The RhoA-binding protein, rhophilin-2, regulates actin cytoskeleton organization.

The Rho GTPases regulate the actin cytoskeleton through interactions with various downstream effector molecules. Here we have identified a ubiquitously expressed human RhoA-binding protein, designated Rhophilin-2. Rhophilin-2 shows 40% amino acid similarity to human Rhophilin-1 and contains an N-terminal Rho-binding, a central Bro1-like, and a C-terminal PDZ domain. Glutathione S-transferase-capture experiments revealed that Rhophilin-1 and Rhophilin-2 interacted with both GDP- and GTP-bound RhoA in vitro. Despite the ability of Rhophilin-1 and Rhophilin-2 to interact with RhoA in a nucleotide-independent fashion, Rho-induced serum response element transcriptional activity was not altered by expression of either of these molecules. Although Rhophilin-2-expressing HeLa cells showed a loss of actin stress fibers, Rhophilin-1 expression had no noticeable effect on the actin cytoskeleton. Coexpression of Rhophilin-2 with a constitutively active Rho mutant reversed the disassembly phenotype, in which the coexpressing cells were more spread and less contracted than Rho alone-expressing cells. Expression of various Rhophilin-2 deletion and point mutants containing the N-terminal RhoA-binding domain but lacking other regions suggested that the disassembly of F-actin stress fibers was not simply caused by Rho sequestration. In addition, the Bro1 and PDZ domains of Rhophilin-2 were required for disassembly. RhoA activity assays also revealed that Rhophilin-2-expressing cells showed increased levels of RhoA-GTP suggesting that the Rhophilin-2-induced disassembly of stress fibers was not mediated by decreased RhoA activity. Based on the biochemical and biological activity, Rhophilin-2 may function normally in a Rho pathway to limit stress fiber formation and/or increase the turnover of F-actin structures in the absence of high levels of RhoA activity.

phorbol ester treatment (8). In addition to its effects on the actin cytoskeleton, Rho also participates in the assembly of focal adhesions following cell-matrix interactions (9 -11). Biochemical studies using the Rho-specific inhibitor C3 toxin, demonstrate that Rho signaling is involved in tyrosine kinase phosphorylation of both focal adhesion kinase and paxillin (12)(13)(14). One likely mechanism for focal adhesion assembly involves an early step in which Rho induces stress fiber formation. The Rho-induced stress fibers generate membrane tension that induces the aggregation of integrins, which in turn recruit other components of focal adhesions and promote their tyrosine phosphorylation (15).
The diversity of Rho-regulated cytoskeletal effects stems from the ability of Rho to interact with a large number of downstream effector proteins (16). The most well studied class of these RhoA effector proteins is the family of Rho-associated kinases (ROCKs) (otherwise known as ROKs or Rho-kinases). ROCKs regulate stress fiber formation (17)(18)(19)(20), cytokinesis (21), and myosin-based contractility (22). In cell culture studies, expression of a constitutively active ROCK mutant induces the formation of thick actin fibers and the assembly of focal adhesions (20,23). The non-kinase Rho effector protein, mDia1, also regulates stress fiber formation, apparently by recruiting, among others, the actin-monomer-binding protein profilin (24,25). Similar to ROCK, expression of constitutively active mutants of mDia1 also induces the formation of actin fibers; however, these actin fibers are thin compared with those induced by ROCK (23). Coexpression of constitutively active mutants of ROCK and mDia1 induces stress fibers and focal adhesions that closely resemble those caused by constitutively active RhoA (23). Like ROCK, mDia1 may also function in cytokinesis and can be detected at the cleavage furrow in mitotic Swiss-3T3 cells (26). Another RhoA effector protein, Citron kinase, also accumulates at the cleavage furrow and participates in the regulation of cytokinesis (27). Constitutively active Citron mutants block cytokinesis of HeLa cells resulting in multinucleated cells (27). These observations indicate that multiple Rho effector proteins may be involved in the regulation of stress fibers, focal adhesions, and cytokinesis induced by RhoA.
Much of what is known about how the effector proteins in Rho signaling pathways function have relied upon the generation and expression of constitutively active and dominant negative mutants. Biochemical and biological analyses of several Rho effector proteins, including ROCK (20), Citron (27), and mDia1 (23), demonstrate that they exist in an inactive state unless they are bound to Rho. The inactive state of these effectors is maintained by intramolecular autoinhibitory interactions, which are interrupted by the binding of GTP-bound Rho, thus inducing conformational changes that expose their active domains. Deletion analysis has identified regions within these Rho effectors that are essential for this autoinhibition. For example, deletion of the N-terminal, Rho-binding domain of ROCK (20) and C-terminal, Rho-binding domain of Citron (27) generates mutants with constitutively active kinase activity. Similarly, deletion of the Rho-binding domain within the N terminus of mDia1 results in constitutive activation of this non-kinase effector resulting in the formation of thin actin stress fibers (23). Interestingly, overexpression of the C-terminal autoinhibitory domain of mDia1 alone in mammalian cells can induce stress fiber formation by activating endogenous mDia1 (28). These results suggest that disruption of autoinhibitory interactions by the binding of GTP-bound Rho is a common mechanism of regulation of these effector proteins.
