Rho small GTPase regulates cell morphology, adhesion and cytokinesis through the actin cytoskeleton. We have identified a protein, p140mDia, as a downstream effector of Rho. It is a mammalian homolog of Drosophila diaphanous, a protein required for cytokinesis, and belongs to a family of formin‐related proteins containing repetitive polyproline stretches. p140mDia binds selectively to the GTP‐bound form of Rho and also binds to profilin. p140mDia, profilin and RhoA are co‐localized in the spreading lamellae of cultured fibroblasts. They are also co‐localized in membrane ruffles of phorbol ester‐stimulated sMDCK2 cells, which extend these structures in a Rho‐dependent manner. The three proteins are recruited around phagocytic cups induced by fibronectin‐coated beads. Their recruitment is not induced after Rho is inactivated by microinjection of botulinum C3 exoenzyme. Overexpression of p140mDia in COS‐7 cells induced homogeneous actin filament formation. These results suggest that Rho regulates actin polymerization by targeting profilin via p140mDia beneath the specific plasma membranes.
The actin cytoskeleton plays a central role in cell motility, morphology, phagocytosis and cytokinesis. It is spatially and dynamically reorganized, providing force for the shape change and surface movement in most eukaryotic cells. Rearrangement of actin is evoked rapidly by extracellular stimuli, and sets of actin‐associated proteins are thought to act cooperatively in the polymerization, cross‐linking and anchorage of actin filaments. The small GTPase Rho has been shown to be required for many actin‐dependent cellular processes such as platelet aggregation (Morii et al., 1992), lymphocyte adhesion (Tominaga et al., 1993), cell motility (Takaishi et al., 1993) and cytokinesis (Kishi et al., 1993; Mabuchi et al., 1993). In cultured fibroblasts, microinjection of Rho causes rapid formation of actin stress fibers and focal adhesions. Conversely, inactivation of Rho by botulinum C3 ADP‐ribosyltransferase prevents this process (Ridley and Hall, 1992). C3 exoenzyme treatment also blocks lysophosphatidic acid‐, endothelin‐ or GTPγS‐induced tyrosine phosphorylation of focal adhesion kinase and paxillin (Kumagai et al., 1993; Rankin et al., 1994; Ridley and Hall, 1994; Seckl et al., 1995). These obsevations indicate that Rho regulates signal transduction pathways linking extracellular stimuli to the reorganization of the actin cytoskeleton.
Recent studies have identified putative downstream target molecules for Rho including citron (Madaule et al., 1995), p150ROK or Rho‐kinase or ROCK‐II (Leung et al., 1995; Matsui et al., 1996; Nakagawa et al., 1996), p160ROCK (ROCK‐I) (Ishizaki et al., 1996), rhophilin and PKN (Watanabe et al., 1996) and rhotekin (Reid et al., 1996). However, the molecular mechanisms by which Rho and these target molecules promote the biological responses described above are poorly understood, although it has been suggested that Rho‐kinase down‐regulates myosin phosphatase through phosphorylation of a regulatory subunit of this enzyme (Kimura et al., 1996). The involvement of phosphoinositide kinases in Rho signaling has also been reported. Rho (Chong et al., 1994) and Rac (Hartwig et al., 1995), another member of the Rho family of GTPases, were shown in different cell systems to stimulate the synthesis of phosphatidylinositol bisphosphate (PIP2). Since the binding of PIP2 is thought to regulate the function of many actin‐associated proteins (reviewed by Janmey, 1994; Jockusch et al., 1995), PIP2 synthesis stimulated by the Rho family of GTPases may induce actin reorganization.
One of the proteins regulated by PIP2 is profilin, which forms a 1:1 complex with G‐actin and which releases actin upon PIP2 binding. Profilin was thought originally to function in the sequestration of unpolymerized actin in the cytoplasm. Recent studies, however, have demonstrated that profilin itself has a promoting effect on actin polymerization. Profilin stimulates the ADP/ATP exchange of actin (Mockrin and Korn, 1980; Nishida, 1985; Goldschmidt‐Clermont et al., 1991) and low amounts of profilin can promote extensive actin assembly in the presence of the thymosin β4–G‐actin complex (Pantaloni and Carlier, 1993). The positive effect of profilin on actin polymerization is also supported by in vivo data, showing that microinjection of the profilin–actin complex increased the content of filamentous actin beyond the amount of injected actin (Cao et al., 1992). Transient membrane localization of profilin has been noted in activated platelets (Hartwig et al., 1989) and in the membrane ruffles of locomoting cells (Buß et al.,1992; Rothkegel et al., 1996). Focal increases in profilin concentration may, therefore, play an important role in promoting actin polymerization at specific sites in cells, although little is known about the mechanism for recruitment of profilin.
In the present study, we have identified a novel Rho target protein, p140mDia, which is a mammalian homolog of Drosophila diaphanous (Castrillon and Wasserman, 1994). p140mDia binds both to the GTP‐bound form of RhoA and to profilin through different regions of the molecule. We show that RhoA, p140mDia and profilin are co‐localized in the membrane ruffles of rapidly spreading cells and in phagocytic cups induced by fibronectin (FN)‐coated beads, both being formed in a Rho‐dependent manner.
