Profilin is an effector for Daam1 in non-canonical Wnt signaling and is required for vertebrate gastrulation

Establishment of the vertebrate body plan depends on cell movements during gastrulation. This process generates the extended axis of the embryo through directed cell migration, termed `convergent extension movements' during which cells polarize, elongate, align and intercalate, resulting in mediolateral narrowing (convergence) and anteroposterior lengthening (extension) of the axis (Keller, 2002). The directionality of these cell movements results from stabilization of protrusions (lamellopodia) on the mediolateral surfaces(Wallingford et al., 2000). Blastopore closure occurs as a result of the sequential circumferential shortening that is mediated by a `purse string' mechanism that constricts the circumference of the blastopore (Keller et al., 2003). In subsequent primary neurulation the lateral edges of the neural plate elevate until they appose at the midline and fuse to form the neural tube; convergent extension movements provide the force for this process(Wallingford and Harland,2002). The molecular mechanisms that control convergent extension movements, blastopore closure and neural tube folding and closure are incompletely understood, but modification of the cytoskeleton via the non-canonical Wnt signaling pathway is required for these processes(Keller, 2002; Wallingford et al., 2002).

The non-canonical Wnt signaling pathway, also termed the β-catenin independent pathway or the planar cell polarity (PCP) pathway, regulates gastrulation cell movements among other processes, and is mediated by the PDZ and DEP domains of Dishevelled (Dvl)(Wallingford and Habas, 2005). This pathway regulates cell movements through modification of the actin cytoskeleton and appears to be independent of transcription(Veeman et al., 2003; Wallingford and Habas, 2005). The specificity of Wnt ligands for distinct branches of the pathway remains poorly deciphered but Wnt5a was first identified as a regulator of gastrulation (Moon et al.,1997). However, Wnt5a can also activate canonical signaling(He et al., 1997; Mikels and Nusse, 2006). Likewise, Wnt3a can activate both canonical and non-canonical signaling(Cadigan and Liu, 2006). Additional known components of the non-canonical pathway include Wnt11, Fz,Dvl, Daam1, Rho, Rac, Jun kinase (JNK), Strabismus and Prickle(Habas and Dawid, 2005; Wallingford and Habas, 2005). In this pathway the Wnt signal is mediated through Fz independently from LRP5/6 (He et al., 2004) and the pathway bifurcates downstream of Dvl into two parallel branches that lead to the activation of the small GTPases Rho and Rac(Habas et al., 2003; Habas et al., 2001; Tahinci and Symes, 2003). Signaling to Rho involves Daam1, which binds to the PDZ domain of Dvl(Habas et al., 2001), leads to the activation of the Rho associated kinase Rock, and mediates cytoskeletal reorganization (Marlow et al.,2002; Veeman et al.,2003; Wallingford et al.,2002). The second branch requires the DEP domain of Dvl but not Daam1 and activates the Rho family GTPase, Rac, which in turn stimulates JNK activity (Habas et al., 2003; Li et al., 1999; Yamanaka et al., 2002).

Daam1 is a member of the Formin family of proteins that are central players in cytoskeletal reorganization (Alberts,2002; Wallar and Alberts,2003). Formin proteins contain three major domains termed the GTPase binding domain (GBD), Formin Homology 1 (FH1) and Formin Homology 2(FH2) domains (Alberts, 2002). It is proposed that Formin proteins exist in the cytoplasm in an autoinhibited state, which is mediated by binding of the C-terminal Diaphanous auto-inhibitory domain (DAD) to the amino terminus(Alberts, 2002; Higgs, 2005). It has been proposed that activated Rho-GTP binds to the GBD domain, releasing the protein from autoinhibition, followed by binding of the FH1 and FH2 domains to effectors to elicit cytoskeletal changes. The FH2 domain can polymerize actin filaments and serves an actin nucleation function(Higgs, 2005; Kovar et al., 2006; Kovar and Pollard, 2004). One molecule known to bind the FH1 domain of the Formin proteins is Profilin,which binds the Formin mDia1 (Evangelista et al., 2002; Frazier and Field, 1997; Severson et al.,2002).

Profilin is an evolutionarily conserved actin binding protein that is involved in actin polymerization (Watanabe and Higashida, 2004; Witke,2004). Genetic studies in Drosophila have uncovered roles for Profilin1 in oogenesis, spermatogenesis, bristle and eye formation(Cooley et al., 1992; Verheyen and Cooley, 1994) and in neuronal cells for axonal guidance and dendritic spine morphology(Witke, 2004). Among the three mammalian profilins, Profilin1 is essential for cytokinesis and mouse knockout mutants die by the two- to eight-cell stage(Witke et al., 2001; Witke, 2004). Profilin1 interacts with the FH1 domain of Formin proteins and serves an actin monomer delivery and capping function (Higgs,2005; Kovar and Pollard,2004; Zigmond,2004). Whether this Profilin-Formin interaction is required for a morphogenetic process in vivo remains unknown. To date, no signaling pathway is known to require Profilin1, and the function of Profilin1 during embryogenesis remains poorly defined.

