Deconvoluting the reactivity of two intermediates formed from modified pyrimidines

Article abstract of DOI:10.1021/ol401472m

Generation of the 5-(2′-deoxyuridinyl)methyl radical (6) was reexamined. Trapping by 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl confirms that 6 is generated. However, trapping by methoxyamine reveals that the respective carbocation (10) is also produce

Full text of DOI:10.1021/ol401472m

ORGANIC
LETTERS
2013
Vol. 15, No. 14
3618–3621
Deconvoluting the Reactivity of Two
Intermediates Formed from Modified
Pyrimidines
Liwei Weng,† Sonia M. Horvat,‡,§ Carl H. Schiesser,‡,§ and Marc M. Greenberg*,†
Department of Chemistry, Johns Hopkins University, 3400 North Charles Street,
Baltimore Maryland 21218, United States, Australian Research Council Centre of
Excellence for Free Radical Chemistry and Biotechnology, Australia, and
School of Chemistry and Bio21 Molecular Science and Biotechnology Institute,
The University of Melbourne, Victoria, 3010, Australia
mgreenberg@jhu.edu
Received May 24, 2013
ABSTRACT
Generation of the 5-(20-deoxyuridinyl)methyl radical (6) was reexamined. Trapping by 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl confirms that
6 is generated. However, trapping by methoxyamine reveals that the respective carbocation (10) is also produced. Examining the effects of these
traps on products in DNA reveals that the carbocation and not 6 yields interstrand cross-links. Cross-link formation from the carbocation is
consistent with DFT calculations that predict that addition at the N1 position of dA is essentially barrierless.
Aryl sulfides and phenyl selenides are useful as photo-
chemical precursors to alkyl radicals. Substituting the aryl
ring with electron-donating substituents enables one to
photolyze sulfides at >300 nm, which is useful for employ-
ing such precursors in the presence of molecules that
absorb in the ultraviolet region, such as nucleic acids.1,2
Our and other groups have used these types of precursors
to independently generate radicals in nucleosides and
oligonucleotides.3ꢀ8 For instance, we recently reported
using 1 to produce the major hydroxyl radical adduct
of thymidine, 5-hydroxy-5,6-dihydrothymidin-6-yl radical
(2, Scheme 1).9 While studying the photochemistry of 1 we
determined that it yields 2and therespectivecarbocation, 3.
We were unable to distinguish between forming 3 directly
from 1 upon photolysis or via electron transfer within the
caged radical pair (Scheme 1). The formation of 3 led us to
reinvestigate the photochemical generation of the 5-(20-
deoxyuridinyl)methyl radical (6) from similar precursors
(4, 5).10ꢀ13 The results of these studies are the subject of this
report.
The 5-(20-deoxyuridinyl)methyl radical (6) was pre-
viously generated by 350 nm photolysis of the phenyl
† Johns Hopkins University.
‡ Australian Research Council Centre of Excellence for Free Radical
Chemistry and Biotechnology.
(7) Tallman, K. A.; Tronche, C.; Yoo, D. J.; Greenberg, M. M.
J. Am. Chem. Soc. 1998, 120, 4903–4909.
(8) Giese, B.; Beyrich-Graf, X.; Erdmann, P.; Giraud, L.;
§ The University of Melbourne.
(1) Kropp, P. J.; Fryxell, G. E.; Tubergen, M. W.; Hager, M. W.;
Harris, J.; Davis, G.; McDermott, J.; Paul, T.; Tornero-Velez, R. J. Am.
Chem. Soc. 1991, 113, 7300–7310.
€
Imwindelried, P.; Muller, S. N.; Schwitter, U. J. Am. Chem. Soc. 1995,
117, 6146–6147.
(2) Fleming, S. A.; Jensen, A. W. J. Org. Chem. 1996, 61, 7040–7044.
(3) Kim, Y.; Hong, I. S. Bioorg. Med. Chem. Lett. 2008, 18, 5054–
5057.
(4) Al-Oudat, B.; Salyer, A.; Trabbic, K.; Bryant-Friedrich, A.
Bioorg. Med. Chem. Lett. 2013, 23, 854–859.
(5) Giese, B.; Dussy, A.; Meggers, E.; Petretta, M.; Schwitter, U.
(9) San Pedro, J. M. N.; Greenberg, M. M. Org. Lett. 2012, 14, 2866–
2869.
(10) Ding, H.; Majumdar, A.; Tolman, J. R.; Greenberg, M. M.
J. Am. Chem. Soc. 2008, 130, 17981–17987.
(11) Hong, I. S.; Ding, H.; Greenberg, M. M. J. Am. Chem. Soc. 2006,
128, 485–491.
J. Am. Chem. Soc. 1997, 119, 11130–11131.
(6) Giese, B.; Beyrich-Graf, X.; Erdmann, P.; Petretta, M.; Schwitter,
U. Chem. Biol. 1995, 2, 367–375.
(12) Hong, I. S.; Greenberg, M. M. J. Am. Chem. Soc. 2005, 127,
3692–3693.
(13) Hong, I. S.; Greenberg, M. M. Org. Lett. 2004, 6, 5011–5013.
r
10.1021/ol401472m
Published on Web 07/03/2013
2013 American Chemical Society

