Full structure assignments of pyrrolizidine alkaloid DNA adducts and mechanism of tumor initiation

Article abstract of DOI:10.1021/tx300292h

Pyrrolizidine alkaloid-containing plants are widespread in the world and are probably the most common poisonous plants affecting livestock, wildlife, and humans. Pyrrolizidine alkaloids are among the first chemical carcinogens identified in plants. Previously, we determined that metabolism of pyrrolizidine alkaloids in vivo and in vitro generated a common set of DNA adducts that are responsible for tumor induction. Using LC-ESI/MS/MS analysis, we previously determined that four DNA adducts (DHP-dG-3, DHP-dG-4, DHP-dA-3, and DHP-dA-4) were formed in rats dosed with riddelliine, a tumorigenic pyrrolizidine alkaloid. Because of the lack of an adequate amount of authentic standards, the structures of DHP-dA-3 and DHP-dA-4 were not elucidated, and the structural assignment for DHP-dG-4 warranted further validation. In this study, we developed an improved synthetic methodology for these DNA adducts, enabling their full structural elucidation by mass spectrometry and NMR spectroscopy. We determined that DHP-dA-3 and DHP-dA-4 are a pair of epimers of 7-hydroxy-9-(deoxyadenosin-N⁶-yl) dehydrosupinidine, while DHP-dG-4 is 7-hydroxy-9-(deoxyguanosin-N²-yl)dehydrosupinidine, an epimer of DHP-dG-3. With the structures of these DNA adducts unequivocally elucidated, we conclude that cellular DNA preferentially binds dehydropyrrolizidine alkaloid, for example, dehydroriddelliine, at the C9 position of the necine base, rather than at the C7 position. We also determined that DHP-dA-3 and DHP-dA-4, as well as DHP-dG-3 and DHP-dG-4, are interconvertible. This study represents the first report with detailed structural assignments of the DNA adducts that are responsible for pyrrolizidine alkaloid tumor induction on the molecular level. A mechanism of tumor initiation by pyrrolizidine alkaloids is consequently fully determined.

Full text of DOI:10.1021/tx300292h

Article
pubs.acs.org/crt
Full Structure Assignments of Pyrrolizidine Alkaloid DNA Adducts
and Mechanism of Tumor Initiation
Yuewei Zhao,† Qingsu Xia,† Goncalo Gamboa da Costa,† Hongtao Yu,‡ Lining Cai,§ and Peter P. Fu*,†
̧
†National Center for Toxicological Research, Jefferson, Arkansas 72079, United States
‡Department of Chemistry and Biochemistry, Jackson State University, Jackson, Mississippi 39217, United States
§Biotranex LLC, Monmouth Junction, New Jersey 08852, United States
S
* Supporting Information
ABSTRACT: Pyrrolizidine alkaloid-containing plants are
widespread in the world and are probably the most common
poisonous plants affecting livestock, wildlife, and humans.
Pyrrolizidine alkaloids are among the first chemical carcino-
gens identified in plants. Previously, we determined that
metabolism of pyrrolizidine alkaloids in vivo and in vitro
generated a common set of DNA adducts that are responsible
for tumor induction. Using LC-ESI/MS/MS analysis, we
previously determined that four DNA adducts (DHP-dG-3,
DHP-dG-4, DHP-dA-3, and DHP-dA-4) were formed in rats dosed with riddelliine, a tumorigenic pyrrolizidine alkaloid. Because
of the lack of an adequate amount of authentic standards, the structures of DHP-dA-3 and DHP-dA-4 were not elucidated, and
the structural assignment for DHP-dG-4 warranted further validation. In this study, we developed an improved synthetic
methodology for these DNA adducts, enabling their full structural elucidation by mass spectrometry and NMR spectroscopy. We
determined that DHP-dA-3 and DHP-dA-4 are a pair of epimers of 7-hydroxy-9-(deoxyadenosin-N6-yl) dehydrosupinidine, while
DHP-dG-4 is 7-hydroxy-9-(deoxyguanosin-N2-yl)dehydrosupinidine, an epimer of DHP-dG-3. With the structures of these DNA
adducts unequivocally elucidated, we conclude that cellular DNA preferentially binds dehydropyrrolizidine alkaloid, for example,
dehydroriddelliine, at the C9 position of the necine base, rather than at the C7 position. We also determined that DHP-dA-3 and
DHP-dA-4, as well as DHP-dG-3 and DHP-dG-4, are interconvertible. This study represents the first report with detailed
structural assignments of the DNA adducts that are responsible for pyrrolizidine alkaloid tumor induction on the molecular level.
A mechanism of tumor initiation by pyrrolizidine alkaloids is consequently fully determined.
hydroxy-1-hydroxymethyl-5H-pyrrolizine (DHP)-derived
DNA adduct formation.17 The levels of the DHP-derived
DNA adducts correlated closely with tumorigenic potencies in
mice fed different doses of riddelliine.17−19 The metabolism of
riddelliine by human liver microsomes resulted in a metabolic
pattern and DNA adduct profile highly similar to those formed
in rat liver, indicating that the results of mechanistic studies
with experimental rodents are highly relevant to humans.20
These overall results suggest that riddelliine can be genotoxic to
humans via DHP-derived DNA adduct formation.17 On the
basis of the results of human exposure,13 tumor potency in
experimental rodents,21 and mechanistic findings on riddelliine,
the U.S. National Toxicology Program (NTP) classified
riddelliine as “reasonably anticipated to be a human carcinogen”
in the NTP 12th Report of Carcinogens.22
INTRODUCTION
■
Pyrrolizidine alkaloids and pyrrolizidine alkaloid N-oxides are
common constituents of hundreds of plant species of different
unrelated botanical families distributed in many geographical
regions of the world.1−10 More than 660 pyrrolizidine alkaloids
and pyrrolizidine alkaloid N-oxides have been identified in over
6000 plants, and about half of them exhibit hepatotoxic
activity.1,6−8 It has been reported that about 3% of the world's
flowering plants contain toxic pyrrolizidine alkaloids.10 It is
highly possible that pyrrolizidine alkaloid-containing plants are
the most common poisonous plants affecting livestock, wildlife,
and humans.3,6,11−14
Pyrrolizidine alkaloids are among the first chemical
carcinogens identified in plants. In 1950, Cook et al.15 reported
that rats treated with alkaloids of Senecio jacobea developed liver
tumors. In 1954, retrorsine, a pyrrolizidine alkaloid, was found
to induce liver tumors in rats.16 Since then, although extensive
studies on liver tumor formation induced by pyrrolizidine
alkaloids spanned over nearly half a century, the elucidation of
the mechanism by which pyrrolizidine alkaloids induce tumors
failed. It was not until 2001 that we found that riddelliine, a
tumorigenic pyrrolizidine alkaloid, induces liver tumors through
a genotoxic mechanism mediated by ( )-6,7-dihydro-7-
Subsequent mechanistic studies on a series of different types
of the pure tumorigenic pyrrolizidine alkaloids, Chinese herbal
plants, and herbal dietary supplements confirmed that under
similar experimental conditions, the same DHP-derived DNA
adducts were generated in vivo and/or in vitro.23−30 The
Received: June 27, 2012
Published: August 2, 2012
© 2012 American Chemical Society
1985
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Scheme 1. Synthesis of DHP-dA Adducts from Reaction of Dehydroriddelliine and dA
formation of the same DHP-derived DNA adducts from all of
these studies indicates that the DHP-derived DNA adducts are
potential biomarkers of pyrrolizidine alkaloid exposure and
tumorigenicity.3,26
be resolved. These include (i) the structures of DHP-dA-3 and
DHP-dA-4 adducts; (ii) because of a limited amount of DHP-
dG-4 available, the structural assignment of DHP-dG-4 has to
be validated; and (iii) whether or not other DHP-derived DNA
adducts are formed in vivo and/or in vitro. To clarify these
matters, alternative synthetic approaches had to be developed
to provide DNA adducts in a sufficient amount for structural
identification.
In this study, we report the successful preparation of DHP-
dG-3, DHP-dG-4, DHP-dA-3, and DHP-dA-4 by reaction of
dehydroriddelliine with dG and dA. UV−visible spectropho-
tometry, circular dichroism (CD) analysis, and mass spectro-
metric and NMR spectroscopic analyses indicated that the
structures of DHP-dA-3 and DHP-dA-4 were a pair of epimers
of 7-hydroxy-9-(deoxyadenosin-N6-yl)-dehydrosupinidine and
that the structure of DHP-dG-4 was 7-hydroxy-9-(deoxygua-
nosin-N2-yl)dehydrosupinidine instead of the previously
a s s i g n e d 7 - h y d r o x y -5 - ( d e o x y g u a n o s i n - N 2 - y l ) -
dehydrosupinidine. We also determined that DHP-dG-3 and
DHP-dG-4, as well as DHP-dA-3 and DHP-dA-4, are
interconvertible.
