PTK7 localization and protein stability is affected by canonical Wnt ligands

Wnt proteins are important for embryonic development and adult tissue homeostasis, and these distinct functions require a precise fine-tuning of downstream signaling events. The demonstration that a single Wnt ligand can activate distinct signaling pathways depending on its choice of receptor (Mikels and Nusse, 2006), suggested that receptor context determines signaling output. It is now acknowledged that combinatorial co-receptor complexes provide a molecular code by which Wnt ligands can cause distinct cellular responses (Niehrs, 2012; van Amerongen et al., 2008). For example the canonical β-catenin-dependent Wnt signaling pathway is activated by binding of Wnt ligands to members of the Frizzled receptor family and the low-density lipoprotein receptor-related proteins (LRP5 or LRP6) (Tamai et al., 2000; Wehrli et al., 2000). Wnt binding results in membrane recruitment of a destruction complex, leading to the stabilization of β-catenin and expression of β-catenin-dependent target genes (MacDonald and He, 2012; MacDonald et al., 2009). In contrast, non-canonical β-catenin-independent Wnt signaling pathways use alternative co-receptors like the receptor tyrosine kinase-like orphan receptor 2 (Ror2) or protein tyrosine kinase 7 (PTK7) (Lu et al., 2004; Nomachi et al., 2008; Schambony and Wedlich, 2007). The best-characterized non-canonical Wnt signaling pathway is the planar cell polarity (PCP) pathway, which determines the coordinated polarity of cells in the plane of an epithelial tissue, thereby regulating diverse developmental processes (Simons and Mlodzik, 2008; Vladar et al., 2009; Wallingford, 2012). Activation of PCP signaling modifies the cytoskeleton and changes cell polarity, for example, by activating small GTPases or JNK-dependent transcription factors (Anastas and Moon, 2013) like ATF2 and their respective target genes (Ohkawara and Niehrs, 2011; Schambony and Wedlich, 2007). Thus, receptor context is crucial to determine signaling outcome.

A Wnt co-receptor of particular interest is the evolutionarily conserved transmembrane receptor PTK7. Originally identified as a gene upregulated in colon carcinomas and named colon-carcinoma kinase 4 (CCK4) (Mossie et al., 1995), PTK7 was later shown to function in a variety of developmental and physiological processes, including the determination of PCP, and the control of cell migration and invasion as well as regeneration (Berger et al., 2017; Dunn and Tolwinski, 2016; Peradziryi et al., 2012). PTK7 is a single-pass transmembrane receptor with extracellular immunoglobulin domains and an intracellular evolutionarily conserved kinase homology domain, which lacks catalytic activity (Kroiher et al., 2001; Miller and Steele, 2000). In vertebrates, PTK7 plays a role in the regulation of PCP, and its loss of function results in classical PCP phenotypes, including convergent extension defects, disruption of stereociliary bundle orientation and inhibition of neural crest cell migration (Hayes et al., 2013; Lu et al., 2004; Paudyal et al., 2010; Shnitsar and Borchers, 2008; Yen et al., 2009). Consistent with a role in activation of PCP signaling, PTK7 has been shown to interact with Wnt ligands as well as the Frizzled 7 (Fz7; also known as Fzd7) and Ror2 receptors (Linnemannstöns et al., 2014; Martinez et al., 2015; Peradziryi et al., 2012; Podleschny et al., 2015; Shnitsar and Borchers, 2008). Furthermore, PTK7 has been reported to recruit Dishevelled (Dsh) proteins to the plasma membrane and to activate JNK and ATF2-dependent signaling (Martinez et al., 2015; Peradziryi et al., 2011; Shnitsar and Borchers, 2008). However, PTK7 function is likely not limited to the regulation of non-canonical Wnt PCP signaling, because PTK7 has also been shown to interact with components of the canonical Wnt pathway including Wnt ligands, the LRP6 receptor and β-catenin (Bin-Nun et al., 2014; Peradziryi et al., 2011; Puppo et al., 2011). Recently, we demonstrated that PTK7 interacts with canonical Wnt ligands Wnt3a and Wnt8, but not non-canonical Wnt5a or Wnt11. Overexpression of PTK7 blocked activation of canonical Wnt signaling by Wnt3a and Wnt8 in Xenopus double axis and luciferase assays. Conversely, PTK7 loss of function activated canonical Wnt activity, suggesting that PTK7 inhibits canonical Wnt signaling (Peradziryi et al., 2011). These findings are also supported by PTK7 loss-of-function studies in zebrafish (Hayes et al., 2013), but are in conflict to others reporting activation of canonical Wnt signaling by PTK7 (Bin-Nun et al., 2014; Puppo et al., 2011). Thus, the specific role of PTK7 in the regulation of Wnt signaling pathways is controversial.

A conundrum is that PTK7 regulates PCP signaling; however, it also serves as a receptor for canonical Wnt ligands. Our previous data indicated that PTK7 interacts with canonical Wnt ligands, leading to inhibition of canonical Wnt signaling (Peradziryi et al., 2011). Currently, it remains unclear how PTK7 is affected by binding to canonical Wnt ligands. Here, we further analyze this by studying PTK7 protein localization and stability in the presence of different Wnt ligands. We find that canonical Wnt ligands lead to caveolin-mediated endocytosis and lysosomal degradation of PTK7, while the localization of PTK7 is not affected by the presence of non-canonical Wnt ligands. Thus, our data suggest that PTK7 is removed from the cell membrane in presence of canonical Wnt ligands. Thereby a Wnt morphogen gradient could direct the cellular polarization of PTK7-expressing cells. Conversely, PTK7 can also trap canonical Wnt ligands and accordingly inhibit their function. Thus, the interaction of canonical Wnt ligands and PTK7 can result in a mutual inhibition, and might be a mechanism to define border regions of active canonical versus non-canonical signaling or for directed cellular polarization in the wider sense.

