DNA damage induced via 5,6-dihydrothymid-5-yl in single-stranded oligonucleotides

Article abstract of DOI:10.1021/ja9533510

5,6-Dihydrothymid-5-yl (4) is generated via Norrish type I cleavage of isopropyl ketone 7. Ketone 7 was site specifically incorporated into chemically synthesized polythymidylates and an oligonucleotide containing all four native deoxyribonucleotides. No

Full text of DOI:10.1021/ja9533510

1828
J. Am. Chem. Soc. 1997, 119, 1828-1839
DNA Damage Induced via 5,6-Dihydrothymid-5-yl in
Single-Stranded Oligonucleotides
Marc M. Greenberg,* Mark R. Barvian, Gary P. Cook, Brian K. Goodman,
Tracy J. Matray, Christopher Tronche, and Hariharan Venkatesan
Contribution from the Department of Chemistry, Colorado State UniVersity,
Fort Collins, Colorado 80523
ReceiVed October 2, 1995. ReVised Manuscript ReceiVed July 8, 1996X
Abstract: 5,6-Dihydrothymid-5-yl (4) is generated via Norrish type I cleavage of isopropyl ketone 7. Ketone 7 was
site specifically incorporated into chemically synthesized polythymidylates and an oligonucleotide containing all
four native deoxyribonucleotides. No damage is induced in oligonucleotides containing 7 upon photolysis under
anaerobic conditions. In the presence of O2, strand breaks and alkaline labile lesions are formed at the original site
of 7, and at nucleotides adjacent to the 5′-phosphate of 7. Kinetic isotope effect experiments reveal that direct
strand scission at the thymidine adjacent to the 5′-phosphate of 4 arises from C1′ hydrogen atom abstraction. The
observed KIE (∼3.9) is attributed to hydrogen atom abstraction from C1′ by the peroxyl radical 35 derived from 4.
Enzymatic end group analysis and measurement of free base release are consistent with a process involving C1′
hydrogen atom abstraction. Cleavage experiments carried out in the presence of t-BuOH (1.05 M) and NaN3 (10
mM) indicate that damage does not result from hydroxyl radical, but that 1O2 is responsible for a significant amount
of the observed strand damage.
Ionizing radiation is the agent that has been used the longest
Scheme 1
to purposefully damage nucleic acids.1 Unlike many of the
synthetic and natural product based nucleic acid damaging
agents, which predominantly react with the carbohydrate
component of nucleic acids, nucleobase-centered reactive
intermediates comprise the majority of reactive intermediates
produced by ionizing radiation (Scheme 1).2-6 In addition to
answering fundamental mechanistic questions regarding the
effects of ionizing radiation on nucleic acids, the elucidation
of the role that individual nucleobase reactive intermediates play
in producing nucleic acid strand breaks presents the possibility
that new reaction pathways will be uncovered. This, in turn,
could provide the impetus for the design of new nucleic acid
damaging agents, or improve ones already in use.
The effects of ionizing radiation on pyrimidine substrates
(such as thymidine and its respective nucleotide and polymeric
congeners) have been the focus of numerous investigations.1
The principal radical species produced via ionization of water,
hydroxyl radical and hydrogen atom, show opposite regiose-
lectivity in their addition to the thymine double bond (Scheme
1). The regioselectivity is maintained for each reactive species,
whether thymine is present as a monomer, or as part of a
biopolymer. The preferred C6 hydrogen atom addition product,
5,6-dihydrothymid-5-yl (4), is also formed via the direct effect
of ionizing radiation. The direct effect of ionizing radiation is
believed to give rise to 4 via addition of an electron, followed
by protonation of the resulting radical anion.7 In order for
nucleobase-centered radicals, such as 4, to induce a direct strand
break, spin must be transferred from the nucleobase to a sugar
moiety. Pyrimidine-based reactive intermediates have been
proposed to induce strand breaks by abstracting a hydrogen atom
from the carbohydrate portions of the nucleotides covalently
bound to their 5′- or 3′-phosphates.1,7a,8 It has also been
suggested that the more reactive σ radical, 2′-deoxyurid-5-yl
(5), translates damage in duplex DNA in the 5′ direction via a
mechanism that involves hydrogen atom abstraction from an
adjacent carbohydrate substrate.9,10
X Abstract published in AdVance ACS Abstracts, February 15, 1997.
(1) von Sonntag, C. The Chemical Basis of Radiation Biology; Taylor
and Francis: London, 1987.
(2) (a) Duff, R. J.; de Vroom, E.; Geluk, A.; Hecht, S. M.; van der Marel,
G. A.; van Boom, J. H. J. Am. Chem. Soc. 1993, 115, 3350. (b) McGall, G.
H.; Rabow, L. E.; Ashley, G. W.; Wu, S. H.; Kozarich, J. W.; Stubbe, J. J.
Am. Chem. Soc. 1992, 114, 4958. (c) Hangeland, J. J.; De Voss, J. J.; Heath,
J. A.; Townsend, C. A. J. Am. Chem. Soc. 1992, 114, 9200. (d) Zein, N.;
Poncin, M.; Nilakantan, R.; Ellestead, G. A. Science 1989, 244, 697. (e)
Goldberg, I. H. Acc. Chem. Res. 1991, 24, 191. (f) Myers, A. G.; Cohen,
S. B.; Kwon, B-M. J. Am. Chem. Soc. 1994, 116, 1670.
(3) (a) Goyne, T. E.; Sigman, D. S. J. Am. Chem. Soc. 1987, 109, 2846.
(b) Sigman, D. S. Acc. Chem. Res. 1986, 19, 180.
(4) (a) Cheng, C-C.; Goll, J. G.; Neyhart, G. A.; Welch, T. W.; Singh,
P.; Thorp, H. H. J. Am. Chem. Soc. 1995, 117, 2970. (b) Neyhart, G. A.;
Cheng, C-C.; Thorp, H. H. J. Am. Chem. Soc. 1995, 117, 1463.
(5) Sitlani, A.; Long, E. C.; Pyle, A. M.; Barton, J. K. J. Am. Chem.
Soc. 1992, 114, 2303.
(7) (a) Symons, M. C. R. J. Chem. Soc., Faraday Trans. 1 1987, 83, 1.
(b) Bernhard, W. A. J. Phys. Chem. 1989, 93, 2187.
(8) (a) Becker, D.; Sevilla, M. D. In AdVances in Radiation Biology,
Volume 17, DNA and Chromatin Damage Caused by Radiation; Lett, J.
T., Sinclair, W. K., Eds.; Academic Press: New York, 1987. (b) Cullis, P.
M.; Langman, S.; Podmore, I. D.; Symons, M. C. R. J. Chem. Soc., Faraday
Trans. 1990, 86, 3267.
(9) (a) Sugiyama, H.; Tsutsumi, Y.; Saito, I. J. Am. Chem. Soc. 1990,
112, 6720. (b) Murray, V.; Martin, R. F. Nucleic Acids Res. 1989, 17, 2675.
(c) Hutchinson, F.; Ko¨hnlein, W. Prog. Subcell. Biol. 1980, 7, 1.
(10) Cook, G. P.; Greenberg, M. M. J. Am. Chem. Soc. 1996, 118, 10025.
(6) Pitie´, M.; Bernadou, J.; Meunier, B. J. Am. Chem. Soc. 1995, 117,
2935.
S0002-7863(95)03351-8 CCC: $14.00 © 1997 American Chemical Society

DNA Damage Induced Via 5,6-Dihydrothymid-5-yl
J. Am. Chem. Soc., Vol. 119, No. 8, 1997 1829
Scheme 2
To help elucidate the mechanisms by which ionizing radiation
cleaves DNA, we have examined the reactivity of nucleobase-
centered radicals under biologically relevant conditions (eqs 1
and 2). Generating 1 and 4 in solution by irradiation of 6 and
Scheme 3a
7, respectively, enables us to investigate the chemistry of these
radicals under conditions in which other radicals derived from
nucleosides are not formed.11,12 This approach offers a distinct
advantage over the use of ionizing radiation, in which multiple
reactive intermediates are formed upon irradiation. Contrary
to previous proposals regarding DNA damage amplification via
hydrogen atom abstraction by 4, we demonstrated that this
species is incapable of inducing strand breaks in single-stranded
polythymidylate via hydrogen atom abstraction under anaerobic
conditions.7a,8,12 Direct strand breaks and alkaline labile lesions
are produced when 4 is generated in the presence of O2. Further
characterization of the nature of this damage and proposals for
its formation are described below.
a
Key: (a) H2, Rh/Al2O3, MeOH/H2O (1:1), (b) TBSCl, pyridine,
(c) sec-BuLi (2.5 equiv), THF and then aroyl chloride, (d) HOAc, H2O.
Scheme 4a
Results and Discussion
Synthesis of Photosubstrates and Reaction Products. The
C5-hydroxyl radical adduct of thymidine (1) was previously
generated from 6 via a photoinduced single electron transfer
process.11,13 This process is less useful for producing radicals
in biopolymers, due to the possibility of electron transfer
between the photosensitizer (e.g. N-methylcarbazole) and
random nucleotides. This method is also incompatible with
studying radical processes under aerobic conditions, because
the benzoate radical anion is readily oxidized by O2. Ketones
contain a weak n,π* transition, which can be populated via direct
irradiation at >300 nm. Consequently, we sought to generate
4 via Norrish type I photocleavage.14,15 In designing ketone
precursors to 4, we wished to minimize the formation of other
reactive intermediates (such as the acyl derivative of 4 (8)
Scheme 2). On the basis of the comparison between 4 and the
tert-butyl radical, we expected decarbonylation of the acyl
a
Key: (a) sec-BuLi (2.5 equiv), THF and then isobutyraldehyde or
benzaldehyde, (b) Dess-Martin periodinane, CH2Cl2, (c) NH4F, MeOH
or HOAc/H2O.
radical progenitor of 4 (8) to be g105 s-1 16
Nonetheless, we
chose to synthesize aryl and alkyl ketones for which the type
I cleavage could be expected to yield 4 directly, and not
via 8.
Formation of the C5-acylated dihydropyrimidines was ac-
complished via acylation of the dianion of 9 (Schemes 3 and
4). The protected forms of aromatic ketones 10a,b were
obtainable directly via trapping of the dianion by the respective
acyl chloride (Scheme 3). However, employing a two-step aldol
condensation/oxidation sequence (Scheme 4) gave rise to more
reproducible yields, and facilitated product purification. Desi-
lylations of the aromatic ketones 10a,b and the isopropyl ketone
13 were carried out using either aqueous acetic acid or NH4F
in MeOH.17
.
(11) Barvian, M. R.; Greenberg, M. M. J. Am. Chem. Soc. 1995, 117,
4894.
(12) (a) Barvian, M. R.; Greenberg, M. M. J. Am. Chem. Soc. 1995,
117, 8291. (b) Barvian, M. R.; Greenberg, M. M. J. Org. Chem. 1995, 60,
1916.
(13) (a) Saito, I.; Ikehira, H.; Kasatani, R.; Watanabe, M.; Matsuura, T.
J. Am. Chem. Soc. 1986, 108, 3115. (b) Masnovi, J. J. Am. Chem. Soc.
1989, 111, 9081.
(14) Bohne, C. In CRC Handbook of Organic Photochemistry and
Photobiology; Horspool, W. M., Song, P.-S., Eds.; CRC Press: Boca Raton,
FL, 1995.
(16) (a) Fischer, H.; Paul, H. Acc. Chem. Res. 1987, 20, 200. (b) Lunazzi,
L.; Ingold, K. U.; Scaiano, J. C. J. Phys. Chem. 1983, 87, 529. (c) Hany,
R.; Fischer, H. J. Chem. Phys. 1993, 172, 131.
(15) Turro, N. J. Modern Molecular Photochemistry; Benjamin/Cum-
mings: Menlo Park, CA, 1978.

