Deletion of mouse Porcn blocks Wnt ligand secretionand reveals an ectodermal etiology of human focaldermal hypoplasia/Goltz syndromeJared J. Barrotta, Gabriela M. Casha, Aaron P. Smithb, Jeffery R. Barrowb, and L. Charles Murtaugha,1

aDepartment of Human Genetics, University of Utah, Salt Lake City, UT 84112; and bDepartment of Physiology and Developmental Biology, Brigham YoungUniversity, Provo, UT 84602

Edited* by Clifford J. Tabin, Harvard Medical School, Boston, MA, and approved June 16, 2011 (received for review May 12, 2010)

TheDrosophila porcupine gene is required for secretion ofwinglessand other Wnt proteins, and sporadic mutations in its uniquehuman ortholog, PORCN, cause a pleiotropic X-linked dominantdisorder, focal dermal hypoplasia (FDH, also known as Goltz syn-drome). We generated a conditional allele of the X-linked mousePorcn gene and analyzed its requirement in Wnt signaling and em-bryonic development. We find that Porcn-deficient cells exhibita cell-autonomous defect in Wnt ligand secretion but remain re-sponsive to exogenous Wnts. Consistent with the female-specificinheritance pattern of FDH, Porcn hemizygous male embryos arrestduring early embryogenesis and fail to generate mesoderm, a phe-notype previously associated with loss of Wnt activity. Heterozy-gous Porcn mutant females exhibit a spectrum of limb, skin, andbody patterning abnormalities resembling those observed in hu-man patients with FDH. Many of these defects are recapitulated byectoderm-specific deletion of Porcn, substantiating a long-standinghypothesis regarding the etiology of human FDH and extendingprevious studies that have focused on downstream elements ofWnt signaling, such as β-catenin. Conditional deletion of Porcn thusprovides an experimental model of FDH, as well as a valuable toolto probe Wnt ligand function in vivo.

epidermis | dermis | hair follicle | skeletal development

Wingless/Wnt signaling has been implicated in the devel-opment of nearly all animal tissues as well as human dis-

eases, such as diabetes and cancer (1, 2). Although the Wntsignaling pathway was first delineated in Drosophila, translatinginsights from this species to vertebrates has been complicated bygenetic redundancy among its components (2). β-catenin is oneof the few nonredundant components of the “canonical” Wntpathway, and its genetic manipulation is widely used to studyWnt signaling in the mouse (3). A requirement for β-catenin isnot necessarily the same as a requirement for Wnt, however,given that each can function independently of the other (4, 5).Drosophila wingless/Wnt secretion and activity require the dedi-

cated function of an endoplasmic reticulum (ER)-localizedacyltransferase enzyme, porcupine (6–8). Porcupine has a sin-gle mammalian ortholog, Porcn (9), and inhibiting this moleculeby RNAi or small molecule antagonists impairs the palmitoyla-tion, secretion, and activity of multiple vertebrate Wnts (10–12).Although Porcn is one of several related membrane-boundO-acyltransferase (MBOAT) enzymes (13), functional studiesreveal no substrate overlap between Porcn and other MBOATs(12, 14–16). These observations suggest that Porcn representsa genetic “bottleneck” in the vertebrate Wnt pathway, compa-rable to β-catenin but regulating ligand production rather thanresponse. Nonetheless, the relationship of Porcn to Wnt functionhas yet to be analyzed genetically in a vertebrate model organism.In fact, the first loss-of-function phenotype of this gene was

described in humans, as the X-linked dominant syndrome focaldermal hypoplasia (FDH, also known as Goltz syndrome; OnlineMendelian Inheritance in Man (OMIM) no. 305600) (17–19).Most patients with FDH are female heterozygotes, and the syn-

drome is never transmitted to male offspring, suggesting male-specific embryonic lethality. The congenital abnormalities asso-ciated with FDH are highly pleiotropic and variable, includingmultiple aspects of skin and skeletal development (20), and con-siderably overlap defects observed in mouse Wnt pathwaymutants (Table S1). Given the connections between porcupine/Porcn and Wnt signaling, we developed a conditional allele ofmouse Porcn that provides a unique genetic tool to block Wntligand biogenesis.

ResultsDeletion of Mouse Porcn Abolishes Wnt Production but Not Respon-siveness. The Porcn targeting vector places loxP sites aroundexons 2 and 3, such that Cre-mediated deletion will eliminate thePorcn protein start codon and the first three predicted trans-membrane domains (Fig. 1A and Fig. S1 A and B). Homologousrecombination inserts an Flp recognition target (FRT)-flanked,promoterless neoR selection cassette downstream of the first(noncoding) exon; neoR should not be expressed if the targetingvector integrates randomly into the genome (21). Although rel-atively few colonies were obtained after electroporation andselection, 12 of 18 clones analyzed had undergone targeting,thereby disrupting the only Porcn allele in these X/Y ES cells(Fig. S1C). We obtained clones either incorporating or omittingthe distal loxP site, based on the position of the 3′ crossoverevent (Fig. S1 A and C), and one clone of each (Porcnneo2lox/Y orPorcnneo1lox/Y) was used for further study (Fig. S1D).As depicted in Fig. S1D, the neoR cassette was excised by

