior half of the palate at E11.5. However, the elaboration of nascent rugae occurring in a generally anterior direction results in the apparent posterior regression of this region of the palate (compare position of red asterisks in Fig 6A). This implies that rostral extension of the anterior palate is accomplished at least in part through recruitment of cells from this more posterior region of the palate.

We next examined the expression of Shh and its receptor Ptc1 at E14.5 in wildtype and 36Pub mutant palates. Compared to the tightly organized bands of Shh seen in the wildtype palate, Shh expression in the mutant palate is defuse and fragmented into disorganized clumps of expressing cells (Fig. 6B&C). To determine whether the loss of Spry2 also impacts other components of SHH signaling we examined the expression of Ptc1 as a downstream target of Shh. In E14.5 wildtype palates, expression of Ptc1 is restricted to the mesenchyme immediately adjacent to the developing rugae but is more broadly expressed throughout the mesenchyme of the presumptive soft palate (Fig. 6D). Ptc1 expression in E14.5 36Pub mutant palates is no longer closely associated with the developing rugae and is highly disorganized with ectopic expression in broad domains of the palate (Fig. 6E). The loss of localized expression of Ptc1 in the 36Pub mutant palate suggests that spatially restricted SHH signaling is impacted. Interestingly, Ptc1 expression underlying the first formed stripe of Shh expression appears to be less impacted than subsequently formed rugae domains. As evidence of the impact of altered Shh expression during palate development, we observed that at E18.5 rugae morphology is severely disrupted in the cleft palates of both 36Pub homozygous and 36Pub/ Spry^2Δ_ORF compound mutants when compared to the orderly array of rugae seen in wildtype littermates (Fig. 6 F&G and data not shown). This data supports a model where SHH signaling from the rugae is required for correct patterning and growth of the palate and that the proper establishment and or maintenance of Shh expression in the palate is dependent on appropriate FGF signaling thresholds. Taken together, these data reveal a novel segmental organization and previously unappreciated dynamic underlying the growth and patterning of the developing palate.

2.6. Altered cellular responses in the Spry2-deficient palate

We next investigated whether morphogenic cellular responses such as proliferation, apoptosis or differentiation were altered in the 36Pub mutant. Typically, models of cleft palate where the shelves fail to make midline contact document diminished growth marked by reduced proliferation or increased cell death. We examined 36Pub mutant palates for changes in cell death as detected by Nile blue sulfate staining and TUNEL assay. Cell death appeared unaltered throughout the E14.5 mutant palate (data not shown). Cellular proliferation assayed by BrdU incorporation however, did reveal that the mitotic index in the E14.5 36Pub mutant palate was increased. Elevated proliferation was observed in both the epithelium and mesenchyme with the most significant increase evident in the mid to posterior palate, consistent with the regions showing altered morphological progression (Fig. 7 and Supplementary Table 1). Additionally, we measured cell proliferation at E13.5 and E15.5 and observed regionally elevated mitotic indices in the 36Pub mutant palate at these time points as well. Interestingly, the increased proliferation within the mutant palate relative to wildtype appears to involve both a temporal and spatial dynamic (Supplementary Table 1). While increased proliferation is not commonly recognized as a mechanistic basis for failed palatal development, such altered proliferation could lead to dysregulated growth and loss of coordinated progression of shelf morphology. The failure to acquire the correct shape at the appropriate time with respect to the overall growth of the face and oral cavity therefore represents a temporal dysmorphology in orofacial development that results in the failed midline contact of the palatal shelves.

Fig. 7.

Altered proliferation in the 36Pub mutant palate. Frontal sections in the mid-posterior region of E14.5 wildtype (A) and E14.5 36Pub mutant (B) palates show increased proliferation in the mutant shelves. Mitotic cells (green) were detected using a BrdU incorporation assay. Nuclei are counterstained with DAPI (blue). Note increased labeling of mitotic cells throughout both the overlying epithelium and underlying mesenchyme particularly along the medial aspect of the palatal shelf and mesenchyme underlying the MEE (white arrows). (C) A regional proliferation map of the mitotic index along the A-P axis of the palate highlights increased cell proliferation in the mutant palate (asterisks p<0.05, see Supplementary Table 1).

