Role of iron and superoxide for generation of hydroxyl radical, oxidative DNA lesions, and mutagenesis in Escherichia coli

We measured the generation of hydroxyl radical (OH(.)) and oxidative DNA lesions in aerobically grown Escherichia coli cells lacking in both superoxide dismutases (SodA SodB) and repressor of iron uptake (Fur) using electroparamagnetic resonance and gas chromatography-mass spectrometry with a selected-ion monitoring method. A specific signal corresponding to OH(.) generation and an increase in oxidative DNA lesions such as 7,8-dihydro-8-oxoguanine and 1,2-dihydro-2-oxoadenine were detected in the strain deficient in sodA sodB fur. We showed that iron metabolism deregulation in fur mutant produced a 2.5-fold iron overload. The sodA sodB fur strain was about 100-fold higher mutability than the wild-type strain. The mutation spectrum in the strain was found to induce GC --> TA and AT --> CG transversions predominantly. The hypermutability of the strain was suppressed by the tonB mutation which reduces iron transport. Thus, excess iron and excess superoxide were responsible for OH(.) generation, oxidative DNA lesion formation, and hypermutability in E. coli.     

Ferric uptake regulator (Fur) mutants of Pseudomonas aeruginosa demonstrate defective siderophore-mediated iron uptake, altered aerobic growth, and decreased superoxide dismutase and catalase activities.

Pseudomonas aeruginosa is considered a strict aerobe that possesses several enzymes important in the disposal of toxic oxygen reduction products including iron- and manganese-cofactored superoxide dismutase and catalase. At present, the nature of the regulation of these enzymes in P. aeruginosa Is not understood. To address these issues, we used two mutants called A4 and C6 which express altered Fur (named for ferric uptake regulation) proteins and constitutively produce the siderophores pyochelin and pyoverdin. Both mutants required a significant lag phase prior to log-phase aerobic growth, but this lag was not as apparent when the organisms were grown under microaerobic conditions. The addition of iron salts to mutant A4 and, to a greater extent, C6 cultures allowed for an increased growth rate under both conditions relative to that of bacteria without added iron. Increased manganese superoxide dismutase (Mn-SOD) and decreased catalase activities were also apparent in the mutants, although the second catalase, KatB, was detected in cell extracts of each fur mutant. Iron deprivation by the addition of the iron chelator 2,2'-dipyridyl to wild-type bacteria produced an increase in Mn-SOD activity and a decrease in total catalase activity, similar to the fur mutant phenotype. Purified wild-type Fur bound more avidly than mutant Fur to a PCR product containing two palindromic 19-bp "iron box" regions controlling expression of an operon containing the sodA gene that encodes Mn-SOD. All mutants were defective in both ferripyochelin- and ferripyoverdin-mediated iron uptake. Two mutants of strain PAO1, defective in pyoverdin but not pyochelin biosynthesis, produced increased Mn-SOD activity. Sensitivity to both the redox-cycling agent paraquat and hydrogen peroxide was greater in each mutant than in the wild-type strain. In summary, the results indicate that mutations in the P. aeruginosa fur locus affect aerobic growth and SOD and catalase activities in P. aeruginosa. We postulate that reduced siderophore-mediated iron uptake, especially that by pyoverdin, may be one possible mechanism contributing to such effect.     

Hypochlorous acid stress in Escherichia coli: resistance, DNA damage, and comparison with hydrogen peroxide stress.

We have investigated the mechanisms of killing of Escherichia coli by HOCl by identifying protective functions. HOCl challenges were performed on cultures arrested in stationary phase and in exponential phase. Resistance to HOCl in both cases was largely mediated by genes involved in resistance to hydrogen peroxide (H2O2). In stationary phase, a mutation in rpoS, which controls the expression of starvation genes including those which protect against oxidative stress, renders the cells hypersensitive to killing by HOCl. RpoS-regulated genes responsible for this sensitivity were dps, which encodes a DNA-binding protein, and, to a lesser extent, katE and katG, encoding catalases; all three are involved in resistance to H2O2. In exponential phase, induction of the oxyR regulon, an adaptive response to H2O2, protected against HOCl exposure, and the oxyR2 constitutive mutant is more resistant than the wild-type strain. The genes involved in this oxyR-dependent resistance have not yet been identified, but they differ from those primarily involved in resistance to H2O2, including katG, ahp, and dps. Pretreatment with HOCl conferred resistance to H2O2 in an OxyR-independent manner, suggesting a specific adaptive response to HOCl. fur mutants, which have an intracellular iron overload, were more sensitive to HOCl, supporting the generation of hydroxyl radicals upon HOCl exposure via a Fenton-type reaction. Mutations in recombinational repair genes (recA or recB) increased sensitivity to HOCl, indicative of DNA strand breaks. Sensitivity was visible in the wild type only at concentrations above 0.6 mg/liter, but it was observed at much lower concentrations in dps recA mutants.     

