Molecular Cloning and Characterization ofWdPKS1, a Gene Involved in Dihydroxynaphthalene Melanin Biosynthesis and Virulence in Wangiella(Exophiala) dermatitidis

ABSTRACT 1,8-Dihydroxynaphthalene (1,8-DHN) is a fungal polyketide that contributes to virulence when polymerized to 1,8-DHN melanin in the cell walls of Wangiella dermatitidis, an agent of phaeohyphomycosis in humans. To begin a genetic analysis of the initial synthetic steps leading to 1,8-DHN melanin biosynthesis, a 772-bp PCR product was amplified from genomic DNA using primers based on conserved regions of fungal polyketide synthases (Pks) known to produce the first cyclized 1,8-DHN-melanin pathway intermediate, 1,3,6,8-tetrahydroxynaphthalene. The cloned PCR product was then used as a targeting sequence to disrupt the putative polyketide synthase gene, WdPKS1, in W. dermatitidis. The resultingwdpks1Δ disruptants showed no morphological defects other than an albino phenotype and grew at the same rate as their black wild-type parent. Using a marker rescue approach, the intactWdPKS1 gene was then successfully recovered from two plasmids. The WdPKS1 gene was also isolated independently by complementation of the mel3 mutation in an albino mutant of W. dermatitidis using a cosmid library. Sequence analysis substantiated that WdPKS1 encoded a putative polyketide synthase (WdPks1p) in a single open reading frame consisting of three exons separated by two short introns. This conclusion was supported by the identification of highly conserved Pks domains for a β-ketoacyl synthase, an acetyl-malonyl transferase, two acyl carrier proteins, and a thioesterase in the deduced amino acid sequence. Studies using a neutrophil killing assay and a mouse acute-infection model confirmed that all wdpks1Δ strains were less resistant to killing and less virulent, respectively, than their wild-type parent. Reconstitution of 1,8-DHN melanin biosynthesis in awdpks1Δ strain reestablished its resistance to killing by neutrophils and its ability to cause fatal mouse infections.

The zoopathogenic fungus Wangiella (Exophiala) dermatitidis is one of many form species of the Fungi Imperfecti, which are darkly pigmented (dematiaceous) owing to the deposition of 1,8-dihydroxynaphthalene (1, melanin in their cell walls (22,45). This fungus has recently become better known as a paradigm for the causative agents of phaeohyphomycosis and other emerging dermatomycoses of humans, because of its increasing detection as a systemic pathogen in both immunocompetent and immunocompromised patients (34,35). Moreover, because W. dermatitidis has a well-defined polymorphic nature and a well-characterized cell wall chemistry, it serves as an excellent model for the more than 100 other dematiaceous fungal pathogens of humans (14,37,42).
Although dark pigments of fungi are often called melanin without regard to mode of enzymatic synthesis or chemical composition, most syntheses of melanin are attributed to either a phenoloxidase, e.g., laccases and tyrosinases, or a polyketide synthase (Pks) of a pentaketide biosynthetic pathway (52). The phenoloxidases have been found mostly among basidiomycete fungi and are usually composed of soluble enzymes with broad substrate specificities (9,33). In contrast, the pentaketide pathway that leads to 1,8-DHN melanin biosynthesis is mostly as-sociated with known or suspected ascomycetes and is very substrate specific (4). Also, while these two kinds of melanins are not essential for fungal growth, they have been documented to have relevance to virulence in W. dermatitidis (15)(16)(17)(18)41) and Cryptococcus neoformans (7,30) as well as in many phytopathogenic fungi, including Colletotrichum lagenarium (26), Magnaporthe grisea (10,54), and Alternaria alternata (25). In addition, disruption of the polyketide synthase gene alb1 of Aspergillus fumigatus reduces its virulence by inhibiting conidial pigmentation, although the end product of the pentaketide pathway in this human pathogen is probably not 1,8-DHN melanin (46).
