Atovaquone and quinine anti-malarials inhibit ATP binding cassette transporter activity

Background Therapeutic blood plasma concentrations of anti-malarial drugs are essential for successful treatment. Pharmacokinetics of pharmaceutical compounds are dependent of adsorption, distribution, metabolism, and excretion. ATP binding cassette (ABC) transport proteins are particularly involved in drug deposition, as they are located at membranes of many uptake and excretory organs and at protective barriers, where they export endogenous and xenobiotic compounds, including pharmaceuticals. In this study, a panel of well-established anti-malarial drugs which may affect drug plasma concentrations was tested for interactions with human ABC transport proteins. Methods The interaction of chloroquine, quinine, artemisinin, mefloquine, lumefantrine, atovaquone, dihydroartemisinin and proguanil, with transport activity of P-glycoprotein (P-gp), breast cancer resistance protein (BCRP), bile salt export pump (BSEP) and multidrug resistance-associated proteins (MRP) 1–4 were analysed. The effect of the anti-malarials on the ATP-dependent uptake of radio-labelled substrates was measured in membrane vesicles isolated from HEK293 cells overexpressing the ABC transport proteins. Results A strong and previously undescribed inhibition of BCRP-mediated transport by atovaquone with a 50% inhibitory concentration (IC50) of 0.23 μM (95% CI 0.17-0.29 μM) and inhibition of P-gp-mediated transport by quinine with an IC50 of 6.8 μM (95% CI 5.9-7.8 μM) was observed. Furthermore, chloroquine and mefloquine were found to significantly inhibit P-gp-mediated transport. BCRP transport activity was significantly inhibited by all anti-malarials tested, whereas BSEP-mediated transport was not inhibited by any of the compounds. Both MRP1- and MRP3-mediated transport were significantly inhibited by mefloquine. Conclusions Atovaquone and quinine significantly inhibit BCRP- and P-gp- mediated transport at concentrations within the clinically relevant prophylactic and therapeutic range. Co-administration of these established anti-malarials with drugs that are BCRP or P-gp substrates may potentially lead to drug-drug interactions.

Background ATP binding cassette (ABC) transporters are membranebound proteins that allocate a wide variety of compounds at the expense of ATP, even against steep concentration gradients [1]. P-glycoprotein (P-gp/ABCB1), bile salt export pump (BSEP/ABCB11), multidrug resistance-associated proteins (MRP1-4/ABCC1-4), and breast cancer resistance protein (BCRP/ABCG2) are among the most important drug transporters of the ABC protein family. ABC transport proteins are known for their capacity to protect the organism from potentially toxic xenobiotics through excretion, thereby decreasing intracellular concentrations. Indeed, typical localization of these export transporters are at the blood-brain barrier, placenta, gut, and at the apical side of liver and kidney cells. Two compounds may interact with the same transport protein through induction of expression, inhibition of protein function or competition of substrates. Pharmacokinetics of co-administered drugs can be critically altered when drug-drug interactions occur at the level of the ABC transport proteins, as distribution and selective excretion of these compounds may depend heavily on ABC protein-mediated transport. This can be reflected either in unexpected high blood plasma concentrations potentially causing toxic effects, or subtherapeutic concentrations at the site of action, diminishing therapeutic effects.
It is essential to assure effective blood plasma concentrations upon treatment with anti-malarial compounds in order to cure severely ill patients and prevent resistance acquisition through exposure of the parasite to sublethal blood plasma concentrations. The first-line treatment as recommended by the World Health Organization (WHO) currently consists of artemisinin-based combination therapy [2]. However, resistance against these regimens has been detected and the number of anti-malarials that can be subsequently applied are limited [3]. Toxic effects by unintended elevated blood plasma concentrations, however, should also be avoided.
Direct interaction with ABC transporter capacity of antimalarial compounds has not been explored in detail. In vitro assays have indicated a possible effect on Pgp-mediated transport or expression after exposure to chloroquine, quinine, mefloquine, primaquine, amodiaquine, piperaquine, artemisinin, and dihydroartemisinin, however, contradictory conclusions concerning the interaction of anti-malarial compounds with ABC transport proteins could be drawn from different experimental set-ups [4][5][6][7][8][9]. A possible interaction of anti-malarial compounds with MRP-type transporters and BCRP has also been described [10][11][12][13]. Co-administration of anti-malarial compounds with other drug types is highly anticipated. For instance, human immunodeficiency virus (HIV) and malaria co-infections are likely to occur, as there is a high overlap in geographical dissemination [14]. Therefore, the effect of anti-malarial compounds on ABC-mediated transport capacity should be explored in more detail in order to secure the most effective treatment strategies for patients receiving multiple drug regimens.

