Release of superoxide from nucleoside peroxyl radicals, a double-edged sword?

Article abstract of DOI:10.1021/ja973200s

5,6-Dihydrothymidin-5-yl (1) and 2'-deoxyuridin-1'-yl (3) were independently generated in solution under aerobic conditions. The release of superoxide (O₂^.-) from the respective peroxyl radicals derived from 1 and 3 was determined spectrophotometrically. Competition studies enable one to estimate that the rate constant for elimination of O₂^.- from the peroxyl radical (4) derived from 3 is ~1 s^-1. This process is competitive with the anticipated rate of trapping of 4 in DNA by glutathione. Relative rate studies indicate that O₂^.- generation reslting from the formation of 1 under aerobic conditions competes effectively with trapping of the peroxyl radical by Bu₃SnH. Superoxide elmination from the peroxyl radical of 1 (2) restores the damaged nucleoside to its unaltered form, implying that this reactive intermediate has a naturally occurring detoxification pathway available to it. However, the freely diffusible superoxide can react further to generate other reactive species capable of damaging nucleic acids, suggesting that the eliminations of O₂^.- from 2 is a potential double- edged sword.

Full text of DOI:10.1021/ja973200s

VOLUME 120, NUMBER 20
MAY 27, 1998
© Copyright 1998 by the
American Chemical Society
Release of Superoxide from Nucleoside Peroxyl Radicals,
a Double-Edged Sword?
Keri A. Tallman, Christopher Tronche, Dong Jin Yoo, and Marc M. Greenberg*
Contribution from the Department of Chemistry, Colorado State UniVersity, Fort Collins, Colorado 80523
ReceiVed September 11, 1997
Abstract: 5,6-Dihydrothymidin-5-yl (1) and 2′-deoxyuridin-1′-yl (3) were independently generated in solution
under aerobic conditions. The release of superoxide (O2•-) from the respective peroxyl radicals derived from
1 and 3 was determined spectrophotometrically. Competition studies enable one to estimate that the rate
constant for elimination of O2 from the peroxyl radical (4) derived from 3 is ∼1 s-1. This process is
•-
competitive with the anticipated rate of trapping of 4 in DNA by glutathione. Relative rate studies indicate
•-
that O2 generation resulting from the formation of 1 under aerobic conditions competes effectively with
trapping of the peroxyl radical by Bu3SnH. Superoxide elimination from the peroxyl radical of 1 (2) restores
the damaged nucleoside to its unaltered form, implying that this reactive intermediate has a naturally occurring
detoxification pathway available to it. However, the freely diffusible superoxide can react further to generate
•-
other reactive species capable of damaging nucleic acids, suggesting that the elimination of O2 from 2 is a
potential double-edged sword.
Alkylperoxyl radicals (ROO•) are involved in a variety of
chemical processes that result in the degradation of biomol-
ecules. Although the reactivity of alkylperoxyl radicals as chain
carriers in lipid autoxidation has been rigorously characterized,
the role of these reactive intermediates in nucleic acid damage
is less understood.1,2 The importance of ROO• in nucleic acid
damage is reflected in the observation that the presence of O2
during γ-radiolysis of DNA results in an approximately 3-fold
enhancement in damage.2a Trapping of the myriad of alkyl
radicals produced in this mechanism by O2 has been referred
to as a damage-fixing event. In some instances, the resulting
peroxyl radicals are believed to contribute to further DNA
damage by abstracting hydrogen atom(s) from the carbohydrate
backbone of the biopolymer.3 Evidence for this brand of DNA
damage amplification has been reported for peroxyl radical 2,
which was produced at defined sites within single-stranded
oligonucleotides from a modified nucleotide (eq 1).4 In contrast,
the trapping of alkyl radicals by thiols is generally considered
to be a protective process which can compete with the fixing
of DNA damage by ROO• formation. Using generic rate
constants for the trapping of alkyl radicals by O2 and by thiols
(in water), these two competing chemical processes are expected
to account for a significant fraction of the reactivity of DNA
radicals.5 In fact, independent studies on 2′-deoxyuridin-1′-yl
(3) produced at defined sites in oligonucleotides revealed that
physiologically relevant levels of thiol (5 mM) compete with
(1) (a) Porter, N. A. In Organic Peroxides; Ando, W., Ed.; John Wiley
& Son, Ltd.: Chichester, England, 1992. (b) Porter, N. A.; Mills, K. A.;
Carter, R. L. J. Am. Chem. Soc. 1994, 116, 6690. (c) Porter, N. A.; Mills,
K. A.; Caldwell, S. E.; Dubay, G. R. J. Am. Chem. Soc. 1994, 116, 6697.
(2) (a) von Sonntag, C. The Chemical Basis of Radiation Biology; Taylor
and Francis: London, 1987. (b) Breen, A. P.; Murphy, J. A. Free Radical
Biol. Med. 1995, 18, 1033.
(4) Greenberg, M. M.; Barvian, M. R.; Cook, G. P.; Goodman, B. K.;
Matray, T. J.; Tronche, C.; Venkatesan, H. J. Am. Chem. Soc. 1997, 119,
1828.
(3) (a) Deeble, D. J.; von Sonntag, C. Int. J. Radiat. Biol. 1986, 49,
927. (b) Schulte-Frohlinde, D.; Bothe, E. Z. Naturforsch. 1984, 39c, 315.
(5) Neta, P.; Grodkowski, J.; Ross, A. B. J. Phys. Chem. Ref. Data 1996,
25, 709.
S0002-7863(97)03200-9 CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/29/1998

4904 J. Am. Chem. Soc., Vol. 120, No. 20, 1998
Tallman et al.
Scheme 1
Scheme 2
O2 (0.2 mM) for 3.6 As with most generalizations, there are
exceptions to these principles. The formation of premutagenic
R-nucleotides via trapping of the respective nucleotide radical
by thiols is contrary to the role that these hydrogen atom donors
play as protecting agents.7 Similarly, one could ask whether
the formation of an alkylperoxyl radical is always a damage-
fixing event. Herein, we describe experiments on two nucleo-
side radicals (1, 3) that address this question.
