R-Ras contains a proline-rich site that binds to SH3 domains and is required for integrin activation by R-Ras.

R-Ras contains a proline-rich motif that resembles SH3 domain-binding sites but that has escaped notice previously. We show here that this site in R-Ras is capable of binding SH3 domains and that the SH3 domain binding may be important for R-Ras function. A fusion protein containing the SH3 domains of the adaptor protein Nck interacted strongly with the R-Ras proline-rich sequence and with the intact protein. The binding was independent of whether R-Ras was in its GDP or GTP form. The Nck binding, which was mediated by the second of the three SH3 domains of Nck, was obliterated by mutations in the proline-rich sequence of R-Ras. The interaction of Nck with R-Ras could also be shown in yeast two-hybrid assays and by co-immunoprecipitation in human cells transfected with Nck and R-Ras. Previous results have shown that the expression of a constitutively active R-Ras mutant, R-Ras(38V), converts mouse 32D monocytic cells into highly adherent cells. Introducing the proline mutations into R-Ras(38V) suppressed the effect of R-Ras on 32D cell adhesion while not affecting GTP binding. These results reveal an unexpected regulatory pathway that controls R-Ras through an SH3 domain interaction. This pathway appears to be important for the ability of R-Ras to control cell adhesion.

Integrins mediate cell adhesion to extracellular matrices and, in some cases, to other cells. Their activity also helps regulate cell survival, growth, differentiation, and migration. Integrin activity is, in turn, regulated by the cell. The most striking examples of regulated integrin activity are the activation of the platelet ␣ IIb ␤ 3 integrin in blood clotting and the activation of the leukocyte ␤ 2 integrins in inflammation (1,2).
The molecular mechanisms of integrin regulation are not well understood. There appear to be two ways of enhancing the cell attachment-promoting activity of integrins; clustering of integrins at the cell surface can enhance the avidity of integrins in cell attachment, whereas conformational changes can increase the affinity of individual integrin molecules (3)(4)(5)(6). Cytoskeletal connections of integrins are likely to be important in the avidity regulation (7), whereas the affinity of integrins may be regulated by other types of proteins. Cytohesins are phosphatidylinositol-binding proteins, at least one of which binds to the cytoplasmic tail of the ␤ 2 integrin subunit and can upregulate the activity of ␤ 2 integrins in leukocytic cells (8). Endonexin is a ␤ 3 -binding protein that increases the activity of the ␣ IIb ␤ 3 integrin when overexpressed in cells (9). ILK is a protein kinase capable of interacting with ␤ 1 , ␤ 2 , and ␤ 3 integrin subunit cytoplasmic tails; it can reduce the activity of integrins containing these subunits (10). The calcium/calmodulin-dependent protein kinase II also down-regulates integrin activity (11).
Certain oncoproteins can change integrin activity, generally by lowering it. Thus, c-Src phosphorylates a tyrosine residue in the ␤ 1 integrin cytoplasmic domain, and this phosphorylation reduces the ligand binding activity of ␤ 1 integrins and changes their subcellular localization (12,13). The small GTPases of the Ras and Rho families are intimately involved in integrin and cytoskeletal regulation. The activation of Rho induces reorganization of the actin cytoskeleton, with consequent cell spreading and effects on integrins (14,15). The related proteins Rac and Cdc42 alter the cytoskeleton differently than Rho, inducing membrane ruffling and microspike formation, respectively. The oncogenic p21 ras (Ras) reduces integrin activity (16).
R-Ras, a small GTPase with a poorly understood function, regulates integrin activity (17)(18)(19)(20)(21). Unlike Ras, R-Ras promotes integrin activity and converts cells that normally grow in suspension into highly adhesive cells (17). Moreover, a dominant negative R-Ras (R-Ras(43N)), when transfected into adherent cells, causes the cells to round up, suggesting that R-Ras is necessary for the maintenance of integrin activity (17). Unlike dominant negative mutants of the oncogenic Ras proteins, R-Ras(43N) is not growth inhibitory (20). Thus, R-Ras may be primarily a regulator of cell adhesion, and it is important to understand how this regulation functions.