In this study, a gene (WdPKS1) that encodes a putative polyketide synthase (WdPks1p) of W. dermatitidis was cloned, sequenced, and disrupted. The WdPKS1 gene contained a single open reading frame, consisting of three exons separated by two short introns. The predicted WdPks1p consisted of 2,177 amino acids and showed significant similarities with other polyketide synthases, but particularly those encoded by the pks gene of C. lagenarium, the alb1 gene of A. fumigatus, and the wA gene of Aspergillus nidulans (8,44,46). The derived protein also contained sequences for the highly conserved ␤-ketoacyl synthase, acetyl-malonyl transferase, acyl carrier protein (ACP), and thioesterase domains, which are all characteristic of a type I polyketide synthase (23). Disruption of WdPKS1 produced strains that had albino phenotypes, which strongly indicated that WdPKS1 was involved in 1,8-DHN melanin biosynthesis. Support for this hypothesis was provided by precursor feeding studies and also by complementation experiments that remelanized the albino wdpks1⌬-1 mutant, as well as mutants described previously (13) with mel3 lesions, but not those with mel4 lesions. When tested in human neutrophils or in an acute mouse model, the albino wdpks1⌬ mutants were less resistant to neutrophil killing and less virulent, respectively, than their wild-type parent, as was found previously for other Mel Ϫ mutants of W. dermatitidis tested similarly (18,41). The importance of WdPKS1 to virulence was further supported by showing that the disruptants had no growth rate defects or other morphological abnormalities. Furthermore, reconstruction of 1,8-DHN melanin biosynthesis in the wdpks1⌬-1 mutant by complementation with WdPKS1 reestablished its ability to resist killing by human neutrophils and to cause fatal mouse infections.
Preparation and analysis of nucleic acid. Genomic DNA of W. dermatitidis was isolated by spheroplasting with Zymolase-20T (ICN Biomedicals, Inc., Aurora, Ohio) followed by detergent lysis, phenol-chloroform extraction, and ethanol precipitation as previously described (55). Total RNA was isolated by the FIG. 1. (A) Predicted structure for the integration of pBF9 at the WdPKS1 locus. The 772-bp PCR product of WdPKS1 was used as a target sequence to disrupt the gene. The whole gene was then cloned by a marker rescue method. The two rescued plasmids pBF10 and pBF32 are also shown. The arrow labeled WdPKS1 indicates the position of the gene and the direction of its transcription. Prior to transformation, pBF9 was linearized with SacII. (B) Southern hybridization analysis of five transformants in which pBF9 was integrated into the W. dermatitidis genome. Genomic DNA of wild type; albino transformants 1, 2, and 3; and black transformants 4 and 5 (lanes 6 and 1 to 5, respectively) was extracted, digested with XhoI, fractionated on a 0.9% agarose gel, and subjected to Southern blotting analysis using the 772-bp WdPKS1 PCR fragment as a probe. Albino transformants 1, 2, and 3 showed hybridization consistent with site-specific integration of pBF9 into the WdPKS1 locus, whereas the plasmid was ectopically integrated at different genomic positions in black transformants 4 and 5.
hot phenol method (1). Southern and Northern blotting were performed using standard methods (1). DNA fragments (25 ng) used for probes in Southern and Northern analysis were labeled with [ 32 P]dATP by using a Prime-a-Gene kit (Promega, Madison, Wis.). Plasmids with WdPKS1 fragments were sequenced by the Institute for Cellular and Molecular Biology of The University of Texas at Austin using an ABI Prism 377 Sequencer. Sequence analysis was performed using Wisconsin Package G software (Genetics Computer Group, Inc., Madison, Wis.). The 772-bp PCR fragment of WdPKS1 was amplified from genomic DNA using primers PKS-1 (5Ј TGAATTCGACACGGCCTGTTCA/CTCCA 3Ј) and PKS-2 (5Ј ACATATGGCGGCACTGAAGTTGTTGA 3Ј). The PCRs were performed as described previously (48). The primers used for reverse transcription (RT)-PCR were RT-PKS1 (5Ј CGTCCACTCGCTCACACTCT 3Ј), RT-PKS2 (5Ј ACCGACTAGTCGAGCAT 3Ј), RT-PKS9 (5Ј GGGTGCTGAAGTCTGT AAA 3Ј), and RT-PKS10 (5Ј TCCCTCTGTGTCGAGAAT 3Ј). First-strand synthesis of the cDNA was achieved using SuperScript II reverse transcriptase (Life Technologies Inc., Rockville, Md.) according to the instructions of the manufacturer. PCR amplification of the first strand synthesized was for 35 cycles as follows: premelt, 94°C for 5 min; denaturation, 94°C for 1 min; hybridization, 50°C for 2 min; and extension, 72°C for 3 min (10 min on the last cycle). For the rescue of fragments flanking the WdPKS1 transgene insertion, DNA from the wdpks1⌬-1 clone was digested to completion with appropriate restriction enzymes and diluted to Յ1 g/ml to favor intramolecular ligation with T4 DNA ligase (New England Biolabs Inc., Beverly, Mass.). After transfection of E. coli XL1-Blue (Stratagene) by electroporation using a Gene Pulse apparatus (Bio-Rad Laboratories, Richmond, Calif.), plasmids were isolated from chloramphenicol-resistant colonies.
