| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 277, Issue 37, 34055-34066, September 13, 2002
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the Department of Microbiology, University of Illinois,
Urbana, Illinois 61801
Received for publication, April 24, 2002, and in revised form, June 14, 2002
When cells are exposed to external
H2O2, the H2O2
rapidly diffuses inside and oxidizes ferrous iron, thereby forming
hydroxyl radicals that damage DNA. Thus the process of oxidative DNA
damage requires only H2O2, free iron, and an
as-yet unidentified electron donor that reduces ferric iron to the
ferrous state. Previous work showed that H2O2
kills Escherichia coli especially rapidly when respiration
is inhibited either by cyanide or by genetic defects in respiratory
enzymes. In this study we established that these respiratory blocks
accelerate the rate of DNA damage. The respiratory blocks did not
substantially affect the amounts of intracellular free iron or
H2O2, indicating that that they accelerated damage because they increased the availability of the electron donor.
The goal of this work was to identify that donor. As expected, the
respiratory inhibitors caused a large increase in the amount of
intracellular NADH. However, NADH itself was a poor reductant of free
iron in vitro. This suggests that in non-respiring cells electrons are transferred from NADH to another carrier that directly reduces the iron. Genetic manipulations of the amounts of intracellular glutathione, NADPH, The appearance of oxygen in the atmosphere allowed some organisms
to develop more efficient metabolic schemes, but it also raised the
possibility that inadvertent chemical oxidations could disrupt
metabolism and damage biomolecules. Molecular oxygen is a triplet
species that is constrained to accept electrons one at a time, and its
univalent redox potential is low enough that it is unable to abstract
electrons from most structural biomolecules. Thus it does not directly
oxidize amino acids or nucleic acids. However, it can slowly oxidize
the reduced flavins of many redox enzymes, thereby generating a mixture
of superoxide (O Still, neither O Thus the process of DNA damage requires three steps as shown in
Reactions 2-4:
Reduced Flavins Promote Oxidative DNA Damage in Non-respiring
Escherichia coli by Delivering Electrons to Intracellular
Free Iron*
ABSTRACT
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
-ketoacids, ferredoxin, and thioredoxin
indicated that none of these was the direct electron donor. However,
cells were protected from cyanide-stimulated DNA damage if they lacked flavin reductase, an enzyme that transfers electrons from NADH to free
FAD. The Km value of this enzyme for NADH is much
higher than the usual intracellular NADH concentration, which explains
why its flux increased when NADH levels rose during respiratory inhibition. Flavins that were reduced by purified flavin reductase rapidly transferred electrons to free iron and drove a DNA-damaging Fenton system in vitro. Thus the rate of oxidative DNA
damage can be limited by the rate at which electron donors reduce free iron, and reduced flavins become the predominant donors in E. coli when respiration is blocked. It remains unclear whether
flavins or other reductants drive Fenton chemistry in respiring cells.
INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
When O
It is the first step (Reaction 2) that remains unclear. The
catalytic iron is thought to exist as a small pool of adventitious iron
that has dissociated from iron trafficking processes or metalloenzymes. This iron is presumably loosely bound to whatever biomolecules are
available, including DNA itself. (Because it is not stably integrated
into proteins, this iron is typically referred to as "free" iron.)
Because most ferrous chelates are rapidly oxidized by molecular oxygen,
the existence of a small pool of ferrous iron that is available
to react with H2O2 indicates that some countervailing reductive process must continuously re-reduce the free
iron. In fact, the efficiency with which exogenous
H2O2 damages DNA provides evidence that iron
re-reduction must be rapid. When Escherichia coli is treated
with 2.5 mM H2O2, DNA damage is
generated at a high rate for up to 20 min (13). At this concentration of H2O2 the Fenton reaction (76 M
1 s1 (17)) should oxidize
pre-existing reduced iron within seconds; the observation that damage
occurs steadily indicates that a reductant recycles the iron very efficiently.
Thus hydroxyl radicals are created when iron catalyzes the transfer of
an electron from a reductant to H2O2. That
reductant has not been identified. Superoxide itself is capable of
reducing free iron and functions in that capacity in some in
vitro Fenton systems (8, 9, 18). However, the intracellular
concentration of superoxide is so low (~1010
M) (19, 20), and its reduction of bound iron so slow
(~105 M1 s1)
(21), that the predicted half-time for this reaction in vivo is >10 h, far too long to be physiologically relevant. Although superoxide is mutagenic (22), this effect apparently stems from its
ability to increase the concentration of free iron by leaching it from
unstable iron-sulfur clusters (16, 23). Other cellular reductants have
been considered. Thiols (~103 M),
-ketoacids (~103 M), and NAD(P)H
(~104 M) are all abundant inside cells, and
each of these can reduce ferric iron in vitro (24-26). The
rates at which they do so have not been compared, so it is difficult to
predict which would be the predominant reductant in
vivo.
E. coli becomes fully resistant to oxidative DNA damage when
it is starved for carbon sources, and sensitivity returns when they are
re-administered (27). This pattern matches the depletion and
replenishment of these reductant pools, and it suggests that the
availability of a reductant for free iron can determine the rate at
which oxidative DNA damage occurs. In fact, when respiration in
E. coli was blocked by the addition of cyanide, the rate at which exogenous H2O2 killed cells was
dramatically potentiated (26). Such a result would be expected if NADH,
or another electron carrier in equilibrium with it, could efficiently
reduce free iron. The purpose of this work was to further explore
this phenomenon in order to identify that reductant.
