Odd-skipped related 2 (Osr2) encodes a key intrinsic regulator of secondary palate growth and morphogenesis

The mammalian secondary palate arises from the medial sides of the maxillary processes flanking the embryonic oral cavity as two bilateral outgrowths that initially grow vertically down the sides of the developing tongue. At a precise developmental stage, the bilateral palatal shelves elevate to a horizontal position above the dorsum of the tongue and fuse with each other at the midline to form the intact secondary palate that separates the nasal cavity from the oral cavity(Ferguson, 1988). Any disturbance of the growth, elevation or fusion of the palatal shelves could result in cleft palate, one of the most common birth defects in humans. Considerable efforts have gone into genetic and epidemiological studies of human orofacial clefting (Murray,2002; Carinci et al.,2003). Several cleft-causing mutations have recently been identified, including mutations in the IRF6, MSX1, PVRL1 and TBX22 genes (van den Boogaard et al., 2000; Suzuki et al.,2000; Braybrook et al.,2001; Sozen et al.,2001; Kondo et al.,2002; Jezewski et al.,2003). In addition, studies of targeted mutations in mice have shown that mutations in a growing number of genes each results in cleft palate in nullizygous mutants (Thyagarajan et al., 2003). Most of these mutant mice also have gross craniofacial developmental defects (e.g. Gendron-Maguire et al., 1993; Rijli et al., 1993; Satokata and Maas, 1994; Martin et al., 1995; Sanford et al., 1997; Peters et al., 1998),indicating that cleft palate is a common secondary effect of other craniofacial abnormalities. Mutations in Msx1 and Tgfb3,which are normally expressed during palate development, have been shown to cause cleft palate as a primary effect, because the gene products normally regulate palatal mesenchyme proliferation and palatal shelf fusion,respectively (Fitzpatrick et al.,1990; Pelton et al.,1990; Satokata and Maas,1994; Kaartinen et al.,1995; Proetzel et al.,1995; Taya et al.,1999; Zhang et al.,2002).

Despite the large number of genes associated with cleft palate formation,strikingly little is known about the molecular processes governing normal palatal growth and patterning. For example, there is little understanding of what controls the initial phases of palatal outgrowth from the medial sides of the maxillary processes, although this occurs at the developmental time when many drugs are administered to experimentally induce cleft palate in animal models (Salomon and Pratt,1979; Shah, 1984; Ferguson, 1988; Diehl and Erickson, 1997). In an attempt to identify genes regulating the initial phases of palate development, we previously carried out an expression screen in mouse embryos and showed that the odd-skipped related 2 (Osr2) gene is specifically activated in the nascent palatal mesenchyme at the onset of palatal outgrowth(Lan et al., 2001). The Osr2 gene encodes a zinc-finger protein with extensive sequence similarity to the Drosophila Odd-skipped family of putative transcription factors (Lan et al.,2001). The odd-skipped gene was initially identified as a pair-rule gene because mutations at this locus cause loss of the odd-numbered segments in the Drosophila embryo(Nusslein-Volhard and Wieschaus,1980; Coulter et al.,1990). Gene expression and phenotypic analyses indicate that the odd-skipped gene product functions to prevent inappropriate expression of other segmentation genes(Coulter and Wieschaus, 1988;DiNardo and O'Farrell, 1987; Mullen and DiNardo, 1995; Saulier-Le Drean et al., 1998). During mouse embryogenesis, Osr2mRNA expression marks the medial maxillary regions where the initial palatal outgrowths occur and persists strongly in the downward growing palatal shelves. Upon palatal shelf elevation and fusion at the midline, Osr2expression in the palate is downregulated(Lan et al., 2001). The sequence homology to the Odd-skipped family of transcription factors and the dynamic expression pattern suggest that Osr2 may play an important role in palate development. Interestingly, recent annotation of the human genome database assigned the orthologous human OSR2 gene to chromosome 8q23,in a chromosomal region strongly associated with non-syndromic orofacial clefting (Prescott et al.,2000). To analyze the function of Osr2 in palate development, we have generated mutant mice carrying a targeted deletion of the Osr2 coding region and found that Osr2 mutants have specific defects in palatal shelf growth and morphogenesis.