Protective role of superoxide dismutases against ionizing radiation in yeast

The protective role of superoxide dismutases (SODs) against ionizing radiation, which generates reactive oxygen species (ROS) harmful to cellular function, was investigated in the wild-type and in mutant yeast strains lacking cytosolic CuZnSOD (sod1Delta), mitochondrial MnSOD (sod2Delta), or both SODs (sod1Deltasod2Delta). Upon exposure to ionizing radiation, there was a distinct difference between these strains in regard to viability and the level of protein carbonyl content, which is the indicative marker of oxidative damage to protein, intracellular H2O2 level, as well as lipid peroxidation. When the oxidation of 2',7'-dichlorofluorescin was used to examine the hydroperoxide production in yeast cells, the SOD mutants showed a higher degree of increase in fluorescence upon exposure to ionizing radiation as compared to wild-type cells. These results indicated that mutants deleted for SOD genes were more sensitive to ionizing radiation than isogenic wild-type cells. Induction and inactivation of other antioxidant enzymes, such as catalase, glucose 6-phosphate dehydrogenase, and glutathione reductase, were observed after their exposure to ionizing radiation both in wild-type and in mutant cells. However, wild-type cells maintained significantly higher activities of antioxidant enzymes than did mutant cells. These results suggest that both CuZnSOD and MnSOD may play a central role in protecting cells against ionizing radiation through the removal of ROS, as well as in the protection of antioxidant enzymes.     

Nitrogen status dependent oxidative stress tolerance conferred by overexpression of MnSOD and FeSOD proteins in Anabaena sp. strain PCC7120

The heterocystous nitrogen-fixing cyanobacterium, Anabaena sp. strain PCC7120 displayed two superoxide dismutase (SOD) activities, namely FeSOD and MnSOD. Prolonged exposure of Anabaena PCC7120 cells to methyl viologen mediated oxidative stress resulted in loss of both SOD activities and induced cell lysis. The two SOD proteins were individually overexpressed constitutively in Anabaena PCC7120, by genetic manipulation. Under nitrogen-fixing conditions, overexpression of MnSOD (sodA) enhanced oxidative stress tolerance, while FeSOD (sodB) overexpression was detrimental. Under nitrogen supplemented conditions, overexpression of either SOD protein, especially FeSOD, conferred significant tolerance against oxidative stress. The results demonstrate a nitrogen status-dependent protective role of individual superoxide dismutases in Anabaena PCC7120 during oxidative stress.     

Regulation of sod genes in Escherichia coli: relevance to superoxide dismutase function

This review is concerned with the effects of environmental perturbations on the expression of the two superoxide dismutase (SOD) genes in Escherichia coli (sodA, MnSOD; sodB, FeSOD). Early studies using SOD activity, showed that MnSOD levels respond to changes in oxygen tension, type of substrate, redox active compounds, iron concentration, the nature of the terminal oxidant, and the redox potential of the medium. FeSOD levels appeared nominally insensitive to these perturbations. More recent molecular genetic studies revealed that sodA expression is subject to regulation by three major regulatory systems: fur (ferric uptake regulation) and arcA arcB (aerobic respiratory control) mediate repression of sodA, while a relatively new system, soxR soxS (superoxide response), mediates activation of sodA expression. By contrast, sodB expression, which is much less studied at this time, appears to be positively activated in trans by fur. A rudimentary gene regulation model is presented which rationalizes past observations, is experimentally testable, and should serve as a guide to future research in this area.     

CO2 exacerbates oxygen toxicity

Reactive oxygen species (ROS) are harmful because they can oxidize biological macromolecules. We show here that atmospheric CO(2) (concentration range studied: 40-1,000 p.p.m.) increases death rates due to H(2)O(2) stress in Escherichia coli in a dose-specific manner. This effect is correlated with an increase in H(2)O(2)-induced mutagenesis and, as shown by 8-oxo-guanine determinations in cells, DNA base oxidation rates. Moreover, the survival of mutants that are sensitive to aerobic conditions (Hpx(-) dps and recA fur), presumably because of their inability to tolerate ROS, seems to depend on CO(2) concentration. Thus, CO(2) exacerbates ROS toxicity by increasing oxidative cellular lesions.     

