XENOBIOTICA,
1992, VOL. 22,
NOS
9/10, 1157-1164
Investigations of mechanisms of reactive metabolite formation from (R)-( + )-pulegone S. D. NELSON?, R. H. McCLANAHAN, D. THOMASSEN, W. PERRY GORDON and N. KNEBEL
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
Department of Medicinal Chemistry, BG-20, University of Washington, Seattle, WA 98195, USA
Received 5 October 1990; accepted 24 April 1992 1. (I?)-( +)-Pulegone is a monoterpene that is oxidized by cytochromes P-450 to reactive metabolites that initiate events in the pathogenesis of hepatotoxicity in mice, rats and humans. 2. Selective labelling of (R)-( )-pulegone with deuterium revealed that menthofuran was a proximate hepatotoxic metabolite formed by oxidation of the allylic methyl groups of pulegone. Incubations of pulegone with mouse liver microsomes in an atmosphere of "0, resulted in the formation of menthofuran that contained only oxygen-18 in the furan moiety. These results are consistent with oxidation of pulegone to an allylic alcohol that reacts intramolecularly with the ketone moiety to form a hemiketal that subsequently dehydrates to generate menthofuran. 3. Studies on the metabolism of menthofuran revealed that it is oxidized by cytochromes P-450 to an electrophilic y-ketoenal that reacts with nucleophilic groups on proteins to form covalent adducts. In addition, diastereomeric mintlactones are formed. Investigations with H,'*0 and "02are indicative of a furan epoxide intermediate, or a precursor, in the formation of the y-ketoenal and mintlactones.
+
Introduction (R)-(+)-Pulegone is a major monoterpene constituent of pennyroyal oil, a mint oil obtained from the leaves of the Mentha pulegium or Hedeoma pulegoides (Grundschober 1979). Pennyroyal oil is widely used as a fragrance and flavouring agent (Hall and Oser 1965), and has been used, based on folklore, as a herbal medicine to control menses and terminate pregnancy (Gleason et al. 1969). The ingestion of large quantities of the oil has been associated with toxic effects (Vallance 1955, Watt and Breyer-Brandwijk 1962, Gunby 1979, Sullivan et al. 1979, Gold and Cates 1980), and hepatotoxicity observed in humans has been modelled in mice (Gordon et al. 1982) and rats (Thomassen et at. 1988). (R)-( )-pulegone was identified as the major hepatotoxic terpene component of pennyroyal oil (Gordon et al. 1982), and studies on the biotransformation of pulegone in rodents have led to the identity of several oxidative and reduced metabolites (Gordon et al. 1982, 1987, Nelson and Gordon 1983, Moorthy et al. 1989). This report will review some of this work, and will describe how stable isotopes have been used to investigate mechanisms of formation of some reactive metabolites of pulegone.
+
t To whom correspondence should be addressed. 0049-8254/92 83.00
0 1992 Taylor & Francis Ltd.
1158
S . D.Nelson
et
al.
Experimental Syntheses, purifications, and analyses of pulegone, menthofuran and their stable isotope labelled compounds have been reported elsewhere (Gordon et al. 1987, McClanahan et al. 1988, 1989). Pulegone epoxides were prepared by oxidation of (R)-( )-pulegone with m-chloroperoxybenzoic acid (m-CPBA) as described previously (Feely and Hargreaves 1970), and we have reported their physical spectral characteristics and g.1.c. analysis elsewhere (Gordon et al. 1982). Piperitenone was synthesized by condensation of methyl vinyl ketone and mesityl oxide with Triton B as described by others (Nakanishi et al. 1980).
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
+
Characterization and quantification of metabolites Incubations were carried out in Teflon-sealed Reacti-flasks (25 ml) that contained 1 mM (R)-(+)pulegone, 6.0 mg of microsomal protein from mouse liver, an NADPH-generating system, and 3 ml of 0.05 M phosphate buffer, pH 7.4. The NADPH-generating system consisted of (final concentration): NADP' (0.35 mM), glucose 6-phosphate (8.77 mM), glucose-6-phosphate dehydrogenase (1 unitlml), and MgC1, (3 mM). Substrate was added in 2Oyl of freshly distilled dimethyl sulphoxide (DMSO). All additions were made into flasks that were maintained at 0°C. Reactions were started by transferring the vials to an incubator shaker that was maintained at 37°C. Reactions were terminated by immersion of the reaction vessels in ice and by the addition of 0.5 ml of acetone to each flask. Metabolites were assayed by the following procedure. ( +)-3-Methylcyclohexanone (0.5 pnol in 2Oyl of DMSO) was added as an internal standard to each incubation flask. Incubation mixtures were then extracted with 6ml of diethyl ether, and the aqueous layer was discarded. The ether extracts were washed with 3 ml of 1 M-HCI and then with 3 ml of 1 M-NaOH. The organic extract was dried over anhydr. Na2S04. A 5-ml sample was taken from each extract and was concentrated to approx. 0.5 ml under a stream of N,. Isooctane (0.5 ml) was added to each concentrate. The remainder of the ether was removed under a gentle stream of N,, and 1.0-2.0y1 of the remaining solution was then injected onto the g.1.c. of a g.1.c.-mass spectrometry system. The g.1.c. conditions were as follows: column-HP SE-54 fused silica capillary column, 0.35 mm x 50m; helium as carrier gas; head pressure, 2Opsi; splitless injection; injector temp. at 250°C at lO"C/min and held at 150°C for 15-20min. The column was connected to a FID detector or the column was run directly into the mass spectrometer source. Mass spectra were obtained on a VG 7070 mass spectrometer coupled to an HP 5840 g.1.c. instrument and a VG 2000 data system. Helium was used as the carrier gas at a head pressure of 15 psi. Samples were introduced by splitless injection at a temp. of 250°C and an oven temp. programme of 3min at 60°C then to 150°C at 10"C/min. The column was connected oia a direct inlet transfer line held at 250°C. Mass spectrometry conditions were: source temp., 200°C; electron impact 70eV; emission current, 200yA; multiplier at 1800-2200 V. Metabolites were identified by a comparison of their g.1.c. retention indices relative to the internal standard and by comparison of their mass spectra, both from pulegone-do and pulegone-d,. For example, 4-hydroxypulegone had a relative retention indice of 2.41 and major ions at m/z168 ( M t ) , 156(Mt-H20), 135(Mt-H,O-CH,), and 107(Mt-CH3-H20-CO). Loss of the methyl group was primarily from the isopropylidine group as noted in the spectrum of the corresponding metabolite from incubation of pulegone-d6 where loss of a -CD3 group was observed. Quantification of metabolites was carried out by peak area comparisons with 3methylcyclohexanone as the internal standard to determine absolute amounts formed over time in separate incubations of pulegone-do and pulegone-d,. Incubations of a 50 : 50 mixture of pulegone-do and pulegone-d, were analysed by g.1.c.