DNA methylation by dimethyl sulfoxide and methionine sulfoxide triggered by hydroxyl radical and implications for epigenetic modifications

Article abstract of DOI:10.1016/j.bmcl.2009.10.124

In this Letter, we demonstrate the formation of m⁵dC from dC or in DNA by dimethylsulfoxide (DMSO) and methionine sulfoxide (MetO), under physiological conditions in the presence of the Fenton reagent in vitro. DMSO reportedly affects the cellular epigenetic profile, and enhances the metastatic potential of cultured epithelial cells. The methionine sulfoxide reductase (Msr) gene was suggested to be a metastatis suppressor gene, and the accumulation of MetO in proteins may induce metastatic cancer. Our findings are compatible with these biological data and support the hypothesis that chemical cytosine methylation via methyl radicals is one of the mechanisms of DNA hypermethylation during carcinogenesis. In addition to m⁵dC, the formation of 8-methyldeoxyguanosine (m⁸dG) was also detected in DNA under the same reaction conditions. The m⁸dG level in human DNA may be a useful indicator of DNA methylation by radical mechanisms.

Full text of DOI:10.1016/j.bmcl.2009.10.124

Bioorganic & Medicinal Chemistry Letters 20 (2010) 260–265
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
DNA methylation by dimethyl sulfoxide and methionine sulfoxide triggered
by hydroxyl radical and implications for epigenetic modifications
*
Kazuaki Kawai, Yun-Shan Li, Ming-Fen Song, Hiroshi Kasai
Department of Environmental Oncology, Institute of Industrial Ecological Sciences, University of Occupational and Environmental Health, 1-1, Iseigaoka, Yahatanishi-ku,
Kitakyushu 807-8555, Japan
a r t i c l e i n f o
a b s t r a c t
In this Letter, we demonstrate the formation of m5dC from dC or in DNA by dimethylsulfoxide (DMSO)
and methionine sulfoxide (MetO), under physiological conditions in the presence of the Fenton reagent
in vitro. DMSO reportedly affects the cellular epigenetic profile, and enhances the metastatic potential
of cultured epithelial cells. The methionine sulfoxide reductase (Msr) gene was suggested to be a meta-
statis suppressor gene, and the accumulation of MetO in proteins may induce metastatic cancer. Our find-
ings are compatible with these biological data and support the hypothesis that chemical cytosine
methylation via methyl radicals is one of the mechanisms of DNA hypermethylation during carcinogen-
esis. In addition to m5dC, the formation of 8-methyldeoxyguanosine (m8dG) was also detected in DNA
under the same reaction conditions. The m8dG level in human DNA may be a useful indicator of DNA
methylation by radical mechanisms.
Article history:
Received 26 September 2009
Revised 23 October 2009
Accepted 27 October 2009
Available online 30 October 2009
Keywords:
5-Methyldeoxycytidine
Methyl radicals
Dimethylsulfoxide
Methionine sulfoxide
Epigenetic
Ó 2009 Elsevier Ltd. All rights reserved.
Alkyl radicals, including methyl radicals, are generated from the
tumor promoters, cumene hydroperoxide and t-butyl hydroperox-
ide in mouse keratinocytes, based on ESR experiments, suggesting
that carbon radicals are involved in cancer induction.1 In addition
to these chemicals, the generation of methyl radicals from various
carcinogens has been observed in vitro and in vivo. The metabo-
lism of 1,2-dimethylhydrazine2 and procarbazine3 produces
methyl radicals. Acetaldehyde generates methyl radicals by treat-
ments with xanthine oxidase,4 peroxynitrite,5 and Fe2+/H2O2.5
The formation of a methyl radical - deoxyguanosine adduct, 8-
methyl-20-deoxyguanosine (m8dG), has been detected in DNA after
a treatment with t-butyl hydroperoxide and ferrous ion in vitro,6
and after the administration of 1,2-dimethylhydrazine to rats.7
Methylation of RNA purine bases at the C-8 position by methyl rad-
icals has been also observed.8 We recently reported that cytosine
C-5 methylation in the monomer dC and DNA occurred via methyl
radicals generated by the tumor promoters, cumene hydroperoxide
and t-butyl hydroperoxide, in the presence of ferrous ion.9 This dis-
covery suggested new mechanisms of aberrant DNA hypermethy-
lation in the epigenetic process during chemical carcinogenesis.
