In Vitro and In Vivo Metabolism and Inhibitory Activities of Vasicine, a Potent Acetylcholinesterase and Butyrylcholinesterase Inhibitor

Vasicine (VAS), a potential natural cholinesterase inhibitor, exhibited promising anticholinesterase activity in preclinical models and has been in development for treatment of Alzheimer’s disease. This study systematically investigated the in vitro and in vivo metabolism of VAS in rat using ultra performance liquid chromatography combined with electrospray ionization quadrupole time-of-flight mass spectrometry. A total of 72 metabolites were found based on a detailed analysis of their 1H- NMR and 13C NMR data. Six key metabolites were isolated from rat urine and elucidated as vasicinone, vasicinol, vasicinolone, 1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3-yl hydrogen sulfate, 9-oxo-1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3-yl hydrogen sulfate, and 1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3-β-D-glucuronide. The metabolic pathway of VAS in vivo and in vitro mainly involved monohydroxylation, dihydroxylation, trihydroxylation, oxidation, desaturation, sulfation, and glucuronidation. The main metabolic soft spots in the chemical structure of VAS were the 3-hydroxyl group and the C-9 site. All 72 metabolites were found in the urine sample, and 15, 25, 45, 18, and 11 metabolites were identified from rat feces, plasma, bile, rat liver microsomes, and rat primary hepatocyte incubations, respectively. Results indicated that renal clearance was the major excretion pathway of VAS. The acetylcholinesterase (AChE) and butyrylcholinesterase (BChE) inhibitory activities of VAS and its main metabolites were also evaluated. The results indicated that although most metabolites maintained potential inhibitory activity against AChE and BChE, but weaker than that of VAS. VAS undergoes metabolic inactivation process in vivo in respect to cholinesterase inhibitory activity.

Metabolic stability and metabolic profile are essential in designing new drugs because the metabolism of drugs can induce many pharmacokinetic (low bioavailability, high clearance, short half-life, first-pass elimination) and toxic problems (reactive metabolites) that can be regulated in the early stages of development by identifying the relationship between metabolism, metabolic toxicity, and compound structure [14,15]. Previous studies have reported that VAS could be metabolized into vasicinone (VAO), desoxyvasicine (DVAS), deoxyvasicinone (DVAO), 1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3-β-D-glucuronid (VAS-3-G), and vasicinone-3-O-glucuronide (VAO-3-G) in rats after an oral dose of 20 mg/kg [16]. However, a discrepancy result was observed in our preliminary experiment, in which DVAS and DVAO were not found among the metabolites.
Characterizing the chemical structure of metabolites is important in metabolism studies [17,18]. Determining the exact structure of metabolites is also the nodus in metabolism studies. Ultra-performance liquid chromatography combined with electrospray ionization quadrupole time-of-flight tandem mass spectrometry (UPLC/ESI-QTOF-MS) and the chemical intelligent software tool Mass Fragment have been used in the metabolic identification of xenobiotics in recent years [19][20][21][22][23]. In the present study, UPLC/ESI QTOF MS was used to perform in vivo and in vitro metabolite identification studies on VAS in rat to elucidate the real metabolism of VAS and understand its whole metabolic route. The metabolites were easily found by comparing fragmentation ions obtained from MS and chromatographic behaviors of the authentic standards. However, identification and characterization of metabolite structures remain a challenge because of the complicated mass fragmentation patterns in the pyrrolo [2,1-b] quinazoline skeleton of VAS. Moreover, certain metabolites are isomers having the same molecular formula with minimal variations presented in their MS spectra. Therefore, we systematically investigated the fragmentation pathways of pyrrolo [2,1-b] quinazoline alkaloids and found that the mass cleavages of the ring skeleton are characteristics of and related to the chemical structure. Fragmentation patterns can be used as a critical clue in investigating VAS metabolism. As a result, a total of 72 metabolites of VAS have been identified in vivo from rat bio-specimens (urine, feces, plasma, and bile) and in vitro from rat liver microsomes (RLMs) and rat primary hepatocytes (RPHs) incubation.