Plasmid construction. Plasmid pBF5 was constructed by ligating a WdPKS1, 772-bp PCR fragment into pGEM-T vector (Promega). The WdPKS1 disruption plasmid pBF9 (see Fig. 1A) was constructed by inserting the 800-bp ApaI-SalI fragment containing the PCR product of WdPKS1 from pBF5 into pCB1004 (provided by J. Sweigard, DuPont Co., Wilmington, Del.), which is a pBluescript SK(ϩ)-based vector that contains the hygromycin phosphotransferase gene (hph) from E. coli, which confers resistance to hygromycin B (HmB), and the tryptophan synthase (trpC) promoter from A. nidulans (6). The complementation plasmid pBF50 (see Fig. 5A) was constructed by ligating the ApaI-and NotIdigested plasmid pCB1532 (provided by Kevin McCluskey, Fungal Genetics Stock Center, Kansas City, Kans.), which is a pBluescript SK(ϩ)-based vector that contains a surfonyl-urea (SUR) resistance gene, with both the ApaI-AclI fragment from pBF10 and the AclI-NotI fragment from pBF32. Plasmid pMO-cosX (provided by Marc J. Orbach, University of Arizona, Tucson, and DuPont Experimental Station, Wilmington, Del.), which contains cos sites from bacteriophage , and plasmid pCB1004, which contains a bacterial origin, a bacterial chloramphenicol resistance gene, and the hph gene, were used to construct the cosmid vector pCB1004cos. The XbaI site in plasmid pCB1004 was removed to

VOL. 69, 2001
CLONING AND CHARACTERIZATION OF WdPKS1 produce the plasmid pCB1004⌬XbaI. A 2.9-kb DNA fragment in plasmid pMOcosX, which contains the bacteriophage cos sequence, was released by ClaI and BamHI double digestion and was inserted into pCB1004⌬XbaI to produce the new cosmid vector pCB1004cos. Construction of cosmid library. A genomic cosmid library of W. dermatitidis was constructed using pCB1004cos by the method of Osiewacz (38). After digesting pCB1004cos with XbaI and dephosphorylating the cohesive ends using calf intestine alkaline phosphatase (Promega), the linearized plasmid was cleaved with BamHI to produce two cosmid arms. The cosmid arms were then ligated to DNA fragments generated from high-molecular-weight genomic DNA (ϳ150 kb) of W. dermatitidis wild-type 8656 (concentration, 100 g/ml), which had been partially digested with Sau3AI to generate fragments of 40 to 50 kb, recovered by ethanol precipitation, and resuspended in double-distilled H 2 O. Incubation of the ligation reaction mixture was for 16 h at 16°C in a total volume of 10 l, using 5 g of partially digested genomic DNA and 1 g of the two cosmid arms. The reaction products were then used immediately for in vitro packaging, which together with the determination of the titer of the cosmid packaging reaction mixture was carried out according to the instruction manual from the Gigapack III XL packaging kit (Stratagene Inc.). The resulting library, consisting of about 5 ϫ 10 4 colonies, was pooled and stored in 25% glycerol at Ϫ70°C, prior to use in experiments designed to rescue Mel Ϫ mutants by complementation.
Measuring of phagocytosis and oxidative burst by flow cytometry. To assess phagocytosis, sedimented (3,000 ϫ g at 4°C for 5 min) and washed (in 5 ml of sterile 0.9% [wt/vol] NaCl) yeast cells from SDA broth (5 ml) were incubated with bis-carboxyethyl-carboxyfluorecein-pentaacetoxy-methylester (BCECF-AM) (final concentration, 1 mol/liter; Roche Biochemicals, Mannheim, Germany) for 30 min at 37°C in phosphate-buffered saline (PBS) (1 ml) as previously described (40,41). The labeled cells (5 ϫ 10 6 ) were then incubated at 37°C for a maximum of 120 min with 1 ml of heparinized (10 units of heparin [as defined by the German Pharmacopeia]/ml [equivalent to ca. 5 units of heparin {as defined by the U.S. Pharmacopeia}/ml]) whole blood from healthy donors in a Thermomixer (Eppendorf, Hamburg, Germany) at 1,000 rpm. At 0, 10, 30, and 60 min, samples (100 l) were removed and immediately mixed with 2 ml of ice-cold lysis buffer (Becton Dickinson, Heidelberg, Germany) to lyse the erythrocytes, and then after 2 h, leukocyte and yeast cell isolations were carried out by centrifugation (10 min, 4°C, 1,300 rpm; Beckman GS-6R centrifuge). After being washed twice in ice-cold PBS, cells were resuspended in PBS (500 l) and analyzed by flow cytometry, using a FACScan flow cytometer (Becton Dickinson) and Cellquest (Becton Dickinson) software (40,41). Instruments setting, linear parameters for forward and side scatter, and logarithmic parameters for FL1 and FL2 with the best test performance were revealed to be the same as those set previously for experiments with W. dermatitidis and its melanin-deficient mutants (41). Oxidative burst was assessed during the phagocytosis by incubation of unlabeled yeast cells in heparinized blood under conditions identical to those described above. Dihydrorhodamine (DHR) (Molecular Probes, Eugene, Oreg.) was added to samples at a final concentration of 10 mg/liter as described previously (41). The association of neutrophils with the labeled yeast cells and the oxidative burst of the neutrophils induced by unlabeled yeast cells in the presence of DHR were detected by an increase in fluorescence of the neutrophils. The increased fluorescence signal was expressed as a percentage of nonfluorescing neutrophils.