Alkylperoxyl radicals can undergo heterolytic cleavage to
form a carbocation and a molecule of superoxide (O2•-).8 The
rate constant of this dissociative pathway is highly dependent
upon the stability of the incipient carbocation. Simple alkyl-
substituted peroxyl radicals (ROO•) eliminate superoxide with
rate constants that are too small to compete with their trapping
by thiols, or possibly even less activated intramolecular sources
of hydrogen atoms, or their bimolecular self-reactions. Het-
eroatom substitution of the carbon atom bearing the peroxyl
moiety significantly increases the rate of superoxide elimination.
The ROO• derived from the dimethyl acetal of acetaldehyde
dihydrouracil was obtained, but not for the peroxyl radical
derived from 5,6-dihydrothymin-5-yl.10
Nucleoside peroxyl radicals 2 and 4 are candidates for the
above homolytic and heterolytic processes, respectively. Elimi-
nation of superoxide from 4, followed by trapping by H2O,
yields the same deoxyribonolactone product (6, Scheme 1) that
one would obtain from its reduction by thiol. When formed
within a biopolymer, the deoxyribonolactone (6) represents an
(5) eliminates O2•- at pH 5 with a rate constant (6.5 × 104 s-1
)
that should enable this unimolecular process to compete
effectively with trapping by physiologically relevant levels of
thiol (5-10 mM) (eq 3).2a,9 An alternative unimolecular
decomposition pathway is available to ROO• containing a
hydrogen atom in the â-position of the molecule. Given a
suitable thermodynamic driving force, such as rearomatization
(eq 4), the elimination of hydroperoxy radical (HOO•), which
alkaline labile lesion.11 Hence, elimination of O2 from 4
•-
would not result in a deviation from the principle of O2 trapping
as a lesion-fixing event. However, elimination of HOO• from
2 reconstitutes thymidine (Scheme 2). Considering that 2 is
•-
derived from a DNA lesion (1), elimination of O2 could be
defined as a naturally occurring detoxification pathway, and
would represent an exception to the principle of O2 fixation of
lesions.
In the above processes, only the fate of the peroxyl radicals
has been considered in the context of the nucleoside components.
The other product of these reactions is O2•-, which in itself is
not deleterious to nucleic acids. However, in the presence of
•-
redox active metal ions, O2 is transformed into the highly
damaging hydroxyl radical.2,12 In addition, in the presence of
•-
nitric oxide (NO), O2 forms peroxy nitrite (ONO2-) which
selectively damages deoxyguanosine in DNA.13 Hence, de-
pending upon the source of the O2 (e.g., 4) and the reaction
essentially completely dissociates under neutral conditions, can
occur with a rate constant greater than 105 s-1 8
. The viability
•-
(10) (a) Deeble, D. J.; von Sonntag, C. Int. J. Radiat. Biol. 1987, 51,
791. (b) Al-Sheikhly, M. I.; Hissung, A.; Schuchmann, H.-P.; Schuchmann,
M. N.; von Sonntag, C.; Garner, A.; Scholes, G. J. Chem. Soc., Perkin
Trans. 2 1984, 601.
(11) (a) Meijler, M. M.; Zelenko, O.; Sigman, D. S. J. Am. Chem. Soc.
1997, 119, 1135. (b) Meschwitz, S. M.; Schultz, R. G.; Ashley, G. W.;
Goldberg, I. H. Biochemistry 1992, 31, 9117.
(12) (a) Walling, C. Acc. Chem. Res. 1975, 8, 125. (b) Haber, F.; Weiss,
J. Proc. R. Chem. Soc. London, Sect. A 1934, 147, 332. (c) Fenton, H. J.
H. J. Chem. Soc. 1894, 65, 899.
of such processes in nucleobase radical chemistry has been
investigated. Support for an analogous reaction in studies on
(6) Tronche, C.; Goodman, B. K.; Greenberg, M. M. Chem. Biol., in
press.
(7) (a) Lesiak, K. B.; Wheeler, K. T. Radiat. Res. 1990, 121, 328. (b)
Ide, H.; Yamaoka, T.; Kimura, Y. Biochemistry 1994, 33, 7127.
(8) von Sonntag, C.; Schuchmann, H.-P. Angew. Chem., Int. Ed. Engl.
1991, 30, 1229.
(9) Schuchmann, M. N.; Schuchmann, H.-P.; von Sonntag, C. J. Am.
Chem. Soc. 1990, 112, 403.
(13) Tamir, S.; Burney, S.; Tannenbaum, S. R. Chem. Res. Toxicol. 1996,
9, 821.

Release of Superoxide from Nucleoside Peroxyl Radicals
J. Am. Chem. Soc., Vol. 120, No. 20, 1998 4905
Table 1. Generation and Reactivity of 1 under Anaerobic
Table 2. Generation and Reactivity of 1 under Aerobic
Conditionsa
Conditionsa
method of
gen of 1
% mass
% conv balance
% yield
method of
gen of 1
7
% yield 8
9
8
9
9:8
% conv
% mass balance
hν
hνb
∆
51 ( 4
78 ( 4
86 ( 5
93 ( 0
2 ( 1
3 ( 0
1 ( 1
1 ( 1
17 ( 3
1 ( 0.1
96 ( 2 74 ( 6
hν
hνb
∆
8 ( 1 48 ( 8 6 ( 2
95 ( 7
62 ( 5
64 ( 3
100
81 ( 4
88 ( 6
94 ( 1
12 ( 1 51 ( 1 4 ( 1 100
0.5 ( 0.1 100
4 ( 1 28 ( 3 7 ( 2
4 ( 1 34 ( 7 9 ( 4
28 ( 7
30 ( 16
103 ( 6
108 ( 7
b
∆
0.7 ( 0.1 100
b
∆
a [12] ) 0.1 mM; [Bu3SnH] ) 1.7 mM. b [(Bu3Sn)2] ) 0.2 mM.
a [12] ) 0.1 mM; [Bu3SnH] ) 1.7 mM. b [(Bu3Sn)2] ) 0.2 mM.
conditions, the elimination of O2•- could result in further DNA
damage. The participation of O2•- in DNA damage processes
makes its elimination from nucleoside peroxyl radicals (in
particular 2) a potential double-edged sword.
pathway and/or alternative hydrogen atom donor (e.g., benzene-
selenol). Only trace amounts of 5,6-dihydro-5-hydroxythymi-
dine (8) are observed under degassed conditions.16 We believe
that 8 is produced as a result of trace amounts of O2 that are
not removed during the freeze-pump-thaw degas cycles.