The identification and quantification of these DHP-derived
DNA adducts were accomplished by the 32P-postlabeling/
HPLC method.18,31 There are several limitations associated
with 32P-postlabeling/HPLC methods for DNA adduct analysis,
in particular the lack of structural information about the DNA
adducts. Subsequently, we developed an LC-ESI/MS/MS
methodology for identification and quantitation of DHP-
derived DNA adducts in vivo and in vitro.1 The DHP-dG and
DHP-dA synthetic standards used for HPLC-ESI-MS/MS
analysis were prepared by reaction of dehydroretronecine
(DHR) with dG and dA, respectively. Reaction of DHR and dG
resulted in four DHP-dG adducts: an epimeric pair of 7-
(deoxyguanosin-N2-yl)dehydrosupinidine (DHP-dG-1 and
DHP-dG-2), 7-hydroxy-9-(deoxyguanosin-N2 -yl)-
dehydrosupinidine (DHP-dG-3), and 7-hydroxy-5-(deoxygua-
nosin-N2-yl)dehydrosupinidine (DHP-dG-4). Reaction of DHR
and dA resulted in four DHP-dA adducts: an epimeric pair of 7-
(deoxyadenosin-N6-yl)dehydrosupinidine (DHP-dA-1 and
DHP-dA-2) and two DHP-dA adducts tentatively assigned as
x-(deoxyadenosin-N6-yl)-dehydrosupinidine (DHP-dA-3), and
y-(deoxyadenosin-N6-yl)dehydrosupinidine (DHP-dA-4), re-
spectively.1
With these synthetical standards available, the DNA adducts
formed in vitro and in vivo were analyzed. We found that
reaction of calf thymus DNA and DHR generated all of the four
DHP-dG and four DHP-dA adducts, with DHP-dA-3 and
DHP-dA-4 being the most abundant, followed by the four
DHP-dG adducts, and DHP-dA-1 and DHPdA-2 the least
abundant.1 Two liver samples were obtained from in vivo
experiments using two different time courses: (i) female F344
rats were gavaged for 3 consecutive days with riddelliine or
monocrotaline and (ii) female F344 rats were treated with
riddelliine for 6 months. We determined that the in vivo liver
DNA samples contained only DHP-dG-3, DHP-dG-4, DHP-
dA-3, and DHP-dA-4 and that they were formed in a dose−
responsive manner.1
EXPERIMENTAL PROCEDURES
Caution: Riddelliine and DHP are carcinogenic in laboratory animals.
They should be handled with extreme care, using proper personal protective
equipment and a well-ventilated hood.
■
Chemicals. o-Bromanil, sodium bicarbonate, magnesium chloride,
calf thymus DNA (sodium salt, type I), 2′-deoxyguanosine (dG), 2′-
deoxyadenosine (dA), glucose-6-phosphate, glucose-6-phosphate
dehydrogenase, and nicotinamide adenine dinucleotide phosphate
(NADP+) were purchased from the Sigma Chemical Co. (St. Louis,
MO). Riddelliine was obtained from Dr. Po-Chuen Chan, NTP.
Alkaline phosphatase was acquired from Roche Diagnostics Corpo-
ration (Indianapolis, IN). Dehydroriddelliine was prepared by
oxidation of riddelliine as previously described.32,33 All solvents used
were HPLC grade.
Synthesis of DHP-dA Adducts. The synthesis involved the
reaction of dehydroriddelliine with dA (Scheme 1). Briefly, a solution
of dehydroriddelliine (10 mg, 28.7 μmol), dA (21.6 mg, 86 μmol), and
NaHCO3 (60 mg, 714 μmol) in 3 mL of DMF was reacted with
stirring at 25 °C for 1 h. Deionized water (0.3 mL) was added
dropwise, and the resulting solution was kept stirring for another 6 h.
The reaction was quenched with deionized water (1.7 mL), and
products were separated by HPLC, first with a Whatman ODS-3
column to remove excess dA and then with a Phenomenex C18 Luna
Nevertheless, there are several features considering the
mechanism of pyrrolizidine alkaloid carcinogenesis that need to
1986
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
5 μm ODS column (250 mm × 4.6 mm) eluted at 1 mL/min; elution
conditions: 0−5 min, 10% acetonitrile in water; 5−50 min, 10−18%
acetrontrile in water.
for LC-MS/MS analysis. For those from metabolism, after enzymatic
hydrolysis of the resulting reaction, the sample was dried by
lypholization. The residue was reconstructed with 100 μL of water.
The reconstituted sample was qualitatively analyzed on an LC-MS/MS
system.
Synthesis of DHP-dG Adducts. Into 3 mL of anhydrous DMF in
a 10 mL round-bottom flask, dehydroriddelliine (10 mg, 28.7 μmol),
dG (24.4 mg, 91.5 μmol), and NaHCO3 (60 mg, 714 μmol) were
added, and the solution was stirred at room temperature for 1 h.
Subsequently, 0.3 mL of deionized water was introduced dropwise into
the solution under stirring. Six hours later, 1.7 mL of deionized water
was added into the reaction mixture, and the solution became clear.
The reaction was monitored by HPLC using a Luna C18 (2) column
(250 mm × 4.6 mm, 5 μm, Phenomenex, Torrance, CA); flow rate, 1
mL/min, monitored at 256 nm; gradient: 0−5 min, 8% acetonitrile in
water; 5−50 min, 8−16% acetonitrile in water.
Interconversion between DHP-dA-3 and DHP-dA-4. Approx-
imately 0.5 mL of a solution of pure DHP-dA-3 or DHP-dA-4 in
acetonitrile/H2O (v/v, 1/10) in a small vial was kept at room
temperature. This solution was monitored by HPLC at 0, 1, 2, 3, 4, 5,
and 6 days. The percentage of DHP-dA-3 and DHP-dA-4 was
calculated from the ratio of their respective peak areas relative to the
sum of their peak areas. The sum of peak areas was stable ( 3%) over
the incubation period, and no additional new peaks were found during
the HPLC analysis, which indicates that these two adducts were
relatively stable under these conditions.
The liquid chromatography system consisted of a Shimadzu
Prominence HPLC system, including a CBM-20A system controller,
two LC-20AD pump, a SIL-20AC HT autosampler, and a SPD-20A
UV/vis detector (Shimadzu Scientific Instrument, Columbia, MD) and
an automated switching valve (TPMV, Rheodyne, Cotati, CA). The
switching valve was used to divert the column effluent to either waste
or a MS instrument. The Shimadzu Prominence HPLC system was
used for sample injection and separation. Each sample (10−30 μL)
was loaded onto a reverse phase column [Ace 3 C18, 2.1 mm × 100
mm, 3 μm, MAC-MOD Analytical, Chadds Ford, PA] with a water/
acetonitrile gradient at 0.2 mL/min, and the sample components were
eluted into the mass spectrometer. The column chamber's temperature
was set to 45 °C. The mobile phases were water and acetonitrile. The
initial gradient consisted of 0% acetonitrile for 0.5 min followed by a
linear gradient up to 15% acetonitrile over 50 min, and acetonitrile was
increased to 95% in 1.5 min. After holding 95% acetonitrile for 5 min,
the instrument was reset to the initial conditions in 1 min. The
analytical column was equilibrated with the starting mobile phase for
12 min. The total run time for analysis was 70 min.
The HPLC elute was coupled with an AB Sciex 4000 QTrap LC/
MS/MS system (AB Sciex, Foster City, CA), equipped with a Turbo V
ion source and a desolvation temperature of 500 °C. Nitrogen was
used as the curtain gas, nebulizer gas, heater gas, and collision gas. The
samples were acquired in positive IonSpray mode using multiple
reaction monitoring methods (MRM). The ion transitions of the
MRM method for specific detection of DHP-dG and DHP-dA adducts
in the samples were m/z 403/269 and m/z 387/253, respectively. The
IonSpray Voltage was 4000 V; the ion source gas 1 and gas 2 were set
to 60 and 50, respectively. The declustering potential was 50 V, and
the collision energy was 19 eV for all of the transitions.
