The YaaA Protein of the Escherichia coli OxyR Regulon Lessens Hydrogen Peroxide Toxicity by Diminishing the Amount of Intracellular Unincorporated Iron^,

Abstract

Hydrogen peroxide (H₂O₂) is commonly formed in microbial habitats by either chemical oxidation processes or host defense responses. H₂O₂ can penetrate membranes and damage key intracellular biomolecules, including DNA and iron-dependent enzymes. Bacteria defend themselves against this H₂O₂ by inducing a regulon that engages multiple defensive strategies. A previous microarray study suggested that yaaA, an uncharacterized gene found in many bacteria, was induced by H₂O₂ in Escherichia coli as part of its OxyR regulon. Here we confirm that yaaA is a key element of the stress response to H₂O₂. In a catalase/peroxidase-deficient (Hpx⁻) background, yaaA deletion mutants grew poorly, filamented extensively, and lost substantial viability when they were cultured in aerobic LB medium. The results from a thyA forward mutagenesis assay and the growth defect of the yaaA deletion in a recombination-deficient (recA56) background indicated that yaaA mutants accumulated high levels of DNA damage. The growth defect of yaaA mutants could be suppressed by either the addition of iron chelators or mutations that slowed iron import, indicating that the DNA damage was caused by the Fenton reaction. Spin-trapping experiments confirmed that Hpx⁻ yaaA cells had a higher hydroxyl radical (HO^•) level. Electron paramagnetic resonance spectroscopy analysis showed that the proximate cause was an unusually high level of intracellular unincorporated iron. These results demonstrate that during periods of H₂O₂ stress the induction of YaaA is a critical device to suppress intracellular iron levels; it thereby attenuates the Fenton reaction and the DNA damage that would otherwise result. The molecular mechanism of YaaA action remains unknown.

INTRODUCTION

The first living organisms appeared on an Earth that was anaerobic and that remained so for another 2 billion years (5). By that point, however, photosynthetic processes had generated enough oxygen to titrate the reduced iron and sulfur species in the ancient seas, and over the next billion years the continued production of oxygen led to its accumulation in the atmosphere until concentrations reached present-day levels. During this period, microbial organisms evolved mechanisms to exploit the metabolic potential of oxygen—but they also had to devise strategies to defray the hazards it posed. Molecular oxygen is a triplet species that reacts rapidly with organic radicals; therefore, once aerobic environments appeared, ancient free-radical-based catalytic mechanisms were either dispensed with or were sequestered in the interiors of enzymes, where oxygen could not penetrate.

A second threat consisted of superoxide and hydrogen peroxide, partially reduced forms of oxygen that are more potent oxidants than is oxygen itself. These species are routinely generated within aerobic cells, apparently when molecular oxygen steals electrons from the reduced flavins or quinones of redox enzymes (27, 28, 36). Their potential for toxicity is implied by the high titers of scavenging enzymes—superoxide dismutases and reductases, and catalases and peroxidases—that are found in virtually all living cells. Mutant bacteria that lack these enzymes do not exhibit phenotypic defects as long as they remain in anaerobic habitats, but upon exposure to oxygen they grow poorly, suffer high rates of mutagenesis, or die (7, 43). Biochemical and physiological studies of such mutants have allowed the identification of some of the targets with which superoxide and H₂O₂ react most avidly. Both superoxide and hydrogen peroxide oxidize the exposed iron-sulfur clusters of aconitase class dehydratases, precipitating their degradation and the loss of enzyme activity (12, 23, 30). Hydrogen peroxide also reacts directly with ferrous iron, and this reaction inactivates enzymes that use mononuclear ferrous iron cofactors (45a). A similar reaction on the surface of DNA produces hydroxyl radicals, and their subsequent oxidation of nucleic acid bases or ribose moieties can cause mutagenesis or cell death (20, 40).

In wild-type Escherichia coli, the best-studied system, the levels of scavenging enzymes are sufficient to drive the steady-state concentration of superoxide below nanomolar levels and the H₂O₂ concentration to about 10⁻⁸ M (21, 44). These concentrations are low enough to avoid any fitness defect, as higher expression of scavenging enzymes provides no further stimulus to aerobic growth. Thus, the basal defenses of bacteria appear to be calibrated to avoid problems from endogenous oxidants.

Nevertheless, bacteria may be critically damaged by H₂O₂ when it enters the cell at high rates from the external environment. Hydrogen peroxide is produced when oxygen reacts with metals and thiols, a phenomenon that occurs in the interfaces between aerobic and anaerobic habitats. Because of its toxicity, H₂O₂ is also used as a weapon in a variety of natural antimicrobial systems: plants, amoebae, and mammals all elaborate an NADPH oxidase to kill invading bacteria (4, 13, 35), and both plants and microbes secrete redox-cycling phenazines and quinones to suppress the growth of their competitors (22, 48). Hydrogen peroxide penetrates membranes at a rate similar to that of water (44), so when bacteria enter an environment that contains H₂O₂, they must activate defensive responses. The first of these to be found, the OxyR system, is triggered when peroxide oxidizes an activated thiolate residue to a sulfenic acid on the OxyR transcription factor (31, 55). The activated OxyR protein then directly stimulates transcription of genes involved in defense. Microarray studies performed in E. coli have suggested that at least two dozen genes lie within the OxyR regulon (57).

These genes presumably signify a list of cellular strategies to fend off the toxic actions of H₂O₂. Not surprisingly, they include katG and ahpCF, which encode the primary catalase and peroxidase in the cell (43). Dps is a ferritin-like protein that sequesters unincorporated iron and thereby reduces its role in damaging DNA (14, 19, 40, 54); the Fur repressor is also induced (56), and it diminishes iron import with the same end effect (49). Hydrogen peroxide directly poisons the housekeeping Isc iron-sulfur cluster assembly system (24), and the induction of the Suf operon provides an H₂O₂-resistant replacement (32, 38). Suf induction is therefore also critical for the repair of damaged iron-sulfur clusters. The induced synthesis of a manganese importer, MntH, enables manganese to replace iron in mononuclear metal enzymes which otherwise would be damaged by Fenton chemistry (2, 25, 45a).

The significance of these adaptations is evident in E. coli mutants that lack both catalase and peroxidase (katG katE ahpCF, or Hpx⁻). These strains accumulate ca. 1 μM H₂O₂, which is slightly above the dose that triggers expression of the OxyR regulon (3, 44). Thus, whereas dps or mntH single mutants can tolerate oxygen, Hpx⁻ dps and Hpx⁻ mntH mutants are viable only in anaerobic medium (2, 40).