| |
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Chemicals--
Ferric chloride, ferrous sulfate heptahydrate,
potassium cyanide (KCN), magnesium sulfate heptahydrate, and magnesium
chloride were purchased from Aldrich. Diethylenetriaminepentaacetic
acid, deferoxamine mesylate (desferrioxamine),
isopropyl-
-D-thiogalactopyranoside (IPTG),1 NADPH, NADH, EDTA,
kanamycin, ampicillin, hydrogen peroxide (H2O2) (30% w/v), tetracyline, reduced glutathione (GSH), chloramphenicol, Vibrio fisheri luciferase, isopropyl alcohol,
methanol, decanol, methanesulfonic acid methyl ester,
L-glutamic dehydrogenase, FAD, deamino-NADH,
5,5'-dithio-bis(2-nitrobenzoic acid), ADP, uracil, and
3-(2-pyridyl)-5,6-bis(2[5-furylsulfonic acid])-1,2,4-triazine (ferene) were purchased from Sigma. Beef liver catalase was from Roche
Molecular Biochemicals. Expand Long Template PCR kit was purchased from
Roche Molecular Biochemicals. dNTPs were from Promega, and PCR primers
were from Operon. Total protein was measured with the Coomassie protein
reagent (Pierce). -Mercaptoethanol and sodium citrate were from
Fisher. Water was purified with a Labconco Water Pro system.
Flavin reductase was overproduced and purified as described (28). The
purified fraction exhibited a single band at the expected molecular
weight, by SDS-PAGE analysis. The enzyme was stored at 
80 °C in 20 mM KPi, pH 7.0, 150 mM NaCl, and
30% glycerol.
Strains and Growth Media-- All strains used in this study were K-12 derivatives (Table I). Mutations were moved into strains by P1 transduction (29). Null mutations in recA, xth, cyo, cyd, ndh, and nuo can, in some backgrounds, cause defective growth in aerobic media. Therefore, to avoid the possible accumulation of suppressor mutations during the outgrowth of mutant colonies, these alleles were transduced anaerobically. Other strains were generated in air. Mutations in xthA were confirmed by sensitivity to H2O2 (30); in recA, by sensitivity to ultraviolet radiation; and in polA, by inability to grow on 0.01% methanesulfonic acid methyl ester LB plates. ndh mutations were screened by the inability of inverted membrane vesicles to oxidize NADH, after incubation at high pH to eliminate NdhI activity (2), and nuo mutations by the inability of their vesicles to oxidize deamino-NADH (2). Mutations in zwf, pnt, and fre were screened by measurement of glucose-6-phosphate dehydrogenase (31), transhydrogenase (32), and flavin reductase activities (33). fre gene disruptions were confirmed by PCR analysis.
|
Defined media contained either M9 or minimal A salts (29), 0.2% glucose, 0.2% acid-hydrolyzed casamino acids, 0.5 mM tryptophan, 1 mM MgSO4·7H2O, and 5 mg of thiamine per liter. LB broth contained 10 g of tryptone, 5 g of yeast extract, and 10 g of NaCl per liter supplemented with 0.2% glucose. Anaerobic experiments were conducted in a Coy chamber under 85% N2, 10% H2, and 5% CO2. Media and plates were made anaerobic by moving the samples into the anaerobic chamber while still hot and allowing them to equilibrate with the anaerobic atmosphere for at least 24 h prior to use. Plasmids and strains were maintained by an antibiotic selection with 50 µg/ml kanamycin, 12 µg/ml tetracycline, or 100 µg/ml ampicillin. Where indicated, 150 µg/ml chloramphenicol was used to inhibit protein synthesis. Genes under control of the Ptac promoter were induced with 50 µM IPTG.
The absorbances of cultures were monitored at 600 nm. Rates of oxygen consumption by aerobic cultures were measured using a Clarke-type oxygen electrode (Rank Brothers, Cambridge, UK).
Killing by H2O2 and Ultraviolet Radiation-- The rate at which 2.5 mM H2O2 killed cells was measured as described previously (27). Briefly, overnight cultures were diluted 1:1000 into fresh media and grown for at least four generations to an A600 of about 0.1-0.2. Immediately prior to a H2O2 challenge, cells were diluted to an A of ~0.1 in 37 °C medium, and H2O2 was added. At timed intervals aliquots were removed, diluted at least 1:100-fold into medium containing 130 units/ml catalase, plated onto LB glucose plates in top agar, and incubated overnight at 37 °C. Where indicated, 3 mM KCN was added 5 min prior to the addition of H2O2. For anaerobic challenges, overnight cultures were grown to mid-exponential phase and then diluted to an absorbance of 0.050 for H2O2 exposure. In these cases all media, diluent, and reagents were anaerobic, and plating was performed anaerobically. Where indicated, the filamentation of non-colony-forming cells was monitored by microscopy (34).
The sensitivity of cells to ultraviolet radiation was measured using exponentially growing cultures. Cells were placed into a sterile Petri dish and illuminated with a far-ultraviolet lamp. At intervals aliquots were removed, diluted into LB medium, plated in top agar, and incubated at 37 °C overnight. Survival was expressed as a percentage of the initial population.
Measurement of Intracellular Free Iron--
Intracellular free
iron was quantified by EPR spectroscopy (16). Exponentially growing
cells (1.5 liters, A600 = 0.1-0.2) in M9 medium
were centrifuged at 7800 × g at room temperature for 7 min and resuspended in 0.7% of the original volume of the same medium
containing 20 mM desferrioxamine. Where indicated 3 mM KCN was added simultaneously with the desferrioxamine.
The sample was incubated at 37 °C with shaking for 10 min,
centrifuged, washed with cold 20 mM Tris-HCl, pH 7.4, and
resuspended in 200 µl of the same buffer containing 10% glycerol.
