Cdc42-interacting protein 4 mediates binding of the Wiskott-Aldrich syndrome protein to microtubules.

The Wiskott-Aldrich syndrome is an inherited X-linked immunodeficiency characterized by thrombocytopenia, eczema, and a tendency toward lymphoid malignancy. Lymphocytes from affected individuals have cytoskeletal abnormalities, and monocytes show impaired motility. The Wiskott-Aldrich syndrome protein (WASP) is a multi-domain protein involved in cytoskeletal organization. In a two-hybrid screen, we identified the protein Cdc42-interacting protein 4 (CIP4) as a WASP interactor. CIP4, like WASP, is a Cdc42 effector protein involved in cytoskeletal organization. We found that the WASP-CIP4 interaction is mediated by the binding of the Src homology 3 domain of CIP4 to the proline-rich segment of WASP. Cdc42 was not required for this interaction. Co-expression of CIP4 and green fluorescent protein-WASP in COS-7 cells led to the association of WASP with microtubules. In vitro experiments showed that CIP4 binds to microtubules via its NH(2) terminus. The region of CIP4 responsible for binding to active Cdc42 was localized to amino acids 383-417, and the mutation I398S abrogated binding. Deletion of the Cdc42-binding domain of CIP4 did not affect the colocalization of WASP with microtubules in vivo. We conclude that CIP4 can mediate the association of WASP with microtubules. This may facilitate transport of WASP to sites of substrate adhesion in hematopoietic cells.

The Wiskott-Aldrich syndrome is an inherited X-linked immunodeficiency with associated thrombocytopenia, eczema, and a tendency toward development of malignancy of the lymphoreticular system (1,2). Affected individuals have lymphocytes that respond poorly to certain stimulants in vitro (3,4) and show abnormalities of cytoskeletal structure (5)(6)(7)(8). Impaired monocyte motility is also observed (9 -11). The gene mutated in Wiskott-Aldrich syndrome was cloned in 1994, and was shown to encode a proline-rich protein of 502 amino acids (12)(13)(14). Although the function of WASP 1 was not immediately apparent from knowledge of its primary structure, a variety of interactions have been discovered showing that WASP plays a role in cytoskeletal organization. The NH 2 terminus of WASP binds to WASP-interacting protein (WIP), a protein with profilin binding and actin binding motifs, overexpression of which causes changes in cytoskeletal structure (15). Missense mutations known to cause Wiskott-Aldrich syndrome impair the WASP-WIP interaction (16). A GBD/CRIB motif for binding the active forms of the small GTPases Rac and Cdc42, proteins that are involved in control of cytoskeletal organization, is present in WASP (18 -20). The proline-rich segment of WASP, between amino acids 310 and 420, interacts with a number of proteins containing SH3 domains, many of which are involved in the regulation of cytoskeletal structure. Among the SH3 proteins known to interact with WASP are the adaptor proteins Nck (21,22) and Grb2 (23,24), Src family kinases (23,(25)(26)(27), phospholipase C␥ (23,25), and Tec family kinases (26,28). Motifs in the COOH terminus of WASP show homology to the actin-binding proteins verprolin (verprolin homology domain, amino acids 430 -446) and cofilin (cofilin homology domain, amino acids 469 -487) (18,29). WASP has been shown to interact directly with actin through the verprolin homology domain (30). A WASP homolog expressed in neural and other tissues, N-WASP, has the ability to depolymerize actin filaments in vitro (28,29). WASP, N-WASP, and the related proteins Scar1 in human and Bee1/Las17 in yeast interact with the Arp2/3 complex, a key regulator of actin polymerization (31,32). WASP-coated microspheres exhibit cytoplasmic actin-based motility, dependent on the presence of the Arp2/3 complex (33). Taken together, these studies suggest that WASP is a mediator of Cdc42/Rac signaling, and has direct effects on the actin cytoskeleton via actin-binding domains and activation of the Arp2/3 complex. SH3 protein interactions may play a role in localization of WASP and/or transmission of extracellular signals to and through WASP.
