Induction of the Manganese-containing Superoxide Dismutase in Escherichia coli Is Independent of the Oxidative Stress (oxyR-Controlled) Regulon*

The synthesis of manganese-superoxide dismutase in response to hydrogen peroxide and to paraquat was examined in strains of Escherichia coli with different mutations in the oxyR gene. Hydrogen peroxide treatment did not induce manganese-superoxide dismutase, but did induce the oxyR regulon. Paraquat induced this enzyme in a strain compromised in its ability to induce the defense response against oxidative stress (oxyR deletion) as well as in a strain that is constitutive and overexpresses the oxyR regulon. Catalase (HPI), but not manganese-superoxide dismutase, was overexpressed under anaerobic conditions in a strain har- boring a constitutive oxyR mutation. The data clearly demonstrate that the induction of manganese-super-oxide dismutase is independent of the oxyR-controlled regulon. mediated partially reduced superoxide radical (02-), hydrogen peroxide (H2O2), and hy-droxyl radical during normal biological reduction of


Induction of the Manganese-containing Superoxide Dismutase in
The synthesis of manganese-superoxide dismutase in response to hydrogen peroxide and to paraquat was examined in strains of Escherichia coli with different mutations in the oxyR gene. Hydrogen peroxide treatment did not induce manganese-superoxide dismutase, but did induce the oxyR regulon. Paraquat induced this enzyme in a strain compromised in its ability to induce the defense response against oxidative stress (oxyR deletion) as well as in a strain that is constitutive and overexpresses the oxyR regulon. Catalase (HPI), but not manganese-superoxide dismutase, was overexpressed under anaerobic conditions in a strain harboring a constitutive oxyR mutation. The data clearly demonstrate that the induction of manganese-superoxide dismutase is independent of the oxyR-controlled regulon.
Superoxide dismutases (EC 1.15.1.1) are metalloenzymes that are widely distributed among living organisms. They provide an essential defense against oxygen toxicity, which is mediated by the partially reduced oxygen intermediates ( i e . superoxide radical (02-), hydrogen peroxide (H2O2), and hydroxyl radical (OH. )) generated during normal biological reduction of dioxygen (1,2). The levels of these reactive oxygen intermediates can be exacerbated by the presence of many environmental factors such as higher pOz (3), ionizing radiation (4), redox-active compounds ( 5 ) , ozone (6), and other oxidants (7). In Escherichia coli, three isozymic forms of superoxide dismutase are found manganese, iron, and hybrid isozymes (8). Currently, the iron-superoxide dismutase is thought to be constitutive with respect to oxygen, whereas manganese-superoxide dismutase is under rigorous control (8). The regulation of manganese-superoxide dismutase has been extensively studied in Enterobacteriaceae (8-17). The enzyme is inducible by oxygen (a), redox-active compounds that generate 0; in presence of oxygen (5, 9), ferrous iron chelators under both aerobic and anaerobic conditions (10, 14, 15), nitrate during anaerobic growth (ll), and several strong oxidants such as ferricyanide, ammonium persulfate, and copper-cyanide complex (12). A model has been proposed for the regulation of manganese-superoxide dismutase in which the sodA gene, the structural gene for the enzyme, is thought to be negatively regulated by an iron-containing redox-sensitive repressor protein (10, 11). Recent findings * This work was supported in part by Grant DMB-8609239 from the National Science Foundation. This is paper 11673 of the Journal Series of the North Carolina Agricultural Service (Raleigh, NC). 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. (11,12,17) support the proposed model and suggest the possibility of more layers of control elements (17). ' The presence of a global regulatory mechanism for coordinate expression of enzymes and proteins needed for cellular protection against hydrogen peroxide and oxidative stress is well documented (18)(19)(20). The induction of the oxidative response regulon is under positive control by the oxyR gene product, whose expression or activity is H202-inducible (18). It has been suggested that manganese-superoxide dismutase is part of the oxyR-controlled regulon (18)(19)(20). The data to support this conclusion came from the fact that its activity is about 2-fold higher in an oxyRl mutant strain of Salmonella typhimurium relative to a wild-type parental strain (18). oxyRl is a dominant mutation that confers resistance to hydrogen peroxide and causes constitutive overexpression of nine HzO2-inducible proteins; manganese-superoxide dismutase is believed to be one of these constitutive proteins (18)(19)(20).
The aim of this study was to explore the role of oxyR in the regulation of manganese-superoxide dismutase in E. coli. We examined the induction of this enzyme in wild-type strains (oxyR+) and compared it with that of mutants affected in the expression of the oxyR locus. The data clearly indicate that the expression of the manganese-superoxide gene (sodA) is not regulated by oxyR.