-mass spectrometry of parent ions, which were major ions generated from all metabolites. The relative ratios of deuterated to non-deuterated metabolites were found to agree well with the ratios obtained by dividing the absolute amounts of the metabolites obtained by g.1.c. analysis of separate incubations of pulegone-do and pulegone-d,. '0-Incorporation studies Menthofuran used in these studies was 99.6% pure and free of mintlactone (g.1.c.-FID). Formation of mintlactone was carried out chemically or enzymically, in the presence of either H,"O or "0,. Reaction conditions were identical to those described above, but in a volume of 0.5ml for H,'80 incubations and 1.0ml for "02 incubations. Incubations with H,l80 were stopped by the addition of 25yl benzoyl chloride, which is rapidly hydrolysed to form benzoic acid. Reaction mixtures were analysed for the isotopic composition of mintlactone and benzoic acid by g.1.c.-mass spectrometry. Conditions of this assay will be reported elsewhere. "0-Incorporation in benzoic acid is proportional to the atom-fraction (a) of l80in water. The percentage of mintlactone formed by incorporation of 2, 1, or 0 water-derived oxygen atom(s) (2W, lW, and OW, respectively) was calculated as follows:
2W=(ML4/a2)x 100%
+
Mechanisms of reactive metabolite formation from (R)- ( ) -pulegone
1159
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
1W =((aML2 -2(1 -o)ML4)/a2) x 100% OW=((dML-a(l -a)ML2+(1 -a)'ML4)/aZ) x 100% where ML, ML2, and ML4 represent the fraction of generated mintlactone with 0, 1, or 2 atoms " 0 . It should be noted that 2W+1W+OW=100~o. For reactions which were carried out under an " 0 , atmosphere the individual components were combined at -10°C to -20°C (methanol/ice). The system was then sealed and the atmosphere was repeatedly ( l o x ) evacuated and flushed with argon. After introduction of " 0 , gas the system was placed in a 37°C waterbath. In one of the vials 5 mM acetanilide (incorporates 0, to form paracetamol) was substituted as substrate to determine the fraction " 0 in molecular oxygen (b). Isotopic content in mintlactone and paracetamol (acetaminophen) were determined by g.1.c.-mass spectrometry. Reactions were terminated after 15min by snap freezing. The value of OW was calculated as (ML2/b) x 1 0 0 ~ o , whereas 1W and 2W could not be assessed. Mintlactone was also formed chemically from 1% a, a'-dimethoxydihydromenthofuran in H, " 0 water with 1yo glacial acetic acid ( 2 min at 200°C; Hirsch and Szur 1972) or from menthofuran and mCPBA (both 0.07 mM) in 50% acetonitrile in H,180 (10min at 0°C; Ruzo et al. 1985). In these systems the reported mechanism of lactone formation proceeds via a y-ketoenal and a furan epoxide intermediate, respectively. In both systems the water composition was probed with late addition of benzoyl chloride.
Results Inducers and inhibitors of cytochromes P-450 were used to establish that pulegone was not hepatotoxic itself, but was oxidized by cytochromes P-450 to toxic metabolites (Gordon et al. 1987, Mizutani et al. 1987, Madyastha and Moorthy 1989). Studies with an analogue of pulegone in which deuterium replaced hydrogens in the isopropylidene moiety identified menthofuran as a proximate toxic metabolite of pulegone based on the fact that the deuterated analogue formed menthofuran at about half the rate as pulegone itself, and that 2-3 times more pulegone-d6 than pulegone had to be administered to mice to cause the same extent of hepatic necrosis (Gordon et al. 1987). The mechanism of formation of menthofuran from pulegone was investigated by the use of I8O2and selective deuterium labelling. Incubations of mouse liver microsomes and pulegone in an atmosphere of 1802 led to the complete incorporation of the oxygen-18 label into the furan ring (Gordon et al. 1987). This result implicated hydroxylation by cytochromes P-450 of the isopropylidene methyl group syn to the ketone followed by intramolecular cyclization to a hemiketal and subsequent dehydration to the furan. If this were the mechanism, an analogue of pulegone in which deuterium was substituted for hydrogen only in the methyl group syn to the ketone should have yielded menthofuran that retained only one deuterium atom. However, we found that approximately 70% of the menthofuran formed from this pulegone-d, substrate retained the methyl group with three deuterium atoms (McClanahan et al. 1988). Subsequent experiments showed that hydroxylation of the methyl group anti to the ketone group had occurred to a considerable extent, and the data indicated that prior to its hydroxylation, topomerization of the isopropylidene group had taken place (figure 1). Consistent with this interpretation is our finding that the kinetic deuterium isotope effects for formation of menthofuran and the E-allylic alcohol are identical and independent of the isotopic composition of the substrate (table 1). Also consistent with initial hydrogen abstraction from the allylic methyl groups is the relatively high intramolecular deuterium kinetic isotope effect on the oxidation of the pulegone-d, substrate to menthofuran and the E-allylic alcohol (table 1).
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
1160
S . D. Nelson et al.
