Inhibitory Potential of Antifungal Drugs on ATP-Binding Cassette Transporters P-Glycoprotein, MRP1 to MRP5, BCRP, and BSEP

Abstract

Inhibition of ABC transporters is a common mechanism underlying drug-drug interactions (DDIs). We determined the inhibitory potential of antifungal drugs currently used for invasive fungal infections on ABC transporters P-glycoprotein (P-gp), MRP1 to MRP5, BCRP, and BSEP in vitro. Membrane vesicles isolated from transporter-overexpressing HEK 293 cells were used to investigate the inhibitory potential of antifungal drugs (250 μM) on transport of model substrates. Concentration-inhibition curves were determined if transport inhibition was >60%. Fifty percent inhibitory concentrations (IC₅₀s) for P-gp and BCRP were both 2 μM for itraconazole, 5 and 12 μM for hydroxyitraconazole, 3 and 6 μM for posaconazole, and 3 and 11 μM for isavuconazole, respectively. BSEP was strongly inhibited by itraconazole and hydroxyitraconazole (3 and 17 μM, respectively). Fluconazole and voriconazole did not inhibit any transport for >60%. Micafungin uniquely inhibited all transporters, with strong inhibition of MRP4 (4 μM). Anidulafungin and caspofungin showed strong inhibition of BCRP (7 and 6 μM, respectively). Amphotericin B only weakly inhibited BCRP-mediated transport (127 μM). Despite their wide range of DDIs, azole antifungals exhibit selective inhibition on efflux transporters. Although echinocandins display low potential for clinically relevant DDIs, they demonstrate potent in vitro inhibitory activity. This suggests that inhibition of ABC transporters plays a crucial role in the inexplicable (non-cytochrome P450-mediated) DDIs with antifungal drugs.

INTRODUCTION

Invasive fungal infections are a leading cause of infection-related mortality in immunocompromised individuals and patients with serious underlying conditions. Although prophylactic or therapeutic use of azole antifungal drugs has substantially improved treatment outcome, these drugs display a significant potential for clinically relevant drug-drug interactions (DDIs) (1, 2). As a result, the pharmacokinetics (PK) of both drugs may alter, resulting in either increased plasma concentrations with subsequent risk of adverse events or subtherapeutic plasma concentrations potentially leading to therapeutic failure.

With regard to phase I and II enzymes, the pharmacokinetic profiles of antifungal drugs have been quite well characterized (2, 3). Yet it also appears that both uptake and efflux transporters are considered a major intervenient in drug PK, and inhibition of such transporters is an important mechanism underlying DDIs (4, 5).

Transporters of the ATP-binding cassette (ABC) transporter protein family are involved in unidirectional, cellular efflux of drugs (6). They are multidomain, integral membrane proteins which all exhibit the capacity to actively transport physiological substrates (e.g., peptides, lipids, and inorganic ions) across extra- and intracellular membranes at the expense of ATP hydrolysis. Expression of these transporters in the cellular membranes of the gastrointestinal tract, blood-brain barrier, liver, and kidneys suggests that they also hold a key position in the cellular protection against toxic compounds and drugs (7). Specifically, P-glycoprotein (P-gp/ABCB1), several isoforms of multidrug resistance-associated proteins (MRP/ABCC) and breast cancer resistance protein (BCRP/ABCG2) have been shown to influence drug PK by extruding a large variety of xenobiotics from cells back to either external medium or blood (8). In addition, although bile salt export pump (BSEP/ABCB11) mediates the canalicular export of bile salts from liver into bile, which can be inhibited by several drugs and may result in cholestatic liver injury (9, 10), it has also been demonstrated to transport drugs (11).

Because ABC transporters have a broad substrate spectrum, interaction of drugs with such a transporter could result in mutual transport inhibition (5). This inhibition could consequently alter the PK of substrates of the inhibited transporter and promote DDIs (5, 6).

Knowledge of the inhibitory potential of antifungal drugs on ABC-mediated transport activity is crucial in understanding the molecular mechanisms underlying non-cytochrome P450 (non-CYP)-mediated DDIs of these drugs. Also, this may be of great importance in explaining variations in local (nonsystemic) concentrations of drugs, thereby increasing safety and efficacy.

