Azole Resistance of Aspergillus fumigatus Biofilms Is Partly Associated with Efflux Pump Activity

Abstract

This study investigated the phase-dependent expression and activity of efflux pumps in Aspergillus fumigatus treated with voriconazole. Fourteen strains were shown to become increasingly resistant in the 12-h (16- to 128-fold) and 24-h (>512-fold) phases compared to 8-h germlings. An Ala-Nap uptake assay demonstrated a significant increase in efflux pump activity in the 12-h and 24-h phases (P < 0.0001). The efflux pump activity of the 8-h germling cells was also significantly induced by voriconazole (P < 0.001) after 24 h of treatment. Inhibition of efflux pump activity with the competitive substrate MC-207,110 reduced the voriconazole MIC values for the A. fumigatus germling cells by 2- to 8-fold. Quantitative expression analysis of AfuMDR4 mRNA transcripts showed a phase-dependent increase as the mycelial complexity increased, which was coincidental with a strain-dependent increase in azole resistance. Voriconazole also significantly induced this in a time-dependent manner (P < 0.001). Finally, an in vivo mouse biofilm model was used to evaluate efflux pump expression, and it was shown that AfuMDR4 was constitutively expressed and significantly induced by treatment with voriconazole after 24 h (P < 0.01). Our results demonstrate that efflux pumps are expressed in complex A. fumigatus biofilm populations and that this contributes to azole resistance. Moreover, voriconazole treatment induces efflux pump expression. Collectively, these data may provide evidence for azole treatment failures in clinical cases of aspergillosis.

INTRODUCTION

The invasive filamentous mold Aspergillus fumigatus is associated with life-threatening pulmonary infections in individuals in an immunocompromised status that can result in unacceptable rates of mortality. Members of the azole class, which includes fluconazole, itraconazole, voriconazole (VRZ), and posaconazole, have been shown to be effective in the treatment of invasive aspergillosis (IA) and are the mainstay treatment for the disease (6, 28, 30, 32). Although the triazoles have proven efficacy with good safety profiles, they have been shown to be associated with resistance through their continuous use (8, 17, 19). Azoles actively target the 14-α-demethylase enzyme, blocking ergosterol biosynthesis and destabilizing the cell membranes of actively growing cells. Mutations within the ergosterol biosynthesis pathway have been reported to cause azole cross-resistance through mutations within the cyp51A gene (8, 16, 33, 34). However, a recent study reported that 43% of azole-resistant isolates did not carry the cyp51A mutation, indicating that other mechanisms of resistance were responsible (2).

Another commonly described azole resistance mechanism is mediated by multidrug resistance (MDR) pumps, which are involved in the active extrusion of antimicrobial molecules, including azole. MDR efflux transporter genes of the ATP-binding cassette (ABC) and the major facilitator superfamily (MFS) classes have been shown to be clinically important in different pathogenic fungi (3, 18). Among members of the ABC transporters, the MDR genes have been widely implicated in Candida albicans azole resistance (36, 38). They have also been shown to be important in C. albicans biofilms, where they have been reported to be transiently expressed and are thought to have a role in detoxification (1, 23, 31).

It has been shown that A. fumigatus also has the capacity to form biofilms encased in polymeric matrix, which is the most likely growth modality with a fungus ball (14, 22). As these biofilms grow, they are differentially sensitive to antifungal drugs (21). Sequence analysis suggests that A. fumigatus has 278 different MFS and 49 ABC transporters (27). A. fumigatus MDR (AfuMDR) pumps have been described in several studies and have been shown to be associated with increased resistance to itraconazole (4, 24). Currently, however, there is little evidence to suggest that they play an active role in clinical resistance (3). Nevertheless, it can be hypothesized that these pumps are activated as the biofilm develops to regulate homeostasis and by exposure to antifungal compounds. This activity may explain previously reported phase-related resistance profiles (21). The aim of our current study was to investigate the biological activity of efflux pumps in A. fumigatus and their differential expression with respect to morphological development and azole treatment.

MATERIALS AND METHODS

Strains and growth conditions.

