Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Dihydroartemisinin (DHA¹; Fig. 1) is the lactol reduction product of artemisinin (qinghaosu) a sesquiterpene lactone endoperoxide that is the principal antimalarial constituent of Artemisia annua. DHA has potent antimalarial activity in vitro and in vivo, but its low aqueous solubility restricts its administration to the enteral routes (Barradell and Fitton, 1995). DHA is also the common metabolite of its methyl (artemether) and ethyl (arteether) ethers, and its hemisuccinate ester artesunate (ARTS) (Lee and Hufford, 1990; Chi et al., 1991; De Vries and Dien, 1996), all three of which are used in the treatment of malaria (Barradell and Fitton, 1995; De Vries and Dien, 1996). Artemether and arteether are dealkylated relatively slowly (t_1/2 = 8 and 24 h, respectively) (Lee and Hufford, 1990; Chi et al., 1991; De Vries and Dien, 1996), and antimalarial activity is most likely due to the parent compounds, rather than to DHA. However in vivo, ARTS is rapidly (t_1/2 = 2-3 min) converted to DHA, which in turn is eliminated from the systemic circulation with a t_1/2 of 40 to 50 min (Yang et al., 1985; Batty et al., 1998a,b; Zhao et al., 1988). Thus, most of the antimalarial activity resulting from ARTS administration is attributable to DHA.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Structure of A, dihydroartemisinin epimers (- or -DHA); B, its -glucuronide (-DHA-G); and C, the furano acetate derivative of DHA.

When administered intravenously to male rats, [12-¹⁴C]DHA was converted principally to the biologically inactive (Ramu and Baker, 1995) -DHA--glucuronide (-DHA-G; Fig. 1) (Maggs et al., 1997, 2000). DHA is also eliminated in bile as minor glucuronides of nonendoperoxide isomers, namely a tetrahydrofurano acetate (Fig. 1), and a 3-hydroxydesoxy rearrangement product (Maggs et al., 1997). Endoperoxides readily undergo biomimetic iron (II)-mediated isomerization as a consequence of a one-electron reduction of the peroxide bridge, which produces carbon-centerd radical intermediates (O'Neill et al., 2001a); the radical-rearrangement pathway producing furano acetate compounds in vitro has been confirmed (Butler et al., 1998). The mechanism(s) and site of isomerization in vivo remain unknown but the two rearrangement pathways for another DHA derivative, artelinic acid, are reported to be catalyzed by human liver microsomes (Idowu et al., 1997). However, Batty et al. (1998c) observed that isolated rat livers perfused with DHA in physiological buffer only very rarely excreted the furano acetate glucuronide in bile, notwithstanding the extensive biliary excretion of -DHA-G (Maggs et al., 2000). The rearrangements might occur partly in blood as artemether is actively isomerized by hemoglobin and blood in vitro (Blum et al., 1998). In the present study, we have investigated both the urinary metabolite profile of ARTS following its administration to patients with falciparum malaria as well as the in vitro metabolism of DHA by human liver microsomes and by purified UDP-glucuronosyltransferases (UGT).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. DHA was synthesized as an epimeric mixture by a published method (Lin et al., 1987) and supplied by Professor R. Haynes, University of Science and Technology, Hong Kong, People's Republic of China. [12-³H]DHA ([³H]DHA; 1.4 Ci mmol¹; Moravek Biochemicals Inc., Brea, CA) was also an epimeric mixture but was radiochromatographically homogeneous (>98.4%) by HPLC using the method employed for assaying glucuronidation of [³H]DHA in human liver microsomes. The sodium salts of -DHA -glucuronide and -DHA -glucuronide were prepared by Dr. A. V. Stachulski, Ultrafine Chemicals (Manchester, UK) (O'Neill et al., 2001b). They were dissolved in deionized water (10 mM), and isomerized to their furano acetate and 3-hydroxydesoxy-DHA forms by reaction with iron (II) chloride as described previously (Maggs et al., 2000). HPLC grade solvents were obtained from Fisher Scientific (Loughborough, UK). Adenosine 3'-phosphate 5'-phosphosulphate (PAPS), 5-andostene-3,17-diol, Brij 58, diazepam, diclofenac, ethinyloestradiol, ketoprofen, oxazepam, mycophenolic acid, 1-naphthol, naloxone, novobiocin, paracetamol, saccharolactone, and uridine diphospho--D-glucuronic acid (UDPGA; tri-sodium salt; catalogue no. U6751) were products of the Sigma-Aldrich Co., Ltd. (Gillingham, Dorset, UK). All other chemicals were of analytical reagent grade.

