Characterization and Cloning of a DictyosteliumSte20-like Protein Kinase That Phosphorylates the Actin-binding Protein Severin*

After receiving an external stimulusDictyostelium amoebae are able to rearrange their actin cytoskeleton within seconds, and phosphorylation is a prime candidate for quick modification of cytoskeletal components. We isolated a kinase from cytosolic extracts that specifically phosphorylated severin, a Ca2+-dependent F-actin fragmenting protein. In gel filtration chromatography severin kinase eluted with a molecular mass of about 300 kDa and contained a 62-kDa component whose autophosphorylation caused a mobility shift in SDS-polyacrylamide gel electrophoresis and stimulated phosphorylation of severin. Severin kinase activity could be specifically precipitated with antibodies raised against the 62-kDa polypeptide. Phosphorylation of severin was strongly reduced in the presence of Ca2+, indicating additional regulation at the substrate level. Peptide sequencing and cloning of the cDNA demonstrated that the 62-kDa protein belongs to the Ste20p- or p21-activated protein kinase family. It is most closely related to the germinal center kinase subfamily with its N-terminal positioned catalytic domain followed by a presumptive regulatory domain at the C terminus. The presence of a Ste20-like severin kinase inDictyostelium suggests a direct signal transduction from the plasma membrane to the cytoskeleton by phosphorylation of actin-binding proteins.

The dynamic rearrangements of the actin cytoskeleton in motile cells are mainly regulated by actin-binding proteins which either interfere directly with the polymerization kinetics of actin or alter the viscoelasticity of the filamentous network (for reviews, see Refs. [1][2][3]. Severin from Dictyostelium discoideum, a model for amoeboid cell motility, belongs to the class of F-actin fragmenting and capping proteins whose members are structurally and functionally related (4). This class includes among others the vertebrate proteins gelsolin (5), villin (6), gCap39 (7), or from Physarum polycephalum the protein fragmin (8). F-actin fragmenting proteins are especially well suited for causing quick rearrangements in the filamentous actin network. At micromolar Ca 2ϩ levels they sever actin filaments by rupturing the noncovalent bonds between actin subunits in a filament. This leads to a rapid increase of short filaments together with a dramatic decrease in viscosity. After having severed the actin filaments, the proteins remain bound at the barbed end of the filaments and thereby prevent filament elongation. It is assumed that this results in solation of the viscous cytoplasm with a large number of short but capped filaments. For several members of this family it has been shown in vitro that uncapping is caused by polyphosphoinositides. In vivo this could then lead to free barbed ends ready for rapid elongation (9).
There is increasing evidence that actin fragmenting proteins might be targets in signaling cascades to the cytoskeleton. Gelsolin has been implicated in the phosphoinositide-mediated F-actin uncapping of human platelets following stimulation of thrombin receptors (10). Fibroblasts of gelsolin null mice have excessive actin stress fibers and migrate more slowly than wild type fibroblasts (11), while overexpression of gelsolin in NIH 3T3 fibroblasts leads to an increase in motility (12). In addition to Ca 2ϩ and polyphosphoinositides, phosphorylation seems to play an important role in regulating proteins from this family as well (13). However, except for a fragmin kinase (14) no other kinase has been described in detail so far.
The intracellular responses to external signals are very often mediated by kinase cascades. The best studied kinase cascade activated by external signals is the mitogen-activated protein kinase (MAPK) 1 system. Its core comprises a module of three kinases in which the most distal MAPK is activated by a MAPK kinase (MAPKK) which itself is activated by a MAPKK kinase (MAPKKK). MAPK modules are ubiquitous among eukaryotes and in recent years it has become clear that in every cell several pathways, responsive to different external stimuli, exist in parallel (15)(16)(17). The protein kinase PAK1 from rat brain was identified based upon its ability to interact with the small GTPases Rac1 and Cdc42. The binding of active Rac1/Cdc42 stimulated autophosphorylation and activity of PAK1. The sequence of PAK1 was found to be closely related to Ste20p, a key regulator in the mating pheromone response pathway in Saccharomyces cerevisiae, that acts via activation of a MAPK cascade (18 -20). The growing family of related kinases is referred to as either the PAK or Ste20-like kinase family. Although their in vivo role has not yet been clearly defined, PAK family members are considered to be promising candidates for the mediation of both Cdc42/Rac-induced effects, cytoskeletal reorganization, and transcriptional activation via a MAPK cascade (21,22).
