EUKARYOTIC CELL, June 2002, p. 420–431 Vol. 1, No. 31535-9778/02/$04.00�0 DOI: 10.1128/EC.1.3.420–431.2002Copyright © 2002, American Society for Microbiology. All Rights Reserved.

The Golgi GDPase of the Fungal Pathogen Candida albicans AffectsMorphogenesis, Glycosylation, and Cell Wall Properties

Ana B. Herrero,1 Daniela Uccelletti,1 Carlos B. Hirschberg,1Angel Dominguez,2 and Claudia Abeijon1*

Department of Molecular and Cell Biology, Boston University Goldman School of Dental Medicine, Boston, Massachusetts 02118,1

and Departamento de Microbiologia y Genetica, IMB/CSIC, Universidad de Salamanca, 37007 Salamanca, Spain2

Received 20 December 2001/Accepted 11 April 2002

Cell wall mannoproteins are largely responsible for the adhesive properties and immunomodulation abilityof the fungal pathogen Candida albicans. The outer chain extension of yeast mannoproteins occurs in the lumenof the Golgi apparatus. GDP-mannose must first be transported from the cytosol into the Golgi lumen, wheremannose is transferred to mannans. GDP is hydrolyzed by a GDPase, encoded by GDA1, to GMP, which thenexits the Golgi lumen in a coupled, equimolar exchange with cytosolic GDP-mannose. We isolated anddisrupted the C. albicans homologue of the Saccharomyces cerevisiae GDA1 gene in order to investigate its rolein protein mannosylation and pathogenesis. CaGda1p shares four apyrase conserved regions with othernucleoside diphosphatases. Membranes prepared from the C. albicans disrupted gda1/gda1 strain had a 90%decrease in the ability to hydrolyze GDP compared to wild type. The gda1/gda1 mutants showed a severe defectin O-mannosylation and reduced cell wall phosphate content. Other cell wall-related phenotypes are present,such as elevated chitin levels and increased susceptibility to attack by �-1,3-glucanases. Our results show thatthe C. albicans organism contains �-mannose at their nonreducing end, differing from S. cerevisiae, which hasonly �-linked mannose residues in its O-glycans. Mutants lacking both alleles of GDA1 grow at the same rateas the wild type but are partially blocked in hyphal formation in Lee solid medium and during induction inliquid by changes in temperature and pH. However, the mutants still form normal hyphae in the presence ofserum and N-acetylglucosamine and do not change their adherence to HeLa cells. Taken together, our data arein agreement with the hypothesis that several pathways regulate the yeast-hypha transition. Gda1/gda1 cellsoffer a model for discriminating among them.

Candida albicans, one of the most frequently isolated fungalhuman pathogens, can cause superficial and systemic infec-tions. Several virulence factors have been described for thisorganism: the secretion of lytic enzymes (28), the ability toundergo the transition from yeast to hyphal growth known asdimorphism (12, 38, 16, 51), and the ability to adhere to hostcells and to penetrate epithelial and endothelial cells. Theadhesion capacity of C. albicans depends on mannoproteinspresent in its cell wall. Recent studies from several laboratorieshave shown that glycosylated outer cell wall mannoproteinsform direct interactions with the host (14, 54) and thereforeglycosylation defects are important for virulence (11, 56). Inyeast, the first O-glycosylation step and the addition of a coreN-linked carbohydrate occurs in the endoplasmic reticulum;and then the glycoproteins move to the Golgi apparatus, whereelongation of O-linked sugar chains and processing of complexN-linked oligosaccharide structures take place (22, 53). In Sac-charomyces cerevisiae, many of the steps have been fully char-acterized at both the biochemical and genetic levels (22, 10).However, for C. albicans only a few genes—MNT1 (11), PMT1(55), and PMT6 (56)—involved in the O-glycosylation pathwayhave been isolated. Deletion of any of the three showed strongalterations of virulence in animal models.

Terminal mannosylation of yeast proteins and lipids occursin the lumen of the Golgi apparatus. The sugar donor for these

reactions, GDP-mannose, must first be transported by a spe-cific transporter from the cytosol, its site of synthesis, into theGolgi lumen, where mannose is transferred to mannans byspecific mannosyltransferases (27). The other reaction prod-uct, GDP, is then hydrolyzed by a GDPase (Gda1p) to GMP,which then exits the Golgi lumen in a coupled, equimolarexchange with cytosolic GDP-mannose (2). This transport/an-tiport cycle was originally described in vitro with Golgi vesiclesfrom rat liver (13). Evidence for its physiological relevance hasbeen obtained in vivo and in vitro with yeast as well as withnematodes and mammals (4, 6). Recently the molecular defectcausing the human disease leukocyte adhesion deficiency syn-drome type II was localized to the gene encoding the GolgiGDP-fucose transporter (26).

S. cerevisiae Gda1p is very active toward GDP, moderatelyactive toward UDP, and inactive toward ADP or any tri- ormonophosphate (59). Deletion of this gene results in markedlyreduced Golgi mannosylation of proteins and lipids in vivo anddecreases fivefold the rate of GDP-mannose entry into Golgivesicles compared to results with the wild type (5). TheKluyveromyces lactis GDA1 gene has also been isolated andcharacterized. Loss of function of the gene results in differentdefects in glycosylation, osmotic stability, and cell wall polymercomposition in the two yeast species (35). Another Golgi nu-cleoside diphosphatase, encoded by the YND1 gene, was re-cently characterized for S. cerevisiae (21, 60). This phosphatasehas a broader spectrum of specificity; nevertheless, it is par-tially redundant with Gda1p in function. The ynd1gda1 doublemutant has more severe glycosylation phenotypes than any of

* Corresponding author. Mailing address: 700 Albany St., W 202A,Boston, MA 02118. Phone: (617) 414-1045. Fax: (617) 414-1041. E-mail: cabeijon@bu.edu.

420

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

the individual mutants (21). It is clear that there is regulationof the glycosylation process at the level of antiporter genera-tion, but the precise relationship between Gda1p and Ynd1p isnot yet understood.

The cell wall glycoproteins of fungal pathogens, such as C.albicans, are recognized during host invasion and modulate theimmune response. Therefore, studying enzymes regulating theglycosylation process in these fungi could help in understand-ing mechanisms of host defense. To determine the in vivo roleof C. albicans Golgi GDPase, the C. albicans GDA1 gene,encoding a protein highly similar to S. cerevisiae and K. lactisGda1p, was cloned, and null mutants were constructed. Thehomozygous C. albicans gda1/gda1 strain was viable andshowed drastically reduced in vitro membrane bound GDPaseactivity. We localized C. albicans Gda1p to the Golgi anddemonstrated that it is implicated in cell wall biogenesis, hy-phal formation, and O-mannosylation.

MATERIALS AND METHODS

Strains, media, and growth conditions. The yeast strains utilized in this studyand their genotypes are listed in Table 1. Strains were grown in yeast extract-peptone-dextrose (YEPD) or synthetic dextrose medium (48), which for Ura�

strains was supplemented with 50 �g of uridine/ml. Solid medium was obtainedby adding agar (2%). Solid medium for inducing the yeast-hypha transition in C.albicans was Lee medium in which glucose was replaced by mannitol (1.25%).The dimorphic transition was induced in Lee medium by changing the temper-ature to 37°C, adding 4% bovine calf serum (GIBCO), or changing the carbonsource (glucose was replaced by 1.25% N-acetylglucosamine [GlcNAc]).

Disruption of GDA1. The plasmid pGDA1 contains the Candida GDA1 genein a 4,120-bp fragment subcloned into pBluescript. The plasmid pGDA1 wasused as a template for “divergent” PCR with the oligonucleotides AH1 (5�-AAAGAATGGTTAAAGAGCTCGAAGCTAGTG) and AH2 (5�-TTTTGGAAGATCTATTTTTCCAGTTGAATC). The oligonucleotide AH1 corresponds tonucleotides �667 to �638 relative to the ATG in the GDA1 sequence and alsoadds a SacI site to the PCR product, while the oligonucleotide AH2 correspondsto nucleotides �103 to �132 and adds a BglII site. The 456-bp PCR fragmentwas cut with SacI and BglII and ligated to the plasmid pMB7 (20), previouslydigested with SacI and BglII. The resulting plasmid was pGDA�1. The plasmidpGDA1 was used again as a template for divergent PCR with the oligonucleo-tides AH3 (5�-CGTATATGTCGACACAAGTTACATTTTAGA) and AH4 (5�-CGTTTGTGTGTGATTGGCAAGCTTTATTGT). The oligonucleotide AH3correspond to nucleotides �31 to �60 relative to the stop codon in the GDA1sequence and also adds a SalI site to the PCR product, while the oligonucleotideAH4 correspond to nucleotides �803 to �832 and adds a HindIII site. The801-bp PCR fragment was cut with SalI and HindIII and ligated to the plasmidpGDA�1, previously digested with SalI and HindIII. From the resulting plasmid,pGDA�2, a 5.2-kb SacI-HindIII fragment was isolated and used to transformstrain CAI4. Correct integration of the cassette into the GDA1 locus of the Ura�

transformants was verified by Southern blot analysis. Spontaneous Ura� deriv-atives of one of the heterozygous disruptants were selected on medium contain-

ing 5-fluoro-orotic acid (U.S. Biological). These clones were screened by South-ern blot hybridization to identify those which had lost the URA3 gene viaintrachromosomal recombination mediated by the hisG repeats. The procedurewas then repeated to delete the remaining functional allele of GDA1. The GDA1gene was then reintroduced into BAB4 by transforming this strain with theplasmid p1041GDA1. This plasmid was constructed by inserting the 4,120-bpBamHI fragment obtained from the plasmid pGDA1 into the BamHI site of theC. albicans p1041 plasmid (23). Although p1041 was described as an integrativeplasmid, we strongly suspect that it integrated into the genome of the mutantgda1 strains because we could neither induce plasmid loss nor recover plasmidfrom the reconstituted strains.