Urinary Metabolites of Dihydroartemisinin following Intravenous Administration of ARTS. Vietnamese adults (n = 26) with uncomplicated falciparum malaria were given ARTS (120 mg i.v. bolus). The pharmacokinetics of ARTS and DHA in the plasma of these patients have been reported previously (Batty et al., 1998b,c). Urine samples (0-2, 2-4, and 4-6 h following the dose) from a subset of 17 of these patients were stored at 80°C for 12 to 15 months prior to being analyzed in the present study. ARTS was also administered orally (120 mg) to a healthy white male volunteer aged 46 years who was not taking concurrent medications and was a nonsmoker. This study was approved by the Ethics Committee of the Mersey Regional Health Authority. Urine was collected before dosing and as 2-h fractions. Aliquots were analyzed immediately by liquid chromatography-mass spectrometry (LC-MS).

Stability of DHA-G in Human Urine. Urine from the malaria patients (200 µl from 2- to 4-h collections) was transferred to capped glass tubes and maintained at 37°C. Aliquots were taken periodically for analysis by LC-MS. In addition, bile containing -[³H]DHA-G (1.6 mM; 17.9 µCi ml¹) produced by a rat liver perfused with [³H]DHA in a single-pass system (Batty et al., 1998c; Maggs et al., 2000) was diluted with either predosing bile from an anesthetized and cannulated rat (3:67, v/v) or urine freshly collected from a nonmedicated adult male volunteer (1:39, v/v). The mixtures were incubated in capped glass tubes at 37°C, and aliquots were analyzed by LC-MS.

Analysis of DHA Urinary Metabolites. Aliquots of urine (50 µl) and bile (2-20 µl) were eluted from an Ultracarb C₈ 5-µm column (Phenomenex, Macclesfield, UK) at room temperature with a gradient of acetonitrile in 0.1 M ammonium acetate, pH 6.9: 20 to 35% over 15 min, 35 to 70% over 10 min. The flow rate was 0.9 ml min¹. Metabolites were characterized by positive-ion electrospray mass spectrometry using a Micromass Quattro II instrument (Micromass UK Ltd., Manchester, UK) as described previously (Maggs et al., 2000). Selected ion monitoring of six channels was performed with a dwell time of 200 ms and an interchannel delay of 20 ms. Data were processed via MassLynx 2.1 software (Micromass UK Ltd.). Proportions of metabolites were estimated from areas of peaks in the mass chromatograms for [M + NH₄]⁺. Radiolabeled compounds were located with a Radiomatic Flo-One model A250 flow detector (Packard Bioscience, Pangbourne, UK) connected in parallel with the mass spectrometer. The eluate was mixed with Ultima Flo AP scintillant (Packard Bioscience B.V., Groningen, The Netherlands) at 1 ml min¹.

Preparation of Human Liver Microsomes and Cytosols. Liver tissue was obtained from a human liver bank (renal transplant patients) maintained by the Department of Pharmacology and Therapeutics, University of Liverpool. Liver samples from eight patients (Table 1) were homogenized in buffer (0.05 M Tris, 1.15% KCl; pH 7.4), and microsomes were prepared by differential centrifugation and resuspended in 0.01 M phosphate buffer (pH 7.4). For three samples, the cytosolic fractions from the final microsomal centrifugation step were retained for evaluation of sulfotransferase activity. Approval for use of these human tissues was obtained from the Ethics Committee of the Mersey Regional Health Authority.

