RAPID COMMUNICATIONS IN MASS SPECTROMETRY

Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242

Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.2579

Detection of reactive free radicals derived

from nucleosides by liquid chromatography coupled
to tandem mass spectrometry of DMPO spin
trapping adducts
Vincent Maurely, Jean-Luc Ravanat* and Serge Gambarelli*
Laboratoire de Résonances Magnétiques et Laboratoire des Lésions des Acides Nucléiques, LCIB (UMR-E 3 CEA-UJF), DRFMC, CEA-
Grenoble, 17 Avenue des Martyrs, 38054 Grenoble Cedex 9, France
Received 12 January 2006; Revised 16 May 2006; Accepted 17 May 2006

In this study, reactive free radicals derived from several nucleosides were spin trapped by 5,5-
dimethyl-1-pyrroline N-oxide (DMPO) and then detected by high-performance liquid chromatog-
raphy coupled to electrospray ionization tandem mass spectrometry (HPLC/ESI-MS/MS). This
method provides a specific detection of spin trapping adducts derived from nucleosides with a very
high sensitivity: quantities as low as 0.5 picomoles of spin trapping adducts corresponding to
concentrations of 2.5 10S8 mol. LS1 were detected. Different spin trapping adducts were characterized
by HPLC/ESI-MS/MS in three well-known systems producing free radicals photochemically: the
photolysis of 5-halo-2(-deoxyuridines, the photolysis of 5-thiophenylmethyl-2(-deoxyuridine and the
photolysis of thymidine with menadione bisulfite as a photosensitizer. A new radical photoreactivity
of uridine derivatives was also detected by this method both at the nucleoside and at the RNA level,
showing that the method is also relevant for studying spin trapping adducts derived from DNA and
RNA strands. Copyright # 2006 John Wiley & Sons, Ltd.

The detection and characterization of radical transients are radical16 and of free radicals produced by lipid peroxi-
important in understanding chemical mechanisms involved dation.15 In these studies, while the combined approach is
in the photo-induced damage of genetic material.1,2 The very efficient, it was found that the presence of other redox
study of these transients is particularly difficult due to their forms of the spin trapping adducts16,17 (see Scheme 2)
short lifetimes. For instance the lifetime of pyrimidine- rendered MS data interpretation difficult. In further MS
derived radical cations can be shorter than 10 ns.3 Spin studies18,19 the same group of authors overcame this problem
trapping4,5 is a very efficient way to overcome this problem through the use of 5,5-dimethyl-1-pyrroline N-oxide
by transforming short-lived radical transients into more (DMPO) as a spin trap in aqueous solution, for which the
stable free radicals named ‘spin trapping adducts’ (see oxidized spin trapping adduct (nitrone, see Scheme 2) is of
Scheme 1). These more stable adducts are usually studied by higher stability.20 The sensitivity of HPLC/EPR/MS is
electron paramagnetic resonance (EPR). The spin trapping limited by the EPR detection of nitroxide adducts, which
technique with EPR detection of adducts has been used in has a lower threshold of 10
6 mol  L
1 in aqueous solution.
numerous studies for the detection of transient radicals The spin trapping approach could be greatly improved by
derived from DNA components.6–14 increasing its sensitivity while maintaining its selectivity for
Hyperfine coupling constants and g factors provided by trapped radicals. This is possible in the case of nucleosidic
the EPR spectrum of a spin trapping adduct are not always radicals by replacing EPR detection by electrospray ioniz-
sufficient, however, to unambiguously determine the ation tandem mass spectrometry coupled with high-
chemical structure of the trapped radical (for examples, performance liquid chromatography (HPLC/ESI-MS/
see Yue Qian et al.15). In these cases mass spectrometry can be MS).21,22 In the present work, potential DMPO spin trapping
useful in the characterization of spin trapping adducts as it adducts of nucleosidic radicals were detected using a specific
provides more complete structural information. Iwahashi fragmentation of nucleosides (loss of 2-deoxyribose moiety)
et al. explored this possibility by designing an on-line HPLC/ and identified on the basis of their molecular weights. Then,
EPR/MS experiment and applied it in the study of the phenyl the more sensitive multiple reaction monitoring (MRM)
mode was found to have a sensitivity about two orders of
*Correspondence to: J.-L. Ravanat or S. Gambarelli, Laboratoire magnitude higher than EPR detection, lowering the detection
des Lésions des Acides Nucléiques et Laboratoire de Résonances threshold to concentrations as low as 10
8 mol  L
1.
Magnétiques, LCIB (UMR-E 3 CEA-UJF), DRFMC, CEA-Greno-
ble, 17 Avenue des Martyrs, 38054 Grenoble Cedex 9, France.
E-mail: jravanat@cea.fr; sgambarelli@cea.fr
y
Present address: Laboratory of Photochemistry, University of
Ottawa, 10 Marie Curie, K1N 6N5, Ottawa, ON, Canada

Copyright # 2006 John Wiley & Sons, Ltd.

