Comparative Study of the Roles of AhpC and KatE as Respiratory Antioxidants in Brucella abortus 2308

ABSTRACT Brucella strains are exposed to potentially toxic levels of H2O2 both as a consequence of their aerobic metabolism and through the respiratory burst of host phagocytes. To evaluate the relative contributions of the sole catalase KatE and the peroxiredoxin AhpC produced by these strains in defense against H2O2-mediated toxicity, isogenic katE, ahpC, and katE ahpC mutants were constructed and the phenotypic properties of these mutants compared with those of the virulent parental strain B. abortus 2308. The results of these studies indicate that AhpC is the primary detoxifier of endogenous H2O2 generated by aerobic metabolism. KatE, on the other hand, plays a major role in scavenging exogenous and supraphysiologic levels of H2O2, although this enzyme can play a supporting role in the detoxification of H2O2 of endogenous origin if AhpC is absent. B. abortus ahpC and katE mutants exhibit wild-type virulence in C57BL/6 and BALB/c mice, but the B. abortus ahpC katE double mutant is extremely attenuated, and this attenuation is not relieved in derivatives of C57BL/6 mice that lack NADPH oxidase (cybb) or inducible nitric oxide synthase (Nos2) activity. These experimental findings indicate that the generation of endogenous H2O2 represents a relevant environmental stress that B. abortus 2308 must deal with during its residence in the host and that AhpC and KatE perform compensatory roles in detoxifying this metabolic H2O2.

Brucella abortus, a facultative intracellular pathogen, causes abortion and infertility in cattle. Humans can also be infected by ingesting contaminated dairy products, through inhalation of infectious aerosols, or via direct contact with an infected fetus (43). Human brucellosis causes flu-like symptoms with a relapsing fever, and this debilitating disease can persist for months or years without appropriate treatment. Although human brucellosis remains a significant zoonotic disease worldwide (47) and a potential bioterrorism threat (70), there is currently no vaccine to prevent human infection, and antibiotic treatment of these infections remains problematic (2).
Prolonged survival and replication in host macrophages play a critical role in the virulence of the Brucella spp. (34,57). Experimental evidence indicates that reactive oxygen species (ROS) such as superoxide (O 2 Ϫ) and hydrogen peroxide (H 2 O 2 ) are important components of the brucellacidal activity of these phagocytes (31). Because the brucellae rely on respiratory metabolism for their energy production (52), these bacteria must also deal with endogenous ROS generated as a by-product of aerobic metabolism (27). Several enzymes that directly detoxify O 2 Ϫ and H 2 O 2 have been identified in Brucella. SodC is a periplasmic Cu-Zn superoxide dismutase (6), and phenotypic evaluation of an isogenic sodC mutant indicates that this enzyme protects B. abortus 2308 from O 2 Ϫ generated by the oxidative burst of host macrophages (22). Brucella strains also produce a single monofunctional catalase that is a structural homolog of Escherichia coli KatE. Although this protein does not possess a standard export signal sequence (63) or a predicted twin arginine transport signal sequence (data not shown), cell fractionation studies with the appropriate controls indicate that this protein resides in the periplasmic compartment (63). B. abortus and Brucella melitensis katE mutants exhibit increased sensitivity to H 2 O 2 compared to their parental strains in in vitro assays (21,33). A B. melitensis katE mutant retains its virulence in experimentally infected goats (21), and B. abortus katE mutants display wild-type virulence in the mouse model (59). These experimental findings suggest that KatE does not play an indispensable role in protecting the brucellae from oxidative killing by host phagocytes. A gene (BAB1_0591) encoding a Mn superoxide dismutase (SodA) has also been identified in B. abortus 2308. SodA activity increases in a B. abortus sodC mutant, suggesting that SodA works in concert with SodC to protect B. abortus 2308 from oxidative damage (65), but the precise role that SodA plays in resistance to oxidative stress in this bacterium remains to be determined experimentally.
