Cloning and Disruption of caPLB1, a Phospholipase B Gene Involved in the Pathogenicity of Candida albicans *

The Candida albicans PLB1gene was cloned using a polymerase chain reaction-based approach relying on degenerate oligonucleotide primers designed according to the amino acid sequences of two peptide fragments obtained from a purified candidal enzyme displaying phospholipase activity (Mirbod, F., Banno, Y., Ghannoum, M. A., Ibrahim, A. S., Nakashima, S., Yasuo, K., Cole, G. T., and Nozawa, Y. (1995) Biochim. Biophys. Acta 1257, 181–188). Sequence analysis of a 6.7-kilobase pairEcoRI-ClaI genomic clone revealed a single open reading frame of 1818 base pairs that predicts for a pre-protein of 605 residues. Comparison of the putative candidal phospholipase with those of other proteins in data base revealed significant homology to known fungal phospholipase Bs from Saccharomyces cerevisiae(45%), Penicillium notatum (42%), Torulaspora delbrueckii (48%), and Schizosaccharomyces pombe(38%). Thus, we have cloned the gene encoding a C. albicans phospholipase B homolog. This gene, designated caPLB1, was mapped to chromosome 6. Disruption experiments revealed that the caplb1 null mutant is viable and displays no obvious phenotype. However, the virulence of strains deleted for caPLB1, as assessed in a murine model for hematogenously disseminated candidiasis, was significantly attenuated compared with the isogenic wild-type parental strain. Although deletion of caPLB1 did not produce any detectable effects on candidal adherence to human endothelial or epithelial cells, the ability of the caplb1 null mutant to penetrate host cells was dramatically reduced. Thus, phospholipase B may well contribute to the pathogenicity of C. albicans by abetting the fungus in damaging and traversing host cell membranes, processes which likely increase the rapidity of disseminated infection.

Infections attributable to the opportunistic pathogen Candida albicans have increased significantly in immunocompromised hosts, including surgical and cancer patients receiving immunosuppressive therapy, as well as in patients requiring intravascular catheterization (1,2). Several traits have been proposed to enhance the virulence of this important nosocomial fungus and include yeast-to-hyphal transformation, adherence to host cells, phenotypic switching, and production of extracellular proteinases and phospholipases (3).
Approaches for evaluating the significance of specific factors to the pathogenicity of C. albicans have traditionally centered around isolating mutants that fail to express a suspected trait and then comparing the virulence of these mutants with known pathogenic strains in animal models. However, such mutants were generated by chemical or UV treatment, processes which likely cause additional mutations at other genetic loci (4,5). The advent of novel genetic strategies, such as the ura-blaster protocol for targeted gene disruption (6,7), and the construction of genetically defined auxotrophic C. albicans mutants (7), have circumvented this problem by facilitating the construction of strain pairs differing only at a specified genetic locus. The virulence of these isogenic strains can be directly compared to their corresponding parental strains without taking heed of potential genetic differences commonly encountered with clinical isolates or mutants generated by chemical or UV treatment. Targeted gene disruption has now been used to evaluate the contribution of several proteins to the virulence of C. albicans (8 -12). Most recently, C. albicans strains deleted for the INT1 gene, which encodes a cell-surface protein with similarity to mammalian integrins, were constructed (13). Adhesion, hyphal formation, and virulence were subsequently found to be correlated with the expression of this gene. Thus, the identification and cloning of genes that encode putative virulence factors represent a key step in furthering our understanding of the pathobiology of C. albicans.
Phospholipases represent a class of lipolytic enzymes, acylhydrolases and phosphodiesterases, that has received extensive characterization (14). Phospholipases A, C, and D have been linked to the virulence of several microbial pathogens (15)(16)(17)(18). For example, two distinct phospholipase Cs of Listeria monocytogenes are required for escape from host vacuoles and cell-to-cell spread (19). C. albicans secretes lysophospholipase, transacylase, and phospholipase B (PLB) 1 activities (20). The candidal phospholipases, like their bacterial counterparts, are also regarded as virulence determinants (21). These enzymes may promote penetration of host cell membranes (22,23), as well as influence adherence characteristics of the fungus (24). The results of a recent investigation that compared the virulence of several blood isolates of C. albicans using a murine model for disseminated candidiasis showed that the level of phospholipase activity secreted by each particular isolate was directly correlated with the pathogenicity of that isolate (21). However, although these findings were encouraging, because genetically unrelated clinical isolates were used, it could only be suggested that phospholipase secretion contributed to the virulence of C. albicans. Thus, to unambiguously determine whether secreted phospholipases contribute to candidal pathogenesis, a molecular genetic approach was initiated by purifying a putative phospholipase from strain 16240, a blood isolate of C. albicans that secretes high levels of phospholipase activity (20). We cloned, sequenced, and disrupted the gene encoding this enzyme and designated it C. albicans phospholipase B (caPLB1). We report here that disruption of caPLB1 had no effect on candidal growth, morphology, or adherence but significantly attenuated candidal virulence in a murine model of disseminated candidiasis and dramatically reduced the ability C. albicans to penetrate host cells.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Culture Media-The C. albicans strains used in this study are listed in Table I. Bacterial strains DH5␣ (25), XL1-Blue (Stratagene, La Jolla, CA), and LE392 (Promega, Madison, WI) were used for the propagation of all plasmids and phage. The plasmid pBSK ϩ was used for subcloning experiments. A C. albicans genomic DNA library (26) cloned into the GEM-12 phage vector (Promega) was used to obtain the full-length caPLB1 gene.