Here we have identified and characterized a novel human RhoA-binding protein, related to mouse Rhophilin-1 (29), which we have named Rhophilin-2. Structurally, Rhophilin-2 contains at least three distinct regions that are homologous to the previously characterized Rho-binding, Bro1, and PDZ (PS.D.-95, Disc-large, ZO-1) domains. Northern analysis revealed that Rhophilin-2 mRNA was expressed in all human tissues examined. Studies in NIH-3T3 fibroblasts showed that expression of Rhophilin-2 or the related molecule, Rhophilin-1, did not alter Rho-induced SRE 1 transcriptional activity. Expression of Rhophilin-2, but not Rhophilin-1 in HeLa cells, caused the loss of stress fibers. This disassembly phenotype induced by Rhophilin-2 was reversed by coexpression with RhoA. Examination of HeLa cells overexpressing various deletion mutants of Rhophilin-2 showed that the Rho-binding, Bro1, and PDZ domains are required for the disassembly of F-actin. RhoA activity assays also revealed that Rhophilin-2expressing cells showed increased levels of RhoA-GTP suggesting that the Rhophilin-2-induced disassembly of stress fibers was not mediated by decreased RhoA activity. Taken together these results suggest that Rhophilin-2 may act as a scaffold protein that limits stress fiber formation or increases the turnover of F-actin in the absence of high levels of RhoA signaling activity.

EXPERIMENTAL PROCEDURES
Identification and Cloning of Rhophilin-2 and Rhophilin-2␤-A human melanocyte cDNA clone encoding a protein with a potential Rho-binding domain was identified by a TBLASTN search of the expressed sequence tag (EST) database at the National Center for Biotechnology Information using the 13-amino acid Rho-binding domain of PKN1 (ELKLKEGAENLRR) as the query. A 1.7-kb human cDNA clone containing this EST (unique IMAGE Consortium identifier 249775 and GenBank TM entry H85494) was obtained and sequenced using a combination of manual sequencing with Sequenase (United States Biochemicals, Cleveland, OH) and automated fluorescent sequencing using sequence-deduced oligonucleotide primers. Using the entire 1.7-kb nucleotide sequence as query, a more abundant, second set of distinct EST clones was identified by additional database searching. DNA sequencing of one of these clones (unique IMAGE Consortium identifier 724324 and GenBank TM entry AA410792) revealed that it was identical to the first clone except for a C-terminal 534-bp segment. We designated the larger clone Rhophilin-2, because it encodes a protein similar to mouse Rhophilin over its entire coding sequence, whereas the shorter form was designated Rhophilin-2␤.
Because both Rhophilin-2 isoforms lacked the 5Ј-end, a rapid amplification of cDNA ends (RACE) PCR strategy was employed. The upstream primer (5Ј-GTTGAACAGGACACTGGCCTTCTC C-3Ј) and the downstream adaptor primer AP1 (5Ј-CCATCCTAATACGACTCACTAT-AGGG C-3Ј) were used in the first round of PCR using cDNA from kidney (Clontech Laboratories, Palo Alto, CA). The conditions of PCR were 30 cycles of 94°C for 45 s, 65°C for 45 s, and 72°C for 2 min. A second round of nested PCR used 1 ⁄50 of the first round reaction product as a template. The new upstream primer (5Ј-ACAAAGTCCAGCTG-GATG-3Ј) and second adaptor primer AP2 (5Ј-ACTCACTATAGGGCTC-GAGCGGC-3Ј) were used with a PCR program of 30 cycles consisting of 94°C for 45 s, 58°C for 45 s, and 72°C for 2 min. Positive clones potentially encoding the 5Ј-end of Rhophilin-2 were identified using restriction enzyme digestion and confirmed through DNA sequencing. The sequence of human Rhophilin-2 and Rhophilin-2␤ are available from GenBank TM under accession number AF268032.
Northern Analysis-A multiple tissue Northern blot containing immobilized poly(A ϩ ) RNA from several human tissues was obtained from Clontech Laboratories (Palo Alto, CA). A probe common to both isoforms derived from the 3Ј-end of Rhophilin-2 was linearized and used in a T3 RNA polymerase transcription reaction with [ 32 P]UTP (3000 Ci/mmol). The resulting high specific activity, antisense riboprobe was hybridized at 65°C under stringent conditions. Following hybridization, the blot was washed and exposed to x-ray film for 72 h.
