Identification of a Filamin Docking Site on PTP-PEST*

PTP-PEST is a cytoplasmic protein-tyrosine phosphatase (PTP) implicated in the regulation of biological processes such as cell motility, cytokinesis, focal adhesion disassembly, and lymphocyte activation. Using a proteomics approach, filamin-A was identified as a novel interacting protein that bound to GST-PTP-PEST. This interaction was confirmed in vitro and in cells by coimmunoprecipitation. The site of filamin interaction on PTP-PEST was mapped to the fourth proline-rich region (Pro4). PTP-PEST has previously been implicated in the regulation of cytokinesis. In further support of this finding, expression of PTP-PEST in HeLa cells resulted in the formation of multinucleated cells. A PTP-PEST mutant lacking Pro4 and unable to bind filamin-A failed to induce the multinucleated phenotype. Further, depletion of filamin-A in HeLa cells was found to reduce the PTP-PEST-dependent multinucleation phenotype. Hence, we conclude that the interaction of PTP-PEST with filamin-A may function in the control of cytokinesis in mammalian cells.

containing 10% fetal bovine serum. 293T cells were transfected using the calcium phosphate method using 10 g of DNA per 10-cm tissue culture plate (27). HeLa cells were transfected using TransIT-TKO according to the manufacturer's recommendation (Mirus, Madison, WI). Cells (7.5 ϫ 10 5 ) were plated in a 6-cm tissue culture plate and incubated overnight at 37°C prior to transfection. Routinely, 30 l of TransIT-TKO were incubated with 7.5 g of DNA in 750 l of serum-free DMEM for 10 min prior to addition to the cells. Adenovirus constructs were generous gifts from Dr. Joan Taylor (UNC) and Dr. Mark Sussman (San Diego State University). The control and filamin (SMARTpool TM ) siRNA were purchased from Dharmacon (Lafayette, CO). For adenovirus infection, 2 ϫ 10 6 cells were incubated with 30 PFU/cell at 37°C for 24 h. Experiments were performed 72-h postinfection. For metabolic labeling, cells were grown to 80% confluence, washed with serum-free medium, then incubated overnight in methionine-and cysteine-free medium containing 10% conditioned media and 1.25 mCi 35 S-Translabel (MP Biomedicals, Inc, Irvine CA). Cell cultures were viewed in phase contrast using a Nikon TMS inverted microscope and imaged using a Nikon Coolpix 950 digital camera (ϫ100 magnification).
Fluorescence Microscopy and Flow Cytometry-For expression analysis of YFP fusion proteins, HeLa cells were fixed in 3.7% formaldehyde for 10 min, washed once in phosphate-buffered saline (PBS), then incubated with PBS containing 100 ng/ml 4,6-diamino-2-phenylindole (DAPI) (Molecular Probes, Eugene, OR) for 3 min. The cells were washed once more in PBS, and mounted in a mixture of 10% Tris, pH 7.5, 90% glycerol, and 25 mg/ml 1,4-diazabicyclo (2,2,2)octane (DABCO) (Sigma). Immunostaining with filamin-A (1:500-Chemicon International, Victoria BC), PTP-PEST (1:100-Exalpha Biologicals, Maynard, MA), and ␤-tubulin (1:100 Sigma) monoclonal antibodies was performed on ice-cold 100% methanol-fixed cells (5 min room temperature) followed by DAPI staining and mounting as described above. For analysis by flow cytometry, trypsinized HeLa cells were washed once in PBS, resuspended in 500 l of PBS, and fixed by addition of 315 l of 95% ice-cold ethanol. The cells were stored at Ϫ20°C until analysis. One hour prior to flow cytometry, the cells were pelleted by centrifugation at 1000 ϫ g for 5 min, washed once in PBS, then resuspended in 500 l of PBS containing 20 g/ml propidium iodide and 30 g/ml RNaseA. The cells were vortexed and incubated at 37°C for 1 h. Flow cytometry was performed on a Becton Dickinson FACScan interfaced to a Cytomation, Inc. Cicero data acquisition system in the UNC Flow Cytometry Facility. Data were analyzed using Summit version 3.1 (Cytomation, Fort Collins, CO).