The morphogenic movements of shelf reorientation and closure occur concurrently with cellular differentiation that gives rise to the palatal processes of the maxilla and palatine bones of the hard palate. Interestingly, analysis of skeletal preparations of E18.5 wildtype, 36Pub−/−;Spry2-BAC-69 transgenic embryos revealed a reduced size and altered morphology of the palatal processes of the maxilla and palatine bones in the transgenic animals, whereas in the 36Pub−/− mutant these bones are vestigial and absent, respectively (Fig. 8A-C). In the wildtype palate, mesenchymal condensation and intramembranous ossification begins around E14.0. We compared skeletogenic differentiation of wildtype and mutant palates between E14.5 and E15.5. In E14.5 wildtype embryos mesenchymal condensations are present in the osteogenic compartment along the anterior-posterior length of the maxilla. By E15.5 these condensations extend into the fused palatal shelves and show signs of significant matrix deposition (see Fig. 1A,B,C & I,J,K). We found very little histological evidence of these mesenchymal condensations within the osteogenic compartments of the 36Pub mutant maxilla at E14.5 (see Fig. 1E,F,&G). The extent of ossification in E15.5 36Pub mutants was greatly reduced when compared to stage matched wildtype embryos (compare Fig. 1 I-K with M-O). We also noted that the degree of skeletal differentiation along the anterior-posterior axis in mutants appeared particularly reduced in the middle and posterior palate (compare Fig. 1I & 1M with Fig. 1K & 1O). This region corresponds to the area of the palate that would normally give rise to the maxillary and palatine processes. Thus, the development of the 36Pub mutant palatal shelves does not appear to be completely arrested. Rather, their morphological and differentiative progression appears to be retarded with the posterior half of the palate being most severely impacted.

Fig. 8.

Altered morphology of skeletal elements in the Spry2-deficient palate. (A-C) E18.5 alcian blue and alizarin red stained skeletal preparations identify a hypomorphic phenotype of reduced skeletal differentiation in Spry2-BAC hemizygous, 36Pub mutant mice. Compared to E18.5 wildtype (A) the palatal processes of the maxilla and palatine bones of the hard palate (outlined in dashed black line in A and B) are reduced in size and show altered morphology in the closed palates of E18.5 Spry2-BAC hemizygous, 36Pub mutant mice (B). In the nontransgenic E18.5 36Pub mutant (C) the palatal process of the maxilla is highly hypoplastic (outlined in dashed white line) while the more posterior process of the palatine bone is completely absent (arrows). Abbreviations: ppmx, palatal process of the maxilla; pppl, palatal process of the palatine bone; vm, vomeronasal.

3. Discussion

The formation of the face and palate is dependent on the highly coordinated outgrowth, movement and fusion of paired facial primordia and palatal shelves. These morphogenic steps are under the control of reciprocal signaling interactions between adjacent epithelia and mesenchyme that involve the integration of the FGF, BMP, and SHH pathways. In this study we have shown that the FGF antagonist Spry2 plays a role in craniofacial development. We investigated Spry2 as a gene required for normal palatogenesis based on its activity as a modulator of FGF/RTK signaling and expression in the developing palate. Morphogenesis and skeletal differentiation of the palate is disrupted in mice homozygous for the 36Pub deletion that removes the Spry2 locus. Furthermore, a targeted mutation of Spry2 fails to complement the 36Pub deletion cleft palate phenotype. Interestingly, mice homozygous for the Spry^2Δ_ORF mutation do not display cleft palate. Differences in genetic background potentially provide a basis for this discrepancy. The 36Pub deletion and Spry^2Δ_ORF mice used in this study are at N12 and N7 generations on a C57BL/6J genetic background, respectively. Additional studies will also be needed to determine whether the loss of another gene or genomic feature within the cleft palate critical interval sensitize the 36Pub mutant mice to the loss of Spry2 function and thereby contribute to the cleft palate phenotype. We also show that a Spry2-BAC transgene rescues craniofacial defects associated with the 36Pub deletion mutant in a dosage dependent manner. Consistent with the loss of SPRY2 antagonism of FGF signaling the 36Pub deletion mutant palates showed excessive proliferation along with elevated, ectopic expression of FGF-responsive genes. This study presents data that highlight a role for Spry2 in the regulation of complex dosage dependent signaling interactions controlling orofacial development.