1830 J. Am. Chem. Soc., Vol. 119, No. 8, 1997
Greenberg et al.
Additional products can be formed via reaction of the radical
pairs formed from 7. Radical pair chemistry can follow two
pathways. Disproportionation is expected to produce dihy-
drothymidine and/or thymidine, while recombination can occur
at two sites in the conjugated radical 4. The product resulting
from recombination on oxygen (16, an N,O-ketene acetal, or
tautomeric 17, an O-alkyl imidate) proved to be too unstable to
be isolable by column chromatography or HPLC. Compounds
similar to 16 and 17 are known to be unstable.19,20 Similar
difficulties were not anticipated for the independent preparation
of recombination product 18. Accordingly, while alkylation of
the dianion of 9 with 2-iodopropane was unsuccessful, 18 was
obtained via sequential alkylation of 19 (eq 3).
Figure 1. X-ray crystal structure of 7.
In general, the diastereoselectivity of the condensation
reactions was approximately 3:1. On the basis of previous
electrophilic reactions with the double bond of protected
thymidine derivatives, the 5S stereoisomer was expected to be
the major product.18 Structural assignments were complicated
by the fact that the four stereoisomers obtained from the
condensation reaction, as well as the diastereomers of protected
(10, 13) and deprotected (7, 11) ketones, were inseparable by
column chromatography. However, separation of the diaster-
eomers of 7 was achieved following formation of the respective
5′-O-dimethoxytritylated ketones 14. Following detritylation
Monomer Photochemistry. As mentioned above, the pho-
tolabile radical precursors were designed to take advantage of
the stepwise nature of the Norrish type I photocleavage
reaction.14,15,21 Aromatic ketones such as 11 can only cleave
to yield 4 and the aroyl species. However, photochemical
cleavage of alkyl ketones (e.g., 7) can yield 4 directly, or via
decarbonylation of 8 (Scheme 2). The acyl radical 8 is not
expected to be proficient at hydrogen atom abstraction. Fur-
thermore, decarbonylation rate constants of acyl radicals that
yield relatively stable alkyl radicals (such as tertiary or benzylic
radicals) are sufficiently great that bimolecular reactions involv-
ing the acyl radical should not compete with this unimolecular
process.16 Comparison of the expected stability of 4 to that of
the tert-butyl radical led us to predict that, if formed, 8 would
decarbonylate at rates equal to, or greater than that of the
pivaloyl system (g1.3 × 105 s-1).16b
of the major diastereomer using aqueous acetic acid, unambigu-
ous assignment of the major stereoisomer of 7 as 5S was
achieved via X-ray crystallography (Figure 1). In addition
to verifying the anticipated stereochemistry of 7, and the
nonplanar nature of the dihydropyrimidine ring, the crystal
structure shows that the nucleoside exists in the C2′-endo,anti
conformation. The anti conformation of 7 suggests that it could
participate in base pairing in duplex DNA. However, the
isopropyl group is projected toward what would be the adjacent
3′-nucleotide, and thus could adversely affect base stacking in
the helix. The effect of 7 on duplex stability awaits experimental
investigation.
It was necessary to verify the fidelity of the photochemical
decomposition of the ketones 7 and 11a prior to incorporation
of these radical precursors into oligonucleotides. This required
the independent preparation of some of the potential reaction
products. While the hydrogen atom trapping product of 4
(5,6-dihydrothymidine) was readily available, other possible
products were unknown. The photoreduction product of 7
(15) was obtained as a complex mixture of diastereomers via
deprotection of 12b (Scheme 4). The analogous product 23
from 11a was isolated from preparative scale photolyses of the
ketone.
Photolysis (λmax) 350 nm) of 11a (10 mM) in the presence
of pseudo-first-order concentrations of cyclohexa-1,4-diene (0.2
M), in a mixture of CH3CN/D2O (1:19), produces 5,6-dihy-
drothymidine (22) in ∼12% yield when the crude photolysate
is analyzed directly by reversed phase HPLC. GC/MS analysis
reveals that all of the 5,6-dihydrothymidine (22) is protonated.
For technical reasons, it was desirable to remove nonpolar
components (e.g., cyclohexa-1,4-diene and byproducts of the
benzoyl radical) prior to HPLC analysis. The aqueous workup
employed to remove organics, as well as heating the sample at
90 °C for 10 min prior to HPLC analysis, increases the yield of
22 to ∼65%. The 5,6-dihydrothymidine (22) formed following
heating contains ∼80% deuterium, indicating that the majority
of this product did not arise from trapping of 4 by the hydrogen
atom donor. Deuterated 22 can arise via radical pair combina-
(19) Hegedus, L. S.; Schultze, L. M.; Montgomery, J. Organometallics
1989, 8, 2189.
(20) Schmir, G. L.; Cunningham, B. A. J. Am. Chem. Soc. 1965, 87,
5692.
(17) Zhang, W.; Robins, M. J. Tetrahedron Lett. 1992, 33, 1177.
(18) Barvian, M. R.; Greenberg, M. M. J. Org. Chem. 1993, 58,
6151.
(21) (a) Engel, P. S. J. Am. Chem. Soc. 1970, 92, 6074. (b) Robbins, W.
K.; Eastman, R. H. J. Am. Chem. Soc. 1970, 92, 6077. (c) Robbins, W. K.;
Eastman, R. H. J. Am. Chem. Soc. 1970, 92, 6079.

DNA Damage Induced Via 5,6-Dihydrothymid-5-yl
J. Am. Chem. Soc., Vol. 119, No. 8, 1997 1831
tion on oxygen (to form 16 and/or 17), followed by hydrolysis,
or photoreduction (e.g., formation of 23) and subsequent
supports, while simultaneously removing the phosphate and
amino protecting groups. This problem was circumvented by
site specifically incorporating 7 as its â-cyanoethyl phosphora-
midite 25 into chemically synthesized oligonucleotides, which
were prepared in a manner that takes advantage of methodology
that obviates the use of NH4OH to effect deprotection.23,24
retrocondensation. As mentioned above, we were unable to
isolate either tautomer of the oxygen recombination product.
Furthermore, neither 16 nor 17 was detectable in crude
1
photolysates by H NMR. Similarly, attempts at deprotecting
12a resulted in formation of 22, illustrating the lability of 23
as well. Fortunately, 23 was isolable from preparative scale
photolyses of 11a in the presence of cyclohexa-1,4-diene (0.2
M). Consequently, we suggest that the majority of 22 formed
via the photolysis of 11a results from retrocondensation of
photoreduction product 23. It was hoped that the p-methox-
yphenyl ketone 11b would undergo less photoreduction in the
presence of hydrogen atom donors, but this was not observed
when 11b was irradiated in the presence of cyclohexa-1,4-
diene.22
An investigation of the photochemistry of 7 was undertaken
to determine if the same complications would arise as was
observed for the aryl ketones 11a and 11b. While the
O-alkylation (O-acylation) products were again undetectable,
the photoreduction (15) and carbon alkylation (18) products were
prepared as described above. Upon irradiation of 7 in CH3-
CN/D2O (1:19), less than 5% of 15 was observed by HPLC,
even in the presence of 0.2 M cyclohexa-1,4-diene. In addition,
the 5,6-dihydrothymidine (22) formed in the presence of
cyclohexa-1,4-diene was completely protonated. In the absence
of exogenous hydrogen atom donor, small amounts (<10%) of
5,6-dihydrothymidine (22) containing large percentages (>33%)
of deuterium were formed.12b Under these conditions, proto-
nated 5,6-dihydrothymidine can arise from disproportionation
within a radical pair, or hydrogen atom abstraction from CH3-
CN, with the former being more likely. It is possible that
deuterated 5,6-dihydrothymidine arises from the analogous
O-alkylation pathway discussed above. If so, then the fact that
none of the radical pair recombination product 18 was observed
could be rationalized on the basis of a large steric preference
for reaction at O4 of 4 within the radical pair.
For most experiments, 7 was introduced into the biopolymer
as a mixture of diastereomers. The separate diastereomers of
7 were incorporated individually, following separation of the
5′-O-dimethoxytritylated ketones 14. All oligonucleotides were
synthesized on one of two photolabile solid phase synthesis
supports (26, 27).23 Oligonucleotides containing only thymidine,
an abasic site model (28), and 7 were treated with anhydrous
diisopropylamine prior to cleavage from the solid phase supports
to remove the â-cyanoethyl protecting groups. Deoxyadenosine,
deoxycytidine, and deoxyguanosine were incorporated into 31
as their allyloxy phosphoramidites.23,24 Thymidine and 7 were
incorporated as their â-cyanoethyl phosphoramidites. The
â-cyanoethyl groups were removed from thymidine and 7 with
anhydrous diisopropylamine prior to removing the allyloxy and
allyloxycarbonyl protecting groups with Pd(0).
The above trapping results are consistent with the generation
and trapping of 4 from 7. Isolation of the diastereomeric
mixture of hydrogen peroxides 24 further supports the generation
of 4, even in the presence of high concentrations of hydrogen
atom donor. Isolated yields of 24 were low due to difficult
purification via reversed phase HPLC, as well as its instability.12b
Structural Effects on Strand Damage. We previously
demonstrated that 5,6-dihydrothymid-5-yl (4) requires O2 to
induce either strand breaks or alkaline labile lesions in 29 and
30. The observed oxygen dependence of damage is independent
of which diastereomer of 7 is present in 29.12a Strand damage
in 29 is readily observed in the 5′ direction up to three
nucleotides away from the original site at which 4 is generated,
Oligonucleotide Synthesis. The isopropyl ketone 7 is stable
to the reagents (trichloroacetic acid, tetrazole, iodine/pyridine/
H2O, and acetic anhydride/N-methylimidazole) used in auto-
mated chemical DNA synthesis. However, 7 is extremely
unstable in concentrated NH4OH, which is the method tradition-
ally used for cleaving oligonucleotides from their solid phase
(23) (a) Matray, T. J.; Greenberg, M. M. J. Am. Chem. Soc. 1994, 116,
6931. (b) Greenberg, M. M.; Gilmore, J. L. J. Org. Chem. 1994, 59, 746.
(c) Venkatesan, H.; Greenberg, M. M. J. Org. Chem. 1996, 61, 525.
(24) Hayakawa, Y.; Wakabayashi, S.; Kato, H.; Noyori, R. J. Am. Chem.
Soc. 1990, 112, 1691.
(22) Yang, N. C.; McClure, D. S.; Murov, S. L.; Houser, J. J.; Dusenbery,
R. J. Am. Chem. Soc. 1967, 89, 5466.

