Trafficking through the Late Endosome Significantly Impacts Candida albicans Tolerance of the Azole Antifungals

Growth conditions.

C. albicans was routinely grown on yeast extract-peptone-dextrose (YPD) at 30°C, supplemented with 50 μg/ml uridine 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 (20) or 50 μg/ml uridine.

For growth curves, overnight cultures in YPD at 30°C were subcultured into 15 ml fresh YNB medium at an initial cell concentration of 1 × 10⁶ cells/ml in the presence of 1 μg/ml fluconazole or 0.5% dimethyl sulfoxide (DMSO) (drug-free control), and incubated at 30°C with shaking. Samples were then taken at 30-min intervals, optical density at 600 nm (OD₆₀₀) was determined spectroscopically, and cell viability was measured by counting the number of CFU on YPD agar plates. A third sample was taken simultaneously to observe vacuolar morphology (see below).

Plasmid construction.

Plasmid pLUX (21) was kindly provided by William Fonzi (Georgetown University). Plasmids pLUXVPS21 (16) and pKE1 (22) have been previously described. All oligonucleotides used in this study are listed in Table S1 in the supplemental material. The construction of additional plasmids is described in the supplemental material.

C. albicans strains.

The vps21Δ/Δ, ypt72Δ/Δ, atg9Δ/Δ, aps3Δ/Δ, ypt52Δ/Δ, ypt53Δ/Δ, vps21Δ/Δ ypt52Δ/Δ, vps21Δ/Δ ypt53Δ/Δ, ypt52Δ/Δ ypt53Δ/Δ and vps21Δ/Δ ypt52Δ/Δ ypt53Δ/YPT53^T27N (the T-to-N change at position 27 encoded by YPT53 [YPT53^T27N]) mutants were constructed in previous studies (16, 22–24). The C. albicans pep12Δ/Δ mutant and isogenic control strains (25) were kindly provided by Samuel Lee (University of New Mexico). The azole-susceptible clinical isolate TW1 (isolate 1 in reference 4) and the matched azole-resistant isolate TW17 (isolate 17 in reference 4) were kindly provided by Theodore C. White (University of Missouri—Kansas City). Control strain YJB6284 (26) was kindly provided by Judith Berman (University of Minnesota), and strain SC5314 has been described previously (27). C. albicans was transformed with DNA constructs using the lithium acetate procedure (28). Gene deletion strains were constructed by the PCR-based approach of Wilson et al. (29), using the ura3Δ/Δ his1Δ/Δ arg4Δ/Δ strain BWP17 (kindly provided by Aaron Mitchell, Carnegie Mellon University). The doxycycline-repressible tetO-ERG11 strains were made using the system described by Nakayama and colleagues (30). For a detailed description of the construction of each C. albicans gene deletion and doxycycline-repressible strain, please refer to the supplemental material.

Antifungal susceptibility testing.

Antifungal susceptibility testing of all the strains included in this study was performed using the broth microdilution method described in CLSI (Clinical and Laboratory Standards Institute) document M27-A3 (31) in a 96-well plate format. All drugs for susceptibility testing used in this study were diluted in DMSO in 2-fold dilutions at 200 times the final concentration. The following antifungal drugs and concentration ranges were tested in this study: fluconazole (Sigma-Aldrich) from 64 μg/ml to 0.0313 μg/ml, miconazole (Sigma-Aldrich) from 50 μg/ml to 0.024 μg/ml, ketoconazole (Sigma-Aldrich) from 32 μg/ml to 0.015 μg/ml, itraconazole (Sigma-Aldrich) from 16 μg/ml to 0.007 μg/ml, and amphotericin B (LKT Laboratories, Inc.) from 16 μg/ml to 0.001 μg/ml. RPMI 1640 medium (Sigma-Aldrich) was prepared according to the CLSI document; the medium was buffered with morpholinepropanesulfonic acid (MOPS), and the pH was adjusted using KOH and HCl. The cell inoculum was ∼1 × 10³ cells per well. The plates were incubated without shaking at 35°C for 24 or 48 h unless otherwise stated. The content of each well was carefully resuspended by pipetting up and down before OD₆₀₀ was measured using a Biotek Synergy Mx plate reader.

Quantification of cellular ergosterol.

Fifty milliliters of YNB broth plus either 4 μg/ml fluconazole or 0.5% DMSO was inoculated with 2 × 10⁶ C. albicans cells/ml and grown at 35°C for 24 h with shaking at 200 rpm. Total cellular sterols were then extracted using n-heptane by the method of Arthington-Skaggs et al. (32). n-Heptane from the organic phase was evaporated under reduced pressure (0.03 mm Hg) at −20°C. The white powdery residue was dissolved in 0.5 ml CDCl₃ (99.8% D contains 1% [vol/vol] tetramethylsilane [TMS]), and it was subjected to proton nuclear magnetic resonance (NMR) analysis. The NMR was recorded on a Varian 500-MHz Inova instrument. The amounts of ergosterol in these samples were estimated by comparing the NMR integrals of the methyl C-18 group at 0.562 ppm to that of the internal TMS standard. The amount of ergosterol in these samples was calculated by standardizing these integrals with the C-18 methyl integral of the standard ergosterol sample (1 mg/ml; Sigma-Aldrich) that was prepared, and the NMR was recorded by the same procedure.