Human umbilical vein endothelial cells (HUVEC) were obtained from discarded human umbilical veins using previously published methods (27). The epithelial cell line HT-29 (ATCC HTB 38) was obtained from ATCC (Rockville, MD).
Ura Ϫ auxotrophs were selected on medium containing 5-fluoroorotic acid (5-FOA) as described (28). The medium was essentially the same as that described except that uridine (50 g/ml) was substituted for uracil. Media used for the growth of Escherichia coli strains were prepared according to standard procedures (29).
Purification of an Extracellular Phospholipase from C. albicans-The putative phospholipase was purified according to a previously described procedure (20). Briefly, C. albicans strain 16240 was grown in Sabouraud dextrose broth (SDB) containing 4% (w/v) glucose for 12 h at 35°C. The cell-free supernatant was obtained by centrifuging the culture at 3000 ϫ g for 10 min and was then filtered. Solid ammonium sulfate was then added to 65% saturation. The salted-out proteins were redissolved and dialyzed against buffer (10 mM Tris-HCl, pH 1.4, 1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 10% glycerol) and then applied to a DEAE-cellulose column. Fractions containing phospholipase activity, as determined by a radiometric assay (20), were pooled and concentrated in an Amicon concentrator fitted with a YM10 membrane. The concentrate was further fractionated on an Ultrogel Ac4 -44 column. The fractions with phospholipase activity were pooled and further purified by sequential chromatography on hydroxyapatite HCA-100S, TSK-gel 3000 HPLC, and Mono Q HPLC columns.
Amino Acid Composition and Sequencing-Amino acid composition analysis of the purified enzyme was performed using the techniques of carboxymethylation and performic acid oxidation (30). The amino acid sequences of both the NH 2 terminus and an internal peptide fragment of the enzyme were determined. NH 2 -terminal sequence analysis was initiated by electrotransferring the SDS-polyacrylamide gel electrophoresis separated enzyme to a membrane support (Millipore Corp., Bedford, MA). The single, 84-kDa band was excised from the membrane, and the bound protein was hydrolyzed and sequenced in an Applied Biosystems model 477A gas-phase sequencer by standard methods (30). To obtain the sequence of the internal peptide fragment, the in-gel protease digestion method was used (30). Briefly, the band containing the enzyme was excised from a polyacrylamide gel, dehydrated, and then digested overnight at 37°C in 75 l of 200 mM NH 4 HCO 3 containing 2.5 g of Promega-modified trypsin. The resulting peptide fragments were recovered by extraction with 60% acetonitrile and 0.1% trifluoroacetic acid, individually separated by HPLC, and their amino acid sequences determined as described above.
Cloning of the C. albicans PLB1 Gene-A portion of DNA from the gene encoding the putative phospholipase was amplified via PCR using degenerate oligonucleotide DNA primers that were designed according to the amino acid sequences present at the NH 2 terminus of the enzyme, as well as at an internal site of the protein. Genomic DNA was extracted (31) from C. albicans strain SC5314 and used as template in PCR reactions. PCR amplification reactions consisted of 100-l volumes containing 10 mM Tris-HCl, pH 8.4, 50 mM KCl, 1.5 mM MgCl 2 , dNTPs (2.5 mM each dNTP), oligonucleotide primers (1 M each), genomic DNA (1 g), and 5 units of Taq DNA polymerase. The PCR conditions were as follows: initial denaturation (95°C, 2 min) followed by 33 cycles of denaturation (94°C, 1 min), annealing (50°C, 1 min), and extension (72°C, 3 min) ending with an 8-min extension step at 72°C. Reaction mixtures were analyzed on 1.2% agarose gels. The desired product, a 261-bp DNA fragment, was isolated (Geneclean, Bio 101), blunt-ended by adding Klenow enzyme and excess dNTPs, cloned into the SmaI site of pBSK ϩ , and sequenced. This PCR-generated fragment was random primer-labeled with [␣-32 P]dCTP using the NEBlot 228 Kit (New England Biolabs) and then used as a probe to screen a C. albicans genomic library cloned into GEM-12. Library screening and purification of vector DNA from positive plaques were carried out as described previously (32).
Molecular Biological Techniques and DNA Sequence Analysis-Standard procedures were used for the propagation and selection of plasmids, the growth of their bacterial hosts, and for the subcloning of DNA fragments (29).
