Stereoisomeric C5-C5'-linked dihydrothymine dimers produced by radiolytic one-electron reduction of thymine derivatives in anoxic aqueous solution: Structural characteristics in reference to cyclobutane photodimers

Article abstract of DOI:10.1021/jo990059s

Radiolytic one-electron reduction of 1-methylthymine (1a) and 1,3- dimethylthymine (1b) in anoxic aqueous solution afforded stereoisomeric C5- C5'-linked dihydrothymine dimers, fractionated into the meso forms of (5R,5'S)- and (5S,5'R)-bi-5,6-dihydrothymine (3a,b[meso]) and a racemic mixture of (5R,5'R)- and (5S,5'S)-bi-5,6-dihydrothymines (3a,b[rac]), along with 5,6-dihydrothymines (2a,b). The meso and racemic dimers were produced in almost equivalent yields, possessing structural similarity with cis-syn- cyclobutane pyrimidine photodimers that are identified as highly mutagenic and carcinogenic photolesions induced by UV light. Similar radiolytic one- electron reduction of thymidine (1c) resulted in the pseudo-meso form of (5R,5'S)- and (5S,5'R)-bi-5,6-dihydrothymidine (3c[RS]) and two diastereormers of (5R,5'R)- and (5S,5'S)-bi-5,6-dihydrothymidine (3c[RR] and 3c[SS]). X-ray crystal structures indicated that two pyrimidine rings of the stereoisomeric dimers except 3a[rac] overlap with each other to a considerable extent, as in the cis-syn-cyclobutane photodimers. The pyrimidine rings of the dimers were twisted around 5-Me-C5-C5'-5'-Me by 51.1(2)°for 3a[meso], -85.4(4)°for 3a[rac], -65(1)°for 3b[meso], 43(2)°for 3b[rac], and 64.9(4)°for 3c[RS], respectively. It was predicted that the C5-C5'-linked dihydrothymine dimers may cause some distortion within a DNA duplex if they were incorporated. The pH dependence of the reactivities was in accord with a mechanism of the C5-C5'-linked dimerization by which electron adducts of la-c are irreversibly protonated at C6 and the resulting 5,6-dihydrothymin-5-yl radicals undergo bimolecular coupling.

Full text of DOI:10.1021/jo990059s

5100
J . Org. Chem. 1999, 64, 5100-5108
Ster eoisom er ic C5-C5′-Lin k ed Dih yd r oth ym in e Dim er s P r od u ced
by Ra d iolytic On e-Electr on Red u ction of Th ym in e Der iva tives in
An oxic Aqu eou s Solu tion : Str u ctu r a l Ch a r a cter istics in Refer en ce
to Cyclobu ta n e P h otod im er s
Takeo Ito, Hideki Shinohara, Hiroshi Hatta, and Sei-ichi Nishimoto*
Department of Energy and Hydrocarbon Chemistry, Graduate School of Engineering, Kyoto University,
Sakyo-ku, Kyoto 606-8501, J apan
Received J anuary 12, 1999
Radiolytic one-electron reduction of 1-methylthymine (1a ) and 1,3-dimethylthymine (1b) in anoxic
aqueous solution afforded stereoisomeric C5-C5′-linked dihydrothymine dimers, fractionated into
the meso forms of (5R,5′S)- and (5S,5′R)-bi-5,6-dihydrothymine (3a ,b[m eso]) and a racemic mixture
of (5R,5′R)- and (5S,5′S)-bi-5,6-dihydrothymines (3a ,b[r a c]), along with 5,6-dihydrothymines (2a ,b).
The meso and racemic dimers were produced in almost equivalent yields, possessing structural
similarity with cis-syn-cyclobutane pyrimidine photodimers that are identified as highly mutagenic
and carcinogenic photolesions induced by UV light. Similar radiolytic one-electron reduction of
thymidine (1c) resulted in the pseudo-meso form of (5R,5′S)- and (5S,5′R)-bi-5,6-dihydrothymidine
(3c[RS]) and two diastereormers of (5R,5′R)- and (5S,5′S)-bi-5,6-dihydrothymidine (3c[RR] and
3c[SS]). X-ray crystal structures indicated that two pyrimidine rings of the stereoisomeric dimers
except 3a [r a c] overlap with each other to a considerable extent, as in the cis-syn-cyclobutane
photodimers. The pyrimidine rings of the dimers were twisted around 5-Me-C5-C5′-5′-Me by
51.1(2)° for 3a [m eso], -85.4(4)° for 3a [r a c], -65(1)° for 3b[m eso], 43(2)° for 3b[r a c], and 64.9(4)°
for 3c[RS], respectively. It was predicted that the C5-C5′-linked dihydrothymine dimers may cause
some distortion within a DNA duplex if they were incorporated. The pH dependence of the
reactivities was in accord with a mechanism of the C5-C5′-linked dimerization by which electron
adducts of 1a -c are irreversibly protonated at C6 and the resulting 5,6-dihydrothymin-5-yl radicals
undergo bimolecular coupling.
In tr od u ction
oxidative damages may arise from the DNA radical
cations induced by the direct effect of radiations. A recent
study has shown that an electron-loss center (hole) at a
given DNA base radical cation can migrate intramolecu-
larly through a π-stacking of DNA bases, thus arriving
at and being trapped most efficiently in a guanine (G)
moiety to cause its oxidative damage.3 Under oxic condi-
tions in the presence of oxygen, these oxidative damages
are significantly enhanced, while a reducing species of
hydrated electron (eaq-) is scavenged by oxygen into less
reactive superoxide radical ion (O2•-).1a Several structures
of oxidative DNA-base modifications, particularly those
of purine modifications, have been identified to account
for the biological lesions.4
Interaction of ionizing radiation with living cells gener-
ates excess free radicals to induce a variety of damages
to cellular DNA that are responsible for carcinogenic,
mutagenic, and lethal effects on the cells.1 The primary
biological effects of ionizing radiation, eventually trig-
gering off the obvious lesions via a sequence of chemical
reactions and biological responses on a wide range of time
scales, are conventionally classified into two subgroups:
the indirect effect and direct effect.1a The indirect effect
is attributable largely to hydroxyl radicals (•OH), hy-
drated electrons (eaq-), and hydrogen atoms (•H) produced
by the radiolysis of excess cellular water (reactions 1 and
3). The direct effect involves absorption of radiation ener-
gy by DNA molecule itself to undergo electron ejection
into DNA radical cation (DNA•+), as is of essential
importance upon exposure to high-linear energy transfer
(LET) radiation (reaction 2).2 The concomitant electron
ejected from DNA is also hydrated as in reaction 3.
Concerning reductive DNA damages by dry electrons
(e- as in reactions 1 and 2) and/or hydrated electrons eaq
-
(reaction 3), the cytosine base moiety is a major site of
(1) (a) von Sonntag, C. In The Chemical Basis of Radiation Biology;
Taylor and Francis: London, 1987. (b) Steenken, S. Free Rad. Res.
Commun. 1992, 16, 349-379. (c) von Sonntag, C.; Schuchmann, H.-P.
Int. J . Radiat. Biol. 1986, 49, 1-34. (d) Bensasson, R. V.; Land, E. J .;
Truscott, T. G. In Excited States and Free Radicals in Biology and
Medicine; Oxford University Press: Oxford, 1993; pp 290-305.
(2) Roots, R.; Chatterjee, A.; Chang, P.; Lommel, L.; Blakely, E. A.
Int. J . Radiat. Biol. 1985, 47, 157-166.
(3) (a) Hall, D. B., Holmlin, R. E., Barton, J . K. Nature 1996, 382,
731-735. (b) Gasper, S. M.; Schuster, G. B. J . Am. Chem. Soc. 1997,
119, 12762-12771.
