Identification and properties of plasma membrane azole efflux pumps from the pathogenic fungi Cryptococcus gattii and Cryptococcus neoformans

Abstract

Objectives

Cryptococcus gattii from the North American Northwest (NW) have higher azole MICs than do non-NW C. gattii or Cryptococcus neoformans. Since mechanisms of azole resistance in C. gattii are not known, we identified C. gattii and C. neoformans plasma membrane azole efflux pumps and characterized their properties.

Methods

The C. gattii R265 genome was searched for orthologues of known fungal azole efflux genes, expression of candidate genes was assessed by RT–PCR and the expressed genes' cDNAs were cloned and expressed in Saccharomyces cerevisiae. Azole MICs and intracellular [³H]fluconazole were measured in C. gattii and C. neoformans and in S. cerevisiae expressing each cDNA of interest, as was [³H]fluconazole uptake by post-Golgi vesicles (PGVs) isolated from S. cerevisiae sec6-4 mutants expressing each cDNA of interest.

Results

Intracellular [³H]fluconazole concentrations were inversely correlated with fluconazole MICs only in 25 NW C. gattii strains. S. cerevisiae expressing three C. gattii cDNAs (encoded by orthologues of C. neoformans AFR1 and MDR1 and the previously unstudied gene AFR2) and their C. neoformans counterparts had higher azole MICs and lower intracellular [³H]fluconazole concentrations than did empty-vector controls. PGVs from S. cerevisiae expressing all six Cryptococcus cDNAs also accumulated more [³H]fluconazole than did controls, and [³H]fluconazole transport by all six transporters of interest was ATP dependent and was inhibited by excess unlabelled fluconazole, voriconazole, itraconazole and posaconazole.

Conclusions

We conclude that C. gattii and C. neoformans AFR1, MDR1 and AFR2 encode ABC transporters that pump multiple azoles out of S. cerevisiae cells, thereby causing azole resistance.

Introduction

The environmental fungus Cryptococcus gattii causes pneumonias and CNS infections in both healthy and immunocompromised people.¹ C. gattii was formerly believed to cause serious infections almost exclusively in tropical or subtropical regions of Africa, Asia and Australia,² but an outbreak of serious C. gattii infections on Vancouver Island, British Columbia was reported in 2002.^3–5 Since the occurrence of these outbreak cases, severe C. gattii infections have been frequently reported in western Canada and the northwestern USA.¹ C. gattii infections in the Pacific Northwest (NW) differ in several important respects from C. gattii infections described in other geographic regions. For example, among people with proven C. gattii infections, 41 of 124 from Canada⁶ and 14 of 52 from Washington and Oregon¹ had serious underlying diseases, compared with 0 of 20 people from Australia.⁷ Similarly, mortality in people with proven C. gattii infections was 15 of 124 in Canada⁶ and 15 of 52 in Washington or Oregon,¹ compared with 0 of 20 in people in Australia.⁷

Slow or incomplete response to antifungal therapy has been reported in some patients infected with C. gattii worldwide^7,8 and in C. gattii patients from the NW treated with azoles.¹ There are likely multiple reasons for these poor responses to azole therapy, but the observations that MICs of several azole antifungals are higher in C. gattii than in Cryptococcus neoformans^1,9,10 and also are higher in C. gattii strains with genotypes prevalent in the NW than in C. gattii with other genotypes^10–12 suggest that resistance to azole antifungals may play an important role. The principal mechanisms by which fungi acquire resistance to azole antifungals are overexpression of the gene (ERG11) encoding the azole target enzyme lanosterol 14-α demethylase (Erg11p),^13–16 mutations in ERG11 that decrease the susceptibility of Erg11p to inhibition by azoles^17–22 and/or overexpression of plasma membrane proteins that pump the drug out of the cells.^{16,21,23–35}

In an earlier study, we found that 25 NW C. gattii clinical isolates had ∼2-fold higher geometric mean MICs for multiple azole antifungals than did 34 C. gattii isolates from other regions (non-NW) or 20 C. neoformans isolates. When we examined whether azole resistance in these strains was caused by overexpression of or mutations in the gene (ERG11) encoding the azole target enzyme, ERG11 mRNA levels were higher in both C. gattii groups than in C. neoformans, but these levels were no higher in the NW than in the non-NW C. gattii strains, nor did they correlate with fluconazole MICs within any group. Lastly, we sequenced ERG11 in the 25 NW C. gattii strains, and we found that five variants encoded deduced products that differed at one or more positions from the product of WT ERG11. When the cDNAs derived from these five variants were expressed in a conditional Saccharomyces cerevisiae erg11 mutant, the resulting transformants' azole MICs were no higher than those of controls expressing the WT cDNA. We concluded from these results that neither ERG11 overexpression nor mutations in the ERG11 coding sequence could explain the increased azole MICs observed in NW C. gattii.¹⁰

