The involvement of PCP proteins in radial cell intercalations during Xenopus embryonic development

Abstract

The planar cell polarity (PCP) pathway orients cells in diverse epithelial tissues in Drosophila and vertebrate embryos and has been implicated in many human congenital defects and diseases, such as ciliopathies, polycystic kidney disease and malignant cancers. During vertebrate gastrulation and neurulation, PCP signaling is required for convergent extension movements, which are primarily driven by mediolateral cell intercalations, whereas the role for PCP signaling in radial cell intercalations has been unclear. In this study, we examine the function of the core PCP proteins Vangl2, Prickle3 (Pk3) and Disheveled in the ectodermal cells, which undergo radial intercalations during Xenopus gastrulation and neurulation. In the epidermis, multiciliated cell (MCC) progenitors originate in the inner layer, but subsequently migrate to the embryo surface during neurulation. We find that the Vangl2/Pk protein complexes are enriched at the apical domain of intercalating MCCs and are essential for the MCC intercalatory behavior. Addressing the underlying mechanism, we identified KIF13B, as a motor protein that binds Disheveled. KIF13B is required for MCC intercalation and acts synergistically with Vangl2 and Disheveled, indicating that it may mediate microtubule-dependent trafficking of PCP proteins necessary for cell shape regulation. In the neural plate, the Vangl2/Pk complexes were also concentrated near the outermost surface of deep layer cells, suggesting a general role for PCP in radial intercalation. Consistent with this hypothesis, the ectodermal tissues deficient in Vangl2 or Disheveled functions contained more cell layers than normal tissues. We propose that PCP signaling is essential for both mediolateral and radial cell intercalations during vertebrate morphogenesis. These expanded roles underscore the significance of vertebrate PCP proteins as factors contributing to a number of diseases, including neural tube defects, tumor metastases, and various genetic syndromes characterized by abnormal migratory cell behaviors.

1. Introduction

The planar cell polarity (PCP) pathway was originally defined as a mechanism allowing epithelial cells to polarize in a plane of a tissue perpendicular to their apicobasal axis (Adler, 2012; Peng and Axelrod, 2012; Wallingford, 2012; Zallen, 2007). Besides the organization of fly and mammalian epithelia, the conserved PCP proteins are involved in the migration of neurons and germ cells (Glasco et al., 2012; Jessen et al., 2002), asymmetric cell division (Bellaiche et al., 2004; Lake and Sokol, 2009; Vladar et al., 2009), branching morphogenesis (Miller et al., 2011; Yates et al., 2010a, 2010b), angiogenesis (Cirone et al., 2008) and ciliogenesis (Gray et al., 2011; Wallingford and Mitchell, 2011). Due to their connection to cilia functions, mutations in the corresponding PCP genes cause diverse ciliopathies. These range from a relatively common polycystic kidney disease to more rare genetic syndromes such as Meckel–Gruber syndrome, Bardet–Biedl syndrome, Oro-facio-digital syndrome and nephronophthisis. (Simons and Mlodzik, 2008; Wang and Nathans, 2007). Additionally, PCP proteins have been associated with multiple congenital abnormalities, such as neural tube and cardiac defects, and implicated in tumor invasiveness (Gray et al., 2011; Hamblet et al., 2002; Luga et al., 2012; Zhu et al., 2012).

In vertebrate early development, PCP pathway components are critical for several cell behaviors, such as cell intercalations and apical constriction during neural tube closure and mesoderm convergent extension (Ossipova et al., 2015a; Sokol, 2000, 2015; Wallingford et al., 2002a). All major components of the Wnt/PCP pathway, including Wnt11 (Heisenberg et al., 2000; Tada and Smith, 2000), Disheveled (Sokol, 1996; Wallingford et al., 2000), Vangl2/Stbm (Goto and Keller, 2002; Jessen et al., 2002; Park and Moon, 2002), Prickle (Pk) (Carreira-Barbosa et al., 2003; Takeuchi et al., 2003; Wallingford et al., 2002b), Fmi/Celsr1 (Formstone and Mason, 2005), have been shown to modulate convergent extension movements during gastrulation and neurulation. As convergent extension is driven largely by mediolateral cell intercalations (Keller, 2002; Shih and Keller, 1992), PCP components were proposed to stabilize mediolateral cell protrusions (Jessen et al., 2002; Wallingford et al., 2000) and/or promote actomyosin contractility at mediolaterally oriented cell junctions (Shindo and Wallingford, 2014). By contrast, the involvement of PCP signaling in radial cell intercalation has been less clear.

During radial intercalation, cells change their position between different tissue layers, along the apical-basal embryonic axis. As a result, the tissue becomes thinner and contains fewer cell layers. In lower vertebrates, radial intercalations are the predominant cellular mechanism of tissue spreading before and during gastrulation, and contribute to the development of the neural tube and epidermis (Keller, 1980; Solnica-Krezel and Sepich, 2012; Walck-Shannon and Hardin, 2014). Specifically, during Xenopus neurulation, the neural plate, consisting of multiple cell layers, is converted into the single-cell-layered neural tube (Hartenstein, 1989; Keller, 1991). Similarly, the two-layered embryonic skin develops from multi-layered epidermal ectoderm (Deblandre et al., 1999; Drysdale, 1992; Stubbs et al., 2006). These processes involve diverse molecular events including changes in cell adhesion and microtubule-dependent vesicular trafficking (Itoh et al., 2014; Kim et al., 2012; Lepage et al., 2014; Marsden and DeSimone, 2001; Solnica-Krezel and Sepich, 2012; Song et al., 2013; Werner et al., 2014).