Based on primary structure and mode of regulation, the PAK family can be subdivided into two main branches. Close relatives of PAK1 and Ste20p (true PAKs) are characterized by a C-terminal kinase domain and an N-terminal regulatory domain of variable length that contains a p21-binding domain (23) and, in some cases, a pleckstrin homology domain as well. Members of the second branch of the PAK family, the so-called GCK subfamily, have their catalytic domain positioned at the N terminus followed by a C-terminal regulatory region (21). Here we describe the isolation and characterization of a severin kinase from Dictyostelium, whose 62-kDa subunit is most closely related to human SOK-1, a member of the GCK subfamily of Ste20-like kinases.

MATERIALS AND METHODS
Protein Purification-Cells of D. discoideum strain AX2 were cultivated axenically at 21°C in 5-liter Erlenmeyer flasks up to a density of 5 ϫ 10 6 cells/ml, harvested without starvation, and homogenized by nitrogen excavitation in a Parr bomb essentially as described (24) in the presence of a mixture of protease inhibitors in the homogenization buffer (25). Usually 50 to 80 g of cells (wet weight) from 12 liters of culture were used for protein purification.
Rabbit actin was prepared from skeletal muscle according to Spudich and Watt (27) and further purified by gel filtration on Sephacryl S300. D. discoideum severin and actin were purified as described (28). The concentration of actin was measured as described (29). All other protein concentrations were determined by the method of Bradford (30) using bovine serum albumin as a standard.
Recombinant regulatory domain of severin kinase was purified from M15[pREP4] cells that had been transformed with the expression plasmid pQE32 containing the corresponding coding region. Cells were grown at 37°C, induced at an OD 580 nm of 0.6 with 0.5 mM isopropyl-1thio-␤-D-galactopyranoside for 2 h, harvested, opened by ultrasonica-tion as described (28), and insoluble material was pelleted (20 min, 30,000 ϫ g). The resulting pellet was stepwise re-extracted, once with TEDABP, pH 8.0, containing 150 mM NaCl and five times with TED-ABP, pH 8.0, containing 6 M urea (TEDABUP). The latter five supernatants were combined and applied onto a Ni-NTA column equilibrated in the same buffer. The column was washed with TEDABUP and then with TEDABUP containing 20 mM imidazole. The regulatory domain was eluted with TEDABUP containing 200 mM imidazole. Regulatory domain purified in this way was slowly dialyzed against TEDABP, pH 8.0, and concentrated by ultrafiltration in a Centricon-10 microconcentrator (Amicon GmbH, Witten, Germany). Two rabbits were immunized with the recombinant protein according to established procedures.
Immunoprecipitation-Polyclonal antibodies directed against the regulatory domain were used for immunoprecipitation of severin kinase from either a partially purified fraction (severin kinase pool after the Mono Q column) or from a crude fraction after opening the Dictyostelium cells (100,000 ϫ g supernatant). Kinase containing fractions (700 l) were incubated overnight at 6°C with 50 l of polyclonal antibody 5196 or polyclonal antibody 5197 in a total volume of 1 ml with a final concentration of 0.1% Triton X-100 in phophate-buffered saline buffer (150 mM NaCl, 100 mM Na 2 HPO 4 , 30 mM KH 2 PO 4 ), pH 8.0. Control immunoprecipitations were carried out in the absence of either polyclonal antibody or severin kinase. Protein-A Sepharose beads (25 l) were added, the suspensions shaken for 1 h, and centrifuged for 2 min at 10,000 ϫ g. The supernatants were removed, the pellets washed three times with 300 l of phosphate-buffered saline, resuspended in 50 l of TEDABP, pH 8.0, and aliquots (20 l) used in a phosphorylation assay with either severin or severin in the actin-severin complex as a substrate. The remaining beads were boiled after addition of 20 l of 3 ϫ SDS sample buffer and analyzed by SDS-PAGE.
