Biochemical Characterization of the Rho GTPase-regulated Actin Assembly by Diaphanous-related Formins, mDia1 and Daam1, in Platelets*

The diaphanous-related formins are actin nucleating and elongating factors. They are kept in an inactive state by an intramolecular interaction between the diaphanous inhibitory domain (DID) and the diaphanous-autoregulatory domain (DAD). It is considered that the dissociation of this autoinhibitory interaction upon binding of GTP-bound Rho to the GTPase binding domain next to DID induces exposure of the FH1-FH2 domains, which assemble actin filaments. Here, we isolated two diaphanous-related formins, mDia1 and Daam1, in platelet extracts by GTP-RhoA affinity column chromatography. We characterized them by a novel assay, where beads coated with the FH1-FH2-DAD domains of either mDia1 or Daam1 were incubated with platelet cytosol, and the assembled actin filaments were observed after staining with rhodamine-phalloidin. Both formins generated fluorescent filamentous structures on the beads. Quantification of the fluorescence intensity of the beads revealed that the initial velocity in the presence of mDia1 was more than 10 times faster than in the presence of Daam1. The actin assembly activities of both FH1-FH2-DADs were inhibited by adding cognate DID domains. GTP-RhoA, -RhoB, and -RhoC, but not GTP-Rac1 or -Cdc42, bound to both mDia1 and Daam1 and efficiently neutralized the inhibition by the DID domains. The association between RhoA and Daam1 was induced by thrombin stimulation in platelets, and RhoA-bound endogenous formins induced actin assembly, which was inhibited by the DID domains of Daam1 and mDia1. Thus, mDia1 and Daam1 are platelet actin assembly factors having distinct efficiencies, and they are directly regulated by Rho GTPases.

Small GTPase Rho proteins regulate various cellular functions such as stress fiber formation, cell shape change, and cytokinesis via rearrangement of the actin cytoskeleton (1). Like other GTPases, Rho acts as a molecular switch by converting between an inactive GDP-bound form and an active GTPbound form. GTP-bound Rho conveys signals by interacting with so-called effector molecules such as Rho-kinase, citron kinase, and protein kinase N (2)(3)(4)(5)(6)(7). Some of the formins classified as diaphanous-related formins (DRFs), 3 which include mDia, Daam, and FRL, are also proposed to be effector molecules for Rho (8).
Formin genes are conserved in species from yeast to mammals (9). All of the formins have domains referred to as forminhomology-1 (FH1) and FH2 domains (9). The FH2 domain is composed of ϳ400 amino acids and forms a donut-shaped ring structure upon homodimerization (10 -12). It directly nucleates actin assembly (13,14) and attaches persistently to the growing barbed ends of actin filaments, protecting these ends from the elongation inhibitory effect of conventional capping proteins. This activity is referred to as "leaky capping" or "processive capping" (13,15,16). The FH1 domain, located adjacent to the FH2 domain, is a 30 -150 proline-rich amino acid sequence that can interact with profilin (17,18). It enhances the actin filament elongation rate by facilitating the availability of globular-actin (G-actin) complexed with profilin (15, 19 -21).
In their resting state DRFs are autoinhibited by an intramolecular interaction between the N-terminal diaphanous inhibitory domain (DID) and the C-terminal diaphanous-autoregulatory domain (DAD) (20,22,23) (Fig. 1C). It is considered that the autoinhibitory interaction between DID and DAD is disrupted when the active GTP-bound Rho binds to the GTPase binding domain (GBD), which is localized at the N-terminal end next to DID, resulting in exposure of the FH1 and FH2 domains (24 -27). However, it has been reported that the active form of Rho, even at excess concentration, could only faintly restore the actin assembly activity of the FH2 domain in the pyrene actin assay when it was autoinhibited by GBD-DID (24).
Among DRFs, mDia1, a mammalian homolog of Drosophila Diaphanous, is the best characterized DRF in mammals (18, 20, 23, 24, 28 -33). In cultured cells, mDia1 co-localizes with Rho and profilin in membrane ruffles and cooperates with ROCK/ Rho-kinase to induce stress fibers in interphase cells (33). Another DRF, Dishevelled-associated activator of morphogenesis-1 (Daam1), has been identified as a binding protein of Dishevelled-2, a downstream signaling molecule for Frizzled that is implicated in signal transduction along the non-canonical Wnt pathway (34,35). Daam1 is reported to regulate the actin cytoskeletal reorganization (34,35), and it is required for proper development of Xenopus (34) and Drosophila (36).