A. fumigatus Af293 and 13 clinical strains (YHCF1 to YHCF13) obtained from the Royal Hospital for Sick Children (Yorkhill Division), Glasgow, United Kingdom, were used throughout this study. The isolates were stored on Sabouraud dextrose slopes (Oxoid, Basingstoke, United Kingdom) at 4°C. Antifungal susceptibility testing with and without efflux pump inhibitor (EPI) was undertaken with all isolates. Af293 and YHCF1 to YHCF4 were used for all fluorescence uptake assays and in quantitative PCR (qPCR) expression analyses. ATCC 1163 was used for in vivo expression analyses. All A. fumigatus strains were grown on Sabouraud dextrose agar at 37°C for 72 h. Conidia were then harvested by flooding the surfaces of plates with phosphate-buffered saline (PBS) (Oxoid, Cambridge, United Kingdom) containing 0.025% (vol/vol) Tween 80 while gently rocking them. Conidia were then counted on a Neubauer hemocytometer and adjusted to a standardized suspension of 1 × 10⁵ conidia ml⁻¹, as previously described (20). All procedures were carried out in a laminar flow cabinet (Hera Safe laminate flow cabinet, model K515; Kendro).

A. fumigatus phase-dependent susceptibility testing.

Voriconazole (Vfend; Pfizer) was prepared in stock solutions of dimethyl sulfoxide (DMSO) (Sigma-Aldrich, Gillingham, United Kingdom) and diluted to the required concentrations in MOPS (morpholinepropanesulfonic acid)-buffered RPMI. For antifungal susceptibility testing for sessile cells, conidia were prepared as described above, and standardized conidial suspensions were inoculated into 96-well microtiter plates (Corning, Oneonta, NY) and incubated for 8, 12, and 24 h at 37°C to form biofilms, as described previously (21). The medium was then aspirated, and any nonadherent cells were removed by thorough washing (3 times with PBS). Residual PBS was then removed before the addition of voriconazole (0 to 256 mg/liter), and the plates were incubated for a further 48 h at 37°C. Sessile MICs (SMICs) were determined at 90% reduction in metabolism compared to untreated controls using an XTT [2,3-bis(2-methoxy-4-nitro-5-sulfo-phenyl)-2H-tetrazolium-5-carboxanilide]-based reduction assay, as previously described (20). Testing of these isolates was performed in quadruplicate on two different days.

Assessment of phase-dependent efflux activity.

An optimized alanine β-naphthylamine (Ala-Nap) fluorescence assay was performed on six isolates (Af293 and YHCF1 to YHCF4) to assess whether different phases of A. fumigatus morphological development exhibited efflux pump activity, as previously described (13). MC-005,556 (Ala-Nap) is enzymatically cleaved inside the cells to produce highly fluorescent β-naphthylamine. Higher levels of fluorescence reflect low efflux pump activity, and vice versa. Selected isolates were washed and standardized to 1 × 10⁵ conidia ml⁻¹ in RPMI, and the standardized conidial suspensions were inoculated into black flat-bottom microtiter plates (Costar 3603; Corning, NY) and incubated for 8, 12, and 24 h at 37°C. The dry weight of each biofilm was also quantified for normalization. The medium was then aspirated, and any nonadherent cells were removed by thoroughly washing the cells three times with buffer solution (K₂HPO₄ [50 mM], MgSo₄ [1 mM], and glucose [0.4%]) at pH 7.0. Mature biofilms (24 h) were also disaggregated and washed three times to take account of lack of diffusion. The reaction was initiated by the addition of Ala-Nap (Sigma-Aldrich, Gillingham, United Kingdom) at a final concentration of 128 mg/liter. The effect of pretreatment with 0.0625 mg/liter of voriconazole, which was the sub-MIC level for Af293, was also investigated. Fluorescence was quantified at 30-s intervals at an excitation wavelength of 320 nm and an emission wavelength of 460 nm over 1 h at 37°C using a fluorescent plate reader (Fluostar Optima; BMG Labtech, Aylesbury, Buckinghamshire, United Kingdom). The relative fluorescence units (RFU) were normalized against the dry weight (mg) of each biofilm and presented as RFU/mg of biofilm.

Assessment of azole sensitivity in the presence of an efflux pump inhibitor.