Phosphorylation Assays-Severin kinase activity was assayed in a reaction mixture (40 l) containing 10 mM Tris/HCl, pH 7.5, 1 mM dithiothreitol, 1 mM EGTA, 1 mM Na 3 VO 4 , 2 mM sodium fluoride, 10 mM MgCl 2 , 0.05 mg/ml bovine serum albumin, 0.1 mM ATP (2-5 Ci of [ 32 P]ATP), 0.01% NaN 3 , and 1-4 M substrate. The reaction was initiated by addition of either the substrate or the kinase to the reaction mixture and carried out at 30°C. Severin (2-4 M final concentration), DS211C (2-4 M), or the 1:1 actin-severin complex (1-2 M each) were used as substrates. The actin-severin complex was allowed to form in G-buffer, followed by the addition of EGTA (1 mM final concentration) resulting in the EGTA stable 1:1 complex (26). Determination of the substrate dependence of severin kinase was carried out with different substrates at a final concentration of 2 M. For testing the Ca 2ϩ dependence of the phosphorylation reaction, the mixture contained, in addition, 2 mM Ca 2ϩ , and in the assays with Mn 2ϩ -ATP, 10 mM MnCl 2 instead of MgCl 2 . The pH dependence was tested by the addition of 1/10 volume of the following buffers to the reaction mixture: 500 mM MES, pH 6.0 and 6.5, 500 mM MOPS, pH 7.0, 500 mM Tris/HCl, pH 7.5, 8.0, and 8.5. Activation by autophosphorylation was tested by preincubating severin kinase with unlabeled ATP for 0, 5, or 20 min in the absence of substrate, then substrate was added and the reaction allowed to proceed for 20 min at 30°C. If not stated otherwise phosphorylation was terminated after 30 min by the addition of 20 l of 3 ϫ concentrated SDS sample buffer and boiling for 3 min. Proteins were separated by SDS-PAGE on minislab gels (110 ϫ 83 ϫ 0.5 mm) using the buffer system of Laemmli (31). Electrophoresis was terminated before the running front reached the lower buffer chamber and the gel was cut just above the running front to remove the lower gel strip which contains most of the non-incorporated radioactive ATP. Protein bands were visualized by staining with Coomassie Brilliant Blue and, after drying of the gels, labeled proteins were detected by autoradiography on Kodak X-AR films. For quantitation of incorporated phosphate, bands were scanned densitometrically and intensities evaluated with the program NIH Image 1.61.
Cloning and DNA Sequence Analysis-Tryptic fragments of the 62-kDa subunit of severin kinase were resolved by reversed-phase chromatography and subjected to Edman degradation on an Applied Biosystems gas-phase sequencer according to Eckerskorn et al. (32).  32 P]dATP and employed to screen a gt11 cDNA library (33) as described (34). From one positive clone the cDNA insert was amplified by PCR using primers of the gt11 flanking regions, cloned into pUC19 vector (35), and sequenced with the chain termination dideoxy method (36) using uni and reverse primers, as well as sequence specific oligonucleotide primers. The isolated cDNA had a size of about 1.2 kb and contained the 3Ј end, but lacked the 5Ј end of the gene. A 5Ј 0.8-kb EcoRI fragment of this clone was used to screen a random primed gt11 cDNA library (CLONTECH Inc., Palo Alto, CA) which yielded another positive clone with an insert of about 1.3 kb harboring the ATG start codon preceded by two in-frame stop codons but lacking the 3Ј end of the gene. The two cDNA clones had about 1-kb sequence in common and internal restriction sites were used to combine the two clones to yield the full-length cDNA in pUC19. In order to exclude possible errors resulting from PCR amplification of the gt11 cDNA clones, we confirmed both sequences at least once with independently amplified and cloned PCR products.
Standard techniques were used for cloning, transformation, and screening (37). Searches for similarities to other protein sequences were done with the program BLAST (38) using the combined non-redundant entries of the Brookhaven Protein Data Bank, Swiss-Prot, PIR, and GenBank at the NCBI. The sequence was analyzed by using the UWGCG (University of Wisconsin Genetic Computer Group; Madison, WI) and PHYLIP (Phylogeny Inference Package, version 3.5c by Joseph Felsenstein, University of Washington) program packages. Northern analysis followed established procedures (39).
The coding sequence for the regulatory domain (aa 277-478) was amplified by PCR (denaturation 94°C, 60 s; annealing 60°C, 60 s; elongation 72°C, 60 s; 25 cycles) with primers Sevkin-Ntreg (5Ј-CGCG-GATCCATATGAGAAGACAAAAATGGTTACAAT-3Ј) and Sevkin-Ct (5Ј-GCGAAGCTTTTATCTTTTAAGGGTTTCAATG-3Ј). Primer sequences corresponding to the coding sequence of severin kinase are shown in italic and restriction sites in the 5Ј overhang sequence for cloning in bold. The resulting PCR product was cloned into the BamHI, HindIII sites of the pQE32 expression vector (Qiagen GmbH, Hilden, Germany). The complete expression construct was sequenced to exclude possible errors resulting from the PCR.

