Identification of 3, N 4-etheno-5-methyl-2′-deoxycytidine in human DNA: A new modified nucleoside which may perturb genome methylation

Article abstract of DOI:10.1021/tx200392a

Methylation of cytidine at dCpdG sequences regulates gene expression and is altered in many chronic inflammatory diseases. Inflammation generates lipid peroxidation (LPO) products which can react with deoxycytidine, deoxyadenosine, and deoxyguanosine in D

Full text of DOI:10.1021/tx200392a

Article
pubs.acs.org/crt
Identification of 3,N4-Etheno-5-methyl-2′-deoxycytidine in Human
DNA: A New Modified Nucleoside Which May Perturb Genome
Methylation
Jagadeesan Nair,† Roger W. Godschalk,*,‡ Urmila Nair,† Robert W. Owen,† William E. Hull,†
and Helmut Bartsch†
†German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany
‡School for Nutrition, Toxicology and Metabolism (NUTRIM), Department of Toxicology, Maastricht University, Universiteitssingel
50, P.O. Box 616, Maastricht, The Netherlands
S
* Supporting Information
ABSTRACT: Methylation of cytidine at dCpdG sequences regulates gene
expression and is altered in many chronic inflammatory diseases. Inflammation
generates lipid peroxidation (LPO) products which can react with deoxycytidine,
deoxyadenosine, and deoxyguanosine in DNA to form pro-mutagenic exocyclic
etheno-nucleoside residues. Since 5-methyl-2′-deoxycytidine (5mdC) residues
exhibit increased nucleophilicity at N3, they should be even better targets for LPO
products. We synthesized and characterized 3,N4-etheno-5-methyl-2′-deoxycyti-
dine-3′-phosphate and showed that LPO products can indeed form the
corresponding etheno-5mdC (ε5mdC) lesion in DNA in vitro. Our newly
developed 32P-postlabeling method was subsequently used to detect ε5mdC
lesions in DNA from human white blood cells, lung, and liver at concentrations 4−
10 times higher than that observed for etheno adducts on nonmethylated cytidine.
Our new detection method can now be used to explore the hypothesis that this
DNA lesion perturbs the DNA methylation status.
phenomenon. It is very difficult to discriminate between (a)
reduction or inhibition of the methylation of a newly
synthesized DNA strand and (b) increased removal of methyl
groups from previously methylated dCpdG sequences.
Evidence for the existence of an active DNA demethylation
process has been obtained,6,7 but other studies have suggested
that hypomethylation may also result when DNMTs are
prevented from accessing DNA.8 Continuous formation of
damaged DNA and its repair during chronic inflammation may
be involved in the latter process. The structural characteristics
of a damaged dCpdG sequence may hinder site-specific access
and methylation by a DNMT. It has been shown that oxidative
DNA lesions at certain positions in a dCpdG unit can also
result in increased DNA methylation,9 probably because some
proteins involved in the methylation machinery can be
considered to be “ancient” DNA repair enzymes. For instance,
the enzymes MBD4 (methyl-dCpdG binding domain protein
4)10 and thymine DNA glycosylase11 can both be involved in
the repair of mismatches at dCpdG sites. Thymine DNA
glycosylase is additionally involved in the repair of the etheno-
adduct residues 3,N4-etheno-2′-deoxycytidine (εdC) in dam-
aged DNA.11,12
INTRODUCTION
■
DNA methyl transferases (DNMT) catalyze the covalent
attachment of a methyl group to the C5 position of cytosine
bases in DNA to form 5-methyl-2′-deoxycytidine (5mdC)
residues.1 Most of these methylations occur at the
-p5′dC3′p5′dG3′p- sequence positions and are found to be
related to gene expression. Such dCpdG sequences in the
promoter region of inactive genes are usually methylated,
whereas demethylation at these sites often results in increased
transcriptional activity. Both hypomethylation of the whole
DNA (global hypomethylation) and site-specific hypermethy-
lation have been observed in common human chronic
degenerative diseases, including cancer.2 Currently, it is still
largely unknown how the balance between methylation and
demethylation rates is maintained or altered. No correlations
have been observed between the degrees of hypo- and
hypermethylation, suggesting that the mechanisms responsible
for alterations of methylation patterns are independent of each
other.3−5 Thus, the overall pattern of DNA methylation is
thought to be the result of a dynamic equilibrium between the
rates of DNA methylation and demethylation. Disturbance of
this balance may lead to changes in gene expression and cell
behavior which initiate or promote the early stages of
carcinogenesis.
Although global hypomethylation has been extensively
studied, it is still unclear what processes underlie this
Received: September 8, 2011
Published: December 13, 2011
© 2011 American Chemical Society
162
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
a
Scheme 1. Synthesis of 3,N4-Etheno-5-methyl-2′-deoxycytidine-3′-phosphate (5) and the 32P-Labeled 5′-Phosphate (6)
a
A brief description of the synthesis is provided in Experimental Procedures; details are provided in Supporting Information. The starting compound
1 is available commercially. DMTr = 4,4′-dimethoxytrityl protecting group; T4-PNK = T4 polynucleotide kinase. The conventional atom numbering
scheme for ribose and the pyrimidine bases has been used. For etheno adducts 5 and 6, the atom numbering in the modified base corresponds to the
IUPAC systematic name 8-methylimidazo[1,2-c]pyrimidin-5(6H)-one for the fused ring system. The adduct formation step 4 → 5 can also be
carried out with lipid peroxidation products such as 4-HNE, derived from arachidonic acid, and represents the primary mechanism for adduct
formation in DNA in vivo. Step 5 → 6 represents the 32P-postlabeling reaction used for ultrasensitive detection of etheno adducts in DNA digests.