transient expression of Flp, and matched sublines were derivedin which the Porcn coding sequence was left intact (referred to asPorcnlox) or subjected to Cre-mediated deletion of exons 2 and 3(PorcnΔ). Because all Porcnlox and PorcnΔ sublines behavedidentically, they are henceforth described collectively rather thanreferring to individual subclones.A custom anti-Porcn antiserum, raised against an epitope

encoded by exons 9 and 10, detected a specific band of ∼40 kDain lysates from WT and Porcnlox ES cells that was absent fromPorcnΔ lysate (Fig. 1B). This protein comigrated with a specificband present in lysates of HeLa cells overexpressing mousePorcn, suggesting that it represents endogenous Porcn. The ab-sence of lower molecular-weight bands in PorcnΔ lysate suggeststhat alternative translation initiation sites are not used and thatdeletion of exons 2 and 3 produces a null allele. PorcnΔ ES cellsgrew normally under standard conditions and maintained ex-

Author contributions: J.R.B. and L.C.M. designed research; J.J.B., G.M.C., A.P.S., J.R.B., andL.C.M. performed research; J.J.B., G.M.C., A.P.S., J.R.B., and L.C.M. analyzed data; and J.J.B.and L.C.M. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1To whom correspondence should be addressed. E-mail: murtaugh@genetics.utah.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1006437108/-/DCSupplemental.

12752–12757 | PNAS | August 2, 2011 | vol. 108 | no. 31 www.pnas.org/cgi/doi/10.1073/pnas.1006437108

pression of the pluripotency marker Oct4 (Fig. S2 A–D), in-dicating that targeted disruption of Porcn does not prevent EScell self-renewal.We analyzed the effects of Porcn disruption on Wnt signaling

by transient transfection of a β-catenin/T-cell factor (TCF)-dependent TOPFlash reporter gene (22, 23). Porcnlox and PorcnΔ

ES cells were assayed for their response to Wnts either addedexogenously, via conditioned medium of L-Wnt3a cells (24), oroverexpressed endogenously, by transfection with Wnt expres-sion plasmids. Although both Porcnlox and PorcnΔ ES cellsresponded robustly to exogenous Wnt3a, only Porcnlox cellsexhibited TOPFlash activation when transfected with Wnt3a orWnt1 (Fig. 2A and Fig. S2E). The defective endogenous Wntresponse of PorcnΔ cells was fully rescued by cotransfection withWT human PORCN, whereas a missense mutant observed ina patient with FDH, affecting the putative active site histidine(H341L) (19), had no rescuing activity (Fig. 2A). This functionaltest of a human PORCN mutant supports the hypothesis thatFDH is a disease of impaired Wnt signaling.

Porcn Is Required for Mesoderm Induction in ES Cells and Embryos.Wnt signaling is required for mesoderm formation in the mouse(25–27), and this requirement is recapitulated in ES cell-derivedembryoid bodies (EBs) (28, 29). As an assay of endogenous Wntactivity, we examined the effect of Porcn deletion on EB de-velopment. Consistent with a defect in mesoderm development,individually picked PorcnΔ EBs failed to form beating myosinheavy chain (MHC)-expressing cardiomyocytes after 13 d of at-tachment culture; by that time, beating foci of MHC+ cells haddeveloped in all EBs derived from parental WT and Porcnlox ES

cells (n = 12 EBs per genotype; Fig. S2 F–H). Neuronal differ-entiation, indicated by βIII-tubulin+ cells, occurred normally inall genotypes (Fig. S2 I and J).Having established that Porcn was required for Wnt signaling

in vitro, we generated mice from Porcnlox (i.e., Porcn2lox) ES cells(Fig. S1D). Porcnlox/Y and Porcnlox/lox mice are viable and fertile,and they are indistinguishable from WT littermates. To deletePorcn in vivo, we crossed Porcnlox females to males carrying Sox2-Cre, an epiblast-specific deleter transgene (30). Embryos wereharvested at an early gastrulation stage [embryonic day (E) 6.5]and analyzed for expression of the early mesoderm markerBrachyury. Brachyury was expressed normally in Porcnlox/Y em-bryos but was undetectable in Porcnlox/Y; Sox2-Cre+ littermates(henceforth referred to as PorcnΔ/Y) (Fig. 1 C–F). Porcn is thusessential for mesoderm formation in vivo, like other Wnt com-ponents (25–27). We have not recovered PorcnΔ/Y males in lateembryogenesis (E15.5–E18.5), and a recently described gene

Fig. 1. Generating and characterizing a conditional mouse Porcn allele. (A)Schematic diagram of WT, loxP-targeted, and deletion alleles of Porcn. Exonsare boxed and numbered, with the coding region indicated in black. (B)Western blot of whole-cell lysates from parental and targeted ES cells (Left)and HeLa cells transfected with empty vector or mouse Porcn (Right) usinga polyclonal antiserum against a C-terminal Porcn epitope. The diamondindicates the band corresponding to overexpressed mouse Porcn and ispresent only in parental and Porcnlox ES cells. (C and D) Whole-mount in situhybridization for Brachyury (purple) on E6.5 control or PorcnΔ/Y embryos.(Scale bar: 100 μm.) (E and F) Sections through Brachyury whole-mountstained embryos, with approximate positions indicated by the dotted lines inC and D. Note that sections were taken from an independent representativepair of embryos, overstained to preserve signal in sections (asterisks indicatenonspecific background). PorcnΔ/Y embryos consistently contain a hollowlumen at this stage, indicating a lack of gastrulation.