The microarray data (57) indicated that the OxyR regulon also includes several genes whose roles are not obvious. One of these, yaaA, is predicted to encode a 29.6-kDa cytoplasmic protein that belongs to a protein family domain of unknown function (DUF328). To explore the role of YaaA during H₂O₂ stress, we have examined the phenotype of a gene deletion in an Hpx⁻ background. The results indicate that during H₂O₂ stress YaaA prevents oxidative damage to both DNA and proteins by diminishing the amount of unincorporated iron within the cell.

MATERIALS AND METHODS

Reagents.

β-Mercaptoethanol, bovine liver catalase, Chelex 100, l-cystine dihydrochloride, deferoxamine mesylate (DFO), diethylenetriaminepentaacetic acid (DTPA), ferric chloride, type II horseradish peroxidase, 30% hydrogen peroxide, ο-nitropheny-β-d-galactopyranoside (ONPG), thiamine, and trimethoprim (TMP) were purchased from Sigma. Potassium cyanide and α-(4-pyridyl-1-oxide)-N-tert-butylnitrone (4-POBN) were from Aldrich. Hanks balanced salt solution without CaCl₂, MgCl₂, MgSO₄, and phenol red (HBSS) was obtained from Gibco. Hy-Case Amino and 2,4-dinitrophenylhydrazine (DNPH) were from Fluka, and Amplex Red was purchased from Molecular Probes. Standard reagents for RNA work were obtained from Ambion. iQ SYBR Green supermix was from Bio-Rad, and real-time PCR plates were purchased from Eppendorf.

Strain constructions.

The strains used in this study are listed in Table 1. Deletions were constructed by the Red recombination deletion method (10) and were transferred to other strains by P1 transduction (37). The antibiotic resistance cassettes were subsequently removed by FLP-mediated excision (10). Gene fusions between the yaaA upstream region (bp −290 to −1) and lacZ were created by standard methods and integrated into the lambda phage attachment site (18), preserving the wild-type yaaA locus at its native chromosomal locus.

Table 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype or characteristics	Source or reference
E. coli strains
AA30	As LC106 plus ΔmntH2::cat	2
AL145	As LC106 plus Δ(lacZ1::cat)1 attλ::[pSJ501::yaaA′-lacZ+]∼cat	This study
AL154	As AL157 plus Δ(ahpC-ahpF′ kan::′ahpF	This study
AL157	As MG1655 plus Δ(lacZ1::cat)1 attλ::[pSJ501:: yaaA′-lacZ+]∼cat	This study
AL194	As LC106 plus Δ(yaaA1::cat)1 fur::kan zbf-507::Tn10	This study
AL221	As LC106 plus Δ(yaaA1::cat)1 ΔmntH2::cat	This study
AL223	As MG1655 plus recA56 srl300::Tn10 Δ(yaaA1::cat)1	This study
AL245	As MG1655 plus Δ(yaaA1::cat)1	This study
AL273	As LC106 plus Δ(feoABC::cat)1 ΔentA1::cat Δ(lacZ1::cat)1 attλ::[pSJ501::fecA′-lacZ+]∼cat	This study
AL285	As LC106 plus Δ(yaaA1::cat)1 Δdps::cat	This study
AL293	As AL273 plus Δ(yaaA1::cat)1	This study
AL344	As AL145 plus ΔoxyR::spec	This study
AL435	As LC106 plus ΔfieF1::cat	This study
AL437	As LC106 plus Δ(yaaA1::cat)1 ΔfieF1::cat	This study
BW25113	lacIqrrnBT14lacZWJ16hsdR514 araBADAH33rhaBADLD78	10
LC106	As MG1655 plus Δ(ahpC-ahpF′) kan::′ahpF Δ(katG17::Tn10)1 Δ(katE12::Tn10)1	45
LEM17	As MG1655 plus recA56 srl300::Tn10	Laboratory stock
MG1655	Wild type	E. coli Genetic Stock Center
SB53	As LC106 plus Δ(yaaA1::cat)1	This study
SP66	As LC106 plus mhpC281::Tn10 lacY1 dps::cat	40
Plasmids
pAH125	Plasmid for lacZ transcriptional fusion construction	18
pCP20	Temperature-sensitive Flp expression plasmid	9
pKD3	FRT-flanked cat	10
pKD46	λ-Red recombinase expression plasmid	10
pACYC184	Cmr Tetr	8
pGSO58	oxyR A233V in pACYC184	29
pSJ501	pAH125 derivative with FRT-flanked cat	24
pWKS30	Plac polylinker Ampr	50
pWKyaaA	pWKS30 containing yaaA insert	This study

Plasmid construction.

The yaaA coding sequence was amplified by PCR from E. coli MG1655 genomic DNA using the forward primer 5′-TTTCAGAATTCTAAATTTCCTGCAAGGACTG-3′ and the reverse primer 5′-GATCCTCTAGACAAACTTAACGCTGCTCGTA-3′. These oligonucleotide primers were designed to engineer EcoRI and XbaI sites at the 5′ and 3′ ends of the gene, respectively. The PCR products were purified, digested with EcoRI and XbaI, and cloned into the pWKS30 vector (50), so that yaaA expression was controlled by the lac promoter. The correctness of the plasmid construct was confirmed by sequencing.

Bacterial growth.

Complex growth medium was LB (10 g Bacto tryptone, 5 g yeast extract, and 10 g NaCl per liter). Defined glucose/amino acids medium contained minimal A salts (37), 0.2% glucose, 0.2% Casamino Acids, 0.5 mM tryptophan, 0.02% MgSO₄ · 7H₂O, and 5 μg/ml thiamine. When antibiotic selection was required, media were supplemented with 100 μg/ml ampicillin, 20 μg/ml chloramphenicol, 30 μg/ml kanamycin, 100 μg/ml spectinomycin, or 12 μg/ml tetracycline.

Experimental protocols were designed to ensure that measurements were performed upon exponentially growing cells. For growth studies, anaerobic overnight cultures were diluted into anaerobic medium and grown for four to five generations to early log phase (optical density at 600 nm [OD₆₀₀] of 0.15 to 0.20). The cultures were then diluted into aerobic medium of the same composition to an OD₆₀₀ of ≈0.0005 to 0.010. All cultures were grown at 37°C. Aerobic LB medium was made 1 day prior to culturing and stored in the dark to avoid the photochemical generation of H₂O₂. Aerobic defined medium was prepared immediately prior to use. Anaerobic medium was transferred immediately after autoclaving to an anaerobic chamber (Coy Laboratory Products), where it was stored under an atmosphere of 5% CO₂–10% H₂–85% N₂ for at least 1 day prior to use. When used, DFO was made fresh and added into the medium at a final concentration of 2 mM before inoculation.