GST-Capture Experiments-GST-capture experiments were performed as described previously (31). Briefly, GST or GST-RhoA fusion proteins were expressed in Escherichia coli, purified on glutathioneagarose resin, and loaded with guanosine 5Ј-3-O-(thio)triphosphate or GDP. N-terminal epitope-tagged mammalian expression vectors for Rhophilin-1, Rhophilin-2, and Rhophilin-2 mutants were transfected into Cos1 cells using FuGENE 6 transfection reagent (Roche Molecular Biochemicals) as described by the manufacturer. Forty-eight h later, cell lysates were prepared and used in binding experiments with immobilized GST recombinant proteins. Following washing, bound proteins were eluted and separated using an 8% SDS-PAGE gel. Proteins were then transferred electrophoretically to nitrocellulose and probed with anti-FLAG M2 monoclonal antibody. The FLAG-tagged Rhophilin proteins were then detected by incubating with rabbit anti-mouse horseradish peroxidase followed by incubation with enhanced chemiluminescence reagents (Pierce) and exposure to x-ray film.
SRE Reporter Assays-NIH-3T3 fibroblasts were maintained in Dulbecco's modified Eagle's medium supplemented with 10% calf serum and seeded to a density of 80,000 cells per well into 12-well culture dishes. NIH-3T3 fibroblasts were transfected with a serum response element luciferase reporter plasmid (Stratagene) containing four SRE sites. SRE-luciferase reporter plasmid (500 ng) was cotransfected with the plasmid for cytomegalovirus-Renilla luciferase (1 ng) and with 500 ng of pCAF parental vector, Rhophilin-1, or Rhophilin-2. In the RhoA-V14 cotransfection experiments, SRE-luciferase reporter plasmid (500 ng) was cotransfected with the plasmid for cytomegalovirus-Renilla luciferase (1 ng) and with 250 ng of pCAF, Rhophilin-1, or Rhophilin-2 and 250 ng of either pcDNA or pcDNA-RhoA-V14. The same total amount of DNA was used in each transfection, which were performed using the LipofectAMINE Plus reagent (Invitrogen) according to the manufacturer's instructions. 24 h post-transfection the medium was removed and incubated in Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum. After 18 h, designated cells were pulsed for 3 h with Dulbecco's modified Eagle's medium containing 10% calf serum. Luciferase activity was measured using the Dual Luciferase TM reporter assay system (Promega) and measured on a LB9501 Berthold luminometer. Transfection experiments were performed in triplicate, and SRE-luciferase activity is shown from two independent experiments.
Indirect Immunofluorescence-Immunofluorescence of HeLa cells was performed essentially as described previously (30) except that FuGENE 6 transfection reagent was used. At 24 h post transfection, the 1 The abbreviations used are: SRE, serum response element; EST, expressed sequence tag; FITC, fluorescein isothiocyanate; GEF, guanine exchange factors; GST, glutathione S-transferase; RACE, rapid amplification of cDNA ends. cells were fixed, permeabilized, and immunostained with either monoclonal or polyclonal antibodies to detect the FLAG epitope in Rhophilin-2-transfected cells. Following washes in phosphate-buffered saline, the coverslips were incubated with secondary antibodies as indicated in the figure legends. F-actin staining in permeabilized cells was performed using Texas Red-conjugated phalloidin (Sigma). Cotransfection experiments were also performed with Myc-tagged RhoA-V14. Confocal microscopy was performed with an Olympus Fluoview confocal microscope (Olympus America, Inc., Melville, NY) attached to an Olympus 1 ϫ 70 inverted fluorescent scope equipped with a 60ϫ oil immersion lens. Digitalized images were captured using Fluoview software (Olympus America, Inc.).
RhoA Activity Assays-RhoA activity assays were conducted as described previously (32,33). Briefly, Cos1 cells were transfected with the appropriate vector using the FuGENE 6 transfection reagent (Roche Molecular Biochemicals) as described by the manufacturer. Twentyfour h post-transfection cells were lysed with buffer containing 50 mM Tris, pH 7.5, 500 mM NaCl, 10 mM MgCl 2 , 1% Triton X-100, 0.1% SDS, 10 g/ml each of leupeptin and aprotinin, and 1 mM phenylmethylsulfonyl fluoride. Cell lysates were clarified by centrifugation at 13,000g at 4°C for 10 min and then incubated immediately with 30 l (20 g) of GST-Rhotekin-RBD (Upstate Biotechnology, Lake Placid, NY) for 45 min at 4°C. A 20-l aliquot of supernatant was also saved for determination of total RhoA levels in each sample. Following incubation, the beads were spun down and washed three times in wash buffer containing 0.1% Triton X-100 and 5 mM MgCl 2 in phosphate-buffered saline. Following washing, bound proteins were eluted and separated using a 4 -20% SDS-PAGE gel. Proteins were then transferred electrophoretically to nitrocellulose and probed with anti-RhoA polyclonal antibody (Upstate Biotechnology, Lake Placid, NY) at 3 g/ml overnight at 4°C. RhoA protein were then detected by incubating with a goat anti-rabbit horseradish peroxidase followed by incubation with enhanced chemiluminescence reagents (Pierce) and exposure to x-ray film.