Enzymatic DNA methylation due to the increased expression of
DNA methyltransferases (DNMTs) is a widely accepted mechanism
of DNA hypermethylation.10,11 However, controversial data show-
ing no clear association between gene hypermethylation in cancer
Fenton Reaction
•OH
COOH
H3C
CH3
CH3
S
S
H2N
O
O
MetO
DMSO
•CH3
NH2
NH2
CH3
N
N
O
N
O
N
HO
HO
O
O
OH
OH
m5dC
dC
* Corresponding author. Tel.: +81 93 691 7469; fax: +81 93 601 2199.
E-mail address: h-kasai@med.uoeh-u.ac.jp (H. Kasai).
Scheme 1. Formation of m5dC via methyl radicals by DMSO and MetO, triggered by
Fenton reaction.
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.10.124

K. Kawai et al. / Bioorg. Med. Chem. Lett. 20 (2010) 260–265
261
and high DNMT expression were also reported.12,13 Therefore,
it would be interesting to examine whether cytosine C-5 methyla-
tion via methyl radicals occurs by other biological/chemical
systems.
Methyl radicals are reportedly generated by the reaction of OH
radicals with dimethylsulfoxide (DMSO),14 from the amino acid
methionine (Met) upon gamma irradiation,15 and by the treatment
of methionine sulfoxide (MetO) with peroxynitrite.16 Particularly,
endogenous MetO formation in proteins is implicated as a patho-
logical biomarker in relation to aging, inflammation and smok-
ing.17–20 It is worth mentioning that DNA hypermethylation
during carcinogenesis is correlated to aging, inflammation and
smoking.21–23 DMSO also reportedly has an effect on the cellular
epigenetic profile, by inducing DNA hypermethylation.24 In this
study, we examined the biological relevance of m5dC formation
in dC and DNA by methyl radicals produced from DMSO and MetO
treated with a Fenton system (Scheme 1).
When dC was reacted with DMSO in the presence of Fenton re-
agent8 at pH 7.3, the formation of m5dC was clearly identified by
HPLC equipped with a photodiode array UV detector (Fig. 1). The
retention time and the UV spectrum of the reaction product were
the same as those of the authentic m5dC. Its formation was depen-
dent on the concentration of DMSO (Fig. 2), and the reaction was
rather rapid, due to its radical character. The reaction was approx-
m5dC
A
4
3
2
1
0
-1
0
10
20
30
40
50
60
B
8
6
4
2
0
0
10
20
30
40
50
60
C
8
6
4
2
0
0
10
20
30
40
50
60
Retention Time (min)
Figure 1. Detection of m5dC in the reaction mixture of dC, DMSO and Fenton reagent by HPLC. The reaction mixture8 (mixed under N2 atmosphere, final volume, 0.225 ml),
containing dC (final concentration, 5.46 mM), DMSO (100 mM),
L
-ascorbic acid (10.7 mM), EDTAꢀ2Na (3.1 mM), FeSO4 (5.3 mM), and H2O2 (19.6 mM) in 110 mM phosphate
buffer (pH 7.3), was reacted in a sealed plastic tube (tube volume, 2 ml) by vigorous shaking at 37 °C. After a 3 h reaction, the solution was centrifuged and an aliquot of the
supernatant was injected into the HPLC apparatus. (A) Chromatogram of the m5dC standard (left) and its UV spectrum (right); (B) chromatogram of the reaction mixture (left)
and UV spectrum of the peak at 50 min (right); (C) chromatogram of the control reaction mixture without DMSO.

