A Transcriptomics Approach To Unveiling the Mechanisms of In Vitro Evolution towards Fluconazole Resistance of a Candida glabrata Clinical Isolate

Candida glabrata is an emerging fungal pathogen. Its increased prevalence is associated with its ability to rapidly develop antifungal drug resistance, particularly to azoles.

asis, favoring the increase in drug-resistant C. glabrata infections (1,4). Triazole drugs are significantly safer and more tolerable in systemic therapy than imidazoles (5), while newer triazoles, such as posaconazole and voriconazole, exhibit broader-spectrum and more-potent activity than fluconazole (6).
The most common cause of clinically acquired azole resistance is the upregulation of genes encoding drug efflux pumps from the ATP-binding cassette (ABC) superfamily and the major facilitator superfamily (MFS). One particular ABC transporter, C. glabrata Cdr1 (CgCdr1), is often involved in the acquisition of fluconazole resistance in C. glabrata isolates (7,8). Additionally, CgCdr2/Phd1 and CgSnq2, two other ABC drug efflux pumps, have also been associated with fluconazole resistance in C. glabrata, their overexpression often resulting from the acquisition of gain-of-function (GOF) mutations in the CgPDR1 gene (8,9). Several MFS multidrug transporters have also been linked to fluconazole resistance in C. glabrata (10). For example, azole resistance has been associated with the overexpression of the drug:H ϩ antiporters (DHA) CgQdr2, CgTpo1_1, CgTpo1_2, and CgTpo3 (11)(12)(13)(14). In the case of posaconazole, a study of seven posaconazole-resistant Candida albicans isolates revealed no changes in the expression of the drug transporters Cdr1, Cdr2, and Mdr1 (15), suggesting that posaconazole resistance may be dissociated from antifungal transport.
In many C. albicans clinical isolates, azole resistance arises from point mutations that lead to conformational changes in Erg11, the primary target of azoles (16). In C. glabrata, a G944A mutation in the ERG11 gene was associated with fluconazole, voriconazole, and polyene resistance in one specific isolate (17). A second fluconazoleresistant isolate of C. glabrata was revealed to have increased expression of ERG11 due to duplication of the entire chromosome containing this gene (18). However, in all other studies on azole-resistant clinical isolates of C. glabrata, no mutation or upregulation of the ERG11 gene was observed, suggesting that this is not an important mechanism for clinical acquisition of resistance to azoles (19)(20)(21). On the other hand, an E139A mutation in the ERG3 gene, also involved in ergosterol biosynthesis, was found to lead to increased resistance to fluconazole in C. glabrata strains (22), while in Candida parapsilosis, a similar mutation appeared in a posaconazole-resistant strain (23). Mutation of the ERG3 gene leads to the formation of ergosta-7,22-dien-3␤-ol as the major sterol produced, instead of ergosta-5,7,24(28)-trienol (24). This alteration prevents azole action, since the toxic sterols that accumulate upon the inhibition of Erg11 can no longer be synthesized by this pathway.
Resistance to azole drugs has mostly been examined as a whole, with little distinction between the mechanisms that may be specific to each azole drug. However, several epidemiological surveys on fluconazole, voriconazole, and posaconazole resistance in C. glabrata have revealed that several clinical isolates display different levels of resistance to each of these drugs (25)(26)(27). In this work, the azole-susceptible C. glabrata isolate 044, recovered from a positive blood culture, was exposed for prolonged periods to serum-level concentrations of fluconazole resulting in multiazole resistance: after 21 days, posaconazole resistance was reached, followed by clotrimazole resistance after 31 days and, finally, fluconazole and voriconazole resistance upon 45 days of induction. A transcriptomics characterization of the evolution of the 044 clinical isolate from azole susceptibility to stepwise acquisition of resistance to multiple azoles was carried out. On the basis of transcriptomics data, the role of adhesin-encoding genes, especially CgEPA3, was investigated in the context of azole drug resistance, establishing a fascinating link between cell-to-cell adhesion, biofilm formation, and drug resistance.
