Protein tyrosine kinase 7 (PTK7) is a transmembrane protein expressed in the developing Xenopus neural plate. PTK7 regulates vertebrate planar cell polarity (PCP), controlling mesodermal and neural convergent-extension (CE) cell movements, neural crest migration and neural tube closure in vertebrate embryos. Besides CE phenotypes, we now show that PTK7 protein knockdown also inhibits Wnt/β-catenin activity. Canonical Wnt signaling caudalizes the neural plate via direct transcriptional activation of the meis3 TALE-class homeobox gene, which subsequently induces neural CE. PTK7 controls meis3 gene expression to specify posterior tissue and downstream PCP activity. Furthermore, PTK7 morphants phenocopy embryos depleted for Wnt3a, LRP6 and Meis3 proteins. PTK7 protein depletion inhibits embryonic Wnt/β-catenin signaling by strongly reducing LRP6 protein levels. LRP6 protein positively modulates Wnt/β-catenin, but negatively modulates Wnt/PCP activities. The maintenance of high LRP6 protein levels by PTK7 triggers PCP inhibition. PTK7 and LRP6 proteins physically interact, suggesting that PTK7 stabilization of LRP6 protein reciprocally regulates both canonical and noncanonical Wnt activities in the embryo. We suggest a novel role for PTK7 protein as a modulator of LRP6 that negatively regulates Wnt/PCP activity.
INTRODUCTION
Wnt signaling regulates many embryonic and adult physiological processes. The most characterized Wnt-network pathway is the Wnt/β-catenin/canonical pathway. Secreted Wnt ligands bind receptor complexes composed of the Frizzled (Fz) seven transmembrane receptor and the low-density lipoprotein receptor-related protein 6 (LRP6). Following Wnt-ligand binding, LRP6 undergoes phosphorylation, recruiting Axin protein to the membrane in a disheveled (Dvl) protein-dependent manner. Dvl and Axin bind GSK3β, preventing β-catenin phosphorylation and degradation. β-catenin translocates to the nucleus, where it binds TCF/LEF proteins to directly activate target gene transcription (MacDonald et al., 2009).
LRP6 function is dependent on dimerization with Fz and the intracellular phosphorylation of its multiple PPPSPXS domains and S/T motifs (Bilic et al., 2007; MacDonald et al., 2008). Upon Wnt/Fz/LRP6 complex formation, Axin and GSK3β are recruited to the membrane, where GSK3β and CK1γ phosphorylate LRP6 (Davidson et al., 2005; Bilic et al., 2007; Zeng et al., 2008). Phosphorylation sites serve as additional docking sites for Axin and GSK3β, freeing more β-catenin. LRP6 phosphorylation triggers auto-oligomerization, creating ‘LRP6-signalosomes’ that recruit more LRP6 to amplify the Wnt signal (Niehrs and Shen, 2010).
In Xenopus, Wnt/β-catenin acts in two distinct temporal patterning pathways. Maternal activation induces the Spemann organizer in the dorsal mesoderm (Niehrs, 2004). Later zygotic Wnt/β-catenin signaling caudalizes anterior neural cells to posterior fates (Elkouby and Frank, 2010). In the posterior neural plate, both zygotic canonical and noncanonical Wnt activities regulate cell fate specification and morphogenesis. Mesodermal Wnt3a protein induces meis3 TALE-class homeobox gene expression in the posterior neural plate, which specifies hindbrain, primary neuron and neural crest fates (Salzberg et al., 1999; Gutkovich et al., 2010). Meis3 protein subsequently activates noncanonical Wnt planar cell polarity (PCP) activity, inducing posterior neural convergent-extension (CE) cell movements (Aamar and Frank, 2004). Thus, canonical Wnt signaling acts upstream of Wnt-PCP via meis3 gene expression (Elkouby et al., 2010).
LRP6 protein bridges canonical- and PCP-Wnt activities to modulate CE movements in the Xenopus neural plate; either ectopic or reduced LRP6 levels inhibit neural CE (Tahinci et al., 2007). Tightly balanced PCP activity optimizes cell polarity. CE is perturbed when PCP is inhibited, but PCP overstimulation by pathway activators also disrupts cell-polarization activity, preventing cell intercalations (Wallingford et al., 2000; Djiane et al., 2000). LRP6 protein modulates CE via Wnt-PCP activity. In excess, LRP6 inhibits PCP, yet when depleted, LRP6 overstimulates PCP; thus CE movements are inhibited (Tahinci et al., 2007; Bryja et al., 2009). Additionally, LRP6 depletion lowers canonical Wnt signaling in the neural plate, disrupting posterior neural cell fate specification, reducing CE. Two possible nonmutually exclusive mechanisms explain neural CE inhibition by LRP6. Posterior neural cells that typically undergo CE are lost, and/or a parallel inhibitory spike of Wnt-PCP activity is released by LRP6 loss of function. This complex dual-regulation of Wnt/β-catenin and PCP activities by LRP6 must be analyzed to elucidate LRP6 function in loss- or gain-of-function phenotypes.