(4) (a) Shibutani, S.; Takeshita, M.; Grollmann, A. P. Nature 1991,
349, 431-434. (b) Cheng, K. C.; Cahill, D. S.; Kasai, H.; Nishimura,
S.; Loeb, L. A. J . Biol. Chem. 1992, 267, 166-172. (c) Tchou, J .;
Grollman, A. P. Mutat. Res. 1993, 299, 277-287. (d) Kouchakdjian,
M.; Bodepudi, V.; Shibutani, S.; Eisenberg, M.; J ohnson, F.; Grollman,
A. P.; Patel, D. J . Biochemistry 1991, 30, 1403-1412.
H2O f •OH + e- + •H
DNA f DNA•+ + e-
(1)
(2)
(3)
e- + nH2O f eaq
-
Oxidative DNA damages including various base modi-
fications by OH radicals have been extensively studied
to elucidate the detailed mechanism of oxidation.1 Similar
10.1021/jo990059s CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/11/1999

Stereoisomeric C5-C5′-Linked Dihydrothymine Dimers
J . Org. Chem., Vol. 64, No. 14, 1999 5101
Sch em e 1
electron attachment at lower temperatures, as character-
ized by ESR spectroscopy:1b,5-9 cytosine radical anion
(C•-) and thymine radical anion (T•-) are thus produced
in 75% and 24% yields, respectively. At higher temper-
atures, however, an equilibrium involving a mutual
electron-transfer process is established between the
thymine and cytosine radical anions, from which ir-
reversible protonation at C6 of the thymine radical anion
T•- takes place to produce a major intermediate of the
5,6-dihydrothymin-5-yl radical (TH•).10,11 In contrast to
oxidative damages, there has been little evidence for the
influence of radiation-induced reductive DNA-base dam-
ages on biological lesions. Previously, a negative result
for the biological effect has been reported that 5,6-
dihydrothymine structures incorporated in DNA are
responsible for neither a blocking nor a premutagenic
lesion,12 while it is one of the typical damage structures
induced by radiolytic reduction of thymine moiety. Al-
though the biological significance of the reductive DNA
damages is still obscure, it may be assumed that a
reduction mechanism is operative to account for the
lethal effect of radiation on hypoxic cells as are involved
in solid tumor tissues.1d
Previously, we reported efficient formation of C5-C5′-
linked dihydrothymidine dimer (3c) in the radiolytic one-
electron reduction of thymidine (1c) in deoxygenated
aqueous solution (Scheme 1), in which 5,6-dihydrothy-
midin-5-yl radicals (5c) were assumed as the most likely
intermediate species (see also Scheme 2).13 Possible
stereoisomers of the C5-C5′-linked dihydrothymidine
dimers have not yet been isolated and characterized,
while hydrogenated thymidine stereoisomers as the
accompanied major products, (5S)-(-)-5,6-dihydrothymi-
dine and (5R)-(+)-5,6-dihydrothymidine, were identified13
by reference to the reported structures.14 It seems of
interest to compare the conformational characteristics of
C5-C5′-linked dihydrothymine dimers with those of
cyclobutane pyrimidine photodimers possessing not only
C5-C5′ but also C6-C6′ linkages, which are afforded
through a formal [2 + 2] cycloaddition between the C5-
C6 double bonds of adjacent pyrimidines upon UV
irradiation of DNA.1a,15 The pyrimidine photodimers have
been identified as highly mutagenic and carcinogenic
lesions that result in miscoding during the DNA replica-
tion due to perturbations of base-pairing interactions and
global structural changes in DNA.15b The local and global
structures of pyrimidine photodimer-incorporated DNA
have been extensively studied by means of computational
simulations,16 NMR spectroscopy,17 X-ray crystallogra-
phy,18 and electrophoresis19 to get molecular insight into
the mechanisms, by which the photodimers induce mu-
tagenicity and in turn interact with a repair enzyme of
DNA photolyase.20 Comparing these structural charac-
terizations with the X-ray crystal structures of DNA-
excision repair enzyme and DNA photolyase, it is the
most likely that the enzymes may recognize the local
structures of the photodimers as well as the whole
pattern of small perturbation of the dimer-incorporated
DNA.20
In light of the biological role of pyrimidine photodimers,
we have a hypothesis that the C5-C5′-linked dihy-
drothymine dimers could be potentially mutagenic and
carcinogenic lesions when formed by chance at a thymine-
thymine (TT) tract in DNA upon exposure of hypoxic cells
to ionizing radiation. In this study, radiolytic reductions
of N1-substituted thymine derivatives such as 1-meth-
ylthymine (1a ), 1,3-dimethylthymine (1b), and thymidine
(1c) (Scheme 1) in deoxygenated aqueous solution have
been performed to isolate stereoisomers of C5-C5′-linked
dihydrothymine dimers: the meso form of (5R,5′S)- and
(5) Symons, M. C. R. In The Early Effects of Radiation on DNA;
Fielden, E. M., O’Neill, P., Eds.; Springer: Berlin, 1991; Vol. H54, pp
111-124.
(6) Sevilla, M. D. In Excited States In Organic Chemistry and
Biochemistry; Pullman, B., Goldblum, N., Eds.; D. Reidel: Dordrecht,
Holland, 1977.
(7) Sevilla, M. D.; Failor, R.; Clark, C.; Holroyd, R. A.; Pettei, M. J .
Phys. Chem. 1976, 80, 353-358.
(8) Sevilla, M. D.; Becker, D.; Yan, M.; Summerfield, S. R. J . Phys.
Chem. 1991, 95, 3409-3415.
(9) (a) Novais, H. M.; Steenken, S. J . Am. Chem. Soc. 1986, 108,
1-6. (b) Cullis, P. M.; McClymont, J . D.; Malone, M. E.; Mather, A.
N.; Podmore, I. D.; Sweeney, M. C.; Symons, M. C. R. J . Chem. Soc.,
Perkin Trans. 2 1992, 1695-1702.
(16) Miaskiewicz, K.; Miller, J .; Cooney, M.; Osman, R. J . Am. Chem.
Soc. 1996, 118, 9156-9163.
(17) (a) Taylor, J .-S.; Garrett, D. S.; Brockie, I. R.; Svoboda, D. L.;
Telser, J . Biochemistry 1990, 29, 8858-8866. (b) Kan, L.-S.; Voituriez,
L.; Cadet, J . Biochemistry, 1988, 27, 5796-5803. (c) Rycyna, R. E.;
Wallace, J . C.; Sharma, M.; Alderfer, J . L. Biochemistry 1988, 27,
3152-3163.
(18) (a) Camerman, N.; Camerman, A. Science 1968, 160, 1451-
1452. (b) Camerman, N.; Camerman, A. J . Am. Chem. Soc. 1970, 92,
2523-2527. (c) Hruska, F. E.; Voituriez, L.; Grand, A.; Cadet, J .
Biopolymers 1986, 25, 1399-1417.
(19) (a) Wang, C.-I.; Taylor, J .-S. Proc. Natl. Acad. Sci. U.S.A. 1991,
88, 9072-9076. (b) Wang, C.-I.; Taylor, J .-S. Chem. Res. Toxicol. 1993,
6, 519-523.
(10) (a) Hayon, E. J . Chem. Phys. 1969, 51, 4881-4899. (b) Theard,
L. M.; Peterson, F. C.; Myers, L. S., J r. J . Phys. Chem. 1971, 75, 3815-
3821. (c) Das, S.; Deeble, D. J .; Schuchmann, M.-N.; von Sonntag, C.
Int. J . Radiat. Biol. 1984, 46, 7-9.