Identification, Cloning, and Sequencing of Rhophilin-2-
Several well characterized Rho effector proteins contain a short amino acid sequence that enables them to bind directly to GTP-bound Rho (34). In an effort to identify additional Rho effector proteins, we searched the human EST database for sequences similar to the Rho-binding domain of human PKN1 (29,35). DNA sequencing of the single clone identified in this search (GenBank TM accession number H85494) revealed a 1.6-kb open reading frame. Subsequent searches of the EST database using the DNA sequence of this putative Rho-binding domain as the query identified multiple additional cDNA clones that could encode an alternatively spliced, longer isoform. DNA sequencing of several of these clones confirmed that they all encode a longer isoform, which we have designated Rhophilin-2, because it is highly homologous to mouse Rhophilin, which we now call Rhophilin-1 (29). The shorter isoform, designated Rhophilin-2␤, is identical to Rhophilin-2 except that it is missing a 178-amino acid region in the C-terminal half of the larger form (Fig. 1). A GenBank TM database search revealed that both isoforms are encoded in a bacterial artificial chromosome clone (CTC-263F14; GenBank TM accession number AC011449) that maps to human chromosome 19. Sequence analysis of this bacterial artificial chromosome clone revealed that the 178 amino ac-ids missing from the potential Rhophilin-2␤ isoform are encoded by five exons (data not shown).
Because all of the identified cDNAs for Rhophilin-2 were missing part of the 5Ј-coding sequences, a 5Ј-RACE strategy was employed. Using human kidney cDNA as template, an additional 186 bp of sequence were cloned thereby producing full-length coding sequences for both Rhophilin-2 and Rhophilin-2␤ (data not shown; GenBank TM accession number AF268032). Analysis of the deduced amino acid sequences revealed that Rhophilin-2 contains 685 amino acids, whereas Rhophilin-2␤ contains 507 amino acids (Fig. 1).
Potential Signaling Domains in Rhophilin Proteins-We have also identified a second human Rhophilin cDNA by searching the human EST database with the mouse Rhophilin-1 cDNA (29). Sequencing (GenBank TM accession number AY082588) suggests that it encodes a human homologue of mouse Rhophilin-1 (29), because these proteins show 75% similarity with each other. Comparison between mouse and human Rhophilin-1 with human Rhophilin-2 revealed that they are similar in size and have three domains in common ( Fig. 2A). Rhophilin-1 and Rhophilin-2 contain a single Rho-binding motif (Fig. 2B), which is known to be capable of high affinity direct binding to Rho but not to other GTPases such as Cdc42, Rac, or Ras (34). Both human Rhophilin proteins also have a central domain of ϳ200 amino acids that are significantly similar to one another ( Fig. 2A) and to Bro1 domains in other proteins (data not shown). Although the biological and biochemical functions of Bro1 domains are not known, the first Bro1-containing protein identified, from Saccharomyces cerevisiae, is linked genetically to a protein kinase C signaling pathway that regulates osmoregulation (36). Bro1 domains have been identified subsequently in a variety of other signaling proteins, including a phosphatase (37), proteins involved in cell death (38,39), pH regulation (40), and oocyte maturation (41).
The greatest amino acid difference between Rhophilin-1 and Rhophilin-2 is observed in the C terminus ( Fig. 2A). Although both Rhophilin-1 and Rhophilin-2 contain PDZ domains, the PDZ domains between both Rhophilin-1 and Rhophilin-2 show limited homology (Fig. 2C). As has been shown for other PDZcontaining signaling molecules (42), the PDZ domains in Rhophilin-1 and Rhophilin-2 could mediate protein-protein interactions and/or play a role in the assembly of larger protein complexes. Several proteins involved in Rho GTPase signaling also contain similar PDZ domains, including Rho effector proteins and Rho guanine exchange factors (GEFs) (Fig. 2C). Two of these GEFs, p-115/PDZRhoGEF (43,44) and , activate Rho and lead to stress fiber formation, whereas one of the effectors, Citron, is involved in cytokinesis (27). In addition, a sperm-specific protein called Ropporin interacts with the PDZ domain of mouse Rhophilin-1 (46).
Tissue Distribution of mRNA Encoding Rhophilin-2-In Northern analysis using a probe complementary to both the Rhophilin-2 and Rhophilin-2␤ mRNA species, we detected a major 3.8-kb transcript in all tissues examined (Fig. 3). We also FIG

Rhophilin-2 Regulates Actin Cytoskeletal Organization
utilized an RT-PCR strategy with primers that could amplify both Rhophilin-2 and Rhophilin-2␤ mRNA transcripts to determine which protein isoform would be synthesized. In all tissues examined, including lung, heart, kidney, liver, and colon, we detected the full-length Rhophilin-2 transcript but not the Rhophilin-2␤ transcript (data not shown). These results suggest that the Rhophilin-2 transcript represents the 3.8-kb species found ubiquitously in various human tissues.