Platelets play important roles in hemostasis and thrombosis, where Rho regulates their shape change (37), adhesion (37,38), and aggregation (39,40) of platelets. To further understand the molecular mechanism of Rho function in platelets, we performed affinity chromatography of a platelet lysate on a GTP-Rho column and identified two DRFs, mDia1 and Daam1. By means of a novel actin assembly assay using beads coated with FH1-FH2 domains, we demonstrated that both formins possess actin assembly activity in the cytosol and that the autoinhibition of both formins was relieved by active Rho GTPases.
Anti-human Daam1 rabbit and rat polyclonal antisera were raised against Daam1 CT, and the rabbit antibody was affinitypurified. Anti-mouse mDia1 rabbit polyclonal antiserum was a generous gift from Dr. S. Narumiya (Kyoto University, Kyoto, Japan). Anti-GST and anti-RhoA monoclonal antibodies were from Santa Cruz Biotechnology. Anti-His 6 monoclonal antibody was from Sigma. Horseradish peroxidase-linked anti-rabbit donkey antibody and anti-mouse sheep antibody (GE Healthcare) were used as secondary antibodies in the immunoblot analysis visualized by enhanced chemiluminescence method (GE Healthcare). Reagents used in this study were purchased from Sigma. The protein concentrations were determined by the Bradford method (Bio-Rad) or from the intensity of protein bands of Coomassie Blue-stained gel using bovine serum albumin as a standard.
Affinity Chromatography of Platelet Lysate Using Active RhoA-Platelet pellet, provided by Kyoto Red Cross Blood Center, was solubilized at 4°C in buffer A containing 0.5% (w/v) Triton X-100 and protease inhibitor mixture P8340 (Sigma) and centrifuged at 100,000 ϫ g at 4°C for 1 h. The supernatant (200 mg of protein) was incubated for 90 min at 4°C with glutathione beads coated with GST-RhoA (200 g) bound to GTP␥S (a non-hydrolysable GTP analog) or GDP, prepared as described (41). The beads were then washed five times with buffer A, and bead-associated proteins were analyzed by SDS-PAGE. Protein bands were excised from Coomassie-stained gels and analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry.
BIAcore Analysis-Kinetic binding analysis was performed using Biacore X (GE Healthcare). Purified recombinant Daam1 NT or mDia1 NT were immobilized onto CM5 sensor chips by amine coupling in 10 mM sodium acetate (pH 4.0), giving 1511.6 and 1673.2 response units, respectively. Subsequent binding experiments were performed in HBS-P buffer (GE Healthcare). The sensor chips were regenerated in 10 mM glycine-HCl (pH 2.0). Dissociation constants (K D ) were determined by fitting all curves at once with the Langmuir binding model using BIAevaluation software.
Actin Assembly Assay in Cytosolic Solution-The platelet pellet was sonicated in buffer A containing protease inhibitor mixture P8340 (Sigma) followed by centrifugation at 100,000 ϫ g at 4°C for 1 h. The supernatant was extensively dialyzed against buffer A at 4°C and stored as platelet cytosol at Ϫ80°C until use. GST-mDia1 CT, GST-Daam1 CT, or their mutants were incubated at 37°C for 10 min with platelet cytosol at 2 mg of proteins/ml. Assays were carried out in the presence of an ATP regeneration system (8 mM creatine phosphate, 50 g/ml creatine phosphokinase, 1 mM ATP). Each solution contained 0.6 M actin, as determined by immunoblot using an anti-actin monoclonal antibody (BD Biosciences Pharmingen) with purified actin as a standard. After incubation, the assembled actin filaments in the solution were adsorbed on the poly-D-lysine-coated coverslip, labeled with rhodamine-phalloidin, and observed by fluorescence microscopy Axioskop2 plus (Zeiss).
