Selective one-electron oxidation of duplex DNA oligomers: Reaction at thymines

Article abstract of DOI:10.1039/b717437c

The one-electron oxidation of duplex DNA generates a nucleobase radical cation (electron hole ) that migrates long distances by a hopping mechanism. The radical cation reacts irreversibly with H₂O or O ₂ to form oxidation products (damaged bases). In normal DNA (containing the four common DNA bases), reaction occurs most frequently at guanine. However, in DNA duplexes that do not contain guanine (i.e., those comprised exclusively of A/T base pairs), we discovered that reaction occurs primarily at thymine and gives products resulting from oxidation of the T-C5 methyl group and from addition to its C5-C6 double bond. This surprising result shows that it is the relative reactivity, not the stability, of a nucleobase radical cation that determines the nature of the products formed from oxidation of DNA. A mechanism for reaction is proposed whereby a thymine radical cation may either lose a proton from its methyl group or H₂O/O₂ may add across its double bond. In the latter case, addition may initiate a tandem reaction that converts both thymines of a TT step to oxidation products. The Royal Society of Chemistry.

Full text of DOI:10.1039/b717437c

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Selective one-electron oxidation of duplex DNA oligomers: reaction at
thymines†
Avik Ghosh,a Abraham Joy,a Gary B. Schuster,*a Thierry Doukib and Jean Cadet*b
Received 12th November 2007, Accepted 3rd January 2008
First published as an Advance Article on the web 30th January 2008
DOI: 10.1039/b717437c
The one-electron oxidation of duplex DNA generates a nucleobase radical cation (electron “hole”) that
migrates long distances by a hopping mechanism. The radical cation reacts irreversibly with H2O or O2
to form oxidation products (damaged bases). In normal DNA (containing the four common DNA
bases), reaction occurs most frequently at guanine. However, in DNA duplexes that do not contain
guanine (i.e., those comprised exclusively of A/T base pairs), we discovered that reaction occurs
primarily at thymine and gives products resulting from oxidation of the T-C5 methyl group and from
addition to its C5–C6 double bond. This surprising result shows that it is the relative reactivity, not the
stability, of a nucleobase radical cation that determines the nature of the products formed from
oxidation of DNA. A mechanism for reaction is proposed whereby a thymine radical cation may either
lose a proton from its methyl group or H2O/O2 may add across its double bond. In the latter case,
addition may initiate a tandem reaction that converts both thymines of a TT step to oxidation products.
The migration and reactions of radical cations in duplex DNA
Introduction
oligomers that contain guanine have been studied thoroughly.1,25–29
The one-electron oxidation of DNA introduces a radical cation
However, the consequence of one-electron oxidation of oligomers
(“hole”) into its stacked nucleobases that results in chemical
that do not contain guanine nucleobases had not been examined
until the report of our recent experiments.30 Previously, there
were scattered reports indicating that reactions of radical cations
in DNA do occur at bases other than guanine.31 For example,
in oligonucleotides that contain both guanine and adenine, the
oxidation product 8-oxo-7,8-dihydroadenine (8-oxoAde) is found
in low yield in comparison with the guanine oxidation product 8-
oxo-7,8-dihydroguanine (8-oxoGua).32,33 This result is consistent
with the idea that relative oxidation potential determines the
reaction site for radical cations in DNA because the Eox of adenine
is somewhat greater than that of guanine. The pyrimidines, T
and C, are much more difficult to oxidize than are the purine
bases.34 Indeed thymine, which has an Eox of ca. 2.1 V vs. NHE,35
is the nucleobase that is most difficult to oxidize. However,
reactions at thymines are observed when a menadione (2-methyl-
1,4-naphthoquinone) group is linked covalently at an internal
position of DNA.36
We recently reported the results of a preliminary study of
the one-electron oxidation of duplex DNA oligomers that do
not contain guanines.30 Surprisingly, reaction occurs primarily at
thymines; no reaction could be detected at adenines. This was
attributed37 to the difference in reactivity between adenine and
thymine radical cations. Specifically, we found by replacement of
T with uracil that the C5-methyl group of thymine is necessary for
the oxidation reactions and strand cleavage to occur.
reactions (“damage”) that may lead to mutations. In recent years,
it has been shown that the reactions of radical cations that
damage DNA need not occur at the site of the initial oxidation.
In duplex DNA, radical cations may migrate long distances
˚
(hundreds of Angstroms) by a reversible hopping process before
being trapped irreversibly by reaction with H2O or O2.1–10 This
process can produce a multiplicity of chemical modifications to
the DNA that mostly consist of base lesions,11–13 and it has been
implicated in carcinogenesis, other diseases14–16 and in aging.17
Apart from its biological relevance, radical cation migration
in DNA is of interest because of its potential application to
molecular electronic devices.18–21
One of the more interesting features of the one-electron oxida-
tion of duplex DNA is that the resulting reaction, which is typically
detected as strand cleavage following chemical or enzymatic
treatment of the damaged DNA,12 commonly occurs at Gn (n =
1–3) sites. It had been generally accepted22 that reaction occurs
at guanines primarily because they are the nucleobases having
the lowest oxidation potential (Eox).23,24 Thus, a migrating radical
cation pauses briefly at a guanine or a multi-guanine site, and this
facilitates trapping by the irreversible reaction with H2O or O2.
aSchool of Chemistry and Biochemistry, Georgia Institute of Technology,
Atlanta, GA, 30332, USA. E-mail: gary.schuster@cos.gatech.edu
bLaboratoire “Le´sions des Acides Nucle´iques”, SCIB-UMR E3 (CEA/
UJF), Institut Nanosciences et Cryoge´nie, CEA/Grenoble, F-38054, Greno-
ble cedex 9, France. E-mail: jean.cadet@cea.fr
Here we report a detailed examination of the one-electron
oxidation of duplex DNA oligomers that do not contain guanines.
These studies include assessment of the distance-dependence of
radical cation migration in duplexes comprised of only A/T
base pairs, the effect of a radical scavenger on the reactions of
thymine, and the role played by the thymine C5-methyl group.
The products of these reactions were identified by means of
† Electronic supplementary information (ESI) available: Melting tempera-
ture data and circular dichroism data for DNA(1–10); histogram indicating
the amount of damage at each TTT step relative to the total damage at all
TTT steps for 15 min of irradiation and 20 min of irradiation of DNA(1);
analysis of the effect of irradiation time on the reaction and strand cleavage
of DNA containing only A/T base pairs. See DOI: 10.1039/b717437c
916 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
sophisticated HPLC–MS/MS analysis under conditions of low
nucleobase conversion. These experiments indicate that there is a
complex mechanism for the reaction of thymine radical cations
in duplex DNA that features a competition between proton loss
from the C5 methyl group and the addition of H2O/O2 across its
C5–C6 double bond, with the former process leading to a tandem
reaction involving an adjacent nucleobase.38,39
ation. After irradiation, the samples were precipitated once with
cold ethanol (100 lL) and 2 lL of glycogen. The precipitated
samples were washed twice with 100 lL of 80% ethanol and dried.
Then dry DNA oligomers were dissolved in 14 lL of water, and
2 lL of sodium phosphate buffer (100 mM), 2 lL of NaCl (1 M)
and 2 lL of Na2IrCl6 (100 lM) were added. After 60 min of
◦
reaction at 37 C, 2 lL of HEPES (20 mM) and 2 lL of EDTA
(100 mM) were added to quench the reaction. The DNA was
precipitated from cold ethanol and dried. For piperidine chemical
cleavage analysis, the samples were mixed with 50 lL of piperidine
Materials and methods
Chemicals were purchased from either Fisher Scientific or Sigma
Aldrich (St Louis, MO). T4 polynucleotide kinase (PNK) enzyme
and [c-32P]-ATP were purchased from GE Healthcare. E. coli
endonuclease III (Endo III) enzyme was purchased from Trevigen
Inc. (Gaithersburg, MD). DNA oligomers were synthesized as
described elsewhere39 on an Expedite 8909 DNA synthesizer. Nu-
cleoside phosphoramidites were obtained from Glen Research and
were used as received. The extinction coefficients of the oligomers
were calculated using a biopolymer properties calculator, and
their concentrations were determined from the absorbance at
260 nm. An adenine is substituted for the anthraquinone group
in the extinction coefficient calculation. The oligonucleotides
were purified by means of reversed phase HPLC on a Hitachi
preparative HPLC system using a Dynamax octadecylsilyl silica
gel column. Purified oligomers were desalted and characterized by
mass spectroscopy. UV melting and cooling curves were recorded
on a Cary 1E spectrophotometer equipped with a multicell
block, temperature controller, and sample transport accessory. CD
spectra were recorded on a JASCO J-720 spectropolarimeter.