nositol; PCR, polymerase chain reaction; HUVEC, human umbilical vein endothelial cells. For Southern blot analyses, total chromosomal DNA isolated from the respective C. albicans strains was digested with KpnI and SacI (6 units/g DNA), electrophoresed through 1.2% agarose, and transferred (29) to nylon membranes (Boehringer Mannheim). DNA was crosslinked to the membrane using a UV Stratalinker 228 (Stratagene) delivering 120,000 J. To cross-link intact, separated candidal chromosomes to nylon supports, an energy of 180,000 J was delivered. The probes used were a 3-kb HindIII fragment containing a portion of the NH 2 -terminal region of the caPLB1 gene (5Ј upstream) and a 736-bp BamHI fragment corresponding to the portion of caPLB1 deleted during the disruption process (BamHI internal). Probes were labeled with digoxigenin-dUTP using the Genius 228 random primed DNA labeling and detection kit (Boehringer Mannheim). Membranes were prehybridized in a rotary oven (Robbins Scientific, model 400) at 65°C for 1 h with 25 ml of 5ϫ SSC containing 0.1% (w/v) N-lauroylsarcosine, 0.02% (w/v) sodium dodecyl sulfate (SDS), and 1.0% (w/v) Blocking Reagent. Digoxigenin-labeled probe was added to the prehybridization solution at a concentration of approximately 30 ng/ml and allowed to hybridize to the membranes overnight at 65°C. Membranes were washed twice with 2ϫ SSC containing 0.1% SDS for 5 min at 65°C, then twice with 0.5ϫ SSC, 0.1% SDS for 15 min at 65°C. Hybridizing DNA fragments were detected using an alkaline phosphatase-conjugated anti-digoxigenin antibody in accordance with the manufacturer's instructions (Boehringer Mannheim).
All DNA sequencing was performed by a custom DNA sequencing service (Retrogen, San Diego, CA). Double-stranded DNA sequences were determined with an Applied Biosystems, Inc. (ABI), model 377, automated sequencer employing the dideoxy chain termination method (33). Nucleotide and protein sequence analyses were performed using the DNA Strider 1.1 program (34), as well as the Wisconsin Genetics Computer Group sequence analysis software package, version 7.0 (35). Homology searches of the non-redundant data base were conducted using the BLASTP program (36).
Disruption of the caPLB1 Gene-A 6.7-kb EcoRI-ClaI genomic fragment harboring the caPLB1 coding region (Fig. 1A) was subcloned into the EcoRI-ClaI site of pCR-Script SK ϩ (Stratagene), in which the BamHI site, originally present in the polylinker region, was destroyed, forming pSPL1. To construct the disruption plasmid pca plb1::hisG-cURA3-hisG, pSPL1 was digested with HindIII and BamHI in order to delete a 1184-bp fragment from the coding region of caPLB1. Into this deleted region was subcloned a 4.2-kb fragment consisting of the candidal URA3 gene (cURA3) flanked by 1.1-kb direct repeats of Salmonella typhimurium hisG DNA (6) (Fig. 3A). The construct was then digested with NotI and XhoI, neither of which have recognition sites in hisG, cURA3, or caPLB1, to release the 9.7-kb disruption cassette from the vector backbone. The linear fragment was then transformed into the ura -C. albicans strain CAI-4 using a lithium acetatebased transformation protocol (31). Ura ϩ prototrophs were selected on synthetic medium lacking uridine. Total chromosomal DNA was isolated (31) from cultures produced by the growth of individual colonies. Targeted integration at the caPLB1 locus was confirmed by Southern blot analysis. Primary disruptants were then grown on 5-fluoroorotic acid (5-FOA) to recycle the cURA3 gene as a selectable marker for disrupting the remaining caPLB1 allele.
Pulsed-field Gel Electrophoresis-Pulsed-field gel electrophoresis was performed using a Bio-Rad contour-clamped homogenous electric field (CHEF) DR III apparatus. Chromosomal DNA was prepared in agarose plugs essentially as described (37). Chromosomes were separated on 0.8% (w/v) Bio-Rad pulsed-field grade agarose in TAE buffer (40 mM Tris, 20 mM acetic acid, 2.5 mM EDTA) at 14°C. The electrophoretic separation conditions were 3 V/cm and a 106°included angle. Chromosomes were resolved using a linear ramped switch time of 120 -480 s over 46 h. Following electrophoresis, gels were stained with ethidium bromide (0.5 g/ml in H 2 O) for 30 min, destained for an additional 30 min in H 2 O, and then photographed under UV illumination. Separated chromosomes were transferred to nylon membranes and hybridized to DNA probes as described above for Southern blot analysis. The only difference was that CHEF gels were subjected to a depurination step (agitation for 20 min in several volumes of 0.25 M HCl) prior to denaturation. Chromosomes prepared from Saccharomyces cerevisiae (Sigma) were electrophoresed in parallel and served as size markers.
Western Blot Analysis-Culture supernatants from the respective C. albicans strains were assayed for presence of PLB by Western blot analysis. Briefly, 100-ml cultures were grown overnight in SDB supplemented with 4% glucose at 30°C. Supernatant from each culture was collected by centrifugation (3,000 ϫ g at 4°C), filtered, and concentrated approximately 500-fold using centrifuge filter units equipped with Biomax-10K membranes (Millipore). Concentrated samples (30 g of total protein) were separated by SDS-polyacrylamide gel electrophoresis on 10% polyacrylamide gels and then electrophoretically transferred to nitrocellulose membranes following standard procedures (29). Membranes were incubated with a guinea pig anti-PLB polyclonal sera (diluted 1:5000) overnight at 4°C and then with a horseradish peroxidase-conjugated goat anti-guinea pig IgG secondary antibody (Chemicon) for 30 min at room temperature. The immune complexes that formed were detected by enhanced chemiluminescence (ECL) according to the manufacturer's instructions (Amersham Pharmacia Biotech).
