Null and hypomorph Prickle1 alleles in mice phenocopy human Robinow syndrome and disrupt signaling downstream of Wnt5a

Embryonic development utilizes a network of transcription regulatory factors and signaling pathways to generate diverse cell types that are organized into functional entities (e.g., tissues). Within each tissue, individual cells require further instructions to establish an ordered architecture by acquiring specific polarity, such as front-rear polarity for migrating mesenchymal cells and apicobasal (AB) and planar cell polarity (PCP) for epithelial cells. Correct establishment of cell polarity is essential for organogenesis, tissue functions and survival.

PCP signaling provides vectorial information for cell arrangement and migration within a large field. First identified in Drosophila, the PCP module consists of six core components: Frizzled, Van Gogh/Strabismus, Dishevelled, Starry Night/Flamingo, Prickle, and Diego (Jenny and Mlodzik, 2006; Jenny et al., 2005; Seifert and Mlodzik, 2007; Wang and Nathans, 2007). In Drosophila wing epithelia, these proteins are segregated into two clusters within the apical domain, with Frizzled-Dishevelled-Diego-Flamingo localized distally, and Prickle-Vangl-Flamingo confined proximally (Jenny et al., 2005; Peng and Axelrod, 2012). A similar pattern is repeated in the neighboring cells and propagated throughout the morphogenetic field, offering directional information for hair positioning through cytoskeleton remodeling. The photoreceptors in Drosophila ommatidia also employ this protein segregation mechanism to facilitate local transmission of planar polarity information (Fanto and McNeill, 2004; Seifert and Mlodzik, 2007; Wang and Nathans, 2007; Zheng et al., 1995). Genetic disruption of PCP components in Drosophila leads to orientation defects in wing hair and ommatidia (Seifert and Mlodzik, 2007).

The PCP signaling pathway is evolutionally conserved. Each of the PCP components characterized in Drosophila has its homolog in vertebrates, and protein segregation and distribution pattern observed in Drosophila hair cells are also retained in some vertebrate epithelia (Deans et al., 2007; Vladar et al., 2012). Gene knockdown of PCP components in lower vertebrates results in defects in convergent-extension, ciliogenesis and neural tube closure (Wallingford, 2006; Wallingford et al., 2002; Wallingford and Harland, 2002). Disruption of PCP component genes in the mouse cause similar but more extensive tissue phenotypes, which include looped tail, abnormally rotated stereociliary bundles of the inner ear hair cells, and misoriented skin hair (Curtin et al., 2003; Hamblet et al., 2002; Montcouquiol et al., 2003; Tissir et al., 2010; Wang et al., 2006; Watanabe et al., 2003). Heart defects (Etheridge et al., 2008; Hamblet et al., 2002; Sinha et al., 2012; Yu et al., 2010) and kidney malformation (Simons et al., 2005; Tissir et al., 2010; Yates et al., 2010) have also been reported in PCP mutant mice.

Among the six core PCP components, the function of Prickle has remained largely elusive in mammals. The mouse genome has four Prickle homologs, with Prickle1 and Prickle2 being most closely related to their Drosophila prototype, whereas Prickle3 and Prickle4 are fairly distant. Prickle1 and Prickle2 were reported to be essential for AB polarity of embryonic epiblasts since ablation of either in mice resulted in early embryonic lethality prior to gastrulation (Tao et al., 2012; Tao et al., 2009). An ENU-induced frameshift mutation in mouse Prickle1 resulted in limb defects and death shortly after birth (Yang et al., 2013), suggesting a yet to be explored late-stage Prickle1 function.

We recently demonstrated that Prickle1 is expressed in the mammalian central nervous system and plays an important role in neuronal morphogenesis (Liu et al., 2013). To further explore a PCP-related function of Prickle1 at both early and late stages of development, we engineered three Prickle1 mutant alleles in mice. Combinations of these with select Cre-drivers were predicted to generate varied gene-dosage in different tissues and bypass early embryonic lethality observed in Prickle1 null mutants (Tao et al., 2009). Here, we report global and pleiotropic PCP defects in Prickle1 null and hypomorphic mutants that closely resemble Wnt5a, Ror or Ryk mutant mice and recapitulate features of human Robinow syndrome (RS; RRS, MIM#268310, http://omim.org/entry/268310; DRS, MIM#180700, http://omim.org/entry/180700), prompting us to explore the link between Prickle1 and the Wnt pathway. Our studies implicate Prickle1 in regulating Wnt/PCP pathway and provide novel insights into RS disease mechanism.