To determine whether azole sensitivity was related to efflux pump activity, an efflux pump inhibitor (l-Phe–l-Arg–β-naphthylamide [MC-207,110]; Sigma-Aldrich, Gillingham, United Kingdom) was used in combination with voriconazole to determine whether antifungal efficacy could be enhanced. For testing azole susceptibility in combination with MC-207,110, Af293 conidia were initially standardized to 1 × 10⁵ conidia ml⁻¹, inoculated into 96-well microtiter plates, and incubated for 8 h at 37°C, as described above. The medium was then aspirated, and any nonadherent cells were removed by thoroughly washing the cells 3 times with PBS. A checkerboard assay was prepared with both voriconazole (0.005 to 4 mg/liter) and MC-207,110 (8 to 512 mg/liter), which were tested within these ranges in combination. An alamar blue colorimetric assay was used to assess viability, according to the manufacturer's instructions (Invitrogen, Paisley, United Kingdom). Following these initial experiments, a defined concentration of MC-207,110 was selected (64 mg/liter) for all subsequent microdilution testing on the entire panel of strains, with voriconazole at 0.005 to 4 mg/liter.

In vitro phase-dependent AfuMDR4 expression analysis.

A. fumigatus Af293 and YHCF1 to YHCF4 biofilms were prepared in triplicate on tissue culture-treated 75-cm² flasks (Nunc, Rochester, NY) for selected time periods (8, 12, and 24 h) on a rocking platform, as described previously for C. albicans (31). In addition, the effect of pretreatment with voriconazole (1 mg/liter) was investigated for Af293, YHCF1, and YHCF2 on each of these populations, and AfuMDR4 expression was quantified at 1, 4, and 24 h. Spent medium containing any nonadherent cells was carefully decanted, and multicellular biofilm material was disaggregated and removed from the flasks using sterile cell scrapers and placed into 2-ml screw-cap vials (Stratech, Amsterdam, Netherlands). Excess medium was removed by centrifugation at 8,000 × g for 2 min. RNA was extracted by mechanical disruption in TRIzol (Invitrogen, Paisley, United Kingdom) and purified using an RNeasy MinElute cleanup kit (Qiagen, Crawley, United Kingdom) according to the manufacturer's instructions. RNA was quantified and the quality was determined using a NanoDrop spectrophotometer (ND-1000; ThermoScientific, Loughborough, United Kingdom). cDNA was subsequently synthesized with Moloney murine leukemia virus (MMLV) reverse transcriptase (Invitrogen, Paisley, United Kingdom) using a MyCycler PCR machine (Bio-Rad Laboratories). The expression of the multidrug efflux pump gene, AfuMDR4, was then assessed by quantitative reverse transcription (RT)-PCR using SYBR GreenER (Invitrogen, Paisley, United Kingdom) according to the manufacturers' instructions. The primer set AfuMDR4-F (5′-GTCGCCGTTACTTTGAGAGC-3′) and AfuMRD4-R (5′-ATGAAGGCAACCACATAGGC-3′) (184 bp) and the β-tubulin housekeeping primer set AfuBtub-F (5′-CAATGGCTCCTCCGATCTCC-3′) and AfuBtub-R (5′-TCCATGGTACCAGGCTCG-3′) (115 bp) were used, as previously described (24). Three independent replicate samples from each strain for each parameter were analyzed in triplicate using an MxProP Quantitative PCR machine and MxProP 3000 software (Stratagene, Amsterdam, Netherlands), and gene expression was normalized to that of the β-tubulin gene according to the 2^−ΔΔCT method (12). The PCR amplification efficiencies of all target genes were optimized prior to analysis.

In vivo phase-dependent AfuMDR4 expression analysis.

A diffusion chamber kit (Millipore, Watford, United Kingdom) was modified to create the biofilm chamber. Briefly, a durapore polyvinylidene fluoride (PVDF) membrane (Millipore, Watford, United Kingdom) with a pore size of 0.45 μm was fixed to one side of Plexiglas rings (Millipore, Watford, United Kingdom), and a silicon sheet was fixed on the other side. The chambers were autoclaved before being implanted in mice. Male CD-1 mice were anesthetized with ketamine and xylazine. The dorsal flank of each mouse was shaved, and a small incision was made in the skin. The diffusion chamber was implanted subcutaneously in such a way that the silicon side of the chamber faced the skin and the durapore membrane was touching the body. The wound was closed using ethicon black braided-silk nonabsorbable sutures. Meloxicam (2 mg/kg of body weight) was administered intraperitoneally twice a day up to 72 h postsurgery. One week after the surgery, A. fumigatus ATCC 1163 (1 × 10⁵ spores/mouse) was injected into the tissue chamber under isoflurane anesthesia. This strain was used because it is less pathogenic than Af293 in this model system and has been used extensively in animal studies by the group. Mice were treated with vehicle (control), low-dose (LD) (3 mg/kg), and high-dose (HD) (10 mg/kg) voriconazole orally, as previously described (37). Mice, in at least triplicate, were euthanized at 0, 1, 4, and 24 h postinfection; the chamber was removed and opened; the biofilm material was processed for RNA extraction; and quantitative RT-PCR was performed as described above.