Evidence has been accumulating that the Wnt/PCP pathway may be involved not only in mediolateral cell intercalations, but also in radial intercalations. Mouse epiblast cells reveal Pk1-dependent apical-basal polarity, an observation possibly relevant to radial intercalations (Tao et al., 2009). Consistent with abnormal radial intercalatory behavior, intestine does not properly elongate in Wnt5-deficient mouse embryos (Cervantes et al., 2009). This has been explained as a cell proliferation defect, but cell intercalations have not been examined. Furthermore, the interference with the activity of Celsr in zebrafish embryos results in epiboly defects, although this role has been attributed to the modulation of cell adhesion rather than PCP signaling (Carreira-Barbosa et al., 2009). Together, these observations suggest that PCP proteins might function in radial cell intercalation.

Our study has addressed this possibility in Xenopus embryonic epidermis, in which some cell types, including multiciliated cells (MCCs), intercalate into the superficial cell layer. Due to tissue-targeted gene manipulation, Xenopus skin has become an established in vivo model for other mucociliary epithelia containing MCCs, such as those of human airways or reproductive tract (Brooks and Wallingford, 2014; Dubaissi and Papalopulu, 2011). A specific technical advantage of this system is to unilaterally manipulate protein function by targeted microinjections, with the uninjected side serving as an internal control. Our analysis of Xenopus epidermal ectoderm revealed an enrichment of the core PCP component Vangl2 at the apical surface of MCCs. We also identified Prickle3 (Pk3), a member of the Prickle family that is mainly expressed in the embryonic skin, and demonstrated the requirement of Vangl2, Pk3 and Disheveled for the radial intercalation of MCC precursors into the superficial cell layer of the skin. Additionally, interference with PCP signaling inhibited radial intercalation of inner layer cells in the neural plate and non-neural ectoderm. To further address the underlying mechanism, we identified the motor kinesin KIF13B as a Disheveled-interacting protein. KIF13B has been previously implicated in cell polarity and cell migration (Horiguchi et al., 2006; Tarbashevich et al., 2011). We demonstrate that KIF13B physically associates with Disheveled and synergizes with PCP signaling to regulate cell intercalatory behavior. Collectively, our data support a general role for PCP signaling in radial cell intercalations during Xenopus gastrulation and neurulation.

2. Methods

2.1. Plasmid constructs and morpholinos

Plasmids encoding GFP-C1 in pXT7, GFP-CAAX in pCS2+ (Kim et al., 2012); nGFP in pCS2+ (Dollar et al., 2005), CFP-Vangl2 in pCS105 (Stbm) (Itoh et al., 2009), mouse HA-Vangl2 (Gao et al., 2011), Mig12-GFP in pCS2+ (Yasunaga et al., 2011) and the Myc-tagged Disheveled constructs Xdsh (Myc-Dvl2), Xdd1 and Xdd2 (Sokol, 1996) have been described. The plasmid encoding Drosophila Pk in pCS105 was a gift from A. Jenny. A cDNA encoding Prickle3 (Pk3) protein missing five aminoacids from the N-terminus (GenBank accession number: BC154995) was amplified by RT-PCR from Xenopus neurula RNA and subcloned into pCS2-Flag. In Flag-Pk3ΔPET, the PET domain has been deleted by PCR according to what was described in Takeuchi et al. (2003).

A cDNA encoding a C-terminal CAP-GLY-domain containing fragment of Xenopus KIF13B was isolated in a yeast-two-hybrid screen from a Xenopus gastrula cDNA library (Brott and Sokol, 2005; Itoh et al., 2000), using the DIX domain of Dvl2 as a bait. pCMVtag2-FlagKIF13B was a gift from A. Chishti. For RNA injections, the FlagKIF13B insert has been subcloned into pCS2-Myc. Details of cloning are available upon request. For lineage tracing, GFP RNA (50–100 pg) was injected along with morpholino anti-sense oligonucleotides (MOs) or RNAs. Capped mRNA was made by in vitro transcription with T7 or SP6 promoter using mMessage mMachine kit (Ambion).

The following morpholino oligonucleotides (MOs) have been used: Vangl2/Stbm MO 5′-GAG TAC CGG CTT TTG TGG CGA TCCA-3′ (Ossipova et al., 2015a), KIF13B MO, 5′-ATCTTGCACAGCGAGCTCCCCTAAC-3′ (Tarbashevich et al., 2011); Pk3 MO1, 5′-GGATGCCGCCCGCTCTCTCCCTTA-3′. Pk3 MO2, 5′-CTCCTCCTGGAATTACGGAACATCC-3′. MO to the Fz8-associated protein phosphatase FRIED (Itoh et al., 2005) with the sequence 5′-GCTTCAGCTAGTGACACATGCAT-3′ has been used as a negative control and produced no detectable phenotype (Itoh et al., 2014). MOs were purchased from Gene Tools (Philomath, OR).

2.2. Xenopus embryos, RNA and morpholino microinjections, in situ hybridization

In vitro fertilization, culture and staging of Xenopus laevis embryos were carried out as previously described (Dollar et al., 2005). For microinjections, four-eight cell embryos were transferred into 2–3% Ficoll in 0.3 × MMR buffer and 5–10 nl of mRNAs or MO solution was injected into one or more blastomeres. Amounts of injected mRNA or MO per embryo have been optimized in preliminary dose–response experiments (data not shown) and are indicated in figure legends. Whole-mount in situ hybridization was carried out using standard techniques (Harland, 1991) with the digoxigenin-labeled antisense and sense RNA probes for Pk3.