A BAC clone containing the Osr2 genomic region was isolated from the RPCI-22 129/SvEvTac mouse BAC library (BACPAC Resources, Children's Hospital of Oakland, Oakland, CA). A targeting vector was made with a 9.4 kb HindIII fragment containing all three coding exons of the Osr2 gene subcloned from the BAC clone(Fig. 1A). The targeting vector contains a 2 kb 5′ homology arm containing the initial 15 codons of the Osr2 open reading frame fused in-frame with a modified bacterial lacZ gene that encodes a nuclearly localized β-galactosidase, a loxP-flanked PGK-neo expression cassette, a 3.3 kb 3′ homology arm and a PGK-DTA expression cassette. Correct targeting of the Osr2 locus with this vector results in the replacement of 2.6 kb of the Osr2-coding region with the lacZ gene and the neo expression cassette, which is expected to produce a β-galactosidase fusion protein containing the N-terminal 15 amino acid residues of the Osr2 protein.

The targeting vector was linearized and electroporated into CJ7 ES cells as previously described (Swiatek and Gridley,1993). G418-resistant ES colonies were screened by Southern hybridization for homologous recombination. Two independently targeted ES cell clones were injected into blastocysts from C57BL/6J mice and the resultant chimeras bred with C57BL/6J females. F1 mice were genotyped by Southern hybridization analysis of tail DNA. Mice and embryos from subsequent generations were genotyped by PCR. PCR with primer 1 (5′-GAT ACG GGT AAG ACA GAA ACT G-3′) and primer 2 (5′-CTA CAA GGA TCT AGC ACA TGC TG-3′) amplified a product of 490 bp from the wild-type Osr2allele. PCR with primer 2 and primer 3 (5′-CTT CTT GAC GAG TTC TTC TGA GG-3′) amplified a mutant allele-specific product of 460 bp. Heterozygous F1 mice were backcrossed with C57BL/6J mice and N2 heterozygous mice were intercrossed for analysis of homozygous phenotype.

X-gal staining of whole-mount embryos and cryostat sections forβ-galactosidase detection was performed as described previously(Hogan et al., 1994). Cryostat sections were counterstained with eosin after X-gal staining. Skeletal preparations were made from newborn mice as described previously(Martin et al., 1995).

For histology, embryos were fixed in Bouin's fixative, dehydrated through graded alcohols, embedded in paraffin wax, sectioned at 7 μm and stained with Hematoxylin and Eosin. For in situ hybridization, embryos were fixed overnight at 4°C in 4% paraformaldehyde in PBS. In situ hybridization of whole-mount embryos and tissue sections were performed as described previously(Lan et al., 2001).

For detection of cell proliferation in the palatal shelves, pregnant Osr2 heterozygous females were injected intraperitoneally on gestational day 12.5 or 13.5 with BrdU (Roche) labeling reagent (45 μg/g body weight). One hour after injection, embryos were dissected, fixed in Carnoy's fixative, dehydrated through graded alcohols, embedded in paraffin wax and sectioned in the coronal plane at 5 μm. Immunodetection of BrdU was performed using the BrdU labeling and detection kit (Roche) according to manufacturer's instructions and the sections were counterstained with Eosin. The total number of mesenchymal cell nuclei as well as the number of BrdU-labeled mesenchymal nuclei in a fixed area of 0.057 mm²beginning at the distal tip and encompassing more than two-thirds of the vertically-oriented palatal shelves were counted using an ocular scale grid. Sections were selected from the middle of the anteroposterior axis of the palatal shelves in comparable positions in the wild-type and mutant embryos and cell counts were recorded for each of the bilateral palatal shelves from five continuous sections of each embryo. The cell proliferation index was calculated as percentage of the cell nuclei with BrdU labeling. Student's t-test was used to analyze the significance of difference and a P value less than 0.01 was considered statistically significant.

Apoptotic cell death in the palatal shelves was assessed by TUNEL labeling of either paraffin wax or cryostat sections using the in situ cell death detection kit (Roche) following the manufacturer's instructions.

