|Home | About | Journals | Submit | Contact Us | Français|
Paracoccidioides brasiliensis is a thermally dimorphic fungus, and causes the most prevalent systemic mycosis in Latin America. Infection is initiated by inhalation of conidia or mycelial fragments by the host, followed by further differentiation into the yeast form. Information regarding gene expression by either form has rarely been addressed with respect to multiple time points of growth in culture. Here, we report on the construction of a genomic DNA microarray, covering approximately 25% of the genome of the organism, and its utilization in identifying genes and gene expression patterns during growth in vitro. Cloned, amplified inserts from randomly sheared genomic DNA (gDNA) and known control genes were printed onto glass slides to generate a microarray of over 12000 elements. To examine gene expression, mRNA was extracted and amplified from mycelial or yeast cultures grown in semi-defined medium for 5, 8 and 14 days. Principal components analysis and hierarchical clustering indicated that yeast gene expression profiles differed greatly from those of mycelia, especially at earlier time points, and that mycelial gene expression changed less than gene expression in yeasts over time. Genes upregulated in yeasts were found to encode proteins shown to be involved in methionine/cysteine metabolism, respiratory and metabolic processes (of sugars, amino acids, proteins and lipids), transporters (small peptides, sugars, ions and toxins), regulatory proteins and transcription factors. Mycelial genes involved in processes such as cell division, protein catabolism, nucleotide biosynthesis and toxin and sugar transport showed differential expression. Sequenced clones were compared with Histoplasma capsulatum and Coccidioides posadasii genome sequences to assess potentially common pathways across species, such as sulfur and lipid metabolism, amino acid transporters, transcription factors and genes possibly related to virulence. We also analysed gene expression with time in culture and found that while transposable elements and components of respiratory pathways tended to increase in expression with time, genes encoding ribosomal structural proteins and protein catabolism tended to sharply decrease in expression over time, particularly in yeast. These findings expand our knowledge of the different morphological forms of P. brasiliensis during growth in culture.
Paracoccidioidomycosis is the most prevalent systemic mycosis in Latin America and is caused by the thermally dimorphic fungus Paracoccidioides brasiliensis (Mackinnon, 1970; Wanke & Londero, 1994). The infection originates with the inhalation of airborne mycelial particles or conidia. The infecting propagules subsequently transform into multiply budding yeasts in tissues. Conversion of the mycelial form (or conidia) to yeast is necessary for establishing infection (Borges-Walmsley et al., 2002; Brummer et al., 1993; Rooney & Klein, 2002). In addition to differences in morphology, yeast and mycelia also differ in nutritional requirements (Arraes et al., 2005; Paris et al., 1985), cell wall composition (more β-glucan in mycelia versus more α-glucan in yeasts, as well as higher amounts of chitin in yeasts) (Kanetsuna et al., 1969; San-Blas & Vernet, 1977; San Blas & San Blas, 1977), respiratory activity (Felipe et al., 2005; Kanetsuna & Carbonell, 1966) and membrane lipid composition (Levery et al., 1996, 1998, 2000; Manocha, 1980; Toledo et al., 1999).
Isolation and characterization of genes is a relatively recent trend in P. brasiliensis research. The first gene to be cloned and characterized was gp43, which encodes the primary diagnostic antigen of P. brasiliensis, a 43 kDa glycoprotein (Cisalpino et al., 1996). In addition, expression levels of gp43 vary among isolates, independently of virulence and morphology (Mattar-Filho et al., 1997). A number of genes, such as hsp70, hsp60, lon, mannosyltransferase, clpB, l35 and hydrophobin, have been identified and characterized through conventional methods, facilitating identification of potential virulence factors and morphology-associated differentially expressed genes (Albuquerque et al., 2004; Barros & Puccia, 2001; Costa et al., 2002; da Silva et al., 1999; Izacc et al., 2001; Jesuino et al., 2002, 2004).
