WAVE3-mediated cell migration and lamellipodia formation are regulated downstream of phosphatidylinositol 3-kinase.

WAVE3 is a member of the WASP/WAVE family of protein effectors of actin reorganization and cell movement. The precise role of WAVE3 in cell migration and its regulation, however, have not been elucidated. Here we show that endogenous WAVE3 was found to be concentrated in the lamellipodia at the leading edge of migrating MDA-MB-231 cells. Platelet-derived growth factor (PDGF) treatment induced lamellipodia formation as well as two-dimensional migration of cells in the wound-closure assay and chemotactic migration toward PDGF in three-dimensional migration chambers. Knockdown of WAVE3 expression by RNA interference prevented the PDGF-induced lamellipodia formation and cell migration. Treatment of cells with LY294002, an inhibitor of phosphatidylinositol 3-kinase (PI3K), also abrogated the PDGF-induced lamellipodia formation and cell migration, suggesting that PI3K may be required for WAVE3 activity. WAVE3 and the PI3K regulatory subunit, p85, were found to interact in a yeast two-hybrid screen, which was confirmed through co-immunoprecipitation. The WAVE3-p85 interaction was mediated by the N-terminal region of WAVE3 and the C-terminal SH2 domain of p85. These results imply that the WAVE3-mediated migration in MDA-MB-231 cells via lamellipodia formation is activated downstream of PI3K and induced by PDGF. The findings of the WAVE3-p85 partnership also suggest a potential regulatory role for p85 in WAVE3-dependent actin-cytoskeleton reorganization and cell migration.

Remodeling of the actin cytoskeleton plays a critical role in altering both cellular morphology and motility (1) by controlling a range of cellular events, including immune defense, embryonic development, and neuronal outgrowth (2,3). In response to extracellular signals such as growth factors, the WASP family of proteins, which contains WASP and N-WASP (4), and the WAVE family of proteins, which contains WAVE1, -2, and -3 (5-7), activate the Arp2/3 complex, leading to the stimulation of actin polymerization and assembly (8,9). The WASP/WAVE families are protein effectors that act downstream of the Rho GTPases to regulate actin polymerization (7). Interactions between members of the WASP family and Cdc42 or phosphatidylinositol 4,5-biphosphate are required for their activation allowing exposure of the verprolin homology, cofilin homology, and acidic (VCA) 1 region, which can then bind to the Arp2/3 complex and initiate actin polymerization (10,11). In contrast, members of the WAVE family function downstream of Rac to induce actin polymerization (12,13). The downstream cellular effects of WASPs and WAVEs are thought to lead to the formation of actin-containing structures known as filopodia and lamellipodia, respectively (7).
WASP and WAVE proteins differ in both the signaling inputs that they receive, as well as in their mode of regulation. WASP proteins are autoinhibited by intramolecular interactions that mask the VCA domain in the absence of signaling molecules such as Cdc42 and phosphatidylinositol 4,5-biphosphate (14,15). On the other hand, WAVE1 and WAVE2 proteins are sequestered in an inactive state through the formation of a complex with four other proteins, PIR121, Nap125, HSPC300, and Abi1 (12,16,17). Extracellular stimuli result in the dissolution of this complex, which subsequently leads to Arp2/3mediated actin polymerization. Whether WAVE3 activity is regulated in the same manner or not, is still unknown. Several other proteins have been shown to regulate the activity of the WASP and WAVE proteins (7). Profilin, which has a high affinity to poly-L-proline, a proline-rich region (or proline-rich domain) that is conserved in all WASP and WAVE proteins (18), was shown to play a critical role in the enhancement of Cdc42-WASP-induced nucleation of actin polymerization (19). On the other hand, the WISH protein, an SH3containing protein, was found to activate N-WASP independently of either Cdc42 or phosphatidylinositol 4,5-biphosphate, by binding to the proline-rich region of N-WASP (20). Other proteins such as Syndapin, WIP, and Nck have been shown to physically interact with the WASP protein through different motifs to induce actin polymerization (21)(22)(23). Although the regulation of WASP proteins is well documented, the regulation of the WAVE proteins, and WAVE3 in particular, has not been well elucidated. The best characterized WAVE protein partner is IRSp53, which has been shown to be an essential intermediate between Rac and WAVE2 in the regulation of membrane ruffling (13). In addition to the involvement of WAVE3 in actin polymerization and cytoskeleton organization (9,24), we have shown that WAVE3 may also be associated with the development of low grade neuroblastoma (25,26).