Here we report the identification of Profilin1 as an interacting partner of Daam1 and a functional component of the non-canonical Wnt signaling pathway. Profilin1 binds to the FH1 domain of Daam1 and colocalizes with Daam1 to actin stress fibers in response to Wnt stimulation. Depletion of Profilin1 inhibits Wnt- and Daam1-mediated stress fiber formation. Xenopus Profilin1 is expressed in the embryo at a time and place consistent with a role in gastrulation. Overexpression or depletion of Profilin1 results in inhibition of blastopore closure but convergent extension, tissue separation and neural fold closure are unaffected. Furthermore, Profilin1 has no role in canonical Wnt signaling and mesoderm specification. Together, our studies reveal a branch point in the non-canonical Wnt pathway that controls a specific aspect of vertebrate gastrulation.

Monoclonal antibodies (mAbs) against HA (F-7), RhoA (26C4), Dvl2 (10B5),Myc (9E10), and polyclonal Abs (pAbs) against RhoA (CAT119) and Myc (N-262)were from Santa Cruz Biotechnology. mAbs against Rac and Cdc42 were from Transduction Laboratories, and against Flag (M2) was from Sigma. Alexa Fluor anti-mouse and anti-rabbit Abs, Texas Red X-Phalloidin and Oregon Green-Phalloidin were from Molecular Probes (Eugene, OR). Anti-β-catenin antibody was from Transduction Laboratories (San Diego, CA).

The human Daam1 and fragments of Daam1 were generated by restriction digestion or a PCR approach, and subcloned in pCS2+MT (for the Myc tag at the N terminus) or pcDNA-HA (for the HA tag at the amino terminus), or pCS2+GFP vector (kindly provided by Dr Jeffrey Miller, University of Minnesota). Rat Profilin1 (isolated from our screen) and Xenopus Profilin1 (isolated by a PCR approach from a Xenopus Stage 10.5 cDNA library) were cloned into pCS2+MT or pCS2+GFP. Details of plasmids are available upon request.

The dsRNAi oligonucleotides for Profilin1 or control GFP oligonucleotides were synthesized using the Dicer Kit (Ambion) and purified following the manufacturer's instructions.

The XProfilin1 Morpholino oligonucleotide (MO) complementary to the translational initiation site, 5′-TGTAGCCGTTCCAAGACATTGTTGT-3′,was synthesized by Gene Tools. A MO with a random sequence was used as the negative control.

A rat brain cDNA library (Clonetech) was screened using the c-Daam1 fragment of Daam1 (Fig. 1A) as the bait. 3.9 million independent clones were screened, and 12 overlapping Profilin1 fragments, in addition to other positives, were obtained.

All were carried out with HEK293T cells or NIH3T3 cells. Cells in a sixwell plate were transfected using the calcium-phosphate method, Polyfect reagent(Qiagen), or Dicer transfection reagent (for RNAi experiments) with 1-2 μg of each indicated plasmid or 500 ng-1 μg annealed RNAi oligo plus 1 μg plasmid. Transfected DNA amounts were equalized via vectors without inserts.

Wnt3a-transfected, Wnt5a-transfected or control L cells were obtained from ATCC and cultured according to the suppliers instructions. Serum-free Wnt3a,Wnt5a and L cell condition medium were prepared according to the manufacturers instructions. Purified Wnt3a protein was purchased from R&D systems and used at a concentration of 250 ng/ml.

Anti-Daam1 antibodies were generated in rabbits against a GST fusion protein containing amino acids 967-1078 of human Daam1. The Daam1 specific antibody was affinity purified using the GST-Daam1 fusion protein by standard methods.

This was carried out as described previously(Capelluto et al., 2002; Habas et al., 2001). Images were obtained using an Olympus IX70 fluorescent microscope with 100X objective lens (Melville, NY) or a Zeiss Axiovert 100 microscope (Oberkochen, Germany). For quantification of localization of Daam1 or Profilin1 to stress fibers, a base line of 10 stress fibers per cell was used and the merged image of Daam1 or Profilin1 onto these fibers was counted as a positive. For quantification of the effects of depletion of Profilin1 and Wnt- and Daam1 mediated stress fiber induction, a base line of 10 stress fibers per cell was used to score,thus any cell containing more than or less than10 fibers was scored as an increase or decrease respectively. These experiments were repeated at least three times and scoring was done in a blind manner so that the scorer had no knowledge of the sample being scored.

