Hyperoxia and the Antimicrobial Susceptibility of Escherichia Coli and Pseudomonas Aeruginosa

Hyperoxia and the Antimicrobial Susceptibility of Escherichia coli and Pseudomonas aeruginosa

K. H. MUHVICH, M. K. PARK, R. A. M. MYERS, AND L. MARZELLA* Department of Pathology, University of Maryland School of Medicine, and Division of Hyperbaric Medicine, Maryland Institute for Emergency Medical Services Systems, 10 South Pine Street, Baltimore, Maryland 21201 Received 22 December 1988/Accepted 26 June 1989

We have tested the ability of hyperoxia (98% 02-2% CO2 at 2.8 atmospheres absolute [ca. 284.6 kPa]) to enhance killing of Escherichia coli (serotype 018 or ATCC 25922) by nitrofurantoin, sulfamethoxazole, trimethoprim, gentamicin, and tobramycin. We have also looked for interactions between hyperoxia and the aminoglycosides against Pseudomonas aeruginosa ATCC 27853. Hyperoxia significantly enhanced bacteriostatic activity of nitrofurantoin and trimethoprim as measured by MIC testing. The possibility exists that these effects might be due to the method required to test MICs under hyperoxic conditions rather than to the effect of hyperoxia itself. In addition, hyperoxia enhanced killing of bacteria by trimethoprim as measured by MBC testing. Hyperoxia decreased numbers of E. coli by 1.3 loglo and P. aeruginosa by 2.7 log1o in cation- supplemented Mueller-Hinton broth medium. The bacteriostatic effects of hyperoxia did not affect MICs of gentamicin or tobramycin. The lack of interaction between hyperoxia and gentamicin or tobramycin was confirmed by determining the number of viable bacteria remaining after 24 h of exposure to hyperoxia by using a pour plate method. We conclude that hyperoxia potentiates the antimicrobial activity of the reduction- oxidation-cycling antibiotic tested (nitrofurantoin) and of one of the antimetabolites tested (trimethoprim). Hyperoxia does not enhance the bactericidal effects of gentamicin and tobramycin, which require oxidative metabolism for transport into bacterial cells.

Hypoxic tissues are particularly susceptible to the estab- (14). Adriamycin, for example, generates fourfold more lishment of mixed aerobic-anaerobic bacterial infections. oxygen-derived free radicals at 100% 02 compared with at These types of infections cause localized abscesses or may 21% 02 (19). spread along fascial planes to cause necrosis of muscle and Finally, hyperoxia may be lethal to bacteria (17) and skin (7). Standard treatment consists ofincision and drainage parasites (5) that lack adequate antioxidant defenses by of infected tissues and intravenous administration of antimi- increasing the generation of 02- and other toxic oxygen crobial agents (6). The rationale for using hyperoxia in these species by cellular metabolic reactions. infections is to restore normal 02 tensions in ischemic tissue, In this work, we have tested representative compounds improve oxidative cellular metabolism (13), and enhance from three classes of antimicrobial agents for interactions phagocytic killing of bacteria (12, 18). with hyperoxia in bacterial killing. The following interac- It has also been suggested that hyperoxia may be benefi- tions may exist between hyperoxia and antimicrobial agents: cial by increasing the activity of specific antimicrobial hyperoxia alone has no effect but potentiates the mode of agents. The best evidence in favor of this proposal was action of the antimicrobial agent, and hyperoxia does not presented by Gottlieb and co-workers (8, 15). In a series of potentiate the mode of action of the antimicrobial agent but in vitro and in vivo experiments, they reported synergy has an effect which is either additive or synergistic with the between sulfonamides and hyperoxia against Vibrio anguil- antimicrobial agent. Synergy exists when oxygen and an larum (15) and Moraxella (Branhamella) catarrhalis (8). antimicrobial agent combine to produce an effect which is Evidence of decreased bacteriostatic and bactericidal activ- greater than the sum of their activities taken separately. ity of aminoglycosides in anaerobic environments has been Additive effects result when the actions of oxygen and an presented by several laboratories (24, 27-29); hyperoxia can antimicrobial agent are equal to the sum of their activities restore the bactericidal activity of tobramycin (1). However, taken separately. The use of a bacterial strain which is it is not