Identification of a New Reactive Metabolite of Pyrrolizidine Alkaloid

Oct 8, 2014 - Identification of a New Reactive Metabolite of Pyrrolizidine Alkaloid Retrorsine: (3H-Pyrrolizin-7-yl)methanol ... The EC cell was coupl...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/crt

Identification of a New Reactive Metabolite of Pyrrolizidine Alkaloid Retrorsine: (3H‑Pyrrolizin-7-yl)methanol Muluneh M. Fashe, Risto O. Juvonen, Aleksanteri Petsalo, Minna Rahnasto-Rilla, Seppo Auriola, Pasi Soininen, Jouko Vepsal̈ aï nen, and Markku Pasanen* School of Pharmacy, Faculty of Health Sciences, University of Eastern Finland, P.O. Box 1627, FI-70211 Kuopio, Finland S Supporting Information *

ABSTRACT: Pyrrolizidine alkaloids (PAs) such as retrorsine are common food contaminants that are known to be bioactivated by cytochrome P450 enzymes to putative hepatotoxic, genotoxic, and carcinogenic metabolites known as dehydropyrrolizidine alkaloids (DHPs). We compared how both electrochemical (EC) and human liver microsomal (HLM) oxidation of retrorsine could produce short-lived intermediate metabolites; we also characterized a toxicologically important metabolite, (3H-pyrrolizin7-yl)methanol. The EC cell was coupled online or offline to a liquid chromatograph/mass spectrometer (LC/MS), whereas the HLM oxidation was performed in 100 mM potassium phosphate (pH 7.4) in the presence of NADPH at 37 °C. The EC cell oxidation of retrorsine produced 12 metabolites, including dehydroretrorsine (m/z 350, [M + H+]), which was degraded to a new reactive metabolite at m/z 136 ([M + H+]). The molecular structure of this small metabolite was determined using highresolution mass spectrometry and NMR spectroscopy followed by chemical synthesis. In addition, we also identified another minor but reactive metabolite at m/z 136, an isomer of (3H-pyrrolizin-7-yl)methanol. Both (3H-pyrrolizin-7-yl)methanol and its minor isomer were also observed after HLM oxidation of retrorsine and other hepatotoxic PAs such as lasiocarpine and senkirkin. In the presence of reduced glutathione (GSH), each isomer formed identical GSH conjugates at m/z 441 and m/z 730 in the negative ESI-MS. Because (3H-pyrrolizine-7-yl)methanol) and its minor isomer subsequently reacted with GSH, it is concluded that (3H-pyrrolizin-7-yl)methanol may be a common toxic metabolite arising from PAs.

■

INTRODUCTION

metabolism of PAs involves three main metabolic reactions: hydrolysis, N-oxidation, and dehydrogenation.17 Cytosolic and microsomal carboxylesterases in the liver catalyze the hydrolysis of PAs to the necine base and necic acid moieties,19,20 whereas flavin-containing monooxygenases and CYP3A4 enzymes catalyze oxidation of PAs to their N-oxides.21,22 The hydrolysis and N-oxidation pathways are generally considered to be detoxification mechanisms.20 On the other hand, in humans, CYP3A4 can catalyze the bioactivation of PAs to reactive pyrrolic esters called dehydropyrrolizidine alkaloids (DHP).20,23,24 DHPs are unstable intermediates and are believed to have two fates in biological matrices in vivo or in vitro: to covalently bind endogenous nucleophiles such as glutathione (GSH), DNA, and proteins or to be hydrolyzed to less reactive but more stable pyrrolic metabolites and the corresponding

Pyrrolizidine alkaloids (PAs) such as retrorsine are naturally occurring common hepatotoxins. Chemically, PAs are bis-esters of amino diols (i.e., amino alcohols) and complex carboxylic acids. 1 Over 6000 plant species mainly belonging to Boraginaceae, Compositae (Asteraceae), and Legumionsae (Fabaceae) are believed to produce PAs or N-oxides of PAs.2 Humans can be exposed to PAs through contaminated foods such as wheat grain, herbal medicines, and dietary supplements,3−5 and this may cause severe health problems such liver veno-occlusive disease (VOD), cirrhosis, and cancer in humans and animals.6−9 Human and livestock poisoning caused by PAs has been recorded in different parts of the world, with liver damage being the main cause of death.10−13 Moreover, recent exposure assessments conducted by several regulatory organizations have stated that PAs remain potential sources of public health risk.4,14 Metabolic activation plays a central role in the toxicity of retrorsine and other PAs.15−18 In general, the primary © 2014 American Chemical Society

