Glutathione and Catalase Provide Overlapping Defenses for Protection against Respiration-Generated Hydrogen Peroxide in Haemophilus influenzae

Abstract

Glutathione is an abundant and ubiquitous low-molecular-weight thiol that may play a role in many cellular processes, including protection against the deleterious effects of reactive oxygen species. We address here the role of glutathione in protection against hydrogen peroxide (H₂O₂) in Haemophilus influenzae and show that glutathione and catalase provide overlapping defense systems. H. influenzae is naturally glutathione deficient and imports glutathione from the growth medium. Mutant H. influenzae lacking catalase and cultured in glutathione-deficient minimal medium is completely devoid of H₂O₂ scavenging activity and, accordingly, substantial amounts of H₂O₂ accumulate in the growth medium. H. influenzae generates H₂O₂ at rates similar to those reported for Escherichia coli, but the toxicity of this harmful metabolite is averted by glutathione-based H₂O₂ removal, which appears to be the primary system for protection against H₂O₂ endogenously generated during aerobic respiration. When H₂O₂ concentrations exceed low micromolar levels, the hktE gene-encoded catalase becomes the predominant scavenger. The requirement for glutathione in protection against oxidative stress is analogous to that in higher and lower eukaryotes but is unlike the situation in other bacteria in which glutathione is dispensable for aerobic growth during both normal and oxidative stress conditions.

Cellular metabolism of molecular oxygen produces reactive and potentially toxic oxygen species such as superoxide radicals, hydrogen peroxide (H₂O₂), and hydroxyl radicals (5). Escherichia coli generates ca. 14 μM H₂O₂ per s when it grows exponentially in air in minimal glucose medium (36). Recently, Seaver and Imlay showed that E. coli mutants completely defective in H₂O₂ scavenging activity grow poorly in air and that growth often stopped entirely when these mutant cells were repeatedly subcultured (36). This implies that aerobically grown E. coli generates enough H₂O₂ to poison scavengerless cells. Indeed, H₂O₂ itself can potentially damage enzymes by oxidizing sulfhydryl and iron-sulfur moieties and, upon conversion to a hydroxyl radical, it produces mutagenic and lethal lesions (4, 21).

E. coli mainly expresses two systems to destroy H₂O₂. The first system is embodied by two types of catalase enzymes, a bifunctional catalase/peroxidase (HPI), encoded by katG (39), and a monofunctional catalase (HPII), encoded by katE (45), which are both heme-containing enzymes involved in the dismutation of H₂O₂ into O₂ and H₂O. Clear evidence shows that the two catalase genes are regulated differently in terms of growth phase and responsiveness to oxidative stress (reviewed in reference 22). HPI is transcriptionally induced during logarithmic growth in response to low micromolar concentrations of H₂O₂. This induction requires the positive transcriptional activator OxyR, which directly senses oxidative stress. On the other hand, HPII is not peroxide inducible and its gene, katE, is transcribed at the transition from the exponential growth phase to the stationary growth phase by RNA polymerase containing the alternative sigma subunit σ^s, the product of the rpoS (or katF) gene.

Next to these catalases, E. coli expresses a two-enzyme H₂O₂-scavenging system, the alkyl hydroperoxide reductase (AhpR) system, which was initially characterized as rapidly reducing diverse organic hydroperoxides such as cumene- and t-butylhydroperoxide (18). Later, Niimura et al. (26) showed that the specific activities of the Salmonella enterica serovar Typhimurium AhpR homolog for the reduction of cumene hydroperoxide and H₂O₂ are within the same order of magnitude. One component of the AhpR system, AhpF, is a flavoprotein that shuttles reducing equivalents from NAD(P)H (with a strong preference for NADH) to AhpC, the actual peroxidase and a member of the peroxiredoxin family of thiol peroxidases. The ahpC and ahpF genes are organized in a two-gene operon, and transcription is OxyR controlled (38).