known whether hyperoxia potentiates the bacteri- sensitive to hyperoxia maximized the chances of seeing an cidal activity of aminoglycosides in comparison with normal effect, whether it was additive or synergistic. Endpoint oxygen conditions. Hyperoxia may also increase bacteri- colony counting of bacteria by a pour plate method was used cidal activity of antimicrobial agents, such as nitrofurantoin in addition to MIC and MBC testing for the detection of and nifurtimox, by increasing the generation of superoxide interactions between oxygen and antimicrobial agents. anion (027). These antimicrobial agents undergo one-elec- (Portions of this study were presented at the Annual tron reduction to free radicals (nitrofurantoin + e-- Scientific Meeting of the Undersea and Hyperbaric Medical nitrofurantoin-). In the presence of oxygen, the antimicro- Society, New Orleans, La., 1988.) bial agent free radical undergoes oxidation to generate 02 (5) (nitrofurantoin- + 02 -- nitrofurantoin + 027). The sum MATERIALS AND METHODS of these reactions is termed reduction-oxidation cycling. Chemicals. Nitrofurantoin, trimethoprim, sulfamethox- The generation of superoxide anion from the drug free azole, gentamicin, tobramycin, nitrofurazone, menadione, radical follows first-order kinetics with respect to oxygen rifamycin SV, and NADPH were purchased from Sigma Chemical Co., St. Louis, Mo. Nitrofurantoin was solubilized * Corresponding author. in dimethylsulfoxide. Sulfamethoxazole and trimethoprim 1526 VOL. 33, 1989 02 INTERACTIONS WITH ANTIMICROBIAL AGENTS AND BACTERIA 1527 were diluted in 0.1 N NaOH and 0.05 N HCI, respectively. pour plates of serial 10-fold dilutions at 24 h of incubation. Stock solutions of gentamicin and tobramycin were made in The growth control well and the wells containing concentra- 0.1 M phosphate buffer (pH 7.8) and supplemented with 25 tions of antibiotics two and four times the MIC for each ,ug of Mg2+ and 50 ,ug of Ca2+ per ml (22). The stock organism were chosen for quantification of bacterial num- solutions of antibiotics were adjusted to pH 7.4, filter steril- bers. ized (pore size, 0.20 ,um), and diluted to desired concentra- Exposure of bacterial cultures to hyperoxia. Bacterial cul- tions. tures were treated in a small (27 by 50 cm) Plexiglas Bacterial strains. A clinical isolate of Escherichia coli hyperbaric chamber (model 1224; Bethlehem Corporation, (serotype 018) was used for MIC and MBC determinations, Bethlehem, Pa.). Cultures were exposed to 98% 02-2% CO2 endpoint quantitation of bacterial numbers, and polaro- at a pressure of 2.8 ATA for 24 h. The pressure at which the graphic measurement of 02 consumption. Two reference 02 was administered (2.8 ATA) is used clinically and does strains, E. coli ATCC 25922 and Pseudomonas aeruginosa not induce toxicity in human patients. Before compression, ATCC 27853, were used for MIC and MBC determinations the chamber was thoroughly flushed with 98% 02-2% CO2. and endpoint colony counting of bacteria. The temperature in the chamber was maintained at 35 to MIC and MBC determinations. MICs were determined by 37°C by circulating heated water through coils of copper broth macrodilution testing (22). MIC testing was performed tubing located in the bottom of the chamber. in 12-well (22-mm-diameter) culture plates (Costar, Cam- Polarographic measurement of antimicrobial agent reduc- bridge, Mass.). Serial twofold antimicrobial agent dilutions tion-oxidation cycling. The ability of nitrofurantoin, ni- were made in either minimal salts-acetate broth medium trofurazone, menadione, and rifamycin SV to generate su- (MSA) as described by Brunker and Brown (3) or Mueller- peroxide anion was measured by oxygen consumption Hinton broth supplemented with 50 ,ug of Ca2' and 25 ,ug of induced in rat liver microsomes and in E. coli serotype 018. Mg2+ per ml. Each well contained 0.5 ml of antimicrobial Oxygen consumption was measured with a YSI model 53 agent diluted in specific medium, 0.4 ml of bacterial biological oxygen monitor equipped with a Clark-type elec- inoculum in the same medium, and 0.1 ml of 1 M HEPES (N- trode, standard bath stirrer assembly (Yellow Springs Instru- 2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) buffer ment Co., Yellow Springs, Ohio), and strip chart recorder (pH 7.4). A growth control well (bacterial inoculum and (Beckman Instruments Co., Irvine, Calif.) by using a