Unraveling the Mechanism for the Viability Deficiency of Shewanella oneidensis oxyR Null Mutant

ABSTRACT

Oxidative stresses triggered by reactive oxygen species (ROS) that damage various cellular components are unavoidable for virtually all living organisms. In defense, microorganisms have evolved sophisticated mechanisms to sense, respond to, and battle against ROS. Shewanella oneidensis, an important research model for applied and environmental microbes, employs OxyR to mediate the response to H₂O₂ by derepressing the production of the major H₂O₂ scavenger KatB as a major means toward these goals. Surprisingly, despite enhanced H₂O₂ degradation, the oxyR mutant carries a viability deficiency phenotype (plating defect), which can be suppressed by the addition of exogenous iron species. Experiments showed that the defect was not due to iron starvation. Rather, multiple lines of evidence suggested that H₂O₂ generated abiotically in lysogeny broth (LB) is responsible for the defect by quickly killing mutant cells. We then showed that the iron species suppressed the plating defect by two distinct mechanisms, either as an H₂O₂ scavenger without involving living cells or as an environmental cue to stimulate an OxyR-independent response to help cells cope with oxidative stress. Based on the suppression of the plating defect by overproduction of H₂O₂ scavengers in vivo, we propose that cellular components that are vulnerable to H₂O₂ and responsible for the defect may reside outside the cytoplasm.

IMPORTANCE In bacteria, OxyR is the major regulator controlling the cellular response to H₂O₂. The loss of OxyR results in reduced viability in many species, but the underlying mechanism is unknown. We showed in S. oneidensis that this defect was due to H₂O₂ generated abiotically in LB. We then showed that this defect could be corrected by the addition of Fe²⁺ or catalase to the LB or increased intracellular production of catalase. Further analyses revealed that Fe²⁺ was able not only to decompose H₂O₂ directly but also to stimulate the activity of OxyR-independent H₂O₂-scavenging enzymes. Our data indicate that iron species play a previously underappreciated role in protecting cells from H₂O₂ in environments.

INTRODUCTION

Oxidative stress, one of the unavoidable crises for the vast majority of aerobic organisms, is essentially an imbalance between the production of reactive oxygen species (ROS) and the ability of cells to counteract or detoxify their harmful effects (1). ROS include superoxide (O₂⁻), hydrogen peroxide (H₂O₂), and hydroxyl radical (·OH), which cause damage to biomolecules such as lipids, proteins, and DNA (1). As an uncharged species that penetrates membranes, H₂O₂ is life threatening by itself and, more importantly, by reacting with Fe²⁺ to generate the deadly toxic ·OH, a process called the Fenton reaction (2). H₂O₂ is ubiquitous because of the existence of diverse ways for its generation, both biologically and abiotically (3). H₂O₂, in concert with O₂⁻, is a metabolic by-product of cellular oxygen respiration, mostly through the accidental auto-oxidation of nonrespiratory flavoproteins and the turnover of committed oxidases (4, 5). Additionally, many organisms synthesize H₂O₂ and secrete it into their environment to suppress the growth of potentially competing and/or detrimental microbes (3). Moreover, H₂O₂ is formed by chemical processes without the involvement of living cells during the interaction of reduced metals and sulfur species with oxygen and, also, by photochemical reactions (3, 6). It is estimated that H₂O₂ is generated through aerobic respiration in Escherichia coli at a rate of 10 to 15 μM/s during growth in air-saturated glucose medium, whereas the amount of H₂O₂ produced by abiotic processes is as much as 20 μM (3, 7). In defense, organisms have evolved various strategies to cope with H₂O₂. The most effective and efficient strategy is to express enzymes that can directly decompose the oxidant, including alkylhydroperoxide reductase (Ahp), catalases, and various peroxidases (8). Ahp serves as the major scavenger when intracellular H₂O₂ levels are low, while catalases dominate when H₂O₂ levels are high. Although a large number of peroxidases have been identified, our understanding of them is much less complete, presumably because their effects are less apparent (9). Another strategy exploited by cells is to express ROS-resistant isoenzymes or pathways to replace the sensitive ones, such as activation of fumarase C and aconitase A, as well as induction of the alternative SUF Fe-S cluster synthesis system, in addition to the ISC housekeeping system (10, 11). Besides these mechanisms, repair systems like the base excision repair (BER) pathway and SOS are activated to address damaged components after the stress has been removed and cellular metabolism resumed (1).

In E. coli and many other bacteria, OxyR, a LysR family transcriptional regulator, is the predominant regulator mediating the cellular response to H₂O₂ (1). The activation of OxyR relies on the oxidation of two conserved cysteine residues and the formation of an intramolecular disulfide bond when H₂O₂ levels are over the limit (12). Upon activation, the regulator turns on dozens of genes, including the H₂O₂ scavenger genes ahpCF and katG (catalase gene), as well as the iron storage protein gene dps (13, 14). Conceivably, E. coli cells lacking the oxyR gene are hypersensitive to H₂O₂ (15, 16). While this phenotype is regarded as the hallmark for the oxyR mutants, the impacts of loss of the gene appear to be broad and diverse (17–20). The oxyR mutants are highly susceptible to redox-cycling agents, such as medadione and paraquat, which induce O₂⁻ generation, but not to organic peroxides, such as tert butyl hydroperoxide (21). In addition, the viability deficiency phenotype (plating defect) was often observed with the oxyR mutants (17, 21, 22). However, the growth rate of cells in liquid medium is not significantly affected by the oxyR mutations in general (22, 23). An exception to the E. coli OxyR model is the OxyR homolog of Neisseria species, whose loss results in enhanced resistance to H₂O₂ killing (24). This is because Neisseria OxyR acts as both a repressor of catalase expression and an activator for other members of its regulon (25, 26).

