Global Control of Dimorphism and Virulence in Fungi

See allHide authors and affiliations

Science  28 Apr 2006:
Vol. 312, Issue 5773, pp. 583-588
DOI: 10.1126/science.1124105


Microbial pathogens that normally inhabit our environment can adapt to thrive inside mammalian hosts. There are six dimorphic fungi that cause disease worldwide, which switch from nonpathogenic molds in soil to pathogenic yeast after spores are inhaled and exposed to elevated temperature. Mechanisms that regulate this switch remain obscure. We show that a hybrid histidine kinase senses host signals and triggers the transition from mold to yeast. The kinase also regulates cell-wall integrity, sporulation, and expression of virulence genes in vivo. This global regulator shapes how dimorphic fungal pathogens adapt to the mammalian host, which has broad implications for treating and preventing systemic fungal disease.

Microbial pathogens that inhabit our environment must undergo a radical change to survive inside a mammalian host. Among the more than 100,000 different species of environmental fungi are six phylogenetically related ascomycetes called the dimorphic fungi: Blastomyces dermatitidis, Coccidioides immitis, Histoplasma capsulatum, Paracoccidioides brasiliensis, Sporothrix schenkii, and Penicillium marneffei. These fungi change morphology once spores are inhaled into the lungs of a mammalian host from hyphal molds in the environment to pathogenic yeast forms. Dimorphic fungi inhabit the soil world-wide and collectively cause over a million new infections a year in the United States alone. They tend to remain latent after infection and may reactivate if the subject becomes immune deficient (15). It has long been believed that phase transition from mold to yeast is obligatory for pathogenicity, but the mechanism that regulates this switch has remained a mystery. In this report, we provide firm genetic evidence that establishes the central role of dimorphism in pathogenicity, and we describe a regulator of this morphologic transition.

It is temperature that induces dimorphic fungi to change phases (6). At 25°C, they grow as mold. At 37°C, the core temperature of humans, they switch into the pathogenic yeast form (7), during which yeast phase–specific virulence genes are induced. Few of these genes have been identified; among the best studied are BAD1 of B. dermatitidis, CBP1 of H. capsulatum, and the 1,3-α-glucan synthase (AGS1) of these fungi and of P. brasiliensis (810). We postulated that deciphering the regulation of phase-specific genes would elucidate the control of morphogenesis.

Forward genetics, a process of inducing mutations randomly in a genome to detect phenotypes and linked genes, has advanced our understanding of microbial pathogenesis. Dimorphic fungi have not yet been manipulated in this way because the classical genetic approaches have proved too cumbersome, and the molecular tools have been unavailable. We previously showed that Agrobacterium tumefaciens transfers DNA randomly into the genomes of B. dermatitidis and H. capsulatum, primarily into single sites and without recombination, which, in theory, allows the identification of recessive mutations (11). Here, we used A. tumefaciens for insertional mutagenesis in a dimorphic fungus to attempt to uncover regulators of yeast phase–specific genes and phase transition from mold to yeast.

BAD1 of B. dermatitidis was used as “bait” in hunting for regulators of dimorphism, because it is expressed during the transition to yeast, regulated transcriptionally, and required for pathogenicity (12, 13). To identify mutants with regulatory defects, we created a B. dermatitidis reporter strain T53-19 harboring a transcriptional fusion between the BAD1 promoter and β-galactosidase reporter, PBAD1-LacZ. We transformed T53-19 with A. tumefaciens and monitored regulatory defects using a color screen (14). As yeast at 37°C, the reporter strain stains blue on media containing 5-bromo-4-chloro-3-indolyl-β-d-galactoside (X-gal). As mold at 22°C, it fails to stain and is white. Because BAD1 expression is up-regulated in yeast, we sought mutants that were white instead of blue at 37°C. In screening ∼15,000 transformants, we found seven with color defects. These transformants were confirmed by colony immunoblot and Northern analysis to have diminished BAD1 and reductions in transcript from <1% to 14% of the reporter strain (14).

