Small GTPase RhoD suppresses cell migration and cytokinesis

Oncogene (1999) 18, 2431 ± 2440 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc Small GTPase RhoD suppresses cell migration and cytokinesis Keisuke Tsubakimoto1, Ken Matsumoto1, Hiroshi Abe1, Junichiro Ishii1, Mutsuki Amano2, Kozo Kaibuchi2 and Takeshi Endo*,1 1 Department of Biology, Faculty of Science, Chiba University, Yayoicho, Inageku, Chiba, Chiba 263-8522, Japan; 2Division of Signal Transduction, Nara Institute of Science and Technology, Ikoma, Nara 630-0101, Japan Rho family small GTPases regulate organization of the actin cytoskeleton. Among them, RhoA plays essential roles in the formation of the actin stress ®bers, the associated focal adhesions, and the contractile rings necessary for cytokinesis. Recently, RhoD, a novel member of Rho family has been identi®ed. The amino acid sequences of its e€ector domain is distinct from those of the other Rho family proteins, suggesting its unique cellular functions. Introduction of the constitutively active form of RhoDG26V into ®broblasts by microinjection or transfection resulted in disassembly of the actin stress ®bers and the focal adhesions, whereas the dominant negative form of RhoDT31K did not a€ect these structures. The degree of cell migration assessed by the phagokinetic tracks on a substrate covered with gold particles was diminished by the expression of RhoDG26V but not by RhoDT31K. Thus, cytoskeletal alterations including the loss of stress ®bers and focal adhesions by RhoD seems to lead to the retardation of cell migration. Transfection of RhoDG26V cDNA into cultured cells also induced multinucleation. Moreover, RhoDG26V microinjected into fertilized eggs and embryos of Xenopus laevis caused cleavage arrest only in the injected cells, and the uncleaved cells contained multiple nuclei. These results imply that RhoD does not a€ect nuclear division but can interfere with cytokinesis presumably by preventing the formation of the actin-based contractile ring. Enhancement of the stress ®bers by RhoA or RhoA-activating lysophosphatidic acid was reversed by the transfection of RhoD cDNA. Accordingly, the cellular functions of RhoD are likely to be antagonistic to those of RhoA. Keywords: small GTPase; Rho family; actin cytoskeleton; cell motility; cytokinesis Introduction The actin cytoskeleton mediates various cellular functions including cell motility, cell adhesion, and cell division. Members of the mammalian Rho family small GTPases have emerged as key regulators that link cell surface receptors to the organization of the actin cytoskeleton. The family comprises RhoA, RhoB, RhoC, Rac1, Rac2, Cdc42, Tc10, RhoG, RhoE, and recently identi®ed RhoD (Narumiya, 1996; Murphy et *Correspondence: T Endo Received 23 June 1998; revised 25 November 1998; accepted 11 December 1998 al., 1996; Van Aelst and D'Souza-Schorey, 1997). The functions of RhoA, Rac1, and Cdc42 and their target e€ector proteins have been intensively examined. In ®broblasts, activation of RhoA by the extracellular ligand lysophosphatidic acid (LPA) leads to the assembly of contractile actin stress ®bers and associated focal adhesions (Ridley and Hall, 1992). Rac1 activated by platelet-derived growth factor or insulin induces the assembly of an actin ®lament meshwork to generate membrane ru‚es (lamellipodia) and speci®c focal complexes (Ridley et al., 1992; Nishiyama et al., 1994; Nobes and Hall, 1995). On the other hand, Cdc42 activated by bradykinin is responsible for the formation of actin ®lamentcontaining microspikes (®lopodia) and associated focal complexes (Kozma et al., 1995; Nobes and Hall, 1995). In addition, Cdc42 can activate Rac1, hence extension of ®lopodia is accompanied by concerted lamellipodial spreading (Kozma et al., 1995; Nobes and Hall, 1995). Sequentially Rac1 can activate RhoA to form stress ®bers, although this is a weak and delayed response in ®broblasts (Ridley et al., 1992; Hall, 1998). Microinjection of Clostridium botulinum C3 ADPribosyltransferase (C3 exoenzyme), which inactivates RhoA by ADP-ribosylating its e€ector domain, or Rho guanine nucleotide dissociation inhibitor (Rho GDI) into Swiss 3T3 ®broblasts causes inhibition of cell migration (Takaishi et al., 1993). Hepatocyte growth factor (HGF) or 12-O-tetradecanoylphorbol-13-acetate (TPA) induces cell migration of a mouse keratinocyte cell line. The migration is also suppressed by microinjection of C3 exoenzyme or Rho GDI (Takaishi et al., 1994). These ®ndings suggest that RhoA mediates migration of these cells. When C3 exoenzyme or Rho GDI is injected into Xenopus embryos (Kishi et al., 1993; Drechsel et al., 1997) or sand dollar eggs (Mabuchi et al., 1993), actin/ myosin-based contractile ring formation is inhibited. In addition, if C3 exoenzyme is injected during cleavage, it causes regression of the cleavage furrow (Mabuchi et al., 1993). Inhibition of RhoA by these proteins results in the interference with cytoplasmic cleavage but not with nuclear division and consequently multinucleation is induced (Kishi et al., 1993; Mabuchi et al., 1993). Accordingly, RhoA is required for progression of cleavage through the formation of contractile ring. RhoD has been cloned by the strategy based on the polymerase chain reaction (Murphy et al., 1996; Matsumoto et al., 1997). The amino acid sequence of its e€ector domain is distinct from those of the other members of the Rho family, suggesting its unique functions among the family. When wild-type or constitutively activated RhoD is transiently expressed in various types of cultured cells by the T7 vaccinia RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2432 virus system, it induces rearrangements of actin cytoskeleton (Murphy et al., 1996). The actin stress ®bers disappear, and instead actin-containing long thin processes are formed at the cell periphery. In addition, any focal complexes containing vinculin or paxillin are