A novel serine/threonine kinase binding the Ras-related RhoA GTPase which translocates the kinase to peripheral membranes.

We previously reported the cloning of a serine/threonine kinase, PAK (for p21 (Cdc42/Rac)-activated kinase), which binds to the Ras-related GTPases Cdc42Hs and Rac1 (Manser, E., Leung, T., Salihuddin, H., Zhao, Z-s., and Lim, L. (1994) Nature 367, 40-46). These p21 proteins together with RhoA comprise the Rho subfamily of proteins that are involved in morphological events. We now report the isolation of a rat cDNA encoding a 150-kDa protein, which specifically binds RhoA in its GTP form and contains an N-terminal serine/threonine kinase domain highly related to the human myotonic dystrophy kinase and a cysteine-rich domain toward the C terminus. The RhoA binding domain is unrelated to other p21 binding domains. Antibody raised against the kinase domain of the predicted protein, termed ROK alpha (for ROK alpha, RhoA-binding kinase), recognized a ubiquitous 150-kDa protein. The brain p150 purified by affinity chromatography with RhoA exhibited serine/threonine kinase activity. In cultured cells, immunoreactive p150 was recruited to membranes upon transfection with dominant positive RhoAV14 mutant and was localized with actin microfilaments at the cell periphery. These results are consistent with a role for the kinase ROK alpha as an effector for RhoA.

The activation of cells by a number of growth factors through their surface receptors specifically affect the cytoskeleton. Some of these effects such as the formation of stress fibers (1, 2) lamellipodia (3), and filopodia (4,5) can be achieved by direct microinjection into cells of specific p21 proteins of the Rho subfamily, indicating that these p21 proteins act downstream of these growth factor receptors. Very little information is available concerning the signaling pathways beyond this point. A recent report has implicated a tyrosine kinase in the RhoA-mediated formation of stress fibers (6). In searching for potential targets of different p21 proteins of the Rho family, we developed methods for detecting direct interaction of the p21 proteins with putative target proteins and isolated cDNAs encoding the tyrosine kinase ACK 1 (7) and the serine/threonine kinase PAK (8). ACK and PAK have related sequences responsible for the interaction with Cdc42 and Cdc42/Rac, respectively (8). We have also demonstrated that there are cellular proteins that can specifically interact with RhoA, with a p150 being ubiquitously present in all tissues examined (8). We now report the isolation and characterization of a cDNA encoding this RhoA-binding protein. The deduced amino acid sequence predicts a N-terminal serine/threonine kinase domain homologous to myotonic dystrophy kinase (9,10). A 90-amino acid region contains a distinctive motif that binds RhoA in the GTP-bound form. Overexpression of RhoA results in the recruitment of ROK␣ to peripheral membranes, which is consistent with the kinase being a specific target for RhoA.

MATERIALS AND METHODS
Expression Screening-A rat brain ZAP library (Stratagene) was used for the expression screening. Plates containing phages 6 h after plating were induced for 16 h with nitrocellulose filters (20 ϫ 20 cm) previously soaked with 10 mM isopropyl-␤-thiogalactoside. Filters were denatured with 6 M guanidinium chloride in decreasing concentrations and returned to a renaturation buffer containing phosphate-buffered saline with 1% bovine serum albumin, 0.5 mM MgCl 2 , 0.1% Triton X-100 and 5 mM dithiothreitol. Filters were probed with solution containing [␥-32 P]GTP-glutathione S-transferase (GST)/RhoA and washed as described (7). Plaques showing binding were detected by autoradiography. Six overlapping clones (rbf-1 to rbf-6) were obtained from 5 ϫ 10 5 plaques after secondary and tertiary screening.
Sequence Analysis and Construction of Expression Vectors-Sequencing of exonuclease III/S1 nuclease nested-deleted subclones was carried out in both directions using the Sequenase sequencing kit (U. S. Biochemical Corp.). The construct rbf-7 was derived from a HincII-HindIII digest of rbf-1 and subcloned in the expression vector pMAL (New England Biolabs) encoding maltose-binding protein (HindIII was from vector polylinker site). Likewise, rbf-8 was derived from a SpeI-HindIII fragment of rbf-1. For raising antibodies, a 5Ј HindIII fragment (corresponding to amino acid residues 33-316) was subcloned in frame into the pMAL vector for expression. Fusion proteins of rbf-7 and rbf-8 in pMAL vector were prepared according to the recommended protocol, separated on 9% SDS-polyacrylamide gels, blot-transferred to nitrocellulose membrane, and renatured for p21 binding assays (8).
