Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Pennyroyal oil is a volatile oil obtained from the leaves of the plants of Mentha pulegium and Hedeoma pulegoides (Guenther, 1949). A tea prepared from the leaves of pennyroyal has been recommended as an aromatic stimulant, carminative, emmenagogue, and headache remedy (Castleman, 1995). Pennyroyal herb and oil are commonly employed in collars to keep cats and dogs free from fleas, and pennyroyal plants hung up to dry are also said to be an effective mosquito repellent. Furthermore, pennyroyal oil has been used in folklore medicine for many years as an abortifacient (Gunby, 1979). However, it does so at lethal or near-lethal doses, so that its effects are unpredictable and dangerous. Recently, several new cases of pennyroyal toxicity have been reported (Anderson et al., 1996; Bakerink et al., 1996). Most cases have occurred in adult women who used pennyroyal as an abortifacient and some of these cases have resulted in death.

The major terpene in pennyroyal oil responsible for its toxicity is pulegone (Gordon et al., 1982). Pulegone is metabolized to several metabolites of which menthofuran appears to be the major proximate toxin based on toxicokinetic studies (Fig. 1) (Thomassen et al., 1988). In this study, we determined the major human liver cytochrome P-450s (P-450s)¹ involved in the metabolism of pulegone and menthofuran. We also identified a new metabolite of menthofuran, 2hydroxymenthofuran, and we determined that it isomerizes to form the detoxication products mintlactone and isomintlactone. In addition, we determined that oxygen from molecular oxygen and water are incorporated into menthofuran metabolites, suggesting that 2-hydroxymenthofuran arises via a dihydrodiol formed from a furanoepoxide.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   General mechanisms of pulegone and menthofuran oxidative metabolism by cytochrome P-450. 2,3-Epoxymenthofuran and -ketoenal are reactive metabolites of menthofuran that give rise to TDC in the presence of semicarbazide, glutathione (GSH) conjugates in the presence of GSH, protein adducts in the presence of proteins, and 2-hydroxymenthofuran, mintlactone, and isomintlactone.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. Glucose 6-phosphate and glucose 6-phosphate dehydrogenase were purchased from Boehringer-Mannheim (Indianapolis, IN). Deuterium oxide was purchased from Cambridge Isotope Laboratories (Andover, MA). H₂¹⁸O was purchased from ISOTEC Inc. (Miamiburg, OH). Pulegone, dilauryl-DL--phosphatidylcholine (DLPC), dextromethorphan, and chlorzoxazone were purchased from Sigma Chemical Co. (St. Louis, MO). Butyllithium (gold label), semicarbazide hydrochloride, 4-methylpyrazole, and sodium cyanide were purchased from Aldrich Chemical Co. (Milwaukee, WI). Sodium sulfate, anhydrous granular powder, was purchased from J. T. Baker Chemical Co. (Phillipsburg, NJ). Bicinchoninic acid protein assay was purchased from Pierce Chemical Co. (Rockford, IL). Menthofuran was purchased from Fluka (Buchs, Switzerland). Coumarin was from Merck (Rahway, NJ). All solvents used were HPLC grade.

Enzymes. Expressed human CYP2C19 and CYP2D6 were purchased from GENTEST Corp. (Woburn, MA). Human CYP1A2 was expressed and purified in our laboratory; details will be reported in a future publication. Human CYP2E1 and CYP2A6 were expressed and purified as previously described (Chen et al., 1996; Koenigs et al., 1997). The P-450 cDNAs for human CYP1A2 and CYP2A6 were kindly provided by Dr. Frank J. Gonzalez, National Institutes of Health, Bethesda, MD. The purified P-450s were reconstituted with recombinant rat NADPH-cytochrome P-450 oxidoreductase, recombinant human cytochrome b₅, and DLPC at molar ratios of 1:2:2:800 for CYP2E1, 1:3:1:800 for CYP1A2, and 1:2:2:600 for CYP2A6 to achieve maximal rates of activity for each isoform. Furafylline was kindly provided by Dr. Jagdish Racha and Dr. Kent L. Kunze in our department. Human liver microsomes (HL), from the National Institutes of Health-supported human liver bank at the University of Washington, were prepared and handled as previously described (Rettie et al., 1989). Microsomal cytochrome P-450 content was quantified by the method of Omura and Sato (1964), and protein concentrations were determined by BCA assay.

