Upregulation of the Adhesin Gene EPA1 Mediated by PDR1 in Candida glabrata Leads to Enhanced Host Colonization

Candida glabrata is an important fungal pathogen in human diseases and is also rapidly acquiring drug resistance. Drug resistance can be mediated by the transcriptional activator PDR1, and this results in the upregulation of multidrug transporters. Intriguingly, this resistance mechanism is associated in C. glabrata with increased virulence in animal models and also with increased adherence to specific host cell types. The C. glabrata adhesin gene EPA1 is a major contributor of virulence and adherence to host cells. Here, we show that EPA1 expression is controlled by PDR1 independently of subtelomeric silencing, a known EPA1 regulation mechanism. Thus, a relationship exists between PDR1, EPA1 expression, and adherence to host cells, which is critical for efficient virulence. Our results demonstrate that acquisition of drug resistance is beneficial for C. glabrata in fungus-host relationships. These findings further highlight the challenges of the therapeutic management of C. glabrata infections in human patients.

C andida glabrata is an opportunistic pathogen that is able to cause invasive infection in susceptible patients. Because C. glabrata is an opportunistic pathogen, host susceptibility is critical to initiate active infection, and over the last half century, the human population at risk has increased (1). This trend is based on an increased number of immunosuppressed patients. Several factors, such as the occurrence of different cancers, the use of organ transplants, the persistence of the AIDS pandemic, the use of broad-spectrum antibiotics, and the expansion of the elderly population, have contributed to the increase of this patient population (2). C. glabrata deploys several distinct factors to infect its host. C. glabrata is less pathogenic than Candida albicans (3); however, it is able to persist in experimental infection models for long periods by eliciting only mild immune responses (4,5). An important virulence attribute of C. glabrata is its capacity to adhere to host tissues. The genome of C. glabrata contains a high number of genes (around 60 in the CBS138 genome) encoding predicted glycosylphosphatidylinositol (GPI)-anchored adhesin-like cell wall proteins (6). Adhesins of C. glabrata are distributed in several groups. The EPA (epithelial adhesion) subfamily is crucial for interaction with host cells. EPA1 is the first member of this family and was identified by Cormack et al. (7) as a major mediator of adherence to epithelial cells. Interestingly, a majority of C. glabrata adhesin genes exhibit subtelomeric locations and are thus under the control of transcriptional silencing by chromatin-based and NADdependent regulation mechanisms (8). Since C. glabrata is a nicotinic acid (NA) auxotroph, NAD-limiting environments release adhesin genes from transcriptional repression when they have a subtelomeric location. Interestingly, the urinary tract is NAD poor and is thus a favorable host environment for adhesin expression. As a matter of fact, urinary tract infections (UTIs) caused by C. glabrata are very common (8).
In recent years, C. glabrata has emerged as an important fungal pathogen in many regions of the world (9). It is believed that antifungal drug usage and the emergence of antifungal drug-resistant C. glabrata isolates are partially responsible for this observation. Currently available antifungal agents for treatment of candidiasis are azoles, candins, and polyenes. Azoles and polyenes such as amphotericin interfere with sterols in these species at different levels. While azoles inhibit an important step in ergosterol biosynthesis (14␣-lanosterol demethylation), amphotericin B forms a complex with ergosterol which compromises fungal viability. Candins target cell wall biosynthesis by inhibiting fungus-specific glucan synthesis. C. glabrata is intrinsically less susceptible to azole antifungals than is C. albicans (10). In addition, when exposed to this class of agents, C. glabrata rapidly develops resistance. It is believed that 20% of clinical isolates are azole resistant (11). Most azole-resistant isolates upregulate ATP-binding cassette (ABC) transporter genes, including C. glabrata CDR1 (CgCDR1) and CgCDR2, which are important transporters for the development of azole resistance (12). This upregulation is mediated by PDR1, a transcriptional activator of the zinc cluster transcription factor family with a Zn(2)Cys(6) domain. Mutations in PDR1 are called gain-of-function (GOF) mutations, since they are responsible for constitutive high expression of ABC transporters in resistant isolates. Multiple GOF mutations have been described, and they occur at different functional domains of the protein (12)(13)(14)(15). PDR1 interacts with the Mediator complex which bridges the activator and the transcriptional machinery in C. glabrata (16). Pdr1 regulates other genes in C. glabrata due to the presence of a pleiotropic drug resistance element (PDRE), a predicted Pdr1 binding target sequence, in the promoters of target genes (17). PDR1 GOF mutations not only regulate ABC transporter genes but also target other genes responsible for other cellular processes. Strikingly, recent reports document increased proportions of multidrug-resistant C. glabrata isolates (i.e., isolates resistant to at least two different classes of agents) (18)(19)(20). It has been established that about 3 to 11% of azole-resistant isolates in the United States also exhibit candin resistance (21).