Platelet-derived growth factor (PDGF) induces a variety of cellular responses, including proliferation, migration, invasion, and cell survival, in several cell types via its receptor (27,28). Although PDGF has clearly been shown to modulate WAVE1and WAVE2-mediated cell motility through activation of PI3K and the production of phosphatidylinositol 3,4,5-triphosphate, which in turn stimulates the formation of lamellipodia (19, 29 -31), the role of PDGF in WAVE3-mediated actin reorganization and cell motility has not been demonstrated. We used the human MDA-MB-231 adenocarcinoma cells to investigate whether PDGF is involved in WAVE3-mediated actin reorganization and cell motility. Here we show that PDGF treatment of MDA-MB-231 cells induced the formation of lamellipodia at the edges of migrating cells, and the accumulation of WAVE3 in these lamellipodia. WAVE3 has also been shown to act downstream of PI3K and is involved in cell migration and the formation of lamellipodia induced by PDGF. We also show that this WAVE3 activity might be regulated through interaction with p85, the regulatory subunit of PI3K.

Materials
The WAVE3 siRNA oligonucleotide, 5Ј-UAGUACUCUUGCCGC-UAUC-3Ј, and the control siRNA, 5Ј-CUUCACAGACCAAGACUGG-3Ј sense strand with symmetric 3Ј-overhangs of two deoxythymidines, were purchased from Dharmacon and annealed according to the manufacturer's instructions. The control siRNA was initially designed to target the ZNF198 gene transcript but failed to affect the expression levels of ZNF198 in all the cell lines we tested. SuperScript II Reverse Transcriptase Kits and Taq polymerase were obtained from Invitrogen. PCR primers were synthesized by Integrated DNA Technologies (Coralville, IA). The primer sequences used for glyceraldehyde-3-phosphate dehydrogenase were 5Ј-GAAGGGAAGGTCGGAGT-3Ј for the forward primer and 5Ј-GAAGATGGTGATGGGATTT-3Ј for the reverse primers, for p85, 5Ј-TACCCGCGCTCTGTACCCGT-3Ј for the forward primer and 5Ј-TCACCTCCTCCCTTGAAATG-3Ј for the reverse primer. Primers for WAVE1, WAVE2, and WAVE3 were as previously reported (6). The antibodies used in this study were as follows: human WAVE3/Scar and GST obtained from Upstate Biotechnology (Charlottesville, VA) and human PI3K p85 obtained from BD Biosciences (San Diego, CA). The secondary antibodies used were donkey anti-rabbit and goat antimouse from Jackson ImmunoResearch (West Grove, PA). Platelet-derived growth factor (PDGF) was obtained from Sigma, and the PI3K inhibitor, LY294002, was obtained from LC Laboratories (Woburn, MA). Gel electrophoresis reagents were from Bio-Rad (Hercules, CA).

Methods
Cell Culture-Human MDA-MB-231 adenocarcinoma cells (ATTC) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. For stimulation with PDGF, the cells were first subjected to serum starvation for 24 h in DMEM containing 0.5% fetal bovine serum and then treated with 50 ng/ml recombinant PDGF. The PI3K-specific inhibitor LY294002 was added to the medium at a concentration of 30 M together with PDGF.
Reverse Transcription-PCR-Cells were lysed in TRIzol reagent (Invitrogen), and total RNA was extracted according to the manufacturer's instructions. RNA was quantified using a Spectrophotometer (Beckman-Coulter Fullerton, CA), and 1 g of RNA was used to generate cDNA with the SuperScript II RT-PCR kit (Invitrogen). Reverse transcription was performed according to the manufacturer's instructions, and the cDNA generated was used as a template for PCR for analysis of gene expression using a PTC-100 Thermal Cycler from MJ Research (Waltham, MA).