Isolation of p140mDia partial cDNA in a two‐hybrid system
A mouse embryo cDNA library was screened to isolate a novel Rho‐binding protein using a yeast two‐hybrid system (Vojtek et al., 1993). LexA DNA‐binding protein fused to the Asn19‐RhoA truncated at Ala181 in the C‐terminus (Asn19–RhoAΔC) was used as bait. Among the His(+) and LacZ(+) yeast clones, 55 were selected which showed no LacZ activity with lamin used as a negative control. The clones yielded cDNA inserts of the same size, several of which were sequenced and found to be identical. When these clones were mated to AMR70 strains bearing a LexA fused to various RhoA mutants, they showed an interaction that was strongest with Val14‐RhoA, weak with Asn19‐RhoA and almost negligible with wild‐type RhoA, although they retained strong interaction with Asn19−RhoAΔC or wild‐type RhoAΔC. This specificity was confirmed by co‐transforming the L40 strain with a plasmid recovered from a representative clone, clone 50, and with various LexA–mutant RhoA fusion constructs (Figure 1). These results suggest that a peptide encoded by clone 50 cDNA preferentially interacts with the activated form of RhoA.
Specific association of p140mDia with the GTP‐bound form of RhoA in vitro
The polypeptide encoded by the cDNA isolated above was expressed in Escherichia coli either as a His‐tagged protein (His6‐cl.50) or as a GST fusion protein (GST–cl.50). An antiserum specific to p140mDia was raised against His6‐cl.50. The specificity of the antiserum was verified by its reactivity with GST–cl.50 expressed in E.coli after isopropyl‐β‐d‐thiogalactopyranoside (IPTG) induction (Figure 2A, lanes 1 and 2). The antiserum detected a single band in Swiss 3T3 cell lysates, which migrated to the 160 kDa position on SDS–PAGE (lanes 3 and 4). We designated this protein as p140mDia (see below). Using this antiserum, we examined the binding specificity of p140mDia for the GTP‐ or GDP‐bound form of Rho family GTPases. p140mDia was precipitated from Swiss 3T3 lysates only by the GTPγS‐bound form of GST–RhoA but not by the GDP‐bound form, GST–Rac1 or GST–Cdc42Hs (Figure 2B). These observations confirm the specificity of the interaction observed in the two‐hybrid system. The specific association of p140mDia with the activated form of RhoA in both assays indicates that p140mDia may work as a downstream effector of Rho.
p140mDia is highly homologous to Drosophila diaphanous and is a formin‐related protein
To obtain a full‐length coding sequence, we sequentially screened two mouse brain cDNA libraries and a mouse embryo cDNA library. Six overlapping clones were isolated (Figure 3A). The composite cDNA sequence from clones 502, 503, 504 and E52 contains an open reading frame which encodes a protein of 1255 amino acids with a calculated mol. wt of 139 336 (Figure 3B). The Rho‐binding region defined by clone 50 cDNA is located in the N‐terminal portion (amino acids 63–260). Between amino acid 571 and 737, there is a region with 14 repeats of polyproline stretches. The repeats are characterized by a motif, IPPPPPLPG, or similar sequences. Five repeats share this motif exactly, while there exist variations, including the extension of the polyproline by one or three prolines. The disruption of the polyproline at position 4 by alanine or serine is also seen in two repeats. We compared the amino acid sequence, excluding polyproline sequences, of this protein with other sequences in the databases using the BLASTP program (Stephen et al., 1990). The search revealed one highly homologous protein, Drosophila diaphanous, required for cytokinesis (Castrillon and Wasserman, 1994). Diaphanous also contains polyprolines in the middle and shows 30 and 39% identity upstream and downstream of the polyproline region to the respective regions of p140mDia (Figure 3C). Several related proteins have also been identified. These proteins include Bni1p of Saccharomyces cerevisiae (DDBJ/EMBL/GenBank L31766), mouse formin (Woychick et al., 1990), Drosophila cappuccino (Emmons et al., 1995), fus1p of Schizosaccharomyces pombe (Petersen et al., 1995) and FigA of Aspergillus nidulans (Marhoul and Adams, 1995). These proteins belong to a family of formin‐related proteins, and contain the polyproline region and the highly conserved portion in the C‐terminal region which Castillon and Wasserman (1994) designated as FH1 and FH2 domains, respectively. Additionally, distant homology is present in sequences from the polyproline region to the C‐terminus of all these molecules. Only diaphanous is highly homologous to p140mDia also in the N‐terminal part, which includes the Rho‐binding domain. p140mDia also shows weak but significant homology in the entire sequence, including the Rho‐binding region, to Bni1p (Figure 3D), which is involved in yeast cell budding. Northern blot analysis revealed that a major 6.3 kb transcript for this novel Rho‐binding protein was expressed ubiquitously in all mouse tissues examined (Figure 3E).
p140mDia binds to profilin in vitro
An actin‐binding protein, profilin, is known to bind also to poly‐l‐proline (PLP) in vitro (reviewed by Machesky and Pollard, 1993) and can be purified selectively by the use of PLP–Sepharose affinity chromatography (Tanaka and Shibata, 1985). Hence, we speculated that p140mDia might bind to profilin. Recently, Reinhard et al. (1995) reported that two polypeptides, VASP and an 81 kDa fragment of an unidentified 160 kDa protein, bound to profilin–agarose and were eluted by PLP solution. We noticed that the sequences of amino acids 731–769 and 1128–1152 of p140mDia were almost identical to the partial amino acid sequences of the fragment of this 160 kDa protein. We therefore examined whether p140mDia can bind to profilin and whether the binding of p140mDia to profilin is dependent on Rho, because the fragment isolated by Reinhard et al. (1995) corresponds to the C‐terminal part of p140mDia. As shown in Figure 4, p140mDia in Swiss 3T3 cell lysates was quantitatively precipitated by the addition of profilin–agarose, while no precipitation was seen with bovine serum albumin (BSA)–agarose. This interaction was not affected by the addition of exogenous RhoA. It also was not affected by the addition of Rac1 or Cdc42Hs or by the addition of GTPγS (data not shown).
