Quantification of the 2-deoxyribonolactone and nucleoside 5′-aldehyde products of 2-deoxyribose oxidation in DNA and cells by isotope-dilution gas chromatography mass spectrometry: Differential effects of γ-radiation and Fe2+-EDTA

Article abstract of DOI:10.1021/ja910928n

The oxidation of 2-deoxyribose in DNA has emerged as a critical determinant of the cellular toxicity of oxidative damage to DNA, with oxidation of each carbon producing a unique spectrum of electrophilic products. We have developed and validated an isotope-dilution gas chromatography-coupled mass spectrometry (GC-MS) method for the rigorous quantification of two major 2-deoxyribose oxidation products: the 2-deoxyribonolactone abasic site of 1′-oxidation and the nucleoside 5′-aldehyde of 5′-oxidation chemistry. The method entails elimination of these products as 5-methylene-2(5H)-furanone (5MF) and furfural, respectively, followed by derivatization with pentafluorophenylhydrazine (PFPH), addition of isotopically labeled PFPH derivatives as internal standards, extraction of the derivatives, and quantification by GC-MS analysis. The precision and accuracy of the method were validated with oligodeoxynucleotides containing the 2-deoxyribonolactone and nucleoside 5′-aldehyde lesions. Further, the well-defined 2-deoxyribose oxidation chemistry of the enediyne antibiotics, neocarzinostatin and calicheamicin γ₁^I, was exploited in control studies, with neocarzinostatin producing 10 2-deoxyribonolactone and 300 nucleoside 5′-aldehyde per 10⁶ nt per μM in accord with its established minor 1′- and major 5′-oxidation chemistry. Calicheamicin unexpectedly caused 1′-oxidation at a low level of 10 2-deoxyribonolactone per 10⁶ nt per μM in addition to the expected predominance of 5′-oxidation at 560 nucleoside 5′-aldehyde per 10⁶ nt per μM. The two hydroxyl radical-mediated DNA oxidants, γ-radiation and Fe²⁺-EDTA, produced nucleoside 5′-aldehyde at a frequency of 57 per 10⁶ nt per Gy (G-value 74 nmol/J) and 3.5 per 10⁶ nt per μM, respectively, which amounted to 40% and 35%, respectively, of total 2-deoxyribose oxidation as measured by a plasmid nicking assay. However, γ-radiation and Fe²⁺-EDTA produced different proportions of 2-deoxyribonolactone at 7% and 24% of total 2-deoxyribose oxidation, respectively, with frequencies of 10 lesions per 10⁶ nt per Gy (G-value, 13 nmol/J) and 2.4 lesions per 10⁶ nt per μM. Studies in TK6 human lymphoblastoid cells, in which the analytical data were corrected for losses sustained during DNA isolation, revealed background levels of 2-deoxyribonolactone and nucleoside 5′-aldehyde of 9.7 and 73 lesions per 10⁶ nt, respectively. γ-Irradiation of the cells caused increases of 0.045 and 0.22 lesions per 10⁶ nt per Gy, respectively, which represents a ～250-fold quenching effect of the cellular environment similar to that observed in previous studies. The proportions of the various 2-deoxyribose oxidation products generated by γ-radiation are similar for purified DNA and cells. These results are consistent with solvent exposure as a major determinant of hydroxyl radical reactivity with 2-deoxyribose in DNA, but the large differences between γ-radiation and Fe²⁺-EDTA suggest that factors other than hydroxyl radical reactivity govern DNA oxidation chemistry.

Full text of DOI:10.1021/ja910928n

Published on Web 04/08/2010
Quantification of the 2-Deoxyribonolactone and Nucleoside
5′-Aldehyde Products of 2-Deoxyribose Oxidation in DNA and
Cells by Isotope-Dilution Gas Chromatography Mass
Spectrometry: Differential Effects of γ-Radiation and
Fe2+-EDTA
Wan Chan,† Bingzi Chen,†,‡ Lianrong Wang,† Koli Taghizadeh,§
Michael S. Demott,† and Peter C. Dedon*,†,§
Department of Biological Engineering and Center for EnVironmental Health Sciences,
Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
Received January 6, 2010; E-mail: pcdedon@mit.edu
Abstract: The oxidation of 2-deoxyribose in DNA has emerged as a critical determinant of the cellular
toxicity of oxidative damage to DNA, with oxidation of each carbon producing a unique spectrum of
electrophilic products. We have developed and validated an isotope-dilution gas chromatography-coupled
mass spectrometry (GC-MS) method for the rigorous quantification of two major 2-deoxyribose oxidation
products: the 2-deoxyribonolactone abasic site of 1′-oxidation and the nucleoside 5′-aldehyde of 5′-oxidation
chemistry. The method entails elimination of these products as 5-methylene-2(5H)-furanone (5MF) and
furfural, respectively, followed by derivatization with pentafluorophenylhydrazine (PFPH), addition of
isotopically labeled PFPH derivatives as internal standards, extraction of the derivatives, and quantification
by GC-MS analysis. The precision and accuracy of the method were validated with oligodeoxynucleotides
containing the 2-deoxyribonolactone and nucleoside 5′-aldehyde lesions. Further, the well-defined
I
2-deoxyribose oxidation chemistry of the enediyne antibiotics, neocarzinostatin and calicheamicin γ1 , was
exploited in control studies, with neocarzinostatin producing 10 2-deoxyribonolactone and 300 nucleoside
5′-aldehyde per 106 nt per µM in accord with its established minor 1′- and major 5′-oxidation chemistry.
Calicheamicin unexpectedly caused 1′-oxidation at a low level of 10 2-deoxyribonolactone per 106 nt per
µM in addition to the expected predominance of 5′-oxidation at 560 nucleoside 5′-aldehyde per 106 nt per
µM. The two hydroxyl radical-mediated DNA oxidants, γ-radiation and Fe2+-EDTA, produced nucleoside
5′-aldehyde at a frequency of 57 per 106 nt per Gy (G-value 74 nmol/J) and 3.5 per 106 nt per µM,
respectively, which amounted to 40% and 35%, respectively, of total 2-deoxyribose oxidation as measured
by a plasmid nicking assay. However, γ-radiation and Fe2+-EDTA produced different proportions of
2-deoxyribonolactone at 7% and 24% of total 2-deoxyribose oxidation, respectively, with frequencies of 10
lesions per 106 nt per Gy (G-value, 13 nmol/J) and 2.4 lesions per 106 nt per µM. Studies in TK6 human
lymphoblastoid cells, in which the analytical data were corrected for losses sustained during DNA isolation,
revealed background levels of 2-deoxyribonolactone and nucleoside 5′-aldehyde of 9.7 and 73 lesions per
106 nt, respectively. γ-Irradiation of the cells caused increases of 0.045 and 0.22 lesions per 106 nt per Gy,
respectively, which represents a ∼250-fold quenching effect of the cellular environment similar to that
observed in previous studies. The proportions of the various 2-deoxyribose oxidation products generated
by γ-radiation are similar for purified DNA and cells. These results are consistent with solvent exposure as
a major determinant of hydroxyl radical reactivity with 2-deoxyribose in DNA, but the large differences
between γ-radiation and Fe2+-EDTA suggest that factors other than hydroxyl radical reactivity govern
DNA oxidation chemistry.
mation, cancer and degenerative diseases.1-3 While studies of
the chemistry of nucleobase damage have dominated the
literature, there is growing evidence that oxidation of 2-deoxy-
ribose in DNA plays a critical role in the toxicity of oxidative
stress, whether as individual lesions, as part of complex DNA
Introduction
DNA damage caused by a variety of endogenous oxidants
and reactive chemicals generated by the innate immune system
may play a significant role in the pathophysiology of inflam-
† Department of Biological Engineering, Massachusetts Institute of
Technology.
(1) Klaunig, J. E.; Kamendulis, L. M. Ann. ReV. Pharmacol. Toxicol. 2004,
44, 239.
