Simultaneous determination of five oxidative DNA lesions in human urine

Article abstract of DOI:10.1021/tx980194k

A method, involving a HPLC prepurification followed by a GC/MS analysis, has been set up for the measurement of nucleic acid oxidation products in human urine. For this purpose, isotopically labeled internal standards have been prepared and used for isotope dilution mass spectrometric detection. Using this approach, four oxidized DNA bases, i.e., 5-hydroxyuracil, 5- (hydroxymethyl)uracil, 8-oxo-7,8-dihydroadenine, and 8-oxo-7,8- dihydroguanine, together with 8-oxo-7,8-dihydro-2'-deoxyguanosine have been simultaneously quantified in human urine samples. The levels of the oxidized nucleic acid constituents, as expressed in picomoles per milliliter, were determined to be, in decreasing order: 8-oxo-7,8-dihydroguanine (583 ± 376) > 5-(hydroxymethyl)uracil (121 ± 56) > 5-hydroxyuracil (58 ± 23) > 8-oxo- 7,8-dihydro-2'-deoxyguanosine (30 ± 15) > 8-oxo-7,8-dihydroadenine (7 ± 4). Attempts to determine the amount of 5,6-dihydroxy-5,6-dihydrothymine, 5- hydroxycytosine, and 2,6-diamino-4-hydroxy-5-formamidopyrimidine using the above HPLC-GC/MS method were unsuccessful.

Full text of DOI:10.1021/tx980194k

8
02
Chem. Res. Toxicol. 1999, 12, 802-808
Sim u lta n eou s Deter m in a tion of F ive Oxid a tive DNA
Lesion s in Hu m a n Ur in e
J ean-Luc Ravanat, Pascale Guicherd, Zorana Tuce, and J ean Cadet*
Laboratoire “L e´ sions des Acides Nucl e´ iques”, Service de Chimie Inorganique et Biologique,
D e´ partement de Recherche Fondamentale sur la Mati e` re Condens e´ e, CEA Grenoble,
1
7 Avenue des Martyrs, F-38054 Grenoble Cedex 9, France
Received August 17, 1998
A method, involving a HPLC prepurification followed by a GC/MS analysis, has been set up
for the measurement of nucleic acid oxidation products in human urine. For this purpose,
isotopically labeled internal standards have been prepared and used for isotope dilution mass
spectrometric detection. Using this approach, four oxidized DNA bases, i.e., 5-hydroxyuracil,
5
8
-(hydroxymethyl)uracil, 8-oxo-7,8-dihydroadenine, and 8-oxo-7,8-dihydroguanine, together with
-oxo-7,8-dihydro-2′-deoxyguanosine have been simultaneously quantified in human urine
samples. The levels of the oxidized nucleic acid constituents, as expressed in picomoles per
milliliter, were determined to be, in decreasing order: 8-oxo-7,8-dihydroguanine (583 ( 376)
>
5-(hydroxymethyl)uracil (121 ( 56) > 5-hydroxyuracil (58 ( 23) > 8-oxo-7,8-dihydro-2′-
deoxyguanosine (30 ( 15) > 8-oxo-7,8-dihydroadenine (7 ( 4). Attempts to determine the
amount of 5,6-dihydroxy-5,6-dihydrothymine, 5-hydroxycytosine, and 2,6-diamino-4-hydroxy-
5
-formamidopyrimidine using the above HPLC-GC/MS method were unsuccessful.
In tr od u ction
Reactive oxygen species, formed continuously in living
oxidation of the nucleoside triphosphates pool, and diet.
In particular, the level of 8-oxoGua determined in rat
urine has been shown to be greatly influenced by the diet
cells, are known to cause oxidation to important biological
molecules (1, 2). Among others, nucleic acids are a critical
target (3, 4) since their integrity is a requisite for cell
survival and development. Despite the protection af-
forded by antioxidant defense, oxidative DNA modifica-
tions are formed in vivo. Therefore, extensive repair of
oxidatively modified DNA has been developed by cells in
order to minimize the biological effects (mutagenesis,
carcinogenesis) of such modifications. The repair products
of these lesions may be excreted in urine as free bases
(
9). From these observations, it seems to be important
to simultaneously determine, in the same urine sample,
several DNA oxidation products, in order to validate the
biological aspect of this approach.
A mass spectrometric detection represents a suitable
approach for the measurement of oxidized DNA compo-
nents in urine. Interestingly, isotopically labeled internal
standards (10) could be used to compensate for eventual
losses of the targeted products to be detected during the
work-up. Therefore, the methodology previously de-
scribed for 5-HMUra (8, 11) has been extended to the
detection of other oxidized nucleic acid constituents in
urine. A HPLC separation is used to prepurify the urine
sample to which the isotopically labeled internal stan-
dards have been added. Thereafter, following derivati-
zation, samples are analyzed by isotope dilution mass
spectrometry at the output of the gas chromatograph. The
method has been successfully applied to determine the
level of five DNA oxidation products including four bases,
(DNA N-glycosylase activities) or nucleosides (nucleotide
excision repair). Therefore, noninvasive methods based
on the measurement of oxidized bases or/and nucleosides
in human urine, used as biomarkers of in vivo DNA
oxidation (5), have been developed during the last decade.
