Fusidic Acid-Resistant Mutants of Salmonella enterica Serovar Typhimurium Have Low Levels of Heme and a Reduced Rate of Respiration and Are Sensitive to Oxidative Stress

Abstract

Mutations in the translation elongation factor G (EF-G) make Salmonella enterica serovar Typhimurium resistant to the antibiotic fusidic acid. Fus^r mutants are hypersensitive to oxidative stress and rapidly lose viability in the presence of hydrogen peroxide. We show that this phenotype is associated with reduced activity of two catalase enzymes, HPI (a bifunctional catalase-hydroperoxidase) and HPII (a monofunctional catalase). These catalases require the iron-binding cofactor heme for their activity. Fus^r mutants have a reduced rate of transcription of hemA, a gene whose product catalyzes the first committed step in heme biosynthesis. Hypersensitivity of Fus^r mutants to hydrogen peroxide is abolished by the presence of δ-aminolevulinic acid, the precursor of heme synthesis, in the growth media and by the addition of glutamate or glutamine, amino acids required for the first step in heme biosynthesis. Fluorescence measurements show that the level of heme in a Fus^r mutant is significantly lower than it is in the wild type. Heme is also an essential cofactor of cytochromes in the electron transport chain of respiration. We found that the rate of respiration is reduced significantly in Fus^r mutants. Sequestration of divalent iron in the growth media decreases the sensitivity of Fus^r mutants to oxidative stress. Taken together, these results suggest that Fus^r mutants are hypersensitive to oxidative stress because their low levels of heme reduce both catalase activity and respiration capacity. The sensitivity of Fus^r mutants to oxidative stress could be associated with loss of viability due to iron-mediated DNA damage in the presence of hydrogen peroxide. We argue that understanding the specific nature of antibiotic resistance fitness costs in different environments may be a generally useful approach in identifying physiological processes that could serve as novel targets for antimicrobial agents.

Translation elongation factor G (EF-G) is a GTP-binding protein that catalyzes the translocation of peptidyl-tRNA on the ribosome from the A site to the P site (30). After GTP hydrolysis and translocation, EF-G · GDP leaves the ribosome and is regenerated by the spontaneous exchange of GDP for GTP off the ribosome. Fusidic acid is an antibiotic that binds to a complex of EF-G and the ribosome and blocks the release of EF-G · GDP from the ribosome (5, 22). This blockage inhibits further protein synthesis.

Resistance to fusidic acid in Salmonella enterica serovar Typhimurium is caused by mutations in fusA, the gene encoding EF-G (21). Similar mutations in fusA are a common cause of fusidic acid resistance in clinical isolates of fusidic acid-resistant Staphylococcus aureus (27). Fus^r mutant forms of EF-G, including that encoded by fusA1 studied here, have a reduced rate of nucleotide exchange, and this causes a reduced rate of protein synthesis and cell growth (26). Fus^r mutants also have reduced levels of the transcription regulator molecule ppGpp (26). Fus^r mutant EF-G apparently interferes with the RelA-dependent synthesis of ppGpp in the ribosomal A site (26). In addition, Fus^r mutants have reduced levels of the stress-related sigma factor RpoS (25). The reduced levels of RpoS may also be a consequence of the reduced levels of ppGpp (15, 23) in Fus^r mutants. Both ppGpp and RpoS are global transcriptional regulators of gene expression, and consequently, Fus^r mutants have phenotypes in addition to translation-rate-related slow growth. These phenotypes include large cell size at division (26) and sensitivity to oxidative stress in vivo (25).

Here we identify reduced production of heme as an important contributing factor to the hypersensitivity of Fus^r mutants to oxidative stress. We also show that this has an additional consequence of reducing the rate of respiration in Fus^r mutants.

MATERIALS AND METHODS

Bacterial strains and media.

The bacterial strains used in this study are listed in Table 1. All strains are derivatives of serovar Typhimurium LT2 (designated as the wild type). This LT2 strain (TT10000) is originally from the lab of John Roth. Transductions were made using P22 HT. Strains were grown at 37°C in Luria-Bertani broth (LB) or, where indicated, in minimal M9 medium or nutrient broth (NB) medium (8). The solid medium was Luria agar (LB supplemented with 1.5% agar; Oxoid, Basingstoke, England).