We performed a yeast two-hybrid screen to look for novel WASP-interacting proteins. One WASP-interacting clone encoded a portion of the protein CIP4. This protein was identified previously as interacting with the active form of Cdc42 (34). CIP4 is a 545-amino acid protein, widely expressed, that has an NH 2 -terminal domain of unknown function with homology to the NH 2 termini of the non-receptor tyrosine kinase Fer and the Fes/Fps proto-oncogene, which was termed the FCH domain. The central region of the protein (amino acids 293-481) was shown to interact with Cdc42, although it does not contain a GBD/CRIB motif. An SH3 domain is found in the COOH terminus of CIP4. We performed a series of studies exploring the mechanism of WASP/CIP4 interaction, and the effects of co-expression of CIP4 on WASP distribution and actin cytoskeletal structure in COS-7 cells.

EXPERIMENTAL PROCEDURES
Two-hybrid Screen-We used the Interaction Trap two-hybrid screen (35). Yeast and vectors were obtained from Roger Brent (University of Massachusetts, Boston, MA). An human T-cell lymphotrophic virus type I-transformed T-cell line cDNA library for this system was purchased from CLONTECH (Palo Alto, CA). The screen was performed as described (16,36) using full-length WASP as bait in the LexA system. Additional methods were obtained from material supplied with the library. The specificity of interaction of clones expressing the interaction phenotype was tested with a bait plasmid containing an open reading frame of Drosophila bicoid protein that was included with the materials obtained from Dr. R. Brent, and with a bait plasmid containing an open reading frame of human IGF1-R␤ cytoplasmic domain that was a gift from Bhakta Dey (NCI Metabolism Branch, Bethesda, MD).
Expression Constructs-A cDNA encoding full-length WASP was cloned into pQBI-25 (Quantum Biotechnologies Inc., Laval, Quebec, Canada) in order to produce GFP-WASP, and into pCR2 (Invitrogen, Carlsbad, CA) for in vitro translation. GST-Cdc42 expression constructs were generously provided by Dr. Anne Ridley (Ludwig Institute for Cancer Research, University College, London, United Kingdom). A full-length CIP4 cDNA PCR product from a B-cell cDNA library (gift from Dr. Colin Duckett, National Institutes of Health, Bethesda, MD) was cloned into pCR2.1 vector (Invitrogen). This CIP4 cDNA was used as a template to clone CIP4 and CIP4 fragments into pCR2.1 for in vitro translation and into pGEX4T-2 for expression of GST-tagged proteins. CIP4 cDNA was cloned into a derivative of the mammalian expression vector pRK5 (PharMingen, San Diego, CA) that was modified to include an NH 2 -terminal myc epitope tag (pRK5-myc) in order to express myctagged CIP4 in mammalian cells. Point mutations were produced by whole vector PCR with primers containing the desired base change using the QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA). Internal deletion mutants of CIP4 and WASP were generated by whole vector PCR using the QuikChange TM kit, with primers flanking the sites of the intended deletion. T4 DNA ligase was then used for blunt end ligation of the PCR products. All constructs were sequenced to confirm the presence of the intended mutations.
In Vitro Binding Assays-In vitro binding assays of WASP with CIP4, and WASP or CIP4 with Cdc42 were performed using the GST pull-down technique. In these assays, an in vitro translated, [ 35 S]methionine-radiolabeled protein was incubated with a potential binding partner in the form of an unlabeled GST fusion protein bound to beads of glutathione-agarose. Specifically, GST fusion protein (10 g) bound to glutathione-agarose was incubated at 4°C for 2 h with 10 l of the in vitro translated protein in PBS, 1% Triton X-100 or, in the case of loaded Cdc42, with 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, 1% Triton X-100, 5 mM MgCl 2 , 0.2% bovine serum albumin in a total volume of 100 l, then washed three times with the same buffer. Proteins adhering to the beads (labeled protein and unlabeled GST protein) were released from the beads, and any interaction disrupted by boiling in SDS gel loading buffer containing 2% 2-mercaptoethanol. This protein mixture was analyzed on an SDS-polyacrylamide gel and the gel exposed to x-ray film. Any labeled protein that had bound to the GST fusion protein will appear as a band on the film.
