Schizosaccharomyces pombe Rgf3p is a specific Rho1 GEF that regulates cell wall  -glucan biosynthesis through the GTPase Rho1p

IntroductionThe fungal cell wall is the essential cellular boundary,controlling many transport processes, cellular metabolismand, indeed, all communications with the extracellular world.Because of its mechanical strength, it allows cells to withstandturgor pressure and consequently prevents cell lysis. In thefission yeast Schizosaccharomyces pombe, the cell wallmainly consists of three polysaccharides, β-1,3-glucan, α-1,3-glucan and galactomannoproteins, all of which form a largecomplex (for a review, see Duran and Perez, 2004). Theircoordinated synthesis represents an essential step for theassembly of a functional cell wall to ensure cell integrity(Ishiguro, 1998).

We have used the biosynthesis of β-1,3-glucan as a modelto study morphogenesis. It has been suggested that β-1,3-glucan is the first polymer to be synthesized (Osumi et al.,1989; Roh et al., 2002) and that the regulation of thispolysaccharide might be a key step in the sequential assemblyof the other cell wall components. β-1,3-Glucan comprises~45% of the cell wall and is the major structural component,as seen by the fact that its enzymatic degradation leads to thesolubilization of the other components. The enzymatic systemthat catalyses the synthesis of this polysaccharide is β-1,3-glucan synthase (GS). GS is composed of at least two fractions:a catalytic moiety of the enzyme and a regulatory component.

In fission yeast, the catalytic subunit of GS is encoded by atleast four genes: cps1+/bgs1+ (Le Goff et al., 1999; Cortes etal., 2002; Liu et al., 2000b; Liu et al., 2002), bgs2+ (Martin etal., 2000; Liu et al., 2000a), bgs3+ (Martin et al., 2003) andbgs4+ (Cortés et al., 2005). All of them code for essentialproteins. In addition to the catalytic subunit, the small GTP-binding protein Rho1p is an essential regulatory subunit(Arellano et al., 1996). Rho1 acts as a binary switch by cyclingbetween an inactive GDP-bound and an active GTP-boundconformational state. Rho1p stimulates GS in its GTP-boundprenylated form, providing a rationale for the understanding ofthe mechanism by which the cell can switch β-1,3-glucansynthesis on and off by interconverting the GDP and GTPforms of Rho1p.

The Rho1p of fission yeast is a functional homologue ofbudding yeast Rho1p (Nakano et al., 1997), and belongs to afamily of small GTPases that are key regulators in polarityprocesses (for reviews, see Mackay and Hall, 1998; Takai etal., 2001; Burridge and Wennerberger, 2004). The fission-yeastRho family includes Cdc42p and Rho1p-Rho5p. The cdc42+

gene is essential and is involved in the establishment of cellpolarity (Miller and Johnson, 1994). The rho2+ gene has beenshown to be involved in the control of cell morphogenesis,probably by regulating the synthesis of Mok1p, the α-1,3-glucan synthase, via a Pck2p pathway (Hirata et al., 1998;

6163

Rho1p regulates cell integrity by controlling the actincytoskeleton and cell-wall synthesis. Here, we describe thecloning and characterization of rgf3+, a member of the Rhofamily of guanine nucleotide exchange factors (Rho GEFs).The rgf3+ gene was cloned by complementation of a mutant(ehs2-1) hypersensitive to drugs that interfere with cell-wallbiosynthesis. The rgf3+ gene was found to be essential forcell viability and depletion of Rgf3p afforded phenotypessimilar to those obtained following depletion of Rho1p.However, the cell death caused by Rgf3p depletion could berescued by the presence of 1.2 M sorbitol, whereasdepletion of Rho1 was lethal under the same conditions.We show that Rgf3p is a specific Rho1-GEF. Thehypersensitivity to drugs affecting the cell wall of the ehs2-1 mutant was suppressed by overexpression of rho1+ but

not by any of the other GTPases of the Rho family. Rgf3pinteracted with the GDP-bound form of Rho1p andpromoted the GDP-GTP exchange. In addition, we showthat overexpression of Rgf3p produces multiseptated cellsand increases β-1,3-glucan synthase activity and theamount of cell wall β-1,3-glucan. Rgf3p localized to theseptum and the mRNA level was regulated in a cell-cycle-dependent manner peaking during septation. Our resultssuggest that Rgf3p acts as a positive activator of Rho1p,probably activating the Rho functions that coordinatecell-wall biosynthesis to maintain cell integrity duringseptation.

Key words: Cell-wall mutants, Rho GEF family, Rho1, Fission yeast,Cytokinesis

Summary

Schizosaccharomyces pombe Rgf3p is a specificRho1 GEF that regulates cell wall β-glucanbiosynthesis through the GTPase Rho1pVirginia Tajadura, Blanca García, Ignacio García, Patricia García and Yolanda Sánchez*Instituto de Microbiología Bioquímica, CSIC/Universidad de Salamanca, and Departamento de Microbiología y Genética, Universidad deSalamanca, Campus Miguel de Unamuno, 37007 Salamanca, Spain*Author for correspondence (e-mail: [email protected])

Accepted 10 September 2004Journal of Cell Science 117, 6163-6174 Published by The Company of Biologists 2004doi:10.1242/jcs.01530

Research Article

JCS ePress online publication date 16 November 2004

6164

Calonge et al., 2000). The rho3+ and rho4+ genes are non-essential and are both involved in cell separation processes.Rho3p interacts with the formin For3p and modulates exocystfunction (Nakano et al., 2002; Wang et al., 2003). Rho4p mightbe involved in septum degradation during cytokinesis (Santoset al., 2003; Nakano et al., 2003).

Fission-yeast Rho1 localizes to sites of polarized growth,the cell poles and the septum (Arellano et al., 1997).Expression of a dominant-active Rho1 mutant (Rho1G15Vor Rho1Q64L) produces swollen cells, branched cells andmultiseptated cells, whereas that of a dominant-negative Rho1mutant (Rho1T20N) produces shrunken or dumpy cells(Arellano et al., 1996; Nakano et al., 1997). These cells havedefects in the organization of the actin cytoskeleton, cellpolarity and cell integrity. Rho1p seems to play severalfunctional roles upon interacting with its targets: it activatesGS (Arellano et al., 1996); it binds directly to the proteinkinase C (PKC) family of protein kinases Pck1p and Pck2p;and it is a positive regulator of these kinases (Arellano et al.,1999; Sayers et al., 2000). In addition, Rho1 regulates thelocalization of F-actin patches (Arellano et al., 1997; Sayerset al., 2000). However, little is known about the proteins thatturn Rho1p on and off in the cell. These proteins might playimportant roles in the specificity of Rho functions. There areat least nine proteins that belong to the Rho GTPase activatingprotein (RhoGAP) family in the S. pombe genome. Three ofthese – Rga1, Rga5 and Rga8 – function as GAPs for Rho1p.None of them is essential for cell viability, althoughdeletion of rga1+ causes a slow-growth defect and severemorphological abnormalities (Nakano et al., 2001). Rga5p isinvolved in the regulation of GS activity and cell integrity(Calonge et al., 2003) and Rga8p is a Shk1p (Cdc42/p21-activated kinase) substrate that negatively regulates Shk1p-dependent growth control pathways, potentially throughinteraction with Rho1p GTPase (Yang et al., 2003). Regardingthe role of RhoGEFs as direct activators of Rho GTPases infission yeast, it has been reported that Scd1p and Gef1p couldactivate Cdc42p (Chang et al., 1994; Coll et al., 2003; Hirotaet al., 2003). Recently, in a search for genomic sequencesbearing a Rho GEF domain, five new genes (rgf1+-rgf5+, forRhoGEF 1-5) have been described and reported to be involvedin the regulation of cell morphology (Iwaki et al., 2003).However, it has not yet been shown whether any of thesefactors act specifically on Rho1p.

Our approach to the study of cell-wall biosynthesis andregulation was to obtain mutants hypersensitive to the cell-wallantifungal drugs Echinocandin (Ech) and Calcofluor White(Cfw) (ehs mutants). In the present work, we cloned the rgf3+

gene as the structural gene that complements the ehs2-1mutation. Genetic and biochemical studies have indicated thatRgf3p is a Rho1p-specific GEF in S. pombe. Moreover, ourdata suggest that, among the different Rho1p essentialfunctions, Rgf3p could specifically regulate β-1,3-glucanbiosynthesis and cell integrity during septation.

Materials and MethodsMedia, reagents and geneticsThe genotypes of the S. pombe strains used in this study are listed inTable 1. Complete yeast growth medium (YES), selective medium(MM) supplemented with the appropriate requirements andsporulation medium (MEA) have been described elsewhere (Morenoet al., 1991). Ech B (LY280949; LILLY Company) (Radding et al.,1998) was stored at –20°C in a stock solution (2.5 mg ml–1) in 50%ethanol and was added to the media at the corresponding finalconcentration after autoclaving. Cfw was prepared (15 mg ml–1) inwater with a few drops of 10 N KOH, filter sterilized and added asabove to EMM or YES medium, the latter previously buffered with50 mM potassium hydrogen phthalate, pH 6.1. Crosses wereperformed by mixing appropriate strains directly on MEA plates.Recombinant strains were obtained by tetrad analysis. Foroverexpression experiments using the nmt1+ promoter, cells weregrown in EMM containing 15 µM thiamine up to the logarithmicphase. Then, the cells were harvested, washed three times with MMand inoculated into fresh medium (without thiamine) at an opticaldensity at 600 nm of 0.01.