Colorimetric Assay for the Presence of Free Fatty Acid-The standard incubation mixture for PLB (acylhydrolase) activity consisted of 200 M dipalmitoyl (C16:0) phosphatidylcholine and 200 M L-palmitoylcarnitine in 0.1% (v/v) Triton X-100. Concentrated culture supernatant was added (100 g of total protein) and the mixture made up to a final volume of 0.25 ml with 0.1 M sodium citrate, pH 4.0. Reactions were incubated at 37°C for 1 h and then stopped by the addition of chloroform/methanol (1:2, v/v). The reaction products were extracted (20), evaporated to dryness under a stream of nitrogen, and taken up in 50 l of 0.1% (v/v) Triton X-100. The relative levels of free fatty acid in each sample was determined using a colorimetric-based acyl-CoA-oxidase system assay kit (Boehringer Mannheim).
Quantification of Candidal Penetration of Host Cell Monolayers by Scanning Electron Microscopy-Equal inocula (1:1, Candida:host cells) of SC5314 or plb1-⌬ 2 were used to infect confluent monolayers of human umbilical vein endothelial (HUVEC) and HT-29 epithelial cells grown on cell culture inserts (1-m pore size, Becton Dickinson). HUVECs were grown in M199 medium supplemented with endothelial cell growth supplement (75 g/ml), heparin (100 mg/ml), and fetal bovine serum (20% v/v), whereas HT29 cells were cultured in RPMI 1640 supplemented with glutamine (1% w/v) and fetal bovine serum (10% v/v). Inserts containing Candida and host cells were then incubated for 2 h at 37°C and 5% CO 2 . Host cell monolayers were fixed and processed for scanning electron microscopy as described previously (38). Candidal hyphae were then viewed using a JOEL 480 scanning electron microscope and scored for the number of fungal hyphae penetrating host cells (at least 200 hyphae were counted in each of three separate experiments).
Immunogold Electron Microscopy-BALB/c mice (7-8-week-old females) were infected with 7.5 ϫ 10 5 blastospores of C. albicans strain SC5314 via the lateral tail vein. Kidneys were harvested 24 h postinfection and processed for immunogold electron microscopy as described previously (39). Grids (400-mesh nickel) containing kidney sections were incubated for 1 h in 1:100 diluted polyclonal PLB antiserum or control goat serum. Following a rinse with Tris-buffered saline (50 mM Tris-HCl, pH 7.5, 150 mM NaCl) containing 0.1% Tween 20, the grids were incubated in 1:5 diluted goat anti-guinea pig IgG secondary antibody conjugated to 25 nm colloidal gold spheres (Electron Microscopy Sciences) for 30 min. The grids were washed in distilled H 2 O, stained for 15 min with aqueous uranyl acetate (2% w/v), and again washed with distilled H 2 O. Prepared sections were viewed with a JOEL transmission electron microscope at 60 kV.
Experimental Infections in Mice-Strains of C. albicans were grown in YPD at 30°C until mid-exponential phase. Cells were harvested, washed, counted, and resuspended at a density of 1.25 ϫ 10 7 /ml in sterile phosphate-buffered saline (PBS). BALB/c mice (7-8-week old females) were infected via the lateral tail vein with 0.2 ml (5 ϫ 10 5 blastospores) of this fungal preparation. Cages were checked twice daily for dead or moribund mice. Mice were categorized as moribund if they displayed the following symptoms: severe lethargy, wasting, and ver-tigo. Such mice were euthanized by CO 2 asphyxiation.
CFU Determinations-Female 7-8-week-old BALB/c mice were infected with 5 ϫ 10 5 blastospores of the respective C. albicans strain as described above. Mice in each group were sacrificed 48 h post-infection, and their livers and kidneys were removed. The organs were weighed, homogenized separately in 5 ml of PBS, and serial dilutions plated on SDA supplemented with chloramphenicol (50 g/ml). The plates were incubated at 37°C for 24 -48 h, after which the number of colony forming units (one colony ϭ 1 CFU) was determined. Values were expressed as log CFU/g of tissue homogenized.
Protein Estimations-Culture supernatants were assayed for total protein content using the Bio-Rad DC Protein Assay kit (Bio-Rad). A standard curve was generated by plotting absorbance values obtained for several dilutions of a 10 mg/ml stock solution of bovine serum albumin.
Statistical Analysis-All statistical comparisons were performed using the StatView 228 version 4.5 software package for Windows 95.

RESULTS
Cloning of the C. albicans PLB1 Gene-The amino acid sequence of the NH 2 -terminal region as well as an internal peptide fragment of the suspected lysophospholipase-transacylase were determined. The amino acid sequence at the NH 2 terminus was determined to be TSPTNGYAPG, whereas that of the internal peptide fragment was AMLTGAAEIS. Degenerate DNA oligonucleotide primers, designed on the basis of these sequences, were used to clone a portion of the gene encoding this enzyme by PCR. A 261-bp DNA fragment was amplified, gel-purified, cloned into pBSK ϩ (Stratagene), and sequenced (Retrogen, San Diego, CA). Comparisons of the deduced amino acid sequence encoded by this fragment against the non-redundant protein data base using the BLASTP program (36) revealed 53 and 49% homology to PLBs from Penicillium notatum and S. cerevisiae, respectively. This fragment was used to probe a C. albicans genomic DNA library, cloned into the GEM-12 vector. Two plaques that hybridized to the labeled fragment were identified. Restriction endonuclease analysis of these two clones revealed that both shared common inserts of approximately 11.3 kb. The region of DNA encoding the putative PLB was further localized by cross-hybridizing the 261-bp PCR product to both genomic DNA fragments, generated by restriction endonuclease digestion of the 11.3-kb genomic clone, and to mRNA transcripts. By using this approach, we identified a 6.7-kb EcoRI-ClaI fragment that harbored the coding region of the gene (Fig. 1A).
