Distinct Domains of Mouse Dishevelled Are Responsible for the c-Jun N-terminal Kinase/Stress-activated Protein Kinase Activation and the Axis Formation in Vertebrates*

Recent studies have shown thatDrosophila Dishevelled (Dsh), an essential component of thewingless signal transduction, is also involved in planar polarity signaling through the c-Jun N-terminal kinase (JNK)/stress-activated protein kinase (SAPK) pathway inDrosophila. Here, we show that expression of a mouse homolog of Dsh (mDvl-1) in NIH3T3 cells activates JNK/SAPK, and its activator MKK7. A C-terminal half of mDvl-1 which contains the DEP domain was sufficient for the activation of JNK/SAPK, whereas an N-terminal half of mDvl-1 as well as the DEP domain is required for stimulation of the TCF/LEF-1-dependent transcriptional activation, a β-catenin-dependent process. A single amino acid substitution (Met for Lys) within the DEP domain (mDvl-1 (KM)) abolished the JNK/SAPK-activating activity of mDvl-1, but did not affect the activity to activate the LEF-1-dependent transcription. Ectopic expression of mDvl-1 (KM) or an N-terminal half of mDvl-1, but not the C-terminal, was able to induce secondary axis inXenopus embryos. Because the secondary axis formation is dependent on the Wnt/β-catenin signaling pathway, these results suggest that distinct domains of mDvl-1 are responsible for the two downstream signaling pathways, the β-catenin pathway and the JNK/SAPK pathway in vertebrates.

c-Jun N-terminal kinase (JNK) 1 /stress-activated protein kinase (SAPK), a member of mitogen-activated protein kinases (MAPKs) (1)(2)(3)(4), is activated by exposure of cells to certain kinds of cytokines and environmental stresses (4 -7). Two MAPK kinases (MAPKKs), MKK4/SEK1 and MKK7, act as direct activators for JNK/SAPK (4 -12). Although several members of MAPKK kinases, Ste20 homologs, protein-tyrosine kinases, and low molecular weight GTPases have been described as upstream components of the JNK/SAPK signaling pathway, players involved in signaling pathways have not been completely listed up, and exact signaling pathways from receptors or sensors to the activation of JNK/SAPK have not been fully understood.
The Wnt proteins constitute a family of secreted glycoproteins, among which Wingless (Wg) in Drosophila is the best characterized member. Genetic evidence reveals that Wg signals through an intracellular cascade that includes Dishevelled (Dsh), Zeste-White3/Shaggy and Armadillo, a ␤-catenin homolog (13). There exist vertebrate counterparts of these genes, and the corresponding cascade is believed to function in a variety of biological processes (14,15). Of the known components, Dsh is thought to act most immediately downstream of the receptor. Alignment of members of invertebrate and vertebrate Dsh family proteins (16 -19) reveals three conserved domains: a DIX domain, a PDZ domain, and a DEP domain (20). Recent analysis of the planar polarity-specific dsh 1 allele in Drosophila revealed a single amino acid point mutation in the DEP domain and demonstrated requirement of the DEP domain for planar polarity signaling but not for wingless signaling (21,22). Moreover, it has been shown that Dsh may act through low molecular weight GTPases and the JNK/SAPK pathway in planar polarity signaling and that overexpression of Dsh results in enhanced c-Jun-N-terminal phosphorylation in cultured cells (21). These results suggested the existence of a signaling pathway from Dsh to activation of JNK/SAPK, which has not been directly tested. In this study, we have first demonstrated that a murine homolog of Dsh, mDvl-1, is able to potently activate JNK/SAPK and its direct activator MKK7 probably through Rac or Cdc42. We then produced various mutant forms of mDvl-1 and tested their abilities to activate JNK/SAPK, to regulate the TCF/LEF-1-dependent transcription, a ␤-catenin-dependent process, in mammalian cells, and to induce the secondary axis in Xenopus embryos. The formation of the secondary axis has been shown to be controlled by the Dsh/␤-catenin pathway (14,15,23,24). The results obtained show that distinct domains of mDvl-1 are responsible independently for the two downstream signaling events, the JNK/SAPK pathway and the ␤-catenin pathway. This mechanism may be important for controlling the bifurcation of signaling pathways from mDvl-1.
