Supplemental Material can be found at: http://dmd.aspetjournals.org/content/suppl/2014/04/02/dmd.114.057414.DC1.html

1521-009X/42/6/1044–1054$25.00 DRUG METABOLISM AND DISPOSITION Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics

http://dx.doi.org/10.1124/dmd.114.057414 Drug Metab Dispos 42:1044–1054, June 2014

Metabolic Activation of the Indoloquinazoline Alkaloids Evodiamine and Rutaecarpine by Human Liver Microsomes: Dehydrogenation and Inactivation of Cytochrome P450 3A4 s Bo Wen, Vikram Roongta, Liling Liu, and David J. Moore Drug Metabolism, Non-Clinical Safety (B.W., L.L., D.J.M.) and Discovery Chemistry (V.R.), Hoffmann-La Roche, Nutley, New Jersey

ABSTRACT Evodiamine and rutaecarpine are the main active indoloquinazoline alkaloids of the herbal medicine Evodia rutaecarpa, which is widely used for the treatment of hypertension, abdominal pain, angina pectoris, gastrointestinal disorder, and headache. Immunosuppressive effects and acute toxicity were reported in mice treated with evodiamine and rutaecarpine. Although the mechanism remains unknown, it is proposed that metabolic activation of the indoloquinazoline alkaloids and subsequent covalent binding of reactive metabolites to cellular proteins play a causative role. Liquid chromatography–tandem mass spectrometry analysis of incubations containing evodiamine and NADPH-supplemented microsomes in the presence of glutathione (GSH) revealed formation of a major GSH conjugate which was subsequently indentified as a benzylic thioether adduct on the C-8 position of evodiamine by NMR analysis. Several other GSH conjugates were also detected, including conjugates of oxidized and demethylated metabolites of

evodiamine. Similar GSH conjugates were formed in incubations with rutaecarpine. These findings are consistent with a bioactivation sequence involving initial cytochrome P450–catalyzed dehydrogenation of the 3-alkylindole moiety in evodiamine and rutaecarpine to an electrophile 3-methyleneindolenine. Formation of the evodiamine and rutaecarpine GSH conjugates was primarily catalyzed by heterologously expressed recombinant CYP3A4 and, to a lesser extent, CYP1A2 and CYP2D6, respectively. It was found that the 3-methyleneindolenine or another reactive intermediate was a mechanism-based inactivator of CYP3A4, with inactivation parameters KI = 29 mM and kinact = 0.029 minute21, respectively. In summary, these findings are of significance in understanding the bioactivation mechanisms of indoloquinazoline alkaloids, and dehydrogenation of evodiamine and rutaecarpine may cause toxicities through formation of electrophilic intermediates and lead to drug-drug interactions mainly via CYP3A4 inactivation.

Introduction

(Jeon et al., 2006; Yang et al., 2006). Although the mechanism of immune-mediated toxicity is not clearly understood, a probable causal link between metabolic activation and the onset of immunotoxicity has been established (Jeon et al., 2006). Rutaecarpine undergoes extensive oxidative metabolism in human and rat liver microsomes mainly by aromatic hydroxylation to form 3-, 10-, 11-, and 12-hydroxyrutaecarpine (Ueng et al., 2005, 2006; Lee et al., 2006b) (Fig. 1). Dihydroxylation and hydroxylation of the aliphatic ring of rutaecarpine were also demonstrated (Lee et al., 2004, 2006b). One of the major biotransformation routes of rutaecarpine is the CYP1A-catalyzed oxidation to the metabolite 10- hydroxyrutaecarpine, which is excreted in rat urine as a glucuronide conjugate (Jan et al., 2006). Recently, evodiamine, another main active indoloquinazoline alkaloid present in Evodia rutaecarpa, was demonstrated to undergo similar cytochrome P450 (P450)–mediated oxidative biotransformation (Sun et al., 2013). In addition to aromatic and aliphatic hydroxylation, other metabolic pathways including N-demethylation of evodiamine were observed (Sun et al., 2013). Of particular interest in the metabolism pathways of rutaecarpine is the detection of 10- and 12-hydroxy metabolites of rutaecarpine (Ueng et al., 2005; Lee et al., 2006b). 10-Hydroxyrutaecarpine and 12-hydroxyrutaecarpine can undergo P450-mediated two-electron oxidations to form electrophilic

Evodiamine and rutaecarpine are the main active indoloquinazoline alkaloids of the herbal medicine Evodia rutaecarpa, which has a variety of pharmacological actions including antithrombotic, antiinflammatory, antianoxic, hypotensive, and vasodilatory effects (Sheu, 1999; Yu et al., 2013). In particular, the cytotoxicity of evodiamine and rutaecarpine on various human cancer cell lines has been extensively studied (Ogasawara at al., 2001; Fei et al., 2003; Liao et al., 2005; Xu et al., 2006; Adams et al., 2007). It has been reported that evodiamine inhibited the proliferation of a wide variety of tumor cells by inducing apoptosis via different mechanisms, including caspase-dependent and -independent pathways, the sphingomyelin pathway, and PI3K/Akt/caspase and Fas-L/NF-kB signaling pathways (Lee et al., 2006a; Kan et al., 2007; Wang et al., 2010; Huang et al., 2011). More recently, highly potent evodiamine derivatives were discovered as novel antitumor agents by systemic structure-activity relationship analysis and biologic evaluations (Dong et al., 2012). Despite the therapeutic benefits, immune-mediated toxicity and acute toxicity were reported in mice treated with evodiamine and rutaecarpine dx.doi.org/10.1124/dmd.114.057414. s This article has supplemental material available at dmd.aspetjournals.org.