Recent evidence indicates that PTK7 is a Wnt receptor; however, it remains unclear how the PTK7 protein is affected by this interaction. Previously, we have shown that PTK7 interacts with canonical Wnt3a and Wnt8, but not with non-canonical Wnt5a and Wnt11 (Peradziryi et al., 2011). As ligand binding can cause receptor endocytosis, we analyzed whether PTK7 protein localization is affected by canonical Wnt ligands. To this end PTK7–GFP was stably expressed in MCF7 cells and its localization in presence or absence of recombinant Wnt proteins was analyzed by confocal microscopy. In the absence of Wnt ligands, PTK7 is predominantly localized at the cell membrane (Fig. 1A; Fig. S1). In contrast, treatment with canonical Wnt3a caused a significant shift of the PTK7 protein from the plasma membrane to the cytoplasm. This translocation did not occur in cells treated with non-canonical Wnt5a, where the PTK7 localization pattern was comparable to that in untreated cells. Next, we asked whether the kinase domain of PTK7, which is required for Dsh recruitment (Shnitsar and Borchers, 2008; Wehner et al., 2011), is necessary for Wnt-mediated PTK7 translocation. Like full-length PTK7, a PTK7 deletion mutant lacking the kinase homology domain (ΔkPTK7) shifted from the plasma membrane to the cytoplasm in the presence of Wnt3a but not in the presence of Wnt5a (Fig. 1B). Thus, the Wnt3a-mediated translocation of the PTK7 protein is independent of its kinase homology domain. To further verify the Wnt3a-dependent shift of the PTK7 receptor from the plasma membrane to the cytoplasm, cell surface biotinylation analyses were performed (Fig. 1C). The fraction of membrane-localized PTK7 and its kinase deletion mutant was determined in untreated cells versus cells treated with Wnt3a or Wnt5a. Consistent with our previous findings, the amount of cell surface PTK7–GFP and ΔkPTK7–GFP significantly decreased after Wnt3a treatment, but remained unchanged after Wnt5a stimulation (Fig. 1C,D).

In order to perform live-cell imaging of the Wnt-mediated PTK7 internalization, total internal reflection fluorescence (TIRF) microscopy was used. TIRF microscopy has the advantage that the basal plasma membrane and the cytoplasmic regions immediately beneath the plasma membrane can be visualized, allowing the analysis of PTK7 vesicle formation and internalization in response to Wnt3a or Wnt5a. MCF7 cells expressing PTK7–GFP were imaged before and after the addition of canonical Wnt3a or non-canonical Wnt5a (Movies 1,2). While few PTK7-positive vesicles with limited mobility were observed in PTK7–GFP-expressing cells, Wnt3a treatment caused a significant increase in the number of PTK7-positive vesicles (Movie 1; Fig. 2A). These vesicles appeared larger and more mobile compared to those of untreated controls (Fig. 2B,C). Furthermore, we also observed a significant increase in the internalization of PTK7-positive vesicles (Fig. 2A). An example of the formation and internalization of a PTK7-positive vesicle in presence of Wnt3a is shown in Fig. 2D. In contrast, cells treated with Wnt5a resembled untreated control cells (Movie 2; Fig. 2A–C). In summary, these data show that in the presence of canonical Wnt ligands membrane-localized PTK7 is internalized and accumulates in vesicle-like structures, while non-canonical Wnt ligands do not affect PTK7 localization.

As PTK7 enters the cell via a Wnt-mediated mechanism, we would expect that the Wnt ligand likely colocalizes with PTK7 in the cytoplasm. To clarify this, MCF7 cells expressing PTK7–RFP were cultured together with cells expressing and secreting canonical Wnt2b–GFP or non-canonical Wnt5a–GFP. Subsequently, colocalization was assessed in living cells (Fig. 3A). As tagging of Wnt proteins may affect their function, we used here GFP-tagged Wnt constructs that had previously been shown to be functionally active in canonical and non-canonical Wnt signaling, respectively (Holzer et al., 2012; Wallkamm et al., 2014). Indeed, we found evidence that a PTK7–Wnt2b complex translocates to the cytoplasm. Secreted Wnt2b–GFP was found to colocalize with PTK7–RFP in vesicle-like structures in the cytoplasm (Fig. 3B,C). An orthogonal cell view confirms that these PTK7–Wnt2b-positive vesicles are located in the cytoplasm (Fig. S2A). In contrast, colocalization of PTK7–RFP with Wnt5a–GFP was only rarely observed (Fig. 3B,C; Fig. S2B). Taken together, these data indicate that PTK7 interacts with canonical Wnts at the plasma membrane and subsequently enters the cell as a PTK7–Wnt complex. In contrast, non-canonical Wnt5a, which does not interact with PTK7, does not affect PTK7 membrane localization.