We generated three mutant Prickle1 alleles in mice by targeting exon 2 with the starting ATG: a gene-trap allele (Prickle1^a), a null allele (Prickle1^b), and a conditional allele (Prickle1^c) (Fig. 1A). The gene-trap allele was predicted to be a hypomorph and produce a trace amount of protein, whereas Prickle1^b and Cre-excised Prickle1^c alleles would be null systemically or in targeted tissues. We first verified that Prickle1 protein was abolished in Prickle1^b/b mutant by immunoblot analysis using two independent antibodies against different domains of Prickle1 (Fig. 1B). In contrast to the reported Prickle1 null mutant that exhibited embryonic lethality prior to gastrulation (Tao et al., 2009), the null mutants (Prickle1^b/b) we produced survived until postnatal day 2 (P2). The Prickle1^c/c conditional deletion generated by a broadly expressed Sox-cre, which effectively represented a null genotype in most tissues, survived for about 2 days longer. The morphological defects of Prickle1^b/b mutant mice were seen as early as embryonic day (E) 15.5 and become more evident at P0 with shortened limbs and snout (supplementary material Fig. S1A,B). A close view of fore- and hind- paws showed blunted digit tips in the mutants (supplementary material Fig. S1C,D). Although the mutant mice are generally slightly smaller at P0, shortening of limb and snout are more pronounced than shortening of the body axis (supplementary material Fig. S1E,F). Additional phenotypes observed at P0 included open eyelid (supplementary material Fig. S2A), bifid and truncated tongue (supplementary material Fig. S2B), cleft palate (supplementary material Fig. S2C, >17 out of 31), curly tail (supplementary material Fig. S2D) and hair orientation defect (supplementary material Fig. S2E,F). Similar phenotypes were seen in Sox2-Cre conditional mutants (supplementary material Fig. S2E,G). Skeletal staining with alcian blue and alizarin red (Fig. 1Ca–e) revealed shortened nasal, premaxillary and dentary bones (Fig. 1Ca), misaligned sternebrae (Fig. 1Cb, 19 out of 31), shorter but thicker long bones (Fig. 1Cc, upper: scapula and humerus; lower: radius and ulna) and absent or delayed ossification centers of phalangeal bones (Fig. 1Cd). Extensive internal organ defects were observed some of which will be described below (and data not shown).

Malformation of the heart, a PCP target organ, causes postnatal death in many instances. In Prickle1 mutants, heart chambers and outflow tracts were developed at E14.5 (Fig. 2A). The aorta (Ao) and pulmonary arteries (Pa), however, were transposed in more than half of the mutant embryos (19 out of 31). Similar phenotypes were observed in newborn (P0) mutant pups, with significantly enlarged right atrium (Fig. 2A, lower panels; 2B, upper panels) and juxtaposed Ao and Pa (Fig. 2B). The atrial septum was also open (Fig. 2C). At the cellular level, H&E staining showed round mutant cardiomyocyte nuclei in contrast to the spindle-shaped nuclei of the control, indicating the loss of polarity (Fig. 2D). Actin filament in myofibrils were disorganized and misaligned (Fig. 2E,F). A significant increase of randomness in fiber orientation was also noticeable in the mutants (Fig. 2F,G), a novel feature that has not been reported in other PCP mutants.

We next examined the kidney, another PCP target organ in which branching and tubule diameter are affected in Vangl2 (Yates et al., 2010) and Wnt9b (Karner et al., 2009) mutant mice, respectively. Cystic kidney has also been observed in Robinow syndrome patients with mutations in Ror2 (Bacino, 1993), a receptor for Wnt5a (Ho et al., 2012). Kidney size in the Prickle1 mutants was comparable to that of the control. There were frequently blood spots on the surface (supplementary material Fig. S3A, 24 out of 30 pulps). The development of renal tubules was largely normal judged by calbindin-D28K labeled ureteric bud (UB) and E-cadherin labeling (supplementary material Fig. S3B). Overt cystic kidneys were observed at birth albeit at very low penetrance (<5%, 3 out of 61) (supplementary material Fig. S3C). H&E staining showed dilated renal tubules in the mutants (Fig. 3A). Lotus tetragonolobus lectin (LTL)-labeled proximal convoluted tubules (PCT) appeared to be distorted, with narrowing of PCT in the outer renal cortex but widening of the inner renal cortex (Fig. 3B). The collecting ducts (CD) of the renal papilla were barely discernable by DAPI staining alone (compare Fig. 3C,D). E-cadherin staining showed irregular elliptical shape of the mutant collecting ducts (Fig. 3C). With simple quantification (Fig. 4A–D), we detected a significant increase in displaced/disarranged cells of the mutant lumen as revealed by E-cadherin and DAPI staining (off plane cells in Fig. 4A,B, arrowheads). No change was obvious in the collecting duct density per unit area (Fig. 4E). Quantification of cell-cell connections from a longitudinal view of collecting ducts (supplementary material Fig. S4A) revealed increased 4-lateral intersections of the mutants (supplementary material Fig. S4B).