ABBREVIATIONS: COSY, homonuclear correlation spectroscopy; EPI, enhanced product ion; ESI-MS, electrospray ionization mass spectrometry; GSH, glutathione; HLM, human liver microsomes; HPLC, high-performance liquid chromatography; kobs, observed first-order rate constants; LC, liquid chromatography; 3MEI, 3-methyleneindolenine; MS/MS, tandem mass spectrometry; MSn, multiple stage mass spectrometry; P450, cytochrome P450; PI, precursor ion. 1044

Downloaded from dmd.aspetjournals.org at Penn State Hershey George T. Harrell Health Sciences Library on May 26, 2014

Received February 7, 2014; accepted April 2, 2014

Bioactivation of Indoloquinazoline Alkaloids para- and ortho-quinone imine intermediates, respectively, which are capable of reacting with cellular proteins and other nucleophiles, such as glutathione. In addition, as depicted in Fig. 1, rutaecarpine and evodiamine are indoloquinazoline alkaloids containing a 3-alkylindole ring system which can potentially undergo P450-mediated dehydrogenation reactions to form an electrophilic 3-methyleneindolenine (3MEI) species. However, to date, no such reactive intermediates have been reported, and the bioactivation mechanism of the indoloquinazoline alkaloids remains unknown. In the present study, we report the detection and identification of several glutathione (GSH) conjugates of evodiamine, derived from the addition of sulfhydryl nucleophile to monohydroxyevodiamine, evodiamine, and N-demethylated evodiamine. Similar conjugates were formed in the incubations of rutaecarpine. In addition, efforts are made to evaluate the relative contributions from individual P450 isoforms to formation of the GSH conjugates. These findings are of significance for understanding the relationship between metabolic activation, enzyme inactivation, and immune-mediated toxicity of evodiamine and rutaecarpine. Materials and Methods The following chemicals were purchased from Sigma-Aldrich (St. Louis, MO): evodiamine (racemic), rutaecarpine, GSH, trichloroacetic acid, and NADPH. Pooled human liver microsomes (HLM) and Supersomes containing cDNA-baculovirus-insect cell–expressed P450s (CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4) were obtained from BD Gentest (Woburn, MA). Formic acid, methanol, and acetonitrile were purchased from Fisher Scientific (Fair Lawn, NJ). All other commercially available reagents and solvents were of either analytical or highperformance liquid chromatography (HPLC) grade. In Vitro Metabolism. All incubations were performed at 37°C in a water bath. Pooled human liver microsomes and human cDNA-expressed P450 isozymes were carefully thawed on ice prior to the experiment. Evodiamine or rutaecarpine (10 and 50 mM) was mixed with human liver microsomal proteins

Fig. 1. Structures of evodiamine and rutaecarpine, and the rutaecarpine hydroxy metabolites.

1045

(1 mg/ml) in 100 mM potassium phosphate buffer (pH 7.4) supplemented with 1 mM GSH. The total incubation volume was 1 ml. After 3-minute preincubation at 37°C, the incubation reactions were initiated by the addition of 1 mM NADPH. Reactions were terminated by the addition of 150 ml of trichloroacetic acid (10%) after 60-minute incubation. Incubations with the recombinant cDNA-expressed P450 isozymes were performed similarly except that liver microsomes were substituted by Supersomes, including CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4 (100 nM). Control samples containing no NADPH or substrates were included. Each incubation was performed in triplicate. Samples were centrifuged at 10,000g for 15 minutes at 4°C to pellet the precipitated proteins, and supernatants were subjected to liquid chromatography/tandem mass spectrometry (LC/MS/MS) analysis of GSH adducts. For human liver microsomal incubations, supernatants were concentrated by solid-phase extraction as described later, prior to LC/MSn analyses. Solid-Phase Extraction. Samples resulting from HLM incubations were desalted and concentrated by solid-phase extraction (SPE) prior to the LC/MS/ MS analyses. Briefly, SPE was performed using Oasis solid-phase extraction cartridges packed with 60 mg of sorbent C18 (Waters, Milford, MA). Cartridges were first washed with 2 ml of methanol and then conditioned with 2 ml of water. Supernatants resulting from centrifugation were loaded onto the cartridges, and cartridges were washed with 2 ml of water and then eluted with 2 ml of methanol. The methanol fractions were dried by nitrogen gas and reconstituted with 150 ml of a water-methanol (70:30) mixture. Aliquots (30 ml) of the reconstituted solutions were subjected to LC/MS/MS analysis. Instrumentation. Metabolite profiling was performed on an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA) coupled with an Agilent 1100 HPLC system (Agilent Technologies, Palo Alto, CA) as described previously, with modifications (Wen et al., 2008a). Separation was achieved using a Polaris C18 column (5 mm, 250  2.1 mm; Varian Inc., Palo Alto, CA) at a flow rate of 0.3 ml/min. A gradient of solvents A (water with 0.1% formic acid) and B (acetonitrile with 0.1% formic acid) was as follows: 10% solvent B for 5 minutes, followed by 10%–80% B in 30 minutes and 80%–95% B in 2 minutes. Formation of metabolites was estimated based on the photodiode array total scan chromatograms using an Agilent 8453 Diode Array UV-Visible Spectrophotometer (Agilent Technologies). Major operating parameters for the ion trap ESI-MS (electrospray ionization mass spectrometry) method were set as follows: capillary temperature, 300°C; spray voltage, 5.0 kV; capillary voltage, 15 V; sheath gas flow rate, 90 (arbitrary value); and auxiliary gas flow rate, 30 (arbitrary value). For a full scan, the automatic gain control was set at 5.0  108, maximum ion time was 100 ms, and the number of microscans was set at 3. For MSn scanning, the automatic gain control was set at 1.0  108, maximum ion time was 400 ms, and the number of microscans was set at 2. For data-dependent scanning, the default charge state was 1, default isolation width was 2.0, and normalized collision energy was 35. Polarity switching was applied to acquire full-scan and data-dependent MSn spectra in both the positive and negative ion mode. Fourier transform MS was set up to acquire high-resolution full-scan MS in the positive ion mode. Complete profiling of reactive metabolites was also carried out using the precursor ion (PI)–enhanced product ion (EPI) method previously described (Wen et al., 2008b). Briefly, the PI scan of m/z 272 was run in the negative mode with 0.2-Da step size, 5-ms pause between mass ranges, and 2-second scan rate or 50-ms dwell. The TurboIonSpray (AB Sciex, Foster City, CA) ion source conditions were optimized and set as follows: curtain gas = 35, collision gas = medium, ion spray voltage = –4500, and temperature = 500. Informationdependent acquisition was used to trigger acquisition of EPI spectra. The EPI scans were run in the positive mode at a scan range for daughter ions from m/z 100 to 1000. Data were processed using Analyst 4.1 software (Applied Biosystems, Foster City, CA). A Shimadzu HPLC system (Shimadzu Scientific Instruments, Columbia, MD) was coupled with an Agilent Eclipse XDB-Phenyl C18 column (3.0  150 mm, 3.5 mm; Agilent Technologies). The HPLC mobile phase A was 10 mM ammonium acetate in water with 0.1% formic acid, and mobile phase B was acetonitrile with 0.1% formic acid. A Shimadzu LC20AD solvent delivery module (Shimadzu Scientific Instruments) was used to produce the following gradient elution profile: 10% solvent B for 2 minutes, followed by 10%–70% B in 20 minutes and 70%–90% B in 2 minutes. LC/MS/ MS analyses were performed on 20-ml aliquots of cleaned samples. For relative comparison of GSH adduct levels, the mass spectrometer was operated in the