The selection of drug-resistant isolates by drug exposure can have a cost (or fitness cost) for the yeast. The cost of antifungal resistance is that strains may be less competitive than wild-type isolates, especially when the drug selection is removed. This potential decrease in fitness may indeed compromise their virulence. In C. glabrata, however, a few studies have addressed the relationship between acquisition of drug resistance and changes in virulence traits and found evidence of increased virulence. As reported by Poláková et al. (22) and Ahmad et al. (23), C. glabrata can form minichromosomes containing genes involved in drug resistance while colonizing the human host, which may selectively favor the expansion of this yeast in vivo. The effects of PDR1 GOF mutations on in vivo fitness and on virulence have been addressed in C. glabrata. It was discovered that PDR1 GOF mutations, in addition to their role as mediators of azole resistance, could enhance virulence of C. glabrata compared to wild-type isolates in mice (12). Enhanced virulence observed in mice was accompanied not only by elevated fungal loads in infected organ but also by treatment failure with azoles. Moreover, the change in virulence was paralleled with gain of in vivo fitness of strains carrying PDR1 GOF mutations. The dogma that development of drug resistance is associated with fitness costs was therefore challenged by the results of Ferrari et al. (12), given that azole resistance had rather a fitness benefit in vivo. CgCDR1, the ABC transporter involved in azole resistance, and the open reading frame (ORF) CAGL0M12947g (PUP1, or PDR1-upregulated gene) are commonly upregulated by PDR1 GOF mutations, and this effect was thought to explain enhanced virulence (24). While deletions of CgCDR1 and PUP1 decreased virulence of C. glabrata in a mouse model of infection, PDR1-independent overexpression resulted in intermediate virulence phenotypes. This suggested that other C. glabrata factors were responsible for the gain of virulence observed in drug-resistant isolates.
To gain insights about these additional factors, we further addressed the role of PDR1 mutations in the interaction with several mammalian cell types, including murine bone marrow-derived macrophages (BMDMs), human acute monocytic leukemia cell line (THP-1)-derived macrophages, and different epithelial cell lines (25). We showed that, interestingly, PDR1 GOF mutations led to decreased adherence to macrophages and to decreased uptake by the same cells. The interaction with epithelial cells revealed, however, an opposite trend. This suggested that PDR1 GOF mutations may help C. glabrata to colonize epithelial host cells by increasing adherence to epithelial cell layers. These data highlighted that PDR1 GOF mutations modulate the interaction with host cells with the consequence of increasing virulence.
Here, we show that specific adhesion attributes of C. glabrata are regulated by PDR1. Among them, the adhesin gene EPA1 contributes to enhanced adhesion to epithelial mammalian cell lines when PDR1 mutations are causing azole resistance. This PDR1mediated effect may help this yeast species to be a more efficient pathogen.