(11) (a) Steenken, S.; Telo, J . P.; Candeias, L. P. J . Am. Chem. Soc.
1992, 114, 4701-4709. (b) Deeble, D. J .; Das, S.; von Sonntag, C. J .
Phys. Chem. 1985, 89, 5784-5788.
(12) Ide, H.; Petrullo, L. A.; Hatahet, Z.; Wallace, S. S. J . Biol. Chem.
1991, 266, 1469-1477.
(13) Nishimoto, S.; Ide, H.; Nakamichi, K.; Kagiya, T. J . Am. Chem.
Soc. 1983, 105, 6740-6741.
(14) Cadet, J .; Balland, A.; Berger, M. Int. J . Radiat. Biol. 1981,
39, 119-133.
(15) (a) Cadet, J .; Vigny, P. In Bioorganic Photochemistry: Photo-
chemistry and the Nucleic Acids; Morrison, H., Ed.; J ohn Wiley &
Sons: New York, 1990; pp 1-272. (b) Taylor, J .-S. Acc. Chem. Res.
1994, 27, 76-82.
(20) (a) Park, H.-W.; Kim, S.-T.; Sancar, A.; Deisenhofer, J . Science
1995, 268, 1866-1872. (b) Vassylyev, D. G.; Kashiwagi, T.; Mikami,
Y.; Ariyoshi, M.; Iwai, S.; Ohtsuka, E.; Morikawa, K. Cell 1995, 83,
773-782.

5102 J . Org. Chem., Vol. 64, No. 14, 1999
Ito et al.
Sch em e 2
F igu r e 1. Typical HPLC analyses of phosphate buffer solutions (pH 7.0) of (a) 1a , (b) 1b, and (c) 1c (5 mM) after γ-irradiation
(2 kGy) in the presence of sodium formate (200 mM) under Ar, as observed by delivering phosphate buffer solution (pH 3) with
(a) 15%, (b) 25%, and (c) 10% of methanol at a flow rate of 0.6 mL min-1. HPLC analysis was performed using (a, b) a Wakosil
5C18 column (φ 4.6 mm × 150 mm) or (c) a Cosmosil 5C18 MS column (φ 4.6 mm × 250 mm).
(5S,5′R)-bi-5,6-dihydrothymines (3a ,b[m eso]) and race-
mic compounds of (5R,5′R)- and (5S,5′S)-bi-5,6-dihy-
drothymines (3a ,b[r a c]) from 1a ,b, the pseudo-meso
form of (5R,5′S)- and (5S,5′R)-bi-5,6-dihydrothymidines
(3c[RS]), and two diasterormers of (5R,5′R)-, and (5S,5′S)-
bi-5,6-dihydrothymidines (3c[RR] and 3c[SS]) from 1c.
The present structural characterization has provided a
prediction that the meso form of the C5-C5′-linked
dihydrothymine dimeric structure could be formed at a
given TT tract in DNA through a small extent of
conformational change.
2-methyl-2-propanol ((CH3)3COH) or (ii) formate ions
(HCOO-), OH radicals and hydrogen atoms are scav-
enged into substantially unreactive 2-methyl-2-propanol
radicals (reaction 4: k(•OH) ) 5 × 108 dm3 mol-1 s-1 as
a major reaction, k(•H) ) 8 × 104 dm3 mol-1 s-1 as a minor
reaction) or converted to a somewhat less reducing
•-
species of carbon dioxide radical anions (CO2•-; G(CO2
)
) 3.5 × 10-7 mol J -1) (reaction 5: k(•OH) ) 3 × 109 dm3
mol-1 s-1, k(•H) ) 3 × 108 dm3 mol-1 s-1), respectively.1a
Consequently, the reaction systems involving the re-
stricted radical species for exclusive reduction can be
achieved: (i) eaq- (reduction potential at pH 7.0: E(nH2O/
eaq-) ) -2.9 V vs NHE22) + •H (E(H+/•H) ) -2.4 V22) and
Resu lts a n d Discu ssion
(ii) eaq + CO2 (E(CO2/CO2•-) ) -1.9 V23).
-
•-
F or m a tion a n d Ch a r a cter iza tion of C5-C5′-
Lin k ed Dih yd r ot h ym in e Dim er s. For investigating
reduction reactivity of 1-methylthymine (1a ), 1,3-di-
methylthymine (1b), and thymidine (1c) exclusively in
the radiolysis of aqueous solution, strongly oxidizing OH
radicals generated along with reducing species of hy-
drated electrons and hydrogen atoms from water radi-
olysis (see reactions 1 and 3) were effectively scavenged
by the aid of well-specified scavengers.1a The G values21
of primary water radicals are known as G(eaq-) ≈
G(•OH) ) 2.8 × 10-7 mol J -1 and G(•H) ) 0.6 × 10-7 mol
J -1 in neutral aqueous solution.1a In a deoxygenated
aqueous solution containing an excess amount of (i)
(CH3)3COH + •OH (•H) f
•CH2(CH3)2COH + H2O (H2) (4)
HCOO- + •OH (•H) f CO2•- + H2O (H2)
(5)
Figure 1a-c shows representative HPLC chromato-
grams as monitored by UV absorbance at 210 nm for Ar-
purged phosphate buffer solutions (5 mM, pH 7.0) of N1-
substituted thymine derivatives 1a -c, respectively, after
(22) Swallow, A. J . In Radiation Chemistry. An Introduction; Long-
man: London, 1973.
(21) The number of molecules produced or changed per 1 J of
radiation energy absorbed by the reaction system.
(23) Schwarz, H. A.; Creutz, C.; Sutin, N. Inorg. Chem. 1985, 24,
433-439.

Stereoisomeric C5-C5′-Linked Dihydrothymine Dimers
J . Org. Chem., Vol. 64, No. 14, 1999 5103
F igu r e 2. ORTEP drawings of C5-C5′-linked dihydrothymine dimers: (a) 3a [m eso]; (b) 3a [r a c]; (c) 3b[m eso]; (d) 3b[r a c].
Ch a r t 1
2 kGy γ-irradiation in the presence of excess (1 M) sodium
formate. Most of the products could not be detected by
UV absorption at 254 nm, indicating the efficient forma-
tion of C5-C6-saturated thymine structures in the
reduction of 1a -c with hydrated electrons and CO2•- as
well. The eluents from the respective major peaks were
collected by preparative HPLC and fractionation. The
fractionated products in aqueous solution containing 10-
25 vol % methanol were evaporated at room temperature
and submitted to a purity check by analytical HPLC
followed by spectroscopic characterization. High-resolu-
tion positive FAB mass spectrometry (FAB-HRMS) of
the isolated HPLC fractions 2a -c (single peak for 2a ,b
in Figure 1a,b and double peaks for 2c in Figure 1c) and
3a -c (double peaks for 3a ,b in Figure 1a,b and triple
peaks for 3c in Figure 1c) provided protonated parent
ions that are in accord with the molecular formulas of
the corresponding 5,6-dihydrothymine derivatives and
C5-C5′-linked dihydrothymine dimers, respectively (see
structures by X-ray crystallography. By reference to the
crystallographs of 3a ,b in Figure 2 (see also Chart 1),
we determined that the faster HPLC fractions are the
meso forms of the (5R,5′S)-bi-5,6-dihydrothymine deriva-
tives (3a ,b[m eso]), while the slower HPLC fractions are
racemic compounds of (5R,5′R)- and (5S,5′S)-bi-5,6-
dihydrothymine derivatives (3a ,b[r a c]) (see also Figure
1a,b). In the present study, further separation of the
racemic compounds 3a ,b[r a c] was not performed. In
contrast to 3a ,b, the crystals of the three HPLC fractions
of 3c were unstable, and therefore, their X-ray crystal-
lographic data were collected by coating with 40 vol %
collodion-ethanol solution to identify only 3c[RS]: the
X-ray crystallography in collodion-coating and in a capil-
lary tube was unsuccessful for the other two products.
