Transport of Artemisinin and Sodium Artesunate in Caco-2 Intestinal EpithelialCells

P. AUGUSTIJNSX, A. D’HULST, J. VAN DAELE, AND R. KINGETReceived January 2, 1996, from the Laboratory of Galenical and Clinical Pharmacy, Catholic University Leuven, Onderwijs en Navorsing,Gasthuisberg, 3000 Leuven, Belgium. Final revised manuscript received February 26, 1996. Accepted forpublication February 29, 1996X.

Abstract 0 Artemisinin and its derivatives are becoming interestingalternatives to the commonly used antimalarial drugs because they areefficient in treating severe and multidrug resistant forms of Plasmodiumfalciparum malaria. A major drawback is the occurrence of recrudescencesome time after treatment. Moderate oral bioavailability has beensuggested as a possible cause. As one of the factors that might limitabsorption after oral administration, we studied the intestinal permeabilityusing an in vitro system of the intestinal mucosa, Caco-2. Concentrationsof artemisinin were determined by UV after alkaline degradation, whilefor sodium artesunate, a capillary electrophoresis method was developed.Artemisinin easily crossed the epithelial cells by passive diffusion (Papp)30.4 ± 1.7 × 10-6 cm s-1, pH 7.4). Permeability of the hemisuccinateanalogue, sodium artesunate, was 8-fold lower (Papp ) 4.0 ± 0.4 × 10-6

cm s-1 at pH 7.4) and strongly dependent on pH, which might result insite dependent resorption in an in vivo situation. Enzyme catalyzed esterhydrolysis of sodium artesunate in Caco-2 monolayers to the biologicallyactive metabolite, dihydroartemisinin, was moderate. The results indicatethat the transepithelial permeability is probably not a limiting factor in theoverall absorption process after oral administration of artemisinin or sodiumartesunate. Solubility, dissolution rate, stability, and first-pass metabolismare suggested as alternative limiting factors.

Introduction

Artemisinin and its analogues have gained increasingattention because of their low toxicity and efficacy againstchloroquine resistant strains of Plasmodium falciparum.They are especially efficient in the treatment of cerebralmalaria.1 The major drawback of these compounds is theoccurrence of a high recrudescence rate after treatment.2 Thereason for this is not known, but the observation thatrecrudescent infections were more common with tablets (totaldose 3 g) than with parenteral formulations (total dose 0.5-1.2 g)3 suggests that a limited oral bioavailability may be oneof the contributing factors. As the oral dosage of antimalarialdrugs offers practical and financial advantages over parenter-al administration, especially in malaria endemic areas devoidof clinical centers, information on all factors that may affectabsorption would be helpful to establish improved oral dosageschemes and possibly reduce the prevalence of recrudescence.The development of dosage schemes or dosage forms thatguarantee sustained blood concentrations is required becauseit is generally accepted that a sufficiently high concentrationover a sufficiently long period of time is required to obtain aparasiticidal action. Although first-pass metabolism has beensuggested to be an important factor accounting for low oralbioavailability,4,5 other phenomena may be invoked as alter-native explanations, such as a low intestinal permeability, alow solubility, and a low dissolution rate. Sodium artesunatewas originally developed as a water soluble analogue in order

to facilitate parenteral administration. It is a soluble hemi-succinate derivative, which, in the blood, is rapidly hydrolyzedto dihydroartemisinin, the active metabolite.6 Sodium arte-sunate is also developed into a tablet for oral administration.It is believed that, after oral administration, the higherprodrug solubility will increase the concentration gradient,resulting in enhanced transport yielding more active com-pound after hydrolysis in the blood. It is, however, not knownwhether the polarity of this prodrug limits its transepithelialpermeability. Pharmacokinetic data of sodium artesunate arescarce, and data on oral absorption are, to our knowledge,missing. As the understanding of all the factors that possiblylimit oral bioavailability may help to develop artemisinin andits analogues into alternatives to the commonly used anti-malarial drugs, we investigated the transepithelial perme-ability of artemisinin and sodium artesunate.7,8 An in vitrocell culture system that is generally accepted to mimic theintestinal lining (Caco-2) was used. In addition, the octanol-water partition coefficient of both compounds and the pKavalue of sodium artesunate were determined in order to betterunderstand the effect of some basic physicochemical charac-teristics on their permeability across Caco-2 mono-layers.

