Food Chemistry 134 (2012) 1096–1105

Contents lists available at SciVerse ScienceDirect

Food Chemistry

journal homepage: www.elsevier.com/locate/foodchem

Metabolic pathways of the psychotropic-carboline , harmaline and , by liquid chromatography/mass spectrometry and NMR spectroscopy

Ting Zhao a, Shan-Song Zheng a, Bin-Feng Zhang a,b,c, Yuan-Yuan Li a, S.W. Annie Bligh d, ⇑ ⇑ Chang-Hong Wang a,b,c, , Zheng-Tao Wang a,b,c, a Institute of Chinese Materia Medica, Shanghai University of Traditional Chinese Medicine, 1200 Cailun Road, Shanghai 201210, China b The MOE Key Laboratory for Standardization of Chinese Medicines and The SATCM Key Laboratory for New Resources and Quality Evaluation of Chinese Medicines, 1200 Cailun Road, Shanghai 201210, China c Shanghai R&D Center for Standardization of Chinese Medicines, 199 Guoshoujing Road, Shanghai 201210, China d Institute for Health Research and Policy, London Metropolitan University, 166-220 Holloway Road, London N7 8DB, UK article info abstract

Article history: The b-carboline alkaloids, harmaline and harmine, are present in hallucinogenic plants and Received 3 June 2011 , and in a variety of foods. In order to establish the metabolic pathway and bioactivities Received in revised form 25 January 2012 of endogenous and xenobiotic bioactive b-carbolines, high-performance liquid chromatography, coupled Accepted 6 March 2012 with mass spectrometry, was used to identify these metabolites in human liver microsomes (HLMs) Available online 16 March 2012 in vitro and in rat urine and bile samples after oral administration of the alkaloids. Three metabolites of harmaline and two of harmine were found in the HLMs. Nine metabolites for harmaline and seven Keywords: metabolites for harmine, from the rat urine and bile samples, were identified. Among them, four Harmaline in vivo metabolites were isolated and fully characterised by NMR analysis. For the first time, harmaline Harmine Metabolites is shown transforming to harmine by oxidative dehydrogenation in rat. Five metabolic pathways were b-Carboline alkaloids therefore proposed, namely, oxidative dehydrogenation, 7-O-demethylation, hydroxylation, O-glucuro- nide conjugation and O-sulphate conjugation. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction such as , harmine, harmaline and can also be found in common plant-derived foodstuffs (e.g. wheat, rice, corn, The b-carboline alkaloids, harmaline and harmine, are the main barely, soy, beans, rye, grapes, mushrooms and vinegar), well- active components of the hallucinogenic plant, Ayahuasca, which is cooked meat, plant-derived beverages (e.g. wine, beer, whisky, used as an ingredient of the popular sacred and psychoactive brandy and sake), and plant-derived inhaled substances (e.g. to- drinks in South American Indian cultures. It is also known as Caapi, bacco) (Agüía, Peña-Farfala, Yáñez-Sedeño, & Pingarróna, 2007; Pinde, Natema or Yaje, which is widely used for prophecy, divina- Alves, Mendes, Oliveira, & Casal, 2010; Conning, 1989; Crotti, tion, and as a sacrament in the northern part of South America Gates, Lopes, & Lopes, 2010; Derakhshanfar, Oloumi, & Mirzaie, (Samoylenko et al., 2010). The b-carboline alkaloids also present 2010; Guan, Louis, & Zheng, 2001; Herraiz, Guillén, & Arán, in Peganum harmala, which has traditionally been used for ritual 2008). Because of their natural presence in the food chain and envi- and medicinal preparations in the Middle East, central Asia, and ronment, there appears to be a risk associated with extra exposure South America. According to previous reports, the entire plant to these b-carbolines from dietary sources, smoking, and consump- and seeds of P. harmala have been separately used to treat diseases tion of alcoholic beverages. In fact, the occurrence of b-carbolines such as cough, asthma, rheumatoid arthritis and swelling pain in human blood and excreta, beef and sardines under normal phys- (Cheng et al., 2010; Hemmateenejad, Abbaspour, Maghamia, Miri, iological conditions has been reported in the literature, which & Panjehshahin, 2006; Kartal, Altun, & Kurucu, 2003; Samoylenko implies that b-carboline alkaloids may be present in biological sys- et al., 2010; Sourkes, 1999; Zheng et al., 2009). The b-carbolines, tems (Guan et al., 2001; Herath, Mikell, Ferreira, & Khan, 2003; Riba et al., 2003). The b-carboline alkaloids have been of interest due to their ⇑ Corresponding authors. Address: The MOE Key Laboratory for Standardization psychotropic properties. These compounds affect the content of of Chinese Medicines, Shanghai University of Traditional Chinese Medicine, 1200 neurotransmitters by strong reversible inhibition of monoamine Cailun Road, Zhangjiang Hi-Tech Park, Shanghai 201210, China. Tel.: +86 21 oxidase (Kim, Sablin, & Ramsay, 1997; Schwarz, Houghton, Rose, 51322511; fax: +86 21 51322519. E-mail addresses: [email protected] (C.-H. Wang), [email protected] Jenner, & Lees, 2003) and the inhibition of (Z.-T. Wang). (Zheng et al., 2009, 2011). They also exhibit some pharmacological

0308-8146/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodchem.2012.03.024 T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105 1097

actions, such as anti-tumour and analgesic effects (Farouk, Laroubi, donors (Mongolian). Reverse-phase silica gel (C18) and MCI Gel Aboufatima, Benharref, & Chait, 2008; Jahaniani, Ebrahimi, CHP20P(75–150 lm) for column chromatography were purchased