◦
(1 M) and heated at 90 C for 30 min. After evaporation of the
piperidine (Speedvac, high heat) and lyophilization twice with
20 lL of water, the samples were dissolved in denaturing loading
dye and subjected to 20% 19 : 1 polyacrylamide gel electrophoresis.
For EcoRIII enzymatic cleavage analysis, the dried samples were
mixed with 12 lL of the enzyme and 8◦lL of buffer solution, heated
◦
first at 37 C for 2 h and then at 90 C for 20 min. The samples
were then reprecipitated with cold ethanol (100 lL) and 2 lL of
glycogen and the precipitated samples were washed with 100 lL of
80% ethanol. After evaporation of the ethanol, the samples were
dissolved in denaturing loading dye and subjected to 20% 19:1
polyacrylamide gel electrophoresis. The gels were dried, and the
cleavage sites were visualized by autoradiography. Quantification
of cleavage bands was performed on a Fuji phosphorimager.
Enzymatic digestion and HPLC–MS/MS analysis
Oligonucleotides, either untreated or exposed to light, were enzy-
matically digested by incubation (2 h, 37 ◦C) at pH 6 in the presence
of phosphodiesterase II and nuclease P1 (Sigma, St Louis, MO).
The pH was then adjusted to 8 by addition of Tris buffer. Treatment
(2 h, 37 ◦C) by phosphodiesterase I and alkaline phosphatase
(Sigma, St Louis, MO) yielded normal and oxidized nucleosides.
The resulting mixture was separated by high performance liquid
chromatography on a Agilent Series 1100 system equipped with
an Uptisphere ODB octadecylsilyl silica gel column (2 × 150 mm
I.D., 3 lm particle size; Montluc¸on, France). The mobile phase
was a gradient of acetonitrile (0 to 20%) in 2 mM ammonium
formate (pH 6.5). Elution of normal nucleosides was monitored
on-line by a UV spectrometer set at 280 nm, while oxidized
nucleosides were detected by a triple quadrupolar mass spec-
trometer (API 3000, Sciex/Perkin Elmer, Thornhill, Canada) used
in the multiple reaction monitoring mode. Negative electrospray
ionization was used for the quantification of the four cis and trans
diastereomers of 5,6-dihydroxy-5,6-dihydrothymidine (ThGly), 5-
(hydroxymethyl)-2ꢀ-deoxyuridine (5-HMdUrd) and 5-formyl-2ꢀ-
deoxyuridine (5-FormdUrd). The respective retention times were
3.7, 4.4, 7.9, 8.1, 15.2 and 19.9 min, respectively. 8-Oxo-7,8-
dihydro-2ꢀ-deoxyadenosine (8-oxodAdo) (retention time 25 min)
was detected in the positive ionization mode. Quantification was
performed by external calibration.
Preparation of radiolabeled DNA
The oligomers were radiolabeled at the 5ꢀ-end using [c-32P]-ATP
and PNK enzyme. A 5 lL sample of desired single stranded DNA
was incubated with 1 lL of [c -32P]-ATP and 2 lL of PNK enzyme
in a total volume of 20 lL at 37 ◦C for 45 min. After incubation,
the DNA sample was suspended in denaturing loading buffer and
was purified on a 20% denaturing polyacrylamide gel. The desired
DNA band was excised from the gel and eluted with 800 lL
of elution buffer (0.5 M NH4OAc, 10 mM Mg(OAc)2/1.0 mM
EDTA/0.1% SDS) at 37 ◦C for 12 h. The DNA was precipitated
from the supernatant by addition of 600 lL of cold ethanol and
2 lL of glycogen solution. The mixture was vortexed, placed on
dry ice for ca. 60 min, and centrifuged at 13 000 rpm for 45 min.
The supernatant was removed, and the residual DNA was washed
with 100 lL of 80% ethanol and air-dried. Suitable volumes of
water were added for further experimentation.
UVA irradiation and cleavage analysis
Samples for irradiation were prepared by hybridizing a mixture of
unlabeled (5.0 lM) and radiolabeled (10 000 cpm) oligonucleotides
with complementary AQ-linked DNA in sodium phosphate buffer
solution (10 mM) and MgCl2 (2 mM) at pH 7.0. Hybridization was
achieved by heating the samples at 90 ◦C for 10 min, followed by
slow cooling to room temperature for 3 h. Samples were irradiated
Determination of quantum yield
The light flux of the Rayonet photoreactor was determined by
using sodium 9,10-anthraquinone-2,6-disulfonate (AQDS(2,6))
actinometry at pH 14.40 The actinometer solutions had an optical
density at 330 nm of 0.1 and were degassed by the freeze–pump–
thaw technique at high vacuum. The extent of reaction of the
actinometer was monitored by UV spectroscopy at various time
◦
at ca. 30 C in microcentrifuge tubes in a Rayonet Photoreactor
(Southern New England Ultraviolet Co., Bransford, CT) equipped
with eight 350 nm lamps. To investigate the effect of glutathione,
different concentrations (0.05 mM, 0.5 mM, 1.0 mM and 5.0 mM)
of glutathione was added to hybridized samples, prior to irradi-
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 916–928 | 917
©

View Article Online
intervals. Conversion was kept below 50%, where the extent of
reaction was linear with irradiation time. The slope of a plot
of extent reaction vs. irradiation time yielded a light flux =
excited state (AQ*3). Significantly, calculations from the Rehm–
Weller42 equation show that AQ*3 has sufficient oxidizing power to
convert either A or T to their radical cations with the concomitant
formation of the AQ radical anion. Although the radical cation
is formed initially at the base pair adjacent to the AQ group, it
is expected that it will migrate through the oligomer before being
trapped by an irreversible reaction. Thus, several of the duplex
oligomers examined contain regularly repeating base sequence
patterns that allow the distance dependence of the reaction
probability to be assessed. For this reason, the irradiation reactions
are carried out to low conversion, “single-hit conditions”, where,
on average, each DNA molecule reacts at best once or not at all.41
Under these conditions, the amount of strand cleavage observed
is directly proportional to the probability of reaction at that site.
After irradiation, the samples were either (a) treated with chemical
or enzymatic reagents that cause strand cleavage to occur at
damaged bases, which enables the identification and quantification
of the radical cation reaction sites, or (b) the UVA-irradiated
oligomers were treated with nuclease P1, phosphodiesterases
and alkaline phosphatase, which digest the oligomer and enable
identification of the products formed from the reactions of the
nucleobase radical cations.
(1.3
0.2) × 10−7 Einstein min−1. DNA(4) (Fig. 1) (10 lM in
air-saturated phosphate buffer solution) was irradiated in the
calibrated Rayonet photoreactor at ca. 30 ◦C. Aliquots were
withdrawn and treated with piperidine at various time intervals.
After evaporation of the piperidine, the samples were suspended in
water. 2ꢀ-Deoxycytidine, for use as an internal standard, was added
to each sample, and the volume was adjusted to 500 lL. These
mixtures were analyzed by reverse phase HPLC on a Hitachi
preparative HPLC system using a Dynamax C18 column. The
extent of reaction of the DNA cleavage was monitored by the
decrease in signal of the TT-containing strand and found to be
linear with irradiation time.
Radical cation reaction at thymine
The first oligomer we examined is DNA(1), which is comprised
exclusively of A/T base pairs. DNA(1) was selected assuming that
reaction of the radical cation would occur at adenine, because
that base has much lower Eox than thymine. The results of
irradiation of DNA(1), its subsequent treatment with piperidine,
and then PAGE analysis revealed that significant strand cleavage
occurred only at thymine bases, which was an unexpected result.