Glucuronidation of [12-³H]DHA Using Human Liver Microsomes. Preliminary experiments were performed to establish optimal reaction conditions and linearity of the assay. Microsomal protein concentration was determined colorimetrically (Bradford, 1976). For the determination of the kinetics of DHA glucuronidation without reference to the stereoselectivity of conjugation, the composition of the final reaction mixture (total volume 250 µl) was as follows: DHA (12.5-500 µM, containing 0.1 µCi [³H]DHA); UDPGA (3 mM); Brij 58 (0.1 mg mg¹ of protein), tris-hydroxymethylaminomethane (0.05 M), MgCl₂ (5 mM), human liver microsomes (0.4-0.5 mg of protein). DHA was solubilized in ethanol, such that the final concentration in the reaction mixture did not exceed 2% (v/v). Reactions were incubated in duplicate at 37°C for 60 min and terminated by addition of 170-µl ice-cold methanol and vortexing for 10 s. Blank incubations contained no UDPGA. After centrifugation (2,500g, 5 min), 100 µl of the supernatant was injected on to the HPLC column, and DHA-G was quantified as picomoles formed per minute per milligram of microsomal protein using radiochemical detection (see below). The Michaelis-Menten equation was fitted to the primary data using the nonlinear regression analysis procedure in SigmaPlot version 6 (Jandel Scientific, San Rafael, CA) to yield estimates of K_m and V_max.

Analysis of [³H]DHA-G Produced by Human Liver Microsomes. Reversed-phase chromatography was performed on a Nucleosil 5µ C₈, 25 cm × 4.6 mm i.d. column (Phenomenex) equipped with a Merck CN precolumn (BDH, Poole, Dorset, UK). The mobile phase consisted of acetonitrile: 0.1M ammonium chloride, 50:50 (v/v), pumped at a rate of 1 ml min¹. Detection was achieved on-line using a Berthold Bioanalytical Instruments BetaFlow radiochemical detection system (Wallac, Milton Keynes, Bucks, UK). Under these conditions, DHA-G and DHA chromatographed at approximate R_t of 4.1 and 8.1 min, respectively. The epimers of DHA were partially resolved, but there was no indication at this short R_t of any corresponding resolution of the product peak, the stereochemistry of which was not specified. The area under the DHA-G peak (in disintegrations per minute) was expressed as a percentage of the total radioactivity in the chromatogram (% turnover).