Radiolabeled WASP or CIP4 was prepared by in vitro translation using the TnT ® coupled transcription-translation system (Promega, Madison WI) in the presence of [ 35 S]methionine. GST fusion proteins were expressed using the protease-deficient Escherichia coli bacterial strain BL21. Cells from induced cultures were collected, resuspended in 50 l of ice-cold PBS/ml of culture and disrupted by sonication. Triton X-100 was added to a final concentration of 1% and the lysate was mixed for 30 min. The lysate was centrifuged at 12,000 ϫ g for 10 min, diluted 5-fold with 1% Triton X-100/PBS, and reduced glutathioneagarose (Sigma) added. After binding of GST fusion proteins, the glutathione-agarose was washed and stored at 4°C. The quantity and purity of the GST fusion proteins was assessed by SDS-PAGE, followed by Coomassie Blue staining. GST-Cdc42 proteins were loaded with the non-hydrolyzable GTP analog GTP␥S immediately before use by adding equal volumes of the purified fusion protein bound to glutathioneagarose in PBS and 100 mM Tris-HCl, pH 7.5, 15 mM EDTA, 1 mg/ml bovine serum albumin, 2 mM dithiothreitol, 1 mM GTP␥S or GDP␤S at 37°C for 30 min and fixed with 12.5 mM MgCl 2 .
Antibodies-Affinity-purified anti-CIP4 antibody was prepared as follows. A PCR product encoding CIP4 residues 118 -481 was cloned into pFLAG-ATS vector (Kodak-IBI) and recombinant FLAG-CIP4 expressed in E. coli BL-21. The FLAG-CIP4 was purified by affinity chromatography on anti-FLAG M2 agarose according to manufacturer's directions. Rabbits were immunized with 100 g of purified FLAG-CIP4 initially, followed by 10 g/week for 10 weeks. Preimmune and immune sera were collected and assayed by enzyme-linked immunosorbent assay using GST or GST-CIP4 as antigens. Anti-CIP4 antibody was purified from the immune sera by affinity chromatography on GST-CIP4 bound to cyanogen bromide-activated Sepharose (Amersham Pharmacia Biotech).
Rabbit anti-tubulin antibody, mouse anti-tubulin monoclonal antibody DM1A, and phalloidin-TRITC were purchased from Sigma. Goat anti-rabbit IgG Alexa TM 594 and goat anti-mouse IgG Alexa TM 488 were purchased from Molecular Probes. Rabbit anti-myc antibody was purchased from Upstate Biotechnology, and mouse monoclonal anti-myc was purchased from Invitrogen. Monoclonal anti-WASP was obtained as described (37).
Cell Transfection-Plasmid DNA was prepared using the Qiagen Maxi Prep kit. COS-7 cells were maintained in RPMI 1640 medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and 2 mM L-glutamine. Cells were transfected by electroporation.
Co-immunoprecipitation-COS-7 cells transfected with GFP-WASP and myc-CIP4 were lysed at 1.2 ϫ 10 7 cells/ml in 1% digitonin, 50 mM Tris-Cl, pH 7.4, 150 mM NaCl supplemented with Complete ® proteinase inhibitor mixture (Sigma) at 4°C for 20 min and centrifuged at 16,000 ϫ g for 10 min. The clarified lysate was diluted 1:3 with washing buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl with proteinase inhibitor), and affinity-purified rabbit anti-CIP4 or preimmune serum was added to a final concentration of 10 g/ml or 5 l/ml, respectively. After incubation at 4°C for 2 h, protein A-Sepharose (50 l of a 50% slurry in wash buffer) was added, and the suspension rocked at 4°C for 1 h. The pellet was washed three times with washing buffer and bound proteins analyzed by SDS-PAGE, followed by Western blot using either monoclonal anti-myc or monoclonal anti-WASP.
In Vitro Microtubule Binding Assays-Microtubules were obtained by incubating 5 mg/ml tubulin (Molecular Probes, Eugene, OR) in G-PEM buffer (100 mM K-PIPES, 1 mM EGTA, 1 mM MgSO 4 , 1 mM GTP, pH 6.8) with 30% glycerol at 35°C for 10 min. Microtubules were stabilized by 10 M paclitaxel (Molecular Probes) as instructed by the manufacturer. The microtubule binding experiments were performed as described (38). Briefly, 50 l of in vitro translated, [ 35 S]methioninelabeled CIP4 or CIP4 mutants was diluted in 200 l of PB (80 mM K-PIPES, pH 6.9, 1 mM EGTA, 1 mM CaCl 2 ) containing Complete ® proteinase inhibitor mixture (Sigma). The mixture was spun at 50,000 ϫ g for 1 h at 4°C in a 70.1Ti rotor (Beckman). 100 l of supernatant was incubated with or without 30 g of microtubules for 30 min at 37°C in the presence of 10 M paclitaxel. The microtubule and CIP4 mixtures were layered over 1 ml of 15% sucrose in PB, and spun for 30 min at 30,000 ϫ g in a SW55 rotor (Beckman). Both supernatants (above the sucrose) and pellets were analyzed for the presence of CIP4 and its mutants by 4 -20% SDS-PAGE, followed by autoradiography. The gels were then stained with Coomassie Blue to confirm that an equal mass of microtubules was loaded on the gel in each sample.