Promutagenic etheno-DNA adducts, which include the
modified nucleoside residues εdC and 1,N6-etheno-2′-deoxy-
adenosine (εdA), are formed inter alia by the major lipid
peroxidation (LPO) product trans-4-hydroxy-2-nonenal (4-
HNE).13 Reaction products of 4-HNE with 5mdC residues
have not been investigated up to now, but their existence has
been postulated since the nucleophilicity of the N3 position
toward electrophilic LPO products is enhanced in 5mdC
relative to dC residues.
Here we report for the first time that the exocyclic adduct
3,N4-etheno-5-methyl-2′-deoxycytidine (ε5mdC) is indeed
formed when DNA containing 5mdC residues is incubated
with electrophilic LPO-derived products. The new model
nucleotide 3,N4-etheno-5-methyl-2′-deoxycytidine-3′-phosphate
(ε5mdCyd-3′-P) with modified cytosine base was synthesized
and structurally characterized by mass spectrometry (MS) and
nuclear magnetic resonance (NMR). An ultrasensitive 32P-
postlabeling method was developed which facilitates the
detection and quantitation of etheno adducts (εdA, εdC,
ε5mdC) as labeled nucleotides in digests of DNA from human
and animal tissues.
EXPERIMENTAL PROCEDURES
■
Monoclonal Antibodies (Mabs). The Mabs used in this study
were provided by M. F. Rajewsky (Institute of Cell Biology, University
of Essen, Essen, Germany), and their characteristics have been
reported previously.14 The Mab EM-C-1 was raised to recognize the
modified nucleoside εdCyd,14 and it was expected that EM-C-1 will
also recognize the structurally similar ε5mdCyd nucleoside and the
corresponding phosphate nucleotides. Similarly, Mab EM-A-1 was
raised to recognize the nucleoside εdA.
3,N4-Etheno-5-methyl-2′-deoxycytidine-3′-phosphate (5).
The synthesis route is summarized in Scheme 1, and details are
presented in the Supporting Information. As previously described,15
commercially available 5′-O-(4,4′-dimethoxytrityl)-thymidine-3′-[(2-
cyanoethyl)-(N,N-diisopropyl)]-phosphoramidite (1) (Amersham
Pharmacia Biotech Europe GmbH, Freiburg, Germany) was first
converted to 5′-O-(4,4′-dimethoxytrityl)-thymidine-3′-bis(2-
cyanoethyl)phosphate (2) and then to O4-ethyl-5′-O-(4,4′-dimethox-
ytrityl)-thymidine-3′-bis(2-cyanoethyl)phosphate (3). The dimethoxy-
trityl protecting group was removed from the 5′ site using ZnBr2.
Normally, the deprotection of the 3′-phosphate is performed by
incubation with ammonia for less than 3 h. However, in Scheme 1 the
incubation was prolonged to 16 h, and the O4-ethyl moiety on thymine
was efficiently replaced by an NH2 group in almost 100% yield. After
the removal of ammonia from the reaction mixture in vacuo, the
solution was lyophilyzed, and the residue was subjected to reverse-
phase HPLC (instrumentation, Hewlett-Packard, C18-reverse phase
column; mobile phase gradient, 0 to 50% acetonitrile in water over 30
163
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
min; flow rate, 1 mL/min) to give pure 5-methyl-2′-deoxycytidine-3′-
phosphate (5mdCyd-3′-P, 4). This compound was subsequently
incubated with 1 M chloroacetaldehyde to form the etheno-adduct
3,N4-etheno-5-methyl-2′-deoxycytidine-3′-phosphate (ε5mdCyd-3′-P,
5), which was purified by reverse-phase HPLC using the method
given above. For C12H16N3O7P, the exact mass M = 345.0726; LC-
ESI-MS (m/z): [M − H]− calcd. 344.0653; found, 344.1; [2M−H]−
calcd. 689.1379; found, 689.1; [M−(C7H7N3O)−H]− calcd. 195.0064;
found, 195.0 (neutral loss of the modified base ε5mCyt).
Bruker’s TopSpin software after Lorentz−Gauss resolution enhance-
ment.
3,N4-Etheno-5-methyl-2′-deoxycytidine-5′-[32P]phosphate
(6). Compound 5 was phosphorylated and 32P-labeled at the 5′ site
using [γ-32P]ATP (>5000 Ci/mmol) and T4 polynucleotide kinase at
pH 6.8. Under these conditions, the kinase also behaves as a 3′-
phosphatase, thus removing the phosphate group from the 3′ site while
labeling the 5′ site.16 The labeling procedure was performed before and
after purification of 5 on immunoaffinity columns prepared with the
Mab EM-C-1. The 32P-labeled 5′-phosphates were resolved on
polyethyleneimine TLC plates using two-dimensional chromatography
[D1 = 1 M acetic acid (pH 3.5) and D2 = saturated ammonium sulfate
(pH 3.5)]. After autoradiography, adduct spots were marked and cut
out, and the radioactivity was measured in a liquid scintillation
counter. Radiolabeling of the standard ε5mdCyd-3′-P (5) before and
after immuno-purification yielded similar results, indicating that the
immunoaffinity columns prepared with EM-C-1 indeed recognized
and retained ε5mdCyd-3′-P.
1
The H NMR data summarized in Table 1 were obtained at 500
MHz (Bruker AM-500 spectrometer, Bruker BioSpin GmbH,
1
Table 1. H NMR Data for 3,N4-Etheno-5-methyl-2′-
a
deoxycytidine-3′-phosphate (5)
b
pos.
1′
δH (ppm)
mult.