Protein tyrosine kinase 7 (PTK7), a transmembrane protein lacking an active kinase domain, is structurally conserved among vertebrates and invertebrates (Miller and Steele, 2000). In PTK7 mutant mice, gastrulation cell movements are disrupted, neural tube closure is perturbed and stereociliary bundle orientation in the inner ear is disturbed (Lu et al., 2004; Yen et al., 2009; Paudyal et al., 2010). The Xenopus ptk7 homolog is expressed in the neural plate and its knockdown disrupts neural tube closure and CE (Lu et al., 2004). These phenotypes resemble known Wnt-PCP mutations, suggesting that PTK7 regulates PCP activity in vertebrates.
We show that PTK7 protein knockdown not only disrupts neural CE, but strongly inhibits posterior neural cell fate specification. PTK7-morphant embryos have a zygotic canonical Wnt loss-of-function phenotype. PTK7 protein depletion inhibits Wnt ligand or LRP6 induction of β-catenin direct-target gene expression in embryos, explants and cultured cells. PTK7 protein depletion strongly reduces LRP6 protein levels in embryos and explants. LRP6 and PTK7 are both transmembrane proteins that physically interact in both Xenopus and cultured cells. PTK7 and LRP6 proteins appear to interact via their transmembrane domains, probably stabilizing LRP6 protein levels. In Xenopus, PTK7 protein depletion phenocopies the LRP6 knockdown phenotype. Although most previous studies suggest an exclusive role for PTK7 as a positive modulator of PCP activity driving cell morphogenesis, we alternatively suggest that PTK7 protein acts to regulate LRP6 protein levels. LRP6 protein acts at a crucial crossroads to fine-tune both canonical- and PCP-Wnt activity levels, coordinating cell fate specification and morphogenesis during early development.
RESULTS
Role of PTK7 protein in early neural patterning
As PTK7 and Meis3 proteins both regulate neural plate CE movements, we examined Meis3/PTK7 epistasis. ptk7 gene expression was unchanged in Meis3-morphant embryos at early-neurula stages (Fig. 1A). However, in PTK7 morphants, meis3 gene expression was severely repressed at similar stages (Fig. 1B). Canonical Wnt/β-catenin-activity directly induces meis3 expression to control posterior neural cell fates (Elkouby et al., 2010), so we determined if posterior neural cell fates were also lost in PTK7 morphants. As assayed by marker expression, both early and late expressed hindbrain, neural crest and primary neuron fates were lost in PTK7 morphants (Fig. 1B,C). At early-neurula stages, similar to meis3, hoxd1 gene expression is also highly reduced (Fig. 1B); hoxD1 is a direct-target gene of both Meis3 and β-catenin proteins (Dibner et al., 2004; Janssens et al., 2010). At later neurula stages, hindbrain (krox20, hoxb3), neural crest (snail2), and primary neuron (n-tub) marker expression is strongly inhibited (Fig. 1C; Fig. 2B; Fig. 4D), suggesting a phenotype more similar to a loss of canonical Wnt/β-catenin signaling than noncanonical Wnt-PCP signaling. These results phenocopy other zygotic canonical Wnt phenotypes, such as Meis3 or Wnt3a protein knockdown, or ectopic Dkk1 protein expression (Elkouby et al., 2010). In PTK7-morphant embryos, we never observe reduced expression of early maternal Wnt direct-target genes (siamois or xnr3) expressed in the organizer (Fig. 1B; Fig. 2A; Fig. 3A,B). PTK7-morpholino oligonucleotide (PTK7-MO) effects are specific; the loss-of-posterior cell fate phenotype is rescued by ectopic expression of Meis3 or human PTK7 proteins (Fig. 1D). These results suggest that in addition to regulating Wnt-PCP, PTK7 knockdown compromises some signaling component of the Wnt/β-catenin pathway.
PTK7 protein depletion inhibits canonical Wnt signaling
To understand PTK7 regulation of Wnt/β-catenin activity, we utilized the Xenopus animal cap (AC) assay. In ACs, ectopic canonical Wnt-pathway components activate transcription of the β-catenin direct-target genes, xnr3 and siamois. In the embryo, xnr3 and siamois gene expression is dependent on maternal, but not zygotic Wnt ligands and receptors. In ACs, transcription of xnr3 and siamois is dependent on zygotic Wnt-pathway components. In ACs, co-injection of the PTK7-MO sharply decreased Wnt3a-induced xnr3 and siamois expression (Fig. 2A, lanes 4-7). Expression of the early neural Wnt target gene, hoxd1 was also strongly inhibited (Fig. 2A, lanes 4-7). Also, in PTK7-depleted ACs, induction of markers by ectopic Dvl is inhibited (Fig. 2A, lanes, 8,9). In PTK7-morphant embryos, endogenous zygotic hoxd1 gene expression is repressed, but maternally regulated siamois and xnr3 gene expression is unchanged (Fig. 2A, lanes 2,3; see also Fig. 3A,B; Fig. 4D). In ACs caudalized by co-injection of Wnt3a and noggin proteins, PTK7 depletion strongly inhibited Wnt-dependent posterior neural marker expression, such as krox20, hoxb3 and n-tubulin, thus recapitulating in ACs the embryonic PTK7-morphant phenotype (Fig. 2B, lanes 4,5).