3.1. Palate morphogenesis relies on growth factor signaling thresholds that are modulated by Spry2 to coordinate regional cell proliferation, patterning, and differentiation

Mutations in a number of genes disrupt various and distinct aspects of palatal development. As a first example, clefting in Tgfβ3 and Egfr mutants results from defective fusion of the palatal shelves. Although Spry2 also acts to modulate EGF signaling, the defect in the Spry2 deficient palate is manifest at earlier stages when FGF signaling has been shown to be critical for palate morphogenesis (Miettinen et al., 1999; Proetzel et al., 1995; Rice et al., 2004). Mutations disrupting Msx1 provide a second example where clefting is a result of reduced proliferation and arrested shelf outgrowth (Satokata and Maas, 1994; Zhang et al., 2002). Our analysis of loss of Spry2 function highlights a third mechanism for cleft palate where increased proliferation results in discoordination of palate growth and the morphogenic movements necessary for midline contact.

Our findings are consistent with increasing evidence that outgrowth and patterning is differentially regulated along the anterior-posterior axis of the palate and is achieved through regional differences in epithelial-mesenchymal signaling competence (Hilliard et al., 2005). The expanded and ectopic expression of Msx1 and Barx1 expression in Spry2-deficient embryos indicates that anterior-posterior patterning is altered. The reduced skeletal differentiation in more posterior skeletal elements potentially reflects the impact of this altered A-P patterning. The highest levels of Barx1 expression in the posterior palatal mesenchyme becomes progressively restricted to the non-osteogenic soft palate. Further studies will be needed to investigate whether the ectopic and elevated Barx1 expression in Spry2-deficient palates could interfere with osteoblastogenesis. One possibility is an alteration of WNT signaling shown to be important for osteoblast lineage development (Day et al., 2005). Targeted disruption of Barx1 has revealed its importance in the regulation of the WNT signaling antagonists sFrp1 and sFrp2 (Kim et al., 2005). Importantly, inhibition of canonical Wnt-β-catenin signaling has also been proposed to be critical for the differentiation of bipotential osteo-chondroprogenitor cells (Hartmann, 2006).

The failure to upregulate Tbx22 in the posterior palate at E14.5 may also be the result of altered patterning of the Spry2-deficient palate. It is interesting to note that putative Msx1 binding sites in conserved sequences of the 5′ regulatory region of Tbx22 have been identified and that Msx1 is typically considered to act as a transcriptional repressor (Herr et al., 2003). The ectopic expression of Msx1 in the mid-posterior palate in a domain that overlaps the normal domain of Tbx22 expression may explain the failure to upregulate Tbx22 at E14.5. Data from a targeted knock out of Tbx22 as to the function of this gene during palate development has not yet been reported. This study presents evidence that suggests Tbx22 is required for the normal growth of the posterior palate. Our data support a model where Spry2-mediated modulation of signal strength plays a key role in establishing thresholds that correspond with differential FGF signaling responsiveness important for patterning the anterior-posterior axis of the palate.

