Requirement for Dlgh-1 in Planar Cell Polarity and Skeletogenesis during Vertebrate Development

The development of specialized organs is tightly linked to the regulation of cell growth, orientation, migration and adhesion during embryogenesis. In addition, the directed movements of cells and their orientation within the plane of a tissue, termed planar cell polarity (PCP), appear to be crucial for the proper formation of the body plan. In Drosophila embryogenesis, Discs large (dlg) plays a critical role in apical-basal cell polarity, cell adhesion and cell proliferation. Craniofacial defects in mice carrying an insertional mutation in Dlgh-1 suggest that Dlgh-1 is required for vertebrate development. To determine what roles Dlgh-1 plays in vertebrate development, we generated mice carrying a null mutation in Dlgh-1. We found that deletion of Dlgh-1 caused open eyelids, open neural tube, and misorientation of cochlear hair cell stereociliary bundles, indicative of defects in planar cell polarity (PCP). Deletion of Dlgh-1 also caused skeletal defects throughout the embryo. These findings identify novel roles for Dlgh-1 in vertebrates that differ from its well-characterized roles in invertebrates and suggest that the Dlgh-1 null mouse may be a useful animal model to study certain human congenital birth defects.


Introduction
The development of specialized organs in vertebrates is tightly linked to the regulation of cell growth, apical-basal cell polarity and cell-cell adhesion during embryogenesis. In addition, the directed movements of cells and their orientation in the same direction within the plane of a tissue, termed planar cell polarity (PCP), appear to be crucial for the proper formation of the body plan. Of great interest has been to decipher the mechanisms involved in regulating these critical aspects of development. From studies in invertebrates it is known that certain PSD95/Dlg/ZO-1 (PDZ) domain containing proteins such as Discs-large (Dlg) play prominent roles in regulating apical-basal polarity [1,2] while different PDZ proteins are part of a genetic network that regulates PCP [3]. Recent work in vertebrates supports the contention that the function of many polarity factors has been conserved crossspecies. However, studies also suggest that certain factors have different or additional roles in vertebrates as compared to invertebrates [4]. In this study, we address the role of Dlgh-1, the mouse homolog of Drosophila dlg, in mouse development.
In Drosophila melanogaster, mutations in dlg and scrib, the gene encoding a second PDZ protein, Scribble, result in neoplastic overgrowth, multilayering and loss of cell shape in various epithelial tissues including the imaginal discs [1], the embryonic epidermis, and the follicular epithelia [2], suggesting a role for these factors in cell proliferation, adhesion and apico-basal polarity. In Drosophila and Caenorhabditis elegans, Dlg and Scrib are thought to be required for properly localizing and maintaining adherens junctions and apical determinants [1,2,5,6,7,8,9]. In Drosophila, Dlg localizes to the septate junctions where it forms a complex with Scrib and Lethal giant larvae (Lgl) [1,2]. This Dlg/ Scrib/Lgl complex antagonizes other PDZ complexes, thereby restricting adherens junctions and apical determinants, such as Crumbs, to the appropriate compartment [10]. Finally, the Partner of Inscuteable/Discs-large complex is required for planar polarity during the asymmetric division of the sensory organ precursor (pl) cell [11].
The role of Dlgh-1 in mouse development has begun to be explored. Caruana and Bernstein [34] reported that mice carrying a gene trap insertion in Dlgh-1 (Dlgh-1 g/gtt mice), which results in a Dlgh-1 protein lacking the SH3, band 4.1, and GUK domains, exhibited a cleft palate and a shorter mandible than Dlgh-1 wild type mice. Additional studies on this Dlgh-1 gt/gt mouse showed that these mice also exhibited a decrease in the number of nephrons in the developing kidney [35] and cell cycle misregulation in the epithelium of the ocular lens [36]. Since the L27 and three PDZ domains of Dlgh-1 were left intact in the gene trap allele [34,37], it is possible that the fusion protein generated from this gene trap allele may retain some or acquire new activities of Dlgh-1 and, therefore, the phenotypes observed may not fully reflect the null phenotype. Consistent with this possibility, lacZ staining of Dlgh-1 gt/gt lenses and control Dlgh-1 +/+ lenses showed similar staining patterns (Rivera and Griep, unpublished observations). Furthermore, two recent reports show that deletion of Dlgh-1 in mice caused abnormalities in the ureteric smooth muscle [38] and absence of vagina and seminal vesicles [39]. Additionally, conditional deletion of Dlgh-1 in the lens led to more severe phenotype than was observed in the lenses of the Dlgh-1 gt/gt mice including epithelial and fiber cell specific defects in cell adhesion and polarity [40].
