Contributions of Aspergillus fumigatus ATP-Binding Cassette Transporter Proteins to Drug Resistance and Virulence

Abstract

In yeast cells such as those of Saccharomyces cerevisiae, expression of ATP-binding cassette (ABC) transporter proteins has been found to be increased and correlates with a concomitant elevation in azole drug resistance. In this study, we investigated the roles of two Aspergillus fumigatus proteins that share high sequence similarity with S. cerevisiae Pdr5, an ABC transporter protein that is commonly overproduced in azole-resistant isolates in this yeast. The two A. fumigatus genes encoding the ABC transporters sharing the highest sequence similarity to S. cerevisiae Pdr5 are called abcA and abcB here. We constructed deletion alleles of these two different ABC transporter-encoding genes in three different strains of A. fumigatus. Loss of abcB invariably elicited increased azole susceptibility, while abcA disruption alleles had variable phenotypes. Specific antibodies were raised to both AbcA and AbcB proteins. These antisera allowed detection of AbcB in wild-type cells, while AbcA could be visualized only when overproduced from the hspA promoter in A. fumigatus. Overproduction of AbcA also yielded increased azole resistance. Green fluorescent protein fusions were used to provide evidence that both AbcA and AbcB are localized to the plasma membrane in A. fumigatus. Promoter fusions to firefly luciferase suggested that expression of both ABC transporter-encoding genes is inducible by azole challenge. Virulence assays implicated AbcB as a possible factor required for normal pathogenesis. This work provides important new insights into the physiological roles of ABC transporters in this major fungal pathogen.

INTRODUCTION

The filamentous fungal pathogen Aspergillus fumigatus is the most common cause of invasive mold infection in humans, and it is associated with an alarmingly high mortality rate. Some triazole antifungal drugs (voriconazole and itraconazole) inhibit the growth of A. fumigatus in vitro and are effective in treatment of A. fumigatus infections; however, development of resistance to these chemotherapeutics is a growing concern (1). While alterations in the cyp51A gene, which encodes the enzymatic target of azole drugs, are commonly found, recent studies have provided evidence that other mechanisms of resistance are also present. One of the most common routes of azole tolerance in other fungal pathogens involves the overproduction of a drug efflux pump, often of the ATP-binding cassette (ABC) transporter family (reviewed in reference 2). These azole resistance transporters are of the ABCG class of ABC transporters and are found in pathogenic yeasts like Candida albicans and Candida glabrata, in addition to Saccharomyces cerevisiae.

The major S. cerevisiae ABCG azole transporter is the Pdr5 protein (3–5). This plasma membrane-localized ABC transporter protein is overproduced in multidrug-resistant cells as a result of transcriptional activation by the related Pdr1 and/or Pdr3 zinc cluster-containing transactivator proteins (reviewed in references 6 and 7). Pdr5 is thought to act as a broad-specificity ATP-dependent drug efflux transporter (8). More recent evidence suggests that Pdr5 acts via control of phospholipid asymmetry in the plasma membrane in cooperation with another plasma membrane-localized ABC transporter called Yor1. The YOR1 gene is also controlled by Pdr1 and Pdr3 but produces an ABCC class transporter (9–11).

Extensive analyses with the pathogenic yeast species C. albicans and C. glabrata have demonstrated that these organisms, like S. cerevisiae, also express ABCG class transporters that are highly transcriptionally upregulated to produce an azole-tolerant phenotype (12–14). These transporters share a high degree of sequence similarity with S. cerevisiae Pdr5 (ScPdr5) and are referred to as C. albicans Cdr1 (CaCdr1) or C. glabrata Cdr1 (CgCdr1).

The role of ABC transporters in azole resistance in A. fumigatus is less clear-cut. A large body of evidence has accumulated demonstrating the occurrence of genetic alterations in the cyp51A gene encoding lanosterol 14α-demethylase, the target enzyme for azole drugs (15). Early analyses of azole-resistant A. fumigatus isolates indicated that the majority of these organisms contained alterations in the cyp51A coding sequence and often in the transcriptional control region (16). However, other experiments determined that changes in ABC transporter gene expression could be linked to increased azole resistance (17, 18, 19). More recent surveys of azole-resistant clinical isolates found that a large fraction of these organisms contained no detectable change at their cyp51A locus (1, 20). Importantly, overexpression of a gene encoding a Pdr5 homologue was found to be required for azole resistance in a strain with a normal cyp51A gene (21). Together, these findings support the view that, as in other fungal pathogens, transcriptional upregulation of ABC transporter gene expression is an important contributor to this clinically key phenotype.

We set out to systematically explore the contributions of various ABC transporters to drug resistance in A. fumigatus. A challenge to this study is the presence of a generally large number of related ABC transporters (50 total) and a specifically increased number of ABCG-type proteins (15 ABCG transporters) (22). To focus our analysis, we restricted the initial experiments to the two proteins sharing the greatest sequence similarity to ScPdr5 and the single protein in A. fumigatus with highest sequence similarity to ScYor1.

