Characterization of RhoA-binding Kinase ROKα Implication of the Pleckstrin Homology Domain in ROKα Function Using Region-specific Antibodies*

Rho-binding kinase α (ROKα) is a serine/threonine kinase with multiple functional domains involved in actomyosin assembly. It has previously been documented that the C terminus part of ROKα interacts with the N-terminal kinase domain and thereby regulates its catalytic activity. Here we used antibodies against different domains of ROKα and were able to reveal some structural aspects that are essential for the specific functions of ROKα. Antibodies against the kinase domain revealed that this part of the protein is highly complex and inaccessible. Further experiments confirmed that this domain could undergo inter- and intramolecular interactions in a complex manner, which regulates the kinase catalytic activity. Other antibodies that raised against the coiled-coil domain, Rho binding domain, and the pleckstrin homology (PH) domain were all effective in recognizing the native proteins in an immunoprecipitation assay. Only the anti-Rho binding domain antibodies could activate the kinase independent of RhoA. The PH antibodies had no apparent effects on the catalytic activity but were effective in blocking actomyosin assembly and cell contractility. Likewise, mutations of the PH domains can abrogate its dominant negative effects on actin morphology. The subsequent disruption of endogenous ROK localization to the actomyosin network by overexpressing the PH domain is supportive of a role of the PH domain of ROK in targeting the kinase to these structures.

ROK␣ belongs to a member of a kinase family that includes myotonic dystrophy kinase, myotonic dystrophy kinase-related Cdc42-binding kinase, and citron kinase. In general, they consist of an N-terminal serine/threonine kinase domain that is followed immediately by an extended coiled-coil region and other functional motifs such as GTPase binding, pleckstrin homology (PH), and cysteine-rich domains. In ROK, the C terminus contains an unconventional PH with an internal cysteine-rich motif (6). These multidomain kinases have been reported to be involved in the regulation of some aspects of actin cytoskeleton rearrangement during different cell stages and cytokinesis through their conserved catalytic activities (11)(12)(13)(14).
The regulation of the catalytic activity of serine/threonine protein kinases often involves kinase phosphorylation in the activation loop and the hydrophobic motif C-terminal to the kinase domain by autophosphorylation and/or phosphorylation by a heterologous kinase (15)(16)(17). Our study on myotonic dystrophy kinase-related Cdc42-binding kinase-␣ has revealed the requirement of a trans-autophosphorylation event for kinase activation; the process is facilitated by kinase dimerization (18). This dimerization/autophosphorylation may be a special feature of this kinase family, in contrast to some kinases such as the AGC protein kinases (Ser/Thr kinases including protein kinase A, B, C, and G), which often require an upstream kinase for activation. Interestingly, a strong correlation of oligomerization with activity has also been described for myotonic dystrophy protein kinase (19). Whether or not this constitutes a fundamental activation mechanism for all members in the family remains to be investigated.
Besides sharing sequence homology in the kinase domains, protein kinases of this family also show similarity in their overall domain arrangements as well as regulation. Both native myotonic dystrophy kinase-related Cdc42-binding kinase and myotonic dystrophy protein kinase have recently been shown to separately exist in multimeric complexes, possibly through parallel intermolecular interaction of their extended coiled-coil domains (18,19). These findings suggest that ROK could also potentially oligomerize through its coiled-coil domain in a similar fashion. In addition, the kinase domains of both myotonic dystrophy kinase-related Cdc42-binding kinase and myotonic dystrophy protein kinase have been shown to be autoinhibited during a resting state by an autoregulatory region C terminus to the kinase domains (18,19). A similar observation was also reported for ROK␣, in which the C terminus including the PH domain has been mapped to be the autoinhibitory region (20). Binding of the GTP-bound form of Rho is known to activate ROK; the interaction is believed to disrupt the negative regulatory interaction between the kinase domain and the C-terminal autoinhibitory region to give rise to an active kinase. Although all results have pointed to the C-terminal end of the coiled-coil domain as the Rho-binding site (6,(21)(22)(23), the exact sequence of elements involved and nature of the interaction remain largely obscured.
