PK synergize with Vangl to suppress Dvl during CE

Injecting 0.1 ng mouse Vangl2 mRNA (mVangl2) into the DMZ results in moderate CE defects and reduction of the length-to-width ration (LWR), and the phenotypes are significantly enhanced by co-injecting a small dose of Xpk (a, a’) or mPk2 (b, b’) mRNA that causes minimal or no CE defect per se. On the other hand, higher dose mVangl2 (0.15 ng) induced more severe CE defects can be rescued by knocking down endogenous Pk using XpkMO (c, c’); whereas knockdown of endogenous vangl2 (XVMO) induced CE defects can be rescued by moderate mPk2 overexpression (d, d’). Conversely, 0.5 ng mouse Dvl2 injection induced CE defects can be dose-dependently rescued by mPk2 (e, e’) and Xpk (f, f’) co-overexpression; and mVangl2-mPk2 co-overexpression induced severe CE defect can be rescued by co-injecting Dvl2 (g, g’).

Vangl2 RH variant diminishes membrane recruitment of Pk and reduces functional synergy with Pk

Schematic illustration showing the structure of the tetraspan protein Vangl2 and location of the R177H variant at the intracellular loop region between transmembrane domain 2 and 3 (a). When expressed in Xenopus animal cap cells, EGFP-Vangl2 RH displays plasma membrane localization, indistinguishable from wild-type EGFP-Vangl2 (b, b’). Immunostaining shows that flag-XPk displays diffuse cytoplasmic localization with some membrane enrichment when expressed alone (c, c’). The plasma membrane localization of flag-XPk is enhanced significantly by co-expression of wild-type Vangl2 (d, d’), but only modestly by Vangl2 RH variant (e, e’). (f) Quantification of the ratio of plasma membrane vs. cytoplasmic flag-XPk in (c), (d) and (e). Co-IP and western blot show that binding to flag-XPk is reduced by Vangl2 RH variant (g). Functionally, co-injection of 0.05 ng Vangl2 and 0.5 ng Xpk can strongly synergize to disrupt CE, but the synergy is significantly reduced when Vangl2 RH variant mRNA is co-injected with Xpk (h). The CE phenotype is quantified by the LWR measurement in (i).

Pk synergizes with Vangl2 to inhibit Wnt11-induced formation of Dvl patches

0.25 ng Wnt11 injection with 0.5 ng of mCh-tagged mouse Dvl2 and membrane-GFP induces formation of distinct Dvl2 patches at the cell-cell contact (a-c’). Co-injection indicates that these Dvl2 patches completely overlap with Fz7 (d-f’). In contrast, Vangl2, when expressed at moderate levels, is distributed more broadly along the plasma membrane (g-i), but also displays enrichment immediately outside and at the edge of Dvl2 patches (g’-I’, arrowheads and arrows, respectively). High level of Vangl2 injection inhibits Wnt11-induced Dvl2 patch formation and makes Dvl2 more evenly distributed with Vangl2 (j-l’). The same effect can also be achieved by co-expressing Pk with moderate level of Vangl2 (m-o’). (p-t) Measurement of the relative intensity of Dvl2 along the patches with either membrane GFP, Fz7, Vangl2 at moderate (0.1 ng) and high (0.5 ng) levels, or Vangl2 (0.1 ng) and XPk (0.5 ng) co-injection.

Pk helps Vangl2 to inhibit Wnt11-induced clustering of Fz7-Dvl2 complexes

Wnt11 induces formation of overlapping Fz7 and Dvl2 patches at the cell-cell contact (a-c’). These patches are not affected by over-expressing moderate level of Vangl2 (d-f’) or Pk (g-i’) individually. Vangl2 and Pk co-expression, however, not only disrupt Dvl2 patches but also disperse Fz7 patches into small puncta (j-l). Enlarged views revealed that some of the Fz7 puncta are on the plasma membrane and remain co-localized with Dvl2, while the others are located in the cytoplasm near the plasma membrane (k’’, l’, arrows) and contain only Fz7 but not Dvl2 (compare arrows in j’ to k’).

Vangl2 R177H variant fails to synergize with Pk to inhibit Fz patch formation and down-regulate Fz stability at the plasma membrane

Wnt11 induced formation of Fz7-GFP patches on the plasma membrane (a, a’) was not affected by moderate expression either wild-type Vangl2 (b, b’) or Vangl2 R177H variant (c, c’) alone. Pk co-injection synergized with wild-type Vangl2 to diminish Fz7 patch formation and induce cytoplasmic Fz7 puncta (d, d’), but the synergy was not observed with Vangl2 R177H (e, e’).

Pk synergize with Vangl2 to prevent Fz-induced phosphorylation of Dvl.

