Introduction

Top Abstract Introduction Materials and Methods Results and Discussion References

The cytochrome P450 (P450²) superfamily of enzymes is encoded by more than 40 genes in humans (Nelson et al., 1996), and at least 12 members in the first three families (CYP1-3) are involved in xenobiotic metabolism and drug-drug interactions (Wrighton and Stevens, 1992). As the biochemistry and enzymology of this enzyme system have evolved and all of the human xenobiotic-metabolizing isoforms have been cloned and expressed, selective substrates have been identified that will yield information about catalytic activity of a specific form when present in a crude mixture (Hickman et al., 1998). Indeed, specific probes have been used in systems such as human liver microsomes, liver slices (VandenBranden et al., 1998), hepatocytes (Li, 2001), and in vivo (Streetman et al., 2000). For example, specific metabolite formation from warfarin and testosterone and metabolism of various alkoxyresorufins have been used as in vitro probes for various P450 forms (Rettie et al., 1995). The enzymology of other drug-metabolizing enzyme systems, such as the UDP-glucuronosyltransferases (UGTs) and sulfotransferases, has not been as well characterized to date, and thus in many cases, a nonspecific or generic substrate has been used (VandenBranden et al., 1998; Hashemi et al., 1999).

7-Ethoxycoumarin (7-EC) is one of the most commonly used in vitro metabolic probes to date. It has been used in hepatocytes and liver slices for in vitro predictions of in vivo hepatic clearance (Carlile et al., 1999), as a positive control for metabolic lability of in vitro systems (VandenBranden et al., 1998; Hashemi et al., 1999), and in addition as a probe substrate for specific P450s (Yamazaki et al., 1996; Shimada et al., 1999). Oxidative deethylation of 7-EC by CYP1A2 (low-K_m component) and by CYP2E1 and 2B6 (high-K_m components) (Yamazaki et al., 1996) generates 7-hydroxycoumarin (7-HC; umbelliferone), which is then susceptible to conjugation by as yet uncharacterized UGT forms when present with the appropriate cofactor. Although sequential metabolism of 7-EC to 7-HC glucuronide has been routinely observed in hepatocytes and liver slices, minimal examination of its coupled oxidation-glucuronidation has been examined in microsomes (Matsubara et al., 1982). This is primarily due to the differential localization of the UGT and P450 enzyme systems on the lumen and surface of the microsomal membrane, respectively. The microsomal membrane imparts a diffusional barrier to the luminal UGT enzyme, and specialized treatments, typically detergents, facilitate cofactor, substrate, and product diffusion but often inhibit P450 activity. The pore-forming peptide alamethicin has recently been applied to microsomal glucuronidation and been shown not to attenuate P450 activity, thus providing a practical means for accessing metabolite generation via sequential oxidation-glucuronidation reactions (Fisher et al., 2000).

The current studies were undertaken to characterize the oxidative-conjugative microsomal metabolites of 7-EC with LC-NMR. LC-NMR, the hyphenation of chromatographic separation with magnetic resonance structural characterization, dates back to the 1970s (Bayer et al., 1979). However, this methodology has only recently come of age, as articles attest to growing application in the field of drug metabolism (Lindon et al., 1996, 1997). Most recently, challenging regiochemistry problems related to acetylenic metabolism (Mutlib et al., 1999, 2000), NIH shift metabolites (Dear et al., 2000), and quinoline oxidation/glucuronidation (Ehlhardt et al., 1998) have been deciphered with LC-NMR. This article reports the application of LC-NMR and LC-MS for the characterization of two sets of coupled phase I and phase II metabolites formed from human liver microsomes perforated with alamethicin and supplemented with the substrate 7-EC and the cofactors NADPH and UDP-glucuronic acid (UDPGA), respectively.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results and Discussion References

Chemicals. UDPGA, saccharic acid-1,4-lactone, alamethicin, trifluoroacetic acid, 7-EC, 7-HC, and coumarin were purchased from Sigma Chemical Co. (St. Louis, MO). HPLC-grade acetonitrile was obtained from Burdick and Jackson (Muskegon, MI). 7-Hydroxycoumarin glucuronide and 7-hydroxycoumarin sulfate were obtained from GENTEST (Woburn, MA) or Salford Ultrafine Chemicals and Research (Manchester, UK). D₂O (99.9% D) and CD₃CN (96-97% D) were obtained from Cambridge Isotopes Laboratories (Cambridge, MA).