Received: July 17, 2014 Published: October 8, 2014 1950

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

Figure 1. General EC/(LC)/MS setup: online EC/ESI-MS (A), online EC/LC/ESI-MS (B), and offline EC/LC/ESI-MS (C). A 100 μM retrorsine sample was prepared in 20 mM ammonium acetate and acetonitrile (3:1), and EC cell oxidation was performed at different applied voltages as per the specific experiments. Abbreviations: EC, electrochemical; LC, liquid chromatography; RP, reverse phase; MS, mass spectrometry; and ESI, electrospray ionization. were prepared by differential ultracentrifugation.32 The microsomal protein concentration was determined by the Bradford method.33 Online EC/(LC)/MS Oxidation of Retrorsine. For the online EC/(LC)/MS oxidation of retrorsine, a Coulochem EC analytical cell (model 5010, ESA Bioscience, Inc., Chelmsford, MA, USA) was coupled online to MS (Figure 1A) or LC/MS (Figure 1B). The EC cell was equipped with porous graphite working electrodes, a Pd counter electrode, and a Pd/H2 reference electrode, and it was controlled by an ESA Coulochem potentiostat (ESA Bioscience, Inc., Chelmsford, MA, USA). In the EC/MS setup, 1 μL of 100 μM retrorsine was directly injected into the EC cell via HPLC solvents (A, ammonium acetate; B, acetonitrile) at a flow rate of 0.2 mL/min, and the data were acquired for 2 min. In the EC/LC/MS setup, 1 μL of 100 μM retrorsine was injected into the EC cell via a stream of HPLC mobile phase solvents, and chromatographic separation was achieved according to HPLC method I described below. Offline EC/LC/MS Oxidation of Retrorsine. The offline EC/ LC/MS (Figure 1C) oxidation of retrorsine was performed by infusing 100 μM retrorsine into the EC cell (at +800 mV vs the Pd/H2 reference electrode) at a 20 μL/min flow rate from an external syringe (Hamilton Company, Reno, NV, USA). The oxidation solution was collected from the outlet tip of the EC cell. The stability of the oxidation products was evaluated by analyzing the oxidized solution by LC/MS using HPLC method I described below. The reactivity of the oxidation products was investigated by incubating the oxidized solution with 4 mM GSH. Furthermore, the effects of GSH concentration and incubation time on the rate of the formation of GSH conjugates were also evaluated. The effect of the concentration of GSH on the rate of the conjugation reaction was studied by incubating the EC cell eluent with different concentrations of GSH (0−33 mM) for 10 min. The effect of incubation time on the conjugation reaction was studied by incubating the oxidized EC cell eluent with 4 mM GSH, and the incubation samples were analyzed by LC/MS in the negative mode at selected time points (10 min to 10 h). The sample injection volume was 2 μL, and the LC conditions were as described in HPLC method II (given below). Incubation of Retrorsine with Human Liver Microsomes (HLM). A 100 μM retrorsine solution was preincubated (Heildoph Incubator 1000, Germany) in a 500 μL reaction mixture containing HLM (0.036 mg of protein) and 5 mM MgCl2 in 100 mM potassium phosphate buffer, pH 7.4, in the presence or absence of 4 mM GSH at 37 °C for 10 min. The reaction was started by the addition of 1 mM NADPH and stopped by addition of 1.5 mL of acetonitrile after 60 min of incubation. The samples were centrifuged (Mini Spin, Eppendorf, Hamburg, Germany) at 10 000g for 10 min, and the resulting supernatant was collected and analyzed by LC/MS in both positive and negative modes. The sample injection volume was 2 μL, and the LC conditions were as described in HPLC method II. For the preparation of a large quantity of [8] for NMR analysis, the incubation system and retrorsine concentration were scaled up to 50 mL and 250 μM, respectively. The incubation was conducted under the same conditions as those described above (without GSH) for 60

carboxylic acids. So far, DHPs have not been detected either in vivo or in vitro; i.e., the formation of these reactive intermediates has been revealed only indirectly via their stable pyrrolic metabolites such as the GSH conjugates.15,25−27 It is conceivable that the presence of endogenous nucleophiles in the biological matrices may be responsible for the difficulties in detecting these reactive electrophilic metabolites such as dehydroretrorsine. Therefore, this study aimed to investigate the metabolism of retrorsine in two different oxidative catalyst environments: an electrochemical (EC) cell and human liver microsomes (HLM) in vitro. Because the EC cell is devoid of any nucleophiles such as GSH and it can be coupled online to a liquid chromatograph/mass spectrometer (LC/MS), it may provide an excellent platform to produce and detect short-lived electrophilic metabolites. Furthermore, its application in the mimicry of cytochrome P450-catalyzed biotransformation reactions has been successfully demonstrated for several organic chemicals including environmental contaminants and therapeutic drugs.28−31 Thus, by applying this method, we demonstrated the formation and detection of reactive and short-lived metabolites such dehydroretrorsine [1] and a new metabolite, (3H-pyrrolizin7-yl)methanol [8]. The latter compound was also detected after metabolism of retrorsine with HLM, and its molecular structure was confirmed by chemical synthesis.

■

MATERIALS AND METHODS

Caution: This chemical is dangerous: retrorsine (CASRN: 480-54-6) is a hepatotoxin and should be handled caref ully. Chemicals and Reagents. All chemicals, reagents, and solvents were of the highest purity available from their commercial suppliers. Retrorsine, ammonium acetate, and GSH were purchased from SigmaAldrich (Steinheim, Germany); acetonitrile and methanol were purchased from VWR International (Leuven, Belgium); and nicotinamide adenine dinucleotide phosphate (NADPH) was purchased from Roche Diagnostics (Mannheim, Germany). Water was purified using a Milli-Q Gradient/Quantum EX system (Millipore, Milford, MA). Ten millimolar retrorsine stock solution was prepared in methanol and stored at −20 °C; 100 μM retrorsine in 20 mM ammonium acetate and acetonitrile (3:1) was prepared on working dates from the stock solution. Human Liver Microsomes. The human liver sample used in this study was obtained from the University Hospital of Oulu as surplus from a cadaver (kidney transplantation donor). The collection of the specimens was approved by the Ethics Committee of the Medical Faculty of the University of Oulu. After surgical excision, the liver samples were immediately transferred to ice, cut into pieces, snapfrozen in liquid nitrogen, and stored at −80 °C until the microsomes 1951

