Identification and molecular cloning of a p21cdc42/rac1-activated serine/threonine kinase that is rapidly activated by thrombin in platelets.

The brain-enriched p21cdc42/rac1-activated serine/threonine kinase, p65PAK, was identified and purified on the basis of overlays with [γ-32P]GTP-Cdc42 onto SDS-fractionated proteins (Manser, E., Leung, T., Salihuddin, H., Zhao, Z.-S., and Lim, L.(1994) Nature 367, 40-46). In this study, the ubiquitously expressed p21cdc42/rac1 binding protein with relative molecular weight of 62,000 was purified from rat testes and shown to contain peptides related to PAK. It has thus been designated as the γ-PAK isoform (α- and β-isoforms being brain enriched). Isolation of γ-PAK cDNAs show that the kinase is highly conserved with α-PAK in both the p21 binding and kinase domains. The purified protein exhibited kinase activity that was activated by GTP-Cdc42 or GTP-Rac1 in vitro. In platelets, a p62 in situ renaturable kinase was recognized by antibodies raised against γ-PAK. This thrombin-activated protein kinase appears to coprecipitate with another kinase of Mr 86,000, suggesting that PAK may be part of a thrombin-responsive signaling complex.

The mammalian p21 ras -related Rho subfamily of GTP binding proteins, including Rho, Rac, and Cdc42, are involved in regulating morphology. Microinjection studies with Swiss 3T3 fibroblasts have shown that RhoA plays a role in stress fiber formation, mediating their formation by growth factors such as platelet-derived growth factor (1). Activation of Rac1 induces membrane ruffling and a subsequent activation of Rho (2). Cdc42, originally identified as being required for development of polarity in Saccharomyces cerevisiae (3), has two mammalian homologues (4,5). Injection of mammalian Cdc42Hs causes filopodia formation in Swiss 3T3 cells (6,7). Cdc42 is required for bradykinin-induced morphological changes (6).
GTPase-activating proteins (GAPs), 1 which stimulate hydrolysis of the GTP bound to Rho subfamily members include the sequence-related Bcr, chimaerin, and RhoGAP (8). These down-regulatory GAPs are generally multi-functional; for example, Bcr has serine/threonine kinase (9) and dbl homology domains (10), while ␣2-chimaerin contains SH2 and phorbol ester binding domains (11). An overlay method that detects GAPs and Rho binding proteins (12) has facilitated the cloning and purification of GAPs including the Bcr-related Abr (13) and ␤-chimaerins (14,15) as well as putative target proteins, including the activated Cdc42-associated tyrosine kinase, p120 ACK (16) and the p21 cdc42/rac1 -activated kinase (p65 PAK ) (17). Brain-enriched p65 PAK is a serine/threonine kinase, activated by GTP-bound forms of Cdc42 and Rac and is homologous to the yeast kinase Ste20p, which forms part of the pheromoneinduced kinase cascade pathway in S. cerevisiae (18). Another class of Rac targets is p67 phox , which regulates the NADPH oxidase activity in phagocytes (19).
The overlay method also disclosed a ubiquitous 62-kDa protein binding Cdc42/Rac (12,17). Here, we show that this p62 represents a PAK isoform. The p62 PAK was purified from rat testes and subsequently cloned from a rat brain cDNA library. The cDNA encoded a protein designated as ␥-PAK, being highly homologous to ␣and ␤-PAK (20) in both the p21 binding and kinase domains. The purified native p62 PAK exhibited kinase activity that was stimulated by GTP-Cdc42 and GTP-Rac in vitro. ␥-PAK is widely expressed in all the rat tissues tested, unlike brain-enriched ␣and ␤-PAKs (20).
Thrombin stimulation of human platelets activates several signaling molecules including GTP binding proteins and phospholipase C-␥, leading to aggregation, cell adhesion, and secretion (21). Phospholipase C-␥ hydrolysis of inositides results in a transient increase of intracellular calcium and diacylglycerol, which subsequently activates protein kinase C (22). Other serine/threonine protein kinases activated by thrombin have been detected by a renaturable in situ kinase assay (23). One of these platelet protein kinases, a 62-kDa (p62) serine/threonine kinase, is recognized by antibodies to recombinant ␥-PAK.

