Depurinating acylfulvene-DNA adducts: Characterizing cellular chemical reactions of a selective antitumor agent

Article abstract of DOI:10.1021/ja0665951

Acylfulvenes (AFs) are a class of semisynthetic agents with high toxicity toward certain tumor cells, and for one analogue, hydroxymethylacylfulvene (HMAF), clinical trials are in progress. DNA alkylation by AFs, mediated by bioreductive activation, is be

Full text of DOI:10.1021/ja0665951

Published on Web 01/27/2007
Depurinating Acylfulvene-DNA Adducts: Characterizing
Cellular Chemical Reactions of a Selective Antitumor Agent
Jiachang Gong,†,‡ V. G. Vaidyanathan,† Xiang Yu,§ Thomas W. Kensler,§
Lisa A. Peterson,‡,| and Shana J. Sturla*,†,‡
Contribution from the Department of Medicinal Chemistry, College of Pharmacy, UniVersity of
Minnesota, Minneapolis, Minnesota 55455, The Cancer Center, UniVersity of Minnesota,
Minneapolis, Minnesota 55455, Johns Hopkins Bloomberg School of Public Health, Baltimore,
Maryland 21205, and School of Public Health, UniVersity of Minnesota,
Minneapolis, Minnesota 55455
Received September 12, 2006; E-mail: sturl002@umn.edu
Abstract: Acylfulvenes (AFs) are a class of semisynthetic agents with high toxicity toward certain tumor
cells, and for one analogue, hydroxymethylacylfulvene (HMAF), clinical trials are in progress. DNA alkylation
by AFs, mediated by bioreductive activation, is believed to contribute to cytotoxicity, but the structures and
chemical properties of corresponding DNA adducts are unknown. This study provides the first structural
characterization of AF-specific DNA adducts. In the presence of a reductive enzyme, alkenal/one
oxidoreductase (AOR), AF selectively alkylates dAdo and dGuo in reactions with a monomeric nucleoside,
as well as in reactions with naked or cellular DNA, with 3-alkyl-dAdo as the apparently most abundant
AF-DNA adduct. Characterization of this adduct was facilitated by independent chemical synthesis of the
corresponding 3-alkyl-Ade adduct. In addition, in naked or cellular DNA, evidence was obtained for the
formation of an additional type of adduct resulting from direct conjugate addition of Ade to AF followed by
hydrolytic cyclopropane ring-opening, indicating the potential for a competing reaction pathway involving
direct DNA alkylation. The major AF-dAdo and AF-dGuo adducts are unstable under physiologically relevant
conditions and depurinate to release an alkylated nucleobase in a process that has a half-life of 8.5 h for
3-alkyladenine and less than approximately 2 h for dGuo adducts. DNA alkylation further leads to single-
stranded DNA cleavage, occurring exclusively at dGuo and dAdo sites, in a nonsequence-specific manner.
In AF-treated cells that were transfected with either AOR or control vectors, the DNA adducts identified
match those from in vitro studies. Moreover, a positive correlation was observed between DNA adduct
levels and cell sensitivity to AF. The potential contributing roles of AOR-mediated bioactivation and adduct
stability to the cytotoxicity of AF are discussed.
Introduction
selectivities are major barriers for success for alkylating agents.
Natural product-derived alkylating agents can exhibit unique
DNA-alkylating agents have contributed to significant im-
provements in cancer therapy in the past five decades and remain
at the front line of efforts to develop improved antitumor
agents.1-5 Alkylation of cellular DNA bases can result in various
biological consequences, including blockage of DNA replication
and transcription, inhibition of gene expression, and DNA strand
cleavage.3,6-8 These processes lead to cytotoxic and cytostatic
effects on cancer cells,2,4 but dose-limiting toxicities and poor
selectivity profiles, and understanding relevant chemical mech-
anisms can reveal new strategies for selectively targeting tumor
cells.9
Acylfulvenes (AFs) are a class of antitumor agents derivatized
from mushroom-derived sesquiterpene illudins.10 Illudin S and
M (1 and 2, Chart 1) possess potent antitumor activities
(IC50 ) 4 nM in MV 522 cell lines) but low therapeutic indices
as high toxicities prevent effective clinical applications.11,12 AFs
have improved selectivity profiles for cancer cells and are
notably nonapoptotic against normal cells.13-18 One analogue,
† Department of Medicinal Chemistry, University of Minnesota.
‡ The Cancer Center, University of Minnesota.
§ Johns Hopkins Bloomberg School of Public Health.
| School of Public Health, University of Minnesota.
(1) Pratt, W. B. The Anticancer Drugs; Oxford University Press: New York,
1994.
(9) Mann, J. Nat. ReV. Cancer 2002, 2, 143-148.
(10) Bando, T.; Narita, A.; Asada, K.; Ayame, H.; Sugiyama, H. J. Am. Chem.
Soc. 2004, 126, 8948-8955.
(2) Hurley, L. H. Nat. ReV. Cancer 2002, 2, 188-200.
(3) Neidle, S.; Thurston, D. E. Nat. ReV. Cancer 2005, 5, 285-296.
(4) Wolkenberg, S. E.; Boger, D. L. Chem. ReV. 2002, 102, 2477-2496.
(5) Nelson, S. M.; Ferguson, L. R.; Denny, W. A. Cell Chromosome 2004,
3, 2-10.
(11) McMorris, T. C.; Anchel, M. J. Am. Chem. Soc. 1964, 87, 1594-1600.
(12) Kelner, M. J.; McMorris, T. C.; Taetle, R. J. Natl. Cancer Inst. 1990, 82,
1562-1565.
(13) McMorris, T. C. Bioorg. Med. Chem. 1999, 7, 881-886.
(14) McMorris, T. C.; Kelner, M. J.; Wang, W.; Yu, J.; Estes, L. A.; Taetle, R.
J. Nat. Prod. 1996, 59, 896-899.
(6) Burrows, C. J.; Muller, J. G. Chem. ReV. 1998, 98, 1109-1151.
(7) Gniazdowski, M.; Denny, W. A.; Nelson, S. M.; Czyz, M. Expert Opin.
Ther. Targets 2005, 9, 471-489.
(15) Kelner, M. J.; McMorris, T. C.; Estes, L.; Starr, R. J.; Rutherford, M.;
Montoya, M.; Samson, K. M.; Taetle, R. Cancer Res. 1995, 55, 4936-
4940.
(8) Bignold, L. P. Anticancer Res. 2006, 26, 1327-1336.
9
10.1021/ja0665951 CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 2101-2111
2101

A R T I C L E S
Gong et al.
Chart 1. Structures of Natural Illudins and Semisynthetic
Derivatives, AFs
(-)-hydroxymethylacylfulvene (4, (-)-HMAF; Chart 1) is
currently in clinical trials and is being evaluated in the treatment
of various human cancers including prostate and advanced solid
and ovarian tumors.19-23 Emerging concerns regarding dose-
limiting toxicities, however, indicate that an improved under-
standing of the detailed mechanism of action is required to
elucidate the novel mechanism of AFs for selective toxicity as
well as for identifying strategies for effectively targeting
sensitive cancers and for developing improved combination
therapies.24,25
Like many well-characterized alkylating antitumor agents,26-29
the cytotoxicities of AFs are thought to originate primarily from
covalent binding to DNA16,17,30,31 DNA damage induction by
AFs has been established by incorporation of radioactivity into
DNA isolated from cells treated with [14C]AFs.16,30 Furthermore,
recent studies indicate that the toxicities of AFs are influenced
by deficiencies in certain DNA repair pathways, such as
transcription-coupled nucleotide excision repair (TC-NER)
enzymes.32,33
Figure 1. Extracted ion chromatographs from HPLC-ESI-MS/MS analysis
of reactions of AF with monomeric nucleosides in the presence of AOR/
NADPH. (a) AF-dAdo, transition of m/z 452 [M + 1]+ to m/z 336 [M -
deoxyribose + 1]+; (b) AF-Ade, transition of m/z 336 [M + 1]+ to m/z
201 [M - Ade]+; (c) AF-Gua, transition of m/z 352 [M + 1]+ to m/z 201
[M - Gua]+; (d) AF-dCyd, transition of m/z 428 [M + 1]+ to m/z 201 [M
- dCyd]+; (e) AF-Thd, transition of m/z 443 [M + 1]+ to m/z 201 [M -
Thd]+.