Most of the assays have focused on the detection of 8-oxo-
,8-dihydro-2′-deoxyguanosine (8-oxodGuo)1 using the
7
widely used electrochemical detection coupled to HPLC
separation (for a review, see ref 6). As an example, active
tobacco smoking has been found to increase the oxidative
DNA damage rate by 35-50% as estimated from the
urinary excretion of 8-oxodGuo (7). However, in a more
recent study, no significant changes have been observed
between smokers and nonsmokers in the urinary levels
of 5-HMUra (8). Oxidative DNA base lesions measured
in urine could have other origins than the repair of
oxidized DNA modifications. These include cell death,
5
-OHUra, 5-HMUra, 8-oxoAde, 8-oxoGua, and one nu-
cleoside, 8-oxodGuo. As a main striking result, large
variations between the levels of the different biomarkers
in human urine were observed.
Ma ter ia ls a n d Meth od s
Ch em ica ls. Cyanoacetic acid, 8-oxoGua, collidine, bromine,
and dithionite were obtained from Sigma (St. Louis, MO).
*
To whom correspondence should be addressed. Telephone: (33) 4
1
5
13
[
3
N , C]Guanidine was obtained from ICON Services INC
7
8
6 88 49 87. Fax: (33) 4 76 88 50 90. E-mail: jcadet@cea.fr.
1
(Summit, NJ ). [ N
paraformaldehyde, and [1,3- N
Eurisotop (Gif-sur-Yvette, France). [2-amino-7,9- N
15
, C]- and [ N
13
15
]urea, [ C]- and [ H
13
2
2
]-
Abbreviations: 8-oxodGuo, 8-oxo-7,8-dihydro-2′-deoxyguanosine;
2
2
1
5
-oxoGua, 8-oxo-7,8-dihydroguanine; 8-oxoAde, 8-oxo-7,8-dihydro-
]uracil were purchased from
2
adenine; 5-HMUra, 5-(hydroxymethyl)uracil; 5-OHCyt, 5-hydroxycy-
tosine; 5-OHUra, 5-hydroxyuracil; FapyGua, 2,6-diamino-5-formyl-
amino-4-hydroxypyrimidine; Thy glycols, 5,6-dihydroxy-5,6-dihy-
drothymine; HPLC-EC, HPLC coupled to electrochemical detection;
TMS, trimethylsilyl.
15
13
3
,8- C]8-
3
,8- C]8-oxoAde, synthesized as
1
5
13
OxoGua and [6-amino-7,9- N
previously described (12), were a gift from Dr. R. Turesky
(Nestl e´ Research Centre, Lausanne, Switzerland).
1
0.1021/tx980194k CCC: $18.00 © 1999 American Chemical Society
Published on Web 08/04/1999

Measurement of Oxidative DNA Lesions in Human Urine
Chem. Res. Toxicol., Vol. 12, No. 9, 1999 803
1
5
13
Ch em ica l Syn t h eses. (1) [2-amino-7,9- N
3
,8- C]8-Oxo-
lesions was found to be, at least, 95% as determined by HPLC
analysis. In addition, no unlabeled derivatives were detected
in the isotopically labeled internal standards as determined by
GC-MS analysis. According to the sensitivity of the detection,
this means that less than 0.2% of the unlabeled product is
contaminating the labeled internal standard.
dGuo. The isotopically labeled nucleoside was enzymatically
1
5
13
synthesized from [2-amino-7,9- N
described (12).
(
3
,8- C]8-oxoGua as previously
1
5
13
2) Isotopically Labeled 5-HMUra. [1,3- N
2
,2- C]-Labeled
1
5
13
uracil was prepared as previously described (13) using [ N
2
, C]-
]5-HMUra was synthesized by
1
5
13
15
18
urea. [1,3- N
2
3
,2,5-methyl- C
2
2
Sa m p le P r ep a r a tion . Typically, 200 pmol of [ N , O]5-
1
15
13
15
13
15
13
15
13
condensing [ C]paraformaldehyde with [1,3- N
2
,2- C]uracil.
2 2 2 2 2
OHUra, [ N , C]5-OHCyt, [ N , C ]5-HMUra, [ N , C ]Thy
1
5
2
15
13
15
13
[
[
1,3- N
2
,5-methyl- H
2
]5-HMUra was prepared by condensing
3 3
glycols, and [ N , C]FapyGua, 100 pmol of [ N , C]8-oxoAde,
150 pmol of [ N , C]8-oxodGuo, and 300 pmol of [ N , C]8-
3 3
2
15
15
13
15
13
H
(
2
]paraformaldehyde with [1,3- N
2
]uracil (14).
1
5
13
3) [1,3- N
2
,2- C]5-OHCyt was prepared as previously de-
oxoGua were added to 3 mL of human urine. The resulting
solution was filtered through a 0.22 µm filtration membrane
and concentrated to 1 mL by evaporation under reduced
pressure prior to injection onto the HPLC system.
scribed (15).
(
1
5
13
4) [1,2-amino-3- N ,2- C]FapyGua. The isotopically labeled
FapyGua was synthesized according to Nelson (16) with the
3
following modifications. Typically, 22.5 mL of ethanol containing
HP LC P u r ifica tion . The HPLC system (Gilson, Middleton,
MI) consisted of two Model 306 pumps controlled by a third
Model 305 pump. Urine samples were injected using a 231 XL
auto injector onto a Hypersil C18 5 µm (250 × 10 mm i.d.)
column (Interchim, Montlu c¸ on, France) equipped with a Hy-
persil C18 5 µm (10 × 2 mm i.d.) guard column. An isocratic
elution consisting of 100% 25 mM ammonium formate was
maintained during 15 min, the flow rate being 2 mL/min.
Thereafter, acetonitrile was added linearly to the mobile phase
to reach an 8% level within 45 min. Then, the column was
washed with 100% acetonitrile for 15 min and was equilibrated
with 25 mM ammonium formate for 15 min prior to the next
injection. The elution was monitored by a 111B UV spectrometer
set at 280 nm connected to a D2500 integrator (Hitachi, Tokyo,
J apan). The output of the UV detector was connected to a FC204
fraction collector (Gilson) used for the collection of 1 min
fractions (2 mL).