Members of the genus Shewanella, a group of Gram-negative facultative gammaproteobacteria, inhabit redox-stratified environments and are renowned for their versatile respiratory abilities (27). These bacteria are more sensitive to H₂O₂ than E. coli, in concert with the high susceptibility to UV and ionizing radiation, understandings derived from studies of the model species Shewanella oneidensis (28–30). Not surprisingly, S. oneidensis has an OxyR homolog, which is demonstrated to be the main regulator mediating the cellular response to H₂O₂ (29). However, several lines of evidence show that this protein has properties significantly different from those of its counterparts studied to date. First, although S. oneidensis OxyR acts as both an activator and a repressor, like the Neisseria OxyR proteins (26), it controls a much larger number of genes. Upon OxyR activation, S. oneidensis mainly relies on derepression of katB (encoding the major catalase) and dps genes to alleviate oxidative damage by degrading H₂O₂ and by sequestering the intracellular free iron, respectively (29). As a result, an S. oneidensis oxyR mutant degrades H₂O₂ more rapidly than the wild type. Meanwhile, a number of genes, including ahpCF and those encoding peroxidases, are positively controlled by OxyR, but their contribution to protection against oxidative stress is much less evident. Second, to date, S. oneidensis OxyR has been the only one that could not functionally complement E. coli OxyR, suggesting that this OxyR has evolved some unique structural features. Third, despite its enhanced ability to decompose H₂O₂, the oxyR mutant bears the plating defect, a phenotype observed in bacteria whose OxyR proteins act as a positive regulator for catalase genes (17, 22, 23). Fourth, the plating defect can be suppressed by supplying rather than limiting exogenous iron. This is in sharp contrast to the established strategy utilized by many other bacteria to combat oxidative stress, where Fe²⁺ binding and iron storage are induced during H₂O₂ stress, leading to reduced free iron levels (2).

The goal of this study was to elucidate the mechanism underlying the colony development defect of the S. oneidensis oxyR mutant. We found that this defect was neither critically linked to aerobic respiration, a major process generating H₂O₂ intracellularly, nor attributable to iron starvation. Rather, cells lacking OxyR became more vulnerable to H₂O₂ and died quickly on lysogeny broth (LB) plates because of H₂O₂ generated abiotically. We then developed evidence suggesting that iron functioned both as an H₂O₂ scavenger without involving living cells and as an environmental cue to stimulate an OxyR-independent response to help cells to combat oxidative stress. Furthermore, suppression of the plating defect by overproduction of H₂O₂ scavengers in vivo suggested that the components responsible may be localized outside the cytoplasm.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

All bacterial strains and plasmids used in this study can be found in Table 1. Information about all of the primers used in this study is given in Table S1 in the supplemental material. E. coli and S. oneidensis were grown in lysogeny broth (LB; Difco, Detroit, MI) under aerobic conditions at 37 and 30°C for genetic manipulation. For testing of the plating defect, both LB and the defined MS medium [KCl, 1.34 mM; NaH₂PO₄, 5 mM; Na₂SO₄, 0.7 mM; NaCl, 52 mM; piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), 3 mM; NH₄Cl, 28 mM; sodium lactate, 30 mM; MgSO₄, 1 mM; CaCl₂, 0.27 mM; and FeCl₂, 3.6 μM, pH 7.0] were used. When necessary, the following chemicals were added to the growth medium: 2,6-diaminopimelic acid (DAP), 0.3 mM; ampicillin, 50 μg/ml; kanamycin, 50 μg/ml; gentamicin, 15 μg/ml; and streptomycin, 100 μg/ml.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
E. coli strains
DH5α	Host for cloning	Laboratory stock
WM3064	ΔdapA, donor strain for conjugation	W. Metcalf, UIUC
S. oneidensis strains
MR-1	Wild type	Laboratory stock
HG0956-8	ΔahpCF, derived from MR-1	29
HG1070	ΔkatB, derived from MR-1	29
HG1328	ΔoxyR, derived from MR-1	29
HG3CAT	ΔkatB ΔkatG1 ΔkatG2, derived from MR-1	29
HGACAT	Δahp ΔkatG1 ΔkatG2, derived from MR-1	This study
HG0956-8-1070	ΔkatB ΔahpCF, derived from MR-1	This study
HG0976-1070	ΔkatB Δohr, derived from MR-1	This study
HG1070-2178	ΔkatB ΔccpA, derived from MR-1	This study
Plasmids
pHGM01	Apr Gmr Cmr, att-based suicide vector	31
pHG101	Kmr, promoterless broad-host vector	32
pHGE-Ptac	Kmr, IPTG-inducible expression vector	33
pHGEI01	Kmr, integrative lacZ reporter vector	34
pBBR-Cre	Spr, helper plasmid for antibiotic cassette removal	56
pHGE-Ptac-ahpCF	Vector for inducible expression of Ahp	This study
pHGE-Ptac-katB	Vector for inducible expression of KatB	This study
pHGEI01-ryhB	pHGEI01 containing the ryhB promoter	This study

Mutagenesis and complementation of mutant strains.

In-frame deletion strains of S. oneidensis were constructed using the att-based fusion PCR method as described previously (31). In brief, two fragments flanking the gene of interest were amplified by PCR with primers containing attB and the gene-specific sequence, which were linked by a second round of PCR. The fusion fragments were introduced into plasmid pHGM01 by using Gateway BP Clonase II enzyme mixture (Invitrogen) according to the manufacturer's instructions. The resultant plasmid was maintained in E. coli strain WM3064 and transferred to S. oneidensis by conjugation. Integration of the deletion constructs into the chromosome was selected by resistance to gentamicin and confirmed by PCR. Verified transconjugants were grown in LB broth in the absence of NaCl and plated on LB supplemented with 10% sucrose. Gentamicin-sensitive and sucrose-resistant colonies were screened by PCR for deletion of the target gene. To facilitate the growth of mutants, thiourea and/or catalase (from bovine liver; Sigma) were added to plates at the final resolution step for genes critical for survival through ROS. Mutants were verified by sequencing the mutated regions.