In the present study, we generated mice carrying a germline null mutation in Dlgh-1 by deleting one of the exons encoding the first PDZ domain and characterized the wide-ranging effects of this mutation on mouse embryogenesis. Loss of Dlgh-1 not only recapitulated the craniofacial and kidney defects observed in the Dlgh-1 gt/gt mouse but also resulted in reduced ossification in the skull, maxilla, and middle ear, and shortening of the long bones. Furthermore, deletion of Dlgh-1 resulted in eyelid closure and neural tube closure defects, and the misorientation of the stereociliary bundles of the cochlear hair cells, all of which suggest defects in PCP. Thus, our analysis identifies novel roles for Dlgh-1 in mouse development, places Dlgh-1 in the group of factors that regulate PCP in the mouse, and provides new insight into the distinct in vivo requirements for this gene in vertebrates, as compared to invertebrates.

Generation of Dlgh-1 Null Mice
To generate Dlgh-1 null mice, mice carrying a mutant allele of Dlgh-1, in which exon 8 and a neomycin resistance-cassette are flanked by loxP sites [40] (Fig. 1A), were mated to EIIA-Cre transgenic mice [41]. The Dlgh-1 null allele was moved onto the FVB/NJ and C57BL/6J genetic backgrounds by backcrossing to either FVB/NJ or C57BL/6J mice for at least three generations before use. Mice heterozygous for the null allele were intercrossed to generate Dlgh-1 2/2 mice and Southern blot analysis was used to confirm the genotype of the progeny (Fig. 1B).

Effect of Loss of Dlgh-1 on Mouse Embryogenesis
It has been reported that mice homozygous for the Dlgh-1 gt allele die at or near the time of birth [12]. To determine if nullizygosity for Dlgh-1 also would be lethal, we intercrossed Dlgh-1 +/2 mice and examined litters for live Dlgh-1 2/2 pups. Of the 47 pups genotyped, 10 were Dlgh-1 +/+ (21%), 22 (47%) were Dlgh-1 +/2 , and 15 (32%) were Dlgh-1 2/2 . Of the 15 Dlgh-1 2/2 pups, all of which exhibited severe craniofacial abnormalities, 13 were found dead on the day of birth and the remaining two died shortly after being found. These data indicate that loss of Dlgh-1 confers perinatal lethality with 100% penetrance and support the hypothesis that Dlgh-1 plays an important role in mouse development.
All of these phenotypes of the Dlgh-1 2/2 mice were observed with 100% penetrance on both the FVB/NJ and C57BL/6J genetic backgrounds and were similar to those observed in the Dlgh-1 gt/gt mice [34]. Additionally, the kidneys of the Dlgh-1 2/2 mice were smaller in size and exhibited a reduced tubule to mesenchyme ratio (data not shown), consistent with a defect in branching morphogenesis, which was reported for the previously described Dlgh-1 mutant mice [35,38]. Finally, defects in the ocular lens were also noted in the Dlgh-1 2/2 mice (data not shown) as reported previously for the Dlgh-1 gt/gt mice [36] and mice in which Dlgh-1 was specifically deleted in the lens [40]. Together, these data show that Dlgh-1 is required for mouse survival, likely in part through its contribution to palate formation, and for multiple aspects of mouse development.

