Portal de Periódicos da CAPES

The Cdc42p GTPase and Its Regulators Nrf1p and Scd1p Are Involved in Endocytic Trafficking in the Fission YeastSchizosaccharomyces pombe

2001; Elsevier BV; Volume: 276; Issue: 5 Linguagem: Inglês

10.1074/jbc.m007389200

1083-351X

Janet Murray, Douglas I. Johnson,

Cellular transport and secretion

Nrf1p was first identified in a screen for negative regulators of the Cdc42p GTPase. Overexpression of Nrf1p resulted in dose-dependent lethality, with cells exhibiting an ellipsoidal morphology and abnormal vacuolar phenotypes including an increase in vacuolar fusion. Green fluorescent protein (GFP)-Cdc42p and GFP-Nrf1p colocalized to vacuolar membranes and GFP-Nrf1p vacuolar localization depended on Scd1p, the Schizosaccharomyces pombe homolog of the Cdc24p guanine nucleotide exchange factor. In this study, site-directed mutagenesis was conducted on Nrf1p to determine its functional domains. Mutations in the three putative transmembrane domains resulted in mislocalization of GFP-Nrf1p and an inability to induce lethality, suggesting a loss of function. Mutations in the second extramembranous loop of Nrf1p also resulted in a loss of function and altered the ability of GFP-Nrf1p to localize to vacuolar membranes. Analysis of Δnrf1 and Δscd1 mutants revealed defects in endocytosis. In addition, overexpression of constitutively active Cdc42G12Vp resulted in an increase in endocytosis and an ability to rescue the endocytic defects in Δnrf1 and Δscd1 cells. These data are consistent with Nrf1p and Scd1p being necessary for efficient endocytosis, possibly through the regulation of Cdc42p. Nrf1p was first identified in a screen for negative regulators of the Cdc42p GTPase. Overexpression of Nrf1p resulted in dose-dependent lethality, with cells exhibiting an ellipsoidal morphology and abnormal vacuolar phenotypes including an increase in vacuolar fusion. Green fluorescent protein (GFP)-Cdc42p and GFP-Nrf1p colocalized to vacuolar membranes and GFP-Nrf1p vacuolar localization depended on Scd1p, the Schizosaccharomyces pombe homolog of the Cdc24p guanine nucleotide exchange factor. In this study, site-directed mutagenesis was conducted on Nrf1p to determine its functional domains. Mutations in the three putative transmembrane domains resulted in mislocalization of GFP-Nrf1p and an inability to induce lethality, suggesting a loss of function. Mutations in the second extramembranous loop of Nrf1p also resulted in a loss of function and altered the ability of GFP-Nrf1p to localize to vacuolar membranes. Analysis of Δnrf1 and Δscd1 mutants revealed defects in endocytosis. In addition, overexpression of constitutively active Cdc42G12Vp resulted in an increase in endocytosis and an ability to rescue the endocytic defects in Δnrf1 and Δscd1 cells. These data are consistent with Nrf1p and Scd1p being necessary for efficient endocytosis, possibly through the regulation of Cdc42p. green fluorescent protein yeast extract and supplements Edinburgh minimal media 2′,7′-dichlorofluorescein diacetate vacuolar-ATPase subunit thiamine Cdc42p is a Rho-like GTPase that is ubiquitously expressed in eukaryotes and has been implicated in many cellular processes including regulation of cellular polarity, transcriptional activation, and phagocytosis of bacteria into mammalian cells (1Johnson D.I. Microbiol. Mol. Biol. Rev. 1999; 63: 54-105Crossref PubMed Google Scholar). Cdc42p acts as a binary switch, active in the GTP-bound state and inactive in the GDP-bound state. The nucleotide-bound state is regulated by guanine-nucleotide exchange factors that mediate the exchange of GDP for GTP, thereby activating Cdc42p and GTPase-activating proteins, which enhance the intrinsic GTPase activity of Cdc42p, thereby inactivating the protein (1Johnson D.I. Microbiol. Mol. Biol. Rev. 1999; 63: 54-105Crossref PubMed Google Scholar). In fission yeast Schizosaccharomyces pombe, Cdc42p is essential and regulates directed growth, with the Δcdc42 allele exhibiting a phenotype of small, round cells and activated alleles conferring a phenotype of large, round cells (2Miller P. Johnson D.I. Mol. Cell. Biol. 1994; 14: 1075-1083Crossref PubMed Scopus (172) Google Scholar). Scd1p, a putative guanine-nucleotide exchange factor for S. pombe Cdc42p, shares amino acid identity with the Dbl family of Rho-type guanine-nucleotide exchange factors including the Saccharomyces cerevisiae guanine-nucleotide exchange factor Cdc24p. A Δscd1 mutant is viable and exhibits a round cell morphology, suggesting that Scd1p plays a role in directed cell growth (3Chang E.C. Barr M. Wang Y. Jung V. Xu H.