Immunofluorescence-COS-7 cells were transfected with pQBI25-GFP-WASP in the presence or absence of pRK5-myc-CIP4 by electroporation. Cells were plated on coverslips in 24-well culture plates. 24 -48 h after transfection, cells were fixed and permeabilized by Cytoperm/Fix kit (PharMingen). Primary human macrophages were prepared and cultured for 7 days as described (39). To visualize microtubules, cells were fixed and permeabilized by 4% formaldehyde, 0.1% Triton X-100, 80 mM K-PIPES, 1 mM EGTA, 1 mM MgSO 4 , 30% glycerol at room temperature for 30 min (40) or by 4% paraformaldehyde, 0.3% Triton X-100, 0.05% glutaraldehyde in cytoskeleton buffer (10 mM PIPES, pH 7.0, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl 2 , 100 g/ml streptomycin) at room temperature for 15 min. Both methods gave similar results. After staining, coverslips were mounted on glass slides by FluorSave ® reagent (Calbiochem). Confocal laser scanning microscopy was performed using an Olympus confocal laser scanning microscope equipped with argon and krypton ion lasers. Series of optical sections through the cells were collected. Images were assembled using Adobe PhotoShop software.

RESULTS
WASP Interacts with CIP4 -Several unique WASP-interacting clones were identified in the two-hybrid screen. One clone, our number 4 -2, encoded a portion of the protein CIP4, a 545 amino acid protein that previously had been identified as an SH3-domain containing protein with partial homology to the non-receptor kinase Fer that interacts with the active form of Cdc42 (33). The nucleic acid sequence of clone 4 -2 was identical to bases 390 -1949 of the published CIP4 sequence (Gen-Bank accession number AJ000414) with the exception of a single C insertion at 1911 in the 3Ј-untranslated region. The cDNA encoded CIP4 amino acids 118 -545. The interaction was specific to the WASP-LexA bait protein; no interaction of CIP4 with either Drosophila bicoid protein or the cytoplasmic domain of IGF1-R␤ was detected in the two-hybrid system (data not shown). The region of WASP responsible for interaction with CIP4 was mapped by deletion mutation analysis in the two-hybrid system. This showed that CIP4 interacted with a WASP construct retaining amino acids 1-442, but failed to interact with a WASP construct retaining amino acids 1-201 (data not shown). Since the WASP 1-442 has both the GBD (230 -260) and the proline-rich region (310 -420), two mechanisms of interaction were thought possible. First, the CIP4 SH3 domain might bind directly to WASP in the proline-rich region. Alternatively, since the yeast homolog of Cdc42 is similar to the human protein, the binding of endogenous yeast Cdc42 to WASP and CIP4 simultaneously could contribute to the interaction. We therefore performed an in vitro binding assay to explore the mechanisms of CIP4/WASP interaction.
The interaction between radiolabeled, in vitro translated WASP of different lengths and GST-CIP4 full-length and deletion mutants is shown in Fig. 1. Full-length CIP4 (GST-CIP4 wild type (WT)) bound strongly to WASP mutants retaining amino acids 1-442 and 1-502, which contain both the polyproline region and the GBD. WASP mutants lacking the polyproline region (1-302 and 1-201) bound CIP4 weakly, whether or not the WASP GBD was present (1-302) or absent (1-201). This indicates that the proline-rich region of WASP is necessary for effective CIP4 binding. However, full-length WASP bound weakly to GST-CIP4 SH3, which contains the isolated CIP4 SH3 domain. WASP mutant 1-442, which retains the polyproline region but lacks the COOH-terminal actin binding region, bound GST-CIP4 SH3 strongly. This difference may be due to a WASP intramolecular interaction between COOHterminal acidic residues and basic residues just NH 2 -terminal of the GBD, which may render the proline-rich region less available for binding to the CIP4 SH3 domain (41,42). WASP mutants lacking the polyproline region (1-302 and 1-201) failed to bind the CIP4 SH3 domain. A CIP4 mutant lacking the SH3 domain (GST-CIP4 118 -481) bound all WASP mutants weakly, if at all. These results demonstrate that the WASP-CIP4 interaction is mediated chiefly by the binding of the CIP4 SH3 domain to the polyproline region of WASP, and that the presence or absence of the WASP GBD has no influence on this binding.
