1,2,5,6-Diepoxyhexane and 1,2,7,8-diepoxyoctane cross-link duplex DNA at 5'-GNC sequences

Article abstract of DOI:10.1021/tx960059c

The carcinogenicity of epoxide compounds has been attributed to covalent binding to DNA. Whereas monoepoxides form only monoadducts, diepoxides can form both monoadducts and interstrand cross-links. The latter are believed to be the more significant cytotoxic lesions as diepoxides are frequently more carcinogenic and mutagenic than their monoepoxide analogues. We therefore examined the relative DNA interstrand cross-linking capabilities of several diepoxides with respect to chain length, molecular flexibility, reported carcinogenic potential, and DNA sequences targeted. Using denaturing polyacrylamide gel electrophoresis, we found that 1,2,5,6- diepoxyhexane and 1,2,7,8-diepoxyoctane share the 5'-GNC target sequence previously found for 1,2,3,4-diepoxybutane [Millard, J. T., and White, M. M. (1993) Biochemistry 32, 2120-2124] and that the efficiency of cross-linking this sequence may reflect carcinogenicity. 1,2,5,6-Diepoxycyclooctane, the biologically inactive rigid analogue of 1,2,5,6-diepoxyhexane, was found to be a poor cross-linker of all DNA sequences examined. Moreover, increasing the diepoxyalkane chain length did not result in enhanced cross-linking ability.

Full text of DOI:10.1021/tx960059c

994
Chem. Res. Toxicol. 1996, 9, 994-1000
1,2,5,6-Diep oxyh exa n e a n d 1,2,7,8-Diep oxyocta n e
Cr oss-Lin k Du p lex DNA a t 5′-GNC Sequ en ces
Michael J . Yunes, Sara E. Charnecki, J ennifer J . Marden, and J ulie T. Millard*
Department of Chemistry, Colby College, Waterville, Maine 04901
Received April 2, 1996X
The carcinogenicity of epoxide compounds has been attributed to covalent binding to DNA.
Whereas monoepoxides form only monoadducts, diepoxides can form both monoadducts and
interstrand cross-links. The latter are believed to be the more significant cytotoxic lesions as
diepoxides are frequently more carcinogenic and mutagenic than their monoepoxide analogues.
We therefore examined the relative DNA interstrand cross-linking capabilities of several
diepoxides with respect to chain length, molecular flexibility, reported carcinogenic potential,
and DNA sequences targeted. Using denaturing polyacrylamide gel electrophoresis, we found
that 1,2,5,6-diepoxyhexane and 1,2,7,8-diepoxyoctane share the 5′-GNC target sequence
previously found for 1,2,3,4-diepoxybutane [Millard, J . T., and White, M. M. (1993) Biochemistry
32, 2120-2124] and that the efficiency of cross-linking this sequence may reflect carcinogenicity.
1,2,5,6-Diepoxycyclooctane, the biologically inactive rigid analogue of 1,2,5,6-diepoxyhexane,
was found to be a poor cross-linker of all DNA sequences examined. Moreover, increasing the
diepoxyalkane chain length did not result in enhanced cross-linking ability.
Sch em e 1. Meta bolism of 1,3-Bu ta d ien e to
Diep oxybu ta n e a n d Oth er Com p ou n d s in th e Liver
by Cytoch r om e P 450 (P 450), Glu ta th ion e (GSH),
Glu ta th ion e S-Tr a n sfer a se (GT), a n d Ep oxid e
Hyd r ola se (EH)
In tr od u ction
Butadiene (C4H6) is an important monomer in the
production of synthetic rubber, nylon, and other polymers
(1). The National Institute for Occupational Safety and
Health has estimated that 65 000 workers are exposed
to this compound annually (2). Butadiene is also a
component of cigarette smoke (3), plastics, and gasoline
(4). Upon inhalation and absorption, butadiene is me-
tabolized in the liver to its mono- and diepoxides (Scheme
1; 5-7). The metabolites 3,4-epoxy-1-butene and 1,2,3,4-
diepoxybutane (DEB,1 1) alkylate DNA, producing point
mutations, deletions, and cancer (8-10). Indeed, epide-
miological studies indicate that excess mortality from
lymphatic and hematopoietic cancers occurs among work-
ers exposed to butadiene (11-13). Other epoxides are
found as pesticides, sterilizing agents, mycotoxins, mam-
malian metabolic products, and air pollutants (9, 14, 15).
Because of the prevalence of potentially hazardous ep-
oxides surrounding living organisms, we are studying
their biochemical mode of action.
stretch only 4 Å, leaving some doubt about its capacity
for interstrand cross-linking (19, 9). However, DNA
demonstrates considerably greater conformational flex-
ibility than previously anticipated: nitrogen mustard
(20-22) and DEB (23) both cross-link duplex DNA
through distal guanines at the sequence 5′-GNC (where
N is any base) in preference to either 5′-GC or 5′-CG
sequences. Despite the different ultimate biological
effects of the carcinogenic DEB and the chemotherapeutic
nitrogen mustard, they share the same genomic target.
Other molecular factors must therefore be involved in
determining the consequences of DNA cross-linkers in
vivo.
Diepoxides were first postulated to act as biological
cross-linking agents in 1951 (8). Later, Brookes and
Lawley proposed that DEB binds to the N7 position of
guanine, similarly to nitrogen mustards, the first clini-
cally utilized antitumor drugs (16). They further sug-
gested that such bifunctional alkylating agents must
span at least 8 Å to cross-link the minimal N7-to-N7
distance, contained at the duplex sequence 5′-GC (17).
Despite the isolation of conjugate 2 from DEB-treated
DNA hydrosylates (18), the alkyl chain of DEB can
With the goal of finding a correlation to biological
activity, we examined the relative cross-linking efficien-
cies of several diepoxides with respect to alkyl chain
length, molecular flexibility, reported carcinogenicity, and
DNA sequences targeted. We used denaturing polyacryl-
amide gel electrophoresis to verify the cross-linking
capability of the putative cross-linkers 1,2,5,6-diepoxy-
hexane (DEH, 3) and 1,2,7,8-diepoxyoctane (DEO, 4) as
well as to demonstrate the poor cross-linking capability
* Author to whom correspondence should be addressed. Email:
jtmillar@colby.edu; Phone: (207) 872-3311; FAX: (207) 872-3555.