The genes designated BAB2_0531 and BAB2_0532 in the B. abortus 2308 genome sequence are predicted to encode the components of an alkyl hydroperoxide reductase complex (AhpC and AhpD, respectively). Peroxiredoxins of the AhpC family detoxify H 2 O 2 , organic peroxides, and peroxynitrite (ONOO Ϫ ) (9,48). AhpD and AhpF are peroxiredoxin reductases that use reducing equivalents generated by cellular metabolism to restore the enzymatic activity of AhpC (10,49). Studies performed with multiple bacterial species indicate that the AhpCD and AhpCF complexes serve as important antioxidants (4,8,11,15,16,36,37,41,44,55,66), and indeed, work in E. coli suggests that AhpC is the major scavenger of H 2 O 2 generated in the cytoplasm of this bacterium as a by-product of aerobic metabolism (61). AhpC has also been shown to play a role in the virulence of several bacterial pathogens, including Helicobacter pylori (45), Mycobacterium bovis (72), and Staphylococcus aureus (15) but does not appear to be required for the virulence of Salmonella enterica serovar Typhimurium (68), Mycobacterium tuberculosis (64), Legionella pneumophila (51), or Porphyromonas gingivalis (32) in experimental models.
In this report, we present evidence that AhpC is the primary antioxidant used by B. abortus 2308 to detoxify endogenous H 2 O 2 generated by respiratory metabolism during routine aerobic cultivation. KatE, on the other hand, plays a major role in scavenging exogenous and supraphysiologic levels of H 2 O 2 , although this enzyme can play a supporting role in the detoxification of H 2 O 2 of endogenous origin if AhpC is absent. Interestingly, AhpC and KatE appear to play complementary roles in protecting B. abortus 2308 from H 2 O 2 of metabolic origin during residence in mice, and the presence of either AhpC or KatE alone is sufficient to allow this strain to maintain a chronic infection.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Brucella abortus 2308 and derivatives of this strain (Table 1) were cultivated on Schaedler agar (SA; Becton, Dickinson and Company) supplemented with 5% defibrinated bovine blood (SBA) at 37°C with 5% CO 2 or in brucella broth (Becton, Dickinson and Company) at 37°C with shaking at 165 rpm. Escherichia coli strains were grown in Luria Bertani (LB) broth (58) or on LB agar at 37°C. Chloramphenicol (15 g/ml for Brucella strains and 30 g/ml for E. coli strains) (Sigma-Aldrich), kanamycin sulfate (45 g/ml) (Invitrogen), and ampicillin or carbenicillin (25 g/ml for Brucella strains and 100 g/ml for E. coli strains) (Sigma-Aldrich) were added to culture media as necessary for selection of bacterial strains carrying antibiotic resistance markers.
Recombinant DNA techniques. Standard methods were employed for the manipulation of recombinant DNA molecules and amplification of DNA by PCR (Table 2) (3,58). Plasmid DNA was introduced into Brucella strains by electroporation (19).
Construction of B. abortus mutants. A previously described gene replacement strategy (19) was used to introduce defined mutations into the genome of B. abortus 2308. ColE1-based plasmids containing cat-disrupted versions of the ahpCD (pKHS3) and katE (pMEK7-9c) loci and a bla-disrupted ahpCD locus (pKHS5) ( Table 1) were independently introduced into B. abortus 2308 by , and KH2 were confirmed by PCR analysis of genomic DNA from these strains by use of ahpCD-, katE-, cat-, and bla-specific primer sets as appropriate and Southern blot analysis with probes for ahpCD, cat, and bla. Plasmid pMEK7-9k, which contains an aph3A-disrupted version of katE (21), was introduced into the B. abortus ahpCD mutant KH16 by electroporation, and transformants were selected on SBA supplemented with kanamycin. A putative B. abortus ahpCD katE double mutant (designated KK21) was selected for further evaluation based on its resistance to kanamycin and chloramphenicol and its sensitivity to ampicillin. A two-step process was also used to construct a second B. abortus ahpCD katE double mutant, designed to meet the regulatory requirement that Brucella strains engineered to possess resistance to chloramphenicol not be introduced into experimentally infected animals. First, pMEK7-9k was used in a gene replacement strategy as described above to construct a katE mutant (MEK6) from B. abortus 2308. Plasmid pKHS5 was then used to introduce a bla-disrupted version of the ahpCD locus into MEK6, resulting in the construction of the B. abortus ahpCD katE double mutant KK9. The genotypes of B. abortus strains KK21 and KK9 were confirmed by PCR analysis of genomic DNA from these strains by use of ahpCD-, katE-, cat-, bla-, and aph3A-specific primer sets as appropriate and Southern blot analysis with probes for ahpCD, cat, bla, and aph3A.