Co‐localization of RhoA, p140mDia and profilin in membrane ruffles of motile cells
The intracellular distribution of p140mDia was first examined in Swiss 3T3 cells (Figure 5A and B). Although the majority of fluorescence obtained by affinity‐purified anti‐p140mDia antibody (AP50) was localized to the thicker regions of the cells, prominent fluorescence was observed in the spreading lamellae, where fine actin ribs are developed. No association of p140mDia with focal adhesions and stress fibers was observed. In mitotic cells, p140mDia was associated with the plasma membrane rather homogeneously. However, in some mitotic cells, p140mDia was concentrated in the cleavage furrow and appeared as a ring‐like structure (Figure 5C and D).
The subcellular localization of p140mDia, profilin and RhoA was then studied in HT1080 human fibrosarcoma cells. In addition to perinuclear staining, the peripheral lamellae of motile cells were enriched with the fluorescent signals for all three proteins. The signals were abolished by prior absorption of the antibodies with the respective antigens (Figure 6). This pattern of profilin distribution is consistent with the pattern demonstrated in the previous study of rat fibroblasts and BHK cells (Buß et al., 1992; Rothkegel et al., 1996). Using these antibodies and phalloidin, localization of the three molecules and F‐actin were compared by dual immunofluorescence. HT1080 cells extend massive ruffles around their periphery, which were strongly stained with phalloidin (Figure 7B). Some p140mDia was detected in these ruffles, and was concentrated most notably at their tips (Figure 7A). Dual fluorescence with anti‐p140mDia and anti‐profilin antibodies revealed co‐localization of the two molecules in membrane ruffles (Figure 7C and D), which is shown more clearly by confocal microscopy (Figure 7C′ and D′). Moreover, double immunofluorescence images produced by the polyclonal anti‐RhoA antibody and anti‐profilin antibody revealed nearly identical patterns at the membrane ruffles of the cells (Figure 7E and F, E′ and F′), suggesting that p140mDia, profilin and RhoA are associated in these highly motile structures.
We also examined the localization of p140mDia, RhoA and profilin in sMDCK2 cells which stably express myc‐tagged RhoA and which rapidly extend membranes in response to 12‐O‐tetradecanoylphorbol‐13‐acetate (TPA) (Takaishi et al., 1995). Both p140mDia and myc‐RhoA were distributed rather homogeneously in the cytoplasm of resting cells. After stimulation by TPA for 15 min, a portion of myc‐RhoA moved to the peripheral membrane ruffles, where p140mDia as well as profilin was co‐localized (data not shown). Because the TPA‐induced membrane extension and ruffles of these cells occur in a Rho‐dependent manner (Takaishi et al., 1995), these results demonstrate the co‐localization of the three proteins, RhoA, p140mDia and profilin, in a Rho‐dependent structure.
Activation‐dependent clustering of RhoA, p140mDia and profilin around fibronectin‐coated beads
Recent studies indicate that integrin ligation by either FN‐ or anti‐integrin antibody‐coated beads recruits RhoA and p190RhoGAP‐B to the plasma membrane beneath the beads (Burbelo et al., 1995b; Miyamoto et al., 1995). We therefore examined if p140mDia and profilin are also recruited to the plasma membrane around FN‐coated beads and co‐localize with RhoA. As shown in Figure 8A and C, we confirmed that RhoA was recruited efficiently by the FN‐coated beads and clustered around the beads. Much less accumulation was found around the PLL‐coated beads (data not shown). Under these conditions, p140mDia also accumulated around the FN‐coated beads and exactly co‐localized with RhoA (Figure 8B). Co‐localization of profilin was also noted (Figure 8D). To test if the recruitment of p140mDia is dependent on Rho, we inactivated endogenous Rho by microinjecting C3 exoenzyme and then carried out the bead assay on the injected cells. As shown in Figure 8E–H, no recruitment of p140mDia and RhoA was found in the C3 exoenzyme‐injected cells, indicating that activation of Rho is required for the recruitment of these molecules to phagocytic cups around the beads.
