XB-ART-1796Curr Biol June 7, 2005; 15 (11): 1039-44.
Subcellular localization and signaling properties of dishevelled in developing vertebrate embryos.
The Dishevelled protein mediates several diverse biological processes. Intriguingly, within the same tissues where Xenopus Dishevelled (Xdsh) controls cell fate via canonical Wnt signaling, it also controls cell polarity via the vertebrate planar cell polarity (PCP) cascade [1, 2, 3, 4, 5, 6, 7, 8 and 9]. The relationship between subcellular localization of Dishevelled and its signaling activities remains unclear; conflicting results have been reported depending upon the organism and cell types examined [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 and 20]. We have approached this issue by developing new reagents to sequester wild-type Dishevelled protein either at the cell membrane or away from the cell membrane. Removal of Dishevelled from the cell membrane disrupts convergent extension by preventing Rho/Rac activation and mediolateral cell polarization. By manipulating the subcellular localization of K-->M (dsh1), we show that this mutation inhibits Dishevelled activation of Rac, regardless of its subcellular localization. These data demonstrate that membrane localization of Dishevelled is a prerequisite for vertebrate PCP signaling. However, both membrane-targeted and cytoplasm-targeted Dishevelled can potently activate canonical Wnt signaling, suggesting that local concentration of Dishevelled protein, but not its spatial localization, is central to canonical Wnt signaling. These results suggest that in vertebrate embryos, subcellular localization is insufficient to account for the pathway specificity of Dishevelled in the canonical Wnt versus PCP signaling cascades.
PubMed ID: 15936275
Article link: Curr Biol
Genes referenced: acta4 akt1 dvl1 dvl2 nodal3.1 nodal3.2 not rac1 rho shh sia1
Article Images: [+] show captions
|Figure 1. Manipulation of Dishevelled Localization (A) Dvl monomers associate via their N-terminal DIX domains to form oligomers that are found in both cytoplasmic and cell-membrane populations (A#). (B) Endogenous Dvl monomers will oligomerize with membrane-tethered Dvl-caax such that the wild-type Dvl is sequestered at the cell membrane (B#). (C) Endogenous Dvl monomers will oligomerize with mitochondrial-localized Xdsh- MA, and the wild-type Dvl will be sequestered in the cytoplasm, away from the cell periphery (C#). (D) Localization of Xdsh-GFP in a normal Xenopus animal cap. (E) Higher-magnification view of Xdsh-GFP (green) localization in normal animal cap; cell membrane is indicated by memRFP (E#, red). (F) Wild-type Xdsh-GFP (green) is translocated to the cell membrane by expression of unlabeled Dvl-KM-caax. (G) Higher-magnification view of a cell expressing wild-type Xdsh-GFP (green) and unlabeled Dvl-KM-caax; cell membrane is indicated by memRFP (G#, red). (H) Wild-type Xdsh-GFP (green) is translocated to clusters in the cytoplasm by expression of unlabeled Xdsh-MA. (I) Higher-magnification view of a cell expressing wild-type Xdsh-GFP (green) and unlabeled Dvl-KM-caax; cell membrane is indicated by memRFP (I#, red).|
|Figure 2. Removal of Dishevelled from the Cell Membrane in the DMZ Blocks Convergent Extension by Disrupting Cell Shape and Cell Polarity (A) Membrane localization of Xdsh-GFP in the DMZ of a normal embryo. Cell membrane can be visualized by memRFP (A#). Colocalization is apparent in the merged image (A$). (B) Xdsh-GFP is entirely sequestered in the cytoplasm in the DMZ of embryos expressing Xdsh-MA. (B#) shows coexpressed memRFP. (B$) shows a merged image. (C) Control embryo showing clone of cells from injection of fluorescent dextran (green) at the 256-cell stage. Labeled clone is more obvious in the fluorescent view in panel (C#). (D) Labeled clones at late gastrulation have undergone significant convergence and extension (compare to panel [C]). Fluorescent view is in panel (D#). (E) Confocal image of cells in the boxed region of panel (D#). Because of mediolateral cell intercalation, labeled cells are intermingled with unlabeled cells. These cells are mediolaterally elongated and aligned. (F) Xdsh-MA-injected embryo showing a labeled clone of cells that is not elongated at early gastrulation (fluorescent view in panel [F#]). (G) Labeled clones at late gastrulation in Xdsh-MA-injected embryos have failed to elongate (fluorescent view in panel [G#]). (H) Confocal image of cells in the boxed region of panel (G#). In embryos expressing Xdsh-MA, labeled cells are not intermingled with unlabeled cells. These cells are not mediolaterally elongated and are randomly oriented (see Figure S3).|
|Figure 3. Removal of Dvl from the Cell Membrane Prevents Nonnal Activation of Rho and Rac during Convergent Extension To assess activation of Rho and Rac, we employed GST-pulldown assays utilizing GST-RDB (Rho binding domain) and GST-PBD (Pak binding domain) fusion proteins, which bind specifically to activated (GTP-bound) Rho and Rac, respectively. (A) Activation of Rho and Rac in dorsal marginal zones expressing constructs listed at top of panel. The mild reduction of Rac activation by Dvi-KM-caax was observed in most but not all cases. (B) localization of RhoA-GFP in cells of the dorsal marginal zone in a control embryo; inset shows Rac-GFP localization in control animal-cap cells. (C) Xdsh-MA expression does not change the subcellular localization of RhoA-GFP or Rac-GFP (inset). (D) Activation of Rho and Rae in ventral marginal zones expressing constructs listed at top of panel.|
