Abstract
β-Artemether (AM), the O-methyl ether prodrug of dihydroartemisinin (DHA), is an endoperoxide antimalarial. The biliary metabolites of AM in adult male Wistar rats were characterized with particular reference to potential antimalarial compounds and stable derivatives of free radical intermediates. [13-14C]-AM (35 μmol kg−1, i.v.) was administered to anesthetized rats. Within 0 to 3 h, 38.6 ± 4.8% (mean ± S.D.,n = 6) of the radiolabel was recovered in bile; the 0- to 5-h recovery was 42.3 ± 4.3%. The major metabolites (0–3 h) were the glucuronides of 9α-hydroxyAM (33.4 ± 6.8% biliary radioactivity) and α-DHA (22.5 ± 4.4%); four stereochemically unassigned monohydroxyAM glucuronides (II, 3.1 ± 0.9;IV, 4.4 ± 1.7%; V, 21.4 ± 3.0%;VI, 3.0 ± 1.1%) and a dihydroxyAM glucuronide (6.0 ± 2.1%) were also identified. A sixth monohydroxyAM glucuronide (VIIa) and desoxyDHA glucuronide were detected in trace amounts. The furano acetate isomer of DHA glucuronide, indicative of the formation of a radical intermediate, was also found in trace amounts. O-methyl substitution of DHA favors ring hydroxylation in vivo. However, the principal hydroxylated metabolite, 9α-hydroxyAM, is unlikely to possess significant antimalarial activity.
β-Artemether (AM)1(Paluther, Rhone-Poulenc Rorer, Antony, France; Fig.1) is a semisynthetic polyoxygenated amorphene containing a peroxide bridge that confers potent antimalarial activity (Meshnick et al., 1996). It is theO-methyl ether of dihydroartemisinin (DHA) and a derivative of artemisinin (qinghaosu), the principal antimalarial constituent of the medicinal plant Artemisia annua. AM is active against the erythrocytic stage of multidrug-resistant strains ofPlasmodium falciparum (Sowunmi and Odulola, 1997). Oral, rectal, and i.m. regimens are generally rapidly effective and well tolerated treatments for both severe and uncomplicated falciparum malaria (de Vries and Dien, 1996; Murphy et al., 1997), although short-term monotherapy has been associated with high rates of relapse. Combinations with nonendoperoxide antimalarials have been used (de Vries and Dien, 1996).
The parasiticidal activity of endoperoxides has been attributed to chemical activation of the drugs within the food vacuole of the intraerythrocytic stage of the parasite: it is proposed that reductive cleavage of the peroxide bridge by heme liberated during digestion of hemoglobin generates free radicals, which induce oxidative stress and alkylate heme and vital parasite proteins (Cumming et al., 1997; Robert and Meunier, 1998). In chemical systems, iron (II) catalyzes the isomerization of artemisininoid peroxides to ring-contracted (furano acetate) and 3α-hydroxydesoxy compounds (Fig. 1) indicative of the sequential intermediacy of alkoxyl and carbon-centered radicals (Jefford et al., 1996; Butler et al., 1998). Such compounds are minor metabolites of DHA and its O-ethyl ether (arteether) in nonparasitized rats (Chi et al., 1991; Maggs et al., 1997) but their putative free radical precursors are not thought to contribute to the drugs' antimalarial activity. Lacking a peroxide functionality, the stable metabolites are pharmacologically inactive. Although the mechanism(s) of activation in mammalian cells has not been determined, it has clear if unsubstantiated implications for the clinical safety of endoperoxides (Park et al., 1998), and especially with respect to the causation of the neurotoxicity associated with this class of drugs (Smith et al., 1998).
AM is regarded as a prodrug of DHA in humans (Teja-Isavadharm et al., 1996) although the two endoperoxides have similar parasiticidal activities in vitro (de Vries and Dien, 1996). Oral doses are absorbed rapidly and converted extensively to DHA in humans (Na Bangchang et al., 1994) and rats (Li et al., 1998). Systemic drug exposure is generally greater when AM is given i.m. (Teja-Isavadharm et al., 1996;Li et al., 1998) but the bioavailability in severe malaria may be highly variable (Murphy et al., 1997). DHA in plasma is the only known human metabolite of AM (Na Bangchang et al., 1994; Teja-Isavadharm et al., 1996) and hitherto plasma DHA has been the only identified metabolite in rats (Li et al., 1998). Demethylation of AM also occurs in mice (China Cooperative Research Group, 1982b; Lee and Hufford, 1990). However, isolated perfused rat livers are reported to metabolize AM to desoxyDHA, 9α-hydroxyAM, and 9β-hydroxyAM, as well as DHA (Lee and Hufford, 1990). The metabolites of arteether (i.v.) in rat plasma (Chi et al., 1991; Ramu and Baker, 1997) and in rat liver microsomes (Hufford et al., 1990; Leskovac and Theoharides, 1991a) include DHA and several hydroxylated derivatives, one of which, 9β-hydroxyarteether, is a highly active antimalarial in vitro (Chi et al., 1991). In contrast, most hydroxyl substituents render the endoperoxide pharmacologically inactive. Because 9- hydroxyAM of unspecified stereochemistry (Lee and Hufford, 1990) and 9β-hydroxyarteether O-glucuronide (Ramu and Baker, 1995,1997) are reported to possess activity in vitro, it is possible that hydroxylated metabolites contribute to the activity of AM in vivo.