Mapping of the ehs2-1 mutantGenetic mapping was carried out by measuring genetic linkage in aswi5 mutant background (Schmidt, 1993) (swi5 strains were a giftfrom H. Schmidt, Institut für Genetik, Braunschweig, Germany).First, the ehs2-1 mutant was shown to map to chromosome III.Second, the position of ehs2-1 on the chromosome was determined.An ehs2-1 swi5-39 leu1-32 h– strain was constructed and crossed witha ura4-294 tps14-5 ade5-36 swi5-39 h90 strain. Tetrad analysis of thiscross revealed that ehs2-1 is localized to the right arm of chromosomeIII, closer to the tps14 gene. Third, the ehs2-1 mutation was mappedby linkage analysis in a swi5+ background. An ehs2-1 leu1-32 h– strainwas crossed with mutants in genes that map to this chromosome arm(tps14 ade6-250 arg1-230 h+). Tetrad analysis of the crosses wasperformed and the ehs2-1 mutant was found to be closely linked tothe ade6 gene (1.5 cM, 45 tetrads analysed). Cosmids in the ade6region were screened for genes that might be related to cell-wallbiosynthesis. We chose mok1+ (which encodes an α-glucan synthase)in cosmid C17A7 and rgf3+ and rgf1+, both in cosmid C645 (genesbearing homology with ROM1 from Saccharomyces cerevisiae, a GEFfor Rho1p). The mok1+ gene was kindly provided by T. Toda (CancerResearch UK, London) and we found no complementation of theehs2-1 phenotypes.

Journal of Cell Science 117 (25)

Table 1. S. pombe strains used in this workStrains Genotypes

PN22 h– leu1-32GI 1 h+ leu1-32, ehs2-1MS38 h– leu1-32 ade6M210 ura4D-18 his3D1MS75 h+/h– leu1-32/leu1-32, ade6M210/ade6M216, ura4D-18/ ura4D-18, his3D1/his3D1YSM373 h+/h– leu1-32/leu1-32 ade6M210/ade6M216, rgf3::ura4+/rgf3+ his3D1/his3D1 ura4D-18/ura4D-18YSM654 h+/h– leu1-32/leu1+ his3+/his3D1 rgf3::ura4+/rgf3+ wee1-50/wee1-50 ura4D-18/ura4D-18YSM656 h+/h– leu1-32/leu1+ his3D1/his3+ rgf3::ura4+/rgf3+ sid2-250/sid2-250 ura4D-18/ura4D-18VT88 h– leu1-32 ade6M210 ura4D-18, his3DI, 81 nmt-rgf3+-ura4+

VT128 h– leu1-32 ade6M210 ura4D-18, his3DI leu1+::EGFP-rgf3+

PPG217 h– leu1-32 ade6M210 ura4D-18 his3D1 rho1::ura4+ + pREP41X-rho1PN35 h+ leu1-32, ura4D-18, cdc25-22

6165GEF for Rho1p in S. pombe

Plasmid and DNA manipulationsThe rgf3 open reading frame (ORF) was obtained from cosmid C645in two pieces. First, we cloned a 2.5 kb SalI-HindIII fragment inpALKS and then we introduced a 3.5 kb HindIII-HindIII fragmentadjacent in the cosmid to obtain pYS10 bearing the entire rgf3 ORF.The pYS8 plasmid, containing the rgf1 ORF, was obtained insertinga 7 kb EcoRI fragment from cosmid C645 into pAUKS. To tag Rgf3pat its N-terminus with enhanced green fluorescent protein (EGFP) andwith the triple repeat of the influenza-virus haemagglutinin epitope(HA) (Craven et al., 1998), pYS10 was modified by site-directedmutagenesis. We created a SalI site at position –2 (before the ATG),a NotI site at position +1 (after the ATG) and a SmaI site at the C-terminus (after the termination codon) (pVT1). The HA and EGFPepitopes were inserted in frame at the NotI site of pVT1. pVT-GFPrgf3 and pVT-HArgf3 fully complemented the ehs2-1phenotypes. Strain VT128, with the GFP-rgf3+ integrated under itsown promoter, was constructed by subcloning the rgf3+ tagged withEGFP (from plasmid pVT-GFPrgf3) into the integrative vector pIJ148(Keeney and Boeke, 1994), resulting in pIJ148-GFP-rgf3+. Thisplasmid was cut with Eco47III and integrated into the leu1 locus ofstrain MS38. The nmt1-promoter-containing vectors pREP3X andpREP41X (Forsburg, 1993) were used to overexpress rho1+ to rho5+,cdc42+ and rgf3+. All GTPases of the Rho family were tagged withtwo HA epitopes at the 5′ end (Calonge et al., 2003). Theoverexpression plasmids were kindly provided by P. Perez and P. M.Coll (Instituto de Microbiología Bioquímica, Salamanca, Spain). Tooverexpress rgf3+, a SalI-SmaI fragment containing the rgf3+ genetagged with the HA epitope from plasmid pVT-HArgf3 was ligatedinto the SalI-SmaI sites of plasmid pREP3X or the Xho1-SmaI sitesof pREP41X and pREP81X. For shut-off, a ura4+-81 nmt1-rgf3+

strain (VT88) was constructed using pVT-HArgf3 and one-step genereplacement. A SalI fragment containing the rgf3+ promoter inplasmid pVT-GFPrgf3 was substituted by another fragmentcontaining 5′ rgf3+ sequences, the ura4+ marker (cloned in SmaI), andthe 81nmt1 promoter (cloned in PstI-SalI). An ApaI fragmentcontaining the regulatory sequences and a 1.5 kb fragment from thergf3+ ORF (up to the ApaI site) was used to transform a haploid strain(MS38). We selected for haploids in MM without thiamine and uracil,and correct integration was analysed by the polymerase chain reaction(PCR).

Construction of rgf3 null mutantsThe rgf3::ura4+ disruption construct was obtained in a two-stepprocess. The 5′ non-coding region of the rgf3+ ORF [nucleotides (nt)–1303 to –9] was obtained by PCR, inserting the SalI and HindIII sites(one at each end), and was ligated into the same sites of the SK-ura4vector to yield pYS52. The 3′ flanking region of the rgf3+ ORF (nt+3769 to +5405) was obtained by PCR, inserting the BamHI and NotIsites as above, and was cloned into the same sites of pYS52 to yieldpYS53. Disruption of rgf3+ was accomplished using the 4.7 kbfragment from pYS53 cut with XhoI and NotI, and transforming theMS75 diploid strain. Transformants were replica-plated five timesconsecutively on YES medium in order to eliminate the cells that hadnot integrated the construct. Then, correct integration was analysedby PCR using the following oligonucleotides: IPCR-b (5′-CACCA-TGCCAAAAATTACACAAGATAGAAT-3′) in the ura4+ gene; R13-e (5′-GGCAGGATTCACCGGATC-3′) downstream from nucleotide–5405 and therefore external to the disruption cassette; GEF-s (5′-CTCTCGTAGAGTCGCGTC-3′) and R15-i (5′-GGCCTTAGCTT-GCCTTG-3′) in the rgf3+ gene. Correct integrations were alsoconfirmed by genomic Southern blotting. Tetrad analysis of theheterozygous diploid disclosed two viable (ura–) and two unviablespores, indicating that rgf3+ is essential for viability.

The rgf3+ gene was isolated from the ehs2-1 mutant by gap repair(Orr-Weaver et al., 1991). Upstream and downstream flankingsequences from rgf3+ were subcloned in pALKS. The plasmid was

linearized with the 5′ and 3′ fragments at the ends and used totransform the ehs2-1 haploid strain (GI 1). The gap in the plasmid wasrepaired using the chromosomal sequences and the plasmids wererecovered from yeast. Transformants were replica plated five timesconsecutively on YES medium, and those able to lose the plasmidwere selected. Based on the rgf3+ sequence, we designedoligonucleotides 400 bp apart and sequenced the entire ORF of fourdifferent clones. In all four clones, there was only one change(cytosine to thymine at position +1834).

Two-hybrid analysesWe performed yeast two-hybrid assays essentially as described byDurfee et al. (Durfee et al., 1993). We created a restriction fragmentcarrying the entire rgf3+ ORF by site-directed mutagenesis,introducing SmaI and SalI sites at the start and termination codon ofrgf3+, respectively. Then, the 3.8 kb fragment was fused in frame tothe GAL4 activation domain of pACT2. GTPases rho1+ to rho5+ andcdc42+ were cloned into pAS2 (Coll et al., 2003) and were used asbait against rgf3+ cloned in the pACT2 plasmid. The S. cerevisiaeY190 strain, which carries the GAL4 recognition sequence and thelacZ and HIS3 reporter genes, was transformed with differentcombinations of plasmids. Expression of the HIS3 reporter gene wasexamined by growth of the host on a –His plate containing 40 mM 3-aminotriazole (3AT).

Pull-down assay for GTP-bound Rho proteinsThe expression vector pGEX-C21RBD (rhotekin-binding domain)(Reid et al., 1996) was used to transform Escherichia coli. The fusionprotein was produced according to the manufacturer’s instructionsand immobilized on glutathione/Sepharose-4B beads (Amersham).After incubation, the beads were washed several times and the boundproteins were analysed by sodium-dodecyl-sulfate polyacrylamide-gel electrophoresis (SDS-PAGE) and Coomassie staining.

