Bacteria Form Intracellular Free Radicals in Response to Paraquat and Streptonigrin DEMONSTRATION OF THE POTENCY OF HYDROXYL RADICAL*

The generation of oxygen reduction products by Neisseria gonorrhoeae FA1090 upon exposure to streptonigrin (SNG) and paraquat (PQ"') and their toxicity was examined. N. gonorrhoeae exhibited max- imal cyanide-insensitive respiration, which was employed as an indicator of superoxide (0;) formation, in the presence of 0.064 mM streptonigrin and 90 mM PQ", respectively. Using the concentrations of SNG and PQ"+ described above, complete lethality (>IOs cells/ml) was observed among cells exposed to SNG, whereas PQ"' reduced viability by only 3 logs. In an attempt to determine the oxygen radical species generated by gonococci when exposed to SNG, dimethyl sulfoxide, Fe3+, KCN, and the spin trap 5,5-dimethyl- 1-pyrroline-N-oxide (DMPO), we were able to detect 'OH manifested as the methyl adduct (DMPO-CH3). The production of the latter species was not inhibited by catalase, suggesting intracellular 'OH generation. When PQ"' was substituted for SNG, only low levels of DMPO-CH3 were observed, the production of which ceased within 8 min. SNG and PQ"', other studies employing an enzyme-mediated free radical-thine oxidase (0.45 milliunits), hypoxanthine (4.2 mM), and DMPO, generating system, concentrations of reagents were as follows: xan-MezSO, and Fe3+ at concentrations described above. contents were mixed and transferred to a flat quartz EPR cell (Wilmad Glass Co., Buena, NJ), fitted into the cavity of a Varian E-9 EPR spectrom- eter (Palo Alto, CA), and the spectrum obtained at 22 EPR instrumentation

In actively respiring cells, the univalent reduction of dioxygen allows formation of superoxide (C&) (1). 0; by itself appears to have significant potential for injury in biological * This work was supported by Chemistry of Life Processes Program of the National Science Foundation Award DCB-8616115, National Institute of Health Grants AM3322-02, AI07001-09, AI15036-07, and HL-33550, and by Grant BC 453 from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
IIRecipient of a Burroughs Wellcome Fellowship from the Infectious Diseases Society of America. systems (2,3). However, & in the presence H202 and copper (4) or iron (5) salts can react to form hydroxyl radical ( 'OH). The capacity of 'OH for nonspecific hydroxylation and/or hydrogen abstraction (6) renders this species particularly 4oxic. However, direct comparison of the effects of a and Paraquat (PQ")' can subvert electron flow from cytochrome pathways in Escherichia coli ( 7 ) , leading to formation of the paraquat-free radical (PQ.+) (8). This free radical, in the presence of dioxygen, produces 0; ( 7 ) . PQ" is toxic to a variety of mammalian cells (9, 10) and demonstrates antimicrobial action (11,12). Streptonigrin (SNG) is a related quinone antibiotic which has also been explored as an anti-tumor agent (13). The lethal action of SNG is enhanced by iron (14,15), implying that an iron-catalyzed Fenton reaction allows formation of 'OH (14, [16][17][18][19][20]. It has also been shown that PQ2+ (21) and other quinones (e.g. doxorubicin (22)) can catalyze 'OH formation under appropriate conditions of incubation.
In work evaluating the effects of SNG on Neisseria gonorrhoeae (23), we were struck by the potency of SNG relative to PQ". Explanations for this difference could include the enhanced capacity of SNG to permeate cells relative to PQ'+ (24) or the inherent ability of SNG to catalyze formation of 'OH in situ; 'OH could damage substrates not susceptible to 0; (25). The present study was undertaken to test this hypothesis and to characterize the differences between PQ'+ and SNG.
OH in biological systems have rarely been undertaken.