Neutrophil killing assay. For quantification of killing by human neutrophils, yeast cells of the strains tested were diluted in PBS to 10 3 to 10 4 CFU per ml. The diluted cells (100 l) and fresh heparinized human blood (900 l) were then mixed and rotated at 37°C for 4 h. Initial viable counts and cell counts were determined after 10 min and after 1, 2, 3, and 4 h of rotation by plating samples of yeast cells either undiluted (20 l) or after dilution with 180 l of PBS (20 l) on SDA.
Microscopy. To ensure the intracellular location of the yeast cells associated with the neutrophils, representative samples used for determination in flow cytrometry were examined by epifluorescence interference contrast microscopy (Leitz DM RB microscope; Leica, Wetzlar, Germany) as described previously (40,41). From three independent assays of each strain studied, we examined 300 yeast cells with respect to their association with the neutrophils at the beginning, after 30 min, and after 60 min of incubation in heparinized blood.
Virulence studies with mice. Test strains (wdpks1⌬-1, wdpks1⌬-2, wdpks1⌬-3, the wild type, and the complemented strains 501 and 502) of W. dermatitidis were cultured in YPD (5 ml) overnight at 30°C with shaking. An aliquot of the overnight culture was used to inoculate 50-ml YPD cultures, which were then grown overnight to mid-log phase. Yeast cells were harvested, washed three times with sterile water, counted on a hemacytometer, and diluted to a final density of 9 ϫ 10 7 cells/ml. Virulence of the strains was then tested in an immunocompetent (normal) mouse model system. Male ICR mice ( ., semiconserved substitutions. The numbers at the right represent the number of amino acids depicted from the first methionine of each derived protein. ethical treatment of animals. Mice (10 per yeast strain) were inoculated via the lateral tail vein with 100 l of the cell suspension (9 ϫ 10 7 cells/ml), such that each mouse received a final dose of 9 ϫ 10 6 cells. To determine the number of viable yeast forms injected into each mouse, an aliquot of the suspension used for injection was diluted and plated in top agar (0.1% Noble agar) onto YPD plates. The plates were incubated at 30°C for 48 to 72 h, and percent viability was determined. Mice were checked three times daily for survival or signs of infection up to 13 days. Visible signs of infection were torticollis, ataxia, or lethargy. Infected mice were considered moribund when they were unable to access food or water. Moribund mice were humanely sacrificed by cervical dislocation under anesthesia.
Statistics. Differences in the extent of phagocytosis and oxidative burst exhibited by the neutrophils and the killing of the W. dermatitidis wild-type strain, the isogenic wdpks⌬ mutants, and complemented revertants were evaluated by the nonparametric Mann-Whitney U test for unpaired samples (P Ͻ 0.005). Survival fractions in virulence tests were calculated by the Kaplan-Meier method, and survival curves were tested for significant difference (P Ͻ 0.01) by the Mantel-Haeszel test using GraphPad Prism software (version 3.00 for Windows). Probability values of Ͻ0.05 were considered significant.
Nucleotide sequence accession number. The nucleotide sequence of the WdPKS1 gene was assigned GenBank accession no. AF 130309.

RESULTS
Disruption of WdPKS1 using a PCR product for gene targeting produced albino strains. PCR primers having a design based on fungal Pks conserved regions allowed amplification of a 772-bp PCR product from genomic DNA of W. dermatitidis, which was then cloned, sequenced, confirmed by sequence similarity to encode a putative polyketide synthase (WdPks1p), and used to produce the disruption plasmid pBF9 (Fig. 1A). Among 53 HmB-resistant transformants obtained after electroporation of pBF9 linearized with SacII, five were albino, suggesting, as expected, that WdPKS1 encoded a Pks involved in melanin biosynthesis. Southern analysis of three albino transformants (wdpks1⌬-1, wdpks1⌬-2, and wdpks1⌬-3) and two melanized transformants (211 and 212), using the 772-bp WdPKS1 PCR fragment as a probe, showed that the WdPKS1hybridized DNA band was suitably shifted from a wild-type position of 4.1 kb to a disruptant position of 10 kb among XhoI-digested fragments (Fig. 1B). These results confirmed that the albino phenotype of the transformants was due to site-specific integration and disruption of WdPKS1, whereas the bands larger than 4.1 kb in strains 211 and 212 were indicative of ectopic plasmid integrations.