Actin Assembly Assay Using Formin-coated Beads and Platelet Cytosol-Glutathione beads coated with GST-mDia1 CT, GST-Daam1 CT, or their mutants were incubated at 37°C with platelet cytosol at 2 mg proteins/ml for 10 min in the presence of ATP regeneration system unless otherwise specified. In some experiments His 6 -tagged mDia1-NT or Daam1-NT and/or His 6 -tagged Rho GTPases loaded with GDP or GppNHp (a non-hydrolysable GTP analog) were also added. After incubation, beads were washed 5 times with buffer A and fixed with 4% formamide at 4°C for 5 min. The beads were then labeled with rhodamine-phalloidin and observed by fluorescence microscopy. Otherwise, the fluorescence intensity of the beads was measured using an Arvo SX 1420 multilabel counter (PerkinElmer Life Sciences). In the experiments shown in Fig.  7A, glutathione beads coated with GST-RhoA bound to GTP␥S or GDP were incubated with platelet lysate to isolate endogenous formins, and the actin assembly was evaluated as described above.
Association of mDia1 NT and Daam1 NT with Various Rho Family GTPases-Glutathione beads coated with GST-mDia1 NT and -Daam1 NT were incubated at 4°C for 2 h with His 6tagged various Rho family GTPases bound to GTP␥S or GDP in 200 l of buffer A containing 4 mg/ml bovine serum albumin. After washing the beads, bead-associated proteins were subjected to SDS-PAGE followed by immunoblot analysis using an anti-His 6 antibody (Sigma).
Association of mDia1 NT and Daam1 NT with Cognate CTs-Glutathione beads coated with GST-mDia1 CT and -Daam1 CT were incubated at 4°C for 2 h with His 6 -tagged mDia1 NT and Daam1 NT in 200 l of buffer A containing 4 mg/ml bovine serum albumin. After washing the beads, bead-associated proteins were subjected to SDS-PAGE followed by immunoblot analysis using an anti-His 6 antibody.
Analysis of the RhoA-Daam1 Interaction in Thrombin-activated Platelets-Platelets (2 ϫ 10 9 platelets/assay) were isolated from freshly obtained whole blood (40) and stimulated with 0.5 units/ml thrombin for various periods at 37°C. Samples were lysed (1:4 v/v) in ice-cold buffer A containing 0.5% Triton X-100 and a protease inhibitor mixture at 4°C for 5 min followed by centrifugation at 300,000 ϫ g for 5 min. The supernatants were incubated with protein A-agarose beads (Roche Applied Science) coated with anti-Daam1 polyclonal antibody at 4°C for 1 h. After washing the beads, bead-associated proteins were analyzed by immunoblotting with anti-RhoA monoclonal and anti-Daam1 rat polyclonal antibodies.

Identification of mDia1 and Daam1 as GTP-RhoA-binding
Proteins in Platelet Lysates-We have previously demonstrated that Rho regulates platelet aggregation (40). To further understand the mechanisms of Rho functions in platelets, we performed an affinity chromatography of a human platelet lysate using immobilized RhoA. As shown in Fig. 1A, two bands were detected as GTP-RhoA-binding proteins. Mass spectrometry analysis revealed that the 120-kDa protein was Daam1 and the 70-kDa protein was the N-terminal half of mDia1. In freshly prepared platelet lysate, the full-length Daam1 and mDia1 proteins were detected by Western blot at their expected molecular masses of 120 and 140 kDa, respectively (Fig. 1B). Thus, the detection of a smaller mDia1 species indicated it was proteolyzed during preparation of the platelet lysate. Both Daam1 and mDia1 belong to formin protein family, and the domain structures of Daam1 and mDia1 are shown in Fig. 1C (9). Both proteins possess the conserved FH1 and FH2 domains as well as N-terminal GBD and DID domains and a C-terminal DAD domain.
