DNA damage by peroxynitrite characterized with DNA repair enzymes
DNA damage by peroxynitrite characterized with DNA repair enzymes
Bernd
Epe*
,
Daniel
Ballmaier
,
Ivan
Roussyn
1
,
Karlis
Briviba
1
and
Helmut
Sies
1
Institut für Pharmazie, Universität Mainz, Staudinger Weg 5, D-55099
Mainz
,
Germany
and
1
Institut für Physiologische Chemie I, Heinrich-Heine-Universität Düsseldorf, Postfach 101007, D-40001
Düsseldorf
,
Germany
Received August 26, 1996;
Revised and Accepted September 25, 1996
ABSTRACT
The DNA damage induced by peroxynitrite in isolated bacteriophage PM2 DNA was
characterized by means of several repair enzymes with defined substrate specificities. Similar results were obtained with peroxynitrite itself and with 3-morpholinosydnonimine (SIN-1), a compound generating the precursors of peroxynitrite, nitric
oxide and superoxide. A high number of base modifications sensitive to Fpg
protein which, according to HPLC analysis, were mostly 8-hydroxyguanine residues, and half as many single-strand breaks were observed, while the numbers of oxidized
pyrimidines (sensitive to endonuclease III) and of sites of base loss
(sensitive to exonuclease III or T4 endonuclease V) were relatively low. This
DNA damage profile caused by peroxynitrite is significantly different from that
obtained with hydroxyl radicals or with singlet molecular oxygen. The effects
of various radical scavengers and other additives (
t
-butanol, selenomethionine, selenocystine, desferrioxamine) were the same for single-strand breaks and Fpg-sensitive modifications and indicate that a single reactive
intermediate but not peroxynitrite itself is responsible for the damage.
INTRODUCTION
Peroxynitrite is a relatively long-lived oxidant that may be produced
in vivo
in the reaction of nitric oxide with the superoxide anion radical (
1
), e.g. by activated macrophages, neutrophils and endothelial cells. Under physiological conditions the major reactive form of peroxynitrite (p
K
a
= 6.8) is the protonated form, peroxynitrous acid (HOONO), which decomposes
with a half-life of ~1 s (
1
). Peroxynitrite can cause strand breaks in plasmid supercoiled DNA (
2
) and in DNA of eukaryotic cells (
3
). Both nitration and nitrosation products [8-nitroguanine (
4
-
6
), 4-hydroxy- 5-nitrosooxy-guanine (
6
)] and various oxidation products [8-hydroxyguanine (8-oxoG), 8-hydroxyadenine (8-oxoA) and an oxazolone] (
7
,
8
) were identified in studies with isolated DNA or nucleosides. In DNA of
macrophages treated with interferon-[gamma] and lipopolysaccharides to produce nitric oxide and superoxide, 8-oxoG, 5-hydroxymethyluracil and formamidopyrimidines were
detected, in addition to DNA deamination products (
9
). Peroxynitrite-induced DNA modifications were found to cause mutations preferentially at
G-C base pairs (
10
).
The mechanism underlying the oxidation of biomolecules by HOONO has not yet been
fully established. Initially, it was suggested that hydroxylations and
nitrations were caused by hydroxyl radicals and nitrogen dioxide radicals that
are generated by homolysis of peroxynitrite (
11
). This has been challenged more recently on the basis of thermodynamic and
kinetic considerations, results obtained with radical scavengers and relatively
low yields (1-4%) of hydroxyl radicals detected in spin trapping experiments (
2
,
12
-
15
). Rather, a reactive intermediate structurally similar to that involved in the
isomerisation to nitrate, possibly an activated form of
trans
-peroxynitrous acid, has been claimed responsible for the reactions (
12
,
16
).
Repair endonucleases, which selectively recognize certain DNA base modifications
and sites of base loss, can be used as sensitive probes to quantify oxidative DNA modifications (
17
-
20
). When several repair endonucleases are used in parallel, DNA damage profiles are
obtained, which can serve as fingerprints of the ultimate DNA damaging species and thus allow the identificaton of the species responsible for the DNA damage.
We studied here the DNA damage profiles induced by peroxynitrite and by 3-morpholinosydnonimine (SIN-1), a compound that decomposes to generate the peroxynitrite precursors nitric oxide and superoxide, and compared the damage profiles with
that induced by hydroxyl radicals. In addition, the effects of various
scavengers were determined. The results indicate that a single species of a
reactivity comparable with hydroxyl radicals, but neither peroxynitrite itself nor hydroxyl radicals, are responsible for the DNA damage observed, which consists predominantly of 8-oxoG and single-strand breaks (SSBs).