13404
This is an Open Access article under the CC BY license.
BWteria-N. gonorrhwae strain FA1090 was provided by Dr. P. Frederick Sparling (University of North Carolina at Chapel Hill). This organism was subcultured daily on GCB agar (GC medium base, Difco) containing 1 and 0.5% (v/v) Kellogg defined Supplements I and I1 (28), respectively. N. gonorrhoeae was chosen since it contains little or no superoxide dismutase (23, 291, which could confuse interpretation of results or limit quinone-mediated damage. All cultures were incubated at 37 "C in an atmosphere of 5% CO2. Broth cultures (GCB broth containing 2% Supplement I and 5 mM sodium bicarbonate) were inoculated from 24-h plates and grown to log phase on a platform shaker (135 strokes/min). Cell density was monitored by the increase in Klett units/h employing a Klett-Summerson colorimeter equipped with a 540-nm filter (Klett Manufacturing Co., New York). Viable cell numbers were determined by dilution plating on GC agar. Plates were inverted and colonies counted after 48 h of incubation at 37 "C in an atmosphere of 5% COZ.
Gonococcal Oxygen Consumption-Oxygen consumption was measured at 37 "C with a Clark oxygen electrode (Yellow Springs Instrument Co., Yellow Springs, OH) employing a 1-ml volume containing approximately l @ organisms/ml as previously described (30). Results were expressed as the percent cyanide-insensitive respiration of a control without cyanide. In other studies employing an enzyme-mediated free radicalthine oxidase (0.45 milliunits), hypoxanthine (4.2 mM), and DMPO, generating system, concentrations of reagents were as follows: xan-MezSO, and Fe3+ at concentrations described above. The contents were mixed and transferred to a flat quartz EPR cell (Wilmad Glass Co., Buena, NJ), fitted into the cavity of a Varian E-9 EPR spectrometer (Palo Alto, CA), and the spectrum obtained at 22 "C. EPR instrumentation settings as well as reaction supplements are given in legends.

RESULTS
Cyanide-insensitive Respiration of FA1090-Our first objective was to compare the microbicidal activities of PQz+ and SNG. In earlier work (23), we had identified the minimum inhibitory concentrations of SNG (49.4 nM) and PQ" (0.125 mM) for N. gorwrrhoeae strain FA1090. However, these differences in minimum inhibitory concentrations do not convey the magnitude of killing, nor do they relate microbial death directly to free radical formation. Oxygen consumption by gonococci is totally inhibited by cyanide (23, 31). However, bacteria incubated with quinone antibiotics develop cyanideinsensitive respiration (CIR) as a manifestation of the diversion of electrons toward formation of O; (32). Therefore, CIR represents an "upper limit" of G and/or other reduction products formed. Exposure of log phase gonococci to a range of PQ" and SNG concentrations resulted in detection of CIR ( Fig. 1). Maximal CIR was detected using 0.064 mM SNG Killing of FA1090 by PQz+ versus SNG and the Effect of Desferrioxamine-Gonococci were incubated in GC broth with concentrations of PQ" and SNG yielding optimal CIR. Significant differences in killing were apparent (Fig. 2). Approximately 3 logs of organisms were killed over a 30-min incubation period with PQ*+, while SNG was lethal to all organisms (10') after 20 min. Concentrations of PQ" greater than those which induced maximal CIR rates did not enhance killing (data not shown).
We and others (20,23) have noted the ability of desferrioxamine to protect bacteria from the antibiotic activities of SNG and an acetaldehydelxanthine oxidase/EDTA O;-generating system containing Fe3+ (33). However, the addition of desferrioxamine (200 p~) afforded no protection against PQ". This reaffirms that media iron (20)   maximal SNG-mediated killing of gonococci (23) but not for PQZ+-mediated killing.

Effect of SNG and P@?+ on Free Radical Formation-Spin
trapping is an integrative method used for the detection of free radicals in biological systems (34-40). The spin trap DMPO reacts with 0;l and 'OH to yield spin-trapped adducts with characteristic EPR spectra (34). Simultaneous generation of these radicals results in a spectrum which is a composite of the individual signals. DMPO reacts with 6 to form 2,2-dimethyl-5-hydroperoxy-l-pyrrolidinyloxyl (DMPO-OOH) which is unstable and decomposes rapidly into three species, including the hydroxyl radical adduct, 2,2-dimethyl-5-hydroxyl-1-pyrrolidinyloxyl (DMPO-OH) (34, 40). Confirmation that DMPO-OH resulted from the spin trapping of 'OH as opposed to the degradation of DMPO-OOH can be obtained by the addition of MezSO to the reaction mixture. The reaction of MezSO with 'OH yields methyl radical (*CH3) which can then be spin trapped to form 2,2,5-trimethyl-1-pyrrolidinyloxyl (DMPO-CH3). Formation of DMPO-CH, occurs at the expense of DMPO-OH (40). Experiments were conducted to directly measure free radicals formed by gonococci exposed to PQ" and SNG. In the absence of bacteria, no EPR spectrum was observed. When SNG was added to a bacterial suspension containing MezSO, KCN, and DMPO, DMPO-OH (Fig. 3A, peak 2 ) and a small nitroxide triplet were detected (peak T ) ; the latter signal has previously been recorded as a direct result of bacterial metabolism of DMPO.' Addition of superoxide dismutase completely inhibited DMPO-OH (scan not shown). These data suggest extracellular formation of 0; (41) detected as DMPO-OH.