Rahbar-Roshandel, & Mahmoudian, 2005; Wang, Liu, & Zheng, from Mitsubishi (Tokyo, Japan). The MG II C18 column (75 mm 2002), vasorelaxant activities (Astulla et al., 2008), and antimicro- 2.0 mm, i.d. 3 lm) was purchased from Shiseido (Tokyo, Japan). bial properties (Arshad, Neubauer, Hasnain, & Hess, 2008). Hence, DMSO-d6 was supplied by Armar Chemicals (Switzerland). b-carboline alkaloids have potential in the treatment of psychiatric A surveyor HPLC system (Thermo-Finnigan, San Jose, CA, USA) disorders and diseases. However, these compounds have some was used for sample separation. A Finnigan LCQ DECA XP, plus toxic effects especially in the central nervous system. For instance, ion-trap mass spectrometer equipped with an electrospray the structures of the N-methylated b-carboline alkaloids resemble ionisation (ESI) source, was used for mass analysis and detection N-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and partic- (Thermo-Finnigan, San Jose, CA, USA). High resolution mass spec- ularly its neurotoxic metabolite, N-methyl-4-phenylpyridinium trometry (HRMS) was recorded using an ion mobility spectrometry (MPP+), may act as an endogenous or environmental neurotoxic (IMS)-mass spectrometry (MS) instrument (SYNAPT G2, Waters agent (Cobuzzi, Neafsey, & Collins, 1994). It was also found that Technologies, Milford, USA). A Bruker AV 400 MHz NMR spectrom- b-carboline alkaloids were co-mutagens in the presence of aro- eter (Faellanden, Switzerland) was used to record 1HNMR matic amines such as aniline, toluidine and inhibitors of mutagens (400 MHz), 13C NMR (100 MHz), HMBC and HSQC spectra. and carcinogens in mutagenic assays (Herraiz et al., 2008). The metabolic outcome and potential bio-activation of endoge- 2.2. Microsomal incubation with NADPH nous and xenobiotic bioactive b-carbolines are matters of research interest. However, not much is currently known about their Each incubation was performed in a 100 mM phosphate buffer . A previous study found that the main metabolic path- at pH 7.4 containing HLMs (final protein concentrations were ways of harmaline and harmine were O-demethylation mediated 1 mg/ml), the NADPH-generating system contained 10 mM glucose by CYP 2D6, CYP1A2 and CYP2C9, and hydroxylation mediated 6-phosphate, 1 mM NADP+, 4 mM magnesium chloride, and 1 unit/ by CYP1A2 and CYP1A1 (Yu, Idle, Krausz, Küpfer, & Gonzalez, ml of glucose 6-phosphate dehydrogenase, and various alkaloids 2003). Both in vivo and in vitro studies have revealed that the major (harmaline, harmine, harmalol, harmol) which were previously fate of harmaline and harmine was demethylation to form harma- dissolved in methanol (final methanol concentration was 1%, v/v) lol and harmol, respectively, which were subsequently excreted as with concentration 200 lM in a total volume of 100 ll. The reac- the glucuronide conjugate and sulphate conjugate (Ho, Estevez, tion system was incubated for 3 min at 37 °C before the reaction Fritchie, & Tansey, 1971; Mulder & Hagedoorn, 1974; Slotkin & was initiated by the addition of NADP+, and the incubation contin- DiStefano, 1970a, 1970b; Slotkin, DiStefano, & Au, 1970). Two ued at 37 °C for 60 min. The reaction was quenched by adding the hydroxylated metabolites were also found in human or mouse liver same volume of ice-cold acetonitrile and the reaction system was microsomes (Tweedie & Burke, 1987; Yu et al., 2003), but the posi- placed in an ice bath for 30 min. The incubation mixtures were tion of the hydroxyl group could not be definitively assigned. These then centrifuged for 10 min at 20,000g and an aliquot of the super- results are of further interest because of the biological features of natant was injected for LC-MSn analysis. these xenobiotic and endogenous bioactive alkaloids and their potential interactions with mutagens and carcinogens. Unambiguous identification of metabolites can usually only be 2.3. Microsomal incubation with UDPGA accomplished by spectroscopic characterisation, especially by MS and NMR experiments (Chen et al., 2010; Zhou et al., 2010). In this HLMs, at 1 mg/ml, and Brij, at 0.05 mg/ml, were mixed together study, liquid chromatography (LC) electrospray ionisation (ESI) and pre-incubated on ice for 5 min. The (harmaline, har- tandem mass spectrometry (MSn) was used as an analytical and mine, harmalol, harmol) was added to give a final concentration characteristic tool. The mass spectral data obtained for the metab- of 200 lM and the incubation mixture, including HLMs at 1 mg/ olites were compared with those of the available standards. MS ml, Brij at 0.05 mg/ml, MgCl2 (4 mM), Tris–HCl buffer with pH data, as well as NMR spectra, were obtained for structural charac- 7.4 (50 mM), was warmed to 37 °C and the reaction was initiated terisation of metabolites of harmaline and harmine in human liver by addition of UDPGA (5 mM). The same volume of ice-cold aceto- microsomes (HLMs) in vitro and in rat urine and bile samples nitrile was added to the reaction mixture after 1 h of incubation. in vivo. Based on the present study, the proposed metabolic path- The samples were centrifuged at 20,000g for 10 min at 4 °C and n ways of harmaline and harmine in rat can be established and the the supernatant was analysed by LC-MS for identification of metabolic pathways in human can also be deduced. metabolites.

2.4. Animals and drug administration 2. Materials and methods Male Sprague–Dawley rats, weighing 250–300 g, were provided 2.1. Materials, chemicals, reagents and instrument by the Experimental Animal Centre, Shanghai University of Tradi- tional Chinese Medicine, China. Animals were kept in an environ- Harmaline and harmine were obtained from TCI EUROPE (TCI, mentally controlled breeding room for 7 days before commencing Tokyo, Japan), harmalol hydrochloride dehydrate and harmol were the experiments. Rats were fed with standard laboratory food obtained from wako Pure Chemical Industries (Ltd., Osaka, Japan). and water ad libitum and fasted for 12 h but with access to water Glucose 6-phosphate, NADP+, UDPGA, Brij and glucose 6-phos- prior to the administration of harmaline or harmine. All studies phate dehydrogenase were obtained from Sigma–Aldrich (St. Louis, on animals were in accordance with the guidelines of the Commit- MO, USA). HPLC-grade acetonitrile was supplied by Merck tee on the Care and Use of Laboratory Animals in China. Harmaline (Damstadt, Germany). Deionised water was purified using a (2 mg/ml) and harmine (4 mg/ml) were dissolved in deionised Milli-Q system (Millipore, Milford, MA, USA). Formic acid, triethyl- water and administered, by gavage, at a dose of 20 mg/kg and amine and potassium dihydrogen phosphate were all of analytical 40 mg/kg body weight, respectively, which were one tenth of the grade. Human liver microsomes (HLMs) were supplied by the median lethal dose (LD50) by our previous acute toxicity test data RuiDe Research Institute for Liver Disease (Shanghai) Co. Ltd (unpublished data) and literature data reported by Massoud, (Shanghai, China) and samples were pooled from 10 individual Hossein, and Pirooz (2002). 1098 T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105