By use of spot assays, the 044 clinical isolate and derived populations were further characterized in terms of antifungal drug resistance. This additional assay confirmed the progressive acquisition of resistance to fluconazole and clotrimazole ( Fig. 2A and B) and showed that these strains had indeed become multiazole resistant, since increased resistance to ketoconazole, miconazole, and tioconazole was also observed ( Fig. 2A). Interestingly, no change in resistance to flucytosine or amphotericin B, two other antifungal drugs of different classes, was observed (Fig. 2C).
Transcriptional remodeling underlying the stepwise acquisition of resistance to posaconazole, clotrimazole, and fluconazole/voriconazole. In order to gain insights into the molecular mechanisms underlying the phenotypic changes leading to differential acquisition of resistance to four azoles, the transcriptome-wide changes occurring on the 21st, 31st, and 45th days of fluconazole induction were analyzed by microarray hybridization. It should be noted that the design of the microarrays used was based on the available CBS138 genome; thus, it is possible that specific characteristics of the transcriptome of the clinical isolate analyzed may have been missed in this study. To make sure that the transcriptome observations reflected the evolving population, and not just a specific isolate, a large number of colonies were mixed and analyzed as a whole. Total RNA was extracted from cell populations growing exponentially in YPD medium in the absence of fluconazole in order to ensure that the transcriptome changes observed corresponded to stable transcriptional modifications. Each sample was compared to the azole-susceptible C. glabrata 044 clinical isolate (control), considering a 1.5-fold threshold, with an associated P value of Ͻ0.05. Overall, the expression of 355 genes was downregulated, whereas the expression of 299 genes was upregulated, in the posaconazole-resistant population obtained after 21 days of fluconazole induction (Tables S2 and S3 at http://ibb.tecnico.ulisboa.pt/Cavalheiro_etal _SuplData.pdf). At day 31, when the population had further acquired clotrimazole resistance, the expression of 73 genes was downregulated, whereas that of 199 genes was found to be upregulated, relative to expression in the parental clinical isolate (Tables S4 and S5). Finally, at day 45, upon the acquisition of fluconazole/voriconazole resistance, the expression of 6 genes was downregulated, whereas that of 27 genes was found to be upregulated, relative to expression in the parental clinical isolate (Tables S6  and S8). There is very little overlap between the differentially expressed genes in the different populations, with only 44 genes whose altered expression was maintained from day 21 to day 31, and 9 upregulated genes on day 31 whose activated expression was maintained until day 45 (Fig. S1 at http://ibb.tecnico.ulisboa.pt/Cavalheiro_etal _SuplData.pdf). GOToolBox was used to identify the Gene Ontology (GO) biologicalprocess terms overrepresented in each data set (Fig. S2).
Acquisition of azole drug resistance is accompanied by decreased azole drug accumulation. Two of the most typical changes that occur in azole-resistant strains are alterations in the concentration of ergosterol (28)(29)(30) and the activation of drug efflux pumps, leading to reduced accumulation of the drug. Indeed, in the strains that evolved from clinical isolate 044, changes at the level of the expression of ergosterol biosynthetic genes were observed, especially in the case of ERG11, whose expression was almost 3-fold higher in the 044Fluco31 strain than in the parental strain but returned to basal levels in the 044Fluco45 strain (Fig. 3A). The total ergosterol concentrations in the susceptible clinical isolate 044 and the derived 044Fluco21, 044Fluco31, and 044Fluco45 cells were therefore determined; they were found to remain constant in the 044Fluco21 and 044Fluco31 cells and to decrease slightly in the 044Fluco45 strain (Fig. 3B). No change in the sequence of the ERG11 gene was identified in the four strains studied.