Diffusion cross-feeding experiments showed that albino wdpks1⌬-1 became blackened and produced melanin when cross-fed by 1,8-DHN melanin precursors produced by the brown strains Mel1 (data not shown) and Mel2 ( Fig. 2A) or with the purified melanin biosynthetic pathway intermediate, scytalone (Fig. 2B). These experiments also showed that no visible intermediates were secreted into the medium on which the wdpks1⌬-1 disruption strain was grown. Thus, WdPks1p is involved in a step before 1,3,6,8-THN reductase functions to produce the compound scytalone or before an oxidase converts 1,3,6,8-THN to the colored shunt product flaviolin (53) and consequently likely participates in the production of 1,3,6,8-THN itself. In W. dermatitidis this process is thought to involve two steps, which are defined by the mutations mel3 and mel4 (13). Therefore, additional experiments using standard protocols (13) were carried out to determine whether strains with either the mel3 or the mel4 lesion could be distinguished from the disruption strain wdpks1⌬-1. From the results of the contact cross-feeding experiment with strains having the mel3 mutation (data not shown) or the mel4 mutation (Fig. 2C) and from the acetate feeding experiment (Fig. 2D), wdpks1⌬-1 was clearly distinguished from the latter (mel4) but not the former (mel3), suggesting for the first time that the mel3 lesion was due to a mutation in WdPKS1.
Cloning of WdPKS1 by a marker rescue approach and by cosmid complementation of the mel3 mutation. Plasmid pBF10 carrying the 5Ј end of WdPKS1 was recovered by digesting genomic DNA of wdpks1⌬-1 with KpnI and then allowing selfligation in dilute solution. The 12-kb plasmid, pBF10, was then rescued by transformation of E. coli, and the 5Ј end of the gene was cloned (Fig. 1A). Using the same strategy, the 3Ј end of WdPKS1 was also recovered in a 13-kb rescue plasmid, pBF32, through SpeI digestion (Fig. 1A).
The transforming cosmid responsible for the reversion of the mel3-2 mutation in strain Mel4 was recovered by in vitro packaging of undigested genomic DNA, followed by transduction of E. coli as described by Yelton et al. (56). As expected, retransformation of Mel3 (mel3-1) and Mel4 (mel3-2) with the cosmid caused most transformants of these two strains to regain melanin production capability (data not shown). PCR analysis using primers for WdPKS1 yielded a 772-bp product, which was consistent with the possibility that WdPKS1 was contained in the cosmid (now called cos-Mel3) and was responsible for complementing the mel3 lesions in strains Mel3 and Mel4 (data not shown). WdPKS1 encodes a type 1 Pks. Restriction enzyme mapping of the cloned inserts in pBF32 and pBF10 (Fig. 3A) and comparisons with the map of the cos-Mel3 (data not shown) confirmed that WdPKS1 had been cloned independently by two methods. Therefore, only the two rescued WdPKS1 gene fragments in pBF32 and pBF10 were completely sequenced after a series of subclonings. A single open reading frame was deduced for WdPKS1 and found to encode 2,177 amino acids. Two putative introns (279 to 337 bp and 6540 to 6596 bp) were also identified in WdPKS1, which were confirmed by RT-PCR (Fig. 3B), and by sequence analysis of the RT-PCR products (data not shown). The deduced amino acid sequence showed 46.8, 45.7 and 44.9% identity to pks1 of C. lagenarium, alb1 of A. fumigatus, and wA of A. nidulans, respectively (Fig. 3C), further indicating that the cloned gene encoded a polyketide synthase. Northern analysis showed constitutive expression of WdPKS1 (data not shown).
Because eukaryotic Pks are generally thought to be large multifunctional proteins (type I Pks) (23), the predicted amino acid sequence was investigated further. The ␤-ketoacyl synthase, acetyl-malonyl transferase, and two ACP and thioesterase domains usually found within a type I Pks were all identified (Fig. 4). The putative WdPks1p also showed complete conservation of the putative active site cysteine residue of the ␤-ketoacyl synthase, the active site serine residue of the acetylmalonyl transferase, and the pantotheine-binding serine residue of the ACP (3,8,20,31,36,44,57), which provided further evidence that WdPKS1 was a type I Pks.