JHH (partner) in Hz
6.560
2.504
2.686
4.834
4.267
3.852
3.921
7.592
7.924
7.679
2.316
dd
6.85 (2′a); 6.40 (2′b), 0.5 (7)
−14.27 (2′b); 6.82 (3′)
3.79 (3′)
2′a (2′)
ddd
ddd
dddd
ddd
dd
32P-Labeling of Nucleotides after DNA Hydrolysis. Native or
pretreated (see below) DNA samples (25 μg) were hydrolyzed to
nucleotide 3′-phosphates using micrococcal endonuclease and spleen
phosphodiesterase.16 Normal nucleotides were quantitated by HPLC
using an aliquot of the digest, while etheno adducts were enriched on
immunoaffinity columns prepared from the Mabs EM-A-1 (recognizes
εdAdo-3′-P) and EM-C-1 (recognizes εdCyd-3′-P and ε5mdCyd-3′-P).
Since EM-C-1 was used to bind the etheno adduct of both methylated
and nonmethylated cytidine, two consecutive columns with EM-C-1
were used to increase total binding capacity, to minimize the
consequences of competitive binding and to ensure the highest
possible retention of both εdC and ε5mdC. The etheno adducts and
the internal standard (2′-deoxyuridine-3′-phosphate) were converted
to radiolabeled 5′-phosphates (5′-P*) using [γ-32P]ATP and T4-
polynucleotide kinase at pH 6.8. The 5′-phosphates were resolved on
polyethyleneimine TLC plates as described above. The absolute
amounts of the adducts were quantitated by liquid scintillation using
standards, and the relative levels or frequencies of the adduct
nucleoside residues (ε5mdC, εdC and εdA) in DNA were calculated
from the measured amounts of the nonadduct nucleotides and listed in
Table 2 as adduct residues per 109 parent residues or per 109 total
residues. The content of 5mdCyd-3′-P in the digest was determined by
labeling an aliquot of the digest after 1000-fold dilution in water and
subsequent separation of 5mdCyd-5′-P* and dCyd-5′-P* by TLC using
the same conditions as those described above (see Figure 2B).
In Vitro Generation of Etheno Adducts in DNA by
Chloroacetaldehyde and by LPO Products of Arachidonic
Acid. Chloroacetaldehyde (CAA) is one of the metabolites of the
human carcinogen vinyl chloride, and it reacts with DNA bases to
2′b (2″)
3′
3.74 (4′); 7.28 (P)
4.84 (5′a); 3.29 (5′b)
−12.61 (5′b)
4′
5′a (5″)
5′b (5′)
dd
2
d
2.00 (3)
3
d
7
q br
d
1.26 (8-Me)
8-Me
a
Five hundred megahertz, 0.5 mg in 0.4 mL of D2O, 30 °C, shifts
relative to HDO = 4.725 ppm; J couplings were estimated from first-
order analysis with a precision of ca. 0.05 Hz. 1H NMR spectrum for
3,N4-etheno-5-methyl-2′-deoxycytidine-3′-phosphate (5) is shown in
b
Supporting Information. Atom numbering as in Scheme 1 and Barrio
et al.;23 for nonequivalent methylene protons, the symbols a and b
refer to the lower and higher ppm values; primes (′ and ″) follow the
conventions used for NMR of deoxyribose derivatives.22 Positions
(pos.), 7 and 8 in the etheno-cytosine base correspond to pos. 6 and 5
in cytosine, respectively, so that 8-Me in the adduct corresponds to 5-
Me in the normal nucleotide.
Rheinstetten, Germany) and 30 °C with a 0.5 mg sample in 0.4 mL
of D2O. Chemical shifts were referenced to the residual HDO signal,
defined as 4.725 ppm. A series of homodecoupling experiments were
performed to assign chemical shifts and couplings. A first-order
analysis of JHH values was made with the multiplet analyis routines in
a
Table 2. Etheno-nucleoside Adduct Residue Levels (Mean SD, n = 3) in Calf Thymus DNA
etheno-nucleoside residue levels [per 109 parent (P) or total (T) residues]
ratios
b
agent [mM]
norm.
εdA
εdC
ε5mdC
ε5mdC/εdA
ε5mdC/εdC
CAA [0]
P
T
P
T
P
T
P
T
P
T
P
T
25 16
403 149
85 31
1362 490
54.5
1.86
15.1
0.51
37.2
1.27
91.0
2.80
68.7
2.22
50.9
1.76
3.38
0.15
7.55
0.36
6.61
0.31
3.40
0.16
3.16
0.15
2.99
0.14
7
5
13
5
CAA [2]
CAA [20]
AA [0]
234 85
68 25
469 278
98 58
3541 1541
35 15
776 102
225 30
4362 1831
916 384
428 44
28841 3853
285 38
16
5
8
1456 227
2
90
9
14
2
AA [0.25]
AA [0.50]
30 21
653 180
137 38
2062 1172
20 12
9
6
74 23
21
1260 789
265 166
3768 1054
37 10
7
a
DNA incubated in vitro with various concentrations of CAA or with LPO products of arachidonic acid (AA), as described in the Experimental
b
Procedure section. Normalization (norm.): adduct residue levels expressed as number per 109 nucleosides of the parent type (P) or total
nucleosides (T).
164
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
form labile hydroxyethano-derivatives which subsequently dehydrate
to give the etheno adducts. Calf thymus DNA (200 μg) in 1 mL of 0.1
M Tris-HCl buffer (pH 7.1) was incubated with increasing
concentrations of CAA (0, 2, and 20 mM) for 30 min at 37 °C.
DNA was precipitated with two volumes of cold ethanol after the
addition of 1/30 volumes of 3 M sodium acetate (pH 5.3). The DNA
was then washed with 70% ethanol, dried, and stored at −20 °C as 25-
μg aliquots.