Chemicals and reagents
VAS and VAO were isolated from P. harmala in our laboratory according to a previously reported method [24]. VAS and VAO structures were elucidated by comparing their spectral six male rats were collected from 0 to 12 h after oral administration of 50 mg/kg VAS. Baseline rat bile was collected before oral administration of VAS. All the samples were stored at −20°C until analysis. Blood was collected from the angular vein of the last six male rats at 0.25, 0.5, 1, and 2 h after oral administration of VAS 50 mg/kg VAS. Plasma was obtained by centrifuging blood at 4000 × g for 10 min, and then stored at −20°C until use.
The urine samples used for isolating and purifying the metabolites were collected from 15 rats that were given VAS orally once a day, over 21 consecutive days, at dose of 100 mg/kg body weight. In total, approximately 5 L of urine was collected and centrifuged at 4000 × g for 15 min. The supernatant was then stored at −20°C until use. At the end of experiment, all rats were sacrificed by CO 2 asphyxiation.

Sample preparation
Up to 1 mL of urine was thoroughly mixed with same volume of acetonitrile, and then centrifuged at 15,000 × g for 10 min. The supernatant was evaporated until dry under nitrogen at 37°C. The residue of the supernatant was re-dissolved in 200 μL of 5% methanol, and then centrifuged at 15,000 × g for 10 min. The supernatant was injected into the chromatographic systems for separation and identification of metabolites.
Approximately 1 g of rat feces was exhaustively extracted with 40 mL of 75% acetonitrile by ultrasonication at room temperature. The supernatant was then evaporated to dryness by a gentle stream of nitrogen at 37°C. The residue was dissolved by 200 μL of 5% methanol and centrifuged at 15,000 × g for 10 min prior to injection.
Plasma or bile (400 μL) was mixed and vortexed with 1200 μL of acetonitrile for 30, then centrifuged at 15,000 × g for 10 min at 4°C. Afterward, the supernatant was evaporated until dry by a gentle stream of nitrogen at 37°C. The residue was dissolved by 80 μL of 5% methanol and centrifuged at 15,000 × g for 10 min. The supernatants (5 μL) were used in the UPLC/ ESI-QTOF-MS analysis.
Semi-preparative HPLC was performed with a ZORBAX SB-C18 (9.4 mm × 25 cm, 5 μm, Agilent, USA) at 30°C in a LC3000 liquid chromatography (Beijing Tong Heng Innovation Technology Co., Beijing, China). The detection wavelength was set at 280 nm. The mobile phase consisted of methanol (A) and 0.1% aqueous formic acid (B). For M16, a flow rate of 4.7 mL/min and 18% A isocratic elution were used to obtain M16 (32 mg). M20-2 (20 mg) was obtained using gradient elution: 0 to 10 min, 5% to 20% A, with a flow rate of 3 mL/min. buffer (50 mM, pH 7.4). After 5 min of pre-incubation at 37°C, the incubation reaction was initiated by the addition of glucose 6-phosphate (10 mM), glucose-6-phosphate dehydrogenase (1 unit/mL), NADPH (1.0 mM), and UDPGA (5.0 mM). The total incubation volume was 100 μL. The reactions were terminated with 300 μL ice-cold acetonitrile after 1 h of incubation. The mixture was centrifuged at 15,000 × g for 10 min, and then 320 μL of supernatant was evaporated to dryness by a gentle stream of nitrogen at 37°C. The residue was dissolved by 80 μL of 5% methanol and centrifuged at 15,000 × g for 10 min. The supernatant (5 μL) was then injected for UPLC/ESI-QTOF-MS system. Control samples without NADPH and UDPGA or substrates were prepared. The incubation was performed in duplicate.