Scheme 1
Scheme 3
error of one another and provide unequivocal support for
carbocation 10 as its source. Having confirmed that irra-
diation of 4yields 10 in addition to the 5-(20-deoxyuridinyl)-
methyl radical (6), we sought traps that react with only
one of the reactive intermediates to determine the species
responsible for ICL formation in DNA.
Scheme 2
4-Hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (HO-
TEMPO, 13) was chosen as a trap for 6, and 14 (Scheme 4)
was obtained (and fully characterized) following photo-
lysis of 4.15 The yield of 14 (8 ( 0.2%) from the phenyl
selenide (4) under anaerobic conditions varied by less than
2% between 0.5 and 20 mM HO-TEMPO (13). Higher
concentrations (up to 200 mM) of 13 were required under
aerobic conditions to reach a maximum yield of 14 (10.1 (
0.7%), due to competition with O2. Photolysis of aryl
sulfide 5 also produced essentially identical yields of 14
under anaerobic (16.7 ( 0.3%) and aerobic (16.6 ( 0.4%)
conditions. The absolute yields of 14 were greater from 5
than from phenyl selenide 4, suggesting that the former
yields a higher proportion of 6.
selenide (4) and aryl sulfide (5, Scheme 2).10,11,13 Irradiating
either nucleoside yields thymidine in the presence of thiol
under anaerobic conditions, and the expected oxygenated
products (7 and 8) under aerobic conditions. Competition
studies between thiol and oxygen also indicated that per-
oxyl radical formation is reversible.11 However, the mass
balances for these reactions were less than 75%. Photolyz-
ing duplex DNA containing 4 (or 5) yielded interstrand
cross-links (ICLs) with the opposing 20-deoxyadenosine
(Scheme 3). Identical ICLs were produced when DNA
containing 4 (but not 5) was exposed to oxidants such as
NaIO4 or 1O2.10,14 Cross-link formation wasascribed to the
intermediacy of 6 upon photolysis and the quinone methide
type species (9) under oxidizing conditions (Scheme 3).
Thiol quenching of ICL formation was also consistent with
the involvement of 6. However, a thiol could also trap a
carbocation, such as 10 (Scheme 4).
Scheme 4
Indeed, photolysis of phenyl selenide 4 in the presence of
tert-butyl thiol (50 mM) produces the expected radical
trapping products (thymidine, 7, and 8) previously de-
scribed. However, sulfide 11, which was independently
synthesized from previously reported 12, was also pro-
duced. The yields of 11 under aerobic (5.1 ( 0.1%) and
anaerobic (6.2 ( 1%) conditions were within experimental
(14) Hong, I. S.; Greenberg, M. M. J. Am. Chem. Soc. 2005, 127,
10510–10511.
(15) See Supporting Information.
Org. Lett., Vol. 15, No. 14, 2013
3619