One mutant, 4-2-2, had a reduction in BAD1 transcript to 14% of the reporter strain, corresponding reduction in BAD1 (Fig. 1A), and pleiotropic defects. At 37°C, 4-2-2 fails to convert to yeast and grows as pseudohyphae (Fig. 1B). Cell wall composition is altered in the mutant (Fig. 1, C and D). BAD1 and chitin are distributed in an aberrant striated pattern in the cell wall, and the amount of 1,3-α-glucan is greatly reduced. As evidenced in other cell-wall mutants (15), mutant 4-2-2 is more sensitive than the parent strain to the cell wall–binding chemicals calcofluor and Congo red (Fig. 1E). After sporulation, mutant 4-2-2 produces nearly 10% as many conidia (the infectious particles) as the parental strain (Fig. 1F), and spore progeny of the mutant retain a pseudohyphal phenotype when grown at 37°C (fig. S1A). Despite these pleiotropic defects, the mutant grows at the same rate as the parent strain (table S1) (14). Mutant 4-2-2 thus has global defects in morphogenesis, cell wall composition, sporulation, and expression of virulence factors BAD1 and 1,3-α-glucan.

Fig. 1.

An insertional mutant of B. dermatitidis with pleiotropic defects in morphogenesis, virulence gene expression, cell wall integrity, and sporulation. (A) (Top) Fungal colony overlay and immunoblot for BAD1. Nitrocellulose overlay of colonies probed with DD5-CB4 monoclonal antibody (mAb) to BAD1. Parental reporter strain T53-19 (WT) is a positive control and bad1Δ strain 55, a negative control. The patches of fungal cells tested are shown below the blot. (Bottom) Northern analysis. By densitometry, BAD1 transcript in the mutant is 14% of that in the wild type. GAPDH, loading control. (B) Microscopic appearance of mutant and wild-type parent strain grown at 37°C in liquid Histoplasma macrophage medium (HMM). The mutant phenotype is stable on serial passage. Scale bar, 10 μm. (C) Surface BAD1 (red), chitin (green), and merged images. (Top) Wild type; (bottom) mutant. BAD1 is stained with mAb DD5-CB4 and goat antibody to mouse phycoerythrin, and chitin with wheat germ agglutinin–fluorescein isothiocyanate. (D) Cell wall composition. Cell walls were obtained and analyzed as described (14). The mutant has half as much 1,3-α-glucan and three times as much 1,6-β-glucan as the wild-type strain. *P < 0.05, ANOVA test. Data are means ± SD of three experiments. (E) Sensitivity to the cell wall–binding agents calcofluor (CF) and Congo red (CR). Growth at 37°C of wild-type reporter strain and mutant was analyzed in liquid HMM alone or with CF or CR (20 μg/ml). Cells were counted to quantify growth rate. Data from one representative experiment of three are shown. Results were similar at 22°C (fig. S1B). (F) Sporulation of mold. (Left) Sporulating hyphae of the wild-type and mutant strains on potato flake agar. (Right) Total number of spores produced after 2 weeks' growth at 22°C (14). Data are representative of two experiments. Scale bar, 10 μm.

By Southern analysis, we determined that transferred DNA (T-DNA) had inserted into one site in the genome of mutant 4-2-2. By adapter polymerase chain reaction (PCR), 256 and 412 nucleotides flanking 5′ and 3′ of the T-DNA insertion, respectively, were amplified and sequenced. Flanking sequence showed identity to contig 52 in the Blastomyces genome (16). Two large putative open reading frames (ORFs) (ORFA and ORFB) were identified near the insertion site (Fig. 2A). The T-DNA transected ORFA, whose transcript is not detectable in mutant 4-2-2 by Northern analysis (Fig. 2B). To assess the functional role of these ORFs in mutant 4-2-2, we complemented the strain with a gene copy of ORFA or ORFB (14) (Fig. 2B). A plasmid containing an intact genomic copy of ORFA and flanking sequence reversed the phenotypic defects in the mutant (Fig. 2C), whereas that containing ORFB and flanking sequence did not.

Fig. 2.