High-Resolution Mass Spectral Data. The samples were
analyzed on an Agilent Technologies Zorbax C8 column (4.6 mm ×
150 mm), 5 μm on a Shimadzu 20 series LC system with two mobile
phases: A, H2O containing 0.1% formic acid, and B, acetonitrile; eluted
with A for 5 min, then increased from A to 35% B in A in 30 min; flow
rate, 0.7 mL/min. This HPLC system was coupled with an LTQ
Orbitrap mass spectrometer from Thermo Corporation (Waltham,
MA), operated at a resolution of 15000.
HPLC conditions: ACE C18 AR column from Mac-Mod analytical,
Inc., PA, 4.6 mm × 250 mm, 5 μm, monitored at 268 nm, with a flow
rate of 1 mL/min; gradient program: 0−5 min, 10% acetonitrile in
water; 5−50 min, 10−18% acetonitrile in water.
For mechanistic study, pure DHP-dA-3 in acetonitrile/H2O18 (v/v,
1/10) was similarly conducted at room temperature for a period of 6
days. The resulting DHP-dA-4 adduct was collected by HPLC for LC/
MS analysis.
Interconversion between DHP-dG-3 and DHP-dG-4. Approx-
imately 0.5 mL of a solution (acetonitrile/H2O = 1/10) containing
pure DHP-dG-3 (or DHP-dG-4) was placed in a small vial sealed with
parafilm and kept at room temperature. The solution was monitored
by HPLC at 0, 1, 2, 3, 4, 5, 6, and 9 days. The analyses were then
conducted under conditions described above for the interconversion
between DHP-dA-3 and DHP-dA-4.
In a separate experiment, pure DHP-dG-3 in acetonitrile/H2O18 (v/
v, 1/10) was similarly conducted at room temperature for a period of 8
days. The resulting DHP-dG-4 adduct was collected by HPLC for LC/
MS analysis.
Instrumentation. A Waters HPLC system, consisting of a model
600 controller, a model 996 photodiode array detector, and a 600
pump, was used for separation and purification of the DHP-derived
Metabolism of Riddelliine by Rat Liver Microsomes in the
Presence of Calf Thymus DNA. The metabolism of riddelliine by
male Fischer 344 rat liver microsomes was conducted in a 1.0 mL
incubation volume containing 100 mM sodium phosphate buffer (pH
7.6), 5 mM magnesium chloride, 1 mM NADP+, 8 mM glucose 6-
phosphate, 2 units of glucose 6-phosphate dehydrogenase, 0.5 mM
riddelliine, 0.5 mg of purified calf thymus DNA, and 0.5 mg of
microsomal protein at 37 °C for 60 min. After incubation, the reaction
was terminated by cooling with ice−water, then sequentially extracted
with 1.0 mL of phenol, 1.0 mL of phenol/chloroform/isoamyl alcohol
(v/v/v, 25/24/1), and 1.0 mL of chloroform/isoamyl alcohol (v/v,
24/1). The DNA in the aqueous phase was precipitated by adding 0.1
mL of 5 M sodium chloride followed by 2 volumes of cold ethanol and
washed three times with 70% ethanol. After this was redissolved in 300
μL of distilled water, the DNA concentration and purity were analyzed
spectrophotometrically, and DNA was stored at −78 °C prior to LC/
MS analysis.
DNA samples (typically ∼100 μg) were then enzymatically
hydrolyzed to deoxynucleosides with micrococcal nuclease (MN),
spleen phosphodiesterase (SPD), and nuclease P1 as previously
described.1 The resulting samples were injected on the HPLC for LC-
ESI/MS/MS analyses.
LC-MS/MS Analysis of DHP-dG and DHP-dA Adducts
Formed from Organic Synthesis and from Metabolism of
Riddelliine by Male Fischer 344 Rat Liver Microsomes in the
Presence of Calf Thymus DNA. The DHP-dG and DHP-dA
adducts formed from the synthetic reactions were dissolved in water
1
DNA adducts. H NMR experiments were carried out at 301 K on a
Bruker Avance III spectrometer equipped with a Bruker BBFO Plus
Smart Probe (Bruker Instruments, Billerica, MA) operating at 500
MHz. Samples were dissolved in DMSO-d6 or DMSO-d6 with a trace
of D2O. Chemical shifts are reported in parts per million downfield
from tetramethylsilane, and coupling constants are reported in hertz.
Two-dimensional NMR homonuclear decoupling (COSY) and NOE
difference (NOESY) experiments were conducted to assist in assigning
proton resonances. CD spectra were recorded on a J-500A
Spectropolarimeter (Japan Spectroscopic Co. Ltd.).
RESULTS
■
Synthesis of DHP-dA Adducts. We previously reported
that the reaction of DHR with dA generated DHP-dA-1, DHP-
dA-2, DHP-dA-3, and DHP-dA-4 adducts; the yields of these
adducts were only 0.1, 0.1, 0.01, and 0.01%, respectively. In the
present study, the synthesis of DHP-dA adducts was attempted
by reaction of dehydroriddelliine with dA under several
different experimental conditions. The condition that provided
the optimal yields was a reaction of dehydroriddelliine with dA
in the presence of NaHCO3 in DMF at 25 °C for 1 h, followed
by the addition of water and further reaction for another 6 h
1987
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
(Scheme 1). The resulting reaction products were separated by
HPLC, first with a Whatman ODS-3 column to remove excess
dA (data not shown) and then with a Phenomenex C18 Luna 5
μm ODS column (Figure 1A). The materials contained in the
major fragments at m/z 136 (DHP-OH), 253 (loss of
deoxyribose and OH), and 369 (loss of OH) (Figure S2 in
the Supporting Information), suggesting that it is a DHP-
derived dA adduct. This adduct had its HPLC retention time
and mass spectral pattern identical to that of the adduct
previously designated as DHP-dA-3 from the reaction of DHR
and dA, whose structure we had previously been unable to
elucidate by 1H NMR due to an insufficient amount of sample.1
With a sufficient quantity of this adduct, also designated here as
DHP-dA-3, available from the new synthetic approach, its
structure was elucidated by NMR spectral analysis. This adduct
1
had a H NMR resonance for H7 (5.09−5.11 ppm) of the
DHP moiety (the necine base) that was nearly identical to H7
(4.99 ppm) of DHR (Table 1 and Figure 2); likewise, H2, H3,
and both pairs of geminal H5 and H6 protons were almost
1
unaffected by adduct formation. The full H NMR spectral
1
assignments are as follows. DHP-dA-3: H NMR (DMSO-d6):
δ 2.19−2.28 (2H, m, dA-H2′b, H6b), 2.61−2.68 (1H, m, H6a),
2.70−2.75 (1H, m, dA-H2′a), 3.50−3.54 (1H, M, dA-H5′b),
3.60−3.64 (1H, m, dA-H5′a), 3.79−3.84 (1H, m, H5b), 3.87−
3.89 (1H, m, dA-H4′), 3.97−4.02 (1H, m, H5a), 4.42 (1H, m,
dA-H3′), 4.46 (1H,w, H9b), 4.58 (1H, w, H9a), 5.09−5.11
(1H, m, H7), 5.22−5.32 (3H, m, dA-5′-OH, 7-OH, dA-3′−
OH), 6.07 (1H, d, H2, J = 2.5), 6.33−6.36 (1H, t, dA-H1′),
6.55 (1H, d, H3, J = 2.5), 7.82 (1H, bs, dA-N6H), 8.20 (1H, bs,
dA-H8), 8.33 (1H, s, dA-H2).
Each of the deoxyribose protons was clearly evident for the
dA moiety. The structural assignment was further confirmed by
both COSY (Figure S3 in the Supporting Information) and
NOESY (Figure S4 in the Supporting Information) NMR
analysis. As evidenced by the COSY and NOESY 2D NMR
spectral results summarized in Figure S5 in the Supporting
Information, the proton−proton interactions through the bond
and space clearly confirmed that reaction occurred at C9 of
dehydroriddelliine and the exocyclic nitrogen of dA. Thus, on
the basis of the UV−vis, mass, and 1H NMR spectroscopic data,
DHP-dA-3 was characterized as 7-hydroxy-9-(deoxyadenosin-
N6-yl)dehydrosupinidine (Scheme 1).