1832 J. Am. Chem. Soc., Vol. 119, No. 8, 1997
Greenberg et al.
Figure 3. Autoradiogram of a 20% denaturing polyacrylamide gel of
photolyzed (6 h) 5′-32P-labeled 31.
Figure 2. Autoradiogram of a 20% denaturing polyacrylamide gel of
photolyzed (6 h) 3′-32P labelled 29.
Table 1. Effect of tert-Butyl Alcohol on Strand Cleavage in
Photolyzed 29 and 31a-c
but is significantly diminished past the site of 28 in 30.12a
Piperidine treatment of photolyzed, 3′-32P-labeled 29 signifi-
cantly alters the cleavage pattern (Figure 2). Whereas two
cleavage products are observed at T11 prior to piperidine
treatment, only the faster moving product is present after alkaline
treatment. In addition, reduction of cleavage at T10 is observed
upon piperidine cleavage, while cleavage at 7 is increased
(Figure 2). The respective behavior of the two cleavage
products at T11 upon piperidine treatment was verified following
their isolation from the gel. The products were isolated and
desalted together as previously described.10 Piperidine treatment
confirmed that the slower moving product is converted into an
oligonucleotide which comigrates with the cleavage product
produced upon NaOH treatment of intact 3′-32P-labeled 29.25
The disparate susceptibilities of the two products produced upon
cleavage at T11 to piperidine are consistent with the interpretation
that the faster moving product contains an intact molecule of 7
(which is stable to piperidine treatment), whereas the slower
moving oligonucleotide fragment contains a more labile mol-
ecule in place of 7. Finally, the disappearance of the product
resulting from cleavage at T10 (Figures 2 and 5) upon piperidine
treatment suggests that it too is associated with the formation
of a product containing an alkaline labile lesion within the
fragmented biopolymer.
oligo-
nucleotide
-
-
-
+
+
-
+
+
piperidine
t-BuOH
29
31
2.70 ( 0.6 2.46 ( 0.1 28.1 ( 3.0 28.7 ( 1.7
2.27 ( 0.2 1.94 ( 0.2 19.2 ( 0.8 20.1 ( 1.6
a Cleavage is expressed in terms of percent of total DNA cleaved.
b [t-BuOH] ) 1.05 M. c hν ) 8 h.
are observed predominantly at the nucleotide (deoxyguanosine;
G in Figure 3) bound to the 5′-phosphate of 7. Direct strand
cleavage at deoxyguanosine is at least 5 times greater than at
the deoxycytidine, which is two nucleotides removed from the
5′-phosphate of 7. The reasons for the different cleavage
patterns observed upon irradiation of 29 and 31 are addressed
below. Regardless of the structure of the biopolymer containing
7 (29 or 31), phosphorimaging analysis reveals that cleavage is
significantly enhanced upon piperidine treatment of photoirra-
diated samples (Table 1).26
Enzymatic End Group Analysis of Oligonucleotide Frag-
mentation Products. Information regarding the nature of the
3′-terminal end groups of products obtained upon irradiation
of oligonucleotides containing 7 was obtained via T4 poly-
nucleotide kinase treatment.27 In the absence of triphosphate,
this enzyme dephosphorylates 3′-terminal phosphates, yielding
Oxygen dependent strand damage is also observed when 4
is produced upon irradiation from 7 in a polymer containing
naturally occurring nucleotides other than thymidine (31; Figure
3). In contrast to 29, alkaline labile and direct strand breaks
(26) Previous estimates of the enhancement of strand damage upon
piperidine treatment were made using densitometric analysis of
autoradiograms.12a These measurements underestimated the extent of
cleavage upon piperidine treatment, due to the limited linear range of the
X-ray film. The current results were obtained using phosphorimaging
analysis.
(25) See the Supporting Information for this paper.

DNA Damage Induced Via 5,6-Dihydrothymid-5-yl
J. Am. Chem. Soc., Vol. 119, No. 8, 1997 1833
Figure 4. Autoradiogram of a 20% denaturing polyacrylamide gel
showing the effects of polynucleotide T4 kinase on photolyzed (6 h)
5′-32P-labeled 29 and 30.
Figure 5. Autoradiogram of a 20% denaturing polyacrylamide gel
showing the effects of calf intestine alkaline phosphatase on photolyzed
(6 h) 3′-32P-labeled 29.
oligonucleotides containing 3′-terminal hydroxyl groups. Oli-
gonucleotides containing the respective end groups are separable
by gel electrophoresis. Although the cleavage patterns observed
in 29-31 are different, enzymatic end group analyses show that
the strand breaks and alkaline labile lesions formed in these
biopolymers consist solely of 3′-phosphate termini (Figure 4).25
The absence of phosphoglycolate termini indicates that hydrogen
atom abstraction from the C4′ position does not occur upon
irradiation of oligonucleotides containing 7.28
observation of a substantial change in migration for two
oligonucleotide fragments having the same sequence and charge,
but differing solely by the presence of a 3′-phosphoglycoalde-
hyde in one case, and a 3′-phosphoglycol in another.5
The structures of the oligonucleotide fragmentation products
at T11 are uncertain and warrant further characterization.
However, the above data (piperidine lability, enzymatic end
group analysis) and other information concerning the results
obtained upon photolysis of 7 lead us to infer that the faster
and slower moving bands arising from cleavage at T11 (Figures
2 and 5) are 33 and 34, respectively (Scheme 5). Aerobic
photolysis of 7 produces 24, which undergoes decomposition
upon storage at -20 °C.12b We expect that an oligonucleotide
containing 24 (34) would migrate more slowly than 33 on
account of its additional hydrogen bonding group, and that this
biopolymer would undergo cleavage upon piperidine treatment.
Hydrogen Atom Abstraction and Free Base Release. Gel
electrophoresis has previously been used to measure kinetic
isotope effects, in order to glean mechanistic information on
nucleic acid damage processes.10,30 The experiment utilized here
enables one to determine an observed KIE by comparing the
Similarly, treatment of 3′-32P-labeled 29 (Figure 5) and 3125
with calf intestine alkaline phosphatase, which dephosphorylates
the 5′-terminus of oligonucleotides, reveals that the 5′-termini
produced via photolysis in the presence of O2 also consist
exclusively of phosphate groups. The observation that the
mobility of both cleavage bands at T11 in 29 is affected upon
treatment with calf intestine alkaline phosphatase is significant.
It suggests that the structures of these cleavage products differ
at a position other than their 5′-termini. We attribute this
surprising behavior to the presence of two oligonucleotide
fragmentation products, each containing a 5′-phosphate, but a
different nucleotide at the position where 7 is incorporated (see
above). As evidenced by the effect of 28 on the migration of
oligonucleotides via gel electrophoresis, structural alterations
of nucleosides and even sequence alterations can exert signifi-
cant effects on the migratory aptitudes of short oligo-
nucleotides.12a,29 Also consistent with this proposal was the
(29) Shibutani, S.; Takeshita, M.; Grollman, A. P. Nature 1991, 349,
431.
(30) (a) Greenberg, M. M.; Dervan, P. B. (California Institute of
Technology). Unpublished results. (b) Kozarich, J. W.; Worth, L., Jr.; Frank,
B. L.; Christner, D. F.; Vanderwall, D. E.; Stubbe, J. Science 1989, 245,
1396. (c) Absalon, M. J.; Wu, W.; Kozarich, J. W.; Stubbe, J. Biochemistry
1995, 34, 2076. (d) Kappen, L. S.; Goldberg, I. H.; Wu, S. H.; Stubbe, J.;
Worth, L., Jr.; Kozarich, J. W. J. Am. Chem. Soc. 1990, 112, 2797. (e)
Kappen, L. S.; Goldberg, I. H.; Frank, B. L.; Worth, L., Jr.; Christner, D.
F.; Kozarich, J. W.; Stubbe, J. Biochemistry 1991, 30, 2034.
(27) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A
Laboratory Manual, 2nd ed.; Cold Spring Harbor Laboratory Press: New
York, 1989.
(28) (a) Stubbe, J.; Kozarich, J. W. Chem. ReV. 1987, 87, 1107. (b) Giese,
B.; Beyrich-Graf, X.; Erdmann, P.; Giraud, L.; Imwinkelried, P.; Mu¨ller,
S. N.; Schwitter, U. J. Am. Chem. Soc. 1995, 117, 6146.