Kinase Assays-15 h after transfection, NIH3T3 cells were lysed, and assays for the activity of HA-tagged protein kinase were performed by the immune complex kinase assay as described (11,26,27). 2 The following proteins were used as substrates in a total volume of 15 l of reaction mixture: 3 g of myelin basic protein (MBP) for MAPK, 3 g of GST-c-Jun (1-79) for SAPK␣, 3 g of His-ATF2 for p38, 10 g of MBP for ERK5, 3 g of GST-KN-MAPK for MAPKK, 3 g of His-KN-MPK2 for MKK3 and MKK6, 1 g of His-SAPK␣ and 3 g of GST-c-Jun (1-79) for SEK1 and MKK7, and 3 g of GST-KN-ERK5(1-457) for MEK5. To determine the amounts of HA-tagged proteins, the immunoprecipitates were subjected to immunoblotting with rabbit anti-HA-antibody (Y-11, Santa Cruz Biotechnology, Inc.). Aliquots of whole cell lysates were subjected to immunoblotting by anti-Myc antibody (9E10, Santa Cruz Biotechnology, Inc.) to confirm appropriate expression of Myc-tagged mDvl-1 proteins. For kinase assays of endogenous JNK/SAPK, the lysates (200 l, ϳ200 g) of NIH3T3 cells expressing each of various mDvl-1 mutants were incubated with 25 l of protein A-Sepharose beads (Amersham Pharmacia Biotech) and 5 l (0.5 mg/ml) of mouse anti-JNK1 antibody (Pharmingen) for 2 h at 4°C. Then, the immune complex kinase assay of endogenous JNK/SAPK was performed as described above, and the amounts of immunoprecipitated endogenous JNK/SAPK were determined by immunoblotting with goat anti-JNK1 antibody (Santa Cruz Biotechnology, Inc.).
Subcellular Fractionation-Cells were washed twice with ice-cold Hepes-buffered saline, harvested by scraping from the culture dishes into ice-cold hypotonic buffer consisting of 10 mM 2-glycerophosphate (pH 7.4), 1 mM MgCl 2 , 2 mM EGTA, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 1% aprotinin (200 l of buffer/60-mm dish), and disrupted by 10 strokes with a homogenizer. The homogenate (200 l) was loaded onto 200 l of 2 M sucrose in hypotonic buffer and centrifuged at 15,000 ϫ g for 30 min to pellet nuclei. To prepare the cytosol fraction, the supernatant was taken above the sucrose cushion and centrifuged at 100,000 ϫ g for 30 min. The pellet fraction was designated the postnuclear membrane fraction and solubilized in hypotonic buffer containing 0.5% Triton X-100 and 0.1% deoxycholate. Protein concentration was determined with a protein assay kit (Bio-Rad) with bovine ␥-globulins (Sigma) as a standard.
Cell Staining-NIH3T3 cells were cultured on glass coverslips and transiently transfected with mDvl-1 expression plasmids. 15 h after transfection, the cells on the coverslips were fixed with 3.7% formaldehyde in phosphate-buffered saline for 10 min at 37°C and treated with 0.5% Triton X-100 in phosphate-buffered saline for 10 min. After blocking with phosphate-buffered saline containing 3% bovine serum albu-min (Sigma) and 0.1% ␥-globulins (Sigma), the coverslips were incubated with the anti-Myc antibody (9E10, Santa Cruz Biotechnology, Inc.) and then washed three times with phosphate-buffered saline. Reacted proteins were detected by Cy3-conjugated goat anti-mouse secondary antibody (Amersham Pharmacia Biotech).