Examining R-Ras for sequence features that might be responsible for its integrin activating function, we noticed that R-Ras differs from Ras and most other members of Ras superfamily small GTPases in that it possesses a distinct prolinerich site. We show here that this site can bind SH3 domains and that the adaptor protein Nck is one of the SH3 domain proteins that interact with R-Ras in cells. We also show that the SH3-binding site is required for integrin activating function of R-Ras. These results reveal an unexpected regulatory interaction for R-Ras that appears to be important for the ability of R-Ras to control cell adhesion. This interaction represents a novel form of cross-talk between a small GTPase and other cellular signaling pathways.

EXPERIMENTAL PROCEDURES
Cells, Antibodies, and Reagents-Mouse monocytic cell line 32D.3 was maintained essentially as described (17) in RPMI 1640 medium supplemented with 1 mM glutamine, 100 units/ml penicillin G, 100 g/ml streptomycin, 10% heat-inactivated fetal calf serum, and 20% conditioned medium from the interleukin-3-producing cell line WEHI-3B. Human embryonic kidney 293 cells were maintained in Dulbecco's modified Eagle's medium plus 10% fetal calf serum. Polyclonal rabbit antibody against the N-terminal 26-amino acid peptide of R-Ras was a gift from Dr. John Reed (21). Mouse monoclonal antibody against hemagglutinin (HA) 1 12CA5 was provided by Dr. J.-L. Guan. Mouse hybridoma cell line producing monoclonal antibody 9E10 against Myc epitope was obtained from American Type Cell Culture. Polyclonal anti-Nck antibody was purchased from Santa Cruz Biotechnology. Human plasma fibronectin was from the Finnish Red Cross.
Preparation of glutathione S-transferase (GST) fusion proteins with SH3 domains from Src and Crk (N) (23), p85␣ (24), Abl (25), Grb2 (C), Fgr, spectrin, and phospholipase C␥ (26) have been described. Nck deletion mutants were constructed by removing SH3 domain individually from the C terminus of NcK by polymerase chain reaction (27). The polymerase chain reaction fragments were then cloned into PGEX-4T1 and expressed as GST fusion proteins. The plasmid for GST fusion protein of the second SH3 domain was obtained from Dr. Sakaue. Expression vectors for GST-R-Ras and GST-R-Ras-(P202A,P203A) were constructed by digesting the plasmids pcDNA3-Myc-R-Ras and pMRR-(P202A,P203A) with EcoRI and XhoI. The released inserts were then cloned into the vector PGEX4T1 (Amersham Pharmacia Biotech). All of the fusion proteins were expressed in bacteria and purified on glutathione-Sepharose (Amersham Pharmacia Biotech) as described (28).
In Vitro Binding of MBP-R-Ras Fusion Proteins to GST-SH3 Fusion Proteins-About 10 g of MBP alone, MBP-RRC, or its mutants were labeled with Na 125 I using Iodogen (Pierce) as described previously (29). In a typical binding assay, 10 g of a GST-SH3 domain fusion proteins in phosphate-buffered saline (PBS) was immobilized on glutathione-Sepharose. The beads were washed twice with PBS and once with the binding buffer (25 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM EDTA, 0.5% Nonidet P-40, 0.05% SDS, 1% BSA, and 0.2 mM phenylmethylsulfonyl fluoride, and resuspended in 20 l of the same buffer at 4°C. About 1 ϫ 10 6 cpm of 125 I-labeled MBP-RRC fusion protein in PBS containing 1% BSA was added to each tube and incubated for 30 min at 4°C. The beads were washed four times with binding buffer without BSA. The bound materials were eluted with 50 l of SDS-polyacrylamide gel electrophoresis (PAGE) loading buffer, separated on 4 -20% gradient SDS-PAGE minigel (Novex, San Diego), and exposed to x-ray films.
