Deletion of Porcn in Mice Leads to Multiple Developmental Defects and Models Human Focal Dermal Hypoplasia (Goltz Syndrome)

Background Focal Dermal Hypoplasia (FDH) is a genetic disorder characterized by developmental defects in skin, skeleton and ectodermal appendages. FDH is caused by dominant loss-of-function mutations in X-linked PORCN. PORCN orthologues in Drosophila and mice encode endoplasmic reticulum proteins required for secretion and function of Wnt proteins. Wnt proteins play important roles in embryo development, tissue homeostasis and stem cell maintenance. Since features of FDH overlap with those seen in mouse Wnt pathway mutants, FDH likely results from defective Wnt signaling but molecular mechanisms by which inactivation of PORCN affects Wnt signaling and manifestations of FDH remain to be elucidated. Results We introduced intronic loxP sites and a neomycin gene in the mouse Porcn locus for conditional inactivation. Porcn-ex3-7flox mice have no apparent developmental defects, but chimeric mice retaining the neomycin gene (Porcn-ex3-7Neo-flox) have limb, skin, and urogenital abnormalities. Conditional Porcn inactivation by EIIa-driven or Hprt-driven Cre recombinase results in increased early embryonic lethality. Mesenchyme-specific Prx-Cre-driven inactivation of Porcn produces FDH-like limb defects, while ectodermal Krt14-Cre-driven inactivation produces thin skin, alopecia, and abnormal dentition. Furthermore, cell-based assays confirm that human PORCN mutations reduce WNT3A secretion. Conclusions These data indicate that Porcn inactivation in the mouse produces a model for human FDH and that phenotypic features result from defective WNT signaling in ectodermal- and mesenchymal-derived structures.


Introduction
Focal dermal hypoplasia (FDH), also known as Goltz syndrome or Goltz-Gorlin syndrome, is an X-linked disorder that predominantly affects females. Individuals with FDH have a pleiotropic phenotype consisting primarily of defects of skeleton, skin, and ectodermal appendages. They often have ectrodactyly, syndactyly, brachydactyly, and oligodactyly of hands and feet, and can have shortened or absent long bones combined with typical streaks of atrophic, hypo-, and hyperpigmented skin with abnormal subcutaneous fat deposition in a thin dermis. Other common features consist of eye and ear abnormalities, brittle and sparse hair, dystrophic nails, hypodontia, and supernumerary nipples. More variable findings in FDH include short stature, pointed chin, cleft lip and palate, osteopathia striata of long bones, diastasis pubis, kidney abnormalities, abdominal wall defects, and papillomas of lips, gingivae, and tonsils. Strikingly, central nervous system abnormalities and intellectual disability are uncommon in FDH [1,2,3,4,5,6]. It has been suggested that the phenotype of FDH results from a developmental defect in signaling between ectoderm and mesoderm [7].
We and others first described that FDH is caused by mutations in the X-linked PORCN gene, which has facilitated the clinical diagnosis of FDH [8,9,10,11,12,13]. Females with FDH have heterozygous or mosaic loss-of-function mutations or large deletions of PORCN and the few affected males have mosaic mutations [14]. This explains the female predominance of the disorder, absence of male-to-male transmission, and presumed male lethality of germline mutations [1,15]. PORCN is the human orthologue of the Drosophila segment polarity gene, porcupine [16]. The Porcupine protein is a multipass transmembrane endoplasmic reticulum (ER) protein and a member of the superfamily of membrane-bound O-acyl transferases (MBOAT) [17]. Drosophila and mouse Porcupine lipid-modify Wnt proteins by covalently linking palmitoleic acid at a conserved serine residue (S209 in mouse Wnt3a), which is required for the attachment of a second acyl group, palmitate at a conserved cysteine (C77 in Wnt3a) [18,19,20]. This is important for processing, secretion, and signaling of Wingless (Wg) in Drosophila, Wnt3a in mouse, and likely several other Wnt proteins (Wnt1, 3, 4, 5a, 6, and 7b) that have wide temporo-spatial expression and signal via the canonical Wnt/b-catenin and the non-canonical planar cell polarity pathway [18,19,20,21,22,23,24,25].