To further establish that the CIP4-WASP interaction can occur independently of Cdc42 binding, the binding of in vitro translated radiolabeled WASP to both CIP4 and the constitutively active Cdc42 mutant V12 was examined. As shown in Fig. 2, a deletion mutant of WASP specifically lacking the GBD retains the ability to bind CIP4, but Cdc42 binding is lost. This shows that CIP4 can bind to WASP in the absence of Cdc42 binding.
To further investigate the CIP4-WASP interaction, COS-7 cells were co-transfected with GFP-WASP and myc-tagged CIP4. GFP-WASP co-immunoprecipitated with CIP4 as shown in Fig. 3. The CIP4 118 -481 mutant did not show binding to WASP, again showing the necessity of the CIP4 SH3 domain. When expressed alone, the CIP4 SH3 domain localized entirely in the cell nucleus as detected by immunofluorescence (data not shown), and was not tested for interaction with WASP by co-immunoprecipitation.
Identification of CIP4 Amino Acids Critical for Cdc42 Binding-The primary sequence of CIP4 does not contain the GBD/ CRIB motif characteristic of the binding site of Cdc42. We therefore tested a series of CIP4 deletion mutants for the ability to interact with the active form of Cdc42. Full-length in vitro translated CIP4 interacted specifically with GST-Cdc42-GTP␥S or with the constitutively active mutant Cdc42-V12-GTP␥S but not to the GDP␤S-loaded forms (not shown). The dominant negative Cdc42 mutant N17 failed to bind CIP4 regardless of GTP or GDP loading (not shown). As shown in Fig. 4, full-length CIP4 (1-545) or CIP4 constructs retaining amino acids 1-417 and 1-423 reacted with Cdc42, but CIP4 1-383 or 1-407 did not. A series of substitution mutations in this region of the CIP4 construct 1-417 revealed that the mutation I398S abrogated binding of CIP4 to active Cdc42 (Fig.   FIG. 5. Effect of CIP4 on WASP colocalization with F-actin. COS-7 cells were transfected with GFP-WASP in the absence (a-c) or presence (d-f) of CIP4. 24 to 48 h after transfection, cells were fixed and permeabilized. WASP distribution was visualized by GFP fluorescence (a and d). F-actin distribution was detected by phalloidin-TRITC staining (b and e). The GFP-WASP and F-actin images were superimposed for assessment of colocalization (c and f).

FIG. 6. Effect of CIP4 on WASP colocalization with microtubules. COS-7 cells were transfected with GFP-WASP in the absence (a-c) or presence (d-f) of CIP4. 24 -48 h after transfection, cells were fixed and permeabilized. WASP distribution was visualized by GFP fluorescence (a and d).
Microtubule staining was performed by rabbit anti-tubulin, followed by goat anti-rabbit Alexa TM 594 (b and e). The GFP-WASP and microtubule images were superimposed for assessment of colocalization (c and f). 4B). We conclude that the CIP4 amino acids 383-417 and the amino acid Ile-398 are critical for Cdc42 binding to CIP4.
CIP4 Changes WASP Colocalization from F-actin to Microtubules-Wiskott-Aldrich syndrome patients show structural abnormalities of platelets and peripheral lymphocytes, and decreased density of surface microvilli (3)(4)(5)(6)(7)(8)(9)(10)(11). Recent studies suggest a role for WASP in actin polymerization since overexpression of WASP induces the formation of actin-containing clusters (18). In order to test the role of CIP4 on WASP-induced cytoskeleton changes, WASP with GFP at the COOH terminus was transfected into COS-7 cells in the absence or presence of myc-tagged CIP4. In the absence of CIP4, the colocalization of GFP-WASP with F-actin was observed in fixed cells stained with phalloidin-TRITC. As reported by others (18), 40 -60% of transfected cells exhibited dot or cluster distribution of GFP-WASP in the perinuclear region, accompanied by the disappearance of stress fibers (Fig. 5, a-c). The overexpressed GFP-WASP protein colocalized in clusters with F-actin (Fig. 5, a-c).