TABLE 1.

Strains used and their genotypes

Straina	Genotype
MM14	Wild-type S. enterica serovar Typhimurium LT2
MM142	fusA1 zhb-736::Tn10
JB2110	fusA1-15 zhb-736::Tn10
MM122	hemA::lacZ (Kanr)
MM123	fusA1 zhb-736::Tn10 hemA::lacZ (Kanr)
MM124	fusA1-15::zhb-736::Tn10 hemA::lacZ (Kanr)
MM125	relA21::Tn10 hemA::lacZ (Kanr)
MM126	relA1 spoT::Tn10 hemA::lacZ (Kanr)
MM46	katG::pRR10 (Ampr)
MM74	fusA1-15 zhb-736::Tn10 katG::pRR10 (Ampr)
MM106	katE::Tn10
MM107	fusA1-15 katE::Tn10

^aAll strains are derived from Salmonella serovar Typhimurium LT2.

Antibiotics.

Fusidic acid sodium salt was a gift from Leo Pharma, Ballerup, Denmark, and had a final concentration of 800 μg/ml when used in the presence of 1 mM EDTA in the solid medium. Tetracycline (Sigma-Aldrich, Stockholm, Sweden) was used in media at a final concentration of 15 μg/ml.

Hydrogen peroxide sensitivity assays.

Bacterial cultures grown in LB to optical densities at 600 nm of 0.2 to 0.4 were treated with 2.5 M hydrogen peroxide (Merck, VWR International, Stockholm, Sweden). To determine viability, aliquots were taken at indicated time points (15, 30, and 45 min), diluted in 0.9% NaCl, and plated onto LB and LB-tetracycline plates. When indicated, cells were exposed to 2,2′-dipyridil (Sigma-Aldrich) for 15 min prior to treatment with hydrogen peroxide. For δ-aminolevulinic acid (ALA; Sigma Aldrich) supplementation experiments, cells were first grown overnight at 37°C in NB medium with and without 50 μg/ml ALA and diluted 100-fold in fresh medium of the same composition. Cells were then grown in NB medium to OD₆₀₀s of 0.2 to 0.4 with and without 50 μg of ALA/ml (8) before being exposed to 2.5 mM hydrogen peroxide. Viability was assayed as described above.

Catalase activity.

Relative catalase activity was quantified by a colorimetric assay based on the use of dicarboxidine ((γγ′-4,-4′-diamino-3,3′-biphenylylenedioxy) dibutyric acid; Sigma-Aldrich), a sensitive substrate for the detection of peroxidase activity (29). Dicarboxidine is converted into a colored product in a reaction catalyzed by the activity of lactoperoxidase (EC 1.11.1.7) (170 U/mg of protein; Fluka/Sigma-Aldrich); the amount of color developed is directly proportional to the amount of H₂O₂ present in the medium. Immediately before each experiment, a solution of 50 μg of lactoperoxidase/ml was mixed with an equal volume of 1 mM dicarboxidine. Relative catalase activity in bacterial cultures was assayed according to a previously established method (36) with minor modifications as described here. Cultures were grown overnight in M9 glucose minimal medium, washed, and then resuspended in prewarmed 37°C M9 medium with 0.2% glycerol as a carbon source (we found that the presence of glucose in the medium interfered with the sensitivity of the reaction). The concentration of cells at the start of the assay was 2 × 10⁸ cells/ml. To begin the reaction, H₂O₂ was added to the bacterial culture to a final concentration of 100 μM. The culture was incubated at 37°C with agitation, and 1-ml samples were taken immediately after the addition of H₂O₂ and at regular time intervals up to 45 min. Each sample was immediately added to a 200-μl aliquot of the dicarboxidine-lactoperoxidase mixture at room temperature. Within 1 min a stable color developed. The absorbance of the samples was measured at 450 nm. The decrease in the amount of color developed as a function of time is proportional to the amount of H₂O₂ degraded by catalase and is thus inversely proportional to catalase activity.

β-Galactosidase assays.