Fig. 2. Wnt activity and processing in Porcn-deficient cells. (A) TOPFlashluciferase reporter assays comparing the response of Porcnlox/Y (blue) andPorcnΔ/Y (red) ES cells with exogenous Wnt3a (10% (vol/vol) L-Wnt3a–con-ditioned medium) or endogenously overexpressed Wnt3a (cells transfectedwith Wnt3a expression plasmid). Where indicated, cells were cotransfectedwith expression plasmids for WT or H341L mutant human PORCN. Relativelight units indicate TOPFlash activity, normalized to an internal transfectioncontrol and plotted on a log scale as fold change relative to untreatedPorcnlox/Y cells (n = 3–8 independent experiments per condition). RLU, rel-ative light unit. *P < 0.05 by Welch’s two-tailed t test. (B) Firefly luciferaseassays of 10T1/2 cells stably infected with a TOPFlash-based lentiviral re-porter, cocultured with immortalized Porcnlox/Y (blue) and PorcnΔ/Y (red)MEFs that were previously infected with an empty retroviral vector (LNCX) orLNC-Wnt3a-HA. Relative light units are plotted as fold change relative toreporter-transduced cells cultured alone (black) (n = 3 independent experi-ments). *P < 0.05 by Welch’s two-tailed t test. (C) Western blots (withantibodies indicated to left of panels) on whole-cell lysates from control andWnt3a-HA–expressing MEFs (Left) and on anti-HA immunoprecipitates ofconditioned media from the same cells (Right, asterisk indicates rabbit IgGheavy chain). Note that the immunoprecipitates represent ∼10-fold moreinput material than the corresponding cellular lysates. Tubulin serves asa loading control. (D) Western blots (antibodies indicated on left) on lysatesfrom control and Wnt3a-HA–expressing MEFs, either unprocessed (whole) orseparated into aqueous and detergent phases by Triton X-114 extraction.GAPDH and β-catenin serve as controls for recovery of nonacylated aqueous-phase proteins, and Ras serves as a control for detergent-phase recovery ofacylated proteins.

Barrott et al. PNAS | August 2, 2011 | vol. 108 | no. 31 | 12753

DEV

ELOPM

ENTA

LBIOLO

GY

trap allele of Porcn was found to cause developmental arrest atgastrulation (31). We therefore conclude that the female-specificinheritance of human FDH results from early loss of PORCN-deficient male embryos.

Loss of Porcn Causes a Cell-Autonomous Defect in Wnt LigandSecretion. Drosophila porcupine is essential for wingless proteinsecretion (7), and inhibiting Porcn function by RNAi or a smallmolecule antagonist prevents Wnt3a secretion (10, 12). Becausetransient transfection of ES cells did not produce sufficient Wntproteins for biochemical analysis, we isolated and immortalizedPorcnlox/Y mouse embryo fibroblasts (MEFs), generated PorcnΔ/Yderivatives by Cre transduction, and stably infected Porcnlox andPorcnΔ cells with either an empty retroviral vector (LNCX) or avector encoding HA-tagged Wnt3a (LNC-Wnt3a-HA), previouslyshown to be biologically active (32). TOPFlash assays indicatedrobust TCF activity in Porcnlox/Wnt3a cells, which was absentin PorcnΔ/Wnt3a cells but fully restored by transfection withhuman PORCN (Fig. S3A). To confirm that Porcn acts cell-autonomously, we stably transduced C3H 10T1/2 fibroblasts witha TOPFlash-based lentivirus (33), generating a WT “responder”cell line, and cocultured these with the above-described MEFs.Coculture with Porcnlox/LNCX cells produced a modest inductionof LEF/TCF activity (∼8-fold over responder cells cultured alone),consistent with endogenous Wnt production by MEFs (34),whereas Porcnlox/Wnt3a cells induced >100-fold activation (Fig.2B). PorcnΔ/Wnt3a MEFs failed to induce significant LEF/TCFactivity in responder cells (Fig. 2B), confirming that Porcn func-tions specifically in Wnt ligand-producing cells.Western blotting of whole-cell lysates indicated that Porcnlox

and PorcnΔMEFs expressed similar levels of Wnt3a-HA, whereasanti-HA immunoprecipitates of conditioned media revealed thatonly Porcnlox cells produced soluble Wnt3a-HA (Fig. 2C). Stain-ingMEFs for colocalization ofWnt3a-HAwith secretory pathwaymarkers, we found in both genotypes that anti-HA immunofluo-rescence completely overlapped the ER marker protein disulfideisomerase but did not detectably overlap the Golgi marker giantin(Fig. S3 B–S). These results suggest that Wnt3a-HA is traffickedrelatively inefficiently through the secretory pathway but that itssecretion nonetheless depends on Porcn function (Fig. 2C). Thisconclusion is in agreement with previous in vitro studies of Porcnfunction, in which impaired Wnt secretion correlated with de-fective fatty acid modification of the ligand (10–12). Using a Tri-ton X-114 partitioning assay to assess Wnt acylation (11, 12), wewere surprised to find that detergent partitioning of Wnt3a-HAwas reduced but not eliminated in PorcnΔ cells (Fig. 2D). Becausethis minor change could not account for the >100-fold reductionin Wnt3a signaling activity of PorcnΔ cells, we hypothesize thatWnt3a is subject to Porcn-independent acylation events that areinsufficient for ligand secretion.

