Glucuronidation represents a major conjugative reaction catalyzed by UDP-glucuronosyltransferases (UGTs). To date, 18 human UGT isoforms, UGT1A1, 1A3, 1A4, 1A5, 1A6, 1A7, 1A8, 1A9, 1A10, 2A1, 2A2, 2B4, 2B7, 2B10, 2B11, 2B15, 2B17, and 2B28, have been identified (Mackenzie et al., 1997; Finel et al., 2005). Most UGT enzymes are expressed in the liver, whereas UGT1A7, 1A8, and 1A10 are reported to be exclusively expressed in extrahepatic tissues (Radominska-Pandya et al., 1999; King et al., 2000; Ritter, 2000; Tukey and Strassburg, 2000). Human UGTs generally exhibit distinct but overlapping substrate specificities, and most substrates are glucuronidated by multiple UGT isoforms (Radominska-Pandya et al., 1999; King et al., 2000; Ritter, 2000; Tukey and Strassburg, 2000; Fisher et al., 2001). Identification of human UGTs responsible for the glucuronidation of existing and novel drugs provides critical information about potential drug-drug interactions (Kiang et al., 2005). Unlike cytochrome P450-mediated biotransformation, UGT reaction phenotyping has attracted less attention. Reasons for this include the limited availability of isoform-selective UGT substrates or antibodies and the relatively low incidence of drug-drug interactions involving UGT-catalyzed reactions in vivo (Williams et al., 2004; Miners et al., 2006). Several compounds, including bilirubin (Bosma et al., 1994), β-estradiol (Senafi et al., 1994), and 17α-ethinylestradiol (Ebner et al., 1993), have been used as in vitro probe substrates for UGT1A1. However, bilirubin is a highly light-sensitive compound and forms multiple glucuronidation products, resulting in difficulties in separation and simultaneous determination (Miners et al., 2004; Zhang et al., 2005). Glucuronidation of β-estradiol and 17α-ethinylestradiol involves contributions of multiple UGT isoforms besides UGT1A1. Accumulating evidence indicates that UGT1A3, 1A4, 1A8, 1A10, 2B7, and 2B15 also metabolize β-estradiol glucuronidation (King et al., 2000; Soars et al., 2004), whereas UGT1A4, 1A8, and 1A9 may also catalyze the glucuronidation of 17α-ethinylestradiol (Cheng et al., 1999; King et al., 2000; Tukey and Strassburg, 2000). The lack of convenient and specific probe substrates for UGT1A1, which appears to have an important role in drug metabolism and drug-drug interactions, remains a problem (Miners et al., 2004).

Etoposide [4′-demethylepipodophyllotoxin-9-(4,6-O-(R)-ethylidene-β-d-glucopyranoside)] (Fig. 1), a commonly used anticancer agent derived from epipodophyllotoxin, is mainly excreted as hydroxyl acid derivatives and glucuronides in humans after oral administration (Fleming et al., 1989). Etoposide glucuronides account for the disposition of 15 to 35% of administered etoposide (D'Incalci et al., 1986; Hande et al., 1988b). Although several possible glucuronidation products of etoposide were found in humans and animals both in vitro and in vivo (Colombo et al., 1985; D'Incalci et al., 1986; Hande et al., 1988a,b), the structures of etoposide glucuronides and related enzyme responsibilities and kinetics involving human UGTs were not systematically or clearly identified. It has been recently reported that UGT1A1 is the specific catalyzing enzyme for the alcoholic glucuronidation of etoposide (Watanabe et al., 2003), but the glucuronidation pathways of etoposide were not completely investigated with all known UGTs, and multiple glucuronidation products were not observed. According to the chemical structure of etoposide, there are three hydroxyl groups (one phenolic and two alcoholic hydroxyls) that can potentially undergo glucuronidation (Fig. 1). To completely understand the glucuronidation pathways of etoposide, we first biosynthesized the phenolic and alcoholic glucuronides of etoposide with human liver microsomes, and preliminarily characterized their structures by liquid chromatography-electrospray ionizationmass spectrometry (LC-ESI-MS) and β-glucuronidase hydrolysis. A sensitive LC-ESI-MS assay for the simultaneous determination of etoposide glucuronides was developed using the biosynthesized standards. The isoforms responsible for the glucuronidation of etoposide were investigated using 12 commercially available recombinant human UGTs (UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15, and 2B17), and human liver and intestinal microsomes. In vitro enzyme kinetics of etoposide glucuronidation in UGT1A1, 1A8, and human liver and intestinal microsomes were also estimated.