Figure 3 illustrates the structure and the three-dimen-
sional lattice packing diagram of 3c[RS], indicating that
a pair of (5S,5′R)-bi-5,6-dihydrothymidine and (5R,5′S)-
bi-5,6-dihydrothymidine monohydrates are packed in a
unit cell: there is a hydrogen bond between the crystal
water and one of the deoxyribose-C(5)-OH groups of the
dimer. Although crystal structures are unknown, the
1
the Experimental Section). According to the H and 13C
NMR spectral data by reference to authentic samples,
the products 2a ,b were identified as 1-methyl-5,6-dihy-
drothymine (2a ) and 1,3-dimethyl-5,6-dihydrothymine
(2b), respectively. Similar products derived from hydro-
genation of 1c were stereoisomeric and identified as 5(S)-
(-)-5,6-dihydrothymidine (2c[S]) and 5(R)-(+)-5,6-dihy-
drothymidine (2c[R]) by comparing the spectral data with
those of authentic samples.13,14 These 5,6-dihydrothymine
derivatives are well documented products obtained in the
radiolytic reduction of thymine derivatives in aqueous
solution. In separate experiments, the HPLC fractions
eluted in shorter retention times (<5 min) in Figure 1a-c
were confirmed to contain relatively low yields of C5- and/
or C6-carboxylated byproducts resulting from addition
(24) Various carboxylated monomeric and dimeric thymine deriva-
tives were separated by HPLC when phosphate buffer solutions (10
mM, pH 3.0) contained higher contents of methanol (10-25 vol %).
The structural characterization by X-ray crystallography of 5,6-
dihydrothymine-6-carboxylic acid as a major product among them was
performed to get insight into the inhibitory effect on dihydroorotate
dehydrogenage and will be reported elsewhere (Ito, T.; Hatta, H.;
Nishimoto, S. Manuscript in preparation).
•-
of CO2 to the C5-C6 double bond of 1a -c.24,25 As
expected, such a carboxylation was excluded in the
radiolysis of aqueous solution containing excess 2-methyl-
2-propanol instead of sodium formate.
Since the FAB-HRMS data predict that the products
3a -c may be isomeric C5-C5′-linked dihydrothymine
dimers, attempts were made to determine their crystal
(25) Wada, T.; Ide, H.; Nishimoto, S.; Kagiya, T. Int. J . Radiat. Biol.
1982, 42, 215-221.

5104 J . Org. Chem., Vol. 64, No. 14, 1999
Ito et al.
dihedral angles are considerably smaller than the cor-
responding values of 152° (N1-N3-C5 planes) and 153°
(C2-C4-C6 planes) for the cis-syn-1,3-dimethylthymine
cyclobutane photodimer and 178° for Thy-[C3]-Thy(c,s).
It was also demonstrated that the pyrimidine rings of
3a -c are not planar where the C6 atoms are out of the
planes of the other atoms. Furthermore, the pyrimidine
rings of 3a -c are twisted around the C5-C5′ linkage,
and the torsion angles through 5-Me-C5-C5′-5′-Me vary
significantly in the range of 43-85°. These values are
much greater than 24° for the cis-syn-1,3-dimethylthy-
mine photodimer because the absence of the C6-C6′
linkage in 3a -c permits more free rotation of pyrimidine
rings around the C5-C5′ linkage. Nevertheless, it is
interesting that two pyrimidine rings of 3a ,b[m eso] are
confronted with each other and thereby cause a stacking
of their carbonyl groups as in the cis-syn-cyclobutane
photodimers. By reference to the twist angle for a
successive thymine base pair about the axis of DNA
duplex,e.g.,33-39°for the structure ofGCGTTTTTTCGC,27
it seems structurally feasible that the meso form of a C5-
C5′-linked dihydrothymine dimer structure like 3a [m eso]
could be produced at a successive thymine base pair by
reduction pathway to cause some distortion within a
DNA duplex. Similar to cis-syn photodimers, such a meso
dimer structure may give rise to insufficient stacking of
the pyrimidine rings that will weaken the hydrogen
bonds of the 3′- or 5′-dihydrothymine moiety with comple-
mentary adenine. In contrast to the meso dimer, the
formation of racemic compounds of C5-C5′-linked dihy-
drothymine dimers such as 3a [r a c] from the normal
stacking conformation of adjacent thymines in DNA is
unlikely, since two pyrimidine rings should be arranged
in the almost identical plane without overlapping of their
carbonyl groups.
F igu r e 3. ORTEP drawing and a three-dimensional lattice
packing diagram of C5-C5′-linked dihydrothymidine dimer
(3c[RS]).
accompanying two products may be assigned to diaster-
eoisomeric (5R,5′R)-bi-5,6-dihydrothymidine (3c[RR]) and
(5S,5′S)-bi-5,6-dihydrothymidine (3c[SS]), respectively,
1
The H and 13C NMR data in D2O or dimethyl sulfox-
1
in accord with the H and 13C NMR data (Tables 5 and
ide-d6 are consistent with the crystallographic charac-
terizations of C5-C5′-linked dihydrothymine dimers
6). As implicated from Figure 3, the C5-C5′-linked
dihydrothymidine dimer bearing hydrophilic deoxyribose
moiety favor the crystallization in the form of hydrate,
while dehydration of the resulting crystal would cause
readily a lattice decomposition. In this context, it is
noteworthy that the pseudo-meso form 3c[RS] is eluted
slower than the diastereomers 3c[RR] and 3c[SS] in the
HPLC, opposite to the elution speeds of 3a ,b[m eso] being
faster than those of 3a ,b[r a c] (Figure 1a-c). It is likely
that the hydrocarbon component of the 2-deoxyribose
moiety is involved in hydrophobic interactions, and
therefore, the accessibility of the furanose ring may play
some role in the HPLC behavior of C5-C5′-linked dihy-
drothymidine dimers.
The X-ray crystallographic data are summarized in
Tables 1 and 2. The C5-C5′-linked dihydrothymine
dimers 3a -c possess the C5-C5′ bond lengths from 1.55
to 1.59 Å. These are comparable to both the C5-C5′ and
C6-C6′ bond lengths of photodimers: 1.58 and 1.60 Å
for the cis-syn-1,3-dimethylthymine cyclobutane pho-
todimer18b and 1.548 and 1.596 Å for 1,1′-trimethylene-
bisthymine photodimer (Thy-[C3]-Thy(c,s)).26 According
to the crystal structures shown in Figures 2 and 3, two
pyrimidine rings of the C5-C5′-linked dihydrothymine
dimers 3a -c overlap with each other, the extent of which
was evaluated by the dihedral angles between the pyri-
midine-containing planes composed of N1, N3, and C5
atoms. The calculation indicates 128° for 3a [m eso], 160°
for 3a [r a c], 134° for 3b[m eso], 136° for 3b[r a c], and 113°
for 3c[RS]. Except for the result for 3a [r a c], these
1
3a -c (Tables 3-6). Table 3 compares the H NMR data
between 3a,b[meso] and 3a,b[r a c] for better understand-
ing of the conformations in solution. For each dimer, the
meso form and the racemic form had similar 1H NMR
chemical shifts except for the H6 and H6′ resonance. It
is thus remarkable that 3a ,b[m eso] showed downfield
shifts in the H6 resonance and in turn upfield shifts in
the counterpart H6′ resonance; therefore, the H6 and H6′
chemical shifts become closer to each other relative to
3a ,b[r a c]. Similarly, the H6 and H6′ chemical shifts of
3c[RS] were also closer to each other, compared with
those of 3c[RR] or 3c[SS], although both protons showed
relatively upfield shifts (Table 5). On the other hand, the
13C NMR chemical shifts showed little difference between
the isomers (Tables 4 and 6).
