Cloning and characterization of a Dictyostelium myosin I heavy chain kinase activated by Cdc42 and Rac.

The motile activities of the small, single-headed class I myosins (myosin I) from the lower eukaryotes Acanthamoeba and Dictyostelium are activated by phosphorylation of a single serine or threonine residue in the head domain of the heavy chain. Recently, we purified a myosin I heavy chain kinase (MIHCK) from Dictyostelium based on its ability to activate the Dictyostelium myosin ID isozyme (Lee, S. -F., and Côté, G. P. (1995) J. Biol. Chem. 270, 11776-11782). The complete sequence of the Dictyostelium MIHCK has now been determined, revealing a protein of 98 kDa that is composed of an amino-terminal domain rich in proline, glutamine, and serine, a putative Cdc42/Rac binding motif, and a carboxyl-terminal kinase catalytic domain. MIHCK shares significant sequence identity with the Saccharomyces cerevisiae Ste20p kinase and the mammalian p21-activated kinase. Gel overlay assays and affinity chromatography experiments showed that MIHCK interacted with GTPgammaS (guanosine 5'-3-O-(thiotriphosphate))-labeled Cdc42 and Rac1 but not RhoA. In the presence of GTPgammaS-Rac1 MIHCK autophosphorylation increased from 1 to 9 mol of phosphate/mol, and the rate of Dictyostelium myosin ID phosphorylation was stimulated 10-fold. MIHCK may therefore provide a direct link between Cdc42/Rac signaling pathways and motile processes driven by myosin I molecules.

The class I myosins (myosin I) are small, single-headed, non-filament-forming myosins present in organisms ranging from protozoans such as Acanthamoeba and Dictyostelium to mammalian cells (1)(2)(3). Myosin I molecules are comprised of a 110 -140-kDa heavy chain and one or more light chains, with the amino-terminal ϳ80 kDa of the heavy chain forming a motor domain, or head, that displays actin-activated Mg-ATPase activity and drives actin-based motile processes. The carboxyl-terminal portion of the myosin I heavy chain forms a globular tail domain that incorporates an acidic phospholipidbinding site and, in some isozymes, a nucleotide-insensitive actin filament-binding site, and a Src homology 3 (SH3) 1 do-main. Detailed biochemical studies on the Acanthamoeba myosin I isozymes have shown that phosphorylation of a single site in the head region of the heavy chain is required for these myosins to express maximal actin-activated MgATPase activity, to contract actin gels, and to move actin filaments (4 -6). Evidence is available to indicate that at least two of the five Dictyostelium myosin I isozymes (myosin IB and ID) are regulated in an analogous manner (7)(8)(9).
Recently, we isolated a Dictyostelium MIHCK based on its ability to activate the Dictyostelium myosin ID isozyme (9). We now report the complete cDNA sequence encoding the Dictyostelium MIHCK and show that it is a member of a protein kinase family that includes the Saccharomyces cerevisiae Ste20p and the mammalian p21-activated kinase (PAK). Members of the PAK kinase family bind to, and in some cases are activated by, Cdc42 and Rac (10 -12), two members of the Rho group of p21 Ras-related GTP-binding proteins (13). Evidence is presented to show that GTP-labeled Cdc42 and Rac bind to the Dictyostelium MIHCK and stimulate its activity.

EXPERIMENTAL PROCEDURES
Preparation of Proteins and Antibodies-Dictyostelium myosin ID and MIHCK were purified as described (8,9). MIHCK was digested with trypsin and peptides were isolated and sequenced as described (8). Human Cdc42, Rac1, and RhoA were expressed as glutathione S-transferase (GST) fusion proteins in bacteria (kindly provided by Dr. Alan Hall, Medical Research Council Laboratory for Molecular Cell Biology, London) and purified over glutathione-Sepharose (Pharmacia Biotech Inc.) (14). Antibodies were raised in a New Zealand White rabbit using MIHCK excised from a Coomassie Blue-stained SDS gel and were affinity-purified using Affi-Gel 15 (Bio-Rad) to which MIHCK was coupled (8). Immunoblot analysis of proteins transferred to Immobilon-P (Millipore) was performed as described (8).
