Loss of Upc2p-Inducible ERG3 Transcription Is Sufficient To Confer Niche-Specific Azole Resistance without Compromising Candida albicans Pathogenicity

ABSTRACT Inactivation of sterol Δ5,6-desaturase (Erg3p) in the prevalent fungal pathogen Candida albicans is one of several mechanisms that can confer resistance to the azole antifungal drugs. However, loss of Erg3p activity is also associated with deficiencies in stress tolerance, invasive hyphal growth, and attenuated virulence in a mouse model of disseminated infection. This may explain why relatively few erg3-deficient strains have been reported among azole-resistant clinical isolates. In this study, we examined the consequences of Erg3p inactivation upon C. albicans pathogenicity and azole susceptibility in mouse models of mucosal and disseminated infection. While a C. albicans erg3Δ/Δ mutant was unable to cause lethality in the disseminated model, it induced pathology in a mouse model of vaginal infection. The erg3Δ/Δ mutant was also more resistant to fluconazole treatment than the wild type in both models of infection. Thus, complete loss of Erg3p activity confers azole resistance but also niche-specific virulence deficiencies. Serendipitously, we discovered that loss of azole-inducible ERG3 transcription (rather than complete inactivation) is sufficient to confer in vitro fluconazole resistance, without compromising C. albicans stress tolerance, hyphal growth, or pathogenicity in either mouse model. It is also sufficient to confer fluconazole resistance in the mouse vaginal model, but not in the disseminated model of infection, and thus confers niche-specific azole resistance without compromising C. albicans pathogenicity at either site. Collectively, these results establish that modulating Erg3p expression or activity can have niche-specific consequences on both C. albicans pathogenicity and azole resistance.

M ortality rates associated with invasive fungal infections (IFIs) remain alarmingly high, despite the availability of three major classes of antifungal drugs (1). One of the most prevalent fungal pathogens, Candida albicans, is estimated to cause more than 400,000 IFIs annually, with attributable mortality rates of 35 to 75% (2). Candida species are also a leading cause of mucosal disease, including a collection of oral manifestations (3), and vaginal candidiasis, which afflicts 70 to 75% of healthy adult women (4). Aside from episodic infections, 5 to 10% of all women suffer from recurrent vulvovaginal Candida infections that cause chronic vaginal pain and require long-term antifungal maintenance therapy to alleviate symptoms (4). Improving patient outcomes for both disseminated and mucosal infections will depend upon a detailed understanding of the causes of treatment failure and the development of countermeasures. The azole antifungals remain the most widely used therapies, acting through inhibition of lanosterol demethylase (Erg11p), an enzyme involved in ergosterol biosynthesis. The resulting depletion of cellular ergosterol and the accumulation of abnormal sterol species, such as 14␣-methylergosta-8,24(28)-dien-3␤,6␣-diol, are both thought to cause plasma membrane dysfunction, ultimately leading to growth arrest (5). Several acquired mechanisms of azole resistance have been characterized in C. albicans, and these mechanisms include overexpression of drug efflux pumps belonging to the major facilitator and ABC transporter superfamilies, as well as the target enzyme itself (6,7). Mutations that alter the target enzymes' (Erg11p) azole binding affinity are also important determinants of susceptibility (8)(9)(10). However, a combination of the above mechanisms are usually necessary to confer clinically relevant levels of azole resistance (7,(11)(12)(13). Notably, the resistance of many fungal isolates is not fully accounted by these mechanisms (14).
An alternative mechanism of azole resistance can arise through inactivation of sterol Δ 5,6 -desaturase (Erg3p), the enzyme responsible for converting 14␣-methylfecosterol, which accumulates upon azole-mediated inhibition of Erg11p, into the dysfunctional diol species (5). However, while azole-resistant mutants devoid of Erg3p activity can be readily selected in vitro (15), only a handful have been reported among azole-resistant clinical isolates of C. albicans (5,(16)(17)(18). This may be because loss of Erg3p activity itself blocks a late step of ergosterol biosynthesis (17), resulting in membrane abnormalities and hypersensitivity to some physiological stresses (19)(20)(21)(22). In C. albicans, loss of Erg3p activity has also been associated with defects in hyphal growth (17,20,23), a major virulence determinant of this species (24,25). Accordingly, several reports have described both clinically derived and lab-engineered erg3 mutants as having attenuated virulence in a mouse model of hematogenously disseminated infection (18,20,23). Thus, while inactivation of Erg3p leads to azole resistance, the associated fitness and pathogenicity defects may disfavor the selection of erg3 mutants in the clinical setting. The significance of ERG3 inactivating mutations to azole resistance in the clinical setting has been further brought into question by the finding that a C. albicans erg3⌬/⌬ mutant is susceptible to azole treatment in a mouse model of disseminated infection, despite in vitro resistance (23). Either way, to date, the consequences of Erg3p inactivation upon C. albicans pathogenicity and azole resistance at mucosal body sites has not been directly tested. This is important, as we previously reported that C. albicans erg2⌬/⌬ and erg24⌬/⌬ ergosterol biosynthetic mutants (lacking C-8 sterol isomerase and C-14 sterol reductase, respectively), are essentially avirulent in the disseminated model but readily colonize the mouse vagina (26).
The purpose of this study was to determine how loss of Erg3p activity affects C. albicans pathogenicity and azole resistance in a mouse model of vaginal candidiasis. In the course of conducting these studies, we serendipitously discovered that altering the regulation of ERG3 transcription is sufficient to induce in vitro and niche-specific in vivo azole resistance, without compromising the pathogenic fitness of the fungus.

RESULTS
Altering ERG3 transcription is sufficient to induce in vitro azole resistance without compromising Candida albicans fitness. To determine whether the loss of Erg3p activity affects C. albicans pathogenicity and azole resistance in a murine model of vaginal candidiasis, we first constructed a C. albicans erg3⌬/⌬ deletion strain using a PCR-based approach (27). A reconstituted strain was then made by reintroducing a wild-type ERG3 allele, including 582 bp of 5= untranslated region (5= UTR) and 303 bp of 3= UTR into the deletion strain using an integrating vector that fully restores the IRO1-URA3 locus (28) to circumvent the well-documented positional effects of URA3 integration (29). A prototrophic (and isogenic) deletion strain was produced by introducing the vector alone into the erg3⌬/⌬ mutant.