To examine the function of Osr2 in palate development, we have generated mice carrying a targeted mutation replacing most of the Osr2-coding region with a modified bacterial lacZ gene through homologous recombination in ES cells(Fig. 1A). The targeted allele, Osr2^tm1Jian, is expected to encode a functionalβ-galactosidase fusion protein containing the N-terminal 15 amino acid residues of the Osr2 protein and to result in loss of Osr2 function. Correctly targeted ES cell clones were used to generate chimeric animals and germline transmission of the targeted mutation was confirmed by Southern hybridization and PCR analysis of tail DNA samples of the F1 and F2 progeny of the chimeric founders (Fig. 1B,C). Embryos heterozygous for the Osr2^tm1Jian allele were isolated and stained with X-gal to reveal domains of β-galactosidase activity. As shown in Fig. 2, β-galactosidase staining was restricted to the mesonephros at E9.5(Fig. 2A). By E10.5,β-galactosidase staining is also detected in the limb buds, in the rostrolateral regions of the mandibular processes, in the mesenchyme posterior to the optic placodes, and in the nascent palatal mesenchyme at the medial sides of the maxillary processes (Fig. 2B,C). At E13.5, β-galactosidase activity was detected throughout the downward growing palatal mesenchyme, with lateral sides of the palatal mesenchyme showing higher β-galactosidase activity than the medial sides (Fig. 2D), which is similar to the Osr2 mRNA expression pattern reported previously(Lan et al., 2001). At E14.75,the palatal shelves have elevated and β-galactosidase expression is downregulated in the palatal mesenchyme, while it is strongly expressed in the tooth bud mesenchyme, the olfactory mesenchyme, the periocular mesenchyme and the eyelids (Fig. 2E). These data indicate that expression of the β-galactosidase reporter fusion protein recapitulates the endogenous Osr2 expression pattern reported previously (Lan et al., 2001),further confirming correct integration of the lacZ gene into the Osr2 locus.

Mice heterozygous for the Osr2^tm1Jian allele are normal and fertile. Heterozygous animals were intercrossed and genotypes of their progeny were determined 2 weeks after birth. No mice homozygous for the mutation were found. Careful examination of staged embryos and newborn mice from heterozygous intercrosses revealed that the Osr2^{tm1Jian/tm1Jian} homozygous mutants could complete embryogenesis but died within 24 hours after birth. All homozygous neonates exhibited open eyelids and bilateral cleft of the secondary palate(Fig. 3B,D), which correlate with Osr2 mRNA expression during normal eyelid and palate development(Lan et al., 2001). As Osr2 also exhibits a dynamic expression pattern during kidney and limb development (Lan et al.,2001), we carefully examined development of these structures by histological and skeletal preparations, but did not find any abnormalities in them (data not shown). In addition to open eyelids and cleft palate, another abnormality observed in the homozygous mutants is that they have thickened tympanic rings (Fig. 3F), which correlates with Osr2 expression in the mesenchymal cells that give rise to the tympanic rings proximal to the first branchial clefts(Fig. 2B). Other craniofacial structures, including the middle ear ossicles that develop from proximal first arch mesenchyme adjacent to the tympanic rings, are unaffected by the Osr2^tm1Jian mutation(Fig. 3G,H).

To investigate which palatal developmental processes require Osr2gene function, we carried out histological analyses of embryos throughout palate development from E12 to birth. As shown in Fig. 4, palatal outgrowth and initial downward palatal growth were normal in the homozygous mutants(Fig. 4A,B). Between E13.5 and E14.5, the vertically oriented palatal shelves undergo rapid growth and initiate reorientation to a horizontal position above the dorsum of the tongue in the wild-type and heterozygous embryos(Fig. 4C,E). By contrast, the Osr2^{tm1Jian/tm1Jian} homozygous mutant littermates exhibit retarded palatal shelves that remain vertically oriented at this stage(Fig. 4D,F). Palatal shelf retardation is observed throughout the anteroposterior axis of the palatal shelves in all homozygous mutants examined at E14.5 (data not shown). By E15.5, the elevated palatal shelves have initiated fusion at the midline in wild-type and heterozygous embryos, whereas the homozygous mutant palatal shelves are elevated but failed to contact each other at the midline(Fig. 4G,H). As the palatal fusion process continues to generate intact secondary palates in wild-type and heterozygous embryos (Fig. 4I),the homozygous mutant palatal shelves are further separated from each other(Fig. 4J), resulting in bilateral cleft of the secondary palate at birth(Fig. 3D).