In Vitro Binding Assays with GTP-bound and GDP-bound Forms of R-Ras-A GST-R-Ras fusion protein containing the entire R-Ras pro-tein was immobilized on glutathione-Sepharose beads. The beads were washed with loading buffer (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM EDTA, and 0.1% BSA) and split into two tubes. The beads were incubated with 1 mM GTP␥S or 1 mM GDP in 200 l of the loading buffer for 15 min at 37°C. After the incubation, MgCl 2 was added to a final concentration of 10 mM to stabilize the binding, 1 ml of cell lysate from 293 cells expressing HA-Nck (full length) was then added to GST-R-Ras beads loaded with GTP␥S or GDP. After incubation at 4°C for 1 h, the beads were washed three times with wash buffer (20 mM Tris-HCl, 130 mM NaCl, 1 mM EDTA, 4 mM MgCl 2 , 0.1% Nonidet P-40, 10% glycerol, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) and resuspended in SDS-PAGE sample buffer. After electrophoresis, proteins were transferred to the polyvinylidene difluoride membrane and blotted with anti-HA antibody to visualize HA-Nck.
Yeast Two-hybrid Assay-The yeast two-hybrid system was used to study protein-protein interactions essentially as described (30,31). A cDNA encoding R-Ras without CAAX box (the CAAX box was removed to allow nuclear localization of R-Ras) was amplified with polymerase chain reaction and cloned into EcoRI and SmaI sites of pBTM116, and the SH3 domains of Nck were cloned into the Klenow-treated BamHI and NotI sites of VP16. As a negative control, lamin cDNA was cloned into BTM116. Yeast strain L40 was co-transfected with 1 g of each plasmid by the LiCl method (32). One-tenth of the transformed yeast cells were plated onto medium lacking uracil, tryptophan, and leucine (UTL) for analysis of transformation efficiency, and nine-tenths were plated onto medium lacking tryptophan, histidine, uracil, leucine, and lysine (THULL). UTL colonies were grown for 2 days at 30°C and then restreaked onto UTL and THULL plates. Filters were loaded on the plates with streaks from yeast transformation, frozen in liquid nitrogen, and warmed to room temperature and then stained with 0.4 mg/ml 5-bromo-4chloro-3-indolyl-␤-D-galactoside in 60 mM Na 2 HPO 4 , 10 mM KCl, 1 mM MgCl 2 , pH 7.0, at 30°C.
Transient Transfection, Immunoprecipitation, and Immunoblot Analysis-Human embryonic kidney 293 cells on 100-mm plates were transiently transfected with 10 g of the indicated vectors using standard calcium phosphate precipitation method. Two days later, cells were lysed in buffer A containing 50 mM Tris-HCl, 50 mM NaCl, 0.5% Triton X-100, 10% glycerol, 0.1% BSA, and protease inhibitors (0.1 unit/ml aprotinin, 10 g/ml leupeptin, and 0.5 mM phenylmethylsulfonyl fluoride). Lysates were clarified by centrifugation at 15,000 ϫ g for 10 min. Antibodies were added to lysates containing equal amount of proteins and incubated for 1 h at 4°C. To precipitate the antibody-antigen complex, Gammabind Sepharose (Amersham Pharmacia Biotech) was added, and incubation was continued for another hour. The immunoprecipitates, pelleted by centrifugation, were washed twice with buffer A containing 0.1% BSA, and once with buffer A alone. The beads were boiled in sample buffer and separated on SDS-PAGE gels. After electrophoresis, proteins were transferred to Immobilon-P nylon membrane (Millipore) and subjected to immunoblotting. Filters were first blocked for 1 h with blocking buffer (3% milk plus 1% BSA in PBS) and incubated with primary antibodies (1 g/ml) for 1 h. Horseradish peroxidase-conjugated secondary antibodies (Sigma) were added at 1:5000 dilutions in blocking buffer. Membrane were developed with Enhanced Luminol Reagents from DuPont.
Stable Transfection of 32D Cells-Log phase 32D.3 cells were transfected with indicated plasmids by as described (16). One day after transfection, 0.7 g/ml Geneticin (G418) (Life Technologies, Inc.) was added. Individual clones were obtained by limited dilution. Level of expression in each clone was examined by immunoblot with an anti-HA epitope monoclonal antibody 12CA5.