Effect of Loss of Dlgh-1 on Eyelid and Neural Tube Closure
The cleft palate and reduced size of the Dlgh-1 2/2 mice (Fig. 2) were observed on both the FVB/NJ or C57BL/6J backgrounds. However, some defects were noted exclusively on the C57BL/6J background. First, eyelid closure defects were apparent in 100% of C57BL/6J Dlgh-1 2/2 mice at E18.5 (Fig. 3). The severity of this phenotype varied. In some cases the epidermis had extended and fused; however, the dermis had not filled in (Fig. 3E). In other cases, the epidermis had failed to extend, leading to a complete failure of eyelid closure (Fig. 3F). To determine if the effect of Dlgh-1 deficiency on eyelid closure may be a direct effect, head sections from Dlgh-1 +/+ and Dlgh-1 2/2 mice were subjected to double immunofluorescence experiments using anti-Dlgh-1 and anti-Vangl2 antibodies (Fig. 4). Staining for Dlgh-1 was observed in the Dlgh-1 +/+ eyelid epithelium (Fig. 4A, B) but was absent in the Dlgh-1 2/2 eyelid (Fig. 4D, E), suggesting that the open eyelid defect may be a direct result of loss of Dlgh-1 function.
Interestingly, the failure of eyelid closure is also noted in the circletail (Crc/Crc) mouse, in which a defective Scrib protein lacking the C-terminal two PDZ domains is expressed [42], and in the looptail (Vangl2 Lp/Lp ) mouse, in which a defective Vangl2 protein carrying an amino acid substitution mutation (S464N) is expressed [43,44]. The Vangl2 protein is normally expressed in the eyelid epithelium [43,44,45] and is primarily localized to the plasma membrane [45]. However, the mutant Vangl2 protein in Vangl2 Lp/ Lp mice fails to localize to the membrane; rather it is cytoplasmic [45]. Immunofluorescent staining of eyelids from Dlgh-1 +/+ mice showed that Vangl2 was tightly localized to the membranes in the eyelid epithelium (arrow, Fig. 4A, C) where there was overlap with Dlgh-1 staining. In contrast, Vangl2 was cytoplasmic in the eyelid of the Dlgh-1 2/2 mice (Fig. 4D, F). Thus, the correct subcellular localization of Vangl2 in the eyelid epithelium depends on Dlgh-1.
Vangl2 Lp/Lp and Crc/Crc mice also exhibit craniorachischisis [42,44,46], a severe neural tube closure defect in which the neural tube is open from the midbrain-hindbrain boundary to the tail. This same neural tube defect was observed in C57BL/6J Dlgh-1 2/ 2 mice at both E16.5 (Figs. 5C, D) and E18.5 (not shown). Unlike the open eyelid phenotype, however, craniorachischisis occurred with low penetrance; only 6.7% (2 out of 30) of the C57BL/6J Dlgh-1 null mice displayed this defect. Dlgh-1 2/2 mice that exhibited craniorachischisis also exhibited gastroschisis, an exter- WT, wild-type allele; C, construct, plasmid containing loxP sites flanking exon 8 of Dlgh-1 and neo cassette, and the TK cassette; F, the floxed allele allele, containing lox P sites flanking Dlgh-19s exon 8 and the neo cassette; R, recombined allele, desired recombination event lacking exon 8 and neo. Abbreviations: 7, exon 7; 8, exon 8; 9, exon 9; P, PstI; triangles, lox P sequences; neo, neomycin; TK, thymidine kinase. 39 probe for Southern blot analysis is denoted by black bars. Fragments expected from PstI digestion and 39 probe hybridization are depicted by double-arrowed lines. (B) Southern blot analysis of PstI digested genomic DNA isolated from mice that were WT (+/+), heterozygote (+/2), or null (2/2) for the Dlgh-1 allele. Sizes of hybridizing bands are shown in kb. (C) Western blot analysis of Dlgh-1 protein levels. Protein lysates of brain, heart, lung, kidney, skin, eye, liver, and mouse embryo fibroblasts (MEF) from Dlgh-1 wild type (+/+), heterozygote (+/2), or null (2/2) mice were resolved by SDS-PAGE and