Pulegone-d6 also was used as a substrate to aid in characterizing the products of microsomal metabolism of pulegone. Interestingly, the overall rates of metabolism of pulegone and pulegone-d6 were not significantly different, but some products were formed at slower rates from pulegone-d6 while others were formed at faster rates (figure 2). This is indicative of isotopically sensitive branching (Jones et al. 1986), also termed metabolic switching (Miwa and L u 1987). We previously reported (McClanahan et al. 1989) that a y-ketoenal is a reactive metabolite formed from the proximate toxic metabolite of pulegone, menthofuran. In addition, we have identified diastereomeric mintlactones as microsomal oxidative metabolites of pulegone and menthofuran. Some possible mechanisms for the formation of mintlactone from menthofuran are depicted in figure 3. These include formation and rearrangement of a furan epoxide to a furanol, direct oxidation of the furan to the furonal, and formation of the furanol via an enonal, either by an intramolecular Cannizzaro reaction or by intramolecular condensation of the hydrated aldehyde with the ketone group. T h e formation of a furanol is proposed as a common intermediate because the amounts of diastereomeric mintlactones that were formed metabolically ( 85 % ( - )-mintlactone and 1 5 yo (+)-isomintlactone) are the same as those formed chemically and in plants (Takahashi et al. 1980). Results of investigations with pulegone-d6 showed retention of only three deuteriums on the lactone methyl group of the mintlactones formed as metabolites. This rules out an intramolecular Cannizzaro reaction in which an additional
-
-
Figure 1. Proposed mechanism of cytochrome P-450-catalysed oxidation of pulegone-d,. The brackets indicate possible oxidation states for the haem iron of cytochrome P-450.
+
Mechanisms of reactive metabolite formation from (R)- ( ) -pulegone
1161
Table 1. Kinetic deuterium isotope effects on the cytochrome P-450-mediated oxidations of deuterated pulegones to an E-allylic alcohol and menthofuran
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
kH/kD (observedy Substrates
R1
RZ
E-allylic alcohol
Pulegone-d3 Pulegone-d, Pulegone-d,
CD3 CD, CD,H
CH3 CD3 CD,H
1.18 f 0.1 2 1.94 f0.09
Menthofuran 1.22 f 0.10 1.89+_0.06 7.72 k0.74
7.69& 0.80
'Values are means f SD ( N = 5 ) .
5.2 (9.4)
t 1.7 (2.6)
X
10.4 (6.0)
1.8 (3.2) Figure 2 . Metabolites formed in mouse liver microsomes from pulegone (X = H) or pulegone-d, (X=D). Numbers are the percentages formed from the non-deuterated and deuterated (in parentheses) substrates. Wavy lines indicate unknown stereochemistry. Several other metabolites are formed that have either been only partially characterized or were not quantitated in this study.
S. D.Nelson et al.
1162
deuterium should have been retained in the lactone bridge position. Results of experiments that utilized either H2'*0 in the aqueous phase of microsomal incubations of menthofuran, or an '*02 atmosphere, revealed that most of the molecules of mintlactone (approximately 80%, table 2) contained two atoms of oxygen from water. Of the remaining molecules, most incorporated one atom of oxygen from molecular oxygen and one atom was retained from the menthofuran substrate.
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
Discussion (R)-(+)-Pulegone is the major constituent terpene of the mint oil obtained from the pennyroyal plant that is responsible for acute tissue organ damage, particularly hepatic centrilobular necrosis, in humans and laboratory animals. Results of studies with analogues of pulegone selectively labelled with deuterium showed that oxidation of the allylic methyl groups by hepatic cytochromes P-450 occurs most likely by hydrogen atom abstraction, based on the high intramolecular deuterium isotope effect observed in the formation of menthofuran and an Eallylic alcohol from pulegone-d, (table l). Although initial hydrogen abstraction apparently occurs from the allylic methyl groups, investigations with a trideuterated analogue (pulegone-d,) revealed that the tertiary radical is probably formed, and allows for topomerization of the allylic methyl groups (figure 1). Additional studies with pulegone-d6 revealed that several oxidative metabolites of pulegone are formed, and that perdeuteration of the allylic methyl groups leads to isotopically sensitive branching or metabolic switching of pathways (figure 2). Investigations with oxygen-18 also have provided insight into mechanisms of the oxidation of pulegone to a proximate toxic metabolite, menthofuran, and into mechanisms of the oxidation of menthofuran to diastereomeric mintlactones. Because the oxygen in the furan ring of menthofuran is derived solely from the atmospheric oxygen, we have proposed that menthofuran is formed by intramolecular condensation of a syn-allylic alcohol of pulegone with the ketone group to form a hemiketal that subsequently dehydrates to the thermodynamically more stable menthofuran (figure 1). Subsequently cytochrome P-450 oxidation of menthofuran to mintlactone is more complicated, and more than one pathway is involved. T h e major pathway,
\
Qmchrumc P-450
Figure 3.
Possible routes for the formation of diastereomeric mintlactones from menthofuran.
Mechanisms of reactive metabolite formation from Table 2.
(R)- ( + ) -pulegone 1163
Incorporation of oxygen atoms from water into mintlactone formed from menthofuran Relative compositionb(%)
Conditionsa
Label
0
1
2
N
Microsomes (from livers of phenobarbital-induced rats) CYP 2 B1 (P-450b)
Hzi80
-4k9 -4k2 -
84+11 __ 84k3
180,
20f3 21 k 2 16f1 15f2
-
3 3 4 3
Hzi80 Hzi80
5*1 9+1
8+1 -4+8
87f2 95f8
3 4
Chemical reactions Dimethoxymenthofuran Menthofuranlm-CPBA
1802
H2"0
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
~~
~
'Conditions are detailed in the Experimental section. bValues represent the relative composition (percentage) of mintlactone formed by the incorporation of 0,1 , or 2 atoms of oxygen from water in the incubation medium; means f SD.
based on results of studies with H,"0 and ''0, (table 2), leads to the incorporation of two atoms of oxygen derived from water. An intermediate in the pathway (figure 3), that is known to hydrate rapidly and incorporate oxygen from water, is the y-ketoenal formed by hydrolysis of dimethoxymenthofuran (Hirsch and Szur 1972). T h e mintlactone formed from dimethoxymenthofuran mostly contains oxygen derived from the aqueous medium (table 2). Although epoxidation of menthofuran may also lead to y-ketoenal formation followed by hydration and cyclization to mintlactone (table 2, figure 3), another pathway that involves either precursors to the epoxide, or epoxide rearrangement that does not proceed through the y-ketoenal, must be operative in systems containing cytochromes P-450. Approximately 20% of the mintlactone formed in incubations of menthofuran with either liver microsomes (from phenobarbitalpretreated rats) or purified cytochrome P-450b, (CUP 2B1) contain one atom of oxygen derived from atmospheric oxygen (table 2). In any mechanism, a final common intermediate is most likely a 2-hydroxyfuran (figure 3), inasmuch as diastereomeric mintlactones are formed in a ratio that is observed in chemical reactions that proceed through this tautomer. Thus, we have used stable isotopes to probe mechanisms of reactive metabolite formation from a natural product, a monoterpene. Substitution of deuterium for hydrogen was used in both tracer and kinetic experiments to characterize metabolite structures and reaction pathways.