2.3. Immunostaining and image analysis

Embryos were fixed at desired stages with Dent's fixative (80% Methanol: 20% DMSO) for 3 h at RT followed by overnight incubation at −20 °C and rehydrated with 1×PBS washing. For cryosectioning, the embryos were embedded in the solution containing 15% cold fish gelatin and 15% sucrose, sectioned at 10–20 μm, and immunostained overnight essentially as described (Dollar et al., 2005). Antibodies against the following antigens were used: GFP (1:200, B-2, Santa Cruz, mouse monoclonal or Invitrogen, rabbit polyclonal), β-catenin (1:200, Santa Cruz, rabbit polyclonal), acetylated tubulin (1:200, Santa Cruz, mouse monoclonal), Rab11 (1:100, BD biosciences, mouse monoclonal), and rabbit polyclonal antibodies for Xenopus centrin-2 (Kim et al., 2012) and Xenopus Vangl2 (Ossipova et al., 2015a). Secondary antibodies were against mouse or rabbit IgG conjugated to Alexa Fluor 488, Alexa Fluor 555 (1:100, Invitrogen) or Cy3 (1:100, Jackson ImmunoResearch). To detect Vangl2 and Rab11, embryos were fixed as described (Ossipova et al., 2015a). Cryosections were mounted for observation with the Vectashield mounting medium (Vector). For en face imaging of multiciliated cells, embryonic epidermis was removed from stage 28 embryos, fixed with MEMFA (Harland, 1991) and epifluorescence was imaged. Alternatively, we used wholemount staining with anti-centrin and anti-GFP antibodies as described (Kim et al., 2012). All fluorescent images were captured using the Axioimager microscope equipped with the Axiovision imaging software (Zeiss). For quantification, results were pooled from two to four independent experiments. At least 4 embryos were analyzed by cross-sections in each experiment. N indicates the total number of ciliated cells quantified for each group. Results are shown as means ± s. e., statistical analysis was carried out with SigmaPlot (Systat Software, Point Richmond, CA). p Values were determined using the two-tailed Student's t-test.

2.4. Cell culture, transfection, immunoprecipitation and western blot analysis

Human embryonic kidney (HEK) 293T cells were grown at 37 °C in Dulbecco's Modified Eagle Medium (Gibco) supplemented with 10% Fetal Bovine Serum (Gibco) and 5 mg/ml penicillin/streptomycin. Flag-KIF13B and Myc-Dvl2 plasmid DNAs or control pCS2 vector DNA were transfected using polyethylenimine (PEI) as previously described (Ossipova et al., 2009). Twenty four hours after transfection, the cells were lysed in 50 mM Tris–HCl, pH7.5, 150 mM NaCl, and 0.5% of Triton-X100, containing protease inhibitors, and the lysates were cleared by centrifugation at 10,000g for 10 min. The lysates were incubated overnight at 4 °C with 10 μl of anti-Flag-agarose beads (Sigma). The bound protein complexes were washed with the lysis buffer and separated by SDS-PAGE using standard techniques. Western blotting was carried out as described previously (Itoh et al., 2000) with the following antibodies: M2 anti-Flag (Sigma), Myc9E10 hybridoma supernatant, α-tubulin (Sigma) and anti-GFP (B2, Santa Cruz). Chemiluminescence was captured by the LAS-3000 imager (Fujifilm).

3. Results

3.1. Vangl2 regulates multiciliated cell migration in the Xenopus skin

One model for radial cell intercalations is the integration of multiciliated cell (MCC) precursors into the superficial cell layer during late neurula stages (Deblandre et al., 1999; Drysdale, 1992; Stubbs et al., 2006). To evaluate a role for PCP signaling in this system, we first assessed the subcellular distribution of the core PCP components Vang-like 2 (Vangl2) and Prickle in the differentiating MCCs. The expression of Vangl2 in Xenopus non-neural and neural ectoderm has been previously reported (Darken et al., 2002; Park and Moon, 2002).

Since Drosophila Vang and Prickle (Pk) form a protein complex (Bastock et al., 2003; Jenny et al., 2003), we coinjected early embryos with low doses of CFP-Vangl2 RNA and Pk RNA, which were determined in separate dose–response experiments (data not shown). At these doses, the injected embryos developed normally, without visible phenotypic defects. Immunostaining revealed a dotted pattern of Vangl2 in the superficial ectoderm at late neurula stages, suggesting a polarized localization in intercalating cells (Fig. 1A). Using antibodies against centrin-2 as a marker for MCCs containing many basal bodies (Kim et al., 2012), we confirmed that the Vangl2/Pk complex is enriched at the apical surface of these cells (Fig. 1B). This apical localization of Vangl2/Pk complexes extends previous studies, describing the distribution of Disheveled near the apical surface of MCCs (Park et al., 2008) and indicates that PCP signaling might regulate the development of MCCs. Although PCP proteins should be present in other intercalating cell types, such as ionocytes (Dubaissi and Papalopulu, 2011) or small secretory cells (Dubaissi et al., 2014; Walentek et al., 2014), they are most readily detected in MCCs, likely being enriched by the associated basal bodies.

Fig. 1.

