The Receptor Protein-tyrosine Phosphatase PTPμ Interacts with IQGAP1*

The receptor protein-tyrosine phosphatase PTPμ is a member of the Ig superfamily of cell adhesion molecules. The extracellular domain of PTPμ contains motifs commonly found in cell adhesion molecules. The intracellular domain of PTPμ contains two conserved catalytic domains, only the membrane-proximal domain has catalytic activity. The unique features of PTPμ make it an attractive molecule to transduce signals upon cell-cell contact. PTPμ has been shown to regulate cadherin-mediated cell adhesion, neurite outgrowth, and axon guidance. Protein kinase C is a component of the PTPμ signaling pathway utilized to regulate these events. To aid in the further characterization of PTPμ signaling pathways, we used a series of GST-PTPμ fusion proteins, including catalytically inactive and substrate trapping mutants, to identify PTPμ-interacting proteins. We identified IQGAP1, a known regulator of the Rho GTPases, Cdc42 and Rac1, as a novel PTPμ-interacting protein. We show that this interaction is due to direct binding. In addition, we demonstrate that amino acid residues 765-958 of PTPμ, which include the juxtamembrane domain and 35 residues of the first phosphatase domain, mediate the binding to IQGAP1. Furthermore, we demonstrate that constitutively active Cdc42, and to a lesser extent Rac1, enhances the interaction of PTPμ and IQGAP1. These data indicate PTPμ may regulate Rho-GTPase-dependent functions of IQGAP1 and suggest that IQGAP1 is a component of the PTPμ signaling pathway. In support of this, we show that a peptide that competes IQGAP1 binding to Rho GTPases blocks PTPμ-mediated neurite outgrowth.

Reversible protein-tyrosine phosphorylation is a primary mode of regulation for several cellular functions including growth, differentiation, adhesion, and protein trafficking. The overall tyrosine phosphorylation state of a protein is regulated by protein-tyrosine kinases (1) and protein-tyrosine phosphatases (2). Similar to the family of protein-tyrosine kinases, the family of protein-tyrosine phosphatases includes both non-receptor and receptor type enzymes. Unlike the tyrosine kinases, however, few substrates and downstream signaling molecules of tyrosine phosphatases have been identified. This manuscript focuses on the receptor protein-tyrosine phosphatase (RPTP), 2 PTP, and the identification of novel PTP-interacting proteins to aid in the elucidation of the PTP signaling pathway.
The extracellular domain of PTP contains sequence motifs similar to those in cell adhesion molecules including: a MAM domain, an immunoglobulin (Ig) domain, and four fibronectin type-III repeats (3). The MAM domain (Meprins, A5, PTP Mu) is a sequence motif that is suggested to play a role in protein dimerization (4). Ig domains are disulfide-bonded structures found in many cell surface proteins and have been shown to mediate homophilic and heterophilic binding between cell adhesion molecules. Fibronectin type-III motifs, originally identified in the extracellular matrix protein fibronectin, are present in many cell adhesion molecules. PTP has been shown to mediate homophilic binding in non-adhesive Sf9 insect cells (5,6). The juxtamembrane domain of PTP contains a region of homology to the conserved intracellular domain of the cadherins. Cadherins are calciumdependent adhesion molecules that mediate cell-cell adhesion and adherens junction formation (7). Cadherins are anchored to the actin cytoskeleton indirectly through the binding of catenins (7). The catenin family of proteins includes ␣-catenin, ␤-catenin, plakoglobin, and p120 catenin. We have demonstrated that PTP interacts with classical cadherins such as N-cadherin, E-cadherin, and R-cadherin and associates with these cadherins in a complex containing ␣-catenin and ␤-catenin (8,9). PTP was also shown to bind p120 catenin (10). PTP has two conserved intracellular catalytic domains, of which only the membraneproximal domain has been shown to be catalytically active (11). The cell adhesion-like extracellular domain of PTP, its intracellular catalytic domain and ability to interact with the cadherin/catenin complex, make PTP an attractive candidate for regulating cadherin-mediated adhesion and migration. In support of this idea, we demonstrated that re-expression of PTP in LNCaP cells restores E-cadherin-mediated cell-cell adhesion (12). Furthermore, PTP expression and catalytic activity are required for N-cadherin mediated neurite outgrowth (13).