Cell Adhesion Assay-Serial dilutions of fibronectin in PBS was coated on 96-well microtiter plates starting at 10 g/ml. Nonspecific binding sites were blocked with 1% BSA in PBS. 32D cells and their transfectants were collected by pipetting or by using brief treatment with 1 mM EDTA in PBS. The suspended cells were washed and plated in duplicate onto fibronectin-coated wells in serum-free medium at 1 ϫ 10 5 cells/well and allowed to adhere for 1 h at 37°C. Adherent cells were fixed with 3.7% paraformaldehyde and stained with 0.5% crystal violet, which was then extracted with 50% ethanol in 50 mM sodium citrate, pH 4.5, and quantitated by measuring absorbance at 595 nm.

R-Ras Contains a Proline-rich Region That Resembles SH3
Domain-binding Sites-In a search for potential mediators for the integrin activating function of R-Ras, we examined sequence features of R-Ras outside the effector domain. As shown in Fig. 1 a proline-rich region toward the C terminus of R-Ras contains three possible PXXP minimal SH3 domain recognition sequences, one of which conforms to the consensus features of a preferred binding site for type II SH3 domains (33,34). No PXXP motif is present in other members of Ras superfamily small GTPases with the exception of TC21 and CDC42Hs. TC21 is most closely related to R-Ras (35) and contains a single PPSP motif (Fig. 1B). CDC42Hs belongs to the Rho subfamily of small GTPases (36), its PPEPKK site also conforms to the consensus type II SH3-binding motif (Fig. 1B). Interestingly, the brain-specific version of CDC42Hs, G25K, (37) lacks the potential SH3-binding site.
Proline-rich Region of R-Ras Selectively Interacts with SH3 Domains in Vitro-To determine whether R-Ras can indeed bind to SH3 domain-containing proteins, we fused the C-terminal portion of R-Ras (amino acids 192-212) to MBP and screened a panel of GST-SH3 domain fusion proteins for binding to this fusion protein, MBP-RRC. A recombinant protein encompassing the three SH3 domains of Nck bound avidly to MBP-RRC ( Fig. 2A). By counting the excised band, it was estimated that about 12% of 125 I-labeled MBP-RRC could be bound by the Nck SH3 domain fragment. None of the other SH3 domains tested exhibited binding above the background level of GST alone. As a control, MBP alone was tested and did not bind to any SH3 domain fusion proteins under the same conditions ( Fig. 2A, lower panel). These results show that the proline-rich domain of R-Ras can serve as selective binding site for SH3 domain-containing proteins, Nck in particular.
SH3 Domain Binding Requires Proline Residues 202 and 203 of R-Ras-To demonstrate that the interaction between the C-terminal portion of R-Ras and Nck is mediated by the proline-rich region, proline to alanine point mutations were introduced to alter the PXXP motifs in R-Ras. Fig. 2B shows that mutation of proline 202 to alanine (P202A), which eliminates the first PXXP site (Fig. 1A), greatly reduced the binding to Nck SH3 domains. Disrupting the second and third PXXP sites with the P203A mutation similarly reduced Nck binding. Mutating both prolines (P202A,P203A) resulted in further reduction of the Nck binding. These results show that the prolinerich domain is required for binding to Nck SH3 domains. Testing with fusion proteins containing various combinations of the three Nck SH3 domains showed that only the middle SH3 domain of Nck binds R-Ras (Fig. 2C).

FIG. 2. Binding of SH3 domains to R-Ras proline-rich site.
