Transcriptome analysis of Aspergillus fumigatus exposed to voriconazole

RESEARCH ARTICLE

Marcia Eliana da Silva Ferreira Æ Iran Malavazi

Marcela Savoldi Æ Axel A. BrakhageMaria Helena S. Goldman Æ H. Stanley Kim

William C. Nierman Æ Gustavo H. Goldman

Transcriptome analysis of Aspergillus fumigatus exposed to voriconazole

Received: 7 February 2006 / Revised: 6 March 2006 / Accepted: 9 March 2006� Springer-Verlag 2006

Abstract For a comprehensive evaluation of genes thathave their expression modulated during exposure of themycelia to voriconazole, we performed a large-scaleanalysis of gene expression in Aspergillus fumigatususing a microarray hybridization approach. By com-paring the expression of genes between the referencetime and after addition of voriconazole (30, 60, 120, and240 min), we identified 2,271 genes differentially ex-pressed in the wild-type strain. To validate the expres-sion of some of these genes during exposure tovoriconazole, we analyzed 13 genes showing higherexpression in the presence of voriconazole by real-time

RT-PCR. Although the magnitudes of induction dif-fered between the two experimental systems, in about85% of the cases they were in good agreement with themicroarray data. To our knowledge this is the first studyof microarray hybridization analysis for a filamentousfungus exposed to an antifungal agent. In our study, wehave observed: (i) a decreased mRNA expression ofvarious ergosterol biosynthesis genes; (ii) increasedmRNA levels of genes involved in a variety of cellfunctions, such as transporters, transcription factors,proteins involved in cell metabolism, and hypotheticalproteins; and (iii) the involvement of the cyclic AMP-protein kinase signaling pathway in the increasedmRNA expression of several of these genes.

Keywords Aspergillus fumigatus Æ Voriconazole ÆMicroarrays Æ Transcriptome

Introduction

Aspergillus fumigatus is the most common speciesof Aspergillus that cause life-threatening pulmonary dis-ease in severely immunocompromised patients (Denning1996). The treatment of these patients has been largelylimited to therapy with the polyene drug amphotericin B,the broad-spectrum triazoles such as itraconazole or vo-riconazole, and/or with the echinocandin caspofungin(Herbrecht et al. 2002; National Committee for ClinicalLaboratory Standards 2002). Amphotericin B therapycan be highly toxic and can result in nephrotoxicity,whereas triazoles are fungistatic and their use is oftenlimited by drug resistance (Denning et al. 1997). In spite ofits safety profile and good therapeutical performance, thecontinuous use of triazoles can result in the developmentof drug resistance and a number of itraconazole-resistantclinical isolates (Denning et al. 1997). Spontaneous andinduced mutants of A. fumigatus have been documented(Dannaoui et al. 2001;Manavathu et al. 1999;Mann et al.2003). Azoles block the ergosterol biosynthesis pathwayby inhibiting the enzyme 14-a-demethylase, product of

Electronic Supplementary Material Supplementary material isavailable for this article at http://dx.doi.org/10.1007/s00294-006-0073-2 and is accessible for authorized users.

Communicated by G. Braus

M. E. da S. Ferreira Æ I. Malavazi Æ M. SavoldiG. H. Goldman (&)Departamento de Ciencias Farmaceuticas,Faculdade de Ciencias Farmaceuticas de Ribeirao Preto,Universidade de Sao Paulo, Av. do Cafe S/N ,CEP 14040-903 Ribeirao Preto, Sao Paulo, BrazilE-mail: ggoldman@usp.brTel.: +55-16-6024280Fax: +55-16-6331092

A. A. BrakhageDepartment of Molecular and Applied Microbiology,Leibniz Institute for Natural Product Research,Infection Biology-HansKnoell Institute (HKI),University of Jena, Jena, Germany

M. H. S. GoldmanFaculdade de Filosofia, Ciencias e Letras de Ribeirao Preto,Universidade de Sao Paulo, Ribeirao Preto, Sao Paulo, Brazil

H. S. Kim Æ W. C. NiermanThe Institute for Genomic Research, RockvilleMD 20850, USA

H. S. Kim Æ W. C. NiermanDepartment of Biochemistry and Molecular Biology,The George Washington University School of Medicine,Washington, DC 20037, USA

Curr Genet (2006)DOI 10.1007/s00294-006-0073-2

the CYP51 gene (Diaz-Guerra et al. 2003). Fungal azoleresistance involves both amino acid changes in the targetsite that alter drug–target interactions and those thatdecrease net azole accumulation (Marichal et al. 1999;Sanglard and Odds 2002; White et al. 1998). Compensa-tory pathways have been documented for themechanismsof resistance to the azole and polyene classes and involvealterations of specific steps in ergosterol biosynthesis(Sanglard et al. 2003a, b; Lupetti et al. 2002). For exam-ple, analysis of the sterol compositions of two separateazole-resistant Candida albicans clinical isolates revealedthe accumulation of ergosta-7,22-dienol, which is con-sistent with the absence of sterol D5,6-desaturase activity,encoded by ERG3 (Kelly et al. 1996, 1997; Nolte et al.1997; Sanglard et al. 2003a, b). The reduced intracellularaccumulation has also been correlated with overexpres-sion of multidrug resistance (MDR) efflux transportergenes of the ATP-binding cassette (ABC) and the majorfacilitator superfamily (MFS) classes (Lupetti et al. 2002).

Mycelia can adjust to changes in the presence ofantifungal drugs by altering gene expression patterns.For a comprehensive evaluation of genes that have theirexpression modulated during exposure of the mycelia tovoriconazole, we performed a large-scale analysis ofgene expression in A. fumigatus using a microarrayhybridization approach. We showed that the mRNAexpression of several of these genes is dependent on theevolutionarily conserved cyclic AMP-protein kinase(cAMP-PKA) signaling pathway. The main objective ofthis study is to initiate research aiming to understandhow A. fumigatus becomes resistant to voriconazole.

Materials and methods

Strains and media methods

Aspergillus fumigatus strains used are CEA17 (pyrG),ATCC 46645 (wild type), DpkaC1 (DpkaC1::hph), andDacyA (DacyA::hph). Media were of two basic types(Kafer 1977). A complete medium with three variants:YAG (2% glucose, 0.5% yeast extract, 2% agar, traceelements), YUU (YAG supplemented with 1.2 g/l eachof uracil and uridine), and liquid YG or YG + UUmedium of the same compositions (but without agar).Vfend, a lyophilized powder containing 200 mg of VRCand 3,200 mg of sulfobutyl ether b-cyclodextrin sodium,was from Pfizer-Roerig (New York, NY, USA). Vfendwas reconstituted with 19 ml of sterile water, as rec-ommended for intravenous use, to give a clear solution,and was stored at �80�C.