Recently, two cDNA sequencing projects for P. brasiliensis were completed by the Brazilian scientific community, and the genomes of three isolates (Pb18, Pb01 and Pb03) are currently being sequenced by the Broad Institute (Cambridge, MA, USA). These cDNA projects have greatly increased the number of gene sequences in the public databases and, through genomic methodologies, have identified genes potentially implicated in the dimorphic switch (e.g. MEKK, MAPK, and Ras-like proteins), respiratory processes (i.e. aerobic in mycelia and anaerobic in yeast), heat-shock responses, drug resistance (e.g. ABC-type transporters), cell wall biosynthesis (e.g. α-glucan synthase), sulfur metabolism, virulence, pathogenicity (e.g. phospholipase, cell wall biosynthesis and adhesion) and as potential drug targets (e.g. chitin deacetylase and 4-hydroxyphenyl pyruvate dehydrogenase) (Felipe et al., 2003, 2005; Ferreira et al., 2006; Goldman et al., 2003; Marques et al., 2004; Nunes et al., 2005). Although these studies have added knowledge about the biology of P. brasiliensis, other approaches can also prove useful to shed light on the genetic composition of this organism. We describe here the construction of a random-shear genomic DNA microarray, and its application in the characterization of gene expression of the morphological forms of P. brasiliensis during growth in semi-synthetic culture media, and in the subsequent identification of differentially expressed genes. These data complement published studies and extend our understanding of the biology of mycelia and yeast forms of P. brasiliensis.
P. brasiliensis isolate Pb01 (ATCC MYA-826) was selected for study because it has been used in a number of molecular biological studies (da Silva et al., 1999; Felipe et al., 2003). Yeast cultures were maintained on modified McVeigh/Morton (MVM) agar slants (Restrepo & Jimenez, 1980) at 36 °C and were transferred every 10–15 days. Mycelial cultures were maintained on MVM slants at ambient temperature and were transferred monthly.
Spent culture filtrate was used as a growth supplement (2:1 ratio of fresh liquid MVM to spent culture filtrate), as previously described (Clemons et al., 1989; Stover et al., 1986) for the growth of mycelial and yeast liquid cultures prior to DNA and RNA extractions.
Genomic DNA (gDNA) was extracted from yeasts collected from 4-day-old liquid cultures (36 °C, shaking at 150 r.p.m.) using previously described methodology (McCullough et al., 2000) (see Supplementary Methods for detailed descriptions). The quantity and quality of the extracted DNA were assessed using a spectrophotometric absorbance ratio (A260:A280) and by electrophoresis through a 0.8% agarose gel stained with ethidium bromide and visualized by UV transillumination.
For total RNA extractions, one yeast or two mycelial 5-day-old slants were used to inoculate 150 ml supplemented liquid MVM medium, which was incubated on a gyratory shaker at 150 r.p.m. at 36 °C for yeast, or at ambient temperature for mycelia, for 5, 8 or 14 days. Liquid cultures were evaluated for contaminating organisms by microscopy and by plating samples on 5% sheep blood agar followed by incubation for 2–3 days at 36 °C. Total RNA was extracted using TRIzol reagent as per the manufacturer's instructions (see Supplementary Methods for details). The quantity and quality of the extracted RNA were assessed by A260:A280 ratio and electrophoresis through a 1.5% agarose gel stained with ethidium bromide (examining the presence of intact 18S and 28S rRNA bands) and visualized by UV transillumination at 302 nm. Contaminating gDNA was removed using RNase-free DNase (RQ1 RNase-free DNase, Promega) as per the manufacturer's instructions.
Genomic DNA was sheared hydrodynamically into ~1.5 kb fragments using a HydroShear apparatus (Genomic Solutions) as described elsewhere (Oefner et al., 1996; Thorstenson et al., 1998). Adequacy of DNA shearing was assessed by 1.5% agarose gel electrophoresis and visualized by UV after ethidium bromide staining. Sheared DNA was used as inserts for the genomic library.
The genomic library was assembled using the λ ZAP II phage as vector (λ ZAP II Predigested EcoRI/CIAP-treated Vector kit, Stratagene) and a modification of the linker-adaptor strategy described by Thorstenson et al. (1998) (also see: http://www-sequence.stanford.edu/protocols/library.html). A diagram of the library construction is presented in Supplementary Fig. S1.
Individual DNA inserts in the phage were amplified using PCR as follows: phage capsids were denatured at 70 °C for 5 min to release viral DNA. DNA inserts were amplified using Platinum PCR Supermix 96 (Invitrogen Life Technologies) with primers complementary to the vector (primer SK, 5′-cgctctagaactagtggatc-3′, and primer KS, 5′-tcgaggtcgacggtatc-3′) following the manufacturer's instructions (see Supplementary Methods for details).