These were performed as described(Habas et al., 2003; Habas et al., 2001; Kato et al., 2002). Embryo injections were done with in vitro transcribed RNAs. Convergent extension assays in explants were performed as described(Habas et al., 2003) using 5 ng/ml activin. Keller explant assays were performed as described(Shih and Keller, 1992). Tissue separation assays were performed as described(Hukriede et al., 2003).

We had shown that a fragment of Daam1 termed C-Daam1 can induce Rho activation and cytoskeletal changes (Habas et al., 2001) (Fig. 1A). In order to identify downstream effectors for Daam1 for these processes, we performed a yeast two hybrid screen using C-Daam1 as bait. From this screen, we isolated 12 overlapping clones of Profilin1 as a C-Daam1-interacting protein.

To delineate interactions between Profilin1 and Daam1 outside of yeast, we examined Profilin1 interaction with Daam1 by coimmunoprecipitation using epitope-tagged wild type and mutant proteins expressed in mammalian HEK293T cells (Fig. 1A). We found that Profilin1 binds to full length Daam1 and C-Daam1, which contains the FH1 and FH2 domains, but not to N-Daam1, which contains the amino-terminal domain(Fig. 1B). Using smaller fragments of C-Daam1 harboring the FH1 or FH2 domains separately(Fig. 1A), we localized the Profilin1 interacting domain to the FH1-containing fragment of Daam1 but not to the FH2-containing fragment (Fig. 1B).

We next examined whether endogenous Profilin1 interacts with epitope-tagged Daam1. For these experiments, we utilized a commercial Profilin1 antibody(Cytoskeletal Labs) that is functional for western blotting but not immunoprecipitation or immunocytochemistry. In agreement with the above results, full-length Daam1 and C-Daam1 but not N-Daam1 interacted with endogenous Profilin1 (Fig. 1C). Additionally the FH1-but not the FH2-containing construct of Daam1 was found to interact with endogenous Profilin1 (Fig. 1C).

As C-Daam1 can mediate Rho activation and cytoskeletal changes(Habas et al., 2001), we tested whether Profilin1 functions in C-Daam1-mediated RhoA activation using the Rhotekin assay (Habas and He,2006; Ren et al.,1999) and extracts of mammalian cells transfected with C-Daam1,Profilin1 or both. As in previous studies, C-Daam1 expression induced Rho activation but expression of Profilin1 did not(Fig. 1D). In addition,co-expression of Profilin1 did not interfere with C-Daam1-mediated Rho activation (Fig. 1D). This suggests that Profilin1 does not function in Daam1-mediated Rho activation.

We next examined the effects of Profilin1 on the actin cytoskeleton in NIH3T3 cells, using Wnt3a and Wnt5a conditioned media (CM) or Wnt3a protein. Previous studies have revealed a dramatic cytoskeletal reorganization of COS or B cells in response to Wnt3a stimulation(Endo et al., 2005; Qiang et al., 2003), but such effects on NIH3T3 cells were not reported. We found that treatment of NIH3T3 cells with Wnt3a CM but not control CM for 3 hours resulted in nuclear accumulation of β-catenin and robust induction of stress fibers(Fig. 2A). An identical effect on stress fiber induction and nuclear accumulation of β-catenin was observed with purified Wnt3a protein (Fig. 2A). We next examined the effects of Wnt5a CM. These studies revealed a robust induction of stress fiber formation but no nuclear accumulation of β-catenin (Fig. 2A). Lastly, a mutant construct of Dishevelled,ΔDIX-Dishevelled that is solely involved in non-canonical Wnt signaling(Habas et al., 2001; Tada and Smith, 2000; Wallingford et al., 2000),could induce the formation of stress fibers(Fig. 2B) without induction of nuclear β-catenin (not shown). These results demonstrate that NIH3T3 cells respond to non-canonical Wnt signaling with the induction of actin stress fibers.

Previous studies have revealed variable results on the effects and localization of Profilin1 to the actin cytoskeleton(Cao et al., 1992; Roy and Jacobson, 2004; Witke, 2004). We examined the localization of GFP-Profilin1 fusion protein in response to Wnt3a stimulation in NIH3T3 cells. Although GFP-Profilin1 was diffusely localized in the cytoplasm of NIH3T3 cells treated with control CM(Fig. 2C), treatment with Wnt3a CM resulted in colocalization of GFP-Profilin1 with actin stress fibers(Fig. 2C,E). These studies suggest that Profilin1 may be a component of non-canonical Wnt signaling that modulates the actin cytoskeleton.