Three-dimensional Migration Assay-Three-dimensional migration was assayed using collagen IV-coated inserts (8-m pore size, BD Falcon) in 6-well plates. Cells were serum-starved for 24 h, and 2 ml of DMEM containing 2 ϫ 10 5 cells was added to the upper chamber. The lower chamber contained 3 ml of DMEM. Where required, PDGF (50 ng/ml) was added to the lower chamber. The inserts were incubated overnight at 37°C, thereafter the nonmigrating cells on the upper surface of the membrane were removed with a cotton swab. The cells that were able to migrate through the pores onto the lower surface of the membrane were fixed and stained with Diff-Quick (American Scientific Products, McGaw Park, IL) and photographed using an inverted Leica DM IRB microscope fitted with a charge-coupled device camera (Leica Microsystems, Germany) and counted.
Wound-closure Assay-This assay was used to measure the ability of cells to close a gap that is introduced in monolayer cells. MDA-MB-231 cells, plated in 6-well plates, were 80 -90% confluent when they were wounded by dragging a plastic pipette tip across the cell surface. The phase contrast images of the wounds were recorded after 24 h, and the wound closure was evaluated by measuring the width of the remaining wound.
Immunofluorescence Microscopy-Cells were grown on glass coverslips and fixed in 4% paraformaldehyde for 20 min in phosphatebuffered saline (PBS) at room temperature and washed with PBS. The cells were then permeabilized in 0.2% Triton X-100 in PBS for 15 min, washed again with PBS, and incubated in the blocking solution containing 5% bovine serum albumin (Sigma) in PBS for 2 h at room temperature. Primary as well as secondary antibodies were diluted at the recommended concentration in 5% bovine serum albumin in PBS. Cells were incubated with the primary antibody for 1 h, washed with PBS, and then incubated with the secondary antibody for 1 h. Actin filaments (F-actin) were stained with rhodamine-conjugated phalloidin (Molecular Probes, Eugene, OR) in PBS. The coverslips were mounted on object slides using Vectashield mounting medium containing DAPI (4Ј,6-diamidino-2-phenylindole, Vector Laboratories, Burlingame, CA). Fluorescent images were captured using a Nikon TE2000-E inverted microscope.
siRNA Treatment-An siRNA oligonucleotide targeting the human WAVE3 mRNA at nucleotide position 93-111 from the start codon was designed. Cells were cultured in DMEM supplemented with 4.5 g/liter glucose, 10% fetal bovine serum, and 2 mM glutamine. For transient transfections, ϳ2 ϫ 10 5 cells were plated in 6-well plates in DMEM without antibiotics 24 h prior to transfection. Cells were transfected with 240 pmol of siRNA using Oligofectamine (Invitrogen) in OPTI-MEM1 medium (Invitrogen) as directed by the manufacturer. Approximately 4 -12 h after transfection, media was supplemented with DMEM containing 10% fetal bovine serum without antibiotics, and cells were incubated at 37°C for an additional 48 h.
Immunoprecipitation and Immunoblotting-For endogenous co-immunoprecipitation, cells were grown in 10-cm diameter Petri dishes and washed twice with ice-cold PBS. Immediately, ice-cold lysis buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and a mixture of proteases inhibitors) was added. The cells were rapidly scraped off the plates, and the crude lysates were transferred to prechilled Eppendorf tubes and centrifuged at 15,000 ϫ g for 20 min at 4°C. Cleared lysates were incubated with rabbit anti-WAVE3 antibody or rabbit anti-serum for 1 h, followed by incubation with protein G-Sepharose beads (Amersham Biosciences) for 45 min at 4°C on a rotary wheel. Beads were then washed twice with lysis buffer. Precipitates were resolved by SDS-PAGE and analyzed by immunoblotting. For immunoprecipitation using GST fusion proteins the cleared cell lysates were immediately mixed with 100 g of either GST, GST-p85, or GST-p85 truncated fusion proteins and the tubes were incubated on a rotary wheel overnight at 4°C. The next morning the protein mixtures were incubated with the anti-WAVE3 antibody, followed by incubation with protein G-Sepharose beads as described above. Beads were then washed twice with lysis buffer. Precipitates were resolved by SDS-PAGE and analyzed by immunoblotting. For immunoblotting cellular lysates containing equivalent amounts of total protein (50 g) were treated with SDS sample buffer and resolved on a 10% SDS-polyacrylamide gel, followed by transfer to nitrocellulose or Immobilon-P (PVDF) membranes using Bio-Rad gel and transfer apparatus (Hercules, CA). Membranes were incubated in 5% whole milk (or bovine serum albumin) for 1 h at room temperature, washed with PBS, followed by incubation with the primary antibody (as specified) overnight at 4°C. Membranes were then washed and incubated in the appropriate secondary antibody at room temperature for 1 h, and the signals were developed using a Western Lights Chemiluminescence Detection kit (PerkinElmer Life Sciences).