The expression levels of the multidrug transporter genes CDR1 and CDR2 were increased Ͼ2-fold only in 044Fluco45 cells (Fig. 4A). This is consistent with the identification of a nonsynonymous point mutation in the sequence of the transcription factor gene PDR1. This point mutation, leading to a Y372C substitution, is in the same position as the Pdr1 Y372N gain-of-function (GOF) mutation identified in other azoleresistant isolates of C. glabrata (31) and is similar to that mutation. It is indeed likely that this Y372C substitution constitutes a new Pdr1 GOF mutation, since Pdr1 was found to control nearly 50% of the upregulated genes in this population in cells growing exponentially in the absence of fluconazole or any other stress. Indeed, in an attempt to identify the transcription factors (TFs) that underlie the observed transcriptome-wide remodeling, the PathoYeastract database (32) was used (Table 1). In this search, the PathoYeastract data used were based solely on data published specifically for C. glabrata, demonstrating experimentally the regulatory association between the transcription factors and their target genes in this yeast. The regulators of the expression of the majority of the genes upregulated during the acquisition of multiazole resistance are unknown. The TF that controls the highest number of upregulated genes is Hal9, a TF of unknown function, for the posaconazole-resistant strain and Pdr1 for the remaining two evolved strains.  The ability of the evolved cells to reduce the intracellular accumulation of azole drugs was then evaluated using [ 3 H]clotrimazole as a model azole drug. Remarkably, the intracellular accumulation of [ 3 H]clotrimazole decreased in all three resistant cell populations (Fig. 4B). Specifically, the 044Fluco21 cell population was found to accumulate 4 times less clotrimazole than the 044 parental clinical isolate. The 044Fluco31 and 044Fluco45 cells exhibited even lower levels of accumulated clotrimazole, reaching levels nearly 10-fold-lower than those registered in the 044 clinical isolate (Fig. 4B). Although reduced drug accumulation is consistent with the expression profile of the multiazole-resistant strain 044Fluco45, no clear changes in the expression of drugresistant transporters were observed in the 044Fluco21 and 044Fluco31 intermediate strains (Fig. 4A), and no change in the sequence of the PDR1 gene was observed.
This observation prompted us to test the ability of the strains in which azole resistance had evolved to adhere and form biofilms. Indeed, 044Fluco31 cells were found by bright-field microscopy to exhibit a significantly higher level of cell aggregates than 044 and 044Fluco21 cells. The percentage of 044Fluco31 cells found to constitute cell aggregates, considering at least 10 cells/aggregate, reached an average of Ͼ50%, in contrast to only 10% for the remaining cell populations (Fig. 6A). Additionally, the number of cells per aggregate, determined by microscopy, was found to be consistently higher for the 044Fluco31 strain (Fig. 6A). Moreover, 044Fluco31 exhibited higher adherence to human vaginal epithelial cells than the other strains ( Fig.  6B). Interestingly, the evolved strain 044Fluco21 also exhibited moderately increased adhesion to the epithelial cells tested, despite the observation that there was no clear upregulation of adhesin-encoding genes in this population (Fig. 6B). This suggests that alternative mechanisms whose nature is unclear mediate increased adhesiveness in these C. glabrata cells. Since no differences in biofilm formation on polystyrene dishes was observed (Fig. 6C), the increased expression of adhesin-encoding genes in 044Fluco31 cells appears to correlate with their ability to display cell-to-cell adherence and to adhere to epithelial cells. The CgEpa3 adhesin is a new determinant of azole drug resistance. Given the observation that a high level of expression of adhesin-encoding genes correlates with increased azole resistance in the evolved strain 044Fluco31, the possible role in azole drug resistance of CgEPA1, CgEPA3, CgEPA9, CgEPA10, CgAWP12, and CgAWP13, accounting for most of the more highly upregulated adhesin-encoding genes, was assessed. Deletion of the CgEPA3 gene, but not the remaining adhesin-encoding genes, increased the susceptibility of C. glabrata to miconazole, ketoconazole, tioconazole, clotrimazole, and fluconazole (Fig. 7A). These results were confirmed by standard assessment of the fluconazole and clotrimazole MICs for the wild-type strain (16 and 2 mg/liter, respectively) and the ⌬cgepa3 deletion mutant strain (8 and 1 mg/liter, respectively). Overexpression of CgEPA3 in the wild-type strain L5U1 was consistently found to increase C. glabrata resistance to miconazole, ketoconazole, tioconazole, clotrimazole, and fluconazole, confirming the role of this gene as a determinant of azole drug resistance (Fig. 7B). To ensure that the deletion of CgEPA3 also influences drug resistance in the 044Fluco31 background, the effect of its deletion on azole resistance was tested. Deletion of CgEPA3 was found to increase the susceptibility of the 044Fluco31 strain to clotrimazole and fluconazole (Fig. 8), suggesting that the effect of this adhesin is not strain dependent.