Depletion of PTK7 protein inhibits LRP6 activity
Canonical Wnt ligands signal via the Fz/LRP6 receptor complex. As PTK7 is a transmembrane protein, we examined potential LRP6/PTK7 interactions. In PTK7-depleted AC explants, both wild-type LRP6 and constitutively activated amino-truncated LRP6ΔN/LRP6ΔE1-4 proteins could not induce expression of Wnt target genes (Fig. 3A). LRP6ΔN protein robustly activates the canonical Wnt pathway independently of Wnt and Dvl proteins (Cong et al., 2004). Indeed, whereas Wnt ligand activity is sensitive to both PTK7 depletion and Dvl protein inhibition by the Xdd1 dominant-negative Dvl protein (Fig. 3B, lanes 8-10), the constitutive-active LRP6 receptor protein is significantly more sensitive to PTK7 protein depletion than Dvl activity inhibition by co-expression of the Xdd1 protein (Fig. 3B, lanes 1-13). In general, the LRP6ΔN protein seems more sensitive to the loss of PTK7 protein than the Wnt ligand. The observation that PTK7 protein knockdown promotes such a severe inhibitory effect on LRP6ΔN activity suggests that wild-type LRP6 protein activity may be regulated by PTK7.
Unlike wild-type LRP6 protein, the LRP6ΔN/ΔE1-4 proteins are Wnt ligand independently hyper-phosphorylated in vivo by CK1-γ and GSK3 (Bilic et al., 2007; Zeng et al., 2008), activating the intracellular Wnt pathway. We determined whether LRP6 phosphorylation was reduced in PTK7-depleted explants using the Tp1479 antibody (Bilic et al., 2007). Surprisingly, overall LRP6ΔN protein levels were highly reduced by PTK7 protein depletion (Fig. 3C).
To complement experiments in Xenopus, we used shRNA-PTK7 to deplete endogenous PTK7 protein in human HEK293 cells (Fig. 3D). PTK7 protein depletion strongly inhibited LRP6 activation of the TOP-Flash (β-catenin/Wnt-dependent) reporter vector (Fig. 3D). LRP6 protein levels were also reduced in these experiments (not shown). In the presence of the shRNA-PTK7, co-expression of the heterologous Xenopus PTK7 protein rescued LRP6 induction of the reporter (Fig. 3D). These results show that in addition to Xenopus, PTK7 is required for LRP6-dependent induction of canonical Wnt activity in human cells.
PTK7 maintains normal LRP6 protein levels: knockdown phenotypes are similar
To examine PTK7 modulation of normal LRP6 protein levels, experiments were performed with wild-type LRP6 protein. PTK7-depletion caused a sharp reduction of ectopically and endogenously expressed wild-type LRP6 protein levels (Fig. 4A,B). In PTK7 morphants, endogenous LRP6 protein levels resembled LRP6 morphants (Fig. 4B). To show PTK7-MO specificity, we performed an AC assay in which we co-injected Wnt3a with low levels of wild-type LRP6 protein (Fig. 4C). Co-injection of the PTK7-MO strongly inhibited Wnt3a/LRP6-induced hoxd1 gene expression (Fig. 4C, lanes 7,8), yet co-expression of PTK7 protein partially rescued hoxd1 levels (Fig. 4C, lanes 7-10). Under these conditions, ectopic PTK7 probably stabilizes some exogenous LRP6 protein, enabling hoxd1 gene expression. Upon comparison, we found that the PTK7-morphant embryos phenocopy LRP6 morphants (Fig. 4D). LRP6 and PTK7 morphants both poorly express posterior neural markers at early/mid-neurula stages, but expression of the anterior forebrain marker, xanf1 was increased (Fig. 4D). Pan-neural marker nrp1 expression is not significantly reduced in either PTK7 or LRP6 morphants, versus controls (Fig. 4D, middle panel). Thus, neural patterning, but not neural induction is perturbed. Both morphant groups have similar morphology, poor neural tube folding, coupled to weak posterior neural marker expression (Fig. 4D, right panel), characteristic of embryos with reduced zygotic canonical Wnt or Meis3 activities.
LRP6 and PTK7 proteins inhibit Wnt-PCP activity in embryos
Studies in Xenopus showed that PCP inhibitors robustly rescue perturbed CE phenotypes induced by excess expression of PCP-activating components (Djiane et al., 2000; Penzo-Mendèz et al., 2003). This assay can determine if a specific protein is an inhibitor or activator of the PCP pathway. In this assay, LRP6 protein was a PCP inhibitor; ectopic LRP6 rescued CE in explants co-expressing high Wnt11 levels, by inhibiting excess PCP activity (Tahinci et al., 2007; Bryja et al., 2009).