LPO products from arachidonic acid were generated in the presence
of ferrous sulfate. The reaction mixture contained 0, 0.25, or 0.5 mM
arachidonic acid, 0.75 mM FeSO4 (freshly prepared), 0.1 mM
NADPH, and 200 μg of calf thymus DNA in 1 mL of 0.1 M Tris-
HCl buffer (pH 7.1). Reactions were allowed to proceed at 37 °C for
30 min and were completed by heating for 30 min at 80 °C. The
mixtures were dialyzed overnight against doubly distilled water. DNA
was precipitated, washed, dried, and stored at −20 °C as 25-μg
aliquots, as described above.
Levels of 3,N4-Etheno-5-methyl-2′-deoxycytidine in DNA
from Human Tissues. DNA from (a) two human white blood cell
samples with predetermined high εdA and εdC levels, obtained from
female volunteers on a very high ω-6 polyunsaturated fatty acid diet,17
(b) 10 nontumorous lung samples from nonsmoking lung cancer
patients,18 and (c) two asymptomatic human liver samples with known
adduct levels of εdA and εdC16 were analyzed for the presence of
ε5mdC using our immuno-enriched 32P-postlabeling method as
described above.
RESULTS
■
Synthetic Reference: 3,N4-Etheno-5-methyl-2′-deoxy-
cytidine-3′-phosphate. The title reference compound
ε5mdCyd-3′-P (5) was synthesized according to Scheme 1,
starting from the commercially available thymidine phosphor-
amidite 1, as described briefly in the Experimental Procedures
section and in more detail in Supporting Information. Note: the
atom numbering scheme for the modified base in compound 5
corresponds to IUPAC nomenclature19 where, for example,
3,N4-etheno-5-methyl-2′-deoxycytidine (ε5mdCyd) is defined
as 8-methyl-6-(2′-deoxy-β-D-ribofuranosyl)-imidazo[1,2-c]-
pyrimidin-5(6H)-one or, equivalently, 5,6-dihydro-8-methyl-5-
oxo-6-(2′-deoxy-β-D-ribofuranosyl)-imidazo[1,2-c]pyrimidine.
LC-MS chromatograms, the negative-ion ESI-MS spectrum,
and the UV spectrum for 5 are displayed in Figure 1. The
HPLC chromatograms and the ESI-MS spectrum demonstrate
the high purity of the preparation and confirm the correct
molecular weight. The [M−H]− peak at m/z = 344.1 agrees
well with the calculated exact mass of 344.0653. The UV
absorption spectrum is characteristic for the etheno adduct with
a shoulder at 215 nm and a well-defined peak at 275 nm.
The 1H NMR data (chemical shifts and scalar couplings) are
summarized in Table 1, and the raw data are presented in
Supporting Information. The 2′-deoxy-β-D-ribose-3′-phosphate
moiety is confirmed on the basis of the shifts and couplings,
which are very similar to those reported for the unmodified
nucleotide dCyd-3′-P.20 Furthermore, the proton coupling
pattern indicates that the conformational equilibrium of the
Figure 1. LC-MS and UV characteristics of ε5mdCyd-3′-P (5). (A)
The LC-MS profiles for total-ion current and UV absorption at 278
nm show retention times of ca. 22.2 min for the HPLC conditions
described in Experimental Procedures. (B) The negative-ion ESI-MS
spectrum confirms the expected molecular weight of 345 and displays
a fragment ion derived by neutral loss of the modified base 3,N4-
etheno-5-methyl-cytosine (ε5mCyt). (C) The UV absorption
spectrum (in water/acetonitrile) shows the characteristic peak for
the adduct at ca. 275 nm.
and 8-Me groups (4J = 1.26 Hz) and the presence of the etheno
protons H2 and H3, which exhibit a pair of doublets with 3J2,3
=
2.0 Hz, analogous to data reported for etheno-cytosine (εCyt),
etheno-cytidine (εCyd), and derivatives21 as well as for etheno-
2′-deoxycytidine (εdCyd) and derivatives.19 The assignment of
H3 as being downfield of H2 for both N1-protonated and
unprotonated forms was made by chemical means by Barrio et
al.21 as a revision of earlier work and is consistent with the
expected deshielding due to the proximity of H3 to C5O in
the ring plane. The reverse assignment, H2 as downfield of H3,
for εdCyd was made by Zhang et al.19 without explanation.
Identification and Chromatographic Characteristics
of Nucleoside Adducts. When calf thymus DNA was treated
2
deoxyribose ring between primarily the S-type (2′-endo, T3)
and N-type (3′-endo, 3T2) forms exhibits an S/N ratio of ca. 1.8
(J1′2′/J3′4′), which is typical for deoxyribose pyrimidine
nucleotides in the anti conformation.15 The parameters J2′3′
=
6.8 Hz and J1′2′ + J3′4′ = 10.6 Hz are also consistent with the
mean values (6.6 and 10.9) found for a series of pyrimidine 3′-
nucleotides in which the base adopts primarily the anti
conformation (as shown in Scheme 1).
The modified base, i.e., the 3,N4-etheno-5-methyl-cytosine
moiety, is confirmed on the basis of the mutually coupled H7
165
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
with CAA or arachidonic acid-derived LPO products, etheno-
adducts of nucleoside residues were formed. Following DNA
digestion, the resulting nucleoside-3′-phosphates were con-
verted to 32P-labeled 5′-phosphates as described in the
Experimental Procedures section. The various normal and
modified nucleotides can be separated using 2D-TLC. For
example, three distinct adducts spots were usually found in the
upper-left-hand corner of the TLC chromatogram (autoradio-
graph in Figure 2A). Spots 1 and 2 had been previously and
summarized in the upper part of Table 2 and are expressed
as the number of adduct residues per 109 residues of the parent
nucleoside type or per 109 total nucleoside residues (ppb). For
the DNA used, methylated deoxycytidine (5mdC) represented
4.7% of the total deoxycytidine content. When nucleoside
adduct frequencies were expressed relative to total nucleoside
residues, the ratio ε5mdC/εdC was 0.15 without CAA
treatment and ca. 0.33 after CAA treatment. However, when
individual adduct levels were expressed relative to the number
of the corresponding parent nucleosides, then the ε5mdC/εdC
ratio was 3.4 in untreated DNA and increased to ca. 7 after
CAA exposure. These results prove that methylation of
deoxycytidine at the C5 position results in a higher frequency
of etheno adduct formation, most likely due to the higher
nucleophilicity of the N3 position compared to normal
deoxycytidine.