Statistics.

Analysis of variance (ANOVA) and t tests were used to investigate independent sample data (for different strains or samples from different animals). A Bonferroni correction for multiple comparisons was applied to the data where appropriate. SPSS (version 11; Chicago, IL) was used for these analyses, and GraphPad Prism (version 4; La Jolla, CA) was used for the production of the figures.

RESULTS

A. fumigatus exhibits phase-dependent resistance to voriconazole.

For all 14 A. fumigatus strains examined in this study, voriconazole consistently exhibited activity against germlings (8 h), with a MIC range between 0.0315 and 1 mg/liter (Table 1). The SMIC₉₀ of voriconazole increased by up to 16- to 256-fold against a monolayer of proliferating mycelia (12 h) compared to germlings. In those multicellular A. fumigatus populations where hyphae were densely intertwined (24 h), the voriconazole SMIC₉₀ was ≥256 mg/liter, which was >512 mg/liter higher than that for germlings.

Table 1.

A. fumigatus exhibits phase-dependent resistance to voriconazole

Strain	8 h			12 h			24 h
	SMIC90 (ΔMICa)	Ala-Nap (RFU/mg)	ΔMDR4c	SMIC90 (ΔMICb)	Ala-Nap (RFU/mg)	ΔMDR4	SMIC90 (ΔMICb)	Ala-Nap (RFU/mg)	ΔMDR4
Af293	0.5 (8)	1,286	1	8 (16)	618	3.89	>256 (>512)	402	2.35
YHCF1	0.0625 (2)	1,263	1	1 (16)	573	4.37	256 (>512)	697	3.38
YHCF2	0.0625 (4)	1,269	1	1 (16)	609	2.31	>256 (>512)	264	2.72
YHCF3	1.0 (4)	1,153	1	32 (32)	634	31.60	>256 (>512)	376	4.50
YHCF4	0.125 (8)	1,161	1	16 (128)	498	65.10	>256 (>512)	346	14.80
YHCF5	0.125 (8)			16 (128)			>256 (>512)
YHCF6	0.125 (4)			16 (128)			>256 (>512)
YHCF7	0.125 (8)			8 (64)			>256 (>512)
YHCF8	0.0312 (4)			4 (128)			>256 (>512)
YHCF9	0.0625 (4)			16 (128)			>256 (>512)
YHCF10	0.0625 (4)			32 (256)			>256 (>512)
YHCF11	0.0625 (8)			8 (64)			>256 (>512)
YHCF12	0.0625 (4)			8 (64)			>256 (>512)
YHCF13	0.0312 (4)			4 (128)			>256 (>512)

^aFold change in the MIC following efflux pump inhibitor challenge.

^bFold change in the MIC compared to the MIC for 8-h germling cells.

^cFold change in expression of AfuMDR4 compared to 8-h germling cells.

A. fumigatus exhibits phase-dependent Ala-Nap uptake that is induced by voriconazole treatment.

Analysis of efflux pump activity is presented as normalized fluorescence units (RFU/mg) in 8-, 12-, and 24-h populations (Fig. 1A and Table 1). The data demonstrated that 8-h germlings exhibited the highest levels of Ala-Nap uptake (∼1,150 RFU/mg), i.e., low levels of efflux. Both the 12-h and 24-h phases of growth demonstrated significantly lower levels of Ala-Nap uptake, i.e., ∼600 and ∼410 RFU/mg, respectively (ANOVA; P < 0.0001). Dispersal and washing of the mature biofilms (24 h) showed results equivalent to those for intact biofilms (data not shown).

Fig. 1.