Since C. glabrata CgEpa3 was identified as conferring resistance to azole drugs, its possible involvement in reducing clotrimazole accumulation in yeast cells was examined. The accumulation of radiolabeled clotrimazole in nonadapted C. glabrata cells suddenly exposed to 30 mg/liter clotrimazole was approximately 2 times higher in cells devoid of CgEpa3 than in wild-type KUE100 cells (Fig. 9A). Overexpression of CgEPA3 in the wild-type strain L5U1 was consistently found to decrease the accumulation of radiolabeled clotrimazole (Fig. 9B). These findings strongly suggest that CgEpa3 contributes to C. glabrata resistance to azole drugs by reducing their accumulation within yeast cells.
To exclude the possibility that the results observed were due to an indirect effect of CgEPA3 expression on the expression of the drug efflux pump CgCdr1, the effect of deletion or overexpression of CgEPA3 on the levels of CgCDR1 transcripts was evaluated. Interestingly, the expression of CgEPA3 was found to have no effect on CgCDR1 transcript levels (Fig. 10), suggesting that CgEpa3 affects azole drug accumulation independently of drug efflux pump activity. The CgEpa3 adhesin promotes biofilm formation. The effect of deleting CgEPA3, CgEPA1, or CgEPA10 on in vitro biofilm formation on 96-well polystyrene microplates was further assessed. Deletion of CgEPA3, but not deletion of CgEPA1 or CgEPA10, resulted in an almost 2-fold decrease in the level of total biofilms formed relative to that for the parental strain (Fig. 11A). The overexpression of CgEPA3 in the wild-type L5U1 strain was consistently found to increase the ability of C. glabrata to form biofilms on polystyrene (Fig. 11B).

DISCUSSION
In this study, a transcriptomics analysis of the evolution of an azole-susceptible clinical isolate toward azole resistance, induced by long-standing incubation with a therapeutic concentration of fluconazole in serum, was carried out. The selected approach enabled the identification of the changes in gene expression occurring with time during 45 days of evolution toward a simple and stable solution. One of the surprising observations is that the number of differentially expressed genes, relative to expression in the initial azole-susceptible strain, decreases with time of evolution. At day 21, when the cells were resistant to posaconazole only, 654 genes were found to be expressed differently, while that number decreased to 272 after 31 days and to just 33 at day 45, when the cells reached multiazole resistance. The evolved azole-resistant populations at each time point were found to display significant differences in the molecular mechanisms that are set in place to develop resistance, as summarized in Fig.  12. The multiazole-resistant strain 044Fluco45 exhibits the upregulation of genes encoding multidrug resistance (MDR) transporters CgCDR1, CgCDR2, and CgTPO1_2, decreased accumulation of azole drugs, and a likely gain-of-function (GOF) mutation in the multidrug transcription factor Pdr1. Although the point mutation observed in the PDR1 sequence, leading to a Y372C substitution, has not been described before, it is indeed very likely to constitute a GOF mutation, given that it is in the same position as the Pdr1 Y372N GOF mutation identified previously (31). Altogether, the results for strain 044Fluco45 are consistent with the recurrent observation that the development of GOF mutations in Pdr1, leading to the upregulation of drug efflux pumps, is the main mechanism of azole resistance acquisition in clinical isolates (8,9). Based on the evolutionary path observed in the transcriptomics analysis, our current model is that  Following C. glabrata Evolution to Azole Resistance Antimicrobial Agents and Chemotherapy the fluconazole-exposed population appears to be iteratively selected toward resistance at minimum cost, which appears to be, in the long term, the acquisition of Pdr1 GOF mutations, associated with drug efflux pump overexpression. Before reaching that optimal solution, the population goes through transcriptome-wide remodeling, likely reflecting the transient selection of more-fit subpopulations. When the Pdr1 GOF solution is reached by part of the population, these optimized cells are selected, leading to the dilution of other subpopulations until their disappearence. Interestingly, the 044Fluco21 and 044Fluco31 populations do not exhibit the typical molecular mechanisms related to azole resistance in C. glabrata but still exhibit increased MICs for all azole drugs, resistance to posaconazole and to both posaconazole and clotrimazole, respectively, and increased ability to limit the intracellular accumulation of [ 3 H]clotrimazole, compared to the 044 clinical isolate. In the 044Fluco21 posaconazole-resistant strain, the expression of cellular processes such as protein synthesis, cell cycle, and DNA damage response, which are apparently unrelated to azole resistance, is upregulated. Thus, although the slight upregulation of CgCDR1 in the 044Fluco21 strain may, at least partially, account for the posaconazole resistance phenotype, further characterization of the mechanisms of posaconazole resistance acquisition in C. glabrata are required.