In this current work, we further analyzed the structure and function relationship of ROK. In agreement with the related protein kinases, we found that native ROK exists in multimeric complexes, and its catalytic activity is dependent on a dimerization/trans-autophosphorylation event. To understand ROK better, we have raised specific antibodies against the various functional domains of ROK as tools to probe its functions in vivo. We show that the antibody against the Rho-binding site could result in ROK activation independent of RhoA. Conversely, the antibody against the C-terminal PH domain produced an inhibitory effect on ROK activity on cell contractility and actomyosin assembly. Combined evidence from immunoprecipitations and mutagenesis studies strongly suggests that the C-terminal PH domain of ROK has a yet undiscovered role on cytoskeleton rearrangement besides its autoinhibitory regulation on the catalytic domain.
Expression and Purification of Recombinant Proteins and Antibody Purification-Recombinant maltose-binding protein-ROK␣-KIN 107-392 (KIN), GST-ROK␣-CC 639 -968 (CC), GST-ROK␣ 1008/1139 (dBD), GST-ROK␣ 1142-1379 (PH), and GST-ROK␤ 1008 -1039 (pBD) were obtained as fusion proteins according to standard protocol. For the preparation of polyclonal antibodies, thrombin-cleaved protein from each GST fusion protein or factor X-cleaved protein from maltose-binding protein fusion protein (400 g) was emulsified in complete Freund's adjuvant for injection into rabbits. Rabbits were bled 10 days after the third and subsequent booster injections. Sera were collected and affinity-purified using 2 mg/ml antigen pre-coupled to cyanogen bromide-activated Sepharose (Sigma) and eluted with buffer containing 100 mM glycine-HCl (pH 2.5), 0.05% Triton X-100. The first 2 eluates were neutralized with Tris/HCl (pH 8.5) and used at a 1:1000 dilution for Western blot analysis. Monoclonal antibody 1A1 against the rat ROK␣ kinase domain has been described previously (24). For cell injection experiments, Triton-free buffer was used for elution, and the antibody concentrations were adjusted to about 0.5-1 mg/ml before use.
Extract Preparation and [␥-32 P]GTP-RhoA Binding-Brain tissue extracts were prepared from fresh tissue by using Dounce homogenizer in lysis buffer containing 25 mM Hepes, pH 7.3, 0.2 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 20 mM ␤-glycerol phosphate, 1 mM sodium orthovanadate, 0.5% Triton X-100, and 5% glycerol with protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride and 1 g/ml each of aprotinin, pepstatin, and leupeptin; Sigma). Extracts were clarified by centrifugation at 30,000 rpm for 30 min at 4°C. Likewise, HeLa cells extracts were obtained from subconfluent cultured cells (90-mm dish) with cells grown either in normal minimum Eagle's medium (MEM) (for direct immunoprecipitation) or methionine-free MEM medium (Sigma) supplemented with [ 35 S]methionine (0.2 mCi/ml; PerkinElmer Life Sciences). Cells were harvested after a 2-h incubation, and sample preparation in lysis buffer and immunoprecipitations were carried out as described before.
Bissulfosuccinimidyl Suberate Cross-linking Assay-Untransfected COS-7 cells or cells overexpressing wild-type ROK␣ were extracted in lysis buffer. Protein concentrations of clarified extracts were adjusted to 1-3 mg/ml. Cross-linking reactions began with addition of 0.05 to 0.5 mM bissulfosuccinimidyl suberate (Pierce) at 4°C for 30 min. Reactions were stopped by the addition of an equal volume of SDS-PAGE sample buffer. Cross-linked products were detected by immunoblotting with either anti-ROK␣ antibody (1A1) for detecting the endogenous ROK␣ or anti-FLAG antibody for overexpressed FLAG-tagged ROK␣ proteins.