To further investigate whether the kidney tubules were generally disorganized, we quantified cell alignment defects in the ascending Henle's loop, which has different origin from the collecting ducts during nephron formation. The thick ascending limb of the Henle's loop was easily identified by E-cadherin staining exhibiting a large plane of cuboidal epithelial cells (Fig. 4F). Cell-cell intersection pattern was significantly altered in the mutant with increased 4-lateral intersections in contrast to predominantly 3-lateral intersections in the control (Fig. 4F,G). Most cuboidal epithelial cells in the control were convex hexagons, but appeared irregular in the mutants with many more quadrilaterals and pentagons (Fig. 4H,I). The altered intersection mode and shape of the mutant epithelial cells were also accompanied by aberrant cell alignment. To quantify this feature, we first define a shortest compound line path between two adjacent columns of cells along the vertical direction of a tubule (supplementary material Fig. S4C,D). We then performed vectorial additions for laterals from each cell vertex on the compound line (supplementary material Fig. S4D) and acquired collections of new vectors. In a plotted polar coordinate (supplementary material Fig. S4E), significantly increased vector variations were observed in the mutants by paired t-test analysis (supplementary material Fig. S4E, P = 2.4E-07, Length: P = 2.02E-06). The vectorial variations in the mutants may indicate a decreased tubule stability and vulnerability to distortion.

Open neural tube is a typical PCP defect identified in Fz, Dvl, Celsr and Vangl mutants (Curtin et al., 2003; Etheridge et al., 2008; Murdoch et al., 2001; Wang et al., 2006). In Prickle1 mutants, the neural tube was largely normal with a slight delay in closure at E10.5 (Fig. 5A, 8 out of 15). We observed brain and axon outgrowth defect as early as E10.5 (3 out of 4). Neural filament staining indicated the failure of innervation of trigeminal (V) and facial (VII) nerves (Fig. 5B). At P0, H&E staining showed enlarged lateral ventricles and misshaped hippocampus (Fig. 5C). The thalamic-cortical axonal tract was also disorganized in the Prickle1 mutants (Fig. 5D, 3 out of 5). However, scanning electron microscopy (SEM) did not reveal stereocilia rotation defects in the inner ear hair cells that are typically observed in other PCP mutants. Rather, the stereociliary actin bundles of the Prickle1 mutant hair cells were shorter and splayed at the tip (Fig. 5E,F, 3 out of 3), consistent with the actin filament defects identified in the heart.

Since PCP components were distributed asymmetrically, we asked whether the localization of other PCP proteins, such as Dvl and Vangl, was perturbed in the Prickle1 mutants. Dvl1-3 and Vangl2 are concentrated at the apical domain of the control ureteric bud cells (Fig. 6A–D, left panels), yet a significant fraction of these was mislocalized to the lateral and basal aspects of the epithelia in the mutants (Fig. 6A–D, right panels). The results are consistent with an earlier report of disrupted apicobasal polarity in Prickle1 mutant embryonic epiblasts (Tao et al., 2009). Actin assembly is a downstream event of PCP, which involves small GTPases (Schlessinger et al., 2009). We found that, rather than apical enrichment, the actin distribution was shifted more laterally in the mutant ureteric buds (Fig. 6E). Consistent with this observation, RhoA also decreased its apical enrichment (Fig. 6F; supplementary material Fig. S5A). Rac1 and Cdc42 are important for lamellilepodia and filopodia formation, respectively, in many types of cultured cells. Interestingly, they are ectopically accumulated in many cells of the mutant ureteric bud and other types of tubules (Fig. 6G,H and supplementary material Fig. S5B,C arrows). Their misexpression/deposition could indicate significantly increased variations in cell motility and/or local rearrangement, either of which could underlie the distorted tubular shapes seen in the mutants.