In the present study, we examined the inhibitory potential of 10 antifungal drugs currently used in the treatment of invasive fungal infections (fluconazole, itraconazole, hydroxyitraconazole, voriconazole, posaconazole, isavuconazole, anidulafungin, caspofungin, micafungin, and amphotericin B) on the efflux of model substrate via ABC transporters P-gp, MRP1 to MRP5, BCRP, and BSEP, using a vesicular overexpression transport assay.

(This work has been presented as a poster at the European Congress on Clinical Microbiology and Infectious Diseases, 2014, Barcelona, Spain.)

MATERIALS AND METHODS

Inside-out membrane vesicles were isolated from human embryonic kidney (HEK) 293 cells (overexpressing a single ABC transporter). These vesicles were used to determine the inhibitory potential of antifungal drugs (250 μM) on ATP-dependent transport of model substrates. Fifty percent inhibitory concentrations (IC₅₀s) were determined if an antifungal drug inhibited transport for >60%.

A detailed description of these individual steps is outlined below.

Materials.

The following antifungal drugs were kindly provided by the manufacturers: hydroxyitraconazole (Janssen Pharmaceuticals, Inc.), voriconazole and anidulafungin (Pfizer, Inc.), posaconazole and caspofungin (Merck, Sharp and Dohme Corp.), isavuconazole (Basilea Pharmaceutica), and micafungin (Astellas Pharma). Fluconazole, itraconazole, and amphotericin B were purchased (Sigma-Aldrich, Zwijndrecht, The Netherlands). Substrates [³H]estrone sulfate ([³H]E₁S), [³H]estradiol 17β-glucuronide ([³H]E₂17βG), and [³H]taurocholic acid ([³H]TCA) were purchased from PerkinElmer, Inc. (Waltham, MA). [³H]methotrexate ([³H]-MTX) was purchased from Moravek Biochemicals (Brea, CA). [³H]N-methyl-quinidine ([³H]NMQ) and unlabeled NMQ were obtained from Solvo Biotechnology (Szeged, Hungary). ATP disodium salt (ATP; from a bacterial source) was purchased from Sigma-Aldrich (Zwijndrecht, The Netherlands). MultiScreen_HTS filter plates and vacuum manifold filtration devices were purchased from Millipore (Etten-Leur, The Netherlands). Opti-Fluor scintillation fluid was purchased from PerkinElmer, Inc. (Waltham, MA).

Transporter-overexpressing membrane vesicles.

Membrane vesicles from HEK 293 cells with overexpression of a single type of ABC transporter (i.e., P-gp, MRP1 to MRP5, BCRP, or BSEP) or enhanced yellow fluorescent protein (eYFP; a cytosolic protein, used as a negative control) were purchased from PharmTox (Nijmegen, The Netherlands).

Vesicular transport assay.

For quality assurance, the vesicles were tested for functionality using a rapid filtration technique (12). Vesicles overexpressing the transporter MRP1, MRP3, or MRP4 were incubated with [³H]E₂17βG (0.1 μM/0.15 μCi), vesicles overexpressing MRP2 or MRP5 with [³H]MTX (0.4 μM/0.15 μCi), and vesicles overexpressing BSEP with [³H]TCA (0.1 μM/0.15 μCi) for 5 min at 37°C. P-gp-overexpressing vesicles were incubated with [³H]NMQ (0.1 μM/0.015 μCi) and BCRP-overexpressing vesicles with [³H]E₁S (0.1 μM/0.1 μCi) for 1 min at 37°C. The activity of the eYFP vesicles (negative control) was measured for all substrates. Incubations were performed using 7.5 μg of protein in Tris sucrose (TS) buffer (pH 7.4) supplemented with 10 mM MgCl₂, 4 mM ATP, or 4 mM AMP (negative control) in a total volume of 30 μl. The process of incubation was stopped on ice. After addition of 150 μl of ice-cold TS buffer, the samples were filtered using a MultiScreen_HTS vacuum manifold filtration device through TS buffer-preincubated 0.45-μm-pore 96-well Millipore filters. The filters were washed twice with 200 μl of TS buffer. After addition of 2 ml of Opti-Fluor scintillation fluid, the samples were counted using a liquid scintillation counter. A ratio was calculated dividing the ATP value by the AMP value. Experiments were performed in triplicate.