Crystal violet exclusion was used to verify that the B. abortus ahpCD and katE mutants and ahpCD katE double mutants retain their smooth lipopolysaccharide phenotypes (1). A solution of 3% H 2 O 2 was also placed on bacterial colonies grown on Schaedler agar to verify the presence or absence of visible catalase activity in the B. abortus strains used in this study.
Quantification of peroxide levels in Brucella abortus cell suspensions. B. abortus strains grown on SBA supplemented with the appropriate antibiotics for 48 h were inoculated into 3 ml brucella broth in 17-by 100-mm tubes and incubated at 37°C with shaking at 165 rpm. Following overnight incubation, the bacterial cells were harvested by centrifugation and resuspended in phosphate-buffered saline (PBS) to an optical density at 600 nm (OD 600 ) of 1.0. A commercial version of the xylenol orange/ferrous iron-based hydrogen peroxide assay originally described by Wolff (73) (National Diagnostics) was used to measure the amount of peroxides in the cell suspensions in accordance with the manufacturer's directions. Briefly, 100 l of bacterial cell suspension was added to 900 l of the assay reagent, the reaction mixtures were incubated at room temperature for 30 min, and the absorbance of the reaction mixtures was measured with a spectrophotometer at 560 nm. Cell-free H 2 O 2 standards (0 M, 1 M, 2 M, 4 M, and 8 M) were used to construct a standard curve, and the levels of peroxides in the cell suspensions were determined by comparison to the standard curve.
This assay was also used to measure the degradation of H 2 O 2 by B. abortus strains after an exogenous exposure. B. abortus strains grown on SBA supplemented with appropriate antibiotics for 48 h were inoculated into 5 ml brucella broth in 17-by 100-mm tubes and incubated at 37°C with shaking at 165 rpm. Following overnight incubation, the bacterial cells were harvested by centrifugation (12,100 ϫ g for 10 min at room temperature) and resuspended in PBS to an OD 600 of 1.0 in three 17-by 100-mm tubes, designated "no exposure," "immediately after exposure," and "time after exposure." For the "immediately after exposure" tubes, cultures were vortexed for 5 s after the addition of hydrogen peroxide, and 100 l of bacterial cell suspension was removed and immediately added to 900 l of the assay reagent. For the "time after exposure" tubes, cultures were vortexed for 5 s after the addition of H 2 O 2 and allowed to incubate at 37°C with shaking at 165 rpm for the appropriate time. Following incubation, 100 l of the bacterial cell suspension was removed and added to 900 l of the assay reagent.
Disk assay for measuring the sensitivity of B. abortus strains to PQ and H 2 O 2 .
B. abortus strains were grown on SBA supplemented with the appropriate antibiotics for 2 days, harvested into brucella broth, and adjusted to a cell density of 10 9 CFU per ml (OD 600 ϭ 0.15). Six hundred microliters of each cell suspension was then added to 18 ml prewarmed (55°C) brucella broth supplemented with 0.7% agar, and 3-ml portions of the resulting cell suspensions were plated onto three Schaedler agar (SA) plates and three SA plates containing 7,800 U/ml bovine catalase (Sigma). A sterile 7-mm Whatman no. 3 filter paper disk was placed in the center of each plate, and 10 l of a fresh 0.5 M solution of paraquat (PQ; Acrōs Organics) was added to each disk. Plates were incubated for 3 days, and the zones of inhibition surrounding each disk were measured in millimeters. This same assay was also used to measure to sensitivity of B. abortus strains to H 2 O 2 , with the exceptions that 10 l of a 30% solution of H 2 O 2 was added to the filter disks instead of PQ, and SA plates supplemented with bovine catalase were not used.
Growth characteristics of the Brucella abortus strains in rich and nutrientlimited media. B. abortus strains were grown overnight in 3 ml brucella broth in 17-by 100-mm culture tubes incubated at 37°C with shaking at 165 rpm. The resulting cultures were inoculated into either 500-ml flasks containing 100 ml of brucella broth at a cell density of approximately 10 3 CFU per ml or 500-ml flasks containing 100 ml Gerhardt's minimal medium (GMM) (23) at a cell density of 10 8 CFU per ml and the flasks incubated at 37°C with shaking at 165 rpm. The number of viable brucellae in these cultures was determined at selected time points after inoculation by serial dilution and plating on SBA or SBA containing the appropriate antibiotic.
