Activation of mitogen-activated protein kinase cascades by p21-activated protein kinases in cell-free extracts of Xenopus oocytes.

In the evolutionarily distant yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe, genetic evidence suggests that activation of pheromone-induced mitogen-activated protein kinase (MAPK) cascades involves the function of the p21-activated protein kinases (PAKs) Ste20 and Shk1, respectively. In this report, we show that purified Ste20 and Shk1 were each capable of inducing p42 activation in cell-free extracts of Xenopus laevis oocytes, while a mammalian Ste20/Shk1-related protein kinase, p65 (Pak1), did not induce activation of p42. In contrast to p42, activation of JNK/SAPK in Xenopus oocyte extracts was induced by both the yeast Ste20 and Shk1 kinases, as well as by mammalian Pak1. Our results demonstrate that MAPK cascades that are responsive to PAKs are conserved in higher eukaryotes and suggest that distinct PAKs may regulate distinct MAPK modules.

In eukaryotes, responses to numerous types of extracellular signals are mediated by highly conserved mitogen-activated protein kinase (MAPK) 1 modules (1)(2)(3). The basic MAPK mod-ule consists of a MAPK kinase kinase (MAPKKK), a MAPK kinase (MAPKK), and a MAPK. MAPKKKs phosphorylate and activate MAPKKs, which phosphorylate and activate MAPKs (4). Homologs of these kinases appear to be both structurally and functionally conserved in evolution (3,5,6). Data from yeast genetic studies, and more recently from studies of mammalian cells, suggest that different types of extracellular signals activate distinct MAPK modules. For example, in the budding yeast Saccharomyces cerevisiae, distinct MAPK modules mediate cellular responses to mating pheromones, osmotic shock, extracellular salt concentration, and nutrients (3). Evidence has been obtained demonstrating the existence of at least four distinct MAPK modules in mammalian cells, although it is likely that the actual number of modules is substantially larger (2,(7)(8)(9)(10)(11)(12).
At present, we have acquired only a limited understanding of how distinct MAPK modules are individually regulated. The best characterized mammalian MAPK cascades are those regulated by receptor tyrosine kinases, such as the epidermal growth factor and platelet-derived growth factor receptors, which signal through Ras proteins via Grb2-related adaptor proteins and Ras guanine nucleotide exchange factors (13). Ras proteins, in turn, directly control the activity of the MAPKKK, Raf-1 (14). Substantially less is known about the regulation of MAPK cascades controlled by other types of receptors, in particular those activated by G protein-coupled seven transmembrane domain receptors (serpentine receptors) (2,15). However, recent studies (16 -18) have provided evidence for the existence of both Ras-and Raf-independent MAPK cascades.
In the evolutionarily distant yeasts S. cerevisiae and Schizosaccharomyces pombe, the p21 cdc42/rac1 -activated protein kinases (PAKs) Ste20 and Shk1, respectively, have been implicated as functioning upstream of pheromone receptor-induced MAPK modules (19,20). 2 Wu et al. (21) recently showed that Ste20 immune complexes phosphorylate the S. cerevisiae MAP-KKK, Ste11, implicating Ste20 as a potential MAPKKK kinase. PAKs structurally related to Ste20 and Shk1 have been identified in mammalian cells (22,23). However, it has not yet been determined whether mammalian PAKs regulate MAPK cascades.
In this report, we show that purified Ste20 and Shk1 protein kinases are each capable of inducing activation of p42 MAPK in cell-free extracts of Xenopus laevis oocytes. In contrast, p65 pak (Pak1), a mammalian PAK structurally related to Ste20 and Shk1, did not induce p42 MAPK activation in Xenopus oocyte extracts. We show that the S. cerevisiae MAPKKK, Ste11, is capable of inducing p42 MAPK activation, and that the N-terminal regulatory domain of Ste11 partially blocked Ste20-induced activation of p42 MAPK . Finally, we show that unlike p42 MAPK , activation of a JNK/SAPK cascade in Xenopus oocyte extracts was induced by both the yeast Ste20 and Shk1 kinases and by mammalian Pak1. Our data demonstrate that MAPK cascades that are activated by PAKs are conserved in higher eukaryotes and suggest that distinct MAPK modules may be regulated by different types of PAKs.