The amount of GTP-bound Rho proteins was analysed using theRho-GTP pull-down assay modified from Ren et al. (Ren et al., 1999).Briefly, wild-type, rgf3+-overexpressing and rgf3-mutant cells (ehs2-1) were transformed with either pREP3X-HArho1+ or pREP3X-HArho2+ and grown for 18 hours in minimal medium withoutthiamine. Extracts from 108 cells were obtained as describedpreviously (Arellano et al., 1997) using 500 µl lysis buffer (50 mMTris, pH 7.5, 20 mM NaCl, 0.5% NP-40, 10% glycerol, 0.1 mMdithiothreitol, 1 mM NaF, 2 mM MgCl2, containing 100 µM p-aminophenyl methanesulfonyl fluoride, leupeptin and aprotinin). 100µg glutathione-S-transferase/RBD (GST-RBD) fusion protein coupledto glutathione-agarose beads was used to immunoprecipitate 1.5 mgof the cell lysates. The extracts were incubated with GST-RBD beadsfor 2 hours. The beads were washed with lysis buffer four times andbound proteins were blotted against 1:2000 diluted 12CA5monoclonal antibody (mAb) as primary antibody to detect HA-Rho1por HA-Rho2p. The total amount of HA-Rho1p or HA-Rho2p levelswere monitored in whole-cell extracts (10 µg total protein), whichwere used directly for western blot and were developed with 12CA5mAb. Immunodetection was accomplished using the ECL detectionkit (Amersham Biosciences).

Cell wall analysesEnzyme preparations and GS assays were performed basically asdescribed previously (Martin et al., 2000). One unit of activity wasmeasured as the amount that catalyses the incorporation of 1 µmolsubstrate (UDP/D-glucose) per minute at 30°C. For labelling andfractionation of cell polysaccharides, exponentially growing culturesof S. pombe cells were supplemented with [U-14C]glucose (1 µCiml–1) and incubated for an additional 4-6 hours at either 28°C or37°C (depending on the experimental conditions assayed). Cells

6166

were harvested and total glucose incorporation was monitored bymeasuring the radioactivity in trichloroacetic-acid-insolublematerial. Mechanical breakage of cells was performed usingprechilled glass beads added to the cells and lysis was achieved ina Fast-Prep System (Bio 101; Savant), using six 15-second intervalsat speed 6. Cell walls were pelleted at 1000 g for 5 minutes andwashed three times with 5% NaCl and three times with 1 mM EDTA.Aliquots (100 µl) of total wall were incubated with 100 UZymolyase 100T or Quantazyme (Quantum Biotechnologies) for 36hours at 30°C. Aliquots without enzyme were included as a control.The samples were centrifuged, and the supernatant and washedpellet were counted separately. The supernatants from theZymolyase 100T reaction were considered to contain β-glucan plusgalactomannan and the pellet was considered to hold α-glucan. Thesupernatants from the Quantazyme reaction were considered toharbour β-glucan and the pellet was considered to hold α-glucanplus galactomannan.

Microscopy techniquesThe localization of EGFP-Rgf3p was visualized in living cells. ForCfw staining, exponentially growing S. pombe cells were harvested,washed once and resuspended in water with Cfw at a finalconcentration 20 µg ml–1 for 5 minutes at room temperature. Afterwashing with water, cells were observed under a DMRXA microscope(Leica, Wetzlar, Germany). Formaldehyde fixation was used beforevisualization of F-actin using rhodamine-conjugated phalloidin asdescribed previously (Balasubramanian et al., 1997).

ResultsTemperature-sensitive ehs2-1 mutant has a defect in GSTo identify the fission-yeast genes involved in glucanbiosynthesis, we searched for mutants hypersensitive to thecell-wall inhibitors Cfw and Ech (Carnero et al., 2000). Therationale behind this approach is that mutants with a weakenedcell wall are unable to withstand the additional disturbancecaused by these drugs and die at concentrations of theantifungal agents that are not lethal for cells with a normal wall(Klis, 1994).

The ehs2-1 mutant (for Ech hypersensitive) was unable togrow at 1 µg ml–1 Ech or 0.1 mg ml–1 Cfw, whereas the wild-type strain was able to withstand concentrations of 7.5 µg ml–1

Ech and 1.5 mg ml–1 Cfw. In addition, the mutant cells showeda lytic thermosensitive phenotype at 37°C, which wassuppressed when an osmotic stabilizer (1.2 M sorbitol) wasadded to the medium (Fig. 1A). All these phenotypes co-segregated as a single Mendelian character in tetrad analysis,and they were found to be recessive by diploid analysis (datanot shown). Some of the ehs2-1 mutant cells were lysed cellsand we found that lysis occurred mainly after cytokinesis. At28°C, the proportion of lysis was less than 10% but, after 6hours at 37°C, more than 60% of the cells showed thatphenotype. To examine the viability of the ehs2-1 mutants,cells from cultures incubated at 28°C or 37°C, or at 37°Csupplemented with sorbitol were counted and plated in richmedium at different times of growth. A rapid loss in viabilitywas observed in the cells growing at 37°C without osmoticsupport (Fig. 1B). The hypersensitivity of the mutant cells tocell-wall-specific drugs (Carnero et al., 2000) and the fact thatthe lytic phenotype (observed at 37°C) could be suppressed byan osmotic stabilizer suggest a defect in cell-wall architecture.To test this possibility, GS activity was measured in ehs2-1 and

wild-type strains grown at 28°C and further incubated for 2hours at either the permissive (28°C) or the restrictivetemperature (37°C). As shown in Table 2, the GS activity ofmutant cells after 2 hours at the restrictive temperature was55%, compared with 100% in the wild-type strain. Even at28°C, the GS activity in the mutant was diminished to 75%.

Journal of Cell Science 117 (25)

Fig. 1. Growth phenotypes of ehs2-1 mutant cells. (A) Morphologyof ehs2-1 mutant cells grown at different temperatures. Differentialinterference-contrast micrographs of S. pombe wild-type (PN22) andehs2-1 (GI 1) grown in YES liquid medium at 28°C or 37°C for 6hours in the presence or absence of 1.2 M sorbitol (S).(B) Proportion of viable cells of the ehs2-1 mutant grown at differenttemperatures for the times indicated (with or without 1.2 M sorbitol)and plated on rich medium at 28°C.

Table 2. β-1,3-Glucan synthase activities from S. pombewild-type (PN22) and mutant (ehs2-1) strains

Temperature Strain Specific activity (%)

28°C Wild type 9.23±1.11 (100)ehs2-1 6.92±1.11 (75)

37°C Wild type 5.35±0.75 (100)ehs2-1 2.98±0.48 (56)

S. pombe wild-type (PN22) and mutant (ehs2-1) strains grown at 28°C and37°C. The 37°C extracts were prepared from cells grown overnight at 28°Cand then for 2 hours at 37°C in rich medium. The strain-specific activity isexpressed as milliunits per mg protein. GTP was added to the assay. Valuesare means±s.d. calculated from at least three independent experiments.

6167GEF for Rho1p in S. pombe

Cloning of the ehs2-1 gene (rgf3)In the process of cloning the ehs2-1 gene by complementation,we isolated a plasmid from a S. pombe genomic library thatwas able to suppress the hypersensitivity of ehs2-1 cells to Echand Cfw. Sequencing of the insert revealed that it contained thebgs3+ gene, encoding one of the bgs family components inS. pombe (Martin et al., 2003). The bgs3+ gene failed tocomplement the lytic phenotype at 37°C of the ehs2-1 mutant,in support of the notion that it was acting as a multicopysuppressor (Martin et al., 2003). Accordingly, we usedpositional cloning as an alternative method to clone the ehs2+

gene. The ehs2-1 mutation mapped very close to the ade6 gene.Cosmids that spanned the region around the ade6 gene wereselected and screened for genes that could be related to cell-wall biosynthesis. We first chose mok1+ (α-glucan synthase) incosmid C17A7, but found no complementation of the ehs2-1phenotypes. Next, we tested two ORFs coding for proteinscontaining Rho-GEF domains, rgf1+ and rgf3+, in cosmidsSPCC645.07C and SPCC645.06c, respectively. The name rgfstands for RhoGEF (http://www.genedb.org/genedb/pombe/index.jsp). The two ORFs are consecutive, with divergentpromoters. The rgf3+ gene completely rescued all phenotypesof the ehs2-1 mutant, whereas rgf1+ partially complementedthe hypersensitivity to Cfw and Ech but did not rescue lysis at37°C (Fig. 2A,B). To determine whether rgf3+ was the trueehs2+ gene or whether it was acting as an extragenic multicopy

suppressor, we subcloned the rgf3+ ORF and flankingsequences in the integrative vector pJK148 (Keeney andBoeke, 1994). The construct was integrated into the genome ofa ehs2-1 mutant at the leu1 locus. The strain created behavedlike the wild type for hypersensitivity to antifungal drugs andheat sensitivity, suggesting that rgf3+ is the structural gene thatcomplements the ehs2-1 mutation (data not shown).

The rgf3+ gene encodes a protein of 1275 amino acids witha predicted molecular size of ~144 kDa. Structural analysis ofRgf3p showed that it contains the putative Dbl homology (DH)domain (amino acid residues 469-653) and a pleckstrinhomology (PH) domain (amino acids 693-855) adjacent to theDH domain characteristic of most RhoGEFs (Fig. 2C) (forreviews, see Zheng, 2001; Schmidt and Hall, 2002). There areseven genes coding proteins with putative RhoGEF domains inS. pombe – Scd1+ and gef1+ both encode GEFs for cdc42+ (Collet al., 2003; Hirota et al., 2003), and rgf1+, rgf2+, rgf3+, gef2+

and gef3+ have been shown to be involved in cell morphologyand the actin cytoskeleton (Iwaki et al., 2003). A comparisonof Rgf3p with Rgf1p and the S. cerevisiae Rom2p is shown inFig. 2C.