The suspension was loaded into EPR tubes, frozen in a dry-ice bath, and
scanned with a Varian Century E-112 X-band spectrophotometer equipped with a Varian TE102 cavity and a Varian temperature controller. The
spectrometer settings used were as follows: field center, 1570 G;
receiver gain, 2,500; field sweep, 400 G; modulation amplitude, 1.25 G;
T, 125 °C; and power, 30 milliwatts. To determine the ability of H2O2 to oxidize intracellular free
iron, 2.5 mM H2O2 was added
following the initial resuspension of cells, and desferrioxamine was omitted.
Measurement of Polymerase-blocking DNA Lesions-- DNA damage can be quantified by the ability of DNA lesions to block PCR amplification of the DNA template (35). Cells were grown to A600 = 0.1-0.2, and they were then treated with H2O2 and/or KCN as described above. When the polA(Ts) strain was studied, cultures were shifted to 42 °C from 30 °C 30 min before the challenge. Catalase (26,000 units/ml) was added to scavenge the H2O2. The high amount was necessary to compensate for its substantial inhibition by cyanide. Total DNA was then extracted and purified according to procedures given for the DNeasy DNA extraction kit (Qiagen). DNA was quantitated using the Picogreen system (Molecular Probes).
The DNA template (25-100 ng) was amplified in a 50-µl reaction that
contained 0.3 µM primers, 350 µM dNTPs
(Promega), 0.2 µCi of [-32P]dATP, and 2.5 units of
the Taq polymerase-based Expand Long Template PCR system
(Roche Molecular Biochemicals). A 10-kb fragment of genomic DNA near
the fumC gene was chosen as the target sequence in our
experiments. The primers were 5' 
3' C-terminal
CAGGGCAACGGAACGGAACACCCGCCCAGAGCATAACC and 5' 3' N-terminal
CGGCGTGAACTCGCAAAATATTAACGATTCAGC. Amplification was accomplished by 30 cycles of denaturation (30 s, 94 °C), annealing (30 s, 62 °C),
and extension (10 min, 65 °C). Control experiments established that
the amount of PCR product was proportional to the amount of template
DNA over the range of template DNA used in these experiments. The PCR
product was resolved on a 1% agarose gel. The dried gels were exposed
to a phosphor screen, and the radioactive product was quantified using
the ImageQuant PhosphorImager program (Amersham Biosciences).
Measurement of Intracellular NADH and NADPH Pools-- The assays of NADH and NADPH were modified from previous procedures (19). In this study, in order to measure levels with sufficient precision, cells were concentrated prior to lysis. Oxygen was bubbled through the cell suspension in order to ensure that the dense culture did not become anaerobic.
Cells were grown in minimal A medium containing 0.2% glucose and 0.2% casamino acids, supplemented with 0.5 mM tryptophan, to A600 = 0.100-0.200. Cells were centrifuged for 3 min at room temperature, and the pellet was resuspended at 37 °C in the same medium containing 1% glucose at A600 = 1.5-2.0. The cell suspension was bubbled with pure oxygen for 5 min to restore aerobic metabolism, and a sample was then removed to determine the absorbance and number of viable cells. Where appropriate, 3 mM KCN was added for 3 min. In some cases, 2 mM H2O2 was added for 3 min. The H2O2 was then scavenged with 3500 units/ml catalase for 7 min. Cells were then lysed instantly with 0.125 M NaOH (final pH = 12.0). Alkaline extracts were incubated at 50 °C for 9 min in order to ensure the inactivation of enzymes and then neutralized with HCl to pH 7.6. Debris was removed by centrifugation at 4 °C (20,000 × g for 20 min). The supernatants were decanted into separate clean vials, and 2.0 mg/ml oxidized glutathione, 1.0 mg/ml chicken egg albumin, and 1.0 mM EDTA were added. The total reduced pyridine dinucleotide pool (NADH + NADPH) of each sample was measured directly in a luciferase reaction (below). NADH alone was measured after the sample was incubated with 40 units/ml glutathione reductase to oxidize the NADPH. NADPH alone was measured after the addition of 0.5 units/ml of E. coli inverted respiratory vesicles to oxidize the NADH. Both glutathione reductase and membranes were added to some samples, to oxidize both NADPH and NADH, thereby providing a depleted extract from which standard curves could be generated upon the addition of defined amounts of NAD(P)H. After 8-10 min incubation of samples with these enzymes at room temperature, the glutathione reductase and/or respiratory membranes were inactivated by the addition of 6 M NaOH to pH 12.0. The pH was then re-adjusted to 7.2.
NAD(P)H levels were determined by the addition of 800 µl of sample to
100 µl of reaction mixture (composed of 10 mM
KPi, 0.05% -mercaptoethanol, 0.1 mg/ml decanal, and 0.2 mg/ml FMN, at pH 7.0) and 200 µl of luciferase (4.5 mg/ml in water)
in a Turner TD-20e luminometer. NADH and NADPH concentrations were
calculated from the integrated light emission. Stock solutions of NADH
and NADPH that were used to establish standard curves were themselves calibrated by absorbance at 340 nm (
= 6.22 mM1 cm1).
To test the stability of NADH and NADPH during sample preparation, known amounts were added to initial cell lysates in control experiments. The ultimate recovery was >90%. Cell viability was checked before and after H2O2 exposure, and the synergistic toxicity of H2O2 and cyanide described in other experiments was reproduced.
Peroxidase Activity-- To test whether H2O2 can act as a respiratory electron acceptor, inverted membrane vesicles were prepared from aerobic cultures (19). All reagents were made anaerobic, including inverted membrane vesicles, which were flushed with pure N2 for 5 min prior to being placed in the anaerobic chamber. Anaerobic reactions contained membrane vesicles (0.01 units of NADH oxidase activity), 100 µM NADH, and 2.5 mM H2O2 in 50 mM KPi, pH 7.4. Peroxidase activity was measured by following the oxidation of NADH by H2O2 at 340 nm in sealed cuvettes.