The possibility that plasma membrane transport proteins may contribute to azole resistance in Cryptococcus was suggested by Venkateswarlu et al.,¹⁷ who showed that C. neoformans clinical strains with high fluconazole MICs had lower intracellular concentrations of [¹⁴C]fluconazole than did strains with lower fluconazole MICs. Since then, two C. neoformans plasma membrane efflux pumps have been identified. C. neoformans MDR1 encodes a deduced membrane transport protein related to eukaryotic multidrug resistance proteins.²⁸ S. cerevisiae cells expressing C. neoformans MDR1 had higher azole MICs than did controls, and a fusion of the C. neoformans Mdr1 protein to GFP localized to the cell surface.³⁶ C. neoformans AFR1 encodes another deduced plasma membrane ABC transporter. C. neoformans cells that overexpressed AFR1 had higher fluconazole MICs than did controls, and mice challenged with AFR1-overexpressing C. neoformans responded less favourably to fluconazole treatment than did mice challenged with WT strains.^29,30,33 Furthermore, targeted disruption of AFR1 in C. neoformans resulted in lower fluconazole MICs and increased susceptibility to fluconazole treatment in infected mice.^30,33 Lastly, S. cerevisiae cells expressing C. neoformans AFR1 showed higher azole MICs than did controls, and a fusion of the C. neoformans Afr1 protein to GFP localized to the cell surface.³⁶

The C. gattii orthologues of C. neoformans AFR1 or MDR1 and their protein products have not been studied, nor is it known if other plasma membrane efflux pumps contribute to azole resistance in C. neoformans or C. gattii. Therefore, we searched the C. gattii genome sequence for orthologues of genes encoding known fungal plasma membrane azole efflux pumps. We found three genes that were expressed and induced by fluconazole in C. gattii clinical strains and whose cDNAs conferred azole resistance and lower intracellular [³H]fluconazole concentrations when they were expressed in S. cerevisiae. We also measured [³H]fluconazole uptake by post-Golgi secretory vesicles (PGVs) isolated from S. cerevisiae AD1-8u-sec6-4 mutants expressing the C. gattii cDNAs of interest and their C. neoformans counterparts to study the catalytic constants, energy requirements and substrate specificities of the C. gattii and C. neoformans proteins of interest.

Materials and methods

Strains and media

The 25 NW and 34 non-NW C. gattii strains and the 24 C. neoformans strains used in this study were collected at the Oregon Health & Science University, or they were obtained from Thomas G. Mitchell (Duke University). Each of the 25 NW C. gattii strains' fluconazole MIC, molecular genotype and deduced Erg11 protein sequence are reported in the paper by Gast et al.¹⁰ Cryptococcus strains were grown on YPD medium (1% yeast extract/2% peptone/2% glucose). S. cerevisiae strains ADΔ and AD1-8u-sec6-4 mutant (Table S1, available as Supplementary data at JAC Online) were obtained from Richard D. Cannon (University of Otago, New Zealand) and were grown YPD or YNB (0.67% yeast nitrogen base without amino acids containing 2% glucose) supplemented with CSM-URA. Plasmids were amplified in Escherichia coli DH5α in LB medium with 100 mg/L ampicillin. Solid media were obtained by adding 1.5% agar.

Antifungal susceptibility testing

The CLSI M27-A3 broth microdilution method³⁷ was used to test S. cerevisiae for antifungal susceptibility, using graded concentrations of fluconazole, voriconazole, itraconazole, posaconazole and amphotericin B. The S. cerevisiae cells were incubated in YNB + glucose that was supplemented with CSM-URA and buffered with 10 mM MOPS/20 mM HEPES–NaOH (pH 7.0). The cultures were scored for presence or absence of visible growth after incubation at 30°C for 48 h.

Intracellular [³H]fluconazole accumulation

Intracellular [³H]fluconazole was quantified as described,²⁶ with modifications. Cryptococcus strains and S. cerevisiae ADΔ transformants were grown in liquid YPD to mid-logarithmic phase. Cells were harvested by centrifugation, washed in YNB and resuspended to OD₆₀₀ = 30 for Cryptococcus or 10 for S. cerevisiae ADΔ in YNB containing [³H]fluconazole (specific activity: 20 Ci/mmol, Amersham Biosciences) at final concentrations of 50 nM for Cryptococcus or 10 nM for S. cerevisiae ADΔ transformants. After the cells were incubated at 30°C for Cryptococcus or 25°C for S. cerevisiae ADΔ, aliquots were removed at intervals and added to cold stop solution (YNB plus 25 μM unlabelled fluconazole), the cells were collected on 0.45 μm nitrocellulose filters, they were washed twice with cold stop solution and intracellular [³H]fluconazole was quantified by liquid scintillation counting.

Identification of transmembrane transporters in C. neoformans and C. gattii

The genome sequences of C. gattii strain R265 and C. neoformans strain H99 (http://www.broadinstitute.org/scientific-community/science/projects/fungal-genome-initiative/fungal-genome-initiative) were queried using tBLASTp for genes encoding deduced proteins similar to the known fungal azole efflux proteins encoded by Candida albicans CDR1, C. albicans CDR2, C. neoformans AFR1 and C. neoformans MDR1. Potential orthologues were identified based on amino acid identity (≥30%) and similarity (≥60%) to the query sequence by global pairwise alignment.