GEF domain is essential for Rgf3p functionWe next examined which sort of mutation in the rgf3+ readingframe was able to confer the ehs2-1 phenotype. The rgf3 ORF

Fig. 2. Complementation of the ehs2-1 thermosensitive and hypersensitive phenotypes by plasmids pAL-rgf3 (pYS10) and pAL-rgf1 (pYS8).GI1 (h+ leu1-32, ehs2-1) cells were transformed with pAL-rgf3, pAL-rgf1 or pAL (empty plasmid). (A) Transformants were selected in MMand the temperature-sensitive phenotype was scored by incubating the cultures for 4 hours at 37°C. Differential-interference-contrast images areshown. (B) Transformants were streaked out on YES plates in the presence or absence of echinocandin (Ech) (1 µg ml–1) or Calcofluor White(Cfw) (1 mg ml–1). Plates were incubated at 28°C for 4 days. (C) Schematic illustration of structural features analysed by the SMART program(Letunic, 2002) (http://smart.embl-heidelberg.de/). Domains are indicated: CNH, citron homology domain (this acts as a regulatory domain andcould be involved in macromolecular interactions); DEP, domain of unknown function present in signalling proteins that contain PH, RasGEF,RhoGEF, RhoGAP, RGS or PDZ domains; PH, pleckstrin-homology domain; RhoGEF, domain conserved among GEFs for Rho/Rac/Cdc42-like GTPases. (D) Alignment of predicted amino acid sequence of ehs2-1 with the corresponding region of known GEF proteins from differentorganisms (S. pombe Scd1, Caenorhabditis elegans unc-73, human Dbl, human Abr, human Bcr, mouse Vav and S. cerevisiae Cdc24). Multiplesequence alignments were performed using the ClustalW program. The site of mutation is located within the RhoGEF domain in a highlyconserved region called CR3 and is marked with ‘611’ over the predicted amino-acid sequence of Ehs2-1p. Asterisks indicate identical aminoacids among all identified gene products. (.) and (:) indicate well-conserved and highly conserved amino acids, respectively.

6168

was rescued from the mutant strain GI 1 (ehs2-1) by gap repair,and the sequences of four different clones were analysed. Allshowed a cytosine-to-thymine change at position 1834. As acontrol for the experiment, two different rescued ehs2-1 cloneswere back-integrated into the leu1 locus of a ehs2-1 mutantstrain. We found that both integrants maintained the mutantphenotype for Ech hypersensitivity and heat sensitivity (datanot shown). The mutation predicts that proline 611 (amino acidnumbering) in the wild-type Rgf3p is converted to a serine inthe mutant Rgf3p (ehs2-1). Proline 611 is located in theRhoGEF domain that extends between amino acids 469 and653, and is one of the few residues conserved in all the proteinsof the RhoGEF family in S. pombe as well as in DH domainsof other GEFs such as human Vav, Bcr, Dbl, Tiam1 and Unc73(Fig. 2D) (Soisson et al., 1998). DH domains contain threeconserved blocks of sequences that have previously beenreferred to as conserved regions 1-3 (CR1-CR3) (Boguski andMcCormick, 1993; Soisson et al., 1998). These three conservedregions form three long helices (H1a, H2b and H8) that packtogether to form the core of the DH domain. Clustal alignment

of the DH domain of Rgf3p with DH domains of several GEFspredicts that proline 611 is located on helix H8 (CR3), whichis the most highly conserved region of the DH domain and towhich many mutations that decrease nucleotide exchangeactivity map (Soisson et al., 1998; Liu et al., 1998). This resultconfirms that rgf3+ is the gene affected in the ehs2-1 mutantand supports the hypothesis that Rgf3p may act as a GEF.

The rgf3+ gene is essential for cell viability and depletionof Rgf3p leads to a lysis phenotype similar to thedepletion of Rho1pRgf3p displays limited homology to yeast Rom1p and Rom2p,both GEFs of Rho1p in S. cerevisiae (Schmidt et al., 1997;Ozaki et al., 1996). Moreover, a mutant in rgf3+ (ehs2-1) isdefective in cell-wall biosynthesis. We therefore attempted todetermine whether Rgf3p is a GEF for Rho1p in S. pombe. Ifthis were indeed the case then the rgf3∆ mutant wouldpresumably show phenotypes similar to those of the rho1∆mutant. To investigate the phenotype resulting from complete

deletion of the rgf3+ gene, we constructed a diploidstrain of the genotype rgf3::ura4+/rgf3+ in which acopy of rgf3+ had been deleted and replaced by ura4+.Tetrad analysis revealed two viable and two nonviablespores, and all the viable spores produced ura–e

colonies (Fig. 3A). The viability of the rgf3::ura4+

mutant spores was not rescued by the presence of 1.2M sorbitol in the medium. Therefore, rgf3+ must beessential for cell viability and must also be requiredfor germination. To further characterize the terminalphenotype of the rgf3::ura4+ mutants, rgf3-null

Journal of Cell Science 117 (25)

Fig. 3. Rgf3p is essential for cell viability and depletion ofRgf3p leads to a lysis phenotype similar to the depletion ofRho1p. (A) Genomic organization of the rgf3+ and rgf1+

loci, and deletion strategy for rgf3+ disruption. Thedirection of transcription is indicated by an arrow. Tetradsfrom a rgf3::ura4+/rgf3+ strain dissected on YES mediumand incubated at 28°C for 4 days. (B) Terminal phenotypeof rgf3-null mutants. Spores prepared from the rgf3::ura4+

strain were inoculated in MM lacking uracil andgerminated for 18 hours. Cells were stained withrhodamine-conjugated phalloidin and DAPI to visualize F-actin and nuclei, respectively (top left) and with CalcofluorWhite (Cfw) to visualize the cell-wall material (top right).Spores with the wee1-50 rgf3∆ and sid2-250 rgf3∆ doublemutations (prepared from strains YSM654 and YSM656,respectively) were inoculated in YES medium andgerminated for 14 hours at 25°C and then for 6 hours at36°C. Cells were stained with Hoechst and Cfw (bottom).(C) Lethal phenotype of the P81 nmt-rgf3 and P41 nmt-rho1 shut-off mutants. Cells grown at 28°C in MM weresupplemented with thiamine to repress the nmt promoter.Nomarsky micrographs were taken after 12 hours in MMwith or without thiamine. (D) Growth phenotypes of P81nmt-rgf3 and P41 nmt-rho1mutants under different growthconditions. Strains VT88 (81 nmt-rgf3+ + pREP81X) andPPG217 (rho1∆ + pREP41X nmt-rho1+) were streakedonto several plate media (YES, YES + Sorbitol and MM-leu) and the plates were incubated for 3 days at 28°C. Thenmt promoter is off in rich medium (YES) and on in MM.Strain VT88 carried pREP81X, an empty plasmid, to allowcells to grow in MM-leu.

6169GEF for Rho1p in S. pombe

spores were germinated, fixed and stained to visualize F-actin,nuclei and septa. Germinated rgf3::ura4+ spores were capableof polarity establishment but appeared to be incapable offinishing the division process, becoming spherical at one end(Fig. 3B). Arrested cells showed two interphase nuclei and astable actomyosin ring, and most of them failed to assemble aseptum (Fig. 3B, Cfw-stained cells). This delay in cytokinesisresembles what has been termed a ‘cytokinesis checkpoint’.The cytokinesis checkpoint depends on a signalling pathwaycalled the septation initiation network (SIN) and the Wee1pkinase (Simanis, 2003; Rajagopalan et al., 2003). We foundthat, in both combinations of double mutants, sid2-250 rgf3∆and wee1-50 rgf3∆, elongated cells with multiple nuclei andmultiple septa were seen frequently during spore germinationindicating a bypass of the checkpoint (Fig. 3B, bottom).

It has been shown previously that fission yeast rho1+ is anessential gene and no conditional mutants are available(Nakano et al., 1997). However, experiments in which theRho1p cellular pool was depleted ended with massive cell lysis(shrinking cells) and actin depolymerization (Fig. 3C)(Arellano et al., 1997). To investigate the lack of function ofRgf3p during vegetative growth, we constructed a rgf3+ geneunder the control of the thiamine-regulatable and reduced-expression-rate nmt1 promoter P81nmt (Forsburg, 1993). Thisconstruct was integrated into the genome of a wild-typehaploid strain (MS38), the endogenous rgf3+ promoter beingreplaced by the P81nmt promoter. As shown in Fig. 3C, thecells displayed a normal cell morphology when rgf3+ wasexpressed in the absence of thiamine. 4 hours after the additionof thiamine to repress rgf3+ expression, a large proportion ofcells had shrunk and, after 9 hours, the whole culture had lysed(Fig. 3C). The phenotype of cells depleted for Rgf3p was verysimilar to that observed in the ehs2-1 mutant at the restrictivetemperature (Fig. 1A) and in cells depleted for Rho1p (Fig.3C). The same phenotype has also been reported (Nakano etal., 1997) in cells expressing the dominant-negative mutantRho1T20N. We next examined whether the Rgf3p shut-offphenotype could be rescued by osmotic support. As shown inFig. 3D, growth of the P81nmt-rgf3+ strain in rich medium(promoter off) was dependent on the presence of 1.2 Msorbitol, whereas Rho1p-depleted cells were unable to grow,regardless of the presence or the absence of 1.2 M sorbitol(Arellano et al., 1997). These results indicate that rgf3 mutantphenotypes are very similar to those of the Rho1p-depletedcells and suggest that Rgf3p and Rho1p function in the samesignal-transduction pathway. The fact that the Rgf3p switch offcould be rescued by sorbitol suggests that Rgf3p would controla subset of the functions of Rho1p, probably those related tocell-wall biosynthesis.