Plasmid Construction and Transfections-WAVE3 constructs were generated using the PCR from the template WAVE3 cDNA IMAGE clone 4838122 (ATTC). All 5Ј-and 3Ј-primers used in the cloning contained an EcoRI restriction site. In the 5Ј-primer for the full-length WAVE3 and the WAVE3 lacking the VCA domain (⌬VCA), the ATG sequence for the first methionine was replaced by a TTG. Amplified fragments were subcloned into the pCR2⅐1 vector using the T-A cloning kit (Invitrogen). The subcloned fragments were inserted into EcoRIdigested pEGFPC1 or pAS2-1 vectors (Clontech), in-frame with the 3Ј terminus of EGFP (GFP) or the GAL4 DNA binding domain (BD), respectively. All constructs were sequence-verified using a 3100 Genetic Analyzer (ABI Prism). The EGFP-recombinant vectors were used for either transient or stable transfections using standard protocols. The GST-p85SH3 (residues 39 -80), -p85NSH2 (residues 333-430), and -p85CSH2 (residues 624 -720) constructs were kindly provided by Dr. L. Rameh (Boston Biomedical Research Institute, Boston, MA).

RESULTS
The WASP and WAVE proteins have been shown to play essential roles in the regulation of actin reorganization and in the formation of filopodia and lamellipodia, respectively, leading to cell migration in several cell types (7,29,(31)(32)(33). WAVE1 and WAVE2 are required for PDGF-induced lamellipodia formation and cell migration (19,29,30). The involvement of WAVE3 in the PDGF-induced cell migration has, however, not been determined. We, therefore, addressed whether WAVE3 is also involved in PDGF-induced lamellipodia formation and cell migration in MDA-MB-231 cells.
WAVE3 Is Required for Cell Migration-To investigate whether WAVE3 plays a role in cell migration, we targeted the WAVE3 transcript in MDA-MB-231 cells using short interfering RNAs (siRNA). Untreated cells show high endogenous levels of WAVE3, and siRNA treatment resulted in Ͼ90% knockdown of the mRNA (Fig. 1A). The expression levels of the closely related WAVE1 and WAVE2 genes were not affected by the siRNA treatment, indicating that, within this family of genes, the siRNA we used is specific for WAVE3. Treatment with WAVE3 siRNA resulted in a comparable reduction in proteins levels (Fig. 1B). Because WAVE3 is predicted to be involved in actin remodeling, we next determined whether reduction in WAVE3 protein levels affect the formation of lamellipodia, which are necessary for cell movement and migration. ϳ30% of cells treated with Oligofectamine alone or with the control siRNA (si-control) formed lamellipodia after being induced with PDGF ( Fig. 1, C and D). The number of cells with PDGF-induced lamellipodia was, however, reduced to Ͻ10% after transfection with siRNA against WAVE3 (Fig. 1, C and  D). Because actin polymerization has also been shown to be involved in cell migration, we investigated the migration ability of MDA-MB-231 cells toward PDGF, after treatment with siRNA using the three-dimensional migration chamber assay. As shown in Fig. 1E, both untreated cells and the cells treated with the control siRNA are easily attracted toward PDGF and pass freely through the collagen-treated barrier, whereas siRNA-treated cells show a marked reduction in their ability to migrate toward PDGF (Fig. 1, E and F). We also determined the effect of WAVE3 knockdown on cell movement using the wound closure assay. Over a 24-h period the control MDA-MB-231 cells can typically completely close the space in the wound assay. However, cells treated with siWAVE3 could not (Fig.  1G), demonstrating a role of WAVE 3 in cell movement. Thus, these assays clearly demonstrate the WAVE 3 is required for cell movement and invasion.