The material contained in the chromatographic peaks eluted
at 37.9 min (Figure 1A) was similarly analyzed. The high-
resolution electrospray positive product ion mass spectrometric
analysis revealed that this adduct had an observed molecular
ion at m/z 387.1779 (calculated mass at m/z 387.1775,
accuracy of 1.07 ppm), and major fragments at m/z 136 (DHP-
OH), 253 (loss of deoxyribose and OH), and 369 (loss of OH)
(Figure S6 in the Supporting Information). Its HPLC retention
time and mass spectral pattern were identical to those of the
adduct previously designated as DHP-dA-4 from the reaction of
DHR and dA.1 Similar to DHP-dA-3, in our previous study, we
had been unable to obtain a sufficient amount of this adduct for
Figure 1. Reversed-phased HPLC profile of (A) DHP-dA adducts
formed from reaction of dA with dehydroriddelliine using a
Phenomenex C18 Luna (2) column (4.6 mm × 250 mm, 5 μm)
eluted at 1 mL/min; elution conditions: 0−5 min, 10% acetonitrile in
water; 5−50 min, 10−18% acetrontrile in water; and (B) DHP-dG
adducts formed from reaction of dehydroriddelliine with dG; HPLC
conditions: Luna C18 (2) column (4.6 mm × 250 4.6 mm, 5 μm);
flow rate, 1 mL/min, monitored at 256 nm; gradient: 0′−5′, 8%
acetonitrile in water; 5′−50′, 8−16% acetonitrile in water.
1
structural identification. The complete H NMR assignment of
chromatographic peaks eluted at 31.7 and 33.3 min exhibited
HPLC retention times (Figure 1A), UV−vis absorption spectra
(Figure S1 in the Supporting Information), and LC/MS spectra
(data now shown) identical to those of DHP-dA-1 and DHP-
dA-2, respectively, previously identified as a pair of epimeric
1
DHP-dA-4 is as follows: H NMR (DMSO-d6): δ 2.22−2.31
(2H, m, dA-H2′b, H6b), 2.64−2.71 (1H, m, H6a), 2.73−2.78
(1H, m, dA-H2′a), 3.55−3.57 (1H, M, dA-H5′b), 3.64−3.67
(1H, m, dA-H5′a), 3.82−3.87 (1H, m, H5b), 3.90−3.92 (1H,
m, dA-H4′), 4.00−4.05 (1H, m, H5a), 4.44−4.45 (1H, m, dA-
H3′), 4.49 (1H, w, H9b), 4.62 (1H, w, H9a), 5.12−5.13 (1H,
m, H7), 5.27−5.35 (3H, m, dA-5′−OH, 7-OH, dA-3′-OH),
6.09 (1H, d, H2, J = 2.0), 6.36−6.39 (1H, t, dA-H1′), 6.58 (1H,
d, H3, J = 2.0), 7.84 (1H, bs, dA-N6H), 8.24 (1H, bs, dA-H8),
DHP-dA adducts
(
)7-(deoxyadenosin-N6 -yl)-
dehydrosupinidine (Scheme 1) synthesized by reaction of
DHR and dA.1
High-resolution electrospray positive product ion mass
spectrometric analysis revealed that the material contained in
the chromatographic peaks eluting at 37.1 min (Figure 1A) had
a protonated molecule ion (M + H)+ at m/z 387.1778
(calculated mass at m/z 387.1775, accuracy of 0.79 ppm) and
1
8.36 (1H, s, dA-H2). The H NMR spectrum of DHP-dA-4 is
similar to that of DHP-dA-3 (Table 1 and Figure 2), suggesting
that it is also 7-hydroxy-9-(deoxyadenosin-N6-yl)-
1988
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Table 1. Proton NMR Spectroscopic Data (500 MHz) of the Necine Base in the Synthetically Prepared DHP-dA and DHP-dG
Adducts Measured in DMSO-d6
adduct
assignment
H6b
DHR1
DHP-dA-3
DHP-dA-4
DHP-dG-3
DHP-dG-4
DHP-dG-5
2.15−2.20 (1H, m)
2.19−2.28 (1H,
2.22−2.31 (1H,
2.18−2.25 (1H, m)
2.21−2.27 (1H, m)
2.15−2.24 (1H, m)
m)
m)
H6a
H5b
H5a
H9b
H9a
H7
2.57−2.65 (1H, m)
3.75−3.79 (1H, m)
3.94−3.98 (1H, m)
4.27 (1H, d, J = 12)
4.33 (1H, d, J = 12)
2.61−2.68 (1H,
2.64−2.71 (1H,
2.55−2.66 (1H, m)
3.78−3.83 (1H, m)
3.98−4.03 (1H, m)
2.63−2.72 (1H, m)
3.83−3.88 (1H, m)
4.02−4.07 (1H, m)
2.58−2.74 (1H, m)
3.78−3.84 (1H, m)
3.98−4.03 (1H, m)
m)
m)
3.79−3.84
3.82−3.87 (1H,
(1H,m)
m)
3.97−4.02 (1H,
4.00−4.05 (1H,
m)
m)
4.46 (1H, w)
4.58 (1H, w)
4.49 (1H, w)
4.62 (1H, w)
4.24 (1H, two sets of d,
J = 14)
4.28 (1H, two sets of d,
J = 14)
4.60 (1H, d, H9b,
J = 15.5)
4.32 (1H, two sets of d)
4.37 (1H, two sets of d)
5.33 (1H, d, H9a,
J = 15.5)
4.99 (1H, dd, J = 6.5, 5.09−5.11 (1H,
5.12−5.13 (1H,
5.03−5.08 (1H, m)
5.08−5.10 (1H, m)
5.20−5.22 (1H, m)
5.98 (1H, d, J = 2.5)
6.58 (1H, d, J = 2.5)
2.5)
m)
m)
H2
6.02 (1H, d, J = 2.5)
6.07 (1H, d,
J = 2.5)
6.09 (1H, d,
J = 2.0)
6.07 (1H, d, J = 2.5)
6.60 (1H, d, J = 2.5)
6.12 (1H, d, J = 2.5)
6.63 (1H, d, J = 2.5)
H3
6.54 (1H, d, J = 2.6)
6.55 (1H, d,
J = 2.5)
6.58 (1H, d,
J = 2.0)
Figure 3. CD spectra of (A) DHP-dA-3 and DHP-dA-4, both adjusted
to 1.5 OD at 270 nm, and (B) DHP-dG-3 and DHP-dG-4, both
adjusted to 0.85 OD at 256 nm, in aqueous solution (acetonitrile/
water = 1/10). The CD spectrometer parameters were 2 m°/cm
sensitivity, 1 cm/min chart speed.
Figure 2. 1H NMR spectra of DHP-dA-3 and DHP-dA-4 measured in
DMSO-d6.
dehydrosupinidine. For further structural confirmation, both
COSY (Figure S7 in the Supporting Information) and NOESY
(Figure S8 in the Supporting Information) spectra were
obtained and analyzed. As shown in Figure S5 in the
Supporting Information, the COSY and NOESY spectra of
DHP-dA-4 and DHP-dA-3 are well correlated.
CD spectral analysis confirmed that DHP-dA-3 and DHP-dA-4
are a pair of epimers differing in the configuration of one
stereocenter in the DHP chromophore (Scheme 1).
Synthesis of DHP-dG Adducts. The four DHP-dG
adducts, assigned as DHP-dG-1, DHP-dG-2, DHP-dG-3, and
DHP-dG-4, were previously synthesized from reaction of DHR
and dG, and their structures were determined by UV−visible
absorption, mass spectrometry, and 1H NMR spectral analysis.1
The yields of these DHP-dG adducts obtained from this
reaction were DHP-dG-1, 0.89%; DHP-dG-2, 0.69%; DHP-dG-
3, 0.15%; and DHP-dG-4, 0.06%, respectively.
Given that only very small amounts of DHP-dG-3 and DHP-
dG-4 were previously obtained for 1H NMR structural analysis,
we felt that their structural assignments warranted validation.
We have now developed an alternative synthetic method,
reaction of dehydroriddelliine with dG, to obtain sufficient
amounts to validate the structural assignment of these two
adducts. After the DHP-dG adducts were synthesized by
reaction of dehydroriddelliine with dG (Scheme 2), the
On the basis of the similarity of the UV spectra and assuming
that the molar extinction coefficients of DHP-dA-3 and DHP-
dA-4 are identical to those of DHP-dA-1 and DHP-dA-2, the
reaction yields of the DHP-dA adducts from the reaction of
riddelliine and dA are as follows: DHP-dA-1, 0.64%; DHP-dA-
2, 0.47%; DHP-dA-3, 2.0%; and DHP-dA-4, 1.98% (Scheme 1).
The yields of DHP-dA-3 and DHP-dA-4 are approximately
200-fold higher than those from the reaction of DHR and dA.