262
K. Kawai et al. / Bioorg. Med. Chem. Lett. 20 (2010) 260–265
11 eV revealed a fragment BH2+ ion at m/z 126, formed by the loss
5
4
3
2
1
0
DMSO
of 20-deoxyribose.9 Therefore, the m5dC in the reaction mixture
was analyzed by LC/MS/MS, by monitoring the m/z 242?126
transition. In Figure 4, chromatograms of the LC/MS/MS analysis
of standard m5dC (A), poly(dG-dC)–DMSO–Fenton reagent (B),
poly(dG-dC)–MetO–Fenton reagent (C), and control poly(dG-dC)
without treatment (D) are shown. In both of the poly(dG-dC)
samples treated with DMSO and MetO in the presence of Fenton
reagent, a 242?126 transition peak appeared at 7.64 min, which
is the same retention time as that of authentic m5dC. The control
MetO
poly(dG-dC) without treatment also showed
a small peak
(Fig. 4D). This means that the commercial poly(dG-dC) contains
0
25
50
mM
75
100
a small amount of m5dC.
Based on the LC/MS/MS analysis, the yields of m5dC in the DNA–
DMSO–Fenton and DNA–MetO–Fenton reactions were calculated
to be 5.00/104 dC and 1.64/104 dC, respectively, which are in the
same range (2–4/104 dC) as the yield produced by the monomer
reactions for 30 min with the DMSO- and MetO–Fenton reagents
(data not shown). The high yield in DNA is unexpected, as com-
pared to the previous result that showing the yield of m5dC forma-
tion by a cumene hydroperoxide/Fe2+ system is 10-fold lower in
DNA than in the dC monomer.9 The following reasons are possible.
(i) The Fenton system used in this study has higher affinity to DNA
than to the monomer dC, thus generating higher levels of ÅOH rad-
icals in DNA;26 (ii) long-range electron transfer along the DNA
chain from the metal binding sites to the dC residues enhanced
the yield of m5dC in DNA.27 Further studies with precise measure-
ments of m5dC using a stable isotope internal standard are
required for the final conclusion.
Figure 2. Dose-dependency of m5dC formation in the dC/DMSO/Fenton reaction
and the dC/MetO/Fenton reaction. The reaction conditions were the same as those
in Figure 1, except that different concentrations of DMSO (25, 50 and 100 mM) and
MetO (50 and 100 mM) were used. Mean values of duplicate experiments are
plotted.
imately 65% complete within 5 min (data not shown). This reaction
may proceed via a free radical mechanism, probably via a methyl
radical. The dose-dependent formation of m5dC from dC was
also observed after a reaction with MetO plus Fenton reagent
(Fig. 2).
In addition, after a double-stranded alternating copolymer, ds
poly(dG-dC), was reacted with DMSO or MetO in the presence of
Fenton reagent at pH 7.3, the formation of m5dC was clearly de-
tected in the poly(dG-dC) after the treatment, by an immuno-dot
blot analysis9,25 (Fig. 3). As a positive control, we analyzed 0.02–
0.2 ng of calf thymus DNA, which contained 1.39 mol % m5dC per
P atoms. The chemiluminescence intensity increased depending
upon the calf thymus DNA concentration. The control poly(dG-
dC) without treatment also showed weak chemiluminescence. This
means that commercial poly(dG-dC) contains a small amount of
m5dC. Since an exact quantitation was difficult with the immu-
no-dot blot analysis, the amount of m5dC in the reaction mixture
was further analyzed by the LC/MS/MS.