We have characterized the biliary metabolites of β-[13-14C]AM ([14C]AM) in rats with particular reference to isomeric compounds having radical precursors and C-9 hydroxylated derivatives, which might retain antimalarial activity.
Materials and Methods
Chemicals.
AM was provided by Dr. P. Buchs (SAPEC SA Fine Chemicals, Lugano, Switzerland). [14C]AM (12.1 mCi/mmol) was provided by Dr. J. A. Kepler (Organic and Medicinal Chemistry, Research Triangle Institute, Research Triangle Park, NC). It had a radiochemical purity of 99%: the drug was eluted (Rt 17.4 min) from a Beckman Ultrasphere 5-μm C18 column (25 cm × 0.46 cm i.d.) with acetonitrile/water (3:2, v/v) at a flow rate of 1.2 ml min−1 and the radioactivity was measured using an on-line detector (Radiomatic Flo-Oneβeta A250; Packard, Pangbourne, Berkshire, UK). [12-3H]DHA (1.4 Ci/mmol; radiochemical purity >99%) was obtained from Moravek Biochemicals Inc. (Brea, CA). Standards of 9α- and 9β-hydroxyAM and of 14-hydroxyAM (Abourashed and Hufford, 1996) were generously donated by Dr. C. D. Hufford, The University of Mississippi (Oxford, Mississippi). The sodium salts of α-DHA β-glucuronide and β-DHA β-glucuronide were provided by Ultrafine Chemicals (Salford, UK). DesoxyDHA and the furano acetate isomer of DHA were prepared from DHA and artemisinin, respectively, as described previously (Maggs et al., 1997); the direct reaction of DHA with iron (II) chloride tetrahydrate consistently yielded an unidentified isomer. The β-glucuronidase preparaton from Helix pomatia (type H-2; 89.4 × 103 U β-glucuronidase/ml) and uridine 5′-diphosphoglucuronic acid (UDPGA) were purchased from Sigma Chemical Co. (Gillingham, Dorset, UK). HPLC grade solvents were products of Fisher Scientific (Loughborough, Leicestershire, UK).
Animal Experiments.
Male Wistar rats (210–240 g) were anesthetized with urethane (1.4 g/ml isotonic saline; 1 ml/kg, i.p.). Cannulas were inserted into the jugular vein and common bile duct and the penis was ligated. [14C]AM (35 μmol/kg) dissolved in dimethyl sulfoxide was administered i.v. (10 μCi/kg; 250 μl). Bile was collected hourly for 5 h at room temperature; it was stored at −30°C until analyzed by liquid chromatography-mass spectrometry (LC-MS) and liquid chromatography-tandem mass spectrometry (LC-MS-MS) with parallel radiometric detection. After 5 h, urine was aspirated from the bladder. Radioactivity in aliquots (50-μl) of bile and urine was assayed by liquid scintillation counting. Portions (ca. 50 mg) of the major internal organs and of skin were solubilized in 1 ml of OptiSolv (Wallac UK, Milton Keynes, Buckinghamshire, UK) at 50°C over 16 h. The solutions were decolorized with H2O2 (400 μl), neutralized with glacial acetic acid (60 μl), left in darkness overnight, and finally mixed with 20 ml of Ultima Gold (Packard Bioscience BV, Groningen, the Netherlands) for scintillation counting. The complete skin was stripped from each carcass, and the remainder immersed in 3 M KOH (200 ml) in a capped plastic bottle, which was maintained at 50°C for 16 h. An aliquot of the resulting liquid fraction (3 ml) was neutralized with concentrated HCl, and samples (1 ml) were taken for scintillation counting.
Enzymic Hydrolysis.