Vangl2 is required for the migration of skin multiciliated cells. (A, B) The Vangl2/Pk complex has a polarized apical distribution in MCC precursors. Four-cell embryos were injected with RNAs encoding CFP-Vangl2 (0.1 ng) and Pk (0.05–0.1 ng), embryos were fixed at stage 18 and immunostained with anti-GFP (A) or anti-GFP and anti-centrin antibodies (B). Arrows point to apically polarized Vangl2. Scale bar in A is 20 μm, bar in B, 10 μm, also applies to C, D. (C–J) Depletion of Vangl2 inhibits MCC intercalation into the superficial cell layer. Four-cell embryos were unilaterally injected with Vangl2 MO (10 ng per blastomere) and GFP RNA (0.1 ng) for lineage tracing with or without HA-Vangl2 RNA (0.1 ng). The embryos were fixed at indicated stages and immunostained for GFP and acetylated tubulin (Ac-tub) (C, D) or centrin (E–J) to mark MCCs. (E–J) Vangl2 RNA injection partly rescues MCC intercalation defect caused by Vangl2 MO. (E–G) En face view. (H–J) Embryo sections. Arrows point to MCCs in the superficial cell layer. Asterisks indicate a deficiency in MCC migration. The apical surface position (at the top) and MCC boundaries are shown by dotted lines. Scale bars in E, H are 10 μm. (K–M) Quantitation of data for the experiments shown in (C–J) and Fig. S1. The percentage of MCCs that fail to migrate and remain in the deep cell layer was calculated separately for Ac-tub (K), centrin (K, L) or Rab11 (M) relative to the total number of MCCs. Number of scored MCCs (n) is shown on top of each bar. All graphs present means ± s. e. Statistical significance was determined by the Student's t-test (**, p=0.014). Data are representative of two to four independent experiments, 4–7 embryos were examined for each treatment.

We next studied the function of Vangl2 in MCC intercalatory behavior by the loss-of-function approach. Four-cell embryos were injected into the animal pole with 20 ng of previously characterized Vangl2-specific morpholino oligonucleotide (MO) (Ossipova et al., 2015a). Epidermal MCCs were visualized by immunostaining of embryo sections with antibodies to acetylated tubulin or centrin (Kim et al., 2012). A specific technical advantage of the Xenopus model system is the ability to unilaterally manipulate protein function by targeted microinjections, with the uninjected side serving as an internal control. Depletion of Vangl2 on one side of the embryo reduced MCC numbers in the superficial ectodermal layer in stage 20–24 embryos as compared with the uninjected side or with the injection of a control MO (Fig. 1C–L). The majority of MCCs remained in the deep ectodermal cell layer, indicating that the depletion of Vangl2 disrupted the ability of MCC to intercalate. Confirming MO specificity, these effects were partly rescued by the injection of RNA encoding mouse Vangl2 (Fig. 1E–J, L). Of note, the epidermal tissues depleted of Vangl2 contained more cell layers than control epidermis (Figs. 1H–J and S1A and B).

This deficiency in MCC intercalation in Vangl2 morphants was observed as early as stage 18. In normal stage 18/19 embryos, migrating MCCs are visible as Rab11-positive cells in the superficial layer (Kim et al., 2012). In embryos depleted of Vangl2, the majority of Rab11-positive MCC cells remained in the deep ectodermal cell layer (Figs. 1M and S1C and D). Interestingly, we observed that Vangl2 depletion not only disrupted the migration of MCCs, but also affected the early alignment of MCC progenitors relative to the embryo surface as described previously for Rab11 interference (Kim et al., 2012). Taken together, these observations are consistent with a role for Vangl2 in MCC radial intercalation behavior.

3.2. Radial intercalation of multiciliated cells requires PCP signaling

We next wanted to assess whether the observed role of Vangl2 in MCC intercalation is specific to Vangl2 or reflects a common function shared by other PCP proteins. The expression of previously identified Xenopus Pk1 homolog is largely limited to the mesodermal and neural tissues (Wallingford et al., 2002b) and we wanted to know whether there are other Prickle homologs expressed in MCCs. Using RT-PCR, we cloned a cDNA for previously uncharacterized Prickle3 (Pk3), and analyzed its expression pattern by whole mount in situ hybridization. We observed that Pk3 transcripts are present in the early ectoderm and later restricted to the Xenopus epidermis and neural plate folds (Fig. 2A–H). At stages 21–26, Pk3 RNA was detectable in the developing otic vesicle, profundal and posterior placodal regions (Schlosser and Ahrens, 2004) and the skin (Fig. 2D–F). The epidermal signal was present in both superficial and deep cell layers (Fig. 2H). Since the epidermis is the major domain of Pk3 expression, we suspected that Pk3 may play a role in cell intercalations during skin development.

Fig. 2.

Expression of prickle3 transcripts in Xenopus early embryos. Embryos were subjected to whole-mount in situ hybridization with Pk3 anti-sense RNA probe or sense probe as a control. Pk3 transcripts are present in the animal pole ectoderm of stage 10 gastrulae (A), and non-neural ectoderm of stage 15 neurulae (B). (C) Stage 18 embryo. Epidermal ectoderm is strongly positive (arrowhead), as compared to neuroectoderm. (D) Stage 21 embryo, anterior region, with staining in the profundal and posterior region placodes (arrowhead). (E) Stage 24 embryo, the otic vesicle and the posterior placodal region are positive (arrowheads). (F) Stage 26 embryos, tailbud. An embryo hybridized with the sense probe is shown on bottom. Anterior is to the left. Lateral view is shown in all panels, except dorsal view in C. (G, H) Cross-sections of stage 18 embryos. Arrowheads point to the staining in neural folds (G) and in both inner and outer layers of the epidermis (H). Positions of nuclei in the epidermis are outlined by dotted lines. Abbreviations: n, notochord, s, somites, np, neural plate.