AFs are less chemically reactive than the parent illudins and
have generally higher IC50’s for cancer cell lines.12,34 On the
basis of the structures of AF metabolites, it has been proposed
that reductive bioactivation, by cytosolic enzymes14,35,36 via
reduction of the carbon-carbon double bond of the R, â-un-
saturated ketone, leads to the formation of the postulated reactive
cyclohexadiene intermediate 5 (Scheme 1). We have obtained
evidence suggesting the biological role of a specific enzyme,
alkenal/one oxidoreductase (AOR),37-40 in the activation of
AFs,41-43 including an association between elevated AOR levels
and increased cellular sensitivity to AFs.41-43
Putative chemical transformations of AFs are illustrated in
Scheme 1. Hydrolysis of 5 gives rise to the major observed
metabolite 8. Intermediate 5 may also lead to monofunctional
DNA adducts, via attack of the cyclopropyl group by a DNA
base, though such monoadducts generally are considered less
lethal than cross-links common to DNA-targeting anticancer
agents, such as nitrogen mustards.44 Alternatively, AFs could
be envisioned to be activated chemically if cellular nucleophiles,
such as glutathione, were to attack the carbon-carbon double
(16) Woynarowska, B. A.; Woynarowski, J. M.; Herzig, M. C. S.; Roberts, K.;
Higdon, A. L.; MacDonald, J. R. Biochem. Pharmacol. 2000, 59, 1217-
1226.
(17) Woynarowska, B. A.; Woynarowski, J. M. Biochim. Biophys. Acta -
Molecular Basis of Disease 2002, 1587, 309-317.
(18) MacDonald, J. R.; Muscoplat, C. C.; Dexter, D. L.; Mangold, G. L.; Chen,
S.-F.; Kelner, M. J.; McMOrric, T. C.; Von Hoff, D. D. Cancer Res. 1997,
57, 279-283.
(19) Senzer, N.; Arsenau, J.; Richards, D.; Berman, B.; MacDonald, J. R.; Smith,
S. Amer. J. Clin. Oncol. 2005, 28, 36-42.
(20) Alexandre, J.; Raymond, E.; Kaci, M. O.; Brain, E. C.; Lokiec, F. O.; Kahatt,
C.; Faivre, S.; Yovine, A.; Goldwasser, F. O.; Smith, S. L.; MacDonald,
J. R.; Misset, J. L.; Cvitkovic, E. Clin. Cancer Res. 2004, 10, 3377-3385.
(21) Raymond, E.; Kahatt, C.; Rigolet, M. H.; Sutherland, W.; Lokiec, F.;
Alexandre, J.; Tombal, B.; Elman, M.; Lee, M. S.; MacDonald, J. R.; Cullen,
M.; Misset, J. L.; Cvitkovic, E. Clin. Cancer Res. 2004, 10, 7566-7574.
(22) Seiden, M. V.; Gordon, A. N.; Bodurka, D. C.; Matulonis, U. A.; Penson,
R. T.; Reed, E.; Alberts, D. S.; Weems, G.; Cullen, M.; McGuire, W. P.
Gynecol. Oncol. 2006, 101, 55-61.
(23) Hilgers, W.; Faivre, S.; Chieze, S.; Alexandre, J.; Lokiec, F.; Goldwasser,
F.; Raymond, E.; Kahatt, C.; Taamma, A.; Weems, G.; MacDonald, J. R.;
Misset, J. L.; Cvitkovic, E. InVest. New Drugs 2006, 24, 311-319.
(24) Woo, M. H.; Peterson, J. K.; Billups, C.; Liang, H.; Bjornsti, M. A.;
Houghton, P. J. Cancer Chemother. Pharmacol. 2005, 55, 411-419.
(25) Serova, M.; Calvo, F.; Lokiec, F.; Koeppel, F.; Poindessous, V.; Larsen,
A. K.; Van Laar, E. S.; Waters, S. J.; Cvitkovic, E.; Raymond, E. Cancer
Chemother. Pharmacol. 2006, 57, 491-499.
(26) Boger, D. L.; Garbaccio, R. M. Acc. Chem. Res. 1999, 32, 1043-1052.
(27) Parrish, J. P.; Kastrinsky, D. B.; Wolkenberg, S. E.; Igarashi, Y.; Boger,
D. L. J. Am. Chem. Soc. 2003, 125, 10971-10976.
(35) McMorris, T. C.; Elayadi, A.; Yu, J.; Hu, Y.; Kelner, M. J. Drug Metab.
Dispos. 1999, 27, 983-985.
(36) McMorris, T. C.; Elayadi, A. N.; Yu, J.; Kelner, M. J. Biochem. Pharmacol.
1999, 57, 83-88.
(28) Gates, K. S. Chem. Res. Toxicol. 2000, 13, 953-956.
(29) Tomasz, M.; Palom, Y. Pharmacol. Ther. 1997, 76, 73-87.
(30) Herzig, M. C. S.; Arnett, B.; MacDonald, J. R.; Woynarowski, J. M.
Biochem. Pharmacol. 1999, 58, 217-225.
(37) Ensor, C. M.; Zhang, H. X.; Tai, H. H. Biochem. J. 1998, 330, 103-108.
(38) Clish, C. B.; Levy, B. D.; Chiang, N.; Tai, H. H.; Serhan, C. N. J. Biol.
Chem. 2000, 275, 25372-25380.
(31) Woynarowski, J. M.; Napier, C.; Koester, S. K.; Chen, S.-F.; Troyer, D.;
Chapman, W.; MacDonald, J. R. Biochem. Pharmcol. 1997, 54, 1181-
1193.
(39) Hori, T.; Yokomizo, T.; Ago, H.; Sugahara, M.; Ueno, G.; Yamamoto,
M.; Kumasaka, T.; Shimizu, T.; Miyano, M. FASEB J. 2004, 18, C281.
(40) Dick, R. A.; Kwak, M.-K.; Sutter, T. R.; Kensler, T. W. J. Biol. Chem.
2001, 276, 40803-40810.
(32) Jaspers, N. G. J.; Raams, A.; Kelner, M. J.; Ng, J. M. Y.; Yamashita,
Y. M.; Takeda, S.; McMorris, T. C.; Hoeijmakers, J. H. J. DNA Repair
2002, 1, 1027-1038.
(41) Gong, J. C.; Neels, J. F.; Yu, X.; Kensler, T. W.; Peterson, L. A.; Sturla,
S. J. J. Med. Chem. 2006, 49, 2593-2599.
(33) Koeppel, F.; Poindessous, V.; Lazar, V.; Raymond, E.; Sarasin, A.; Larsen,
A. K. Clin. Cancer Res. 2004, 10, 5604-5613.
(42) Dick, R. A.; Kensler, T. W. J. Biol. Chem. 2004, 279, 17269-17277.
(43) Dick, R. A.; Yu, X.; Kensler, T. W. Clin. Cancer Res. 2004, 10, 1492-
1499.
(34) McMorris, T. C.; Yu, J.; Estes, L. A.; Kelner, M. J. Tetrahedron 1997, 53,
14579-14590.
(44) Noll, D. M.; Mason, T. M.; Miller, P. S. Chem. ReV. 2006, 106, 277-301.