To investigate the role of Profilin1 in mediating non-canonical Wnt signaling, we explored the localization of Profilin1 and Daam1 in mammalian cells. Staining with affinity purified α-Daam1 polyclonal sera showed that Daam1 is localized mainly in the cytoplasm of NIH3T3 cells(Fig. 2D). In response to Wnt3a CM stimulation, Daam1 relocalized predominantly to actin stress fibers and to a lesser extent to the plasma membrane(Fig. 2D,E). As this response was similar to that of Profilin1 (Fig. 2C), we examined whether Daam1 and Profilin1 are colocalized in response to Wnt stimulation and found this to be the case for both transfected and endogenous Daam1 (Fig. 3A,B). This colocalization in response to Wnt stimulation suggest that Daam1 and Profilin1 may function in a common molecular pathway to mediate effects on the actin cytoskeleton

We next determined whether Profilin1 is required for Wnt- and Daam1-mediated cytoskeletal changes. We employed dsRNA-mediated interference(RNAi) to deplete endogenous Profilin1 by more than 60% without affecting the levels of β-catenin or Daam1 (Fig. 3C). Transfection of Profilin1 or control siRNA did not inhibit Top-flash reporter activation in NIH3T3 cells mediated by Wnt3a CM or Dishevelled (Fig. 3D). Importantly in cells depleted of Profilin1 stress fiber induction in response to Wnt3a CM, purified Wnt3a or ΔDIX-Dishevelled, but not in response to serum stimulation, was markedly decreased whereas the nuclear accumulation ofβ-catenin was unaffected (Fig. 4A-C). Additionally in cells transfected with C-Daam1, which induces stress fiber formation (Habas et al., 2001), a significant reduction of stress fiber formation was observed in cells depleted of Profilin1(Fig. 5A,B). These studies demonstrate that Profilin1 is required for Wnt- and Daam1-mediated cytoskeletal changes.

To help elucidate the in vivo role of Profilin1, we examined the expression pattern of Profilin1 during Xenopus embryogenesis. XProfilin1 shares 48.7% identity with the rat and human proteins (see Fig. S1A in the supplementary material). RT-PCR analysis showed that Profilin1 is expressed maternally and throughout development (see Fig. S1B in the supplementary material). The spatial pattern of Profilin1 gene expression visualized by in situ hybridization revealed a dynamic expression profile in the developing embryo especially in regions associated with morphogenetic movements. Profilin1 was expressed in the animal pole of the fertilized egg (see Fig. S1C in the supplementary material). At the blastula stage, Profilin1 was observed circumferentially around the blastopore lip and in the involuting mesodermal cells (see Fig. S1C in the supplementary material). During the neurula stage Profilin1 was expressed in the neural folds and anterior neural plate and during later development was expressed at higher levels in the brain, eyes and spinal cord (see Fig. S1C in the supplementary material). We note that the expression pattern of Profilin1 overlaps with that of Daam1(Nakaya et al., 2004).

To elucidate the function of Profilin1 in vivo, we examined the effects of misexpression of Profilin1 during Xenopus development. Injection of Profilin1 RNA into the two ventral marginal blastomeres of the four-cell embryo had no effect on Xenopus development in a concentration range of 100 pg to 1 ng (Fig. 6A,E). In contrast injection of Profilin1 RNA into the dorsal marginal zone of the fourcell embryo resulted in severe gastrulation defects in a dosage dependent manner whereas injection of LacZ RNA had no significant effect(Fig. 6A,E). In Profilin1-injected embryos anterior structures including the head and eyes were reduced and the neural folds failed to close. This phenotype is suggestive of a role of Profilin1 in gastrulation cell movements and blastopore closure (Wallingford et al.,2002).

To study the role of Profilin1 in early Xenopus development by a loss-of-function approach, we designed an antisense MO to deplete the endogenous Profilin1 protein (Fig. 6B). As the commercial Profilin1 antisera did not recognize endogenous Xenopus Profilin1 by western blot analysis (not shown), we tested the efficiency of the Profilin1 MO to inhibit translation of a Myc-Profilin1 construct in the Xenopus embryo. We found a dose dependent suppression of translation of Myc-Profilin1, but not of Rho and Rac by injection of 25 and 50 ng of the Profilin1 MO and no effect was observed using a control MO (Fig. 6C,E). Injection of the Profilin1 MO or a control MO at a level of 50 ng into the marginal zone of the ventral two cells of the four-cell embryo had little effect on Xenopus development(Fig. 6D,E). Similarly injection of control MO into the dorsal cells had little effect but injection of 10-50 ng of the Profilin1 MO resulted in embryos with open neural folds in a dose dependent manner (Fig. 6D,E). The phenotype induced by injection of the maximal dose of 50 ng Profilin1 MO could be rescued by coinjection of 100 pg of aΔN-Profilin1 RNA, a construct in which the MO binding site was deleted,but not by 100 pg of LacZ RNA (Fig. 6D,E). These experiments demonstrate the specificity of the Profilin1 MO and also demonstrate that Profilin1 is required for gastrulation.