However, the absence of significant strand cleavage at adenine
under these conditions does not demonstrate conclusively that
reaction of the radical cation does not occur at these bases in
DNA(1). This is because the major product expected from the
reaction of an adenine radical cation in DNA, namely 8-oxo-
7,8-dihydroadenine (8-oxoAde), does not readily result in strand
cleavage when treated with piperidine.43 To resolve this issue, the
irradiated samples were treated with Na2IrCl6 before their reaction
with piperidine. It has been shown that the oxidation of 8-oxoAde
by Na2IrCl6 gives products that do result in strand cleavage on
piperidine treatment.44 The results of these experiments, shown
in Fig. 2, reveal that reaction of the radical cation introduced by
UVA irradiation of the covalently linked AQ occurs primarily at
thymine, not at adenine; a result confirmed by analysis of the
reaction products (see below). More specifically, the thymines
closer to the AQ react more often, which reveals that there is a
distance dependence to the reaction efficiency. Furthermore, the
central and 5ꢀ-T of the TTT sequences are more reactive than the
3ꢀ-T, which shows that the site of the radical cation reaction is
governed by subtle electronic or steric factors. This point will be
addressed in greater detail below.
Fig. 1 Structures of DNA oligomers used in the study of long distance
charge migration through adenine and thymine containing DNA duplexes.
* = [32P]-radiolabel.
Results
Charge migration and reaction in A/T-containing duplex DNA
We prepared the series of DNA oligonucleotides shown in Fig. 1
to probe the results of their one-electron oxidation reactions.
Each of the duplexes is fully complementary and exhibits the
expected melting and spectroscopic properties.41 These duplex
oligomers contain a covalently linked anthraquinone group (AQ)
photosensitizer14 and a [32P]-radiolabel (indicated by “*” in Fig. 1),
to permit analysis of strand cleavage by PAGE and phospho-
rimagery. Irradiation of the AQ group at 350 nm, where DNA
does not absorb significantly, results in its electronic excitation
and subsequent rapid intersystem crossing to form the triplet
Quenching of the thymine reaction by GG steps
The surprising finding that the one-electron oxidation of DNA(1)
results primarily in reaction at thymine was probed further by
investigation of DNA(2), which is identical to DNA(1) except that
918 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Replacement of TTT by UUU
The surprising result that the one-electron oxidation of DNA(1)
causes reaction at thymine was probed further by the investigation
of DNA(3), which is identical to DNA(1) except that every other
TTT sequence is replaced by UUU. Of course, the difference
between thymine and uracil is that the latter lacks a methyl
group at its C5-position. It has been previously reported that the
thymidine radical cation in aqueous solution follows two paths
to the formation of the three sets of oxidation products that
are shown in Scheme 1.32,38 Apparently, the thymidine radical
cation can lose a proton from its C5-methyl group to form a
radical that is trapped by O2 and eventually generates, through
the intermediacy of 5-(hydroperoxymethyl)-2ꢀ-deoxyuridine, 5-
(hydroxymethyl)-2ꢀ-deoxyuridine (5-HMdUrd) and 5-formyl-2ꢀ-
deoxyuridine (5-FormdUrd). Alternatively, the thymidine radical
cation can be attacked in two subsequent steps by H2O and O2,
which results eventually in formation of the four cis and trans
diastereomers of 5,6-dihydroxy-5,6-dihydrothymidine (c- and t-
ThdGly).
The results of irradiation of DNA(3) are also shown in Fig. 2.
Just as in the case of DNA(1), there is no detectable reaction at any
nucleobase other than thymine. In particular, there is no significant
strand cleavage at the uracils in the UUU segments. This finding
suggests that the thymine methyl group plays an important role in
the reaction of the radical cation in DNA. We carried out control
experiments to confirm this view.
The most common oxidation product of uracil is 5,6-dihydroxy-
5,6-dihydrouracil (uracil glycol).45 Although the reaction rates of
KMnO4 with T and U to produce thymine glycol and uracil glycol,
respectively, are comparable,46 we noticed in sequencing lanes that
piperidine-induced strand cleavage is less efficient for U than for
T. This suggested the possibility that a one-electron oxidation of
uracil in DNA could be yielding uracil glycol that is not revealed
as strand cleavage by the reaction with Na2IrCl6 and piperidine.
We explored this possibility by replacing the piperidine treatment
of the irradiated DNA by reaction with E. coli endonuclease III
(Endo III), which is known to cause strand cleavage at uracil
glycol.47 The results are unchanged—the damage pattern revealed
by Endo III is the same as the one revealed by piperidine. Clearly,
the one electron oxidation of DNA(3) results in the reaction of
radical cations at TTT but not at UUU sequences.
Fig. 2 Autoradiograms of DNA(1–3). D and D1 are control lanes
(no UVA irradiation, and UVA irradiation but no piperidine treatment,
respectively). The labels above the lanes identify the DNA oligomer and
show the time of irradiation in min. Lanes labeled A/G and T are
the Maxim–Gilbert sequencing lanes. The figure is a composite formed
by editing a larger PAGE gel. All quantitative data were obtained by
phosphorimagery of unedited gels.
it contains a single GG step. The GG step of DNA(2) is positioned
˚
22 base pairs (ca. 82 A) from the AQ group. The irradiation of
DNA(2) results in quenching of the reaction at the thymines that
is seen in DNA(1); instead, most of the observed strand cleavage
occurs at the GG step, see Fig. 2. This result indicates that the
radical cation introduced by irradiation of the AQ residue migrates
reversibly through the DNA duplex and is trapped eventually in
an irreversible chemical reaction at the most reactive site.6 For
DNA(1) the most reactive sites are at the thymines, for DNA(2)
the most reactive site is the GG step.
Quantum yield of reactions
The significance of our discovery that the reactions of radical
cations in DNA comprised of A/T bases occur primarily at
Scheme 1
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 916–928 | 919
©

View Article Online
thymine depends on the quantum yield for these processes. We
examined the reaction of DNA(4) to measure the quantum yield
for loss of DNA (U−DNA). The AQ-linked strand of DNA(4)
consists only of A and U nucleobases. For this reason, cleavage of
this strand is expected to be negligible. The complementary strand
of DNA(4) is comprised of A and T bases. This strand has four TT
steps that are separated by AA steps. The one-electron oxidation
of DNA with this sequence results in detectable strand cleavage
only at the TT steps (see below).
Optically matched samples of DNA(4) and an anthraquinone-
2,6-disulfonate actinometer40 were irradiated at 350 nm for 6 min in
a Rayonet photoreactor. The irradiated DNA samples were treated
with piperidine, to cause strand cleavage, and then subjected
to HPLC analysis to determine the amount of intact A/T
strand remaining. The irradiation reaction was carried out to low
conversion to minimize errors that would result from the damage
to more than one thymine on each DNA molecule. Such over-
irradiation would lead to an underestimation of the quantum
yield. This experiment indicates that for DNA(4), U−DNA = (2.1
0.2)%. This value is similar to that obtained for the AQ-sensitized
oxidation of DNA containing GG steps,48 which shows that the
reaction of thymine radical cations leading to strand cleavage is not
a particularly rare event for DNA oligomers that do not contain
guanines.
systematic reduction in the amount of strand cleavage. These
findings are presented as a histogram of strand cleavage yield
in Fig. 3. It should be noted that inhibition of strand cleavage
by GSH suggests that whatever product is formed in its reaction
with a thymine radical does not result in DNA strand cleavage at
that site upon reaction with hot piperidine even after treatment
with Na2IrCl6. Finally, it should also be noted that control
experiments show that GSH does not itself inhibit piperidine-
induced strand cleavage at damaged thymines. Significantly, a plot
of the reciprocal of total strand cleavage yield at the TT sequences
of DNA(5) against GSH concentration is non-linear. This suggests
that GSH is quenching more than one radical intermediate that
leads, eventually, to strand cleavage at T. These findings support
the reaction mechanism for this process, suggested below.
Quenching of strand cleavage at thymines by glutathione
The reaction of thymine in these oligomers is initiated by the
one-electron oxidation of a nucleobase to form a radical cation.
The participation of the thymine methyl group in this process
is implied by the result obtained when UUU is substituted for
TTT. Previous reports indicate that the thymidine radical cation
in aqueous solution loses a proton from the methyl group to
form a 5-(2ꢀ-deoxyuridinyl)methyl radical, which is trapped by
molecular oxygen in a subsequent step.38,49 The radical formed by
proton loss from the thymine methyl group may also play a role
in its reaction in DNA. We examined the effect of glutathione
(GSH), a tripeptide having a cysteine residue that is known to be a
free radical scavenger,50,51 on the strand cleavage that results from
irradiation of an AQ-linked oligomer.