To confirm the expected in vitro azole resistance of the erg3⌬/⌬ strain, we tested its susceptibility to fluconazole using the standard CLSI broth microdilution protocol (30). As anticipated, the erg3⌬/⌬ mutant was insensitive to fluconazole (Fig. 1). Unexpectedly, the ERG3 reconstituted strain (from here on referred to as shprERG3 [shpr stands for shorter promoter]), was also fully resistant to fluconazole (Fig. 1), voriconazole (see Fig. S1A in the supplemental material), itraconazole (Fig. S1B), and ketoconazole (Fig. S1C). This result was confirmed by testing multiple independently constructed strains of the same genotype. Sequence analysis confirmed the absence of mutations within the cloned ERG3 sequences. Subsequently, a second ERG3 reconstituted strain (from here on referred to as lnprERG3 [lnpr stands for longer promoter]) was produced by reintroducing ERG3 with 1,000 bp of the 5= UTR, which was sufficient to restore fluconazole susceptibility (Fig. 1). These results initially suggested that insufficient 5= UTR had been reintroduced into the shprERG3 strain and that some promoter element The 5= UTR of ERG3 is an important determinant of C. albicans azole susceptibility. The C. albicans erg3Δ/Δ mutant was complemented by reintroduction of a wild-type copy of the ERG3 ORF with either 582 nucleotides (nt) (strain shprERG3) or 1,000 nt (strain lnprERG3) of the 5= UTR and 303 nt of the 3= UTR. The fluconazole susceptibility of the wild-type (WT), erg3Δ/Δ, shprERG3, and lnprERG3 strains and an azole-resistant clinical isolate (TW17) was evaluated using the standard CLSI broth microdilution protocol. Following 24 h (A) or 48 h (B) of incubation, growth was measured as OD 600 and expressed as a percentage of the growth in the control wells with no drug (DMSO alone) for each strain. The means Ϯ standard deviations (error bars) for three biological replicates are indicated.
ERG3, Azole Resistance, and C. albicans Pathogenicity ® essential for ERG3 transcription lies between 1,000 and 582 bp upstream of the start codon. Surprisingly, this was not supported by quantitative reverse transcription-PCR (qRT-PCR) analysis, which revealed similar levels of the ERG3 transcript in wild-type, shprERG3, and lnprERG3 strains (Fig. 2). However, while the addition of 5 g/ml fluconazole increased the ERG3 transcript three-to fourfold in both wild-type and lnprERG3 strains, this response was not observed in the shprERG3 strain. As expected, no ERG3 transcript was detected in the erg3⌬/⌬ strain.
Given that the shprERG3 strain expresses apparently normal levels of the ERG3 transcript in the absence of the azoles, we next examined several other phenotypes that have been associated with Erg3p deficiency (22,31). The erg3⌬/⌬ mutant was hypersensitive to membrane stress caused by the detergent SDS, as well as ionic stresses induced by high concentrations of CaCl 2 or NaCl (Fig. 3A). However, these phenotypes were not observed in either the shprERG3 or lnprERG3 strains, which behaved as the wild-type control.
Several C. albicans Erg3p-deficient strains have also been described as having defects in hyphal growth (17,18,20,23), which is essential for pathogenicity (24,25). Similarly, we found our erg3Δ/Δ mutant to be deficient in hyphal growth on medium 199 (M199) and 10% fetal bovine serum (FBS) agar (Fig. 3B). Under these conditions, both the shprERG3 and lnprERG3 strains formed apparently normal hyphae. Notably, the C. albicans erg3Δ/Δ mutant can form hyphae when grown in liquid M199 medium at 37°C (Fig. S2) or when Ume6p, a transcription factor that activates the expression of hypha-specific genes, is overexpressed (Fig. S3). Thus, the C. albicans erg3Δ/Δ mutant is capable of forming hyphae under at least some conditions. As hyphal growth is intimately linked to the ability of C. albicans to form biofilms (32), we compared each strain's capacity to produce biofilms using a simple in vitro assay. While the erg3⌬/⌬ mutant has diminished capacity to form biofilms, both shprERG3 and lnprERG3 strains were unaffected (Fig. 3C).
In summary, reintroducing ERG3 with the shorter promoter into the erg3⌬/⌬ strain is sufficient to restore normal stress tolerance and hyphal growth but insufficient to confer in vitro fluconazole susceptibility. The phenotypes of the shprERG3 strain demonstrate that reducing ERG3 transcription or eliminating its azole-inducible expression (rather than complete loss of Erg3p function) is sufficient to dramatically enhance C. albicans survival following azole exposure, and without compromising fungal fitness or virulence-related attributes.
Azole-induced ERG3 transcription is dependent upon the Upc2p transcription factor. To provide a convenient reporter of transcriptional activity, either 582 bp (short FIG 2 Azole-inducible ERG3 transcription is lost in the shprERG3 strain. Wild-type (WT), erg3Δ/Δ, shprERG3, and lnprERG3 strains were grown for 24 h in YPD at 37°C in the presence or absence of 5 g/ml fluconazole. RNA was then extracted, ERG3 transcript abundance was quantified by qRT-PCR using the ΔC T method, and ERG3 transcript levels were normalized to that of ACT1. The mean plus standard error of the mean (error bar) for three biological replicates are shown for each strain. Each treatment group was compared using a two-way analysis of variance (ANOVA) and Tukey's multiple-comparison test, and the P values are indicated.
