Migfilin Interacts with Vasodilator-stimulated Phosphoprotein (VASP) and Regulates VASP Localization to Cell-Matrix Adhesions and Migration*

Cell migration is a complex process that is coordinately regulated by cell-matrix adhesion and actin cytoskeleton. We report here that migfilin, a recently identified component of cell-matrix adhesions, is a biphasic regulator of cell migration. Loss of migfilin impairs cell migration. Surprisingly, overexpression of migfilin also reduces cell migration. Molecularly, we have identified vasodilator-stimulated phosphoprotein (VASP) as a new migfilin-binding protein. The interaction is mediated by the VASP EVH1 domain and a single L104PPPPP site located within the migfilin proline-rich domain. Migfilin and VASP form a complex in both suspended and adhered cells, and in the latter, they co-localize in cell-matrix adhesions. Functionally, migfilin facilitates VASP localization to cell-matrix adhesions. Using two different approaches (VASP-binding defective migfilin mutants and small interfering RNA-mediated VASP knockdown), we show that the interaction with VASP is crucially involved in migfilin-mediated regulation of cell migration. Our results identify migfilin as an important regulator of cell migration and provide new information on the mechanism by which migfilin regulates this process.

Cell migration is a tightly controlled process that is crucially involved in embryonic development, wound repair, and other physiological processes (1)(2)(3)(4). Abnormal cell migration is an important causal factor for the pathogenesis and/or progression of a wide variety of diseases. Understanding how cells control their motility therefore is an important topic in molecular biology.
Migfilin is a recently identified widely expressed focal adhesion protein that provides a link between cell-matrix adhesions and the actin cytoskeleton (5). Migfilin consists of three structurally distinct regions. The C-terminal region of migfilin is composed of three LIM domains, which mediates the interaction with Mig-2, a focal adhesion protein that is critically involved in cell-matrix adhesion and shape modulation (5,6). The N-terminal region of migfilin interacts with filamin (5), an actin-binding protein whose deficiency or mutations cause defects in cell migration (7,8). Between the N-terminal and the C-terminal regions lies a proline-rich domain. Unlike the N-and C-terminal domains, however, neither the binding partners nor the functions of the migfilin proline-rich domain were known. Interestingly, some human cells express not only migfilin but also a splicing variant (termed as migfilin(s)) lacking the prolinerich domain (5).
In this study, we have sought to identify proteins that interact with the proline-rich domain of migfilin and investigated the roles of migfilin in regulation of cell migration. Our results show that vasodilator-stimulated phosphoprotein (VASP), 2 an actin cytoskeletal regulatory protein (reviewed in Refs. 9 -11), interacts with the proline-rich domain of migfilin. Depletion of migfilin diminished VASP localization to cell-matrix adhesions and impairs cell migration. Surprisingly, overexpression of migfilin also reduces cell migration. Using two different approaches (overexpression of VASP-binding defective migfilin mutants and siRNAmediated VASP knockdown), we show that the interaction with VASP is crucially involved in migfilin regulation of cell migration.
Yeast Two-hybrid Assays-A cDNA fragment encoding human migfilin residues 1-189 was inserted into the pGBKT7 vector (Clontech). The construct was used as bait to screen a human keratinocyte MATCHMAKER cDNA library following the manufacturer's protocol (Clontech). Sixteen positive clones were obtained from the library screening. Five of them were sequenced, among which three were found to encode VASP. To identify the VASP domain that mediates migfilin binding, cDNA fragments encoding human VASP sequences were inserted into the pGADT7 vector. The cDNA fragments encoding migfilin sequences were inserted into the pGBKT7 vector. The pGADT7 and pGBKT7 vectors containing VASP or migfilin sequences were introduced into yeast cells (Saccharomyces cerevisiae strain AH109) and the interaction was analyzed following the manufacturer's protocol (Clontech).