disassembled in the cells. RhoD also interferes with the formation of large endosomes by the homotypic fusion between the early endosomes in response to the constitutively active form of Rab5 (Murphy et al., 1996). This is likely to be primarily due to an inhibition of intracellular motility of the early endosomes through disrupting cytoskeletal ®laments. In the present study, we examined the e€ects of RhoD on the cell migration and mitosis. The degree of cell migration was diminished by the expression of activated RhoD. Cytoskeletal changes including the loss of stress ®bers and focal adhesions by RhoD may lead to the inhibition of cell migration. Transfection of the activated RhoD cDNA into cultured cells also induced multinucleation. Microinjection of the activated RhoD protein into Xenopus eggs and embryos gave rise to cleavage arrest and generated the uncleaved cells containing multiple nuclei. Thus, RhoD does not a€ect mitosis but can impede cytokinesis possibly by preventing the formation of the contractile ring. Furthermore, enhancement of the stress ®bers by RhoA or LPA was reversed by the transfection of RhoD cDNA. These results suggest that RhoD antagonizes the e€ects of RhoA on the actin cytoskeleton. Results Structural properties of RhoD We have cloned a cDNA encoding RhoD by screening a cDNA library of the mouse skeletal muscle C2 myotubes with a product of the 3' rapid ampli®cation of cDNA ends (3' RACE) (Matsumoto et al., 1997) based on the polymerase chain reaction. This cDNA includes the entire coding region as well as the 5' and 3' untranslated regions (DDBJ/EMBL/GenBank Data Bank accession number D89821). The predicted amino acid sequence of this cDNA is identical to that reported by Murphy et al. (1996). Comparison of the entire amino acid sequence with those of the other members of the Rho family indicates that RhoD is distinct from the other members over the whole sequence (data not shown). This is corroborated by the evolutional relationship that RhoD diverged from the other Rho family proteins at an early stage of evolution (Matsumoto et al., 1997). RhoD is distinct from the other Rho family members as many as seven amino acids with respect to the nature of their side chains within the core e€ector domain and the putative extended e€ector domain (Self et al., 1993) (Figure 1). This sequence divergence of RhoD suggests that it interacts with speci®c target proteins and that consequently it exerts unique cellular functions. Destruction of stress ®bers and focal adhesions by RhoD To examine the e€ects of RhoD in ®broblasts, the bacterially expressed recombinant proteins were microinjected into serum-starved subcon¯uent Swiss 3T3 cells. These quiescent cells still contained many actin stress ®bers although the stress ®bers were thinner than those in proliferating cells (Kozma et al., 1995). Fluorescein-conjugated dextran was coinjected with the proteins to identify injected cells. When the cells were microinjected with the constitutively active form of RhoDG26V, most stress ®bers disappeared within 30 min, as detected by rhodamine-phalloidin staining (Figure 2a). This result is consistent with the previous report (Murphy et al., 1996), although thin processes containing actin ®laments were not de®nitely formed in the injected cells compared with uninjected cells under these conditions. Injection of the dominant negative RhoDT31K did not cause appreciable cytoskeletal changes (Figure 2b). cDNAs encoding the constitutively active and the dominant negative forms of RhoD, which were fused with Myc-epitope tag at their N-termini, were expressed by transient transfection in proliferating C3H/10T1/2 ®broblasts. We used this cell line for transfection instead of Swiss 3T3 cells owing to much higher transfection eciency. The cells were doubly stained with the monoclonal antibody (mAb) Myc19E10 recognizing the Myc-tag and rhodamine-phalloidin. The RhoDG26V-expressing cells often exhibited disappearance of the stress ®bers and instead di€used phalloidin staining throughout the cytoplasm within 18 h after the transfection (Figure 2c). This di€used phalloidin staining appears to represent disassembled or further fragmented actin ®laments. Double staining with Myc1-9E10 and anti-vinculin antibody showed that the focal adhesions also became scarcely detectable in most of the RhoDG26V-expressing cells (Figure 2e). Thin processes were not always evident in the expressing cells under these conditions either. Whole appearance, the actin cytoskeleton, and the focal adhesions of the RhoDT31K-expressing cells were indistinguishable from those of the untransfected cells (Figure 2d and f). Figure 1 Comparison of the amino acid sequence of RhoD with those of known Rho family members around the core and the extended e€ector domains. The origins of known Rho family proteins are human RhoA (Yeramian et al., 1987), mouse RhoB (Nakamura et al., 1996), mouse RhoC (Segade et al., 1995), human RhoE (Foster et al., 1996), human RhoG (Vincent et al., 1992), mouse Rac1 (Moll et al., 1991), mouse Rac2 (Shirsat et al., 1990), mouse Cdc42 (Marks and Kwiatkowski, 1996), and human Tc10 (Drivas et al., 1990). Amino acids at positions of 450% identity are shown in white on black. Double overlines and dashed overlines denote the core and the putative extended e€ector domains, respectively. Amino acids in RhoD distinct from those in the other Rho family proteins in the core and extended e€ector domains are pointed out by daggers RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2433 Figure 2 Destruction of the stress ®bers and the focal adhesions by the activated RhoD introduced by microinjection and transfection. (a and b) Serum-starved subcon¯uent Swiss 3T3 cells microinjected with the recombinant constitutively active RhoDG26V (a) or dominant negative RhoDT31K (b) protein, 30 min after the injection. (c ± f) Transfected 10T1/2 cells expressing Myc-tagged RhoDG26V (c