PDGF Induces Lamellipodia Formation and Migration of MDA-MB-231 Cells-The role of PDGF in cell migration has already been determined in several cell types (3,31,34). To establish a link between PDGF and cell migration in our system, we investigated the effect of PDGF treatment on the ability of MDA-MB-231 cells to form lamellipodia and migrate toward PDGF. When MDA-MB-231 cells were serumstarved for 24 h, the majority of cells showed elongated and flattened shapes, and only ϳ10% of cells formed lamellipodia (Fig. 2, A and B). Treatment of these cells with 50 ng/ml PDGF for 30 min induced lamellipodia formation and a polarized shape in more than 30% of cells (Fig. 2, A and B). We next examined the effect of PDGF treatment on cell motility using the wound-closure assay. PDGF-stimulated cells migrated into the wounded area and were able to close the wound within 24 h, whereas, in the serum-starved cells, the wound remained open after 24 h (Fig. 2C). As an alternative measure of cell motility, we also examined the effect of PDGF on the ability of cells to migrate in a three-dimensional migration chamber assay. We observed an increase of ϳ6-fold in the number of cells that were able to migrate toward PDGF, compared with the control cells not treated with PDGF (Fig.  2, D and E). Thus, PDGF is also capable of stimulating the migration of MDA-MB-231 cells, in both nondirectional (wound-closure) and directional (three-dimensional migra-tion chamber) assays, probably by inducing lamellipodia formation at the leading edges.
WAVE3 Co-localizes with the Actin Structures in the Lamellipodia-Immunoblotting with a specific WAVE3 antibody showed that WAVE3 protein is present in both the serum-starved and PDGF-induced cells (Fig. 3A). WAVE3 is usually predominantly expressed in normal brain tissue (6,26,35). However, we have shown that WAVE3 is also expressed in several advanced stage neuroblastoma tumors and cell lines as well as in metastatic breast cancer cell lines, including MDA-MB-231 cells (26), which may account for the relationship between overexpression of WAVE3 and the metastatic phenotype.
We next investigated whether treatment of MDA-MB-231 cells with PDGF affects the intracellular distribution of the WAVE3 protein. Immunocytochemistry analyses showed that WAVE3 was largely concentrated in the nucleus and the cytoplasm of the untreated cells (Fig. 3B). Additionally, the unstimulated cells showed elongated and flattened shapes and numerous stable actin fibers (Fig. 3B). On the other hand, cells induced with PDGF were able to form lamellipodia at the leading edges of migrating cells (Fig. 3B), and, in addition to the diffuse distribution of WAVE3 in the cytoplasm and the nucleus, WAVE3 was also concentrated in the lamellipodia and co-localized with the actin structures in the lamellipodia (Fig. 3B).
Interaction between WAVE3 and p85 May Be Required for WAVE3-mediated Lamellipodia Formation and Cell Migration-Analysis of the WAVE1 and WAVE2 proteins demonstrates that their function is mediated through specific proteinprotein interactions (12, 13, 16, 17). To investigate whether the same is true for WAVE3 we used the yeast two-hybrid system In C the wound-closure assay (magnification, ϫ100) of serum-starved cells shows a vastly increased ability of the PDGF-treated cells to close the gap. In D a three-dimensional migration chamber assay shows that serum-starved cells have a greatly increased ability to pass through the chamber in the presence of PDGF (right), compared with untreated cells (left), which is ϳ6-fold greater (E). These values were derived from the analysis of cell numbers in ten microscopic fields in each experiment.
to identify potential binding partners for WAVE3. As a result, p85, the regulatory subunit of the PI3K, was identified as a potential WAVE3-interacting protein. This interaction between WAVE3 and p85 was confirmed by co-immunoprecipitation (IP) of WAVE3 and p85 from lysates of MDA-MB-231 cells. The presence of both the WAVE3 (Fig. 4A) and p85 (Fig. 4A) proteins in the lysate was confirmed by immunoblotting using anti-WAVE3 and anti-p85 antibodies. Using the anti-p85 antibody for IP and the anti-WAVE3 antibody for Western blot analysis, a 66-kDa band, consistent with the size of the WAVE3 protein, was detected (Fig. 4A). The rabbit pre-immune IgG alone was not able to precipitate the WAVE3-p85 complex, thus demonstrating the specificity of the co-IP.