Glucuronidation of DHA by Recombinant UGTs. DHA was incubated with recombinant human UDP-glucuronosyltransferases expressed in V79 (Chinese hamster lung fibroblasts) cells (Ethell et al., 2001). Cells expressing UGT1A1, UGT1A6, UGT1A9, and UGT2B7 were resuspended in phosphate-buffered saline, pH 7.4, and were then disrupted by sonication (MSE Soniprep150; Sanyo Gallenkamp, Loughborough, UK): 4 × 5-s bursts allowing a minute on ice between bursts. Total cellular protein was determined colorimetrically (Bradford, 1976). Typical assay composition was 100 mM tris/maleate (pH 7.4), 5 mM MgCl₂, 2 mM UDPGA, and 200 to 250 µg of sonicated cell-protein in a volume of 100 µl. DHA was added in dimethyl sulfoxide to a final concentration of 50 to 1000 µM (final dimethyl sulfoxide was <2.5%). Assays were terminated after 40 min by the addition of 100 µl of prechilled methanol, and the precipitated protein removed by centrifugation (1,000g, 10 min). DHA glucuronide was measured using an LC-MS-MS method; eluent was delivered by an HP1100 chromatograph (Agilent Technologies, Stockport, UK) and selected reaction monitoring performed with a Quattro LC tandem quadrupole instrument (Micromass UK Ltd.). The transition of m/z 460.17 (parent ion) to m/z 113 (daughter ion) was monitored in negative ion mode; MS conditions consisted of a nebulizer gas (nitrogen), 100 l/min; desolvation gas (nitrogen), 300 l/min; source block temperature, 100°C; desolvation temperature, 500°C; capillary voltage, 3.5 kV; cone voltage, 30 V; and collision energy, 25 eV using argon as the collision gas. The ion of m/z 113 was the most abundant daughter ion obtained by fragmentation of both the - and -epimeric forms of synthetic DHA-G. The HPLC conditions comprised a gradient of acetonitrile in 0.1% formic acid (20-35% over 15 min, then from 35-70% over 10 min) using a Spherisorb 15 cm × 2.1 mm 5 µM C₈ column (HPLC Technology; Little Mundells, Welwyn Garden City, UK) with a flow rate of 0.3 ml min¹. Under these conditions, the standards of -DHA-G and DHA-G eluted at 11.3 and 13.2 min, respectively. Authentic DHA-G was used for calibration. Assays using control substrates and the recombinant UGTs were run in parallel with the DHA assays to ensure that the sonicated cell lines were active. This was done using a previously published HPLC assay, which measures ¹⁴C UDPGA incorporation into the glucuronide by heterogenous radiochemical detection (Ethell et al., 1998). The activities from these assays and the substrates used are listed with the DHA activity data in Table 2. Kinetic parameters for the glucuronidation of DHA by expressed human UGT1A9 and UGT2B7 were measured by assaying at 50, 100, 250, 500, and 1,000 µM DHA for UGT1A9 and 10, 20, 50, 100, 150, 200, and 250 µM DHA for UGT2B7. The Michaelis-Menten equation was fitted to the experimental data using nonlinear regression analysis (Prism, Graphpad Software, San Diego, CA).

Oxidative Metabolism of [³H]DHA by Human Liver Microsomes. The possibility of cytochrome P450-dependent metabolism of DHA was explored using a published method with minor modifications and albendazole as a positive control substrate (Rawden et al., 2000). Briefly, the final reaction mixture (total volume 250 µl) contained NADP (4 mM), glucose 6-phosphate (40 mM), MgCl₂ (20 mM), glucose-6-phosphate dehydrogenase (10 units), K₂HPO₄ (20 mM, pH 7.4), [³H]DHA (100 µM), and human liver microsomes (2.5 mg of protein). Incubations were carried out at 37°C for 30 min and terminated by the addition of 170-µl ice-cold methanol and vortexing for 10 s. After centrifugation (12,500g, 5 min), supernatant (100 µl) was subjected to HPLC analysis with radiochemical detection as above.

Sulfation of [³H]DHA by Human Liver Cytosol. Assays for sulfotransferase activity were carried out essentially as described previously (Foldes and Meek, 1973); sulfation of paracetamol was included as a positive control. Briefly, the reaction mixture (total volume 250 µl) contained [³H]DHA (100 µM) or paracetamol (1 mM), PAPS (0.4 mM), Na₂HPO₄ (10 mM, pH 7.4), and human liver cytosol (0.4-1 mg of protein). Blank incubations contained no PAPS. Incubations were carried out at 37°C for 60 min and terminated by the addition of 170-µl ice-cold methanol and vortexing for 10 s. After centrifugation (12,500g, 5 min), supernatant (100 µl) was subjected to HPLC analysis with radiochemical detection as above.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Patient Characteristics. The characteristics of the subset of 17 patients in the present study were similar to those of the larger group of 26 from which it was drawn (Batty et al., 1998c), male/female 16:1, age (mean ± S.D.) 28 ± 8 years, weight 50 ± 5 kg, hematocrit 37 ± 7%, hemoglobin 126 ± 21 g l¹, plasma creatinine 102 ± 27 µM, bilirubin 17 ± 11 µM, alanine aminotransferase 36 ± 20 units l¹, parasite count on admission (geometric mean) 9,616 parasites/µl.