modi- HEPES buffer) and a sterility control well (growth medium fication of the polarographic assay described by Hassan (11). and HEPES buffer) were included with each MIC test. The A microsomal fraction was prepared from rat liver (2) to total volume per well was 1.0 ml, with a depth of 3 mm. This test for reduction of nitrofurantoin, nitrofurazone, menadi- depth was selected to maximize 02 penetration into the one, and rifamycin SV by cytochrome P-450 reductase. For medium as shown by Gudewicz et al. (10). The final bacterial microsomal assays, each antimicrobial agent (0.7 mM) and inoculum was approximately 5 x 105 CFU/ml and was NADPH (0.4 mM) were added to 50 mM potassium phos- confirmed for each assay by counting colonies in pour plates phate buffer (pH 7.0) to achieve a final volume of 3.0 ml. of 10-fold dilutions. The number of CFU in pour plates was Oxygen consumption began after the addition of 10 to 20 mg determined with a semiautomatic colony counter (Acculite of microsomes per ml. In bacterial assays, E. coli was grown model 133-8002; Fisher Scientific, Pittsburgh, Pa.). overnight in tryptic soy broth, centrifuged, and suspended in The cultures were incubated at 35 to 37°C either at 1 50 mM potassium phosphate buffer containing 2 mM MgCl2 atmosphere absolute (ATA) (101.3 kPa) in a gaseous envi- (pH 7.0) at 4°C. The final concentration of the bacterial cell ronment of 21% 02, 5% C02, and the balance N2, or in 98% suspension was 100 to 200 jig/ml. The 02 uptake reaction 02-2% CO2 at 2.8 ATA (ca. 284.6 kPa) for 24 h. The MIC was started by adding glucose (0.17%) to the reaction tests performed in normal oxygen pressure were read as mixture as a carbon source. NaCN (2.0 mM) was added after soon as the growth control showed suitable turbidity (ap- 1 to 2 min of 02 consumption to inhibit bacterial cellular proximately 24 h after the beginning of the experiment). The respiration. Cyanide was used to inhibit cytochrome oxi- MICs determined under hyperoxic conditions were read 24 dase, which transfers electrons to oxygen. Cyanide-insensi- to 30 h after removal from the hyperbaric chamber (approx- tive respiration continued and resulted in the formation of imately 48 h after the beginning of the experiment). The MIC hydroperoxides. After a steady state was established, the was the lowest antimicrobial agent concentration that pro- antimicrobial agent was added (0.7 mM) and 02 consumption duced a clear well (no growth) at the time that turbidity was was measured. The net rate of 02 consumption was deter- apparent in the growth control well. MICs of nitrofurantoin, mined by subtracting 02 consumption rates before addition trimethoprim, sulfamethoxazole, and tobramycin for E. coli of the antimicrobial agent from rates after addition of the serotype 018 were determined in MSA. MICs of gentamicin antimicrobial agent. and tobramycin for E. coli ATCC 25922 and P. aeruginosa Data analysis. Statistical evaluation of the data was done ATCC 27853 were determined in Mueller-Hinton broth. The by using Student's t test (paired comparison). MBC was the lowest antimicrobial agent concentration that decreased the original inoculum by >99.9%. To determine RESULTS MBCs, 10-fold dilutions of all antimicrobial agent-containing wells lacking turbidity were plated onto Mueller-Hinton agar Hyperoxia and a reduction-oxidation-cycling antimicrobial medium with 5% sheep blood added. After 24 h of incuba- agent (nitrofurantoin). We hypothesized that hyperoxia tion, MBC break points were determined by using rejection would increase antimicrobial activity through increased 02 values as described by Pearson et al. (23). MBCs were production by reduction-oxidation-cycling antibiotics such determined for the same bacteria and antimicrobial agent as nitrofurantoin. Reduction of these compounds can occur combinations as described above. through NADH- and NADPH-dependent reductases. Auto- Endpoint enumeration of bacteria. In order to compare oxidation of the reduced compound results in 02 produc- MICs of aminoglycosides for bacteria exposed to hyperoxia tion. In order to investigate this hypothesis, we first tested with MICs determined in a normal oxygen environment, the ability of oxygen to increase 02 consumption after bacteria in the culture wells were enumerated by making reduction of nitrofurantoin, nitrofurazone, menadione, and 1528 MUHVICH ET AL. ANTIMICROB. AGENTS CHEMOTHER.