Complementation analysis. To express CaGDA1 in S. cerevisiae gda1�, thep426-CaGDA1 plasmid was constructed. The CaGDA1 open reading frame wasamplified by utilizing pGDA1 as a template with S (5�-GGATTCGTTTCATAAATGATAAACCC) and A (5�-CGTAATATGTTTTGAAATTAAAACTTC)as primers. The amplified fragment was cloned into the pCRII-TOPO vector(Invitrogen), and the resulting plasmid, pTCaG, was sequenced. The CaGDA1open reading frame was subcloned from pTCaG to p426 (39) using XhoI andKpnI restriction sites for 5� and 3� ends, respectively. The S. cerevisiae null mutantwas then transformed with p426-CaGDA1 for functional analysis experiments.

Production of antibodies against CaGda1p. The pMAL Protein Fusion andPurification system (New England Biolabs) was used for expressing and purifyinga fragment of CaGda1p. The EcoRI fragment corresponding to nucleotides�579 to �1436 relative to CaGDA1 ATG was subcloned from pGDA1 topMAL-c2X. This construction resulted in a fusion protein between the 5� end ofthe malE gene and the fragment from CaGDA1. The resulting plasmid, pMAL-GDA1, was then used for production of recombinant protein (maltose bindingprotein) according to the manufacturers’ instructions. About 2 mg of fusionprotein was sent to Lampire Biologicals for the production of rabbit polyclonalantibodies.

Preparation of vesicle fractions. Vesicle fractions were prepared from yeastcells (1.5 liters) grown to exponential phase in Lee Medium and from hypha cellsinduced by a change in the pH of Lee Medium over a 16-h period. After beingwashed with cold l0 mM sodium azide as previously described (2), cells wereresuspended in spheroplast buffer (50 mM potassium phosphate [pH 7.5], 1.4 Msorbitol, 10 mM sodium azide, 0.3% �-mercaptoethanol) containing 0.5 mg ofZymolyase 100T (Seikagaku)/ml and incubated at 37°C for 40 min. The sphero-plast suspension was centrifuged, and the pellet was washed with buffer 1 (10 mMtriethanolamine acetic acid [pH 7.2], 0.8 M sorbitol, 1 mM EDTA, 1 �g ofleupeptin/ml, 0.5 mM phenylmethylsulfonyl fluoride) and resuspended in thesame buffer. Spheroplasts were then broken by passage through a narrow-bore10-ml glass pipette, diluted in membrane buffer, and centrifuged at 1,000 � g for10 min to obtain P1 “The supernatant was centrifuged at 100,000 � g for 30 minto obtain P100” Vesicles were stored at �70°C. Protein was measured using theBCA method (Bio-Rad).

Nucleotide diphosphate hydrolysis assay. Hydrolysis of GDP, UDP, and ADPwas measured in vesicle fractions as previously described (59) in buffer contain-ing 0.2 M imidazole [pH 7.5], 10 mM CaC12 or 10 mM MnC12, 0.1% TritonX-100, and 2 mM GDP, UDP, or ADP. One hundred microliters of this solution,containing 5 to 10 �g of sample protein, was incubated at 30°C for 30 min. Thereaction was stopped by transferring the tubes to ice and adding 10 �l of 10%sodium dodecyl sulfate (SDS). To determine the amount of phosphate releasedduring hydrolysis, 200 �l of water and 700 �l of AMES reagent (1:6 mixture of10% ascorbic acid and 0.42% ammonium molybdate in 1 N sulfuric acid) were

TABLE 1. Yeast strains used in this study

Strain Genotype or description Reference or source

C. albicansSC5314 PrototrophicCAI4 �ura3::imm434/�ura3::imm434 Fonzi and Irwin (20)SS19-4 ura3/ura3 �mnn9::hisG/mnn9::hisG Southard et al. (50)BAB1 ura3/ura3 GDA1/gda1::hisG-URA3-hisG This workBAB2 ura3/ura3 GDA1/�gda1::hisG This workBAB3 ura3/ura3gda1::hisG-URA3-hisG/�gda1::hisG This workBAB4 ura3/ura3 �gda1::hisG/�gda1::hisG This work

S. cerevisiaeG2-6 MAT� ura3-52 lys2-801 am ade2-101 oc trp1-�1 his3-�200 leu2-�1 Abeijon et al. (2)G2-7 MAT� ura3-52 lys2-801 am ade2-101 oc trp1-�1 his3-�200 leu2-�1 gda1::LEU2 Abeijon et al. (2)

VOL. 1, 2002 CANDIDA GOLGI GDPase 421

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

added to each tube; following incubation at 45°C for 20 min, absorbance wasmeasured at 660 nm.

�-1,2-mannosyltransferase assays. Reactions were performed in a 50-�l finalvolume containing 2 to 4 �g of membrane protein, 50 mM HEPES (pH 7.2),0.1% Triton X-100, 10 mM MnCl2, and 10 to 200 �M GDP- [3H]mannose (0.1�Ci). �-Methyl-D-mannonoside (10 mM) was added as an exogenous acceptor.After incubations at 30°C for 10 min, reactions were stopped by adding 0.5 ml of5 mM EDTA. Radioactive substrates were separated from acceptors by bindingof the former to a 1-ml Dowex-1 column; the radioactivity in the eluate (con-taining the acceptor) was measured.

Sucrose gradient fractionation. Wild-type cells exponentially grown overnightin 500 ml of YEPD were used. Spheroplasts were prepared as described aboveand suspended in buffer 1. The P1 supernatant fraction was prepared as de-scribed previously, and aliquots (15 ml) were placed on top of two 30-ml pre-formed, 25 to 50% continuous sucrose gradients in Beckman SW28 centrifugetubes. The sucrose solutions contained 1 mM MgCl2 and 10 mM triethanolamineacetic acid (pH 7.2). Gradients were centrifuged at 4°C for 90 min, at 27,000 rpmin an L8-90 Beckman Ultracentrifuge as described previously. Ten 2.5-ml frac-tions were collected from the top of each gradient, diluted, and centrifuged at100,000 � g for 30 min; pellets were resuspended in buffer 1 and kept at �70°C.The �-1,2-mannosyltransferase assay was used to detect Golgi enrichment infractions as reported previously, except that 5 to 10 �l of each fraction was usedfor the reactions. In order to measure endoplasmic reticulum (ER) enrichmentin the fractions, NADPH cytochrome c reductase was assayed as describedpreviously (32).

O-linked carbohydrate analysis. The method of Haselbeck and Tanner (24)was followed for the isolation of total O-linked carbohydrates from 2-3H man-nose (18 Ci/mmol; New England Nuclear [NEN]) radiolabeled cells. �-Elimina-tion was achieved in 0.1 M NaOH for 24 h at room temperature, after which thereaction was stopped by addition of HCl to a final concentration of 0.15 M andthe protein was removed by centrifugation. Radiolabeled species in the super-natant were subjected to thin-layer chromatography on Silica Gel G plates(Merck) with two ascents, in ethyl acetate-butanol-acetic acid-water (3:4:2.5:4).The thin-layer chromatograms were treated with EN3HANCE reagent (NEN)for fluorography and exposed to Kodak X-OMAT X-ray film at �70°C. Bio-gel-P4 chromatography was performed as described by Ferguson (18).