RNA isolation

Approximately 1.0 · 107 conidia/ml of A. fumigatusstrains were used to inoculate 50 ml of liquid culturesthat were incubated in a reciprocal shaker at 37�C/150 rpm for 16 h. Mycelia were aseptically transferred to

fresh YG medium in the presence or absence of 0.5 lg ofvoriconazole for 30, 60, 120, and 240 min. Mycelia wereharvested by filtration through Number 1 Whatman fil-ter, washed thoroughly with sterile water, quickly frozenin liquid nitrogen, disrupted by grinding, and total RNAwas extracted with Trizol (Life Technologies, USA). Tenmicrograms of RNA from each treatment were thenfractionated in 2.2 M formaldehyde, 1.2% agarose gel,stained with ethidium bromide, and then visualized withUV-light. The presence of intact 28S and 18S ribosomalRNA bands was used as a criterion to assess the integrityof the RNA. RNAse-free DNAse treatment was done aspreviously described by Semighini et al. (2002).

Real-time PCR reactions

All the PCR and RT-PCR reactions were performedusing an ABI Prism 7700 Sequence Detection System(Perkin-Elmer Applied Biosystem,USA). Taq-Man� EZRT-PCR kits (Applied Biosystems,USA) were used forRT-PCR reactions. The thermal cycling conditionscomprised an initial step at 50�C for 2 min, followed by30 min at 60�C for reverse transcription, 95� for 5 min,and 40 cycles at 94�C for 20 s and 60�C for 1 min. Taq-Man Universal PCR Master Mix kit was used for PCRreactions. The thermal cycling conditions comprised aninitial step at 50�C for 2 min, followed by 10 min at95�C and 40 cycles at 95�C for 15 s and 60�C for 1 min.The reactions and calculations were performed accord-ing to Semighini et al. (2002). Table 1 in the Supple-mental data describes the primers and Lux� fluorescentprobes (Invitrogen) used in this work.

Gene expression methods

The DNA amplicon microarray for A. fumigatus Af293was fabricated as follows. For PCR targets, we selected a700-bp region immediately upstream of the predicted stopcodon from each gene. If the gene was smaller than700 bp, we took the entire gene. Then, we included 150 bpof sequence downstream of the gene or as much as there isin the intergenic region when shorter than 150 bp. Thesetarget sequences provided a maximum of 850 bp for eachgene. We conducted automated selection of PCR primerpairs by feeding the target sequences to Primer 3.0 (http://www-genome.wi.mit.edu/genome_software) with opti-mized design parameters that can be used to amplifygreater than 5/6 of the targets. The predicted resultingPCR products were on average 710 bp in length. Usingthis approach we were able to design primers for 9,516genes (96% of the predicted number of open readingframes present in theA. fumigatus genome).We amplifiedthese target gene regions from genomic DNA. Theresulting PCR products were purified and spotted intriplicate at high density on Corning (Acton, MA, USA)UltraGAPS� aminosilane-coated microscope slidesusing a robotic spotter built by Intelligent Automatic

Systems (Cambridge, MA, USA) and cross-linked byultraviolet illumination.

Hybridized slides were scanned using the AxonGenePix 4000B microarray scanner and the TIFF imagesgenerated were analyzed using TIGR Spotfinder(<http://www.tigr.org/software/>) to obtain relativetranscript levels. Data from TIGR Spotfinder were storedinMADAM, a relational database designed to effectivelycapture and store microarray data. Data was normalizedusing a local regression technique LOWESS (LOcallyWEighted Scatterplot Smoothing) for hybridizationswith RNA-based samples using a software tool MIDAS(<http://www.tigr.org/softlab>). The resulting datawasaveraged from triplicate genes on each array, fromduplicate flip-dye arrays for each experiment, and frombiological replicates, taking a total of 12 intensity datapoints for each gene. Differentially expressed genes at the95% confidence level were determined using intensity-dependent Z-scores (with Z = 1.96) as implemented inMIDAS and the union of all genes identified at each timepoint were considered significant in this experiment. Theresulting data (average of three independent experiments)were organized and visualized based on similar expressionvectors in genes using Euclidean distance and hierarchicalclustering with average linkage clustering method to viewthe whole data set (Figs. 2, 3) and K-means to group thegenes in 60 clusters (Fig. 1) with TIGR MEV (<http://www.tigr.org/software>).

Results

Identification of genes that have their expressionmodulated during exposure to voriconazole

To identify A. fumigatus genes that were differentiallyregulated during contact with voriconazole, we

determined the transcriptional profile of A. fumigatusexposed to 0.5 lg/ml of this azole for different periods oftime at 37�C. This concentration was chosen based on arecent manuscript describing that the distribution ofMICs reported by eight laboratories on ten separatedays for two A. fumigatus strains was in the range of0.25–1.0 lg/ml of voriconazole (Espinel-Ingroff et al.2005). Microarray hybridizations were carried out withRNA obtained from mycelia harvested at 30, 60, 120,and 240 min after adding 0.5 lg/ml of voriconazole tothe cultures. Aiming to verify if voriconazole could af-fect viability consequently causing cell death, instead ofexposing mycelia we exposed A. fumigatus germlings(8 h germinated) to 0.5 lg/ml of voriconazole for240 min. Under these conditions, A. fumigatus germlingsretained more than 80% viability (data not shown),suggesting low levels of A. fumigatus cell death duringshort time exposure to voriconazole. The RNA obtainedfrom the original mycelial culture before adding voric-onazole was taken as reference (t = 0 h). By comparingthe expression of genes between the reference time andafter addition of voriconazole (30, 60, 120, and240 min), we identified 2,271 differentially expressedgenes in the wild-type strain (for details, see Sect. ‘‘Geneexpression methods’’). A direct analysis of the modu-lated genes identified through our microarray hybrid-ization experiments allowed us to detect several geneswith increased or decreased mRNA involved in a varietyof cellular processes. Their specific modulation is likelyto be implicated with the A. fumigatus adaptation to thepresence of voriconazole.

