Metabolism of diosbulbin B in vitro and in vivo in rats: Formation of reactive metabolites and human enzymes involved

Article abstract of DOI:10.1124/dmd.114.058222

Diosbulbin B (DB), a major constituent of the furano-norditerpenes in Dioscorea bulbifera Linn, exhibits potential antineoplasmic activity and hepatotoxicity. The metabolism and reactive metabolites of DB in vitro (with human and animal liver microsomes) and in vivo in rats were investigated. The human enzymes involved in DB metabolism were identified. DB was first catalyzed into reactive metabolites of 2-butene-1,4-dial derivatives dependent on NADPH and then trapped by Tris base or oxidized to hemiacetal lactones (M12 and M13) in microsomal incubations. Tris base was used as buffer constituent and as a trapping agent for aldehyde. Methoxylamine and glutathione (GSH) were also used as trapping agents. DB metabolism in vivo in rats after oral administration was consistent with that in vitro. The structures of M12 and M13, as well as mono-GSH conjugates of DB (M31), were confirmed by nuclear magnetic resonance spectroscopy of the chemically synthesized products. The bioactivation enzymes of DB were identified as CYP3A4/5, 2C9, and 2C19. CYP3A4 was found to be the primary enzyme using human recombinant cytochrome P450 enzymes, specific inhibitory studies, and a relative activity factor approach for pooled human liver microsomes. Michaelis-Menten constants K_mand V_maxwere determined by the formation of M31. The reactive metabolites may be related to the hepatotoxicity of DB. The gender difference in CYP3A expression in mice and rats contributed to the gender-related liver injury and pharmacokinetics in mice and rats, respectively. Copyright

Full text of DOI:10.1124/dmd.114.058222

Supplemental material to this article can be found at:
http://dmd.aspetjournals.org/content/suppl/2014/07/22/dmd.114.058222.DC1.html
1521-009X/42/10/1737–1750$25.00
DRUG METABOLISM AND DISPOSITION
http://dx.doi.org/10.1124/dmd.114.058222
Drug Metab Dispos 42:1737–1750, October 2014
Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics
Metabolism of Diosbulbin B In Vitro and In Vivo in Rats: Formation of
Reactive Metabolites and Human Enzymes Involved s
Baohua Yang, Wei Liu, Kaixian Chen, Zhengtao Wang, and Changhong Wang
Ministry of Education Key Laboratory for Standardization of Chinese Medicines, State Administration of Traditional Chinese
Medicine Key Laboratory for New Resources and Quality Evaluation of Chinese Medicines, Institute of Chinese Materia Medica,
Shanghai University of Traditional Chinese Medicine, Shanghai, China (B.Y., W.L., K.C., Z.W., C.W.); and Shanghai R&D Centre for
Standardization of Chinese Medicines, Shanghai, China (K.C., Z.W., C.W.)
Received March 25, 2014; accepted July 21, 2014
ABSTRACT
Diosbulbin B (DB), a major constituent of the furano-norditerpenes in vitro. The structures of M12 and M13, as well as mono-GSH con-
in Dioscorea bulbifera Linn, exhibits potential antineoplasmic ac-
jugates of DB (M31), were confirmed by nuclear magnetic resonance
tivity and hepatotoxicity. The metabolism and reactive metabolites spectroscopy of the chemically synthesized products. The bioactiva-
of DB in vitro (with human and animal liver microsomes) and in vivo
in rats were investigated. The human enzymes involved in DB
tion enzymes of DB were identified as CYP3A4/5, 2C9, and 2C19.
CYP3A4 was found to be the primary enzyme using human recom-
metabolism were identified. DB was first catalyzed into reactive binant cytochrome P450 enzymes, specific inhibitory studies, and
metabolites of 2-butene-1,4-dial derivatives dependent on NADPH a relative activity factor approach for pooled human liver microsomes.
and then trapped by Tris base or oxidized to hemiacetal lactones (M12 Michaelis-Menten constants Km and Vmax were determined by the
and M13) in microsomal incubations. Tris base was used as buffer
constituent and as a trapping agent for aldehyde. Methoxylamine and
formation of M31. The reactive metabolites may be related to the
hepatotoxicity of DB. The gender difference in CYP3A expression in
glutathione (GSH) were also used as trapping agents. DB metab- mice and rats contributed to the gender-related liver injury and
olism in vivo in rats after oral administration was consistent with that
pharmacokinetics in mice and rats, respectively.
Introduction
Xenobiotics are usually converted to chemicals that can be more readily
eliminated from the body by metabolic biotransformation once they enter
the body. Xenobiotic metabolites may be classified into the following broad
classes according to their pharmacological properties: 1) inactive metab-
olites, which possess no pharmacological activity; 2) active metabolites,
which may have pharmacological properties of lower, equal, or greater
magnitude than their parent compounds; and 3) reactive metabolites that
can covalently react with and alter the functional macromolecular structure
of endogenous targets in vivo, resulting in unwanted toxic effects (Njuguna
et al., 2012). For example, teucrin A, a neo-clerodane diterpene in germander
(Teucrium chamaedrys), is a structural analog of DB. Oral administration
of teucrin A or extract from germander can induce significant hepatic
necrosis in mice, which has been attributed to the metabolic activation
of a furan moiety to 1,4-enedials reactive metabolites (Druckova and
Diosbulbin B (DB) is a major constituent of the furano-norditerpenes
in Dioscorea bulbifera Linn, which has been used in traditional Chinese
medicine as a remedy for sore throat, struma, and tumor (Zhang et al.,
2008). Severe liver injuries, including liver failures, and even death
have been reported in patients who took D. bulbifera for more than
eight consecutive days (Niu and Chen, 1994). Many diosbulbins, which
include diosbulbin A–H (Kawasaki et al., 1968; Ida et al., 1978),
diosbulbin I–J (Wang et al., 2009), and diosbulbin K–M, have been
separated and identified from D. bulbifera (Liu et al., 2010). DB has
been confirmed as one of the main hepatotoxic constituents in
D. bulbifera. This compound also causes gender-related liver injury
in mice (Wang et al., 2011) as well as gender-related pharmaco-
kinetic behavior in rats (Yang et al., 2013). DB also exhibits
remarkable antitumor activity in S180 cell–bearing mice (Komori, Marnett, 2006).
1997; Wang et al., 2012), but no activity in a tumor cell line in
Available information on DB metabolism is insufficient. Thus, in the
vitro (Komori, 1997). These data imply that DB metabolism plays present study, DB metabolism in vitro was investigated using liver
an important role in the pharmacological activity and hepatotoxicity of
DB in vivo.
microsomes from humans and animals, such as rats, mice, rabbits, and
guinea pigs, using ultra-performance liquid chromatography–tandem mass
spectrometry (UPLC-MS/MS). DB metabolites in urine, feces, bile,
and plasma of rats after oral administration of DB were also identified.
Glutathione (GSH) and methoxylamine were used as trapping agents
in the incubation in vitro once the reactive metabolites were dis-
tinguished by the formation of Tris base conjugates of DB. To confirm
This work was supported by the Doctoral Fund of Ministry of Education of
China [20133107110012] (awarded to C.W.).
dx.doi.org/10.1124/dmd.114.058222.
s This article has supplemental material available at dmd.aspetjournals.org.
ABBREVIATIONS: AQ4N, 1,4-bis-{[2-(dimethylamino-N-oxide)ethyl]amino}-5,8-dihydroxy-anthracene-9,10-dione; DB, diosbulbin B; DMDO,
dimethyldioxirane; DMSO, dimethylsulfoxide; GSH, glutathione; HLM, human liver microsomes; HMBC, heteronuclear multiple-bond correlation;
HPLC, high-performance liquid chromatography; IS, internal standard; L-739,010, (1S,5R)-3-cyano-1-(3-furyl)-6-{6-[3-(3a-hydroxy-6,8-dioxabicyclo
[3.2.1]octanyl)]pyridine-2-yl-methoxyl}naphthalene; MGCD, mono-GSH conjugates of DB; P450, cytochrome P450; Q-TOF, quadrupole–time-of-flight;
RAF, relative activity factor; RLM, rat liver microsomes; UPLC-MS/MS, ultra-performance liquid chromatography–tandem mass spectrometry.
1737