Expression analysis

Cryptococcus strains were grown in YPD medium to mid-logarithmic phase and exposed to fluconazole at 30°C under conditions previously described.¹⁰ Cells were grown in the absence of the drug until mid-log phase and incubated in 8 mg/L fluconazole for 2 h. RNA was extracted with the Aurum Total RNA Fatty and Fibrous Tissue Kit (Bio-Rad). RT–PCR was carried out using the iScript One-Step RT–PCR Kit with SYBR Green (Bio-Rad), and PCR products were amplified using oligonucleotides listed in Table S2.

Plasmid construction and expression in S. cerevisiae strains ADΔ and AD1-8u-sec6-4 mutant

RT–PCR with the designed oligonucleotides (Table S2) were used to amplify: the C. neoformans AFR1, AFR2, MDR1, CNAG_06348 cDNAs from total RNA from C. neoformans H99 and the C. gattii AFR1, AFR2, MDR1 and CNBG_5107 cDNAs from total RNA from C. gattii R265. The PCR products were ligated into the PacI and NotI restriction sites in plasmid pABC3 (Table S3).³⁶ GFP tagging of the C-terminus of each protein was performed by amplifying PCR fragments using the specific oligonucleotides (Table S2) and ligated into PacI and NotI sites on pABC3-GFP (Table S3). Recombinant pABC3- and pABC3-GFP plasmids were digested with AscI and transformed into ADΔ and AD1-8u-sec6-4 S. cerevisiae cells using the Alkali-Cation™ Yeast Kit (MP Biomedicals). Colony PCR was performed to verify proper integration of each transformation cassette at the chromosomal PDR5 locus.

Isolation and properties of PGVs

S. cerevisiae AD1-8u-sec6-4 mutants transformed with pABC3-CnAFR1, pABC3-CnAFR2, pABC3-CneMDR1, pABC3-CgAFR1, pABC3-CgAFR2, pABC3-CgMDR1 or pABC3 empty plasmid (Table S1) were grown for vesicle accumulation as described by Ruetz and Gros.^38,39 Transformants were grown overnight at 25°C in 0.67% YNB supplemented with Complete Supplement Mixture uracil (YNB-ura) and expanded in YPD at 25°C to mid-logarithmic phase (OD₆₀₀ = 2.0). PGV accumulation was induced by shifting the temperature to 37°C and grown for 2 h. Growth was stopped by the addition of 10 mM NaN₃ and cultures were transferred to ice. Cells were collected by centrifugation (10 000 g for 10 min), washed once in 10 mM Tris-HC1, pH 7.5, 5 mM NaN₃ and the cell pellets stored at –80°C.

PGVs were isolated as described by Ruetz and Gros,^38,39 with modifications. Frozen cells were resuspended in 100 mM Tris-SO₄ (pH 9.4) at 25°C and collected by centrifugation, then incubated in SM buffer (1.4 M sorbitol/20 mM HEPES–KOH, pH 7.0) with Zymolyase 20T (25 mg/g wet cells) supplemented with 10 mM NaN₃, 2 mM EDTA and 40 mM β-mercaptoethanol for 1 h, 37°C. Spheroplasts were washed twice in SM buffer with 10 mM NaN_3, and incubated on ice in SM buffer (5 mL/g of cells) with 1 mM CaCl₂, 5 mM MnSO₄ and concanavalin A (1.5 mg/g of cells) for 15 min. Spheroplasts were collected by centrifugation and washed twice with cold SM buffer, after which they were incubated in hypotonic lysis buffer [0.6 M sorbitol/20 mM HEPES–KOH, pH 7.0/2 mM EDTA/PI cocktail for use with fungal and yeast extracts (Sigma-Aldrich, diluted 1: 100)] for 10 min on ice. The spheroplasts were disrupted by Dounce homogenization (25 strokes) and the lysate was centrifuged twice (10 000 g, 10 min, 4°C). The PGV-enriched fraction was collected by centrifugation twice (100 000 g, 45 min, 2°C) and resuspended in nitrate vesicle buffer (50 mM sucrose/10 mM Tris-HEPES, pH 7.5/100 mM potassium nitrate) with 5 mM EGTA. The final pellet was resuspended in nitrate vesicle buffer. Total protein concentrations in PGV samples were determined by the Bradford method.⁴⁰

PGV samples were adjusted to 0.5 mg of total protein/mL in nitrate buffer. [³H]Fluconazole uptake was initiated by adding PGV samples to pre-warmed (37°C) nitrate buffer supplemented with NaATP (2.5 mM), creatine phosphate (10 mM), creatine phosphokinase (3 mg/L) and [³H]fluconazole (0.05 μM). [³H]Fluconazole transport was stopped by adding ice-cold stop solution (200 mM sucrose, 10 mM Tris-HCl, pH 7.5, 25 μM fluconazole), after which the PGVs were collected as described previously. After two additional washes with stop solution, the radioactivity was determined by liquid scintillation counting.

Results

[³H]Fluconazole accumulation in Cryptococcus cells

When the 25 NW C. gattii, 34 non-NW C. gattii and 24 C. neoformans strains described in an earlier study¹¹ were incubated in 50 nM [³H]fluconazole for 30 or 60 min, the mean of intracellular [³H]fluconazole concentrations did not differ significantly in the three groups (Figure 1a). However, there was a significant inverse correlation between the intracellular [³H]fluconazole concentrations and the fluconazole MICs in the NW C. gattii strains at 30 min (Figure 1b) and 60 min (data not shown), but not in the non-NW C. gattii or in C. neoformans. In contrast, intracellular [³H]fluconazole concentrations did not correlate with amphotericin MICs in any of the three groups (data not shown).