In a continuing effort to understand the PTP signaling pathway, we set out to identify potential PTP substrates and interacting proteins using a previously published strategy (14). PTPs have a conserved aspartate residue, which serves as a general acid during catalysis. The structural data on the PTP catalytic domain suggest that mutation of the aspartate residue to alanine (D-A) creates a "substrate trap" by affecting only the V max and not the K m of the phosphatase (15). Holsinger et al. (14) used a fusion protein consisting of an intracellular D1205A mutant of the RPTP, DEP-1, fused to glutathione S-transferase (GST), to isolate substrates from lysates of cells treated with the protein-tyrosine phosphatase inhibitor pervanadate. They found that DEP-1 associates with p120-catenin. Palka et al. (16) performed similar experiments with DEP-1 and identified additional substrates including the hepatocyte growth factor receptor (met) and Gab1 as well as p120-catenin. We generated a D1063A mutant using the intracellular domain of PTP (iPTPDA) and expressed it as a bacterial GST fusion protein. The iPTPDA fusion protein was used in pull-down assays with A549 cell lysates and associated proteins were identified by immunoblotting. Using this strategy, we identified IQGAP1 as a PTP-interacting protein.
It is well established that members of the Rho-family of GTPases regulate actin cytoskeleton remodeling. The most well known members of this family include Cdc42, Rac1, and RhoA, which generate the formation of filopodia, lamellipodia, and stress fibers, respectively (32). IQGAP1 represents a protein that can link cytoskeletal proteins to Rhofamily GTPases, specifically Cdc42 and Rac1 (20,(27)(28)(29)33). In this manuscript, we demonstrate that IQGAP1 is a novel PTP-interacting protein. We show that the interaction between PTP and IQGAP1 is direct and enhanced by activated Cdc42 and to a lesser extent, Rac1. Furthermore, we demonstrate PTP regulates neurite outgrowth via IQGAP1.

EXPERIMENTAL PROCEDURES
Antibodies-Polyclonal antibodies to IQGAP1 (used for immunocytochemistry) and ERK2 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal antibodies to IQGAP1 (used for immunoblotting), ␤-catenin, and E-cadherin were purchased from Transduction Laboratories (San Diego, CA). An antibody to phospho-p44/42 MAPK (p-ERK1/2) was purchased from New England Biolabs (Beverly, MA). An antibody to calmodulin was purchased from Upstate Biotechnology (Lake Placid, NY). An anti-HA-peroxidase antibody used to detect the Rho GTPase fusion proteins was purchased from Roche Applied Science. A polyclonal antibody generated against the extracellular domain of PTP (494) and a monoclonal antibody generated against the intracellular domain of PTP (SK18) have been described previously (5,34). A monoclonal anti-phosphotyrosine antibody was a gift from Nick Tonks (Cold Spring Harbor Laboratory) and has been described previously (35). A monoclonal antibody to N-cadherin was a gift from Margaret Wheelock (University of Nebraska) and has been described previously (36).
Expression of Fusion Proteins in Escherichia coli-Plasmids containing intracellular PTP GST fusion proteins (iPTP-JMD, iPTP-⌬765-815, iPTPWT, iPTPCS, iPTPDA, iPTPWT-⌬D2, iPTPDA-⌬D2) or GST alone were expressed in E. coli under the con-trol of the lac promoter. The iPTP-⌬765-816 (B4), iPTPWT (B5), and iPTPCS (B5m) have been described (11). We generated the iPTPDA construct (D1063A mutation) using site-directed mutagenesis. We generated the iPTPWT-⌬D2 and iPTPDA-⌬D2 constructs by restriction digestion of iPTPWT and iPTPDA, respectively, with BamHI, which cuts at 3532 bp. iPTP-JMD was generated by restriction digest of iPTPWT-⌬D2 with AccI and ClaI. To isolate functional iPTP-GST fusion proteins, proteins were isolated from E. coli as follows. Bacteria were resuspended in 10 ml of resuspension buffer (0.1 M NaCl, 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA) and incubated on ice for 15 min. To lyse bacterial cells 1 ml of 0.5 M EDTA, 1.1 ml of 20% Triton X-100, 55 l of 1 M dithiothreitol, 10 l of ␤-mercaptoethanol, 100 l of 100 mM phenylmethylsulfonyl fluoride, and 30 l of protease inhibitor cocktail (Sigma) was added to 10 ml of resuspended cells. Cells were sonicated and spun at 15,000 rpm for 25 min. GST fusion proteins were isolated from the cleared supernatant using glutathione-Sepharose 4B beads (Amersham Biosciences). Expression and protein concentration of GST fusion proteins was determined by Coomassie stain. Isolated fusion proteins adsorbed onto glutathione-Sepharose were used in the GST pull-down experiments as described below. For pulldown assays performed in the absence of pervanadate treatment, GST fusion proteins were isolated from E. coli using PBST (1% Triton X-100, 1 mM benzamidine, and protease inhibitor cocktail in PBS).