1738
Yang et al.
from Fisher Scientific (Fair Lawn, NJ). NMR solvents used were 99.8%
ampoule dimethylsulfoxide (DMSO)-d6 or pyridine-d5 (Cambridge Isotope
Laboratories Inc., Tewksbury, MA). All other chemicals used were of analytical
grade.
Animals. Male Sprague-Dawley rats (weighing 200–250 g) were purchased from
Shanghai Slac Laboratory Animal Co. Ltd. (Shanghai, China). All experiments
performed in rats were in accordance with the People’s Republic of China
legislation on the use and care of laboratory animals and were approved by the
Experimental Animal Ethical Committee, Shanghai University of Traditional
Chinese Medicine.
the structures of the reactive metabolites, mono-GSH conjugates
of DB (MGCD) were chemically synthesized, purified by semipre-
parative high-performance liquid chromatography (HPLC), and iden-
tified by NMR spectroscopy. The bioactivation enzymes of DB were
identified using human recombinant cytochrome P450 (P450) en-
zymes. Michaelis-Menten constants Km and Vmax were estimated by
the formation of MGCD. Moreover, the importance of P450 isoforms
for the bioactivation of DB was determined using specific inhibitory
studies and relative activity factor (RAF) approach for pooled
human liver microsomes (HLM). This study could contribute to the
understanding of the pharmacological activity and hepatotoxicity of
DB.
Preparation of Liver Microsomes
Liver microsomes were prepared from Sprague-Dawley rats, mice, rab-
bits, and guinea pigs according to the reported method (Lu and Levin,
1972). The method used uninduced animals as well as animals that were
pretreated with phenobarbital (50 mg/kg i.p. for 3 days). Protein concen-
trations were determined according to the method described by Bradford
(1976).
Materials and Methods
Chemicals and Reagents
DB (.98% purity) was purchased from Shanghai Forever Biologic
Technology Co., Ltd. (Shanghai, China). Glucose 6-phosphate, glucose-6-
phosphate dehydrogenase, NADPH, Tris base, reduced GSH, methoxylamine,
sulfaphenazole, and (S)-(+)-N-3-benzylnirvanol were from Sigma-Aldrich
(St. Louis, MO). Human enzymes CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C18, 2C19,
2D6, 2E1, 3A4, 3A5, and human P450-reductase coexpressed in Escherichia
coli, supplemented with purified cytochrome b5, were from Shanghai Ruide
Institute of Liver Disease (Shanghai, China). Pooled HLM of 32 humans (with
equal numbers of males and females) were from BD Biosciences (Woburn, MA).
Ketoconazole and jatrorrhizine hydrochloride [internal standard (IS)] were pur-
chased from the National Institute for Food and Drug Control (Beijing, China).
Ultrapure water (.18 mV) was obtained from a Milli-Q water purification
system (Millipore, Billerica, MA). HPLC-grade acetonitrile was purchased
Microsomal Incubations
The incubation system (100 ml) contained DB (100 mM), microsomal pro-
teins (1 mg/ml), and a NADPH-regenerating system that included NADPH
(1 mM), glucose 6-phosphate (10 mM), and glucose-6-phosphate dehydrogenase
(1 unit/ml) in buffer A (50 mM Tris-HCl, pH 7.4, containing 4 mM MgCl2) or
buffer B (100 mM phosphate buffer, pH 7.4, containing 4 mM MgCl2). DB was
dissolved in acetonitrile, and the final concentration of acetonitrile in the in-
cubation system was ,1%. The system was preincubated for 5 minutes at 37°C
and initiated by adding NADPH. Control incubations were conducted without
DB or NADPH. Incubations were stopped after 1 hour by the addition of 500 ml
Fig. 1. The typical total ion current (TIC) (A) and extract ion spectra of m/z 464 (B), m/z 360 (C), m/z 446 (D), m/z 378 (E), and m/z 377 (F) from microsomal metabolites of
diosbulbin B in Tris-HCl buffer (M1 to M11) and phosphate buffer (M12 and M13), using an Agilent 6410 Triple Quad LC-MS/MS system with eluting program A.

Reactive Metabolites of Diosbulbin B and Human Enzymes Involved
1739
Fig. 2. Metabolic profile of diosbulbin B in vitro in Tris-HCl buffer (M1 to M11) and phosphate buffer (M12 and M13).
of ice-cold acetonitrile and then vortex-mixed for 1 minute. Denatured proteins UPLC-MS/MS system for analysis (see UPLC-MS/MS Analysis to Screen
were separated by centrifugation at 16,000g for 10 minutes. The supernatant
(540 ml) was evaporated to dryness under a nitrogen stream, reconstituted
using 90 ml of 10% acetonitrile, vortex-mixed for 1 minute, and centrifuged
at 16,000g at 4°C for 10 minutes. The supernatant (5 ml) was injected into the
Metabolites In Vitro and In Vivo (Analysis A)).
Trapping experiments were conducted in the presence of nucleophiles,
including GSH and methoxylamine (each at 10 mM), and the samples were
prepared by the aforementioned procedure.

1740
Yang et al.
Fig. 3. Proposed mechanisms for the formation of Tris base conjugates with the enedial derivatives of DB (M1 to M3).
Metabolism In Vivo in Rats
and cannulated via the bile duct. DB was suspended in 0.5% methylcellulose and
injected into the stomach in a single dose of 75 mg/kg. Drug-free bile was
collected for 30 minutes before DB administration. Bile was then collected from
0 to 4 hours and 4 to 8 hours after administration. Bile samples were prepared
similar to the plasma and urine samples as described earlier.
Investigation of Metabolites in Plasma of Rats. Male rats were fasted for
12 hours and had free access to water prior to the experiments. Blank blood
samples were drawn from the retro-orbital plexus before administration. Then
DB formulated in 0.5% methylcellulose suspension was orally administrated by
gavage to rats in single doses of 32, 75, and 150 mg/kg. Blood samples (;300 ml)
were drawn at 0.5, 1, 2, 4, 8, 12, and 24 hours after DB administration. Plasma was
generated by centrifugation at 10,000g for 5 minutes. Acetonitrile (5 ml) was added
to the combined plasma (1 ml) to precipitate the protein, then vortex-mixed for
1 minute and centrifuged at 16,000g for 10 minutes. The supernatant was evaporated
to dryness under nitrogen stream, reconstituted using 200 ml of 10% acetonitrile,
vortex-mixed for 1 minute, and centrifuged at 16,000g at 4°C for 10 minutes. The
supernatant (5 ml) was injected into the UPLC-MS/MS system for analysis (see
UPLC-MS/MS Analysis to Screen Metabolites In Vitro and In Vivo (Analysis A)).
Investigation of Metabolites in Urine and Feces of Rats. Rats were treated
as described earlier. Blank urine, feces before dosing, and urine and feces from
0 to 24 hours after dosing were collected and stored at 280°C until analysis.
Urine (1 ml) was processed using the procedure used to process the plasma.
Feces were moistened with little water and then homogenized. All homogenated
feces were extracted using 6-fold acetonitrile by ultrasonic wave for 1 hour and
then centrifuged (Yang et al., 2013). The supernatant (100 ml) was diluted with
100 ml of water, and a 5-ml aliquot of the diluted solution was injected into the
UPLC-MS/MS system for analysis (see UPLC-MS/MS Analysis to Screen
Metabolites In Vitro and In Vivo (Analysis A)).
UPLC-MS/MS Analysis to Screen Metabolites In Vitro and In Vivo
(Analysis A). Samples from incubation and biologic specimens (plasma, bile, urine,
and feces) were analyzed on an Agilent 6410 Triple Quad liquid chromatography–
tandem mass spectrometry system (Agilent Technologies Inc., Santa Clara, CA)
or a Waters ACQUITY UPLC-Micromass quadrupole–time-of-flight (Q-TOF)
system (Waters, Milford, MA) using an Acquity UPLC BEH C18 column (50 ꢀ
2.1 mm, 1.7 mm; Waters) at 40°C. The analytes were separated by a gradient
mobile phase consisting of acetonitrile (A) and an aqueous solution of 0.1%
formic acid (B) at a flow rate of 0.2 ml/min. The gradient elution program was as
follows: 0–15.0 minutes, 5–25% A; 15.0–16.0 minutes, 15–80% A; 16.0–17.0
minutes, 80–5% A (program A) or 0–5.0 minutes, 10–20% A; 5.0–12.0 minutes,
20–27% A; 12.0–15.0 minutes, 27–95%; 15.0–17.0 minutes, 95–5% A (program
B). The temperature of the autosampler was maintained at 3–6°C. MS was
operated in positive mode under the following operating Triple Quad parameters:
gas temperature, 325°C; gas flow, 10 l/min; nebulizer, 35 psi; capillary voltage,
4.0 kV; fragmentor, 130 V; cell accelerator, 3 V; and dwell, 200 ms. For Q-TOF,
the parameters were as follows: capillary voltage, 3.0 kV; sampling cones, 20 V;
extraction cone, 4.0 V; source temperature, 120°C; desolvation temperature, 350°C;
cone gas, 50 L/h; and desolvation gas, 600 L/h.
Elemental composition calculations for the exact masses were performed off-
Investigation of Biliary Metabolites in Rats. Rats were terminally line using MassLynx V4.1 (Waters), which was used to apply the spectrum that
anesthetized with urethane (1.4 g/ml in isotonic saline; 1.0 ml/kg intraperitoneally) was manually generated for every peak. Potential assignments were calculated
Fig. 4. Structural formula of M3, including its fragmentation, as shown by its mass spectrum using an Agilent 6410 Triple Quad LC-MS/MS system.