9
2102 J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007

Acylfulvene−DNA Adducts
A R T I C L E S
Scheme 1. Proposed Mechanism for AF Metabolism and Formation of the Major Metabolite 8 or DNA Adducts
Scheme 2. AOR-Mediated Reactions of AF with dAdo and dGuo
bond to form 5. Elucidating which chemical transformations
may underlie the biological activities of AFs depends on the
characterization of AF-DNA adducts and assessing their
biological significance. The structures and chemical properties
of DNA adducts can influence cellular responses including DNA
repair, transcription, and cell death. Herein, we report the first
characterization of reactions between AF (3, Chart 1) and
nucleosides, on the level of individual nucleosides, in naked
DNA and in cells. These results show that AF alkylates purine
bases, giving rise to chemically labile adducts. AF-DNA
adducts observed in cells treated with AF match those identified
from in vitro reactions, and relative levels correlate with cell
sensitivity and reductase expression.
MS/MS is an ideal tool to distinguish products resulting from
different reaction pathways. Monomeric deoxynucleosides were
reacted individually with AF in the presence or absence of AOR/
NADPH, at 37 °C for 24 h in a phosphate-buffered solution
(pH 7.4). After solid-phase extraction, reaction mixtures were
analyzed by HPLC-ESI-MS/MS. Both deoxyadenosine and
deoxyguanosine (dAdo and dGuo, respectively) were readily
alkylated by AF in the presence of AOR/NADPH (Figure 1a-
c), but under the same reaction conditions, no adducts were
observed in reaction mixtures from deoxycytidine or thymidine
(dCyd or Thd, respectively) reactions (Figure 1d, e). In the
absence of AOR/NADPH, we examined the direct reaction of
AF with monomeric deoxynucleosides (Scheme 1). No nucleo-
side adducts were detected by HPLC-MS/MS, and AF remained
unchanged. To examine whether AOR or NADPH alone
activates AF, control experiments were carried out in which
dAdo was reacted with AF in the presence of AOR or NADPH;
no adducts were observed in either case.
Results
Reactions of AF with Nucleosides in the Presence or
Absence of AOR. As an initial step toward characterizing DNA
damage induced by AF, reactions with monomeric deoxy-
nucleosides were performed as simple and easily analyzed model
systems. Additionally, these studies were used to establish
analytical methods required to probe adduct formation in double-
stranded DNA and in cells. On the basis of proposed mecha-
nisms of DNA alkylation by AF (Scheme 1), DNA adducts
formed from competing direct vs bioactivation-coupled path-
ways are predicted to contain isomeric indene moieties with
characteristic MS/MS fragmentation patterns. Therefore, HPLC-
HPLC-MS/MS analysis of the dAdo reaction mixture displays
a minor peak eluting at 13.5 min with m/z 452 and a major
peak eluting at 19.3 min with m/z 336 (Figure 1a, b). MS/MS
analysis of the peak at 13.5 min yields the fragments m/z 336
(M - deoxyribose + 1) and m/z 201 (M - dAdo), consistent
with AF-dAdo (Scheme 2). The peak eluting at 19.3 min
displays a major fragment ion m/z 201, consistent with depu-
rination of AF-dAdo. The positively charged AF-dAdo is
9
J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007 2103

A R T I C L E S
Gong et al.
Figure 2. Selected proton-carbon correlations observed in the 1H-13C gHMBC NMR spectrum, and chemical shifts of selected protons and carbons of 20.
Scheme 3. Chemical Synthesis of the Major AF Metabolite 8a
a Reagents and conditions: (a) HCl/AlCl3, methacronitrile; (b) 10%
NaOH, reflux; (c) AlCl3, 200 °C; (d) SnCl4, MOMCl; (e) KCN, DMF; (f)
10% NaOH, reflux; (g) BH3/THF; (h) MeOH, H3O+.
MS/MS data. In this regard, two-dimensional heteronuclear
NMR experiments are useful for distinguishing potential
isomers.48 A common structural feature of AF-DNA adducts
resulting from AOR-mediated reactions is the indene moiety.
Figure 3. Extracted ion chromatographs from HPLC-ESI-MS/MS analysis
We proposed that a useful synthetic intermediate for probing
the identity of bioactivation-derived AF adducts would be AF
metabolite 8, with the primary alcohol converted into a leaving
group. Metabolite 8 can be obtained by enzyme-mediated
reduction of AF followed by hydrolysis or by treating AF with
zinc in acid. These approaches are inefficient because they
involve the synthesis of the tricyclic AF core structure, only to
destroy the three-membered ring and aromatize the six-
membered ring. We therefore desired an electrophilic precursor
that could be prepared in abundant quantities from readily
available reagents and would be amenable to structural modi-
fications that would be useful in the subsequent studies of related
AF derivatives.
The synthesis of 8 was accomplished in eight steps starting
from 2,4-dimethylphenol 9 (Scheme 3). Thus, 9 was treated with
methacronitrile in the presence of anhydrous AlCl3 and dry HCl,
affording 10 in 60% yield. The cyano group in 10 was
hydrolyzed in aqueous NaOH to yield acid 11 in 90% yield.
Cyclization of 11 was accomplished in 85% yield by treating
with AlCl3/NaCl at 200 °C. The desired chloromethylation of
12 was problematic because common chloromethylation condi-
tions involving formaldehyde in concentrated HCl resulted in
the formation of an acid-catalyzed aldol reaction product, in
which the R-carbon of the ketone was substituted. Using more
vigorous conditions (MOM-Cl, SnCl4) however, 12 is con-
of hydrolyates of DNA samples treated with AF in the presence of AOR/
NADPH. (a) 20, transition of m/z 336 [M + 1]+ to m/z 201 [M - Ade]+;
(b) 21 and 25, transition of m/z 352 [M + 1]+ to m/z 201 [M - Gua]+;
(c) 27, transition of m/z 352 [M + 1]+ to m/z 217 [M - Ade]+ ; (d) Co-
injection of synthetic 20 with DNA sample, transition of m/z 336 [M +
1]+ to m/z 201 [M - Ade]+.
thermally unstable: heating the reaction mixture (90 °C, 30 min)
results in the disappearance of the peak corresponding to AF-
dAdo and produces a chromatogram in which AF-Ade is the
only detected product, consistent with the conversion of AF-
dAdo to AF-Ade.
HPLC-MS/MS analysis of a dGuo reaction mixture displays
a peak eluting at 18.2 min with m/z 352 and a major MS/MS
fragment ion m/z 201 (Figure 1c), consistent with AF-Gua
(Scheme 2). These data are consistent with a similar pathway
as that indicated for AF-Ade adduct, i.e., formation of a
positively charged dGuo adduct which readily depurinates
resulting in an AF-base adduct. The difference between the
reactions with dAdo and dGuo is that the postulated intermediate
AF-dGuo (Scheme 2) was not detected by HPLC-MS/MS. The
absence of AF-dGuo may reflect decreased relative stability.
Identification of the AF-dAdo and AF-Gua Adduct. Levels
of adduct derived from the AF-treated nucleosides described
above were insufficient for definitive structural characterization
by NMR analysis. Therefore we sought a synthetic route for
preparing the depurination product AF-Ade and AF-Gua.
Alkylation at both 3- and 7-positions of dAdo or dGuo labilizes
the modified bases,45-47 and both would be consistent with the
(46) Singer, B.; Grunberger, D. Molecular biology of mutagens & carcinogens;
Plenum Press: New York, 1983.
(47) Gates, K. S.; Nooner, T.; Dutta, S. Chem. Res. Toxicol. 2004, 17, 839-
856.
(45) Golankiewicz, B.; Ostrowski, T.; Folkman, W. Nucleic Acids Res. 1990,
18, 4779-4782.
(48) Cheng, G.; Shi, Y.; Sturla, S. J.; Jalas, J. R.; McIntee, E. J.; Villalta,
P. W.; Wang, M.; Hecht, S. S. Chem. Res. Toxicol. 2003, 16, 145-152.
9
2104 J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007

Acylfulvene−DNA Adducts
A R T I C L E S
Scheme 4. Chemical Synthesis of Adduct 20 and 21a
a Reagents and conditions: (a) 2-mesitylenesulfonyl chloride/pyridine; (b) p-toluenesulfonyl chloride/pyridine; (c)NaBr, DMF; (d) NaI, acetone, reflux;
(e) Ade, DMA, 80 °C; (f) dGuo, DMA, 80 °C.