The promoterless plasmid pHG101 was used in genetic complementation of mutants as described previously (32). For inducible gene expression, the gene of interest was generated by PCR and introduced into the inducible plasmid pHGE-Ptac under the control of the promoter Ptac (33). After sequencing verification, the resulting vectors were transferred into the relevant strains via conjugation.

β-Galactosidase activity assay.

The β-galactosidase activity assay was performed to determine gene expression using the integrative lacZ reporter pHGEI01 (34). In brief, the sequence of ∼400 bp upstream from the gene of interest was amplified and placed in front of the full-length E. coli lacZ gene. The resulting vector was maintained in E. coli WM3064 and transferred to S. oneidensis after verification by sequencing. Cells of the log-phase culture (optical density at 600 nm [OD₆₀₀] of ∼0.4) were harvested by centrifugation, washed with PBS, treated with lysis buffer (0.25 M Tris-HCl, 0.5% Triton X-100, pH 7.5) for 30 min, and subjected to the o-nitrophenyl-β-d-galactopyranoside (ONPG)-based assay as described previously (35). β-Galactosidase activity was determined by monitoring the color development at 420 nm using a Synergy 2 multimode microplate reader (M200 Pro; Tecan), and the results are presented as Miller units.

H₂O₂ assay.

To measure the H₂O₂ concentrations in samples without exogenous iron species, the ferrous ion oxidation-xylenol orange (FOX) method was used (36). Mid-log-phase cells in liquid medium or cells grown overnight on plates were collected, washed twice in 50 mM NaHPO₄ buffer (pH 7.0), and resuspended in the same buffer to an OD₆₀₀ of 0.1. H₂O₂ was added to a final concentration of 0.5 mM, and the cells were incubated at 30°C. After 5 min, cells were filtered out and the elutions were assayed for the remaining H₂O₂. To assay the H₂O₂ concentrations in samples containing iron species, the leuco crystal violet (LCV) method was used (37). In brief, reagents were added for a total volume of 2 ml in the following order: 50 μM H₂O₂, compounds tested at the required concentrations, 100 mM KH₂PO₄ buffer (pH 4.0), 41 mM LCV dissolved with HCl, and 4 mg/50 ml horseradish peroxidase containing 1.5 mM sodium azide. The mixture was kept in the dark at room temperature for 30 min, to stabilize absorbance. Aliquots of 200 μl of the mixture were transferred into a 96-well plate, and the absorbance of crystal violet cation (CV⁺) was measured using a Tecan M200 pro microplate reader.

Viability assay.

Cells of the log-phase cultures (OD₆₀₀ of ∼0.4) were collected by centrifugation and adjusted to 10⁹ cell/ml, which was set as the undiluted (dilution factor 0) level. Tenfold serial dilutions were prepared with fresh LB. Five microliters of each dilution was spotted onto LB plates, on which additives (metal ions, catalase, thiourea, etc.) were added at the concentrations indicated in the figures. For a 15- by 100-mm plate containing approximately 25 ml of medium, metal ions and thiourea sterilized by filtration were added before pouring, while 2,000 units of catalase was spread evenly over the surface after solidification. The plates were incubated for 24 h or longer in the dark before being read. All experiments were repeated at least three times.

Quantification of intracellular total iron.

Quantification of total iron was performed as described previously (38). Cells grown overnight on LB plates were collected, washed with phosphate-buffered saline (PBS), and adjusted to similar densities (OD₆₀₀ of ∼0.6). Aliquots of 50 ml were mixed with 5 ml of 50 mM NaOH, sonicated on ice using 60 3-s bursts with a 4-s cooling period between each burst, and centrifuged at 5,000 × g for 10 min. The cell lysates (100 μl) were then mixed with 100 μl 10 mM HCl and 100 μl iron-releasing reagent (1.4 M HCl and 4.5% [wt/vol] KMnO₄ [1:1]) and treated at 60°C for 2 h. After cooling, the iron detection reagents (6.5 mM ferrozine, 6.5 mM neocuproine, 2.5 M ammonium acetate, and 1 M ascorbic acid in water) were added. The absorbance of samples was measured at 550 nm 30 min later. The standard curve was constructed using 0 to ∼300 μM ferric chloride.

Calcein assay to measure intracellular free iron.

Quantification of intracellular free iron was performed as described previously (39). S. oneidensis strains grown overnight on LB plates with or without 2 mM iron ions were collected by scraping cells with a metal loop, washing with PBS, and adjusting them to the same OD₆₀₀. Aliquots of cells were mixed with 5 μM calcein-AM (acetomethoxy derivate of calcein) for 30 min at 37°C. At the same time, other aliquots were treated with the iron chelator dipyridyl for 10 min prior to treatment with calcein-AM for comparison. Cells were then washed with PBS to remove the extracellular calcein-AM and monitored for signal changes using a Tecan M200 pro microplate reader (excitation at 485 nm and emission at 535 nm).

RESULTS

Plating defect of ΔoxyR mutant is alleviated on minimal medium agar plates.