In Xenopus extract, DMZ injected Dvl2 migrates as two bands, with the slower migrating band increasing in intensity from stage 10 to 12 as CE starts and progresses during gastrulation (a). The slower migrating form of Dvl2 is increased by Fz7 co-injection, but eliminated by phosphatase treatment (b). High level Vangl2 injection can reduce the phosphorylated form of Dvl2, while Fz7 can counter Vangl2’s effect to increase Dvl2 phosphorylation (c). Co-IP experiment indicates that only the faster migrating, unphosphorylated form of Dvl2 can be co-immunoprecipitated by Vangl2 (d). High level Xpk (1 ng) or moderate Vangl2 (0.25 ng) cannot significantly reduce phosphorylated form of Dvl2 when injected individually, but their co-injection can suppress Dvl2 phosphorylation (e). (f) Quantification of the ratio between the fast migrating/ unphosphorylated and the slow migrating/ phosphorylated forms of Dvl2 in (e).

Ror2 is an obligatory component of the Fz/Dvl cluster complex induced by Wnt11.

(a-c) Co-injection of Wnt11 with EGFP-tagged Xror2 (0.1 ng) and Dvl2-mCh shows that Ror2 and Dvl2 both form patches that largely overlap, but unlike Dvl2, Ror2 additionally displays broad distribution along the entire cell cortex. (c’-e’) The enlarged views show that both Ror2 and Dvl2 accumulate at the center of the patches (red arrowhead), but Ror2 patches are slightly longer and extend beyond the border of Dvl2 patches (green arrowheads). (d) Measurement of the relative intensity of Dvl2-mCh along the patches with Ror2-EGFP. (e) Injection of Wnt11 induces co-injected Dvl2-mCh to cluster into patches on the cell cortex. (b) Co-injection of 25 ng Xror2 morpholino (Xror2-MO) blocks Wnt11-induced Dvl2 clustering.

Vangl2/Pk exert bimodal regulation of Ror2 in non-canonical Wnt signaling.

(a-c) Wnt11 induces injected XRor2-td-Tomanto and Fz7-GFP to co-cluster into patches on the cell cortex. Overexpression of Vangl2 (d-f) or Pk (g-i) individually does not perturb co-clustering of Ror2 with Fz7 in response to Wnt11, but their co-overexpression (j-l) diminishes Ror2-Fz7 co-clustering into patches and cause Fz7 to form cytoplasmic puncta whereas Ror2 remained on the plasma membrane (compare arrows in j’-l’). a’-l’ are enlarged view of a-l, respectively. Conversely, partial knockdown of endogenous XVangl2 with moderate level of XVMO (14 ng) also diminished Ror2/Fz7 patches formed in response to Wnt11 (Fig.8m-o’) with simultaneous formation of intracellular vesicles around the plasma membrane which contain Fz7 but not Ror2 (Fig.8m’-o’, arrows).

An integrated model for how core PCP proteins mediates non-canonical Wnt signaling to regulate CE.

(a, a’) In the absence of non-canonical Wnt, Pk helps Vangl to act as an adaptor that brings together Dvl and Ror, through either simultaneous binding of both Dvl and Ror to a single Vangl (a) or oligomerization of Vangl proteins bound separately to Dvl and Ror (a’), and keeps both Dvl and Ror inactive to prevent ectopic non-canonical Wnt signaling. (b, b’) Non-canonical Wnt initiates signaling by triggering Fz-Ror heterodimerization, and in turn, the complexes consisting of Ror/Vangl/Dvl are brought close to Fz to deliver Dvl for activation of downstream targets. (c, c’) Non-canonical Wnt also induces other events such as Dvl phosphorylation, likely through CK1, that may cause Dvl dissociation from Vangl and transition to Fz in a spatially and temporally controlled manner.

In the DMZ, both over-expression (a, a’) and morpholino knockdown (b, b’) of Xpk can dose-dependently block CE, resulting in shorten body axes and reduced LWR. In the animal cap, XpkMO knockdown can also block activin induced CE (c, c’), but the defect can be rescued by co-injecting RNA encoding mouse Pk2 (d, d’).

EGFP-Vangl2, flag-Dvl2 and GFP-mPk2 mRNA were injected into the DMZ, either alone or together, at 4 cell stage, and embryos were harvested at stages 10.5-11 for westernblot. Co-overexpressing Vangl2 with mPk2, or Dvl2 with mPk2, does not significantly alter each other’s protein level (a). Similarly, co-overexpressing Flag-XPk with EGFP-Vangl2 in activin treated animal cap explant also does not alter each other’s protein level (b). Knocking down endogenous Xvangl2 does not alter the protein level of injected Flag-XPk either (c). Interestingly, however, co-transfecting Vangl2 with XPk into 293T cells significantly reduced XPk protein level (d).