DNA Constructs, Transfection, and Immunoprecipitation-DNA vectors encoding FLAG-migfilin or FLAG-migfilin(s) were described (5,12). Deletion or substitution mutations (as specified in each experiment) were introduced into the migfilin coding sequence by PCR. To generate the vector encoding the GFP-tagged migfilin proline-rich domain, a cDNA fragment encoding migfilin residues 84 -180 was ligated into the pEGFP-C2 vector (Clontech). DNA vectors encoding wild type or mutant forms of migfilin were generated by ligating the wild type or mutant forms of migfilin cDNAs (including the stop codon) into the p3xFLAG-CMV-14 vector (Sigma). The migfilin coding sequences were confirmed by DNA sequencing. Cells were transfected with the vectors encoding various forms of migfilin using Lipofectamine reagents (Invitrogen). For immunoprecipitation * This work was supported by National Institutes of Health Grants GM65188 and DK54639 (to C. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  analyses, the transfectants were lysed with the lysis buffer (1% Triton X-100 in 50 mM Tris-HCl, pH 7.4, containing 150 mM NaCl, 2 mM Na 3 VO 4 , 100 mM NaF, and protease inhibitors). The cell lysates were mixed with agarose beads conjugated with anti-FLAG mAb M2 (Sigma) or a rabbit anti-GFP Ab (Santa Cruz Biotechnology) and UltraLink Immobilized Protein G (Pierce). The beads were washed four times and the immunoprecipitates were analyzed by Western blotting with Abs as specified.
Co-precipitation of Endogenous Migfilin and VASP-WI-38 cells were grown in DMEM containing 10% FBS in culture plates for 1 day. The cell monolayers (adhered) were lysed with the lysis buffer. In parallel, WI-38 cells were harvested by trypsinization and washed with DMEM containing 10% FBS. The cells were re-suspended in DMEM containing 10% FBS, and maintained in suspension on a rocking platform for 45 min (suspension). The cells were then washed with phosphate-buffered saline and lysed with the lysis buffer. Migfilin was precipitated from the lysates with GST-FLN21 fusion protein (containing filamin A residues 2282-2395) as described (5).
siRNA and Transfection-The migfilin siRNA and control RNA were described (5). A second migfilin Stealth TM siRNA (target sequence: 5Ј-gaagaggguggcaucgucugucuuu-3Ј) was generated by Invitrogen. The migfilin Stealth siRNA was used in some experiments as specified. Two different VASP siRNAs were used. The first one (termed VASP siRNA1 herein) was purchased from Santa Cruz Biotechnology (catalog number: sc-29516) and the second one (termed VASP siRNA2 herein, target sequence: 5Ј-aaagaggaaatcattgaagcc-3Ј) was purchased from Invitrogen. Cells were transfected with the siRNA or the control RNA using Oligofectamine or Lipofectamine 2000 (Invitrogen) following the manufacturer's protocols. The cells were analyzed 2 days after transfection. In double DNA/RNA transfection experiments, MDA-MB-231 cells were transfected with the DNA vector encoding migfilin or a control vector using Lipofectamine Plus (Invitrogen). One day after DNA transfection, the cells were transfected with VASP siRNA or a control RNA using Lipofectamine 2000. The double transfectants were analyzed 2 days after the RNA transfection.
Immunofluorescent Staining-Immunofluorescent staining was performed as described (13). Briefly, cells were plated on fibronectincoated coverslips, fixed, and dually stained with primary mouse mAbs and rabbit polyclonal Abs as specified. The primary mouse and rabbit Abs were detected with secondary fluorescein isothiocyanate-conjugated anti-mouse IgG and Rhodamine Red TX -conjugated anti-rabbit IgG Abs, respectively.
Preparation of Triton X-100-soluble and -insoluble Fractions-Cells were plated on collagen I-coated 60-mm plates and incubated at 37°C under a 5% CO 2 , 95% air atmosphere for 2 h to allow formation of cell-matrix adhesions. The cells were rinsed once with a PIPES buffer (100 mM PIPES, pH 6.9, 0.1 mM EDTA, 0.5 mM MgCl 2 , 4 M glycerol) containing protease inhibitors, and then extracted with the same buffer supplemented with 0.75% Triton X-100. The cell extracts were collected, vortexed, and centrifuged at 20,800 ϫ g at 4°C for 5 min. Protein concentrations in the Triton X-100-soluble (cytosol) fractions were determined using a BCA protein assay (Pierce). The pellets (the Triton X-100-insoluble fractions) were extracted with 1% SDS in phosphatebuffered saline. The Triton X-100-soluble and -insoluble fractions were mixed with SDS-PAGE sample buffer and analyzed by Western blotting with anti-migfilin and anti-VASP Abs.