Isolation, culture, and incubation of RPHs
Male rats (150 ± 20 g) were anesthetized with pentobarbital (100 mg/kg, ip) and their livers were perfused using the two-step collagenase digestion method described previously [27]. Cell vitality was assessed using the trypan blue method; only hepatocytes with a viability greater than 85% were seeded in a 6-well plate in M199 medium (Gibco, Laboratories Inc.; Grand Island, NY) supplemented with 5% fetal bovine serum, 10 −9 mol/L dexamethasone, and insulin. Cells were cultured for 24 h before use in metabolism studies, and then added with VAS (200 μM) thereafter. After 1, 3, 5, 12, and 24 h of incubation, 100 μL supernatant was terminated with 300 μL of ice-cold acetonitrile. The mixture was centrifuged at 15,000 × g for 10 min, and 320 μL of supernatant was evaporated until dry by a gentle stream of nitrogen at 37°C. The residue was dissolved by 80 μL of 5% methanol and centrifuged at 15,000 × g for 10 min. Up to 5 μL of the supernatant was injected into the UPLC/ESI-QTOF-MS system.

NMR instrumentation for identification of the metabolite structures
NMR spectra ( 1 H and 13 C NMR, HSQC, HMBC) were recorded on a Bruker AV 400 MHz spectrometer (400 MHz for 1 H and 100 MHz for 13 C). Chemical shifts were recorded in parts per million using tetramethylsilane as an internal standard. Deuterium reagents (DMSO-d6, CD 3 OD, D 2 O) were used as solvents for the metabolites.

In vitro anticholinesterase assays
The AChE and BChE inhibitory activities of VAS and its main metabolites, namely, VAO, VASL, VAOL, VAS-3-S, VAO-3-S, and VAS-3-G, were evaluated based on our previously established method [8]. A 50 μL incubation system composed of 10 μL of the test compound solution (all the test compound solutions were diluted to a series of concentrations with buffer solution before each experiment) and 40 μL of the enzyme solutions (with final concentrations: 0.0035 unit/mL for AChE, or 0.008 unit/mL for BChE) were mixed and pre-incubated for 15 min. Up to 50 μL of substrate solution (final concentration of 5.505 μM for ACh, or 7.152 μM for BCh) was added into the mixture, which was then incubated for 20 min at 25°C. The reaction was terminated by addition of 300 μL of ice acetonitrile solution (0°C), which was immediately dissolved with 1.899 μM IS (chlormequat). The solution was then centrifuged (15,000 × g, 10 min), and the supernatant was used for analysis by UPLC-ESI MS/MS. The inhibitive IC 50 values of VAS and its key metabolites on AChE and BChE were calculated using the Prism software (GraphPad Software Inc., San Diego, CA).

UPLC Chromatography and Quadrupole Time-of-flight (Q/TOF) Mass Spectrometer
Chromatographic analysis was performed on an HSS T3 column (100 mm × 2.1 mm, 1.8 μm particle size; Waters Corporation, Milford, MA, USA) for separation of samples using an ACQUITY UPLC system equipped with a binary solvent manager system and an autosampler. The column was eluted with a gradient mobile phase of methanol (A) and 0.1% formic acid in deionized water (B): 0-6 min 5% A; 6-12 min, linear from 5% to 15% A; 12-18 min, linear from 15% to 30% A; 18-24 min, linear from 30% to 50% A; 24-28 min, linear from 50% to 90% A; 28-30 min, 5% A for equilibration of the column. The flow rate was 0.4 mL/min. The column and sample-tray temperatures were maintained at 40°C and 10°C, respectively.
A Waters ACQUITY Synapt G2 Q/TOF tandem mass spectrometer was combined with the UPLC system through an electrospray ionization (ESI) interface and controlled by Masslynx software (version 4.1, Waters Corp., Manchester, UK). The positive ionization mode of ESI source was operated. The optimized conditions for maximum detection of metabolites were as follows: source temperature, 120°C; desolvation temperature, 400°C; capillary voltage, 3.0 kV; sample cone, 30 V and extraction cone, 4.0 V. The cone and desolvation gas (N 2 ) flows were set at 75 and 800 (L/h), respectively. The leucine-enkephalin (2 ng/mL) was used as the lock mass to generate a reference ion in positive mode at m/z 556.2771 and the injection rate was 5 μL/min for accurate mass acquisition. Data were collected in centroid mode and the MS E approach using a dynamic ramp of collision energy was carried out in two scan functions. Function 1 (low energy): mass scan ranged from 100-1000,with 0.2 s scan time and 0.05 s inter-scan delay; Function 2 (high energy): mass scan ranged from 100-1000 with 0.2 s scan time, 0.05 s inter-scan delay, and collision energy ramp of 10-40 V. MS/MS analysis were operated for major metabolites to obtain additional information from product ions since the same collision energy of MS E in Function 2 could not match each precursor ion. The conditions for the MS/ MS functions, sample cone voltages, and sample collision energies were summarized in Table 1.