Peroxynitrite resistance assay. B. abortus strains were grown on SBA at 37°C with 5% CO 2 for 48 h. Bacterial cells were harvested and resuspended to a cell density of 10 8 in 1 ml PBS in 17-by 100-mm culture tubes. The peroxynitrite generator SIN-1 (3-morpholinosydnonimine HCl; Sigma Aldrich) at a final concentration of 15 mM and 1,000 U/ml of bovine catalase were added to the cell suspensions and the mixtures incubated for 1 h at 165 rpm at 37°C. The numbers of viable brucellae in these cultures and in parallel cultures that were not exposed to SIN-1 were then determined by serial dilution and plating on SBA.
Experimental infection of cultured murine macrophages. A modification of the methods described by Gee et al. (22) was used to evaluate the capacity of the B. abortus strains to survive and replicate in cultured murine resident peritoneal macrophages. Briefly, macrophages obtained from 6-to 8-week-old female BALB/c mice were seeded at a density of 1.5 ϫ 10 5 cells per well in sterile 96-well plates in Dulbecco's modified Eagle's medium (ATCC) with fetal calf serum (DMEM-FCS) containing 20 g/ml gentamicin. After an overnight incubation, 100 U/ml gamma interferon (IFN-␥; Peprotech) was added to cultured macrophages for 30 min before macrophages were washed and coincubated with the Brucella strains (multiplicity of infection [MOI], 50:1) that were previously opsonized with a subagglutinating concentration (1:1,000) of hyperimmune mouse serum. After allowing phagocytosis to occur for 2 h, extracellular bacteria were removed by incubating the macrophage monolayer with DMEM-FCS containing 50 g/ml gentamicin for 1 h. The cell culture medium was then replaced with DMEM-FCS containing 20 g/ml gentamicin, and this medium was replaced at 24-h intervals for incubation times that extended to 48 h. At 2, 24, and 48 h postinfection (p.i.), the phagocyte monolayer was washed with PBS-FCS and lysed with 0.1% deoxycholate, and the number of viable intracellular brucellae present was determined by serial dilution and plating on SBA. Triplicate wells of phagocytes infected with each strain were evaluated at every time point.
Experiments were also performed to evaluate the effects of treatment of the cultured macrophages with the NADPH oxidase inhibitor apocynin (Sigma) (42) and the inducible nitric oxide synthase (iNOS) inhibitor N G -methyl-L-arginine (L-NMMA; Fisher) (18). The protocol described above was used, except that 500 M apocynin, L-NMMA, or both were included in the DMEM-FCS throughout the experiment. Microscopic analysis of nitroblue tetrazolium reduction was used to monitor the oxidative burst capacity of the cultured macrophages (7), and control experiments were performed to ensure that apocynin and/or L-NMMA at this concentration in DMEM was not toxic for the Brucella strains.
Experimental infection of mice. The procedures previously described by Robertson and Roop (53) were used to evaluate the spleen colonization profiles of the B. abortus strains in C57BL/6 and BALB/c mice from Harlan Laboratories. These methods were also used to compare the virulence of B. abortus strains in C57BL/6J mice and derivatives of these mice that lack a functional phagocyte NADPH oxidase (B6.129S6-Cybb tm1Din /J) or inducible NO synthase (B6.129P2-Nos2 tm1lau /J) obtained from Jackson Laboratories. Briefly, mice were infected with 5 ϫ 10 4 brucellae via the intraperitoneal route, and at each sampling point postinfection, the mice were euthanized, their spleens aseptically removed, and spleen homogenates serially diluted and plated on SBA to determine the number of viable brucellae present.

RESULTS
Identification of an alkyl hydroperoxide reductase complex (AhpCD) in B. abortus 2308. The genes designated BAB2_0531 and BAB2_0532 in the B. abortus 2308 genome sequence are annotated as ahpC and ahpD, respectively. The products of these two genes are predicted to be components of the alkyl hydroperoxide reductase complex AhpCD. In many bacteria, the peroxiredoxin AhpC serves as an important antioxidant that detoxifies hydrogen peroxide, organic peroxides, and/or peroxynitrite (9,26,61). AhpD is a peroxiredoxin reductase that uses reducing equivalents generated by cellular metabolism to recycle the enzymatic activity of AhpC (10,26). The Brucella AhpC shares 47% amino acid identity with the Mycobacterium tuberculosis AhpC, and the Cys-61, Cys-174, and Cys-176 residues that have been shown to be important for activity in the latter protein (12,26) are conserved as Cys-57, Cys-171, and Cys-173 in the Brucella AhpC ortholog. Likewise, Brucella AhpD displays 44% amino acid identity with its M. tuberculosis counterpart, and amino acid sequence alignment indicates that the Cys-131 and Cys-134 residues in this protein are equivalent to the Cys-130 and Cys-133 residues that are required for the peroxiredoxin reductase activity of mycobacterial AhpD (10,26). Reverse transcriptase PCR analysis indicates that the ahpC and ahpD genes in B. abortus 2308 are cotranscribed as an operon (data not shown), which is consistent with the predicted function of their products in an enzymatic complex and the genetic organization of the ahpCD operons in other bacteria (10,26).