Luna-Tapia et al. ® promoter) or 1,000 bp (long promoter) of the ERG3 5= UTR was amplified from strain SC5314 and cloned upstream of the green fluorescent protein gamma variant (GFP␥) coding sequence (33). Each reporter construct was then introduced into the wild-type strain CAI4 (34). Consistent with the qRT-PCR data, in the absence of fluconazole, GFP␥ expression was similar for either construct. While fluconazole treatment increased the GFP␥ signal approximately eightfold in the strain carrying the longer promoter, no increase was observed in the strain with the shorter ERG3 promoter (Fig. S4). This further supports the existence of a promoter element, located between 582 and 1,000 bp upstream of the ERG3 open reading frame (ORF) that is required to activate transcription in response to azole treatment. To determine whether ERG3 transcription may be affected by changes in growth conditions, we examined the effects of temperature (30°C, 35°C, and 37°C) as well as pH (pH 3, 4, 5, 6, and 7) using the P ERG3 -GFP␥ FIG 3 Loss of azole-inducible ERG3 transcription does not affect C. albicans stress tolerance or hyphal growth. (A) C. albicans cell suspensions were prepared at 1 ϫ 10 7 cells/ml, and serial 1:5 dilutions prepared in sterile water. Cell suspensions were then applied to YPD agar plates or YPD agar supplemented with the indicated concentrations of fluconazole (Fluc), SDS, NaCl, or CaCl 2 using a sterile multipronged applicator. The plates were incubated at 30°C for 48 h and then imaged. (B) C. albicans strains were suspended at 1 ϫ 10 7 cells/ml in sterile water, and 2.5 l spotted onto either an M199 or 10% FBS agar plate. The resulting colonies were imaged after 96-h incubation at 37°C. (C) The capacity to form biofilms was also evaluated in flat-bottomed 96-well plates. Each C. albicans strain was grown in RPMI 1640 medium at 37°C for 24 h and stained with crystal violet before biofilm mass was quantified as OD 570 . Values are means plus standard errors of the means (SEM) from four independent experiments. The biofilm mass of each group was compared using one-way ANOVA and Tukey's multiple-comparison test. The value for the erg3Δ/Δ mutant group was significantly different (P Ͻ 0.0001) from the values for the WT, shprERG3, and lnprERG3 groups as indicated by the two asterisks.
ERG3, Azole Resistance, and C. albicans Pathogenicity ® reporter constructs. However, neither parameter had a significant impact on GFP␥ expression from either the long or short promoters (less than twofold, data not shown).
In C. albicans, the Upc2p transcription factor enhances the expression of several ergosterol biosynthetic genes in response to azole treatment, including ERG11 (35). Although ERG3 appears to be a target of Upc2p in Saccharomyces cerevisiae (36), this had not been confirmed in C. albicans. To address this question, we introduced the azole-responsive GFP␥ reporter construct (with the longer promoter) into an upc2Δ/Δ deletion strain. In the absence of fluconazole, GFP␥ expression was similar in the upc2Δ/Δ mutant and wild-type strain backgrounds, indicating similar levels of basal transcription (Fig. 4). Fluconazole treatment increased GFP␥ expression about 10-fold in the wild-type strain, but not in the upc2Δ/Δ mutant. These results confirm that the induction of ERG3 transcription that occurs in C. albicans upon azole treatment is completely dependent upon the Upc2p transcription factor.
Using the program matrix-scan at the Regulatory Sequence Analysis Tool RSAT-Fungi webserver (http://rsat-tagc.univ-mrs.fr/rsat/matrix-scan_form.cgi) (37), we searched for motifs that may represent putative Upc2p-binding sites within the 5= UTR of ERG3. A cis-regulatory element-enriched region (CRER) was identified between positions Ϫ751 to Ϫ373. Within this region, four motifs represent putative Upc2p-binding sites (Fig. S5). Another element located at positions Ϫ379 to Ϫ373 (TCGTTTA) that coincides with a putative Upc2-binding site resembles the classical sterol-responsive element (SRE) TCGTATA, present in the 5= UTRs of UPC2 and several ergosterol biosynthetic genes (38)(39)(40); hence, we deemed this sequence an "imperfect" SRE (iSRE) (Fig. S5). The ERG3 promoter also contains two elements that match the TATA box consensus sequence TATAWAWR (W means A or T, and R means A or G) (41); the first element is located at positions Ϫ679 to Ϫ672, and the second element is located at positions Ϫ343 to Ϫ336 (Fig. S5). Nonetheless, the role of each of the above elements in ERG3's transcriptional upregulation in response to the azoles has not yet been determined.
Loss of azole-inducible ERG3 transcription is sufficient to eliminate "toxic" diol production in fluconazole-treated C. albicans. In order to compare levels of Erg3p activity, we analyzed the composition of sterols extracted from both untreated and fluconazole-treated cultures of each strain. As expected, untreated wild-type and lnprERG3 cells produced an abundance of ergosterol (Table 1). In contrast, the erg3⌬/⌬ FIG 4 Azole-inducible ERG3 transcription is dependent upon the Upc2p transcription factor. The GFP␥ coding sequence was cloned downstream of 1 kb of the ERG3 promoter sequence, and the resulting construct introduced into C. albicans wild-type strain CAI4 (WT) or a upc2Δ/Δ mutant. The resulting strains were grown in YNB broth in the presence or absence of 5 g/ml fluconazole for 24 h at 35°C with shaking, before GFP␥ fluorescence was quantified (excitation wavelength of 488 nm [9 nm bandpass]; emission wavelength of 507 nm [9 nm bandpass]). Fluorescence intensity was then normalized to the cell density (OD 600 ) and expressed relative to the DMSO control. The means Ϯ the standard deviations for four biological replicates are shown. The relative fluorescence of each treatment group was then compared using a two-way ANOVA and Tukey's multiple-comparison test, and the P values are indicated. mutant did not produce detectable levels of ergosterol under untreated conditions, instead accumulating high levels of ergosta-7,22-dienol. The addition of fluconazole reduced ergosterol content in the wild-type and lnprERG3 strains by more than 25-fold and induced the formation of significant levels of 14␣-methylergosta-8,24(28)-dien-3␤,6␣-diol (Table 2). However, erg3⌬/⌬ cells did not form the dysfunctional diol species upon fluconazole treatment. Interestingly, while the shprERG3 strain formed near normal levels of ergosterol in the absence of drug (Table 1), it failed to produce detectable levels of the sterol-diol species upon fluconazole treatment ( Table 2). The capacity to produce ergosterol in the absence of drug explains the lack of stress and hyphal growth phenotypes for the shprERG3 strain. The azole resistance of this strain can also be explained by the absence of abnormal diols upon antifungal treatment.