Instrumentation. NMR studies were performed on a Varian VXR-300 spectrometer (Varian Analytical Instruments, Sunnyvale, CA). Infrared spectra were recorded on a Perkin-Elmer 1600 series Fourier transform infrared instrument (Perkin-Elmer Cetus Instruments, Norwalk, CT). Ultraviolet spectra were recorded on a Cary 3E UV-visible by Varian analytical instruments. Gas chromatographic analyses were performed on a Hewlett-Packard 5890 (Hewlett-Packard, Palo Alto, CA) using a 30 m × 0.32 mm WCOT DB-5 fused silica capillary column (J & W Scientific, Folsom, CA). Electron impact-mass spectrometry was carried out on a VG Micromass Trio-2000 quadruple mass spectrometer (Cheshire, U.K.). HPLC was performed on a Hewlett-Packard 1090 equipped with UV (Palo Alto, CA).

Synthesis. Dimethoxymenthofuran and 5,6,7,8-tetrahydro-4,7-dimethyl-7H-cinnoline (TDC) were synthesized by a method previously described (McClanahan et al., 1989), as was dimethyldioxirane (Oishi and Nelson, 1992).

1

Synthesis of 2-Hydroxymenthofuran. Dimethoxymenthofuran (200 mg, 940 µmol) in 10 ml of ether was added to 10 ml of 0.1 M sulfuric acid solution under argon and the reaction mixture was stirred at room temperature for 4 h. The biphasic reaction mixture was extracted three times with 10 ml of ether. The organic phase was dried over anhydrous sodium sulfate and transferred to a dry flask. The ether was evaporated under a stream of argon and a yellow oily product was obtained. Any attempt to further purify the product (column chromatography and/or vacuum transfer) led to additional decomposition. The half-life of 2-hydroxymenthofuran was determined by gas chromatography-mass spectrometry (GC/MS; Table 1). Addition of D₂O to 2-hydroxymenthofuran increased the molecular ion from m/z 166 to 167. The spectral data for 2-hydroxymenthofuran were as follows: Infrared: broad band at 3500 cm¹ (O-H stretching) and 1694 and 1650 cm¹ (C = C stretching vibration of unsymmetrical conjugated diene), Ultraviolet: _max 278 nm (2-hydroxy group attached to furan ring causes bathochromic shift of menthofuran _max 221 nm). ¹H NMR (CDCl₃, 300 MHz):  1.25 to 1.30 (3H, d, 6-CH₃), 1.87 to 1.90 (3H, s, 3-CH₃) and 2.5 to 1.5 (7H, m). MS m/z (% relative intensity) 166 ([M]⁺, 70), 138 ([M-CO]⁺, 20), 137 ([M-HCO]⁺, 10), 124 ([M-CO-CH₂]⁺, 10), 123 ([M-HCO-CH₂]⁺, 35), 110 ([M-CO-C₂H₄]⁺, 25), 109 ([M-HCO-C₂H₄]⁺, 14), 95 (75), 81 (100), 67(35). Exposure of this metabolite to D₂O shifts the molecular ion from m/z 166 to 167, whereas the remainder of the spectrum is unchanged. The mass spectrum of the trimethylsilylated derivative gave a parent ion at m/z 238 ([M]⁺, 70) and a base peak ion at m/z 196 ([M-C₃H₆]⁺, 100) (Fig. 2).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   The mass spectrum of the trimethylsilylated (TMS) derivative of 2-hydroxymenthofuran.

Synthesis of 2-(²H)-Menthofuran. In a round-bottom flask, menthofuran (300 mg, 2.0 mmol) was dissolved in 100 ml of dry tetrahydrofuran under argon. The flask was cooled to 78°C by adding dry ice/acetone and butyllithium (2.5 mmol) dropwise over a period of 15 min. After 30 min, D₂O (5 ml) was added to the reaction mixture and the temperature was kept at 78°C for another 30 min. The flask was left to warm up to room temperature and the product was extracted with ethyl acetate (3 × 25 ml). The ¹H NMR was the same as menthofuran except for the disappearance of the peak at 7.02 ppm, corresponding to the hydrogen atom at the 2 position. The percentage of incorporation of deuterium was 85% as determined by GC/MS.