A total of 34 genes from P. brasiliensis were selected for inclusion as known controls on the array. Genes were chosen based on known or expected expression patterns in mycelia or yeast or during form transition from mycelia to yeast. Template sequences were obtained from GenBank and primer pairs were designed for each gene using Primer 3 software, considering an amplicon size of ~1 kb (Rozen & Skaletsky, 2000). Nine primer pairs were designed from regulatory gene sequences provided by Dr Adrian R. Walmsley (University of Durham, UK) (see Supplementary Table S1 for genes and primers). The microarray contained a total of 40 control features based on the above-mentioned genes.
All amplified fragments were electrophoresed through 1.0% agarose gels for 12 min using an E-Gel 96 High-Throughput Agarose Electrophoresis System (Invitrogen) and photographed. Amplicon lengths were estimated using a 1 kb Plus DNA Ladder (Invitrogen) (Supplementary Fig. S1). Amplicons were purified using a Nucleofast 96 PCR Clean-up kit (BD Biosciences).
Purified PCR products from the cloned fragments and controls (12243 inserts from the library plus 40 features representing 34 known genes from P. brasiliensis for a total of 12283 elements) were spotted onto aminosilane-coated glass slides (UltraGAPS Slides, Corning) using custom-built robots at the Stanford Microarray Core Facility (http://www.microarray.org).
Cyanine 5 (Cy5)-labelled cDNA was synthesized from mycelial or yeast total RNA as per a protocol provided by the Stanford Microarray Core Facility, with modifications (see Supplementary Methods for full details).
Cyanine 3 (Cy3)- or Cy5-labelled P. brasiliensis gDNA was synthesized from 250 ng sheared gDNA using the BioPrime Array CGH Genomic Labelling System (Invitrogen) following the manufacturer's instructions. Cy3-labelled gDNA was used as a control to facilitate comparisons between slides and data integration in future experiments (Gadgil et al., 2005; Talaat et al., 2002; Williams et al., 2006). Cy5-labelled gDNA was used in self-to-self hybridizations along with Cy3-labelled gDNA during microarray validation procedures.
Samples of labelled nucleic acids for hybridization were concentrated using a Microcon YM-30 Centrifugal Filter Device (Millipore) and suspended to 40 μl with 3× saline-sodium citrate buffer (Sigma–Aldrich) containing 0.1% SDS (Sigma–Aldrich). These samples were incubated at 100 °C for 2 min, 37 °C for 15–25 min, transferred onto a microarray slide and covered with a coverslip and placed in a hybridization chamber (see Supplementary Methods for details). Hybridization was carried out at 50 °C for 14–18 h followed by washing and scanning within 24 h (GenePix 4000B Microarray Scanner, Axon Instruments). All hybridization experiments were done in duplicate from each of two independent cultures.
Cy3-labelled gDNA was used as the control for comparisons, as it provides a stable and practical reference to hybridize against labelled cDNA in competitive hybridizations (Gadgil et al., 2005; Talaat et al., 2002) and enables determination of an absolute value of gene expression. Furthermore, the use of a more stable reference, such as gDNA, as opposed to experiment-specific cDNA pools, provides a useful link to compare arrays across different experiments, adding to the usefulness of any given array dataset (Williams et al., 2006). Cy5-labelled cDNA was synthesized from mycelial or yeast total RNA obtained after days 5, 8 and 14 of culture.
Spots with more than 50% of their pixel values below the background median plus one standard deviation on the gDNA channel (i.e. Cy3), that were shown to contain more than one insert by agarose gel electrophoresis of amplified clones, or whose insert PCR failed, were eliminated from the analysis. There were 1846 spots with these criteria. Overall, this microarray contained 10437 elements, which we estimate covers approximately 25% of the P. brasiliensis genome considering an average insert size of about 1 kb and a genome size for P. brasiliensis of 30 Mb (Almeida et al., 2007; Cano et al., 1998; Montoya et al., 1999).