Anaerobic photolysis of 12 in the absence of (Bu3Sn)2 yields
significant amounts of thymidine (9), which are reduced to trace
amounts when photolyses are carried out in the presence of the
ditin reagent. The majority of 9 produced under these conditions
is believed to result from a side reaction involving the
disproportionation of two radicals, presumably the radical pair
produced upon direct irradiation of 12. Minor amounts of 9
may also result from an O2 dependent pathway which will be
discussed in detail below. The significant reduction in the
amount of 9 produced upon photolysis of 12 in the presence of
(Bu3Sn)2 is attributed to the more efficient photoinitiation of
radical chain reactions involving Bu3Sn• as the chain transfer
agent. Thermolysis (37 °C) of 12 in THF-D2O (1:1 by volume)
under anaerobic conditions using a water soluble azo initiator
(14) in the presence of Bu3SnH produces high yields of 5,6-
Results and Discussion
Methods for Nucleoside Radical Generation. The efficient
generation of 3 via Norrish type I photocleavage of 10 was
described previously.14 No evidence for singlet oxygen (1O2)
formation during UV irradiation of an oligonucleotide containing
10 was detected. This is fully consistent with the short lifetime
for the excited state of 10 detected by Stern-Volmer analysis.
Norrish type I photocleavage of 11 was considerably less
efficient, and irradiation of 11 resulted in the production of 1O2.4
Consequently, phenyl selenide 12 was designed as a potential
photochemical and/or thermal precursor to 1. The synthesis of
12 was completed in a straightforward manner starting from
bis-silyl-protected dihydrothymidine (13), as previously de-
scribed for the synthesis of 11.4 Although the diastereomers
of 12 were separable, this proved to be unnecessary for the
experiments described below.
dihydrothymidine (7) as a 1:1 mixture of diastereomers (Table
1). The thermolysis conditions produce only trace amounts of
8 and 9 regardless of whether (Bu3Sn)2 is present, consistent
with the expected efficient chain reactions involving Bu3Sn•.
Mass spectral analysis of bis-silylated dihydrothymine, produced
via formic acid hydrolysis and subsequent persilylation of crude
reaction mixtures, confirms that the majority of 7 formed under
anaerobic conditions is due to trapping of 1 by Bu3SnH.
Regardless of the reaction conditions (see Table 1), the
deuterium content of dihydrothymine was never greater than
8%. The deuterated dihydrothymine detected could be attributed
to small amounts of benzeneselenol produced during the
reactions.
Dihydrothymidine (7) is absent from product mixtures
resulting from decomposition of 12 under aerobic conditions.
This is consistent with the strongly favored trapping of 1 by O2
in competition with Bu3SnH, resulting in the formation of a
1:1 diastereomeric mixture of 5,6-dihydro-5-hydroxythymidine
(8, Table 2).16 When 12 is decomposed under aerobic condi-
tions in H218O, none of the heavier isotope is incorporated in
8, indicating that O2 is the sole source of the hydroxyl group in
this product. Although thermolyses were carried out using
stoichiometric initiator (14), the extent of conversion of 12 was
significantly reduced from that achieved under anaerobic
conditions (100%), or upon photolysis.17 This is presumably
due to O2 quenching of radical chain initiation and propagation.
It is important to note that the ratio of 9 to 8 is significantly
greater than one in the presence of low, but pseudo-first-order,
Phenyl selenide 12 proved to be a versatile precursor to 1.
Photolysis of 12 under anaerobic conditions resulted in a
moderate yield of 5,6-dihydrothymidine (7) as a 1:1 mixture of
diastereomers (Table 1). The reactions were carried out in a
mixture of THF and H2O (D2O) (1:1 by volume) in order to
strike a balance between the solubility requirements of the tin
hydride and the desire to carry out the experiments in an aqueous
environment.15 Employing an aqueous solvent system makes
extrapolation of the experimental results to biological systems
more reasonable. In addition, utilizing D2O in conjunction with
a nonexchangeable hydrogen atom donor (Bu3SnH) provides
an additional mechanistic handle regarding the origin of 7.
Deuterated 7 is indicative of involvement of a nonradical
(14) Goodman, B. K.; Greenberg, M. M. J. Org. Chem. 1996, 61, 2.
(15) Experiments carried out using a water soluble tin hydride reagent
(tris[3-(2-methoxyethoxy)propyl]stannane) complicated the product analysis,
and will be reported on at a later date. For the synthesis of this water soluble
tin hydride reagent see: Light, J.; Breslow, R. In Organic Syntheses; Coffen,
D. L., Ed.; Wiley: New York, 1994; Vol. 72, p 199.
(16) Barvian, M. R.; Greenberg, M. M. J. Org. Chem. 1993, 58, 6151.
(17) Mayer, S.; Prandi, J. Tetrahedron Lett. 1996, 37, 3117.

4906 J. Am. Chem. Soc., Vol. 120, No. 20, 1998
Tallman et al.
Figure 2. Ratio of thymidine (9) versus 5,6-dihydro-5-hydroxythy-
midine (8) formed upon photolysis of 12 (50 µM) as a function of
[Bu3SnH]-1. The slope of the line ) kElim/kRed ) 1.3 × 10-2 M.