‡ Current address: S & T Global, Inc., Woburn, MA 01801.
§ Center for Environmental Health Sciences, Massachusetts Institute of
Technology.
(2) Dedon, P. C.; Tannenbaum, S. R. Arch. Biochem. Biophys. 2004, 423,
12.
(3) Beckman, K. B.; Ames, B. N. Phys. ReV. 1998, 78, 547.
9
10.1021/ja910928n  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 6145–6153 6145

A R T I C L E S
Chan et al.
Scheme 1. Elimination Products of the 2-Deoxyribonolactone and Nucleoside 5′-Aldehyde Products of 2-Deoxyribose in DNA
lesions with closely opposed strand breaks and oxidized abasic
sites,4,5 or as participants in protein-DNA cross-links, and
protein and DNA adducts.6 For hydroxyl radical and other
reactive oxygen and nitrogen species, the oxidation of each of
carbon in 2-deoxyribose in DNA occurs with an initial hydrogen
atom abstraction to form a carbon-centered radical that, under
normoxic conditions, adds molecular oxygen to form a peroxyl
radical;6 conditions of low oxygen concentration promote the
formation of cyclic nucleotide products such as the 5′,8-
cyclopurine product of 5′-oxidation.7 The degradation of the
peroxyl radical results in the formation of a variety of electro-
philic and genotoxic products that differ for each position.6 We
have previously quantified hydroxyl radical-induced formation
of several of these products, including the 3′-phosphoglycolate
and 2-deoxypentos-4-ulose abasic site of 4′-oxidation,8 the 5′-
(2-phosphoryl-1,4-dioxobutane) residue of 5′-oxidation9 and the
3-phosphoglycolaldehyde that arises in 3′-oxidation.6,10-12
The present studies focus on the oxidation of the 1′- and 5′-
positions in 2-deoxyribose in DNA, as shown in Scheme 1.
Oxidation of the 1′-carbon produces a 2′-deoxyribonolactone
abasic site that undergoes a rate-limiting ꢀ-elimination reaction
to form a butenolide species with a half-life of 20 h in single-
stranded DNA (32-54 h in duplex DNA) followed by a rapid
δ-elimination to release 5-methylene-2(5H)-furanone (5MF).13-16
An analogous ꢀ-/δ-elimination reaction with the 3-oxonucleotide
product of 3′-oxidation yields 2-methylene-3(2H)-furanone
(2MF).17,18 The chemistry of 5′-oxidation partitions to form two
sets of products: a 3′-formylphosphate-ended fragment and a
2-phosphoryl-1,4-dioxo-2-butane residue that undergoes ꢀ-e-
limination to form a trans-butenedialdehyde species; or a
nucleoside 5′-aldehyde residue that can undergo ꢀ-/δ-elimina-
tions to produce furfural.19-21 Under conditions of low oxygen
concentration, the 5′-radical can react to form 5′,8-cyclopurine
products.7
The goals of the present studies were two-fold. The first was
to develop a method to rigorously quantify the 5MF and furfural
elimination products of the 2-deoxyribonolactone and nucleoside
5′-aldehyde, respectively. With formation half-lives on the order
of 25-100 h at 37 °C,13,22 these elimination products should
arise spontaneously in cells and tissues and thus have potential
for use as biomarkers of oxidative stress and as probes for the
fate of DNA damage products in biological systems. The second
goal was to rigorously define the contributions of these
2-deoxyribose oxidation products to the spectrum of products
arising from hydroxyl radical reactions with DNA. While
previous studies of the nucleoside 5′-aldehyde have been
descriptive in nature, attempts to quantify the 2-deoxyribono-
lactone have led to the conclusion that the 2-deoxyribonolactone
represents one of the most abundant DNA oxidation products
caused by hydroxyl radicals.15,16,23 This stands in contrast to
studies of 2-deoxyribose oxidation by putative hydroxyl radicals
generated by Fe2+-EDTA, in which deuterium isotope effects
indicated relative reactivity in the order 5′ > 4′ > 3′ ≈ 2′ ≈
1′.24 We sought to define the absolute and relative contributions
of 2-deoxyribonolactone and nucleoside 5′-aldehyde to the
chemistry of 2-deoxyribose oxidation in DNA by hydroxyl
radical, with the conclusion that the level of 2-deoxyribono-
lactone varies significantly for Fe2+-EDTA and γ-radiation.
We have also performed studies in irradiated human cells, with
(4) Povirk, L. F.; Goldberg, I. H. Proc. Natl. Acad. Sci. U.S.A. 1985, 82,
3182.
(5) Weinfeld, M.; Rasouli-Nia, A.; Chaudhry, M. A.; Britten, R. A. Radiat.
Res. 2001, 156, 584.
(6) Dedon, P. C. Chem. Res. Toxicol. 2008, 21, 206.
(7) Boussicault, F.; Kaloudis, P.; Caminal, C.; Mulazzani, Q. G.; Chat-
gilialoglu, C. J. Am. Chem. Soc. 2008, 130, 8377.
(8) Chen, B.; Zhou, X.; Taghizadeh, K.; Chen, J.; Stubbe, J.; Dedon, P. C.
Chem. Res. Toxicol. 2007, 20, 1701.
(17) Bryant-Friedrich, A. C. Org. Lett. 2004, 6, 2329.
(18) Lahoud, G. A.; Hitt, A. L.; Bryant-Friedrich, A. C. Chem. Res. Toxicol.
2006, 19, 1630.
(9) Chen, B.; Bohnert, T.; Zhou, X.; Dedon, P. C. Chem. Res. Toxicol.
2004, 17, 1406.
(10) Greenberg, M. M.; Kreller, C. R.; Young, S. E.; Kim, J. Chem. Res.
Toxicol. 2006, 19, 463.
(19) Pratviel, G.; Pitie, M.; Bernadou, J.; Meunier, B. Angew. Chem., Int.
Ed. Engl. 1991, 30, 702.
(11) Collins, C.; Awada, M. M.; Zhou, X.; Dedon, P. C. Chem. Res. Toxicol.
2003, 16, 1560.
(20) Barciszewski, J.; Siboska, G. E.; Pederson, B. O.; Clark, B. F. C.;
Rattan, S. I. S. Biochem. Biophys. Res. Commun. 1997, 238, 317.
(21) Pitie, M.; Burrows, C. J.; Meunier, B. Nucleic Acids Res. 2000, 28,
4856.
(12) Collins, C.; Zhou, X.; Wang, R.; Barth, M. C.; Jiang, T.; Coderre,
J. A.; Dedon, P. C. Radiat. Res. 2005, 163, 654.
(13) Zheng, Y.; Sheppard, T. L. Chem. Res. Toxicol. 2004, 17, 197.
(14) Hwang, J. T.; Tallman, K. A.; Greenberg, M. M. Nucleic Acids Res.
1999, 27, 3805.
(22) Kodama, T.; Greenberg, M. M. J. Org. Chem. 2005, 70, 9916.
(23) Roginskaya, M.; Bernhard, W. A.; Marion, R. T.; Razskazovskiy, Y.
Radiat. Res. 2005, 163, 85.
(15) Sato, K.; Greenberg, M. E. J. Am. Chem. Soc. 2005, 127, 2806.
(16) Xue, L.; Greenberg, M. M. Angew. Chem., Int. Ed. 2007, 46, 561.
(24) Balasubramanian, B.; Pogozelski, W. K.; Tullius, T. D. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 9738.
9
6146 J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010

Quantifying 2-Deoxyribonolactone and Nucleoside 5′-Aldehyde
A R T I C L E S
Scheme 2. Reaction of 5MF and Furfural with
Pentafluorophenylhydrazine
made to synthesize the PFPH derivative of 2MF, with the latter
synthesized from thymidine by the method of Stubbe and co-
workers.26 A single product was identified by GC-MS, with a
molecular ion at m/z 276 as the base peak, together with
characteristic fragment ions at m/z 93 [C3F3]+, 117 [C5F3]+, 148
[C6F4]+, and 167 [C6F5]+. However, this conjugate proved to
be too unstable for use as a standard, and we could not identify
a species with the same retention time and m/z value in DNA
samples exposed to γ-radiation or Fe2+-EDTA (data not
shown), perhaps due to the instability of 2MF-PFPH derivative.