MATERIALS AND METHODS

A. fumigatus strains, growth conditions, and transformation.

Three strains were used in this study: the A. fumigatus Af293 strain, for which the entire genomic sequence is available (23); the akuBΔ (A1151) strain (24); and the akuAΔ (AfS35(strain (25)). These strains were typically grown at 37°C in rich medium (YG; 0.5% yeast extract, 2% glucose) or in minimal medium (MM; 1% glucose, nitrate salts, trace elements, 2% agar [pH 6.5]; trace elements, vitamins, and nitrate salts are as described in the appendix of reference 26. For solid medium, 1.5% agar was added. Uracil auxotrophs were supplemented with 1.2 g/liter each of uracil and uridine. The Af293 strain was transformed using Agrobacterium tumefaciens-mediated transformation (27). The A. fumigatus strains lacking either of the Ku70/80 subunits were transformed by generating protoplasts as described previously (28). For regeneration of protoplasts upon transformation, 182 g/liter of sorbitol was added, along with 200 mg/liter of Hygromycin Gold (Invivogen) to select for transformants. The strains used in this study are listed in Table 1.

Table 1.

A. fumigatus strains used in this study

Strain	Parent	Description	Source or reference
Af293		Wild type	S. Klutts
A1151	CEA17	akuBΔ::pyrG	G. Goldman
AfS35	D141	akuAΔ::loxP	FGSC
SPF13	Af293	abcAΔ::hph	This study
SPF17	Af293	yorAΔ::hph	This study
SPF20	Af293	abcBΔ::hph	This study
SPF21	A1151	abcAΔ::hph	This study
SPF27	A1151	abcBΔ::hph	This study
SPF59	AfS35	abcAΔ::hph	This study
SPF62	AfS35	abcBΔ::hph	This study
SPF56	AfS35	hph::hspA-abcA	This study
SPF44	AfS35	pyrG::hph-abcA-luc::pyrG	30
SPF46	AfS35	pyrG::hph-abcB-luc::pyrG	30
SPF48	AfS35	pyrG::hph-gpdA2-luc::pyrG	30
SPF24	A1151	abcA-GFP::pyrG	This study
SPF79	AfS35	abcB-GFP::hph	This study

For every A. fumigatus disruption mutant, multiple independent isolates (typically 3) were generated, with the exception of the abcBΔ disruption in the Af293 background (only 1 was obtained). We also generated three different versions of the AfS35 strain containing the hspA-abcA gene fusion. This was critical to ensure that the behavior of a given genetic background was consistent and not representative of a rare isolate. In each case, our multiple isolates produced indistinguishable phenotypes, and we report the behavior of a representative clone here.

Plasmids.

DNA manipulations were done using standard procedures (29) or according to the manufacturer's instructions. A list of plasmids used in this study is provided in Table 2. The abcA, abcB, and yorA knockout constructs were made taking advantage of recombination cloning in S. cerevisiae in pSP19 (30), amenable to propagation in S. cerevisiae and A. fumigatus transformation mediated by Agrobacterium tumefaciens. The disruption constructs were generated in pSP19, as described below. Sequences of all primers are available on request. Each cassette had the following components: the loxP-hph-loxP selection cassette, flanked by a 1-kb region immediately upstream and downstream of the coding sequence of the gene to be deleted. The primers used for amplifying DNA upstream and downstream to the targeted gene were designed in pairs such that one had a 40-bp overlapping sequence with the selection cassette at one end and a 40-bp overlap sequence with the termini of pSP19, gapped by XbaI and XhoI digestion, at the other end. These primer pairs were designed to also each contain 20 nucleotides to permit amplification of the desired segment of the A. fumigatus genome. The loxP-hph-loxP selection cassette was released from the plasmid pSKB57 (provided by Stacey Klutts, University of Iowa) by SpeI/EcoRV cleavage and gel purified. The yorA, abcA, and abcB deletion plasmids were named pSP36, pSP44, and pSP45, respectively. These plasmids were transformed in A. tumefaciens EHA10 for subsequent transformation into A. fumigatus Af293 strain. The abcA and abcB deletion alleles in the A1151 and AfS35 strains were generated as follows. Plasmid pSP44 was digested with SnaBI/KpnI to release the abcAΔ::hph disruption cassette, while pSP45 was digested with AleI/KpnI to release the abcBΔ::hph disruption cassette from the vector backbone, which was subsequently transformed into protoplasts. We did not generate gene reconstitution strains but instead constructed multiple independent disruption strains to ensure that phenotypic changes were due to loss of a gene of interest rather than a secondary change occurring in a rare isolate.

Table 2.

