Introduction

Top Abstract Introduction Materials and methods Results Discussion References

Coumarin (1,2-benzopyrone), a natural plant product, is widely used as a perfume additive in consumer products (Cohen, 1979; Opdyke, 1974). Coumarin was also used as a food additive, but this practice was discontinued because of potential hepatotoxicity (Hazleton et al., 1956; Hagan et al., 1967). In recent years, coumarin, combined with cimetidine, has been used in clinical trials for the treatment of malignant melanoma, metastatic renal cell carcinoma (Marshall et al. 1994), and prostatic carcinoma (Mohler et al., 1994).

Recent studies in this laboratory indicated that i.p. administration of coumarin at 25 to 50 mg/kg resulted in tissue-selective cytotoxicity in the olfactory mucosa (OM) of Wistar rats and C57BL/6 mice (Gu et al., 1997). In comparison, higher doses were needed to produce hepatotoxicity in rats (Lake, 1984; Lake et al., 1989). Dose-response analysis of the olfactory toxicity of coumarin indicated that Wistar rats were more sensitive than C57BL/6 mice (Gu et al., 1997). Similar species differences were also found in the sensitivity to coumarin-induced hepatotoxicity in earlier studies and were thought to be derived from species differences in the biotransformation of coumarin in the liver (Lake and Grasso, 1996; Ratanasavanh et al., 1996). In humans, 7-hydroxycoumarin is the predominant coumarin metabolite in vivo and in vitro in liver microsomes (Lake et al., 1992a; Moran et al., 1987). In rats, however, o-hydroxyphenylacetic acid and o-hydroxyphenylacetaldehyde (o-HPA) are the major in vivo and in vitro products, respectively (Cohen, 1979; Fentem et al., 1991; Lake et al., 1992b). Coumarin metabolism in mice was found to be strain dependent (Raunio et al., 1988; Wood and Taylor, 1979), and mice with low hepatic coumarin 7-hydroxylase activity were much more susceptible to coumarin-induced hepatotoxicity than those with high activity (Endell and Seidel, 1978).

High levels of coumarin 7-hydroxylase activity were observed in the OM in both rats and mice in previous studies (Béréziat et al., 1995; Liu et al., 1996; Su et al., 1996), but little is known about the metabolic activation of coumarin in OM microsomes, the P-450 isoforms involved, the role of target tissue activation or detoxification in the OM-selective toxicity, or the basis of species differences in coumarin toxicity in the OM. In this study, coumarin metabolism in rat and mouse OM microsomes was analyzed and compared with that in liver microsomes using reverse phase high-performance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS). The role of CYP2G1 and CYP2A5 [the nomenclature used in this report is according to Nelson et al. (1996)] in OM microsomal coumarin metabolism was examined immunochemically and by kinetic analysis of coumarin metabolism with purified, heterologously expressed mouse CYP2G1 and CYP2A5 in reconstituted systems. In addition, the activity of human P-450s in the metabolic activation of coumarin was examined using microsomes of SF9 cells heterologously expressing 13 individual human P-450 isoforms.

	Materials and Methods

Top Abstract Introduction Materials and methods Results Discussion References

Chemicals. [4-¹⁴C]Coumarin, reduced glutathione (GSH), -NADPH, -NADP, D-glucose-6-phosphate, and glucose-6-phosphate dehydrogenase were obtained from Sigma Chemical Co. (St. Louis, MO). Coumarin, 4-hydroxycoumarin, 7-hydroxycoumarin, 3,4-dihydrocoumarin, 2-hydroxyphenylacetic acid, 3-(2-hydroxyphenylpropionic acid), o-coumaric acid, and iodomethane were purchased from Aldrich (Milwaukee, WI). 3-Hydroxycoumarin, 6,7-dihydroxycoumarin, and 7,8-dihydroxycoumarin were obtained from Indofine Chemical Company (Somerville, NJ). o-HPA was kindly provided by Dr. Lois D. Lehman-McKeeman of the Procter & Gamble Company (Cincinnati, OH). Insect cell microsomes expressing recombinant human P-450 (Supersomes) were purchased from Gentest (Woburn, MA).

Determination of Catalytic Activity toward Coumarin. [4-¹⁴C]Coumarin (14.4 mCi/mmol), which was 99.7% pure based on HPLC radiochromatography, was used as substrate in all microsomal and reconstituted reaction systems unless indicated otherwise. Contents of various incubation mixtures are indicated in the legends to tables and figures. For metabolite determination, reactions were terminated by the addition of 1 volume of ice-cold methanol. After centrifugation at 12,000 rpm at 4°C for 10 min, an aliquot of the supernatant was analyzed by reverse phase HPLC as described below. Protein adduct formation was determined as described by Ding et al. (1996).