MATERIALS AND METHODS
DNA repair endonucleases and chemicals
DNA from bacteriophage PM2 (PM2 DNA) was prepared according to the method of
Salditt
et al
. (
21
). More than 97% was in the supercoiled form, as determined by the method
described below. Formamidopyrimidine-DNA glycosylase (Fpg protein) (
22
) and endonuclease III from
Escherichia coli
were kindly provided by S. Boiteux, Fonteney aux Roses, France. T4 endonuclease V was partially purified by the method described by Nakabeppu
et al
. (
23
) from the
E.coli
strain A 32480 (
uvrA, recA
, F'lac IQ1) carrying the plasmid ptac-denV (kindly provided by L. Mullenders, Leiden, Netherlands) after induction with isopropyl- [beta]-D-thiogalactopyranoside. Exonuclease III was
purchased from Boehringer, Mannheim, Germany. All repair endonucleases were
tested for their incision at reference modifications (i.e. thymine glycols
induced by OsO
4
, AP sites by low pH and 8-oxoG by methylene blue plus light) under the applied assay conditions to ensure that the correct substrate modifications are fully recognized and no incision at non-substrate modifications takes place (
24
). 3-morpholinosydnonimine (SIN-1),
N
-hydroxypyridine-2-thione (2-HPT), desferrioxamine mesylate and most other chemicals
were obtained from Sigma-Aldrich Chemie, Deisenhofen, Germany.
Synthesis of peroxynitrite
Peroxynitrite was synthesized from sodium nitrite and hydrogen peroxide (H
2
O
2
) using a quenched-flow reactor as previously described (
1
,
25
) with minor modifications (
26
). Residual H
2
O
2
was eliminated by passage of the peroxynitrite solution over powdered manganese
dioxide. Peroxynitrite was concentrated by freeze fractionation. Its
concentration was determined spectrophotometrically at 302 nm ([epsilon] = 1670/M/cm).
Modification of PM2 DNA
For the reaction with SIN-1, PM2 DNA (10 [mu]g/ml) was incubated in phosphate buffer (10 mM NaH
2
PO
4
, 50 mM NaCl, pH 7.4) with SIN-1 (1-10 [mu]M) for 1 h at 37oC. The DNA was precipitated by ethanol-sodium acetate and redissolved in BE
1
buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA) for damage analysis.
For the reaction with preformed peroxynitrite, an alkaline solution of
peroxynitrite (see above) was diluted with water and added at room temperature
under vigorous stirring to PM2 DNA (10 [mu]g/ml) in phosphate buffer. The calculated final peroxynitrite concentrations
were between 2.5 and 100 [mu]M. In all cases, the pH of the reaction mixtures did not increase >7.6.
After 3 min at room temperature, the DNA was precipitated by ethanol-sodium acetate and redissolved in BE
1
buffer. After a second precipitation and dissolving in BE
1
buffer, the DNA was used for damage analysis. In control experiments, the
peroxynitrite solution was incubated with phosphate buffer for 3 min at room
temperature to decompose the peroxynitrite, before the DNA was added.
In some of the experiments, 10-500 mM
t
-butanol, 1-300 [mu]M desferrioxamine mesylate or other additives were present
during the exposure to peroxynitrite or SIN-1.
Quantification of DNA modifications in PM2 DNA
The DNA relaxation assay used to quantify endonuclease-sensitive modifications and strand breaks in PM2 DNA has been described earlier (
19
,
27
). It makes use of the fact that supercoiled PM2 DNA is converted by either a
SSB or the incision of a repair endonuclease into a relaxed (nicked) form which
migrates separately from the supercoiled form in agarose gel electrophoresis. Quantification of both forms of DNA by fluorescence scanning after staining of
the gel with ethidium bromide allows the calculation of the average number of
SSBs per PM2 molecule (10
4
bp) or-if an incubation with a repair endonuclease precedes the gel
electrophoresis-the number of SSBs plus endonuclease-sensitive sites. All data were corrected for the numbers of sites
observed in untreated control DNA (<0.1 site/10
4
bp in all cases).
