The Novel Rho GTPase-activating Protein Family Protein, Rga8, Provides a Potential Link between Cdc42/p21-activated Kinase and Rho Signaling Pathways in the Fission Yeast, Schizosaccharomyces pombe

The PAK family kinase, Shk1, is an essential regulator of polarized growth in the fission yeast, Schizosaccharomyces pombe . Here we describe the characterization of a novel member of the RhoGAP family, Rga8, identified from a two-hybrid screen for proteins that interact with the Shk1 kinase domain. Although deletion of the rga8 gene in wild type S. pombe cells results in no obvious phenotypic defects under normal growth conditions, it partially suppresses the cold-sensitive growth and morphological defects of S. pombe cells carrying a hypomorphic allele of the shk1 gene. By contrast, overexpression of rga8 is lethal to shk1 -defective cells and causes morphological and cytokinesis defects in wild type S. pombe cells. Consistent with a role for Rga8 as a downstream target of Shk1, we show that the Rga8 protein is directly phosphorylated by Shk1 in vitro and phosphorylated in a Shk1-dependent fashion in S. pombe cells. Fluorescence photomicroscopy of the GFP-Rga8 fusion protein indicates that Rga8 is localized to the cell ends during interphase and to the septum-forming region during cytokinesis. In S. pombe cells carrying the orb2–34 allele of shk1 , Rga8 exhibits a monopolar pattern of localization, providing

p21-activated kinases (PAKs) 1 comprise a family of highly conserved serine/threonine kinases that bind to GTP-bound forms of the Rho-type p21 GTPases, Cdc42 and Rac (reviewed in Refs. 1 and 2). In some cases, this interaction leads to the stimulation of PAK kinase activity, the characteristic for which this family of protein kinases derives its name. PAKs have been implicated in the regulation of diverse processes in eukaryotic organisms, including growth factor-induced signaling pathways (3)(4)(5)(6), cytoskeletal organization (7)(8)(9), cell morphology and motility (4, 7, 10 -12), cell cycle control (13)(14)(15), apoptosis (13,16), and neurological function (17). Although a number of different PAK substrates have been identified in mammalian cells, their contributions and, in some instances, biological relevance to the diverse biological functions of the PAK kinases remain to be clarified. The rod-shaped fission yeast, Schizosaccharomyces pombe, possesses two genes, shk1 (also known as pak1 and orb2) (4,7,15) and shk2 (also known as pak2), encoding members of the PAK kinase family (18,19). shk1 is an essential gene required for normal cytoskeletal organization, polarized growth and morphology, proper control of cell cycle progression, completion of cytokinesis, and normal mating response of S. pombe cells (4, 7, 14, 20 -23), whereas shk2 is a nonessential gene that appears to be largely redundant in function with shk1 (18,19). Genetic and molecular studies indicate that Shk1 is an effector of the single Cdc42 GTPase homolog in S. pombe (4,7,18). Like Shk1, Cdc42 is essential for viability, polarized growth, and normal mating response of S. pombe cells (4,24). Cdc42 and Shk1 are components of a multiprotein complex that functions downstream of Ras1, the single Ras GTPase homolog in S. pombe (4,7,25,26). Like Cdc42 and Shk1, Ras1 participates in the regulation of cell morphology and mating in S. pombe, but unlike Cdc42 and Shk1, it is not essential for cell viability (27,28). Ras1 interacts with Cdc42 via complex formation with the presumptive Cdc42 guanine nucleotide exchange factor, Scd1 (26). Like Ras1, Scd1 is required for normal cell shape and mating, but not viability of S. pombe cells.
In addition to Cdc42, three other nonessential proteins, Scd2 (21,26), Skb1 (14,25), and Skb5 (29), positively modulate Shk1 function in S. pombe cells. Scd2 interacts directly with Scd1, Cdc42, and Shk1, and appears to function as a scaffold that positively modulates protein-protein interactions between Scd1 and Cdc42 and between Cdc42 and Shk1 (4,26). scd2⌬ cells, like ras1⌬ and scd1⌬ mutants, are viable but ovoid in shape and mating-defective. Skb1 and Skb5 are apparently specialized regulators of Shk1, as they are required for proper control of cell polarity under conditions of high osmolarity but not under normal growth conditions (29,30). Shk1 is negatively regulated by an essential and highly conserved WD repeat protein, Skb15 (20).
Although substantial insights have been obtained relating to the molecular regulation of Shk1, relatively little is known about its downstream effector pathways. Genetic data suggest * This work was supported by research grants from the National Institutes of Health (Grant R01GM53239) and the University of Texas M. D. Anderson Cancer Center. 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) AY375444.
␤-Galactosidase Filter Assay for Two-hybrid Interactions-The filter assay for testing two-hybrid interactions was performed as described previously (42).