Acknowledgements This work was supported by N I H Grant No. GM25418 (S.D.N.) and a fellowship from Deutsche Forschungsgemeinschaft (N.K.). References FEELY, T. M., and HARCREAVES, M. K., 1970, Circular dichroism studies. The pulegone epoxides. Journal of the Chemical Society (0,1745-1750. GLEASON, M. N., GOSSELIN, R. E., HODGE,H. C., and SMITH, R. P., 1969, Therapeutics Index, in Clinical Toxicology of Commercial Products, 3rd edn. (Baltimore: Williams & Wilkins), p. 109. GOLD,J., and CATES, W., 1980, Herbal abortifacients. Journal of the American Medical Association, 243, 1365-1366. GORDON, W. P., FORTE,A. J., MCMURTRY, R. J., GAL,J., and NELSON,S. D., 1982, Hepatotoxicity and pulmonary toxicity of pennyroyal oil and its constituent terpenes in the mouse. Toxicology and Applied Pharmacology, 65, 41 3-424.
+
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
1164 Mechanisms of reactive metabolite formation from (R)- ( ) -pulegone GORDON, W. P., HUITRIC,A. C., SETH, C. L., MCCLANAHAN, R. H., and NELSON,S. D., 1987, The metabolism of the abortifacient terpene, (R)-(+)-pulegone, to a proximate toxin, menthofuran. Drug Metabolism and Disposition, 15, 589-594. GRUNDSCHOBER, F., 1979, Literature review of pulegone. Perfumer FIaworist, 4, 15-1 7. GUNBY,P., 1979, Plant known for centuries still causes problems today. Journal of the American Medical Association, 241, 2246-2247. HALL,R. A,, and OSER,B. L., 1965, Recent progress in the consideration of flavoring ingredients under the food additives amendment 111 GRAS substances. Food Technology, 19, 253-271. HIRSCH,J . A., and SZUR,A. J., 1972, The hydrolysis of a,a’-dimethoxydihydrofurans.Journal of Heterocyclic Chemistry, 9, 523-529. JONES,J. P., KORZEKWA, K. R., RETTIE,A. E., and TRAGER, W. F., 1986, Isotopically sensitive branching and its effect on the observed intramolecular isotope effects in cytochrome P-450 catalyzed reactions: A new method for the estimation of intrinsic isotope effects. Journal of the American Chemical Society, 108, 7074-7078. MADYASTHA, K. M., and MOORTHY, B., 1989, Pulegone mediated hepatotoxicity: Evidence for covalent binding of (R)-(+)-[‘TI pulegone to microsomal proteins in witro. Chemico-Biological Znteractions, 12, 325-333. MCCLANAHAN, R. H., HUITRIC, A. C., PEARSON, P. G., DESPER,J. C., and NELSON, S. D., 1988, Evidence for a cytochrome P-450 catalyzed allylic rearrangement with double bond topomerization. Journal of the American Chemical Society, 110, 1979-1981. MCCLANAHAN, R. H., THOMASSEN, D., SLATTERY, J. T., and NELSON, S. D., 1989, Metabolic activation of (R)-(+)-pulegone to a reactive enonal that covalently binds to mouse liver proteins. Chemical Research in Toxicology, 2, 349-355. MIWA,C. T., and Lu, A. Y. H., 1987, Kinetic isotope effects and ‘metabolic switching’ in cytochrome P-450-catalyzed reactions. BioAssays, 7 , 21 5-219. MIZUTANI, T., NOMURA, H., NAKANISHI, K., and FUJITA, S., 1987, Effects of drug metabolism modifiers on pulegone-induced hepatotoxicity in mice. Research Communications in Chemical Pathology and Pharmacology, 58, 75-83. MOORTHY, B., MADYASTHA, P., and MADYASTHA, K. M., 1989, Metabolism of a monoterpene ketone, (R)-(+)-pulegone, a hepatotoxin in rat. Xenobiotica, 19, 217-224. NAKANISHI, O., FUJITANI, M., ICHIMOTO, I., and UEDA,H., 1980, An improved process for the synthesis of piperitenone from mesityloxide and methyl vinyl ketone. Agricultural and Biological Chemistry, 44, 1667-1671. NELSON, S . D., and GORDON, W. P., 1983, Mammalian drug metabolism. Journal ofNatural Products, 46, 71-78. Ruzo, L. O., CASIDA, J. E., and HOLDEN,I., 1985, Direct N.M.R. detection of an epoxyfuran intermediate in peracid oxidation of the fungicide methfuroxam. Journal of the Chemical Society, Chemical Communications, 1642-1643. SULLIVAN, J. B., RUMACK, B. H., THOMAS, H., PETERSON, R. G., and BRYSCH, P., 1979, Pennyroyal oil poisoning and hepatotoxicity. Journal of the American Medical Association, 242, 2873-2874. TAKAHASHI, K., SOMEYA, T., MURAKI,S., and YOSHIDA, T., A new keto-alcohol, (-)-mintlactone, ( )-isomintlactone and minor components in peppermint oil. Agricultural and Biological Chemistry, 44, 1535-1543. THOMASSEN D., SLATTERY, J. T., and NELSON,S. D., 1988, Contribution of menthofuran to the hepatotoxicity of pulegone: Assessment based on matched area under the curve and on matched time course. Journal of Pharmacology and Experimental Therapeutics, 224, 825-829. VALLANCE, W. B., 1955, Pennyroyal oil poisoning: a fatal case. Lancet, ii, 850-851. WATT,J . M., and BREYER-BRANDWIJK, M. G., 1962, The Medicinal and Poisonous Plants of Southern and Eastern Africa, 2nd edn (London: E. & S. Livingstone), p. 523.