X Abstract published in Advance ACS Abstracts, August 1, 1996.
Abbreviations: DEB, 1,2,3,4-diepoxybutane; DEH, 1,2,5,6-diep-
oxyhexane; DEO, 1,2,7,8-diepoxyoctane; DECO, 1,2,5,6-diepoxycyclooc-
tane; PAGE, polyacrylamide gel electrophoresis.
1
S0893-228x(96)00059-8 CCC: $12.00 © 1996 American Chemical Society

Diepoxyalkanes Cross-Link at 5′-GNC
Chem. Res. Toxicol., Vol. 9, No. 6, 1996 995
Ta ble 1. DNA Du p lexes Used in Th ese Stu d ies
Ta ble 2. In ter a tom ic Dista n ces betw een Ep oxid e Gr ou p s
(fr om r ef 19)
compd
greatest separation (Å)
diepoxybutane
diepoxycyclooctane
diepoxyhexane
diepoxyoctane
4.0
4.3
6.6
8.9
DNA Clea va ge Rea ction s. The DNA GGGCGGG (Table
1) was 3′- or 5′-end radiolabeled on the top strand. Gel-purified
cross-linked DNA or monoalkylated DNA (an aliquot of a cross-
linking reaction used without further purification) was heated
at 90 °C in aqueous piperidine to cleave at sites of N7 alkylation
(29). The resulting fragments were purified to single-base
resolution through 25% denaturing PAGE run as above at 65
°C. Gels were dried (Hoefer Drygel Sr.) onto Whatman 3MM
paper, and autoradiograms were obtained through overnight
exposure. Densitometry (Hoefer GS-300) data were smoothed,
plotted, and integrated (Hoefer Scientific GS370 Densitometry
Program, Version 2.0) to obtain plots of cleavage intensity for
each nucleotide. Bands were assigned through reference to
Maxam-Gilbert G-lanes (29). Total monoalkylation efficiency
was calculated by subtracting the intensity of the peak corre-
sponding to the remaining single strands from the sum of the
intensities of all the peaks.
of 1,2,5,6-diepoxycyclooctane (DECO, 5). Our results
suggest that cross-linking efficiency may be related to
the carcinogenic potential of these agents.
Syn th esis of Diep oxid es. We used the epoxidation method
of Paul and Tchelitcheff (30) to synthesize 1,2,5,6-diepoxyhex-
ane. A solution of 1,2,5,6-hexadiene (4.7 g, 57 mmol) and
m-chloroperoxybenzoic acid (33.9 g, 196 mmol) in 150 mL of
dichloromethane was mixed at 4 °C for 15 h. Following vacuum
filtration, a dilute NaOH solution was added to neutralize the
acid byproduct of the reaction. The organic phase was sepa-
rated, refiltered, and dried with anhydrous sodium sulfate. The
residue was run through a silica gel column with a 3:7 ether/
hexane solvent system, and three fractions were collected. IR
analysis showed the desired 1,2,5,6-diepoxyhexane to be in
fraction three. Purity was 99% as monitored by GC/MS,
neglecting response factors. Synthesis of 1,2,4,5-diepoxypentane
was unsuccessful by this method, as well as through epoxidation
with magnesium monoperoxyphthalate hexahydrate (31), so this
compound could not be used in our studies. 1,2,3,4-Diepoxybu-
tane, 1,2,7,8-diepoxyoctane, and 1,2,5,6-diepoxycyclooctane were
all purchased from Aldrich (Milwaukee, WI).
Exp er im en ta l P r oced u r es
P r ep a r a tion of Ra d iola beled DNA Du p lexes. Oligo-
nucleotides shown in Table 1 were purchased from Operon
Technologies, Inc. (Alameda, CA), and purified through dena-
turing polyacrylamide gel electrophoresis (PAGE) as described
previously (23). Radioisotopic labeling of 3′-ends was with
[R-32P]dATP/Klenow fragment; 5′-ends, with [γ-32P]ATP/T4 poly-
nucleotide kinase under standard conditions followed by ethanol
precipitation with sodium acetate (24).
Rea ction w ith Diep oxid es. Radiolabeled duplex DNA (0.5
OD260) was incubated in 0.3 mM sodium acetate (pH 5) and 250
mM diepoxide (100 µL total volume) for 2 h at 37 °C. Identical
products, although in lower yield, were found when the reaction
medium was Tris-EDTA buffer (pH 7.2). Epoxide reactions have
been reported to be acid-catalyzed (25). The slow reaction rate
of diepoxides relative to other cross-linkers, such as nitrogen
mustards, necessitated the high concentrations used (26).
Incubation was followed by ethanol precipitation and lyophiliza-
tion. The resulting pellet was dissolved in 100 µL of purified
water and a 5 µL aliquot removed for Cerenkov counting to
determine total cpm in the reaction. The remaining sample was
lyophilized for denaturing PAGE.
P u r ifica tion of Cr oss-Lin k ed DNA th r ou gh Den a tu r in g
P AGE. Cross-linked samples were dissolved in 5 M aqueous
urea/xylene cyanole and loaded onto a 20% polyacrylamide gel
(19:1 acrylamide/bis(acrylamide), 50% urea, 0.35 mm thick, 41
× 37 cm), run on a Hoefer Poker Face sequencer at 65 W.
Autoradiography was used to visualize single-stranded and
cross-linked DNA, of roughly half the mobility of the corre-
sponding single strands. The major cross-linked band was
excised from the gel and Cerenkov counted. For poorly cross-
linked duplexes, the region of the gel where centrally cross-
linked product would be expected was located by comparison to
a mitomycin C cross-linked DNA standard (23, 27, 28). The
percentage of cross-linked DNA was determined from the ratio
of cpm in the gel slices to the total cpm in the reaction. DEB,
DEH, DEO, and DECO reactions were all run simultaneously
in duplicate with the same DNA sequence on each gel to
minimize variability between samples. Multiple trials were
performed and standard deviations calculated as reported.