1
g of cyanoacetic acid and 150 µL of sulfuric acid were refluxed
for 2.5 h. Thereafter, 20 mL of water was added, and the solution
was neutralized with 1 N NaOH. Extraction with diethyl ether
afforded 1 g of ethyl cyanoacetate (yield 70%). In a subsequent
step, 170 µL of ethyl cyanoacetate was added to 130 mg of
1
5
13
[
3
N , C]guanidine hydrochloride in 10 mL of sodium ethanolate
previously obtained by adding 0.1 g of sodium to 10 mL of
ethanol. The resulting solution was refluxed for 3 h. Thereafter,
unreactive ethyl cyanoacetate was removed by extraction with
diethyl ether. The aqueous phase was collected and dried, and
the resulting precipitate was solubilized into 5 mL of cold (4
°C) 10% acetic acid. Then, 200 mg of sodium nitrate was added,
and vigorous stirring was maintained for 1 h at 4 °C. The
resulting purple nitrate derivative was collected by centrifuga-
tion and washed with cold water in order to remove sodium
chloride and the excess of sodium nitrate. The precipitate was
suspended into 5 mL of H
2
O, and the solution was heated up to
However, to determine the retention time of the different
lesions, 300 pmol of each of the DNA modifications was injected
onto the column, and all the 1 mL fractions collected between 5
and 45 min were analyzed by GC/MS. This allowed us to
determine the retention time of the different DNA lesions under
the HPLC conditions used. Thereafter, for sample analysis, only
the fractions of interest containing the analyzed DNA modifica-
tions were collected and subsequently lyophilized. The resulting
products were dissolved in 100 µL of water and transferred into
silylation vials. Samples were lyophilized again and derivatized
in 100 µL of a (1:1) mixture of acetonitrile (silylation grade,
Pierce, Rockford, IL) and N,O-bis(trimethylsilyl)trifluoro-
acetamide for 30 min at 130 °C, prior to GC/MS analysis. Prior
to silylation, in order to improve the sensitivity of the detection
of 8-oxodGuo, fractions containing the latter lesion were further
treated by 88% formic acid for 30 min at 130 °C in order to
achieve a quantitative conversion of the nucleoside into the
corresponding free base 8-oxoGua.
8
0 °C. Then, dithionite was slowly added until the solution
becomes colorless. After cooling to room temperature and
subsequent drying, the precipitate was solubilized into 1 mL of
formic acid, and the resulting solution was stirred for 30 min
at room temperature. Then, formic acid was removed by
1
5
evaporation under reduced pressure, and [1,2-amino-3- N
3
,2-
1
3
C]FapyGua was purified by HPLC using an NH
Hypersil 5 µm NH
2
column
(
2
, 250 × 6 mm i.d.) from Interchim (Montlu-
c¸ on, France). For this purpose, the product was dissolved in a
minimum of the HPLC buffer consisting of a mixture (1:10) of
2
3
5 mM ammonium formate and CH CN before injection onto
the HPLC column. HPLC analysis (UV detection set at 270 nm)
showed that the solution contains almost exclusively [1,2-amino-
1
5
13
15
3
- N
3
,2- C]FapyGua (95%) and about 5% [1,2-amino-3- N
3
,2-
1
3
C]guanine. Collected fractions (k′ ) 5) were found to contain
1
5
13
isotopically pure [1,2-amino-3- N
GC/MS analysis mass (relative importance) m/ z 461 (100%) [M
4TMS], m/ z 446 (95%) [M + 4TMS-Me], m/ z 372 [M + 4TMS-
OTMS].
5) [1,3- N
]Thymine was obtained almost quantitatively by hydrogena-
3
,2- C]FapyGua (yield 60%):
+
GC/MS An a lysis. GC/MS analysis was performed on an HP
5
890 series II gas chromatograph (Hewlett-Packard, Les Ulis,
1
5
13
15
(
2
2 2 2
,2,5-methyl- C ]Thy Glycols. [1,3- N ,2,5-methyl-
France) equipped with a capillary column (0.25 mm × 30 m)
coated with a 0.25 µm film of methylsiloxane substituted by 5%
of phenylsiloxane (HP5-MS, Hewlett-Packard). The injection (1
µL) was performed in the splitless mode with the injector port
maintained at 260 °C. The constant flow rate (helium) was 1.6
L/min, and the transfer line was maintained at 280 °C. For the
detection of the oxidized pyrimidine bases and 8-oxoAde, the
oven temperature was maintained at 70 °C for 1 min. Then,
the temperature was subsequently increased to 300 °C at 20
°C/min, and then left at 300 °C for 2 min. Different conditions
were used for the GC analysis of 8-oxoGua and FapyGua.
Typically, the initial oven temperature was set at 120 °C and
then raised to 300 °C at a rate of 20 °C/min and left at the latter
temperature for 3 min. Detection of positive ions was provided
by an HP 5972 Mass Selective Detector operating in the electron
impact ionization mode (Hewlett-Packard). As previously de-
scribed (19, 20), the following target ions (and qualifiers) were
used for the detection of the different lesions eluted at the
following retention times: FapyGua, 6.69 min, m/ z 442 (457);
13C
tion of [1,3- N
1
5
13
2
2
,2,5-methyl- C ]5-HMUra (14). The resulting
labeled thymine was subsequently converted into Thy glycols
following the procedure described by Lustig et al. (17).