Chart 2. Structures of AF-DNA Adducts Proposed on the Basis
of MS and MS/MS Data
mediated reaction of dAdo with AF, indicating that they are
identical. 1H and 1H-13C gHMQC NMR analysis (Supporting
Information) resulted in a spectrum consistent with the formula-
tion of an adduct that contains both indene and Ade moieties.
To unambiguously assign the substitution site of Ade, a 2-D
1H-13C gHMBC NMR spectrum was obtained (Supporting
Information). These data indicate correlations of methylene
protons of indene H1′ (4.3 ppm) with C-2 (143 ppm) and C-4
(150 ppm) of Ade (Figure 2), demonstrating that the final
product of the reaction of AF with dAdo is the 3-substituted
Ade adduct.
In contrast to the reaction of Ade, reaction of bromide 18
with dGuo did not result in any products consistent with an
AF-Gua adduct. However, the iodide 19 readily reacts with
dGuo at 80 °C in DMA to afford a product that coelutes with
the AF-Gua adduct that results from the reaction between dGuo
and AF in the presence of AOR (Scheme 4). This compound
(21) was isolated in 5% yield in a similar manner to that for
20. Further analysis by HPLC-MS/MS indicates that 21 is
identical to the AF-Gua adduct, and NMR analysis (Supporting
Information), including a 2-D 1H-13C gHMBC NMR, demon-
strated that compound 21 is a 7-substituted guanine adduct.
verted into 13 in 70% yield. Chloride 13 was treated with KCN
followed by base-catalyzed hydrolysis to generate acid 15. Both
carboxyl and ketone groups of 15 were reduced in one step
using excess BH3. Acidic workup resulted in metabolite 8 in
65% yield.
To enable the comparison of the relative abundances of Ade
adduct 20 and Gua adduct 21 on the basis of HPLC-MS/MS
peak areas, a sample containing equal concentrations of 20 and
21 (0.1 µM) were analyzed by HPLC-MS/MS and relative
sensitivities for ESI-MS detection were determined. The relative
response factor of adduct 20 is 2.5 times higher than that of
21, and the HPLC-MS/MS peak areas of 20 generated from
the AOR-mediated reaction of AF with dAdo is 250 times higher
than that of 21 generated from the AOR-mediated reaction of
AF with dGuo. This indicates that, under the same conditions,
the reaction between dAdo and AF in the presence of AOR is
about 100 times more efficient than that of dGuo.
The primary alcohol of 8 was converted into various leaving
groups, and their reactivities toward Ade were tested (Scheme
4). Neither a p-toluenesulfonyl or a 2-mesitylenesulfonyl nor
an iodide derivative of 8 reacted with Ade, up to 80 °C for 48
h. On the other hand, under the same conditions, when Ade
was reacted with bromide derivatives 18, which was obtained
in 50% yield by treating 17 with NaBr, product 20 was
genereated with chromatographic properties consistent with
those of AF-Ade. This material was purified crudely by normal-
phase flash chromatography, followed by a Biotage Sp1 high
performance flash chromatography system equipped with a
reversed-phase flash column, yielding 1-2% of the product.
The yield could not be improved by adding AgOTf or by varying
the reaction temperature, time, and relative equivalents of
reactants, but the amount of product generated was sufficient
for structural characterization. The synthetic material had the
same HPLC retention time, m/z, and MS/MS fragmentation
patterns as those observed from AF-Ade produced in the AOR-
Reaction of AF with Calf Thymus DNA. The chemical
reactivities of nucleotides in DNA are influenced by the
structural conformation of the double-stranded helix. Further-
more, cellular DNA is confined to the nucleus, forming a
complex with histone proteins with highly ordered structures.49
Despite the differences between the chemical environments of
naked vs cellular DNA, naked DNA can serve as an effective
(49) van Holde, K. E. Chromatin; Springer-Verlag: New York, 1989.
9
J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007 2105

A R T I C L E S
Gong et al.
Table 1. Extracted Ion Chromatographs of Reactions of Calf
Thymus DNA with AF
adducts
AOR
MS peaks
MS/MS
AF-dAdo
AF-Ade
AF-Gua
+
+
+
452
336
352
336/201
201
201
(two peaks)
model, often yielding similar patterns of adduct formation.50-53
We therefore investigated the reaction of AF with naked DNA
as a model to characterize potential cellular reactions of AF.
Calf thymus DNA was allowed to react with AF under the
same conditions as those described for monomeric nucleoside
reactions. Alkylated DNA was precipitated, isolated, and
hydrolyzed enzymatically with DNase I, phosphodiesterase I,
and alkaline phosphatase. Both the supernatant of DNA reaction
mixtures and hydrolyzed DNA samples were analyzed by
HPLC-MS/MS. Results indicate that, in the presence of AOR/
NADPH, AF reacts with DNA giving rise to four main products
(Figure 3). In both supernatant and hydrolyzed DNA samples,
20 appears to be the most abundant peak accounting for 98%
of the total peak areas for adducts either in the hydrolysate or
in the supernatant. The presence of 20 in the supernatant
indicates that AF-dAdo adduct 22 (Chart 2) is chemically labile
in double-stranded DNA and depurinates at 37 °C and pH 7.4,
releasing the modified nucleobase and generating an abasic site
in the DNA strand.46,47 A peak corresponding to 22 also was
observed in hydrolyzed DNA samples but slowly converted to
20 in neutral aqueous solution at 25 °C. Consistent with results
from nucleoside studies discussed above, dGuo adducts are
minor DNA alkylation products in both supernatant and DNA
hydrolysate. In addition to the adduct 21 that elutes at 18.2 min,
another peak corresponding to a mass consistent with AF-
modified Gua was observed to elute at 17.1 min (Figure 3b).
The identical MS and MS/MS spectra (Table 1) of both Gua
adducts indicate that they are isomeric and result from alkylation
at different sites on dGuo, inducing depurination of modified
Gua from double-stranded DNA. Potential reactions at the 7-
and 3-position of dGuo are consistent with this type of reactivity,
such that the potential structure for the new guanine adduct may
be 25 (Chart 2).
To test whether DNA reacts with AF in the absence of a
metabolic enzyme, DNA was allowed to react with AF in a
phosphate buffer (pH 7.4) at 37 °C for 24 h. DNA was isolated
and enzymatically hydrolyzed as described above. HPLC-MS/
MS analysis of DNA hydrolysate displays a peak that eluted at
15.1 min with m/z 352 (M + 1) (Figure 3c) and MS/MS of 217
(M-dAdo), consistent with an adduct resulting from conjugate
addition to the R,â-unsaturated ketone of AF, followed by
hydrolytic cyclopropyl ring-opening illustrated by 27 (Scheme
5). An adduct with this m/z and MS/MS pattern was not detected
in the direct reaction of AF with dAdo or in the reaction of AF
with DNA in the presence of AOR/NADPH. The relative
abundance of 27, on the basis of the resulting HPLC-MS/MS
spectrum, is estimated to be about 0.1% that of 20 resulting
Figure 4. Neutral thermal hydrolysis of 22 from double-stranded DNA.
Figure 5. Cleavage of pBR322 DNA (300 ng) by AF in the presence of
AOR (10 ng)/NADPH (200 µM) in TBE at 37 °C. Lane 1: Control pBR322
DNA (300 ng); lane 2: DNA (300 ng) + AF (250 µM); lane 3: DNA +
AOR (10 ng); lane 4: DNA + AOR (10 ng) + NADPH (200 µM); lane
5-11: DNA + AOR (10 ng) + NADPH (200 µM) + AF (10, 20, 30, 40,
50, 60, 70 µM, respectively).
from the reaction between AF and DNA, indicating the
diminished direct reactivity of AF (Scheme 5, pathway A)
compared to that of the intermediate generated in the bioacti-
vation-mediated process (Scheme 5, pathway B). In the same
reaction mixture, DNA adducts that may have been formed by
direct chemical reaction of AF with other bases in DNA were
not observed. Peaks corresponding to AF-dGuo (transition of
m/z 484 to m/z 217), AF-dCyd (transition of m/z 444 to m/z
217), and AF-Thd (transition of m/z 459 to m/z 217) were not
detected in HPLC-MS/MS analysis. DNA cross-links such as
Ade-AF-Ade (m/z 471) and Ade-AF-Gua (m/z 487), that may
have been formed through the nucleopilic attack of a DNA base
to putative intermediate 5, which would result from the direct
reaction of AF with DNA, were not observed.