A and B, the C-terminal region of R-Ras containing the proline-rich sequence (residues 119 -212) was fused to MBP in pPR997 vector. This fusion protein (MBP-RRC) and its proline to alanine mutant derivatives (P202A, P203A, and double mutant P202A,P203A) were labeled with 125 I, and equal amounts of each protein were tested for binding to SH3 domains from various proteins, produced as GST fusion proteins in bacteria, and immobilized on glutathione-Sepharose. Bound materials were eluted with SDS-PAGE sample loading buffer and separated on a 4 -20% gradient gel. A, MBP-RRC but not MBP binds strongly to Nck. B, the proline to alanine R-Ras mutants show reduced binding to the Nck SH3 domain fragment. C, R-Ras interacts with the second SH3 domain of Nck. The full-length of GST-Nck fusion protein and GST fusion proteins of Nck mutants in which the second or third SH3 domain had been deleted, as well as a GST fusion protein of the second SH3 domain, were immobilized on glutathione-Sepharose beads. The beads were incubated with equal amounts of cell lysates from Myc-Ras transfected 293T cells at 4°C for 1 h. After washing, the bound proteins (including the GST fusion protein) were eluted with SDS-PAGE sample buffer and blotted with anti-R-Ras antibody (Santa Cruz Biotechnology). Protein staining (lower panel) shows that equal amounts of the fusion proteins had bound to the beads. D, both GTP and GDP forms of R-Ras bind Nck. GST-R-Ras immobilized on glutathione-Sepharose beads was preloaded with GTP␥s or GDP␥s. The beads were incubated with lysates prepared from cells expressing HA-Nck. Bound proteins were eluted with SDS-PAGE sample buffer, separated on a 4 -20% gradient gel and transferred to a polyvinylidene difluoride membrane. Immunoblotting of the membrane with anti-HA showed Nck binding to GST-R-Ras loaded with either GTP or GDP but not to GST. GTP-bound and GDP-bound R-Ras-GST fusion protein was examined. Nck was found to bind both the GTP and GDP forms of R-Ras, indicating that the Nck interaction is not regulated by the R-Ras nucleotide binding (Fig. 2D). (38,39). A Nck fragment encoding the SH3 domains was the "prey," and R-Ras with a deleted CAAX box (R-Ras-CAAX Ϫ ) was the "bait." R-Ras and Nck interacted strongly in this growth assay (Fig. 3) as well as in an assay based on ␤-galactosidase activity (not shown). As expected (40), R-Ras interacted also with Raf-1 (not shown). A negative control using lamin as bait showed no signal with the Nck prey. Thus, the SH3 domain is active in yeast cells, and R-Ras can associate specifically with Nck in these cells.

R-Ras and Nck Form a Complex in Mammalian
Cells-To study R-Ras-SH3 domain interaction in mammalian cells, we transiently transfected Nck into 293 cells together with HAtagged R-Ras or R-Ras-(P202A,P203A). Immunoprecipitation and immunoblot shows that wild-type R-Ras co-immunoprecipitated with Nck and that P202A,P203A mutation significantly reduced that association (Fig. 4A, left half). Similar results were obtained when cells were transfected with HAtagged Nck and Myc-tagged R-Ras or R-Ras-(P202A,P203A) ( Fig. 4A right half). These results indicate that R-Ras and Nck can associate in intact cells and that proline-rich domain is necessary for the association.
The Proline-rich Domain of R-Ras Is Required for Integrin Regulation by R-Ras-We have previously demonstrated that an activated form of R-Ras, R-Ras(38V), can induce substrate attachment of 32D mouse monocytic cells, which normally grow in suspension, and that the increased adhesiveness is caused by integrin activation (17). We next used this assay to determine whether the proline-rich site contributes to the integrinregulating activity of R-Ras. A major proportion of transfected 32D cells expressing the activated R-Ras(38V) became adherent and spread on tissue culture dishes (Fig. 5A), whereas 32D cells expressing an R-Ras(38V) mutated in the two SH3-binding domain proline residues (R-Ras(38V)-(P202A,P203A)) grew in suspension (Fig. 5B). Control cells, including those transfected with vector alone (Fig. 5C), parental cells (not shown), and R-Ras(wt)-transfected cells (below), also remained nonadherent.
To quantitate the effects of the P202A,P203A mutation on integrin-mediated cell adhesion, 32D transfectants were plated onto wells coated with varying concentrations of fibronectin.
R-Ras(38V)-transfected cells adhered to fibronectin, but clones expressing the R-Ras(38V)-(P202A,P203A) mutant either did not adhere to fibronectin or adhered poorly to it (Fig. 5D). That the various transfectants expressed equivalent levels of the transfected R-Ras was confirmed by immunoblotting cell extracts with anti-HA antibody (not shown). No adhesion was observed with R-Ras(wt)-or vector-transfected cells. Similar results were obtained when another adhesive protein, vitronectin, was used to coat the substrate. The lack of cell attachmentpromoting activity by the R-Ras(38V)-(P202A,P203A) mutant was unlikely to be due to a loss of GTP binding capability, because controls showed the P202A,P203A mutation to have no effect on GTP binding by R-Ras (Fig. 6). These results indicate that the proline-rich domain contributes to the integrin activating function of R-Ras, possibly by interacting with the SH3 domains of Nck. DISCUSSION Our results reveal an unexpected regulatory interaction of R-Ras that is mediated through binding to SH3 domains. The results single out the adaptor protein Nck as a candidate cellular protein for SH3 domain-dependent binding of R-Ras and show that the integrity of SH3 domain binding is necessary for the ability of R-Ras to regulate integrin-mediated cell attachment.