To investigate whether Profilin1 can regulate canonical Wnt signaling, we performed secondary axis induction assays(McMahon and Moon, 1989; Sokol et al., 1991). Expression of Xwnt8 or Dsh ventrally in Xenopus embryos induced the formation of a secondary axis, and co-expression of Profilin1 had no effect on this induction (see Fig. S2A,B in the supplementary material). Furthermore,Profilin1 or Profilin1 MO had no effect on Xwnt-8 or Dsh induction of Xnr3 and Siamois(Harland and Gerhart, 1997)(see Fig. S2C in the supplementary material). These results support the view that Profilin1, like Daam1, functions specifically in noncanonical Wnt signaling.

One mechanism by which Profilin1 overexpression or depletion may affect gastrulation might involve the disruption of mesodermal cell fate specification. We tested this possibility by examining marker gene expression in animal cap explants treated with activin. Neither injection of Profilin1 RNA nor Profilin1 MO inhibited the expression of the panmesodermal marker brachyury (Xbra), dorsal mesodermal marker chordin,and ventrolateral mesodermal marker Xwnt8 (see Fig. S2D in the supplementary material).

We next examined the expression and localization of mesendodermal and neural markers by in situ hybridization. Embryos were injected dorsally at the 4-cell stage with Profilin1 RNA, Profilin1 MO, control MO, ΔN-Profilin1 RNA or Profilin1 MO plus ΔN-Profilin1 RNA. The Profilin1 and Profilin1 MO injected embryos were abnormal, and the latter could be rescued by co-injecting ΔN-Profilin1 RNA as described above(Fig. 6D,E). At stage 10.5 Profilin1 and Profilin1 MO-injected embryos showed normal mesoderm and dorsal axis formation as assayed by the mesodermal marker Xbra, and dorsal Gsc and Otx2 expression(Fig. 7). At stage 12 such embryos expressed Xbra surrounding a large blastopore that failed to close (Fig. 7). At stage 13 Gsc expression in control embryos was observed in the prechordal plate in the deep anterior mesendoderm whereas in Profilin1 and Profilin1 MO-injected embryos, Gsc expression remained near the open blastopore(Fig. 7). Otx-2 was expressed anteriorly in both mesodermal and overlying neural tissues in control embryos at stage 13. However, in Profilin1 and Profilin1 MO-injected embryos, two separate Otx-2 expression domains were seen(Fig. 7), one next to the open blastopore, which probably reflects expression in the anterior mesoderm that fails to involute, and the other in neural ectoderm that may be induced via planar neural induction. Sox-2 is a pan-neural marker that marks the neural plate at stage 14. In Profilin1 and Profilin1 MO-injected embryos, Sox-2 expression was seen in a broad dorsal region that surrounds the open blastopore but lacks neural plate morphology(Fig. 7).

These experiments demonstrate that overexpression or depletion of Profilin1 does not interfere with mesoderm specification.

The phenotypes observed with overexpression or depletion of Profilin1 suggest a role in gastrulation. To delineate whether Profilin1 functions in convergent extension movements, we first examined the effects of overexpression or depletion of Profilin1 on activin-treated animal explants which exhibit morphogenetic elongation characteristic of gastrulation(Symes and Smith, 1987). We observed that expression or depletion of Profilin1 at doses that resulted in severe gastrulation phenotypes (Fig. 5A,D,E) had no effect on elongation of the explants(Fig. 8A,B).

This result was surprising as all known components of the noncanonical Wnt pathway including Wnt11 (Tada and Smith,2000), Fz7 (Djiane et al.,2000), Dsh (Sokol,1996), Daam1 (Habas et al.,2001), Rho (Habas et al.,2003; Tahinci and Symes,2003) and Rac (Habas et al.,2003; Tahinci and Symes,2003) potently inhibit convergent extension movements. We therefore tested the role of Profilin1 in convergent extension movements using dorsal marginal zone (Keller) explants, which more closely reflect the tissue undergoing convergent extension movements in vivo(Keller et al., 1985). Again we observed that expression or depletion of Profilin1 had no effect on elongation of the Keller explants (Fig. 8C,D), whereas dominant negative Dsh, Xdd1, potently inhibited elongation (Fig. 8C,D). These results demonstrate that Profilin1 is not required for convergent extension movements during gastrulation.