MATERIALS AND METHODS
in Table I.
Bacterial Strains-The E. coli strains used in this study are listed Media and Growth Conditions-A minimal salts medium (21) supplemented with 0.5% glucose, 0.05% yeast extract, and 10 Mg/ml each arginine and methionine was used and is designated as GY medium. The rich medium, TSY, contained 3% Trypticase soy broth plus 0.5% yeast extract. Addition of the amino acids and yeast extract to the glucose minimal medium was required to support the growth of the oxyR deletion mutant and was used to grow all four strains. Overnight cultures grown in GY medium at 37 "C on a rotary shaker at 200 rpm were used to inoculate fresh GY and TSY media. Growth was estimated in terms of turbidity measured at 600 nm (ODw). The experimental cultures were allowed to grow for about two generations from an initial ODsw of 0.03 or 0.05 before addition of hydrogen peroxide or paraquat from sterile stock solutions. Anaerobic growth conditions were maintained by growing the slant cultures, the overnight cultures, and the experimental cultures in the designated media equilibrated At the end of the specified growth period, chloramphenicol (150 rg/ in an anaerobic environment maintained in a Coy anaerobic chamber. ml) was added, and the cultures were allowed to stand for 15 min before removing from the anaerobic chamber (10). Assays-Cell-free extracts were prepared by sonication followed by dialysis (22). Protein was estimated by method of Lowry et al. (23) using bovine serum albumin as a standard. Superoxide dismutase (24) and catalase (25) were assayed as described previously. Superoxide dismutase isozymes, i.e. manganese, iron, and hybrid, were separated ' H. M. Hassan and D. Touati, unpublished data. 14808  (26), visualized by the nitro blue tetrazolium activity stain (27), and quantitated by linear scanning densitometry (8). Catalase activity was localized in 10% polyacrylamide gels and stained as described previously (28). Western blotting was done according to the method described by Towbin et al. (29). Antisera prepared against pure manganese-superoxide dismutase was a kind gift of Dr. D. Clare (Department of Animal Science, North Carolina State University). The peroxidase conjugate of goat anti-rabbit immunoglobulin was purchased from Cappel/Cooper-Biomedical. Sensitivity of each strain to hydrogen peroxide and paraquat was tested using disc inhibition assays. Cells were grown in nutrient broth to midlog phase. Aliquots were then diluted 10-fold with soft agar and plated onto LB medium. BBL filter discs were placed in four quadrants of each plate, and 10 pl of hydrogen peroxide or paraquat stock solutions were applied to the discs, giving final concentrations of 60 and 300 pg of Hz02 or 38 and 190 pg of paraquat. The diameter of the zone of inhibition was measured after 24 h at 37°C. Chemicals-Paraquat, bovine serum albumin, xanthine, xanthine oxidase, and horse heart cytochrome c were purchased from Sigma. Hydrogen peroxide (30%), Tris-HC1, EDTA, and other chemicals were obtained from Fisher.

Effects of Hydrogen Peroxide on Catalase and Superoxide
Dismutase Biosynthesis-Exposure of E. coli and S. typhimurium to 60 PM hydrogen peroxide for 60 min has been shown to induce the oxidative stress regulon via the activation of oxyR (18,19). Fig. 1 compares the catalase and superoxide dismutase activities of the four test strains growing in GY medium 1 h after initial exposure to hydrogen peroxide (60 PM). Superoxide dismutase biosynthesis in each of the four strains was unaffected by H202 treatment, whereas catalase levels, in agreement with findings by Christman et al. (18), increased 3.5-fold in oxyR+ strains RK4936 and K12, but not in the oxyA3 deletion strain TA4112. In TA4110, which contains the dominant oxyR2 allele, catalase levels were 15fold higher than those seen in the parental strain, K12. Furthermore, in this mutant strain, catalase was uninducible by H202, as previously reported (18). Fig. 2 shows the gels developed for catalase after aliquots of the extracts from the H2Oz-treated cultures (oxyR+ and oxyA3) were electrophoresed. The induction pattern of the two bands shows that HPI, the faster mobility band, was the catalase induced by hydrogen peroxide in strains with a functional oxyR locus. The superoxide dismutase isozymes in these extracts were also separated by electrophoresis, and the gels were stained for superoxide dismutase activity. Densitometric scans of these gels revealed no significant differences in manganesecontaining superoxide dismutase in the four test strains following exposure to hydrogen peroxide (Table 11). Similar results were seen after 2 h of exposure (data not shown). The activity of the iron-and hybrid-superoxide dismutase was not affected by hydrogen peroxide treatment (Table II;