Urinary Metabolites of DHA Following Administration of ARTS. When the urine of a healthy volunteer administered artesunate orally was analyzed by LC-MS on the day of collection, it was found to contain -DHA-G (R_t, 14.1 min) but neither -DHA-G (R_t, 13.7 min) nor the glucuronides' furano acetate isomers (Fig. 2). The metabolite was identified by chromatographic (Fig. 2) and mass spectral (Maggs et al., 2000) comparisons with the two synthetic glucuronides. Although a very small peak in the mass chromatogram for m/z 478 ([M + NH₄]⁺), which was also found in the urine of patients, coeluted with synthetic -DHA-G, it did not yield the diagnostic fragments of DHA glucuronide (Maggs et al., 2000).

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Selected-ion chromatogram (m/z 478 for [M + NH4]+) for A, -DHA-G in urine from the white volunteer who took 120 mg of ARTS orally; and B, synthetic standards for -DHA-G (Rt, 13.7 min) and -DHA-G (Rt, 14.2 min). See Materials and Methods for LC-MS conditions.

Isomerisation of DHA-G. The -DHA-G in urine from six of the malaria patients (isomer/DHA-G ratio at t = 0, 0.23-0.91) underwent spontaneous isomerization (6-57%; mean = 26%) over 2 h at 37°C. No reaction occurred in the urine of five other patients (ratio at t = 0, 0.15-3.3). In contrast to the conjugate urine, the radiolabeled -DHA-G in rat bile was stable at 37°C for at least 2.5 h and at 28°C for 38 h. When bile was diluted 40-fold with freshly collected human urine and incubated at 37°C for 2 h, the glucuronide underwent 99 and 95% isomerization, as determined by LC-MS and radiometric analysis, respectively, but no hydrolysis. No rearrangement occurred when the bile was incubated in either acetate (100 mM, pH 5.2) or HEPES (50 mM, pH 7.4) buffer.

Metabolism of [³H]DHA by Human Liver Microsomes and Cytosol. There was no observable turnover of [³H]DHA incubated with either human liver microsomes (L17, L22, L34) and NADPH or human liver cytosols (L36, L37, L38, L39) and PAPS; appropriate control assays run in parallel showed substrate turnover indicating that the assay systems were functional. However, microsomes from all livers tested (L16, L17, L20, L22, L34, L38) could turnover between 5 and 12% of the [³H]DHA during a 1-h incubation with UDPGA. The reaction product had an R_t of approximately 4.1 min on HLPC with radiochemical detection. Blank incubations without microsomes or cofactor produced no turnover of substrate. Preliminary experiments using microsomes from livers L17 and L22 established that 5 mM MgCl₂ and 0.1 mg of the detergent Brij 58 per milligram of protein gave maximal DHA-G production. Addition of the -glucuronidase inhibitor saccharolactone (2 mM) did not increase the reaction yield. Linearity of the assay was demonstrated for microsomal protein concentration (0.5-1 mg per incubation) and duration of incubation (up to 2 h). The K_m and V_max for the formation of DHA-G were investigated in microsomes from four individual human livers, and these data are summarized in Table 3. The mean K_m was 90 µM (range 64-126 µM) while the mean V_max was 177 pmol min¹ mg¹ (range 64-253 pmol min¹ mg¹).

Glucuronidation of DHA by Recombinant UGT. In preliminary experiments, disrupted cell pellets from V79 cells expressing the human UGT1A1 or UGT1A6 did not form DHA-G whereas preparations containing UGT1A9 and UGT2B7 resulted in significant metabolism to -DHA-G (Table 2). A detailed examination of the kinetics of DHA-G formation showed that the K_m was lower for UGT1A9 (32 µM) than for UGT2B7 (438 µM), whereas V_max was similar (8.9 and 10.9 pmol min¹ mg¹, respectively) (Fig. 3, A and B). Incubations with UGT1A9 appeared to generate only the glucuronide of -DHA (Fig. 4B), a finding that is consistent with the in vivo and other in vitro data, but selected reaction monitoring at least suggested the possibility of trace production of -DHA-G by UGT2B7 (Fig. 4A).