Cultures were grown overnight in LB at 37°C and diluted 100-fold in fresh LB medium. Samples were collected from exponentially growing cells (OD₆₀₀, 0.3) and analyzed as described previously (25). OD₄₂₀ and OD₅₄₀ were measured for 16 h at intervals of 5 min in a Bioscreen C machine (Labsystems, Helsinki, Finland). β-Galactosidase activity units were calculated from the linear part of the OD₄₂₀ curve for all the samples analyzed, using the formula OD₄₂₀ − [1.75 × (OD₅₄₀/OD₆₆₀)] × time (minutes) × volume of the sample (milliliters).

Competitions in M9 minimal medium supplemented with hydrogen peroxide.

A total of 10⁶ cells of the wild type and the mutant strain per ml were inoculated into M9 minimal medium containing 0.2% glucose supplemented with 70 μM H₂O₂. For amino acid complementation assays, 1% Casamino Acids, 0.2% glutamine, or 0.2% glutamate (all from Sigma-Aldrich) were added to the minimal medium with 70 μM H₂O₂. Cultures were incubated at 37°C for 18 h. Counts of viable wild-type and mutant cells were made by taking aliquots, diluting them, and plating them onto LB and LB-tetracycline plates. The competition index is the ratio of mutant/wild-type cells measured in CFU at the end of one cycle of growth competition, where the ratio at time zero is normalized to 1:1.

Heme measurements.

The relative amount of heme in strain MM14 (wild-type strain LT2) versus that in MM142 (fusA1) was measured by a fluorescence heme assay (37). Cultures of the wild type and the fusA1 mutant were grown overnight with shaking in LB, washed, and resuspended in Dulbecco phosphate-buffered saline. OD₆₀₀s of the cultures were carefully measured, and appropriate dilutions were made so that different 50-μl aliquots contained numbers of cells corresponding to 0.05 to 1.0 OD₆₀₀ unit (1 OD₆₀₀ unit is defined as the number of cells that give an OD₆₀₀ of 1 in 1 ml). We determined that measurements by OD₆₀₀ units and dry weight (biomass) gave equivalent results for these strains. Each 50-μl aliquot of cells was mixed with 500 μl of heated 2 M oxalic acid (Sigma-Aldrich) and incubated in an oven at 100°C for 30 min. This treatment strips iron from heme, and the resultant protoporphyrin can be measured by fluorescence. A 2 M concentration of oxalic acid was prepared by dissolving 10.08 g of oxalic acid with heating and stirring in 30 ml of water and then bringing the final volume to 40 ml. After 30 min at 100°C, 500 μl of Dulbecco phosphate-buffered saline (or water) was added to the cell-oxalic acid mixture, and the new mixture was allowed to cool to room temperature. Fluorescence was measured on an SPF-500 corrected-spectrum spectrofluorometer (Aminco-Bowman, Silver Spring, Md.), using Hellma semimicrofluorescence cuvettes (Sigma-Aldrich). Samples were excited at 400 nm, and emission was scanned from 200 to 800 nm. Two emission peaks for porphyrin fluorescence were observed, at 608 and 662 nm. We also made, in parallel, fluorescence measurements with crude membrane preparations prepared by French pressing and ultracentrifugation and found the results to be indistinguishable from those made with whole cells treated with oxalic acid (data not shown). We found the oxalic acid method to be simpler, more rapid, and better suited for assaying multiple samples.

Respiration assay.