The 2,271 modulated genes have been analyzed withthe aid of a K-means algorithm, in an attempt to clustergenes according to the similarities in their expressionprofiles. Their distribution into 60 distinct clusters showsa large number of genes with minor alterations in theirexpression levels, while others were more severely up- or

Table 1 Comparison of gene expression values obtained with the microarray hybridization and real-time RT-PCR

Genes 30 min 60 min 120 min 240 min Correlation coefficienta

Array/RT Array/RT Array/RT Array/RT RP/RS

Afu5g06070 1.00/�1.09 0.75/�0.53 1.80/0.89 4.60/2.27 0.93/0.80Afu1g10390 1.80/2.74 1.10/2.78 0.50/3.69 1.10/2.24 �0.61/�0.63Afu1g14330 �0.15/�0.02 �0.15/1.93 1.00/3.22 2.00/4.49 0.91/0.95Afu6g03470 1.00/�1.51 0.80/0.83 2.00/1.69 1.75/3.19 0.61/0.69Afu7g00480 1.30/0.81 0.30/2.84 1.10/3.39 1.60/4.31 0.13/0.40Afu8g05710 2.00/3.43 �0.10/2.31 �0.05/0.21 �0.20/0.73 0.55/0.40Afu1g15490 �0.20/�0.86 0.15/1.68 0.10/3.18 �1.60/4.23 �0.67/�0.40Afu1g03200 0.10/�0.42 0.10/1.37 0.60/3.99 2.00/4.85 0.82/0.95Afu1g14050 �0.15/0.02 0.10/2.89 2.50/5.99 3.40/9.23 0.96/1.00Afu7g06680 0.10/�0.34 �0.15/2.05 1.90/1.78 5.60/6.46 0.91/0.40Afu8g05800 0.40/1.59 �0.20/2.18 2.50/2.79 4.25/5.38 0.91/0.80Afu4g12470 1.20/0.71 �0.20/�0.01 1.00/1.49 1.90/2.42 0.90/0.79Afu8g05010 1.75/1.40 0.70/0.66 0.60/1.74 1.10/5.89 0.10/0.00

The values are shown by the log2 ratio between the experimental time point and the reference value (0 h). The table also shows thecorrelation coefficient calculated after comparison of the array and the RT-PCR data for each geneaWe calculated both Pearson’s (RP) and Spearman’s (RS) correlation coefficients for each pair of curves by using Sigma Stat software(Jandel Scientifics Corp., Erkrath, Germany). Alpha was set to 0.05. The former is more adequate for comparisons in which the dataassume a normal distribution, while the latter is more appropriate for datasets which do not follow a normal distribution. The highest R ishighlighted in bold. The P-value obtained for all the correlations was £ 0.05

down-regulated at one or more steps of the exposure tovoriconazole (Fig. 1). We focused our attention on thefour clusters that seemed to contain genes with the mostintense and consistent increased and decreased mRNAexpression (counting from the top left to right, clusternumbers 3 and 10, and 28 and 35, respectively, indicatedin Fig. 1). In cluster 3, genes were highly expressed after30 min (Fig. 2a), while cluster 10 contained genes thathad increased mRNA expression at 120 and 240 min(Fig. 2b). In clusters 28 and 35, we observed genes thathave their mRNA expression decreased at all exposuretimes. In cluster 3, we have observed the A. fumigatusppoA gene (Afu4g10770) that encodes a fatty acid oxy-genase (Tsitsigiannis et al. 2005b). This gene has beenproposed as an activator of mammalian immune re-sponses contributing to enhanced resistance to oppor-tunistic fungi and as a factor that modulate fungaldevelopment contributing to resistance to host defenses(Tsitsigiannis et al. 2005b). In addition, analysis of the

Aspergillus nidulans genome has led to the identificationof three fatty acid oxygenases (PpoA, PpoB, and PpoC)predicted to produce psi factors that have been shown toalter the ratio of asexual to sexual sporulation in thefilamentous fungus A. nidulans (Tsitsigiannis et al.2005a). The increased expression of the ppoA gene dur-ing exposure to voriconazole could suggest a role for thisgene in signaling during stressing conditions. We alsoobserved increased expression in this cluster of twogenes, one that encodes the theta class of glutathione S-transferase (gstA, Burns et al. 2005; Afu7g05500) and anABC multidrug transporter (Afu1g10390), respectively.Both genes could be involved in voriconazole detoxifi-cation. The gstA has already shown increased expressionat mRNA level in the presence of H2O2 or 1-chloro-2,4-dinitrobenzene (Burns et al. 2005) and our results sug-gest a role for these enzymes in the response of A. fu-migatus to antifungal agents, such as voriconazole. Incluster 10, we identified mdr1 (Afu5g06070) that encodes

Fig. 1 Clusters of gene expression generated by the K-meansalgorithm. The 2,271 genes that showed modulation in expressionduring exposure to voriconazole were evaluated by a Figure ofMerit algorithm. The obtained results supported their sub-divisioninto 60 clusters, which was achieved with the aid of a K-meansalgorithm. Groups of genes with similar modulation of geneexpression during the exposure to voriconazole are located in eachcluster. The figure shows, in the Y-axis, the variation in the

Log2(Cy5/Cy3) ratios (from �7 to 7) along with the different timepoints of the exposure to voriconazole (X-axis), taking as areference their respective expression levels at time zero. Clusters 3and 10, and 28 and 35, containing genes that displayed the mostintense and consistent increased and decreased mRNA expression,respectively are indicated by bold lines and shown in more detail inFig. 2

an ABC multidrug transporter, and two genes that en-code a cytochrome P450 (Afu8g00560) and a lipase/esterase (Afu1g15430), respectively. Expression of mdr1in Saccharomyces cerevisiae conferred increased resis-tance to the antifungal agent cilofungin (LY121019), anechinocandin B analog (Tobin et al. 1997). The in-creased expression of the cytochrome P450 and the li-pase/esterase could be related to detoxification andmodifications in the viscosity and lipid composition ofthe plasma membrane that occur during exposure tovoriconazole.

On the other hand, clusters 28 and 35 contain a seriesof genes that display consistent decreased expressionduring exposure to voriconazole (Fig. 2c, d). Among the

genes found in these two clusters, we observed severalgenes related to the composition of the cell membrane,such as, a esterase family protein, a S-adenosyl-L-methi-onine:C-24-D-sterol-methyltransferase (ERG6) and asterol-o-acyltransferase (Afu8g06350, Afu4g09190, andAfu1g06040). Intriguingly, two conidial hydrophobins(rodA and rodB; Afu5g09580 and Afu1g17250) have theirexpression decreased during exposure to voriconazole.