Pooled bile (0- to 2-h collections; 80–200 μl) from two or three rats was diluted with 0.1 M sodium acetate, pH 5.0, (final volume, 1 ml) and incubated with β-glucuronidase preparation (ca. 3 × 103 U β-glucuronidase) in a capped glass test tube at 37°C for 16 h. Thereafter, the incubation was extracted with methyl tert-butyl ether (5 volumes × 2). The pooled extracts were evaporated to dry residue at 35°C under a stream of oxygen-free nitrogen, reconstituted in methanol (150 μl), and analyzed immediately by LC-MS. Recovery of biliary radioactivity in the reconstituted extracts was 85 to 95%.
Isomerization of Endoperoxide Glucuronides.
Aqueous solutions of synthetic α- and β-DHA β-glucuronides (100 μl, 10 mM) were mixed with iron (II) chloride tetrahydrate in water (50 mM; final concentration, 1 mM) and incubated at 37°C for 3 h. The parent conjugate and its furano acetate and hydroxydesoxy isomers were resolved and identified by LC-MS. Bile (2 μl; 100 × 103 dpm) from an isolated rat liver perfused with [3H]DHA in a single-pass system (Batty et al., 1998b) contained only [3H]DHA glucuronide (3.8 mM) by radiochromatographic and LC-MS analyses. It was diluted (2:93, v/v) with bile collected from an anesthetized rat before the animal was administered [14C]DHA. The mixture was incubated with iron (II) chloride in water (40 mM; final concentration, 10 mM) in a capped glass test tube at 37°C for 2 h. Bile (0- to 1-h collections; combined volumes, 120 μl) from two of the rats administered [14C]AM was treated in the same manner for up to 2.5 h. Each incubation was centrifuged periodically to sediment a precipitate, and the supernatant (30–50 μl) analyzed immediately by LC-MS.
Microsomal Glucuronylation.
Hepatic microsomes from adult male Wistar rats were prepared by the standard differential centrifugation technique. They were incubated (final protein concentration, 3.5 mg/ml) with either [3H]DHA (0.1–1.0 mM; 0.1 μCi), desoxyDHA (0.1 mM), or the furano acetate isomer of DHA (0.1 mM) in polypropylene microcentrifuge tubes using a modified form of the method of Batty et al. (1998b). The complete mixtures contained MgCl2 (5 mM) and UDPGA (3 mM) in Tris-HCl buffer (50 mM, pH 7.4; final volume, 250 μl). Reactions were initiated by adding the UDPGA, performed at 37°C for up to 30 min, and terminated with ice-cold methanol (170 μl). The mixtures were placed on ice. The precipitate was sedimented by centrifugation, and aliquots (100-μl) of the supernatant were analyzed by LC-MS.
LC-MS and LC-MS-MS Analyses.
Positive- and negative-ion electrospray mass spectra were obtained by LC-MS using a Quattro II tandem quadrupole instrument fitted with the standard in-line source (Micromass, Manchester, UK). Eluent was delivered by twin Jasco PU-980 pumps (Jasco UK, Great Dunmow, Essex, UK). Samples were eluted from an Ultracarb 5-μm C8 column (25 cm × 0.46 cm i.d.; Phenomenex, Macclesfield, UK) at 0.9 ml min−1 as follows: bile (20–100 μl), with a gradient of acetonitrile (10 to 20% over 10 min; 20% for 10 min; 20 to 30% over 15 min; 30 to 50% over 10 min) in 0.1 M ammonium acetate, pH 6.9, except that a modified gradient (10 to 30% over 20 min; 30 to 50% over 10 min; 50 to 70% over 5 min) was used to locate nonpolar metabolites; deconjugated biliary metabolites and coinjected standards, with a gradient of methanol (40 to 75% over 30 min) in the acetate buffer; microsomal glucuronylation mixtures, authentic DHA glucuronides, and isomers, with a gradient of acetonitrile (20 to 35% over 15 min; 35 to 70% over 10 min) in the acetate buffer. Eluate was split between a radioactivity flow detector and the LC-MS interface (ca. 40 μl min−1). The source temperature was 70°C; the capillary voltage, 3.9 × 103 V (3.1 × 103 V for anion analysis); the high voltage lens and radio frequency lens voltage, 0.2 × 103V (0.0 V) and 0.1 V (0.0 V), respectively. Spectra were acquired between m/z 100 to 1050 at one scan/5 s. Analyte fragmentation was achieved by increasing the cone voltage from 30 V (45 V). Collision-induced dissociation (CID) of the ammonium adducts of major metabolites in bile was achieved at a collision energy of 20 eV using argon at 9 × 10−3 mBar. The cone voltage was 20 V. Daughter spectra were acquired betweenm/z 100 to 550 at one scan/5 s. All data were processed via MassLynx 2.1 software (Micromass Ltd., Manchester, UK).
