Control of Redox Balance by the Stringent Response Regulatory Protein Promotes Antioxidant Defenses of Salmonella*

We report herein a critical role for the stringent response regulatory DnaK suppressor protein (DksA) in the coordination of antioxidant defenses. DksA helps fine-tune the expression of glutathione biosynthetic genes and discrete steps in the pentose phosphate pathway and tricarboxylic acid cycle that are associated with the generation of reducing power. Control of NAD(P)H/NAD(P)+ redox balance by DksA fuels downstream antioxidant enzymatic systems in nutritionally starving Salmonella. Conditional expression of the glucose-6-phosphate dehydrogenase-encoding gene zwf, shown here to be under DksA control, increases both the NADPH pool and antioxidant defenses of dksA mutant Salmonella. The DksA-mediated coordination of redox balance boosts the antioxidant defenses of stationary phase bacteria. Not only does DksA increase resistance of Salmonella against hydrogen peroxide (H2O2), but it also promotes fitness of this intracellular pathogen when exposed to oxyradicals produced by the NADPH phagocyte oxidase in an acute model of infection. Given the role of DksA in the adjustment of gene expression in most bacteria undergoing nutritional deprivation, our findings raise the possibility that the control of central metabolic pathways by this regulatory protein maintains redox homeostasis essential for antioxidant defenses in phylogenetically diverse bacterial species.

We report herein a critical role for the stringent response regulatory DnaK suppressor protein (DksA) in the coordination of antioxidant defenses. DksA helps fine-tune the expression of glutathione biosynthetic genes and discrete steps in the pentose phosphate pathway and tricarboxylic acid cycle that are associated with the generation of reducing power. Control of NAD(P)H/NAD(P) ؉ redox balance by DksA fuels downstream antioxidant enzymatic systems in nutritionally starving Salmonella. Conditional expression of the glucose-6-phosphate dehydrogenase-encoding gene zwf, shown here to be under DksA control, increases both the NADPH pool and antioxidant defenses of dksA mutant Salmonella. The DksA-mediated coordination of redox balance boosts the antioxidant defenses of stationary phase bacteria. Not only does DksA increase resistance of Salmonella against hydrogen peroxide (H 2 O 2 ), but it also promotes fitness of this intracellular pathogen when exposed to oxyradicals produced by the NADPH phagocyte oxidase in an acute model of infection. Given the role of DksA in the adjustment of gene expression in most bacteria undergoing nutritional deprivation, our findings raise the possibility that the control of central metabolic pathways by this regulatory protein maintains redox homeostasis essential for antioxidant defenses in phylogenetically diverse bacterial species.
Bacterial pathogens have the remarkable ability to survive harsh conditions encountered during the course of infection. For instance, Gram-negative bacteria of the genus Salmonella must endure the extreme acidity of the stomach, osmotic stress in the intestinal lumen, as well as oxygen-dependent and -independent cytotoxicity in the intracellular environment of macrophages. Of all of these stresses, reactive oxygen species are arguably the most extensively studied host defenses. Salmonella infection triggers the assembly of an NADPH phagocyte oxidase respiratory complex (1) for the production of superoxide, a radical that rapidly dismutates to give rise to hydrogen peroxide (H 2 O 2 ). Membrane-permeable H 2 O 2 causes extensive damage to DNA bases and [4Fe-4S] prosthetics groups of dehy-dratases via Fenton-like chemistries that are dependent on the ferrous iron-catalyzed univalent or divalent reduction of H 2 O 2 (2,3). H 2 O 2 is also thought to mediate cytotoxicity through the oxidation of thiol groups in redox active cysteines (4). The importance of reactive oxygen species in host defense against Salmonella, as well as an assortment of other fungal and bacterial pathogens, is underscored by the increased incidence of recurrent infections in chronic granulomatous disease patients who lack a functional NADPH phagocyte oxidase (5). A murine model deficient in the gp91phox membrane-bound component of the NADPH oxidase accurately recapitulates the clinical importance of this enzymatic complex in the innate response of humans during Salmonella infections (6).
Bacteria, including Salmonella, have a diverse arsenal of antioxidant defenses. Superoxide dismutases, catalases, alkylhydroperoxidases, glutathione, and thioredoxins all directly or indirectly detoxify reactive oxygen species (7,8). Because of the membrane-remodeling activity of effectors secreted into the host cell by the type III secretion system encoded within Salmonella pathogenicity island 2, Salmonella can avoid NADPH phagocyte oxidase-containing vesicles (9 -11). Recent investigations on the electron transport chain highlight the importance of reduced nucleotides for the resistance of Salmonella against oxidative stress (12). Maintaining cellular reductive power may provide the fuel needed for detoxification of oxyradicals because many of the antioxidant enzymatic systems consume reduced pyridine nucleotides. It is still uncertain whether and to what extent central metabolic pathways contribute reducing power for classical antioxidant defenses. In the present work, we have elucidated a novel role for DksA 3 in the control of discrete steps of central metabolism that fuel antioxidant defenses.