To understand the cause of limb outgrowth defects in the mutants, we examined two polarity proteins, Vangl2 and Dvl2, which were reported to mediate Wnt5a signaling crucial for limb outgrowth (Gao et al., 2011; Ho et al., 2012). Vangl2 localizes asymmetrically in chondrocytes at E12.5 (Gao et al., 2011). We found a similar, proximally enriched pattern of Vangl2 in the control chondrocytes but this asymmetry was diminished in the mutant (Fig. 7A; supplementary material Fig. S6A and Fig. S7A). Asymmetrical distribution of Vangl2 protein was also observed at E14.5 in the circumferential areas of the bone ossification center of the control, but the localization pattern was broadened in the mutants (Fig. 7B; supplementary material Fig. S6B), and mutant cells with polarized Vangl2 were reduced (supplementary material Fig. S7B). Further analysis demonstrated a reduction of alignment of the mutant cells with Vangl2 staining (supplementary material Fig. S7C). Although Vangl2 localization was reported to polarize along proximal-distal axis in general, it appeared to polarize radially surrounding the ossification center (supplementary material Fig. S6B). Dvl2 polarization was not detected at E12.5 in the control digit chondrocytes (data not shown) but emerged at E14.5 appearing as a distally enriched pattern and the opposite of Vangl2 (Fig. 7C and supplementary material Fig. S6C, left panel). This polarization was completely lost in the mutant (Fig. 7C and supplementary material Fig. S6C, right panel). These results indicated that similar to the kidney tubules, the cell polarity machinery in chondrocytes was also perturbed, which could explain the shortened limb phenotype.

To explore the relationship between Prickle1 and Wnt5a signaling, we first turned to in vitro assays in cultured HEK293 cells. We found that ubiquitination of a GFP-tagged Prickle1 was enhanced by Wnt5a and Ryk in presence of MG132, a proteasome inhibitor (Fig. 8A, compare lane2 with lane1). As expected, accumulation of ubiquitinated Prickle1 was less significant without MG132 treatment (Fig. 8A, compare lanes 2 and 3), presumably due to its rapid proteasomal degradation. These results suggest a negative regulation of Prickle1 by Wnt5a signaling. Accordingly, increased Prickle1 accumulation was observed in Wnt5a null mutant tissues (Fig. 8B). Wnt5a enhances Dvl2 phosphorylation in several cell lines and in embryonic fibroblasts as well (González-Sancho et al., 2004; Ho et al., 2012) (Fig. 8C). We therefore examined Dvl2 expression in Prickle1 mutants. A disproportional increase of phosphorylated and shifted form of Dvl2 (ps-Dvl2) was detected in both mutant kidney and distal paw tissues at P0 (Fig. 8D), suggesting that loss of Prickle1 potentiates Wnt5a signaling. To further corroborate that Dvl2 phosphorylation regulated through Wnt5a-Prickle1 is a critical event during limb outgrowth, we examined Dvl2 phosphorylation status in Wnt5a mutant distal paws. A decreased Dvl2 phosphrylation was observed (Fig. 8E), consistent with the finding that Wnt5a signaling degrades Prickle1. Taken together (Fig. 8F), our data indicate that Prickle1 mediates Wnt5a signaling to modulate Dvl2 phosphorylation/activation; the latter regulates actin assembly, a process affecting multiple aspects of tissue morphogenesis including cell shape, migration and arrangement.

Previous studies (Afzal et al., 2000a; Afzal et al., 2000b; Ali et al., 2007; Butler and Wadlington, 1987; Patton and Afzal, 2002) suggested that the tissue outgrowth and craniofacial defects in Prickle1 mutants were reminiscent of human Robinow syndrome (RS). Since many RS patients have normal life expectancy, we examined whether Prickle1 hypomorph mutants would survive to adulthood and display RS-like features. We generated mice with gene-trap (Prickle1^a/a) alleles alone and with combined gene-trap and germline knockout (Prickle1^a/b) alleles. Both Prickle1^a/a and Prickle1^a/b hypomorphs were born at the expected Mendelian ratio and survived to adulthood, but developed shorter stature (Fig. 9A) and craniofacial defects with wide spaced eyes, flat nose, short snout (Fig. 9B), prominent forehead (Fig. 9C) (30 out of 30), abnormal appearance of eyelid and eyelashes (supplementary material Fig. S8A–D, 15 out of 15), and fused mandibular incisors (Fig. 9D, 14 out of 30). Computed tomography (CT) scans detected incomplete and distorted vertebrae formation in the mutant spinal cord (Fig. 9E, 3 out of 3). Significantly shortened long bones (Fig. 9F, Femurs) and digital bones were consistently observed (Fig. 9G, metatarsals, 3 out of 3). Thus Prickle1 hypomorphs recapitulate most phenotypic features of human RS patients (Butler and Wadlington, 1987; Patton and Afzal, 2002). Our findings uncover an unexpected aspect of RS pathophysiology and suggest that loss-of-function mutations in human PRICKLE1 might also be a cause of this condition.