MATERIALS AND METHODS
Nucleic Acid Manipulation and Analysis-The plasmids pRD56 (for expression of GST fusion proteins in S. cerevisiae), pGSTSHK1⌬N (for expression of a GST-Shk1⌬N (residues 183-450) in S. cerevisiae), pRD56STE11 (for expression of a GST-Ste11 fusion protein in S. cerevisiae), pRP259 and pRP259STE11⌬C (for expression of a GST and GST-Ste11⌬C (residues 1-415) fusion proteins, respectively, in Escherichia coli), and pTrcHisHRAS[G12V] (for expression of a polyhistidinetagged H-Ras[G12V] protein in E. coli) have been described previously (20, 24 -26). The plasmid pRDSTE20RI, which was used for expression of the Ste20⌬N (residues 496 -939) GST fusion protein (GSte20⌬N) in S. cerevisiae, was constructed by cloning an EcoRI-KpnI fragment of the STE20 gene into the corresponding sites of the plasmid pRD56. The oligonucleotide primers 5Ј-ATGTCTCAGTGGCCATTAGGCAAATGA-ATCTC-3Ј and 5Ј-TCGTGTAATGGTAACCGACAATCGATGAGAAT-3Ј were used in a polymerase chain reaction to amplify a fragment of the STE20 gene carrying the K649R mutation. The resulting product was digested with MscI and KpnI and used to replace the corresponding fragment of the wild type STE20 gene, producing the mutant gene STE20 [K649R]. An EcoRI-KpnI fragment of STE20⌬N[K649R] was cloned into the corresponding sites of pRD56, producing the plasmid pGSTE20⌬N[K649R], which was used to express GST-Ste20⌬N-[K649R] fusion protein in S. cerevisiae. pGEXKGPAK1-232, which was used for expressing GST-Pak1⌬N (residues 232-544) in E. coli, will be described elsewhere. 3 Expression and Purification of Fusion Proteins-AN43-5A (MATa ade1 arg4 leu2-3, 112 trp1 ura3-52 mfa1::FUS1::lacZ his3::FUS1:: HIS3) was used for expression of GST fusion proteins in S. cerevisiae. Yeast cultures expressing GST fusion proteins were grown to about 3 ϫ 10 6 cells/ml in drop out medium (27) at 30°C with shaking. Cells were harvested by centrifugation, resuspended at about 10 6 cells/ml in 1 liter of drop out medium without glucose and containing 2% galactose, 2% glycerol, and 1% ethanol, and grown to a density of about 10 7 cells/ml. Cell extracts were prepared as described previously (24). The yeast cell lysate (5 ml) was mixed with 2 ml of glutathione-agarose (Sigma) and the resulting slurry incubated at 4°C for 30 min. The beads were washed three times with phosphate-buffered saline (PBS) and resuspended in 2 ml of PBS. GST fusion protein-bound beads were stored at 4°C in PBS. Polyhistidine-tagged H-Ras[G12V] protein was expressed and purified as described previously (26).
Phosphorylation and MAPK Activation Assays-Kinase assays were performed by mixing 5 l of purified protein suspension with 25 l of kinase buffer (30 mM HEPES, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol, 1 mM Na 3 VO 4 ), adding 1 l of [␥-32 P]ATP (6000 Ci/mmol, 100 Ci/ml (DuPont NEN)), and incubating for 15 min at 30°C. Reactions were stopped by adding 10 l of 4 ϫ SDS-PAGE sample buffer and boiling for 5 min. Samples were resolved by electrophoresis in 12% polyacrylamide gels and then exposed to film for autoradiography. MAPK assays were performed by incubating fusion proteins with extracts prepared from stage 5 to stage 6 Xenopus oocytes as described previously (26). Protein blots were probed using anti-ERK polyclonal antibody, 691 (10), and developed using the ECL chemiluminescent detection kit as described by the manufacturer (Amersham Corp.).

