p116Rip is a novel filamentous actin-binding protein.

p116Rip is a ubiquitously expressed protein that was originally identified as a putative binding partner of RhoA in a yeast two-hybrid screen. Overexpression of p116Rip in neuroblastoma cells inhibits RhoA-mediated cell contraction induced by lysophosphatidic acid (LPA); so far, however, the function of p116Rip is unknown. Here we report that p116Rip localizes to filamentous actin (F-actin)-rich structures, including stress fibers and cortical microfilaments, in both serum-deprived and LPA-stimulated cells, with the N terminus (residues 1-382) dictating cytoskeletal localization. In addition, p116Rip is found in the nucleus. Direct interaction or colocalization with RhoA was not detected. We find that p116Rip binds tightly to F-actin (Kd approximately 0.5 microm) via its N-terminal region, while immunoprecipitation assays show that p116Rip is complexed to both F-actin and myosin-II. Purified p116Rip and the F-actin-binding region can bundle F-actin in vitro, as shown by electron microscopy. When overexpressed in NIH3T3 cells, p116Rip disrupts stress fibers and promotes formation of dendrite-like extensions through its N-terminal actin-binding domain; furthermore, overexpressed p116Rip inhibits growth factor-induced lamellipodia formation. Our results indicate that p116Rip is an F-actin-binding protein with in vitro bundling activity and in vivo capability of disassembling the actomyosin-based cytoskeleton.

more than 100 that bind polymeric F-actin, and many of them induce cross-linking or bundling of F-actin (10).
In our ongoing studies to delineate Rho signaling by the lipid growth factor lysophosphatidic acid (LPA) (11,12), we previously identified a ubiquitously expressed protein of 116 kDa, provisionally named p116 Rip , which binds relatively weakly to activated RhoA in a yeast two-hybrid assay (13). The p116 Rip sequence predicts several protein interaction domains, including at least one PH domain, two proline-rich stretches, and a C-terminal region predicted to form a coiled-coil domain. This suggests that p116 Rip may have a scaffolding role recruiting different proteins into a RhoA-regulated macromolecular complex. When overexpressed in N1E-115 neuroblastoma cells, p116 Rip promotes cell flattening and process extension and inhibits cytoskeletal contraction in response to LPA (13). The p116 Rip phenotype was reminiscent of what is observed after RhoA inactivation (using dominant-negative RhoA or C3 toxin), which led to the suggestion that p116 Rip may negatively regulate RhoA signaling (13). However, the function of p116 Rip remains unknown; importantly, no evidence that p116 Rip binds directly to RhoA in mammalian cells has been discovered (13).
In the present study, we set out to characterize p116 Rip in further detail. We find that p116 Rip , rather than directly binding to RhoA, interacts with F-actin via its N-terminal region and colocalizes with dynamic F-actin structures such as stress fibers, cortical microfilaments, filopodia, and lamellipodial ruffles. Furthermore, we show that p116 Rip induces bundling of F-actin in vitro, with the bundling activity residing in the N-terminal region. Yet overexpression of p116 Rip or its Nterminal actin-binding domain disrupts the actin cytoskeleton and thereby interferes with growth factor-induced contractility and lamellipodia formation. Our studies specify p116 Rip as a novel F-actin-binding protein and demonstrate that p116 Rip can affect, either directly or indirectly, the integrity of the actomyosin-based cytoskeleton.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-N1E-115 and COS-7 cells were grown in Dulbecco's modified Eagle's medium containing 7.5% fetal calf serum. NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium containing 10% newborn calf serum. N1E-115 cells were transiently transfected using the calcium phosphate precipitation method described previously (14). NIH3T3 cells were transfected using Lipo-fectAMINE Plus (Invitrogen), and serum was added to the cells after the transfection procedure as described by the manufacturer. COS-7 cells were transfected by the DEAE-dextran method as described previously (15).