Co-precipitation Experiments-Co-precipitation experiments were performed essentially as described (29). S. pombe wild type cells co-transformed with pART1CMShk1⌬N118 together with either pAAUGST (for expression of GST) or pAAUGSTRga8 (for expression of GST-Rga8) were grown to early log phase. Cell lysates were prepared using glass beads as described previously (29). GST complexes were purified from cell lysates (1 mg of total protein) using glutathioneagarose beads (Amersham Biosciences) and resolved by SDS-PAGE and subsequent immunoblot analyses.
In Vitro Kinase Assays and in Vivo Labeling Experiments-In vitro kinase assays were performed with bacterially purified proteins as described (29). Approximately 13.5 g of GST or 0.1 g of GST-Rga8 were incubated with ϳ20 ng of His 6 -Shk1 or 1 g of His 6 -Ras in a total volume of 25 l of kinase buffer and incubated at 30°C for 20 min. Reactions were terminated by adding 25 l of 2ϫ sample buffer and boiling for 5 min, then resolved by SDS-PAGE and subsequent autoradiography.
In vivo P-32 labeling experiments for the analysis of Rga8 phosphorylation in S. pombe cells were performed essentially as described previously (46). S. pombe cells were cultured in phosphate-free EMM supplemented with 1 mM phosphate until they reached early log phase. Cells were then resuspended at 2 ϫ 10 6 cells/ml in fresh phosphate-free EMM medium supplemented with 50 M phosphate. Five milliliters of cells was incubated with 1 mCi of [ 32 P]orthophosphate (10 4 Ci/mmol) (NEX054, PerkinElmer Life Sciences) for 4 h and then mixed with 50 ml of unlabeled carrier cells. Cells were harvested and lysed in yeast lysis buffer as described previously (29). An equal amount of lysates was incubated with 100 l of protein A beads and 10 l of anti-HA antibody 12CA5 for 2 h at 4°C. Immune complexes were divided into two halves and resolved by 6% SDS-PAGE. One gel was dried for detection of 32 P-labeled proteins by autoradiography, and the other was subjected to immunoblot analyses for detection of HARga8 protein.
Subcellular Localization of GFP-Rga8 -S. pombe strains transformed with pREP41GFPRga8, for expression of GFP-Rga8, were grown on EMM plates containing 50 M thiamine to repress expression of GFP-Rga8. Fresh cultures were inoculated into EMM lacking thiamine to induce derepression of GFP-Rga8 expression. Cultures were monitored for the first detectable GFP signals (typically about 16 h at 25°C) and subjected to fluorescence photomicroscopy. All cultures were incubated at 25°C. cdc10 -129 and cdc25-22 cultures were shifted to 36°C for 4 h prior to photomicroscopy.

RESULTS
Cloning and Sequence Analysis of the rga8 Gene-A previously described two-hybrid screen for proteins that interact with the Shk1 kinase domain (20) resulted in the isolation of multiple cDNAs corresponding to a novel S. pombe gene, which we have designated rga8 (GenBank TM accession number AY375444). The rga8 gene contains an uninterrupted open reading frame encoding a predicted 88-kDa protein that is highly similar in sequence and predicted structural organization to the S. cerevisiae RhoGAP, Rgd2 (E value of 8 ϫ 10 Ϫ87 ), as well as a hypothetical RhoGAP protein predicted from the genome sequence of Neurospora crassa (E value of 8 ϫ 10 Ϫ127 ) (Fig. 1A). These proteins each possess a highly homologous C-terminal RhoGAP domain (38 -44% pairwise identity), an N-terminal FCH (Fes/CIP4 homology) domain (30 -40% pairwise identity), and a DEP (Dishevelled/Egl-10/pleckstrin homology) domain (38 -48% pairwise identity) positioned between the FCH and RhoGAP domains (Fig. 1B). FCH and DEP domains are found in a variety of known signal transduction and cytoskeletal regulatory proteins (47,48). A BLAST search of the S. pombe genome data base identified a second novel gene (S. pombe cosmid sequence SPAC1952.16) encoding a predicted RhoGAP protein that is more distantly related to Rga8 (E value of 2 ϫ 10 Ϫ28 ) than either Rgd2 or the N. crassa RhoGAP. This protein, which we shall refer to as Rga9, possesses N-terminal FCH and C-terminal RhoGAP domains but lacks the DEP domain found in Rga8, Rgd2, and the Rga8related N. crassa RhoGAP (Fig. 1A). Rga8 and Rga9 exhibit little sequence similarity to the seven previously described S. pombe RhoGAP proteins, Rga1-7 (49).