To identify which particular domain of p85 mediates the WAVE3-p85 interaction, a series of truncated fragments of p85, fused to GST, were used in co-IPs from protein extracts of MDA-MB-231 cells. Immunoprecipitation of WAVE3 with the anti-WAVE3 antibody resulted in the co-IP of the C-terminal Src homology 2 domain of p85 (p85CSH2), but did not co-IP either the Src homology 3 (p85SH3), or the N-terminal Src homology 2 (p85-NSH2) domains, or GST alone (Fig. 4B). These results clearly indicate that p85 interacts directly with WAVE3 through the C-terminal SH2 domain.
A series of WAVE3-truncated fragments fused to the C terminus of either EGFP or the GAL4 DNA binding domain (GAL4BD) were generated to determine which particular domain of WAVE3 was involved in the interaction with p85. HEK-293 cells were transfected with either the GFP-tagged WAVE3 constructs or the pEGFPC1 vector alone, and clones were selected that showed stable expression of the fusion proteins. These clones were used for IP using an anti-p85 antibody, which resulted in the co-IP of both the full-length WAVE3 protein (W3F) and the WAVE3 construct lacking the VCA domain (W3⌬VCA), but not GFP (Fig. 4C). These results indicate that the VCA domain of WAVE3 is not required for the interaction with p85.
The WAVE3 domain that binds to p85 was further refined using the ␤-galactosidase assay with yeast clones co-transformed with the GAL4BD-WAVE3 constructs and the AD-p85 construct. As expected, those clones containing W3F or W3⌬VCA constructs generated ␤-galactosidase activity indicating an interaction with AD-p85 (Fig. 4D), further confirming the observations in the co-IP assays. On the other hand, yeast clones containing either the DNA BD domain alone, or other truncated forms of WAVE3, did not yield blue colonies (Fig.  4D), indicating that these WAVE3 domains are not directly responsible for the interaction with p85.
PI3K Activity May Be Required for the Regulation of WAVE3mediated Lamellipodia Formation and Cell Migration-The direct interaction between WAVE3 and the regulatory subunit  4. WAVE3 and the regulatory subunit of PI3K, p85,

physically interact, and the interaction is mediated through the basic domain of WAVE3 and the C-terminal SH2 domain of p85.
A, in vivo complex formation between WAVE3 and p85 in MDA-MB-231 cells. Cell extracts were immunoprecipitated with either an anti-p85 antibody or preimmune control IgG, and the immunocomplexes were resolved on 10% polyacrylamide gels. Western blot analysis demonstrates the presence of a 66-kDa protein, corresponding to WAVE3, in the initial cell lysates (top). Similarly, an 85-kDa protein, which corresponds to p85, is also present in the cell lysates (center). Using an anti-WAVE3 antibody (bottom), the 66-kDa WAVE3 band can be seen in the IP using anti-p85, but not in the IP using the preimmune IgG. The 54-kDa band corresponds to a background cross-reacting band. B, localization of the WAVE3-interacting domain in p85. MDA-MB-231 cell extracts were mixed with either protein extracts containing GSTtagged truncated forms of p85 or GST alone and were immunoprecipitated using the anti-WAVE3 antibody. Western blot analysis (top) using anti-GST antibodies identifies the various expected sized GST constructs. IP using the anti-WAVE3 antibody (bottom) demonstrates that only the 45-kDa p85CSH2 derivative protein interacts with full-length WAVE3. The nonspecific 54-kDa band is again seen in all lanes. C, localization of the p85-interacting domain in WAVE3. Protein extracts from HEK 293 cells with stable expression of either GFP alone, GFP-W⌬ VCA, or GFP-W3F were immunoprecipitated using the anti-p85 antibody. Western blot analysis using anti-GFP antibody identifies proteins corresponding to both the full-length WAVE3 and WAVE3 lacking the VCA domain, but not GFP alone (top). The corresponding GFP-tagged proteins were detected in the crude lysate in all three lanes (center), as was p85 using anti-p85 antibodies (bottom). D, ␤-galactosidase assays indicate that the WAVE3-p85 interaction is mediated by the N-terminal region of WAVE3. CG1945 yeast cells were co-transformed with p85-AD and either BD or WAVE3-BD fusion constructs. Only the clones carrying W3F and W3⌬ VCA generated blue colonies (dark color) indicative of ␤-galactosidase activity. The pCL1 vector was used as a positive