Reactive Metabolites of Diosbulbin B and Human Enzymes Involved
1741
TABLE 1
Mass spectral data for the metabolites and diosbulbin B
Compound
m/z
MS2
HRMS (Measured)
HRMS (Calculated)
Chemical Formula
Error
ppm
23.9
23.9a
1.9a
+
+
Diosbulbin B
M1–M3
M4–M6
362.2 (M+NH4
464.2 (M+H+)
360.1 (M+H+)
446.2 (M+H+)
378.2 (M+NH4
377.2 (M+H+)
360.2 (M+H+)
390.2 (M+H+)
390.2 (M+H+)
392.2 (M+H+)
419.2 (M+H+)
632.2 (M+H+)
650.2 (M+H+)
939.3 (M+H+)
)
)
317.2, 299.3, 271.4, 253.3
446.3, 428.3, 418.0, 400.1, 372.3, 360.3, 332.1
332.0, 314.2, 286.0,
362.1618
362.1604
464.1921
360.1447
446.1815
378.1553
377.1236
360.1447
390.1553
390.1553
392.1709
419.1818
632.1914
650.2020
939.2752
C19H20O6
464.1927-464.1939
360.1438-360.1440
446.1790-446.1797
378.1553, 378.1537
377.1232, 377.1239
360.1425-360.1525
390.1562, 390.1534
390.1561, 390.1544
392.1718, 392.1737
419.1820-419.1838
632.1899-632.1933
650.1993-650.2010
939.2762
C23H29NO9
C19H21NO6
C23H27NO8
C19H20O7
M7–M9
5.6a
M10–M11
M12–M13
M14–M16
M17–M18
M19–M20
M21–M22
M23–M26
M27–M31
M32–M34
M35
333.3, 315.1, 287.2
359.1, 331.1, 303.1, 274.9, 257.1, 228.9
332.2, 314.2
362.2, 344.2
362.2, 330.2, 312.1
4.2a
C19H20O8
1.1a
C19H21NO6
C20H23NO7
C20H23NO7
C20H25NO7
C21H26N2O7
C29H33N3O11
C29H35N3O12
C39H50O17N6S2
6.1a
22.3a
2.3a
374.2, 343.2
27.1a
24.8a
23.5a
1.5a
388.2, 387.2, 359.2, 349.1, 341.2, 309.1
557.2, 335.6 (M+K++H+)
632.2, 575.2
S
S
864.2, 789.2, 761.2, 470.1 (M+2H+)
21.1
HRMS, High Resolution Mass Spectrometry.
aThe absolute value of the error is the biggest.
using monoisotopic masses with a tolerance of 10 ppm deviation. The number DB in phosphate buffer solutions. The concentrations of GSH and the NADPH-
and types of expected atoms were set as follows: carbons # 60, hydrogens
# 120, oxygens # 30, nitrogens # 30, and sulfurs # 5.
regenerating system were similar to those in microsomal incubations. The in-
cubations (100 ml), which were conducted in triplicate, were stopped by the addition
of 500 ml of cold acetonitrile containing jatrorrhizine hydrochloride (IS) following
the steps of the microsomal incubations. Microsomes from the same cell line that
contain only the vector were used as control. The formation rate of metabolite M31,
Chemical Synthesis of Main Metabolites (GSH Conjugates of DB)
Dimethyldioxirane. Dimethyldioxirane (DMDO) was prepared according one of the main MGCDs, was determined by UPLC-MS/MS (see UPLC-MS/MS
to the reported method (Murray and Singh, 1997) with a slightly modified
generator apparatus in which another Dewar condenser was connected to the
receiving flask instead of the air condenser. In brief, peroxymonosulfate (oxone,
50 g, 0.0813 mol) was added in portions to a stirred mixture of water (20 ml), acetone
Analysis for MGCD (M31) Formation (Analysis B)).
Incubations of DB with Pooled HLM and Specific Chemical Inhibitors
To estimate the importance of P450 enzymes responsible for the bioactivation
(12 ml, 0.163 mol), and sodium bicarbonate (24 g) under nitrogen atmosphere. An of DB, DB (60 mM) was incubated with pooled HLM (0.2 mg/ml) in the
acetone-water (14 ml, 20ml) mixture was simultaneously added dropwise. The
reaction proceeded for 15 minutes, and a slight vacuum was applied to the reaction
assembly. The distillate was collected for 20 minutes in the receiving flask
containing Na2SO4 and cooled in a dry ice/acetone bath. The receiving flask was
also connected to a trap with potassium iodide solution. The distillate solution
(3–4 ml), which typically affords DMDO concentrations of 0.02–0.04 M, was
transferred to a dry ice bath and used immediately for the next step.
Enedial Derivative of DB. DB (10 mg, 0.029 mmol) was dissolved in
2 ml of acetone and precooled in a dry ice bath for 1 minute. Then 1.2 ml
of DMDO was added immediately. The reaction mixture was allowed to
warm up to room temperature and left to stand for 2 hours. A 200-ml
aliquot of the reaction mixture was evaporated to dryness under nitrogen
stream and reconstituted using 200 ml of 10% acetonitrile for UPLC-MS/MS
analysis (see UPLC-MS/MS Analysis to Screen Metabolites In Vitro and In Vivo
(Analysis A)) to confirm the disappearance of DB. The residual DMDO was
bubbled off in the stream of nitrogen gas. The total volume was reduced to
0.5 ml and used immediately for the next step.
Reaction of Enedial Derivative of DB with GSH. GSH (17.8 mg, 0.058
mmol) was dissolved in 2 ml of phosphate buffer (pH 7.4), and a solution of
enedial derivative of DB (0.5 ml) was added dropwise while shaking. The
resulting solution was left to stand at room temperature for 15 minutes. A 10-ml
aliquot of the solution was diluted with 1 ml of water and then subjected to the
UPLC-MS/MS system (see UPLC-MS/MS Analysis to Screen Metabolites In
Vitro and In Vivo (Analysis A)) to confirm the production of GSH adducts of
DB. The final solution was stored at 280°C until the following separation.
Semipreparative HPLC Separation of Synthetic Products. The synthetic
products were separated by a reverse phase–HPLC system using a DIKMA
Spursil C18 column (250 ꢀ 10 mm, 5 mm; Dikma Technologies, Lake Forest,
CA) at 25°C, with mobile phase consisting of 25% acetonitrile and 0.1% formic
acid at the following gradient flow rates: from 0 to 15 minutes, 1.8 ml/min; and
from 15 to 35 minutes, 1.6 ml/min. The wavelengths were set at 210 and 254 nm.
Fig. 5. The typical total ion current (TIC) (A) and extract ion spectra of m/z 360 (B),
m/z 390 (C), m/z 392 (D), and m/z 419 (E) from microsomal metabolites of diosbulbin
B trapped by methoxylamine in Tris-HCl buffer, using a Waters ACQUITY UPLC-
Micromass Q-TOF system with eluting program A.
Incubations of DB with Recombinant Human P450 Enzymes
Recombinant human liver P450 isoenzymes of 100 pmol/ml (CYP1A2, 2A6,
2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 3A4, and 3A5) were incubated with 50 mM