1046

Wen et al.

multiple reaction monitoring mode. Multiple reaction monitoring transitions were simultaneously monitored for detecting evodiamine GSH conjugate EM1: m/z 609→308; for EM2: m/z 625→496; for EM3: m/z 625→336; for EM4: m/z 595→308; for rutaecarpine GSH conjugate RM1: m/z 593→286; for RM2: m/z 609→480; and for RM3: m/z 625→496. Data were analyzed using Analyst 4.1 version software (Applied Biosystems). Metabolite Isolation and NMR Characterization. Human liver microsomal incubations with evodiamine (50 mM) were performed as described earlier on a 50-ml scale. Trichloroacetic acid (10%) was added to terminate the reaction after 60-minute incubation. Samples were centrifuged at 10,000g for 15 minutes at 4°C to pellet the precipitated proteins, and supernatants were concentrated by solid-phase extraction as described earlier. The major GSH conjugate EM1 was therefore isolated from the cleaned and concentrated reaction mixture. Separation was achieved using a Polaris C18 column (5 mm,

250  4.6 mm; Varian Inc.) at a flow rate of 1 ml/min. A gradient of solvents A (water with 0.1% formic acid) and B (acetonitrile with 0.1% formic acid) was as follows: 10% solvent B for 5 minutes, followed by 10%–50% B in 30 minutes and 50%–90% B in 2 minutes. All NMR spectra were acquired on a Bruker Avance 500 Ultrashield NMR spectrometer equipped with a 5-mm CPQNP z-gradient cryoprobe (Bruker, Rheinstetten, Germany) operating at a 1 H frequency of 500.13 MHz and 13 C frequency of 125.76 MHz. Each NMR sample was dissolved in 180 ml of methanol-d4 with tetramethylsilane as an internal reference standard. Water suppression was carried out on the samples as needed. All assignments were proven by conventional NMR experiments including the two-dimensional homonuclear correlation spectroscopy (COSY). Enzyme Inactivation. Midazolam 19-hydroxylase activity was determined to quantify time- and concentration-dependent loss of CYP3A4 activity of

Fig. 2. UV chromatograms (photodiode array total scan) of metabolites in human liver microsomal incubations of evodiamine (A) and rutaecarpine (B) in the presence of GSH and NADPH.

1047

Bioactivation of Indoloquinazoline Alkaloids

Fig. 3. Mass spectra of evodiamine GSH conjugate (EM1) obtained in the positive ion mode. (A) Full-scan mass spectrum of EM1. (B) MS/MS spectrum of EM1 at m/z 609 ([M + H]+). (C) MS3 mass spectrum of the fragment ion of EM1 at m/z 308.

HLMs in the presence of evodiamine, as described previously with modifications (Wen et al., 2009). Primary incubations included evodiamine (0, 1, 3, 10, 30, 50, and 100 mM), 1 mM NADPH, 1 mg/ml HLM, 3 mM MgCl2, and 0.1 M potassium phosphate buffer (pH 7.4). The mixture was incubated in

a 37°C shaking water bath for various time points (0, 2, 6, 12, and 20 minutes). At each preincubation time point, 10-ml aliquots of the primary incubation mixtures were transferred to a secondary incubation to a final volume of 100 ml, which included 10 mM midazolam, 1 mM NADPH, 3 mM MgCl2, and 0.1 M

TABLE 1 1

H NMR data of evodiamine and its GSH conjugate (EM1) (d in CD3OD)

Refer to Fig. 7 for signal assignments of the thioether moiety. Proton Signals Evodiamine

C-1 C-2 C-3 C-4 C-9 C-10 C-11 C-12 C-13b C-15 C-7a C-7b C-8

7.24 7.56 7.16 7.98 7.57 7.09 7.19 7.42 6.03 2.68 4.81 3.31 2.99

(1H, (1H, (1H, (1H, (1H, (1H, (1H, (1H, (1H, (3H, (1H, (1H, (2H,

bd, J = 8.1 Hz) ddd, J = 8.1, 7.9 and 1.8 Hz) bdd, J = 8.2 and 7.9 Hz) dd, J = 8.2 and 1.8 Hz) bd, J = 8.5 Hz) ddd, J = 8.5, 8.2 and 2.1 Hz) bdd, J = 8.4 and 8.2 Hz) dd, J = 8.4 and 2.1 Hz) s) s) dd, J = 5.8 and 3.1 Hz) dd, J = 5.8 and 3.2 Hz) dd, J = 3.1 and 3.2 Hz)

EM1

7.30 7.59 7.26 8.03 7.70 7.16 7.22 7.42 6.03 2.58 5.03 3.55 4.68

(1H, (1H. (1H, (1H, (1H, (1H, (1H, (1H, (1H, (3H, (1H, (1H, (1H,

bd, J = 8.2 Hz) bdd, J = 8.2, 7.9 and 1.8 Hz) bdd, J = 8.2 and 7.9 Hz) dd, J = 8.2 and 1.8 Hz) bd, J = 8.5 Hz) ddd, J = 8.5, 8.2 and 2.1 Hz) bdd, J = 8.3 and 8.2 Hz) dd, J = 8.3 and 2.1 Hz) s) s) dd, J = 5.3 and 1.4 Hz) dd, J = 5.3 and 1.6 Hz) dd, J = 1.4 and 1.6 Hz)

cysteine 19 –H, d 2.9/3.4; cysteine 29 –H, d 4.8; glutamate 39 –H, d 2.6; glutamate 49 –H, d 2.1/2.3; glutamate 59 –H, d 3.6; glycine 69 –H, d 3.8.