Solubility Assay-Cells were lysed in ice-cold lysis buffer (0.1% Triton X-100, 50 mM Tris, pH 7.4, 150 mM NaCl, and 1 mM EDTA, supplemented with protease inhibitors) and were left on ice for 15 min. Lysates were centrifuged for 30 min (13,000 rpm; Eppendorf table centrifuge, 4°C). Pellet and supernatant fractions were collected, dissolved in sample buffer, and subjected to SDS-PAGE. Proteins were detected by Western blotting using the polyclonal anti-p116 Rip antibody (1:1000 dilution).
Expression and Purification of Recombinant Fusion Proteins-The E. coli strains BL21 or DH5␣ were transformed with plasmids encoding GST-NT, GST-CT, GST-⌬5, GST-⌬6, GST⌬7, GST⌬8, GST⌬9, or GST, respectively. Colonies were obtained and used to inoculate Luria broth/ ampicillin. Cultures were grown and isopropyl ␤-D-thiogalactoside was added overnight to induce expression of the fusion proteins when the cultures reached an OD between 0.4 and 0.6. Bacteria were harvested by centrifugation at 4000 ϫ g, resuspended in cold lysis buffer (50 mM Tris-HCl, pH 7.6, 50 mM NaCl, 5 mM MgCl 2 , 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride), and lysed by sonification followed by the addition of 0.5% Nonidet P-40. Lysates were cleared by centrifugation at 4000 ϫ g, and resulting supernatants were incubated with glutathione-Sepharose 4B beads (Amersham Biosciences). Beads with affinity-bound proteins were washed five times with lysis buffer, and bound proteins were eluted with 50 mM Tris-HCl, pH 8.0, containing 10 mM reduced glutathione. In some cases, GST was cleaved off by incubating beads with affinity-bound proteins with thrombin (Amersham Biosciences) according to manufacturer's protocol.
Purified full-length p116 Rip fused to GST or purified full-length p116 Rip fused to Myc were obtained by transfection of COS-7 cells with the constructs pMT2sm-FLp116 Rip -GST and pMT2sm-FLp116 Rip -myc, respectively. Cells were lysed in ice-cold lysis buffer (1% Nonidet P-40, 50 mM Tris, pH 7.4, 200 mM NaCl, 2.5 mM MgCl 2 , and 10% glycerol, supplemented with protease inhibitors). Lysates were cleared by centrifugation (13,000 rpm; 10 min). Further purification of the full-length GST fusion proteins occurred in the same manner as the purification of GST fusion proteins produced in bacteria. FLp116 Rip -myc fusion protein was purified using a column of monoclonal myc antibodies (9E10) chemically cross-linked to protein G-Sepharose beads (Amersham Biosciences). The protein was eluted off with 0.1 M glycine, pH 2.5, and fractions were collected in tubes containing 0.1 volume of 1 M Tris-HCl, pH 8.0, to make an end pH of 7.0. Eluted proteins were subjected to SDS-PAGE, followed by protein staining with Coomassie Blue to estimate the purity and concentration of the proteins in the fractions. Some of the proteins were concentrated using Centricon 10-kDa cutoff devices (Millipore). Protein concentration was also determined by the Bradford method using BSA as a standard. Purified proteins were stored in aliquots at Ϫ80°C in 10% glycerol.
F-actin Cosedimentation Assay-COS-7 cells were transfected with the indicated expression vectors and lysed for 48 h after transfection in ice-cold lysis buffer (1% Nonidet P-40, 50 mM Tris, pH 7.4, 200 mM NaCl, 2.5 mM MgCl 2 , and 10% glycerol, supplemented with protease inhibitors). Lysates were cleared by centrifugation (13,000 rpm; 10 min) and supernatant aliquots were run out on SDS-PAGE gels to check for expression (data not shown). 10-l aliquots were used in the in vitro actin-binding assay.