1
5
18
(
6) [1,3- N
2
,6-hydroxy- O]5-OHUra. The synthesis was
adapted from the already described synthesis (18). Typically,
1
5
1
1
0 µL of bromine was added to [1,3- N
2
]uracil in suspension in
1
8
mL of H
2
O. Then, the solution was maintained under
vigorous stirring at 5 °C for 15 min. Thereafter, unreactive
bromine was removed by air bubbling, and 200 µL of collidine
was added. The resulting solution was heated up to 120 °C for
1
h. The solution was then freeze-dried after a chloroform
1
5
extraction step aimed at removing collidine. [1,3- N
2
,6-hydroxy-
O]5-OHUra (90%) was obtained from the lyophilized product
by HPLC purification (k′ ) 1) using a Hypersil 5 µm C18 (250
6 mm i.d.) column eluted with 5% of methanol in water
1
8
×
(
detection set at 260 nm).
The same procedures were applied for the synthesis of the
corresponding unlabeled derivatives, except for those com-
mercially available. The purity of all the synthesized DNA
1
5
13
[ N
3
, C]FapyGua, 6.69 min, m/ z 446 (461); Thy glycols, 6.57
1
5
13
min, m/ z 259 (433); [ N
2
2
, C ]Thy glycols, 6.57 min, m/ z 262

8
04 Chem. Res. Toxicol., Vol. 12, No. 9, 1999
Ravanat et al.
F igu r e 1. Chemical structures of the DNA lesions to be
measured in human urine by a HPLC GC/MS assay.
1
5
,18O]5-OHUra,
2
(
437); 5-OHUra, 6.99 min, m/ z 329 (344); [ N
6
[
.99 min, m/ z 333 (348); 5-OHCyt, 8.0 min, m/ z 328 (343);
1
5
13
3
N , C]5-OHCyt, 8.0 min, m/ z 331 (346); 5-HMUra, 7.42 min,
1
5
13
m/ z 358 (343); [ N
2
, C
2
]5-HMUra, 7.42 min, m/ z 362 (347);
F igu r e 2. Panel A: GC-MS calibration curve for 8-oxoAde
obtained by plotting the relative ratio of the target ion of
8-oxoAde (ion 367) and of the isotopically labeled compound (ions
371) versus increasing relative amounts of both compounds.
Panel B: Typical GC/MS chromatograms obtained for the
urinary detection of 8-oxoAde.
1
5
2
[
N
2
, H
min, m/ z 352 (367); [ N
-oxoGua, 6.72 min, m/ z 440 (455); [ N
min, m/ z 444 (459).
2
]5-HMUra, 7.41 min, m/ z 362 (347); 8-oxoAde, 9.06
1
5
13
3
, C]8-oxoAde, 9.06 min, m/ z 356 (371);
1
5
13
8
3
, C]8-oxoGua, 6.72
Resu lts
of the targeted compounds to occur. This observation
particularly applies to FapyGua and Thy glycols. 5-O-
HCyt was found to be eluted later than the two other
above mentioned lesions, and, therefore, derivatization
of the oxidized base was possible. However, the GC/MS
analyses indicated the overwhelming presence of con-
taminants that prevents the quantification of 5-OHCyt,
even by using the highly specific selected ion monitoring
mode.
HP LC Sep a r a tion . To determine the retention time
of the different DNA modifications, 300 pmol of each
standard was injected onto the HPLC column, and the
retention time was determined by GC/MS analysis of the
collected fractions. This procedure has been used since a
simpler method consisting of UV detection of the stan-
dards at the output of the HPLC column would have
required the injection of large amounts of these modified
bases or nucleosides. Therefore, the latter method has
been excluded in order to avoid contamination of the
HPLC system with a large excess of the standard that
could affect the accuracy of the measurements. The
different studied lesions (Figure 1) were collected in the
following fractions: FapyGua and Thy glycols, 6-7 and
GC-MS An a lysis. The advantage of using an isotopi-
cally labeled internal standard is illustrated by the
precision of the quantification. As an example, the
calibration curve obtained for 8-oxoAde is shown in
Figure 2. The slope of the curve close to 1 indicates that
the labeled and unlabeled compounds have similar GC-
MS behaviors. In addition, no unlabeled compound could
be detected in the labeled DNA lesion. Similar calibration
curves (Figures 3-6) were obtained for the other DNA
bases with a correlation coefficient always higher than
0.99. The limit of the detection of the assay is close to
100 fmol for the bulk of the different DNA lesions, with
the exception of 8-oxoGua which is about 10 fold higher.
Since the measurement was performed on 3 mL of urine,
this represented, at the worst, a level of about 0.3 pmol/
mL of urine, far lower than the levels detected in urine
samples. To evaluate the accuracy and reproducibility of
the assay, the quantification of the different DNA lesions
was resumed on the same urine sample. The average
variation coefficient (Table 1) of the assay was around
5%, with the highest value (9%) for 8-oxoAde which was
found to be present in the smallest amount in urine. In
addition, a spike experiment was performed to evaluate
the accuracy of the analytical method. Typically, a urine
sample was spiked with a known amount (20, 50, 100,
7
5
-8 min; 5-OHCyt, 8-9 min; 5-OHUra, 9-10 min;
-HMUra, 13-14 and 14-15 min; 8-oxoGua, 21.5-22.5
and 22.5-23.5 min; 8-oxoAde, 34.5-35.5 and 35.5-36.5
min; 8-oxodGuo, 42-43 and 43-44 min. The correspond-
ing isotopically labeled internal standards were collected
in the same fractions as the unlabeled derivatives since
no differences in the retention time between the labeled
and unlabeled derivatives were observed except for
deuteriated 5-HMUra (vide infra). The slowest HPLC-
eluting modified DNA bases and nucleosides were col-
lected into two fractions. Both fractions were analyzed
separately, and the fraction that contained the highest
amount of the lesion (usually the first collected one) was
used for quantification.