To investigate whether palatal shelf retardation in the Osr2^{tm1Jian/tm1Jian} mutant embryos was due to impaired cell proliferation, we analyzed BrdU incorporation in E12.5 and E13.5 embryos. No difference in the cell proliferation index between wild-type and mutant palatal shelves was observed in E12.5 embryos (data not shown). However,whereas the wild-type and heterozygous embryos showed no significant differences in palatal cell proliferation at E13.5, the homozygous mutants exhibited a 26% reduction (P<0.01) in cell proliferation index in the palatal mesenchyme (Fig. 5C,D). The wild-type and heterozygous palatal shelves exhibit faster growth in the medial halves than the lateral halves(Fig. 5A,B,D), which is consistent with previous reports and results in the characteristic erectile shape of the palatal shelves before elevation(Ferguson, 1988). Remarkably,when the percentage of BrdU-labeled cells was recorded separately for the medial and lateral halves of the palatal shelves, there was a 37% reduction(P<0.01) in the cell proliferation index in the medial halves of homozygous mutant palatal shelves, while the cell proliferation index was not significantly different (P>0.05) in the lateral halves of the palatal shelves in wild-type, heterozygous and homozygous mutant embryos(Fig. 5D). This preferential reduction in cell proliferation in the medial halves of the palatal shelves results in the appearance of retarded, mediolaterally symmetric palatal shelves in the homozygous mutants at E14.5(Fig. 4D). We also examined whether abnormal cell death contributed to the palatal shelf retardation by in situ TUNEL assays, but did not find any differences in cell death between wild-type and homozygous mutant palatal shelves (data not shown).

As Osr2 mRNA expression is activated at the onset of palatal outgrowth and persists throughout the downward growing palatal shelves during mouse embryogenesis (Lan et al.,2001), the normal initiation of palatal outgrowth and preferential reduction in cell proliferation in the medial halves of the palatal shelves in the Osr2^{tm1Jian/tm1Jian} mutants are surprising observations. One possible explanation for the mutant phenotype is that loss of Osr2 function in the initial stage of palatal outgrowth is compensated by expression of a related gene product. The Osr1 gene encodes a protein with 65% overall amino acid sequence identity and 98%sequence identity in the zinc-finger domain with the Osr2 protein(So and Danielian, 1999; Lan et al., 2001). No other odd-skipped related gene has been found in the mouse or human genomes. Thus, we examined Osr1 mRNA expression and compared that with the expression pattern of Osr2 mRNA during palate development. We also compared the expression patterns of Osr1 and Osr2with that of Pax9, which encodes a paired-class homeodomain-containing transcription factor required for proper palatal patterning (Peters et al.,1998). From E12.5 to E13.5, Osr2 mRNA is expressed abundantly throughout the palatal mesenchyme, with lateral regions expressing higher levels than the medial regions (Fig. 6A-C). Osr1 mRNA is strongly expressed in the developing tongue and at the maxillary-mandibular junction but is only weakly expressed in the palatal mesenchyme at E12.5 and E13.0(Fig. 6D,E). By E13.5, Osr1 mRNA expression is significantly up-regulated in the lateral halves and is completely downregulated in the medial halves of the palatal shelves, forming a sharp boundary in the middle of the palatal shelves(Fig. 6F). Pax9 mRNA is expressed in a pattern similar to that of Osr2 mRNA in the palatal mesenchyme at E12.5 and E13.0 (Fig. 6G,H). However, at E13.5, Pax9 mRNA expression is upregulated in the mediodistal regions and downregulated in the lateral regions of the palatal shelves (Fig. 6I). These data indicate that dynamic molecular changes occur between E13.0 and E13.5 during palate development. Interestingly, the defect in palate development in the Osr2^{tm1Jian/tm1Jian} mutants is first detectable at this stage and the specific reduction in cell proliferation in the mutant medial palatal mesenchyme correlates with the spatiotemporal downregulation of Osr1 mRNA expression in those cells.