To interfere with the function of Pk3, we made a construct, in which the PET domain has been deleted. A similar form of the Drosophila Pk has been shown to act in a dominant interfering manner (Lin and Gubb, 2009). Wild-type Pk3 and Pk3ΔPET RNAs were overexpressed in the Xenopus embryonic ectoderm. Whereas wild-type Pk3 did not have major effects on MCCs in the surface epithelium, the surface of Pk3ΔPET-expressing embryos was devoid of MCCs (Fig. S2A–C), suggesting that MCCs do not intercalate into the superficial ectoderm layer. Similar results were obtained using an independent approach, in the epidermal tissue depleted of Pk3 with two non-overlapping Pk3 MOs. In embryos injected with either MO, more than 80% of MCCs remained in the inner cell layer (Figs. 3A–C and S2D and E). Both Pk3 MOs effectively blocked Pk3 RNA translation in vivo (Fig. S2F). Embryo sections confirmed that MCCs formed in the deep ectodermal layer, but failed to migrate (Fig. 3A–C). In contrast to Pk3 MOs, control MO did not significantly affect MCC behavior. These observations suggest that, similar to Vangl2, Pk3 functions to regulate MCC intercalatory behavior.

Fig. 3.

Interference with Pk3 and Disheveled leads to MCC intercalation defects. (A–C) Role of Pk3 in MCC intercalation. Four-cell embryos were injected with control morpholino (COMO) or Pk3 MO1 (25 ng per blastomere), along with GFP RNA (0.1 ng) as a lineage tracer. Stage 22 embryos were cryosectioned and immunostained for Centrin and GFP. Arrows indicate MCCs integrated into the superficial cell layer. Asterisks point to MCCs that failed to migrate. Dashed lines mark the boundaries of the epidermis, apical is up. (D–G) Effects of Disheveled mutant constructs on MCC intercalation. Xdd1 and Xdd2 RNAs (0.5–1 ng each) were coinjected with GFP RNA into four-cell embryos, embryos were either lysed at stage 11 for protein analysis (D) or fixed at stage 22 and immunostained for GFP and Centrin (E, F). (D) Immunoblot analysis with anti-Myc antibodies showing Xdd1 and Xdd2 expression levels. Scale bar in B is 20 μm, also applies to (A, E, F). (C, G) Quantification of data. Scoring has been done as described in Fig. 1. All graphs show means ± s. e. At least 6–10 embryos were examined per each treatment. n, number of examined cells.

Disheveled is another core PCP component that has been implicated in cell behavior during vertebrate morphogenesis. To extend our observations, we wanted to examine the effects of Xenopus Dvl2 constructs Xdd1 and Xdd2 on MCC intercalation. Following the original study (Sokol, 1996), Xdd1 has been frequently used to interfere with Disheveled signaling due to its strong effects on convergent extension, whereas Xdd2 did not significantly influence body axis elongation. We further characterized the effects of Xdd1 and Xdd2 on embryo morphology in targeted RNA injections, which result in similar levels of protein expression (Fig. 3D). As previously reported, neural folds did not close in Xdd1 RNA-injected embryos, whereas embryos expressing Xdd2 had milder neural tube defects (Fig. S3A–C). By contrast, both Xdd1 and Xdd2 reduced non-neural ectoderm spreading and caused the lateral bending of the embryo towards the site of RNA injection (Fig. S3D–H). At these doses, both constructs inhibited MCC intercalation in stage 22 embryos (Fig. 3E–G), supporting a role of Disheveled in MCC migration into the superficial layer.

Together, these findings indicate that PCP signaling is required for MCC intercalatory behavior in embryonic skin.

3.3. KIF13B physically interacts with Disheveled and synergizes with PCP proteins to promote MCC intercalation

Radial intercalations of MCCs have been shown to involve vesicular and microtubule-dependent trafficking (Kim et al., 2012; Werner et al., 2014). Moreover, polarized microtubule-dependent trafficking has been implicated in PCP signaling (Matis et al., 2014; Shimada et al., 2006; Vladar et al., 2012). To better understand how the PCP pathway modulates radial cell intercalations, we searched for new proteins physically interacting with Disheveled. Our yeast two-hybrid screen using a gastrula cDNA library identified the motor kinesin KIF13B as a Dvl2 interacting partner (see Materials and Methods). KIF13B, containing the microtubule-binding CAP-Gly domain, has been previously implicated in neuronal polarity (Horiguchi et al., 2006) and the migration of germ cells in Xenopus embryos (Tarbashevich et al., 2011). We confirmed the biochemical interaction of Dvl2 and KIF13B by immunoprecipitation of tagged proteins from lysates of transfected HEK293T cells (Fig. 4A). Based on these observations, we hypothesized that KIF13B might be involved in the control of radial cell intercalation by PCP proteins.

Fig. 4.