In E. coli, AhpR and HPI have discrete roles in scavenging H₂O₂. AhpR is the primary scavenger of endogenous H₂O₂. HPI contributes little when H₂O₂ concentrations are low, but it becomes the more effective scavenger when H₂O₂ concentrations are high or, presumably, when the absence of a carbon source depletes the cell of the NADH necessary for AhpR activity (36).

Analogously, catalases are believed to be indispensable in eukaryotic cells to protect compartments subjected to high H₂O₂ concentrations, e.g., the peroxisomal matrix and membrane (6), whereas the H₂O₂ resulting from the cellular metabolism of molecular oxygen is primarily scavenged by thiol-dependent peroxidases (13, 29). In animals, these peroxidases are fueled by either the glutathione or the thioredoxin redox cycle. The tripeptide-reduced glutathione (GSH; γ-l-glutamyl-l-cysteinyl glycine) is an abundant and ubiquitous low-molecular-weight thiol with a presumed role in many cellular processes, ranging from oxidative stress defense to cysteine storage (reviewed in reference 31). During these processes, GSH is often converted into its symmetrical disulfide form, oxidized glutathione (GSSG), and regeneration occurs via the NADPH-dependent action of glutathione reductase. Glutathione deficiency in higher eukaryotes is at the basis of several diseases caused by the degradation of mitochondria (8, 19), an observation which allowed us to conclude that H₂O₂ generated through respiration is scavenged inside the mitochondria by GSH peroxidase (GPx). Furthermore, the finding that glutathione-deficient yeast strains are hypersensitive to H₂O₂ points to the importance of glutathione-based peroxide removal in eukaryotes (10, 17).

Haemophilus influenzae is a strictly human commensal organism. As unencapsulated species, the bacterium is found in the upper respiratory tracts of up to 80% of healthy adults and, as encapsulated species, in 3 to 5% of normal individuals (25). Encapsulated strains are capable of invasive infections, including meningitis, pneumonia, and epiglottitis.

An acatalasemic H. influenzae Rd mutant was constructed by transposon-mediated inactivation of one structural gene, hktE. Consequently, hktE accounts for total catalase activity in H. influenzae cells (3). The hktE gene product, HktE, is strongly homologous to E. coli KatE. However, the H. influenzae catalase is downregulated in the stationary growth phase and is upregulated by exposure to H₂O₂. As such, hktE regulation is similar to that of katG (3). Consistent with findings in other bacteria, disruption of catalase production in H. influenzae causes hypersensitivity to high H₂O₂ concentrations (2). On the other hand, deletion of the hktE gene produces only a modest reduction in the ability to cause lethal sepsis after parenteral challenge and causes no change in the ability to colonize after intranasal inoculation in the infant rat model of infection (2). Consequently, the H. influenzae hktE gene is not of major importance for the process of colonization and invasive infection.

Because H. influenzae catalase mutants do not show any growth disadvantage either in vitro in rich medium or in vivo after intranasal inoculation in infant rats, it is anticipated that at least one other H₂O₂-scavenging system supports peroxide removal in H. influenzae. AhpR is not involved because neither an ahpC nor an ahpF homolog is apparent from the H. influenzae Rd genome sequence (7).

Recently, we reported that H. influenzae is naturally deficient in glutathione biosynthesis (42) and showed that H. influenzae instead imports glutathione from the growth medium. In this way, extracellular glutathione (supplemented as GSSG to prevent chemical reactions with the oxidizing test chemicals) protects cultures against methylglyoxal, S-nitrosoglutathione, and t-butylhydroperoxide toxicity. Interestingly, an open reading frame (HI0572) was isolated that encodes glutathione-based t-butylhydroperoxide removal. On the other hand, the presence of GSSG in the growth medium of H. influenzae NCTC 8143 cultures does not influence the sensitivity to exogenous H₂O₂ stress (42). We show here, however, that glutathione is a crucial component in the metabolism of H₂O₂ endogenously generated during aerobic growth.

MATERIALS AND METHODS

Media.

Mueller-Hinton broth (MH broth) was prepared from a dehydrate (Fluka at Sigma-Aldrich, St. Louis, Mo.) and autoclaved. MH medium broth was MH broth supplemented with Haemophilus test medium supplement (Oxoid, Hampshire, United Kingdom), containing V-factor (NAD) and X-factor (hemin), according to the manufacturers' instructions.