To investigate the role of Osr2 in palate development further, we examined the expression patterns of several genes implicated in regulating palatal growth and/or patterning. Msx1, a homeobox gene required for anterior palate mesenchyme proliferation, is expressed weakly in the anterior but not posterior regions of the palatal mesenchyme and strongly in maxillary and mandibular mesenchyme at E14.5 (Zhang et al., 2002). No differences in Msx1 expression were found between wild-type and Osr2^{tm1Jian/tm1Jian} mutants,although the palatal shelves are retarded in the anterior region in the homozygous mutants (Fig. 7A,B). Bmp4 has recently been implicated in regulating palate mesenchyme growth because a Bmp4 transgene driven by the Msx1 promoter was able to rescue palate growth defects in the Msx1^-/-mutant mice (Zhang et al.,2002). However, Bmp4 expression appears unaltered in the Osr2^{tm1Jian/tm1Jian} mutant palatal shelves compared with the wild-type embryos (Fig. 7C,D). Expression of Tbx22, the mouse ortholog of the human X-linked cleft palate gene (Braybrook et al., 2002; Bush et al.,2002; Herr et al.,2003), is also unaltered in the mutant palatal shelves(Fig. 7E,F). Pax9,encoding a paired-class transcription factor required for proper palate patterning (Peters et al.,1998), exhibits a dynamic expression pattern during palate development (Fig. 6G-I). In the Osr2^{tm1Jian/tm1Jian} mutants, Pax9 expression is similar to that in wild-type embryos during early palatal outgrowth (data not shown). However, by E13.5 when the wild-type palatal shelves showed a strong mediolateral gradient of Pax9 mRNA expression, the Osr2^{tm1Jian/tm1Jian} mutants exhibited a uniform lower level of Pax9 mRNA expression in the palatal mesenchyme(Fig. 8A,B). By E14.5, when strong Pax9 expression is observed in the palatal shelves in wild-type littermates (Fig. 8C), significantly reduced levels of Pax9 mRNA are found throughout the retarded palatal shelves in Osr2^{tm1Jian/tm1Jian} mutants(Fig. 8D). The alteration in Pax9 mRNA expression in the Osr2^{tm1Jian/tm1Jian}mutants is specific to the palatal mesenchyme, as Pax9 mRNA levels in the tooth bud mesenchyme and in other craniofacial regions are similar in wild-type and mutant embryos (Fig. 8A-D). Tgfb3, which encodes a growth factor required for normal palatal shelf fusion (Kaartinen et al., 1995; Proetzel et al.,1995), is expressed in the medial and distal epithelia of the vertically oriented palatal shelves at E14.5 in wild-type embryos(Fig. 8E). In the homozygous Osr2^{tm1Jian/tm1Jian} mutants, Tgfb3 expression in the distal palatal epithelia is lost (Fig. 8F). Complementary to the Tgfb3 expression pattern in the epithelia, Osr1 is expressed in the mesenchyme of the proximolateral one-third of the palatal shelves in wild-type embryos at E14.5(Fig. 8G). In the homozygous mutant palatal shelves, the Osr1 expression domain extends to the distal tip of the palatal mesenchyme (Fig. 8H), most probably as a result of the severe reduction in medial palatal mesenchyme proliferation. By E15.0, the palatal shelves have elevated to the horizontal position above the tongue and the Osr1 expression domain is still restricted in the same regions of the palatal mesenchyme in the wild-type and heterozygous embryos(Fig. 8I). By contrast, the homozygous mutant littermates exhibit a low level of Osr1 expression throughout the retarded, still vertically oriented palatal shelves(Fig. 8J). These data indicate,in addition to the role in palatal shelf growth, that Osr2 function is required for the normal mediolateral patterning and elevation of the palatal shelves.

Cleft palate is among the most common birth defects in humans and has been observed in a number of mice carrying mutations in genes encoding transcription factors (Gendron-Maguire et al., 1993; Rijli et al.,1993; Satokata and Maas,1994; Martin et al.,1995; Qiu et al.,1997; De Felice et al.,1998; Peters et al.,1998; Zhao et al.,1999), growth factors and their receptors(Kaartinen et al., 1995; Proetzel et al., 1995; Sanford et al., 1997; Miettinen et al., 1999), other signaling molecules (Matzuk et al.,1995; Jiang et al.,1998), and extracellular matrix molecules(Pace et al., 1997; Lavrin et al., 2001). Whereas the cleft palate phenotypes at birth may appear similar, the underlying mechanisms in cleft pathogenesis are very different. Palate development occurs at the time of significant growth and morphogenesis of the entire craniofacial complex and is dependent on normal development of other craniofacial structures (Ferguson, 1988). The cleft palate phenotype observed in many mutant mice, such as those with mutations in Dlx1, Dlx2, Hoxa2, Mhox (Prrx1 – Mouse Genome Informatics) and δEF1 (Zfhx1a – Mouse Genome Informatics) genes (Gendron-Maguire et al., 1993; Rijli et al.,1993; Martin et al.,1995; Qiu et al.,1997; Takagi et al.,1998), is accompanied by gross craniofacial and skeletal malformations and can be attributed, at least partly, to secondary effects of those malformations. Several mutations cause mechanical hindrance to palatal shelf elevation. For example, the palatal shelves fuse aberrantly to the lateral sides of the tongue in the Jag2^-/- mutant mice and the tongue fails to descend during palatal shelf elevation in the Foxf2^-/- mutant mice, causing cleft palate(Jiang et al., 1998; Wang et al., 2003). Another class of mutations specifically interferes with palatal shelf fusion. For example, Titf2 and Tgfb3 are normally expressed in palatal epithelium and mutations in these genes specifically disrupt palatal shelf fusion (Kaartinen et al.,1995; Proetzel et al.,1995; De Felice et al.,1998).