DNA Sequence Analysis-The nucleotide sequence of a 4.8-kb region present on the 6.7-kb EcoRI-ClaI genomic DNA fragment was determined (Fig. 1A). A single open reading frame of 1818 bp, encoding a putative 605-amino acid protein with a molecular mass of 66,441 Da, was identified (Fig. 1B). The amino acid sequence encoded by the 261-bp PCR product was identical to the corresponding sequence predicted by the genomic clone. The only discrepancy was that the genomic clone predicted for a glycine instead of an alanine at position 105 (Fig. 1B). The genomic DNA sequence encodes 17 amino acid residues that are absent from the NH 2 terminus of the mature protein. This stretch of residues was identified as a possible signal sequence using DNA analysis programs (33, 34).
Hydropathy analysis of the predicted protein sequence according to Kyte and Doolittle (40) (Fig. 1C) revealed the presence of a single stretch of hydrophobic amino acids present at the amino terminus (residues 1-18). This segment of amino acids most likely functions as a signal peptide that targets the protein to the endoplasmic reticulum for subsequent processing and, ultimately, secretion.
Since the candidal PLB and the gene encoding it were obtained from different candidal isolates (16240 and SC5314, respectively) raises the possibility that these strains may harbor multiple phospholipase genes with differing degrees of similarity and that a gene other than that which encodes the purified enzyme may have been cloned. We addressed this issue by performing Southern blot analyses on genomic DNA extracted from C. albicans strains SC5314, 16240, and CAI-4, as well as two additional clinical isolates (strains 15153 and 36082) under varying conditions of hybridization stringency. As expected, only a 3.6-kb KpnI-SacI fragment containing a portion of the caPLB1 gene was detected in each strain following hybridization with a DNA probe specific for caPLB1 (data not shown).
Comparison of the candidal protein against the non-redun-dant data base using the BLASTP program revealed 45, 42, and 48% homology to known PLBs from S. cerevisiae (Gen-Bank accession number L23089), P. notatum (accession number P39457), and Torulaspora delbrueckii (accession number Q11121), respectively (Fig. 2). Recent searches have also identified similarity (38%) to PLB from Schizosaccharomyces pombe (accession number Z99258) (Fig. 2). The predicted size of the protein encoded by the candidal gene (605 amino acids) is consistent with the sizes of PLBs from other fungi (protein sizes in these organisms ranged between 612 and 664 amino acids). All of these enzymes, with the exception of those from S. pombe and P. notatum, contain potential signal sequences at their NH 2 -terminal regions that most likely direct the proteins to the secretory pathway. Additionally, a hydrophobic COOHterminal region was also identified in PLBs from S. cerevisiae, S. pombe, P. notatum, and T. delbrueckii. Interestingly, these COOH-terminal regions contain possible consensus sequences that may signal for the addition of a glycosylphosphatidylinositol (GPI) anchor (41,42). In contrast, neither a hydrophobic COOH terminus nor a GPI attachment site was identified in the candidal PLB. Targeted Disruption of caPLB1-Sequential disruption of each caPLB1 allele was verified by Southern blot analysis. The initial construction of caplb1 null mutants suggested that CAI-4 was triploid for the caPLB1 locus. As expected, following transformation with the linearized pcaplb1::hisG-cURA3-hisG disruption cassette, a 7.6-kb band representing the disrupted caPLB1 allele (caplb1::hisG-cURA3-hisG) and a 3.6-kb band indicative of an intact caPLB1 locus were detected in strain plb⌬1, in which one caPLB1 allele had been deleted (Fig. 3B). Following growth on 5-FOA, strain plb⌬1 was again transformed with pcaplb1::hisG-cURA3-hisG to disrupt the remaining caPLB1 allele. Interestingly, three bands were detected by Southern blot analysis of genomic DNA isolated from strain plb⌬2 (Fig. 3B). A 4.7-kb band representing a caplb1::hisG fragment that had been generated following growth on 5-FOA and the expected 7.6-kb caplb1::hisG-cURA3-hisG fragment were both detected. However, a 3.6-kb fragment, which can only arise from an intact caPLB1 allele, also hybridized to labeled probe DNA and confirmed the presence of at least three copies of caPLB1 in the CAI-4 strain background. The remaining caPLB1 allele was deleted as described above. Strain plb⌬3, generated in this way, no longer retained a wild-type caPLB1 locus since only the disrupted caPLB1 alleles, caplb1::hisG-cURA3-hisG and caplb1::hisG, and not a 3.6-kb (caPLB1) band could be detected (Fig. 3B). Chromosomal analysis of each of the disruptants was performed to determine if all three copies of caPLB1 localized to the same chromosome or were present on different ones and to confirm whether parental strains CAI-4 or SC5314 were triploid for this gene. Labeled probe (5Ј upstream) DNA (Fig. 3A) hybridized only to chromosome 6 in strains SC5314, CAI-4, and plb⌬1. However, this same probe hybridized to both chromosome 5 and 6 in strains plb⌬2 and plb⌬3 (data not shown). This finding confirmed that the tri-allelic state likely arose from a mutational event (duplication/translocation, improper targeting of transforming DNA, or by gene conversion) which occurred after the construction of plb⌬1.