-P. Wigler M.H. Cell. 1994; 79: 131-141Abstract Full Text PDF PubMed Scopus (244) Google Scholar). S. pombe Nrf1p is a 122-amino acid protein with three putative transmembrane domains that was previously identified in a screen for negative regulators of Cdc42p (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar). A Δnrf1 mutant was viable; however, high level expression of Nrf1p was lethal, resulting in an ellipsoidal morphology and abnormal vacuolar phenotypes including vacuolar coalescence around the nucleus and subsequent fusion of the vacuoles. In S. pombe, vacuoles are numerous, with greater than 50 individual organelles sometimes present within a cell. Green fluorescent protein (GFP)1-Nrf1p and GFP-Cdc42p colocalized to the plasma membrane, nuclear membrane, septum, and vacuolar membranes. The localization of GFP-Nrf1p to the vacuolar membrane and subsequent vacuolar coalescence and fusion depended on Scd1p (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar). To characterize the functional domains of Nrf1p, site-directed mutations were generated. Mutations generated in the three putative transmembrane domains affected GFP-Nrf1p localization, consistent with the prediction that these transmembrane domains were essential for membrane anchoring. Mutations created in predicted extramembranous domains of Nrf1p had varying effects on GFP-Nrf1p localization and function, with the second loop domain of Nrf1p having a role in targeting Nrf1p to the vacuolar membrane. Studies were also conducted to further examine the role of Scd1p in Nrf1p localization to the vacuole. These studies led to the discovery that Δscd1 and Δnrf1 mutants had a defect in endocytosis, and this defect could be reversed by expression of activated Cdc42p. Together, these data suggest that Cdc42p-dependent signaling pathways play a role in endocytosis in S. pombe. S. pombe cells were grown in yeast extract and supplements (YES) complex media or in Edinburgh minimal media (EMM) lacking uracil, leucine, or both (5Moreno S. Klar A. Nurse P. Methods Enzymol. 1991; 194: 795-823Crossref PubMed Scopus (3143) Google Scholar). EMM and EMM agar were purchased from Bio101 (Vista, CA). Thiamine was added to S. pombe growth media at 5 μg/ml to repress transcription from the nmt1 promoter. The S. pombe strains are listed in Table I. Yeast transformations were performed as described previously (5Moreno S. Klar A. Nurse P. Methods Enzymol. 1991; 194: 795-823Crossref PubMed Scopus (3143) Google Scholar,6Sherman F. Fink G.R. Hicks J.B. Methods in Yeast Genetics: Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1986Google Scholar).Table IS. pombe strains usedStrainGenotypeSourceED668h+, ade6.M216, leu1.32, ura4-D18P. FantesSPSCD1Uh90, ade6.M210, leu1.32, ura4-D18, scd1::ura4+Ref. #3Chang E.C. Barr M. Wang Y. Jung V. Xu H.-P. Wigler M.H. Cell. 1994; 79: 131-141Abstract Full Text PDF PubMed Scopus (244) Google ScholarJM2h+, ade6.M216, leu1.32, ura4-D18, nrf1::ura4+Ref. #4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholarypt7−h+, ade6.M216, leu1.32, ura4-D18, ypt7::ura4+Ref. #11Bone N. Millar J.B.A. Toda T. Armstrong J. Curr. Biol. 1998; 8: 135-144Abstract Full Text Full Text PDF PubMed Scopus (116) Google ScholarTP319–13Ch−, leu1.32, ura4-D18, pmkl::ura4+Ref. #25Toda T. Dhut S. Superti-Furga G. Gotoh Y. Nishida E. Sugiura R. Kuno T. Mol. Cell. Biol. 1996; 16: 6752-6764Crossref PubMed Scopus (188) Google Scholar Open table in a new tab Enzymes, polymerase chain reaction kits, and other reagents were purchased from standard commercial sources and used as specified by the