Role for the oxyS Gene in Regulation of Intracellular Hydrogen Peroxide in Escherichia coli

Abstract

Intracellular hydrogen peroxide is regulated in Escherichia coli by OxyR in response to the metabolic production of H₂O₂. Here, we show that the untranslated oxyS RNA controlled by OxyR has a role in this regulation. The oxyS transcript appears to affect the metabolic output of H₂O₂ rather than the removal of H₂O₂ by catalases-hydroperoxidases.

The intracellular steady-state concentration of hydrogen peroxide (H₂O₂) depends on its rates of generation and dismutation. The major source of superoxide (O₂⁻) and H₂O₂ in Escherichia coli cells is the respiratory chain, which accounts for as much as 90% of the total production of H₂O₂ (10, 12). Even though the metabolic generation of H₂O₂ varies >10-fold during growth, E. coli cells maintain the intracellular concentration of hydrogen peroxide within a narrow range (0.20 ± 0.05 μM). This regulation is accomplished by a continuous variation in the degree of activation of the OxyR protein, which in turn governs transcription of katG, the gene encoding catalase-hydroperoxidase I (catalase–HP-I) (8).

Although catalase–HP-I constitutes the major detoxification system for endogenous H₂O₂ in exponential-phase E. coli cultures (8, 14), other regulated activities are involved in controlling H₂O₂ levels: ΔoxyR strains have a higher concentration of H₂O₂ and are more susceptible to exogenous oxidative stress than strains mutated only in katG (8).

In addition to catalase–HP-I, OxyR controls the expression of seven to eight other proteins in Salmonella typhimurium and E. coli (4, 11, 19). One of these, the ahpFC-encoded alkyl hydroperoxide reductase, was reported to react with H₂O₂ in vitro (20), but the activity did not contribute measurably to the regulation of the H₂O₂ concentration in vivo (8). Another member of the OxyR regulon, oxyS, was recently reported to have posttranscriptional regulatory functions (1). Strains overexpressing oxyS had a reduced rate of spontaneous and H₂O₂-induced mutagenesis, and cells carrying a deletion had near-wild-type resistance to challenge with exogenous H₂O₂ (1). These results prompted us to evaluate the possible role of oxyS in H₂O₂ homeostasis.

Bacterial strains and experimental procedures.

The strains of E. coli used in this study are listed in Table 1. Strains BGF416 and BGF420 were constructed by transduction (18) of the ΔoxyS2::Cm allele from strain GS035 into strains AB1157 and BGF611, respectively. Plasmid poxyS has the oxyS gene under the control of the tac promoter in a multicopy vector; even without induction by isopropyl-β-d-thiogalactoside, the level of the oxyS transcript expressed from poxyS is similar to the level seen in H₂O₂-treated wild-type E. coli (1). Plasmid psyxO has the gene in the reverse orientation. Cells were inoculated into Luria-Bertani (LB) broth (18) containing the appropriate antibiotic and incubated overnight at 37°C with gentle shaking (100 rpm). For experimental measurements, the saturated cultures were diluted 100-fold into fresh LB broth and incubated at 37°C for 3 h (optical density at 600 nm, ∼1). Antibiotics were used at the following concentrations (in micrograms per milliliter): tetracycline, 12.5; streptomycin, 50; chloramphenicol, 25; and ampicillin, 100. The intracellular concentration of H₂O₂ was assessed with peroxidase-mediated scopoletin oxidation as previously described (9). Total catalase activity was assayed by monitoring the disappearance of H₂O₂ at 240 nm in cell homogenates as described previously (8, 23) and normalized to the protein concentration determined with bovine serum albumin as the standard (15). The rate of H₂O₂ production was calculated from the experimental values for H₂O₂ and catalase concentrations as previously described (3, 8). The rate of O₂⁻ production was measured in membrane preparations by monitoring the superoxide dismutase (SOD)-sensitive rate of cytochrome c reduction at 550 nm (ɛ_reduced − ɛ_oxidized = 21 mM⁻¹ cm⁻¹) (2, 12). The reaction mixtures consisted of 50 mM potassium phosphate buffer (pH 7.4), 20 μM cytochrome c, 100 μM NADH, and membrane protein (∼0.2 mg/ml), with or without 50 U of bovine CuZn SOD. The rate of respiration by cell cultures was measured with a Clark-type electrode (5). Three-hour cultures were placed in the chamber of an oxygraph, and O₂ concentration was monitored for 5 to 10 min. Values are expressed in nanoatoms of O₂ per minute per 10⁶ cells. The temperature was 37°C. The cellular sensitivity to H₂O₂ was assessed on plates. Three-hour cultures were plated on LB plates, and a filter disk (10-mm diameter, Whatman no. 1) was soaked with 150 mmol of H₂O₂ in water and placed in the center of the plate. After 12 to 16 h, the diameter of growth inhibition was measured (11). The frequency of spontaneous mutations to rifampin resistance was measured by plating on LB plates containing 125 μg of rifampin per ml; Rif^r colonies were scored after 24 h of incubation at 37°C (11) and normalized to the number of cells plated (8).