Exogenous induction of the iron dicitrate transport system of Escherichia coli K-12.

Streptonigrin was used to select mutants impaired in the citrate-dependent iron transport system of Escherichia coli K-12. Mutants in fecA and fecB could not transport iron via citrate. fecA-lac and fecB-lac operon fusions were constructed with the aid of phage Mu dl(Ap lac). Strains deficient in ferric dicitrate transport which were mutated in fecB were as inducible as transport-active strains. They expressed the FecA outer membrane protein and beta-galactosidase of the fecB-lac operon fusions. In contrast, all fecA::lac mutants and fecA mutants induced with N-methyl-N'-nitro-N-nitrosoguanidine did not respond to ferric dicitrate supplied in the growth medium. tonB fecB mutants which were lacking all tonB-related functions were not inducible. We conclude that binding of iron in the presence of citrate to the outer membrane receptor protein is required for induction of the transport system. In addition, the tonB gene has to be active. However, iron and citrate must not be transported into the cytoplasm for the induction process. These data support our previous conclusion of an exogenous induction mechanism. Mutants in fur expressed the transport system nearly constitutively. In wild-type cells limiting the iron concentration in the medium enhanced the expression of the transport system. Thus, the citrate-dependent iron transport system shares regulatory devices with the other iron transport systems in E. coli and, in addition, requires ferric dicitrate for induction.     

Superoxide dismutase activity in Pseudomonas putida affects utilization of sugars and growth on root surfaces

To investigate the role of superoxide dismutases (SOD) in root colonization and oxidative stress, mutants of Pseudomonas putida lacking manganese-superoxide dismutase (MnSOD) (sodA), iron-superoxide dismutase (FeSOD) (sodB), or both were generated. The sodA sodB mutant did not grow on components washed from bean root surfaces or glucose in minimal medium. The sodB and sodA sodB mutants were more sensitive than wild type to oxidative stress generated within the cell by paraquat treatment. In single inoculation of SOD mutants on bean, only the sodA sodB double mutant was impaired in growth on root surfaces. In mixed inoculations with wild type, populations of the sodA mutant were equal to those of the wild type, but levels of the sodB mutant and, to a great extent, the sodA sodB mutant, were reduced. Confocal microscopy of young bean roots inoculated with green fluorescent protein-tagged cells showed that wild type and SOD single mutants colonized well predominantly at the root tip but that the sodA sodB double mutant grew poorly at the tip. Our results indicate that FeSOD in P. putida is more important than MnSOD in aerobic metabolism and oxidative stress. Inhibition of key metabolic enzymes by increased levels of superoxide anion may cause the impaired growth of SOD mutants in vitro and in planta.     

Viability of rep recA mutants depends on their capacity to cope with spontaneous oxidative damage and on the DnaK chaperone protein

Replication arrests due to the lack or the inhibition of replicative helicases are processed by recombination proteins. Consequently, cells deficient in the Rep helicase, in which replication pauses are frequent, require the RecBCD recombination complex for growth. rep recA mutants are viable and display no growth defect at 37 or 42 degrees C. The putative role of chaperone proteins in rep and rep recA mutants was investigated by testing the effects of dnaK mutations. dnaK756 and dnaK306 mutations, which allow growth of otherwise wild-type Escherichia coli cells at 40 degrees C, are lethal in rep recA mutants at this temperature. Furthermore, they affect the growth of rep mutants, and to a lesser extent, that of recA mutants. We conclude that both rep and recA mutants require DnaK for optimal growth, leading to low viability of the triple (rep recA dnaK) mutant. rep recA mutant cells form colonies at low efficiency when grown to exponential phase at 30 degrees C. Although the plating defect is not observed at a high temperature, it is not suppressed by overexpression of heat shock proteins at 30 degrees C. The plating defect of rep recA mutant cells is suppressed by the presence of catalase in the plates. The cryosensitivity of rep recA mutants therefore results from an increased sensitivity to oxidative damage upon propagation at low temperatures.     

Pseudomonas aeruginosa sodA and sodB mutants defective in manganese- and iron-cofactored superoxide dismutase activity demonstrate the importance of the iron-cofactored form in aerobic metabolism.