suppliers. Oligonucleotide primers were obtained from Genosys Biotechnologies Inc. (The Woodlands, Texas). Standard methods were used for recombinant DNA manipulations (7Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1989Google Scholar). The pREP3X-GFP-A8-nrf1 +construct was generated by inserting a nrf1 +fragment isolated from pREP41X-GFP-A8-nrf1 + (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar) into pREP3X-GFP-A8 (provided by A. Merla) at theNot I and Bam HI sites. Site-directed mutations (Fig. 1 A) were created in pREP41X-GFP-A8-nrf1 + using the QuikChange site-directed mutagenesis kit obtained from Stratagene (La Jolla, CA). The DNA sequence of all polymerase chain reaction products was verified through the Vermont Cancer Center DNA Sequencing Facility. Vacuolar staining was conducted using carboxy-DCFDA or FM 4-64 (Molecular Probes, Eugene OR). The cells were grown to mid-log phase under derepressing conditions (described above) and incubated for 30 min in fresh media containing 10 μmcarboxy-DCFDA or 20 μg/ml FM 4-64. These cells were then washed with media, and half of the cells were resuspended in H20 for carboxy-DCFDA staining or allowed to incubate for 30 min post-washing before visualization for FM 4-64 staining. GFP images were captured at mid-log phase after 18–28 h under derepressing conditions (described above). Digital images were captured and analyzed as described previously (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar). Total cellular protein was isolated from cells containing nmt1 promoter-driven GFP-Nrf1p fusion proteins. The cells were grown in EMMS-Leu−Thi liquid media to mid-log phase, collected, washed with H20, resuspended in 100 μl of 1× phosphate-buffered saline, and spheroplasted at 37 °C in the presence of 100 μg/ml zymolyase until a sample showing greater than 80% lysis was observed upon the addition of SDS to 0.1%. The spheroplasts were then collected and resuspended in lysis buffer (0.3m sorbitol, 140 mm NaCl, 50 mm Tris pH 8.0) with protease inhibitors (1:100 dilutions of 5 mg/ml aprotinin, 5 mg/ml leupeptin in water, 6 mg/ml phenylmethylsulfonyl fluoride, and 5 mg/ml pepstatin in methanol). SDS (0.1%) was added, and the samples were vortexed briefly. Protein samples were diluted 1:2 in SDS lysis buffer (8Laemmli U.K. Nature. 1970; 227: 680-685Crossref PubMed Scopus (207208) Google Scholar) containing 40% β-mercaptoethanol, heated at 100 °C for 5 min, and separated on an 10% SDS-polyacrylamide gel, and protein was transferred to nitrocellulose paper (BA-S83, 0.02 μm; Schleicher & Schuell). 20 μg of protein was loaded in each lane, as indicated by the Bradford protein assay method. Mouse anti-GFP antibody (Roche Molecular Biochemicals) was used at a 1:1000 dilution, and horseradish peroxidase-conjugated goat anti-mouse IgG secondary antibody (Sigma) was used at a dilution of 1:3000. Antibody-antigen complexes were visualized using the Renaissance system (PerkinElmer Life Sciences). One ml of cells grown in minimal media and under derepressing conditions (when necessary) were harvested and washed twice with media, resuspended in 500 μl of media containing 5 mg/ml of lucifer yellow carbonyl hydrazine (Sigma), and incubated for 60 min at 23 °C. Cells were washed three times with media and visualized. Lucifer yellow uptake assays were conducted as described previously (9Goode N.T. Hajibagheri M.A.N. Warren G. Parker P.J. Mol. Biol. Cell. 1994; 5: 907-920Crossref PubMed Scopus (29) Google Scholar) with the following changes: 5 ml of cells were grown as above and washed twice in ice-cold media and resuspended in 200 μl of ice-cold media containing 5 mg/ml of lucifer yellow carbonyl hydrazine, and half (100 μl) of each sample was incubated at 32 °C with the other half kept on ice. After 90 min, the cultures were washed three times with ice-cold H2O, treated with zymolyase for 30 min at 37 °C, then lysed with acid-washed glass beads for 1 min. 400 μl of media was added to the beads, and the liquid was transferred to a fresh tube, then centrifuged for 15 min at 10,000 × g. The fluorescence (excitation, 426 nm; emission, 550 nm) was determined using a fluorimeter and quantified by comparison to a standard curve of lucifer yellow fluorescence. GFP-Nrf1p was previously shown to localize to the plasma membrane, nuclear membrane, septum, and vacuolar membrane (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar). To determine if the three predicted transmembrane domains in Nrf1p were functional, three amino acids in the middle of each putative domain were mutated to charged residues (Fig. 1 A). These mutations (nrf1 tm1, nrf1 tm2, andnrf1 tm3 ) were constructed in pREP3X-GFP-A8-nrf1 +, a high level expression vector, and expressed as GFP-Nrf1p fusion proteins. Total protein was isolated from wild-type ED668 cells containing GFP-Nrf1p, GFP-Nrf1tm1p, GFP-Nrf1tm2p, or GFP-Nrf1tm3p, and all were expressed at similar levels (Fig. 1 B). Although high level overexpression of GFP-Nrf1p was lethal (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar), the three transmembrane mutant proteins did not confer lethality when overexpressed (Fig.2 A). The localization of these mutant proteins was abnormal with no vacuole localization, variable levels of diminished plasma and nuclear membrane localization, and a general increase in the presence of cytosolic GFP aggregates (Fig.2 B). These data suggest that all three of these domains are necessary for efficient localization of GFP-Nrf1p to plasma, nuclear, and especially vacuolar membranes, supporting the hypothesis that these are functional transmembrane domains. Site-directed mutagenesis was performed on the putative extramembranous loops of Nrf1p to determine which were necessary for function. The "positive inside" observation was followed, which predicts that the nontranslocated loops of a protein are enriched in positively charged residues as compared with translocated loops (10von Heijne G. J. Mol. Biol. 1992; 225: 487-494Crossref PubMed Scopus (1402) Google Scholar). Three charged residues within a stretch of five amino acids in the predicted nontranslocated (i.e. cytoplasmic) loops were changed to alanine residues in pREP3X-GFP-A8-nrf1 + (Fig.1 A). Total protein was isolated, and the level of expression of GFP-Nrf1nt-1p, GFP-Nrf1L2–1p, GFP-Nrf1L2–2p was comparable with wild-type GFP-Nrf1p (Fig. 1 B). Overexpression of GFP-Nrf1nt-1p was lethal, and localization of the protein appeared similar to GFP-Nrf1p, suggesting that these charged residues were not necessary for the proper localization of GFP-Nrf1p or its lethality when overexpressed. Overexpression of GFP-Nrf1L2–1p did not confer lethality, and no fluorescence was localized to the vacuole, although the plasma membrane and nuclear membrane localization appeared enhanced compared with overexpression of wild-type GFP-Nrf1p (Fig.3, A and B). These data indicate that this portion of the second extramembranous loop was necessary for efficient localization to the vacuoles and was also necessary for the protein to cause a lethal defect when overexpressed. Overexpression of GFP-Nrf1L2–2p also did not confer lethality, but it localized similarly to wild-type GFP-Nrf1p (Fig. 3,A and B). These data indicate that thenrf1 L2–2 mutation affects the ability of Nrf1p to cause a lethal phenotype when overexpressed; however, it does not appear to affect the localization of the protein in wild-type cells. The localization of Nrf1p to vacuoles and subsequent coalescence of vacuoles around the nucleus and vacuolar fusion depended on Scd1p (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar). To determine if Scd1p may affect the localization of the GFP-Nrf1 mutant proteins, we expressed them in Δscd1 cells. Overexpression of wild-type GFP-Nrf1p or GFP-Nrf1nt-1p was lethal in Δscd1 cells; however, overexpression of the other GFP-Nrf1 mutant proteins did not affect the viability of this strain (data not shown). GFP-Nrf1nt-1p was localized to the plasma and nuclear membranes but was not observed in the vacuolar membranes similarly to wild-type GFP-Nrf1p, suggesting that this mutation does not affect localization of GFP-Nrf1p in Δscd1 cells (Fig. 4). The three