Figure 2. Effect of methoxyamine on interstrand cross-link
formation from 5 (17) under aerobic conditions.
Figure 1. Effect of methoxyamine on product distribution from
5 under aerobic conditions.
Sodium azide was examined as a potential trap for the
carbocation (10), as it was successfully employed to probe
for 9. However, the photochemistry of 16 complicated
product analysis. (The azide was partially converted to
the aldehyde under the aqueous conditions, presumably
through the nitrene.) Consequently, we utilized methoxy-
amine to trap 10. Adduct 15 was independently prepared
from 12 and was produced upon irradiation of 4 and 5
under aerobic and anaerobic conditions. Importantly,
while the yield of 15 from 5 increased in the presence of
greater methoxyamine concentration, the amounts of ra-
dical trapping products were unaffected. Under aerobic
conditions the yield of oxygenated products (7, 8) was
unchanged (Figure 1), as was that of thymidine under
anaerobic conditions.15 Similar to the situation involving 1
(Scheme 1), we are unable to distinguish between direct
heterolysis and formation of 10 via electron transfer within
the initially formed radical cage (Scheme 4).
cross-link yield in the presence trap from the ICL formed
in the absence of methoxyamine (eq 1). The ratio of
trapped carbocation (10) to ICL varied linearly with
respect to methoxyamine concentration (Figure 2). The
slope of this line corresponds to the ratio of the rate
constant for trapping of 10 by the nucleophile (kT) to that
for cross-linking (kICL). The average slope of this line
obtained from experiments carried out under aerobic
conditions (4.5 ( 0.2 Mꢀ1, 5 experiments) was within
experimental error of that obtained under anaerobic con-
ditions (3.2 ( 1.2 Mꢀ1, four experiments).15 Extracting a
rate constant for kICL from these data requires approx-
imating kT. Hydrazines and alkoxyamines have similar
nucleophilicities. Rate constants for reactions between
benzyhydrilium ions and hydrazine range from ∼2 ꢁ 102
to 5 ꢁ 103 Mꢀ1 s
.
ꢀ1 16 We expect less conjugated 10 to react
more rapidly than benzyhydrilium ions. However, we do
not know how much faster, making it difficult to estimate
kICL. Carbocations can react orders of magnitude faster
with nucleophiles than the examples cited. Regardless of
the magnitude of kICL, these data clearly indicate that
cross-linking is due to the carbocation (10), which is
formed either upon direct photolysis and/or via electron
transfer within the radical pair, as was previously proposed
for 1 (Scheme 1).
The orthogonal traps, methoxyamine, and 13 were then
tested separately as competitors for ICL formation upon
irradiation of 5 in 17. In the absence of competitor, the
average cross-link yield was independent of O2, as pre-
viously reported and ranged from 21.8 to 23.2% in multi-
ple experiments, each consisting of three replicates.11
Addition of 13 up to 100 mM had no effect on ICL
formation when 17 was photolyzed under aerobic or anae-
robic conditions, indicating that the 5-(20-deoxyuridinyl)-
methyl radical (6) is not responsible for cross-link
formation.15 However, 14 was detected by UPLC follow-
ing enzyme digestion of the photolysate, confirming for-
mation of 6.15 In contrast, methoxyamine competed with
ICL formation, and 15 was detected in digested samples
by UPLC.15 In a typical experiment the cross-link yield
decreased from 22.6 ( 0.7% to 15.8 ( 0.2% from 0 to
100 mM of methoxyamine. The amount of 10 trapped
by methoxyamine was determined by subtracting the
In order to shed further light on this intriguing reaction,
we examined the fundamental radical reaction between
adenine 17 and the uridinylmethyl radical 18 to afford
adduct 19 using computational techniques (Scheme 5).
This study showed that the energy barrier (ΔE‡) exceeds
50 kJ molꢀ1 at all levels of theory employed, with the M06-
2X/6-311G (d,p) level providing a value of 58.8 kJ molꢀ1
.
Several other levels of theory provided similar data.15 It is
interesting to note the M06-2X/6-311G (d,p) optimized
transition state (21) for the formation of 19 is assisted by
hydrogen bonding (Figure 3). However, given that the
energy barrier is in excess of 50 kJ molꢀ1, it is unlikely that
3
the cross-linking reaction that occurs when DNA contain-
ing 4 or 5 is photolyzed is radical in nature.
(16) Nigst, T. A.; Antipova, A.; Mayr, H. J. Org. Chem. 2012, 77,
8142–8155.
3620
Org. Lett., Vol. 15, No. 14, 2013