Given the above finding, we proceeded to disrupt the caPLB1 gene a second time starting with the original CAI-4 strain. Chromosomal analysis was performed at every stage of this second disruption as a means to screen newly constructed strains for genetic translocations or incorrect targeting of FIG. 4. Immunodetection of PLB in concentrated culture filtrates. Concentrated supernatants were prepared from the indicated strains, and portions containing equal amounts of total protein were subjected to immunoblot analysis using guinea pig anti-PLB polyclonal antisera as detailed under "Experimental Procedures." Lane 1, purified PLB from which antisera was raised. Lane 2, CAI-4. Lane 3, plb1-⌬ 1 . Lane 4, plb1-⌬ 2 . The size of the purified candidal PLB (labeled phospholipase B) is approximately 84 kDa.

FIG. 3. Targeted disruption of the caPLB1 gene.
A, construction of pcaplb1::hisG-cURA3-hisG. DNA fragments used as probes are underlined. A 1184-bp HindIII-BamHI fragment was replaced with the ura-blaster cassette containing direct repeats of hisG DNA and the selectable marker caURA3. The entire cassette was released from plasmid backbone, purified, and used to transform C. albicans strain CAI-4 as detailed under "Experimental Procedures." Southern blot analysis was used to confirm correct integration of caplb1::hisG-cURA3-hisG at the chromosomal caPLB1 locus. Total chromosomal DNA was isolated and digested with KpnI and SacI. Following agarose gel electrophoresis and transfer to nylon membranes, the DNA was probed with a digoxigenin-labeled DNA fragment (5Ј upstream) containing a 5Ј-portion of the caPLB1 gene. The expected sizes of the hybridizing KpnI-SacI fragments from the caPLB1, caplb1::hisG-cURA3hisG, and caplb1::hisG alleles are 3.6, 7.6, and 4.7 kb, respectively. B, blot revealing that the first disruption process unexpectedly required three rounds of transformation to completely delete all alleles of caPLB1. C, blot showing the results of a second independent disruption of caPLB1, in which only two rounds of transformation were needed. Strain plb1-⌬ 1uraϪ , the parent to plb1-⌬ 2 , was a ura Ϫ isolate recovered following growth of plb1-⌬ 1 on 5-FOA. transforming DNA. Initial transformation of CAI-4 resulted in nine isolates that were disrupted in one caPLB1 allele since a 7.6-kb band representing the disrupted caPLB1 allele (pcaplb1::hisG-URA3-hisG) and a 3.6-kb band indicative of an intact caPLB1 allele were detected (Fig. 5C). Strain plb1-⌬ 1 was selected for loss of the cURA3 gene by growth on medium containing 5-FOA (Fig. 3C), and then again transformed with the disruption cassette. One isolate (plb1-⌬ 2 ) produced a Southern hybridization pattern, a 7.6-kb caplb1::hisG-cURA3-hisG band and a 4.7-kb caplb1::hisG band, indicative of complete disruption of caPLB1 (Fig. 3C). Interestingly, examination of seven other transformants revealed that the caplb1::hisG allele, generated following the first round of transformation and 5-FOA selection, may have been targeted for integration instead of the remaining caPLB1 allele. Thus, these isolates were indistinguishable from strain plb1-⌬ 1 . Chromosomal analysis of strains disrupted for one or both caPLB1 alleles showed that only chromosome 6 hybridized to the 5Ј upstream probe. However, no signal was detected from chromosomes prepared from strain plb1-⌬ 2 when the same blot was reprobed with a portion of the gene (BamHI internal) (Fig. 3A) that would have been deleted from chromosomal caPLB1 loci following successful disruption.
Secretion of PLB by the parental as well as each sequential caplb1 null mutant was determined by Western blot analysis using anti-PLB polyclonal sera. PLB was detected only in concentrated culture filtrates produced by parental and plb1-⌬ 1 strains but not in supernatant produced by strain plb1-⌬ 2 (Fig.  4).
PLB catalyzes the concomitant removal of both acyl chains from dipalmitoylphosphatidylcholine. Given that the end products of this reaction are fatty acid and glycerylphosphorylcholine, a colorimetric method (43) that detects the presence of free fatty acids was used to measure the relative PLB (acylhydrolase) activity in concentrated culture filtrates (Table II). Lpalmitoylcarnitine was added to these reactions since it was shown previously to increase the hydrolase activity of PLB by 2-fold (20). This was necessary since lysophospholipase and transacylase are the predominant phospholipase activities exhibited by the enzyme (20). The levels of free fatty acid, relative to CAI-4, released from dipalmitoylphosphatidylcholine following incubation with culture filtrates obtained from CAI-4 and plb1-⌬ 1 was 100 and 54%, respectively. In contrast, culture filtrate produced by the caplb1 null mutant, plb1-⌬ 2 , liberated less than 1% of the free fatty acid level relative to CAI-4. This level of free fatty acid was consistent with the level released by uninoculated concentrated culture media, which served as a negative control in these assays.