The 044Fluco31 strain was found to display upregulation of the ERG11 gene; however, the concentration of ergosterol was found to remain constant in this strain. Assuming that the increased ERG11 gene expression results in increased Erg11 protein expression, this may at least prevent the decrease in the ergosterol content that fluconazole exposure is bound to induce and may thus decrease azole susceptibility by maintaining the Erg11/drug molecule ratio. Although upregulation of the ERG11 gene (33,34) and the augmentation of ergosterol levels (30) are associated with azole resistance in Candida albicans, in the case of C. glabrata, the expression level or amino acid substitutions of the ERG11 gene do not seem to correlate with azole resistance acquisition in the clinical setting (19,20,35). Given that there is no corresponding increase in ergosterol levels (as observed), the increased expression of ERG11 may only partially, not completely, explain the observed gain in azole resistance in strain 044Fluco31. This observation prompted us to analyze in more detail the molecular basis underlying the posaconazole and clotrimazole resistance exhibited by strains 044Fluco31. Strain 044Fluco31 was found to exhibit upregulation of several adhesinencoding genes, accompanied by an increased ability to adhere to other C. glabrata cells and to epithelial cells. Among the adhesin-encoding genes upregulated in the 044Fluco31 strain, we focused our research on three adhesins of the EPA family, encoded by the CgEPA1, CgEPA3, and CgEPA10 genes. Significantly, the expression of Epa3 was found to decrease C. glabrata susceptibility to azole drugs, directly or indirectly leading to decreased accumulation of azole drugs. These results indicate Epa3 as an important, though unexpected, player in azole resistance. Our current model is that the role of CgEpa3, and possibly that of other adhesins, in azole resistance, might be to protect the cells from the extracellular concentration of the drug by promoting cell aggregation. Interestingly, comparing the genome of an azole-susceptible C. glabrata isolate with that of an azole-resistant C. glabrata isolate showed a higher number of adhesin-like genes in the resistant isolate (36). As expected, CgEpa3 was also found to play a role in C. glabrata adhesion and biofilm formation, a finding consistent with the predicted role of CgEpa3 and its upregulation in C. glabrata biofilms in vitro (37). Altogether, the analysis of the evolution toward multiazole resistance of the 044 clinical isolate suggests that prolonged exposure to fluconazole progressively selects the subpopulation that evolves to higher resistance with lower costs, leading to what appears to be a unique response to fluconazole induction. Indeed, the final transcrip- Following C. glabrata Evolution to Azole Resistance Antimicrobial Agents and Chemotherapy tional profile reached by the 044Fluco45 strain gives evidence of the important role of Pdr1 GOF mutations and the activation of MDR transporters in this context. Nevertheless, in the path to full resistance, several other, eventually more subtle, mechanisms of azole resistance may be employed by the evolving population, including the overexpression of adhesin-encoding genes. This study highlights the role of one of these genes, CgEpa3, in azole drug resistance, further supporting the notion that azole resistance is a multifactorial process, composed of different molecular mechanisms that should be considered in the design of better-suited therapeutic strategies.
The VK2/E6E7 human epithelial cell line (ATCC CRL-2616) was used for adhesion assays. This cell line is derived from the vaginal mucosa of a healthy premenopausal female subjected to vaginal repair surgery and was immortalized with human papillomavirus 16/E6E7. Cells maintenance was achieved with keratinocyte-serum-free medium, containing 0.1 ng/ml human recombinant epidermal growth factor FIG 12 Current model for the mechanisms underlying the evolution of the 044 clinical isolate toward multiazole resistance. Under the timeline of prolonged fluconazole exposure, indicating the days at which resistance to each azole drug was achieved, the main biological processes found to be upregulated in each of the azole-resistant strains are highlighted. Below, the conclusions of the experimental results obtained are given, suggesting that while for 044Fluco21, drug resistance appears to rely mostly on decreased drug accumulation due to increased CgCdr1 expression, the 044Fluco31 strain displays increased drug tolerance due to increased cell-to-cell adhesion, and 044Fluco45 exhibits multiazole resistance due to the acquisition of a CgPdr1 GOF mutation leading to increased expression of CgCdr1 and CgCdr2.