GST Pull-down Experiments-A549 cells were grown to 85-95% confluence. Cells were collected by scraping into lysis buffer containing 20 mM Hepes, pH 7.5, 1% Nonidet P-40, 150 mM NaCl, 1 mM EDTA, 1 mM benzamidine, and protease inhibitor cocktail. Cells were incubated on ice for 30 min and then centrifuged at 3000 rpm for 3 min. The supernantant was saved and its protein concentration determined using the Bradford method (37). Equal amounts of protein (1 mg) were added to equal amounts of the iPTP GST constructs or GST alone adsorbed to glutathione-Sepharose. Samples were rocked at 4°C for 2.5 h, washed four times with lysis buffer, and incubated at 95°C for 5 min in 2ϫ SDS sample buffer. One-third of the sample was resolved by SDS-PAGE and transferred to nitrocellulose for immunoblot analysis as described previously (8). Pull-down experiments using purified His-IQGAP1 from Sf21 cells were performed in PBS, 1% Triton X-100, 1 mM benzamidine, and protease inhibitor cocktail.
Isolation of TAT Fusion Proteins-TAT fusions of the Rho family GTPases Cdc42, Rac1, and RhoA were obtained from Dr. Steven Dowdy (University of California San Diego, La Jolla, CA). The Rho GTPase fusion proteins contain an N-terminal His 6 -tag for purification, a TATtag for protein transduction, and a HA-tag for protein detection (38). The fusion proteins were expressed in E. coli under the control of the lac promoter and purified on TALON metal affinity resin (Clontech, Palo Alto, CA). Expression and protein concentration of TAT fusion proteins was determined by Bradford assay and verified by immunoblot using anti-HA-tag antibodies.
Purification of His-IQGAP1-IQGAP1 was cloned into pFast Bac HT as follows. The gene for full-length IQGAP1 was cut from pcDNA3-Myc-IQGAP1 (39) with XbaI and partially digested with BamHI. The construct was inserted into pFast Bac HT at BamHI and XbaI sites. pFast Bac HT-IQGAP1 was transformed into DH10 Bac. The recombinant Bacmid DNA of IQGAP1 was isolated and used to infect Sf21 insect cells. Forty-eight h post-infection, Sf21 cells were lysed and His-IQGAP1 purified using the QIAexpress Ni-NTA Fast Start kit (Qiagen, Inc., Valencia, CA) as indicated in the manufacturer's protocol with the modification that 1% Triton was included in all buffers with the exception of the elution buffer. Purity of the His-IQGAP1 protein was determined by silver stain and verified by immunoblot with a monoclonal anti-IQGAP1 antibody.
Immunocytochemistry-A549 cells were plated in Lab Tech II chamber slides (Nalge Nunc International Corp., Rochester, NY). When the cells reached the desired confluence (1-3 days post-plating) they were fixed with 4% paraformaldehyde for 10 min. Cells were then washed with PBS three times and permeabilized using 0.5% saponin in blocking buffer (20% goat serum, 1% bovine serum albumin in PBS) for 30 min. Primary antibodies were diluted in blocking buffer plus 0.5% saponin and incubated overnight at 4°C. After incubation with primary antibody, cells were rinsed five times with TNT buffer (0.1 M Tris-HCl, pH 7.5, 0.15 M NaCl, 0.05% Tween 20). Secondary antibodies (either goat anti-mouse fluorescein isothiocyanate, goat anti-mouse Texas Red, or goat anti-rabbit fluorescein isothiocyanate from ICN Biochemicals, Irvine, CA) were diluted in blocking buffer plus saponin and incubated for 1 h at room temperature. After incubation with secondary antibody, cells were washed five times with TNT buffer and once with distilled water. Molecular Probes SlowFade Light Antifade kit was used to minimize quenching. Slides were imaged using a Nikon (Tokyo, Japan) TE200 inverted microscope, and images were collected with a Spot RT digital camera and image acquisition software (Diagnostic Instruments, Inc., Sterling Heights, MI).