+
1992, VOL. 22,
NOS
9/10, 1157-1164
Investigations of mechanisms of reactive metabolite formation from (R)-( + )-pulegone S. D. NELSON?, R. H. McCLANAHAN, D. THOMASSEN, W. PERRY GORDON and N. KNEBEL
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
Department of Medicinal Chemistry, BG-20, University of Washington, Seattle, WA 98195, USA
Received 5 October 1990; accepted 24 April 1992 1. (I?)-( +)-Pulegone is a monoterpene that is oxidized by cytochromes P-450 to reactive metabolites that initiate events in the pathogenesis of hepatotoxicity in mice, rats and humans. 2. Selective labelling of (R)-( )-pulegone with deuterium revealed that menthofuran was a proximate hepatotoxic metabolite formed by oxidation of the allylic methyl groups of pulegone. Incubations of pulegone with mouse liver microsomes in an atmosphere of "0, resulted in the formation of menthofuran that contained only oxygen-18 in the furan moiety. These results are consistent with oxidation of pulegone to an allylic alcohol that reacts intramolecularly with the ketone moiety to form a hemiketal that subsequently dehydrates to generate menthofuran. 3. Studies on the metabolism of menthofuran revealed that it is oxidized by cytochromes P-450 to an electrophilic y-ketoenal that reacts with nucleophilic groups on proteins to form covalent adducts. In addition, diastereomeric mintlactones are formed. Investigations with H,'*0 and "02are indicative of a furan epoxide intermediate, or a precursor, in the formation of the y-ketoenal and mintlactones.
+
Introduction (R)-(+)-Pulegone is a major monoterpene constituent of pennyroyal oil, a mint oil obtained from the leaves of the Mentha pulegium or Hedeoma pulegoides (Grundschober 1979). Pennyroyal oil is widely used as a fragrance and flavouring agent (Hall and Oser 1965), and has been used, based on folklore, as a herbal medicine to control menses and terminate pregnancy (Gleason et al. 1969). The ingestion of large quantities of the oil has been associated with toxic effects (Vallance 1955, Watt and Breyer-Brandwijk 1962, Gunby 1979, Sullivan et al. 1979, Gold and Cates 1980), and hepatotoxicity observed in humans has been modelled in mice (Gordon et al. 1982) and rats (Thomassen et at. 1988). (R)-( )-pulegone was identified as the major hepatotoxic terpene component of pennyroyal oil (Gordon et al. 1982), and studies on the biotransformation of pulegone in rodents have led to the identity of several oxidative and reduced metabolites (Gordon et al. 1982, 1987, Nelson and Gordon 1983, Moorthy et al. 1989). This report will review some of this work, and will describe how stable isotopes have been used to investigate mechanisms of formation of some reactive metabolites of pulegone.
+
t To whom correspondence should be addressed. 0049-8254/92 83.00
0 1992 Taylor & Francis Ltd.
1158
S . D.Nelson
et
al.
Experimental Syntheses, purifications, and analyses of pulegone, menthofuran and their stable isotope labelled compounds have been reported elsewhere (Gordon et al. 1987, McClanahan et al. 1988, 1989). Pulegone epoxides were prepared by oxidation of (R)-( )-pulegone with m-chloroperoxybenzoic acid (m-CPBA) as described previously (Feely and Hargreaves 1970), and we have reported their physical spectral characteristics and g.1.c. analysis elsewhere (Gordon et al. 1982). Piperitenone was synthesized by condensation of methyl vinyl ketone and mesityl oxide with Triton B as described by others (Nakanishi et al. 1980).
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
+
Characterization and quantification of metabolites Incubations were carried out in Teflon-sealed Reacti-flasks (25 ml) that contained 1 mM (R)-(+)pulegone, 6.0 mg of microsomal protein from mouse liver, an NADPH-generating system, and 3 ml of 0.05 M phosphate buffer, pH 7.4. The NADPH-generating system consisted of (final concentration): NADP' (0.35 mM), glucose 6-phosphate (8.77 mM), glucose-6-phosphate dehydrogenase (1 unitlml), and MgC1, (3 mM). Substrate was added in 2Oyl of freshly distilled dimethyl sulphoxide (DMSO). All additions were made into flasks that were maintained at 0°C. Reactions were started by transferring the vials to an incubator shaker that was maintained at 37°C. Reactions were terminated by immersion of the reaction vessels in ice and by the addition of 0.5 ml of acetone to each flask. Metabolites were assayed by the following procedure. ( +)-3-Methylcyclohexanone (0.5 pnol in 2Oyl of DMSO) was added as an internal standard to each incubation flask. Incubation mixtures were then extracted with 6ml of diethyl ether, and the aqueous layer was discarded. The ether extracts were washed with 3 ml of 1 M-HCI and then with 3 ml of 1 M-NaOH. The organic extract was dried over anhydr. Na2S04. A 5-ml sample was taken from each extract and was concentrated to approx. 0.5 ml under a stream of N,. Isooctane (0.5 ml) was added to each concentrate. The remainder of the ether was removed under a gentle stream of N,, and 1.0-2.0y1 of the remaining solution was then injected onto the g.1.c. of a g.1.c.-mass spectrometry system. The g.1.c. conditions were as follows: column-HP SE-54 fused silica capillary column, 0.35 mm x 50m; helium as carrier gas; head pressure, 2Opsi; splitless injection; injector temp. at 250°C at lO"C/min and held at 150°C for 15-20min. The column was connected to a FID detector or the column was run directly into the mass spectrometer source. Mass spectra were obtained on a VG 7070 mass spectrometer coupled to an HP 5840 g.1.c. instrument and a VG 2000 data system. Helium was used as the carrier gas at a head pressure of 15 psi. Samples were introduced by splitless injection at a temp. of 250°C and an oven temp. programme of 3min at 60°C then to 150°C at 10"C/min. The column was connected oia a direct inlet transfer line held at 250°C. Mass spectrometry conditions were: source temp., 200°C; electron impact 70eV; emission current, 200yA; multiplier at 1800-2200 V. Metabolites were identified by a comparison of their g.1.c. retention indices relative to the internal standard and by comparison of their mass spectra, both from pulegone-do and pulegone-d,. For example, 4-hydroxypulegone had a relative retention indice of 2.41 and major ions at m/z168 ( M t ) , 156(Mt-H20), 135(Mt-H,O-CH,), and 107(Mt-CH3-H20-CO). Loss of the methyl group was primarily from the isopropylidine group as noted in the spectrum of the corresponding metabolite from incubation of pulegone-d6 where loss of a -CD3 group was observed. Quantification of metabolites was carried out by peak area comparisons with 3methylcyclohexanone as the internal standard to determine absolute amounts formed over time in separate incubations of pulegone-do and pulegone-d,. Incubations of a 50 : 50 mixture of pulegone-do and pulegone-d, were analysed by g.1.c.