1742
Yang et al.
presence of a NADPH-regenerating system, GSH (10 mM), and chemical 60, 80, 100, 120, 160, 200, and 300 mM; CYP3A4, 10 pmol; CYP3A5,
inhibitors at 37°C for 1 hour. The chemical inhibitors used were as follows:
sulfaphenazole (10 mM) for CYP2C9, (S)-(+)-N-3-benzylnirvanol (1 mM) for
CYP2C19 (Walsky and Obach, 2003), and ketoconazole (1 mM) for CYP3A4/5
(Venkatakrishnan et al., 2001). Control incubations were conducted with vehicles
but without chemical inhibitors. (S)-(+)-N-3-Benzylnirvanol was dissolved in
DMSO, sulfaphenazole in water, and ketoconazole in methanol. Organic solvents
in the incubations were controlled within 1% of the incubation volume. Incubations
in triplicate were processed as described for incubations with recombinant human
P450 enzymes. Inhibition was calculated as the M31 formation rate in the inhibited
(vinhibited) samples (n = 3) relative to the control (vcontrol) samples (n = 3):
15 pmol; CYP2C9, 25 pmol; CYP2C19, 15 pmol; and pooled HLM, 0.2 mg/ml.
Incubation solutions (100 ml), which were conducted in triplicate, were
transferred to an ice bath for 5 minutes after 1 hour at 37°C, followed by
the sequential addition of 10 ml of acetonitrile containing jatrorrhizine
hydrochloride (IS), 10 ml of DMSO, and 480 ml of acetonitrile. The
following processing steps were similar to those of the microsomal in-
cubations, except that the residual was reconstituted with 90 ml of
aqueous solution containing 10% DMSO and 10% acetonitrile. The
samples were analyzed by UPLC-MS/MS (see UPLC-MS/MS Analysis
for MGCD (M31) Formation (Analysis B)). The enzyme kinetic in-
vestigations were performed within the range of linearity with respect to
concentrations of P450 enzymes and DB, which were determined in
preliminary experiments. Parent compound conversions were within 5%
at all DB concentrations. Apparent Km and Vmax were estimated by non-
inhibition ¼ ð1–vinhibited=vcontrolÞ ꢀ 100:
Enzyme Kinetics Studies on the Formation of MGCD (M31)
Incubations used to derive kinetic constants by the formation of M31
had the following DB and P450/HLM concentrations: DB, 5, 10, 20, 40, linear regression using GraphPad Prism 5.0 (GraphPad Software Inc., San
Fig. 6. Metabolic profile of diosbulbin B in vitro
trapped by methoxylamine in Tris-HCl buffer.

Reactive Metabolites of Diosbulbin B and Human Enzymes Involved
1743
Diego, CA). The data were fitted to the one-enzyme Michaelis-Menten
equation:
formed via reactive metabolites (Fig. 2) with a proposed mechanism
(Fig. 3) similar to the bioactivation of teucrin A (Druckova and Marnett,
2006). The furan moiety of DB was catalyzed by P450 to form reactive
metabolites, 2-butene-1,4-dial derivatives of DB, and one of the
aldehyde groups of the reactive metabolites reacted with the amine
group of Tris base to form the Schiff base. The Schiff base then
reacted with the other aldehyde group of the reactive metabolite via
addition reaction, and intramolecular rearrangement occurred to form
substituted pyrrol derivatives. Signals of fragments with loss of H2O,
CO, or C (CH2OH)3 were observed (Fig. 4; Table 1), supporting the
deduced structures of M1 to M3. According to the synthesis of the
conjugates of teucrin A with N-acetyl lysine (Druckova and Marnett,
2006), the configuration of C12 in DB could be changed in the me-
tabolites. The 3-substituted pyrroline-2-one could be the major me-
tabolites (M3), similar to the main conjugates of teucrin A with
N-acetyl lysine. M1 to M3 might be composed of four compounds
which have two structures, 3-substituted pyrroline-2-one and 4-substituted
pyrroline-2-one, with different configurations in C12 (Fig. 2). However,
it is difficult to distinguish each one from mass data except the major
metabolites (M3).
V ¼ Vmax ꢀ S=ðKm þ SÞ;
or the Michaelis-Menten model with uncompetitive substrate inhibition
(Venkatakrishnan et al., 2001):
ꢀÀ
ꢀ
Á
V ¼ Vmax ꢀ S Km þ S þ S2 Ks
:
Eadie-Hofstee transformations were used when multiple enzymes were in-
volved or Hill equation was more suitable.
Prediction of Relative Contributions for P450 Enzymes in HLM to MGCD
(M31) Formation
The relative contribution of P450 isoforms in HLM to the formation of M31
was predicted using RAFs (Crespi and Miller, 1999) calculated as
ꢀ
fið%Þ ¼ RAFiviðsÞ +RAFiviðsÞ ꢀ 100;
where fi is the relative contribution of a specific P450 isoform, RAFi is the
relative activity factor, and vi is the reaction velocity of the P450 isoform at a
specific DB concentration. vi was calculated by the Michaelis-Menten equation
using the estimated Vmax and Km. RAFi was calculated as the rate for a probe
substrate in HLM divided by the rate for a probe substrate with cDNA expressed.
The probe substrates were testosterone, tolbutamide, and (S)-mephenytoin for
CYP3A4/5, CYP2C9, and CYP2C19, respectively.
M4 to M6. Metabolites M4 to M6 exhibited a signal at m/z 360.1
(M + H+), 15 Da higher than DB in molecular weight. Signals of
fragments m/z 332.0 (M + H+-CO), m/z 314.2 (M + H+-CO-H2O), and
m/z 286.0 (M + H+-CO-H2O-CO) were observed, which demonstrated
similar DB fragmentation. From their chemical formula (C19H21NO6),
metabolites M4 to M6 were deduced as the N-dealkylated products of
M1 to M3 with a loss of C (CH2OH)3 (Fig. 2).
UPLC-MS/MS Analysis for MGCD (M31) Formation (Analysis B)
Chemically synthesized M31 was used as a standard substance. The UPLC-
MS/MS method was fully validated according to the Bioanalytical Method
Validation Guidance for Industry enacted by the Food and Drug Administration
(http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInfor-
mation/Guidances/UCM368107.pdf ). Analysis was performed on an Agilent
6410 Triple Quad LC-MS/MS system (Agilent Technologies Inc.) using an
Acquity UPLC BEH C18 column (50 ꢀ 2.1 mm, 1.7 mM; Waters) at 40°C.
The analytes were separated by a gradient mobile phase consisting of
acetonitrile (A) and an aqueous solution of 0.1% formic acid and 5 mM
ammonium acetate (B) at a flow rate of 0.3 ml/min. The gradient elution
program was as follows: 0–4.0 minutes, 10–16.4% A; 4.0–6.0 minutes, 16.4–
40% A; 6.0–8.0 minutes, 40–80% A; and 8.0–10.0 minutes, 80–10% A. The
temperature of the autosampler was maintained at 4°C. The injection volume was
5 ml. The tandem mass spectrometry system was operated in positive mode under
the following operating parameters: gas temperature, 325°C; gas flow, 10 l/min;
nebulizer, 35 psi; capillary voltage, 4.0 kV; fragmentor, 190 V for M31 and 170 V
for IS; cell accelerator, 3 V for M31 and 1V for IS; and dwell, 200 ms.
Quantification transitions for multiple reaction mode were set at m/z 632.1→m/z
557.2 for M31 and m/z 230.1→m/z 120.9 for IS, with collision energy set to 20 V
for both substances. The data were processed using the Agilent MassHunter
Qualitative Analysis B.04.00 Software (Agilent Technologies Inc.). Calibration
standards with seven levels (0.083, 0.166, 0.332, 0.83, 1.66, 3.32, and 6.64 mM)
were prepared daily by adding stock solution of M31 (86 mM) by serial dilutions.
Calibration standards were analyzed in triplicate, and quality control samples at
three levels (0.083, 0.332, and 3.32 mM) were run daily and checked to confirm
that they were at nominal value.
M7 to M9. Metabolites M7 to M9 were observed at m/z 446.2 (M + H+),
18 Da lower than that of M1 to M3, suggesting that an intramolecular
reaction might occur when Tris base is conjugated to the reactive me-
tabolites of DB (Fig. 2). This reaction is probably similar to the formation
of M27 to M31 (see M27 to M31). However, no fragment signals were
observed, which resulted in the deduction with slight uncertainty.
M10 and M11. Metabolites M10 and M11 showed signals at m/z
378.2 (M + NH4+), with a similar parent ion of DB (m/z 362.2, M +
NH4+) and similar DB fragmentation patterns (Table 1). Thus, M10 to
M11 were identified as the mono-oxydates of DB.
Results
Identification of Metabolites in Microsomal Incubations
Thirteen metabolites dependent on the NADPH-regenerating
system were detected in the microsomal incubations (Fig. 1).
These metabolites included 11 metabolites in buffer A and two in
buffer B.
M1 to M3. Metabolites M1 to M3 were observed at m/z 464.2
(M + H+), with a chemical formula of C23H29NO9 by Q-TOF mass, which
indicated that one molecule of Tris base (C4H11NO3) might be con-
Fig. 7. The typical total ion current (TIC) (A) and extract ion spectra of m/z 632 (B),
m/z 650 (C), and m/z 939 (D) from microsomal metabolites of diosbulbin B trapped
by GSH, using a Waters ACQUITY UPLC-Micromass Q-TOF system with eluting
jugated to DB. M1 to M3 were identified as Tris base conjugates of DB program B.