and e) or RhoDT31K (d and f), 24 h after the transfection in the growth medium. Cells were stained with rhodamine ± phalloidin to detect actin ®laments (a ± d) or with anti-vinculin antibody to detect focal adhesions (e and f). Arrows indicate the RhoDG26V-injected cells (a) or the Myc ± RhoDG26V-expressing cells (c and e), in which the stress ®bers and focal adhesions disappeared. Arrowheads point to the RhoDT31K-injected cells (b) or the Myc ± RhoDT31K-expressing cells (d and f), in which the stress ®bers and focal adhesions remained. Injected cells were detected by coinjection of ¯uorescein ± dextran. Transfected cells were detected by immuno¯uorescent staining with the mAb Myc1-9E10. Scale bars, 20 mm Suppression of cell migration by RhoD Since it has been suggested that actin-associated structures are involved in cell migration (Lauffenburger and Horwitz, 1996; Mitchison and Cramer, 1996), destruction of the stress ®bers and focal adhesions by RhoD prompted us to examine the e€ect of the protein on cell migration. The degree of cell migration was assayed by phagokinetic tracks of 10T1/2 cells on gold particle-coated glass coverslips (Albrecht-Buehler, 1977). The cells were transfected with the Myc-tagged cDNA of RhoDG26V or RhoDT31K and replated on the particle-coated coverslips. The tracks free of the particles generated by phagocytotic ingestion or attachment of the particles on the cell surface correspond to migration trails. The degree of migration was markedly retarded in most of the RhoDG26Vexpressing cells compared with untransfected cells (Figure 3a ± c). On the other hand, RhoDT31K-expressing cells migrated to the extent comparable to untransfected cells (Figure 3d and e). In fact, the area of the tracks that the RhoDG26V-expressing cells migrated (4.86103 mm2) was about one third and one fourth of what the untransfected and RhoDT31Kand expressing cells formed (13.36103 mm2 20.06103 mm2, respectively) (Table 1). These results evidently indicate that migration ability of the ®broblasts is suppressed by the forced expression of the activated RhoD. Induction of multinucleation in cultured cells by RhoD When Myc ± RhoDG26V was expressed by transfection in Balb/3T3 and 10T1/2 ®broblasts and N1E-115 neuroblastoma cells, the cells with two to four nuclei became evident by 24 h after the transfection (Figure 4). The cytoplasm of such multinucleated cells tended to be larger than that of mononucleated cells. In Balb/3T3 cells, such multinucleated cells were *10% of the RhoDG26V-expressing cells at this time (Table 2). On the other hand, essentially all of the untransfected cells and the cells transfected with the vacant Myc-tagging vector pCMVmyc contained single nuclei (Table 2). The ratio of multinucleated cells elevated up to *20% RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2434 Figure 3 Phagokinetic tracks of 10T1/2 cells expressing RhoDG26V and RhoDT31K. (a ± c) Phagokinetic tracks of Myc ± RhoDG26Vexpressing cells. The tracks were examined by dark ®eld microscopy (a). Arrows indicate the expressing cells, which were detected by immuno¯uorescent staining with Myc1-9E10 (b and c). (d and e) Phagokinetic tracks of Myc ± RhoDT31K-expressing cells examined by dark ®eld microscopy (d). An arrowhead points to the expressing cell detected by the immuno¯uorescent staining with Myc19E10 (e). The transfected cells were replated on gold particle-coated coverslips 6 h after the transfection and ®xed 18 h after the replating. Scale bars, 100 mm (a and d) and 20 mm (b, c and e) Table 1 Retardation of cell migration by the activated RhoD estimated by phagokinetic tracks Transfected cDNA None RhoDG26V RhoDT31K 3 2 Area of phagokinetic tracks (610 mm ) 13.3+5.3 4.8+3.9 20.0+8.3 The phagokinetic tracks of 10T1/2 cells transfected with Myc-tagged RhoDG26V or RhoDT31K cDNA were recorded as micrographs and traced on graph paper. The area of the tracks (n=12 for each transfection experiment) was calculated and expressed as mean+s.d. by 48 h, but most of them contained two nuclei and few cells contained more than four nuclei (data not shown). Although a few per cent of 10T1/2 and N1E115 cells already contained two nuclei prior to transfection, the percentage of multinucleated cells elevated after the transfection (Table 2). Forced expression of a constitutively active mutant of Cdc42G12V in human HeLa-derived cells has been shown to induce multinucleation and enlargement of the cytoplasm (Dutartre et al., 1996). We also transfected Myc ± Cdc42G12V cDNA to Balb/3T3, 10T1/2, and N1E-115 cells. The cells became multinucleated similarly to those transfected with Myc ± RhoDG26V cDNA. The ratio of the multinucleated cells was also *10% in the Balb/3T3 cells expressing Cdc42G12V at 24 h after the transfection (data not shown). This ratio is almost comparable to that reported by Dutartre et al. (1996) using the stably transfected cells. These results suggest that the forced expression of the activated RhoD induces multinucleation in the cultured cells by preventing cytokinesis as has been postulated for Cdc42. Interference with cleavage and induction of multinucleation in Xenopus embryos by RhoD To con®rm the postulation that the activated RhoD induces multinucleation by preventing cytokinesis, we microinjected the recombinant RhoDG26V or RhoDT31K protein into one blastomere of Xenopus laevis embryo at the two-cell stage. A cell injected only with the injection bu€er cleaved normally (Figure 5Aa). The RhoDG26V-injected cell, however, usually cleaved only once and ceased cleavage after that, whereas the opposite uninjected blastomere cleaved normally (Figure 5Ab). This injection consequently gave rise to hemiblastulation. Occurrence of one cleavage after the injection seems to be ascribable to a short period between the ®rst and the second cleavage (*30 min) and a time lag between the injection and activation of the protein by the C-terminal modi®cations