FIG. 2.
Nondenaturing polyacrylamide gel of catalase from samples exposed (lunes b) and not exposed (lanes u) to hydrogen peroxide. Aliquots of extracts containing 50 pg of protein from the HzOz-treated deletion strain (TA4112) and its parent (RK4936) (Fig. 1) were loaded onto a 10% polyacrylamide gel which was developed and stained for activity as described under "Materials and Methods." HPI and HPII, hydroperoxidase isozymes. ence between the total and Mn-superoxide dismutase represents the iron-plus-hybrid form). About 5 mM Hz02 is normally required to inactivate the iron-isozyme (30).
Induction of Manganese-Superoxide Dismutase by Paraquat in oxyA3 and oxyR2 Mutants-Paraquat has been shown to increase the intracellular flux of 0;; and in turn, the cells induce manganese-superoxide dismutase to scavenge 0; (5,9,10,13,14). As expected, the synthesis of this enzyme in the wild-type (oxyR+) strains, RK4936 and K12, was inducible by the presence of paraquat in the growth medium (Table 111).
If induction of the manganese-containing superoxide dismutase is a function of the oxidative stress regulon, then a mutation altering the expression of the regulatory gene, oxyR, or its product should prevent or alter its induction by paraquat. This was not the case. The data ( Table 111) clearly showed that it was induced by paraquat in the oxyA3 mutant (strain TA4112) as well as in the oxyR2 mutant (strain TA4110). Thus, strain TA4112, which lacks the functional oxyR gene, did not appear to be hindered in its ability to induce manganese-superoxide dismutase as compared to the parental stain, RK4936. TA4110 contained greater amounts of this activity at both paraquat levels compared to its parent, K12. However, the percentage of manganese-isozyme of the total activity was about the same in both strains, indicating a similar level of induction.
Effect of oxyRP on Anaerobic Expression of Manganese-Superoxide Dismutase and HPI-Manganese-superoxide dismutase is not expressed in anaerobically grown cultures (8) except when iron chelators (10, 14, 15) or oxidants (11,12) are added to such cultures. Therefore, it was of interest to see if the dominant oxyR2 mutation plays a regulatory role under anaerobiosis. For this experiment, strain TA4110 and its parent, K12, were grown anaerobically in GY and TSY media, and cell-free extracts were prepared as described under "Materials and Methods." Analysis of superoxide dismutase revealed that only the iron form was synthesized anaerobically. Furthermore, Western blot analysis did not indicate the presence of manganese-superoxide dismutase antigen, thus suggesting the absence of inactive apoprotein (data not shown). On the other hand, under anaerobic conditions, catalase (HPI) levels in a midlogarithmic phase culture of TA4110 were 8.5-fold higher than those of K12 (data not shown).
oxyR2 Does Not Confer Protection against Paraquat Toxicity-We used the disc inhibition assay to determine the inhibitory effects of H202 and paraquat on the four strains

TABLE I1
Expression of manganese-superoxide dismutase after exposure to 60 P M hydrogen peroxide Gels were prepared from the extracts as described in the legend for Fig. 1 and were scanned to allow calculations of manganese-isozyme. (data not shown). The data indicated, in agreement with findings by Christman et al. (18), that the oxyR2 mutant (overproducer strain) was the most resistant to killing by hydrogen peroxide (60 and 300 pg/disc), whereas the oxyA3 mutant (deletion strain) was the least resistant. On the other hand, all four strains were equally sensitive to paraquat (38 and 190 pg/disc).