Figure 1. Cytochrome c reduction upon photolysis of (b) 12 (50 µM)
in the presence of cyt c (50 µM), EDTA (0.1 mM), and catalase (12
µg), (2) 12 (50 µM), cyt c (50 µM), EDTA (0.1 mM), catalase (12
µg), and superoxide dismutase (5 units), and (9) cyt c (50 µM), EDTA
(0.1 mM), and catalase (12 µg).
of the expected amount of O2•- formed is accounted for by this
spectrophotometric assay. We believe that the remainder of
the O2•- released upon formation of thymidine is consumed by
disproportionation. These observations were qualitatively con-
concentrations of Bu3SnH. This indicates that the formation
of thymidine (9) is not due to a depletion of the hydrogen atom
donor.
firmed using the oxidation of epinephrine to adrenochrome (λmax
Hydroperoxyl Radical (Superoxide) Elimination from 2.
The large amount of thymidine (9) formed during decomposition
of 12 under aerobic conditions is consistent with the elimination
of HOO• from 2 (Scheme 2). The formation of thymidine (9)
from 2 is drawn as a single-step homolytic process, rather than
•- 19,20
) 480 nm) by O2
.
To gauge the facility of the formal elimination of O2•- from
2, the ratio of 9 to 8 was determined as a function of Bu3SnH
concentration (Figure 2). The tin hydride was present in pseudo-
first-order concentration in all experiments. In contrast to the
data reported above (Table 2), these experiments were carried
out at 55 °C in a mixture of THF-H2O (19:1 by volume) in
order to ensure reduction of the hydroperoxide to 8 and complete
solubility of Bu3SnH over its entire concentration range. The
ratio of 9 to 8 (12.8 ( 1.9) formed in the presence of the Bu3-
SnH (1 mM) under these less polar solvent conditions is slightly
larger than the respective ratio of products formed under
comparable reaction conditions at 37 °C in 1:1 THF-H2O in
the presence of 1.7 mM tin hydride (Table 2). Due to the
alteration of more than one experimental variable (trap con-
centration, solvent, temperature), the significance of solvent and/
or temperature effects on these reactions from this single
comparison will not be overinterpreted. However, the change
in the ratio of 9 to 8 is consistent with the expected reduction
in kRed (Scheme 2) with decreasing solvent polarity, and cannot
be accounted for solely by the change in tin hydride concentra-
tion.21
•-
a two-step process involving loss of O2 followed by depro-
tonation. This proposal is based upon the expected instability
of the intermediate carbocation R to the carbonyl moiety.
Furthermore, the absence of any 18O incorporation in 8 from
water (∼27.7 M) does not support the two-step heterolytic
process. Admittedly, a two-step heterolytic process cannot
definitively be ruled out, as the same labeling result would be
obtained if deprotonation overwhelmed hydration. The possible
involvement of a carbocation does not alter the discussion below.
The low pKa (4.8) of HOO• ensures that O2•- will be the ultimate
product whether the concerted process (as shown in Scheme 2)
or two-step process is operative.
•-
Evidence for the release of O2 from 2 was obtained using
a spectrophotometric assay involving the reduction of cyto-
chrome c.18 The reduced form of cytochrome c possesses a
strong chromophore at 550 nm. Irradiation of 12 (0.05 mM)
in the presence of 1 equiv of cytochrome c results in steady
growth of the reduced product (Figure 1). Photolysis of 12 in
the presence of superoxide dismutase (5 units) quenches 74.3%
of the amount of cytochrome c reduction, supporting the
The dependency of the 9:8 ratio as a function of Bu3SnH
concentration is consistent with the postulated competing
reactive pathways available to 2 (Scheme 2). The data clearly
demonstrate that elimination of HOO• effectively competes with
physiologically relevant concentrations of an efficient hydrogen
atom donor. The slope of Figure 2 (1.3 × 10-2 M) represents
the ratio kElim/kRed. The rate constant for the elimination process
(kElim ≈ 65 s-1) is estimated assuming that the bimolecular rate
constant for the reaction between Bu3SnH and 2 is ap-
proximately the same as for tert-butyl thiol reacting with ROO•
•-
contention that O2 is responsible for the observed reaction.
Additional superoxide dismutase (100 units, data not shown)
results in no further reduction in the amount of reduced
cytochrome c. We suggest that the remainder of the reduced
product is due to a superoxide independent process involving
the phenylselenyl group released upon photolysis. The amount
of cytochrome c reduced that is attributable to O2•- corresponds
to 36% of the observed thymidine (as determined by reversed-
phase HPLC). Considering that more than one-third of the
thymidine formed under aerobic photolysis of 12 is attributable
to radical cage processes (see Tables 1 and 2), more than half
(19) Misra, H. P.; Fridovich, I. J. Biol. Chem. 1972, 247, 3170.
(20) See the Supporting Information.
(21) Tronche, C.; Martinez, F. N.; Horner, J. H.; Newcomb, M.; Senn,
M.; Giese, B. Tetrahedron Lett. 1996, 37, 5845.
(18) (a) Armitage, B.; Yu, C.; Devadoss, C.; Schuster, G. B. J. Am. Chem.
Soc. 1994, 116, 9847. (b) McCord, J. M.; Fridovich, I. J. Biol. Chem. 1969,
244, 6049.

Release of Superoxide from Nucleoside Peroxyl Radicals
J. Am. Chem. Soc., Vol. 120, No. 20, 1998 4907
Figure 4. Ratio of [16O]6 to [18O]6 formed upon photolysis of 10 (0.05
mM) in H218O as a function of â-mercaptoethanol concentration. The
slope of the line ) kT/kE ) 4.3 × 103 M-1
.