F a p yGu a , Th y Glycols, a n d 5-OHCyt. These DNA
modifications were found to be eluted in the close vicinity
of the void volume of the C18 reverse phase column.
Therefore, the lyophilized fractions were found to contain
a large amount of salts that prevent the derivatization

Measurement of Oxidative DNA Lesions in Human Urine
Chem. Res. Toxicol., Vol. 12, No. 9, 1999 805
F igu r e 3. Panel A: GC-MS calibration curve for 5-OHUra.
Panel B: Typical chromatogram obtained for the urinary
detection of 5-OHUra.
F igu r e 5. Panel A: GC-MS calibration curve for 8-oxoGua.
Panel B: Typical chromatogram obtained for the urinary
detection of 8-oxoGua.
F igu r e 6. Panel A: GC-MS calibration curve for 8-oxodGuo.
Panel B: Typical chromatogram obtained for the urinary
detection of 8-oxodGuo. The ions used for detection are those of
8-oxoGua since the nucleoside is hydrolyzed by formic acid
treatment prior to GC-MS analysis.
F igu r e 4. Panel A: GC-MS calibration curve for 5-HMUra.
Panel B: Typical chromatogram obtained for the urinary
detection of 5-HMUra.
1
8
0, and 20 pmol of 5-OHUra, 5-HMUra, 8-oxoGua,
-oxoAde, and 8-oxodGuo per milliliter urine, respec-
5-HMUr a . The level of 5-HMUra was determined in
tively) of the targeted DNA bases. Analysis of the
collected fraction obtained after HPLC purification indi-
cates a good recovery for the five DNA lesions (97 ( 6,
one human urine sample using either the deuteriated
1
5
2
labeled internal standard [ N
2
, H
2
]5-HMUra or
1
5
2
[ N , C ]5-HMUra. The results, representing the average
2 2
9
5
6 ( 7, 99 ( 5, 101 ( 2, and 98 ( 4 for 5-OHUra,
-HMUra, 8-oxoGua, 8-oxoAde, and 8-oxodGuo, respec-
and standard deviation of 10 analyses of the same urine
sample, are reported in Figure 7. A large variation in the
results is observed when the deuteriated standard is
tively).

8
06 Chem. Res. Toxicol., Vol. 12, No. 9, 1999
Ravanat et al.
Ta ble 1. Rep lica te An a lyses of th e F ive DNA Lesion s
Deter m in ed in th e Sa m e Ur in e Sa m p lea
nucleosides detected in urine could have other origin(s)
than the repair of oxidized DNA. In this respect, it has
been shown that the level of 8-oxoGua in rat urine is
greatly influenced by diet (9).
A few DNA modifications have been already detected
in urine. These include 8-oxoGua, Thy glycols, and the
related 2′-deoxyribonucleosides, together with 5-HMUra
DNA
lesion
5-OHUra 5-HMUra 8-oxoGua 8-oxoAde 8-oxodGuo
n
8
6
138.5
5.6
5
722
26
8
6
average
stdv
variation
64.1
3.9
6.1
8.0
0.7
8.8
36.2
2.1
5.8
%
4.0
3.6
(
8
for a review, see ref 6). It should also be mentioned that
-oxoAde has been detected in human urine but not
a
Results are expressed in picomoles of lesion per milliliter of
urine.
quantified (21). Most of the methods employed for the
measurement of oxidative DNA modifications in urine
have not involved mass spectrometric detection, and thus
no isotopically labeled internal standards have been used.
Therefore, the accuracy of the quantification could be
questioned since measurement of such molecules in urine
represents a challenging analytical problem.
Therefore, in order to validate the approach, a method
has been developed for the simultaneous determination
of several oxidative DNA lesions (Figure 1) in human
urine. This is based on the GC/MS measurement of the
modifications after initial HPLC purification as already
described for 5-HMUra (8, 11, 22). One of the major
advantages of the assay is the accuracy provided by the
mass spectrometric detection. The GC-MS chromato-
grams of the different DNA lesions obtained after initial
HPLC prepurification of urine samples are illustrated in
Figures 2-6. They show the quasi-absence of contami-
nating peaks that could affect the quantification. In order
to improve the specificity of the assay, two specific ions
of each DNA lesion were used for the GC-MS detection.
The target ion (the most intense one) was used for
quantification purpose, whereas for each urine sample
the ratio between this ion and the second detected one
was compared to the ratio obtained for the standard. The
fact that the same ratio was observed strongly suggests
that the eventuality of a contamination by a coeluting
impurity is very low. Furthermore, the detection by mass
spectrometry allows the use of isotopically labeled inter-
nal standards that could correct for any eventual losses
of the compounds during work-up. However, care has to
be taken in the choice of the isotopically labeled internal
standard, as illustrated in Figure 7. The standard devia-
tion of 10 independent determinations of 5-HMUra in a
F igu r e 7. Relative ratio between the detected amount of the
internal standard versus 5-HMUra determined in a human
1
5
2
urine sample using either the [ N
2
, H
2
] (white column) or the
1
5
13
[
N , C ] (gray column) internal standard. The results repre-
2 2
sent the average and standard deviation of 10 independent
determinations.
Ta ble 2. Am ou n ts of Oxid ized DNA Lesion s In clu d in g
5
-OHUr a , 5-HMUr a , 8-OxoGu a , 8-OxoAd e, a n d 8-Oxod Gu o
a
in Hu m a n Ur in e
DNA
lesion 5-OHUra 5-HMUra 8-oxoGua 8-oxoAde 8-oxodGuo
pmol/mL 58 ( 23 121 ( 56 583 ( 376
7 ( 4
30 ( 15
a
Results, expressed in pmol/mL, represent the average (
standard deviation of the analyses of, at least, 10 different urine
samples.
used, while a more accurate determination of 5-HMUra
1
5
13
is obtained using the [ N
2 2
, C ] internal standard.