Overexpressed p140mDia induces homogeneous fine actin filaments
Finally, we examined the effect of p140mDia overexpression on the actin cytoskeleton in COS‐7 cells. Cells overexpressing p140mDia showed the homogeneous anti‐p140mDia staining with clear cell contours (Figure 9A), indicating that some of the p140mDia accumulated on the plasma membrane. This membrane association was a persistent finding in transfected cells, although a portion of p140mDia formed aggregates in the cytoplasm of cells expressing a very high level of p140mDia or co‐expressing Val14‐RhoA (data not shown). Double staining with phalloidin revealed reduction in stress fibers and enhancement of fine F‐actin staining in almost all cells showing p140mDia overexpression (Figure 9B and B′). The homogeneous distibution of fine F‐actin was observed even after p140mDia was co‐expressed with C3 exoenzyme. Expression of C3 exoenzyme alone almost abolishes F‐actin staining in COS‐7 cells (see examples of C3 exoenzyme phenotype of COS‐7 cells; arrowheads in Figure 9D and D′). When p140mDia and C3 exoenzyme were expressed together at a higher level, the cells showed homogeneous F‐actin staining (arrows in Figure 9C and D, and Figure 9D′) similar to that obeserved in cells overexpressing p140mDia alone. The induction of fine actin filaments was persistently observed in all cells that expressed a high level of p140mDia and C3 exoenzyme, indicating that overexpressed p140mDia can promote actin polymerization in the absence of Rho activity. To evaluate the nature of F‐actin induced by overexpressed p140mDia, we extracted the cells with 0.1% Triton in the presence of phalloidin before fixation. Because this procedure extracted most of the expressed p140mDia, we examined F‐actin staining only in the cells containing overexpressed p140mDia as insoluble aggregates. F‐actin staining remained homogeneous in these cells (data not shown).
p140mDia is a novel target protein of Rho
This two‐hybrid screening initially was aimed at isolating a novel exchange factor for RhoA. Asn17‐Cdc42Hs binds strongly to a dbl oncogene product bearing a catalytic domain for the GDP/GTP exchange of Cdc42Hs and RhoA (Hart et al., 1994). Indeed, Asn19‐RhoA and its C‐terminal truncated form interact with Dbl far more strongly than does wild‐type RhoA in a two‐hybrid system (N.Watanabe, unpublished results). However, we did not obtain any proteins with Dbl homology in this screening. This may be due to the short size of the cDNA in the library used in the two‐hybrid screening, which we found up to ∼650 bp. Hart et al. (1994) reported that the minimum Dbl domain consisted of 260 amino acids. By this screening, we have isolated the Rho‐binding domain of p140mDia instead. Strong interaction of p140mDia with Val14‐RhoA in the two‐hybrid system and its specific association with the GTPγS‐bound form of RhoA in vitro indicate that p140mDia is a downstream target molecule of Rho. Among the effectors of the Rho family of GTPases reported thus far, PKN, rhophilin and rhotekin possess a conserved Rho‐binding motif (Reid et al., 1996), and PAK, STE20 and WASP of Rac/Cdc42 effectors share a conserved Rac/Cdc42‐binding motif (Burbelo et al., 1995a; Symons et al., 1996). However, p140mDia does not show any significant homology to other effectors of Rho.
p140mDia is a mammalian homolog of Drosophila diaphanous
Structural elucidation and a comparison of p140mDia with sequences in databases has revealed that p140mDia belongs to a family of formin‐related proteins. The functions of all of these proteins are related to the regulation of cell structure and polarity. Drosophila diaphanous is required for cytokinesis. Bni1p is involved in yeast bud formation. Formin is a product of a mouse gene for proper limb pattern formation. Cappuccino is essential for the polarity of Drosophila egg formation. Fus1 of S.pombe is required for cell wall fusion during conjugation. These proteins share homology in sequences extending from the polyproline region to the C‐terminus, while the N‐terminal regions are divergent. For example, formin is spliced in its N‐terminal region, yielding variants bearing N‐termini of quite different isoelectric points (Jackson‐Grusby et al., 1992). Hence, it appears that formin‐related molecules have preserved the C‐terminal half during evolution, which suggests that this region is associated with functions common to all of these molecules. In particular, the polyproline region (FH1) and the highly‐conserved FH2 domain may serve as domains which interact with some cytoskeletal elements. In contrast, the divergent N‐termini may provide the differentially evolved function for each molecule. p140mDia is highly homologous only to diaphanous in the N‐terminus, which includes the Rho‐binding domain. Relative enrichment of p140mDia in the cleavage furrow of some dividing cells may support the functional conservation between p140mDia and diaphanous. Taken together, these data suggest that p140mDia is a mammalian homolog of Drosophila diaphanous. In addition, Bni1p, which retains distant homology to p140mDia also in the Rho‐binding domain, may be a yeast homolog of p140mDia.
p140mDia, a ligand for profilin
Castillon and Wasserman (1994) noted the similarity between the mutant phenotypes of diaphanous and chickadee, a Drosophila profilin, and suggested the possibility that diaphanous binds to chickadee through its polyproline motifs. Disruption of the profilin gene results in a cytokinesis‐defective phenotype in Drosophila (Verheyen and Cooley, 1994), S.pombe (Balasubramanian et al., 1994) and Dictyostelium (Haugwitz et al., 1994). Similarly, inactivation of Rho by C3 exoenzyme also prevents the formation and maintenance of the cleavage furrow in dividing oocytes (Mabuchi et al., 1993). Moreover, both Rho (Aullo et al., 1993) and profilin (Theriot et al., 1994) are required for the movement of intracellularly infected Listeria monocytogenes. Hence, there are certain cellular processes requiring all three proteins, Rho, diaphanous and profilin. Here, we have demonstrated that intact p140mDia can bind to profilin in vitro, and its distribution in vivo largely overlaps with that of profilin in cells. These results strongly indicate that profilin associates physically with p140mDia not only in vitro but also in vivo. VASP, showing a similar in vivo co‐localization to profilin, has been the only ligand of profilin identified thus far among many polyproline‐containing proteins (Reinhard et al., 1995). VASP and p140mDia appear to be two major profilin ligands in cells, because they were the two major proteins isolated by profilin–agarose affinity chromatography (Reinhard et al., 1995). Recently, another polyproline protein, WASP, was identified as a putative effector for Cdc42Hs (Symons et al., 1996). While overexpression of WASP induces actin polymerization, no link between WASP and profilin has been reported yet.