Mech a n ism of C5-C5′-Lin k ed Dim er iza tion by
Ra d iolytic Red u ction . Previously, we reported a brief
account for the mechanism by which thymidine under-
•-
goes one-electron reduction by hydrated electron or CO2
to give C5-C5′-linked dihydrothymidine dimer and 5,6-
dihydrothymidines.13 Thymine is known to undergo one-
electron reduction into its radical anion by hydrated
electron at a diffusion-controlled rate (k(eaq-) ) 1.7 × 1010
dm3 mol-1 s-1 at pH 6.0)28 and similarly by CO2•- at more
(26) Leonard, N. J .; Golankiewicz, K.; McCredie, R. S.; J ohnson, S.
M.; Paul, I. C. J . Am. Chem. Soc. 1969, 91, 5855-5862.
(27) Nelson, H. C. M.; Finch, J . T.; Luisi, B. F.; Klug, A. Nature
1987, 330, 221-226.

Stereoisomeric C5-C5′-Linked Dihydrothymine Dimers
J . Org. Chem., Vol. 64, No. 14, 1999 5105
Ta ble 1. Cr ysta l Da ta a n d Exp er im en ta l Deta ils
3a [m eso]
3a [r a c]
3b[m eso]
3b[r a c]
13.24(1)
15.50(1)
14.73(1)
3c[RS]
a (Å)
b (Å)
c (Å)
8.618(5)
11.898(4)
13.020(5)
97.23(5)
1324(1)
4
20.645(6)
6.414(4)
12.186(4)
121.29(2)
1378.9(9)
4
8.375(4)
10.602(6)
17.480(3)
100.79(2)
1524(1)
4
6.000(3)
14.619(2)
13.280(2)
102.12(2)
1138.8(6)
2
â (°)
V (Å3)
3022(3)
8
Z
µ (Mo KR) (cm-1
space grp
cryst syst
)
1.08
P21/c
monoclinic
4.3; 4.9
1.55
1.04
Cc
monoclinic
5.5; 6.6
2.17
1.00
P21/n
monoclinic
7.7; 4.9
2.64
1.01
Pbcn
orthorhombic
6.0; 3.7
1.74
1.20
P21
monoclinic
4.8; 6.0
2.58
residualsa (R; Rw)
GOF
a
2
R ) Σ||Fo| - |Fc||/Σ|Fo|; Rw ) [(Σw(|Fo| - |Fc|)2/Σw|Fo| )]1/2
.
Ta ble 2. Selected Bon d Len gth , P la in An gles, a n d
Ta ble 5. 1H NMR (D2O, 400 MHz) Ch em ica l Sh ifts (p p m )
Tor sion An gles ofC5-C5′-Lin k ed Dih yd r oth ym in e Dim er s
3a [m eso] 3a [r a c] 3b[m eso] 3b[r a c] 3c[RS]
bond length (Å)
of C5-C5′-Lin k ed Dih yd r oth ym id in e Dim er s
3c[RS]
3c[RR] or 3c[SS]
5-CH3
H
H
1.278 (s), 1.305 (s)
3.275 (d), 3.323 (d)
3.578-3.680a (m)
1.139 (s), 1.222 (s)
3.335 (d), 3.176 (d)
4.061 (d), 4.090 (d
-12.2, -13.0
C5-C5′ 1.591(3) 1.568(4) 1.57(2)
1.55(2) 1.583(6)
1.35(2) 1.348(5)
1.41(2) 1.393(6)
1.35(2) 1.379(5)
1.60(2) 1.534(6)
1.53(2) 1.544(5)
1.48(2) 1.459(5)
1.35(2) 1.354(5)
1.44(2) 1.385(5)
1.38(2) 1.372(6)
1.56(2) 1.524(6)
1.55(2) 1.542(6)
1.44(2) 1.467(5)
6A, H6′A
6B, H6′B
N1-C2 1.347(3) 1.39(1)
C2-N3 1.364(3) 1.34(1)
1.36(2)
1.41(2)
J
6AB, J 6′AB (Hz) -13.6, -13.2
N3-C4 1.395(3) 1.410(8) 1.37(2)
C4-C5 1.530(3) 1.541(10) 1.56(2)
deoxyribose
H(1)
6.127 (t), 6.077 (t)
1.982-2.189 (m)
6.174 (t), 6.233 (t)
1.985-2.221 (m),
2.059-2.252 (m)
4.247-4.282 (m),
4.285-4.310 (m)
3.778-3.806 (m),
3.797-3.826 (m)
3.568-3.666 (m),
3.552-3.667 (m)
C5-C6 1.530(3) 1.55(1)
1.50(2)
H(2)
C6-N1 1.457(3) 1.455(9) 1.49(2)
N1′-C2′ 1.339(2) 1.27(1)
C2′-N3′ 1.385(2) 1.47(1)
1.35(2)
1.40(2)
H(3)
H(4)
H(5)
4.232-4.280 (m)
3.764-3.794 (m)
3.548-3.645a (m)
N3′-C4′ 1.366(2) 1.348(9) 1.36(2)
C4′-C5′ 1.528(3) 1.50(1)
C5′-C6′ 1.537(3) 1.54(1)
1.57(2)
1.51(2)
C6′-N1′ 1.463(2) 1.457(9) 1.46(2)
plane angle (deg) between pyrimidine rings
a
A part of the H6 peaks overlapped with the H(5) peaks.
128
torsion angle (deg) about C5-C5′
51.1(2) -85.4(4) -65(1)
160
134
136
113
Ta ble 6. 13C NMR (D2O, 100 MHz) Ch em ica l Sh ifts (p p m )
43(2)
64.9(4)
of C5-C5′-Lin k ed Dih yd r oth ym id in e Dim er s
Ta ble 3. 1H NMR (DMSO-d 6, 270 MHz) Ch em ica l Sh ifts
(p p m ) of C5-C5′-Lin k ed Dih yd r oth ym in e Dim er s
3c[RS]
3c[RR] or 3c[SS]
5-CH3
C6
C2
C4
20.394, 20.504
46.235, 46.528
156.366, 156.622
178.220, 178.330
47.497, 47.698
17.505, 17.359
47.333, 47.278
155.525, 156.257
179.828, 179.812
46.894, 47.845
3a [m eso]
3a [r a c]
3b[m eso]
3b[r a c]
5-CH3
1-CH3
3-CH3
3-NH
1.184 (s)
2.794 (s)
1.136 (s)
2.875 (s)
1.167 (s)
2.804 (s)
2.908 (s)
1.084 (s)
2.814 (s)
2.892 (s)
C5
deoxyribose
C(1)
C(2)
C(3)
C(4)
10.143 (s)
3.208 (d)
3.440 (d)
10.100 (s)
3.003 (d)
4.143 (d)
86.798, 86.835
37.860, 38.116
73.594, 73.631
87.987, 88.060
64.139, 64.286
86.652, 86.908
38.006, 38.281
73.503, 73.448
88.060, 88.006
63.993, 64.048
H
H
6A, H6′A
6B, H6′B
3.263 (d)
3.383 (d)
-13.8
3.127 (d)
3.734 (d)
-12.7
J AB (Hz) -13.5
-12.2
C(5)
Ta ble 4. 13C NMR (DMSO-d 6, 67.5 MHz) Ch em ica l Sh ifts
(p p m ) of C5-C5′-Lin k ed Dih yd r oth ym in e Dim er s
For the clarity of the mechanism of C5-C5′-linked
dimerization, a further attempt was made to investigate
the influence of pH on the radiolytic reduction of 1a -c
in anoxic buffer solution. As shown in Figure 4, upon
increasing the pH values from 3.0 to 11.0, the G values
for both the decomposition of 1a ,b and the formation of
C5-C5′-linked dimers 3a ,b increased in acidic solution,
attained their maxima in neutral solution, and then
decreased in basic solution. Similar behavior was also
observed for the radiolytic reduction of 1c into 3c, while
the pH value resulting in the maximum G values shifted
to somewhat basic region of pH 9.0. In a wide pH range
the stereoisomeric C5-C5′-linked dimers as the meso
form and the racemic compounds were produced in
almost equivalent yields. The maximum G values of the
respective C5-C5′-linked dimers thus obtained were
3a [m eso] ) 1.07 × 10-7 mol J -1 (23% selectivity based
on the decomposed substrate), 3a [r a c] ) 0.97 × 10-7 mol
J -1 (20%), 3b[m eso] ) 1.02 × 10-7 mol J -1 (30%), and
3b[r a c] ) 0.91 × 10-7 mol J -1 (27%) for 3c[RS] ) 1.22
× 10-7 mol J -1 (20%) and for 3c[RR] + 3c[SS] ) 1.13 ×
10-7 mol J -1 (18%).