Luciferase Assays-10 h after transfection, NIH3T3 cells were washed once in ice-cold Hepes-buffered saline, scraped in 150 l of reporter lysis buffer (Promega), and centrifuged by 12,000 rpm for 5 min after vortex for 15 s. The supernatant was used as a cell extract to detect the luciferase activity. Luciferase assay was carried out using a luciferase assay system (Promega). In brief, 20 l of the room temperature cell extract was mixed with 100 l of room temperature luciferase assay reagent containing the substrate. The reaction was performed and measured in a luminometer (LB9507, Berthold). The protein concentration was determined with a protein assay kit (Bio-Rad) and used for normalization of the luciferase assays.
Xenopus Injection-Sense mRNAs were obtained from linearized templates using SP6 mMessage Machine kit (Ambion). Fertilized eggs were dejellied, and each of various mDvl-1 mutants mRNAs (1.5 ng) was injected into dorsal or ventral blastomeres at the four-cell stage. Injection was performed with 3% Ficoll in 0.1ϫ modified Barth's solution, and embryos were cultured for 1 or 2 days in 0.1ϫ modified Barth's solution. For kinase assays of JNK/SAPK in Xenopus embryos, animal caps were dissected from stage 8.5 embryos that had been injected with 0.3 g of mRNA encoding Myc-SAPK␣ together with 1.2 g of control mRNA or mRNA encoding HA-tagged wild-type mDvl-1 or mDvl-1(KM) at the animal pole. Each of the caps was crushed by pipetting in a buffer (40 l) consisting of 50 mM Tris-Cl, pH 7.5, 25 mM 2-glycerophosphate, 10 mM EGTA, 2 mM MgCl 2 , 2 mM dithiothreitol, 1 mM vanadate, 1 mM phenylmethylsulfonyl fluoride, and 0.2% aprotinin, and then centrifuged at 15,000 ϫ g for 30 min. The supernatant was used as an extract for kinase assay.

Activation of JNK/SAPK and MKK7 by Expression of mDvl-1-Expression of Drosophila
Dsh has been reported to induce c-Jun phosphorylation (21). To investigate a role of Dsh as activators for MAPK pathways, we tested whether four distinct classes of the MAPK superfamily molecules could be stimulated by a mouse homolog of Dsh, mDvl-1. Each of four HAtagged MAPK superfamily molecules (HA-MAPK, HA-SAPK␣, HA-p38␣, and HA-ERK5) was coexpressed with Myc-tagged mDvl-1 in NIH3T3 cells and immunoprecipitated with anti-HA antibody. The immunoprecipitates were subjected to an immune complex kinase assay. JNK/SAPK was strongly activated by expression of mDvl-1 (Fig. 1A, JNK/SAPK). p38␣ was also weakly activated, but MAPK or ERK5 was not activated by mDvl-1 (Fig. 1A). It was confirmed that all these transfected kinases could be activated by appropriate stimuli such as hyperosmolarity (data not shown). These results suggest that JNK/SAPK and p38 lie downstream of mDvl-1. We next determined which of MAPKK family molecules could be stimulated by mDvl-1. Each of six HA-tagged MAPKK family molecules was coexpressed with Myc-mDvl-1, and the kinase activities of MAPKKs were determined. Consistent with the activation of JNK/SAPK by mDvl-1, MKK7, a specific activator of JNK/ SAPK, was activated by coexpression of mDvl-1, whereas classical MAPKK and MEK5, specific activators of MAPK and ERK5, respectively, were not activated (Fig. 1B). Rather surprisingly, activation of MKK3 or MKK6, specific activators of p38, was not detected. One reason for this may be the low sensitivity of the MAPKK assay, compared with the MAPK assay, although it was confirmed that all the transfected kinases could be activated by appropriate stimuli (data not shown). The decrease of the amount of HA-tagged proteins by coexpression of Myc-mDvl-1 (see Fig. 1B) may further increase the difficulty in detecting the possible activation. Anyway, it was found that expression of mDvl-1 leads to strong activation of the MKK7-JNK/SAPK pathway.