The etheno-deoxyadenosine adduct εdA was also quantified,
and its levels were similar to those of ε5mdC when evaluated
on the basis of total nucleoside residues. However, on the basis
of parent nucleosides, εdA levels were significantly lower than
those observed for εdC and ε5mdC, with or without CAA
treatment (ε5mdC/εdA ratios in the range 15 to 55). A clearly
positive dose−response relationship for CAA treatment was
observed for all three etheno adducts.
Figure 2. Chromatographic characteristics (2D-TLC) of etheno-
nucleotide adducts (A) and unmodified nucleotides (B) obtained from
DNA digests using the 32P-postlabeling technique. Calf thymus DNA
was treated in vitro with adduct-forming compounds as described in
the text. The nucleoside-3′-phosphates obtained from digests were
converted to 32P-labeled 5′-phosphates as described in the Exper-
imental Procedures. The numbered autoradiographic spots correspond
to (1) 1,N6-etheno-2′-deoxyadenosine-5′-[32P]phosphate (εdAdo-5′-
P*); (2) 3,N4-etheno-2′-deoxycytidine-5′-[32P]phosphate (εdCyd-5′-
P*); (3) 3,N4-etheno-5-methyl-2′-deoxycytidine-5′-[32P]phosphate
(ε5mdCyd-5′-P*, 6); (4) unmodified 2′-deoxycytidine-5′-[32P]-
phosphate (dCyd-5′-P*); and (5) 5-methyl-2′-deoxycytidine-5′-[32P]-
phosphate (5mdCyd-5′-P*) corresponding to compound 4 in Scheme
1.
Lipid peroxidation (LPO) of arachidonic acid (AA) in the
presence of ferrous sulfate and NADPH results in the
production of reactive agents, such as 4-HNE, which can
generate in situ in native DNA the same three etheno-
nucleoside adducts described above, including the newly
identified ε5mdC. The data summarized in the bottom portion
of Table 2 demonstrate that in vitro incubations of LPO
products with calf thymus DNA result in the formation of
etheno adducts in a dose-dependent manner. For AA
concentrations of 0, 0.25, and 0.5 mM, the ε5mdC/εdC ratio
was nearly constant at ca. 0.15, based on total nucleoside
residues, or ca. 3.2, based on parent nucleosides. The
corresponding ε5mdC/εdA ratios decreased from 2.8 to 1.8 or
from 91 to 51, respectively, with increasing AA concentration.
Statistical analysis of the results in Table 2 demonstrates that
the levels of the three described adducts formed during in vitro
DNA incubations exhibit strong linear correlations with the
following coefficients (p < 0.05): εdC vs ε5mdC, r = 0.98; εdA
vs ε5mdC, r = 0.97; εdA vs εdC, r = 0.94. These correlations
suggest that closely related mechanisms are involved in the
formation of these adducts.
Levels of ε5mdC in DNA from Human Tissue and
White Blood Cells. Three types of human samples (stored in
the Biobank of the German Cancer Research Center, DKFZ),
namely, white blood cells (n = 2), liver (n = 2), and lung tissue
(n = 10), had been previously analyzed for εdA and εdC levels
in DNA. These samples were reanalyzed using our ultra-
sensitive 32P-postlabeling method, and ε5mdC was unequiv-
ocally detected in all samples. The number of ε5mdC residues
per 109 parent nucleosides were as follows: 65 and 78 in two
asymptomatic human liver samples; 6415 and 9468 in two
samples of white blood cells from women on a diet high in ω-6
polyunsaturated fatty acid (PUFA, linoleic acid); 31−1865 in
ten nontumorous lung specimens from lung cancer patients
(mean SEM: 397 19). The frequency of the etheno adduct
ε5mdC in DNA from human lung tissue was about 10 times
higher than that for εdC.
unequivocally identified as 1,N6-etheno-2′-deoxyadenosine-5′-
[32P]phosphate (εdAdo-5′-P*) and 3,N4-etheno-2′-deoxycyti-
dine-5′-[32P]phosphate (εdCyd-5′-P*), respectively,16 while
spot 3 was unknown. By chromatographic comparison with
our newly synthesized and labeled standard, compound 6, we
were able to identify spot 3 as 3,N4-etheno-5-methyl-2′-
deoxycytidine-5′-[32P]phosphate (ε5mdCyd-5′-P*).
For quantitative analysis and comparisons, nucleoside adduct
levels in DNA can be expressed relative to the total number of
residues or to the number of unmodified parent nucleoside
residues. As described in Experimental Procedures, the normal
nucleotides in a DNA digest were quantitated by HPLC
methods. The number of unmodified 5mdC residues was
determined by 32P-postlabeling of a highly diluted fraction of
the DNA digest followed by 2D-TLC. In Figure 2B, the small
spot 5 in the upper-right-hand corner of the TLC chromato-
gram could be assigned to 5-methyl-2′-deoxycytidine-5′-[32P]-
phosphate (5mdCyd-5′-P*), while the larger spot 4 corre-
sponds to normal 2′-deoxycytidine-5′-[32P]phosphate (dCyd-5′-
P*).