We used this strategy to address whether PTK7 inhibits Wnt-PCP activity regulating neural plate folding. Excess Wnt11 or Wnt5 expression strongly inhibited neural plate closure (compare Fig. 5Aa-c). Co-expression of PTK7 strongly rescued this perturbation phenotype (compare Fig. 5Ab,c and 5Ad,e). Control embryos had normal neural plate closure (95%), versus Wnt11-injected (20%) and Wnt5-injected (0%) embryos (Fig. 5B). Reciprocally, there was a robust perturbed open neural plate phenotype in Wnt11-expressing (56%) and Wnt5-expressing (87%) embryos (Fig. 5B). PTK7 strongly rescued neural plate closure defects in Wnt11 (67% normal, 30% partial, n=33) and Wnt5 (26% normal, 43% partial, n=47) co-expressing embryos (Fig. 5B). The ratio of ‘completely’ open neural plates was reduced nearly 20-fold in the PTK7/Wnt11 rescue group and nearly threefold in the PTK7/Wnt5 group versus the non-PTK7 expressing sibling embryos (Fig. 5B). Ectopic Wnt5 gave a stronger phenotype than Wnt11, and was thus harder to rescue (Fig. 5B). Like PTK7, LRP6 also rescued neural plate closure (not shown).
Additionally, the distance between the neural crest-specific snail2 expression domains at neurula stage (st.) 16-17 is used to measure the width of the neural plate, being diagnostic for reduced CE. Ectopic Wnt5/11 expression expands this distance (compare Fig. 5Ca and 5Cb,c), which was significantly reduced by PTK7 co-expression (Fig. 5Cd,e). These experiments show that PTK7 expression robustly rescues neural plate CE and folding perturbations triggered by excess Wnt-PCP activity, strong proof that PTK7 acts to antagonize and not enhance Wnt-PCP activity.
In Xenopus embryos, the ATF2-luc reporter plasmid is a readout for Wnt-PCP activity, being activated by co-expression with Fz7, Wnt5a or Wnt11 proteins (Ohkawara and Niehrs, 2011). We activated this reporter with Fz7, and then co-expressed ectopic levels of either PTK7 or LRP6 protein (Fig. 5D). Both proteins similarly inhibited Fz7 induction of the reporter (Fig. 5D). Under similar experimental conditions to the reporter assay, PTK7 also rescues the open neural plate phenotype induced by ectopic Fz7 expression. Fz7 expression induces open neural plates in 47% (n=100) of the injected embryos versus 11% of the controls (n=132), whereas PTK7 overexpression reduces open neural plates by over fourfold, returning levels to 11%, like the controls (n=72; three independent experiments). Also, in embryos in which Fz7/Wnt5 proteins are co-expressed at lower additive concentrations to perturb neural plate closure, ectopic PTK7 strongly rescues neural folding, while also inhibiting the PCP-reporter assay (not shown).
We also performed experiments similar to those described for LRP6 rescue of excess Wnt11 inhibition of activin-induced CE in ACs (Tahinci et al., 2007). We compared CE in ACs co-expressing the activated-activin receptor, Alk4 + Wnt11 and either LRP6 or PTK7 proteins (supplementary material Fig. S1). PTK7 or LRP6 co-expression both strongly rescued Wnt11 inhibition of mesodermal CE in AC explants, whereas the PTK7-MO enhances CE perturbation by Wnt11 (supplementary material Fig. S1). These results show that PTK7, like LRP6 protein, negatively modulates Wnt-PCP activity.
LRP6 and PTK7 proteins inhibit Wnt-PCP activity in cultured cells
Ultraviolet (UV) irradiation induces JNK-activation and phosphorylation of c-Jun in cultured HEK293 cells (Weidenfeld-Baranboim et al., 2011). Wnt-PCP pathway components also efficiently induce c-Jun phosphorylation (Funato et al., 2008). To determine if LRP6 or PTK7 inhibits endogenous c-Jun phosphorylation, these proteins were transfected into HEK293 cells that were UV irradiated. Expression of dominant-negative PCP-inhibitory Wnt11 and Fzl7 proteins (supplementary material Fig. S2A), or PTK7 and LRP6 proteins (supplementary material Fig. S2B) severely reduced UV-induced c-jun phosphorylation. Thus, LRP6 and PTK7 proteins act as inhibitors and not activators of Wnt-PCP dependent JNK-activation.
PTK7 and LRP6 proteins physically interact
To identify a PTK7 interacting domain in the cytoplasmic region, we systematically deleted the LRP6 protein intracellular domain. The LRP6ΔC protein has 33 amino acids of intracellular region starting with MLCP amino acid residues following the transmembrane domain (Fig. 6A). The LRP6ΔC protein was further deleted to constructs having only 16 (LRP6ΔC-M), and four (LRP6ΔC-S) intracellular domain amino acid residues (Fig. 6A). These truncated LRP6 proteins were all unstable in the absence of PTK7 protein (Fig. 6B). This instability suggests that the LRP6 intracellular domain may not be interacting with the PTK7 protein solely through these motifs. To address PTK7/LRP6 protein-protein interactions, co-immunoprecipitation (co-IP) assays were performed in Xenopus embryos and HEK293 cells. In both Xenopus and HEK293 cells, full-length PTK7 and LRP6 proteins physically interact, undergoing co-IP (Fig. 6C,D lanes 1,2). In HEK293 cells, activated LRP6ΔE1-4 and LRP6ΔC, LRP6ΔC-M and LRP6ΔC-S truncated constructs all co-IP with PTK7 protein (Fig. 6D,E). The various LRP6 constructs lacking nearly all of the extracellular (LRP6ΔN, LRP6ΔE1-4) or intracellular domains (LRP6ΔC-S) were unstable in the absence of PTK7 protein (Fig. 3C; Fig. 6B) and all co-IP with wild-type PTK7 protein (Fig. 6D,E). The intracellular truncated dominant-negative PTK7ΔK protein also co-immunoprecipitates with wild-type LRP6 protein (Fig. 6F). By contrast, The CK1-γ and GSK3β components of the LRP6-signalosome complex do not co-immunoprecipitate with wild-type PTK7 protein in HEK293 cells (not shown).