|Figure 4. Constitutive Membrane Localization of Dishevelled Enhances Its Gain-of- Function Disruption of Convergent Extension (A) Normal control embryo. (B) Embryo expressing Dvl-KM-caax and displaying a mild phenotype. (C) Embryo expressing Dvl-KM-caax and displaying a severe phenotype. The phenotypes (B and C) are independent of effects on dorsal cell-fate specification, given that normal expression of Sonic Hedgehog and the somite marker 12/101 was observed (data not shown). (D and E) Frequency of normal, mild, and severe phenotypes (as indicated in panels [A]– [C]) after injection of indicated mRNAs.|
|Figure 5. Both Membrane-Targeted and Constitutively Cytoplasmic Dishevelled Are Potent Activators of Canonical Wnt Signaling (A) Injection of 50 pg Xdsh-MA or Dvl-KM-caax is sufficient to induce secondary axes when injected ventrally into 8-cell Xenopus embryos. Injection of equivalent doses of wild-type Xdsh-GFP or Dvl-1 fails to induce secondary axes. Many of the secondary axes induced by Dvl- KM-caax or Xdsh-MA were complete, as judged by the presence of a secondary cement gland (B, arrow). (C) Injection of 50 pg Xdsh-MA or Dvl-KM-caax in animal caps is sufficient to activate transcription of the direct Wnt target genes Xnr-3 and siamois. Injection of equivalent doses of wild-type Xdsh-GFP or Dvl-1 fails to activate transcription of these genes.|
|Figure S1. Translation and Rho/Rac Activation of Dishevelled Construct (A) Anti-HA Western blotting to assess translation of injected mRNAs, as indicated above the panels. (B) Failure of Rho and Rac activation by Dvl- MA is overcome by the addition of Dvl-1 or Dvl-Caax.|
|Figure S2. Quantification of Xdsh-GFP Membrane Association To quantify the degree of membrane association of Xdsh-GFP, we calculated its colocalization with membrane-tethered RFP (memRFP [S9]). The plot in this figure shows the ratio of pixels in which GFP and RFP signals are colocalized to the total pixels. A threshold is applied to remove low-intensity signal that is likely to be nonspecific. This ratio, the “colocalization coefficient,” was attained with Zeiss LSM5 software. First, as a control, we measured colocalization of Xdsh-GFP and memRFP in control animal caps (column 1). Expression of unlabeled Dvl-KM-caax resulted in a dramatic increase in Xdsh-GFP colocalization with memRFP (column 2), demonstrating the efficacy of this quantification method (see also Figures 1D and 1F in the main text). We next quantified colocalization in DMZ cells. As compared to normal animal caps cells, DMZ cells display a much higher fraction of Xdsh-GFP colocalized with memRFP (column 3). However, in DMZ cells expressing unlabeled Xdsh-MA, we observe as much as a 10-fold reduction in the colocalization of Xdsh-GFP and memRFP (column 4). These data indicate that Xdsh-MA effectively removes wild-type Dvl from the cell membrane in cells undergoing convergent extension.|
|Figure S3. Effects of Xdsh-MA on Convergent Extension (A) Dorsal view of control embryo at stage 24. (B) Control embryo at stage 10.5, hybridized to the Xnot probe (dorsal is up). (C) Control embryo (stage 24), hybridized to the xSHH probe. (D) Dorsal view of Xdsh-MA-injected embryo at stage 24; note the reduced axis elongation and open neural tube. (E) Xdsh-MA-injected embryo at stage 10.5; expression of Xnot is normal (dorsal is up). (F) Xdsh-MA-injected embryo expression of xSHH is normal at stage 24. It should be noted that despite the ability of Xdsh-MA to activate canonical Wnt signaling (see Figure 5 in the main text), expression of the dorsal markers Xnot and xSHH is not significantly affected (axial-protocadherin expression was also unaffected but was not shown). Expression of Dvl-caax likewise activates canonical Wnt signaling (Figure 5) but did not significantly affect expression of dorsal markers (xSHH and Xnot, data not shown). Both constructs also disrupt convergent extension, but activation of the canonical Wnt pathway is not likely to be the cause of the convergentextension defects because hyperactivation of canonical Wnt signaling downstream of Dishevelled does not inhibit convergent extension [S12, S15]; instead, enhanced canonical signaling likely accelerates convergent extension by activating expression of Xnr3 [S16]. (G) Morphometrics of cells in the notochord and somites of stage-13 embryos. Plot shows length-to-width ratio (LWR) of cells plotted against the orientation of the long axis of that cell. Control cells (red) are highly polarized; cell axes are oriented mediolaterally, and cells have high LWR (2.06 0.087; mean SEM). Cells in Xdsh-MA-injected embryos (black) are not polarized; cell axes are randomly oriented, and cells have low LWR (1.62 0.067; mean SEM). (H) All cells in control embryos have their long axes oriented mediolaterally (e.g., within 45 of perpendicular to the embryonic anteroposterior axis). Cells in embryos expressing Xdsh-MA have their long axes distributed evenly between mediolateral and anteroposterior orientations.|