F-actin cosedimentation assays were performed according to the manufacturer's protocol (Cytoskeleton, Denver, CO). Briefly, prespun aliquots of COS-7 cell lysates (100,000 ϫ g, 30 min) or purified proteins GST-NTp116 Rip , GST, FLp116 Rip -GST, BSA, ␣-actinin (Cytoskeleton), or GST-⌬5p116 Rip ; GST-⌬6p116 Rip ; GST-⌬7p116 Rip ; GST-⌬8p116 Rip ; GST-⌬9p116 Rip were incubated for 1 h at room temperature with 40 g of pure actin filaments. The final concentration of F-actin was 18 M. Filaments were subsequently pelleted by centrifugation 100,000 ϫ g (Beckman airfuge). As a control for actin-independent sedimentation, the various proteins were also centrifuged under conditions in which filamentous actin was omitted from the mix. Cosedimenting proteins were resolved by SDS-PAGE and detected by either Coomassie Blue staining or by Western blot analysis using anti-p116 Rip antibodies, anti-GFP rabbit polyclonal antibodies (16), or an anti-actin mouse monoclonal antibody (Mab 1501R; Chemicon).
For quantitative analysis, a fixed concentration of FLp116 Rip -GST (1 M) was mixed with increasing amounts of F-actin (0 -3.5 M) in polymerization buffer and incubated at room temperature for 30 min. Proteins were centrifuged as above and total pellets and supernatants were loaded separately on SDS-polyacrylamide gels. Protein bands were detected by Coomassie Blue staining and were scanned and quantified using the software program TINA. The amount of p116 Rip bound to different concentrations of F-actin was fit to a single rectangular hyperbola using Prism (ver. 3; GraphPad Software, San Diego, CA). In all cases, entire pellet and supernatant fractions were loaded separately on SDS-polyacrylamide gels and detected by either Coomassie Blue staining or by Western blot (above).
Electron Microscopy-To test for bundling activity, actin filaments (5 M) were incubated for 30 min with purified proteins GST, ␣-actinin , and CT-p116 Rip (0.5 M) at room temperature. Samples were absorbed on to glowdischarged carbon-coated formvar film on a copper grid and negatively stained with 1% uranyl acetate and examined with a Philips CM10 electron microscope.
Metabolic Labeling-N1E-115 cells were incubated in methionine/ cysteine-free media for 30 min and labeled for 4 h with medium containing [ 35 S]methionine/cysteine (200 Ci/ml; Amersham Biosciences). Labeled medium was aspirated and cells were washed once in ice-cold PBS. Cells were scraped in ice-cold lysis buffer (1% Nonidet P-40, 50 mM Tris pH 7.4, 200 mM NaCl, 2.5 mM MgCl 2 , and 10% glycerol, supplemented with protease inhibitors). Lysates were cleared by centrifugation (13,000 rpm; 10 min) and supernatants were tumbled with protein A-Sepharose beads precoupled to either preimmune rabbit serum, anti-p116 Rip antibodies (13), or anti-myosin IIA antibodies (BTI, Oklahoma City, OK) for 1 h at 4°C. GST pull-down assays were performed with 20 l of GSH-Sepharose beads loaded with 20 g of either GST alone or the GST-NT-fusion protein containing the isolated actin-binding domain. Beads were washed five times in ice-cold lysis buffer and resolved by SDS-PAGE. Proteins were detected by autoradiography. In some cases the gels were blotted and further analyzed by Western blotting to assess the identity of labeled proteins using the polyclonal anti-p116 Rip antibody (1:1000 dilution) and anti-myosin II antibodies (1:500 dilution).

RESULTS
We originally isolated p116 Rip through its interaction with activated RhoA-V14 in yeast two-hybrid assays in which the RhoA isoprenylation site was mutated to prevent membrane targeting (13); no interaction was found with other small GTPases, notably activated RhoB, RhoE, Rac1, Cdc42, and Ras. 2 The interaction between p116 Rip and RhoA-V14 was relatively weak, however, and in subsequent studies, we have been unable to confirm that p116 Rip interacts with RhoA in mammalian cells (13). 2 Furthermore, overexpression of p116 Rip in COS-7 cells did not significantly influence the activation state of endogenous RhoA. 3 We therefore conclude that p116 Rip is unlikely to be a high-affinity binding partner and/or negative regulator of RhoA.