Porcn Heterozygosity Provides a Mouse Model of FDH Defects. Todetermine the in vivo requirements for Porcn after gastrulation,we used Sox2-Cre to generate PorcnΔ/+ females, all of whichwere recognizably abnormal in late embryogenesis (E15.5–E18.5,n = 42 total). The phenotypes, summarized in Table S1, closelyresembled human FDH and partially recapitulated one or morepreviously described Wnt pathway mutants (Fig. 3). As in thehuman syndrome, PorcnΔ/+ phenotypes varied widely in severity,most likely attributable to stochastic X-inactivation (20). Forexample, almost every heterozygous embryo examined exhibitedone or more abnormal limbs, with defects ranging from digit lossor fusion to complete absence of autopod and ulna (Fig. 3 D–F).The defining phenotype of FDH in humans is thin or absent

dermis, which typically manifests at birth in discrete lesionsranging in size from millimeters to centimeters (20). Similarly, weobserved large areas of dermal atrophy in a subset of PorcnΔ/+

embryos, such that internal organs, including liver and heart, were

visible through an epidermal monolayer (Fig. 3C). In severe cases,these defects were accompanied by hypoplasia of the sternum aswell as ventral body wall closure defects (Fig. 3 G–J). Thesephenotypes recapitulate those of dermal-specific β-catenin KOs(35), as well as those attributed to severe cases of human FDH(20, 36).Other grossly obvious defects have not been reported in FDH

but resembled those of other mouse Wnt pathway mutants(Table S1). For example, almost all PorcnΔ/+ embryos exhibiteda range of tail defects (Fig. 3 B and C) similar to those of Wnt5anulls (37), Wnt3a and Lrp6 hypomorphs (38, 39), and Dvl2;Dvl3double mutants (40). Taken together, the concordance of PorcnΔ/+,FDH, and mouse Wnt mutant phenotypes provides independentsupport for a central role of Porcn in Wnt signaling.

Porcn Is Required for Ectodermal Expression of Lef1 and Hair FollicleDevelopment. FDH is commonly associated with localized defects

Fig. 3. FDH-like phenotypes in PorcnΔ/+ heterozygous embryos. (A–C)Comparing E17.5 PorcnΔ/+ embryos with WT reveals a range of phenotypesthat include cleft palate, tail hypoplasia, omphalocele, atrophic dermisthrough which the liver is visible, and tail/posterior axis truncation. cp, cleftpalate; de, atrophic dermis; om, omphalocele; tr, tail/posterior axis trunca-tion; tl, tail hypoplasia. (D–F) Alcian blue/alizarin red skeletal stains of WTand PorcnΔ/+ forelimbs. The arrow in F indicates lack of autopod, accom-panied by absence of ulna. (G–H) Alcian blue/nuclear fast red staining toreveal skeletal elements, including sternum, of E15.5 WT and PorcnΔ/+ ven-tral body walls sectioned at the level of the heart. he, heart; st, sternum. (Iand J) Immunostaining of sections semiadjacent to G and H (approximatepositions indicated by red boxes) for E-cadherin (red) and the dermis/mes-enchyme marker PDGF receptor-α (green). de, dermis; ep, epidermis. (Scalebars: G and H, 500 μm; I and J, 50 μm.)

12754 | www.pnas.org/cgi/doi/10.1073/pnas.1006437108 Barrott et al.

in ectodermal appendages, such as hair, teeth, and nails (20), thedevelopment of which requires canonical Wnt signaling (41, 42).Our efforts to study postnatal development of these tissues werefrustrated by perinatal lethality, because several litters producedonly a single viable but runted PorcnΔ/+ pup. This mouse exhibitedfocal hairlessness on its ventral skin but was otherwise WT inappearance (Fig. 4 A and B), suggesting that postnatal survivalselects for individuals at the mild end of a phenotypic spectrum.The first hair follicle primordia are detected at E14.5 and are

marked by Wnt-dependent expression of the TCF factor Lef1,which, in turn, is required for hair development (41, 43). Whole-mount in situ hybridizations on embryos of this stage revealeda decreased density of Lef1+ placodes in the dorsolateral skin ofPorcnΔ/+ embryos (Fig. 4 C and D). This phenotype was morepronounced at E17.5; at that stage, PorcnΔ/+ embryos exhibitedlarge patches of abnormally smooth, hair follicle-deficient epi-dermis (Fig. 4 E–G). In WT E17.5 embryos, Lef1 was stronglyexpressed throughout the basal epidermis as well as in nascenthair follicles (Fig. 4 H and J), and both of these expressiondomains were reduced or absent in hairless patches of Porcnheterozygotes (Fig. 4 I and K). Other basal markers, such askeratin-14 and p63, were expressed normally in these regions, aswas the suprabasal marker keratin-10, indicating normal kerati-nocyte differentiation (Figs. 4 L and M and Fig. S4). Thesephenotypes closely resemble those induced by overexpression of

Dkk1, a canonical Wnt inhibitor (41), suggesting that Porcnmediates the Wnt-dependent up-regulation of epidermal Lef1.