Stabilities of AF-DNA Adducts in Double-Stranded DNA.
The release of AF adducts from double-stranded DNA creates
abasic sites in DNA, which raises the question of whether DNA-
interacting proteins encounter DNA-AF adducts or the resulting
abasic sites. Because the chemical stabilities of adducts may
be useful for elucidating their potential biological relevance,
we determined the t1/2 of 22 in double-stranded DNA. Calf-
thymus DNA was treated with AF in the presence of AOR/
NADPH. Alkylated DNA was isolated and redissolved in
phosphate-buffered solution (pH 7.4). The resulting solution was
heated at 37 °C, and the release of 20 was monitored by
removing aliquots and analyzing by HPLC-MS/MS. An estimate
of the initial concentration of 22 was established by heating
the alkylated DNA at 90 °C for 1 h. Further heating of DNA
(50) MacLeod, M. C. Carcinogenesis 1995, 16, 2009-2014.
(51) Millard, J. T.; Spencer, R. J.; Hopkins, P. B. Biochemistry 1998, 37, 5211-
5219.
(52) Smith, B. L.; Bauer, G. B.; Povirk, L. F. J. Biol. Chem. 1994, 269, 30587-
30594.
(53) Trzupek, J. D.; Gottesfeld, J. M.; Boger, D. L. Nat. Chem. Biol 2006, 2,
79-82.
9
2106 J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007

Acylfulvene−DNA Adducts
A R T I C L E S
Scheme 5. Putative Competing DNA Alkylation Pathways
does not give increased levels of adducts, indicating complete
release of adducts from DNA. The analysis of modified DNA
was carried out until over 90% of adduct underwent depurination
(ca. 24 h). For MS analysis, N2-Bn-dG54 was used as an
analytical injection standard and the peak area of 20 in each
sample was normalized according to this standard. Data obtained
from HPLC-MS/MS analysis yielded the rate constant k ) 0.081
formic acid/piperidine to induce cleavage at all Gua sites or at
all Gua and Ade sites, respectively. For DNA treated with AF
alone, faint bands were observed consistent with cleavage at
Gua and Ade sites but lacked any evidence for sequence
specificity; i.e., the relative intensity of the bands was the same
as that for the dimethyl sulfate/piperidine control. Likewise, in
the presence of AOR, cleavage did not appear to be sequence
specific (Figure S3, Supporting Information), but AOR greatly
enhanced the efficiency of Gua + Ade cleavage, consistent with
increases in adduct levels observed in reactions of AF/AOR
with nucleosides and calf thymus DNA (Scheme 5).
h-1 (Figure 4) for depurination of AF-dAdo 22 in DNA; t1/2
)
8.5 h. Using the same analytical method, which was optimized
for detection and resolution of adduct 20, the peaks correspond-
ing to the two dGuo adducts were not resolved. After heating
at 37 °C for 8 h, however, the unresolved peak corresponding
to 21 and 25 on HPLC-MS/MS spectra saturated such that the
Gua adduct levels were the same as that from a sample heated
at 90 °C for 1 h. Complete depurination within 8 h, assuming
approximately four to five-half-lives, provides a crude estimate
of the stabilities of 23 and 24 at under 2 h.
Relative Levels of Adducts in Cells Transfected with
pCEP and AOR Vectors. In previous studies, we observed a
positive correlation between sensitivity to AFs and levels of
AOR expression in a large panel of tumor cell lines. To
determine whether this relationship is consistent with increases
in AF-specific adducts and to probe whether the reaction
products characterized in cell-free systems are consistent with
cellular processes, we examined adduct formation in cells treated
with AF and compared adduct levels in AF-sensitive and
resistant cells. Human embryonic kidney cells were transfected
with AOR vectors or blank vectors, as described previously.41
Each cell line was treated with AF (500 nM) for 3 h, cells were
lysed, and DNA was isolated. The DNA was hydrolyzed and
analyzed as described for calf thymus DNA reactions. Consistent
with the results from naked DNA experiments, Ade adduct 20
and Gua adducts 21 and 25 were observed in hydrolysates of
DNA samples isolated from cells (Figure S4, Supporting
Information). Adduct 20 is the predominant DNA adduct in AF-
treated cells, accounting for 98% of the total peak area
attributable to 20, 21, and 25. While rigorous comparisons
between absolute levels of different adducts cannot be made
by HPLC-MS/MS analysis without suitable analytical standards,
the data obtained allowed us to compare relative levels of a
given adduct in AOR-transfected vs control cells. It is therefore
significant that 5-10 times higher levels of each adduct were
observed in cells that overexpressed AOR vs those in control
cells (Figure 6). A peak corresponding to adduct 27, resulting
from the direct conjugate reaction of DNA with AF, was also
observed. However, because of the near-detection-limit levels
of this adduct in all samples, a comparison of levels between
cell types could not be made (Figure S4, Supporting Informa-
tion). To verify that differences in adduct levels reflected those
present in cellular DNA, rather than were influenced by loss
during isolation and hydrolysis, cell culture media and DNA
wash solutions were also analyzed for the presence of depuri-
nation adducts, and no adducts were detected in these samples.
DNA Relaxation Assay. There is a substantial precedent for
DNA-alkylating agents to result in depurinating adducts.47 The
formation of labile adducts can induce single-strand breaks in
double-stranded DNA under basic conditions.6 To probe whether
labile AF adducts result in DNA strand cleavage, relaxation
assays of supercoiled plasmid DNA were carried out. Single-
stranded cleavage of circular pBR322 DNA (form I) can lead
to the formation of open-circular DNA (form II), whereas
double-stranded cleavage results in linear DNA (form III).
Supercoiled plasmid DNA (form I) was reacted with AF at 37
°C for 48 h under alkaline conditions (pH 8.3) in the presence
or absence of AOR (10 ng) and NADPH (200 µM). In the
resulting gel electrophoretogram, no cleavage was observed for
the DNA treated with AF (250 µM), AOR, or NADPH alone
(Lanes 1-4, Figure 5). Addition of decreasing amounts of AF
(10-70 µM) in the presence of AOR (10 ng) and NADPH (200
µM), however, resulted in cleavage of DNA from form I to
form II (Lanes 5-11). A linear form of DNA, resulting from
double-stranded cleavage, was not observed. A similar experi-
ment was carried out in phosphate buffer (pH 7.4) in order to
analyze whether AF can induce DNA cleavage in the presence
of AOR and NADPH under neutral conditions, but the cleavage
efficiency in neutral conditions is less than that in alkaline
conditions (Figure S1, Supporting Information).
To probe whether DNA sequence influences the sites targeted
by AF, we examined the cleavage of a 377 bp sequence
(pBR322) of DNA using a standard Maxam-Gilbert55 sequenc-
ing approach. pBR322 DNA was reacted with either 20 or 40
µM AF in the presence of AOR/NADPH, or 40 µM AF alone
(Figure S2, Supporting Information). As comparison standards,
reactions were carried out using dimethyl sulfate/piperidine or
(54) Choi, J. Y.; Guengerich, F. P. J. Biol. Chem. 2004, 279, 19217-19229.
(55) Maxam, A. M.; Gilbert, W. Methods Enzymol. 1980, 65, 499-559.
9
J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007 2107

A R T I C L E S
Gong et al.