immunoblotted with an anti-Dlgh-1 antibody, as described in Materials and Methods. Membranes were reprobed with an anti-b-actin antibody as a loading control. doi:10.1371/journal.pone.0054410.g001 nalization of the gut (Fig. 5A, C). Closure of the eyelid and neural tube involves a process of polarized cell movement known as convergent extension [3,47]. Thus, the defects in eyelid and neural tube closure exhibited in the Dlgh-1 2/2 mice suggest a new role for Dlgh-1 in the mouse in the regulation of convergent extension. Recent studies in vertebrates have shown that convergent extension is regulated, at least in part, by a group of proteins that are involved in planar cell polarity (PCP) in Drosophila [3]. PCP refers to the polarization of cells within the plane of a cell sheet. In Drosophila, PCP regulates wing hair, body bristle, and eye ommitidial cluster orientation. In mammals, the uniform orientation of stereociliary bundles of the hair cells within the cochlea represents a distinctive example of PCP. In Vangl2 Lp/Lp mice, the stereociliary bundles in the inner (IHC) and outer (OHC) hair cell layers are misoriented [4]. Similarly, in Crc/Crc mice the stereociliay hair bundles in the second and third row of OHCs are misoriented [4], Together, these data identify a role for Vangl2 and Scrib in the regulation of the PCP pathway in vertebrates.
Because dlg and scrib are known to interact in Drosophila [2] and because the open eyelid, neural tube, and gut phenotypes of the Dlgh-1 2/2 embryos (Fig. 4) resemble that of Crc/Crc mutant mice [42], we asked if stereociliary bundle orientation was affected in the Dlgh-1 null mice. Cochlear sensory epithelia (organs of Corti) from E18.5 Dlgh-1 +/+ and Dlgh-1 2/2 embryos were stained with phalloidin to label hair cell sterocilia, a structure rich in filamentous actin, and the specimens analyzed by confocal microscopy. Normally, the stereociliary bundles are arranged uniformly around the primary cilia and oriented toward the outer edge of the organ. The IHC bundles are arranged in a curved shape while the OHC bundles are arranged in a v-shape [48]. The stereociliary bundles in the IHC row of the Dlgh-1 2/2 mice were uniformly polarized, as in the Dlgh-1 +/+ mice (Fig. 6A, B). The stereociliary bundles in all three rows of OHCs of the Dlgh-1 +/+ mice were uniformly oriented (Fig. 6A). However, many stereociliary bundles in the third row of OHCs in the Dlgh-1 2/2 cochleae were misoriented (Fig. 6B, arrows). Occasionally, stereociliary bundles in the second OHC layer were also misoriented (Fig. 6B). The angle of orientation of individual  stereociliary bundles in the OHC3 layer relative to the neuralabneural axis in Dlgh-1 +/+ and Dlgh-1 2/2 cochlea were measured and the degrees of deviation from this axis calculated (Fig. 6C). The average degrees of deviation for bundles in the Dlgh-1 2/2 mice (33.7+/21.87 degrees) was significantly higher (p = 1.5610 29 ) than in the Dlgh-1 +/+ mice (17.3+/20.87 degrees). To determine if the effect of Dlgh-1 deficiency on the cochlea might be a direct effect, head sections from day E15.5 Dlgh-1 +/+ embryos were immunostained with an anti-Dlg-1 antibody (Fig. 6D). Immunoreactivity was observed in the epithelial cells of the duct, which give rise to the sensory cells, and cartilage within the cochlea, indicating the presence of Dlgh-1 protein in these structures. To further determine if the effect of Dlgh-1 deficiency on hair cell orientation might be a direct effect, cochlear explants from P2 control mice were subjected to immunofluorescence for Dlgh-1. Immunoreactivity was observed in the basal-lateral membrane domain of the outer hair cells (Fig. 6F, arrows).