The content of sterols indicative of Erg3p activity is shown in boldface type. ND, not detected. b Sterols that have not been identified (the mass ion and/or fragmentation pattern was unclear).  In some cases (e.g., 14-methyl-ergosta-trienol), we cannot be certain of the arrangement of double bonds in the sterol, but it is an ergosta type sterol with three double bonds and a 14-methyl group.
ERG3, Azole Resistance, and C. albicans Pathogenicity ® Loss of azole-inducible ERG3 transcription does not compromise C. albicans pathogenicity, but it is insufficient to confer fluconazole resistance in a mouse model of disseminated infection. It has been previously reported that deletion of ERG3 in the SC5314 strain background reduces C. albicans virulence in the mouse model of disseminated candidiasis (20,21,23). We therefore compared the pathogenicity of our erg3⌬/⌬, shprERG3, and lnprERG3 strains to a wild-type control using the same model. This confirmed that our erg3⌬/⌬ strain is severely impaired in this model, as it failed to cause mortality within 14 days of inoculation (Fig. 5A). In contrast, the virulence of both the shprERG3 and lnprERG3 strains was indistinguishable from that of the wild-type strain, with each causing 100% mortality. This further supports our earlier conclusion that the fitness and pathogenicity of the shprERG3 strain are not compromised, despite its in vitro azole resistance. Interestingly, while the erg3⌬/⌬ deletion strain did not cause mortality, it was able to persist within the mouse tissues to the end of the experiment {(~2.03 Ϯ 1.32) ϫ 10 6 CFU/g [mean Ϯ standard error of the mean (SEM)]; day 14 postinoculation}. However, the erg3Δ/Δ mutant showed defects in hypha formation in this model of infection (Fig. S6).
To compare the in vivo susceptibility of these strains to fluconazole, a second experiment was conducted in which groups of mice (four mice in each group) were inoculated intravenously with each strain and then either treated for 7 days with fluconazole (25 mg/kg of body weight) or sham treated with phosphate-buffered saline (PBS) via daily intraperitoneal injections. Consistent with the previous experiment, nearly all mice infected with the wild type, shprERG3, and lnprERG3 and treated with , erg3Δ/Δ, shprERG3, or lnprERG3 strain via lateral tail vein injection. The mice were then monitored three times daily for 14 days, and those showing distress were humanely euthanized. The survival of each group was compared using the log rank test (P ϭ 0.0051). (B) Groups of BALB/c mice were inoculated with~5 ϫ 10 5 CFU of either WT (GP1), erg3Δ/Δ, shprERG3, or lnprERG3 strain as described above and then split into two treatment groups (four mice in each group). One group was treated with 25 mg/kg fluconazole (Fluc), and another group was treated with vehicle alone (PBS) by daily intraperitoneal injection starting 24 h after infection. Animals were monitored for 8 days, and the survival of each group was compared using the log rank test (P Ͻ 0.0001).

®
PBS succumbed to infection, whereas those mice treated with fluconazole all survived to day 8 (Fig. 5B). Analysis of fungal colonization of the kidneys revealed that fluconazole-treated mice had only low levels of fungal colonization {(8.77 Ϯ 1.33) ϫ 10 2 CFU/g [mean Ϯ SEM] for the wild-type, (2.97 Ϯ 1.26) ϫ 10 3 CFU/g for shprERG3, and (2.64 Ϯ 0.14) ϫ 10 2 CFU/g for lnprERG3}, indicating that all three strains are susceptible to fluconazole in this model. Thus, while the shprERG3 strain is resistant in vitro, it is apparently susceptible to fluconazole in the mouse model of disseminated infection. The erg3⌬/⌬ deletion strain failed to cause lethality in either the PBS or fluconazole treatment groups. However, kidney CFU counts from mice surviving to day 8 were not significantly different between the groups treated with fluconazole [(3.66 Ϯ 2.34) ϫ 10 4 CFU/g] and PBS [(2.22 Ϯ 0.49) ϫ 10 5 CFU/g]. Thus, while the erg3⌬/⌬ strain is much less virulent than the other three strains in the mouse model of disseminated infection, the persistent levels of CFU in the kidney indicate that it is apparently much more resistant to fluconazole treatment.
Complete or partial loss of Erg3p activity is sufficient to confer fluconazole resistance in a mouse model of vaginal candidiasis. Previous investigations of Erg3p's role in C. albicans pathogenicity and azole resistance have focused exclusively on the disseminated model of infection (20,21,23). However, it is unknown whether loss of Erg3p activity is sufficient to impair C. albicans virulence or confer azole resistance in other host niches, such as the mucosal membranes that are its primary habitat. To test this, we inoculated estrogen-treated mice intravaginally with either wild-type, erg3Δ/Δ, shprERG3, or lnprERG3 strains. Each group was then split into two treatment groups, with half the mice treated with fluconazole (50 mg/kg) via intraperitoneal injection on days 4 and 7 postinoculation, and the other half receiving PBS vehicle alone. The levels of fungal colonization were then determined as CFU from vaginal lavage fluid obtained on days 7 and 10. All four strains were able to colonize the PBS-treated mice to similar levels at both time points (Fig. 6A and B). Fluconazole treatment significantly reduced fungal colonization by both wild-type and lnprERG3 strains but had less impact in animals infected with the erg3Δ/Δ or shprERG3 strain ( Fig. 6A and B). This indicates that both erg3Δ/Δ and shprERG3 strains are not only proficient at colonizing the mouse vagina but also have reduced susceptibility to fluconazole treatment at this site.