Previous studies in Escherichia coli have revealed DksA to have pleiotropic regulatory roles in gene expression, cell division, amino acid biosynthesis, quorum sensing, stress resistance, and virulence (13)(14)(15)(16)(17)(18)(19). Some of the global regulatory effects of DksA are dependent on alarmone guanine tetraphosphate (ppGpp) and the stringent response. The stringent response was originally identified as a global adaptation of bacteria to nutritional stress. Starvation for amino acids, glucose, phosphate, or fatty acids leads to the accumulation of ppGpp (20,21). This alarmone helps the cell adapt to starving conditions by down-regulating transcription of translational machinery, thereby balancing protein production with nutrient availability. Some of the regulatory actions of ppGpp on the activity of RNA polymerase are achieved in conjunction with DksA, a protein with structural homology to Gre transcription factors (22). However, independent and opposing roles for DksA have also been observed, supporting unique functions for DksA (23,24). DksA has been shown to play a role in the virulence of several pathogens including Salmonella; however, it is still unclear how adjustments in gene expression by DksA contribute to Salmonella pathogenesis (25). We report herein that DksA exerts transcriptional control of discrete steps in central metabolic pathways of the pentose phosphate pathway and tricarboxylic acid cycle during the stationary phase of Salmonella.
Our investigations indicate that DksA helps maintain a redox balance critical for Salmonella virulence in response to reactive oxygen species generated intracellularly by the NADPH phagocyte oxidase.

EXPERIMENTAL PROCEDURES
Bacterial Strains-Salmonella enterica sv. typhimurium strain ATCC 14028s was used as wild-type and as a background in the construction of mutants following the method previously described by Datsenko and Wanner (26) (supplemental Table  1). PCR amplicons containing kanamycin or chloramphenicol resistance cassettes, or a 3ϫFLAG tag, flanked by the flippase (Flp) recognition target (FRT) were generated from pKD13, pKD3, and pSUB11 templates using LA Taq high fidelity polymerase (TaKaRa, Madison, WI) and 60-base-long primers containing homology to the target gene (supplemental Table 2). Purified PCR products were electroporated into S. typhimurium strain TT22236 harboring the plasmid pTP223 that expresses an isopropyl 1-thio-␤-D-galactopyranoside-inducible Red recombinase system. Mutations were moved into S. typhimurium strain 14028s via P22-mediated transduction, and pseudolysogens were eliminated by streaking on Evans blue uranine agar. The antibiotic cassette was removed by recombining the flanking FRT sites with the Flp recombinase expressed from the temperature-sensitive pCP20 plasmid (27). In-frame deletions were verified by PCR analysis. Strain AV09294 lacking dksA was complemented with the low copy vector pWSK29 encoding a wild-type dksA allele expressed from its native promoter. Transcriptional fusions were constructed by pCP20-mediated integration of pCE36 encoding a promoterless lacZY into FRT scars of selected genes (28).
Conditional Expression of zwf-Construction of a zwf conditional mutant was achieved by placing the zwf open reading frame under the control of the tetracycline promoter/ operator. The tetracycline promoter/operator containing plasmid pASK75 was modified for the vector to be inserted into the Salmonella chromosome. Briefly, the origin of replication in pASK75 was replaced with the R6K origin amplified from plasmid pKD13. The amplicon containing the R6K origin and pASK75 were ligated after SpeI and PfIMI digestion, generating a pir-dependent suicide vector. The FRT site from pCE36 was directionally cloned into the MscI unique site of the modified pASK75. The ampicillin resistance cassette of the modified pASK75 plasmid was disrupted by directionally cloning a gentamicin cassette from pPS856 into ScaI and PvuI unique sites to generate pTX (see Fig. 6A). A zwf conditional mutant was con-structed by placing the gene with its native ribosomal binding site into EcoRI and BamHI restriction sites placed downstream of the tet promoter. The final construct was confirmed by sequence analysis. The suicide vector was recombined by Flp expressed from pCP20 into a unique FRT site engineered downstream of putP in the chromosome of Salmonella strain AV09422. The P tet -controlled zwf was transduced into wildtype and dksA mutant Salmonella strain AV09294. Expression of the allele was initiated with 0.2 g/ml anhydrotetracycline (Sigma-Aldrich).
Real Time PCR-Bacterial cultures grown in Luria Bertani (LB) broth for 20 h were diluted to 10 8 cfu/ml in PBS and treated with 1 mM H 2 O 2 for 30 min at 37°C with shaking. Even though the concentration of H 2 O 2 used in these experiments is 10 times higher than that used for the killing assays described below, the high bacterial densities used for the isolation of RNA prevented the loss of bacterial viability. Total bacterial RNA was isolated in a bead-beater (Biospec Products, Inc.) as described previously (29) using TRIzol reagent (Invitrogen) containing silicon beads. RNA was extracted with chloroform, precipitated with isopropyl alcohol, washed with ethanol, and dried in a speed vacuum. Pellets were resuspended in RNasefree water and treated with RNase-free DNase (Ambion, Austin, TX). Complimentary cDNA was synthesized at 42°C using Moloney murine leukemia virus reverse transcriptase (Promega), RNasin (Promega), dNTPs, and random hexamers. cDNA was used as a template for real time and reverse transcription PCR. Transcripts of rpsM and rplN were normalized for the housekeeping gene rpoD.