control for ␤-galactosidase activity. of PI3K strongly supports the involvement of PI3K in the regulation of WAVE3-mediated cell migration and lamellipodia formation. To investigate whether PI3K is involved in PDGFinduced lamellipodia formation and cell migration in MDA-MB-231 cells, as it has been shown in a variety of cell types (1, 30, 36 -40), MDA-MB-231 cells were treated with the PI3K inhibitor LY294002. Only ϳ10% of the serum-starved cells were able to form lamellipodia (Fig. 5, A and B), whereas treatment of the cells with 50 ng/ml PDGF for 30 min resulted in the formation of lamellipodia in Ͼ30% of cells (Fig. 5, A and  B). Addition of LY294002 to cells at the concentration of 30 M, together with PDGF, suppressed the formation of lamellipodia in Ͼ90% of cells (Fig. 5, A and B). Similarly, PDGF treatment resulted is a ϳ6-fold increase in the number of cells able to migrate through the collagen-coated membrane (Fig. 5C), and treatment with LY294002 inhibited cell migration to levels comparable with the untreated cells (Fig. 5C). These results show that inhibition of PI3K activity interferes with the PDGFinduced lamellipodia formation and cell migration. Because we have also shown that WAVE3 is required for PDGF-mediated lamellipodia formation and cell migration (Fig. 1), these results also support an essential role for PI3K in WAVE3-mediated lamellipodia formation and cell migration by recruiting WAVE3 to the edges of polarized cells. Together, the results presented here demonstrate that WAVE3 is essential for PDGF-induced lamellipodia formation and migration of MDA-MB-231 cells, that PI3K activity is required for this WAVE3mediated process, and that a direct interaction between WAVE3 and p85 may be required for WAVE3 activity. DISCUSSION The ability of a cell to migrate is a critical component for both normal cellular processes as well as for tumorigenesis (1,41,42). WAVE3 is a member of the WASP/WAVE family of proteins, which mediate cellular motility through the Rac and Cdc42 pathways (12,13,17,43,44). WASP and N-WASP are activated downstream of Cdc42 and phosphatidylinositol 4,5biphosphate to induce filopodia formation (43,44), whereas activation of Rac leads the formation of lamellipodia and cell migration by the WAVE proteins (12,13,17). Although much is known about the pathways leading to the activation of WASP, N-WASP, WAVE1, and WAVE2, relatively little is known about the involvement of WAVE3 and its regulation in the processes of actin reorganization and cell migration. Furthermore, several studies have recently reported the involvement of PDGF in the N-WASP-and WAVE2-mediated lamellipodia formation and cell migration (19,29,30). Thus, we investigated whether WAVE3 has a role in the formation of lamellipodia and cell migration in MDA-MB-231 cells in response to PDGF.
As a first step toward understanding the involvement of WAVE3 in actin reorganization that leads to cell motility, we determined whether WAVE3 is associated with actin struc- tures in the leading edges of migrating MDA-MB-231 cells. WAVE3 protein was detected in MDA-MB-231 cells regardless of PDGF stimulation. Treatment of cells with PDGF, however, resulted an increase in the number of cells with lamellipodia at the edges of migrating cells, which is consistent with the previously published data on the effect of both PDGF as well as other growth factors on different cells types (3,31,34). PDGF treatment also resulted in the accumulation of WAVE3 at these cell edges where lamellipodia were formed, suggesting a role of WAVE3 in the formation of lamellipodia. Knockdown of WAVE3 expression not only abrogated the formation of lamellipodia, but also inhibited MDA-MB-231 migration, which implies that WAVE3 expression is essential for both the formation of lamellipodia as well as cell migration. Both WAVE1 and WAVE2 are expressed in MDA-MB-231 cells (not shown), and both WAVE1 and WAVE2 were found to be involved in cell motility and actin organization in response to PDGF (19,29,30,45). Changes in the ability of MDA-MB-231 cells to form lamellipodia and to migrate in response to PDGF are, however, specifically associated with WAVE3, because these changes were the result of the inhibition of WAVE3 expression, regardless of the presence of WAVE1 and WAVE2, whose their expression was not affected by the WAVE3 siRNA.