Glutathione Measurements-- Reduced glutathione (GSH) levels were inferred from measurements of total acid-soluble thiols (36), because >90% of this pool is GSH (37). Cells were grown in LB to A600 = 0.2-0.3 and, where indicated, exposed to 3 mM KCN for 5 min. Cultures were then centrifuged, washed in the same medium, and resuspended in 3% of the original volume of cold 50 mM KPi, pH 7.4, 1 mM EDTA. The sample was lysed by French press and clarified by centrifugation. An aliquot was removed for protein determination. Trichloroacetic acid (2.5%) was added to the remaining sample, and the sample was chilled on ice for 10 min. The sample was clarified by centrifugation and assayed. One-ml reactions consisted of 0.06 mg/ml 5,5'-dithiobis(2-nitrobenzoic acid) dissolved in 50 mM KPi, pH 7.4, and 100 µl of sample. Final absorbance was determined at 412 nm. A standard curve was generated by spiking the sample with known quantities of GSH.
Phage Inactivation--
The ability of an in vitro
system to generate DNA damage was assessed by the inactivation of
bacteriophage 
(13). Typically, phage were diluted to
105 plaque-forming units/ml into a 0.8% saline reaction
mixture at room temperature containing 1-10 µM ferric
chloride, 100 µM reductant, and 250 µM
H2O2. At intervals aliquots were diluted into
saline and infected into wild-type E. coli, which were
plated in top agar. After overnight growth, the number of plaques were
counted. The fraction of surviving phage was calculated relative to
control phage that were not exposed to the Fenton system.
In Vitro Measurements of the Reduction of Ferric Iron-- The rate at which various reductants converted ferric to ferrous iron in vitro was measured with a ferene dye-based assay. Reactions were performed at room temperature. They contained 10-20 µM ferric chloride, NAD(P)H, or glutathione, and, where indicated, 15 µM FAD and purified flavin reductase in a final volume of 1.5 ml of 20 mM Tris, pH 7.4. All buffers and reagents were anaerobic.
In reactions that measured iron reduction by crude extracts, reactions were conducted in air, and buffers were not added, because they promote the oxidation of Fe2+ by molecular oxygen. Approximately 3 mM KPi was carried over into the reaction system from the cell extracts. Because the concentration of ferric chloride was low, the pH of this system was about 6.0; it did not change significantly over the course of the reaction.
At time points, 60 µl of a freshly made mixture of 100 mM
ADP and 30 mM ferene were added to stop the iron-reduction
reactions and chelate the ferrous iron. Ferrous iron was quantified at
562 nm using an extinction coefficient of 26.6 mM1 cm1.
Generation of DNA Strand Breaks in Vitro--
Plasmid pACYC184
DNA was prepared by Qiagen column; the final elution was performed
using anaerobic 10 mM Tris, pH 8.0, and the sample was
frozen. DNA damage reactions were performed in an anaerobic chamber
with reagents that had been degassed or freshly dissolved in anaerobic
water. Reactions contained 200 µM NADH, 15 µM FAD, 10 µM FeCl3, Fre, 10 ng
of DNA, and 200 µM H2O2, added in
that order. In some cases 3 mM glutathione, 20 mM desferrioxamine, or 5000 units of catalase were
included. Reactions were terminated by the addition of 5000 units of
catalase. Samples were then electrophoresed on a 1% agarose gel,
stained for 2 h with ethidium bromide, destained for 4 h, and
scanned by a PhosphorImager to quantify the bands of supercoiled and
relaxed DNA.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Most Intracellular "Free Iron" Is in the Ferrous Form--
The
vulnerability of E. coli to Fenton chemistry suggests that
at least some of its free iron is in the ferrous form. Because ferric
iron has a sharp and clear EPR signal, whereas that of ferrous iron is
broad and indistinct (38), we used EPR spectroscopy to determine the
predominant redox state of the free iron. The EPR signal of untreated
cells was small, consistent with the spectrum of ferrous iron (Fig.
1). However, the signal could be enlarged and sharpened by incubating cells with desferrioxamine, a
cell-permeable iron chelator that binds free iron and stimulates
its autoxidation to the ferric form (Fig. 1). (Desferrioxamine chelates
iron that is loosely bound to biomolecules but does not remove iron
from metalloproteins (16).) A very similar transition occurred when desferrioxamine was omitted and cells were treated instead with high
concentrations of H2O2. The similarity of the
EPR signals implied that most of the iron that could be visualized in
the desferrioxamine-treated cells had been in the ferrous form,
accessible to oxidation by H2O2.
|
Cyanide Increases the Rate at Which the Cell Is Killed by
Fenton-driven DNA Damage--
A previous study (26) reported that
cyanide greatly accelerates the rate at which
H2O2 kills E. coli. This result
was replicated (Fig. 2A).
Cyanide was added to the cells immediately prior to H2O2 exposure; after exposure, the cells were
diluted and plated, and cell death was defined by the inability to form
colonies. The short-term treatment with cyanide alone did not kill any
cells. The phenomenon did not require protein synthesis, because the sensitization also occurred in cells treated with chloramphenicol. The
dead cells formed long filaments in the hours after
H2O2 exposure (data not shown). This behavior
suggested that the damaged cells retained the metabolic capacity to
grow but failed to divide because they could not replicate their DNA
(27).
|
Maximal killing occurred when 2.5 mM H2O2 was used, with less killing occurring at higher concentrations (Fig. 2B). This pattern has been ascribed to the ability of excess H2O2 to scavenge the ferryl radical before it dissociates into ferric iron and the hydroxyl radical (13). As such, these kinetics suggested that the lethal cell damage arose from iron-mediated Fenton chemistry. Indeed, cells that were pretreated with the cell-permeable iron chelators desferrioxamine and o-phenanthroline were fully protected from the killing effects of the combined cyanide-H2O2 exposure (Fig. 2A).