Eighty-two genes that displayed the highestexpression when A. fumigatus mycelium was exposedto voriconazole have been manually selected from oursignificant genes list (see Supplemental data, Table 2)and organized in hierarchical clusters, which is shownin Fig. 3. These selected voriconazole-induced genes

Fig. 2 Hierarchical clusteringshowing the pattern ofexpression of A. fumigatusgenes contained in clusters 3 (a),10 (b), 28 (c), and 35 (d) whichshowed the increased anddecreased levels of mRNAexpression during exposure tovoriconazole. Clusternumbering is considered fromthe upper left corner of Fig. 1.The color code displays theLog2(Cy5/Cy3) ratio for eachtime point, having Cy3 as thereference value (time point=0)

identified by the microarray studies could be broadlyclassified into several groups according to their pre-dicted function. There are several genes that encodetransporters, transcription factors, proteins involved incell metabolism, and hypothetical proteins. To validatethe expression of some of these genes during exposureto voriconazole, we designed Lux probes and usedreal-time RT-PCR analysis (Fig. 4). The results wereexpressed as the relative number of A. fumigatustranscripts, where the reference time zero transcriptcopy number was given a value of 1. Based on thegenes more expressed displayed in Fig. 3, we chose 13genes that encode five ABC-transporters (abcA =Afu5g06070; abcB = Afu1g10390; abcC = Afu1g14330;abcD = Afu6g03470; and abcE = Afu7g00480), threeMFS-transporters (mfsA = Afu8g05710; mfsB =Afu1g15490; and mfsC = Afu1g03200), a F-box protein(fbpA = Afu1g14050), an AAA-protease (aaaA =Afu7g06680), a C6 finger protein (finA = Afu8g05800),

CpcA (cpcA = Afu4g12470), and a Zn finger protein(zfpA = Afu8g05010). We also compared the gene expres-sion variation estimated by these two methodologies bycalculating both Pearson’s (RP) and Spearman’s (RS) cor-relation coefficients for the Log2 ratios obtained by the twoapproaches. As shown in Table 1, positive correlation wasobserved for both RP and RS in 11 out of 13 genes (about85%of the cases). Furthermore, the value of eitherRP orRS

was above 0.50 (indicating moderate to strong correlation)in 9 out of 13 genes (70% of the cases). Thus, although wewere able to detect some discrepancies between the twomethodologies, it seems that our microarray hybridizationapproach is capable of providing information aboutAspergillus fumigatus gene expression modulation with aconsiderably high level of confidence. Our results suggestthat all these 13 genes described might be involved in A.fumigatus adaptation to vorizonazole, considering that theywere more expressed when A. fumigatus mycelium was ex-posed to voriconazole.

Fig. 2 (Contd.)

Gene expression induced by voriconazole is dependenton the cAMP-PKA signaling pathway

Although it is not well comprehended, there are evi-dences showing the cAMP-PKA signaling pathwaymodulates susceptibility of Candida spp. and S. cerevi-siae to antifungal azoles (Kontoyiannis and Rupp 2000;Jain et al. 2003). The cAMP-PKA signaling pathwayalso mediates resistance to dicarboximide and aromatichydrocarbon fungicides in Ustilago maydis (Rameshet al. 2001). Liebmann et al. (2003, 2004) have isolatedand deleted the A. fumigatus adenylate cyclase (acyA)

and cAMP-dependent protein kinase A (pkcA1) genes.As a preliminary step to investigate the involvement ofthe cAMP-PKA signaling pathway in tolerance toantifungal drugs, first we have grown the wild type,DacyA, and DpkcA1 strains in the presence of differentconcentrations of voriconazole, itraconazole, ampho-tericin, and caspofungin. These mutant strains haveshown the same degree of susceptibility to these drugsthan the wild-type strain (data not shown).

Next, we verified whether the induction of the dif-ferent genes confirmed by real-time RT-PCR as moreexpressed in response to voriconazole was dependent

Fig. 3 Hierarchical clusteringshowing the pattern ofexpression of A. fumigatusselected genes duringadaptation to voriconazole. Thecolor code displays theLog2(Cy5/Cy3) ratio for eachtime point, having Cy3 as thereference value (hyphae at timepoint=0)

on the A. fumigatus DacyA and DpkcA1 background.Thus, the wild-type and the mutant strains weregrown in the presence of 0.5 lg/ml of voriconazole,RNA was extracted and real-time RT-PCR performedwith the Lux probes described in Fig. 4. We observedthat abcA mRNA expression was not dependent onboth acyA and pkcA1 mutant background, i.e., thisgene showed comparable mRNA levels in both wild-type and mutant background (Fig. 4a). In contrast,abcB, abcC, abcD, abcE, mfsB, fbpA, finA, cpcA, andzfpA mRNA expression were decreased in both mu-tants when compared to the wild-type strain (Fig. 4b–e, g, i, k–m). Interestingly, we have observed thatsome of the genes are dependent upon adenyl cyclasebut not protein kinase A mutant backgrounds, andvice versa. For example, mfsA, and mfsC mRNAexpression were not dependent on acyA mutantbackground (Fig. 4f, h) while aaaA mRNA expressionwas not dependent on pkcA1 mutant background(Fig. 4j). This probably could reflect differences attranscriptional level regulation of these genes.

Taken together, these results suggest that the cAMP-PKA signaling pathway plays a role in the transcrip-tional regulation of genes that are induced duringadaptation to voriconazole in A. fumigatus.

Expression of genes that encode transportersand enzymes involved in the ergosterolbiosynthesis pathway

As mentioned above, the reduced intracellular accumu-lation of azoles has been correlated with changes ingenes in the ergosterol pathway and overexpression ofABC and MFS transporters. Thus, we identified all thegenes among the 2,271 genes differentially expressed thatcould possibly belong to these categories and organizedthem in hierarchical clusters as shown in Figs. 5 and 6.There are 12, 14, and 15 transporter-encoding genes thatbelong to different transporter categories displaying highexpression at 30–240, 120–240, and 30 min (Fig. 5).Curiously, there are 84 transporter genes that showeddecreased mRNA expression during adaptation to vo-riconazole. The differential expression of these trans-porter-encoding genes could reflect the differentstrategies used by the mycelia to adapt to the presence ofvoriconazole.