EGFP tagged mPk2 display uniform cytoplasmic distribution with some enrichment at the plasma membrane when expressed alone in animal cap cells (a, c, and enlarged view in c). Plasma membrane enrichment of mPk2 is enhanced by co-injection of mVangl2 (b), but eliminated by morpholino knockdown of endogenous XVangl2 (d). Injecting a small amount of mVangl2 (50 pg) could restore mPk2 plasma membrane enrichment in XVangl2 morphants (e).

Dvl2-EGFP display uniform cytoplasmic distribution when expressed alone in Xenopus animal cap cells (a, a’), but co-expression with either wild-type Vangl2 (b, b’) or Vangl2 R177H variant (c, c’) recruits Dvl2-EGFP to the plasma membrane. (d) Quantification of the ratio of plasma membrane vs. cytoplasmic Dvl2-EGFP in (a), (b) and (c). Co-IP and westernblot show that binding to Dvl2 is not affected by Vangl2 RH variant (e).

When injected into the DMZ, 0.2 ng and 0.1 ng wild-type mouse Vangl2 mRNA causes severe and moderate CE defect, respectively. Vangl2 R177H mRNA, however, causes significantly milder CE defect (a). (a’) Quantification of the LWR of embryos in (a). Fz7 and Dvl2 co-injection caused severe CE defects can be rescued by 0.1 wild-type Vangl2, but not by Vangl2 R177H (b). (b’) Quantification of the LWR of embryos in (b).

In both animal cap and DMZ cells, Fz7 can recruit Dvl2 from the cytoplasm to the plasma membrane (compare b to a, and e to d). Over-expression of Pk2 does not inhibit Fz7-mediated plasma membrane recruitment of Dvl2 (c, f).

(a) In DMZ explant, binding between flag-Dvl2 and EGFP-Vangl2 is reduced by Wnt11. But over-expression of Pk2 can prevent Wnt11-induced dissociation of Dvl2 and Vangl2. (b) Quantification of EGFP-Vangl2 co-IP efficiency from 3 separate experiments.

Moderate level of EGFP-Vangl2 (0.1 ng) was co-injected with Wnt11 and mCherry tagged Dvl2. 3D reconstruction of confocal images shows that under this condition, all Dvl2 proteins on the plasma membrane formed discrete patches at the apical cell-cell junctions (a), while Vangl2 was distributed more diffusely on the plasma membrane (b) with enrichment overlapping with Dvl2 patches (c, white arrows). Enlarged views (a’-c’) show that enriched Vangl2 forms rings that encircle Dvl2 patches.

Xenopus explants were incubated with FM4-64FX, a fluorescent dye that is membrane impermeable and can only be internalized through endocytosis. In explants in which Fz7-GFP was co-injected only with Wnt11, Fz7 primarily formed patches on the cell cortex. Most of the internalized FM4-64FX puncta were devoid of Fz7, and the few Fz7 puncta were devoid of FM4-64FX (arrowheads in a-c, and enlarged views a’-c’). In contrast, in explants in which Fz7-GFP and Wnt11 were co-injected with Vangl2/Pk, majority of the Fz7-GFP puncta are positive for FM4-64FX (arrows in d-f and enlarged views d’-f’), indicating that they were largely formed through endocytosis.

Wnt11 induces formation of Fz7 patches along the plasma membrane (a). Over-expression of DshMA, a mitochondria tethered form of Dvl that can sequester endogenous Dvl/Dsh away from the plasma membrane, inhibits Fz7 patch formation, reduces membrane level of Fz7, and results in Fz7 cytoplasmic puncta (b).

DMZ injection of 0.2 ng of Vangl2 mRNA induces severe CE defects, which can be rescued by co-injecting 0.05 or 0.1 ng of Xror2 mRNA (a). The severity of the CE phenotype is quantified by measuring the length-to-width ratio (b).

In Xenopus embryo extract, co-IP experiment shows that binding between flag-Dvl2 and HA-Vangl2 is reduced by Wnt11 (compare lane2 and 3). Knocking down endogenous XRor2 with morpholino (xRor2 MO) can prevent Wnt11-induced dissociation of Dvl2 and Vangl2 (compare lane 4 and 5) (a). (b) Co-IP experiment shows that Ror2-Flag can pull down co-injected EGFP-Vangl2 but not Myc-Dvl2. (c) Fluorescence-detection size exclusion chromatography (FSEC) with protein extract from Xenopus embryos injected with Ror2-EGFP, HA-Vangl2 and Flag-Dvl2. A peak of EGFP signal is detected around fraction 15 (lower panel). Western blot analyses show co-fractionation of Ror2-EGFP, HA-Vangl2 and Flag-Dvl2 in fractions 14 (approximate molecular weight 773-1717 kD), 15 (∼348-773 kD) and 16 (∼166-348 kD).