Partial Purification and Characterization of Severin
Kinase-To identify protein kinases from D. discoideum that phosphorylate cytoskeletal proteins, we screened DEAE column fractions of soluble homogenates for kinase activities by adding actin, severin, or a 1:1 complex of both proteins as a substrate. We detected an activity that phosphorylated the actin-fragmenting protein severin either on its own or in a complex with actin. This severin kinase activity was further purified by additional chromatographic steps including gradient elution from S-Sepharose or Mono Q, and gel filtration on Superose 12 (Fig. 1). In the final Superose 12 gel filtration step the kinase eluted at a position corresponding to a molecular mass of about 300 kDa (Fig. 1D, inset, Superose 12). These active fractions contained a polypeptide of about 62 kDa which (i) coeluted with kinase activity, (ii) was strongly phosphorylated in the presence of [␥-32 P]ATP and (iii) shifted almost completely to a higher molecular mass in SDS-PAGE after preincubation with unlabeled Mg 2ϩ -ATP ( Fig. 2A). The low percentage gel (7.5% acrylamide) used in this experiment resolved the rather broad 62-kDa signal shown in Fig. 1 into at least three distinct bands (Fig. 2A, lane 3) which suggests multiple autophosphorylation of the 62-kDa protein. It is not yet clear whether native severin kinase is composed of only the 62-kDa subunit or whether it constitutes a heteromer. Autophosphorylation of the 62-kDa polypeptide exactly followed severin kinase activity during all purification steps and was therefore likely to represent the severin kinase (Fig. 1). In addition, polyclonal antibodies directed against the regulatory domain specifically precipitated the 62-kDa polypeptide from either a partially purified severin kinase fraction or from the soluble fraction after opening the Dictyostelium cells. In phos-phorylation assays, the 62-kDa polypeptide in the immunoprecipitate as well as added severin either on its own or in complex with actin were strongly phosphorylated (Fig. 2B, lanes 2 and  3). Phosphorylation of severin in the actin-severin complex was more pronounced than phosphorylation of severin alone. Results obtained with a second independently generated antiserum were very similar. Control immunoprecipitations were carried out in the absence of either polyclonal antibody or severin kinase fraction as described above. In the absence of antibodies (Fig. 2B, lane 1) or severin kinase (data not shown) there was no phosphorylation of severin. These results clearly demonstrate that the 62-kDa polypeptide is essential for severin kinase activity.
We used the peak fractions from the gel filtration column, the last purification step, to biochemically characterize severin kinase and to obtain sequences from tryptic peptides. Fig. 3 shows the time dependence (A), the activation by autophosphorylation (B), and the pH dependence (C) of severin kinase as measured by phosphorylation of domains 2 and 3 of severin (DS211C; see below). During early time points, there was an almost exponential increase of incorporation of phosphate into the substrate which could be attributed to self-activation of the kinase and excess of substrate (Fig. 3A). Autophosphorylation for 5 min increased the activity of severin kinase more than 3-fold and a nearly 6-fold increase in activity was observed after 20 min of in vitro autophosphorylation (Fig. 3B). The activity of severin kinase decreased rapidly at pH values below pH 7.0, while pH values above 7.5 decreased its activity only moderately (Fig. 3C). Routinely, phosphorylation assays were carried out for 30 min at pH 7.5 and 30°C.
The substrate specificity of severin kinase has been tested with domain 1 (DS151), domain 2 (DS111M), and domains 2 and 3 (DS211C) of severin (26), the 1:1 actin-severin complex, and with the Dictyostelium actin-binding proteins ␣-actinin, ABP120 gelation factor, and hisactophilin. Besides severin, on its own as well as in the 1:1 complex with actin, only DS151 (residues 1-151 of severin) and DS211C (residues 152-362 of severin) turned out to be substrates of severin kinase (data not shown). In particular DS211C was very strongly phosphorylated being an even better substrate than native severin. Since DS151 and DS211C have no overlapping amino acids, severin kinase must either phosphorylate native severin at two or more sites, or alternatively, there must be a cryptic phosphorylation site in the constructs that is not accessible in the native molecule.