We also analyzed the m8dG in these reaction products by LC/
MS/MS. The standard m8dG28 exhibited an MH+ ion at m/z 282,
and product ion analysis from m/z 282 revealed a fragment
+
BH2 ion at m/z 166, formed by the loss of 20-deoxyribose. There-
fore, the m8dG was analyzed by monitoring the m/z 282?166
transition. In Figure 5, chromatograms of the LC/MS/MS analysis
of standard m8dG (A), poly(dG-dC)–DMSO–Fenton reagent (B),
poly(dG-dC)–MetO–Fenton reagent (C), and control poly(dG-dC)
without treatment (D) are shown. In the poly(dG-dC) treated
with DMSO and MetO in the presence of Fenton reagent, a
282?166 transition peak appeared at 18.6 min (Fig. 5B and C),
which is the same retention time as that of authentic m8dG
(Fig. 5A). The control poly(dG-dC) without treatment also dis-
played a small peak of m8dG (Fig. 5D), which shows that the
commercial poly(dG-dC) contains a small amount of m8dG in
addition to m5dC. The formation of m5dC and m8dG in DNA poly-
mers is summarized in Table 1. The yield of m5dC formation in
DNA was found to be 15–30-fold higher than that of m8dG by
the reaction with DMSO and MetO in the presence of Fenton
reagent.
In the LC/MS analysis, the standard m5dC exhibited an MH+
ion at m/z 242, and product ion analysis from m/z 242 with
Calf thymus
ds poly(dG-dC)
DNA
Control
0.02 ng
DMSO, 5 min
MetO, 30 min
0.05 ng
0.1 ng
0.2ng
DMSO is known to induce hypermethylation of various genetic
loci and affects the epigenetic profile in mouse embryoid bodies.
Although those authors ascribed DNA methylation to the increase
of DNMT3a activity,24 other mechanisms of DNA methylation can-
not be ruled out. DMSO also enhanced the invasiveness and meta-
static potential of cultured epithelial cells, with an epigenetic
effect.29
DNA hypermethylation is increased with age, inflammation and
smoking.21–23 The MetO/Met ratio in proteins also reportedly in-
creased with age, inflammation and smoking.17–20 MetO in proteins
is repaired by methionine sulfoxide reductase (Msr). Three reports
have suggested that the Msr gene is a tumor suppressor.30–32
The chemical methylation of cytosine C-5 may occur in vivo. A
Met residue in a chromatin protein reportedly interacts with DNA
by the intercalation of the Met side chain into a GC pair, based on
NMR studies.33 If the Met residue is oxidized to MetO, then the
cytosine residues in the vicinity could undergo C-5 methylation
Figure 3. Detection of m5dC in ds poly(dG-dC) by an immuno-dot blot analysis.9,25
The reaction mixture (mixed under N2 atmosphere, final volume, 0.45 ml)
contained ds poly(dG-dC) (final concentration, 8.9 A260 OD units/ml), DMSO or
MetO (100 mM),
L
-ascorbic acid (10.7 mM), EDTAꢀ2Na (3.1 mM), FeSO4 (5.3 mM),
and H2O2 (19.6 mM), in 110 mM phosphate buffer (pH 7.3), and was reacted in a
sealed plastic tube (tube volume, 2 mL) by vigorous shaking at 37 °C. After 5 min or
30 min, the ds poly(dG-dC) was recovered from the reaction mixture and was used
for the analysis. As positive controls, m5dC in various amounts of calf thymus DNA
was visualized. As negative controls, untreated ds poly(dG-dC) was analyzed.

K. Kawai et al. / Bioorg. Med. Chem. Lett. 20 (2010) 260–265
263
Figure 4. LC/MS/MS analysis of m5dC in the reaction products. (A) Standard m5dC (1.78 mg/mL); (B) hydrolysate of poly(dG-dC)/DMSO/Fenton reaction product (reaction
conditions were the same as those in Figure 3, with 30 min reaction time); (C) hydrolysate of poly(dG-dC)/MetO/Fenton reaction product (reaction conditions were the same
as those in Figure 3, with 30 min reaction time); and (D) control ds poly(dG-dC) without treatment. A four mL portion of each sample was injected. The transition m/z
242?126 was monitored. Chromatograms B, C and D are shown on the same scale.