2236 V. Maurel, J.-L. Ravanat and S. Gambarelli

(photosensitizer) corresponds to an absorbance of 0.3 at

312 nm for an optical length of 1 mm; at this wavelength the
R + N
+
R N absorbances of Thd (saturated solutions) and of DMPO
-
O O (10 mM) are negligible.
For high-power experiments, UV-C irradiations were
DMPO
performed using a single shot of a pulsed (5 ns) Nd-Yag
Scheme 1. Formation of stable adducts by spin trapping laser (Quantel, Les Ulis, France) at 266 nm. The samples were
from a radical transient R. held in an open plastic vial (Eppendorf, Le Pecq, France) of
200 mL and 5 mm diameter and received the laser beam
through the vial input.
In order to validate this method, our work focused first on
5-halopyrimidine nucleosides, which are well-known photo- Chemicals
precursors of nucleoside free radicals. We then applied this 5-IdUrd, 5-BrdUrd, dUrd, 5-BrUrd, Urd, Thd and menadione
method to the detection of thymidine-based free radicals bisulfite were of the highest available purity, purchased from
produced by well-described mechanisms such as the Sigma (Lyon, France) and used as received. An aqueous
photosensitization of thymidine by menadione2,11 and the solution of sodium deuterium oxide (8.5 M) was purchased
photolysis of 5-thiophenylmethyl-20 -deoxyuridine, which is from Euriso-top (Paris, France). DMPO was purchased from
a photoprecursor of the 5-(20 -deoxyuridinyl)methyl Sigma. Aqueous solutions of DMPO were stirred in the
radical.23,24 This method also enabled us to detect an presence of active charcoal for 2 h, and the charcoal was then
unexpected spin adduct derived from an uracil-centered removed by filtration. This purification procedure was
radical generated upon UV-C irradiations of uridine, 20 - repeated three times in order to minimize the EPR signals
deoxyuridine and RNA strands. due to paramagnetic impurities in commercial DMPO. 5-
Thiophenylmethyl-20 -deoxyuridine was synthesized as
described by Romieu et al.25
EXPERIMENTAL
Enzymatic hydrolysis of RNA
UV irradiation for spin trapping experiments Quantitative digestion of RNA into the corresponding
For low-power light irradiations, aqueous samples of 250 mL ribonucleosides was performed under identical conditions
were held in a 1 mm optical path quartz UV-visible cell to those used to digest DNA.26
(Hellma, Paris, France) and irradiated using either a Rayonet
photoreactor (The Southern New England Ultraviolet HPLC/ESI-MS/MS measurements
Company, Handen, CT, USA) or a UV lamp (Bioblock, The HPLC/ESI-MS/MS system used in the present work,
Illkirch, France) equipped with Hg-vapour tubes. The light consisting of an API 3000 (Applied Biosystems, Toronto,
power was measured using a VLX3W radiometer (Vilbert- Canada) electrospray tandem mass spectrometer connected
Lourmat, Marne-la-Vallée, France). For UV-C (254 nm) to a HP1100 HPLC system (Agilent, Massy, France), has been
experiments the light power used was 0.65 mW/cm2 for described in detail in previous work.22 For the detection of
the UV lamp and 5 mW/cm2 for the photoreactor. The potential DMPO-nucleoside adducts, the mass spectrometer
nucleoside concentrations were adjusted to obtain an was used in the neutral loss MS/MS mode with a constant
absorbance close to 1.0 for an optical path of 1.0 mm. Typical loss of 116 Da, in positive ion mode. The first quadrupole was
concentrations used are 1.0 mM for 20 -deoxyuridine (dUrd), scanned over a range of 50 Th around the m/z value of the
3.2 mM for 5-iodo-20 -deoxyuridine (5-IdUrd), 3.6 mM for 5- [MþH]þ ion of the expected adducts. The ion spray voltage
bromo-20 -deoxyuridine (5-BrdUrd), 1.9 mM for 5-thiophe- was 5500 V, the collision energy was 25 eV and the total scan
nylmethyl-20 -deoxyuridine, 1.0 mM for uridine (Urd), and time was 1.5 s. A MRM method was set up for the detection of
3.6 mM for 5-bromouridine (5-BrUrd). DMPO was chosen as the dUrd-DMPO adduct (nitrone c) using the transition
a spin trap because of its low extinction coefficient in UV-C: m/z 340 ! 224. The spray voltage was maintained at 5500 V,
according to our measurements in aqueous solution, the optimized collision energy was 20 eV, and the tempera-
e254 nm ¼ 360 mol
1 cm
1 L. The DMPO concentration was ture of the auxiliary turbo gas was 4008C. For detection of the
low enough (1.0 mM) to ensure an absorption lower than 0.04 Urd-DMPO adduct nitrone c(, the transition used was
for an optical length of 1.0 mm at 254 nm. For UV-B m/z 356 ! 224 using the same MS parameters as those
experiments (300 nm), the light power used was 5 mW/ optimized for the detection of the corresponding 2-deoxyr-
cm2. The aqueous solution used was saturated in thymidine ibose derivative nitrone c. In addition, to improve the
(Thd). The concentration of 6.5 mM of menadione bisulfite specificity of detection of nitrone c(, additional transitions
were used, m/z 356 ! 206, 356 ! 163 and 356 ! 135, with a
collision energy of 40 eV. HPLC analysis was performed for
R N R N R
+
N
30 min after injection and an additional 15 min period was
OH O O
- used to equilibrate the HPLC column between successive
hydroxylamine nitroxide nitrone
injections. Separations were performed using an Uptisphere-
ODB (3 mm, 0.2  15 cm) column obtained from Interchim
Scheme 2. The three possible redox forms of a spin trapping (Montluçon, France). The elution was achieved at a flow rate
adduct obtained with DMPO. of 0.2 mL/min in the gradient mode. The proportion of
Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm
 Detection of spin-trapped radicals of nucleosides 2237