Plasmids used in this study

Plasmid	Description	Source or reference
pHL83	pCR2.1 TOPO GA5-GFP::loxP-pyrGAfu-loxP	FGSC
SKB57	pCR2.1 TOPO loxP-hph-loxP	S. Klutts
pSP19	pDHT ScCEN6 ScARSH4 ScURA3	30
pSP36	pSP19 yorAΔ::loxP-hph-loxP	This study
pSP44	pSP19 abcAΔ::loxP-hph-loxP	This study
pSP45	pSP19 abcBΔ::loxP-hph-loxP	This study
pSP46	pRS316 loxP-hph-loxP::hspA-abcB	This study
pSP47	pRS316 loxP-hph-loxP::hspB-abcB	This study
pSP38	pRS316 abcA-GFP::loxP-hph-loxP	This study
pSP58	pRS316 abcB-GFP::loxP-hph-loxP	This study
pSP40	pET42a GST-abcA	This study
pSP41	pET42a GST-abcB	This study

To put abcA and abcB under the control of strong hspA promoter, the loxP-hph-loxP selection cassette was placed between a 1.5-kb region immediately upstream of the target gene and 1-kb hspA promoter (30), followed by a 1.5-kb region corresponding to the 5′ end of the target gene and inserted into pRS316 (31) by recombinational cloning in S. cerevisiae. While the loxP-hph-loxP selection cassette was generated as described above, the flanking regions for homologous integration into the A. fumigatus genome and the hspA promoter were generated by PCR amplification using primers that had a 40-bp overlap sequence with pRS316 linearized by XbaI and the hph-loxP selection cassette. The plasmids carrying the hspA-abcA and hspA-abcB promoter fusions were named pSP46 and pSP47, respectively. The pSP46 and pSP47 inserts were released from their respective vector backbones by NotI and SpeI digestion for protoplast transformation.

The abcA gene was tagged using a green fluorescent protein (GFP)-loxP-pyrG-loxP fragment generated from pHL83 (Fungal Genetic Stock Center) digested with XbaI and SpeI. The above-mentioned DNA was flanked with 1 kb corresponding to the 3′ end of abcA and 1.5 kb of DNA immediately downstream of the abcA gene. These two segments of DNA were generated by primer pairs that had a 40-bp overlap with the XbaI-linearized pRS316 and 20 nucleotides of the abcA gene or 40 nucleotides of the GFP-loxP-pyrG-loxP fragment and 20 nucleotides of the abcA gene. These were cloned by homologous recombination in S. cerevisiae to generate pSP38. Digestion of pSP38 by NheI and MluI allowed the release of the abcA-GFP tagging construct for protoplast transformation into the pyrG⁻ A1151 strain with selection for uracil prototrophy. The abcB gene was tagged with GFP using the loxP-hph-loxP selection cassette, flanked by GFP and a 1.5-kb region corresponding to DNA immediately downstream of the abcB gene. The GFP gene was PCR amplified from pHL83 and was preceded by 1 kb of PCR-amplified DNA corresponding to the 3′ end of the abcB gene in the tagging construct. All the PCR products had a 40-bp overlap with the adjoining DNA fragments to generate the abcB-GFP-loxP-hph-loxP-abcB tagging construct in XbaI-linearized pRS316 by homologous recombination in S. cerevisiae as described above for abcA. The resulting plasmid was named pSP58. The tagging construct was released from this plasmid by SpeI and XhoI digestion for protoplast transformation into the AfS35 strain with selection for hygromycin resistance.

Generation of polyclonal antibodies against AbcA and AbcB.

Fragments of 360 and 330 bp (corresponding to the N-terminal, nontransmembrane protein regions spanning 120 and 110 amino acids [aa]) from AbcA and AbcB, respectively, were PCR amplified and cloned in frame as BamHI/SalI fragments downstream of the glutathione S-transferase (GST) tag in pET42a (EMD Millipore, Inc.) to form pSP40 and pSP41 and transformed into the bacterial strain Escherichia coli BL21(DE3). Two liters of transformed bacteria was grown to log phase and induced with IPTG (isopropyl-β-d-thiogalactopyranoside) for 90 min. Cell lysates were prepared using a French press. Protein purification was accomplished using glutathione Sepharose 4B resin (GE Healthcare, Inc.) as described by the manufacturer. Protein fractions were analyzed by staining them with Coomassie blue and by Western blotting using GST-specific antibodies. The purified proteins (GST-AbcA and GST-AbcB) were then lyophilized and sent to Cocalico Biologicals, Inc., PA, for injection into rabbits to generate polyclonal antibodies against AbcA and AbcB. Antiserum generated from these rabbits were received and tested for immunoreactivity against appropriate GST fusion proteins. The antiserum was then purified using an AminoLink plus coupling resin (ThermoScientific, Inc.) according to manufacturer instructions, and the affinity-purified antiserum was then used to detect AbcA and AbcB proteins from A. fumigatus cell lysates.

Antifungal susceptibility testing.