Analysis of Rga8 Two-hybrid Interactions-We used the twohybrid system to test whether Rga8 forms complexes with the full-length Shk1 protein, the Shk1 regulatory domain, itself, the S. pombe Rho1 GTPase (41,50), or with other known components of the Ras1⅐Cdc42⅐Shk1 complex in S. pombe. Complex formation was detected between full-length Shk1 and Rga8 hybrid proteins (Fig. 2, A and B) and between Rga8-Rga8 hybrids (Fig. 2B), the latter finding of which suggests that Rga8 forms a homomeric complex. Rga8 did not interact detectably with the Shk1 regulatory domain ( Fig. 2A) nor did it interact with Ras1, Cdc42, Cdc42G12V (dominant active Cdc42 mutant protein), Rho1, Scd1, Scd2, Skb1, Skb5, or Skb15 (representative tests are shown in Fig. 2B). We determined further that Rga8 did not interact with the mammalian Pak2 kinase and interacted only weakly with the other S. pombe PAK, Shk2 (Fig. 2C). Cumulatively, the results of these two-hybrid exper- iments suggest that Rga8 forms complexes with itself and with Shk1 that are substantially specific in nature.
Shk1 Associates with Rga8 in S. pombe Cells-Co-precipitation experiments were performed to determine whether Rga8 and Shk1 interact in S. pombe cells. We co-expressed c-myc epitope-tagged Shk1 protein (CMShk1⌬N118) together with either glutathione S-transferase Rga8 fusion protein (GST-Rga8) or unfused GST in S. pombe cells. Shk1⌬N118 can substitute for full-length Shk1 in S. pombe and was used for these experiments because the full-length CMShk1 protein expresses poorly in S. pombe cells (29). Cells were lysed, and the supernatant fractions of the lysates were incubated with glutathione-Sepharose beads to isolate GST complexes, which were analyzed by immunoblotting. As shown in Fig. 3, CMShk1⌬N118 was found to be associated with GST-Rga8, but not with GST, providing evidence that the Shk1 and Rga8 proteins interact in S. pombe cells.
Deletion of the rga8 Gene Partially Rescues the Cold-sensitive Growth and Morphological Defects of S. pombe Cells Deficient in Shk1 Function-An rga8::ura4 cassette was constructed in which most of the N-terminal half of the rga8 protein coding sequence was deleted by the S. pombe ura4 gene ("Experimental Procedures"). The rga8::ura4 cassette was transformed into wild type S. pombe diploid strain SP870D. rga8 ϩ /rga8::ura4 heterozygotes were identified by Southern blotting, sporulated, and subjected to tetrad analysis. Most tetrads produced two viable Uraϩ colonies and two viable UraϪ colonies, demonstrating that rga8 is a nonessential gene (data not shown). We found that rga8⌬ mutants lacked significant growth or mor-phological defects under normal culturing conditions at temperatures ranging from 20°C to 35°C in either rich or minimal medium (Fig. 4, A and B). In addition, F-actin staining experiments revealed no obvious defects in actin cytoskeletal organization in rga8⌬ cells (Fig. 4C).
We next determined the effect of the rga8⌬ mutation on S. pombe cells defective in Shk1 function. To do this, we crossed the rga8⌬ strain to an S. pombe mutant, nmt1-shk1K415R, in which the shk1 gene is replaced by a sequence that expresses a kinase-deficient Shk1 mutant protein, Shk1K415R (21), from a weak allele of the nmt1 promoter (22). Two of the resulting diploid strains were sporulated and subjected to tetrad analysis. Approximately one-fourth of the progeny were rga8⌬ nmt1-shk1K415R double mutants, indicating that the rga8⌬ mutation is not lethal when combined with the shk1K415R mutation. Because the nmt1-shk1K415R mutant is cold-sensitive for growth (22), we compared the growth of rga8⌬ nmt1-shk1K415R mutants to that of the respective single mutants and wild type S. pombe cells at 20°C. Whereas nmt1-shk1K415R cells were arrested for growth at 20°C, the rga8⌬ nmt1-shk1K415R mutant was capable of growing at this temperature, although not as well as wild type S. pombe cells (Fig.  5A). We observed further that, in comparison to nmt1-shk1K415R cells, which exhibit a rounded morphology, rga8⌬ nmt1-shk1K415R cells tended to be more polarized in shape, although they were not as elongated in shape as wild type S. pombe cells (Fig. 5B). We conclude from these experiments that the rga8⌬ mutation can partially rescue the nmt1-shk1K415R mutation.
Effects of rga8 Overexpression in Wild Type and shk1-defective S. pombe Cells-The above results showing that the rga8⌬ mutation partially rescues the shk1K415 mutant suggest that Shk1 and Rga8 may have opposing functions. If this is the case,  Shk1⌬N, lamin, and Byr2) or Gal4 DNA binding domain (GBD) (Rga8) fusion proteins were tested for interaction with GAD fusions of Rga8, Scd1, Shk1, Cdc42, Cdc42G12V, Rho1, and Ras1, as indicated. L40 was used for hosting LBD interactions, and HF7c was used as the host for GBD interactions. The dark gray indicates a positive interaction (␤galactosidase reporter gene induced). C, test for interaction of Rga8 with S. pombe Shk2 and mammalian Pak2 proteins. GBD-Rga8 was tested for interaction with GAD fusions of Shk1, Shk2, and Pak2 as indicated. A faint positive signal (light gray) was detected for the GBDRga8-GADShk2 test, but not for GBDRga8-GADPak2 test.