Smears from vaginal lavage fluid samples were also made on days 7 and 10 to compare the levels of polymorphonuclear leukocyte (PMN) infiltration, as high levels of PMN influx have been associated with symptomatology (42). Seven days postinoculation, mice infected with the erg3⌬/⌬ mutant had only a slight reduction in PMN counts compared to the other three strains, further supporting that the mutant has no major virulence defect in this model of infection (Fig. 6C). On day 7, the fluconazole treatment groups had reduced PMN influx in animals infected with the wild-type, shprERG3, and lnprERG3 strains compared to the untreated groups. In contrast, fluconazole had no impact on PMN influx in animals infected with the erg3⌬/⌬ mutant (Fig. 6C). By day 10, only mice infected with the wild-type strain had a statistically significant reduction in PMN influx in the treated versus untreated groups (Fig. 6D). Similar levels of PMNs were observed in the treated and untreated animals infected with either the erg3⌬/⌬ or shprERG3 strain, indicating little or no benefit to treatment. For mice infected with the lnprERG3 strain, there was a trend toward reduced PMN influx in the fluconazoletreated group; however, this was not statistically significant. Finally, the hyphae observed in vaginal lavage fluid from mice infected with the erg3⌬/⌬ mutant were indistinguishable from hyphae formed by the other three strains in this model (Fig. S7). Thus, the erg3⌬/⌬ mutant is not deficient in hyphal growth at this mucosal site. These findings indicate that the erg3Δ/Δ mutant has no colonization defect and is able to induce pathology in the mouse model of vaginal candidiasis. Furthermore, in contrast to the wild type, the capacity of the erg3Δ/Δ mutant to induce pathology is not compromised by fluconazole treatment.

DISCUSSION
In vitro selection studies with several fungal species have demonstrated that azole resistance frequently results from mutations that inactivate the sterol C-5-desaturase encoded by ERG3 (15,43,44). However, relatively few Erg3p-deficient mutants have been identified among azole-resistant clinical isolates of C. albicans (5,(16)(17)(18), which instead seem to favor mechanisms involving alterations of the target protein's amino acid sequence, increased target protein expression, and/or elevated expression of drug efflux pumps (6)(7)(8)(9)(10). This is somewhat surprising, as the acquisition of clinically relevant levels of azole resistance usually requires some combination of these mechanisms, and is thought to be a stepwise process (7,(11)(12)(13). In contrast, inactivation of Erg3p can offer a single step mechanism to complete azole resistance. Several factors may help explain the apparent disparity between the high frequency with which azole-resistant erg3 mutants can be selected in vitro and their infrequent occurrence in the clinical setting.
First, the ploidy of the fungal species in question is likely to affect the frequency with which erg3 mutants arise, as erg3 loss-of-function mutations are recessive (15). As such, while azole-resistant erg3 mutants can be readily selected in a haploid species such as S. cerevisiae (15,(44)(45)(46), they are expected to occur less frequently in a constitutively diploid species such as C. albicans, as inactivation of both ERG3 alleles is necessary to increase MICs. This would require either two independently occurring mutations in either allele, or mutation in one allele followed by a gene conversion event. In contrast, activating mutations in the Upc2p transcription factor, which controls expression of the target enzyme  Erg11p, or those controlling expression of the Candida drug resistance (CDR)-and multidrug resistance (MDR)-type drug efflux pumps (Tac1p and Mrr1p), can be either semidominant or fully dominant (13,(47)(48)(49). Nonetheless, in vitro selection of Candida dubliniensis, a diploid species, with itraconazole primarily selects for Erg3p-deficient resistant mutants (43). Thus, ploidy alone does not satisfactorily account for the low frequency with which azole-resistant erg3 mutants are isolated in the clinical setting.
A second factor may be the concentration of drug to which the fungus is exposed. Anderson et al. reported that in vitro exposure of S. cerevisiae isolates to high doses of fluconazole primarily selects for erg3 mutants, while progressive selection with stepwise increases in fluconazole concentration favors resistance mechanisms that involve elevated expression of drug efflux pumps (15). As such, it is conceivable that patient dosing and drug availability at the site of infection are important determinants of the adaptive mechanisms that arise.
A third, possible explanation for the infrequent isolation of erg3 mutants in the clinical setting may be the associated fitness costs. It is clear from our data, as well as several other studies, that while complete loss of Erg3p activity leads to in vitro azole resistance, it concurrently causes several phenotypic abnormalities. These abnormalities include hypersensitivity to various physiological stresses, as well as deficiencies in hyphal growth under at least some culture conditions. The resulting loss of pathogenic fitness may oppose or disfavor the selection of erg3 mutants in patients receiving azole treatment. While the alternative sterol species that accumulate in erg3 null mutants are sufficient to support propagation in lab culture, they may be less compatible with fungal survival, proliferation, or invasiveness during infection of the mammalian host. This is supported by several studies that have shown that erg3 mutants isolated from the clinical setting and lab-engineered erg3⌬/⌬ strains are less virulent than wild-type isolates in murine models of disseminated infection (18,20,23). Interestingly, Vale-Silva et al. reported an azole-resistant C. albicans clinical isolate with an erg3 null mutation, which is apparently virulent, and had no obvious defects in stress tolerance or hyphal growth (17). The authors suggest that this isolate may harbor compensatory mutations that counteract the effects of Erg3p inactivation. The nature of the compensatory mechanism(s) and the frequency with which they can coincide with erg3 inactivating mutations are unknown. Nonetheless, the weight of evidence, including the in vitro data presented herein, suggests that in the absence of compensatory mechanisms or mutations, complete loss of ERG3 function is detrimental to C. albicans pathogenic fitness. This notion was only partially supported by our in vivo data, which revealed that the erg3⌬/⌬ mutant has niche-specific defects in pathogenicity.
In both disseminated and vaginal models of infection, wild-type C. albicans is pathogenic but susceptible to azole treatment. In contrast, the erg3⌬/⌬ mutant has severely attenuated virulence in the mouse model of disseminated candidiasis but is proficient at colonizing and capable of inducing pathology in the vaginal model of infection. Thus, the fitness defects incurred upon Erg3p inactivation are apparently of greater consequence during disseminated infection than mucosal propagation and/or infection. This may reflect the fact that C. albicans is highly adapted for a commensal lifestyle upon mucosal surfaces and typically only disseminates in a highly debilitated mammalian host. Our previous finding that C. albicans erg2⌬/⌬ and erg24⌬/⌬ mutants (lacking C-8 sterol isomerase and C-14 sterol reductase, respectively) are avirulent in the mouse model of disseminated infection but readily colonize the mouse vagina (26) further supports the notion that mucosal environments may be more permissive of alterations in the ergosterol biosynthetic pathway than deeper infection sites. Either way, the deletion mutant is relatively resistant to fluconazole treatment in both models of infection. As such, while the erg3⌬/⌬ mutant has niche-specific virulence defects, it is more azole-resistant than the wild type is at both mucosal and disseminated sites of infection (Table 3).