TABLE 1.

Bacterial strains and plasmids of E. coli used in this study

Strain or plasmid	Genotype or relevant characteristic	Source or reference
Strain
AB1157	F−thr-1 leuB6 proA2 his-4 thi-1 argE2 lacY1 galK2 rpsL supE44 ara-14 xyl-15 mtl-1 tsx-33	Laboratory stock
BGF611	As AB1157, but katG17::Tn10	8
BGF416	As AB1157, but ΔoxyS2::Cm	This study
BGF420	As BGF611, but ΔoxyS2::Cm	This study
GS035	K-12 ΔoxyS2::Cm	1
RK4936	araD139 (argF-lac)205 glbB5301 non-(gyrA219) relA1 rpsL150 metE70 btuB::Tn10	Laboratory stock
TA4112	As RK4936, but Δ(oxyRS-btuB)3	4
Plasmid
pKK177-3	Ampr vector	Pharmacia
poxyS	pKK177-3 derivative carrying oxyS	1
psyxO	pKK177-3 derivative carrying oxyS in the reverse orientation	1

Regulation of intracellular concentrations of hydrogen peroxide.

The level of H₂O₂ in the oxyS-deficient strain BGF416 (in exponential phase) was ∼2-fold higher than the level measured for the parental oxyS⁺ strain (Fig. 1A), consistent with regulation mediated by oxyS. This increase was similar to that observed for a katG strain and less than that for a ΔoxyRS strain (Fig. 1A). We therefore tested whether mutations in both oxyS and katG would act synergistically to elevate the intracellular H₂O₂ concentration. A ΔoxyS katG double mutant (BGF420) had an ∼3-fold increase in H₂O₂ concentration (Fig. 1A), which shows that katG and oxyS play independent roles in the OxyR-dependent regulation of H₂O₂.

FIG. 1.

Effect of genetic deficiency in oxyS or katG on the intracellular concentration of hydrogen peroxide or catalase activity. (A) Steady-state H₂O₂ concentrations in intact cells. (B) Total catalase activity in cell extracts. Values are the means of four to six independent experiments ± SEMs. Strain abbreviations: WT1, AB1157 (oxyRS⁺ katG⁺); oxyS, BGF416 (oxyR⁺ oxyS2::Cm); katG, BGF611 (oxyRS⁺ katG17::Tn10); oxyS katG, BGF420 (oxyR⁺ oxyS2::Cm katG17::Tn10); WT2, RK4936 (oxyRS⁺ katG⁺); oxyR, TA4112 [Δ(oxyRS-btuB)3].

Increased steady-state concentrations of H₂O₂ can result from decreases in the rate of its decomposition or increases in the rate of H₂O₂ production. The catalase–HP-I is the major H₂O₂-decomposing activity in exponentially growing E. coli (8, 14), and it is possible that oxyS regulates katG by a posttranscriptional mechanism (1) distinct from the transcriptional activation of katG mediated by OxyR. However, the deletion of oxyS did not alter the total catalase activity in either the wild-type or the katG background (Fig. 1B). As previously reported, the katG mutant strain had an ∼70% lower catalase activity due to the lack of a functional catalase–HP-I (the remainder is the katE-encoded enzyme [8, 14]).