Figure 1.

Intracellular [³H]fluconazole accumulation in NW C. gattii, non-NW C. gattii and C. neoformans. (a) Intracellular [³H]fluconazole concentrations after incubation in 50 nM [³H]fluconazole for 30 min were not significantly different in the three groups by paired t-test (data shown as means ± SD, n = 3 experiments). (b) Intracellular [³H]fluconazole concentrations were inversely correlated with fluconazole MICs in NW C. gattii strains, but not for the strains from other groups. The two-tailed Mann–Whitney U-test was used to compare intracellular [³H]fluconazole concentrations between groups, and regression analysis was performed to correlate intracellular [³H]fluconazole concentrations with fluconazole MICs within groups.

Identification and heterologous expression of potential azole efflux pumps

Since these results suggested that plasma membrane efflux pumps may contribute to fluconazole resistance in C. gattii, we searched the genome sequence of NW C. gattii strain R265 for ORFs that encode deduced proteins similar to the C. albicans azole efflux pumps Cdr1p or Cdr2p or to the C. neoformans azole efflux pumps Afr1p or Mdr1p. We found seven ORFs that encoded deduced proteins that were at least 60% similar and 30% identical by global pairwise alignment to one or more of the query sequences (Figure 2a and Table 1). To determine whether these seven genes of interest were expressed in NW C. gattii, we quantified the corresponding mRNAs by RT–PCR in total RNA extracted from four NW C. gattii clinical isolates with fluconazole MICs ≤16 mg/L and four with fluconazole MICs ≥32 mg/L. The C. gattii genes CNBG_1200 (which is orthologous to C. neoformans AFR1 and is referred to as C. gattii AFR1 hereafter) and CNBG_6088 (referred as C. gattii AFR2 hereafter) were expressed in all eight strains in the absence of fluconazole, and expression of both of these genes was induced by exposure to 8 mg/L fluconazole for 2 h. The genes CNBG-1138 (which is orthologous to C. neoformans MDR1 and is referred to as C. gattii MDR1 hereafter) and CNBG_5107 were also expressed in all eight strains in the absence of fluconazole, but their expression was not induced by fluconazole. In contrast, the mRNAs encoded by C. gattii genes CNBG_5117, CNBG-4708 and CNBG_5385 were not detected in total RNA extracted from any strain under the conditions tested (Figure 2b). The AFR1, AFR2 and MDR1 orthologues were also expressed in both the presence and absence of fluconazole in four of four non-NW C. gattii strains with fluconazole MICs ≤16 mg/L, four of four non-NW C. gattii strains with fluconazole MICs ≥32 mg/L, four of four C. neoformans strains with fluconazole MICs ≤16 mg/L and four of four C. neoformans strains with fluconazole MICs ≥32 mg/L. However, AFR1 expression was not induced by fluconazole exposure in non-NW C. gattii, and neither AFR1 nor AFR2 expression was induced by fluconazole exposure in C. neoformans (Figure 2c). Lastly, expression of AFR1, AFR2 and MDR1 did not correlate with the fluconazole MICs in the groups of NW C. gattii, non-NW C. gattii or C. neoformans strains we studied (data not shown).

Figure 2.

Membrane transporter candidates in C. gattii. C. gattii orthologues of known fluconazole efflux pumps. (a) CLUSTALW (http://align.genome.jp/) protein phylogenetic tree of seven putative ABC transporters from the Broad Institute's R265 Genome Database with proteins encoded by CaCDR1, CaCDR2, CnAFR1 and CnMDR1. The scale bar corresponds to 0.1 amino acid changes per site. (b) The ratios of each mRNA of interest to ACT1 mRNA were determined in four azole-susceptible and four azole-resistant NW C. gattii strains incubated in the absence or presence of fluconazole (8 mg/L) for 2 h. CNBG_5117, CNBG_4708 and CNBG_5385 mRNAs were not detectable under the conditions tested. (c) The ratios of each mRNA of interest to ACT1 mRNA were determined in four azole-susceptible and four azole-resistant NW C. gattii, non-NW C. gattii and C. neoformans strains incubated in the absence or presence of fluconazole (8 mg/L) for 2 h. mRNA ratios that were higher in cells exposed to fluconazole than in fluconazole-unexposed controls are shown by asterisks (P < 0.05 by paired t-test; data shown as means ± SD, n = 3 experiments).

Table 1.