Fluorescence microscopy to determine vacuole morphology.

Vacuole morphology was determined using C. albicans strains expressing either green fluorescent protein (GFP)-Ypt72p (16) or Cpy1p-GFP (Cpy1p prepropeptide [CPP]-GFP; see supplemental material) fusion protein. In some experiments, cells were prelabeled with the dye FM4-64 to label the vacuolar membrane as previously described (33). Cells were observed with an Olympus BX51 fluorescence microscope with a 100× objective and a fluorescein isothiocyanate (FITC) filter set to detect GFP fluorescence or tetramethyl rhodamine isothiocyanate (TRIT-C) to observe FM4-64. For each field, matching phase-contrast and fluorescence images were acquired using a QImaging MicroPublisher 3 real-time viewing (RTV) camera and the CellSens digital imaging software (Olympus). Adjustments in the brightness and/or contrast of images were made using the CellSens digital imaging software. All images presented for a given time point/condition in each experiment were scaled identically to permit direct comparison.

Inhibition of lanosterol demethylase causes significant vacuolar defects in C. albicans.

Using a high-throughput screening assay, we recently identified a variety of imidazole antifungals, including miconazole and ketoconazole as disrupting the integrity of the C. albicans vacuole (unpublished results). Some imidazoles have been proposed to possess a secondary antifungal mechanism that is independent of their inhibition of ergosterol biosynthesis and that accounts for their fungicidal activity at high concentrations (34–37). We therefore tested whether the fungistatic triazole fluconazole caused similar perturbation of the C. albicans vacuole. This was investigated using a strain expressing a GFP-tagged GTPase, Ypt72p, that localizes to the surface of the vacuole (16). Significant vacuolar disruption was observed at fluconazole concentrations of ≥0.0625 μg/ml. In order to examine the relationship between azole-mediated vacuolar disruption and growth inhibition, a time course experiment was performed. The GFP-Ypt72p-tagged strain was subcultured into YNB medium with and without 1 μg/ml fluconazole, and the growth, cell viability, and vacuolar morphology were compared at 30-min intervals. When growth was measured as OD₆₀₀, statistically significant growth inhibition was initially observed in the presence of fluconazole after 420 min (Fig. 1A). However, when cell viability was compared in the same cultures as CFU, fluconazole was found to significantly reduce CFU versus the drug-free control at the 330-min time point (Fig. 1B). This apparent inconsistency may be explained by the tendency of fluconazole-treated cells to form clumps, indicating a defect in cell separation (38). These changes presumably occur before the rate of biomass production (OD₆₀₀) is affected. In the presence of fluconazole, vacuolar fragmentation was initially observed after 240 min, and severe disruption occurred in all cells by the 270-min time point (Fig. 1C). This precedes the reduction in CFU and OD₆₀₀, suggesting that vacuolar degeneration is an early event following azole treatment and not merely a secondary consequence of prolonged growth arrest. Similar effects on vacuolar integrity were observed using a C. albicans strain expressing a GFP-tagged version of carboxypeptidase Y (see Fig. S1 in the supplemental material), which labels the vacuolar lumen, or when vacuoles were prelabeled with FM4-64 (Fig. S2), a dye that is taken up by endocytosis and accumulates to label the vacuolar membrane.

In order to confirm that the observed vacuolar defects were a direct consequence of Erg11p inhibition rather than an “off-target” effect of the azoles, we constructed a strain in which the transcription of the ERG11 gene could be shut down using a doxycycline-repressible promoter. In the presence of doxycycline, the tetO-ERG11 strain exhibits vacuolar defects similar to those observed following fluconazole treatment (Fig. 1D). Thus, Erg11p inhibition is sufficient to cause loss of vacuolar integrity, presumably as a consequence of ergosterol depletion.

Vps21p-mediated trafficking through the prevacuolar compartment affects fungal growth in the presence of azoles.