Resu lts
Diepoxide reactivity toward DNA was monitored ac-
cording to length of the carbon chain (Table 2). We
compared interstrand cross-linking by DEB, DEH, and
DEO. Moreover, we considered conformational flexibility
by including the rigid diepoxycycloalkane DECO. We
also examined the monofunctional compound ethylene
oxide (CH2CH2O) in addition to these bifunctional agents
although diepoxides are frequently more carcinogenic and
mutagenic than their monoepoxide analogues (19, 9). For
example, the mutagenicity of 3,4-epoxy-1-butene is much
lower than that of DEB (10, 32). Moreover, DEB has
been found to be some 30 times more effective, per
primary alkylation, at generating mutations than eth-
ylene oxide, despite similar reaction rates (9).
The first step in a cross-linking event is the formation
of a monoalkylated product. We tested the importance
of the initial alkylation step for a variety of epoxides on
a
32P end-labeled DNA with a core sequence of 5′-
d(GGGCGGG). Because cross-linking efficiency is low,
we used the distribution of total alkylation to ap-
proximate monoalkylation frequency. Following reaction
with epoxide and subsequent piperidine treatment (to
cleave at N7-alkylated guanines), we purified the result-
ing fragments to single-base resolution through 25%
denaturing PAGE and assigned bands via a Maxam-

996 Chem. Res. Toxicol., Vol. 9, No. 6, 1996
Yunes et al.
Ta ble 3. Mon oa lk yla tion by Ep oxid es: P er cen t Tota l
Alk yla tion of th e GGGCGGG Du p lex (See Ta ble 1) a n d
Aver a ge Rela tive P er cen ta ges of Mon oa d d u ct F or m a tion
a s a F u n ction of Nu cleotid e P osition a
Ta ble 4. Aver a ge Cr oss-Lin k in g Efficien cies of DNA’s
w ith Sequ en ces Sh ow n in Ta ble 1a
core sequence
agent
TGCAb
TCGAb
GGCC
GATC
% cleavage at
DEB
0.08 ( 0.02 0.09 ( 0.01 0.98 ( 0.37
0.09 ( 0.04 0.02 ( 0.00 0.34 ( 0.16
0.05 ( 0.05 0.07 ( 0.06 1.08 ( 0.50
0.86 ( 0.33
0.08 ( 0.02
0.30 ( 0.11
% total
G1 G2 G3 G4 G5 G6 alkylation
DEH
DEO
DECO
agent
ethylene oxide
DEB
DEH
DEO
DECO
20
22
21
20
16
16
18
18
16
17
17
13
13
15
20
17
17
19
17
14
15
19
16
16
17
15
11
13
16
16
NDb
37
ND
15
0.016 ( 0.004 0.014 ( 0.003
a
Values indicate percent interstrand cross-linked product
relative to total starting material (see Experimental Procedures).
b
From Cerenkov counting of gel region where centrally cross-
12
linked material is expected.
a
Numbering is from 5′-3′. Piperidine cleavage products of both
b
5′- and 3′-end-labeled DNA were averaged. Not determined.
F igu r e 2. Densitometric analysis of piperidine cleavage prod-
ucts of DEO-cross-linked DNA (3′-end radiolabel: * ) 32P).
Lettering denotes the residue cleaved. Cleavage bands of equal
intensities indicated linkage between G8 and G9 on opposite
strands, or cross-linking at the sequence 5′-d(GNC) as shown.
Oligonucleotides containing central 5′-TGCA and 5′-
TCGA sequences were cross-linked extremely poorly,
with yields of less than 0.1%, for all diepoxides tested
(Table 4). This agrees with previous results for DEB,
which also demonstrates poor reactivity for cross-linking
5′-GC and 5′-CG sequences (23).
The oligonucleotide containing the core 5′-GGCC showed
substantial reactivity (Table 4). This DNA is cross-linked
through distal guanines at the sequence 5′-GNC by DEB
(23). We verified this connectivity for DEO and DEH
through isolation and subsequent piperidine cleavage of
the major cross-linked band. The electrophoretic mobili-
ties of the resulting radiolabeled fragments corresponded
to those of a Maxam-Gilbert G reaction. Equal intensi-
ties of the piperidine cleavage bands for both guanines
confirmed that N7 alkylation occurred at the sequence
5′-GNC in this duplex (Figure 2).
The duplex containing 5′-GATC, corresponding to
reaction of distal guanines at the sequence 5′-GNNC, also
showed significant cross-linking by DEB and DEO (Table
4). Complete piperidine cleavage of the major cross-
linked band verified the guanine N7 connectivity. DEH
(∼0.1%) and DECO (∼0.01%) both showed extremely poor
cross-linking of the 5′-GNNC sequence.
All the agents studied demonstrated greater cross-
linking of the duplex containing the sequence 5′-GNC
than those containing 5′-GNNC, 5′-GC, or 5′-CG. How-
ever, DEO and DEB showed substantial reactivity toward
the 5′-GNNC sequence as well. To verify the core
preference and control for the experimental variability
involved in comparing distinct duplexes, we presented
all of these sequences in a single duplex with the core
5′-d(GGGCGGG). DEB- and DEO-cross-linked DNA’s
were piperidine cleaved, and the resulting products were
analyzed through single-base resolution denaturing PAGE.
Integration of the densitometry data produced relative
percentages of the cleavage products, which were aver-
aged for both 5′- and 3′-radiolabeled duplex DNA (Table
5). Relative product abundance followed the order G2 >
F igu r e 1. Denaturing PAGE analysis of radiolabeled DNA
GGCC (see Table 1) incubated for 2 h with 250 mM DEB (lane
1), 250 mM DEO (lane 2), and 0 mM diepoxide in 0.3 M NaOAc
(pH 5) (lane 3).
Gilbert G sequencing reaction. Relative cleavage intensi-
ties at each position correspond to alkylation frequency.
