Generation of Free Radicals from Metronidazole and Other Nitroimidazoles by

Metronidazole, ronidazole, secnidazole, beddazole, and misonidazole are reduced by intact I>itriehonuwurs fkefus cells to nitro anion radicals that can be detected by electron spin resonance spectroscopy. This activity appears to be related to the cellular content of reducing substrates, since nitm anion radical formation is stim-ulated in the presence of glucose and pyruvate. The nitro anion radicals could not be detected under aerobic conditione. Anaerobic homogenates of T. fahrs also reduce metronidazole to the nitro anion radical when pyruvate, NADH, or NADPH is added as the ultimate source of reducing equivalents. Free radical formation may be the basic cause of nitroimidazole toxicity in trichomonads.

mediates are possible, including the nitro anion radical, and the nitroso and hydroxylamine derivatives: The nitro anion radical and the hydroxylamine derivative were suggested as the most likely candidates for the toxic intermediates ( 7,8).
Electron spin resonance studies of rat liver microsomal reduction of a wide variety of nitro derivatives including nitrobenzene (9). nitrofurantoin (lo), nifurtimox (ll), benznid d e (X?), ronidazole (13), and metronidazole (13) have unambiguously demonstrated the presence of a nitro anion radical metabolite. The protozoan parasite Trypanosoma cruzi can also reduce nifurtimox to a nitro anion radical (14). However, the pathway by which metronidazole is metabolized in trichomonads has been postulated to be different from that observed in mammalian systems (4). In this paper, we provide ESR spectroscopy evidence that the reduction of metronidazole and other nitroimidazoles by trichomonads also occurs via nitro anbn radical metabolites.

MATERIALS AND METHODS
Tritrichomonas foetus was obtained through the courtesy of F. Costa e Silva and W. De Souza of the Institute of Biophysics, Federal University of Rio de Janeiro, Brazil, and was maintained in Diamond's trypticase/yeast extract/maltose medium (15) without agar and antibiotics, pH 7.0, with 10% fetal calf s e m , at 37 "C. The cultures were subcultivated at intervals of 48 h. For experiments, cells were harvested after different periods of cultivation, in most cases &er 24 to 30 h. The cells were collected by centrifugation at 1500 X g for 15 min and washed once in 0.1 Y potassium phosphate buffer, pH 7.5. The number of cells was determined with a Coulter Counter.
Glass powder (5 g/g of cells, wet weight) was added to the washed pellet, and the mixture was ground in a mortar for 3 min at 4 "C. This procedure resulted in complete breakage of the cells, as revealed by phase contrast micrascopy. Most of the glass powder was separated by decantation. The disrupted cells were then suspended in 0.1 M potassium phosphate buffer, pH 7.5, and homogenized by several passages through a No. 24 gauge hypodermic needle attached to a syringe.
Gluwxe, sodium pyruvate, NADH, NADPH, glucose &phmphate, glucose-6-phmphate dehydrogenase (type XXIII), metronidazole, and glass beads were obtained from Sigma. Benznidazole (N-benzyl-2nitro-I-imidazole acetamide) and mhsonidazoie (1-(2-nitro-l-imidazolyi)-3-methoxy-2-propanol) were obtained from Hoffman-La Roche ESR measurements were made at room temperature, 24 "C, with a Varian E-9 spectrometer equipped with a TMllo cavity as previously described (10)(11)(12). For the experiments with T. foetus homogenates, the reaction mixture (3 ml final volume) contained the drug, at the concentration stated under "Results," and 10 m~ sodium pyruvate or an NADPH-or an NADH-generating system of NADPH or NADH (1 m~) , glucose &phosphate (5 m~) , and glucose-&phosphate dehydrogenase (1 unit/d). For the experiments with intact cells, the reaction mixture (3 ml final volume) contained the drug, at the concentration stated under "Results," with or without 10 n w glucose or 10 m~ pyruvate. A 0.1 M potassium phosphate buffer, pH 7.5, was used throughout. The incubations were gapsed with nitrogen for 5 min prior to initiating with either pyruvate, NADH, NADPH, or glucose. The protein concentration was determined as previously described (11). Fresh homogenates or cells (less than 9 h old) that were stored on ice were used in all experiments.
Oxygen uptake was measured in the Gilson polarograph using a Clark electrode. Assays of oxygen consumption were made at 37 "C in a medium containing 0.1 M potassium phosphate buffer, pH 7.5, and additions as stated under "Results."