Figure 3. Adrenochrome formation (∆Abs ) Abst - Abs0) upon
photolysis of (b) 10 (0.1 mM) in the presence of (R)-(-)-epinephrine
(0.1 mM) and desferal (1 mM), (2) 10 (0.1 mM), (R)-(-)-epinephrine
(0.1 mM), desferal (1 mM), and superoxide dismutase (100 units), and
(9) (R)-(-)-epinephrine (0.1 mM) and desferal (1 mM).
investigated by measuring the extent of 18O incorporation in 6
(resulting from aerobic photolysis of 10 in H218O) as a function
of â-mercaptoethanol concentration (Scheme 1). It was antici-
pated that the deoxyribonolactone (6) produced during photolysis
of 10 in H218O via superoxide elimination would contain 18O,
whereas reduction of the peroxyl radical by thiol would yield
[16O]6 (Scheme 1, Figure 4).26 Alternative pathways for the
formation of 6 were considered. For example, disproportion-
ation of 3 and the tert-butyl radical to form the unstable 1′,2′-
dehydronucleoside (15) would yield 6 upon hydrolysis.27 This
(5 × 103 M-1 s-1).22 The estimated rate constant for elimination
of HOO• from 2 is more than 1000 times slower than that
measured for rearomatization of the peroxyl radical derived from
the hydrogen atom adduct of benzene (eq 4), and is reflective
of the smaller thermodynamic driving force for this process.8
This estimate is indeed a rough one, and is based upon the fact
that alkanethiols and Bu3SnH react with alkyl radicals at similar
rates.23 One should note that these respective families of radical
traps show opposite reactivity trends for electrophilic perfluo-
roalkyl radicals.24 Therefore, the above bimolecular rate
constant for the reaction between Bu3SnH (kRed) and electro-
philic 2 (and in effect kElim) may be underestimated.
pathway was discounted as a major source of 6 on the basis of
the anaerobic photolysis of 10 in D2O, which produced 6
containing only 6% deuterium.26 In general, O2 independent
pathways for the formation of 6 were eliminated by carrying
out all photolyses in the presence of a minimum level of
â-mercaptoethanol ()0.5 mM). No deoxyribonolactone (6) is
produced upon photolysis of 10 (50 µM) under these conditions
in the absence of O2. Therefore, any 6 formed in the presence
of O2 and thiol must result from decomposition of the alkyl-
peroxyl radical (4).28 Formation of [18O]6 from radical-radical
reactions involving 4 (such as dimerization, followed by
decomposition of the tetroxide) are inconsistent with the
observed dependence of the [16O]6:[18O]6 ratio on thiol
concentration. In addition, processes involving tertiary alkyl-
peroxyl radicals are relatively slow (∼104 M-1 s-1), and would
not compete with thiol trapping of 4 under the photolysis
conditions.29 Nonetheless, photolysis experiments were carried
out under the least intense conditions possible, to minimize the
steady-state concentration of 3 (and in effect 4). Furthermore,
Superoxide (O2•-) Elimination from 4. Quantitative de-
termination of superoxide release using cytochrome c during
irradiation of 10 was not possible. Control experiments
demonstrated that one or more of the alkyl radicals were capable
of inducing reduction of the cytochrome.25 However, as was
•-
the case for 2, qualitative determination that 4 releases O2
was addressed using the oxidation of epinephrine as a probe
(Figure 3). Irradiation of 10 in the presence of superoxide
dismutase results in significant diminution of adrenochrome
formation. However, increasing the amount of superoxide
dismutase (200 units) does not completely quench adrenochrome
formation (data not shown). This indicates that a second oxidant
is present which is capable of oxidizing epinephrine. We
suggest that the tert-butylperoxyl and/or pivaloylperoxyl radicals
(also derived from Norrish type I photocleavage of 10) is this
other oxidant. However, we do not believe that these other
ROO• radicals are responsible for an appreciable amount of the
indirectly detected O2•-. In general, geminal heteroatoms are
•-
(26) It was independently demonstrated that the deoxyribonolactone (1)
does not undergo deuterium or oxygen exchange under the reaction
conditions.
(27) Robins, M. J.; Trip, E. M. Tetrahedron Lett. 1974, 3369.
(28) The yield of 6 was >90% of the amount of 10 consumed during
photolysis. Consumption of 10 was established by reversed-phase HPLC.
The yield of 6 was determined by GC/MS analysis following persilylation.
Both compounds were quantitated using internal standards.
(29) (a) Bennett, J. E. J. Chem. Soc., Faraday Trans. 1990, 86, 3247.
(b) During the photolysis (2 h), ∼25% of a solution of 10 (50 µM) is
converted in the RPR-100 photoreactor (equipped with one lamp). On the
basis of this level of conversion, we estimate that the steady-state
concentration of 4 during photolysis is at the nanomolar level. Consequently,
the rate of reaction of two molecules of 4 (using the rate constant in ref
29a) is significantly slower than trapping of 4 by thiol.
required in order for O2 elimination to compete with dimer-
ization of the corresponding peroxyl radicals.8,9
•-
The viability for competition of O2 elimination in the
presence of biologically relevant reductants (e.g., thiols) was
(22) (a) Schulte-Frohlinde, D.; Behrens, G.; O¨ nal, A. Int. J. Radiat. Biol.
1986, 50, 103. (b) Chenier, J. H. B.; Furimsky, E.; Howard, J. A. Can. J.
Chem. 1974, 52, 3682.
(23) Newcomb, M. Tetrahedron 1993, 49, 1151.
(24) Rong, X. X.; Pan, H.-Q.; Dolbier, W. R. Jr.; Smart, B. E. J. Am.
Chem. Soc. 1994, 116, 4521.
(25) EPR spin trapping experiments using 5,5-dimethylpyrroline N-oxide
were also unfruitful. Multiple trapping products were observed. Furthermore,
the superoxide adduct is unstable under the photolysis conditions.

4908 J. Am. Chem. Soc., Vol. 120, No. 20, 1998
Table 3. Effect of NaCl on the Reactivity of 4
Tallman et al.