The average monoalkylation frequency is fairly uniform
among dG residues in this duplex for all epoxides tested
(Table 3). Any interstrand cross-linking preference
exhibited by these epoxides must therefore occur at the
second step of cross-link formation: the conversion of
monoadducts to cross-links. This agrees with our previ-
ous results for DEB (23). However, the relative efficien-
cies for monoalkylation differed for the diepoxides tested,
with DEB > DEO > DECO.
We then incubated several duplex DNA sequences with
DEH, DEO, and DECO to verify their cross-linking
capabilities. Oligonucleotides with different guanine-
containing four-base cores (TGCA, TCGA, GGCC, and
GATC; see Table 1 for sequences) were exposed to
diepoxides and analyzed by 20% denaturing PAGE.
These reactions were extremely heat-sensitive, so samples
were loaded directly onto gels without heat denaturing.
In addition to recovering chiefly single strands and
presumably monoadducts, less mobile products were also
formed in these reactions (Figure 1). These somewhat
diffuse bands of reduced mobility are indicative of agents
with some flexibility in their sequence specificity, such
as the pyrrolizidine alkyloids (33). The major product
was quantified in order to distinguish between the
relative cross-linking efficiencies of the variable core
sequences in the duplexes studied. A mitomycin C
centrally cross-linked reference was used to pinpoint the
appropriate area of the gel for poorly cross-linked du-
plexes (23, 28).

Diepoxyalkanes Cross-Link at 5′-GNC
Chem. Res. Toxicol., Vol. 9, No. 6, 1996 997
Ta ble 5. Aver a ge Rela tive P er cen ta ges of Cr oss-Lin k in g
by Diep oxybu ta n e a n d Diep oxyocta n e a s a F u n ction of
Nu cleotid e P osition in th e Du p lex GGGCGGG (See Ta ble
1)a
of a DNA cross-linker, it contains a chain of only four
carbon atoms, which is unlikely to span the putative 8 Å
distance between reactive groups in B-DNA. Lawley and
Brookes pose this dilemma, but do not resolve it (19). It
has been reiterated in the literature that N7-to-N7 cross-
linking between deoxyguanines is impossible for diep-
oxides shorter than at least seven carbon atoms (40). For
example, the high frequency of multilocus deletions
among DEO-induced Neurospora crassa mutants relative
to diepoxypentane-induced mutants has been attributed
to DNA cross-linking: diepoxypentane was considered
unlikely to cross-link N7 of guanines because its alkyl
chain can extend only 5.2 Å (41). We therefore conducted
a systematic study on the effect of diepoxide alkyl chain
length to resolve the question of their capacities for
interstrand cross-linking. DNA’s of defined sequence
were used to determine possible sequence preferences for
cross-linking, which has not been possible in previous
studies demonstrating DEB cross-linking in whole cells
(42-44).
We compared cross-linking efficiencies of diepoxybu-
tane, diepoxyhexane, diepoxyoctane, and diepoxycyclooc-
tane for several DNA sequences. Somewhat diffuse
bands on denaturing polyacrylamide gels indicated that
cross-links resulted from linkage of central dG residues
and also, to a lesser extent, of other residues. Hyperre-
activity with duplex termini has been noted for several
cross-linking agents which do not possess a rigid se-
quence specificity, presumably because the reactivity of
the terminal residues is higher than that of those in the
duplex regions (33). Terminal cross-links were present
to a larger extent in DNA’s that were poorly cross-linked.
The major product was quantified to determine the
relative reactivities of the tested core sequences.
Quantifying the centrally dG-cross-linked material led
to the following order of reactivity for all of the diepoxides
tested: 5′-GGCC > 5′-GATC . 5′-TGCA, 5′-TCGA. This
is consistent with a shared preference for the sequence
5′-GNC, although other sequences are also cross-linked
less efficiently. Therefore, despite their different chain
lengths, DEB, DEH, DEO, and the nitrogen mustard
mechlorethamine all target 5′-GNC preferentially.
The relative cross-linking efficiencies for the preferred
5′-GNC sequence followed the order DEB, DEO > DEH
> DECO. Nitrogen mustard interstrand cross-linking
efficiency (45) and sequence specificity (46) have been
correlated with biological activity. It is difficult to make
this comparison for diepoxides because of some discrep-
ancies in the literature as to the relative carcinogenicities
of these compounds. Mutagenicity has been correlated
to carcinogenic activity (10, 47, 48), and several reports
confirm both the carcinogenicity and mutagenicity of
DEB (9, 49). For example, DEB is a moderately active
carcinogen in rodents after skin application (14, 50),
intraperitoneal injection (51), and subcutaneous injection
(52, 53). DEH has been classified as a weak carcinogen
upon subcutaneous injection in mice and rats (52) and
skin painting in mice, yet it induced about twice as many
cancers as DEO, although at twice the concentration (54).
The mutagenicity of DEH is greater than that of DEO
in Salmonella typhimurium (55). Furthermore, biological
activity has been reported to decrease with chain length
from the four- to the ten-carbon diepoxides (56-58).
However, it has also been reported that DEB, DEH, and
DEO have about the same carcinogenic effectiveness (9).
It is clear that diepoxycyclooctane is inactive in mam-
malian systems (14, 19, 54, 59) as well as in the Ames
% cleavage at
agent
G1
G2
G3
G4
G5
G6
DEB
DEO
23
24
44
31
9
14
8
9
7
9
9
14
a
Numbering is from 5′ to 3′ end. Piperidine cleavage products
of both 5′- and 3′-end-labeled DNA were averaged.
G1 > G3, G4, G5, G6, indicating that both diepoxides
again showed preferential cross-linking of the 5′-GNC
sequence. The next most favored site for cross-linking
was at 5′-GNNC. The other sequences, 5′-GC, 5′-CG, 5′-
CGG, and 5′-CGGG, were all poorly cross-linked. Thus,
despite its significantly longer chain length, DEO was
confirmed to share the sequence preference of DEB.
However, DEO showed only a 1.5-fold preference for
cross-linking 5′-GNC relative to 5′-GNNC in this duplex,
reduced from the 2-fold preference demonstrated by DEB.
Radiolabeling of the bottom strand verified linkage to the
single central G; virtually complete piperidine fragmen-
tation in all cases verified that cross-linking was in fact
from guanine N7-to-N7 positions.