Quantitation of Etheno-Nucleoside Residues in DNA
Treated with Adduct-Forming Agents in Vitro. The 32P-
postlabeling technique was used to determine the levels of three
etheno-nucleoside residues (ε5mdC, εdC, and εdA) in
untreated calf thymus DNA and after its exposure to 2 mM
and 20 mM chloroacetaldehyde (CAA). The results are
For the 14 samples studied, a statistically significant positive
correlation was observed between the formation of εdC and
166
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
ε5mdC (r = 0.97, p < 0.01, Figure 3). A weaker but significant
correlation was also found for εdA and ε5mdC (r = 0.79, p <
etheno-DNA adducts) were found in one study to be affected
by the methylation status of dCpdG islands.27
Since etheno adducts can be formed in dCpdG sequences
(on dC, 5mdC, and dG residues), one could argue that these
types of DNA damage play a role in the alteration of the global
methylation status of DNA and, thus, a role in the onset of
chronic degenerative diseases. To this end, we synthesized and
characterized the new etheno adduct ε5mdCyd-3′-P (3,N4-
etheno-5-methyl-2′-deoxycytidine-3′-phosphate), compound 5,
and detected the presence of the corresponding modified
ε5mdC residues in tissue DNA and in DNA treated in vitro
with CAA or LPO-derived products of AA. The frequency of
ε5mdC residues was relatively high (6−9 per 106 5mdC) in two
WBC samples from women on a high ω-6 PUFA diet. In 10
nontumorous lung samples from lung cancer patients, the
ε5mdC/εdC ratio was about 10:1. Such “massive” DNA
damage could alter the methylation profile in different ways.
For instance, the interaction between DNMT and dCpdG
islands could be affected, thus altering the methylation rate, or
the removal of the damaged base by DNA repair processes
would simultaneously remove 5mdC, for instance by (long-
patch) nucleotide excision repair, as described for the removal
of the 4-HNE-induced etheno-dG adduct.28
Etheno adducts (εdC and εdA) have been found to be
strongly elevated in human tissues with premalignant
conditions related to chronic inflammatory processes, including
cirrhosis, ulcerative colitis, Crohn’s disease, and familial
adenomatous polyposis.13 These patients are at high risk for
developing cancer in the affected organ. Chronic inflammatory
processes that occur in hepatocellular cirrhosis and ulcerative
colitis are accompanied by marked methylation changes in
dCpdG islands in the affected but still nonmalignant tissues.29
Taken together, these findings support the hypothesis that pro-
inflammatory stimuli leading to LPO-derived DNA damage
could account, in part, for the epigenetic changes observed
during carcinogenesis. In many types of cancer, the expression
of DNMT1 was found to be increased,30 although global
hypomethylation was still observed. It is currently not
understood why increased DNMT1 expression does not lead
to normal methylation levels or even hypermethylation. The
presence of ε5mdC residues may reduce the accessibility of
DNA for DNMT’s and, thus, result in hypomethylation.
In the present work, we did not determine the frequency of
ε5mdC in colon tissue from cancer-prone patients. However,
εdA and particularly εdC have been found to be significantly
increased in subjects suffering from ulcerative colitis, Crohn’s
disease, and familial adenomatous polyposis.13 Previously, we
observed in DNA digests from colon samples an adduct spot
that migrated at the same position as the ε5mdCyd-3′-P
standard, but our new ultrasensitive detection method was not
then available to allow quantitative analysis. Still, from these
observations it can be inferred that high levels of ε5mdC are
present in colon DNA from these patients. In the future, our
32P-postlabeling method will allow the reliable determination of
etheno adduct levels in the DNA obtained from small tissue
biopsies.
Figure 3. Correlation between the formation of the etheno adduct
residues ε5mdC and εdC in human DNA in vivo. Numbers of adduct
9
▲
)
residues per 10 parent residues are plotted on a log−log scale for (
WBC of female volunteers on a high ω-6 polyunsaturated fatty
(linoleic) acid diet (n = 2); ( ) asymptomatic human liver tissue
biopsies (n = 2); and ( ) nontumorous surgical lung specimens from
●
○
nonsmoking lung cancer patients (n = 10). For all 14 samples, the
correlation coefficient is r = 0.97 (p < 0.01).
0.01, data not shown). The correlations observed for these
three DNA lesions suggest that they are formed by a common
pathway in vivo, i.e., via a LPO product such as 4-HNE. This
conclusion is further supported by the strong correlations
found for in vitro adduct formation (Table 2). The steady-state
levels of these adducts in DNA in vivo are determined by the
rates of their formation and their removal by DNA repair
processes. While distinct repair pathways and kinetics have
been described for εdA and εdC residues,22 nothing is currently
known about the removal of ε5mdC from DNA. Studies are
now warranted to elucidate the possible role for this lesion in
miscoding and/or perturbation of methylation patterns. Our
data established that the etheno-adduct on 5mdC should be
added to the known major endogenous DNA damage products.
DISCUSSION
■
A change in the methylation patterns of genomic DNA is a
hallmark for many types of disease, and both hyper- and
hypomethylation have been observed.2 The processes that lead
to these changes have not yet been fully elucidated, but there is
evidence that DNA damage may play a role.9 DNA damage in
dCpdG sequences disturbs the proper methylation of cytidine
in the newly synthesized strands.9 In fact, dCpdG sequences
have been found to be mutational hotspots for several
carcinogens, including endogenous carcinogens resulting from
oxidative stress and lipid peroxidation, which in turn lead to the
formation of the etheno adduct residues εdA and εdC in
DNA.23 These adducts have been consistently detected at
background levels in normal DNA, but their frequencies are
increased several fold in human and animal tissues in which
tumors subsequently develop.24 These adducts have also been
detected in atherosclerotic lesions.25 The etheno adducts are
highly pro-mutagenic lesions,26 and mutations at dCpdG sites
by chloroacetaldehyde (a compound that specifically forms
In colorectal cancer, gene-specific hypermethylation is also
often observed, for example, in the gene for the DNA repair
enzyme O6-methylguanine-DNA methyltransferase (MGMT).