Liver Specimens. Human liver specimens were obtained, and microsomes were prepared as described previously (Fisher et al., 2000). Microsomes from a mixture of nine human livers (HL samples A, B, C, D, F, G, H, L, and M) were used for all studies.

Ethoxycoumarin Metabolism. For microsomal metabolism of 7-EC, 0.1 to 0.3 mg of human liver microsomes, 0.1 M sodium phosphate buffer, pH 7.4, 50 µg of alamethicin/mg of microsomal protein, 1 mM MgCl₂, 5 mM saccharic acid-1,4-lactone, and 250 µM 7-EC were incubated with 1 mM NADPH, with and without 5 mM UDPGA, in a final volume of 200 µl for 15 to 45 min. Reactions were stopped by the addition of 50 µl of 5% (v/v) HCl and 20 µl of 0.2 mM coumarin as internal standard. Blank incubations were performed without cofactor, and cofactor was added after quenching the incubation. Samples were centrifuged to pellet-precipitated material, and 25 to 30 µl was injected for HPLC analysis. Analysis was performed on a Shimadzu (Kyoto, Japan) HPLC system, equipped with two LC-10AD pumps, an SCL-10A system controller, an SIL-10A autoinjector, an SPD-10A UV/Visible detector set to 320 nm, and a 3-µm C₁₈(3) Luna 100 × 4.6 column (Phenomenex, Torrance, CA). In some chromatographic runs, a Shimadzu RF-10A fluorescence detector set to excitation and emission wavelengths of 323 and 463 nm, respectively, was used to detect analytes. The mobile phase was 0.05% trifluoroacetic acid in water (solvent A) and 0.05% trifluoroacetic acid in acetonitrile (solvent B). Initial conditions were 88% solvent A and 12% solvent B. Analytes were eluted with a linear gradient to 65% solvent B over 12.5 min. Metabolite formation was quantitated by comparing peak area ratios (metabolite/internal standard) in incubations with ratios obtained from a standard curve containing known amounts of 7-HC (M3) and 7-HC glucuronide (M4). Standard curve correlation coefficients (r²) were 0.99. Quantities of the unknown metabolites M1 and M2 were estimated using standard curves for M3 and M4, respectively. Incubations for LC-NMR analysis were performed as described above; however these incubations were stopped with an equal volume of cold acetonitrile, lyophilized overnight, reconstituted in mobile phase (90% solvent A/10% solvent B), and centrifuged to remove insoluble material.

	Results and Discussion

Top Abstract Introduction Materials and Methods Results and Discussion References

In earlier studies on the oxidation/conjugation of 7-EC in microsomes, alamethicin was applied using 50 µg/mg of microsomal protein (Fisher et al., 2000). When supplemented with NADPH as the sole cofactor, 7-EC was formed without impairment of P450 function, whereas in the presence of both NADPH and UDPGA as cofactors sequential dealkylation and glucuronidation occurred. Characterization of the product profile by HPLC-UV allowed not only the known products 7-HC (M3) and 7-HC glucuronide (M4) [at retention times (t_R) of 7.9 and 4.0 min, respectively] to be quantitated (Table 1) but also unearthed two unexpected microsomal metabolites, M1 and M2 (t_R at 12.0 and 8.9 min, respectively) (Table 1).

View this table: [in this window] [in a new window]   TABLE 1 Characteristics of microsomal metabolites of 7-ethoxycoumarin with various treatments Metabolites M3 and M4 were identified as 7-HC and 7-HC glucuronide, respectively. Formation rates for metabolites M1 and M2 were estimated using the standard curves for 7-HC and 7-HC glucuronide, respectively. Some of these data have been published previously (Fisher et al., 2000).