Our results show that PI3K is required for the formation of lamellipodia in MDA-MB-231 cells in response to PDGF, and for their migration toward PDGF, because inhibition of PI3K activity with LY294002, in PDGF-treated cells, resulted in the inhibition of both the formation of lamellipodia and cell migration. The inhibition of PI3K activity also disturbed the accumulation of WAVE3 at the edges of migrating cells, which resulted in the distribution of WAVE in the cell center and the nucleus (not shown). Similar studies have recently reported that the formation of lamellipodia and migration of PDGFtreated fibroblasts, as well as HGF-treated myogenic cells, also require the production of phosphatidylinositol 3,4,5-triphosphate by PI3K, leading to the recruitment of WAVE2 to the polarized plasma membrane (30,31). It is also possible that the production of phosphatidylinositol 3,4,5-triphosphate, as a result of PI3K activity, is essential for WAVE3-mediated filopodia formation and cell migration, as it has been shown for WAVE2 (30,31). These results support the previously reported data on the involvement of PI3K in the formation of lamellipodia and cell migration in other cell types in response to specific growth factors (3,19,29,31).
PI3K is a heterodimer consisting of a 110-kDa catalytic subunit and an 85-kDa regulatory subunit. PI3K activity depends on the production of phosphatidylinositol 3,4,5-triphosphate, a messenger molecule produced in response to a variety of growth factors. Upon activation, PI3K is targeted to specific locations within the cell where changes in the cytoskeleton organization take place (38). Although the downstream targets of PI3K are still not completely understood, it has been shown that the binding of the PI3K complex to activated plateletderived growth factor receptor, via the p85 SH2 domains, leads to the internalization of the complex into endocytic vesicles, where cellular trafficking is regulated (46). We have demonstrated a direct interaction between the N-terminal region of WAVE3 and the C-terminal SH2 domain of p85. SH2 domains are known to interact with specific motifs containing tyrosines only when they are phosphorylated. Indeed, WAVE proteins have been shown to become hyperphosphorylated in response to PDGF (47). Additionally, we found that treatment of MDA-MB-231 cells with PDGF resulted in WAVE3 phosphorylation at tyrosine residues using anti-phosphotyrosine antibody (not shown). Several tyrosine residues are present in the N-terminal region of WAVE3 that interacts with the p85-SH2 domain.
Sequence analysis of WAVE3 identified only a single potential target for tyrosine kinase phosphorylation at amino acid 151 (Tyr-151). Furthermore, this tyrosine residue is conserved in all three WAVE proteins, and a recent study has shown that, in WAVE2, it is the only tyrosine residue to undergo phosphorylation by abelson tyrosine kinase (48), further supporting our findings WAVE3 Tyr-151 is involved in mediating the interaction between WAVE3 and p85.
We suggest two possible functional consequences of the direct interaction between p85 and WAVE3 that is described here. First, the interaction may positively regulate actin cytoskeletal remodeling by providing a necessary link between Rac and WAVE3, with the resulting activation of the Arp2/3 complex and the reorganization of the cytoskeleton. The product of PI3K activity, phosphatidylinositol 3,4,5-triphosphate, was shown to be required for the regulation of Rac activity (49). The activation of Rac is also mediated by a direct p85-Rac interaction (50,51). We have found that p85 binds to WAVE3 through the C-terminal SH2 domain. This association is direct, as indicated by the one-to-one interaction in the yeast twohybrid assay, and occurs both in vivo and in vitro, as suggested by the co-IP assays. WAVE3 and p85 also co-localize to sites of active actin polymerization in the cell membrane. These results suggest that p85 may be an essential intermediate between Rac and WAVE3 in the regulation of the actin-cytoskeleton, much in the same way IRSp53 funnels signals from Rac to WAVE1 and WAVE2 proteins (13). Alternatively, the interaction between WAVE3 and p85 may play a role in negatively regulating the activity of WAVE3. A direct interaction between p85 and Abi1 has recently been suggested to be necessary for Abi1-dependent Rac activation, which leads to actin reorganization (52). Interestingly, Abi1, a component in a multiprotein complex, was shown to negatively regulate WAVE1-and WAVE2-mediated actin polymerization and reorganization (12,16,17). The demonstration that WAVE3 and p85 interact also suggests that p85, in association with an inhibitory multiprotein complex that may include Abi1 as well as other proteins, may regulate the activity of WAVE3 by sequestering it in an inactive form until appropriate extracellular stimuli are received. It is possible, however, that both of these possibilities are part of the same regulatory pathway.