TABLE 2. Effects of hyperoxia on MICs and MBCs of selected antimicrobial agents for E. coli serotype 018 0.17%Gluco Oxygen exposurea

'a Antimicrobial agent (n) 21% 02b 98% 02c o £ CN-I20 mM MIC MBC MIC MBC c (jug/ml) (jLg/ml) (,ug/ml) (jIg/ml) E 0 0.Q7mM NF \ Nitrofurantoin (10) 2.94 5.28 1.07d 3.43 0 Sulfamethoxazole (3) 2.08 608 1.62 608 ?U Trimethoprim (9) 0.58 0.86 0.31e 0.40" Tobramycin (6) 5.57 12.8 6.4 12.8 a Results are geometric means of MICs or MBCs calculated by taking the at nth root of the product of n determinations. Determinations were performed in duplicate. b 21% 0r5% C02-74% N2 at 1 ATA for 24 h. c 98% 02-2% CO2 at 2.8 ATA for 24 h. I I I d Significantly different (P = 0.01) from the value found with normal oxygen 0 2 4 6 conditions. Time (min) e Significantly different (P = 0.005) from values found with normal oxygen conditions. FIG. 1. Tracing of a polarographic recording of oxygen consumption by nitrofurantoin in the presence of E. coli serotype 018. After equilibration ofE. coli in phosphate buffer (pH 7.0) in ambient available. It has been shown that the generation of 02- by air at 37°C, the recorder was calibrated. Oxygen consumption began is potentiated by increasing the upon addition of glucose (0.17%). NaCN was then added to inhibit reduction-oxidation cycling 02 consumption. Addition of nitrofurantoin resulted in increased 02 oxygen concentration (19). Hyperoxia (98% 02) would thus consumption. Net 02 consumed was derived by subtracting the rate be expected to increase bacterial killing by nitrofurantoin of 02 consumption before the addition of nitrofurantoin from the through increased production of 027. This is supported by rate Of 02 consumption in the presence of nitrofurantoin. the data shown in Table 2. The MIC of nitrofurantoin in hyperoxic conditions was significantly decreased (P = 0.01) to one-third of the MIC in normal oxygen conditions. The rifamycin SV by E. coli serotype 018. Oxygen consumption MBC of nitrofurantoin for E. coli serotype 018 was also was measured with a Clark-type electrode. Figure 1 is a decreased, but this decrease did not achieve statistical tracing from a typical polarographic experiment. Cellular significance. respiration of E. coli was induced by the addition of glucose In summary, under normal oxygen conditions, nitrofuran- (0.17%). In the presence of NaCN, nitrofurantoin caused an toin undergoes reduction-oxidation-cycling reactions. Hy- increase in the rate Of 02 consumption. This is indirect peroxia increases bacteriostatic and bactericidal effects of evidence for the production of 027- Cyanide only slowed the nitrofurantoin, probably through increased production of rate of cellular respiration, because it does not inhibit 02- hydroperoxide respiration of the bacterial cells. 02 con- Hyperoxia and antimetabolites (sulfamethoxazole and tri- sumption by nitrofurantoin and nitrofurazone with endoge- methoprim). It has also been hypothesized that hyperoxia nous E. coli reductases was somewhat reduced compared may increase antibacterial effects of sulfamethoxazole and with values obtained with microsomal NADPH reductase trimethoprim by maintaining folate synthetic enzymes in an (Table 1). Menadione generated similar quantities of 02 oxidized state (8). Sulfamethoxazole and trimethoprim act with both endogenous E. coli reductases and microsomal synergistically to inhibit bacterial growth by blocking se- NADPH reductases. Rifamycin SV failed to consume 02 quential steps in the production of folic acid. Hyperoxia with either E. coli or microsomal NADPH reductase. 