Porcn has an essential role in early embryonic development [26]. It is first expressed in epiblast cells of peri-gastrulation-stage embryos and later becomes more restricted dorsally, laterally, and in the primitive streak region, with reduced expression in the anterior visceral endoderm. At later stages, expression is stronger dorsally, in the neural tube, cranial region, and optic vesicles and largely overlaps expression patterns of Wnt proteins [27]. It has also been demonstrated in studies on embryos formed by aggregating Porcn-mutant embryonic stem (ES) cells with tetraploid blastocysts as well as in vitro in cultured cells that the gene is essential for gastrulation and that its inactivation causes failure of endoderm and mesoderm differentiation [25,27].
To better understand the molecular mechanisms that cause the multiple defects of FDH and to study the function of Porcn in an in vivo mammalian model, we generated a conditional allele of Porcn in mouse ES cells to produce a tissue-specific deletion of the gene. We report here that hemizygous constitutive inactivation of Porcn causes early embryonic lethality of male embryos, while heterozygous female embryos have developmental defects of the neural tube and body wall, with most not surviving to birth. Ectodermal-specific inactivation in the skin causes alopecia, thin skin, and tooth anomalies. Inactivation in mesenchyme of the limbs causes shortening of long bones and digits. Cell-based assays demonstrate reduced secretion of WNT3A in the presence of mutant PORCN. These combined findings support that defective Wnt signaling is at the basis of the phenotypic features of FDH.

FDH-like developmental defects in a chimeric Porcn genetrap mutant mouse
We performed two sets of injections into blastocysts of mouse ES cells (line CSD256) that were targeted with a genetrap cassette containing a splice-acceptor sequence upstream of a b-galactosidaseneomycin fusion (b-geo) reporter gene inserted 39 to exon 2, which contains the translation initiation site of Porcn ( Figure 1A). Consistent with the prediction that this allele would result in inactivation of Porcn, only a single liveborn male chimera ( Figure 1B-I) was found in 28 offspring. This animal had fused and hypoplastic digits on all 4 extremities ( Figure 1B and 1E), 2 midline ventral wall defects, and a lower dorsal skin defect with alopecia ( Figure 1C). The skin lesions showed a thin epidermis with presence of a dermoid cyst in one of the ventral lesions ( Figure 1F). Interestingly, this mouse also had external male genitalia, but was a hermaphrodite with a single ovary with oviduct and a hypoplastic testis ( Figure 1G). It developed a cystic intra-abdominal structure, suspected to be a dilated obstructed vas deferens ( Figure 1D), and had a hydronephrotic kidney ( Figure 1H). We confirmed the presence of the targeted allele in skin, liver, and tail of this mouse by PCR amplification of lacZ ( Figure 1I). The phenotype in this mosaic male chimeric animal is consistent with the clinical presentation of FDH in males, who have somatic mosaic mutations [8,9,28]. The low survival of male chimeras is also in agreement with recent data showing that injection of CSD256 ES cells into tetraploid blastocysts results in early embryonic lethality and extensive gastrulation defects [27,29]. This data provided the rationale for generation of a conditionally targeted Porcn allele to study in vivo effects of mammalian Porcn inactivation.
Intronic targeting of a neomycin resistance minigene into the Porcn locus results in developmental defects in chimeric mice We next inserted an FRT-flanked neomycin (Neo) gene, a loxP site in intron 2, and a second loxP site in intron 7 of Porcn ( Figure  S1) and electroporated them into mouse ES cells to generate X Porcn-ex3-7Neo-flox /Y ES cell clones. We obtained 5/288 (1.7%) correctly targeted clones from which 3 independent lines of chimeric mice were generated after successful blastocyst injections with different clones. Of 17 chimeric mice, there were 9 high-level male and female chimeras from 2 different clones that displayed a mild phenotype reminiscent of human FDH (Figure 2A-G). These mice had absent, fused, and shortened digits on 1 to all 4 of their extremities (Figure 2A and B), vertebral anomalies of the tail ( Figure 2C), hydronephrosis ( Figure 2D), small testicles in males ( Figure 2E), and uteri with rudimentary uterine horns in females ( Figure 2F). We confirmed the presence of the targeted allele in various tissues by PCR amplification with primers P1 and P2 ( Figure 2G). Therefore, intronic integration of the Neo gene and loxP sites created at least a hypomorphic Porcn allele. The phenotype observed in chimeric mice further confirmed that as in human FDH, mosaic expression of mutant Porcn is sufficient to generate developmental abnormalities. We excised the Neo gene by mating X Porcn-ex3-7Neo-flox /Y chimeras with mice expressing Flp recombinase to generate X Porcn-ex3-7flox /X and X Porcn-ex3-7flox /Y mice ( Figure S1D), who had no observable phenotype and were fertile.