The addition of Fe3+
M, no DETAPAC) to this system led to an EPR spectrum which was predominantly DMPO-CH3 (Fig. 3B, peak I) and DMPO-OH (Fig. 3B, peak 2); the concentration of the latter was so small that it disappeared during the scan, perhaps as a result of decomposition. The addition of superoxide dismutase to the SNG-Fe3+ system had no effect on the amplitude of DMPO-CH, but did diminish the low-field DMPO-OH signal (Fig. 3C). When catalase was added to the reaction mixture, a depression in all peak amplitudes was noted (Fig. 3D). These data imply that 'OH (spin trapped as DMPO-CH3) was generated both intra-and extracellularly. Alternatively, it is possible that DMPO-CH3 was produced before catalase could eliminate HzOz. To test this hypothesis, extracellular DMPO-CH3 was generated by the action of xanthine oxidase on hypoxanthine in the presence of Fe3+-DETAPAC, Me2S0, DMPO, and bacteria. When catalase was added to this reaction mixture prior to the addition of bacteria, DMPO-CH3 was eliminated (data not shown). The spin trapping experiments described above were repeated substituting PQz+ for SNG. Results were similar except that, in the presence of Fe3+, DMPO-CH3 was detected in smaller concentrations and the duration of production appear to be significantly less (Fig. 3E).
There are several possible explanations as to how supplemental Fe3+ allowed formation of 'OH by SNG. It has been suggested that SNG can bind iron (42) and other transition metals (43) directly, and that the proximity of the SNG.iron complex to critical substrates accounts for its toxicity. The addition of Fe3+ to a superoxide-generating system including MezSO results in the detection of a small concentration of DMPO-CH3 (peak I ) , DMPO-OH (peak 2), and DMPO-OOH (peak 3), the latter of which builds with time (Fig. 4A). SNG, added to a xanthine oxidase/hypoxanthine superoxide-generating system, led to formation of DMPO-CH3 which was increased at the expense of DMPO-OOH (Fig. 4B). When we substituted PQz+ for SNG, similar results were obtained (Fig.  4C). DMPO-CH3 was completely inhibited by catalase (Fig.   4, D and E ) , confirming formation of 'OH. In the presence of catalase, neither DMPO-OH nor DMPO-OOH are detected, most likely because 0; is reacting with excess Fe3+ (40). These results suggest that both SNG and PQ" complex with iron allowing formation of 'OH via a Fenton reaction.
The Effect of DETAPAC on Extracellular 'OH Formation in the Presence of P@+-When SNG or PQ2+ were used to stimulate free radical production by gonococci in the presence of DETAPAC, a doubling of signal amplitudes was observed, and DETAPAC, DMPO-CH, formation (Fig. 5A, peak 1) was markedly increased relative to that seen in the absence of DETAPAC (Fig. 3E). Superoxide dismutase enhanced DMPO-CH3 but did not diminish DMPO-OOH or DMPO-OH (Fig. 5B). These results demonstrate the formation of intracellular 4, and that both the location and detection of spin-trapped @ and 'OH are influenced by the presence of Fe3+/DETAPAC. DMPO-CH, was almost completely eliminated by catalase (Fig. 5C), suggesting that the 'OH generated in the presence of PQ2+/Fe3+/DETAPAC is primarily extracellular. Catalase also produced a large increase in DMPO-OH at the expense of DMPO-OOH, most likely secondary to a bioreductive phenomenon (40).
Experiments were undertaken to determine if extracellular 'OH enhanced the lethality of PQ". The addition of exogenous Fe3+ DETAPAC (100 PM) had no effect on PQ2+-mediated killing (Fig. 6). Because EDTA-Fe3+ is a superior catalyst in the Fenton reaction (6), the effects of PQ" were also tested in the presence of Fe3+-EDTA; no enhancement of killing was observed (Fig. 6).