As shown in Supporting Information Figure 1, the chromato-
graphically well resolved 1a, 1b and 2 all produced a strong
molecular ion signal at m/z 276 that was used for subsequent
quantitative analyses.
The next step was to define analytical parameters for the
GC-MS quantification of 1a, 1b, and 2. Following definition
of the GC retention times and mass spectral behavior of 1a,
1b, and 2, calibration curves were prepared by mixing heated,
PFPH-treated samples of calf thymus DNA with fixed amounts
of isotopically labeled 1a, 1b, and 2 and variable amounts of
unlabeled forms and extracting 1a, 1b, and 2 into dichlo-
romethane. Plots of the peak area ratios for unlabeled and labeled
PFPH derivatives were linear with slopes of 0.0016 pmol-1 and
0.0033 pmol-1 for 1a+1b and 2, respectively (see Supporting
Information Figure 2). Lines fitted by linear regression do not
pass through the origin due to the presence of background levels
of DNA oxidation products in the calf thymus DNA and other
possible sample processing contaminants. The background level
is equivalent to 1.65 pmol of 1a+1b and 0.485 pmol of 2 in
250 µg of DNA or 2.2 1a+1b per 106 nt and 6.3 2 per 107 nt.
These values represent the practical limit of quantification.
The final step was to validate the analytical method and
determine the overall efficiency of the elimination, derivatization
and extraction steps using oligodeoxynucleotides containing
defined quantities of 2-deoxyribonolactone and nucleoside 5′-
aldehyde damage products. A 17-mer oligodeoxynucleotide, 5′-
TGTGCCXAACTTACCGT-3′, containing 2-deoxyribonolac-
tone at X (3) was prepared with >92% purity by UV irradiation
of a nitrobenzyl cyanohydrin nucleoside-containing precursor,
as described by Zheng and Sheppard13 (Supporting Information
Figure 3). A 3-mer oligodeoxynucleotide, 5′-TGC-3′, with a
nucleoside 5′-aldehyde terminus at T (4) was isolated in >95%
purity by HPLC purification of the major product of a reaction
of a self-complementary, duplex oligodeoxynucleotide, 5′-
GCATGC-3′, with the enediyne neocarzinostatin (Supporting
Information Figure 3), as described by Sugiyama et al.27 The
nucleoside 5′-aldehyde-containing single-strand oligodeoxy-
nucleotide had a half-life of 34 h at ambient temperature and
pH 7.4 (Supporting Information Figure 4), which is in reasonable
agreement with the half-life in double-stranded DNA of 101 h
at 37 °C determined by Greenberg and co-workers22 given the
stability conferred by duplex DNA for many damage products
(e.g., 2-deoxyribonolactone13).
the observation of similar proportions of 2-deoxyribose oxidation
products arising in Vitro and in cells, and a ∼250-fold protective
effect against γ-radiation-induced 2-deoxyribose oxidation in
DNA provided by the cellular environment.
Results
Development and Optimization of a GC-MS Method for
the Quantification of 2-Deoxyribonolactone and Nucleoside
5′-Aldehyde in Oxidized DNA. As shown in Schemes 1 and 2,
the overall strategy of quantifying the 2-deoxyribonolactone and
nucleoside 5′-aldehyde products entails: (1) their elimination
as 5MF and furfural, respectively, from oxidized DNA (Scheme
1); (2) derivatization of 5MF and furfural with pentafluorophe-
nylhydrazine (PFPH) (Scheme 2); and (3) analysis of the
derivatives by gas chromatography-coupled mass spectrometry
(GC-MS) using isotopomeric internal standards.
The first step in method development involved synthesis and
characterization of isotopically labeled and unlabeled forms of
the PFPH derivatives of 5MF and furfural for use as standards.
The reaction of PFPH with furfural led to the expected geometric
isomers of the resulting hydrazone (1a, 1b in Scheme 2), as
observed previously by Ho and Yu.25 1a and 1b were well-
resolved chromatographically (Figure 1) and showed identical
electron ionization (EI) mass spectra (Supporting Information
Figure 1A), with the combined signal for both isomers used to
quantify the furfural-PFPH derivative from DNA samples.
Reaction of PFPH with 5MF afforded an unexpected single
product, 6-methyl-2-(perfluorophenyl)pyridazin-3(2H)-one (2),
the structure of which is shown in Scheme 2. An attempt was
These well-characterized standards were used to validate the
analytical method by spiking 250 µg of calf thymus DNA with
3 (88 pmol) or 4 (293 pmol) followed by heat and polylysine-
catalyzed elimination of 5MF and furfural, respectively, de-
rivatization with PFPH and GC-MS analysis of methylene
chloride extracts, as described in Experimental Methods. The
(25) Ho, S. S.; Yu, J. Z. EnViron. Sci. Technol. 2004, 38, 862.
(26) Harris, G.; Ator, M.; Stubbe, J. Biochemistry 1984, 23, 5214.
(27) Sugiyama, H.; Fujiwara, T.; Kawabata, H.; Yoda, N.; Hirayama, N.;
Saito, I. J. Am. Chem. Soc. 1992, 114, 5573.
Figure 1. GC-MS chromatogram of 1a, 1b, and 2 with selected ion
monitoring at m/z 276.
9
J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010 6147

A R T I C L E S
Chan et al.
Figure 3. Quantification of 5MF (b) and furfural (O) in DNA exposed to
γ-radiation (A), Fe2+-EDTA (B). DNA was treated with the oxidants and
quantified by GC-MS as described in the Experimental Methods. The data
represent mean ( SD for three independent experiments. Fitting the data
by linear regression yielded lines with the following equations: (A) 5MF:
y ) 9.7x + 59 (r2 ) 0.98); furfural: y ) 57x + 268 (r2 ) 0.99); (B) 5MF:
y ) 2.4x + 30 (r2 ) 0.97); furfural: y ) 3.5x + 1.7 (r2 ) 0.94).
Figure 2. Quantification of 5MF (b) and furfural (O) in DNA exposed to
neocarzinostatin (A), and calicheamicin γ1I (B). DNA was treated with the
oxidants and quantified by GC-MS as described in the Experimental
Methods. The data represent mean ( SD for three independent experiments.
Fitting the data by linear regression yielded lines with the following
equations: (A) 5MF: y ) 8.4x + 11 (r2 ) 0.96); furfural: y ) 299x - 94
(r2 ) 0.99); (B) 5MF: y ) 10x + 0.26 (r2 ) 1.0); furfural: y ) 564x + 7.0
(r2 ) 0.99).
5′-chemistry with 564 ((49) nucleoside 5′-aldehyde residues
per 106 nt per µM (Figure 2B). This is consistent with selectivity
rather than specificity in the oxidation of 2-deoxyribose by
enediynes in DNA, and with the minor-groove location of 1′-,
4′-, and 5′-hydrogen atom targets for all of the enediynes.6
Studies with the hydroxyl radical generators γ-radiation and
Fe2+-EDTA revealed dose-dependent formation of 2-deoxyri-
bonolactone and nucleoside 5′-aldehyde damage products. As
shown in Figure 3, γ-radiation produced 10 ((0.75) 2-deoxy-
ribonolactone and 57 ((3.7) nucleoside 5′-aldehyde residues
per 106 nt per Gy, which represents radiation chemical yields
(G-values) of 13 and 74 nmol/J, respectively. Fe2+-EDTA
produced 2.4 ((0.20) 2-deoxyribonolactone and 3.5 ((0.46)
nucleoside 5′-aldehyde per 106 nt per µM.