8
-OxoGu a , 8-oxod Gu o, 8-oxoAd e, a n d 5-OHUr a .
The GC/MS analysis assay after initial HPLC prepuri-
fication was found to be a suitable method for the
•
15
13
measurement of five OH -mediated DNA lesions in hu-
2 2
urine sample is about 6% when [ N , C ]5-HMUra is
man urine samples. Typical chromatograms obtained for
the analysis of the different DNA lesions are shown in
Figures 2-6.
The urinary levels of the oxidized nucleic acid con-
stituents, expressed in picomoles per milliliter, in de-
creasing order are the following: 8-oxoGua (583 ( 376)
used as the internal standard. However, this deviation
is close to 65% when the deuteriated labeled internal
standard is utilized (Figure 7), and larger variations are
observed without labeled internal standard (data not
shown). The higher variation observed when the deuteri-
ated internal standard is used can be explained by a
significant difference in the HPLC retention time of the
deuteriated derivative compared to unlabeled 5-HMUra.
This is rationalized in terms of an important isotopic
effect exerted by the deuterium atoms. Therefore, the
collected fraction (1 min) could be enriched in one of the
two 5-HMU derivatives, and thus the quantification may
be over- or underestimated. One possibility to minimize
this difficulty would be to collect larger HPLC elution
volumes in order to have both labeled and unlabeled
HMU in the same fraction. However, the collected
fraction will also contain additional contaminants which
could affect the GC/MS analysis. A more accurate ap-
>
5-HMUra (121 ( 56) > 5-OHUra (58 ( 23) > 8-oxo-
dGuo (30 ( 15) > 8-oxoAde (7 ( 4) (Table 2).
Discu ssion
The biological relevance of using the measurement of
oxidized nucleobases and nucleosides in urine as bio-
markers of in vivo DNA oxidation is still a matter of
debate. It is well established that DNA oxidation occurs
within cells and that DNA repair systems are efficient
to prevent the accumulation of these lesions in the
genome. The latter modified bases and nucleosides, after
being removed from DNA, are considered as poor sub-
strates for the enzymes involved in nucleotide synthesis
and thus are generally excreted in urine without further
metabolism (5, 6). However, oxidized DNA bases and
1
5
13
2 2
proach consists of using a [ N , C ]-labeled internal
standard which minimizes differences in retention times
between the labeled and the unlabeled derivatives (10).
Thus, no enrichment of the normal oxidized base versus

Measurement of Oxidative DNA Lesions in Human Urine
Chem. Res. Toxicol., Vol. 12, No. 9, 1999 807
the isotopically internal standard is expected during the
HPLC purification, even if small fractions are collected.
This explains why the synthesis of the [ N , C ]-labeled
2 2
internal standard has been made in addition to the
deuteriated labeled internal standard already available
inter-individual variation for 8-oxoGua that could be
explained, at least partly, from the fact that the presence
of this oxidized DNA base in urine could have a dietary
origin (9). It should be noted that the values obtained
for 8-oxoGua in the present study are about 20-fold
higher than those previously published (27). The origin
of this discrepancy may be attributed, at least partly, to
an underestimation of the level of 8-oxoGua in the
previous study since no internal standard was used since
the measurements were performed by HPLC-EC. This
could also be explained by the poor recovery of 8-oxoGua
from urine samples as previously reported (28). Results
obtained by HPLC-EC measurement after improving the
recovery of the modified base from the precipitate of
stored frozen urine (28) are within the same range as
the presently reported data. The amounts of the other
DNA modifications detected are within the same range
as those already obtained (for a review, see ref 6).
However, the level of 8-oxodGuo using the HPLC GC/
MS approach, described in the present study, is slightly
higher than those obtained using the HPLC-EC detection
method (7, 29, 30).
Results obtained using the HPLC GC/MS approach
indicate an almost constant ratio between the level of
the oxidized lesion in each of the analyzed human urine
samples, except for 8-oxoGua (data not shown). There-
fore, a urine sample with a higher level of 8-oxodGuo
usually exhibits high levels of 5-OHUra, 5-HMUra, and
8-oxoAde. Additional work is required in order to confirm
these preliminary observations and to determine intra-
individual variations for each of the detected DNA
lesions. This will have to be extended to the study of
inter-individual variations of the level of nucleobases in
human urine under well-defined conditions of oxidative
stress.
It may be concluded that we have developed an
accurate analytical method that enables the detection of
five oxidative DNA lesions in human urine. The main
advantage of the assay as compared with previous
methodologies is the potentiality of using isotopically
labeled internal standards since the measurement is
provided by a mass spectrometric detector. However, the
approach also has limitations since the assay is restricted
to compounds which are significantly retained on the
ODS column used for the prepurification. This work
constitutes a major effort in order to validate the biologi-
cal relevance of the measurement of oxidized DNA bases
and nucleosides in urine as biomarkers of oxidative
stress. The development of HPLC coupled to a tandem
mass spectrometric detection using isotopically labeled
internal standards (31) will certainly facilitate such a
goal.
1
5
13
(
11).