3.2. Spry2 as part of a signaling network that guides palate and facial development

Our results provide insight into FGF pathway control of tissue interactions involved in craniofacial development. Cellular proliferation during palate development has been shown to depend on multiple signaling pathways. Reciprocal epithelial-mesenchymal communication integrates this spatially distributed signaling activity into a network of interactions that coordinate the morphogenesis of the palate (Fig. 9). A mesenchymal Bmp4/Msx1 feedback loop is required for Shh expression leading to proliferation in the anterior palate (Zhang et al., 2002). Mesenchymal Fgf10 signaling through Fgfr2b has also been shown to regulate Shh expression and cell proliferation, particularly in the epithelium. The loss of normal expression of Shh and its receptor Ptc1 in the Spry2 deficient palate underscores the importance of maintaining appropriate levels of FGF signaling input to this network. Furthermore, the localized and dynamic expression of Shh suggest that the developing rugae act as signaling centers where the FGF, BMP and SHH pathways are integrated to form a network that controls palatal growth. Evidence supporting our model that Shh expression and new rugae formation represent landmarks for the organization of signaling interactions that direct the rostral extension of the palate comes from recent work by Li and Ding (Li and Ding, 2007). Based upon the expansion of the anteriorly restricted Shox-2 and regression of the posteriorly restricted Meox-2 expression domains from E12.5 to E14.5 without an accompanying regional difference in proliferation, the authors hypothesize that the expansion of the anterior palate involves recruitment of once posterior Meox-2 expressing cells and their conversion into anterior Shox-2 expressing cells.

Fig. 9.

Model of epithelial-mesenchymal signaling regulating cell proliferation during palatogenesis. In the anterior palate, mesenchymal proliferation is controlled by a Bmp4/Msx1 feedback loop required for Shh and Bmp2 expression. Proliferation and Shh expression in the epithelium is also dependent on Fgf10 signaling through Fgfr2b. Fgf9 is a candidate for an epithelial source of FGF regulating mesenchymal proliferation and gene expression. The palatal mesenchyme is patterned into anterior Msx1 and posterior Barx1 domains. The expression of these genes is responsive to thresholds of FGF signaling. Elevated FGF activity in the absence of Spry2 antagonism leads to increased proliferation, elevated and expanded Etv5 and Barx1 expression, ectopic expression of Msx1 (dashed arrow), and altered Shh expression. Therefore FGF signaling thresholds must be properly modulated to coordinate shelf outgrowth and loss of the FGF antagonist Spry2 results in perturbations of gene expression, cell proliferation and timing of palatal development.

Loss of the inhibitor Spry2 results in expanded expression of FGF responsive genes in both the epithelium and mesenchyme, increased proliferation in both the epithelium and mesenchyme, and altered morphodifferentiative progression of palate growth. Notably, Etv5 and Barx1 expression is expanded in the anterior epithelium adjacent to Fgf10 expressing mesenchyme and the expression of Barx1 and Msx1 is expanded and ectopic in the mesenchyme of Spry2-deficient mutants. These results strongly implicate a role for Spry2 inhibition of FGF signaling in both the epithelium and mesenchyme. A reciprocal source of FGF in palatal epithelium regulating mesenchymal proliferation has not been reported. However, based upon spatial gene expression and the 40% incidence of cleft palate in loss of function mouse mutants, Fgf9 is a strong candidate as an epithelial FGF signaling source in the palate (Colvin et al., 1999; Colvin et al., 2001). Fgf9 has been shown to be able to induce expression of Msx1 in oral mesenchyme during odontogenesis (Kettunen and Thesleff, 1998). It is therefore possible that the ectopic and expanded Msx1 expression in the Spry2-deficient palate is a result of the loss of antagonism leading to unregulated Fgf9 signaling through its high affinity receptor Fgfr2c in the mesenchyme (Ornitz et al., 1996) (Fig. 9). In support of this hypothesis Klein et al. report that the loss of Spry2 antagonism during tooth development leads to ectopic expression of FGF target genes in tissue that is normally refractory to this signaling input and that this defect can be fully rescued by genetically reducing Fgfr2 dosage (Klein et al., 2006). Additionally, an engineered gain of function mutation in Fgfr2c phenocopies the craniofacial defects of Crouzon and Pfeiffer syndromes and results in cleft palate and altered skeletal differentiation (Eswarakumar et al., 2004; Ibrahimi et al., 2004). This activating mutation provides an analogous model to the loss of Spry2 antagonism. Both the Fgfr2c gain of function and Spry2 loss of function reflect the resulting impact of altered FGF signaling strength on craniofacial development. Taken together, our data support a model where reciprocal tissue interactions involving FGF signaling controls proliferation, patterning, and the necessary shape changes required to close the palate utilizing different epithelial and mesenchymal FGF signaling effectors and a common antagonist, Spry2.