Baculovirus Infection of A549 Cells-Baculovirus containing a mammalian expression plasmid for GFP-tagged wild-type PTP was generated as previously described using the pBacMam-2 vector from Novagen (40). For co-localization studies, A549 cells were plated in Lab Tech II chamber slides 24 h prior to infection with baculovirus. For infection, 200 l of viral supernatant was added to chamber slide wells containing cells in 200 l of medium and incubated 2 h at 37°C, 5% CO 2 . After the 2-h incubation, all medium was removed from the cells and replaced with media containing 150 nM trichostatin A (Sigma). Cells were incubated at 37°C, 5% CO 2 overnight. The following day immunocytochemistry for IQGAP1 was performed as described above.
Immunoprecipitations-A549 cells were grown to 90% confluence. Cells were washed twice with PBS and treated with the cross-linking reagent DSP (1 mM, Pierce) for 10 min at room temperature. Tris, pH 7.5, was added to 50 mM for 15 min at room temperature, then the cells were washed once with PBS. Cells were lysed in 50 mM Tris, pH 7.5, 1% Triton X-100, 150 mM NaCl, 1 mM benzamidine, and protease inhibitor cocktail. Samples were sonicated and centrifuged at 10,000 rpm for 5 min. Supernatants were saved and protein concentrations determined using the Bradford method. Immunoprecipitations were performed with equal amounts of supernatant protein (400 g) using protein A-Sepharose (Amersham Biosciences) preloaded with a rabbit polyclonal antibody directed to an extracellular epitope of PTP (494) or non-immune rabbit serum. Samples were rocked at 4°C for 3-4 h. Beads were washed four times with lysis buffer and heated to 37°C for 15 min and 95°C for 5 min in 2ϫ SDS sample buffer. Immunoprecipitated proteins were resolved by SDS-PAGE and immunoblot.
Bonhoeffer Stripe Assay-The stripe assay used was a modified version of the Bonhoeffer method (41) described previously (42). Briefly, tissue cultures dishes were coated with nitrocellulose (43) and allowed to dry before applying a silicon lane matrix to the dish. Alternating stripes of laminin and PTP-Fc chimera containing Texas Red-conjugated bovine serum albumin (for visualization of the lanes) were generated as follows. 80 ng of PTP-Fc chimera (described previously; Ref. 44) was injected into the channels of the silicon lane matrix, incubated for 10 min, aspirated, and then replaced with a fresh aliquot of the same substrate two additional times. All remaining binding sites within the lanes were blocked with bovine serum albumin (fraction V; Sigma) and rinsed with calcium-magnesium-free phosphate buffer. The matrix was removed, and 1.75 g of laminin (Biomedical Technologies Inc., Stoughton, MA) was spread across the lane area and incubated for 20 min. Explants from E8 chick nasal retina were grown on the alternating stripes of laminin and PTP for 48 h. Quantitation of the stripe assays was performed using a rating scale previously described (45,46). Neurites that show no preference for either substrate are assessed at 0. Neurites that grow exclusively on one substrate are assessed at 3. Neurites that grow mainly on the laminin lanes with an occasional neurite crossing over the PTP lanes are assessed at 2, while an assessment of 1 is given when there is a significant amount of neurite crossing but a tendency to fasciculate on laminin. To perturb IQGAP1 function, a 6 M concentration of either a TAT-tagged, scrambled peptide or an IQGAP1 peptide containing an N-terminal TAT protein transduction sequence was added at the time of explant. The IQGAP1 peptide corresponds to amino acids 1054 -1077 of IQGAP1 plus the N-terminal TAT sequence (GRKKRRQRRRMVVSFNRGARGQNALRQILAP-VVK, synthesized by Genemed Synthesis, San Fransico, CA). A peptide comprising amino acids 1054 -1077 of IQGAP1 (MK24) is known to compete the interaction of IQGAP1 with Cdc42 and Rac1 (47).