1744
Yang et al.
M12 and M13. Microsomal incubations of DB in buffer B (phos- [(1S,5R)-3-cyano-1-(3-furyl)-6-{6-[3-(3a-hydroxy-6,8-dioxabicyclo[3.2.1]
phate buffer) generated two other products, M12 and M13 (m/z 377.2, octanyl)]pyridine-2-yl-methoxyl}naphthalene] (Zhang et al., 1996),
M + H+), in addition to M10 and M11, with the disappearance of M19, and M20 were deduced as a 1-O-methyloxime-2-butene
M1 to M9 when buffer A (Tris buffer) was replaced. M12 and M13 4-aldehyde derivative of DB (Fig. 6), and the double bond between
were also involved in the formation of reactive metabolite and were carbon and nitrogen in the structures could turn into a triple bond after
identified as hemiacetal lactones (see Identification of the Synthesized losing CH3OH. Their reduction products, M21 and M22, also sup-
Products). The 2-butene-1,4-dial moiety of the reactive metabolite ported the deduced structures. M19 and M20 could transform into
turned into hemiacetal and was oxidized to hemiacetal lactones (Fig. 2). M17 and M18 by intramolecular reaction.
M1 to M13 were all observed in the incubations with liver microsomes
of rabbits and guinea pigs, and M11 was absent in the microsomal than M19 and M20 and were eluted earlier than M19 and M20,
incubations of rats, mice, and humans. suggesting they were reduction products of M19 and M20. Thus, M21
M21 and M22. Metabolites M21 and M22 had an m/z 2 Da higher
Trapping experiments using methoxylamine in buffer A with rat and M22 were deduced as 1-O-methyloxime-2-butene 4-alcohol deri-
liver microsomes (RLM) resulted in a series of metabolites (M14 to vatives of DB.
M26; Fig. 5) along with the disappearance of M1 to M9.
M23 to M26. Metabolites M23 to M26 were observed at m/z 419.2
M14 to M16. Metabolites M14 to M16 were observed at m/z 360.2 (M + H+), 29 Da higher than that of M19 and M20. The chemical
(M + H+), with chemical formulae (C19H21NO6) and fragments (Table 1) composition (C21H26N2O7) of M23 to M26 indicated that two mol-
similar to M4 to M6, as well as similar DB relative retention times. M14 ecules of methoxylamine were conjugated to DB. Fragmentation with
to M16 were identified as N-dealkylated products of conjugates of loss of one molecule of CH3OH (32 Da) from m/z 419.2 to m/z 387.2
methoxylamine and DB (Fig. 6), consistent with M4 to M6.
suggested that M23 to M26 might have structures similar to M19
M17 to M20. Metabolites M17 to M18 and M19 to M20 had the and M20. M23 to M26 were identified as the bis-O-methyloxime deriv-
same parent ion signals at m/z 390.2 (M + H+), with different fragment atives of DB, supported by their fragmentation with loss of another
signals. Fragments with loss of CH3OH (32 Da) were found in the MS molecule of CH3OH (32 Da) from m/z 341.2 to m/z 309.1 (Table 1).
spectra of M19 and M20, but were absent in M17 and M18, indicating
The same GSH adducts (M27 to M35; Fig. 7) were observed in
that they had different structures. M17 and M18 were assigned as other trapping experiments with GSH as a capture reagent in buffer A
3-substituted or 4-substituted N-methoxy pyrroline-2-one, which was or buffer B with RLM.
deduced from their fragmentation pattern similar to that of M14 to
M27 to M31. The major metabolites M27 to M31 exhibited an
M16. According to the NMR analysis of similar metabolites, L-739,010 m/z 632.2 (M + H+) with a chemical composition of C29H33N3O11S, which
Fig. 8. Metabolic profile of diosbulbin B trapped by GSH.

Reactive Metabolites of Diosbulbin B and Human Enzymes Involved
1745
Fig. 9. Metabolites found in the feces of rats
that were orally administered DB at 32 mg/kg
using an Agilent 6410 Triple Quad LC-MS/MS
system with eluting program A (A) and 150 mg/kg
using a Waters ACQUITY UPLC-Micromass
Q-TOF system with eluting program B (B).
indicated that one molecule of GSH conjugated to DB and was supported
M32 to M34. The metabolites M32 to M34 with signals at
by the fragment signal (m/z 557.2) with loss of glycine. These metabolites m/z 650.2 (M + H+) were generated by GSH conjugated to DB,
were generated by intramolecular reaction after GSH conjugated to the similar to the formation of M1 to M3 (Fig. 8). These metabolites are
2-butene-1,4-dial derivatives of DB (Lu et al., 2009) (Fig. 8). The main supported by their chemical formula (C29H35N3O12S) and the fragment
metabolite, M31, was composed of 3-mercapto-pyrrol (M31a) and signal (m/z 575.2) with loss of glycine.
2-mercapto-pyrrol derivatives of DB (M31b). Similar to M1 to M3, the
M35. A minor metabolite M35 was observed at m/z 939.2
configuration of C12 could be changed in GSH conjugates of DB. (M + H+). Its chemical formula (C39H50O17N6S2), as well as fragment
Configurations of chiral carbons in GSH moieties also could be changed signals with loss of one glycine (m/z 864.2) and two glycines (m/z 789.2)
(see Identification of the Synthesized Products). Therefore, M27 to M30 from the parent molecule, indicated that two molecules of GSH were
might be composed of isomers of M31a and M31b with configurations conjugated to DB. According to the formation of bis-GSH adducts of
changed in C12 or in the GSH moiety.
furan that were chemically synthesized (Chen et al., 1997), M35 was
Fig. 10. The typical total ion current (TIC) spectra of the chemically synthesized products M12 to M13 (A), MGCD, and M31 (B), using an Agilent 6410 Triple Quad
LC-MS/MS system with eluting program A, and their structures.