Molecular Biology-GST fusion proteins containing fragments of PTP-PEST were engineered using PCR to amplify segments of the PTP-PEST open reading frame. Amplified product was inserted into pGEX2TK (Amersham Biosciences) in-frame with the GST coding sequence. Two deletion mutants of PTP-PEST were created, ⌬Pro4, in which residues 674 through 682 were deleted, and ⌬Pro5, in which the C-terminal 20 amino acids were deleted. This latter mutation is similar to the ⌬CTH mutant of PTP-HSCF/BDP, which deletes the C-terminal 24 amino acids and abolishes PSTPIP binding (11). These were created by amplification of the entire plasmid, except for the sequences to be deleted, followed by blunt end ligation. Point mutations were engineered into PTP-PEST using the QuikChange strategy (Stratagene, La Jolla CA). Sequence analysis was performed on each mutant to verify the intended point mutations/deletions and that no unintended mutations were present. These analyses were performed in the UNC-CH Genome Analysis Facility on a model 3730 DNA Analyzer (PerkinElmer Life Sciences) using the ABI PRISM TM Dye Terminator Cycle Sequencing Ready Reaction kit with AmpliTaq DNA Polymerase, FS (PerkinElmer, Applied Biosystems Division).
Expression of Recombinant Proteins-Expression of the fusion proteins in Escherichia coli was induced by incubation with 0.3 mM isopropyl ␤-D-thiogalactopyranoside for 4 -5 h at room temperature. The bacteria were lysed by sonication, and the proteins purified by binding to glutathione-agarose beads. The fusion proteins were quantified by SDS-PAGE and Coomassie Blue staining.
Protein Analysis-Cells were lysed in 1% Triton X-100 lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100) or 0.1% Triton X-100 lysis buffer (0.1% Triton X-100, 0.3 M sucrose, 50 mM Tris.HCl (pH 7.5), 100 mM KCl, 1 mM CaCl 2 , 2.5 mM MgCl 2 ) containing protease and phosphatase inhibitors and the protein concentration determined using the bicinchoninic acid assay (Pierce). Immunoprecipitations were performed using 1-2 mg of protein and 10 l of polyclonal antiserum or 2 l of purified antibody. Immune complexes were precipitated using protein A-Sepharose or protein G-Sepharose beads, washed twice with lysis buffer, twice with Tris-buffered saline and boiled in sample buffer. For GST pulldown experiments, 1-2 mg of lysate was precleared with 20 g of GST bound to glutathione beads for 45 min at 4°C. The cleared supernatant was then incubated with 10 -20 g of fusion protein immobilized on glutathione beads for 3 h at 4°C. The beads were washed twice in lysis buffer, twice in Tris-buffered saline and boiled in sample buffer. Immune complexes and proteins bound to GST fusion proteins were analyzed by Western blotting using a variety of primary antibodies, secondary antibodies conjugated to horseradish peroxidase and enhanced chemiluminescence for detection (Amersham Biosciences). Commercially available antibodies against filamin (Chemicon International, Victoria BC) and ␤-tubulin (Sigma) were used for these analyses. The polyclonal PTP-PEST antiserum used in this study has been described (20).
Mass Spectrometry-Samples were run on an SDS-polyacrylamide gel, and the bands visualized by silver staining. Protein identification by mass spectrometry was done in the Michael Hooker Proteomics Core Facility at UNC-CH. Briefly, the band of interest was excised and digested with trypsin at ϳ1:50 enzyme/substrate ratio. Tryptic peptides were eluted from the gel, and MALDI analysis was performed on a Bruker Reflex III time-of-flight mass spectrometer (Bruker). The mass spectrum was analyzed using the RADARS bioinformatics program (28).