Reconstitution of melanin biosynthesis. To confirm that the albino phenotype was not due to secondary, hidden mutations, melanin biosynthesis in the albino strain, wdpks1⌬-1, was reconstituted by integration of WdPKS1 into its genome at the WdPKS1 endogenous site. This was accomplished with the vector pBF50 constructed by subcloning and ligating the 5Ј end of the gene from pBF10 to the 3Ј end of the gene from pBF32 in vector pCB1532, which contained the sulfonyl-urea (SUR) and the ampicillin (amp) resistance genes for selection in W. dermatitidis and E. coli, respectively. Plasmid pBF50 (Fig. 5A), which contained the full-length WdPKS1 with its endogenous promoter, was then trans-formed without being linearized into wdpks1⌬-1 by electroporation. Among the resulting SUR-resistant transformants, 2 of 11 were restored to melanin synthesis (data not shown). Southern analysis of the two complemented strains showed the expected DNA band shifts (Fig. 5B), which confirmed that the cloned WdPKS1 had integrated into the 5Ј portion of the disrupted WdPKS1 gene.
To ensure that any differences detected in susceptibility to killing by neutrophils or loss of virulence in mice was due only to loss of melanin in the wdpks⌬ mutants, the growth rates of wdpks⌬-1, wild-type 8656, and the WdPKS1-complemented strain 501 were also compared in rich YPD (Fig. 6). The results confirmed that wdpks1⌬-1 grew at the same rate as the wild- type 8656 and as the WdPKS1-complemented strain 501, both at 25°C (generation time, 3.3 h) and 37°C (generation time, 2.8 h). Microscopic comparisons of these strains with the wild type grown identically also showed no apparent morphological differences (data not shown).
Fluorescence staining of W. dermatitidis and phagocytosis by human neutrophils. Staining of the W. dermatitidis strains by incubation for 30 min in PBS containing 1 mol of BCECF-AM per liter resulted in a stable green fluorescence in the parent strain and all of its isogenic mutants and complemented strains (501 and 502) (data not shown). Although the darkly pigmented, wild-type strain and the respective complemented strains exhibited a slightly lower level of green fluorescence compared to the albino wdpks1⌬-1 mutant (data not shown), accurate quantitative comparisons of the rates of phagocytosis by the human neutrophils of the strains were still possible. The results showed that phagocytosis by the neutrophils was essentially the same for all the strains during three independent assays (Fig. 7). By 10 min of coincubation, Ͼ98% of yeast cells of all the strains were localized within the phagocytes as judged by microscopy. No significant difference with respect to the proportion of budding yeast cells in relation to their extracellular or intracellular location was observed by epifluorescence interference contrast microscopy (data not shown), which suggested that a good correlation existed between the flow cytometric assay and microscopic observations. Oxidative burst evoked by and killing of W. dermatitidis in human neutrophils. Phagocytosis of yeast cells was paralleled by an oxidative burst in the phagocytosing neutrophils, as suggested by the increased fluorescence detected by flow cytometry. In no case was a significant difference in oxidative bursts detected among the responses evoked after neutrophil phagocytosis of the wild-type strain 8656, the mutant wdpks1⌬-1, and its respective complemented strain 501 (Fig. 8). In contrast, the wdpks1⌬ mutants were considerably more susceptible to killing by the neutrophils than the melanized strains (Fig. 9). In every case, the killing of the melanin-deficient mutants was significantly greater (P Ͻ 0.005) after 120, 180, and 240 min of coincubation with whole blood compared to the wild-type strain 8656 and the respective complemented strains. However, no differences were detected among the rates of killing of the three wdpks1⌬ mutants tested in five independent experiments (Fig. 9).

Disruption of WdPKS1 causes loss of virulence in mice.
Animal studies were carried out to compare the degrees of virulence of the wild type, the wdpks1⌬ disruptants, and the WdPKS1-complemented strains. In at least three independent experiments, mice injected with the wdpks1 disruptants showed a dramatic reduction in mortality compared to that of the wild type and the WdPKS1-complemented strains. Lethality in mice infected with strain 8656 began on days 4 and 5, with mortality rates of 90 to 100% by days 6 and 7 (Fig. 10). In contrast, mice injected with the wdpks1 disruptants showed only 0 to 10% mortality at the end of the experiments (day 13). However, the WdPKS1-complemented strains all recovered full virulence; mortality started on day 5 and reached 70 to 80% by the termination date (Fig. 10). In addition, the Mel4 (mel3-2) cos-WdPKS1-complemented strain was also found to have recovered full virulence (data not shown).

DISCUSSION
Prior studies have strongly indicated that virulence of dematiaceous fungi is influenced by the presence of 1,8-DHN melanin (13, 15-18, 24, 41). The results of this study provide additional support for this concept. After the melanin biosyn- FIG. 9. Percent killing of W. dermatitidis as determined by colony counts prior to and after incubation in heparinized blood. The killing is displayed as the percentage decrease in the number of CFU during the incubation times indicated (n ϭ 5; mean values ϩ standard deviations [error bars]). The asterisk identifies a significant difference from the wild type (wt) as determined by the Mann-Whitney U test (P Ͻ 0.005), whereas n.s. identifies the absence of significant differences from wild type as determined by the same test.