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Michaelis-Menten kinetic plots for the formation of -DHA-G using A, expressed UGT1A9 (Vmax = 8.9 pmol min1 mg1, Km = 32 µM); and B, expressed UGT2B7 (Vmax = 10.9 pmol min1 mg1, Km = 438 µM) isoforms in vitro.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   LC-MS-MS selected-ion reaction monitoring of the transition of parent ion (m/z 460.17) to daughter ion (m/z 113) of -DHA-G and -DHA-G. A, supernatant from in vitro incubation of DHA with UGT2B7 and appropriate cofactors; B, supernatant from in vitro incubation of DHA with UGT1A9 and appropriate cofactors; and C, reference standards for -DHA-G (Rt, 11.3 min) and -DHA-G (Rt, 13.2 min). See Materials and Methods for LC-MS-MS conditions.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Our study is the first to identify DHA-G as the primary urinary metabolite of DHA in humans. Using a combination of in vivo and in vitro approaches, DHA-G and its tetrahydrofuran isomer were the only significant metabolites found in the urine of patients with malaria who were treated with ARTS; thus, no deoxygenated or hydroxylated glucuronides analogous to known metabolites of other artemisinin derivatives (Lee and Hufford, 1990; Chi et al., 1991; Maggs et al., 1997; Maggs et al., 2000) were found by LC-MS analysis. DHA epimerizes freely in solution and both epimers have been found in the plasma of ARTS-treated patients (Batty et al., 1996). Irrespective of whether DHA is administered to volunteers and patients as ARTS or to rats as either the lactol (Maggs et al., 1997) or -O-methyl ether (Maggs et al., 2000), it is found as -DHA glucuronide by LC-MS. Stereospecific glucuronidation of DHA also occurs in the isolated perfused rat liver (Batty et al., 1998a). As might be expected from the 40 to 50 min t_1/2 of DHA in patients (Batty et al., 1998b,c), these conjugates were in greatest concentration in the urine collected in the first 2 h after dose. The ratio of -DHA-G to its tetrahydrofuran isomer in the urine varied widely between patients. Since the furano acetate isomer of DHA-G can be absent from freshly collected urine following administration of ARTS, and -DHA-G isomerizes spontaneously in urine, the isomeric conjugate found in patients' urine was likely to have been at least partly a product of reactions occurring ex vivo. The documented chemistry of reactions between iron (II) and endoperoxides (Maggs et al., 2000; O'Neill et al., 2001a) and the effect of EDTA observed in the present study suggest an action of iron in urine [urinary Fe²⁺ in healthy adult males is 0.73 ± 0.32 µM (Hjortso et al., 1990)]. The stability of -DHA-G in bile might reflect the presence of iron-binding protein (Regoeczi and Chindemi, 1995). Caution is required in the interpretation and quantification of the urinary metabolites of endoperoxide drugs. Thus, we conclude that between-patient variation in the ratio of -DHA-G to its tetrahydrofuran isomer might be due primarily to varying levels of Fe²⁺ in the urine.