To investigate the stereochemistry, CD spectra of DHP-dA-3
and DHP-dA-4 were measured. As shown in Figure 3A, the CD
Cotton effects of DHP-dA-3 were approximate mirror images
of those of DHP-dA-4. It is noteworthy that the CD Cotton
effects of DHP-dA-3 and DHP-dA-4 are much weaker than
those of DHP-dA-1 and DHP-dA-2.1 The overall results of
UV−vis, LC/MS, 1D and 2D NMR (COSY and NOESY), and
1989
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Scheme 2. Synthesis of DHP-dG Adducts from Reaction of Dehydroriddelliine with dG
resulting reaction products were separated by HPLC, first with
a Whatman ODS-3 column to remove excess dG (data not
shown) and then with a Phenomenex C18 Luna 5 μm ODS
column (Figure 1B). The chromatographic peaks eluting at
25.8, 27.7, 29.1, 30.8, and 31.5 min (Figure 1B) had their UV−
vis absorption spectra and mass spectra with identical
protonated molecules (M + H)+ at m/z 403, closely similar
to those previously reported for DHP-dG-1, DHP-dG-2, DHP-
dG-3, and DHP-dG-4.1 On the basis of their HPLC profile
(deoxyguanosin-N2-yl)dehydrosupinidine. This confirms that
the previous structure assignment is indeed correct. However,
we determined that the full proton 1H NMR spectrum of DHP-
1
dG-4 adduct should be assigned as follows: DHP-dG-4: H
NMR (DMSO-d6): δ 2.21−2.27 (2H, m, dG-H2′b, H6b),
2.63−2.72 (2H, m, dG-H2′a, H6a), 3.50−3.60 (2H, m, dG-
H5′), 3.83−3.88 (2H, m, dG-H4′, H5b), 4.02−4.07 (1H, m,
H5a), 4.28 (1H, H9, two set of d, J = 14 Hz), 4.37 (1H, H9a,
two set of d, J = 14 Hz), 4.41 (1H, m, dG-H3′), 5.03 (1H, w,
OH), 5.08−5.10 (1H, m, H7), 5.34 (1H, m, OH), 6.12 (1H, d,
H2, J = 2.5), 6.23 (1H, m, dG-H1′), 6.63 (1H, d, H3, J = 2.5),
7.92 (1H, s, dG-H8), 8.57 (1H, s, OH/NH), 10.72 (1H, bs,
dG-NH1). On the basis of this full proton NMR assignment
and the NMR data (Table 1) and shown in Figure 4, the NMR
spectral data of DHP-dG-4 are similar to those of DHP-dG-3.
Analysis of the COSY NMR spectrum (Figure S11 in the
Supporting Information) and the NOESY NMR spectrum
(Figure S12 in the Supporting Information) also indicated that
the structure of the DHP-dG-4 adduct should also be 7-
hydroxy-9-(deoxyguanosin-N2-yl)dehydrosupinidine. The CD
spectra of DHP-dG-3 and DHP-dG-4 adducts (Figure 3B)
further validated that they are a pair of epimeric DHP-dG
adducts with the asymmetric center at the C7 position of the
necine base (Scheme 2). Thus, the previous assignment of
DHP-dG-4 as 7-hydroxy-5-(deoxyguanosin-N2-yl)-
dehydrosupinidine1 is incorrect.
1
(Figure 1B), UV−visible absorption spectra, H NMR spectra
(data not shown), and mass spectral patterns, the chromato-
graphic peaks eluting at 27.7, 29.1, 30.8, and 31.5 min were
identified as the previously determined DHP-dG-1, DHP-dG-2,
DHP-dG-3, and DHP-dG-4 adducts.1 On the basis of 1H NMR
(Figure 4) and 2D NMR (COSY and NOESY) spectral analysis
(Figure S9 and S10 in the Supporting Information), we
confirmed that the structure of DHP-dG-3 is 7-hydroxy-9-
An additional DHP-dG adduct, assigned as 7-hydroxy-9-
(deoxyguanosin-7N-yl)dehydrosupinidine (DHP-dG-5), was
also obtained from this new synthetic approach (Figure 1B).
Its mass spectrum had a protonated molecule (M + H)+ at m/z
403, identical to those of DHP-dG-1, DHP-dG-2, DHP-dG-3,
1
and DHP-dG-4. On the basis of its mass, H NMR (Figure 5
and Table 1), 2D COSY NMR (Figure S13 in the Supporting
Information), and spectral analyses (Figure S14 in the
Supporting Information), this adduct was assigned as DHP-
dG-5. The proton NMR assignment of DHP-dG-5 is as follows:
DHP-dG-5: H NMR (DMSO-d6): δ 2.15−2.24 (2H, m, dG-
H2′b, H6b), 2.58−2.74 (2H, m, dG-H2′a, H6a), 3.46−3.55
1
Figure 4. 1H NMR spectra of DHP-dG-3 and DHP-dG-4 measured in
DMSO-d6. The absolute stereochemistries depicted at C7 are arbitrary.
1990
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
dA-3 was converted into DHP-dA-4. After 6 days at room
temperature, DHP-dA-3 and DHP-dA-4 reached an approx-
imately equal ratio (Figure 6). Similar results were obtained
starting with pure DHP-dA-4 (Figure S15 in the Supporting
Information; Figure 6). It is noteworthy that in this 6 day
interconversion study, neither DHP-dA-3 nor DHP-dA-4
decomposed. In contrast, interconversion did not occur
between DHP-dA-1 and DHP-dA-2 (data not shown).
Interconversion between DHP-dG-3 and DHP-dG-4
Adducts. We found that DHP-dG-3 and DHP-dG-4 in
acetonitrile/water (v/v, 1/10) are interconvertible at room
temperature. As analyzed by HPLC, DHP-dG-3 was converted
into DHP-dG-4. After 9 days, DHP-dG-3 and DHP-dG-4
reached an equal ratio (Figure 6). Similar results were obtained
starting with pure DHP-dG-4 (Figure 6). Different from the
DHP-dA adducts, during the 9 day interconversion study, a
small part, less than 5% of DHP-dG-3 and DHP-dG-4,
decomposed. In contrast, interconversion did not occur
between DHP-dG-1 and DHP-dG-2 (data not shown).
Mechanism of Interconversion between DHP-dA-3
and DHP-dA-4 Adducts. For determining the mechanism by
which DHP-dA-3 interconverts to DHP-dA-4 adduct, DHP-dA-
3 dissolved in acetonitrile/H2O18 (v/v, 1/10) was kept at room
temperature for 6 days, and then, the resulting DHP-dA-4 was
collected by HPLC for LC/MS analysis. The mass spectrum of
the unlabeled DHP-A-4 (Figure 7A) had the protonated
molecular ion at m/z 387 and the characteristic ions at m/z 369
([M − H2O + H]+), m/z 253, and m/z 136, respectively. The
mass spectrum of the [O18]DHP-dA-4 (Figure 7B) had the
protonated molecular ion at m/z 389 and the characteristic ions
at m/z 369 ([M − H2O18 + H]+), m/z 253, and m/z 138,
respectively. Comparison of the molecular ions and the mass of
these fragment ions as well as the assigned structures shown in
1
Figure 5. H NMR spectra of DHP-dG-5 measured in DMSO-d6.
(2H, m, dG-H5′), 3.78−3.84 (2H, m, dG-H4′, H5b), 3.98−
4.03 (1H, m, H5a), 4.32−4.34 (1H, m, dG-H3′), 4.60 (1H, d,
H9b, J = 15.5), 4.93 (1H, w, NH/OH), 5.20−5.22 (1H, m,
H7), 5.28 (1H, w, NH/OH), 5.33 (1H, d, H9a, J = 15.5), 5.63
(1H, w, NH/OH), 5.98 (1H, d, H2, J = 2.5), 6.09−6.12 (1H,
m, dG-H1′), 6.58 (1H, d, H3, J = 2.5), 6.95 (2H, b, NH2), 7.94
(1H, s, dG-H8). It is noteworthy that, different from other
deoxyguanosin-7N-yl adducts that undergo spontaneous
depurination, this adduct was highly stable to the point that
it could be isolated by HPLC, taken to dryness, and evaluated
by NMR without depurination.