A. fumigatus displays phase-dependent efflux activity and is induced by voriconazole. (A) Efflux activities of 8-, 12-, and 24-h biofilms. (B) Eight-hour biofilms treated with 0.0625 mg/liter of voriconazole and evaluated at 0, 1, 4, and 8 h. Standardized A. fumigatus conidia were grown in black flat-bottom 96-well plates for 8, 12, and 24 h and washed in buffer, and Ala-Nap (128 mg/liter), which is enzymatically cleaved to produce fluorescent β-naphthylamine, was added. The relative fluorescence normalized to the dry weight of the biofilms (RFU/mg) was determined over 120 min at an excitation wavelength of 320 nm and an emission wavelength of 460 nm using a fluorescent plate reader. High values indicate low efflux, and vice versa. Five different strains were used, and the experiments were performed on three separate occasions. The error bars represent the standard deviations of the means. Ala-Nap internal cleavage was normalized (RFU/mg) and shown to be maximal after 8 h growth, while it was significantly reduced at 12 and 24 h (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

The 8-h germling growth phase was subsequently examined to evaluate the effect of pretreatment with voriconazole (0.0625 mg/liter) on efflux activity (Fig. 1B). This phase was selected because quantifiable efflux activity was constitutively low among the isolates examined, allowing quantification of any changes following voriconazole treatment. The data demonstrated that after voriconazole treatment, the numbers of RFU/mg were reduced from ∼1,100 to ∼980 after 1 h and significantly reduced to ∼750 (P < 0.05) and ∼550 (P < 0.01) after 4 and 8 h, respectively.

Competitive inhibition of efflux pumps improves A. fumigatus sensitivity to voriconazole.

An EPI (MC-207,110) was used to assess the contribution of efflux pumps to voriconazole sensitivity. MC-207,110 is a competitive substrate of efflux pumps. A checkerboard assay was initially developed to test combinations of voriconazole and EPI. It was demonstrated that EPI did not exhibit any antifungal activity at any of the concentrations tested (>512 mg/liter). Therefore, 64 mg/liter was selected for all combination experiments. In the presence of EPI, the MIC range was significantly reduced, from 2- to 8-fold (t test; P < 0.05) (Fig. 2 and Table 1).

Fig. 2.

A. fumigatus biofilm sensitivity to voriconazole is increased in the presence of an efflux pump inhibitor. The data show that the MICs of A. fumigatus are significantly reduced in the presence of the efflux pump inhibitor MC-207,110, by approximately 2- to 8-fold (t test; *, P = 0.0196). A. fumigatus biofilms were grown in flat-bottom 96-well plates for 8 h and washed, and voriconazole (0.005 to 4 mg/liter) was added with or without MC-207,110 (64 mg/liter). Viability was assessed using alamar blue. Fourteen different strains were used, and the experiments were performed on three separate occasions. The error bars represent the standard deviations of the means.

A. fumigatus exhibits phase-dependent efflux pump expression (AfuMDR4) that is induced by in vitro voriconazole treatment.

The in vitro expression of AfuMDR4 transcripts was assessed in 8-, 12-, and 24-h phases of growth, compared to the fold increase in the SMIC₉₀ for five strains, and analyzed as a fold change in expression by normalization to the 8-h phase data (Table 1). Expression of AfuMDR4 was shown to be differentially upregulated in both the 12- and 24-h phases of growth in a strain-dependent manner, with maximal upregulation observed in the 12-h phase. The levels of upregulation appeared to correlate with the change in the SMIC₉₀, e.g., Af293 showed a 3.89-fold change in gene expression and a 16-fold change in MIC, whereas YHCF4 showed a 65.1-fold change in gene expression and a 128-fold change in MIC. At 24 h, gene expression was reduced, which did not correlate with reduced MIC levels.

These three populations were subsequently pretreated with voriconazole, and expression analysis of AfuMDR4 was performed at 1, 4, and 24 h posttreatment. Data were analyzed at each time point in comparison to matched untreated populations (Fig. 3). These data demonstrated that AfuMDR4 within the 8-h phase was significantly influenced by voriconazole exposure, increasing by 1.75-fold (P < 0.05), 2.5-fold (P < 0.001), and 4.2-fold (P < 0.001) after 1, 4, and 24 h, respectively. AfuMDR4 within the 12-h phase was also significantly influenced by voriconazole exposure, initially increasing 2.6-fold (P < 0.001), 1.9-fold (P < 0.01), and 4.4-fold (P < 0.001) after 1, 4, and 24 h, respectively. AfuMDR4 in the 24-h phase was also influenced by voriconazole exposure, but less dramatically, with increases of 1.65-fold (P < 0.05), 1.62-fold (P < 0.05), and 1.13-fold (P > 0.05) after 1, 4, and 24 h, respectively.

Fig. 3.