1048

Wen et al.

potassium phosphate (pH 7.4). The midazolam mixture was incubated in a 37°C shaking water bath for 5 minutes and stopped by the addition of ice-cold acetonitrile in a 1:1 ratio (v/v). The mixture was centrifuged at 10,000g for 15 minutes. The supernatant (10 ml) was injected and analyzed by a Sciex API 4000 triple quadrupole mass spectrometer (Applied Biosystems) coupled with a Shimadzu HPLC system (Shimadzu Scientific Instruments). 19-Hydroxymidazolam was separated using a Hypersil BDS C18 column (50  2.1 mm, 5 mm; Thermo Fisher Scientific) at a flow rate of 0.3 ml/min. Selective reaction monitoring experiments in the positive ionization mode were performed using a dwell time of 150 ms per transition to detect the following precursor (Q1) to product (Q3) ion pairs: 326–291 and 342–203 for midazolam and 19hydroxymidazolam, respectively. The HPLC mobile phase A was 5 mM ammonium acetate in water with 0.1% formic acid, and mobile phase B was acetonitrile with 0.1% formic acid. A Shimadzu LC10AD solvent delivery module (Shimadzu Scientific Instruments) was used to produce the following gradient elution profile: 10% solvent B for 0.5 minute, 10%–90% B in 3.5 minutes, followed by 90% B for 1 minute.

Results Characterization of GSH Conjugates in Human Liver Microsomes. For the LC/MS/MS analysis of GSH conjugates, samples generated from incubations with human liver microsomes were desalted and concentrated by solid-phase extractions, and resulting samples were subjected to both data-dependent LC/MSn scanning and PI-EPI experiments described earlier. The metabolic profiles of evodiamine and rutaecarpine in human liver microsomes showed several GSH conjugates EM1–4 and RM1–3, respectively (Fig. 2). There were several

other components in the metabolic profile of evodiamine—for instance, the peaks eluted at 12.6, 14.2, and 14.9 minutes (Fig. 2A) showed an [M + H]+ of m/z 320, suggesting that they are likely monohydroxylated metabolites of evodiamine (Sun et al., 2013). Similar monohydroxylated metabolites eluted at 12.0 and 12.3 minutes were observed in the metabolic profile of rutaecarpine (Fig. 2B). The major adduct EM1 displayed a molecular ion [M + H]+ of m/z 609, suggesting a GSH adduct with attachment of a glutathionyl group to evodiamine (Fig. 3A). Fragmentation of EM1 molecular ions resulted in a major product ion at m/z 302 derived from a neutral loss of GSH (307 Da) and afforded a protonated GSH product ion at m/z 308 (Fig. 3B), suggesting the presence of an aliphatic and/or benzylic thioether motif in this GSH adduct (Baillie and Davis, 1993). The MS2 spectrum of EM1 showed product ions at m/z 480 and 534 derived from a neutral loss of 129 and 75 Da, corresponding to elimination of the pyroglutamate and glycine of GSH, respectively (Fig. 3B). Further fragmentation of the product ion at m/z 308 afforded several fragment ions, including ions at m/z 290, 233, 179, and 162 solely derived from the GSH moiety (Fig. 3C). These data suggested that EM1 is a GSH adduct derived from the addition of the sulfhydryl nucleophile to the 3MEI species to form an aliphatic and/or benzylic thioether conjugate. This assignment was supported by the negative MS spectra of EM1 (Supplemental Fig. 1). Fragmentation of the [M – H]– ions of EM1 at m/z 607 afforded product ions at m/z 589, 306, 288, 272, and 254. The anion at m/z 272 corresponds to deprotonated g-glutamyl-dehydroalanylglycine originated from the glutathionyl moiety (Dieckhaus et al., 2005). To identify the chemical structure of EM1, this metabolite was

Scheme 1. Proposed mechanisms for the bioactivation and formation of GSH conjugates of evodiamine in human liver microsomes.

Bioactivation of Indoloquinazoline Alkaloids isolated, purified, and subjected to NMR analysis. Eight aromatic proton signals at 8.03 ppm (C-4), 7.70 ppm (C-9), 7.59 ppm (C-2), 7.42 ppm (C-12), 7.30 ppm (C-1), 7.26 ppm (C-3), 7.22 ppm (C-11), and 7.16 ppm (C-10) were present in the 1H NMR spectrum of EM1, respectively (Supplemental Fig. 2; Table 1), revealing the presence of all eight aromatic proton signals in evodiamine. Glutathionyl proton signals appeared in the region from 2.0 to 4.8 ppm. The 1H NMR spectrum of EM1 revealed the only proton signal at C-8 with a “downfield” shift to 4.68 ppm, compared with two proton signals (2.99 and 2.78 ppm) at C-8 in evodiamine (Supplemental Fig. 2; Table 1). The COSY spectrum of EM1 revealed strong cross-peaks between aromatic protons at C-3 and C-4 and between C-9 and C-10, consistent with cross-peaks observed in that of evodiamine (Fig. 7). It also revealed cross-peaks between aliphatic protons at C-7a and C-8 and between C-7b and C-8, in contrast to the multiple cross-peaks between protons at C-7a and C-8a,b and between C-7b and C-8a, b observed in the COSY spectrum of evodiamine (Supplemental Fig. 3). The COSY spectra of evodiamine and EM1 both showed magnetic inequivalence of the two aliphatic protons at C-7 with a difference in chemical shifts of approximately 1.5 ppm, presumably due to the different electron shielding effect of the adjacent nitrogen atom, whereas the protons at C-8 of evodiamine are shown to be equivalent. Comparison of the COSY spectra of evodiamine and EM1 also

1049

revealed a significant ”downfield” of ;1.7 ppm of the proton at C-8 of EM1 due to the inductive effects of the adjacent sulfur atom, with much less effect on the two protons at C-7 (Fig. 7; Supplemental Fig. 3). These data supported that the GSH conjugation occurred at C-8, instead of C-7 of evodiamine. Full assignment of proton signals of EM1 was achieved with the aid of the COSY experiment, and confirmed the structural assignment of EM1. Taken together, the MS and NMR data identified EM1 as an evodiamine GSH conjugate with the glutathionyl moiety attached at the C-8 position of evodiamine (Scheme 1). The MS spectrum of EM2 revealed a molecular ion [M + H]+ of m/z 625, suggesting a GSH adduct with the addition of the sulfhydryl nucleophile to monohydroxylated evodiamine (Fig. 4A). Fragmentation of EM2 molecular ions resulted in neutral loss of 129 and 75, corresponding to elimination of the pyroglutamate and glycine of GSH, respectively (Fig. 4B). The ion at m/z 350 was formed via cleavage of the sulfur-carbon bond of the glutathionyl moiety, and its occurrence suggested the presence of an aromatic thioether motif in this GSH adduct (Baillie and Davis, 1993). The presence of product ion at m/z 465 suggested both monohydroxylation and glutathione conjugation on the 3-alkylindole structural motif (Fig. 4B). Further fragmentation of the ion at m/z 496 afforded several fragment ions, including ions at m/z 478, 421, 350, 336, and 219 (Fig. 4C). These