VOL. 69, 2001
CLONING AND CHARACTERIZATION OF WdPKS1 thetic pathway gene, WdPKS1, of W. dermatitidis was cloned and disrupted, the resulting wdpks1⌬ mutants were shown to be more susceptible to killing by neutrophils and to be less virulent in an acute mouse model. One wdpks1⌬ mutant was then shown to regain resistance to neutrophil killing and to exhibit normal virulence in mice when melanin biosynthesis was reconstituted by complementation with WdPKS1. To our knowledge, WdPKS1 is the first melanin biosynthetic pathway gene cloned in W. dermatitidis, or any other dematiaceous pathogen of humans, and shown to contribute directly to virulence.
The WdPKS1 gene was cloned by a PCR method coupled with a marker rescue approach and then sequenced. The predicted WdPks1p showed highly significant sequence similarity with type I Pks (23) and shared domains with other Pks for a ␤-ketoacyl synthase, an acetyl-malonyl transferase, two ACPs, and a thioesterase (3,8,20,31,36,44,57). This result strongly suggested that the WdPKS1 gene encodes a polyketide synthase involved in melanin biosynthesis in W. dermatitidis. In addition, the WdPKS1 gene was also found to be constantly expressed throughout the W. dermatitidis growth cycle at both 25 and 37°C and confirmed to be nonessential, as would be expected of a gene involved in melanin biosynthesis by secondary metabolism. Phenotypic analyses of wdpks disruption mutants further showed that these albino strains had both wildtype growth rates and cellular morphologies in vitro.
Recently, flow cytometry of BCECF-AM-stained W. dermatitidis was found to be suitable for determination of phagocytosis rates of yeast cells by incubation with neutrophils in whole, heparinized, human blood: BCECF leads to a stable intracellular stain of the yeast cells (41). To ensure an intra-cellular location of the yeast cells found associated with neutrophils, epifluorescence microscopy was carried out in combination with interference contrast microscopy of representative experimental samples. In accordance with the prior results (41), no differences were found in the degrees and rates of phagocytosis between the W. dermatitidis wild-type strain and the WdPKS1-complemented and noncomplemented wdpks1⌬ mutants. Evoked oxidative burst, estimated by a DHR method that was also recently validated for W. dermatitidis (41), similarly showed that the amounts and rates of the oxidative bursts induced by the same strains were comparable. This again was taken as confirmation of equal internalization of unstained yeast cells by the human granulocytes, because all the neutrophils involved exhibited a bright green fluorescence in the presence of DHR, which is indicative of a comparable degree of oxidative burst.
The rates of phagocytosis and subsequent amounts of killing of W. dermatitidis determined in a bioassay were also in excellent agreement with the prior results (41) but were obtained with the genetically uncharacterized W. dermatitidis albino mutant Mel3 (mel3-1), which was generated previously by UV mutagenesis (22). Statistically significant differences between the darkly pigmented wild-type strain and the wdpks1⌬-1-complemented strain and the albino wdpks1⌬ mutants were clearly detected after exposure of yeast cells of these strains in whole heparinized human blood for more than 10 min. This observation clearly demonstrated that the presence of intact melanin is the main factor contributing to the difference in killing of the different strains by human neutrophils. Thus, the albino mutant strains of W. dermatitidis are killed by neutrophils in a comparable fashion to Candida albicans and Saccharomyces cerevisiae (unpublished data), which can be taken as an indication that melanin synthesis contributes directly to the virulence of this black yeast species.
The most intriguing phenotype of the wdpks1⌬ disruption strains beyond their loss of pigmentation was their significant loss of virulence in a mouse model of acute infection compared to those of their wild-type parent and the two complemented strains in the wdpks1⌬-1 background. Such reduced virulence in the mouse model was also previously observed with the albino strain Mel3 (mel3-1) (15). However, because the UV used to generate this mutant could have given rise to multiple gene defects not easily detected in an asexual fungus, it might be argued that firm conclusions about the relationship of melanin with virulence in this strain are not be warranted (16)(17)(18). This concern is now moot because of our ability to complement the mel3 lesion with WdPKS1, which clearly established the equivalency of strains with mel3 lesions with our newly derived and less virulent wdpks1⌬ disruption strains.