Studies using human liver microsomes showed that [³H]DHA-G was the only detectable metabolite of [³H]DHA. Other pathways such as cytosolic sulfation and microsomal CYP450 oxidation did not have a significant role in the disposition of DHA. The mean calculated unbound intrinsic clearance for the reaction to DHA-G was 3.1 × 10⁶ l min¹ mg¹ of microsomal protein (Cl_int = V_max/K_m; corrected for a free fraction of DHA in human liver microsomal incubations of 0.64; P. Gibbons, personal communication). If it is assumed that the liver is approximately 2.2% of body weight in humans and that the yield of microsomal protein is around 25 mg g¹ of liver (G. T. Tucker, personal communication), in vivo hepatic plasma clearance calculated from the equation for the well stirred model is approximately 0.05 l h¹ kg¹
where CL_h = total hepatic clearance, Q_h = hepatic plasma flow (0.78 l h¹ kg¹), and f_u = DHA fraction unbound in plasma (0.56; Batty, 1999). This is 8.4-fold lower than the mean unbound plasma clearance for DHA after i.v. ARTS administration to patients (0.42 l h¹ kg¹) (Batty et al., 1998a,b). Our finding is consistent with recent data for a range of glucuronidated drugs where CL_h calculated from in vitro human liver microsomal metabolism data as above, consistently underestimated in vivo clearance by around10-fold (Soars et al., 2002).

Inhibitor studies of the glucuronidation of DHA in human liver microsomes showed low IC₅₀ values (<500 µM) for substrates such as ethinyloestradiol, diclofenac, ketoprofen, oxazepam, and diazepam. Since they are substrates for UGT1A1, UGT1A8, UGT1A9, or UGT2B7 (Ebner et al., 1993; Sabolovic et al., 2000), we reasoned that these isoforms might be involved in the glucuronidation of DHA. Metabolism by UGT1A4, UGT1A6, UGT1A10, or UGT2B15 can probably be excluded because high concentrations of the respective preferred substrates mycophenolic acid (Mojarrabi and Mackenzie, 1997), lamotrigine (Magdalou et al., 1992), paracetamol (Bock et al., 1993), or 5-andostene-3,17-diol (Belanger et al., 1995) were weak inhibitors of DHA glucuronidation. Using expressed human UGTs and a sensitive LC-MS-MS assay system, we were able to show a lack of DHA-G formation by UGT1A1 and UGT1A6 as well as significant, highly stereoselective -DHA-G formation by UGT1A9 and UGT2B7. Such a high level of stereoselectivity for glucuronidation has been described only rarely (Aumont et al., 2001). These authors claimed stereospecific conjugation of cis-or trans-resveratrol by certain UGTs, but a stereoselective reaction at the stereogenic center of epimers does not appear to have been reported. While the K_m for DHA glucuronidation by UGT1A9 (32 µM) was much lower than that for UGT2B7 (438 µM), these values were nevertheless comparable with K_m values from the human liver microsomal experiments (90 µM) and suggest that UGT1A9 may be the dominant isoform for the metabolism of DHA. While the formation of DHA-G observed in human liver microsomes is consistent with metabolism by UGT1A9 in this tissue, it should be noted that this isoform is expressed in colon (Strassburg et al., 1998) and a variety of extrahepatic tissues (Albert et al., 1999), as well as in liver (Strassburg et al., 1998; Ren et al., 2000), and that other isoforms also may be involved.

In summary, our study has identified -DHA--G as the major metabolite, and UGT1A9 and UGT2B7 as the predominant isoforms involved in the clearance of DHA in humans. Apart from a role in the rapid clearance of DHA, the glucuronide pathway may have implications for drug interactions with DHA. There are examples of induction and of inhibition interactions arising from two substrates competing for UGT pathways in vivo. With induction, the oral clearance of propafenone is markedly increased during cotreatment with the inducer rifampicin, leading to a clinically significant interaction (Dilger et al., 2000). With inhibition, tacrolimus can significantly increase the area under the plasma concentration-time curve for mycophenolic acid in transplant patients and require dose reduction of the latter to avoid toxicity (Zucker et al., 1997). Thus, it is possible that mechanistically similar drug interactions also might occur with DHA. Induction or inhibition could, respectively, decrease or increase antimalarial efficacy. Currently, the most common cotreatment used with ARTS or DHA is mefloquine, but there are no reports of any interaction. In vitro data from the present study also indicate that the antipyretic paracetamol is a very poor inhibitor of DHA glucuronidation and is therefore unlikely to result in a significant interaction in vivo.