Previously, we showed that the plating defect of the S. oneidensis oxyR mutant can best be rescued by adding either thiourea or iron species into LB medium (29). As shown by the results in Fig. 1A, visible growth of the wild-type culture following 10⁵× dilution (dilution factor 5, ∼5 × 10³ cell/ml) was observed, whereas the ΔoxyR strain could show growth only when it was undiluted (dilution factor 0, ∼5 × 10⁸ cell/ml). In the presence of thiourea at 5 and 10 mM, the viability was almost recovered to the wild-type level. Thiourea has been used routinely as a scavenger of hydroxyl radicals, but it was recently suggested that the ability of the chemical to help combat ROS should be attributed at least partially to its inhibitory effects on growth and/or metabolic rates (40). Hence, we reasoned that a medium allowing metabolism to run more slowly than in LB should alleviate the plating defect. To test this, we plated ΔoxyR cells on MS minimal medium agar plates (Fig. 1B). Indeed, the ability for colony development of the mutant was just one degree less robust than that of the wild type. However, the ΔoxyR colonies were still significantly smaller (see Fig. S1A in the supplemental material), indicating that the MS medium helps correct the plating but not the slow growth phenotype. To further explore the idea that the increased viability was associated with the reduced growth and/or metabolic rate, we assessed the effect of sucrose on colony formation of the ΔoxyR strain on LB plates. Sucrose, as an osmotic agent, hinders growth significantly at high concentrations (see Fig. S1B), a scenario observed previously (41). However, the viability defect was not evidently rescued by the addition of sucrose (see Fig. S1C). These data imply that the suppression of the plating defect of the ΔoxyR strain on MS plates may not be mainly attributable to the reduced metabolic rate and that there must be an unknown factor(s) responsible for the plating defect on LB plates.

FIG 1.

Plating defect of the oxyR mutant. The mutant was validated by genetic complementation in our previous study (29). Cultures at mid-log phase were collected, adjusted to 10⁹ cell/ml (dilution factor 0), and then serially diluted, and 5 μl of each dilution was dropped onto agar plates as indicated. The results shown are from after growth of the wild type plated at the lowest cell density (10⁴ cell/ml, dilution factor 5) became evident. Experiments were performed at least three times, and representative results are presented. (A) Effects of thiourea on the plating defect on LB plates. (B) Results on MS plates.

Catalase suppresses the plating defect of the ΔoxyR mutant.

The finding that the ΔoxyR strain has a plating defect on LB but not on MS plates prompted us to screen for agents that prevent survival/growth of the strain on LB plates. Although loss of OxyR derepresses the expression of the katB gene, conferring an enhanced H₂O₂ removal capacity on cells (29), we examined the effect of exogenous catalase (2,000 U [the same concentration was used in all experiments where exogenous catalase was added]) on the viability and growth of ΔoxyR cells on LB plates. As shown by the results in Fig. 2A, the plating defect was fully rescued, implying that H₂O₂ was present in the medium. We therefore reasoned that a katB mutant would be affected with respect to survival and growth on LB plates. Indeed, the loss of the katB gene greatly impaired viability; no growth was observed with dilutions of 100× or higher. As expected, the plating defect was not observed on MS plates (data not shown). We then performed an H₂O₂ assay to confirm the presence of the oxidant. In LB broth (8 h after autoclaving), the concentrations of H₂O₂ were approximately 5.2 ± 0.6 μM (mean ± standard deviation), and they were fairly stable with time (Fig. 2B). When catalase (2,000 U) was added, the H₂O₂ levels were reduced to the detection limit of the method. Similarly, H₂O₂ in the supernatants of ΔkatB cultures remained at relatively high levels, about 4 ± 0.5 μM at both the exponential (OD₆₀₀ of ∼0.3) and stationary (OD₆₀₀ of ∼1.2) phase. In contrast, the levels of H₂O₂ in the supernatants of cultures of the other strains tested decreased significantly. The wild-type cultures at exponential and stationary phases had values of 1 ± 0.2 and 3.8 ± 0.4 μM, respectively. Because of derepressed production of KatB, the ΔoxyR strain restricted the H₂O₂ levels by no more than 1.3 ± 0.3 μM. In the case of the MS medium, the levels of H₂O₂ were below 1 μM under all test conditions. These data imply a possibility that the plating defect is due at least in part to H₂O₂ generated in LB without the involvement of living cells. Moreover, the plating defect of the ΔoxyR strain was observed on LB plates lacking either tryptone or yeast extract (see Fig. S2 in the supplemental material). While this result indicates that both ingredients may be involved in H₂O₂ generation, tryptone alone evidently results in a defect more severe than is seen with yeast extract alone, suggesting that peptides and amino acids are the main source of abiotic H₂O₂ generation. It is worth mentioning that the slow-growth phenotype of the ΔoxyR strain remained the same in the presence of catalase (see Fig. S1A), implying that these two defects may result from functional impairment of different sets of proteins.

FIG 2.

LB contains H₂O₂. (A) Effects of catalase (2,000 U) on the plating defect of indicated strains on LB plates. Experiments were performed at least three times, and similar results were obtained each time. (B) H₂O₂ assay. H₂O₂ concentrations in indicated media (No cell) and in cultures at the indicated cell densities (OD₆₀₀ of 0.3 and 1.2) were determined. The data reported represent the mean results ± standard deviations (SD) (n = 4).

Cells of the ΔoxyR strain on LB plates die quickly.

To address the severe plating defect of the ΔoxyR strain, the catalase rescue assay was performed. We dropped wild-type and ΔoxyR cultures at 1,000× and 10× dilutions onto LB plates such that the wild-type but not the ΔoxyR strain could proceed to colony development (Fig. 3A). Catalase was then applied to the ΔoxyR culture drops at different times. We found that cells could form visible colonies only when the application was carried out immediately (0 h). These results show that diluted ΔoxyR cells not promptly protected by the catalase treatment simply die out.

FIG 3.