Results
Fragmentation of the parent VAS and important inter-metabolites (VAO, VASL, and VAOL) Proper identification of metabolites using the on-line UPLC/QTOF-MS approach required comprehensive understanding of the fragmentation behavior of the parent compound and related compounds to be tested. Table 2 showed the elemental composition, experimental and calculated masses, and mass errors of the adduct ion and the fragment ions of VAS, VAO, VASL, and VAOL. The maximum mass errors between the measured and calculated values were less than 10 ppm ( 2.0 mDa), which indicated high resolution and good accuracy. The dynamic energy of MS 2 from protonated VAS [M+H] + at m/z 189.1030 afforded the product ion at m/z 171.0925 by the loss of water molecules (18.0105 Da, exacted 18.0106 Da). As a result, a carbon-carbon double bond between C-2 and C-3 through elimination reaction was formed (Fig 2A). Thereafter, the product ion at m/z 154.0657 (C 11 H 8 N + , calculated m/z 154.0657), 144.0812 (C 10 H 10 N + , calculated m/z 144.0813), and 143.0726 (C 10 H 9 N + , calculated m/z 143.0735) were formed through the breakdown of rings B and C, as well as the elimination of ammonia (NH 3 , 17.0268 Da, exact 17.0265 Da), CHN unit (27.0113 Da, exact 27.0109 Da), and CH 2 N radicals (28.0189 Da, exact 28.0187 Da). The fragmentation process was similar to that reported by Rashkes, who studied the fragmentation of natural and synthetic quinazoline derivatives [28]. In addition, a heptatomic and octatomic ring was produced in the fragmentation of natural and synthetic quinazoline derivatives (Fig 2A) VAO is a metabolite of VAS [2] resulting from the ketonization of VAS in C-9 site. However, the fragmentation pathways of VAO were different from VAS. ESI-MS spectral analysis of VAO revealed a protonated molecule ion at m/z 203.0824. The MS 2 spectra showed identical  Fig 2D. The fragmentation pathways of VAOL were very similar to that of VAO, which is accustomed to the successive loss of H 2 O, CO unit and the CHN unit and NH 3 . The CO unit could be used to identify the metabolites, whose parent compound undergoes ketonization. Furthermore, the special fragment ion at 146.0610 (C 9 H 8 NO + , calculated m/z 146.0606) showed a hydroxyl group of ring A, which could be used to identify the metabolites that undergoes hydroxylation in ring A.

Characterization of the metabolites in vivo and in vitro
Figs 3-8 showed the extracted and total ion chromatograms of the parent compound and their metabolites in vivo (rat urine, feces, bile, plasma) and in vitro (RLMs and RPHs). A total of 72 metabolites were observed in rat urine, and 15, 45, 25, 18, and 11 metabolites were found in vivo in rat feces, bile, and plasma and in vitro in RLMs and RPHs, respectively. The metabolites were numbered according to their molecular weight and chromatographic retention times. Table 3 listed the detailed information (including of retention times, presume formula, measured mass, calculated mass, fragment ions and presence in different samples) of the metabolites. The structures of metabolite were presumed by the comparison of mass spectra and chromatographic retention times with available reference standards, or by its mass spectral fragmentation patterns. The proposed structures of the metabolites were showed in Fig 9, which included monohydroxylation, dihydroxylation, trihydroxylation, dehydrogenation, methylation, acetylation, sulfation, and glucuronation of the parent compound.