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

Scheme 1. Preparation of the Reactive Isomer (3H-Pyrrolizin-7-yl)methanol [8]a

a

(A) Synthetic preparation of [8] and its rearrangement to isomer [20] during storage, (B) possible isomers of [8], and (C) the proposed reaction mechanism between [8] and reduced GSH. min. The reaction was terminated by acetonitrile addition, and the samples were centrifuged at 10 000g for 10 min. The supernatant was collected and evaporated to dryness under nitrogen gas and recovered in NMR solvent (CDCl3). HPLC Conditions. An Agilent 1200 HPLC system (Agilent Technologies, Waldbronn, Germany) was used, and the solvents used in this study were 5 mM ammonium acetate (A) and acetonitrile (B). In order to achieve optimum chromatographic separation, two different HPLC methods were used. In method I, chromatographic separation was achieved with an Xterra MS C8, 3.5 μm 2.1 × 20 mm IS column (Waters, Milford, MA, USA). The flow rate was 0.2 mL/ min, and a linear gradient from 10% B to 100% B in 10 min was used. Then, the column was allowed to re-equilibrate at 10% B for 4 min before the next injection. In method II, chromatographic separation was achieved with a Hypersil graphite column (100 × 4.6 mm, Cheshire, England). The flow rate was 0.4 mL/min, and a linear gradient from 10% B to 100% B in 10 min was used. Then, the column was allowed to re-equilibrate at 10% B for 4 min prior to the next injection. ESI-MS and ESI-MS/MS Conditions. A Finnigan LTQ ion-trap mass spectrometer (Thermo Electron Corp., Waltham, MA, USA) in both positive and negative modes was used for the MS and MS/MS analyses with a mass scan range of m/z 130−1000. The MS conditions in the positive mode were as follows: sheath gas flow rate, 30 AU; auxiliary gas flow rate, 2 AU; spray voltage, 4.00 kV; capillary temperature, 250 °C; and capillary voltage, 20 V. The MS conditions in negative mode were as follows: sheath gas flow rate, 30 AU; spray voltage, 3.80 kV; capillary temperature, 250 °C; and capillary voltage, −20 V. MS and MS/MS data were acquired and analyzed by Xcalibur 2.0 SUR1. Accurate Mass Acquisition. For accurate mass measurements, an Agilent 6540 time-of-flight (Q-ToF) (Agilent Technologies, Palo Alto, CA, USA) equipped with and AJS ESI ion source was used. The QToF MS conditions were as follows: capillary voltage, 3.5 kV; fragmentor voltage, 100 V; sheath gas at flow rate, 11 L/min at 350 °C (nitrogen); drying gas flow rate, 10 L/min at 325 °C (nitrogen); and nebulizer (adjusted to a 45 psi). As a lock mass, Agilent reference mass solution (m/z 121.05087 and m/z 922.00979) was used. The LC/MS data for this specific experiment were acquired and analyzed by Agilent MassHunter workstation software (B.04.00). The Q-ToF MS was coupled with LC (Agilent 1290 Series Rapid Resolution, Waldbronn, Germany). The LC/MS analysis was achieved on an Agilent Zorbax SB-C18 column (2.1 × 50 mm, 1.8 μm particle size) together with an Agilent Infinity in-line filter (0.2 μm). HPLC mobile phase solvents were 0.1% formic acid (A) and acetonitrile (B) at a flow rate of 0.3 mL/min and a column oven temperature of 50 °C. A linear gradient elution from 5% B to 20% in 5 min was used followed by a rapid

increase to 90% B in 2 min and a 3 min re-equilibration phase between each run. Identification of the Oxidation Products. The EC cell oxidation products and the supernatants of the incubation samples were directly analyzed by LC/MS in the negative and positive modes; MS, MS/MS, and accurate masses were acquired for each metabolite. For the previously known metabolites of retrorsine, such as retrorsine N-oxide [5], retrorsic acid [12], 7-GSH-DHP [13], and 7,9-diGSHDHP [14], the MS/MS data from the literature were obtained and used during the identification, whereas the new metabolites were identified mainly by the MS, MS/MS, and accurate mass data. Where possible, the accurate mass measurement errors were calculated and used to assist in the identification process, particularly in the elemental composition analysis. Because of its toxicological importance, [8] was chemically synthesized, and its identification and structural characterization were achieved by extensive NMR and HRMS analysis of the biological and synthetic samples. NMR Experiments. NMR spectra from the synthetic compounds were recorded on a Bruker AVANCE 500 NMR spectrometer (Bruker-Biospin GmbH, Karlsruhe, Germany) equipped with a 5 mm quadronuclear probe and Bruker TopSpin software version 1.3 running on a standard PC, and NMR spectra from the biological samples were recorded on a Bruker AVANCE IIIHD 600 NMR spectrometer equipped with a 5 mm cryogenically cooled triple resonance probe head (CryoProbeTM Prodigy TCI) and Bruker TopSpin software version 3.2. The 1H NMR spectra were collected with a 90° pulse, relaxation delay of 10 s, and 16 scans at 300 K. The proton decoupled 13C NMR spectra were measured by using the zgdc pulse program with standard parameters. Assignments of the signals were verified with 2D 1H−1H COSY (correlation spectroscopy) optimized for long-range couplings, 2D 1H−1H TOCSY (total correlation spectroscopy), 2D 13C−1H HSQC (heteronuclear single quantum correlation spectroscopy), and 2D 13C−1H HMBC (heteronuclear multiple bond correlation spectroscopy) spectra, which were acquired for representative samples using routine settings. Preparation of the Reactive Metabolite [8] (m/z 136). The target metabolite, [8], was obtained after lithium aluminum hydride reduction from the previously known ethyl 3H-pyrrolizine-7carboxylate [19].34 The starting compound [19] was prepared from ethyl 2-pyrrolylglyoxylate [17], obtained from freshly distilled pyrrole [15] and ethoxalyl chloride [16], and vinyltriphenylphosphonium bromide [18], leading to the formation of a mixture of two isomers, i.e., [19] and its isomer, which were separated on neutral alumina (Scheme 1A). Preparation of [17]. Compound [17] was prepared using a known method34 in 57% yield. 1H NMR (500.1 MHz, CDCl3) δ 10.1 (1H, bs), 7.40 (1H, bs), 7.20 (1H, bs), 6.37 (1H, m), 4.42 (2H, t, J = 1952

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

Table 1. LC/MS and MS/MS Data of the Oxidation Products and the Glutathione Conjugatesa PDs

RT (min)

MS/MS fragments

RET A [1] [2] [3] [4] [5]d [6] [7] B [8]d [9]d [10] [11] [12]d C [13]d [14]d

4.51

334, 324, 138, 120

6.01 3.45 4.37 4.54 3.68 4.35 4.52

350, 348, 348, 348, 350, 350, 366,

332, 152, 338, 152, 138, 324, 152,

136, 134 320, 134 136, 136, 134

7.31 8.48 6.64 5.65 4.01

118, 118, 136, 152, 213,

108, 108, 126, 122, 187,

94, 80 94, 80 110, 94, 82 108 169

5.60 5.17

423, 272, 254, 143, 128 712, 601, 457, 254, 272

118 152, 134 120, 118 118

observedb mass (m/z)

predicted mass (m/z)

errorc (ppm)