2.5. Samples collection and processing procedures MSn modes in the range of m/z 100–1000. All operations were controlled by the Xcalibur software version 1.2 (Finnigan). The urine samples were collected from 0 to 24 h after oral A Bruker AV 400 NMR spectrometer (Faellanden, Switzerland) administration of harmaline or harmine, and the blank urine sam- was used to record 1H NMR (400 MHz), 13C NMR (100 MHz), HMBC ples were collected before oral administration. For bile sampling, and HSQC spectra in DMSO at 25 °C. an abdominal incision was made and the common bile duct was cannulated with a Closed IV Catheter System (ID = 0.07 cm, Becton 3. Results and discussion Dickinson, USA) for the collection of bile samples. Bile samples were collected from 0 to 6 h after oral administration of harmaline 3.1. Mass spectral analysis of harmaline, harmine, harmol, harmalol or harmine. Blank rat bile was collected before oral administration of alkaloids. All the samples were stored at 20 °C until analysis. The MSn fragmentation patterns of harmaline and harmine The same volume of acetonitrile was mixed with the urine were examined to aid a better understanding of the MSn spectra (1 ml), thoroughly, and then centrifuged at 3000g for 10 min. The of the metabolites. As indicated in Fig. 1, the protonated parent supernatant (1 ml) was evaporated to dryness under nitrogen at molecules, at m/z 215, for harmaline and m/z 213 for harmine, were 37 °C. The residue of the supernatant was reconstituted in 200 ll observed. The MS2 spectra of m/z 215 displayed the ions at m/z 200 of 5% acetonitrile, and then centrifuged at 20,000g for 10 min. and 174 (Fig. 1A1), and MS3 spectra of the ion at m/z 174 gave the The supernatant was injected into the chromatographic systems product ion at m/z 159 (Fig. 1A2). The MS2 spectra of the ion at m/z for separation and identification of the metabolites. The bile 213 displayed product ions at m/z 198 (Fig. 1B1), and the fragment samples were processed with the same pre-treatment method as ion at m/z 170 (Fig. 1B2) was the MS3 product ion. As the main the urine samples. metabolites of harmaline and harmine, fragmentation behaviors The urine samples used for isolating and purification metabo- of harmalol and harmol were also studied. The protonated molec- lites were collected from 30 rats for each group which were admin- ular ion showed m/z 201 for harmalol and m/z 199 for harmol. The istered with harmaline and harmine twice a day, over 30 protonated molecular ion, m/z 201, gave two major product ions at consecutive days, by gavage, at dose rates of 20 mg/kg and m/z 160 and m/z 184 in the MS2 spectra (Fig. 1C1), and the MS3 40 mg/kg body weight, respectively. In total, about 8 l of urine spectra of m/z 160 displayed an ion at m/z 132 (Fig. 1C2). A full for each group were collected and centrifuged at 4000g for product ion scan of harmol with the precursor ion m/z 199, corre- 15 min, and then the supernatant was stored at 20 °C until used. sponding to [M+H]+, resulted in the main MS2 fragment at m/z 171 (Fig. 1D1). 2.6. Isolation and purification of urinary metabolites

About 8 l of urine samples (collected from the rats after oral 3.2. Metabolism of harmaline and harmine by human liver administration of harmaline or harmine) were subjected to MCI microsomes (HLMs) Gel CHP 20P column chromatography (4.5 60 cm; 1000 ml) and The b-carboline alkaloids, harmaline (M00) and harmine (M0) eluted with a gradient of MeOH-H2O (0:100; 5:95; 10:90; 20:80; 30:70). The metabolites were mainly obtained from 5%, 10% and (200 lM, respectively), were incubated separately with HLMs. The various metabolites were detected by HPLC-MS (electrospray 20% MeOH–H2O fractions. The fractions were concentrated in va- cuo to yield residues. The residues were dissolved in a small ionisation) as O-demethylated metabolites and hydroxylation amount of water, and then subjected to a RP-18 silica gel column products. Thus, harmaline (M00) metabolised by CYP450 isozymes in HLMs gave three major metabolites M1, M2 and M3 (Figs. 2 and and washed with MeOH–H2O by gradient elution (5% MeOH for M7 and M8, 8% MeOH for M9 and M12). The metabolites were fur- 3). M1 was identified on the basis of its mass spectra (ESI), UV ther purified by semi-preparative HPLC. absorption spectra (photodiode array detection of chromato- Semi-preparative HPLC was performed with an ODS column graphic peaks), and co-elution with standard of harmalol. This also ZORBAX SB-C18 (9.4 mm 25 cm, 5 lm, Agilent, USA) at 30 °Cin suggested that M2 and M3 may be hydroxylated metabolites (the a Waters 600 liquid chromatography apparatus equipped with a exact hydroxylation positions were not determined), by mass and Waters 2996 PDA detector (Waters, Milford, MA). The detection UV absorption spectra, consistent with previous studies (Herraiz wavelength was set at 245 nm. The mobile phase consisted of A et al., 2008; Tweedie & Burke, 1987; Yu et al., 2003). It was found (acetonitrile) and B (aqueous 0.1% formic acid) at a flow rate of that the parent compound, harmaline, did not participate in the 3 ml/min, using isocratic elution: 7% A for M7 and M8, 12% A for phase II reaction. However, harmalol (M1), the phase I metabolite M9, and 13% A for M12. As a result, 5 mg of M7, 5 mg of M8, of harmaline, conjugated with glucuronate and was transformed to 10 mg of M9 and 40 mg of M12 were obtained. a metabolite (M6) in the HLMs incubation system with UDPGA, which was consistent with a previous report (Mulder & Hagedoorn, 2.7. Chromatographic and spectroscopic methods 1974). The same metabolic pattern was observed for biotransforma- The metabolites were separated and measured on a Finnigan tion of harmine (M0) by CYP450 isozymes in HLMs that afforded Deca XP plus LCQ HPLC-ESI(APCI)-Ion trap MS using a maintained M4 which was identified as harmol. In addition, CYP450 isozymes oxidised harmine to hydroxyl-harmine (M5). Harmol glucuronide MG II C18 column (75 mm 2.0 mm, i.d. 3 lm) at 30 °C. The mobile phase consisted of A (acetonitrile) and B (aqueous 0.1 % formic (M7) could be detected in the HLMs incubation system of harmol acid) at a flow rate of 0.3 ml/min, using gradient elution: 0– with UDPGA (Figs. 2 and 3). 30 min (5–20% A). A Finnigan LCQ DECA XP plus ion-trap mass spectrometer, equipped with an electrospray ionisation (ESI) inter- 3.3. Metabolites detected in the urine and bile of the rat after oral face operating in positive ion mode, was used for analytical detec- administration of harmaline and harmine tion. The tuning parameters were optimised and set as follows: electrospray needle voltage, 5 kV; sheath gas flow, 20 arbitrary 3.3.1. General units; auxiliary nitrogen gas flow, 5 arbitrary units; capillary volt- UV spectra and the full-scan mass spectra of rat urine and bile age, 9 V; heated capillary temperature of 300 °C. Data acquisition samples after oral administration of harmaline or harmine were was performed in full-scan, selective ions monitoring (SIM) and compared with those of blank urine and bile samples for the T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105 1099