3a [m eso]
3a [r a c]
3b[m eso]
3b[r a c]
5-CH3
1-CH3
3-CH3
C5
C6
C2
18.781
33.743
15.957
34.366
19.225
34.818
27.814
45.572
50.509
153.471
172.435
17.338
35.590
27.843
44.689
50.442
152.389
173.368
44.982
51.932
152.870
173.223
43.822
52.482
151.972
175.295
C4
than 5 orders of magnitude slower rate (k(CO2•-) ) ∼5
× 104 dm3 mol-1 s-1).28 By reference to the reduction
potential23 vs the normal hydrogen electrode at pH 7.0,
CO2•- (E(CO2/CO2•-) ) -1.9 V) is of less reducing ability
than hydrated electron (E(nH2O/eaq-) ) -2.9 V). Never-
theless, we confirmed in the previous study13 that CO2
•-
had practically the same one-electron reducing ability as
hydrated electron toward 1c, thus resulting in similar
product distribution of 2c[R], 2c[S], and 3c (the stere-
oisomers were not separated previously).
(28) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. J .
Phys. Chem. Ref. Data 1988, 17, 513-886.

5106 J . Org. Chem., Vol. 64, No. 14, 1999
Ito et al.
F igu r e 5. Influence of the concentration of thymidine (1c)
on the yields of C5-C5′-linked dihydrothymidine dimers (O,
3c[RS]; 9, 3c[RR] + 3c[SS]) as observed at about 30%
decomposition of 1c in the X-radiolysis of Ar-purged phosphate
buffer solution (pH 7.0) containing sodium formate (200 mM).
drogen abstraction by OH radical from formic acid as in
reaction 6 produces a radical species of •COOH that has
a stronger acidity (pKa ) 1.4) than the parent formic acid
•-
F igu r e 4. pH-dependent variation of G values for decomposi-
tion of thymine derivatives (b, 1a -c) and formation of C5-
C5′-linked dihydrothymine dimers ([, 3a ,b[m eso] or 3c[RS];
0 3a ,b[r a c] or 3c[RR] + 3c[SS]) in the γ-radiolysis of Ar-
purged phosphate buffer solution containing sodium formate
(200 mM): C5-C5′-linked dimerization of (a) 1a to 3a , (b) 1b
to 3b, and (c) 1c to 3c.
and is dissociative into CO2 under the present condi-
tions at pH >2.0. However, the rate constant of reaction
5 (k(•OH) ) 3 × 109 dm3 mol-1 s-1) is 1 order of magnitude
greater than that (k(•OH) ) 1 × 108 dm3 mol-1 s-1) of
reaction 6;28 therefore, the net efficiency of scavenging
OH radicals to reducing CO2-• being lowered to consider-
able extent in acidic solution at pH < 3.75.
In light of previous studies on the reaction of hydrated
electrons with pyrimidines,10 the pH dependence of the
C5-C5′-linked dimerization may be rationalized by the
reaction pathways given in Scheme 2. The initial and key
step involves one-electron reduction of 1a -c by not only
HCOOH + •OH (•H) f •COOH + H2O (H2) (6)
Hydrogen atoms and OH radicals readily add to the
C5-C6 double bond of pyrimidines to produce both 5-yl
and 6-yl radicals.1a We therefore examined a hypothetical
radical reactivity whether the C5 radical can add to the
C5-C6 double bond of the parent thymine moiety,
producing the C5-C5′-linked dimer 6-yl radical (path 5
in Scheme 2), in the representative radiolytic reduction
of 1c in phosphate buffer. As shown in Figure 5, while
the yields of 3c were almost constant at concentrations
of 1c lower than 0.5 mM in formate-containing phosphate
buffer (pH 7.0), they were decreased with increasing
concentration of 1c. This result may rule out the above
hypothesis that the addition reaction of the 5-yl radical
5c to the C5 position of the C5-C6 double bond of 1c
will be a notable route to dimeric product 3c, because it
should favor the higher concentrations of 1c contrary to
the results in Figure 5. Figure 5 suggests that contribu-
tion of possible hydrogen abstraction of 5c from the
deoxyribose moiety of 1c may become more important
and thereby inhibit the radical coupling to 3c, upon
increasing the concentration of 1c. In the separate
experiments using methanol-d4 as an OH radical scav-
enger, radiolytic reduction of 1a ,b in phosphate buffer
also afforded the C5-C5′-linked dimers 3a ,b. Assuming
a radical addition mechanism by which the dihydrothy-
mine dimer 6-yl radicals (6a ,b) will be produced from the
reaction of C5 radicals 5a ,b with 1a ,b and thereafter
abstract deuterium atom from CD3OD (Scheme 3), the
dimeric products 3a ,b should possess the monodeuter-
ated structures. The FAB-MS data of 1a and 1b provided
the respective parent ions at 283 and 311, demonstrating
that the incorporation of deuterium atom into the dimer
did not occur. This result is also consistent with the
conclusion that the radical coupling mechanism is pre-
dominant for the C5-C5′-linked dimerization by radi-
olytic reduction (path 4).
•-
hydrated electron but also by CO2 to form the corre-
sponding radical anions as the primary intermediates
(4a -c) (path 1). It is likely that the electron adducts
4a -c are readily protonated at O4 by water (path 2) to
form the reducing radicals (5′a -c) followed by very slow
self-termination.1a In view of the pKa ≈ 7.2 for the
protonated electron adduct of thymine,10 such protona-
tions of 4a -c will become more significant with decreas-
ing the pH, and inversely, deprotonation of 5′a -c (path
2) will be more facilitated with increasing pH value. In
competition with the fast protonation at O4 to oxygen-
protonated radicals, 4a -c may also undergo another type
of protonation at C6 to form 5,6-dihydrothymin-5-yl
radicals (5a -c) (path 3) that have oxidizing properties
and more stability.29 This reaction occurs irreversibly to
increasing extents as the pH decreases; therefore, both
the one-electron reductive decomposition and the C5-
C5′-linked dimerization by a radical coupling mechanism
of 1a -c (path 4) are enhanced. In basic solution at pH
> 10, however, the yields of electron adducts 4a -c should
significantly decrease in correspondence with the pH
effect on the yield of primary active species in the water
radiolysis: the yield of OH radicals responsible for the
formation of CO2•- in reaction 5 rapidly decreases, while
that of hydrated electrons slightly increases, with in-
creasing pH.30 There is an additional pH effect on the
acidic reaction system in the presence of excess formate
ions. In accord with the pKa ) 3.75 for formic acid, the
formate ions involved in the reaction system for conver-
sion of OH radicals to reducing CO2•- (see reaction 5) are
largely protonated in acidic solution at pH <3.75. Hy-
(29) Whillans, D. W.; J ohns, H. E. J . Phys. Chem. 1972, 76, 489-493.