MICs of the azoles itraconazole and voriconazole against Aspergillus were determined using a broth microdilution (BMD) method. Itraconazole (Janssen) and voriconazole (Pfizer) were obtained as reagent-grade powders from their respective manufacturers. The BMD method was performed according to the CLSI M38-A2 standard (32). Trays containing a 0.1-ml aliquot of the appropriate drug solution (2× the final drug concentration) in each well were sealed and stored at −70°C until used in the study. A stock conidial suspension (10⁶ spores/ml) was diluted to a final inoculum concentration of 0.4 × 10⁴ to 5 × 10⁴ CFU/ml and dispensed into the microdilution wells. The final concentrations of drugs in the wells ranged from 0.007 to 8 μg/ml. The inoculated microdilution trays were incubated at 35°C and read at 24 and 48 h. The MIC endpoint was defined as the lowest concentration that produced complete inhibition of growth. Quality control was ensured by testing the following strains recommended in CLSI standard M38-A2: Candida parapsilosis ATCC 22019, Candida krusei ATCC 6258, and Aspergillus flavus ATCC 204304.

Spot assay.

Fresh spores of A. fumigatus were suspended in 1× phosphate-buffered saline (PBS) supplemented with 0.01% Tween 20 (1× PBST). The spore suspension was counted using a hemocytometer to determine the spore concentration. The spores were then appropriately diluted in 1× PBST so that ∼20 spores (in 5 μl) were spotted on YG medium with or without the drug. The plates were incubated at 37°C and inspected for growth every 12 h.

GFP localization.

A. fumigatus spores were grown overnight at 37°C in rich medium. The mycelia were harvested by filtration, washed in deionized water, and visualized for GFP fluorescence and Nomarski optics using an Olympus (Tokyo, Japan) BX-60 microscope with a 100× oil immersion objective. Images were captured using a Hamamatsu (Shizuoka, Japan) ORCA charge-coupled-device camera.

Luciferase assay.

Firefly luciferase assays were done using a firefly luciferase assay kit (obtained from Biotium, Inc.) as described in reference 30. Luminescence was measured using an IVIS 100 imaging system (Caliper Life Sciences). The number of photons emitted per second (luminescence) was then normalized to the amount of total protein present in the cell lysate per ml using a Bradford assay. To estimate fold induction of promoters in the presence of voriconazole, one relative luminescent unit was calculated as luminescence emitted by the promoter-reporter gene immediately (0 h) after treatment with drug. A sublethal dose of voriconazole was used to minimize any effects of cell death on the observed luciferase activity measured.

Western blotting.

A. fumigatus spores were grown overnight at 37°C in rich medium. The mycelia were harvested by filtration, washed in deionized water, and used to prepare crude cell lysates. The mycelia (5 mg [wet weight]) were ground in liquid nitrogen using a mortar and pestle and then resuspended in 1 ml of A. fumigatus extraction buffer (50 mM HEPES [pH 7.7], 1 M sucrose, 50 mM NaCl, 1 mM dithiothreitol [DTT], 1 mM EDTA). The cell suspension was vortexed in the presence of 0.5-mm glass beads for 5 min and subjected to low-speed centrifugation (850 × g for 5 min) at 4°C. The supernatant was taken and subjected to high-speed centrifugation (16,000 × g for 15 min) at 4°C. The pellet was taken and resuspended in 250 μl of 1× Laemmli buffer and incubated at 37°C for 15 min. Twenty-five microliters of this suspension was electrophoresed by 7.5% SDS-PAGE, transferred to a nitrocellulose membrane, blocked with 5% nonfat dry milk in phosphate-buffered saline, and then probed with appropriate polyclonal rabbit antiserum (1:500). Horseradish peroxidase-conjugated secondary antibody and an ECL kit (Pierce) were used to visualize immunoreactive protein.

Galleria mellonella killing assay.

Fresh spores of A. fumigatus were suspended in 1× PBST. The spore suspension was diluted and counted under a hemocytometer to determine the spore concentration. The spore suspension was then diluted in 1× PBST to reach a concentration of 5 × 10⁷ spores/ml and used to infect G. mellonella caterpillars (Wax Worm Store) in the final instar larval stage. These worms were stored in the dark and used within 7 days from the day of shipment. Ten to 12 randomly chosen caterpillars (light colored without gray marking on the cuticle) were used per group. A 10-μl Hamilton syringe was used to inject 5-μl aliquots of the spore suspension into the hemocoel of each caterpillar via the last left proleg. When voriconazole (5 μg) was used, the antifungal drug was injected in a 5-μl volume into the last right proleg, 1 h after infection with A. fumigatus spores. After injection, caterpillars were incubated at 37°C in petri plates with wood chips and the number of dead caterpillars was scored daily. Caterpillars were considered dead when they displayed no movement in response to touch. A mock injection with PBS was performed in each experiment to monitor killing due to physical injury or infection by pathogenic contaminants. Killing curves were plotted and estimation of differences in survival was analyzed by the Kaplan-Meier method using SigmaPlot software. A P value of 0.05 was considered significant. Each experiment was repeated at least three times. The survival assay represented here was plotted by incorporating data points from 3 independent experiments. We also tested two independent isolates of each mutant strain reported here and found similar results. Results from one isolate are presented here.