transmembrane mutant proteins GFP-Nrf1tm1p, GFP-Nrf1tm2p, and GFP-Nrf1tm3p all showed a similar localization pattern with no vacuole localization and variable levels of diminished plasma membrane and nuclear membrane localization, comparable with the localization pattern in wild-type cells (data not shown). Localization of GFP-Nrf1L2–1p was as predicted, with localization to the plasma and nuclear membranes only (Fig. 4). However, GFP-Nrf1L2–2p localization was similar to that observed in wild-type cells, with localization to the plasma membrane, nuclear membrane, and the vacuolar membranes, indicating a bypass of the Scd1p requirement for vacuolar membrane localization (Fig. 4). These data further implicate the second extramembranous loop of Nrf1p in the localization of GFP-Nrf1p to the vacuolar membranes and suggest that there may exist a Scd1p-independent mechanism for Nrf1p localization. These above-mentioned localization results suggested that Nrf1p may first be targeted to the plasma membrane and subsequently localize to the vacuolar membrane through endocytosis, implicating Scd1p and/or Nrf1 in the endocytosis pathway. To examine this possibility, lucifer yellow uptake was assayed in wild-type, Δscd1, and Δnrf1 cells as well as Δypt7 cells, which are deficient in vacuolar fusion and, hence, endocytosis. Δscd1 cells showed a decrease in endocytosis similar to what was observed with Δypt7 cells, whereas Δnrf1 cells showed a lesser deficiency in uptake of lucifer yellow (Fig. 5 A). These data are consistent with Scd1p and Nrf1p having a positive function in endocytosis. To determined whether endocytosis in wild-type or Δscd1 cells could be affected by overexpression of Nrf1p or the Nrf1 mutant proteins, the nrf1 mutations described previously were inserted into pREP1-nrf1, a high level expression plasmid. As with overexpression of GFP-Nrf1p fusion proteins, overexpression of the wild-type Nrf1p or Nrf1nt-1p from these plasmids led to lethality after extended induction periods in both strains, whereas overexpression of the other mutant proteins did not (data not shown). Overexpression of wild-type Nrf1p or Nrf1nt-1p caused a significant decrease in lucifer yellow uptake in wild-type cells under conditions where the cells were still viable, whereas overexpression of the other mutant proteins led to at least as much lucifer yellow uptake as the vector control (Fig. 5 B). These data suggest that overexpression of functional Nrf1p leads to a decrease in endocytosis. Overexpression of Nrf1p or any of the mutant proteins did not affect lucifer yellow uptake in the Δscd1 strain (data not shown). Overexpression of Nrf1L2–2p, which localizes to the vacuole in Δscd1 cells, did not increase endocytosis in these cells, suggesting that targeting of Nrf1p to the vacuole (at least of this mutant protein) could occur independent of Scd1p-dependent endocytosis. To examine whether Cdc42p may be involved in endocytosis, dominant-activated Cdc42G12Vp and dominant-negative Cdc42T17Np were overexpressed in wild-type cells, and the ability to uptake lucifer yellow was analyzed. Overexpression of dominant-activated Cdc42G12Vp produced a 3–5-fold increase in lucifer yellow uptake (Fig.6 A), with no vacuolar abnormalities or coalescence observed (data not shown). A modest increase in lucifer yellow uptake was also observed with overexpression of dominant-negative Cdc42T17Np (Fig. 6 A). These results suggest that overexpression of Cdc42 mutant proteins can influence endocytosis. A small decrease in lucifer yellow uptake was also observed upon overexpression of kinase-inactive Pak1K415R,K416Rp, suggesting that Pak1p may be involved in the endocytic process. These data raise the possibility that the endocytic defects observed in the Δscd1 and Δnrf1 mutants may be the result of decreased Cdc42p activity. To test this possibility, dominant-activated Cdc42G12Vp was overexpressed in the Δscd1 and Δnrf1 strains. Overexpression of Cdc42G12Vp rescued the endocytosis defect in both the Δscd1 and Δnrf1 strains (Fig. 6 B) but not the defect observed in cells overexpressing Nrf1p (data not shown), suggesting that the endocytosis defect observed in these strains was due to a decrease in Cdc42p activity (see "Discussion"). Vacuoles fuse in response to osmotic stress (i.e. placement in H2O), and this fusion has been shown to be dependent on the Pmk1p/Spm1p mitogen-activated protein kinase cascade as well as the conserved Rab-like GTPase Ypt7p, which is necessary for vacuolar fusion in S. cerevisiae (11Bone N. Millar J.B.A. Toda T. Armstrong J. Curr. Biol. 1998; 8: 135-144Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar, 12Haas A. Scheglman D. Lazer T. Gallwitz D. Wickner W. EMBO J. 1995; 14: 5258-5270Crossref PubMed Scopus (233) Google Scholar). To determine whether the vacuolar coalescence around the nucleus and subsequent vacuole fusion observed in cells overexpressing Nrf1p act by a similar mechanism, Nrf1p was overexpressed in Δpmk1 and Δypt7 mutant strains. As previously shown in wild-type cells, overexpression of Nrf1p was lethal in these strains (data not shown). The vacuolar morphology was observed by carboxy-DCFDA staining and confirmed by subsequent localization of GFP-Nrf1p. The Δpmk1 strain showed the abnormal vacuolar phenotypes associated with Nrf1p overexpression, but no vacuolar changes were observed in the Δypt7 cells (TableII). GFP-Nrf1p localization appeared normal in the Δpmk1 cells, but the Δypt7 strain showed localization only to the plasma membrane (data not shown). These data suggested that the localization of Nrf1p to the vacuole and the subsequent abnormal vacuolar morphology was independent of Pmk1p but was dependent on Ypt7p.Table IINrflp-induced vacuolar abnormalitiesStrainPlasmid% normal% abnormal2-aRepresents cells with vacuoles coalesced around the nucleus. n = 200.ED668pREP1982ED668pREP1-nrfl +6535Δscd1pREP11000Δscd1pREP1-nrf1+991Δypt7pREP11000Δypt7pREP1-nrf1 +1000Δpmk1pREP1964Δpmk1pREP1-nrf1 +68322-a Represents cells with vacuoles coalesced around the nucleus. n = 200. Open table in a new tab To determine whether Scd1p or Nrf1p may be necessary for vacuolar fusion induced by osmotic stress, Δscd1 and Δnrf1 cells were grown in YES media, stained with carboxy-DCFDA, and shifted into H20 to observe the vacuolar morphology (Fig. 7). Fusion was observed in Δscd1 cells as well as Δnrf1 cells, suggesting that the vacuolar fusion induced by osmotic stress is independent of Scd1p and Nrf1p. These data are consistent with the vacuolar fusion induced by overexpression of Nrf1p occurring through a separate mechanism. Nrf1 proteins that contained mutations in the putative transmembrane domains designed to perturb the predicted hydophobicity of this domain no longer conferred a lethal phenotype and were mislocalized, with a decrease in membrane localization and an increase in GFP aggregates, suggesting that these domains were necessary for efficient targeting to membranes. The nrf1 nt-1mutation did not affect the GFP localization or lethality conferred by overexpression, whereas mutations in the predicted second extramembranous loop did, suggesting that this region of the protein was involved in localization of GFP-Nrf1p to the vacuolar membrane. Both GFP-nrf1 L2–1p and GFP-nrf1 L2–2p no longer conferred lethality when overexpressed, and GFP-nrf1 L2–1p did not localize to the vacuolar membrane, whereas GFP-nrf1 L2–2p localized to the vacuolar membrane independent of Scd1p. Both deletion and overexpression of nrf1 led to similar defects in endocytosis. This seemingly paradoxical result is reminiscent of results with Cdc24p in S. cerevisiae in which both deletion and overexpression of Cdc24p led to a loss-of-function phenotype (13Ziman M. Johnson D.I. Yeast. 