Kinetic Analyses of the Interaction of RhoA to Daam1 NT and mDia1 NT-To evaluate the affinity between Daam1 GBD and RhoA, we performed kinetic analyses using surface plasmon resonance with a Biacore X instrument. We purified His 6tagged RhoA and His 6 -tagged GBD-DID domain of Daam1, which we designated as Daam1 NT (Fig. 1C). The purities of the proteins used were more than 90% (Fig. 1D). The Daam1 NT protein was immobilized on a CM5 sensor chip, and GTP␥Sbound RhoA was used as the analyte. Kinetic analyses were performed with 10 -200 nM GTP␥S-RhoA on the Daam1 NTimmobilized chip. The dissociation constant (K D ) of the binding of GTP␥S-RhoA to Daam1 NT was determined to be 4.0 nM (Fig. 1E). Next, 2-8 M GDP-RhoA was used as the analyte, and K D of the binding of GDP-RhoA to Daam1 NT was determined to be 770 nM. Thus, the interaction between RhoA and Daam1 NT was GTP-dependent, and the affinity of RhoA in its GTPform with Daam1 is considerably high. Similarly, His 6 -tagged GBD-DID of mDia1 (designated as mDia1 NT) was purified (Fig. 1D), and K D of the binding of GTP␥S-RhoA to mDia1 NT and that of GDP-RhoA to mDia1 NT were determined to be 4.8 and 430 nM, respectively. The affinities of mDia1 NT to GTPbound RhoA and GDP-bound RhoA were similar to those of Daam1 NT.
Evaluation of Actin Filament Assembly by a Semiquantitative Assay Using Formin-coated Beads-The DID domains of DRFs have been demonstrated to suppress their actin assembly activity of their FH2 domains by intramolecular interaction with the DAD domains that are immediately on the C-terminal side of the FH2 domains (20). GTP-bound Rho GTPase is thought to relieve this inhibition by binding to the GBD adjacent to the DID domain. However, it has never been shown biochemically in vitro by assays using purified actin that physiological concentrations of Rho GTPases efficiently activate the formins (24). It has been proposed that some additional factor(s) is required for the full recovery of autoinhibited formin proteins (24). Here, we tried to establish an in vitro assay system capable of evaluating the actin assembly in crude cytosolic solutions to investigate the effect of Rho GTPases on relief of the autoinhibition.
We first examined whether we could detect actin assembly by formins in the crude cytosol. We purified recombinant GST-Daam1 FH1-FH2-DAD (Daam1 CT) and GST-mDia1 FH1-FH2-DAD (mDia1 CT) and incubated them individually with ATP and platelet cytosol. After incubation, we adsorbed the assembled actin filaments on the poly-lysine-coated coverslip and observed them by rhodamine-phalloidin staining. Both Daam1 CT and mDia1 CT induced filamentous structures, whereas GST alone had a much less pronounced effect ( Fig.  2A), indicating that we could observe cytosolic actin assembly in a cell-free system.
Next, we tried to assemble the actin filaments in cytosol using these formins immobilized on beads. As shown in Fig. 2B, beads coated with the GST-Daam1 CT and GST-mDia1 CT, but not GST-coated beads, were uniformly stained by rhodamine-phalloidin. Many linear filamentous structures were observed on the beads coated with both mDia1 CT and Daam1 CT (Fig. 2B). Without the addition of ATP, those filamentous structures were not observed (data not shown). The addition of 2 M cytochalasin D and 5 M latrunculin A, both monomeric actinsequestrating agents, to cytosol completely abolished the fluo-rescence on the beads coated with either Daam1 or mDia1 (data not shown). These results suggest that actin assembly was actually induced on the formin CT-coated beads. Quantification of the fluorescence intensity revealed time-dependent increases in actin assembly. The initial velocity of mDia1 CT exceeded that of Daam1 CT by more than 10-fold (Fig. 2C), similar to the results obtained using a conventional pyrene-actin assay as shown previously by others (10) and by us (12).
To confirm that the fluorescence intensity of the stained beads reflected the actin assembly activity of formins, we examined the activity of several Daam1 CT mutants using the bead assay. The FH2 domain forms a reciprocally tethered homodimer structure with the monomers connected by a flexible "linker," a subdomain within FH2 consisting of ϳ40 residues. We recently showed that the length of the linker region affects the actin assembly activity of the FH2 domain (12). In that study we examined the actin assembly activity of several Daam1 CT mutants having various linker lengths (Fig. 3A) using the pyrene actin assay and obtained the following results. 1) ⌬23 and ⌬29 mutants completely lost their assembly activity, whereas ⌬11, ⌬17, ⌬20N, and ⌬20C mutants retained activity similar to that of wild type protein.
2) The assembly activity of ⌬17d was half that of wild type protein. In this study, we examined the same linker mutants in the bead assay. We obtained similar results as shown in Fig. 3B. Thus, we concluded that the fluorescence intensity of the beads in the bead assay reflects the actin assembly activity.