Inhibition of the mDvl-1-induced Activation of JNK/SAPK by Dominant-Negative Cdc42 or Rac1-Genetic analysis in Drosophila showed that Drosophila RhoA and Rac, but not Cdc42, function downstream of Dsh in planar polarity signaling (21). To investigate potential roles of low molecular weight GTPases in the mDvl-1-induced JNK/SAPK activation in mammalian cells, dominant-negative forms of Ha-Ras, Rac1, Cdc42, and RhoA were coexpressed with HA-SAPK␣ and Myc-mDvl-1. mDvl-1-induced JNK/SAPK activation was significantly suppressed by dominant-negative Cdc42 or Rac1 (Fig. 1C), indicating that Rac1 and Cdc42 act downstream of mDvl-1 in the JNK/SAPK activation pathway. In contrast, dominant-negative Ha-Ras or RhoA did not interfere with the activation of JNK/SAPK (Fig. 1C). It is possible that functions of the Rho family GTPases in vertebrates are slightly different from those in Drosophila, although it is suggested that the Rho family GTPases lie downstream of Dsh or mDvl-1 in both vertebrates and invertebrates.
A C-terminal Half of mDvl-1 Is Required and Sufficient for JNK/SAPK Activation-It has previously been suggested that the DEP domain of Drosophila Dsh is important for the Dshinduced phosphorylation of c-Jun, because substitution of a single amino acid (Lys-417 to Met) within the DEP domain or deletion of the DEP domain abolished the c-Jun-phosphorylation-inducing activity of Dsh (21). To investigate the importance of DEP domain in the mDvl-1 induced activation of JNK/ SAPK in mammalian cells, NIH3T3 cells were cotransfected with HA-SAPK and each of various mutant forms of mDvl-1 (see Fig. 2A), and the activity of JNK/SAPK was determined. A mutant mDvl-1 lacking the DIX domain (⌬DIX) activated JNK/ SAPK as efficiently as did wild-type mDvl-1 (Fig. 2B, wt and  ⌬DIX), whereas a mutant lacking the DEP domain (⌬DEP) and a mutant in which Lys-438 (which corresponds to Lys-417 of Drosophila Dsh) within the DEP domain was replaced by Met (KM) failed to activate JNK/SAPK (Fig. 2B, ⌬DEP and KM). Moreover, the DEP domain-containing C-terminal portion of mDvl-1 (mDvl-1(DEP)) which lacks both the DIX and PDZ domains activated JNK/SAPK as efficiently as did wild-type mDvl-1 (Fig. 2B, DEP). We then measured the activity of endogenous JNK/SAPK by an immune complex kinase assay using anti-JNK1(p46 JNK/SAPK ) antibody, and found that endogenous JNK/SAPK was activated by overexpression of wild-type mDvl-1, mDvl-1(⌬DIX), or mDvl-1(DEP), but not by that of mDvl-1(⌬DEP) or mDvl-1(KM) (Fig. 2C). These results indicate that the C-terminal portion of mDvl-1, which contains the DEP domain, is required and sufficient for the JNK/SAPK activation.
An N-terminal Half of mDvl-1 Is Required for TCF/LEF-1mediated Transcription-Expression of Wnt-1 or ␤-catenin with LEF-1 in NIH3T3 cells is shown to activate TCF/LEF-1dependent transcription (28). To address whether mDvl-1 is able to activate the Wnt/␤-catenin pathway in NIH3T3 cells, we used a luciferase reporter assay. NIH3T3 cells were transfected transiently with pTOPFLASH, a luciferase reporter gene containing multimeric TCF/LEF-1 binding sites (29), together with expression plasmids of LEF-1 and mDvl-1. Expression of wild-type mDvl-1 with LEF-1 strongly increased transcription from pTOPFLASH (Fig. 2D, WT). This transcriptional activation was not observed with pFOPFLASH, a reporter gene containing multimeric mutant TCF/LEF-1 binding sites (data ting. Similar results were obtained in three different experiments. C, NIH3T3 cells were transfected with pSR␣HA-SAPK␣ and CS4-Myc-mDvl-1 together with an expression vector encoding Ha-RasN17, Rac1N17, Cdc42N17, or RhoAN19. The kinase activity of JNK/SAPK was measured by the immune complex kinase assay, and the amounts of immunoprecipitated HA-SAPK␣ (␣HA) and the presence of mDvl-1 in the lysates (␣Myc) were detected by immunoblotting. Similar results were obtained in three different experiments.