Tissue and Cell Preparation, Detection of p21 Binding Activities, and Western Blotting-Rat C6 glial, human SK-N-SH neuroblastoma, mouse NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 4.5 g/liter glucose and maintained with 5% CO 2 . Rat cerebellar granule cells were cultured as described by Leung (11). Adult Sprague-Dawley rat tissues and cultured cells were homogenized in extraction buffer containing 50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 0.1% Triton X-100, 2 g/ml each of leupeptin, pepstatin, and aprotinin. Soluble extracts (150 g) obtained by centrifuging at 100,000 ϫ g for 30 min at 4°C were separated on 9% SDS-polyacrylamide gels, blot-transferred to nitrocellulose membrane, and renatured for RhoA binding assays as described earlier. For immunological analysis, blots were first blocked with 5% skim milk before using mouse polyclonal antisera or a monoclonal antibody 1A1 raised against the N-terminal kinase domain.
To determine the specificity and nucleotide dependence of p21 binding, GST/RhoA and GST/Cdc42 (8) were phosphorylated using chicken * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) U38481.
Purification of a p150 Kinase from Rat Brain-Rat brain cytosol from 20 adult male rats (about 40 g) in Q buffer containing 25 mM Tris-HCl, pH 8.0, 0.5 mM MgCl 2 , 0.05% Triton X-100, and 0.1 mM EDTA was loaded onto an 80-ml Sepharose-Q column (Pharmacia Biotech Inc.), washed with 120 ml of buffer Q containing 0.1 M NaCl. The p150enriched 0.3 M NaCl eluent, as determined by RhoA binding and immunoblotting assays, was diluted to 0.1 M NaCl with S buffer containing 25 mM MES-NaOH, pH 6.0, 0.5 mM MgCl 2 , 0.05% Triton X-100 and loaded onto a 20-ml column with Sepharose-S (Pharmacia). This was washed with 50 ml of S buffer with 80 mM NaCl and eluted at 0.2 M NaCl. This p150-enriched fraction was then loaded onto a 1-ml glutathione-Sepharose column preloaded with 5 mg/ml GTP␥S-GST/RhoA. After washing with three column volumes of S buffer, binding proteins were eluted with Q buffer of increasing pH (pH 6 -9) before finally debinding with 5 mM glutathione. The presence of p150 ROK␣ was confirmed by immunoblotting with the specific antibody with the fraction eluted at pH 8 (E3) giving the best recovery of purified p150. Kinase assay and subsequent phosphoamino acid analysis were performed as described previously (8).
Transient Transfection and Morphological Analyses-Cos-7 cells were grown in DMEM plus 10% fetal bovine serum (FBS) (Life Technologies, Inc.) in 5% CO 2 . Cells at 80% confluence were transfected with vector pXJ40 (13) containing the hemagglutinin (HA) tag as control or vectors with wild type RhoA or activated RhoA V14 using Lipofectamine (Life Technologies, Inc.). After 5 h in serum-free medium and 12 h in 5% FBS, cells were harvested in lysis buffer containing 20 mM HEPES, pH 7.3, 150 mM NaCl, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, 0.5 mM phenylmethylsulfonyl fluoride, 2 g/ml each of pepstatin, aprotinin, and leupeptin. After centrifugation at 100,000 ϫ g for 30 min, supernatants were collected, and the pellets were extracted with lysis buffer containing 0.5% Triton X-100 and 1% sodium deoxycholate to give particulate fractions. Soluble and particulate fractions (100 g of protein) were resolved on a 10% polyacrylamide gel for Western blotting with either antibodies to ROK␣ or to HA (Boehringer, Mannheim, Germany). For cell staining, HeLa cells were grown in multichamber tissue culture slides (Lab-Tek) in DMEM supplemented with 10% FBS and maintained at 37°C in 5% CO 2 . After 16 h of transfection with pXJ40 vector with either RhoA V14 or Cdc42 V12 , cells were fixed with 4% paraformaldehyde and immunostained for ROK␣ expression using monoclonal antibody 1A1/fluorescein isothiocyanateconjugated anti-mouse IgG, and for p21 expression using rabbit anti-HA/rhodamine-anti-rabbit IgG. Tetramethylrhodamine ␤-isothiocyanate-phalloidin (Sigma; 0.5 g/ml) was used for staining filamentous actin. Confocal imaging was with a MRC600 system with a Zeiss microscope.