Hypersensitivity of the ehs2-1 mutant to cell-wall drugsis suppressed by overexpression of rho1+ but not otherGTPasesIf rgf3+ functions as an upstream regulator of rho1+,overexpression of rho1+ would be expected to suppress thehypersensitivity to Ech and Cfw as well as the temperature-sensitive growth phenotype of the ehs2-1 mutant. The GI 1strain (ehs2-1, leu 1-32, h+) was transformed with plasmidsbearing rho1+, rho2+, rho3+, rho4+, rho5+ and cdc42+ under thecontrol of the nmt1 promoter or with an empty vector as a

control. As shown in Fig. 4, the Ech hypersensitivity of theehs2-1 mutant was suppressed by rho1+ in minimal mediumwithout thiamine (promoter on). In medium with thiamine(promoter off), no suppression was observed. The rho1+ genewas also partially able to suppress the temperature-sensitivephenotype and the hypersensitivity to Cfw (data not shown).None of the other genes was able to suppress the phenotypes,this being consistent with the idea that rgf3+ would act in thesame pathway as rho1+ (Fig. 4). Overexpression of rho2+ inwild-type cells was lethal by itself, as well as in the ehs2-1mutant background (Fig. 4). To avoid this problem, we usedrho2+ driven by the P41nmt promoter (medium level). Thisconstruct produced viable cells. No complementation of thehypersensitivity to Ech or Cfw was found either (datanot shown). Interestingly, overexpression of cdc42+ wasdeleterious in a ehs2-1 background, whereas, in a wild-typebackground, it was perfectly viable (Fig. 4) (Miller andJohnson, 1994). It has recently been described that Rga8p, anovel Rho1-GAP, is an effector of Cdc42p, providing a linkbetween the Cdc42p and Rho1p signalling pathways (Yang etal., 2003).

Rgf3p specifically interacts with the GDP-bound form ofRho1p and promotes GDP-GTP exchangeUsing the yeast two-hybrid system, we investigated whetherRgf3 interacts with Rho1 or any of the Rho-family proteins.Plasmids for GTPases were kindly provided by P. Perez and

Fig. 4. Suppression of the echinocandin-hypersensensitive growthphenotype of the ehs2-1 mutant by overexpression of rho1+. MS38(rgf3+) was transformed with pREP3X (empty vector) and GI 1(ehs2-1/rgf3) was transformed with pREP3X-rho1 (rho1+), pREP3X-rho2 (rho2+), pREP3X-rho3 (rho3+), pREP3X-rho4 (rho4+),pREP3X-rho5 (rho5+), pREP3X-cdc42 (cdc42+) and pREP3X(empty vector). Transformants were streaked onto MM, MM plusthiamine, MM plus 1.5 µg ml–1 echinocandin and MM plus thiamineand 1.5 µg ml–1 echinocandin plates, and incubated at 32°C for 4days.

6170

P. M. Coll. For each Rho protein, point mutations that trappedthe GTPase in the GTP-bound form (rho1-G15VC199S) or theGDP-bound form (rho1-T20NC199S) fused to the DNA-binding domain were assayed. The entire ORF of rgf3+ wasfused to the transcriptional activation domain. The interactionwas examined by growth of the host on a –histidine platecontaining 40 mM 3-aminotriazole. As shown in Table 3 andFig. 5A, Rgf3p specifically interacted with the GDP-boundform of Rho1p (rho1-T20NC199S) but not with GTP-bound

Rho1p (rho1-G15VC199S). There was also interaction withRho5p bound to GDP (rho5-TC199S) but not with any of theother Rho proteins bound to GDP or GTP. The rho5+ gene isa new member of the Rho family of unknown function and isthe closest homologue to rho1+.

To investigate further the possible role of Rgf3p as an Rho1activator, we analysed the in vivo amount of GTP-bound Rho1pin cells with different amounts of Rgf3p. Wild-type cellscarrying the control plasmid pREP4X, ehs2-1 mutant cells(pREP4X) and wild-type cells overexpressing rgf3+ (carryingpREP4X-rgf3+) were transformed with plasmid pREP3X-HA-rho1. After induction of the nmt1 promoter for 18 hours, theamount of Rho1p bound to GTP was analysed by precipitationwith GST-C21RBD, the rhotekin-binding domain (which hadpreviously been obtained and purified from bacteria) andblotting with anti-HA antibody (Fig. 5B). Western blotsof whole-cell extracts (10 µg protein) showed that the totalamount of Rho1p was similar in wild-type, mutant and cellstransformed with pREP3X-rgf3+ (Fig. 5B). The amountof active Rho1p increased considerably in the strainoverexpressing rgf3+ compared with the control strain withnormal amounts of Rgf3p. No differences between the ehs2-1mutant and the wild type were observed, possibly because underthe conditions assayed (32°C) the mutant phenotype was not asstrong as it was at 37°C. As a control, we also analysed theamount of GTP-Rho2p in wild-type and ehs2-1 mutant cells andin cells overexpressing rgf3+ (Fig. 5B). These cells weretransformed with the plasmid pREP3X-HA-rho2 and GTP-bound Rho2p was pulled down from the extract by binding toGST-C12RBD. No changes in the levels of Rho2p bound toGTP were observed among the three strains (Fig. 5B). Theseresults indicate that Rgf3p acts as a specific Rho1p activator inS. pombe.

Overexpression of Rgf3p interferes with septation andincreases cell-wall synthesis.It has been shown that overexpression of rho1+ produces fourtypes of cell: swollen, branched, multiseptate and mixtures ofthese phenotypes. It has also been reported that both the cellwall and the septum are very thick in such cells (Arellanoet al., 1996; Nakano et al., 1997). We overproduced rgf3+

to determine whether the effect was similar to rho1+

overexpression or the overexpression of any of the other Rhoproteins. The rgf3+ gene was cloned under the thiamine-repressible nmt1 promoter in the pREP3X vector. After 20hours of induction, overexpression of rgf3+ produced cellscontaining multiple septa; the same phenotype has beendescribed before (Iwaki et al., 2003) (Fig. 6A). DAPI stainingrevealed that, in most multiseptate cells, each compartmentcontained one nucleus, indicative of a defect in cell separationafter septum assembly (not shown). Cfw mainly stains septa inwild-type S. pombe. Cells overexpressing rgf3+ showeda general increase in Cfw fluorescence, which was stillconcentrated in the septum. Therefore, we analysed thepossible role of Rgf3p as an activator of cell-wall biosynthesis.Because GS is one of the Rho1p effector proteins, we examinedthe activity of the GS in cells that overexpressed Rgf3p. Asexpected, an increase in enzymatic activity was detected incells overexpressing rgf3+ compared with the activity observedin the wild-type strain (Fig. 6B). Consistently, rgf3 mutant cells

Journal of Cell Science 117 (25)

Table 3. Two-hybrid analysis of the interactions betweendifferent Rho GTPases (pAS2) and Rgf3p (pACT2) used

as baitGene (pAS2)* rgf3 (pACT2) Empty (pACT2)

Empty + –rho1G15VC199S (GTP) + +rho1T20NC199S (GDP) +++ +rho2G17VC197S (GTP) ++ ++rho2T22NC197S (GDP) ++ ++rho3G24VC198S (GTP) ++ ++rho3T29NC198S (GDP) + +rho4G15VC198S (GTP) – –rho4T28NC198S (GDP) – –rho5G15VC197S (GTP) + +rho5T20NC197S (GDP) +++ +cdc42G12V∆C (GTP) – –cdc42T17N∆C (GDP) + +

* ‘GTP’ or ‘GDP’ indicates that this mutant emulates the GTP- or GDP-bound form, respectively.

Fig. 5. Rgf3p is a specific Rho1-GEF. (A) Rgf3 binds directly to theGDP-bound form of Rho1p (Rho1-T20N). Y190 cells expressing theindicated proteins were cultured on a SD plate with histidine (left) orwithout histidine plus 40 mM 3AT (right) at 30°C for 3 days. (B) TheRgf3p level modulates the amount of GTP-bound Rho1p in vivo.Wild-type (MS38) cells expressing pREP4X or pREP4X-rgf3, andehs2-1 (GI 1) mutant cells were transformed with either pREP3X-HA-rho1 or pREP3X-HA-rho2. GTP-Rho1p or GTP-Rho2p werepulled down from the cell extracts with GST-C21RBD and blottedagainst 12CA5, anti-HA monoclonal antibody.

6171GEF for Rho1p in S. pombe

(ehs2-1) showed a severe reduction (50%) in GS enzymaticactivity (Table 2), indicating that changes in Rgf3p levelscaused changes in GS activity.