FIG. 3. Rga8 associates with Shk1 in fission yeast. Wild type S. pombe cells expressing CMShk1⌬N118 in combination with GST or GST-Rga8 were lysed, and GST complexes were purified and resolved by SDS-PAGE followed by immunoblotting using either anti-c-myc antibody to detect bound CMShk1 (top) or anti-GST antibody to detect GST fusion proteins (bottom). CMShk1⌬N118 was associated with GST-Rga8 but not with GST. GST degradation products were detected for GST-Rga8. Such degradation is commonly observed for large GST fusion proteins expressed in S. pombe.
then it would be predicted that overexpression of rga8 might induce phenotypes similar to those resulting from loss of Shk1 function. To determine whether this is the case, we constructed the plasmid pREP1Rga8 for overexpressing the rga8 proteincoding sequence from the thiamine-repressible nmt1 promoter. pREP1Rga8 and the empty control plasmid, pREP1, were transformed into wild type and nmt1-shk1K415R cells. The resulting transformants were grown on thiamine-containing minimal medium (EMMϩthi), then subcultured twice onto thiamine-free minimal medium (EMMϪthi) to determine the effects of rga8 overexpression on cell growth and morphology. As shown in Fig. 6A, whereas overexpression of rga8 caused only a modest growth inhibitory phenotype in wild type S. pombe cells, it was lethal to nmt1-shk1K415R cells. Microscopic analyses revealed that rga8 overexpression in wild type S. pombe cells resulted in a high incidence of morphological defects (e.g. spheroidal, bottle-shaped, bent, and enlarged cells) as well as lower frequencies of cytokinesis (multiseptated cells) and cell lysis defects (Fig. 6B). Intriguingly, this phenotype bears some resemblance to that caused by overexpression of the rho1 gene in S. pombe cells (50). In nmt1-shk1K415R cells, rga8 overexpression led to a very strong cell lysis phenotype without affecting the already spheroidal morphology of this mutant (Fig.  6B). These results are consistent with a model in which Rga8 functions as a negative regulator of Shk1-dependent growth control processes in S. pombe cells.

FIG. 6. Effects of rga8 overexpression on the growth and morphology of wild type and nmt1-shk1K415R cells. A, wild type (top)
and nmt1-shk1K415R (shk1K415R) (bottom) strains transformed with pREP1 (left) or pREP1Rga8 (right) were grown on EMMϩthi, subcultured overnight onto EMMϪthi, then streaked onto EMMϪthi and incubated at 30°C. Wild type cultures were incubated for 4 days and nmt1-shk1K415R cultures for 5 days prior to scanning the plates. rga8 overexpression was lethal to nmt1-shk1K415R cells, but only modestly inhibitory to the growth of wild type cells. B, wild type (top panels) and nmt1-shk1K415R (shk1K415R) (bottom panels) strains transformed with pREP1 (left) or pREP1Rga8 (right) were grown on EMMϩthi, then subcultured into EMMϪthi liquid at low density and incubated at 30°C for 2 days, with subculturing performed as necessary when cultures reached mid-log phase. rga8 overexpression caused a high incidence of aberrant morphology in wild type S. pombe cells, as well as a low incidence of cytokinesis defects (multiseptated cells) (inset panel of upper right panel). A higher incidence of cell lysis was also detected in wild type cultures overexpressing rga8 than in control cultures. rga8 overexpression caused a very high incidence of cell lysis in nmt1-shk1K415R cultures. incubated together with recombinant polyhistidine-tagged Shk1 protein (His 6 -Shk1) or recombinant His 6 -Ras protein in kinase reactions in vitro. GST-Rga8 was found to be phosphorylated in kinase reactions containing His 6 -Shk1 but not in reactions lacking His 6 -Shk1 (Fig. 7A). In control assays, GST was not phosphorylated by Shk1, as we have shown in previously published studies (23, 29) (data not shown). The level of Shk1 kinase activity, as measured by autophosphorylation, was not markedly affected by GST-Rga8 (Fig. 7A). These results demonstrate that Rga8 is a direct in vitro substrate of the Shk1 kinase.