Interconversion between DHP-dA-3 and DHP-dA-4
Adducts. We found that DHP-dA-3 and DHP-dA-4 in
acetonitrile/water (v/v, 1/10) are interconvertible. As analyzed
by HPLC (Figure S15 in the Supporting Information), DHP-
Figure 6. Interconversion between DHP-dA-3 and DHP-dA-4 in acetontrile/water (v/v, 1/10) starting from (A) pure DHP-dA-3 and (B) pure
DHP-dA-4 and interconversion between DHP-dG-3 and DHP-dG-4 in acetontrile/water (v/v, 1/10) starting from (C) pure DHP-dG-3 and (D)
pure DHP-dG-4. Data presented in this figure are means SDs from three experiments. The relative percentages were calculated from their HPLC
peak area ratios.
1991
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Figure 7. The LC/MS profiles of (A) DHP-dA-4, (B) [O18]DHP-dA-4, (C) DHP-dG-4, and (D) [O18]DHP-dG-4.
Figure 7A,B indicates that the O18 atom is specifically labeled at
the 7-hydroxyl group of the necine base.
(Figure 7D) had the protonated molecular ion at m/z 405
and the characteristic ions at m/z 385 ([M − H2O18 + H]+),
m/z 289, m/z 269, m/z 152, and m/z 118, respectively.
Comparison of their mass spectra (Figure 7C,D) indicates that
the O18 atom is specifically labeled at the 7-hydroxyl group of
the necine base.
HPLC-ES-MS/MS Analysis of DHP-dG and DHP-dA
Adducts Formed from Microsomal Metabolism of
Riddelliine in the Presence of Calf Thymus DNA. The
DHP-dG and DHP-dA adducts formed in vitro were assessed
by HPLC-ES-MS/MS through multiple reaction monitoring
Mechanism of Interconversion between DHP-dG-3
and DHP-dG-4 Adducts. DHP-dG-3 dissolved in acetoni-
trile/H2O18 (v/v, 1/10) at room temperature was similarly
conducted for 9 days, and the resulting [O18]DHP-dG-4 was
collected by HPLC for LC/MS analysis. The mass spectrum of
the unlabeled DHP-dG-4 (Figure 7C) had the protonated
molecular ion at m/z 403 and the characteristic ions at m/z 385
([M − H2O + H]+), m/z 287, m/z 269, m/z 152, and m/z 118
respectively. The mass spectrum of the [O18]DHP-dG-4
1992
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Figure 8. LC-ESI/MS/MS profiles of DHP-dG and DHP-dA adducts. (A) Chromatogram obtained using synthetic standards and (B)
chromatogram obtained from in vitro incubation of DNA with riddelliine in the presence of rat liver microssomes.
Figure 9. Proposed general mechanism leading to DHP-derived DNA adduct formation from the metabolism of riddelliine in vivo and in vitro.
(MRM). The MRM profile of the five DHP-dG adducts and
the four DHP-dA adducts standards is shown in Figure 8A, and
the MRM profile of the adducts formed from the metabolism is
shown in Figure 8B, respectively. As the yields calculated based
on their molar extinction coefficients and HPLC peak areas,
these results indicate that (i) DHP-dG-1, DHP-dG-2, DHP-
dG-3, and DHP-dG-4 adducts were formed, of which DHP-dG-
1 and DHP-dG-2 adducts were formed to a similar extent; (ii)
DHP-dG-5 adduct was not formed; (iii) DHP-dA-1, DHP-dA-
2, DHP-dA-3, and DHP-dA-4 adducts were formed predom-
inantly, while DHP-dA-1 and DHP-dA-2 adducts were formed
to a similar extent; and (iv) among all of these adducts, DHP-
dA-3 and DHP-dA-4 were the most abundant, followed by
DHP-dG-3 and DHP-dG-4 adducts.
DISCUSSION
■
In the present study, we successfully synthesized DHP-dA-3,
DHP-dA-4, DHP-dG-3, and DHP-dG-4 by reaction of
1993
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
dehydroriddelliine with dG and dA, respectively, for structural
identification. On the basis of mass spectral, 1D H NMR, 2D
Thus, these results indicate that dehydroriddelliine preferen-
tially binds to the exocyclic amino group, but not to the N7
position, of the dG nucleic base of the cellular DNA or the
isolated calf thymus DNA.
1
COSY NMR, and NOESY NMR spectral analysis, we
unequivocally determined that DHP-dA-3 and DHP-dA-4 are
a pair of epimers, ( )7-hydroxy-9-(deoxyadenosin-N6-yl)-
dehydrosupinidine (Scheme 1). With the same approach, we
also determined that DHP-dG-3 and DHP-dG-4 are a pair of
epimers ( )9-(deoxyguanosin-N2-yl)dehydrosupinidine
(Scheme 2). These structural assignments were further
validated by the observation that the CD spectra of DHP-dA-
3 and DHP-dA-4 are mirror images (Figure 3A) and that the
CD spectra of DHP-dG-3 and DHP-dG-4 are also mirror
images (Figure 3B). Thus, all of these DNA adducts were
formed by the attack of the exocyclic amino group (of dG and
dA) to the C9 position of the dehydroriddelliine necine base.
The results indicate that the reaction of dehydroriddelliine with
dG and dA is highly regioselective, with predominant binding at
the C9 position of the necine base, which is in contrast to the
reaction of DHR with dG and dA, predominantly producing
DHP-dG-1, DHP-dG-2, DHP-dA-1, and DHP-dA-2, formed
through the binding at the C7 position of the necine base.1
There are two possible pathways that lead to DHP-derived
DNA adducts from metabolism of riddelliine in vivo and in
vitro.17 The first pathway is covalent binding of the
dehydroriddelliine metabolite to cellular DNA, forming
dehydroriddelliine-derived DNA adducts, which are subse-
quently hydrolyzed to DHP-derived DNA adducts. The second
pathway is hydrolysis of dehydroriddelliine, catalyzed by
esterases, to DHP,5,34−36 which subsequently binds to DNA17
(Figure 9). Because dehydropyrrolizidine alkaloid (pyrrole)
metabolites are highly unstable and DHP is the most stable
pyrrolic metabolite, we and other investigators6,17,36 previously
believed that these DHP-derived DNA adducts should be
preferentially formed by the second pathway rather than the
first pathway. Because we found that DHP-dG-3, DHP-dG-4,
DHP-dA-3, and DHP-dA-4 are the DNA adducts formed in
vivo,1 the first pathway should be predominant. On the basis of
our present findings, the metabolic activation pathway of
riddelliine leading to DHP-derived DNA adduct formation is
hypothesized to be that illustrated in Figure 9. The metabolism
of riddelliine catalyzed by P450 enzymes first generates 3-
hydroxyriddelliine and/or 8-hydroxyriddelliine, which upon
enzymatic dehydration produces dehydroriddelliine. Dehydror-
iddelliine possesses two alkylating positions at the C7 and C9
positions of the necine base capable of electrophillic reaction
with cellular DNA bases in vivo. We determined that the
reaction is highly regioselective, with the cellular DNA
predominantly attacking to the C9 position, leading to the
formation of C9-substituted DHP-DNA adducts, DHP-dG-3,
DHP-dG-4, DHP-dA-3, and DHP-dA-4 (Figure 9). The
secondary carbonium ion intermediate formed at the C7
position is more reactive than the primary carbonium ion
formed at the C9 position of the necine base, and thus, it
should preferentially form. The preferential regioselective
reaction that occurred at the C9 position is apparently due to
steric hindrance since the C9 position is less hindered than the
C7 position.1 Alternatively, it is also possible that the
conformation of the bases in the double helix determines the
positions available to the nescine base.
As the structure of DHP-dG-5 was firmly determined by
mass, 1H NMR, and 2D COSY NMR spectra as DHP-dG-5, its
high stability is quite different from other deoxyguanosin-7N-yl
adducts that undergo spontaneous depurination. The reason for
this stability warrants further investigation.
Dehydropyrrolizidine alkaloids, the primary pyrrolic metab-
olites, are highly unstable and biologically/chemically reactive
and thus are unable to be isolated from any biological systems.