In cells transfected with myc-CIP4 alone, a general decrease in cellular F-actin was observed with a few stress fibers and subcortical F-actin remaining (data not shown). This has been reported by others (33). When GFP-WASP was cotransfected with myc-CIP4, no dot or cluster structures of GFP-WASP and F-actin were seen (Fig. 5d). F-actin in co-transfected cells did not co-localize with GFP-WASP (Fig. 5, d-f), and assumed a pattern similar to that seen in cells transfected with myc-CIP4 alone. About 40% of the cells showed a diffuse cytoplasmic distribution of GFP-WASP, and the other 60% showed GFP-WASP in cytoplasmic linear structures reminiscent of a microtubular pattern. This linear pattern was not observed in cells transfected with GFP-WASP alone. We then analyzed the relationship between the linearly distributed GFP-WASP and microtubules by staining the cells with anti-tubulin antibody. Without cotransfection with CIP4, GFP-WASP-expressing cells had a normal microtubule distribution, with microtubules radiating from a perinuclear microtubule organizing center to the periphery (Fig. 6b). No distortion of microtubule structure was seen in cells overexpressing GFP-WASP alone, and there was no correlation between GFP-WASP distribution and microtubules (Fig. 6, a-c). In cells co-expressing GFP-WASP and CIP4, however, a linear distribution of GFP-WASP is seen that stained with anti-tubulin antibody (Fig. 6, d-f). Microtubules in WASPϩCIP4 cotransfected cells appeared either thicker than normal or in networks of shorter stretches of microtubules. Staining for CIP4 in cells with or without WASP revealed a cytoplasmic distribution of CIP4 with some membrane association; strong localization of CIP4 to microtubules was not observed in cells overexpressing CIP4.
CIP4 NH 2 Terminus Binds Microtubules in Vitro-We tested the ability of CIP4 and CIP4 deletion constructs to bind to microtubules by an in vitro co-sedimentation experiment. In vitro translated, [ 35 S]methionine-labeled CIP4 or CIP4 deletion mutants were incubated with or without paclitaxel-polymerized microtubules. Microtubules were then pelleted by ultracentrifugation. COOH-terminal deletion mutants of CIP4 retaining amino acids 1-118 and 1-290, as well as full-length CIP4, cosedimented with microtubules as shown in Fig. 7. NH 2 -terminal deletion mutants CIP4 291-481 and CIP4 393-481 did not cosediment with microtubules (Fig. 7). These results indicate that the NH 2 terminus of CIP4 mediates microtubule binding. CIP4 1-118 and CIP4 1-290 were more completely removed from the supernatant by microtubules than full-length CIP4, suggesting that the NH 2 terminus of CIP4 alone binds to microtubules more strongly than fulllength CIP4. Additional experiments showed that CIP4 1-417 and ⌬383-481 (lacking the binding site for Cdc42) also bound to microtubules (data not shown). [ 35 S]Methionine-labeled CIP4 (full-length or deletion mutants) was prepared by in vitro translation. Radiolabeled CIP4 retaining the amino acids listed at the right of the figure was incubated with (ϩ) or without (Ϫ) paclitaxel stabilized-microtubules. Supernatants (s) and pellets (p) were collected after ultracentrifugation over a 15% sucrose cushion and analyzed for presence of radiolabeled CIP4 by SDS-PAGE and autoradiography.

FIG. 8. Co-localization of CIP4 with microtubules in untransfected cells.
COS-7 (a-c) and primary human macrophages (d-f) were cultured and fixed for staining as described under "Experimental Procedures." Cells were stained with mouse monoclonal anti-tubulin, followed by goat anti-mouse IgG Alexa TM 488 (a and d), and with affinity-purified anti-CIP4 followed by goat anti-rabbit IgG Alexa TM 594 (b and e). The tubulin and CIP4 images were superimposed for assessment of colocalization (c and f).