1834 J. Am. Chem. Soc., Vol. 119, No. 8, 1997
Greenberg et al.
Scheme 5
fact that the diminution of strand damage at this nucleotide does
not affect strand damage at nucleotides displaced further in the
5′ direction from 7. This observation eliminates a possible free
radical chain propagation mechanism, analogous to that observed
in lipid peroxidation, as the reason for the cleavage pattern
observed for 29.12a As explained below, we believe that a
portion of the strand damage at sites distal to 7 in polythymidy-
lates is due to the participation of an additional pathway.
The significant KIE observed upon deuteration of C1′ strongly
suggests that hydrogen atom abstraction from this position of
the deoxyribonucleotide covalently bound to the 5′-phosphate
of 7 is involved in the direct strand scission produced during
aerobic photolysis of 32. The absence of a KIE upon dideu-
teration of the C2′ and deuteration of the C4′ positions suggests
that the reactive species responsible for hydrogen atom abstrac-
tion is a very selective one. We expect that the peroxyl radical
derived from 4 (35) would react selectively with an adjacent
Scheme 6
deoxyribonucleotide. On the basis of recent calculations of
enthalpies for hydrogen atom abstraction from deoxyribose, and
the approximate O-H bond strength in a hydroperoxide (∼90
kcal/mol), the C1′ position may be the only thermodynamically
favorable position within a deoxyribonucleotide for hydrogen
atom abstraction by a peroxyl radical.33 Further justification
for proposing that strand breaks are produced via C1′ hydrogen
atom abstraction by 35 is provided by the observation that the
analogous peroxyl radical derived from aerobic photolysis of
5-bromodeoxyuridine (36; eq 5) also selectively abstracts a
hydrogen atom from the C1′ position of the 5′ adjacent
nucleotide in single-stranded DNA.34
Table 2. Observed Kinetic Isotope Effects in 32
position
deuterated
non-piperidine
piperidine
C1′
C2′
C4′
3.9 ( 0.5
1.0 ( 0.1
1.0 ( 0.1
1.3 ( 0.2
1.0 ( 0.2
1.0 ( 0.1
measured ratio of strand breaks at two cleavage sites in two
different oligonucleotides (Scheme 6, eq 4).10,30a One oligo-
nucleotide (control) contains two molecules of 7 where neither
carbohydrate of the adjacent 5′-nucleotides is isotopically
enriched. The ratio of cleavage at the two nucleotides adjacent
to 7 in the control is compared to the relative cleavage of an
oligonucleotide of identical sequence, where the nucleotide
adjacent to the 5′-phosphate of one of the molecules of 7 is
enriched in deuterium at a single position in the carbohydrate
moiety. By carrying out the experiment in this manner
(preferably to a small extent of cleavage), one does not have to
measure the extent of conversion, because the protio site serves
as an internal standard.
The kinetic isotope effects were measured using the above
method with deuterium enrichment at C1′, C2′, and C4′ using
32a-d. Measurements were made for direct strand breaks, as
well as alkaline labile lesions. The deuterated thymidines were
prepared via known procedures, and incorporated into 32b-d
as their phosphoramidite analogues.30a,31,32 In a typical experi-
ment, 4-5 samples of each oligonucleotide were irradiated. The
KIEs reported are derived from the averages of the cleavage
ratios measured in these independent experiments (Table 2).32
Essentially no KIE is observed upon deuteration of C2′ or C4′
of the thymidine at position X in 32, while a significant effect
is observed upon deuteration of C1′. Also of interest is the
Despite the arguments presented above in favor of participa-
tion of 35 in the observed strand scission, the magnitude of the
KIE (∼4.0) observed upon deuteration of C1′ is lower than one
might expect for the involvement of a peroxyl radical.35 The
observed KIE is also smaller than one might expect if an alkoxyl
radical was the species responsible for hydrogen atom abstrac-
tion.35 The selectivity of the process implied by the absence
of a KIE upon deuteration at C4′ and C2′ is more consistent
with the involvement of a peroxyl radical than a more reactive
alkoxyl radical.36,37 Finally, while the magnitude of the
observed KIE could be considered low for hydrogen atom
abstraction by a peroxyl radical, one must realize that the
(33) (a) Benson, S. W. Thermochemical Kinetics, 2nd ed.; John Wiley:
New York, 1976. (b) Miaskiewicz, K.; Osman, R. J. Am. Chem. Soc. 1994,
116, 232.
(34) Cook, G. P.; Koppisch, A. T.; Greenberg, M. M. (Colorado State
University) Unpublished results.
(35) (a) Lewis, E. S.; Ogino, K. J. Am. Chem. Soc. 1976, 98, 2264. (b)
Howard, J. A. In Free Radicals, Volume II; Kochi, J. K., Ed.; John Wiley
and Sons, New York, 1973. (c) Green, M. K.; Boyle, B. A.; Vairamani,
M.; Mukhopadhyay, T.; Saunders, W. H., Jr.; Bowen, P.; Allinger, N. L. J.
Am. Chem. Soc. 1986, 108, 2381. (d) For an example where KIEs for peroxyl
radicals are relatively low see: Evans, C.; Scaiano, J. C.; Ingold, K. U. J.
Am. Chem. Soc. 1992, 114, 4589.
(31) (a) Cook, G. P.; Greenberg, M. M. J. Org. Chem. 1994, 59, 4704.
(b) Kohn, P.; Samaritano, R. H.; Lerner, L. M. J. Am. Chem. Soc. 1965,
87, 5475. (c) Vorbru¨ggen, H.; Bennua, B. Chem. Ber. 1981, 114, 1279.
(32) A sample autoradiogram demonstraing the KIE experiment is
included in the Supporting Information.

DNA Damage Induced Via 5,6-Dihydrothymid-5-yl
J. Am. Chem. Soc., Vol. 119, No. 8, 1997 1835
1
observed KIE in this experiment is a weighted average of the
true KIE for the process, multiplied by the fraction of total
product (strand break) measured that is attributable to this
pathway.30e Alternate reactive species and/or pathways would
result in a reduction of the observed KIE for a process involving
35.
Since strand scission observed upon irradiation of oligonucle-
otides containing 7 is dependent upon oxygen, alternative
pathways might also be expected to involve oxygen. Further-
more, any reactive species in addition to 35 is (are) not required
to induce strand damage via hydrogen atom abstraction from
the carbohydrate moiety. In fact, the significant diminution of
the observed KIE (Table 2), as well as the large overall increase
in strand scission (Table 1) observed upon piperidine treatment,
is consistent with the production of a second reactive species
that does not react via a pathway similar to that followed by
35.
In order to corroborate the KIE experiments, we investigated
the degradation products formed upon irradiation of 29.
Abstraction of the C1′ hydrogen atom of a nucleotide in DNA
typically results in concomitant release of the respective
nucleobase.2e,3b,4 The yield of thymine was determined by
reversed phase HPLC as a function of direct strand break
formation in 29. Irradiation of unlabeled 29 resulted in the
release of ∼40% thymine compared to direct strand breaks,
which were measured by anion exchange HPLC. We were
unable to determine whether thymine degradation products were
formed, due to limitations of HPLC detection. Methylenefura-
none (37) has also been associated with DNA oxidation at C1′
type I cleavage of 7, and (3) production of O2 via quenching
of the excited state of 7 by O2.
We previously discounted the involvement of the excited state
of 7 in strand damage, on the basis of the independence of strand
scission on the stereochemistry of the ketone.12a The observed
dependence of the cleavage pattern on the oligonucleotide
sequence is also inconsistent with this explanation, as it has
recently been shown that the excited states of all four nucleotides
have very similar energies, and are all sensitized by ketones.39
Isopropyl and/or the respective acyl radical is produced in
solution via the Norrish type I photocleavage. These radicals
are capable of inducing nucleic acid damage, and might well
contribute in some manner to the observed cleavage. However,
carbon-centered radicals do not require O2 in order to damage
nucleic acids.40 Hence, we do not believe that such species, or
the peroxyl radicals derived from them (which one would expect
would be even less reactive), are major contributors to the strand
damage derived from photolysis of 7.
The above arguments against hydroxyl, alkyl, and peroxyl
radicals do not rule out all diffusible species as being participants
in strand damage. Various excited states, including those of
ketones, can be quenched by O2 to produce singlet oxygen (1O2).
Indeed, photolysis of 7 in the presence of 1O2 trap 38 indicated
that this reactive species is produced under conditions in which
DNA strand damage is observed.41 The product observed was
40 and not the initially formed endoperoxide 39, which was
shown to decompose to the phenol 40 under the conditions in
which 7 is photolyzed (eq 6). Singlet oxygen is most often
in some, but not all, instances.3a,4,6 We did not detect significant
photodegradation of independently synthesized 37 under our
irradiation conditions.4,38 However, our detection limits (∼25
nmol via GC/MS) prohibited direct observation of the methyl-
enefuranone (37) when 29 was irradiated.
associated with damage of guanine-containing nucleosides.42
Absolute rate measurements show that O2 reacts at least an
1
order of magnitude faster with deoxyguanosine than with other
nucleosides.43 However, it has also been shown that thymidine
monophosphate deactivates 1O2 with a bimolecular rate constant
greater than 105.43b Furthermore, early studies showed that
1
Photochemical Generation of O2 and Its Role in Strand
Damage. The observations regarding the cleavage pattern in
32 that were made while KIE experiments were carried out, as
well as the very different cleavage patterns produced for
irradiation of polythymidylates 29, 30, and 32 and 31, led us to
investigate the possibility that a reactive species other than 35
is responsible for a portion of the O2 dependent strand damage
of irradiated oligonucleotides containing 7.12a The different
cleavage patterns observed in 29-32, as well as the dependence
on O2, suggest that a highly reactive, nonselective diffusible
species, such as hydroxyl radical, was not responsible for the
oligonucleotide damage. This proposal was supported by
photolyses carried out in the presence of t-BuOH (1.05 M), a
hydroxyl radical trap, which had no effect upon the extent of
damage in either 29 or 31 (Table 1).1 Other pathways
considered by us included (1) energy transfer to the neighboring
nucleotides from the excited state of 7, (2) involvement of the
isopropyl and/or respective acyl radical produced via the Norrish
1
thymidine was degraded by O2.44 Both direct and indirect
measurements indicate that thymidine reacts more rapidly than
deoxycytidine and deoxyadenosine with 1O2.43,44 Hence, while
1O2 does react preferentially with deoxyguanosine, it is errone-
ous to assume that this oxidant does not react with other
nucleotides.
The selectivity observed for cleavage in 31, relative to that
observed in polythymidylates (e.g., 29), is consistent with the
1
involvement of O2 in the observed strand damage. Conse-
quently, we probed for the possible involvement of 1O2 in strand
damage, by examining the effect of azide on the direct and
alkaline labile strand breaks in 29 (Table 3).42,43 Addition of
(39) (a) Wood, P. D.; Redmond, R. W. J. Am. Chem. Soc. 1996, 118,
4256. (b) Gut, I. G.; Wood, P. D.; Redmond, R. W. J. Am. Chem. Soc.
1996, 118, 2366.
(40) (a) Milligan, J. R.; Ward, J. F. Radiat. Res. 1994, 137, 295. (b)
Hiramato, Z.; Johkoh, H.; Sako, K-I.; Kikugawa, K. Free Radical Res.
Commun. 1993, 19, 323.
(41) Nardello, V.; Azaroual, N.; Cervoise, I.; Vermeersch, G.; Aubry,
J-M. Tetrahedron 1996, 52, 2031.
(42) (a) Sies, H.; Menck, C. F. M. Mutat. Res. 1992, 275, 367. (b) Sheu,
C.; Foote, C. S. J. Am. Chem. Soc. 1995, 117, 474. (c) Devasagayam, T. P.
A.; Steenken, S.; Obendorf, M. S. W.; Schulz, W. A.; Sies, H. Biochemistry
1991, 30, 6283.
(36) On the basis of the calculated enthalpies for hydrogen atom
abstraction from deoxyribose, and the approximate bond dissociation energy
for an O-H bond in an alcohol (∼104 kcal/mol), we expect that an alkoxyl
radical will exhibit much less selectivity in its reactions with deoxyribo-
nucleotides than a peroxyl radical.33 In a related study, less selective
reactivity has been observed for 5 in single-stranded DNA.34
(37) (a) Foti, M.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc. 1994,
116, 9440. (b) Ingold, K. U. In Free Radicals, Volume I; Kochi, J., Ed.;
John Wiley: New York, 1973. (c) Be´rces, T.; Dombi, J. Int. J. Chem. Kinet.
1980, 12, 183.
(43) (a) Lee, P. C. C.; Rodgers, M. A. J. Photochem. Photobiol. 1987,
45, 79. (b) Rougee´, M.; Benasson, R. V. C. R. Acad. Sci., Ser. II 1986,
302, 1223.
(38) Grundmann, C.; Kober, E. J. Am. Chem. Soc. 1955, 77, 2332.
(44) Simon, M. I.; Van Vunakis, H. J. Mol. Biol. 1962, 4, 488.