Severin phosphorylation appeared to be regulated also at the substrate level as incorporation of phosphate was strongly reduced in the presence of Ca 2ϩ . The activity of severin kinase itself was not affected under these conditions because its autophosphorylation and also phosphorylation of DS211C remained nearly unaltered in the presence of Ca 2ϩ (Fig. 4). The addition of Ca 2ϩ triggers a conformational change in severin (28) that might render either the target amino acid or the complete protein inaccessible for severin kinase. Since severin kinase accepted also Mn 2ϩ -ATP as phosphate donor we tested the Ca 2ϩ dependence of the phosphorylation reaction under these conditions as well. A similar Ca 2ϩ dependence of severin phosphorylation was found with Mn 2ϩ -ATP, autophosphorylation of severin kinase, and phosphorylation of DS211C, however, overall phosphorylation was not as pronounced as with Mg 2ϩ -ATP (Fig. 4). region. Northern blot analysis of growth phase Dictyostelium cells showed one mRNA band with a size of approximately 1.7 kb (data not shown). The difference between the apparent molecular mass of 62 kDa in SDS-PAGE and the calculated molecular mass of approximately 53 kDa could be explained by reduced mobility of the polypeptide in SDS-PAGE. Similar differences in apparent and calculated molecular mass were reported for the two related kinases Krs-1 and Krs-2 (see below). Krs-1 and 2 have an estimated molecular mass of 63 and 61 kDa on SDS-polyacrylamide gels while the predicted molecular mass is 56.3 kDa for Krs-1 and 55.6 kDa for Krs-2 (40). In this report the authors suggest that a highly acidic region in the C-terminal domain could be responsible for the slightly aberrant migration behavior in SDS-PAGE. The calculated pI of the kinase is 6.7; all microsequences collected from the protein were present in the cDNA deduced amino acid sequence.
The predicted protein sequence indicates a two-domain organization of the protein (Fig. 5). The N-terminal part with 276 residues constitutes the catalytic domain characteristic of Ser/ Thr-and Tyr-protein kinases. All 11 subdomains typically found in these protein kinases (41) are present. The C-terminal domain encompasses 202 residues and is rich in glutamine, threonine, and proline residues. Most obvious are two glutamine-rich stretches between aa 323 and 347 and one threonine-rich stretch between aa 359 and 369 (9 out of 11 residues). Several proline residues (17 in total) are scattered throughout the central part of the C-terminal domain between residues 312 and 429. These regions could constitute binding interfaces for regulatory proteins. In addition, a highly acidic region is present between aa 290 and 306 with 10 negatively charged amino acids out of 17. Short acidic regions of unknown function have also been found in other proteins from the PAK family including mammalian PAKs, Ste20p, Krs-1 and 2, and SOK-1 (21,40).
In data base searches with the program BLAST (38) the highest degree of sequence similarity was observed in members of the PAK family of protein kinases. These kinases share a highly conserved catalytic domain and have the so-called PAK signature "GTPY/FWMAPE" in common (Fig. 5). They can be subdivided into two groups based on their structure and regulation. Ste20p, PAK1, MIHCK (42), and related PAKs have a C-terminal kinase domain and a p21 binding motif in the The proteins were separated by SDS-PAGE in 7.5% gels, stained with Coomassie Blue (lanes 1 and 2), or stained and processed for autoradiography (lane 3). The positions of the 62-kDa polypeptide before (*) and after (*Ј) autophosphorylation are indicated. The presence of at least three distinct bands in the autoradiogram suggests multiple autophosphorylation of the 62-kDa polypeptide. B, partially purified severin kinase from the Mono Q column was incubated in the absence (lane 1) or presence of polyclonal antibodies (lanes 2 and 3) that were raised against the regulatory domain of the 62-kDa polypeptide. After precipitation with protein-A Sepharose, the beads were used in phosphorylation reactions with either severin (lanes 1 and 2) or the actin-severin complex (lane 3) as substrates. Proteins were separated by SDS-PAGE and processed for autoradiography as described above. Please note that the specifically precipitated material showed the characteristic autophosphorylation and phosphorylation of the substrates .   FIG. 3. Time dependence (A), activation by autophosphorylation (B), and pH dependence of severin kinase activity (C). The phosphorylation reactions with DS211C as a substrate were carried out for the indicated periods of time (A), for 20 min after allowing severin kinase to autophosphorylate for 0, 5, or 20 min in the presence of unlabeled ATP (B), or for 30 min at the pH values stated (C). The reaction mixtures were separated by SDS-PAGE in 12% gels and the dried and stained gels subjected to autoradiography. The radioactive DS211C bands were scanned densitometrically and intensities evaluated with the program NIH image 1.61. N-terminal part. In the GCK branch of the PAK family the catalytic domain is positioned at the extreme N terminus (21). Based on its primary structure and sequence homology, the 62-kDa subunit of severin kinase clearly belongs to the GCK subfamily (Fig. 6A). Sequence comparisons of its catalytic domain with the catalytic domains of the other kinases revealed that the kinase subunit of severin kinase is most closely related to human SOK-1 (75% identity) and the open reading frame T19A5.2 from Caenorhabditis elegans (72% identity). However, even with the most distant member in this comparison, Ste20p from S. cerevisiae, the kinase from Dictyostelium shared 42% sequence identity in the catalytic domain.