A B. abortus ahpCD mutant exhibits higher levels of endogenous cellular peroxides than the parental strain. Studies performed with E. coli indicate that AhpC plays a major role in removing the H 2 O 2 that is generated in the cytoplasm of this bacterium as a by-product of aerobic metabolism (61). Phenotypic analysis of the B. abortus mutant ahpCD KH16 suggests that Brucella the AhpC performs a similar function. Significantly higher levels of endogenous peroxides are detected in KH16 than in the parental 2308 strain following aerobic growth (Fig. 1), and the levels of these ROS are significantly diminished in a derivative of the ahpCD mutant carrying a plasmidborne copy of the ahpCD locus. Endogenous peroxide levels also return to approximately wild-type levels in a derivative of KH16 carrying a plasmid that overexpresses katE (Fig. 1). Because monofunctional catalases such as the Brucella KatE detoxify H 2 O 2 but not organic peroxides (13,38), these findings indicate that the elevated levels of endogenous peroxides detected in the ahpC mutant are predominantly made up of H 2 O 2 . It is also important to note that overexpression of katE or the addition of exogenous catalase reduces the levels of endogenous peroxides detected in B. abortus 2308 cell suspensions below the baseline levels shown in Fig. 1 (data not  shown), indicating that this assay provides a reliable indication of the levels of endogenous H 2 O 2 generated by the Brucella strains examined in this study.
The biochemical properties of catalases allow these enzymes to degrade H 2 O 2 across a broad range of concentrations (67). Accordingly, catalases often provide bacteria with a second line of defense against the buildup of endogenous H 2 O 2 of metabolic origin when primary detoxifiers such as AhpC are absent (15,61). The levels of peroxides detected in the B.
abortus katE mutant KH2 are substantially lower than those detected in the isogenic ahpCD mutant and not significantly different from those detected in the parental 2308 strain. Moreover, although the levels of endogenous peroxides detected in the B. abortus ahpCD katE double mutant KK21 are consistently higher than those detected in the isogenic ahpC mutant KH16, these differences are not statistically significant. These experimental findings suggest that KatE plays a limited role in protecting B. abortus 2308 from the buildup of endogenous H 2 O 2 during routine aerobic cultivation.
AhpCD is required for the wild-type resistance of B. abortus 2308 to endogenous H 2 O 2 generated by the redox cycling agent paraquat. Paraquat (PQ) reacts with components of the respiratory chain in bacterial cells, leading to the univalent reduction of O 2 and the generation of O 2 Ϫ in these cells (24). This O 2 Ϫ then serves as a substrate for cytoplasmic superoxide dismutases such as SodA, which can convert this ROS to H 2 O 2 and O 2 (20). Spontaneous nonenzymatic dismutation of O 2 Ϫ to H 2 O 2 and O 2 also occurs under physiologic conditions (27). Thus, one of the consequences of treating respiring bacterial cells with PQ is the generation of increased intracellular levels of H 2 O 2 . The B. abortus ahpCD mutant KH16 and the ahpCD katE mutant KK21 consistently and reproducibly exhibit larger zones of inhibition around disks containing PQ in a disk sensitivity assay than does the parental 2308 strain or the isogenic katE mutant (Fig. 2A). Introduction of a plasmid-borne copy of the ahpCD locus reduces the sensitivity of KH16 and KK21 to PQ to approximately the same levels displayed by B. abortus 2308.