H. influenzae specific minimal medium (MIc medium) was prepared essentially as described by Herriott et al. (15). l-Cystine (Sigma-Aldrich) was added to a final concentration of 50 μM, and Haemophilus test medium supplement was added according to the manufacturers' instructions. GSSG-supplemented MIc medium [MIc(GSSG)] was prepared by replacing l-cystine with GSSG (Sigma-Aldrich) at the same final concentration. Both l-cystine and GSSG were added from a sterile concentrated stock solution. For anaerobic growth, the recipe for the preparation of MIc minimal medium was adjusted as follows: (i) sodium lactate was replaced by disodium fumarate (final concentration, 5 mM), and (ii) the final glycerol concentration was raised from 0.3 to 0.5%. This medium is referred to as anMIc or anMIc(GSSG) medium throughout the text, depending on whether l-cystine or GSSG was added, respectively; 1.8% agar was added to prepare anMIc and anMIc(GSSG) agar plates.

Bacterial strains and growth conditions.

Wild-type strain H. influenzae Rd was purchased from the American Type Culture Collection (Manassas, Va.). The construction of the H. influenzae catalase-negative strain AB2593 (Rd hktE::mini-Tn10Cm) was reported previously (2). Strain AB2593 was kindly provided by William R. Bishai (Department of Medicine, Division of Infectious Diseases, Johns Hopkins University School of Medicine).

Cultures were routinely grown at 37°C under anaerobic conditions in anMIc medium. Anaerobic cultures were prepared in a Coy chamber (Coy Laboratory Products, Inc.) under an atmosphere of 85% N₂-10% H₂-5% CO₂ in culture tubes that fit tightly into the cuvette holder of a Shimadzu 1240 Mini Single-Beam UV-VIS spectrophotometer (Shimadzu, Duisburg, Germany). The system enables cell density measurements without the need to transfer culture samples into regular cuvettes. The tubes were closed with a silicone stopper and were shaken outside the Coy chamber. As such, anaerobiosis was preserved, a finding confirmed in preliminary experiments by adding 0.002% resazurine to the growth medium (resazurine is a redox-sensitive dye added to media as a simple, qualitative indicator of the redox conditions of the media).

Tn10-linked mutations in hktE were confirmed by assaying catalase activity in whole cells as described earlier (12).

For aerobic shift experiments, precultures grown anaerobically overnight were diluted 1:100 in anMIc or anMIc(GSSG) medium to an optical density at 600 nm (OD₆₀₀) of ∼0.005 and then grown anaerobically to an OD₆₀₀ of ∼0.15. The silicone stopper was then removed in order to shift the cells into air. In the case of aerobic-shift experiments to monitor growth in the presence of nonenzyme H₂O₂ scavenging, pyruvate was added to the anMIc or anMIc(GSSG) cultures from a buffered sterile stock solution to a final concentration of 0.75%. In the case of aerobic-shift experiments to evaluate catalase activity in response to aerobiosis, cell extracts were prepared, and catalase activity was measured, as described previously (42), 20 min after the removal of the silicone stopper. These catalase activities were then compared to those derived from continuously anaerobically grown counterparts.

Heat-killed bacteria were prepared by incubating suspensions at 60°C for 1 h.

Disk diffusion.

Overnight anaerobically grown precultures were diluted 1:100 times in anMIc or anMIc(GSSG) medium to an OD₆₀₀ of ∼0.005 and then grown anaerobically to an OD₆₀₀ of 0.5 (late exponential phase). The following manipulations were performed inside the Coy chamber. Using a sterile cotton swab, cells were inoculated onto the entire surface of anMIc or anMIc(GSSG) plates. Round sterile filters (5.2-mm diameter) were placed in the center of the plates and spotted with either 5 μl of 3% H₂O₂ or 5 μl of 500 mM t-butyl hydroperoxide. The plates were placed in an anaerobic jar and incubated for 2 days at 37°C. The diameter of the zone of complete inhibition was recorded in millimeters. The experiment was performed in triplicate; mean values are reported.