Considering the relatively high effective concentration of
superoxide generated by the decomposition of peroxyl radicals
in nucleic acids (compared to its production in bulk solution),
such processes could result in DNA damage amplification when
effected in the presence of NO or redox active metal ions.12,13
When a radical such as 2 is the progenitor to superoxide,
elimination of this reactive oxygen species could very well be
a double-edged sword.
[NaCl] (mM)
kT/kE (×10-3, M-1
)
kE (s-1
)
0
25
50
4.3
3.8
3.5
1.2
1.3
1.4
it is more likely that dimerization of 4, or reaction of 4 with an
alkyl radical (e.g., tert-butyl or 3), would yield [16O]6.8 Hence,
we propose that the [18O]6 formed in H218O during photolysis
Experimental Section
•-
of 10 results from O2 elimination from 4.
General Methods. All reactions were carried out in oven-dried
glassware, under a nitrogen atmosphere, unless specified otherwise.
DMSO, pyridine, and CH2Cl2 were distilled from CaH2. Bu3SnH was
distilled from itself immediately prior to use. THF was distilled from
Na/benzophenone ketyl. D2O and H218O (>97%) were from Cambridge
Isotope Labs. BSTFA, â-mercaptoethanol, and KO2 were from Sigma-
Aldrich. Doubly distilled H2O (dd H2O) was obtained from a Barnstead
Nanopure still. Azo initiator 14 was from Wako. Ketone 10 was
prepared as previously described.32 All photolyses of oligonucleotides
were carried out in Pyrex tubes (0.25 in. i.d.) using a Rayonet
photoreactor (RPR-100) equipped with lamps having a maximum output
at 300 or 350 nm.
The ratio of the rate constants for peroxyl radical trapping
•-
by â-mercaptoethanol relative to O2 elimination (3.5 × 103
M-1) is extracted from the data in Figure 4. However, due to
the limited body of absolute rate constants available concerning
the trapping of alkylperoxyl radicals by thiols as discussed above
(kT ≈ 5.0 × 103 M-1 s-1), the absolute rate constant for
superoxide elimination (kE) from 4 can only be estimated to be
∼1.2 s-1 22
.
The possibility that the heterolytic fragmentation
of 4 might be significantly accelerated by increasing the ionic
strength was investigated (Table 3). The isotopic partitioning
in 6 as a function of thiol concentration varied only slightly
upon adding NaCl to the photolysis mixtures. The lack of a
strong salt effect on the isotope content of 6 could be a
consequence of the fact that both competing processes (thiol
trapping and O2•- elimination, Scheme 1) are accelerated upon
increasing the ionic strength of the solvent.5,21 Overall, these
results suggest that the additional stabilization by the pyrimidine
nitrogen of the incipient carbocation formed upon O2•- elimina-
tion from 4 is less than that of a second ethereal group of an
acetal (as in 5).8
5-(Phenylseleno)-3′,5′-O,O-bis(tert-butyldimethylsilyl)-5,6-dihy-
drothymidine. sec-BuLi (17.5 mL (1.3 M), 22.8 mmol) was added
to a solution of bis-TBDMS-5,6-dihydrothymidine (13, 3.67 g, 7.77
mmol) in THF (26 mL) at -78 °C.4 After 2 h, DMPU (4.98 g, 38.9
mmol) was added to the reaction mixture, followed by a solution of
PhSeBr (2.05 g, 8.70 mmol) in THF (12 mL). The reaction mixture
was slowly warmed to room temperature. After 6 h, the reaction
mixture was quenched with NH4Cl (4 mL), poured into H2O (100 mL),
and extracted with ether (3 × 100 mL). The organics were washed
with brine (100 mL) and dried over MgSO4. Flash chromatography
(19:1 CH2Cl2-EtOAc to 9:1 CH2Cl2-EtOAc) yielded the bis-silyl-
protected phenyl selenide as a white foam (3.03 g, 62%) as a (∼2:1)
mixture of diastereomers. Running a slower gradient enables one to
separate the diastereomers. Diastereomer A (minor): 1H NMR (CDCl3)
δ 7.75 (br s, 1H) 7.59-7.23 (m, 5H), 6.23 (t, 1H, J ) 7.2 Hz), 4.34
(m, 1H), 3.91-3.70 (m, 4H), 1.93 (m, 1H), 1.44 (s, 3H), 0.90 (s, 9H),
Conclusions
The experiments described above indicate that nucleoside
peroxyl radicals participate in unimolecular reactions that
produce O2•-. At first glance, the estimated rate constant for
0.87 (s, 9H), 0.10 (s, 3H), 0.09 (s, 3H), 0.06 (s, 3H), 0.05 (s, 3H); 13
C
O2 elimination from 4 (∼1 s-1, Table 3) may appear to be
•-
NMR (CDCl3) δ 170.6, 151.4, 138.0, 129.6, 128.8, 125.5, 86.7, 83.9,
72.5, 63.2, 47.4, 43.5, 37.4, 25.9, 25.7, 20.8, 18.3, 17.9, -4.7, -4.8,
-5.3, -5.4; IR (KBr) 3202, 3079, 2954, 2928, 2895, 1712, 1472, 1438,
1377, 1362, 1253, 1222, 1116, 1092, 1029, 834 cm-1. Diastereomer
B (major): 1H NMR (CDCl3) δ 7.73 (br s, 1H), 7.55-7.23 (m, 5H),
6.35 (dd, 1H, J ) 6.2, 8.1 Hz), 4.38 (m, 1H), 3.71-3.47 (m, 5H),
2.24-2.15 (m, 1H), 2.06-1.99 (m, 1H), 1.40 (s, 3H), 0.91 (s, 9H),
too slow to contribute significantly to DNA damage. However,
it has recently been shown that glutathione (which is present in
vivo) reacts with DNA peroxyl radicals at <2 × 102 M-1 s-1 30
.