Rhophilin-2 Binds to Both GTP-and GDP-bound RhoA-Previously, a recombinant protein containing part of mouse Rhophilin-1 was shown to bind RhoA in a GTP-dependent fashion (29). In light of these findings and the fact that the Rho-binding domain of all the Rhophilin homologues are quite similar, we were interested in determining whether Rhophilin-2 bound RhoA in a GTP-dependent fashion and whether we could generate a specific Rho-binding deficient mutant. For these experiments we generated N-terminally FLAG-tagged mammalian expression vectors for Rhophilin-1, Rhophilin-2, and a potential Rhophilin-2-Rho-binding mutant (Rhophilin-2-E55A,N56A) and expressed these constructs in HeLa cells. As a positive control, we utilized lysates from Myc-tagged ROCKtransfected cells in binding experiments and observed only GTP-dependent binding (Fig. 4). Using the different Rhophilin cell lysates in GST-capture assays revealed that both Rhophilin-1 and Rhophilin-2 bound to GST-RhoA-GTP and GST-RhoA-GDP but not detectably to GST (Fig. 4). Interestingly, we depleted GTP/GDP nucleotide bound to RhoA using a published method (47) and still observed Rhophilin-2 binding to RhoA (data not shown). This nucleotide-independent binding of Rhophilin-2 for RhoA was specific for this GTPase, because no binding of Rhophilin-2 was observed with Rac1 or RhoE loaded with either GDP or GTP (data not shown). Finally, a Rhophilin-2 mutant (Rhophilin-2-E55A,N56A), containing two alanine substitutions within two conserved residues of the Rho-binding domain, showed no detectable binding to RhoA-GTP or RhoA-GDP (Fig. 4).

Rhophilin-1 and Rhophilin-2 Do Not Alter Rho-induced SRE Transcriptional Activity-Previous studies have identified
RhoA (48) and a number of RhoA effector proteins including PKN (49), Rhotekin (50), and mDia (51) as activators of SRE transcriptional activity. To test whether Rhophilin-1 and Rhophilin-2 might alter SRE activity, we performed SRE-luciferase reporter assays in NIH-3T3 cells. In these experiments, cells were transfected with vector control, Rhophilin-2, or Rhophilin-1 constructs and then cultured in 0.5% serum. Under these conditions, a 3-h serum treatment of cells transfected with control vector led to a small induction of SRE activity (Fig.  5A). In these experiments, the control pCAF vector blocked serum-induced SRE activity as compared with the pcDNA vector control, which showed a 7-fold increase in SRE activity following serum induction (Fig. 5B). Rhophilin-1 and Rhophlin-2 did not alter SRE activity in low serum but increased SRE activity in the presence of serum (Fig. 5A).
Cells Expressing Rhophilin-2, But Not Rhophilin-1, Show Actin Disassembly-Next we examined the cellular distribution and biological effects of Rhophilin-2 and Rhophilin-1 on the actin cytoskeleton. In HeLa cells, epitope-tagged Rhophilin-2 was detected as diffuse cytoplasmic staining (Fig. 6A). Examination of the F-actin distribution in these Rhophilin-2transfected HeLa cells showed a marked loss of F-actin from stress fibers and the cortical membrane region (Fig. 6B). In many of the transfected HeLa cells there was an increase in short protrusions emanating from the cell surface. Expression of Rhophilin-2 in Cos1 cells showed a similar disassembly phenotype, whereas in primary human keratinocytes there was a loss of F-actin from adherens belt (data not shown). Additionally we tested whether the Rhophilin-2 disassembly phenotype was similar to that induced by C3. In HeLa cells, C3 overexpression resulted in a general cell rounding, in addition to disassembly of F-actin (data not shown). The disassembly phenotype induced by Rhophilin-2 was similar to C3 with respect to loss of polymerized actin; however, the numerous cellular projections were not seen in C3-overexpressing cells. To test whether this F-actin disassembly phenotype was unique to Rhophilin-2, we also used a mammalian expression vector to express human Rhophilin-1 in HeLa cells. Although Rhophilin-1 was highly expressed in these cells (Fig. 6C), no alteration in the F-actin cytoskeleton was observed in these expressing cells (Fig.  6D). These results suggest that Rhophilin-1 and Rhophilin-2 may have distinct roles in regulating the actin cytoskeleton.