Isolation and Sequencing of cDNA Clones-DNA manipulations were done according to standard procedures (15). A Dictyostelium gt11 library prepared from 4-h starved cells (Clonetech) was plated at a density of ϳ2.5 ϫ 10 4 plaques/150-mm LB agar plate, grown for 4 h at 37°C, and induced for 6 h using nitrocellulose filters (Schleicher & Schuell) impregnated with 1 mM isopropyl-␤-thiogalactosidase (Promega). The filters were probed with the affinity-purified anti-MIHCK polyclonal antibodies and positive clones detected using a goat antirabbit secondary antibody conjugated to alkaline phosphatase and colorimetric substrates (Bio-Rad). Positive phage clones were purified over a second round of screening, isolated from plate lysates using the Wizard DNA kit (Promega), and subcloned into pGEM7Z. Plasmid templates prepared using Qiagen columns (Qiagen Inc.) were sequenced on both strands using Sequenase (version 2.0, U. S. Biochemical Corp.) and/or an Applied Biosystems model 373A automated sequencer (Core Facility for Protein and DNA Chemistry, Queen's University, Kingston, Ontario, Canada). Two independent, overlapping clones were identified that contained sequences coding for MIHCK tryptic peptides. The Dictyostelium cDNA library was rescreened with these clones or with polymerase chain reaction-amplified DNA fragments prepared from the clones to obtain clones containing the entire MIHCK cDNA sequence (complete details of the cloning strategy are available upon request). Nucleotide sequences were analyzed and aligned to form the full-length DNA sequence using the DNASTAR program. Data bank searches were performed using the BLAST algo-* This work was supported by the Medical Research Council of Canada and by National Institutes of Health Grant GM50009. 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 /EBI Data Bank with accession number(s) U67715.  p21 Binding Assays-Cdc42, Rac1, and RhoA were exchanged with either GDP or GTP␥S as described (16). Overlay assays were performed by incubating Immobilon-P membranes containing MIHCK with p21 proteins (2 g/ml) labeled with [ 35 S]GTP␥S (specific activity: 50 Ci/mol) (Amersham Corp.) as described (16). Co-precipitation experiments were performed in 50 mM NaCl, 1 mM dithiothreitol, 2 mM MgCl 2 , 25 mM Tes, pH 7.0, using glutathione-Sepharose beads containing 0.5 mg/ml of either GDP-or GTP␥S-labeled Rac1 GST fusion protein. The glutathione beads were incubated with MIHCK (50 g/ml) for 30 min at room temperature, collected by centrifugation, washed extensively with the above buffer, and eluted with 5 mM glutathione in 50 mM Tris, pH 8.
Phosphorylation Assays-MIHCK was autophosphorylated at a con-centration of 30 g/ml by incubation in kinase buffer (2 mM MgCl 2 , 1 mM dithiothreitol, 0.25 mM ATP, 0.1 mg/ml bovine serum albumin, 20 mM Tes, pH 7.0) with or without 0.2 mg/ml GTP␥S-Rac1. [␥-32 P]ATP (Du-Pont NEN) at a specific activity of 1000 Ci/mol was included in assays where phosphate incorporation into MIHCK was directly measured, but was absent when MIHCK was used to phosphorylate myosin ID. Phosphorylation of myosin ID (90 g/ml) was performed in kinase buffer containing 1000 Ci/mol [␥-32 P]ATP with 0.6 g/ml MIHCK. Reactions were stopped at the indicated times by taking aliquots of 10 -20 l from the assays and immediately adding them to a one-fifth volume of boiling hot SDS sample buffer (9). Following SDS-gel electrophoresis the gel was stained with Coomassie Blue and dried for autoradiography. To quantify 32 P incorporation the appropriate protein bands were excised from the gel and counted in a scintillation counter.

RESULTS AND DISCUSSION
The affinity-purified anti-Dictyostelium MIHCK polyclonal antibody was found to recognize a single major band of ϳ110 kDa when used to probe crude Dictyostelium cell extracts and to react strongly with purified Dictyostelium MIHCK (Fig. 1A). An initial screen of a gt11 Dictyostelium cDNA expression library with the affinity-purified anti-MIHCK antibody yielded two overlapping cDNA clones, which were then used to reprobe the same cDNA library. Several additional overlapping cDNA clones were isolated that could be aligned to yield a single open reading frame of 2,685 nucleotides encoding a predicted polypeptide of 895 amino acids (Fig. 1B). The sequences of three peptides obtained from tryptic digests of MIHCK were all found within the deduced amino acid sequence (Fig. 1B). The predicted molecular mass of MIHCK (98 kDa) is in good agreement with the molecular mass of 110 kDa estimated from SDS gels (9) and is comparable with results obtained with Ste20p, a kinase related to MIHCK (see below). Ste20p has a molecular mass of 110 kDa on SDS gels and a predicted size of 102 kDa (17,18).