Gene expression data underwent global normalization using the R statistical package to perform quantile normalization of the global reference channel by computing the centroid of all global reference signatures, and performing a LOESS fit of each individual global reference signature to the centroid. The LOESS curve for a given sample was then used to scale both channels for each gene in that sample. Data were further normalized using GeneSpring GX 7.3.1 (Agilent Technologies) as follows: per spot to the gDNA channel, and per gene to the respective median of each gene (Talaat et al., 2002). For mycelial versus yeast gene expression, per-slide normalization consisted of selecting a list of spots where gDNA intensities showed a linear relationship with cDNA intensities in M versus A plots [M=log2(Cy5/Cy3) versus A=log2(Cy5×Cy3)½] across all time points in both mycelia and yeast, as well as being below a twofold change in gene expression values. A total of 102 spots satisfied these criteria and were used as positive controls for the per-slide normalization process.
Principal components analysis (PCA) using the conditions of time and morphology was used to find primary tendencies in the dataset (Alter et al., 2000). A twofold cutoff for mycelial versus yeast differences was applied to all time points to focus on broader changes in expression levels. Array features were identified as representing differentially expressed genes after parametric two-way ANOVA analysis on log-transformed values (P<0.05), considering time and morphology as the parameters analysed and using Benjamini and Hochberg's method for multiple-test control of false discovery rate (Manly et al., 2004). Clones of interest were sequenced and analysed as described in Supplementary Methods. Sequenced genes were further analysed by hierarchical clustering using Pearson's correlation as the similarity measure and the unweighted pair group method with arithmetic mean (UPGMA) as the linkage method (Eisen et al., 1998).
For interspecies microarray data comparison, protein sequences were obtained from GenBank corresponding to the accession numbers listed by Hwang et al. (2003) and Johannesson et al. (2006). Each protein sequence was compared against our sequenced microarray probes using blastx. The top-scoring blastx hit was considered a match if its E-value was less than 1e−5.
Microarray results were verified using real-time RT-PCR (Brilliant QRT-PCR Core Reagent kit and MX3000P QPCR System, Stratagene). Eight genes of known expression pattern were examined (see Supplementary Methods for details) and results were extracted using MX3000P v1.20 software (Stratagene). These results were analysed by two-way ANOVA with Bonferroni correction (GraphPad Prism v3.0, GraphPad Software), as well as for Spearman and Pearson's correlation coefficients in GeneSpring GX 7.3.1 (Agilent). GeneSpring data transformations included normalizing each gene to its corresponding median across all samples to centre the data around zero and used l35 (for comparisons with microarray data) or α-tubulin (for comparisons between l35- and α-tubulin-normalized real-time RT-PCR datasets) as reference genes. No differences were observed using either gene as reference. All real-time RT-PCRs were done in triplicate and used the same RNA samples that were used in microarray hybridizations.
During the course of this study a total of 19800 clones of sheared gDNA were isolated and stored. The insert sizes ranged from 500 to 3000 bp (Supplementary Fig. S1b). A total of 481 clones were selected for sequencing and Fig. 1 presents an overview of the gene classifications after sequence analysis. The predicted functions contained a variety of categories, such as transposable elements, transporters, respiratory processes, metabolism (protein, amino acid, sugar and lipid), cell division, morphogenesis, RNA processing and transcription. DNA sequences with a match were generally closely related to genes of fungal origin, and most matched with P. brasiliensis or Ajellomyces capsulatus expressed sequence tags (ESTs). These results demonstrated that the library construction methodology worked as expected.
A gDNA microarray was assembled containing 12243 inserts from the library plus 40 features representing known genes from P. brasiliensis. Supplementary Fig. S2 presents a scatter plot of log2-transformed intensities of hybridized Cy3-labelled gDNA versus Cy5-labelled gDNA. Test results routinely showed ~99% of spots with fluorescence levels above the background and similar fluorescence intensities for both channels in each spot, validating that the microarray was functioning. Single-probe hybridization results showed that when a single cloned insert was labelled and hybridized to the array, the highest fluorescence reading was found at its corresponding spot, indicating that the microarray elements were correctly spotted at their assigned positions (results not shown). Quality control filters described above (see Methods, Microarray data analysis) were used to eliminate 1846 spots from the analysis process.