Fig. 1. MSn spectra of alkaloids harmaline (A1,A2), harmine (B1,B2), harmalol (C1,C2) and harmol (D1), using ESI in positive ion mode. identification of possible metabolites. Carefully screening the data the rat urine, except for M11, which was only found in rat bile. collected in the HPLC-PDA-ESI/MSn experiments resulted in the In addition, M1, M5–M9 were also detected in rat bile. Seven discovery of nine metabolites (M0, M1, M4–M9, M11) for harma- metabolites (M4, M5, M7–M10, and M12) for harmine in the rat line and seven metabolites (M4, M5, M7–M10, M12) for harmine urine were found. However, only M7, M8 and M9 could be detected (Figs. 2 and 4). All the metabolites of harmaline were detected in in the rat bile. Their [M+H]+ ions were at m/z 213, 201, 199, 229, 1100 T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105

Fig. 2. Structures of harmaline (M00), harmine (M0) and metabolites as well as the proposed major metabolic pathways of harmaline and harmine.

377, 375, 405, 279, 309, 281, 295 for M0, M1, M4, M5, M6, M7, M8, oxygen atom, suggesting that M5 was hydroxyl-harmine (the exact

M9, M10, M11, M12, respectively. The kmax values observed in the hydroxylation positions were not determined) (Fig. 2). However, it UV spectra of harmaline (M00) and harmalol (M1) were at 260 and may be possible to deduce the existence of phase I metabolites M5- 370 nm, harmine (M0), harmol (M4) and M5 were at 245 and a (5-hydroxylate harmine) and M5-b (3-hydroxylate harmine) 320 nm, suggesting that the metabolites were similar to the chro- from the results for the chemical structures of M8 and M12. Only mophore framework as the parent compound. However, the other one hydroxyl-harmine was detected in the urine; this may be metabolites (M6–M10) had the same absorption wave- due to the instability of hydroxyl-harmine and immediately subse- lengths at 245, 320 and 360 nm which may be influenced by the quent to the reaction of phase II. glucuronate or sulphate. All metabolites were identified by com- D n paring UV spectra, the changes in observed mass ( m/z) and MS 3.3.4. Metabolite M6 spectra. The mass spectrum of M6, at a retention time of 4.30 min, gave an ion [M+H]+ at m/z 377 (Fig. 4A). The product ion, m/z 201 3.3.2. Metabolites M1 and M4 [M+H176]+ in the MS2 spectrum, also indicated the presence of MS and MS2 spectra of the protonated molecular ion at m/z 201 a glucuronic acid unit, and the MS3 product ion at m/z 160 was and 199 were the same as those of the standard harmalol and har- the same as one of the fragment ions of harmalol. Therefore, M6 mol. Therefore, metabolites M1 and M4 were identified as harma- was deduced as harmalol glucuronide (Fig. 2). lol and harmol on the basis of their mass spectra (ESI), and UV absorption spectra and comparing with the standards. 3.3.5. Metabolite M7 M7 was observed at retention time of 6.07 min with a proton- 3.3.3. Metabolite M5 ated molecular ion at m/z 375 (m/z 199 + 176) (Fig. 4A). The pro- M5 showed a protonated ion [M+H]+ at m/z 229 with retention tonated molecular ion of M7 was 176 Da (a glucuronic acid unit) time of 14.55 min (Fig. 4A). The MS2 spectrum of M5 displayed the more than that of harmol (M4). The MS2 spectra of M7 showed a characteristic product ion at m/z 213 formed by the loss of an major product ion at m/z 199, and the product ion showed MS3 T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105 1101

Fig. 3. Chromatograms of reaction products of alkaloids harmaline (M00), harmine (M0), harmalol (M1), harmol (M4) with HLMs in vitro. product ion at m/z 171 which was consistent with the ion observed for M8, corresponding to the same molecular formula 2 in the MS spectrum of harmol (M4) (Fig. 1D1). The HRMS gave a C19H20N2O8 (Fig. 2). protonated molecular ion at m/z 375.1186 (calculated 375.1192) The 1H and 13C NMR data of M8 are shown in Table 1. The pro- which corresponded with the molecular formula of M7 ton signals at d 5.09(1H, br. s) and the carbonyl signals at d 101.2,