As previously noted, the [32P]-labeled strand of DNA(2) con-
tains four TTT sequences separated by AA steps and a GG
step following the last TTT sequence. This oligomer was used
to differentiate between possible quenching of a precursor to a
thymine radical by GSH, which would also quench the reaction
at the GG step, and the quenching of a thymine-based radical
itself. In the absence of GSH, irradiation of DNA(2) results in
some strand cleavage at the TTT sequences, but the major reaction
occurs at the GG step; see Fig. 2. Addition of GSH (up to 5 mM)
to solutions of DNA(2) before irradiation results in reduction in
the amount of strand cleavage detected at the thymines without
meaningfully affecting the reaction at the GG step.52 This result
indicates that a thymine radical, not a precursor to the thymine
radical that also results in strand cleavage at the GG step, is
quenched by GSH.
Fig. 3 Histogram showing the effect of glutathione (GSH) on thymine
damage of DNA(5). The relative amount of strand cleavage (counts per
second, cps) observed by phosphorimagery at the four TT sequences as a
function of glutathione concentration, as is indicated.
The complex effect of UT and TU steps
The experimental examination of the one-electron oxidation of
DNA leading to strand cleavage at thymine described thus far
reveals a process that proceeds through at least two trappable
radical intermediates. To further probe the nature and identity
of these intermediates, we investigated a set of DNA oligomers,
DNA(6) and DNA(7), that contain TT steps with uracils in place
of thymines at strategic locations. The pattern of strand cleavage
that emerges in these experiments reveals important details about
the reactive intermediates and subtle insight into the structural
control of the reaction mechanism.
The [32P]-labeled strand of DNA(5) contains four TT steps each
separated by two adenine nucleobases. For comparison, DNA(6)
is identical with DNA(5) except that the TT steps are replaced
by two sets of alternating TU and UT steps, and DNA(7) has,
starting from the 3ꢀ-end, UT, TT, TU and UU steps in place of TT
steps. These oligomers were irradiated, then treated with Na2IrCl6
and piperidine, and subjected to PAGE analysis. The results are
shown in Fig. 4.
DNA(5) contains four TT steps in the labeled strand. Its irra-
diation and subsequent treatment with Na2IrCl6 and piperidine
results in strand cleavage at each of the TT steps. The addition
of GSH to solutions of DNA(2) before irradiation results in a
As expected, strand cleavage is detected at each of the four TT
steps of DNA(5), and no strand cleavage is observed at the UU
step of DNA(7). But, surprisingly, strand cleavage in DNA(6) and
DNA(7) is strongly dependent on the order of nucleobases in the
920 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
steps. Experiments with radiolabeled versions of DNA(4) show, as
expected, that no strand cleavage occurs in the AQ-linked strand
and that the expected strand cleavage, after piperidine treatment,
is seen at the TT steps of the complementary strand. Unlabeled
DNA(4) was irradiated for various times under the standard
conditions; see Fig. 5. Rather than the usual piperidine treatment
after irradiation, these samples were treated with nuclease P1,
phosphodiesterases II and I and alkaline phosphatase, and the
resulting nucleosides were separated by means of HPLC. A
typical HPLC profile is shown in Fig. 6. The four product
peaks were identified mass spectroscopically by comparison with
authentic samples58 to be cis-thymidine glycol (c-ThdGly), trans-
thymidine glycol (t-ThdGly), 5-(hydroxymethyl)-2ꢀ-deoxyuridine
(5-HMdUrd) and 5-formyl-2ꢀ-deoxyuridine (5-FormdUrd). Sig-
nificantly, only a small amount of 8-oxodAdo is detected in these
experiments (see Fig. 5). This finding confirms the conclusion
drawn from the strand cleavage results that reaction occurs
primarily at T in the A/T base pairs.
Fig. 4 Autoradiograms of DNA(5–7) showing the complex effect of TU
and UT steps. D and 15 represent 0 and 15 min of irradiation, respectively.
T indicates the thymidine-sequencing lanes. The figure is a composite
formed by editing a larger PAGE gel. All quantitative data were obtained
by phosphorimagery of unedited gels.
UT and TU steps. Both the bases are damaged in a 3ꢀ-TU-5ꢀ steps
but neither reacts in 3ꢀ-UT-5ꢀ steps. The same pattern is observed in
both oligomers DNA(6) and DNA(7). Evidently, the reactions of
the thymine radical cation that lead, eventually, to strand cleavage
depend not only on the presence of the thymine methyl groups
but also upon the precise placement of those groups. Substitution
of a uracil for the T at 5ꢀ-side of a TT step has only a modest
effect on the relative amount of strand cleavage observed, but
replacement of the 3ꢀ-T with U nearly completely inhibits reaction
at both bases at that step. This finding indicates that the process
leading to reaction of thymine radical cation and strand cleavage
at TT steps, at least in part, involves two adjacent nucleotides in
the DNA duplex—a process that has been referred to as a tandem
reaction.38,53
Fig. 5 The yield of different thymidine and 2ꢀ-deoxyadenosine oxidation
products relative to the amount of unreacted thymidine as a function of
the irradiation time.
The products of thymine oxidation resulting from the one-
electron oxidation of DNA(4) fall clearly into two sets. The first
set, c- and t-ThdGly, results from the oxidation of the thymine 5,6-
double bond, and the two products 5-HMdUrd and 5-FomdUrd,
result from reaction at the thymine methyl group. To distinguish
between primary and eventual secondary oxidation, the yield of
each of these products (relative to thymidine) was determined as
a function of the extent of reaction. The results (Fig. 5) show that
the product ratios are essentially invariant with time and that the
yield of all products extrapolates to zero at zero irradiation time.
This finding shows that these oxidized 2ꢀ-deoxyribonucleosides are
primary thymidine oxidation products, each being formed from
reaction originating from either hydration of the thymine radical
cation or addition of molecular oxygen to the related deprotonated
methyl centered radical. The major product is 5-FormdUrd,
which is formed in 63% yield, and the yields of 5-HMdUrd,
cis-ThdGly and trans-ThdGly are 13, 20, and 4%, respectively.
The identification of these products and their quantification in
relative yields help to identify the mechanism for the one-electron
Identification of the products formed from the reaction of T radical
cation in duplex DNA
A key to understanding the nature of the reactions of thymine
radical cation in duplex DNA is to identify the products that are
formed in this process. The results reported thus far indicate that at
least one of the reaction products (a damaged nucleobase) initiates
strand cleavage when treated with piperidine. Since it is known that
thymidine glycols and 5-formyl-2ꢀ-deoxyuridine are alkali-labile
lesions likely to lead to strand cleavage,53–57 further investigation
was required to identify the precise oxidation products that are
formed. We carried out a series of HPLC–MS/MS experiments to
identify and to quantify the products that result from the one-
electron oxidation of duplex DNA oligomers that contain no
guanine nucleobases.
The AQ-linked strand of DNA(4) contains only A and U
bases, and its complementary strand has four equally spaced TT
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 916–928 | 921
©

View Article Online
Fig. 7 Autoradiogram showing crosslinking in the UVA-irradiation of
DNA(5) (lanes 1–4) and DNA(8) (lanes 5–8). Samples in lanes 1, 2, 5 and 6
were not treated with Na2IrCl6 or piperidine, while the samples in lanes 3, 4,
7 and 8 have been treated with those reagents. Lanes 2, 4, 6 and 8 have been
irradiated for 20 min and lanes 1, 3, 5 and 7 are the corresponding dark
controls. The figure is a composite formed by editing a larger PAGE gel.
All quantitative data were obtained by phosphorimagery of unedited gels.
gels of samples irradiated for a few min when the autoradiogram is
exposed for the standard amount of time in either samples treated
with Na2IrCl6 and piperidine or in untreated samples. However, for
samples of DNA(5) irradiated for 20 min and with long exposure
of the gel to photographic film, weak crosslink bands are detected
in the samples untreated with Na2IrCl6 and piperidine. Although
precise quantification of the crosslink yield is not possible with
these experiments, it is clearly very low. No crosslinking is visible
with a sequence consisting of only A and U nucleobases on the
labeled strand, DNA (8). These observations assist the creation of
a proposed mechanism for the reaction of thymine radical cation
in DNA.