02 significantly reduced the MICs and MBCs of trimethoprim consumption with concomitant generation of 02- was for E. coli serotype 018, as shown in Table 2. The MIC of shown to be linearly correlated to the concentration of trimethoprim forE. coli under hyperoxic conditions was 0.31 nitrofurantoin in the range of 0.2 to 3.3 mM (data not ,ug/ml, and the MIC determined in normal oxygen conditions shown). These data demonstrate that nitrofurantoin is capa- was 0.58 p.g/ml. MBCs of trimethoprim were significantly ble of reduction-oxidation cycling when reducing equiva- decreased (P = 0.005) from 0.86 ,ug/ml in normal oxygen lents, reducing enzymes, and molecular oxygen (21% 02) are conditions to 0.40 pug/ml in hyperoxia. MICs were determined in MSA medium to maximize the possibility offinding interactions between hyperoxia and antimicrobial agents. TABLE 1. Oxygen consumption by reduction-oxidation-cycling MSA medium with acetate added as a carbon source is compounds in the presence of microsomal NADPH reductase or sufficient for growth of bacteria but lacks nutrients such as E. coli serotype 018 thymidine which would otherwise allow bacteria to bypass 02 consumption (nmol of 02/min blocks in folate metabolism. per ,imol of compound)a by: Hyperoxia and antimicrobial agents with oxygen-dependent Compound re- E. coli uptake (gentamicin and tobramycin). Aminoglycosides Microsomes serotype 018 quire oxidative metabolism for transport into bacterial cells. Normal oxygen conditions restore the bactericidal activity of Nitrofurantoin 103 34 aminoglycosides, which is diminished in anoxic environ- Nitrofurazone 72 28 ments. We wished to test the ability of hyperoxia to increase Menadione 81 72 bacterial killing over that seen in normal oxygen conditions. Rifamycin SV 0 0 MIC and MBC testing oftobramycin forE. coli serotype 018 0 Results are means of three determinations tested in triplicate. performed in MSA medium showed no significant differ- VOL. 33, 1989 02 INTERACTIONS WITH ANTIMICROBIAL AGENTS AND BACTERIA 1529

TABLE 3. Effects of hyperoxia on colony counts of E. coli serotype 018 in the presence of gentamicin or tobramycin CFU' of E. coli serotype 018 with: 02 Gentamicin (>ig/ml) Tobramycin (,ug/ml) 0 2 4 0 2 4 21%b 8.9 ± 0.3 6.3 ± 0.4 4.9 + 0.4 8.9 ± 0.3 6.3 ± 0.3 4.6 ± 0.6 98%c 7.8 0.5d 6.1 ± 0.3 3.4 ± 0.4 7.8 0.5d 5.3 ± 0.3 3.0 ± 0.3 a Data are means of log CFU ± standard error. n = 12 determinations; determinations were performed in duplicate. b 21% 02-5% C02-74% N2 at 1 ATA for 24 h. c 98% 02-2% C02 at 2.8 ATA for 24 h. d Significantly different (P = 0.0001) from the value found in normal oxygen conditions. ences between normal oxygen conditions and hyperoxia itantly reduced to 02- (20). This oxygen-derived free radical (Table 2). No significant differences were seen in MICs or may undergo further reduction to form H202, which is toxic MBCs of gentamicin or tobramycin in MHB for either E. coli for bacteria. Hydrogen peroxide may take part in the Fenton 25922 or P. aeruginosa 27853 (data not shown). Endpoint reaction, where it is converted to the highly reactive hy- colony counting did not reveal increased bactericidal effects droxyl radical (16). These toxic oxygen species may inacti- of gentamicin or tobramycin against E. coli serotype 018 vate bacterial enzymes, promote lipid peroxidation of bac- under hyperoxic conditions, as shown in Table 3. In addi- terial cell membranes, and alter bacterial DNA. Bacteria tion, no enhanced bactericidal effects of gentamicin or lacking antioxidant enzymes are particularly susceptible to tobramycin were demonstrated against E. coli ATCC 25922 the effects of oxygen-derived free radicals. Moreover, nitro- or P. aeruginosa ATCC 27853 under hyperoxic conditions furantoin or (100% 02 at 4.2 ATA [425.5 kPa]) can (data not shown). hyperoxia Bacteriostatic effects of hyperoxia as shown by endpoint each inhibit the growth of E. coli by inducing stringency (25, colony counting. During MIC testing, marked bacteriostatic 26). The stringency response is defined as intracellular effects were seen in growth control wells of all bacteria accumulation of guanosine 5'-diphosphate 3'-diphosphate, exposed to hyperoxia. Typically, turbidity was seen in the which is a regulatory inhibitor of biosynthetic processes in culture wells at 24 to 30 h after the multiwell plates were bacteria (25). Nitrofurantoin blocks synthesis of branched- removed from the hyperoxic environment. To