RESULTS

A. fumigatus contains a large family of ABCG and ABCC class ABC transporters.

While circumstantial evidence exists that ABC transporter proteins are involved in azole resistance in A. fumigatus, no systematic investigation of this question has previously been undertaken. Given that the Cdr1 ABC transporter plays a major role in azole susceptibility in Candida species, we set out to analyze the ABCG class transporters in A. fumigatus that shared the most sequence similarity with this protein. We used the sequence of S. cerevisiae Pdr5 as the prototype fungal ABCG transporter since this intensively studied protein represents the best-understood and founding member of these membrane proteins. The characteristic structure of ABCG transporters resembling Pdr5 is shown in Fig. 1A.

Fig 1.

ABCG transporter family in A. fumigatus. (A) The typical topology of a full-length ABCG transporter like Pdr5 is diagrammed here. The location of each set of 6 transmembrane domains is labeled “MSD” (for membrane spanning domains), while “NBD” is used to indicate the signature ATP-binding cassette motif called the nucleotide-binding domain. The region of the proteins used to raise antibodies is designated “Ab.” (B) Phylogenetic tree diagram of ABCG family in A. fumigatus. This alignment was generated using BLAST Tree View at the NCBI database (maximum sequence difference = 0.9).

A BLAST search comparing the ScPdr5 sequence against the A. fumigatus genome was performed (Fig. 1B). This comparison identified 15 proteins showing significant sequence conservation to Pdr5. The two proteins with the highest degree of sequence similarity were those with GenBank accession numbers XP_755847 and XP_752803. We selected these for further analyses and designated them AbcA and AbcB, respectively. The abcB gene has also been designated cdr1B by Fraczek and colleagues (21).

Along with these ABCG-type transporters, we also characterized the A. fumigatus homologue of S. cerevisiae Yor1. ScYor1 is the only known ABCC class transporter in this yeast that is localized to the plasma membrane (33). The other ABCC transporters are found in the vacuolar membrane (34, 35). A BLAST search using the ScYor1 protein sequence identified 10 ABCC class transporters in A. fumigatus but, interestingly, only a single protein that shared the unique structure of ScYor1. Typically, ABCC transporters contain two copies of a monomeric structure consisting of 6 transmembrane domains followed by the signature ABC nucleotide-binding domain. ABCC proteins precede this duplicated structure with a set of 5 transmembrane domains, first described for the mammalian Mrp1 protein (36). ScYor1 and the A. fumigatus protein with GenBank accession number XP_751910 both possess the duplicated ABCC transmembrane domains and ABC domains but lack this amino-terminal set of transmembrane segments. Since the A. fumigatus protein mirrored this unusual structure, this was taken as evidence that it represents an authentic homologue of ScYor1, and we designated the gene yorA.

abcB and abcA knockout strains are sensitive to azole drugs.

In order to assess the physiological roles of the three selected ABC transporter proteins, our first goal was to construct a series of isogenic mutant strains lacking the corresponding coding sequences in the genome. For each gene, a disruption allele was prepared in which 1 kb of 5′ and 3′ noncoding DNA was present flanking a hygromycin resistance determinant. We first constructed null alleles of abcA, abcB, and yorA in the A. fumigatus strain Af293. These null alleles were introduced into Af293 using Agrobacterium tumefaciens-mediated transformation, and hygromycin-resistant isolates were recovered. We confirmed the correct integration of these disruption cassettes by PCR. Representative isolates were grown and tested for the ability to tolerate a series of drug challenges by placing spores on a plate containing a gradient of drug as described previously (37, 38). These plates were incubated at 37°C and then photographed (Fig. 2).

Fig 2.

Azole resistance phenotypes of mutants in ABC transporter-encoding genes. (A) The mutations shown here were produced in the wild-type strain Af293 (WT) using A. tumefaciens-mediated transformation. Each null allele lacked all coding sequences for the indicated ABC transporter and was marked with a hygromycin resistance cassette. Strains were grown to make fresh spore suspensions. These suspensions were diluted in PBST buffer, and ∼20 spores were placed on each spot. Plates were incubated at 37°C and then photographed. Increasing drug concentrations are indicated by increased bar width. (B) Same as panel A, but these strains are derivatives of A1151 (akuB^KU80Δ). (C) The strain used to generate these mutants is designated AfS35 (akuA^KU70Δ). Drug phenotypes were determined as for the other panels, but the final row of spots corresponds to a strain containing the hspA-abcA fusion gene that overproduces abcA. Panels A and C were photographed after 2 days, while panel B was photographed after 3 days.

Loss of either abcA or abcB led to an increase in azole susceptibility in Af293 cells. A null allele of yorA had no significant effect on azole tolerance. However, the data for abcB were weakened by the difficulty in obtaining disruption mutants of this gene in the Af293 background. Since this raised the concern that the behavior we observed was not representative of a typical abcB null allele, we constructed multiple abcBΔ::HYG alleles in two other strain backgrounds to address this issue.