Modulation of the interaction with host cells is PDR1
GOF mutation and C. glabrata strain dependent. In a previous report, we showed that PDR1 hyperactivity in C. glabrata mediates evasion from phagocytosis by murine and human macrophages (25). This was reproduced across three different PDR1 GOF mutations (L280F, R376W, and T588A) expressed in the azole-susceptible C. glabrata oropharyngeal clinical isolate DSY562. Surprisingly, we also found that the GOF mutation L280F led to increased adherence of DSY562 to three different epithelial cell lines (25). To follow up, we first addressed the reproducibility of these effects in different C. glabrata strain backgrounds, including the reference strain CBS138 (ATCC 2001), whose genome was sequenced and is publicly available (26); BG2, a well-known genetic background used in several studies (7,8,27,28); and DSY2235, another azole-susceptible oropharyngeal clinical isolate from our strain collection (12). We introduced the same PDR1 alleles (the wild-type PDR1 allele from clinical isolate DSY562, PDR1 WT , and the hyperactive allele from matched isolate DSY565, PDR1 L280F ) in these three additional C. glabrata strains and repeated competition phagocytosis assays using RAW 264.7 macrophage-like cells. Mixed suspensions of a given genetic background bearing either PDR1 WT or PDR1 L280F (and additionally expressing either green fluorescent protein [GFP] or red fluorescent protein [RFP], in order to distinguish between strains) were used to inoculate the macrophage cultures. Representative microscopy pictures are shown in Fig. S2 in the supplemental material. Our results confirmed the differences between DSY562 expressing the two alleles that were previously observed using primary murine macrophages and THP-1-derived human macrophages (Fig. 1A). However, this effect depended on the C. glabrata genetic background and was reproduced only by strain DSY2235 and not by CBS138 or BG2 (Fig. 1A). We next tested the adherence of the constructed strains to CHO-Lec2 epithelial cells and found that the increase in adherence was reproduced in all tested strain backgrounds (Fig. 1B). Our previous work showed that other tested PDR1 alleles with GOF mutations R376W and T588A expressed in DSY562 reproduced the differences in the interaction with macrophages (25). Here, we addressed whether they also mediated increased adherence to CHO-Lec2 cells. Our results indicate that these GOF mutations could mediate a similar increase in adherence when expressed in DSY2235 ( Fig. 1D) but not in DSY562 (Fig. 1C).
PDR1 hyperactivity leads to adhesin gene overexpression. In light of these PDR1-mediated changes in the interaction of C. glabrata with mammalian innate immune cells and epithelial cells, we reasoned that PDR1 might regulate the expression of cell wall proteins with a role in adherence to host cells. We combined information from a previous genome-wide microarray study of PDR1 hyperactivity in DSY562 performed in our laboratory (24), together with other transcriptomic studies in azoleresistant C. glabrata strains (14,29,30) and available functional knowledge on C. glabrata cell wall adhesins, in order to compile a small group of candidate genes for further testing. They included EPA1, EPA7, EPA12, and PWP4. We performed real-time quantitative PCR (qPCR) to test the impact of PDR1 hyperactivity on the expression of these adhesins. Adhesin genes, particularly members of the EPA family, are generally difficult to target by qPCR due to the presence of tandem repeat sequence regions and a relatively high sequence similarity between different genes. We retrieved publicly available C. glabrata adhesin gene sequences from strains CBS138 and BG2 and sequenced EPA1 from DSY562 and DSY2235. After qPCR optimization, efficient relative quantification of adhesin transcript levels could be achieved, including mutant and overexpression strains (illustrated for EPA1 in Fig. S3 in the supplemental material). Our qPCR results revealed that the GOF mutation L280F mediated overexpression (at least 2-fold) of EPA1 in all strain backgrounds ( Fig. 2A to D). Other GOF mutations did not lead to EPA1 overexpression in all strain backgrounds. Strikingly, comparison of transcript levels with adherence phenotypes for any given strain revealed a perfect correspondence between EPA1 overexpression and increased adherence to epithelial cells ( Fig. 1B to D and 2A to D). Transcript levels of EPA7 and EPA12 were either increased or not affected, with no obvious association with increased adherence to epithelial cells. PWP4 expression was not affected by PDR1 hyperactivity (Fig. 1B to D and 2A to D).
Pdr1 regulates EPA1 expression. The association of EPA1 overexpression with increased adherence to epithelial cells in C. glabrata strains with PDR1 GOF mutations first indicated a role for EPA1 as a Pdr1-mediated effector of the adherence phenotypes. Our results showed that PDR1 regulates the expression of EPA1, but the question remains, however, whether this molecular regulation is direct or indirect. Interestingly, we noticed the presence of a putative PDRE in the promoter of EPA1 (TCCACGCA, position starting Ϫ571 upstream of the EPA1 start codon [ Fig. 3A]) (30) matching the PDR1 binding site (TCCACGGA) (17). The deletion of 72 bp centered on the putative PDRE in the EPA1 promoter in a DSY562-derived strain containing the hyperactive PDR1 L280F allele led to expression of EPA1 being decreased by approximately 4-fold compared to a wild-type EPA1 promoter (Fig. 3B). Interestingly, data published by Paul et al. (17) reporting the genomic binding sites of Pdr1 suggest that the EPA1 promoter is targeted by Pdr1 (see Fig. S4A in the supplemental material). Chromatin immunoprecipitation (ChIP) with anti-Pdr1 on EPA1 showed, however, that Pdr1 does not bind tightly the EPA1 promoter (see Fig. S4B), thus raising the possibility of an indirect effect of Pdr1 on EPA1.