Materials and MethodsMaterialssArtemisinin and artesunic acid were kindly provided

by Profarma (Geel, Belgium) and Mediplantex (Hanoi, Vietnam).Hank’s balanced salt solution, Dulbecco’s modified eagle medium(DMEM) containing glutaMAX, N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid (Hepes), 0.05% trypsin and 0.02% EDTA in PBS,nonessential amino acids (NEAA), penicillin-streptomycin (10 000IU/mL-10 000 µg/mL), and fetal bovine serum were from Gibco BRL(N.V. Life Technologies, Paisley, Scotland). MeOH and NaH2PO4were from BDH (Poole, England) and UCB (Leuven, Belgium),respectively. All chemicals were used as received.Cell CulturesThe Caco-2 cell line was generously provided by

Dr. Y. Schneider (Laboratory of Cellular Biochemistry, UniversiteCatholique de Louvain, Belgium). Cells were used between passages110 and 130. The Caco-2 cells were maintained in high-glucose (4.5g/L) DMEM containing glutaMAX, 100 units/mL of penicillin, 100 µg/mL of streptomycin, 1% NEAA, and 10% fetal bovine serum andgrown in tissue culture flasks (75 cm2, Nunc, Roskilde, Denmark) at37 °C in 5% CO2. They were trypsinized at a ratio of 1:5 after reaching90% of confluence using 0.05% trypsin in PBS containing 0.02%EDTA. For transport experiments, cells were seeded in tissue cultureinserts (Anopore membranes, Nunc, 25 mm).The medium was changed the day after seeding and every other

day thereafter. Apical (AP) and basolateral (BL) chamber volumeswere maintained at 1.5 and 2 mL, respectively.Experiments were conducted 17-22 days postseeding.Transepithelial electrical resistance (TEER) was measured using

an EVOM voltohmmeter (WPI, Aston, England). Only monolayershaving TEER values above 250 Ω cm2 were used in studies. Sodiumfluorescein was used as a hydrophilic marker for cell monolayerintegrity.9 Typical sodium fluorescein flux values across Caco-2monolayers after the transport experiments with test compound werebelow 0.5% h-1.Transepithelial TransportsFor the determination of the trans-

epithelial flux of artemisinin and sodium artesunate across Caco-2X Abstract published in Advance ACS Abstracts, April 1, 1996.

© 1996, American Chemical Society and S0022-3549(96)00001-9 CCC: $12.00 Journal of Pharmaceutical Sciences / 577American Pharmaceutical Association Vol. 85, No. 6, June 1996

+ +

cells, the polarized monolayers were preincubated with transportmedium (Hank’s balanced salt solution, supplemented with 10 mMHepes and 25 mM glucose) for 30 min, after which TEER values weremeasured to check monolayer integrity. The transport medium wasthen replaced by transport medium with test compound at the donorside (1.5 mL if apical, 2 mL if basolateral). To maintain sinkconditions, the inserts were transferred to fresh 6-well plates after15, 30, 45, and 60 min for artemisinin, while for sodium artesunate,samples were taken from the apical (10 µL) and basolateral sides (100µL), and the volume withdrawn was replaced with fresh transportmedium, for which a correction was made in further calculations. Allflux experiments were conducted in triplicate. The transport wasexpressed as the percentage of drug added to the donor compartmentor calculated as apparent permeability coefficient (Papp) according tothe equation

with ∆Q/∆t the permeability rate (µg/min), C0 the initial concentration(µg/mL), and A the surface area of the monolayer (4.9 cm2).8Artemisinin AssaysConcentrations of artemisinin were meas-