In vitro induction of multiple azole resistance. Three randomly selected colonies of the 044 clinical isolate, exhibiting susceptibility to all azoles tested, was incubated in 10 ml of YPD medium overnight in a rotating drum at 150 rpm and 35°C. A 1-ml aliquot of this culture, containing 10 6 blastoconidia, was transferred to different vials, each containing 9 ml of culture medium with or without 16 mg/ml fluconazole, a concentration of the drug that corresponds to therapeutic levels in serum obtained during antifungal treatment (38), and was incubated overnight as described above. The following day, aliquots from each culture containing 10 6 blastoconidia were again transferred to fresh medium containing the same antifungal and were reincubated as described above. Each day, for the 80 days of the assay, a 1-ml aliquot from each subculture was mixed with 0.5 ml of 40% glycerol and was frozen at -70°C for later testing. To assess resistance stability, the resistant isolates obtained were subcultured daily in the absence of the drug for 30 days. Three colonies from each isolate were incubated in 10 ml drug-free YPD medium at 35°C and 150 rpm. The following day, aliquots were transferred to fresh medium. At each subculture, a 1-ml aliquot of the suspension was mixed with 0.5 ml of 40% glycerol, and the mixture was frozen at -70°C for further testing.
Drug susceptibility assays. The MIC values of each antifungal drug were determined according to the M27-A3 protocol and the M27-S4 supplement of the Clinical and Laboratory Standards Institute (CLSI) (39). Interpretative criteria for fluconazole and voriconazole were those of the CLSI: for fluconazole, a susceptible-dose-dependent (S-DD) MIC of Յ32 mg/ml and a resistance (R) MIC of Ն64 mg/ml; for voriconazole, a wild-type MIC of Յ0.5 mg/ml and a non-wild-type MIC of Ն1 mg/ml. Although susceptibility breakpoints have not yet been established for posaconazole or clotrimazole, strains inhibited by Յ4 mg/ml were considered to be susceptible to posaconazole or clotrimazole, respectively, considering that their breakpoints should be 4-fold higher in C. glabrata than in C. albicans, as is the case for fluconazole (40,41). Every 5 days of incubation, with or without the antifungal, MIC values were redetermined for fluconazole, voriconazole, posaconazole, and clotrimazole. Candida glabrata type strain CBS138 was used in each testing assay, as recommended.
The antifungal drug susceptibilities of the 044 clinical isolate and the derived azole-resistant strains (1), of the KUE100 parental strain and the derived ⌬cgepa1, ⌬cgepa3, and ⌬cgepa10 deletion mutants (2), and of the C. glabrata wild-type strain L5U1 harboring the pGREG576 vector or the pGREG576_PDC1_CgEPA3 plasmids (3) were evaluated by spot assays, as described previously (11).
Transcriptomic analysis. The C. glabrata 044 clinical isolate and the derived 044Fluco21 (pozaconazole-resistant), 044Fluco31 (pozaconazole-and clotrimazole-resistant), and 044Fluco45 (pozaconazole-, clotrimazole-, fluconazole-, and voriconazole-resistant) strains were harvested in the mid-exponential phase of growth in YPD medium. Three independent cultures from each strain were used for transcriptional profiling. RNA extraction was performed as described elsewhere (42). The quality and integrity of the purified RNA were confirmed using a bioanalyzer. The DNA chips used for this microarray analysis were manufactured by Agilent using a design for C. glabrata (43). The microarray was designed using eArray by Agilent Technologies, based on the annotation of C. glabrata CBS138 available at the Yeast Gene Order Browser in 2014 (44). cDNA synthesis, hybridization, and scanning were performed using protocols similar to those described in reference 43, except that hybridization was carried out using an Agilent hybridization oven at 65°C for 17 h at 100 rpm, according to a previously described protocol (42). Data were analyzed using the LIMMA package in Bioconductor (www .bioconductor.org), as described before (42). Each gene was represented by two probes spotted in duplicate, which were used separately to calculate the log fold change (FC) (Tables S1 to S3 at http://ibb.tecnico.ulisboa.pt/Cavalheiro_etal_SuplData.pdf). Only genes exhibiting a log 2 FC of Ͼ1 and a P value of Յ0.05 for at least one probe were selected for further analysis. Gene Ontology enrichment analysis was performed with the GOToolBox Web server (45) for each group of upregulated and downregulated genes, considering C. glabrata genes or their Saccharomyces cerevisiae homologs. Predictive analysis of the transcription factors controlling the observed transcriptional alterations was conducted using the PathoYeastract database (32).