The skeletal defects in the Dlgh-1 2/2 mice on the FVB/NJ background extended to a reduction in the ossification of the ribs, as well as the second, third, and fourth sternebrae (S) (Fig. 7E). The long bones of the fore-and hindlimbs were also shortened by approximately 20% and 18%, respectively, in the Dlgh-1 2/2 mice when compared to their Dlgh-1 +/+ counterparts (Figs. 7F, G). Calcification in the phalanges, while apparent in the limbs of the Dlgh-1 +/+ mice, was absent in the limbs of the Dlgh-1 2/2 mice (Figs. 6F, G, arrow). Taken together, these results suggest that Dlgh-1 is required for proper skeletogenesis of craniofacial, axial and long bones. Furthermore, as the calvarium forms through intramembranous bone formation whereas the other craniofacial bones and the axial bones form through endochondral ossification [49], these data show that both forms of bone development are affected when Dlgh-1 is ablated in the mouse.

Discussion
The MAGUK protein Dlgh-1, which is highly conserved crossspecies, has emerged as an important factor in the regulation of cell-cell adhesion, apical-basal polarity, and cell proliferation in Drosophila [1,2]. The regulation of these cellular aspects is critical for the establishment of specialized organs during embryogenesis. In the current study, we assess the role of Dlgh-1 in mouse embryogenesis (see Table 1 for a summary of phenotypes). We provide evidence that loss of Dlgh-1 affects PCP during mouse embryogenesis. We further provide evidence that loss of Dlgh-1 causes defects in skeletal structures arising through both endochondral and intramembranous ossification. These findings demonstrate newly discovered roles for Dlgh-1 in vertebrate development and suggest that the Dlgh-1 2/2 mouse may serve as an animal model to study major congenital birth defects in humans such as defects in neural tube closure and skeletal malformations.

Role of Dlgh-1 in Mouse Organogenesis
The results presented in this study suggest that Dlgh-1 plays a vital role during mouse organogenesis. Defects in the mandible,   A, B) Representative images of the OHC and IHC layers in the cochlea of Dlgh-1 +/+ and Dlgh-1 2/2 mice. The organ of Corti was isolated from the cochleae of E18.5 Dlgh-1 +/+ (A) and Dlgh-1 2/2 (B) embryos, stained with phalloidin to visualize the actin bundles of the inner and outer hair cells, and viewed by confocal microscopy. Arrows indicate cells with misoriented stereociliary bundles in OHC3 row of the Dlgh-1 2/2 mice. (C) Measurement of misorientation of sterociliary bundles in OHC3 layer of Dgh-1 +/+ and Dlgh-1 2/2 mice. The angle of sterociliary bundles in OHC3 of Dlgh-1 +/+ and Dlgh-1 2/2 mice was measured in reference to a line parallel to the neuralabneural axis and perpendicular to the plane of the pillar cells. Plotted is the deviation in degrees from the neural-abneural axis for each OHC3 bundle measured in (4 cochleae from 3 mice of each genotype, 36 cells for Dlgh-1 +/+ and 33 cells for Dlgh-1 2/2 mice). No deviation from the neural-abneural axis was assigned a value of 0u. The average angle deviation from 0u for OHC3 bundles in Dlgh-1 2/2 cochlea (33.7u+/21.869, red bar) was significantly higher (p = 1. the kidneys, and the lens have been previously described in the Dlgh-1 gt/gt mice [34,35,36]. The formation of these structures and organs was likewise affected in our Dlgh-1 null mice ( Fig. 2 and data not shown). More recently, novel defects in the ureteric smooth muscle and the urogenital tract have been observed in Dlgh-1 2/2 mice [38,39]. In the present study, we provide evidence for Dlgh-1 playing a role in the skeletogenesis of trunk and limb structures, in neural tube and eyelid closure, and in the organization of the stereociliary bundles in the cochlea. These  [34,37], it is likely that the function of one or more of these protein-protein interaction domains is required for Dlgh-19s role in the novel phenotypes we have observed.