HPLC Analysis of Coumarin Metabolites. A Waters (Milford, MA) µBondapak C18 (3.9 × 150 mm) column was used to analyze coumarin metabolites with a Waters HPLC system consisting of a model 600S controller, a model 626 pump, a model 717 plus autosampler, and a model 996 photodiode array detector. A FLO-ONE A-500 radiochromatography detector (Packard; Downers Grove, IL) was used to detect radioactive metabolites in the HPLC effluent. The mobile phase consisted of 25% (v/v) tetrahydrofuran in water (A), water (B), and 1% (w/v) formic acid in water (C), and the metabolites were eluted according to Peters et al. (1991). On-line radioactivity detection and quantification of radiolabeled metabolites were performed as described previously (Ding et al., 1996).

GC-MS Analysis of Coumarin Metabolites. Microsomal reactions were carried out as described above with 0.5 mM unlabeled coumarin as substrate in a final volume of 250 µl. Reactions were stopped by the addition of 750 µl of ethyl acetate and organic compounds were extracted after the addition of 7-ethoxycoumarin as an internal standard. The organic phase was dried under nitrogen, dissolved in 100 µl of dry acetone, and derivatized with iodomethane based on the method described by de Vries et al. (1985). An isooctane solution (10 µl) containing derivatized and underivatized compounds was analyzed through a Hewlett-Packard (Wilmington, DE) GC-MS system consisting of a model 5890A gas chromatograph, with a model 7673A automatic sampler, interfaced to a model 5970 series mass selective detector. A 30 m × 0.25 mm DB-1 column (J & W Scientific; Folsom, CA) with 0.25-µm film thickness was used for separation of metabolites with helium as the carrier gas at a flow rate of 0.75 ml/min. The GC oven temperature was maintained at 80°C initially for 5 min, followed by a ramp of 10°C/min to 300°C, and then a hold at 300°C for 3 min. The mass spectrometer was operated in full scan mode (from m/z 65 to m/z 600 in 1-s intervals) or in selective ion monitoring (SIM) mode. The GC conditions afforded resolution of derivatized authentic standards including 3-, 4-, and 7-hydroxycoumarins; 6,7- and 7,8-dihydroxycoumarins; o-hydroxyphenylacetic acid; o-hydroxyphenylpropionic acid; and o-coumaric acid.

Other Methods. Microsomes were prepared from OM and liver of male C57BL/6 mice (about 20 g b.wt.) and Wistar rats (about 200 g b.wt.) according to Ding and Coon (1990). Purification of heterologously expressed mouse CYP2A5 and CYP2G1 and preparation of a polyclonal antibody to purified CYP2A5 have been described recently (Gu et al., 1998). The sources of purified rabbit hepatic NADPH-P-450 reductase and cytochrome b₅ (b₅) have been described elsewhere (Ding et al., 1996). Protein concentrations were determined using BCA protein assay reagents (Pierce, Rockford, IL) with bovine serum albumin as the standard. Microsomal P-450 concentrations were determined according to Omura and Sato (1964). Statistical significance was determined using Student's t test.