Fig. 6 HPLC trace and structures of the four thymidine oxidation
products identified by HPLC–MS/MS analysis.
oxidation of thymine in duplex DNA. This topic is discussed in
detail below.
Distance dependence of thymine damage and strand cleavage
Crosslinking of DNA by reaction of 5-(uracilyl)methyl radical
One of the hallmarks of the long-distance oxidation of normal,
guanine-containing DNA is that the amount of strand cleavage at
a particular guanine is sequence dependent and characteristically
decreases exponentially with the distance between the site of
initial oxidation (at the AQ group, for example) and a particular
guanine.6,63 This behavior has been interpreted in terms of the
phonon-assisted polaron-hopping model.1,6,63,64
We investigated three DNA oligomers, DNA(5,9,10), contain-
ing a uniform set of TT steps separated by AA, ATA, or ATATA
sequences, respectively. These experiments reveal a familiar expo-
nential distance dependence for the reaction of thymine radical
cations in duplex DNA. The data from analyses of high resolution
PAGE gels for these experiments are shown in the form of semilog
plots in Fig. 8. These experiments show that as the distance from
the AQ charge injector to the TT increases, the amount of strand
cleavage observed at that step decreases. In each case, the semilog
plot of strand cleavage yield with distance is linear, but the slope
The formation of 5-FormdUrd and 5-HMdUrd implicates forma-
tion of the thymine methyl radical from the initial deprotonation
reaction of thymine radical cation in duplex DNA.49 It is easily
seen how this radical could react with O2 and lead naturally to
the formation of 5-HMdUrd and 5-FormdUrd.59,60 It has been
reported that a specially formed thymine methyl radical reacts in
DNA with its paired adenine to form a crosslink between the two
DNA strands.61,62 We searched for evidence of crosslinking in the
one-electron oxidation of DNA(5) to gain additional support for
the intermediacy of the thymine methyl radical in this process.
The covalently linked AQ group of DNA(5), which contains
only A/T nucleobases, was irradiated in the usual manner.
However, to preserve crosslinks that might be susceptible to
cleavage, these samples were analyzed both with and without
treatment with Na2IrCl6 and piperidine (see Fig. 7). There is no
evidence of detectable crosslink formation in the electrophoretic
922 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
The surprising reactions of thymine radical cation in DNA
DNA can be oxidized by ionizing radiation,12 photochemically,11
or chemically with a variety of reagents.66,67 Independent of the
means of introducing a radical cation into DNA, it resides pri-
marily on the nucleobases7 and it is trapped by reaction with H2O
or O210,68 at guanines,6 or more often at Gn sites. The reason usually
offered to explain this G-selective reaction based upon relative Eox
seems obvious, but it is incomplete. It is certainly true that guanines
in DNA have the lowest Eox of the four DNA bases, and it seems
certain that the Eox of GG and GGG sequences is somewhat
lower than that of an “isolated” G (one without a neighboring
guanine).69 It is for these reasons that it has been concluded that
radical cations hopping through DNA pause briefly at guanines
where they are more “stable” due to the low Eox, and, consequently,
it is at guanine that they are irreversibly trapped.
In oligomers containing only A/T base pairs the radical cation
must reside primarily on either an adenine or a thymine. The
difference in Eox between A and T measured in acetonitrile solution
is 0.15 V.35 The Eox of these nucleobases will certainly be affected
by their incorporation into DNA and by changes in solvation, but
it seems extraordinarily unlikely that the equilibrium Eox of T will
shift to be less than A. Thus, based simply upon the experimentally
determined Eox of A and T in solution, the population of the
thymine radical cation at equilibrium is expected to be less than
1% of that for the adenine radical cation. Thymine is not the most
stable site for the radical cation, but evidently it is the most reactive.
It has been understood for more than 50 years37 that product
yields are not necessarily correlated with the stability of precursor
reactive intermediates. This fact is enshrined in the Curtin–
Hammett principle, and this principle bears additional discussion
in the context of the reactions of thymine radical cations in
DNA. The Curtin–Hammett principle states that, for reactions
passing rapidly and reversibly through more than one reactive
intermediate each leading to a different product, the ratio of
products depends on the difference in the free energy of the
transition states leading to each product, not specifically on the
relative energies of the intermediates.
The Curtin–Hammett principle was formulated to explain the
reactivity of molecules that exist in two interconverting forms
(conformers), each of which gives a different product.70 In the
current context, the reactive intermediates of the Curtin–Hammett
principle can be considered to be the nucleobase radical cations (or
radical cations delocalized as a polaron over several bases)71 that
hop rapidly and reversibly from site to site in duplex DNA. The
products, of course, are the familiar damaged nucleosides resulting
from trapping of the radical cation; among these are 8-oxo-7,8-
dihydro-2ꢀ-deoxyguanosine, 8-oxodAdo, the thymidine oxidation
products identified above, and others.12,70,72 The key concept of the
Curtin–Hammett principle is that the relative abundances of the
intermediates (determined by the relative Eox of the nucleobases
in the current case) cannot be used to predict the ratio of products
formed. The relative yield of products is determined by the
difference in free energies of the transition states (DDG‡) leading
to their formation. In DNA that contains guanine, the energy of
the transition state (DG‡) for its reaction with H2O or O2 evidently
is lower than that for the reaction of any other base radical cation.
For DNA that contains only A/T base pairs, DG‡ for reaction at
the thymine radical cation is lower than DG‡ for reaction at the
Fig. 8 Semilog plots of strand cleavage at a particular TT step (TTn)
divided by the total strand cleavage observed at all TT steps (TTt) for
DNA(5), DNA(9) and DNA(10) as a function of the distance of each TT
˚
step from the covalently-linked AQ (assuming 3.4 A per base pair).
of the plot is dependent upon the sequence of bases separating
the TT steps. The magnitude of the slope in such a plot is linked
to the relative rate constants for radical cation hopping (khop) and
irreversible trapping (ktrap) by its reaction with H2O or O2.64 These
findings reveal that hopping and reaction of the radical cation at
thymine exhibit fundamentally the same pattern as the hopping
and reaction at guanine-containing sites in duplex DNA.
Discussion
The experimental results reported above reveal surprising and
interesting features about the one-electron oxidation of DNA
oligomers that do not contain guanine or whose guanine nucle-
obases are far removed from the site of initial oxidation. First,
reaction occurs at thymine despite the fact that it has the highest
oxidation potential of the four DNA nucleobases. This observation
contradicts the long-held view that one-electron oxidation occurs
at guanine because it is the base with the lowest Eox. Clearly,
other factors are at play in determining the site of reaction, and
these will be considered in detail below. Second, the reaction of
the thymine radical cation in DNA leads to a set of products
that reveal a complex reaction mechanism involving proton loss
from the methyl group of the thymine radical cation, specific
water addition across the thymine radical cation 5,6-double bond
at C6 and a delicately balanced interplay of distance and steric
interactions between adjacent nucleotides. DNA is a complex
molecule, and the reactions of the thymine radical cation in DNA
reflect this complexity. However, the experiments reported here
lead to the proposal of a reaction mechanism that accommodates
the product data, the glutathione quenching results, and the
complex consequences of uracil substitution for thymine in TT
steps. This mechanism has important implications for the oxidative
reactions of genomic DNA.65 Finally, the experiments on the
distance dependence of thymine damage underscore the emergent
nature of long-distance charge transport in DNA.6 In certain
sequences, runs of A/T base pairs are barriers to radical cation
migration. However, in sequences that contain no guanines (which
are low-energy radical cation traps), hopping through these A/T
sequences occurs very efficiently.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 916–928 | 923
©

View Article Online
Fig. 9 Possible mechanism for the tandem formation of T-damaged products at the 3ꢀ- and 5ꢀ-thymines of a TT step starting with the radical cation
localized on the 5ꢀ-T.
adenine radical cation. Thus despite the fact that A has a much
lower Eox than T, very little 8-oxodAdo is formed and nearly all of
the products arise from reactions of the thymine radical cation.