KIF13B is a Disheveled-interacting protein that synergizes with PCP signaling to regulate MCC intercalation. (A) KIF13B co-immunoprecipitates with Disheveled. HEK293T cells were transfected with Myc-Dvl2 and FLAG-KIF13B. KIF-13B complexes were pulled down from cell lysates using anti-FLAG antibodies coupled to agarose, and both KIF13B and Myc-Dvl2 were visualized by western blotting. (B) Depletion of KIF13B affects MCC integration into the superficial cell layer. Control MO or KIF13B MO (40 ng each) were injected into 8-cell stage embryos. MCC migration was analyzed in cryosections of stage 22 embryos by immunostaining for centrin. Depletion of KIF13B induced MCC migration defects (arrow). Scale bar is 15 μm. (C). Quantification of MCC migration defects. Rescue experiments were performed by co-injecting KIF13B MO with 150 pg of human MYC-tagged KIF13B RNA. Data are presented as means ± s. e. Statistical significance was determined by the Student's t-test (*, p<0.05). (D) Synergistic effects of KIF13B, Vangl2 and Disheveled on MCC intercalation. Eight-cell stage embryos were injected with Vangl2 MO (5 ng), KIF13B MO (10 ng), or Xdd2 mRNA (0.25 ng), separately or in combination as indicated. MCC migration was quantified as in (C).

We next assessed whether KIF13B functions in MCC intercalation. Embryos were depleted of KIF13B using a previously characterized MO (Tarbashevich et al., 2011). The epidermis of KIF13B MO-injected embryos, but not of control MO-injected embryos, revealed a defect in MCC migration (Fig. 4B and C). This defect could be partially rescued by the injection of human KIF13B RNA (Figs. 4C and S4). To test whether KIF13B might functionally interact with the PCP pathway, a low dose of KIF13B MO was injected together with a low dose of Vangl2 MO or Xdd2 RNA. Coinjection of KIF13B MO with Vangl2 MO or Xdd2 RNA caused enhanced defects as compared to separate injections, indicating that KIF13B synergizes with PCP proteins to control MCC migration (Fig. 4D). Together, these studies suggest that KIF13B acts by promoting PCP signaling.

3.4. Polarized distribution of the Vangl2/Pk complex in deep layer neuroectoderm

MCC progenitor integration into the superficial epidermal cell layer is one example of radial cell intercalation. Radial intercalations are also known to take place in embryonic ectoderm during early epiboly and later, during neural tube formation (Davidson and Keller, 1999; Hartenstein, 1989; Keller, 1991). To further evaluate a general role for PCP signaling in radial cell interacalations during neural tube development, we first assessed the distribution of Vangl2/Prickle complexes in the deep layer neuroectodermal cells (Fig. 5A). Embryos were coinjected with low doses of HA-Vangl2 RNA (0.05–0.1 ng) and Pk RNA (0.05 ng) to avoid potential defects in neural plate morphogenesis. Membrane-targeted GFP RNA (GFP-CAAX, 0.1 ng) was used to trace cell boundaries. Examination of individual mosaically-labeled cells revealed the striking polarization of Vangl2/Pk complexes near the outer surface of inner layer cells at stage 15 (Fig. 5B and C). In the absence of Pk, HA-Vangl2 was not polarized, suggesting that in cells overexpressing Vangl2, Pk becomes the limiting component of the PCP protein complex (Fig. 5D), consistent with our recent observations (Ossipova et al., 2015b). Occasionally, we observed that individual deep layer cells labeled with GFP-CAAX RNA contained a thin polarized cell protrusion pointing to the apical surface, possibly reflecting the onset of their intercalation into the superficial layer (Fig. 5E). Vangl2/Pk complexes were observed in such protrusions in single cells (Fig. 5F and F′). Similar results were obtained with a CFP-Vangl2 construct containing a different tag. The observed localization of Vangl2/Pk complexes likely reflects the existing polarity of endogenous proteins, as evidenced by immunostaining of embryo sections with a Vangl2-specific antibody (Fig. 5G), which has been recently characterized by our group (Ossipova et al., 2015a). The appearance of the Vangl2/Pk-complex-containing monopolar protrusions directed towards the apical surface suggests that deep layer cells undergo radial intercalations, driven by PCP signaling.

Fig. 5.

Polarized distribution of PCP proteins in embryonic neuroectoderm. (A) Scheme of neurula stage embryo and a transverse section of the Xenopus neural plate at the hindbrain level. Dashed line indicates section plane. The boxed area indicates an approximate location of images. (B) Neural plate section with apically directed Vangl2/Pk complexes shown at low magnification. Eight-cell embryos were injected into animal-dorsal blastomeres with RNAs encoding HA-Vangl2, Pk and GFP-CAAX (0.1 ng each). The embryos were cultured until stage 15, cross-sectioned and immunostained with indicated antibodies. The apical surface is at the top as indicated by dashed lines. Midline (M) is to the right. Arrows point to the polarized Vangl2/Pk complex in deep neuroctodermal cells. (C–C″) The localization of the Vangl2/Pk complex in embyos coinjected with HA-Vangl2 and Pk RNAs as compared to the lineage tracer GFP-CAAX. np, neural plate; s, somite; n, notochord. Scale bar in C” is 10 μm, also applies to (D–D″).

(D–D″) HA-Vangl2 is not significantly polarized in the neural plate in the absence of overexpressed Pk. (E) A protrusion directed towards the outer surface (arrow) is detectable in a single GFP-CAAX-marked cell. (F, F′) A single deep layer cell containing an apical protrusion with enriched Vangl2/Pk complex (arrow). Scale bar is 10 μm, also applies to E, G. (G) Staining of neuroectoderm with anti-Vangl2 antibodies reveals apically enriched endogenous Vangl2 in an inner layer cell.