The magnitude by which cyanide accelerated the killing of wild-type cells by H2O2 could not easily be quantified because little killing occurred in the absence of cyanide. However, using xthA mutants that are deficient in DNA repair, it was possible to see that cyanide caused approximately a 5-fold increase in killing rate (Fig. 2C). The impact of cyanide upon the killing of wild-type cells was larger than this, presumably because the number of DNA lesions produced in the absence of cyanide were few and able to be repaired, but the addition of cyanide generated enough lesions to saturate the DNA-repair process and cause death.
Cyanide Specifically Increases the Rate at Which H2O2 Damages Intracellular DNA-- It was formally possible that cyanide accelerated cell death either by increasing the rate of DNA damage or by inhibiting the efficiency of DNA repair. However, cyanide increased the sensitivity of strains that were already genetically deficient in each of the pathways known to execute the repair of oxidized DNA. These included recA mutants, which lack recombinational repair capacity, as well as xthA and polA mutants, which are defective in excision and nick repair because of the absence of exonuclease III and DNA polymerase I, respectively (data not shown).
These results indicated that cyanide increased killing by accelerating
the rate of DNA damage, rather than by inhibiting repair. We sought
confirmation by using quantitative PCR to appraise the amount of DNA
damage formed by H2O2. This method detects any
lesions that block progression of the DNA polymerase. Data from
wild-type cells confirmed that H2O2 created far
more DNA damage when cyanide was present (Fig.
3A). To guard against the
possibility that cyanide inhibited rapid excision repair during the
period of DNA extraction, the experiment was repeated using
polA mutants. The same results were obtained (Fig.
3B). Note that cyanide itself did not damage DNA.
|
Cyanide Treatment Does Not Substantially Increase Free Iron Levels-- According to the standard scheme of oxidative DNA damage (Reactions 2-4), only H2O2, iron, and a reductant are required for the generation of DNA damage. Therefore, a treatment that accelerates this process must do so by providing one of these reactants. Cyanide does not accelerate damage by affecting the intracellular concentration of H2O2, because the intracellular and extracellular concentrations of H2O2 are essentially equivalent at the doses used in this experiment (39). Although cyanide inhibits catalase, catalase is not sufficiently active to affect the intracellular H2O2 concentration during the period of this exposure (39). Indeed, with this challenge regime, catalase-deficient strains exhibit the same sensitivities to H2O2 as do wild-type cells ((34) data not shown).
EPR experiments were then performed to test whether cyanide promotes damage by increasing the amount of free iron. When cells were treated with cyanide, the free-iron pool increased only slightly, about 2-fold (from 30 to 60 µM). This amount seemed too small to explain the large effect upon oxidative damage. Furthermore, when cells were quickly washed and suspended in fresh medium, respiration resumed, and sensitivity to H2O2 was immediately and completely relieved. In contrast, the levels of free iron remained at 60 µM even 15 min later (data not shown). Thus the sensitivity to H2O2 correlated with respiratory inhibition and was not connected to the modest changes in free iron.
Because the major effect of cyanide is not to change the amount of H2O2 or free iron, we concluded that cyanide must increase the availability of the electron donor that drives the Fenton reaction (Reaction 2). Our subsequent efforts were devoted to the identification of that reductant in these cells.
Respiratory Blocks Accelerate the Fenton Reaction by Forcing the Accumulation of NADH-- An earlier study (13) showed that cyanide sensitizes cells to H2O2 specifically because it blocks respiration. The key observation was that mutants that lacked either ubiquinone or respiratory NADH dehydrogenase activity, and therefore could not respire, were as sensitive to H2O2 as were cyanide-treated wild-type cells (26). The most obvious effects of these respiratory blocks are the accumulation of NADH and a dissipation of protonmotive force. Because NADH could plausibly serve as an electron donor, it was speculated that its accumulation facilitated DNA damage because NADH directly reduced free iron.
Fortunately, genetic experiments can be used to discern whether NADH
concentration or pmf is the critical effector of
H2O2 sensitivity. At the time of the earlier
study, it was not realized that E. coli contains two
respiratory NADH dehydrogenases; in fact, the Ndh mutant
that was used in that work had mutations that eliminated both enzymes.
NADH dehydrogenase I is a proton-translocating complex that is
structurally and functionally analogous to the mitochondrial complex I. NADH dehydrogenase II is a non-translocating enzyme whose physiological
role is to reoxidize excess NADH when the protonmotive force is high or
when the NdhI activity is inadequate to handle the NADH flux. During
growth on glucose both enzymes are present. In this medium the NdhII
enzyme is thought to have a major role in recycling NADH, because the
rate of NADH formation is very high.
We examined the sensitivity to H2O2 of single
mutants that lacked one or the other dehydrogenase. Unlike the double
mutant, both single mutants grow well in aerobic medium. With glucose as the carbon source, NADH and ATP are rapidly formed by glycolysis, and NdhII probably handles the majority of the NADH respiratory flux.
The nuo null mutants, which lacked the proton-translocating complex, exhibited wild-type resistance when cyanide was absent and,
like wild-type cells, were greatly sensitized when it was added (Fig.