Rho‐dependent targeting of profilin: possible mechanism for focal actin reorganization
The present study has examined extensively the subcellular localization of endogenous RhoA, p140mDia and profilin, and has shown that the three proteins are co‐localized in membrane ruffles, especially at the tip of ruffles, of motile cells. Co‐localization of these three proteins was also observed around the phagocytic cups induced by FN‐coated beads. It is noteworthy that the co‐localization in both structures is dependent on the action of Rho. Firstly, on TPA stimulation, sMDCK2 cells rapidly extend their membranes with fine ruffles at their edges, where myc‐RhoA is co‐localized with p140mDia and profilin. Microinjection of C3 exoenzyme or Rho‐GDI abolished membrane extension and ruffle formation of these cells (Takaishi et al., 1995). Secondly, the recruitment of RhoA and p140mDia around FN‐coated beads is also abolished by prior injection of C3 exoenzyme. These results strongly suggest that p140mDia and profilin are recruited to these dynamic membrane structures by the action of Rho.
As Grinnel and Geiger (1986) noted, the phagocytic membranes around the FN‐coated beads are morphologically similar to the membrane ruffles seen in the edges of cultured cells. Recently, similar Rho‐dependent membrane folding was reported at the entry of Shigella into epithelial cells (Adam et al., 1996). Rho is also required for HGF‐ and TPA‐induced membrane ruffles in KB cells, which are morphologically different from Rac‐dependent, insulin‐induced ruffles (Nishiyama et al., 1994). All of these structures are associated with dynamic actin polymerization, and appear to be highly related in view of the morphological similarities and the functional connections with Rho. Profilin is also thought to be involved in actin polymerization beneath the dynamic plasma membranes. It is transiently translocated to the plasma membrane of stimulated platelets and leukocytes (Hartwig et al., 1989) and is concentrated in the ruffling membranes of locomoting fibroblasts (Buß et al., 1992). Since a small amount of profilin can promote extensive actin filament assembly from the G‐actin–thymosin β4 precursor pool (Pantaloni and Carlier, 1993), the targeting of profilin has been thought to be important in the enhanced actin polymerization at these membranes. One candidate responsible for this targeting is PIP2 (Hartwig et al., 1989). While the synthesis of PIP2 can be stimulated by Rho (Chong et al., 1994), it is not known yet whether Rho activates PIP2 synthesis locally in the above membranes. The identification of p140mDia suggests that Rho and p140mDia are the long‐sought targeting vehicles for profilin, and that the p140mDia–profilin complex targeted by Rho possibly acts on focal actin polymerization. This idea would be in agreement with the results of transient p140mDia expression in COS‐7 cells. The overexpressing cells showed the diffuse staining for p140mDia and the enhanced fine actin filament assembly in the entire cell. Co‐expression with C3 exoenzyme did not alter this diffuse localization of p140mDia. These results showed that overexpressed p140mDia was not regulated by Rho and was distributed evenly in the cell. The homogeneous distribution of the induced F‐actin in both transfectants may thus reflect the loss of a precise mechanism for its localization.
On the bases of these observations, we have devised a model for a p140mDia‐mediated action of Rho as depicted in Figure 10. In this scheme, Rho is activated locally and recruits p140mDia–profilin beneath a specific site of the plasma membrane. The locally increased concentration of profilin then promotes actin polymerization. It is likely that multiple profilin molecules bind simultaneously to a single p140mDia. Since it has been shown that profilin bound to actin exposes the PLP‐binding site (Tanaka and Shibata, 1985; Schutt et al., 1993), the profilin–actin complex may also bind to p140mDia. Whether p140mDia influences profilin‐catalyzed actin polymerization is an interesting question to be investigated. PIP2 possibly also works cooperatively in this context.
It is well known that microinjection of Rho induces focal adhesion and stress fiber formation in cultured fibroblasts (Ridley and Hall, 1992). However, neither Rho nor p140mDia has ever been observed to be concentrated at focal adhesions or stress fibers (Adamson et al., 1992; Takaishi et al., 1995; this study). Recent reports showed that Rho‐induced focal adhesion and actin polymerization are inhibited differentially by staurosporine and cytochalasin D (Nobes and Hall, 1995). We presume, therefore, that two or more Rho effectors work in combination to induce Rho effects and that some may work at earlier steps before the final structures are stabilized. Based on the experiments on molecular clustering around the FN‐coated beads (similar to experiments employed in this study), Miyamoto et al. (1995) proposed a multi‐step model for focal complex formation.