Figure 6. Relative levels of DNA adducts in cells transfected with an AOR vector or a pCEP vector. (A) Ade adduct 20; (B) Gua adduct 21; (C) Gua adduct
25. Error bars indicate standard deviations from average values shown for three independent analytical runs.
Discussion
were to act as the nucleophile. Cross-linking is a common mode
of action for antitumor agents, but such adducts could not be
detected in our analyses. While this observation is consistent
with previous studies, it is possible that cross links are formed
at presently undetected levels. An alternative reaction pathway
consistent with the MS/MS data would involve addition of H2O
at the C8 position of AF, followed by homoallylic substitution
by DNA. However, there is no supporting evidence to suggest
that H2O can react directly with AF.34 The direct reaction
between AF and DNA, but not between AF and nucleosides,
as well as differences in products formed in reactions with
monomeric nucleosides vs DNA, may indicate the role of
noncovalent interactions between AF and DNA that activate
AF for nucleophilic attack.26 Furthermore, alkylation at two
dGuo sites in DNA compared to one in the isolated dGuo
reaction may indicate that AF or its reactive intermediate may
also result from noncovalent interactions within DNA.56,57
This study describes the first chemical characterization of
DNA adducts induced by AF and the first evidence that the
associated chemical transformations are relevant in cells. In the
current study, we focus on reactions of AF as the parent structure
in a class that includes the clinically relevant HMAF in which
the core AF structure appears to be a key factor for selectivity.
AF readily alkylates monomeric nucleosides and DNA in the
presence of a bioreductive enzyme, AOR. The formation of two
dAdo adducts 20 and 22 and one dGuo adduct 21 was observed
in reactions of AF with monomeric nucleosides in the presence
of AOR/NADPH, with 20 as the predominant species. These
adducts have a common major MS fragment of m/z 201
indicative of an indene moiety and consistent with previous
results suggesting that AF is activated by reduction of the
carbon-carbon double bond of the R,â-unsaturated ketone
(Scheme 1). Reactions between AF and naked or cellular DNA
resulted in the formation of adducts common to those from
monomeric nucleoside reactions. Additionally, a Gua adduct
not present in monomeric nucleoside reactions was observed
in the calf thymus and cellular DNA reaction mixtures. We
observed the same pattern of adducts in AF-treated cells as those
in reactions between AF and naked DNA.
On the basis of previous studies, AF alone is unreactive
toward common nitrogen-based nucleophiles34 but covalently
associates with isolated proteins and DNA.30 Using a sensitive
HPLC-MS/MS analytical method, we observed the formation
of a directly formed Ade adduct. On the basis of its MS/MS
fragmentation, the direct adduct is structurally distinct from that
formed in the presence of AOR and is consistent with a structure
resulting from the conjugate addition of adenine to the R,â-
unsaturated ketone of AF, followed by hydrolytic cyclopropane
cleavage (6, Scheme 5). The postulated intermediate could lead
to DNA cross-links, i.e., 29, if a DNA base instead of H2O
Structural characterization of the major AF adduct and a Gua
adduct was achieved by independent chemical synthesis of
standards for 20 and 21, revealing that when bioactivated by
AOR, AF mainly alkylates dAdo at the 3-position with only a
small portion of activated AF alkylating dGuo at either the 3-
or 7-position. The preferential alkylation of DNA at the
3-position of dAdo is observed with other antitumor agents
including CC-1065 and duocarmycins,26,58,59 as well as the
closely related toxin ptaquiloside.60 These agents share an allylic
cyclopropane as a key structural feature, and the postulated
activated intermediate of AF (5) is most structurally similar to
the toxin ptaquiloside. Adducts from ptaquiloside, however, are
formed indiscriminately at both the 7-position of dGuo and the
3-position of dAdo.60 While sequence context does not appear
to influence the reactivity of AF, discrepancies observed in
reactions of nucleosides vs naked/cellular DNA and between
(56) Dervan, P. B.; Edelson, B. S. Curr. Opin. Struct. Biol. 2003, 13, 284-
(58) Boger, D. L. Angew. Chem., Int. Ed. 1996, 35, 1438-1474.
299.
(59) Boger, D. L.; Garbaccio, R. M. Bioorg. Med. Chem. 1997, 5, 263-276.
(60) Kushida, T.; Uesugi, M.; Sugiura, Y.; Kigoshi, H.; Tanaka, H.; Hirokawa,
J.; Ojika, M.; Yamada, K. J. Am. Chem. Soc. 1994, 116, 479-486.
(57) Smela, M. E.; Currier, S. S.; Bailey, E. A.; Essigmann, J. M. Carcinogenesis
2001, 22, 535-545.
9
2108 J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007

Acylfulvene−DNA Adducts
A R T I C L E S
Scheme 6. Schematic Representation of AOR-Mediated Reaction of AF with Double-Stranded DNA, Illustrated for a Random DNA
Sequence
AF and structurally similar agents may indicate differences in
corresponding DNA binding modes.56,57
these adducts to the cytotoxicity of AF, which is consistent with
previous results.41,43 Furthermore, AF-DNA adducts may serve
as useful chemical markers of the sensitivity of cells to AF.
Structural elucidation of AF-DNA adducts provides a
chemical tool for probing their biological significance. Bulky
DNA adducts, such as the initial nucleoside adducts induced
by AF, can be cytotoxic, blocking both DNA replication and
transcription. Quickly generated abasic sites, through depuri-
nation of labile alkylated purine bases, may also contribute to
cytotoxicity; however, abasic sites typically are well tolerated
by cells and rapidly repaired mainly through the process of base
excision repair (BER).63 It is therefore important to understand
the degree to which AF-DNA adducts or abasic lesions are
the major inducer of cell responses. Despite the chemical lability
observed in cell free studies, these adducts appear to persist in
the presence of cellular proteins. Furthermore, previous work
has demonstrated that AF-induced DNA damage is exclusively
repaired by transcription-coupled NER enzymes, which are
anticipated to process bulky adducts but not be significantly
influenced by BER pathways.32,33 The results of this study,
combined with data from cell-based experiments, indicate that
the initially formed chemically labile AF-DNA adducts con-
tribute to the cytotoxicity of AF, but further studies are required
to correlate the specific structures of AF adducts, i.e., 22 vs 23,
24, with their interactions with biological systems, such as NER
proteins.
In summary, these studies provide the first direct chemical
evidence for DNA adducts specific to AF using nucleosides
and naked DNA as well as in cells. A bioactivation-generated
3-substituted dAdo adduct is estimated to be the most abundant
adduct in reactions with cellular DNA as well as with nucleo-
sides or calf thymus DNA in the presence of AOR. Furthermore,
AF is capable of directly damaging DNA without enzymatic
bioactivation. These adducts appear to be formed at compara-
tively low levels in both cell-free and cell-based systems, but
their impacts on biological processes are unknown. A positive
correlation between cell sensitivity and levels of structurally
specific bioactivation-generated DNA adducts was observed,
which indicates that these DNA adducts may have medicinal
significance.
Purine alkylation labilizes the modified base in DNA, and
the chemical stability depends on the structure.46,47 We observed
facile depurination of AF-DNA adducts from double-stranded
DNA (Scheme 6). In the case of dGuo, only depurinated AF-
Gua adducts were detected, reflecting the instability of the
corresponding nucleoside adduct. The half-life of 22 under
physiological pH was determined to be 8.5 h, and the corre-
sponding dGuo adducts, 23 and 24, have half-lives under 2 h
in DNA. Depurination of AF adducts results in abasic DNA
sites and single-stranded DNA cleavage under weakly basic
conditions (pH 8.3), as evidenced by supercoiled DNA relax-
ation assays. The approximate degree of DNA cleavage cor-
relates with adduct levels: treatment with AF alone leads to
minimal DNA cleavage, while addition of AOR/NADPH results
in high levels of DNA adducts and substantial DNA cleavage.
The bioactivation-mediated process of strand cleavage results
in increases in reactivity but does not impact the pattern of
cleavage with respect to DNA sequence. Both processes appear
to have no sequence-based selectivities, at least in the case of
sequences presented by the pBR322 fragment. Finally, reactions
of AF with individual nucleosides, calf thymus DNA, and a
377 bp DNA sequence demonstrate that AF alkylates dGuo and
dAdo but not dCyd and Thd and that the efficiency of this
process is increased dramatically by AOR-mediated bioactiva-
tion.