Cultures were grown overnight at 37°C in LB supplemented with 0.2% glucose. Respiration assays were carried out in a 10-liter (working volume) stirred-tank bioreactor equipped with Wonderware 2000-based software (Belach Bioteknik, Stockholm, Sweden). The bioreactor was filled with 10 liters of LB supplemented with 2% glucose and 3 ml of polypropylene glycol (antifoaming agent; BDH, Poole, United Kingdom). The temperature and pH were adjusted to 37°C and 7, respectively. The bioreactor was inoculated with an appropriate amount of culture to achieve an OD₆₀₀ of 0.08. The culture used had been grown overnight in a shaking flask of LB supplemented with 0.25% glucose. During the respiration experiment, the OD₆₀₀ was recorded every 20 min, while dissolved oxygen, carbon dioxide content, and pH were registered continuously. The respiration (oxygen consumption) of the cell population was expressed as the specific respiration rate (q) during one doubling time of the population (the time for the culture's OD₆₀₀ to increase from 0.2 to 0.4). q is measured in micromoles of O₂ per liter per min per OD₆₀₀. The overall respiration rate of the population, qX (where X is the cell concentration), is in equilibrium with the oxygen transfer rate (OTR) of the bioreactor, i.e., OTR = qX. The bioreactor OTR was calculated from the formula OTR = kLa(c* − c), where kLa is the volumetric mass transfer coefficient, c* is the saturated oxygen concentration, and c is the measured oxygen concentration at a certain time point. Both kLA and c* were determined prior to the experiment by the sulfite oxidation method (2) and by use of the same value parameters (volume, temperature, aeration, stirrer speed, and pressure) applied to the actual culture. During the experiment, the dissolved oxygen concentration was registered continuously and read when the cell concentration (X) corresponded to an OD₆₀₀ of 0.4. The specific respiration rate was determined by calculating OTR/X.

RESULTS

Fus^r mutants reduce HPI and HPII activity.

We have previously shown that Fus^r mutants have reduced catalase activity relative to the wild type (25). The relative catalase activity levels associated with the wild type and the strains with the mutations fusA1-15 and fusA1 are 1.00, 0.45, and 0.35, respectively. We asked whether the reduced catalase activity in Fus^r mutant JB2110 (fusA1-15) is associated with the HPI or HPII catalases. Isogenic wild-type and fusA1-15 strains (Table 1) were constructed carrying functional knockouts (katG::Tn10 and katE::Tn10) affecting HPI and HPII, respectively. Catalase activity was measured for each strain as the rate of clearance of H₂O₂ from the growth medium (see Materials and Methods). In strains carrying either one of the catalase genes intact, the presence of the fusA1-15 mutation reduced the catalase activity to less than half the value found in the equivalent fusA⁺ strain (Table 2). This showed that the fusA1-15 mutation reduced both HPI and HPII catalase activities. We hypothesized that a possible cause of the reduced catalase activity might be the molecule heme, which is a cofactor for both HPI and HPII (3).

TABLE 2.

Impact of fusA1-15 genotype on catalase activity in Serovar Typhimurium strains

Strain	Genotype	Active catalase(s)	Catalase activity (%)a
MM14	Wild type	HPI + HPII	100
JB2110	fusA1-15	HPI + HPII	45
MM106	katE::Tn10	HPI	43
MM107	katE::Tn10 fusA1-15	HPI	12
MM46	katG::pRR10	HPII	73
MM74	katG::pRR fusA1-15	HPII	31

^aResults are an average of at least 10 independent assays with a standard error of 5%.

Fus^r mutants have reduced expression of hemA.

We wanted to determine whether Fus^r mutants were defective in the production of heme. The first committed step in the biosynthesis of heme is the synthesis of the heme precursor ALA. In Salmonella, glutamyl-tRNA reductase, encoded by hemA, converts a small fraction of the cells' charged glutamyl-tRNA^Glu to glutamate-semialdehyde. This in turn is converted by glutamate-1-semialdehyde aminotransferase, encoded by hemL, to ALA. We measured expression of hemA in the wild type and in two Fus^r mutants (fusA1 and fusA1-15) by use of a hemA-lacZ operon fusion from Tom Elliott (8). Each Fus^r mutant is associated with a decreased level of transcription of the fusion relative to that of the wild type. In strain MM123, the strain with fusA1, the level of hemA-lacZ transcription is only 64% of the wild-type level, while in MM124 (fusA1-15), it is 85% (Table 3). We conclude that Fus^r mutants have reduced transcription of hemA.

TABLE 3.

Gene expression from a hemA-lacZ operon fusion in wild-type, Fus^r, and relA/spoT mutants

Strain	Genotypea	β-Galactosidase activityb	SE
MM14	Wild type	98.0	1.1
MM124	fusA1-15	83.6	1.1
MM123	fusA1	62.5	2.6
MM125	relA21::Tn10	87.4	0.8
MM126	relA1 spoT::Tn10	83.9	1.4

^aRelevant genotype; full genotypes are listed in Table 1.