Zygotic inactivation of Porcn by EIIa-Cre and Hprt-Cre causes early embryonic lethality
To generate Porcn-ex3-7del alleles ( Figure S1E) in the zygote and earliest stages of embryonic development, X Porcn-ex3-7flox /X mice were bred to homozygous X/Y;EIIa-Cre/EIIa-Cre mice, which express Cre recombinase under control of the EIIa viral promoter [30]. From 13 litters, we obtained 38 female and 40 male offspring, but only 3 male and 3 female liveborn mice carried the EIIa-Cre transgene together with a Porcn-ex3-7flox allele ( Table 1) and had low-level mosaicism for the deleted Porcn-ex3-7del allele (not shown). The recovery of these mosaic floxed/ deleted survivors can be explained by mosaic expression of the EIIa-Cre transgene as previously described [31]. We then bred these survivors to wild type mice, but only obtained 2 heterozygous females carrying the Porcn-ex3-7del allele out of 133 offspring total. These mice were viable and showed diminished hair growth, which was also observed in their mosaic parents, but they had no other defects. We propose that skewed X-inactivation patterns may explain the fairly mild phenotype in these heterozygous offspring. Timed matings yielded 14 female and 4 male live embryos at E10.5, but only 5 out of 14 female embryos carried the Porcn-ex3-7del allele; we did not recover any X Porcn-ex3-7del /Y embryos or liveborn mice, indicating that lethality of embryos with this genotype occurs before E10.5. This is consistent with data that became available during the course of this work that support an essential role for Porcn in gastrulation [25,27].
Because of the observed mosaic inactivation of Porcn with the EIIa-Cre transgene [31] and to further examine the embryonic defects in heterozygous mutant females in this animal model for FDH, we next bred floxed mice to mice that express Cre recombinase under control of the endogenous X-linked Hprt promoter that drives expression of Cre in oocytes and zygotes. Ubiquitous non-mosaic deletion of floxed alleles can be obtained when conditionally targeted mice are bred with either female or male Hprt-Cre mice [32]. We reasoned that breeding floxed X Porcn-ex3-7flox /X females to X Hprt-cre /Y males would yield heterozygous deleted females and allow us to study the full phenotypic spectrum of Porcn mutations. It would also allow us to address whether germline heterozygous mutations in females are usually embryonic lethal, since it has been proposed that most surviving human females with FDH have either non-random X-inactivation or postzygotic mosaic mutations [9].
From 5 litters, we obtained 17 liveborn male and 9 liveborn female offspring, none of which carried the Porcn-ex3-7del allele ( Table 2), confirming that embryos with the X Porcn-ex3-7del /X genotype are lost prenatally. Timed matings revealed that 15 out of 43 live embryos at E9.5, 5 out of 18 at E10.5, and 3 out of 8 at E11.5 had the X Porcn-ex3-7del /X genotype. Most of these (.80%) had open neural tubes, defects in ventral body wall closure, and axial/ tail truncation ( Figure 3A). And mutant embryos appeared progressively smaller compared to wild type counterparts at advancing embryonic stages. By E12.5 most X Porcn-ex3-7del /X embryos were resorbed. These findings indicate that the majority of females with heterozygous inactivation of Porcn are lost prenatally, consistent with some clinical data in human FDH [9]. The axial/tail truncation and neural tube defect partially resemble Wnt3a mutants and the phenotype of conditional inactivation of b-catenin in 3 germ layers by Cre recombinase under control of a promoter fragment of the caudal-related homeobox gene (Cdx) [33,34]. The severe ventral body wall defects in heterozygous female mice are very interesting in light of recent case reports of pentalogy of Cantrell in infants or fetuses with FDH and mutations in PORCN [35,36]. The open ventral body wall was also reported in dermal b-catenin mutant mice and Dvl2;Dvl3 double mutant mice [37,38].