Pulegone and Menthofuran Metabolism Assays. The metabolism of pulegone and menthofuran were measured by GC/MS as follows. HLs (50 µg) or expressed P-450s (30 pmol) were incubated at 37°C in 250 µl (final volume) incubation mixture containing potassium phosphate buffer (100 mM, pH 7.4 or pH 6.5), NADP⁺ (0.5 mM), glucose 6-phosphate (10 mM), glucose 6-phosphate dehydrogenase (1 U/ml), and pulegone or menthofuran [DLPC (0.01% w/v final concentration)]. Reactions were initiated by addition of the NADPH-generating system. The incubation time for pulegone was 10 min and for menthofuran was 15 min. The reaction rates were linear for both substrates over these time periods. The reaction mixtures were quenched by placing the tubes in dry ice/acetone. Acetophenone (2.5 nmol/incubation) was added as an internal standard. The incubations were then extracted with ethyl acetate (100 µl) by vortexing for 30 s, followed by centrifugation in a microfuge for 3 min at 16,000g. The tubes were placed in dry ice/acetone to freeze the aqueous phase and the organic phase was transferred and dried over anhydrous sodium sulfate for 5 min. The dried organic phase was transferred to an insert and analyzed by GC/MS.

Inhibition Studies. Four sets of HLs (HL126, HL129, HL133, and HL146) were pooled together and used for this study. Mechanism-based inhibitors [furafylline, diethyldithiocarbamate (DDC), and troleandomycin] were preincubated with microsomes, buffer, and a NADPH-generating system and incubated for 10 min at 37°C. Pulegone or menthofuran was added to the reactions, which then were incubated for an additional 10 min for pulegone or 15 min for menthofuran. Competitive inhibitors were coincubated with pulegone or menthofuran under the same conditions mentioned above.

GC/MS Procedure. For pulegone metabolism the GC temperature program used was as follows: the temperature was kept at 65°C for 3 min and then raised to 135°C at a constant rate of 10°C/min. Menthofuran was quantified by selected ion monitoring of the peak area at m/z 150 (retention time 8.1 min) divided by the peak area for acetophenone at m/z 120 (retention time 6.4 min).

CYP2E1 Activity Assay. Chlorzoxazone 6-hydroxylase activity was used as a marker for CYP2E1 activity. The incubation mixture contained expressed and purified human CYP2E1 or HLs, potassium phosphate buffer (100 mM, pH 7.4), NADP⁺ (0.5 mM), glucose 6-phosphate (10 mM), glucose 6-phosphate dehydrogenase (1 U/ml), chlorzoxazone (10, 20, 40, 80, 100, and 500 µM), and pulegone or menthofuran (0-200 µM with DLPC [0.01% (w/v) final concentration]. Reactions were initiated by addition of the NADPH-generating system. The metabolite, 6-hydroxychlorzoxazone was measured by an HPLC method previously described (Peter et al., 1990).

Kinetic Analysis. The kinetic parameters (K_m, V_max, and K_i) were determined by SYSTAT 5.2.1, a computer program designed for nonlinear regression analysis.

¹⁸O-Incorporation Studies. The incorporation of ¹⁸O from H₂¹⁸O was performed under the conditions described for menthofuran incubations with the following modifications. The total volume was changed to 100 µl and H₂¹⁸O (30 µl) was added to the incubation in place of unlabeled H₂O. The ¹⁸O content of the reaction was determined to be 24.5% by direct infusion of H₂¹⁸O in the mass spectrometer source.

Investigation of Pathways Involved in Metabolism of Menthofuran. Semicarbazide (1 mM) was incubated separately in an aqueous solution with dimethoxymenthofuran (1 mM) and 2-hydroxymenthofuran (1 mM) in 1.0 ml final volume. Aliquots (50- µl), of the reaction mixtures were transferred to microfuge tubes containing ethyl acetate (100 µl) and vortexed for 30 s. The organic phase was dried over sodium sulfate and analyzed by GC/MS.