H 2 O 2 is an uncharged ROS and can readily cross cellular membranes by diffusion. This allows extracellular catalase to serve as an efficient detoxifier of intracellular H 2 O 2 (62). Consequently, since the addition of paraquat results in the generation of both superoxide and hydrogen peroxide, an important control in these assays is the addition of exogenous catalase to the test medium to relieve hydrogen peroxide toxicity. This allows for determination of whether or not the increased susceptibility of the B. abortus ahpCD and ahpCD katE mutants to PQ is due to the increased intracellular accumulation of H 2 O 2 . As shown in Fig. 2B, the addition of exogenous catalase to the test medium reduces the zone of inhibition around disks containing PQ exhibited by B. abortus KH16 (ahpCD) and KK21 (ahpCD katE) to the same size as those exhibited by the parental 2308 strain. This indicates that the increased sensitivity of the B. abortus ahpCD and ahpCD katE mutants to PQ is due to the increased intracellular accumulation of H 2 O 2 and not a differential sensitivity of these mutants to O 2 Ϫ. More importantly, these experimental findings further support the contention that AhpC serves as a primary detoxifier of endogenous H 2 O 2 produced by respiratory metabolism in B. abortus 2308, while KatE plays a limited and secondary role in this regard. The ahpCD locus is required for maintenance of stationaryphase viability of B. abortus 2308 during aerobic growth in a defined minimal medium. B. abortus 2308, KH16 (ahpCD), KH2 (katE), and KK21 (ahpCD katE) exhibit equivalent growth kinetics and viability during exponential growth and stationary phase when these strains are cultivated aerobically in brucella broth (Fig. 3A). When these strains are grown in Gerhardt's minimal medium (GMM), however, the B. abortus ahpCD mutant KH16 and the isogenic ahpCD katE double mutant KK21 both exhibit a significant loss of stationary-phase viability compared to their respective parental strains 2308 and KH2. Similarly increased levels of endogenous peroxides are present in the B. abortus ahpCD mutant KH16 and the ahpCD katE double mutant KK21 compared to those present in 2308 and the katE mutant KH2 during growth in GMM (data not shown). The loss of stationary-phase viability in GMM exhibited by the B. abortus ahpCD mutant KH16 can also be rescued to a significant degree by the introduction of a plasmid containing either ahpCD or katE into this strain (Fig. 3C), and this phenotype in the B. abortus ahpC katE double mutant KK21 can be rescued by a plasmid carrying katE. These data suggest that AhpC plays a particularly important role in detoxifying endogenous H 2 O 2 generated during stationary phase in B. abortus 2308 during in vitro cultivation under nutrient-limiting conditions.
KatE is the major detoxifier of exogenous hydrogen peroxide in B. abortus 2308. While AhpC appears to be the major detoxifier of endogenous H 2 O 2 in B. abortus 2308, the results presented in Fig. 4 indicate that KatE is the major detoxifier of exogenous H 2 O 2 in this strain. Even at levels of exogenous H 2 O 2 as low as 5 M, the B. abortus katE and ahpCD katE mutants exhibit a marked defect in their capacity to degrade exogenous H 2 O 2 compared to their parental strains (Fig. 4), and these defects are much more dramatic when these strains are exposed to 50 and 100 M H 2 O 2 (Fig. 5). The role of KatE in the degradation of exogenous H 2 O 2 is further reflected in the differences in the sensitivities to H 2 O 2 exhibited by the B. abortus ahpC and katE mutants and the ahpCD katE double mutant in a disk sensitivity assay (Table 3), where the strains lacking KatE display a much more pronounced phenotype than the ahpCD mutant.