3.5. PCP proteins regulate the intercalatory behavior of deep layer ectoderm cells

Since PCP proteins become polarized along the apicobasal axis in inner neuroectoderm cells, we hypothesized that Vangl2 may function in radial cell intercalations. To test this hypothesis, 4–8-cell embryos were unilaterally injected with Vangl2 MO and GFP RNAs as lineage tracer and tissue morphology was studied in stage 15–17 neurulae. Cryosections revealed the failure of the neural fold to close at stage 17, consistent with the known function of Vangl2 in neural tube closure. We also observed a concomitant increase in the thickness of the neural plate at stage 15as a result of Vangl2 MO injections (Fig. 6A and B). Whereas the control un-injected or control MO-injected areas of the neural plate consisted of two-to-three layers of elongated cells at the hindbrain level, the corresponding neural plate area depleted of Vangl2 contained three to five layers of non-polarized cells with more round morphology.

Fig. 6.

Radial intercalation of deep layer cells is inhibited in embryos deficient in PCP signaling. Four-to-eight-cell embryos were unilaterally injected into prospective dorsal or ventral ectoderm with control MO (A, CO MO, 15–20 ng) or Vangl2 MO (B, 15–20 ng). Coinjected GFP RNA (0.1 ng) is a lineage tracer. Embryos were cross-sectioned at stage 15 and immunostained for β-catenin to mark cell boundaries and GFP to identify the injected side of the embryo. Dashed lines mark notochord and somite boundaries. Vertical white bars indicate the thickness of the neural plate, which is increased in Vangl2-depleted tissues. Scale bar is 20 μm. M, midline (dashed line); Np, neural plate; s, somites; n, notochord. (C–E) Thickness of non-neural ectoderm is controlled by PCP signaling. (C) Injected embryo shown at low magnification. White boxed areas approximately correspond to those shown at higher magnification in C′ (injected side) and C″ (control uninjected side). Scale bar is 100 μm in C, 20 μm in C′, also applies to C″, D, E. Abbreviations: e, ectoderm; n, notochord; s, somites. (D) Control MO does not change ectoderm morphology. (E) The effect of Vangl2 MO on ectoderm thickness in stage 10.5 embryo. Up to 5–6 cell layers are visible after Vangl2 depletion (white bar) as compared with the uninjected side, containing 2 to 4 cell layers. Bar, 40 μm. (F–H) Thick ectoderm in embryos expressing the dominant interfering Disheveled constructs Xdd1 and Xdd2. Shown are representative sections of a control embryo (F) or embryos injected with 1 ng of Xdd1 (G) or Xdd2 (H) RNA into a ventral animal blastomere and immunostained for GFP and β-catenin to mark cell boundaries. White bars indicate ectoderm thickness (C–H).

In addition to the increase in neural plate thickness, we observed that nonneural ectoderm of stage 15 embryos contained three to five cell layers at the Vangl2 MO-injected side, as compared to two cell layers at the uninjected side (Fig. 6C and D). This defect was partly rescued by the injection of Vangl2 RNA (Fig. 1H–J). Pk3 depletion also increased the number of epidermal layers, causing the appearance of thick epidermis (Fig. 3A and B). Since no differences were reported in the number of cell divisions between normal and Vangl2-depleted tissues in Xenopus embryos (Darken et al., 2002), it is unlikely that the thickness of ectoderm originates from a primary defect in cell division.

Based on the above observations, we propose that the altered thickness of the epidermis in Vangl2 and Pk3 morphants reflects a role of these PCP proteins in early radial intercalations that occur during epiboly both in neural and non-neural ectoderm. In Xenopus, epiboly takes place during late blastula stages and throughout gastrulation to allow ectoderm spreading over the entire embryo surface (Keller, 1991). Consistent with this view, the effects of Vangl2 depletion on ectoderm thickness were observed as early as stage 10.5 (Fig. 6E). Reinforcing this possibility, sections of embryos microinjected with Xdd1 and Xdd2 RNAs contained multiple ectodermal layers as compared to the two layers of the uninjected control tissues (Fig. 6F–H). Collectively, our observations support the hypothesis that PCP signaling has a role in radial cell intercalation.

4. Discussion

Radial cell intercalations represent an important model for understanding directed cell migration in vivo and is relevant to cancer mechanisms. In the Xenopus larval skin, radial intercalation is essential for the development of several cell types, including MCCs (Deblandre et al., 1999), ionocytes (Dubaissi and Papalopulu, 2011) and the small secretory cells (Dubaissi et al., 2014; Walentek et al., 2014). Our study demonstrates unique apically-directed localization of Vangl2/Pk complexes in radially intercalating cells in vivo, however, its relationship to PCP protein distribution in migrating cancer cells (Anastas et al., 2012; Luga et al., 2012; Zhu et al., 2012) and in the plane of the epithelium is currently unknown. We also show that MCCs fail to integrate into the superficial cell layer when the PCP pathway is inhibited. This cell intercalation defect may contribute to abnormal basal body apical docking and defective cilio-genesis previously documented for the Xenopus epidermis depleted of Vangl2 and Disheveled (Mitchell et al., 2009; Park et al., 2006, 2008). The earlier studies focused on the analysis of MCCs that reached the outer cell layer and did not report intercalation defects, possibly due to less efficient interference with PCP signaling. Of note, genetic mutants for Disheveled and Vangl functions in mice and zebrafish do not display nodal cilia defects (Borovina et al., 2010; Hashimoto et al., 2010; Song et al., 2010), indicating that these proteins per se are not necessary for ciliary growth. Although we only studied MCCs, it is likely that other radially intercalating cells of the epidermis, such as ionocytes and small secretory cells (Dubaissi and Papalopulu, 2011; Dubaissi et al., 2014; Walentek et al., 2014), would be similarly affected in PCP morphants.