ured according to a slight modification of the method of Zhao.10 Inshort, 100 µL samples were alkalinized with 200 µL of 0.4% NaOH.After 5 min at 50 °C, samples were acidified with 50 µL of 1 M aceticacid (in 50:50 MeOH/H2O). The resulting degradation product showeda UV maximum at 260 nm and could be analyzed using HPLC. TheHPLC system consisted of an M-6000A pump, a UV detector Model440, a Model U6K injector, and a Novapak C18 column (8 mm Radial-Pak cartridge, 4 µm packing, Waters). The mobile phase consistedof 10 mM phosphate pH 4.5 and MeOH (50:50); the flow rate was 1.5mL/min. The retention time was 8.5 min. The intraday and interdayprecisions for concentrations ranging from 0.31 to 10 µM were lowerthan 8% RSD.Artesunate AssaysSamples of sodium artesunate were directly

analyzed using capillary electrophoresis with UV detection at 185 nm.All separations were performed in a fused silica capillary (φ ) 50 µm,l ) 49 cm) using a Quanta 4000 CE system (Waters, Milford, MA).Samples were injected by gravity-induced siphoning and run underconstant high voltage (30 kV) conditions. CE running electrolyteconsisting of 40 mM lithium phosphate (pH 6.8) was freshly preparedin Milli-Q water, filtered and degassed immediately prior to use.Artesunate (t ) 1.8 min) was well separated from its degradationproducts, i.e., dihydroartemisinin, which comigrated with the electro-osmotic flow (t ) 1.5 min), and succinic acid (t ) 5.6 min). Interdayas well as intraday RSD values were lower than 5% (25-333 µMartesunate).Octanol-Water Partition CoefficientsArtemisinin and sodium

artesunate were dissolved in transport medium (pH 7.4) to give finalconcentrations of 100 µM and 10 mM, respectively; 3 mL of theaqueous solution and 3 mL of 1-octanol were shaken for 2 h at 37 °C.After separation of the two phases by centrifugation, the drugconcentration was determined in the aqueous layer, and the concen-tration in the octanol phase was calculated from the initial and finalconcentrations in the aqueous phase. Transport medium and octanolwere mutually saturated before the experiment.Acid Dissociation Constant of Artesunic AcidsThe acid

dissociation constant of artesunic acid was determined by potentio-metric titration using an automatic titrator (Metrohm 655 Dosimat).Artesunic acid was dissolved in water to give a final concentration of0.25 mM and was titrated with 0.005 N NaOH.

Results and DiscussionAs a direct measure of intestinal transepithelial perme-

ability, transport of artemisinin and sodium artesunate(Figure 1) was studied in an in vitro cell culture system ofthe intestinal mucosa (Caco-2). The Caco-2 cell culture systemis commonly used to study the transepithelial permeabilityof nutrients and drugs, and the cells express most of theenzymes and carrier systems present in the intestinal mu-cosa.7,8 Transport of artemisinin and sodium artesunate wasstudied at concentrations of 100 µM and 10 mM, respectively.Concentrations were chosen on the basis of their solubility

and the detection limit of the analytical method. Transportof artemisinin and sodium artesunate across Caco-2 mono-layers was linear with time when incubated for 60 min at 37°C. At pH 7.4, the permeability of artemisinin was 8-foldhigher than that of sodium artesunate (Table 1), which wasexpected because of ionization of sodium artesunate. The acid-dissociation constant of artesunic acid determined by poten-tiometric titration was found to be 4.6, which is 1 pH unithigher than the value reported in the literature.11 It isobvious that artesunic acid is almost completely ionized at apH of 7.4 (99.8%) and that this might limit the transepithelialpermeability. In contrast to artemisinin, which does not haveany ionizable group, a low permeability is also expected forsodium artesunate in view of the results of the octanol-waterpartitioning (Table 2). Due to the experimental conditions(pH 7.4), the transport of sodium artesunate may be under-estimating the in vivo situation; indeed, at pH 6.6 ( 0.5, theestimated pH of the intestinal lumen,12 the ionized fractionwill be considerably reduced, enhancing diffusion throughmembranes. When the transport of sodium artesunate (10mM) was studied for 2 h at pH 6.0 and 7.4, it proved to betoxic to the cells at pH 6.0, based on the observation thatTEER values dropped to 176 ( 123 Ω cm2 compared to 473 (37 Ω cm2 for the pH 7.4 condition. When sodium artesunatetransport was monitored for up to 3 h at a concentration of 1mM, TEER values of both pH conditions remained the same

Papp ) ∆Q∆t(60)AC0

(cm‚s-1)Figure 1sStructure of artemisinin (R ) dO), dihydroartemisinin (R ) OH), andartesunic acid (R ) OCCH2CH2CO2H).