Toxicity Study of Mintlactone and Isomintlactone. Hepatotoxicity of mintlactones in vivo was assessed by determining alanine aminotransferase levels in plasma after i.p. injection of mintlactones (0, 50 and 150 mg/kg rat) into two Sprague-Dawley rats (250 g) at each dose level.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Pulegone Metabolism by Human Liver Cytochrome P-450s. Pulegone (200 µM) was incubated separately with several human cytochrome P-450s (1A2, 2A6, 2C9, 2C19, 2D6, 2E1, and 3A4). Only CYP2E1, CYP1A2, and CYP2C19 oxidized pulegone to menthofuran with measurable velocities of 8.5, 2.1, and 1.7 nmol/min/nmol P-450, respectively (Fig. 3A).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   A, rates of metabolism of pulegone (200 µM) to menthofuran in incubations of expressed human liver cytochrome P-450s over a time interval (10 min) in which formation rate of metabolite is linear. B, Lineweaver-Burk plots of pulegone metabolism by expressed human liver cytochrome P-450 (expressed CYP2E1, ; expressed CYP1A2, ; and CYP2C19, ).

Menthofuran Metabolism. In this study, a new metabolite of menthofuran was detected. The infrared spectrum of this metabolite demonstrated the presence of a broad band at 3500 cm¹ corresponding to the presence of a hydroxyl group. An exchangeable hydrogen was detected when the metabolite was allowed to exchange with deuterium oxide; the mass spectrum showed an increase of molecular ion from m/z 166 to 167. The mass spectrum of this metabolite was identical with the mass spectra of mintlactones except for the abundance of the m/z peaks found, which indicates that the structure of this new metabolite is very similar to mintlactones. There was no observable peak in the region of resonance absorption for a furanyl hydrogen. However, the ¹H NMR indicated the presence of a singlet at  1.9, which corresponded to the methyl group at the C-3 position. The position of resonance absorption indicates that the hybridization of C-3 is sp², which is an indication for the presence of furan ring in this metabolite. In addition, the ultraviolet spectrum had a _max at 278 nm corresponding to a conjugated system. The trimethylsilylated derivative of this metabolite demonstrated the presence of two prominent peaks at m/z 238, the molecular ion and m/z 196, the base peak (Fig. 2). The neutral loss of 42 atomic mass units corresponds to the loss of propylene, which was also observed with menthofuran. The loss of propylene apparently occurs via a retro-Diels-Alder reaction and is favored by retention of the furan ring. In conclusion, we propose that this new metabolite is 2-hydroxymenthofuran.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   A, rates of metabolism of menthofuran (200 µM) to 2-hydroxymenthofuran, mintlactone, and isomintlactone in incubations of expressed human liver cytochrome P-450s over a time interval (15 min) in which formation rates of metabolites were linear. B, Lineweaver-Burk plots of menthofuran metabolism by expressed human liver cytochrome P-450 (expressed CYP2E1, ; expressed CYP1A2, ; and CYP2C19 ().

¹⁸O-Incorporation Studies. Based on studies using ¹⁸O₂ or H₂¹⁸O, we determined that 2-hydroxymenthofuran, mintlactone, and isomintlactone incorporate the same percentage of oxygen from each source. These results are consistent with the observation that the mintlactones arise by isomerization of 2-hydroxymenthofuran (Table 4). All metabolites incorporate one oxygen atom from H₂O in 35 to 45% of the product molecules and one oxygen atom from O₂ in 60 to 70% of the product molecules. Based on the mass spectral fragmentation patterns of the metabolites, these oxygens are incorporated exclusively at the C-2 position of the furan ring. The other oxygen that was not accounted for by ¹⁸O-incorporation is the furanyl oxygen atom of menthofuran.