The interactions of the R-Ras proline-rich domain have the hallmarks of an SH3 domain interaction. Three lines of evidence: direct binding assays with fusion proteins, two-hybrid analysis, and coimmunoprecipitation from cells, demonstrate the specificity and high avidity of the R-Ras binding to an SH3 domain of Nck. Nck is an adaptor protein comprising three consecutive SH3 domains and a C-terminal SH2 domain (41). Our use of SH3 domain fusion proteins in the binding assays places the interaction site in Nck to the SH3 domains, specifically to the middle SH3 domain. Moreover, our mutational analysis showed that the proline-rich segment of R-Ras, which includes three adjacent copies of the SH3 domain-binding consensus sequence PXXP, is the binding site on the R-Ras side. To our knowledge, this is the first demonstration of an SH3 domain interaction by a small GTPase. The independence of the SH3 domain binding on the GTP/GDP regulation of R-Ras suggests that Nck (and other SH3 domain proteins that might bind to this site) functions as an adaptor, rather than as downstream effector, of R-Ras.
The Nck SH3 domain fragment was the only efficient binder of the R-Ras C-terminal proline-rich sequence; fusion proteins of SH3 domains from nine other proteins showed no significant binding. The specificity of the SH3 domain binding was further underscored by the fact that only one of the three SH3 domains of Nck bound the R-Ras fragment. Moreover, we were also able to show that the Nck-R-Ras interaction takes place in cells. Specific binding of these proteins was seen in yeast two-hybrid tests, and we were also able to co-immunoprecipitate R-Ras and Nck from human cell extracts. The co-immunoprecipitation of Nck and R-Ras was only seen after transfection of the two proteins into the test cells, and only a small fraction of the transfected R-Ras appeared to be associated with Nck. This is not surprising, given the many other interactions of Nck with cellular signaling proteins. The interaction may also be regulated in such a way that only a subfraction of R-Ras is bound to Nck at any given time. Although we cannot exclude the possibility that some other SH3 domain proteins might also interact with R-Ras, these results suggest that R-Ras may be a physiological ligand of Nck.
The SH3 domain-binding site of R-Ras identified in this study may serve to target R-Ras to appropriate subcellular locations. The site is located in the 20 amino acids preceding the conserved C-terminal tetrapeptide motif, the CAAX box. This stretch of amino acids is highly variable among the various Ras proteins (42). However, it is conserved in the same Ras protein among species. The C-terminal regions of the only two R-Ras proteins for which the sequences are available, human and mouse, differ by one amino acid, and the proline-rich sequences are identical. This conservation suggests that the C-terminal region plays an important role in defining the divergent functions of the individual Ras proteins. In R-Ras, one such function appears to be SH3 domain binding.
Nck, a candidate binder of R-Ras in cells, interacts through its SH2 domain with p130 Cas (43) and with activated receptor protein tyrosine kinases (44), including Eph receptors (45). p130 Cas is a docking protein that accumulates in focal adhesions, which also contain clustered integrins (46,47). Nck also associates with the focal adhesion kinase (48). Many tyrosine kinase receptors, including the insulin, platelet-derived growth factor and vascular endothelial growth factor receptors are also associated with integrins (49). Thus, Nck could, by binding to these molecules through an SH2 domain interaction and to R-Ras through an SH3 domain, bring R-Ras close to integrins. In addition, Nck may be able to bind R-Ras with its middle SH3 domain while simultaneously binding another protein with its other two SH3 domains. The cytoskeleton-associated proteins WASP and dynamin are possible candidates for such binding, because they interact with a different Nck SH3 domain (the C-terminal one) than R-Ras (50). How such interactions might contribute to cell adhesion regulation by R-Ras remains to be determined.