Cell culture and transduction
HEK293 cells were grown to 40% confluency in Dulbecco's modified Eagle's medium-GlutaMAX-I containing 10% foetal calf serum at 5% CO 2 in 500 cm 2 triple flasks. Culture medium was removed and 25 mL of medium combined with 10 mL virus was added and incubated at RT for 20 min, followed by the addition of another 40 mL of complete medium including 5 mM sodium butyrate to enhance protein expression.

Membrane vesicle isolation and protein analysis
Cells were harvested three days post transduction by a 5min centrifugation step at 3,000 g. Cells were resuspended in ice-cold hypotonic buffer (0.5 mM sodium phosphate, 0.1 mM EDTA, pH 7.0) containing protease inhibitors (100 mM phenylmethylsulfonyl fluoride, 5 mg/ml aprotinin, 5 mg/ml leupeptin, 1 mg/ml pepstatin and 1 mg/ml E-64) and shaken at 4°C for 30 min. This lysate was centrifuged 100,000 xg for 30 min at 4°C, after which the pellet was homogenized in ice-cold TS buffer (10 mM Tris-HEPES and 250 mM sucrose, pH 7.4) supplemented with protease inhibitors described before using a tight-fitting Dounce homogenizer for 25 strokes. Two subsequent centrifugation steps at 4°C of firstly 20 min at 4,000 g followed by supernatant centrifugation for 60 min at 100,000 g ensured harvesting of the membrane fraction. The pellet was resuspended in ice-cold protease free TS buffer and passed 25 times through a 27-gauge needle to enhance membrane vesicle formation. Protein concentration in these vesicles was determined using the Bio-Rad protein assay, vesicles were flash-frozen in N 2 and stored at −80°C.

Vesicular transport assays
A rapid filtration technique that has been described earlier was applied to evaluate uptake of transporter specific substrates into the vesicles; NMQ for P-gp, E1S for BCRP, E 2 17βG for MRP1-4 and TCA for BSEP [20]. Briefly, 0.015-0.15 μCi of labelled substrate was combined with unlabelled substrates to a concentration of 0.1-1 μM in a 30 μL reaction mixture with 4 mM ATP, 10 mM MgCl 2 and 7.5 μg total protein membrane vesicles in TS buffer. Transport was allowed by transfer of the plates to 37°C during 1-5 min, a time-point within the linear phase of time-dependent transport, as previously determined [15][16][17][18][19]. Hereafter, the reaction was rapidly stopped by placing the plates back on ice and the addition of 150 μL ice-cold TS buffer. Samples were subsequently transferred to a 96-well filter plate that had been pre-incubated with TS buffer, and filtered using a multiscreen HTS-vacuum manifold filtration device (Millipore). Filters were washed and extracted, after which 2 mL scintillation fluid was added to each filter. Radioactive signal on the filters was determined by liquid scintillation counting. Negative controls included eYFP-transduced vesicles and AMP instead of ATP in the reaction mixture.
In the first screen, all anti-malarial compounds were added to the reaction mixture to evaluate transport inhibition at a concentration of 100 μM. Solvents were used as negative controls, as CQ was dissolved in milliQ, Q and ART in methanol, MQ, L, ATO and DHA in DMSO and PG in 50% ethanol. When ATP-dependent uptake was reduced more than 66.7%, the compound was considered a potential inhibitor, and multiple concentrations were tested in the reaction mixture to determine the IC 50 value. All concentrations were tested in duplicates or triplicates in two individual biological replicates containing vesicles of independent transductions. Results were depicted and statistically analysed using Graphpad Prism, version 5.03. IC 50 values were determined by nonlinear regression analysis of (log) inhibitor-response curves with variable slope. Maximal transport was restricted to 100%, and the minimum was set to be equal or greater than 0%. Statistical analysis was performed using IBM SPSS Statistics 20, applying one-way ANOVA (Analysis of variance).