^bβ-Galactosidase activities are given in Miller units (10³) and are the average of results of at least four independent assays.

To determine whether the decreased transcription of hemA was due to low levels of ppGpp in the Fus^r mutants (26), we measured the expression of the hemA-lacZ fusion in mutants with low ppGpp levels due to mutations in relA (31) and spoT. The relA mutation causes a small reduction in hemA-lacZ expression (Table 3). However, while the fusA1 and relA21::Tn10 strains have similar basal levels of ppGpp, 5 and 7 pmol/OD₆₄₀ unit, respectively (26), they have very different levels of hemA-lacZ expression, i.e., 64and 89% of wild-type levels, respectively. Also, whereas the fusA1-15 and relA1 spoT::Tn10 strains have very different basal levels of ppGpp, 20 and <1 pmol/OD₆₄₀ unit, respectively (25, 26), they have similar levels of hemA-lacZ expression, i.e., 85 and 86% of wild-type levels, respectively. We conclude that the decreased level of transcription of hemA in Fus^r mutants is not fully explained by their decreased levels of ppGpp.

ALA reverses Fus^r sensitivity to oxidative stress.

We wanted to determine whether the increased sensitivity of Fus^r mutants to oxidative stress caused by exposure to hydrogen peroxide (25) could be attributed to a reduced flow through the heme biosynthesis pathway. A recent study has shown that a Salmonella hemA mutant is very susceptible to hydrogen peroxide (11). We measured the viability of the wild-type strain MM14 and the fusA1-15 strain JB2110 after exposure to hydrogen peroxide when grown with or without the precursor of heme synthesis, ALA. The wild-type bacteria are relatively insensitive to 30 min of exposure to hydrogen peroxide, and, as expected, the presence of ALA did not significantly affect their survival (Fig. 1). In contrast, the fusA1-15 strain (JB2110) was sensitive to this level of hydrogen peroxide, with less than 20% surviving 15 min of exposure. However, the fusA1-15 strain grown with ALA survived as well as the wild type (Fig. 1). The fusA1 strain MM142 was as sensitive in this assay as the strain carrying fusA1-15 (data not shown), and its sensitivity was also reversed by growth in the presence of ALA. The effect of ALA in reversing sensitivity to hydrogen peroxide supports the hypothesis that this sensitivity is due to a low level of heme in the Fus^r mutant.

FIG. 1.

Effect of H₂O₂ on bacterial viability of the wild type and a Fus^r mutant as a function of the presence or absence of ALA, a precursor in the heme biosynthesis pathway. The standard error of these measurements is ±10% for the wild type and ±4% for the fusA1-15 mutant.

Media dependence of Fus^r sensitivity to hydrogen peroxide.

In the ALA assay described above, cells were grown in a rich LB medium. We noted that in LB, the extreme sensitivity to H₂O₂ of strains carrying fusA1, previously noted in M9 medium (25), is partly ameliorated. We asked which constituent of LB medium was responsible for this reversal of sensitivity. We made these assays as pairwise competitions between the wild type and Fus^r mutants because this is a very sensitive way to measure relative improvements in survival or growth rate. Note that no ALA was added in these assays. We found that the addition of Casamino Acids to M9 glucose minimal medium ameliorated the fusA1 mutant's sensitivity to H₂O₂ (Table 4). The pathway of heme biosynthesis begins with the conversion of glutamyl-charged tRNA^Glu by the product of the hemA gene to glutamate-1 semialdehyde. We considered the possibility that in Fus^r mutants, the ameliorating effect of Casamino Acids might be due to the fact that they increase the level of charged glutamyl-tRNA^Glu available for the heme biosynthesis pathway. Accordingly, we added several individual amino acids as a supplement to M9 glucose minimal medium with H₂O₂ and measured the relative fitness of the wild-type and fusA1 strains in a growth competition assay. The addition of glutamine or glutamate, but not proline, histidine, or tryptophan, had a large effect in reducing the sensitivity of the fusA1 strain to H₂O₂ (Table 4). The magnitude of the increase in competitive ability (fitness) is as great with glutamine alone as it is with Casamino Acids. These data further support the suggestion that the sensitivity of Fus^r mutants to H₂O₂ is, at least in part due to a reduced level of heme synthesis. Furthermore, the positive effect of glutamine and glutamate addition suggested that the earliest step in the synthesis of heme is limited by the availability of charged glutamyl-tRNA^Glu for conversion by the hemA product.