p116 Rip Localizes to Dynamic Actin-rich Structures and the Nucleus-As a first step in elucidating the function of p116 Rip , we examined its subcellular localization in N1E-115 and NIH3T3 cells using a polyclonal anti-p116 Rip antibody raised against the putative "RhoA-binding domain" (RBD; Ref. 13). Cells were simultaneously double-stained with rhodaminephalloidin to visualize F-actin. In serum-deprived N1E-115 cells, endogenous p116 Rip colocalizes with F-actin structures, especially the actin-rich microspikes (Fig. 1, top). After stimulation with LPA, a potent activator of RhoA, N1E-115 cells rapidly round up and neurites retract (17,18). In those contracted cells, p116 Rip is found relocalized to the contractile actomyosin ring at the cell cortex (Fig. 1). In NIH3T3 cells, maintained either in serum-free medium or stimulated with LPA, p116 Rip colocalizes with F-actin-rich structures, particularly along stress fibers, at cortical microfilaments, and at the leading edge of lamellipodia (Fig. 1, bottom). Of note, p116 Rip staining is also detected in the cytoplasm and the nucleus (Fig.  1, bottom). Specificity of the observed immunostaining was confirmed by using the GST-RBD polypeptide antigen (previously termed ⌬2; ref. 13), which blocked the p116 Rip fluorescence signal. Furthermore, p116 Rip transfected into COS-7 or N1E-115 cells showed the same subcellular distribution pattern as endogenous p116 Rip : colocalization with F-actin-rich structures as well as nuclear and cytoplasmic staining ( Cells were grown on glass coverslips and treated as follows: N1E-115 or NIH3T3 cells were cultured in serum-free medium and were either left untreated or were stimulated with LPA (1 M) for 10 min to induce RhoA-mediated cytoskeletal contraction and cell rounding (N1E-115 cells) or stress fibers (3T3 cells). NIH3T3 cells were used to visualize lamellipodia and membrane ruffles. Abundant filopodia (microspikes) are observed in serum-starved N1E-115 cells. The distribution of endogenous p116 Rip was visualized by immunofluorescence using polyclonal anti-p116 Rip antibody (raised against part of the coiled-coil region, aa 545-823 (13)), whereas F-actin was detected by rhodamineconjugated phalloidin. Merges are shown in green (p116 Rip ) and red (F-actin). p116 Rip is seen to colocalize with filopodia, cortical actin, membrane ruffles, and stress fibers. In addition, p116 Rip is found in the nucleus (best visible in the two bottom rows) and the cytoplasm. Scale bars, 10 m.

FIG. 2. Association of p116
Rip with the actin cytoskeleton. A, expression constructs used for transfection in N1E-115 cells encode FLp116 Rip and three deletion mutants encompassing the complete C terminus (construct CT; aa 545-1024) or part of the C terminus (construct RBD; aa 545-823) including the "RhoA-binding domain," as indicated, or p116 Rip encompassing the N-terminal PH domain (construct NT; aa 1-382.). P-rich, proline-rich regions. The N-terminal PH domain was not recognized earlier (13). B, immunofluorescence analysis of transfected N1E-115 cells using anti-p116 Rip antibody and rhodamine-conjugated phalloidin (red staining only in left). Transfected full-length p116 Rip colocalizes with cortical actin, whereas truncation mutants CT and RBD are found in the cytoplasm and in the nucleus. C, solubility of transfected full-length or truncated p116 Rip in a buffer containing 0.1% Triton-X-100, as examined by subcellular fractionation into a supernatant (s) and pellet (p) fraction. The p116 Rip antibody was used for detection on Western blot. Full-length p116 Rip is partially insoluble whereas the truncation mutants CT and RBD are largely soluble. IB, immunoblot. D, N1E-115 cells were transfected with an HA-tagged construct encoding a truncated version of p116 Rip encompassing the N-terminal PH domain (construct HA-NT; aa 1-382). Cells were fixed 24 h after start of transfection, and the localization of the expressed NT construct was analyzed by immunofluorescence using anti-HA antibody. F-actin was stained with rhodamine-conjugated phalloidin (red staining). It is seen that the NT protein colocalizes with F-actin.  (13)); two proline-rich regions, and a C-terminal coiled-coil region ( Fig.  2A). The putative "RhoA-binding domain" (RBD) that was isolated in yeast two-hybrid screens (13) overlaps with the coiledcoil region, as indicated in Fig. 2A.