␤-Galactosidase Assays-Strains carrying chromosomal lacZY fusions to genes encoding enzymes of pentose phosphate pathway or tricarboxylic acid cycle known to generate reducing power were constructed as described previously (28). To test whether these metabolic pathways are under the control of DksA, the lacZY fusions were transduced into the dksA mutant strain AV09294. LacZ enzymatic activity was quantified spectrophotometrically in bacterial cultures grown overnight in LB broth as ␤-galactosidase activity by following the conversion of o-nitrophenyl-␤-D-galactopyranoside to o-nitrophenyl. ␤-Galactosidase activity is expressed as Miller units calculated according to the equation: 1000 ϫ ((A 420 nm Ϫ 1.75 ϫ A 550 nm )/ (T(min) ϫ V(ml) ϫ A 600 nm )). Under the experimental conditions used in our assays, the background ␤-galactosidase activity was ϳ1 Miller unit.
Western Blotting-An overnight culture of dksA::3ϫFLAG grown in LB broth was diluted to 10 8 cfu/ml and treated for 1 h with 1 mM H 2 O 2 in PBS. Samples were taken every 15 min, pelleted, resuspended in lysis buffer, and disrupted by sonication. Samples normalized to 100 ng were resolved using 12% (v/v) SDS-PAGE, transferred electrophoretically to a nitrocellulose membrane, and immunoblotted with anti-FLAG M2 monoclonal antibody (Sigma-Aldrich).
In Vitro Susceptibility to Reactive Oxygen Species-Wild-type and dksA mutant bacteria grown overnight in LB broth were diluted in PBS to a final concentration of 10 5 cfu/well. Where indicated, the expression of zwf was induced by adding 0.2 g/ml anhydrotetracycline (Sigma-Aldrich) to bacteria diluted in PBS. Bacteria  Pyridine Nucleotide Quantification-Pyridine nucleotides were extracted and quantified using a modified cycling method described by Husain et al. (12). Briefly, oxidized and reduced nucleotides were extracted in 0.2 M HCl or 0.2 M NaOH, respectively, from bacterial cultures grown for 20 -22 h in LB medium or diluted in PBS with or without 0.4% glucose (w/v) or 0.1% casamino acids (w/v). Quantification of oxidized and reduced forms of pyridine nucleotides was achieved utilizing a modified thiazolyl tetrazolium blue cycling assay (12) using NADH/ NAD ϩ -specific alcohol dehydrogenase or NADPH/NADP ϩ glucose-6-phosphate dehydrogenase (30 -34). Nucleotide concentrations calculated by regression analysis of known standards were normalized for bacterial density as measured spectrophotometrically at A 600 nm . Intracellular nucleotide concentrations were calculated taking into account a bacterial cell volume of 10 Ϫ15 liters (35).
Glutathione Quantification-Total (GSH-GSSG) or oxidized glutathione (GSSG) was extracted and measured using a glutathione recycling method described by Baker et al. (36). Samples were prepared by pelleting 250 l of cultures grown 20 -22 h in LB broth. For experiments aimed at measuring the recovery of GSH after oxidative stress, bacterial cultures grown overnight in LB broth were pelleted, resuspended in PBS, and treated with 1 mM H 2 O 2 . 200 units of catalase were added after 5 min to eliminate the remaining H 2 O 2 . 250-l samples collected 5 and 15 min after the addition of catalase were resuspended in an equal volume of 20 mM EDTA. Samples used for the determination of GSSG were resuspended in 20 mM EDTA containing 2 mM N-ethylmaleimide. Bacteria were lysed by sonication using a 40-s pulse. One volume of 10% HClO 4 was added to the lysates to precipitate proteins. Lysates were neutralized with 93.5 l of KOH, freeze/thawed, and centrifuged to remove KClO 4 and obtain cleared lysates. N-Ethylmaleimide was removed by ether extraction. A reaction mixture was freshly prepared by adding 2.8 ml of 1 mM 5,5Ј-dithiobis (2-nitrobenzoic acid), 3.75 ml of 1 mM NADPH, 5.85 ml of 100 mM NaH 2 PO 4 -5 mM EDTA phosphate-EDTA buffer, and 20 units of glutathione reductase (Sigma-Aldrich). 100 l of reaction mixture was immediately added to 50 l of sample in a 96-well microtiter plate and read at A 412 nm for 5 min. Concentrations were obtained by regression analysis of known standards. Intracellular GSH-GSSG concentrations were calculated taking into account the number of colony-forming units and a bacterial cell volume of 10 Ϫ15 liters.