In summary, we have identified p140mDia as a new Rho target. It provides a direct molecular link between Rho and profilin, which have been characterized separately as a regulator for the actin cytoskeleton. The functions and subcellular localization of the three molecules suggest that they work cooperatively in actin reorganization beneath the dynamic plasma membranes. Future studies will reveal how these molecules work together with other Rho effectors and the actin cytoskeleton in dynamic movement of the plasma membranes.
Materials and methods
pGEX‐RhoA has been described previously (Watanabe et al., 1996). pGEX‐Rac1 and pGEX‐Cdc42Hs were gifts of Y.Takai. pBTM‐RhoA was prepared as described previously (Watanabe et al., 1996). To generate Val14 and Asn19 mutations on RhoA, the BamHI–EcoRV fragment of the pGEX‐RhoA encoding the RhoA N‐terminus was ligated into pBluescript (Stratagene), and mutagenized according to the method of Kunkel (1985). The corresponding wild‐type fragment of pGEX‐RhoA was then replaced by each mutagenized fragment, and the BamHI–EcoRI fragments encoding the full‐length coding region of these mutant RhoAs were ligated into pBTM116 to yield pBTM‐Val14‐RhoA and ‐Asn19‐RhoA, respectively. The C‐terminal deletion of these mutants at Ala181 (pBTM‐Val14‐RhoAΔC and ‐Asn19‐RhoAΔC) was prepared as described previously (Reid et al., 1996). The BamHI–BamHI fragment of pGEX‐Cdc42Hs was also ligated into pBTM116. All the other plasmids used in the two‐hybrid system, including the cDNA library, were kindly provided by Stan Hollenberg, Rolf Sternglanz, Stan Fields and Paul Bartel (Vojtek et al., 1993).
Yeast two‐hybrid screening
The yeast L40 strain harboring pBTM‐Asn19‐RhoAΔC was transformed with pVP16 fused with a mouse embryo cDNA library. Initial transformation yielded 2.2×107 clones, which were amplified seven times during the 6 h culture before spreading on His (−) plates. Among 1.5×108 transformants, 978 clones were isolated as His+ and LacZ+, and 220 clones were cured from the bait plasmid. Interactions with other proteins were evaluated by mating with yeast strain AMR70 harboring various test baits. Clones interacting with a negative control, lamin, were eliminated. When the remaining 55 clones were mated to AMR70 strains bearing LexA fused to various RhoA mutants, all of them showed an interaction that was strongest with Val14‐RhoA, less strong with RhoAΔC and wild‐type RhoAΔC, weak with Asn19‐RhoA and almost negligible with wild‐type RhoA. The plasmids recovered from these clones were identical. A representative clone, clone 50, was used for further analysis. To confirm the specificity of the interactions in a two‐hybrid system, a pVP16 plasmid recovered from yeast clone 50 (pVP‐cl.50) was co‐transformed into L40 with various pBTM116 plasmids. These transformants were plated as patches on selective medium, transferred on cellulose filters, and β‐galactosidase activity was assayed as described (Vojtek et al., 1993).
Two mouse brain libraries (936309 in λZAP II, Stratagene, and ML3000a in λgt‐10, Clontech) were screened with the 32P‐labeled 0.6 kb cDNA insert of pVP‐cl.50. One positive clone, clone 502, and two positive clones, clones 503 and 504, were obtained from the former and the latter library, respectively. Using the 5′ part of 504 and the 3′ part of 503 as probes, clones E51, E52 and E73 were isolated from a mouse embryo library (Nakagawa et al., 1996). Nucleotide sequence was determined using the dideoxy chain termination method.
Northern blot analysis
Poly(A)+ RNA was prepared from several tissues of adult mice using oligo(dT) latex beads according to the standard procedure (Sambrook et al., 1989). Six μg of each poly(A)+ RNA was separated on a 1.0% agarose gel containing 2.1% formaldehyde, and transferred to a Biodyne A filter (Pall BioSuport, NY). The filter was then hybridized with the 32P‐labeled 0.6 kb cl.50 cDNA. The filter was washed finally with 0.4× SSC and 0.1% SDS at 65°C, and subjected to autoradiography.
Anti‐p140mDia antibody was prepared as follows. The BamHI and EcoRI sites of pVP‐cl.50 flanking the cDNA insert were used to ligate cl.50 cDNA into pQE11 (QIAGEN) and pGEX‐3X (Pharmacia) vectors. His6‐tagged cl.50 was expressed in E.coli JM109 strain and was purified with Ni‐NTA resin (QIAGEN) according to the manufacturer‘s protocol. The purified protein mixed with Freund's adjuvant was injected into rabbits, and anti‐p140mDia antiserum was raised. The antibody was then affinity purified using GST–cl.50 fusion protein immobilized on nitrocellulose membranes essentially as described (Reinhard et al., 1992). Briefly, inclusion bodies containing GST–cl.50 were isolated from E.coli and solubilized in Laemmli's buffer. The solubilized proteins were separated by SDS–PAGE and transferred to the nitrocellulose membranes. A band of GST–cl.50 was excised as a strip, and antibodies absorbed to this strip were eluted at 4°C successively with 100 mM glycine–HCl buffer (pH 2.3), 100 mM monoethanolamine buffer (pH 11.5) and 100 mM glycine–HCl containing 10% 1.4‐dioxane (pH 2.5). The eluates were neutralized immediately with 0.25 vol. of 250 mM sodium phosphate buffer (pH 8.8 for the first and third eluates and pH 7.0 for the second eluate). These eluates were combined and used as the affinity‐purified antibody, AP50.