PTK7 is a Wnt co-receptor that interacts with other known Wnt receptors (Bin-Nun et al., 2014; Martinez et al., 2015; Peradziryi et al., 2011; Podleschny et al., 2015). As the Fz7 receptor is required for the interaction of PTK7 with canonical Wnt ligands (Peradziryi et al., 2011), we analyzed whether the Wnt-mediated translocation of PTK7 is dependent on Fz7. To this end, we used Xenopus ectodermal (animal cap) cells expressing PTK7–Myc and combined them either with control or Wnt-expressing animal cap cells to create larger cell aggregates (Fig. 3D,E). Confirming our mammalian cell culture data, we observe that PTK7 is localized at the plasma membrane in control cells, but is internalized from the membrane in the presence of Wnt2b (Fig. 3Da). PTK7 localized in vesicle-like structures, where it also colocalized with Wnt2b. Conversely, Wnt5a did not affect the membrane localization of PTK7. In contrast, co-injection of PTK7 with a morpholino oligonucleotide (MO), which blocks the translation of the Fz7 protein, significantly inhibited the Wnt2b-mediated internalization of PTK7 (Fig. 3Db,E). This suggests that Fz7 is required for the Wnt2b-mediated internalization of PTK7. As PTK7 can also interact with the non-canonical Wnt receptor Ror2 (Martinez et al., 2015; Podleschny et al., 2015), we also tested whether Ror2 is required. However, Ror2 loss of function did not affect the Wnt2b-mediated translocation of PTK7 (Fig. 3Dc,E). Compared to the results obtained with MCF7 cells (Fig. 3C), fewer vesicles per cell were detected in Xenopus ectodermal cells (Fig. 3E). Furthermore, in addition to an increase in Wnt and PTK7 double-positive vesicles there is also an increase in the PTK7-positive but Wnt-negative vesicle population after Wnt2b-treatment, which was not observed in control or Wnt5a-treated cells. An explanation for these observations is that, in contrast to the co-culture experiment using MCF7 cells, the Xenopus ectodermal cells were incubated for a longer time period in the presence of Wnt-secreting cells. Thus, Wnt2b and PTK7 double-positive vesicles may have lost the Wnt2b signal over time. To test this hypothesis, we examined whether we would obtain similar results when we also prolonged the co-culture experiment with MCF7 cells. Indeed, we find that this is the case. In cells incubated with Wnt ligands for 2 or 4 h, the number of PTK7-positive but Wnt-negative vesicles per cell is comparable between the different treatments (Fig. S3). However, for the 8 h co-culture this population increases in the cells co-cultured with Wnt2b in comparison to cells co-cultured with Wnt5a. This suggests that Wnt2b and PTK7 double-positive vesicles lose their Wnt2b signal over time, either because Wnt2b is faster degraded or it leaves the vesicles. Taken together, these data suggest that PTK7 interacts with canonical Wnt ligands and requires Fz7 to enter the cytoplasm, but this internalization is independent of the non-canonical Wnt receptor Ror2.

Ligand-mediated endocytosis is a common process through which transmembrane receptors are transported into the cytoplasm. To examine whether PTK7 is internalized in the presence of canonical Wnts through a clathrin- or caveolin-mediated route, MCF7 cells expressing PTK7–GFP were co-stained with antibodies detecting endogenous caveolin-1α or clathrin. In control cells or cells treated with Wnt5a, PTK7–GFP and caveolin-1α colocalized at the membrane, but cytoplasmic colocalization was rarely observed (Fig. 4A,B). Conversely, in the presence of Wnt3a, PTK7–GFP colocalized with caveolin-1α in vesicle-like structures in the cytoplasm. In addition, we analyzed whether the kinase deletion mutant of PTK7, ΔkPTK7–GFP, colocalizes with caveolin-1α in the cytoplasm. Consistent with the findings for full-length PTK7–GFP, ΔkPTK7–GFP colocalized with caveolin-1α in the cytoplasm after Wnt3a but not after Wnt5a treatment (Fig. S4A,B). In addition to the observed colocalization of fluorescent PTK7 and caveolin-1α in human and Xenopus cells, biochemical interaction of these proteins was analyzed by performing co-immunoprecipitation experiments. In lysates of MCF7 cells Myc-tagged PTK7 proteins (full-length PTK7 and ΔkPTK7) co-precipitated HA-tagged caveolin-1α (Fig. 4C) and vice versa (Fig. S4C), suggesting that caveolin-1α interacts with PTK7 independently of its kinase homology domain. Taken together, these data indicate that PTK7 is endocytosed by a caveolin-mediated pathway.

To assess the role of clathrin-mediated endocytosis of PTK7, endogenous clathrin was visualized in MCF7 cells expressing PTK7–GFP by immunostaining. Colocalization of PTK7–GFP with clathrin was infrequently detected (Fig. S5). Stimulation of the cells with Wnt3a, only led to a minor increase in the PTK7–clathrin colocalization in comparison to that seen in untreated or Wnt5a-treated cells (Fig. S5A,B). Similar results were obtained for ΔkPTK7–GFP in respect to colocalization with clathrin (Fig. S5C,D). These experiments confirm that Wnt3a-induced PTK7 internalization is mediated by caveolin-1α rather than by clathrin. In contrast to caveolin-1α, which is an integral membrane protein and therefore stably associated with the plasma membrane, clathrin localizes dynamically to the plasma membrane. Hence, after vesicle release into the cytoplasm clathrin disassembles quickly from the vesicular membrane (Mundy et al., 2012; Parton and del Pozo, 2013). Consequently, cytoplasmic colocalization of a cargo protein with caveolin-1α is clearly detectable. However, clathrin colocalization might not be detected as it may have dissociated from the vesicles. To further analyze whether PTK7 receptor internalization is exclusively mediated by caveolin, its colocalization with PTK7 was studied by TIRF microscopy. This allows the visualization of the plasma membrane and adjacent cytoplasmic regions and would therefore detect clathrin if it was still attached to the vesicles. MCF7 wild-type cells stained with antibodies detecting endogenous PTK7, caveolin-1α and clathrin, respectively, were analyzed by TIRF microscopy. PTK7-positive dots were analyzed for colocalization with either caveolin-1α or clathrin, and quantified through determining the Pearson correlation coefficients (PCCs) of single dots. Consistent with the previous findings obtained using confocal microscopy, colocalization of PTK7 with caveolin-1α increased significantly upon treatment with Wnt3a (Fig. 5A,C). In contrast, PTK7-clathrin colocalization remained largely unchanged in the presence or absence of Wnt3a (Fig. 5B,C). These results confirm that PTK7 internalization is mediated by caveolin-1α and seems to be independent of the clathrin-mediated endocytosis pathway.