1746
Yang et al.
identified as a bis-GSH conjugate of DB. This metabolite was formed (d 169.99 and 117.31 for M12a; d 169.38 and 118.29 for M12b). M12a
by the mercapto group of one GSH added to the Schiff base formed by
the other GSH added to the aldehyde group of the reactive metabolites
(Fig. 8).
and M12b were identified as hemiacetal lactones, with the hemiacetal
group connected to the quaternary carbon of the alkene group. In
M13a and M13b, the hemiacetal group (d 99.00 for M13a and M13b)
was linked to the methyne carbon of the alkene group (d 135.70 and
147.46 for M13a; d 137.15 and 147.33 for M13b). The configurations
of C16 in M12a and M12b were determined to be R and S using the
Chem3D Ultra 8.0 software (CambridgeSoft Corporation, Waltham,
MA), according to the differences in the chemical shifts of H11. The
configurations of C15 in M13a and M13b were elucidated to be R and
S by the same method (Fig. 10). The chemical shifts of C and H of
M12 and M13, attributed by the 1H-1H correlation spectroscopy,
heteronuclear single quantum coherence, and heteronuclear multiple-
bond correlation (HMBC) spectra, are shown in Table 2.
M31 was eluted from the column at 28.6 minutes, and then lyophilized
to afford white powder. Signals similar to those of DB in the 13C-NMR
spectrum were all doubled, which indicated that M31 is composed of
two isomers (see Fig. 10; Supplemental Fig. 1). The proportions of
M31a and M31b were estimated to be approximately 1.2:1 according
to the peak area of similar 1H-NMR signals. M31a is 3-mercapto
substituted at the pyrrole moiety supported by the HMBC correlation
between H12 (d 5.36, dd) and the quaternary carbon (d 107.11) (see
Supplemental Fig. 2). M31b is 2-mercapto substituted at the pyrrole
moiety because no similar HMBC correlation was observed. Correlations
between H12 and H20 (CH3) in the nuclear Overhauser effect spec-
troscopy of M31a and M31b suggested that the configurations of C12 of
M31a and M31b were consistent with that of DB (see Supplemental Fig. 3).
NMR data of M31a and M31b are shown in Table 3.
Identification of Metabolites of DB In Vivo in Rats
No metabolite but DB was found in the plasma of rats after oral
administration at three dosages, compared with the blank plasma. A
signal of mono-oxydate (M10) in the urine of rats was observed when
DB was administered at 32 and 75 mg/kg. Several metabolites (M36
to M38) with signals at m/z 377(M + H+), in addition to M12 and
M13, were observed in the feces of rats when DB was administered at
32 mg/kg (Fig. 9). M36 to M38, which could not be identified by mass
data, might be the isomers of M12 and M13. Mono-GSH adducts
(m/z 632, M + H+), including M27, M30, and M31, were detected in the
bile and feces of rats at DB dosages of 75 and 150 mg/kg.
Identification of the Synthesized Products
The reaction mixture, exposed to air or not, would result in different
products before the residual DMDO was bubbled off in the stream
of nitrogen gas (see Enedial Derivative of DB). M12 and M13 (m/z
377) were the main products if the reaction mixture was exposed to air for
a few minutes before it was concentrated under nitrogen. Otherwise, GSH
adducts (m/z 632 and 939) were primarily produced if the reaction mixture
was kept from air as long as possible (Fig. 10). M27, M31, and M35 could
be observed in the total ion current spectrum, whereas minor M28 to M30
could only be observed in the extract ion spectrum.
M12 and M13 (at a proportion of 5:1) were eluted from the column
at 21.3 and 24.6 minutes, respectively. M12 and M13 were then
lyophilized to afford white powder, and each was composed of two
isomers. Most signals in the 13C-NMR spectrum of M12 are similar
to those of DB (Yonemitsu et al., 1993), except that the signals
belonging to the furan moiety of DB change to those of one carbonyl
of ester (d171.23 for M12a, d171.38 for M12b), one hemiacetal group
Another MGCD, an analog of M31, was eluted at 30.3 minutes and
1
then lyophilized to afford white powder. The H-NMR spectrum of
this MGCD (data not shown) was very similar to that of M31.
Correlations between H12 and H20 (CH3) in the nuclear Overhauser
effect spectroscopy spectrum were also observed, which suggests
that the MGCD was composed of isomers of M31a and M31b
(d 99.49 for M12a, d 98.50 for M12b), and one alkene group with configurations changed in the GSH moiety. The 13C-NMR
TABLE 2
1H (600 MHz) and 13C-NMR (150 MHz) data for M12a/M12b and M13a/M13b (pyridine-d5)
M12a/M12b
M13a/M13b
C/H Number
C-1/H-1
dH
J-H,H Values
dC
dH
J-H,H Values
dC
1.50(m)
1.84(m)
4.76(dd)
1.72(d)
2.34(dd)
2.79(d)
1.99(d)
4.78(d)
2.13(d)
2.53(dd)
29.52/29.92
1.23(m)/1.43(m)
1.71(m)/1.82(m)
4.74(dd)
29.68/29.22
C-2/H-2
C-3/H-3
5.5, 5.5
11.7
5.4, 11.7
4.4
12.0
5.7
76.94
39.20
5.5, 5.5
11.5
5.5, 11.5
5
11.7
6.0
76.74
39.00
1.67(d)
2.33(dd)
2.77(d)
1.92(d)
4.73(d)
2.07(d)
C-4/H-4
C-5/H-5
C-6/H-6
C-7/H-7
42.42
42.33
78.11/77.9
37.19
42.19
42.07/41.98
77.66/77.60
37.12
11.6
5.8, 11.6
11.6
5.5, 11.6
2.51(dd)/2.49(dd)
C-8
C-9
C-10/H-10
C-11/H-11
90.02/90.22
45.73
40.98
40.52/39.63
90.05/89.96
45.55/45.47
38.89
2.50(m)
2.04(m)/1.92(m)
1.22(dd)/1.65(dd)
1.98(dd)/2.23(dd)
5.20(dd)
2.0(dd)/1.82(dd)
2.24(dd)/2.08(dd)
5.39(dd)/5.20(dd)
11.7, 12.1/11.4, 11.7
6.0, 12.1/5.3, 11.7
5.5, 10.5
10.5, 11.5
5.5, 11.5
5.5, 10.5
40.75/40.38
C-12/H-12
C-13
C-14
C-15/H-15
C-16/H-16
C-17
76.55/76.83
169.99/169.38
117.31/118.29
171.38/171.23
99.49/98.5
75.46/75.16
135.7/137.15
147.46/147.33
99.00
6.56(s)/6.79(s)
6.46(s)/6.71(s)
7.90(s)
6.52(s)/6.62(s)
170.82
176.77
175.88
177.08/176.75
176.03
C-18
C-20/3H
1.09(s)/1.06(s)
16.21/15.94
0.98(s)/0.99(s)
15.87/15.75
C/H, Carbon/Hydrogen; d, doublet; dd, doublet doublet; m, multiplet; s, singlet.

Reactive Metabolites of Diosbulbin B and Human Enzymes Involved
1747
TABLE 3
spectrum of the MGCD was not acquired because of insufficient
compound.
1H (600 MHz) and 13C-NMR (150 MHz) data for M31a and M31b (DMSO-d6)
M31a/M31b
Incubations of DB with Recombinant Human P450 Enzymes
C/H Number
dH
J-H,H Values
dC
CYP3A4 and 3A5, as well as 2C19 and 2C9, were capable of
bioactivating DB to M31, with CYP3A4 having the highest catalytic
activity (Fig. 11).
C-1/H-1
1.67(m)
1.93(m)
4.82(dd)
29.23/29.33
C-2/H-2
C-3/H-3
5.2, 5.2
11.3
77.34
38.76
1.93(d)
Incubations of DB with Pooled HLM and Specific Chemical
Inhibitors
2.4(dd)
2.71(d)
2.19(d)
4.73(d)/4.75(d)
2.11(d)/2.21(d)
2.32(dd)/2.43(dd)
5.4, 11.3
4.6
10.1
5.2
10.7
C-4/H-4
C-5/H-5
C-6/H-6
C-7/H-7
41.98/41.90
41.53/41.36
77.78/78.51
37.51/37.33
The inhibited reaction activities for the formation of M31 caused
by ketoconazole, sulfaphenazole, and (S)-(+)-N-3-benzylnirvanol
were 87, 15, and 27%, respectively. The results indicated that
CYP3A4 was the major enzyme responsible for the bioactivation
of DB.
5.4, 10.7
C-8
C-9
C-10/H-10
C-11/H-11
89.53/90.08
45.77/45.71
39.23/39.10
41.24
1.67(m)
1.94(dd)
2.00(dd)
5.36(dd)/5.28(dd)
11.3, 11.7
5.4, 11.7
5.2, 10.4
5.2, 10.1
Enzyme Kinetics Studies on MGCD (M31) Formation
C-12/H-12
C-13
C-14/H-14
C-15/H-15
C-16/H-16
C-17
75.68/74.90
126.39/125.15
107.11/128.64
121.82/107.73
134.01/122.38
177.89
176.84
16.86/16.73
179.22
The best fit of data was obtained using the Michaelis-Menten
equation for enzyme kinetics with pooled HLM (Fig. 12). Eadie-
Hofstee transformations showed no evidence of two-enzyme kinetics.
CYP3A4, 3A5, 2C9, and 2C19 also exhibited Michaelis-Menten
kinetics with no evidence of uncompetitive substrate inhibition. The
enzyme kinetics parameters are shown in Table 4.
6.99(bs) for 3b
6.64(bs) for 3a
6.96(bs) /6.83(bs)
C-18
C-20/3H
1.22(s)
Glu (-COOH)
Glu a-CH
Glu b-CH2
Glu g-CH2
Glu (-CO-NH-)
Cys a-CH
Cys b-CH2
4.81(m)
2.1-2.3(m)
2.1-2.3(m)
61.75
29.33
29.92
Prediction of Relative Contributions for P450 Enzymes in HLM to
MGCD (M31) Formation
4.23(m)/4.01(m)
2.97(m)/3.21(m)
2.58(m)/2.81(m)
58.16/56.45
35.9/33.57
The RAF predictions were made at 5.0 and 40 mM DB. P3A5 was
not included in the RAF calculations. Considering the low Vmax and
large Km, as well as the low expression in HLM, the contribution of
CYP3A5 was lower than that of CYP2C19. The relative contributions
of CYP3A4 were approximately 83.3 and 81.3%. CYP2C9 contrib-
uted about 10.5 and 12.2%, which were higher than the contributions
of CYP2C19 (6.2 and 6.6%).
Cys (-CO-NH-)
Gly CH2
Gly (-COOH)
171.26/171.40
41.77
171.26/171.40
3.62 (overlapped)
bs, broad singlet; C/H, Carbon/Hydrogen; d, doublet; dd, doublet doublet; m, multiplet.
MGCD was finally synthesized because of the extremely low yield
of metabolites in microsomal incubations.
Discussion
The concentration of DMDO was not measured in the procedure
because DMDO rapidly escaped from the solution, and DMDO mea-
surement is time-consuming (Murray and Singh, 1997). A 1.2-ml-volume
of DMDO solution was sufficient to oxidize 0.029 mmol of DB, sug-
gesting that the concentration of DMDO was .0.024 M. UPLC-MS/MS
DB metabolism in vitro and in vivo in rats, as well as the human
enzymes involved, was first investigated. The main metabolic pathway
of DB was catalyzed by CYP3A4/5, 2C9, and 2C19 to generate the
reactive metabolites, 2-butene-1,4-dial derivatives of DB, which were
captured by nucleophiles such as Tris base, methoxylamine, and GSH or
oxidized to hemiacetal lactones. Tris base conjugates of DB (M1 to M3)
that were observed at the total ion current spectrum of microsomal
incubations in the Tris-HCl buffer incubation system provided important
information on the metabolism pathway of DB, compared with the few
data on metabolites in the phosphate buffer system. Tris base was used
as a buffer constituent and as a trapping agent for the reactive me-
tabolites of DB because Tris base could react with aldehyde to form the
Schiff base (Cannon and Davison, 1976; Niedernhofer et al., 1997). The
metabolites of DB found in vivo in rats were consistent with those
observed in RLM incubations in vitro. M10 was found in rat urine and
MGCD in rat bile and feces, as well as M12 and M13 in rat feces, which
indicated that DB was also transformed into 2-butene-1,4-dial deri-
vatives in vivo. DB metabolism in the liver and intestine decreased the
bioavailability of DB when the compound was orally administered. In
addition, species metabolic difference was observed in the formation of
mono-oxydates; the two mono-oxydates (M10 and M11) were found in
rabbits and guinea pigs and one (M10) in rats, mice, and human micro-
somal incubations. The results might have been caused by the varying
expression of flavin-containing mono-oxygenases among these species
(data not shown).
1
was used, instead of H-NMR as previously reported (Druckova and
Marnett, 2006), to monitor the disappearance of DB that was oxidized
Fig. 11. Formation of metabolite M31 by recombinant human cytochrome P450
isoenzymes after incubation with 50 mM DB for 60 minutes. Data shown are the
area obtained in the mass spectrum for M31 relative to the internal standard and
normalized to the amount of P450 (average of triplicate determinations 6 S.D.).