The deduced amino acid sequence of the Dictyostelium MI-HCK predicts a protein composed of an ϳ400-residue aminoterminal domain in which ϳ50% of the residues are proline, serine, or asparagine; a putative binding motif for Cdc42 and Rac; a linker region rich in proline, glutamine, and asparagine; and a carboxyl-terminal protein kinase catalytic domain (Fig.  1B). Data base searches using the program BLASTP revealed no significant similarity between the amino-terminal domain of MIHCK and other proteins; however, the presence of multiple proline-rich sequences containing the motif PXXP (where X is any amino acid; double underlined in Fig. 1B) suggests that this domain may mediate interactions with SH3 domain-containing proteins (19). Data base searches with residues 393-455 of MIHCK identified several protein kinases, including the S. cerevisiae Ste20p (17) and Cla4p (20) protein kinases and the mammalian PAK kinase (12), with homology to this region ( Fig. 2A). This region includes a 14 -16-residue motif that has been defined as the minimal sequence required for the GTPdependent binding of the Ras-related GTP-binding proteins Cdc42 and Rac (10). MIHCK retains only four of the eight consensus Cdc42/Rac binding motif residues ( Fig. 2A), but, nevertheless, specifically binds the GTP-bound forms of Cdc42 and Rac (see below).
The MIHCK Cdc42/Rac binding motif is separated from the kinase domain by a linker region that contains a 16-residue asparagine repeat and two potential SH3 domain-binding sites (Fig. 1B). The MIHCK kinase catalytic domain shares the conserved features of primary structure typical of protein Ser/ Thr kinases, including all residues that have been recognized as invariant or nearly invariant (21). A BLASTP search of the data base using the MIHCK catalytic domain as a query yielded the highest similarity scores with the kinase domains of Ste20p, Cla4p, PAK, and other members of this family (Fig.  2B). The catalytic domains of the yeast and human kinases are more similar to each other (52-68% sequence identity) than to MIHCK (42-50% sequence identity) (Fig. 2C), perhaps reflecting the phylogenetic position of Dictyostelium, which, based on protein data sets, diverged prior to the lineage leading to animals and fungi (22,23). A second subgroup of protein kinases have been described that are similar to Ste20p and PAK throughout the catalytic domain but do not contain a recognizable Cdc42/Rac binding site (24,25). The catalytic domains of these kinases, represented in the dendogram by the human MST1 kinase, are more divergent from Ste20p, PAK, and Cla4p than is MIHCK (Fig. 2C).
The ability of Dictyostelium MIHCK to interact directly with the Rho group of GTP-binding proteins was tested using an overlay assay. Human Cdc42, Rac1, and RhoA, expressed as GST fusion proteins, were labeled with the nonhydrolyzable GTP analog [ 35 S]GTP␥S and overlaid onto filters containing immobilized MIHCK. Autoradiograms indicated that MIHCK interacted with GTP␥S-labeled Cdc42 and Rac1, but not with RhoA (Fig. 3A). Rac1, which consistently exhibited the highest levels of binding to immobilized MIHCK, was examined for its interaction with native, soluble MIHCK using a glutathione-Sepharose bead co-precipitation assay. No significant co-pre-cipitation of MIHCK was detected with beads containing either GTP␥Sor GDP-labeled Rac1 (Fig. 3B). Studies were then performed on MIHCK that had been activated by autophosphorylation. Autophosphorylated MIHCK (P-MIHCK) contains 1 mol of phosphate/mol and exhibits a 40-fold greater rate of myosin ID phosphorylation than unphosphorylated MIHCK (9). P-MIHCK co-precipitated with GTP␥S-Rac1, but not GDP-Rac1 (Fig. 3B). The results suggest that autophosphorylation promotes the interaction of the soluble, native MIHCK with GTP␥S-Rac1, although it does not seem to be required for the denatured, immobilized MIHCK to bind Rac1. In studies using immobilized p65 PAK , autophosphorylation was shown to decrease the affinity for GTP-Rac1 (12).