Radiometric HPLC.
Eluate was mixed with Ultima Flo AP scintillant (1.0 ml min−1; Packard Bioscience) in a Radiomatic detector, and the radiolabeled components were quantified after background subtraction using A250-1.6 software.
Results
Excretion and Distribution of Radioactivity.
After the i.v. administration of [14C]AM (35 μmol/kg) to male rats, 23.4 ± 4.7 (mean ± S.D.,n = 6), 11.0 ± 3.2, and 4.2 ± 1.0% of the radioactivity was excreted into bile during the first, second, and third hours, respectively; the biliary recovery over 5 h was 42.3 ± 4.3%. Only 5.4 ± 0.9% of the radioactivity was found in the bladder urine 5 h after dosing.
The liver, kidneys, and intestinal tract contained 5.5 ± 1.6, 4.4 ± 1.3, and 6.9 ± 1.6% of the dose, respectively; the skin, 8.7 ± 3.5%. Brain, heart, lungs, and spleen each contained less than 0.5%. The recovered carcass residues represented 7.6 ± 2.8%. Total recoveries from the soft tissues, skin, and carcass equaled 33.9 ± 7.1%. Fifth-hour plasma samples were devoid of quantifiable radioactivity.
Microsomal Glucuronylation.
[3H]DHA incubated with UDPGA was metabolized to a single glucuronide (Rt 12.5 min; 75% conversion of 0.1 mM [3H]DHA over 30 min), which gave an electrospray spectrum containingm/z 478 ([M + NH4]+, 37), 267 ([478 − NH3 − dehydroglucuronic acid (DHG) − H2O]+, 31), 221 ([267 − CO − H2O]+, 16) and 163 ([221 − (CH3)2CO]+, 100). DesoxyDHA underwent incomplete turnover to a glucuronide (Rt 12.5 min), which yieldedm/z 462 ([M + NH4]+, 46) andm/z 251 ([462 − NH3− DHG − H2O]+, 100). The fragmentation pathways of ammonium adducts of artemisinin derivatives involving loss of ammonia, water, and CO have been characterized by Chi and Baker (1993). DHA furano acetate did not undergo glucuronylation although, in separate incubations, coincubated equimolar DHA was conjugated.
Isomerization of DHA Glucuronides.
The synthetic α- and β-DHA β-glucuronides (Rt 12.5 and 12.0 min, respectively) each yielded only two isomeric products when reacted with iron (II) chloride (Rt 5.5 and 10.0 min for the α/β-, and 4.5 and 10.0 min for the β/β-glucuronide, respectively). Analogous reactions of artemisinin compounds produce furano acetate and 3α-hydroxydesoxy isomers (Jefford et al., 1996). The major furano acetate products were identified by their qualitatively identical electrospray spectra; specifically by the diagnostic loss of the acetate group (Chi and Baker, 1993; Maggs et al., 1997): for the α/β conjugate, m/z 478 ([M + NH4]+, 100), 401 ([M + NH4 − NH3 − CH3CO2H]+, 7), 373 ([401 − CO]+, 8), 267 ([M + NH4 - NH3 − DHG − H2O]+, 24), 225 ([401 − DHG]+, 33), 207 ([225 − H2O]+, 23). The minor products, taken to be the hydroxydesoxy (epoxide) isomers, gave only one fragment: for the α/β conjugate, m/z 478 ([M + NH4]+, 27) andm/z 267 (100).
Identification of the Biliary Metabolites of [14C]AM.
The metabolites of [14C]AM in bile (0- to 3-h pooled collections; 38.6 ± 4.8% of dose) were resolved by HPLC and quantified radiometrically (Table 1). Three major radiolabeled components (III, Rt 14.0 min; V, 22.0 min;VII, 30.5 min) were found (Fig.2); they and the lesser metabolitesI (9.0 min), II (13.0 min), IV (15.5 min), and VI (25.0 min) were all present in the first hourly collection. Small amounts of a nonpolar compound (Rt 41.5 min; ≤2% chromatographed radioactivity), which was isobaric (m/z 302, [M + NH4]+) with AM (Rt 44 min) were found occasionally.