acetonitrile in 5 mM ammonium formate, starting from 0%, monodeuterated D6-dUrd 20 -deoxyuridine in a 70:30 ratio, as
reached 50% within 30 min, the column being maintained at determined by ESI-MS.
288C.

EPR measurements RESULTS

EPR measurements were performed on a Bruker (Wissem-
bourg, France) EMX X-band continuous wave spectrometer Validation of the method: detection of free
with a Bruker ER 4116 DM rectangular cavity operating at radicals produced by photolysis
9.66 GHz. Experiments were performed at room tempera- of 5-halo-2(-deoxyuridine
ture. Due to the high dielectric loss of aqueous solutions, a Aqueous solutions of 5-IdUrd containing DMPO as a spin
1 mm i.d. glass tube was filled with 30 mL of sample and trap were UV-C irradiated and analyzed by HPLC/ESI-MS/
inserted in a standard EPR tube. In the reported spectra, EPR MS. The neutral loss MS/MS mode with a constant loss of
settings for the measurements are: microwave power 10 mW, 116 Da was used to detect potential DMPO-nucleoside
modulation amplitude 3 G, time constant 10 ms, acquisition adducts. Such a detection method has been already
rate 42 s/scan. successfully used for the detection of radiation-induced
modified nucleosides,22 the loss of 116 Da corresponding to
NMR characterization of nitrone c the loss of the 2-deoxyribose moiety. The major component
Nitrone c (1H, 500 MHz, D2O) d, ppm: 9.92 (s, HC6), 6.33 (t, 2 detected had a molecular weight of 339 Da, that could
HC10 , J ¼ 6.0 Hz), 4.43–4.46 (m, HC30 ), 4.06–4.09 (m, HC40 ), 3.75– correspond to a DMPO-dUrd adduct. Therefore, to improve
3.79 (m, 2 HC50 ), 3.15–3.29 (m, 2 Hallyl, Jgeminal ¼ 18.5 Hz, the sensitivity of detection, a reaction monitoring method,
J ¼ 7.5 Hz, Dd ¼ 0.06 ppm), 2.33–2.39 (m, 2HC20 ), 2.10 (t, 2 using the transition m/z 340 ! 224, was set up to specifically
HCH2, J ¼ 7.5 Hz), 1.40 (s, 6 HCH3). detect this compound. The chromatogram corresponding to
DMPO (1H, 200 MHz, D2O) d, ppm: 7.18 (t, Hvinyl, J ¼ 2.4 Hz), the transition m/z 340 ! 224 exhibits one main peak at
2.69 (t of d, 2 Hallyl, J3 ¼ 7.4 Hz, J ¼ 2.4 Hz), 2.17 (t, 2 HCH2, 16.1 min (Fig. 1). The detected product was attributed to the
J ¼ 7.4 Hz), 1.40 (s, 6 HCH3). expected nitrone c, which could be obtained from the
2(-Deoxyuridine (1H, 200 MHz, D2O) d, ppm: 7.82 (d, HC6, nitroxide spin adduct b produced by the spin trapping
J ¼ 4.0 Hz), 6.26 (t, 2 HC10 , J ¼ 6.8 Hz), 5.86 (d, HC5, J ¼ 4.0 Hz), reaction of radical a with DMPO (Scheme 3).
4.40–4.47 (m, HC30 ), 3.99–4.06, (m, HC40 ), 3.67–3.86 (m, 2 HC50 ), A large amount of the product was then prepared to
2.33–2.40 (m, 2HC20 ). confirm the proposed structure on the basis of extensive
NMR and MS investigations. The main features of the
Synthesis of monodeuterated D6-dUrd recorded NMR spectrum, compared with those of DMPO
The synthesis was adapted from the work of Rabi and Fox.27 and dUrd (see Experimental section), are: (i) the proton at the
dUrd (101 mg) was dissolved in D2O and then D2O was C5 position in the spectrum of dUrd and the vinylic proton of
evaporated in order to remove exchangeable protons. This DMPO have no corresponding signals in the spectrum of the
operation was repeated twice, then dUrd was dissolved in nitrone c; (ii) the proton on the C6 position on the uracil
2 mL of deuterated DMSO and 200 mL of a 3 M NaOD moiety has a downfield chemical shift (9.92 ppm) in the
solution in D2O were added. This solution was refluxed for nitrone c compared with that of dUrd (7.82 ppm), in
3 h at 1358C then cooled by adding 7 mL of distilled water. accordance with the withdrawing effect of the nitrone
The pH was adjusted to 7 by addition of HCl. After solvent moiety in c; (iii) the allylic protons of the pyrrolidine N-oxide
evaporation the product was purified by preparative HPLC moiety of nitrone c are not equivalent, most probably due to a
using a Phenomenex Ultramax 5 C18 column, 250  10 mm, steric hindrance from the uracil ring. Additional information
with methanol/water 5:95 as an eluent. The overall yield was was gained from ESI-MS. The successively observed product
59% (60 mg). NMR (1H, 200 MHz, D2O) d, ppm: 6.04 (t, 2 HC10 , ions m/z 340 ! 224, 224 ! 206, 206 ! 163, and 163 ! 135 are
J ¼ 6.8 Hz), 5.63 (s, HC5), and other signals were unchanged
compared with the spectrum described for dUrd. The ESI
mass spectrum yielded a [M–Hþ]
ion, m/z 228 (100%), and
m/z 229 (22%).