Discu ssion
Goldacre et al. (34) first suggested almost fifty years
ago that the cytotoxicity of bifunctional alkylating agents
might be related to their DNA cross-linking reactions.
Although attempts to correlate DNA binding with car-
cinogenicity have generally been successful, some me-
tabolites bind DNA without being carcinogenic (35).
Clearly, investigations of chemically similar agents with
differing biological effects are needed to understand the
basis of carcinogenesis. Previously, we compared the
DNA cross-linking preferences of diepoxybutane to the
chemically analogous nitrogen mustard mechlorethamine
(23). We found that, despite their vastly different bio-
logical effects, these compounds target the same genomic
site: 5′-GNC. Interstrand cross-linking sequence prefer-
ences alone therefore cannot explain their differing
biological activities. In this study, we compared the DNA
cross-linking efficiencies of a variety of diepoxides, also
with differing biological consequences, to elucidate a
molecular rationale for their resulting activities.
We first tested monoalkylation by epoxides to deter-
mine the relative importance of this step and the
subsequent formation of cross-links. Monoalkylation can
be sequence specific, such as for mitomycin C (36, 37),
or sequence random. Many carcinogens, including ni-
trogen mustard, demonstrate a preference for runs of
contiguous guanines, attributed to increased nucleophi-
licity (38, 39). However, epoxides do not seem to follow
this pattern. The monoepoxide ethylene oxide and the
diepoxides DEB, DEH, DEO, and DECO all showed fairly
uniform alkylation patterns among the guanines in the
sequence 5′-GGGCGGG. Any sequence preference for
cross-linking must therefore arise at the step of monoad-
duct conversion to cross-link. Interestingly, there was a
difference in relative monoalkylation abilities, with DEB
yielding 2- to 3-fold more alkylated product than DEO
and DECO.
The effect of the chain length of bifunctional alkylating
agents was first pondered almost thirty years ago (18).
Although diepoxybutane exhibits the biological properties

998 Chem. Res. Toxicol., Vol. 9, No. 6, 1996
Yunes et al.
geometry of the cross-linked adducts for these diepoxides
to test this theory.
test (60). Therefore, carcinogenic potential seems to
follow the order DEB > DEH > DECO, with DEO
somewhere in the middle. Cross-linking efficiency also
follows this order, with the possible exception of DEO.
In summary, we have demonstrated that the putative
cross-linkers DEH and DEO cross-link DNA through the
N7 of distal deoxyguanines at the 5′-d(GNC) sequence,
also the preferred site for DEB and mustard cross-
linking. DECO was a poor cross-linker of all sequences
tested. Cross-linking efficiencies, and not the sequences
targeted, of these diepoxides correlate with their carci-
nogenicities. Moreover, increasing the alkyl chain length
did not result in enhanced diepoxide cross-linking ef-
ficiency.
It is perhaps surprising that cross-linking efficiency
does not increase with diepoxyalkane length. Therefore,
it appears that chain length does not dictate the weak
interstrand cross-linking capacity of DEB relative to
nitrogen mustard, as previously postulated (42), and that
other factors must be involved in mustard’s enhanced
efficiency.
Diepoxycyclooctane can be viewed as a diepoxyhexane
molecule containing epoxide functions held rigid with
respect to one another by an ethylene bridge. The
chemical reactivity of these two compounds is similar
(19). However, whereas DECO is completely inactive as
a carcinogen, its open-chain flexible analogue DEH is
somewhat carcinogenic (52, 54). Moreover, we found that
DECO is a poor cross-linker relative to DEH. DECO
cross-links the sequence 5′-GGCC about 5% as well as
does DEH. Therefore, flexibility of reaction centers
seems to be important for both carcinogenicity and DNA
cross-linking.
Ack n ow led gm en t. We thank the following for gen-
erous financial support: the donors of the Petroleum
Research Fund, administered by the ACS (28004-GB4);
a Bristol-Myers Squibb Company Award of Research
Corporation (C-3279); the Clare Boothe Luce Fund; a
Supplemental Award of the Camille and Henry Dreyfus
Foundation Scholar/Fellow Program for Undergraduate
Institutions; and the Howard Hughes Medical Institute.
We also thank Professor Bradford Mundy for his kind
assistance with organic syntheses and Professor Thomas
Shattuck for helpful suggestions.
To control for the inherent errors involved in cutting
and quantifying gel bands, the cross-linking sequence
preference of DEO was confirmed with a single DNA
duplex containing a 5′-GGGCGGG core. The relative
sequence reactivities in this duplex were about 3 GNC :
2 GNNC : 1 GC, CG (Table 5). The longer alkyl chain
length of DEO appeared to reduce the absolute cross-
linking specificity, as DEB showed a 2-fold preference for
cross-linking GNC relative to GNNC and a 4-fold prefer-
ence relative to GC and CG (Table 5 and ref 23). Alley
et al. recently reported the reduced cross-linking specific-
ity of 2,5-bis(1-azridinyl)-1,4-benzoquinone relative to
mechlorethamine, also attributed to its longer tether
length (61, 62). The slightly different relative reactivities
in this duplex from those found when comparing the
individual duplexes described above can be attributed to
the impact of flanking bases as well as experimental error
in cutting gel slices. Neighboring bases have been shown
to have a significant impact on cross-linking by other
agents [for example, mitomycin C (37, 63)], although the
influence of flanking sequence is secondary to the abso-
lute core sequence preference for cross-linking.
Refer en ces
(1) Morrow, N. L. (1990) The industrial production and use of 1,3-
butadiene. Environ. Health Perspect. 86, 7-8.
(2) Fajen, J . M., Roberts, D. R., Ungers, L. J ., and Krishnan, E. R.
(1990) Occupational exposure of workers to 1,3-butadiene. Envi-
ron. Health Perspect. 86, 11-18.
(3) Brunnemann, K. D., Kagan, M. R., Cox, J . E., and Hoffmann, D.
(1990) Analysis of 1,3-butadiene and other selected gas-phase
components in cigarette mainstream and sidestream smoke by
gas chromatography-mass selective detection. Carcinogenesis 11,
1863-1868.