Finally, to examine the role of caveolin for PTK7 internalization, MCF7 cells were treated with methyl-β-cyclodextrin (MβCD). MβCD removes cholesterol from membranes leading to a disruption of lipid rafts and consequently destroys caveolae (Rodal et al., 1999). Inhibition of caveolin-mediated endocytosis by MβCD treatment prevented the Wnt3a-induced PTK7 (Fig. 6A,B) internalization. Consistent with these findings, the Wnt3a-dependent endocytosis of the PTK7 construct lacking the kinase homology domain, ΔkPTK7, was also abolished in cells treated with MβCD (data not shown). Furthermore, loss of function of caveolin-1α in Xenopus ectodermal cells also prevented the Wnt2b-mediated endocytosis of PTK7. Ectodermal aggregates composed of cells co-injected with PTK7–Myc and control MO in combination with Wnt2b-expressing cells showed PTK7 internalization. As expected, PTK7 largely remained at the membrane in aggregates combined with control or Wnt5a-expressing cells (Fig. 6C). This is also reflected by the quantification of Wnt and PTK7 double-positive and PTK7-positive vesicles in these aggregates (Fig. 6D). In contrast, co-injection of caveolin-1α MO with PTK7–Myc significantly decreased the number of Wnt and PTK7 double-positive and PTK7-positive vesicles in Wnt2b-containing aggregates and PTK7 remained at the membrane (Fig. 6C,D). Thus, these data indicate that the Wnt-mediated PTK7 endocytosis is dependent on caveolin-1α.

Next, we examined whether caveolin-1α-mediated PTK7 internalization affects PTK7 protein stability. MCF7 cells were incubated with increasing concentrations of canonical Wnt3a or non-canonical Wnt5a, and endogenous PTK7 protein levels normalized to actin were determined. Endogenous PTK7 protein intensity decreased in the presence of increasing concentrations of Wnt3a but not Wnt5a (Fig. 7A,B). Similar results were obtained when testing the effect of different canonical (Wnt3a and Wnt8) and non-canonical (Wnt11) Wnt proteins on the PTK7 protein in Xenopus ectodermal explants (Fig. S6). To analyze whether the decrease of PTK7 protein in the presence of canonical Wnt is the result of protein degradation, lysosomal degradation was inhibited using chloroquine. Chloroquine raises the lysosomal pH and thereby inhibits lysosomal enzymes that require an acidic pH for their activity (Steinman et al., 1983). Endogenous PTK7 levels in MCF7 cells were determined in the presence or absence of Wnt3a, and in addition when lysosomal degradation was inhibited using chloroquine. While the relative PTK7 signal intensity decreased with increasing concentrations of Wnt3a, chloroquine treatment prevented this effect (Fig. 7C,D). Similar effects were observed when using ammonium chloride to inhibit lysosomal degradation (Fig. S7), while a proteasomal inhibitor (MG 132) did not prevent PTK7 degradation (data not shown). Furthermore, immunostaining detects a significant increase in colocalization of PTK7 with the lysosomal marker Lamp-2 in intracellular vesicles of cells treated with Wnt3a, compared to that seen in Wnt5a-treated cells or controls (Fig. 7E,F). These results indicate that canonical Wnt proteins lead to endocytosis and subsequently lysosomal degradation of PTK7 in a dose-dependent manner.

Previously, we observed that PTK7 inhibits canonical Wnt signaling (Peradziryi et al., 2011); however, the molecular mechanism is unclear. Our data suggest that PTK7 interacts with canonical Wnt proteins and removes them via caveolin-mediated endocytosis from the extracellular matrix, thereby likely preventing their interaction with canonical Wnt co-receptors. Thus, to determine whether caveolin function contributes to PTK7-mediated inhibition of canonical Wnt signaling we performed Xenopus second axis assays. Secondary axes were induced by overexpressing Wnt8 on the ventral side of Xenopus embryos. As expected, co-injection of PTK7 significantly inhibited second axis induction. However, this effect was less severe if the embryos were co-injected with caveolin-1α MO (Fig. 8), indicating that PTK7 requires caveolin-1α for successful inhibition of canonical Wnt signaling. Thus, these data suggest that caveolin-mediated endocytosis supports the inhibition of canonical Wnt signaling by PTK7.

During development and tissue homeostasis Wnt signals control various cellular behaviors by activating distinct signaling pathways. The establishment of PCP is an evolutionarily conserved process in development and organogenesis, and requires the polarized localization of core PCP proteins. Wnt ligands have been shown to serve as instructive signals affecting PCP protein localization and polarity axis orientation (Gao, 2012; Yang and Mlodzik, 2015). For vertebrate PCP, a role of non-canonical Wnt ligands, like Wnt5a and Wnt11, has been well documented (Gao et al., 2011; Heisenberg et al., 2000; Kilian et al., 2003; Qian et al., 2007; Tada and Smith, 2000). Concerning the molecular mechanism, recent data indicate that Wnt gradients provide directional information to a field of cells by forming Wnt-induced receptor complexes, thereby modulating protein activity or stability (Andre et al., 2012; Gao et al., 2011). However, vertebrates express 19 Wnt ligands and diverse Wnt receptors, allowing the formation of various distinct receptor complexes. Therefore, analyzing the role of Wnt ligands and their respective receptor complexes in the establishment of vertebrate PCP will likely still hold some surprises.