In the frog epidermis, PCP proteins may function in radial intercalations by coordinating the position of MCCs relative to their neighboring cells or by affecting MCC intrinsic polarization. One cellular process that underlies MCC intercalatory behavior is the alignment of microtubules needed for apical positioning of basal bodies (Werner et al., 2014). In the attempt to understand how this microtubule-dependent process is regulated, we identified the motor kinesin KIF13B, which has been previously shown to control germ cell migration (Tarbashevich et al., 2011) and neuronal polarity (Horiguchi et al., 2006). We found that KIF13B is essential for MCC migration, suggesting that it represents the cellular machinery responsible for the microtubule-dependent trafficking. Since KIF13B physically and functionally interacts with Disheveled, this kinesin might mediate the vesicular trafficking of PCP proteins themselves. Similarly, Rab11-positive endosomes are required for MCC intercalation (Kim et al., 2012) and could be involved in PCP protein trafficking (Devenport et al., 2011; Mahaffey et al., 2013; Ossipova et al., 2014). Other critical cargo molecules for Rab11 vesicles may be E-cadherin or N-cadherin (Desclozeaux et al., 2008; Piloto and Schilling, 2010), which have been also implicated in cell intercalations (Hong and Brewster, 2006; Kane et al., 2005; Song et al., 2013; Warrington et al., 2013). Together, our results suggest that KIF13B is involved in PCP signaling by targeting PCP proteins, such as Disheveled, to their specific location in the cell. This hypothesis is consistent with the requirement for microtubule dynamics and polarity in PCP-dependent cell polarization in Drosophila and mammalian cells (Matis et al., 2014; Shimada et al., 2006; Vladar et al., 2012).

In addition to the involvement in skin morphogenesis and MCC migration, our findings establish a more general role for vertebrate PCP proteins in radial cell intercalations. We reveal highly polarized Vangl2/Pk complexes in deep layer neuroectoderm cells. The Vangl2/Pk complex is enriched in monopolar protrusions, which are directed towards the outer surface and are likely to function in radial intercalations. Of note, medially oriented monopolar protrusions of deep layer neuroectoderm cells have been previously described and hypothesized to play a role in mediolateral intercalations (Elul and Keller, 2000). Since radial cell intercalation is mainly observed at late stages of neural tube closure (Davidson and Keller, 1999), we propose that the observed defects of PCP protein depletion reflect the early process of cell spreading during gastrulation (Keller, 1980; Warga and Kimmel, 1990). Consistent with their function in epiboly, we demonstrate the requirement of Vangl2 and Pk3 for the thinning of embryonic ectoderm as early as stage 10.5. Similar function was recently proposed for GEF-H1, a Rho-specific GEF (Itoh et al., 2014). Additional studies are warranted to assess how PCP proteins cross-talk with GEF-H1, which has been shown to physically interact with Disheveled and Daam1 (Tsuji et al., 2010), as well as components of the vesicular trafficking machinery (Pathak et al., 2012).

Another possible mechanism that explains the involvement of PCP proteins in radial intercalations is through the control of mitotic spinde orientation. Spindle positioning is tightly coordinated during epiboly in Xenopus ectoderm (Woolner and Papalopulu, 2012), and PCP signaling controls the cell division plane in the zebrafish epiblast during gastrulation (Gong et al., 2004; Morin and Bellaiche, 2011) and during neural tissue development in both zebrafish and mammals (Ciruna et al., 2006; Lake and Sokol, 2009; Segalen et al., 2010). Of note, Rab11 is similarly involved in the positioning of the mitotic spindle in mammalian cells (Hehnly and Doxsey, 2014). Branching morphogenesis and tubulogenesis have been also linked to oriented cell divisions, therefore, morphogenetic defects in PCP mutants could be associated with the abnormal orientation of the mitotic spindle (Hukriede and Dawid, 2011; Miller et al., 2011; Yates et al., 2010a, 2010b). These potential mechanisms reiterate the important roles of both cell division and cell shape in the regulation of morphogenesis (Xiong et al., 2014).

The proposed general function of PCP signaling in radial cell intercalations helps to understand the established participation of PCP proteins in the processes that involve kidney or lung tubulo-genesis and branching morphogenesis, and during nervous and vascular system development (Miller et al., 2011; Tissir and Goffinet, 2013; Yates et al., 2010a, 2010b). Our observations are likely to be extended to other models, such as radial cell intercalations in zebrafish (Carreira-Barbosa et al., 2009; Yin et al., 2008). Supporting our conclusions, zebrafish embryos depleted of the core PCP protein Celsr/Flamingo revealed epiboly defects (Carreira-Barbosa et al., 2009), although maternal zygotic trilobite/Vangl2 mutants undergo normal epiboly (Ciruna et al., 2006). Therefore, further studies are warranted to examine the radial cell intercalation behavior in zebrafish mutant embryos. In Drosophila embryos, radial cell intercalations are essential for germ band extension, however, this process does not require core PCP signaling (Zallen, 2007; Zallen and Wieschaus, 2004). Our findings suggest that the PCP system has evolved in chordates to take up a general role in cell intercalations, extending the number of developmental processes that involve PCP proteins. This expanded role underscores the significance of vertebrate PCP proteins as factors contributing to the multitude of human diseases, including neural tube defects, tumor invasiveness, and various ciliopathies.