The HPLC-UV and HPLC-fluorescence properties of M1 and M2 suggested that they were metabolites related to 7-EC. All four metabolites were fluorescent, although M1 and M4 were minor peaks with this detection method. When quantities of M1 and M2 were estimated by HPLC-UV using standard curves for 7-HC and 7-HC glucuronide, respectively, they appeared to be similar in abundance to the known metabolites (Table 1). LC-MS data supported M1 and M2 being the result of sequential hydroxylation and glucuronidation. M+H⁺ ions of M1 and M2 at m/z 207 and 383, respectively, were 16 amu (hydroxylation) and 192 amu (hydroxylation/glucuronidation) higher than that of the parent drug. By comparison, the M+H⁺ for M3 (M+H⁺ at m/z 163) and M4 (M+H⁺ at m/z 339) were consistent with metabolites formed by sequential oxidative deethylation and glucuronidation, respectively. Additionally, the LC-MS source-induced loss of 176 amu from the putative glucuronide M2 afforded a fragment consistent with M+H⁺ of the corresponding aglycone M1. Collision-induced dissociation of the M+H⁺ of M2 led to a very facile loss of 176 amu, whereas collision-induced dissociation of the M+H⁺ of M1 yielded loss of CO as the only discernible fragment. Unfortunately, LC-tandem mass spectrometry gave no information on molecular connectivity. Taken together, these data suggest that M1 arises through hydroxylation of 7-EC without dealkylation, and M2 arises via glucuronidation of M1. However, ascertaining the regiochemistry of these biotransformations necessitated LC-NMR (Table 2).

Microsomal incubations for analysis by HPLC-NMR were supplemented with NADPH alone (M1 and M3 only metabolites) or NADPH and UDPGA (M2 and M4 major metabolites). As illustrated in Fig. 1, the NMR spectrum of M1 shows the absence of the resonance for H3 and the collapse of the H4 doublet to a singlet compared with the parent, 7-EC (Fig. 1). Furthermore, relative to the parent, there is an up-field shift in the H4 resonance (0.78 ppm) and, to a lesser extent, the H5 resonance 0.13 ppm). There is little change in the chemical shifts of H6 and H8. Interpreted in conjunction with LC-MS data, the NMR properties of M1 indicate a 3-hydroxy substituent on 7-EC.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   LC-NMR spectra of 7-EC and metabolites M1 and M2. *, residual methanol from the sample preparation; **, residual protio solvents of acetonitrile and water.

The NMR spectrum of M2 (Fig. 1) supported glucuronidation at position 3, with the absence of resonance at 6.30 ppm (compared with parent compound) and the presence of resonances at 3.3 to 4.2 ppm, attributable to protons H2' to H5'. Of the resonances that were retained in M2 compared with the parent, the change in H4 was the most pronounced, its up-field shift (0.39 ppm) considered attributable to the shielding glucuronide ether. Another diagnostic feature of M2 was the isolated resonance ascribed to H6', which, occurring at 5.10 ppm, was characteristic for the anomeric proton of glucuronides (Kaspersen and Van Boeckel, 1987). Homonuclear decoupling experiments yielded further support to position-3 substitution by verifying J-coupling between the chemical shifts at 7.52 and 6.99 ppm (Fig. 2). Saturation of the resonance in the region of 6.99 ppm resulted in the collapse of the doublet at 7.54 ppm to a singlet. This collapsed singlet was assigned to H5, and the other singlet at 7.52 ppm was assigned to H4. Furthermore, in the converse experiment (data not shown), saturation of the resonance at 7.52 to 7.54 ppm deconvoluted the partially overlapping peaks at 6.98 to 6.99 ppm into the apparent single coincident resonance assigned to H6 and H8.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Partial LC-NMR spectrum of metabolite M2 (6.7 to 7.7 ppm) illustrating selective decoupling at the frequency range associated with H6 and H8.