Modification of calf thymus DNA and HPLC analysis
Calf thymus DNA (50 [mu]g) was incubated with SIN-1 (12 [mu]M) for 1 h at 37oC in 200 [mu]l phosphate buffer. The DNA was precipitated, redissolved in 100 [mu]l phosphate buffer and incubated with 2.5 [mu]g Fpg protein for 1 h at 37oC. The excised bases were separated from the DNA on an anion exchange minicolumn
(Qiagen, Hilden, Germany) and applied to a reversed-phase HPLC column (Nucleosil 100-5 C18, 250-4, Macherey-Nagel, Düren, Germany) equipped with an electrochemical detector (ESA
Coulochem II; detection potential 300 mV) and eluted with 50 mM phosphate buffer pH 5.5, which contained 5% methanol, at a flow rate of 1.2 ml/min. When authentic 8-hydroxyguanine was added to unmodified DNA before the incubation with Fpg protein, the recovery was found to be 78%. Control incubations were carried out without SIN-1 or without Fpg protein.
Fe(III)-EDTA-dependent nitration of 4-hydroxyphenyl- acetate by peroxynitrite
The iron complex Fe(III)-EDTA was prepared by mixing solutions of iron-(III)-sulfate and sodium-EDTA in the ratio 1:1. Peroxynitrite (final concentration 50 [mu]M) was added to 4-hydroxyphenylacetate (4-HPA, 1 mM) in 80 mM sodium phosphate
buffer containing Fe(III)-EDTA (0.5 mM) and various concentrations of
t
-butanol while vigorous stirring to start the reaction at room temperature,
pH 7.2. Samples were incubated for 30 min at room temperature. Alternatively,
as a control, 4-HPA was added 5 min after peroxynitrite to buffer alone. The final pH was
measured after the reaction to account for the slight alkaline shift caused by
the NaOH used to stabilize the stock solution of peroxynitrite. The pH was then
adjusted to 10.0-10.6 with 3 M aqueous NaOH. The yield of 4-hydroxy-3-nitrophenylacetate (NO
2
-HPA) was calculated from its absorbance at 430 nm ([epsilon] = 4400/M/cm) (
25
).
Assay of peroxynitrite-mediated oxidation of dihydro- rhodamine 123
Peroxynitrite-mediated oxidation of dihydrorhodamine 123 in presence of various
concentrations of
t-
butanol was carried out as described previously (
28
) with minor modifications (
29
). Fluorescence intensities were measured with a spectrophotometer (LS-5, Perkin-Elmer, Norwalk, USA) with excitation and emission wavelengths of 500 and 536 nm, respectively, at room temperature. The fluorescence intensity was linearly related to the rhodamine
concentration between 0 and 400 nM.
RESULTS
DNA damage profiles
PM2 DNA in phosphate buffer was exposed to peroxynitrite or to SIN-1. The latter compound upon thermal decomposition under aerobic conditions
generates superoxide and nitric oxide, which combine to yield peroxynitrite (
30
,
31
). Subsequently, the DNA was analysed for SSBs and modifications sensitive to various repair endonucleases (Table
1
). In the case of peroxynitrite, the extent of DNA damage with increasing peroxynitrite concentration was found
to follow a saturation curve, but the ratio of the DNA modifications was
constant (Fig.
1
, upper panel). In contrast, a linear dose-response was observed with SIN-1 (Fig.
1
, lower panel). No DNA modifications were detected when DNA was treated with the
decomposition products of peroxynitrite (Materials and Methods).
Recognition of DNA modifications by repair endonucleases used in this study
Repair endonuclease
Concentration applied
Recognition spectrum
a
Sites of base loss
b
Base modifications
Fpg protein
c
1 [mu]g/ml
+
8-oxoG
d
; Fapy
e
Endonuclease III
c
10 ng/ml
+
5,6-dihydropyrimidines; hyd
f
T4 endonuclease V
20 ng/ml
+
Py<>Py
g
Exonuclease III
1 [mu]g/ml
+
-
a
See refs 32-36.
b
For the recognition of sites of base loss oxidized in the 1' or 4' position, see ref. 27.
c
5-hydroxycytosine and 5-hydroxyuracil have recently been described as new substrates of both
Fpg protein and endonuclease III (37).
d
7,8-dihydro-8-oxoguanine (8-hydroxyguanine).
e
Formamidopyrimidines (imidazole ring-opened purines).
f
5-hydroxy-5-methylhydantoin and other ring-contracted and fragmented pyrimidines.
g
Cyclobutane pyrimidine photodimers.