Rga8 Is Phosphorylated in S. pombe Cells in a Shk1-dependent Fashion-Experiments were next performed to determine whether Rga8 is also phosphorylated in S. pombe cells and, if so, whether its phosphorylation is dependent on Shk1. The plasmid pREP4xHARga8 was constructed for expressing HA epitope-tagged Rga8 protein (HARga8) from the nmt1 promoter. Wild type and nmt1-shk1K415R S. pombe cells transformed with either pREP4xHARga8 or empty vector were metabolically labeled with [ 32 P]orthophosphate. 32 P-Labeled cells were lysed, and HA immune complexes were purified from the cell lysates and subjected to SDS-PAGE, autoradiography, and immunoblot analysis. As shown in Fig. 7B, HARga8 was phosphorylated in S. pombe cells and the degree of phosphorylation was significantly greater in wild type cells than in nmt1-shk1K415R cells. These results demonstrate that Shk1 is required for Rga8 phosphorylation in vivo and, when considered together with the above results showing that Rga8 is directly phosphorylated by Shk1 in vitro, implicate Rga8 as a Shk1 substrate in S. pombe. Subcellular Localization of Rga8 -To investigate the subcellular localization of Rga8, we constructed the plasmid, pREP41GFP-Rga8, for expressing the Rga8 protein as a fusion to green fluorescent protein (GFP-Rga8) from a weak allele of the nmt1 promoter (36). pREP41GFP-Rga8 was transformed into wild type S. pombe cells, and the resulting transformants were cultured on EMMϩthi, then subcultured into EMMϪthi and monitored microscopically to detect the first observable GFP fluorescence. We found that GFP-Rga8 was concentrated at the cell ends in interphase cells and at the septum forming region in dividing cells (Fig. 8A). Analysis of nuclear stained cells revealed that GFP-Rga8 disappears from the cell ends early in mitosis and becomes concentrated at the developing septum early after the completion of mitosis, where it remains until the completion of cell division (data not shown). In a temperature-sensitive S. pombe mutant, cdc10 -129, which at the restrictive temperature arrests in G 1 prior to the activation of bipolar growth (new end take-off, or NETO) (51), GFP-Rga8 was localized to the single growing cell end (Fig. 8B). In cdc25-22 cells, which at high temperature arrest in G 2 with both ends growing, Rga8 was predominantly localized to both cell ends at the restrictive temperature (Fig. 8C). These observations indicate that the Rga8 protein is localized to areas of cellular growth in S. pombe.
Because Shk1 also localizes to the cell ends and septumforming region of S. pombe cells (22), experiments were carried out to investigate whether Rga8 localization is regulated by Shk1. For this analysis, we utilized a shk1 mutant strain, orb2-34, which is monopolar for growth at 25°C. In contrast to the predominantly bipolar localization exhibited by GFP-Rga8 in wild type interphase cells, we found that GFP-Rga8 exhibited an exclusively monopolar pattern of localization in orb2-34 cells (Fig. 8D). These results demonstrate that Shk1 is required for bipolar localization of Rga8. GFP-Rga8 was also localized predominantly to one cell end in tea1⌬ cells (Fig. 8E), which exhibit a monopolar growth defect similar to orb2-34 mutant cells (23,66). In the "T"-shaped cells that are commonly found FIG. 7. Shk1 directly phosphorylates Rga8 in vitro and is required for Rga8 phosphorylation in S. pombe cells. A, GST (Ϸ14 g), GST-Rga8 (Ϸ0.1 g), His 6 -Shk1 (Ϸ0.02 g), and His 6 -Ras (Ϸ1 g) fusion proteins purified from bacterial cells were incubated alone or together, as indicated, in in vitro kinase reactions and resolved by SDS-PAGE (6% gels) followed by autoradiography. GST-Rga8 was phosphorylated in reactions containing His 6 -Shk1, but not in reactions containing His 6 -Ras1. Shk1 autophosphorylation was also detected in the assay, as we have shown previously (29). B, wild type and nmt1-shk1K415R S. pombe cells transformed with pREP4xHARga8, for expression of HARga8, or the empty vector, pSLF173 (wild type cells only), were labeled with [P 32 ]orthophosphate and lysed. HA immune complexes were isolated from the cell lysates by immunoprecipitation. Half of each immune complex was resolved by SDS-PAGE followed by autoradiography (top panel), whereas the other half was subjected to SDS-PAGE and immunoblotting to detect HARga8 (bottom panel). The degree of 32 P incorporation into HARga8 was significantly greater in wild type cells than in nmt1-shk1K415R cells. in tea1⌬ cultures (33), GFP-Rga8 was usually localized to the mid-body growth tip, but not to the normal cell ends (see cell marked by the arrow in Fig. 8E). We conclude that Tea1, like Shk1, is required for bipolar localization of Rga8.