Role of Dlgh-1 in Planar Cell Polarity
Planar cell polarity is the mechanism through which the orientation of cells is coordinated within the plane of a cell sheet. In vertebrates, PCP has been suggested to be required for several developmental processes that involve convergent extension, such as lengthening of the body axis, eyelid closure, neural tube closure, branching morphogenesis in the kidney and lung, and the sterociliary cell hair bundle orientation in cochlea [3]. Dlgh-1 2/2 mice, with variable penetrance, exhibited a reduced body length to width ratio, open eyelids (Fig. 3), craniorachischisis (Fig. 5), and misorientation of stereocilia in the third row of cochlear OHCs (Fig. 6). Moreover, Dlgh-1 is expressed in the eyelid epithelium (Fig. 4A, B) and cochlear hair cells (Fig. 6F) and Dlgh-1 deficiency led to mislocalization of the core PCP protein, Vangl2, in the mutant eyelid (Fig. 4D, F). Thus, this study provides the first evidence in support of a role for Dlgh-1 in the regulation of PCP in the mouse in vivo.
Although much remains to be learned about the molecular and genetic mechanisms regulating PCP, from studies in the Drosophila eye and wing, factors including Frizzled (Fz), Dishevelled (Dsh),  Strabismus/Van Gogh (stbm/Vang), and Starry night/Flamingo (Stan/Fmi) are thought to mediate PCP through non-canonical Wnt signaling [3]. This pathway appears to be highly conserved in vertebrates. Wnt signaling has been shown to be required for mediating the organization of cochlear OHCs into their distinctive unidirectional orientation during development [48,50]. Fz3 2/ 2 ;Fz6 2/2 , Dvl1 2/2 ;Dvl2 2/2 , and Lp/Lp mice show reduced body length to width ratio, craniorachischisis, failure of eyelid closure, and misorientation of stereociliary bundles at the apices of sensory hair cells [4,42,51,52,53] Additionally, this small group of factors in vertebrates includes Celsr1 [54], a protocadherin and homolog of Stan/Fmi, and PTK 7, a protein tyrosine kinase [55]. Interestingly, Fz3, Fz6, and Vangl2 contain C-terminal PDZ binding motifs whereas Dvl1 and Dvl2 are PDZ domain proteins, suggesting that PDZ domain proteins and their ligands interact to play a major role in PCP. Dvl2 genetically interacts with Vangl2 and Vangl2 is required for the proper membrane localization of Dvl2 in the cochlear hair cells [52]. Vangl2 interacts with Fz3 and is required for targeting of Fz3 to the membrane of the cochlear hair cells [56]. It has recently been shown that these core PCP proteins also interact with other PDZ proteins in vertebrate systems. Fz4 and Fz7 have been shown to interact with the MAGUK protein, MAGI-3, through the PDZ binding motif of Fz and the first PDZ domain of MAGI-3 in vitro and in epithelial cells [57] and Dlgh-1 has been shown to interact through its PDZ domain with the C-terminal PDZ binding motif in Fz1, Fz4, and Fz7 at least in yeast two hybrid experiments [58,59]. Mice defective for the PDZ protein Scrib (Crc/crc mice) exhibit the same phenotypes as mouse Fz, Dvl, and Vangl2 mutants. Scrib been shown to genetically interact with Vangl2 to modulate PCP in the cochlea [4] and Scrib binds to Vangl2 [56]. In this study, we have shown that the phenotype of Dlgh-1 null mice bears many similarities to the phenotypes of PCP mutant mice. Furthermore, we have shown that Dlgh-1 is expressed in the eyelid epithelium (Fig. 4A, B) and the outer hair cells of the cochlea (Fig. 6F) and is required for correct subcellular localization of Vangl2 in the eyelid (Fig. 4D, F). Therefore, we suggest that Dlgh-1 also is a member of this group of factors involved in the regulation of PCP in the mouse, which is has very recently been shown to include another mouse Dlg, Dlg-3 [60]. The role for Dlgh-1 in PCP in properly localizing Vangl2 to the membrane may be a critical function for PCP as Vangl2 has been shown to be required for proper localization of Fz3 [56] and Dvl2 [61,62].