rule out the chain amino acids such as valine and leucine (26). The loss of effects of hyperoxia on aminoglycosides caused by observed decrease in the MIC of nitrofurantoin for a clinical the delay in appearance of turbidity in growth controls, bacterial isolate (E. coli serotype 018) during hyperoxic endpoint colony counts were performed. A 1-log1o difference exposure may be due to a combination of these mechanisms. (P = 0.0001) was seen between growth controls exposed to However, there remains a small possibility that this differ- normal oxygen conditions and growth controls exposed to ence could be due to the method required to test MICs under hyperoxia for E. coli serotype 018 (Table 3). A 1.5-log1o hyperoxic conditions and not to the effect of hyperoxia itself. difference (P = 0.0001) was seen between growth controls of We next examined the effects of hyperoxia and the anti- E. coli (ATCC 25922) exposed to normal oxygen conditions metabolites sulfamethoxazole and trimethoprim against E. and hyperoxia. A 2.7-log10 difference (P = 0.006) was seen coli serotype 018. Sulfonamides such as sulfamethoxazole between growth controls of the reference P. aeruginosa block the conversion of para-aminobenzoic acid into dihy- strain (ATCC 27853) exposed to normal oxygen conditions drofolic acid by competitive inhibition. It is thought that and hyperoxia (data not shown). These findings are impor- sulfonamides may have a higher affinity for bacterial tetrahy- tant, since bacteriostatic effects of hyperoxia have been dropteroic acid synthetase than the naturally occurring sub- shown to be strain specific (9). When looking for additive or strate para-aminobenzoic acid. Trimethoprim blocks the synergistic effects of hyperoxia and antimicrobial agents, next step in bacterial folic acid synthesis, the conversion of one needs to use a bacterium for which hyperoxia is at least dihydrofolate to tetrahydrofolate, by inhibition of the en- bacteriostatic. Endpoint enumeration of bacteria revealed zyme dihydrofolate reductase. The sequential blockage in that all three bacterial strains were sensitive to the bacteri- the folate biosynthetic pathway results in synergy between ostatic effects of hyperoxia. these antimicrobial agents against a number of pathogenic bacteria. Gottlieb (8) suggested that oxidation of metabolic DISCUSSION intermediates by hyperoxia might alter the kinetics of the There are several possible mechanisms by which hyper- folate synthetic pathway. In the studies with V. anguillarum oxia may influence the actions of antimicrobial agents. We (15), hyperoxia increased bacterial killing when combined explored the possibility that hyperoxia could interact with with sulfisoxazole and trimethoprim. In the present study, reduction-oxidation-cycling compounds such as nitrofuran- we saw a significant decrease in MICs and MBCs of trimeth- toin to increase bacterial killing. Nitrofurantoin [N-(5-nitro- oprim for E. coli serotype 018. These findings are in 2-furfurylidene)-1-amino hydantoin] is a nitro compound agreement with our previous work (21), which showed that which is toxic for a broad spectrum of bacteria known to hyperbaric oxygen combined with sulfamethoxazole- cause acute urinary tract infections. The exact mode of trimethoprim decreases bacterial numbers in the peritoneal action of nitrofurantoin is not known. However, like other cavity of rats with E. coli serotype 018-induced sepsis. nitro compounds, nitrofurantoin produces 02 in the pres- Similar interactions between hyperoxia and sulfamethox- ence of molecular oxygen. The first step in the reduction- azole were not seen. Vibrio spp. are known to be more oxidation-cycling reaction of nitrofurantoin is a one-electron highly susceptible to effects of hyperoxia than are other reduction to the nitro anion (NO2-). When molecular oxy- gram-negative bacteria, such as the ones that we have gen is present, N02- is oxidized to NO2 and 02 iS concom- examined. 1530 MUHVICH ET AL. ANTIMICROB. AGENTS CHEMOTHER.