Evidence That Rga8 Contains a Rho-specific GAP Activity-The budding yeast homolog of Rga8, Rgd2, has been shown to possess in vitro GAP activity for both Cdc42 and Rho GTPases (52). However, the in vivo specificity of Rgd2 GAP activity has not been described. As a means of assessing whether Rga8 may function as a GAP for Cdc42 or Rho GTPases in S. pombe cells, we expressed the S. pombe Cdc42 and Rho1 GTPases as epitope-tagged proteins in S. pombe cells either alone or in combination with overexpression of rga8. Rho1 was selected for this analysis because the rho1 gene, like the shk1 and cdc42 genes, is essential for viability and for proper regulation of actin cytoskeletal organization and polarized cell growth in S. pombe (41,50,53) and because overexpression of rga8 causes phenotypes similar to those resulting from rho1 overexpression (see above). Cells expressing c-myc-tagged Rho1 (CMRho1) or HA-tagged Cdc42 (HACdc42) were lysed, and the lysates were incubated with agarose beads bound to GST fusion proteins that bind selectively to activated (GTP-bound) forms of Cdc42 (GSTShk1-(145-380)) or Rho (ROCK-(831-1010); "Experimental Procedures"). As shown in Fig. 9A, although cells that overexpressed rga8 together with CMRho1 expressed a level of total CMRho1 protein similar to that detected in cells expressing only endogenous Rga8, they had a significantly reduced level of CMRho1-GTP. By contrast, rga8 overexpression did not affect Cdc42-GTP levels in S. pombe cells (Fig. 9B). These results provide evidence that Rga8 has Rho-specific GAP function in vivo, and, specifically, that it may function as a GAP for the S. pombe Rho1 GTPase.
Evidence for Positive Functional Interaction between Rga8 and Rho1 in S. pombe Cells-Given the above results suggesting that Rga8 may function as a GAP for the Rho1 GTPase in S. pombe, we sought to determine whether Rga8 interacts functionally with Rho1 in S. pombe cells. Overexpression of the wild type rho1 gene has been shown to cause partial inhibition of cell growth, aberrant morphology, and cytokinesis defects (multiseptated cells) in wild type S. pombe cells (41,50), whereas expression of a dominant negative allele of rho1, rho1T20N, causes a lethal, shrunken cell phenotype (41). Genetic experiments were performed to examine whether overexpression of rga8 affects the phenotypes induced by overexpression of wild type and dominant negative rho1 genes. To determine whether overexpression of rga8 affects the growth inhibitory phenotype induced by expression of rho1T20N, wild type S. pombe cells were transformed with pREP1Rho1T20N, which allows for thiamine-repressible expression of rho1T20N, together with either an empty vector (pAAUCM) or with a plasmid from which rga8 is overexpressed as a c-myc epitopetagged protein (CMRga8) from the strong adh1 promoter (pAAUCMRga8). The transformants were streaked onto EMM plates without thiamine to derepress rho1T20N expression and on EMM plates containing either 0.01 or 0.1 M thiamine to allow for partial repression of rho1T20N expression and incubated at 30°C for 4 days. As summarized in Table I, nmt1 promoter-driven overexpression of rho1T20N resulted in complete inhibition of cell growth on EMM lacking thiamine, and this growth inhibitory phenotype was attenuated with increasing concentrations of thiamine. We found that the severity of growth inhibition caused by rho1T20N was unaffected by rga8 overexpression both on media with thiamine and on media without thiamine. Thus, despite possessing an apparent Rho1 GAP activity, overproduction of Rga8 does not exacerbate the cytotoxic phenotype induced by expression of the Rho1T20N protein.
We next determined the effects of rga8 overexpression on phenotypes induced by overexpression of the wild type rho1 gene. As shown in Table II, we found that S. pombe wild type cells that co-overexpressed rga8 and rho1 together grew more slowly than transformants that overexpressed rga8 or rho1 alone. Microscopic analyses revealed that the frequencies of abnormally shaped and multiseptated cells were also higher in cultures that co-overexpressed rga8 and rho1 than in cultures overexpressing either gene alone (data not shown). Thus, gain of Rga8 function exacerbates phenotypes caused by gain of Rho1 function and vice versa, suggesting the possibility that Rho1 and Rga8 may have a positive functional interaction in S. pombe cells.
Overexpression of rho1 Is Lethal to Shk1-defective S. pombe Cells-If, as suggested by the above results, Rga8 is a potentiator rather than negative regulator of Rho1 function in S. pombe cells, then we would predict that overexpression of rho1, like overexpression of rga8, might be inhibitory to the growth of S. pombe cells deficient in Shk1 function. To test  Ϫ ϩ ϩ ϩ rho1T20N and rga8 Ϫ ϩ ϩ ϩ a The S. pombe host strain used for this experiment was SP870. rho1T20N was overexpressed from the plasmid pREP1Rho1T20N. rga8 was overexpressed from the plasmid pAAUCMRga8. Cells overexpressing only rho1T20N were co-transformed with empty pAAUCM plasmid, and cells overexpressing only rga8 were co-transformed with empty pREP1 plasmid so that the growth of all transformants could be compared on the same EMM medium.
b Cultures were grown on EMMϩthi, then subcultured onto EMM without thiamine or with the indicated concentrations of thiamine. Relative growth rates at 30°C are indicated by plus (ϩ) symbols.