Metabolite VII coincided with the predominant peak in the selected-ion (m/z 478) chromatogram corresponding to an ammonium adduct of DHA glucuronide (Fig.3b). It formed weaker anions atm/z 459 ([M − 1]−; 30) and m/z 519 ([M + CH3CO2H − 1]−; 100) but did not yield any negatively charged fragments under conditions that provided abundant cationic fragments of the aglycone (Table 2). Neither AM nor DHA yielded anionic species. The metabolite cochromatographed with the glucuronide of DHA formed by rat liver microsomes. It also cochromatographed with the DHA glucuronide excreted into bile by isolated perfused rat livers (Batty et al., 1998b), the corresponding metabolite of sodium artesunate (β-DHA hemisuccinate) in human urine (Batty et al., 1997) and synthetic α-DHA β-glucuronide but not with the synthetic β-DHA β-glucuronide (Rt 30.0 min), which had an electrospray spectrum qualitatively identical with the spectra of its isomer (data not shown) and the metabolite. The electrospray spectra of VII andIII (Fig. 4a; Table 2) differed somewhat from the thermospray spectra of synthetic DHA glucuronide and 9β-hydroxyarteether glucuronide, respectively (Ramu and Baker, 1997); principally in the order of elimination of neutral fragments. Finally, the aglycone moiety of VII underwent the major iron (II)-mediated isomerization—no reaction occurred in the absence of the iron—characteristic of an artemisininoid endoperoxide (Jefford et al., 1996), although the reaction required a high concentration of iron (>1 mM; 10 mM used here) to reach near completion within 1 h. DHA glucuronide in human urine undergoes the same conversion spontaneously (Maggs et al., 1998), presumably because of the relatively high ratio of iron to iron-binding protein in normal urine. The furano acetate glucuronide (Rt21.0 min) was identified by its electrospray spectrum, which was identical with that of the isomer generated from [3H]DHA glucuronide in the bile of isolated perfused rat livers (data not shown). A minor metabolite (ca. 1% chromatographed radioactivity) corresponding to this conjugate was found sporadically in the 0- to 1-h bile collections by radiometric and mass spectrometric analyses, and otherwise by the latter method alone. It cochromatographed with the isomer obtained chemically from synthetic DHA β-glucuronide and yielded characteristic mass spectra (Fig.5a; Tables 2 and3).
[14C]DHA glucuronide and its aglycone liberated enzymically from three pooled bile samples (Rt32.0 min; m/z 302 ([M + NH4]+, 100), 284, 267, 221, 207, 163; 22% eluted radioactivity) eluted immediately before trace amounts of desoxyDHA glucuronide (peak-to-peak resolution, 0.2 min) and desoxyDHA, respectively, which were identified by LC-MS but not locatable by radiometric profiling. For desoxyDHA glucuronide:m/z 462 ([M + NH4]+, 10;m/z 478 from DHA glucuronide in scans, 100) andm/z 251 (24). For desoxyDHA:m/z 286 ([M + NH4]+, 4) andm/z 251 ([M + NH4 − NH3 − H2O]+, 25).
Metabolites II to VI each yielded, although at a distinctively low relative intensity in the case of II, an ion of m/z 508 equivalent to [M + NH4]+ for a glucuronide of monohydroxylated AM (Fig. 3a); III, V, andVI were also detected as weak anions atm/z 489 ([M − 1]−). Except for two of the lesser metabolites,II and VI, the fate of which could not be discerned, the monohydroxyl glucuronides were all shown to possess an endoperoxide linkage by their iron (II)-mediated conversion to a radiolabeled isomeric product (Rt 13.0 min), which was taken to consist of coeluting glucuronides. The isomerizations were slower than that of DHA glucuronide although they reached completion within 2.5 h. The electrospray spectrum of the product peak was analogous to that of the DHA glucuronide isomer (Table2) and diagnostic of the furano acetate form of an artemisinin compound: m/z 508 ([M + NH4]+, 100), 431 ([M + NH4 − NH3 − CH3CO2H]+, 54), 403 ([431 − CO]+, 27), 399 ([431 − CH3OH]+, 39), 371 ([399 − CO]+, 24), 255 ([431 − DHG]+, 23), 223 ([399 − DHG]+, 75), 205 ([223 − H2O]+, 21). AsII and VI fragmented in a similar fashion toIII and IV they are also taken to be endoperoxides. Furthermore, none of the metabolites underwent the simplified fragmentation of hydroxydesoxyDHA glucuronide and other artemisininoid epoxides (Chi et al., 1991).