Metabolite M1. Metabolite M1 (retention time t R = 13.90 min) is a metabolite detected in the in vivo and in vitro samples. Metabolite M1 exhibited a [M + H] + ion at m/z 203.0839, which was 14 Da (13.9807 Da) higher than that of the protonated parent compound, indicating the introduction of an oxygen atom and dehydrogenation (+ 13.9793 Da). The metabolic site for the parent compound may be at C-1, C-2, or C-9. The MS 2 spectra showed identical fragment ions at m/z 185.0725 (C 11  . Given the clearance of the methyl group during the first step, no further information could help determine the position of the methyl group; thus, the position of the methyl group was unclear. Metabolites M3. M3 was detected with a protonated molecular weight of 205.0977, which was 16 Da (15.9949 Da) higher than that of the protonated parent compound, indicating the introduction of one hydroxylation metabolite. In addition, four independent chromatographic peaks with protonated ions at m/z 205.0977 were detected.   The fragment ions at m/z 159.0683 and 134.0618 were attributed to the hydroxylation of ring A, suggesting that the +16 Da modification occurred on the aromatic moiety. The molecular formula of the metabolite was further supported by the 1 H and 13 C NMR spectral data ( Table 4). The 1 H-NMR spectra of M3-2 in CD 3 OD-d4 indicated the existence of three aromatic protons at δ 7.04 (1H, d, J = 8.7 Hz, H-5), 6.83 (1H, dd, J = 8.7, 2.5 Hz, H-6) and 6.69 (1H, d, J = 2.4 Hz, H-8), respectively. Compared with the parent compound, the spectra revealed that metabolic substitution occurred in the benzene ring (position C-6 or C-7). Meanwhile, the 1 H and 13 C NMR spectra of M3-2 was similar to that of VASL [29,30]. Thus, M3-2 was confirmed as VASL.
Metabolite    (36 Da) and another molecule of H 2 O and a neutral loss of CO unit, respectively. Our data indicated that hydroxylation occurred on ring B or ring C (position C-2, C-9). LogP value, an octanol/water partition coefficient, was used to determine the site of hydroxylation. As an important parameter for estimating the hydrophobicity of a compound, LogP also be used to describe the tendency for distribution from aqueous phase into hydrophobic phase. Compounds with a high LogP would have long retention time on reverse phase high performance liquid chromatography [31,32]. Considering that 1,2,3,9-tetrahydropyrrolo [ The key ion of m/z 173.0728 was generated by the loss of CO at ring B, which indicated that an oxygen atom underwent dehydrogenation at ring B (C-9). Meanwhile, the key ion of m/z 155.0608 and 130.0674 were generated by loss of H 2 O and a molecule from the CHN unit. These data were attributed to an oxygen atom at ring C (position C-1 or position N). The chromatographic retention time of M5-3 (16.18 min) was longer than that of the precursor compound M1 (VAO) under the acidic elution condition (0.1% formic acid watermethanol), indicating that M5-3 was an N-oxide of VAO. Thus, M5-3 was identified as 3-hydroxy-9-oxo-2, 3, 9, 10-tetrahydro-1H-pyrrolo[2,1-b] quinazoline 10-oxide.
Metabolites M6. M6 was detected with a protonated molecular weight of 219.1134, which was 30 Da (30.0106 Da, CH 2 O) higher than that of the protonated parent compound. These metabolites had a common special fragment ion 187.0825 and lost a molecule of methanol (CH 3 OH), indicating the introduction of the methoxylation metabolite. In addition, three independent chromatographic peaks with protonated ions at m/z 219.1134 were detected.
M7-2, M7-3, and M7-4 were detected at 2.62, 3.60, and 4.10 min, respectively. The MS 2 spectra showed fragment ions at m/z 203.0820 (C 11  M7-5, M7-6, and M7-7 were detected at 5.81, 10.66, and 13.57 min, respectively. These retention times were longer than that of the parent compound and all given information of monohydroxylation metabolites. Thus, M7-5, M7-6, and M7-7 were identified as the N-oxides after hydroxylation or di-N-oxides of the parent compound. M7-5 was detected at 5.81 min. The MS 2 spectra showed fragment ions at m/z 203.0848 (C 11  O, and CO unit, indicating that another hydroxyl group was combined in ring C (position C-3). Therefore, M8 was identified as 1,2,3-trihydroxy-2,3-dihydro-pyrrolo [2,1-b] quinazolin-9 (1H)-one. Metabolites M9. M9 was detected with a protonated molecular weight of 235.1083, which was 46 Da (46.0057 Da) higher than that of the protonated parent compound, indicating the introduction of two oxygen atoms and a methyl group into the molecule. In addition, two independent chromatographic peaks with protonated ions at m/z 235.1083 were detected.