Cell Migration-The cells (HeLa, HT-1080, or MDA-MB-231) were transfected with the siRNA duplexes or DNA vectors as specified in each experiment. Two days after the transfection, cell migration was analyzed using Transwell motility chambers as described (14). Briefly, the undersurfaces of the 8-mm pore diameter Transwell motility chambers (Costar) were coated with 20 g/ml fibronectin. The cells were suspended in 0.1 ml of DMEM containing 5 mg/ml bovine serum albumin and added to the upper chambers (2 ϫ 10 4 cells/chamber). After incubation at 37°C for the indicated period of time, the cells on the upper surface of the membrane were removed. The membranes were fixed and the cells on the undersurface were stained with Gills III hematoxylin. The cells from five randomly selected microscopic fields were counted.
Madin-Darby canine kidney (MDCK) cell migration was assessed by the ability of the cells to migrate into a cell-free area as described (15). Briefly, the cells were plated in DMEM containing 10% FBS in plates that were pre-coated with 30 g/ml collagen I. The cell monolayers were wounded by scratching with a plastic pipette tip. After washing, the cells were incubated in DMEM containing 1% FBS and 0.5 g/ml mitomycin C for the indicated periods of time. Images of three different segments of the cell-free area were recorded, and the distances traveled by the cells at the front in three different segments of the wound were measured.
For statistical analyses, paired t test was used to compare two groups of data. For comparison among three or more groups of data, one-way analysis of variance followed by Tukey's post-test were used. All values are presented as the mean Ϯ S.D. from n experiments as indicated. p values Ͻ0.05 were considered statistically significant.
Cell Adhesion-MDCK cells were transfected with a control vector lacking the migfilin sequence or vectors encoding different forms of migfilin as specified in the experiments. Two days after the transfection, the cells (5 ϫ 10 4 /well) were seeded in quadruplicate in 96-well plates that were pre-coated with 30 g/ml collagen I. After incubation at 37°C for 40 min, the wells were washed three times with phosphate-buffered saline. The cells were quantified using crystal violet as described (16). The percentages of the adhered cells were presented as the absorbance at 570 mm from the adhered cells divided by the absorbance at 570 mm from the total seeded cells. The data from two experiments were analyzed by one-way analysis of variance followed by Tukey's post-test. p values Ͻ0.05 were considered statistically significant.

Identification of VASP as a Binding
Protein of Migfilin-We used two approaches to identify proteins that interact with the migfilin prolinerich domain. One is a structure-based "candidate" approach. VASP was a strong candidate because of the presence of multiple LPPPP motifs, potential docking sites for VASP (9,10), in migfilin proline-rich domain. To test whether VASP interacts with migfilin, we transfected HeLa cells with a vector encoding FLAG-migfilin and a vector lacking migfilin sequence as a control, respectively. Expression of FLAG-migfilin in the FLAG-migfilin transfectants (Fig. 1A, lane 2) but not the control transfectants (Fig. 1A, lane 1) was confirmed by Western blotting. FLAGmigfilin was immunoprecipitated with anti-FLAG mAb M2 (Fig. 1A, lane 4) and the immunoprecipitates were probed by Western blotting with an anti-VASP Ab. VASP was readily detected in the FLAG-migfilin immunoprecipitates (Fig. 1B, lane 4). In control experiments, no VASP (Fig. 1B, lane 3) was detected in control precipitates lacking FLAGmigfilin (Fig. 1A, lane 3). In the second approach, we screened a human yeast two-hybrid cDNA library (containing 2.5 ϫ 10 6 independent clones) with a migfilin fragment containing the proline-rich domain as bait. Five positive clones were sequenced, among which three encoded VASP. To further analyze this, we performed yeast two-hybrid binding assays using vectors encoding different domains of VASP and migfilin. The results showed that the VASP EVH1 domain is both necessary and sufficient for interacting with migfilin ( Fig. 1C). On the other hand, no interaction was detected between VASP and the C-terminal LIM region of migfilin (Fig. 1C). These results identify VASP as a migfilin-binding protein and map the migfilin-binding site to the EVH1 domain of VASP.

Endogenous Migfilin and VASP Form a Complex in Cells-We next tested whether endogenous migfilin and VASP form a complex in cells.
To do this, we precipitated migfilin from human cells with a GST fusion protein containing filamin repeat 21 (GST-FLN21), which recognizes the N-terminal region of migfilin (5). As expected, migfilin ( Fig  formation of cell-matrix adhesions. VASP is phosphorylated by protein kinase A on its first phosphorylation site upon cell detachment (17), which causes a characteristic shift of VASP from 46 to approximately 50 kDa on SDS-PAGE gels. Consistent with the finding that the formation of the migfilin-VASP complex occurs prior to the formation of cellmatrix adhesions, the 50-kDa phosphor-VASP also formed a complex with migfilin (Fig. 2B, lane 4).