MATERIALS AND METHODS
Purification and Peptide Analysis of p62 PAK -p62 PAK was purified by column chromatography at 4°C using a modification of the method described previously (17). Briefly, adult rat testes with the outer membrane removed were quickly frozen and stored at Ϫ70°C. The tissue (60 g) was thawed, minced, and resuspended in 300 ml of buffer Q (25 mM MES-NaOH, pH 6.5, 0.5 mM MgCl 2 , 0.05 mM ZnCl 2 , and 0.05% Triton X-100) containing 0.5 mM phenylmethylsulfonyl fluoride and 1 g/ml each of leupeptin, pepstatin, and aprotinin. The tissue was homogenized twice, 10 strokes each, with a Dounce glass homogenizer and centrifuged at 100,000 ϫ g for 1 h. The supernatant was loaded onto a 100-ml S-Sepharose column in Q buffer, in which the column was eluted stepwise with 0.8 column volumes of Q buffer with NaCl gradient from 0.1 to 0.4 M. The p62 PAK -containing fractions (eluted with 0.1 and 0.2 M NaCl) were pooled, diluted with an equal volume of Q buffer, and loaded onto a 30-ml Q-Sepharose column in Q buffer. The p62 PAK -containing fractions (eluted with Q buffer containing 0.2 and 0.3 M NaCl) were pooled and adjusted to pH 7.5 with 1 M Tris and loaded onto a 15-ml zinc-chelating Sepharose precharged with zinc (according to manufacturer's protocol). The column was washed with Z buffer (100 mM NaCl, 0.05% Triton X-100), pH 7.5. The p62 PAK was eluted with Z buffer, pH 6.0, and with added 0.1 mM GTP␥S was loaded onto a 1-ml glutathione-Sepharose column containing GTP␥S-glutathione S-transferase (GST)/ Cdc42 (17). The column was washed with 3 volumes of Z buffer, pH 6.0; proteins were eluted with 3 volumes of Z buffer, pH 8.8, and lastly with 2 volumes of 10 mM glutathione. The flow-through was reloaded onto another GST-Cdc42 column to recover unbound p62 PAK . The p62 PAKcontaining fraction was resolved on 9% SDS-polyacrylamide gel electrophoresis, and the p62 bands corresponding to ϳ20 g of protein were excised. Gel pieces were subjected to total tryptic digestion, and peptides were recovered by washing with 50% acetonitrile. Peptides were separated by reverse-phase chromatography on a C18 column under standard conditions.
Cloning of p62 PAK cDNA-PCR was performed by using degenerate oligonucleotides corresponding to peptides IVSIGDP (5Ј-GATCTA-GAT(T/C/A)GTNAG(C/T)AT(A/C/T)GGNGA(C/T)CC-3Ј) and WMAPEV (5Ј-GAC(T/C)TCIGGNGCCATCCA-3Ј) shown in Fig. 3B. 30 cycles (1 min at 94°C, 2 min at 50°C, and 3 min at 72°C) were used to amplify rat testis cDNA. An amplified DNA fragment of ϳ600 base pairs (probe A in Fig. 3A) was subcloned and sequenced to confirm its identity. This was used to screen 1 ϫ 10 5 clones of a rat brain cDNA library (Stratagene), which yielded three independent phage clones with overlapping inserts containing the PCR cDNA sequence (insert size,ϳ 1.5, 2.5, and 4 kb). The largest insert (clone A1 in Fig. 3A) was found to be truncated just before the expected p21 binding region present in the ␣-PAK cDNA sequence. The cDNA library was rescreened with an upstream 5Ј-probe (probe B in Fig. 3A) derived from clone A1 to obtain clones containing the missing upstream sequence. Of four phage clones with overlapping inserts isolated, only one clone contained the entire ␥-PAK sequence (clone B4 in Fig. 3A). The double-stranded cDNA was sequenced on both strands using Sequenase (version 2.0) as recommended by the manufacturer (U. S. Biochemical Corp.).
