Role of P Glycoprotein in Absorption of Novel Antimalarial Drugs

Abstract

Bidirectional transport of four novel antimalarial compounds was determined using Caco-2 cell monolayers. P glycoprotein-mediated efflux was greatest for pyronaridine (5 to 20 μM) and low for naphthoquine (5 μM). With 20 μM naphthoquine, net efflux was blocked, suggesting saturation of the transporter. Piperaquine and dihydroartemisinin were not transported by the system.

Permeability glycoprotein (P-gp) ATP-dependent transporters influence the passage of many drugs across epithelial barriers in the intestine, brain, liver, and kidney (1) and may alter their pharmacokinetic and/or pharmacodynamic properties. Several antimalarial drugs are substrates and/or inhibitors of P-gp (9, 10, 15). In the intestine, P-gp has a secretory function which can contribute to low bioavailability and substrate-inhibitor interactions (14). The antimalarials dihydroartemisinin, naphthoquine, piperaquine, and pyronaridine are highly lipid soluble, with logP (log octanol-water partition coefficient) values of 2.6, 6.16, 6.16, and 4.98, respectively (ACD I Lab Software version 8.14 for Solaris; ACD Labs, Toronto, Canada). Evidence suggests that dihydroartemisinin and piperaquine have oral bioavailabilities of <50% (6, 7), while pyronaridine may interfere with P-gp-mediated transport (10, 11). We therefore investigated whether these four drugs are P-gp substrates and whether P-gp-mediated efflux explains the apparently low bioavailabilities of dihydroartemisinin and piperaquine.

Drug transport was studied with an in vitro gastrointestinal model by using a monolayer of a CLEFF9 subclone of human Caco-2 cells with high P-gp-mediated efflux (4) and previously reported experimental protocols (3-5). Briefly, cells were seeded onto 0.6-cm² polycarbonate filters in 24-well plates and grown in high-glucose Dulbecco's modified Eagle's medium for 3 weeks. Following incubation in buffered Hanks balanced salt solution (HBSS) with or without specific efflux inhibitors for 30 min at 37°C, transepithelial electrical resistance was measured and assay medium/inhibitors were placed in the receiver chambers. Antimalarial drugs were added to the donor chamber of each well. The apical and basolateral chambers received 0.3 and 0.6 ml medium, respectively. Transepithelial electrical resistance measurements were repeated at the conclusion of the studies to ensure continued monolayer integrity.

Naphthoquine was assayed using high-performance liquid chromatography (HPLC) with UV detection at 260 nm after separation through an Eclipse XDB-C₈ column (Agilent Technologies, Forest Hill, Australia) with a mobile phase of 22% acetonitrile, 10% methanol, and 68% 20 mM KH₂PO₄, with 11 mM triethylamine and 0.1% trifluoroacetic acid pumped at 1.2 ml/min. The limit of detection (LOD) was 30 nM. Pyronaridine was assayed by HPLC with detection at 277 nm after separation through a 100- by 4.6-mm (inside diameter), 3.5-μm XTerra C₁₈ column (Waters Associates, Milford, MA), with acetonitrile buffer (80 mM KH₂PO₄, 22 mM triethylamine HCl, and 0.1% trifluoroacetic acid, pH 2.95) mixture (18:82) pumped at 1.1 ml/min (LOD, 60 nM). Piperaquine was assayed by HPLC with detection at 345 nm using the XTerra C₁₈ column and a modified acetonitrile buffer mixture (8:92) (12) (LOD, 120 nM). A dual-label β counter was used to detect [³H]dihydroartemisinin and [¹⁴C]mannitol, the latter added during transport studies to monitor monolayer integrity.