4). In contrast, the ndh
mutants were hypersensitive even when cyanide was absent. The further
addition of cyanide to the ndh mutant increased sensitivity
slightly or not at all (Fig. 4), confirming that the effect of cyanide
upon DNA damage is solely mediated by its inhibition of
respiration.
|
Thus the impact of cyanide could be mimicked by the lack of the NADH dehydrogenase whose purpose is to reoxidize NADH but not by the lack of the pmf-generating dehydrogenase. We inferred from these results that the effect of the respiratory block on H2O2 sensitivity was not due to a loss of protonmotive force. Instead, it likely resulted from an accumulation of NADH. Direct measurements determined that cyanide treatment increased intracellular NADH pools 16-fold (Table II). When cells were additionally exposed to H2O2 (to mirror the cell-killing protocol), the NADH levels were diminished, possibly from damage to glucose catabolic enzymes; but they were again far higher with cyanide present. In contrast, the intracellular concentration of NADPH, which in growing cells is more abundant than NADH, increased less than 2-fold.
|
We were initially surprised to observe that cells did not become sensitive to H2O2 when they were challenged anaerobically, despite the presumptive lack of respiration. However, by using respiratory vesicles we determined that in the absence of oxygen H2O2 serves as an alternative substrate for both cytochrome oxidases o and d (see under "Materials and Methods"). Thus, during H2O2 challenge, anaerobic cells respire. Indeed, sensitivity was conferred either by mutations that eliminated cytochrome oxidases or by the addition of cyanide (data not shown). This result emphasizes that the sensitization by respiratory blocks does not require disruption of central metabolism, which proceeds normally in anaerobic cells. Our data indicated that the Fenton reaction is rapid whenever NADH levels are high. This result seemed to support the simple model (26) that NADH drives DNA damage by directly reducing ferric iron.
NADH Is Unlikely to be the Direct Reductant of Free Iron--
The
free iron that participates in Fenton reactions in
vivo is probably bound to a variety of biomolecules, including
DNA. For this reason workers have regarded small molecules, such as superoxide, NAD(P)H and glutathione, as the most probable electron donors, because these electron carriers are more likely than redox enzymes to be able to make electronic contact with iron atoms that are
affixed to the surfaces of proteins, membranes, and nucleic acids. As
mentioned, both experimental and quantitative arguments indicate that
superoxide is not a predominant iron reductant in vivo (40).
We therefore compared the abilities of NADH and glutathione to drive a
Fenton-dependent DNA-damaging system in vitro.
During incubation of 20 µM ferric iron with 400 µM NADH or 3 mM glutathione, levels similar
to those seen in non-respiring cells, the reduction by glutathione
(0.09 µM/min) proceeded much faster than that by NADH
(0.01 µM/min). The experiment was repeated in the
presence of H2O2 and bacteriophage , and the
loss of phage infectivity was measured. Again, glutathione drove the
Fenton reaction far more effectively than did NADH (data not shown).
This result implied that NADH must be a comparatively insignificant
direct reductant of free iron, even in cells in which the NADH pool is
enlarged by respiratory blocks. Therefore, electron flow from NADH to
iron must be mediated by an intermediate carrier.
Glutathione, Thioredoxin, Ferredoxin, and NADPH Are Not Responsible for Iron Reduction in Non-respiring Cells-- NADH is the ultimate source of electrons for all intracellular reductants, so increases in the level of NADH could elevate the levels of these other reductants, too. They, or electron carriers downstream of them, might directly reduce iron.
When the concentration of reduced glutathione was measured in untreated and cyanide-exposed cells, virtually no difference was detected (2.4 versus 2.6 mM). Furthermore, like wild-type cells, gshA mutants were greatly sensitized to H2O2 by the addition of cyanide, with >80% survival when challenged without cyanide and <1% survival when challenged with it. Thus glutathione is not involved in cyanide-accelerated oxidative DNA damage. Similar experiments showed that neither thioredoxin nor ferredoxin was required for sensitivity (data not shown).
NADPH is the conduit for electron flow from NADH to biosynthetic pathways and to electron-carrying proteins, such as thioredoxin. One can exploit mutations to sever the connections between the NADPH and NADH pools, so that respiratory blocks no longer affect the NADPH level. NADPH is primarily formed in E. coli in two ways: by oxidation of glucose 6-phosphate via the pentose-phosphate shunt, and by the action of the pmf-driven transhydrogenase (32). Because flux through the pentose-phosphate shunt is allosterically controlled by NAD+, and the transhydrogenase uses NADH as a substrate, we suspected that cyanide caused the moderate increase in NADPH titers by increasing the turnover of one or both of these processes. We therefore generated a mutant that lacked both transhydrogenase and glucose-6-phosphate dehydrogenase, the first enzyme in the pentose-phosphate pathway. As hoped, the NADPH levels of this strain were unusually low (~70 µM) and did not rise at all when cyanide was added, even though NADH pools rose 12-fold, from 35 to 430 µM. Yet cyanide still increased sensitivity to H2O2; 90% survived a 6-min H2O2 challenge without it but only 0.5% survived a challenge in its presence. Thus we concluded that neither NADPH nor any of the processes that derive electrons from it are responsible for reducing free iron in cyanide-treated cells.
Flavin Reductase Drives DNA Damage in Non-respiring Cells-- The only other small electron carrier of which we are aware is free flavin. Whereas FADH2 most commonly serves as a redox cofactor when it is tightly bound to enzymes, free FADH2 is used as a co-substrate for monooxygenase reactions (42-46), including several from E. coli strains (28, 47, 48). Indirect evidence has suggested that FADH2 may also assist in the activation of ribonucleotide reductase (33). E. coli contains an enzyme, called flavin reductase, that transfers electrons efficiently from NADH to free FAD (49).
We observed that mutants that lacked the structural gene for flavin
reductase, fre, exhibited normal
H2O2 sensitivity (Fig. 5A). However, in contrast to
all other strains, these mutants did not become very sensitive to
H2O2 when cyanide was added. When the
fre mutation was transduced into unrelated E. coli K-12 strains, they too became resistant to the synergistic
effects of cyanide and H2O2. Complementation
with fre on a plasmid restored wild-type behavior (Fig.