Determination of inhibitory potency of strong inhibitors
Subsequently, transport inhibition assays were performed for a larger concentration range of Q, MQ, ATO and PG to evaluate P-gp or BCRP activity. Inhibition of transport was measured in a similar fashion applying the same specific radio-labelled substrates. Drug concentrations were logarithmically depicted, and a sigmoidal, inhibitorresponse, variable slope equation was fitted to the data to determine the inhibition curve. Maximal inhibition to 0% transport was not always reached, which might be due to endogenous transport present in the vesicular membranes.
The strongest inhibitory effect for ATO on BCRPmediated transport was found at median nanomolar range. Transport of estrone sulphate was inhibited with 50% by this compound at 0.23 μM (95% CI 0.17-0.29 μM) (Figure 2A), whereas MQ and PG required the addition of 18 μM (95% CI 17-20 μM) ( Figure 2B) and 118 μM (95% CI 93-148 μM) ( Figure 2C) to achieve a similar effect on BCRP activity, respectively. Also for the other compound-transporter combinations, IC 50 values were found in the low to median micromolar range. The effect of Q on P-gp-mediated NMQ transport inhibition was the strongest, and the IC 50 was defined at 6.8 μM (95% CI 5.9-7.8 μM) ( Figure 2D). MQ was a less potent inhibitor with an IC 50 of 72 μM (95% CI 49-104 μM) ( Figure 2E). The inhibitory concentration of ATO and Q transport were within the therapeutic range of blood plasma concentrations after both prophylactic and curative anti-malarial dosing.

Discussion
In this study, the interaction of anti-malarial compounds CQ, Q, ART, MQ, L, ATO, DHA and PG with the activity of P-gp, BCRP, MRP1-4 and BSEP ABC transport proteins were investigated. ATO was found to be a strong inhibitor of BCRP-mediated transport, which has not been described Figure 1 Inhibitory effect of anti-malarial drugs on ABC transport activity. The inhibitory effect of 100 μM of CQ, Q, ART, MQ, L, ATO, DHA and PG on ABC transporter activity was assessed. Transport was measured in pmol/mg protein/min and expressed as percentage of solvent controls, which represent 100% transport. Bars with * are significantly different from solvent controls, p < 0.05. A P-gp-mediated transport of NMQ was significantly inhibited by CQ, Q, MQ and PG, and increased by ART and DHA. B BCRP-mediated transport of E1S was significantly inhibited by all compounds, most pronounced inhibitors were MQ and ATO. C BSEP-mediated transport of TCA was significantly inhibited by ATO, but not by the other anti-malarials. Induction of transport was observed for CQ, ART and DHA. D-G MRP1-4-mediated E217βG transport. MQ significantly inhibited MRP1 and MRP3 transport activity. Furthermore, induction of MRP1 mediated transport was found for ATO, which, together with ART, also stimulated MRP2 transport activity. MRP3 mediated transport was stimulated by both ART and DHA. Inhibition larger than 66.7% was found for Q and MQ on P-gp transport, as well as MQ, ATO and PG on BCRP transport activity (highlighted bars).
previously. Also Q was identified as a potent inhibitor of P-gp-mediated transport. In addition, subtle alterations on transporter activity have also been identified for other compound-transporter combinations, both inhibitory and stimulating. These interactions can be either competitive or non-competitive. Allosteric interactions that stimulate transport have been observed for several ABC transporters and are substrate dependent, due to which translation of these results to other transporter-substrate combinations is difficult.
A 50% inhibition of BCRP-mediated transport activity could be achieved with 0.23 μM ATO. This concentration is easily reached in blood plasma during both prophylactic and therapeutic use of ATO, as maximal ATO blood plasma concentrations are around 14 μM (range 8-26 μM) after a daily prophylactic dose of 250 mg [21]. Although the free concentration of ATO is reduced due to its high plasma protein binding, intracellular concentrations at the target site may be higher. ATO is used in a fixed combination with PG in Malarone®, which is prophylactively prescribed to travellers, and at higher dosages to treat falciparum malaria, especially in regions of ACT failure [2,22]. BCRP is located primarily on the apical side of excretory organs, and it is highly involved in excretion of xenobiotics from the body [23,24]. As ATO is excreted into bile against steep concentration gradients, involvement of ABC transport proteins such as BCRP is likely [25] and interactions with ATO can occur when elimination of co-administered therapeutics is inhibited.
Indeed, cases of interactions with ATO have been reported. The azithromycin AUC (area under the curve) and maximal concentrations were lower in all patients when taken in combination with ATO by HIV-1 positive children [26]. Although direct interaction of azithromycin with BCRP has not been investigated, interaction at this level cannot be excluded. Moreover, a clear increase in plasma concentration of etravirine, a reverse transcriptase inhibitor, and saquinavir, a protease inhibitor, was observed in a Caucasian female who started malaria prophylaxis with ATO/PG (250/100 mg) fixed dose combination [27]. Etravirine and saquinavir were prescribed to treat HIV1 subtype B in an antiretroviral combination therapy, supplemented with raltegravir and maraviroc. AUCs, during a 12-hour measurement interval, were increased 55% for etravirine and 274% for saquinavir, and peak concentrations after administration of the antiretrovirals