In this respect, PTK7 is a very interesting Wnt co-receptor in that it regulates PCP signaling in vertebrates, but also interacts with canonical Wnt ligands. Here, we find that PTK7 interacts with canonical Wnt ligands at the plasma membrane and is subsequently endocytosed via a caveolin-mediated process. Interestingly, non-canonical Wnt ligands did not affect PTK7 localization. Consistent with this, canonical-Wnt-dependent PTK7 internalization required the Fz7 receptor, but not the non-canonical Ror2 receptor. These data are consistent with our previous findings that PTK7 interacts with canonical Wnt3a and Wnt8, but not non-canonical Wnt5a and Wnt11. Furthermore, the PTK7–Wnt interaction is not direct, but is mediated by Fz7 (Peradziryi et al., 2011); hence, Fz7 is also required for Wnt-mediated endocytosis of PTK7. The intracellular kinase homology domain of PTK7 is not required for interaction with Wnt and Fz7, and indeed we confirm here that this domain is also not necessary for Wnt-mediated endocytosis of PTK7. As the kinase domain recruits Dsh and PKCδ to the plasma membrane (Shnitsar and Borchers, 2008; Wehner et al., 2011), a function associated with the activation of PCP signaling (Kinoshita et al., 2003; Park et al., 2005), this suggests that the Wnt-mediated endocytosis of PTK7 is independent of its interaction with Dsh. Furthermore, we note that PTK7 is subject to lysosomal degradation in the presence of canonical Wnt ligands, while it is not affected by non-canonical Wnt ligands. Thus, canonical Wnt gradients may affect PTK7 localization and stability, thereby likely modulating its role in PCP signaling. In respect to Wnt/β-catenin signaling, we find that PTK7 inhibits this signaling pathway suggesting that PTK7 may trap canonical Wnt ligands in a non-canonical Wnt co-receptor complex, thereby preventing their interaction with receptors favoring Wnt/β-catenin signaling. Consistent with this hypothesis, PTK7 inhibits canonical Wnt-induced Xenopus double axis formation (Peradziryi et al., 2011). Interestingly, this PTK7-mediated inhibition is significantly less severe if caveolin-1α function is knocked down. Furthermore, we observed here that fluorescently labeled canonical Wnt2b colocalized with PTK7 in intracellular vesicles, while this was only rarely observed for Wnt5a. This suggests that PTK7 efficiently removes canonical Wnt ligands from the extracellular space, thereby preventing their interaction with bona fide canonical Wnt receptor complexes. Thus, the PTK7–Wnt interaction may provide a molecular mechanism of mutual regulation of PCP and Wnt signaling pathways.

Currently, we can only speculate about the endogenous role of the PTK7 receptor at the intersection of Wnt signaling pathways. PTK7 is an acknowledged regulator of vertebrate PCP and, as such, is involved in various morphogenetic processes (Hayes et al., 2013; Podleschny et al., 2015; Shnitsar and Borchers, 2008; Williams et al., 2014; Xu et al., 2016; Yen et al., 2009). Concerning the functional relevance of the canonical Wnt-mediated endocytosis of PTK7, this could be a mechanism by which a Wnt gradient affects PTK7 localization. In this respect, one would expect that PTK7 is not evenly distributed at the cell membrane, but shows some – possibly also dynamic – polarization. A system where the dynamic localization of PCP proteins has been demonstrated are migrating neural crest cells. PCP signaling components are, for example, required for contact inhibition of locomotion, which is a phenomenon whereby PCP components are transiently localized at cell contacts and mediate the change of cell polarization and the subsequent movement of cells in the opposite direction (Mayor and Theveneau, 2014). Interestingly, we also find that PTK7 is localized at these cell–cell contact zones, but is removed if cell contacts are broken (data not shown). Thus, Wnt ligands may affect PTK7 dynamics in migrating neural crest cells. For example, Wnt2b, which we have shown leads to internalization of PTK7 and colocalizes with PTK7 in intracellular vesicles, is expressed in the Xenopus branchial arches at the time of neural crest migration (Rankin et al., 2012). Neural crest cells also provide an explanation for the relevance of the inhibition of canonical Wnt signaling by PTK7. Canonical Wnt activity decreases at the onset of neural crest migration, and ectopic activation inhibits neural crest migration (Maj et al., 2016; Rabadán et al., 2016). Thus, neural crest cells require controlled basal levels of canonical Wnt signaling to enable their migration. PTK7 is expressed in migrating neural crest cells and also has the ability to inhibit canonical Wnt signaling; therefore, it could be a molecular tool to achieve these controlled levels. As PTK7 affects diverse processes ranging from embryonic morphogenesis to wound repair it likely also contributes to the mutual interaction of Wnt signaling pathways in these systems.