Cell viability.

To determine the cell viability, cell samples were collected at designated time points and mixed with 250 U/ml catalase (final concentration) to scavenge residual hydrogen peroxide. Samples were then serially diluted, mixed with top agar, and spread onto the surface of anaerobic plates. Colonies were allowed to grow anaerobically for 24 to 48 h before counting.

DIC microscopy.

Cultures were grown in anaerobic LB medium to exponential phase before dilution into aerobic LB medium to an OD₆₀₀ of 0.0005. They were then grown for 5 h. Images were captured by differential interference contrast (DIC) microscopy (17).

thyA forward mutagenesis assay.

In order to determine the frequency of mutagenesis, Thy⁻ mutants were selected using TMP (37). Strains were transferred from anaerobic LB to aerobic LB as described above. At intervals aliquots were removed, mixed with catalase to stop oxidative stress, and washed with minimal glucose/amino acids medium at room temperature three times. The cells were placed in an anaerobic chamber and diluted into minimal A salts. The number of viable cells was determined by dilution and plating in 2 ml F-top agar containing 1 mg/ml thymidine on defined glucose/amino acids plates. thyA mutants were selected by including 0.2 mg/ml TMP in the F-top agar, and CFU were determined after anaerobic incubation at 37°C for 48 h. For each experiment, a subset of the trimethoprim-resistant colonies was tested and confirmed to be Thy⁻ by examining their growth on minimal glucose/amino acids plates with and without thymidine supplementation.

H₂O₂ killing assay.

Cells were grown in minimal A glucose medium containing 0.5 mM tryptophan, phenylalanine, and tyrosine to an OD₆₀₀ of 0.15 as described above. The cultures were then diluted to an OD₆₀₀ of 0.025 in the same medium and incubated for 5 min with either 3.3 mM potassium cyanide (53) or 0.5 mM cystine (39). Hydrogen peroxide (2.5 mM) was then added. At the indicated time points aliquots were removed, and 2,500 U/ml catalase (final concentration) was added in order to end the hydrogen peroxide stress. After serial dilution, samples were mixed with top agar and spread on minimal A defined medium plates. CFU were determined after anaerobic incubation at 37°C for 48 h.

EPR spin trapping of hydroxyl radicals.

Cells were cultured in aerobic LB as described above. The cells were then harvested at an OD₆₀₀ of 0.2, when the growth of Hpx⁻ yaaA cells slowed down. The cells were washed twice with 25 ml of ice-cold Chelex-treated HBSS and resuspended in 0.5 ml of ice-cold HBSS. The cell suspension was added to a room temperature mixture (total volume, 1 ml) containing 100 mM DTPA, 10 mM 4-POBN, 170 mM ethanol, 1 mM H₂O₂, and HBSS, and the POBN hydroxyl radical adducts were quantified by electron paramagnetic resonance spectroscopy (EPR) as described previously (33). The final cell densities were equivalent in all samples (OD₆₀₀ of ∼20).

EPR measurement of intracellular unincorporated iron.

The cells were cultured in 600 to 1,000 ml of aerobic LB medium as described above. Cells were centrifuged and resuspended at a 100-fold-higher cell density in LB medium containing 10 mM DTPA (pH 7.0) and 20 mM DFO (pH 8.0). The suspension was incubated at 37°C aerobically for 15 min. The purpose of the cell-impermeable chelator DTPA was to block further iron import, while the cell-penetrating iron chelator DFO facilitated oxidation of unincorporated ferrous iron to EPR-detectable ferric iron, as described previously (52). The cell pellet was washed twice with 5 ml of ice-cold 20 mM Tris-HCl (pH 7.4) and finally resuspended in 200 μl of ice-cold 20 mM Tris-HCl–30% glycerol (pH 7.4). The EPR analysis was performed as described previously (33). The measured EPR signals were normalized by cell density and converted to intracellular concentrations by using the following conversion: 1 ml of E. coli culture at an OD₆₀₀ of 1 comprises 0.52 μl of intracellular volume (21).

β-Galactosidase activity.

Cell extracts were prepared in ice-cold 50 mM Tris-HCl (pH 8.0) by passage through a French pressure cell (SIM Aminco FA-003). Cell debris was removed by centrifugation, and β-galactosidase activity in the supernatant was measured by standard procedures (37).

Protein concentrations were determined by the Braford assay (Coomassie protein assay reagent; Thermo Scientific) using bovine serum albumin (albumin standard; Thermo Scientific) as the standard.

H₂O₂ measurement.

Cells were cultured in defined glucose/amino acids medium under anaerobic conditions to log phase, and then they were diluted into aerobic medium to an OD₆₀₀ of 0.005 and grown to an OD₆₀₀ of ∼ 0.15. These cells were centrifuged, suspended in fresh defined medium at an OD₆₀₀ of 0.020, and incubated aerobically at 37°C with vigorous aeration. At time points cells were removed by centrifugation, and H₂O₂ concentrations were measured in the supernatants by using the Amplex Red/horseradish peroxidase method (43) in a Shimadzu RF Mini-150 fluorometer.

Visualization of protein carbonylation.

Cells were cultured in aerobic LB as described above. Hpx⁻ yaaA cells were harvested by centrifugation at the beginning (OD₆₀₀, 0.2) and in the middle of the growth lag (OD₆₀₀, ∼0.25). Other strains were collected at an OD₆₀₀ of ∼0.2. Cell pellets were washed with ice-cold 50 mM potassium phosphate buffer, pH 7.0, and resuspended in the same buffer with 5 mM DTPA, which prevented further protein oxidation in cell extracts. After lysis using a French press, cell debris was removed by centrifugation, and protein concentrations in the cell extracts were determined using the Bradford assay. All samples were then diluted to the same protein concentration (6 mg/ml). Protein carbonyl groups were derivatized with DNPH, and the adducts were visualized by Western blotting (2).

Rapid amplification of cDNA ends (5′-RACE).