The importance of a polyketide synthase involved in melanin biosynthesis in the human pathogen W. dermatitidis is now documented for the first time to be equivalent to that of similar fungi found among the plant pathogens. In the plant pathogenic fungi C. lagenarium and M. grisea, deficient 1,8-DHN melanin biosynthesis resulted in nonmelanized appressoria (27,32,39). In both cases, those mutants lost their ability to penetrate plant leaf tissue and thus became avirulent due to the fact that melanin is required for the rigidity of appressoria (2). Similarly, it was reported recently that invasive hyphal growth in W. dermatitidis is dependent on melanin biosynthesis FIG. 10. Mouse survival analysis after injection with wild type and wdpks mutants. Groups of 10 mice received injections of log-phase yeast cells of W. dermatitidis wild type (wt), wdpks1⌬-1, wdpks1⌬-2, wdpks1⌬-3, or the melanin-reconstituted strains 501 and 502 (both are in a wdpks1⌬-1 background). The injections contained 9 ϫ 10 6 cells per mouse, and the mice monitored for 10 to 15 days to determine survival rate. The data presented are average data from at least three experiments, all of which showed very similar results. Survival fractions were calculated by the Kaplan-Meir method, and survival curves were tested for significant difference (P Ͻ 0.01) by the Mantil-Haeszel test using GraphPad Prism software (version 3.00 for Windows). (5). Brush and Money hypothesized that melanized hyphae exert larger turgor-derived forces at their apices than nonmelanized cells in W. dermatitidis, explaining their propensity for fast substrate invasion (5).
Previous studies demonstrated by parasexual genetic methods that strains with the mel3 and mel4 genotypes are representative of mutants with defects in two different enzymes that function during the conversion of acetyl-CoA or malonyl-CoA to 1,3,6,8-THN (12,13). Strains with mutations in these genes can be differentiated also by physiological methods (13). Thus, comparisons of wdpks1⌬-1 with strains with one or the other of these two mutations, plus complementation studies with the cloned WdPKS1 gene itself, can be used to distinguish such mutants. For example, although none of the strains with the mel3, mel4, or wdpks1⌬ mutations secrete visible intermediates into the culture medium, the wdpks1⌬-1 strain contact crossfeeds with strains with the mel4 mutation (Fig. 2C), whereas wdpks1⌬-1 does not do the same with strains with the mel3 mutation (13). Also, whereas the latter two mutant types are both very white, strains with the mel4 mutation are more beige than white with time on YPD or MCD agar, which previously suggested that mel3 represents a lesion in an enzyme that functions prior to mel4 (13). Finally, when grown on 1% (wt/ vol) acetate-supplemented MCD agar, wdpks1⌬-1 and strains with the mel3 lesion remained white, whereas strains with the mel4 lesion turned black (Fig. 2D). Although the basis of these contact cross-feeding and acetate effects is not clear, these physiological tests support the hypothesis that two enzymes are involved in the conversion of acetyl-CoA or malonyl-CoA to 1,3,6,8-THN in W. dermatitidis and that the wdpks1⌬ disruption mutants are equivalent to the previously described mutants with mel3 lesions (13). Furthermore, although WdPKS1 complemented strains with the mel3 lesion, it did not complement a strain with a mel4 lesion. Thus, although the nature of the mel3 mutation is now clear, that of mel4 remains unknown. In this respect, in most fungal 1, 8-DHN melanin systems, there is only one structural polyketide synthase gene known to contribute at this step in the pathway (19,21,29,50). The cloning of the MEL4 gene in the future will enable us to understand better 1,8-DHN melanin biosynthesis in the human pathogen W. dermatitidis and the differences between the pathways in W. dermatitidis and in other black fungi. Unfortunately all attempts to clone this gene by complementation or by insertional mutagenesis and marker rescue have failed (unpublished data).
Genes involved in 1,8-DHN melanin synthesis are clustered in some fungi and not in others. In A. alternata, a melanin pathway gene cluster has been identified that contains at least three pathway biosynthetic genes within a 30-kb region (25), which encode the polyketide synthase, the scytalone dehydratase, and the 1,3,8-THN reductase. Cloning of these three genes in C. lagenarium was also reported (28,39,44). However, in contrast to A. alternata it appears that these same genes are not closely linked in C. lagenarium (27). In M. grisea, classical genetic analysis with melanin mutants indicates none of the pathway genes are closely linked (10). Although pathway genes have not been isolated for C. miyabeanus and Cochliobolus heterostrophus, classical genetics analysis indicates that in both organisms the polyketide synthase and the 1, 3, 8-THN reductase genes but not the scytalone dehydratase gene, are linked (28). In W. dermatitidis, two independently isolated cosmid clones containing WdPKS1 were used to transform mutants with mel1, mel2, and mel4 lesions. However, in no case was complementation observed. Thus, it appears that the mel4 mutation (representative of the unknown enzyme in the pathway) and the genes for the scytalone reductase and for the 1,8-DHN oxidase (polymerase) are not closely linked with Wd-PKS1. More extensive searches to identify these genes and others related to 1,8-DHN melanin biosynthesis in W. dermatitidis are in progress.