Cells of the ΔoxyR strain on LB plates die quickly. Experiments were performed at least three times, and similar results were obtained. (A) Effects of catalase on the plating defect of indicated strains on LB plates. Cultures were serially diluted, and the indicated dilution of each strain was dropped on LB plates. Catalase (2,000 U) was applied to the drops at the indicated times. Production of AhpCF and KatB at various levels was induced by IPTG. (B) Catalase staining analysis. Cells were collected just before and 30 min after the addition of 0.2 mM H₂O₂. Proteins from the indicated cell lysates were separated by native PAGE and stained for catalase activity. Overproduction of KatB induced by 0.2 mM IPTG resulted in KatB aggregates, forming the supercomplex.

In view of the protective effect of catalase on the viability of the ΔoxyR strain, we proposed that the plating defect was likely due to its impaired defense system. In addition to KatB, Ahp, a complex formed by the two subunits AhpC and AhpF, is another major H₂O₂ scavenger in S. oneidensis (29, 30). Because the transcription of ahp relies on OxyR activation, the production of Ahp is extremely low in the ΔoxyR strain (29). To test whether Ahp plays a role in the reduced viability conferred by the oxyR mutation, we placed the ahpCF genes under the control of the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible P_tac promoter within pHGE-Ptac (33). AhpCF produced at various levels by IPTG induction significantly improved the viability of the ΔoxyR strain (Fig. 3A). The ability of the mutant to achieve enhanced viability appeared proportional to the level of IPTG; with IPTG at 0.2 mM, cultures of 100× dilution were able to grow in the absence of catalase. However, the impaired viability of the ΔoxyR strain was apparently not due to reduced Ahp production alone, as an ahpCF deletion mutant exhibited normal survival and growth on LB plates (Fig. 3A).

Because of the derepression in the ΔoxyR strain, KatB is produced at a substantially elevated level compared to its level in the wild-type strain under noninducible conditions (Fig. 3B). Given that overproduction of Ahp could improve the viability of the ΔoxyR strain, we wondered whether increasing KatB production further would have a similar effect. Consistent with its reduced viability (Fig. 2A), the ΔkatB strain at 1,000× dilution survived and grew only when exogenous catalase was applied onto culture drops within 3 h (Fig. 3A). By using the same vector, the level of KatB production was controlled by IPTG. While the inducer at 0.05 mM did not show any noticeable effect on the viability of the ΔoxyR strain, the defect was completely corrected with 0.2 mM IPTG (Fig. 3A). To confirm overproduction of KatB, we performed catalase staining (Fig. 3B). Besides the fact that the expression pattern of KatB was in excellent agreement with the results of our previous study (29), we found that the levels of KatB increased drastically in the presence of IPTG at 0.2 mM, leading to the formation of a KatB supercomplex. These data collectively indicate that the ΔoxyR strain was vulnerable to H₂O₂ generated extracellularly and that this vulnerability could be rescued either by the removal of the oxidant with exogenous catalase or by enhancement of the intracellular scavenging ability.

The plating defect of the oxyR mutant is not caused by iron starvation.

One of the most striking findings about the S. oneidensis oxyR mutant was that the plating defect could be corrected by adding iron species into LB plates (29). To assess whether this phenomenon is specific to iron, we tested the ΔoxyR strain with other metals. As shown by the results presented in Fig. 4A, in contrast to the results for Fe²⁺, Zn²⁺, Mg²⁺, and K⁺ hardly improved the viability of the ΔoxyR and ΔkatB strains. Mn²⁺, however, appeared as effective as Fe²⁺ in elevating the viability of the katB mutant. While an effect of Mn²⁺ was evident in the oxyR mutant, it could not fully correct the defect in this strain. Compared to Fe²⁺, Fe³⁺ displayed a significantly reduced ability to correct the defect, to a level similar to that of Mn²⁺ (Fig. 4). These results imply a possibility that the ΔoxyR strain suffers from an iron shortage. However, the odds are low given that the defect is not observed on low-MS medium, which contains 3.6 μM Fe³⁺, about 5 times lower than the ∼17 μM iron concentration in LB (42). This is confirmed by the finding that levels of total iron in the wild-type and ΔoxyR strains were comparable under all test conditions (see Fig. S3 in the supplemental material).

FIG 4.

Plating defect of the oxyR mutant is not caused by iron starvation. Effects of indicated metal ions (Fe²⁺, FeSO₄; Mn²⁺, MnCl₂; Zn²⁺, ZnSO₄; Mg²⁺, MgCl₂; K⁺, KCl; and Fe³⁺, FeCl₃) on the plating defect of indicated strains on LB plates. For each metal ion, five different concentrations (2× increases) were tested, and the results for the concentrations that had the most significant effect are shown.

One explanation for this dilemma may be that the levels of free Fe²⁺ are low in the ΔoxyR strain. Similar to KatB, the iron storage protein Dps is produced at substantially increased levels upon the loss of OxyR (29). As a result, more free Fe²⁺ is sequestered by Dps, triggering a requirement for iron. To test this notion, we evaluated the intracellular free-iron content by measuring the expression of the ryhB gene, whose product is a small RNA responding to free Fe²⁺ to regulate the expression of genes involved in iron metabolism (43, 44). The results demonstrated that the ryhB promoter activities of the wild-type and ΔoxyR strains were not significantly different under the same conditions, suggesting that their free-iron levels are comparable (see Fig. S3A in the supplemental material). Similar results were obtained for the ΔkatB strain (see Fig. S3A). To further confirm this, we measured the intracellular levels of free iron in cells grown on LB plates with or without iron supplement by using the calcein assay (39). Consistent with the results for the ryhB promoter assay, the same pattern of response was displayed in all wild-type and mutant strains, including the ΔoxyR and ΔkatB strains (see Fig. S3B). These data collectively indicate that the intracellular iron content, either total or free, was not significantly affected by the oxyR mutation.