Tissue-Specific Porcn Deletion Implicates Ectodermal Wnt Defects inFDH Etiology. FDH has been speculatively attributed to defectivegene function in the ectoderm (44), a hypothesis that we havebegun to test by crossing Porcnlox/lox females to additional Cretransgenic males to produce tissue-specific KO male offspring.Our initial experiments focused on the limb, because of theavailability of Cre lines and well-characterized Wnt KO pheno-types (Table S1). Using a mesenchyme-specific Prx1-Cre driver(45), we obtained a dramatically shortened limb phenotype atE17.5, which almost perfectly reproduced that of Wnt5a null mice(37) (n = 10 of 10 Porcnlox/Y; Prx1-Cre embryos; Fig. 5 A andB). In addition to confirming that Porcn is required for a well-characterized “noncanonical” Wnt signaling process, this resultsuggests that the direct effects of Wnt5a on limb outgrowth aremediated by its mesenchymal expression domain rather than itsexpression in the overlying apical ectodermal ridge (AER) (37).Like Wnt5a nulls, Porcnlox/Y; Prx1-Cre mice exhibit loss of distaldigits but otherwise preserve all individual skeletal elements, andtherefore do not reproduce the syndactyly or truncation phe-notypes commonly seen in patients with FDH and in PorcnΔ/+

mice (Table S1).Fusion and loss of skeletal elements are observed whenWnt3 is

deleted in the hind-limb ectoderm with Msx2-Cre (46, 47). Basedon that finding, we used Msx2-Cre to ablate Porcn and observeda variably penetrant Wnt3-like phenotype at E18.5, including

Fig. 4. Focal absence of hair follicles in PorcnΔ/+ heterozygotes. (A and B)Ventral view of WT and PorcnΔ/+ littermates at weanling stage (postnatalday 22). (C–E) Whole-mount in situ hybridization for Lef1 (brown) onshoulder region of E14.5 control or PorcnΔ/+ embryos. The asterisk indicatesa fragment of skin accidentally removed during dissection. (E) Dorsal view ofan E17.5 PorcnΔ/+ embryo, indicating patches of unusually smooth skin(arrowheads). (F–M) Semiadjacent sections of dorsal skin from E17.5 WT orPorcnΔ/+ embryos, stained with H&E or immunostained for Lef1 or p63. Thebrackets in I indicate patches devoid of hair follicles and expressing low or noLef1. hf, hair follicle. (Scale bars: 100 μm.)

Fig. 5. Tissue-specific Porcn deletion phenotypes. (A and B) Alcian blue/alizarin red-stained forelimbs of E17.5 control or Porcnlox/Y; Prx1-Cre(Prx1KO) embryos. Mutant embryos exhibit shortening of all skeletal ele-ments and loss of distal digits. hu, humerus; ra, radius; sc, scapula; ul, ulna.(C–F) Hind-limb skeleton preparations of E18.5 control or Porcnlox/Y; Msx2-Cre (Msx2KO) embryos and whole-mount in situ hybridization for the AERmarker Fgf8 on E11.5 hind limbs. Note the almost complete absence ofautopod in E18.5 mutant and focal loss of Fgf8 expression at E11.5. (G and H)H&E-stained sections of the ventral body wall from E18.5 control or Msx2KOembryos. The brackets in H indicate an area of severe dermal thinning inmutant body wall, such that liver almost directly abuts surface ectoderm. li,liver. (I and J) Lef1 immunostaining of dorsal skin from E18.5 control orMsx2KO embryos, revealing extensive domain of hairless Lef1-devoid epi-dermis in mutant (arrowhead in J indicates isolated patch of Lef1+ basalcells). (Scale bars: G and H, 500 μm; I and J, 100 μm.)

Barrott et al. PNAS | August 2, 2011 | vol. 108 | no. 31 | 12755

DEV

ELOPM

ENTA

LBIOLO

GY

syndactyly and autopod truncation (n = 6 of 14 Porcnlox/Y; Msx2-Cre hind limbs; Fig. 5C andE). Similar variability is observed withWnt3 ablation because of inefficient deletion across the dorsal-ventral boundary of the distal limb ectoderm (46). As with Wnt3deletion, the variable skeletal phenotype of Porcn deletion wasprefigured bymosaic AERdefects in the hind limb at E11.5 (Fig. 5D and F). In addition to the limb, Msx2-Cre drives mosaic ecto-dermal recombination throughout the trunk (48), where we ob-served focal skin and dermal defects at E18.5 that reproducedthose of PorcnΔ/+ mice (n= 3 of 3 Porcnlox/Y; Msx2-Cre embryos;Fig. 5 G–J). Skin-specific deletion therefore appears to be suffi-cient to account for many of the structural and differentiationdefects of human and mouse Porcn heterozygotes, supportinga major role for this gene in the ectoderm.