Since oxyS did not seem to affect expression of the primary H₂O₂ scavenging activity, we hypothesized that oxyS might influence the cellular generation of H₂O₂, most of which arises from the dismutation of O₂⁻ by SOD (10). We therefore tested the ΔoxyS, katG, and ΔoxyRS strains for changes in the rate of O₂⁻ production. The rates of superoxide anion production in membrane preparations (12) increased 6-fold in the ΔoxyS strain and 2.5-fold in the ΔoxyRS strain compared to their wild-type counterparts (Table 2). Expression of oxyS from the multicopy plasmid poxyS (1) complemented the ΔoxyS phenotype by preventing the increased superoxide production (Table 2). In the wild-type strain, the rate of O₂⁻ production was not changed significantly by poxyS, the vector plasmid, or a plasmid with oxyS in the reverse orientation (psyxO [1]) (Table 2 and data not shown).

TABLE 2.

Effect of oxyS mutation on the rates of H₂O₂ production in intact cells and O₂⁻ production in isolated membranes^a

Strain (relevant genotype)	d[H2O2]/dt (μM/s)	d[O2−]/dt (nmol/min/mg of protein)
		Without poxyS	With poxyS
AB1157 (oxyRS+)	2.1 ± 0.7	0.3 ± 0.1	0.2 ± 0.1
BGF416 (oxyR+ ΔoxyS)	3.6 ± 0.7	1.7 ± 0.5	0.3 ± 0.1
BGF611 (oxyRS+ katG)	0.9 ± 0.7	0.15 ± 0.1	0.08 ± 0.02
RK4936 (oxyRS+)	2.2 ± 0.5	0.20 ± 0.03	0.3 ± 0.1
TA4112 (ΔoxyRS)	1.9 ± 0.5	0.5 ± 0.1	0.16 ± 0.04

^aThe rates of H₂O₂ production were calculated with the experimental values for H₂O₂ and catalase concentrations (Fig. 1). The rates of O₂⁻ production were measured in membranes isolated from 3-h cultures of the strains listed. Values are means of four to six independent experiments ± SEMs. 

Unexpectedly, there seemed to be a small decrease in O₂⁻ production in the katG strain BGF611, but this was not statistically significant (Table 2). However, this strain is oxyR proficient, and it may be that there is diminished O₂⁻ production due to OxyR-dependent induction of oxyS as a result of the increased H₂O₂ concentration in this strain (Fig. 1A). Indeed, forcing the increased expression of oxyS by itself with poxyS was sufficient to decrease O₂⁻ production in BGF611 as well as in the ΔoxyRS strain TA4112 (Table 2).

The rates of H₂O₂ production were calculated from the experimental values for the steady-state H₂O₂ concentration and the total catalase concentration (3, 8). Deletion of oxyS in strain BGF416 increased H₂O₂ production 1.7-fold, compared to a 6-fold increase in O₂⁻ generation in this strain (Table 2). We have not determined whether H₂O₂ production is decreased by poxyS in the ΔoxyS strain. In principle, the 2:1 O₂⁻-H₂O₂ stoichiometry of superoxide dismutation (3) would predict a maximal threefold increase in H₂O₂ generation in the ΔoxyS strain. However, reactions other than SOD can consume O₂⁻ (7), and such reactions may contribute to this difference. The lack of a significant increase in H₂O₂ production in the ΔoxyRS strain (TA4112) (Table 2) could also be related to non-SOD pathways consuming superoxide.

The metabolic production of O₂⁻ and H₂O₂ in E. coli depends on the number of active respiratory chain units per cell and on the energetic state of the chains, which can be altered by directing the electron flux through components with higher or lower energetic efficiency (coupled versus uncoupled components) (10, 22). Therefore, changes in growth conditions or in the proportions of coupled and uncoupled components would determine the rate of free radical production at the respiratory chain. The effect of oxyS on the metabolic production of superoxide and hydrogen peroxide reported here could be due to this type of regulation. Three of the eight oxyS-regulated genes reported so far, fhlA, gadB, and uhpT, are involved in energy metabolism, although they have not been related directly to electron transport processes. The fhlA gene encodes a transcriptional activator of the hydrogenase pleiotropic operon (hypABCDE) (16, 17). The gadB gene encodes a glutamic acid decarboxylase (21), and uhpT encodes the sugar phosphate transporter (6, 13). Perhaps oxyS represses some energy pathways that leak more O₂⁻ and are deleterious to cells under oxidative stress.