Cell-based studies provide direct evidence for the generation
of AF-specific DNA damage and further support the role of
AOR in the bioactivation of AF and sensitization of cells. AOR
is a cytosolic enzyme, and it has therefore been debated whether
the reactive metabolite of AF is stable enough, in the presence
of cellular nucleophiles, to reach the nucleus and alkylate
chromosomal DNA.43,61 In the current studies, we observed a
positive correlation between cytosolic levels of AOR and DNA
adducts induced by AF. Cellular DNA samples were obtained
using a nuclear extract method indicating that the activated
intermediate of AF appears sufficiently stable to reach DNA
targets.62 The positive correlation between specific adduct levels
and cell sensitivities to AF may indicate a contributing role of
Experimental Section
(61) McMorris, T. C.; Kelner, M. J.; Wang, W.; Moon, S.; Taetle, R. Chem.
Res. Toxicol. 1990, 3, 574-579.
Chemicals and Enzymes. All chemicals were purchased from the
Sigma-Aldrich chemical company (Milwaukee, WI) and used without
further purification. AOR was purified according to the procedure
(62) Dignam, J. D.; Lebovitz, R. M.; Roeder, R. G. Nucleic Acids Res. 1983,
11, 1475-1489.
(63) Boiteux, S.; Guillet, M. DNA Repair 2004, 3, 1-12.
9
J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007 2109

A R T I C L E S
Gong et al.
previously described.42 All other enzymes were purchased from Sigma
(St. Louis, MO). The synthesis of AF was described previously.41 N2-
Bn-dGuo was synthesized according to the published method.54
Reaction of AF with Calf Thymus DNA in the Presence of AOR/
NADPH. To a 1.5 mL eppendorf tube were added 0.87 mL water,
phosphate buffer (100 µL, 1.0 M, pH 7.4), calf thymus DNA (2 mg),
AF (0.5 µmol, 10 µL of 50 mM solution in DMSO), NADP+ (5.0
µmol), glucose-6-phosphate (10.0 µmol), glucose-6-phosphate dehy-
drogenase (5 units), and AOR (0.2 nmol). The mixture was allowed to
react at 37 °C for 24 h. Modified DNA was precipitated by adding
cold ethanol. It was washed with ethanol, dried under nitrogen for 15
min, and enzymatically hydrolyzed using the general DNA hydrolysis
method described above. The supernatant and hydrolyzed DNA samples
were analyzed using the general qualitative method.
1
Apparatus and Assay Conditions. 1-D H and 13C NMR spectra
were acquired on a 500 MHz Varian NMR spectrometer, and chemical
shifts were reported relative to the residual nondeuterated solvent
signals. 2-D gHMQC and gHMBC were recorded on a 600 MHz Varian
NMR spectrometer. HPLC analysis was carried out on an Agilent 1100
series instrument with a diode array detector. Compounds were eluted
using a solvent gradient from 10 to 75% acetonitrile in water over a
course of 30 min at a flow rate of 0.8 mL/min. The HPLC column
used was a Phenomenex Luna 5 µm C18(2) 100 A 250 mm × 4.60
mm (Phenomenex, Torrance, CA). HRMS were recorded on a Bruker
BioTOF II mass spectrometer with an Electron-spray ionization source.
Polypropethylene glycol (PPG) was used as matrix in all HRMS
experiments. Flash chromatography (silica gel 200-400 mesh) was used
for product purification.
Qualitative HPLC-ESI-MS/MS analyses were carried out with a
Finnigan LCQ Deca instrument (Thermo Finnigan LC/MS Division,
San Jose, CA) interfaced with a Waters Alliance 2690 HPLC multi-
solvent delivery system and equipped with an SPD-10A UV detector
(Shimadzu Scientific Instruments, Columbia, MD), using the HPLC
solvent elution system and column as described above. The ESI source
was set in positive ion mode as follows: voltage, 4 kV; current,
10 µA; and capillary temperature, 275 °C. MS data were collected for
a mass range of 100-1000 amu, and MS/MS data were acquired with
the following parameters: isolation width, 1.5 amu; normalized collision
energy, 40%; activation Q, 0.25; and activation time, 30 ms. MS
samples were purified by solid-phase extraction prior to analysis as
follows: A solid-phase extraction cartridge (Phenomenex, Strata-X 33
µm cartridge) was activated by eluting with 1 mL of MeOH followed
by 1 mL of H2O. The sample was loaded on the cartridge and was
washed successively with 1 mL of H2O, 1 mL of 10% aqueous MeOH,
and 2 mL of 70% aqueous MeOH. The MeOH-containing portions were
combined and concentrated to a volume of approximately 1 mL.
Semiquantitative HPLC-ESI-MS/MS analysis was carried out with
an Agilent 1100 capillary flow HPLC (Agilent Technologies, Palo Alto,
CA) interfaced with a Finnigan Discovery Max triple quadrupole mass
spectrometer. Chromatographic separation was achieved with a 150
mm × 0.5 mm 5 µm particle size C18 column (Agilent Zorbax SB-
C18), using the HPLC solvent elution system described in each
experiment. The ESI source was set in the positive ion mode as
follows: voltage, 3.7 kV; current, 3 µA; and heated ion transfer tube,
275 °C. Samples were purified by solid-phase extraction as described
above but with a final sample volume of 50 µL.
Reaction of AF with Calf Thymus DNA in the Absence of AOR/
NADPH. To a 1.5 mL eppendorf tube were added 0.99 mL water,
phosphate buffer (100 µL, 1.0 M, pH 7.4), calf thymus DNA (2 mg),
and AF (0.5 µmol, 10 µL of 50 mM solution in DMSO). The mixture
was allowed to react at 37 °C for 24 h. Modified DNA was precipitated
by addition of cold ethanol, washed with ethanol, dried under nitrogen
for 15 min, and enzymatically hydrolyzed using the general DNA
hydrolysis method. The supernatant and hydrolyzed DNA samples were
isolated and analyzed using the general qualitative method.
Kinetics of Depurination of DNA-AF Adducts in Double-
Stranded DNA. To a 1.5 mL eppendorf tube were added 0.87 mL
water, phosphate buffer (100 µL, 1.0 M, pH 7.4), calf thymus DNA
(2 mg), AF (10 µL, 50 mM in DMSO), NADP+ (50 µL, 100 mM),
glucose-6-phosphate (50 µL, 200 mM), and glucose-6-phosphate
dehydrogenase (10 µL, 500 units/mL) and AOR (10 µL, 0.5 mg/mL).
The mixture was allowed to react at 37 °C for 24 h. The modified
DNA was precipitated by addition of cold ethanol and then washed
with ethanol. After drying under nitrogen for 15 min, DNA was
redissolved in 1.0 mL of 10 mM Tris-HCl/5 mM MgCl2 buffer (pH
7.0) and was heated to 37 °C. Aliquots (50 µL) were removed following
incubation for 0.2, 1, 3, 5, 8, 12, and 24 h. The aliquot was immediately
treated with cold ethanol to precipitate DNA, and the supernatant was
isolated and analyzed using the general semiquantitative method. To
estimate the total concentration of adduct in the initial sample, an aliquot
(50 µL) was heated at 90 °C for 1 h to completely depurinate. HPLC
analysis was performed using a gradient system consisting of 100%
solvent A (0.1% acetic acid in water) for 2 min, followed by a 25 min
linear gradient to 20% A, 80% B (33% ACN in isopropanol) that was
held for 10 min. The flow rate was 10 µL/min. An aq solution of N2-
Bn-dGuo (50 µL, 0.5 µM) was added to each sample as an injection
standard. AF adducts and the injection standard were monitored in the
selected reaction monitoring (SRM) mode with the following ion
transitions: N2-Bn-dGuo, m/z 458.2 [M + 1]+ to m/z 242.2 [M -
deoxyribose + 1]+; 20, m/z 336.2 [M + 1]+ to m/z 201.2 [M - Ade]+;
AF-Gua, m/z 352.2 [M + 1]+ to m/z 201.2 [M - Gua]+. The peak
area of each adduct was normalized to that of the injection standard.