The metabolites M3 and M4, identified by LC-MS as the O-dealkylated and phenolic glucuronide, respectively (vide supra), were characterized by the loss of resonances H9 and H10 (Fig. 3). Furthermore, the up-field shift of H6 and H8 for M3 (0.05 and 0.1 ppm relative to the parent, respectively) can be attributed to the stronger shielding of the unmasked phenol in M3 compared with the parent ether. The resonance for H6 of M3 occurred as a doublet of doublets (J_5,6 = 8.6 Hz; J_6,8 = 2.3 Hz) and H8 as a doublet (J_8,6 = 2.3 Hz). Similar coupling constants were observed for M4, as was the loss of resonances for H9 and H10. Typical glucuronide resonances were observed also (H2'-H6'). These data support M4 as a glucuronide attached via the ether linkage at position 7. Interestingly, the anomeric proton resonance (H6') of M4 occurred as two overlapped doublets. This may be attributed to some conformational averaging of two states that is slow on the NMR timescale rather than additional J-coupling (H6' is only J-coupled to H5'). It is not known why this is observed for M3 compared with M1, and further experiments are required to investigate this phenomenon.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   LC-NMR spectra of metabolites M3 and M4. *, residual methanol from the sample preparation; **, residual protio solvents of acetonitrile and water.

The formation of M1 and M2 from 7-EC was unexpected and clearly illustrative of the coupled oxidation-glucuronidation activity of alamethicin-treated microsomes. The proposed pathway for the formation of these four metabolites is shown in Fig. 4. During the course of P450 catalysis, the heme-bound intermediate proceeding from olefin -electron abstraction from 7-EC could conceivably give rise to either an epoxide or a ketone (Liebler and Guengerich, 1983) that tautomerizes to the more stable allylic alcohol. It is noteworthy that in the P450-mediated oxidation of coumarin, the 3,4-epoxide is not an intermediate in the formation of 3-hydroxycoumarin (Born et al., 1997). Interestingly, incubation of 250 µM 7-EC with cryopreserved human hepatocytes, thawed and in suspension, generated significant quantities of 7-hydroxycoumarin glucuronide (data not shown). However, no evidence for 3-hydroxy-derived metabolites was observed. It is clear that significant differences exist between coupled oxidation-glucuronidation in microsomes and hepatocytes for 7-EC, and further studies are necessary to elucidate these differences.

View larger version (10K): [in this window] [in a new window]   Fig. 4.   Proposed metabolic pathway for 7-EC in alamethicin-treated microsomes supplemented with NADPH and UDPGA.

In the work presented here, LC-NMR was applied to the characterization of 7-HC (M3) and 7-HC glucuronide (M4), previously reported as metabolites of 7-EC in alamethicin-treated microsomes (Fisher et al., 2000). More importantly, however, LC-NMR proved to be a powerful tool in the structural elucidation of metabolites M1 and M2 derived from a novel 3-hydroxylation pathway in human microsomes. In fact, metabolites from the latter pathway were apparently as abundant as those formed from dealkylation as judged by HPLC-UV. Metabolites from 7-alkoxycoumarins that proceed via routes other than dealkylation have been observed previously. The metabolism of 7-butoxycoumarin has been shown to proceed predominantly through 3- and 4-hydroxylation pathways, although this was performed with rat CYP2B1 and these alternative metabolites were not seen with 7-EC (Kobayashi et al., 1998). Also, 6-hydroxylation of 7-EC by rat liver P450 has been reported, occurring as a consequence of metabolic switching upon deuteration of the ethoxymethylene of 7-EC (Harada et al., 1984). Alternative unidentified metabolites have been observed after incubation of 7-EC with rat liver slices (Ball et al., 1996), and apparent lactone ring oxidation was observed in guinea pig and dog liver slices (Terada et al., 1996). However, to our knowledge neither 3-hydroxy nor 3-hydroxyglucuronide formation from 7-EC has been previously reported from human liver microsomes.

The formation of these alternative metabolites could add complexity to data derived from 7-EC metabolism generated with microsomes. Indeed, rapid HPLC analysis of 7-EC incubations could obscure retention time differences and thus confound metabolite determinations. Additionally, as some assays of 7-EC biotransformation (Ball et al., 1996; Hashemi et al., 1999) may use total fluorescence with and without -glucuronidase treatment to quantitate metabolism, the observed fluorescence of the 3-hydroxyglucuronide could distort experimental conclusions. Also, since 7-EC can be used as an in vitro probe for CYP1A1 or 1A2 at low concentrations (Yamazaki et al., 1996) and the enzymology of 7-EC 3-hydroxylation has not yet been evaluated, care must be taken to ensure unambiguous metabolism data depending upon the analytical method used.