MG1655 cells and Hpx⁻ cells were cultured in LB as described above and harvested at an OD₆₀₀ of ∼0.2. RNA was extracted by the hot phenol method. Contaminating DNA was removed by DNase I digestion (Turbo DNA-free kit; Ambion), and the DNA-free RNA samples were converted to cDNA according to the manufacturer's instructions (SuperScript II reverse transcriptase; Invitrogen). The 5′ end of yaaA transcripts was determined by using a FirstChoice RLM-RACE Kit (Applied Biosystems). The PCR primers used were yaaA outer 5′-race, 5′-CTCGTTCAGCTTGTTGGTGAT-3′, and yaaA inner 5′-race, 5′-AGCCGGTGTAGACATCACCTTT-3′. Thermal cycling was performed for 35 cycles in three steps: 94°C for 30 s, 65°C for 30 s, and 72°C for 1 min.

Real-time PCR.

Cells were cultured in LB as described above. When cultures achieved an OD₆₀₀ of ∼0.2, they were treated with H₂O₂ (0, 0.25, 1, and 2 mM) for 10 min at room temperature. mRNA was harvested and converted to cDNA as described above. The following primers were used to analyze expression of yaaA and the housekeeping gene gapA: yaaA forward, 5′-GCTGTTAGACAATTCCCAGCAG-3′; yaaA reverse, 5′-AAGCCGGTGTAGACATCACCTT-3′; gapA forward, 5′-CGTTCTGGGCTACACCGAAGATGACG-3′; gapA reverse, 5′-AACCGGTTTCGTTGTCGTACCAGGA-3′. The 25-μl real-time PCR mixture contained 1 ng of cDNA, 12.5 μl of SYBR Green supermix (Bio-Rad), and 1.6 μM each primer. Thermal cycling was performed in a Mastercycler ep realplex apparatus (Eppendorf) for 40 cycles in three steps: 95°C for 20 s, 55°C for 20 s, and 68°C for 20 s.

RESULTS

The gene yaaA is induced by OxyR.

The gene yaaA is predicted to encode a 29.6-kDa conserved cytoplasmic protein that belongs to a family domain of unknown function, DUF328 (11). YaaA does not possess amino acid similarity to proteins that have defined functions, and it does not have any recognizable motifs. DNA microarray data from Zheng et al. showed that yaaA was induced in an OxyR-dependent manner when 1 mM H₂O₂ was applied (57). Our analysis of the DNA sequence upstream of yaaA revealed a potential OxyR binding site approximately 74 bp upstream of the yaaA start codon (Fig. 1). To verify these observations, we constructed a yaaA′-lacZ transcriptional fusion and measured the expression of yaaA in different strains.

Fig. 1.

Genomic context of yaaA in E. coli. In E. coli, yaaA is located at 0.12 min and is flanked by two genes with undefined functions, yaaJ and yaaX. The genomic context for yaaA is not conserved among bacteria that possess this gene, and no obvious gene partners were found by computational analyses. A computational search using an OxyR consensus sequence (58) identified one potential OxyR binding site (bp −74 to −38), and 5′-RACE analysis mapped the 5′ end of yaaA transcripts to 22 bp upstream of the yaaA start codon. TSS, transcriptional start site.

When transferred from anaerobic to aerobic conditions, the expression of yaaA was not induced in wild-type cells (E. coli MG1655). However, under the same conditions its expression was induced 10-fold in a catalase/peroxidase-deficient strain (Hpx⁻), which accumulates up to ∼1 μM endogenous H₂O₂ and in which the OxyR response is fully activated (43) (Fig. 2A). Induction did not occur if an oxyR deletion was present (Fig. 2B). 5′-RACE revealed that the 5′ end of yaaA transcripts were identical for both wild-type cells and Hpx⁻ cells, mapping to 22 bp upstream of the yaaA start codon (Fig. 1). When wild-type cells were transformed with the plasmid pGSO58, which contains a mutated oxyR allele (A233V) that constitutively expresses OxyR-regulated genes even in the absence of H₂O₂ stress (29), the expression of yaaA was induced under both anaerobic (data not shown) and aerobic conditions (Fig. 2A). These results confirmed that yaaA is part of the OxyR regulon.

Fig. 2.

A transcriptional yaaA′-lacZ fusion confirms that yaaA is part of the OxyR regulon. (A) Cultures were grown in anaerobic or aerobic LB medium for five generations before collection and β-galactosidase measurements (OD₆₀₀, ∼0.2). WT, MG1655; vector, empty vector control (pACYC184); pGSO58, plasmid-borne mutant OxyR A233V that constitutively overexpresses OxyR-regulated genes; ahp⁻ (AL154), NADH peroxidase mutant; Hpx⁻ (AL145), catalase/peroxidase-deficient mutant. (B) Anaerobic exponential-phase cultures were harvested at an OD₆₀₀ of ∼0.2 (−O₂) or were aerated for 30 min before collection (+O₂). The OD₆₀₀s of the Hpx⁻ oxyR⁺ (AL145) and Hpx⁻ oxyR mutant (AL344) strains increased from ∼0.2 to ∼0.35 during the period of aeration. In this and other figures, the error bars represent the standard errors of the means.

Interestingly, in ahp-deficient (catalase-proficient) mutants that experience minor H₂O₂ stress (44), the expression of yaaA was at the basal level during aerobic culture (Fig. 2A), even though katG, another member of the OxyR regulon, was strongly induced (data not shown). The result suggests either that the yaaA promoter region has a lesser affinity for oxidized OxyR, or that a modifying effect is imposed by another regulator that is affected by more severe H₂O₂ stress.

Hpx⁻ yaaA cells show a growth defect in LB during H₂O₂ stress.

Since yaaA was induced in Hpx⁻ strains during aerobic culture, we constructed a yaaA deletion in the catalase/peroxidase-deficient background (Hpx⁻ yaaA). The Hpx⁻ yaaA cells grew normally under anaerobic conditions, but they stopped growing about five generations after shifting to aerobic conditions, lagging for several hours before recovering (Fig. 3A).

Fig. 3.

The Hpx⁻ yaaA strain exhibits a growth defect in aerobic LB medium. Exponential-phase cultures in anaerobic LB medium were diluted into aerobic LB medium at time zero, and the optical density was monitored. (A) Wild-type cells (MG1655), black squares; ΔyaaA (AL245), gray triangles. The symbols for Hpx⁻ (catalase/peroxidase-deficient strain, LC106) and Hpx⁻ yaaA (SB53) are defined in the figure. (B) Symbols (in addition to those defined for panel A): open circles, Hpx⁻ yaaA cells with 2 mM DFO in the medium; stars, Hpx⁻ tonB yaaA (AL179).