RhoA Expression Can Rescue the Rhophilin-2-induced Disassembly Phenotype-Because Rhophilin-2 can bind RhoA, it seemed possible that Rhophilin-2 could cause a loss of F-actin stress fibers by sequestering Rho and thus inhibiting its activity. To test this possibility, HeLa cells were transfected with a constitutively active RhoA mutant (RhoA-V14) alone or in combination with Rhophilin-2. Cells overexpressing RhoA were contracted highly (Fig. 7A) and showed numerous stress fibers (Fig. 7B). Although cells overexpressing Rhophilin-2 and RhoA-V14 showed an abundance of F-actin stress fibers, these cotransfected cells were more spread and less contracted (Fig. 7,  C and D). Additionally, we observed a similar phenotype with wild-type RhoA expression experiments (data not shown). These results suggest that RhoA expression can rescue the Rhophilin-2 disassembly phenotype and additionally suggest that Rhophilin-2 might modify the biological activity of RhoA by sequestering the GTPase or altering some downstream pathway.
The Rho-binding, Bro1, and PDZ Domains Are Required for Inducing the Actin Disassembly Phenotype-To further examine the mechanism of Rhophilin-2-induced actin disassembly, we generated a variety of deletion and point mutants of Rhophilin-2. Immunofluorescence and Western blot analysis of HeLa cells transfected with these mutants revealed that they were expressed at relatively similar levels and migrated as expected for their predicted molecular weights (data not shown). As might be predicted, the ability of the different mutant proteins to bind RhoA required an intact Rho-binding domain (Fig. 8). For example, one Rhophilin-2 deletion mutant missing the N-terminal RhoA-binding domain, Rhophilin-2-⌬1, did not bind to RhoA (Fig. 8). In contrast, three Rhophilin-2 deletion mutants (Rhophilin-2-N1, Rhophilin-2-⌬Bro1, and Rhophilin-2-⌬2) and one Rhophilin-2 point mutant (Rhophilin-2-R517A,L525A, G526A) that all contained an intact RhoAbinding domain showed specific Rho-A binding activity (Fig. 8). Taken together these results suggest that Rhophilin-2 binds both GTP-and GDP-bound RhoA and requires an intact Rhobinding sequence for this binding activity in vitro.
The various deletion and point mutants of Rhophilin-2 were FIG. 3. Rhophilin-2 is expressed ubiquitously. Northern blot analysis of poly(A ϩ ) RNA from a variety of human tissues using a probe derived from the 3Ј-untranslated region of Rhophilin-2. A 3.8-kb transcript corresponding to Rhophilin-2 was detected in all tissues examined. FIG. 4. Rhophilin-1 and Rhophilin-2 bind both GTP-and GDPbound RhoA. Cell extracts of HeLa cells transfected with FLAG epitope-tagged Rhophilin-1, FLAG epitope-tagged Rhophilin-2, FLAG epitope-tagged Rhophilin-2-E55A,N56A, or Myc epitope-tagged ROCK were prepared and incubated with immobilized GST or GST-RhoA loaded with either guanosine 5Ј-3-O-(thio)triphosphate or GDP. Following washing, bound proteins were analyzed by Western blot using the M2 mouse anti-FLAG TM or anti-Myc 9E10 monoclonal antibody followed by an anti-mouse horseradish peroxidase secondary antibody. The blots were developed using enhanced chemiluminescence reagents (Sigma). One-twentieth of the inputs are also shown for each of the sets of experiments.

Rhophilin-2 Regulates Actin Cytoskeletal Organization
tested next for their ability to disrupt the actin cytoskeleton. Expression of a double point mutant (Rhophilin-2-E55A,N56A) that was unable to bind RhoA in vitro did not induce disassembly (Fig. 9, A and B). A deletion mutant, Rhophilin-2-⌬1 (containing amino acids 123-685), lacking a functional Rho-binding domain was also unable to bind RhoA in vitro and did not induce F-actin disassembly (see Fig. 8). The phenotype of these mutants suggested the possibility that the N-terminal Rho-binding domain of Rhophilin-2 might be responsible for the disassembly phenotype by sequestrating RhoA,; however, characterization of additional mutants suggested that multiple regions of Rhophilin-2 were required. Cells expressing the Nterminal Rho-binding domain fragment (Rhophilin-2-N1 deletion mutant, containing amino acids 1-157) or a C-terminal truncation mutant of Rhophilin-2 (Rhophilin-2-⌬2, containing amino acids 1-588) did not induce actin disassembly in vivo   FIG. 5. Rhophilin-1 and Rhophilin-2 do not alter RhoA-induced SRE activity significantly. A, NIH-3T3 fibroblasts were cotransfected with the SRE-luciferase reporter and 0.5 g of expression vectors for pCAF, pCAF-Rhophilin-1, and pCAF-Rhophilin-2 as indicated under "Experimental Procedures." -Fold induction was determined as the activity of the reporter construct with respect to vector controls. The mean of three experiments is represented. Error bars indicate standard deviation. B, NIH-3T3 fibroblasts were cotransfected with the SRE-luciferase reporter and 0.5 g of expression vectors for pcDNA-RhoA-V14, pCAF-Rhophilin-1, and pCAF-Rhophilin-2 as indicated under "Experimental Procedures." -Fold induction was determined as the activity of the reporter construct with respect to vector controls. The mean of three experiments is represented. Error bars indicate standard deviation.