RESULTS
Identification of a Novel PTP-PEST Binding Partner-The noncatalytic domains of PTPs serve as docking sites for other proteins including substrates (29). Whereas a number of substrates and adaptor proteins are known to bind the C-terminal noncatalytic domain of PTP-PEST, it seems likely that additional binding partners remain to be identified. As a first step in the identification of novel binding partners, recombinant GST fusion proteins containing three different fragments of PTP-PEST, immobilized on glutathione-agarose beads, were incubated with lysates of HEK 293 cells metabolically labeled with [ 35 S]methionine and [ 35 S]cysteine. After washing, bound proteins were segregated by SDS-PAGE and visualized by fluorography. A number of proteins bound to all of the fusion proteins including the GST negative control. Each of the regions of PTP-PEST bound uniquely to several proteins. The catalytic domain of PTP-PEST (amino acids 1-305) specifically bound to four proteins with molecular masses of 300, 90, 68, and 55 kDa. The N-terminal region of the noncatalytic domain, containing amino acids 297-493, bound specifically to two proteins, a 305 and an 80 kDa proteins. The most striking associated protein was the 280-kDa protein that uniquely bound to the fusion protein containing the C-terminal region of the noncatalytic domain, containing amino acids 470 -775 ( Fig. 1). Two additional proteins of molecular mass 80 and 58 kDa were also associated with this fusion protein. A proteomics approach was chosen to pursue the identity of novel PTP-PEST binding partners. Twenty milligrams of HEK 293 cell lysate were incubated with 600 g of GST fusion protein immobilized on glutathioneagarose beads. The beads were washed, bound proteins eluted with sample buffer and run on a preparative SDS-polyacrylamide gel. Proteins were visualized by silver staining. The 280-kDa protein was again found to associate with the fusion protein containing amino acids 471-775 of PTP-PEST, but was not found in complex with other fusion proteins (data not shown). The protein band was excised and subjected to in-gel trypsin digestion. Peptides were eluted from the gel and analyzed by matrix-assisted laser desorption/ionization mass spectrometry at the Michael Hooker Proteomics Core Facility at UNC-CH. The spectrum of the 280-kDa PTP-PEST associated protein was analyzed using the RADARS bioinformatics software and identified as filamin-A (expectation value 4.3 ϫ 10 Ϫ5 , 13% peptide coverage). Thus, based upon its mass fingerprint, PTP-PEST appeared to associate with filamin (also known as actinbinding protein (ABP-280).
Validation That Filamin Binds PTP-PEST-To validate the finding that filamin binds PTP-PEST, GST pull-down and coimmunoprecipitation experiments were performed. Lysates of HEK 293 cells were incubated with GST alone or GST-PTP-PEST fusion proteins, which were immobilized on glutathione beads. The beads were washed and bound filamin was detected by Western blotting. Whereas filamin did not bind to GST alone, filamin was associated with the fragment of PTP-PEST extending from residue 470 -775 ( Fig. 2A). This result validated the identification of filamin as a binding partner of recombinant PTP-PEST. To further verify this interaction, endogenous PTP-PEST was immunoprecipitated from HEK 293 lysates using a polyclonal antiserum and the immune complex analyzed by Western blotting with a filamin antibody. Endogenous filamin was detected in the PTP-PEST immune complex, but was not detected in preimmune control immunoprecipitations (Fig.  2B). The reciprocal co-immunoprecipitation experiments were performed using lysates of Hela cells infected with adenoviruses designed to express GFP or PTP-PEST (Fig. 2C). Filamin was immunoprecipitated and the immune complex blotted for associated PTP-PEST using the polyclonal antiserum (Fig. 2D). PTP-PEST was detected in the filamin immune complex from PTP-PEST-expressing cells, but was not found in the isotype control immune complex (Fig. 2D). A small amount of PTP-PEST could be detected in filamin immune complexes from GFP-expressing cells. This reflected interaction between endogenous PTP-PEST and filamin (Fig. 2D). These reciprocal co-immunoprecipitations suggest that filamin is a bona fide binding partner of PTP-PEST and that the two proteins are likely to associate in vivo.
Identification of the Filamin Binding Site on PTP-PEST-To map the region of PTP-PEST required for filamin binding, GST fusion proteins containing different fragments of PTP-PEST were used. Lysates were incubated with fusion proteins immobilized to glutathione-agarose beads and bound filamin was detected by Western blotting. A recombinant protein containing amino acids 560 -700 of PTP-PEST bound filamin, whereas fusion proteins containing PTP-PEST residues 471-560 or 700 -775 did not (Fig. 3A). This experiment narrowed the filamin binding site to within residues 560 -700 of PTP-PEST. Two additional fusion proteins encoding residues 560 -650 and 600 -700 were tested for filamin binding. The former failed to bind filamin, whereas the latter was able to bind filamin (Fig.  3B). This finding suggested that the filamin binding site on PTP-PEST was likely between residues 650 and 700, a region that contains the fourth proline-rich sequence of PTP-PEST, Pro4. As numerous proteins dock to PTP-PEST via prolinerich sequences in its C-terminal noncatalytic domain, Pro4 appeared as a good candidate binding site for filamin. To test this hypothesis, a 9-amino acid deletion removing Pro4 was engineered into full-length PTP-PEST (⌬Pro4). For this experiment, PTP-PEST constructs were transiently expressed in HEK 293 cells as yellow fluorescent protein (YFP) fusion proteins to assess the ability of these mutants to co-immunoprecipitate with endogenous filamin. YFP-PTP-PEST, YFP-⌬Pro4, and YFP-⌬Pro5 (a mutation deleting the C-terminal 20 resi-  2 and 4) or isotype control (lanes 1 and 3) antibodies. Immune complexes were collected using protein G-Sepharose, washed, and bound PTP-PEST detected by Western blotting. dues which contains Pro5) were expressed to similar levels, as detected by Western blotting (Fig. 3C). Filamin was immunoprecipitated from lysates and associated PTP-PEST detected by Western blotting using a PTP-PEST polyclonal antibody. Whereas YFP-PTP-PEST and YFP-⌬Pro5 were co-immunoprecipitated with filamin, the YFP-⌬Pro4 mutant was defective for filamin binding (Fig. 3D). This result supports the contention that this proline-rich region is the site of interaction between PTP-PEST and filamin.