The consumption of molecular oxygen by Pseudomonas aeruginosa can lead to the production of reduced oxygen species, including superoxide, hydrogen peroxide, and the hydroxyl radical. As a first line of defense against potentially toxic levels of endogenous superoxide, P. aeruginosa possesses an iron- and manganese-cofactored superoxide dismutase (SOD) to limit the damage evoked by this radical. In this study, we have generated mutants which possess an interrupted sodA (encoding manganese SOD) or sodB (encoding iron SOD) gene and a sodA sodB double mutant. Mutagenesis of sodA did not significantly alter the aerobic growth rate in rich medium (Luria broth) or in glucose minimal medium in comparison with that of wild-type bacteria. In addition, total SOD activity in the sodA mutant was decreased only 15% relative to that of wild-type bacteria. In contrast, sodB mutants grew much more slowly than the sodA mutant or wild-type bacteria in both media, and sodB mutants possessed only 13% of the SOD activity of wild-type bacteria. There was also a progressive decrease in catalase activity in each of the mutants, with the sodA sodB double mutant possessing only 40% of the activity of wild-type bacteria. The sodA sodB double mutant grew very slowly in rich medium and required approximately 48 h to attain saturated growth in minimal medium. There was no difference in growth of either strain under anaerobic conditions. Accordingly, the sodB but not the sodA mutant demonstrated marked sensitivity to paraquat, a superoxide-generating agent. P. aeuroginosa synthesizes a blue, superoxide-generating antibiotic similar to paraquat in redox properties which is called pyocyanin, the synthesis of which is accompanied by increased iron SOD and catalase activities (D.J. Hassett, L. Charniga, K. A. Bean, D. E. Ohman, and M. S. Cohen, Infect. Immun. 60:328-336, 1992). Pyocyanin production was completely abolished in the sodB and sodA sodB mutants and was decreased approximately 57% in sodA mutants relative to that of the wild-type organism. Furthermore, the addition of sublethal concentrations of paraquat to wild-type bacteria caused a concentration-dependent decrease in pyocyanin production, suggesting that part of the pyocyanin biosynthetic cascade is inhibited by superoxide. These results suggest that iron SOD is more important than manganese SOD for aerobic growth, resistance to paraquat, and optimal pyocyanin biosynthesis in P. aeruginosa.     

Heterologous protective immunization elicited in mice by Pasteurella multocida fur ompH

Different strategies have been developed to produce vaccines against Pasteurella multocida. The approach described herein involves overexpression on the bacterial cell surface of Fur-regulated IROMPs (iron-regulated outer-membrane proteins). Accordingly, the ability of fur mutants to promote heterologous protection was examined in a Swiss mouse animal model. Twofur mutants derived from P. multocida were isolated, one of which was also defective in the OmpH protein. In mice challenged with virulent P. multocida, outer-membrane protein (OMP) extracts of fur cells conferred the same protection as obtained with wild-type cells grown in iron-depleted medium. Total protection was achieved with 40 microg of OMP extract from the fur ompH mutant. Mice administered heat-inactivated fur ompH cells were 60% cross-protected. The presence of a galE mutation in these cells did not further increase the protection level. Additionally, cell disruption by sonication provoked a higher level of protection than conferred by heat-treated cells. Taken together, the results showed that P. multocida fur ompH cells offer a simple and suitable approach for cross-protecting animals against infection with P. multocida.     

Cells Lacking Pfh1, a Fission Yeast Homolog of Mammalian Frataxin Protein,Display Constitutive Activation of the Iron Starvation Response

Friedreich ataxia is a genetic disease caused by deficiencies in frataxin. This protein has homologs not only in higher eukaryotes but also in bacteria, fungi, and plants. The function of this protein is still controversial. We have identified a frataxin homolog in fission yeast, and we have analyzed whether its depletion leads to any of the phenotypes observed in other organisms. Cells deleted in pfh1 are sensitive to growth under aerobic conditions, display increased levels of total iron, hallmarks of oxidative stress such as protein carbonylation, decreased aconitase activity, and lower levels of oxygen consumption compared with wild-type cells. This mitochondrial protein seems to be important for iron and/or reactive oxygen species homeostasis. We have analyzed the proteome of cells devoid of Pfh1, and we determined that gene products up- and down-regulated upon iron depletion in wild-type cells are constitutively misregulated in this mutant. Because of the particular signaling pathway components governing the iron starvation response in fission yeast, our experiments suggest that cells lacking Pfh1 display a decrease of cytosolic available iron that triggers activation of Grx4, the common regulator of the iron starvation gene expression program. Our Schizosaccharomyces pombe Δpfh1 strain constitutes a new and useful model system to study Friedreich ataxia.     