Mouse Infections-Six-to 8-week old C57BL/6 or congenic gp91phox-deficient (37) mice bred in our animal facility according to Institutional Animal Care and Use Committee guidelines were used to assess the role of DksA in Salmonella virulence. Briefly, individual animals were inoculated intraperitoneally with ϳ400 cfu of Salmonella grown overnight to stationary phase in LB broth. Mouse survival was monitored over time. Mice manifesting signs of distress (i.e. low spontaneous activity and ruffled coat) were humanely euthanized by CO 2 inhalation.
Macrophage Killing Assays-The anti-Salmonella activity of macrophages was evaluated as described previously (29). Briefly, peritoneal exudate cells were harvested from C57BL/6 or gp91phox-deficient mice 4 days after intraperitoneal inoculation of 1 mg/ml sodium periodate. The cells were cultured for 36 -48 h in RPMI ϩ 1640 medium (Cellgro, Herndon, VA) supplemented with 10% heat-inactivated fetal bovine serum (Bio-Whittaker, Walkersville, MD), 15 mM Hepes, 2 mM L-glutamine, 1 mM sodium pyruvate (Sigma-Aldrich), and 100 units/ml penicillin and 100 mg/ml streptomycin (Cellgro). Macrophages washed with media lacking antibiotics were challenged at a multiplicity of infection of 2 with Salmonella that had been opsonized with 10% normal mouse serum for 20 min. After 25 min of infection, the medium was exchanged with prewarmed RPMI ϩ 1640 medium containing 25 g/ml gentamicin. Macrophages were lysed with 0.25% deoxycholic acid (w/v) at the indicated time points after infection and the surviving intracellular bacteria enumerated on LB agar plates. Killing is expressed as the fraction of bacteria recovered at the indicated time relative to the bacterial burden isolated after 25 min of internalization.
Statistical Analysis-The data were analyzed using Student's paired t test. Determination of statistical significance between multiple comparisons was achieved using analysis of variance (ANOVA) followed by a Bonferroni post test using transformed data. Data were considered statistically significant when p Ͻ 0.05.

RESULTS
Hydrogen Peroxide Induces a Stringent Response-Salmonella, as is true for many bacteria, have several defenses with overlapping roles in the detoxification of reactive oxygen and nitrogen species. Published data indicate that the stringent response is induced in Salmonella exposed to reactive nitrogen species (38). In this work, we investigated whether H 2 O 2 , another reactive species produced in the innate response to Salmonella, may induce the stringent response. The stringent response is characterized by a down-regulation of translational machinery, including ribosomal protein gene expression. Consequently, transcription of the 30 S ribosomal protein gene rpsM was quantified by real time PCR as a readout of stringent regulation. The expression of rpsM was decreased (p Ͻ 0.01) 10-fold in wild-type Salmonella exposed to 1 mM H 2 O 2, whereas its transcription remained relaxed in a strain lacking the stringent response regulatory protein DksA (Fig. 1A). The transcription of the 50 S ribosomal protein rplN gene also remained relaxed in H 2 O 2 -treated dksA mutant bacteria (Fig.  1B). These findings indicate that H 2 O 2 induces DksA-dependent stringent regulation. H 2 O 2 did not, however, alter the amount of DksA protein in the cell (Fig. 1C). Moreover, the differences in ribosomal protein transcription between wildtype bacteria and the dksA mutant do not appear to reflect differences in growth rates as determined in a Bioscreen C growth analyzer (Oy Growth Curves AB Ltd., Helsinki, Finland) (Fig. 1D).
DksA Increases the Resistance of Salmonella to Oxidative Stress-Because the dksA mutant failed to repress transcription of rpsM and rplN upon exposure to H 2 O 2 , we tested whether a lack of dksA predisposes this bacterium to reactive oxygen species. H 2 O 2 , which is generated during the respiratory burst of macrophages in response to Salmonella (1), was chosen in these studies. Cytotoxicity assays were initially conducted using 400 M H 2 O 2 . However, this concentration killed all dksA mutant bacteria. Consequently, lower concentrations of H 2 O 2 were tested. The dksA mutant exhibited a time-dependent hypersusceptibility to 100 M H 2 O 2 ( Fig. 2A). Thirty times fewer dksAdeficient bacteria survived 2-h exposure to H 2 O 2 than did wildtype controls. A wild-type dksA allele expressed from the low copy plasmid pWDKS restored the survival of dksA-deficient Salmonella to levels seen in H 2 O 2 -treated wild-type bacteria.
At the low concentrations of bacteria used in these studies (i.e. 10 5 cfu/well), the decay of H 2 O 2 was insignificant over 2 h. The susceptibility of dksA mutant bacteria to H 2 O 2 -mediated killing is growth-dependent. Logarithmically grown dksA mutant bacteria were remarkably resistant to H 2 O 2 , whereas they became highly susceptible upon reaching stationary phase (Fig. 2B). The contribution of DksA to the antioxidant defenses of Salmonella appears to be independent of the stringent response alarmone ppGpp because a relA spoT (ppGpp°) mutant was significantly (p Ͻ 0.05) less susceptible to H 2 O 2 than the isogenic dksA-deficient strain. Moreover, a triple mutant lacking dksA relA spoT was as susceptible to H 2 O 2 as the single dksAdeficient Salmonella (p ϭ 0.170). The hypersusceptibility of dksA mutant bacteria to oxidative stress does not appear to be due to a defect in the consumption of H 2 O 2 (p ϭ 0.637 compared with wild-type bacteria) (Fig. 2C).