Mapping the VASP-binding Site to a Single LPPPP Site within the Proline-rich Domain of Migfilin-To facilitate structure-function analyses, we sought to define the VASP-binding site on migfilin. Toward this end, we first tested whether the migfilin proline-rich domain is necessary for interaction with VASP. To do this, we transfected HeLa cells with vectors encoding FLAG-migfilin(s), which lacks the proline-rich domain (5), and FLAG-migfilin, respectively. The expression of FLAGmigfilin (Fig. 3A, lane 2) and FLAG-migfilin(s) (Fig. 3A, lane 3) was confirmed by Western blotting. FLAG-migfilin (Fig. 3A, lane 5) and FLAG-migfilin(s) (Fig. 3A, lane 6) were immunoprecipitated from the corresponding transfectants but not the control transfectants (Fig. 3A, lane 4) with an anti-FLAG mAb. As expected, VASP was readily coimmunoprecipitated with FLAG-migfilin (Fig. 3B, lane 5). By marked contrast, no VASP was co-immunoprecipitated with FLAG-migfilin(s) lacking the proline-rich domain (Fig. 3B, lane 6) or with the anti-FLAG mAb in the absence of migfilin (Fig. 3B, lane 3). Thus, the proline-rich domain of migfilin is required for the interaction with VASP. Although migfilin contains several polyproline motifs that can potentially serve as binding sites for profilin, it was not detected in the migfilin immuno- precipitates (Fig. 3C, lane 5), suggesting that the proline-rich domain of migfilin is preferentially recognized by VASP.
The migfilin proline-rich domain contains multiple polyproline sites including two LPPPP sites (starting from Leu residues at positions 104 and 166, respectively). To map the VASP-binding site, we introduced deletion mutations (⌬84 -112 and ⌬142-170) into the migfilin prolinerich domain (Fig. 3F). Cells were transfected with vectors encoding FLAG-tagged ⌬84 -112 or ⌬142-170 and VASP binding was analyzed by co-immunoprecipitation as described in Fig. 3, A and B. Deletion of residues 84 -112, but not that of residues 142-170, eliminated VASP binding (Fig. 3F). To define the VASP-binding site, we introduced Ala   C, E, and G) and a rabbit anti- VASP Ab (B, D, F, and H). Bar in D, 20 m. substitution mutations into the Leu 104 polyproline site. Cells were transfected with vectors encoding the FLAG-tagged Leu 104 polyproline substitution mutant or FLAG-migfilin (as a positive control), or a vector lacking the migfilin sequence as a negative control. Expression of the FLAG-Leu 104 polyproline mutant (Fig. 3G, lane 3) and FLAG-migfilin (Fig. 3G, lane 2) were confirmed by Western blotting. Co-immunoprecipitation assay showed that the Leu 104 polyproline mutant (Fig. 3G,  lane 6), unlike the wild type migfilin (Fig. 3G, lane 5), failed to bind to VASP (Fig. 3H, compare lanes 5 and 6). These results demonstrate that VASP binding is mediated by a single LPPPPP site located at positions 104 -109 within the proline-rich domain.

Migfilin Facilitates VASP Localization to Cell-Matrix Adhesions-
Consistent with the interaction between migfilin and VASP, migfilin and VASP proteins co-localized in cell-matrix adhesions (Fig. 4, compare A and B). To test whether migfilin plays a role in VASP localization to cell-matrix adhesions, we suppressed migfilin expression by transfection with the migfilin siRNA (5). Immunofluorescent staining with an anti-migfilin mAb revealed that the level of migfilin at cell-matrix adhesions was substantially reduced in the migfilin siRNA transfectants compared with that in the control transfectants (compare Fig. 4, C with A). As expected, clusters of VASP were readily detected in focal adhesions in control cells (Fig. 4, A and B). To analyze the effect of migfilin knockdown on VASP localization, we randomly selected 50 migfilin knockdown cells and found that all of them exhibited diminished focal adhesion localization of VASP (a representative migfilin knockdown cell is shown in Fig. 4, C and D). To confirm this, we transfected HeLa cells with a second migfilin siRNA (the migfilin Stealth siRNA, Invitrogen). Transfection of the cells with the migfilin Stealth siRNA effectively knocked down migfilin and diminished the localization of VASP to cellmatrix adhesions (data not shown). Collectively, these results suggest an important role of migfilin in facilitating VASP localization to cell-matrix adhesions. In contrast to VASP, we found that focal adhesion kinase was clustered at cell-matrix adhesions in both the control (Fig. 4, E and F) and migfilin knockdown cells (Fig. 4, G and H).