RESULTS
Genetic experiments have suggested that the S. cerevisiae and S. pombe PAKs, Ste20 and Shk1, respectively, function upstream of mating pheromone-induced MAPK modules (19,20). 2 To examine whether PAK-responsive MAPK cascades are conserved in evolution, we utilized a cell-free system for examining the activation of p42 MAPK in extracts of X. laevis oocytes (26). This and similar systems have previously been used to demonstrate in vitro activation of p42 MAPK induced by purified Ras, Mos, protein kinase A, and protein kinase C (26, 28 -30). Using this assay system, a direct correlation between the mobility shift induced by these various agents and activation of the kinase activity of p42 MAPK has been demonstrated (26). A GST-Ste20⌬⌵ fusion protein (GSte20⌬N), lacking amino acid residues 1-495 of the Ste20 regulatory domain, and a GST-Shk1⌬N fusion protein (GShk1⌬N), lacking residues 1-182 of the Shk1 regulatory domain, were expressed in yeast and purified from cell lysates using glutathione-agarose beads (see "Materials and Methods"). Both GSte20⌬N (Fig. 1A) and GShk1⌬N (not shown) were judged to be catalytically active, based on the ability of each to autophosphorylate in vitro. As shown in Fig. 1B, GSte20⌬N rapidly induced activation of p42 MAPK in Xenopus oocyte extracts, as measured by a decrease in the mobility of p42 MAPK . Using less than 1% of the amount of GSte20⌬N protein compared with Ras protein, we detected Ste20⌬N-induced p42 MAPK activation in 30 min, while p21 rasinduced p42 MAPK activation required 2 h. GShk1⌬N also induced activation of p42 MAPK when added to Xenopus extracts, although not as rapidly as GSte20⌬N (Fig. 1B). A bacterially expressed Ste20 protein that contained a functional regulatory domain was also capable of inducing p42 MAPK activation in Xenopus oocyte extracts, albeit not as effectively as either GSte20⌬N or GShk1⌬N (data not shown).
To determine whether p42 MAPK activation was dependent on Ste20 catalytic function, we introduced a mutation into the GSte20⌬N coding sequence at Lys 649 , a residue that is conserved in protein kinases (31). GSte20⌬N[K649R] was catalytically inactive, as determined by its inability to autophosphorylate, and did not induce p42 MAPK activation when added to Xenopus oocyte extracts (Fig. 2). Thus, p42 MAPK activation induced by Ste20 requires that the protein be catalytically active.
Genetic data suggest that the Ste11 protein kinase functions downstream of Ste20 in the S. cerevisiae pheromone response pathway (19). Therefore, we tested whether the N-terminal regulatory domain of Ste11 could inhibit activation of p42 MAPK induced by GSte20⌬N in Xenopus oocyte extracts. The Ste11 regulatory domain (residues 1-415) was purified from a bacterial expression system as a GST-fusion protein. As shown in Fig. 3A, GSte11⌬C partially blocked p42 MAPK activation induced by GSte20⌬N in Xenopus oocyte extracts. This result is consistent with two possible interpretations: (i) the Ste11 N terminus inhibits the activity of a protein activated by Ste20 in Xenopus oocyte extracts or (ii) Ste11 is a substrate of Ste20.  Ste20⌬N (GSte20⌬N) proteins were expressed in yeast and purified from cell extracts using glutathione agarose beads as described under "Materials and Methods." Kinase assays were performed in the presence (ϩ) or absence (Ϫ) of 5 ng of GSte20⌬N attached to glutathione agarose beads and/or 250 ng of control GST beads (to ensure that the GST moiety was not phosphorylated by GSte20⌬N). Samples were resolved by 12% SDS-PAGE, then exposed to film for autoradiography. Having determined that the Ste11 N-terminal regulatory domain could partially block Ste20-induced p42 MAPK activation, we tested whether Ste11 was capable of inducing activation of p42 MAPK in Xenopus oocyte extracts. A full-length GST-Ste11 fusion protein (GSte11) was expressed in yeast and purified from cell lysates (see "Materials and Methods"). As shown in Fig. 3B, GSte11 induced p42 MAPK activation when added to Xenopus oocyte extracts. While a Ste11 homolog has not yet been identified in X. laevis, several MAPK and MAPKK homologs have been isolated from this organism (32)(33)(34). Our results suggest the existence of a MAPK cascade in Xenopus oocytes that strongly resembles the mating pheromone-responsive MAPK cascades found in yeasts.