Transmembrane transporters analysed in this study

Name	GI/Broad	Size (amino acids)	MW (kDa)	Identity (%) to CaCdr1p	Identity (%) to CaCdr2p	Orthologue	Orthologue identity (%)	ABC	TMD
CnAFR1	18478280	1543	172.49	33	33	CgAFR1	95	2	12
CgAFR1	CNBG_1200	1542	172.93	33	33	CnAFR1	95	2	12
CneMDR1	2668555	1408	152.10	11	9	CgMDR1	92	2	10
CgMDR1	CNBG_1138	1408	152.18	12	8	CneMDR1	92	2	10
CnAFR2	CNAG_00869	1529	169.77	48	46	CgAFR2	92	2	12
CgAFR2	CNBG_6088	1529	170.36	48	46	CnAFR2	92	2	12
CNAG_06348	CNAG_06348	1425	158.69	30	31	CNBG_5107	91	2	11
CNBG_5107	CNBG_5107	1410	157.07	31	32	CNAG_06348	91	2	11
CaCDR1	68465695	1501	169.94	100	84	CaCDR2	84	2	12
CaCDR2	68465615	1499	168.95	84	100	CaCDR1	84	2	12
CNBG_5117	CNBG_5117	1409	156.73	31	32	CNAG_06338	91	2	11
CNBG_4708	CNBG_4708	1441	161.08	32	33	CNAG_07799	86	2	10
CNBG_5385	CNBG_5385	1403	156.41	47	46	CNAG_04098	91	2	11
CNAG_06338	CNAG_06338	1420	158.25	31	32	CNBG_5117	91	2	11
CNAG_07799	CNAG_07799	1448	161.20	32	33	CNBG_4708	86	2	10
CNAG_04098	CNAG_04098	1509	168.26	47	45	CNBG_5385	91	2	12

GI/Broad, GenBank identification number in the protein database (http://www.ncbi.nlm.nih.gov/protein) or identification number in the Broad Institute database (http://www.broadinstitute.org/scientific-community/science/projects/fungal-genome-initiative/fungal-genome-initiative). MW, molecular weight calculated by Bioedit Sequence Alignment Editor. ABC, ATP-binding cassette domain predicted by the SMART program (http://smart.embl-heidelberg.de/). TMD, conserved transmembrane domain predicted by the TMHMM program (http://www.cbs.dtu.dk/services/TMHMM/).

Orthologue identity was calculated by CLUSTALW (http://align.genome.jp/).

We next examined whether the four expressed C. gattii genes or their C. neoformans orthologues encode functional azole transporters by: (i) cloning these genes' cDNAs by RT–PCR and ligating them into plasmid pABC3; (ii) introducing the resulting plasmids into an S. cerevisiae strain (ADΔ) that is highly susceptible to azoles because it lacks seven plasma membrane ABC transporters;³⁶ and (iii) testing the resulting transformants for susceptibility to antifungal drugs. S. cerevisiae ADΔ cells expressing the C. gattii AFR1, C. gattii MDR1 and CNBG_6088 (referred to as C. gattii AFR2 hereafter) cDNAs and also the corresponding C. neoformans cDNAs (AFR1, MDR1 and AFR2, respectively) were more resistant to all antifungal azoles tested than were empty-vector controls, whereas S. cerevisiae ADΔ cells expressing the C. gattii CNBG_5107 cDNA or its C. neoformans counterpart (CNAG_06348) were not. In contrast, expression of none of these cDNAs affected the S. cerevisiae transformants' susceptibility to the polyene antifungal amphotericin B (Table 2).

Table 2.

Effects of AFR1, AFR2 or MDR1 overexpression (with or without C-terminal GFP tagging) on the susceptibility of S. cerevisiae ADΔ transformants to antifungal drugs

Plasmid	MIC (mg/L)
	fluconazole	voriconazole	itraconazole	posaconazole	amphotericin B
pABC3 (no insert)	0.5	0.0625	0.25	0.125	2
pABC3-CNAG_06348	0.5	0.0625	0.25	0.125	2
pABC3-CNBG_5107	0.5	0.0625	0.25	0.125	2
pABC3-CnAFR1	4	0.5	2	1	2
pABC3-CnAFR1-GFP	1	0.125	0.5	0.5	2
pABC3-CnAFR2	2	0.25	1	0.5	2
pABC3-CnAFR2-GFP	2	0.25	1	0.5	2
pABC3-CneMDR1	8	1	>4	>4	2
pABC3-CneMDR1-GFP	2	0.5	1	1	2
pABC3-CgAFR1	4	0.5	>4	2	2
pABC3-CgAFR1-GFP	1	0.125	0.5	0.5	2
pABC3-CgAFR2	8	1	2	2	2
pABC3-CgAFR2-GFP	4	0.25	1	1	2
pABC3-CgMDR1	4	0.5	>4	>4	2
pABC3-CgMDR1-GFP	2	0.125	1	1	2

To localize the proteins of interest, the C. neoformans or C. gattii AFR1, AFR2 and MDR1 cDNAs were ligated into plasmid pABC3-GFP, which resulted in in-frame fusions of GFP to each protein's C-terminus. When S. cerevisiae ADΔ cells transformed with these plasmids were examined by confocal fluorescence microscopy, all six fluorescent proteins localized primarily to the cell surfaces, with partial localization to other intracellular structures (Figure 3). Lastly, S. cerevisiae expressing the GFP-labelled forms of the proteins of interest often had lower azole MICs than did S. cerevisiae expressing the unlabelled forms of the same proteins, but these MICs were higher than those empty-vector controls in all cases (Table 2).

Figure 3.