Whereas II to IV and VI produced qualitatively similar spectra, which included prominent ions attributable to elimination of the O-methyl group as methanol (Chi and Baker, 1993; Ramu and Baker, 1997), and a peak atm/z 265 arising from loss of the glucuronic acid moiety (DHG; 176 amu), V alone yielded a (base) peak atm/z 251 (Fig. 4b; Table 2). The distinctive character of the fragmentation of V was confirmed by LC-MS-MS (Fig. 5b; Table 3). Formation of m/z 251 is most easily represented as expulsion of the peroxide bridge with formation of a keto alkene (Fig. 6). By implication, this process is dependent on the position of the hydroxyl function. Previously, loss of molecular oxygen from artemisinin endoperoxides has been observed only during electron impact analysis of artemisinin and DHA (Fales et al., 1990). One of the metabolites (Rt 30.0 min; 24% eluted radioactivity) recovered from pooled bile after enzymic hydrolysis of the conjugates was identified as the aglycone of V by the presence in its electrospray spectrum (Fig. 7b) of a peak at m/z 251; at a cone voltage of 50 V:m/z 332 ([M + NH4]+, 78), 300 ([M + NH4 − CH3OH]+, 21), 283 (100), 251 (26), 219 (39), 205 (7), 179 (7), 161(10).
Although the smaller fragments of III andIV (m/z 283, 265, 247, 237, 219, 205, and 179) corresponded to those of 9α- and 9β-hydroxyAM, the epimers of the standard compound were indistinguishable and the same match was achieved with 14-hydroxyAM. The principal radiolabeled aglycone liberated from bile by enzymic hydrolysis (Rt25.0 min; 38% eluted radioactivity) coeluted with 14-hydroxyAM but only with the α-epimer of 9-hydroxyAM (Rt9β-hydroxyAM, 24.0 min); none of the aglycones coeluted with 9β-hydroxyAM. Its spectrum (Fig. 7a) matched that of the α-epimer; for the metabolite: m/z 332 ([M + NH4]+, 100), 300 (34), 283 (53), 269 (7), 247 ([283–2 H2O]+, 10), 237 (9), 219 (27), 207 (9), 205 (4), 201 (5), 179 (32), 161 (24). Additionally, two ions (m/z 214 and 231) were found uniquely in the spectrum of 14-hydroxyAM (14-hydroxyarteether is not a metabolite of arteether in rats; Chi et al., 1991). III is identified as 9α-hydroxyAM glucuronide.
A minor peak in the m/z 508 chromatogram (VIIa; 29.5 min) for bile coeluted with DHA glucuronide (VII) but was identifiable as an hydroxyAM glucuronide from its daughter spectrum (Table 3).
Metabolite I coincided with a relatively weak peak in the selected-ion chromatogram (m/z 524) for ammonium adducts of dihydroxyAM glucuronides. As with V, its fragmentation included a second loss of 32 amu, which could be rationalized in terms of the elimination of the peroxide bridge as molecular oxygen.
None of the following potential metabolites of AM was located in rat bile as a glucuronide: hydroxyDHA, hydroxydesoxyAM, and hydroxydesoxyDHA (Rt for hydroxydesoxyDHA conjugate obtained by isomerization of synthetic DHA glucuronide, 9.5 min).
Discussion
The biliary metabolites of AM in male rats given the drug i.v. were predominantly the glucuronylated products of two oxidative pathways, namely O-demethylation and hydroxylation, operating in parallel (Fig. 8). Previous studies have established the role of biliary excretion in the elimination of artemisininoid metabolites from rats (Maggs et al., 1997; Batty et al., 1998b) and mice (China Cooperative Research Group, 1982b). An early report on the distribution of AM in rats noted that the highest levels of the drug were in brain 5 min after i.v. administration (Lee and Hufford, 1990) but the present measurements of total tissue radioactivity after 5 h indicate that this was a transitory concentration.
The glucuronylation of DHA in rats and humans was highly selective for the α-epimer: DHA epimerizes rapidly in solution and both epimers are plasma metabolites of artesunate in patients (Batty et al., 1996) but only α-DHA β-glucuronide was found in human urine and rat bile.Ilett et al. (1998) have partially characterized the glucuronylation of DHA by human liver microsomes. Malaria infection reduces the glucuronylation of DHA by rat liver (Batty et al., 1998a). Notwithstanding the dealkylation of arteether in vivo (Chi et al., 1991; Li et al., 1998) and the conjugation of DHA derived metabolically from AM, DHA glucuronide is not detected in either the plasma or urine of rats given arteether (Ramu and Baker, 1997). In the rat, its excretion appears to be confined to bile. α-DHA β-glucuronide given i.v. to cannulated rats is eliminated rapidly in bile as a mixture of the parent conjugate and a small amount of furano acetate glucuronide (J.L.M., unpublished observation), and in the absence of hydrolysis by plasma β-glucuronidase might undergo direct uptake into the liver.