(4) Mehlman, M. A., and Legator, M. S. (1991) Dangerous and cancer-
causing properties of products and chemicals in the oil refining
and petrochemical industrysPart II: carcinogenicity, mutage-
nicity, and developmental toxicity of 1,3-butadiene. Toxicol. Ind.
Health 7, 207-220.
(5) Malvoisin, E., and Roberfroid, M. (1982) Hepatic microsomal
metabolism of 1,3-butadiene. Xenobiotica 12, 137-144.
(6) Dahl, A. R., Bechtold, W. E., Bond, J . A., Henderson, R. F.,
Mauderly, J . L., Muggenburg, B. A., Sun, J . D., and Birnbaum,
L. S. (1990) Species differences in the metabolism and disposition
of inhaled 1,3-butadiene and isoprene. Environ. Health Perspect.
86, 65-69.
(7) Henderson, R. F., Bechtold, W. E., Sabourin, P. J ., Maples, K.
R., and Dahl, A. R. (1993) Species differences in the metabolism
of 1,3-butadiene in vivo. In Butadiene and Styrene: Assessment
of Health Hazards (Sorsa, M., Peltonen, K., Vainio, H., and
Hemminki, K., Eds.) pp 57-64, IARC Scientific Publications No.
127, IARC, Lyon.
(8) Hendry, J . A., Homer, R. F., Rose, F. L., and Walpole, A. L. (1951)
Cytotoxic agents: II, bis-epoxides and related compounds. Br. J .
Pharmacol. 6, 235-255.
(9) Ehrenberg, L., and Hussain, S. (1981) Genetic toxicity of some
important epoxides. Mutat. Res. 86, 1-113.
(10) De Meester, C. (1988) Genotoxic properties of 1,3-butadiene.
Mutat. Res. 195, 273-281.
(11) Divine, B. J . (1990) An update on mortality among workers at a
1,3-butadiene facilityspreliminary results. Environ. Health Per-
spect. 86, 119-128.
(12) Divine, B. J ., Wendt, J . K., and Hartman, C. M. (1993) Cancer
mortality among workers at a butadiene production facility. In
Butadiene and Styrene: Assessment of Health Hazards (Sorsa,
M., Peltonen, K., Vainio, H., and Hemminki, K., Eds.) pp 345-
362, IARC Scientific Publications No. 127, IARC, Lyon.
(13) Landrigan, P. J . (1993) Critical assessment of epidemiological
studies on the carcinogenicity of 1,3-butadiene and styrene. In
Butadiene and Styrene: Assessment of Health Hazards (Sorsa,
M., Peltonen, K., Vainio, H., & Hemminki, K., Eds.) pp 375-388,
IARC Scientific Publications No. 127, IARC, Lyon.
Considerable DNA distortion has been postulated to
result from an N7-to-N7 cross-link at 5′-GNC by nitrogen
mustard (20, 21, 64) and DEB (23). DNA bending upon
mustard interstrand cross-linking has been experimen-
tally verified (65). The need for a narrowed major groove
may in fact dictate the preferred cross-linking site of
these compounds: X-ray crystallographic analysis con-
firms the intrinsic bending of the sequence 5′-GGCC (66).
It is likely that diepoxide cross-linking also contributes
to net DNA bending, which may have significance in vivo.
The underwinding and bending triggered by the antitu-
mor agent cisplatin are believed to be responsible for
binding of both its intrastrand and interstrand cross-links
by high-mobility group proteins (67, 68). Furthermore,
binding of such proteins could shield cross-linked lesions
from repair and contribute to the mechanism of cytotoxic
action (69). Therefore, the unique three-dimensional
structures generated by diepoxides of varying chain
lengths might result in differential repair processes in
vivo, which could in turn mediate cytotoxicity at the
cellular level. We hope to determine experimentally the
(14) Van Duuren, B. L., Orris, L., and Nelson, N. (1965) Carcinogenic-
ity of epoxides, lactones, and peroxy compounds. Part II. J . Natl.
Cancer Inst. 35, 707-717.

Diepoxyalkanes Cross-Link at 5′-GNC
Chem. Res. Toxicol., Vol. 9, No. 6, 1996 999
(15) Takahashi, A., Endo, T., and Nozoe, S. (1992) Repandiol, a new
cytotoxic diepoxide from the mushrooms Hydnum repandum and
H. repandum var. album. Chem. Pharm. Bull. 40, 3181-3184.
(16) Brookes, P., and Lawley, P. D. (1961) The alkylation of guanosine
and guanylic acid. J . Chem. Soc., 3923-3928.
(17) Brookes, P., and Lawley, P. D. (1961) The reaction of mono- and
di-functional alkylating agents with nucleic acids. Biochem. J . 80,
496-503.
(18) Lawley, P. D., and Brookes, P. (1967) Interstrand cross-linking
of DNA by difunctional alkylating agents. J . Mol. Biol. 25, 143-
160.
(19) Van Duuren, B. L. (1969) Carcinogenic epoxides, lactones, and
halo-ethers and their mode of action. Ann. N.Y. Acad. Sci. 163,
633-651.
(39) Hartley J . A., Bingham, J . P., and Souhami, R. L. (1992) DNA
sequence selectivity of guanine-N7 alkylation by nitrogen mus-
tards is preserved in intact cells. Nucleic Acids Res. 20, 3175-
3178.
(40) Van Duuren, B. L., and Goldschmidt, B. M. (1966) Carcinogenicity
of epoxides, lactones, and peroxy compounds. III. Biological
activity and chemical reactivity. J . Med. Chem. 9, 77-79.
(41) Ong, T.-M., and de Serres, F. J . (1975) Mutation induction by
difunctional alkylating agents in Neurospora crassa. Genetics 80,
475-482.
(42) Verly, W. G., and Brakier, L. (1969) The lethal action of mono-
functional and bifunctional alkylating agents on T7 coliphage.
Biochim. Biophys. Acta 174, 674-685.
(43) J elitto, B., Vangala, R. R., and Laib, R. J . (1989) Species
differences in DNA damage by butadiene: role of diepoxybutane.
Arch. Toxicol., Suppl. 13, 246-249.