HPLC analysis of Fpg-sensitive base modifications induced by SIN-1
The analysis of the modified bases excised by Fpg protein from calf-thymus DNA pretreated with SIN-1 by means of HPLC and an electrochemical detector established the
presence of 8-oxoG in the damaged DNA (Fig.
3
). After correction for the recovery of authentic 8-oxoG (78%; see Materals and Methods) and for the small amount of 8-oxoG observed in untreated calf thymus DNA (0.29 pmol; Fig.
3
C), the amount detected (1.4 pmol; Fig.
3
A) corresponded to a generation by SIN-1 of 1.8 8-oxoG residues/ 10
4
bp. These are 75% of the Fpg-sensitive modifications generated under the same reaction conditions in
PM2 DNA. According to the damage profile induced by SIN-1 (Fig.
2
), another 10% of the Fpg-sensitive modifications were sites of base loss (sensitive to exonuclease
III and T4 endonuclease V). Either experimental errors or a minor fraction of
Fpg-sensitive base modifications which are not 8-oxoG (Table
1
) could account for the 15% residual modifications.
Effects of
t
-butanol
The inhibition by
t
-butanol of DNA damage by peroxynitrite and SIN-1 and, for comparison, hydroxyl radicals generated by photodecomposition of 2-HPT, are shown in Figure
4
. For all three damaging agents, 50% inhibition was observed at ~20 mM
t
-butanol. Fpg-sensitive modifications and SSBs were affected to the same extent.
Both these lesions, therefore, most probably emanate from the same reactive
intermediate that is scavenged by
t
-butanol.
Effects of other scavengers
As shown in Table
2
, SOD inhibited DNA damage by SIN-1, which generates peroxynitrite via superoxide, but not by peroxynitrite itself. Catalase had no effect on the damage. The iron chelator desferrioxamine mesylate inhibited the generation by SIN-1 and peroxynitrite of both SSBs and Fpg-sensitive modifications by 50-70% (Table
2
). Under similar reaction conditions, desferrioxamine had much less influence on
the DNA damage induced by hydroxyl radicals, generated by photodecomposition of
2-HPT (Table
2
).
.
Effects of various scavengers on DNA damage induced by peroxynitrite, SIN-1 and photodecomposition of 2-HPT
Presence of
Relative number of modifications (%)
a
Peroxynitrite (40 [mu]M)
b
SIN-1 (10 [mu]M)
b
2-HPT (2 mM) + light
c
SSB
Fpg
d
SSB
Fpg
SSB
Fpg
SOD (20 [mu]g/ml)
101 +-19
95 +- 16
26 +- 3
27 +- 10
120 +- 16
85 +- 21
Catalase (315 U/ml)
96 +-15
93 +- 2
90 +- 12
113 +- 19
107 +- 11
115 +- 23
DSF
e
(1 [mu]M)
27 +-10
30 +- 7
78 +- 17
107 +- 22
89 +- 21
93 +- 8
(30 [mu]M)
32 +- 4
29 +- 3
45 +- 10
49 +- 7
77 +- 5
83 +- 3
(300 [mu]M)
25 +- 3
27 +- 5
31 +- 3
41 +- 15
65 +- 13
76 +- 12
Cystine (1 mM)
123 +- 22
73 +- 25
Selenocystine (1 mM)
58 +- 23
48 +- 25
Methionine (0.1 mM)
57 +-13
41 +- 21
51 +- 5
31 +- 7
70 +- 7
81 +- 7
Selenomethionine (0.1 mM)
23 +- 9
9 +- 6
8 +- 2
5 +- 3
79 +- 11
85 +- 10
a
Number of modifications observed without additives defined as 100%.
b
Values represent means (+- SD) of four to six independent experiments.
c
Values represent means (+- SD) of two to three independent experiments.
d
DNA modifications sensitive to Fpg protein.
e
Desferrioxamine mesylate.
The selenium compounds, selenomethionine (0.1 mM) or selenocystine (1 mM), decreased the number of Fpg-sensitive DNA modifications caused by 40 [mu]M peroxynitrite more effectively than methionine (0.1 mM) or cysteine (1 mM), respectively (Table
2
).
DISCUSSION
The results provide evidence that peroxynitrite generates a distinct type of DNA
damage that deviates significantly both from that induced by hydroxyl radicals
and that induced by singlet oxygen (Fig.