	Results

Top Abstract Introduction Materials and methods Results Discussion References

Tissue Differences in Microsomal Coumarin Metabolism in Rats and Mice. The results of HPLC analysis of microsomal coumarin metabolites are shown in Fig. 1. The metabolites were identified, when possible, on the basis of comigration with authentic standards. In rats, several products were detected in OM microsomal reactions (Fig. 1A), including o-HPA, 7-hydroxycoumarin, 3-hydroxycoumarin, and three unidentified metabolites, designed X1, X2, and X3, with o-HPA being the most abundant. However, only one product, o-HPA, was detected in liver microsomal reactions (Fig. 1B), consistent with earlier reports that o-HPA is the predominant metabolite in rat liver (Lake et al., 1992b). In mice, the HPLC profile of OM microsomal coumarin metabolites (Fig. 1C) was very similar to that of rats, with no indication of a significant species difference in metabolism. With mouse liver microsomes, however, although o-HPA accounted for 78% of the total metabolites produced (Fig. 1D), several other metabolites, including X1, X2, X3, and 3-hydroxycoumarin, were also detectable, which is different from the results seen in rats. No metabolites were detected in control reactions in which NADPH was omitted. The 3-hydroxycoumarin peak was not well resolved from neighboring small peaks, which may also represent minor products. The HPLC system is capable of resolving available standards, including 3-, 4-, and 7-monohydroxycoumarins; 6,7- and 7,8-dihydroxycoumarins; 3,4-dihydrocoumarin; and o-HPA.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   HPLC radiochromatograms of [4-14C]coumarin metabolites formed in microsomal reactions. The reactions were carried out as described in the legend to Table 1 with microsomes from rat OM (A), rat liver (B), mouse OM (C), and mouse liver (D). Final protein concentrations were different in reactions with different microsomal preparations, as indicated in Table 1, and the substrate was added at 72 µM for olfactory microsomes and 144 µM for liver microsomes. Metabolites were separated using HPLC and detected with an on-line radioactivity detector. Metabolite peaks, including 3-hydroxycoumarin (3-OH), 7-hydroxycoumarin (7-OH), o-HPA, and three unidentified products, designed X1, X2, and X3, and the substrate are indicated.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   GC-MS analysis of derivatized monohydroxycoumarins. Microsomal coumarin metabolism was carried out as described in the legend to Table 1 but using unlabeled substrate at a concentration of 500 µM. Reaction mixtures contained olfactory microsomal proteins (0.5 mg/ml) or liver microsomal proteins (2 mg/ml) in a final volume of 250 µl. Hydroxycoumarins formed by microsomal P450s were then extracted and methylated by reactions with CH3I to form the methoxycoumarins. GC-MS analysis was performed in SIM mode (m/z 176) with methylated metabolites generated in reactions with rat OM (A), rat liver (B), mouse OM (C), and mouse liver microsomes (D). Five peaks were detected, including 3-methoxycoumarin (3-OH), 7-methoxycoumarin (7-OH), and three unidentified monomethoxycoumarins, designated M1, M2, and M3, which represent 5-, 6-, and 8-methoxycoumarins (5-, 6-, and 8-OH). Procedures for derivatization and GC-MS analysis are described in Materials and Methods. Note that the relative abundances are in different scales in each panel.

Coumarin Metabolism by Purified Mouse CYP2A5 and CYP2G1. The ability was determined of purified CYP2A5 and CYP2G1, major P-450 isoforms in mouse OM, to metabolize coumarin. As shown in Fig. 3, both enzymes produced more than one metabolite. 7-Hydroxycoumarin was the predominant product generated by CYP2A5, but low amounts of o-HPA, X1, and X2 were also formed. The metabolite profile of mouse CYP2G1 is similar to that of mouse OM microsomes, with o-HPA being the primary product. Kinetic parameters of o-HPA and 7-hydroxycoumarin formation by purified mouse CYP2G1 and CYP2A5 were also determined (Table 3). Although CYP2A5 was much more active than CYP2G1 in 7-hydroxycoumarin formation, with a 10-fold higher catalytic efficiency and a 4-fold higher V_max, it was much less active than CYP2G1 in the formation of o-HPA. With the substrate at 72 µM, the turnover numbers for o-HPA formation were 4.5 and 1.4 nmol/min/nmol of P-450 by CYP2G1 and CYP2A5, respectively (data not shown), and those for X1, X2, X3, and 3-hydroxycoumarin formations by CYP2G1 were 0.66, 0.32, 0.34, and 0.95 nmol/min/nmol of P-450, respectively. Kinetic parameters for o-HPA formation by CYP2A5 were not obtained because of difficulties in metabolite quantification at low substrate concentrations. Interestingly, CYP2G1 had a lower K_m for o-HPA formation than for coumarin 7-hydroxylation, suggesting a preferential role of CYP2G1 in metabolic activation. The apparent K_m values for mouse OM microsomal 7-hydroxycoumarin and o-HPA formations (3.1 µM and 2.7 µM, respectively; experiments not shown) were lower than the values found for the purified enzymes. This may be partly because of involvement of b₅ because in other experiments not shown, the addition of b₅ at a ratio of 4 nmol/nmol of P-450 increased the rates of coumarin metabolite formation by 30 to 36% with either CYP2G1 or CYP2A5 at a substrate concentration of 72 µM.

View larger version (17K): [in this window] [in a new window]   Fig. 3.   HPLC profiles of coumarin metabolites generated by purified mouse CYP2A5 and CYP2G1. Reaction mixtures contained 0.2 µM purified P-450, 50 mM phosphate buffer, pH 7.4, 0.6 µM purified NADPH-P-450 reductase from rabbit liver, 30 µg/ml dilauroylphosphatidylcholine, 72 µM [4-14C]coumarin, and 1 mM NADPH in a final volume of 125 µl. The reactions were carried out at 37°C for 10 min, and the metabolites were analyzed by radiometric HPLC as described in Materials and Methods.