PDR1 hyperactivity-mediated increase in adherence to epithelial cells is dependent on EPA1. The data described above hint at a role for EPA1 in the PDR1-mediated modulation of the interaction of C. glabrata with host epithelial tissues. In order to confirm this hypothesis, we inactivated EPA1 in DSY562-derived strains and repeated the competition assays for adherence to epithelial CHO-Lec2 cells. As shown in Fig. 4A, the absence of EPA1 in DSY562 with a PDR1 L280F allele eliminated the increased adherence compared to the strain harboring a wild-type PDR1 allele. This epa1Δ strain showed no difference in adherence to epithelial cells from the strain harboring the wild-type EPA1 and PDR1 alleles. Reversion of EPA1 rescued the increase in adherence (Fig. 4A). C. glabrata is often found causing infection in nicotinic acid (NA)-deficient host environments (notably in the urinary tract). This host niche is known to lead to derepression of EPA1 and other subtelomeric adhesin genes (8), thus signifying that EPA1 can be upregulated even in the absence of PDR1 GOF mutations. We recapitulated this situation by addressing adherence of C. glabrata grown either in complete medium or in NA-limited medium (5% of the regular NA concentration). When the wild type was grown under NA limitation, i.e., PDR1-independent adhesin expression-inducing conditions (as shown by qPCR for EPA1 [see Fig. S5 in the supplemental material]), adherence to epithelial cells matched that of the strain with the hyperactive PDR1 L280F allele grown under a normal NA concentration (Fig. 4B). Nevertheless, the increased adherence mediated by PDR1 hyperactivity was still detected when both strains were grown under NA limitation (Fig. 4B). Taken together, our data confirm that EPA1 mediates the difference in adherence between C. glabrata strains with wild-type and hyperactive PDR1 alleles.
Impact of EPA1 on adherence to epithelial cells is C. glabrata strain dependent. Inactivation of EPA1 in DSY562 with the hyperactive PDR1 L280F allele had an unexpectedly low impact on adherence to epithelial cells (reduction of adherence by less than 30%), thus making it almost indistinguishable from the wild type. Inactivating EPA1 in DSY562 had a similar low impact on adherence (Fig. 4A). This suggests that EPA1 was not required to maintain basal adherence levels in this strain. This is in contrast to the large adherence reduction (Ͼ90%) reported previously in BG2 (7). We tested the BG2 epa1Δ strain and observed the same large reduction in adherence (data not shown), thus ruling out inappropriate test conditions. This suggests an inherently low contribution of Epa1 for basal adherence of DSY562 to epithelial cells. We then inactivated EPA1, tested the adherence of DSY2235 containing the hyperactive PDR1 L280F allele, and compared the results to the wild type. Adherence of the DSY2235 epa1Δ mutant dropped by more than 75% compared to the wild type (Fig. 4C), reminiscent of observations in BG2. While performing these experiments, we noticed that basal EPA1 expression levels were highly variable between different C. glabrata strain backgrounds. Interestingly, these basal expression levels were proportional to the degree of adherence to epithelial cells. However, our results indicated that DSY562 stands as an exception. Even if this yeast expresses EPA1 at the lowest levels among the tested isolates of this study, it was still second in terms of adherence to epithelial cells (Fig. 5). This, together with the low impact of EPA1 on adherence, suggests that, unlike other C. glabrata strains (which rely heavily on Epa1 for adherence to epithelial cells), the DSY562 strain background relies primarily on alternative adhesins for basal adherence. Nevertheless, and even if PDR1 hyperactivity mediates upregulation of additional adhesins in DSY562, PDR1 hyperactivity still mediates increased adherence to epithelial cells through EPA1 upregulation in this strain background.