We generated HA-tagged p116 Rip and three truncated versions (HA-tagged) encoding the CT coiled-coil region, the RBD and the NT half (NT-p116 Rip ; Fig. 2A). The various cDNA constructs were transiently transfected into N1E-115 cells and the subcellular localization and detergent solubility of the resulting proteins were analyzed. Transfected HA-p116 Rip , like endogenous p116 Rip , localizes to cortical F-actin (and the nucleus; data not shown). In contrast, the p116 Rip -CT and RBD polypeptides display nuclear and cytoplasmic localization (Fig.  2B). In keeping with these results, the CT and RBD truncation mutants are largely Triton-soluble, whereas full-length p116 Rip (transfected and endogenous) is about 50% insoluble ( Fig. 2C and results not shown), consistent with association with the cytoskeleton.
Similar to full-length p116 Rip , the N terminus of p116 Rip (NT-p116 Rip ) colocalizes with F-actin and is also detectable in the nucleus (Fig. 2D and results not shown). When the NT-p116 Rip -expressing cells were analyzed at Ͼ48 h after transfection, however, the F-actin cytoskeleton was largely disrupted (see below). Collectively, these results indicate that the N-terminal part of p116 Rip (aa 1-382) determines its subcellular localization.
Binding of p116 Rip to F-actin-F-actin associates with the motor protein myosin-II to generate contractile forces in nonmuscle cells. In metabolically labeled N1E-115 cells, we found that endogenous p116 Rip as well as the purified polypeptide NT-p116 Rip (fused to GST) coprecipitated with proteins of 43 and 200 kDa (Fig. 3A, lanes 2 and 4, respectively). Immunoblot analysis confirmed that the 43-kDa protein is actin (not shown), and revealed that the 200 kDa protein represents the heavy-chain of non-muscle myosin-II (Fig. 3B). Although the reciprocal precipitations yielded variable results, our data support the notion that p116 Rip associates with actomyosin in vivo.
We next investigated the actin-binding properties of NT-p116 Rip . As shown in Fig. 4A, NT-p116 Rip (fused to GST) cosediments with purified F-actin, as did ␣-actinin, whereas GST alone and BSA did not. Fusion proteins containing the Cterminal regions of p116 Rip (CT and RBD) failed to cosediment with F-actin (results not shown). It thus seems that the Nterminal region of p116 Rip contains an F-actin-binding domain. We next examined the binding affinity of full-length p116 Rip for F-actin. Increasing concentrations of purified F-actin (0 -3.5 M) were mixed with a fixed amount of p116 Rip (1 M). After high-speed centrifugation, the amount of p116 Rip cosedimenting with F-actin was determined. From the resulting binding curve we estimate that p116 Rip binds to F-actin with an apparent dissociation constant (K d ) of about 0.5 M (Fig. 4B).
To define the N-terminal regions mediating F-actin binding in further detail, we made various deletion mutants and determined their F-actin cosedimentation properties, as illustrated in Fig. 4C. Whereas p116 Rip has no obvious sequence homology to known actin-binding proteins, potential actin-binding motifs include the N-terminal PH domain (19) as well as a stretch of positively charged residues (KKKRK, aa 157-161) that could interact with the highly anionic actin filament (20). We found that one deletion mutant (⌬6; aa 1-212), comprising both the PH domain and the positive stretch, can bind F-actin, whereas the other mutants cannot (Fig. 4C). Thus, the extreme N terminus (aa 1-43), the PH domain, and the adjacent cationic residues are all necessary to mediate F-actin binding. Further definition of the critical actin-binding motif(s) within the Nterminal region must await future studies.