C18H25NO6

352.1754

352.1755

−0.2

C18H23NO6 C18H23NO7 C18H23NO7 C18H23NO7 C18H25NO7 C18H23NO7 C18H23NO8

ND 366.1556 ND 366.1551 368.1711 368.1694 384.1657

350.1598 366.1547 366.1547 366.1547 368.1704 368.1704 384.1653

ND 3.0 ND 1.1 1.9 −2.7 1.0

C8H9NO C8H9NO C8H11NO2 C8H11NO3 C10H16O6

136.0757 136.0757 154.0865 170.0807 231.0878

136.0757 136.0757 154.0863 170.0812 231.0874

0.0 0.0 1.2 −2.9 1.7

C18H26N4O7S C28H41N7O12S2

441.1451 730.2187

441.1449 730.2182

0.4 0.7

molecular formula

a

Abbreviations: PDs, products; RT, retention time; RET, retrorsine; ND, not determined. bData generated by ESI-Q-ToF. cEstimated as ((observed mass − predicted mass)/observed mass) × 106. dIdentified after both electrochemical and human liver microsomal oxidation of retrorsine. (A) Primary oxidation products of retrorsine ([1]−[7]) in the online EC/LC/MS setup, (B) secondary EC cell oxidation products of retrorsine ([8]− [12]) in the offline EC/LC/MS setup, and (C) glutathione conjugate [13] and [14].

Figure 2. Electrochemical oxidation of retrorsine. (A) Effect of the applied voltage (mV) on the oxidative degradation of retrorsine and the formation of oxidation products [1] and [6]. (B) Effect of the applied voltage on the formation of [2]−[5] and [7]. 7.1 Hz), 1.41 (3H, t, J = 7.1 Hz); 13C NMR (125.8 MHz, CDCl3) δ 172.3, 162.3, 129.5, 128.2, 122.4, 112.2, 62.4, 14.1. Preparation of [19]. Compound [19] was prepared using a known method34 in 12% yield (major isomer). 1H NMR (500.1 MHz, CDCl3) δ 6.93 (1H, bs), 6.85 (1H, d, J = 2.7 Hz), 6.65 (1H, d, J = 2.7 Hz), 6.45 (1H, bs), 4.43 (2H, bs), 4.28 (2H, q, J = 7.1 Hz), 1.35 (3H, t, J = 7.1 Hz); 13C NMR (125.8 MHz, CDCl3) δ 164.9, 145.9, 131.6, 124.2, 117.4, 113.2, 105.0, 59.5, 52.7, 14.6. Preparation of [8]. Powdered LiAlH4 (480 mg, 12.6 mmol) in dry ether (10 mL) was refluxed for 20 min and cooled to −5 °C followed by dropwise addition of [19] (720 mg, 4.1 mmol) at this temperature in dry ether (10 mL) in 30 min. After stirring at −5 °C for 1 h, water (1.2 mL) was added followed by NaOH (5 M, 4.2 mL), and the reaction mixture was allowed to react for 30 min. The ether layer was separated, and the water phase was extracted with dry ether (2 × 10 mL). The combined ether layers were dried over Mg2SO4 and evaporated to dryness to yield [8] (490 mg, 89%), a pale yellow viscous liquid (purity 95%, Figure S1a−d). 1H NMR (500.1 MHz, CDCl3) δ 6.89 (1H, m), 6.67 (1H, dt, J = 6.1 and 2.3 Hz), 6.23 (1H, d, J = 2.3 Hz), 6.20 (1H, m), 4.60 (2H, bs), 4.42 (2H, bs); 13C NMR (125.8 MHz, CDCl3) δ 139.6, 127.5, 122.5, 117.0, 112.5, 111.8, 58.5, 52.0. Preparation of (3H-Pyrrolizin-1-yl)methanol [20]. This isomer was easily formed from compound [8] during storage or handling under light. 1H NMR (500.1 MHz, CDCl3) δ 6.94 (1H, m), 6.26 (1H, m), 6.09 (1H, s), 5.94 (1H, d, J = 3.2 Hz), 4.56 (2H, bs), 4.43 (2H,

m); 13C NMR (125.8 MHz, CDCl3) δ 140.1, 137.8, 121.8, 117.2, 111.8, 96.3, 59.1, 51.4.

■

RESULTS Electrochemical Oxidation of Retrorsine. In the preliminary EC cell oxidation of 100 μM retrorsine in the direct EC/MS setup (Figure 1A) at +500 mV or higher, four different protonated molecular [M + H]+ ions at m/z 350, 366, 368, and 384 were detected. None of these products were detected at +0.00 mV or in the blank sample (0 μM retrorsine). According to the ESI-MS ion count, the highest yields of the oxidation products were achieved at +800 mV; hence, +800 mV was used as an optimal voltage for the EC oxidation of retrorsine in 20 mM ammonium acetate buffer in our experimental setups. In order to precisely investigate the above molecular ions, the EC cell was coupled online with LC/ ESI-MS (Figure 1B) to allow for chromatographic separation and subsequent identification. This online EC/LC/MS experiment revealed the formation of seven different oxidation products ([1]−[7]) of retrorsine at +800 mV (Table 1A). The highest yields of the primary oxidation products were achieved at the applied potentials between +700 and +900 mV (Figure 2A,B). At +1000 mV or higher potentials, we could not observe [1] and [6], although this may be due to the fact that the 1953

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

Chart 1. Proposed Molecular Structures of the Electrochemical Oxidation Products of Retrorsine

eluent by LC/MS. The results indicated that [1] and [3] were unstable and could not be detected in the offline EC/LC/MS setup; [2] and [7] were stable for the first 20 and 30 min after oxidation at room temperature, respectively; and [4]−[6] remained stable over the entire time period studied (Figure 3).