Materials and Methods

Chemicals. Etoposide [4′-demethylepipodophyllotoxin-9-(4,6-O-(R)-ethylidene-β-d-glucopyranoside)], 3′-azido-3′-deoxythymidine (AZT; internal standard for the quantification of etoposide and etoposide glucuronides), bilirubin, β-estradiol, 17α-ethinylestradiol, uridine 5′-diphosphoglucuronic acid (UDPGA), magnesium chloride, d-saccharic acid 1,4-lactone (d-SL; specific β-glucuronidase inhibitor), alamethicin, Tris-HCl (pH 7.4) and β-glucuronidase (type B-10 from bovine liver) were purchased from Sigma-Aldrich (St. Louis, MO). Etoposide glucuronides (EGs), including phenolic (EPG) and alcoholic glucuronides (EAG1 and EAG2), were biosynthesized using human liver microsomes, and separated and purified by a preparative HPLC method. Recombinant human UGTs (UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15, and 2B17; BD Supersomes enzymes) were purchased from BD Gentest (Woburn, MA). Pooled human liver microsomes (HLMs) were obtained from In Vitro Technologies (Baltimore, MD). Pooled human intestinal microsomes (HIMs) were purchased from XenoTech (Lenexa, KS). The protein contents of HLMs, HIMs, and recombinant human UGTs were used as described in the data sheets provided by the manufacturers. Enzyme activities of recombinant human UGTs, HLMs, and HIMs were confirmed by HPLC-UV assays as described by the manufacturers with modifications, using 7-hydroxy-4-trifluoromethylcoumarin as a substrate for 11 UGT isoforms (except UGT1A4), HLMs and HIMs, and trifluoperazine as a substrate for UGT1A4. Acetonitrile (ACN; HPLC grade) was obtained from Mallinckrodt (Phillipsburg, NJ). All other chemicals and reagents were of analytical grade.

Glucuronidation of Etoposide. In vitro glucuronidation of etoposide was conducted in an incubation volume of 200 μl. Incubation conditions were initially optimized using HLMs for linear product formation with respect to substrate concentrations (0–2000 μM), microsomal protein concentrations (0.25, 0.5, and 1 mg/ml), incubation time (15, 30, and 60 min), and activation of microsomes by alamethicin (10, 25, 50, and 100 μg/mg protein). The stock solution of etoposide was prepared in dimethyl sulfoxide (DMSO), and the final concentration of DMSO in the incubations was 1%. (Preliminary experiments showed that 1% DMSO in the final incubation has no influence on the glucuronidation of etoposide.) Preliminary experiments indicated that the glucuronide formation of etoposide was linear up to 30 min of incubation time, with 100 μg alamethicin/mg protein and 0.5 mg/ml microsomal protein.

Incubation mixtures containing 0.5 mg of protein/ml of microsomes (recombinant human UGTs, HLMs, and HIMs), 0.1 M Tris-HCl buffer (pH 7.4), and 50 μg alamethicin/mg protein in the final incubation (200 μl) were preincubated on ice for 10 min. After the addition of MgCl₂ (10 mM), d-SL (10 mM), and etoposide, the incubation mixture was preincubated at 37°C for 5 min. The reaction was initiated by the addition of UDPGA (2 mM) and incubated at 37°C in a shaking water bath for 30 min. The reaction was terminated by the addition of 800 μl of ice-cold ACN containing internal standard AZT (6.25 μg). After the removal of protein by centrifugation at 15,000g for 10 min at 4°C, the supernatants were transferred and dried with a stream of nitrogen at 45°C in a water bath. The residue was reconstituted in 125 μl of ACN/0.1% formic acid (20:80 v/v). The reconstituted samples were then centrifuged at 10,000g for 5 min at 4°C, and 20 μl of the final supernatants were introduced for LC-ESI-MS assay. Preliminary experiments indicated that the reconstituted samples containing etoposide, EGs, and AZT in the reconstituting solvent were stable for at least 72 h at 4°C.