Assay for in situ acid phosphatase activity. Acid phosphatase was analyzed asdescribed by Schweingruber et al. (47) with some modifications. The cells weregrown in 50 ml of Lee medium to log phase (optical density at 600 nm [OD600],1.5), and acid phosphatase activity was induced by overnight incubation in thesame medium, except that H2PO4 was replaced by KCl at the same concentra-tion. Cells were collected by centrifugation, washed with water, and suspended in240 �l of ice-cold lysis buffer (0.05 M sodium citrate [pH 5.5], 1 mM EDTA, 2mM phenylmethylsulfonyl fluoride, 0.1 mM dithiothreitol, and 10% glycerol).Cell lysates were prepared by vortexing for 4 min with 0.5-mm-diameter glassbeads at 4°C. After addition of 120 �l of lysis buffer, the lysates were recoveredand centrifuged at 14,000 � g for 10 min (two times). Ten- to thirty-microliteraliquots were mixed with bromophenol blue (final concentration, 0.01%) and15% glycerol and subjected to electrophoresis on a 5% native polyacrylamide gel.The upper running buffer contained 5.16 g of Tris and 3.48 g of glycine/liter, andthe lower buffer contained 14.5 g of Tris and 60 ml of 1 M HCl/liter. Acidphosphatase activity in gels was detected as described previously (47). For en-zymatic deglycosylation, 500 U of endoglycosidase H (Endo H) (New EnglandBiolabs) was added to the samples (without heat or detergent) and incubated for18 h at 37°C.

Zymolyase sensitivity phenotypic test. Cultures of the wild-type (CAI4), het-erozygous (BAB2), and homozygous (BAB4) C. albicans strains were grown inYEPD medium until the exponential phase. Cells were washed twice in waterand resuspended in 10 mM Tris-HCl (pH 7.5)–0.3% �-mercaptoethanol. Ap-proximately 2.5 � 107 cells were resuspended in the same buffer containingZymolyase 100T at a concentration of 0.01 mg/ml. The optical density at 600 nmwas measured at the start of the incubation and every 20 min thereafter. Thedecreased in optical density reflected the portion of cells that had lysed.

NaCl, Calcofluor White, caffeine, SDS, sodium orthovanadate, and hygromy-cin B sensitivities. Methods for testing the C. albicans strains were similar for allthe effectors. Cultures were grown in 100 ml of YEPD medium with 1% glucoseuntil the exponential phase and diluted to an OD600 of 0.1. Four microliters ofpure and 1/5 serial dilutions of each cell culture were spotted onto YEPD platescontaining NaCl (0.5 to 1.5 M), Calcofluor White (10 to 25 �g/ml), caffeine (5 to15 mM), SDS (0.005% to 0.05%), sodium orthovanadate (10 to 20 mM), andhygromycin B (100 to 300 �g/ml). Differences in growth were recorded afterincubation of the plates at 28°C for 72 h.

Cell wall analysis. For the analysis of cell wall carbohydrate, C. albicans yeastcells (100-ml cultures in 0.1% Glc YEPD) were labeled with 50 �Ci of [U14C]glucose (310 mCi/mmol; NEN) during a 24-h period. Cell wall polysaccharideswere fractionated and quantified as previously described (15). Chitin was mea-sured enzymatically according to the method of Bulawa (9).

Alcian blue binding assay. Alcian blue binding assays were carried out usingthe method of Odani et al. (41) with minor modifications. A standard curve wascreated by making a serial dilution of 0.1% alcian blue 8GX (Electron Micros-copy Sciences) in 0.02 N HCl, measuring the absorbance at 600 nm, and plottingthe OD600 versus micrograms of alcian blue. Five milliliters of cells at an OD600

of 1 (cells grown in YEPD until exponential phase) was washed once with 2 mlof 0.02 N HCl and resuspended in 1 ml of staining solution (0.005% alcian bluein 0.02 N HCl; a total of 50 �g in the staining solution). Cells were allowed tostand for 10 min in the tubes and then centrifuged for 3 min, and the OD600 ofthe supernatants was measured. The amount of dye (x �g) in the supernatantfluid was determined using the standard curve generated above. The amount ofdye bound to the cells was calculated as 50 �g minus x �g.

Cell wall phosphate determination. Phosphate determination assays were car-ried out using a published procedure (55). Cells grown until exponential phase inYEPD were centrifuged, resuspended in water, and broken with glass beads. Thelysate was centrifuged at 900 � g for 10 min, and the cell wall pellet was washedtwice with water. Approximately 15 mg (wet weight) of cell wall was resuspendedin 300 �l of 10% Mg(NO3)2 · 6H20 in ethanol; the mixture was evaporated todryness inside a 13- by 100-mm Pyrex test tube over a strong flame with rapidshaking and further heated in the flame until the brown fumes had disappeared.After the tube had cooled, 0.3 ml of 1 N HCl was added. The tube was cappedwith a marble and heated in a boiling bath for 15 min to hydrolyze to inorganicphosphate any pyrophosphate formed in the ashing procedure. The inorganicphosphate was then measured with the Ames method, as we described for theNDPase assay.

Adherence to epithelial cells. Human cervical epithelial cells (HeLa) weregrown to confluency in Dulbecco’s modified Eagle’s medium containing 10%fetal bovine serum and 100 �g of ciprofloxacin/ml at 37°C (5% CO2). Monolayersestablished in six-well culture dishes were used for adhesion studies. Adhesionwas determined according to the method of Timpel et al. (55). Briefly, mono-layers were washed twice with 2 ml of Dulbecco’s phosphate buffered saline(DPBS), overlaid with either 75 or 150 individual C. albicans cells in 1 ml ofDPBS, and incubated at 37°C for 45 min in an atmosphere of air containing 5%CO2. Following the incubation, monolayers were washed twice with 2 ml of warmDPBS to remove nonadhering cells; the monolayer in each well was then coveredby 2 ml of YEPD agar (1% agar). Yeast colonies appearing after 48 h of growthat 28°C were counted (each colony is assumed to be derived from a single cell).The exact amount of fungal cells (100%) applied to the monolayers was deter-mined by direct plating on YEPD. Adherence was determined as the percentageof fungal cells attached to monolayers of HeLa cells.

Southern blot analysis. Genomic DNA was prepared using the DNeasy Kit(QIAGEN) following the manufacturer’s instructions. About 30 �g of genomicDNA was digested with BclI and electrophoresed through an 0.8% agarose gel.The fractionated DNA was transferred to positively charged nylon membranes(Hybond N�; Amersham), and the membranes were fixed by UV radiation.Prehybridization, hybridization, and labeling of the probe were done followingthe specifications of the DIG Luminescent labeling and detection kit (Roche).

Western blot analysis. Protein fractions were separated by SDS-polyacryl-amide gel electrophoresis and transferred to polyvinyl difluoride membranes(Bio-Rad). Proteins were visualized by standard procedures after reaction withthe Renaissance chemiluminescence detection kit (NEN Life Science) accordingto the manufacturer’s instructions. Polyclonal antibodies against S. cerevisiaechitinase and CPY were kindly provided by C. Specht and R. Gilmore and usedat a final dilution of 1:3,000 and 1:2,000, respectively. Polyclonal anti-CaGda1pantibody was used at 1:3,000.

RESULTS

Isolation of a Golgi apyrase from C. albicans (CaGDA1). A4.2-kb clone containing the complete CaGDA1 gene was ob-tained by hybridization of a partial BamHI library with a frag-ment of the CaGDA1 gene. This fragment was isolated by PCRusing oligonucleotides designed on the basis of sequences pres-ent in the C. albicans database (http://www.sequence.stanford.edu/group/candida). The sequence of the CaGDA1 gene wasdetermined using standard primers and sequence-specific oli-

422 HERRERO ET AL. EUKARYOT. CELL

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

gonucleotides. This sequence was 100% identical to theCaGDA1 sequence that later appeared in the C. albicans ge-nome project (and has EMBL accession No. AJ421721). Plas-mid pGDA1 contained a single open reading frame of 1,800bp. The predicted 599-amino-acid protein was 61% identical toS. cerevisiae ScGda1p. Interestingly, CaGda1p showed no po-tential N-glycosylation sites. The Kyte-Doolittle algorithm (notshown) revealed a single hydrophobic domain close to theamino terminus, consistent with a type II membrane protein,like the S. cerevisiae counterpart. The Candida GDPase con-tains a unique sequence in the stem region, between aminoacids 60 and 100, predicted with high probability to adopt acoil-coil conformation that favors dimerization (44). In spite of

not having that sequence, ScGda1p has been shown to beactive as a homodimer in situ (5). Alignment of the CaGda1pwith other proteins in the data bank showed structural simi-larity to a family of apyrases, sharing four conserved apyrasemotifs (Fig. 1).