TABLE 4.

Medium composition reverses the sensitivity of the Fus^r mutation fusA1 to hydrogen peroxide^a

Growth competition medium	CI (fusA1 mutant/WT)b
M9	3.2 × 10−2
M9 + 70 μM H2O2	<1 × 10−6
M9 + 70 μM H2O2 + 1% Casamino Acids	1.4 × 10−3
M9 + 70 μM H2O2 + 0.2% glutamine	1.4 × 10−3
M9 + 70 μM H2O2 + 0.2% glutamic acid	3.5 × 10−4

^aMedia (10 ml) were inoculated with (per milliliter) 10⁶ cells of the wild-type (MM14) and fusA1 (MM142) strains each and incubated in M9 minimal medium with 0.2% glucose (and the indicated additions) for 18 h at 37°C with shaking. The final cell number was approximately 10⁹/ml, which corresponds to about 10 generations of growth.

^bThe competition index (CI) is the ratio of the number of viable fusA1 mutant CFU to the number of wild-type (WT) CFU after 18 h of competition in each medium. A value of <1 × 10⁻⁶ means that no viable cells were recovered. Results are the means of results of three independent assays.

Fluorescence assay of heme levels.

The experiments described above suggest that Fus^r mutants have defects in the synthesis of heme. We decided to measure, using afluorescence assay (see Materials and Methods), the relative amount of heme in strain MM14 (wild type) and the Fus^r mutant MM142 (fusA1). For each strain, we assayed 10 independent cultures, using 1 OD₆₀₀ unit of cells for each measurement. The assay is based on stripping the iron from the heme moiety, which leaves a protoporphyrin molecule which fluoresces following excitation at 400 nm (see Materials and Methods). We normalized fluorescence to biomass, measured as either optical density of the starting culture or dry weight of the cells; the two methods gave the same result. We found that the Fus^r(fusA1) strain has significantly less heme than the wild-type strain (mean [±standard error of the mean] values of 0.69 [±0.02]and 1.00 [±0.05], respectively). This provides direct evidence that the Fus^r phenotype can cause a heme deficiency.

Respiration rate.

Salmonella serovar Typhimurium is a facultative anaerobe (28). It grows faster aerobically because a proton motive force is generated across the bacterial membrane with oxygen as a terminal electron acceptor (19, 28). This process drives the efficient synthesis of ATP (13). Heme-containing cytochromes are an essential part of this respiratory process. We considered the possibility that Fus^r mutants, being limited in the amount of heme they produce, might also be defective in aerobic respiration. We measured the rate of oxygen consumption for the wild type and for two Fus^r mutants, strains MM142 (fusA1) and JB2110 (fusA1-15), during exponential growth in LB medium. We found that each of the mutants had a significantly lower respiration rate than the wild type (Fig. 2). The respiration rate for the fusA1 mutant is only 41% of that measured for the wild type, and for the fusA1-15 mutant, the respiration rate is 74%. We conclude that these Fus^r mutants have a significantly reduced rate of respiration.

FIG. 2.

Respiration (rate of oxygen consumption) in the wild type and two different Fus^r mutants. The standard error of these measurements is less than ±2 U for each strain.

Iron-mediated DNA damage.