To further analyze this, we dually stained the migfilin knockdown cells and control cells with mouse anti-vinculin and rabbit anti-VASP Abs. Clusters of vinculin and VASP were readily detected in control cells (Fig. 4, I and J). As expected, VASP was largely diffusely distributed in migfilin knockdown cells. We randomly selected 30 migfilin knockdown cells. Clusters of vinculin were detected at cell-matrix adhesions in all these cells, despite the diminished level of VASP in these adhesion structures (a representative cell is shown in Fig. 4, K and L). The vinculin clusters, however, appeared somewhat smaller in migfilin knockdown cells compared to those in the control cells (Fig. 4, compare I and K).
Protein clusters at cell-matrix adhesions and cytoskeleton are often resistant to nonionic detergent (e.g. Triton X-100) extraction (12). To further analyze the effect of migfilin on VASP distribution, we extracted the migfilin siRNA transfectants and the control transfectants with a buffer containing 0.75% Triton X-100. As expected, transfection of the cells with the migfilin siRNA reduced the level of migfilin in the Triton X-100-soluble (cytosol) as well as the Triton X-100-insoluble (cytoskeletal) fractions (Fig. 4M). VASP was detected in both the Triton X-100soluble (Fig. 4N, lane 3) and Triton X-100-insoluble (Fig. 4N, lane 1) fractions. Knockdown of migfilin reduced the level of VASP in the Triton X-100-insoluble fraction (Fig. 4N, compare lane 1 with lane 2) and concomitantly increased the level of VASP in the Triton X-100-soluble fraction (Fig. 4N, compare lane 3 with lane 4). Re-probing the same membrane with an anti-vinculin Ab showed that the level of vinculin in these fractions was not significantly altered (Fig. 4O). The biochemical extraction results, together with those of the immunofluorescent staining, suggest that depletion of migfilin selectively attenuates the localization and clustering of VASP at cell-matrix adhesions.
Migfilin Is Not Required for VASP Localization to Lamellipodia-We next tested whether migfilin is required for VASP localization to lamellipodia. Human HT-1080 fibrosarcoma cells were used in these studies as a high percentage of them form extensive VASP-positive lamellipodia that were often present in a polarized fashion (Fig. 5, B and F). Among 53 cells that were analyzed, 34 (64%) of them exhibited strong VASP-positive lamellipodia (two representative images are shown in Fig. 5, B and  F). Despite the presence of VASP in lamellipodia, migfilin was not detected in this structure. Instead, it was highly concentrated in cellmatrix adhesions (Fig. 5, A and E), where VASP clusters were also detected (Fig. 5, B and F). Transfection of HT-1080 cells with migfilin siRNA substantially reduced the level of migfilin (Fig. 5, C and G). Immunofluorescent staining of migfilin knockdown cells with the mouse anti-migfilin mAb and rabbit anti-VASP Ab showed that, as expected, depletion of migfilin reduced the clustering of VASP at cellmatrix adhesions (Fig. 5, D and H). We analyzed 36 migfilin knockdown cells and detected VASP-positive lamellipodia in 25 (69%) of them. However, they often appeared shorter or un-polarized (Fig. 5, D and H). These results suggest that migfilin is not absolutely required for VASP localization to lamellipodia, albeit it may influence the length and polarity of lamellipodium.