-mass spectrometry of parent ions, which were major ions generated from all metabolites. The relative ratios of deuterated to non-deuterated metabolites were found to agree well with the ratios obtained by dividing the absolute amounts of the metabolites obtained by g.1.c. analysis of separate incubations of pulegone-do and pulegone-d,. '0-Incorporation studies Menthofuran used in these studies was 99.6% pure and free of mintlactone (g.1.c.-FID). Formation of mintlactone was carried out chemically or enzymically, in the presence of either H,"O or "0,. Reaction conditions were identical to those described above, but in a volume of 0.5ml for H,'80 incubations and 1.0ml for "02 incubations. Incubations with H,l80 were stopped by the addition of 25yl benzoyl chloride, which is rapidly hydrolysed to form benzoic acid. Reaction mixtures were analysed for the isotopic composition of mintlactone and benzoic acid by g.1.c.-mass spectrometry. Conditions of this assay will be reported elsewhere. "0-Incorporation in benzoic acid is proportional to the atom-fraction (a) of l80in water. The percentage of mintlactone formed by incorporation of 2, 1, or 0 water-derived oxygen atom(s) (2W, lW, and OW, respectively) was calculated as follows:
2W=(ML4/a2)x 100%
+
Mechanisms of reactive metabolite formation from (R)- ( ) -pulegone
1159
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
1W =((aML2 -2(1 -o)ML4)/a2) x 100% OW=((dML-a(l -a)ML2+(1 -a)'ML4)/aZ) x 100% where ML, ML2, and ML4 represent the fraction of generated mintlactone with 0, 1, or 2 atoms " 0 . It should be noted that 2W+1W+OW=100~o. For reactions which were carried out under an " 0 , atmosphere the individual components were combined at -10°C to -20°C (methanol/ice). The system was then sealed and the atmosphere was repeatedly ( l o x ) evacuated and flushed with argon. After introduction of " 0 , gas the system was placed in a 37°C waterbath. In one of the vials 5 mM acetanilide (incorporates 0, to form paracetamol) was substituted as substrate to determine the fraction " 0 in molecular oxygen (b). Isotopic content in mintlactone and paracetamol (acetaminophen) were determined by g.1.c.-mass spectrometry. Reactions were terminated after 15min by snap freezing. The value of OW was calculated as (ML2/b) x 1 0 0 ~ o , whereas 1W and 2W could not be assessed. Mintlactone was also formed chemically from 1% a, a'-dimethoxydihydromenthofuran in H, " 0 water with 1yo glacial acetic acid ( 2 min at 200°C; Hirsch and Szur 1972) or from menthofuran and mCPBA (both 0.07 mM) in 50% acetonitrile in H,180 (10min at 0°C; Ruzo et al. 1985). In these systems the reported mechanism of lactone formation proceeds via a y-ketoenal and a furan epoxide intermediate, respectively. In both systems the water composition was probed with late addition of benzoyl chloride.
Results Inducers and inhibitors of cytochromes P-450 were used to establish that pulegone was not hepatotoxic itself, but was oxidized by cytochromes P-450 to toxic metabolites (Gordon et al. 1987, Mizutani et al. 1987, Madyastha and Moorthy 1989). Studies with an analogue of pulegone in which deuterium replaced hydrogens in the isopropylidene moiety identified menthofuran as a proximate toxic metabolite of pulegone based on the fact that the deuterated analogue formed menthofuran at about half the rate as pulegone itself, and that 2-3 times more pulegone-d6 than pulegone had to be administered to mice to cause the same extent of hepatic necrosis (Gordon et al. 1987). The mechanism of formation of menthofuran from pulegone was investigated by the use of I8O2and selective deuterium labelling. Incubations of mouse liver microsomes and pulegone in an atmosphere of 1802 led to the complete incorporation of the oxygen-18 label into the furan ring (Gordon et al. 1987). This result implicated hydroxylation by cytochromes P-450 of the isopropylidene methyl group syn to the ketone followed by intramolecular cyclization to a hemiketal and subsequent dehydration to the furan. If this were the mechanism, an analogue of pulegone in which deuterium was substituted for hydrogen only in the methyl group syn to the ketone should have yielded menthofuran that retained only one deuterium atom. However, we found that approximately 70% of the menthofuran formed from this pulegone-d, substrate retained the methyl group with three deuterium atoms (McClanahan et al. 1988). Subsequent experiments showed that hydroxylation of the methyl group anti to the ketone group had occurred to a considerable extent, and the data indicated that prior to its hydroxylation, topomerization of the isopropylidene group had taken place (figure 1). Consistent with this interpretation is our finding that the kinetic deuterium isotope effects for formation of menthofuran and the E-allylic alcohol are identical and independent of the isotopic composition of the substrate (table 1). Also consistent with initial hydrogen abstraction from the allylic methyl groups is the relatively high intramolecular deuterium kinetic isotope effect on the oxidation of the pulegone-d, substrate to menthofuran and the E-allylic alcohol (table 1).
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
1160
S . D. Nelson et al.