Our results with the shprERG3 strain indicate that complete loss of Erg3p activity is not necessary to confer azole resistance. Specifically, the loss of azole-inducible transcription of ERG3 apparently confers fluconazole resistance in vitro without compro-mising C. albicans pathogenic fitness. Like the wild type, the shprERG3 strain was pathogenic in both models of infection and susceptible to azole treatment in the disseminated model. However, the shprERG3 strain was more resistant to fluconazole treatment in the vaginal model than the wild type, and therefore exhibits niche-specific azole resistance (Table 3). These findings imply that the level of expression and specific activity of Erg3p directly impact C. albicans pathogenicity and susceptibility to azole therapy. Moreover, titrating Erg3p expression/activity has niche-specific consequences upon these vitally important characteristics.
The fourth possible explanation for the dearth of Erg3p-deficient mutants in the clinical setting may be that many are overlooked, as C-5 desaturase activity is not routinely examined. Point mutations and polymorphisms within ERG3 can be readily detected through sequence analysis, and isolates bearing nonsense mutations-resulting in a truncated (nonfunctional) polypeptide are easily identified. However, the consequences of missense mutations or polymorphisms within the 5= and 3= noncoding regions upon Erg3p expression, specific activity, or azole susceptibility are rarely investigated. A detailed biochemical analysis of cellular sterol content is required to confirm deficiencies in C-5 desaturase activity-an analysis that is not routinely performed on all azole-resistant isolates. If this analysis is performed, isolates with a substantive reduction in C-5 sterol desaturase activity can be detected as having significantly reduced ergosterol content (or its complete absence) and an accumulation of ergosta-7,22-dienol. Accordingly, to date, most reports have focused upon describing clinical isolates with mutations that completely inactivate the ERG3 gene. However, isolates with only a partial reduction in Erg3p activity will have more subtle shifts in sterol profiles that may not be as obvious. This is important, as the results described herein suggest relatively small adjustments to ERG3 transcription may be sufficient to induce azole resistance in C. albicans-at least within some host niches. The Kelly lab has previously shown that many of the azole-resistant S. cerevisiae erg3 mutants they selected in vitro are in fact "leaky" mutants that only partially impair Erg3p catalytic activity (45). Through thorough analysis of sterol profiles, at least two leaky erg3 mutants were also identified among a panel of azole-resistant clinical C. albicans isolates (16). Such mutations permit the production of at least some ergosterol but are also sufficient to induce full azole resistance in vitro (16). Notably, the two leaky erg3 C. albicans mutants show phenotypic profiles similar to the profile of the shprERG3 strain reported here, i.e., full azole resistance in vitro but normal hyphal growth and stress tolerance (see Fig. S7 in the supplemental material). Our findings beg a series of important questions: To what degree does Erg3p activity need to be suppressed to induce azole resistance? To what degree can Erg3p activity be suppressed before the fitness or pathogenicity of C. albicans is compromised? Are there levels of Erg3p activity that can support full azole resistance without compromising C. albicans pathogenicity? Do the Erg3p activity levels required to support C. albicans pathogenicity and/or determine azole susceptibility/resistance vary between different host niches (e.g., oral versus vaginal versus disseminated) or according to host immune status? Our results also raise the question of whether, and how, altered transcriptional regulation might
confer Erg3p-dependent clinical azole resistance that may not be apparent in vitro using the standard CLSI MIC assay. Similarly, are there host relevant factors that impinge upon the regulation of Erg3p expression or activity? In this regard, using the P ERG3 -GFP␥, we found that ␤-estradiol (1.25 to 10 g/ml), which was used in the vaginal model, did not affect transcription from the ERG3 promoter in the presence or absence of fluconazole (not shown). Finally, and perhaps most importantly, given the almost exclusive focus thus far on erg3 null mutants-might the role of Erg3p as a determinant of C. albicans azole resistance have been grossly underestimated?
In summary, we have demonstrated that relatively minor adjustments in the regulation of ERG3 transcription are sufficient to induce azole resistance without compromising C. albicans hyphal growth or stress tolerance. Furthermore, even complete loss of Erg3p activity does not compromise C. albicans pathogenic fitness in a murine model of vaginal candidiasis. On the basis of these results and those presented elsewhere (16), we propose that a more thorough investigation of the role of the ERG3 gene as a determinant of azole susceptibility in the clinical setting is warranted. These investigations should not only seek to identify erg3 null mutants but also consider the effects of variation in ERG3 transcription, protein expression, and missense mutations on the catalytic activity of Erg3p. Furthermore, the consequences of altered Erg3p activity upon fungal pathogenicity should be simultaneously examined.

MATERIALS AND METHODS
Growth and media conditions. All C. albicans strains were routinely grown on yeast extractpeptone-dextrose (YPD) agar plates or in liquid medium at 30°C and supplemented with uridine (50 g/ml) when necessary. Transformant selection was carried out on minimal YNB medium (6.75 g/liter yeast nitrogen base without amino acids, 2% dextrose, 2% Bacto agar), supplemented with the appropriate auxotrophic requirements as described for S. cerevisiae (50) or uridine at 50 g/ml. All C. albicans strains were stored frozen at Ϫ80°C in 20% glycerol.
For sterol extraction, strains were grown overnight in YNB, and 100 l of a solution with 1 ϫ 10 6 cells/ml were used to inoculate a final volume of 10 ml YNB broth plus 5 g/ml fluconazole or 0.5% dimethyl sulfoxide (DMSO) (minus fluconazole control) to give a starting concentration of 1 ϫ 10 4 cells/ ml. These cultures were grown at 35°C for 24 h with shaking at 160 rpm. All strains were grown in triplicate.