H₂O₂ detection.

The Amplex red hydrogen peroxide/peroxidase assay kit (Molecular Probes, Eugene, Oreg.) was used to detect H₂O₂ according to the manufacturer's instructions. This assay can be performed either fluorometrically or spectrophotometrically. We chose the latter approach because the extinction coefficient of the Amplex red reagent oxidation product (resorufin) is high (54,000 cm⁻¹ M⁻¹), resulting in a detection limit of ca. 0.2 μM. The latter value is low enough to complete the experiments described here. Absorbance was measured at 563 nm by using a Shimadzu 1240 Mini Single-Beam UV-VIS spectrophotometer. The assay procedure was performed in a total volume of 400 μl. For each assay, no-H₂O₂ controls were monitored.

H₂O₂ scavenging by whole cells.

Overnight anaerobically grown precultures were diluted 1:100 times in anMIc or anMIc(GSSG) medium to an OD₆₀₀ of ∼0.005 and grown anaerobically to an OD₆₀₀ of 0.15. Cells were pelleted in a microcentrifuge, washed twice, and resuspended in room temperature phosphate-buffered saline (PBS) at an OD₆₀₀ of 0.15. H₂O₂ was added to the appropriate final concentration (see the figure legends). At intervals, 200-μl samples were removed, and the reactions were terminated (i.e., the cells were removed) by filtering the reaction mixtures with sterile Millex-GV13 0.22-μm-pore-size filter units (Millipore Products Division, Bedford, Mass.). H₂O₂ was then assayed by the Amplex red method.

Measurement of H₂O₂ accumulation in cell cultures.

Cultures were grown as described for the aerobic-shift experiments. Immediately before and at intervals after the shift to air, the OD₆₀₀s were measured, and 200-μl samples were removed. Cells were immediately removed from the samples by filtration with sterile Millex-GV13 0.22-μm-pore-size filter units. The cell-free culture media were then assayed for H₂O₂. H₂O₂ levels were also determined in sterile medium that was incubated at 37°C for an equivalent time period. These background levels, however, did not exceed the detection limit of ca. 0.2 μM. The rate of H₂O₂ production was normalized to the cytoplasmic volume of the suspended cells by using a standard ratio of 0.94-μl internal volume per 1 ml of H. influenzae at 1 OD₆₀₀. The latter value is derived from the assumption that H. influenzae cells can be modeled as approximate cylinders with a length of 1.2 μm and a radius of 0.25 μm; as such, the volume per cell is 0.236 × 10⁻¹⁵ liters. The number of cells present in 1 ml of a 1-OD₆₀₀ anMIc culture was determined by dilution plating to be 4.0 × 10⁹.

RESULTS

H. influenzae Rd catalase null mutant strain AB2593 exhibits a severe aerobic growth defect in minimal medium.

The H. influenzae strains examined here, the HktE-negative strain AB2593, and the wild-type isogenic parent strain Rd are identical to those studied by Bishai et al. (2), who used them to characterize catalase mutants of H. influenzae in terms of colonization and invasion abilities. This primary report illustrates that these strains share identical growth characteristics when grown aerobically in rich medium. We found that the growth curve of strain AB2593, grown in rich medium (MHs), is indeed similar to the growth curve obtained with wild-type cultures. No variation was observed when AB2593 cells were repeatedly subcultured in MHs medium. Also, no aberrations were detected for aerobic growth in minimal medium (MIc) inoculated with a stationary-phase MHs medium culture. However, these latter MIc cultures appeared to have lost one or more protective factors for surviving aerobic conditions. Indeed, second-generation AB2593 MIc subcultures showed severe growth defects when grown aerobically (Fig. 1), whereas growth in the absence of oxygen proceeded at wild-type levels (data not shown). Growth even stopped entirely in third-generation AB2593 MIc subcultures (Fig. 1).

FIG. 1.