A mammalian Ste20/Shk1-related protein kinase, Pak1, was previously purified as a Cdc42/Rac1-binding protein (23). We tested the ability of Pak1 to activate p42 MAPK in Xenopus oocyte extracts. The addition of a constitutively activated Pak1⌬N protein (residues 232-544) to Xenopus oocyte extracts failed to induce activation of p42 MAPK (Fig. 4). This result demonstrates that the p42 MAPK cascade induced by Ste20 and Shk1 in Xenopus oocytes is selectively activated by a limited subset of PAKs. To further examine the relationship between PAKs and the activation of MAPK cascades, we determined whether PAKs induce activation of a JNK/SAPK cascade in Xenopus oocytes. JNK/SAPKs comprise a subfamily of stressactivated MAPKs distinct from the p42/p44 MAPK subfamily (35). As shown in Fig. 5, Pak1⌬N, as well as the yeast Ste20 and Shk1 kinases, each induced activation of JNK/SAPK in Xenopus oocyte extracts. MEKK, a potent activator of JNK/ SAPK in mammalian systems, also induced activation of JNK/ SAPK in Xenopus oocyte extracts (Fig. 5). In contrast, JNK/ SAPK activation was only weakly induced by H-Ras protein (Fig. 5). DISCUSSION We have shown that yeast PAKs-Ste20 from S. cerevisiae and Shk1 from S. pombe trigger the activation of p42 MAPK in cellfree extracts of X. laevis oocytes. In contrast, we found that the mammalian kinase Pak1, a Ste20/Shk1-related protein originally isolated as a Cdc42/Rac1-binding protein (23), did not induce p42 MAPK activation in Xenopus oocyte extracts. While not capable of inducing activation of p42 MAPK , mammalian Pak1 did induce activation of JNK/SAPK, as did the yeast Ste20 and Shk1 kinases. Our results demonstrate that PAKresponsive MAPK cascades are conserved in evolution and suggest that in higher organisms distinct MAPK modules might be regulated by different types of PAKs. In particular, our results suggest that protein kinases more closely related to Ste20 and Shk1 than Pak1 might regulate p42 MAPK signaling Samples were resolved by electrophoresis in 4 -20% gradient SDSpolyacrylamide gels, then exposed to film for autoradiography. Duplicate samples were blotted to nitrocellulose and probed with an anti-GST antibody (Santa Cruz) and developed using ECL. JNK/SAPK activation is expressed as the ratio of c-Jun⌬C phosphorylation relative to c-Jun⌬C protein present as determined by scanning densitometry. All values are expressed relative to GST, which was normalized to a value of 1.0. FIG. 2. Ste20⌬N-induced p42 MAPK activation is dependent on Ste20 catalytic function. Fusion proteins were purified from yeast cell extracts as described under "Materials and Methods." Kinase assays (left side of panel) were performed as described in Fig. 1A using GSte20⌬N or GSte20⌬N[K649R] fusion proteins. The ability of GSte20⌬N or GSte20⌬N[K649R] fusion proteins to induce p42 MAPK activation in Xenopus oocyte extracts (right side of panel) was measured as described in Fig. 1B.   FIG. 3. Effect of S. cerevisiae Ste11 and Ste11⌬C proteins on p42 MAPK activation. A, Xenopus oocyte extracts were incubated with 2 ng of GSte20⌬N glutathione-agarose beads and either 32 ng of GSte11⌬C glutathione-agarose beads or 8 l of glutathione-agarose beads alone. p42 MAPK activation was measured as described in the legend to Fig. 1 and is expressed as the percentage of p42 MAPK in activated form (p42 MAPK activated/total p42 MAPK ) as determined by densitometry. B, full-length GST-Ste11 fusion protein was purified from yeast extracts using glutathione-agarose, and 400 ng of the protein was assayed for its ability to induce p42 MAPK activation in Xenopus oocyte extracts as described in the legend to Fig. 1.   