PDR1 hyperactivity mediates an EPA1-dependent increase in colonization of the murine urinary tract. In order to determine the biological significance of PDR1-mediated regulation of EPA1 expression, we employed a murine model of urinary tract infection (UTI). Besides strains expressing either PDR1 WT or PDR1 L280F and epa1Δ strains, we also addressed the impact of PDR1-independent EPA1 overexpression. To this end, we constructed DSY562-derived strains expressing EPA1 from an episomal plasmid under the control of a strong constitutive promoter (the Saccharomyces cerevisiae PGK1 [ScPGK1] promoter). The resulting strains displayed a strong increase in EPA1 transcript levels, irrespective of the PDR1 allele (see Fig. S6A in the supplemental material). This was accompanied by an increase in the measured adherence to epithelial cells to close to 100% of the inoculum under our test conditions (see Fig. S6B). We tested all these genetic variants from the DSY562 strain background in the UTI model, as well as a subset of them in the DSY2235 strain background. The results illustrate an increased ability of strains expressing PDR1 L280F to colonize the bladder and then to reach the kidneys of infected mice compared to strains with PDR1 WT (Fig. 6A to D). Just as in the case of adherence to epithelial cells, these phenotypes depended on EPA1, since PDR1 L280F epa1Δ strains showed CFU counts indistinguishable from those of the PDR1 WT strains in both DSY562 and DSY2235. The ScPGK1 promoter-driven EPA1 overexpression in DSY562 led to the highest observed CFU counts, with no statistically significant difference between PDR1 alleles ( Fig. 6A and C). With regard to epa1Δ mutants in DSY562 with PDR1 L280F , even if they show a decrease in tissue burden in infected animals, they still show a higher burden than the epa1Δ PDR1 WT strain. This suggests an additional, EPA1-independent effect of PDR1 hyperactivity in the colonization of host urinary tract organs by C. glabrata.

DISCUSSION
Effect of PDR1 on EPA1 expression. We have previously established a link between PDR1 hyperactivity and enhanced virulence in C. glabrata with mouse models of infection (12). A follow-up work showed that a decreased phagocytosis by macrophages and an increased adherence to host epithelial cells contribute to these phenotypes (25). Here, we report that regulation of these phenotypes by PDR1 seems to be complex. PDR1 hyperactivity per se does not guarantee differences in the interactions with host cells, which additionally depend on the C. glabrata genetic background and specific PDR1 GOF mutations. Changes in phagocytosis by macrophages depend on the C. glabrata strain background, as phagocytosis of two of the four tested strains was not affected by replacement of the wild-type PDR1 allele by PDR1 L280F . Further work will be required to explain this observation. The interaction with epithelial cells, on the other hand, was affected in all strain backgrounds but depended on the specific GOF mutation. GOF mutations have profound and distinct effects on the C. glabrata transcriptome, and not all GOF mutations will lead to upregulation of EPA1 (24). Until now, the molecular basis behind the transcriptome heterogeneity due to PDR1 GOF mutations has not been elucidated. Among possible explanations, PDR1 may recruit different cofactors or additional zinc cluster transcriptional regulators to form heterodimers, which may depend on the type of GOF mutation, thus shaping differences in target binding domains. On the other hand, PDR1 may be subjected to selective degradation (for example, ubiquitylation) with different kinetics depending on the nature of GOF mutations, thus creating distinct transcriptional complexes and distinct transcriptional profiles. Transcriptional control by ubiquitylation is known for yeast transcription factors and especially when they are in an active state (31). Last, PDR1 is known to interact with subunits of the Mediator complex, which is composed of a sophisticated architecture of proteins (32). Specific GOF mutations may interfere with the Mediator complex and alter transcription specificity and efficiency. Our data confirm that GPImodified cell wall adhesin genes are part of the PDR1 extensive regulon and show that PDR1 modulates the interaction with epithelial cells through the transcriptional regulation of adhesins, among which EPA1 was found to play a major role. In fact, increased adherence to epithelial cells in strains with hyperactive PDR1 alleles strictly depends on EPA1. The question of whether EPA1 regulation by PDR1 is direct or indirect remains open. While deleting the putative PDRE on the promoter of EPA1 resulted in a decrease in EPA1 expression, chromatin immunoprecipitation (ChIP) experiments failed to detect a direct binding of Pdr1 to the EPA1 promoter. It is possible that the subtelomeric location of EPA1 makes Pdr1 difficult to access experimentally, due to the presence of other factors that are part of the subtelomeric silencing complex. These factors include Sir2 and other members (HST1 to -4) of the bulky subtelomeric silencing