Fig. 4. Mass spectra of evodiamine GSH conjugate (EM2) obtained in the positive ion mode. (A) Full-scan mass spectrum of EM2. (B) MS/MS spectrum of EM2 at m/z 625 ([M + H]+). (C) MS3 mass spectrum of the fragment ion of EM2 at m/z 496.

1050

Wen et al.

data suggested that EM2 is a GSH adduct with attachment of the glutathionyl moiety to the monohydroxylated alkylindole ring of evodiamine. The MS spectrum of EM3 revealed a molecular ion [M + H]+ of m/z 625, suggesting a GSH conjugate of monohydroxylated evodiamine (Fig. 5A). Fragmentation of EM3 molecular ions resulted in a neutral loss of 129 and 75, corresponding to elimination of the pyroglutamate and glycine of GSH, respectively (Fig. 5B). Similar to EM2, the ion at m/z 350 was formed presumably via cleavage of the sulfur-carbon bond of the glutathionyl moiety, suggesting the presence of an aromatic thioether motif in this GSH adduct (Fig. 5B). Further fragmentation of the ion at m/z 336 afforded fragment ions, including ions at m/z 319, 233, and 192 (Fig. 5C). In particular, the fragment ion at m/z 192 suggested that both monohydroxylation and glutathione conjugation occurred on the 3-alkylindole structural motif of evodiamine (Fig. 5C). A proposed structure for EM3, which is consistent with the collision-induced dissociation cleavage, is shown in Fig. 5. The minor adduct EM4 revealed a molecular ion [M + H]+ of m/z 595, suggesting a GSH adduct of the demethylated evodiamine (Fig. 6A). Similar to EM1, fragmentation of EM4 molecular ions resulted in a major product ion at m/z 288 derived from a neutral loss of GSH (307 Da) and afforded a protonated GSH product ion at m/z 308 (Fig. 6B), suggesting an aliphatic and/or benzylic thioether conjugate

(Baillie and Davis, 1993). The MS2 spectrum of EM4 also showed product ions at m/z 466 and 520 derived from a neutral loss of 129 and 75 Da, respectively (Fig. 6B). Further fragmentation of the product ion at m/z 308 afforded fragment ions at m/z 290, 233, 179, and 162 derived from the GSH moiety (Fig. 6C). These data suggested that EM4 is a benzylic thioether conjugate derived from addition of the sulfhydryl moiety to demethylated evodiamine. Rutaecarpine, another main active indoloquinazoline alkaloid of the herbal medicine Evodia rutaecarpa and a structural analog of evodiamine, formed similar GSH conjugates RM1, RM2, and RM3 in incubations with human liver microsomes. The major adduct RM1 displayed a molecular ion [M + H]+ of m/z 593, suggesting a GSH adduct with attachment of the sulfhydryl group to rutaecarpine (Supplemental Fig. 4). Fragmentation of RM1 molecular ions resulted in a major product ion at m/z 286 derived from a neutral loss of GSH (307 Da) and afforded a protonated GSH product ion at m/z 308. Similar to evodiamine GSH adducts EM1 and EM4, these data suggested that RM1 is an aliphatic and/or benzylic thioether conjugate (Baillie and Davis, 1993). The product ions at m/z 464 and 518 were derived from a neutral loss of 129 and 75 Da, corresponding to elimination of the pyroglutamate and glycine of GSH, respectively (Supplemental Fig. 4). Further fragmentation of the product ion at m/z 286 afforded several fragment ions, including ions at m/z 258, 185,

Fig. 5. Mass spectra of evodiamine GSH conjugate (EM3) obtained in the positive ion mode. (A) Full-scan mass spectrum of EM3. (B) MS/MS spectrum of EM3 at m/z 625 ([M + H]+). (C) MS3 mass spectrum of the fragment ion of EM3 at m/z 336.

Bioactivation of Indoloquinazoline Alkaloids

1051

Fig. 6. Mass spectra of evodiamine GSH conjugate (EM4) obtained in the positive ion mode. (A) Full-scan mass spectrum of EM4. (B) MS/MS spectrum of EM4 at m/z 595 ([M + H]+). (C) MS3 mass spectrum of the fragment ion of EM4 at m/z 308.

and 167 derived from the rutaecarpine moiety. A proposed structure for RM1, which is consistent with the collision-induced dissociation cleavage, is shown in Supplemental Fig. 4. This structural assignment was supported by the negative MS spectra of RM1 (Supplemental Fig. 5). The MS spectrum of RM2 revealed a molecular ion [M + H]+ of m/z 609, suggesting a GSH adduct with the addition of the sulfhydryl nucleophile to monohydroxylated rutaecarpine (Supplemental Fig. 6). The product ions at m/z 480 and 534 were derived from a neutral loss of 129 and 75, corresponding to elimination of the pyroglutamate and glycine of GSH, respectively. The ion at m/z 334 was formed via cleavage of the sulfur-carbon bond of the glutathionyl moiety, suggesting the presence of an aromatic thioether motif in RM2. The presence of product ion at m/z 232 suggested both monohydroxylation and sulfhydryl addition on the 3-alkylindole structural motif (Supplemental Fig. 6). Another minor adduct RM3 showed a molecular ion at m/z 625, suggesting a GSH adduct with the addition of the sulfhydryl nucleophile to dihydroxylated rutaecarpine (Supplemental Fig. 7). The major product ion at m/z 496 was derived from a neutral loss of 129. The fragment ions at m/z 248 and 230 suggested that both dihydroxylation and glutathione conjugation occurred on the 3-alkylindole structural motif of rutaecarpine. Taken together, the results suggested that RM2 and RM3 are aromatic thioether conjugates with addition of