Organization of similar and/or different cell types throughout a large field is crucial for systemic functions. PCP is among the best-known signaling pathway managing large-scale cell organization. In mammals, disruption of different homologues of insect core-PCP proteins suggest variable functions of PCP components in distinct cellular processes during embryonic development. For example, disruption of Frizzleds (Wang et al., 2006), Dvls (Etheridge et al., 2008), Celsr1(Curtin et al., 2003) and Vangl2 (Kibar et al., 2001) exhibit some phenotypic commonalities including open neural tube and misoriented inner ear stereociliary bundles, which are considered classic PCP phenotypes. Disruption of Diego/Inversin lead to left-right asymmetry defects without significant classic PCP phenotypes (Morgan et al., 1998; Simons et al., 2005; Watanabe et al., 2003).

Wnts are suggested to be functionally linked to the activity of core-PCP components regardless of whether these are part of or parallel to the core-PCP. Wnts have been demonstrated in flies and lower vertebrates as critical morphogens for the core PCP complex (Takeuchi et al., 2003; Veeman et al., 2003; Wu et al., 2013). However, disruption of Wnt5a, a known PCP-related morphogen in mammals, gives rise to characteristic shortening of distal tissue axes (Gao et al., 2011; Ho et al., 2012; Qian et al., 2007), which is barely seen in other PCP mutants. Furthermore, mutations in either Wnt5a or its receptor Ror2 in humans lead to Robinow syndrome, distinct from that of other PCP gene mutations (Kibar et al., 2007; Murdoch et al., 2001; Wang et al., 2011). In the present study, we show that null and hypomorph Prickle1 mutations in mice manifest clinical features of human RS caused by mutations of Wnt5a pathway. Wnt5a regulates Prickle1 degradation to signal Dvl2 phosphorylation further suggest that Prickle1, unlike other identified core PCP component, is dedicated to Wnt pathway, which is particularly important for dynamic cell arrangement and migration during tissue outgrowth. The phenotypic similarity and its role in mediating Wnt5a/Ror2 signaling make Prickle1 a likely candidate gene for Robinow syndrome.

We initially generated different Prickle1 alleles in order to bypass early embryonic lethality reported previously in a Prickle1 null mutant (Tao et al., 2009). Our goal was to study Prickle1 late-stage PCP function, which cannot be addressed using conventional knockout mice. Surprisingly, the Prickle1 null embryos in our colonies survived through gastrulation, and mutant mice were born at the expected Mendelian ratio. Our Prickle1 gene targeting strategy was identical to that of the previous study (Tao et al., 2009). In both studies, exon2 carrying the protein translation start codon was deleted. This should result in a null allele, which we were able to confirm directly by showing absence of residual or aberrant proteins on immunoblots using both N and C-terminal Prickle1 antibodies. Furthermore, we replaced exon2 by an eYFP reporter carrying its own poly(A) signal, which should cause RNA transcription to stop and thus a further disruption to Prickle1 gene expression. Therefore the stark discrepancy in mutant survival between the two studies cannot be attributed to a differential degree of gene ablation. Rather, this difference is most likely due to the difference in genetic background, which in our case is an equal mixture of C57Bl/6 and Sv/129 but a pure C57Bl/6J through backcrossing in the prior study.

The pleiotropic tissue outgrowth defects in Prickle1 mutants indicate a disruption of PCP pathway that govern morphogenesis. The vectorial cell disarrangement in Prickle1 mutant tissues including in kidney and heart demonstrates a role of Prickle1 in PCP function. However, PCP defects in Prickle1 mutants appear to mechanistically differ from those in other PCP genes. Mislocalization of PCP proteins, small GTPases and F-actin are reminiscent of disrupted AB polarity, which is not usually seen in other PCP mutants. This part of our data is consistent with the suggested role of Prickle1 in embryonic epiblast AB polarity (Tao et al., 2009). These observations indicate that Prickle1 may also be involved in setting up apico-basal polarity.