Aerobic growth deficiency of acatalasaemic H. influenzae in minimal medium. Overnight precultures of wild-type H. influenzae Rd and mutant AB2593 (hktE) grown in rich medium (MHs) were diluted 1:100-fold in MIc medium [Rd (1), ▪; AB2593 (1), ▵], and growth was monitored at 600 nm. After 8 h of growth, these cultures were subcultured (1:100) in MIc broth, and growth was recorded [Rd (2), ▪; AB2593 (2), ▿). Again, 8 h later, growth was monitored of the third-generation subcultures in MIc broth [Rd (3), ▪; AB2593 (3), ⋄].

A clear representation of the aerobic growth defect of catalase-negative AB2593 cells could be obtained by shifting cultures, after repeated anaerobic subculturing in anMIc medium, from anaerobic to aerobic conditions (Fig. 2). Whereas Rd cultures immediately responded to the shift by the expected growth acceleration due to the initiation of aerobic respiration, AB2593 cultures responded by no more than a slight increase of the doubling time as a step toward complete growth arrest and death.

FIG. 2.

Both exogenous pyruvate and GSSG conceal the aerobic growth defect in minimal medium of H. influenzae AB2593. The assay was done by aerobic shift growth measurements. Overnight precultures of wild-type H. influenzae Rd and mutant AB2593 (hktE) anaerobically grown in anMIc broth were diluted 1:100-fold in anMIc medium (Rd, ▪; AB2593, ▿) and anMIc(GSSG) broth (Rd, ▴; AB2593, ⋄) (A) or in anMIc medium (Rd, ▪; AB2593, ▿) and in anMIc medium supplemented with 0.75% pyruvate (Rd, ▴; AB2593, ⋄) (B). Anaerobic growth was monitored. At an OD₆₀₀ of ∼0.15, the cultures were shifted to air (indicated by an arrow), and aerobic growth was recorded.

As mentioned before, catalases are ubiquitous among aerobic organisms, where they serve to detoxify the H₂O₂ that arises from exposure to oxygen. So, we first studied whether it is this H₂O₂ that poisons AB2593 grown aerobically in minimal medium. Therefore, we repeated the aerobic-shift experiments in the presence of sodium pyruvate, a nonenzyme scavenger of peroxides (41). We observed that catalase-negative H. influenzae cells grown in pyruvate-supplemented anMIc medium doubled at wild-type rates after the aerobic shift (Fig. 2). A question arises as to why endogenously generated H₂O₂ kills catalase-negative H. influenzae cells in MIc medium but fails to do so in rich MHs medium. Note that we here refer to endogenously generated H₂O₂. We indeed tested anMIc medium and anMIc(GSSG) medium for the nonenzymatic production of H₂O₂ and found that concentrations were below the detection limit (ca. 0.2 μM). For comparison, Luria-Bertani medium chemically generates ca. 1.2 μM H₂O₂ per min (14).

Exogenous glutathione conceals the aerobic growth defect in minimal medium of H. influenzae AB2593.

The fact that even fairly dense cultures of AB2593 died as a result of an aerobic shift (Fig. 2) rules out the possibility that a dilution effect causes growth inhibition in minimal medium. More likely, the aerobic growth defect is the result of the different composition of the growth media. One particular component that is not present in (an)MIc medium compared to rich medium is glutathione. When supplied to the growth medium in oxidized form, this thiol tripeptide was recently shown to strengthen the oxidative defense of naturally glutathione-deficient H. influenzae (42). We therefore repeated the aerobic-shift experiments in the presence of GSSG (Fig. 2) and observed that AB2593 cultures grown in glutathione-supplemented minimal medium are completely relieved from defective growth under aerobic conditions.

Effects of intracellular glutathione-based peroxide reduction on the metabolism of H₂O₂ produced during aerobic respiration.