RNA from embryos collected at E9.5 and E10.5 was extracted to evaluate gene expression of Porcn, Wnt pathway, and Wnt target genes, as well as other developmental markers by quantitative RT-PCR. Expression of Porcn itself and the canonical Wnt signaling pathway gene Axin2 and the target gene c-myc were significantly reduced at E9.5, indicating that reduced Porcn levels affects Wnt signaling and its targets ( Figure 3B). Reduced expression of Porcn and c-myc persisted in E10.5 mutant embryos. There was also decreased expression of the neuroectoderm marker Gbx2 ( Figure 3C). Gbx2 plays an essential role in cerebellum development and maintenance of the mid/hindbrain organizer and it is also required to pattern the neural tube [39,40,41]. The decreased levels of Gbx2 suggest that reduced function of Porcn in X Porcn-ex3-7del /X embryos affects neuroectoderm development. This could in part explain the observed open neural tube defects. It also implies that, even though central nervous system anomalies and neural tube defects are not typically seen in human FDH patients, mutations of Porcn impair the development of these structures. We speculate that in humans these defects are too severe or associated with additional developmental defects to permit survival to birth.

Conditional inactivation of Porcn in limb mesenchyme causes mesomelic limb shortening and digital abnormalities
To further investigate the mechanisms by which loss of Porcn results in the limb defects of FDH, and to initiate studies into the role of Porcn in mesenchyme-derived tissues, we bred Porcn-ex3-7flox mice with Prx-Cre mice, reported to drive Cre expression in mesenchyme at early stages of forelimb development ( Figure 4) [42]. Analysis of 6 litters ( Table 1) indicated preserved Mendelian ratios in liveborn offspring. Male offspring (14/14, 100%) with X Porcn-ex3-7flox /Y;Prx-Cre genotypes and predicted Porcn-ex3-7del in tissues where Prx-Cre is expressed, have significantly shorter limbs  Table 1. Liveborn offspring from breeding of X Porcn-ex3-7flox /X mice to 3 different Cre strains. as early as E18.5 ( Figure 4A). They are also smaller than their littermates at postnatal day 7 (P7) ( Figure 4B) and become increasingly stunted by P28 ( Figure 4C). Their limbs remain very short with apparent syndactyly of digits of both fore-and hindlimbs in some animals (3/14, 21%) ( Figure 4D). Skeletal preparations show that long bones of all extremities and digits are shortened, but we did not observe bony fusion of digits, indicating that the syndactyly is limited to soft tissues and does not include skeletal elements ( Figure 4E-J). As expected, the Porcn-ex3-7del allele was detected by PCR in affected tissues from limbs of these mice ( Figure S2A). These data indicate that Porcn deletion in mesenchymal derivatives is sufficient to cause severe defects of fore-and hindlimbs. As expected from the known expression pattern of Prx-Cre [42], defects were more severe in forelimbs. The shortened limb phenotype is similar to that of Wnt5a null except that our mice did not show loss of distal digits as Barrott et al.  observed [25,43]. This observation supports that Porcn may be required for ''noncanonical'' Wnt signaling; yet, the reason we observe a milder phenotype in X Porcn-ex3-7flox /Y;Prx-Cre mice remains to be discovered. The shortened limb phenotype models the features of severe cases of FDH, except for absence of bony syndactyly [44].
Conditional inactivation of Porcn by Krt14-Cre-mediated deletion causes thinning of the skin with absent hair follicle development and dental defects To identify the origin of defects of the skin and ectodermal appendages present in FDH and to investigate the contribution of altered ectodermal PORCN function to the FDH phenotype, we bred X Porcn-ex3-7flox /X mice with X/Y;Krt14-Cre/-mice. The Krt14 promoter drives Cre expression in the basal layers of developing epithelia, causing deletion of floxed alleles in the epidermis and other ectodermal derivatives [45]. Mendelian ratios were preserved among liveborn offspring of these matings ( Table 1). Heterozygous X Porcn-ex3-7flox /X;Krt14-Cre/-females had barely detectable hair loss, but males with Krt14-Cre-driven Porcn deletion had large areas of thin skin with alopecia ( Figure 5A and B). The mosaic pattern of this phenotype likely results from variable expression of the Krt14-Cre transgene, as we have previously observed. Consistent with Krt14-driven Cre expression in multiple ectodermal derivatives, the mutant mice also had abnormally formed incisors ( Figure 5C-E).