Uptake of aminoglycoside antibiotics such as gentamicin ganisms. Annu. Rev. Microbiol. 25:111-152. and tobramycin is an energy-dependent process. Electron 10. Gudewicz, T. M., J. T. Mader, and C. P. Davis. 1987. Combined transport involving quinone oxidation-reduction cycles is an effects of hyperbaric oxygen and antifungal agents on the important source of energy for aminoglycoside entry into growth of Candida albicans. Aviat. Space Environ. Med. 58: bacteria (4). This may account for the resistance of anaero- 673-688. 11. Hassan, H. M. 1984. Exacerbation of superoxide radical forma- bic and facultative anaerobic bacteria to aminoglycosides in tion by paraquat. Methods Enzymol. 105:523-532. anaerobic environments. Our results indicate that no en- 12. Hohn, D. C., R. D. MacKay, B. Halliday, and T. K. Hunt. 1976. hanced activity of aminoglycosides is achieved under hyper- Effect of 02 tension on microbicidal function of leukocytes in oxic conditions in vitro. These results do not exclude the wounds and in vitro. Surg. Forum 27:18-20. possibility that hyperoxia could restore oxygen tensions in 13. Hunt, T. K., and M. P. Pai. 1972. The effect of varying ambient tissues to levels required for aminoglycoside uptake by oxygen tensions on wound metabolism and collagen synthesis. bacteria. A similar restoration of 02 tensions allows poly- Surg. Gynecol. Obstet. 135:561-567. morphonuclear leukocytes to kill bacteria in previously 14. Jones, D. P. 1985. The role of oxygen concentration in oxidative hypoxic tissues (12, 18). Finally, we have used stress: hypoxic and hyperoxic models, p. 151-195. In H. Sies endpoint (ed.), Oxidative stress. Academic Press, Inc. (London), Ltd., enumeration of viable bacteria to look for interactions be- London. tween hyperoxia and aminoglycosides. This was necessary 15. Keck, P. E., S. F. Gottlieb, and J. Conley. 1980. Interaction of because the cultures exposed to hyperoxia had not grown to increased pressures of oxygen and sulfonamides on the in vitro visual turbidity in growth control wells at the time they were and in vivo growth of pathogenic bacteria. Undersea Biomed. removed from the hyperbaric chamber. Endpoint colony Res. 7:95-106. counts confirmed the bacteriostatic effects of hyperoxia in a 16. Korbashi, P., R. Kohen, J. Katzhendler, and M. Chevion. 1986. quantitative manner but did not show any interactions with Iron mediates paraquat toxicity in Escherichia coli. J. Biol. aminoglycosides. Enidpoint colony counting may be more Chem. 261:12472-12476. suitable than and MBC testing for 17. Krieg, N. R., and P. S. Hoffman. 1986. Microaerophily and MIC examining the oxygen toxicity. Annu. Rev. Microbiol. 40:107-130. interactions between hyperoxia and antimicrobial agents, 18. Mader, J. T., G. L. Brown, J. C. Guckian, C. H. Wells, and because bacteria are quantified at the same standard time J. A. Reimarz. 1980. A mechanism for the amelioration by point that MICs in air are normally read (24 h). hyperbaric oxygen of experimental staphylococcal osteomyeli- tis in rabbits. J. Infect. Dis. 142:915-921. ACKNOWLEDGMENTS 19. Mimnaugh, E. G., M. A. Trush, and T. E. Gram. 1981, Stimu- lation by adriamycin of rat heart and liver microsomal NADPH- This study was supported by the Maryland Institute for Emer- dependent lipid peroxidation. Biochem. Pharmacol. 30:2797- gency Medical Services Systems Research Fund and by Public 2804. Health Service grant 5T35 HL07612-02 from the National Institutes 20. Moreno, S. N. J., R. P. Mason, and R. Docampo. 