oxidation products had been further oxidized. Product [1] has been hypothesized to be a pyrrolic ester intermediate, and [5] is identified as retrorsine N-oxide. Products [2], [3], [4], [6], and [7] were new. These are the first experiments to demonstrate the formation of several products after EC cell oxidation of retrorsine. Identification of the EC Oxidation Products. The Proposed molecular structures of all products ([1]−[14]) are depicted in Chart 1, and their MS/MS fragments, accurate masses, and accurate mass measurement errors are presented in Table 1A−C. Product [1] (m/z 350; [M + H]+) had a molecular weight 2 Da less than the parent alkaloid retrorsine (m/z 352; [M + H]+), and it was fragmented predominantly to m/z 118 in the ESI-MS/MS. We observed [1] only after the online EC/LC/MS oxidation of retrorsine but not in the offline EC/LC/MS setup. When the LC column fraction of [1] was collected and analyzed by LC/MS, only [8] was detected in the positive mode. On the basis of the molecular weight, MS/MS data, and instability profile, we concluded that [1] was dehydroretrorsine. [2]−[4] (m/z 366) and [7] (m/z 384) had characteristic MS/MS fragment ions at m/z 134, 152, and 348. However, since [2]−[4] and [7] were not detected after microsomal metabolism of retrorsine, they do not seem to be toxicologically relevant. Products [5] and [6] (m/z 368) were both 16 Da higher than retrorsine, evidence for addition of an oxygen atom. Because compound [5] showed characteristic MS/MS fragment ion clusters at m/z 118−120 and m/z 136− 138,35 we identified it as the N-oxide of retrorsine. However, the rate of [5] formation was very low. For instance, according to the ESI-MS ion counts, [7] (m/z 384) was over 140 times more abundant than that of [5]. Stability of the Oxidation Products. We studied the stability of the oxidation products of retrorsine in the absence or presence of reduced GSH. In an attempt to evaluate the intrinsic instability of the products, we oxidized 100 μM retrorsine at +800 mV, collected the eluent, and analyzed the

Figure 3. Stability of the electrochemical oxidation products of retrorsine. [1] and [3] were not detected in the collected sample, and [2] and [7] were stable for 20 and 30 min, respectively. [4]−[6] were stable for the entire study period. The chromatographic separation of the oxidation products was achieved with an Xterra MS C8, 3.5 μm 2.1 × 20 mm IS column using HPLC method I as described in the Materials and Methods.

In addition, four new products, [8]−[11], as well one known compound, [12], were observed in the collected sample. The molecular weights of these products were smaller compared to those of [1]−[7], showing extensive ester bond cleavage at the C7 and C9 facilitated by [12] as a good leaving group. The accurate MS, MS/MS, and mass measurement error data by the Q-ToF analysis (Table 1B) showed that [8]−[11] shared similarities, i.e., all originated from the amino alcohol moiety of retrorsine. Product [12] (at m/z 231 [M − H]−) was identified as necic acid based on the MS/MS data, i.e., the carboxylic acid side chain of retrorsine. 1954

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

Figure 4. Reactivity of electrochemical oxidation products of retrorsine. (A) Effect of glutathione (GSH) concentration and (B) effect of the incubation time on the rates of formation of GSH conjugates [13] and [14]. The chromatographic separation of the oxidation products was achieved with a Hypersil graphite column using HPLC method II as described in the Materials and Methods.

When the collected eluent was incubated with 4 mM GSH at room temperature, two GSH conjugates were detected at m/z 441 [13] and m/z 730 [14] in the negative ESI-MS mode (Table 1C). The rates of [13] and [14] formation were dependent on both the GSH concentration (Figure 4A) and the incubation time (Figure 4B). Inclusion of GSH did not affect the stability of [4], [5], [6], [10], [11], and [12]; however, [8] and [9] (both at m/z 136) completely disappeared. In order to confirm whether [8] and [9] were reactive products with GSH in the oxidized solution, we collected and incubated the column fractions of [8] or [9] with 4 mM GSH. The results confirmed that [8] and [9] were the products responsible for the formation of [13] and [14]. The same GSH conjugates, [13] and [14], were also observed when retrorsine was incubated with HLM in the presence of 4 mM reduced GSH. DHPs are believed to react, in vivo and in vitro, with GSH to produce these conjugates.20,36 However, our results showed that these conjugates could be formed during the reaction between GSH and [8] or its isomer [9] at room temperature. [8] and [9] (m/z 136) as Metabolites of Retrorsine with HLM. Because the EC cell is purely an instrumental method, we investigated the formation of these small reactive metabolites ([8] and [9]) after the metabolism of retrorsine with HLM in vitro and observed the formation of two metabolites at m/z 136 that had the same LC/MS properties (retention time, MS/MS fragment ions, and accurate mass) as those of [8] and [9], respectively (Figure 5). In the presence of 4 mM reduced GSH, two GSH conjugates were detected at m/ z 441 and 730 in the negative ESI-MS mode that were identified as 7-GSH-DHP [13] and 7,9-di-GSH-DHP [14], respectively. Moreover, the rates of [8] and [9] formation were significantly reduced in the presence of GSH in the incubation mixture. Although [5] and [12] could be detected after incubation of retrorsine with HLM, inclusion of GSH in the incubation mixture did not affect the stability of products [5] and [12]. Determination of the Molecular Structure of [8]. In an attempt to determine the molecular structure of [8], we utilized three different approaches. First, on the basis of the accurate mass (136.0757 [M + H]+) and its MS/MS fragment ions (Table 1B), the elemental composition of [8] was determined to be C8H9NO. Theoretically, this molecular formula could represent [8] or its isomers, [20] and [21] (Scheme 1B). Second, a large quantity of [8] was prepared by the scaled-up incubation of 4 mg of retrorsine with HLM, and its NMR spectral data represented the target molecule, [8]. Finally, the target metabolite was chemically synthesized (Scheme 1A), and its NMR (see Materials and Methods, preparation of [8]) and

Figure 5. Formation [8] and [9] (m/z 136) after oxidative metabolism of retrorsine with human liver microsomes (A) and an electrochemical cell (B). LC/MS analysis was performed by LC/MS in the positive mode, and chromatographic separation was achieved with a Hypersil graphite column using HPLC method II as described in the Materials and Methods.