Antifungal inhibitory effects on model substrate transport.

The vesicular transport assay was performed as described above. Antifungal drugs were dissolved in dimethyl sulfoxide (DMSO; fluconazole, itraconazole, hydroxyitraconazole, voriconazole, posaconazole, isavuconazole, anidulafungin, and amphotericin B) or in ultrapure water (caspofungin and micafungin) to a final concentration of 12.5 mM. At higher concentrations, the maximum solubility of itraconazole in DMSO was reached. Therefore, concentration-dependent analysis of itraconazole was limited to 100 μM. Antifungal drugs (250 μM) or solvents (DMSO or ultrapure water) were added to the mixture before incubation. For all mixtures, ATP was added and eYFP was used as a negative control. Screens were performed in triplicate. After subtraction of eYFP, mean values of substrate transport were expressed as percentages (100% for samples with solvent). The concentration of 250 μM was used to ascertain that most inhibitory effects would be observed. Antifungal drugs inhibiting transport of the model substrate for more than 60% at 250 μM (empirically chosen based on previous experiments) were selected to further determine concentration-dependent inhibition and, ultimately, 50% inhibitory concentrations.

Concentration-dependent inhibitory effects.

Transport studies were performed as described above, but with antifungal drug concentrations ranging from 0 to 250 μM. For each concentration, four data points were recorded and the control values (eYFP) were subtracted.

Data analysis.

Mean values for substrate-specific transport were expressed as percentages relative to the control (after subtraction of eYFP background transport), with the solvent control representing 100% transport. Mean percentages of relative transport per transporter interaction were pooled. All transport data were expressed as means ± standard deviations. Drug concentrations were plotted as logarithmic concentrations. Concentration-dependent inhibition data were analyzed by nonlinear regression analysis according to a one-site binding model with variable slope (also called the Hill equation), using GraphPad Prism software, version 5.03 (Graphpad Software Inc., San Diego, CA).

RESULTS

Functionality of the transporter-overexpressing membrane vesicles.

The vesicular transport assay showed functionality of all transporters compared to that of the negative control. The AMP-ATP ratios of the transporters P-gp, MRP1 to MRP5, BCRP, and BSEP are shown in Table S1 in the supplemental material.

Antifungal inhibitory effects on model substrate transport.

The inhibitory potentials of antifungal drugs (250 μM) on the transport of radiolabeled model substrates via ABC transporters P-gp, MRP1 to MRP5, BCRP, and BSEP are shown in Fig. 1.

FIG 1.

Inhibitory potential of antifungal drugs (250 μM) on ABC transporter activity.

The azole antifungals fluconazole and voriconazole did not inhibit the transport of model substrate of any selected transporter. Itraconazole, hydroxyitraconazole, and posaconazole all significantly inhibited (>60%) P-gp, BCRP, and BSEP. Isavuconazole was capable of significant inhibition of P-gp and BCRP.

The echinocandin anidulafungin inhibited P-gp-, BCRP-, and MRP4-mediated transport of model substrates. Caspofungin inhibited P-gp-, BCRP-, BSEP-, MRP1-, and MRP3-mediated transport. Micafungin was capable of inhibiting substrate transport of all transporters for more than 60%.

Amphotericin B was capable of inhibiting substrate transport only below the threshold of 60% for BCRP.

Stimulation of transport was observed in some cases (Fig. 1). This uptake of the model substrate for >100% compared to the value for the control was especially seen with MRP2 (all antifungals with exception of micafungin and amphotericin B).

Concentration-dependent inhibitory effects.

IC₅₀s of the concentration-dependent inhibition by selected antifungal drugs (showing >60% inhibition at a concentration of 250 μM) are presented in Fig. 2 to 4. Subsequent IC₅₀s for each interaction are shown in Fig. 5.

FIG 2.

Concentration-dependent inhibition of substrate transport by the azole antifungals itraconazole (A), hydroxyitraconazole (B), posaconazole (C), and isavuconazole (D).

FIG 4.