There are about 20 genes involved in the biosynthesisof ergosterol in A. fumigatus (Ferreira et al. 2005). Wewere able to identify nine of these genes as having theirexpression modulated during the adaptation to vorico-nazole (Fig. 6). Interestingly, three of them, two para-

Fig. 4 Fold increase in RNAlevels in response to theadaptation to voriconazole.Mycelia were grown in theabsence of any drug and thentransferred to voriconazole0.5 lg/ml for 30, 60, 120, and240 min. Real-time RT-PCRwas the method used toquantify the mRNA. Themeasured quantity of the abcA(a), abcB (b), abcC (c), abcD (d),abcE (e), mfsA (f), mfsB (g),mfsC (h), fbpA (i), aaaA (j), finA(k), cpcA (l), and zfpA (m)mRNA in each of the treatedsamples was normalized usingthe CT values obtained for thetubC RNA amplifications runin the same plate. The relativequantitation of all the genesand tubulin gene expression wasdetermined by a standard curve(i.e., CT values plotted againstlogarithm of the DNA copynumber). Results of four sets ofexperiments were combined foreach determination; means areshown. The values represent thenumber of times the genes areexpressed compared to the wild-type control grown without anydrug (represented absolutely as1.00)

logs of erg24 and an ortholog of erg25, have theirmRNA expression increased. In contrast, the two or-thologs that encode the target protein Erg11 have theirmRNA expression decreased. Moreover, erg1, one ofthe paralogs of erg3, erg4, -7, -10, and the two paralogsof erg13 have also their mRNA expression decreased.

Discussion

The results of the present work demonstrate the use ofhigh throughput microarray hybridization analysis toexamine gene expression during the adaptation of A.fumigatus mycelium to voriconazole. As a source forvoriconazole, we have used Vfend (Pfizer) which is morethan 90% sodium sulfobutyl ether cyclodextrin. Al-though unlikely, considering that no controls were per-formed to determine which changes in the transcriptomewere due to sodium sulfobutyl ether cyclodextrin andwhich were due to voriconazole, it cannot be completelydiscarded the possibility that some of the observedchanges might be due to the carrier rather than to

voriconazole. To our knowledge that is the first study ofmicroarray hybridization analysis for a filamentousfungus exposed to an antifungal agent. Several studieshave already been performed in S. cerevisiae and C.albicans (Bammert and Fostel 2000; De Backer et al.2001; Zhang et al. 2002; Agarwal et al. 2003; Rogers andBarker 2003; Karababa et al. 2004; Barker et al. 2004;Liu et al. 2005). In our study, we have observed: (i) adecreased mRNA expression of various ergosterol bio-synthesis genes; (ii) increased mRNA expression ofgenes involved in a variety of cell functions, such astransporters, transcription factors, proteins involved incell metabolism, and hypothetical proteins; and (iii) theinvolvement of the cAMP-PKA signaling pathway in theoverexpression of several of these genes.

Genes involved in the ergosterol biosynthesis pathway

Azoles block the ergosterol biosynthesis pathway byinhibiting the enzyme 14-a-demethylase, product of theCYP51/erg11 gene (Diaz-Guerra et al. 2003). Fungal

Fig. 4 (Contd.)

azole resistance involves both amino acid changes in thetarget site that alter drug–target interactions and thosethat decrease net azole accumulation (Marichal et al.1999; Sanglard and Odds 2002; White et al. 1998). An-other aspect of the azole resistance conferred by theCYP51 genes refers to the fact that these genes canconfer resistance when overexpressed. Osherov et al.(2001) showed that the overexpression of the A. nidulansCYP51A/pdmA/erg11 gene into A. fumigatus resulted inincreased resistance to itraconazole. These results

indicate that triazole resistance in clinical isolates ofmoulds may result from amplification or overexpressionof the P-450 14-a-demethylase.

About 20 genes are involved in the ergosterol bio-synthesis pathway in A. fumigatus spp. (Ferreira et al.2005). There are several duplicated genes in this path-way, such as erg3 and erg11 that showed three and twocopies, respectively. Besides that, erg4, erg7, erg10,erg13, erg24, and erg25 showed two, three, two, two,two, and two copies, respectively. Very little is known

Fig. 5 Hierarchical clustering showing the pattern of expression ofA. fumigatus genes encoding transporters during adaptation tovoriconazole. The color code displays the Log2(Cy5/Cy3) ratio for

each time point, having Cy3 as the reference value (hyphae at timepoint=0)

Fig. 6 Hierarchical clusteringshowing the pattern ofexpression of A. fumigatusgenes encoding proteins fromthe ergosterol biosynthesispathway during adaptation tovoriconazole. The color codedisplays the Log2(Cy5/Cy3)ratio for each time point,having Cy3 as the referencevalue (hyphae at time point=0)

about the genetics and biochemistry of the ergosterolbiosynthesis pathway in A. fumigatus. Mellado et al.(2001) have shown that both copies of the A. fumigatuserg11 were expressed. The existence of several duplicatedgenes in the ERG pathway could be a good strategy tomodulate the composition and fluidity of the cell mem-brane. This could confer several adaptative advantagesto colonize new environments and also to counteractantifungal drugs. We have observed a decrease in theexpression of several genes in this pathway duringadaptation to voriconazole, including both erg11 genesand one of the erg3 paralogs. Interestingly, the erg24and erg25 genes showed increased mRNA expressionduring adaptation to voriconazole. These genes encodeD14-sterol reductase and 4-methylsterol oxidase,respectively. They act in two subsequent steps in theergosterol biosynthesis pathway by converting 4,4-di-methyl-cholesta-8,12,24-trienol to 4,4-dimethyl-8,24-cholestadienol (by Erg24 enzymatic activity) andsubsequently to 4-methyl-8,24-cholestadienol (by Erg25enzymatic activity). Contrasting results have beenobserved by De Backer et al. (2001) in C. albicans, whenthese authors verified adaptation to voriconazole. Aglobal mRNA increase of ERG genes in response toazole treatment was observed by these authors. ERG11and ERG5 mRNA expression was found to be increasedapproximately 12-fold. In addition, a significant increasein mRNA expression was observed for ERG6, ERG1,ERG3, ERG4, ERG10, ERG9, ERG26, ERG25, ERG2,IDII, HMGS, NCP1, and FEN2, all of which are genesknown to be involved in ergosterol biosynthesis.Bammert and Fostel (2000) also observed comparableresults by genome-wide transcript profiles followingexposure of S. cerevisiae to a number of antifungalagents targeting ergosterol biosynthesis (clotrimazole,fluconazole, itraconazole, ketoconazole, voriconazole,terbinafine, and amorolfine). In addition to ergosterolbiosynthesis genes, 36 mitochondrial genes and a num-ber of other genes with roles related to ergosterol func-tion were responsive. The increased mRNA expressionof ergosterol biosynthesis genes was also observed byAgarwal et al. (2003) and Liu et al. (2005) upon expo-sure to azoles in S. cerevisiae and C. albicans, respec-tively. It is possible these differences among A. fumigatusand yeasts could reflect intrinsic differences in the reg-ulation of genes expressing proteins involved in theergosterol biosynthesis when the cells are exposed toazoles. Since this is a voriconazole-susceptible strain,what is being seen by microarrays and RT-PCR indecreasing the mRNA expression of these genes may notbe an adaptive process but possibly a mal-adaptive onethat contributes to sensitivity.