The aerobically grown Hpx⁻ yaaA cultures filamented profusely (Fig. 4). Both anaerobic cultures and aerobic Hpx⁺ yaaA cultures appeared as normal unit-sized cells (data not shown), confirming that H₂O₂ was needed to create this phenotype. Filamentation is a hallmark of difficulty in DNA replication, and it commonly results when chemical stress damages DNA but does not substantially disrupt protein or membrane biogenesis.

Fig. 4.

Hpx⁻ yaaA cells exhibit a severe filamentous phenotype. Exponentially growing cultures in anaerobic LB medium were diluted into aerobic LB medium and grown for 5 h. Images were captured by DIC microscopy using a 40× oil immersion objective.

A chromosomal insertion of the yaaA⁺ allele at the λ phage attachment site suppressed the growth defect and the filamentation of Hpx⁻ yaaA cells (data not shown). Deletion of yaaJ, which terminates at 70 bp upstream of yaaA, in the Hpx⁻ strain, does not result in a growth defect during H₂O₂ stress (data not shown). These results indicated that the growth defect of the Hpx⁻ yaaA mutant was due to the loss of YaaA function and that YaaA is involved in the cellular response to H₂O₂ stress.

Hpx⁻ yaaA cells lose viability during H₂O₂ stress.

The viability of Hpx⁻ yaaA strains and Hpx⁻ parent strains were monitored upon aeration (Fig. 5). Samples were harvested at different time points, and CFU were determined by plating in LB top agar. After a brief lag, the number of viable cells increased steadily in the Hpx⁻ parent strain. In contrast, the viable cell count of the Hpx⁻ yaaA strain did not change during the first 2 h of aerobic culture, confirming that the increase in the OD₆₀₀ was due to filamentation rather than cell division. Approximately 3 h after aeration, concomitant with the onset of the growth lag, the viability of the Hpx⁻ yaaA mutant decreased by an order of magnitude.

Fig. 5.

The growth defect of the Hpx⁻ yaaA mutant occurred concomitantly with cell death. Exponentially growing cultures in anaerobic LB medium were diluted into aerobic LB medium at time zero. (A) Growth rates, based on the optical density. (B) Viable cell counts for Hpx⁻ (LC106) and Hpx⁻ yaaA (SB53).

Ultimately, the number of viable cells began to increase again, albeit at a slower pace than in the yaaA⁺ parent. This recovery was not due to suppressor mutations, since the isolated outgrowers reproduced the same growth pattern when the experiment was repeated. Some of the outgrowth depended upon conditioning of the medium, since a lag resumed when the outgrowing cells were centrifuged and then diluted into fresh aerobic medium. The Hpx⁻ strains can gradually degrade H₂O₂ that is found in LB medium, because the H₂O₂ is a weak substrate for the cytochrome oxidases of the respiratory chain (A. L. Woodmansee and J. A. Imlay, unpublished data); a similar recovery has been observed in Hpx⁻ strains that lack fur or dps, two other genes that are critical during H₂O₂ stress (40, 49).

The expression of yaaA was also strongly induced when Hpx⁻ strains were aerated in defined glucose medium (data not shown). However, in contrast to the results in LB broth, the Hpx⁻ yaaA mutant did not exhibit any phenotype in defined glucose medium (data not shown). The growth defect reemerged when yeast extract was added to the defined medium (see Discussion).

yaaA mutants show higher levels of DNA damage and polypeptide damage during H₂O₂ stress.

Previous work has shown that H₂O₂ directly inactivates enzymes that employ solvent-exposed iron atoms either in iron-sulfur clusters or as mononuclear cofactors (23, 45a). Further, H₂O₂ generates hydroxyl radicals when it reacts with unincorporated intracellular ferrous iron, and these radicals then covalently damage both DNA and proteins (2, 20, 40). We sought to identify the cellular injury that arrests the growth of Hpx⁻ yaaA mutants, in order to gain insight into the function of YaaA.

The filamentation and cell death of Hpx⁻ yaaA strains suggested that DNA might be damaged. Indeed, a thyA forward mutagenesis assay revealed that the mutation frequency of aerobically grown Hpx⁻ yaaA cells was 4- to 8-fold higher than that of the Hpx⁻ strain (Fig. 6). The effect depended upon oxidative stress, since the yaaA mutant in the wild-type (scavenger-proficient) background exhibited a mutation frequency similar to that of the parent strain. The yaaA mutation did not sensitize cells to UV radiation, an alternative source of DNA damage (data not shown).

Fig. 6.

Aerobic mutation frequencies are elevated in the Hpx⁻ yaaA mutant compared to its Hpx⁻ parent strain. Exponentially growing cells in anaerobic LB medium were diluted into aerobic LB medium, and the mutation frequencies were measured at subsequent time points as described in Materials and Methods. The zero time point data were obtained immediately prior to aeration. The mutation frequencies in Hpx⁺ strains (barely visible above) were about 0.5 to 1 mutants/10⁶ cells.

In a scavenger-proficient background, the yaaA mutation did create a growth defect in recA56 strains (Fig. 7A and B), which are defective in the repair of oxidative DNA lesions (1). This defect was apparent, once again, only in LB medium and only under aerobic conditions. The defect was blocked when catalase was added to the medium (Fig. 7C), which implied that the damage was due to the trace H₂O₂ that is present in aerobic LB medium. A similar result was obtained with polA1 mutants (see Fig. S1 in the supplemental material). The defect is transient, due to the decomposition of H₂O₂ by cellular enzymes. These data indicate that (i) YaaA has a role independent of the DNA recombination pathway or the SOS response, which are already absent from recA56 mutants, and (ii) even very low levels of H₂O₂ can cause debilitating DNA damage in the yaaA mutant. The data, however, do not resolve whether YaaA has a role in preventing DNA damage or in repairing it.

Fig. 7.

Hpx⁺ recA yaaA mutants exhibited growth and viability defects. Anaerobic exponential-phase cultures were diluted into aerobic LB medium at time zero. (A, C, and D) Growth rates, based on the optical density. The recA yaaA (AL223) mutants exhibited a growth (A) and viability (B) defect that was suppressed by ∼50 U/ml catalase (C) or 2 mM DFO (D).

During H₂O₂ stress, the hydroxyl radicals that are produced by the Fenton reaction can damage both DNA and polypeptides. The latter injury can be tracked by the formation of carbonyl adducts, which arise when hydroxyl radicals oxidize some amino acid side chains. An enzyme-linked immunosorbent assay-based carbonylation assay did not detect any difference in the extent of protein carbonylation in Hpx⁻ yaaA and Hpx⁻ yaaA⁺ mutants when they were grown in defined glucose medium (data not shown). However, when they were cultured in aerobic LB medium, the Hpx⁻ yaaA cells gradually accumulated higher levels of protein carbonyls than did the Hpx⁻ parent strain. These modifications were apparent at the later time point, which coincides with the growth lag (Fig. 8). While the most straightforward interpretation is that the dysfunction of YaaA directly causes the higher carbonylation, it remains possible that the carbonylation is an indirect consequence of the growth defect.