Identification of Etoposide Glucuronides. Structural identification of EGs was performed by LC-ESI-MS both in the positive and negative ionization modes with full-scan detection using incubation samples of HLMs. HPLC separation was carried out using an Agilent (Palo Alto, CA) HP 1050 LC system with a guard column (Zorbax RX-C8, 12.5 × 4.6 mm i.d., 5 μm; Agilent) and an analytical column (Zorbax RX-C8, 150 × 12.1 mm i.d., 5 μm; Agilent). HPLC conditions were: mobile phase, A: 0.1% formic acid, B: ACN; gradient elution, 0 to 10 min: 10% B to 50% B, 10 to 12 min: 50% B, 12 to 14 min: 50% B to 10% B, 14 to 18 min: 10% B; column temperature, ambient; flow rate, 0.3 ml/min; injection volume, 20 μl; run time, 18 min. Typical retention times (t_R) of AZT, EPG, EAG1, EAG2, and etoposide under the experimental conditions used were 6.0, 8.0, 9.0, 9.5, and 10.5 min (Fig. 2), respectively.

MS analysis was performed by an Applied Biosystems/MDS Sciex (South San Francisco, CA) API 100 liquid chromatography/mass spectrometry system with a TurboIonspray interface. MS parameters in the positive ESI ionization mode were: ionspray voltage, 4500 V; ionspray temperature, 450°C; orifice voltage, 25 V; focusing ring voltage, 250 V; nebulizer gas, 10 l/min; curtain gas, 8 l/min; dwell time, 1 ms; pause time, 5 ms; scan time, 9 s; scan mode, full scan at the range of 100 to 1200 m/z. MS parameters in the negative ESI ionization mode were: ionspray voltage, –3300 V; ionspray temperature, 450°C; orifice voltage, –50 V; focusing ring voltage, –250 V; nebulizer gas, 10 l/min; curtain gas, 8 l/min; dwell time, 1 ms; pause time, 5 ms; scan time, 9 s; scan mode, full scan at the range of 100 to 1200 m/z.

Biosynthesis of Etoposide Glucuronides. Enzymatic biosynthesis of EGs was conducted using HLMs. In brief, 1 mM etoposide was incubated with HLMs (2 mg protein/ml), 0.1 M Tris-HCl pH 7.4, 10 mM MgCl₂, Brij 35 (0.5 mg/mg protein), 10 mM d-SL, and 2 mM UDPGA twice in 10 ml of final incubations for 7 h at 37°C. The stock solution of etoposide (0.5 M) was prepared in DMSO/methanol (50:50 v/v). The concentration of organic solvent in the final incubation was 0.2%. The reaction was terminated by the addition of 30 ml of ice-cold ACN, twice. After the removal of protein by centrifugation at 15,000g for 15 min at 4°C, the supernatants were pooled and dried with a stream of nitrogen at 45°C. The residue was reconstituted in 3 ml of ACN/1% acetic acid (20:80 v/v) and stored at –20°C until use.

Separation of EGs was performed by a preparative HPLC method using an Econosil C18 column (250 × 10 mm i.d., 10 μm; Alltech, Deerfield, IL) on a Shimadzu (Kyoto, Japan) LC-6A HPLC system with UV detection. HPLC conditions used were as follows: mobile phase, ACN/1% acetic acid (23:77 v/v); elution mode, isocratic; flow rate, 4 ml/min; detection, 254 nm; injection volume, 800 μl; run time, 20 min. The retention times of EPG, EAG1, and EAG2 were 7, 10, and 13 min, respectively. After collection of the fractions of EGs (0.5-min intervals), the individual fractions corresponding to EPG, EAG1, and EAG2 confirmed by LC-ESI-MS were pooled and lyophilized. The lyophilized powders were dissolved in ACN/H₂O (50:50 v/v) and used as the standards for quantification. The purities of synthesized EGs estimated by both LC-ESI-MS and HPLC-UV (254-nm) assays were at least 97%.

The concentrations of synthesized EGs were estimated by LC-ESI-MS after β-glucuronidase cleavage. Briefly, aliquots of the diluted EG standards were incubated with β-glucuronidase (10,000 U/ml in the final incubation) in 0.1 M sodium acetate buffer (pH 5) at 37°C for 24 h. Preliminary experiments indicated that, under the enzymatic hydrolysis conditions tested, all forms of EGs could be completely cleaved to the parent compound, etoposide. The generated etoposide was quantified by LC-ESI-MS using an etoposide calibration curve, and then the individual concentrations of EG standards were estimated according to the molecular weight ratios of EGs to etoposide.