A. fumigatus AfuMDR4 expression is induced by exposure to voriconazole for 1, 4, and 24 h. Exposure to voriconazole induced AfuMDR4 expression in the 8-, 12-, and 24-h phases of A. fumigatus growth. Quantitative RT-PCR was performed on each cDNA population, using the β-tubulin gene as a housekeeping gene to calculate the relative expression of AfuMDR4 according to the 2^−ΔΔCT method (12). Three isolates were tested on three separate occasions for each time point. The error bars represent the standard deviations of the means. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

A. fumigatus exhibits phase-dependent efflux pump expression (AfuMDR4) that is increased by voriconazole treatment in vivo.

The in vivo expression of AfuMDR4 transcripts was assessed in vehicle control- and in 3 mg/kg and 10 mg/kg voriconazole-treated animals at 0, 1, 4, and 24 h posttreatment and analyzed as a fold change in expression by normalization to the baseline control. Expression of AfuMDR4 was shown to be significantly upregulated according to ANOVA following treatment with voriconazole after 1, 4, and 24 h (Fig. 4). Significant upregulation was observed at 1 and 4 h with 3- and 10-mg/kg doses (P < 0.01, P < 0.05, P < 0.01, and P < 0.01, respectively). At 24 h, upregulation of AfuMDR4 was still observed (P < 0.05 and P < 0.01, respectively).

Fig. 4.

A. fumigatus AfuMDR4 is expressed in vivo and is induced by voriconazole. Treatment with voriconazole (3 and 10 mg/kg) induced AfuMDR4 expression 1, 4, and 24 h posttreatment. Following removal of the subcutaneous A. fumigatus ATCC 1163 biofilm and RNA processing, quantitative RT-PCR was performed on each cDNA population, using the β-tubulin gene as a housekeeping gene to calculate the relative expression of AfuMDR4 according to the 2^−ΔΔCT method (12). Three animals were used as a baseline for comparison to vehicle (n = 3 at 1, 4, and 24 h), 3 mg/kg (n = 3 at 1 and 4 h; n = 5 at 24 h), and 10 mg/kg (n = 3 at 1 and 4 h; n = 4 at 24 h). The error bars represent the standard deviations of the means. *, P < 0.05; **, P < 0.01.

DISCUSSION

A. fumigatus triazole resistance has become increasingly associated with the ergosterol biosynthetic pathway through mutation of the cyp51A gene locus. Increasing evidence suggests, however, that efflux pump-mediated mechanisms may also be important determinants of resistance, especially given their characterized role in C. albicans. This study has demonstrated for the first time that efflux pump activity is associated with biofilm development and is also induced by triazole exposure, both of which contribute to azole resistance.

We first demonstrated in a panel of clinical isolates, using modified MIC methodologies, that A. fumigatus in different phases of filamentous growth showed increased resistance to voriconazole, a phenomenon previously demonstrated for Af293 by our group (21). Even when biofilm cells were disaggregated and standardized to similar cell counts using absorbance readings, reduced sensitivity to VRZ was observed, which we also previously reported for C. albicans (31). Recent studies are in agreement with these observations, as it has recently been shown that triazole sensitivity is limited by complex filamentous growth (35). Whether this is because the physical quantity of drug is too low to elicit an effect or because subtle transcriptional events occur in response to antifungal challenge, leading to reduced sensitivity, still remains to be determined. It is possible that the inoculum effect plays a role in limiting the effectiveness of VRZ (10). A recent study reported that voriconazole treatment of C. albicans biofilms induced a calcineurin-dependent response, resulting in changes in cell wall integrity that led to antagonism of the cell wall-active micafungin (9). Moreover, global transcriptional analyses of voriconazole-treated A. fumigatus mycelia demonstrated that over 2,000 genes were differentially expressed, and among these there were increased levels of transporter mRNA (5).

Based on this and our own unpublished microarray studies, we decided to focus on efflux as a potential mechanism of resistance and analyzed the biochemical activity of efflux pumps in our model system. Active extrusion of a fluorescent molecule was shown to be elevated at 24 h and 12 h in comparison to 8-h germling cells. Voriconazole treatment of the 8-h germling cells resulted in a time-dependent increase in efflux pump activity. The 12-h and 24-h cells were not tested, as efflux was constitutively elevated, making differentiation of efflux activity difficult. This was potentially because of the role of extracellular-matrix material, which has been shown to impede diffusion of molecules (25). Disruption and standardization of the 24-h biofilm demonstrated fluorescence equivalent to that of the undisrupted biofilm, suggesting that poor penetration of Ala-Nap was not a factor in the low levels of fluorescence detected in these assays. To demonstrate that efflux was involved in the reduced sensitivity to voriconazole, we used a broad-spectrum efflux pump inhibitor, MC-207,110, a substrate of efflux pumps. While no studies have previously used this compound in eukaryotic studies, we demonstrated a significant reduction in the MIC of voriconazole of 2- to 8-fold with a panel of isolates. This suggests that this molecule has utility for future efflux pump inhibitor studies.