Loss of the Ku70/80 heterodimeric protein has been found to strongly stimulate homologous recombination in several organisms, including A. fumigatus (24, 39, 40). We employed two different strains with different defects in the Ku80 complex to increase the frequency of obtaining homologously targeted integration events. These strains were A1151 (akuBΔ) and AfS35 (akuAΔ). Homologous integration was found to be high enough in these genetic backgrounds that the use of the laborious and time-intensive A. tumefaciens-mediated transformation was replaced with a more facile protoplast transformation technique. We were able to obtain at least two isolates of all disruption alleles (including abcBΔ::HYG) using these backgrounds. We compared the abilities of these disruption mutants to grow in the presence of azole drugs using the gradient plate assay described above (Fig. 2B and C).

Loss of abcB caused a large decrease in azole sensitivity in all backgrounds that were tested, consistent with this gene being an important participant in tolerance to these drugs. The absence of abcA had more variable effects on voriconazole resistance, as its removal from A1151 caused a negligible increase in susceptibility. However, an abcAΔ::HYG allele in AfS35 did elicit a modest increase in voriconazole sensitivity.

We also assayed the relative azole tolerance of these strains using a liquid growth assay. Spores of the relevant strains prepared in the Af293 genetic background were grown in liquid media containing 2-fold serial dilutions of either voriconazole or itraconazole. Plates were observed every 24 h, and the lowest drug concentration at which no growth could be seen was noted (Fig. 3).

Fig 3.

MIC measurement for A. fumigatus strains lacking ABC transporter genes. Strains with the indicated genotypes were inoculated into 96-well microtiter plates containing serial 2-fold dilutions of either voriconazole (top panel) or itraconazole (bottom panel). Each plate was incubated at 37°C and evaluated after 24 or 48 h. The MIC endpoint was defined as the lowest concentration that produced complete inhibition of growth.

As in the radial growth assay on solid media, strains lacking abcB were the most sensitive to either azole challenge. The presence of the abcAΔ::HYG allele produced a more subtle increase in azole susceptibility, while the effect of the yorAΔ::HYG mutation could be detected only after extended incubations. These data confirm that the drug phenotypes indicated by our radial growth experiments are accurately assigned.

While these manipulations tested the effect of the loss of these transporters, we also wanted to explore the consequence of elevated expression of abcA and abcB. To accomplish this, we prepared constructs in which the strong hspA^SSA1 promoter was placed upstream of the ATG for either abcA or abcB. 5′ to this promoter, we placed the hygromycin resistance gene. This cassette was flanked by 1 kb of DNA from the amino-terminal coding sequence and the promoters of abcA and abcB. Both of these constructs were introduced into AfS35 cells and hygromycin-resistant transformants selected. We recovered only hspA^SSA1-driven abcA clones and tested these for their resistance to voriconazole (Fig. 2C).

The presence of the hspA^SSA1-abcA fusion gene consistently increased voriconazole resistance. The inability to recover similar constructs for abcB is not understood presently.

Analysis of abcA and abcB expression.

The results of the genetic analyses indicated that AbcB was an important contributor to azole tolerance, while AbcA played a minor role that could be uncovered when a strong promoter drove its expression. To correlate the level of expression of these two different proteins with their azole resistance phenotypes, we generated rabbit polyclonal antibodies against each ABC transporter protein. Since these proteins contain multiple transmembrane domains, we produced amino-terminal fragments of each transporter in bacteria. The amino terminus of each protein is predicted to be soluble and is associated with the first ABC nucleotide-binding domain. These antigens were successful in raising monospecific antibodies against each ABC transporter.

To validate the specificity of the antibodies and examine levels of AbcA and AbcB protein expression, we carried out Western blotting experiments. Whole-cell protein extracts were prepared from wild-type AfS35 cells or isogenic variants lacking AbcA or AbcB or containing the hspA-abcA construct. Equal amounts of protein were resolved by SDS-PAGE and then subjected to Western analysis using the affinity-purified antibody directed against AbcA (Fig. 4A) or AbcB (Fig. 4B).

Fig 4.

Expression of AbcA and AbcB. (A) Whole-cell protein extracts were prepared from A. fumigatus strains of the indicated genotypes. The hspA-abcA-containing fusion strain was either left unchallenged (−) or induced with 5% ethanol addition (+). Equal amounts of protein were resolved by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were stained with Ponceau S dye to visualize total protein (top) or with a polyclonal antiserum directed against AbcA (bottom). Molecular mass markers, in kilodaltons, are indicated on the right side. (B) Whole-cell protein extracts prepared from the A. fumigatus strains of the indicated genotypes were processed as for panel A, except that the Western analyses were carried out using an antibody directed against recombinant AbcB.

Detection of AbcA protein produced from the endogenous gene was not possible using our anti-AbcA antiserum. However, in strains containing the hspA-abcA fusion gene, the correctly sized protein was readily apparent. This finding is consistent with the ability of this fusion gene to increase voriconazole resistance shown in Fig. 2C. Treatment of these cells with ethanol, a stress known to increase hspA expression, led to a modest further elevation of AbcA expression.