To further clarify the relationship between PAK family members and the kinase subunit of severin kinase, we calculated multiple sequence alignments of the catalytic domains of PAK family members. In the evolutionary tree derived from these alignments the members of the two PAK branches are separated as expected, and for GCK subfamily members the tree is split into two main branches, one formed by KHS1, Rab8ip, HPK1, NIK, the other one by MST1, MST2, MESS1, NRK1, T19A5.2, SOK-1, and the kinase subunit of severin kinase. SOK-1, T19A5.2, and the kinase subunit of severin kinase are most closely related and listed together (Fig. 6B).
A sequence alignment with human SOK-1 is shown in Fig. 7. The two proteins are 75% identical and 84% similar in their catalytic domains. In addition, they share significant sequence similarity in the C-terminal domain. Two long regions of 44 and 58 amino acids with approximately 31% identity and 40% similarity, respectively, are present in this domain; the first one is adjacent to the catalytic domain and the second one is located at the extreme C terminus. A third short stretch of 16 amino acids in the central part of the C-terminal domains displays 33% sequence similarity and is flanked in the 62-kDa subunit of severin kinase by two insertions of 22 (amino acid 320 -341) and 51 (amino acid 358 -408) residues. Interestingly, when we compared the entire C-terminal domains of the Dictyostelium kinase and Rab8ip from mouse (43) or GCK from human (44) with the program Bestfit we found only one short homologous region of 19 residues with about 37% sequence similarity. All 16 residues of the central C-terminal homology region of the Dictyostelium kinase and SOK-1 were contained in these 19 residues, thus raising the possibility for this sequence to constitute an as yet unknown p21-binding motif. DISCUSSION Based on in vitro phosphorylation assays we have partially purified a severin kinase from cytoplasmic extracts of D. dis- coideum. Severin kinase has a molecular mass of about 300 kDa and harbors a 62-kDa subunit that is closely related to p21-activated protein kinases. Sequence comparisons of the catalytic domains of selected PAKs and the severin kinase subunit clearly identified the kinase as a new member of the GCK subfamily. It displays the highest similarity to human SOK-1 that is activated by oxidant stress (45). Both proteins are 75% identical in their catalytic and 31% identical in their regulatory domains and could therefore fulfill a similar or even identical in vivo function.
Severin kinase is the first example of a GCK subfamily kinase with a cytoskeletal protein as a possible in vivo target. In contrast to GCK subfamily members, several true PAKs have recently been implicated in cytoskeletal reorganization or the regulation of cytoskeletal proteins. Ste20p was found to bind to Bem1p which associates with actin (46) and PAK1 is thought to regulate actin organization in mammalian cells via an as yet unknown effector (47,48). MIHCK from Dictyostelium and its homologue from Acanthamoeba phosphorylate the heavy chain of some of the myosin I isozymes on a single serine or threonine residue and thereby stimulate their actin-activated Mg-ATPase activity 30 -50-fold (49,50). Cloning of the corresponding genes revealed that MIHCK is a member of the PAK family and closely related to mammalian PAK and yeast Ste20p molecules (42,51). In gel overlay assays and affinity chromatography experiments, MIHCK from Dictyostelium interacted with GTP␥S-labeled Rac1 and Cdc42, which probably bind to a conserved p21-binding domain commonly found in the N-terminal regulatory domain of true PAKs. Interestingly, in the presence of active Rac1 and Cdc42, autophosphorylation of MIHCK increased from 1 up to 9 mol of phosphate per mol of kinase concomitant with an approximately 10-fold stimulation of the rate of myosin ID phosphorylation. These results suggest that MIHCK directly links Cdc42/Rac signaling pathways to motile processes driven by myosin I molecules (42).