Other Pathways Involved in Oxidation of Menthofuran. To determine whether 2-hydroxymenthofuran and mintlactones could be formed from proposed reactive metabolites of menthofuran, products of the hydrolysis of dimethoxymenthofuran and 2,3-epoxymenthofuran were characterized. Previously, TDC was shown to be a product of semicarbazide with a reactive -ketoenal intermediate of menthofuran formed in these reactions (Thomassen et al., 1992). When dimethoxymenthofuran (a chemical precursor of the -ketoenal) was incubated in aqueous solution, 2-hydroxymenthofuran and mintlactones were formed. When dimethoxymenthofuran was incubated in the presence of semicarbazide, TDC was formed, indicating the formation of a -ketoenal as previously observed (McClanahan et al., 1989; Thomassen et al., 1992). However, semicarbazide had no effect on the formation of 2-hydroxymenthofuran or mintlactones. Moreover, there was no reaction between semicarbazide and either 2-hydroxymenthofuran or mintlactones.

Toxicity of Mintlactone and Isomintlactone. Administration of up to 150 mg/kg i.p. (a severely hepatotoxic dose of menthofuran in rats) of mintlactones caused no mortality over a 48-h period and caused no hepatic necrosis as evidenced by lack of histopathologic changes in the liver and by lack of increases in serum alanine aminotransferase above normal.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Pulegone is the major constituent of pennyroyal oil that is responsible for hepatotoxicity in humans (Anderson et al., 1996). Studies in vivo and in vitro with inhibitors and inducers of cytochrome P-450 have demonstrated an association between hepatotoxicity of pulegone and its metabolic activation by P-450s (McClanahan et al., 1989). Pulegone is metabolized to a proximate hepatotoxic metabolite, menthofuran, that contributes significantly to the toxic effects of pulegone (Thomassen et al., 1988). Like pulegone, menthofuran requires oxidation by cytochrome P-450s to cause hepatotoxicity. As part of this study, we identified the P-450s in human liver responsible for these oxidative reactions.

The major human P-450 responsible for the oxidation of pulegone to menthofuran was found to be CYP2E1, with minor contributions by CYP1A2 and CYP2C19 (Fig. 3 and Tables 2 and 3). Expressed and purified human CYP2E1 had the lowest K_m and the highest V_max, resulting in the highest value for intrinsic clearance, V_max/K_m = 0.29 nmol/min/nmol P-450/µM. This value is about an order of magnitude greater than the intrinsic clearances for either CYP1A2 or CYP2C19. Concentrations of pulegone and menthofuran in humans after toxic doses of pennyroyal have not been determined, so we do not know the relevance of the K_ms to in vivo concentrations. However, based on the in vivo studies reported here, CYP2E1 would be the major isoform involved in the oxidation of both terpenes.

Studies with selective inhibitors of various P-450 isoforms using pooled HLs corroborate these results and suggest that >80% of the metabolism of pulegone to menthofuran is inhibited by either 4-methylpyrazole or DDC. Both of these compounds have been shown to inhibit CYP2E1. Although 4-methylpyazole also inhibits CYP2D6 (Newton et al., 1995) and DDC inhibits CYP2A6 (Chang et al., 1994), CYP2A6 and CYP2D6 did not oxidize pulegone to menthofuran. Therefore the inhibition of pulegone metabolism by 4-methylpyrazole and DDC is due to inhibition of CYP2E1. Furafylline also inhibited pulegone metabolism up to 15%, indicating some involvement of CYP1A2 at the concentrations of pulegone used (200 µM).

Investigations of the metabolism of menthofuran by human liver P-450s led to the identification of 2-hydroxymenthofuran as a new metabolite by infrared, UV, and ¹H NMR. 2-Hydroxymenthofuran isomerizes to the stable lactones, mintlactone, and isomintlactone in a pH-dependent manner (Table 1). Therefore, all three products were quantified in P-450 incubations. The major P-450s involved in oxidation of menthofuran are the same ones involved in the oxidation of pulegone, with the addition of CYP2A6. CYP2E1 again is the most efficient catalyst, having a K_m value similar to that for pulegone, although the V_max value is about 20 times less than that of pulegone.