Q-ToF spectral data were acquired and confirmed that it represented the molecular structure of [8]. [9], a minor metabolite of retrorsine at m/z 136, was also reactive toward reduced GSH and formed identical GSH conjugates ([13] and [14]), and its MS/MS fragmentation pattern was the same as that of [8].

■

DISCUSSION In vitro and in vivo metabolism studies have proved that hepatotoxic PAs such as retrorsine are bioactivated by CYPs to very reactive and unstable DHPs.15,20 Our observations were consistent with this hypothesis, as dehydroretrorsine [1] was detected only during the online EC/LC/MS oxidation of retrorsine. Instead, [1] was degraded to a new metabolite that was subsequently identified as [8], and the EC product proved to be identical to metabolites found in HLM incubations with retrorsine and other hepatotoxic PAs such as lasiocarpine and senkirkin (unpublished data) in vitro. Furthermore, [8] readily reacted with reduced GSH, producing previously known monoand di-GSH−DHP conjugates,25,37 showing that it might be a toxicologically important metabolite. As far as we are aware, [8] has not previously been reported to be a metabolite of PAs or even an organic chemical entity endowed with a specific CAS number. Hence, the main achievement of the current study was the identification of [8] as a new metabolite with potential toxicological implications. 1955

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

interface. Instead, two well-separated LC/MS peaks were observed at m/z 136 ([8] and [9]) in both matrices (Figure 5). One may assume that DHR might have lost a water molecule during the ionization process at the ESI interface and produced a pseudomolecular ion at m/z 136 [154 − H2O]. In fact, this phenomenon has previously been reported in the ESI-MS analysis of DHR.41 In contradiction to these observations, the 1 H NMR analysis of the synthetic product confirmed that [8] was not formed during the ionization process at the ESI-MS interface. Moreover, the retention times of synthetic and biological [8] were obtained and found to be identical (Figure S2). In conclusion, the current study both identified and clarified the structural characterization of [8], a new reactive metabolite of retrorsine. The molecular structure of this novel metabolite was determined by state-of-the-art techniques such as Q-ToF and 1H and 13C NMR spectroscopy. The metabolite, [8], was a degradation product of dehydroretrorsine, a putative toxic metabolite of retrorsine, and it readily reacted with reduced GSH, resulting in the formation of two GSH conjugates: monoand di-GSH−DHP. Furthermore, [8] was also detected after metabolism of lasiocarpine and senkirkin, showing that it is a common, and also potentially toxic, metabolite of hepatotoxic PAs. In addition, the study also showed that the use of EC made it possible to produce and identify short-lived metabolites of retrorsine, such as [1].

In order to characterize the molecular structure of [8], we examined two previously presented hypotheses: (1) hepatotoxic PAs are biotransformed to very unstable and reactive intermediate pyrrolic esters called DHPs15 and (2) DHPs form stabilized carbonium ions each at C7 and C9 assisted by necic acid as a good leaving group.9 Thus, we hypothesized that [8] (m/z 136) would be formed through resonance stabilization of the carbonium ion at C7 driven by anchimeric assistance from the lone pair of electrons at the N atom of the pyrrolic ring to form a more stable metabolite [8]. On the basis of these assumptions as well as the MS/MS data and the known molecular structures of the GSH conjugates, the proposed molecular structure of [8] was predicted. Moreover, from the point of view of the reaction mechanism, the only isomer to lead to C7-substituted GSH derivatives is isomer [8]. In this isomer, the double bond between C6 and C7 is activated due to the presence of a nitrogen lone electron pair, leading to a resonance structure in which the negative charge is located at C6. In the next step, a four-centered reaction is possible (Scheme 1C) in which C6 reacts with proton and C7 with sulfur in the GSH molecule, leading to the final product, e.g., [13]. Therefore, the molecular structure of [8] was confirmed to be (3H-pyrrolozin-7-yl)methanol. The second reactive metabolite of retrorsine, [9], which was detected after both microsomal and EC oxidation of retrorsine, was not observed in the synthetic sample and hence it was not possible to determine its molecular structure. Because [8] and [9] had very similar MS/MS fragment ions, had the same accurate masses, and formed identical GSH conjugates, we concluded that [9] was an isomer of [8]. In fact, it seemed that compound [8] isomerism was a common phenomenon. For instance, compound [9] both after microsomal and EC cell oxidation of retrorsine and compound [20] after chemical synthesis were identified as isomers of [8]. Because we did not observe [9] as a degradation product of [1], we could not confirm whether it had been formed from the same biotransformation route as that for [8] or if it had a different but unidentified intermediate precursor metabolite. The toxic metabolites of PAs are very reactive toward nucleophiles such as DNA and proteins, resulting in VOD, cancer, and genotoxicity.38,39 The mutagenic actions of PAs are mediated mainly through the alkylation of DNA,39 whereas hepatic VOD may be primarily initiated after sinusoidal endothelial cell damage either via a direct metabolite−protein interaction or through increased oxidative stress as a result of GSH depletion.38,40 In order to elicit such detrimental actions, the putative ultimate toxic metabolites, DHPs, might have to be transported to the nucleus of hepatocytes or endothelial cells from the site of production, i.e., the endoplasmic reticulum in the hepatocytes. However, DHPs are short-lived and may not be stable enough to allow them to cross intact through the membrane barriers. The new metabolite, [8], not only seemed to be reactive but also was sufficiently stable to be transported to the putative sites of toxicity. Moreover, consistent with the previous observation that the reactive metabolites of PAs have two reactive sites at the C7 and C9, the reaction between [8] resulted in the formation of di-GSH conjugate, a metabolite conjugated to two GSH molecules, i.e., it is a powerful GSH depletor. Earlier studies have indicated that DHR is one of the major metabolites of hepatotoxic PAs.15,41 DHR was not detected after EC or microsomal oxidation of retrorsine in our analytical LC/MS setups such as Q-ToF and LTQ coupled with ESI