1836 J. Am. Chem. Soc., Vol. 119, No. 8, 1997
Table 3. Effect of Azide on Strand Scission in 29a,b
Greenberg et al.
independent generation of the radical produced upon C1′
hydrogen atom abstraction.45
[azide] (mm)
non-piperidine
piperidine
1
Despite being partially obscured by the generation of O2,
0
10
0.66 ( 0.22
0.25 ( 0.04
15.0 ( 1.1
8.1 ( 1.1
the experiments described here support the proposal that two
covalently linked nucleotides in a single strand of DNA can be
damaged via the transposition of spin from an initially damaged
nucleobase to an adjacent carbohydrate moiety.46 This nucleic
acid damage pathway is distinct from those examined involving
other agents, and could provide the impetus for the design of a
new family of nucleic acid damaging agents.2-4
a Cleavage is expressed in terms of percent of total DNA cleaved.
b hν ) 6 h.
Table 4. Effect of D2O on Strand Scission in 29a,b
solvent
non-piperidine
piperidine
H2O
D2O
2.4 ( 0.6
1.7 ( 0.1
27.4 ( 3.1
22.7 ( 2.4
Experimental Section
a Cleavage is expressed in terms of percent of total DNA cleaved.
General Methods. 1H NMR spectra were recorded at 300, 270, or
200 MHz. IR spectra were obtained using a Perkin-Elmer 1600 Series
FT-IR. All reactions were carried out in oven-dried glassware, under
a nitrogen atmosphere, unless specified otherwise. N,O-Bis(trimeth-
ylsilyl)(trifluoromethyl)acetamide containing 1% TMSCl (BSTFA/
TMS) was obtained from Sigma. T4 polynucleotide kinase and calf
intestine alkaline phosphatase were from New England Biolabs.
Terminal deoxynucleotidyl transferase was from US Biochemical.
[R-32P]ddATP and [γ-32P]ATP were from Amersham. Pyridine, N,N-
diisopropylethylamine (Hu¨nig’s base), and CH2Cl2 were distilled from
CaH2. THF was distilled from Na0/benzophenone ketyl. Acetonitrile
was purified and dried by passing over anhydrous CuSO4, followed by
distillation from CaH2.
Oligonucleotides were synthesized on an Applied Biosystems Inc.
380B DNA synthesizer using previously reported cycles.23a,b Allyloxy-
protected phosphoramidites and photolabile solid phase synthesis
supports 26 and 27 were prepared as described.23,24 DNA manipulations
were carried out using standard procedures.27 Chemical sequencing
reactions were performed as described.23a,47 Photolyzed oligonucleotides
were electrophoresed on 20% polyacrylamide denaturing gels (5%
cross-link, 45% (urea by weight)). The gels were quantitated either
using a Molecular Dynamics phosphorimager equipped with Imagequant
Version 3.3 software or by densitometry.
All photolyses of nucleosides and oligonucleotides were carried out
in Pyrex tubes (0.25 in. i.d.) using a Rayonet photoreactor equipped
with lamps having a maximum output at 350 nm. Oligonucleotide
photolyses were carried out in 10 mM phosphate buffer (pH 7.0) and
100 mM NaCl. In experiments involving photolysis of monomeric
materials, yields of 7, 11a, and 22 were determined by reversed phase
HPLC analysis (Rainin Microsorb-MV C18, 0.4 × 25 cm column; A,
H2O; B, CH3CN; 0-35% B linearly over 35 min; flow rate, 1.0 mL/
min) using 5-bromo-2′-deoxyuridine as an internal standard (added after
irradiation). The response factors versus 5-bromo-2′-deoxyuridine were
determined using independently prepared 7 (0.9), 11a (0.5), and 22
(1.4) at 205 nm. The yield of free thymine released during the
photolysis of 29 was determined by reversed phase HPLC using the
above C18 column (A, 10 mM KH2PO4 (pH 6.8), 2.5% MeOH; B, 10
mM KH2PO4 (pH 6.8), 20% MeOH; 0-100% B linearly over 15 min;
flow rate, 1.0 mL/min). Uracil was added as an internal standard after
photolysis (response factor, 1.2 at 254 nm). The loss of 29 was
determined using anion exchange HPLC (Vydac weak anion-exchange
oligonucleotide column; 0.4 × 25 cm; A, 0.1 M (NH4)2HPO4 (pH 6.7),
20% CH3CN; B, 0.3 M (NH4)2HPO4 (pH 6.7), 20% CH3CN; 0-67%
B linearly over 20 min; flow rate, 1.5 mL/min). Singlet oxygen trapping
experiments were analyzed by reversed phase HPLC using the above
C18 column (A, H2O; B, CH3CN; 5% B from t ) 0 to t ) 5 min;
5-50% B linearly over 25 min; 50-5% B linearly over 10 min; flow
rate, 1 mL/min).
b hν ) 8 h.
NaN3 (10 mM) to aerobic photolyses of 29 results in ap-
proximately a 2-fold diminution in overall strand damage.
Independent experiments showed that the azide does not affect
the quantum yield for disappearance of 7. These observations
suggest that 1O2 is partially responsible for the observed strand
damage. In contrast to these experiments, photolytic cleavage
in D2O (which typically enhances strand damage by increasing
1
the lifetime of O2) appears to slightly reduce the extent of
cleavage at 7 and the nucleotides adjacent to its 5′-phosphate
(Table 4). Statistically, the extents of cleavage in H2O and D2O
are within experimental error of each other. A possible reason
for this lack of an effect by D2O can be explained by considering
1
the fact that O2 is generated in the vicinity of the oligonucle-
otides that are cleaved. This is far different from the situation
when 1O2 is produced in the bulk of a dilute solution of DNA.
1
The lifetime of O2 in H2O is >40 µs, and it is estimated to
have a diffusion radius of ∼100 Å.42a Consequently, 1O2
diffuses many angstroms greater than that required to span the
distance (<20 Å) of the nucleotides (7 to T9) over which damage
is measured in 29. Hence, increasing the 1O2 lifetime by adding
D2O does not necessarily have to increase the amount of strand
damage at nucleotides 7 through T9 in 29.
1
Finally, the involvement of O2 in strand damage is also
consistent with the cleavage pattern observed for 3′-32P-labeled
oligonucleotides, as well as the enzymatic end group analysis
of oligonucleotides labeled at this terminus. Production of 1O2
has no effect on the structure of 7. Hence, the formation of
two cleavage (33, 34) products containing 5′-phosphate termini
upon cleavage at the nucleotide adjacent to the 5′-phosphate of
7 (Scheme 5) is consistent with cleavage induced via 35
(yielding a cleavage product, 34, that is itself alkaline labile)
1
and O2 (yielding a product, 33, containing an intact molecule
of 7, which is not susceptible to piperidine treatment).
Summary and Conclusions. Isopropyl ketone 7 was chosen
as a precursor to 4 in order to influence the initial bond cleavage
during the Norrish type I photocleavage. However, this choice
of ketone inadvertently complicated the photochemistry. Previ-
ously observed strand damage, and that characterized above,
are consistent with the involvement of 35 (from 4) and 1O2 (due
to quenching of the excited state of 7 by O2).12
Kinetic isotope effect experiments and enzymatic end group
analysis are consistent with a reaction mechanism involving
hydrogen atom abstraction by 35 from the C1′ position at the
5′ adjacent nucleotide (Scheme 7).3,6 Estimates of the thermo-
dynamics of this process are consistent with the observed
selectivity of this process.33 The details for the formation of a
strand break following hydrogen atom abstraction from the C1′
position, and the nature of the deoxyribose degradation product-
(s), are currently under investigation. Utilizing photochemical
precursors to 4 that are less prone to 1O2 generation (e.g., tert-
butyl and/or benzyl ketones) will simplify these studies, as will
Quantitation of Thymine Release as a Function of Strand Break
Formation in 29. Oligonucleotide 29 (56 µM, total volume 100 µL)
was photolyzed for 28 h as described above. Prior to, and after
photolysis, 5 µL of the photolysis sample was removed, spiked with a
24 nucleotide long oligonucleotide (11.8 µM, 5 µL), and analyzed by
(45) (a) Goodman, B. K.; Greenberg, M. M. J. Org. Chem. 1996, 61, 2.
(b) Gimisis, T.; Chatgilialoglu, C. J. Org. Chem. 1996, 61, 1908.
(46) Box, H. C.; Budzinski, E. E.; Freund, H. G.; Evans, M. S.; Patrzyc,
H. B.; Wallace, J. C.; Maccubbin, A. E. Int. J. Radiat. Biol. 1993, 64, 261.
(47) Maxam, A. M.; Gilbert, W. S. Methods Enzymol. 1980, 65, 499.