The expression of eight genes was validated using real-time RT-PCR. These were hydrophobin, y20, mannosyltransferase, hsp70, α-tubulin, l35, pfr1 and a toxin efflux pump [major facilitator superfamily (MFS) pump] gene. Fig. 2 presents the estimated quantities of mRNA transcripts for each of these genes as determined by real-time RT-PCR. All selected genes showed significant differences in expression between mycelia and yeast. l35, pfr1, α-tubulin and hydrophobin had significant differences over time and all, except the MFS pump, showed significant changes with time and morphological form. Statistical analyses showed that all, except the MFS pump, had similar expression levels in both forms for at least one time point. l35 and α-tubulin, genes previously used as controls in P. brasiliensis gene expression studies (Goldman et al., 2003; Jesuino et al., 2002, 2004), were shown to have similar expression levels only on day 5. Correlation coefficients of real-time RT-PCR and microarray data using l35 as the normalizer were all positive and in general above 0.7 (Table 1), indicating that the gDNA microarray platform is useful for gene expression experiments.
PCA of expression using log-transformed data was performed to determine general tendencies present in our dataset. The first two principal components captured 81.75% of the variance present in the data. The PCA plot shown in Fig. 3 indicates that mycelial gene expression was generally different from that in yeast, and that mycelial expression was generally unchanged over the selected time points. In contrast, the day 8 and day 14 time points of gene expression by yeast were different from the day 5 time point (Fig. 3).
A more detailed view of general gene expression is presented in Fig. 4, where hierarchical clustering of all genes present on the array showed that yeast had more upregulated genes than mycelia (cluster 1). Clusters 2, 3 and 4 illustrate that yeast tends to vary expression more with time than do mycelia. Upregulated genes in mycelia tended to concentrate in clusters 3 and 4. The control genes of known expression present in these clusters were used to determine their general nature. Thus, clusters 1 and 2 contained controls known to have higher expression in yeast (hsp104/clpa, mannosyltransferase, kex2 and lon1), while clusters 3 and 4 contained known mycelial controls (m51, hydrophobin, 40S ribosomal protein and gp43). Clones were chosen for sequencing based on ANOVA analysis for significance, using time and morphology as parameters and their presence in each of these clusters. We focused on genes that were overexpressed at all time points in either mycelia or yeast (morphology related genes) and genes that showed variations with time (metabolism- and stationary-phase-related genes).
Fig. 5 shows the clustering results of 521 sequenced features present in the array and illustrates the gene expression patterns that were investigated in this study, while a more detailed description of some differentially expressed genes is given in Table 2. Mycelia and yeast are known to differ in cell wall structure and composition (Kanetsuna et al., 1969; San-Blas & Vernet, 1977; San Blas & San Blas, 1977), nutritional requirements (yeast requires sulfur-containing amino acids) (Arraes et al., 2005; Paris et al., 1985), membrane lipid composition (Levery et al., 1996, 1998, 2000; Manocha, 1980; Toledo et al., 1999) and respiratory activity (Felipe et al., 2005; Kanetsuna & Carbonell, 1966).
Analysing the yeast cluster in light of these functional categories we found a cysteine desulfurase, an aminopeptidase (homocysteine catabolism) and two genes (sah1 and a SAM-dependent methyltransferase) involved in metabolism of sulfur-containing amino acids. Cysteine desulfurase, aminopeptidase and sah1 were upregulated in yeast only on day 5, whereas SAM-dependent methyltransferase was upregulated in yeast at all time points. Interestingly we also found an N-terminal peptidyl-methionine acetyltransferase (nat2), required for survival under heat stress in Saccharomyces cerevisiae, as well as two clones containing the dipeptide transporter ptr2. We also found that a homologue of met17, which catalyses the addition of inorganic sulfur into O-acetyl-l-homoserine generating homocysteine, was upregulated in mycelia only on day 5, with yeast catching up at later time points. The expression of inorganic sulfur assimilation genes in yeast has been previously reported, but it is still unknown whether the pathway is functional (Ferreira et al., 2006). Similarly, overexpression of sulfur metabolism genes and amino acid transporters has been detected in the pathogenic form of Histoplasma capsulatum (Hwang et al., 2003).