There is a possibility that actin assembly occurred in the cytosol solution through the mechanism other than nucleation by formin, and the GST-formin CT domains on beads just bound the preformed filaments. To exclude this possibility, we used F-actin binding fragments of two actin binding proteins, talin and gelsolin, as negative controls (designated as talin CT and gelsolin CT, respectively). GST-talin CT-and GST-gelsolin CT-coated beads showed little fluorescence similar to GSTcoated beads (Fig. 3C), suggesting that formin CT domains on beads does not bind preformed F-actin but nucleates de novo actin assembly.

Daam1 and mDia1 Are Autoinhibited by the DID-DAD
Interaction-As for mDia1 and FRL␣, the actin assembly activity of the FH1 and FH2 domains is reportedly autoinhibited by the DID domain. The regulation of Daam1, however, has not been assessed biochemically. Here we tried to confirm the autoinhibition model of mDia1, and we examined the regulation of Daam1 using our bead assay.
First, we examined the binding ability of purified His 6 -tagged Daam1 NT and mDia1 NT with Daam1 CT and mDia1 CT in vitro. As shown in Fig. 4A, Daam1 NT interacted with Daam1 CT but not with mDia1 CT. Similarly, the mDia1 NT interacted with mDia1 CT but not with Daam1 CT (Fig. 4B). The mDia1 NT A256D, with a single-amino acid mutation in its DID domain, did not interact with DID-DAD, as shown previously (20). The corresponding mutant Daam1 NT A229D also failed to interact with Daam1 CT (Fig. 4A).
We next analyzed the effects of these N-terminal peptides on the actin assembly activity. As shown in Fig. 4D, mDia1 CT-mediated actin assembly was inhibited in a concentration-dependent manner by the addition of mDia1 NT but not Daam1 NT. Similarly, Daam1 CT-mediated filament assembly was efficiently inhibited by the addition of Daam1 NT but not by mDia1 NT (Fig. 4C). The mDia1 NT A256D and the Daam1 NT A229D had no effect on actin assembly activity (Fig. 4, C and D). These data indicated that the NT fragments inhibited the actin  assembly activity of Daam1 CT and mDia1 CT through binding to the CT fragments via the DID domains.
It has been shown that Phe-1195 in the DAD domain of mouse mDia1 is critical for interaction with the DID domain (42). We generated human mDia1 CT harboring an alanine substitution at Phe-1203 that corresponds to mouse mDia1 Phe-1195. This mDia1 CT mutant (mDia1 CT F1203A) assembled actin filaments with efficiency similar to that of the wild type protein (Fig. 5B). However, it did not interact with mDia1 NT (Fig. 5B), and the actin assembly was not inhibited by mDia1 NT (Fig. 5D). The corresponding mutant Daam1 CT F1046A also did not bind to Daam1 NT (Fig. 5A), and Daam1 NT did not inhibit the actin assembly generated by Daam1 CT F1046A (Fig. 5C). Taken together, these results indicate that the actin assembly activities of Daam1 and mDia1 were inhibited by their intramolecular DID-DAD interactions.
GTP-bound Rho Relieved the Autoinhibition of Daam1 and mDia1-We systematically examined which Rho family GTPases interacted with the GBD domains of Daam1 and mDia1. We prepared GST-tagged recombinant human RhoA, RhoB, RhoC, RhoD, RhoF (also designated as Rif), RhoG, RhoJ, Rac1, Rac2, Rac3, Cdc42, Cdc42like, Rnd1, and TC10 in both GDPand GTP␥S-bound forms. We then examined the interaction of these GTPases with His 6 -tagged Daam1 NT and mDia1 NT. As shown in Fig.  6A, mDia1 NT interacted with GTP␥S-bound but not GDP-bound RhoA, RhoB, or RhoC as shown previously (33). It also interacted weakly with RhoD and RhoF. Daam1 NT also interacted with RhoA, RhoB, and RhoC as well as weakly with RhoD when it was in its GTP␥S-bound form. Even when they were in their GTP␥S-bound form, Rac1, Cdc42, and other Rho subfamily GTPases did not bind to the GBD domain of either mDia1 or Daam1 (Fig. 6A). It was confirmed that GST-Rac1 and -Cdc42 used in this study were functional because beads coated with them could bind their common effector molecule, p21-activated kinase 1 (PAK1) (43,44), in platelet cytosol in a GTP-dependent manner (data not shown).