We detected the Porcn-ex3-7del allele in affected skin ( Figure  S2B) and confirmed reduced Porcn expression by RNA in situ hybridization with a Porcn riboprobe in the skin of E16.5 mutant embryos ( Figure 5F-I). Histological analysis of skin at P9 showed that affected areas are completely devoid of hair follicles ( Figure 5J and K). This phenotype is a phenocopy of that obtained when b-catenin is conditionally inactivated in skin by the same Krt14-Cre ( Figure 5L and M), supporting the conclusion that it results from defective Wnt signaling. Analysis at E16.5 revealed that early budding of epidermal cells to form the hair placodes that should give rise to hair follicles did not take place in these embryos ( Figure 5N and O). Absent staining in mutant skin for the early hair follicle marker P-cadherin and for Sox9, which is present in hair follicles from the placode stage, confirmed that placode formation was drastically compromised ( Figure 5P-S), and indicates that hair follicle morphogenesis is blocked. This suggests that these aspects of the epidermal phenotype can result from ectodermal loss of Porcn ( Figure S2B).
Therefore, although Prx drives Cre expression primarily in mesenchymal derivatives and Krt14 drives it primarily in ectodermal derivatives, mice with Porcn inactivation in either of these Cre lines have phenotypes that are remarkably similar to those seen in human patients, implicating that disruption of both mesenchymal and ectodermal PORCN contributes to the FDH phenotype. Interestingly, we also noted that in the areas of thinned skin, the subcutaneous fat was directly adjacent to the outermost epidermal layers of the skin ( Figure 6). While this will require further detailed investigation, this finding may be relevant to the observed areas of apparent herniating fat found in the skin defects of patients with FDH.

Effects of overexpressed wild type and mutant PORCN on cellular WNT3A secretion
It has been shown that Drosophila Porcupine and, in cell culture assays, its murine ortholog lipid-modify Wnt proteins (in particular Wnt3a) to promote their secretion from the ER and consequently from Wnt-producing cells, allowing them to act as signaling molecules that activate receptors on Wnt-responding cells [18,22,23]. However, a direct effect of Porcupine on WNT-protein secretion and accordingly a link between FDH-causing mutations and WNT signaling had not yet been demonstrated in human cells when we initiated our studies. We therefore performed a cell-based assay to investigate how human wild type (wt) and mutant PORCN influence secretion of WNT proteins (Figure 7). Co-expression of wt PORCN and WNT3A in human embryonic kidney (HEK293T) cells showed that, as expected, transiently overexpressed wt PORCN reduced WNT3A in the cell to virtually undetectable levels, indicating increased secretion of WNT3A. In contrast, PORCN with mutations p.M1I (c.3G.A) and p.R124X (c.370C.T), known to cause human FDH, resulted in WNT3A retention in cells compared to wt PORCN ( Figure 7A). However, other PORCN mutations found in FDH, p.S136F (c.407C.T), p.G168R (c.502G.A), and p.Y359X (c.1077C.A), did not or only mildly affected WNT3A retention in cells. It remains to be investigated how these mutations affect WNT protein signaling. Interestingly, other recent studies also indicate that Porcupine proteins without lipid adducts are still secreted but have defective Wnt signaling [25,46]. When the same assay was performed with overexpressed WNT1, which is also known to be modified by Porcupine, we found that disease-causing mutations of PORCN did not reduce WNT1 secretion as much as was seen with WNT3A ( Figure S3). These results are in agreement with other non-human data demonstrating that the functional consequences of lipid modification of Wnt proteins by Porcupine varies between different Wnt ligands and that implications for Wnt signaling are complex and not limited to their secretion from the ER [19].