Iron species decompose H₂O₂ directly.

It is clear that the oxyR mutation has little influence on intracellular iron levels, ruling out the possibility that the addition of exogenous iron functions mainly to rescue an iron shortage. Alternatively, iron may scavenge H₂O₂ without involving cells, given that both ferrous and ferric ions were reported to react with H₂O₂ many years ago (6). To investigate this, the effects of Fe²⁺ and Fe³⁺ on H₂O₂ decomposition were examined using leuco crystal violet (LCV) reagent, which can reliably quantify H₂O₂ in the presence of iron species (37). In the presence of H₂O₂ and horseradish peroxidase, LCV forms a crystal violet cation, CV⁺, which absorbs at 590 nm and remains stable for days (37). As shown by the results in Fig. 5, in the presence of 50 μM Fe²⁺, only 28% of the initially added H₂O₂ (200 μM) remained 30 min after the addition (the same treatment time below unless otherwise noted). When Fe²⁺ was added to final concentrations of 150 and 200 μM, the levels of H₂O₂ remaining were reduced to ∼15 and less than 1 μM, respectively. For the ferric ion, the addition had no significant effect on H₂O₂ degradation when its concentration was no more than 600 μM (Fig. 5). However, degradation of H₂O₂ became evident with Fe³⁺ at 800 μM or higher. These results indicate that both ferrous and ferric ions are able to react with H₂O₂ but that Fe²⁺ is substantially more effective than Fe³⁺. We then examined whether Mn²⁺ and thiourea have similar effects given that both suppress the plating defect of the ΔoxyR strain to some extent. Interestingly, although thiourea at low concentrations (as shown by the results at 100 μM) displayed an ability to remove H₂O₂, it was less effective than iron species, leaving a significant amount of the oxidant in the medium even when the agent was added at levels greater than 1 mM. In contrast, degradation of H₂O₂ was not observed with Mn²⁺ at levels up to 2 mM. These results, in line with those showing that catalase is the most effective agent to suppress the plating defect of the ΔoxyR strain, suggest that direct H₂O₂ removal is likely the major mechanism involved. In parallel, other strategies, including at least one that is exploited by Mn²⁺, must exist.

FIG 5.

Iron species decompose H₂O₂ directly. LB broth containing 200 μM H₂O₂ was mixed with the indicated chemicals at various concentrations. After 30 min of incubation, the amount of H₂O₂ remaining was determined using the LCV method. The data were normalized to the initial concentration. The concentrations of the chemicals tested in the assay were as follows: Fe²⁺, 0.02 to 0.2 mM; Fe³⁺, 0.2 to 1 mM; thiourea, 0.1 to 2 mM; and Mn²⁺, 0.4 to 2 mM. The data reported represent the mean results ± SD (n = 4).

Iron ions facilitate degradation of H₂O₂ by living cells.

Although exogenous iron can scavenge H₂O₂ in cultures directly, the possibility that it also influences the ability of S. oneidensis cells to decompose the oxidant could not be excluded. To test this, we examined the capacities of relevant strains to decompose H₂O₂ when grown on LB plates in the absence and presence of exogenous iron. Cells grown on plates for 18 h were collected, washed, suspended to similar OD values, and subjected to the H₂O₂ degradation assay. Compared to cells on LB plates, the wild-type cells grown with exogenous Fe²⁺ and Fe³⁺ displayed faster H₂O₂ degradation (Fig. 6A). However, the enhanced capacities for H₂O₂ removal induced by the two iron ions were similar, in contrast with their substantially different abilities to remove H₂O₂ in cell-free environments. Interestingly, the increased H₂O₂ removal capacities are, at least in part, independent of KatB, as cells of the katB mutant grown with Fe²⁺ and Fe³⁺ also degraded H₂O₂ more rapidly than those grown without (Fig. 6A). To determine whether the increased H₂O₂-degrading capacities depend on the living cells, we performed the H₂O₂ assay with the same samples which were disrupted by sonication. All samples displayed reduced capacities for degrading H₂O₂ compared to the capacity of the untreated control (Fig. 6A). As catalases are H₂O₂-scavenging enzymes that can function independently of living cells (8), the results suggest that the increased H₂O₂-degrading capacity in cells grown with iron species is likely due to noncatalase enzymes, such as peroxidase. The experiment was then repeated with the addition of thiourea and Mn²⁺. While thiourea did not show any influence on the H₂O₂-degrading capacity, Mn²⁺ induced an increase in the ability to decompose the oxidant to levels similar to those observed with Fe²⁺ and Fe³⁺ (see Fig. S4 in the supplemental material).

FIG 6.

Iron ions facilitate degradation of H₂O₂ by living cells. (A) Cells of the indicated strains grown overnight on LB plates with or without the indicated additives were collected with a metal loop, washed, and suspended to an OD₆₀₀ of ∼0.1. Five minutes after the addition of 0.5 mM H₂O₂, the remaining H₂O₂ was determined. Con, initial concentration; −, LB without additive; LC, living cell; DC, disrupted cell. Asterisks indicate statistically significant differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001). Error bars indicate SD (n = 4). (B) Effects of indicated mutations on the plating defect on LB plates without or with 2 mM Fe²⁺. (C) Expression of ahp and katB in the indicated strains under indicated conditions. Cells grown on the LB plates without or with 2 mM Fe²⁺ for 16 h were collected, washed, and resuspended in PBS. Aliquots were treated with 200 μM H₂O₂ for 10 min, and ahp and katB promoter activities assayed using an integrative lacZ reporter.