Migfilin Is a Biphasic Regulator of Cell Migration-The role of migfilin in regulation of cell migration had not been previously investigated. To test whether migfilin is required for cell migration, we transfected HeLa cells with the migfilin siRNA and the control RNA, respectively. As expected, the level of migfilin in the migfilin siRNA transfectants were substantially reduced (Fig. 6A). The levels of other cellular proteins, including actin (Fig. 6B) and other focal adhesion proteins such as Mig-2, filamin, VASP, paxillin, focal adhesion kinase, ILK, PINCH, and ␣-parvin (not shown) were not altered, confirming the specificity of the migfilin siRNA. We next compared the migration of the migfilin-deficient cells with that of the control cells. The results showed that depletion of migfilin significantly reduced cell migration (Fig. 6, C-E). To further test this, we transfected HT-1080 fibrosarcoma cells and MDA-MB-231 breast carcinoma cells with the migfilin siRNA and the control RNA, respectively. Transfection with the migfilin siRNA effectively reduced the level of migfilin in HT-1080 cells (Fig. 6F) and MDA-MB-231 cells (Fig. 6K). Equal loading was confirmed by probing the same samples with an anti-actin Ab (Fig. 6, G and L). Next, we compared migration of the migfilin knockdown cells with that of the control cells. In both HT-1080 (Fig. 6, H-J) and MDA-MB-231 (Fig. 6, M-O) cells, depletion of migfilin impaired cell migration. Knockdown of migfilin with the migfilin Stealth siRNA also impaired cell migration (not shown). Taken together, these results suggest that migfilin is required for proper cell migration.
Next, we sought to test the effect of up-regulation of migfilin on cell migration. To do this, we overexpressed migfilin in HeLa cells (Fig. 7A,  compare lane 1 with 3). Surprisingly, overexpression of migfilin in HeLa cells reduced cell migration (Fig. 7, B, C, and E). To further test this, we overexpressed migfilin in MDCK epithelial cells (Fig. 8A, lane 1). Again, overexpression of migfilin reduced MDCK cell migration (Fig. 8, B-D). Similarly, overexpression of migfilin in MDA-MB-231 cells reduced cell migration (see below). Thus, the effect of migfilin on cell migration is biphasic. On the one hand, migfilin is required for supporting cell migration and therefore loss of it impairs cell migration. On the other hand, expression of an excessive amount of migfilin also suppresses cell migration.
The VASP Binding Is Crucial for Migfilin-mediated Suppression of Cell Migration-To test whether the VASP binding is involved in migfilin-mediated suppression of cell migration, we overexpressed migfilin(s), which lacks the VASP-binding proline-rich domain in HeLa cells (Fig. 7A, lane 2). Overexpression of migfilin(s), unlike that of migfilin, did not significantly reduce cell migration (Fig. 7, B-E). To further test this, we overexpressed migfilin(s) in MDCK cells (Fig. 8A, lane 3). Consistent with the results in HeLa cells (Fig. 7), overexpression of the VASP-binding defective migfilin(s) in MDCK cells did not significantly reduce cell migration (Fig. 8, B-E). These results raised an interesting possibility that VASP binding is required for migfilin-mediated suppression of cell migration. To test this, we overexpressed the VASP-binding defective Leu 104 polyproline mutant, and the wild type migfilin as a control, in MDCK cells. The expression of migfilin ( Fig. 9, lane 2) and the VASP-binding defective mutant (Fig. 9, lane 3) was confirmed by Western blotting. Overexpression of the VASP-binding defective mutant, unlike that of the wild type migfilin, did not significantly reduce MDCK cell migration (Fig. 9B), suggesting that the interaction with VASP is required for migfilin-mediated suppression of cell migration. To further analyze this, we tested the effect of overexpression of migfilin on cell-matrix adhesion. The results showed that overexpression of migfilin substantially increased cell-matrix adhesion (Fig. 9C). By contrast, overexpression of the VASP-binding defective Leu 104 polyproline mutant or migfilin(s) failed to enhance cell-matrix adhesion (Fig. 9C).
The forgoing mutational experiments provide strong evidence for a role of the VASP binding in migfilin-mediated suppression of cell migration. If VASP is indeed involved in this process, depletion of VASP should abolish or significantly weaken the migfilin-mediated suppression of cell migration. Because siRNA that targets canine VASP was unavailable, we used human MDA-MB-231 cells to test this. Consistent with the results obtained with MDCK cells, overexpression of migfilin, but not that of the VASP-binding defective migfilin mutant, significantly reduced MDA-MB-231 cell migration (Fig. 10, A and B). To test whether VASP is involved in migfilin-mediated suppression of cell  migration, we dually transfected MDA-MB-231 cells with VASP siRNA1 and the migfilin expression vector or VASP siRNA1 and the control vector lacking the migfilin sequence. In control experiments, we dually transfected MDA-MB-231 cells with the control RNA and the migfilin expression vector, or the control RNA and the control DNA vector lacking migfilin sequence. Western blotting showed that the level of VASP was reduced in both VASP siRNA1 transfectants (Fig. 10C,  compare lanes 1 and 3 with lanes 2 and 4), irrespective of the level of migfilin. On the other hand, migfilin was overexpressed in both migfilin transfectants (Fig. 10C, compare lanes 3 and 4 with lanes 1 and 2). Knockdown of VASP in MDA-MB-231 cells reduced cell migration (Fig. 10E). Importantly, overexpression of migfilin in the VASP knockdown cells (Fig. 10F), unlike that in cells expressing a normal level of VASP (Fig. 10G), did not significantly reduce cell migration. Using a similar approach, we confirmed that knockdown of migfilin with a second VASP siRNA (VSAP siRNA2) also abolished migfilin-mediated suppression of cell migration (Fig. 11). Collectively, these results suggest that VASP is crucially involved in migfilin-mediated regulation of cell migration.