p116 Rip Induces Bundling of F-actin in Vitro via its N-terminal Region-We next examined the ability of p116 Rip and NT-116 Rip to induce actin cross-linking in vitro, using ␣-actinin as a positive control. Myc-p116 Rip and GST-p116 Rip were isolated from transfected COS-7 cells using affinity chromatography, and protein purity was determined by Coomassie Blue stain-

FIG. 3. Association of p116 Rip with actomyosin complexes in vivo.
A, N1E-115 cells were starved in methionine/cysteine-free medium for 30 min and then labeled for 4 h with [ 35 S]methionine/cysteine. Endogenous p116 Rip was immunoprecipitated (IP) using polyclonal anti-p116 Rip antibodies (arrow). Normal rabbit serum (NRS) was used as a control for immunoprecipitation (left). GST pull-down assays were performed with either GST alone or the GST fusion protein containing the isolated actin-binding domain (GST-NT-p116 Rip ; right lanes). Precipitates were subjected to SDS-PAGE and analyzed by autoradiography. B, Western blot analysis of precipitates and GST pull-down assays using polyclonal anti-p116 Rip and anti-myosin-II antibodies, endogenous p116 Rip , and myosin-II were immunoprecipitated using polyclonal anti-p116 Rip antibodies and polyclonal anti-myosin-II antibodies, respectively (left lanes). Normal rabbit serum (NRS) was used as a control for immunoprecipitation. GST pull-down assays were performed with either GST alone or the GST fusion protein containing the isolated actin-binding domain (GST-NT-p116 Rip ; right lanes). Myosin coprecipitates in p116 Rip immuno-complexes and in the GST-NT-p116 Rip pulldown assay.
p116 Rip , A Novel F-Actin-binding Protein ing. Myc-p116 Rip , like GST-p116 Rip , binds F-actin, as shown by cosedimentation assays using lysates from transfected COS-7 cells (see Fig. 6A, left). Because the dimeric nature of GST could mediate artifactual actin cross-linking by GST-p116 Rip , we also used Myc-p116 Rip . Purified GST-p116 Rip , Myc-p116 Rip , ␣-actinin, or GST alone were incubated with F-actin, and the samples were subsequently analyzed by electron microscopy. In the absence of p116 Rip or in the presence of GST alone, long actin filaments were randomly distributed all over the grid and no organized actin bundles were observed (Fig. 5A). In the presence of either GST-p116 Rip or Myc-p116 Rip , however, F-actin became organized into thick bundles similar to those formed by the actin-bundling protein ␣-actinin (Fig. 5, B, C, and D). The bundles consisted of many actin filaments closely aligned in juxtaposition, with no branching of filaments observed.
We also tested the isolated N-terminal actin-binding domain (NT-p116 Rip ; aa 1-382) and the C-terminal coiled-coil region (CT-p116 Rip ; aa 545-1024) for bundling activity. In these experiments, GST-fusion proteins were produced in bacteria followed by GST cleavage. As expected, the NT-p116 Rip protein induced actin bundling similar to full-length p116 Rip , whereas no actin bundles were observed after incubation of F-actin with the CT polypeptide (Fig. 5, E and F). Thus, p116 Rip induces bundling of F-actin in vitro through its N-terminal actin-binding domain.