RESULTS
Using GTP-Cdc42 as a probe for expression screening, we previously isolated a novel brain Cdc42-binding tyrosine kinase (7). A similar approach was employed to identify cDNAs encoding proteins binding GTP-RhoA in a rat brain cDNA expression library. Six clones were obtained, which contained overlapping sequences, all of which encoded the RhoA binding domain (Fig.  1A). We then derived a small region (about 90 amino acids; rbf-8) responsible for the p21 binding (Fig. 1B). The sequence was dissimilar to the binding motifs for Cdc42/Rac1 (8). RhoA was only bound in its GTP form. Little or no Cdc42 was bound (Fig. 1C).
The complete sequence of a putative 159-kDa protein was obtained from analysis of the overlapping cDNAs ( Fig. 2A). It contained a novel serine/threonine kinase most closely related to the human myotonic dystrophy kinase (9, 10) (53% identity) and a p180 Cdc42-binding kinase (53% identity) recently identified in our laboratory, 2 as well as to the product of the fungal cot-1 gene (14) essential for hyphal elongation and the Drosophila "warts" gene (15) implicated in cell growth and morphology (Fig. 2B). The RhoA-binding kinase (termed ROK␣) also contains a cysteine/histidine-rich domain at the C terminus (Fig.  2C); this domain had some similarity to those of the PKC and chimaerin families (16), but the spacing of the invariant cysteines is not consistent with its being a diacylglycerol receptor. Between the kinase and RhoA binding domains, the sequence is predicted to assume an ␣-helical coil-coil structure (data not shown).
We have been unable to express full-length recombinant ROK␣ protein in Escherichia coli. However, the kinase domain could be expressed as a fusion protein with maltose-binding protein which was used to raise polyclonal antiserum. An immunoreactive 150-kDa protein was detected in all tissues, including brain as well as cultured rodent and human cells (Fig.  3A), which appeared to correspond to a ubiquitous p150 RhoAbinding protein previously described (8). The native 150-kDa RhoA-binding protein was then purified from rat brain cytosol by affinity chromatography with GST/RhoA fusion protein. The purified protein bound RhoA and was recognized by the antiserum to ROK␣ (Fig. 3A, lane E3). The p150 was capable of autophosphorylation and of phosphorylating myelin basic protein at serine and threonine residues (Fig. 3B). Although GTP-RhoA did not appear to stimulate p150 kinase activities in vitro, it is possible that the p150 kinase (ROK␣) becomes activated by affinity chromatography with RhoA during the purification and thus incapable of responding further to RhoA.
We then examined the effect of overexpressing RhoA on the intracellular distribution of the p150 ROK␣ in Cos-7 cells. About 20% of the cells were efficiently transfected. ROK␣ immunoreactivity, detected with a specific monoclonal antibody, was mainly present in the cytosolic fraction in control cells 2 T. Leung and L. Lim, unpublished observations. Membrane Association of RhoA-binding Kinase 29052 transfected with empty vector (Fig. 4A). Upon transfection with either RhoA or RhoA V14 , both of which are effective in promoting stress fiber formation when microinjected into cells (1), there was an increase in p150 ROK␣ in the pellet fractions, suggesting that RhoA promoted its association with membranes. This association was investigated at the cytological level using HeLa cells transfected with the dominant-positive RhoA V14 mutant. In HeLa and other cells examined, ROK␣ was distributed in the cytoplasm. RhoA V14 -transfected cells showed a generally rounded morphology with increased actin microfilaments mainly at the cell periphery (Fig. 4B, panel b), but also present in stress fibers adjacent to the substratum (Fig. 4B,  panel c). Much of the endogenous ROK␣ immunoreactivity co-localized with the actin microfilaments at the peripheral cell membrane (Fig. 4B, panel a), but not with stress fibers (data not shown). By contrast, in cells expressing Cdc42 V12 no such membrane translocation occurred with ROK␣ immunoreactivity being generally cytosolic (Fig. 4B, panel f). Neither was PKC␣ (analyzed as another control) membrane-associated in RhoA V14 -transfected cells (data not shown). We conclude that there is a specific recruitment of ROK␣ to plasma membrane with activated RhoA. DISCUSSION ROK␣ is a member of a serine/threonine kinase family, which includes the myotonic dystrophy kinase (9, 10) Neurospora cot-1, and the Drosophila warts gene (15), as well as a recently isolated Cdc42-binding kinase. 