A more informative context for these results involves their
expression as a percentage of total number of 2-deoxyribose
oxidation events. This is achieved by quantifying total 2-deox-
yribose oxidation by plasmid topoisomer analysis, which is a
well established technique that accounts for all 2-deoxyribose
oxidation events by exploiting the conversion of supercoiled
plasmid DNA to nicked and linear forms following direct strand
breaks and, after derivatization with agents such as putrescine,
yields of 1a+1b and 2 were found to be 43 ( 2.4% and 21 (
0.7% of the theoretical value, respectively, which represents
the overall efficiency for the analytical method. Measurements
of furfural and 5MF were thus corrected by a factor of 2.3 and
4.8, respectively, to arrive at the quantity of nucleoside
5′-aldehyde and 2-deoxyribonolactone, respectively.
Quantification of 2-Deoxyribonolactone and Nucleoside
5′-Aldehyde in DNA Oxidized by Enediyne Antibiotics,
γ-Radiation, and Fe2+-EDTA. The validated method was
applied to quantify 2-deoxyribonolactone and nucleoside 5′-
aldehyde residues in DNA and cells exposed to the enediyne
antibiotics, neocarzinostatin and calicheamicin, as positive
controls and to γ-radiation and Fe2+-EDTA. Neocarzinostatin
is known to cause 1′-, 4′-, and 5′-oxidation of 2-deoxyribose in
DNA,28 which is consistent with the observation of 8.4 ((0.99)
2-deoxyribonolactone and 298 ((14) nucleoside 5′-aldehyde
residues per 106 nt per µM (Figure 2A). Calicheamicin γ1I has
been proposed to produce only 4′- and 5′-oxidation,28 yet we
observed a low level of 10 ((0.34) 2-deoxyribonolactone
residues per 106 nt per µM in addition to the expected
(28) Dedon, P. C.; Goldberg, I. H. Chem. Res. Toxicol. 1992, 5, 311.
9
6148 J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010

Quantifying 2-Deoxyribonolactone and Nucleoside 5′-Aldehyde
A R T I C L E S
oxidized abasic sites.29-38 We previously established the
frequency of total 2-deoxyribose oxidation at 141 events per
106 nt per Gy for γ-irradiation and at 10 events per 106 nt per
µM for Fe2+-EDTA using an identical irradiation source, DNA
concentration, buffer and temperature conditions employed for
the present studies.8,9,11,12 Linear regression analysis of plots
of the quantities of 2-deoxyribonolactone and nucleoside 5′-
aldehyde lesions versus calculated total 2-deoxyribose oxidation
events at the various doses of γ-radiation revealed that 2-deoxy-
ribonolactone accounted for ∼7% of γ-radiation-induced 2-deox-
yribose oxidation (y ) 0.069x + 0.059, r2 ) 0.98), while
nucleoside 5′-aldehyde residues accounted for ∼40% of the
damage (y ) 0.40x + 0.27, r2 ) 0.99). On the other hand, ∼24%
of the 2-deoxyribose oxidation by Fe2+-EDTA was composed
of 2-deoxyribonolactone (y ) 0.24x + 0.03, r2 ) 0.97), while
∼35% was nucleoside 5′-aldehyde residues (y ) 0.35x +
0.0017, r2 ) 0.94).
Quantification of 2-Deoxyribonolactone and Nucleoside 5′-
Aldehyde in DNA in Human Cells Exposed to γ-Radiation. To
assess the effects of the cellular environment on the formation
of 2-deoxyribonolactone and nucleoside 5′-aldehyde in oxidized
DNA, TK6 human lymphoblastoid cells were γ-irradiated and
the quantity of 5MF and furfural determined in isolated DNA.
An important facet of this experiment involved an assessment
of the effect of the DNA isolation process on the level of 5MF
and furfural measured in the isolated DNA. To this end, a control
experiment was performed in which purified and irradiated
DNA, containing a defined quantity of 2-deoxyribonolactone
and nucleoside 5′-aldehyde, was added to cells during the DNA
isolation process and the quantities of 5MF and furfural
determined in the total isolated DNA. The measured quantities
of 5MF and furfural were expressed relative to the quantity of
purified and irradiated DNA that was added to the cells, with
the background contribution of 5MF and furfural from the
cellular genomic DNA removed by analysis of the slope of a
γ-radiation dose-response curve. As shown in Supporting
Information Figure 6, linear regression analysis revealed ef-
ficiencies of 5.3 5MF per 106 nt per Gy and 44 furfural per 106
nt per Gy. These yields correspond to 53% and 77%, respec-
tively, of the yields obtained with irradiated purified DNA used
to spike the cell samples (10 5MF and 57 furfural per 106 nt
per Gy). These results reveal a tangible loss of 2-deoxyribono-
lactone and nucleoside 5′-aldehyde during the cell processing
steps and provide correction factors of 1.9 and 1.3 for cellular
yields of 5MF and furfural, respectively, in addition to the
correction factors of 4.8 and 2.3, respectively, for the analytical
method.
Figure 4. Formation of 5MF (b) and furfural (O) in γ-irradiated human
TK6 cells. Cells were exposed to γ-radiation at ambient temperature and
processed for GC-MS analysis as described in the Experimental Methods.
Data were corrected for losses during DNA isolation and sample processing
for GC-MS analysis as described in the text and background levels were
subtracted. The data represent means ( SD for three independent experi-
ments. Fitting the data by linear regression yielded lines with the following
equations: 5MF: y ) 0.045x + 1.4 (r2 ) 0.99); furfural: y ) 0.22x + 1.1
(r2 ) 0.99). The slopes are statistically significant at p < 0.003, and the
results for 5MF and furfural at 300 and 500 Gy are significantly different
from each other by t test with p < 0.0045 and p < 0.0001, respectively.
The results of the analysis of 2-deoxyribonolactone and
nucleoside 5′-aldehyde in irradiated TK6 cells are shown in
Figure 4. After correction for losses during DNA isolation and
processing, we observed background levels of 2-deoxyribono-
lactone and nucleoside 5′-aldehyde of 9.7 ((0.22) and 73 ((1.5)
lesions per 106 nt, respectively. Exposure of TK6 cells to
γ-radiation resulted in corrected damage frequencies of 0.045
((0.0028) 5MF per 106 nt per Gy and 0.22 ((0.012) furfural
per 106 nt per Gy (Figure 4).
Discussion
The emerging appreciation for the biological importance of
2-deoxyribose oxidation in DNA has motivated efforts to define
the quantities of the various DNA-bound and diffusible products,
in an effort to identify the chemistry relevant to the biological
situation.6 We have now expanded the repertoire of analytical
methods with the development of an accurate, precise and
sensitive method to quantify two major products of 2-deoxyri-
bose oxidation: the 2-deoxyribonolactone abasic site arising from
1′-oxidation of DNA and the nucleoside 5′-aldehyde residue
arising from 5′-oxidation of DNA. This approach exploits the
simple chemical release of quantifiable electrophiles from
oxidized DNA. The application of this method has revealed
important quantitative features of 2-deoxyribose oxidation in
purified DNA and in cells.
The isotope-dilution GC-MS approach developed here
represents the most quantitative and chemically specific method
for analysis of both the nucleoside 5′-aldehyde and 2-deoxyri-
bonolactone products of 2-deoxyribose oxidation in DNA. We
have established the overall efficiency of the elimination,
derivatization and analytical steps using chemically and quan-
titatively well-defined substrates, which provides a critical
correction for artifacts arising during sample processing. The
accuracy of the methods was further validated in studies with
the DNA-cleaving enediyne antibiotics, neocarzinostatin and
calicheamicin. Both drugs undergo thiol-induced cyclization to
(29) Dedon, P. C.; Goldberg, I. H. Biochemistry 1992, 31, 1909.
(30) Dedon, P. C.; Salzberg, A. A.; Xu, J. Biochemistry 1993, 32, 3617.
(31) Yu, L.; Golik, J.; Harrison, R.; Dedon, P. J. Am. Chem. Soc. 1994,
116, 9733.
(32) Yu, L.; Mah, S.; Otani, T.; Dedon, P. J. Am. Chem. Soc. 1995, 117,
8877.