Fig. 8.

Protein carbonylation in Hpx⁻ yaaA cells. Cells were cultured in aerobic LB medium as described in Materials and Methods. The Hpx⁻ yaaA cells were harvested at the beginning (OD₆₀₀, 0.2; 3 h) and in the middle of the growth lag (6 h). Other strains were collected at an OD₆₀₀ of ∼0.2. Strains used were as follows: wild type (WT; MG1655); Hpx⁻ dps (SP66), a positive control for carbonylation; Hpx⁻ (LC106); Hpx⁻ yaaA (SB53).

We also evaluated whether YaaA has a role in defending mononuclear iron-containing enzymes or solvent-exposed Fe-S clusters against H₂O₂ stress. The activities of ribulose-5-phosphate epimerase and isopropylmalate isomerase, respectively, were compared in Hpx⁻ and Hpx⁻ yaaA mutants. Both enzymes exhibited a reduction in activities relative to scavenger-proficient strains, but the yaaA allele had no further impact (data not shown).

The Fenton reaction causes injuries in yaaA mutants during H₂O₂ stress.

The preceding results suggested that yaaA might suppress cellular injuries that arise from the Fenton reaction. Consistent with this idea, the growth defect of Hpx⁻ yaaA mutants was suppressed by maneuvers that diminished the amount of intracellular iron. The exogenous addition of DFO, a cell-permeable iron chelator that blocks Fenton chemistry (20), eliminated the growth lag (Fig. 3B). A similar effect was obtained by a tonB mutation, which prevents iron import by the siderophore and ferric-citrate systems (51) (Fig. 3B). Furthermore, DFO also eliminated the growth defect of recA yaaA mutants (Fig. 7D). These results confirm that the key injury in the yaaA mutant during H₂O₂ stress is due to the Fenton reaction.

YaaA protects cells by attenuating the Fenton reaction.

In principle, YaaA might protect cells either by attenuating the Fenton reaction and reducing HO^• formation or by helping cells to tolerate the damage that it creates. In order to distinguish between these possibilities, we employed EPR to evaluate the rates of HO^• formation.

Intracellular HO^• was trapped by ethanol/POBN and detected by room temperature EPR. Wild-type cells exhibited a vanishingly small signal (data not shown), whereas the level of hydroxyl radical formation was substantial in Hpx⁻ mutants, as reported previously (40). In the early period after aeration, the Hpx⁻ yaaA cells exhibited a signal similar to that of Hpx⁻ cells (data not shown). However, the signal continually increased in the Hpx⁻ yaaA cells, ultimately becoming about twice that of the Hpx⁻ strain (Fig. 9). Pretreatment with 2 mM dipyridyl (an iron chelator) for 5 min eradicated the EPR signal (data not shown), confirming that it represented hydroxyl radicals produced by the Fenton reaction. These results indicated that YaaA protects cells by attenuating the Fenton reaction.

Fig. 9.

The Hpx⁻ yaaA mutant exhibited higher levels of hydroxyl radicals. Cells were cultured in aerobic LB as described earlier in this report, and they were harvested at an OD₆₀₀ of 0.2, which corresponded to the beginning of the growth lag of Hpx⁻ yaaA cells. The final cell densities were equivalent in all samples. The six peaks represented the HO^• signal of Hpx⁻ cells and Hpx⁻ yaaA cells.

YaaA does not affect H₂O₂ levels.

Continuous hydroxyl radical formation results when the cellular pool of unincorporated iron cycles through two consecutive reactions, the Fenton reaction (termed reaction 1 here; Fe²⁺ + H₂O₂ → Fe³⁺ + OH⁻ + HO^•) followed by the regeneration of ferrous iron mediated by cellular reductants (reaction 2; Fe³⁺ + red → Fe²⁺ + ox).

Both intracellular cysteine and reduced flavins have been shown to serve as the predominant reductants in reaction 2, depending upon the growth conditions (39, 53). To minimize oxidative injuries, cells must control the levels of H₂O₂, unincorporated iron, and cysteine/flavin. The spin-trapping data suggested that YaaA suppresses the level of one of these reactants.

The rate of intracellular H₂O₂ formation depends upon the titers of autoxidizable enzymes in the cell, and in Hpx⁻ strains the steady-state level of H₂O₂ also depends upon minor scavenging mechanisms. We sought to test whether YaaA might affect either of these processes. In Hpx⁻ strains the H₂O₂ concentration rapidly equilibrates between the cytoplasm and the external medium; therefore, cocultured Hpx⁻ strains experience equivalent intracellular concentrations of H₂O₂, and a strain that efficiently scavenges H₂O₂ can suppress the growth defects that would otherwise arise in a cocultured strain that cannot (44). To this end, we cocultured Hpx⁻ cells and Hpx⁻ yaaA cells together in LB medium under aerobic conditions and then analyzed the viability at designated time intervals for the two individual strains, which were distinguished upon plating by using drug resistance markers. In mixed culture the Hpx⁻ yaaA strain continued to exhibit its full aerobic viability defect, while the Hpx⁻ strain did not. The result indicates that YaaA does not alter the intracellular levels of H₂O₂ (data not shown).

The rate of cellular H₂O₂ formation was also directly measured as described in Materials and Methods. These experiments were performed in defined media, as components in LB interfere with H₂O₂ measurements. Under these conditions, the rate of H₂O₂ accumulation in Hpx⁻ cells was 11 ± 1 nM/min (mean ± standard error of the mean), versus 11 ± 0.3 nM/min in Hpx⁻ yaaA cells, which further indicates that YaaA does not affect the rate of H₂O₂ accumulation.

YaaA does not inhibit ferric iron reduction.

FADH₂ and cysteine can each reduce ferric iron and drive the Fenton reaction forward in vivo. Reduced free flavin concentrations are elevated by respiratory blocks, through the action of an NADH-dependent flavin reductase (Fre), and under these conditions H₂O₂ can create hydroxyl radicals and DNA lesions at a very high rate (53). However, in defined medium neither deletion nor induction of the yaaA allele affected the rate at which cyanide/H₂O₂ killed cells (data not shown). The same was true when cells were shifted from sulfate to cystine medium (data not shown), a protocol that elevates intracellular cysteine levels and similarly amplifies H₂O₂ sensitivity (39). These results suggest that YaaA is unlikely to attenuate the Fenton reaction by inhibiting the rate of ferric iron reduction (reaction 2).