Genes encoding transporters

The reduced intracellular accumulation of azoles has alsobeen correlated with overexpression of MDR effluxtransporter genes of the ABC and MFS classes (Lupetti

et al. 2002). A. fumigatus has proportionally more ABCand MFS transporter-encoding genes than S. cerevisiae,Schizosaccharomyces pombe, and Neurospora crassa(Ferreira et al. 2005). In contrast to the extensive numberof genes encoding transporters in A. fumigatus, there arevery few studies characterizing these transporters andtheir relationship with azole resistance. We have ob-served several genes encoding both ABC and MFStransporters as more expressed during A. fumigatusadaptation to voriconazole. Tobin et al. (1997) identifiedtwo genes (AfuMDR1 and AfuMDR2) in A. fumigatusencoding proteins of the ABC superfamily. Expression ofAFUMDR1 in S. cerevisiae conferred increased resis-tance to the antifungal agent cilofungin (LY121019), anechinocandin B analog. Slaven et al. (2002) cloned agene, atrF, from A. fumigatus that has characteristicMDR motifs. Dot blot analysis revealed that AF72 hasapproximately fivefold higher levels of atrF mRNA thansusceptible isolates AF10 and H06-03 in cultures withsub-minimum inhibitory concentration (sub-MIC) levelsof itraconazole. Nascimento et al. (2003) identified Af-umDMR3 and AfumDMR4 that encode an MFS and anABC transporter, respectively. Real-time quantitativePCR with molecular beacon probes was used to assessexpression levels of AfumDMR3 and AfumDMR4. Itr-aconazole-resistant mutants showed either constitutivehigh-level expression of both genes or induction ofexpression upon exposure to itraconazole. These resultssuggest that overexpression of one or both of thesetransporters is linked to high-level itraconazole resis-tance in A. fumigatus. Langfelder et al. (2002) identifiedan ABC transporter-encoding gene designated abcA. AnabcA deletion mutant did not show increased sensitivityto itraconazole, amphotericin B, voriconazole, posaco-nazole, ravuconazole, and echinocandins.

From this list of genes, we were able to identify onlymdr1 (Afu5g06070) as one of the most expressed genes inour microarray experiments, suggesting the other trans-porter-encoding genes are not playing a role in adaptationto voriconazole. Large-scale inactivation experiments forA. fumigatus transporter-encoding genes could help todefine which of these genes are involved in antifungaldrug resistance in this species.

Effects of voriconazole adaptation on the mRNAexpression of other genes

There are several genes encoding transcription factorsthat display increase in the mRNA level duringadaptation to voriconazole, such as two C2H2 fingerdomain (Afu8g05010 and Afu3g08010), three C6 fingerdomain (Afu8g05800, Afu3g00930, and Afu3g01980),and cpcA, that encodes a bZIP transcription factorwhich is a functional ortholog of S. cerevisiae Gcn4p,the transcriptional activator of the cross-pathwaycontrol system of amino acid biosynthesis (Krappmannet al. 2004). We have also observed genes encoding anF-box protein (fbpA; Afu1g14050), a AAA family

ATPase (aaaA; Afu7g06680), and calmodulin (calA;Afu4g10050). These transcription factors could be in-volved in the response to the stress conditions that areoccurring during the adaptation to voriconazole. Theycould have as targets, genes involved in the detoxifi-cation such as genes that encode transporters. TheAAA family ATPase is an ATPase associated withvarious cellular activities, but it seems to share thecommon behavior of inducing conformational changesin target proteins. These conformational changes leadto substrate remodeling and, in some cases, perturbprotein structure sufficiently to promote unfolding (fora review, see Hanson and Whiteheart 2005). It ispossible the A. fumigatus aaaA is mediating protein–protein reactions including unfolding for proteolysis,the disassembly of protein aggregates, and the disas-sembly of otherwise stable protein complexes, that aredirectly or indirectly involved in adaptation to voric-onazole. The increased mRNA expression of the cal-modulin encoding gene could also suggest that thecalcium signaling pathway can play a role in A. fu-migatus adaptation to voriconazole. Actually, it hasbeen shown in C. albicans that calcineurin is essentialfor virulence and survival during membrane perturba-tion by azoles (Sanglard et al. 2003a, b).

Cyclic AMP signaling pathway and adaptationto voriconazole

The cAMP-PKA signaling pathway has been shown tobe involved in susceptibility of Candida spp. and S. ce-revisiae to antifungal azoles (Kontoyiannis and Rupp2000; Jain et al. 2003). C. albicans and S. cerevisiaestrains mutated in the adenylate cyclase and cyclase-associated protein were hypersusceptible to fluconazole,itraconazole, or miconazole (Jain et al. 2003). The sameauthors showed that a defect in azole-dependentupregulation of the C. albicans multidrug transporterCDR1 contributes to hypersusceptibility of the adenyl-ate cyclase mutant. Contrasting with these results,Kontoyiannis and Rupp (2000) showed that cAMPgeneration in S. cerevisiae results in some protectionfrom fluconazole toxicity in a way that is independent ofthe efflux transporter Pdr5p. We have shown that al-though A. fumigatus deletion mutants are not moresensitive to azoles (itraconazole and voriconazole),amphotericin, and caspofungin, they showed decreasedmRNA expression levels of several genes shown as moreexpressed during adaptation to voriconazole, includingseveral ABC and MFS transporters. These resultsdemonstrated that cAMP-PKA signaling pathway couldplay at least an indirect role in azole-dependent geneexpression.