C18H18N2O7 (Fig. 2). 77.1, 75.6, 73.7, 72.4, 172.8 suggested that this metabolite had a b- The molecular formula of the metabolite was further supported D-glucuronic acid chain. The presence of a proton signal at d by the 1H and 13C NMR spectral data (Table 1). The 1H-NMR spec- 4.11(3H, s) in the 1H NMR spectrum and a carbonyl signal at d 13 trum of M7 in DMSO-d6 revealed the presence of five aromatic pro- 56.2 in the C NMR spectrum of M8 indicated that the metabolic tons, three protons at d 8.06 (1H, br. d, J = 8.2 Hz, H-5), 7.19 (1H, s, reaction of 7-O-demethylation did not happen to M8. The 13C- H-8) and 6.92 (1H, br. d, J = 8.2 Hz, H-6) and revealed an ABX sys- NMR spectrum of M8 was similar to that of M7, except the shift tem for a 1,2,4-trisubstituted benzene ring. The proton signals at d to downfield of the carbonyl signal of C-5 at 20 ppm. This indicated 1 5.03 (1H, d, J = 6.0 Hz, C-10) in the H-NMR spectrum and the that the b-D-glucuronic acid chain was linked to C-5. So M8 was carbonyl signals at d 101.2, 76.7, 75.5, 73.6, 72.3, 172.6 in the identified as harmine-5-O-b-D-glucuronic acid. Based on the struc- 13 C-NMR spectrum confirmed that this metabolite had a b-D-glucu- ture of M8, it could be deduced that M5-a would be found in phase ronic acid. The 13C-NMR spectrum of M7 was similar to the spec- I metabolism (Fig. 2). trum of harmol (Coune, Angenot, & Denoel, 1980), except for the carbonyl signals of the glucuronic residue, indicating that the 3.3.7. Metabolite M9 parent nucleus of M7 was harmol (M4). The HMBC correlations M9 was observed at a retention time of 12.78 min with a pro- between the carbonyl signal at d 158.5(C-7) and proton signal at tonated molecular ion [M+H]+ at m/z 279 (Fig. 4A). The product 0 d 5.03(1H, d, J = 6.0 Hz, H-1 ) indicated that the glucuronic acid ion, m/z 199 [M+H80]+ in the MS2 spectrum, clearly showed the was linked to the C-7. So M7 was identified as harmol-7-O-b-D-glu- presence of a sulphide. The MS2 product ion at m/z 199 showed a curonic acid (Fig. 2). major MS3 product ion at m/z 171 which was also the MS2 product ion of harmol (M4). Based on the mass spectra data, M9 was de- duced to be harmol sulphate. The HRMS of the protonated molec- 3.3.6. Metabolite M8 ular ion was at m/z 279.0438 (calculated 279.0440) corresponded

The mass spectrum of M8 at a retention time of 8.00 min gave to the same molecular formula of M9 C12H10N2O4S(Fig. 2). an ion [M+H]+ at m/z 405 (Fig. 4A). The MS2 spectrum of M8 dis- The 1H- and 13C-NMR data are presented in Table 1. The 1H- + played a characteristic product ion at m/z 229 [M+H176] , formed NMR spectrum of M9 in DMSO-d6 revealed the presence of five by the loss of a glucuronic acid unit, suggesting that M8 was iden- aromatic protons, three protons at d 7.51 (1H, s, H-8), 6.98 (1H, tical to harmine glucuronide, and the MS3 product ion at m/z 214, dd, J = 8.6, 1.8 Hz, H-6) and 8.04 (1H, d, J = 8.4 Hz, H-5) and re- the same as the characteristic fragment ion of M5, suggesting that vealed an ABX system for a 1,2,4-trisubstituted benzene ring and harmine went through a hydroxylation reaction and subsequently the two constituted an AB system with a meta coupling constant. formed M8 through glucuronide conjugation. The HRMS showed a This was similar to the 1H-NMR spectrum of harmol (M4) (Coune protonated molecular ion at m/z 405.1290 (calculated 405.1298) et al., 1980). There were differences between the 13C-NMR spectra 1102 T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105

Fig. 4. Chromatograms of metabolites detected in rat urine after oral administration of alkaloids of (A) harmaline (M00) and (B) harmine (M0). of M9 and harmol; only one of the carbonyl signals at d 122.0 (C-5) substituent group of the M9 had been changed. Apart from the was similar to harmol, and the other carbonyl signal shifted to substitution differences, the structure of the rest of the compound downfield with a difference of 3–4 ppm. This indicated that the was confirmed by the analysis of HMBC and HMQC spectra. Using T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105 1103

Table 1 1H- and 13C-NMR signals of metabolites M7, M8, M9 and M12 (ppm).

No. Carbon signals Proton signals M7 M8 M9 M12 M7 M8 M9 M12 1 141.9 141.6 141.9 140.5 3 138.0 137.1 138.0 163.6 8.16 (1H,br. s) 8.10 (1H, br.s) 8.16 (1H, d, J = 5.2 Hz) 4 112.7 112.7 112.7 113.5 7.84 (1H,br. s) 7.73 (1H, br. s) 7.84 (1H, d, J = 5.2 Hz) 8.11 (1H, d, J = 5.4 Hz) 4a 127.6 127.9 127.4 129.3 4b 116.3 117.1 117.1 116.6 5 122.0 142.0 122.0 122.9 8.06 (1H, br. d, J = 8.2 Hz) 8.04 (1H, d, J = 8.4 Hz) 8.16 (1H, d, J = 8.6 Hz) 6 110.8 108.3 113.9 114.7 6.92 (1H, br. d, J = 8.2 Hz) 7.83 (1H, s) 6.98 (1H, dd, J = 8.6, 1.8 Hz) 7.06 (1H, dd, J = 8.6, 1.8 Hz) 7 158.5 151.9 154.3 155.4 8 98.6 95.7 103.4 103.1 7.19 (1H, s) 7.06 (1H, s) 7.51 (1H, s) 7.59 (1H, d, J = 1.8 Hz) 8a 142.1 134.8 141.6 142.8 8b 135.2 129.7 135.3 134.7 9 11.60 (1H, s, N–H) 11.60 (1H, s, N–H) 11.48 (1H, s, N–H) 11.95 (1H, s, N–H) Me 19.1 19.1 19.1 19.1 2.72 (3H, s) 2.72 (3H ,s) 2.73 (3H, s) 2.83 (3H, s) O–Me 56.2 4.11 (3H, s) 10 101.2 101.2 5.03 (1H, d, J = 6.0 Hz) 5.09 (1H, br. s) 20–50 72.3, 73.6, 75.5, 76.7 72.4, 73.7, 75.6, 77.1 60 172.6 172.8