IQGAP1 Associates with PTP in
Pull-down Assays-To identify potential PTP-interacting proteins we used a previously published strategy employing GST fusion proteins to pull down associated proteins (14,16). PTPs have a conserved aspartate residue, which serves as a general acid during catalysis. The structural data on the PTP catalytic domain suggest that mutation of the aspartate residue to alanine (Asp 3 Ala) creates a "substrate trap" by affecting only the V max and not the K m of the phosphatase (15). The "substrate trapping" mutant (Asp 3 Ala) retains normal affinity for its substrate but catalysis is reduced; therefore it binds irreversibly to its substrate. We used a series of intracellular PTP-GST fusion proteins (Fig. 1), including a D1063A mutant (iPTPDA), immobilized on glutathione-Sepharose in pulldown assays to identify PTP-interacting proteins. Lysates from confluent A549 cells treated with or without the tyrosine phosphatase inhibitor, pervanadate, were incubated with the GST fusion proteins, and associated proteins were resolved by SDS-PAGE and identified by immunoblot. The iPTPCS construct possesses a mutation in which an essential cysteine residue is mutated to a serine (C1095S), leaving the protein catalytically inactive. The iPTP-⌬765-815 construct lacks part of the PTP juxtamembrane domain. In our system, both the C1095S and D1063A mutant constructs (iPTPCS and iPTPDA) act equally well to "trap" potential substrates based on immunoblot data using a phosphotyrosine antibody to detect associated proteins in the presence of pervanadate (data not shown).
Based on the molecular weight of the proteins identified by the antiphosphotyrosine immunoblot described above, we made an educated guess as to what proteins may be interacting with PTP and probed for a number of these proteins. Among the proteins we identified by immunoblot was IQGAP1 (Fig. 2), a regulator of the Rho GTPases, Cdc42 and Rac1. All the PTP constructs pulled down IQGAP1 with approximately equal affinity with the exception of iPTP-⌬765-815, which has reduced affinity for several of the proteins we examined (Fig. 2 and data  not shown), suggesting the juxtamembrane domain may alter protein folding or the affinity of protein-protein interactions. In addition to IQGAP1, other known IQGAP1-interacting proteins were detected in the pull-down assays including calmodulin and ERK2 (p42 MAPK), in both its unphosphorylated and phosphorylated, active form, suggesting PTP interacts with IQGAP1 in a complex. These data do not provide evidence either for or against the possibility that IQGAP1 is a substrate for PTP due to the fact that substrates can bind in both the phosphorylated and unphosphorylated form (10,48,49). The slight variation in binding of IQGAP1 and ERK2 seen in the absence or the presence of pervanadate is not observed in all experiments. Equal loading of GST fusion proteins was confirmed by Ponceau staining of the nitrocellulose membrane (data not shown). After our initial pull-down experiments, we generated both a wild-type and D1063A mutant construct lacking the second phosphatase domain, iPTPWT-⌬D2 and iPTPDA-⌬D2, respectively. These constructs are more efficiently expressed in bacteria due to their smaller size and behave similarly to the full-length intracellular constructs with respect to the binding of IQGAP1 (Fig. 3A). These data indicate that PTP interacts with this complex via its juxtamembrane or first phosphatase domain.
Constitutively Active Cdc42 Enhances the Binding of IQGAP1 to PTP-Members of the Rho family of GTPases are known to induce cytoskeletal changes involved in cell adhesion and motility (50,51). The most well studied members of this family are Cdc42, Rac1, and RhoA, which induce filopodia, lamellipodia, and stress fibers, respectively (32,(52)(53)(54). IQGAP1 is a known regulator of Cdc42, and several IQGAP1 functions show dependence on activated, GTP-bound Cdc42. We were interested in determining the effect of activated Cdc42 on the interaction of IQGAP1 with PTP. To address this question, we added purified, constitutively active (CA) or dominant negative (DN) Cdc42, Rac1 and RhoA to pull-down assays using the iPTPWT-⌬D2 and iPTPDA-⌬D2 constructs. Constitutively active mutants of the Rho GTPases have defective GTPase activity, keeping them in a GTP-bound  state (53). Dominant negative constructs of the Rho GTPases preferentially bind GDP over GTP and inhibit endogenous GTPases by titrating out guanine nucleotide exchange factors required for their activation (52,55). The data in Fig. 3 show that in the presence of CA-Cdc42 there is a striking increase in the binding of IQGAP1 to both iPTPWT-⌬D2 and iPTPDA-⌬D2 fusion proteins, 2.4-fold (average of 2.4, n ϭ 3) and 2.6-fold (average of 2.9, n ϭ 3), respectively. CA-Rac1 also caused an increase in the PTP/IQGAP1 interaction but to a lesser extent, 1.6fold (average of 2.0, n ϭ 2) and 1.3-fold (average of 1.5, n ϭ 3), respectively. No significant change in binding is seen with CA-RhoA, consistent with the lack of interaction between IQGAP1 and RhoA (19). Similarly, none of the DN constructs significantly altered the PTP/ IQGAP1 interaction (Fig. 3). The association of the known PTP-interacting protein, RACK1 (56), is not altered by the addition of the Rho GTPases (Fig. 3). These data indicate that CA-Cdc42 and CA-Rac1 association with IQGAP1 enhances its interaction with PTP. These data also suggest that PTP may regulate Cdc42-and/or Rac1-dependent functions of IQGAP1. Fig. 3D shows that CA-Cdc42 causes an increase in IQGAP1 binding to the full-length intracellular PTP constructs similar to that seen with the truncated PTP constructs, demonstrating that the increase in binding is not unique to the truncated constructs.