FIG. 6. Rhophilin-2, but not Rhophilin-1, induces the loss of F-actin structures. Cells were transfected with either FLAG-tagged Rhophilin-2 (A and B) or FLAG-tagged Rhophilin-1 (C and D). Twenty-four h after transfection, the cells were fixed, permeabilized, and stained. Rhophilin-2 (A) and Rhophilin-1 (C) were detected with a mouse monoclonal anti-FLAG antibody followed by staining with a FITC-conjugated goat anti-mouse antibody. F-actin was detected using Texas Red-conjugated phalloidin (B and D). even though both proteins bound RhoA effectively (see Fig. 8).
We also tested whether mutations within the Bro1 and PDZ domains would affect the ability of Rhophilin-2 to induce the disassembly phenotype. In the case of the Bro1 mutant, we generated an in-frame Rhophilin-2 deletion mutant missing amino acids 185-302 within the Bro1 domain, whereas a triple point mutant (Rhophilin-2-R517A,L525A,G526A) was generated within the PDZ domain. Cells overexpressing the Rhophilin-2-⌬Bro1 (Fig. 9, C and D) and Rhophilin-2-R517A,L525A,G526A mutants were unable to induce the actin disassembly phenotype (Fig. 9, E and F), although both mutants still bound RhoA in vitro (see Fig. 8). These results suggest that at least three different regions of Rhophilin-2 appear to be required for inducing the disassembly phenotype: the Rho-binding, the Bro1, and the C terminus containing the PDZ domain.
Rhophilin-2 Increases Cellular Levels of GTP-bound RhoA-To formally disprove the possibility that Rhophilin-2 was altering levels of GTP-bound RhoA in vivo, we measured the levels of active RhoA in cells using the Rhotekin-affinity precipitation assay (32,33). Cos1 cells were transiently transfected with vector control, pCAF-Rhophilin-2, or pCAF-Rhophilin-2-R517A,L525A,G526A, and transfection efficiency was estimated around 30%. Using these lysates, similar amounts of total RhoA were observed in cells expressing the different expression vectors (Fig. 10). Interestingly, Rhophilin-2 increased the amount of RhoA-GTP compared with the vector control (Fig. 10). The triple PDZ mutant of Rhophilin-2 also increased the amount of GTP-bound RhoA but to a lesser extent than the wild-type Rhophilin-2 (Fig. 10). These results suggest that Rhophilin-2-induced actin disassembly does not involve lowering of RhoA-GTP levels.

DISCUSSION
The presence of a large number of RhoA effector proteins suggests the possibility that some of these effectors might cooperate to regulate Rho-induced actin cytoskeletal changes (52). In fact, studies with two RhoA effector proteins, ROCK and mDia1, show that they cooperate to increase induction of Rho-like stress fibers (23). In this article, we describe the identification and characterization of human Rhophilin-2, a new member of the Rhophilin family of Rho-binding proteins expressed ubiquitously. The domain structure of human Rhophilin-2 and Rhophilin-1 share 49% amino acid similarity. The amino acid similarity within the Rho-binding domain of Rhophilin-1 and Rhophilin-2 also correlates with their observed RhoA binding properties. GST-capture experiments showed that Rhophilin-1 and Rhophilin-2 bound to both GDPand GTP-bound RhoA but not to other related GTPases such as Rac1 or RhoE. Furthermore, a double point mutation in the Rho-binding site of Rhophilin-2 effectively blocked binding to both GDP-and GTP-bound RhoA. Although these binding re-FIG. 7. Rhophilin-2-induced loss of F-actin structures is reversed by coexpression of Rho. Cells were transfected with Myc-tagged Rho-V14 (A and B) or Myc-tagged Rho-V14 plus FLAGtagged Rhophilin-2 (C and D). Twentyfour h after transfection, the cells were fixed, permeabilized, and stained. Myctagged Rho-V14 was detected with a mouse monoclonal anti-Myc antibody followed by staining with a FITC-conjugated goat anti-mouse antibody (A). In the cotransfected cells, Rhophilin-2 was detected with a rabbit polyclonal anti-FLAG antibody followed by staining with FITCconjugated goat anti-rabbit antibody (C), and Myc-tagged Rho-V14 was detected with a mouse monoclonal anti-Myc antibody followed by staining with a Cy5-conjugated goat anti-mouse antibody (data not shown). Texas Red-conjugated phalloidin was used to detect F-actin (B and D).