To determine whether the CaGda1p is localized to the Golgiapparatus, a mixed vesicle population was subjected to centrif-ugation through a continuous sucrose gradient, and its distri-bution was compared with that of ER and Golgi markers.Fractions enriched in an ER marker, NADPH cytochrome creductase, migrated in the heavy end of the gradient (Fig. 2).CaGda1p was most abundant in lighter-density fractions mi-grating between 27 and 36% sucrose; the same fractions were

FIG. 1. Multiple alignment of proteins related to Cagda1p. The BOXSHADE program was used to compare sequences of proteins, from fungito mammals. Identical residues are shaded in black, and conserved residues are shaded in gray. Asterisks indicate apyrase conserved regions.GenBank accession numbers for the sequences used in the alignment are as follows: S. cerevisiae GDPase, NP_010872; K. lactis GDPase, AJ401304;C. albicans GDPase, AJ421721; Pea NTPase, BAA89275; potato ATPase, P80595; human CD39, NP_001767; mouse CD39, NP_033978;Caenorhabditis elegans (C. elegans), Q18411; S. cerevisiae YND1, NP_010920; human Golgi UDPase, AF016032.

VOL. 1, 2002 CANDIDA GOLGI GDPase 423

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

enriched in �-1,2-mannosyltransferase, commonly used asGolgi marker (2).

We next determined whether the CaGda1p was able to com-plement S. cerevisiae gda1� by transformation of this strainwith the plasmid p426-CaGDA1. We previously showed with S.cerevisiae gda1� cells that the more rapid migration of chiti-nase on SDS gels, relative to the wild type, was the result ofdecreased O-mannosylation (2). In transformants carrying theCaGDA1 gene, the wild-type migration of chitinase was re-stored, demonstrating the functional homology between thetwo genes (Fig. 3A). CaGda1p was also able to fully comple-ment the N-glycosylation defect of the S. cerevisiae gda1�strains as monitored by the electrophoretic migration of theN-glycosylated vacuolar glycoprotein CPY (Fig. 3B).

Disruption of GDA1 alleles. The Ura-blaster technique (20)was used to disrupt both alleles of CaGDA1 in the C. albicansGDA1/GDA1 strain CAI4. Chromosomal DNA of the wildtype and transformants was digested with BclI and analyzed bySouthern hybridization using a 1.7-kb BamHI-EcoRI-specificprobe. The wild-type GDA1 allele displayed 1.5- and 2.1-kb

bands (Fig. 4B, lane 1). The 1.5-kb band originated with thehybridization of the probe to sequences upstream of the GDA1locus; that area was expected to remain unchanged during thedisruption process (Fig. 4B, lanes 1 to 5). A new 1.8-kb bandappeared in transformants with the Ura blaster integrated intoone allele (Fig. 4B, lane 2). After selection on 5-fluorooroticacid-containing medium, the loss of the URA3 gene and onecopy of the hisG element resulted in a 4.8-kb band (Fig. 4B,lane 3). The remaining intact GDA1 allele was disrupted sim-ilarly, leading to homozygous gda1::hisG/gda1�::hisG-URA3-hisG strains (Fig. 4B, lane 4) and corresponding Ura� deriv-atives (Fig. 4B, lane 5). The phenotypes reported below wereobserved in at least two independently isolated disrupted orreconstituted strains.

Cagda1 strains are defective in GDPase in vitro activity. Todetermine whether CaGDA1 encodes a GDPase, we measuredthe abilities of membrane fractions isolated from the wild-typestrain, CAI4, as well as heterozygous and homozygous mutantstrains, BAB2 and BAB4, to hydrolyze GDP (Fig. 5). Mem-branes prepared from the homozygous mutant strains showeda 90% decrease in their ability to hydrolyze GDP in the pres-ence of calcium with respect to wild-type strains (Fig. 5A).Activities of the hemizygous GDA1/gda1 strains were interme-diate, demonstrating a clear gene dosage effect. Concomi-

FIG. 2. CaGda1p is localized in the Golgi apparatus. (A) Enzymeactivities were assayed after velocity sucrose gradient centrifugation asdescribed in Materials and Methods. The maximum specific activity forNADPH cytochrome c reductase (ER marker) was 559 mmol/min,while the maximum activity for �-1,2-mannosyltransferase (Golgi mark-er) was 1.23 mmol/min. Dashed line, sucrose concentration; black dots,amount of protein; black squares, NADPH cytochrome c reductase;gray triangles, �-1,2-mannosyltransferase. (B) Western blotting againstCaGda1p was performed as described in Materials and Methods.

FIG. 3. CaGDA1 complements S. cerevisiae glycosylation pheno-types. SDS-polyacrylamide gel electrophoresis separation and Westernblotting with antibodies against secreted chitinase, an O-glycosylationreporter (A), and carboxypeptidase-Y, an N-glycosylation reporter (B).

FIG. 4. Deletion of CaGDA1 alleles. (A) Structure of differentalleles. The wild-type GDA1 gene and the alleles disrupted by thehisG-URA3-hisG cassette or by hisG alone are shown. (B) Southernblot analysis of genomic DNA was performed with the following strainsdigested with BclI. Lane 1, CAI4 (�ura3::imm434/ura3::imm434); lane2, BAB1 (ura3/ura3 GDA1/gda1::hisG-URA3-hisG); lane 3, BAB2(ura3/ura3 GDA1/�gda1::hisG); lane 4, BAB3 (ura3/ura3gda1::hisG-URA3-hisG/�gda1::hisG); and lane 5, BAB4 (ura3/ura3 �gda1::hisG/�gda1::hisG). The 1,727-bp BamHI-EcoRI fragment was used as aprobe.

424 HERRERO ET AL. EUKARYOT. CELL

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

tantly, the UDPase activity also seemed to decrease while thatof ADPase did not change, as we had previously found with S.cerevisiae gda1 strains (2) (Fig. 5). The GDPase activity in C.albicans doubled in the presence of manganese (Fig. 5B), dif-fering from S. cerevisiae, where calcium is the preferred cationfor GDP hydrolysis (59). Comparing the activities of yeast andhypha forms, we observed that the GDPase and UDPase ac-tivities are quite similar whereas the ADPase activity is slightlyincreased in the hypha. �-1,2-mannosyltranferase, a Golgi en-zyme used as a control, was similar in all the membrane prep-arations (Fig. 5C).

CaGda1p has a role in O-glycosylation. S. cerevisiae gda1mutants have severe defects in mannosylation of O- and N-linked glycoproteins and glycosphingolipids (2). In order todetermine whether CaGDA1 mutants have defects in O-glyco-sylation, we labeled wild-type and mutant cells with 2-3H man-nose, released their O-linked carbohydrate chains by �-elimi-nation, and analyzed them by thin-layer chromatography (Fig.

6A). The oligosaccharide pattern from the C. albicans wild-type strains showed five major species containing between oneand five mannose residues (M1 to M5) (Fig. 6A, lane 2) as inS. cerevisiae (Fig. 6A, lane 1). Some minor slower-migratingspecies, not present in S. cerevisiae, were seen in C. albicansthat could correspond to chains with six or seven mannoseunits according to Hayette et al. (25). These results differ fromthe trimannose O-linked mannan structure reported by Buur-man et al. (11) and prompted us to look into other C. albicanswild-type strains. We observed the same pattern of 5- to 7-res-idue-long O-linked mannose chains not only for the CAI4strain and its parental strain SC5314 but also for an unrelatedwild-type strain, ATCC 10261 (data not shown), strongly sug-gesting that these chain lengths are a characteristic of C. albi-cans.

C. albicans gda1/gda1 strains showed a partial block in theirability to extend O-linked mannan chains manifested by accu-mulation of M1, slight reduction of M2, and a drastic reductionof M3, while the amount M4 and M5 were quite similar toamounts with the wild type. Further analysis of products usinga Biogel-P4 column allowed us to fractionate by size five clearpeaks in wild-type and mutant strains and quantify the relativeabundance of each O-liked mannose chain (Fig. 6B and C).While in the wild-type M1 represents 13% of the total O-linked pool (M1 to M5), in the gda1/gda1 mutant M1 repre-sents 42%, showing that a high proportion of the initiatedchains cannot be further extended. The most severe reductionwas found in chains with three mannoses (M3), with gda1/gda1strains having less than half of wild-type levels. The high-molecular-weight material remaining at the origin with thin-layer chromatography and appearing at the void volume ofBiogel-P4 columns (Fig. 6A, B, and C) might represent pir-likeproteins that are covalently attached to the cell wall and can bereleased by mild alkali treatment (30).

C. albicans O-linked chains were treated with Jack bean�-mannosidase in order to determine the glycosydic linkagebetween the mannose units (Fig. 6D). All of the most abundantM2 chains, as well as most of the M3 chains, were converted tofree mannose. Neither prolonged incubation nor addition offresh �-mannosidase could convert the small remnant to man-nose. These observation, together with the fact that more-abundant M2 chains were completely digested in the sameincubation, strongly suggest that C. albicans has at least twokinds of O-linked chains. Surprisingly, M4 and M5 chains wereresistant to the exomannosidase attack, indicating that themannose at their nonreducing end is not in �-linkage. S. cer-evisiae has only �-linked mannose residues in its O-glycans.