DISCUSSION
Cell migration is a complex process that is coordinately regulated by cell-matrix adhesion and actin cytoskeleton (1)(2)(3)(4)18). The studies presented in this paper have demonstrated that migfilin is a key component of the cellular machinery that controls cell migration. Using several different types of cells (HeLa, HT-1080, and MDA-MB-231 cells), we have shown that depletion of migfilin impairs cell migration. How does migfilin contribute to cell migration? Migfilin contains no intrinsic catalytic activities but possesses multiple protein-binding domains (19). Through different protein-binding domains migfilin interacts with multiple components of cell-extracellular matrix adhesions and the actin cytoskeleton, including Mig-2 (5), filamin (5,20), and as reported  3 and 4). One day after DNA transfection, the cells were transfected with VASP siRNA1 (lanes 1 and 3) or the control RNA (lanes 2 and 4). The cell lysates were analyzed by Western blotting (20 g of protein/lane) with Abs that recognize VASP (C) or migfilin (D). Two days after siRNA transfection, cell migration was analyzed as described in the legend to Fig. 6, M-O. E, the migration of the VASP knockdown cells expressing a normal level of migfilin was compared with that of the control cells (normalized to 100%). F, the migration of the VASP knockdown cells overexpressing migfilin was compared with that of the VASP knockdown cells expressing a normal level of migfilin (normalized to 100%). G, the migration of the migfilin-overexpressing cells that express a normal level of VASP was compared with that of the cells expressing normal levels of migfilin and VASP (normalized to 100%). Bars represent mean Ϯ S.D. (n ϭ 3). **, p Ͻ 0.05 versus control.
in this paper, VASP (Figs. 1 and 2). We have previously shown that migfilin is involved in linking the cell-matrix adhesions to the actin cytoskeleton (5,21). Thus, the migratory defect induced by the loss of migfilin is probably caused, at least in part, by the impaired connection between cell-matrix adhesions and the actin cytoskeleton. Additionally, migfilin could contribute to other steps that are pertinent to cell migration. For example, lamellipodia in migfilin knockdown cells appeared shorter or un-polarized (Fig. 5). Thus, migfilin may contribute to cell migration by influencing the formation or polarization of lamellipodia. It has been well documented that VASP regulates actin polymerization in lamellipodia (10,(22)(23)(24)(25). Given the interaction of migfilin with VASP, a simple model would be that migfilin influences lamellipodium formation via VASP. However, two lines of evidence suggest that the interaction of migfilin with VASP probably is not directly involved in the organization of lamellipodium. First, migfilin co-localizes with VASP in cell-matrix adhesions but not in lamellipodia. Thus, it is unlikely that migfilin directly regulates VASP activity in lamellipodia. Second, despite its effect on the length and polarization of lamellipodia, loss of migfilin does not prevent VASP localization to lamellipodia. Instead, loss of migfilin compromised the localization of VASP to cell-matrix adhesions, suggesting that migfilin, together with other focal adhesion proteins such as zyxin (26 -28), contributes to the localization of VASP to cell-extracellular matrix adhesions. The fact that migfilin is concentrated at the cell-extracellular matrix adhesions suggests that migfilin likely influences lamellipodium formation and polarization indirectly, possibly through its effect on the structure or signaling of the cell-matrix adhesion sites.