Elucidation of the genotype of mutant 4-2-2. (A) A single T-DNA insertion is present in the mutant. Adapter PCR identified 256 bases on the left border and 412 bases on the right. Flanks have homology to contig 52 in the Blastomyces genome. Putative ORFs (A and B) are located near the insertion. T-DNA was inserted into the 3825-nucleotide ORFA coding sequence 522 nucleotides upstream of the stop codon. The insertion interrupts the first β sheet of the protein's receiver domain (see Fig. 2D). ORFB is 1.4 kb away from the T-DNA insertion. LB denotes the T-DNA left border, and RB the right border. (B) Northern analysis of ORFA transcript. A1-D is a transformant of mutant 4-2-2 complemented with an intact genomic copy of ORFA and flanking sequence. (C) ORFA complements the defects in 4-2-2. (Left) ORFA restores yeast morphology to mutant 4-2-2, whereas ORFB does not. (Middle) Fungal colony immunoblot for BAD1. In transformants reexpressing ORFA (A1-R, A1-K, A1-O), BAD-1 is detectable. Negative controls are bad1Δ strain 55 and a transformant of strain 4-2-2 that received a control vector lacking ORFA (“vector”). The patches of fungal cells tested are shown below the blot. (Right) Growth of mutant, wild type, and a complemented strain A1-D on HMM in the presence of calcofluor (CF) or Congo red (CR) (20 μg/ml). Top line shows growth of the strains on HMM alone. Scale bar, 10 μm. (D) ORFA has the domain structure and sequence of histidine kinase and is conserved in dimorphic fungi. ORFA has a histidine-containing H-box, an aspartate-containing D-box, and G- and N-boxes (32). Two putative transmembrane domains (TM) and an aspartate-containing receiver domain at the C terminus are also present. Sequences homologous to the S. cerevisiae (Sc) histidine kinase SLN1 and B. dermatitidis (Bd) histidine kinase are present in other dimorphic fungi H. capsulatum (Hc) and C. immitis (Ci). (E) Blastomyces ORFA complements an sln1 defect in S. cerevisiae. S. cerevisiae JF2007 [sln1::LEU2, ura3-52, trp1Δ63, his3Δ200, leu2Δ1, lys2, pRSPTP2 (URA3)] was transformed with a galactose-inducible vector containing a c-Myc–tagged SLN1 (pGalSln1-A1 and A2) or a FLAG-tagged ORFA (pESCTRP-422-A1 and A2). Both vectors contained TRP1 for selection. Transformants were initially plated on medium lacking uracil, to select for pRSPTP2, and lacking tryptophan, to select for the expression vector. Transformants were then plated on medium containing 5-FOA to select against pRSPTP2 and containing galactose for induction. pRSPTP2 rescues the lethal sln1 defect. Only transformants with a functional histidine kinase that complements the snl1 defect can grow on 5-FOA media under inducing conditions (14, 20). Transformants were plated on 5-FOA–containing medium with glucose as a control for gene induction. (F) Kinase activity detected by a luminescence assay. Decreasing relative light units (RLUs) indicate increasing kinase activity. Protein was immunoprecipitated from S. cerevisiae JF2007 transformed with c-Myc–tagged SLN1 expression vector (Sln1p), FLAG-tagged ORFA expression vector (orfAp), or untransformed JF2007 (JF2007), by using Myc-specific or FLAG-specific antibody (14). Bovine serum albumin (BSA) and reaction buffer (background) are negative controls. Data are the means ± SD of three experiments.

ORFA encodes a protein of 1274 residues (on the basis of transcript size) and predicted by gene-finding software (Softberry, Mount Kisco, NY). The gene has three exons totaling 3825 base pairs (bp) and two introns of 140 and 40 bp, and it displays homology to domains of histidine kinase by BLAST analysis and CD search. Histidine kinases are signal transduction proteins that organisms in all three domains of life use to respond to environmental signals (17) and control developmental processes (18, 19). ORFA is predicted to have two transmembrane domains and the necessary elements for histidine kinase function, including the histidine-containing H-box and aspartate-containing D-box involved in phosphorelay (Fig. 2D). The sequence also contains the N- and G-boxes used in ATP-binding and catalytic function, and an aspartate-containing receiver domain. The B. dermatitidis sequence is homologous to the hybrid histidine kinase SLN1 in Saccharomyces cerevisiae and to sequences in the genomes of H. capsulatum and C. immitis, dimorphic fungi for which extensive genome sequence is available (Fig. 2D).