To gain further insight into the interaction between the antifungal activity of the azoles and vacuolar integrity, we tested the susceptibility of C. albicans mutants blocked in four distinct vacuolar trafficking steps (see Fig. S3 in the supplemental material), using the standard CLSI (Clinical and Laboratory Standards Institute) protocol for antifungal susceptibility testing (31). Mutants blocked in either autophagy (atg9Δ/Δ) (23) or in a direct Golgi-to-vacuole trafficking pathway (aps3Δ/Δ) (24) have susceptibilities to fluconazole (Fig. 2A and B) and miconazole (see Fig. S4A and S4B in the supplemental material) that are similar to those of the isogenic controls. However, a vps21Δ/Δ mutant lacking a Rab GTPase that regulates membrane fusion and trafficking through the prevacuolar compartment (PVC) is significantly less sensitive to fluconazole (Fig. 2C), miconazole, itraconazole, and ketoconazole (see Fig. S4C, S5A, and S5B in the supplemental material). Interestingly, a ypt72Δ/Δ mutant lacking a Rab GTPase that localizes to the vacuole also shows slightly elevated growth in the presence of fluconazole (Fig. 2D), but the difference is less dramatic than for the vps21Δ/Δ mutant. This suggests that trafficking through the PVC is more important with respect to determining azole sensitivity than trafficking to the vacuole per se.

The susceptibility of the vps21Δ/Δ mutant to amphotericin B, which acts by directly binding to ergosterol, is not significantly different from those of the controls (see Fig. S5C in the supplemental material). Furthermore, the vps21Δ/Δ mutant has a slight increase in susceptibility to the allylamine antifungal terbinafine (Fig. 2E), which inhibits an earlier step of the ergosterol biosynthetic pathway, and to amorolfine (Fig. 2F), which acts downstream of Erg11p. Thus, the phenotype of the vps21Δ/Δ mutant seems to be specific to the azoles, rather than a general “drug resistance” phenotype.

The VPS21 paralogs YPT52 and YPT53 do not significantly affect azole susceptibility.

We recently described two additional Rab GTPases that localize to the PVC, are closely related to, and share significant functional overlap with Vps21p (22). However, the ypt52Δ/Δ and ypt53Δ/Δ mutants lacking these GTPases are unaffected in their susceptibility to miconazole (see Fig. S6A and S6B in the supplemental material), as is a ypt52Δ/Δ ypt53Δ/Δ double mutant (Fig. S6C). While the ypt52Δ/Δ single mutant shows marginally increased growth in the presence of fluconazole compared to the isogenic control strain (Fig. 3A), the ypt53Δ/Δ mutant is unaffected (Fig. 3B). Furthermore, the vps21Δ/Δ ypt52Δ/Δ mutant (Fig. S6E), vps21Δ/Δ ypt53Δ/Δ mutant (Fig. S6F), and a triple GTPase mutant (Fig. 3C) are all affected by fluconazole and miconazole (data not shown) to an extent similar to that of the vps21Δ/Δ single mutant. Thus, despite significant functional overlap between these three Rab GTPases, it seems that the Vps21p GTPase has the greatest effect upon the ability of C. albicans to grow in the presence of the azoles under standard testing conditions.

To further determine whether this phenotype represents a specific function of the Vps21p GTPase or is a feature of the trafficking step controlled by Vps21p, we tested the azole susceptibility of a C. albicans pep12Δ/Δ mutant (25). Pep12p is a t-SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) protein that is a downstream effector of Vps21p and that also facilitates membrane fusion at the PVC (39). The pep12Δ/Δ mutant shows a similar dose response to fluconazole as that described above for the vps21Δ/Δ mutant (Fig. 3D), suggesting that this phenotype is the result of a block in PVC transport rather than a specific function of the Vps21p GTPase.

Elevated growth of the vps21Δ/Δ mutant in the presence of the azoles is independent of the Cdr1p and Mdr1p efflux pumps.

We previously reported that the C. albicans vps21Δ/Δ mutant has reduced rates of endocytosis (16). Many plasma membrane proteins are targeted for degradation within the fungal vacuole following endocytic trafficking via the PVC. Thus, we considered that reduced rates of uptake and degradation may extend the “half-lives” of the well-described Cdr1p and Mdr1p drug efflux pumps at the cell surface, and this may contribute to the reduced sensitivity of the vps21Δ/Δ mutant to fluconazole. To test this, we examined the susceptibility of vps21Δ/Δ cdr1Δ/Δ and vps21Δ/Δ mdr1Δ/Δ double mutants to fluconazole. However, loss of either CDR1 (Fig. 4A) or MDR1 (data not shown) does not significantly affect the growth of the vps21Δ/Δ mutant in the presence of fluconazole. Thus, the elevated growth of the vps21Δ/Δ mutant in the presence of the azoles does not depend upon the activity of either the Mdr1p or Cdr1p efflux pump.