Constitutively activated LRP6 proteins require endogenous LRP6 protein
Like LRP6ΔN/LRP6EΔ1-4, the mouse myr-LRP6-ICD is a constitutively activated protein that has a complete intracellular domain, but the LRP6 transmembrane domain has been swapped to an N-terminal myristoylation target sequence (Cselenyi et al., 2008). LRP6ΔN or myr-LRP6-ICD induction of Wnt target genes was inhibited by PTK7 knockdown (Fig. 7A, lanes 4,5,7,8). When myr-LRP6-ICD is expressed in PTK7-morphant ACs, it appears somewhat more stable than other mutant LRP6 proteins, as levels are highly variable between experiments (compare Fig. 7B, lanes 2,3 and 7D, lanes 4,5), suggesting that additional regions outside the membrane domain may also function in protein stability. Regardless of myr-LRP6-ICD protein levels, its induction of Wnt target gene expression was strongly inhibited by PTK7 depletion (Fig. 7A,C). Similar to PTK7 depletion, the LRP6ΔN and myr-LRP6-ICD protein activities were both also inhibited identically by endogenous LRP6 protein depletion (Fig. 7A), suggesting a requirement for interaction with the endogenous LRP6 protein. In support, we found that the constitutively activated truncated LRP6 proteins co-immunoprecipitates with wild-type LRP6 protein in HEK293 cells (supplementary material Fig. S3). Surprisingly, the LRP6-MO not only inhibits Wnt target gene activity by myr-LRP6-ICD, but also elevates myr-LRP6-ICD protein levels, in contrast to the PTK7-MO, which lowers them (Fig. 7B). These results give novel insights into the activity mechanism of constitutively activated LRP6 proteins, demonstrating their dependence on endogenous wild-type LRP6 protein for activity and degradation. Wnt cascade activation enhances LRP6 degradation (Li et al., 2010), and myr-LRP6-ICD protein associated with endogenous LRP6 appears to be degraded. After LRP6 protein depletion and signalosome loss, inactive myr-LRP6-ICD protein ‘alone’ in the membrane is more stable (Fig. 7B), undergoing slower turnover.
DISCUSSION
Canonical and noncanonical PCP-Wnt signaling pathways regulate cell morphogenesis and cell fate in different systems (Wallingford et al., 2002; Elkouby and Frank, 2010). The canonical and noncanonical Wnt pathways also mutually interact. PCP signaling inhibited canonical activity (Kühl et al., 2001; Yan et al., 2001; Saneyoshi et al., 2002), whereas the canonical Wnt antagonist Dkk1 protein stimulated PCP activity (Caneparo et al., 2007). During both Xenopus and mouse development, high LRP6 protein levels suppress Wnt-PCP activity, whereas LRP6 protein knockdown activates high Wnt-PCP activity, blocking CE movements in explants (Tahinci et al., 2007; Bryja et al., 2009). LRP6 protein seems to balance optimal canonical versus PCP Wnt signaling levels.
Neural expressed meis3 is a direct target of mesodermal Wnt3a signaling, and Meis3 protein acts downstream to specify posterior neural cell fates, inducing Wnt-PCP dependent CE (Elkouby et al., 2010; Aamar and Frank, 2004). PTK7 protein regulates CE movements in both mouse and Xenopus embryos (Lu et al., 2004; Yen et al., 2009; Paudyal et al., 2010). PTK7 is expressed in the Xenopus neural plate; its knockdown impairs neural CE (Lu et al., 2004). We speculated that PTK7 is positioned downstream of Meis3 protein to regulate neural CE. However, Meis3/PTK7 epistasis experiments showed that ptk7 gene expression is normal in Meis3 morphants, but meis3 expression is severely inhibited in PTK7 morphants. Moreover, PTK7 knockdown embryos phenocopy Wnt loss-of-function and Meis3-morphant phenotypes. Besides inhibited neural CE, PTK7 knockdown causes a broader loss of posterior neural cell fates, including hindbrain, neural crest and primary neurons. In Xenopus, the PTK7-morphant phenotype is zygotic and not maternal, as expression of maternally regulated Wnt direct-target genes such as xnr3 and siamois is never inhibited, whereas expression of early zygotic Wnt direct-target genes, such as meis3 and hoxd1, is severely inhibited.
As PTK7 is a transmembrane protein acting zygotically upstream of Meis3, we speculated that PTK7 regulated Wnt signaling via ligands or receptors. Indeed, in AC assays, PTK7 protein knockdown inhibited Wnt pathway activation, as measured by target-gene marker expression, when either Wnt ligands or LRP6 proteins were ectopically expressed. We found that LRP6 protein levels are severely reduced in PTK7-morphant embryos or explants. Canonical Wnt signaling is highly compromised by this loss of LRP6 protein levels. Indeed, PTK7 morphants identically phenocopied LRP6 morphants; embryos and explants lose posterior neural cell fates and do not undergo proper neural CE.