Role of CYP2A and CYP2G1 in OM Microsomal Metabolism of Coumarin. To determine whether mouse CYP2A5 and CYP2G1 play important roles in coumarin metabolism in the olfactory tissue, an antibody to mouse CYP2A5, which cross-reacts with mouse CYP2G1, was included in microsomal reaction mixtures. In experiments not shown, the antibody recognized heterologously expressed human CYP2A6 but did not recognize a number of other human P-450 isoforms on immunoblots, including CYP1A1, CYP1A2, CYP2B6, CYP2E1, CYP3A4, CYP1B1, CYP2C8, CYP2C9-Arg, CYP2C19, CYP2D6-Val, CYP3A5, and CYP4A11. Moreover, the antibody, when used at a level of 5 mg IgG/nmol of P-450, inhibited both 7-hydroxycoumarin and o-HPA formation by purified mouse CYP2G1 completely and by purified CYP2A5 by more than 90%. As shown in Fig.4A, with anti-CYP2A5 IgG added at 1 mg/mg of mouse olfactory microsomal proteins and coumarin at 72 µM, formation of X3 and 7-hydroxycoumarin was totally inhibited. The addition of anti-CYP2A5 IgG also resulted in large decreases (64%) in o-HPA formation and similar decreases in X1 formation but no decreases in X2 formation. In other experiments (not presented), when coumarin concentration was reduced to 9 µM and the antibody was added at 3 mg/mg of OM microsomal proteins, a greater inhibition of microsomal o-HPA formation (75%) was observed, indicating that CYP2G1 plays a greater role in this coumarin metabolic activation at low than at high substrate concentrations.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Role of CYP2A5 and CYP2G1 in microsomal metabolism of coumarin. The effects of an inhibitory anti-CYP2A5 antibody, which also inhibits mouse CYP2G1, on coumarin metabolism were examined with microsomal preparations from mouse OM (A), rat OM (B), and rat or mouse liver (C). Reaction mixtures contained 72 µM [4-14C]coumarin for olfactory microsomal reactions or 144 µM for liver microsomal reactions, 50 mM phosphate buffer, pH 7.4, and anti-CYP2A5 IgG added at 0 to 3 mg/mg of microsomal proteins in a final volume of 125 µl. Reactions were carried out at 37°C for 10 min. The formations of X1, X2, X3, 7-hydroxycoumarin (7-OH), and o-HPA were determined for olfactory microsomal reactions, but only o-HPA was determined for liver microsomal reactions. The addition of preimmune IgG did not affect metabolic activity. The values presented are averages of two experiments with differences less than 15% of the mean.

Coumarin Metabolism by Heterologously Expressed Human P-450s. Of 13 human P-450s examined, only six were active toward coumarin, including CYP1A1, CYP1A2, CYP2A6, CYP2B6, CYP2E1, and CYP3A4; the other forms studied were CYP1B1, CYP2C8, CYP2C9-Arg, CYP2C19, CYP2D6-Val, CYP3A5, and CYP4A11 (data not shown). o-HPA was the only product detected in reactions with CYP1A1, CYP1A2, CYP2B6, and CYP2E1, whereas 7-hydroxycoumarin was the only product detected in reactions with CYP2A6. CYP3A4, however, produced relatively small amounts of X1 in addition to o-HPA. The activity of CYP2B6 was very low; product formation was detectable only after a 120-min reaction with reductase/P-450 = 12.5:1. No metabolite was detected by HPLC analysis in reactions with microsomes from control SF9 cells.

	Discussion

Top Abstract Introduction Materials and methods Results Discussion References

Our previous study indicated that coumarin is a tissue-selective toxicant in rodent OM and that rats are more sensitive than mice to coumarin olfactory toxicity (Gu et al., 1997). However, it is not clear whether the differential sensitivity resulted from tissue and species differences in coumarin metabolism. Metabolic activation of coumarin is thought to involve the formation of the reactive coumarin 3,4-epoxide (Cohen, 1979; Fentem and Fry, 1993; Lake, 1984), which can react with GSH to form GSH conjugates or react with protein to form covalent adducts (Huwer et al., 1991; Lake et al., 1992a). Alternatively, the epoxide will undergo rapid degradation to form o-HPA (Born et al., 1997), which can also cause toxicity in the cell (Feron et al., 1991). On the other hand, the formation of monohydroxy and dihydroxycoumarins represents detoxification (Gu et al., 1997; Lake et al., 1989). The results of this study indicate that the rates of metabolic activation of coumarin in rat and mouse OM microsomal preparations (per mg microsomal protein) were much higher than those in the liver, consistent with the role of target tissue activation in tissue-selective toxicity. Although there were marked species differences in hepatic coumarin metabolic rates, with mice being much more active than rats, there was no significant difference in the rate of metabolic activation or of total coumarin metabolite formation in OM microsomes of rats and mice. Thus, the species difference in coumarin olfactory toxicity may be mainly because of differences in the rates of systemic clearance of this compound rather than because of differences in target tissue activation or detoxification.