DksA Controls the Redox Balance of Stationary Phase Salmonella-Recently, it has been suggested that NADH plays a role in resistance of Salmonella to reactive oxygen species (12). Consequently, the pools of oxidized and reduced pyridine nucleotides were quantified in dksA mutant bacteria. Logarithmically grown wild-type and dksA-deficient Salmonella harbored similar levels of pyridine nucleotides ( Table 1). The reduced NADH and NADPH pyridine pools were nonetheless significantly (p Ͻ 0.001) lower in stationary phase dksA mutant bacteria than in wild-type controls. Conversely, the oxidized pyridine conjugates were increased in the former ( Table 1). The dramatic decrease in reducing power seen in stationary phase dksA-deficient bacteria raises the possibility that DksA controls central metabolic pathways that supply most of the reducing power in the cell. The majority of the NAD(P)H-reducing equivalents are generated in discrete enzymatic reactions of the pentose phosphate pathway and tricarboxylic acid cycle. To test whether steps associated with the production of reducing equivalents may be under control of DksA, lacZY transcriptional fusions were engineered to selected genes. Genes encoding glucose-6-phosphate dehydrogenase (zwf), isocitrate dehydrogenase (icdA), and malic enzyme (maeB) responsible for the generation of NADPH, and the sucAB-encoded subunits of ␣-oxaloacetate dehydrogenase associated with generation of NADH all showed DksA-dependent regulation (Fig. 3A, bold  red font). Several other genes of the pentose phosphate pathway, glycolysis, and tricarboxylic acid cycle such as gnd, aceEF, and mdh, which are also associated with production of reductive power, do not appear to be regulated by DksA.   Mutations were constructed in genes whose expression was decreased 2-fold or more in a dksA background. Strains deficient in sucAB or zwf showed decreases in reductive power compared with wild-type bacteria (Fig. 3, B and C). The sucAB and zwf mutants contained significantly (p Ͻ 0.05) lower NADPH intracytoplasmic concentrations than wild-type bacteria. These two strains were also hypersusceptible to 100 M H 2 O 2 (Fig. 3D). On the other hand, strain AV08257 lacking icdA harbored slight increases in NADH/NAD ϩ and NADPH/ NADP ϩ ratios and was hyperresistant to H 2 O 2 (Fig. 3, B-D), whereas the maeB-deficient strain was hypersusceptible to H 2 O 2 . Reduced cycling of metabolites from the tricarboxylic acid cycle into glycolysis may explain the hypersusceptibility of the maeB mutant to H 2 O 2 because pyruvate is known to be an efficient scavenger of reactive oxygen species (39). Similar to overnight cultures grown in LB medium, the dksA mutant strain cultured in PBS for 1 h harbored lower NAD(P)H/ NAD(P) ϩ ratios than wild-type controls ( Table 2). The addition of glucose or casamino acids to PBS not only raised the pools of reduced nucleotides but also increased the resistance of dksA mutant Salmonella to H 2 O 2 ( Table 2 and Fig. 3E). Together, these findings are interpreted as evidence in favor of a model in which carbon fluxes through central metabolic pathways increase the pool of reduced nucleotides that can be used for antioxidant defenses. Nonetheless, excessive carbon fluxing through these metabolic pathways appears to predispose against oxidative stress as shown by the fact that glucose and casamino acids enhanced the susceptibility of wild-type Salmonella to H 2 O 2 (Fig. 3E).

Redox Balance Associated with a Functional DksA Supports the Proper Function of Antioxidant Defenses-Intracellular redox potentials can be calculated with the Nernst equation
(E ϭ E°-RT/nF ϫ ln(reduced/oxidized)), where E°is the potential of the redox couple, R is the gas constant (8.3145 J⅐K Ϫ1 ⅐mol Ϫ1 ), n is the number of electrons transferred in the reaction, and F is Faraday's constant (9.6485 ϫ 10 4 C⅐mol Ϫ1 ) (40).