■

ASSOCIATED CONTENT

S Supporting Information *

Purity (Figure S1) and LC/MS chromatography (Figure S2) of compound [8]. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: markku.pasanen@uef.fi. Funding

This study was funded by the Academy of Finland (grant no. 137589), FPDP-Toxicology section, and by University of Eastern Finland Strategic Funding. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We thank H. Jaatinen, P. Hänninen, and M. Salminkoski for technical assistance. ABBREVIATION COSY, correlation spectroscopy; DHPs, dehydropyrrolizidine alkaloids; DHR, dehydroretronecine; EC, electrochemistry (or electrochemical, depending on the context); HSQC, heteronuclear single quantum correlation spectroscopy; HLM, human liver microsomes; HRMS, high-resolution mass spectrometer; IPCS, International Program on Chemical Safety; Q-ToF, timeof-flight; TOCSY, total correlation spectroscopy; VOD, venoocclusive disease

■

REFERENCES

(1) Clark, A. M. (1960) The mutagenic activity of some pyrrolizidine alkaloids in Drosophila. Z. Vererbungsl. 91, 74−80. (2) Smith, L. W., and Culvenor, C. C. J. (1981) Plant sources of hepatotoxic pyrrolizidine alkaloids. J. Nat. Prod. 44, 129−152.

1956

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957

Chemical Research in Toxicology

Article

(3) Betz, J. M., Eppley, R. M., Taylor, W. C., and Andrzejewski, D. (1994) Determination of pyrrolizidine alkaloids in commercial comfrey products (Symphytum sp.). J. Pharm. Sci. 83, 649−653. (4) EFSA Panel on Contaminants in the Food Chain (CONTAM) (2011) Scientific opinion on pyrrolizidine alkaloids in food and feed. EFSA J. 9, 2406. (5) Edgar, J. A., Colegate, S. M., Boppre, M., and Molyneux, R. J. (2011) Pyrrolizidine alkaloids in food: a spectrum of potential health consequences. Food Addit. Contam., Part A 28, 308−324. (6) Tandon, H. D., Tandon, B. N., and Mattocks, A. R. (1978) An epidemic of veno-occlusive disease of the liver in Afghanistan. Pathologic features. Am. J. Gastroenterol. 70, 607−613. (7) Kim, H. Y., Stermitz, F. R., Molyneux, R. J., Wilson, D. W., Taylor, D., and Coulombe, R. A., Jr. (1993) Structural influences on pyrrolizidine alkaloid-induced cytopathology. Toxicol. Appl. Pharmacol. 122, 61−69. (8) Fu, P. P., Xia, Q., Lin, G., and Chou, M. W. (2002) Genotoxic pyrrolizidine alkaloidsmechanisms leading to DNA adduct formation and tumorigenicity. Int. J. Mol. Sci. 3, 948−964. (9) Wiedenfeld, H., and Edgar, J. (2011) Toxicity of pyrrolizidine alkaloids to humans and ruminants. Phytochem. Rev. 10, 137−151. (10) Kakar, F., Akbarian, Z., Leslie, T., Mustafa, M. L., Watson, J., Van Egmond, H. P., Omar, M. F., and Mofleh, J. (2010) An outbreak of hepatic veno-occlusive disease in Western Afghanistan associated with exposure to wheat flour contaminated with pyrrolizidine alkaloids. J. Toxicol. 2010, 313280. (11) Molyneux, R. J., Gardner, D. L., Colegate, S. M., and Edgar, J. A. (2011) Pyrrolizidine alkaloid toxicity in livestock: a paradigm for human poisoning? Food Addit. Contam., Part A 28, 293−307. (12) Stegelmeier, B. (2011) Pyrrolizidine alkaloid-containing toxic plants (Senecio, Crotalaria, Cynoglossum, Amsinckia, Heliotropium, and Echium spp.). Vet. Clin. North Am.: Food. Anim. Pract. 27, 419− 428. (13) Robinson, O., Want, E., Coen, M., Kennedy, R., van den Bosch, C., Gebrehawaria, Y., Kudo, H., Sadiq, F., Goldin, R. D., Hauser, M. L., Fenwick, A., Toledano, M. B., and Thursz, M. R. (2014) Hirmi Valley liver disease: A disease associated with exposure to pyrrolizidine alkaloids and DDT. J. Hepatol. 60, 96−102. (14) (2013) Pyrrolizidine alkaloids in herbal teas and teas, Bundesinstitut für Risikobewertung, http://www.bfr.bund.de/cm/ 349/pyrrolizidine-alkaloids-in-herbal-teas-and-teas.pdf. (15) Mattocks, A. R., and White, I. N. H. (1971) The conversion of pyrrolizidine alkaloids to N-oxides and to dihydropyrrolizine derivatives by rat-liver microsomes in vitro. Chem.−Biol. Interact. 3, 383−396. (16) White, I. N. H., Mattocks, A. R., and Butler, W. H. (1973) The conversion of the pyrrolizidine alkaloid retrorsine to pyrrolic derivatives in vivo and in vitro and its acute toxicity to various animal species. Chem.−Biol. Interact. 6, 207−218. (17) International Programme on Chemical Safety (1988) Environmental health criteria 80: pyrrolizidine alkaloids, WHO, Geneva, Switzerland, http://www.inchem.org/documents/ehc/ehc/ehc080. htm. (18) Reid, M. J., Lamé, M. W., Morin, D., Wilson, D. W., and Segall, H. J. (1998) Involvement of cytochrome P450 3A in the metabolism and covalent binding of 14C-monocrotaline in rat liver microsomes. J. Biochem. Mol. Toxicol. 12, 157−166. (19) Dueker, S. R., Lame, M. W., Morin, D., Wilson, D. W., and Segall, H. J. (1992) Guinea pig and rat hepatic microsomal metabolism of monocrotaline. Drug Metab. Dispos. 20, 275−280. (20) Fu, P. P., Xia, Q., Lin, G., and Chou, M. W. (2004) Pyrrolizidine alkaloidsgenotoxicity, metabolism enzymes, metabolic activation, and mechanisms. Drug Metab. Rev. 36, 1−55. (21) Williams, D. E., Reed, R. L., Kedzierski, B., Ziegler, D. M., and Buhler, D. R. (1989) The role of flavin-containing monooxygenase in the N-oxidation of the pyrrolizidine alkaloid senecionine. Drug Metab. Dispos. 17, 380−386. (22) Miranda, C. L., Chung, W., Reed, R. E., Zhao, X., Henderson, M. C., Wang, J., Williams, D. E., and Buhler, D. R. (1991) Flavin-