Western blot comparison of extracts from wild-type or abcAΔ cells using the anti-AbcB antiserum detected a polypeptide species of 170 kDa. A similar experiment using proteins derived from abcBΔ cells failed to detect this protein. This antiserum was capable of recognizing AbcB protein produced from the wild-type gene.

Having established that full-length AbcA and AbcB can be detected in cells, we wanted to determine where these membrane transporter proteins might be localized. To begin to examine this issue, we prepared fusions of each ABC transporter to the green fluorescent protein (GFP). In both cases, GFP was placed at the C terminus of the ABC transporter with expression controlled by the wild-type promoter. Transformants were obtained and mycelia visualized using Nomarski optics and fluorescence microscopy (Fig. 5).

Fig 5.

Subcellular localization of AbcA and AbcB. (A) Isogenic wild-type (WT; A1151) or integrated abcA-GFP fusion strains were grown overnight in rich medium at 37°C. Mycelia were harvested by filtration, washed with water, and then visualized using Nomarski optics (DIC) or fluorescent light (GFP). (B) Same as panel A, but wild-type and abcB-GFP-containing strains were AfS35 cells.

Both GFP fusion proteins were found to be associated with the cell surface. Notably, we found no detectable intracellular fluorescence, which contrasted with our previous studies of heterologously expressed AbcA-GFP in S. cerevisiae (38). In this foreign environment, we could readily detect perinuclear AbcA-GFP. In the native A. fumigatus setting, AbcA-GFP (and AbcB-GFP) seemed to efficiently reach the plasma membrane.

There were differences observed in the fluorescence from these two membrane proteins. The AbcB-GFP fusion generated a more punctate appearance, while the AbcA-GFP fusion was more evenly distributed. Both of these patterns were clearly above the background seen in cells lacking a GFP fusion and consistent with both ABC transporters exhibiting steady-state localization to the plasma membrane. Further experiments are required to confirm this suggestion.

Transcriptional induction during voriconazole stress.

We have previously described development of a firefly luciferase reporter gene system that works in A. fumigatus (30) and use of this system to produce fusion genes to several fungal promoters. Our finding of the roles of AbcA and AbcB in voriconazole resistance prompted us to test the ability of this azole drug to influence expression levels of these transporters using the luciferase fusion genes already generated. The promoters for abcA, abcB, and the glycolytic gene gpdA2 were fused to firefly luciferase to make transcriptional reporters for each gene. These were integrated into the A. fumigatus genome at the pyrG locus. Appropriate integrants were grown to log phase and then treated with voriconazole for 45 or 90 min. Whole-cell protein extracts were prepared and assayed for their luciferase enzyme activities (Fig. 6).

Fig 6.

Voriconazole induction of abcA and abcB expression. A. fumigatus transformants containing fusions between the indicated promoters and the firefly luciferase gene were grown to early log phase and then exposed to 20 ng/ml of voriconazole for 0, 45, or 90 min. Luciferase activity was assayed from representative aliquots and normalized to the untreated control. The error bars represent standard deviations in fold induction of luciferase activity upon drug treatment based on 3 independent experiments done on a representative strain.

Both the abcA and abcB promoters were found to be inducible upon voriconazole treatment. Importantly, the gpdA2 promoter was unresponsive to voriconazole, consistent with azole drug induction being restricted to genes involved in its detoxification. These data suggest that control of abcA and abcB expression by voriconazole occurs via induction of elements within the promoters of these genes.

Role of AbcA and AbcB in virulence during Galleria infection.

To analyze the role of AbcA and AbcB during pathogenesis, we utilized the Galleria mellonella (greater wax moth) infection model. The G. mellonella caterpillar has been established to be a successful model in identifying genes crucial for virulence in many pathogenic fungi, such as the yeasts C. albicans and Cryptococcus neoformans as well as filamentous fungi such as different species of Aspergillus, including A. fumigatus. Briefly, equal numbers of spores from various A. fumigatus strains were injected into G. mellonella larvae along with a buffer control. The outcome of this infection was determined by monitoring killing of the larvae, in both the presence and absence of drug (voriconazole) administration. The percentages of larvae that survived infection were plotted with time (Fig. 7).

Fig 7.

Virulence effects of AbcA and AbcB. G. mellonella larvae were injected with the indicated A. fumigatus strains (2.5 × 10⁵ spores) or a buffer control (PBST). All these experiments employed the AfS35 background. These injections were done either without (A) or with (B) an initial injection with voriconazole to assess the ability of this azole drug to protect the larvae from infection. Survival of injected larvae was assessed every 24 h after injection.

In the absence of drug, spores lacking the abcB gene were found to be hypovirulent (P < 0.001). No significant differences were seen between the other strains, including strains lacking or overproducing AbcA. This finding suggests the possibility that AbcB is required for normal pathogenesis of A. fumigatus, irrespective of its role in azole resistance.