DiscussionIn the past 2 decades, homologs of nearly every developmentalgene identified in Drosophila have been mutated in the mouse,with porcupine representing an unusual exception. We have nowgenerated a conditional mutant allele of its unique mouseortholog, Porcn, for which our studies indicate a conserved rolein Wnt ligand secretion (7, 8). Our genetic analyses confirm andextend previous studies, using RNAi or small molecules to knockdown Porcn function in vitro (10–12), and indicate that Porcnlox

represents a potent tool to probe Wnt ligand function in vivo.This study does not address the scope and specificity of Porcn

function: Is it required for all 19 mammalian Wnts, and does itregulate any other ligands? With respect to the first question, allmammalian Wnts contain the conserved serine residue that is thesubstrate for Porcn palmitoylation in Wnt3a (10). DrosophilaWntD, an orphan family member, is the only Wnt that lacks thisresidue and the only Wnt that has been shown to be secreted inthe absence of porcupine (49). PorcnΔ cells can be used directlyto test the role of Porcn in the biogenesis of diverse Wnt ligands,as well as to address open questions regarding additional Wntmodifications and their involvement in ligand trafficking andactivity (6). We were surprised to find that although Wnt3a-HAsecretion is abolished in PorcnΔ cells, its acylation is only slightlyreduced. Prior studies have identified multiple sites of Wnt3aacylation (10, 24), and we speculate that loss of Porcn arrestsligand trafficking in MEFs only after completion of a Porcn-independent acylation step. PorcnΔ cells may provide a usefulsystem to identify the site, nature, and biological relevance of thisposttranslational modification.With respect to Porcn substrates beyond Wnts, it has been

shown that porcupine is dispensable for hedgehog and bonemorphogenetic protein (BMP) signaling inDrosophila (8), and wefind that limb mesenchyme-specific deletion of Porcn causes de-fective outgrowth, mimicking loss ofWnt5a (37), without ablationof specific skeletal elements as seen in mesenchyme-specific Shhor Bmp2/Bmp4mutants (50, 51). Future studies, in PorcnΔ cells aswell as tissue-specific KOs, will address the possible requirementfor this gene in other pathways.The Porcn heterozygous phenotype closely resembles human

FDH, particularly its more severe manifestations (20). Studies ofinheritance and X-inactivation in human patients have suggestedthat only the least-affected patients with FDH survive beyondbirth (19), and our data strongly support this model: Most

PorcnΔ/+ mice appear to die perinatally, and the only postnatalsurvivor yet obtained had an extremely mild phenotype com-pared with embryos. As in FDH, the defects in PorcnΔ/+ embryosare typically discrete and asymmetrical, presumably reflectingdomains of Porcn-deficient cells established by X-inactivation.The characteristic skin defects of patients with FDH frequentlyfollow the so-called “lines of Blaschko,” stripes and whorls hy-pothesized to represent the clonal descendants of embryonicectoderm progenitors (20). In turn, this observation suggests thatFDH results from loss of PORCN function in the ectoderm, withnon-cell-autonomous effects on underlying mesodermal cells(44). Our tissue-specific deletion studies provide experimentalsupport for this hypothesis and, in addition to illuminating theetiology of FDH, exemplify the utility of Porcnlox as a tool todissect Wnt signaling in vivo. Given that Porcn is X-linked, tissue-specific KO males can be obtained in a single generation bycrosses between Cre-transgenic males and Porcnlox females. Itshould therefore be facile and economical to test the role ofPorcn-dependent Wnt signals in any tissue for which an appro-priate Cre driver is available.

Materials and MethodsDetailed methods are provided in SI Materials and Methods. In brief, thePorcnneo2lox targeting vector (Fig. S1A) was constructed via recombineering(52, 53), and ES cells were electroporated, selected, and analyzed usingstandard techniques (54). ES cells used to generate mice were culturedcontinuously on MEF feeders, whereas ES cells used for in vitro experimentswere grown under feeder-free conditions, in the presence of serum andleukemia inhibitory factor (LIF) (54). A polyclonal rabbit anti-Porcn antise-rum was generated by Covance against a C-terminal peptide epitope(TEEKDHLEWDLTVSR, encoded by exons 9 and 10). Wnt reporter gene assaysused the pSuper8 × TOPFlash and pSuper8 × FOPFlash plasmids (23), as wellas the 7TFP lentiviral reporter construct (33). Triton X-114 partitioning assayswere performed essentially as described (12, 55). Chimeric mice were pro-duced by the University of Utah Transgenic Core Facility, via ES cell injectioninto C57BL/6 blastocysts. In all experiments, mutant embryos were com-pared with littermate controls retaining full Porcn coding regions (i.e.,WT or floxed) on all alleles. Sox2-Cre (30) and Prx1-Cre mice (45) wereobtained from The Jackson Laboratory, and Msx2-Cre mice (47) were pro-vided by Mark Lewandoski (National Cancer Institute, Frederick, MD). Allanimal experiments were performed according to protocols approved byinstitutional committees of the University of Utah and Brigham YoungUniversity.

Note Added in Proof. While this paper was in press, Biechele et al. (56)published a study characterizing ES cells harboring a Porcn gene trap loss-of-function allele, and demonstrated defects in Wnt production and mesodermdifferentiation very similar to those described here.

ACKNOWLEDGMENTS. We thank the following individuals for reagents,equipment, and advice: Mario Capecchi, Mike Howard, Kristen Kwan, KirkThomas, Yukio Saijoh, Sabine Fuhrmann, Diane Ward, Aubrey Chan, NealCopeland, Mark Lewandoski, and Randall Moon. We are particularly grate-ful to Susan Tamowski for deriving germ-line-chimeric mice. We thank SuziMansour, Daniel Kopinke, and Kristen Kwan for comments on the manu-script. L.C.M. thanks Norbert Perrimon for originally drawing his interest toporcupine. This work was supported by Grant 06-B-116 from the SearleScholars Foundation (to L.C.M.), Grant R01-DK075072 from the NationalInstitutes of Health (to L.C.M.), the University of Utah Seed Grant Program(to L.C.M.), and Grant R15-HD060087 from the National Institutes of Health(to J.R.B.).