As to the involvement of CYP2A6, we have shown that menthofuran is a potent mechanism-based inactivator of this isoform (Khojasteh-Bakht et al., 1998). The involvement of CYP2A6 in the metabolism of menthofuran would be expected to be limited, because the enzyme is inactivated with a partition ratio of products formed to enzyme inactivation of approximately 4:1. We also determined that pulegone is not a mechanism-based inactivator of CYP2A6 (data not shown).

The results of these studies are interesting from the standpoint that two inducible isoforms of P-450 are responsible for most of the oxidation of pulegone and menthofuran. Although the major mintlactone end products are not toxic, presumably the precursor primary metabolites such as a furan epoxide and/or -ketoenal are hepatotoxic (McClanahan et al., 1989; Thomassen et al., 1992). Therefore ingestion of alcohol, which induces CYP2E1, and/or smoking cigarettes, which induces CYP1A2, may contribute to hepatotoxicity after pennyroyal ingestion.

Investigation of the mechanism of formation of 2-hydroxymenthofuran from menthofuran by CYP2E1 provided evidence for a dihydrodiol intermediate that resulted from hydration of an initial furan epoxide. Based on studies with ¹⁸O₂, a minimum of 60% of the oxygen incorporated at the C-2 position of 2-hydroxymenthofuran and its mintlactone isomers is derived from O₂ and based on studies with H₂¹⁸O the remainder of the oxygen is derived from H₂O. Meanwhile, the furanyl oxygen of menthofuran is retained as the endocyclic oxygen in the products. These results are consistent with pathway a in Fig. 5 in which a furanyl epoxide is formed by CYP2E1 and then is hydrolyzed spontaneously with addition of water at either C-2 or C-3. If hydrolysis occurs at C-2, oxygen in the final products will be derived from water following dehydration of an intermediate dihydrodiol via pathway d (Fig. 5). If addition occurs at C-3, the oxygen at C-2 from molecular oxygen will be retained. Alternatively, spontaneous ring opening of the epoxide assisted by the furanyl oxygen lone pair of electrons, followed by proton loss and rearomatization (Fig. 5, pathway e) would give the same result. This is consistent with the finding that a larger fraction of the oxygen at C-2 is derived from O₂.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Proposed mechanisms for formation of the -ketoenal, 2-hydroxymenthofuran, and mintlactones from 2,3-epoxymenthofuran.

Ring opening of a furanyl epoxide could also lead to formation of a -ketoenal as shown in pathway c and/or pathway b (Fig. 5). Results of studies with 2-hydroxymenthofuran showed that this metabolite does not rearrange to the -ketoenal, because no TDC adduct was formed from 2-hydroxymenthofuran or mintlactones. Presumably, the -ketoenal could hydrate and cyclize as shown. However, under the conditions of the short incubation periods used in this study, this occurred only to a limited extent and addition of semicarbazide did not significantly decrease the formation rates of 2-hydroxymenthofuran and mintlactones. Over greater time periods, a significantly greater amount of oxygen from water is incorporated into the furanyl oxygen and C-2 oxygen as previously found (Thomassen et al., 1992).

Finally, there was no deuterium isotope effect on the formation rates of 2-hydroxymenthofuran or mintlactones when deuterium was substituted for hydrogen at C-2 of menthofuran. This result provides indirect evidence that there is no direct hydrogen atom abstraction from this position in the oxidation mechanism.

In conclusion, pulegone and menthofuran are metabolized primarily by CYP2E1 and to a lesser extent by CYP1A2 and CYP2C19. We identified a new metabolite of menthofuran, 2-hydroxymenthofuran, which isomerizes to form mintlactone and isomintlactone. Finally, results of investigations with ¹⁸O₂, H₂¹⁸O, and C-2 deuterium-labeled menthofuran are consistent with the formation of a furan epoxide intermediate in the oxidation of menthofuran by CYP2E1. Previously, we had shown that dimethyldioxirane could chemically oxidize menthofuran to a furanyl epoxide (Oishi and Nelson, 1992). This epoxide may directly rearrange to form 2-hydroxymenthofuran and a -ketoenal and/or form these metabolites by hydration to a diol followed by dehydration reactions. Whether or not the epoxide, the -ketoenal, or both are responsible for the hepatotoxicity caused by menthofuran requires further investigation.