including isoprenylation (Glomset and Farnsworth, 1994) or accession of the modi®ed RhoD to its target proteins. Microinjection of c-Mos RNA (Sagata et al., 1989), the anti-XAC antibody, or excess amounts of the constitutively active XAC protein (Abe et al., 1996) to one blastomere of the two-cell stage embryo similarly cleaved only once or a few times and generates hemiblastula. RhoDT31K-injected cell cleaved normally and was indistinguishable from the uninjected embryo (Figure 5Ac). To examine whether nuclear division took place or not in the RhoDG26V-injected blastomere, paranembedded sections of the embryo was stained with bisbenzimide H33258. Two or three nuclei were detected in each section of the uncleaved blastomere (Figure 5Ba). Thus, the RhoDG26V-injected cell became multinucleated. Mabuchi et al. (1993) have shown that RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2435 Figure 4 Induction of multinucleation in cultured cells by the activated RhoD. (a ± c) A binucleated Balb/3T3 cell generated by the transfection of Myc-tagged RhoDG26V cDNA. (d ± f) A binucleated 10T1/2 cell generated by the expression of Myc ± RhoDG26V. (g ± h) A tetranucleated N1E-115 cell generated by the expression of Myc ± RhoDG26V. Phase-contrast micrographs (a, d and g), corresponding ¯uorescent micrographs detecting RhoDG26V-expressing cells by staining with Myc1-9E10 (b, e and h), and those detecting nuclei by staining with bisbenzimide H33258 (c, f and i). The cells were ®xed 24 h after the transfection. Scale bar, 20 mm Table 2 Ratio of multinucleation ®broblasts Cell line Transfected cDNA Balb/3T3 None (pCMVmyc) RhoDG26V None (pCMVmyc) RhoDG26V C3H/10T1/2 in RhoDG26V-expressing Number of nuclei in a cell (%) One Two Three 100 90.9 94.4 86.6 0 8.5 5.6 12.3 0 0.6 0 1.1 Balb/3T3 and 10T1/2 ®broblasts were transfected with the vacant pCMVmyc vector (none) or the vector harboring RhoDG26V cDNA. The cells were ®xed 24 h after the transfection and doubly stained with Myc1-9E10 and H33258. More than 200 transfected cells in each transfection were examined to count the number of the nuclei microinjection of C3 exoenzyme into sand dollar eggs induce multinucleation by preventing cytokinesis but not nuclear division. To compare the e€ects of RhoDG26V and C3 exoenzyme, each of these proteins was injected into Xenopus fertilized eggs before cleavage. The eggs injected with each protein cleaved only once and arrested cleavage after that. Because Xenopus embryos are opaque and thus the nuclei in intact whole embryos cannot be observed by microscopy, they were squashed after the staining with H33258 to count the total number of the nuclei. Both the RhoDG26V-injected embryo and the C3 exoenzymeinjected embryo contained about 32 nuclei at the time corresponding to stage 6 of control embryo (Figure 5Bb and c). Taken together, these results imply that the Figure 5 Interference with cleavage and induction of multinucleation in Xenopus embryos by the activated RhoD. (A) Hindrance to cleavage by microinjected RhoDG26V. One blastomere of the two cell stage embryo was microinjected with the injection bu€er as a control (a) and with the recombinant RhoDG26V (b) or RhoDT31K (c) protein. The embryos were ®xed at stage 6 ± 7 and viewed from the animal poles. (B) Induction of multinucleation by microinjected RhoDG26V. (a) The embryo was injected with RhoDG26V and ®xed as above. A paran-embedded section was stained with H33258 to detect the nuclei. (b and c). The fertilized eggs were injected with RhoDG26V (b) or C3 exoenzyme (c) and ®xed at the time corresponding to stage 6 of control embryo. The squashed embryos were stained with H33258. Only one microscopic ®eld, which usually contains 1 ± 6 nuclei, is shown for each embryo. Scale bar, 100 mm RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2436 activated RhoD as well as C3 exoenzyme interferes with cytokinesis of the blastomeres without a€ecting nuclear division and that consequently the protein induces multinucleation in the blastomeres and in the cultured cells as well. Antagonistic e€ects of RhoD to RhoA action The above e€ects of RhoD appeared to be antagonistic to those of RhoA. To assess this possibility, constitutively activated RhoAG14V, RhoDG26V, or both the cDNAs were transfected to 10T1/2 cells. The stress ®bers were enhanced by the expression of hemagglutinin (HA)-tagged RhoAG14V in as much as 93% of the expressing cells (Figure 6a and b, Table 3). When HA ± RhoAG14V and Myc ± RhoDG26V) cDNAs were cotransfected, the ratio of cells containing the enhanced stress ®bers decreased to 80% (Figure 6c ± e, Table 3). Thus, RhoD interferes at least in some of the cells with the stress ®ber formation induced by RhoA. The interference with the stress ®ber formation only in a subset of the cotransfected cells might be due to the dual expression of the constitutively active RhoAG14V and RhoDG26V. Therefore, we next transfected only Myc ± RhoDG26V cDNA to 10T1/2 cells and then stimulated the cells with LPA to activate endogenous RhoA. More than 70% of the RhoDG26Vexpressing cells contained diminished stress ®bers, whereas the unexpressing cells contained the stress ®bers enhanced by the LPA treatment (Figure 6f and g). These results indicate that RhoD has an antagonistic e€ect to RhoA at least in the stress ®ber formation. Table 3 Interference of RhoD with RhoA-induced stress ®ber formation Transfected cDNA RhoAG14V RhoAG14V+RhoDG26V RhoDG26V Enhancement of stress ®bers (%) Cell numbers counted (+) (7) 93.1 80.9 0 6.9 19.1 100 116 139 123 10T1/2 cells were either singly transfected with HA ± RhoAG14V or Myc ± RhoDG26V cDNA or cotransfected with both the cDNAs. HA ± RhoAG14V- and Myc ± RhoDG26V-expressing cells were detected by anti-HA-tag antibody and Myc1-9E10 staining, respectively. Stress ®bers were detected by the staining with coumarin ± phallacidin. The expressing cells with (+) or without (7) enhancement of the stress ®bers were counted Figure 6 Antagonistic e€ects of RhoD to RhoA action inducing the stress ®ber formation in 10T1/2 cells. (a and b) Cells transfected with HA ± RhoAG14V cDNA, 24 h after the transfection. Stress ®bers were enhanced in the RhoAG14V-expressing cell (b, arrow). (c ± e) Cells cotransfected with HA ± RhoAG14V and Myc ± RhoDG26V cDMAs, 24 h after the transfection. RhoAG14Vinduced enhancement of the stress ®bers was suppressed by the coexpression of RhoDG26V (e, arrow). (f and g) Cells transfected with Myc ± RhoDG26V cDNA and stimulated with LPA. Twenty-four hours after the transfection, the cells were treated with 200 ng/ ml LPA for 10 min. RhoDG26V-expressing cells (arrows) contained diminished stress ®bers, whereas unexpressing cells contained enhanced stress ®bers (g). HA ± RhoAG14V-expressing cells were detected by anti-HA-tag antibody staining (a and c). Myc ± RhoDG26V-expressing cells were detected by Mcy1-9E10 staining (d and f). Actin ®laments were detected by the staining with coumarin ± phallacidin (blue ¯uorescence) (b and e) or with rhodamine ± phalloidin (g). Scale bar, 20 mm RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2437 The above results urged us to determine how RhoD interferes with the e€ect of RhoA. One possible mechanism is that RhoD binds to e€ectors of RhoA to sequester them away from RhoA. The Rhoassociated kinase isozymes, Rho-kinase/ROCK-II/ ROKa and ROCK-I/ROKb, are the e€ector proteins of RhoA mediating the assembly of the stress ®bers and focal adhesions (Leung et al., 1996; Amano et al., 1997; Ishizaki et al., 1997). To examine whether RhoD a€ects Rho-kinase, interaction of RhoDG26V and RhoAG14V with each domain of Rho-kinase (Amano et al., 1997) was analysed by ligand overlay assay. RhoAG14V strongly bound to the Rho-binding domain, whereas RhoDG26V did not bind to any domains of Rho-kinase in this assay (Figure 7). Consequently, RhoD is likely to antagonize RhoA action inducing the stress ®bers through a way other than direct prevention of Rho-kinase. Discussion The amino acid sequence within the core e€ector domain and the putative extended e€ector domain of RhoD is distinct from those of the other Rho family members. As has been expected from its sequence divergence, RhoD exerted unique cellular e€ects di€erent from those of the known Rho family proteins. Microinjection of the constitutively active RhoDG26V protein into ®broblasts and transfection of its cDNA to ®broblasts resulted in disassembly of the actin stress ®bers and associated focal adhesions as has been reported by Murphy et al. (1996). RhoA mediates the formation of the stress ®bers and focal adhesions in cultured cells stimulated by extracellular signaling molecules such as LPA (Ridley and Hall, 1992). Introduction of the activated RhoAG14V protein in cultured cells by microinjection or transfection brings about the same cytoskeletal changes (Ridley and Hall, 1992; Amano et al., 1997; Uehara et al., 1997; Sahai et Figure 7 Ligand overlay assay for interaction between RhoAG14V or RhoDG26V and Rho-kinase. (a) SDS ± PAGE of recombinant proteins of each domain of Rho-kinase. CAT, catalytic domain; COIL, coiled-coil domain; RB, Rho-binding domain; RB (TT), Rho-binding domain with loss-of-function mutation; PH, PH domain. (b) Ligand overlay assay for interaction between RhoAG14V and each domain of Rho-kinase. RhoAG14V strongly bound to RB domain. (c) Ligand overlay assay for interaction between RhoDG26V and each domain of Rho-kinase. No interaction was detected al., 1998). Cotransfection of RhoAG14V and RhoDG26V cDNAs brought about interference with enhancement of the stress ®bers induced by RhoAG14V in a subset of the coexpressing cells. Moreover, transfection of RhoDG26V cDNA followed by LPA treatment not only prevented LPA-induced enhancement of the stress ®bers but also diminished preexisting stress ®bers. These results imply that RhoD has an antagonistic e€ect to RhoA at least in the stress ®ber formation. Transfection of RhoDG26V cDNA to ®broblasts resulted in retardation of cell migration as detected by phagokinetic track analysis. Since it has been postulated that actin-associated structures are required for cell migration (Lau€enburger and Horwitz, 1996; Mitchison and Cramer, 1996), the alterations in cytoskeletons and cell adhesions including the stress ®bers and focal adhesions by the activated RhoD may lead to the interference with cell migration. Inactivation of RhoA and related proteins by C3 exoenzyme or Rho GDI causes inhibition of cell migration (Takaishi et al., 1993,1994). These ®ndings corroborate the above notion. Cells transformed with the activated RhoA or Rac1 as well as with the oncogenic guanine nucleotide exchange factors (GEFs) (Dbl, Ost, Tiam-1, and Vav) for Rho family members exhibit invasive and metastatic potential either in vivo or in vitro (Habets et al., 1994; Michiels et al., 1995; del Peso et al., 1997). RhoA and Rac1 are therefore likely to play important roles in aquisition of invasive and metastatic properties of tumor cells. In this context, RhoD might be applicable to prevent invasive and metastatic potential of the transformed cells. Multinucleation was induced in several cell types including ®broblasts and neuroblastoma cells transfected with RhoDG26V cDNA. Induced expression of the constitutively active Cdc42G12V in HeLa cells also results in multinucleated cells (Dutartre et al., 1996). The ratio of multinucleated cells induced by the transfection of the RhoD cDNA was almost comparable to that in the Cdc42-expressing cells. Multinucleated cells are also generated by forced expression of the oncogenic forms of Vav, Dbl, and Tiam-1 (Katzav et al., 1989; Ron et al., 1991; Habets et al., 1994), which exhibit GEF activity for Cdc42 (Hart et al., 1991; Michiels et al., 1995; Han et al., 1997). Because Cdc42 is implicated in the reorganization of actin cytoskeleton (Kozma et al., 1995; Nobes and