At the pH of the experiments reported here, DNA is a
polyanion that exists in a continually fluctuating solvation (H2O)
and counter-ion (Na+) environment. It has been found that the
instantaneous positions of the Na+ ions and solvent molecules
have a very large effect on the energy and the localization of
radical cations in DNA.9 The Curtin–Hammett principle explains
the composition of product mixtures for reactions that proceed
through rapidly equilibrating intermediate conformers. In the
current context, it is the hopping of the radical cation (polaron)
from one site to the next, driven by motions of H2O and Na+
ions, that corresponds to the classical equilibration of conformers.
At equilibrium, the fraction of the radical cation that resides on
thymines must be very small. But it is not the composition of this
equilibrium mixture that determines the product yield, it is DDG‡.
In the case of reaction of the guanine radical cation with H2O in
DNA, it has been found that the DG‡ is controlled by solvation
and counter-ion association, and that the product is stabilized by
transfer of a proton through water by a Grotthaus73 mechanism to
a nearby phosphate group.7 It is likely that similar considerations
apply to the reaction of thymine radical cation in DNA. That is,
there are configurations of solvent and counter-ions that cause the
activation free energy for reaction of the thymine radical cation to
be lower than that for the reaction of the adenine radical cation.
And as a consequence, the vast majority of the products formed
come from reaction of the thymine radical cation.
A mechanism for reaction of the thymine radical cations in DNA
The results reported above show that the one-electron oxidation
of DNA containing only A/T base pairs proceeds through
intermediate free radicals that can be trapped with GSH and
give products resulting from reactions at the methyl group and
the 5,6-double bond of thymine.74 Also, for 3ꢀ-TT-5ꢀ segments,
substitution of U for T at one base can affect reaction at the
other, which points to the operation, at least in part, of a tandem
reaction mechanism. Based on these findings, we developed a set of
plausible mechanistic pathways, which are shown in Fig. 9 and 10.
In both cases, the pathways shown commence after the migrating
924 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Fig. 10 Possible mechanism for the (tandem?) formation of T-damaged products at the 3ꢀ- and 5ꢀ-thymines of a TT step starting with the radical cation
localized on the 3ꢀ-T.
radical cation has been “localized” at a particular thymine by some
configuration of solvent molecules and counter-ions.
radical, which is shown as 3 in Fig. 9. Of course, this methyl
radical can be trapped by O2, leading eventually to the formation
of 5-HMdUrd or 5-FormdUrd. The net result of this sequence of
reactions is the conversion of the 5ꢀ-T to thymidine glycols and, by
a tandem reaction, the 3ꢀ-T is converted to the coupled products
5-HMdUrd or 5-FormdUrd. This sequence of reactions accounts
for the observation that no strand cleavage is observed at either
base when there is a U in place of the 3ꢀ-T of a TT step. Clearly,
there is no methyl hydrogen atom on the uracil for the 6-hydroxy-
5,6-dihydrothymidine-5-peroxy radical (2) to abstract. Thus, in
3ꢀ-UT-5ꢀ steps, reversibly formed peroxy radical 2 either simply
reverts eventually to the radical cation, where it is annihilated to
reform thymine, or it gives an undetected product that does not
lead to strand cleavage.
Consider first the case of a TT step where the radical cation is
localized on the 5ꢀ-T, as in 1 in Fig. 9 with R = CH3. A possible
reaction of this intermediate is the loss of a proton to form a methyl
radical (see 6 in Fig. 10 for a similar structure). Another possibility
is that the 5ꢀ-T radical cation is trapped by the reversible addition
of H2O (addition of H2O may be reversible but that of O2 is not)
and O2 across its 5,6-double bond, which will give an intermediate
6-hydroxy-5,6-dihydrothymidine-5-peroxy radical (2 in Fig. 9). In
this case, our findings suggest that addition across the 5,6-double
bond is more likely to lead to strand cleavage than is the loss of a
proton from the methyl group of this thymine radical cation. This
view is supported by the observation that replacement of the 3ꢀ-T
by U in TT steps inhibits strand cleavage at both the 3ꢀ-U and
at the 5ꢀ-T. As shown in Fig. 9, the peroxy radical 2 formed by
addition across the double bond may abstract a hydrogen atom
from the C5-methyl group of the adjacent 3ꢀ-T to form a methyl
To be valid, this sequence of reactions must also account for the
observation that strand cleavage occurs at both U and T when a TT
step is replaced by 3ꢀ-TU-5ꢀ. Again, we begin by first considering
the case where the radical cation is localized on the 5ꢀ-U, (1 in
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 916–928 | 925
©

View Article Online
Fig. 9 with R = H). Of course, there is no methyl proton to
lose and thus the only likely reaction is addition of H2O/O2 to
form the 6-hydroxy-5,6-dihydrouracil-5-peroxy radical, shown as
2 in Fig. 9 with R = H. The peroxy radical thus formed can
abstract a C5-methyl hydrogen atom from the adjacent 3ꢀ-T, and
this tandem reaction will yield 2ꢀ-deoxyuridine glycols from the 5ꢀ-
U and the coupled products 5-HMdUrd or 5-FormdUrd from the
3ꢀ-T. These products are expected to result in strand cleavage upon
hot piperidine treatment at both the nucleobases in the 3ꢀ-TU-5ꢀ
sequence, as is observed experimentally.
in DNA by a photochemical reaction. Crosslinking is observed
from the one-electron oxidation of DNA oligomers comprised
only of A/T bases, which we postulate involves this radical
intermediate.
DNA is a complex molecule, and this complexity is reflected in
the reactions described here. The mechanisms postulated in Fig. 9
and 10 account for the experimental results, but they are certainly
incomplete. For example, the explanation offered for the effect of
substituting U for T of TT steps takes account only of the missing
methyl group. We have recently shown in the reaction of GG steps76
that nucleobase substitution at adjacent locations affects local
solvation and the controlling steric environment, which control
the reaction outcome. These effects may also play a role in the
reactions discussed here. We are continuing to test the mechanistic
hypothesis offered here.
The migrating radical cation may also localize on the 3ꢀ-T
of the TT steps; the sequence of reactions we propose must
also fit the experimental observations for this possibility. Just
as in the previous case, the 3ꢀ-T radical cation, 5 in Fig. 10,
may react by losing a proton from its methyl group or by the
reversible addition of H2O/O2 across the double bond. The latter
reaction generates a peroxy radical that is related to 2 of Fig. 9.
Molecular modeling studies75 on B-form DNA show that this
Long-distance radical cation migration in A/T DNA
ꢀ
˚
peroxy radical generated at the 5 -T is 2.9 A from the C5-methyl
hydrogen of 3ꢀ-T; putting it within reaction distance. However,
The mechanism of long-distance charge migration has been under
intensive investigation since the now-discredited claim was made
that DNA is a “molecular wire”.71,77 These studies resulted in the
development of the phonon-assisted polaron hopping model for
long-distance radical cation transport in DNA.71,78 Briefly, radical
cations are localized over a few adjacent bases by small distortions
of the DNA structure and by stabilizing changes to the solvation
environment.7 This self-trapped radical cation is referred to as
a polaron and the magnitude of the stabilization is referred to
as its binding energy.79 The polaron hops from site to site when
thermal motions (phonons) provide sufficient activation energy to
overcome their binding. The primary driver for polaron hopping
is the motions of the Na+ counter-ions.9
these calculations reveal that for the peroxy radical generated
ꢀ
ꢀ
˚
at the 3 -T, the C5-methyl hydrogen of the 5 -T is 4.2 A away;
a distance sufficiently great to render the hydrogen abstraction
reaction unlikely. Thus in this view, addition across the double
bond of the 3ꢀ-T radical cation does not lead to a tandem reaction
or to observable strand cleavage. However, the radical formed by
loss of a proton from the 3ꢀ-T radical cation, 6, may react with O2
to form the 5-methylperoxy radical 7 and eventually the coupled
products 5-HMdUrd or 5-FormdUrd, as is shown in Fig. 10. This
analysis indicates that no strand cleavage at either T of the TT
step is expected when the 3ꢀ-T is replaced by U and the radical
cation is localized at the 3ꢀ-position, because formation of the 5-
methylperoxy radical is not possible. This is consistent with the
experimental observations.