Transport data are summarized in Table 1. Naphthoquine (5 μM; pH 7.4 on both sides of monolayer) had a moderate 1.7-fold basolateral-to-apical efflux (P < 0.01). At 4 μM each, the potent P-gp inhibitors PSC-833 and GF120918 abolished the net efflux gradient, suggesting a role for P-gp in naphthoquine efflux. At 20 μM, transport stabilized at 25 × 10⁻⁶ cm/s in both directions (P > 0.08), suggesting saturation. Concentrations of <5 μM were not tested due to the low naphthoquine LOD. A pH gradient was applied across the monolayer to simulate P-gp uptake in the upper gastrointestinal tract (10 mM 2-morpholinoethanesulfonic acid-buffered HBSS, pH 6 in apical chambers, and 10 mM HEPES-buffered HBSS, pH 7.4 in basolateral chambers). Naphthoquine efflux increased 83-fold over that in the uptake direction. P-gp inhibitors only partially reduced the efflux gradient (to 30- to 40-fold), indicating that ionization factors and/or H⁺ antiporter activity underlies rapid efflux with a pH 6.0/7.4 differential. As the predicted pK_a for naphthoquine is 8.02 (ACD I Lab software), pH gradient differences may reflect 88% and 99.5% ionization of the drug at pH 7.4 and 6.0, respectively. The logD (log octanol-water partition coefficient for a given pH) for naphthoquine changes from 3.1 to 4.0 between pH 6.0 and 7.4, which should not significantly alter lipid bilayer permeability. These differences in ionization and logD are unlikely to influence the efflux gradient.

TABLE 1.

Transport rates for 5 and 20 μM antimalarial drugs through Caco-2 subclone (CLEFF9) monolayers in both apical-to-basolateral and basolateral-to-apical directions

Antimalarial drug	pH in Ap/Bas chambers	Concn (μM) (efflux modifier)	Transport rate (cm/s) (10−6)a		Fold difference (net flow direction)
			Ap→Bas	Bas→Ap
Naphthoquine	7.4/7.4	5	13.3 ± 0.6	22.6 ± 1.9b	1.7 (efflux)
	7.4/7.4	5 (4 μM PSC-833)	22.3 ± 1.3c	22.2 ± 2.8	1.0 (no net flux)
	7.4/7.4	5 (4 μM GF120918)	15.1 ± 0.8c	15.7 ± 0.6c	1.0 (no net flux)
	7.4/7.4	5 (25 μM MK571)	15.6 ± 0.8c	25.7 ± 2.0b,c	1.7 (efflux)
	6.0/6.0	5	12.9 ± 0.8c	37.0 ± 1.9b,c	2.9 (efflux)
	6.0/7.4	5	1.8 ± 0.1	151.5 ± 16.8b,c	82.8 (efflux)
	6.0/7.4	5 (4 μM PSC-833)	3.1 ± 0.1c	94.8 ± 4.8b,c	30.8 (efflux)
	6.0/7.4	5 (4 μM GF120918)	2.2 ± 0.1	81.8 ± 4.8b,c	38.1 (efflux)
		7.4/7.4	20	27.3 ± 0.6	22.4 ± 2.0	0.8 (no net flux)
		7.4/7.4	20 (4 μM PSC-833)	23.8 ± 0.4	18.8 ± 2.0	0.8 (no net flux)
Pyronaridine	7.4/7.4	5	4.2 ± 0.6	28.0 ± 3.4b	6.7 (efflux)
	7.4/7.4	5 (4 μM PSC-833)	1.7 ± 0.0c	12.4 ± 1.2b,c	7.3 (efflux)
	7.4/7.4	5 (4 μM GF120918)	3.7 ± 0.2	16.7 ± 0.5b,c	4.5 (efflux)
	6.0/7.4	5	3.8 ± 0.1	52.9 ± 3.0b	13.9 (efflux)
	6.0/7.4	5 (4 μM GF120918)	4.0 ± 0.1	26.7 ± 0.5b,c	6.7 (efflux)
	7.4/7.4	20	10.3 ± 0.3	18.8 ± 0.3b	1.8 (efflux)
	7.4/7.4	20 (4 μM PSC-833)	15.2 ± 0.3	16.3 ± 1.1	1.1 (no net flux)
	7.4/7.4	20 (4 μM GF120918)	11.5 ± 0.6	13.2 ± 0.7	1.1 (no net flux)
		7.4/7.4	20 (500 μM probenecid)	12.8 ± 0.6	22.4 ± 0.5b	1.8 (efflux)
		6.0/7.4	20	1.7 ± 0.0	32.1 ± 2.0b	19.0 (efflux)
Piperaquine	7.4/7.4	20	0.0 ± 0.6	0.1 ± 0.1b	NAd (detection too low)
	7.4/7.4	20 (4 μM PSC-833)	0.9 ± 0.1	0.4 ± 0.0b	0.4 (uptake)
Dihydroartemisinin	7.4/7.4	1	32.0 ± 2.6	21.9 ± 1.7b	0.7 (uptake)
	7.4/7.4	1 (4 μM PSC-833)	30.8 ± 1.5	21.1 ± 2.4b	0.7 (uptake)
	7.4/7.4	20	38.6 ± 5.0	29.9 ± 0.6	0.8 (no net flux)