2
). The inhibitory effect of
t
-butanol is similar to that observed under the same conditions for DNA
damage by hydroxyl radicals, generated either by photodecomposition of 2-HPT (Fig.
4
) or by ionizing radiation (
40
). This indicates that a single intermediate of a reactivity comparable with
hydroxyl radicals is responsible for the generation of both SSBs and Fpg-sensitive base modifications, the two principal modifications observed in
this analysis. Apparently, peroxynitrite itself is not scavenged by
t
-butanol, since
t
-butanol increased rather than decreased the yield of the products
generated by peroxynitrite from 4-HPA and dihydrorhodamine. Therefore, the reactions of peroxynitrite with these substrates involve intermediates which are distinct from those involved in the reaction of peroxynitrite
with DNA. That the presence of hydroxyl radical scavengers can cause such
increases in products of reactions of peroxynitrite has been reported previously (
2
,
41
). The experiments with catalase and SOD (Table
2
) demonstrate that a Fenton reaction catalysed by traces of transition metals is not involved in the DNA damage
observed. Yet, desferrioxamine inhibits the formation of both SSBs and Fpg-sensitive modifications by 50-70% (Table
2
). The effect is in agreement with results by Denicola
et al
. (
42
) which indicated that the reaction intermediate similar to hydroxyl radicals (probably derived from
trans
-peroxynitrous acid), but neither peroxynitrite itself nor
cis
- peroxynitrous acid acts as a one-electron oxidant of desferrioxamine. As evident from the data obtained with 2-HPT plus light (Table
2
), a similar reaction of desferrioxamine with hydroxyl radicals is inefficient or does not occur. The non-linear dose-dependence observed with peroxynitrite, but not with SIN-1 (Fig.
1
), suggests that the concentration of the reactive intermediate responsible for
the DNA damage is diminished at high peroxynitrite concentrations (which are
not reached during SIN-1 decomposition).
The very efficient inhibition of peroxynitrite-induced DNA damage by selenomethionine and selenocystine confirms the recently demonstrated
protection by selenocompounds from SSB formation in supercoiled DNA (
43
) and may be of relevance
in vivo
. The big difference betwen the selenocompounds and their sulfur analogues is not
observed for the damage generated by hydroxyl radicals (Table
2
). Most probably, the selenocompounds-in contrast with
t
-butanol and desferrioxamine-react with peroxynitrite itself, as has been demonstrated previously for ebselen (
26
,
44
).
The HPLC analysis reveals that most of the bases excised by Fpg protein from SIN-1-modified calf thymus DNA are 8-oxoG residues. This lesion is known to be premutagenic, giving
rise predominantly to GC -> TA transversions (
45
,
46
). Recently, GC -> TA transversion were indeed found to be the prevailing type of mutation
when a plasmid pretreated with peroxynitrite was replicated in bacteria or
mammalian cells (
10
).
The Fpg protein concentration dependence of the incisions at the peroxynitrite-induced modifications gives no indication that other (unknown) base modifications serve as additional substrates for Fpg protein in the case of the DNA damage induced by peroxynitrite.
The recognition by Fpg protein of oxazolones or 4-hydroxy-5-nitrosooxyguanine, which were demonstrated to be generated
from high concentrations (30 mM) of peroxynitrite and DNA or desoxyguanosine (
6
,
8
), has not been established yet. 8-Hydroxyadenine, which has been identified as a product of peroxynitrite
with DNA under the same conditions, is not a good substrate for Fpg protein (
35
,
47
). 8-Nitroguanine, another purine modification detected in calf thymus DNA treated with peroxynitrite, but not with SIN-1 (
5
), does not seem to be a major reaction product under the reaction conditions
employed in the present study, since this modification undergoes rapid
depurination at 37oC (
5
), while no thermolabile modifications were detected in PM2 DNA (Fig.
2
).
ACKNOWLEDGEMENTS
We thank S. Boiteux for providing Fpg protein and endonuclease III. This work
has been supported by the Deutsche Forschungsgemeinschaft (SFB 519-B2 and SFB 503-B1) and by the National Foundation for Cancer Research, Bethesda.
REFERENCES
1 Beckman,J.S., Beckman,T.W., Chen,J., Marshall,P. and Freeman,B.A. (1990) Proc. Natl. Acad. Sci. USA, 87, 1620-1624.MEDLINE Abstract