1994; 10: 463-474Crossref PubMed Scopus (37) Google Scholar). One explanation could be that loss of a protein and excess amounts of a protein both adversely affect the stability and/or function of a multiprotein complex essential for the process in question. Deletion of scd1 also led to a decrease in endocytosis, whereas overexpression of dominant-activated Cdc42G12Vp led to a 3–5-fold increase in endocytosis and rescued the defect in Δscd1 and Δnrf1 cells. These data implicate the Cdc42p-dependent signaling pathway in the endocytic process in S. pombe. Overexpression of dominant-negative Cdc42T17Np also led to a modest increase in endocytosis, suggesting that this allele may not function negatively in the endocytosis pathway. In mammalian cell lysates, endocytic vesicles have been shown to move at the ends of actin tails, and the Cdc42p effector, Wiskott-Aldrich syndrome protein (WASp), has been shown to be involved in the nucleation of actin and subsequent propulsion of these vesicles (14Merrifield C.J. Moss S.E. Ballestrem C. Imhof B.A. Giese G. Wunderlich I. Almers W. Nat. Cell Biol. 1999; 1: 72-74Crossref PubMed Scopus (266) Google Scholar, 15Taunton J. Rowning B.A. Coughlin M.L. Wu M. Moon R.T. Mitchison T.J. Larabell C.A. J. Cell Biol. 2000; 148: 519-530Crossref PubMed Scopus (344) Google Scholar, 16Rozelle A.L. Machesky L.M. Yamamoto M. Driessens M.H.E. Insall R.H. Roth M.G. Luby-Phelps K. Marriott G. Hall A. Yin H.L. Curr. Biol. 2000; 10: 311-320Abstract Full Text Full Text PDF PubMed Scopus (429) Google Scholar). Two Rho-type GTPase inhibitors, ToxB and RhoGDI, interfered with this process, and purification of a soluble factor capable of rescuing the ToxB defect was identified as a Cdc42p-RhoGDI complex (15Taunton J. Rowning B.A. Coughlin M.L. Wu M. Moon R.T. Mitchison T.J. Larabell C.A. J. Cell Biol. 2000; 148: 519-530Crossref PubMed Scopus (344) Google Scholar). These data are consistent with the Cdc42p-dependent signaling pathway being involved in the trafficking of endocytic vesicles. In mammalian cells, Cdc42p localizes to Golgi membranes and has been shown to regulate targeted secretion to the basolateral membrane in polarized epithelial cells (17Erickson J.W. Zhang C.J. Kahn R.A. Evans T. Cerione R.A. J. Biol. Chem. 1996; 271: 26850-26854Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar, 18Kroschewski R. Hall A. Mellman I. Nat. Cell Biol. 1999; 1: 8-13Crossref PubMed Scopus (286) Google Scholar). Microinjection of plasmids encoding dominant-negative Cdc42T17Np resulted in a mistargeting of proteins normally destined for the basolateral membrane as well as a mistargeting of basolateral endocytosed proteins. Microinjection of activated Cdc42Q61Lp-encoding plasmids resulted in a complete loss of plasma-membrane polarity (18Kroschewski R. Hall A. Mellman I. Nat. Cell Biol. 1999; 1: 8-13Crossref PubMed Scopus (286) Google Scholar). Cdc42p has also been shown to interact with the COPI coatomer complex, specifically through the γCOP subunit (19Wu W.J. Erickson J.W. Lin R. Cerione R.A. Nature. 2000; 405: 800-804Crossref PubMed Scopus (186) Google Scholar). COPI subunits have been purified from isolated endosomes, suggesting that COPI is involved in targeting in the endocytic pathway as well (20Whitney J.A. Gomez M. Sheff D. Kreis T.E. Mellman I. Cell. 1995; 83: 703-713Abstract Full Text PDF PubMed Scopus (266) Google Scholar, 21Aniento F. Gu F. Parton R.G. Gruenberg J. J. Cell Biol. 1996; 133: 29-41Crossref PubMed Scopus (315) Google Scholar). Thecdc42 F28L-transforming mutant showed a defect in COPI recruitment to vesicles; however, a triple Cdc42F28L,K183S,K184S mutant protein that could no longer interact with γCOP resulted in a lack of transformation ability (19Wu W.J. Erickson J.W. Lin R. Cerione R.A. Nature. 2000; 405: 800-804Crossref PubMed Scopus (186) Google Scholar). This mutant protein contained two altered C-terminal lysine residues necessary for COPI binding, suggesting that Cdc42p interaction with COPI was necessary for its ability to transform cells. Altogether, these data implicate mammalian Cdc42p in both exocytic and endocytic trafficking, and the data herein implicate S. pombe Cdc42p in the endocytic pathway, suggesting a conserved role for this GTPase in these processes. In S. pombe, vacuoles are numerous, with greater than 50 individual organelles sometimes present within a cell. Under hypo-osmotic stress conditions, the vacuoles can fuse into one or two large organelles (11Bone N. Millar J.B.A. Toda T. Armstrong J. Curr. Biol. 1998; 8: 