GTP-Rac1 and GTP-Cdc42 stimulate the autophosphorylation of PAK and increase its activity toward exogenous substrates (12,26). Similar effects of Cdc42 (data not shown) and Rac1 were observed with MIHCK. The amount of phosphate incorporated into MIHCK via autophosphorylation increased dramatically in the presence of GTP␥S-Rac1 (from 1 to 9 mol of phosphate/mol), resulting in a decreased mobility of the kinase on SDS gels (Fig. 4A-C). MIHCK autophosphorylated to 9 mol/mol in the presence of GTP␥S-Rac1 displayed an activity 10-fold higher than MIHCK autophosphorylated in the absence of GTP␥S-Rac1 when assayed both with synthetic peptides (9) (data not shown) and myosin ID as substrates (Fig. 4D). It seems that the spontaneous autophosphorylation of MIHCK to 1 mol of phosphate/mol is sufficient to induce only partial kinase activity; subsequent interaction with Cdc42 or Rac1 is required for further autophosphorylation and expression of full kinase activity.
Multiple members of the Rho family of GTP-binding proteins have been identified in Dictyostelium and one, RacE, is specifically required for cytokinesis (27,28). In mammalian cells Cdc42 and Rac stimulate the polymerization of actin, affect the organization of the actin cytoskeleton, and promote a variety of actin-based motile processes such as membrane ruffling, filopodia extension, pinocytosis, and invasive growth (13, 29 -31). In yeast Cdc42 has a role in controlling actin filament distribution and polarized cell growth (32,33). The results reported here identify the Dictyostelium MIHCK as a target for Cdc42 and Rac and suggest that in Dictyostelium one consequence of activating Rac-related GTP-binding proteins is to stimulate motile processes driven by the single-headed myosin I isozymes. Myosin I molecules in a variety of cell types have been localized to regions of the actin-rich cortex undergoing active movement, such as membrane ruffles, leading lamella, phagocytic cups, and neuronal growth cones (34 -37), and have been implicated, largely through the analysis of knockout mutations in lower eukaryotes, in cellular processes such as endocytosis, phagocytosis, pinocytosis, secretion, pseudopod extension, and polarized cell growth (38 -42). Such studies suggest that the myosin I molecules play a central role in actin-dependent membranebased motile processes and thus represent attractive candidates to mediate some of the effects of Cdc42 and Rac.
The substrate specificity and regulatory properties of the Dictyostelium MIHCK are remarkably similar to those of the 97-kDa Acanthamoeba MIHCK (3). Both MIHCKs can activate Dictyostelium myosin I isozymes and require autophosphorylation for activity (7,9,43). In both cases autophosphorylation is accelerated by acidic phospholipids and the enhancement of the rate of autophosphorylation by acidic phospholipids is completely inhibited by Ca 2ϩ /calmodulin (43)(44)(45). Recent partial sequence analysis of the Acanthamoeba MIHCK reveals that, like Dictyostelium MIHCK, it contains a catalytic domain related to Ste20p and PAK (50). This raises the question as to whether Ste20p and PAK, which share considerable sequence identity to the Dictyostelium MIHCK, both in the kinase and Cdc42/Rac binding domain, may function as myosin I-activating kinases. At present there is little direct evidence to show that myosin I molecules in organisms other than Acan-thamoeba and Dictyostelium require phosphorylation of the heavy chain for activity, and indeed, sequence alignments indicate that the serine/threonine residue mapped as the site of heavy chain phosphorylation in the head domain of the Acanthamoeba myosin I isozymes (46) is replaced in the large majority of myosins, including all the known mammalian myosins I, by a glutamate or aspartate residue, which, it has been argued, may relieve the requirement for phosphorylation (47). Myosins that retain the regulatory serine/threonine residue in the head domain and so are the most likely to require heavy chain phosphorylation for enzymatic and mechanochemical activity include the S. cerevisiae and Aspergillus class I myosins and the Drosophila and mammalian class VI myosins (47). Interestingly, PAK and the Acanthamoeba MIHCK can activate smooth muscle myosin II by phosphorylating a site on the light chain, raising the intriguing possibility that members of this kinase family may coordinate the regulation of more than one class of myosins (48,49).

FIG. 4. Effect of GTP␥S-Rac1 on MIHCK autophosphorylation and activity.
A-C, MIHCK was autophosphorylated with (ϩ) or without (Ϫ) GTP␥S-Rac1 in the presence of [␥-32 P]ATP as described under "Experimental Procedures." Aliquots were taken at the times indicated and subjected to SDS-gel electrophoresis. A shows the Coomassie Bluestained gel, B the corresponding autoradiogram, C the amount of 32 P incorporated into the MIHCK band determined by excising the MIHCK band from the SDS gel and counting it in a scintillation counter. D, Dictyostelium myosin ID was phosphorylated with MIHCK autophosphorylated in the presence (ϩ) or absence (Ϫ) of GTP␥S-Rac1. Samples were taken at the times indicated, subjected to SDS-gel electrophoresis and autoradiography.