However, in ulcerative colitis patients prone to colorectal
cancer, the promoter hypermethylation of the MGMT gene
was significantly less frequent than reported for sporadic
colorectal cancers.31 This suggests that continuous DNA
167
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
(4) Ehrlich, M., Jiang, G., Fiala, E., Dome, J. S., Yu, M. C., Long, T. I.,
Youn, B., Sohn, O. S., Widschwendter, M., Tomlinson, G. E.,
Chintagumpala, M., Champagne, M., Parham, D., Liang, G., Malik, K.,
and Laird, P. W. (2002) Hypomethylation and hypermethylation of
DNA in Wilms tumors. Oncogene 21, 6694−6702.
damage to (methylated) dCpdG sequences by LPO products
could occur, leading to hypomethylation, rather than hyper-
methylation. Thus, one can speculate that the continuous
formation and removal of etheno adducts, including ε5mdC, at
dCpdG sites could result in a gradual loss of methylation. A link
between the repair of etheno adduct residues and DNA
methylation processes has already been considered because
MBD4 (methyl-dCpdG binding domain protein 4) is involved
in the repair of 5-methylcytosine CpG-TpG mismatches.10 At
present, the repair processes that would recognize and remove
ε5mdC are still unknown and warrant further investigation.
Also, studies on whether etheno adducts influence the coiling
and packaging of DNA (e.g., by affecting histone binding) will
help to elucidate the complex interaction among methylation,
demethylation, adduct formation, and repair.
In conclusion, we show for the first time that inflammation
generated lipid peroxidation (LPO) products can react with 5-
methyl-2′-deoxycytidine (5mdC) to form ε5mdC in vitro and in
vivo, and therefore, ε5mdC should be added to the list of
endogenous types of DNA damage. The availability and
characterization of the synthetic reference compounds 4 and
5 and the development of an ultrasensitive and specific assay for
the quantitation of ε5mdC lesions in small (ca. 25 μg) human
DNA samples provide a novel tool for unraveling the possible
role of ε5mdC in the progression of chronic degenerative
human diseases.
(5) Florl, A. R., Steinhoff, C., Muller, M., Seifert, H. H., Hader, C.,
̈
Engers, R., Ackermann, R., and Schulz, W. A. (2004) Coordinate
hypermethylation at specific genes in prostate carcinoma precedes
LINE-1 hypomethylation. Br. J. Cancer 91, 985−994.
(6) Mayer, W., Niveleau, A., Walter, J., Fundele, R., and Haaf, T.
(2000) Demethylation of the zygotic paternal genome. Nature 203,
501−502.
(7) Santos, F., Hendrich, B., Reik, W., and Dean, W. (2002) Dynamic
reprogramming of DNA methylation in the early mouse embryo. Dev.
Biol. 241, 172−182.
(8) Weitzman, S. A., Turk, P. W., Milkowski, D. H., and Kozlowski,
K. (1994) Free radical adducts induce alterations in DNA cytosine
methylation. Proc. Natl. Acad. Sci. U.S.A. 91, 1261−1264.
(9) Turk, P. W., Laayoun, A., Smith, S. S., and Weitzman, S. A.
(1995) DNA adduct 8-hydroxyl-2′-deoxyguanosine (8-hydroxygua-
nine) affects function of human DNA methyltransferase. Carcinogenesis
16, 1253−1255.
(10) Hendrich, B., and Bird, A. (1998) Identification and
characterization of a family of mammalian methyl-CpG binding
proteins. Mol. Cell. Biol. 18, 6538−6547.
(11) Cortazar, D., Kunz, C., Saito, Y., Steinacher, R., and Schar, P.
́
̈
(2007) The enigmatic thymine DNA glycosylase. DNA Repair 6, 489−
504.
(12) Gros, L., Ishchenko, A. A., and Saparbaev, M. (2003)
Enzymology of repair of etheno-adducts. Mutat. Res. 531, 219−229.
(13) Nair, U., Bartsch, H., and Nair, J. (2007) Lipid peroxidation-
induced DNA damage in cancer-prone inflammatory diseases: a review
of published adduct types and levels in humans. Free Radical Biol. Med.
43, 1109−1120.
ASSOCIATED CONTENT
* Supporting Information
The details of the synthesis shown in Scheme 1. This material is
available free of charge via the Internet at http://pubs.acs.org.
■
S
(14) Eberle, G., Barbin, A., Laib, R. J., Ciroussel, F., Thomale, J.,
Bartsch, H., and Rajewsky, M. F. (1989) 1,N6-etheno-2′-deoxyadeno-
sine and 3,N4-etheno-2′-deoxycytidine detected by monoclonal
antibodies in lung and liver DNA of rats exposed to vinyl chloride.
Carcinogenesis 10, 209−212.
AUTHOR INFORMATION
Corresponding Author
*Department of Toxicology, Maastricht University, Universi-
teitssingel 50, P.O. Box 616, Maastricht, The Netherlands. Tel:
+31 433881104. Fax: +31 433884146. E-mail: r.godschalk@
maastrichtuniversity.nl.
■
(15) Godschalk, R. W., Nair, J., Kliem, H. C., Wiessler, M., Bouvier,
G., and Bartsch, H. (2002) Modified immunoenriched 32P-HPLC
assay for the detection of O4-ethylthymidine in human biomonitoring
studies. Chem. Res. Toxicol. 15, 433−437.