FIG. 1. Activation of the JNK/SAPK pathway by overexpression of mDvl-1.
A, an expression plasmid of an HA-tagged MAPK family molecule (1.0 g of pSR␣HA-MAPK, -SAPK␣, -p38␣, or -ERK5) was transiently cotransfected with indicated amounts of a Myc-tagged mDvl-1 expression plasmid (Myc-mDvl-1) into NIH3T3 cells. The empty vector was used to achieve equal amounts of each vector in each transfection. 15 h after transfection, HA-tagged proteins were immunoprecipitated by mouse anti-HA antibody. The immunoprecipitates were subjected to kinase assays as described under "Experimental Procedures." Samples were analyzed by SDS-polyacrylamide gel electrophoresis and molecular imager system (Bio-Rad). The amounts of immunoprecipitated HA-tagged proteins were detected by immunoblotting with rabbit anti-HA antibody (␣HA). The presence of mDvl-1 in the lysates was verified by immunoblotting with anti-Myc antibody (␣Myc). Similar results were obtained in three different experiments. B, an expression plasmid of an HA-tagged MAPKK family molecule (1.0 g of pSR␣HA-MAPKK, -MKK3b, -SEK1, -MEK5, -MKK6, and -MKK7) was transiently cotransfected with indicated amounts of a Myc-tagged mDvl-1 expression plasmid into NIH3T3 cells. 15 h after transfection, the kinase activity of HA-tagged proteins was measured by the immune complex kinase assay as described under "Experimental Procedures." The amounts of immunoprecipitated HA-tagged proteins (␣HA) and the presence of mDvl-1 in the lysates (␣Myc) were detected by immunoblot-not shown), indicating that transcriptional response to mDvl-1 is dependent on the presence of functional TCF/LEF-1 binding sites. Activation of the JNK/SAPK pathway did not increase the TCF/LEF-1-dependent transcription (Fig. 2D, MKK7/JNK).
Although mDvl-1 (⌬DIX), mDvl-1(⌬DEP), or mDvl-1(DEP) did not significantly increase the luciferase activity, mDvl-1(KM) did increase the activity to the same extent as did wild-type mDvl-1 (Fig. 2D). This result is consistent with the previous reports in Drosophila (21,30), which demonstrated that the DIX and PDZ domains were required for Armadillo (␤-catenin) stabilization, whereas the KM mutation in the DEP domain did not affect the ␤-catenin pathway. However, our result is different from the result reported by Yanagawa et al. (30) in which the DEP domain was shown to be unnecessary for ␤-catenin stabilization in Drosophila cells. This discrepancy might result from the differences in cell types, dishevelled proteins used (Drosophila or mouse), or assay systems.
Effect of the Addition of CAAX Motif on the JNK/SAPKactivating Activity of mDvl-1-Because the DEP domain has previously been shown to be required for Dsh relocalization to the plasma membrane (22), we considered the possibility that the translocation of Dsh protein might be involved in the activation of the JNK/SAPK pathway. Then, we made several constructs that expressed wild-type and mutant forms of mDvl-1 as fusions to the C terminus of CAAX motif of Ras to obtain membrane-bound forms of mDvl-1. We checked subcellular localization of these mDvl-1 mutants. Transfected NIH3T3 cells were fractionated into postnuclear membrane and cytosol fractions, and the fractions were subjected to immunoblotting. As shown in Fig. 3A, wild-type mDvl-1 protein localized in both the cytosol (lane C) and membrane fraction (lane M), whereas mDvl-1-CAAX protein was only in the membrane fraction. The DEP domain-containing C-terminal portion of mDvl-1 localized predominantly in the membrane fraction (Fig. 3A, DEP). The two mutants, mDvl-1(⌬DEP) and mDvl-1(KM), localized in both the cytosol and the membrane fractions, but, unlike wild-type mDvl-1, they were mainly present in the cytosol fraction. In contrast, mDvl-1(⌬DEP)-CAAX and mDvl-1(KM)-CAAX proteins localized predominantly in the membrane fraction (Fig. 3A, ⌬DEP-CAAX and KM-CAAX). The cell staining with indirect immunofluorescence showed that wild-type mDvl-1 and mDvl-1(KM) proteins were detected throughout the cytoplasm (Fig. 3B and data not shown), whereas CAAX-fused mDvl-1 proteins were concentrated in the plasma membrane ( Fig. 3B and data not shown). Thus, the CAAX motif targeted mDvl-1 proteins to the plasma membrane. It should be noted that CAAX-fused mDvl-1 proteinsexpressing cells underwent drastic morphological changes: rounding, shrinkage, and induction of protrusions and processes (see Fig. 3B).