(30) Spinks, J . W.; Woods, R. J . In Introduction to Radiation
Chemistry, 3rd ed.; J ohn Wiley & Sons: New York, 1990.

Stereoisomeric C5-C5′-Linked Dihydrothymine Dimers
J . Org. Chem., Vol. 64, No. 14, 1999 5107
cytosine site and 25% for the thymine site, respectively.8
The electron adduct of the thymine site seems to be
protonated at the C6 to form the 5-yl radical, in accord
with the formation of the 5,6-dihydrothymine structure
as a major damage fragment in the γ-irradiated aqueous
solution of E. coli DNA.36 In the model reactions, the C6-
protonation of the electron-thymine adduct occurs more
slowly than the corresponding O4-protonation (k < 2 ×
Sch em e 3
37
38
104 s-1
)
and the N3 protonation (k ) 3.3 × 106 s-1
)
of
the electron-cytosine adduct. Recently, it has also been
suggested that the 5,6-dihydrothymin-5-yl radical gener-
ated within the DNA duplex may be a precursor to
alkaline-labile 2′-deoxyribonolactone generated via radi-
cal abstraction of a hydrogen atom at C1′ of the adjacent
deoxyribose unit.39
According to the present mechanistic study in a model
reaction system, it may be postulated that the C5-C5′-
linked dihydrothymine dimer structure could be produced
by coupling of a pair of thymin-5-yl radicals when
generated thorough migration of two electrons to a given
adjacent thymine-thymine sequence of DNA. Such a
hypothetical C5-C5′-linked dimerization in DNA may
not occur by sparsely ionizing low LET radiations under
usual conditions due to a mechanistic restriction, even
though it seems to be structurally feasible. However, the
densely ionizing high LET radiations may permit gen-
eration of two electrons as the net for migration and
reduction to form an adjacent electron-thymine adduct
pair. Recently, the effect of high LET O7+ heavy ion beam
on thymidine in the solid state has been investigated to
identify several dimeric structures among the thymidine
decomposition products.40 These dimeric products are
possibly attributed to coupling of several types of mon-
omeric radicals that should be more enhanced by densely
ionizing high LET radiations to generate nonhomoge-
neously high concentration of radicals, relative to sparsely
ionizing low LET radiations.
From the redox properties of the oxygen-protonated
and carbon-protonated pyrimidine radicals,10c it may be
expected that disproportionation of radicals by electron
transfer from reducing radicals 5′a -c to oxidizing 5-yl
radicals 5a -c (path 6) may occur especially in acidic and
neutral solutions to reproduce 5,6-dihydrothymines 2a -c
along with regeneration of 1a -c, as shown in Scheme 4.
However, this disproportionation would not be an im-
portant route to 2a -c because the 5-yl radical does not
readily undergo electron-transfer reactions,29 thus being
rather of neutral radical nature than oxidizing. Such a
neutral radical nature of 5a -c accounts for ready bimo-
lecular radical coupling to form C5-C5′-linked dihy-
drothymine dimers 3a -c as the major products.
It is likely that the 5-yl radicals are a common
intermediate for competition between radical coupling
into C5-C5′-linked dimer and hydrogen abstraction into
5,6-dihydrothymine. This competition kinetics was con-
firmed in the radiolytic reduction of 1c (0.1 mM) in Ar-
saturated phosphate buffer (pH 7.0) with various con-
centrations of formate ions as a hydrogen donor. As
shown in Figure 6, the yields of 5,6-dihydrothymidines
2c[S] and 2c[R] increased, and inversely, those of the
C5-C5′-linked dimers 3c decreased upon increasing the
concentration of sodium formate. Another support was
that the yields of 2a ,b were 2-fold depressed, i.e., about
60% yield with formate to 30% yield with 2-methyl-2-
propanol, when the radiolytic reduction of 1a ,b was per-
formed in Ar-purged phosphate buffer (pH 7.0) containing
2-methyl-2-propanol (100 mM) instead of sodium formate.
Biologica l Im p lica tion of C5-C5′-Lin k ed Dih y-
d r oth ym in e Dim er s. Radiation biological studies have
shown that direct ionization of DNA generates DNA
radical cations (holes) and electrons to result in fixation
of DNA damages.1 There have also been several results
indicating that DNA is able to transport hole31 and
electron32-35 through a DNA π-stack.3 In association with
the intra-DNA electron migration, previous ESR studies
evaluated the relative abundance of DNA-base radical
anions (electron-base adducts) as about 75% for the
Exp er im en ta l Section
Ma t er ia ls. 1-Methylthymine (Sigma Chemical), thymine
(Kohjin), and thymidine (Kohjin) were used as received.
Purified 1,3-dimethylthymine was kindly supplied by Fujii
Memorial Research Institute, Otsuka Pharmaceutical. Sodium
formate and 2-methyl-2-propanol were purchased from Nacalai
Tesque and were used without further purification. NaH2PO4
and methanol (HPLC grade) were used as received from Wako
Pure Chemical Industries.
(33) (a) Murphy, C. J .; Arkin, M. R.; J enkins, Y.; Ghatlia, N. D.;
Bossmann, S. H.; Turro, N. J .; Barton, J . K. Science 1993, 262, 1025-
1029. (b) Arkin, M. R.; Stemp, E. D. A.; Holmlin, R. E.; Barton, J . K.;
Ho¨rmann, A.; Olson, E. J . C, Barbara, P. F. Science 1996, 273, 475-
480. (c) Holmlin, R. E.; Dandliker, P. J .; Barton, J . K. Angew. Chem.,
Int. Ed. Engl. 1997, 36, 2714-2730. (d) Olson, E. J . C.; Hu, D.;
Ho¨rmann, A.; Barbara, P. F. J . Phys. Chem. B 1997, 101, 299-303.
(34) Murphy, C. J .; Arkin, M. R.; Ghatlia, N. D.; Bossmann, S.;
Turro, N. J .; Barton, J . K. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 5315-
5319.
(35) Netzel, T. L. J . Chem. Educ. 1997, 74, 646-651.
(36) Te´oule, R.; Bonicel, A.; Bert, C.; Fouque, B. J . Am. Chem. Soc.
1978, 100, 6749-6750.
(37) Shragge, P. C.; Hunt, J . W. Radiat. Res. 1974, 60, 233-249.
(38) Hissung, A.; von Sonntag, C. Int. J . Radiat. Biol. 1979, 35, 449-
458.
(39) (a) Barvian, M. R.; Greenberg, M. M. J . Org. Chem. 1995, 60,
1916-1917. (b) Barvian, M. R.; Greenberg, M. M. J . Am. Chem. Soc.
1995, 117, 8291-8292. (c) Greenberg, M. M.; Barvian, M. R.; Cook, G.
P.; Goodman, B. K.; Matray, T. J .; Tronche, C.; Venkatesan, H. J . Am.
Chem. Soc. 1997, 119, 1828-1839. (d) Kotera, M.; Bourdat, A.-G.;
Defrancq, E.; Lhomme, J . J . Am. Chem. Soc. 1998, 120, 11810-11811.
(40) Gromova, M.; Balanzat, E.; Gervais, B.; Nardin, R.; Cadet, J .