Funding
Financial support for the work of R.W.O. was in part provided
by the Division of Preventive Oncology, National Center for
Tumor Diseases, Heidelberg (Director, Professor Cornelia
Ulrich).
(16) Nair, J., Barbin, A., Guichard, Y., and Bartsch, H. (1995) 1,N6-
ethenodeoxyadenosine and 3,N4-ethenodeoxycytine in liver DNA from
humans and untreated rodents detected by immunoaffinity/32P-
postlabeling. Carcinogenesis 16, 613−617.
(17) Nair, J., Vaca, C. E., Velic, I., Mutanen, M., Valsta, L. M., and
Bartsch, H. (1997) High dietary omega-6 polyunsaturated fatty acids
drastically increase the formation of etheno-DNA base adducts in
white blood cells of female subjects. Cancer Epidemiol. Biomarkers Prev.
6, 597−601.
DEDICATION
We dedicate this work to the late Dr. Jagadeesan Nair.
■
ABBREVIATIONS
■
(18) Godschalk, R. W., Nair, J., van Schooten, F. J., Risch, A., Drings,
P., Kayser, K., Dienemann, H., and Bartsch, H. (2002) Comparison of
multiple DNA adduct types in tumor adjacent human lung tissue:
effect of cigarette smoking. Carcinogenesis 23, 2081−2086.
(19) Zhang, W., Rieger, R., Iden, C., and Johnson, F. (1995)
Synthesis of 3,N4-etheno, 3,N4-ethano, and 3-(2-hydroxyethyl)
derivatives of 2′-deoxycytidine and their incorporation into oligomeric
DNA. Chem. Res. Toxicol. 8, 148−156.
ε5mdC, 3,N4-etheno-5-methyl-2′-deoxycytidine; 5mdC, 5-
methyl-2′-deoxycytidine; εdA, 1,N6-etheno-2′-deoxyadenosine;
εdC, 3,N4-etheno-2′-deoxycytidine; LPO, lipid peroxidation;
AA, arachidonic acid; CAA, chloroacetaldehyde; 4-HNE, trans-
4-hydroxy-2-nonenal
REFERENCES
■
(20) Davies, D. B. (1978) Conformations of nucleosides and
nucleotides. Progr. NMR Spectrosc. 12, 135−225.
(1) Kim, J. K., Samaranayake, M, and Pradhan, S. (2009) Epigenetic
mechanisms in mammals. Cell. Mol. Life Sci. 66, 596−612.
(2) Robertson, K. D. (2005) DNA methylation and human disease.
Nat. Rev. Genet. 6, 597−610.
(3) Widschwendter, M ., Jiang, G., Woods, C., Muller, H. M., Fiegl,
H., Goebel, G., Marth, C., Muller-Holzner, E., Zeimet, A. G., Laird, P.
(21) Barrio, J. R., Sattsangi, P. D., Gruber, B. A., Dammann, L. G.,
and Leonard, N. J. (1976) Species responsible for the fluorescence of
3,N4-ethenocytidine. J. Am. Chem. Soc. 98, 7408−7414.
(22) Hang, B., Chenna, A., Rao, S., and Singer, B. (1996) 1,N6-
ethenoadenine and 3,N4-ethenocytosine are excised by separate human
DNA glycosylases. Carcinogenesis 17, 155−157.
̈
̈
W., and Ehrlich, M. (2004) DNA hypomethylation and ovarian cancer
biology. Cancer Res. 64, 4472−4480.
168
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169

Chemical Research in Toxicology
Article
(23) El Ghissassi, F., Barbin, A., Nair, J., and Bartsch, H. (1995)
Formation of 1,N6- ethenoadenine and 3,N4-ethenocytosine by lipid
peroxidation products and nucleic acid bases. Chem. Res. Toxicol. 8,
278−283.
(24) Bartsch, H., and Nair, J. (2006) Chronic inflammation and
oxidative stress in the genesis and perpetuation of cancer: role of lipid
peroxidation, DNA damage, and repair. Langenbecks Arch. Surg. 391,
499−510.
(25) Nair, J., De Flora, S., Izzotti, A., and Bartsch, H. (2007) Lipid
peroxidation-derived etheno-DNA adducts in human atherosclerotic
lesions. Mutat. Res. 621, 95−105.
(26) Barbin, A. (2000) Etheno-adduct-forming chemicals: from
mutagenicity testing to tumor mutation spectra. Mutat. Res. 462, 55−
69.
(27) Choi, J. H., and Pfeifer, G. P. (2004) DNA damage and
mutations produced by chloroacetaldehyde in a CpG-methylated
target gene. Mutat. Res. 568, 245−256.
(28) Feng, Z., Hu, W., Amin, S., and Tang, M. S. (2003) Mutational
spectrum and genotoxicity of the major lipid peroxidation product,
trans-4-hydroxy-2-nonenal, induced DNA adducts in nucleotide
excision repair-proficient and -deficient human cells. Biochemistry 42,
7848−7854.
(29) Issa, J. P. (2002) Epigenetic variation and human disease. J.
Nutr. 132 (Suppl 8), 2388S−2392S.
(30) Kanai, Y., and Hirohashi, S. (2007) Alterations of DNA
methylation associated with abnormalities of DNA methyltransferases
in human cancers during transition from a precancerous to a malignant
state. Carcinogenesis 28, 2434−2442.
(31) Matsumura, S., Oue., N., Ito., R., Nakayama, H., Kitadai, Y.,
Yokozaki, H., Chayama, K., and Yasui, W. (2003) The promoter
methylation status of the DNA repair gene O6-methylguanine-DNA
methyltransferase in ulcerative colitis. Virchows Arch. 443, 518−523.
169
dx.doi.org/10.1021/tx200392a | Chem. Res. Toxicol. 2012, 25, 162−169