Quantification of Etoposide Glucuronides. Quantification of EGs was carried out by LC-ESI-MS in the negative ionization mode with selective ion monitoring (SIM) detection. HPLC conditions were the same as described in the identification of etoposide glucuronides. MS parameters used for quantification were as follows: ionspray voltage, –3300 V; ionspray temperature, 450°C; orifice voltage, –30 V; focusing ring voltage, –200 V; nebulizer gas, 10 l/ml; curtain gas, 8 l/ml; dwell time, 300 ms; pause time, 5 ms; scan time, 1 s; scan mode, SIM with [M – H]^– for EGs (m/z 763), etoposide (m/z 587), and AZT (m/z 266), respectively. Calibration curves were set up using the biosynthesized and purified standards of EGs. Standard solutions containing EPG, EAG1, and EAG2 were spiked into the incubation media and then treated as described in the glucuronidation of etoposide. Concentrations of EGs in the samples were estimated with 1/x² weighted least-squares regression equations derived from the peak area ratios of EGs to that of AZT.

UGT Reaction Screening of Etoposide. UGT reaction screening of etoposide was conducted with HLMs, HIMs, 12 commercially available recombinant human UGTs, and insect cell control (negative control) microsomes using a constant amount of microsomal protein (0.5 mg/ml) and a fixed concentration of etoposide (250 μM). Incubation conditions were the same as described in the glucuronidation of etoposide. Incubations without UDPGA or without substrate were also carried out. The incubation samples were treated as described in the glucuronidation of etoposide, and then analyzed by LC-ESI-MS.

Enzyme Kinetics for the Glucuronidation of Etoposide. Apparent enzyme kinetic parameters (K_m and V_max) for in vitro glucuronidation of etoposide were measured with recombinant human UGT1A1, 1A8, HLMs, and HIMs, using various concentrations of etoposide (0–2000 μM) at fixed concentrations of microsomal proteins (0.5 mg/ml), MgCl₂ (10 mM), alamethicin (50 μg/mg protein), d-SL (10 mM), and UDPGA (2 mM) in a 200-μl final incubation. Incubation conditions and sample preparation were similar to those of glucuronidation of etoposide described previously. The concentrations of EGs were quantified by LC-ESI-MS. Apparent K_m and V_max values were estimated by fitting the duplicate experimental data to a Michaelis-Menten equation: V = (V_max × S)/(K_m + S), or a substrate inhibition equation (Houston and Kenworthy, 2000): V = (V_max × S)/(K_m + S + S²/K_si), where K_m is the Michaelis-Menten constant, V_max is the maximum velocity, S is the substrate concentration, and K_si is the substrate inhibition constant, by a nonlinear least-squares regression using SigmaPlot (SPSS Inc., Chicago, IL). The apparent kinetic parameters were reported as mean ± S.E. from individual duplicates.

Inhibitory Effects of Typical Probe Substrates on Etoposide UGT Activity in HLMs. Bilirubin (Bosma et al., 1994), β-estradiol (Senafi et al., 1994; Soars et al., 2003), and 17α-ethinylestradiol (Ebner et al., 1993), UGT1A1 substrates commonly used, were selected to investigate their potential inhibitory effects on etoposide UGT1A1 activity. The stock solutions of all substrates tested were prepared in DMSO, and the final concentration of DMSO in the incubation was 1%. The formation rates of etoposide glucuronides in HLMs (0.5 mg protein/ml) at a fixed concentration of etoposide (250 μM) with and without inhibitor (0–2000 μM) were determined by LC-ESI-MS. Incubation conditions were the same as those described previously in glucuronidation of etoposide, and the IC₅₀ values were estimated graphically.