(33) LaMarr, W. A.; Sandman, K. M.; Reeve, J. N.; Dedon, P. C. Chem.
Res. Toxicol. 1997, 10, 1118.
(34) LaMarr, W. A.; Nicolaou, K. C.; Dedon, P. C. Proc. Natl. Acad. Sci.
U.S.A. 1998, 95, 102.
(35) Tretyakova, N. Y.; Burney, S.; Pamir, B.; Wishnok, J. S.; Dedon, P. C.;
Wogan, G. N.; Tannenbaum, S. R. Mutat. Res. 2000, 447, 287.
(36) Lopez-Larraza, D. M.; Moore, K., Jr.; Dedon, P. C. Chem. Res. Toxicol.
2001, 14, 528.
(37) Dong, M.; Wang, C.; Deen, W. M.; Dedon, P. C. Chem. Res. Toxicol.
2003, 16, 1044.
(38) Zhou, X.; Liberman, R. G.; Skipper, P. L.; Margolin, Y.; Tannenbaum,
S. R.; Dedon, P. C. Anal. Biochem. 2005, 343, 84.
9
J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010 6149

A R T I C L E S
Chan et al.
produce a diradical species that, when positioned in the minor
groove of DNA, abstracts hydrogen atoms from 2-deoxyribose
on each strand.28 Neocarzinostatin produces 2 single-strand
lesions for every double-strand lesion, with 5′-oxidation occur-
ring on one strand of a double-strand lesion and in most single-
strand damage events.28 Calicheamicin produces mainly double-
stranded DNA damage, with 4′-oxidation on one strand and 5′-
oxidation on the other.28 Both drugs are also very efficient at
producing DNA damage, with the concentration of damage
events closely approximating the concentration of added drug.28
These features of calicheamicin- and neocarzinostatin-induced
formation of nucleoside 5′-aldehyde residues are consistent with
our observation of high levels of furfural at 564 and 299 lesions
per 106 nt per µM, respectively, which translates into 0.4 and 1
µM concentrations of nucleoside 5′-aldehyde residues per µM
concentration of each drug. The lesser quantity of 1′-oxidation
apparent with neocarzinostatin (8 5MF per 106 nt per µM) is
consistent with the occurrence of 1′-chemistry only in double-
stranded lesions and only in a small subset of damage sites (GC-
containing damage sites).28 The observation of a minor amount
of 2-deoxyribonolactone associated with calicheamicin-induced
damage suggests a degree of flexibility in the binding of the
activated drug or in the selectivity of the diradical species for
minor groove-accessible 2-deoxyribose hydrogen atoms. These
results verify the accuracy of the methods.
As the basis for our method, Razskazovskiy and co-workers
took a different approach with greater specificity for the
2-deoxyribonolactone abasic site than the probe-based approach.
This entailed elimination of the ribonolactone as 5MF (Scheme
1) followed by HPLC resolution and quantification of 5MF by
UV spectroscopy.23 They exploited their observation of a
catalytic effect of polyamines and several metal ions for
increasing the efficiency of heat-induced release of 5MF from
irradiated DNA, with maximal release occurring with poly-
lysine.23 Our finding of 10 2-deoxyribonolactone abasic sites
per 106 nt per Gy of γ-radiation is significantly higher than the
1 lesion per 106 nt per Gy measured by Roginskaya and co-
workers using DNA subjected to 40 kVp X-rays.23 However,
when expressed as radiation chemical yield that takes into
account energy deposition in the irradiated solution, the results
are in better agreement, with G-values for 5MF formation of
13 nmol/J in the present studies and 6.7 nmol/J for Razska-
zovskiy and co-workers.23 The variance is probably not due to
the different radiation sources since the low energy X-rays
employed by Ronginskaya et al. would be expected to produce
more damage per unit dose than 60Co γ-rays due to the higher
relative biological effectiveness of the X-rays.42 Further, the
dose rates used in the two studies are similar (163 Gy/min in
the present studies versus 180 Gy/min in the studies of
Roginskaya et al.). One possible explanation is that the
discrepancy is due to different product elimination conditions
(shorter heating time, weaker catalyst), and the fact that no
internal standard was employed to correct for the loss of 5MF
during their workup process. Another discrepancy lies in the
conclusion that 2-deoxyribonolactone represents 30% of total
2-deoxyribose oxidation in X-ray-irradiated DNA,23 in com-
parison to our observation of a 7% frequency. This may be
related to their use of free base release as a measure of total
2-deoxyribose oxidation, which misses sugar damage involving
the nucleoside 5′-aldehyde, while we employed total strand
breaks and abasic sites.
The studies in purified DNA reveal quantitative insights into
mechanisms of 2-deoxyribose oxidation in DNA. Table 1
contains a compilation of the contributions of 2-deoxyribose
oxidation products to total 2-deoxyribose oxidation in DNA by
γ-radiation and Fe2+-EDTA, including data from the present
studies. We observed that nucleoside 5′-aldehyde comprises 40%
and 35% of the total 2-deoxyribose oxidation in DNA caused
by γ-radiation and Fe2+-EDTA, respectively. In addition to
our previous observation that a product of the other pathway
of 5′-oxidation, the 2-phosphoryl-1,4-dioxobutane residue,
represents ∼4% and ∼10% of total 2-deoxyribose oxidation by
γ-radiation and Fe2+-EDTA, respectively,9 5′-chemistry thus
represents nearly half of 2-deoxyribose oxidation in DNA. This
is consistent with theoretical and experimental evidence that
the 5′-position represents a major target for oxidants as a result
of solvent exposure.6,24 Solvent exposure may also account for
the relatively high percentage of damage arising from 4′-
oxidation (Table 1), with 13% of γ-radiation-induced 2-deoxy-
ribose oxidation accounted for by products from the two
pathways of 4′-chemistry: 3′-phosphoglycolate and the 2-deoxy-
pentos-4-ulose abasic site.8
Studies of the nucleoside 5′-aldehyde have to date been
primarily descriptive in nature,19,22 as have earlier studies of
the 2-deoxyribonolactone using various combinations of gas,
liquid and other types of chromatography with mass spectro-
metric or UV spectroscopic detection.39-41 More recent methods
for quantifying 2-deoxyribonolactone have improved sensitivity,
but were not chemically specific or rigorously quantitative.15,16,23
Greenberg and co-workers developed a biotinylated cysteine
derivative that reacts with the butenolide ꢀ-elimination product
of the 2-deoxyribonolactone (Scheme 1) to form a heat-stable
cyclic amide.15,16 The method can detect the 2-deoxyribono-
lactone lesion in samples of oxidized DNA with an apparent
limit of quantification in the high fmol/low pmol range.
However, the accuracy of the method has not been rigorously
established with quantitatively defined standards and the
specificity of the probe has not been established exhaustively
in reactions with the other R,ꢀ-unsaturated mono- and dicarbonyl
species generated during DNA oxidation (e.g., nucleoside 5′-
aldehyde and 1,4-dioxo-2-phosphorylbutane from 5′-oxidation,
3′-phosphoglycolaldehyde and the 3-oxonucleotide from 3-ox-
idation, etc.). These problems may be reflected in the conclusion
that the 2-deoxyribonolactone represents 75-80% of all alde-
hyde- or ketone-containing 2-deoxyribose oxidation products
arising with γ-radiation and Fe2+-EDTA.15,16 This would mean
that aldehyde- and ketone-containing 2-deoxyribose oxidation
products, which are produced from 2′-, 3′-, 4′-, and 5′-oxidation,
represent less than one-quarter of total 2-deoxyribose oxidation
products. Such a high proportion of the 2-deoxyribonolactone
lesion is at odds with the site-specific deuterium isotope effect
studies of Tullius and co-workers and with our chemically
specific observation of the 2-deoxyribonolactone as 7% and 24%
of 2-deoxyribose oxidation produced by γ-radiation and
Fe2+-EDTA, respectively.