PTP-PEST Expression Induces a Multinucleated Phenotype-HeLa cell populations stably expressing PTP-PEST variants as YFP fusion proteins were established by drug selection and sorting by FACS. Examination of wild-type YFP-PTP-PESTexpressing HeLa cells revealed a high frequency of cells with a multinucleated phenotype (Fig. 4A). This phenotype is intriguing, given the reported role of PTP-PEST in controlling cytokinesis (5). The wild-type YFP-PTP-PEST-expressing cells exhibited a severe cell proliferation defect that resulted in a rapid loss of protein expression (data not shown). The multinucleated phenotype was recapitulated using two different methods of transient expression; infection with adenoviral vectors and transient expression of YFP fusion proteins. HeLa cells were infected with an adenovirus designed to express GFP (Ad-GFP), which served as a negative control, or with the adenovirus-PTP-PEST (Ad-PTP-PEST). Seventy-two hours postinfection, nuclei were stained with DAPI, and the cells visualized by immunofluorescence using a PTP-PEST or a ␤-tubulin antibody, or by visualizing GFP fluorescence. Ad-GFP-infected cells were rarely observed to contain more than a single nucleus, whereas Ad-PTP-PEST infected cells were frequently observed to contain multiple nuclei (Fig. 4B). Multinucleated cells were scored in multiple experiments and expressed as percentage of cells containing multiple nuclei. Whereas a small percentage of control cell populations contained more than a single nucleus, ϳ25% of the Ad-PTP-PEST-infected cells were multinucleated (Fig. 4C). This phenotype was further validated by examining the ploidy of Ad-GFP and Ad-PTP-PEST infected cells by flow cytometry. Asynchronously growing cells were fixed, stained with propidium iodide, and analyzed by flow cytometry. Control cells exhibited the typical profile of cycling cells with a peak of cells with 2N DNA content, a peak of cells with 4N DNA content and a population of cells with DNA content between 2N and 4N indicative of cells in S phase (Fig. 4D). Ad-PTP-PEST cells exhibited the same basic profile with two notable differences. First, the ratio of cells with 4N DNA content to cells with 2N DNA content was increased. Second, there was a population of cells with greater than 4N DNA content that was not present in the control cells. The flow cytometric analysis also supports the finding that PTP-PEST-expressing cells exhibit a higher frequency of multiple nuclei.
The Pro4 Region Is Required to Induce the Multinucleated Phenotype-To begin to address the mechanism by which PTP-PEST generates a multinuclear phenotype, mutants of PTP-PEST were analyzed for their ability to induce this phenotype. YFP-PTP-PEST fusion proteins were transiently expressed in HeLa cells. Seventy-two hours post-transfection, cells were fixed, and stained with DAPI, and cells examined by fluorescent microscopy. Cells expressing exogenous protein were identified as YFP-positive cells and the number of nuclei per YFPpositive cell counted (Fig. 5A). Less than 5% of control cells, expressing the empty YFP vector, exhibited multiple nuclei. Approximately 17% of the YFP-PTP-PEST-expressing cells exhibited the multinucleated phenotype (Fig. 5B). Thus the results obtained using adenovirus to express exogenous PTP-PEST were recapitulated using transient transfection of YFP fusion proteins. To address the role of phosphatase activity in inducing the formation of multinucleated cells, a catalytically defective mutant was used. Two mutations were engineered into PTP-PEST, C231S, which removes the catalytic cysteine and abolishes activity, and R237D, which dramatically reduces the affinity of phosphatases for their substrates (30,31). This latter mutation ensured that the mutant would not exhibit neomorphic effects because of substrate trapping in vivo. In contrast to YFP-PTP-PEST, YFP-RD/CS induced a multinucleated phenotype in ϳ6% of cells, which was not statistically different from the level seen in control cells (Fig. 5B). This result demonstrates that catalytic activity is essential for induction of the phenotype. The two PTP-PEST binding partners most likely involved in controlling cell division are filamin and PSTPIP. To address the importance of these interactions, the ⌬Pro4 and ⌬Pro5 mutants were tested for their ability to produce multinucleated cells. The ⌬Pro5 mutant was engineered to mimic the previously published mutation deleting the CTH region of PTP-PEST family of PTPs, which disrupts PSTPIP binding (11). YFP-⌬Pro4-expressing cells contained ϳ6% multinucleated cells, which again was not significantly different from the frequency seen in control cells, demonstrating that the Pro4 sequence was required for PTP-PEST to induce this phenotype (Fig. 5B). Expression of YFP-⌬Pro5 in cells produced a multinucleated phenotype in ϳ12% of the cells. This result is statistically different from both the vector control cells (Tukey's Multiple Comparison Test, p Ͻ 0.001) and wild-type-expressing cells (p Ͻ 0.01). This result suggests that both the Pro4 and Pro5 regions function in controlling this phenotype, and that the Pro4 region is essential for inducing multinucleation.