Immunoprecipitation and Kinase Assays-COS-7 cells grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum were transfected with different ROK␣ constructs using LipofectAMINE (Invitrogen). 16 h after transfection, cell extracts were obtained with lysis buffer (25 mM HEPES, pH 7.7, 0.15 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 1 mM sodium vanadate, 20 mM glycerol phosphate, 5% glycerol, and 0.2% Triton X-100). Clarified cell extracts were incubated with anti-FLAG antibody-conjugated agarose beads (Sigma) or anti-HA antibody and protein A-conjugated Sepharose beads (Sigma) for 2 h at 4°C. After extensive washing, the immunoprecipitates were either continued with kinase assays or separated by SDS-PAGE for Western blot analysis with anti-HA or anti-FLAG antibody. Kinase assays were carried out at 30°C for 10 min using GST-MLC2 as substrates (6) in buffer containing 10 Ci of [␥-33 P]ATP, 10 M ATP, 25 mM Hepes (pH 7.3), 25 mM KCl, 5 mM ␤-glycerol phosphate, 2.5 mM sodium fluoride, 5 mM MgCl 2 , 1 mM MnCl 2 , and 0.025% Triton X-100. Total soluble extracts from rat brain and COS-7 cells were used for Western blotting using mouse monoclonal antibody 1A1 or rabbit anti-ROK␣ antibodies raised against the different domains of ROK␣. For immunoprecipitation, rat brain soluble extract (50 mg) was incubated with antibodycoupled Sepharose beads for 30 min. Immunoprecipitated products after extensive washing were resolved on 7.5% PAGE-SDS and immunoblotted with 1A1 antibody.
Microinjection, Cell Staining, and Immunofluorescence Microscopy-HeLa cells were maintained and transfected as described (6). For mi-croinjection of HeLa and Swiss 3T3 cells, subconfluent cells were plated on coverslips for 48 h before microinjecting with the different ROK␣ constructs (50 ng/l) or anti-ROK␣ antibodies (0.5-1 g/l). C3 exoenzyme (0.2 g/l; Upstate Biotechnology, Inc.) was co-injected in some experiments. 2-4 h after injection, cells were fixed with 4% paraformaldehyde and stained with a combination of various primary antibodies (anti-HA (12CA5; Roche Molecular Biochemicals), anti-FLAG (M2; Sigma), fluorescein isothiocyanate (FITC) anti-mouse, or FITC anti-rabbit antibodies. Stained cells were analyzed with a Bio-Rad Radiance 2000 Confocal Imager adapted to a Nikon microscope. Other antibodies used for cell staining were obtained from various commercial sources (mouse anti-myosin light chain (Sigma), rabbit anti-myosin heavy chain (Biomedical Technologies Inc.), and mouse anti-vinculin (Sigma).
Sphingosine 1-Phosphate Treatment and Cell Contractility Assay-Subconfluent HeLa cells were serum-starved for 6 h before injecting with the various antibodies. After incubating for 30 min, cells were treated with 0.3 g/ml sphingosine 1-phosphate (Sigma). Time lapse phase-contrast microscopy of the injected and non-injected cells was followed immediately after treatment. Images at 0 time and 15 min were taken for comparison.

Specificity of Various Anti-ROK Antibodies in
Recognizing the Native Form of ROK-To estimate the endogenous ROK activities in cells and tissues, we have attempted to raise antibodies against the various regions of ROK␣ for immunoprecipitation assays. An obvious phenomenon from these experiments was the inability of the antibodies against the kinase domain (both monoclonal antibody 1A1 and polyclonal antibodies KIN-Ab) in recognizing the native endogenous full-length protein in rat brain extracts (Fig. 1A). Antibodies against the coiled-coil domain of ROK␣ (CC-Ab), distal RhoA binding domain (BD-Ab), and the PH domain (PH-Ab) were all effective in immunoprecipitating the endogenous ROK (Fig. 1, A and C).