Hall, 1995; Dutartre et al., 1996), formation of the contractile rings or some cytoskeletal structures required for cytokinesis may be abrogated by deregulated expression of Cdc42. Similarly, overexpression of the activated RhoD is likely to prevent cytokinesis by suppressing the formation of these cytoskeletal structures. The prevention of cytokinesis without a€ecting nuclear division seems to result in the formation of multinucleated cells. Recently, citron kinase, an e€ector protein of RhoA, has been shown to be involved in cytokinesis (Madaule et al., 1998) Transfection of cDNAs encoding the truncation mutants of citron kinase results in a failure in cytokinesis to generate multinucleated cells. This ®nding also supports the possibility that deregulated expression of RhoDG26V gives rise to multinucleated cells through interfering with RhoA action. Although the RhoDG26V-expressing cells as well as the Cdc42G12Vexpressing cells contained two or three nuclei, they RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2438 rarely contained more than four nuclei even after long cultivation. This might be due to intrinsic restrictions of cells to accommodate too many nuclei such as limited cytoplasmic volume without cytokinesis. Microinjection of RhoDG26V protein into the Xenopus eggs and embryos demonstrated the above possibility that the multinucleation in RhoDG26V-expressing cells is induced by the prevention of cytokinesis without a€ecting nuclear division. When C3 exoenzyme or Rho GDI is injected into Xenopus embryos (Kishi et al., 1993; Drechsel et al., 1997) or sand dollar eggs (Mabuchi et al., 1993), formation of the contractile rings is impaired. Inhibition of RhoA by these proteins results in the interference with cytoplasmic cleavage but not with nuclear division and consequently multinucleation is induced (Kishi et al., 1993; Mabuchi et al., 1993). These ®ndings indicate that the e€ects of the activated RhoD on cytokinesis are similar to those of the inhibitors of RhoA activity and further suggest that RhoD antagonizes RhoA also in cytokinesis. Exerting the e€ects of RhoD in Xenopus embryos imply that its target proteins and possibly a protein homologous to mammalian RhoD are present in Xenopus embryos. According to the previous report (Murphy et al., 1996), thin long processes containing actin ®laments were commonly (70%) formed in cells exogenously expressing wild-type RhoD or RhoDG26V. In the present study, however, such processes were not de®nitely detected in the microinjected cells or in the transfected cells compared with uninjected or untransfected cells. One possible explanation for the discrepancy between their results and ours might stem from the di€erences in the expression systems or in the cells used. Indeed, the activated RhoAG14V did not stimulate the stress ®ber formation in their T7 vaccinia virus system (Murphy et al., 1996), whereas we constantly observed that RhoAG14V stimulates the formation of stress ®bers and focal adhesions in both microinjection and transfection experiments (see Figure 6b) as reported by many investigators (Ridley and Hall, 1992; Amano et al., 1997; Uehara et al., 1997; Sahai et al., 1998). Several e€ector proteins of RhoA have been identi®ed (Narumiya, 1996; Narumiya et al., 1997; Tapon and Hall, 1997; Van Aelst and D'SouzaSchorey, 1997). The Rho-associated kinase isozymes, Rho-kinase/ROCK-II/ROKa and ROCK-I/ROKb, mediate the assembly of the stress ®bers and focal adhesions (Leung et al., 1996; Amano et al., 1997; Ishizaki et al., 1997), presumably by inhibitory phosphorylation of myosin phosphatase subunit and direct phosphorylation of myosin light chain leading to actin ± myosin interaction (Kimura et al., 1996; Amano et al., 1996; Chihara et al., 1997) and by phosphorylation of the Na-H exchanger NHE1 leading to increased binding of integrin to extracellular matrix proteins (Tominaga et al., 1998). Another e€ector protein mDia is involved in actin polymerization by recruiting pro®lin (Watanabe et al., 1997). Citron kinase is likely to play essential roles in RhoA-mediated cytokinesis (Madaule et al., 1998). To address how RhoD interferes with the e€ects of RhoA, we assessed the possibility that RhoD sequester Rho-kinase from RhoA by binding to the kinase. Although RhoD did not appreciably interact with any domains of Rhokinase, the interaction between RhoD and the other e€ector proteins remains to be examined. Alternatively, either direct sequestration of RhoA by RhoD or regulation by RhoD of upstream regulators for RhoA, i.e., inhibition of GEFs or constitutive activation of Rho GDI or GTPase-activating proteins (GAPs), might be plausible for simultaneous interference with multiple e€ects of RhoA. Recently identi®ed Rho family proteins, Rnd1 and Rnd3/RhoE, which may exist as intrinsically active forms, also disassemble the stress ®bers and focal adhesions and prevent the formation of these structures induced by LPA (Nobes et al., 1998). These proteins and RhoD, however, may antagonize RhoA action through di€erent ways because the core e€ector domains of Rnd1 and Rnd3 are identical to that of RhoA and distinct from that of RhoD. In any case, RhoD and Rnd might serve as novel types of regulators for RhoA and other Rho family proteins. Materials and methods Expression and puri®cation of recombinant proteins Point mutations to generate the constitutively active form of RhoDG26V and the dominant negative form of RhoDT31K were introduced in the cDNA with Transformer sitedirected mutagenesis kit (Clontech Laboratories, Inc.). Coding sequences of the wild-type and the mutated RhoD were fused in frame to glutathione S-transferase (GST) in pGEX-2T vector (Pharmacia Biotech). These recombinant fusion proteins were expressed in E. coli strain XL1-Blue and anity-purifed with glutathione-Sepharose (Pharmacia Biotech) as described (Matsumoto et al., 1997). GST moiety was removed from the fusion proteins