the molecular formula obtained from HRMS data, the substituent H-8), 7.06 (1H, dd, J = 8.6, 1.8 Hz, H-6) and an ABX system for a group at C-7 of M9 could be a sulphonic acid. Hence, M9 was iden- 1,2,4-trisubstituted benzene ring, and a single proton signal d tified as harmol-7-sulphonic acid (Fig. 2). 8.11(1H, d, J = 5.4 Hz) indicated that there must be a substitution at the 3 or 4-position of M12. Compared with the 1H-NMR spec- 3.3.8. Metabolite M10 trum of harmine (Carmona-Guzmanm, Balon-Ameida, Idalgo-Tele- The metabolite M10 was detected in rat urine samples at a do, & Munoz-Perez, 1989), M12 did not show the methoxyl proton retention time of 13.38 min (Fig. 4B). A full product ion scan of and carbonyl signals. Thus, the substitution in the 7-position of 13 M10, with the precursor ion m/z 309 corresponding to [M+H]+, re- M12 was a hydroxyl group. Comparing the C-NMR spectrum of sulted in the main fragment at m/z 229. The ion at m/z 229 (80 Da M12 with harmine, an upfield shift of C-3 by 25.6 ppm indicated from 309) could be formed by the loss of a sulphonic acid, suggest- that there was an oxygen substituent linked to C-3. This was con- ing that M10 was hydroxylated harmine-sulphate (Fig. 2). firmed by the HMBC correlations of the proton at d 8.11 (1H, d, J = 5.4 Hz, H-4) with the carbonyl signal at d 116.6 (C-4b) and 134.7 (C-8b). The absorbances at 1240 and 1041 cm1 indicated 3.3.9. Metabolite M11 that M12 had a sulphonic group. Hence, its structure was sug- With protonated ion [M+H]+ at m/z 281, the metabolite M11 of gested to be 3-hydroxyl-7-sulphonic harmol through IR, mass, harmaline (M00) was only found in rat bile sample at a retention 1H- and 13C-NMR spectra. Based on the structure of M12, it can time of 10.74 min. The product ion m/z 201 [M+H80]+ in the be deduced that M5-b may be found in phase I metabolites of MS2 spectrum clearly showed the presence of a sulphide. Based harmine (Fig. 2). on the mass spectra, M11 was therefore deduced to be harmalol sulphate (Fig. 2). 3.4. A summary of metabolic pathways for harmaline and harmine 3.3.10. Metabolite M12 The urinary metabolite, M12, of harmine was isolated and puri- The biotransformations of harmaline and harmine were investi- fied from urine samples, and we found a protonated ion [M+Na]+ at gated using a combination of HPLC-UV/MSn and NMR techniques. m/z 317 by coupling a direct insertion probe with the mass spec- From the above results, the metabolic pathways of harmaline trometer. The product ion m/z 237 [M+Na80]+ in the MS2 spec- and harmine in rat were proposed, as in Fig. 2. The observation trum clearly showed the presence of a sulphide, suggesting that of these metabolites can be explained by five proposed pathways; harmine was metabolised by a hydroxylation reaction to form 7-O-demethylation, hydroxylation, oxidative dehydrogenation, O- M7 and subsequently M12 through sulphate conjugation. How- glucuronide conjugation and O-sulphate conjugation. It was note- ever, the protonated ion [M+Na]+ at m/z 317 could not be detected worthy that harmaline (M00) could be transformed to harmine in the urine or bile samples by chromatographic and spectroscopic (M0) by oxidative dehydrogenation. It was found that the purity methods, which was explained in Section 2.7. The protonated ion of harmaline was about 99.32% and the impurity in it was harmine, [M+H]+,atm/z 295, of M12 was obtained using a maintained MGII with a content of 1.58% detected by HPLC-MS. However, the con-