Phenotypic characterization of strains deleted for caPLB1 revealed that both growth rate and germ tube formation were similar to the corresponding isogenic parental strains SC5314 and CAI-4 (data not shown). Furthermore, disruption of caPLB1 did not alter the ability of C. albicans to adhere to cultured endothelial or epithelial cell monolayers (data not shown). Thus, caPLB1 is not essential for normal candidal growth, morphology, or adherence.
Expression of PLB during Candidal Infection-Immunogold electron microscopy was used to show that PLB is expressed in vivo during candidal infection. Sections of kidney tissue prepared from mice infected with candidal strain SC5314 were reacted with either PLB antiserum or goat serum, which served as a negative control. Immunogold complexes were formed following incubation of the tissue sections with PLBantiserum but not with control goat serum (Fig. 5, A and B). Secretion of PLB was diffuse and did not appear to be localized to a specific hyphal region.
Deletion of caPLB1 Attenuates Candidal Virulence-The effect of caPLB1 disruption on virulence was evaluated in a murine model of hematogenously disseminated candidiasis. All mice infected with parental strain SC5314 succumbed to candidal infection within 9 days. In contrast, 50 and 60% of mice infected with either the PLB-deficient strain plb1-⌬ 1 or plb1-⌬ 2 , respectively, were alive at day 15 (Fig. 6). The mean survival time (days, ϮS.D.) for mice infected with SC5314 was 4.4 Ϯ 2.1 compared with 12.7 Ϯ 2.7 and 13.3 Ϯ 2.6 for mice infected with plb1-⌬ 1 or plb1-⌬ 2 , respectively. Statistical comparisons of the survival curves by Logrank analysis revealed that mice infected with either strain plb1-⌬ 1 or plb1-⌬ 2 survived significantly longer than mice infected with strain SC5314 (p Ͻ 0.0001 for both comparisons).
PLB-deficient Strains Are Cleared More Rapidly from Host Tissues-Tissue fungal burden experiments were carried out to assess the severity of infection caused by wild-type and mutant strains. Consistent with the results of the survival experiment, candidal strains plb1-⌬ 1 and plb1-⌬ 2 were cleared significantly faster from the kidneys and brain compared with SC5314 (Table II). The relative rates of clearance are as follows: plb1-⌬ 2 Ͼ plb1-⌬ 1 Ͼ SC5314. Gross inspection of the kidneys prior to homogenization showed that numerous, visible candidal foci covered the renal cortex of mice infected with strain SC5314. In contrast, no candidal foci were visibly detectable on renal surfaces of mice infected with either of the PLB-deficient strains. Southern blot analysis of total chromosomal DNA prepared from several representative colonies confirmed that the genotypes of the isolates recovered from tissue homogenates were identical to the corresponding strains used to infect the mice (data not shown).
The PLB-deficient Mutant Is Less Able to Penetrate Host Cells-Strains SC5314 and plb1-⌬ 2 were compared for their ability to penetrate HUVEC and HT-29 host cell monolayers using scanning electron microscopy. As expected, both the parent and PLB-deficient mutant formed germ tubes and adhered to the host cell monolayers. However, the capacity of the PLBdeficient strain to penetrate HUVEC and HT-29 host cell was reduced relative to SC5314. The percentage (ϮS.D.) of penetrating SC5314 hyphae was 66.7 Ϯ 1.7 for the endothelial cell line and 57.8 Ϯ 2.5 for the epithelial cell line. In contrast, the percentages of penetrating plb1-⌬ 2 hyphae were 37.3 Ϯ 1.3 and 29.0 Ϯ 2.9 for the HUVEC and HT-29 cell lines, respectively. Statistical analysis of the data showed that the differences were significant (p Ͻ 0.0002 for HUVEC; p Ͻ 0.0003 for HT-29).

DISCUSSION
In a previous study we obtained evidence supporting the notion that phospholipase secretion is a virulence determinant of C. albicans (21). However, the candidal strains used were genetically unrelated clinical isolates. Therefore, to overcome the problem of variation in strain background, we initiated a genetic approach to evaluate the relevance of phospholipases to candidal virulence. We determined the amino acid sequences of two peptides from a recently purified (20) extracellular C. albicans enzyme exhibiting phospholipase activity and cloned the gene encoding it by PCR. The availability of the cloned gene, caPLB1, which encodes a C. albicans phospholipase B homolog, enabled us to construct genetically defined PLB-deficient mutants by targeted gene disruption.
When analyzed by SDS-polyacrylamide gel electrophoresis, the purified candidal enzyme has a molecular mass of 84 kDa (20). The predicted molecular mass of the protein encoded by the caPLB1 gene is 66,441 Da. The observed difference in molecular mass between the native enzyme and the protein predicted from the nucleotide sequence of caPLB1 may be due to post-translational glycosylation since we have previously shown that the native enzyme is a glycoprotein (20). In this regard, all known fungal PLBs studied thus far are highly glycosylated (44 -46).