For members of the GCK subfamily the putative regulatory role of the C-terminal non-catalytic domain is not clear. In the case of MST1, MST2, and SOK-1 it apparently has an inhibitory function because its removal resulted in an increase in kinase activity (45,52). Furthermore, it has been shown that the C-terminal domains of MST1 and MST2 mediate homo-and heterodimerization (52). Rab8ip, the murine homologue of human GCK has been isolated in a two-hybrid screen as a Rab8 interacting protein (43). This finding was surprising because members of the GCK subfamily lack the conserved p21-binding domain of 16 amino acids found in the N-terminal regulatory domains of true PAKs (21,23). Thus it is possible that also other GCK subfamily members may be regulated by small GTPases as well, but that a common binding motif is not identified yet. We compared the sequences of the non-catalytic domain of GCK, Rab8ip, and the kinase subunit of severin kinase and found a stretch of 19 similar amino acids (amino acids 341-359 in the 62-kDa subunit of severin kinase), that could constitute a binding site for a small GTPase.
Like MIHCK and other kinases of the PAK family, severin kinase showed strong and possibly multiple autophosphorylation ( Fig. 2A) which resulted in a severalfold activation of kinase activity (Fig. 3B). In addition, severin phosphorylation seemed to be regulated at the substrate level since Ca 2ϩ strongly reduced phosphorylation of severin, whereas auto- FIG. 7. Sequence alignment of the kinase subunit of severin kinase (62 kDa) and human SOK-1 (SOK-1). The sequence alignment was done with the program Clustal from the UWGCG program package. Identical and similar residues are indicated by a star or a point, respectively. The arrowhead marks the start of the regulatory regions. Both sequences are closer related in their catalytic domains than in their regulatory regions. A small stretch of similar amino acids (341-359 in 62 kDa) is also present in GCK and Rab8ip, and might be a putative binding site for small GTPases. Residue numbers of both proteins are shown on the right. phosphorylation of the kinase and phosphorylation of DS211C were nearly unchanged (Fig. 4). In two-dimensional gel electrophoresis, purified severin resolved in three bands suggesting that also in vivo severin is subject to phosphorylation. Treatment of purified severin with severin kinase resulted in an additional more acidic spot (data not shown). It is at present not clear whether phosphorylation influences one or more of the in vitro activities of severin. This important issue is difficult to resolve because one has to be able to obtain not only fully phosphorylated severin, but also distinguish and characterize the phosphorylation sites in native severin as opposed to recombinant domains 1 (DS151) and domains 2 ϩ 3 (DS211C). Phosphorylation of fragmin, the Physarum homologue of severin, by a casein kinase II enzyme had no effect on the in vitro activity of fragmin (14). The authors speculate that phosphorylation of fragmin could be associated with an intracellular redistribution similar to gCAP39 which was shown to be preferentially associated with nuclear preparations in the phosphorylated state (13).
Several members of the GCK subfamily are responsive to cellular stress. Sps1p has been shown to become activated in response to nutrient deprivation (53). Human Krs-1 and Krs-2, which are identical with MST1 and MST2, are activated upon treatment of cells with staurosporine, okadaic acid, high concentrations of sodium arsenite, and extreme heat shock at 55°C (40,54,55). The activity of another member, human SOK-1, was shown to be induced severalfold by oxidant stress, but not by growth factors, alkylating agents, cytokines, heat shock, and osmotic stress. It most likely controls a novel stress response pathway since it is not involved in already defined MAPK cascades (45). This leads to the assumption that members from the GCK subfamily are important for the response of eukaryotic cells to environmental stresses. The in vivo regulation of severin kinase is not yet known. However, its close similarity to human SOK-1 and other kinases of the GCK subfamily suggests that it might also be activated in response to cellular stress. Possibly its activation leads to phosphorylation of severin and connects an extracellular signal to the cytoskeleton via an as yet unknown regulatory cascade. Disruption of the severin kinase gene and in vivo labeling experiments should help to unravel the in vivo role of Dictyostelium severin kinase.