135-144Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). This osmotic stress-induced fusion is dependent on the Rab-type GTPase Ypt7p and the Sty1p and Pmk1p mitogen-activated protein kinase cascade pathways. Deletion of the Sty1p or Pmk1p mitogen-activated protein kinases or Ypt7p resulted in a severe defect in vacuole fusion when S. pombe cells were isolated from rich media and placed in H2O (11Bone N. Millar J.B.A. Toda T. Armstrong J. Curr. Biol. 1998; 8: 135-144Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). The acidification of these vacuoles did not appear to be affected, as the pH-dependent fluorescent vital stain carboxy-DCFDA stained vacuoles in these cells. In this study, it was shown that the vacuolar fusion induced by overexpression of Nrf1p depended on Ypt7p but not on Pmk1p, and osmotically induced vacuolar fusion was not dependent on Nrf1p or Scd1p, suggesting that there exist at least two distinct vacuolar fusion mechanisms. Recently the S. cerevisiae Nrf1p homolog (Vtc1p/Nrf1p) was copurified with the vacuolar-ATPase subunit (v-ATPase) Vma10p (22Cohen A. Perzov N. Nelson H. Nelson N. J. Biol. Chem. 1999; 274: 26885-26893Abstract Full Text Full Text PDF PubMed Scopus (75) Google Scholar). A Δvtc1 mutant showed no morphological defect; however, a reduction in quinacrine uptake into purified vacuoles suggested that the vacuoles were not properly acidified, which depends on the v-ATPase. v-ATPase deletion mutant cells can survive a lack of acidification by taking up external fluid via endocytosis but are inviable in media at a pH higher than 7.0 (23Nelson N. Harvey W.R. Physiol. Rev. 1999; 79: 361-385Crossref PubMed Scopus (370) Google Scholar). A Δvma8Δvtc1 double mutant was viable at a higher pH, and vacuoles isolated from a Δvtc1 strain showed a decrease in the level of v-ATPase units associated with these membranes (22Cohen A. Perzov N. Nelson H. Nelson N. J. Biol. Chem. 1999; 274: 26885-26893Abstract Full Text Full Text PDF PubMed Scopus (75) Google Scholar). Vtc1p appears to be a member of a conserved family of proteins (Vtc1p-4) that are found in a complex on the vacuolar membrane in S. cerevisiae (22Cohen A. Perzov N. Nelson H. Nelson N. J. Biol. Chem. 1999; 274: 26885-26893Abstract Full Text Full Text PDF PubMed Scopus (75) Google Scholar). 2A. Mayer, personal communication. These data suggest a possible role for Nrf1p/Vtc1p in the proper targeting of v-ATPase subunits to the vacuole. There is growing evidence for Cdc42p being involved in vesicle trafficking in mammalian cells. In this study, overexpression of constitutively GTP-bound Cdc42p led to an increase in endocytosis inS. pombe and rescued the endocytosis defect in the Δscd1 and Δnrf1 strains. Interestingly, a recent study has also implicated Cdc42p in the control of endocytosis in mammalian dendritic cells (24Garrett W.S. Chen L. Kroschewski R. Ebersold M. Turley S. Trombetta S. Galan J.E. Mellman I. Cell. 2000; 102: 325-334Abstract Full Text Full Text PDF PubMed Scopus (366) Google Scholar). These results suggest that a function for Cdc42p in vesicle trafficking may be conserved throughout eukaryotes. GFP-Cdc42p also localizes to the vacuole and may be involved in vacuolar fusion (4Murray J.M. Johnson D.I. Genetics. 2000; 154: 155-165PubMed Google Scholar). Although endocytosis and vacuolar fusion may not be distinct processes (for example, deletion ofypt7 leads to a defect in vacuolar fusion and therefore shows a defect in lucifer yellow uptake), the fact that Nrf1p and Scd1p are not required for osmotically induced vacuolar fusion suggests that the endocytic defect in Δscd1 and Δnrf1 deletion strains is not a result of a general defect in vacuolar fusion. Further examination of the functions of Nrf1p, Scd1p, and Cdc42p in S. pombe will help to uncover whether these proteins act during vesicle formation, trafficking, and/or vacuolar fusion. We thank J. Armstrong, E. Chang, P. Fantes, A. Merla, T. Toda, and A. Mayer for strains and reagents. We also thank G. Ward and members of the Johnson lab for thoughtful discussions and critical reading of this manuscript.