1. Logan CY, Nusse R (2004) The Wnt signaling pathway in development and disease.Annu Rev Cell Dev Biol 20:781–810.

2. MacDonald BT, Tamai K, He X (2009) Wnt/beta-catenin signaling: Components,mechanisms, and diseases. Dev Cell 17:9–26.

3. Grigoryan T, Wend P, Klaus A, Birchmeier W (2008) Deciphering the function ofcanonical Wnt signals in development and disease: Conditional loss- and gain-of-function mutations of beta-catenin in mice. Genes Dev 22:2308–2341.

4. Veeman MT, Axelrod JD, Moon RT (2003) A second canon. Functions and mechanismsof beta-catenin-independent Wnt signaling. Dev Cell 5:367–377.

5. NelsonWJ, Nusse R (2004) Convergence of Wnt, beta-catenin, and cadherin pathways.Science 303:1483–1487.

6. Hausmann G, Bänziger C, Basler K (2007) Helping Wingless take flight: How WNTproteins are secreted. Nat Rev Mol Cell Biol 8:331–336.

7. van den Heuvel M, Harryman-Samos C, Klingensmith J, Perrimon N, Nusse R (1993)Mutations in the segment polarity genes wingless and porcupine impair secretion ofthe wingless protein. EMBO J 12:5293–5302.

8. Kadowaki T, Wilder E, Klingensmith J, Zachary K, Perrimon N (1996) The segmentpolarity gene porcupine encodes a putative multitransmembrane protein involved inWingless processing. Genes Dev 10:3116–3128.

9. Tanaka K, Okabayashi K, Asashima M, Perrimon N, Kadowaki T (2000) The evolu-tionarily conserved porcupine gene family is involved in the processing of the Wntfamily. Eur J Biochem 267:4300–4311.

12756 | www.pnas.org/cgi/doi/10.1073/pnas.1006437108 Barrott et al.

10. Takada R, et al. (2006) Monounsaturated fatty acid modification of Wnt protein: Itsrole in Wnt secretion. Dev Cell 11:791–801.

11. Galli LM, Barnes TL, Secrest SS, Kadowaki T, Burrus LW (2007) Porcupine-mediatedlipid-modification regulates the activity and distribution of Wnt proteins in the chickneural tube. Development 134:3339–3348.

12. Chen B, et al. (2009) Small molecule-mediated disruption of Wnt-dependent signalingin tissue regeneration and cancer. Nat Chem Biol 5:100–107.

13. Hofmann K (2000) A superfamily of membrane-bound O-acyltransferases with im-plications for wnt signaling. Trends Biochem Sci 25:111–112.

14. Buglino JA, Resh MD (2008) Hhat is a palmitoylacyltransferase with specificity forN-palmitoylation of Sonic Hedgehog. J Biol Chem 283:22076–22088.

15. Gutierrez JA, et al. (2008) Ghrelin octanoylation mediated by an orphan lipid trans-ferase. Proc Natl Acad Sci USA 105:6320–6325.

16. Yang J, Brown MS, Liang G, Grishin NV, Goldstein JL (2008) Identification of theacyltransferase that octanoylates ghrelin, an appetite-stimulating peptide hormone.Cell 132:387–396.

17. Grzeschik KH, et al. (2007) Deficiency of PORCN, a regulator of Wnt signaling, is as-sociated with focal dermal hypoplasia. Nat Genet 39:833–835.

18. Wang X, et al. (2007) Mutations in X-linked PORCN, a putative regulator of Wntsignaling, cause focal dermal hypoplasia. Nat Genet 39:836–838.

19. Bornholdt D, et al. (2009) PORCN mutations in focal dermal hypoplasia: Coping withlethality. Hum Mutat 30:E618–E628.

20. Temple IK, MacDowall P, Baraitser M, Atherton DJ (1990) Focal dermal hypoplasia(Goltz syndrome). J Med Genet 27:180–187.

21. Schwartzberg PL, Goff SP, Robertson EJ (1989) Germ-line transmission of a c-ablmutation produced by targeted gene disruption in ES cells. Science 246:799–803.

22. Korinek V, et al. (1997) Constitutive transcriptional activation by a beta-catenin-Tcfcomplex in APC-/- colon carcinoma. Science 275:1784–1787.

23. Veeman MT, Slusarski DC, Kaykas A, Louie SH, Moon RT (2003) Zebrafish prickle,a modulator of noncanonical Wnt/Fz signaling, regulates gastrulation movements.Curr Biol 13:680–685.

24. Willert K, et al. (2003) Wnt proteins are lipid-modified and can act as stem cell growthfactors. Nature 423:448–452.

25. Haegel H, et al. (1995) Lack of beta-catenin affects mouse development at gastrula-tion. Development 121:3529–3537.

26. Liu P, et al. (1999) Requirement for Wnt3 in vertebrate axis formation. Nat Genet 22:361–365.

27. Barrow JR, et al. (2007) Wnt3 signaling in the epiblast is required for proper orien-tation of the anteroposterior axis. Dev Biol 312:312–320.