the sulfhydryl moiety to the mono- and dihydroxylated metabolites of rutaecarpine, respectively. Formation of GSH Conjugates with Recombinant P450s. To investigate the roles of individual human P450 isozymes in the bioactivation of evodiamine and rutaecarpine, incubations were carried out with insect cell–expressed recombinant P450s. The rates of metabolite formation obtained from individual incubations with recombinant P450 enzymes were multiplied by the mean specific content of the corresponding P450 enzyme in human liver microsomes to obtain the normalized reaction rates for each P450 enzyme (Rodrigues, 1999; Supplemental Table 1). After normalization for the relative hepatic abundance of P450 isozymes, CYP3A4 was the predominant enzyme in the formation of the major GSH conjugate EM1 in incubations of evodiamine (Fig. 8A). Only trace amounts or no EM1 was detected in incubations with other P450 enzymes, including CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, and CYP2E1. Similarly, formation of EM2, EM3, and EM4 was predominantly carried out by CYP3A4 (Fig. 8A). CYP1A2 and CYP2C19 also catalyzed EM2 formation, and the levels were approximately 19% and 10% of those formed by CYP3A4, respectively. Similar contributions from CYP1A2 and CYP2C19 were observed with formation of EM3. For rutaecarpine GSH adducts, CYP3A4 was also the predominant enzyme in the formation of RM1–3

1052

Wen et al.

Fig. 7. 1H-1H COSY spectrum of EM1 showing the coupling between aliphatic protons at C-7 and C-8.

(Fig. 8B). Only trace amounts or no RM1 was detected in incubations with other P450 enzymes. In contrast, CYP2D6 contributed significantly to formation of RM2 and RM3, with levels of approximately 39% and 37% of those formed by CYP3A4, respectively (Fig. 8B). These results suggested that CYP3A4 is the major P450 enzyme involved in the formation of EM1, EM4, and RM1, presumably via a 3MEI intermediate via a dehydrogenation mechanism (Scheme 1). CYP3A4 Inactivation Kinetics of Evodiamine. Preincubation of evodiamine with human liver microsomes showed that CYP3A4 was inactivated in a time- and concentration-dependent manner (Fig. 9A). The observed first-order rate constants (kobs) of the inactivation reaction, at a specific evodiamine concentration, were calculated from the slopes of these lines. The hyperbolic plot of kobs versus evodiamine concentrations is shown in Fig. 9B, from which rate constants were obtained. The kinact was determined to be 0.029 minute–1 and KI was 29 mM, respectively. Discussion The results from the current investigation constitute the first report on the cytochrome P450–catalyzed bioactivation of evodiamine and rutaecarpine, two main active indoloquinazoline alkaloids from the herbal medicine Evodia rutaecarpa. Several GSH conjugates were formed in human liver microsomal incubations and characterized by LC/MS/MS and/or NMR experiments. Formation of the benzylic thioether conjugates EM1, EM4, and RM1 was mediated primarily by CYP3A4. It was demonstrated that the 3MEI and/or another reactive

intermediate of evodiamine was a mechanism-based inactivator of CYP3A4, with inactivation parameters KI = 29 mM and kinact = 0.029 minute21, respectively. These findings are of importance to understanding the bioactivation pathways of indoloquinazoline alkaloids and potential links to their mechanism of toxicity and drug-drug interactions. Bioactivation of evodiamine was found to be mediated likely via multiple mechanisms by human liver microsomes. For the major benzylic thioether adduct EM1, GSH conjugation occurred specifically at C-8 of the aliphatic ring of evodiamine, which was confirmed by LC/MSn and NMR analysis. A proposed mechanism for the formation of evodiamine GSH adducts EM1 and EM4 is depicted in Scheme 1. Upon an initial hydrogen atom abstraction from the 3-methylene position of the indole ring, evodiamine undergoes dehydrogenation reactions to generate a reactive 3MEI species. Such dehydrogenation mechanisms have been proposed for the bioactivation of several 3-alkylindole–containing compounds, such as 3-methylindole (Skiles and Yost, 1996; Thornton-Manning et al., 1996; Yan et al., 2007), zafirlukast (Kassahun et al., 2005), and more recently, a tumor necrosis factor-a inhibitor SPD-304 (Sun and Yost, 2008). In this mechanism, the cytochrome P450 iron-oxo (compound I) abstracts a hydrogen atom from the 3-methylene group at the C-8 position of evodiamine and, subsequently, a second electron to generate the highly electrophilic 3MEI intermediate (Scheme 1). The dehydrogenation mechanism generally competes with the hydroxyl rebound mechanism that would produce the corresponding alcohol. This was in parallel with a previous in vitro study in which an aliphatic hydroxyl

Bioactivation of Indoloquinazoline Alkaloids

Fig. 8. Formation of GSH conjugates of evodiamine (A) and rutaecarpine (B) in incubations with cDNA-expressed recombinant P450 isozymes. The enzyme activities were expressed as the percentage of CYP3A4 activity and shown as an average of three measurements.

metabolite of evodiamine was detected in human liver microsomes (Sun et al., 2013). Alternatively, hydration of 3MEI would lead to formation of the alcohol metabolite 8-hydroxyevodiamine and vice versa (Scheme 1). These findings suggested that 3-alkylindole moieties are potential toxicophores through the generation of highly electrophilic 3MEI species by dehydrogenation reactions. However, further investigation is needed to confirm such proposed mechanisms using techniques including stable isotope labeling and 18O incorporation (Skiles and Yost, 1996). Formation of EM2 and EM3 was presumably via a quinone imine intermediate. Upon initial hydroxylation by P450 enzymes, 10- or 12hydroxyevodiamine metabolite undergoes an overall two-electron oxidation to yield their respective iminoquinone species, which can be trapped by GSH via a nucleophilic addition to generate the GSH adducts EM2 and EM3, respectively (Scheme 1). Such a two-step oxidation mechanism has been proposed for the bioactivation of 3-methylindole (Yan et al., 2007) and other compounds forming p-aminophenol

1053

Fig. 9. (A) Time- and concentration-dependent inactivation of CYP3A4 by evodiamine. ♦, 0 mM;D, 1 mM;d, 3 mM;s, 10 mM; j, 30 mM; u, 50 mM;m, 100 mM. (B) The hyperbolic plot of kobs versus evodiamine concentrations in HLM.