For RNA extraction, 1 ϫ 10 4 cells/ml were inoculated in 5 ml YPD and grown at 37°C with shaking for 24 h in the presence of 4 g/ml of fluconazole or 0.5% DMSO. All strains were grown in triplicate.
Plasmid construction. Plasmid pLUX (28), was kindly provided by William Fonzi (Georgetown University). Plasmid pKE1 (51) was previously described. All oligonucleotides used in this study are listed in Table S1 in the supplemental material.
Construction of the reporter plasmids for transcriptional analysis, plnERG3prGFP␥ and pshERG3prGFP␥, was performed as follows. Different lengths of the ERG3 promoter were amplified from SC5314 gDNA with primer pairs ERG3lngpF-KpnI plus ERG3prR-SalI (1,000 bp upstream of the start codon) or ERG3shrpF-KpnI plus ERG3prR-SalI (582 bp upstream of the start codon). These promoter fragments were then cloned between the KpnI and SalI sites of pKE1 in place of the ACT1 promoter. The GFP␥ coding sequence was then amplified using primer pair GFPAMPF-SalI plus GFPAMPR-MluI and cloned downstream of the ERG3 promoter between SalI and MluI sites.
C. albicans strain construction. Strains used in this study are listed in Table S2. Gene deletion strains were constructed by the PCR-based approach of Wilson et al. (27), using the ura3⌬/⌬ his1⌬/⌬ arg4⌬/⌬ strain BWP17 (kindly provided by Aaron Mitchell, Carnegie Mellon University).
ERG3 deletion cassettes were amplified by PCR with primer pair ERG3DISF plus ERG3DISR, using pRS-ARG4⌬SpeI or pGEM-HIS1 (27) as the template. Each ERG3 allele was sequentially deleted from strain BWP17 using ARG4 and HIS1 markers to generate erg3⌬/⌬ ura3⌬/⌬ gene deletion mutants. Correct integration of deletion cassettes was confirmed at each step by PCR with the following primer sets: ARG4INTF2 plus ERG3AMPR-SacI and ARG4INTR2 plus ERG3AMPF-SacI (ARG4 integration) or HIS1INTF2 plus ERG3AMPR-SacI and HIS1INTR2 plus ERG3AMPF-SacI (HIS1 integration). The lack of an intact ERG3 allele was confirmed using primer pair ERG3DETF and ERG3DETR.
For construction of the C. albicans upc2Δ/Δ ura3Δ/Δ mutant, each UPC2 allele was sequentially deleted from strain BWP17 using the ARG4 and HIS1 markers. Deletion cassettes were amplified with primer pairs UPC2DISF plus UPC2DISR, using pRS-ARG4⌬SpeI or pGEM-HIS1 (27)  HIS1INTR2 and UPC2AMPR-SacI plus HIS1INTF2 (HIS1 integration). The absence of an intact UPC2 allele was confirmed using primer pair UPC2SEQF1 and UPC2SEQR3. For the construction of the strains used in the GFP␥ reporter assays, plasmids plnERG3prGFP␥ and pshERG3prGFP␥ were linearized with NheI and transformed into wild-type CAI4 and upc2⌬/⌬ ura3⌬/⌬ strains.
Stress resistance and hyphal growth assays. For stress resistance analysis, C. albicans was grown overnight in YPD at 30°C. Cells were washed in sterile deionized water, cell density was adjusted to 10 7 cells/ml, and 1:5 serial dilutions were performed in a 96-well plate. Cells were then applied to agar plates using a sterile multipronged applicator. Resistance to different stresses was determined on YPD agar containing 8 g/ml of fluconazole, 0.05% SDS, 1.5 M NaCl, or 500 mM CaCl 2 , with incubation at 30°C during 48 h.
For hyphal growth analysis, 2.5 l from a cell suspension with 10 7 cells/ml was spotted on M199 agar or 10% FBS (fetal bovine serum) agar plates, followed by incubation for 96 h at 37°C.
Quantification of biofilm biomass. Each C. albicans strain was grown overnight in YPD broth at 37°C with shaking at 200 rpm. Each culture was then washed two times in sterile phosphate-buffered saline (PBS), cell density was adjusted to 1 ϫ 10 6 per ml in RPMI 1640 medium, and 200 l was dispensed into each well of a flat-bottomed, 96-well plate (six wells per strain). After incubation at 37°C for 24 h, the plate was rinsed three times with sterile PBS and then stained with 0.01% crystal violet (MP Biomedicals) for 15 min. The plate was again rinsed three times with sterile water, and biofilms were eluted with 95% ethanol (200 l/well). One hundred twenty-five microliters of resolubilized dye was transferred to a new microtiter plate, and the optical density at 570 nm (OD 570 ) was measured using a microplate reader. Four independent experiments were performed.
Antifungal susceptibility testing. Antifungal susceptibility testing of all the strains included in this study was performed by using the broth microdilution method described in the CLSI document M27-A3 (30) in a 96-well plate format. All drugs for susceptibility testing used in this study were diluted in DMSO in twofold dilutions at 200 times the final concentration. Fluconazole (Sigma-Aldrich) concentrations tested ranged from 64 g/ml to 0.0313 g/ml. RPMI 1640 medium (Sigma-Aldrich) was prepared according to the CLSI document. The medium was buffered with 0.165 M morpholinepropanesulfonic acid (MOPS), and its pH was adjusted using NaOH and HCl. Cell inoculum was~1 ϫ 10 3 cells per well. Plates were incubated without shaking at 35°C for 24 or 48 h. The content of each well was carefully resuspended by pipetting up and down before OD 600 was measured using a Cytation 5 cell imaging multimode reader (Bio-Tek Instruments, Inc.).
Sterol extraction and quantitation. Nonsaponifiable lipids were extracted using alcoholic KOH as reported previously (52). Samples were dried in a vacuum centrifuge (Heto) and derivatized by the addition of 100 l of 90% N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA)-10% trimethylsilyl (TMS) (Sigma), 200 l of anhydrous pyridine (Sigma), and heating for 2 h at 80°C. TMS-derivatized sterols were analyzed and identified using gas chromatography-mass spectrometry (GC-MS) (Thermo 1300 GC coupled to a Thermo ISQ mass spectrometer; Thermo Scientific) with reference to retention times and fragmentation spectra for known standards. GC-MS data files were analyzed using Xcalibur software (Thermo Scientific) to determine sterol profiles for all isolates and for integrated peak areas. Percentages of total sterols are given as the mean for three replicates.