Results

Structural Characterization of Etoposide Glucuronides. Three isomeric glucuronidation products of etoposide were found in the incubations with HLMs and UGT1A1 (Fig. 2) and identified by LC-ESI-MS both in the positive and negative ion modes. In the positive ion mode, etoposide and its glucuronides exhibited relatively abundant and multiple fragments, including protonated podophyllotoxin ion (m/z 383, the loss of glucose or sugar moiety from etoposide) and its subfragmentation ion (m/z 229), and several adducts of ammonium, sodium, and potassium (Fig. 3). The proposed in-source MS fragmentation pathway of etoposide in the positive ion mode is shown in Fig. 4, which is consistent with published reports (Chen and Uckun, 2000; Pang et al., 2001). In the negative ion mode, all etoposide glucuronides exhibited a strong and exclusive base peak corresponding to the deprotonated intact glucuronide molecule ion ([M – H]^–, m/z 763), whereas etoposide showed the deprotonated podophyllotoxin ion (m/z 381, the loss of glucose or sugar moiety from etoposide) as the predominant peak (Fig. 5). The diagnostic fragmentations for the formation of etoposide glucuronides were based on the fragmentation ion at m/z 589 ([M + H – 176]⁺; the neutral loss of glucuronide moiety, 176 Da) and several adducts of ammonium ([M + NH₄]⁺, m/z 782), sodium ([M + Na]⁺, m/z 787), and potassium ([M + K]⁺, m/z 803) in the positive ion mode, and the fragmentation ion at m/z 587 ([M – H – 176]^–; the neutral loss of glucuronide moiety, 176 Da); the deprotonated intact glucuronide molecule ion ([M – H]^–, m/z 763); and the loss of H₂O from [M – H]^– (m/z 745) in the negative ion mode, respectively. In addition, enzyme hydrolysis showed that all conjugation products of etoposide could be cleaved into the parent compound by β-glucuronidase, further confirming the formation of glucuronide conjugates. The distinguishing fragments for the phenolic (EPG) and alcoholic glucuronides (EAG1 and EAG2) of etoposide were elucidated by the presence (phenolic glucuronide) or absence (alcoholic glucuronide) of the subfragmentation ions of m/z 435 and 247 resulting from the loss of glucuronide moiety and then the loss of phenolic side chain (Figs. 3 and 4b), and subsequently, the loss of glucose or sugar moiety (Figs. 3 and 4f) in the positive ion mode, and the distinctive cleavage of glucuronic acid moiety (m/z 175 and 113 for EPG; m/z 193, 175, and 113 for EAG1 and EAG2) in the negative ion mode (Figs. 5 and 6), respectively. Because the C–O bond to the aliphatic chain of alcoholic glucuronides is more easily cleaved than the C–O bond to the aromatic ring of phenol-linked glucuronides (Fenselau and Johnson, 1980; Niemeijer et al., 1991), the MS spectra of EAG1 and EAG2 were very similar and did exhibit the deprotonated glucuronic acid ion (m/z 193), whereas EPG did not. These different in-source MS fragmentation pathways of glucuronic acid moieties in etoposide phenolic and alcoholic glucuronides in the negative ion mode (Fig. 6) were in agreement with previous findings about phenolic and alcoholic glucuronides (Niemeijer et al., 1991; Gu et al., 1999; Levsen et al., 2005). Additional evidence for the identification of the positional isomers of phenolic and alcoholic glucuronides is their different chromatographic retentions on reversed-phase HPLC, since phenolic glucuronides (e.g., EPG, t_R = 8.0 min) are generally more hydrophilic and, consequently, have earlier elution than those of alcoholic glucuronides (e.g., EAG1 and EAG2, t_R = 9.0 and 9.5 min, respectively). The similar examples include estradiol 3- and 17-glucuronides (Alkharfy and Frye, 2002), ethinylestradiol 3- and 17-glucuronides (Ebner et al., 1993), and the phenolic and alcoholic glucuronides of denopamine (Kaji and Kume, 2005) and labetalol (Niemeijer et al., 1991). However, the conjugation sites of the two alcoholic glucuronides of etoposide could not be individually identified by MS methods, and this needs to be further elucidated. In addition, in this study, the formation of di- or tri-glucuronides of etoposide was not observed under the experimental conditions used.