1748
Yang et al.
Fig. 12. Enzyme kinetics after incubation with 5–300 mM DB for 60 minutes. Data show the mean of triplicate determinations.
by DMDO. Unexpectedly, several products with signals at m/z 377, that M7 to M9 were temporarily inferred due to their few mass data.
including M12 and M13, were detected. Thus, M12 and M13 were Chemical synthesis of M12, M13, M35, and M27 to M31 proved the
successfully synthesized. The reaction was completed in 15 minutes, and existence of reactive metabolites, and finally, the NMR data of M12,
the major products (M31a and M31b) transferred to their isomers (M27 M13, and M31 confirmed the structures of reactive metabolites.
and M28) when the reaction continued, which made the separation more
UPLC-MS/MS with multiple reaction mode was used to improve
difficult. M27 and minor M28 to M30 were hard to separate because selectivity and sensitivity to determine MGCD in incubations. MGCD
they were always eluted together. M35 (m/z 939), having poor stability was insoluble in acetonitrile, but soluble in DMSO. DMSO (10%) was
because of the mercapto group, was hard to obtain as well.
used to reconstitute the residual to determine MGCD in enzyme kinetic
The structures of the reactive metabolites of DB were confirmed samples, which resulted in high recoveries (90–110%) of MGCD. Prior
by the structure elucidations of metabolites in the trapping experiments to protein precipitation, 10 ml of DMSO was added to keep the same
of Tris base, methoxylamine, and GSH. Although it was difficult procedure with calibration samples.
to distinguish metabolites with different configurations, the struc-
CYP3A4 was identified as the major bioactivation enzyme of DB
ture elucidations from mass data and literatures made sense except both by chemical inhibition studies and RAF approach. Considering