M9-1 and M9-2 were detected at 13.92 and 19.56 min, respectively. The MS 2 spectra were not acquired because of the interference of endogenous or other metabolites with same retention time and near molecular weight. Therefore, no fragment information could help to identify the structures of M9-1 and M9-2.
Metabolites M10. M10 was detected with a protonated molecular weight of 237.0875, which was 48 Da (47.9747 Da) higher than that of the protonated parent compound, indicating the introduction of three oxygen atoms into the molecule. In addition, two independent chromatographic peaks with protonated ions at m/z 237.0875 were detected.
Metabolites M11. M11 was detected with a protonated molecular weight of 245.1290, which was 56 Da (56.0262 Da) higher than that of the protonated parent compound, indicating the introduction of acetyl and methyl groups into the molecule. In addition, three independent chromatographic peaks with protonated ions at m/z 245.1290 were detected.  189.1038 molecular. Other fragment ion information was as same as those of the parent compound, indicating that M15 was 1,2,3,9-tetrahydro-pyrrolo [2, 1-b] quinazolin-3-yl hydrogen sulfate.
The molecular formula of the metabolite was further supported by the 1 H and 13 C NMR spectral data (Table 4). The 1 H-NMR spectra of M15 indicated the existence of four aromatic protons and the metabolic substitution did not occur in the benzene ring. Meanwhile, the 13 C NMR spectra of M15 were similar to that of VAS (M0). Difference was observed between the 1 H-NMR spectra of M15 and VAS. Only one of the protons at δ 5.45 (H-3) shifted to downfield by 0.90 ppm; the other protons signals shifted to a downfield by 0.28−0.54 ppm, which indicated that metabolic substitution occurred in C-3. Therefore, M15 was confirmed as 1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3-yl hydrogen sulfate. The molecular formula of the metabolite was further supported by the 1 H and 13 C NMR spectral data (Table 4). Similar to M15, the 1 H-NMR spectra of M16 indicated the existence of four aromatic protons, indicating that metabolic substitution did not occur in the benzene ring. Meanwhile, the 13 C NMR spectra of M16 were similar to the VAO (M1). Differences were observed between the 1 H-NMR spectra of M16 and VAO. Only one of the protons at δ 5.73 (H-3) shifted to downfield by 0.76 ppm; the other protons signals shifted to downfield by 0.14-0.45 ppm. This finding indicated that the metabolic substitution was happened in the C-3. Therefore, M16 was confirmed as 9-oxo-1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3-yl hydrogen sulfate.
Metabolites M17. M17 was detected with a protonated molecular weight of 285.0545, which was 96 Da (95.9517 Da) higher than that of the protonated parent compound, indicating the introduction of oxygen atom and sulfonic group into the molecule. In addition, two independent chromatographic peaks with protonated ions at m/z 285.0545 were detected.
M17-1 and M17-2 were detected at 2.24 and 4.69 min, respectively. The MS 2 spectra showed fragment ions at m/z 205.0981 (C 11  Other fragment ion information was as same as that of M3-2 (m/z 205 Da). Therefore, M17-1 and M17-2 might be sulfates of M3-2, but the same fragment ion information with M3-2 might be found all of the hydroxylation occurred on ring A. Thus, M17-1 and M17-2 were identified as the sulfates of the parent compound that underwent hydroxylation at ring A. Metabolite M18. M18 was detected (t R = 10.56 min) with a protonated molecular weight of 297.0185, which was 108 Da (107.9157 Da) higher than that of the protonated parent compound. The MS 2 spectra showed identical fragment ions at m/z 217.0618 (C 11 ) indicated that M20 was a glucuronide conjugate of the parent compound. Other fragment ion information of M20 was as same as that of parent compound. These results revealed that M20 was a tentatively identified glucuronide conjugate of VAS (position C-3, N-1 or N-2).