The cleft palate phenotype in the Osr2^{tm1Jian/tm1Jian}mutant mice is clearly different from those of all previously reported mutant mice. All Osr2^{tm1Jian/tm1Jian} homozygous mutants examined after E15.5 exhibited cleft palate. In addition to cleft palate, Osr2^{tm1Jian/tm1Jian} mutants exhibited open eyelids and thickened tympanic rings, which correlated with Osr2 mRNA expression during normal development of these structures. Open eyelids have been observed in several mouse mutant strains without cleft palate(Keeler, 1935; Bennett and Gresham, 1956;Jeriloff et al., 2000), indicating that it does not cause cleft palate. The thickened tympanic ring phenotype is unique to the Osr2^{tm1Jian/tm1Jian} mutants, but it most probably reflects a developmental role for Osr2 that is independent from that in palate development as other craniofacial structures immediately adjacent to the tympanic rings are unaffected by the mutation. Moreover, Osr2 mRNA is normally expressed during palatal outgrowth and Osr2^{tm1Jian/tm1Jian} mutants exhibited specific impairment of palatal mesenchymal cell proliferation, which resulted in retarded,mediolaterally symmetric palatal shelves before palatal shelf elevation. Thus,whereas Osr2 is required for the normal development of several craniofacial structures, the cleft palate phenotype of the Osr2^{tm1Jian/tm1Jian} mutants resulted from a primary defect in palatal shelf growth.

Impairment of palatal mesenchyme cell proliferation has been reported previously in the Msx1^-/- mutant mice and recently in mice with neural crest-specific inactivation of the Tgfbr2 gene(Zhang et al., 2002; Ito et al., 2003). Msx1 is strongly expressed in the maxillary processes and tooth buds,but only very weakly expressed in the anterior region of the palatal shelves(Zhang et al., 2002). Reduction in cell proliferation was found in the anterior but not in the posterior regions of the palatal shelves in Msx1^-/-mutants (Zhang et al., 2002). In mice with neural crest-specific inactivation of the Tgfbr2 gene,palatal shelves grow normally up to E13.5 but palatal mesenchyme growth is reduced after palatal shelf elevation at E14.5(Ito et al., 2003). These mutant mice also exhibited gross craniofacial skeletal growth defects(Ito et al., 2003), consistent with expression of Tgfbr2 mRNA in migrating cranial neural crest cells (Wang et al., 1995). By contrast, Osr2 mRNA expression is specifically activated in the nascent palatal mesenchyme at the onset of palatal outgrowth and persists strongly in the palatal mesenchyme during the downward palatal growth(Lan et al., 2001). Palatal growth retardation is observed throughout the anteroposterior axis of the palatal shelves before palatal shelf elevation in the Osr2^{tm1Jian/tm1Jian} mutants. Therefore, Osr2expression is a specific marker of the early palatal mesenchyme and is essential for the rapid downward growth of the palatal shelves before elevation.