metabolites, such as diclofenac (Shen et al., 1999; Poon et al., 2001) and lumiracoxib (Li et al., 2008; Kang et al., 2009). This was in parallel with previous studies wherein such hydroxylated metabolites of evodiamine and rutaecarpine were detected in liver microsomes (Ueng et al., 2005, 2006; Sun et al., 2013). The reactive iminoquinone species would covalently modify nucleophilic residues of cellular proteins and/or P450 enzymes leading to toxicity and enzyme inactivation. However, apart from these oxidative activation pathways, it is noteworthy that the role of phase II metabolism (e.g., glucuronidation, sulfation) of the hydroxylated metabolites, e.g., 10- or 12-hydroxyrutaecarpine, remains to be elucidated (Lee et al., 2005; Jan et al., 2006), and phase II conjugation of the hydroxylated metabolites would serve as an alternative clearance and/or detoxification pathway prior to further oxidations. The NADPH-dependent formation of GSH conjugates indicated that one or more P450 enzymes were involved in the generation of

1054

Wen et al.

reactive intermediates of evodiamine and rutaecarpine by human liver microsomes. Experiments with recombinant P450 enzymes revealed that formation of benzylic thioether GSH conjugates EM1, EM4, and RM1 was predominantly mediated by CYP3A4, suggesting that CYP3A4 is the predominant P450 isozyme involved in the dehydrogenation mechanism. Formation of EM2 and EM3 was primarily carried out by CYP3A4 and, to a lesser extent, CYP1A2 and CYP2C19, respectively. It was noteworthy that CYP2D6 contributed significantly to formation of rutaecarpine GSH conjugates RM2 and RM3, with levels of approximately 39% and 37% of those formed by CYP3A4, respectively. This was in line with previous findings that CYP2D6 played a significant role in rutaecarpine 10- and 12-hydroxylations (Ueng et al., 2006), which are the initial steps leading to formation of the iminoquinone species and corresponding GSH conjugates. In conclusion, we found that the 3-alkylindole moiety in the indoloquinazoline alkaloids evodiamine and rutaecarpine is a potential toxicophore through the generation of highly electrophilic 3MEI species by P450-mediated dehydrogenation reactions. CYP3A4 is the predominant P450 isozyme involved in the dehydrogenation mechanism, whereas CYP2D6 also played a significant role in oxidation and bioactivation of rutaecarpine via formation of iminoquinone species. It is our hypothesis that 3MEI or another reactive intermediate was a mechanism-based inactivator of CYP3A4. In summary, findings from the current study are of significance in understanding the bioactivation mechanisms of indoloquinazoline alkaloids, and dehydrogenation of evodiamine and rutaecarpine may cause toxicities through formation of electrophilic intermediates and lead to drug-drug interactions. Authorship Contributions Participated in research design: Wen, Roongta, Liu. Conducted experiments: Wen, Roongta, Liu. Contributed new reagents or analytic tools: Wen, Roongta. Performed data analysis: Wen, Roongta, Liu. Wrote or contributed to the writing of the manuscript: Wen, Moore. References Adams M, Mahringer A, Kunert O, Fricker G, Efferth T, and Bauer R (2007) Cytotoxicity and p-glycoprotein modulating effects of quinolones and indoloquinazolines from the Chinese herb Evodia rutaecarpa. Planta Med 73:1554–1557. Baillie TA and Davis MR (1993) Mass spectrometry in the analysis of glutathione conjugates. Biol Mass Spectrom 22:319–325. Dieckhaus CM, Fernández-Metzler CL, King R, Krolikowski PH, and Baillie TA (2005) Negative ion tandem mass spectrometry for the detection of glutathione conjugates. Chem Res Toxicol 18:630–638. Dong G, Wang S, Miao Z, Yao J, Zhang Y, Guo Z, Zhang W, and Sheng C (2012) New tricks for an old natural product: discovery of highly potent evodiamine derivatives as novel antitumor agents by systemic structure-activity relationship analysis and biological evaluations. J Med Chem 55:7593–7613. Fei XF, Wang BX, Li TJ, Tashiro S, Minami M, Xing DJ, and Ikejima T (2003) Evodiamine, a constituent of Evodiae Fructus, induces anti-proliferating effects in tumor cells. Cancer Sci 94:92–98. Huang H, Zhang Y, Liu X, Li Z, Xu W, He S, Huang Y, and Zhang H (2011) Acid sphingomyelinase contributes to evodiamine-induced apoptosis in human gastric cancer SGC-7901 cells. DNA Cell Biol 30:407–412. Jan WC, Lin LC, Don MJ, Chen CF, and Tsai TH (2006) Elimination of rutaecarpine and its metabolites in rat feces and urine measured by liquid chromatography. Biomed Chromatogr 20: 1163–1171. Jeon TW, Jin CH, Lee SK, Jun IH, Kim GH, Lee DJ, Jeong HG, Lee KB, Jahng Y, and Jeong TC (2006) Immunosuppressive effects of rutaecarpine in female BALB/c mice. Toxicol Lett 164: 155–166. Kan SF, Yu CH, Pu HF, Hsu JM, Chen MJ, and Wang PS (2007) Anti-proliferative effects of evodiamine on human prostate cancer cell lines DU145 and PC3. J Cell Biochem 101:44–56. Kang P, Dalvie D, Smith E, and Renner M (2009) Bioactivation of lumiracoxib by peroxidases and human liver microsomes: identification of multiple quinone imine intermediates and GSH adducts. Chem Res Toxicol 22:106–117.