CYP2G1 and CYP2A5 are major P-450 isoforms of mouse OM microsomes (Gu et al., 1998), and both can generate o-HPA and 7-hydroxycoumarin from coumarin. The results of immunoinhibition assays and kinetic analyses suggest that in mice, CYP2G1 plays a major role in coumarin metabolic activation, whereas CYP2A5 plays a major role in coumarin detoxification in the OM. The incomplete inhibition of microsomal o-HPA formation by the anti-CYP2A5 antibody suggested that other P-450s, such as CYP2E1, CYP1A2, and CYP3A, which are known to be expressed in mouse OM (Genter et al., 1998), may also contribute to the metabolic activation, particularly at high substrate concentrations. Conversely, the lack of significant inhibition of hepatic coumarin metabolism by the anti-CYP2A5 IgG not only documents the specificity of the antibody but also confirms that the high activity in metabolic activation of coumarin in mouse olfactory microsomes is due to the presence of the tissue-specific CYP2G1.

The P-450 isoforms involved in coumarin metabolism in rat OM have not been as thoroughly studied. A previous report indicated that rat CYP1A1 and CYP2B1 are active in o-HPA formation from coumarin (Peters et al., 1991); these isoforms have been detected in rat OM, although their levels have not been determined (Ding and Coon, 1993). Our recent studies indicate that CYP2A3 is an abundant P-450 in rat OM (Su et al., 1996); however, SF9 microsomes containing heterologously expressed rat CYP2A3 had very low activity toward coumarin 7-hydroxylation (Liu et al., 1996). Preliminary experiments with a partially purified rat CYP2A3 preparation indicated that CYP2A3 is active in o-HPA formation from coumarin, although the activity is quite low (X. Zhuo, C. Liu, and X. Ding, unpublished results). Rat CYP2G1, the OM-specific isoform, is not yet available for metabolic studies, although the cDNA has been cloned (Nef et al., 1989). Nevertheless, the antibody to mouse CYP2A5 and CYP2G1 inhibited up to 85% of rat OM microsomal activity in o-HPA formation, suggesting that CYP2G1 and CYP2A3 may play major roles in coumarin metabolic activation in rat OM, although the relative importance of the two isoforms remains to be determined. The lack of inhibition of hepatic microsomal o-HPA formation is consistent with the absence of CYP2A3 and CYP2G1 in rat liver (Kimura et al., 1989; Nef et al., 1989; Su et al., 1996).

The results from the present study add coumarin to a growing list of compounds, such as 2,6-dichlorobenzonitrile and acetaminophen, that are preferentially activated by tissue-predominant P-450 isoforms (i.e., CYP2A or CYP2G1) and cause tissue-selective toxicity in rodent OM (Ding et al., 1996; Genter et al., 1998; Gu et al., 1998; Liu et al., 1996). It is unclear, however, whether tissue-specific or predominant P-450s are present in human nasal mucosa, and it has not been determined whether human nasal microsomes are active in the metabolic activation of coumarin. Nevertheless, multiple P-450 isoforms, including CYP2A6, CYP1B1, CYP3A5, CYP2E1, and CYP4B1, have been detected in human nasal mucosa (Getchell et al., 1993; Su et al., 1996; J. Gu, T. Su, and X. Ding, unpublished) and could potentially impart toxic consequences on exposure to these chemicals. In the liver, CYP2A6 has been well documented as the major coumarin 7-hydroxylase (Fernandez-Salguero and Gonzalez, 1995), although it was not known which human P-450s are active in the metabolic activation of coumarin. Our results demonstrate that at least five human P-450 isoforms, including CYP1A1, CYP1A2, CYP2B6, CYP2E1, and CYP3A4, can generate o-HPA from coumarin. Interestingly, all five isoforms are known to be inducible by xenobiotic compounds, thus providing a mechanism for potential drug-drug interactions that may augment coumarin toxicity in the liver and possibly other tissues. Consistent with a previous report that 7-hydroxycoumarin is the major urinary coumarin metabolite in humans (Moran et al., 1987), CYP2A6 appears to have a lower K_m in coumarin 7-hydroxylation than metabolic activation by the other isoforms. However, individuals with less active or inactive CYP2A6 variants (Fernandez-Salguero et al., 1995) may be less protected and could be subjected to higher risk of coumarin toxicity, particularly when high doses are given for therapeutic purposes.