FIG. 8. Activity of Rhophilin-2 and its mutants. The structures of Rhophilin-2 and mutant constructs are shown. All constructs used in the transient transfection studies contain a FLAG epitope tag at the N terminus. For the Rho-binding assays the various constructs were transfected into HeLa cells, and cell lysates were prepared and tested for RhoAbinding using a GST-capture approach as described in the legend for Fig. 4. Specific RhoA binding is denoted by a plus sign. F-actin disassembly was scored as positive when greater than 80% of the transfected cells showed a lack of stress fibers as detected by Texas Red-conjugated phalloidin staining.
sults contrast studies of mouse Rhophilin-1, which showed that a bacterially expressed fragment of mouse Rhophilin-1 interacted only with GTP-bound RhoA in an overlay assay (29), these results are consistent with two studies showing that PKN, containing a similar Rho-binding domain, can bind both GDP and GTP-bound forms of RhoA. However, we cannot exclude the possibility that Rhophilin-1 and Rhophilin-2 interact only with the GTP-bound form of RhoA in vivo or that binding to the GDP-bound RhoA has only regulatory function. These data also add Rhophilin-1 and Rhophilin-2 to the list of known effector proteins, such as PRK2 (53), the 68-kDa human erythrocyte phosphatidylinositol 5-kinase type 1 (54), and IQGAP1 (55), that interact with both GDP-and GTP-bound Rho GTPases.
Transfection experiments revealed that Rhophilin-2, but not Rhophilin-1, induces the disassembly of F-actin structures in many different cell types. To help define the mechanism by which Rhophilin-2 induces actin disassembly, we examined a number of Rhophilin-2 mutant proteins for RhoA binding and for effects on cytoskeletal organization. The observation that Rhophilin-2 mutant proteins containing the N-terminal Rhobinding sequence but missing other regions failed to induce the disassembly phenotype strongly suggests that a simple mechanism such as RhoA sequestration is not sufficient to explain the disassembly phenotype. Our results are also in contrast to studies with two other RhoA-interacting proteins, mDia1 (23) and Nir2 (47), where overexpression of deletion mutants containing only the Rho-binding domain, but not overexpression of full-length proteins, was able to induce F-actin stress fiber disassembly. Further proof for Rhophilin-2 exerting its actin disassembly effect was demonstrated by RhoA-GTP activation assays that showed increased levels of RhoA-GTP in response to Rhophilin-2 expression. Although we do not know the reason for the increase in RhoA-GTP levels with Rhophilin-2 expression, it may be a compensatory mechanism in response to the loss of stress fibers similar to that observed with cytochalasin D and colchicine treatment (32). Furthermore, Rhophilin-2 had no obvious effect on RhoA-induced SRE transcriptional activity. The ability of Rhophilin-2 to disrupt the actin cytoskeleton specifically, but not alter Rho-induced transcriptional activity, contrasts other agents such as C3 toxin (48) or a dominant negative Rho mutant (56) that inhibit both stress fiber formation and transcriptional activity.
Based on our deletion and point mutants, we propose that multiple regions within Rhophilin-2 are involved in stress fiber disassembly consistent with Rhophilin-2 acting as a scaffold protein. The deletion mutant within the Bro1 domain of Rhophilin-2 suggests that this region is involved in stress fiber disassembly. Interestingly, several studies have now linked Bro1 domain-containing proteins such as Bro1/Npi3 (57), Rim20P (58), and ALG-2 (59, 60) to signal transduction-regulated proteolysis. Presently, we are exploring the possibility that Rhophilin-2-mediated stress fiber disassembly may involve proteolysis of signaling or cytoskeletal components. Additionally, we propose a critical requirement for the PDZ domain in Rhophilin-2 for this activity, which likely involves protein-protein interactions (42). This inference is based on our studies showing that overexpression of a Rhophilin-2 PDZ domain triple point mutant was unable to cause F-actin disassembly. Using a yeast two-hybrid screen, ␣-actinin-4, a major actin-binding and -bundling domain and known PDZ domaininteracting protein (61), was identified as an in vitro binding partner for Rhophilin-2, and additional studies showed that the Rhophilin-2 PDZ domain triple point mutant was unable to interact with ␣-actinin-4. 2 Further studies are aimed at validating and determining the biological significance of the Rhophilin-2-␣-actinin-4 interaction.
Recent results show that a Drosophila knock-out of the single Rhophilin gene is embryonic lethal; this suggests an important biological function of Rhophilin (62). From the work presented here, we propose that the two human Rhophilin members, Rhophilin-1 and Rhophilin-2, have different activities. Although the exact normal function of Rhophilin-2 is not clear, one function of Rhophilin-2 may be to limit stress fiber formation in the absence of high levels of activated RhoA. Alternatively, Rhophilin-2 may normally be activated by RhoA and promote actin stress fiber disassembly, which may be required in vivo for biological processes such as cell motility 2 J. W. Peck and P. D. Burbelo, unpublished data.  and cytokinesis and whose activity may be normally masked by the action of other effector proteins (e.g. ROCK and Dia) that promote stress fiber assembly. Future studies are aimed at identifying additional protein targets in the cell responsible for mediating Rhophilin-2-induced cytoskeletal changes to clarify its normal function.