A feature that distinguishes the candidal enzyme from the other known fungal PLBs is the absence of a hydrophobic COOH terminus. Such hydrophobic COOH-terminal regions may ultimately be replaced by a GPI anchor. Indeed, potential GPI-anchoring sites, immediate to the hydrophobic COOH terminus, were identified in fungal PLBs from S. cerevisiae, S. pombe, T. delbrueckii, and P. notatum (41,42). Proteins modified with a GPI anchor may be transiently tethered to the plasma membrane or ultimately cross-linked to the insoluble glucan component of the cell wall (47)(48)(49). Release of proteins associated with the plasma membrane would require the action of a GPI-specific phospholipase. In this view a GPI anchor may serve to regulate the release of the enzyme to the surroundings. Unlike PLBs from nonpathogenic fungi, the candidal PLB, escaping GPI anchoring, would likely be directly secreted. Such a characteristic may enhance the virulence of C. albicans. Further characterization of PLBs from each fungal species will be necessary to clarify whether any are GPI-anchored, and what effect this modification may have on the function and subcellular localization of PLBs.
Initial disruption of caPLB1 was complicated by the finding that three rounds of transformation were required, suggesting that strain CAI-4 was triploid for this locus. This finding seemed plausible since disruption of the C. albicans chitin synthase (CHS2) and catalase (CAT1) genes also revealed that they are triploid (9,50). However, since a DNA probe specific for caPLB1 sequences hybridized to chromosomes 5 and 6 in both strains plb⌬2 and plb⌬3, but only to chromosome 6 in strains plb⌬1, SC5314, and CAI-4, suggests that the triallelic state of the caPLB1 genetic locus likely originated from a gene duplication/translocation event rather than an inherent triploid state in CAI-4. Because the unforeseen event occurred during the construction of strain plb⌬2, either growth on 5-FOA or the transformation procedure itself could have induced the mutation.
Redisruption of caPLB1 showed that the caplb1::hisG allele, generated following growth of strain plb1-⌬ 1 on 5-FOA, may have been integrated with transforming DNA at a higher frequency than the wild-type caPLB1 allele. This finding was also observed during the first disruption process since only a small number of plb⌬3 isolates were produced following transformation of plb⌬2, whereas the majority of isolates retained a wildtype allele. In this view the caplb1::hisG allele may, therefore, compete with wild-type loci as a site for recombination with transforming DNA. Alternatively, the caplb1::hisG allele could have reverted to a wild-type allele by a gene conversion event with the remaining caPLB1 allele prior to the next round of transformation.
Although we cannot discount the fact that some candidal genes may be aneuploid (9,50), our study with caPLB1 highlights the importance of fully characterizing the nature of a suspected triploidy even if mutant strains are constructed by the generally reliable technique of targeted gene disruption. Intensive molecular genetic analysis may disclose potential mutations that otherwise go undetected, especially if disruption of the gene in question produces no obvious phenotypic defect.
caPLB1 gene product is not essential to growth, germination, or adherence characteristics of the fungus. However, the virulence of strains partially or completely abolished for PLB activity was significantly attenuated compared with the parental strain SC5314. The finding that the virulence of strain plb1-⌬ 1 , which harbors only a single deleted caPLB1 allele, was also significantly reduced compared with SC5314 suggests that there may be a dose-dependent relationship between PLB and virulence. Thus, a threshold level of PLB may be required to FIG. 6. Survival of mice following experimentally induced candidiasis. The virulence of heterozygous (plb1-⌬ 1 ) and homozygous (plb1-⌬ 2 ) disruptants was compared with the parental strain SC5314 in an intravenous murine model of hematogenously disseminated candidiasis. BALB/c mice (10 per group) were infected via the lateral tail vein with 5 ϫ 10 5 blastospores of the respective C. albicans strain. Cages were checked twice daily for dead or moribund mice. Survival was monitored for 14 days. effectively enhance candidal virulence. The fact that deletion of caPLB1 did not render C. albicans strains completely avirulent underscores the notion that candidal pathogenicity is multifactorial and is regulated by more than one determinant (51).
Increased clearance rates of PLB-deficient strains, combined with their decreased ability to penetrate host cell monolayers, indicate that PLB may play a role in candidal virulence by causing direct damage to host cell membranes (possibly by enzymatic degradation of phospholipid constituents). Such injury would allow fungal hyphal elements to more effectively traverse the vascular endothelium, ultimately increasing the rapidity of dissemination to and colonization of target organs.
The availability of genetically defined mutants that fail to secrete PLB has enabled us to unambiguously show that this enzyme contributes to the virulence of C. albicans. Our data suggest that PLB likely enhances candidal virulence by damaging host cell membranes. However, given the substrate specificity of the enzyme, PLB could also be involved in an as yet unidentified signal transduction pathway potentiated by lysophosphatidylcholine levels (52). The PLB-deficient mutants should contribute to the overall understanding of candidal pathobiology by providing clues to the mechanism(s) by which this fungal pathogen interacts with and invades host tissues. A better perception of the processes operative during candidal infection may well stimulate the development of new and improved antifungal agents and therapeutic approaches for the treatment and, more importantly, prevention of candidiasis.