Synthesis of bideuterated 2D5,6-dUrd

After the removal of exchangeable protons, 500 mg of dUrd
were dissolved in 10 mL of deuterated DMSO, then 1 mL of a
6 M NaOD solution in D2O was added. This solution was
refluxed for 6 h at 1358C. Two phases appeared, the aqueous
one was brown and the organic one was yellow. The reaction
was stopped by adding 10 mL of water and the pH of the
solution was neutralized. After evaporation, the resulting
solid was stirred in water (30 mL). The orange solution
obtained by filtration was purified by HPLC as described Figure 1. HPLC/ESI-MS/MS chromatogram obtained for the
above. The overall yield was 0.2% (1 mg) of a white solid transition m/z 340 ! 224 of an aqueous solution of 5-IdUrd
which corresponds to a mix of bideuterated 2D5,6-dUrd and containing DMPO photolyzed by UV-C (dose 50 mJ).
Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm
2238 V. Maurel, J.-L. Ravanat and S. Gambarelli
O O
X
HN HN

O N O N

HO hν HO
O X + O

OH OH a

X = Br or I

+
N
-
O

O O
+
HN N HN N
-
O O
O N O N
HO oxidation HO
O O

OH c OH b

Scheme 3. Reaction of spin trapping following the photolysis

of 5-halo-20 -deoxyuridine and production of the nitrone c by
oxidation of the spin adduct b.

compatible with the structure of the nitrone c proposed in

Scheme 4. In particular the neutral losses of 43 Da
(m/z 206 ! 163) and 28 Da (m/z 163 ! 135) are characteristic
of pyrimidine nucleosides.21 The experimental data are in
agreement with the structure proposed for the nitrone c,
whose formation involves the initial generation of a 20 -
deoxyuridin-5-yl radical a that could be trapped by DMPO
(Scheme 3).
1
H-NMR was also used to determine the concentration of
nitrone c using dUrd as an internal standard as described Figure 2. (a) Calibration curve obtained for the detection of
previously.28 The calibrated solution was used to determine nitrone c by HPLC/ESI-MS/MS using the transition m/z
the molecular absorption coefficients of nitrone c: e247nm ¼ 340 ! 224. (b) Amounts of nitrone c detected by HPLC/
19800 mol
1 cm
1 L and e311nm ¼ 10250 mol
1 cm
1 L. Then, ESI-MS/MS for UV-C-irradiated aqueous solutions of DMPO
calibrated solutions were prepared and used as external and 5-IdUrd (*) and DMPO and 5-BrdUrd (&) as a function of
standards for a quantitative determination of nitrone c the UV-C irradiation dose. Quantities of nitrone c are given in
detected by HPLC/ESI-MS/MS. The detection response was picomoles detected in the injected volume (20 mL) and calcu-
found to be linear with the injected amount of nitrone c over at lated using the area of the main peak and the calibration
least two orders of magnitude, starting from 0.5 pmol (limit of shown in (a).
quantification) up to 50 pmol.
The HPLC/ESI-MS/MS approach was then used to nitrone c in these experiments. Calculated quantum yields
determine the quantity of nitrone c produced by photolysis were found to be 1.1  10
3 and 1.3  10
4 for 5-IdUrd and
of 5-IdUrd and 5-BrdUrd. The amount of nitrone c was found 5-BrdUrd, respectively. The determined quantum yields are
to increase linearly with the dose of UV-C irradiation for both one order of magnitude lower than those given in the
5-IdUrd and 5-BrdUrd (Fig. 2(b)). The slope of the curve was literature for the disappearance of 5-IdUrd (quantum yield29
then used to determine the quantum yields of formation of 2.84  10
2) and 5-bromouracil (quantum yield30 1.8  10
3)
in photolysis experiments at 254 nm in aqueous solutions.
These lower quantum yields were expected since only a
-43 fraction of the produced radicals react with DMPO to
O generate nitrone c. Interestingly, the formation of nitrone c
+
HN N was found to be ten times less efficient upon UV-C-induced
-
-28 O N
O
Nitrone c photolysis of 5-BrdUrd than of 5-IdUrd, as expected from the
HO
mass 339 Da lower quantum yield reported in the literature for the
O
-116 brominated nucleoside than for the iodinated one.
OH
Applications to the detection of free radicals
Scheme 4. Proposed fragmentations to explain transitions derived from Thd
observed for nitrone c analyzed by electrospray mass spec- Aqueous aerated thymidine solutions containing DMPO as a
trometry. spin trap and menadione bisulfite as a photosensitizer were
Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm
 Detection of spin-trapped radicals of nucleosides 2239