We assayed the histidine kinase activity of ORFA using genetic and biochemical approaches. The ORFA of B. dermatitidis was expressed heterologously in S. cerevisiae to see if it functionally complements a sln1 defect in strain JF2007 (20). S. cerevisiae has a single hybrid histidine kinase, Sln1p, which regulates an osmosensing mitogen-activated protein kinase (MAPK) cascade, an oxidative stress-response pathway, and cell wall biosynthesis (21, 22). The lethal sln1 defect in JF2007 is viable because of the presence of a plasmid containing the phosphatase gene PTP2. Ptp2p dephosphorylates the Hog1 protein that accumulates in the absence of the functional histidine kinase (23). After lithium acetate transformation of JF2007 with an expression vector containing either ORFA or SLN1, we selected against maintenance of the PTP2 transgene by examining growth on 5-fluoroorotic acid (5-FOA). Transformants receiving either SLN1 or ORFA survived the loss of PTP2, which implies that ORFA functionally complements the sln1 defect (Fig. 2E and fig. S2). In biochemical studies, the B. dermatitidis ORFA protein product, immunoprecipitated from S. cerevisiae transformants, exhibited histidine kinase activity similar to that of Sln1p in a luminescence assay (Fig. 2F). ORFA thus encodes a protein that functions genetically and biochemically as a histidine kinase.

To test the role of B. dermatitidis histidine kinase unambiguously in the global defects observed in mutant 4-2-2, we created a targeted knockout by allelic replacement (14) (fig. S3A). The knockout is locked in the mold form at 37°C (Fig. 3A) and has all of the pleiotropic defects of mutant 4-2-2 [impaired BAD1 and 1,3-α-glucan expression, sensitivity to calcofluor and Congo red, and failure to sporulate] to a more extreme extent (fig. S3, B to D). Complementing the knockout corrected these defects (Fig. 3A and fig. S3). Henceforth, we refer to the gene here as DRK1 for dimorphism-regulating histidine kinase. We were unable to test virulence of DRK1 knockout strains in mice because the hyphae could not be reliably quantified and no spores were made. The more severe phenotype of the knockout compared with the insertion mutant 4-2-2 suggests that there is residual gene activity in the latter, perhaps due to the partial function of a truncated protein or to minimal DRK1 transcript beneath the level of detection.

Fig. 3.

The histidine kinase DRK1 regulates dimorphism from mold to yeast and virulence gene expression in B. dermatitidis and H. capsulatum. (A) (Left) Knockout strain (14081 drk1Δ) grown at 37°C is locked in the mold morphology. (Right) Complemented strain 14081 drk1Δ (pJNA1) regains the parental yeast phenotype at 37°C. Scale bar, 10 μm. (B) (Left) Gene silencing of DRK1 by RNAi in B. dermatitidis 60636 (drk1-RNAi) induces pseudohyphal morphology at 37°C. Scale bar, 10 μm. (Right) Northern analysis of virulence factors BAD1 and AGS1 and yeast phase-specific gene BYS1 in three independent DRK1-silenced transformants of B. dermatitidis parental strain 60636. GAPDH, loading control. (C) (Left panel) Northern analysis of two independent DRK1-silenced tranformants of H. capsulatum strain 186AR ura5, probing for the expression of DRK1 and virulence genes CBP1 and AGS1. (Middle) Ruthenium red stain of CBP in culture supernatant. Four independent DRK1-silenced transformants of H. capsulatum (1-5, 1-13, 1-15, and 1-6) show decreased staining compatible with reduced CBP. Parental strain 186AR ura5 is a positive control, and medium (-) a negative control. (Right) Surface 1,3-α-glucan in a DRK1-silenced strain of H. capsulatum (Hc186 drk1-RNAi) and parental 186AR ura5 (WT). Cells were stained with mAb MOPC104e and goat antibody to mouse fluorescein isothiocyanate (14). Light image is on the right. Scale bar, 10 μm. (D) Pigmentation of H. capsulatum wild-type (HcKD) and DRK1-silenced strain (HcKD drk1-RNAi). Mold on agar plates (left) and a suspension of the harvested spores (right).