was markedly increased. Saquinavir and etravirine have previously been described as potent BCRP inhibitors, but not substrates, with IC 50 concentrations of 19.5 and 1.0 μM [28,29]. Both raltegravir and maraviroc do not inhibit BCRP, indicating that interaction with BCRP is specific for saquinavir and etravirine [30]. An alternative or complementary explanation could be interaction at the level of Cytochrome P450 (CYP) enzymes, as PG is mainly metabolized by CYP2C19 but also partly by CYP3A4, saquinavir by CYP3A4 and etravirine mainly by CYP3A4 and to a minor extent by CYP2C9 and CYP2C19 [31][32][33]. Raltegravir is not metabolized by members of the CYP family, however, maraviroc is a substrate of CYP3A4 [34,35]. A strong correlation at this level of drug interaction could therefore not be observed, stressing the plausible role of transporter-mediated drug interactions.
Another study demonstrated a significant decrease in ATO plasma concentration when taken in combination with efavirenz, lopinavir/ritonavir or atazanavir/ritonavir therapy [36]. Interaction at the level of metabolism through glucuronidation was proposed. However, as ATO is only marginally glucuronidated but mostly excreted unchanged into the bile, interaction at the level of ABC transport proteins and more specifically BCRP could play an important role [25]. Indeed, efavirenz, lopinavir and atazinavir have been described as inhibitors of BCRP-mediated transport [37]. Lopinavir and efavirenz were found to be stronger inhibitors, and correspondingly, ATO concentration was decreased more drastically in these two combinations compared to atazinavir co-administration.
A 50% inhibition of P-gp-mediated transport by Q was found at a concentration of 6.8 μM. Indeed, in other in xvitro cellular uptake experiments Q has been described to be both an inhibitor and a substrate of P-gp [4,8,9,[45][46][47][48]. The concentration at which Q was effective was lower in the current study than previously described. Most likely this can be attributed to the difference in substrates used. Maximal plasma concentrations reach 30 μM during a seven-day regimen of 10 mg/kg oral dose three times daily of quinine sulphate, and although Q is bound to plasma-proteins to some extent, clinically relevant interactions at the level of P-gp-mediated transport during quinine treatment may be expected [49].
Interactions with Q have been described for ritonavir/ lopinavir combination therapy as well as ritonavir monotherapy, and for nevirapine, rifampicin, cyclosporine, and digoxin. Q co-administration with digoxin decreased billary excretion of the latter, indicating specific involvement of transport processes [50]. When co-administered with ritonavir, Q blood plasma concentrations were increased [51]. Ritonavir indeed is both a substrate and inhibitor of P-gp, therefore interaction at this level may explain the increase in Q concentration [52,53]. After rifampicin, nevirapine and lopinavir co-administration, Q blood plasma concentrations were decreased [49,[54][55][56]. Rifampicin interacts with P-gp as substrate, inhibitor and inducer, and lopinavir has been found to inhibit P-gp [57][58][59]. However, this has not been shown for nevirapine. Q is one of the oldest anti-malarial drugs still in use, and although it is not used any more in first-line treatment strategy, its use has increased as it is often applied as an alternative treatment after ACT stock-outs [60]. Furthermore, for treatment of malaria infections in pregnant women it is one of the few compounds that can be applied safely [61]. Adherence to this compound is known to be low due to the large range of common and often plasma concentration-dependent side effects [62]. For these reasons, establishing effective but non-toxic blood plasma concentrations is essential in the treatment of malaria, and interaction with co-administered compounds that mediate P-gp transport should be tightly monitored.
Especially, the interaction of both ATO and Q with antiretroviral medication could have severe implications on treatment strategies for both infections, as HIV is another major contributing factor to morbidity, especially in sub-Saharan regions of Africa [14]. Many different antiretroviral compounds are being prescribed, depending on personal characteristics and resistance status, and for many of these compounds interactions with BCRP have been described.

Conclusions
Anti-malarial compounds can reduce ABC transporter activity. ATO appeared to be a potent inhibitor of BCRP and Q of P-gp in vitro. Both compounds inhibited ABC transporter activity at concentrations equalling prophylactic and effective blood plasma concentrations. Potential involvement in interactions with antiretroviral and antibiotic compounds have been described for ATO and Q, which can be explained by the observed inhibitory effects on BCRP and P-gp transport activity.

Competing interests
The authors declare that they have no competing interests.
Authors' contributions SR contributed to the design of the study, the performance of the vesicle transport assays, the analysis of the data and drafted the manuscript. JH executed the vesicle transport assays and helped to draft the manuscript. MV participated in the transport studies and data analysis. RS contributed to the design of the study and critical revision of the manuscript. FR advised in the design of the study and adjusted the manuscript. JK conceived of the study, coordinated its execution and helped in the draft of the manuscript. All authors read and approved the final manuscript.