GppNHp-RhoA, but not GDP-RhoA, efficiently rescued the inhibition of the Daam1 CT-mediated actin filament assembly by the Daam1 NT in the bead assay (Fig.  6B). Another efficient rescue was achieved when GppNHp-RhoB or -RhoC was added instead of GppNHp-RhoA (Fig. 6B). This was also the case for mDia1 (Fig. 6C). The efficient rescue of the actin assembly activity by GppNHp-RhoA was also confirmed by the observation of each bead for Daam1 (data not shown) and mDia1 (Fig. 6D) by fluorescence microscopy. Neither GppNHp-Rac1 nor GppNHp-Cdc42 relieved the autoinhibition of either Daam1 or mDia1 significantly (Fig. 6, B and C, and data not shown). The addition of GppNHp-RhoA alone to GST-coated beads did not induce any enhancement of fluorescence intensity (data not shown). Thus, Daam1 and mDia1 could be activated by RhoA, RhoB, and RhoC.
Endogenous Formin-mediated Actin Assembly Regulated by Rho in Platelets-So far we showed recombinant Daam1 and mDia1 have efficient actin assembly activities regulated by Rho GTPases. In the last set of experiments we examined whether endogenous formins are activated in isolated platelets. Thrombin is a potent agonist for platelets and induces rapid RhoA activation (38,45), causing shape change and aggregation. As shown in Fig. 7A, anti-Daam1 antibody precipitated little RhoA before thrombin stimulation, whereas RhoA was clearly detected in the immunoprecipitates with anti-Daam1 antibody after 20 s of stimulation. Thus, thrombin induced the associa- tion of RhoA with Daam1, suggesting that Daam1 could function under control of RhoA in platelets. Activated RhoA would recruit Daam1 and mDia1 in platelets to assemble actin filaments. Next, we isolated endogenous formins from platelet lysate using the GTP␥S-bound GST-RhoA-coated beads. Actin filaments were observed on the beads when the beads were incubated with platelet cytosol (Fig. 7B). The actin assembly was mediated by Daam1 and mDia1 because the addition of either DID domain efficiently inhibited the actin assembly (Fig.  7B). Thus, Daam1 and mDia1 would function as actin assembly factors under the control of RhoA in the activated platelets.

DISCUSSION
We have identified two DRFs, mDia1 and Daam1, that are GTP-RhoA-binding proteins found in platelets. We characterized these formins using a novel bead assay. We found that the mDia1 CT containing its FH1, FH2, and DAD domains and the Daam1 CT were able to assemble actin filaments in the platelet cytosol. We then showed that the activities of mDia1 and Daam1 were regulated by autoinhibitory intramolecular interactions between the DID domain and the DAD domain. These interactions are unlocked by active Rho GTPases.
In this study we tried to detect the actin assembly on Daam1and mDia1-coated beads in the cytosol using an in vitro assay system. We obtained the following results. 1) We observed rhodamine-phalloidin-stained filamentous structures on the beads (Fig. 2B). 2) F-actin binding fragments of talin and gelsolin did not bind F-actin in the same experimental condition. 3) Monomeric actinsequestering agents, such as cytochalasin D or latrunculin A, abolished the formation of these structures and the increases in fluorescence intensity (data not shown). 4) The bead assay results (Fig. 3B) obtained using the linker deletion mutants were well correlated with the actin assembly activity measured by the pyrene-actin assay (12). 5) ATP was required to form fluorescent filamentous structures on the beads (data not shown). Considering all of these results, we concluded that the fluorescence intensity of the beads reflects the actin assembly activity of formins.
The pyrene assay has traditionally been used to examine the actin assembly activity of formin proteins, and it is suitable for assessing nucleating and elongating activities. However, it is difficult to detect actin assembly activity by the pyrene assay in a crude cytosolic solution. On the other hand, in the bead assay established in this study, we used cytosol as a monomeric actin source. Using this assay, we can observe actin assembly in a crude cytosolic solution, where the actin assembly could be modulated by many cytosolic factors. Thus, the bead assay can reconstitute the actin assembly reaction, which is modified by yet unidentified cellular actin-binding proteins. This method could be useful in trying to better understand actin assembly in the cytoplasmic environment.