protein complex (33). This might conceivably lead to a false-negative ChIP result, but more evidence is required to solve this question. SIR2 downregulation is known to mediate derepression of EPA1 (33). Genome-wide microarray data previously gathered by our laboratory did not involve SIR2 downregulation in DSY565 (the clinical strain containing the PDR1 L280F allele) as a possible mediator of increased EPA1 expression in azoleresistant isolates (24). As we mentioned above, PDR1 may recruit the ubiquitylation machinery, which itself might be able to modify histones and therefore influence subtelomeric silencing and consequently EPA1 expression. Ubiquitylation as an indirect regulatory mechanism was described in S. cerevisiae for ubiquitin-conjugating enzyme Rad6 and the histone H2B (34). This working model is under investigation in our laboratory.
Challenges in addressing EPA1 expression. Adhesins have been notably absent from previous genome-wide studies of the Pdr1 regulon generally focused on analysis of azole drug resistance. The reason for this may be that adhesin genes are particularly difficult to analyze. Adhesin genes are characterized by the presence of megasatellites (tandemly repeated sequences with variable size), high intraspecies sequence variability, and high intergenic sequence similarity within the EPA family (35). These characteristics make these genes difficult to analyze in genome-wide transcriptional studies. In previous microarray experiments, EPA gene transcript signals may have been largely undetected and/or amalgamated with different transcript signals in the same family. A notable exception is the study by Caudle et al. (30), in which EPA1 was upregulated by a specific PDR1 GOF mutation (K274N). Interestingly, this work also highlighted a second PDR1 GOF mutation (L946S), which did not result in EPA1 upregulation and was thus in agreement with our observations on the different effects of GOF mutations on EPA1 expression. Even in RNA sequencing experiments, adhesin genes may pose important challenges to mapping of short reads, which may exclude them from further analysis during quality control steps. Previous studies by the Cormack laboratory have mostly relied on radioactive probe-based S1 nuclease protection assay for specific and sensitive relative quantification of EPA1 transcript levels (8,28,36,37). Here, we have developed and used an alternative qPCR method. We have used strains harvested in mid-log phase following dilution of overnight saturated cultures and grown for two generations in rich medium. Testing EPA1 expression under these conditions requires a strict attention to growth conditions to ensure reproducibility. This is especially relevant since EPA1 expression changes rapidly during in vitro growth. It is transiently induced upon dilution in fresh growth medium under standard laboratory growth conditions and is then transcriptionally repressed by subtelomeric silencing (28).
C. glabrata genetic backgrounds and EPA1-related phenotypes. EPA1 has long been considered the major adhesin of C. glabrata, enabling most of the capacity for adherence to epithelial cells, even though C. glabrata contains more than 20 EPA-like genes and other types of adhesin-like genes (6). We showed here that virulence-enhancing PDR1 hyperactivity depends on EPA1 to mediate increased adherence to epithelial cells. There are intraspecies differences in homogeneity of EPA1 expression. Two C. glabrata groups of isolates have been identified, including the CBS138-like group of strains, in which EPA1 expression is homogeneous in yeast cell populations, and BG2-like strains, in which expression varies between individual cells (38). These patterns also seem to be associated with differences in expression ratios between noninducing and inducing (NA limitation) conditions (38). DSY562 belongs to the CBS138-like group, as it shows homogeneous expression (data not shown). In this work, we also noticed intraspecies differences in basal levels of EPA1 expression: transcript levels are highly variable between different C. glabrata strains (Fig. 5). Furthermore, there is yet another layer of complexity, since some C. glabrata strains are independent of EPA1 for their high basal adherence to epithelial cells. While BG2 and DSY2235 depend almost exclusively on EPA1 for adherence, DSY562 displays basal adherence levels independent of EPA1. In the DSY562 background, EPA1 plays a role only in increased adherence. We have indications that the DSY562 genome contains a higher number of adhesin genes than BG2 or CBS138 (L. A. Vale-Silva and D. Sanglard, unpublished data). This feature may explain the relatively low dependence of this strain on EPA1. Interestingly, a recent study showed that C. glabrata isolates of clinical origins may harbor distinct patterns of cell wall adhesins contributing to the intrinsic adherence capacities of such strains (39). This reinforces the notion that C. glabrata isolates may differ between each other in their repertoire of adhesins by either differential expression and/or the presence of distinct genomic patterns.