Concentration-dependent inhibition of substrate transport by the polyene antifungal amphotericin B.

FIG 5.

Overview of antifungal drug concentrations (micromolar) at 50% inhibition (95% CI) of ATP-dependent uptake of the model substrate (IC₅₀). Inhibition potentials range from relatively strong inhibition (red; 0 to 5 μM) to relatively mild inhibition (yellow; 20 to 25 μM) and eventually relatively low inhibition (blue; 50 μM and up), using intermediate colors with increments of 5 μM. FLZ, fluconazole; ITZ, itraconazole; hITZ, hydroxyitraconazole; PSZ, posaconazole; VCZ, voriconazole; ISA, isavuconazole; ANF, anidulafungin; CAS, caspofungin; MCF, micafungin; AMB, amphotericin B.

We found a 50% inhibitory effect on P-gp-mediated transport of NMQ for the azole antifungals itraconazole, hydroxyitraconazole, posaconazole, and isavuconazole at the low micromolar concentrations of 2 (95% confidence interval [95% CI], 1.1 to 2.3), 5 (2.9 to 7.7), 3 (2.2 to 4.1), and 3 (2.7 to 3.9) μM, respectively (Fig. 2 and 5). In addition, IC₅₀s (95% CI) for P-gp were seen at somewhat higher concentrations for the echinocandin antifungals anidulafungin, caspofungin, and micafungin at 38 (23 to 61.8), 34 (26.8 to 43.7), and 45 (29.2 to 70.5) μM, respectively (Fig. 3 and 5).

FIG 3.

Concentration-dependent inhibition of substrate transport by the echinocandin antifungals anidulafungin (A), caspofungin (B), and micafungin (C).

The echinocandins were the only antifungals capable of inhibition of MRP transporters. For MRP1, MRP3, and MRP5, the IC₅₀s (95% CI) of micafungin were 21 (19.9 to 23.0), 42 (38.2 to 46.7), and 22 (19.3 to 26.0) μM, respectively. Although less potent, caspofungin inhibited transport of E₂17βG via MRP1 and MRP3 for 50% (95% CI) at 112 (88.7 to 140.7) and 158 (91.6 to 272.1) μM, respectively (Fig. 3 and 5).

Transport of MTX via MRP2 was inhibited for 50% (95% CI) at a relatively high concentration of micafungin (148 [81.8 to 266] μM) compared to the IC₅₀s for other efflux pumps (Fig. 3 and 5). Micafungin showed strong inhibitory effects (95% CI) on MRP4-mediated transport (4 [3 to 5.4] μM), whereas anidulafungin had an IC₅₀ that was somewhat higher, 28 (21.7 to 36.3) μM (Fig. 3 and 5).

Inhibition of substrate transport via BCRP at low micromolar concentrations was found for itraconazole, hydroxyitraconazole, posaconazole, isavuconazole, anidulafungin, and caspofungin (IC₅₀ [95% CI] = 2 [1.6 to 3.6], 12 [7.4 to 20.2], 6 [3.9 to 8.1], 11 [5.2 to 2.3], 7 [6.1 to 8.3], and 6 [2.5 to 13.2], respectively) (Fig. 2, 3, and 5). Inhibitory effects (95% CI) on transport at higher concentrations, 21 (17.4 to 25.4) and 127 (86.8 to 186.3) μM, were found for micafungin and amphotericin B (Fig. 2, 4, and 5).

For BSEP, the IC₅₀s (95% CI) of itraconazole and hydroxyitraconazole were 3 (2.7 to 3.4) and 17 (16.1 to 18.6) μM, respectively (Fig. 2 and 5). Posaconazole, caspofungin, and micafungin all showed higher IC₅₀s (95% CI), i.e., 34 (29 to 40.4), 186 (113.5 to 305.6), and 85 (72.8 to 100) μM, respectively (Fig. 2, 3, and 5).