We exploited RNA interference (RNAi) for gene silencing in B. dermatitidis to knock down DRK1 function and to circumvent the extreme phenotypes of the knockout (14). RNAi experiments were carried out in two different B. dermatitidis strains: 60636 and 14081. DRK1-silenced transformants from both strains exhibit rough colony morphology and pseudohyphal growth at 37°C, reduced sporulation, and sensitivity to calcofluor and Congo red (Fig. 3B; and fig. S4 and table S2). To explore the relation between B. dermatitidis DRK1 and expression of yeast-phase virulence genes, we analyzed transcript for DRK1, BAD1, and AGS1 by Northern analysis in DRK1-silenced strains. BAD1 and AGS1 transcripts are absent in parallel with that of DRK1, and the transcript for BYS1, a yeast-phase gene of unknown function, is inconsistently reduced in the strains (Fig. 3B).

The DRK1 sequence and its key domains are highly conserved in H. capsulatum and C. immitis (Fig. 2D). To test whether DRK1 acts as a global regulator of dimorphism and yeast-phase virulence gene expression in other dimorphic fungi, we used RNAi to silence gene expression in H. capsulatum. DRK1 was silenced in strain 186ura5AR and a clinical isolate, HcKD (14). Transformants were initially screened for those that grew as pseudohyphae at 37°C. DRK1-silenced strains showed concomitant reduction in the expression of virulence factors CBP and 1,3-α-glucan (9, 24), sensitivity to calcofluor, and reduced sporulation (Fig. 3C; and fig. S5 and table S2). Transcript levels for the silenced strains were correspondingly reduced for DRK1, CBP1, and AGS1 (Fig. 3C). Strikingly, the brown pigment indicative of melanin in mycelia and spores of wild-type Histoplasma strains was absent in the DRK1-silenced strains (Fig. 3D). Melanin is linked with virulence in other fungal pathogens, including C. neoformans and A. fumigatus (25). The histidine kinase DRK1 thus regulates global functions, including dimorphism and virulence gene expression in H. capsulatum.

Because mold-to-yeast transition and expression of the yeast-phase genes BAD1, CBP1, and AGS1 are required for pathogenicity, we postulated that silencing DRK1 expression would impair virulence of B. dermatitidis and H. capsulatum. We investigated virulence of DRK1-silenced strains in a mouse model of lethal pulmonary infection (14). After intratracheal infection with spores of B. dermatitidis, DRK1-silenced strains from two independent isolates were sharply attenuated compared with wild-type strains, as measured by survival and lung colony-forming units (CFUs) (Fig. 4A). In a murine model of histoplasmosis after intratracheal infection with spores, DRK1-silenced strains of H. capsulatum also were sharply reduced in virulence compared with wild-type strains (Fig. 4B). The growth rate of DRK1-silenced strains in all genetic backgrounds was similar to that of the respective parent strain (table S1). Silencing expression of the histidine kinase DRK1, therefore, reduced pathogenicity markedly in two dimorphic fungi.

Fig. 4.

Effect of knocking down DRK1 expression on the in vivo pathogenicity of B. dermatitidis and H. capsulatum in murine models of pulmonary infection. (A) Spores of DRK1-silenced transformants from Blastomyces strains 60636 and 14081 were used to infect C57BL6 mice. Mice (n = 10 per group) received 104 spores intratracheally. The wild-type strain, three independent DRK1-silenced transformants, and two control (CTRL) transformants that received an RNAi vector lacking the target sequence were studied. (Top and middle) Survival; (bottom) burden of lung infection (CFUs) in mice 14 days after infection. P < 0.001, for survival and lung CFU in the gene-silenced transformants versus wild-type and control strains. Results were similar when mice were infected with 100 times as many spores (106) of DRK1-silenced transformants (fig. S6). (B) Spores of DRK1-silenced transformants of Histoplasma clinical isolate KD were used to infect C57BL6 mice. Mice (n = 10 per group) received 108 spores intratracheally. The wild-type strain, three independent DRK1-silenced transformants, and two control (CTRL) transformants that received an RNAi vector lacking target sequence were studied. (Top) Survival; (bottom) lung infection 8 and 27 days after infection. P < 0.001, for survival and lung CFU in gene-silenced transformants versus wild-type and control strains.