C18 column with a mixture of acetonitrile and water as mobile tent of harmine and its metabolites in the urine sample collected phase in a gradient mode. The appearance of the product ion, m/z after oral administration of harmaline was 23.4%, using the peak 279 [M+H16]+, formed by the loss of a hydroxyl group, indicated area normalisation method. This suggested that oxidative dehydro- that the change of the mass spectrometric fragmentations and genation was one of the major metabolic pathways for harmaline characteristics of M12 were influenced by the properties of the in rat. samples and mobile phase. According to the in vivo metabolic data in rat and in vitro met- The structure of M12 was also established by IR, 1H- and 13C- abolic data in HLMs, it can be deduced that the biotransformation NMR spectra (Table 1). The structure of M12 was established by of harmaline and harmine in the human body was similar to that of a combination of one-dimensional and two-dimensional NMR the phase I metabolic pathways in rat, but different in phase II met- techniques, operating at 400 MHz. The 1H-NMR spectrum of M12 abolic pathways. 7-O-Demethylated metabolites and hydroxylated in DMSO-d6 revealed the presence of four aromatic protons, three b-carboline (hydroxylation sites may be different from the hydrox- protons at d 8.16 (1H, d, J = 8.6 Hz, H-5), 7.59 (1H, d, J = 1.8 Hz, ylation action in rat) were the major metabolites, subsequently 1104 T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105 excreted as the glucuronide conjugates and sulphate conjugates. U1130303), the National Nature Science Foundation of China (No. However, some hydroxylation sites were still uncertain. The differ- 81173119), the Key Project of Ministry of Science and Technology ence in phase II metabolism of harmaline and harmine between rat of China (2012ZX09103201-051), the Shanghai Science & Technol- and human mainly depended on the metabolic activity dif- ogy Development Foundation (08JC1418600 and 08DZ1972101), ferences of glucuronosyltransferase and sulfatase in rat and hu- the Program for Shanghai Innovative Research Team in University man. The glucuronide conjugation pathway is particularly well (2009) and the Program for Changjiang Scholars and Innovative developed in man and the sulphate conjugation pathway predom- Research Team in University (IRT1071) to Prof. Chang-Hong Wang inates in rat (Slotkin et al., 1970). Possibly the O-glucuronide for the financial support of this study. conjugate was the major phase II metabolite in human, but the O-sulphate conjugate mainly in rat. The b-carboline alkaloids (harmaline, harmine) and almost all of their metabolites were detected in urinary excretion; only the References polar metabolites such as M1, M5, M6, M7, M8, M9 and M11, could Agüía, L., Peña-Farfala, C., Yáñez-Sedeño, P., & Pingarróna, J. M. (2007). be excreted in bile, and these data support to the idea that Determination of b-carboline alkaloids in foods and beverages by high- high-molecular-weight and polar metabolites tend mainly to be performance liquid chromatography with electrochemical detection at a excreted in bile. Interestingly, the metabolites of hydroxylated glassy carbon electrode modified with carbon nanotubes electrode modified with carbon nanotubes. Analytica Chimica Acta, 585, 323–330. b-carboline-sulphates (M10, M12) and hydroxylated b-carboline- Alves, R. C., Mendes, E., Oliveira, B. P. P., & Casal, S. (2010). Norharman and harman glucuronic acid (M8) are herein reported for the first time. In addi- in instant coffee and coffee substitutes. Food Chemistry, 120, 1238–1241. tion, the oxidative dehydrogenation of harmaline (transforming to Arshad, N., Neubauer, C., Hasnain, S., & Hess, M. (2008). Peganum harmala can minimize Escherichia coli infection in poultry, but long-term feeding may induce harmine as one of the major metabolic pathways for harmaline in side effects. Poultry Science, 87, 240–249. rat) is observed for the first time. The contents of the metabolites Astulla, A., Zaima, K., Matsuno, Y., Hirasawa, Y., Ekasari, W., Widyawaruyanti, A., were also calculated, roughly, by using the peak area normalisation et al. (2008). Alkaloids from the seeds of Peganum harmala showing antiplasmodial and vasorelaxant activities. Nature Medicine, 62, 470–472. method. It was found that phase I metabolites of M1 (harmalol) Carmona-Guzmanm, M. C., Balon-Ameida, M., Idalgo-Teledo, J. H., & Munoz-Perez, and M4 (harmol) were the main metabolite in rat urine after oral M. A. (1989). Sulfonation reactions of b-carbolines. Canadian Journal of administration of harmaline, and the M4 (harmol) was the main Chemistry, 67, 720–726. metabolites in rat urine after oral administration of harmine. The Chen, J. Z., Chou, G. X., Wang, C. H., Yang, L., Bligh, S. W. A., & Wang, Z. T. (2010). Characterization of new metabolites from in vivo biotransformation of content of M5 (hydroxyl-harmine) was about one eighth that of norisoboldine by liquid chromatography/mass spectrometry and NMR M4 in the hamine-treated groups, but not in the harmaline-treated spectroscopy. Journal of Pharmaceutical and Biomedical Analysis, 52, 687–693. groups. In addition, the in vivo data revealed that the abundance of Cheng, X. M., Zhao, T., Yang, T., Wang, C. H., Bligh, S. W. A., & Wang, Z. T. (2010). HPLC fingerprints combined with principal component analysis, hierarchical M9 (harmol-7-sulphonic acid) was greatest in phase II metabolites cluster analysis and linear discriminant analysis for the classification and in harmine-treated groups. Meanwhile, the two O-sulphate conju- differentiation of Peganum sp. indigenous to China. Phytochemical Analysis, 21, gation of M9 and M11 (harmalol-7-sulphonic acid) were plentifully 279–289. Cobuzzi, R. J., Neafsey, E. J., & Collins, M. A. (1994). Differential cytotoxicities of N- found in harmaline treated groups. This indicated that the O-sul- methylated beta-carbolinium analogues of MPP+ in PC12 cells: insight into phate conjugation was a major phase II characteristic of harmaline potential neurotoxicants in Parkinson’s disease. Journal of Neurochemistry, 62, and harmine in rat. It was also found that M8 (harmine-5-O-b-D- 1503–1510. Coune, C. A., Angenot, L. J. G., & Denoel, J. (1980). 13C NMR spectra of Strychnos glucuronic acid) was one of major phase II metabolites in harma- alkaloids: harman and usambarensine derivatives. Phytochemistry, 19, line- and harmine-treated groups. Harmine (M0) and a series of 2009–2011. characteristic metabolites, such as M4, M8, and M9, related to Conning, D. M. (1989). Food composition and human behaviour. Food Chemistry, 33, 33–41. harmine, could be detected in the urine of harmaline-treated Crotti, A. E., Gates, P. J., Lopes, J. L., & Lopes, N. P. (2010). Electrospray MS-based groups, which indicated that harmine might be formed by characterization of beta-carbolines–mutagenic constituents of thermally oxidative dehydrogenation from harmaline and then excreted as processed meat. Molecular Nutrition & Food Research, 54, 433–439. glucuronide conjugates and sulphate conjugates. Derakhshanfar, A., Oloumi, M. M., & Mirzaie, M. (2010). Study on the effect of Peganum harmala extract on experimental skin wound healing in rat: Due to their lower abundances, M2, M3, M5, and M10 could not pathological and biomechanical findings. Comparative Clinical Pathology, 19, be collected in sufficient amounts to be separated and purified, 169–172. which resulted in failure to elucidate metabolites structure and Farouk, L., Laroubi, A., Aboufatima, R., Benharref, A., & Chait, A. (2008). Evaluation of the analgesic effect of alkaloid extract of Peganum harmala L. Possible metabolic pathways. In order to illustrate the dynamic metabolism mechanisms involved. Journal of Ethnopharmacology, 115, 449–454. profiles of harmaline and harmine in vivo, further research would Guan, Y., Louis, E. D., & Zheng, W. (2001). Toxicokinetics of tremorogenic natural be required on pharmacokinetics of harmaline and harmine to products, harmane and harmine, in male Sprague–Dawley rats. Journal of Toxicology and Environmental Health Part A, 64, 645–660. understand the absorption, distribution and excretion of these Hemmateenejad, B., Abbaspour, A., Maghamia, H., Miri, R., & Panjehshahin, M. R. two alkaloids and their metabolites in organisms. (2006). Partial least squares-based multivariate spectral calibration method for simultaneous determination of beta-carboline derivatives in Peganum harmala seed extracts. Analytica Chimica Acta, 575, 290–299. 4. Conclusion Herath, W., Mikell, J. R., Ferreira, D., & Khan, I. A. (2003). Microbial metabolites of harman alkaloids. Chemical & Pharmaceutical Bulletin (Tokyo), 51, 646–648. Oxidative dehydrogenation, 7-O-demethylation, hydroxylation, Herraiz, T., Guillén, H., & Arán, V. J. (2008). Oxidative metabolism of the bioactive O-glucuronide conjugation, and O-sulphate conjugation were and naturally occurring b-carboline alkaloids, norharman and harman, by human cytochrome P450 . Chemical Research in Toxicology, 21, established as the main metabolic pathways for the b-carboline 2172–2180. alkaloids, harmaline and harmine, by HPLC-UV/MSn and NMR Ho, B. T., Estevez, V., Fritchie, G. E., & Tansey, L. W. (1971). Metabolism of harmaline analysis. The results indicated that harmaline and harmine were in rats. Biochemical Pharmacology, 20, 1313–1319. Jahaniani, F., Ebrahimi, S. A., Rahbar-Roshandel, N., & Mahmoudian, M. (2005). mainly excreted in urine and bile as glucuronide and sulphate Xanthomicrol is the main cytotoxic component of Dracocephalum kotschyii and conjugates from the body of rat. a potential anti-cancer agent. Phytochemistry, 66, 1581–1592. Kartal, M., Altun, M. L., & Kurucu, S. (2003). HPLC method for the analysis of harmol, harmalol, harmine and harmaline in the seeds of Peganum harmala L.. Journal of Acknowledgements Pharmaceutical and Biomedical Analysis, 31, 263–269. Kim, H., Sablin, S. O., & Ramsay, R. R. (1997). Inhibition of A by The authors gratefully acknowledge the award from the Key beta-carboline derivatives. Archives of Biochemistry and Biophysics, 337, 137–142. Program of Joint Funds of the National Natural Science Foundation Massoud, M., Hossein, J., & Pirooz, S. (2002). Toxicity of Peganum harmala: review of China and Xinjiang Uygur Autonomous Region of China (No. and a case report. Iran Journal of Pharmacology and Therapeutics, 1, 1–4. T. Zhao et al. / Food Chemistry 134 (2012) 1096–1105 1105