Cloning of the C. glabrata CgEPA3 gene (ORF CAGL0E06688g). The pGREG576 plasmid from the Drag&Drop collection (46) was used to clone and express the C. glabrata open reading frame (ORF) CAGL0E06688g in S. cerevisiae, as described before for other heterologous genes (47). pGREG576 was acquired from Euroscarf and contains a galactose-inducible promoter (GAL1) and the yeast selectable marker URA3. The CgEPA3 gene was cloned in two sections, both generated by PCR, using CBS138 genomic DNA and the primers listed in Table 2. To enable expression of the EPA3 gene in C. glabrata, the GAL1 promoter was replaced by the constitutive C. glabrata PDC1 promoter. The PDC1 promoter DNA was generated by PCR using the primers listed in Table 2. The recombinant plasmids pGREG576_CgEPA3 and pGREG576_PDC1_CgEPA3 were obtained through homologous recombination in S. cerevisiae and were verified by DNA sequencing.
Disruption of the C. glabrata CgEPA1, CgEPA3, CgEPA9, CgEPA10, CgAWP12, and CgAWP13 (ORF CAGL0E06644g, CAGL0E06688g, CAGL0A01366g, CAGL0A01284g, CAGL0G10219g, and CAGL0H10626g) genes. The deletion of the CgEPA1, CgEPA3, CgEPA9, CgEPA10, CgAWP12, and CgAWP13 genes was carried out in the parental strain KUE100, using the method described in reference 48. The target genes were replaced, through homologous recombination, by a DNA cassette including the CgHIS3 gene. The pHIS906 plasmid including CgHIS3 was used as a template, and transformation was performed as described previously (48). The deletion of CgEPA3 in the 044Fluco31 background was carried out using the method described in reference 49. The gene was replaced through homologous Following C. glabrata Evolution to Azole Resistance Antimicrobial Agents and Chemotherapy recombination by a SAT1 flipper cassette. PCR was used to prepare the replacement cassettes and to verify recombination loci and gene deletions, using the primers listed in Table 2. Gene expression analysis. The levels of CgEPA1 and CgEPA3 transcripts in the azole-susceptible isolate 044 and the azole-resistant derived strains, as well as the levels of CgCDR1 transcripts in the wild-type strain KUE100 and its derived ⌬cgepa3 deletion mutant strain, and in strain L5U1 harboring the cloning vector pGREG576 or the CgEPA3 expression plasmid pGREG576_PDC1_CgEPA3, were assessed by quantitative real-time PCR. Total-RNA samples were obtained from cell suspensions harvested under control conditions, in the absence of drugs. Synthesis of cDNA for real-time RT-PCR experiments, from total-RNA samples, was performed using the Multiscribe reverse transcriptase kit (Applied Biosystems), following the manufacturer's instructions, and using 10 ng of cDNA per reaction. The RT-PCR step was carried out using SYBR green reagents. Primers for the amplification of the CgEPA1, CgEPA3, CgCDR1, and CgACT1 cDNA were designed using Primer Express software (Applied Biosystems) ( Table 2). The RT-PCRs were conducted in a thermal cycler block (7500 Real-Time PCR system; AGAGCCGTCTTCCCTTCCAT CgACT1_Rev TTGACCCATACCGACCATGA Applied Biosystems). The CgACT1 mRNA level was used as an internal control. The relative values obtained for the wild-type strain under control conditions were set at 1, and the remaining values are presented relative to that control.