Subcloning of ␥-PAK-An oligonucleotide primer (LLMT53, 5Ј-GCTAGATCTATCATGTCTGATAACGGG-3Ј) was designed with a Bgl2 site in frame and upstream of the methionine start site. Using the clone B4 template, cDNA was amplified by PCR using this oligonucleotide primer and a reverse primer (LLMT52, 5Ј-CCCACCAGCAAGGTAC-3Ј) downstream of the unique HpaI site (Fig. 3A). 30 cycles of reaction (1 min at 94°C, 2 min at 50°C, and 3 min at 72°C) gave an amplified 0.9-kb product, which was subcloned into pGEX-2TK vector (Pharmacia, BamHI and SmaI digested). Most of this 0.9-kb PCR-derived cDNA insert was replaced with ␥-PAK cDNA sequence (clone B4 in Fig. 3A) by replacement of 0.8 kb with a 1.4-kb BamHI-EcoRI fragment yielding the full-length ␥-PAK cDNA. The 100-bp of amplified cDNA was sequenced. The full-length ␥-PAK cDNA in pGEX-2TK was cut with SalI and EcoRI to remove the C-terminal kinase domain (Fig. 3A), flushed, and religated to generate a construct able to express ␥-PAK 1-210 as a GST fusion protein.
Expression and Purification of ␥-PAK-The full-length ␥-PAK and ␥-PAK 1-210 in pGEX-2TK was transformed into XL-1 and BL-21 competent cells. Overnight cultures of the transformants were diluted 10fold with Lucia broth containing 50 g/ml ampicillin and grown to an optical density (at 600 nm) of ϳ0.6. The culture was then induced with 0.5 mM isopropylthiogalactoside for 4 -6 h at room temperature. The cells were pelleted by centrifugation at 10,000 ϫ g for 20 min, washed with PBS, repelleted, and stored at Ϫ20°C. Thawed cells were resuspended in buffer G (50 mM Tris, pH 8.0, 0.1% Triton X-100, 0.5 mM MgCl 2 in PBS) containing 1 mg/ml lysozyme, 5 mM dithiothreitol and 0.5 mM phenylmethylsulfonyl fluoride. The lysed cells were sonicated and centrifuged at 100,000 ϫ g for 30 min. The supernatant was loaded onto a glutathione-Sepharose (Pharmacia) column and washed with 2 column volumes of buffer G. Proteins were eluted with 5 mM glutathione in buffer G.
Isolation and Thrombin Activation of Rat Platelets-Fresh rat blood was collected in 50-ml tubes containing 0.1 volume of 100 mM citrate, 50 mM EDTA, and 1.5 units/ml of apyrase (Sigma). Platelets were recovered from this blood using NycoPrep 1.063 (Nycomed) according to manufacturer's protocol. The platelets were resuspended in buffer P2 (137 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 0.4 mM NaH 2 PO 4 , 5.6 mM glucose, 5 mM HEPES, pH 7.3, and 1 unit/ml apyrase) at a concentration of about 5 ϫ 10 8 cells/ml. To activate platelets, bovine thrombin (Sigma) was added at 1 unit/ml platelets, and the mixture was incubated at 37°C for 5 min or at times indicated. The cells were pelleted at 14,000 ϫ g on the microfuge for 1 min and stored at Ϫ70°C.
Immunoprecipitations-Thrombin-activated platelets were solubilized in IP buffer (50 mM HEPES, pH 7.3, 150 mM NaCl, 0.5% Triton X-100) containing 0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml each of leupeptin, pepstatin, and aprotinin for 10 min, sonicated for 10 s, and centrifuged for 10 min at 4°C. The clarified lysate was incubated with rabbit polyclonal antiserum raised against GST/␥-PAK 1-210 fusion protein for 2 h at 4°C. The mixture was loaded onto a column containing protein A-Sepharose (Sigma), which was washed with 3 volumes of IP buffer. Proteins were eluted with 2 ϫ SDS-sample buffer, boiled for 5 min, and subjected to gel electrophoresis and Western blot analysis.
p21 Binding, GAP Overlay, and in Vitro Kinase Assays-These assays were performed essentially as described previously (12,17).
Western Blot Analysis-Following transfer of proteins to PVDF or nitrocellulose membranes, these filters were blocked overnight at 4°C in PBS containing 5% skimmed milk. All subsequent steps were at room temperature. Filters were washed (3 ϫ 5 min) with PBS containing 0.1% Tween-20 (BDH) and incubated with primary antibody (at 1/1000 dilution) in 1% skimmed milk for 2 h. After washing (5 ϫ 5 min) in PBS containing 0.1% Tween-20, the filters were incubated with horseradish peroxidase-conjugated secondary antibody for 1 h. The filters were washed, and enhanced chemiluminescence signals were detected by placing to film (Amersham) for 10 -60 s, depending on exposure.