Expression changes in genes related to respiratory processes, such as NAD-dependent isocitrate dehydrogenase, which has been reported to be more highly expressed in yeast (Kanetsuna & Carbonell, 1966), were also detected. Our results are in agreement, as this gene showed constant overexpression in yeast in comparison with mycelia. Contrasting data have shown this gene to be upregulated in mycelia (Felipe et al., 2005), perhaps due to differences in experimental design, as reported elsewhere (Marques et al., 2004). Alternatively, the different results may be attributable to variations between fungal isolates, as observed in Coccidioides posadasii (Johannesson et al., 2006). In addition, we found that a cytochrome c1 haem protein precursor, which is related to aerobic respiratory processes, was constantly overexpressed in yeast. Our data also indicated the overexpression of two subunits of NADH:ubiquinone oxidoreductase, involved in electron transport for oxidative phosphorylation by yeast. Although components of complexes I and III were found to be upregulated in yeast, complex IV components (cytochrome c oxidase subunits 1, 2 and 5a) were upregulated in mycelia at earlier time points or had equivalent levels. Malate synthase and malate dehydrogenase homologues belonging to the glyoxylate cycle were also detected. The first was constantly upregulated in yeast and the latter upregulated in yeast on day 5 only, which is in agreement with previously published data. These findings further support the importance to P. brasiliensis of the glyoxylate cycle, which allows the utilization of two-carbon compounds for the production of glucose through gluconeogenesis during systemic infection and within phagocytic cells, thus improving pathogen survival chances against host defences in the nutrient-poor environment of the phagolysosome (Derengowski et al., 2008; Felipe et al., 2005; Ferreira et al., 2006; Nunes et al., 2005; Tavares et al., 2007). The importance of the glyoxylate cycle has been established in plant and animal pathogens such as Candida albicans, Penicillium marneffei, Trichophyton rubrum and Magnaporthe grisea. For fungi such as Candida albicans, the glyoxylate cycle is thought to be most important during the early stages of infection after internalization by macrophages. However, studies with Aspergillus fumigatus have indicated that the glyoxylate cycle is not necessary for virulence (Ebel et al., 2006; Schobel et al., 2007). As the glyoxylate cycle is not present in humans it is considered an attractive target for drug development (Barelle et al., 2006; Fradin et al., 2003; Lorenz & Fink, 2001, 2002; Lorenz et al., 2004; Thirach et al., 2008; Wang et al., 2003; Zaugg et al., 2009).
Among the genes involved in lipid metabolism we identified a transcription factor homologue hac1, upregulated at all times in yeast. In S. cerevisiae this gene is involved in the unfolded protein response in the endoplasmic reticulum (ER), as well as in phospholipid biosynthesis. The unfolded protein response can be caused by stresses such as lipid deprivation, heat, drug treatment, and mutations in or overexpression of wild-type secretory proteins, and involves increased production of proteins that increase lipid biosynthesis or restore homeostasis in the ER, as needed (Kaufman, 1999). The unfolded protein response has also been associated with virulence and susceptibility to antifungals in A. fumigatus (Richie et al., 2009). Upregulation of the gene in yeast could be a response to growth at higher temperature or could be related to the distinct lipid composition of yeast cells (Manocha, 1980). We also found a homologue of A. fumigatus fab1 that encodes 1-phosphatidylinositol-3-phosphate 5-kinase, responsible for the generation of phosphatidylinositol 3,5-bisphosphate. In S. cerevisiae, fab1 is involved in vacuolar sorting and homeostasis (Cooke et al., 1998). Also of interest is the requirement for fab1 for growth at elevated temperatures in S. cerevisiae (Yamamoto et al., 1995), which may explain its constitutively high expression in the yeast form. We also found a homologue to S. cerevisiae erg28, whose protein product is reported to be the anchor to the ER of the erg25, erg26 and erg27 protein products and is also involved in interaction with the protein of erg6; all are involved in ergosterol biosynthesis (Mo et al., 2004). These findings, along with the previously reported upregulation of erg25 in the yeast form of P. brasiliensis, further support the hypothesis that membrane lipid reorganization in yeast is a response to a higher-temperature environment or oxidative stress (Marques et al., 2004). Interestingly, we found a glycerol-3-phosphate dehydrogenase. This gene is overexpressed in H. capsulatum yeasts (Table 3), with an important role in glycerol metabolism during growth on non-fermentable carbon sources and in nutrient-poor environments, as may occur during infection of the mammalian host (Hwang et al., 2003; Sprague & Cronan, 1977).
Other genes differentially regulated in yeast included those involved in mRNA transcription, processing and degradation, as our results indicated that mRNA populations are very different between mycelia and yeast. The isolation of differentially regulated transcription factors from yeast (YL1, serine/threonine protein kinases and cAMP-dependent protein kinase, among others) provides a number of candidates for future experiments on genes encoding proteins that regulate morphogenesis of the pathogenic form of this fungus. Serine/threonine kinases, for example, have been found to be upregulated in the pathogenic form of Coccidioides posadasii (Johannesson et al., 2006). We also found genes directly associated with transcription from RNA polymerase II and III promoters (rpc4, srb5 and tfc3), mRNA export from the nucleus (sac3), splicing (prp39) and degradation (rrp4), which may have roles in yeast having a larger number of genes differentially expressed than did mycelia.