[ 3 H]clotrimazole accumulation assays. [ 3 H]clotrimazole transport assays were carried out as described before (11). To estimate the accumulation of clotrimazole (intracellular/extracellular [ 3 H]clotrimazole), yeast cells were grown in BM medium until the mid-exponential phase and were harvested by filtration. Cells were washed and resuspended in BM medium to obtain dense cell suspensions (optical density at 600 nm [OD 600 ], 5.0 Ϯ 0.2, equivalent to approximately 2.2 mg [dry weight] ml Ϫ1 ). After a 5-min incubation at 30°C, with agitation (150 rpm), 0.1 M [ 3 H]clotrimazole (American Radiolabelled Chemicals; 1 mCi/ml) and 30 mg/liter of unlabeled clotrimazole were added to the cell suspensions. The intracellular accumulation of labeled clotrimazole was followed for 30 min, as decribed elsewhere (11). To calculate the intracellular concentration of labeled clotrimazole, the internal cell volume (Vi) of the exponential cells was considered constant and equal to 2.5 l (mg [dry weight] Ϫ1 ) (50).
Quantification of total cellular ergosterol. Ergosterol was extracted from cells using the following method adapted from reference 51, as described before (52). Cells were cultivated in YPD medium with orbital agitation (250 rpm) until stationary phase, harvested by centrifugation, and resuspended in methanol. Cholesterol (Sigma) was added as an internal standard to estimate the ergosterol extraction yield. The extracts were analyzed by high-pressure liquid chromatography (HPLC) with a 250-mm by 4-mm C 18 column (LiChroCART Purospher STAR RP-18, end-capped, 5 mm) at 30°C. Samples were eluted in 100% methanol at a flow rate of 1 ml/min. Ergosterol was detected at 282 nm with a retention time of 12.46 Ϯ 0.24 min, while cholesterol was detected at 210 nm with a retention time of 15.36 Ϯ 0.35 min. Results are expressed in micrograms of ergosterol per milligram (wet weight) of cells.
Biofilm quantification. C. glabrata strains were tested for their capacity for biofilm formation by use of the crystal violet method. For that, the C. glabrata strains were grown in SDB medium and harvested by centrifugation at mid-exponential phase. The cells were inoculated with an initial OD 600 of 0.05 Ϯ 0.005 in 96-well polystyrene microtiter plates (Greiner) in either SDB (pH 5.6) or RPMI (pH 4) medium. Cells were cultivated at 30°C for 15 Ϯ 0.5 h or 4 Ϯ 0.5 h with mild orbital shaking (70 rpm). After the incubation time, each well was washed three times with 200 l of deionized water to remove cells not attached to the biofilm matrix. Then 200 l of a 1% crystal violet (Merck) alcoholic solution was used to stain the biofilm present in each well. Following 15 min of incubation with the dye, each well was washed with 250 l of deionized water. The stained biofilm was eluted in 200 l of 96% (vol/vol) ethanol, and the absorbance of each well was read in a microplate reader at a wavelength of 590 nm (SPECTROstar Nano; BMG Labtech).
Human vaginal epithelial cell adherence assay. For the adhesion assays, VK2/E6E7 human epithelial cells were grown and inoculated in 24-well polystyrene plates (Greiner) with a density of 2.5 ϫ 10 5 cells/ml a day prior to use. Additionally, C. glabrata cells were inoculated with an initial OD 600 of 0.05 Ϯ 0.005 and were cultivated at 30°C for 16 Ϯ 0.5 h with orbital shaking (250 rpm) in YPD medium. In order to initiate the assay, the culture medium of mammalian cells was removed and replaced by new culture medium in each well, and, subsequently, Candida glabrata cells were added to each well, with a density of 12.5 ϫ 10 5 CFU/well, corresponding to a multiplicity of infection (MOI) of 10. Then the plate was incubated at 37°C under 5% CO 2 for 30 min. Afterwards, each well was washed 3 times with 500 l of 1ϫ phosphate-buffered saline (PBS) (pH 7.2), followed by the addition of 500 l of 0.5% Triton X-100 and incubation at room temperature for 15 min. The cell suspension in each well was then recovered, diluted, and spread onto agar plates to determine the CFU count, which represents the proportion of cells adherent to the human epithelium.
Accession number(s). The data sets were deposited at the Array Express Database with reference number E-MTAB-6787.