^aEach value represents the mean ± standard error of the mean for triplicate independent determinations. Ap, apical; Bas, basolateral.

^bSignificant difference between basolateral-to-apical and apical-to-basolateral transport for the particular set of drug transport conditions (P < 0.05).

^cSignificant difference between drug transport in a particular direction for a given concentration and the equivalent transport with the addition of an efflux pump inhibitor (P < 0.05).

^dNA, not applicable.

Pyronaridine showed greater flux with an apical-basolateral pH gradient of 6.0/7.4. The calculated pK_a of pyronaridine is 10.22, with interpolated logD values of 0.6 and 1.7 at pH 6.0 and 7.4, respectively (ACD I Lab Software), suggesting that changes in membrane partitioning could be structurally related when the pH gradient conditions are applied. Pyronaridine has been shown to inhibit P-gp-mediated efflux (10, 11) but was not thought to be transported itself (10). Our data suggest that pyronaridine is a weak P-gp substrate, as the 1.8-fold efflux gradient for 20 μM concentrations (P < 0.001) increased to 6.7-fold at 5 μM (P < 0.003). Moreover, 20 μM concentrations completely abolished efflux during coincubation with the P-gp blocker PSC-833 or GF120918 (Fig. 1 and Table 1).

FIG. 1.

Bidirectional transport of 20 μM pyronaridine through the Caco-2 CLEFF9 subclone. Shown are the apical-to-basolateral direction (□ and ▪) and basolateral-to-apical direction (⋄ and ⧫), with (▪ and ⧫) and without (□ and ⋄) the presence of 4 μM PSC-833, a potent P glycoprotein inhibitor, on both sides of the membrane.

Neither piperaquine nor dihydroartemisinin exhibited P-gp-mediated transport. While chloroquine, mefloquine, quinine, and pyronaridine are weak P-gp substrates, they inhibit the efflux of other P-gp substrates (8, 9, 11, 14). Although our experiments did not assess whether the present antimalarial drugs behave in the same way, our results suggest limited P-gp-mediated pyronaridine and naphthoquine efflux at low micromolar concentrations. As P-gp-mediated efflux appeared to increase as the drug concentration was lowered, organs, such as the brain, with extensive P-gp expression might not accumulate these drugs, as the greatest plasma concentrations after single dosing rarely exceed 0.3 to 0.4 μM for either napthoquine (13) or pyronaridine (2). Since oral doses of each drug are 400 to 600 mg, concentrations at the gastrointestinal wall are likely to be at least in the mid-micromolar range, where initial absorption would occur based on the physicochemical properties of the drugs and the fact that P-gp-mediated efflux would be saturated. As the luminal concentration falls to the low-micromolar range, P-gp-mediated efflux could attenuate the continued oral absorption of these compounds, especially pyronaridine.

In summary, it is unlikely that P-gp has a significant role in the gastrointestinal absorption of piperaquine and dihydroartemisinin, while the effect of P-gp on the absorption of naphthoquine will be limited due to saturation at low doses. Pyronaridine may exhibit variable absorption on the basis of significant P-gp-mediated efflux. It is interesting that oral piperaquine has proved effective in clinical trials (6) regardless of our inability to detect significant transport in our in vitro model. Lack of transport suggests that a drug should not be given orally. Future piperaquine cell transport studies should address this apparent inconsistency.