Thus, the major detected product by HPLC/MS/MS was

attributed to the nitrone g derived from spin trapping of
transient thymidyl-5-yl generated radical e (Scheme 5) as
observed for the photolysis of 5-thiophenylmethy-20 -deox-
yurine (vide infra). In addition, another main product was
detected with the transition m/z 356 ! 240. Such a product
was also detected when photosensitization was performed
using benzophenone as a photosensitizer in the presence of
UV-A (not shown). Work is in progress to identify this
compound that could correspond to the addition of DMPO
onto the C5–C6 double bond of the thymidine radical cation
to form a C5 or C6 DMPO adduct h of saturated thymidine.
In the absence of DMPO, deprotonation of the thymidine
radical cation is in competition with hydration onto the C5–
C6 double bond, and, under our experimental conditions, we
could postulate that the addition of DMPO, instead of water,
generates the detected DMPO spin adduct h.
For this system, the EPR signal of one nitroxide spin trap
adduct was measured. This spectrum (Fig. 4(a)) is typical of a
DMPO adduct obtained from a carbon-centered radical with
main hyperfine coupling constants in the range aN ¼ 16–16.5
Figure 3. (a) HPLC/ESI-MS/MS chromatogram obtained G and aH,Cb ¼ 23–24 G. However, any improvement of the
using a constant neutral loss of 116 Da of an aqueous solution spectral resolution and therefore any further analysis of this
saturated with Thd, containing menadione bisulfite and spectrum were made difficult by the solvent used (water)
DMPO (10
2 mol  L
1) photolyzed by UV-B (dose and by the low stability of the observed species. In addition,
3750 mJ). dR ¼ 2-deoxyribose moiety. (b) Superposition even with better stability, the resolution of superhyperfine
of HPLC/ESI-MS/MS chromatograms for transitions m/z couplings would be unusual for this type of adduct. The
356 ! 240 and m/z 354 ! 238 of an aqueous solution intensity of the EPR signal shown in Fig. 4(a) decreases with a
of 5-thiophenylmethyl-20 -deoxyuridine containing DMPO half-life time in the range of 20 min after the end of
(10
2 mol  L
1) photolyzed by UV-C (dose 3750 mJ). irradiation. This decay was slow enough to run a HPLC
purification and still to detect an EPR signal for one of the
collected fractions (Fig. 4(b)). The analysis of this fraction by
UV-B irradiated with a low light intensity and the solutions ion trap MS exhibited one main ion at m/z 356 and one
were analyzed by HPLC/ESI-MS/MS using a constant product ion at m/z 240, consistent with the previous
neutral loss of 116 Da. Two main peaks were observed, at observation of the transition m/z 356 ! 240 by HPLC/ESI-
16.37 min for the transition m/z 354 ! 238 and at 20.25 min for MS/MS.
the transition m/z 356 ! 240 (Fig. 3(a)). In order to check if the peak observed at 16.37 min for the
According to the previously described interpretation of transition m/z 354 ! 238 can be attributed to the nitrone g
EPR spin trapping experiments,11 the absorption of a UV-B derived from the radical e, aqueous solutions of 5-
photon by menadione bisulfite leads to an electron transfer thiophenylmethyl-20 -deoxyuridine (a photoprecursor of
from Thd to the photosensitizer by the so-called type I radical e, Scheme 6) containing DMPO were UV-C irradiated
photosensitization reaction.31 Then, the initially generated and then analyzed by HPLC/ESI-MS/MS. The chromato-
thymidine radical cation d rapidly deprotonates to produce grams corresponding to the transitions m/z 354 ! 238 and
the neutral radical e or could be hydrated onto C5 or C6. 356 ! 240 are shown in Fig. 3(b). The main peak is obtained

O O O O
° N
HN UV-B HN - H+ HN DMPO HN
+° O°
O N O N O N O N
Menadione
dR dR d dR e dR f

DMPO
O

HN O +
N
O N HN
+ O
dR N
O N
O g
h dR

Scheme 5. Radicals produced by UV-B irradiation of thymidine photosensitized

by menadione bisulfite and corresponding spin trapping adducts.
Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm
2240 V. Maurel, J.-L. Ravanat and S. Gambarelli

Figure 4. (a) EPR spectrum (1 scan) measured 10 min after

the UV-B irradiation of a saturated aqueous solution of Thd
containing menadione bisulfite and DMPO (10
2 mol  L
1).
Dose: 3750 mJ. (b) Most intense EPR spectrum (10 scans)
measured in the fractions collected after HPLC purification.