(44) Vangala, R. R., Laib, R. J ., and Bolt, H. M. (1993) Evaluation of
DNA damage by alkaline elution technique after inhalation
exposure of rats and mice to 1,3-butadiene. Arch. Toxicol. 67, 34-
38.
(45) Sunters, A., Springer, C. J ., Bagshawe, K. D., Souhami, R. L.,
and Hartley, J . A. (1992) The cytotoxicity, DNA crosslinking
ability and DNA sequence selectivity of the aniline mustards
melphalan, chlorambucil, and 4-[bis(2-chloroethyl)amino] benzoic
acid. Biochem. Pharmacol. 44, 59-64.
(46) Wyatt, M. D., Lee, M., Garbiras, B. J ., Souhami, R. L., and
Hartley, J . A. (1995) Sequence specificity of alkylation for a series
of nitrogen mustard-containing analogues of distamycin of in-
creasing binding site size: evidence for increased cytotoxicity with
enhanced sequence specificity. Biochemistry 34, 13034-13041.
(47) Ong, T.-M., and de Serres, F. J . (1972) Mutagenicity of chemical
carcinogens in Neurospora crassa. Cancer Res. 32, 1890-1893.
(48) McCann, J ., Choi, E., Yamasaki, E., and Ames, B. N. (1975)
Detection of carcinogens as mutagens in the Salmonella/mi-
crosome test: assay of 300 chemicals. Proc. Natl. Acad. Sci. U.S.A.
72, 5135-5139.
(49) IARC (1976) Evaluation of Carcinogenic Risk of Chemicals to Man;
Cadmium, Nickel, and Some Epoxides, Miscellaneous Industrial
Chemicals and General Considerations on Volatile Anaesthetics,
Vol. 11, pp 115-121, IARC, Lyon.
(50) Van Duuren, B. L., Nelson, N., Orris, L., Palmes, E. D., and
Schmitt, F. L. (1963) Carcinogenicity of epoxides, lactones, and
peroxy compounds. J . Natl. Cancer Inst. 31, 41-55.
(51) Shimkin, M. B., Weisburger, J . H., Weisburger, E. K., Gubareff,
N., and Suntzeff, V. (1966) Bioassay of 29 alkylating chemicals
by the pulmonary-tumor response in strain A mice. J . Natl.
Cancer Inst. 36, 915-935.
(52) Van Duuren, B. L., Langseth, L., Orris, L., Teebor, G., Nelson,
N., and Kuschner, M. (1966) Carcinogenicity of epoxides, lactones,
and peroxy compounds. IV. Tumor response in epithelial and
connective tissue in mice and rats. J . Natl. Cancer Inst. 37, 825-
838.
(20) Ojwang, J . O., Grueneberg, D. A., and Loechler, E. L. (1989)
Synthesis of a duplex oligonucleotide containing
a nitrogen
mustard interstrand DNA-DNA cross-link. Cancer Res. 49,
6529-6537.
(21) Millard, J . T., Raucher, S., and Hopkins, P. B. (1990) Mechlor-
ethamine cross-links deoxyguanosine residues at 5′-GNC se-
quences in duplex DNA fragments. J . Am. Chem. Soc. 112, 2459-
2460.
(22) Rink, S. M., Solomon, M. S., Taylor, M. J ., Rajur, S. B., McLaugh-
lin, L. W., and Hopkins, P. B. (1993) Covalent structure of a
nitrogen mustard-induced DNA interstrand cross-link: an N7-
to-N7 linkage of deoxyguanosine residues at the duplex sequence
5′-d(GNC). J . Am. Chem. Soc. 115, 2551-2557.
(23) Millard, J . T., and White, M. M. (1993) Diepoxybutane cross-links
DNA at 5′-GNC sequences. Biochemistry 32, 2120-2124.
(24) Sambrook, J ., Fritsch, E. G., and Maniatis, T. (1989) Molecular
Cloning: A Laboratory Manual, 2nd ed., Cold Spring Harbor
Laboratory, Cold Spring Harbor, NY.
(25) Ross, W. C. J . (1950) The reactions of certain epoxides in aqueous
solutions. J . Chem. Soc., 2257-2272.
(26) Stacey, K. A., Cobb, M., Cousens, S. F., and Alexander, P. (1958)
The reactions of the “radiomimetic” alkylating agents with
macromolecules in vitro. Ann. NY Acad. Sci. 68, 682-701.
(27) Millard, J . T., Weidner, M. F., Kirchner, J . J ., Ribeiro, S., and
Hopkins, P. B. (1991) Sequence preferences of DNA interstrand
crosslinking agents: quantitation of interstrand crosslink loca-
tions in DNA duplex fragments containing multiple crosslinkable
sites. Nucleic Acids Res. 19, 1885-1891.
(28) Millard, J . T., and Beachy, T. M. (1993) Cytosine methylation
enhances mitomycin C cross-linking. Biochemistry 32, 12850-
12856.
(29) Maxam, A. M., and Gilbert, W. (1980) Sequencing end-labeled
DNA with base-specific chemical cleavages. Methods Enzymol. 65,
499-560.
(30) Paul, R., and Tschelitcheff, S. (1948) (Beta)-diethylenic hydro-
carbons. II. synthesis of 3,5-dihydroxy-1-substituted piperidines
from 1,4 pentadiene. Bull. Soc. Chim., 896-900.
(31) Brougham, P., Cooper, M. S., Cummerson, D. A., Heaney, H., and
Thompson, N. (1987) Oxidation reactions using magnesium
monoperphthalate: a comparison with m-chloroperoxybenzoic
acid. Synthesis 11, 1015-1017.
(32) Cochrane, J . E., and Skopek, T. R. (1993) Mutagenicity of 1,3-
butadiene and its epoxide metabolites in human TK6 cells and
in splenic T cells isolated from exposed B6C3F1 mice. In Buta-
diene and Styrene: Assessment of Health Hazards (Sorsa, M.,
Peltonen, K., Vainio, H., and Hemminki, K., Eds.) pp 195-204,
IARC Scientific Publications No. 127, IARC, Lyon.