We took advantage of the fact that for C. albicans 88% ofcell wall carbohydrates are N-linked (17), and we decided toanalyze the amount of cell wall mannan in order to monitorpossible N-glycosylation defects. Unlike the case with the S.cerevisiae gda1� strain, the amount of mannoproteins in the C.albicans gda1/gda1 mutants was unchanged from that in wild-type cells. In both cases mannoproteins constituted approxi-mately 38% of total cell wall polymers. We also looked, byactivity staining on native gels, at the size of the heavily N-glycosylated and secreted acid phosphatase (46, 47). We foundno difference in the migration of acid phosphatase secreted bythe wild type and by gda1 mutants (Fig. 7, lane 3 versus lane 4).Even when the gel was run for 20 h in order to allow detection

FIG. 5. GDPase, UDPase, and ADPase activities in membranes ofyeast and hypha forms of C. albicans in the presence of Ca2� (A) andMn2� (B). (C) �-1,2-mannosyltransferase activity. Results are aver-ages of three independent determinations; bars indicate standard de-viations.

VOL. 1, 2002 CANDIDA GOLGI GDPase 425

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

of minor changes, no difference was found (Fig. 7, lane 6 versuslane 7). The MNN9 mutant, defective in outer chain glycosyl-ation, was used as a positive control (Fig. 7, lanes 5 and 8).Acid phosphatase can be also used as a reporter of O-glyco-sylation after enzymatic removal of N-linked chains (40). Wewanted to see if the general defect in O-glycosylation that wedetected was evident also in this individual protein. After treat-ment with Endo H, no difference in the migration of the acidphosphatase produced by the wild type and by gda1 mutantswas detected (Fig. 7, lane 1 versus lane 2). We see two possi-bilities: either the O-glycosylation of this particular reporter isnot affected, or the N-deglycosylation done under native con-ditions was extensive but not total and small differences aremasked. The absence of CaGda1p does not lead to severetruncation of large N-linked oligosaccharides, as reported forK. lactis gda1 mutants, which are affected in O- but not inN-glycosylation (35). Although Cagda1 mutants did not seemto display a defect in N-glycosylation, the CaGDA1 gene com-plements defects in N-glycosylation of S. cerevisiae gda1 nullmutants (Fig. 3B).

CaGda1p has a role in cell wall morphogenesis. In order todetermine whether the absence of CaGda1p affects cell wallmorphogenesis in C. albicans, we initially compared the sensi-tivities of the wild type and mutants to different effectors.Cagda1/Cagda1 mutants showed only a very slight increase insensitivity toward Calcofluor white and hygromycin B, but not

FIG. 6. O-glycosylation: total alkali-releasable saccharides. Base-sensitive oligosaccharides from various strains radiolabeled with [2-3H]man-nose were separated by thin-layer chromatography (butanol, ethanol, water [5:3:2]) and subjected to autoradiography (50,000 cpm/lane) (A andD) or Biogel P4 chromatography (0.8 � 106 and 1 � 106 cpm) (B and C). S.c WT is the S. cerevisiae wild-type strain G2-6; C.a WT is the C. albicanswild-type strain CAI4. C.a gda1/gda1 is C. albicans strain BAB4. �Mase is Jack bean �-mannosidase.

FIG. 7. N-glycosylation: activity staining of acid phosphatase.Twenty-microliter samples from C. albicans wild type (WT) (CAI4),gda1/gda1 (BAB4), and mnn9/mnn9 (SS19-4), prepared as described inMaterials and Methods, were run in a 5% native gel for either 2 h(lanes 1 to 5) or 20 h (lanes 6 to 8) at 100 V. Endoglycosidase H (EndoH)-treated samples are indicated by �.

426 HERRERO ET AL. EUKARYOT. CELL

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

toward NaCl or caffeine (data not shown). We also analyzedthe sensitivity of the Cagda1/Cagda1 mutant toward cell lysisinduced by a �-1,3-glucanase preparation (Zymolyase). Nullmutant cells were more sensitive to enzymatic lysis than thewild type and the hemizygous strains when yeast cells weregrown either at 23 oC (Fig. 8), or 37°C (data not shown).Analysis of the cell wall composition of wild-type and hemizy-gous strains showed that the amounts of glucan and manno-protein in wild-type and homozygous mutant strains were sim-ilar, approximately 45% glucan and 38% mannoprotein. Theonly quantitative difference was observed at the level of chitin:in the gda1/gda1 mutant it was 8.71 0.8 nmol GlcNAc/mg ofcells (wt/wt), nearly double that of the wild type (3.77 0.2nmol GlcNAc/mg of cells [wt/wt]) and hemizygous strains (4.01 0.5 nmol GlcNAc/mg of cells [wt/wt]).

CaGda1p affects cell surface charge and phosphate content.Cell surface hydrophobicity is considered to be an importantvirulence determinant of C. albicans, and it has been linked tothe level of cell wall protein glycosylation (37). In S. cerevisiae,mannosylphosphate gives a net negative charge to cell surfacemannoproteins which in turn allows binding, under acidic con-ditions, to the dye alcian blue (41). We therefore determinedthe level of alcian blue binding for wild-type, hemizygous, andmutant strains of C. albicans and also for the Camnn9 mutant,which is severely affected in glycosylation (50). Binding of thedye to gda1/gda1 cells was 70% of that of the parental strain,whereas for hemizygous and reconstituted strains, it was thesame (Fig. 9A). To determine if these differences in bindingcorresponded to differences in phosphate content, cell wallswere isolated and the level of phosphate was measured (Fig.9B). The amount of phosphate in the cell wall of gda1/gda1strains was nearly half of that of the wild-type cells, completelyexplaining the differential binding of alcian blue.

CaGda1p is not required for adhesion to epithelial cells.Several studies have shown that mannoproteins are necessaryfor adhesion of C. albicans to the surfaces of host cells (38, 51).Because CaGDA1 mutants showed less cell surface charge anda severe defect in O-glycosylation, we measured their adher-ence to monolayers of HeLa cells. In these assays, yeast cells

were placed on an epithelial monolayer for 45 min followed byremoval of nonadhering cells by washing. The number of ad-hering cells was determined by growth in a YEPD agar overlayafter washing. Adherence was determined as the percentage offungal cells attached to monolayers of HeLa cells. No signifi-cant difference was found between the adherence of wild-typeCandida cells (49% 8%) and gda1/gda1 mutant cells (47% 10%). Each experiment was done in triplicate, with two start-ing amounts of fungal cells.

CaGda1p is required for hyphal morphogenesis. It has beenreported that the C. albicans Capmt1 and Capmt6 mutants,both affected in O-glycosylation, are partially defective in hy-phal development (55, 56). To determine whether CaGda1pexerts an effect on germ tube formation, experiments of hyphalinduction in solid and liquid media were carried out. With thelatter we also analyzed the effect of different stimuli on fila-mentous growth: pH, temperature, serum, and N-acetylglu-cosamine.

For the analysis of hyphal formation in solid medium, C.albicans strains were grown in yeast form in liquid Lee mediumand then plated onto solid Lee medium with mannitol. Theappearance of colonies after 6 days of growth at 37°C can beseen in Fig. 10A. As expected, the C. albicans wild-type strainCAI4 formed colonies with long hyphae. In contrast, the mu-tant BAB4 showed a clear defect in filamentation. The hemi-zygous and reconstituted strains BAB2 and BAB4-GDA dis-played an intermediate phenotype (Fig. 10A). In liquid

FIG. 8. C. albicans cell lysis sensitivity following treatment with�-1,3-glucanase. Cells were treated with Zymolyase for different times,and lysis was measured by the decrease in the OD600. Results aremeans standard deviations from three independent experiments.

FIG. 9. Cell surface charge in C. albicans strains. (A) Alcian bluebinding assays. Relative dye binding was calculated as the percentageof dye bound compared with results for the parental strain (CAI4).(B) Cell wall phosphate content. Results are average of three inde-pendent determinations; bars indicate standard deviations. WT, wildtype.

VOL. 1, 2002 CANDIDA GOLGI GDPase 427

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

medium the gda1/gda1 mutants grow at the same rate as thewild type but show a partial block in hyphal formation follow-ing changes in temperature (not shown) and pH (Fig. 10B).The mutants formed normal hyphae in the presence of serumand N-acetylglucosamine (not shown).