To corroborate these results, we also studied the activity incells that overexpressed rho1+ and rgf3+ at the same time(transformed with the pREP3X-rho1 and pREP4X-rgf3plasmids). As described previously (Arellano et al., 1996),cells overexpressing rho1+ showed an increase in GS activity.This increase was considerably (ten times) higher in cells thatoverexpressed rgf3+ at the same time (Fig. 6B). These resultsclearly indicate that Rgf3p is involved in the regulation of β-1,3-glucan biosynthesis.

We also analysed the cell-wall composition of cells thatoverexpressed rgf3+, rho1+ or both. As shown in Fig. 6C, therewas an increase in the amount of β-glucan in cells thatoverexpressed rgf3+ compared with wild-type cells (16% and10%, respectively), and that increase was similar to that seenin cells that overexpressed rho1+ (15%). There was also ageneral increase in cell-wall biosynthesis in cells thatoverexpresed rgf3+ compared with wild-type and rho1+ cells(37%, 24.5% and 33%, respectively). In these cells, the ratiobetween β- and α-glucan fractions was the same as that foundin the wild-type S. pombe cells, indicating a simultaneousincrease in both α- and β-glucan polymers. Additionally, theamount of galactomannan was not significantly affected. Cellsthat overexpressed rgf3+ and rho1+ at the same time did notshow any further increase in β-glucan biosynthesis with respectto the overexpression of each gene separately. This could bedue to the limited amounts of other factors needed for cell-wallassembly. These results suggested that Rgf3p was specificallyactivating Rho1p, the GTPase that directly regulates thebiosynthesis of β-1,3-glucan and, through Pck2p, thebiosynthesis of the two main polymers α-1,3-glucan and β-1,3-glucan.

Rgf3p localizes to the septumTo gain further insight into the function of Rgf3p, wedetermined its subcellular localization. We constructed a Rgf3-EGFP fusion protein (at the 5′ end of the rgf3+ ORF) under thecontrol of the rgf3+ promoter (pVT-GFPrgf3). The GFP-Rgf3pfusion was functional and restored the ability of the ehs2-1mutant to grow in Ech and Cfw. Rgf3p localization wasexamined in strains carrying a EGFP-rgf3+ gene integrated atthe leu1 locus of a wild-type strain. The staining patternobserved was consistent with the localization of Rgf3p mainlyto the septum (Fig. 7A). EGFP-Rgf3p fluorescence appearedin the medial region even before the septum was stained withCfw (see enlarged cells in Fig. 7A, bottom). This stage wasvery transient. As the septum developed, the EGFP-Rgf3pfluorescence extended further towards the centre of the celluntil it formed a band across the cell. This band was notcontinuous (Fig. 7A), with dots of fluorescence being seen.Finally, as cell separation began by digestion of the primaryseptum, the EGFP-Rgf3p fluorescence began to disappear.These observations indicate that Rgf3p is targeted to thedeveloping septum early in the septation process and persiststhroughout cell separation. Some cells showed dots of greenfluorescence at one of the poles; this could reflect a smallamount of protein remaining there after cell separation. To testthe possibility that EGFP-Rgf3p concentrates at the cell endsin interphase cells, we analysed the localization of the proteinin a cdc25-22 mutant strain carrying the pVT-GFPrgf3plasmid. In cdc25-22 cells, which arrested in G2 phase at hightemperature with both ends growing, no Rgf3 fluorescence waspresent at the poles. The cells were then released (at 25°C)from the block (at 37°C) and the signal appeared in septatingcells (not shown). We also examined cells overexpressingEGFP-rgf3 (from the nmt1 promoter) to see whether the fusionwas also localized to other weakly stained structures but wefailed to detect any other cellular area to which it was localized(not shown).

Rgf3p was visualized only in cells with a developing

Fig. 6. Phenotypes of Rgf3p overexpression. (A) Micrographs ofCalcofluor White (Cfw) stained wild-type cells transformed withpREP3X (empty plasmid) or pREP3X-rgf3+ (rgf3+ overexpression)grown without thiamine for 20 hours. (B) In vitro glucan synthase(GS) activity assayed with the membrane fraction of wild-type cells(MS38) transformed with pREP3X, pREP4X-rgf3 (rgf3+

overexpression), pREP3X-rho1 (rho1+ overexpression) or bothpREP4X-rgf3 and pREP3X-rho1 (rgf3+ and rho1+ overexpression).Extracts were prepared from cells grown in MM without thiamine at32°C for 18 hours. Specific activity is expressed as milliunits per mgprotein. Values are the means of at least three independentexperiments with duplicated samples, and error bars representstandard deviations (s.d.). (C) Cell-wall composition in cells thatoverexpress rgf3+. The relative levels of [14C]-glucose radioactivityincorporated into each cell-wall polysaccharide are shown for thesame strains as above: wild-type (MS38) transformed with pREP3X,pREP4X-rgf3 (rgf3+ overexpression), pREP3X-rho1 (rho1+

overexpression) or both at the same time. Cells were grown in theabsence of thiamine for 18 hours and then [14C]-glucose was added 6hours before harvesting the cells. Values are the means of threeindependent experiments with duplicate samples. Standard deviationsfor total carbohydrate values are shown.

6172

septum. We therefore considered thepossibility that rgf3+ levels might beregulated in a cell-cycle-dependentmanner. Thus, we determined the levels ofrgf3+ mRNA in a synchronous culture,using a cdc25-22 strain. Cells weresynchronized as described above. Wefound that rgf3+ mRNA levels weresharply periodic, rising to a peak beforeseptation at 100 minutes, and decreasingwhen most of the cells had a septum (Fig.7B). Recently, a wide-ranging analysis ofcell-cycle periodic expression in S. pombehas shown that rgf3+ mRNA is periodicallytranscribed and its expression is dependenton Ace2p, a transcription factor that alsocontrols other genes with predicted rolesin cell division (Rustici et al., 2004)(http://www.sanger.ac.uk/). The results ofthese experiments show that both thelocalization of Rgf3p and the mRNA levelsfluctuate during the cell cycle, peakingduring septation.

DiscussionYeast morphogenesis and cell growthare coupled to the biosynthesis anddegradation of the cell wall. Therefore, allthese processes must be strictly controlledby, and linked to, general signal-transduction pathways (Ishiguro, 1998;Rajagopalan et al., 2003). Here, we used aclassical genetic approach to identify newS. pombe genes involved in maintainingcell-wall integrity. The rgf3+ gene wascloned by complementation of the ehs2-1mutant phenotypes. The ehs2-1 mutantcells were hypersensitive to the cell-wallinhibitors Ech and Cfw, and also displayeda thermosensitive lytic phenotype thatcould be suppressed by an osmoticstabilizer.

The predicted Rgf3p and Rgf1p (locatedadjacently in the chromosome) proteinsshow a RhoGEF domain characteristic ofproteins that act as GDP-GTP exchangefactors for Rho GTPases (Zheng, 2001;Schmidt and Hall, 2002; Hoffman andCerione, 2002). Genetic and biochemicalevidence reported here support the notion that Rgf3p is a GEFfor Rho1p. Disruption of rgf3+ and deletion of rho1+ wereunviable. Germinating spores with the rgf3::ura4+ doublemutation arrested as single elongated cells with no visibleseptum, and lethality was not rescued in the presence of 1.2 Msorbitol. Rgf3p depletion in vegetative cells caused cell lysis,with a morphology very similar to those of cells devoid ofRho1 or Pck1/2 activity. This suggests that the main functionof Rgf3p would be regulation of the Rho1p GTPase. Consistentwith this idea, the hypersensitivity to Ech (Fig. 4) andthe temperature-sensitive phenotype were suppressed by

overexpression of rho1+ but not of any other of the Rho genes.Interestingly, overexpression of cdc42+ was lethal in an ehs2-1 mutant background.

Our biochemical data strongly support the view that Rgf3pacts as a specific positive regulator of Rho1p. The full-lengthRgf3p interacted specifically with Rho1p in its GDP-boundstate but not with other Rho proteins (except for Rho5p), anda high level of Rgf3p increased the level of GTP-Rho1p invivo. Moreover, the phenotype seen in the ehs2-1 mutant cellsat the restrictive temperature (almost identical to the lack offunction of Rho1p) was due to a mutation located on helix H8

Journal of Cell Science 117 (25)

Fig. 7. Rgf3p localizes to the septum. (A) Wild-type cells were transformed with anintegrative plasmid expressing EGFP-rgf3+ under its own promoter (strain VT128). Cellswere grown at 28°C. (Top) EGFP-Rgf3p localization in a population of living cells atdifferent stages of the cell cycle is shown in the centre. The corresponding differentialinterference contrast images are shown on the left and Calcofluor White (Cfw) staining isshown on the right. (Bottom) A square with cells that had not yet developed a visibleseptum is enlarged. These cells already showed EGFP-Rgf3p dots of fluorescence. (B) Thergf3 mRNA levels were followed by northern-blot analyses in a cdc25-22 block-releaseexperiment. RNA samples were collected every 20 minutes along two consecutive cellcycles after release to 25°C. The blot was probed for rgf3 and for act1, the latter as anmRNA loading control.

6173GEF for Rho1p in S. pombe

(CR3), which is the most highly conserved region of the DHdomain and to which many mutations that decrease nucleotideexchange activity are mapped. Additionally, overexpression ofrgf3+ and rho1+ at the same time produces very refringent cellswith a phenotype similar to that of the constitutively activeallele Rho1G15V (Arellano et al., 1996) (data not shown).