Like the ΔkatB strain, the ΔoxyR strain also displayed elevated H₂O₂ removal capacities in the presence of Fe²⁺ and Fe³⁺, although the effects were much more modest than the effects on the wild type (Fig. 6A). This observation implies that the auxiliary H₂O₂-scavenging enzymes are unlikely to be tightly regulated by OxyR. Unfortunately, this could not be tested directly because we failed to remove the katB gene from the ΔoxyR strain. As alternatives, we examined the contributions of Ahp, KatG1, and KatG2 to the increased H₂O₂ removal capacities given that these proteins are a major peroxidase and additional catalases, respectively (29). ΔkatB ΔkatG1 ΔkatG2 (ΔCAT), and ΔkatB Δahp mutants were generated and characterized with respect to viability when grown with Fe²⁺ or without (Fig. 6B). Consistent with a previous finding (29), KatG1 and KatG2 were dispensable. However, the removal of ahp from the ΔkatB strain further reduced viability and, more importantly, the impaired viability could only be partially rescued by exogenous Fe²⁺. This result implies that Ahp may be responsible for the difference in the H₂O₂-degrading capacities of living and disrupted cells shown by the results in Fig. 6A. To test this, we assayed the H₂O₂ removal capacities of the ΔkatB Δahp strain grown with exogenous Fe²⁺ and Fe³⁺. As expected, the double mutant exhibited a further-reduced H₂O₂-degrading capacity but it was still able to decompose H₂O₂ (Fig. 6A), showing that additional systems exist for H₂O₂ removal. Additionally, we assessed the contributions to viability of CcpA and Ohr, two peroxidases that have been implicated in H₂O₂ scavenging in previous reports (30, 45). However, the viabilities of the ΔkatB ΔccpA and ΔkatB Δohr strains were indistinguishable from that of the ΔkatB strain under both conditions (Fig. 6B).

Furthermore, we determined whether exogenous Fe²⁺ affects the expression of these genes given that both katB and ahp contribute to viability. By using an integrative lacZ reporter system (34), we found that, in the wild-type strain, both genes were induced by the addition of Fe²⁺ into the medium (Fig. 6C). In the ΔoxyR strain, the katB gene was no longer responsive but the ahp gene displayed enhanced expression, albeit less substantial than its enhancement in the wild type, indicating that katB may be solely controlled by OxyR, while ahp is subject to regulation by an additional factor(s). Overall, these results indicate that KatB and Ahp are the predominant forces for H₂O₂ removal and that S. oneidensis is also equipped with OxyR-independent ROS-scavenging enzymes.

DISCUSSION

OxyR is now regarded as a pleiotropic regulator, and its loss causes multiple defects in various biological processes (1, 12). One of these is the viability deficiency. Since in most bacteria, OxyR functions as an activator, at least for H₂O₂-scavenging genes, the defect is attributed to reduced ability to scavenge endogenous H₂O₂, as most if not all of the enzymes carrying out the task require OxyR for expression. In S. oneidensis, OxyR acts as a repressor of the katB gene such that the production level of KatB in the oxyR mutant is comparable to the level in the wild type under H₂O₂-induced conditions, leading to rapid H₂O₂ degradation (29). We were surprised when we found that the S. oneidensis oxyR mutant also carries the viability defect. The purpose of this study was to identify the factors that account for this phenotype of the oxyR mutant.

The S. oneidensis oxyR mutant is not vulnerable to endogenous H₂O₂, as aerobic growth at various rates appears irrelevant to the plating defect. This notion is supported by the finding that the katB gene could not be removed from the oxyR mutant, given its predominance in limiting the accumulation of endogenous H₂O₂ (29). In contrast, the oxyR mutant cells suffer severe damage from the stress imposed by ambient H₂O₂, based on the unparalleled efficacy of exogenous catalase in suppressing the defect and its enzymatic specificity. Although abiotic generation of H₂O₂, up to 20 μM, in sterile laboratory medium has been mentioned in a previous review (3), such a finding remains unreported in research articles (personal communication with J. A. Imlay). Here, we showed that the amounts of H₂O₂ generated in LB are sufficient to differentiate mutants with impaired ability to respond to oxidative stress from the wild type, whereas the defined MS medium does not generate H₂O₂ to detectable levels.

Based upon the revelation of exogenous H₂O₂ as the major factor for the plating defect, we suggest a model to explain what we have observed in this study (Fig. 7). In diluted cultures, cells on plates are likely separated from one another and, thus, have to cope with H₂O₂ produced abiotically in the medium individually. The oxyR mutant cells carry pleiotropic defects, and one of the consequences is that they are vulnerable to extracellular H₂O₂. By inhibiting cellular components that are essential for survival and/or growth, ambient H₂O₂ rapidly kills these oxyR mutant cells (Fig. 7A). In contrast, in cultures with high cell densities, the cells exist in close proximity, resulting in patches of cells within which some cells are sequestered from a lethal dose of H₂O₂ and can proceed to replicate (Fig. 7B). In addition, H₂O₂ inhibition and killing are conceivably alleviated with higher starting cell numbers, as there is less H₂O₂ per cell and so less oxidative damage. In the case of exogenously added catalase, growth is allowed simply because the catalase effectively and efficiently removes H₂O₂. While this mechanism is applicable for presumably any agents that are able to scavenge H₂O₂, such as thiourea tested in this study, the degrees of efficacy can be critically different depending on the chemical features of the agents. Alternatively, the plating defect can be suppressed by enhancing the intracellular H₂O₂-scavenging ability of cells (Fig. 7C). When extracellular H₂O₂ is present, an H₂O₂ gradient is formed across the cytoplasmic membrane; the internal concentration is approximately an order of magnitude lower than the external concentration (46). In overabundance, KatB or Ahp confers upon cells a substantially elevated capacity to decompose H₂O₂, leading to low internal H₂O₂ concentrations. A direct consequence is that the flux of H₂O₂ into the cell is accelerated, promptly lowering the concentration of H₂O₂ in the proximity to a level permissive for survival and growth. It is worth mentioning that the oxyR mutation in S. oneidensis does not significantly affect growth in liquid LB, a scenario also found in some other bacteria (22, 23). A possible explanation is that some of the ΔoxyR cells lyse after inoculation, releasing catalase, which decomposes H₂O₂, into the liquid medium. On the plates, catalase from dead cells could not make such a contribution.