MCF7 (provided by the Department of Hematology, University Medical Center Göttingen, Göttingen, Germany) cell lines were grown in RPMI 1640 medium (Biochrom) supplemented with 10% fetal calf serum (FCS) and 5% penicillin-streptomycin (Biochrom) at 37°C and 5% CO₂. For the generation of stable and inducible PTK7–GFP or ΔkPTK7–GFP MCF7 cells, the tetracycline-regulated mammalian expression system T-REx™ (Invitrogen, Life Technologies) was used. Cells were transfected with the regulatory plasmid pcDNA™6/TR and selected using blasticidin (6 μg/ml). Cells that responded strongly to tetracycline were additionally transfected with the expression plasmid pcDNA™5/TO carrying the PTK7–GFP or ΔkPTK7–GFP gene. Resistant cells were selected with blasticidin and hygromycin (1 μg/ml). For cloning of PTK7–GFP into the pcDNA™5/TO, first the coding sequence of hPTK7 was amplified using the primers 5′-CACGTGGCTAGCGCCCTCAGCTCCTTTTCCTGA-3′ and 5′-GACGAATTCGCGGCTTGCTGTCCACGGT-3′, and introduced into pEGFPN1 at the NheI and EcoRI restriction sites. For cloning of ΔkPTK7–GFP into the pcDNA™5/TO, ΔkPTK7 was amplified using the primers 5′-CACGTGGCTAGCATGGGAGCTGCGCGGGGATCC-3′ and 5′-CGAGAATTCGGTGCATCTTATCACTTGTGC-3′, and introduced into the NheI and EcoRI restriction sites of the pEGFPN1 vector. From the pEGFPN1 vectors, PTK7–EGFP or ΔkPTK7–EGFP were amplified using the primers forward 5′-GTCGATATCATGGGAGCTGCGCGGGGATCC-3′ and reverse 5′-ACGGCGGCCGCCTTGTACAGCTCGTCCATGC-3′. The PCR products were introduced respectively into the EcoRV and NotI restriction sites of the pcDNA™5/TO vector. For cultivation of stable transfected MCF7 cell lines, RPMI medium was additionally supplemented with 1 μg/ml hygromycin and 6 μg/ml blasticidin. PTK7–GFP or ΔkPTK7–GFP expression in stably transfected MCF7 cell lines was induced by the addition of doxycycline to a final concentration of 1 µg/ml.

Cells were seeded on 10 mm glass coverslips (Menzel-Gläser, Thermo Scientific), and PTK7–GFP or ΔkPTK7–GFP expression was induced by the addition of doxycycline at a final concentration of 1 µg/ml for at least 15 h. For treatment with Wnt proteins, cells were washed three times with PBS, and 200 ng/ml recombinant human (rh)Wnt3a (R&D Systems, 5036-WN) or rhWnt5a (R&D Systems, 645-WN) diluted in RPMI medium was added to the cells. For inhibition of caveolin-mediated endocytosis, cells were incubated in 5 mM MβCD (Sigma-Aldrich, C4555) for 4 h before Wnt treatment. Cells were washed with PBS, fixed in 4% paraformaldehyde in PBS for 20 min at room temperature, permeabilized with 0.2% Triton X-100 in PBS and blocked in blocking buffer [1% bovine serum albumin (BSA) in PBS] for 1 h. Cells were incubated with anti-GFP (Roche, 11814460001, 1:1000; Abcam, ab290, 1:2000), anti-caveolin-1 (Abcam, ab2910, 1:500), anti-clathrin (Abcam, ab14409, 1:500) or anti-Lamp-2 (BD Bioscience, 555803, 1:500) antibodies overnight at 4°C. Cells were washed with blocking buffer and subsequently incubated with Alexa Fluor-conjugated secondary antibodies [Alexa Fluor 594-conjugated goat anti-rabbit-IgG (Life Technologies, A-11012, 1:400), Alexa Fluor 488-conjugated goat anti-rabbit-IgG (Life Technologies, A-11008, 1:400), Alexa Fluor^® 488-conjugated goat anti-mouse-IgG (Life Technologies, A-11029,1:400), Alexa Fluor 594-conjugated goat anti-mouse-IgG (Life Technologies, A-11005, 1:400)] for 1 h at room temperature. Cells were mounted in fluorescence mounting medium (Dako, Agilent Technology) supplemented with DAPI (Carl Roth) to a final concentration of 1 µg/ml. Stained cells were imaged by confocal laser-scanning fluorescence microscopy (LSM 780, Carl-Zeiss or TCS SP5, Leica Microsystems). For quantification, the PTK7 protein localization was classified into three different categories: membrane localization, membrane and cytoplasmic localization, or cytoplasmic localization. The category ‘membrane localization’ was used if PTK7 was clearly localized at the plasma membrane and weakly or not at all in the cytoplasm. An equal distribution of PTK7 at both membrane and cytoplasm was classified as ‘membrane and cytoplasmic localization’. ‘Cytoplasmic localization’ was determined as mainly cytoplasmic PTK7 localization with weak or no localization at the plasma membrane.

For co-culture experiments cells were seeded in two- or four-well Lab-Tek Chambers (Nunc™ Lab-Tek™ Chambered Coverglass, Thermo Scientific) and transfected with PTK7–RFP. Further cells were seeded on six-well plates and transfected with Wnt2b–GFP (Holzer et al., 2012) or Wnt5a–GFP (Walkamm et al., 2014). Transfections were performed using Lipofectamine^® 2000 (Life Technologies). Cells expressing Wnt2b–GFP or Wnt5a–GFP were subsequently added to the cells expressing PTK7–RFP (Podleschny et al., 2015) and incubated together for 4 h. Live-cell imaging was performed by using a spinning disc confocal microscope (AxioObserver Z1, Zeiss) with a Plan-Apochromat 63×, NA 1.40 oil objective. Protein colocalization was quantified with ImageJ software (coloc 2 plugin).