The lack of sequential dealkylation and hydroxylation of AM, as deduced from the biliary metabolite profile, contrasts with the biotransformations of artemisinin endoperoxides in vitro. Thus DHA is converted extensively to four undefined hydroxylated products by rat liver microsomes (Leskovac and Theoharides, 1991a) and one of the partly characterized metabolites of arteether in a crude hepatic subcellular preparation is an isomer of hydroxyDHA (Baker et al., 1989). Arteether is deethylated by a number of human hepatic P-450s (Grace et al., 1998). A plasma metabolite of i.v. arteether has been assigned tentatively to 9α-hydroxyDHA (Chi et al., 1991) but no hydroxylated DHA (glucuronide or aglycone) was located in bile using LC-MS during either this study or earlier studies on the biotransformations of DHA in vivo (Maggs et al., 1997) and in isolated perfused rat livers (Batty et al., 1998b). Subcellular incubations do not provide a balanced representation of the metabolic fate of artemisinin endoperoxides in vivo; in particular, the ratio of dealkylation to hydroxylation can be much higher in vitro (Baker et al., 1989; Chi et al., 1991). Demethylation of AM in mice is reported to be greater after i.v. versus i.m. dosing (Lee and Hufford, 1990).
Unlike microsomal hydroxylation of DHA, the relatively slow (NAD/NADH-dependent) deoxygenation of the endoperoxide function of DHA obtained with rat liver cytosol (Baker et al., 1989; Leskovac and Theoharides, 1991b) is reproduced in the biotransformations of AM in vivo. DHA rather than the parent drug is likely to be the substrate for reduction, as this seems to be the case with arteether (Baker et al., 1989; Chi et al., 1991). DesoxyDHA is also formed from AM by isolated rat livers (Lee and Hufford, 1990), and deoxygenation is a minor pathway for DHA in rats (Maggs et al., 1997) if not in isolated livers (Batty et al., 1998b). Desoxyartemisinin is a urinary metabolite of artemisinin in humans (Lee and Hufford, 1990). These epoxide products are pharmacologically inactive in vitro (Cumming et al., 1997) and in vivo (China Cooperative Research Group, 1982a).
The biliary metabolites of AM were examined for conjugates of two stable DHA isomers, namely 3α-hydroxydesoxyDHA and the furano acetate, because their biomimetic iron (II)-mediated synthesis generates free radical intermediates via distinct pathways of one-electron reduction (Jefford et al., 1996; Butler et al., 1998). Examples of these derivatives are minor metabolites of arteether (Chi et al., 1991) and DHA (Maggs et al., 1997) in rats and of arteether (Baker et al., 1989), artelinate (Idowu et al., 1997), and DHA (Baker et al.,1989) in subcellular fractions. Arteflene, an unrelated synthetic antimalarial endoperoxide, undergoes chemically analogous but much more extensive bioactivation in the rat (Bishop et al., 1999). The present failure to locate a biliary glucuronide of either hydroxydesoxyAM or hydroxydesoxyDHA and the excretion of just trace amounts of the furano acetate of DHA as a glucuronide would appear to confirm the essential stability of the artemisininoid peroxide bridge in vivo. This stability is additionally emphasized by the absence from bile of glucuronides of AM and DHA diols, i.e., the products of reduction analogous to those produced by prostaglandin endoperoxide reductase (Burgess and Reddy, 1997); indeed, no diol metabolite of an artemisinin compound has been described to date. The restrained metabolism of the endoperoxides contrasts with the compounds' reactivity toward inorganic iron (II), which exceeds that of simple dialkyl peroxides (Butler et al., 1998). The mechanism(s) of rearrangement of endoperoxides in biological systems is unknown. The prostacyclin and thromboxane synthase hemethiolate enzymes (Ullrich and Brugger, 1994) appear the most likely enzymic candidates (Maggs et al., 1997) but only microsomal P-450 (Idowu et al., 1997) and hemoglobin (Blum et al., 1998) have been implicated experimentally; AM is isomerized when incubated with either blood or hemoglobin (Blum et al., 1998). A predominant role for blood is implied by the observation ofBatty et al. (1998b) that isolated rat livers perfused with DHA in a synthetic medium only very rarely excrete isomers of DHA glucuronide into bile. Hepatic cytochrome P-450 seemingly catalyzes extensive hydroxylation and O-dealkylation without obtaining effective access to the peroxide bridge.