The RhoGAP, Rga8, Is a Potential PAK Target in S. pombe whether this is the case, wild type and nmt1-shk1K415R cells were transformed with the plasmid pREP1Rho1, which allows for thiamine-repressible expression of rho1 from the nmt1 promoter, or the empty vector, pREP1, and analyzed for growth on thiamine-free EMM. As shown in Fig. 10A, whereas overexpression of rho1 had only a modest inhibitory effect on the growth of wild type S. pombe cells, it was lethal to the nmt1-shk1K415R mutant. To rule out the possibility that the strong growth inhibitory phenotype caused by overexpression of wild type rho1 was due to high levels of Rho1-GDP, analogous experiments were performed to determine the effects of overexpression of a dominant active, GTPase-defective allele of rho1, rho1G15V, in wild type and nmt1-shk1K415R cells. Because rho1G15V is highly inhibitory to the growth of S. pombe cells when expressed at high levels (41), we expressed this sequence using the plasmid pREP41 (36), which contains an attenuated allele of the nmt1 promoter and derepressed its expression on media containing a low concentration of thiamine (10 nM). Under these conditions, we observed that, similar to overexpression of wild type rho1, rho1G15V expression partially inhibited the growth of wild type S. pombe cells and completely inhibited growth of the nmt1-shk1K415R mutant (Fig. 10B). Taken together, these results suggest that gain of Rho1 function, like gain of Rga8 function, is inhibitory to the growth of S. pombe cells deficient in Shk1 function, thereby providing support for a model in which Rga8 and Rho1 interact in a positive fashion. DISCUSSION In this report, we have provided genetic and molecular evidence for functional interaction between the fission yeast PAK family kinase, Shk1, and a novel member of the RhoGAP protein family, Rga8. Deletion of the rga8 gene, although causing no obvious phenotypic defects under normal growth conditions, partially rescues the cold-sensitive growth and morphological defects of S. pombe cells carrying a hypomorphic allele of the shk1 gene. This finding suggests that Rga8 may play a role antagonistic to that of the Shk1 kinase. Consistent with this possibility, we found that overexpression of rga8 is lethal to shk1-defective cells and causes defects in morphology and cytokinesis in wild type S. pombe cells. Rga8 was found to be directly phosphorylated by the Shk1 kinase in vitro and was phosphorylated in a Shk1-dependent fashion in S. pombe cells. GFP-Rga8 fusion proteins localize to the cell ends and septumforming region of S. pombe cells, a pattern of subcellular localization that overlaps with that of Shk1. In a shk1 mutant strain, orb2-34, which is defective in the activation of bipolar growth, GFP-Rga8 was localized to only one cell end during interphase, indicating that Shk1 is required for bipolar localization of Rga8. Taken together, our results suggest that Rga8 is a Shk1 substrate that functions as a negative regulator of Shk1-mediated growth control pathway(s) in S. pombe cells. Our finding that rga8 overexpression is lethal to S. pombe cells expressing a kinase-deficient Shk1 protein (Shk1K415R), but only modestly inhibitory to wild type S. pombe cells, suggest that Rga8 may be negatively regulated by Shk1 phosphorylation.
Rga8 is one of at least nine RhoGAP family proteins encoded by the S. pombe genome. Seven of these genes, rga1-7, have been previously described (49,54). Like rga8, none of the previously described RhoGAP genes is essential for cell viability, although deletion of one gene, rga1, causes a slow growth defect and severe morphological abnormalities (49). Rga8 shares no significant sequence similarity with the predicted Rga1-7 proteins (pairwise E values in all cases Ͼ0.1) but exhibits a modest degree of homology (E value of 2 ϫ 10 Ϫ28 ) to a previously uncharacterized S. pombe RhoGAP gene, rga9, which we identified from a BLAST search of the S. pombe genome data base. We have found that rga9, like rga8, is a nonessential gene and that simultaneous disruption of both genes results in no obvious phenotypes under normal growth conditions. 2 The lack of phenotypic defects resulting from deletion of rga8 and rga9 may be an indication that they share overlapping functions with one or more of the previously described rga RhoGAP genes, despite the lack of significant sequence homology be-2 Y. Qyang, unpublished observations. None ϩϩϩ rho1 ϩϩ rga8 ϩϩ rho1 and rga8 ϩ a The S. pombe strain used for this experiment was SP870. rho1 was overexpressed from the plasmid pAAUCMRho1. rga8 was overexpressed from the plasmid pREP1Rga8. Cells overexpressing only rho1 were co-transformed with empty pREP1 plasmid, and cells overexpressing only rga8 were co-transformed with empty pAAUCM plasmid so that the growth of all transformants could be compared on the same EMM medium.
b Cultures were grown on EMMϩthi, then subcultured twice on EMMϪthi to derepress Rga8 expression. Relative growth rates 30°C are indicated by plus (ϩ) symbols.