Scheme 5
Figure 3. M06-2X/6-311G (d,p) optimized structure of the
transition state 21 involved in the reaction of 17 with 18.
Natural bond orbital analysis at the BHandHLYP/
6-311G (d,p) level of theory on 21 reveals significant
involvement of the lone pair on N1.15,21 Indeed the LPN f
5-(20-deoxyuridinyl)methyl radical (6) is generated via
hole migration in A T sequences, but cross-links are not
3
observed.17,18 These experiments also suggest that other
purported precursors to the 5-(20-deoxyuridinyl)methyl
radical (6) and related molecules that produce ICLs may
also generate 10 and that other phenyl selenides used to
generate radicals in DNA produce the corresponding
carbocations.3,4,6,19 Finally, the recent report by Li show-
ing that 6 can be oxidized raises the possibility that the
radical may indirectly lead to interstrand cross-links.20
SOMO interaction is calculated to be worth 1584 kJ molꢀ1
,
while the SOMO f π*NC is worth 1325 kJ molꢀ1. This
indicates that resonance structure 18b is a significant con-
tributor to this chemistry and that the key step is one in
which the radical predominately acts as a Michael acceptor
for the N1 nitrogen of 17.
With this in mind, it came as no surprise that we were
unable to locate a transition state for the reaction of the
analogous cation 20 with 17. This reaction is predicted to
be barrierless in the gas phase, with an exothermicity of
190.8 kJ molꢀ1 at the B3LYP/6-311G(d,p) level of theory,
consistent with the above experiments indicating that the
observed cross-links are produced by the carbocation.
Determination of the fact that carbocation 10 is respon-
sible for cross-linking when DNA containing 4 or 5 is
photolyzed clarifies and reconciles a number of observa-
tions in the literature. Cross-linking from 10 provides a
simple explanation for why ICL formation is independent
of O2 since the ionic intermediate (10) does not react with
O2. The formation of identical products from photolysis or
oxidation of DNA containing 4 is also consistent with the
intermediacy of 10, which the above computational studies
indicate should face a significantly lower barrier to adding
to N1 of dA than radical 6. The formation of ICLs from 10
also reconciles observations in the literature in which the
Acknowledgment. We are grateful for support of this
research by the National Institute of General Medical
Sciences (GM-054996) and the Australian ResearchCoun-
cil through the Centers of Excellence Scheme. We thank
Professor Lei Li (IUPUI) for sharing his results with us
prior to publication.
Supporting Information Available. All experimental
procedures. Spectroscopic characterization of all new
compounds. Effect of methoxyamine on products from
5 under anaerobic conditions, and effect 13 and methoxy-
amine (under anaerobic conditions) on ICL formation.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(19) Kuang, Y.; Sun, H.; Blain, J. C.; Peng, X. Chem.;Eur. J. 2012,
18, 12609–12613.
(20) Lin, G.; Li, L. Angew. Chem., Int. Ed. 2013, 52, 5594–5598.
(21) BHandHLYP/6-311G(d,p) was chosen for consistency with
previous calculations. See: Schiesser, C. H.; Wille, U.; Matsubara, H.;
Ryu, I. Acc. Chem. Res. 2007, 40, 303–313.
(17) Barnett, R. N.; Joseph, J.; Landman, U.; Schuster, G. B. J. Am.
Chem. Soc. 2013, 135, 3904–3914.
(18) Ghosh, A.; Joy, A.; Schuster, G. B.; Douki, T.; Cadet, J. Org.
Biomol. Chem. 2008, 6, 916–928.
The authors declare no competing financial interest.
Org. Lett., Vol. 15, No. 14, 2013
3621