for the transition m/z 354 ! 238 at 16.24 min. A much weaker
peak is observed for the transition m/z 356 ! 240 than that
observed for the photosensitization of Thd in the presence of
menadione, confirming the specific formation of the
thymidyl-5-yl radical from photolysis of 5-thiophenyl-
Figure 5. (a) HPLC/ESI-MS/MS chromatogram obtained for
methyl-20 -deoxyuridine (Scheme 6). The lack of an intense
the transition m/z 340 ! 224 of an aqueous solution of dUrd
peak for that m/z 356 ! 240 transition is in agreement with
containing DMPO photolyzed by UV-C using a UV lamp.
the proposed mechanism of formation of the DMPO adduct
Dose: 250 mJ. Inset: Quantities of nitrone c detected during
h detected upon photosensitization of Thd by menadione
similar experiments as a function of the UV-C irradiation dose
(vide supra).
received. (b) Quantity of generated nitrone c vs. power of a
According to these experiments and to the well-estab-
single UV-C laser flash used for irradiation of an aqueous
lished homolytic rupture of the C–S bond by UV-C
solution of dUrd and DMPO.
photolysis of the photoprecursor 5-thiophenylmethyl-20 -
deoxyuridine,23,24 the peak observed at 16.3 min on the
chromatograms shown in Fig. 3 could be unambiguously Therefore, additional experiments were carried out to confirm
attributed to the nitrone g produced after spin trapping of that the detected product corresponded to nitrone c.
radical e by DMPO. First, it was checked that the nitrone c is not produced
following an over-irradiation of dUrd. This was done by
using a single shot of a nanosecond pulsed laser to irradiate
Detection of spin adducts from dUrd, Urd, solutions of dUrd and DMPO. In such an experiment all
and RNA strands photons are absorbed in 5 ns, before the completion of
Aqueous solutions of dUrd containing DMPO were UV-C photochemical reactions. The corresponding chromatograms
irradiated and then analyzed by HPLC/ESI-MS/MS using are similar to those recorded for samples irradiated by UV
the MRM method developed for the detection of nitrone c lamps and they confirm the presence of nitrone c. In addition,
(Fig. 5(a), to be compared with Fig. 1). Interestingly, UV-C the amount of nitrone c formed was also found to be linear
irradiation of dUrd was found to generate nitrone c with a with the light intensity (Fig. 5(b)) at least for laser intensities
quantum yield determined to be 5.5  10
5. The quantum lower than 1.5  107 W  cm
2 per pulse. Such a linear
yield of formation of this product from UV-C-induced relationship indicates that nitrone c is produced via a
decomposition of dUrd is only 2.2 and 20 times lower than monophotonic mechanism. For higher intensities, the
those determined for 5-BrdUrd and 5-IdUrd, respectively. relationship is no longer linear, most probably due to the
The UV-C irradiation of 20 -deoxyuridine was not expected2 overwhelming biphotonic ionization of dUrd.
to produce radical e, at least not in significant amounts. To confirm that upon UV-C irradiation of dUrd the
detected product could be attributed to nitrone c, with a
DMPO residue linked to the C5 position of dUrd, deuterated
O O
analogues of dUrd were prepared. Irradiation of solutions of
Ph
HN S HN D6-dUrd (deuterium on C6) and DMPO led to chromato-
UV-C
O N O N
grams for the transition m/z 341 ! 225 that were identical to
dR dR e
the chromatograms obtained for the transition m/z 340 ! 224
with non-deuterated dUrd (data not shown). This indicates
Scheme 6. Formation of radical e by photolysis of the photo- that DMPO is not bound to the C6 position of dUrd. In
precursor 5-thiophenylmethyl-20 -deoxyuridine. addition, when a sample containing bideuterated 2D5,6-dUrd
Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm
 Detection of spin-trapped radicals of nucleosides 2241