In order to elucidate whether glutathione-deficient AB2593 cells indeed exhibit problems in detoxifying the H₂O₂ that is produced during aerobic respiration, we repeated our aerobic-shift experiment and monitored the H₂O₂ concentrations in the medium (Fig. 3). For all conditions tested, we measured higher H₂O₂ concentrations immediately after the aerobic shift. This most likely represents the adaptation time required to build up adequate antioxidant activity. Table 1 illustrates that the shift to oxidative metabolism caused a substantial 6.6-fold increase in catalase activity in glutathione-deficient H. influenzae Rd cells, indeed suggesting that, in response to the shift to air, H₂O₂ is generated at a rate greater than what could be metabolized to a steady-state level (i.e., <0.2 μM) by the antioxidant machinery present in anaerobically growing cells. This hypothesis, however, appears not to be valid for glutathione-supplemented cells. Note that the experiment illustrated in Table 1 only focuses on catalase activity, which, as described below, appears to be less efficient compared to glutathione-based peroxide removal. Thus, in response to the aerobic shift, induction of the latter system may be sufficient to alleviate the toxicity of respiratory-generated H₂O₂ to levels too low to induce the synthesis of catalase.

FIG. 3.

H₂O₂ production by glutathione- and catalase-deficient H. influenzae. Wild-type H. influenzae Rd and mutant AB2593 (hktE) were grown anaerobically in anMIc(GSSG) medium (Rd, ▪; AB2593, ▿) or anMIc medium (Rd, ▴; AB2593, ⋄) to an OD₆₀₀ of 0.15. The cultures were then shifted to air, and the H₂O₂ concentration of the medium was measured at various time points after the shift.

TABLE 1.

Effect of shift to oxidative metabolism on catalase activity monitored in H. influenzae Rd extracts derived from cultures grown in the presence or absence of GSSGa

Growth medium (growth condition)	Mean catalase (HktE)b activity (μmol of H2O2 decomposed/min/mg of protein) ± SEM
anMIc(GSSG) (anaerobiosis)	34.0 ± 2.3
anMIc(GSSG) (aerobiosis)	33.6 ± 4.5
anMIc (anaerobiosis)	34.4 ± 1.2
anMIc (aerobiosis)	228 ± 22

^aSee Materials and Methods for experimental details.

^bResults are expressed as the mean values from three independent experiments.

Twenty minutes after the shift to air, the growth of glutathione-supplemented AB2593 cultures continued, metabolizing H₂O₂ to concentrations below the detection limit. On the other hand, glutathione-deficient catalase-negative cells clearly experienced problems in controlling H₂O₂ production, as noticed by the steady H₂O₂ accumulation in the growth medium.

Once we had determined that both glutathione-based H₂O₂ removal and catalase activity are important during aerobic growth, we wondered whether these two systems represent the entire H₂O₂-scavenging activity of H. influenzae. Therefore, we set up a time-response experiment exemplifying the in vivo H₂O₂-scavenging activities of whole metabolically active cells (Fig. 4). We observed that, whereas glutathione-supplemented AB2593 scavenged 1.5 μM H₂O₂ as well as did its wild-type parent, the glutathione-deficient hktE mutant exhibited virtually no H₂O₂-scavenging activity. Also, heat-killed glutathione-supplemented AB2593 cells failed to remove H₂O₂ from solution. Thus, H. influenzae relies on two systems for the destruction of H₂O₂ generated through respiration: a glutathione-based system that requires metabolically active cells on the one hand and catalase on the other.

FIG. 4.

Glutathione-supplemented acatalasaemic H. influenzae and wild-type H. influenzae Rd remove low micromolar concentrations of extracellular H₂O₂ at similar rates. Cultures of Rd and AB2593 were grown anaerobically in anMIc(GSSG) medium (Rd, ▪; AB2593, ▿) and in anMIc medium (Rd, ▴; AB2593, ⋄), and resuspended in PBS at an OD₆₀₀ of 0.075. Heat-killed bacteria (○) were prepared by incubating the PBS suspension at 60°C for 1 h. H₂O₂ was added to a final concentration of 1.5 μM. The H₂O₂ concentration was measured at various time points after addition of H₂O₂ as described in Materials and Methods.

HktE and glutathione-based peroxidatic activities have discrete roles in scavenging H₂O₂.