Putative virulence factors, such as cbp1, were overexpressed in yeast at all time points. In H. capsulatum, this gene encodes a calcium-binding protein identified as overexpressed in yeasts and considered a virulence factor. It would be interesting to study whether P. brasiliensis cbp1 aids survival within macrophages in the same way as the equivalent protein in H. capsulatum (Hwang et al., 2003; Patel et al., 1998; Sebghati et al., 2000). Other interesting virulence candidates were the nir gene, encoding a nitrate reductase also found in the parasitic form of Coccidioides posadasii, and urease (a nickel-dependent enzyme) (Table 3) (Johannesson et al., 2006; Rappleye & Goldman, 2006). Urease activity has been detected before in P. brasiliensis, giving further support to the suggestion that such a gene is functional and important in this species (Bagagli et al., 1998).
Comparisons with related fungal species also pointed towards common expression by pathogenic forms of a number of transporters, such as drug-resistance proteins and copper, nickel and calcium transporters, all of which may have essential roles in pathogen survival during infection (Table 3).
The mycelial form of P. brasiliensis had few differentially overexpressed genes. The mycelial genes included trichothecene C-15 hydroxylase and a putative gene encoding a trichothecene efflux pump; the MFS pump was also shown to be upregulated in mycelia by real-time RT-PCR. Whether P. brasiliensis produces trichothecene, a mycotoxin, remains to be determined. However, these genes have been identified in the ongoing P. brasiliensis genomic sequencing project of the Broad Institute (entries PADG_03582.1, PABG_01811.1, PAAG_01242.1 and PABG_05447.1). We also detected a hypothetical protein containing RNA recognition motifs and a putative splicing factor, which may influence the overall mycelial gene expression program. Other genes expressed by mycelia included a sugar transporter, phosphoribosylformylglycinamidine synthase (ade6), involved in de novo purine nucleotide biosynthesis, and a pseudouridine synthase. Comparisons with related fungal pathogens showed that the MFS pump and an adhesin/tyrosinase were also overexpressed in the saprobic phase of H. capsulatum, while ape3 (involved in protein catabolic processes) was detected in the mycelial phase of Coccidioides posadasii (Table 3) (Hwang et al., 2003; Johannesson et al., 2006).
Interspecies comparison also yielded genes that differed in their expression patterns across species, including nitrogen assimilation transcription factor, succinyl-CoA:3-ketoacid-coenzyme A transferase 1 and ccg9, among others. These differences may reflect changes caused by experimental conditions or reflect specific metabolic processes of each fungal species (Table 3).
Genes that changed expression with time were also selected for analysis. The first group comprised genes that were upregulated with time in both forms and were also more highly expressed in yeast (i.e. the transposon cluster). These were the two above-mentioned subunits of NADH:ubiquinone oxidoreductase, 3-isopropyl malate dehydrogenase A (leucine biosynthesis), glutamate dehydrogenase (amino acid transport and metabolism) and rxt2. The most notable feature of this cluster was the presence of the majority of clones (29 out of 35) that contain transposable element sequences. Similarly, transposable elements have been detected as overexpressed in H. capsulatum yeasts (Hwang et al., 2003), although their role in morphogenesis and growth in culture remains unclear. Recently, it has been established that S. cerevisiae Ty long terminal repeat (LTR) retrotransposon transcription is activated by adenine starvation, among other stresses, and in some cases can also affect transcription of neighbouring genes (Servant et al., 2008).
The second cluster with time-related variations was the ribosomal structure cluster, containing most of the ribosomal proteins detected in this study (seven genes plus one control). This cluster showed a marked decrease in gene expression with time, especially in yeast. Components of the proteasome, pre1 and pup1, and a ubiquitin ligase, ubr1, behaved in the same manner, which could be a reaction to medium depletion or reaching stationary phase, especially in the yeast form.