As our results above show, Profilin1 does not regulate convergent extension movements. The phenotypes we observed with expression or depletion of Profilin1 nevertheless suggest a role for this protein in gastrulation, which thus might function in tissue separation, blastopore closure, neural fold closure, or any combination thereof. To examine tissue separation, we injected Profilin1 RNA or MO and a tracer GFP RNA into the dorsal cells of the 4-cell embryo, explanted the axial mesodermal region of the injected embryos, and cultured the explants on wild type animal caps. No effect on tissue separation was observed (Fig. 8E,F). As a positive control we injected Xfz7, which was demonstrated to interfere with tissue separation (Winklbauer et al.,2001), and could confirm these observations(Fig. 8E,F). These studies indicate that Profilin does not regulate tissue separation during gastrulation.

We next examined the role of Profilin1 in neural fold closure. For this purpose we employed the strategy of Wallingford and Harland(Wallingford and Harland,2002) by injecting Profilin1 RNA or MO into the dorsal medial or dorsal marginal cells of the 16-cell embryo. These injections selectively target the medial or lateral neural plate, respectively(Fig. 9A). These studies revealed that Profilin1 RNA and Profilin1 MO did affect neural fold closure when injected medially but not laterally(Fig. 9B). The effects of Profilin1 MO could be rescued by co-expression of the ΔN-Profilin1 RNA(Fig. 9B), and control MO injection dorsally or medially had no effect (not shown). However, the neural fold closure defects in these embryos were localized to the posterior of the embryos suggesting that the phenotype may reflect a defect in blastopore closure rather than a direct effect on neural fold closure. Indeed we observed that Profilin1 RNA or Profilin1 MO-injected embryos displayed a delay in blastopore closure, and in a majority of injected embryos blastopore closure failed (Fig. 6A,D). These data strongly suggests a role for Profilin1 specifically in blastopore closure rather than neural fold closure.

We next examined the process of blastopore closure directly using time lapse imaging after injection with Profilin1 RNA into the dorsal marginal zone of the four-cell embryo. These studies revealed a dramatic delay and in the majority of injected embryos a complete failure of blastopore closure(Fig. 9C). Depletion of endogenous Profilin1 using Profilin1 MO revealed a similar failure of blastopore closure and this phenotype could be rescued by co-injection ofΔN-Profilin1 (Fig. 9C).

Lastly, we performed experiments to simultaneously deplete Profilin1 and Daam1 to test for interactive effects on gastrulation and blastopore closure,which would be expected if the two factors functions in the same pathway. Embryos were injected with subthreshold levels of Daam1 MO or Profilin1 MO,which individually induce failure of blastopore closure and gastrulation in only a minority of the embryos (Fig. 9D). However, when both Daam1 and Profilin1 MO were co-injected,the number of affected embryos increased substantially(Fig. 9D). Taken together these studies suggest that Profilin1 functions with Daam1 in the non-canonical signaling pathway during gastrulation.

In this study we demonstrated that Profilin1 is an effector of Daam1 in non-canonical Wnt signaling. Profilin1 binds to the FH1 domain of Daam1 and colocalizes with Daam1 to actin stress fibers in response to Wnt stimulation. Depletion of Profilin1 specifically inhibits stress fiber formation induced by Wnt stimulation and Daam1 overexpression. We further showed that Profilin1 is required downstream of Daam1 for blastopore closure during Xenopusgastrulation. These findings illuminate a molecular pathway linking Wnt signaling to the regulation of the cellular cytoskeletal architecture.

Daam1 is a Formin protein that is required for non-canonical Wnt signaling(Habas et al., 2001). Daam1 functions as a link between Dishevelled and the Rho GTPase in mediating cytoskeletal changes required for gastrulation cell movements but how Daam1 accomplishes these effects is not known. It is likely that Daam1 acts as a scaffolding protein and utilizes independent domains (GBD, FH1, FH2, etc.) to recruit and regulate factors required for Rho activation and cytoskeletal changes.

Profilin1 was biochemically purified as one of the first actin binding proteins and Profilin1 can stimulate the polymerization of actin filaments in vitro but the signaling pathways that require Profilin1 for their function and morphogenetic process regulated by Profilin1 remained unknown(Witke, 2004). Here we report a functional role for Profilin1 in Wnt- and Daam1-mediated cytoskeletal changes and for blastopore closure during embryogenesis.