Relevance of EPA1 in the UTI model. We confirmed using the UTI model that EPA1 is important for in vivo colonization of the bladder and kidneys. It is also crucial for PDR1-mediated increased colonization. This likely contributes to enhanced virulence. Previous data from the Cormack laboratory found no influence of an epa1Δ mutant in a vaginal and gastrointestinal infection model (7,40). Later, the same laboratory found lower colonization of the mouse bladder by a mutant lacking three adhesion genes, including EPA1, EPA6, and EPA7 (8,40). We observed here that the single EPA1 deletion is sufficient for decreasing C. glabrata fungal loads in the UTI mouse model. One reason for this discrepancy could be due to differences in genetic backgrounds of the investigated yeasts (BG2 versus DSY562/2235) and/or of the mice. It is likely that several other adhesins may play a role in vivo rather than in vitro. For example, deletion of EPA1 in DSY2235 has a higher impact on adherence in vitro than on bladder/kidney colonization in vivo. On the other hand, overexpressing EPA1 alone, even in the presence of PDR1 WT , is enough to increase CFU to levels that may even reach saturation of the model.
In conclusion, we found that some adhesins, including EPA1, are regulated by PDR1. Elevated expression of EPA1 is the major mechanism that leads to increased adherence to epithelial cells in strains bearing hyperactive PDR1 alleles. EPA1 has a clear impact in vivo and allows colonization of the bladder and kidneys in the UTI murine model. EPA1 is likely to be an important contributor to increased virulence of azole-resistant C. glabrata strains with GOF PDR1 alleles as we earlier published.

MATERIALS AND METHODS
C. glabrata strain culture and growth media. All C. glabrata strains used in this study are listed in Table S1 in the supplemental material. Strains with the prefix SFY are from previously published collections (12,24). All strains were stored in 20% glycerol stocks at Ϫ80°C and cultured in either YPD (1% yeast extract, 2% peptone, 2% D-glucose) rich medium or appropriate selective media at 30°C. For solid media, 2% agar was added. Selective medium for growth of transformed strains was YPD containing either 200 g/ml of nourseothricin (clonNAT; Werner BioAgents, Germany) or 600 g/ml of hygromycin B (PAA Laboratories, Austria). Additionally, YNB minimal medium (0.67% yeast nitrogen base plus 2% glucose) with appropriate amino acids and bases and without uracil was used to select uracil prototrophs. To obtain sensitive derivatives after recycling of the dominant selection marker cassettes, resistant strains were incubated for a minimum of 4 h in YCB-BSA medium (23.4 g/liter yeast carbon base and 4 g/liter bovine serum albumin; pH 4.0), and between 100 and 200 CFU was plated on YPD agar plates containing either 20 g/ml of nourseothricin or 200 g/ml of hygromycin B. YPD containing 30 g/ml of fluconazole (Sigma) was used when required. Escherichia coli DH5␣ was used as a host for plasmid construction and propagation. E. coli DH5␣ was grown in Luria-Bertani broth or on Luria-Bertani agar plates, supplemented with 0.1 mg/ml of ampicillin (AppliChem GmbH, Germany) when required.
Plasmids and C. glabrata strain constructions. All yeast transformations were performed using an adapted lithium acetate (LiAc) procedure (41). Strains expressing fluorescent proteins were constructed using targeted gene disruption of CgURA3, followed by transformation with episomal plasmids complementing uracil auxotrophy as previously described (25). Inactivation of URA3 in the DSY2235 strain background led to a detectable growth defect even after episomal URA3 complementation. There was no difference between strains expressing different PDR1 alleles and different fluorescent proteins (see Fig. S1 in the supplemental material). Other strain backgrounds had no detectable change after deletion of URA3.