In Drosophila, a mutation in the DEP domain (the KM mutation) impairs both the membrane localization of Dsh and the function of Dsh in planar polarity signaling, suggesting that translocation of Dsh is important for function (22). Conimmune complex was determined by immunoblotting with anti-JNK1 antibody (␣JNK). Similar results were obtained in four different experiments. D, NIH3T3 cells were transfected with pHA-LEF1 (0.03 g) and the reporter plasmid pTOPFLASH (0.3 g) together with an expression plasmid (0.5 g) for various mDvl-1 proteins (empty vector (control), WT, ⌬DIX, ⌬DEP, DEP, or KM) or together with expression plasmids for MKK7(DE) and JNK (0.3 g each), as indicated. The total amount of the DNA was adjusted to 1.2 g/35-mm dish with CS2 vector plasmid. 10 h after transfection, the cells were harvested, and the luciferase activity was measured as described under "Experimental Procedures." Average relative luciferase activities from three independent experiments were calculated relative to the activity of the control, which was set at 1. Error bars represent mean Ϯ S.D.  1.2 g of ⌬DEP). The endogenous JNK/ SAPK kinase activity was measured by an immune complex kinase assay (upper panel). The amount of endogenous JNK/SAPK in each sistent with this idea, the two mutants, mDvl-1(KM) and mDvl-1(⌬DEP), had no ability to activate JNK/SAPK (see Fig. 2, B and C and Fig. 3C). Interestingly, mDvl-1(KM)-CAAX was able to activate JNK/SAPK to some extent (Fig. 3C). In addition, a mDvl-1(KM) mutant, which was fused to the myristylation motif derived from chick c-Src at its N terminus, could also activate JNK/SAPK (data not shown). These results suggest that mDvl-1(KM) has a potential to activate JNK/SAPK if properly localized to membrane. In contrast, mDvl-1(⌬DEP)-CAAX could not activate JNK/SAPK (Fig. 3C). Thus, the DEP domain may have a role in localizing mDvl-1 to the plasma membrane to activate the JNK/SAPK pathway, but the function of the DEP domain is not solely the targeting of mDvl-1 to the plasma membrane. In the case of the Raf/MAPK pathway, growth factor stimulation recruits Raf-1 to the plasma membrane, and the membrane-bound form of Raf-1 (RafCAAX) is constitutively active (31,32). However, mDvl-1-CAAX had a lower activity than mDvl-1 to stimulate the activity of JNK/ SAPK (Fig. 3C). These results suggest that the mode of the CAAX-mediated membrane localization of mDvl-1 may be different from that of the DEP domain-mediated membrane localization, the latter being required for optimal activation of the JNK/SAPK pathway.
We then examined the effect of the addition of CAAX-motif to mDvl-1 on the TCF/LEF-1-mediated transcriptional activation. Because the membrane localization-disrupting mutant mDvl-1(KM) was able to increase the LEF-1-dependent transcriptional activation to the same extent as did wild-type mDvl-1 (see Fig. 2D), it was possible that the addition of CAAX might not have significant effects on the ability of mDvl-1 to activate the LEF-1-dependent transcription. In fact, the addition of the CAAX motif did not affect significantly the ability of various mDvl-1 mutants to induce the LEF-1-dependent transcription in luciferase assays (Fig. 3D). The most significant effect was the reduction of the ability of wild-type mDvl-1, the reason of which is not known at present. These results taken together indicate that the DEP domain-mediated membrane localization of mDvl-1 is important for the JNK/SAPK pathway but not for the ␤-catenin/LEF-1 pathway, although the DEP domain may have other roles in both pathways.