Profilin1 interacts with Daam1 as assayed by coimmunoprecipitation(Fig. 1B), and we demonstrate binding between endogenous Profilin1 and Daam1 indicating a physiological interaction (Fig. 1C). Profilin1 binds to the FH1 domain of Daam1(Fig. 1B,C). An interaction between Profilin and the FH1 domain of Formins such as mDia1 has been reported previously and Profilin was implicated as an effector for Formin proteins in mediating actin polymerization (Wallar and Alberts, 2003; Watanabe et al., 1997). Recent studies have shown that the FH2 domain of Formins including mDia1 can stimulate the polymerization of actin filaments in vitro (Krebs et al., 2001) and it likely that this activity is coordinated with factors such as Profilin that bind to the FH1 domain for morphogenesis.

The non-canonical Wnt pathway plays important roles in cell polarization and cytoskeletal reorganization. In mammalian cultured cells, stimulation through the non-canonical Wnt pathway induces shape changes and regulates motility (Endo et al., 2005; Qiang et al., 2003; Shibamoto et al., 1998; Torres and Nelson, 2000). Dishevelled is required in vivo in the formation and stabilization of lamellopodial protrusions that regulate cell movements during gastrulation,and the small GTPase Rho acts downstream of Dishevelled in this signaling cascade (Endo et al., 2005; Wallingford et al., 2000). We showed that Profilin1 does not induce or inhibit Rho activation(Fig. 1D), suggesting that Profilin1 is required for cytoskeletal changes in addition to the Rho pathway downstream of Dishevelled and Daam1.

Characterizing the requirement for Profilin1 in cytoskeletal changes, we showed that Profilin1 and Daam1 are localized to actin stress fibers in response to Wnt stimulation in NIH3T3 cells(Fig. 2C,D). To our knowledge,this is the first time that a Formin protein or Profilin has been localized to actin stress fibers in mammalian cells. These observations suggest that the activity of Profilin1 may be regulated by a Wnt-dependent mechanism. Furthermore, depletion of Profilin1 abrogates stress fiber formation induced by Wnt, ΔDIX-Dishevelled or C-Daam1, demonstrating a requirement for Profilin1 in cytoskeletal changes mediated by non-canonical Wnt signaling(Fig. 4A-C and Fig. 5A,B). It is important to note that in these depletion studies dramatic effects on the actin cytoskeleton but no effects on the accumulation of nuclear β-catenin were observed (Fig. 4A), supporting the notion that morphological changes are independent of the canonical Wnt pathway.

Wnt signaling branches into three main pathways downstream of Dishevelled and considerable effort has been expended to decipher how Dishevelled channels signaling into these pathways (Habas and Dawid, 2005; Wallingford and Habas, 2005). Previously we have demonstrated that Daam1 is not a component of canonical signaling (Habas et al., 2001) and we have shown here that Profilin1 did not induce a secondary axis, did not interfere with Wnt- or Dishevelled-induced secondary axis formation and did not inhibit target genes of the canonical pathway in animal explants (see Fig. S1A-C in the supplementary material). We therefore conclude that Profilin1 does not play a role in canonical Wnt signaling.

The gastrulation defects observed with overexpression or depletion of Profilin1 did not result from a failure of mesodermal specification, as expression levels of all mesodermal marker genes tested were unaffected by manipulating Profilin1. Likewise, expression levels of neural markers were not affected by manipulating Profilin1 levels (see Fig. S1D in the supplementary material and Fig. 7), although gastrulation defects led to spatial mislocalization of both mesodermal and neural markers. These results are consistent with those of previous studies showing that mesodermal specification is unaltered by inhibition of non-canonical Wnt signaling (Djiane et al.,2000; Habas et al.,2001). We therefore conclude that Profilin1 does not affect mesodermal and neural specification in the embryo.

Vertebrate gastrulation involves a dynamic series of cell polarization and migration events that mediate blastopore closure, axial extension and neural fold closure (Wallingford et al.,2002). This morphogenetic process is regulated by non-canonical Wnt signaling, and Dishevelled is an important component of this pathway(Tada and Smith, 2000; Wallingford et al., 2000). We have previously shown that a Wnt-11/Fz/Xdsh/Daam1/Rho signaling pathway regulates convergent extension movements during gastrulation(Habas et al., 2001), and many studies have implicated the non-canonical Wnt pathway in blastopore closure,convergent extension movements, tissue separation and neural fold closure(Veeman et al., 2003; Wallingford et al., 2002). Whether distinct effectors are specifically utilized in the different morphogenetic processes has remained unclear(Keller, 2002; Keller et al., 2003).