Metabolites M21. M21 was detected with a protonated molecular weight of 379.1141, which was 190 Da (190.0113 Da) higher than that of the protonated parent compound, indicating the introduction of oxygen atom, dehydrogenation, and addition of a C 6 H 8 O 6 unit (glucuronidation) into the molecule. In addition, two independent chromatographic peaks with protonated ions at m/z 379.1141 were detected. However, no more fragment information was acquired because the peak abundance of M24 was too low or the interference was too serious. Therefore, the concrete substitution position was unclear. Metabolites M25. M25 was detected with a protonated molecular weight of 397.1247, which was 208 Da (208.0219 Da) higher than that of the protonated parent compound, indicating the introduction of two oxygen atom and addition of a C 6 H 8 O 6 unit (glucuronidation) into the molecule. In addition, two independent chromatographic peaks with protonated ions at m/ z 397.1247 were detected. M25

Anticholinesterase activity of VAS and metabolites
The inhibitory activities of VAS and its key metabolites of VAO, VASL, VAOL, VAS-3-S, VAO-3-S, and VAS-3-G against AChE and BChE were evaluated in vitro and summarized in

Discussion
According to results of bioavailability studies of VAS and metabolites in our laboratory (no published data), the absolute bioavailability of VAS after oral administration was about 50%, and the elimination half-life of VAS after intravenous and oral administration was about 5 h and 1.5 h, respectively. In the process of urine samples collection, 15 rats have been administrated with VAS orally at dose of 100 mg/kg body weight once a day for 21 consecutive days and no any observable side effects have been seen. In addition, the LD 50 value of VAS has been determined as 308.25 mg/kg (no published data). It illustrated that VAS is safe as a potential candidate drug.
In the present study, we systematically investigated the metabolites of VAS in vivo (rat urine, feces, plasma, and bile) and in vitro (RLMs and RPHs) and the cholinesterase inhibitory activities of its main metabolites. By using the analysis of UPLC/ESI-QTOF MS scan, 1 H-NMR, and 13 C NMR, a total of 72 metabolites were identified. Among them, 72 metabolites were found from the urine, 15 from the feces, 25 from the plasma, 45 from the bile, 18 from RLMs, and 11 from RPHs. These results indicated that VAS encountered a strong liver first pass effect and the liver was the momentous metabolic organ for the metabolism of VAS.
VAS undergoes extensive sequential metabolism in vivo and in vitro. Both the phases I and II metabolites were observed. The metabolic pathways of VAS were proposed in Fig 9. The formation of the phase I metabolites included hydroxylation (M3, M7, M10) and desaturation (M1, M4, M5, M8). Among the phase I metabolites, the product of monohydroxylation and desaturation (VAO) was the main metabolite. This compound could be separated from P. harmala and A. vasica. Other hydroxylation or hydroxylation and desaturation products were also extensively observed. The reaction involved the addition of one, two, or even three hydroxyl groups or/and desaturation from the parent drug. In some metabolites, two metabolic transformations simultaneously occurred. For example, M8 was formed by the desaturation reaction followed by a hydroxylation reaction. This metabolic pattern is widespread in many cases [20]. The phase II biotransformation observed as methylation, acetylation, sulfation, and glucuronidation. Among the reactions, sulfation and glucuronidation metabolites (M15, M16, M20-2 and M21-2) were most common among phase II metabolites. To obtain the reference substances of VAS metabolites, 15 rats were given continuous oral administration of VAS for 21 days and urine samples were collected within a period of 0 h to 528 h (48 h after the last dosing) for chemical separation. M15, M16, M20-2, and M21-2 were then isolated from the urine. However, because of an unavoidable lapsus, most of M20-2 and M21-2 were hydrolyzed due to urine sample being exposed at room temperature for an excessively long time at a high pH environment (pH > 10). Only M15, M16, and a small quantity M20-2 were isolated and acquired. The structures were determined as VAS-3-S, VAO-3-S, and VAS-3-G by NMR and UPLC/ ESI-QTOF-MS. It indicated that when we needed to isolate the metabolites from urine and feces, the pH value was a vital factor that must be controlled. The structure of metabolites indicated the 3-hydroxyl group and C-9 site are the main metabolic soft spots, and VAO was an important intermediate metabolite.