To ensure that the vps21Δ/Δ mutant had not acquired any point mutations within the gene encoding the target enzyme, Erg11p, that could impact azole susceptibility, we amplified and sequenced the ERG11 open reading frame (ORF) from two independently constructed vps21Δ/Δ mutant strains and their isogenic controls. No polymorphisms were detected, and the sequence matched the sequence in the C. albicans genome sequence database. Interestingly, vacuolar disruption was observed in the vps21Δ/Δ mutant upon fluconazole treatment (Fig. 4B), indicating that the mechanisms that enable the vps21Δ/Δ mutant to grow in the presence of the azoles do not depend upon maintaining an intact vacuole.

The C. albicans vps21Δ/Δ mutant exhibits an exaggerated “trailing growth” phenotype.

The dose-response profile of the vps21Δ/Δ mutant to fluconazole is distinct from that of a well-characterized azole-resistant isolate (4) (see Fig. S7A and S7B in the supplemental material). When growth was compared at the earlier time point of 24 h, it could be discerned that the growth of the vps21Δ/Δ mutant was inhibited at fluconazole concentrations similar to those of the control strains, but the degree of growth inhibition was significantly less (Fig. 5A). In addition, above concentrations of 0.5 μg/ml, no further “dose-dependent” decrease in growth was observed. The differential dose-response profiles at 24 versus 48 h (Fig. 5B) are characteristic of the “trailing growth” phenomenon (40). Clinical isolates exhibiting “trailing growth” typically exhibit MICs of ≤1 μg/ml at the 24-h time point, but they show MICs of ≥64 μg/ml at 48 h. Using a strict cutoff of 80% growth inhibition (MIC₈₀), the vps21Δ/Δ mutant would be considered azole resistant at both 24- and 48-h time points; however, if we adopt a less-stringent criteria of 50% growth inhibition (MIC₅₀), the differential response of the vps21Δ/Δ mutant at 24 h versus 48 h conforms to that of “trailing growth.” While trailing isolates are generally thought to be susceptible to azole treatment in vivo (41, 42), the phenomenon can complicate in vitro antifungal susceptibility testing. Furthermore, the mechanisms underlying trailing growth are not well understood. We therefore examined whether the increased growth of the vps21Δ/Δ mutant in the presence of the azoles was true resistance or an exaggerated “trailing growth” phenotype. It has been previously reported that adjusting the pH of the RPMI 1640 growth medium from 7 to a more acidic pH is sufficient to eliminate trailing growth, but not true azole resistance (40). At pH 3, the vps21Δ/Δ mutant still grows slightly more than the control strains in the presence of both fluconazole (Fig. 5C) and miconazole (not shown), but the differential was substantially diminished. It has also been reported that adjusting the incubation temperature from 35°C to either 25 or 42°C eliminates trailing growth (43). The vps21Δ/Δ mutant was completely resensitized to fluconazole at either 25 or 42°C (Fig. 5D and E), further supporting the idea that the elevated growth of the vps21Δ/Δ mutant in the presence of the azoles is an exaggerated form of the trailing growth phenomenon, rather than true azole resistance.

The vps21Δ/Δ mutant is more tolerant of Erg11p inhibition and ergosterol depletion than VPS21⁺ strains.

The dose-response data for the 24-h time point (Fig. 5A) suggest that the potency of the azoles with respect to Erg11p inhibition is unchanged in the vps21Δ/Δ mutant but that the in vitro antifungal efficacy (i.e., impact on growth) is reduced under standard testing conditions. We therefore considered the possibility that the vps21Δ/Δ mutant has a decreased dependence upon ergosterol for growth or that it is more tolerant than the wild type of the toxic sterol intermediates that accumulate upon azole treatment. To test this, we compared cellular ergosterol levels in the vps21Δ/Δ mutant and isogenic control strain in the presence and absence of fluconazole. This confirmed that both the vps21Δ/Δ mutant and control strains are similarly depleted of ergosterol following treatment with fluconazole (Fig. 5F). Furthermore, NMR analysis of cellular n-heptane extracts from mutant and control cultures indicates that there are no significant differences in the sterol intermediates that accumulate upon fluconazole treatment (see Fig. S8A and S8B in the supplemental material). Thus, the elevated growth of the vps21Δ/Δ mutant in the presence of the azoles is unlikely to be accounted for by altered drug uptake, efflux, or target inhibition.

The above results suggest that despite fluconazole inhibiting Erg11p (and ergosterol biosynthesis), the vps21Δ/Δ mutant is able to grow more than its isogenic control under standard MIC testing conditions. To provide further support for this argument, we deleted both VPS21 alleles from the tetO-ERG11 strain described above. This enabled us to repress Erg11p activity at a transcriptional level and independently of the azole drugs. Doxycycline suppressed the growth of the VPS21⁺ control strains, but growth was suppressed to a much lesser extent in the vps21Δ/Δ background (Fig. 5G). Together, these data suggest that the elevated growth of the vps21Δ/Δ mutant in the presence of the azoles relates to a tolerance of Erg11p inhibition and ergosterol depletion, rather than a reduction in Erg11p inhibition.