Int. J . Radiat. Biol. 1998, 74, 81-97.
(31) (a) Saito, I.; Takayama, M.; Sugiyama, H.; Nakatani, K.;
Tsuchida, A.; Yamamoto, M. J . Am. Chem. Soc. 1995, 117, 6406-6407.
(b) Sugiyama, H.; Saito, I. J . Am. Chem. Soc. 1996, 118, 7063-7068.
(c) Breslin, D. T.; Schuster, G. B. J . Am. Chem. Soc. 1996, 118, 2311-
2319.
(32) Dandliker, P. J .; Holmlin, R. E.; Barton, J . K. Science 1997,
275, 1465-1468.

5108 J . Org. Chem., Vol. 64, No. 14, 1999
Ito et al.
Sch em e 4
MHz) Fourier transform NMR spectrometer. The C5-C5′-
linked dimers produced from thymidine were measured in D2O
with reference to 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid
(TSP) sodium salt (Aldrich) using EX-400, and those from
1-methylthymine and 1,3-dimethylthymine in dimethyl sul-
foxide-d6 (Nacalai Tesque) were measured using GSX-270.
Chemical shifts in DMSO-d6 were expressed from the residual
1
proton signals of DMSO (δ ) 2.49 ppm) for H NMR and from
the carbon signals (δ ) 39.50 ppm) for 13C NMR, respectively.
High-resolution positive FAB mass spectra (FAB-HRMS)
were recorded on a J EOL J MS-SX102A spectrometer, using
glycerol matrix: 5,5′-bi-5,6-dihydro-1-methylthymines calcd for
C
12H18N4O4 283.141 (M + H)+, found 283.142 (3a [m eso]) and
F igu r e 6. Variation of the yields of dihydrothymidines (b,
2c[S]; O, 2c[R]) and C5-C5′-linked dihydrothymidine dimers
(0, 3c[RS]; 9, 3c[RR] + 3c[SS]) in the γ-irradiation (150 Gy)
of Ar-purged phosphate buffer solutions (pH 7.0) of 1c (0.1 mM)
with increasing concentrations of sodium formate.
283.143 (3a [r a c]); 5,5′-bi-5,6-dihydro-1,3-dimethylthymines
calcd for C14H22N4O4 311.172, found 311.173 (3b[m eso]) and
311.173 (3b [r a c]); 5,5′-bi-5,6-dihydrothymidines calcd for
C
20H30N4O10 487.204, found 487.204 (3c[RS]), 487.206, and
487.206 (3c[RR] and 3c[SS]).
X-r a y Cr ysta llogr a p h y. The colorless needle-shaped crys-
tals of C5-C5′-linked dihydrothymine dimers were grown from
aqueous solutions (∼3 mg mL-1). While the dihydrothymidine
dimers were coated with 40 vol % collodion-ethanol solution,
the other dihydrothymine dimers were mounted directly on a
glass fiber. All measurements were made on a Rigaku AFC7R
diffractometer with graphite-monochromated Mo-KR radia-
tion (λ ) 0.71609 Å) and a 12-kW rotating anode generator.
The diffraction data were collected at room temperature using
the ω - 2θ scan mode to a maximum 2θ value of 60.0° for
3a [m eso] and 3a [r a c], 55.1° for 3b[m eso], and 55.0° for
3b[r a c] and 3c[RS]. The intensities of three representative
reflections that were measured after every 150 reflections
remained constant throughout the data collection, indicating
electronic stability of the sample crystals except for 3c[RS].
The structures were solved by direct methods (SHELXS 86,41
SIR 8842) and expanded using a Fourier technique. The
positions of the hydrogen atoms were calculated. While
hydrogen atoms of 3a [m eso] were refined isotropically, only
isotropic B values were refined for hydrogen atoms of the other
dimers.
Ra d iolytic Red u ction of N1-Su bstitu ted Th ym in e De-
r iva tives. Aqueous solutions of N1-substituted thymine de-
rivatives (1a -c; 5 mM) containing an excess amount (1 M) of
sodium formate were prepared with water ion-exchanged using
a Corning Mega-Pure System MP-190 (>16 MΩ cm). Typically,
the pH of the aqueous solutions was adjusted to 7.0 ( 0.1 with
phosphate buffer (2 mM). For the experiments on pH depen-
dence of the radiolytic reduction, aqueous solutions of 1a -c
(1 mM) containing sodium formate (200 mM) or 2-methyl-2-
propanol (100 mM) were buffered at pH 3-11 with phosphate
buffer (2 mM) and sodium hydroxide. All the solutions were
purged with Ar and irradiated in a sealed glass ampule at
room temperature with a 60Co γ-ray (dose rate: 0.89 Gy min-1
)
or an X-ray source (5 Gy min-1) (Rigaku RADIOFLEX-350).
HP LC. Aliquots (10 µL) of the irradiated solutions were
analyzed by high-performance liquid chromatography (HPLC),
using a Shimadzu 10A HPLC system equipped with a Rheo-
dyne 7725 sample injector. Sample solutions were injected onto
a reversed-phase column (Wakosil 5C18, φ 4.6 mm × 150 mm)
containing C18 chemically bonded silica gel (5 µm particle size)
for analysis of 3a ,b. The C5-C5′-linked dihydrothymidine
dimers (3c) were analyzed using a reversed-phase column
(Nacalai Cosmosil 5C18 MS, φ 4.6 mm × 250 mm). The
phosphate buffer solutions (10 mM, pH 3.0) containing varying
concentrations of methanol (10-25 vol %) were delivered as
the mobile phase at a flow rate of 0.6 mL min-1. The column
eluents were monitored by the UV absorbance at 210 nm. For
isolation of the products, the irradiated solutions were evapo-
rated to a minimum volume and were subjected to preparative
HPLC, using a Tosoh Preparative HPLC system equipped with
a Chromatocorder 12 (System Instruments). The isolation was
performed on a reversed-phase column (Wakosil 10C18, φ 10
mm × 300 mm) containing C18 chemically bonded silica gel
(10 µm particle size), and phosphate buffer solution (10 mM,
pH 3.0) containing 5-20 vol % methanol was delivered at a
flow rate of 3 mL min-1. After aqueous fractions of the
respective products were collected automatically and evapo-
rated, the resulting residues were lyophilized and subjected
to spectroscopic measurements.
Ack n ow led gm en t. This work was supported by
Grants-in-Aid for J SPS Fellows (No. 00086200) and for
Scientific Research (C) (No. 09640632) from the Ministry
of Education, Science, Sports, and Culture of J apan. We
are grateful to Dr. Toˆru Ueno for performing fast atom
bombardment mass spectral analysis.
1
Su p p or tin g In for m a tion Ava ila ble: Copies of H NMR
and 13C NMR spectra and X-ray crystallographic data of
3a ,b[m eso], 3a ,b[r a c], 3c[RS], 3c[RR], and 3c[SS] (except
for X-ray structural data of 3c[RR] and 3c[SS]). This material
is available free of charge via the Internet at http://pubs.acs.org.
J O990059S
(41) Sheldrick, G. M. In Crystallographic Computing 3; Sheldrick,
G. M., Kruger, C., Goddard, R., Eds.; Oxford University Press: Oxford,
1985; pp 175-189.
(42) Burla, M. C.; Camalli, M.; Cascarano, G.; Giacovazzo, C.;
Polidori, G.; Spagna, R.; Viterbo, D. J . Appl. Crystallogr. 1989, 22, 289-
303.
1
Sp ectr oscop ic Mea su r em en ts. H and 13C NMR spectra
were recorded on a J EOL GSX-270 (270 MHz) or EX-400 (400