Discussion
PORCN was first identified as the mutated gene in Goltz-Gorlin syndrome or FDH in 2007 [8,9]. Since then, at least 80 different mutations and large genomic deletions of the PORCN gene have been found in FDH patients [36]. The resulting phenotype in human patients varies widely from mild skin and distal skeletal defects to severe forms of FDH, with multiple defects that include severe limb abnormalities involving long bones, aplasia cutis, limbbody wall complex anomaly, pentalogy of Cantrell, and Van Allen-Myhre syndrome [35,36,47]. Identification and characterization of PORCN mutations as the cause of FDH was important for genetic counseling and diagnostic testing in suspected cases. However, an animal model for the condition and functional cell-based assays are needed to determine the functional in vivo consequences of Porcn mutations on Wnt signaling and to investigate new therapies.
We first attempted to generate mice from ES cells with a genetrap insertion in the Porcn locus. This produced a single chimeric animal, but not unexpectedly, there was no germline transmission of this null allele. We then generated a conditional allele of Porcn by introducing loxP sites flanking exons 3 and 7. When targeted mice are bred with early embryonic Cre-expressing mice (EIIa-Cre, Hprt-Cre), ectoderm-specific Krt14-Cre, and limb mesenchyme-specific Prx-Cre mice, a variety of interesting phenotypes are observed that recapitulate various aspects of the human condition and prove that these mice provide a reliable animal model for human FDH. Our findings are in agreement with and significantly expand on those reported by Barrott and colleagues during the course of this work in a mouse model with a deletion of exons 2 and 3 of Porcn [25]. We also made important novel observations that may be relevant for other human conditions potentially resulting from Porcn mutations or other defects in the Wnt signaling pathway.
Although lethality of Porcn mutations in hemizygous males has already been confirmed in studies with embryos resulting from aggregation of CSD256 ES cells with tetraploid blastocysts [25] and in vivo [27], our data provides the first in vivo evidence that many heterozygous females with inactivating mutations of PORCN also die in utero. We also demonstrate that, even though neural tube defects are not typically present in liveborn humans with FDH, severe cranial neural tube defects are common in the heterozygous embryos. This is consistent with expression of Porcupine and Wnt proteins in the developing central nervous system and suggests that associated defects could be responsible for the lack of surviving humans with FDH that have NTDs or other severe brain abnormalities [34]. We also demonstrated in different ways that mosaicism for a Porcn mutation is important for the phenotype: we observed limb, skin, and internal organ abnormalities in the CSD256 genetrap chimera and in chimeric animals carrying the Porcn-ex3-7Neo-flox allele, as well as mosaic skin defects in mice with Porcn-ex3-7 deletion driven by Krt14-Cre and by EIIa-Cre. Having the current mouse model will also allow us to study the contribution of X-inactivation patterns for the mosaic expression of the phenotype in females.
Krt14-Cre-mediated Porcn inactivation also causes complete absence of hair follicle development and differentiation. This is in agreement with the lack of hair follicles described with ubiquitous Cre recombinase-mediated inactivation [25], but our observation that inactivating Porcn only in ectoderm using Krt14-Cre is sufficient to cause this phenotype is novel. In addition, we found that the defects in hair follicle development phenocopy those observed after Krt14-Cre-driven b-catenin inactivation, implying that they are caused by defective canonical Wnt signaling in ectodermal derivatives. The absence of P-cadherin and Sox9 indicates a very early defect in specification of the hair placode. Our findings of abnormal dentition in the animals are also novel and may result from similar disruption in ectodermal Wnt signaling in the tooth placode.
Another novel observation is that the chimeric animals have some of the rarer phenotypic features occasionally described in FDH, such as urogenital anomalies. Considering the known role of Wnt proteins in the development of the urogenital tract, the Porcn-ex3-7flox mice will be valuable to study how Porcupine regulates Wnt signaling during the development of the urogenital tract.