FIG . 10. rho1 overexpression is lethal to shk1-defective cells. A, wild type and nmt1-shk1K415R cells transformed with pREP1 or pREP1Rho1 (pRho1) were grown on EMMϩthi, then streaked onto EMMϪthi and incubated for 5 days at 30°C prior to photographing the plates. B, wild type and nmt1-shk1K415R cells transformed with pREP41 or pREP41Rho1G15V (pRho1G15V) were grown on EMMϩthi, subcultured onto EMM containing 10 nM thiamine, then streaked onto EMM containing 10 nM thiamine and incubated for 5 days at 30°C prior to photographing the plates.
The RhoGAP, Rga8, Is a Potential PAK Target in S. pombe tween the proteins encoded by the respective genes.
Two genes encoding Rho GTPase homologs, rho1 and rho2, have been described in S. pombe (41,50,55). Similar to shk1, rho1 is an essential gene required for normal actin cytoskeletal organization, cell polarity, and cell integrity (41,50,53). rho2, although not essential for cell viability, also contributes to the regulation of these processes (55). Given the apparent similarities of phenotypes resulting from shk1 and rho1 null mutations, as well as the similar phenotypes caused by overexpression of the rga8 and rho1 genes, we investigated whether the Shk1 and Rga8 proteins interact functionally with Rho1. Consistent with its predicted structure, we obtained molecular evidence that Rga8 functions as a Rho1GAP in S. pombe cells. However, despite possessing an apparent RhoGAP activity, the results of our genetic analyses suggest that Rga8 may potentiate, rather than antagonize, Rho1 function in S. pombe cells and that Rho1, like Rga8, has functions that are antagonistic to those of the Shk1 pathway. One model suggested by our genetic data is that Rga8 is a downstream effector of Rho1. If this model is correct, then Rga8 likely functions as a point of convergence for PAK and Rho signaling pathways in S. pombe cells. This is an intriguing possibility that may be of relevance to our understanding of mammalian PAK function, given the fact that antagonistic interactions between Cdc42/Rac/PAK and Rho pathways have been well documented in mammalian cells (56 -60).
There is ample precedent for considering a possible role for Rga8 as an effector of Rho GTPase function in S. pombe. Roles for GTPase-activating proteins as effectors of p21 GTPase function were originally suggested from studies conducted more than a decade ago in which it was shown that RasGAP interacts with the effector binding domain of Ras (61) and that Ras inhibits muscarinic receptor-induced opening of potassium channels in a RasGAP-dependent fashion (62). Evidence that some mammalian RhoGAPs may function as effectors of Rho GTPases has been provided from studies on the mammalian RhoGAP family proteins p85 (a regulatory subunit of phosphatidylinositol 3-kinase) and n-chimaerin (63,64). Indeed, n-chimaerin, despite having a GAP activity specific for the Rho family GTPases Cdc42 and Rac, appears to function cooperatively with Cdc42 and Rac to induce the formation of lamellipodia and filopodia in mammalian cells (64). The idea that RhoGAPs could serve as both convergent and divergent points in Rho GTPase-mediated signaling pathways is further supported by structural features of RhoGAP family proteins, most of which contain one or more functional motifs commonly found in signal transduction and/or cytoskeletal regulatory factors, including DEP, FCH, PH, and SH3 domains (65). Future studies on the mechanisms of Rga8 regulation and function may provide insights into underlying mechanisms of regulation and function for RhoGAP family proteins in mammalian cells. It will likewise be of substantial interest to investigate whether RhoGAP family proteins are targets of PAK kinases in mammalian cells.
We recently provided evidence that the cell polarity regulator, Tea1, is a potential downstream substrate-effector of Shk1 in S. pombe cells (23). Unlike Rga8, which appears to function as a negative regulator of Shk1-mediated growth control pathway(s) in S. pombe, Tea1 appears to be a positive effector of Shk1. Although both Rga8 and Tea1 appear to be direct substrate targets of Shk1, we nevertheless found that Tea1 is required for bipolar localization of Rga8 during interphase growth of S. pombe cells. This finding supports our previously proposed idea that interactions among components of the Shk1 complex may be very dynamic in nature (23). Indeed, it may be that some components, such as Tea1 and Rga8, cannot be strictly defined as having "upstream" and "downstream" functions, per se, in Shk1-mediated growth control processes. Rather, Shk1 may be a central regulatory factor of a macromolecular complex comprised of at least some elements that interact in dynamic and, in some cases, phosphoregulated fashions with multiple components of the complex. Such interactions would allow for a dynamic and complex regulation of Shk1 complex function, which might be required for efficient regulation and coordination of dynamic processes involved in regulating cytoskeletal organization, plasma membrane synthesis and turnover, and cell wall synthesis and degradation required for polarized cell growth and cell division in S. pombe. Further characterization of the underlying mechanisms of Shk1 function in S. pombe may provide insights into what are likely to be complex mechanisms of PAK function in regulating cellular morphogenesis and motility in higher eukaryotes.