The two formins identified here as GTP-RhoA-binding proteins are classified as DRFs, which are thought to be autoinhibited in their resting state. The DID domains are known to inhibit actin assembly activity mediated by the FH2 domain for mDia1 through interaction with the DAD domains (20). We demonstrated here that Daam1 is regulated in the same manner. There is a possibility that this DID-DAD interaction of full-length Daam1 or mDia1 is intermolecular because mDia1 and mDia2 can heterooligomerize and inhibit the Rho-mediated serum-responsive factor activation activity of each other (46). To determine whether the inhibitory interaction of DID and DAD is intermolecular or intramolecular, the structural analysis of full-length formin would be required. It is interesting to note that for Daam1 and mDia1 the DID domain of each formin interacted with its cognate DAD domain and inhibited actin assembly activity of its cognate FH1-FH2-DAD domains in the cytosol (Fig. 4, A-D), suggesting that exogenously expressed DID has a dominant negative activity with respect to its cognate DRF. DRFs are considered to be regulated by Rho GTPases, although the efficient activation of formins has never been demonstrated biochemically. Using the bead assay, we clearly demonstrated that GTP-RhoA, -RhoB, and -RhoC could restore the actin assembly activity of Daam1 and mDia1. The addition of even excess concentration of active RhoA resulted in only a faint recovery of actin assembly activity of the autoinhibited-form DRFs in a pyrene-actin assay (24). Therefore, it has been proposed that some additional factor(s) could be required for full recovery of autoinhibited DRFs (24). Here, we used platelet cytosol, which may contain such an additional factor, as a G-actin source, and that is why we could demonstrate the effect of Rho in reversing the autoinhibition of Daam1 and mDia1. Further investigation is required to unveil the nature of the additional factor(s).
We determined the K D of the binding of RhoA to Daam1 and mDia1 in both nucleotide forms by surface plasmon resonance method. K D of the binding between GTPform RhoA and mDia1 was 4.8 nM, consistent with the previous reports of 6 nM using the stopped-flow fluorescence method (27) and 9 nM using isothermal titration calorimetry method (25). K D of the binding of GTP␥S-RhoA and Daam1 was 4.0 nM, and it is comparable with that between mDia1 and GTP␥S-RhoA. The K D value of the interaction of GDP-RhoA to each formin was 100 -200-fold higher than that of GTP␥S-RhoA, suggesting that both formins function as effectors of RhoA in living cells. We systematically examined which Rho family GTPases interact with the GBD domains of Daam1 and mDia1. Interestingly, the partners of both mDia1 and Daam1 with Rho GTPases were very similar among the Rho GTPases tested (Fig. 6A). Both mDia1 and Daam1 interacted strongly with RhoA, RhoB, and RhoC and weakly with RhoD but not with other GTPases including Rac1 and Cdc42 except for RhoF, which interacts with mDia1 weakly. Phylogenetically, the three proteins RhoA, RhoB, and RhoC are closely related, and they form the Rho subgroup. RhoD and RhoF are the most homologous proteins to the Rho subgroup except for Rnd group (47). These data suggest that DRFs, at least Daam1 and mDia1, are regulated by the RhoA, -B, -C group. Although Cdc42 has been reported to bind Daam1 (48), we could not detect a high affinity interaction with this protein using our system (Fig. 6A).
In platelets, RhoA is rapidly activated upon stimulation (38,45), and botulinum C3 exoenzyme treatment of platelet inhibited the shape change, adhesion, and aggregation, which are mediated by actin cytoskeletal reorganization (37)(38)(39)(40). In this study we demonstrated that Daam1 rapidly forms a complex with RhoA in thrombin-activated platelets (Fig. 7B). Because Daam1 and mDia1 are the actin assembly factors directly regulated by Rho, they could contribute to shape change and aggregation in activated platelets via their actin assembly activity. Because mDia1 is 10-fold more potent in actin assembly activity than Daam1, these two DRFs might have distinct functions. Both Daam1 and mDia1 are ubiquitously expressed and may play important roles in RhoA-dependent reactions in various types of cells.