Polyclonal rabbit anti‐profilin antibody was described previously (Buß et al., 1992) and monoclonal mouse 2H11 antibody against bovine profilin was described elsewhere (Mayboroda et al., 1997). Both anti‐profilin antibodies selectively interact with profilin I and not with profilin II. Rabbit polyclonal 119 anti‐RhoA antibody and mouse monoclonal 26C4 anti‐RhoA antibody (Lang et al., 1993) were obtained from Santa Cruz Biotechnology, Inc. Mouse monoclonal 9E10 anti‐myc antibody (Evan et al., 1985) was a gift of S.Nishikawa, Kyoto University.
These antibodies were depleted by incubating an aliquot of AP50, 2H11 and 119 solution successively with five pieces of membranes blotted with GST–cl.50, purified profilin and GST–RhoA, respectively. The antibody‐depleted solution was used at the same dilution as the original antibody.
Affinity precipitation of p140mDia by Rho family proteins in vitro
GST fusion proteins of RhoA, Rac1 and Cdc42Hs were expressed and prepared according to the manufacturer's protocol. Confluent Swiss 3T3 cells, ∼1×107 cells, were collected and disrupted by sonication (5 s, four times) in 3.2 ml of buffer A [10 mM MES pH 6.5, 150 mM NaCl, 2 mM MgCl2, 0.5 mM EDTA, 0.5% Triton X‐100, 5 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 μg/ml leupeptin]. Sonicated homogenates were centrifuged at 10 000 g for 20 min, and the supernatant was saved. Loading of each nucleotide was carried out by incubating 10 μM GST–Rho GTPases with 1 mM GTPγS or GDP under the same conditions as described previously (Ishizaki et al., 1996). One‐tenth of the supernatant was then incubated with 400 pmol of each nucleotide‐loaded GST–Rho GTPase. After incubation at 30°C for 30 min, 5 μl of glutathione–Sepharose4B was added to the solution and the mixture was incubated at 4°C for 1 h. The beads were washed twice with 1 ml of buffer A, and boiled in Laemmli sample buffer. The solubilized extracts were subjected to immunoblotting with anti‐cl.50 antiserum according to the procedure previously published (Kumagai et al., 1993).
Cells and immunofluorescence
Swiss 3T3 cells, sMDCK2 cells stably expressing myc‐tagged RhoA (Takaishi et al., 1995) (a gift of K.Takaishi and Y.Takai) and HT1080 human fibrosarcoma cells (a gift of K.Sekiguchi) were grown in Dulbecco‘s modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS). sMDCK2 cells were maintained in 500 μg/ml G‐418. For morphological analysis, HT1080 and Swiss 3T3 cells were seeded onto glass coverslips at a density of 1×105 cells per 35 mm dish, cultured overnight and fixed. sMDCK2 cells were seeded onto glass coverslips at a density of 1×104 cells per 35 mm culture dish in DMEM containing 10% FCS, and incubated for 16 h. The medium was then changed to DMEM without FCS, and the cells were incubated for another 24 h. Serum‐starved cells were stimulated with or without 10−7 M TPA (Sigma) for 15 min at 37°C, and fixed.
For indirect immunofluorescence, cells were fixed in phosphate‐buffered saline (PBS) containing 3.7% paraformaldehyde for 20 min at room temperature, and then permeabilized with 0.2% Triton X‐100 in PBS for 10 min. After several washes with PBS, cells were incubated in buffer B (20 mM Tris pH 7.4, 50 mM NaCl) containing 5% BSA at room temperature for >30 min. For staining with rabbit polyclonal antibodies, the cells were incubated either with a 1:10 dilution of AP50 for p140mDia staining, with a 1:40 polyclonal anti‐RhoA antibody, or with a 1:80 dilution of anti‐profilin antiserum in the blocking solution at room temperature for 1 h, and washed three times with buffer B containing 0.1% Triton X‐100. The cells were then stained with Cy2‐labeled goat anti‐rabbit IgG (Amersham Life Science) and washed five times with buffer B plus 0.1% Triton X‐100. For dual immunofluorescence, either 9E10 anti‐Myc antibody at 10 μg/ml, 2H11 monoclonal anti‐profilin antibody at a 1:2 dilution or 26C4 monoclonal anti‐RhoA antibody at a 1:50 dilution, was added to the primary antibody incubation. Rhodamine‐conjugated anti‐mouse IgG (Organon Technica Corp.) was then used at a 1:50 dilution. For F‐actin staining, rhodamine‐conjugated phalloidin (Molecular Probe) was added to the second antibody incubation. Nuclear staining with DAPI (4′,6‐diamidino‐2‐phenylindole) was performed as described (Dyck et al., 1994). Fluorescence images were photographed with a conventional fluorescence microscope (Axiophoto, Zeiss) or with a confocal laser scanning microscope (MRC1024, Bio‐Rad).
In some experiments, COS cells overexpressing p140mDia were first washed with a buffer containing 10 mM MES, pH 6.1, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl2 and 5 mM glucose, and extracted with 0.1% Triton X‐100 in this buffer in the presence of 10 μM phalloidin for 10, 30 and 300 s before fixation.