5A). The fre mutation also protected ndh mutants from H2O2 (Fig.
5B). Interestingly, strains that overproduced flavin
reductase anaerobically were even more sensitive to
H2O2 than were wild-type cells (Fig.
6). (In this experiment
catalase-deficient mutants were used in order to prevent O2
formation from the decomposition of H2O2.)
|
|
In all experiments the fre mutation abolished most, but not all, of the sensitization that cyanide causes. The small fre-independent sensitization may be caused by the 2-fold increase in free-iron concentration, which persisted in fre mutants (data not shown).
FADH2 That Is Produced by Flavin Reductase Efficiently Transfers Electrons to Ferric Iron-- The genetic and metabolic data suggested that respiratory blocks enlarge the NADH pools and thereby accelerate the rate at which flavin reductase transfers electrons to free flavins, which then reduce free iron. This scheme could only be true if normal NADH pools were insufficient to saturate flavin reductase activity. In fact, the apparent Km value of the enzyme for NADH is unusually high, 300 µM (50). This substantially exceeds the NADH concentration of respiring cells. Michaelis-Menten calculations suggest that an increase from 20 to 330 µM NADH (Table II) could potentially increase the flux through flavin reductase by 8.4-fold. Thus the kinetic properties of the enzyme conform with the impact of the physiological inhibitor.
Flavin reductase was purified to homogeneity. In the presence of NADH
and FAD, Fre rapidly catalyzed the reduction of ferric iron (Fig.
7A). The reaction rate was far
lower when iron was incubated with glutathione or with NADH in the
absence of the enzyme. When H2O2 and plasmid
DNA were added to the flavin reduction system, the DNA was
efficiently damaged (Fig. 7B).
|
The presumptive reaction series in this system is shown in Reactions
5-8 as follows:
|
|
|
|
|
|
|
|
The titer of Fre in wild-type cells is sufficiently high that cell
extracts, NADH, and FAD efficiently reduce ferric iron (Fig.8). The same reaction was much
slower in extracts from a fre mutant. Fre provides about
80% of the flavin reductase activity of wild-type extracts; the low
residual activity is presumably due to the redox enzymes that reduce
free flavins (51).
|
Because flavins autoxidize in aerobic solution, these reactions were all conducted anaerobically to confirm that superoxide need not mediate the electron transfer from flavin to iron. Superoxide does not mediate the effect of flavin reductase in vivo either, because it accelerated damage even in anaerobic cells (Fig. 6).
Thus flavin reductase constitutes the most efficient in vitro iron reduction system that we have yet found. Our results parallel those of Coves and Fontecave (52), who recovered flavin reductase when they purified an activity that efficiently reduced iron-siderophore chelates.
Flavin Reductase Promotes DNA Damage in Vivo Only When Respiratory
Blocks Increase the Pool of NADH--
We wanted to determine whether
flavin reductase was also responsible for the DNA damage that occurs
when H2O2 is added to cells that are respiring
normally. The rate of killing is not easily quantified in wild-type
cells, which efficiently repair the lesions. The fre
mutation was therefore transduced into repair-defective xth
and recA strains, in which killing by
H2O2 is substantial. The mutation did not
diminish the rate of cell death when H2O2 was
added in the absence of cyanide (Fig. 9).
The mutation did, however, diminish the acceleration of killing that
occurred when cyanide was added. Therefore, flavin reductase is the
preponderant source of electrons for free iron only when the pool of
NADH is large enough to satisfy its high Km. In
respiring cells flavin reductase is less active, and another, slower
pathway of iron reduction predominates. That other pathway has not been
identified.
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Free FADH2 Is an Efficient Reductant of Free
Iron--
These results indicate that the rate of oxidative DNA damage
depends upon how quickly free iron is reduced, and they show that free
FADH2 can be an efficient reductant. In non-respiring E. coli, Fenton chemistry occurs when electrons are
transferred from the enlarged NADH pools to FAD, from FADH2
to iron, and finally from iron to H2O2 (Fig.
10). This adventitious pathway
resembles enzymic electron-transport chains, which also employ iron to
break the oxygen-oxygen bond and flavins to bridge electron flow from the two-electron nicotinamide donor to the metal center acceptor.
|
Metal catalysts may be required for electron transfer to H2O2 because they weaken the oxygen-oxygen bond. Destabilization is presumably accomplished by interaction of an H2O2 oxygen atom with the metal d-orbitals; this is suggested by the fact that, in vivo and in vitro, the immediate product of the Fenton reaction is a ferryl radical, FeO2+, rather than a free hydroxyl radical (13, 53). An "organic Fenton reaction" reported recently (54) appeared to be exceptional, but further study has suggested that the reaction may proceed by addition of H2O2 to an unusual quinone reductant, rather than by direct electron transfer.
FADH2 efficiently reduces iron because reduced flavins are
facile univalent reductants. This trait is a consequence of the low
reduction potential of the flavosemiquinone/dihydroflavin pair (0.15
V) (55). Enzymes that transfer electrons from divalent donors to
internal metal centers, such as succinate dehydrogenase, typically use
flavins to do so; therefore, it is not surprising that
FADH2 also efficiently reduces free iron. In contrast, the NADH
/NADH potential is +0.93 V (56) and that of the
glutathione GS·/GSH couple is +0.90 V (57). Although it is
formally true that subsequent reactions of these radical products can
pull the initial reactions forward, these high reduction potentials
apparently establish a kinetic barrier that prohibits either NADH or
GSH from being as effective a univalent reductant as is
FADH2.