Our study suggests that Prickle1 is a major component of Wnt5a signaling. Several lines of evidence support this conclusion: i) Prickle1 is targeted to proteasomal degradation upon Wnt5a signaling; ii) Prickle1 is upregulated in Wnt5a null mutant limb tissue; and iii) Dvl2 phosphorylation, a hallmark of Wnt5a activation (González-Sancho et al., 2004; Ho et al., 2012), is altered in Prickle1 mutant tissues. Published studies demonstrating ubiquitination of Prickle1 by Smurf/Par6 complex upon Wnt5a signaling (Narimatsu et al., 2009) and Dvl2 stability regulated by Prickle1 (Carreira-Barbosa et al., 2003) provide additional support for notion that Prickle1 transduce Wnt5a signaling to downstream targets. A previous study demonstrated that Vangl2 is a phosphorylation target upon Wnt5a/Ror2 signaling, and Vangl2 polarization is lost upon disruption of Wnt5a signaling in chondrocytes (Gao et al., 2011). However, in Ror compound mutant MEF cells, Dvl rather than Vangl2 phosphorylation is found to be a critical event (Ho et al., 2012). Similarly, our data showed that Dvl2 phosphorylation is a downstream target of Prickle1-mediated Wnt5a signaling. We also found that Vangl2 polarization is partially retained in Prickle1 mutant chondrocytes, suggesting that Prickle1 and Dvl2 may act downstream of Vangl2.

Loss of Prickle1 leads to enhanced Wnt5a signaling in biochemical assays yet the mutant phenotypes mimic those of both gain- and loss-of-function mutants of Wnt5a (van Amerongen et al., 2012; Yamaguchi et al., 1999) (supplementary material Fig. S9). This paradox could be explained by the nature of polarity signaling. Namely, Wnt5a gradient confers a vectorial (directional) cue that drives cell behavior. Both absence of Wnt5a or excess Wnt5a that saturates signaling around the entire cell will lead to loss of the vectorial information (supplementary material Fig. S10). Loss of Prickle1 disrupts Wnt5a signaling to downstream targets so that cells cannot interpret or respond to the vetorial information. Hence all three genetic alterations lead to a similar defect in tissue morphogenesis.

All procedures involving the use of mice were approved by National Eye Institute Animal Care and Use Committee (ACUC). Prickle1 gene-trap mutant strain was generated as described previously (Liu et al., 2013). The knockout construct was designed as a gene-trap allele first and was brought into straight and conditional knockout alleles upon excisions by Cre and Flp recombinase, respectively. Briefly, the following elements were engineered into Prickle1 locus in order: Frt, En2 splicing acceptor, eYFP, polyadenylation signal (poly (A)), loxP, PGK-neo, Frt, and loxP right before Prickle1 exon 2, and a third loxP after Exon2. This allele allows eYFP reporter expression under the control of endogenous Prickle1 promoter, and was designated as Prickle1^a (Fig. 1A). A complete Cre and Flp recombination will convert Prickle1^a to Prickle1^b and Prickle1^c alleles, respectively (Fig. 1A). Southern analysis and genotyping were conducted as described previously (Liu et al., 2013).

Wnt5a knockout mouse strain was obtained from JAX laboratory, and maintained by standard protocol that complies with ACUC regulations at NEI.