PTP and IQGAP1 Bind Directly-To gain additional information regarding the PTP-IQGAP1 interaction, we constructed a GST-PTP fusion protein (iPTP-JMD) containing amino acid residues 765-958, which includes the juxtamembrane domain and the first 35 residues of phosphatase domain one (Fig. 1). This construct was used in pull downs from A549 cells. Data shown in Fig. 4A indicate the iPTP-JMD construct binds IQGAP1, indicating that the IQGAP1 binding site on PTP lies within amino acid residues 765-958. As expected, GST alone was unable to pull down IQGAP1.
To address whether the PTP-IQGAP1 interaction is direct, we purified His-tagged IQGAP1 from Sf21 insect cells and performed pulldown experiments using purified, bacterially expressed GST-iPTP-JMD or GST alone. Data in Fig. 4B show that GST-iPTP-JMD binds directly with His-IQGAP1. GST alone was unable to pull down IQGAP1. These data suggest IQGAP1 directly interacts with PTP, and this interaction is mediated by amino acid residues 765-958 of PTP. iPTPWT was included as a positive control. To determine whether we could recapitulate our previous data with CA-Cdc42 using purified proteins, we performed pull-down experiments using purified His-IQ-GAP1 and purified iPTPWT in the presence of CA-Cdc42. Similar to the data presented in Fig. 3, CA-Cdc42 enhances the interaction between purified PTP and purified IQGAP1 2-fold. These data suggest the effect of CA-Cdc42 is due to a conformational change in either PTP or IQGAP1 upon Cdc42 binding (Fig. 4B).
Endogenous PTP and IQGAP1 Interact in Vivo-To further characterize the PTP/IQGAP1 interaction, we examined the localization of endogenous PTP and IQGAP1 in A549 cells as well as their ability to co-immunoprecipitate. In A549 cells PTP accumulates at cell-cell contacts in high cell density cultures (Fig. 5) as described previously in other cell types (8,57). At low cell density, PTP stains diffusely in the cytoplasm of single cells and concentrates where cell-cell contacts have formed in adjacent cells (Fig. 5). IQGAP1 is known to associate with cortical actin filaments (19,29) and to localize at cell-cell contacts (24,58). In agreement with these data, IQGAP1 accumulates at cell-cell contacts in high cell density cultures of A549 cells (Fig. 5). At low cell density, IQGAP1 remains mostly cytoplasmic but concentrates in the lamellipodia (Fig. 5) as described in other cell types (19,58).
To get a better understanding of where PTP and IQGAP1 might associate, we did co-localization studies using A549 cells expressing Equal amounts of protein from A549 cell lysates were incubated with iPTP-JMD fusion protein or GST alone immobilized on glutathione-Sepharose. iPTP-JMD interacts with IQGAP1, whereas GST alone does not interact with IQGAP1 (A). iPTP-JMD, iPTPWT, or GST alone purified from bacteria was incubated with purified His-IQGAP1 isolated from Sf21 insect cells. Both iPTP-JMD and iPTPWT bind purified His-IQGAP1 as detected by an anti-IQGAP1 monoclonal antibody. GST alone did not bind purified His-IQGAP1. iPTPWT and GST were incubated with purified His-IQGAP1 in the absence or presence of CA-Cdc42. The presence of CA-Cdc42 increased the binding of iPTPWT to purified His-IQGAP1. Values listed below the panel represent densitometry readings relative to the lane where no GTPase was added (B). GFP-tagged full-length PTP. When adjacent cells are just touching, PTP-GFP is concentrated at the tips of filopodia extensions making initial contacts (Fig. 6a), whereas IQGAP1 is concentrated in the lamellipodia (Fig. 6b). As cell contacts form, PTP-GFP accumulates in a broad band spanning the area of contact (Fig. 6d). At this point, a portion of the IQGAP1 redistributes to the nascent contact (Fig. 6e) and co-localizes with PTP-GFP (Fig. 6g). As suggested by immunocytochemistry of endogenous proteins (Fig. 5), PTP-GFP and IQGAP1 co-localize at cell-cell contacts (Fig. 6, g and k).