For analysis of protein colocalization, the PCC was calculated with ImageJ. PCCs for single vesicles or areas to be analyzed within each cell were determined by drawing a region of interest (ROI) of equal size. PCC values range from 1 to –1, whereby 1 reflects a perfect correlation between the pixels in two channels (here, red and green), 0 stands for a random correlation and -1 means perfect but negative correlation (Adler and Parmryd, 2010). PCC values of ≥0.5 are considered as positive. According to this, colocalization of single vesicles or cells was counted as positive if the calculated PCC was ≥0.5.

For TIRF microscopy, PTK7–GFP-expressing MCF7 cells were seeded on 42 mm glass coverslips (H. Saur Laborbedarf). For analysis of endogenous PTK7, caveolin-1α and clathrin protein expression, wild-type MCF7 cells were fixed and blocked as described above. Cells were incubated with anti-caveolin-1 (Abcam, ab2910, 1:500), anti-clathrin (Abcam, ab1440, 1:500) or anti-PTK7 (CCK4) (R&D systems, AF4499, 1:200) antibodies overnight at 4°C. As secondary antibodies, Alexa Fluor 647-conjugated chicken anti-rabbit-IgG and Alexa Fluor^® 488 goat anti-mouse-IgG were used. For TIRF live-cell imaging, PTK7–GFP expression was induced by the addition of doxycycline at a final concentration of 1 µg/ml for at least 15 h. Cells were imaged for 15 or 30 min and then either canonical Wnt3a or non-canonical Wnt5a protein was added and cells were further imaged for 15 or 30 min. TIRF imaging was performed by using a Leica DMI6000B microscope with a HCX Plan-Apochromat 100×, NA 1.47 oil objective. The penetration depth was 130 nm for the experiments using fixed cells and 150 nm for live-cell imaging.

For transfection, MCF7 cells were seeded in six-well plates with a surface area of 10 cm² per well. Transfection of PTK7–Myc, ΔkPTK7–Myc (Shnitsar and Borchers, 2008) and caveolin-1α–HA was carried out by using Lipofectamine^® 2000 Reagent (Life Technologies). Caveolin-1α was amplified by PCR from Xenopus caveolin-1α cDNA (RZPD, catalogue number IRBMp990B0725D) with the forward primer, 5′-TTGAATTCAGCATGTCTGGTGGCAAATACATAG-3′, which contains a 5′ EcoRI restriction site and the reverse primer, 5′-TTCTCGAGCACTTCTTTGCGTAAGGAA-3′, which contains a 3′ XhoI restriction site. The PCR product was cut with EcoRI and XhoI and inserted into the respective sites of the pCS2+/HA vector. At 48 h after transfection, cells were washed in Tris-buffered saline (TBS; 50 mM Tris-HCl pH 7.5 and 150 mM NaCl), scraped and lysed in NP-40 lysis buffer [50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.5% (v/v) NP-40 containing Complete EDTA-free protease inhibitor cocktail tablet (Roche)] and supplemented with SDS to a final concentration of 0.1%. A total volume of 800 µl lysate was pre-cleared for 1 h with Protein-A–Sepharose CL-4B (GE Healthcare) at 4°C with end-over-end mixing. After pre-clearing, 50 µl of the lysate were collected as input control. For the antigen–antibody reaction, supernatants were incubated with anti-HA.11 (Covance, MMS-101P, 1:150) or anti-Myc 9E10 (Sigma Aldrich, M4439, 1:250) antibodies for 2 h at 4°C. Complexes were precipitated with Protein A–Sepharose (GE Healthcare) for 1 h at 4°C, washed five times for 5 min with NP-40 lysis buffer and boiled for 5 min at 95°C in 6× Laemmli loading buffer (350 mM Tris-HCl pH 6.8, 9.3% Dithiothreitol, 30% (v/v) glycerol, 10% SDS, 0.02% Bromphenol Blue) and loaded on 10% or 12% SDS-PAGE gels.

MCF7 cells stably expressing PTK7–GFP or ΔkPTK7–GFP cultured in six-well plates were treated with 200 ng/ml rhWnt3a or rhWnt5a for 1 h. Subsequently, cells were cross-linked with 0.25 mg/ml EZ-Link-Sulpho-NHS-SS-biotin for 30 min (Thermo Scientific) and subsequently quenched with Quenching solution (Thermo Scientific). Cells were washed in TBS, and scraped and lysed in 800 µl NP-40 lysis buffer. 50 µl of the sample was collected as input control. Cell surface proteins were affinity-purified using NeutrAvidin–agarose beads (Thermo Scientific) for 2 h with end-over-end mixing. Beads were washed five times for 5 min with NP-40 lysis buffer, boiled at 95°C for 5 min in 6× Laemmli loading buffer and loaded onto 10% or 12% SDS-PAGE gels. Quantification of the relative signal intensities of the cell surface to total protein levels was performed by using ImageJ.