Mouse skeletal visualization entirely followed protocol by Ovchinnikov (Ovchinnikov, 2009). Images were taken with dissection Zeiss microscope equipped with Axiovision digital camera. Tissues were dissected and fixed with 4% PFA and subjected to sectioning at 15 um for frozen sections, or 50 um for vibratome sections. H&E staining was performed on fixed frozen sections (FD Neurotechnologies, Inc, PS103-1, PS-104-1), and on methacrylate plastic sections generated at NEI Histology core. Immunohistochemistry were performed as previously described (Liu et al., 2012). Skin flat mount was prepared exactly following the description by Wang et al. (Wang et al., 2010). Briefly, skin was dissected and cleaned away of fat tissues, fixed with 4% PFA, and cleared with benzyl benzoate:benzyl alcohol (BBBA), imaged using a Zeiss dissection microscope (Liu et al., 2008). Antibodies and fluorescent dyes were used in this study are as follows: Phalloidin-Alexa 488, (Invitrogen, A12379); Lotus tetragonolobus lectin, (Vector Laboratories, FL-1321); rat anti-E-cadherin, (Sigma, U3254); mouse anti-calbindin D28K, (Sigma, C9848, CB-955); mouse anti-α–acetylated tubulin, (Sigma, T6793); mouse anti-Dvl1, (3F12, Santa Cruz Sc-8025); mouse anti-Dvl3, (4D3, Santa Cruz : SC8027); mouse anti-Dvl2 (Santa Cruz, sc-8026); mouse anti-Ror2 (DSHB), rabbit anti-Vangl2, (gift from Dr. Matthew Kelley at NIDCD, NIH); mouse anti-RhoA, (Sigma, SAB1400017; Cytoskeleton, Inc. Cat. ARH03 ); β-catenin (610154, BD transduction laboratories); rabbit anti-NF-145, (Chemicon, AB1987); mouse anti-Cdc42, (Cytoskeleton, Inc, Cat # ACD03); mouse anti-Rac1, and (Cytoskeleton, Inc, Cat # ARC03). Fluorescent images were collected using Leica SP5 and Olympus FV1000 confocal microscope at NNRL and NEI imaging core. Fluorescent intensity measurement was conducted using NIH ImageJ software.

The cardiac myofibrils orientation is defined by a group of arrows arbitrarily drawn along Phalloidin stained actin. Each arrow marks at least 4 continuous A-bands (Fig. 2H). Angles of arrows are measured referring to epicardium surface, on which rays from any given points are defined as 0 and 180 degree angles respective to opposite directions (right versus left). All angles are forced into quadrant I–II (0–180°) by subtracting 180° if they happen to fall into quadrant III–VI (180–360°). Arrows are binned into 0°–90° and 91°–180°, and presented as percentage of total in graph. About 35 angular arrows/section were drawn to illustrate myofibrils' orientations. Three tissue sections from each animal were used for quantification. Three animals for each genotype were used for analysis. Total 108 and 111 angular arrows from respective control and the mutant tissue sections were measured.

Thickness of a collecting duct (CD) wall was quantified as cell number. E-cadherin and DAPI staining was used to identify cells making up the CD wall on cross sections. Each imaged plane is composed of 7×1 µm-projected stacks. Cells out of an imaged plane are named off-plane cells. Three tissue sections from three animals of each genotype were used for analysis. About 90 tubules were counted for each genotype. Student t-test was used to detect the P value for all groups of data.

Cell arrangement was quantified by examination of cell intersection pattern in the thick ascending limb of the Henle's loop. Depending on number of cell laterals (junctions) intersect, cell arrangement is defined as 3-lateral and 4-lateral patterns. Cell alignment was quantified as a transformation of lateral vectorial lengths of two adjacent columns of cells. A shortest compound line path between two adjacent columns of cells along the vertical direction of a tubule was drawn (supplementary material Fig. S4C). Vectorial additions were then performed for laterals from each cell vertex on the compound line using parallelogram method to create collections of new vectors (supplementary material Fig. S4D). These vectors were plotted in a polar coordinate (supplementary material Fig. S4E). The orientations and lengths of the vectors variations were subjected to paired t-test to detect statistical differences. 68 vectors acquired roughly equally from three animals of either wild type or the mutant Henle's loop were analyzed and plotted.

Cell interfaces of a Henle's loop cell were quantified by counting cells directly contacting that cell. Because the shapes of many mutant cells are not regular, we therefore named the cells by their interfaces (int). Three tissue sections from each animal were used for quantification. Three animals for each genotype were used for analysis. Total 331 wild type and 416 mutant cells were cataloged.

Cells with positive Vangl2 staining were counted. Fractions of polarized Vangl2 cells were presented as percentage. A set of three sections derived from either wild type or mutant embryos were used for analysis. P-values were computed by T-test function. The aligned cell groups were defined only if three center points of Vangl2 staining of adjacent cells make an angle equal or large than 165 degree (arbitrarily defined), for example angles a, b and c (supplementary material Fig. S7). By this definition, the aligned cells do not deviate more than 15 degree from each other in terms of stained Vangl2 orientations. Three wild type and three mutant embryos were analyzed, each of which 24 cell groups were analyzed. A t-test was conducted using the Excel program to compute p values.