Assessing Wnt-PCP levels in CE assays is complex, because the depletion or excess of Wnt-PCP is incompatible with CE (Wallingford et al., 2000; Djiane et al., 2000). Opposite effectors give similar outputs. When activin induces CE in ACs, ectopic Wnt11 protein blocks it. In these ACs, co-expression of Wnt-PCP pathway inhibitors rescues CE by reducing overwhelmingly excess PCP levels (Djiane et al., 2000). CE in Keller explants and ACs was rescued by excess LRP6 levels, canceling out PCP overstimulation by Wnt11. Thus, LRP6 protein represses Wnt-PCP activity; reciprocally, LRP6 depletion overstimulates Wnt-PCP activity, inhibiting CE (Tahinci et al., 2007; Bryja et al., 2009).
We show in embryos, explants and cultured cells that PTK7, like LRP6, acts as a Wnt-PCP inhibitor. Both proteins rescue neural plate CE and folding inhibited by ectopic Wnt5, Wnt11 or Fz7 expression that overstimulates PCP. In ACs, we show that PTK7, similar to LRP6, rescues activin-induced CE inhibited by PCP overstimulation. Studies suggested that PTK7 protein is required for embryo morphogenesis to positively modulate Wnt-PCP activity (Lu et al., 2004; Shnitsar and Borchers, 2008; Hayes et al., 2013). We suggest that during Xenopus development, PTK7 stabilization of LRP6 protein negatively modulates Wnt-PCP activity. The reduction of LRP6 protein levels via the LRP6-MO or via PTK7-MO triggers an increase in Wnt-PCP signaling that inhibits CE.
Both LRP6 and PTK7 are transmembrane proteins. In co-IP assays, these proteins physically interact in both Xenopus and human cells. Deletion of the entire LRP6 intracellular domain does not inhibit PTK7 binding. LRP6 proteins lacking the extracellular domain also co-immunoprecipitate with PTK7, suggesting that PTK7 and LRP6 proteins interact via their transmembrane domains. PTK7 also binds dominant-negative LRP6 protein; thus LRP6 protein need not be engaged in Wnt signaling to bind PTK7. LRP6 protein is the gatekeeper of the canonical Wnt pathway. Components that activate β-catenin/Wnt signaling, such as Wnts and Dvl all require functional Fz/LRP6 receptor complexes; these proteins cannot induce Wnt target gene transcription in the absence of PTK7 or LRP6 proteins. In either PTK7- or LRP6-morphant cells, we suggest that the Wnt ligand fails to signal, Dvl cannot form signalosomes and β-catenin protein will be less stable.
Recent studies in Xenopus and zebrafish concluded differently concerning the interactions between PTK7 and canonical Wnt signaling (Puppo et al., 2011; Peradziryi et al., 2011; Hayes et al., 2013). One Xenopus study suggests that PTK7 is required for canonical Wnt signaling and PTK7 knockdown blocks maternal Wnt target gene expression in embryos (Puppo et al., 2011). The initial activation of Spemann organizer-specific genes is dependent on intracellular Wnt signaling components. Maternal Wnt 5a/11 proteins regulate organizer formation (Cha et al., 2008). Only by MO depletion of maternal Wnt pathway components in oocytes are organizer ablation phenotypes observed (Heasman et al., 2000; Kofron et al., 2007). Our results show that PTK7 and LRP6 protein depletion does not affect organizer formation; maternally regulated xnr3 and siamois gene expression is not reduced in PTK7- or LRP6-morphant embryos, whereas expression of the zygotic direct Wnt target genes hoxd1 and meis3 is severely repressed. Furthermore, PTK7-MO injected embryos had phenotypes resembling zygotic Wnt loss of function. Our neural phenotype resembles embryos described in other PTK7 protein knockdown studies in Xenopus (Lu et al., 2004). PTK7 morphants do not resemble β-catenin knockdown embryo phenotypes, in which the body axis is severely perturbed and xnr3 and siamois gene expression is lost (Heasman et al., 2000). Our present work and similar studies utilize simultaneous co-injection of two PTK7-MOs at relatively low concentrations (Shnitsar and Borchers, 2008). The organizer phenotype study used different PTK7-MOs, at a fourfold higher concentration per/blastomere (Puppo et al., 2011). This variation in MO strategies might explain the differences in their observed phenotypes, which resemble maternal LRP6 depletion (Kofron et al., 2007). In zebrafish, maternal PTK7 knockdown was performed (Hayes et al., 2013). In contrast to the zygotic Xenopus phenotype (Lu et al., 2004), the maternal zebrafish phenotype has a more profound effect on mesodermal CE versus neural CE. There are a few possibilities to explain potentially different PTK7 function and phenotypes between species and embryonic stages. Studies in cancer cells have uncovered five unique proteolytic fragments of PTK7, in addition to the full-length protein. These PTK7 isoforms differ in both subcellular localization and apparent function (Golubkov and Strongin, 2012; Na et al., 2012). These PTK7 isoforms could differentially regulate canonical or noncanonical Wnt activities in a cell-context-specific manner. Unraveling the mechanistic function of other PCP components during development has also been difficult. Like PTK7, Frd1/2-Dapper proteins also gave conflicting results as to their role regulating canonical and noncanonical Wnt signaling, especially in overexpression assays (reviewed by Brott and Sokol, 2005). Determining the cellular mechanism by which Vangl2/Pk1 overexpression or knockdown modulates PCP has also been elusive (reviewed by Roszko et al., 2009). Further studies in different model systems should clarify these points.