2 The myotonic dystrophy kinase may be involved in membrane functions related to ion channels (17,18) while cot-1 and warts mutants show abnormal cell growth with morphological consequences. It is intriguing that the related mammalian ROK␣ contains a RhoA binding domain. This domain is unrelated to the other known p21 binding domains and does not affect intrinsic or p190 RhoGAP-stimulated GTPase activities of RhoA (data not shown), unlike PAK or ACK, which inhibit the GTPase activi-FIG. 2. Sequence of a novel kinase containing a RhoA binding domain and other domains. A, predicted amino acid sequence of rat brain RhoA-binding protein. Sequences were obtained from both directions of two overlapping clones as well other cDNAs isolated by expression screening (rbf-1 to rbf-6). Restriction sites refer to those shown in Fig. 1. The N-terminal kinase, p21 binding (BD) and cysteine-rich (CR) domains are shown in bold. A diagrammatic representation of the predicted protein is also depicted. B, alignment of ROK␣ kinase domain and related kinases. ROK␣ was aligned with human myotonic dystrophic kinase (DMK, GenBank™ accession no. L08835), Neurospora cot-1 (N.Cot-1, accession no. P38679), Drosophila warts gene product (wts, accession no. L39837), and rat PKC␣ (accession no. P28867) using the CLUSTAL method (DNASTAR). C, alignment of cysteine/histidine-rich domains (CRD) of ROK␣, rat PKC␣, human Raf-1, and rat n-chimaerin. The invariant residues present in most CRDs of this class (16) are in bold and marked with asterisks. The position of His as a potential substitute for consensus Cys is marked (ϩ).

FIG. 3.
Expression of ROK␣ in various tissues, its purification from brain, and its kinase activities. A, RhoA binding and ROK␣ immunoreactivity of soluble extracts and purified ROK␣. Soluble protein (150 g) from rat brain (BR) and cultured cells (NB, human neuroblastoma; C6, rat C6 glioma; Gr, rat cerebellar granule cells; FB, mouse NIH3T3 fibroblasts) separated on a 9% polyacrylamide gel were probed with [␥-32 P]GTP-GST/RhoA (upper panel) or polyclonal antibodies raised against ROK␣ kinase domain (lower panel). The last lane is Coomassie staining of purified rat brain p150 isolated by RhoA-affinity chromatography (see "Materials and Methods"), which exhibits ROK␣ immunoreactivity and RhoA binding (lane E3). B, kinase activity of p150 ROK␣. Kinase assays were carried out in the absence (lanes 1 and  3) and presence of GTP␥S-RhoA (lanes 2 and 4). Assays in lanes 3 and 4 were performed with myelin basic protein (MBP) as exogenous substrate. For phosphoamino acid analysis, 32 P-phosphorylated p150 (lane 1) and myelin basic protein (lane 2) were subjected to analysis as described previously (8), using phosphoserine (PS), phosphothreonine (PT), and phosphotyrosine (PY) as standards.
Membrane Association of RhoA-binding Kinase 29053 ties of their p21 partners (7,8). RhoA, which shares 50% identity with Cdc42 and Rac1, binds to a different protein population than the latter p21 proteins, with p150 being ubiquitous and most prominent in all tissues (8). This p150 RhoAbinding kinase (putatively identified as ROK␣) is readily purified using a GST-RhoA affinity column. In relation to the in vivo effects of RhoA in promoting membrane association of ROK␣, we have found that HA tagged-ROK␣ is activated when co-transfected with RhoA on enzymatic analysis of HA-immunoprecipitates (data not shown). Most cells with membrane-associated ROK␣ showed a more rounded morphology, with the kinase being co-localized with peripheral actin microfilaments. No correlation of ROK␣ localization with stress fibers was observed, although the latter are increased in RhoA-overexpressing cells. This suggests that ROK␣ may not be directly involved in regulating or maintaining stress fibers. However, apart from increased stress fibers a frequent consequence of RhoA transfection is the occurrence of cells with rounded morphology. Filamentous actin reorganization is essential in the rounding up process and in the Rho-dependent formation of the contractile ring, which precedes cell division (19), and it is possible that ROK␣ participates in these events. In addition to morphology and cytokinesis (19,20), evidence is accumulating for the involvement of Rho proteins in motility (21), transformation (22,23), and apoptosis (24). Our finding of a RhoA-binding kinase should prove helpful in determining the mechanisms underlying these important cellular activities as well as the pathological basis of myotonic dystrophy because of the similarity of the dystrophic kinase to ROK␣.