Hpx⁻ yaaA mutants have higher unincorporated iron levels than the Hpx⁻ parent strain.

The rate of oxidative DNA damage is directly determined by the amount of unincorporated iron inside the cell (26, 46). To learn whether YaaA controls iron levels, we measured intracellular unincorporated iron by whole-cell EPR. Samples were cultured aerobically in LB and collected at an OD₆₀₀ of 0.1, 0.2, and 0.3; these time points represented Hpx⁻ yaaA cells prior to the growth lag, at its onset, and at its midpoint. The data showed that deletion of yaaA in unstressed Hpx⁺ cells had no effect on the iron levels of approximately 65 μM (Fig. 10A). The level of unincorporated iron in H₂O₂-stressed Hpx⁻ strains was substantially higher (150 μM), an effect that may reflect the oxidative degradation of iron-sulfur clusters. The key observation, however, was that the Hpx⁻ yaaA cells consistently accumulated 2-fold-higher levels of unincorporated iron (300 μM) by the beginning (OD₆₀₀, 0.2) of the growth lag (Fig. 10A). This effect matches the 2-fold effect of yaaA upon hydroxyl radical formation. The iron returned to Hpx⁻ yaaA⁺ levels when yaaA was expressed in trans on a plasmid (Fig. 10B).

Fig. 10.

Levels of unincorporated iron are elevated in Hpx⁻ yaaA cells. The strains were cultured in aerobic LB medium, and intracellular unincorporated iron was determined by whole-cell EPR analysis. Data are reported as the means and standard errors of the means from at least three independent experiments. (A) Cells were harvested at an OD₆₀₀ of 0.1, 0.2, and 0.3. (B) Cells were harvested at an OD₆₀₀ of 0.2. pWKS30, empty vector; pWKyaaA, plasmid expressing yaaA under control of the lac promoter.

Hpx⁻ yaaA cells do not show a growth defect compared to Hpx⁻ parent strains in defined medium during aerobic culture. Consistent with the growth curve results, the EPR data showed that the level of intracellular free iron in Hpx⁻ yaaA cells was similar to that of Hpx⁻ cells in defined medium, approximately 100 μM for both strains at an OD₆₀₀ of 0.2 (data not shown).

These results indicate that YaaA suppresses hydroxyl radical formation, and thus the rate of oxidative DNA damage, by diminishing the amount of unincorporated iron inside oxidatively stressed cells.

YaaA does not repress expression of iron importers.

The factors that control iron import, trafficking, and disposition within the cell are largely mysterious. The best-understood facet of iron regulation is the transcriptional control upon the synthesis of iron importers and of iron-containing enzymes (6, 34). The primary iron importers active in lab cultures of E. coli are the ferric enterobactin uptake (Ent) system, the ferric citrate uptake (Fec) system, and the ATP-dependent ferrous import (Feo) system. We observed that Hpx⁻ fec yaaA, Hpx⁻ entA yaaA, and Hpx⁻ feo entA yaaA mutants all exhibited aerobic growth defects compared to their yaaA⁺ counterpart strains (data not shown), which implied that the yaaA phenotype did not depend on a specific iron importer. However, all three systems are controlled at the transcriptional level by the Fur repressor, and it seemed plausible that YaaA might perturb the function of this regulator. More generally, in order to evaluate whether YaaA controls iron levels by controlling transcription of the importer(s), we constructed Δfeo ΔentA derivatives, which rely upon the Fec system as the sole mechanism of iron uptake. Using a fecA′-lacZ transcriptional fusion integrated at the lambda attachment site, we found that mutation of the yaaA allele had no impact upon fecA transcription (Fig. 11). Thus, YaaA does not control the expression of iron importers. Other possible points of iron control exist, such as an influence on transporter activity.

Fig. 11.

fecA′-lacZ induction in Hpx⁻ feo entA strains (yaaA⁺ versus yaaA⁻). Cultures were grown in anaerobic LB medium for five generations and either harvested at an OD₆₀₀ of ∼0.2 (-O₂) or diluted into aerobic LB medium and grown aerobically for five more generations before collection at an OD₆₀₀ of ∼0.2 (+O₂). The yaaA⁺ strain is Hpx⁻ feo entA fecA′-lacZ (AL273); the yaaA⁻ strain is Hpx⁻ feo entA yaaA fecA′-lacZ (AL293).

Hpx⁻ mntH yaaA mutants showed a strong growth defect in defined medium.

Our previous data suggested that YaaA helped to control the iron level during peroxide stress; hence, we wanted to check whether YaaA worked with other proteins that also control iron homeostasis during the OxyR response, including the iron uptake regulator (Fur), iron-scavenging protein (Dps), and manganese transporter (MntH). The yaaA deletion exacerbated the growth and viability defects of Hpx⁻ fur and Hpx⁻ dps strains in LB (see Fig. S2 in the supplemental material), but not in defined medium (data not shown). However, Hpx⁻ mntH yaaA mutants showed a strong growth defect in both LB and defined glucose medium (Fig. 12). During H₂O₂ stress, manganese is imported via MntH to replace the iron cofactor in certain peroxide-sensitive enzymes (2, 45a). In defined medium, Hpx⁻ mntH yaaA cells stopped growing almost immediately after aeration (Fig. 12B), but the number of viable cells remained constant (Fig. 12C). These results suggested that this particular phenotype might be caused by inhibition of metabolism instead of DNA damage. Collectively, these data show that YaaA has functions independent of Fur, Dps, and MntH.

Fig. 12.

Hpx⁻ mntH yaaA cells exhibited growth defects in both LB and defined glucose medium. Anaerobic exponential-phase cultures were diluted into the same aerobic medium at time zero. Strains used included the following: Hpx⁻ (LC106), Hpx⁻ mntH (AA30), Hpx⁻ yaaA (SB53), and Hpx⁻ mntH yaaA (AL221). (A) Optical density in LB. (B and C) Optical density (B) and cell viability (C) in defined glucose medium.

Additional experiments revealed that Hpx⁻ ftn and Hpx⁻ bfr mutants grow as well as Hpx⁻ strains, indicating that under our conditions ferritin and bacterioferritin do not play a key role during H₂O₂ stress and are not involved in the critical function of YaaA (data not shown). Similarly, although FieF is suspected as acting as an iron efflux system (15), an fieF deletion did not diminish the growth of an Hpx⁻ strain, and the further addition of a yaaA mutation recapitulated the Hpx⁻ yaaA phenotype (see Fig. S3 in the supplemental material).