Pulegone-d6 also was used as a substrate to aid in characterizing the products of microsomal metabolism of pulegone. Interestingly, the overall rates of metabolism of pulegone and pulegone-d6 were not significantly different, but some products were formed at slower rates from pulegone-d6 while others were formed at faster rates (figure 2). This is indicative of isotopically sensitive branching (Jones et al. 1986), also termed metabolic switching (Miwa and L u 1987). We previously reported (McClanahan et al. 1989) that a y-ketoenal is a reactive metabolite formed from the proximate toxic metabolite of pulegone, menthofuran. In addition, we have identified diastereomeric mintlactones as microsomal oxidative metabolites of pulegone and menthofuran. Some possible mechanisms for the formation of mintlactone from menthofuran are depicted in figure 3. These include formation and rearrangement of a furan epoxide to a furanol, direct oxidation of the furan to the furonal, and formation of the furanol via an enonal, either by an intramolecular Cannizzaro reaction or by intramolecular condensation of the hydrated aldehyde with the ketone group. T h e formation of a furanol is proposed as a common intermediate because the amounts of diastereomeric mintlactones that were formed metabolically ( 85 % ( - )-mintlactone and 1 5 yo (+)-isomintlactone) are the same as those formed chemically and in plants (Takahashi et al. 1980). Results of investigations with pulegone-d6 showed retention of only three deuteriums on the lactone methyl group of the mintlactones formed as metabolites. This rules out an intramolecular Cannizzaro reaction in which an additional
-
-
Figure 1. Proposed mechanism of cytochrome P-450-catalysed oxidation of pulegone-d,. The brackets indicate possible oxidation states for the haem iron of cytochrome P-450.
+
Mechanisms of reactive metabolite formation from (R)- ( ) -pulegone
1161
Table 1. Kinetic deuterium isotope effects on the cytochrome P-450-mediated oxidations of deuterated pulegones to an E-allylic alcohol and menthofuran
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
kH/kD (observedy Substrates
R1
RZ
E-allylic alcohol
Pulegone-d3 Pulegone-d, Pulegone-d,
CD3 CD, CD,H
CH3 CD3 CD,H
1.18 f 0.1 2 1.94 f0.09
Menthofuran 1.22 f 0.10 1.89+_0.06 7.72 k0.74
7.69& 0.80
'Values are means f SD ( N = 5 ) .
5.2 (9.4)
t 1.7 (2.6)
X
10.4 (6.0)
1.8 (3.2) Figure 2 . Metabolites formed in mouse liver microsomes from pulegone (X = H) or pulegone-d, (X=D). Numbers are the percentages formed from the non-deuterated and deuterated (in parentheses) substrates. Wavy lines indicate unknown stereochemistry. Several other metabolites are formed that have either been only partially characterized or were not quantitated in this study.
S. D.Nelson et al.
1162
deuterium should have been retained in the lactone bridge position. Results of experiments that utilized either H2'*0 in the aqueous phase of microsomal incubations of menthofuran, or an '*02 atmosphere, revealed that most of the molecules of mintlactone (approximately 80%, table 2) contained two atoms of oxygen from water. Of the remaining molecules, most incorporated one atom of oxygen from molecular oxygen and one atom was retained from the menthofuran substrate.
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
Discussion (R)-(+)-Pulegone is the major constituent terpene of the mint oil obtained from the pennyroyal plant that is responsible for acute tissue organ damage, particularly hepatic centrilobular necrosis, in humans and laboratory animals. Results of studies with analogues of pulegone selectively labelled with deuterium showed that oxidation of the allylic methyl groups by hepatic cytochromes P-450 occurs most likely by hydrogen atom abstraction, based on the high intramolecular deuterium isotope effect observed in the formation of menthofuran and an Eallylic alcohol from pulegone-d, (table l). Although initial hydrogen abstraction apparently occurs from the allylic methyl groups, investigations with a trideuterated analogue (pulegone-d,) revealed that the tertiary radical is probably formed, and allows for topomerization of the allylic methyl groups (figure 1). Additional studies with pulegone-d6 revealed that several oxidative metabolites of pulegone are formed, and that perdeuteration of the allylic methyl groups leads to isotopically sensitive branching or metabolic switching of pathways (figure 2). Investigations with oxygen-18 also have provided insight into mechanisms of the oxidation of pulegone to a proximate toxic metabolite, menthofuran, and into mechanisms of the oxidation of menthofuran to diastereomeric mintlactones. Because the oxygen in the furan ring of menthofuran is derived solely from the atmospheric oxygen, we have proposed that menthofuran is formed by intramolecular condensation of a syn-allylic alcohol of pulegone with the ketone group to form a hemiketal that subsequently dehydrates to the thermodynamically more stable menthofuran (figure 1). Subsequently cytochrome P-450 oxidation of menthofuran to mintlactone is more complicated, and more than one pathway is involved. T h e major pathway,
\
Qmchrumc P-450
Figure 3.
Possible routes for the formation of diastereomeric mintlactones from menthofuran.
Mechanisms of reactive metabolite formation from Table 2.
(R)- ( + ) -pulegone 1163
Incorporation of oxygen atoms from water into mintlactone formed from menthofuran Relative compositionb(%)
Conditionsa
Label
0
1
2
N
Microsomes (from livers of phenobarbital-induced rats) CYP 2 B1 (P-450b)
Hzi80
-4k9 -4k2 -
84+11 __ 84k3
180,
20f3 21 k 2 16f1 15f2
-
3 3 4 3
Hzi80 Hzi80
5*1 9+1
8+1 -4+8
87f2 95f8
3 4
Chemical reactions Dimethoxymenthofuran Menthofuranlm-CPBA
1802
H2"0
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
~~
~
'Conditions are detailed in the Experimental section. bValues represent the relative composition (percentage) of mintlactone formed by the incorporation of 0,1 , or 2 atoms of oxygen from water in the incubation medium; means f SD.
based on results of studies with H,"0 and ''0, (table 2), leads to the incorporation of two atoms of oxygen derived from water. An intermediate in the pathway (figure 3), that is known to hydrate rapidly and incorporate oxygen from water, is the y-ketoenal formed by hydrolysis of dimethoxymenthofuran (Hirsch and Szur 1972). T h e mintlactone formed from dimethoxymenthofuran mostly contains oxygen derived from the aqueous medium (table 2). Although epoxidation of menthofuran may also lead to y-ketoenal formation followed by hydration and cyclization to mintlactone (table 2, figure 3), another pathway that involves either precursors to the epoxide, or epoxide rearrangement that does not proceed through the y-ketoenal, must be operative in systems containing cytochromes P-450. Approximately 20% of the mintlactone formed in incubations of menthofuran with either liver microsomes (from phenobarbitalpretreated rats) or purified cytochrome P-450b, (CUP 2B1) contain one atom of oxygen derived from atmospheric oxygen (table 2). In any mechanism, a final common intermediate is most likely a 2-hydroxyfuran (figure 3), inasmuch as diastereomeric mintlactones are formed in a ratio that is observed in chemical reactions that proceed through this tautomer. Thus, we have used stable isotopes to probe mechanisms of reactive metabolite formation from a natural product, a monoterpene. Substitution of deuterium for hydrogen was used in both tracer and kinetic experiments to characterize metabolite structures and reaction pathways.