To test the hypothesis that oxyS deletion affects the energy metabolism of the cells, we measured the rates of respiration in intact wild-type and ΔoxyS cells. Oxygen uptake by exponentially growing cells was significantly increased in the ΔoxyS strain (mean rate of O₂ consumption ± standard error of the mean [SEM] for three independent experiments, 0.77 ± 0.02 nanoatoms of O₂/min/10⁶ cells; wild type, 0.51 ± 0.06 nanoatoms of O₂/min/10⁶ cells), and this effect was largely suppressed by introduction of the multicopy plasmid poxyS (0.63 ± 0.07 nanoatoms of O₂/min/10⁶ cells). As seen for the rate of O₂⁻ production (Table 2), expression of additional oxyS from poxyS in the wild-type strain may slightly decrease the rate of respiration (from 0.51 ± 0.06 to 0.44 ± 0.04 nanoatoms of O₂/min/10⁶ cells). We therefore conclude that oxyS does affect cellular respiration. More-detailed studies will be required to delineate the mechanism underlying this regulation.

We previously reported that twofold increases in the rate of production of H₂O₂ are enough to trigger a substantial OxyR-dependent transcription of katG (10) and that twofold increases in the steady-state concentration of H₂O₂ significantly increased the frequency of spontaneous mutation (8). We therefore tested whether the lack of oxyS or oxyS and katG resulted in phenotypic changes. We measured the frequency of spontaneous mutation (a sensitive marker of oxidative DNA damage [8]) and the cellular sensitivity to exogenous H₂O₂ in the various strains. Compared to the wild-type strain (AB1157), both parameters were unchanged in the ΔoxyS strain (BGF416), but there were significant increases in spontaneous mutation (2.8-fold) and H₂O₂ sensitivity (1.5-fold) in the ΔoxyS katG strain (BGF420) (Table 3). Multicopy oxyS had an antimutagenic effect in the ΔoxyS and katG strains and possibly in the oxyRS⁺ strain (Table 3). Multicopy oxyS in the double mutant strain BGF420 (ΔoxyS katG), decreased the frequency of spontaneous mutation to almost the same value as obtained the katG strain without poxyS (Table 3). Interestingly, multicopy oxyS did not complement the mutator phenotype of the ΔoxyRS strain (Table 3). This result suggests that, in addition to oxyS, either katG or some other OxyR-dependent activity is critical for limiting mutagenesis by endogenous H₂O₂. Alternatively, oxyS in poxyS is not regulated in response to oxidative stress (1), so the level of the RNA does not adjust to changing H₂O₂ production.

TABLE 3.

Oxidative damage and sensitivity to H₂O₂ in exponentially growing E. coli^a

Strain (relevant genotype)	Mutation frequency (Rifr colonies/109 cells)		H2O2 sensitivity (killing zone, mm)
	Without poxyS	With poxyS
AB1157 (oxyRS+)	10 ± 2	6 ± 5	8 ± 1
BGF416 (oxyR+ ΔoxyS)	11 ± 2	5 ± 1	9 ± 1
BGF611 (oxyRS+ katG)	16 ± 1	8 ± 1	10 ± 1
BGF420 (oxyS katG)	28 ± 2	20 ± 2	12 ± 1
RK4936 (oxyRS+)	7 ± 1	5 ± 1	7 ± 2
TA4112 (ΔoxyRS)	80 ± 8	100 ± 10	17 ± 1

^aThree-hour cultures were plated on LB plates with rifampin or LB plates with a filter paper soaked in H₂O₂, as described in the text. Values are means of four to six independent experiments ± SEMs. 

Multicopy oxyS did not significantly alter the sensitivity to hydrogen peroxide in any of the strains listed (data not shown), in agreement with the lack of regulation of katG by oxyS (Fig. 1B).

Our results show an oxyS-dependent regulation of the intracellular production of oxygen free radicals. In this way, the oxyS pathway and the OxyR-dependent induction of catalase–HP-I would provide two independent and complementary mechanisms to limit the levels of H₂O₂ during aerobic growth and possibly under oxidative stress. Elimination of either pathway is still compatible with almost normal aerobic growth, but elimination of both oxyS and katG produces a ΔoxyRS-like phenotype with significant increases in the frequency of spontaneous mutations and sensitivity to H₂O₂. Defining the mechanism(s) by which oxyS limits O₂⁻ production in respiring E. coli may reveal general pathways to avoid oxidative damage.