3.3. FGF signaling and the genetic basis of craniofacial malformations

Orofacial anomalies are among the most prominent class of birth defects, occurring in ∼1/500 to 1/2,500 births. The rapid proliferative expansion of the facial primordia and the need to coordinate the overlapping processes of complex morphogenic movements with cellular differentiation are defining features of the sensitivity of craniofacial development to disruption by genetic and environmental insult. Genetic analysis of human patients affected by craniofacial syndromes as well as gene perturbations in mouse has generated a great deal of insight into the genes involved in craniofacial development. Mutations leading to orofacial clefting can be categorized as either syndromic or nonsyndromic depending on the presence or absence of additional developmental anomalies (Jugessur and Murray, 2005). That a significant majority of cleft lip and palate defects appear to be nonsyndromic underscores the molecular and genetic complexity of the underlying developmental program. Precise and efficient communication between adjacent cell layers is a prerequisite for the ordered progression of such a program. Therefore, progress in the prevention of orofacial clefting will require not only the identification of additional genes, but also the placement of the various components into molecular pathways and interacting networks. Genetic network models will provide a framework for studying how various mutations and allelic combinations together with environmental factors impact the morphogenetic events required for normal orofacial development. The data presented here identify Spry2 as a key component of this network, acting to modulate FGF signaling strength and defining regional responsiveness to epithelial-mesenchymal signaling interactions that direct the outgrowth and patterning of the palatal shelves.

Identifying Spry2 as a gene critical for palate development provides further insight as to how altered FGF signaling contributes to craniofacial malformations and will help make this class of birth defects more genetically tractable. Further support for a role of Spry2 during palate development comes from medical sequence analysis of patients with cleft palate accompanied by a familial history of cleft palate. In a recent report Vieira and coworkers identified mutations of Spry2 in nucleotides that are conserved from Xenopus to human that along with mutations identified in 5 other genes may collectively account for as much as 6% of nonsyndromic clefting (Vieira et al., 2005). The dosage-dependent role of Spry2 also provides insight into the extent to which developmental processes can tolerate signaling perturbations. Mice carrying a deletion of Spry2 on an inbred C57BL/6J background have a high probability of developing clefts of both the primary and secondary palate. However, restoring Spry2 expression to ∼25% normal levels is enough to rescue the palatal defects by shifting shelf growth into a window that is compatible with the overall timing of orofacial development. Our findings emphasize that developmental control is achieved by a network of interactions that elicit morphogenic responses that are sensitive to signaling thresholds. Thus, variations in the overall signaling balance can lead to differences in final form that range from unique facial characteristics to birth defects.

4. Experimental procedures

4.1. Mice and production of Spry2-BAC transgenics

36Pub deletion mice were maintained on a C57BL/6J genetic background (currently N12). 36Pub mice were genotyped in a PCR assay using the deletion flanking markers D14Mit265 and D14Mit177 as previously described (Roix et al., 2001). Spry2^Δ_ORF mice were genotyped as described (Shim et al., 2005).

The Spry2-BAC used for transgenic mouse production was isolated by PCR screening the CITB-CJ7 (129S1/Sv) BAC library (Research Genetics). The size of the BAC was determined by matching BAC-end sequences against the mouse genome (Ensembl MGSC v15.30.1) and pulse field gel electrophoresis analysis as previously described (Peterson et al., 2002). Gene content and conserved sequence features on the BAC were characterized using computational methods as previously described (Peterson et al., 2002). C57BL/6J Spry2-BAC transgenic animals were produced by pronuclear injection of purified BAC DNA. Transgenic animals were genotyped using a PCR assay based on the (CA)_n repeat marker D14J42 that detects the 129S1/Sv BAC-derived sequence in the C57BL/6J inbred genomic background. Southern blot analysis was used to determine a copy number of 1-2 integration events per genome with strictly Mendelian inheritance for the Spry2-BAC lines 2 and 69 used in this study (data not shown). All primer sequences are available upon request.