Confocal microscopy images of S. cerevisiae ADΔ transformed with membrane pumps from C. gattii and C. neoformans. S. cerevisiae cells transformed with pABC3-CnAFR1-GFP (a and d), pABC3-CnAFR2-GFP (b and e), pABC3-CneMDR1-GFP (c and f), pABC3-CgAFR1-GFP (g and j), pABC3-CgAFR2-GFP (h and k) and pABC3-CgMDR1-GFP (i and l) showed GFP localization on the cell surface. Differential interference contrast (a–c and j–l) images of the transformed cells are shown above their respective confocal microscopy images (d–f and j–l).

Intracellular [³H]fluconazole concentrations in S. cerevisiae

The observations that heterologous expression of the C. neoformans or C. gattii AFR1, AFR2 and MDR1 proteins of interest in S. cerevisiae resulted in increased azole MICs and that fluorescently tagged versions of all six proteins localized mostly to the cell surface suggested that these proteins functioned as azole efflux pumps, but this evidence was indirect. Therefore, to determine more directly if the Cryptococcus proteins of interest pump azoles out of the cell, we quantified intracellular [³H]fluconazole at intervals after the S. cerevisiae ADΔ transformants of interest were incubated in 10 nM [³H]fluconazole. We found that S. cerevisiae expressing C. neoformans or C. gattii AFR1, C. neoformans or C. gattii MDR1, and C. gattii or C. neoformans AFR2 cDNAs accumulated substantially less intracellular [³H]fluconazole than did empty-vector controls, whereas S. cerevisiae cells expressing the CNAG_06348 or CNBG_5107 cDNAs did not (Figure 4).

Figure 4.

Intracellular [³H]fluconazole accumulation in S. cerevisiae cells overexpressing Cryptococcus membrane transporters. Intracellular [³H]fluconazole concentrations after incubation in [³H]fluconazole (10 nM, 25°C) for the intervals shown were higher in S. cerevisiae cells transformed with empty plasmid (pABC3), pABC3-CNAG_06348 (CNAG_06348) and pABC3-CNBG_5107 (CNBG_5107) than cells transformed with pABC3-CnAFR1 (CnAFR1), pABC3-CnAFR2 (CnAFR2) and pABC3-CneMDR1 (CneMDR1) (a) and pABC3-CgAFR1 (CgAFR1), pABC3-CgAFR2 (CgAFR2) and pABC3-CgMDR1 (CgMDR1) (b) (data shown as means ± SD, n = 3 experiments).

[³H]Fluconazole uptake by S. cerevisiae PGVs

We concluded from the results above that C. gattii or C. neoformans Afr1p, Afr2p or Mdr1p can pump azoles out of intact S. cerevisiae cells, but the inaccessibility of the cytoplasmic face of the plasma membranes of whole cells precludes detailed examination of the transport properties of the proteins of interest. Since the plasma membrane and PGV membranes in eukaryotic cells are oriented in the opposite directions, ABC and MFS transporters that pump their substrates out of whole fungal cells pump their substrates into the lumens of isolated PGVs.^23,38,39 Therefore, we expressed the C. neoformans or C. gattii AFR1, AFR2 and MDR1 cDNAs in a S. cerevisiae ADΔ strain that also bears the temperature-sensitive sec6-4 mutation,³⁶ shifted the transformants to 37°C to block fusion of PGVs to the plasma membrane, isolated the PGVs that accumulated in the cytoplasm from lysed spheroplasts, purified them by differential centrifugation, and verified that the final subcellular fraction contained abundant intact PGVs transmission electron microscopy (data not shown).

PGVs isolated from S. cerevisiae that overexpressed C. neoformans or C. gattii AFR1, AFR2 or MDR1 all accumulated substantially more [³H]fluconazole than did PGVs from empty-vector controls (Figure 5a and c). Transport of [³H]fluconazole into isolated PGVs by all six C. gattii and C. neoformans proteins of interest conformed to Michaelis–Menten kinetics (Figure 5b and d). C. gattii Afr1p, Afr2p and Mdr1p had 2.5–2.8-fold lower K_M values and 13–20-fold higher V_max values than did C. neoformans Afr1p, Afr2p and Mdr1p (Table 3). [³H]Fluconazole uptake by PGVs isolated from S. cerevisiae overexpressing C. neoformans or C. gattii AFR1, AFR2 and MDR1 was much less in the absence of ATP than in its presence, and it was inhibited by the non-hydrolysable ATP analogue 5′-adenylyl-β-γ-imidodiphosphate (AMP-PNP) and by the ATPase inhibitor orthovanadate, but not by the proton ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP) (Figure 6a). Lastly, 50-fold molar excesses of unlabelled fluconazole, voriconazole, itraconazole and posaconazole substantially inhibited uptake of [³H]fluconazole uptake by PGVs isolated from S. cerevisiae expressing all six Cryptococcus cDNAs of interest, whereas amphotericin B did not (Figure 6b).

Figure 5.

[³H]Fluconazole accumulation by PGVs. PGVs isolated from S. cerevisiae AD1-8u-sec6-4 cells transformed with pABC3-CnAFR1 (CnAFR1), pABC3-CnAFR2 (CnAFR2) and pABC3-CneMDR1 (CneMDR1) (a) and pABC3-CgAFR1 (CgAFR1), pABC3-CgAFR2 (CgAFR2) and pABC3-CgMDR1 (CgMDR1) (c) accumulated substantially more [³H]fluconazole after incubation in 0.05 μM [³H]fluconazole for the intervals shown than did PGVs from pABC3-transformed controls (pABC3) (data shown as means ± SD, n = 3 experiments). Lineweaver–Burk plots of initial rates of [³H]fluconazole uptake by PGVs containing C. neoformans (b) or C. gattii (d) Afr1p, Afr2p and Mdr1p.