However, factors other than solvent exposure must account
for the spectrum of 2-deoxyribose oxidation products. For
example, 2-deoxyribonolactone accounts for 24% of total
2-deoxyribose oxidation by Fe2+-EDTA (Table 1), which
(39) Dizdaroglu, M.; Schulte-Frohlinde, D.; von Sonntag, C. Z. Naturforsch.
1977, 32C, 1021.
(40) Goyne, T. E.; Sigman, D. S. J. Am. Chem. Soc. 1987, 109, 2846.
(41) Kappen, L. S.; Goldberg, I. H. Biochemistry 1989, 28, 1027.
(42) Hill, M. A. Radiat. Prot. Dosim. 2004, 112, 471.
9
6150 J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010

Quantifying 2-Deoxyribonolactone and Nucleoside 5′-Aldehyde
A R T I C L E S
Table 1. Relative Quantities of γ-Radiation- and Fe2+-EDTA-Induced 2-Deoxyribose Oxidation in DNA
% total 2-deoxyribose oxidation in isolated DNA
% solvent accessible % reaction with
damage per 106 nt per
Gy in TK6 cells
surfacea
hydroxyl radicala
2-deoxyribose oxidation product
γ-rad
Fe2+-EDTA
ref
1′
2′
3′
1
11
14
11
13
17
2-deoxyribonolactone abasic site
erythrose abasic site
7
-
1
-
10
3
24
-
0.045
-
0.002
-
-
-
0.22
-
present study
3′-phosphoglycolaldehyde
-
11, 12
3′-oxo-nucleotide
-
4′
5′
28
46
22
57
3′-phosphoglycolate
-
8
8
2-deoxypentos-4-ulose abasic site
5′-nucleoside 5′-aldehyde
5′-(2-phosphoryl-1,4-dioxobutane)
-
40
4
35
10
present study
9
a Data taken from Balasubramanian et al.24
reflects a significantly higher reactivity than solvent exposure
(1%) and the isotope effect studies of Tullius and co-workers
(11% relative reactivity) would predict.24 This difference may
reflect the higher stability of the 1′-radical, and thus the chemical
reactivity of the site, which is calculated to fall in the order 1′
> 4′ > 2′ > 3′ > 5′ by Sevilla and co-workers.43 The observation
of a 3.4-fold difference between γ-radiation and Fe2+-EDTA
in the proportion of 2-deoxyribonolactone (7% versus 24%,
respectively) also suggests that factors other than formation of
hydroxyl radical influence the reactivity of the oxidizing agents,
such as the negative charge of the Fe2+-EDTA complex as it
affects interactions with the polyanionic DNA helix. Some
caution is necessary in interpreting the data in Table 1, given
the potential for contributions to total 2-deoxyribose oxidation
from yet-to-be identified lesions and from damage not involving
strand breaks and abasic sites, such as 5′,8-cyclopurines.7,44
The results of studies in cells subjected to ionizing radiation
are also revealing. With corrections for losses during DNA
isolation, the cellular environment caused a 260-fold decrease
in the formation of nucleoside 5′-aldehyde from 57 to 0.22
lesions per 106 nt per Gy and a 220-fold reduction in 2-deoxy-
ribonolactone formation from 10 to 0.045 lesions per 106 nt
per Gy relative to purified DNA. This protective effect is similar
to ∼1000-fold reduction of γ-radiation-induced formation of
8-oxo-7,8-dihydro-dG in human cells45 and of the 3-phospho-
glycolaldehyde product of 3′-oxidation12 (neither study corrected
for artifactual losses), presumably due to the radical quenching
effects of chromatin proteins and other species in the cell.46
Some care must be exercised in interpreting the results of studies
in cells due to the relatively labile nature of the 2-deoxyribono-
lactone (24 h half-life at 37 °C13) and the nucleoside 5′-aldehyde
(30-100 h half-life22) and the potential for DNA repair, which
could potentially lead to loss of the lesions during the time
between exposure and DNA isolation. However, the rapid
irradiation and DNA workup (several hours), the correction for
losses during isolation, and the consistency of the results with
other types of DNA damage all support the quantitatively
rigorous nature of the cellular data.
formation rates of 1.5,12 9.7 and 57 lesions per 106 nt per Gy,
respectively, which amounts to a ratio of 1:6.5:38. In cells, the
rates were reduced to 0.002,12 0.045 and 0.22 lesions per 106
nt per Gy, respectively, and a ratio of 1:22:110. The relative
quantities are thus quite similar, especially in consideration of
the fact that the data for 3′-phosphoglycolaldehyde quantification
were not corrected for losses during DNA isolation from cells.12
These results raise the possibility of consistent in Vitro and in
ViVo results for other 2-deoxyribose oxidation products.
In conclusion, we have developed and validated a chemically
specific and sensitive isotope-dilution GC-MS method for the
quantification of DNA damage representing 1′- and 5′-oxidation
chemistry in Vitro and in cells: the 2-deoxyribonolactone and
nucleoside 5′-aldehyde residues. Our observation that the 5′-
carbon in 2-deoxyribose represents a major target for oxidants
is in general agreement with the solvent-accessible surface area
model of Balasubramanian et al.24 The studies reveal quantitative
insights into mechanisms of 2-deoxyribose oxidation as well
as a comprehensive compilation of the contributions of 2-deoxy-
ribose oxidation products to total 2-deoxyribose oxidation in
DNA by γ-radiation and Fe2+-EDTA. The GC-MS method
provides a means for accurate determination of these elimination
products as biomarkers of oxidative stress and as probes for
the fate of DNA damage products in cells and tissues.
Experimental Methods
Materials. All chemicals and reagents were of the highest purity
available and were used without further purification unless noted
otherwise. Calf thymus DNA, pentafluorophenylhydrazine (PFPH),
furfural, and poly-L-lysine were purchased from Sigma Chemical
Co. (St. Louis, MO). 2-Deoxy-D-[13C5]ribose was obtained from
Omicron Biochemicals, Inc. (Southbend, IN), 3,4,5-[2H]-furfural
from Medical Isotopes, Inc. (Pelham, NH), acetonitrile from
Honeywell Burdick & Jackson (Muskegon, MI), and dichlo-
romethane from Merck (Gibbstown, NJ). Deionized water was
further purified with a Milli-Q system (Millipore Corporation,
Bedford, MA) and was used in all experiments.
Instrumentation. Gas chromatography-mass spectrometry
(GC-MS) analyses were performed on a Hewlett-Packard 6890
series GC system equipped with a Hewlett-Packard 5973 mass
selective detector. High mass accuracy experiments were performed
on an Agilent 6510 series Q-TOF mass spectrometer with a standard
ESI interface. 1H and 1H-decoupled-13C NMR spectroscopic studies
were performed on a Bruker 400 MHz spectrometer. Samples were
irradiated in a Gammacell 220 Excel Irradiator (MDS Nordion,
Canada) with an annular 60Co source at a rate of 163.1 Gy/min.
Synthesis of (E)- and (Z)-1-(Furan-2-ylmethylene)-2-(per-
fluorophenyl)hydrazine (1a and 1b). The PFPH derivatives of
furfural and 3,4,5-[2H]3-furfural (1a, 1b) were prepared in overnight
reactions with a 10-fold molar excess of PFPH in hexane at ambient
temperature, as described by Ho and Yu.25 The products were
characterized by GC-MS and high mass accuracy mass spectrom-
A comparison of the relative quantities of the 3′-phospho-
glycolaldehyde, 2-deoxyribonolactone and nucleoside 5′-alde-
hyde products in isolated DNA and cells suggests that the
cellular environment does not greatly alter the selectivity of
hydroxyl radicals for the various positions in 2-deoxyribose.
Irradiation dose response studies in isolated DNA caused
(43) Colson, A. O.; Sevilla, M. D. J. Phys. Chem. 1995, 99, 3867.
(44) Murata-Kamiya, N.; Kamiya, H.; Muraoka, M.; Kaji, H.; Kasai, H. J.