DISCUSSION
In enteric bacteria, hydrogen peroxide adaptation (31) has been shown to coincide with the induction of 30 proteins (18). Nine of these HpOz-inducible proteins are coordinately and positively regulated by the oxyR gene product (18). Catalase, alkylhydroperoxide reductase, and the manganese-containing superoxide dismutase are three of the enzymes identified to be regulated by the oxyR locus (18-20). Furthermore, these three antioxidant enzymes are reported to be overexpressed in oxyR constitutive mutants (18)(19)(20). The manganese-superoxide dismutase is known to be induced by oxygen (3, 8), redox-active compounds (5,9), and many oxidants (6, 11, 12). Therefore, the goal of this study was to elucidate the role of the oxyR locus in the regulation of this enzyme.
Superoxide Dismutase Is Not Induced during HZOz Adaptation-In this part of the study, we examined the biosynthesis of superoxide dismutase in oxyR+ (wild type), oxyA3 (deletion) and oxyR2 (constitutive) strains growing in GY medium in the presence of 60 p~ Hz02. This concentration of H202 was found, as expected (18), to induce catalase activity 2.5-3.5-fold in the oxyR+ strain, but had no effect on the oxyA3 or the oxyR2 strain. The data, however, clearly showed that the activities of superoxide dismutase isozymes were not affected by this treatment. These observations are in agreement with Demple and Halbrook (31), who also observed no change in superoxide dismutase activity and only a moderate 2-4-fold increase in catalase activity in extracts from wildtype E. coli exposed to hydrogen peroxide. The concentration of HzOz used in this study (60 p~) was very low and did not cause significant inhibition of the ironor hybrid-superoxide dismutase (Table I1 and data not shown), nor did it cause a significant change in the redox potential of the cells.' Recently, we have shown that 0.5 mM HzOz can induce manganese-superoxide dismutase only in a catalase-deficient mutant of E. coli due to a positive change in the cells' redox potential (12). The observation that catalase was induced by H202 treatment whereas manganese-superoxide dismutase was not affected suggests that sodA is probably not a part of the oxyR regulatory network. However, in agreement with   Christman et al. (18), we found that a strain constitutive in the expression of the oxidative stress regulon (oxyR2) did contain somewhat higher levels of this enzyme activity compared to the wild-type strain when both were grown in minimal media (see further discussion below). Induction of Manganese-Superoxide Dismutase Is Not Influenced by Mutations in oxyR Locus-In this part of the study, we used paraquat to induce manganese-superoxide dismutase in the different oxyR mutants (Table 111). It is clear that the oxyR deletion strain as well as the oxyR overproducer strain were not affected in their abilities to induce manganesesuperoxide dismutase in response to added paraquat. These results, together with the findings that all the test strains were equally sensitive to paraquat, as determined by disc inhibition assays, clearly suggest that the oxyR locus is not involved in the induction of manganese-superoxide dismutase by paraquat or in providing enhanced resistance to this compound.
oxyR Locus Is Functional during Anaerobiosis but Does Not Control Manganese-Superoxide Dismutase-It was of interest to observe that the oxyR2 mutant possessed higher levels of catalase compared to oxyR+ when grown in the absence of oxygen. This suggested that the oxidative stress regulon is anaerobically overexpressed in this strain, TA4110. However, if oxyR2 in TA4110 also affects the synthesis of manganesesuperoxide dismutase, then we should have seen the expression of sodA in anaerobic cultures of this strain. This was not the case. We were unable to detect any manganese-superoxide dismutase activity or antigen in cell-free extracts prepared from anaerobic cultures of K12 or TA4110.
Why Does oxyR Constitutive Mutant Have Slightly Higher Basal Level of Manganese-Superoxide Dismutase Than Its Parent Strain?-As indicated above, we found that strain TA4110 contained about 2-fold more manganese-superoxide dismutase than strain K12 when grown in GY medium (Table  11). This difference was not seen, however, when these two strains were grown in rich TSY medium (Table 111, 0 mM paraquat). The increased level of this enzyme seen in the oxyR constitutive strain (TA4110) in GY (but not in TSY) medium may be explained by some differences in their metabolism in the minimal medium. We examined the rate of oxygen uptake and the rate of endogenous 0; generated (32) by cells grown in GY and TSY media and found no significant difference^.^ The possibility still remains that TA4110 may have a lower internal concentration of iron when grown in a minimal medium, and/or other nutritional factors may also be involved.
In conclusion, the data demonstrate that the biosynthesis and induction of manganese-superoxide dismutase in E. coli are independent of the oxyR-controlled regulon. Recent work using sodA::lacZ protein fusions (17) is in agreement with our conclusion. We have previously reported (33) that this enzyme is not part of the inducible DNA repair (SOS) system. Furthermore, recent studies ( 17)4 have shown that the sodA gene is not part of the heat shock regulon. The possibility of a special superoxide-inducible (soi) regulon is being examined (34).