Other studies suggest that PTK7 protein inhibits canonical Wnt signaling by interacting with Wnt ligands (Peradziryi et al., 2011; Hayes et al., 2013). These studies are mainly based on ectopic PTK7 expression and inhibition of Wnt-activated reporter constructs. Our study mainly utilizes PTK7 protein knockdown. Identifying LRP6 as the nexus of PTK7 protein activity is a key discovery in our experiments. In many systems, ectopic gene expression will give phenotypes resembling gene knockdown phenotypes. Ectopic excess PTK7 protein levels above a threshold could titrate out Wnt or LRP6 proteins from normal complexes, disrupting canonical Wnt signaling, not necessarily recapitulating endogenous physiological function. We demonstrate that PTK7-morphant embryo phenotypes resemble canonical Wnt knockdowns, supporting our data showing LRP6 protein depletion. These observations do not support the idea that PTK7 protein inhibits canonical Wnt signaling. If PTK7 acted as a canonical Wnt inhibitor, we would not expect it to rescue posterior neural marker gene expression. We would also not expect that ectopic PTK7 would rescue of neural folding, nor not inhibit expression of the snail2 neural crest marker, as snail2 expression is extremely sensitive to the loss of canonical Wnt signaling.
In Xenopus, ectopically expressed PTK7 mediated Dvl and RACK membrane recruitment during PCP (Shnitsar and Borchers, 2008; Wehner et al., 2011). These results differ from those in mouse and zebrafish, in which Ptk7 knockout mice and zebrafish show no genetic interactions with, or disruptions of, endogenous Dvl protein localization, despite having strong PCP phenotypes (Yen et al., 2009; Lee et al., 2012; Hayes et al., 2013). The Xenopus studies do not examine endogenous protein. This same Xenopus study also claims that the PTK7-MO blocks neural CE in ACs induced by ectopic FoxD1 (XBF-2) protein. This experiment does not determine if the PTK7-MO blocks canonical Wnt caudalizing activity, preventing neural caudalization and CE by FoxD1 in ACs (Fonar et al., 2011).
In mouse auditory epithelium morphogenesis, PCP activity is required for the apical formation of stereociliary bundles in sensory hair cells (Lee et al., 2012). In mice, Ptk7 or the noncanonical Fz3/Fz6 receptor knockouts give a similar PCP-phenotype; hair cell bundle misorientation is accompanied by a loss of vinculin planar asymmetry. However, in double homozygous mutants of Ptk7/Fz3 or Ptk7/Fz6, bundle orientation and vinculin asymmetry are restored (Lee et al., 2012). Ptk7 and Fz3/6 proteins act in an apparent antagonist manner to control epithelial cell PCP. These results strongly support our observations in Xenopus that PTK7 acts to restrain PCP during neural CE.
A fine-tuned balance of canonical and PCP activities is essential for correct nervous system development. In Xenopus, canonical Wnt activity initially induces posterior fates; these cells then undergo neural CE movements via activation of the Wnt-PCP pathway. Thus, early inhibition of canonical Wnt activity prevents later neural CE by eliminating posterior neural cell fates. Also, LRP6 and PTK7 proteins positively modulate canonical activity, while simultaneously inhibiting noncanonical PCP activity to fine-tune optimal Wnt signaling levels (Tahinci et al., 2007; Bryja et al., 2009). These two LRP6/PTK7-requiring functions, neural fate specification and Wnt-PCP inhibition, are not necessarily mutually exclusive. Earliest induction of posterior neural cell fates may initially require a higher canonical to PCP activity ratio, whereas later, during CE movements, this balance may shift to favor PCP activity. At later stages, wnt3a is expressed locally and not ubiquitously in the neural plate (Fonar et al., 2011). In the absence of canonical Wnt signaling, ectopic expression of the Meis3 protein robustly induces neural marker expression and CE (Elkouby et al., 2010; Elkouby et al., 2012), suggesting that after posterior neural specification, canonical Wnt activity is reduced to lower levels, permissive for neural CE.
PTK7 protein binds the LRP6 protein and is crucial for maintaining its stability. PTK7 knockdown depletes LRP6 protein levels crucial for activating caudalizing transcription factor targets, such as meis3 and hoxd1. PTK7 and LRP6 protein knockdown both cause similar phenotypes; posterior neural cell fates and resultant neural CE are both lost. Both PTK7 and LRP6 proteins inhibit Wnt-PCP signaling when in excess. These findings suggest an exciting new role for PTK7. PTK7 regulates LRP6 protein levels, which maintain the canonical/PCP Wnt activity ratio, thus controlling posterior neural cell fate specification. Defects in both canonical and noncanonical Wnt signaling are linked to cancer, as are modulations in LRP6 and PTK7 protein activities (MacDonald et al., 2009; Wang, 2009; Golubkov and Strongin, 2012). Future experiments should determine if disrupted PTK7/LRP6 interactions modulating Wnt-network signaling play a functional role during carcinogenesis.