The experiments reported in this study indicate that rgf3+ isinvolved in the regulation of cell wall biosynthesis and cellintegrity. The rgf3 mutant cells (ehs2-1 mutant) werehypersensitive to cell-wall antifungal drugs and showed atemperature-sensitive lytic phenotype that could be rescued bythe presence of 1.2 M sorbitol. Cells with a mutation in thergf3 gene (ehs2-1) were defective in GS activity at 37°C andcells that overexpressed rgf3+ showed a GS activity that wastwice that of wild-type cells and similar to the GS activity incells overexpressing Rho1p (Calonge et al., 2003).Furthermore, cells overexpressing rgf3+ together with rho1+

showed a huge increase in GS activity (approximately seven-to tenfold) compared with the wild-type level. Even withoutGTP added to the reaction, the GS activity was seven timeshigher than in the wild type, indicating that an excess of Rgf3phad raised the intracellular pool of GTP-bound Rho1p (alreadyactivated). Regarding cell-wall composition, an increase in theamount of Rgf3p simultaneously increased the amount of α-and β-glucan polymers. This is consistent with the notion thatRho1p binds and activates Pck2p, which in turn activates thesynthesis of the two main structural polymers of the cell wall:α- and β-glucans (Arellano et al., 1999; Calonge et al., 2000).

In previous work, we reported that the rgf3 mutation (ehs2-1 mutation) was suppressed by bgs3+, a putative β-1,3-GSsubunit. Multiple copies of bgs3+ complemented thehypersensitivity to Ech and Cfw but not the temperature-sensitive phenotype (Martin et al., 2003). The cps1+/bgs1+ andbgs2+ gene, which encode the other GS subunit homologues,also suppressed the Ech hypersensitivity but not the Cfw ortemperature-sensitive phenotypes. These genes, which actdownstream from rgf3+, are key components in the final stepsof β-glucan biosynthesis, and their suppression providesevidence that one of the main functions of rgf3+ is β-1,3-glucansynthesis activation. The fact that a high dose of Mok1p (theα-glucan synthase) was not able to complement any of theehs2-1 mutant phenotypes supports the hypothesis that β-1,3-glucan is the most important polymer affected in the cell wallof the rgf3 mutants.

Previous studies have shown that Rho1p depletion causescell death concomitant with a decrease in β-1,3-GS activity.Lysis is not prevented by an osmotic stabilizer and occursmainly after cytokinesis (Arellano et al., 1997), probablybecause correct cell-wall assembly is essential at that point ofthe cell cycle. Here, we found that the cell-lysis phenotypeproduced by Rgf3p depletion was prevented by 1.2 M sorbitol.Furthermore, the protein localized to the septum region in theearly stages of cytokinesis and remained there until cellseparation. The model that we propose considers that Rgf3pwould activate Rho1p during cytokinesis, when cell-wallintegrity is compromised. It is possible that a spatial ortemporal localization of Rgf3p might be necessary for the localproduction of an active (GTP-bound) form of Rho1p. In theabsence of Rgf3p, but in the presence of osmotic support,Rho1p could be activated in some other ways.

GS activity must be strictly regulated in time and in

synchrony with the cell cycle, and Rho1p might be the finalcomponent of a GTPase cascade linking cell-cycle controls tocell-wall biosynthesis. Our data support this hypothesis. Thergf3+ mRNA levels peaked during septation and the proteinaccumulated at the contractile ring. It is known that Bgs1p isthe GS involved in primary septum assembly and that mutantsin bgs1 can engage the cytokinetic checkpoint. Several mutantsin bgs1+ (cps1-12, drc1-191) arrest with two interphase nucleiand a stable actomyosin ring (Le Goff et al., 1999; Liu et al.,1999). We found that germinating spores lacking Rgf3pshowed a similar phenotype with stable actomyosin rings andmost of them lack a septum (Fig. 3B), suggesting that the lackof Rgf3p activates the cytokinetic checkpoint. In fact, sporesfrom a double mutant such as wee1-50 ∆rgf3 or sid2-250 ∆rgf3formed several nuclei and septa, suggesting that they canbypass the septation checkpoint (Fig. 3B) (Simanis, 2003;Rajagopalan et al., 2003). Identification of upstream regulatorsof rgf3+ will be necessary to understand how Rho1p regulatescell-wall integrity during cytokinesis in fission yeast.

We thank P. Perez, P. Coll and V. Martin for plasmids, strains andall types of help throughout the work, and T. Toda and H. Schmidtfor strains. A. Durán, H. Valdivieso, J. C. Ribas and C. Roncero areacknowledged for helpful discussions. V. Tajadura acknowledgessupport from a fellowship granted by the MEC, Spain, and I. Garcíawas supported by a fellowship from the Junta de Castilla y León. Thiswork was supported by grants BIO2001-1663 from the ComisiónInterministerial de Ciencia y Tecnología, Spain (CSI7/01) from theJunta de Castilla y León.

ReferencesArellano, M., Duran, A. and Pérez, P. (1996). Rho1 GTPase activates the (1-

3)β-D-glucan synthase and is involved in Schizosaccharomyces pombemorphogenesis. EMBO J. 15, 4584-4591.

Arellano, M., Duran, A. and Perez, P. (1997). Localization of theSchizosaccharomyces pombe Rho1 GTPase and its involvement in theorganization of the actin cytoskeleton. J. Cell. Sci. 110, 2547-2555.

Arellano, M., Valdivieso, M. H., Calonge, T. M., Coll, P. M., Duran, A. andPerez, P. (1999). Schizosaccharomyces pombe protein kinase Chomologues, Pck1p and Pck2p, are targets of Rho1p and Rho2p anddifferentially regulate cell integrity. J. Cell Sci. 112, 3569-3578.

Balasubramanian, M., McCollum, D. and Gould, K. L. (1997). Cytokinesisin fission yeast Schizosaccharomyces pombe. Methods Enzymol. 283, 494-506.

Boguski, M. S. and McCormick, F. (1993). Proteins regulating Ras and itsrelatives. Nature 366, 643-654.

Burridge, K. and Wennerberger, K. (2004). Rho and Rac take centre stage.Cell 116, 167-179.

Calonge, T. M., Nakano, K., Arellano, M., Arai, R., Katayama, S. Toda,T., Mabuchi, I. and Perez, P. (2000). Schizosaccharomyces pombe Rho2GTPase regulates the cell wall α-glucan biosynthesis, through the proteinkinase Pck2. Mol. Biol. Cell 11, 4393-4401.

Calonge, T. M., Arellano, M., Coll, P. M. and Perez, P. (2003). Rga5p is aspecific Rho1p GTPase-activating protein that regulates cell integrity inSchizosaccharomyces pombe. Mol. Microbiol. 47, 507-518.

Carnero, E., Ribas, J. C., Garcia, B., Duran, A. and Sanchez, Y. (2000).Schizosaccharomyces pombe Ehs1p is involved in maintaining cell wallintegrity and in calcium uptake. Mol. Gen. Genet. 264, 173-183.

Chang, E. C., Barr, M., Wang, Y., Jung, V., Xu, H. P. and Wigler, M. H.(1994). Cooperative interaction of S. pombe proteins required for matingand morphogenesis. Cell 79, 131-141.

Coll, P. M., Trillo, Y., Ametzazurra, A. and Perez, P. (2003). Gef1p, a newguanine nucleotide exchange factor for Cdc42p, regulates polarity inSchizosaccharomyces pombe. Mol. Biol. Cell 14, 313-323.

Cortes, J. C., Ishiguro, J., Duran, A. and Ribas, J. C. (2002). Localizationof the (1,3)beta-D-glucan synthase catalytic subunit homologueBgs1p/Cps1p from fission yeast suggests that it is involved in septation,

6174

polarized growth, mating, spore wall formation and spore germination. J.Cell Sci. 115, 4081-4096.

Cortés, J. C. G., Carnero, E., Ishiguro, J., Sánchez, Y., Durán, A. andRibas, J. C. (2005). The novel (1,3)β-D-glucan synthase catalytic subunitBgs4p from fission yeast is essential during both cytokinesis and polarizedgrowth. J. Cell Sci. (in press).

Craven, R. A., Griffiths, D. J., Sheldrick, K. S., Randall, R. E., Hagan, I.M. and Carr, A. M. (1998). Vectors for the expression of tagged proteinsin Schizosaccharomyces pombe. Gene 221, 59-68.

Duran, A. and P. Perez. (2004). Cell wall synthesis. In The Molecular Biologyof Schizosaccharomyces pombe. Genetics, Genomics And Beyond (ed. R.Egel), pp. 269-276. New York, NY: Springer.

Durfee, T., Becherer, K., Chen, P.-L., Yeh, S.-H., Yang, Y., Kilburn, A. E.,Lee, W.-H. and Elledge, S. J. (1993). The retinoblastoma protein associateswith the protein phosphatase type 1 catalytic subunit. Genes Dev. 7, 555-569.

Forsburg, S. L. (1993). Comparison of Schizosaccharomyces pombeexpression systems. Nucleic Acids Res. 21, 2955-2956.

Hirata, D., Nakano, K., Fukui, M., Takenaka, H., Miyakawa, T. andMabuchi, I. (1998). Genes that cause aberrant cell morphology byoverexpression in fission yeast: a role for a small GTP-binding protein Rho2in cell morphogenesis. J. Cell. Sci. 111, 149-159.

Hirota, K., Tanaka, K., Ohta, K. and Yamamoto, M. (2003). Gef1p andScd1p, the two GDP-GTP exchange factors for Cdc42p, form a ringstructure that shrinks during cytokinesis in Schizosaccharomyces pombe.Mol. Biol. Cell 14, 3617-3627.