1984. The of Health. reduction of nifurtimox and nitrofurantoin to free radical metab- We gratefully acknowledge the assistance of Paul P. Rodier and olites by rat liver mitochondria. Evidence for an outer mem- Bret Holliday in the maintenance and operation of the hyperbaric brane-located nitroreductase. J. Biol. Chem. 259:6298-6305. chamber. 21. Muhvich, K. H., R. A. M. Myers, and L. Marzella. 1988. Effect of hyperbaric oxygenation, combined with antimicrobials and LITERATURE CITED surgery, in a rat model of intra-abdominal infection. J. Infect. 1. Adams; K. R., T. E. Sutton, and J. T. Mader. 1987. In vitro Dis. 157:1058-1061. potentiation of tobramycin under hyperoxic conditions. Under- 22. National Committee for Clinical Laboratory Standards. 1985. sea Biomed. Res. Suppl. 14:37. Methods for dilution antimicrobial susceptibility tests for bac- 2. Bernacchi, A., R. Myers, B. F. Trump, and L. Marzeha. 1984. teria that grow aerobically. National Committee for Clinical Protection of hepatocytes with hyperoxia against carbon tetra- Laboratory Standards, Villanova, Pa. chloride-induced' injury.. Toxicol. Pathol. 12:315-323. 23. Pearson, R. D., R. T. Steigbigel, H. T. Davis, and S. W. Chap- 3. Brunker, R. L., andO. R. Brown. 1971. Effects of hyperoxia on man. 1980. Method for reliable determination of minimal lethal oxidized and reduced NAD and NADP concentrations in Esch- antibiotic concentrations. Antimicrob. Agents Chemother. 18: erichia coli. Microbios 4:193-203. 699-708. 4. Bryan, L. E., and S. Kwan. 1983. Roles of ribosomal binding, 24. Reynolds, A. V., J. M. T. Hamilton-Miller, and W. Brumfitt. membrane potential, and electron transport in bacterial uptake 1976. Diminished effect of gentamicin under anaerobic or hy- of streptomycin and gentamicin. Antimicrob. Agents Chemo- percapnic conditions. Lancet i:447-449. ther. 23:835-845. 25. Seither, R. L., and 0. R. Brown. 1982. Induction of stringency 5. Docampo, R., and S. N. J. Moreno. 1986. Free radical metabo- by. hyperoxia in Escherichia coli. Cell. Mol. Biol. 28:285-291. lism of antiparasitic agents. Fed. Proc. 45:2471-2476. 26. Selther, R. L., and 0. R. Brown. 1984. Paraquat and nitrofuran- 6. Finegold, S. M., and H. M. Wexier. 1988. Therapeutic implica- toin inhibit growth of Escherichia coli by inducing stringency. J. tions of bacteriologic findings in mixed aerobic-anaerobic infec- Toxicol. Environ. Health 14:763-771. tions. Antimicrob. Agents Chemother. 32:611-616. 27. Tack, K. J., and L. D. Sabbath. 1985. Increased minimal 7. Gorbach, S. L. 1982.,Interactions between aerobic and anaero- inhibitory concentrations with anaerobiosis for tobramycin, bic bacteria. Scand. J. Infect. Dis. Suppl. 31:61-67. gentamicin, and amikacin compared to latamoxef, chloram- 8. Gottlieb, S. F. 1970. Interaction of oxygen, temperature, and phenicol, and clindamycin. Chemotherapy 31:204-210. drugs on two species of Neisseria as concerns the mechanisms 28. Vandaux, P. 1981. Peripheral inactivation of gentamicin. J. of oxygen toxicity, p. 288-296. In J. Wada and T. Iwa (ed.), Antimicrob. Chemother. Suppl. 8:17-25. Proceedings of the Fourth International Congress on Hyper- 29. Verklin, K. J., and G. L. Mandeli. 1977. Alteration of the baric Medicine. Igaku Shoin Ltd., Tokyo. effectiveness of antibiotics by anaerobiosis. J. Clin. Lab. Med. 9. Gottlieb, S. F. 1971. Effect of hyperbaric oxygen on microor- 89:65-71.