Although their bindings with cellular DNA and proteins that
form DNA adducts, protein adducts, and DNA−protein cross-
linking adducts are responsible for hepatotoxicity, mutagenicity,
and tumorigenicity of pyrrolizidine alkaloids,3,6,35,38,39 the
structures of these resulting adducts have never been fully
elucidated. Our present study represents the first report to
assign unequivocally the detailed chemical structures of the
DNA adducts that are responsible for the etiology of
pyrrolizidine alkaloid-induced tumors. Many published reports,
including ours, indicate that the most reactive site of
dehydropyrrizidine alkaloid metabolites for binding with
cellular constituents is the C7 position of the necine base.6,17
For instance, we mistakenly synthetically prepared 3′-dGMP-
DHP adducts, with the linking at the C7-position of DHP, as
the authentic standards for identification and quantification of
DHP-derived DNA adducts by 32P-postlabeling/HPLC anal-
ysis.17−20,24−26,30,31,40 In our previous report using LC-ESI/
MS/MS for DNA adducts analysis, we also focused on the
synthesis of C7-substituted DNA adduct standards, for
example, DHP-dG-1, DHP-dG-2, DHP-dA-1, and DHP-dA-
2.1 However, our present and previous studies1 clearly show
that cellular DNA, calf thymus DNA (Figure 8), dG, and dA
preferentially bind to dehydropyrrizidine alkaloid, for example,
dehydroriddelliine, at the C9 position of the necine base. The
drastic different regioselective reaction of these nucleophiles
with DHP at the C7 position and with dehydropyrrolizidine
alkaloids at the C9 position is hypothesized to be mainly
because of steric hindrance at the C7 position. At present, it is
not known how cellular proteins or single amino acids bind to
dehydropyrrolizidine alkaloids. The only published study
related to this matter is the binding of DHR with cysteine
and glutathione.41,42 Thus, further investigation is warranted.
We have also determined that DHP-dA-3 and DHP-dA-4 are
interconvertible in acetonitrile/water (v/v, 1/10) (Figure 6).
DHP-dA-3 was converted into DHP-dA-4 and from DHP-dA-4
to DHP-dA-3, to reach an equal ratio in 6 days without
decomposition (Figure 6). Similar interconversion occurred
between DHP-dG-3 and DHP-dG-4, although the adducts
decomposed to a small extent (Figure 6). We also found that in
the presence of [O18]H2O, a fraction of the hydroxyl groups at
the C7 position contained [O18]OH, indicating that during
interconvension, the OH group at the C7 position was removed
from the molecule. On the basis of our findings, we propose the
mechanism depicted in Scheme 3. This reaction scheme is
similar to that proposed by ref 36 on the reaction of
nucleophiles with the dehydropyrrolizidine alkaloids (pyrrolic
metabolites), particularly the diester dehydropyrrolizidine
alkaloids. While an ester group at the C7 position of the
necine is a good leaving group on the nucleophilic reaction, the
hydroxyl group at the C7 position is a much weaker leaving
group. This is consistent with the interconversion of the DHP-
We have also found that DHP-dG-5 was not detected in the
NTP liver sample (rats dosed with riddelliine in vivo) (data not
shown) or from the incubation of riddelliine with rat liver
microsomes in the presence of calf thymus DNA (Figure 8).
1994
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Scheme 3. Proposed Mechanism of Interconversion between
DHP-dA-3 and DHP-dA-4 and between DHP-dG-3 and
DHP-dG-4
ACKNOWLEDGMENTS
■
We thank Dr. Po-Chuen Chan of NTP for supplying riddelliine
and Frederick A. Beland for critical review of this manuscript.
ABBREVIATIONS
■
CD, circular dichroism spectroscopy; DHR, dehydroretrone-
cine or (−)-R-6,7-dihydro-7-hydroxy-1-hydroxymethyl-5H-pyr-
rolizine); DHP, ( )-6,7-dihydro-7-hydroxy-1-hydroxymethyl-
5H-pyrrolizine; PNK, cloned T4 polynucleotide kinase; MN,
micrococcal nuclease; SPD, spleen phosphodiesterase; dG, 2′-
deoxyguanosine; dA, 2′-deoxyadenosine; DHP-dG-1 and DHP-
dG-2, epimeric pairs of 7-(deoxyguanosin-N2-yl)-
dehydrosupinidine; DHP-dG-3 and DHP-dG-4, epimeric pairs
of 7-hydoxy-9-(deoxyguanosin-N2-yl)dehydrosupinidine; DHP-
dG-5, 7-hydroxy-9-(deoxyguanosin-7N-yl)dehydrosupinidine;
DHP-dA-1 and DHP-dA-2, epimeric pairs of 7-(deoxyadeno-
sin-N6-yl)dehydrosupinidine; DHP-dA-3 and DHP-dA-4, 7-
hydroxy-9-(deoxyadenosin-N6-yl)-dehydrosupinidine; LC-ESI-
MS/MS, high-performance liquid chromatography electrospray
ionization tandem mass spectrometry; NCTR, National Center
for Toxicological Research; NTP, National Toxicology
Program
REFERENCES
■
(1) Fu, P. P., Chou, M. W., Churchwell, M., Wang, Y., Zhao, Y., Xia,
Q., Gamboa da Costa, G., Marques, M. M., Beland, F. A., and Doerge,
D. R. (2010) High-performance liquid chromatography electrospray
ionization tandem mass spectrometry for the detection and
quantitation of pyrrolizidine alkaloid-derived DNA adducts in vitro
and in vivo. Chem. Res. Toxicol. 23, 637−652.
dA and DHP-dG adducts requiring 6 and 9 days, respectively,
to reach an equal ratio of both epimers, DHP-dA-3 and DHP-
dA-4, as well as DHP-dG-3 and DHP-dG-4 (Scheme 3).
In summary, we have fully elucidated the structures of DHP-
dG-3, DHP-dG-4, DHPdA-3, and DHP-dA-4. They are the
DNA adducts formed in vivo, and the level of their formation
correlates with the tumor potency in liver of rats administered
with riddelliine. Thus, we propose that these DNA adducts are
potential DNA biomarkers of pyrrolizidine alkaloid-induced
tumorigenicity as well as pyrrolizidine alkaloid exposure.
(2) Fu, P. P., Xia, Q., Chou, M. W., and Lin, G. (2007) Detection,
hepatotoxicity, and tumorigenicity of pyrrolizidine alkaloids in Chinese
herbal plants and herbal dietary supplements. J. Food Drug Anal. 15,
400−415.
(3) Fu, P. P., Xia, Q., Lin, G., and Chou, M. W. (2004) Pyrrolizidine
alkaloidsGenotoxicity, metabolism enzymes, metabolic activation,
and mechanisms. Drug Metab. Rev. 36, 1−55.
(4) Fu, P. P., Yang, Y.-C., Xia, Q., Chou, M. W., Cui, Y., and Lin, G.
(2002) Pyrrolizidine alkaloidsTumorigenic components in Chinese
herbal medicines and dietary supplements. J. Food Drug Anal. 10, 198−
211.
(5) IPCS (1989) Pyrrolizidine Alkaloids Health and Safety Guides.
Health and Safety Criteria Guides 26, WHO, Geneva, Switzerland.
(6) Mattocks, A. R. (1986) Chemistry and Toxicology of Pyrrolizidine
Alkaloids, Academic Press, London, NY.
ASSOCIATED CONTENT
* Supporting Information
Figures S1−S15. This material is available free of charge via the
■
S
Internet at http://pubs.acs.org.
(7) Roeder, E. (1995) Medicinal plants in Europe containing
pyrrolizidine alkaloids. Pharmazie 50, 83−98.
AUTHOR INFORMATION
Corresponding Author
*Tel: +1-870-543-7207. Fax: +1-870-543-7136. E-mail: peter.
fu@fda.hhs.gov.
■
(8) Roeder, E. (2000) Medicinal plants in China containing
pyrrolizidine alkaloids. Pharmazie 55, 711−726.
(9) Roeder, E., and Wiedenfeld, H. (2009) Pyrrolizidine alkaloids in
medicinal plants of Mongolia, Nepal and Tibet. Pharmazie 64, 699−
716.
Funding
(10) Smith, L. W., and Culvenor, C. C. (1981) Plant sources of
hepatotoxic pyrrolizidine alkaloids. J. Nat. Prod. 44, 129−152.
(11) Betz, J. M., Eppley, R. M., Taylor, W. C., and Andrzejewski, D.
(1994) Determination of pyrrolizidine alkaloids in commercial
comfrey products (Symphytum sp.). J. Pharm. Sci. 83, 649−653.
(12) Edgar, J. A., Roeder, E., and Molyneux, R. J. (2002) Honey from
plants containing pyrrolizidine alkaloids: a potential threat to health. J.
Agric. Food Chem. 50, 2719−2730.
This research was supported in part by appointment (Y.Z.) to
the Postgraduate Research Program at the NCTR administered
by the Oak Ridge Institute for Science and Education through
an interagency agreement between the U.S. Department of
Energy and the FDA.
Notes
This article is not an official U.S. Food and Drug
Administration (FDA) guidance or policy statement. No
official support or endorsement by the U.S. FDA is intended
or should be inferred.