MCF7 cells were cultured in six-well plates for at least 15 h. For treatment with Wnt proteins, cells were washed three times with PBS before rhWnt proteins were added at different concentrations (50–400 ng/ml) diluted in RPMI medium. For inhibition of caveolin-mediated endocytosis, cells were incubated with 5 mM MβCD for 4 h prior to and during Wnt incubation. For inhibition of lysosomal degradation cells were incubated with 100 µM chloroquine (or 50 mM NH₄Cl) for four hours before and during Wnt treatment. Cells were washed three times in TBS, lysed in NP-40 lysis buffer and homogenized using a 30G syringe. Protein extracts of MCF7 cells were separated by 10% or 12% SDS-PAGE, transferred to a nitrocellulose membrane (Whatman) by electroblotting and blocked in TBST buffer [50 mM Tris-HCl pH 7.5, 150 mM NaCl and 0.5% (v/v) Tween 20] containing 5% nonfat dried milk. The following antibodies were used for detection of proteins: anti-PTK7 (CCK4) (R&D Systems, AF4499, 1:1000), anti-HA.11 (Covance, MMS-101P, 1:1000), anti-GFP (Roche, 11814460001, 1:2000), anti-Myc (abcam, ab19234, 1:1000), anti-actin (Merck Millipore, MAB1501, 1:2000) antibodies. Horseradish peroxidase-conjugated secondary antibodies used were anti-mouse-IgG (Santa Cruz Biotechnology, sc-2005, 1:5000), anti-goat-IgG (Santa Cruz Biotechnology, sc-2020, 1:10,000) and anti-rabbit-IgG (Cell Signaling, 7074, 1:2000). Proteins were detected using Pierce™ ECL Western Blotting Substrate (Thermo Scientific) and exposed to an X-ray film or Odyssey^® Fc Imaging System (LI-COR Bioscience).

Xenopus embryo microinjections were performed as described previously (Borchers et al., 2001). All animal experiments were performed according to approved guidelines. Capped sense RNA for microinjections was prepared by using the mMessage mMachine kit (Ambion, Life Technologies). The following plasmids and morpholino oligonucleotides (MOs) were used for sense mRNA synthesis: PTK7 (Shnitsar and Borchers, 2008), PTK7–Myc (Shnitsar and Borchers, 2008), Wnt2b–GFP (Holzer et al., 2012), Wnt5a–GFP (Wallkamm et al., 2014) and lacZ (Smith and Harland, 1991), caveolin-1α MO (5′-CATCTATGTATTTGCCACCAGACAT-3′, Gene Tools, LLC) and standard control MO (5′-CCTCTTACCTCAGTTACAATTTATA-3′, Gene Tools, LLC).

Axis duplication was induced by the injection of 5 pg Wnt8 mRNA into four-cell-stage embryos, marginally in one ventral blastomere. 250 pg of PTK7 RNA, 10 ng of control MO or caveolin-1α MO were co-injected. Embryos were quantified for second axis generation at early tadpole stages.

One-cell-stage Xenopus embryos were injected with 500 pg PTK7-Myc mRNA alone or in combination with 10 or 20 ng standard control MO, 20 ng caveolin-1α MO, 10 ng Fz7 MO (Abu-Elmagd et al., 2006) or 10 ng Ror2 MO (5′-GTCAGGCGAGGTAAGGGGCAACACT-3′). Additionally, one-cell embryos were injected with 100 pg Wnt2b-GFP or 100 pg Wnt5a-GFP mRNA. Ectodermal explants (animal caps) were cut at stage 8 as described previously (Wallingford and Harland, 2001). Ten PTK7–Myc-expressing animal caps were mixed with ten Wnt-expressing animal caps in 24-well plates and afterwards dissociated in Ca²⁺-free 0.8× MBS buffer [10 mM Hepes (pH 7.0), 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.82 mM MgSO₄ and 0.66 mM KNO₃] by gentle pipetting. For controls, 20 PTK7-expressing caps were used. At 1 h after dissociation, the Ca²⁺-free buffer was removed and the cells were allowed to reaggregate in 0.8× MBS [10 mM Hepes (pH 7.0), 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 0.82 mM MgSO₄, 0.41 mM CaCl₂, 0.66 mM KNO₃]. The cells were incubated until the control Xenopus embryos reached stage 14–15, fixed in MEMFA (4% formaldehyde, 0.1 M MOPS, 2 mM EGTA, 1 mM MgSO₄) for 20 min and blocked for 1 h in PTw containing 10% FCS. PTK7–Myc was stained using a Cy3-tagged anti-Myc antibody (Sigma-Aldrich, C6594, 1:100), Wnt proteins were visualized by using anti-GFP (Abcam, ab290, 1:1000) primary antibodies and secondary Alexa Fluor^® 488-conjugated goat anti-rabbit-IgG (Life Technologies, A-11008, 1:400) antibodies. The reaggregated animal caps were analyzed by spinning disk confocal microscopy (AxioObserver Z1, Zeiss). The total number of PTK7-positive vesicles in PTK7-expressing cells was determined in cells adjacent (in close proximity) to Wnt-expressing cells. Wnt and PTK7 colocalization of specific vesicles was determined by analyzing the PCC.

We thank Dietmar Gradl (Dept. of Cell and Developmental Biology, Zoological Institute, Karlsruhe Institute of Technology, Karlsruhe, Germany) for supplying plasmids. Furthermore, we thank Ingrid Bohl-Maser and Christiane Rohrbach for technical assistance and Melanie Bernhardt for excellent care of the Xenopus colony. Spinning disk experiments were performed using the Core Facility Cellular Imaging, Philipps University Marburg and TIRFM images were recorded in the Bioimaging Facility Marburg. The results on PTK7- and ΔkPTK7-mediated caveolin and clathrin endocytosis in MCF7 cells (colocalization and precipitation, TIRF imaging of fixed cells, biotinylation assay, MβCD and chloroquine treatment) as well as the Xenopus second axis experiments were previously presented as part of a PhD thesis by Hanna Berger, Georg-August University Göttingen, Germany, 2016.