In those cases where the DNA has a regularly repeating pattern
of nucleobases, the distance dependence of radical cation reaction
The observation that the central T in 3ꢀ-TTT-5ꢀ segments of
DNA(2) are damaged most heavily is consistent with the proposed
mechanism. The radical cation at the central T is expected
to participate in a tandem reaction with the 3ꢀ-T to generate
piperidine-labile thymine glycol residues. Also, when the radical
cation is localized at the 5ꢀ-T, it may similarly participate in a
tandem reaction generating piperidine-labile 5ꢀ-HMdUrd or 5-
FormdUrd, again, at the central T. Thus there are two reaction
routes that lead to strand cleavage at the central T of TTT
segments.
efficiency is controlled by two parameters: khop, and ktrap
.
In
64
these cases, linear semilog plots of reaction efficiency (strand
cleavage yield) with distance from the site of initial oxidation
are observed, and the slope of that line is determined by kratio
,
which is defined as the ratio of khop to ktrap. More generally,
the distance dependence of radical cation reactivity in DNA
with non-repetitive nucleobase sequences can be understood
only by considering the set and sequence of bases in the entire
oligonucleotide. In these cases, linear semilog plots are typically
not observed and the distance dependence of reactivity emerges
by consideration of the interactions of all nucleobases of the DNA
oligomer among themselves.6
The proposed reaction sequences give two sets of products:
glycols arising from water addition to the thymine 5,6-double
bond, and 5-HMdUrd or 5-FormdUrd that arise from reaction
of molecular oxygen to the 5-(2ꢀ-deoxyuridinyl)methyl radical.
The reactions outlined in Fig. 9 produce both sets of products
in a 1 : 1 ratio by a tandem process. The reactions outlined in
Fig. 10 produce only 5-HMdUrd or 5-FormdUrd. Experimentally,
we observe that these two sets are formed in a ca. 5 : 1 ratio in
favor of 5-HMdUrd and 5-FormdUrd. This suggests that proton
loss from the thymine radical cation in DNA leads to observable
products about four times more frequently than does addition to
the 5,6-double bond.
The DNA oligomers examined in this work fall into the first
linear distance dependence category. The data, shown in Fig. 8, for
DNA(5,9,10) reveal linear semilog plots for these oligomers where
four TT steps are separated by AA, ATA and ATATA sequences,
respectively. The slopes of the lines for DNA(5,9,10) are −0.009
−1
˚
0.001, −0.02
0.004, and −0.03
0.001 A , and the derived
values for kratio are ca. 20, 10, and 3, respectively. For each of
these oligomers, the rate of hopping is somewhat faster than the
rate of the irreversible trapping reactions. This behavior parallels
precisely that observed for numerous DNA oligomers that contain
regularly spaced guanines, and this fact confirms the central role
played by nucleobase radical cations in the reactions at thymines
that lead to strand cleavage upon hot piperidine treatment.
Finally, it has been reported61,62 that interstrand crosslinking
between thymine and its paired adenine is observed when a 5-(2ꢀ-
deoxyuridinyl) methyl radical (6 of Fig. 10) is directly generated
926 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
6 J. Joseph and G. B. Schuster, J. Am. Chem. Soc., 2006, 128, 6070–
6074.
Conclusions
7 R. N. Barnett, A. Bongiorno, C. L. Cleveland, A. Joy, U. Landman and
G. B. Schuster, J. Am. Chem. Soc., 2006, 128, 10795–10800.
8 A. Joy and G. B. Schuster, Chem. Commun., 2005, 2778–2784.
9 R. N. Barnett, C. L. Cleveland, A. Joy, U. Landman and G. B. Schuster,
Science, 2001, 294, 567–571.
10 A. K. Ghosh and G. B. Schuster, J. Am. Chem. Soc., 2006, 128, 4172–
4173.
11 B. Armitage, Chem. Rev., 1998, 98, 1171–1200.
12 C. J. Burrows and J. G. Muller, Chem. Rev., 1998, 98, 1109–1151.
13 A. Sancar, Annu. Rev. Biochem., 1996, 65, 43–81.
14 S. Kanvah and G. B. Schuster, J. Am. Chem. Soc., 2004, 126, 7341–
7344.
15 B. Demple and L. Harrison, Annu. Rev. Biochem., 1994, 63, 915–948.
16 H. E. Poulsen, H. Prieme and S. Loft, Eur. J. Cancer Prev., 1998, 7,
9–16.
17 P. Hasty and J. Vijg, Science, 2002, 296, 1250–1251.
18 A. Pike, B. Horrocks, B. Connolly and A. Houlton, Aust. J. Chem.,
2002, 55, 191–194.
19 H. W. Fink and C. Schonenberger, Nature, 1999, 398, 407–410.
20 M. Di Ventra and M. Zwolak, in Encyclopedia of Nanoscience and
Nanotechnology, ed. H. S. Nalwa, American Scientific Publishers,
Stevenson Ranch, California, 2004, vol. 2, pp. 475–493.
21 E. Braun and K. Keren, Adv. Phys., 2004, 53, 441–496.
22 I. Saito, T. Nakamura, K. Nakatani, Y. Yoshioka, K. Yamaguchi and
H. Sugiyama, J. Am. Chem. Soc., 1998, 120, 12686–12687.
23 S. Steenken and S. V. Jovanovic, J. Am. Chem. Soc., 1997, 119, 617–
618.
The irradiation of AQ-linked DNA oligomers with UVA light
absorbed by the anthraquinone leads to the one-electron oxidation
of DNA with the concomitant formation of a radical cation that
is localized on the nucleobases. The radical cation hops long
distances by the phonon-assisted polaron-hopping mechanism
and it is trapped irreversibly by reaction with H2O and O2 at the
most reactive sites. In oligonucleotides that contain guanine, the
most reactive sites are the guanines or Gn steps. Coincidentally,
guanines (Gn steps) are also the sites having the lowest Eox. In
contrast, for DNA oligomers that are comprised only of A/T
base pairs, the irreversible trapping reactions occur at T, not at
A. Adenine has a much lower Eox than thymine and for that reason
the radical cation is more stable at A than at T, but evidently the T
radical cation is more reactive than the A radical cation and, as is
explained by the Curtin–Hammett principle, the major products
observed are those that come from the most reactive site.
The products formed from the reaction of the thymine radical
cation in DNA fall into two categories: those resulting from
deprotonation of the methyl group, and those resulting from
hydration at the 5,6-double bond. The proposed mechanism for
formation of these products begins from an orientation of solvent
and counter-ions that localizes the radical cation on a particular
thymine and enables its reaction. That thymine radical cation in a
TT step may either lose a proton from its methyl group or H2O/O2
may add across its double bond. In the latter case, this addition
may initiate a tandem reaction that converts both thymines of
the TT step to oxidation products. However, the major products
observed originate with proton loss from the methyl group to form
a methyl radical that is subsequently trapped by reaction with O2.
The findings reported here may have important implications
for the oxidative reactions of genomic DNA where there are long
stretches of base pairs that contain no or few guanines. In these
circumstances, oxidatively induced damage is expected to generate
lesions at thymines.
24 K. Senthilkumar, F. C. Grozema, C. F. Guerra, F. M. Bickelhaupt, F. D.
Lewis, Y. A. Berlin, M. A. Ratner and L. D. A. Siebbeles, J. Am. Chem.
Soc., 2005, 127, 14894–14903.
25 B. Giese, Acc. Chem. Res., 2000, 33, 631–636.
26 H. A. Wagenknecht, Nat. Prod. Rep., 2006, 23, 973–1006.
27 F. D. Lewis, H. H. Zhu, P. Daublain, T. Fiebig, M. Raytchev, Q. Wang
and V. Shafirovich, J. Am. Chem. Soc., 2006, 128, 791–800.
28 K. Kawai and T. Majima, Pure Appl. Chem., 2005, 77, 963–975.
29 Y. Osakada, K. Kawai, M. Fujitsuka and T. Majima, Proc. Natl. Acad.
Sci. U. S. A., 2006, 103, 18072–18076.
30 A. Joy, A. K. Ghosh and G. B. Schuster, J. Am. Chem. Soc., 2006, 128,
5346–5347.
31 I. Saito, M. Takayama and S. Kawanishi, J. Am. Chem. Soc., 1995, 117,
5590–5591.