Table 3.

Catalytic properties of C. gattii and C. neoformans Afr1p, Afr2p and Mdr1p in [³H]fluconazole transport

	KM (μM)	Vmax (pmol/mg total protein/min)
CnAfr1p	0.26 ± 0.05	3.22 ± 0.21
CnAfr2p	0.13 ± 0.02	2.60 ± 0.29
CneMdr1p	0.11 ± 0.04	1.46 ± 0.90
CgAfr1p	0.65 ± 0.11	65.60 ± 0.27
CgAfr2p	0.36 ± 0.07	33.98 ± 2.23
CgMdr1p	0.31 ± 0.04	23.40 ± 0.39

Figure 6.

Energy dependence and azole inhibition of [³H]fluconazole transport. PGVs isolated from S. cerevisiae AD1-8u-sec6-4 cells transformed with pABC3-CnAFR1 (CnAFR1), pABC3-CnAFR2 (CnAFR2), pABC3-CneMDR1 (CneMDR1), pABC3-CgAFR1 (CgAFR1), pABC3-CgAFR2 (CgAFR2) and pABC3-CgMDR1 (CgMDR1) were tested for energy requirements (a) and antifungal inhibition tests (b). (a) PGVs overexpressing Afr1p, Afr2p and Mdr1p accumulated much more [³H]fluconazole after 10 s in buffer + ATP than in buffer without ATP or in buffer + ATP + AMP-PNP or the ATP inhibitor sodium orthovanadate (VAN). In contrast, the proton ionophore CCCP had little effect on [³H]fluconazole transport by Afr1p, Afr2p or Mdr1p (data shown as means ± SD, n = 3). (b) [³H]Fluconazole uptake after 10 s by PGVs overexpressing Afr1p, Afr2p and Mdr1p was quantified in the presence of 50-fold molar excesses of the compounds listed (data shown as means ± SD, n = 3 experiments).

Discussion

The major objectives of this study were to identify plasma membrane drug efflux pumps that may contribute to azole resistance in C. gattii and to characterize the catalytic properties of these pumps and their C. neoformans counterparts. We decided to study azole efflux pumps for several reasons. First, although several groups have reported more azole resistance in NW C. gattii than in C. neoformans or non-NW C. gattii,^11,12 the mechanisms responsible for azole resistance in NW C. gattii are not known. Second, we found in an earlier study that neither overexpression of the azole target gene ERG11 nor mutations in ERG11 coding sequences could explain the high azole MICs observed in many NW C. gattii strains.¹⁰ Third, although there is considerable evidence that the C. neoformans proteins Afr1p and Mdr1p pump azole antifungals out of the cell, azole efflux pumps have not been identified in C. gattii. Lastly, there have been no detailed studies of the catalytic properties of any C. neoformans or C. gattii plasma membrane azole efflux pump.

To explore whether azole efflux pumps play a role in azole resistance, we measured intracellular [³H]fluconazole concentrations in 25 NW C. gattii strains, 34 non-NW C. gattii strains and 24 C. neoformans strains. We found no significant differences in mean [³H]fluconazole concentrations in the three groups, but there was a significant inverse correlation between the intracellular [³H]fluconazole concentrations and the fluconazole MICs in the NW C. gattii strains, but not the other groups. Previous studies have correlated azole resistance with intracellular concentrations of rhodamine 6G (a fluorescent substrate for fungal azole efflux pumps)^41,42 or [³H]fluconazole,²³ but we are unaware that intracellular concentrations of labelled fluconazole have previously been shown to correlate with azole MICs in a large group of clinical isolates.

Since our results supported the possibility that azole efflux pumps may contribute to azole resistance in NW C. gattii, we searched the NW C. gattii strain R265 genome for genes whose deduced products were similar to known fungal azole efflux pumps. We found seven candidate genes, and four of these were expressed in eight of eight C. gattii clinical isolates. Lamping et al.³⁶ showed in 2007 that overexpression of several fungal genes or cDNAs encoding azole efflux pumps (including C. neoformans Afr1p and Mdr1p) in S. cerevisiae ADΔ resulted in increased azole MICs and decreased intracellular concentrations of the fluorescent dye rhodamine 6G and also that GFP-labelled versions of fungal proteins of interest localized to the surfaces of the S. cerevisiae transformants. We used the system developed by Lamping et al.³⁶ to overexpress the C. gattii AFR1, AFR2 and MDR1 cDNAs and the orthologous C. neoformans cDNAs in S. cerevisiae ADΔ, and we found that the resulting transformants had higher MICs for multiple azoles and lower intracellular [³H]fluconazole concentrations than did empty-vector controls. We also found that the GFP-labelled versions of all six Cryptococcus proteins of interest: (i) localized mostly to the peripheries of the S. cerevisiae ADΔ transformants; and (ii) retained at least some of their abilities to increase azole MICs. These results constitute strong evidence that the proteins encoded by C. gattii and C. neoformans AFR1, AFR2 and MDR1 increased azole MICs by pumping multiple azoles out of the S. cerevisiae cells, but they did not rule out completely the alternative possibility that these proteins acted by interfering with azole uptake.