DISCUSSION

In this work we addressed the consequences of the elimina-tion of the guanosine diphosphohydrolase of C. albicans, en-coded by CaGDA1, on mannosylation, cell wall construction,and filamentation. Previous studies of the S. cerevisiae GolgiGda1p provided strong evidence for the role of this enzyme inthe nucleotide sugar transport/antiport cycle. After proteinand lipid mannosylation, the reaction product GDP is con-verted by Gda1p to GMP, which can exit the lumen in an

exchange, coupled to the entry of additional GDP-mannosefrom the cytosol (2). This cycle has a pivotal function in reg-ulating posttranslational modifications that occur in the lumenof the Golgi apparatus of all eukaryotes (27).

CaGDA1 was cloned based on its similarity to the ScGDA1gene and was shown to be its functional homologue. CaGda1p,like ScGda1p and KlGda1p, is localized to the Golgi apparatusand is also predicted to be a type II membrane protein with ashort NH2-terminal cytosolic tail and the bulk of the catalyticdomain facing the lumen of the organelle (1, 35). CaGda1p isdifferent from the K. lactis and S. cerevisiae GDPases in that itcontains no N-linked glycosylation sites. Gda1p is responsiblefor about 90% of the membrane-bound, calcium-dependent, invitro GDPase activity of C. albicans yeast and hyphal forms aswell as that of other yeasts, such as K. lactis and S. cerevisiae.Rather active manganese-dependent GDPases remain in the

FIG. 10. Hypha formation of C. albicans strains. (A) Colonies grown for 6 days at 37°C in Lee solid medium; same magnification for all panels.(B) Hypha induction in liquid medium; cells were grown in Lee medium, pH 4.5, at 37°C and then incubated under conditions that promote hyphalgrowth in Lee medium at pH 6.8.

428 HERRERO ET AL. EUKARYOT. CELL

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

gda1/gda1 strains, as indicated by a 50% decrease in the GDPhydrolysis capabilities in hyphae and a 70% decrease in yeastforms of the homozygous mutant. CaGda1p appears to be anactive UDPase in vitro, and manganese greatly stimulates theUDP hydrolysis activity of these and other enzyme(s) remain-ing in the Candida gda1/gda1 strains; again the phenomenon ismore evident with hyphae, suggesting the existence of hypha-specific GDPases and/or UDPases not yet identified.

Only two nucleotide diphosphatases (GDA1 and YND1)have been found in the entire genome of S. cerevisiae, and bothhave their active sites facing the Golgi lumen (1, 60). Thegda1� ynd1� double deletion has a synthetic effect on glyco-sylation, cell growth, and cell shape phenotype (21). Ynd1pappears to be regulated by the binding of its COOH-terminalcytoplasmic domain to an activator subunit of vacuolar H�

ATPase (61). The precise relationship between Ynd1p, Gda1p,and glycosylation in the Golgi is not understood. Our datasuggest that either the CaYnd1p enzyme (we have identifiedthe YND1 Candida gene, which shows 46% identity with the S.cerevisiae gene [unpublished results]) remaining in Cagda1/Cagda1 strains is more active in the hyphae or that there areother membrane bound nucleoside diphosphases in C. albicansunknown at present. Investigating both possibilities will furtherour understanding of the relationship between glycosylation,wall construction, and filamentation.

Loss of CaGDA1 function leads to cell wall-associated de-fects and to “medium-conditional” loss of hyphal development.Defects in the cell wall were not gene dosage dependent. Weobserved a significant reduction in binding of the positivelycharged dye alcian blue that was completely accounted for bya reduction of cell wall phosphate content with Cagda1 mu-tants. Although in C. albicans negatively charged sialylatedglycoconjugates have been proposed to exist (3), we found noevidence for them. Cell walls of gda1/gda1 mutants containedtwice as much chitin as those of the wild type, providing inde-pendent evidence for cell surface alteration. Recent work on S.cerevisiae (43, 31) shows that stress to the cell wall resultingfrom mutations or from environmental factors often leads tostrengthening of the wall by intercalation of a substantialamount of chitin fibers. Our results indicate that mechanismsfor induction of stress-related chitin synthesis are also presentin C. albicans. While mutations in GDA1 have consequences inthe structure and composition of the cell wall of the three yeastspecies studied so far, the alterations are not identical. C.albicans and K. lactis gda1 mutants become hypersensitive toattack by �-1,3-glucanase, while S. cerevisiae gda1 mutants be-come glucanase resistant. The mechanism underlying theabove changes in GDA1 null mutants from the three yeastspecies are not yet understood, but the differences among themhighlight the importance of direct studies with C. albicans. Onthe other hand, in Camnn9 and Capmt1 mutants, which aredefective in N- and O-glycosylation of proteins, respectively,enhanced levels of chitin together with increased coupling ofcell wall proteins through �-1,6-glucan to chitin were observed(31). It is likely that C. albicans and S. cerevisiae respond to cellwall weakening in similar but not identical fashions.

It has been shown that the ability of C. albicans to switchfrom yeast to hyphal forms is important for pathogenicity (34).There is rapid reshaping and expansion of the cell wall duringhyphal formation; for this reason, mutations in genes involved

in cell wall construction have filamentation defects. Some ofthem are as follows: KRE9 required for �-1,6-glucan synthesis(36); PHR1 and PHR2, a pair of pH-regulated �-1,3-glucano-syltransferases responsible for the elongation of �-1,3-glucan(19); SRB1 encoding GDP-mannose pyrophosphorylase (58);MNN9, required for N-linked outer chain glycosylation (50);and PMT1 and PMT6, responsible for the initiation of O-linkedmannan chains (55, 56). We thus examined whether Cagda1mutants could undergo a yeast-hypha transition. Loss ofCaGDA1 function blocked pH- and temperature-induced hy-phal formation but had no effect on filamentation induced byserum or N-acetylglucosamine as the sole carbon source. Aclear gene dosage effect was seen for this partial block inhyphal formation.

The morphogenetic switch from budding yeast to hyphalgrowth occurs in response to a variety of stimuli and growthconditions. Recently it has been shown that environmentallyinduced filamentous growth can occur with C. albicans even inthe absence of EGF1, CPH1, and TUP1, providing evidence fora fourth regulatory pathway (8). It appears that genes turnedon during filamentous growth do not respond to a centralregulator; rather, they respond individually to at least fourpathways that regulate filamentous growth (52, 33, 7), stronglysuggesting that a network of signaling pathways extends downto target genes. Thus, even morphologically similar phenotypesmay be different at the molecular level, especially at the cellwall (8). Serum still stimulates hyphal formation not only inCagda1 mutants but also in Capmt1 O-glycosylation mutants,as well as in mutants defective in elements of a conservedmitogen-activated protein kinase signaling pathway, all ofwhich manifest a partial block in filamentation (55). For thisreason it was suggested that not morphogenesis per se butrather sufficient levels of an unidentified O-glycosylated pro-tein critical for the mitogen-activated protein kinase signalingpathway that is operative in some media could be defective inpmt1 mutants (55). Similar reasoning could be applied toCagda1 mutants, which are also defective in O-glycosylation.

The O-glycosylation defects in gda1 mutants are not theresult of a single missing mannosyltransferase. ShorterO-linked mannan chains are a consequence of the reducedavailability of the substrate GDP-mannose in the lumen of theGolgi. In the absence of Gda1p there is insufficient generationof the antiporter molecule GMP to sustain the nucleotidesugar transport cycle. We found a large increase in singleO-linked mannose residues in the Cagda1 mutants, indicatingthat the O-linked mannan chains were initiated normally in theER but that elongation in the Golgi failed. This was expected,since it is well established that transfer of the first O-linkedmannose to serine and threonine by protein-O-mannosyltrans-ferases occurs in the ER with dolichol-P-mannose as the donor(24). GDP-mannose is the direct donor for the subsequentmannose residues, which are added stepwise to complete O-linked chains in the lumen of the Golgi (42). Analyses of bulkcell mannoproteins of Cagda1 mutants also showed a smalldecrease in chains with two mannose residues, a severe de-crease in chains with three, and normal amounts of longerchains with four and five mannose residues. It is possible thatthe Km for GDP-mannose of some mannosyltransferases couldbe very low. Therefore, the reactions catalyzed by these en-

VOL. 1, 2002 CANDIDA GOLGI GDPase 429

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

zymes would occur at maximum velocity even with a limitedsupply of substrate.

It was surprising to find that gda1 mutants exhibited appar-ently normal N-glycosylation together with a significant de-crease in cell wall phosphate content. Although the amounts ofN-glycans present in wild-type and gda1 mutant strains areabout the same, alteration in the structure of N-linked oligo-saccharides cannot be ruled out. It has been recently shownthat both, N- and O-linked glycosylation are affected by theloss of function of the essential Golgi apparatus GDP-mannosetransporter, CaVRG4 (40). It is not clear why phenotypic con-sequences when the transport/antiport cycle that suppliesGDP-mannose to the Golgi lumen is affected by reduced ex-pression of the transporter (CaVrg4p) are different from whenthe same cycle is disturbed by the absence of one of the anti-porter generating enzymes (CaGda1p). One possibility couldbe a differential compartmentalization within the CandidaGolgi of Gda1p and other enzymes able to convert GDP toGMP, which could in turn locally drive the transport/antiportcycle to some extent.