(33) Weidner, M. F., Sigurdsson, S. Th., and Hopkins, P. B. (1990)
Sequence preferences of DNA interstrand cross-linking agents:
dG-to-dG cross-linking at 5′-CG by structurally simplified ana-
logues of mitomycin C. Biochemistry 29, 9225-9233.
(34) Goldacre, R. J ., Loveless, A., and Ross, W. C. J . (1949) Mode of
production of chromosome abnormalities by the nitrogen mus-
tards. Nature 163, 667-669.
(35) MacLeod, M. C., Evans, F. E., Lay, J ., Chiarelli, P., Geacintov,
N. E., Powell, K. L., Daylong, A., Luna, E., and Harvey, R. G.
(1994) Identification of a novel, N7-deoxyguanosine adduct as the
major DNA adduct formed by a non-bay-region diol epoxide of
benzo[a]pyrene with low mutagenic potential. Biochemistry 33,
2977-2987.
(53) McCammon, C. J ., Kotin, P., and Falk, H. L. (1957) The cancero-
genic potency of certain diepoxides. Proc. Am. Assoc. Cancer Res.
2, 229-230.
(54) Van Duuren, B. L., Langseth, L., Goldschmidt, B. M., and Orris,
L. (1967) Carcinogenicity of epoxides, lactones, and peroxy
compounds. VI. Structure and carcinogenic activity. J . Natl.
Cancer Inst. 39, 1217-1228.
(55) El-Tantawy, M. A., and Hammock, B. D. (1980) The effect of
hepatic microsomal and cytosolic subcellular fractions on the
mutagenic activity of epoxide-containing compounds in the sal-
monella assay. Mutat. Res. 79, 59-71.
(56) Ross, W. C. J . (1953) The chemistry of cytotoxic alkylating agents.
Adv. Cancer Res. 1, 397-449.
(57) Haddow, A. (1958) Chemical carcinogens and their modes of
action. Br. Med. Bull. 14, 79-92.
(58) Kotin, P., and Falk, H. L. (1963) Organic peroxides, hydrogen
peroxide, epoxides, and neoplasia. Radiat. Res. Suppl. 3, 193-
211.
(59) Huang, S. L., Rader, D. N., and Lee, C.-Y. (1978) The association
between mutagenicity and adduct formation of 1,2,7,8-diepoxy-
octane and 1,2,5,6-diepoxycyclooctane. Chem.-Biol. Interact. 20,
333-340.
(60) Frantz, S. W., and Sinsheimer, J . E. (1981) Bacterial mutagenicity
and toxicity of cycloaliphatic epoxides. Mutat. Res. 90, 67-78.
(61) Alley, S. C., Brameld, K. A., and Hopkins, P. B. (1994) DNA-
DNA interstrand cross-linking by 2,5-bis(1-aziridinyl)-1,4-benzo-
quinone: nucleotide sequence preferences and covalent structure
(36) Li, V.-S., and Kohn, H. (1991) Studies on the bonding specificity
for mitomycin C-DNA monoalkylation processes. J . Am. Chem.
Soc. 113, 275-283.
(37) Kumar, S., Lipman, R., and Tomasz, M. (1992) Recognition of
specific DNA sequences by mitomycin C for alkylation. Biochem-
istry 31, 1399-1407.
of the dG-to-dG cross-links at 5′-d(GN
nC) in synthetic oligonucle-
otide duplexes. J . Am. Chem. Soc. 116, 2734-2741.
(38) Said, B., and Shank, R. C. (1991) Nearest neighbor effects on
carcinogen binding to guanine runs in DNA. Nucleic Acids Res.
19, 1311-1316.
(62) Alley, S. C., and Hopkins, P. B. (1994) DNA interstrand cross-
linking by 2,5-bis(1-aziridinyl)-3,6-bis(carbethoxyamino)-1,4-
benzoquinone: covalent structure of the dG-to-dG cross-links in

1000 Chem. Res. Toxicol., Vol. 9, No. 6, 1996
Yunes et al.
calf thymus DNA and a synthetic DNA duplex. Chem. Res.
Toxicol. 7, 666-672.
(63) Millard, J . T., Weidner, M. F., Raucher, S., and Hopkins, P. B.
(1990) Determination of the DNA cross-linking sequence specific-
ity of reductively activated mitomycin C at single-nucleotide
resolution: deoxyguanosine residues at CpG are cross-linked
preferentially. J . Am. Chem. Soc. 112, 3637-3641.
(64) Hopkins, P. B., Millard, J . T., Woo, J ., Weidner, M. F., Kirchner,
J . J ., Sigurdsson, S. Th., and Raucher, S. (1991) Sequence
preferences of DNA interstrand cross-linking agents: importance
of minimal DNA structural reorganization in the cross-linking
reactions of mechlorethamine, cisplatin, and mitomycin C. Tet-
rahedron 47, 2475-2489.
bending models. Proc. Natl. Acad. Sci. U.S.A. 90, 2930-2934.
(67) Bruhn, S. L., Pil, P. M., Essigman, J . M., Housman, D. E., and
Lippard, S. J . (1992) Isolation and characterization of human
cDNA clones encoding a high mobility group box protein that
recognizes structural distortions to DNA caused by binding of the
anticancer agent cisplatin. Proc. Natl. Acad. Sci. U.S.A. 89, 2307-
2311.
(68) Kasparkova, J ., and Brabec, V. (1995) Recognition of DNA
interstrand cross-links of cis-diamminedichloroplatinum (II) and
its trans isomer by DNA-binding proteins. Biochemistry 34,
12379-12387.
(65) Rink, S. M., and Hopkins, P. B. (1995) A mechlorethamine-
induced DNA interstrand cross-link bends duplex DNA. Biochem-
istry 34, 1439-1445.
(66) Goodsell, D. S., Kopka, M. L., and Cascio, D. (1993) Crystal
structure of CATGGCCATG and its implications for A-tract
(69) Brown, S. J ., Kellett, P. J ., and Lippard, S. J . (1993) Ixr1, a yeast
protein that binds to platinated DNA and confers sensitivity to
cisplatin. Science 261, 603-605.
TX960059C