Fatty acid-mediated intracellular iron translocation: a synergistic mechanism of oxidative injury

Fatty acid has been reported to be associated with cardiovascular diseases and cancer, but the possible mechanism remains unclear. Here, we reported a novel mechanism for the permissive role of fatty acid on iron intracellular translocation and subsequent oxidative injury. In vitro study from endothelial cells showed that iron alone had little effect, whereas in combination with PA (palmitic acid), iron-mediated toxicity was markedly potentiated, as reflected in mitochondrial dysfunction, cell death, apoptosis, and DNA mutation. We also showed that PA not only facilitated iron translocation into cells through a transferrin-receptor (TfR)-independent mechanism, but also translocated iron into mitochondria; the subsequent intracellular iron overload resulted in reactive oxygen species (ROS) overgeneration and lipid oxidation. Further investigation revealed that PA-facilitated iron translocation is due to Fe/PA-mediated extracellular oxidative stress and the subsequent membrane damage with increased membrane permeability. Fe/PA-mediated toxic effects were reduced in rho0 cells lacking mitochondrial DNA or by antioxidant enzyme SOD, especially mitochondrially localized MnSOD, suggesting a permissive role of PA for iron deposition on the vascular wall and its subsequent toxicity via mitochondrial oxidative stress. This observation was confirmed in vivo in mice, wherein higher vascular iron deposition and accompanying superoxide release were observed in the presence of a high-fat diet with iron administration.     

Expression of the Maize Mnsod (Sod3) Gene in Mnsod-Deficient Yeast Rescues the Mutant Yeast under Oxidative Stress

Superoxide dismutases (SOD) are ubiquitous in aerobic organisms and are believed to play a significant role in protecting cells against the toxic, often lethal, effect of oxygen free radicals. However, direct evidence that SOD does in fact participate in such a protective role is scant. The MnSOD-deficient yeast strain (Sod2d) offered an opportunity to test the functional role of one of several SOD isozymes from the higher plant maize in hopes of establishing a functional bioassay for other SODs. Herein, we present evidence that MnSOD functions to protect cells from oxidative stress and that this function is conserved between species. The maize Sod3 gene was introduced into the yeast strain Sod2d where it was properly expressed and its product processed into the yeast mitochondrial matrix and assembled into the functional homotetramer. Most significantly, expression of the maize Sod3 transgene in yeast rendered the transformed yeast cells resistant to paraquat-induced oxidative stress by complementing the MnSOD deficiency. Furthermore, analyses with various deletion mutants of the maize SOD-3 transit peptide in the MnSOD-deficient yeast strain indicate that the initial portion (about 8 amino acids) of the maize transit peptide is required to direct the protein into the yeast mitochondrial matrix in vivo to function properly. These findings indicate that the functional role of maize MnSOD is conserved and dependent on its proper subcellular location in the mitochondria of a heterologous system.     

Induction of manganese-containing superoxide dismutase is required for acid tolerance in Vibrio vulnificus

The manganese-containing superoxide dismutase (MnSOD) of Vibrio vulnificus, normally detected after the onset of the stationary phase, is expressed during the lag that immediately follows the transfer of cells grown exponentially to a fresh medium acidified to pH 5.0, whereas Fe-containing SOD is constitutively expressed. The signal triggering the growth lag and MnSOD induction therein is not low pH but intracellular superoxide accumulated under these conditions, since addition of a superoxide scavenger not only shortened the lag but also abrogated the MnSOD induction. If the lysine decarboxylase reaction proceeds in the presence of sufficient lysine, the broth is rapidly neutralized to abolish the generation of oxidative stress. Accordingly, the acid tolerance response was examined without the addition of lysine. SoxR regulates MnSOD induction. Lack of MnSOD caused by mutations in soxR or sodA resulted in low tolerance to low pH. The fur mutant derepressing MnSOD showed better tolerance than the wild type. Thus, an increase in total cytosolic SOD activity through MnSOD induction is essential for the cell to withstand the acid challenge. The contribution of cuprozinc-containing SOD to acid tolerance is not significant compared with those of cytosolic SODs.     