DISCUSSION
Quinones, viologens, and related compounds are reduced to unstable radical intermediates as a consequence of their interaction with a variety of cells (47). In the presence of 02, these intermediates are reoxidized, resulting in the formation of 4, which eventually leads to cell injury (2, 3). G spontaneously dismutates to H202 (44) and may lead to formation of other oxygen reduction products as well (1). The specific oxygen reduction products and target site(s) ultimately responsible for the injury mediated by these compounds remains in doubt. In the present work the antibacterial mechanisms of SNG and PQ" were compared to provide additional insight into this matter.
Differences in the antimicrobial potency of quinone-like drugs could occur for a variety of reasons unrelated to the nature of oxygen reduction products formed (e.g. differences in membrane permeability, decreased K , of a quinone reductase, etc. (24)). Therefore, it was essential to use each compound examined at a concentration which induced similar magnitudes of 0; formation. Using E. coli, Hassan and coworkers (7,11,45) demonstrated that CIR provides an accurate measurement of bacterial 0; production resulting from quinone-like drugs. For N. gonorrhoeae, 90 mM PQ" and 64 p~ SNG elicited maximal CIR. Comparable CIR with E. coli has been observed with 2.0 mM PQ" or 0.1 mM SNG, although concentrations inducing maximal CIR have not been determined (32).
At concentrations yielding maximal CIR, SNG was far more microbicidal for N . gonorrhoeae than PQ". Killing observed with PQz+ was similar to earlier reports using E. coli (7, 41).
DF decreased the microbicidal activity of SNG (20) to the level seen with PQz+, but had no effect on PQ2+-mediated killing. Although intracellular iron pools are decreased following growth of some bacteria in the presence of DF (14), our work (23) suggests that the inhibitory effect of DF on SNGmediated killing involves its effect on extracellular iron. SNG has been shown to induce the iron-dependent formation of 'OH in a cell-free system using NADH as an electron donor (46, 47). Previous work suggests that SNG cannot only bind iron (42), but other transition metals (43) as well, and this ability may add greatly to its toxicity. Since Fe3+ bound to DF is unable to catalyze formation of 'OH (6), DF-mediated inhibition of the action of SNG likely reflected elimination of the contribution of 'OH.
To evaluate directly the free radical events surrounding the microbicidal activity of PQ" and SNG, spin trapping was used. In the presence of exogenous Fe3+, high levels of DMPO-CH3 (representing 'OH formation) were observed. When PQz+ was substituted for SNG, lower levels of 'OH (i.e. DMPO-CH,) were seen as would be predicted from our results with DF. It seemed possible that the differences in 'OH formed by these two agents was due to the ability of SNG to bind iron in a form capable of catalyzing an iron-driven Fenton reaction (14,20). However, addition of both PQ2+ and 6. Graf, E., Mahoney, J. R., Bryant, R. G., and Eaton, J. W. SNG resulted in similar 'OH production when added to an Fe3+-supplemented &-generating system. Previous work with PQ" suggested that an iron chelator was required for optimal formation of 'OH (21, 48). These studies showed that PQ" reduced by a xanthinelxanthine oxidase system reacted with HzOz to cause oxidation of deoxyribose. 'OH scavengers inhibited this reaction 20-3796 suggesting that some 'OH production can occur with PQ" in the presence of unchelated iron.
Reduced paraquat is believed to traverse the bacterial membrane and react with extracellular 02, leading to formation of extracellular 0; (49). In our studies, the persistance of DMPO-CH, in the presence of catalase suggests that SNG (and to a lesser extent PQ") stimulated formation of intracellular 'OH. The strain of N . gonorrhoeae used in this study contains 2611.2 catalase unitslmg protein (23). It seems possible that these levels of catalase helped to limit the concentration of 'OH formed; experiments to test this hypothesis are in progress. Control experiments eliminated the possibility that extracellular DMPO spin-trapped adducts were diffusing into the cell. These results represent the first report of successful spin trapping of intracellular 'OH. They contrast to those of DiGuiseppi and Fridovich (3) who observed only extracellular 'OH formed by Streptococcus sanguis exposed to the quinone plumbagin. The site at which free radicals are generated may also determine their toxicity (50). Experiments were conducted to correlate site of 'OH formation with microbicidal activity.
DETAPAC and EDTA increased formation of extracellular 'OH but did not enhance killing of gonococci by PQ". These results are consistent with earlier work in which formation of extracellular 'OH generated with plumbagin exerted little microbicidal activity (3). Experiments evaluating the microbicidal effects of extracellular 'OH and 0; have emphasized the toxicity of 'OH, but the systems employed are markedly different than those in this report (33). Because of its highly reactive nature, 'OH diffuses only short distances (10-20 A) (51) before reacting with a hydrogen donor. Therefore, generation in close proximity to critical bacterial target sites is probably more important for the toxicity of 'OH than the concentration formed. 02-derived free radicals have been implicated as important mediators of damage to a variety of biological systems (1, 2, 25, 42, 52, 53). The present work demonstrates directly the toxicity of intracellular 'OH relative to &. PQ2+ and SNG can be used to dissect these issues in other biological systems.