We observed differences in the phenotypes of Dlgh-1 2/2 embryos depending on the genetic background. On the C57BL/ 6J background, eyelid closure defects were present in 100% and craniorachischisis was present in 6.7% of Dlgh-1 2/2 embryos (Figs. 3, 5). Neither of these phenotypes was observed in Dlgh-1 2/2 embryos on the FVB/NJ background. The low penetrance of craniorachischisis in Dlgh-1 2/2 embryos on the C57BL/6J background is similar to observations of low penetrance of craniorachischisis in Dvl1 2/2 ;Dvl2 2/2 mice on the C57BL/6J background. By contrast the penetrance of craniorachischisis is high on the 129 background [52]. The observation that the penetrance of some phenotypes associated with PCP is dependent on mouse genetic background suggests either that there are additional genetic modulators of PCP that have yet to be identified or that polymorphisms in one or more of the already identified genes on these different genetic backgrounds influences PCP.

Role of Dlgh-1 in Skeletogenesis
The cleft palate and hypotrophic mandible in the Dlgh-1 gt/gt mice indicated that this PDZ domain protein might be important in the development of craniofacial skeletal structures [34]. The Dlgh-1 2/2 mice not only recapitulated the craniofacial defects observed in the Dlgh-1 gt/gt mice, but also displayed reduced ossification of the maxilla, middle ear, frontal, and parietal bones and overall reduced size of the skeleton (Fig. 7).
One mechanism through which loss of Dlgh-1 may affect craniofacial development is through disruption of WNT/PCP signaling [63]. Recently, it has been shown that certain core Wnt/ PCP proteins are expressed in cranial neural crest (CNC) cells [64,65], which give rise to most of the skeletal structures of the head and neck, and these core PCP proteins are required for CNC migration [66]. As noted above, Dlgh-1 has been shown to bind to several Fz receptors, including those commonly associated with Wnt/PCP signaling [58,59]. Likewise, Wnt5a, which has been suggested to be a regulator of Wnt/PCP signaling in the mouse [50], is expressed in outgrowing regions of the facial primordia and Wnt5a mutant mice display abnormally shaped head and shortened snout and mandible [67]. Finally, Wnt5a deficiency in mice results in cleft palate formation due to inhibition of the directional migration of cells that is required for palate formation [68]. Thus, the similarity of the craniofacial defects in Wnt5a 2/2 , Vangl2 Lp/Lp , and Dgh-1 2/2 mice suggests that Dlgh-1 may be a regulator of craniofacial skeletogenesis through a mechanism that involves modulating Wnt/PCP.
The Dlgh-1 2/2 mice exhibited defects not only in skeletal structures of the head, but also of the trunk and limbs and the skeletons were reduced in size overall. In particular, the forelimbs and hindlimbs of Dlgh-1 2/2 mice were approximately 20% shorter than those of Dlgh-1 +/+ mice and the humerus was clearly wider in the mutants as compared to controls (Fig. 6). One possibility is that the reduced size is due to an inhibition of proliferation, as Dlgh-1 has been associated with regulating proliferation [22]. However, Wnt/PCP also has been shown to play a role in limb development [69,70]. Wnt5a mutant mice exhibit truncated limbs [67] and the directional outgrowth of the limb has been shown to require Wnt5a/JNK signaling in the distal mesenchyme [71,72]. Similarly, Vangl2 Lp/Lp mice display shorter and wider limbs [73]. Recent studies have demonstrated a genetic and mechanistic link between Wnt5a and Vangl2 through the frizzled-like receptor, Ror2 [73], which binds Wnt5a [74], clearly establishing a connection between Wnt5a and a core PCP protein. Thus, the similarities in the limb phenotypes of these mouse mutants and the Dlgh-1 2/2 mice combined with the known mechanistic link between Dlgh-1 and Vangl2 suggest that the mechanism through which Dlgh-1 exerts its effects may be through modulating Wnt/PCP.