when triggered by ÅOH radicals. Various oxidized proteins and
their degradation products may accumulate in the cytosol during
aging and inflammation and may react with dCTP to form
m5dCTP, which could become incorporated into DNA and induce
gene silencing.34
If chemical methylation via methyl radicals is one of the
mechanisms of epigenetic change during carcinogenesis, and
m8dG is formed in a specific ratio to m5dC, then m8dG formation
analyzed by LC/MS/MS would be a good marker of chemical DNA
methylation, because the background level of m8dG would be
very low, as compared to that of m5dC, in mammalian cell
DNA. Therefore, epi-mutagens inducing DNA hypermethylation
can be assayed by in vitro experiments using cultured cells.
The analysis of the accumulation status of m8dG in human
DNA may also be a good indicator of the increase of m5dC in
DNA by radical mechanisms, and would be useful for human
cancer risk assessment.
In the present study, in addition to m5dC, the formation of a
small amount of m8dG was detected in the poly(dG-dC) after
the treatment. The higher level of m5dC formation than that of
m8dG may be due to a steric effect. Namely, the cytosine C-5
position may be present in a more open structure in the DNA,
while the guanine C-8 is sterically hindered by the phosphate
backbone.

264
K. Kawai et al. / Bioorg. Med. Chem. Lett. 20 (2010) 260–265
Figure 5. LC/MS/MS analysis of m8dG in the reaction products. (A) Standard m8dG28 (2.08
lg/mL); (B, C, D) the same samples as B, C, D in Figure 4.
References and notes
Table 1
Formation of m5dC and m8dG in ds poly(dG-dC) by DMSO and MetO in the presence
of Fenton reagent
1. Taffe, B. G.; Takahashi, N.; Kensler, T. W.; Mason, R. P. J. Biol. Chem. 1987, 262,
12143.
2. Augusto, O.; Du Plessis, L. R.; Weingrill, C. L. Biochem. Biophys. Res. Commun.
1985, 126, 853.
3. Goria-Gatti, L.; Iannone, A.; Tomasi, A.; Poli, G.; Albano, E. Carcinogenesis 1992,
13, 799.
Reaction
m5dC/104 dC
m8dG/104 dG
DMSO + Fenton reagent
MetO + Fenton reagent
5.00
1.64
0.27
0.06
4. Nakao, L. S.; Kadiiska, M. B.; Mason, R. P.; Grijalba, M. T.; Augusto, O. Free
Radical Biol. Med. 2000, 29, 721.
5. Nakao, L. S.; Ouchi, D.; Augusto, O. Chem. Res. Toxicol. 1999, 12, 1010.
6. Hix, S.; Morais Mda, S.; Augusto, O. Free Radical Biol. Med. 1995, 19,
293.
Acknowledgments
7. Netto, L. E.; RamaKrishna, N. V.; Kolar, C.; Cavalieri, E. L.; Rogan, E. G.; Lawson,
T. A.; Augusto, O. J. Biol. Chem. 1992, 267, 21524.
8. Kang, J. O.; Gallagher, K. S.; Cohen, G. Arch Biochem. Biophys. 1993, 306, 178.
9. Kasai, H.; Kawai, K. Chem. Res. Toxicol. 2009, 22, 984.
10. Benbrahim-Tallaa, L.; Waterland, R. A.; Dill, A. L.; Webber, M. M.; Waalkes, M.
P. Environ. Health Perspect. 2007, 115, 1454.
This work was supported by grants from the Ministry of Health,
Labor and Welfare of Japan, and a grant from Asia International
Educational Program for Graduate Students in the Field of Occupa-
tional Health.

K. Kawai et al. / Bioorg. Med. Chem. Lett. 20 (2010) 260–265
265
11. Robertson, K. D.; Uzvolgyi, E.; Liang, G.; Talmadge, C.; Sumegi, J.; Gonzales, F.
A.; Jones, P. A. Nucleic Acids Res. 1999, 27, 2291.