General DNA Hydrolysis Procedure. Modified DNA (1.0 mg) was
dissolved in 1 mL of 10 mM Tris-HCl/5 mM MgCl2 buffer (pH 7.0).
DNA was first digested with 1500 units of DNase I (type II, from bovine
pancreas) at 37 °C for 10 min. To the resulting mixture were added
0.1 unit of phosphodiesterase I (type II, from Crotalus adamanteus
venom) and 750 units of alkaline phosphatase. The mixture was
incubated at 37 °C for an additional 60 min.
Reactions of AF with Nucleosides in the Presence of AOR/
NADPH. To a 1.5 mL eppendorf tube were added 0.87 mL of water,
phosphate buffer (100 µL, 1.0 M, pH 7.4); dGuo, dAdo, dCyd, or Thd
(2.0 µmol); AF (0.5 µmol, 10 µL of 50 mM solution in DMSO);
NADP+ (5.0 µmol); glucose-6-phosphate (10.0 µmol); glucose-6-
phosphate dehydrogenase (5 units); and AOR (0.2 nmol). The mixture
was allowed to react at 37 °C for 24 h. The reaction mixture was
analyzed using the general qualitative method.
Half-life time was calculated from the first-order rate constant (t1/2
)
0.693/k), which was obtained from the slope of the plot of ln[c/c0] )
-kt, where t is the reaction time, c is the concentration of adducts
remaining in DNA, and c0 is the initial concentration of adducts.
Cell Culture and Transfection. Human embryonic kidney cells (293
cells) were obtained from ATCC (American Type Culture Collection,
Manassas, VA) and maintained in Eagle’s medium (high glucose,
Invitrogen, Inc., Carlsbad, CA), supplemented with 10% heat-
deactivated fetal bovine serum (Invitrogen, Inc.). Cells were incubated
at 37 °C in a humidified atmosphere containing 5% CO2. Plasmid
pCEP4-rAOR was cloned as previously described (1) and transfected
into 293 cells with Lipofectamine 2000 reagents (Invitrogen, Inc.).
Genomic DNA Extraction. pCEP4- or pCEP4-AOR-transfected 293
cells were seeded onto two 10-cm culture dishes at 6 million cells/
dish. After incubating for 16 h, cells were treated with fresh media
containing 500 nM acylfulvene for 3 h under growth conditions
described above. After washing with phosphate-buffered saline, cells
were lysed in TES buffer (10 mM Tris pH 8.0, 100 mM EDTA, 0.5%
SDS) and incubated at 55 °C for 30 min with the addition of protease
Reactions of AF with Nucleosides in the Absence of AOR/
NADPH. To a 1.5 mL eppendorf tube were added 0.99 mL of water;
phosphate buffer (100 µL, 1.0 M, pH 7.4); dGuo, dAdo, dCyd, or Thd
(2.0 µmol); and AF (0.5 µmol, 10 µL of 50 mM solution in DMSO).
The mixture was allowed to react at 37 °C for 24 h. The reaction
mixture was analyzed using the general qualitative method.
9
2110 J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007

Acylfulvene−DNA Adducts
A R T I C L E S
K at 100 µg/mL. Digested lysates were extracted with an equal volume
of phenol/chloroform and centrifuged at 10 000 × g for 15 min. The
resulting aqueous layer was saved and extracted twice with equal
volumes of ether to remove residual phenol/chloroform. DNA was
precipitated with the addition of 0.8 volume of isopropanol before
centrifugation at 12 000 × g for 15 min. After washing twice with
70% ethanol, DNA pellets were dried under nitrogen for 15 min. DNA
was hydrolyzed using the general DNA hydrolysis method and analyzed
using the general semiquantitative method. HPLC analysis was
performed using a gradient system consisting of 90% MeOH in H2O
for 2 min, followed by a 20 min linear gradient to 50% MeOH in H2O.
A 5 min linear gradient to 5% MeOH in H2O followed, and 5% MeOH
in H2O was held for 20 min. The flow rate was 10 µL/min. AF adducts
were monitored by the selected reaction monitoring (SRM) mode with
the ion transitions as follows: 20, m/z 336.2 [M + 1]+ to m/z 201.2
[M - Ade]+; AF-Gua, m/z 352.2 [M + 1]+ to m/z 201.2 [M - Gua]+;
27, m/z 352.2 [M + 1]+ to m/z 217.2 [M - Ade]+.
reported procedure. The resulting 377 bp DNA was purified on 5%
nondenaturing polyacrylamide gel. AF (20 and 40 µM) was reacted
with 5′-end labeled DNA (2000 cpm) in Hepes buffer (pH 7.4, 10 mM)
in the absence or presence of AOR (10 ng), NADPH (200 µM), and
calf thymus DNA (0.25 mM) as a carrier DNA for 48 h. The reaction
was stopped by addition of absolute ethanol and stored at -20 °C.
DNA was precipitated by cooling the sample in dry ice and centrifuga-
tion at 12 000 rpm. The supernatant was removed, and the DNA pellet
was washed with 70% ethanol (0.5 mL) and dried under a vacuum.
The DNA sample was treated with piperidine (50 µL of 0.1 M) and
heated at 90 °C for 30 min. G and G + A sequencing was carried out
as per Maxam-Gilbert workup.55 The sample was lyophilized using
speedVac and redissolved in 20 µL of water and lyophilized again.
The dried fragmented DNA was dissolved in formamide loading buffer,
heated at 90 °C for 5 min, and loaded in 11% denaturing polyacrylamide
gel. The gel was electrophoresed at 1500 V for 4 h, and the resolved
fragments were analyzed using a phosphorimager.
Supercoiled DNA Relaxation. Supercoiled pBR322 DNA (25 µM
bp) was incubated with AF (10-70 µM) in the absence and presence
of AOR (10 ng) and NADPH (200 µM) in TBE (80 mM) (pH 8.3).
The samples were incubated for 48 h at 37 °C, and the reaction was
stopped by adding loading dye containing 0.1% bromophenol blue,
0.1% Xylene cyanol, and 80% (v/v) formamide solution. The gel was
electrophoresed at 100 V for 3 h in TBE buffer using 0.8% agarose
gel containing 0.5 µg/mL ethidium bromide. The bands were illuminated
under UV light and photographed using Biorad Gel Doc 1000.
DNA Cleavage. Plasmid pBR322 DNA was digested with EcoRI
followed by treatment with alkaline phosphatase, and 5′-end labeling
of linearized DNA was performed using T4 polynucleotide kinase with
γ-32P-ATP at 37 °C for 30 min according to the manufacturer’s protocol.
The reaction was stopped by heating at 65 °C for 10 min. Labeled
DNA was purified by Amersham G-50 spin columns by following the
manufacturer’s protocol. The 5′-end labeled linearized DNA was further
digested with restriction enzyme BamH1 according to the previously
Acknowledgment. We acknowledge support from the Office
of the Dean of the Graduate School of the University of
Minnesota and from the American Cancer Society. J.G. was
supported by a Postdoctoral Fellowship from the Carcinogenesis
and Chemoprevention Program of The Cancer Center, Univer-
sity of Minnesota. S.J.S. was supported in part by Career
Transition Award CA-108604 from the National Cancer Insti-
tute. We acknowledge a reviewer for helpful suggestions
regarding the use of 19 as an adduct precursor.
Supporting Information Available: Synthesis and charac-
terization of 9-21, gHMQC and gHMBC spectra of 20 and
21, Figure S1-S4. This material is available free of charge via
the Internet at http:/pubs.acs.org.
JA0665951
9
J. AM. CHEM. SOC. VOL. 129, NO. 7, 2007 2111