Other Assays-Protein concentrations were measured using the Bradford assay with protein dye reagent (Bio-Rad) according to manufacturer's protocol.

Purification of the Ubiquitous p62 cdc42/rac1 Binding Protein-
The p68 cdc42/rac1 and p65 cdc42/rac1 binding isoforms of PAK (␣-and ␤-PAK, respectively) (17,20) are preferentially found in the brain while the 62-kDa binding protein is ubiquitous (17). This p62 could be partially purified from rat testes (containing little of the larger binding proteins) by chromatography on S-, Q-, and zinc chelating-Sepharose columns employing [␥-32 P]GTP-Cdc42 overlay for detection (Fig. 1A). Affinity chromatography with GTP␥S-Cdc42 (Fig. 1B) then yielded two Cdc42-associated bands, the lower one being a proteolytic product of the p62 (17). Although the p62 did not bind strongly, appearing in all fractions of the GST/Cdc42 affinity column, microgram quantities were obtained from the pH 8.8 eluted fraction (shown on a Coomassie-stained gel in Fig. 1B, lane 23). The 68-kDa protein is residual albumin (added to the nucleotide exchange buffer in the GST-Cdc42 column). Gel-purified p62 and p60 proteolytic bands on tryptic digestion and reversephase chromatography yielded identical fingerprints (data not shown). The sequence of five peptides derived from p62 is shown in Fig. 3B. Only one (STMVGTPYWMAPE) was identical to the predicted ␣-PAK sequence with all others containing at least one amino acid substitution. The p62 thus represented a PAK isoform, which we designated as ␥-PAK.
Activation of p62 PAK in Vitro-As with p65 PAK , purified p62 PAK was activated by Rac and Cdc42 ( Fig. 2A). Cdc42 was more effective than Rac as determined by the increased autophosphorylation and myelin basic protein (MBP) phosphorylation. Both p62 PAK and MBP (Fig. 2B) were phosphorylated on serine and threonine residues. GTP␥S-Cdc42 induced a larger mobility shift than GTP␥S-Rac1, as monitored by Cdc42 overlays of the in vitro activated kinase (Fig. 2C, lanes 3-5). Phosphorylated p62 PAK (pp62) still bound Cdc42.
Isolation and Sequencing of ␥-PAK cDNAs-Degenerate oligonucleotides corresponding to the sequences IVSIGDP and WMAPEV (Fig. 3B) were used as primers for PCR of rat testis cDNA. The expected 600-bp PCR product encoded one PAKrelated sequence including NVLLGMEGSVK of purified p62 PAK . This PCR product (probe A, Fig. 3A) was used to screen a rat brain cDNA library, which yielded three clones with overlapping inserts at the C-terminal end (Fig. 3A). The longest clone of 4 kb appeared to be truncated at its 5Ј-end corresponding to a position before the p21 binding domain. Further screening with upstream sequences from the 4-kb clone (probe B, Fig. 3A) yielded clones containing upstream sequences that included the p21 binding domain. The cDNA of clone B4 contained 1756 nucleotides with a 1575-nucleotide open reading frame encoding a predicted 58-kDa protein of 524 amino acids (Fig. 3A). The putative translation start site began at nucleotide 49 (Fig. 3B), with upstream flanking sequences conforming to a translation initiation consensus sequence (24). The first five amino acids of ␥-PAK showed homology to the predicted N-terminal sequence of ␣-PAK (Fig. 4).
Predicted Amino Acid Sequence of ␥-PAK-The ␥-PAK open reading frame encoded all five peptide sequences derived from purified p62 PAK (underlined in Fig. 3B). ␥-PAK showed 76% identity with ␣-PAK, with particular divergence in the linker region between the p21 binding domain (BD) and the kinase domain (Fig. 3A). The latter two shared Ͼ90% identity. Two features in the linker region are conserved, polyacidic (8 consecutive residues) and proline-rich sequences. ␥-PAK showed 94% identity to human hPAK65 (25), suggesting it to be the rat homologue. However, the N-terminal hPAK65 sequence appeared truncated relative to the rat ␣and ␥-isoforms. A database search revealed hPAK65 5Ј-non-coding and 38 bp of pu-tative coding sequence to be identical to human placental lactogen (somatomammotropin) cDNA (26). Thus, it is likely that the N-terminal 12 amino acids of hPAK65, which do not match the rat PAKs, are not present in the bona fide hPAK65 protein.