Using the intracytoplasmic NADPH/ NADP ϩ concentrations found in our studies, we calculated that wildtype and dksA mutant bacteria have an intracellular redox environment of Ϫ305 and Ϫ287 mV, respectively (Fig. 4A). Although noticeably lower than wild-type bacteria, the values in the dksA mutant should still be thermodynamically favorable for fueling glutathione (GSH) or thioredoxin redox couples with E values of Ϫ240 and Ϫ270 mV, respectively (41,42). Given this potential paradox, we investigated whether the decreased reductive power seen in dksA mutant Salmonella may be of consequence to the redox state of GSH. Compared with wild-type controls grown to stationary phase in LB broth, the dksA mutant had significantly (p Ͻ 0.001) lower concentrations of reduced GSH, al-   NOVEMBER 19, 2010 • VOLUME 285 • NUMBER 47 though it contained higher (p Ͻ 0.05) pools of oxidized GSH (GSSG) (Fig. 4B). Overall, the dksA mutant had a 3.5-fold lower GSH/GSSG ratio than wild-type bacteria (Fig. 4C). Next, we studied the effect of H 2 O 2 on glutathione levels in wild-type and dksA-deficient Salmonella. These studies were performed with 1 mM H 2 O 2 , which at the high bacterial densities used to biochemically analyze GSH/GSSG ratios, did not affect bacterial viability. The GSH/GSSG ratio decreased in both wild-type and dksA-deficient Salmonella upon challenge with H 2 O 2 in PBS (Fig. 4D), demonstrating that GSH is oxidized in response to H 2 O 2 . To test the turnover of GSSG, 200 units/ml catalase were added to the bacteria 5 min after 1 mM H 2 O 2 treatment. The redox GSH/GSSG balance quickly recovered to resting levels in wild-type Salmonella (Fig. 4D). This recovery was accompanied by a 2-fold increase in the total glutathione pool (GSHϩGSSG) (Fig. 4E). In contrast, the GSH/GSSG ratio only recovered partially in dksA-deficient Salmonella, which exhibited slower kinetics compared with wild-type bacteria. Moreover, its already smaller GSHϩGSSG pool dropped by half in the dksA mutant 10 min after challenge with 1 mM H 2 O 2 , rebounding to resting levels 10 min later (Fig. 4E). The pool of GSHϩGSSG did not decrease appreciably in wild-type bacteria 10 min after exposure to H 2 O 2 , whereas it increased (p Ͻ 0.01) by close to 2-fold 10 min after the addition of catalase.

DksA-regulated Antioxidant Defenses
The lower GSHϩGSSG pool seen in dksA mutant bacteria suggests that de novo GSH biosynthesis may also be regulated by DksA. To test this hypothesis, lacZY fusions were constructed to the genes gshA and gshB encoding glutamate-cysteine ligase and glutathione synthetase, respectively. Fig. 4F shows that the transcription of both steps in the biosynthesis of GSH, and especially the second step catalyzed by GshB, are under control of DksA. Collectively, these data indicate that DksA does not only help maintain a proper reduced nucleotide pool needed in the recycling of GSSG, but also coordinates the expression of glutathione biosynthetic genes.
Conditional Expression of the Glucose-6-phosphate Dehydrogenase Increases the Resistance of the dksA Mutant to H 2 O 2 -The majority of NADPH-reducing equivalents are generated in the pentose phosphate pathway by Zwf glucose-6-phosphate and Gnd 6-phosphogluconate dehydrogenases. The step catalyzed by Zwf rate limits the flux of carbon through the pentose phosphate pathway. Consequently, it is not surprising that a mutation in the zwf locus lowers cellular reductive power needed for resistance to H 2 O 2 . We therefore selected zwf as a candidate for the conditional complementation of the dksA mutation. The zwf gene containing its native ribosomal binding site was put under the control of the heterologous tetracycline promoter/operator, generating plasmid pTZWF (Fig. 5A). The pTZWF and pTX suicide plasmids containing a unique FRT site were integrated into the Salmonella chromosome in an intergenic region downstream of the putP gene (43). The constructs were transduced into the dksA mutant strain AV09294, generating AV10080 and AV10158, respectively. The induction of pTZWF with 0.2 g/ml anhydrotetracycline resulted in the production of large amounts of glucose-6-phosphate dehydrogenase as indicated by Coomassie Blue staining of cytoplasmic soluble proteins (data not shown). Conditional expression of zwf increased the survival of the dksA mutant expressing put::pTZWF by 10-fold when treated with 250 M H 2 O 2 (Fig. 5B). The increased resistance of the dksA mutant expressing zwf under the tet promoter coincided with lower (p ϭ 0.0027) NADP ϩ (dksA put::pTX versus dksA put::pTZWF; 1388 Ϯ 7 M versus 1255 Ϯ 26 M) and higher (p ϭ 0.0306) NADPH (dksA put::pTX versus dksA put::pTZWF; 153 Ϯ 4 M versus 186 Ϯ 12 M) levels at 30 and 60 min after induction. Consequently, the NADPH/NADP ϩ ratios increased upon induction with anhydrotetracycline (Fig. 5C).