Quantification of Etoposide Glucuronides. Experiments indicated that all etoposide glucuronides had higher MS responses in the negative ionization mode than that in the positive ionization mode. Quantification of EGs was then performed with SIM detection in the negative ionization mode. The limit of detection was 0.5 ng/ml for all forms of EGs with injection volume of 20 μl. The quantitative linear ranges were 2 to 1500 ng/ml for EPG, and 1 to 1000 ng/ml for EAG1 and EAG2. The intraday precisions (relative standard deviation, n = 4) were 1.5 to 5.6%, 1.0 to 6.4%, and 1.1 to 7.0% for EPG, EAG1, and EAG2, respectively. The interday precisions (relative standard deviation, n = 4) were 1.5 to 6.8%, 1.2 to 4.0%, and 2.9 to 13% for EPG, EAG1, and EAG2, respectively. Experimentally, the most important MS parameters for the quantification of intact glucuronides were the cone voltages, including orifice and focusing ring voltages, in this study. Increasing cone voltages significantly enhanced MS intensities of molecular ions of etoposide and EGs (data not shown). However, at high cone voltages (>–40 and –300 V of orifice and focusing ring voltages, respectively), the MS intensity of EGs significantly decreased, because some of the intact glucuronides underwent insource dissociation. Increasing the ionspray voltage (from –2000 to –3300 V) resulted in a slight increase of MS responses of etoposide and EGs, but the influence was much less than that of cone voltages. Changing the ionspray temperature (from 250–450°C) showed no or little effect on the determination sensitivity. No significant in-source thermal degradation of EGs was observed at high ionspray temperature (450°C) when the optimal cone voltages were used in this study. It is noted that the MS responses of EPG, EAG1, and EAG2 were apparently different from each other at the same concentrations (EAG2 > EPG > EAG1), and also quite different from their parent compound, etoposide. The cone voltage-dependent sensitivities for the quantification of intact etoposide glucuronides by LC-ESI-MS are consistent with the analysis of other glucuronides in biological samples (Yan et al., 2003). Collectively, the effects of cone voltages on the determination of sensitivity by LC-ESI-MS are chemical structure- and molecular weight-dependent, since relatively low cone voltages are beneficial to the quantification of intact glucuronides (e.g., EPG, EAG1, and EAG2), whereas high cone voltages are helpful for the determination of the parent or unconjugated compounds (e.g., etoposide).

UGT Reaction Screening of Etoposide. Incubation of etoposide (250 μM) with 12 commercially available recombinant human UGTs, insect cell control (negative control) microsomes, HLMs, and HIMs demonstrated that three UGT isoforms (UGT1A1, 1A8, and 1A3), HLMs, and HIMs catalyzed the glucuronidation of etoposide in vitro (Fig. 7). No glucuronidation of etoposide was observed with the other nine UGT isoforms tested and the insect cell control microsomes. Three etoposide glucuronides were found in vitro, with the predominant form of etoposide glucuronide being the phenolic glucuronide (EPG). Alcoholic glucuronides (EAG1 and EAG2) were the minor metabolites, with an approximately 8 to 10% glucuronidation rate of EPG formation (Figs. 2 and 7). Among the three UGT isoforms that catalyzed the glucuronidation of etoposide, UGT1A1 had the highest glucuronide formation rates (62, 7.1, and 6.9 pmol/min/mg protein for EPG, EAG1, and EAG2, respectively). The activities of UGT1A8 (6.1, 0.37, and 3.7 pmol/min/mg protein for EPG, EAG1, and EAG2, respectively) and UGT1A3 (0.60, 0.64, and 0.05 pmol/min/mg protein for EPG, EAG1, and EAG2, respectively) were much lower than those of UGT1A1.

Enzyme Kinetic Parameters. Apparent enzyme kinetic parameters were estimated using various concentrations of etoposide (0–2000 μM) with recombinant human UGT1A1 and 1A8, HLMs, and HIMs. The kinetics of UGT1A1, HLMs, and HIMs fitted to the Michaelis-Menten equation at 5 to 2000 μM etoposide (Fig. 8; Table 1). For the formation of EPG catalyzed by UGT1A1, the V_max, K_m, and V_max/K_m values were 124 pmol/min/mg protein, 285 μM, and 0.44 μl/min/mg protein, respectively. In comparison with UGT1A1, the V_max of EPG with HLMs (110 pmol/min/mg protein) was very similar, but its K_m (530 μM) was approximately 2-fold higher, resulting in a 2-fold lower value of V_max/K_m (0.21 μl/min/mg protein). The formation rate of EPG in the incubation containing HIMs (54.4 pmol/min/mg protein) was approximately 50% of that of HLMs and UGT1A1. The enzyme activities toward the formation of alcoholic glucuronides of etoposide (EAG1 and EAG2) by UGT1A1, HLMs, and HIMs were much lower, with 4- to 13-fold lower V_max values than that of EPG (Figs. 2 and 7; Table 1). The K_m of EAG1 with UGT1A1 (230 μM) was slightly lower than that of EPG (285 μM), whereas the K_m values of EAG1 with HLMs (330 μM) and HIMs (324 μM) were very similar. The K_m values of EAG2 with UGT1A1 (753 μM), HLMs (1010 μM), and HIMs (922 μM) were approximately 2- to 3-fold higher than that of EPG and EAG1.