Expression of ␥-PAK-␥-PAK mRNA was detected in all rat tissues examined, primarily as a 4.6-kb transcript (Fig. 5A). Two weaker ubiquitous transcripts of 7 and 4 kb may represent alternate spliced mRNAs or cross-hybridizing transcripts of other PAK isoforms. The mRNA expression pattern resembles that of the ubiquitous p62 cdc42/rac1 binding protein (Fig. 5B), which is also similar in size to the in vitro translation product of clone B4 (Fig. 5B). We were unable to generate a full-length recombinant kinase as a GST-fusion protein. However, we were able to express truncated GST/␥-PAK 1-210 as a 50-kDa fusion protein, which exhibited robust Cdc42 binding in overlay assays (Fig. 6A). GST/␥-PAK 1-210 and GST/␣-PAK 1-250 bound [␥-32 P]GTP-Cdc42 in a similar fashion, within the range of 0 -500 ng of protein, consistent with the high conservation within the p21 binding domain. Both GST/␣-PAK 1-250 and GST/␥-PAK 1-210 also inhibited intrinsic GTPase activity of Cdc42 similarly (Fig. 6B), doubling the half-life of bound GTP PCR was performed by using degenerate primers corresponding to peptides IVSIGDP and WMAPEV shown in Fig. 3B to amplify rat testis cDNA. An amplified 600-bp PCR product (probe A) was sequenced and used to screen a rat brain cDNA library. Three phage clones with insert sizes of 4, 1.5, and 2.5 kb (clones A1, A2, and A3, respectively) contained overlapping C-terminal sequences. The longest clone (A1) was truncated at the 5Ј-end, and an upstream probe (probe B) of 400 bp from clone A1 was used to rescreen the same library. Four clones of sizes 4, 0.9, 3, and 2 kb (clones B1, B2, B3, and B4, respectively) were obtained. Clone B1 was identical to clone A1, and clones B2 and B3 contained overlapping N-terminal sequences. The 2-kb clone (B4) contained the entire ␥-PAK cDNA sequence including the translational start site. The restriction map for clone B4 is also shown. Bottom, a schematic diagram of the B4 open reading frame and the relative positions of the p21 binding and kinase domains, and untranslated regions. B, the cDNA and predicted amino acid sequences of ␥-PAK. The five peptide sequences derived from purified p62 PAK are underlined. from 12 to 24 min at 20°C. Full-length GST-␣-PAK and GST-␣-PAK 1-250 had the same effect on intrinsic hydrolysis, suggesting the kinase domain does not influence the p21 interaction under these conditions.
Thrombin Activates p62 PAK in Platelets-Thrombin treatment of platelets activates several serine/threonine kinases, leading to processes including aggregation (21). Of the 12 rapidly activated serine/threonine kinases detected using a renaturable in situ kinase assay, the p170, p86, and p60 showed highest activity (23). The identities of these kinases, except p42-44 mitogen-activated protein kinase (27), are unknown. Fig. 7A illustrates that addition of thrombin (1 unit/ml) rapidly increased the activities of three platelet kinases of 170, 86, and 62 kDa. The major Cdc42 binding protein of 62 kDa in platelet lysates had identical migration characteristics as the 62-kDa kinase (Fig. 7B). Comparison with brain and testes (left panel, Fig. 8A) showed platelets to have no novel Cdc42 interacting proteins.
However, as we observed strong GAP activity in p85 and p50 in platelets even in the binding assays (Fig. 7B, seen as white bands), we subjected platelets and rat tissues to the GAP overlay technique (12). Brain cytosol contained GAPs (white bands) of ϳ 150, 100, 85, 50, and 45 kDa as previously reported (Fig. 8A, lane 4, marked) with a faint band of 75 kDa (the white bands of 60 -68 kDa result from [␥-32 P]GTP-Cdc42 sequestration (binding) by the PAK isoforms rather than from GAP activity). Platelets contained Cdc42 GAPs of sizes 85, 75, 50, and 45 kDa (Fig. 8A, lane 6). To determine if these were due to proteins containing Rho GAP-like domains such as Abr or the p85␣ regulatory subunit of phosphatidylinositol 3-kinase, an immunoblot was assayed with appropriate antibodies. Abr, present in brain, was not detected in platelets; both tissues, however, apparently contained the p85 subunit (Fig. 8B).