To assess the modification of the course of infection by azole drugs, voriconazole was injected 1 h after the initial exposure to A. fumigatus spores. Treatment with voriconazole strongly improved larval survival. Interestingly, the presence of the hspA-abcA overexpression cassette strain showed a trend toward enhanced virulence of cells in this assay. The significance of this effect was modest (P values of 0.052 for hspA-abcA versus abcBΔ and 0.1 for hspA-abcA versus wild type). Taken together, these data support the view that AbcB is required for normal virulence and that overexpression of AbcA may enhance azole resistance during infection.

DISCUSSION

Azole resistance in A. fumigatus appears to be entering a dynamic phase in its development. Most azole-resistant mutants previously described appeared to be associated with alterations in the cyp51A locus, including changes in the amino acid sequence of the enzyme and duplications of promoter DNA segments (41). More recently, this picture has become more complex as continued characterization of azole-tolerant isolates found that these strains frequently possessed wild-type versions of cyp51A (1, 20). Molecular characterization of some of these strains provided evidence that expression of an ABC transporter-encoding gene called cdr1B (referred to in this work as abcB) was elevated and likely the cause of increased azole resistance (21). The earlier work described in reference 21 provided the first evidence for the role of cdr1B (abcB) in azole resistance in wild-type A. fumigatus. In this study, we have extended our understanding of this ABC transporter protein with the finding that Cdr1B/AbcB is critical for azole resistance in several common laboratory strains, detecting the polypeptide chain for the first time, localizing the protein to the plasma membrane, and providing evidence that this protein may be required in pathogenesis. Combined with the previous study (21), these data implicate Cdr1B/AbcB as an important participant in the biology of drug resistance in A. fumigatus.

Other fungal pathogens have clearly established roles for ABC transporters in acquisition of drug resistance in patients. Azole-resistant isolates of C. albicans and C. glabrata are readily obtained that can be shown to exhibit elevated expression of ABC transporter-encoding genes. Extensive analyses of these azole-tolerant organisms have implicated alterations in the transcriptional control of the relevant ABC transporter genes leading to large elevations in transcript and protein level (13, 14). In both Candida species, single amino acid substitutions in a zinc cluster-containing transcription factor leads to this constitutive induction of ABC transporter expression (42, 43). The increased azole tolerance in these strains is genetically quite stable. The more recent identification (21) of strains overproducing cdr1B (also called abcB or abcC) in a wild-type cyp51A background will be interesting to examine for the nature of the transcriptional activation of this key azole resistance-conferring ABC transporter-encoding gene.

We also provide evidence that the promoters of both abcA and abcB can drive azole-inducible gene expression. Previous microarray experiments indicated that a number of different azole-inducible genes are present in A. fumigatus (19). Several ABC transporter-encoding genes were profiled in this work, and abcB (also called abcC or cdr1B) was found to be highly inducible by voriconazole challenge. The voriconazole induction seen in the microarray experiments was much greater than we report here, but this could be due to differences in strain backgrounds or experimental manipulations. Irrespective of the magnitude of induction, some fraction of this is supported by the promoter regions of both these ABC transporter-encoding loci. Additionally, this inducibility may reduce the generation of transcription factor mutations, such as those seen in Candida species and S. cerevisiae, and instead allow an acute and reversible response to azole challenge. Once the drug concentration has dropped sufficiently, the inducing stimulus may be removed and expression of drug resistance genes returned to baseline.

The data reported here also provide the first insight into the subcellular localization of both AbcA and AbcB. Both of these proteins were found on the A. fumigatus plasma membrane but did not exhibit a uniform distribution. The punctate fluorescence pattern of each transporter suggests that these may be enriched in discrete subdomains of the A. fumigatus plasma membrane. Since very little is known of the cell biology of membrane trafficking in this organism, an important subsequent goal will be to further analyze localization of AbcA and AbcB to the plasma membrane in comparison with other protein to establish the specificity of this observed localization. It was surprising to see no evidence for intracellular fluorescence of either ABC transporter protein. With the yeast Pdr5 homologues, it is typical to see the protein localized to the plasma membrane but also to the intracellular vacuole, the site of degradation of this and other proteins (44). Using our antibodies and GFP fusion proteins, we will address the itinerary of the A. fumigatus ABC transporters.

While it is common for ABC transporters to play significant roles in drug resistance, we also found that loss of AbcB led to a decrease in virulence in the G. mellonella infection model. This defect was seen irrespective of the presence of azole drug and suggests that AbcB is required for normal pathogenesis of the fungus. This effect is not seen with AbcA; instead, AbcA appears to work in a more typical fashion by increasing tolerance to azole drugs (certainly when overproduced). Another striking difference between abcA and abcB was the relative ease of genetically modifying either locus. We could easily obtain strains overproducing or lacking AbcA but had to go to some length to recover adequate numbers of isolates of abcB null alleles and were not able to recover a strain that overproduced AbcB from the hspA promoter. These findings are consistent with AbcB playing a fundamentally important role in A. fumigatus physiology behind drug resistance that is dosage sensitive. Our future experiments will be aimed at testing this idea for the unique role of AbcB in this important filamentous fungal pathogen.