Acknowledgements This work was supported by N I H Grant No. GM25418 (S.D.N.) and a fellowship from Deutsche Forschungsgemeinschaft (N.K.). References FEELY, T. M., and HARCREAVES, M. K., 1970, Circular dichroism studies. The pulegone epoxides. Journal of the Chemical Society (0,1745-1750. GLEASON, M. N., GOSSELIN, R. E., HODGE,H. C., and SMITH, R. P., 1969, Therapeutics Index, in Clinical Toxicology of Commercial Products, 3rd edn. (Baltimore: Williams & Wilkins), p. 109. GOLD,J., and CATES, W., 1980, Herbal abortifacients. Journal of the American Medical Association, 243, 1365-1366. GORDON, W. P., FORTE,A. J., MCMURTRY, R. J., GAL,J., and NELSON,S. D., 1982, Hepatotoxicity and pulmonary toxicity of pennyroyal oil and its constituent terpenes in the mouse. Toxicology and Applied Pharmacology, 65, 41 3-424.
+
Xenobiotica Downloaded from informahealthcare.com by University of Sydney on 08/27/14 For personal use only.
1164 Mechanisms of reactive metabolite formation from (R)- ( ) -pulegone GORDON, W. P., HUITRIC,A. C., SETH, C. L., MCCLANAHAN, R. H., and NELSON,S. D., 1987, The metabolism of the abortifacient terpene, (R)-(+)-pulegone, to a proximate toxin, menthofuran. Drug Metabolism and Disposition, 15, 589-594. GRUNDSCHOBER, F., 1979, Literature review of pulegone. Perfumer FIaworist, 4, 15-1 7. GUNBY,P., 1979, Plant known for centuries still causes problems today. Journal of the American Medical Association, 241, 2246-2247. HALL,R. A,, and OSER,B. L., 1965, Recent progress in the consideration of flavoring ingredients under the food additives amendment 111 GRAS substances. Food Technology, 19, 253-271. HIRSCH,J . A., and SZUR,A. J., 1972, The hydrolysis of a,a’-dimethoxydihydrofurans.Journal of Heterocyclic Chemistry, 9, 523-529. JONES,J. P., KORZEKWA, K. R., RETTIE,A. E., and TRAGER, W. F., 1986, Isotopically sensitive branching and its effect on the observed intramolecular isotope effects in cytochrome P-450 catalyzed reactions: A new method for the estimation of intrinsic isotope effects. Journal of the American Chemical Society, 108, 7074-7078. MADYASTHA, K. M., and MOORTHY, B., 1989, Pulegone mediated hepatotoxicity: Evidence for covalent binding of (R)-(+)-[‘TI pulegone to microsomal proteins in witro. Chemico-Biological Znteractions, 12, 325-333. MCCLANAHAN, R. H., HUITRIC, A. C., PEARSON, P. G., DESPER,J. C., and NELSON, S. D., 1988, Evidence for a cytochrome P-450 catalyzed allylic rearrangement with double bond topomerization. Journal of the American Chemical Society, 110, 1979-1981. MCCLANAHAN, R. H., THOMASSEN, D., SLATTERY, J. T., and NELSON, S. D., 1989, Metabolic activation of (R)-(+)-pulegone to a reactive enonal that covalently binds to mouse liver proteins. Chemical Research in Toxicology, 2, 349-355. MIWA,C. T., and Lu, A. Y. H., 1987, Kinetic isotope effects and ‘metabolic switching’ in cytochrome P-450-catalyzed reactions. BioAssays, 7 , 21 5-219. MIZUTANI, T., NOMURA, H., NAKANISHI, K., and FUJITA, S., 1987, Effects of drug metabolism modifiers on pulegone-induced hepatotoxicity in mice. Research Communications in Chemical Pathology and Pharmacology, 58, 75-83. MOORTHY, B., MADYASTHA, P., and MADYASTHA, K. M., 1989, Metabolism of a monoterpene ketone, (R)-(+)-pulegone, a hepatotoxin in rat. Xenobiotica, 19, 217-224. NAKANISHI, O., FUJITANI, M., ICHIMOTO, I., and UEDA,H., 1980, An improved process for the synthesis of piperitenone from mesityloxide and methyl vinyl ketone. Agricultural and Biological Chemistry, 44, 1667-1671. NELSON, S . D., and GORDON, W. P., 1983, Mammalian drug metabolism. Journal ofNatural Products, 46, 71-78. Ruzo, L. O., CASIDA, J. E., and HOLDEN,I., 1985, Direct N.M.R. detection of an epoxyfuran intermediate in peracid oxidation of the fungicide methfuroxam. Journal of the Chemical Society, Chemical Communications, 1642-1643. SULLIVAN, J. B., RUMACK, B. H., THOMAS, H., PETERSON, R. G., and BRYSCH, P., 1979, Pennyroyal oil poisoning and hepatotoxicity. Journal of the American Medical Association, 242, 2873-2874. TAKAHASHI, K., SOMEYA, T., MURAKI,S., and YOSHIDA, T., A new keto-alcohol, (-)-mintlactone, ( )-isomintlactone and minor components in peppermint oil. Agricultural and Biological Chemistry, 44, 1535-1543. THOMASSEN D., SLATTERY, J. T., and NELSON,S. D., 1988, Contribution of menthofuran to the hepatotoxicity of pulegone: Assessment based on matched area under the curve and on matched time course. Journal of Pharmacology and Experimental Therapeutics, 224, 825-829. VALLANCE, W. B., 1955, Pennyroyal oil poisoning: a fatal case. Lancet, ii, 850-851. WATT,J . M., and BREYER-BRANDWIJK, M. G., 1962, The Medicinal and Poisonous Plants of Southern and Eastern Africa, 2nd edn (London: E. & S. Livingstone), p. 523.
+