In addition to showing that migfilin is required for proper cell migration, we have demonstrated that an excessive amount of migfilin suppresses cell migration. Thus, migfilin regulates cell migration in a biphasic fashion. How does overexpression of migfilin suppress cell migration? We have used two different approaches to address this question. The first is a mutational approach. This is based on our initial observation that overexpression of migfilin(s), a naturally occurring migfilin splicing variant lacking the proline-rich domain (and hence VASP-binding defective), unlike that of migfilin, fails to suppress cell migration (Figs. 7 and 8). This finding is interesting for two reasons. First, it suggests a new mechanism by which cells can regulate their motility. It is well known that different types of cells often exhibit vastly different motilities. Cell motility is influenced by many factors including the strength of cellmatrix adhesion, lamellipodial protrusion, etc., and ultimately, the level and subcellular localization of molecules that control one or more of these processes (1)(2)(3). The presence of two naturally occurring migfilin splicing variants with vastly different motility regulating activities provides a system by which cells could control their motility through RNA splicing. Second, this finding suggests that migfilin likely suppresses cell migration through proteins that interact with its proline-rich domain. We have demonstrated that VASP binds to the proline-rich domain of migfilin. Furthermore, we have mapped the VASP binding to a single LPPPPP site (residues 104 -109) within the proline-rich domain. Substitution mutations within the L 104 PPPPP site abolished the VASP binding (Fig. 3). Importantly, overexpression of the VASP-binding defective migfilin mutant, like that of migfilin(s), failed to suppress cell migration ( Figs. 9 and 10). These results provide strong evidence for a role of the VASP binding in migfilin-mediated suppression of cell migration. In the second approach, we have suppressed VASP expression by RNA interference. The results showed that, unlike cells that express a normal level of VASP, VASP knockdown cells were unresponsive to migfilin overexpression (Figs. 10F and 11E). This result is highly consistent with the results obtained with the mutational approach. Collectively, they suggest that VASP is likely involved in the suppression of cell migration induced by the overexpression of migfilin.
A large body of genetic, cellular, and biochemical evidence has demonstrated that VASP is a key regulator of actin cytoskeletal dynamics  1 and 3) or the migfilin expression vector (lanes 2 and 4). One day after the DNA transfection, the cells were transfected with VASP siRNA2 (lanes 3 and 4) or the control RNA ( lanes  1 and 2). The cell lysates were analyzed by Western blotting (12 g of protein/lane) with Abs that recognize VASP (A) or migfilin (B). The membrane used in panel A was re-probed with an anti-actin Ab (C). D-F, cell migration was analyzed as described in the legend to Fig. 10, E-G. Bars represent mean Ϯ S.D. (n ϭ 3). **, p Ͻ 0.05 versus control.
(reviewed in Refs. 9 -11). For example, VASP, through its EVH1 domain, binds to the proline-rich motifs in the ActA protein of Listeria monocytogenes (29 -32). The binding of VASP to the ActA protein exerts several effects on actin assembly including stimulation of the Arp2/3-mediated actin nucleation and reduction of the number of filamentous actin branches (33)(34)(35)(36)(37). The consequences of VASP-mediated protein interactions are complex and often contextually (and subcellular localization) dependent. For example, in the case of L. monocytogenes, the binding of VASP to the bacterial surface of the ActA protein promotes Listeria motility within the cells (reviewed in Refs. 37 and 38). In Dictyostelium, DdVASP promotes cell adhesion, filopodia formation, and directional cell movement (39). In the nervous system, members of the Ena/VASP protein family are required for proper neuronal migration and axon guidance (40 -48). Using mouse fibroblasts, Bear et al. (22) have shown that VASP negatively regulates the motility of these cells. VASP localizes to both cell-matrix adhesions and lamellipodia. However, lamellipodia are the primary sites in which VASP suppresses fibroblast motility (22). These studies demonstrate an important mechanism by which VASP suppresses cell migration. It remained to be determined, however, whether in other cell types VASP could regulate cell migration through other mechanisms. Theoretical and experimental analyses have shown that the effect of the cell-matrix adhesion strength on cell migration is biphasic (reviewed in Refs. 1 and 2). Abundant VASP is present at cell-matrix adhesions but the function of VASP in these sites has been a puzzling question. In this study, we have found that overexpression of migfilin, which localizes to cell-matrix contacts but not lamellipodia, enhances MDCK cell-matrix adhesion and concomitantly reduces cell migration. Importantly, neither migfilin(s) nor the VASP-binding defective mutant can do so. Thus, in these cells the interaction of VASP with migfilin at cell-matrix adhesions likely enhances cell-matrix adhesion and consequently suppresses cell migration. These results provide, for the first time, functional evidence suggesting a role of VASP (via its interaction with migfilin) at cell-matrix adhesions.