DISCUSSION

In the present study, we assessed the in vitro inhibitory potential of 10 established and novel antifungal drugs currently used in the treatment of invasive fungal infections on the transport of radiolabeled model substrates via ABC transporters shown to influence drug pharmacokinetics due to their involvement in drug transport, i.e., P-gp, MRP1 to MRP5, BCRP, and BSEP. Inhibition of these transporters may be an important mechanism underlying clinically significant DDIs. We found that the azole antifungals fluconazole and voriconazole were unable to inhibit transport of the model substrate for any of the studied transporters for more than 60% at 250 μM. Itraconazole, hydroxyitraconazole, posaconazole, and isavuconazole inhibited a selection of transporters, but in a very potent fashion. In contrast, echinocandin antifungals seemed appreciable inhibitors of transport activity throughout the range of transporters investigated.

The clinical context of our findings is outlined in the following examples. The azole antifungal itraconazole showed strong inhibition of P-gp-mediated transport, with an IC₅₀ of 2 μM (Fig. 5). Inhibitory effects of similar magnitude on P-gp have been previously reported (13, 14). This in vitro interaction supports the findings of a clinically significant DDI of itraconazole and digoxin, a cardiac glycoside mainly excreted unchanged in the urine and a known substrate of P-gp in vitro and in vivo (15, 16). During concomitant treatment, itraconazole was thought to inhibit P-gp-mediated digoxin secretion in the renal tubular cells (17, 18), which can now be confirmed by the conducted experiment.

Itraconazole showed strongest inhibition of BCRP-mediated transport (IC₅₀ = 0.4 μM[Fig. 5]). This inhibition at concentrations below 1 μM is in agreement with other in vitro cellular uptake experiments (19). In clinical practice, the inhibition of BCRP by itraconazole may be the mechanism behind the unexplained increases in exposure and peak concentrations of rosuvastatin (a known substrate of BCRP but only a poor substrate for CYP enzymes) during concomitant itraconazole use (20, 21).

Itraconazole also showed strong inhibition of BSEP (IC₅₀ = 3 μM [Fig. 5]), a transporter mediating secretion of bile salts into bile (22). As itraconazole has been shown to induce cholestatic liver injury (23), inhibition of either MDR3 (24) or BSEP (9, 10) has been reported as an underlying mechanism. Our data indicate that itraconazole only inhibits BSEP and is likely capable of obstructing efflux of bile acids out of the hepatocyte and inducing cholestasis. This emphasizes not only that inhibition of efflux transporters contributes to DDIs with concomitant medications but also that interactions of endogenous substrates are possible with the consequence of toxicity.

Similar to the case with itraconazole, our data demonstrate strong inhibition of P-gp and BCRP activity by hydroxyitraconazole and posaconazole (Fig. 5). Because the molecular structures and physicochemical properties of these three azole antifungals are closely related (25–27), they may exert similar inhibitory potentials for ABC transporters. This may also be hypothesized for the structural analogues fluconazole and voriconazole (28, 29), which are both incapable of inhibiting transport for more than 60% at 250 μM (Fig. 1). The absence of inhibitory potential of fluconazole for P-gp and both fluconazole and voriconazole for BCRP has been demonstrated by others (13, 14, 19, 30, 31). Itraconazole, hydroxyitraconazole, and posaconazole show a more pronounced lipophilic character than do fluconazole and voriconazole (32). A positive correlation between the lipophilicity of azole antifungals and inhibition of P-gp was reported (31), underpinning our findings. Substrates that are translocated by P-gp can be presented to the transporter binding site directly from the lipid bilayer, which is different for the other transporters, where compounds reach the external loops from the cytosol (5), possibly explaining the high affinity of lipophilic azoles itraconazole, hydroxyitraconazole, and posaconazole for P-gp. Lastly, this study is the first to report that isavuconazole showed strong inhibition of P-gp and BCRP (Fig. 5). This is in agreement with in vitro studies of isavuconazole, although this inhibition does not alter the PK of digoxin and methotrexate (probe substrates of P-gp and BCRP, respectively) in clinical studies performed with healthy volunteers (33–35).

For the echinocandins, we demonstrated inhibition over a wide range of ABC transporters (Fig. 1). In contrast, echinocandins show relatively few clinically significant DDIs in vivo, of which the underlying pharmacological mechanisms are mostly unknown. Because echinocandins show a low potential for inhibition of CYP enzymes, inhibition of the ABC transporters as found in our study might be a plausible underlying mechanism causing the few reported DDIs. For example, concomitant use of micafungin (a weak inhibitor of CYP3A4 [14]) and sirolimus resulted in an increased sirolimus area under the plasma concentration time curve (AUC) of 22% (36). This clinically significant increase could possibly involve the inhibition of P-gp, as sirolimus is reported to be a substrate of this transporter (37).