Although, the most likely explanation for the defects in skeletogenesis in the Dlgh-1 mutant mice is that Dlgh-1 is required the Wnt/PCP-dependent development of these skeletal structures, analysis of mouse mutants has demonstrated that numerous other signaling pathways play a role in skeletogenesis. These include FGFR [75,76], TGFbII [75,77], BMP [78], and Wnt/b-catenin [79,80,81,82]. Further studies will be required to determine if the mechanism through which Dlgh-1 affects bone formation is through an effect on Wnt/PCP and how Wnt/PCP signaling coordinates with these other signaling pathways to direct the proper formation of the skeleton.
In summary, we provide evidence of new roles for Dlgh-1 in the regulation of PCP and both endochondral and intramembranous bone formation in the mouse.

Ethics Statement
All procedures using mice conformed to the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The protocol covering these studies (protocol #M00712) was approved by the Institutional Animal Care and Use Committee of the University of Wisconsin School of Medicine and Public Health (Animal Welfare Assurance #A3368-01).

Generation of the Dlgh-1 Null Mouse
The generation of the mice carrying the gene targeted Dlgh-1 conditional allele (Fig. 1A) has been described in detail previously (Rivera et al., 2009). To generate mice carrying a germline null mutation in Dlgh-1, female Dlgh-1 fn/+ mice were mated to male EIIA-Cre transgenic mice [41] (Fig. 1). Removal of exon 8, which encodes a portion of the first PDZ domain of the Dlgh-1 protein, results in a frameshift mutation that generates a termination codon 3 codons downstream of the deletion. Male F1 progeny were mated to stock C57BL/6J or FVB/NJ females. Tail DNAs from the F2 progeny were screened by PCR for both the WT allele and the null allele using the Dlgh-1 59 primer CATCATGGTT-GAAGTGCTCTGGGC paired to the Dlgh-1 39 primer GGAAGGAAACTCACGGATGGTCC (Fig. 1).

Gross and Microscopic Analyses
For examination of morphological defects at the gross level, newborn mice were examined for overt defects and photographed under a dissecting microscope. To examine palate formation, E18.5 mice were collected from pregnant dams and an incision made through the temporal-mandibular joint to expose the palate surface. Embryos were staged by designating the morning of the vaginal plug as day E0.5.

Cochlear Hair Cell Stereociliary Bundle Staining
The intact organ of Corti from E18.5 Dlgh-1 +/+ , Dlgh-1 +/2 , and Dlgh-1 2/2 embryos was isolated in Hank's Balanced Salt Solution (HBSS, Gibco, Invitrogen) and fixed in 4% paraformaldehyde for 3 hours at 4uC. After initial fixation, the stria vascularis and tectorial membrane were removed and then the explants were fixed in 4% paraformaldehyde for overnight at 4uC. The fixed explants were permeabilized with 0.5% triton-X100 in HBSS for 30 minutes at room temperature, then stained with Alexa Fluor 568-phalloidin (1 mg/ml, Molecular Probes) for 1 hour, and finally mounted as a surface preparation on a glass slide. Stained sections were viewed on a Nikon Diaphot 200 confocal microscope and images were captured using BioRad 1024 software. The orientation of individual stereociliary bundles was determined relative to a line parallel to the neural-abneural axis and perpendicular to the row of pillar cells, as described by Montcouquiol et al. [4]. The deviation from the neural-abneural axis was measured in degrees with a protractor. A cell with the vertex of the sterociliary bundle parallel to the neural-abneural axis was assigned 0 0 deviation. A total of 4 cochleae from 3 different mice of each genotype were analyzed. The data were subjected to statistical analysis using the two-sided Wilcoxon Rank Sums test and p,.05 was considered statistically significant. For detection of Dlgh-1 in the hair cells by immunofluorescence, fixed explants were immersed in 150 mM Tris-HCl, pH 9.0 at 65uC for 15 minutes followed by acetone treatment at 220uC for 20 minutes. After washing, explants were blocked in 10% serum diluted in 0.5% Triton X-100 in PBS for 2 hours at room temperature. Explants then were incubated with anti-Dlgh-1 antibody (1:500 dilution, Novus Biologicals) overnight at 4uC followed by incubation with fluorescein conjugated horse anti-mouse secondary antibody for 2 hours at room temperature. Immunostained explants were viewed and photographed as described above.