Mulder, G. J., & Hagedoorn, A. H. (1974). UDP glucuronyltransferase and Tweedie, D. J., & Burke, M. D. (1987). Metabolism of the beta-carbolines, harmine phenolsulfotransferase in vivo and in vitro. Conjugation of harmol and and harmol, by liver microsomes from phenobarbitone- or 3- harmalol. Biochemical Pharmacology, 23, 2101–2109. methylcholanthrene-treated mice. Identification and quantitation of two Riba, J., Valle, M., Urbano, G., Yritia, M., Morte, A., & Barbanoj, M. J. (2003). Human novel harmine metabolites. Drug Metabolism and Disposition, 15, 74–81. pharmacology of ayahuasca: subjective and cardiovascular effects, monoamine Wang, C. H., Liu, J., & Zheng, L. M. (2002). Analysis of harmine and harmaline of metabolite excretion, and pharmacokinetics. Journal of Pharmacology and Peganum harmala in different parts of different localities. Chinese Pharmaceutical Experimental Therapeutics, 306, 73–83. Journal, 37, 212–215. Samoylenko, V., Rahman, M. M., Tekwani, B. L., Tripathi, L. M., Wang, Y. H., Khan, Yu, A. M., Idle, J. R., Krausz, K. W., Küpfer, A., & Gonzalez, F. J. (2003). Contribution of S. I., et al. (2010). , a unique combination of MAO individual cytochrome P450 isozymes to the O-demethylation of the inhibitory and antioxidative constituents for the activities relevant to psychotropic b-carboline alkaloids harmaline and harmine. Journal of neurodegenerative disorders and Parkinson’s disease. Journal of Pharmacology and Experimental Therapeutics, 305, 315–322. Ethnopharmacology, 127, 357–367. Zheng, X. Y., Zhang, Z. J., Chou, G. X., Wu, T., Cheng, X. M., Wang, C. H., et al. (2009). Schwarz, M. J., Houghton, P. J., Rose, S., Jenner, P., & Lees, A. D. (2003). Activities of Acetylcholinesterase inhibitive activity-guided isolation of two new alkaloids extract and constituents of banisteriopsis caapi relevant to parkinsonism. from seeds of Peganum nigelastrum bunge by an in vitro TLC-bioautographic Pharmacology Biochemistry and Behavior, 75, 627–633. assay. Archives of Pharmacal Research, 32, 1245–1251. Slotkin, T. A., & DiStefano, V. (1970a). Urinary metabolites of harmine in the rat Zheng, X. Y., Zhang, L., Cheng, X. M., Zhang, Z. J., Wang, C. H., & Wang, Z. T. (2011). and their inhibition of monoamine oxidase. Biochemical Pharmacology, 19, Identification of acetylcholinesterase inhibitors from seeds of plants of genus 125–131. Peganum by thin-layer chromatography–bioautography. Journal of Planar Slotkin, T. A., & DiStefano, V. (1970b). A model of harmine metabolism in the rat. Chromatography – Modern TLC, 24, 470–474. Journal of Pharmacology and Experimental Therapeutics, 174, 456–462. Zhou, G., Shi, S., Zhang, W., Tan, Z., Chen, Y., Guo, D., et al. (2010). Identification of Slotkin, T. A., DiStefano, V., & Au, W. Y. (1970). Blood levels and urinary excretion of ilaprazole metabolites in human urine by HPLC-ESI-MS MS and HPLC-NMR harmine and its metabolites in man and rats. Journal of Pharmacology and experiments. Biomedical Chromatography, 24, 1130–1135. Experimental Therapeutics, 173, 26–30. Sourkes, T. L. (1999). ‘‘Rational hope’’ in the early treatment of Parkinson’s disease. Canadian Journal of Physiology and Pharmacology, 77, 375–382.