Radiat. Res. (Tokyo) 1997, 38, 121.
(45) Frelon, S.; Douki, T.; Ravanat, J. L.; Pouget, J. P.; Tornabene, C.;
Cadet, J. Chem. Res. Toxicol. 2000, 13, 1002.
(46) Ljungman, M.; Hanawalt, P. C. Mol. Carcinog. 1992, 5, 264.
9
J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010 6151

A R T I C L E S
Chan et al.
etry. The reaction of furfural with PFPH leads to a hydrazone with
cis- and trans- isomers (1a, 1b) that were well resolved by GC-MS
(Figure 1) and showed identical EI mass spectra. As shown in
Supporting Information Figure 1A, the molecular ion [M]+. at m/z
276 was detected as the base peak, with abundant fragment ions at
m/z 117, 155, and 182 that corresponded to [C5F3]+, [C5F5]+, and
[C5F5NH]+, respectively. Standard stock solutions of defined
concentration were prepared as described elsewhere.25
hexadeoxynucleotide, 5′-GCATGC-3′, in potassium phosphate
buffer (50 mM, pH 7.4) was treated with 46 µM neocarzinostatin
and 10 mM glutathione, the latter to activate the drug for DNA
cleavage. After 12 h at 0 °C, the cleavage mixture was resolved by
reverse-phase HPLC using the conditions described above, with
collection of the nucleoside 5′-aldehyde-containing TGC fragment
eluting at 15.2 min. The product was characterized by high mass
accuracy mass spectrometry with an [M + H]+ ion detected at m/z
859.1850 (Supporting Information Figure 3). The presence of
nucleoside 5′-aldehyde was further confirmed by converting it to a
hydrazone derivative with a 10-fold molar excess of PFPH. The
resulting PFPH hydrazone derivative had an HPLC elution time of
28.5 min and an expected [M + H]+ at 1039.2 m/z (data not shown).
The oligodeoxynucleotide was stable at -80 °C for more than 1
mo and was determined to have a half-life of 34 h at 21 °C
(Supporting Information Figure 4), which compares favorably with
the half-life of 100 h at 37 °C in the studies of Greenberg and
co-workers.22
Synthesis of 6-Methyl-2-(perfluorophenyl)pyridazin-3(2H)-
one (2). The synthesis of 2 was achieved by reaction of 5MF,
prepared from 2-deoxy-D-ribose as described elsewhere,47 with a
5-fold molar excess of PFPH at 21 °C for 15 h. The reaction mixture
was resolved by reverse-phase HPLC on a Phenomenex Synergi
Column (250 mm ×10 mm, 4 µM) using water and acetonitrile as
the mobile phase, with a single product eluting at 11.6 min.
Following repurification with a shallower acetonitrile gradient (24.4
min), the chromatographically pure product was determined to be
6-methyl-2-pentafluorophenyl-3(2H)-pyridazinone (2) (Scheme 2).
GC-MS analysis revealed a molecular ion [M]+. at m/z 276 as the
base peak (Supporting Information Figure 1B) and characteristic
fragment ions at m/z 93 [C3F3]+, 117 [C5F3]+, 167 [C6F5]+, 229
Oxidation of Purified DNA. Buffered solutions of calf thymus
DNA were subjected to oxidation by various agents as follows.
Solutions of DNA (0.6 mL, 420 µg/mL) in Chelex-treated potassium
phosphate buffer (50 mM, pH 7.4) were subjected to γ-irradiation
at a dose-rate of 161.3 Gy/min over a range of 10-108 Gy. DNA
1
[M - CO - F]+, 248 [M - CO]+ and 257 [M - F]+. H NMR
(CDCl3): δ 7.20 (d, 1H, J ) 9.7 Hz), 6.99 (d, 1H, J ) 9.7 Hz),
I
2.38 (s, 3H, CH3). 13C NMR (CDCl3): δ 158.7 (s), 146.9 (s), 143.9
damaging reactions by neocarzinostatin, calicheamicin γ1 , and
Fe2+-EDTA were conducted as reported elsewhere by Chen
et al.9,48 In brief, DNA samples were exposed to neocarzinostatin
and calicheamicin and freshly prepared Fe2+-EDTA at concentra-
tions of 3.4-19.6 µM, 0.4-4.4 µM, and 25-360 µM, respectively,
with 10 mM glutathione used to activate neocarzinostatin and
calicheamicin. Following Sephadex G25 chromatography (GE
Healthcare), the DNA was processed for GC-MS analysis.
(d, JCF ) 260.3 Hz), 142.2 (d, JCF ) 259.5 Hz), 138.1 (d, JCF
)
259.9 Hz), 135.0 (s), 130.8 (s), 116.7 (s), 21.0 (s). UV absorbance
spectrophotometry revealed a λmax ) 295 nm, with an extinction
coefficient of 4000 M-1 cm-1. High mass accuracy mass spec-
trometry revealed a close correlation between theoretical (277.0400)
and measured (277.0396) m/z values of the [M + H]+ ion, which
is consistent with a molecular formula of C11F5H5N2O. Using a
similar strategy, the 13C5-labeled version of 2 was synthesized from
2-deoxy-D-[13C5]ribose. Standard stock solutions of defined con-
centration were prepared by weighing and verified by UV absorbance.
Synthesis and Characterization of the PFPH Derivatives of
2MF. 2MF was synthesized from thymidine by the method of
Stubbe and co-workers,26 which entails oxidation of 5′-O-p-
tolylsonylthymidine to 3′-keto-5′-O-p-tolylsonylthymidine with
pyridinium dichromate and release of 2MF by base-catalyzed
elimination reaction. 2MF was reacted with a 5-fold molar excess
of PFPH in dichloromethane for 15 h at 21 °C. Analysis of the
reaction mixture by GC-MS revealed a single product with a
molecular ion at m/z 276 as the base peak and characteristic
fragment ions at m/z 93 [C3F3]+, 117 [C5F3]+, 148 [C6F4]+, and
167 [C6F5]+. The product proved to be unstable during storage at
-80 °C, so an internal standard could not be prepared for GC-MS
analyses.
GC-MS Quantification of the PFPH Derivatives of 5MF
and Furfural in Oxidized DNA. The general strategy for
quantification of dRL and NA lesions in oxidized purified DNA
involves release of their ꢀ-/δ-elimination products, 5MF and
furfural, respectively, derivatization of the products with PFPH,
addition of internal standards, extraction of the PFPH derivatives,
and quantification by GC-MS. To each DNA sample (0.6 mL)
was added an aliquot of an aqueous solution of poly-L-lysine (15
µL, 0.1 M) followed by incubation at 90 °C for 1 h. After cooling
to ambient temperature, an aliquot of an aqueous solution of PFPH
(30 µL, 0.2 M) was added and the reaction allowed to proceed
overnight at ambient temperature. Following addition of a 20 µL
mixture of isotope-labeled internal standards (9 µM [2H]3-1a/1b,
15 µM [13C]5-2), the solutions were extracted three times with 300
µL of dichloromethane. The combined organic layers were dried
under N2 and the residue was dissolved in 20 µL of dichloromethane.
GC-MS analysis was performed with 1 µL of the dichlo-
romethane solution of each sample using positive electron impact
ionization (EI; 70 eV). Samples were injected onto a DB-35MS
capillary column (30 m × 0.25 mm ×0.25 µm film thickness) in
a Hewlett-Packard 6890/5973 GC-MS system operating with the
following parameters: injector temperature set at 250 °C in the
splitless mode for the first 1.5 min, and transfer line temperature
at 280 °C; carrier gas (He) 1 mL/min; GC oven at 40 °C for 2
min, increased to 200 at 7 °C/min, ramped to 310 at 20 °C/min,
and held at 310 °C for 5 min; solvent delay, 16 min (to allow the
elution of excess PFPH reagent prior to mass spectrometer scan).