�-1,2-linked oligomannosides were first described by Shibataet al. (49) as associated with C. albicans cell wall phosphopep-tidomannan. Our data strongly suggest that O-linked chainswith 4 and 5 mannose units have �-linked mannose at thereducing end. Recently phospholipomannan, a family of cellsurface glycolipids, was found to contain long linear chains of�-1,2-linked mannose residues (57). Several groups have in-vestigated the recognition of �-1,2-oligomannosides by the im-mune system in relation to the pathogenic character of C.albicans. These molecules have been shown to elicit specificantibodies in humans (45). Moreover, �-1,2-oligomannosideshave been shown to act as C. albicans adhesins for macro-phages and stimulate the production of tumor necrosis factoralpha (29). A complete study of the structure and biosynthesisof wall glycans will help in understanding the immunologicalproperties of the cell wall and its relation to the pathogenesisof C. albicans infections.

ACKNOWLEDGMENTS

We thank Gretchen Carney for expert typing and P. Berninsone andP. Robbins for helpful discussions. We also thank Stuart M. Levitz andMichael Mansour for their help with the cell adhesion studies.

D.U. is a recipient of a fellowship from the Pasteur Institue-CenciBolognetti foundation. This work was supported by NIH grantsGM59773 to C.A., GM 30365 to C.B.H., DGICYT grant PM 98-0317to A.D., and by grant no. 99196 from the Fulbright Commission forCultural, Educational and Scientific Exchange between the UnitedStates and Spain to A.D. and C.B.H.

REFERENCES

1. Abeijon, C., P. Orlean, P. W. Robbins, and C. B. Hirschberg. 1989. Topog-raphy of glycosylation in yeast: characterization of GDPmannose transportand lumenal guanosine diphosphatase activities in Golgi-like vesicles. Proc.Natl. Acad. Sci. USA 86:6935–6939.

2. Abeijon, C., K. Yanagisawa, E. C. Mandon, A. Hausler, K. Moremen, C. B.Hirschberg, and P. W. Robbins. 1993. Guanosine diphosphatase is requiredfor protein and sphingolipid glycosylation in the Golgi lumen of Saccharo-myces cerevisiae. J. Cell Biol. 122:307–323.

3. Alviano, C. S., L. R. Travassos, and R. Schauer. 1999. Sialic acids in fungi:a minireview. Glycoconj. J. 16:545–554.

4. Berninsone, P., H. Y. Hwang, I. Zemtseva, H. R. Horvitz, and C. B. Hirsch-berg. 2001. SQV-7, a protein involved in Caenorhabditis elegans epithelialinvagination and early embryogenesis, transports UDP-glucuronic acid,UDP-N-acetylgalactosamine, and UDP-galactose. Proc. Natl. Acad. Sci.USA 98:3738–3743.

5. Berninsone, P., Z. Y. Lin, E. Kempner, and C. B. Hirschberg. 1995. Regu-lation of yeast Golgi glycosylation. Guanosine diphosphatase functions as ahomodimer in the membrane. J. Biol. Chem. 270:14564–14567.

6. Berninsone, P. M., and C. B. Hirschberg. 2000. Nucleotide sugar transport-ers of the Golgi apparatus. Curr. Opin. Struct. Biol. 10:542–547.

7. Braun, B. R., and A. D. Johnson. 1997. Control of filament formation inCandida albicans by the transcriptional repressor TUP1. Science 277:105–109.

8. Braun, B. R., and A. D. Johnson. 2000. TUP1, CPH1 and EFG1 makeindependent contributions to filamentation in Candida albicans. Genetics155:57–67.

9. Bulawa, C. E. 1992. CSD2, CSD3, and CSD4, genes required for chitinsynthesis in Saccharomyces cerevisiae: the CSD2 gene product is related tochitin synthases and to developmentally regulated proteins in Rhizobiumspecies and Xenopus laevis. Mol. Cell. Biol. 12:1764–1776.

10. Burda, P., and M. Aebi. 1999. The dolichol pathway of N-linked glycosyla-tion. Biochim. Biophys. Acta 1426:239–257.

11. Buurman, E. T., C. Westwater, B. Hube, A. J. Brown, F. C. Odds, and N. A.Gow. 1998. Molecular analysis of CaMnt1p, a mannosyl transferase impor-tant for adhesion and virulence of Candida albicans. Proc. Natl. Acad. Sci.USA 95:7670–7675.

12. Calderone, R. A., and P. C. Braun. 1991. Adherence and receptor relation-ships of Candida albicans. Microbiol. Rev. 55:1–20.

13. Capasso, J. M., and C. B. Hirschberg. 1984. Mechanisms of glycosylationand sulfation in the Golgi apparatus: evidence for nucleotide sugar/nucleo-side monophosphate and nucleotide sulfate/nucleoside monophosphate an-tiports in the Golgi apparatus membrane. Proc. Natl. Acad. Sci. USA 81:7051–7055.

14. Casanova, M., J. L. Lopez-Ribot, C. Monteagudo, A. Llombart-Bosch, R.Sentandreu, and J. P. Martinez. 1992. Identification of a 58-kilodalton cellsurface fibrinogen-binding mannoprotein from Candida albicans. Infect. Im-mun. 60:4221–4229.

15. Castro, C., J. C. Ribas, M. H. Valdivieso, R. Varona, F. del Rey, and A.Duran. 1995. Papulacandin B resistance in budding and fission yeasts: iso-lation and characterization of a gene involved in (1,3)beta-D-glucan synthesisin Saccharomyces cerevisiae. J. Bacteriol. 177:5732–5739.

16. Cutler, J. E. 1991. Putative virulence factors of Candida albicans. Annu. Rev.Microbiol. 45:187–218.

17. Elorza, M. V., A. Marcilla, and R. Sentandreu. 1988. Wall mannoproteins ofthe yeast and mycelial cells of Candida albicans: nature of the glycosidicbonds and polydispersity of their mannan moieties. J. Gen. Microbiol. 134:2393–2403.

18. Ferguson, M. A., P. Murray, H. Rutherford, and M. J. McConville. 1993. Asimple purification of procyclic acidic repetitive protein and demonstrationof a sialylated glycosyl-phosphatidylinositol membrane anchor. Biochem. J.291:51–55.

19. Fonzi, W. A. 1999. PHR1 and PHR2 of Candida albicans encode putativeglycosidases required for proper cross-linking of beta-1,3- and beta-1,6-glucans. J. Bacteriol. 181:7070–7079.

20. Fonzi, W. A., and M. Y. Irwin. 1993. Isogenic strain construction and genemapping in Candida albicans. Genetics 134:717–728.

21. Gao, X. D., V. Kaigorodov, and Y. Jigami. 1999. YND1, a homologue ofGDA1, encodes membrane-bound apyrase required for Golgi N- and O-glycosylation in Saccharomyces cerevisiae. J. Biol. Chem. 274:21450–21456.

22. Gemmill, T. R., and R. B. Trimble. 1999. Overview of N- and O-linkedoligosaccharide structures found in various yeast species. Biochim. Biophys.Acta 1426:227–237.

23. Goshorn, A. K., S. M. Grindle, and S. Scherer. 1992. Gene isolation bycomplementation in Candida albicans and applications to physical and ge-netic mapping. Infect. Immun. 60:876–884.

24. Haselbeck, A., and W. Tanner. 1983. O-glycosylation in Saccharomyces cer-evisiae is initiated at the endoplasmic reticulum. FEBS Lett. 158:335–338.

25. Hayette, M. P., G. Strecker, C. Faille, D. Dive, D. Camus, D. W. Mackenzie,and D. Poulain. 1992. Presence of human antibodies reacting with Candidaalbicans O-linked oligomannosides revealed by using an enzyme-linked im-munosorbent assay and neoglycolipids. J. Clin. Microbiol. 30:411–417.

26. Hirschberg, C. B. 2001. Golgi nucleotide sugar transport and leukocyteadhesion deficiency II. J. Clin. Investig. 108:3–6.

27. Hirschberg, C. B., P. W. Robbins, and C. Abeijon. 1998. Transporters ofnucleotide sugars, ATP, and nucleotide sulfate in the endoplasmic reticulumand Golgi apparatus. Annu. Rev. Biochem. 67:49–69.