Finally, the cell culture assays provide evidence that PORCN containing mutations that cause human FDH can influence the secretion and signaling of WNT proteins from the cell. Consistent with prior data [18,25,27,46], we demonstrate that knockdown of PORCN by siRNA, as well as some of the FDH-causing mutations that we studied, cause retention of WNT3a in the cell. In contrast, similar experiments with WNT1 show a lesser and different effect on WNT protein secretion and indicate that the effects of identical mutations in PORCN cannot be generalized to all WNT proteins. A recent study also showed that differential palmit(e)oylation of two residues of Wnt1 has both overlapping and distinct consequences, whereby one is Porcn-dependent and the other is Porcn independent. It has been proposed that this differential residue modification may play a role in determining participation in the canonical versus non-canonical Wnt-signaling pathway [48]. This data provides further evidence that Porcupine has a complex role in the regulation of secretion and signaling of Wnt ligands and that individual Wnt-proteins are differently influenced. This is consistent with data by others that indicate that effects of Porcupine on Wnt signaling can be uncoupled from Wnt ligand secretion [19].
The mouse Porcn alleles that we generated, supported by cellbased assays, provide evidence for an important role of Porcupine in ectodermal and mesodermal signaling of Wnt proteins. This and other generated mouse models [25] will be important tools for further detailed characterization of the function of PORCN in development and disease, and have provided a model for investigating potential therapies for some of the features of FDH that progress postnatally, such as skin defects and peri-orificial papillomas.

Materials and Methods
All animal procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee. All animals are monitored daily by experienced personnel, animal staff and veterinarians. Animals were euthanized when they were unable to stand or eat, or when they displayed agonal breathing, vocalization, self-mutilation, had dehiscent wounds, hypothermia, weight loss of .20%, or seemed otherwise to be suffering. For animals that were impaired in reaching standard food supply due to skeletal abnormalities, food pellets were placed on the bottom of the cage, and we confirmed that animals were reaching these pellets and were also able to reach drinking water.

Generation and analysis of the Porcn genetrap mouse
Mouse embryonic stem (ES) cells (line CSD256) carrying a genetrap mutation in intron 2 of Porcn ( Figure 1A) previously generated as described [49] were purchased from the Mutant Mouse Regional Resource Centers (MMRRC). ES cells were injected into blastocysts that were transferred into pseudopregnant females, yielding 1 male chimera out of 28 offspring. This mouse was sacrificed at 3.5 months of age and tissues were snap frozen for genetic analysis or fixed in 10% formalin and embedded in paraffin for sectioning and hematoxylin-eosin (H&E) staining for histopathology. Skeleton reconstruction of the bony skeleton (without cranium) was performed by micro-CT using a Gamma Construction of the targeting vector, embryonic stem cell (ES) culture, and gene targeting The pFRT-LoxP plasmid used to construct the targeting vector was kindly provided by Dr. James Martin (Institute of Biosciences and Technology, Houston). The pFRT-LoxP contains 2 loxP sites, a neomycin gene flanked by two FRT sites, and a TK cassette for negative selection. The 4.7 kb 59 targeting arm, 1.7 kb mid arm, and 5 kb 39 targeting arm of the Porcn targeting vector and diagnostic probes for correct targeting were amplified from BAC clone bMQ-207H16 (129S7/AB2.2, Sanger Center) using Ta-KaRa LA Taq TM DNA Polymerase (Takara Bio, Inc.) and cloned into a TA-vector (TOPOH TA Kit, Invitrogen). A ScaI site for future characterization was introduced into the forward PCR primer used to amplify the 59 arm. The 59 arm was released from the TA-vector with PacI and AscI restriction digestion and cloned into PacI/AscI sites of the pFRT-LoxP plasmid; the mid arm was released with NotI and Acc65I, and digested and cloned into NotI/ Acc65I sites of the pFRT-LoxP plasmid; the 39 arm was released by XhoI restriction digestion and cloned into a XhoI site of the pFRT-LoxP plasmid to generate the Porcn-ex3-7Neo-flox targeting construct. This construct was sequenced to confirm the genomic DNA sequence and insertion sites, linearized by PacI restriction digestion and electroporated into mouse ES cells of a 129S5/ SvEvBrd genetic background. ES cells were cultured in medium with G418 (geneticin) and FIAU [1-(2-deoxy-2-fluoro-beta-Darabinofuranosyl)-5-iodouracil] for positive and negative selection, respectively. DNA from surviving ES cell clones was isolated for restriction digestion with EcoRI and Southern analysis by hybridization with an external 39 probe located downstream of the 39 arm. Because of the presence of 59 repeated sequences, the 59 probe was located inside the 59 targeting arm and was used after restriction digestion with ScaI for confirmation of results from the 39 probe.