CIP4 expression on WASP distribution suggest that CIP4 should co-localize with microtubules. We therefore examined the distribution of native CIP4 in untransfected COS-7 cells and in primary human macrophages. Fig. 8 shows a COS-7 cell and a differentiated human macrophage after 7 days in culture stained with anti-tubulin and anti-CIP4. There is strong association of CIP4 with microtubules in both cell types. We conclude that CIP4 does indeed associate with microtubules in vivo, and that the cytoplasmic distribution seen in the cells overexpressing CIP4 is the result of the large excess of protein present.
NH 2 Terminus of CIP4 Mediates WASP Localization with Microtubules-It has been reported that the F-actin and WASP cluster formation in WASP-overexpressing cells could be inhibited by the actin polymerization inhibitor cytochalasin D (18). A decrease of overall F-actin content in cells overexpressing CIP4 alone has also been observed by us (data not shown) and others (33). It is possible that in the presence of CIP4, actin polymerization is generally inhibited or actin depolymerization is enhanced. This may leave WASP in an F-actin-free status, allowing it to move to microtubules. The PH domain of Dbl family member Lfc has been reported to associate with tubulin (43). Since some authors have suggested that WASP has a PH domain in its NH 2 -terminal 108 amino acids (28), we asked if the microtubule association of WASP in the presence of CIP4 was mediated by direct binding of WASP to microtubules through its PH domain, or indirectly through the NH 2 -terminal microtubule binding site of CIP4. Myc-tagged CIP4 deletion mutants were expressed alone or with GFP-WASP in COS-7 cells (Fig.  9). These CIP4 constructs had been previously tested for WASP and microtubule binding by in vitro experiments, summarized in the upper half of Fig. 9. Full-length CIP4, which has the ability to bind to both WASP and microtubules, caused WASP to assume the linear distribution characteristic of microtubules (Fig. 9a). The NH 2 -terminal deletion mutant CIP4 118 -545, which lacks the microtubule binding region, binds to WASP in a co-immunoprecipitation experiment (see Fig. 3). When ex-pressed in COS-7 cells alone, CIP4 118 -545 showed the same effect in decreasing F-actin staining as the full-length protein (data not shown). When cotransfected with GFP-WASP, CIP4 118 -545 had the same effect as full-length CIP4 in abolishing the GFP-WASP clusters that are characteristic of WASP colocalization with F-actin (Fig. 9b). However, GFP-WASP did not form the linear structures characteristic of microtubules. This suggests that CIP4 microtubule binding is required for causing WASP to associate with microtubules. CIP4 1-417, which lacks the COOH-terminal SH3 domain, bound to microtubules but could not bind to WASP in vitro (data not shown). This mutant was unable to cause WASP to assume a microtubular distribution (Fig. 9c). CIP4⌬(383-481) is an internal deletion mutant of CIP4 lacking amino acids 383-481. We have shown by in vitro binding studies that CIP4 amino acids between 383-417 were necessary for binding active Cdc42 (see Fig. 4). CIP4⌬(383-481) preserved both microtubule and WASP binding ability in vitro (data not shown) and caused WASP to assume the linear distribution characteristic of microtubules (Fig. 9d). These results strongly suggest that both the WASP-binding and microtubule-binding portions of CIP4 are required to cause the association of WASP with microtubules in co-transfected cells. Cdc42 binding by CIP4 did not seem to be necessary for CIP4-mediated localization of WASP to microtubules. DISCUSSION CIP4 was originally identified in a two-hybrid screen as a protein interacting with active Cdc42 and was found to have partial homology in the NH 2 -terminal region to the protooncogene non-receptor cytoplasmic kinase Fes/Fps and to the related kinase Fer (33). This region of homology was termed the FCH domain. The FCH domain is found in the NH 2 terminus of a variety of proteins from yeast to mammals (44,45). Although the function of this conserved domain is not known, we present evidence here that the FCH domain in CIP4 mediates its binding to microtubules. We also report that CIP4 FIG. 9. The amino terminus of CIP4 mediates association of WASP with microtubules. Top, the FCH and SH3 domains of CIP4 are diagrammed at the top. Full-length CIP4 and CIP4 mutant constructs used in the co-transfection experiments are also shown. Binding of CIP4 and CIP4 mutants to microtubules was tested by the in vitro microtubule binding assay as described under "Experimental Procedures." CIP4-WASP