23. Kim, J. S.; Kim, H.; Shim, Y. M.; Han, J.; Park, J.; Kim, D. H. Carcinogenesis 2004,
25, 2165.
12. Eads, C. A.; Danenberg, K. D.; Kawakami, K.; Saltz, L. B.; Danenberg, P. V.; Laird,
P. W. Cancer Res. 1999, 59, 2302.
24. Iwatani, M.; Ikegami, K.; Kremenska, Y.; Hattori, N.; Tanaka, S.; Yagi, S.; Shiota,
K. Stem Cells 2006, 24, 2549.
13. Ehrlich, M.; Woods, C. B.; Yu, M. C.; Dubeau, L.; Yang, F.; Campan, M.;
Weisenberger, D. J.; Long, T.; Youn, B.; Fiala, E. S.; Laird, P. W. Oncogene 2006,
25, 2636.
14. Eberhardt, M. K.; Colina, R. J. Org. Chem. 1988, 53, 1071.
15. Makino, K. J. Phys. Chem. 1979, 83, 2520.
16. Nakao, L. S.; Iwai, L. K.; Kalil, J.; Augusto, O. FEBS Lett. 2003, 547, 87.
17. Wells-Knecht, M. C.; Lyons, T. J.; McCance, D. R.; Thorpe, S. R.; Baynes, J. W. J.
Clin. Invest. 1997, 100, 839.
25. Leuratti, C.; Singh, R.; Lagneau, C.; Farmer, P. B.; Plastaras, J. P.; Marnett, L. J.;
Shuker, D. E. Carcinogenesis 1998, 19, 1919.
26. Floyd, R. A. Biochem. Biophys. Res. Commun. 1981, 99, 1209.
27. Arkin, M. R.; Stemp, E. D.; Pulver, S. C.; Barton, J. K. Chem. Biol. 1997, 4, 389.
28. Maeda, M.; Nushi, K.; Kawazoe, Y. Tetrahedron 1974, 30, 2677.
29. Bentel, J. M.; Rhodes, G. C.; Markus, I.; Smith, G. J. Int. J. Cancer 1990, 46, 251.
30. Bilsborough, J.; Van Pel, A.; Uyttenhove, C.; Boon, T.; Van den Eynde, B. J. J.
Immunol. 1999, 162, 3534.
18. Onorato, J. M.; Thorpe, S. R.; Baynes, J. W. Ann. N.Y. Acad. Sci. 1998, 854, 277.
19. Hosako, M.; Ogino, T.; Omori, M.; Okada, S. Free Radical Biol. Med. 2004, 36, 112.
20. Carp, H.; Miller, F.; Hoidal, J. R.; Janoff, A. Proc. Natl. Acad. Sci. U.S.A. 1982, 79,
2041.
31. Lei, K. F.; Wang, Y. F.; Zhu, X. Q.; Lu, P. C.; Sun, B. S.; Jia, H. L.; Ren, N.; Ye, Q. H.;
Sun, H. C.; Wang, L.; Tang, Z. Y.; Qin, L. X. BMC Cancer 2007, 7, 172.
32. De Luca, A.; Sacchetta, P.; Nieddu, M.; Di Ilio, C.; Favaloro, B. BMC Mol. Biol.
2007, 8, 39.
21. Kim, S. K.; Jang, H. R.; Kim, J. H.; Kim, M.; Noh, S. M.; Song, K. S.; Kang, G. H.;
Kim, H. J.; Kim, S. Y.; Yoo, H. S.; Kim, Y. S. Carcinogenesis 2008, 29, 1623.
22. Suzuki, H.; Toyota, M.; Kondo, Y.; Shinomura, Y. Methods Mol. Biol. 2009, 512,
55.
33. Allain, F. H.; Yen, Y. M.; Masse, J. E.; Schultze, P.; Dieckmann, T.; Johnson, R. C.;
Feigon, J. EMBO J. 1999, 18, 2563.
34. Holliday, R.; Ho, T. Somat. Cell Mol. Genet. 1991, 17, 537.