Based on our immunocytochemistry data, we performed immunoprecipitation studies on A549 cells at high cell density. Confluent A549 cells were treated with the DSP cross-linking reagent, and endogenous PTP was immunoprecipitated using an antibody to the extracellular domain of PTP (494). Fig. 7 shows PTP and IQGAP1 can be coimmunoprecipitated from A549 cells. PTP is proteolytically processed in the endoplasmic reticulum yielding two fragments of ϳ100 kDa each, which remain tightly associated via a non-covalent interaction (34). The intracellular fragment is detected in Fig. 7 as well as full-length, unprocessed PTP. Furthermore, the IQGAP1-interacting proteins N-cadherin, E-cadherin, and ␤-catenin are also detected in a complex that immunoprecipitates with PTP, suggesting PTP may regulate the cadherin/catenin complex via IQGAP1.
IQGAP1 Is Required for PTP-mediated Neurite Outgrowth-PTP is a permissive substrate for E8 chick nasal retinal ganglion cells (RGCs) (42,46). Bonhoeffer stripe assays were used to determine the effect of perturbing IQGAP1 on PTP-mediated crossing of nasal retinal ganglion cells. Stripe assays measure the preference of neurites to grow on different substrates such as laminin and PTP. Both laminin and PTP are permissive substrates for E8 chick nasal RGC neurites. As shown in Fig. 8A (Control), nasal RGC neurites freely cross between the laminin and PTP lanes. Addition of a membrane-permeable, TAT-tagged IQGAP1 competitive peptide corresponding to the Cdc42 and Rac1 binding site on IQGAP1 completely abolishes nasal RGC neurite crossing onto the PTP substrate without affecting growth on the laminin lanes, suggesting that IQGAP1 is required for PTP-mediated crossing of nasal RGC neurites (Fig. 8A, IQGAP1). This peptide, comprising amino acids 1054 -1077 of IQGAP1 (MK24), is known to compete the interaction of IQGAP1 with Cdc42 and Rac1 (47). Addition of a membrane-permeable, TAT-tagged scrambled peptide did not inhibit PTP-mediated crossing (Fig. 8A, Scrambled). Quantitation of the stripe assays was performed as described previously (45,46) and is shown in Fig. 8B. These data show that the IQGAP1 competitive peptide induces nasal RGC avoidance of the PTP substrate lanes, demonstrating that IQGAP1 is required for PTP-mediated neurite outgrowth.

DISCUSSION
Protein tyrosine phosphorylation is a key post-translational modification involved in the regulation of several cellular processes. Although several substrates and interacting proteins are known for tyrosine kinases, in comparison, few have been identified for tyrosine phosphatases, especially the RPTP family of enzymes. We have identified IQGAP1 as a novel PTP-interacting protein and propose that it is part of a PTP signaling pathway leading to changes in the cytoskeleton, cell adhesion, and neurite outgrowth.