RESULTS
Incubation of metronidazole with intact T. foetus in the presence of glucose generates a radical with a multi-line ESR spectrum characteristic of the nitro anion radical (Fig. L4). Analysis of the nuclear hyperfine parameters of the radical agreed well with those determined for both the pulse radiolysis-generated radical in its unprotonated form (16) and the rat liver microsome-generated radical anion (13).
No ESR signal could be detected using heat-denatured T.
foetus (70 "C, 30 min) or when the drug was omitted (Fig. 1B). The spectrum of the anion radical could not be observed under aerobic conditions; however, the signal did appear once the dissolved oxygen in the incubation had been consumed. Even after bubbling with nitrogen for 5 min, there was still a lag before the signal appeared. This delay was attributed to the presence of residual oxygen and indicates that strict anaerobiosis is necessary for the buildup of the radical. Following the lag period, a steady state radical concentration was achieved.
The effects of glucose and pyruvate on the metronidazole steady state ESR signal are shown in Table I. The signal was observed in the absence of added exogenous substrates, indicating that endogenous substrates could be used by cells as the ultimate electron donors for metronidazole reduction; pyruvate was the most effective exogenous substrate.

TABLE I Effect of different substrates on the steady state concentration of metronidazole anion radical in incubations of Tritrichomonas
The same ESR cell remained in the cavity throughout the experiment to minimize any artifact due to differences in cell position. To maximize the signal-to-noise ratio, the instrument settings were: microwave power 20 mW, 5 G modulation amplitude. The values are the average f S.D. of three incubations. Other conditions were as in icals were analyzed by computer simulation (Fig. 3). The nuclear hyperfine couplings of these radicals agreed well with those determined previously (12, 13, 16). The hyperfine coupling constants of secnidazole were indistinguishable from those of metronidazole, as might be expected. Only the larger hypefine coupling constants are required to simulate these ESR spectra (Fig. 3) because of the poor resolution caused by the high modulation amplitude needed to detect the radical concentrations in these incubations.
Homogenates of T. foetus were also able to reduce metronidazole to a nitro anion radical (Table 11)

Effect of different substrates on the steady state concentration of metronidazole anion radical in incubations of homogenates
The same ESR cell remained in the cavity throughout the experiment to minimize any artifact due to differences in cell position. To maximize the signal-to-noise ratio, the instrument settings were: microwave power 20 mW ,  effective substrate for metronidazole reduction by T. foetus homogenate as determined by steady state radical concentrations (Table 11). The signal was also observed in the presence of NADH or NADPH as electron donor. In the absence of the homogenates, the NADH-or the NADPH-generating system or pyruvate did not produce observable concentrations of the metronidazole anion radical. The spectrum of the anion radical could not be observed under aerobic conditions or after heating the homogenate in a steam bath for 10 min. Metronidazole (0.5-6 mM) did not stimulate oxygen consumption by incubations of T. foetus homogenates (in the presence of NADH, NADPH, or pyruvate) or cells (either alone or in the presence of pyruvate or glucose).

DISCUSSION
For the first time, an intact living organism has been used to obtain evidence of nitroimidazole anion radical formation. Intact T. foetus cells are able to reduce metronidazole and other nitroimidazoles to their respective anion radicals. This activity appears to be related to the cellular content of reducing substrates, since nitro anion radical formation was stimulated in the presence of glucose or pyruvate, the two main sources of T. foetus reducing equivalents (22,23).
Anaerobic homogenates of T. foetus also reduce metronidazole to the nitro anion radical when pyruvate is added as an electron donor. Trichomonad homogenates contain a pyruvate-ferredoxin oxidoreductase (22) which reduces nitroimidazoles via systems containing ferredoxin-or flavodoxin-type electron transport proteins (24). These proteins, one of which has been isolated recently, are reduced enzymatically by the substrates and, in turn, reduce the nitroimidazoles' (24). With homogenates, glucose was not a good electron donor for metronidazole reduction. This is probably due to the dilution of enzymes and cofactors necessary for the generation of pyruvate. Nitro anion radical formation could not be detected under aerobic conditions. This could be attributed either to the reaction between oxygen and the nitro anion radicals (Reaction l), ArNO2-+ 0 2 "-f ArNOz + 6,- (1) as has been postulated for rat liver microsome incubations of metronidazole (13), or to a competition for the electrons between the nitroimidazoles and 0 2 . The competitive nature of this inhibition is suggested by the observation that metronidazole did not stimulate oxygen consumption in T. foetus intact cells or homogenates even in the presence of pyruvate.
The metronidazole anion radical signal observed in the homogenate in the presence of an NADPH-or an NADHgenerating system demonstrates that pyruvate synthase is not the sole system in trichomonads capable of reducing metronidazole (25).
The biological consequences of nitroimidazole anion radical formation in trichomonads are unknown, but the known chemistry of the anion free radicals (10) suggests that the reactions of the nitroimidazole radical metabolites may be of toxicological significance. At the low oxygen tension where the facultative anaerobic trichomonads live, the anion radicals would establish a higher steady state concentration, which would allow either their direct interaction with DNA and/or proteins or, more probably, their subsequent reduction to another active intermediate like the hydroxylamine, which could then react with DNA and/or proteins. This interaction of the reactive metabolite or metabolites with DNA and/or proteins is the most widely held explanation for their antimicrobial activity (6, 7, 25). Our results give evidence for the formation of a reactive intermediate of nitroimidazole reduction by trichomonads.