Kassahun K, Skordos K, McIntosh I, Slaughter D, Doss GA, Baillie TA, and Yost GS (2005) Zafirlukast metabolism by cytochrome P450 3A4 produces an electrophilic a,b-unsaturated iminium species that results in the selective mechanism-based inactivation of the enzyme. Chem Res Toxicol 18:1427–1437. Lee TJ, Kim EJ, Kim S, Jung EM, Park JW, Jeong SH, Park SE, Yoo YH, and Kwon TK (2006a) Caspase-dependent and caspase-independent apoptosis induced by evodiamine in human leukemic U937 cells. Mol Cancer Ther 5:2398–2407. Lee SK, Lee DW, Jeon TW, Jin CH, Kim GH, Jun IH, Lee DJ, Kim SI, Kim DH, and Jahng Y, et al. (2005) Characterization of the Phase II metabolites of rutaecarpine in rat by liquid chromatography-electrospray ionization-tandem mass spectrometry. Xenobiotica 35:1135–1145. Lee SK, Lee J, Lee ES, Jahng Y, Kim DH, and Jeong TC (2004) Characterization of in vitro metabolites of rutaecarpine in rat liver microsomes using liquid chromatography/tandem mass spectrometry. Rapid Commun Mass Spectrom 18:1073–1080. Lee SK, Lee JH, Yoo HH, Kim DH, Jahng Y, and Jeong TC (2006b) Characterization of human liver cytochrome P450 enzymes involved in the metabolism of rutaecarpine. J Pharm Biomed Anal 41:304–309. Li Y, Slatter JG, Zhang Z, Li Y, Doss GA, Braun MP, Stearns RA, Dean DC, Baillie TA, and Tang W (2008) In vitro metabolic activation of lumiracoxib in rat and human liver preparations. Drug Metab Dispos 36:469–473. Liao CH, Pan SL, Guh JH, Chang YL, Pai HC, Lin CH, and Teng CM (2005) Antitumor mechanism of evodiamine, a constituent from Chinese herb Evodiae fructus, in human multiple-drug resistant breast cancer NCI/ADR-RES cells in vitro and in vivo. Carcinogenesis 26:968–975. Ogasawara M, Matsubara T, and Suzuki H (2001) Inhibitory effects of evodiamine on in vitro invasion and experimental lung metastasis of murine colon cancer cells. Biol Pharm Bull 24: 917–920. Poon GK, Chen Q, Teffera Y, Ngui JS, Griffin PR, Braun MP, Doss GA, Freeden C, Stearns RA, and Evans DC, et al. (2001) Bioactivation of diclofenac via benzoquinone imine intermediatesidentification of urinary mercapturic acid derivatives in rats and humans. Drug Metab Dispos 29:1608–1613. Rodrigues AD (1999) Integrated cytochrome P450 reaction phenotyping: attempting to bridge the gap between cDNA-expressed cytochromes P450 and native human liver microsomes. Biochem Pharmacol 57:465–480. Sheu JR (1999) Pharmacological effects of rutaecarpine, an alkaloid isolated from Evodia rutaecarpa. Cardiovasc Drug Rev 17:237–245. Shen S, Marchick MR, Davis MR, Doss GA, and Pohl LR (1999) Metabolic activation of diclofenac by human cytochrome P450 3A4: role of 5-hydroxydiclofenac. Chem Res Toxicol 12:214–222. Skiles GL and Yost GS (1996) Mechanistic studies on the cytochrome P450-catalyzed dehydrogenation of 3-methylindole. Chem Res Toxicol 9:291–297. Sun HZ, Fang ZZ, Cao YF, Sun XY, and Hong M (2013) Investigation of the in vitro metabolism of evodiamine: characterization of metabolites and involved cytochrome p450 isoforms. Phytother Res 27:705–712. Sun H and Yost GS (2008) Metabolic activation of a novel 3-substituted indole-containing TNF-alpha inhibitor: dehydrogenation and inactivation of CYP3A4. Chem Res Toxicol 21: 374–385. Thornton-Manning J, Appleton ML, Gonzalez FJ, and Yost GS (1996) Metabolism of 3-methylindole by vaccinia-expressed P450 enzymes: correlation of 3-methyleneindolenine formation and protein-binding. J Pharmacol Exp Ther 276:21–29. Ueng YF, Don MJ, Jan WC, Wang SY, Ho LK, and Chen CF (2006) Oxidative metabolism of the alkaloid rutaecarpine by human cytochrome P450. Drug Metab Dispos 34:821–827. Ueng YF, Yu HJ, Lee CH, Peng C, Jan WC, Ho LK, Chen CF, and Don MJ (2005) Identification of the microsomal oxidation metabolites of rutaecarpine, a main active alkaloid of the medicinal herb Evodia rutaecarpa. J Chromatogr A 1076:103–109. Wang C, Li S, and Wang MW (2010) Evodiamine-induced human melanoma A375-S2 cell death was mediated by PI3K/Akt/caspase and Fas-L/NF-kappaB signaling pathways and augmented by ubiquitin-proteasome inhibition. Toxicol In Vitro 24:898–904. Wen B, Chen Y, and Fitch WL (2009) Metabolic activation of nevirapine in human liver microsomes: dehydrogenation and inactivation of cytochrome P450 3A4. Drug Metab Dispos 37:1557–1562. Wen B, Coe KJ, Rademacher P, Fitch WL, Monshouwer M, and Nelson SD (2008a) Comparison of in vitro bioactivation of flutamide and its cyano analogue: evidence for reductive activation by human NADPH:cytochrome P450 reductase. Chem Res Toxicol 21:2393–2406. Wen B, Ma L, Nelson SD, and Zhu M (2008b) High-throughput screening and characterization of reactive metabolites using polarity switching of hybrid triple quadrupole linear ion trap mass spectrometry. Anal Chem 80:1788–1799. Xu ML, Li G, Moon DC, Lee CS, Woo MH, Lee ES, Jahng Y, Chang HW, Lee SH, and Son JK (2006) Cytotoxicity and DNA topoisomerase inhibitory activity of constituents isolated from the fruits of Evodia officinalis. Arch Pharm Res 29:541–547. Yan Z, Easterwood LM, Maher N, Torres R, Huebert N, and Yost GS (2007) Metabolism and bioactivation of 3-methylindole by human liver microsomes. Chem Res Toxicol 20: 140–148. Yang XW, Zhang H, Li M, Du LJ, Yang Z, and Xiao SY (2006) Studies on the alkaloid constituents of Evodia rutaecarpa (Juss) Benth var. bodinaieri (Dode) Huang and their acute toxicity in mice. J Asian Nat Prod Res 8:697–703. Yu H, Jin H, Gong W, Wang Z, and Liang H (2013) Pharmacological actions of multi-targetdirected evodiamine. Molecules 18:1826–1843.

Address correspondence to: Dr. Bo Wen, Drug Metabolism and Pharmacokinetics, GlaxoSmithKline, 709 Swedeland Road, King of Prussia, PA 19406. E-mail: [email protected]