Cellular functions known to be regulated by PTP include cadherinmediated cell adhesion (12), N-cadherin-mediated neurite outgrowth (13), and axon guidance (42,46,59). Similar to PTP, IQGAP1 has been shown to regulate cadherin-medited cell adhesion (23,24,60,61) and to induce neurite outgrowth (62), suggesting PTP and IQGAP1 share a common signaling pathway. In addition, both PTP homophilic bind-FIGURE 6. Co-localization of PTP-GFP and IQGAP1. Subconfluent A549 cells were infected with baculovirus expressing WT PTP-GFP. One day after infection immunocytochemistry was performed for IQGAP1. Phase images are shown of three different fields (d, h, and l). When cells initially contact one another PTP-GFP is concentrated at the tip of filopodial extentions (a), IQGAP1 is concentrated in lamellipodia (b), and there is little or no co-localization of PTP-GFP and IQGAP (c). As contacts begin to form, PTP-GFP is localized in a broad band spanning the contact site (e), and a portion of IQGAP1 redistributes to this site (f) and co-localizes with PTP-GFP (g, yellow). Once a cell contact has been established PTP-GFP (i) and IQGAP1 (j) are both concentrated at the cell-cell contact (k, yellow). FIGURE 7. PTP and IQGAP1 interact in vivo. A549 cells were grown to 90% confluence and treated with 1 mM DSP cross-linking reagent. Immunoprecipitations were performed using a polyclonal antibody to an extracellular epitope of PTP or non-immune rabbit serum (NIRS). Proteins were resolved by SDS-PAGE and immunoblotted for the presence of the indicated proteins. IQGAP1, N-cadherin (130 kDa), E-cadherin (120 kDa), and ␤-catenin (92 kDa) specifically co-immunoprecipitate with PTP. The PTP (200 and 100 kDa) immunoblot is a control for PTP immunoprecipitation.
ing (44) and IQGAP1 overexpression (20) induce a filopodial dominant phenotype dependent on active Cdc42. Many IQGAP1 functions are dependent on active Cdc42, including cross-linking of actin filaments (29,33), capture and stabilization of microtubule plus ends via binding to CLIP-170 (27), stabilization of E-cadherin-mediated adhesion junctions via release of ␤-catenin (60), promotion of cell migration (63), and the inhibition of E-cadherin endocytosis (61). Our data show activated Cdc42 enhances the binding of IQGAP1 to PTP suggesting PTP can regulate this complex and, therefore, potentially IQGAP1 activities.
Although we have shown IQGAP1 is a PTP-interacting protein, we have no evidence that IQGAP1 is a substrate of PTP. Recently, it was reported that IQGAP1 is heavily phosphorylated in MCF-7 cells grown in the presence of serum (62). Phosphorylation is enhanced in response to phorbol 12-myristate 13-acetate stimulation, and Ser-1443 was identified as the predominant phosphorylated residue (62). Two groups independently identified the potential kinase as the novel PKC, PKC⑀ (62,64). Li et al. (62) demonstrated that IQGAP1-induced neurite outgrowth in N1E-115 cells is augmented by phosphorylation at Ser-1443. Phosphorylation of Ser-1443 has also been shown to influence the binding of nucleotide-depleted Cdc42 in vitro by opening up the C terminus of IQGAP1 and revealing novel Cdc42 binding sites (64). The implications of IQGAP1 binding to nucleotide-depleted Cdc42 are currently unclear. The fact that Ser-1443 enhances neurite outgrowth induced by IQGAP1 is of interest because both PTP-mediated outgrowth and axon guidance of retinal ganglion cells involve the novel PKC, PKC␦ (46,65). We propose that PKC and IQGAP1 are part of a common signaling pathway initiated by PTP homophilic binding. Future studies in our laboratory will explore this possibility.
Regardless of whether IQGAP1 is a tyrosine-phosphorylated substrate for PTP, PTP could modify IQGAP1 activity by regulating IQGAP1 binding partners. This could be accomplished by conformational changes induced by the binding of IQGAP1 to PTP. We have provided evidence that active Cdc42 enhances the binding of IQGAP1 to PTP independent of phosphatase activity (Fig. 4). Alternatively, several IQGAP1-associated proteins are potential tyrosine-phosphorylated substrates for PTP such as classical cadherins, ␤-catenin and Cdc42. Changes in the tyrosine phosphorylation status of these proteins could alter their ability to interact with IQGAP1.
Immunocytochemistry data predict that PTP and IQGAP1 interact initially at nascent cell-cell contacts. It is intriguing to postulate that IQGAP1, downstream of PTP homophilic binding, facilitates changes in the actin cytoskeleton leading to cell adhesion and adherens junction formation. This hypothesis is supported by reports indicating filopodia extensions initiate cell-cell adhesion (66). This scenario is applicable to neurite outgrowth and axon guidance as well as cell adhesion, since all of these processes involve sensing adjacent cells and responding to the environment. This idea is supported by the data presented in Fig. 8, which clearly show IQGAP1 is involved in PTP-mediated neurite outgrowth of nasal RGC neurons.