Table 1sEffect of Various Conditions on the Permeability of SodiumArtesunate and Artemisinin across Caco-2 Monolayers a

Conditions T (°C)Papp (×106)(cm s-1) SD

(a) Artemisinin (100 µM, 1 h)Apical−basolateral direction 37 30.4 1.7

20 22.5 0.52 20.0 0.3

Basolateral−apical direction 37 30.9 2.6Sodium azide (1 mM) 37 30.8 1.9

(b) Artesunate (10 mM, 1 h)pH 7.4 37 4.0 0.4

a Three monolayers were used in each experiment. They were incubated withsodium artesunate alone, artemisinin alone, or artemisinin in the presence of 1mM sodium azide.

Table 2sPhysicochemical Properties of Artemisinin and SodiumArtesunate

DrugAqueousSolubility pKa

Octanol−TransportMedium Partition Coeff

Artemisinin 113 µg/mLa 160Sodium artesunate >200 mg/mL 4.6 0.31

a Reference 17.

578 / Journal of Pharmaceutical SciencesVol. 85, No. 6, June 1996

+ +

after 3 h (501 ( 28 and 454 ( 51 Ω cm2 for pH 7.4 and 6.0,respectively); a 4-fold increase in transport was observed atpH 6.0 compared to pH 7.4 (46 ( 4% versus 10 ( 2%), clearlyillustrating the pH dependent transmembrane transport ofsodium artesunate. On the basis of the pH and pKa values ofsodium artesunate, the permeability is expected to be en-hanced by a factor of 24 if only the undissociated moleculecrosses the membranes. However, due to the limited sensitiv-ity of the detection method, sink conditions could not bemaintained in this experiment (1 mM added to the apicalside), and some back diffusion may have occurred, offsettingthe expected permeability increase. As the transport ofsodium fluorescein, a hydrophilic marker substance alsohaving one negative charge and a similar molecular weight(376 compared to 384 for sodium artesunate), was lower than0.5% h-1, sodium artesunate (4.73% h-1) is believed to crossthe monolayers predominantly by transcellular transport.The transport of artemisinin showed a slight decrease with

decreased temperature (Table 1). This decrease was muchsmaller than the decrease observed for other molecules whichare transported by an active mechanism, such as for examplebile acids.13 Therefore, the limited temperature dependenceof transport was not considered indicative of an energy-requiring transport mechanism. This was confirmed byrepetition of the experiments in the presence of the metabo-lism inhibitor sodium azide: no decrease of transport couldbe detected (Table 1). In addition, no polarity in artemisininflux was observed (Table 1), corroborating the thesis thatartemisinin crosses biological membranes by passive diffusion.Sodium artesunate has been described to act as a prodrug,

probably not reaching the blood in its intact form due tohydrolysis into dihydroartemisinin and succinic acid.14 How-ever, no major enzymatic degradation was observed in thepresent study, with a total (apical + basolateral) recovery of91% after 1 h (10 mM, pH 7.4). The observation of sodiumartesunate appearing unchanged at the basolateral side mightbe due to enzyme saturation or due to a low enzyme affinitycombined with a short residence time in the cells. Andersonet al.,15 who studied the bioconversion of the hemisuccinatederivative of methylprednisolone in plasma, indeed suggestedthat anionic carboxylate-containing prodrugs have low enzymebinding affinity, resulting in ester hydrolysis resistance.Neutral carboxylic acid ester prodrugs have been shown toundergo intracellular enzymatic metabolism in Caco-2 mono-layers,16 illustrating that this model can be used to screenester hydrolysis during transport. If extensive luminal hy-drolysis of sodium artesunate occurred, its solubility anddissolution rate advantage would be lost, as its hydrolysisproduct, dihydroartemisinin, is expected to be much lesssoluble, and would cause a much lower concentration gradient.If, however, hydrolysis occurred at the membrane, this wouldbe an advantage because the locally formed more lipophilicmetabolite would subsequently easily cross the membrane.The high permeability of artemisinin across biological