Polyclonal antibodies against ␥-PAK 1-210 , lacking the kinase domain, detected only the p62 PAK and did not significantly cross-react with the ␣and ␤isoforms, as assessed by [␥-32 P]GTP Cdc42 binding to immunoprecipitates from brain extract (Fig. 9C, Cdc42 overlay panel, cf. lanes 1 and 2). The platelet p62 was efficiently immunoprecipitated, exhibiting both Cdc42 binding and serine/threonine kinase activity in situ (Fig. 9, A and B, lanes 2). It was not tyrosine phosphorylated (data not shown). A serine/threonine kinase of about 86 kDa appeared to be coimmunoprecipitated with p62 PAK and which did not bind Cdc42 (Fig. 9, A and B, lanes 2). This p86 was not detected by the ␥-PAK anti-serum (data not shown). Preincubation of the antibodies with ␥-PAK 1-210 fusion protein drastically decreased immunoprecipitation of both p62 PAK and p86 (Fig. 9B, lanes 3). The p86 kinase was not detected in brain immunoprecipitates (Fig. 9C) (the brain p90 with weak kinase activity was not affected by the preincubation of the antibodies with ␥-PAK fusion proteins, cf. lanes 2 and 3). Another renaturable immunoprecipitated brain protein kinase (p65) was detected (Fig. 9C, in situ kinase panel, lane 2). Although it did not bind Cdc42 (Fig. 9C, Cdc42 overlay panel, lane 2), its activity was diminished when the antibodies were preincubated with ␥-PAK proteins (Fig. 9C, in situ kinase panel, lane 3). The identities of the p62 PAK -associated protein kinases, p86 and p65, are under study. DISCUSSION In this study, we report the purification and subsequent cloning of a 62-kDa Cdc42/Rac binding protein, which corresponds to an isoform of PAK, termed ␥-PAK. ␥-PAK is ubiquitously expressed in rat tissues, unlike ␣and ␤-PAK, which are found in few tissues other than brain (20). Thus, ␥-PAK reflects the widespread distribution of Rac and Cdc42 (28) and is prob- ably a common target protein for signaling by these proteins. Purified native p62 PAK activity toward MBP is increased on binding to GTP-Cdc42 and GTP-Rac in vitro. Unlike purified p65 PAK , the p62 PAK does not show decreased binding to Cdc42 upon activation. This is similar to results using recombinant hPAK65 (25), except that we find PAK to be phosphorylated in both threonine and serine residues. hPAK65 and rat ␥-PAK appear to be homologues. The only non-conservative amino acid difference is at ␥-PAK G349 , where hPAK65 contain an arginine; otherwise, this glycine is present in all mammalian and yeast PAK-like kinases that are known. Curiously, although the predicted hPAK65 contains 18 amino acid residues less than p62 ␥-PAK , its reported mobility represents a molecular mass of 65 kDa. This difference could be due to the presence of 12 amino acids derived from the placental lactogen mRNA sequence. It is likely that the reported hPAK65 cDNA does not exist as a bona fide mRNA in vivo, although a splicing event between two mRNAs is possible. The rat ␥-PAK N terminus shows blocks of homology to ␣-PAK and has been identified in two independent clones. The use of a probe derived from the conserved kinase domain of hPAK65 (25) will result in detection of mRNA of various PAK isoforms. The use of a probe sequence corresponding to the more variable linker domain allowed us to determine that the 4-kb mRNA is the dominant ␥-PAK mRNA in rat tissues. Although the ϳ170-kDa Cdc42/ Rac binding protein might be thought to arise by alternate splicing from the larger 7-kb mRNA, this family of binding proteins is not sequence related to PAK. 2 In platelets, the thrombin-activated p62 kinase (23) probably corresponds to p62 PAK , suggesting the involvement of Rac and 2 T. Leung and L. Lim, unpublished data.  Fig. 1. The tissue samples are as follows: brain (B), heart (H), kidney (K), liver (Li), lung (Lu), spleen (S), testis (Te), and thymus (Thy). An in vitro translation reaction with B4 cDNA was performed using a TNT-coupled reticulocyte lysate system according to manufacturer's protocol (Promega). The ␥-PAK translated products (5 and 10 l of lysate, lanes 1 and 2) were separated on 9% SDS gel and exposed to Hyperfilm (Amersham). Cdc42 in platelet function. Thrombin also appears to require Rho proteins to mediate cell aggregation, which can be inhibited by C3 transferase, which ribosylates Rho (29). Rho has been reported to activate a downstream target, p85 phosphatidylinositol 3-kinase, in platelets (30). The role of Cdc42 and Rac is likely to involve the dramatic changes in cell shape that accompany platelet activation, since mammalian Cdc42 has been associated with cell spreading in monocytes (3l), and microinjected Cdc42 (6, 7) and Rac (2) proteins can induce filopodia and membrane ruffling, respectively, in Swiss 3T3 cells. In yeast, the PAK-related kinase Ste20p is thought to lie at the top of the pheromone-induced mitogen-activated protein kinase cascade (18). This kinase has been implicated as a target for the Ste4/Ste18 ␤␥ subunits and interacts with Cdc42 (32), suggesting that mammalian Rho p21s may regulate mitogen-activated kinase cascades.