Since the concentration of glutathione in cells is estimated to
be ∼5 mM, O2•- elimination from 4 within biopolymers could
compete with its reduction by thiol (kT[RSH]/kE ∼1).2a
0.89 (s, 9H), 0.08 (s, 3H), 0.07 (s, 3H), 0.06 (s, 3H), 0.05 (s, 3H); 13
C
In absolute terms, elimination of O2•- from 2 is considerably
faster than from 4. Superoxide elimination from 2 is also more
facile than from the analogous radical derived from the free
base, thymine.10a This process competes favorably with bimo-
lecular trapping by Bu3SnH in an aqueous solvent system. The
competition between hydrogen atom transfer to 2 and its
unimolecular reconstitution of thymidine is directly relevant to
the efficiency of the recently reported DNA damage amplifica-
tion mechanism involving this alkylperoxyl radical.4 While Bu3-
SnH is a superior hydrogen atom donor compared to molecules
such as THF (a model for a deoxyribose ring) in bimolecular
reactions, it is unclear at this time what the effective molarity
of an adjacent nucleotide’s carbohydrate is in a biopolymer
containing 2.22,31 Hence, the viability of the process described
in Scheme 2 requires further investigation within biopolymers,
and would constitute a naturally occurring detoxification
pathway.
NMR (CDCl3) δ 170.7, 152.3, 137.8, 129.7, 129.0, 125.1, 86.4, 83.4,
71.9, 62.9, 48.5, 43.8, 37.9, 25.9, 25.7, 20.7, 18.3, 17.9, -4.6, -4.8,
-5.3, -5.4; IR (KBr) 3212, 3076, 2954, 2928, 2897, 2857, 1715, 1697,
1472, 1437, 1377, 1361, 1255, 1216, 1115, 1092, 1028, 835 cm-1
.
5-(Phenylseleno)-5,6-dihydrothymidine (12). NH4F (1.81 g, 48.8
mmol) was added to a solution of the above bis-silyl compound (3.05
g, 4.85 mmol) in anhydrous MeOH (65 mL) and stirred at 60 °C for
27 h. The reaction mixture was quenched with NaHCO3 (1 mL) and
the solvent removed in vacuo. The crude product was dissolved in
MeOH, filtered, and purified by flash chromatography (19:1 CH2Cl2-
MeOH to 4:1 CH2Cl2-MeOH). The product was isolated as a white
foam (1.64 g, 85%). Reaction of the separated diastereomers in the
same manner enabled their spectral characterization. Diastereomer A:
1H NMR (MeOH-d4) δ 7.62-7.59 (m, 2H), 7.44-7.39 (m, 1H), 7.34-
7.29 (m, 2H), 6.19 (dd, 1H, J ) 8.3, 6.0 Hz), 4.31 (m, 1H), 3.90-3.83
(m, 2H), 3.72 (d, 2H, J ) 4.1 Hz), 3.7 (d, 1H, J ) 13.6 Hz), 2.19-
2.10 (m, 1H), 2.0-1.93 (m, 1H), 1.43 (s, 3H); 13C NMR (MeOH-d4)
δ 173.3, 154.1, 150.2, 139.4, 138.5, 130.8, 130.1, 127.2, 87.8, 85.6,
72.9, 63.7, 45.4, 37.5, 21.2; IR (KBr) 3400 (br), 3057, 2922, 1684,
1473, 1436, 1379, 1291, 1220, 1086, 1049, 744 cm-1. Diastereomer
(30) Hildenbrand, K.; Schulte-Frohlinde, D. Int. J. Radiat. Biol. 1997,
71, 377.
(31) Howard, J. A. In Free Radical Chain Reactions, Part II; Kochi, J.
K., Ed.; John Wiley & Son, Inc.: New York, 973.
(32) Greenberg, M. M.; Yoo, D. J.; Goodman, B. K. Nucleosides
Nucleotides 1997, 16, 33.

Release of Superoxide from Nucleoside Peroxyl Radicals
J. Am. Chem. Soc., Vol. 120, No. 20, 1998 4909
1
B: mp 171-173.5 °C; H NMR (MeOH-d4) δ 7.58-7.55 (m, 2H),
were reheated for 1 h if analysis was not completed within 3 h of
silylation. Note: The formic acid treatment steps were omitted for
sample preparation from experiments involving 10. Pyridine was
omitted from the persilylation procedures involving 10.
7.45-7.39 (m, 1H), 7.34-7.29 (m, 2H), 6.30 (dd, 1H, J ) 6.6, 7.8
Hz), 4.31 (m, 1H), 3.71-3.53 (m, 5H), 2.39-2.30 (m, 1H), 2.10-
2.00 (m, 1H), 1.42 (s, 3H); 13C NMR (MeOH-d4) δ 173.2, 154.9, 139.2,
130.9, 130.2, 126.9, 87.5, 84.9, 72.4, 63.4, 45.7, 38.1, 21.1; IR (KBr)
3400 (br), 2923, 2861,1703, 1477, 1438, 1380, 1364, 1286, 1218, 1156,
GC/MS analysis was carried out directly on the BSTFA solution
containing the persilylated nucleosides using a 30 m 5% phenyl methyl
silicone capillary column. Data were collected in the SIM mode for
isotopic studies with a dwell time of 100 ms. Analysis of cleaved
nucleosides from experiments involving 12 was carried out using the
following temperature program: T1, 150 °C (2 min); Tf, 190 °C at 10
°C/min. Ions monitored were 255.1 (thymine), 257.1 (5,6-dihy-
drothymine), 258.1, 345.1 (5,6-dihydro-5-hydroxythymine), and 347.1.
Ions 258.1 and 347.1 were detected to determine 2H and 18O incorpora-
tion, respectively. Analysis of experiments involving 10 was carried
out using the following temperature program: T1, 120 °C (2 min); Tf,
300 °C at 10 °C/min. Ions monitored in bis-silylated 6 were 261.1
(16O) and 263.1 (18O).