Several studies have examined the expression of efflux pumps in C. albicans biofilms, where constitutive expression of CaMDR1 was reported during early adhesion and in a phase-dependent manner thereafter (15, 23, 31). However, to date, no studies have reported this in A. fumigatus biofilms. Based on previous work on itraconazole-resistant mutants, we decided to focus our investigations on AfuMDR4, which served as a quantifiable example for these studies (24). qPCR analysis demonstrated strain-dependent upregulation of the gene, particularly in the 12-h cells. Interestingly, those strains that exhibited the most elevated resistance and active biochemical efflux showed the highest levels of gene expression. Furthermore, voriconazole was shown to induce significant upregulation of AfuMDR4 transcripts in treated 8-, 12-, and 24-h cells in a time-dependent manner. Ferreira and colleagues also observed azole-constitutive and -induced efflux pump expression on 24-h cultures following treatment with itraconazole (4). Here, it was shown that AfuMDR1 to -4 were differentially expressed following itraconazole treatment and that AfuMDR3 and AfuMDR4 were preferentially induced on exposure to the azole, which has also been reported for C. albicans (7).

The clinical relevance of these transcriptional changes remains to be seen. Therefore, to evaluate whether efflux pump expression occurred in vivo, we devised an implanted A. fumigatus biofilm model to investigate the effects of voriconazole treatment on AfuMDR4 expression. Expression was shown to be constitutive, and the expression was upregulated in response to treatment with two different doses of voriconazole. Similarly, the expression of CaMDR1 in a catheter-based C. albicans biofilm model has been reported, but at a lower constitutive level (26). Moreover, the same group reported induction of CaCDR1 and CaCDR2 in their in vivo model following fluconazole treatment but no change in the expression of CaMDR1 (11).

Azole resistance cannot be explained by the cyp51A mutation (2) or drug efflux, as is proposed here. In this study, we demonstrated maximal strain-dependent transcript expression at 12 h, yet voriconazole resistance increased to ≥256 mg/liter for all strains tested. Based on previous work by Ramage and coworkers (31) on C. albicans biofilms with CaMDR and CaCDR mutants, it was shown that the resistant phenotype existed despite high sensitivity of planktonic cells to fluconazole. However, disruption and washing of the biofilms to remove extrapolymeric matrix significantly reduced the sessile MIC (31). Expression of these pumps was shown to be upregulated in the biofilm, suggesting that efflux pump activity played a contributory role in azole resistance but was not the principal mechanism. Given that the gene expression profile of A. fumigatus was generally downregulated at 24 h in comparison to 12 h, it is likely that the presence of extracellular-matrix material within the biofilm plays an active role in resistance, as has been recently shown for C. albicans biofilms (25). These studies have shown that Fks1p is able to bind and sequester azoles and other antifungal classes. It is therefore likely that A. fumigatus is able to utilize its own polysaccharides for this purpose. It is plausible that resistance is a highly regulated process, with efflux pump mechanisms playing an early role in resistance during colonization, which is then reduced as extrapolymeric matrix is produced during biofilm maturation.

Collectively, the data presented here, in addition to the available literature, support the hypothesis that efflux pumps are an important, but not exclusive, determinant of resistance to azoles (3, 18). Their primary role may be in homeostasis within complex environments to protect themselves from acute toxicity (29), but within clinical environments, such as an aspergilloma, exposure to azoles dugs may enhance the levels of efflux pump expression, thereby either contributing to or inducing clinical resistance (2). Indeed, expression of these pumps has been shown biochemically, indicating the transcripts are translated into protein. These observations may explain why treatment failure with azoles may occur clinically and should raise doubts as to whether sequential azole therapy is an appropriate treatment strategy for aspergillosis. Further studies are required to establish how A. fumigatus efflux pumps are regulated and whether inhibitors of them could augment azole therapies.