Regulation of YaaA.

To obtain further insight into the function of YaaA, we measured the expression of the yaaA′-lacZ fusion under different conditions (Table 2). Although previous DNA microarray data showed that yaaA was induced 4-fold by treatment with 1 mM H₂O₂ in an ΔoxyR strain (57), we did not observe this induction in either a β-galactosidase assay or by real-time PCR measurements (data not shown). Our data suggest that OxyR is the sole transcriptional regulator of yaaA in response to oxidative stress.

Table 2.

Regulation of yaaA

Test condition	Mutants (medium used)	Regulator	Resulta
H2O2 stress	Hpx− (LB and defined medium)	OxyR	Yes
Redox-cycling drug	WT (LB + 200 μM paraquat)	SoxR	No
Iron rich	fur (LB)	Fur	No
Iron starved	tonB feo (defined medium + DFO)		No
Mn rich	mntR (LB)		No
Mn starved	mntH (defined medium)		No
Small RNA	hfq (LB)	Small RNAs	No
Stationary phase	WT (LB and defined medium)	RpoS	No
iscR deficient	iscR (LB)	IscR	No
Autoregulation	Hpx−yaaA (LB)	YaaA	No

^aA negative result was defined as a less-than-2-fold change in expression of yaaA.

The expression of yaaA was unaltered (<2-fold induction/repression) under other tested conditions. It did not respond to manganese availability; it was not induced by redox-cycling drugs, which indicates that it is not regulated by SoxR (16, 47); it was not affected by Fur, IscR, Hfq, RpoS, or YaaA itself.

DISCUSSION

In some natural environments bacteria must cope with extended periods of exposure to low levels of H₂O₂, an experience that was recreated in this study through the use of strains that lack scavenging enzymes. The H₂O₂ concentration that accumulates in the cytoplasm of aerobic Hpx⁻ mutants, approximately 1 μM, is expected to be reached in the cytoplasm of wild-type cells when they enter environments that contain approximately 5 μM H₂O₂ (44). The primary effect of these doses upon E. coli is to block growth. The particular damage mechanisms that have been identified thus far all appear to involve Fenton-type reactions between iron and H₂O₂: the creation of DNA lesions and the inactivation of mononuclear iron enzymes, of iron-sulfur dehydratases, of the Isc Fe/S assembly system, and of the Fur repressor (2, 20, 23, 24, 49). The second-order rate constants for these reactions are high, ca. 10³ to 10⁴ M⁻¹ s⁻¹, which means that when H₂O₂ is present at 1 μM they occur with half-times on a minutes time scale. No other target molecules are known that react with H₂O₂ so readily, and so it seems fitting that many of the genes controlled by OxyR serve to avoid or to alleviate the damage done by Fenton reactions. YaaA now joins Dps (40, 54) and Fur (49) as a mediator that minimizes the amount of unincorporated iron inside H₂O₂-stressed cells. Like Dps and Fur, YaaA has the effect of minimizing damage both to DNA and to proteins. Like Dps, but unlike Fur, YaaA seems not to play much of a role during routine growth; it is only during H₂O₂ stress, when YaaA levels rise, that its influence is apparent. For this reason, yaaA mutants resemble dps mutants in that they do not exhibit unusual sensitivity in experiments that involve a sudden bolus of H₂O₂; a period of OxyR regulon expression is required for it to become functional. Indeed, mntH and suf mutants only evince a phenotype during protracted, low-grade stress, too (2, 24).

Little is known about the DUF328 family of proteins. They do not have recognizable motifs or absolutely conserved residues, and no crystal structures are available. YaaA homologues are found in only a few eukaryotes, all of which are photosynthetic (four green algae and Ricinus communis). YaaA genes are widely distributed among bacteria, including both anaerobes and aerobes. Their presence in anaerobes does not necessarily indicate that they have a nonoxidative stress role, since even anaerobes are periodically exposed to oxygenated waters and customarily encode antioxidant enzymes, including scavengers of H₂O₂. A yaaA gene is identifiable in Lactobacillus gasseri, which uses H₂O₂ as a chemical weapon to inhibit competitors (45b), but we did not find any homologue in Borrelia burgdorferi, a bacterium that seems not to have iron-dependent enzymes or to import iron (42). Thus, the circumstantial data are broadly consistent with a role in iron metabolism and antioxidant activity.

The yaaA phenotype was pronounced in medium that included yeast extract. EPR measurements showed that E. coli generally contains higher levels of unincorporated iron when yeast extract is in the medium. This effect cannot be achieved simply by adding high levels of iron to defined medium, which leads us to suspect that a component of yeast extract makes iron more available for import. Interestingly, yaaA was not as easily induced by H₂O₂ as was katG. We do not know whether the mechanistic basis of this difference stems from a difference in OxyR affinity, but the end effect resembles the graded responses of members of the SOS regulon, some of which are turned on rapidly and others of which are activated later (41). Thus, catalase synthesis might comprise the first response to slight H₂O₂ stress, whereas alterations in iron metabolism are engaged only if stress is severe.

So how does YaaA control iron levels? This question is challenging, because the details of iron metabolism remain foggy. In principle, intracellular iron levels might be suppressed in a few ways: by slowing iron import, speeding delivery to iron enzymes, sequestering iron in storage proteins, preventing the leakage of iron from labile enzymes, or pumping excess iron from the cell (Fig. 13). However, YaaA is not an integral membrane protein, meaning that it could at best be an accessory to efflux: it was effective no matter which iron import system was active, it did not seem to affect the damage to Fe/S dehydratases, and it retained a phenotype whether or not cells contained Dps, the primary iron storage system during oxidative stress. Although one can contrive arguments around these facts, they collectively hint that perhaps the remaining option—that YaaA is somehow involved in iron trafficking—is the most likely one. Pull-down experiments did not identify any convincing partner proteins (data not shown). As our understanding of iron metabolism progresses, perhaps a specific hypothesis for the YaaA mechanism will become apparent, as will experimental methods that would test it.

Fig. 13.

Potential mechanisms by which YaaA might diminish levels of unincorporated iron. YaaA might slow down the increase of unincorporated iron by slowing iron import (1) or preventing the leakage of iron from H₂O₂-damaged enzymes (2), or YaaA might accelerate the decrease of unincorporated iron by sequestering iron (3), speeding iron delivery to target proteins (4), or increasing iron efflux (5).