DNA Damage Induced Via 5,6-Dihydrothymid-5-yl
J. Am. Chem. Soc., Vol. 119, No. 8, 1997 1837
Scheme 7
anion exchange HPLC. After photolysis, the sample was filtered
through an Amicon membrane filter. The membrane was washed with
H2O (100 µL), prior to speed-vacing the filtrate to dryness. The filtrate
was resuspended in an aqueous solution of uracil (9.8 µM, 100 µL),
and analyzed by reversed phase HPLC.
organic layers were washed with brine (25 mL) and dried over MgSO4.
Flash chromatography (CH2Cl2/EtOAc; 12:1) afforded 13 (229 mg,
74%) as a hygroscopic foam: 1H NMR (CDCl3) δ 7.68 (br d s, 1H,
minor), 7.63 (br d s, 1H, major), 6.25 (dd, 1H, J ) 6.2, 6.3 Hz, minor),
6.12 (dd, 1H, J ) 6.0, 6.0 Hz, major), 4.39 (m, 1H, minor), 4.29 (m,
1H, major), 4.08 (d, 0.55H, J ) 13 Hz, major), 3.84-3.79 (m, 1H),
3.75-3.65 (m, 2.45H, major + minor), 3.15-2.98 (m, 2H), 2.30-
2.21 (m, 0.45H, minor), 2.01-1.81 (m, 1.55H, major + minor), 1.37
(s, 3H), 1.07-0.95 (m, 6H), 0.92-0.84 (m, 18H), 0.08-0.02 (m, 12H);
IR (film) 3083 (br d), 2947, 2920, 2857, 1691, 1473, 1437, 1383, 1360,
Detection of Singlet Oxygen upon Photolysis of 7. A methanol
solution (0.2 mL) of 7 (3 mM) was photolyzed in a RPR-100
photoreactor for 6 h at 25 °C in the presence of 38 (20 mM).41 The
crude photolysate was analyzed by reversed phase HPLC and 1H NMR.
In a control experiment, 38 (10 mM) was photolyzed in the presence
of Rose Bengal (1 mg), as described in methanol (5 mL).41 An aliquot
of the crude photolysate was analyzed as described above by reversed
phase HPLC and 1H NMR. Following HPLC analysis, the photolysis
mixture was subjected to the identical photolysis conditions used for
7, and the crude photolysate was again analyzed by reversed phase
1257, 1224, 1089, 1021 cm-1
.
Preparation of 7. Anhydrous NH4F (128 mg, 3.45 mmol) and 13
(229 mg, 0.42 mmol) were stirred in anhydrous MeOH (10 mL) at 50
°C. After 24 h, the mixture was poured into saturated NaHCO3 (100
mL) and extracted with a mixture of isopropyl alcohol and CHCl3 (3:7
(v/v), 5 × 50 mL). The organic layers were pooled, washed with brine
(25 mL), and dried over MgSO4. Flash chromatography (CH2Cl2/
MeOH; 9:1) yielded 92 mg (69%) of 7 as a hygroscopic solid: 1H
NMR (MeOH-d4) δ 6.21 (dd, 1H, J ) 7.4, 6.8 Hz, major), 6.10 (dd,
1H, 6.3, 8.0 Hz, minor), 4.29 (m, 1H), 4.08 (d, 1H, J ) 12.8 Hz, minor),
3.86 (d, 1H, J ) 13.3 Hz, major), 3.78-3.70 (m, 1H), 3.67-3.60 (m,
2H), 3.26 (d, 1H, J ) 13.2 Hz, minor), 3.20 (d, 1H, J ) 12.8 Hz,
major), 3.18-3.12 (m, 1H), 2.41-2.35 (m, 1H, major), 2.2-2.15 (m,
1H, minor), 1.98-1.88 (m, 1H), 1.39 (s, 3H, minor), 1.35 (s, 3H, major),
1.09 (d, 3H, J ) 6.9 Hz, minor), 1.07 (d, 3H, J ) 6.9 Hz, major), 1.04
(d, 3H, J ) 6.9 Hz, minor), 0.96 (d, 3H, J ) 6.9 Hz, major); 13C NMR
(MeOH-d4) δ 213.1 (minor), 211.6 (major), 173.0 (minor), 172.7
(major), 154.7 (major), 154.1 (minor), 87.7 (minor), 87.5 (major), 85.7
(minor), 84.9 (major), 72.5 (minor), 72.3 (major), 63.3, 57.1 (minor),
56.9 (major), 45.6 (minor), 44.9 (major), 37.4, 37.2 (major), 37.0
(minor), 20.6 (minor), 20.0 (major), 18.7 (major), 17.9 (minor); IR
(KBr) 3493 (br d), 3014, 2979, 2961, 1704, 1702, 1485, 1437, 1384,
1291, 1222, 1091, 1053, 1020 cm-1; HRMS (FAB) calcd 315.1537
(M + H), found 315.1556.
1
HPLC and H NMR.
Preparation of 9. Rhodium on alumina (10%, 100 mg) was added
to a solution of thymidine (10 g, 41.28 mmol) in a mixture of methanol
and water (1:1, 75 mL) in a high-pressure bottle. The mixture was
stirred under 50 psi of hydrogen gas for 48 h. The catalyst was removed
by filtration through a pad of Celite and the filtrate lyophilized to give
9.6 g (96%) of dihydrothymidine. TBSCl (5.65 g, 37.50 mmol) was
dissolved in pyridine (25 mL) and added to dihydrothymidine (3.0 g,
12.38 mmol) that had been previously dried from pyridine (3.0 mL)
under vacuum. The reaction was allowed to stir at 40 °C for 24 h,
quenched with MeOH (0.5 mL), poured into H2O (250 mL), and
extracted with diethyl ether (3 × 150 mL). The combined organic
layers were washed with brine (50 mL) and dried over MgSO4. Flash
chromatography (CH2Cl2/EtOAc; 9:1) yielded 9 (5.1 g, 88%): mp 103-
104 °C; 1H NMR (CDCl3) δ 7.31 (br d s, 1H), 6.26 (dd, 1 H, J ) 6.6,
6.6 Hz), 4.34 (m, 1H) 3.77 (m, 1H), 3.69 (m, 2H), 3.37 (dd, 1H, J )
5.4, 12.9 Hz), 3.25 (dd, 1H, J ) 8.9, 13.2 Hz), 2.60 (m, 1H), 1.97 (m,
2H), 1.24 (d, 3H, J ) 7 Hz), 0.89 (s, 9H), 0.87 (s, 9H), 0.06 (s, 6H),
0.05 (s, 6H); IR (film) 3211 (br d), 3077, 2955, 2922, 2888, 2844,
1705, 1472, 1383, 1250, 1111, 1022 cm-1. Anal. Calcd for C22H44N2O5-
Si2: C, 55.88; H, 9.40; N, 5.92. Found: C, 55.67; H, 9.15; N, 5.94.
Preparation of 12b. A cyclohexane solution of sec-BuLi (1.98 mL,
1.2 M) was added via syringe to 9 (450 mg, 0.96 mmol) in THF (4.5
mL) at -78 °C. After 1 h, isobutyraldehyde (104 mg, 1.44 mmol)
was added dropwise via syringe. After stirring for an additional 2 h at
-78 °C, the reaction was quenched with saturated NH4Cl (0.5 mL),
warmed to room temperature, and poured into H2O (50 mL). The
aqueous layer was extracted with ether (2 × 125 mL), and the combined
organic layers were washed with brine (50 mL) and dried over MgSO4.
Flash chromatography (CH2Cl2/EtOAc; 9:1) yielded 12b (314 mg, 60%)
Preparation of 15. Alcohol 12b (210 mg, 0.39 mmol) was stirred
in a mixture of HOAc/H2O (4:1 (v/v), 3 mL) for 24 h. The solvents
were removed in Vacuo. Flash chromatography (CH2Cl2/MeOH, 9:1)
yielded 15 (67 mg, 54%) as a mixture of four diastereomers: 1H NMR
(MeOH-d4) δ 6.10-6.35 (m, 1H), 4.25-4.35 (m, 1H), 3.50-4.00 (m,
4H), 3.15-3.40 (m, 2H), 1.75-2.55 (m, 3H), 1.15-1.25 (m, 3H), 0.85-
0.95 (m, 6H); HRMS (FAB) calcd 317.1712 (M + H), found 317.1710.
Preparation of Hydroperoxides 24. A mixture of 7 (50 mg, 0.159
mmol) and cyclohexa-1,4-diene (276 mg, 3.0 mmol) in a mixture of
water and acetonitrile (3:2 (v/v), 15 mL) was loaded into a Pyrex tube
fitted with a Teflon stopcock. The mixture was photolyzed in a Rayonet
reaction chamber equipped with λmax ) 350 nm bulbs. After 16 h, the
solvent was removed using a Savant Speed-Vac concentrator. The
gummy residue was dissolved in water (500 µL) and filtered through
a 0.45 µm nylon filter. Hydroperoxides 24a and 24b were isolated
using reversed phase HPLC (Rainin C18, 0.4 × 25 cm column; A, H2O;
B, CH3CN; 3-35% B linearly over 35 min), with retention times of
5.6 and 6.8 min, respectively. 24a: 1H NMR (D2O) δ 6.15 (dd, 1H,
J ) 6.8, 6.8 Hz), 4.26 (m, 1H), 3.78 (m, 2H), 3.59 (m, 2H), 3.40 (d,
1H, J ) 13.2 Hz), 2.24-2.00 (m, 2H), 1.34 (s, 3H); HRMS (FAB)
calcd 277.1035 (M + H), found 277.1050. 24b: 1H NMR (D2O) δ
6.16 (dd, 1H, J ) 6.8, 6.8 Hz), 4.26 (m, 1H), 3.81 (m, 2H), 3.75 (d,
1H, J ) 13.5 Hz), 3.61 (m, 2H), 3.39 (d, 1H, J ) 13.5 Hz), 2.17-2.02
(m, 2H), 1.35 (s, 3H); HRMS (FAB) calcd 277.1035 (M + H), found
277.1046.
1
as a mixture of diastereomers: mp 58-62 °C; H NMR (CDCl3) δ
7.62 (br d s,1H), 6.34 (dd, 1H, J ) 6.2, 7.0 Hz, minor), 6.24 (dd, 1H,
J ) 7,7 Hz, major), 4.31 (m, 1H), 3.77 (m, 1H), 3.64-3.22 (m, 5H),
1.91-1.73 (m, 3H), 1.29 (s, 2H), 1.19 (s, 1H), 0.98 (m, 6H), 0.88 (s,
9H), 0.86 (s, 9H), 0.06 (s, 3H), 0.05 (s, 3H), 0.04 (s, 3H), 0.03 (s, 3H);
IR (film) 3214 (br d), 3084 (br d), 2955, 2927, 2853, 1699, 1469, 1384,
1356, 1215, 1120, 1092 cm-1. Anal. Calcd for C26H52N2O6Si2: C,
57.30; H, 9.64; N, 5.14. Found: C, 57.17; H, 9.48; N, 5.04.
Preparation of 13. Dess-Martin periodinane reagent (650 mg, 0.87
mmol) was added to a solution of 12b (314 mg, 0.58 mmol) in CH2Cl2
(6.5 mL) at 0 °C. The reaction was stirred and allowed to warm to
room temperature over 24 h, at which time the reaction was quenched
with saturated NaHCO3 (20 mL) containing Na2S2O3 (2.5 g). The
aqueous layer was extracted with CH2Cl2 (3 × 100 mL). The combined