membranes confirms its usefulness in the treatment ofcerebral malaria, as good blood-brain barrier permeabilityis expected. Artesunate, on the other hand, is rapidly

hydrolyzed in the blood, resulting in the formation of dihy-droartemisinin, which is also expected to easily cross theblood-brain barrier.From the results of this study, it is concluded that arte-

misinin, which is a low-solubility drug, easily crosses theintestinal monolayers by passive diffusion. Permeability wasfound to be much lower for sodium artesunate and, in theexperimental conditions used, highly pH dependent, so its invivo absorption might be site dependent. Its high solubilityin combination with the extensive area of the intestinalmucosa might compensate for the observed low permeability.It is, however, premature to draw any conclusion on the oralabsorption, because, in an in vivo situation, drug absorptionis the result of an interplay between many factors, such asdrug solubility, dissolution rate, regional pH differences, andmembrane permeability. We are presently investigating theinfluence of solubility and dissolution rate. Luminal trans-formation of sodium artesunate into dihydroartemisinin willalso be investigated.

References and Notes1. Klayman, D. L. Science 1985, 228, 1049-1055.2. Woerdenbag, H. J.; Pras, N.; van Uden, W.; Wallaart, T. E.;

Beekman, A. C.; Lugt, C. B. Pharm. World Sci. 1994, 16 (4),169-180.

3. Hien, T. T.; White, N. J. Lancet 1993, 341, 603-608.4. Titulaer, H. A. C.; Zuidema, J.; Kager, P. A.; Wetsteyn, J. C. F.

M.; Lugt, Ch.B.; Merkus, F. W. H. M. J. Pharm. Pharmacol.1990, 42, 810-813.

5. Xinyi, N.; Liyi, H.; Zhihong, R.; Zhenyu, S. Eur. J. Drug Metab.Pharmacokinet. 1985, 10 (1), 55-59.

6. White, N. J. Trans. R. Soc. Trop. Med. Hyg. 1994, 88 (S1), 41-43

7. Hidalgo, I. J.; Raub, T. J.; Borchardt, R. T. Gastroenterology1989, 96, 736-749.

8. Artursson, P. J. Pharm. Sci. 1990, 79, 476-482.9. Walter, E.; Kissel, T. Eur. J. Pharm. Sci. 1995, 3, 215-230.10. Zhao, S. Analyst 1987, 112, 661-664.11. Edlund, P. O.; Westerlund, D.; Carlqvist, J.; Wu, B. L.; Jin, J.

H. Acta Pharm. Suec. 1984, 21, 223-234.12. Evans, D. F.; Pye, G.; Bramley, R.; Clark, A. G.; Dyson, T. J.;

Hardcastle, J. D. Gut 1988, 29, 1035-1041.13. Hidalgo, I. J.; Borchardt, R. T. Biochim. Biophys. Acta 1990,

1035, 97-103.14. White, N. J. Trans. R. Soc. Trop. Med. Hyg. 1994, 88, 3-4.15. Anderson, B. D.; Conradi, R. A.; Spilman, C. H.; Forbes, A. D.

J. Pharm. Sci. 1985, 74, 382-387.16. Hovgaard, L.; Bronsted, H.; Buur, A.; Bundgaard, H. Pharm.

Res. 1995, 12, 387-392.17. Trigg, P. I. In Economic and Medicinal Plant Research; Wagner,

H., Hikino, H., Farnsworth, N. R., Eds.; Academic Press: SanDiego, 1989; pp 20-55.

AcknowledgmentsThis study was supported by a grant from the Belgian Nationaal

Fonds voor Wetenschappelijk Onderzoek (NFWO). The technicalassistance of S. Colson and H. Herbots is greatly acknowledged. Mrs.Ngo Thu Hoa (College of Pharmacy, Hanoi, S.R.V.) is gratefullyacknowledged for providing us with artesunate.

JS960001I

Journal of Pharmaceutical Sciences / 579Vol. 85, No. 6, June 1996

+ +