Culture medium used for MEFs and HEK293 cells is based on DMED/F12 (Invitrogen, Cat. 12660-012) supplemented with or without 10% FBS and penicillin/streptomycin (Invitrogen, Cat. 15070-063). Cell transfection was performed in 6-well dishes with total amount 2 ug DNA/well using Lipofectamine 2000 (Invitrogen, Cat. 11668-019) for different combinations of GFP-Pk1, Wnt5a and Ryk plasmids. Biological triplicates and technical replicates were made for each transfection. 36 hr after transfection, cells were treated with 100 nM DMSO or MG-132 for 20 hr. Cells were then harvested for preparation of protein extracts.

For culture primary mouse embryonic fibroblasts (MEFs), E13.5 embryos were dissected out from the uterus of pregnant females in HBSS (Invitrogen, 14025-076), internal organs and head were removed, remained tissues were chopped into pieces using a razor blade, digested with 2.5% trypsin at room temperature for 15 min with soft taps every 5 min, pipetted through 1 ml pipette tip for multiple times, added DMEM/F12 (Invitrogen, Cat. 12660-012) medium with supplement of 10% FBS and penicillin/streptomycin (Invitrogen, Cat. 15070-063), and filtered through 70 µm cell mesh strainer (Biologixresearch, 15-1070). Cells were collected by centrifuge at 1000× g, plated into 10 cm culture dish, and used at passage 2.

Protein extracts from tissue or transfected HEK293 cells were prepared with RAPI or 2×SDS sample buffers according to the standard protocols. Protein A Dynabeads (Invitrogen, Cat.# 1000D) were used for pulling down anti-GFP antibody (Rabbit anti-GFP , Torrey Pines Biolabs, Inc, TP401) associated Prickle1 protein complex at 4C overnight. Proteins were dissociated from the beads by heating in 2×SDS sample buffer at 98C for 5 min, then run a western blot and probed with anti-Ubiquitin antibody and anti-GFP antibody.

Custom-made antibodies for Prickle1 were generated in our laboratory by immunizing rabbits with bacterially expressed recombinant proteins corresponding to aa 344–464 (Pk1C) and aa 711–849 (Pk1N), respectively. Antisera were affinity purified and were used at 1:1000 dilutions for Western blots. Mouse anti-γ-tubulin (Sigma, T-6557) was used at 1:3000; Rabbit anti-Wnt5a was used at 1:1000 (Abcam, ab72583); RhoA, Cdc42 and Rac1 (Cytoskeleton, Inc) were used at 1:1000; rabbit anti-Dvl2 (Chemicon, AB5972; Cell Signaling, Cat# 3216) and rabbit anti-GFP were used at 1:2000; mouse anti-Ubiquitin antibody (ab7254) was used at 1:1000.

Samples were prepared for SEM as described (May-Simera and Kelley, 2012). Dissected cochlea sensory epithelia were fixed in EM fix (2.5% glutaraldehyde, 4% PFA, 10 mM CaCl₂ in 0.1M Hepes) for 2 hr at room temperature. After coating with repeated washes of 1% OsO₄ and 1% tannic acid, samples were dehydrated through an ethanol series, critical point-dried, and imaged on a S-4800 (Hitachi) field emission scanning electron microscope.

X-ray micro computed tomography (micro-CT imaging) is performed in the Mouse Imaging Facility. The mouse is placed under general anesthesia in an induction chamber with 3–5% isoflurane delivered by a gas mixture of oxygen, medical air and nitrogen. Once the animal is unconscious, anesthesia is maintained with 1–2% isoflurane administered via nosecone. Sterile ophthalmic ointment is applied to the corneas to prevent desiccation under anesthesia. The animal is placed on the imaging platform secured into the necessary imaging position using transpore and autoclave tape. Depth of anesthesia is monitored by direct visualization of the animal respiratory rate. Body temperature is maintained by the imaging systems warm air blower. Anesthetic gasses are scavenged in a centralized vacuum system approved by the Division of Safety. Total imaging time is approximately 5–30 minutes for a 180 or 360 degree scan.

We thank Dr. Jeremy Nathans at Johns Hopkins for experimental and intellectual inputs, Dr. Lijin Dong at National Eye Institute (NEI) Genetic Engineering Facility (GEF) for mutant mouse generation, Dr. Robert Fariss at NEI Imaging Facility, and Brenda Klaunberg at NIH Mouse Imaging Facility for help with imaging analysis, Yide Mi, Linn Giester, and Balakrishna Archana for technical assistance, Dr. Tudor Baeda at NEI for helpful discussions, and Sarah Cheng for initial English Editing. This work is supported by intramural research program of the NEI.

This research is partly or wholly funded by the NIH.