The HPLC GC/MS approach was found to be suitable
for the measurement of 5-OHUra, 5-HMUra, 8-oxoAde,
-oxoGua, and 8-oxodGuo (Table 2). Repeated analyses
8
of the same urine sample were achieved (Table 1) in order
to demonstrate the accuracy of the detection. In addition,
a spike experiment has been performed (vide supra) to
evaluate the precision of the quantification. In both cases,
low variation coefficients were obtained, showing the
advantage of using isotope dilution mass spectrometry
for the measurement of oxidized DNA bases in urine.
However, poorly retained DNA bases (under the HPLC
conditions used) are eluted in fractions containing high
amounts of salts that prevent accurate GC/MS analysis.
Therefore, polar oxidized DNA bases such as Thy glycols
and FapyGua could not be measured using this approach.
Two different attempts were made to overcome this
difficulty. First, boronate columns (23) were used to
isolate Thy glycols after the initial HPLC purification,
as previously used for urine samples (24). Normal phase
amino column chromatography (25) was also applied for
the purification of FapyGua. The GC/MS analysis of the
collected fractions using the two above-described methods
has clearly indicated the presence of overwhelming
contaminants. The other approach consisted of chemically
transforming the polar lesions into a more hydrophobic
derivative that could be eluted later under the HPLC
conditions used. For this purpose, fractions that con-
tained Thy glycols and FapyGua were either treated by
HI to convert Thy glycols into Thy (24), or treated by
formic acid to initiate the cyclization of FapyGua into Gua
(26). Unfortunately, the GC/MS measurement of the
resulting unmodified DNA bases (Thy and Gua) in the
treated samples was poorly sensitive and reproducible.
In addition, 5-OHCyt, which is more retained on the C18
column than the two above-mentioned lesions, was found
to be eluted in a fraction containing impurities. Again,
the presence of these contaminants prevents the quan-
tification of the modification by GC/MS. Therefore, this
does not make possible the application of the HPLC GC/
MS assay for the quantitation of these DNA lesions. The
latter examples illustrate the difficulties in detecting
polar oxidized DNA bases in urine.
However, the present HPLC GC/MS approach allows
the quantification of five DNA modifications including
5
(
5
-OHUra, 5-HMUra, 8-oxoGua, 8-oxoAde, and 8-oxodGuo
Table 2). To our knowledge, this is the first time that
-OHUra and 8-oxoAde are measured and quantified in
Refer en ces
human urine, even if the latter oxidized purine base has
already been detected (21). It could be noted that the five
DNA base lesions were measured simultaneously. Inter-
estingly, in order to increase the sensitivity of the
detection of 8-oxodGuo, the nucleoside could be hydro-
lyzed into its base derivative, 8-oxoGua, using formic acid
treatment prior to GC/MS analysis.
The amounts of the different oxidized bases and
nucleosides detected in human urine are reported in
Table 2. Comparison of the results shows that 8-oxoGua
is present in larger amounts by comparison with the
other oxidative DNA lesions. This is followed by 5-HMU-
ra, 5-OHUra, 8-oxodGuo, and 8-oxoAde. There is a larger
(1) Ames, B. N., Shigenaga, M. K., and Hagen, T. M. (1993) Oxidants,
antioxidants, and the degenerative diseases of aging. Proc. Natl.
Acad. Sci. U.S.A. 90, 7915-7922.
(2) Halliwell, B., and Gutteridge, M. C. (1984) Oxygen toxicity, oxygen
radicals, transition metals and disease. Biochem. J . 219, 1-14.
(3) Halliwell, B., and Aruoma, O. I. (1991) DNA damage by oxygen-
derived species: Its mechanism and measurement in mammalian
systems. FEBS Lett. 281, 9-19.
(
4) Cadet, J ., Berger, M., Douki, T., and Ravanat, J .-L. (1997)
Oxidative damage to DNA: formation, measurement and biologi-
cal significance. Rev. Physiol. Biochem. Pharmacol. 31, 1-87.
5) Shigenaga, M. K., Gimeno, C. J ., and Ames, B. N. (1989) Urinary
(
8
-hydroxy-2′-deoxyguanosine as a biological marker of in vivo
oxidative DNA damage. Proc. Natl. Acad. Sci. U.S.A. 86, 9697-
9701.

8
08 Chem. Res. Toxicol., Vol. 12, No. 9, 1999
Ravanat et al.
(
6) Loft, S., and Poulsen, H. E. (1996) Cancer risk and oxidative DNA
base damage in DNA by stable isotope-dilution mass spectrom-
etry. FEBS Lett. 315, 1-6.
(21) Stillwell, W. G., Xu, H.-X., Adkins, J . A., Wishnok, J . S., and
Tannenbaum, S. R. (1989) Analysis of methylated and oxidized
purines in urine by capillary gas chromatography-mass spec-
trometry. Chem. Res. Toxicol. 2, 94-99.
(22) Bianchini, F., Hall, J ., Donato, F., and Cadet, J . (1996) Monitoring
urinary excretion of 5-hydroxymethyluracil for assessment of
oxidative DNA damage and repair. Biomarkers 1, 178-184.
(23) J erkovic, B., Kung, H. C., and Bolton, P. H. (1998) Purification
of thymine glycol DNA and nucleosides by use of boronate
chromatography. Anal. Biochem. 255, 90-94.
(24) Cathcart, R., Schweirs, E., Saul, R. L., and Ames, B. N. (1984)
Thymine glycol and thymidine glycol in human and rat urine: a
possible assay for oxidative DNA damage. Proc. Natl. Acad. Sci.
U.S.A. 81, 5633-5637.
(25) Ravanat, J .-L., Douki, T., Incardona, M.-F., and Cadet, J . (1993)
HPLC separations of normal and modified nucleobases and
nucleosides on an amino silica gel column. J . Liq. Chromatogr.
16, 3185-3202.