28. Ibrahim, A. S., F. Mirbod, S. G. Filler, Y. Banno, G. T. Cole, Y. Kitajima,J. E. Edwards, Jr., Y. Nozawa, and M. A. Ghannoum. 1995. Evidence im-plicating phospholipase as a virulence factor of Candida albicans. Infect.Immun. 63:1993–1998.

29. Jouault, T., C. Fradin, P. A. Trinel, A. Bernigaud, and D. Poulain. 1998.Early signal transduction induced by Candida albicans in macrophagesthrough shedding of a glycolipid. J. Infect. Dis. 178:792–802.

30. Kandasamy, R., G. Vediyappan, and W. L. Chaffin. 2000. Evidence for thepresence of pir-like proteins in Candida albicans. FEMS Microbiol. Lett.186:239–243.

31. Kapteyn, J. C., L. L. Hoyer, J. E. Hecht, W. H. Muller, A. Andel, A. J.

430 HERRERO ET AL. EUKARYOT. CELL

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from

Verkleij, M. Makarow, H. Van Den Ende, and F. M. Klis. 2000. The cell wallarchitecture of Candida albicans wild-type cells and cell wall-defective mu-tants. Mol. Microbiol. 35:601–611.

32. Kubota, S., Y. Yoshida, H. Kumaoka, and A. Furumichi. 1977. Studies on themicrosomal electron-transport system of anaerobically grown yeast. V. Pu-rification and characterization of NADPH-cytochrome c reductase. J. Bio-chem (Tokyo) 81:197–205.

33. Liu, H., J. Kohler, and G. R. Fink. 1994. Suppression of hyphal formation inCandida albicans by mutation of a STE12 homolog. Science 266:1723–1726.

34. Lo, H. J., J. R. Kohler, B. DiDomenico, D. Loebenberg, A. Cacciapuoti, andG. R. Fink. 1997. Nonfilamentous Candida albicans mutants are avirulent.Cell 90:939–949.

35. Lopez-Avalos, M. D., D. Uccelletti, C. Abeijon, and C. B. Hirschberg. 2001.The UDPase activity of the Kluyveromyces lactis Golgi GDPase has a role inuridine nucleotide sugar transport into Golgi vesicles. Glycobiology 11:413–422.

36. Lussier, M., A. M. Sdicu, S. Shahinian, and H. Bussey. 1998. The Candidaalbicans KRE9 gene is required for cell wall beta-1, 6-glucan synthesis and isessential for growth on glucose. Proc. Natl. Acad. Sci. USA 95:9825–9830.

37. Masuoka, J., and K. C. Hazen. 1999. Differences in the acid-labile compo-nent of Candida albicans mannan from hydrophobic and hydrophilic yeastcells. Glycobiology 9:1281–1286.

38. Mitchell, A. P. 1998. Dimorphism and virulence in Candida albicans. Curr.Opin. Microbiol. 1:687–692.

39. Mumberg, D., R. Muller, and M. Funk. 1995. Yeast vectors for the con-trolled expression of heterologous proteins in different genetic backgrounds.Gene 156:119–122.

40. Nishikawa, A., J. B. Poster, Y. Jigami, and N. Dean. 2002. Molecular andphenotypic analysis of CaVRG4, encoding an essential Golgi apparatusGDP-mannose transporter. J. Bacteriol. 184:29–42.

41. Odani, T., Y. Shimma, X. H. Wang, and Y. Jigami. 1997. Mannosylphosphatetransfer to cell wall mannan is regulated by the transcriptional level of theMNN4 gene in Saccharomyces cerevisiae. FEBS Lett. 420:186–190.

42. Orlean, P. 1997. Biogenesis of yeast wall and surface components, vol. 3.Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

43. Osmond, B. C., C. A. Specht, and P. W. Robbins. 1999. Chitin synthase III:synthetic lethal mutants and “stress related” chitin synthesis that bypassesthe CSD3/CHS6 localization pathway. Proc. Natl. Acad. Sci. USA 96:11206–11210.

44. Parry, D. A. 1982. Coiled-coils in alpha-helix-containing proteins: analysis ofthe residue types within the heptad repeat and the use of these data in theprediction of coiled-coils in other proteins. Biosci. Rep. 2:1017–1024.

45. Poulain, D., C. Faille, C. Delaunoy, P. M. Jacquinot, P. A. Trinel, and D.Camus. 1993. Probable presence of beta(1–2)-linked oligomannosides thatact as human immunoglobulin G3 epitopes and are distributed over a Can-dida albicans 14- to 18-kilodalton antigen. Infect. Immun. 61:1164–1166.

46. Schonholzer, F., A. M. Schweingruber, H. Trachsel, and M. E. Schweingru-ber. 1985. Intracellular maturation and secretion of acid phosphatase ofSaccharomyces cerevisiae. Eur. J. Biochem. 147:273–279.

47. Schweingruber, A. M., F. Schoenholzer, L. Keller, R. Schwaninger, H.

Trachsel, and M. E. Schweingruber. 1986. Glycosylation and secretion ofacid phosphatase in Schizosaccharomyces pombe. Eur. J. Biochem. 158:133–140.

48. Sherman, F., G. R. Fink, and J. B. Hicks. 1986. Methods in yeast genetics.Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

49. Shibata, N., T. Ichikawa, M. Tojo, M. Takahashi, N. Ito, Y. Okubo, and S.Suzuki. 1985. Immunochemical study on the mannans of Candida albicansNIH A-207, NIH B-792, and J-1012 strains prepared by fractional precipita-tion with cetyltrimethylammonium bromide. Arch. Biochem. Biophys. 243:338–348.

50. Southard, S. B., C. A. Specht, C. Mishra, J. Chen-Weiner, and P. W. Rob-bins. 1999. Molecular analysis of the Candida albicans homolog of Saccha-romyces cerevisiae MNN9, required for glycosylation of cell wall mannopro-teins. J. Bacteriol. 181:7439–7448.

51. Steele, C., J. Leigh, R. Swoboda, H. Ozenci, and P. L. Fidel, Jr. 2001.Potential role for a carbohydrate moiety in anti-Candida activity of humanoral epithelial cells. Infect. Immun. 69:7091–7099.

52. Stoldt, V. R., A. Sonneborn, C. E. Leuker, and J. F. Ernst. 1997. Efg1p, anessential regulator of morphogenesis of the human pathogen Candida albi-cans, is a member of a conserved class of bHLH proteins regulating mor-phogenetic processes in fungi. EMBO J. 16:1982–1991.

53. Strahl-Bolsinger, S., M. Gentzsch, and W. Tanner. 1999. Protein O-manno-sylation. Biochim. Biophys. Acta 1426:297–307.

54. Sundstrom, P. 1999. Adhesins in Candida albicans. Curr. Opin. Microbiol.2:353–357.

55. Timpel, C., S. Strahl-Bolsinger, K. Ziegelbauer, and J. F. Ernst. 1998.Multiple functions of Pmt1p-mediated protein O-mannosylation in the fun-gal pathogen Candida albicans. J. Biol. Chem. 273:20837–20846.

56. Timpel, C., S. Zink, S. Strahl-Bolsinger, K. Schroppel, and J. Ernst. 2000.Morphogenesis, adhesive properties, and antifungal resistance depend onthe Pmt6 protein mannosyltransferase in the fungal pathogen Candida albi-cans. J. Bacteriol. 182:3063–3071.

57. Trinel, P. A., Y. Plancke, P. Gerold, T. Jouault, F. Delplace, R. T. Schwarz,G. Strecker, and D. Poulain. 1999. The Candida albicans phospholipoman-nan is a family of glycolipids presenting phosphoinositolmannosides withlong linear chains of beta-1,2-linked mannose residues. J. Biol. Chem. 274:30520–30526.

58. Warit, S., N. Zhang, A. Short, R. M. Walmsley, S. G. Oliver, and L. I.Stateva. 2000. Glycosylation deficiency phenotypes resulting from depletionof GDP-mannose pyrophosphorylase in two yeast species. Mol. Microbiol.36:1156–1166.

59. Yanagisawa, K., D. Resnick, C. Abeijon, P. W. Robbins, and C. B. Hirsch-berg. 1990. A guanosine diphosphatase enriched in Golgi vesicles of Saccha-romyces cerevisiae. Purification and characterization. J. Biol. Chem. 265:19351–19355.

60. Zhong, X., and G. Guidotti. 1999. A yeast Golgi E-type ATPase with anunusual membrane topology. J. Biol. Chem. 274:32704–32711.

61. Zhong, X., R. Malhotra, and G. Guidotti. 2000. Regulation of yeastectoapyrase ynd1p activity by activator subunit Vma13p of vacuolar H�-ATPase. J. Biol. Chem. 275:35592–35599.

VOL. 1, 2002 CANDIDA GOLGI GDPase 431

on April 16, 2020 by guest

http://ec.asm.org/

Dow

nloaded from