The B. abortus ahpCD mutant KH16 displays increased sensitivity to peroxynitrite. Biochemical studies have shown that in addition to H 2 O 2 , AhpC can also detoxify organic peroxides such as tert-butyl hydroperoxide (t-BOOH) (26), cumene hydroperoxide (CHP) (28,50), and peroxynitrite (ONOO Ϫ ) in vitro (9), and genetic studies have shown that this peroxiredoxin provides bacterial cells with an important defense against environmental exposure to these compounds (14,39,66). The results shown in Fig. 6 suggest that AhpC plays an important role in protecting B. abortus 2308 from exposure to ONOO Ϫ . Compared to the parental strain, the ahpC mutant KH16 displays an increased sensitivity to the ONOO Ϫ generator SIN-1 in an in vitro assay, and genetic complementation of KH16 with a plasmid-borne wild-type version of the ahpCD locus restores the resistance of the mutant to ONOO Ϫ to the same levels as those exhibited by the parent strain. In contrast, the extent to which AhpC contributes to the detoxification of organic peroxides in B. abortus 2308 is unclear. The B. abortus ahpCD mutant KH16 exhibits variable and inconsistent sensitivity to t-BOOH and CHP in standard in vitro assays, and the levels of lipid hydroperoxides present in B. abortus 2308 and  (57). To determine to what extent AhpC and KatE protect B. abortus 2308 from H 2 O 2 of exogenous and endogenous origin in the host, the virulence properties of B. abortus 2308 and isogenic ahpCD, katE, and ahpCD katE mutants in cultured murine macrophages and experimentally infected mice were evaluated. Only the B. abortus ahpCD katE mutant KK9 exhibited significant and stable attenuation compared to the parental 2308 strain in cultured murine macrophages (Fig. 7), and this attenuation was consistently observed only when these phagocytes were stimulated with IFN-␥. Notably, the addition of apocynin (a NADPH oxidase inhibitor), L-NMMA (an iNOS inhibitor), or both of these inhibitors in combination to the phagocyte cultures failed to alleviate the attenuation exhibited by the B. abortus ahpCD katE mutant in the IFN-␥-treated macrophages (Fig. 7). These experimental findings suggest that neither AhpC nor KatE is playing a role in protecting B. abortus 2308 from exogenous H 2 O 2 produced as a result of the oxidative burst of these phagocytes. They also suggest that AhpC does not play a prominent role in protecting B. abortus 2308 from exogenous ONOO Ϫ generated by the NADPH oxidase and iNOS activity of host macrophages.
The B. abortus ahpCD katE double mutant KK9 was also the only mutant to display significant attenuation compared to the parental 2308 strain in C57BL/6 ( Fig. 8A) or BALB/c (Fig. 8B) mice, and the severe attenuation exhibited by B. abortus KK9 in C57BL/6J mice was not alleviated in congenic NADPH oxidase and iNOS knockout mice (Fig. 8C and D). These data provide further evidence that neither AhpC nor KatE plays a direct role in protecting B. abortus 2308 from the oxidative or nitrosative bursts of host phagocytes. Instead, they support the contention that the buildup of endogenous H 2 O 2 is a biologically relevant environmental stress encountered by B. abortus 2308 during its residence in the murine host (57). Moreover, they demonstrate that the presence of either AhpC or KatE alone is sufficient to alleviate this stress.

DISCUSSION
The experimental findings presented here show that AhpC plays a major role in scavenging H 2 O 2 that is generated as a by-product of respiratory metabolism in B. abortus 2308. This function is similar to that reported for the E. coli AhpC and consistent with the reported biochemical properties of this class of peroxiredoxins in general, which work most efficiently on low levels of H 2 O 2 (61). The capacity of AhpC to scavenge metabolic H 2 O 2 appears to be especially important to B. abortus 2308 for the maintenance of stationary-phase viability when this strain is cultured under nutrient-limited conditions. This function is consistent with the observation that maximum expression of an ahpC-lacZ fusion is observed during stationary phase in B. abortus 2308 (K. Steele, unpublished observations), and AhpC has been proposed to be an important stationaryphase antioxidant (56,60,71). The basis for the H 2 O 2 -dependent loss of viability of the B. abortus ahpC mutant during stationary phase is not known. But the fact that the ahpC mutant does not exhibit this phenotype during growth in a nutritionally replete medium suggests that certain key biosyn-   (29).
In contrast to what was found for the ahpC mutant, the phenotypes exhibited by the B. abortus katE mutant and ahpCD katE mutant suggest that the sole catalase produced by this bacterium plays a minimal role in detoxifying endogenous H 2 O 2 of metabolic origin during routine aerobic cultivation in vitro. These results are intriguing considering the compensa-  abortus 2308 remains to be determined experimentally. Despite the fact that KatE appears to play a minimal role in protecting B. abortus 2308 from endogenous H 2 O 2 during routine aerobic cultivation in vitro, the studies performed with experimentally infected mice suggest that this enzyme plays a pivotal backup role in protecting this bacterium from the metabolic H 2 O 2 it generates during replication in the host. This proposition is based on two observations. First, although our in vitro studies suggest that the Brucella AhpC has the capacity to degrade H 2 O 2 , ONOO Ϫ , and possibly organic peroxides (see below), the only described function for monofunctional catalases such as KatE that the authors are aware of is the degradation of H 2 O 2 . This strongly suggests that H 2 O 2 toxicity plays a key role in the attenuation exhibited by the B. abortus ahpCD katE mutant. Second, this mutant displays the same level of attenuation in NADPH oxidase-deficient mice that it does in wild-type mice, indicating that AhpC and KatE do not provide protection from exogenous H 2 O 2 produced as a by-product of the oxidative burst of host phagocytes. The fact that the presence of either AhpC or KatE alone allows B. abortus 2308 to maintain persistent infection in mice suggests that brucellae possess functionally redundant systems to protect themselves from the metabolic H 2 O 2 they generate endogenously during replication in the host. This is perhaps to be expected of a bacterium that must deal with exposure to ROS of both endogenous origin as well as those generated by the NADPH oxidase and iNOS activity of host phagocytes (57) during residence in this environment.