MATERIALS AND METHODS
Xenopus embryos
Ovulation, in vitro fertilization, culture and explant dissections were as described (Re’em-Kalma et al., 1995).
RNA and MO injections and plasmid constructs
Capped in vitro transcribed mRNA constructs of ptk7, ptk7ΔK, huPTK7 (Shnitsar and Borchers, 2008); meis3, noggin, wnt3a (Elkouby et al., 2010); dvl, dd1, wnt11 (Aamar and Frank, 2004), activated-alk4 receptor (R. Harland Lab); lrp6, lrp6ΔN, lrp6ΔC, lrp6ΔE1-4 (Bilic et al., 2007; Zeng et al., 2008) were injected into one- or two-cell-stage embryos. Antisense MOs (Gene Tools): Meis3-MO, PTK7-MO, LRP6-MO (Dibner et al., 2001; Lu et al., 2004; Tahinci et al., 2007). Two-mixed PTK7-MOs were injected (Lu et al., 2004). The LRP6ΔC deletion constructs were deleted from the intracellular carboxy-end with two different 3′ end polymerase chain reaction (PCR) primers yielding constructs with 16 (LRP6ΔC-M) or four (LRP6ΔC-S) amino acid residues of the intracellular domain.
In situ hybridization
Whole-mount in situ hybridization was performed with digoxigenin-labeled probes: krox20, hoxB3, n-tubulin (n-tub) and snail2 (Gutkovich et al., 2010).
Semi-quantitative reverse transcription (sqRT) PCR analysis
sqRT-PCR was performed (Snir et al., 2006). In all sqRT-PCR experiments, three to six independent experimental repeats were performed. Samples are assayed at least twice for each marker. sqRT-PCR primers: ef1a, his4, hoxD1, krox20, hoxB9, hoxB3, hoxB4, foxd3, n-tub (Gutkovich et al., 2010); siamois and xnr3 primer sequences (R. Harland Lab database). Xenopus ptk7 primers: A: 5′-TACCGTGTGACCAATGAGGA-3′; B: 5′-CATGCCTC - CTAGCCTTTCTG-3′.
Cell culture and transfection
Human embryonic kidney cells (HEK293), were grown in Dulbecco’s modified Eagle medium, fetal calf serum, glutamine and antibiotics. Cells were transfected by the calcium phosphate procedure.
Immunoprecipitation (IP) in Xenopus and HEK293 cells
RNAs (2 ng) encoding PTK7-HA and VSVG-LRP6 proteins were co-injected into one-cell-stage embryos. Twenty-five st.12 embryos from each group were lysed in hypotonic buffer. Lysates were centrifuged three times at 13,000 rpm. Protein A-Sepharose beads (Sigma) were washed three times with lysis buffer. Anti-HA antibody (Covance) was added to beads and rocked for 2 hours, at 4°C. Beads were rewashed and lysates were added. Rocking continued overnight, at 4°C. Beads were washed three times in lysis buffer + Triton 0.1%. Electrophoresis loading buffer was added. Proteins were detected by anti-HA and anti-VSVG antibodies. HEK293 cells were lysed on ice, 48 hours after transfection, with 500 μl IP buffer. Cells were vortexed for 10 seconds, every 5 minutes, for 15 minutes. Supernatant was collected after spinning the tubes at 10,000 rpm, for 15 minutes. For IP, the relevant antibody was added to the lysates and rocked for 1 hour, at 4°C; recombinant protein G-Agarose beads (Invitrogen) were added and rocked overnight. Beads were spun down for 5 minutes at 2500 rpm, and washed four times with IP buffer. Electrophoresis loading buffer was added.
Western analysis
Western analysis was performed as previously described (Zetser et al., 2001). Antibodies used were: LRP6-T1479 (1 μg/ml), LRP6-Sp1490 (1 μg/ml), α-Tubulin (Sigma T9026; mouse, 1:10,000), VSVG (Sigma V4888; rabbit, 1:2000), HA (Covance MMS101R; mouse, 1:1000), myc (Santa Cruz SC-789; rabbit, 1:1000). Secondary antibodies: goat anti-mouse (Thermo Scientific 31432; 1:4000), or goat anti-rabbit (Thermo Scientific 31460; 1:4000). α-Tubulin is the protein loading control.
Acknowledgements
We thank Drs A. Borchers, C. Niehrs, X. He, R. Harland and E. Lee for plasmids and antibodies.
Author contributions
N.B.-N., H.L., A.M. and D.F. designed the project, carried out experiments, analyzed data and wrote the paper. M.L. and S.E. designed and performed experiments and analyzed data.
Funding
D.F. was supported by grants from the Israel-Niedersachsen Fund [ZN2319] and the Israel Science Foundation [658/09].
References
Competing interests
The authors declare no competing financial interests.