Thermostability of manganese- and iron-superoxide dismutases from Escherichia coli is determined by the characteristic position of a glutamine residue.

The structurally homologous mononuclear iron and manganese superoxide dismutases (FeSOD and MnSOD, respectively) contain a highly conserved glutamine residue in the active site which projects toward the active-site metal centre and participates in an extensive hydrogen bonding network. The position of this residue is different for each SOD isoenzyme (Q69 in FeSOD and Q146 in MnSOD of Escherichia coli). Although site-directed mutant enzymes lacking this glutamine residue (FeSOD[Q69G] and MnSOD[Q146A]) demonstrated a higher degree of selectivity for their respective metal, they showed little or no activity compared with wild types. FeSOD double mutants (FeSOD[Q69G/A141Q]), which mimic the glutamine position in MnSOD, elicited 25% the activity of wild-type FeSOD while the activity of the corresponding MnSOD double mutant (MnSOD[G77Q/Q146A]) increased to 150% (relative to wild-type MnSOD). Both double mutants showed reduced selectivity toward their metal. Differences exhibited in the thermostability of SOD activity was most obvious in the mutants that contained two glutamine residues (FeSOD[A141Q] and MnSOD[G77Q]), where the MnSOD mutant was thermostable and the FeSOD mutant was thermolabile. Significantly, the MnSOD double mutant exhibited a thermal-inactivation profile similar to that of wild-type FeSOD while that of the FeSOD double mutant was similar to wild-type MnSOD. We conclude therefore that the position of this glutamine residue contributes to metal selectivity and is responsible for some of the different physicochemical properties of these SODs, and in particular their characteristic thermostability.     

Effect of R factors and other plasmids on ultraviolet susceptibility and host cell reactivation property of Escherichia coli

Some R factors, like some colicin factors, confer partial protection against the bactericidal effect of ultraviolet (UV) irradiation. Of 31 plasmids (17 R, 3 col, and 11 R-col factors) tested in Escherichia coli K-12, 15 protected, 11 had little or no effect, and 5 caused increased UV susceptibility. The effect of representative plasmids was qualitatively the same in K-12 of wild-type UV sensitivity, lambda-lysogenic or non-lysogenic, and in UV-sensitive mutants of classes uvrA, uvrB, uvrC, and recA (except that a sensitizing factor did not increase the sensitivity of two recA hosts). It is inferred that the UV-protecting effect of some plasmids does not result from their specifying enzymes similar to those deficient in such mutants. UV killing of multiply auxotrophic K-12, of wild-type sensitivity or recA or uvrC, was reduced by deprivation of required amino acids for 2 hr before irradiation, and further reduced if "starvation" was continued for 2 hr after irradiation. The plasmids tested in these conditions produced qualitatively the same effects as in nonstarved cells-except that in K-12 of wild-type UV sensitivity the effect of protecting plasmids was reversed (i.e. they caused decreased survival) when the cells were starved after irradiation. Two UV-protecting R factors reduced the ability of HCR(+) K-12 to support growth of irradiated phage T1.     

Isolation of superoxide dismutase mutants in Escherichia coli: is superoxide dismutase necessary for aerobic life?

Mu transposons carrying the chloramphenicol resistance marker have been inserted into the cloned Escherichia coli genes sodA and sodB coding for manganese superoxide dismutase (MnSOD) and iron superoxide dismutase (FeSOD) respectively, creating mutations and gene fusions. The mutated sodA or sodB genes were introduced into the bacterial chromosome by allelic exchange. The resulting mutants were shown to lack the corresponding SOD by activity measurements and immunoblot analysis. Aerobically, in rich medium, the absence of FeSOD or MnSOD had no major effect on growth or sensitivity to the superoxide generator, paraquat. In minimal medium aerobic growth was not affected, but the sensitivity to paraquat was increased, especially in the sodA mutant. A sodA sodB double mutant completely devoid of SOD was also obtained. It was able to grow aerobically in rich medium, its catalase level was unaffected and it was highly sensitive to paraquat and hydrogen peroxide; the double mutant was unable to grow aerobically on minimal glucose medium. Growth could be restored by removing oxygen, by providing an SOD-overproducing plasmid or by supplementing the medium with the 20 amino acids. It is concluded that the total absence of SOD in E. coli creates a conditional sensitivity to oxygen.