FIG. 4. The mammalian Pak1 kinase does not induce activation of p42 MAPK in Xenopus oocyte extracts. GST-PAK1⌬N (GPak1⌬N) fusion protein was expressed and purified from bacterial extracts as described under "Materials and Methods." Xenopus oocyte extracts were incubated with either 1 g of GPak1⌬N or 20 ng of GSte20⌬N, and p42 MAPK activation was measured as described in the legend to Fig. 1. pathways in higher eukaryotes. We have, in fact, cloned rat cDNA sequences corresponding to six distinct Ste20/Shk1-related protein kinases, including Pak1 and Pak2. 4 One of these encodes a protein kinase that is more closely related to Ste20 and Shk1 than Pak1 or Pak2. It is possible that through evolution, distinct PAKs have evolved to regulate distinct MAPK pathways. Given their capacity to activate both p42 MAPK and JNK/SAPK pathways, it can be speculated that the yeast Ste20 and Shk1 kinases represent ancestral forms of PAKs.
While this manuscript was in preparation, other investigators provided evidence that in mammalian cells Cdc42 and Rac1 induce activation of JNK/SAPKs (36 -38). In one of these studies (38), it was shown that the N-terminal regulatory domain of Pak1 blocked Cdc42 and Rac1-induced JNK/SAPK activation. However, since Pak1 binds directly to Cdc42 and Rac1 (23), it could not be concluded from this experiment whether Pak1 itself mediated Cdc42/Rac1-induced activation of JNK/SAPK. Our results demonstrate that Pak1 does indeed signal to the JNK/SAPK pathway. It is likely, therefore, that Pak1 represents a link between Cdc42/Rac1 and JNK/SAPK cascades.
It remains to be determined exactly how PAKs fit into the regulatory networks that lead to the activation of MAPK modules. In fission yeast, Shk1 and Cdc42 are components of a Ras-dependent signaling complex that regulates both cytoskeletal-dependent cellular morphology and a MAPK module (20). We have recently obtained genetic data suggesting that Shk1 acts upstream of the MAPKKK, Byr2. 2 Genetic data also indicate that in S. cerevisiae Ste20 functions upstream of Ste11 (19), which is both structurally and functionally related to Byr2 (5). Our results are consistent with biochemical data from others showing that Ste20 immune complexes are capable of phosphorylating Ste11 in vitro (21). It is possible that Ste20 and Shk1 are MAPKKK kinases, although direct demonstration of this will require the use of purified proteins.
Recent studies have shown that in mammalian cells, as in fission yeast, Cdc42 proteins participate in the regulation of cytoskeletal-dependent cell morphology (39,40). The structurally related small G proteins, Rac and Rho, also control cytoskeletal organization (41,42). It is not yet known whether Cdc42-activated protein kinases related to Ste20 and Shk1 participate in cytoskeletal regulation in mammalian cells. However, we have obtained results consistent with this possibility. We have found that expression of the N-terminal regulatory domain of Pak2 alters cytoskeletal organization in Swiss 3T3 fibroblasts and disrupts normal morphology in fission yeast. 4 Thus, it appears likely that in both yeasts and higher eukaryotes, Ste20/Shk1-related protein kinases participate both in cytoskeletal regulation and in MAPK signaling pathways. It remains to be determined whether there is a link between MAPK signaling and control of the actin cytoskeleton. Further study and characterization of PAKs using metazoan cell systems in combination with yeast genetic and molecular biological approaches should provide a better understanding of the roles these types of protein kinases play in eukaryotic signal transduction pathways.