We have described a highly conserved hybrid histidine kinase, DRK1, that is indispensable for dimorphism, virulence gene expression, and pathogenicity in dimorphic fungi. Our finding that DRK1 gene disruption locks a dimorphic fungus in the mold form uncovers a long-sought regulator of phase transition. The observation that phase-locked cells lose virulence extends the biochemical studies of Medoff et al. (7) and offers genetic proof that conversion of mold to yeast is required for pathogenicity in dimorphic fungi. A change in shape alone probably does not explain why the conversion is required, because mold and yeast differ in the expression of many genes and phenotypes, including some that are linked with virulence.

Two-component signaling systems are widespread in the prokaryotes. Eukaryotes have been thought to rely mainly on serine, threonine, and tyrosine kinases for signal transduction, but histidine kinase two-component systems have recently been shown to play a role in environmental sensing and cell development in eukaryotes (26), for example, in Candida albicans, where they regulate filamentation (18, 19). We show that a histidine kinase regulates sensing of environmental changes needed for mold-to-yeast transition in at least two dimorphic fungal pathogens. Histidine kinase homologs were identified in three dimorphic species for which the most complete genome sequence is available: B. dermatitidis, H. capsulatum, and C. immitis. The presence of this gene in multiple species and its conserved role in B. dermatitidis and H. capsulatum suggests that it may control phase transition and virulence gene expression, as well as cell wall development and sporulation, in the other systemic dimorphic fungi. DRK1 shares limited sequence similarity with histidine kinases that regulate filamentation in the more distantly related fungus C. albicans, although the functional domains are conserved. Nevertheless, the finding that histidine kinases regulate changes in shape for diverse fungal species points to a potentially broad role of these environmental sensors in the fungal kingdom.

What is the environmental signal that DRK1 of Blastomyces and Histoplasma senses to regulate phase transition and virulence gene expression? In S. cerevisiae, Sln1p detects osmotic stress, whereas in Schizosaccharomyces pombe, the histidine kinase–regulated SPC1 MAPK cascade senses osmotic stress as well as oxidative and heat stress and nutrient deprivation (27). Potential signals for histidine kinase sensing in dimorphic fungi include temperature, osmotic or oxidative stress, nutrient deprivation, redox potential, and host-derived factors such as hormones like 17-β-estradiol, which induces germ tubes in C. albicans (28) and blocks mold-to-yeast transition of P. brasiliensis (29).

In S. cerevisiae, the hybrid histidine kinase Sln1p transfers a phosphoryl group to the histidine residue of the phosphotransfer (HPt) domain in Ypd1p (30). Ypd1p transfers a phosphoryl group to one of two response regulators, Ssk1p or Skn7p, which control MAPK cascades and gene expression. Ypd1p, Ssk1p, and Skn7p homologs are present in both the Blastomyces and Histoplasma genomes; three other putative histidine kinases also are present (16). The four histidine kinases may sense different environmental signals that all lead through Ypd1p to the same output of morphogenesis and virulence gene expression. Alternatively, multiple downstream response regulators could respond to stimulation from Ypd1p, each controlling a distinct program involved in phase transition.

Histidine kinases linked with two-component relays have been identified in all three domains of life, but none have been established in any of the fully sequenced vertebrate genomes. The lack of such a homolog in humans suggests that these proteins may serve as antifungal drug targets. Previously identified bacterial histidine-kinase inhibitors have had general antifungal activity that is not kinase-specific; instead, it results in general membrane damage (31). Greater knowledge of eukaryotic histidine kinase function could assist in the development of better-targeted inhibitory compounds. Dimorphic fungi attenuated by knocking out histidine kinase might also be used for vaccination purposes.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Tables S1 and S2

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article