PDR1 deletion and replacement of PDR1 alleles were performed using the recyclable dominant selection marker SAT1 (42), encoding nourseothricin resistance (SAT r ) as previously described (12).
To delete the putative PDRE on positions Ϫ571 to Ϫ564 relative to the EPA1 start codon (a possible Pdr1 binding site), a 1-kb EPA1 5=-flanking sequence was amplified by PCR using primers EPA1-KpnI-2 (5=-GCAGGGTACCAAAAAGAACATC-3=) and EPA1-NcoI (5=-GTCACCATGGAATAGAGTGGATCTGGTACTTTG-3=). The PCR product was digested by KpnI and NcoI and ligated to a KpnI/NcoI fragment from pVS43, thus generating pVS47. This makes use of a naturally occurring NcoI restriction site 35 bp downstream of the putative PDRE and generates a plasmid carrying a deletion of 72 bp (between positions Ϫ600 and Ϫ527) centered on the putative PDRE on the 5= UTR of EPA1. pVS47 was digested by SacII and KpnI and used to transform epa1Δ C. glabrata strains.
All constructed plasmids were confirmed by DNA sequencing. Constructed strains were confirmed by sequencing or PCR. EPA1 deletions and reversions were confirmed by Southern blotting (data not shown).
Phagocytosis assays. Competition phagocytosis assays were performed as previously described (25). Briefly, harvested bone marrow-derived macrophages (BMDMs) were suspended in complete Iscove modified Dulbecco medium (IMDM) at a density of 3.0 ϫ 10 5 cells/ml. RAW 264.7 macrophage-like cells were suspended in complete DMEM at a density of 1.0 ϫ 10 5 cells/ml in complete IMDM. Macrophage suspensions were transferred to 24-well plates in a volume of 1 ml per well on top of round cover slides and incubated overnight at 37°C in a humid atmosphere with 5% CO 2 . Macrophages were then infected with 1:1 mixtures of 3.0 ϫ 10 5 yeast cells of each of the two C. glabrata strains in 100 l of phosphate-buffered saline solution (PBS). The plates were centrifuged for 1 min at 200 ϫ g and incubated for 30 min at 37°C in the presence of 5% CO 2 . Cocultures were washed to remove nonmacrophage-associated yeasts, and the round cover slides with adherent cells were stained with 100 g/ml of calcofluor white for 10 min. The cover slides were mounted onto microscopy slides and observed using a Zeiss Axioplan 2 epifluorescence microscope. Images were recorded using a Visitron Systems HistoScope camera and VisiView imaging software (Visitron Systems, Puchheim, Germany).
UTI model. Urinary tract infection (UTI) experiments were performed according to a previously described model (46). Briefly, for tissue burden experiments, each C. glabrata strain was grown in 10 ml of YPD broth under agitation for 18 h at 37°C. After growth, cells were centrifuged, washed and resuspended in 10 ml of sterile PBS, and then adjusted to reach a concentration of 5 ϫ 10 8 cells/ml. For each strain, a group of 10 isoflurane-anesthetized female BALB/c mice were infected via intraurethral catheterization (polyethylene catheter, about 4 cm long; outer diameter, 0.61 mm; Becton, Dickinson, Sparks, MD) using 100 l of the corresponding C. glabrata cell suspension for each animal. Mice were sacrificed 7 days after the transurethral challenge, and for each animal, bladders and kidney pairs were harvested, weighed, and homogenized in 1 and 5 ml of sterile saline, respectively. C. glabrata inocula and burdens were enumerated by performing serial dilutions and counting CFU on YPD agar. The C. glabrata detection limits were 50 and 10 CFU/ml for kidneys and bladder homogenates, respectively. CFU counts were analyzed by unpaired t tests, and a P value of less than 0.05 was considered significant.
Nucleotide sequence accession numbers. The EPA1 gene sequences of C. glabrata strains DSY562 and DSY2235 determined in the present study have been deposited in GenBank under accession numbers KR296804 and KR296805, respectively.