To further investigate the specificity of these antibodies, immunoprecipitations using [ 35 S]methionine-labeled HeLa cell extracts were carried out. Here we showed that except for the kinase antibodies (KIN-Ab), other antibodies (CC-Ab, dBD-Ab, and PH-Ab) could specifically immunoprecipitate both the native and metabolically labeled ROK protein (Fig. 1B).
To see if the anti-kinase domain antibodies are effective in recognizing the kinase domain alone, we used the antibodies to immunoprecipitate the kinase domain overexpressed in transfected COS-7 cells. As shown in Fig. 2A, the overexpressed kinase domain protein was readily immunoprecipitated by the kinase antibody 1A1. Furthermore, the antibody apparently inhibited the catalytic activity of the immunoprecipitated protein. This inhibitory effect could also be observed when 1A1 antibody was co-injected with ROK catalytic domain construct in HeLa cells (Fig. 2B). Injection of this antibody alone into HeLa cells resulted in no obvious effect on the overall actin morphology (Fig. 2B, a and b). However, it was able to almost completely block the aberrant actin filament formation induced by the active ROK-CAT when co-injected (Fig. 2B, c and d), indicating that the antibody had an inhibitory effect against ROK in vivo. These results suggest that the failure of this antibody in recognizing the native full-length ROK protein could be due to inaccessibility of the kinase antibodies to the epitopes.
Native ROK Exists in High Molecular Weight Complexes-It is therefore of great interest to find out why the kinase domain in the full-length protein is inaccessible to the antibodies. We found that full-length ROK␣ can exist in high molecular weight multimeric forms. As shown in Fig. 3A, endogenous ROK␣ co-migrated with high molecular weight markers in a gel filtration analysis, and no monomeric form of the protein was detectable. A molecular size of about 600 kDa is suggestive of a tetrameric structure. Likewise, chemical cross-linking experiments with bissulfosuccinimidyl suberate also revealed a slow migrating band, confirming that ROK can exist as multimeric complexes (Fig. 3B).

Intra-and Intermolecular Interactions
Regulate ROK␣ Catalytic Activities-To define which region(s) of ROK␣ is involved in the oligomerization event, we have examined the various regions of ROK␣ that may have contributed to this event. Both the proximal and distal coiled-coil regions (including the Rho binding domain, which is located at the end of the extended coil regions) were effective in forming oligomers (Fig. 4B). Surprisingly, the kinase domain alone can also form homophilic dimers. In contrast, the PH domain at the C terminus was totally ineffective in dimer formation.
Apart from the intermolecular interaction, we also attempted to determine if intramolecular interactions may also take place. Indeed, the kinase domain can also interact with two independent regions, one at the distal coiled-coil region and an additional interaction at the C-terminal PH domain (Fig.  4C). Hence, our results have revealed that ROK␣ can form complexes through intermolecular interactions through its coiled-coil and kinase domains and that the N-terminal kinase domain can also interact intramolecularly with the C-terminal coiled-coil and PH domains. To see if these interactions are functional, we co-expressed the FLAG-tagged kinase domain with the various HA-tagged interacting partners to check for their associations and effects on the kinase domain activity. As shown in Fig. 4D, both the distal coiled-coil and PH domains can partially block the catalytic activities of the co-precipitated kinase domain. Interestingly, dimerization of the wild-type kinase with the kinase-dead ROK␣ kinase domain also resulted in an inactive kinase, suggesting that trans-autophosphorylation plays an essential role in regulating the kinase activity.
Antibodies to the Distal Rho Binding Domain Block RhoA Binding and Activate ROK Independent of RhoA-Previous work on ROK isoforms reveal that the Rho binding domain is located at the end of the coiled-coil region (6,(21)(22)(23) and that binding of RhoA results in kinase activation (7). Because the antibodies are specifically raised against different domains, we have attempted to determine if binding of these antibodies to the RhoA binding domain of ROK competes with RhoA binding.