Finally, among the 481 sequenced clones, the presence of phase-specific hypothetical and novel genes with no matches in analysed databases was assessed, and the results are shown in Fig. 1. Phase-specific hypothetical genes represent 13% of the identified genes (11% in yeast and 2% in mycelia), and may have roles in morphogenesis that are also important in other dimorphic pathogenic fungi, such as H. capsulatum and Coccidioides spp. Genes with no match in analysed databases may represent species-specific biochemical processes, with 6% present in yeast and 1% present in mycelia. These sets of genes are possible candidates for further studies on P. brasiliensis morphogenesis.
The primary goal of these studies was that of building a randomly sheared genomic DNA library for use in a microarray-based approach to assess gene expression patterns in the pathogenic fungus P. brasiliensis. We examined gene expression of both mycelia and yeast during growth in semi-synthetic media over a period of 14 days to identify differentially expressed genes. Our results demonstrate that the methodology used for library and microarray construction is useful and applicable to studies of P. brasiliensis. Furthermore, we found previously undescribed genes that were differentially expressed by the two forms and gained insights regarding the behaviour of this fungus during growth in culture. These results are similar to those for other genomic shotgun approaches for microarray construction, which have been used successfully in studies involving pathogens such as Plasmodium falciparum and H. capsulatum (Hayward et al., 2000; Hwang et al., 2003).
The utilization of gDNA for library construction reduces the cloning bias seen in cDNA libraries. The adaptor–linker strategy minimizes the generation of chimeric inserts and is useful for the cloning of blunt-end DNA fragments. However, such gDNA libraries, and the microarrays assembled from them, also contain non-coding sequences. These may be uninformative in gene expression studies, but may nevertheless be useful in studies such as comparative genomic hybridization. Another drawback is that two genes may be present in a single insert if they are positioned closely in the genome (~6.4% of sequenced clones in our study contained two genes), which makes it difficult to distinguish expression of the adjacent genes.
Global gene expression and gene-specific analysis showed that mycelia and yeast forms had distinct behaviours during growth in culture. Yeast showed a larger number of upregulated genes than mycelia, and also showed more rapid and distinct changes in gene expression with time. Although not previously reported for P. brasiliensis, these differences might be predictable, since mycelia and yeast exist in thermally distinct environments and differ in cell wall structure and composition, nutritional requirements, membrane lipid composition and respiratory activity. Form-specific genes related to these differences, as well as genes related to growth in culture, were detected, and require further experimentation for full characterization. For the most part, our results correlate with previous genome-wide studies in P. brasiliensis and add further knowledge about the biology of this fungus. The decreased number of differentially expressed genes by yeast on day 14 is likely associated with the cultures attaining stationary phase due to exhaustion of essential nutrients (Arraes et al., 2005; Ferreira et al., 2006; Paris et al., 1985), and indicate that differentially expressed genes are more readily observed at earlier phases of growth.
In conclusion, we assembled a random-shear shotgun gDNA microarray that is useful for studies of gene expression in P. brasiliensis. Our validation of the methods used, and our findings of differentially expressed genes involved in various functional categories, demonstrate the utility of the array. We found that yeast has a larger number of upregulated genes than do mycelia, particularly early in growth at day 5. Evidence presented here suggests that the thermal dimorphism of P. brasiliensis and growth in culture, in the mycelial or yeast form, are complex phenomena that involve a variety of biological processes. The identified genes include potential cell surface proteins, genes belonging to pathways that do not occur in humans, and transcription regulators that may be used as targets for the development of future therapies. Comparisons among different fungal species demonstrated a number of common metabolic processes across species in the morphological form found in tissues, as well as in the saprobic mycelial form, contributing to the growing set of general characteristics of these closely related dimorphic fungi. Further experimentation and the application of gene disruption tools are essential to assess the exact functions and behaviours of the reported genes and how they relate to dimorphism.
This work was supported in part by a Fogarty Training Grant (NIH/ERID) 5D43TW923-5. We thank Drs Stanley Falkow, Gary K. Schoolnik, Rafael A. Irizarry, Theodore White and Elena Seraia for valuable discussions and insights, as well as Gaurav Singh for technical assistance.
The GenBank accession numbers (dbGSS id) for the DNA sequences reported in this paper are FI778766–FI779357.
The microarray data discussed in this publication have been deposited in the NCBI Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO series accession number GSE15511 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE15511).
Supplementary methods, two supplementary figures and a supplementary table are available with the online version of this paper.