(26) Douki, T., Martini, R., Ravanat, J .-L., Turesky, R. J ., and Cadet,
J . (1997) Measurement of 2,6-diamino-4-hydroxy-5-formamidopy-
rimidine and 8-oxo-7,8-dihydroguanine in isolated DNA exposed
to gamma radiation in aqueous solution. Carcinogenesis 18,
2385-2391.
(27) Suzuki, J ., Inoue, Y., and Suzuki, S. (1995) Changes in the urinary
excretion level of 8-hydroxyguanine by exposure to reactive
oxygen-generating substances. Free Radical Biol. Med. 18, 431-
436.
(28) Helbock, H. J ., Beckman, K. B., Shigenaga, M. K., Walter, P. B.,
Woodall, A. A., Yeo, H. C., and Ames, B. N. (1998) DNA oxidation
matters: the HPLC-electrochemical detection assay of 8-oxo-
deoxyguanosine and 8-oxo-guanine. Proc. Natl. Acad. Sci. U.S.A.
95, 288-293.
damage in man. J . Mol. Med. 74, 297-312.
(7) Loft, S., Vistisen, K., Ewertz, M., Tjonneland, A., Overvad, K.,
and Poulsen, H. E. (1992) Oxidative DNA damage estimated by
8
-hydroxydeoxyguanosine excretion in humans: influence of
smoking, gender, and body mass index. Carcinogenesis 13, 2241-
247.
2
(
8) Bianchini, F., Donato, F., Faure, H., Ravanat, J .-L., Hall, J ., and
Cadet, J . (1998) Urinary excretion of 5-(hydroxymethyl)uracil in
healthy volunteers: effect of active and passive tobacco-smoke.
Int. J . Cancer 77, 40-46.
(
9) Park, E. M., Shigenaga, M. K., Degan, P. D., Korn, T. S., Kitzler,
J . W., Wehr, C. M., Kolachana, P., and Ames, B. N. (1992) Assay
of excised oxidative DNA lesions: isolation of 8-oxoguanine and
its nucleoside derivatives from biological fluids with a monoclonal
antibody column. Proc. Natl. Acad. Sci. U.S.A. 89, 3375-3379.
(
10) Mamer, O. A. (1996) Stable isotope dilution techniques. J .
Chromatogr. B 682, 182-183.
(
11) Faure, H., Incardona, M.-F., Boujet, C., Cadet, J ., Ducros, V., and
Favier, A. (1993) Gas chromatographic-mass spectrometric de-
termination of 5-hydroxymethyluracil in human urine by stable
isotope dilution. J . Chromatogr. 616, 1-7.
(
12) Stadler, R. H., Staempfli, A. A., Fay, L. B., Turesky, R. J ., and
1
5
13
Welti, D. (1994) Synthesis of multiply labeled [
3 1
N , C ]-8-oxo-
substituted purine bases and the corresponding 2′-deoxynucleo-
sides. Chem. Res. Toxicol. 7, 784-791.
(
13) Harada, K., and Suziki, S. (1976) The new synthesis of uracil and
1
,3-dimethyluracil. Tetrahedron Lett. 27, 2321-1322.
14) Cline, R. E., Fink, R. M., and Fink, K. (1959) Synthesis of
-substituted pyrimidines via formaldehyde addition. J . Am.
(
5
Chem. Soc. 81, 2521-2527.
(
15) Douki, T., Delatour, T., Bianchini, F., and Cadet, J . (1996)
Observation and prevention of an artifactual formation of oxidized
DNA bases and nucleosides in the GC-EIMS method. Carcino-
genesis 17, 347-353.
16) Nelson, V. C. (1996) Synthesis of isotopically labelled DNA
degradation products for use in mass spectrometric studies of
cellular DNA damage. J . Labelled Compd. 38, 713-723.
17) Lustig, M. J ., Cadet, J ., Boorstein, R. J ., and Teebor, G. W. (1992)
Synthesis of the diastereomers of thymidine glycol, determination
of concentrations and rates of interconvention of their cis-trans
epimers at equilibrium and demonstration of different alkali
lability within DNA. Nucleic Acids Res. 20, 4839-4845.
18) Eaton, M. A. W., and Hutchinson, D. W. (1973) Poly(5-hydroxy-
cytidylic acid). Biochim. Biophys. Acta 319, 281-287.
(29) Tagesson, C., K a¨ llberg, M., Klintenberg, C., and Starkhammar,
H. (1995) Determination of urinary 8-hydroxydeoxyguanosine by
automated coupled-column high performance liquid chromatog-
raphy: A powerfull technique for assaying in vivo oxidative DNA
damage in cancer patients. Eur. J . Cancer 31A, 934-940.
30) Loft, S., Fischer-Nielsen, A., J eding, I. B., Vistisen, K., and
Poulsen, H. E. (1995) 8-Hydroxydeoxyguanosine as a urinary
biomarker of oxidative DNA damage. J . Toxicol. Environ. Health
(
(
(
4
0, 391-404.
31) Ravanat, J .-L., Duretz, B., Guiller, A., Douki, T., and Cadet, J .
1998) Isotope dilution high-performance liquid chromatography-
(
(
(
(
19) Dizdaroglu, M. (1985) Application of capillary gas chromatogra-
phy-mass spectrometry to chemical characterization of radiation-
induced base damage of DNA: implications for assessing DNA
repair processes. Anal. Biochem. 144, 593-603.
electrospray tandem mass spectrometry assay for the measure-
ment of 8-oxo-7,8-dihydro-2′-deoxyguanosine in biological samples.
J . Chromatogr. B 715, 349-356.
(
20) Dizdaroglu, M. (1993) Quantitative determination of oxidative
TX980194K