As expected, only antibodies against the distal binding domain (dBD-Ab in Fig. 5A) are effective in blocking RhoA binding in vitro. Interestingly, full-length ROK␣ immunoprecipitated with this antibody showed significant increases in the catalytic activity in comparison with ROK protein pulled down by other antibodies (Fig. 5B). This suggested to us that the dBD-Ab not only blocked RhoA binding but also activated ROK␣ catalytic activity.
To further examine the effectiveness of these antibodies in ROK activation in vivo, we have microinjected these antibodies into HeLa cells and checked the effects on actin morphology. As shown in Fig. 5C, cells injected with dBD-Ab gave significant increases in actin stress fibers. These actin filaments were not sensitive to C3 toxin, which is a potent inhibitor of RhoA but is readily disassembled in the presence of the kinase inhibitor HA1077. These results give further support of the in vitro data that the antibodies to the Rho binding domain can activate ROK catalytic activity independent of Rho.
Antibodies against the C-terminal PH Domain Block Cell Contractility and Myosin Assembly-Thus far our results and others clearly show that the C terminus of ROK␣ can form part of intramolecular interaction with the kinase domain and this can negatively regulate the kinase activity ( Fig. 4; Ref. 20). However, although antibodies against the Rho binding domain

FIG. 3. Gel filtration chromatography and chemical cross-linking reveal the multimeric nature of ROK␣.
A, rat brain extract was applied onto a Sepharose CL-6B-200 gel filtration column. Fractions were collected, and the relative amounts of ROK in each fraction were determined by Western blotting with antibody 1A1 and quantified with a Bio-Rad laser densitometer. The elution profile for the standard molecular weight markers are indicated by arrowheads. B, soluble rat brain extract were exposed to increasing concentrations of the crosslinking agent bissulfosuccinimidyl suberate (BS 3 , from 0.05 to 0.5 mM). Cross-linked endogenous ROK␣ was determined by immunoblotting with antibody 1A1. The arrow indicates the monomeric ROK␣. are effective in activating ROK␣ catalytic activity, the antibodies against the C terminus are apparently ineffective. A consistent observation was the lack of contractility of cells injected with the PH-Ab, in contrast to those injected with dBD-Ab (Fig.  6A). Cell contractility was also observed when HeLa cells were treated with Rho-activating agents such as lysophosphatidic acid or sphingosine 1-phosphate (SPP; Fig. 6A), the responsiveness to these agents was lost when cells were injected with PH-Ab (Fig. 6A). To further characterize this phenotypic effect upon PH-Ab, we have stained injected cells for myosin light chain, polymerized actin filaments, and focal adhesions. As shown in Fig. 6B, cells injected with PH-Ab exhibited a prominent defect on myosin assembly with moderate changes in polymerized actin arrangement (Fig. 6B, a and b). By contrast, cells injected with dBD-Ab showed marked enhancement in myosin staining, whereas CC-Ab was relatively ineffective (Fig.  6B, c and d). PH-Ab-injected cells also showed mild but consistent decreases in focal adhesion staining (Fig. 6C, a and b) in and PH domain (ROK␣-PH) can independently interact and inhibit kinase domain catalytic activity. COS-7 cells were transfected with FLAG-ROK␣-CAT construct alone or co-transfected this construct with either HA-tagged ROK␣-pCC, ROK␣-dCC, ROK␣-BD, ROK␣-PH, or ROK␣-CT construct. Immunoprecipitations were carried out using anti-HA antibodies, and the IP products were immunoblotted with anti-FLAG or anti-HA antibody as described in B. The immunoprecipitation products were subjected to kinase assays using GST-MLC2 as substrate, and phosphorimaging was used to quantify the radiolabeled substrate on SDS-PAGE. The means and S.E. from three experiments were calculated as a percentage of the immunoprecipitated FLAG-ROK␣-CAT activity and are shown in D.
contrast to cells injected with dBD-Ab, which showed marked increases (c and d), and to those with CC-Ab, which remained unchanged (c-f). Thus, PH-Ab apparently exerted a potent effect on myosin assembly with relatively milder effects on actin and focal adhesions. These cytoskeletal effects of the PH-Ab may account for the observed lack of contractility in PH-Ab-injected cells upon sphingosine 1-phosphate treatment.