Adventitious Redox Chemistry May be Surprisingly Common--
Our
current understanding holds that a series of five adventitious
univalent electron transfers comprises the process of DNA damage;
molecular oxygen oxidizes reduced flavoproteins to generate superoxide
and H2O2, the superoxide then oxidizes
iron-sulfur clusters and releases free iron, free FADH2
reduces the free iron, the reduced iron transfers an electron to
H2O2, and the resultant hydroxyl radical
abstracts an electron from DNA. It may seem surprising that a
nonenzymatic process with so many steps should progress with enough
efficiency to have an impact upon the cell, particularly because the
intermediates can be drawn off by side reactions and scavenging
processes. However, the rate constants of the individual reactions are
very high, with the lowest (k = 76 M1 s1) being that of the Fenton
reaction. Univalent redox reactions often have low rearrangement
energies and therefore proceed more readily than other adventitious
reactions, requiring primarily that the donor and acceptor be brought
into electronic contact and that the reduction potentials favor the
reaction. Investigators have long taken advantage of this principle by
using small dyes as efficient artificial substrates for redox enzymes.
Oxidative stress is an unfortunate manifestation of the same idea;
oxygen species are good oxidants and are small enough that they cannot easily be excluded from the active sites of enzymes. Their ability to
intercept electrons is a problem that nature cannot easily solve.
What Is the Role of Flavin Reductase?-- The ability of Fre to reduce free flavins serves no obvious physiological purpose. Mutants that lack fre are sensitive to hydroxyurea, which deactivates ribonucleotide reductase; because reduced free flavins can reactivate the enzyme in vitro, it has been suggested that the production of diffusible FADH2 might be the real function of Fre (33). However, this enzyme efficiently reduces the free FAD pool only if NADH accumulates to unusually high levels, and those levels are achieved only when respiration is blocked. It is hard to understand why an enzyme would have such a high Km that it could not turn over rapidly under normal physiological conditions.
An explanation may be suggested by the fact that the fre gene exhibits homology to the flavin reductases that provide FADH2 to monooxygenases. In at least one case, kinetic data indicate that the partner reductase and monooxygenase physically interact (58). If so, FADH2 may be directly channeled between the enzymes, and the production of free FADH2 by small amounts of uncomplexed reductase may be adventitious. That scenario might explain the unfavorable kinetics of NADH binding by isolated Fre.
Furthermore, it is notable that the overproduction of Fre increased the rate of oxidative DNA damage above that of wild-type cells. That observation suggests that the normal metabolic contribution of Fre is not sufficient to fully sensitize cells, again as if it is the aberrant over-reduction of the free FAD pool that causes sensitivity.
The fre gene lies immediately downstream of ubiD, which encodes an enzyme in the ubiquinone biosynthetic pathway. Aerobic ubiquinone biosynthesis involves oxygenation steps whose enzymes have not been genetically defined (59), and we wondered whether fre might encode a reductase that provides FADH2 to one of them. However, other workers (60) recently demonstrated that ubiquinone synthesis is normal in fre strains. To guard against the possibility that a suppressor mutation had arisen in those fre mutants, we transduced the fre mutation anaerobically into fresh backgrounds and then re-checked the aerobic respiration rate. Our data support the conclusion that fre does not affect ubiquinone synthesis. For now its role remains uncertain.
What Is the Predominant Reductant of Free Iron Under Usual Biological Conditions?-- Reaction 3 dictates that the rate at which H2O2 damages DNA depends directly upon the concentration of free ferrous iron. This idea underpinned much early work on superoxide, which was thought to damage cells by serving as a reductant of free iron and thereby driving the Fenton reaction. It turned out that superoxide is too scarce to do so (23, 40). However, the results of this study confirm that when H2O2 concentrations are high, the rate of iron reduction does determine the pace of oxidative DNA damage. Furthermore, we have found that free FADH2 can be an important intracellular iron reductant, although its concentration may be high in E. coli only when respiration has been blocked.
We have recently determined that the synergistic killing of E. coli by NO and H2O2 (62) is primarily due to the inhibition of cytochrome oxidase by NO and the consequent flux through flavin reductase, leading to rapid DNA damage.2 Further experiments will be necessary to reveal whether this damage pathway is a significant killing mechanism inside phagocytes.
We remain interested in identifying the electron donor that drives DNA damage under normal conditions, when cells respire freely. Whereas our data indicate that flavin reductase itself is not involved (Fig. 9), it remains possible that FADH2 that is generated by other enzymes is the predominant reductant. Many redox enzymes can adventitiously reduce free flavins, albeit at a lower rate than does Fre. Sulfite reductase, for example, has a significant flavin reductase activity (2, 63). Alternatively, in the absence of much free FADH2, other small electron carriers like thiols may become the predominant donors. Further work will be necessary to resolve this issue.
Whatever the reductant(s), the cytosolic compartment of E. coli is a sufficiently reducing environment that free iron
accumulates in the ferrous form, despite the oxidizing force of
molecular oxygen and endogenous H2O2. For this
reason, free iron is poised to participate in Fenton reactions and
thereby contribute to mutation rates. Interestingly, this is not true
of Saccharomyces cerevisiae, whose pools of free iron are
predominantly oxidized (64). It is possible that in yeast this iron is
localized in a compartment that lacks univalent reductants, thereby
minimizing the impact of the iron upon oxidative DNA damage.
| |
ACKNOWLEDGEMENTS |
|---|
We thank Alex Smirnoff and Tatiana Smirnoff at the Illinois EPR center, a National Institutes of Health Biomedical Research Technology Resource supported by Grant P41-RR01811. Strains and plasmids were generously provided Luying Xun, Marc Fontecave, David Clark, Larry Vickery, and John Cronan, Jr.
| |
FOOTNOTES |
|---|
* This work was supported by National Institutes of Health Grant GM59030.The costs of publication of this article were defrayed <