1838 J. Am. Chem. Soc., Vol. 119, No. 8, 1997
Greenberg et al.
Preparation of 14. Ketone 7 (60 mg, 0.18 mmol) and 4,4′-
dimethoxytrityl chloride (77 mg, 0.22 mmol) were stirred in pyridine
(3 mL) at 0 °C. After 16 h, the reaction was quenched with MeOH
(0.1 mL), poured into saturated NaHCO3 (20 mL), and extracted with
EtOAc (3 × 20 mL). The organic layers were combined, washed with
brine (25 mL), dried over Na2SO4, and concentrated. Flash chroma-
tography (hexanes/EtOAc/MeOH, 4.5:5.0:0.5) yielded 14 (67 mg,
54%): mp 90-92 °C; 1H NMR (CDCl3) δ 7.41-7.17 (m, 10H), 6.82
(d, 3H, J ) 8.2 Hz), 6.80 (d, 1H, J ) 6.7 Hz), 6.27 (dd, 1H, J ) 6.2
Hz, minor), 6.10 (dd, 1H, J ) 6.6 Hz, major), 4.50-4.38 (m, 1H),
4.05 (d, 1H, J ) 12.7 Hz, major), 3.85 (m,2H, minor), 3.76 (s, 6H),
3.36-3.29 (m, 2H), 3.00 (m, 2H), 2.20-2.05 (m, 2H), 1.54 (s, 3H),
1.25-0.92 (m, 6H); IR (KBr) 3199 (br d), 3077, 2973, 2926, 2832,
Preparation of 11a. Ketone 10a (91 mg, 0.16 mmol) was stirred
in HOAc/H2O (4:1 (v/v), 3 mL). After 24 h, the solvents were removed
in Vacuo. Flash chromatography (CH2Cl2/MeOH, 9:1) yielded 11a (38
mg, 70%) as a hygroscopic solid: 1H NMR (MeOH-d4) δ 7.87 (m,
2H), 7.53 (m, 1H), 7.43 (m, 2H), 6.17 (m, 1H), 4.24 (m, 1H), 4.06 (d,
1H, J ) 12 Hz), 3.73 (m, 1H), 3.66 (m, 1H), 3.56-3.38 (m, 2H), 2.17-
1.81 (m, 2H), 1.62 (s, 3H, major), 1.56 (s, 3H, minor); IR (film) 3188,
3056, 2924, 1682, 1475, 1447, 1432, 1385, 1357, 1235, 1089, 1042
cm-1 13C NMR (MeOH-d4) δ 202.1 (minor), 201.6 (major), 176.3
;
(minor), 175.8 (major), 157.1 (major), 156.7 (minor), 139.6 (minor),
139.5 (major), 136.8 (major), 136.5 (minor), 132.4, 132.2 (major), 132.1
(minor), 90.0 (minor), 89.9 (major), 88.1 (minor), 87.5 (major), 75.2
(minor), 74.9 (major), 65.9, 58.5 (major), 58.4 (minor), 49.5 (minor),
49.2 (major), 40.1 (minor), 40.0 (major), 22.0 (major), 21.9 (minor);
HRMS (FAB) calcd 349.1399 (M + H), found 349.1416.
Preparation of 23. A solution of ketone 11a (20 mg, 57 µmol) in
cyclohexa-1,4-diene (200 mM) in CH3CN/H2O (3:2; 4 mL) was freeze-
pump-thaw-degassed three times and irradiated for 20 h in the
Rayonet photoreactor. The solvents were removed in Vacuo. Flash
chromatography (MeOH/CH2Cl2; 3:97) yielded a mixture (11 mg, 65%)
of four diastereomers of 23 as an oil: 1H NMR (MeOH-d4) δ 7.4-7.3
(m, 5H), 6.35-6.05 (m, 1H), 5.03 (m, 1H), 4.24 (m, 1H), 3.9-3.32
(m, 4H), 3.10-3.00 (m, 1H), 2.35-1.95 (m, 2H), 1.28 (s, 3H, major),
1.15 (s, 3H, minor), 1.03 (s, 3H); HRMS (FAB) calcd 351.1556 (M +
H), found 351.1571.
1701, 1602, 1504, 1475, 1442, 1377, 1297, 1245, 1174, 1029 cm-1
.
Anal. Calcd for C35H40N2O8: C, 68.17; H, 6.54; N, 4.54. Found: C,
67.97; H, 6.44; N, 4.39.
Separation of (5R)- and (5S)-14. The diastereomers of 14 were
prepared as described above and subsequently separated by flash
chromatography (hexanes/EtOAc/MeOH, 4.8:5.0:0.2). (5S)-14: Rf )
1
0.35; H NMR (CDCl3) δ 7.76 (s, 1H), 7.42-7.14 (m, 10H), 6.83-
6.76 (m, 4H), 6.27 (dd, 1H, J ) 7, 7 Hz), 4.48 (m, 1H), 3.87-3.83
(m, 2H), 3.76 (s, 6H), 3.30 (m, 1H), 3.09 (m, 1H), 3.02 (d, 1H, J )
12.3 Hz), 2.56 (m, 1H), 2.05-1.97 (m, 1H), 1.92 (s, 1H), 1.15 (s, 3H),
1.09 (d, 3H, J ) 6.6 Hz), 0.92 (d, 3H, J ) 6.7 Hz). (5R)-14: Rf )
1
0.28; H NMR (CDCl3) δ 7.40-7.19 (m, 11H), 6.83-6.80 (m, 4H),
6.10 (dd, 1H, J ) 6.8, 6.8 Hz), 4.41 (m, 1H), 4.05 (d, 1H, J ) 12.7
Hz), 3.87 (m, 1H), 3.77 (s, 6H), 3.36 (m, 1H), 3.03-2.94 (m, 2H),
2.56 (m, 1H), 2.17-2.08 (m, 1H), 1.85 (s, 1H), 1.25 (s, 3H), 1.02 (d,
3H, J ) 6.6 Hz), 0.91 (d, 3H, J ) 6.7 Hz).
Preparation of 10b. The dianion of 9 (500 mg, 1.05 mmol) was
prepared as described above for the preparation of 12b, and acylated
with p-anisoyl chloride (223 mg, 1.31 mmol). After the reaction was
worked up, as described for 12b, flash chromatography (CH2Cl2/EtOAc,
9:1) afforded 421 mg (70%) of 10b, as a 3:1 mixture of diastereo-
mers: 1H NMR (CDCl3) δ 7.93 (d, 2H, J ) 8.7 Hz, major), 7.86 (d,
2H, J ) 8.7 Hz, minor), 7.48 (br d s, 1H, minor), 7.39 (br d s, 1H,
major), 6.88 (d, 2H, J ) 8.7 Hz, minor), 6.87 (d, 2H, J ) 8.7 Hz,
major), 6.25 (t, 1H, J ) 6.2 Hz), 4.42-4.25 (m, 1H), 4.12 (d, 1H, J )
13.3 Hz, minor), 3.97 (d, 1H, J ) 12.9 Hz, major), 3.84 (s, 3H), 3.80-
3.55 (m, 3 H), 3.25 (d, 1H, J ) 12.9 Hz, major), 3.16 (d, 1H, J ) 13.3
Hz, minor), 2.35-1.85 (m, 2H), 1.60 (s, 3H, major), 1.57 (s, 3H, minor),
0.88 (s, 9H), 0.87 (s, 9H), 0.10-0.08 (m, 12H); IR (thin film) 3202,
3074, 2954, 2929, 2898, 2857, 1711, 1601, 1575, 1511, 1472, 1384,
Preparation of Phosphoramidite 25. 2-Cyanoethyl N,N-diisopro-
pylchlorophosphoramidite (35 mg, 0.15 mmol) was added to a stirred
solution of 14 (74 mg, 0.12 mmol) and Hu¨nig’s base (23 mg, 0.18
mmol) in CH2Cl2 (3 mL) at 0 °C. After 30 min, the reaction was poured
into saturated NaHCO3 (20 mL) and extracted with EtOAc (2 × 15
mL). The combined organic layers were washed with brine (10 mL),
dried over Na2SO4, and concentrated. Flash chromatography (CH2-
Cl2EtOAc, 3:1) yielded 5 (50 mg, 51%) as a foam: 1H NMR (CDCl3)
δ 7.42-7.14 (m, 10H), 6.83-6.76 (m, 4H), 6.27 (m, 1H, major), 6.12
(m, 1H, minor), 4.60 (m, 1H), 3.75 (s, 6H), 3.64-3.48 (m, 4H), 3.36-
3.02 (m, 4H), 2.58 (m, 2H), 2.39 (m, 2H), 2.25-2.02 (m, 2H), 1.30-
0.91 (m, 21H); 31P NMR (CDCl3) δ 149.17 (s), 149.08 (s), 149.02 (s).
Phosphoramidites consisting of a single stereoisomer at C5 were
prepared as described above using single diastereomers of the dimethox-
ytritylated ketones 14.
Preparation of 12a. The dianion of 9 (100 mg, 0.211 mmol) was
prepared and acylated with benzaldehyde (28 mg, 0.26 mmol) using
the procedure described for the preparation of 12b. Flash chromatog-
raphy (CH2Cl2/EtOAc, 9:1) afforded 85 mg (70%) of 12a as an oil:
1H NMR (CDCl3) δ 7.36 (m, 5H), 6.29 (dd, 1H, J ) 6, 8 Hz, minor),
6.16 (dd, 1H, J ) 5.5, 8.5 Hz, major), 5.00 (s, 1H, major), 4.94 (s, 1H,
minor), 4.33 (m, 1H, minor), 4.13 (m, 1H, major), 3.64 (m, 1H), 3.25
(dd, 1H, J ) 4.5, 10.5 Hz), 3.15 (d, 1H, J ) 12 Hz), 3.05 (dd, 1H, J
) 7, 10. 5 Hz), 2.72 (d, 1H, J ) 12 Hz, minor), 2.64 (d, 1H, J ) 12
Hz), 1.84 (m, 2H), 1.30 (s, 3H, major), 1.18 (s, 3H, minor), 0.89-0.86
(m, 18H), 0.08-0.02 (m, 12H); IR (thin film) 3216 (br d), 3064 (br
d), 2953, 2929, 2885, 2857, 1704, 1472, 1387, 1361, 1252, 1222, 1091,
1057, 1029 cm-1; HRMS (FAB) calcd 579.3286 (M + H), found
579.3264.
Preparation of 10a. Alcohol 12a (85 mg, 0.14 mmol) was oxidized
with Dess-Martin periodinane reagent (140 mg, 0.29 mmol) as
previously described for the preparation of 13. Flash chromatography
(CH2Cl2/EtOAc; 9:1) afforded 79 mg (94%) of 10a as an oil: 1H NMR
(CDCl3) δ 8.88-7.79 (m, 2H), 7.60-7.58 (m, 1H), 7.55-7.37 (m, 2H),
6.25 (dd, 1 H, J ) 8, 8 Hz), 4.37-4.34 (m, 1H, major), 4.31-4.27 (m,
1H, minor), 4.15 (d, 1H, J ) 14.2 Hz, minor), 3.97 (d, 1H, J ) 13 Hz,
major), 3.76-3.60 (m, 3H), 3.28 (d, 1H, J ) 13 Hz, major), 3.17 (d,
1H, J ) 14.2 Hz, minor), 2.19-1.86 (m, 2H), 1.62 (s, 3H, major),
1.58 (s, 3H, minor), 0.87 (m, 18H), 0.07 (s, 3H), 0.06 (s, 3H), 0.05 (s,
3H), 0.04 (s, 3H); IR (thin film) 3200 (br d), 3140 (br d), 2954, 2929,
2857, 1709, 1687, 1472, 1383, 1360, 1252, 1095 cm-1. Anal. Calcd
for C29H48N2O6Si2: C, 60.37; H, 8.40; N, 4.85. Found: C, 60.15; H,
8.18; N, 4.70.
1360, 1253, 1174, 1096, 1029, 971 cm-1
.
Prepartion of 11b. Ketone 10b (100 mg, 0.165 mmol) was stirred
in HOAc/H2O (4:1 (v/v), 3 mL) for 24 h. The solvents were removed
in Vacuo. Flash chromatography (CH2Cl2/MeOH, 9:1) yielded 11b (41
mg, 66%) as a hygroscopic solid: 1H NMR (CDCl3) δ 8.01-7.90 (m,
3H), 6.86 (d, 2H, J ) 8 Hz), 6.18 (t, 1H, J ) 7 Hz, major), 6.04 (t,
1H, J ) 7 Hz, minor), 4.52 (m, 1H, minor), 4.42 (m, 1H, major), 4.18
(d, 1H, J ) 14 Hz, minor), 4.03 (d, 1H, J ) 14 Hz, major), 3.81 (s,
3H), 3.80-3.65 (m, 4H), 3.28 (d, 1H, J ) 14 Hz, minor), 3.24 (d, 1H,
J ) 14 Hz, major), 2.54-2.12 (m, 3H), 1.58 (s, 3H, major), 1.56 (s,
3H, minor); 13C NMR (MeOH-d4) δ 197.0 (minor), 196.5 (major), 173.8
(minor), 173.7 (major), 165.2 (major), 165.1 (minor), 154.6 (major),
154.2 (minor), 132.8, 129.0, 114.9 (major), 114.8 (minor), 88.6 (minor),
88.3 (major), 87.5 (minor), 87.3 (major), 86.2 (minor), 85.8 (major),
85.6 (minor), 85.0 (major), 72.6 (minor), 72.3 (major), 63.3, 56.1
(major), 55.6 (minor), 48.4 (minor), 48.1 (major), 37.5, 19.7 (minor),
19.5 (major); IR (film) 3336, 2932, 2829, 1696, 1596, 1568, 1474,
1455, 1441, 1385, 1347, 1309, 1248, 1173, 1088, 1032 cm-1
.
Preparation of 19. 2′-Deoxyuridine was hydrogenated as described
above for the preparation of 9. The dihydronucleoside (500 mg, 2.14
mmol) was dried by repeated azeotroping from anhydrous pyridine (2
mL). TBSCl (825 mg, 5.47 mmol) was dissolved in pyridine (5 mL)
and added to the nucleoside. The reaction was stirred for 48 h at 40
°C, quenched with MeOH (0.1 mL), and poured into H2O (50 mL).
The aqueous layers were extracted with ether (3 × 50 mL). The organic
layers were combined, washed with brine (25 mL), and dried over
MgSO4. Flash chromatography (CH2Cl2/EtOAc, 12:1) yielded 875 mg
1
(88%) of 19: mp 81-85 °C; H NMR (CDCl3) 7.33 (br d, 1H), 6.27
(dd, 1H, J ) 7, 7 Hz), 4.34 (m, 1H), 3.77 (m, 1H), 3.71-3.60 (m,
3H), 3.29 (m, 1H), 2.58 (m, 2H), 1.96 (dd, 2H, J ) 4.6, 7 Hz), 0.88 (s,
9H), 0.87 (s, 9H), 0.06 (s, 6H), 0.05 (s, 6H); IR (film) 3215 (br d),
3086 (br d), 2954, 2928, 2888, 2857, 1713, 1472, 1462, 1337, 1283,
1254, 1218, 1192, 1103, 1032, 1005 cm-1
.