Mouse genotype analysis by PCR
The positions of the genotyping primers are indicated in Figure  S1. Primer sequences are P1: CAAGTCCTTCCGCATATGGT, P2: TGCATGCTTCAGGTAAGACG, P3: TACCGCACAGA-GACCAAGTG and P4: CCACACTCAGGGGAGTCAGT. PCR reactions were performed on DNA extracted from tail-tip, skin, or other tissues using the following conditions: 1 cycle at 95uC for 3 minutes, 35 cycles at 95uC for 30 seconds, 58uC for 30 seconds, and 72uC for 40 seconds, and 1 cycle at 72uC for 5 minutes. Primer sequences for genotyping EIIA-Cre, Prx-Cre, and FLP mice were obtained from Jackson Labs. Primers for genotyping Krt14-Cre mice are CreF: TGCTGTTTCACTGGT-TATGCGG and CreR: TTGCCCCTGTTTCACTATCC.
Skeleton preparations, hematoxylin-eosin staining, Oil red O staining, RNA in situ hybridization and immunofluorescence Skeletons were prepared following the standard Alcian Blue and Alizarin Red staining protocol. Hematoxylin-eosin and immunofluorescence staining were done on 4% PFA-fixed 6-10 mm sections of OCT-frozen dorsal skin or embryos. Oil red O staining was performed on Krt14cKO and control 4% PFA-fixed 6-10-mm sections of OCT-frozen dorsal P9 skin following standard protocol. Automated non-radioactive RNA in situ hybridization with an antisense Porcn riboprobe that is predicted to detect all known transcript variants and a sense control riboprobe was performed on skin at mouse embryos at E16.5 as described [27]. Primers for generating the riboprobe were prepared according to Biechele, et al. [27]. For immunofluorescence analysis, block-diluent solution of 5% normal donkey serum, 2% gelatin, and 0.2% Triton X-100 in PBS was used with primary antibodies to Sox9 (1:100, Santa Cruz Biotechnology), b4-integrin (1:200, BD Biosciences), Pcadherin (1:200, R&D Systems), and Keratin 5 (1:500, gift from Colin Jamora) and FITC-or RRX-conjugated donkey secondary antibodies (1:150 and 1:200, respectively, Jackson Labs). Nuclei were stained with Hoechst.

WNT3A and WNT1 secretion assay
Human PORCN isoform B (pDONR223-PORCN, ID:7828), WNT3A (pCR-BluntII-WNT3A, ID: 40007188) and WNT1 (pCR4-WNT1, ID: 30915309) cDNA clones were purchased from Open Biosystems. All inserts were PCR-amplified and subcloned to pcDNA3.3-topoTA vector (Invitrogen). Since PORCN isoform D is the longest PORCN transcript, we used it to investigate the function of PORCN. Isoform D clones were generated by inserting an 18 bp fragment into the PORCN B sequence using the QuickChangeH II Site-Directed Mutagenesis Kit (Agilent Technologies). PORCN isoform D mutants (p.M1I, p.R124X, p.S136F, p.G168R and p. Y359X) were also generated using this kit. Control plasmid pcDNA3.1-GFP was generously provided by Laura W. Burrus (San Francisco State University). HEK293T/17 cells (CRL-11268 TM , ATCC) were cultured in 6-well plates to 50% confluence. For each well, 500 ng of WNT3A or WNT1 was co-transfected with 500 ng of wild type or mutant PORCN isoform D plasmids using the PolyJet TM DNA In Vitro Transfection Reagent (SignaGen). After 48 hours, cells were harvested and cell lysates were prepared for Western blot analysis. The assay was performed at least 4 times with 2 or 3 independent replicates for each. Rabbit polyclonal antibodies to Wnt3a (ab28472, Abcam), Wnt1 (ab15251, Abcam) and b-tubulin (ab6046, Abcam) were used for Western blotting. Quantification was done using ImageJ software (U.S. National Institutes of Health).