by digesting with 10 U/ml bovine thrombin (Sigma) (Matsumoto et al., 1997). Microinjection into cultured cells Swiss 3T3 ®broblasts (Todaro and Green, 1963) were grown on glass coverslips in Dulbecco's modi®ed Eagle's (DME) medium containing 10% fetal calf serum (growth medium). Serum-starved subcon¯uent cells were prepared by culturing the cells for 48 h in DME medium containing 0.2% NaHCO3 (Kozma et al., 1995). The recombinant proteins of the constitutively active and the dominant negative RhoD were microinjected into the cells as described previously (Matsumoto et al., 1997). Fluorescein-conjugated dextran (Molecular Probes, Inc.) at a ®nal concentration of 0.5 mg/ml was coinjected with the proteins to allow identi®cation of injected cells. The cells were ®xed and permeabilized and actin ®laments were detected by staining with rhodamine-phalloidin (Molecular Probes, Inc.) as described (Endo et al., 1996). The specimens were observed with a Zeiss Axioskop microscope equipped with phase-contrast and epi¯uorescence optics. Microinjection into Xenopus embryos and sectioning of the embryos About 10 nl of the recombinant RhoDG26V or RhoDT31K protein at 3 mg/ml in the injection bu€er (80 mM NaCl, 4 mM MgCl2, 4 mM HEPES ± NaOH, pH 7.0, and 0.1 mM dithiothreitol) were microinjected into one blastomere of Xenopus laevis embryos at the 2-cell stage. They were microinjected with glass capillary needles by using a micromanipulator (Narishige MO-102R) and a microinjector (Shimadzu CIJ-1). The embryos were ®xed as described (Abe et al., 1996) at appropriate stages after the injection. RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2439 They were observed with a Zeiss 2000-C dissection microscope equipped with overhead illumination. The ®xed embryos in 100% ethanol were embedded in paraffin and 5 ± 6 mm-thick sections were prepared. They were mounted on ovalbumin-coated slide glasses and stained with 1 mg/ml bisbenzimide H33258 (Hoechst 33258). C3 exoenzyme expression plasmid pET-3a/C3 was presented by Dr S Narumiya, and the C3 exoenzyme was prepared according to Morii and Narumiya (1995). The recombinant RhoDG26V or C3 exoenzyme was injected into Xenopus fertilized eggs. At the time corresponding to stage 6 of control embryo, the injected embryos were pricked with a tungsten needle and ®xed with methanol. They were mounted on slide glasses, stained with H33258, and squashed with coverslips. The specimens were observed with the Zeiss Axioskop microscope with a 106PlanNeo¯uar lens as stated above. pCMVmycRhoD and pEF-BOS-HA3/RhoA, they were incubated with Myc1-9E10 plus anti-HA-tag and then with FITC-anti-mouse IgG plus rhodamine-anti-rabbit IgG and coumarin ± phallacidin (emitting blue ¯uorescence through UV ®lters). The specimens were observed with Zeiss Axioskop microscope as stated above. Phagokinetic tracks Gold particles were prepared and glass coverslips were coated with the particles according to Albrecht-Buehler (1977). The mutated RhoD cDNAs in pCMVmyc vector were transfected to 10T1/2 cells. Six hours after the glycerol shock, the cells were replated on the gold particle-coated coverslips. They were ®xed 18 h after the replating and processed for immuno¯uorescent staining with Myc1-9E10. The specimens were observed by dark ®eld microscopy and ¯uorescence microscopy. Epitope tagging and transfection The cDNAs containing the entire coding region of the wild-type and the mutated RhoD were fused in frame to the N-terminal Myc-tag in pCMVmyc vector (Matsumoto et al., 1997). RhoAG14V cDNA was subcloned in pEF-BOSHA3 vector (Amano et al., 1997). Cdc42G12V cDNA subcloned in pEF-BOS-myc vector was presented by Dr Y Takai. The mouse C3H/10T1/2 ®broblasts (Rezniko€ et al., 1973), mouse Balb/3T3 ®broblasts (Aaronson and Todaro, 1968), and mouse N1E-115 neuroblastoma cells (Amano et al., 1972) were grown on glass coverslips in the growth medium. They were transfected with these plasmids by the calcium phosphate-mediated method as described (Endo et al., 1996). The transiently transfected cells were processed for immuno¯uorescence microscopy (Endo and Nadal-Ginard, 1998). The ®xed and permeabilized cells were incubated with the mAb Myc1-9E10 recognizing the Myc-tag (Evan et al., 1985) (American Type Culture Collection) or with anti-HA-tag rabbit IgG (MBL) and then with ¯uorescein isothiocyanate (FITC)-conjugated goat anti-mouse or anti-rabbit IgG (anity-puri®ed, Cappel Laboratories). To detect actin ®laments, rhodamine ± phalloidin or coumarin ± phallacidin (Molecular Probes, Inc.) was included in the secondary antibody. To examine the focal adhesions, anity-puri®ed anti-chicken gizzard vinculin (Endo and Masaki, 1982; Fukami et al., 1994) was included in the primary antibody. To observe the nuclei, cells were further stained with bisbenzimide H33258. When 10T1/2 cells were cotransfected with Ligand overlay assay Coiled-coil (COIL), Rho-binding (RB), loss-of-functionmutated Rho-binding (RB(TT)), and PH domains of Rhokinase were expressed as recombinant GST-fusion proteins in E. coli (Amano et al., 1997). GST ± catalytic (CAT) domain was produced in Sf9 cells by the baculovirus system (Amano et al., 1997). These anity-puri®ed proteins were subjected to SDS ± PAGE and transferred to Immobilon PVDF membrane (Millipore). They were probed with a-32PGTP-labeled GST ± RhoAG14V or GST ± RhoDG26V according to Manser et al. (1995). Washed membrane was exposed to an X-ray film RX (Fuji). Acknowledgements We are grateful to Dr Shuh Narumiya for the generous gift of the C3 exoenzyme expression plasmid, to Dr Yoshimi Takai for the Cdc42 expression plasmid, and to Dr Takashi Obinata for permitting us to use laboratory facilities. This study was supported by research grants to T Endo from the Ministry of Education, Science, Sports, and Culture of Japan and from the Ministry of Health and Welfare of Japan for Nervous and Mental Disorders (8A-1). K Matsumoto is a Research