(deuterium on C6 and C5 positions, the sample containing additional transitions observed for nitrone c( obtained from
also 30% of contaminating D6-dUrd) was irradiated in the irradiation of 5-BrUrd, i.e. transitions m/z 356 ! 206,
presence of DMPO, only monodeuterated nitrone c was 356 ! 163 and 356 ! 135 (Fig. 6). The presence of these
detected using the transition m/z 341 ! 225 (data not shown). transitions upon UV-C-mediated photolysis of RNA in the
These experiments demonstrate that the hydrogen in the C6 presence of DMPO confirms that nitrone c( can be produced
position is conserved and that the hydrogen in the C5 in RNA, and that the product can be detected by HPLC/ESI-
position of dUrd is lost during the formation of the nitrone c. MS/MS subsequent to RNA hydrolysis.
These results are consistent with the proposed formation of
nitrone c from the initially generated 20 -deoxyuridin-5-yl
DISCUSSION
radical a.
In order to check if UV-C irradiation of RNA could also The objective of the present work was to evaluate the
induce the formation of a uridin-5-yl radical, a HPLC/ESI- potential of HPLC/ESI-MS/MS to detect spin trapping
MS/MS method was developed for the detection of the adducts derived from nucleosides, as already described
corresponding ribonucleoside of nitrone c. Aqueous for amino acids.18,19 First, photolysis of 5-halogenated
solutions containing 5-BrUrd or Urd and DMPO were UV- pyrimidine derivatives, a well-known system for the
C irradiated and analyzed by HPLC/ESI-MS/MS using a generation of the pyrimidin-5-yl radical, was performed in
constant neutral loss of 132 Da corresponding to a ribose the presence of DMPO, a classic spin trap. Potential spin
moiety. After irradiations of 5-BrUrd (Fig. 6, left panel) and trapping adducts were then detected by HPLC/ESI-MS/MS
Urd (Fig. 6, middle panel), only the transition m/z 356 ! 224 using a neutral loss scan method. The overwhelming
produces a signal and the corresponding chromatograms detected adduct was found to correspond to the expected
exhibited peaks at similar retention times. All these nitrone c in which the pyrimidine moiety is linked at C5 to
observations are consistent with the formation of the nitrone DMPO. The proposed structure of the adduct was confirmed
c(, an analogue of the nitrone c but containing a ribose moiety by NMR analysis. Application of the HPLC/ESI-MS/MS
instead of a 20 -deoxyribose moiety. A more sensitive MRM method was then extended to other systems known to induce
method was then developed for the detection of nitrone c( several pyrimidine radicals. The obtained results are in
using the transition m/z 356 ! 224 and additional specific agreement with the previously described photoreactivities of
transitions (Fig. 6). 5-thiophenylmethyl-20 -deoxyuridine and thymidine photo-
To detect the formation of nitrone c( in RNA, RNA sensitized by menadione bisulfite.
solutions were UV-C irradiated in the presence of DMPO. The detection of neutral losses of 116 Da or of 132 Da, (mass
Thereafter, RNA was precipitated to eliminate the excess of of deoxyribose and ribose moieties, respectively) in HPLC/
DMPO and then digested enzymatically to corresponding ESI-MS/MS is very specific and ensures that only com-
ribonucleosides. Using the specificity and sensitivity of the pounds derived from nucleosides are detected. Application
developed MRM method, a peak was observed at a retention of more sensitive MRM methods enabled us to detect
time similar to that of nitrone c(. Its intensity was found to amounts of spin trapping adducts as low as 0.5 pmol
increase with the irradiation time. The peak was not detected corresponding to a concentration in spin trapping adducts
in the absence of irradiation or when irradiation was as low as 10
8 mol  L
1.
performed in the absence of DMPO. Confirmation that this Difficulties in the mass spectrometric studies of spin
peak corresponds to nitrone c( was obtained by monitoring trapping adducts due to the different possible redox forms of
these adducts are well documented.16,17 In the absence of any
reducing agent, due to the higher stability of the nitrone
oxidized forms, the interpretation of the HPLC/ESI-MS/MS
data of spin trapping adducts derived from nucleosides is
simple, as described for those obtained for amino acids.19
Under our experimental conditions, only the oxidized forms
of the spin trapping adducts were detected, which leads to a
straightforward interpretation of mass spectral data.
The detection of nitrone c or its analogue c( after
irradiation of either dUrd or Urd and RNA is very puzzling
since the formation of the radical e by UV-C irradiation of the
pyrimidine has never been reported.2 However, our
experimental results clearly establish the formation of
nitrone c and the corresponding analogue c(. First, the
obvious similarity between the chromatograms obtained
Figure 6. HPLC/ESI-MS/MS detection of nitrone c( gener- after irradiation of halogenated and non-halogenated nucleo-
ated from (i) UV-C irradiation of 5-BrUrd in the presence of sides demonstrates that the same adducts are produced in
DMPO, left panel; (ii) UV-C irradiation of Urd in the presence both cases and the NMR analysis of nitrone c unambiguously
of DMPO, middle panel; (iii) UV-C irradiation of RNA in the confirms the proposed structure. Secondly, the experiments
presence of DMPO (right panel) subsequent to enzymatic performed with mono- and bideuterated dUrd confirm that
digestion. For all chromatograms, the four transitions used to the hydrogen at the C5 position is lost during the formation
detect nitrone c( are shown. of the spin trapping adducts. Thirdly, the observation of the
Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm
2242 V. Maurel, J.-L. Ravanat and S. Gambarelli