PH Domain of ROK␣ Has Multiple Functional Roles-The observation of an inhibitory effect of the PH-Ab has prompted us to investigate if the C-terminal PH domain of ROK␣ may have other functional roles apart from its intramolecular kinase inhibition. A mutant with two conserved tryptophan residues (ROK␣-CT AL ; Refs. 25 and 26) was used to examine its effects on kinase inhibition. As shown in Fig. 7A, this mutant was as effective as the wild-type protein in interacting and inhibiting the kinase domain catalytic activity (Fig. 7A). However, the dominant negative effect of this mutant protein on actin cytoskeleton is lost (Fig. 7B), suggesting that the PH domain may have a role that is unrelated to the intramolecular kinase inhibition. Because both PH-Ab and overexpression of minimal PH domain apparently had similar cytoskeletal effects, we attempted to examine the effect of the PH domain on the localization of the endogenous ROK␣ protein in Swiss 3T3 fibroblasts using the CC-Ab. As shown in Fig. 7C, interphase fibroblast showed endogenous ROK co-localized with myosin filaments (a and b), in agreement with recent data showing co-localization of Rho regulatory proteins in these structures (27). When overexpressing the PH domain, the distribution of endogenous ROK was more diffused, and myosin arrangement was in a random fashion (c and d). Expression of the mutant PH domain (ROK␣-PH AL ) apparently had no such effect, and endogenous ROK showed the normal close alignment with assembled myosin. Similar effects were observed with a PH domain mutant at the cysteine-rich domain region within the PH domain (C1284S/C1287S; data not shown). We therefore conclude that an intact PH domain is required for the blockage of endogenous ROK protein to localize to the site of myosin assembly. DISCUSSION We have described in this report that antibodies against the kinase domain of ROK␣ can recognize the kinase domain alone but not the full-length protein. Further experiments have indicated that the kinase can exist in multimeric form, as the result of extensive intermolecular interactions of the various coiled-coil domains. A tetrameric structure can be predicted based on the molecular size of about 600 kDa on gel filtration chromatography and chemical cross-linking. Both the proximal and distal coiled-coil domains can form independent homophilic dimers. Furthermore, the C terminus can also interact intramolecularly with the kinase domain. These results are similar to an earlier report (20), but our data suggest that a major interaction also include the distal coiled-coil region that overlaps with the RhoA binding domain, although the C-terminal PH domain may also contribute to the overall effect. Interestingly, the kinase domain is also able to form homodimers. The myotonic dystrophy kinase family including ROK␣ consists of a conserved activation loop and an extended C-terminal hydrophobic loop that are known for catalytic activation (15)(16)(17). In some cases the phosphorylation within these two regions can be achieved by other kinases or through an autocatalytic event. Our results of an inactive ROK␣ kinase when an active kinase domain dimerized with a kinase-dead counterpart may suggest that a trans-autophosphorylation event upon kinase dimerization is essential for kinase activation. The observation that autophosphorylation site mutants ROK␣ T240A in the activation loop region and ROK␣ T405A in the extended hydrophobic region that exhibited significant losses of catalytic activity also provides further support of this regulatory event (data not shown). Furthermore, co-expression of wild-type ROK␣ with PDK1 (both active and kinase-dead forms) did not result in alterations in ROK␣ catalytic activity, implying that ROK␣ is not regulated by this kinase (data not shown). It is tempting to speculate that the dimerization and autophosphorylation of ROK␣ may well be a major event for the regulation of these kinases. Our recent data on the related Cdc42-binding kinase myotonic dystrophy kinase-related Cdc42-binding kinase-␣ has also provided strong evidence for this notion. In this case the dimerization and activation of the kinase domain (which depends on the conserved N-terminal extended region) and the inhibitory interaction between the kinase domain and the distal coiled-coil domain are mutually exclusive and crucial for regulating the catalytic activity (18). A similar conserved mechanism may also be involved in the dimerization of the kinase domain of ROK␣ and its distal coiled-coil domain, with an additional requirement that a larger C-terminal region is required for effective interaction and kinase inhibition (20). We therefore conclude that ROK␣ can form multimeric complexes essentially with the central coiled-coil region. The homophilic dimer formation of the kinase domain is crucial for the catalytic activity, and this can be abrogated by interaction with the C-terminal region, in particular the distal coiled-coil region, which overlaps with the Rho-A binding domain and forms an essential part of the autoinhibitory motif.