Expression of p116 Rip or the N-Terminal Actin-binding Domain Promotes Stress Fiber Disassembly and Process
Outgrowth-We investigated the effects of overexpression of p116 Rip and its N-terminal region on cell morphology and cytoskeletal organization in NIH3T3 cells. To this end, we used HA-tagged p116 Rip and a p116 Rip -GFP fusion protein (its direct binding to F-actin was confirmed; Fig. 6A). Contrary to expectations raised by the actin-bundling studies, overexpression of p116 Rip , A Novel F-Actin-binding Protein p116 Rip in NIH3T3 cells resulted in loss of stress fibers and outgrowth of long dendrite-like processes (Fig. 6B). This phenotype was observed with wild-type p116 Rip , Myc-, HA-and p116 Rip -GFP (Fig. 6, B and C, and results not shown). Less than 10% of the p116 Rip -transfected NIH3T3 cells contained stress fibers, compared with Ͼ60% of the GFP-expressing control cells (Fig. 6D). LPA stimulation of NIH3T3 cells leads to rapid RhoA-mediated cell contraction (albeit less dramatic than in N1E-115 cells). However, no contractile response to LPA was seen in the p116 Rip -overexpressing NIH3T3 cells, similar to what we previously observed in p116 Rip -overexpressing N1E-115 cells (13). Loss of stress fibers was already detectable at 6 to 8 h after transfection, whereas process extension appeared at later time points (Ͼ12-16 h, when p116 Rip levels were more elevated).
Expression of the actin-binding region (HA-NT-p116 Rip ) in NIH3T3 cells led to the same dramatic loss of stress fibers and induction of dendrite-like extensions. In contrast, cells expressing the C-terminal domain only (HA-CT-p116 Rip ) displayed a normal stress fiber pattern (Fig. 7, A and B). Thus, the NT region of p116 Rip is necessary and sufficient for stress-fiber p116 Rip , A Novel F-Actin-binding Protein disruption and consequent loss of contractility in p116 Rip -overexpressing cells.
Finally, we examined the cytoskeletal response of p116 Ripoverexpressing NIH3T3 cells to platelet-derived growth factor (PDGF), which is a potent inducer of Rac-mediated lamellipodia formation and membrane ruffling. PDGF induced prominent lamellipodia formation in the control cells but not in the p116 Rip -GFP-expressing cells (Fig. 8, A and B). We conclude that although p116 Rip has actin-bundling activity in vitro, overexpression of p116 Rip in fibroblasts and neuronal cells disrupts F-actin assembly and thereby interferes with Rho/Rac-controlled cytoskeletal remodeling. DISCUSSION We originally isolated p116 Rip as a RhoA-interacting protein in a yeast two-hybrid screen (13). Binding to activated RhoA was relatively weak, however, and our initial conclusion that p116 Rip interacts directly with RhoA in mammalian cells turned out to be premature (13). In fact, we have since found that p116 Rip is unlikely to be a direct binding partner of RhoA. 2 In the present study, we provide the first insights into the function of p116 Rip . We show here that p116 Rip is an F-actinbinding protein that has bundling activity in vitro, with the actin-binding domain residing in the N-terminal region (aa 1-212; construct ⌬6-p116 Rip ; Fig. 4C). This conclusion is based on the following observations: 1) p116 Rip and NT-p116 Rip associate with actomyosin complexes in vivo; 2), the N-terminal region of p116 Rip , but not its C-terminal half, cosediments with F-actin in vitro; and 3), purified full-length p116 Rip and NT-p116 Rip induce the assembly of actin filaments into thick bundles in vitro. In addition, we show that endogenous p116 Rip localizes to dynamic F-actin-rich structures that are normally under the control of Rho family GTPases, notably along stress fibers, in cortical microfilaments as well as in filopodia and lamellipodia. Furthermore, p116 Rip is also detected in the nucleus, consistent with p116 Rip containing several potential nuclear localization signals (between residues 43-587; not shown). There is growing evidence for the presence of actin and actin-binding proteins in the nucleus, but very little is still known about their importance for normal cell function (21). One challenge for future studies is to determine how nuclear targeting of p116 Rip is normally regulated.