28. Lindsley RC, Gill JG, Kyba M, Murphy TL, Murphy KM (2006) Canonical Wnt signaling isrequired for development of embryonic stem cell-derived mesoderm. Development133:3787–3796.

29. ten Berge D, et al. (2008) Wnt signaling mediates self-organization and axis forma-tion in embryoid bodies. Cell Stem Cell 3:508–518.

30. Hayashi S, Lewis P, Pevny L, McMahon AP (2002) Efficient gene modulation in mouseepiblast using a Sox2Cre transgenic mouse strain. Gene Expr Patterns 2:93–97.

31. Cox BJ, et al. (2010) Phenotypic annotation of the mouse X chromosome. Genome Res20:1154–1164.

32. Shimizu H, et al. (1997) Transformation by Wnt family proteins correlates with reg-ulation of beta-catenin. Cell Growth Differ 8:1349–1358.

33. Fuerer C, Nusse R (2010) Lentiviral vectors to probe and manipulate the Wnt signalingpathway. PLoS ONE 5:e9370.

34. Yoon JC, et al. (2010) Wnt signaling regulates mitochondrial physiology and insulinsensitivity. Genes Dev 24:1507–1518.

35. Ohtola J, et al. (2008) beta-Catenin has sequential roles in the survival and specifi-cation of ventral dermis. Development 135:2321–2329.

36. Hancock S, et al. (2002) Probable identity of Goltz syndrome and Van Allen-Myhresyndrome: Evidence from phenotypic evolution. Am J Med Genet 110:370–379.

37. Yamaguchi TP, Bradley A, McMahon AP, Jones S (1999) A Wnt5a pathway underliesoutgrowth of multiple structures in the vertebrate embryo. Development 126:1211–1223.

38. Greco TL, et al. (1996) Analysis of the vestigial tail mutation demonstrates that Wnt-3a gene dosage regulates mouse axial development. Genes Dev 10:313–324.

39. Kokubu C, et al. (2004) Skeletal defects in ringelschwanz mutant mice reveal that Lrp6is required for proper somitogenesis and osteogenesis. Development 131:5469–5480.

40. Etheridge SL, et al. (2008) Murine dishevelled 3 functions in redundant pathways withdishevelled 1 and 2 in normal cardiac outflow tract, cochlea, and neural tube de-velopment. PLoS Genet 4:e1000259.

41. Andl T, Reddy ST, Gaddapara T, Millar SE (2002) WNT signals are required for theinitiation of hair follicle development. Dev Cell 2:643–653.

42. Widelitz RB (2008) Wnt signaling in skin organogenesis. Organogenesis 4:123–133.43. van Genderen C, et al. (1994) Development of several organs that require inductive

epithelial-mesenchymal interactions is impaired in LEF-1-deficient mice. Genes Dev 8:2691–2703.

44. Moss C (1999) Cytogenetic and molecular evidence for cutaneous mosaicism: Theectodermal origin of Blaschko lines. Am J Med Genet 85:330–333.

45. Logan M, et al. (2002) Expression of Cre Recombinase in the developing mouse limbbud driven by a Prxl enhancer. Genesis 33:77–80.

46. Barrow JR, et al. (2003) Ectodermal Wnt3/beta-catenin signaling is required for theestablishment and maintenance of the apical ectodermal ridge. Genes Dev 17:394–409.

47. Sun X, et al. (2000) Conditional inactivation of Fgf4 reveals complexity of signallingduring limb bud development. Nat Genet 25:83–86.

48. Pan Y, et al. (2004) gamma-secretase functions through Notch signaling to maintainskin appendages but is not required for their patterning or initial morphogenesis.Dev Cell 7:731–743.

49. Ching W, Hang HC, Nusse R (2008) Lipid-independent secretion of a Drosophila Wntprotein. J Biol Chem 283:17092–17098.

50. Bandyopadhyay A, et al. (2006) Genetic analysis of the roles of BMP2, BMP4, andBMP7 in limb patterning and skeletogenesis. PLoS Genet 2:e216.

51. Scherz PJ, McGlinn E, Nissim S, Tabin CJ (2007) Extended exposure to Sonic hedgehogis required for patterning the posterior digits of the vertebrate limb. Dev Biol 308:343–354.

52. Wu S, Ying G, Wu Q, Capecchi MR (2008) A protocol for constructing gene targetingvectors: Generating knockout mice for the cadherin family and beyond. Nat Protoc 3:1056–1076.

53. Liu P, Jenkins NA, Copeland NG (2003) A highly efficient recombineering-basedmethod for generating conditional knockout mutations. Genome Res 13:476–484.

54. Nagy A, Gertsenstein M, Vintersten K, Behringer R (2003) Manipulating the MouseEmbryo: A Laboratory Manual (Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY), 3rd Ed, p x.

55. Bordier C (1981) Phase separation of integral membrane proteins in Triton X-114solution. J Biol Chem 256:1604–1607.

56. Biechele S, Cox BJ, Rossant J (2011) Porcupine homolog is required for canonical Wntsignaling and gastrulation in mouse embryos. Dev Biol 355:275–285.

Barrott et al. PNAS | August 2, 2011 | vol. 108 | no. 31 | 12757

DEV

ELOPM

ENTA

LBIOLO

GY