Distinct Domains of mDvl-1 Are Important for JNK/SAPK Activation and the Axis Formation in Xenopus Embryos-Several molecules in the Wnt signaling pathway including Dsh and ␤-catenin have been shown to be able to cause duplication of the body axis in Xenopus embryos (23,24). We then examined the abilities of mDvl-1 mutants to activate the JNK/SAPK and to induce the secondary axis in Xenopus embryos. Microinjection of wild-type mDvl-1 mRNA, but not that of mDvl-1(KM) mRNA, induced activation of JNK/SAPK in animal caps (Fig.  4A). Expression of mDvl-1(⌬DIX) or mDvl-1(DEP) in animal caps could also induce the activation of JNK/SAPK to a variable extent (data not shown). The reason of this variability is  Fig. 2D. 10 h after transfection, the cells were harvested and the luciferase activity was measured as described under "Experimental Procedures." Average relative luciferase activities from three independent experiments were calculated relative to the activity of the control, which was set at 1. Error bars represent mean Ϯ S.D. not known at present. Injection of wild type mDvl-1 mRNA into ventral blastomeres at the four-cell stage embryo triggered the formation of a secondary axis (Fig. 4B, WT). mDvl-1 mRNA injections into dorsal blastomeres at the same stage did not alter normal development (data not shown). Ventral injection of mDvl-1(⌬DEP) (data not shown) or mDvl-1(KM) mRNA, but not that of mDvl-1(⌬DIX) or mDvl-1(DEP) mRNA, triggered the secondary axis formation (Fig. 4B). These results demonstrate that in Xenopus embryos as well as in mammalian cultured cells, the KM mutation in the DEP domain has no effect on the ␤-catenin pathway whereas the intact DEP domain is required for JNK/SAPK activation.
In our preliminary experiments, dominant-negative or constitutively active forms of Xenopus MKK6 (XMEK3) and MKK7 did not interfere with normal axis formation or cause the axis duplication. 3 These results suggest that JNK/SAPK is not involved in the axis formation in Xenopus. Then, it should be elucidated what roles JNK/SAPK plays downstream of Dishevelled in Xenopus. The Wnt genes can be classified into at least two subfamily members, Wnt1 and Wnt5a members (33,34). In Xenopus embryos, expression of members of the Wnt1 subfamily induces axis duplication, whereas expression of the other subfamily alters morphogenetic movements during gastrulation. It is tempting to speculate that JNK/SAPK functions downstream of members of the Wnt5a subfamily.
In summary, this study has shown that mDvl-1 is able to induce the activation of both the ␤-catenin pathway and the JNK/SAPK pathway in both mammalian cultured cells and Xenopus embryos. As the KM mutation in the DEP domain did not affect the LEF-1-dependent transcription or the axis duplication, the DEP domain-mediated membrane localization may not be important for the activation of the ␤-catenin pathway. On the other hand, the KM mutation abolished the ability of mDvl-1 to activate the JNK/SAPK pathway, and the intact DEP domain is sufficient for activation of this pathway in mammalian cells. Therefore, it is suggested that the DEP domain-containing C-terminal half and the DIX and PDZ domaincontaining N-terminal half are important for the JNK/SAPK pathway and for ␤-catenin pathway, respectively. Thus, bifurcation of the Wnt signaling pathway at the point of Dishevelled may be a conserved mechanism in both vertebrates and invertebrates. This study may provide the molecular basis by which the bifurcation is achieved and assured.
After the completion of this study, a paper of Li et al. (35) appeared, which showed the regulation of LEF-1 and JNK/ SAPK by Dvl proteins in mammalian cells.