DksA Contributes to Salmonella Virulence in a Murine
Model of Acute Salmonellosis-The induction of stringent regulation by H 2 O 2 and the high susceptibility of dksA mutant bacteria to H 2 O 2 raise the possibility that DksA may play a role in resistance against reactive species generated by NADPH oxidase. To evaluate the role of DksA during acute salmonellosis, C57BL/6 and congenic gp91phox-deficient mice were challenged intraperitoneally with wild-type or its isogenic dksA mutant Salmonella strain AV09294. Bacteria lacking dksA were highly attenuated in C57BL/6 mice (Fig. 6A). About 8,000 cfu of the dksA-deficient strain could be recovered from spleens of infected mice 14 days after inoculation. The isolated bacteria did not grow in E salts minimal medium supplemented with glucose, indicating that suppressor mutations had not developed during the course of the infection. In addition, PCR analysis indicated that the bacteria recovered from viscera of infected mice carried a mutated dksA allele. Although wild-type bacteria killed gp91phox-deficient mice 2 days earlier than the isogenic dksA-deficient Salmonella, the latter regained virulence in gp91phoxdeficient mice (Fig. 6B). Together, these data indicate that DksA helps Salmonella withstand the innate host response associated with a functional NADPH oxidase.
DksA Mediates Salmonella Fitness inside Professional Phagocytes-The early innate response of macrophages against Salmonella is dominated by reactive oxygen species generated through the enzymatic activity of the NADPH phagocyte oxidase. Given the dramatic attenuation of a dksA mutant in mice bearing a functional NADPH oxidase, the intracellular survival of wild-type and dksA mutant Salmonella was compared in primary macrophages. The number of Salmonella lacking dksA that were isolated from macrophages 2 h after infection was 3-fold lower than that of wild-type bacteria. Differences in bacterial burden of wildtype and dksA mutant Salmonella augmented at later times in the course of the infection, when wildtype bacteria increased whereas the number of dksA-deficient bacteria remained unchanged (Fig. 6C). The early disadvantage of dksA-deficient Salmonella in the intracellular environment of macrophages suggests that DksA helps antagonize the antimicrobial activity emanating from NADPH oxidase. To test this hypothesis, the intracellular survival of wild-type and dksA mutant bacteria was studied in gp91phox-deficient macrophages. Similar numbers of wildtype and dksA mutant bacteria were recovered from gp91phoxdeficient macrophages (Fig. 6D), supporting a role for DksA in defense against products of the respiratory burst.

DISCUSSION
Repression of genes encoding ribosomal proteins, rRNA, and tRNA is a hallmark of the stringent response of bacteria undergoing nutritional deprivation. Findings reported herein indicate that oxidative stress also represses transcription of the rpsM-and rplN-encoded 30 S and 50 S ribosomal proteins. In analogy to NO (38), the apparent induction of stringent regulation by H 2 O 2 may be related to nutritional stress. By modifying redox active centers, NO and H 2 O 2 affect a variety of cellular processes. In particular, the [4Fe-4S] prosthetic group of the IlvD dihydroxyacid dehydratase has been shown to react with  . DksA increases fitness of Salmonella exposed to the NADPH phagocyte oxidase. A and B, C57BL/6 (A) and congenic NADPH phagocyte oxidase-deficient (phox) (B) mice were inoculated intraperitoneally with ϳ400 cfu of either wild-type (WT) or dksA-deficient Salmonella. The percent of mice surviving the infection was evaluated over time. The data represent 10 mice/group from two separate experiments. C and D, the anti-Salmonella activity of periodate-elicited peritoneal macrophages isolated from C57BL/6 mice (C) or their congenic phox-deficient mice (D) was evaluated over time. The number of surviving intracellular bacteria was estimated at the indicated times after culture on LB agar plates as described under "Experimental Procedures. " The data represent the mean percent survival Ϯ S.E. (error bars) of 6 -12 independent observations from at least two independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.01 compared with WT controls. NOVEMBER 19, 2010 • VOLUME 285 • NUMBER 47 both reactive oxygen and nitrogen species, whereas the redox active thiol group of the MetE cobalamin-independent methionine synthase is a primary target of oxidative stress (2,44). Oxidation of redox active centers in these two proteins is expected to decrease the biosynthesis of branch-chain amino acids and methionine, thereby providing a mechanism for the H 2 O 2 -dependent induction of the stringent response.

DksA-regulated Antioxidant Defenses
Investigations presented herein indicate that DksA improves the resistance of Salmonella to the antimicrobial activity of the NADPH phagocyte oxidase. The antioxidant defenses mediated by DksA in Salmonella are functional within macrophages and in an acute model of infection. By exerting transcriptional regulation of central metabolic pathways, DksA controls the production of reducing power that helps detoxify oxyradicals and repair biomolecules damaged by reactive oxygen species. Our investigations indicate that the antioxidant defenses associated with DksA are mostly independent of the traditional stringent response alarmone ppGpp. This is in keeping with the idea that both DksA and ppGpp can have different modes of action and can compensate for amino acid auxotrophies, cellcell aggregation, motility, filamentation, and stimulation of RpoS accumulation when the other is absent (24). As is the case for the latter examples, further studies are needed to delineate the mechanisms by which DksA regulates redox homeostasis independently of ppGpp.