DNA Damage Induced Via 5,6-Dihydrothymid-5-yl
J. Am. Chem. Soc., Vol. 119, No. 8, 1997 1839
Preparation of 20. A cyclohexane solution of sec-BuLi (3.0 mL,
1.2 M) was added to 19 (654 mg, 1.43 mmol) and DMPU (549 mg,
4.29 mmol) in THF (5 mL) at -78 °C. After stirring for 1 h, the
reaction was warmed to -20 °C and 2-iodopropane (364 mg, 1.56
mmol) was added. After 2 h at -20 °C, the reaction was quenched
with saturated NH4Cl (0.5 mL) and poured into H2O (30 mL). The
aqueous layers were extracted with ether (3 × 50 mL). The combined
organic layers were washed with brine (25 mL) and dried over MgSO4.
Flash chromatography (CH2Cl2/EtOAc, 12:1) yielded 380 mg (55%)
3 Hz, minor), 4.29 (m, 1H, major), 4.01 (m, 1H, minor), 3.78 (m, 1H),
3.66-3.52 (m, 3H), 3.41 (d, 1H, J ) 13 Hz, minor), 3.26 (d, 1H, J )
13 Hz, minor), 3.08 (d, 1H, J ) 13 Hz, major), 2.20-1.88 (m, 3H),
1.10 (s, 3H, minor), 1.05 (s, 3H, major), 0.99-0.90 (m, 6H); 13C NMR
(MeOH-d4) δ 177.5, 154.5, 87.5 (major), 87.4 (minor), 85.4 (minor),
85.2 (major), 72.9 (major), 72.6 (minor), 63.5 (major), 63.4 (minor),
49.9, 45.3, 45.0 (major), 44.6 (minor), 37.8 (minor), 37.4 (major), 31.5
(minor), 31.0 (major), 17.8 (minor), 17.6 (major), 17.4 (minor), 17.2
(major); IR (KBr) 3385 (br d), 2968, 2879, 1698, 1487, 1439, 1395,
1374, 1226, 1092, 1050 cm-1; HRMS (FAB) calcd 287.1606 (M +
H), found 287.1606.
1
of 20: mp 124-126 °C; H NMR (CDCl3) δ 7.29 (br d, 1H), 6.26
(dd, 1H, J ) 6, 6 Hz), 4.32 (m, 1H) 3.80 (m, 1H), 3.77-3.54 (m, 3H),
3.17 (m, 1H), 2.40-2.11 (m, 2H), 1.93 (m, 2H), 1.02 (m, 6H), 0.87 (s,
9H), 0.85 (s, 9H), 0.06 (s, 3H), 0.05 (s, 3H), 0.04 (s, 3H), 0.03 (s, 3H);
IR (film) 3120 (br d), 2970, 2923, 2891, 2852, 1713, 1694, 1682, 1473,
1454, 1371, 1253, 1096 cm-1. Anal. Calcd for C25H50N2O5Si2: C,
57.55; H, 9.68; N, 5.59. Found: C, 57.71; H, 9.74; N, 5.54.
Acknowledgment. Financial support of this work from the
National Institutes of Health (Grant GM-46534) and the National
Science Foundation (Grant CHE-9424040) and seed funding
from the Elsa U. Pardee Foundation are appreciated. M.R.B.,
G.P.C., and B.K.G. thank the U.S. Department of Education
for partial support under the Graduate Assistance in Areas of
National Need Program (Grant No. P200A10210). M.M.G. is
an Alfred P. Sloan Research Fellow. We are grateful to
Professor Oren Anderson and Ms. Susie Miller for solving the
crystal structure of 7, as well as to the NSF for the purchase
and upgrade of the X-ray diffraction system. Mass spectra were
obtained on instruments supported by the National Institutes of
Health Shared Instrumentation Grant GM-49631.
Preparation of 21. A cyclohexane solution of sec-BuLi (1.9 mL,
1.2 M) was added to 20 (380 mg, 0.76 mmol) and DMPU (293 mg,
2.28 mmol) in THF (3 mL) at -78 °C. After stirring for 1 h,
iodomethane (215 mg, 0.915 mmol) was added and the reaction was
warmed to 25 °C. After 2 h at 25 °C, the reaction was quenched with
saturated NH4Cl (0.5 mL) and poured into H2O (30 mL). The aqueous
layers were extracted with ether (3 × 50 mL). The combined organic
layers were washed with brine (25 mL) and dried over MgSO4. Flash
chromatography (CH2Cl2/EtOAc, 11:1) yielded 277 mg (71%) of 21
as an oil: 1H NMR (CDCl3) δ 7.31 (br d, 1H), 6.29 (m, 1H), 4.32 (m,
1H), 3.85-3.54 (m, 4H), 3.28-2.85 (m, 2H), 2.20-1.63 (m, 3H), 1.25-
0.70 (m, 27H), 0.30-0.06 (m, 12H); IR (film) 3189 (br d), 2957, 2923,
2889, 2845, 1705, 1468, 1438, 1387, 1249, 1219, 1090, 1026 cm-1
.
Supporting Information Available: Autoradiograms of
enzymatic end group analysis of 31, a sample KIE experiment,
the results of piperidine treatment of cleavage products from
[3′-32P]29 at T11, the X-ray crystal structure data of 7, and data
analysis of phosphorimager results (17 pages). See any current
masthead page for ordering and Internet access instructions.
Anal. Calcd for C26H52N2O5Si2: C, 58.31; H, 9.80; N, 5.44. Found:
C, 58.24; H, 9.62; N, 5.29.
Preparation of 18. Protected dihydro derivative 21 (220 mg, 0.43
mmol) was stirred in HOAc/H2O (4:1 (v/v), 3 mL) for 24 h, after which
the solvents were removed in Vacuo. Flash chromatography (CH2Cl2/
MeOH, 9:1) yielded 18 (67 mg, 55%) as a hygroscopic solid: 1H NMR
(MeOH-d4) δ 6.26 (dt, 1H, J ) 9, 3 Hz, major), 6.21 (dt, 1H, J ) 8,
JA9533510