(13) Huxtable, R. J., Ed. (1989) Human Health Implications of
Pyrrolizidine Alkaloids and Herbs Containing them, Vol. 1, CRC Press
Inc., Boca Raton, FL.
(14) Stegelmeier, B. L., Edgar, J. A., Colegate, S. M., Gardner, D. R.,
The authors declare no competing financial interest.
Schoch, T. K., Coulombe, R. A., and Molyneux, R. J. (1999)
1995
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996

Chemical Research in Toxicology
Article
Pyrrolizidine alkaloid plants, metabolism and toxicity. J. Nat. Toxins 8,
95−116.
(33) Zhao, Y., Xia, Q., Yin, J. J., Lin, G., and Fu, P. P. (2011)
Photoirradiation of dehydropyrrolizidine alkaloids-Formation of
reactive oxygen species and induction of lipid peroxidation. Toxicol.
Lett. 205, 302−309.
(34) IARC (1976) Pyrrolizidine Alkaloids. In IARC Monograph on the
Evaluation of Carcinogenic Risk of Chemicals to ManSome Naturally
Occurring Substance, Vol. 10, International Agency for Research in
Cancer, Lyon, France.
(35) Kim, H. Y., Stermitz, F. R., Li, J. K., and Coulombe, R. A., Jr.
(1999) Comparative DNA cross-linking by activated pyrrolizidine
alkaloids. Food Chem. Toxicol. 37, 619−625.
(36) Mattocks, A. R. (1968) Toxicity of pyrrolizidine alkaloids.
Nature 217, 723−728.
(15) Cook, J. W., Duffy, E., and Schoental, R. (1950) Primary liver
tumours in rats following feeding with alkaloids of Senecio jacobaea.
Br. J. Cancer 4, 405−410.
(16) Schoental, R., head, M. A., and Peacock, P. R. (1954) Senecio
alkaloids: Primary liver tumours in rats as a result of treatment with
(1) a mixture of alkaloids from S. jacobaea lin.; (2) retrorsine; (3)
isatidine. Br. J. Cancer 8, 458−465.
(17) Yang, Y. C., Yan, J., Doerge, D. R., Chan, P. C., Fu, P. P., and
Chou, M. W. (2001) Metabolic activation of the tumorigenic
pyrrolizidine alkaloid, riddelliine, leading to DNA adduct formation
in vivo. Chem. Res. Toxicol. 14, 101−109.
(37) Galloway, S. M., Armstrong, M. J., Reuben, C., Colman, S.,
Brown, B., Cannon, C., Bloom, A. D., Nakamura, F., Ahmed, M., Duk,
S., et al. (1987) Chromosome aberrations and sister chromatid
exchanges in Chinese hamster ovary cells: Evaluations of 108
chemicals. Environ. Mol. Mutagen. 10 (Suppl. 10), 1−175.
(38) Hincks, J. R., Kim, H. Y., Segall, H. J., Molyneux, R. J., Stermitz,
F. R., and Coulombe, R. A. (1991) DNA cross-linking in mammalian
cells by pyrrolizidine alkaloids: structure-activity relationships. Toxicol.
Appl. Pharmacol. 111, 90−98.
(18) Chou, M. W., Jian, Y., Williams, L. D., Xia, Q., Churchwell, M.,
Doerge, D. R., and Fu, P. P. (2003) Identification of DNA adducts
derived from riddelliine, a carcinogenic pyrrolizidine alkaloid. Chem.
Res. Toxicol. 16, 1130−1137.
(19) Chou, M. W., Yan, J., Nichols, J., Xia, Q., Beland, F. A., Chan, P.
C., and Fu, P. P. (2003) Correlation of DNA adduct formation and
riddelliine-induced liver tumorigenesis in F344 rats and B6C3F(1)
mice. Cancer Lett. 193, 119−125.
(20) Xia, Q., Chou, M. W., Kadlubar, F. F., Chan, P. C., and Fu, P. P.
(2003) Human liver microsomal metabolism and DNA adduct
formation of the tumorigenic pyrrolizidine alkaloid, riddelliine.
Chem. Res. Toxicol. 16, 66−73.
(21) Chan, P. C., Haseman, J. K., Prejean, J. D., and Nyska, A. (2003)
Toxicity and carcinogenicity of riddelliine in rats and mice. Toxicol.
Lett. 144, 295−311.
(22) National Toxicology Program (2011) NTP 12th Report on
Carcinogens. Rep. Carcinog. 12, iii−499.
(23) Chou, M. W., and Fu, P. P. (2006) Formation of DHP-derived
DNA adducts in vivo from dietary supplements and chinese herbal
plant extracts containing carcinogenic pyrrolizidine alkaloids. Toxicol.
Ind. Health 22, 321−327.
(24) Wang, Y. P., Fu, P. P., and Chou, M. W. (2005) Metabolic
activation of the tumorigenic pyrrolizidine alkaloid, retrorsine, leading
to DNA adduct formation in vivo. Int. J. Environ. Res. Public Health 2,
74−79.
(39) Kim, H. Y., Stermitz, F. R., and Coulombe, R. A. (1995)
Pyrrolizidine alkaloid-induced DNA-protein cross-links. Carcinogenesis
16, 2691−2697.
(40) Fu, P. P., Chou, M. W., Xia, Q., Yang, Y. C., Yan, J., Doerge, D.
R., and Chan, P. C. (2001) Genotoxic pyrrolizidine alkaloids and
pyrrolizidine alkaloid N-oxides - mechanisms leading to DNA adduct
formation and tumorigenicity. Environ. Carcinogen. Ecotoxicol. Rev. 19,
353−386.
(41) Mattocks, A. R., and Jukes, R. (1990) Recovery of the pyrrolic
nucleus of pyrrolizidine alkaloid metabolites from sulphur conjugates
in tissues and body fluids. Chem.-Biol. Interact. 75, 225−239.
(42) Robertson, K. A., Seymour, J. L., Hsia, M. T., and Allen, J. R.
(1977) Covalent interaction of dehydroretronecine, a carcinogenic
metabolite of the pyrrolizidine alkaloid monocrotaline, with cysteine
and glutathione. Cancer Res. 37, 3141−3144.
(25) Wang, Y. P., Yan, J., Beger, R. D., Fu, P. P., and Chou, M. W.
(2005) Metabolic activation of the tumorigenic pyrrolizidine alkaloid,
monocrotaline, leading to DNA adduct formation in vivo. Cancer Lett.
226, 27−35.
(26) Xia, Q., Chou, M. W., Edgar, J. A., Doerge, D. R., and Fu, P. P.
(2006) Formation of DHP-derived DNA adducts from metabolic
activation of the prototype heliotridine-type pyrrolizidine alkaloid,
lasiocarpine. Cancer Lett. 231, 138−145.
(27) Xia, Q., Chou, M. W., Lin, G., and Fu, P. P. (2004) Metabolic
formation of DHP-derived DNA adducts from a representative
otonecine type pyrrolizidine alkaloid clivorine and the extract of
Ligularia hodgsonnii hook. Chem. Res. Toxicol. 17, 702−708.
(28) Xia, Q., Yan, J., Chou, M. W., and Fu, P. P. (2008) Formation of
DHP-derived DNA adducts from metabolic activation of the prototype
heliotridine-type pyrrolizidine alkaloid, heliotrine. Toxicol. Lett. 178,
77−82.
(29) Yan, J., Nichols, J., Yang, Y. C., Fu, P. P., and Chou, M. W.
(2002) Detection of riddelliine-derived DNA adducts in blood of rats
fed riddelliine. Int. J. Mol. Sci. 3, 1019−1026.
(30) Yan, J., Xia, Q., Chou, M. W., and Fu, P. P. (2008) Metabolic
activation of retronecine and retronecine N-oxideFormation of
DHP-derived DNA adducts. Toxicol. Ind. Health 24, 181−188.
(31) Yang, Y., Yan, J., Churchwell, M., Beger, R., Chan, P., Doerge, D.
R., Fu, P. P., and Chou, M. W. (2001) Development of a 32P-
postlabeling/HPLC method for detection of dehydroretronecine-
derived DNA adducts in vivo and in vitro. Chem. Res. Toxicol. 14, 91−
100.
(32) Mattocks, A. R., Jukes, R., and Brown, J. (1989) Simple
procedures for preparing putative toxic metabolites of pyrrolizidine
alkaloids. Toxicon 27, 561−567.
1996
dx.doi.org/10.1021/tx300292h | Chem. Res. Toxicol. 2012, 25, 1985−1996