If the PTP-PEST/filamin complex was biologically significant during cytokinesis, it was postulated that cells devoid of filamin-A would be resistant to PTP-PEST-induced multinucleation. A filamin-A-deficient human melanoma cell line (M2) and a derivative re-expressing wild-type filamin (A7) (32) seemed appropriate for these studies. However, the M2 cells exhibited a multinucleated phenotype (Ͼ20%) while A7 cells were predominantly mononuclear (data not shown). These findings preclude the use of the M2 cells to address this question. As an alternative strategy, filamin-A expression in HeLa cells was reduced using siRNA (Fig. 6A). It was only possible to partially deplete filamin-A in transfected HeLa cells and following transfection, cells expressing high levels of filamin-A, and cells with suppressed levels of filamin-A were clearly distinguished by immunofluorescence (Fig. 6B). siRNA-mediated knockdown of filamin did not cause a significant change in the degree of multinucleation of HeLa cells (3.3 Ϯ 0.9%) in comparison to control HeLa cells (2.0 Ϯ 0.35%), possibly because of the incomplete inhibition of expression.
Optimum siRNA-mediated suppression of filamin levels was at 72 h, and could be maintained beyond 120 h, while optimum expression of YFP-PTP-PEST fusion constructs was at 72 h. Hence, HeLa cells were transfected with filamin siRNA and 48 h later the cells were transfected with plasmids encoding the YFP fusion proteins. The multinucleated phenotype was assessed 72 h following transfection with the GFP plasmids. The cells were fixed and stained for filamin-A to identify cells expressing high and low levels of filamin-A. The nuclear phenotype of YFP-positive cells was scored (Fig. 6C). In cells containing normal levels of filamin-A, 15% of YFP-PTP-PEST-expressing cells were multinucleated. In contrast, the frequency of multinucleated YFP-PTP-PEST-expressing cells was reduced to 5% in cells with depleted filamin (Fig. 6C). The proportion of control YFPexpressing cells exhibiting a multinucleated phenotype was similar in cells exhibiting normal or depleted levels of filamin-A and was not different to the proportion seen in untransfected HeLa cells, i.e. Ͻ5%. Furthermore, cells expressing the filaminbinding defective PTP-PEST mutant (⌬Pro4) exhibited similar proportions of multinucleated cells as control cells with either normal or suppressed levels of filamin. These results support the hypothesis that the interaction between PTP-PEST and filamin is important for the induction of the multinucleated phenotype by PTP-PEST.

DISCUSSION
In this article, we have identified filamin as a novel binding partner for PTP-PEST. Whereas it is not yet clear if this reflects a direct or indirect interaction, the Pro4 sequence within PTP-PEST is required for the interaction. Establishing cells stably expressing exogenous PTP-PEST proved difficult apparently because of a growth defect and early passage HeLa cells stably expressing PTP-PEST had an increased propensity for multinucleation. The selective disadvantage conferred by PTP-PEST-expressing cells led to rapid loss of protein expression. The multinucleated phenotype was also induced in HeLa cells when PTP-PEST was expressed using an adenoviral vector or transiently expressed as a YFP fusion protein.