adduct after only one laser shot indicates that nitrone c is not Acids, vol. 1, Morrison H (ed). Wiley-Interscience: New York,
produced by the secondary oxidation of an initially 1990; 1.
3. Görner H, Schulte-Frohlinde D. Radiat. Phys. Chem. 1995; 45:
generated photoproduct. In spite of all these indications, 677.
the results described here concerning dUrd and Urd are not 4. Janzen EG. Acc. Chem. Res. 1971; 4: 31.
sufficient by themselves to propose a consistent mechanism. 5. Tordo P. Spin trapping: recent developments and appli-
cations. In Electron Paramagnetic Resonance, Specialist Period-
Such a mechanistic study is beyond the scope of the ical Reports, vol. 16, Gilbert BC, Davies JD, McLauchlan AD
methodological study of the detection of spin trapping (eds). Royal Society of Chemistry: Cambridge, 1998; 116.
adducts by HPLC/ESI-MS/MS reported in this work. 6. Rustgi S, Riesz P. Int. J. Radiat. Biol. 1978; 33: 21.
7. Lagercrantz C. J. Am. Chem. Soc. 1973; 95: 220.
Finally, the detection of the nitrone c( after irradiation of 8. Ho WF, Gilbert BC, Davies MJ. J. Chem. Soc., Perkin Trans. 2
RNA subsequent to its enzymatic digestion to nucleosides 1997; 2533.
shows that HPLC/ESI-MS/MS of DMPO spin trapping 9. Davies MJ, Gilbert BC, Hazlewood C, Polack NP. J. Chem.
Soc., Perkin Trans. 2 1995; 13.
adducts of nucleoside free radicals is a versatile method 10. Catterall H, Davies MJ, Gilbert BC, Polack NP. J. Chem. Soc.,
which can be used to detect transient radicals in the Perkin Trans. 2 1993; 2039.
chemistry and in the photochemistry of RNA and DNA. 11. Murali Krishna C, Decarroz C, Wagner JR, Cadet J, Riesz P.
Photochem. Photobiol. 1987; 46: 175.
12. Hazlewood C, Davies MJ. J. Chem. Soc., Perkin Trans. 2 1995;
895.
CONCLUSIONS 13. Luxford C, Morin B, Dean RT, Davies MJ. Biochem. J. 1999;
344: 125.
HPLC/ESI-MS/MS is a very specific, sensitive and infor- 14. Hawkins CL, Davies MJ. Chem. Res. Toxicol. 2002; 15: 83.
mative method for the detection of DMPO spin trapping 15. Yue Qian S, Tomer KB, Yue GH, Guo Q, Kadiiska MB, Mason
adducts of reactive free radicals derived from nucleosides. RP. Free Radical Biol. Med. 2002; 33: 998.
16. Iwahashi H, Parker CE, Mason RP, Tomer KB. Anal. Chem.
This method was successfully applied to detect free radicals 1992; 64: 2244.
from three well-established radical decomposition pathways 17. Iwahashi H. J. Chromatogr A 1996; 753: 235.
involving nucleoside derivatives. In addition, a spin trapping 18. Detweiler CD, Deterding LJ, Tomer KB, Chignell CF,
Germolec D, Mason RP. Free Radical Biol. Med. 2002; 33: 364.
adduct from a new radical pathway was detected in the 19. Deterding LJ, Ramirez DC, Dubin JR, Mason RP, Tomer KB.
photolysis of Urd and dUrd. J. Biol. Chem. 2004; 279: 11600.
Experiments with RNA strands show that DMPO spin 20. McIntire GL, Blount HN, Stronks HJ, Shetty RV, Janzen EG.
J. Phys. Chem. 1980; 84: 916.
adducts of base-centred radicals can also be detected by 21. Frelon S, Douki T, Ravanat J-L, Pouget J-P, Tornabene C,
HPLC/ESI-MS/MS after enzymatic hydrolysis, which opens Cadet J. Chem. Res. Toxicol. 2000; 13: 1002.
a wide range of applications for this technique. 22. Regulus P, Spessotto S, Gateau M, Cadet J, Favier A, Ravanat
J-L. Rapid Commun. Mass Spectrom. 2004; 18: 2223.
23. Bellon S, Gasparutto D, Romieu A, Cadet J. Nucleosides,
Acknowledgements Nucleotides Nucleic Acids 2001; 20: 967.
The authors gratefully acknowledge Dr Sophie Bellon and Dr 24. Bellon S, Ravanat J-L, Gasparutto D, Cadet J. Chem. Res.
Didier Gasparutto for the gift of 5-thiophenylmethyl-20 -deox- Toxicol. 2002; 15: 598.
25. Romieu A, Bellon S, Gasparutto D, Cadet J. Org. Lett. 2000; 2:
yuridine, Dr Colette Lebrun for technical assistance concern- 1085.
ing mass spectrometry, and Dr Jean Cadet for helpful 26. Ravanat J-L, Douki T, Duez P, Gremaud E, Herbert K, Hofer
comments and discussions. T, Lasserre L, Saint-Pierre C, Favier A, Cadet J. Carcinogenesis
(London) 2002; 23: 1911.
27. Rabi JA, Fox JJ. J. Am. Chem. Soc. 1973; 95: 1628.
28. Ravanat J-L, Remaud G, Cadet J. Arch. Biochem. Biophys. 2000;
REFERENCES 374: 118.
1. Ravanat J-L, Douki T, Cadet J. J. Photochem. Photobiol., B: Biol. 29. Rahn RO, Sellin HG. Photochem. Photobiol. 1982; 35: 459.
2001; 63: 88. 30. Campbell JM, Schulte-Frohlinde D, von Sonntag C. Photo-
2. Cadet J, Vigny P. The photochemistry of nucleic acids. In chem. Photobiol. 1974; 20: 465.
Bioorganic Photochemistry, Photochemistry and the Nucleic 31. Foote CS. Photochem. Photobiol. 1991; 54: 659.

Copyright # 2006 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2235–2242
DOI: 10.1002/rcm