Bacteria lacking DksA exhibit a normal rate of consumption of H 2 O 2 at concentrations of the reactive species frequently seen in inflammatory processes. Thus, DksA does not appear to have a major effect on high rates of H 2 O 2 detoxification. In contrast, DksA helps regulate the production of reducing power in stationary phase bacteria. Transcriptional analysis has revealed that DksA enhances the expression of certain steps of the pentose phosphate pathway and tricarboxylic acid cycle that generate NAD(P)H. DksA was found to fine-tune the transcription of zwf, icdA, sucAB, and maeB. Conditional expression of zwf encoding the glucose-6-phosphate dehydrogenase boosted the redox potential and antioxidant defenses of dksA mutant bacteria. Given the fact that DksA plays a global regulatory role and is involved in the transcription of many steps of central metabolism, we found it remarkable that the conditional expression of zwf has such a profound effect on the resistance to H 2 O 2 in the dksA mutant. The resistance to H 2 O 2 associated with the heterologous expression of zwf was nonetheless delayed 30 min after NADPH accumulated in the bacteria, suggesting that NADPH fuels downstream enzymes. Together, the work presented herein provides evidence in favor of a model in which the generation of reductive power through the DksA-mediated control of central metabolic pathways boosts antioxidant defenses.
Application of the Nernst equation to the calculated intracellular concentrations of the NADPH/NADP ϩ redox couple indicates that the intracellular environment of dksA-deficient bacteria is Ϫ287 mV compared with Ϫ305 mV of wild type. The diminished redox power of the dksA mutant should still be thermodynamically favorable for fueling glutathione and thioredoxin reductases. Nonetheless, the cytoplasmic environment of dksA-deficient Salmonella is poor in reduced GSH, whereas it harbors high concentrations of GSSG. This observa-tion suggests that the reduced nucleotide pool in dksA mutant bacteria is indeed detrimental to downstream antioxidant defenses, including the glutathione oxidoreductase. Further supporting this idea, the dksA mutant exhibits lower rates in the reduction of GSSG than wild-type bacteria. The pool of ϳ200 M NADPH in dksA mutant Salmonella is still above the estimated K m of 20 M that glutathione reductase has for NADPH (45). According to Michaelis-Menten kinetics, the velocity of an enzymatic reaction is dependent on substrate level (V ϭ V max [S]/K m ϩ [S], where V is velocity, V max is maximal velocity, and S is substrate concentration). Considering V max and K m constant and taking into account the large pool of GSSG in the dksA mutant, it follows that limited NADPH availability likely drives the slow reduction of GSSG seen in this mutant. Collectively, these findings indicate that low NAD(P)H/NAD(P) ϩ ratios negatively affect the antioxidant defenses of dksA-deficient bacteria.
In addition to low pools of NADPH and NADH, the dksA mutant contains low levels of reduced GSH. This tripeptide is synthesized from cysteine, glutamate, and glycine. Auxotrophies for several amino acids, including glycine, have been described for dksA mutant bacteria (46). Therefore, the reduced transcription of glutathione synthase reported herein and limited glycine availability likely underlie the low GSH pool in the dksA mutant. The defect in GSH biosynthesis might shift the prime targets of oxidative stress from the tripeptide to more essential biomolecules. This idea is supported by the observed drop in measurable GSHϩGSSG (possibly through the formation of mixed disulfides) when dksA mutant bacteria were challenged with H 2 O 2 . A defect in GSH biosynthesis might aggravate susceptibility to oxidative stress of nutritionally deprived bacteria because they appear to synthesize GSH de novo upon treatment with H 2 O 2 . Because of the global role for DksA in the regulation of gene transcription, it is still possible that DksA controls the expression of other antioxidant defenses as well.
Control of antioxidant defenses by DksA is growth-dependent. Rapidly growing, dksA-deficient Salmonella are even more resistant to H 2 O 2 than wild-type bacteria. In contrast, stationary phase dksA mutant bacteria suffer a considerable loss of viability when exposed to H 2 O 2 . These findings are consistent with the idea that DksA exerts transcriptional control in stationary phase when nutrients become limited. Accordingly, the addition of carbon in the form of glucose or amino acids enhances both the reductive power and the resistance of dksAdeficient Salmonella to H 2 O 2 . The pool of reduced pyridine nucleotides must nonetheless be tightly controlled to minimize collateral damage. On one hand, NADH and NADPH reductive power fuels the antioxidant activities of alkyhydroperoxidase, and glutathione and thioredoxin reductases. On the other, NADH can act as a pro-oxidant that feeds reducing equivalents to flavoproteins, which can adventitiously generate the Fenton catalyst ferrous iron (47). Accordingly, experiments presented herein show that glucose and amino acids increase the cytotoxicity of H 2 O 2 against stationary phase wild-type Salmonella. Therefore, the fine-tuning of central metabolism by DksA during periods of nutritional starvation helps maintain a delicate balance of reduced nucleotides for use in antioxidant defenses while limiting excessive buildup of reducing power that could generate undesired damage.
In conclusion, our investigations have elucidated a novel role for DksA in the antioxidant defenses of Salmonella. The regulation of central metabolic pathways by DksA coordinates the supply of reducing power that fuels antioxidant defenses crucial for resistance of Salmonella to reactive oxygen species produced during the innate host response.