Since Ras and Rac can co-operate in transformation (33) and a Ras target, phosphatidylinositol 3-kinase, has been suggested to mediate platelet-derived growth factor-induced membrane ruffling in fibroblasts (34), these p21s may signal both morphological and transcriptional changes in cells. A candidate Rho target is phosphatidylinositol 4-phosphate 5-kinase, which may mediate the effects of Rho on the cytoskeleton in the integrin signaling pathway (35). It is likely that Rho family FIG. 7. Thrombin stimulation of renaturable kinases in platelets. A, platelets were incubated with 1 unit/ml of thrombin at 37°C for 0, 0.5, 1, and 3 min as indicated. The platelet lysates were separated by 9% SDS gels, transferred onto PVDF membranes, and assayed for kinase activity in situ. B, The p62 kinase comigrated with the major [␥-32 P]GTP-Cdc42 binding activity in these extracts.
FIG. 9. Immunoprecipitated p62 protein exhibits both Cdc42 binding activity and in situ kinase activity. A, and B, immunoprecipitation of p62 PAK from platelets. Thrombin-activated platelet lysates (lanes T) were immunoprecipitated with either ␥-PAK antibodies (raised to GST/␥-PAK 1-210 fusion proteins) alone (lanes IP) or antibodies preincubated with GST/␥-PAK 1-210 fusion proteins (lanes IPϩC). The lysates or immunoprecipitates were separated on 9% SDS gels, transferred, and assayed for [␥-32 P]GTP-Cdc42 binding activity or for kinase activity in situ. Strong signals in lane IPϩC of the Cdc42 overlay correspond to the GST/␥-PAK 1-210 , which does not include the kinase domain. C, immunoprecipitation of p62 PAK from brain supernatant. Rat brain supernatant (lanes T) was immunoprecipitated with ␥-PAK antibodies alone (lane IP) or in the presence of GST/␥-PAK 1-210 fusion proteins (lanes IPϩC). The purified GST/␥-PAK 1-210 protein was also shown as a reference (lanes C). The brain lysate and immunoprecipitates were separated on 9% SDS gels, Coomassie stained, or transferred onto PVDF membranes, and assayed for [␥-32 P]-Cdc42 binding activity or for kinase activity in situ.
p21s as with Ras p21s function in part by activating protein and lipid kinases. Although transforming tyrosine kinases, Ras and its target Raf kinases, are generally thought to act in cell proliferation and differentiation, it is noteworthy that proteins such as Src and mitogen-activated protein kinase are present in high concentrations in platelets (36,27). Platelets undergo extensive cytoskeletal reorganization during activation (21); the rapidity of ␥-PAK activation suggests a primary role for the kinase in these events. Additionally in mast cells, Rac and Rho have been reported to play a role in regulating secretion (37). The putative p62 PAK -associated p86 kinase in platelets, which is also a rapidly activated serine/threonine kinase (23), may form a signaling complex perhaps mediated by the heterotrimeric G-proteins activated by thrombin. We have yet to determine whether this p86 kinase is responsible for the Cdc42 GAP activity observable in overlay assays. Further analysis of proteins associated with ␥-PAK are likely to throw light on the role of PAK since some of these may well be downstream targets for the kinase.