Latex bead binding and microinjection in Swiss 3T3 cells
Polystyrene latex beads (11.9 μm average diameter, Sigma) were coated with either 50 μg/ml human FN (Collaborative Research, Inc.) or 100 μg/ml poly‐l‐lysine (PLL) (Sigma) as described (Grinnel and Geiger, 1986). Trypsinized Swiss 3T3 cells were plated onto FN‐coated coverslips and allowed to attach to the slips for 2 h at 37°C in DMEM containing 10% FCS. Each different type of beads was then placed onto the cells. After incubation for 15 min at 37°C, the cells were fixed. For microinjection of C3 exoenzyme, recombinant C3 exoenzyme was prepared as described (Morii and Narumiya, 1995). C3 exoenzyme at 150 ng/μl in 10 mM HEPES pH 7.2, 2 mM MgCl2, 20 mM KCl and 0.1 mM DTT was injected into the cells, plated and attached as described, with 0.5 μg/μl rabbit IgG or mouse IgG (Zymed) for the detection of injected cells. After the cells were incubated for 20–30 min, binding of FN‐coated beads was then carried out as described above.
Human platelet profilin was purified using PLP affinity chromatography, as described previously (Janmey, 1991). Briefly, 250 mg of PLP (Mr 12 000, Sigma) was coupled to CNBr‐activated Sepharose 4B (Pharmacia). Washed human platelets were prepared from the buffy coat fraction of 100 U of blood as described (Ishizaki et al., 1996). The platelets were disrupted in 200 ml of extraction buffer, and the supernatants were applied to the PLP–Sepharose. The homogenous preparation of profilin was obtained by the elution with 7 M urea after 4 M urea washes. Profilin, 0.96 mg, was then conjugated with 1 ml of NHS‐HiTrap (Pharmacia) according to the manufacturer‘s protocol (immobilized profilin). As a control, the same amount of BSA was similarly coupled to NHS‐HiTrap. Rho family GTPases were prepared as GST fusion proteins as described, cleaved from GST according to the manufacturer's protocol, and loaded with GTPγS or GDP as described above. Confluent Swiss 3T3 cells obtained from twelve 6 cm dishes were solubilized in 2.4 ml of buffer C (10 mM Tris–HCl, pH 7.0, 150 mM NaCl, 50 mM NaF, 2 mM MgCl2, 0.5 mM EDTA, 1 mM Na3VO4, 0.1% Triton X–100, 1 mM PMSF, 1 mM benzamidine, 5 μg/ml leupeptin) and centrifuged at 10 000 g for 10 min. A one‐tenth aliquot of the supernatant was then incubated with 20 μl of immobilized profilin with the addition of free GTPγS or GDP, a GTPγS‐ or GDP‐loaded Rho family GTPase, or a vehicle. The final concentration of Rho GTPases added was 1 μM. After the mixture was incubated for 30 min at 25°C, the beads were spun down at 1000 g for 2 min. The supernatant was saved. The beads were washed once with 100 μl of buffer C containing 300 mM NaCl. Half of the washed beads and one‐tenth of the saved supernatant were boiled in Laemmli buffer and subjected to anti‐p140mDia immunoblotting using the ECL system (Amersham).
Construction of the expression vector for p140mDia and transfection
pCMX‐p140mDia was constructed by sequentially ligating in pCMX vector (Dyck et al., 1994) the EcoRI–BamHI fragment of clone 504 cDNA subcloned in pBluescript, the BamHI–EcoRI fragment of clone 503 cDNA in pBluescript and the EcoRI–EcoRI fragment of clone E51 cDNA in pBluescript. Construction of a mammalian expression vector of C3 exoenzyme, pEFBOS‐C3, will be described elsewhere. Transfection of COS‐7 cells with either pCMX‐p140mDia, pEFBOS‐C3, pCMX‐FLAG‐Val14‐RhoA, or in combination, was performed as described (Ishizaki et al., 1996; Watanabe et al., 1996). The cells were fixed 25 h after the addition of plasmid DNA, and stained with anti‐p140mDia, anti‐FLAG M2 antibody (Kodak) or rhodamine–phalloidin as described above.
Note added in proof
Bni1p was recently reported to be a putative target of Rho1p or Cdc42p of Saccharomyces cerevisiae [ ; ]. The ROCK/ROKα/Rho‐kinase family of protein kinases have been shown to mediate Rho‐induced formation of focal adhesions and stress fibers [ ; ; ].
We are indebted to Stan Hollenberg, Rolf Sternglanz, Stan Fields and Paul Bartel for the yeast two‐hybrid system and to Kenji Takaishi and Yoshimi Takai for sMDCK2 cells, pGEX‐Rac1 and pGEX‐Cdc42Hs. We thank S.Nishikawa for 9E10 antibody, K.Sekiguchi for HT1080 cells and M.Symons for discussion. We are most grateful to Y.Kishimoto and H.Fuyuhiro for their skilled assistance and to K.Okuyama for secretarial work. We also thank T.Murata and H.Boku for help in the preparation of antibody, O.Nakagawa for a mouse cDNA library and K.Fujisawa for pEFBOS‐C3. This work was supported in part by a Grant‐in‐Aid for Specially Promoted Research from the Ministry of Education, Science and Culture of Japan and by a HFSP grant.
- Copyright © 1997 European Molecular Biology Organization