Data were acquired by an HP5973 Mass Selective Detector operated
in selected ion monitoring mode (SIM) at m/z values of 276, 279,
and 281. Using calibration curves that were validated daily by
running a standard mixture before the samples, quantities of 1a+1b
were determined from the sum of the peak area ratios of 1a and
1b (m/z 276) to [2H]3-1a and [2H]3-1b (m/z 279), respectively, while
2 was quantified from the peak area ratio of 2 (m/z 276) to [13C]5-2
(m/z 281). Calibration curves were prepared using mixtures of
Synthesis of 2-Deoxyribonolactone- and 5′-Aldehyde-Con-
taining Oligonucleotides. A 17-mer oligodeoxynucleotide, 5′-
TGTGCCXAACTTACCGT-3′, containing a 2-deoxyribonolactone
at X (3) was prepared by UV irradiation of a precursor containing
a nitrobenzyl cyanohydrin nucleoside analogue, as described by
Zheng and Sheppard.13 The conversion efficiency was determined
to be 92% by reverse-phase HPLC with a system consisting of a
Thermo Hypersil GOLD aQ 150 mm × 2.1 mm column with 3
µM particle size eluted at 200 µL/min with a gradient of acetonitrile
in 10 mM ammonium acetate (1% acetonitrile for 2 min, 1-20%
over 23 min, 20-100% over 3 min, held at 100% acetonitrile for
10 min); the retention times of the oligodeoxynucleotides were 23.9
and 25.4 min for the deoxyribonolactone-containing and parent
species, respectively. The final products were identified by high
mass accuracy mass spectrometry (Supporting Information Figure 3)
and by polyacrylamide gel electrophoretic analysis (Supporting
Information Figure 5).
An oligodeoxynucleotide standard containing a nucleoside 5′-
aldehyde terminus, 5′-TGC-3′ (4), was prepared as described by
Sugiyama et al.27 Briefly, a 90 µM solution of a self-complementary
(48) Wang, L.; Chen, S.; Xu, T.; Taghizadeh, K.; Wishnok, J. S.; Zhou,
(47) Grundmann, C.; Kober, E. J. Am. Chem. Soc. 1955, 77, 2332.
X.; You, D.; Deng, Z.; Dedon, P. C. Nat. Chem. Biol. 2007, 3, 709.
9
6152 J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010

Quantifying 2-Deoxyribonolactone and Nucleoside 5′-Aldehyde
A R T I C L E S
variable quantities of 1a+1b (2.92-2920 pmol) and 2 (0.432-432
pmol) and fixed quantities of 2H3-(1a+1b) (180 pmol) and 2 (300
pmol) with calf thymus DNA (0.6 mL, 420 µg/mL) in Chelex-
treated potassium phosphate buffer (50 mM, pH 7.4).
nucleotide of the purified, irradiated DNA added to the cell sample.
As shown in the radiation dose-response curve in Supporting
Information Figure 6, fitting the data by linear regression yielded
lines for 5MF and furfural with slopes of 5.3 5MF per 106 nt per
Gy (r2 ) 0.99) and 44 furfural per 106 nt per Gy (r2 ) 1.0). On the
basis of analyses in purified DNA (10 5MF and 57 furfural per 106
nt per Gy), all data from cellular experiments were subsequently
corrected by factors of 1.9 for 5MF and 1.3 for furfural to account
for losses during DNA isolation.
Validation of the GC-MS Method. The efficiency of the
elimination, derivatization, and analytical steps was determined
using the oligodeoxynucleotides 3 and 4, which provides a critical
correction for artifacts arising during sample processing. The
validation entailed spiking 250 µg of calf thymus DNA with 3 (88
pmol) or 4 (293 pmol) followed by heat and polylysine-catalyzed
elimination of 5MF or furfural, respectively, derivatization with
PFPH, and GC-MS analysis of extracts, as described above. The
overall efficiency of the analytical method was calculated as the
measured quantities of 2 or 1a + 1b divided by the quantity of
added 3 or 4, respectively. The yields of 1a+1b and 2 were found
to be 43 ( 2.4% and 21 ( 0.7% of the theoretical value,
respectively. All data were subsequently corrected by factors of
2.3 and 4.8, respectively.
Quantification of the PFPH Derivatives of 5MF and Fur-
fural in γ-Irradiated TK6 Cells. Studies were undertaken to
quantify the formation of 2-deoxyribonolactone and nucleoside 5′-
aldehyde in TK6 human lymphoblastoid cells exposed to γ-radia-
tion. Cultures of TK6 human lymphoblastoid cells, grown to 1 ×
106 cells per milliliter as described previously,12 were washed three
times with PBS and resuspended in PBS at a concentration of 5 ×
106 cells per milliliter. The suspensions (10 mL) were exposed to
γ-radiation (0-500 Gy), at ambient temperature at a dose-rate of
161.3 Gy/min. Genomic DNA was isolated using a Qiagen DNA
isolation kit and quantified by UV absorption at 260 nm. The DNA
was processed for GC-MS analysis of 5MF and furfural as
described earlier.
Acknowledgment. We extend thanks to Dr. Terry Sheppard
(Nature Publications) for providing the precursor to 2-deoxyri-
bonolactone-containing oligonucleotide, to Prof. Jacquelyn Yanch
(Dept. of Nuclear Science and Engineering, MIT), Dr. Dennis H. W.
Chan (Dept. of Chemistry, Hong Kong University of Science and
Technology), and Dr. Liang Cui (Dept. of Biological Engineering,
MIT) for helpful discussions and insights into data interpretation,
to Drs. Jeff Simpson and Bob Kennedy for help with NMR
experiments (Spectroscopy Laboratory, Dept. of Chemistry, MIT),
and to Ms. Bahar Edrissi for the cell culture. Special thanks are
extended to Prof. Yuriy Razskazovskiy (East Tennessee State
University) for providing samples of 5MF and for critical insights
into radiation chemistry. The chromatographic and mass spectro-
metric analyses were performed in the Bioanalytical Facilities Core
of the MIT Center for Environmental Health Sciences. This work
was supported by grants from the National Institute of Environ-
mental Health Sciences (ES002109), the National Center for
Research Resources (RR023783-01, RR017905-01), and the Na-
tional Cancer Institute (CA103146).
To control for effects of the cellular environment on the quantities
of 2-deoxyribonolactone and nucleoside 5′-aldehyde during DNA
isolation, a control experiment was performed in which purified,
irradiated DNA containing defined quantities of 2-deoxyribono-
lactone and nucleoside 5′-aldehyde was added to lysed cells during
the DNA isolation procedure and 5MF and furfural were quantified
as described earlier. Samples of purified genomic DNA from TK6
cells, isolated using the Qiagen kits described earlier, were
γ-irradiated with 0, 8.4, 22.4, or 36.4 Gy. A 122 µg quantity of the
irradiated DNA was added to 5 × 107 unexposed TK6 cells (∼250
µg DNA) at the nucleus isolation stage of the Qiagen genomic DNA
isolation protocol. Following purification of the total DNA mixture,
5MF and furfural were quantified by GC-MS in the purified DNA
and spiked-cell DNA, as described earlier. The quantities of 5MF
and furfural in the spiked-cell DNA were expressed as lesions per
Supporting Information Available: Electron impact ionization
mass spectra of the PFPH derivatives of furfural (1a, 1b) and
5MF (2); calibration curves for the gas chromatography-mass
spectrometry analysis of 1a, 1b, and 2; reverse-phase HPLC
analysis of the 2-deoxyribonolactone- and nucleoside 5′-
aldehyde-containing oligodeoxynucleotides; reverse-phase HPLC
analysis of the stability of the 5′-TGC-3′ containing nucleoside
5′-aldehyde terminus (4); polyacrylamide gel electrophoretic
analysis of the 2-deoxyribonolactone-containing oligodeoxy-
nucleotide (3). This material is available free of charge via the
Internet at http://pubs.acs.org.
JA910928N
9
J. AM. CHEM. SOC. VOL. 132, NO. 17, 2010 6153