containing monooxygenase: a major detoxifying enzyme for the pyrrolizidine alkaloid senecionine in guinea pig tissues. Biochem. Biophys. Res. Commun. 178, 546−552. (23) Mattocks, A. R., and Bird, I. (1983) Pyrrolic and N-oxide metabolites formed from pyrrolizidine alkaloids by hepatic microsomes in vitro: Relevance to in vivo hepatotoxicity. Chem.−Biol. Interact. 43, 209−222. (24) Dai, J., Zhang, F., and Zheng, J. (2010) Retrorsine, but not monocrotaline, is a mechanism-based inactivator of P450 3A4. Chem.− Biol. Interact. 183, 49−56. (25) Lin, G., Cui, Y., and Hawes, E. M. (1998) Microsomal formation of a pyrrolic alcohol glutathione conjugate of clivorine: firm evidence for the formation of a pyrrolic metabolite of an otonecine-type pyrrolizidine alkaloid. Drug Metab. Dispos. 26, 181−184. (26) Yan, J., Xia, Q., Chou, M. W., and Fu, P. P. (2008) Metabolic activation of retronecine and retronecine N-oxide formation of DHPderived DNA adducts. Toxicol. Ind. Health 24, 181−188. (27) Zhao, Y., Xia, Q., Gamboa Da Costa, G., Yu, H., Cai, L., and Fu, P. P. (2012) Full structure assignments of pyrrolizidine alkaloid DNA adducts and mechanism of tumor initiation. Chem. Res. Toxicol. 25, 1985−1996. (28) Jurva, U., Wikström, H. V., Weidolf, L., and Bruins, A. P. (2003) Comparison between electrochemistry/mass spectrometry and cytochrome P450 catalyzed oxidation reactions. Rapid Commun. Mass Spectrom. 17, 800−810. (29) Lohmann, W., and Karst, U. (2006) Simulation of the detoxification of paracetamol using on-line electrochemistry/liquid chromatography/mass spectrometry. Anal. Bioanal. Chem. 386, 1701− 1708. (30) Odijk, M., Baumann, A., Olthuis, W., van den Berg, A., and Karst, U. (2006) Electrochemistry-on-chip for on-line conversions in drug metabolism studies. Biosens. Bioelectron. 26, 1521−1527. (31) Jahn, S., and Karst, U. (2012) Electrochemistry coupled to (liquid chromatography/) mass spectrometrycurrent state and future perspectives. J. Chromatogr. A 1259, 16−49. (32) Pelkonen, O., Kaltiala, E. H., Larmi, T. K., and Kärki, N. T. (1974) Cytochrome P-450-linked monooxygenase system and druginduced spectral interactions in human liver microsomes. Chem.−Biol. Interact. 9, 205−216. (33) Bradford, M. M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal. Biochem. 72, 248−254. (34) Brandänge, S., and Lundin, C. (1971) A synthesis of endo- and exo-1-ethoxycarbonylpyrrolizidne. Acta Chem. Scand. 25, 2447−2450. (35) Ruan, J., Li, N., Xia, Q., Fu, P. P., Peng, S., Ye, Y., and Lin, G. (2012) Characteristic ion clusters as determinants for the identification of pyrrolizidine alkaloid N-oxides in pyrrolizidine alkaloid-containing natural products using HPLC-MS analysis. J. Mass Spectrom. 47, 331−337. (36) Tamta, H., Pawar, R. S., Wamer, W. G., Grundel, E., Krynitsky, A. J., and Rader, J. I. (2012) Comparison of metabolism-mediated effects of pyrrolizidine alkaloids in a HepG2/C3A cell-S9 coincubation system and quantification of their glutathione conjugates. Xenobiotica 42, 1038−1048. (37) Chen, Z., and Huo, J. (2010) Hepatic veno-occlusive disease associated with toxicity of pyrrolizidine alkaloids in herbal preparations. Neth. J. Med. 68, 252−260. (38) Chen, T., Mei, N., and Fu, P. P. (2010) Genotoxicity of pyrrolizidine alkaloids. J. Appl. Toxicol. 30, 183−196. (39) Wang, X., Kanel, G. C., and DeLeve, L. D. (2000) Support of sinusoidal endothelial cell glutathione prevents hepatic veno-occlusive disease in the rat. Hepatology 31, 428−434. (40) Robertson, K. (1982) Alkylation of N2 in deoxyguanosine by dehydroretronecine, a carcinogenic metabolite of the pyrrolizidine alkaloid monocrotaline. Cancer Res. 41, 8−14. (41) Lin, G., Cui, Y., and Hawes, E. M. (2000) Characterization of rat liver microsomal metabolites of clivorine, an hepatotoxic otonecinetype pyrrolizidine alkaloid. Drug Metab. Dispos. 28, 1475−1483.

1957

dx.doi.org/10.1021/tx5002964 | Chem. Res. Toxicol. 2014, 27, 1950−1957