It has been reported that PTP-PEST-null cells also exhibit a cytokinesis defect as pairs of cells joined by an actin-rich remnant of the cleavage furrow are frequently observed (5). At first glance the findings that exogenous PTP-PEST expression and ablation of PTP-PEST expression produce a similar phenotype are counterintuitive, i.e. that gain of function and loss of function result in similar phenotypes. Interestingly, similar seemingly disparate results between null cells and cells expressing exogenous PTP-PEST are observed in the motility of cells (5,8). Cells expressing exogenous PTP-PEST exhibit reduced cell motility, as do PTP-PESTnull fibroblasts. These observations can be reconciled if cycles of tyrosine phosphorylation/dephosphorylation are required to regulate these biological processes. Cells expressing exogenous protein exhibit insufficient phosphorylation of key substrates, and thus a failure to regulate function resulting in defective cytokinesis. Conversely, null cells exhibit constitutive phosphorylation of key substrates and thus a failure to regulate function resulting in defective cytokinesis.
The PTP-PEST mutant that exhibits deficient filamin binding (⌬Pro4) does not induce a multinucleated phenotype implicating a Pro4 binding partner in the induction of this phenotype. Pro4 is a binding site for three different proteins, Grb2, Csk, and filamin (12,18,19). It is quite unlikely that Grb2 is the relevant Pro4 binding partner associated with this phenotype as Grb2 can associate with three additional proline-rich sequences in the C-terminal noncatalytic domain of PTP-PEST (12). Csk is a second Pro4 binding partner (18,19), and the major cellular function of this tyrosine kinase is to negatively regulate the catalytic activity of the Src family of tyrosine kinases (33). The c-Src kinase is activated during mitosis (34) and inhibition of Src family kinases blocks the initiation of cytokinesis (35,36). However, PTP␣, rather than Csk, appears to be the major enzyme regulating phosphorylation of the negative regulatory site during mitosis (37). The third Pro4 binding partner that could function in the control of cytokinesis is filamin. A role for filamin in cytokinesis has been inferred by its localization to the cleavage furrow/midbody (38). This original finding was recently confirmed in a proteomics study identifying filamin as a component of isolated midbodies, although RNAi-mediated knockdown of filamin did not induce a cytokinesis phenotype in Caenorhabditis elegans (39). Because the induction of a multinucleated phenotype by PTP-PEST requires filamin-A expression, filamin-A is the most likely known Pro4-docking protein controlling this phenotype. The catalytic activity of PTP-PEST is required to promote the multinucleated phenotype, suggesting that dephosphorylation of select substrates is necessary to alter cell division. Filamin has been reported as a substrate for tyrosine phosphorylation by Lck in vitro (40), but we have been unable to detect phosphotyrosine changes on filamin upon transient expression of PTP-PEST (data not shown). It seems more likely that this interaction serves to localize a fraction of the cellular PTP-PEST in the vicinity of other proteins, which serve as direct substrates. There is recent evidence for the involvement of phosphotyrosine-containing proteins during cytokinesis. Tyrosine-phosphorylated proteins have been found to accumulate in cleavage furrows in sea urchin embryos (41). The molecular targets for tyrosine phosphorylation in regulation of cytokinesis have not been conclusively established and the relevant PTP-PEST substrates are unknown. There are a number of reasons why RhoGAPs or RhoGEFs are currently the favored targets. Filamin-A, which can in turn associate with Rho (26), may act as a scaffold to tether these proteins in complex to facilitate regulation. PTP-PEST has already been implicated as a regulator of Rac in fibroblasts (9,10). Trio is an exchange factor for Rho that associates with filamin-A and is tyrosine-phosphorylated, but it is unclear if its activity is regulated by phosphorylation (42,43). Vav3 is a RhoGEF that is regulated in a cell cycledependent manner in HeLa cells and constitutive expression of Vav3 promotes a multinucleated phenotype (44). Tyrosine phosphorylation of the regulatory region of the Vav family of GEFs promotes a conformational change and activation of catalytic activity (45). Mutation of this tyrosine to phenylalanine in Vav3 (tyrosine 173) abolishes its ability to induce multinucleated cells, indirectly suggesting that catalytic activity is required to promote this phenotype (44). The activity of p190RhoGAP is regulated in vivo by tyrosine phosphorylation (46). P190RhoGAP has been localized to the region of the cleavage furrow in dividing cells and overexpression of p190RhoGAP induces a multinucleated phenotype (47). Clearly there are a number of plausible mechanisms via which PTP-PEST could serve to regulate cytokinesis. Future studies will focus upon the elucidation of molecular targets and the detailed mechanism of how PTP-PEST controls cytokinesis.