The N-terminal region p116 Rip shows no obvious sequence similarity to known F-actin-binding proteins. Therefore, p116 Rip does not classify as a member of the superfamily of actin-binding proteins that includes ␣-actinin/spectrin members, plectin, filamin, and dystrophin (22)(23)(24). At least three distinct mechanisms could account for the actin-bundling activity of NT-p116 Rip . One possibility would be that NT-p116 Rip is able to dimerize and thereby induces actin bundling. However, using transfected COS cells, we did not detect an interaction between NT-p116 Rip and full-length p116 Rip (results not shown), which argues against the possibility that the NT domain can form dimers. The second possibility is that NT-p116 Rip might bundle F-actin through the polycationic KKKRK motif (residues 157-161), just after the first PH domain (20). However, mutational analysis reveals that neither the first PH domain nor the cationic motif is sufficient for F-actin binding (Fig. 4C). A third possibility is that NT-p116 Rip may harbor two actin-binding domains, each of which binds a separate actin filament; as yet, we have no evidence for or against this notion. Further studies are required to identify the N-terminal sequence motifs in p116 Rip that determine F-actin binding and bundling.
Contrary to what one would expect for a protein with actinbundling activity, overexpression of p116 Rip in NIH3T3 cells causes loss of stress fibers and produces a dendrite-like morphology. This phenotype, which is reminiscent of cells expressing dominant-negative RhoA (18,25) requires the N-terminal actin-binding domain of p116 Rip but not the C-terminal coiledcoil region. The importance of the N terminus in determining cytoskeletal architecture can also be inferred from the observation that overexpressed p116 Rip causes cell flattening in N1E-115 cells, whereas an N-terminally truncated version does not (13). Loss of stress fibers and other actin-rich structures is a common feature of overexpressed actin-monomer (G-actin) sequestering proteins (26 -28), but our efforts to test whether NT-p116 Rip can bind G-actin yielded negative results (not shown). However, there is precedent for actin cross-linking proteins to cause F-actin disassembly in vivo. In particular, overexpression of the actin-binding region of neurabin, an Factin cross-linking protein, causes collapse of stress fibers and promotes filopodial outgrowth, apparently by recruiting protein phosphatase I to F-actin-rich structures (29). Furthermore, overexpression of villin, a protein that can bundle, cap, nucleate, or sever actin in vitro, results in the disappearance of stress fibers and enhanced microvilli elongation, a phenotype that strictly correlates with the actin-bundling activity of villin (30).
Overexpressed p116 Rip not only induces an inactive RhoA phenotype but also interferes with PDGF-induced lamellipodia p116 Rip , A Novel F-Actin-binding Protein formation, which is a typical Rac-mediated response. The present findings lead us to suggest that, rather than being a negative regulator of Rho/Rac, p116 Rip can destabilize F-actin-rich structures by competing with and displacing other actin-crosslinking proteins. An alternative or additional possibility is that p116 Rip may recruit regulatory proteins that disassemble the F-actin network (such as actin-severing proteins or protein phosphatases; see Ref. 29). As for the displacement model, the neuronal F-actin-binding protein drebrin induces the formation of highly branched processes, similar to that observed with p116 Rip . It does so by interfering with the actin binding and bundling activities of fascin, ␣-actinin, and tropomyosin (31,32). A similar mechanism might underlie the p116 Rip overexpression phenotype.
Finally, we note that a recently identified actin-binding protein named Tara (593 residues) shows a high degree of similarity to p116 Rip (46% overall amino acid identity (33)). In common with p116 Rip , Tara contains an N-terminal PH domain and a C-terminal coiled-coil region, but it lacks the N-terminal actin-binding region of p116 Rip . No actin cross-linking activity has been reported for Tara until now; nevertheless, overexpression of Tara in HeLa cells leads to enhanced formation of stress fibers and cortical F-actin (33). Thus, despite their structural similarities, p116 Rip and Tara have opposing actions on F-actin organization.
In conclusion, our studies specify p116 Rip as a novel F-actinbinding protein with bundling activity in vitro and demonstrate that p116 Rip can affect, directly or indirectly, the integrity and contractility of the actomyosin-based cytoskeleton. Further insight into the physiological role of p116 Rip in cytoskeletal regulation will rely on the identification of additional binding partners of p116 Rip as well as on interference approaches by using RNA interference-expressing vectors. These studies are currently in progress.