Hoffman, G. R. and Cerione, R. (2002). Signaling to the Rho GTPases:networking with the DH domain. FEBS Lett. 513, 85-91.

Ishiguro, J. (1998). Genetic control of fission yeast cell wall synthesis: thegenes involved in wall biogenesis and their interactions inSchizosaccharomyces pombe. Genes Genet. Syst. 73, 181-191.

Iwaki, N., Karatsu, K. and Miyamoto, M. (2003). Role of guaninenucleotide exchange factors for Rho family GTPases in the regulation ofcell morphology and actin cytoskeleton in fission yeast. Biochem. Biophys.Res. Commun. 312, 414-420.

Keeney, J. B. and Boeke, J. D. (1994). Efficient targeted integration at leu1-32 and ura4-294 in Schizosaccharomyces pombe. Genetics 136, 849-856.

Klis, F. M. (1994). Cell wall assembly in yeast. Yeast 10, 851-869.Le Goff, X., Woollard, A. and Simanis, V. (1999). Analysis of the cps1 gene

provides evidence for a septation checkpoint in Schizosaccharomycespombe. Mol. Gen. Genet. 262, 163-172.

Letunic, I., Goodstadt, L., Dickens, N., Doerks, T., Schultz, J., Mott, R.,Ciccarelli, F., Copley, R., Ponting, C. and Bork, P. (2002). Recentimprovements to the SMART domain-based sequence annotation resource.Nucleic Acids Res. 30, 242-244.

Liu, X., Wang, H., Eberstadt, M., Schnuchel, A., Olejniczak, E. T.,Meadows, R. P., Schkeryantz, J. M., Janowick, D. A., Harlan, J. E.,Harris, E. A. et al. (1998). NMR structure and mutagenesis of the N-terminal Dbl homology domain of the nucleotide exchange factor Trio. Cell95, 269-277.

Liu, J., Tang, X., Wang, H. and Balasubramanian, M. K. (2000a). Bgs2p,a 1,3-β-glucan synthase subunit is essential for maturation of ascospore wallin Schizosaccharomyces pombe. FEBS Lett. 478, 105-108.

Liu, J., Wang, H. and Balasubramanian, M. K. (2000b). A checkpoint thatmonitors cytokinesis in Schizosaccharomyces pombe. J. Cell Sci. 113, 1223-1230.

Liu, J., Tang, X., Wang, H., Oliferenko, S. and Balasubramanian, M. K.(2002). The localization of the integral membrane protein Cps1p to the celldivision site is dependent on the actomyosin ring and the septation-inducingnetwork in Schizosaccharomyces pombe. Mol. Biol. Cell 13, 989-1000.

Mackay, D. J. and Hall, A. (1998). Rho GTPases. J. Biol. Chem. 273, 20685-20687.

Martin, V., Ribas, J. C., Carnero, E., Durán, A. and Sánchez, Y. (2000).Bgs2+, a sporulation-specific glucan synthase homologue is required forproper ascospore wall maturation in fission yeast. Mol. Microbiol. 38, 308-321.

Martin, V., Garcia, B., Carnero, E., Duran, A. and Sanchez, Y. (2003).Bgs3p, a putative 1,3-β-glucan synthase subunit, is required for cell wallassembly in Schizosaccharomyces pombe. Eukaryotic Cell 2, 159-169.

Miller, P. J. and Johnson, D. I. (1994). Cdc42p GTPase is involved incontrolling polarized growth in Schizosaccharomyces pombe. Mol. Cell.Biol. 14, 1075-1083.

Moreno, S., Klar, A. and Nurse, P. (1991). Molecular genetic analysis of thefission yeast Schizosaccharomyces pombe. Methods Enzymol. 194, 795-823.

Nakano, K., Arai, R. and Mabuchi, I. (1997). The small GTP binding protein

Rho1 is a multifunctional protein that regulates actin localization, cellpolarity, and septum formation in the fission yeast Schizosaccharomycespombe. Genes Cells 2, 679-694.

Nakano, K., Mutoh, T. and Mabuchi, I. (2001). Characterization of GTPase-activating proteins for the function of the Rho-family small GTPases in thefission yeast Schizosaccharomyces pombe. Genes Cells 6, 1031-1042.

Nakano, K., Imai, J., Arai, R., Toh-E, A., Matsui, Y. and Mabuchi, I.(2002). The small GTPase Rho3 and the diaphanous/formin For3 functionin polarized cell growth in fission yeast. J. Cell Sci. 115, 4629-4639.

Nakano, K., Mutoh, T., Arai, R. and Mabuchi, I. (2003). The small GTPaseRho4 is involved in controlling cell morphology and septation in fissionyeast. Genes Cells 8, 357-370.

Orr-Weaver, T. K., Szostack, J. W. and Rothstein, R. J. (1991). Geneticapplications of yeast transformation with linear and gapped plasmids.Methods Enzymol. 101, 228-245.

Osumi, M., Yamada, N., Kobori, H., Taki, A., Naito, N., Baba, M. andNagatani, T. (1989). Cell wall formation in regenerating protoplasts ofSchizosaccharomyces pombe: study by high resolution, low voltagescanning electron microscopy. J. Electron Microsc. 38, 457-468.

Ozaki, K., Tanaka, K., Imamura, H., Hihara, T., Kameyama, T., Nonaka,H., Hirano, H., Matsuura, Y. and Takai, Y. (1996). Rom1p and Rom2pare GDP/GTP exchange proteins (GEPs) for the Rho1p small GTP bindingprotein in Saccharomyces cerevisiae. EMBO J. 15, 2196-2207.

Radding, J. A., Heidler, S. A. and Turner, W. W. (1998). Photoaffinityanalog of the semisinthetic echinocandin LY303366: identification ofechinocandin targets in Candida albicans. Antimicrob. Agents Chemother.42, 1187-1194.

Rajagopalan, S., Wachtler, V. and Balasubramanian, M. (2003).Cytokinesis in fission yeast. a story of rings, rafts and walls. Trends Genet.19, 403-408.

Reid, T., Furiyashiki, T., Ishizaki, T., Watanabe, G., Watanabe, N.,Fujisawa, K., Morii, N., Madaule, P. and Nayumira, S. (1996). Rhotekin,a new putative target for Rho-bearing homology to a serine/threonine kinase,PKN, and rhophilin in the Rho-binding domain. J. Biol. Chem. 271, 13556-13560.

Ren, X. D., Kiosses, W. B. and Schwartz, M. A. (1999). Regulation of thesmall GTP-binding protein Rho by cell adhesion and the cytoskeleton.EMBO J. 18, 578-585.

Roh, D.-H., Bowers, B., Riezman, H. and Cabib, E. (2002). Rho1pmutations specific for regulation of β(1,3)glucan synthesis and the order ofassembly of the yeast cell wall. Mol. Microbiol. 44, 1167-1183.

Rustici, G., Mata, J., Kivinen, K., Lio, P., Penkett, C. J., Burns, G., Hayles,J., Brazma, A., Nurse, P. and Bahler, J. (2004). Periodic gene expressionprogram of the fission yeast cell cycle. Nat. Genet. 36, 809-817.

Santos, B., Gutierrez, J., Calonge, T. M. and Perez, P. (2003). Novel RhoGTPase involved in cytokinesis and cell wall integrity in the fission yeastSchizosaccharomyces pombe. Eukaryotic Cell 2, 521-533.

Sayers, L. G., Katayama, S., Nakano, K., Mellor, H., Mabuchi, I., Toda,T. and Parker, P. J. (2000). Rho-dependence of Schizosaccharomycespombe Pck2. Genes Cells 5, 17-27.

Schmidt, H. (1993). Effective long-range mapping in Schizosaccharomycespombe with the help of swi5. Curr. Genet. 24, 271-273.

Schmidt, A. and Hall, A. (2002). Guanine nucleotide exchange factors forRho GTPases: turning on the switch. Genes Dev. 16, 1587-1609.

Schmidt, A., Bickle, M., Beck, T. and Hall, M. N. (1997). The yeastphosphatidylinositol kinase homolog TOR2 activates RHO1 and RHO2 viathe exchange factor ROM2. Cell 88, 531-542.

Simanis, V. (2003). Events at the end of mitosis in the budding and fissionyeast. J. Cell Sci. 116, 4263-4275.

Soisson, S. M., Nimnual, A. S., Uy, M., Bar-Sagi, D. and Kuriyan, J. (1998).Crystal structure of the Dbl and pleckstrin homology domains from thehuman SON OF SEVENLESS protein. Cell 95, 259-268.

Takai, Y., Sasaki, T. and Matozaki, T. (2001). Small GTP-binding proteins.Phys. Rev. 81, 153-208.

Wang, H., Tang, X. and Balasubramanian, M. K. (2003). Rho3p regulatescell separation by modulating exocyst function in Schizosaccharomycespombe. Genetics 164, 1323-1331.

Yang, P., Qyang, Y., Bartholomeusz, G., Zhou, Z. and Marcus, S. (2003).The novel Rho1 GTPase-activating protein family, Rga8, provides apotential link between Cdc42/p21-activated kinase and Rho signalingpathways in the fission yeast, Schizosaccharomyces pombe. J. Biol. Chem.278, 48821-48830.

Zheng, Y. (2001). Dbl family guanine nucleotide exchange factors. TrendsBiochem. Sci. 26, 274-732.

Journal of Cell Science 117 (25)