It was proposed that one consequence of a respiratory deficiency could be the accumulation of NADH, which ultimately provides to free iron the electrons that drive the Fenton reaction (11). Hydroxyl radicals are generated through the Fenton reaction, in which ferrous iron transfers an electron to hydrogen peroxide. These radicals then directly oxidize DNA, producing strand breaks and changes in DNA topology (16). Thus, inhibition of respiration results in an increased amount of cytosolic reductants available to reduce free ferric iron. In the presence of hydrogen peroxide, the ferrous iron thus formed will donate an electron and generate a hydroxyl radical that can damage DNA. This argument is supported by findings that inhibition of any step in the respiratory chain dramatically accelerates the rate at which hydrogen peroxide kills Escherichia coli (16, 33). To test this hypothesis with respect to Fus^r strains defective in respiration, we incubated cultures of the wild type and a Fus^r mutant with 2,2′-dipyridyl (an iron chelator that penetrates cells and chelates intracellular ferrous iron as well as iron outside the cells). Cultures were incubated together with 2,2′-dipyridyl for 30 min prior to exposure to 2.5 mM H₂O₂ for 45 min. In cultures treated directly with 2.5 mM H₂O₂, the percentages of cells viable after 45 min were 41% (wild type) and 20% (Fus^r mutant). In contrast, over 70% of the cells in each of the cultures pretreated with 2,2′-dipyridyl survived 45 min of exposure to 2.5 mM H₂O₂. These results suggest that the decreased survival of the Fus^r mutant in the presence of H₂O₂ is, at least in part, associated with iron-mediated DNA damage, since a decrease in the ferrous iron concentration improves survival.

Previous studies have shown that killing by micromolar concentrations of H₂O₂ (mode I bacterial killing) is mediated by iron-dependent DNA damage, while killing by millimolar H₂O₂ concentrations (mode II killing) is caused by oxidative damage to multiple targets (17). To compare the relative sensitivities of serovar Typhimurium Fus^r mutants to mode I and mode II killing, low (750 μM) and high (4 mM) H₂O₂ concentrations were used according to a published method (35). After exposure to 750 μM H₂O₂, 50% of the wild-type cells (MM14) and 3% of cells of the fusA1 mutant (MM142) and the fusA1-15 mutant (JB2110) survived after 3 h of incubation(Fig. 3A). In contrast, exposure to 4 mM H₂O₂for 2 h reduced the viability of the wild type to 1%, of JB2110 to 0.2%, and of MM142 to a level below the detection limit (<0.001%). Loss of viability after exposure to 4 mM H₂O₂ is believed to involve oxidative modification of lipids and proteins (17). These results suggest that MM142 (the fusA1 mutant) is hypersensitive to the oxidative damage to DNA, to proteins, and to lipids caused by mode II killing, whereas in the case of JB2110 (the fusA1-15 mutant), the DNA damage caused by Mode I killing might be the more important factor for survival.

FIG. 3.

Mode I versus mode II killing of the wild type and two different Fus^r mutants.

DISCUSSION

Fus^r strains of Salmonella serovar Typhimurium with mutations in EF-G have reduced catalase activity and are sensitive to oxidative stress caused by H₂O₂ (25). Detoxification of H₂O₂ in E. coli and Salmonella is due mainly to two distinct species of catalase. HPI, a bifunctional catalase-hydroperoxidase, is encoded by the katG gene and is active as a tetramer of 81-kDa subunits (9). HPI is transcriptionally induced by OxyR as part of a genetic response to H₂O₂ (34). HPI activity is observed in the periplasm and in cytoplasmic membrane fractions (14). HPII, a monofunctional catalase, is encoded by the katE gene (24), is active as a tetramer of 78-kDa subunits (10), and is under the control of rpoS (32). HPII is localized in the cytoplasm (14). Here we have shown that the reduced catalase activity in Fus^r mutants is associated with a reduction in activity for both HPI and HPII. This association suggested to us that it might reflect a deficiency in something common to both enzymes, such as the iron-binding cofactor heme. We found that expression of hemA, as measured from a hemA-lacZ fusion, is reduced in Fus^r mutants, which supports this hypothesis. Salmonella Serovar Typhimurium uses the enzyme glutamyl-tRNA reductase (encoded by hemA) to convert a small fraction of the cells' glutamyl-tRNA^Glu to glutamate semialdehyde. This in turn is converted by the enzyme glutamate-1-semialdehyde aminotransferase (encoded by hemL) to ALA (1, 12). We found that providing cells with ALA, the precursor in the biosynthesis of heme, improved the viability of Fus^r strains exposed to H₂O₂ to the wild-type level. It has been reported for E. coli that induction of the stringent response (7) results in the RelA-dependent excretion of glutamate from cells (6). Because the levels of ppGpp are disturbed in Fus^r mutants (26), we wondered whether the ALA deficiency might be due in part to excretion of glutamate, which possibly leads to insufficient charging of tRNA^Glu. We hypothesized that the external addition of either glutamate or glutamine (which can be converted into glutamate inside cells) might decrease the sensitivity of Fus^r mutants to H₂O₂. In support of this, we found that the addition of Casamino Acids, glutamate, or glutamine dramatically improved the survival and growth of the least fit Fus^r mutant, relative to the wild type, in the presence of H₂O₂. Other amino acids tested had no positive or negative effect on the Fus^r mutant. Because glutamyl-tRNA^Glu is the substrate of the hemA product, a reasonable interpretation of our results is that the insufficient availability of glutamate is, together with reduced transcription of hemA, a factor limiting heme synthesis in Fus^r mutants. Finally, we measured relative heme levels by a fluorescence assay and found that a Fus^r mutant (with a fusA1 genotype) has significantly less heme than the wild type (69% of that of the wild type). Thus, heme deficiency is a phenotype of Fus^r mutants.