binding was assessed either by in vitro pull-down experiments or by coimmunoprecipitation. Bottom, COS-7 cells were cotransfected with GFP-WASP and myc-CIP4 or CIP4 mutants (panels a-d). Subcellular distribution of GFP-WASP was detected by GFP fluorescence. Coexpression and subcellular distribution of myc-tagged CIP4 and mutants was detected by rabbit anti-Myc antibody (Upstate Biotechnology) followed by goat anti-rabbit Alexa TM 594 as shown in the insets at the lower right corner of each picture. binds to the Wiskott-Aldrich syndrome protein, a hematopoietic restricted protein critical for development of normal immunity and platelet function. Co-expression of WASP and CIP4 in COS-7 cells led to WASP localization on microtubules. CIP4 appears to act as an adaptor that mediates the association of WASP with microtubules, by binding microtubules via its NH 2terminal FCH domain, and WASP via its COOH-terminal SH3 domain. In the overexpression system, Cdc42 binding was not important for this adaptor function, since deletion of the CIP4 region that binds Cdc42 had no effect on CIP4 mediated WASP association with microtubules. It is interesting, however, that both CIP4 and WASP bind active Cdc42, suggesting that they are both regulated by this molecule, which is an important regulator of cytoskeletal structure (46,47). Further experiments are needed to determine if Cdc42 activation has any influence on CIP4 adaptor function in more physiologic conditions.
Is there evidence that other FCH domain proteins bind microtubules and serve as adaptors similar to CIP4? The mammalian proteins PACSIN (protein kinase C and casein kinase 2 substrate in neurons) and PSTPIP (proline, serine, threonine phosphatase-interacting protein) are similar to CIP4 in that they have an NH 2 -terminal FCH domain and a COOH-terminal SH3 domain (44,48). PSTPIP was shown to bind WASP via its SH3 domain, and, when co-expressed in Chinese hamster ovary cells, caused redistribution of WASP from perinuclear F-actin clusters to a cytoplasmic filamentous network (49). The closely related protein PSTPIP2, which is very similar to PST-PIP except that it lacks the SH3 domain, was found to colocalize with PSTPIP in a similar cytoplasmic "meshwork-like structure" (50). Syndapin I, which is the rat homolog of the human PACSIN, binds the neuronal WASP-related protein N-WASP via its SH3 domain (51). PACSIN2, which is highly similar to PACSIN/syndapin I, when overexpressed in NIH3T3 fibroblasts, assumed a cytoplasmic reticular distribution that partly overlapped both the microtubule and actin filament networks (52). These studies suggest that the FCH domain proteins PSTPIP and PACSIN/syndapin I may also associate with microtubules, and that these proteins might serve as adaptors linking microtubules with proline-rich proteins via the SH3 domain.
Microtubules are dynamic cytoskeletal structures that contribute to cell shape and polarity, motility, intracellular transport, and signal transduction (for reviews, see Refs. 53 and 54). What cell function might be served by the association of WASP with microtubules, mediated by CIP4? WASP function is known to be required for normal monocyte (macrophage) motility (9 -11). Recently, WASP was shown to be an essential component of the macrophage podosome, which is the principal adhesive structure of this cell type (39). The osteoclast, derived from the same progenitor as the macrophage, also adheres to the substrate by podosomes. In the osteoclast, podosomes form at the cell periphery adjacent to the ends of microtubules, and if the microtubules are disrupted, the podosomes form in random patterns (55). This suggests that the microtubules may deliver podosome assembly components or signals to the sites where podosomes form. In fibroblasts, the principal adhesive structure is the focal adhesion. In these cells, the microtubules appear to deliver signals regulating the size and distribution of the focal adhesions (56,57). It may be that CIP4 is a component of a pathway directing WASP and possibly other proteins onto the microtubules for delivery to sites of adhesive structure assembly. Further experiments aimed at disrupting the activity of CIP4 in cells undergoing formation of adhesive structures will investigate this possibility.
In summary, we have shown that the two Cdc42 effector molecules CIP4 and WASP interact, and that CIP4 mediates the association of WASP with microtubules. This reinforces the emerging understanding of the cytoplasmic microtubule network as a dynamic surface which plays a role in facilitation of cell adhesion, signaling, and macromolecular assembly.