Our results provide a first step toward a globalunderstanding of the mechanisms that are involved inthe adaptation of filamentous fungi to azoles.Furthermore, this work demonstrates the potentialutility of gene expression profiling in antifungal studies.

Acknowledgments We would like to thank the Fundacao de Am-paro a Pesquisa do Estado de Sao Paulo and Conselho Nacional deDesenvolvimento Cientıfico e Tecnologico, both from Brazil, forfinancial support for our research, and SPP1160 of the DeutscheForschungsgemeinschaft for A.A.B. Microarray analysis work wassupported by NIAID grant to W.C.N. We also thank the twoanonymous reviewers and the editor for the suggestions.

References

Agarwal AK, Rogers PD, Baerson SR, Jacob MR, Barker KS,Cleary JD, Walker LA, Nagle DG, Clark AM (2003) Genome-wide expression profiling of the response to polyene, pyrimi-dine, azole, and echinocandin antifungal agents in Saccharo-myces cerevisiae. J Biol Chem 12:34998–35015

Bammert GF, Fostel JM (2000) Genome-wide expression patternsin Saccharomyces cerevisiae: comparison of drug treatmentsand genetic alterations affecting biosynthesis of ergosterol.Antimicrob Agents Chemother 44:1255–1265

Barker KS, Crisp S, Wiederhold N, Lewis RE, Bareither B, Eck-stein J, Barbuch R, Bard M, Rogers PD (2004) Genome-wideexpression profiling reveals genes associated with amphotericinB and fluconazole resistance in experimentally induced anti-fungal resistant isolates of Candida albicans. J AntimicrobChemother 54:376–385

Burns C, Geraghty R, Neville C, Murphy A, Kavanagh K, Doyle S(2005) Identification, cloning, and functional expression ofthree glutathione transferase genes from Aspergillus fumigatus.Fungal Genet Biol 42:319–327

Dannaoui E, Borel E, Monier MF, Piens MA, Picot S, Persat F(2001) Acquired itraconazole resistance in Aspergillus fumiga-tus. J Antimicrob Chemother 47:333–340

De Backer MD, Ilyina T, Ma XJ, Vandoninck S, Luyten WH,Vanden Bossche H (2001) Genomic profiling of the response ofCandida albicans to itraconazole treatment using a DNAmicroarray. Antimicrob Agents Chemother 45:1660–1670

Denning DW (1996) Diagnosis and management of invasiveaspergillosis. Curr Clin Top Infect Dis 16:277–299

Denning DW, Venkateswarlu K, Oakley KL, Anderson MJ,Manning NJ, Stevens DA, Warnock DW, Kelly SL (1997)Itraconazole resistance in Aspergillus fumigatus. AntimicrobAgents Chemother 41:1364–1368

Diaz-Guerra TM, Mellado E, Cuenca-Estrella M, Rodrıguez-Tu-dela JL (2003) A point mutation in the 14a-sterol demethylasegene cyp51A contributes to itraconazole resistance in Asper-gillus fumigatus. Antimicrob Agents Chemother 47:1120–1124

Espinel-Ingroff A, Fothergill A, Ghannoum M, Manavathu E,Ostrosky-Zeichner L, Pfaller M, Rinaldi M, Schell W, Walsh T(2005) Quality control and reference guidelines for CLSI brothmicrodilution susceptibility method (M 38-A document) foramphotericin B, itraconazole, posaconazole, and voriconazole.J Clin Microbiol 43:5243–5246

Ferreira ME, Colombo AL, Paulsen I, Ren Q, Wortman J, HuangJ, Goldman MH, Goldman GH (2005) The ergosterol biosyn-thesis pathway, transporter genes, and azole resistance inAspergillus fumigatus. Med Mycol 43:S313–S319

Hanson PI, Whiteheart SW (2005) AAA+ proteins: have engine,will work. Nat Rev Mol Cell Biol 6:519–529

Herbrecht R, Denning D, Patterson TF, Bennett JE, Greene RE,Oestmann JW, Kern WV, Marr KA, Ribaud P, Lortholary O,Silvestre R, Rubin RH, Wingard JR, Stark P, Durand C,Caillot D, Thiel E, Chandrasekar PH, Hodges MR, SchlammHT, Troke PF, de Pauw B (2002) Voriconazole versusamphotericin B for primary therapy of invasive aspergillosis. NEngl J Med 347:408–415

Jain P, Akula I, Edlind T (2003) Cyclic AMP signaling pathwaymodulates susceptibility of candida species and Saccharomycescerevisiae to antifungal azoles and other sterol biosynthesisinhibitors. Antimicrob Agents Chemother 47:3195–3201

Kafer E (1977) Meiotic and mitotic recombination in Aspergillusand its chromosomal aberrations. Adv Genet 19:33–131

Karababa M, Coste AT, Rognon B, Bille J, Sanglard D (2004)Comparison of gene expression profiles of Candida albicansazole-resistant clinical isolates and laboratory strains exposedto drugs inducing multidrug transporters. Antimicrob AgentsChemother 48:3064–3079

Kelly SL, Lamb DC, Kelly DE, Loeffler J, Einsele H (1996)Resistance to fluconazole and amphotericin in Candida albicansfrom AIDS patients. Lancet 348:1523–1524

Kelly SL, Lamb DC, Kelly DE, Manning NJ, Loeffler J, Hebart H,Schumacher U, Einsele H (1997) Resistance to fluconazole andcross-resistance to amphotericin B in Candida albicans fromAIDS patients caused by defective sterol delta 5,6-desaturation.FEBS Lett 400:80–82

Kontoyiannis DP, Rupp S (2000) Cyclic AMP and fluconazoleresistance in Saccharomyces cerevisiae. Antimicrob AgentsChemother 44:1743–1744

Krappmann S, Bignell EM, Reichard U, Rogers T, Haynes K,Braus GH (2004) The Aspergillus fumigatus transcriptionalactivator CpcA contributes significantly to the virulence of thisfungal pathogen. Mol Microbiol 52:785–799

Langfelder K, Gattung S, Brakhage AA (2002) A novel methodused to delete a new Aspergillus fumigatus ABC transporter-encoding gene. Curr Genet 41:268–274