In a previous study, VAO, DVAS, DVAO, VAS-3-G, and VAO-3-G have been identified as metabolites of VAS in rats after a single oral dose of 20 mg/kg has been reported [16]. In our study, however, DVAS, DVAO, and their related metabolites could not be found either in vivo or in vitro. We believed that several factors could have caused the discrepancies. Firstly, VAS with impurity might be used in previous studies. We all know that the natural compounds isolated from plants could be easily mixed with some analogs. We found that the initial separation and preparation of VAS easily being resulted in some impurities (including DVAS and DVAO), when the parent compound VAS being isolated from P. harmala. In order to obtain pure VAS (not contain DVAS and DVAO), purification should be performed further. Secondly, the animal strain or individual differences could be a source of discrepancy. The metabolism of many drugs was also affected by the species, strain, sex, age, and individual differences [33]. Thirdly, the stereochemical structure of VAS would be given consideration for the discrepancy results. In addition, (-)-VAS was also isolated from A. visica and (±)-VAS was isolated from P. harmala [34]. The different of stereoselective characteristics might lead to different metabolism profiles [35].
For the 23 metabolites found in plasma, VAO, 1,2,3,9-tetrahydropyrrolo [2,1-b] quinazolin-3,9-diol (a hydroxylation metabolite of VAS, but the substituted site was not as same as VASL), VAS-3-S, VAO-3-S, VAS-3-G, and VAO-3-G were found with high abundance. VAO was reported to cause more bronchodilatory effects than parent VAS both in vitro and in vivo [36]. In addition, VAO was reported to have more potent anti-asthmatic activity than VAS comparable to that of sodium cromoglycate [37]. Meanwhile, as another important metabolic pathway, the sulfation metabolites were first reported for VAS and related compounds.
From the anti-cholinesterase assay results (Table 5), we can see that the parent compound VAS exhibited strong inhibition both on AChE and BChE activities, and that the tested metabolites exhibited inhibitory activity on both AChE and BChE, especially in the IC 50 values of VASL (2.649 ± 0.432 μM) against AChE. It was slightly stronger than the parent compound (VAS, 3.242 ± 0.079 μM). Interestingly, the IC 50 values of VAOL (46.890 ± 2.835 μM) against AChE was also stronger than VAO (76.597 ± 8.458 μM), which implied that the hydroxysubstituted benzene ring of VAS analogs could increase the inhibitory activity on AChE. However, the AChE selectivity index (SI, IC 50 of BChE/IC 50 of AChE) of VASL and VAOL were 0.930 and 1.545, which indicated that the hydroxy-substituted on benzene ring of VAS analogs obviously could reduce the selectivity for BChE inhibition. In comparison with VAS and VAS-3-S, the inhibitory activities of VAO and VAO-3-S on AChE and BChE were significantly reduced. VAO and VAO-3-S were reduced by 22.7-fold and 8.0-fold on AChE inhibitory activity and 101.0-fold and 2.8-fold on BChE inhibitory activity, respectively. However, the AChE SI of VAO and VAO-3-S were unchanged compared with VAS and VAS-3-S, which implied that the ketonization of VAS analogs might cause reduced inhibitory activity both on AChE and BChE. In addition, to be compared with VAS, the activities of VAS-3-G against AChE and BChE have reduced 6.7-fold and 1027.7-fold, respectively, with reduced selectivity for BChE inhibition. Overall, VAS has undergone inactivation in vivo in respect to anticholinesterase activity.
In conclusion, our study revealed that VAS is a metabolic unstable compound that is extensively metabolized in vivo and in vitro in rat. The 3-hydroxyl group and C-9 site are the main metabolic soft spots, and VAO is an important metabolic intermediate. The result of anticholinesterase activity assay of VAS and its main metabolites indicated that VAS undergoes metabolic inactivation process in vivo in consideration of anticholinesterase activity. Results from our study are important in understanding the metabolism and anticholinesterase activity of VAS and related quinazoline type alkaloid and provide useful information that can be used as reference for pharmacochemistry and clinical pharmacology.