In this respect, it is intriguing to find that antibodies to the Rho binding domain can activate ROK kinase activity. It is conceivable that the interaction of the antibody with an exposed surface at this region close to the Rho binding and the overlapping kinase inhibitory interaction region is sufficient to cause a conformational change leading to activation of the kinase in a Rho-independent manner. We have previously mapped the Rho binding domain of ROK␣ to residues 968 -1047 (6), and more recent work (21)(22)(23) has documented a more likely binding site at a nearby N-terminal region. Antibodies directed specifically to the Rho-binding site of ROK␤ were ineffective in recognizing the protein, indicating this region may not be exposed and, hence, inaccessible to the antibodies (see Fig. 1). The requirement of antibodies against an extended region beyond the Rho binding domain for the recognition and activation of ROK␣ may imply multiple binding sites for Rho. A weaker site at the C-terminal side of the Rho binding region may be necessary for opening up the distal coiled-coil region, where a more pertinent Rho binding can occur at a subsequent stage. It is likely that such perturbation may be crucial for releasing the inhibitory effect of this region upon the kinase domain, whose subsequent dimerization described earlier is essential for kinase activation.
Although the PH domain can also form part of the autoinhibitory domain, it is not known whether it can serve other functional roles because PH domains are known to interact with phospholipid and membrane components (28). Interestingly, the PH domain of ROK␣ is unique, consisting of an internal cysteine-rich motif. Previous reports indicate that this motif alone can block stress fibers and focal adhesion formation, probably through the autoinhibitory effect on the kinase catalytic activity (6,20). However, our observation that the anti-kinase antibodies are unable to recognize the full-length protein may suggest that overexpression of the PH domain may not have major effects on the full-length protein because this could only be effective intramolecularly. Support of this view comes from the microinjection experiments with the PH antibodies that resulted in defective cell contractility and myosin assembly. Furthermore, PH mutants at the PH/cysteine-rich domain regions are significantly less effectively in blocking actin stress fiber and focal adhesion formation in comparison with the wild-type PH domain, although they still retained the ability to interact with the kinase domain protein and inhibited the catalytic activity. This provides strong evidence that this region of ROK␣ may have an additional functional role. The observation of a blockage of translocation of endogenous ROK to myosin assembly site may imply that this region of ROK␣ may play a role in kinase targeting to and subsequent assembling myosin.
In summary, we have raised antibodies to the various regions of ROK␣. Antibodies against the kinase domain have revealed that this part of the kinase is cryptic. This phenomenon is probably the result of extensive inter-and intramolecular interactions of the different domains that also form part of the kinase activation mechanism. Antibodies to the Rho binding domain can essentially replace Rho in terms of kinase activation, and antibodies to the C-terminal PH domain also revealed that this part of the kinase is essential for targeting of the kinase to actomyosin compartment. These antibodies are therefore useful in defining functions to the various structural domains of the protein.