We have examined the role of Profilin1 in each of these morphogenetic processes. In animal cap explants treated with activin or in Keller explants,expression or depletion of Profilin1 has no effect on elongation(Fig. 8A-D). We tested for a role of Profilin1 in tissue separation and observed no effects(Fig. 8E,F). We further tested the role of Profilin1 in neural fold closure using targeted injections to direct overexpression or depletion of Profilin1 in different regions and observed effects on neural fold closure exclusively in the posterior region of the embryo (Fig. 9B),indicative of an indirect effect through inhibition of blastopore closure. Embryos overexpressing or depleted in Profilin1 show delayed blastopore closure (Fig. 9C), which can explain the defects observed in such embryos. A failure of blastopore closure will result in a failure of the neural fold to close in the posterior although axial extension will be normal if convergent extension is unaffected.

These results are intriguing in the context that no single component of the non-canonical Wnt pathway has been shown previously to be required for blastopore closure alone. Force generation for blastopore closure probably involves the actin cytoskeleton (Keller et al., 1985), and we suggest that Profilin1 mediates a signal derived from the non-canonical Wnt pathway to the actin cytoskeleton. The identification of Profilin1 as a molecular component specifically required for blastopore closure provides a branch point in the non-canonical pathway for this specific morphogenetic event.

It is clear that cell motility during gastrulation requires dynamic changes to the cytoskeleton and to cell polarity, involving polarization of the migrating cells for mediolateral intercalation and convergent extension movements (Keller, 2002; Keller et al., 2003; Wallingford et al., 2002). Additionally these movements are dependent on the stabilization of protrusions termed lamellipodia, which are controlled by Dishevelled(Wallingford et al., 2000). However, what factors control the active assembly and disassembly of the microtubule network and cellular actin cytoskeleton are poorly resolved. Recent studies have identified a dynamic requirement for changes to the microtubule cytoskeleton and a Rho-GEF was identified for this process(Kwan and Kirschner, 2005). Furthermore, the Formin protein Daam1 mediates signaling from Dishevelled to the actin cytoskeleton for convergent extension movements(Habas et al., 2001). We propose here that Daam1 utilizes the effector molecule Profilin1 to mediate actin polymerization for cell motility during gastrulation. As Wnt signaling induces a colocalization of Daam1 and Profilin1 to actin stress fibers and depletion of Profilin1 abrogates induction of actin stress fibers in response to Wnt stimulation, this Daam1/Profilin1 complex is probably required for reorganization of the actin cytoskeleton during blastopore closure. The `purse string' mechanism of blastopore closure requires force generation executed by the actin cytoskeleton (Keller et al.,2003) and we propose this is mediated by the action of Daam1 and Profilin1. Indeed a recent study has revealed a role for Daam1 in Drosophila in actin polymerization during tracheal development(Matusek et al., 2006). The mechanism of this dynamic control over the actin cytoskeleton by Daam1 and Profilin1 will require a detailed investigation of the contribution of the individual domains within Daam1 to its function during gastrulation as well as the identity of other effector molecules such as the Rho-GEF, which triggers Rho activation in response to Wnt stimulation.

We propose a model of non-canonical Wnt signaling during gastrulation in which Dishevelled binds to Daam1 and Profilin1 is recruited to a Dishevelled/Daam1 complex (Fig. 9E). As Profilin1 is an actin polymerization factor, it can mediate cytoskeletal changes required for blastopore closure. In Xenopus gastrulation, Daam1 also leads to Rho and ROCK activation that independently results in modulation of the actin cytoskeleton(Fig. 9E). Wnt signaling through Dishevelled but not involving Daam1 also activates Rac, which is required independently for execution of the full array of gastrulation movements (Habas et al.,2003). Our results suggest that different aspects of gastrulation movements require different combinations of separate or overlapping signals that are generated as branches of the non-canonical Wnt pathway.

We thank Sunita Kramer, Michael Tsang and Michael Shen for discussion and critical comments, and members of the Dawid laboratory for stimulating discussions. We thank Drs Jeffrey Miller, Sergei Sokol and Maszumi Tada for reagents and we are grateful to Drs. William Wadsworth and Michael Matisse for use of their microscopes. This work is supported in part by intramural funds of the National Institute of Child Health and Human Development; by grants from the Uehara Memorial Foundation and the New Jersey Commission on Cancer Research to A.S.; and by the Foundation of UMDNJ, the American Heart Association, a Basil O'Connor Starter Scholar Award from the March of Dimes and an NSF grant (#0544061) to R.H.