In the Msx1^-/- mutant mice, impairment of mesenchyme growth in the anterior palate was shown to be due to lack of Bmp4expression and transgenic expression of Bmp4 under the control of the Msx1 promoter was able to restore anterior palatal mesenchyme proliferation in the Msx1^-/- mutant mice(Zhang et al., 2002). We found that Msx1 and Bmp4 are expressed normally in the developing palatal shelves in Osr2^{tm1Jian/tm1Jian} mutant mice(Fig. 7). Furthermore, we observed no differences in Shh and Bmp2 expression, which are believed to function downstream of the Msx1-BMP4 pathway in the palatal shelves (Zhang et al., 2002),between wild-type and Osr2^{tm1Jian/tm1Jian} mutant palatal shelves (data not shown). The differences in phenotype and molecular marker expression patterns in Msx1^-/- and Osr2^{tm1Jian/tm1Jian} mutants indicate that Osr2 and Msx1 function in distinct molecular pathways to regulate palate development. The fact that impairment of palatal mesenchyme proliferation occurs earlier in the Osr2^{tm1Jian/tm1Jian} mutants than in mice lacking Tgfbr2 in the neural crest derivatives suggests that Osr2 function in the palatal mesenchyme is also independent of Tgfβ signaling. The identification of Osr2 as an essential regulator of palatal mesenchyme growth warrants future investigation of how Osr2 expression is activated in the nascent palatal mesenchyme and how it controls palatal mesenchyme proliferation. Interestingly, the human OSR2 gene is located at chromosome 8q23, in a region with strong association to non-syndromic orofacial clefting (Prescott et al.,2000), suggesting that OSR2 is a new candidate gene for human cleft palate formation.

Whereas the Osr2 gene is activated at the onset of palatal outgrowth and is expressed throughout the downward growing palatal mesenchyme,the Osr1 gene exhibits a unique, mediolaterally differentially regulated pattern of expression during palate development. During early palatal outgrowth, Osr1 mRNA is expressed weakly throughout the palatal mesenchyme. By E13.5 of mouse development, Osr1 mRNA expression is strongly upregulated in the lateral halves and completely downregulated in the medial halves of the palatal shelves(Fig. 6D-F). As the mediolateral pattern is laid down, Pax9 mRNA expression is upregulated in the medial palatal mesenchyme(Fig. 6I). In the Osr2^{tm1Jian/tm1Jian} mutants, early Osr1 and Pax9 expression patterns are unaltered during palatal outgrowth (data not shown). However, by E13.5, the upregulation of Pax9 mRNA expression seen in the medial regions of the wild-type palatal shelves is not observed in the mutants (Fig. 8B), indicating a molecular patterning defect in the mutant palatal mesenchyme. At the same developmental stage, a significant reduction in BrdU incorporation is observed specifically in the medial palatal mesenchyme in the Osr2^{tm1Jian/tm1Jian} mutants(Fig. 5). Although mice deficient in the Pax9 gene have cleft palate(Peters et al., 1998), it is not known whether Pax9 directly regulates palatal mesenchyme proliferation. Thus, it remains to be investigated whether the reduction in Pax9mRNA expression in the medial palatal mesenchyme is the cause of the region-specific reduction in palatal cell proliferation in the Osr2^{tm1Jian/tm1Jian} mutants. Nevertheless, these data indicate that Osr2 plays an important role in the mediolateral patterning of the palatal mesenchyme. The differential regulation of Osr1 mRNA expression in the medial and lateral halves of the palatal shelves suggests that Osr1 may also have a role in mediolateral palatal patterning.

In addition to alterations in gene expression in the palatal mesenchyme, we observed alteration of Tgfb3 mRNA expression in the palatal epithelium at E14.5 in the Osr2^{tm1Jian/tm1Jian} mutants. Interestingly, the epithelial Tgfb3 expression domain complements that of Osr1 mRNA expression in the palatal mesenchyme in both wild-type and the Osr2^{tm1Jian/tm1Jian} mutant embryos(Fig. 8E-H). Because the Osr1 and Tgfb3 mRNAs are expressed normally up to E13.5 in the Osr2^{tm1Jian/tm1Jian} mutants (data not shown), the alterations in their expression in the palatal shelves at E14.5 are most probably secondary effects of the reduction in medial palatal mesenchyme proliferation. However, the corresponding shift in epithelial Tgfb3and mesenchymal Osr1 expression domains suggest that Osr1expression in the palatal mesenchyme may regulate mesenchymal-epithelial interactions that in turn regulate Tgfb3 expression in the palatal epithelium. Further characterization of the molecular pathways through which Osr2 and Osr1 regulate palate development will provide novel insights into the molecular mechanisms governing palate growth and patterning.

We thank Bob Angerer and Lin Gan for comments on the manuscript; Tom Gridley for the CJ7 ES cell line; YiPing Chen for cDNA probes; and the University of Rochester Transgenic Mouse Facility for generation of chimeric mice from targeted ES cells. This work was supported by a NIH/NIDCR grant(DE13681) to R.J.