Liebmann B, Gattung S, Jahn B, Brakhage AA (2003) cAMPsignaling in Aspergillus fumigatus is involved in the regulationof the virulence gene pksP and in defense against killing bymacrophages. Mol Genet Genomics 269:420–435

Liebmann B, Muller M, Braun A, Brakhage AA (2004) The cyclicAMP-dependent protein kinase a network regulates develop-ment and virulence in Aspergillus fumigatus. Infect Immun72:5193–5203

Liu TT, Lee RE, Barker KS, Lee RE, Wei L, Homayouni R,Rogers PD (2005) Genome-wide expression profiling of theresponse to azole, polyene, echinocandin, and pyrimidine anti-fungal agents in Candida albicans. Antimicrob Agents Chemo-ther 49:2226–2236

Lupetti A, Danesi R, Campa M, del Tacca M, Kelly S (2002)Molecular basis of resıstance to azole antifungals. Trends MolMed 8:76–81

Manavathu EK, Vazquez JA, Chandrasekar PH (1999) Reducedsusceptibility in laboratory-selected mutants of Aspergillus fu-migatus to itraconazole due to decreased intracellular accumu-lation of the antifungal agent. Int J Antimicrob Agents 12:213–219

Mann PA, Parmegiani RM, Wei S-Q, Mendrick CA, Li X, Loen-berg D, DiDomenico B, Hare RS, Walker SS, McNicholas PM(2003) Mutations in Aspergillus fumigatus resulting in reducedsusceptibility to posaconazole appear to be restricted to a singleamino acid in the cytochrome P-450 14a-demethylase. Anti-microb Agents Chemother 47:577–581

Marichal P, Koymas L, Willlemsens S, Bellens D, Verhasselt P,Luyten W, Borgers M, Ramaekers FCS, Odds FC, Vanden-Bossche H (1999) Contribution of mutations in the cytochromeP-450 14-a-demethylase (Erg11p, Cyp51p) to azole resistance inCandida albicans. Microbiology 145:2701–2713

Mellado E, Diaz-Guerra TM, Cuenca-Estrella M, Rodriguez-Tu-dela JL (2001) Identification of two different 14-alpha steroldemethylase-related genes (cyp51A and cyp51B) in Aspergillusfumigatus and other Aspergillus species. J Clin Microbiol39:2431–2438 (Erratum in: J Clin Microbiol 2001; 39:4225)

Nascimento AM, Goldman GH, Park S, Marras SA, Delmas G,Oza U, Lolans K, Dudley MN, Mann PA, Perlin DS (2003)Multiple resistance mechanisms among Aspergillus fumigatusmutants with high-level resistance to itraconazole. AntimicrobAgents Chemother 47:1519–1526

National Committee for Clinical Laboratory Standards (2002)Reference method for broth dilution antifungal susceptibilitytesting of conidium-forming filamentous fungi. Proposed stan-dard M38-A. National Committee for Clinical LaboratoryStandards, Wayne

Nolte FS, Parkinson T, Falconer DJ, Dix S, Williams J, Gilmore C,Geller R, Wingard JR (1997) Isolation and characterization offluconazole- and amphotericin B-resistant Candida albicansfrom blood of two patients with leukaemia. Antimicrob AgentsChemother 44:196–199

Osherov N, Kontoyannis DP, Romans A, May GS (2001) Resis-tance to itraconazole in Aspergillus nidulans and Aspergillus fu-migatus is conferred by extra copies of theA. nidulans P-450 14a-demethylase gene, pdmA. J Antimicrob Chemother 48:75–81

Ramesh MA, Laidlaw RD, Durrenberger F, Orth AB, KronstadJW (2001) The cAMP signal transduction pathway mediatesresistance to dicarboximide and aromatic hydrocarbon fungi-cides in Ustilago maydis. Fungal Genet Biol 32:183–193

Rogers PD, Barker KS (2003) Genome-wide expression profileanalysis reveals coordinately regulated genes associated withstepwise acquisition of azole resistance in Candida albicansclinical isolates. Antimicrob Agents Chemother 47:1220–1227

Sanglard D, Odds FC (2002) Resistance of Candida species toantifungal agents: molecular mechanisms and clinical conse-quences. Lancet Infect Dis 2:73–85

Sanglard D, Ischer F, Parkinson T, Falconer D, Bille J (2003a)Candida albicans: mutations in the ergosterol biosyntheticpathway and resistance to several antifungal agents. Antimic-rob Agents Chemother 47:2404–2412

Sanglard D, Ischer F, Marchetti O, Entenza J, Bille J (2003b)Calcineurin A of Candida albicans: involvement in antifungaltolerance, cell morphogenesis and virulence. Mol Microbiol48:959–976

Semighini CP, Marins M, Goldman MHS, Goldman GH (2002)Quantitative analysis of the relative transcript levels of ABCtransporter Atr genes in Aspergillus nidulans by real-time re-verse transcription-PCR assay. Appl Environ Microbiol68:1351–1357

Slaven JW, Anderson MJ, Sanglard D, Dixon GK, Bille J, RobertsIS, Denning DW (2002) Increased expression of a novelAspergillus fumigatus ABC transporter gene, AtrF, in the pres-ence of itraconazole in an itraconazole resistant clinical isolate.Fungal Genet Biol 36:199–206

Tobin MB, Peery RB, Skatrud PL (1997) Genes encoding multipledrug resistance-like proteins in Aspergillus fumigatus andAspergillus flavus. Gene 200:11–23

Tsitsigiannis DI, Kowieski TM, Zarnowski R, Keller NP (2005a)Three putative oxylipin biosynthetic genes integrate sexual andasexual development in Aspergillus nidulans. Microbiology151:1809–1821

Tsitsigiannis DI, Bok JW, Andes D, Nielsen KF, Frisvad JC,Keller NP (2005b) Aspergillus cyclooxygenase-like enzymes areassociated with prostaglandin production and virulence. InfectImmun 73:4548–4559

White TC, Marr KA, Bowden RA (1998) Clinical, cellular, andmolecular factors that contribute to antifungal drug resistance.Clin Microbiol Rev 11:382–402

Zhang L, Zhang Y, Zhou Y, An S, Zhou Y, Cheng J (2002) Re-sponse of gene expression in Saccharomyces cerevisiae toamphotericin B and nystatin measured by microarrays. J An-timicrob Chemother 49:905–915