Because heme is also a component of the cytochromes involved in respiration (3), we suspected that heme-deficient Fus^r mutants might also be defective in respiration. We measured respiration as the rate of oxygen consumption and found that the Fus^r mutants had a significantly reduced rate compared to the wild type. Similarly, it was previously shown that a heme-deficient mutant (hemA::Tn10) of Salmonella serovar Typhimurium was both hypersensitive to hydrogen peroxide and respired at half the rate of wild-type cells (11). Mutants with defects in respiration have been shown to suffer from an increased rate of iron-mediated DNA damage (11, 16). Inhibition of any step in the respiratory chain, either by cyanide or by genetic defects in respiratory enzymes, dramatically accelerates the rate at which hydrogen peroxide kills E. coli (18, 33, 38). It has also been established that respiratory blocks accelerate the rate of DNA damage (38). The respiratory blocks do not substantially affect the levels of intracellular free iron or H₂O₂, indicating that they accelerate damage because they increase the availability of the electron donor. Indeed, a respiratory deficiency causes the accumulation of NADH, which provides to free iron the electrons that drive the Fenton reaction. It was proposed that respiratory inhibitors, by blocking the oxidation of NADH, increase the amount of cytosolic reductants available to reduce free ferric iron. In the presence of hydrogen peroxide, the ferrous iron thus formed will donate an electron and generate a hydroxyl radical that can damage DNA. We found that the presence of 2,2′-dipyridyl, a chelator that penetrates cells and sequesters intracellular ferrous iron as well as iron outside the cells, prior to exposure to H₂O₂, increases the viability of a Fus^r mutant.

Thus, amino acid alterations giving an antibiotic resistance phenotype are, at least in the case of some fusidic acid-resistant mutants, associated with pleiotropic phenotypes which reduce bacterial fitness in a variety of environments. These phenotypes, including the defect in heme biosynthesis and the reduced heme levels identified in this study, are interesting in that they provided insight into the Achilles' heel of resistant strains. Importantly, this Achilles' heel is not limited to a Fus^r mutation associated with a severe growth defect in vitro, such as fusA1. We have shown here that a Fus^r mutation, fusA1-15, with an in vitro growth rate equivalent to that of the wild type, is also sensitive to oxidative stress. Sensitivity to oxidative stress is of interest because (i) aerobic growth is associated with the generation of reactive oxygen species and (ii) intracellular pathogens such as Salmonella are exposed to reactive oxygen species during the respiratory burst in macrophages, as reviewed previously (20). It has been suggested that a hemA mutation in Salmonella retards growth in an aerobic environment in vivo (4). This shows that the pleiotropic phenotypes associated with antibiotic resistance may result in exploitable weaknesses if one can identify the nature of the fitness costs in relevant environments. The approach taken here, i.e., investigation of the physiological consequences for the bacteria of becoming antibiotic resistant, may be generally useful in providing information that can be applied to identify novel targets or modify treatment therapies and therefore in dealing more effectively with the increasing problem of antibiotic-resistant bacteria.