Fellow of the Japan Society for the Promotion of Science. References Aaronson SA and Todaro GJ. (1968). J. Cell. Physiol., 72, 141 ± 148. Abe H, Obinata T, Minamide LS and Bamburg JR. (1996). J. Cell Biol., 132, 871 ± 885. Albrecht-Buehler G. (1977). Cell, 11, 395 ± 404. Amano T, Richelson E and Nirenberg M. (1972). Proc. Natl. Acad. Sci. USA, 69, 258 ± 263. Amano M, Ito M, Kimura K, Fukata Y, Chihara K, Nakano T, Matsuura Y and Kaibuchi K. (1996). J. Biol. Chem., 271, 20246 ± 20249. Amano M, Chihara K, Kimura K, Fukata Y, Nakamura N, Matsuura Y and Kaibuchi K. (1997). Science, 275, 1308 ± 1311. Chihara K, Amano M, Nakamura N, Yano T, Shibata M, Tokui T, Ichikawa H, Ikebe R, Ikebe M and Kaibuchi K. (1997). J. Biol. Chem., 272, 25121 ± 25127. del Peso L, HernaÂndez-Alcoceba R, Embade N, Carnero A, Esteve P, Paje C and Lacal C. (1997). Oncogene, 15, 3047 ± 3057. Drechsel DN, Hyman AA, Hall A and Glotzer M. (1997). Curr. Biol., 7, 12 ± 23. Drivas GT, Shih A, Coutavas E, Rush MG and D'Eustachio P. (1990). Mol. Cell. Biol., 10, 1793 ± 1798. Dutartre H, Davoust J, Gorvel J-P and Chavrier P. (1996). J. Cell Sci., 109, 367 ± 377. Endo T and Masaki T. (1982). Biochem. Biophys. Res. Commun., 107, 1467 ± 1474. Endo T and Nadal-Ginard B. (1998). J. Cell Sci., 111, 1081 ± 1093. Endo T, Matsumoto K, Hama T, Ohtsuka Y, Katsura G and Obinata T. (1996). J. Biol. Chem., 271, 27855 ± 27862. Evan GI, Lewis GK, Ramsay G and Bishop JM. (1985). Mol. Cell. Biol., 5, 3610 ± 3616. Foster R, Hu K-Q, Lu Y, Nolan KM, Thissen J and Settleman J. (1996). Mol. Cell. Biol., 16, 2689 ± 2699. Fukami K, Endo T, Imamura M and Takenawa T. (1994). J. Biol. Chem., 269, 1518 ± 1522. RhoD regulates cell migration and cytokinesis K Tsubakimoto et al 2440 Glomset JA and Farnsworth CC. (1994). Annu. Rev. Cell Biol., 10, 181 ± 205. Habets GGM, Scholtes EHM, Zuydgeest D, van der Kammen RA, Stam JC, Berns A and Collard JG. (1994). Cell, 77, 537 ± 549. Hall A. (1998). Science, 279, 509 ± 514. Han J, Das B, Wei W, van Aelst L, Mosteller RD, KhosraviFar R, Westwick JK, Der CJ and Broek D. (1997). Mol. Cell. Biol., 17, 1346 ± 1353. Hart MJ, Eva A, Evans T, Aaronson SA and Cerione RA. (1991). Nature, 354, 311 ± 314. Ishizaki T, Naito M, Fujisawa K, Maekawa M, Watanabe N, Saito Y and Narumiya S. (1997). FEBS Lett., 404, 118 ± 124. Katzav S, Martin-Zenca D and Barbacid M. (1989). EMBO J., 8, 2283 ± 2290. Kimura K, Ito M, Amano M, Chihara K, Fukata Y, Nakafuku M, Yamamori B, Feng J, Nakano T, Okawa K, Iwamatsu A and Kaibuchi K. (1996). Science, 273, 245 ± 248. Kishi K, Sasaki T, Kuroda S, Itoh T and Takai Y. (1993). J. Cell Biol., 120, 1187 ± 1195. Kozma R, Ahmed S, Best A and Lim L. (1995). Mol. Cell. Biol., 15, 1942 ± 1952. Lau€enburger DA and Horwitz AF. (1996). Cell, 84, 359 ± 369. Leung T, Chen X-Q, Manser E and Lim L. (1996). Mol. Cell. Biol., 16, 5313 ± 5327. Mabuchi I, Hamaguchi Y, Fujimoto H, Morii N, Mishima M and Narumiya S. (1993). Zygote, 1, 325 ± 331. Madaule P, Eda M, Watanabe N, Fujisawa K, Matsuoka T, Bito H, Ishizaki T and Narumiya S. (1998). Nature, 394, 491 ± 494. Manser E, Leung T and Lim L. (1995). Methods Enzymol., 256, 130 ± 139. Marks PW and Kwiatkowski DJ. (1996). Genomics, 38, 13 ± 18. Matsumoto K, Asano T and Endo T. (1997). Oncogene, 15, 2409 ± 2417. Morii N and Narumiya S. (1995). Methods Enzymol., 256, 196 ± 206. Michiels F, Habets GGM, Stam JC, van der Kammen RA and Collard JG. (1995). Nature, 375, 338 ± 340. Mitchison TJ and Cramer LP. (1996). Cell, 84, 371 ± 379. Moll J, Sansig G, Fattori E and van der Putten H. (1991). Oncogene, 6, 863 ± 866. Murphy C, Sa€rich R, Grummt M, Gournier H, Rybin V, Rubino M, Auvinen P, LuÈtcke A, Parton RG and Zerial M. (1996). Nature, 384, 427 ± 432. Nakamura T, Asano M, Shindo-Okada N, Nishimura S and Monden Y. (1996). Biochem. Biophys. Res. Commun., 226, 688 ± 694. Narumiya S. (1996). J. Biochem., 120, 215 ± 228. Narumiya S, Ishizaki T and Watanabe N. (1997). FEBS Lett., 410, 68 ± 72. Nishiyama T, Sasaki T, Takaishi K, Kato M, Yaku H, Araki K, Matsuura Y and Takai Y. (1994). Mol. Cell. Biol., 14, 2447 ± 2456. Nobes CD and Hall A. (1995). Cell, 81, 53 ± 62. Nobes CD, Lauritzen I, Mattei M-G, Paris S, Hall A and Chardin P. (1998). J. Cell Biol., 141, 187 ± 197. Rezniko€ CA, Brankow DW and Heidelberger C. (1973). Cancer Res., 33, 3231 ± 3238. Ridley AJ and Hall A. (1992). Cell, 70, 389 ± 399. Ridley AJ, Paterson HF, Johnson CL, Diekmann D and Hall A. (1992). Cell, 70, 401 ± 410. Ron D, Zannini M, Lewis M, Wickner RB, Hunt LT, Graziani G, Tronick SR, Aaronson SA and Eva A. (1991). New Biol., 3, 372 ± 379. Sagata N, Watanabe N, Vande Woude GF and Ikawa Y. (1989). Nature, 342, 512 ± 518. Sahai E, Alberts AS and Treisman R. (1998). EMBO J., 17, 1350 ± 1361. Segade F, Claudio E, Wrobel K, Ramos S and Lazo PS. (1995). J. Immunol., 154, 2384 ± 2392. Self AJ, Paterson HF and Hall A. (1993). Oncogene, 8, 655 ± 661. Shirsat NV, Pignolo RJ, Kreider BL and Rovera G. (1990). Oncogene, 5, 769 ± 772. Takaishi K, Kikuchi A, Kuroda S, Kotani K, Sasaki T and Takai Y. (1993). Mol. Cell. Biol., 13, 72 ± 79. Takaishi K, Sasaki T, Kato M, Yamochi W, Kuroda S, Nakamura T, Takeichi M and Takai Y. (1994). Oncogene, 9, 273 ± 279. Tapon N and Hall A. (1997). Curr. Opin. Cell Biol., 9, 86 ± 92. Todaro GJ and Green H. (1963). J. Cell Biol., 17, 299 ± 313. Tominaga T, Ishizaki T, Narumiya S and Barber DL. (1998). EMBO J., 17, 4712 ± 4722. Uehara M, Ishizaki T, Satoh H, Ono T, Kawahara T, Morishita T, Tamakawa H, Yamagami K, Inui J, Maekawa M and Narumiya S. (1997). Nature, 389, 990 ± 994. Van Aelst L and D'Souza-Schorey C. (1997). Genes Dev., 11, 2295 ± 2322. Vincent S, Jeanteur P and Fort P. (1992). Mol. Cell. Biol., 12, 3138 ± 3148. Watanabe N, Madaule P, Reid T, Ishizaki T, Watanabe G, Kakizuka A, Saito Y, Nakao K, Jockushu BM and Narumiya S. (1997). EMBO J., 16, 3044 ± 3056. Yeramian P, Chardin P, Madaule P and Tavitian A. (1987). Nucleic acids Res., 15, 1869.