To address this last possibility and also to study the catalytic properties of the Cryptococcus proteins of interest, we quantified [³H]fluconazole uptake by PGVs isolated from S. cerevisiae AD1-8u-sec6-4 mutants that overexpressed each of the six C. gattii and C. neoformans cDNAs of interest. We found that [³H]fluconazole transport into the PGV lumens by all six proteins of interest was rapid, conformed to Michaelis–Menten kinetics, required ATP, was inhibited by the non-hydrolysable ATP analogue AMP-PNP, and was not inhibited by the proton ionophore CCCP. It is clear from these results that C. gattii and C. neoformans Afr1p, Afr2p and Mdr1p are plasma membrane ABC transporters that pumped multiple azoles out of S. cerevisiae cells.

One unexpected finding was that the three C. gattii azole transporters had much higher V_max values than the corresponding C. neoformans transporters, whereas the two species' K_M values were similar. The V_max values of the six Cryptococcus azole efflux pumps were also substantially higher than those of C. albicans Cdr1p and Cdr2p (0.91 ± 0.15 and 0.52 ± 0.10 pmol/mg protein/min, respectively), which are the only other fungal azole efflux pumps for which comparable catalytic constants are known.²³ It is possible that the C. gattii transporters' high V_max values may contribute to the higher azole MICs in NW C. gattii than in C. neoformans that we observed. However, these V_max values were expressed in terms of total protein in the PGV preparations. Insufficient amounts of the recombinant proteins of interest were present in extracts of whole S. cerevisiae cells or in isolated S. cerevisiae PGVs to be detected by SDS–PAGE. Efforts to quantify the GFP- or FLAG-labelled forms of each protein of interest by probing immunoblots with the monoclonal antibodies to GFP and to the FLAG epitope were also unsuccessful. Therefore, although we used the same method and a single host S. cerevisiae strain to generate the PGVs we used in the study, the observed differences in V_max values between may have been due to differences in such factors as translation efficiency or protein stability in a heterologous host rather than inherent differences in these proteins' abilities to transport azoles across membranes.

The observation that overexpression of all six Cryptococcus cDNAs of interest conferred resistance to multiple azoles in S. cerevisiae suggested that the efflux pumps encoded by these six cDNAs transport multiple azoles out of the cell. Since radiolabelled voriconazole, itraconazole and posaconazole are not available, we could not test this hypothesis directly. However, excess voriconazole, itraconazole and posaconazole inhibited [³H]fluconazole uptake by PGVs isolated from S. cerevisiae overexpressing all six Cryptococcus cDNAs of interest at least as well as did unlabelled fluconazole, as would be expected for alternative transport substrates. Thus, we concluded that C. gattii and C. neoformans Afr1p, Afr2p and Mdr1p can all transport multiple azoles across fungal membranes. In an earlier study, we found that heterologous expression of the recombinant Erg11 proteins from NW C. gattii, non-NW C. gattii and C. neoformans in a tetracycline-repressible S. cerevisiae erg11 mutant resulted in similar MICs for multiple azoles;¹⁰ our studies to date of three C. gattii azole efflux pumps and of the azole target enzyme Erg11p do not support the practice of using alternative azole antifungals in an attempt to overcome high fluconazole MICs in NW C. gattii infections.

In summary, we have shown the C. gattii and C. neoformans proteins Afr1p, Afr2p and Mdr1p pump fluconazole and other azoles out through the plasma membranes of whole S. cerevisiae cells, thereby causing azole resistance. We also showed that transport of [³H]fluconazole into S. cerevisiae PGVs by the six Cryptococcus proteins of interest required ATP, conformed to Michaelis–Menten kinetics, was inhibited by AMP-PNP but not by CCCP, and was inhibited by excess unlabelled fluconazole, voriconazole, itraconazole and posaconazole. Previous studies by others have shown that Afr1p contributes to azole resistance and virulence in C. neoformans, but the functions of Afr1p in C. gattii and of Afr2p and Mdr1p in C. gattii and C. neoformans are not yet known. Since we examined only the effects of heterologous production of recombinant Cryptococcus proteins in S. cerevisiae, our results do not establish a role for any of these proteins in azole resistance in C. gattii or C. neoformans. The expression studies suggested that one or more of the pumps of interest may contribute to azole resistance, but we did not find a statistically significant correlation between expression of any of the genes of interest and azole MICs in limited numbers of strains. Therefore, to define these proteins' functions in C. gattii and to assess their contributions to azole resistance, we are currently endeavouring to construct and characterize the phenotypes of C. gattii afr1, afr2 and mdr1 null mutants. We also plan to analyse in detail the effects of fluconazole exposure on expression of AFR1, AFR2 and MDR1 in WT C. gattii and in the afr1, afr2 and mdr1 null mutants.

Funding

This work was supported by NIH U54 AI-811680 and by a grant from the Medical Research Council of Oregon.

Transparency declarations

None to declare.

Supplementary data

Tables S1–S3 are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).