The kinetics of etoposide glucuronidation in UGT1A8 at 5 to 500 μM etoposide fitted to the typical Michaelis-Menten kinetics, with the V_max and K_m values of 13.4 pmol/min/mg protein and 65 μM for EPG, and 5.28 pmol/min/mg protein and 83 μM for EAG2 (Fig. 9; Table 2), respectively. However, when the etoposide concentration exceeded 500 μM, the formation rates of etoposide glucuronides (EPG and EAG2) gradually decreased with the increase of etoposide concentration, exhibiting a substrate inhibition kinetics (Fig. 9). When the apparent enzyme kinetic parameters were estimated by fitting to the substrate inhibition kinetics at 5 to 2000 μM etoposide, the V_max, K_m, and K_si (substrate inhibition constant) values were 19.1 pmol/min/mg protein, 113 and 1004 μM for EPG, and 6.89 pmol/min/mg protein, 125 and 1421 μM for EAG2 (Table 2), respectively. The enzyme kinetics for the formation of EAG1 catalyzed by UGT1A8 and the kinetics of UGT1A3 for etoposide glucuronidation were not estimated because of their very low activities.

Inhibition of Etoposide Glucuronidation by UGT1A1 Probe Substrates. Inhibitory effects of three commonly used UGT1A1 probe substrates, bilirubin, β-estradiol, and 17α-ethinylestradiol, on etoposide UGT activity in HLMs are shown in Fig. 10. The estimated IC₅₀ values are summarized in Table 3. The substrates tested exhibited different inhibitory influence on the formation of individual etoposide glucuronides in vitro. Bilirubin, the putatively specific substrate of UGT1A1, strongly inhibited the formation of EPG (IC₅₀ = 75 μM) and EAG1 (IC₅₀ = 54 μM), and significantly inhibited the formation of EAG2 (IC₅₀ = 140 μM). 17α-Ethinylestradiol strongly inhibited the formation of EAG2 ((IC₅₀ = 42 μM) but showed very similar and moderate inhibition on EPG (IC₅₀ = 200 μM) and EAG1 (IC₅₀ = 210 μM). β-Estradiol also exhibited strong inhibition on EPG (IC₅₀ = 68 μM) and EAG2 (IC₅₀ = 86 μM). Interestingly, the inhibitory effect of β-estradiol on the formation of EAG1 was not so significant in comparison with the other two etoposide glucuronide isomers, even at very high concentration of β-estradiol (∼80% at 2000 μM). Collectively, these inhibitory results suggested that etoposide is the selective substrate of UGT1A1.

Discussion

Human UGTs generally exhibit distinct but overlapping substrate specificities, and most substrates are glucuronidated by more than one isoform (Radominska-Pandya et al., 1999; King et al., 2000; Ritter, 2000; Tukey and Strassburg, 2000; Fisher et al., 2001). Due to the lack of specific antibodies for human UGT isoforms, it is important to alternatively explore some selective substrates for the evaluation of the responsibility of individual UGT isoforms. Currently, several compounds, including bilirubin (Bosma et al., 1994), β-estradiol (Senafi et al., 1994), and 17α-ethynylestradiol (Ebner et al., 1993), have been used as the probe substrates for UGT1A1 in vitro. Bilirubin is putatively considered as the most specific substrate for UGT1A1 (Bosma et al., 1994; Burchell et al., 1995; King et al., 2000). However, simultaneous and accurate determination of bilirubin and its multiple glucuronidation isomers (including one di-glucuronide, and two cis/trans and two positional isomers of mono-glucuronides) is difficult because of the light sensitivity of bilirubin, and the instability or interconversion of mono- and di-glucuronide of bilirubin in vitro and in vivo (Jansen et al., 1977; Chowdhury et al., 1981; Gordon et al., 1984; Miners et al., 2004; Zhang et al., 2005). In addition, it has been reported that bilirubin may also inhibit UGT1A4 (Ghosal et al., 2004). UGT1A1 predominantly catalyzes the phenolic glucuronidation of β-estradiol (Senafi et al., 1994; Soars et al., 2003) at position 3 with atypical kinetics, whereas UGT2B7 mainly catalyzes the alcoholic glucuronidation of β-estradiol at position 17 (Gall et al., 1999). However, UGT1A8, 1A10, and 1A3 have also been reported to catalyze the glucuronidation of β-estradiol-3 with significant activities (Soars et al., 2004), whereas UGT1A3, 1A4, 1A8, 1A10, and 2B7 may also be involved in the glucuronidation of β-estradiol-17 (Soars et al., 2004). Although 17α-ethinylestradiol glucuronidation is primarily catalyzed by UGT1A1 (Ebner et al., 1993), other UGT isoforms (e.g., UGT1A4, 1A8, and 1A10) may also metabolize the glucuronidation of 17α-ethinylestradiol at lower but significant activities (Cheng et al., 1999; King et al., 2000; Tukey and Strassburg, 2000). This overlapping substrate specificity for “probes” of UGT1A1 complicates the interpretation of in vitro data, particularly in human intestinal tissues where relatively high levels of UGT1A8 and 1A10 are expressed (Cheng et al., 1999; Radominska-Pandya et al., 1999; Tukey and Strassburg, 2000). Since multiple UGT enzymes catalyze the glucuronidation of β-estradiol and 17α-ethinylestradiol, caution should be taken when these two compounds are chosen as the probe substrates of UGT1A1.