The major metabolite of AM in rat bile was identified as 9α-hydroxyAM glucuronide. A regioselective hydroxylation at C-9 on the artemisinin skeleton is catalyzed by rat liver microsomes (Leskovac and Theoharides, 1991a) and is also suggested by the plasma metabolites of arteether in male rats (Ramu and Baker, 1997). However, an earlier analysis found a derivative identified tentatively as 3α-hydroxyarteether at a higher plasma concentration than 9α-hydroxyarteether (Chi et al., 1991). Because the former has a much shorter retention time on a reversed-phase column than the latter, whereas metabolites II to VIIa of AM elute either immediately before or after the isomeric 9α-hydroxyAM glucuronide, it is thought improbable that one of the unassigned hydroxyAM conjugates is 3α-hydroxyAM glucuronide. Although arteether is metabolized to 2- and 14-hydroxyarteether, as well as 9α- and 9β-hydroxyarteether, by rat liver microsomes (Hufford et al., 1990; Leskovac and Theoharides, 1991a), neither 2- nor 14-hydroxyAM glucuronide would be accounted prospective metabolite candidates because neither 2- nor 14-hydroxyarteether nor their conjugates has been detected in the plasma of rats given arteether (Chi et al., 1991; Ramu and Baker, 1997). Stereoselective C-9 hydroxylation occurs in rat hepatic microsomes (Leskovac and Theoharides, 1991a), and was first demonstrated in vivo when the α- and β-epimers of 9-hydroxyarteether were found in plasma at the ratios 3.4 (Ramu and Baker, 1997) and 6.5 (Chi et al., 1991) after i.v. administration of the drug. Similarly, both epimers of 9-hydroxyAM have been recovered from the perfusate of isolated rat livers (Lee and Hufford, 1990), although without quantification. Despite these findings, only the glucuronide of 9α-hydroxyarteether could be detected in plasma (Ramu and Baker, 1997) and only the α-epimer of 9-hydroxyAM was recovered from biliary conjugates, suggesting that the C-9 hydroxyl group is glucuronylated with a high degree of stereoselectively in rat liver.
The biliary metabolite profile of the rats differed somewhat from that of mice, which includes, in addition to DHA glucuronide (the major metabolite) and III to VI, a hydroxyAM glucuronide substituted on the O-methyl carbon (Bell et al., 1998).
Studies in volunteers suggest that AM and DHA–the only known metabolite of AM in humans–are responsible for the greater part of the plasma antimalarial activity (Teja-Isavadharm et al., 1996). Nevertheless, additional contributions to this activity cannot be discounted because 9β-hydroxyarteether, although neither 9α-, 2α-, nor 14-hydroxyarteether, is strongly parasiticidal in vitro (Chi et al., 1991; Ramu and Baker, 1997). Furthermore, 9β-hydroxyarteether glucuronide retains considerable antimalarial activity whereas DHA is essentially inactivated by glucuronylation (Ramu and Baker, 1995,1997). The C-9 isomer of hydroxyAM is said to be active in vitro but the potency of the individual epimers has not been specified (Lee and Hufford, 1990). It is anticipated that the stereoselectivity obtained with 9-hydroxyarteether will apply to 9-hydroxyAM. In this case, male Wistar rats, which apparently achieve complete α-selectivity, i.e., pharmacological deactivation, with AM's C-9 hydroxylated biliary metabolite, are unlikely to form a major pharmacologically active plasma metabolite in addition to DHA.
Acknowledgments
We thank Dr. A. V. Stachulski (Ultrafine Chemicals) for preparing glucuronides of DHA. Bile from isolated perfused rat livers and urine from patients given artesunate were provided by Dr. K. F. Ilett and Dr. K. T. Batty, Dept. of Pharmacology, University of Western Australia. We thank J. O. Bell, S. Newby, and D. J. Boocock for their assistance.
Footnotes
-
Send reprint requests to: B.K. Park, Wellcome Principal Research Fellow, Department of Pharmacology and Therapeutics, University of Liverpool, Ashton Street, Medical School, Liverpool L69 3GE, UK. E-mail: b.k.park{at}liv.ac.uk
-
This study was supported by the B. and V. Ax:son Johnson Foundation, the Wellcome Trust (L.P.D.B.), and F. Hoffmann-La Roche Ltd. (P.M.O'N.). The LC-MS system was purchased and maintained by means of grants from the Wellcome Trust. B.K.P. is a Wellcome Principal Research Fellow. A preliminary communication was presented at the Spring 1998 meeting of the British Pharmacological Society and published in abstract form [Bell et al. (1998)Br J Pharmacol 124 (Proceedings Suppl)42P].
- Abbreviations used are::
- AM
- β-artemether
- [14C]AM
- β-[13-14C]AM
- CID
- collision-induced dissociation
- DHA
- dihydroartemisinin
- DHG
- dehydroglucuronic acid
- LC-MS
- liquid chromatography-mass spectrometry
- LC-MS-MS
- liquid chromatography-tandem mass spectrometry
- UDPGA
- uridine 5′-diphosphoglucuronic acid
- Received March 15, 1999.
- Accepted October 18, 1999.
- The American Society for Pharmacology and Experimental Therapeutics