4.2. Histology and in situ hybridization

Embryos were obtained from timed matings with E0.5 defined as noon of the day of the vaginal plug. Tissue samples for histological analysis were dissected in cold PBS (phosphate buffered saline) and fixed overnight in either Bouin's fixative or 4% paraformaldehyde (PFA) and processed and stained with Hematoxylin and Eosin following standard protocols.

Samples for whole mount in situ hybridization analysis were dissected in cold PBS and immediately fixed in 4% PFA overnight at 4°C followed by several rinses of PBST (PBS with 0.1% Tween-20), dehydrated to 100% methanol and stored at −20°C. Palates were dissected free of surrounding tissue and processed intact or sectioned into 1-2 mm tissue slices using a microtome blade. A detailed in situ hybridization protocol is available upon request. In order to gain a qualitative assessment of relative expression levels all wildtype and mutant tissues were processed in a single vial and detected for an identical length of time.

The following genes were PCR amplified, T/A cloned (Invitrogen) and used to generate anti-sense riboprobes for the expression analyses: Barx1 (nt 175-1334 of the coding sequence); Etv5 (nt 1591-3245 of the coding sequence); Fgf10 (nt 1-630 of the coding sequence); Fgfr2 (nt 1427-2465 of the coding sequence); Msx1 (nt 1-1213 of the coding sequence); Ptc1 (nt 2903-4262 of the coding sequence); Spry2 (nt 16-1465 of the coding sequence); Shh (full length cDNA clone from A. McMahon); Tbx22 (nt 307-1547 of the coding sequence).

4.3. Cell proliferation and survival assays

Cell proliferation was determined using a BrdU labeling kit (Roche) and cell death was determined by TUNEL analysis using an in situ cell death detection kit (Roche) according to the manufacturer's instructions. For BrdU incorporation pregnant females were injected with 20 μl/gram of body weight with 10 mM BrdU and embryos collected one hour later. Tissues were dissected in cold PBS, fixed overnight in 4% PFA at 4°C and embedded in paraffin. 10 μm serial frontal sections were collected at 6 sections per slide and grouped into four regions each representing one-fourth the total anterior to posterior length of the individual palate (anterior, mid-anterior, mid-posterior, and posterior). Serial sets of slides from each region were processed for BrdU labeling and TUNEL analysis. Nuclei were counterstained with DAPI. Mitotic index and cellularity data were obtained by counting the number of labeled cells and total number of cells per unit area in an identical region of the palatal shelf. Statistical analysis was performed using ANOVA and student's t-test. Embryos for Nile Blue Sulfate (NBS) staining were dissected and rinsed in cold PBS and then incubated in filtered 10 mg/ml of NBS in PBS with 0.1% Tween-20 for 45 minutes at 37°C.

4.4. RNA extraction and Quantitative RT-PCR analysis

E14.5 palatal tissue was dissected from surrounding tissue, snap frozen in 200 μl of RNAlater (Ambion) and stored at −80°C. Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. RNA was DNase I treated (Ambion) and RNA quantity and quality was checked using an Agilent Bioanalyzer 2100. 1 μg of total RNA was primed using random hexamers and reverse transcribed using Superscript II reverse transcriptase (Invitrogen) and diluted 1:10 prior to PCR analysis.

Relative quantification of expression levels for Spry2 in the palate were assayed in E14.5 wildtype, 36Pub mutant, and Spry2-BAC hemizygous line 2 and line 69 36Pub mutant samples. Four samples of each genotype were run in triplicate and β–actin expression was used to normalize relative expression levels. Real-time PCR was performed on an ABI 7500 using the following TaqMan gene expression assays: Spry2 Mm00442344_m1; mouse ACTB 4352341E in a 20 μl reaction solution containing TaqMan univesral mastermix with UNG (Applied Biosystems) and 5 ng of cDNA. Data were analyzed as described (Peirson et al., 2003). Cycling conditions were 50 °C for 2 minutes, 95°C for 10 minutes followed by 95°C for 15 seconds and 60°C for 1 minute for 40 cycles.