Superoxide Detection with Epinephrine. A solution of 10 or 12
(0.1 mM) and (R)-(-)-epinephrine (0.1 mM) was prepared in phosphate
buffer (50 mM, pH 7.5) treated with desferal (1 mM). The epinephrine
was prepared in HCl (10 mM, pH 2). SOD was prepared in phosphate
buffer. The samples were prepared in eppendorf tubes (1 mL total
volume), vortexed, and transferred to a UV cell. The initial absorbance
was measured and the sample photolyzed at 350 nm (one lamp). The
UV absorbance was measured at 480 nm at the specified time intervals.
When experiments were conducted with KO2, a sample of epineph-
rine (0.1 mM) and/or Bu3SnH (0.17 or 1.7 mM) was prepared in 1:1
THF-H2O (1 mM desferal) and transferred to a UV cell. The initial
absorbance was measured at 480 nm. A solution of KO2 (0.1 mM) in
anhydrous DMSO was added dropwise at 0.06 mL/h over 2 h, and the
final absorbance at 480 nm was measured.
Superoxide Detection with Cytochrome c. A solution of 12 (50
µM), cytochrome c (50 µM), and catalase (12 µg) was prepared in
phosphate buffer (50 mM, pH 8) treated with EDTA (0.1 mM). SOD
was prepared in phosphate buffer. The samples were prepared in
eppendorf tubes (1 mL total volume), vortexed, and transferred to a
UV cell. The initial absorbance was measured at 550 nm. The samples
were photolyzed at 350 nm (four lamps) for 2 h, measuring the
absorbance in 10 min intervals. The percentage of reduced cytochrome
c was calculated as previously described.18a Aliquots of the sample
before and after photolysis were analyzed by reversed-phase HPLC as
described above, to determine the extent of conversion of 12 and the
yield of thymidine.
1087, 1048, 1000, 980 cm-1
.
General Procedure for Thermolysis of 12. Stock solutions of 12,
Bu3SnH, and (Bu3Sn)2 were prepared in THF. Stock solutions of water
soluble azo initiator, 14, were prepared in H2O. Thermolysis samples
were prepared with the appropriate reagents having a final volume of
100-400 µL (1:1 or 19:1 THF-H2O). Isotopic studies were carried
out in D2O or H218O. Aerobic experiments were conducted so that O2
was not the limiting reagent. The reaction tubes were left open to the
atmosphere and heated at 37 °C (water bath) or 55 °C (heat block) for
24 h. Anaerobic thermolyses were carried out in sealed tubes, which
were degassed with four freeze-pump-thaw cycles (3 min each).
Anaerobic thermolyses were run at 37 °C for 1.5 h, unless otherwise
specified.
General Procedure for Photolyses. Samples were prepared with
10 (or 12) and the appropriate trap in (1:1 or 19:1) THF-H2O (for 12)
or H2O (for 10). Samples for experiments under anaerobic conditions
were degassed with at least four freeze-pump-thaw cycles (3 min
each). Aerobic samples were left open to the atmosphere. Photolyses
of samples containing 12 were carried out in Pyrex reaction tubes in a
Rayonet photoreactor at 300 nm (four lamps) for 45 min, with an
internal temperature of 25 °C. Photolyses of samples containing 10
were carried out in a similar manner using one lamp (350 nm).
General Workup of Thermolysis and Photolysis Samples. Samples
containing Bu3SnH were transferred to 1 dram vials fitted with Teflon
septa. The tubes were rinsed with dd H2O (200 µL) and THF (2 ×
200 µL), and lyophilized. The residue was subsequently lyophilized
from hexanes (500 µL) to facilitate the removal of tin. The residue
was taken up in dd H2O (200 µL), vortexed, and filtered through a
0.45 µm syringe filter. The vials were rinsed with dd H2O (3 × 200
µL), filtered, and pooled. The internal standard was added, and the
samples were lyophilized on a Savant Speed Vac concentrator. The
residue was taken up in dd H2O (100 µL) for HPLC analysis.
All other samples were transferred directly to eppendorf tubes (1.5
mL). The reaction tubes were washed with dd H2O (200 µL) and THF
(2 × 200 µL). The internal standard was added and the sample
lyophilized. For HPLC analysis the residue was taken up in dd H2O
(100 µL).
HPLC Analysis of Thermolysis and Photolysis Samples. The
products were detected at 205 and/or 254 nm using a gradient solvent
program of H2O and MeOH, unless otherwise specified. For analysis
of experiments involving 12, the gradient was 5% MeOH at t0 for 10
min, linearly to 70% MeOH at 30 min, and linearly to 5% MeOH at
40 min at a flow rate of 1.0 mL/min. For analysis of experiments
involving 10, the gradient was 0% MeOH at t0, linearly to 50% MeOH
at 22 min, and held at this solvent mixture for 13 min, at a flow rate
of 1.0 mL/min.
Mass Spectral Analysis of Persilylated Derivatives. An aliquot
of the HPLC sample (50 µL) was transferred to a 1 mL Kontes conical
reaction vial fitted with a Teflon septa. The sample was treated with
96% formic acid (450 µL) and heated at 100 °C for 1 h. The formic
acid was removed in vacuo, and the samples were lyophilized from
absolute ethanol (3 × 50 µL). The residue was dissolved in BSTFA
(40 µL) and anhydrous pyridine (10 µL). The solution was heated at
100 °C for 2 h. GC/MS analysis was carried out immediately. Samples
Acknowledgment. Financial support of this work from the
National Institutes of Health (Grant GM-54996) and the
Petroleum Research Fund, administered by the American
Chemical Society (Grant AC-30027), is greatly appreciated.
M.M.G. is an Alfred P. Sloan Research Fellow. We thank Brian
K. Goodman for carrying out initial experiments on 3.
Supporting Information Available: 1H and 13C NMR
spectra of the separated diastereomers of 12 and the respective
bis-silyl phenyl selenides and oxidation of epinephrine upon
aerobic photolysis of 12 (9 pages, print/PDF). See any current
masthead page for ordering information and Web access
instructions.
JA973200S