FIG 7.

Model illustrating the mechanism underlying the plating defect of the oxyR mutant. Cells face H₂O₂ at ∼6 μM (blue background) generated abiotically in LB but have intracellular levels lower than 0.05 μM (white background). An H₂O₂ gradient forms from extracellular space to the cytoplasm membrane because of the H₂O₂ influx. (A) The oxyR mutant cells, whose H₂O₂ scavengers are able to keep intracellular H₂O₂ levels below 0.05 μM, carry pleiotropic defects and are vulnerable to extracellular H₂O₂. OM, outer membrane; IM, inner membrane. (B) A high cell density allows the oxyR mutant cells to form aggregates. As a result, cells within aggregates are not accessible to extracellular H₂O₂, and thus, they survive and are able to grow. (C) Increased H₂O₂-degrading capacity promptly removes intracellular H₂O₂ and speeds up the H₂O₂ influx, reducing the concentration of H₂O₂ in the proximity and facilitating survival and growth.

This logic implies that many cellular components outside the S. oneidensis cytoplasm are vulnerable to H₂O₂. In support of this idea, many predicted targets of H₂O₂, including mononuclear iron proteins, Fe-S cluster-containing enzymes, proteins carrying active cysteine residues, and lipids (1, 47–50), are extensively localized outside the cytoplasm. This is particularly true with S. oneidensis, as the bacterium expresses more than 40 c-type cytochromes and many Fe-S cluster-containing proteins that reside exclusively and largely in the membranes and the periplasm, respectively. Conceivably, some of these proteins may carry out a function essential for growth or viability, and thus, the cell would stop growing and die if they were inactivated by ROS.

Among the metal ions tested, Fe²⁺ was the most effective in H₂O₂ removal. The addition of this iron species would be expected to reduce peroxide to hydroxyl radicals, as it does inside the cell. As such, the damage caused by hydroxyl would be limited to LB constituents or the exopolymeric substances on the surface of the cell. We are not certain of the mechanism underlying the direct decomposition of H₂O₂ by Fe³⁺ in the abiotic system used in this study. Compared to that of Fe²⁺, the efficiency was considerably lower. However, the impacts of Fe³⁺ and Fe²⁺ on the plating defect of the oxyR mutant were similar. One explanation is that S. oneidensis, which is renowned for metal reduction, is able to reduce Fe³⁺ to Fe²⁺ under aerobic conditions (51). It is also possible that LB also contains some superoxide, which can reduce Fe³⁺ to Fe²⁺ through the first step of the Harber-Weiss reaction (52). The subsequent second step is the Fenton reaction, which converts H₂O₂ to hydroxyl radicals. The presence of superoxide, generated biotically, abiotically, or both, may also offer an explanation for the plating defect and the effect of Mn²⁺, which can largely rescue the plating defect of the oxyR and katB mutants, although it is responsible for a marginal amount of decomposition of H₂O₂. Mn²⁺, established as the oxidative stress reliever, can scavenge superoxide either by cofactoring superoxide dismutase or functioning abiotically (53). Additionally, it is proposed that Mn²⁺ acts as an oxidant-resistant substitute for iron in the active site of many nonheme iron enzymes, which are vulnerable during oxidative stress (54). The S. oneidensis genome encodes additional catalases and a large number of peroxidases, some of which remain uncharacterized, implying the presence of OxyR-independent H₂O₂-scavenging enzymes. Our finding that katB-, katB ahp-, and oxyR-deficient cells grown in the presence of iron and manganese species had increased H₂O₂-scavenging capacities supports this notion. In E. coli, excess copper is found to be able to help protect cells from H₂O₂ damage by enhancing the activity of hydroperoxidase I (KatG), hydroperoxidase II, and superoxide dismutase, which is dependent on neither OxyR nor SoxS (55). However, it is clear that the additional help from proteins beyond the OxyR regulon is minor. Notably, the results presented here exclude the involvement of Ohr or CcpA. This is somewhat unexpected because previous work has shown (i) that the organic peroxide (OP)-responding regulator OhrR is also responsive to H₂O₂ in S. oneidensis and has a larger regulon than those studied (30) and (ii) that CcpA is a periplasmic cytochrome c peroxidase capable of removing H₂O₂ (45). In contrast, the increased H₂O₂-scavenging capacities are due to some of the OxyR regulon members which can be activated by a different regulator in the absence of OxyR. Our data indicate that Ahp is such a member. Based on the fact that its production increases in the presence of both Fe²⁺ and Mn²⁺, it is conceivable that the unknown regulator may be involved in the response to heavy metals. Efforts to test these notions are under way.

The work presented here also raises unanswered questions about the slow growth of the oxyR mutant. In bacteria in which OxyR functions solely as an activator, this universal phenotype resulting from the oxyR mutation has been conveniently attributed to the reduced H₂O₂-scavenging capacity (20–23). Our data suggest that this may not be the case, at least in the context of S. oneidensis. Although exogenous catalase perfectly suppresses the plating defect, the oxyR mutant still grows significantly more slowly than the wild type, indicating that this growth defect does not appear to be associated with H₂O₂-scavenging capacity. Addressing the underlying mechanisms for this represents an important challenge for future work.