Reactive Metabolites of Diosbulbin B and Human Enzymes Involved
1749
TABLE 4
Enzyme kinetic parameters for diosbulbin B to M31
HLM and P450s
Vmax
Km
Vmax/Km
pmol·min21·mg21 protein or
pmol·min21·pmol21 P450
mM
l·min21·mg21 protein or
l·min21·pmol21 P450
HLM
3A4
3A5
2C9
2C19
776.2
6.3
2.0
1.9
6.3
254.0
68.3
98.5
133.8
92.0
3.06 ꢀ 1023
9.22 ꢀ 1025
2.03 ꢀ 1025
1.42 ꢀ 1025
6.85 ꢀ 1025
that the RAF approach provides a satisfactory prediction of the roles of antitumor agents that depend on P450 activation may also be activated
P450 isoforms compared with inhibition studies, contribution of in tumor regions. For example, AQ4N (1,4-bis-{[2-(dimethylamino-N-
CYP2C9 to the bioactivation of DB was higher than that of CYP2C19. oxide)ethyl]amino}-5,8-dihydroxy-anthracene-9,10-dione), a CYP3A
However, the result in the RAF approach was contrary to those from substrate, is activated to a cytotoxic metabolite specifically in hypoxic
the inhibition studies. The contrasting results may have been caused tumor regions (Patterson and Murray, 2002). The relationship between
by the compensation effects from other enzymes when CYP2C9 was reactive metabolites of DB and antitumor activity should be inves-
inhibited. The Km of CYP3A4 in the formation of M31 was 68.1 mM, tigated in future studies.
indicating that DB had moderate affinity to CYP3A4.
DB and teucrin A both are bioactivated in the furan moiety. Furan
has already been proved to transform into cis-2-butene-1,4-dial by
P450 (Chen et al., 1995). This reactive substance, cis-2-butene-1,4-dial,
is likely responsible for the ultimate toxic and carcinogenic properties of
furan because it readily reacts with amino acids (Chen et al., 1997) and
DNA (Byrns et al., 2006). Mono-GSH adduct of furan, as well as
metabolites derived from cysteine cis-2-butene-1,4-dial lysine cross-
links, was found in rat urine, which suggested that furan was also
bioactivated to form cis-2-butene-1,4-dial in vivo (Peterson et al., 2006;
Lu and Peterson, 2010). A substantial amount of radioactive materials
remained in the liver bound to proteins at 24 hours after dosing 8 mg/kg
[14C]furan to rats, which could cause hepatotoxicity in rats (Burka et al.,
1991). Hepatotoxicity is correlated with covalent binding to hepatic
proteins with reactive metabolites, which also has been clarified from
numerous drugs. Acetaminophen is very safe at therapeutic doses. How-
ever, acetaminophen overdose may produce fulminating hepatic necrosis
that can be fatal. Covalent binding resulting from reactive metabolites is
the most important factor (Hinson et al., 2004). Troglitazone, an oral
antidiabetic agent, has been removed from the market because it has been
associated with cases of severe hepatotoxicity and drug-induced liver
failure. Reactive metabolites may play a causative role (Kassahun et al.,
2001). The bioactivation of DB that has been inferred to form reactive
metabolites by liver P450 might be primarily responsible for the hepato-
toxicity of DB and D. bulbifera.
Authorship Contributions
Participated in research design: Yang, Chen, Z. Wang, C. Wang.
Conducted experiments: Yang, Liu.
Contributed new reagents or analytic tools: Yang, Liu, C. Wang.
Performed data analysis: Yang, Liu, C. Wang.
Wrote or contributed to the writing of the manuscript: Yang, Liu, C. Wang.
References
Bradford MM (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.
Byrns MC, Vu CC, Neidigh JW, Abad JL, Jones RA, and Peterson LA (2006) Detection of DNA
adducts derived from the reactive metabolite of furan, cis-2-butene-1,4-dial. Chem Res Toxicol
19:414–420.
Burka LT, Washburn KD, and Irwin RD (1991) Disposition of [14C]furan in the male F344 rat.
J Toxicol Environ Health 34:245–257.
Cannon DJ and Davison PF (1976) A stabilised tris(hydroxymethyl)aminomethane adduct in
reduced collagen. Connect Tissue Res 4:187–191.
Chen LJ, Hecht SS, and Peterson LA (1995) Identification of cis-2-butene-1,4-dial as a micro-
somal metabolite of furan. Chem Res Toxicol 8:903–906.
Chen LJ, Hecht SS, and Peterson LA (1997) Characterization of amino acid and glutathi-
one adducts of cis-2-butene-1,4-dial, a reactive metabolite of furan. Chem Res Toxicol
10:866–874.
Chen Y, Ferguson SS, Negishi M, and Goldstein JA (2004) Induction of human CYP2C9 by
rifampicin, hyperforin, and phenobarbital is mediated by the pregnane X receptor. J Pharmacol
Exp Ther 308:495–501.
Crespi CL and Miller VP (1999) The use of heterologously expressed drug metabolizing enzymes
—state of the art and prospects for the future. Pharmacol Ther 84:121–131.
Druckova A and Marnett LJ (2006) Characterization of the amino acid adducts of the enedial
derivative of teucrin A. Chem Res Toxicol 19:1330–1340.
Harmsen S, Meijerman I, Beijnen JH, and Schellens JHM (2009) Nuclear receptor mediated
induction of cytochrome P450 3A4 by anticancer drugs: a key role for the pregnane X receptor.
Cancer Chemother Pharmacol 64:35–43.
Previous studies have demonstrated that extract from D. bulbifera,
in which DB is the main ingredient, could cause higher hepatotoxicity
in male compared with female mice (Wang et al., 2011). Pharmaco-
kinetic studies revealed that the oral absolute bioavailability of DB in
female rats was nearly seven times higher than that in males (Yang
et al., 2013). Given that DB was mainly metabolized by CYP3A4 in
humans, higher expression of 3A, including 3A1 and 3A2, in male
than in female mice and rats (Kato and Yamazoe, 1992) could enhance
more metabolized DB, resulting in higher hepatotoxicity in male mice
and lower oral bioavailability in male rats than in females. Concerning
drug interaction, it should be noted that D. bulbifera may cause fatal
toxicity when used with inducers of CYP3A4 such as rifampicin and
hyperforin that also induce CYP2C9 expression (Chen et al., 2004;
Harmsen et al., 2009).
Hinson JA, Reid AB, McCullough SS, and James LP (2004) Acetaminophen-induced hepato-
toxicity: role of metabolic activation, reactive oxygen/nitrogen species, and mitochondrial
permeability transition. Drug Metab Rev 36:805–822.
Ida Y, Kubo S, Fujita M, Komori T, and Kawasaki T (1978) Furanoid-norditerpene aus Pflanzen
der Familie Dioscoreaceae, V. Struktur der Diosbulbine-D, -E, -F, -G und H. Liehigs Ann Chem
1978:818–833.
Kassahun K, Pearson PG, Tang W, McIntosh I, Leung K, Elmore C, Dean D, Wang R, Doss G,
and Baillie TA (2001) Studies on the metabolism of troglitazone to reactive intermediates in
vitro and in vivo. Evidence for novel biotransformation pathways involving quinone methide
formation and thiazolidinedione ring scission. Chem Res Toxicol 14:62–70.
Kato R and Yamazoe Y (1992) Sex-specific cytochrome P450 as a cause of sex- and species-
related differences in drug toxicity. Toxicol Lett 64-65:661–667.
Kawasaki T, Komori T, and Setoguchi S (1968) Furanoid- norditerpenes from Dioscorea plants,
I. diosbulins A, B, and C from Dioscorea bulbifera L. forma spontanea Makino et Nemoto.
Chem Pharm Bull (Tokyo) 16:2430.
Komori T (1997) Glycosides from Dioscorea bulbifera. Toxicon 35:1531–1536.
Liu H, Chou GX, Guo YL, Ji LL, Wang JM, and Wang ZT (2010) Norclerodane diterpenoids
from rhizomes of Dioscorea bulbifera. Phytochemistry 71:1174–1180.
Lu AYH and Levin W (1972) Partial purification of cytochromes P-450 and P-448 from rat liver
microsomes. Biochem Biophys Res Commun 46:1334–1339.
Lu D, Sullivan MM, Phillips MB, and Peterson LA (2009) Degraded protein adducts of cis-
2-butene-1,4-dial are urinary and hepatocyte metabolites of furan. Chem Res Toxicol
22:997–1007.
The relationship of the reactive metabolites with the antitumor
activity of DB remains unknown. P450 expression, particularly 1A,
1B, 2C, 3A, and 2D subfamily members, has been identified in a
wide range of human cancers (Patterson and Murray, 2002). Prodrug
Lu D and Peterson LA (2010) Identification of furan metabolites derived from cysteine-cis-2-
butene-1,4-dial-lysine cross-links. Chem Res Toxicol 23:142–151.

1750
Yang et al.
Murray RW and Singh M (1997) Synthesis of expoxides using dimethyldioxirane: trans-stilbene
oxide: [dioxirane, dimethyl - and oxirane, 2,3-diphenyl-, trans]. Org Synth 74:91–100.
Niedernhofer LJ, Riley M, Schnetz-Boutaud N, Sanduwaran G, Chaudhary AK, Reddy GR,
Wang J, Liang Q, Ji L, Liu H, Wang C, and Wang Z (2011) Gender-related difference in liver
injury induced by Dioscorea bulbifera L. rhizome in mice. Hum Exp Toxicol 30:1333–1341.
Wang G, Liu JS, Lin BB, Wang GK, and Liu JK (2009) Two new furanoid norditerpenes from
Dioscorea bulbifera. Chem Pharm Bull (Tokyo) 57:625–627.
and Marnett LJ (1997) Temperature-dependent formation of
a conjugate between tris
(hydroxymethyl)aminomethane buffer and the malondialdehyde-DNA adduct pyrimidopurinone.
Chem Res Toxicol 10:556–561.
Niu ZM and Chen AY (1994) 16 cases report of toxic hepatitis caused by Dioscorea bulbifera.
Yang B, Wang X, Liu W, Zhang Q, Chen K, Ma Y, Wang C, and Wang Z (2013) Gender-related
pharmacokinetics and absolute bioavailability of diosbulbin B in rats determined by ultra-
performance liquid chromatography-tandem mass spectrometry. J Ethnopharmacol 149:810–815.
Yonemitsu M, Fukuda N, Kimura T, and Komori T (1993) Diosbulbin-B from the leaves and
stems of Dioscorea bulbifera: 1H-1H and 13C-1H COSY NMR studies. Planta Med 59:577.
Zhang YP, Gao W, and Gao HY (2008) Research progress of Dioscorea bulbifera Linn. Mod
Chinese Med 10:34–37.
Chin J Integr Tradit Western Liver Dis 4:55–56.
Njuguna NM, Masimirembwa C, and Chibale K (2012) Identification and characterization of
reactive metabolites in natural products-driven drug discovery. J Nat Prod 75:507–513.
Patterson LH and Murray GI (2002) Tumour cytochrome P450 and drug activation. Curr Pharm
Des 8:1335–1347.
Zhang KE, Naue JA, Arison B, and Vyas KP (1996) Microsomal metabolism of the
Peterson LA, Cummings ME, Chan JY, Vu CC, and Matter BA (2006) Identification of a cis-2-
butene-1,4-dial-derived glutathione conjugate in the urine of furan-treated rats. Chem Res
Toxicol 19:1138–1141.
Venkatakrishnan K, Von Moltke LL, and Greenblatt DJ (2001) Human drug metabolism and the
cytochromes P450: application and relevance of in vitro models. J Clin Pharmacol 41:
1149–1179.
Walsky RL and Obach RS (2003) Verification of the selectivity of (+)N-3-benzylnirvanol as
a CYP2C19 inhibitor. Drug Metab Dispos 31:343.
Wang JM, Ji LL, Branford-White CJ, Wang ZY, Shen KK, Liu H, and Wang ZT (2012) Anti-
tumor activity of Dioscorea bulbifera L. rhizome in vivo. Fitoterapia 83:388–394.
5-lipoxygenase inhibitor L-739,010: evidence for furan bioactivation. Chem Res Toxicol 9:547–554.
Address correspondence to: Changhong Wang, Institute of Chinese Materia
Medica, Shanghai University of Traditional Chinese Medicince, 1200 Cailun Rd,
Shanghai 201210, China. E-mail: wchcxm@hotmail.com; or Zhengtao Wang,
Institute of Chinese Materia Medica, Shanghai University of Traditional Chinese
Medicince, 1200 Cailun Rd, Shanghai 201210, China. E-mail: wangzht@hotmail.com