The fact that neither AhpC nor KatE appears to play a role in protecting the brucellae from the oxidative burst of macrophages is intriguing, especially considering the role that ROS and IFN-␥ have been proposed to play in the brucellacidal activity of these phagocytes (31). Moreover, the temporal nature of the attenuation exhibited by the B. abortus ahpC katE double mutant KK9 in cultured macrophages (e.g., 24 h p.i.) and the observation that attenuation of this mutant was observed only when these phagocytes were stimulated by IFN-␥ are what would be predicted for a Brucella mutant with an increased sensitivity to the oxidative burst (31). The apparent lack of correlation between macrophage NADPH oxidase activity and the attenuation exhibited by the B. abortus ahpC katE double mutant is also perplexing given the documented role that Brucella SodC plays in detoxifying O 2 Ϫ produced by host macrophages (22), a process that generates H 2 O 2 . One possibility is that the exogenous H 2 O 2 produced as a by-product of the oxidative burst of host macrophages is less of a threat to the brucellae than the primary product of this reaction (e.g., O 2 Ϫ). This would be analogous to the situation observed for Salmonella strains, where mutants lacking SODs that detoxify exogenous O 2 Ϫ (e.g., sodC mutants) are more attenuated in experimental hosts than the corresponding catalase-or AhpCdeficient mutants (17,25,68). Clearly, a more comprehensive evaluation of the gene products that protect the brucellae from the respiratory burst of host macrophages is warranted. It will also be important to define the Brucella cellular components that are prone to damage by exogenous and endogenous ROS during replication in host macrophages.
The increased sensitivity of the B. abortus ahpC mutant to the ONOO Ϫ generator SIN-1 in in vitro assays suggests that the Brucella AhpC, like its counterparts in Salmonella enterica serovar Typhimurium, Mycobacterium tuberculosis, and Helicobacter pylori, has peroxynitrite reductase activity (9). This enzymatic activity has been proposed to be important as a bacterial defense against ONOO Ϫ production by host macrophages (39), but the results obtained in this study with the B. abortus ahpCD mutant KH40 and the ahpCD katE double mutant KK9 in iNOS-deficient mice suggest that AhpC does not play a prominent role in protecting the parental 2308 strain from ONOO Ϫ produced by host phagocytes.
In addition to their ability to detoxify H 2 O 2 and ONOO Ϫ , bacterial AhpC proteins have also been shown to be able to degrade organic peroxides. Indeed, the name alkyl hydroperoxide reductase reflects the fact that degradation of organic peroxides was the first property identified for many members of this class of bacterial enzymes (28). Thus, it is notable that no conclusive evidence was obtained from the studies described in this report supporting a role for AhpC in the detoxification of organic peroxides in B. abortus 2308. One possible explanation for these findings is that this bacterial strain also possesses the organic peroxide scavenger Ohr (40). Phenotypic analysis of an ohr mutant indicates a role for Ohr in the detoxification of the organic peroxides tert-butyl hydroperoxide and cumene hydroperoxide in B. abortus 2308 (J. Baumgartner, unpublished). Consequently, further phenotypic analysis of ahpC and ohr mutants and ahpC ohr double mutants will be needed to determine whether or not AhpC can detoxify organic peroxides in B. abortus 2308.
In summary, the results presented in this report indicate that AhpC and KatE play distinct but complementary roles in protecting B. abortus 2308 from exposure to H 2 O 2 . In order to better understand the contributions of these antioxidants to the physiology and intracellular lifestyle of this bacterium, it will be important in future studies to examine the coordinate regulation of ahpCD and katE during different stages of growth and in response to oxidative stress. It will also be important to determine how Brucella KatE is exported to the periplasm.