It has been reported that UGT1A1 only catalyzes the alcoholic glucuronidation of etoposide (Watanabe et al., 2003), but the glucuronidation pathways of etoposide were still unclear, considering its potential multiple glucuronidation products (Fig. 1). In addition, since etoposide glucuronide standards were not used, true Michaelis-Menten parameters were not determined. In this study, we reported that three positional isomeric glucuronides of etoposide (Fig. 2), including one phenolic (EPG) and two alcoholic glucuronides (EAG1 and EAG2), were formed in the incubations containing recombinant human UGTs (UGT1A1, 1A3, and 1A8), HLMs, and HIMs in vitro (Fig. 7). The main glucuronidation product of etoposide was the phenolic glucuronide (EPG), whereas the two alcoholic glucuronides (EAG1 and EAG2) were the minor metabolites. The formation of multiple etoposide glucuronides in vitro found in this study is consistent with the previous in vivo results observed in rats and rabbits after administration of etoposide (Colombo et al., 1985; Hande et al., 1988a), although the structures of etoposide glucuronides were not completely characterized in the previous publications. In this report, the structures of etoposide glucuronides were preliminarily identified by LC-ESI-MS both in the positive and negative ionization modes, and confirmed by β-glucuronidase cleavage, although the conjugation sites of the two alcoholic glucuronides of etoposide need to be further distinguished. Our data also indicated that, at the same concentrations, etoposide and its three glucuronides had different MS responses (EAG2 > EPG > EAG1 > etoposide), when measured in the negative ionization mode. Therefore, accurate quantification of etoposide glucuronides using the biosynthesized standards is necessary. UGT reaction screening (Fig. 7) and enzyme kinetic study (Figs. 8 and 9; Tables 1 and 2) indicated that three UGT isoforms, UGT1A1, 1A3, and 1A8, were responsible for the formation of etoposide glucuronides in vitro. UGT1A1 is the principal enzyme catalyzing etoposide glucuronidation, whereas the enzyme activities of UGT1A3 and 1A8 were only 1 and 10% of UGT1A1, respectively, although the individual contribution of the related UGT isoforms needs to be further confirmed and quantitatively compared considering the expressed levels among the different isoforms. The Eadie-Hofstee plots for HLMs and HIMs (Fig. 8, D, E, and F) showed the involvement of a single UGT isoform, UGT1A1, in the etoposide glucuronidation reactions, suggesting that the contributions of UGT1A8 or 1A3 might be minor. In addition, inhibitory study also indicated that etoposide is the selective substrate of bilirubin glucuronidating isoform, UGT1A1 (Fig. 10; Table 3).

In conclusion, we demonstrated that three isomeric glucuronides of etoposide were formed in vitro, including one phenolic and two alcoholic glucuronides. UGT1A1 is the principal enzyme responsible for the formation of etoposide glucuronides in vitro, mainly in the form of phenolic glucuronide. Considering that the catalyzing activities of UGT1A8 and 1A3 for the glucuronidation of etoposide are much lower than that of UGT1A1, and UGT1A8 is mainly expressed in extrahepatic tissues, etoposide could be used as an alternative or better selective substrate for human UGT1A1 instead of other nonspecific probes in hepatic systems.