Comparing IC₅₀s with plasma concentrations should be done with greatest caution, since the obtained IC₅₀s represent free-drug concentrations. Most antifungal drugs are all highly bound to plasma proteins, reducing the free concentration able to inhibit transport of the substrate (Table 1). In addition, due to possible accumulation of antifungal drugs in tissue compartments, intracellular concentrations at the target site may be higher than plasma concentrations.

TABLE 1.

Overview of antifungal drug peak concentrations and protein binding at steady state during typical adult dosing^a

Drug	Cmax (mg/liter)	Cmax (μM)	Protein binding (%)
FLZ	6–20	19.6–65.3	10.0
ITZ	0.5–2.3	0.7–3.3	99.8
hITZ	0.6–2.1	0.8–2.9	99.5
VCZ	3.0–4.6	8.6–13.2	58.0
PCZ	1.5–2.2	2.1–3.1	99.0
ISA	7.5	21.6	>99
ANF	7.2	6.3	99
CAS	12.1	11.1	97
MCF	8.8	6.9	>99
AMB	0.5–2.0	0.5–2.2	>95.0

^aInformation obtained from references 33, 38, and 44. C_max, antifungal drug peak concentration; FLZ, fluconazole; ITZ, itraconazole (oral solution formulation); hITZ, hydroxyitraconazole (oral solution formulation); PSZ, posaconazole; VCZ, voriconazole; ISA, isavuconazole; ANF, anidulafungin; CAS, caspofungin; MCF, micafungin; AMB, amphotericin B.

Amphotericin B was only found to inhibit BCRP for >60% at high concentrations (Table 1). Although amphotericin B is in widespread clinical use, little is known about clinically relevant transporter-mediated DDIs. Because amphotericin B is associated with nephrotoxicity, it is known to decrease clearance of other renally eliminated drugs (38). In agreement with previous literature (14), P-gp-mediated transport of substrate was not inhibited by amphotericin B in our study. Moreover, we did not observe any inhibition of other transporters at relevant concentrations (Fig. 1).

Besides inhibition of transport, stimulation was also observed in some cases (Fig. 1), especially for MRP2. This is in concordance with other transport assays, in which increased transport of E₂17βG via MRP2 was seen. It was suggested that MRP2 can be allosterically stimulated by drugs, causing increased transport of the model substrate (39). These stimulatory effects on ABC transporters have also been observed for other drug classes, e.g., tuberculosis drugs (40).

A limitation of this study is that the observed inhibition of the transporters may be dependent on the translocated substrate. Different probe substrates can shift IC₅₀s, as previously reported for P-gp (41, 42). Therefore, caution should be exercised by translating these results to different transporter-substrate combinations. Regarding the mode of inhibition, the inside-out vesicles are less suitable for study of inhibition by hydrophilic compounds, as these might not be able to reach the inside of the vesicles. Since micafungin and amphotericin B have the lowest logP values, we demonstrated that these compounds were present inside the vesicles at high concentrations (data not shown). Moreover, the current study does not provide insight into whether or not the investigated antifungal agents are substrates of these transporters themselves.

Interestingly, despite the wide range and variability of CYP-mediated DDIs by azole antifungals in the clinical setting (1, 3), these drugs exhibited selective inhibition of efflux transporters in our study. Strikingly, despite the low potential for clinically relevant DDIs by echinocandins (38), an unprecedented inhibitory potential of the echinocandin antifungals on efflux transporters is observed. This suggests that the studied efflux pumps might play a crucial role in the inexplicable (non-CYP-mediated) DDIs of antifungal drugs, specifically for the echinocandins.

This study demonstrates the inhibitory potential of various established and novel antifungal drugs on the transport of model substrates via ABC drug efflux transporters. We are of the opinion that data generated in this research provide new insights into molecular mechanisms and help in further resolution of inexplicable (non-CYP-mediated) DDIs with antifungal drugs.