RNA extraction. Cells were collected by centrifugation at 3,500 rpm for 5 min. Supernatant was poured off, and cells pellets were stored at Ϫ80°C. RNA was extracted from the cell pellets using the hot phenol method of Schmitt et al. (53). RNA pellets were eluted in 20 l of nuclease-free water, and quantity and purity were determined spectrophotometrically by measuring absorbance at 260 nm and 280 nm.
Quantitative RT-PCR (qRT-PCR). RNA concentrations were adjusted to 1 g, and aliquots were treated with DNase I as indicated by the manufacturer (Thermo Scientific). cDNA was synthesized using the Verso cDNA synthesis kit (Thermo Scientific) with random hexamers according to the manufacturer's instructions. Quantitative PCR was performed with primer pairs that amplify 100 bp of the ACT1 (ACT1FWDS2 plus ACT1REVS2) or ERG3 (ERG3qPCR_F2 plus ERG3qPCR_R2) (Table S1) coding sequences using the Maxima SYBR green/ROX quantitative PCR (qPCR) master mix (2ϫ) (Thermo Scientific) as indicated by the manufacturer. Reactions were completed in a 7500 real-time PCR system (Applied Biosystems). Expression levels of ERG3 were compared to those of ACT1 (normalizing gene) using the ΔC T method as described previously (54).
GFP␥ reporter assays. C. albicans strains expressing GFP␥ from the ERG3 promoter were grown overnight in YPD at 30°C, washed, and subcultured into 10 ml YNB at 1 ϫ 10 6 cells/ml in the presence of 5 g/ml fluconazole or 0.5% DMSO, and then incubated at 35°C for 24 h with shaking. Cells were pelleted and transferred to a round-bottomed 96-well plate, and GFP␥ fluorescence intensity was quantified using a Cytation 5 cell imaging multimode reader (Bio-Tek Instruments, Inc.) with excitation at 488 nm (9 nm bandwidth) and emission at 507 nm (9 nm bandwidth). Growth was determined by measuring OD 600 for normalization of the fluorescence signal.
Ethics statement. The animals used in this study were housed in American Association for Accreditation of Laboratory Animal Care (AAALAC)-approved facilities at the University of Tennessee Health Science Center (UTHSC). The Institutional Animal Care and Use Committee (IACUC) at UTHSC approved use of all animals and procedures (IACUC protocol numbers 15-081 and . Mice were given standard rodent chow and water ad libitum. Mice were monitored daily for signs of distress, including noticeable weight loss, lethargy, and body condition score. The IACUC at UTHSC uses the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals (55) and the Guide for the Care and Use of Laboratory Animals (56) as a basis for establishing and maintaining an institutional program for activities involving animals. To ensure high standards for animal welfare, the IACUC at UTHSC remains compliant with all applicable provisions of the Animal Welfare Act (AWAR) (57), guidance from the Office of Laboratory Animal Welfare (OLAW), and the American Veterinary Medical Association Guidelines on Euthanasia (58).
Mouse model of vaginal candidiasis. C57BL/6 mice were administered 0.1 mg of estrogen (␤-estradiol 17-valerate; Sigma) dissolved in 0.1 ml sesame oil subcutaneously 72 h prior to inoculation with C. albicans. Estrogen injections were administered weekly thereafter. Stationary-phase cultures of C. albicans isolates were washed three times in sterile, endotoxin-free PBS and resuspended in PBS at 2.5 ϫ 10 8 cells/ml. Estrogen-treated mice (eight mice in each experimental group) were then inoculated intravaginally with 20 l of the cell suspension (final inoculum, 5 ϫ 10 6 cells). Mice were monitored daily for signs of distress, including noticeable weight loss and lethargy. Fluconazole treatment was administered at 50 mg/kg via intraperitoneal injection on days 4 and 7 postinoculation (p.i.). Vaginal lavage was performed on day 7 and 10 p.i., and fungal colonization was quantified as CFU by plating serial dilutions of the lavage fluid onto YPD agar plates containing 50 g/ml of chloramphenicol. PMN counts within the lavage fluid samples were also enumerated as an indicator of inflammation. Smears of lavage fluid samples were stained by the Papanicolau technique, and the number of PMNs in five nonadjacent fields was counted by using a light microscope with a 40ϫ objective.
Mouse model of disseminated candidiasis. C. albicans strains were grown overnight in YPD cultures at 30°C with shaking. Stationary-phase cultures of C. albicans strains were washed twice in sterile, endotoxin-free PBS and resuspended in PBS. BALB/c mice were inoculated via tail vein injection with 5 ϫ 10 5 cells in 100 l of cell suspension. Viable-cell counts of each inoculum were confirmed by plating appropriate dilutions onto YPD agar plates and counting the colonies formed after 48 h.
Pathogenesis experiment. Groups of six animals per strain were infected and monitored for 14 days p.i., and those showing signs of distress were humanely sacrificed. Surviving animals by day 14 p.i. were euthanized. The kidneys from each mouse were extracted, weighed, and homogenized in PBS. Serial dilutions of kidney homogenate were plated on YPD agar plates containing 50 g/ml of chloramphenicol. The number of CFU/gram of kidney tissue was determined from the number of colonies on the plates after 48 h.
In vivo fluconazole treatment experiment. Groups of eight animals per strain were inoculated via tail vein injection with 100 l of cell suspension. After 24 h p.i., each group was split into two groups of four animals each. Either group was treated intraperitoneally with 200 l of 25 mg/kg/day of fluconazole or with vehicle alone (PBS plus 1% DMSO) daily for 6 days (until day 7 p.i.). Mice were monitored daily, and those showing distress were euthanized. Kidneys were removed from each mouse at the time of death, weighed, and homogenized in PBS, and their fungal burdens were determined as described earlier in this article. On day 8 p.i., all surviving animals were sacrificed, their kidneys were removed, and their fungal burdens were determined as explained previously.