32 T. Douki and J. Cadet, Int. J. Radiat. Biol., 1999, 75, 571–581.
33 P. O’Neill, A. W. Parker, M. A. Plumb and L. D. A. Siebbeles, J. Phys.
Chem. B, 2001, 105, 5283–5290.
34 S. Fukuzumi, H. Miyao, K. Ohkubo and T. Suenobu, J. Phys. Chem.
A, 2005, 109, 3285–3294.
35 C. A. M. Seidel, A. Schulz and M. H. M. Sauer, J. Phys. Chem., 1996,
100, 5541–5553.
36 F. Bergeron, D. Houde, D. J. Hunting and J. R. Wagner, Nucleic Acids
Res., 2004, 32, 6154–6163.
37 D. Y. Curtin, Rec. Chem. Prog., 1954, 15, 111–128.
38 K. N. Carter and M. M. Greenberg, J. Am. Chem. Soc., 2003, 125,
13376–13378.
Acknowledgements
The authors at Georgia Institute of Technology are grateful
for financial support of this work from the National Science
Foundation and by the Vassar Woolley Foundation. Dr Joshy
Joseph prepared the DNA oligomers used in this work and
contributed useful discussion and expertise in the completion of
this publication. The authors at the De´partement de Recherche
Fondamentale sur la Matie`re Condense´e wish to thank the Euro-
pean EU Marie Curie Training and Mobility program (project no.
MRTN-CT2003 “CLUSTOXDNA”) for partial support.
39 S. M. Gasper and G. B. Schuster, J. Am. Chem. Soc., 1997, 119, 12762–
12771.
40 A. Harriman and A. Mills, Photochem. Photobiol., 1981, 33, 619–
625.
41 See ESI†.
42 D. Rehm and A. Weller, Isr. J. Chem., 1970, 8, 259.
43 M. H. Chung, H. Kiyosawa, E. Ohtsuka, S. Nishimura and H. Kasai,
Biochem. Biophys. Res. Commun., 1992, 188, 1–7.
44 J. G. Muller, V. Duarte, R. P. Hickerson and C. J. Burrows, Nucleic
Acids Res., 1998, 26, 2247–2249.
45 M. R. Dhananjeyan, R. Annapoorani and R. Renganathan, J. Pho-
tochem. Photobiol., A, 1997, 109, 147–153.
References
46 C. T. Bui and R. G. H. Cotton, Bioorg. Chem., 2002, 30, 133–137.
47 A. A. Purmal, G. W. Lampman, J. P. Bond, Z. Hatahet and S. S. Wallace,
J. Biol. Chem., 1998, 273, 10026–10035.
48 A. K. Dotse, E. K. Boone and G. B. Schuster, J. Am. Chem. Soc., 2000,
122, 6825–6833.
49 T. Delatour, T. Douki, C. D’Ham and J. Cadet, J. Photochem.
Photobiol., B, 1998, 44, 191–198.
50 D. Pogocki and C. Schoneich, Free Radic. Biol. Med., 2001, 31, 98–107.
1 G. B. Schuster, Acc. Chem. Res., 2000, 33, 253–260.
2 B. Giese, M. Spichty and S. Wessely, Pure Appl. Chem., 2001, 73, 449–
453.
3 M. A. O’Neill and J. K. Barton, J. Am. Chem. Soc., 2004, 126, 11471–
11483.
4 D. Dee and M. E. Baur, J. Chem. Phys., 1974, 60, 541–560.
5 G. B. Schuster and U. Landman, in Long-range charge transfer in DNA
I, Springer, Berlin/New York, 2004.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 916–928 | 927
©

View Article Online
51 C. Schoneich, U. Dillinger, F. Vonbruchhausen and K. D. Asmus, Arch.
Biochem. Biophys., 1992, 292, 456–467.
52 It has been shown that the guanine radical cation and/or the related
(G–H)• oxidizing radical can be easily reduced; see S. V. Jovanovic and
M. G. Simic, Biochim. Biophys. Acta, 1989, 1008, 1939–1944.
53 I. S. Hong, K. N. Carter, K. Sato and M. M. Greenberg, J. Am. Chem.
Soc., 2007, 129, 4089–4098.
54 M. J. Lustig, J. Cadet, R. J. Boorstein and G. W. Teebor, Nucleic Acids
Res., 1992, 20, 4839–4845.
55 T. Berthod, Y. Petillot, A. Guy, J. Cadet, E. Forest and D. Molko,
Nucleosides Nucleotides, 1996, 15, 1287–1305.
56 K. N. Carter and M. M. Greenberg, J. Org. Chem., 2003, 68, 4275–4280.
57 K. N. Carter, T. Taverner, C. H. Schiesser and M. M. Greenberg, J. Org.
Chem., 2000, 65, 8375–8378.
58 S. Frelon, T. Douki, J. L. Ravanat, J. P. Pouget, C. Tornabene and J.
Cadet, Chem. Res. Toxicol., 2000, 13, 1002–1010.
59 C. Decarroz, J. R. Wagner, J. E. van Lier, C. M. Krishna, P. Riesz and
J. Cadet, Int. J. Radiat. Biol., 1986, 50, 491–505.
65 K. A. Friedman and A. Heller, J. Phys. Chem. B, 2001, 105, 11859–
11865.
66 J. Termini, Mutat. Res., 2000, 450, 107–124.
67 A. E. Aust and J. F. Eveleigh, Proc. Soc. Exp. Biol. Med., 1999, 222,
246–252.
68 J.-L. Ravanat and J. Cadet, Chem. Res. Toxicol., 1995, 8, 379–388.
69 F. D. Lewis, R. S. Kalgutkar, Y. W. Xiaoyang, J. Liu, R. T. Hayes,
S. E. Miller and M. R. Wasielewski, J. Am. Chem. Soc., 2000, 122,
12346–12351.
70 J. I. Seeman, Chem. Rev., 1983, 83, 83–134.
71 P. T. Henderson, D. Jones, G. Hampikian, Y. Z. Kan and G. B. Schuster,
Proc. Natl. Acad. Sci. U. S. A., 1999, 96, 8353–8358.
72 W. K. Pogozelski and T. D. Tullius, Chem. Rev., 1998, 98, 1089–1107.
73 C. J. T. de Grotthuss, Ann. Chim. (Cachan, Fr.), 1806, 58, 54–74.
74 M. M. Greenberg, Org. Biomol. Chem., 2007, 5, 18–30.
75 The molecular modeling studies were carried out using HyperChem 7.0
and the distances were measured after optimizing the DNA structure
by geometrical optimization.
60 C. M. Krishna, C. Decarroz, J. R. Wagner, J. Cadet and P. Riesz,
Photochem. Photobiol., 1987, 46, 175–182.
76 C. L. Cleveland, R. N. Barnett, A. Bongiorno, J. Joseph, C. Liu, G. B.
Schuster and U. Landman, J. Am. Chem. Soc., 2007, 129, 8408–8409.
77 C. J. Murphy, M. R. Arkin, Y. Jenkins, N. D. Ghatlia, S. H. Bossman,
N. J. Turro and J. K. Barton, Science, 1993, 262, 1025; E. D. A. Stemp
and J. K. Barton, Metal Ions in Biological Systems, Marcel Dekker,
New York, 1996, vol. 33, pp. 325–365.
61 I. S. Hong, H. Ding and M. M. Greenberg, J. Am. Chem. Soc., 2006,
128, 485–491.
62 I. S. Hong and M. M. Greenberg, J. Am. Chem. Soc., 2005, 127, 10510–
10511.
63 P. Das and G. B. Schuster, Proc. Natl. Acad. Sci. U. S. A., 2005, 102,
14227–14231.
78 E. M. Conwell and S. V. Rakhmanova, Proc. Natl. Acad. Sci. U. S. A.,
2000, 97, 4556–4560.
64 C. S. Liu, R. Hernandez and G. B. Schuster, J. Am. Chem. Soc., 2004,
79 C. M. Chang, A. H. C. Neto and A. R. Bishop, Chem. Phys., 2004, 303,
189–196.
126, 2877–2884.
928 | Org. Biomol. Chem., 2008, 6, 916–928
This journal is
The Royal Society of Chemistry 2008
©