Since we had now shown that glutathione-based scavenging activity alone is sufficient to scavenge H₂O₂ derived from the own metabolism, we sought to determine to what extent this type of peroxidatic activity is beneficial for H. influenzae at high, supranormal levels of H₂O₂. Therefore, we set up disk diffusion assays, which test the ability of strains to degrade high concentrations of H₂O₂ to a level that permits growth (Table 2). The glutathione-deficient wild type appeared to be somewhat more sensitive than its glutathione-supplemented counterpart. On the other hand, the glutathione-supplemented hktE mutant turned out to be hypersensitive, indicating that HktE represents the more effective scavenging system at supranormal H₂O₂ concentrations in H. influenzae.

TABLE 2.

Effect of glutathione deficiency on sensitivity to exogenous supranormal H₂O₂ concentrations in wild-type and acatalasaemic H. influenzaea

Strain (medium)	Zone of inhibition (mm)
Rd (MIc)	26
Rd [MIc(GSSG)]	21
AB2593 (MIc)	46
AB2593 [MIc(GSSG)]	33

^aEach assay was done by disk diffusion of 5 μl of 3% H₂O₂. See Materials and Methods for experimental details.

To verify this result in terms of kinetic behavior of the HktE and glutathione-based peroxidatic systems, we set up a concentration-response experiment exemplifying the rate of H₂O₂ decomposition achieved by a steady number of metabolically active whole cells (Fig. 5). The peroxidatic activity inside glutathione-supplemented AB2593 cells became saturated when extracellular H₂O₂ concentrations exceeded 25 μM. In contrast, HktE-expressing cells (glutathione-deficient wild type) were not saturated by the highest H₂O₂ concentrations tested, a finding consistent with the millimolar K_m values reported for other catalases (1, 20, 33). Furthermore, the H₂O₂ decomposition rates of whole glutathione-deficient wild-type cells at high extracellular H₂O₂ concentrations were much greater than those for whole cells that only relied on the glutathione-based peroxidatic system. On the other hand, at extracellular H₂O₂ concentrations of up to 5 μM, glutathione-based scavenging was significantly more efficient than that propelled by catalase.

FIG. 5.

Dependence of scavenging rate on extracellular H₂O₂ concentration. Rates of H₂O₂ decomposition were measured in dilute suspensions (0.075 OD₆₀₀) of H. influenzae Rd grown in anMIc medium (▪) and of AB2593 (hktE) grown in anMIc(GSSG) medium (▵). Rates were normalized to a value of 1.0 OD₆₀₀.

Calculation of endogenous H₂O₂ production during aerobic growth.

As shown in Fig. 3, substantial H₂O₂ accumulated in the medium of glutathione-deficient AB2593 cultures. These cultures were shown to be completely deficient in H₂O₂ scavenging at low micromolar concentrations, so that virtually all of the H₂O₂ that enters or is formed within these cells diffuses out without being scavenged (36, 37). For this reason, the measurement of excreted H₂O₂ is a valid proxy for measurement of endogenous H₂O₂ formation. Between 20 and 30 min after the shift to air, glutathione-deficient AB2593 cultures at an OD₆₀₀ of 0.15 accumulated 1.05 μM H₂O₂ (mean of three independent experiments). Using the relation that 1 ml of bacteria at an OD₆₀₀ of 1.0 contains 0.94 μl of cytoplasmic volume (see Materials and Methods), we calculated that aerobically grown log-phase H. influenzae forms H₂O₂ at a rate of ∼12.4 μM per s.

DISCUSSION

In 1983, Meister and Anderson (23) proposed that glutathione protects mammalian cells from oxidative damage. Two decades and several studies later, it has become definite that glutathione is indispensable for eukaryotic cells to deal with reactive oxygen species generated by aerobic respiration. For the most part, this results from the fact that GSH functions as a cofactor for peroxidatic H₂O₂ removal (16, 35). In contrast, Greenberg and Demple (11) reported that a glutathione-deficient mutant of E. coli K-12 has normal resistance to H₂O₂ and cumene hydroperoxide. Thus, even though intracellular GSH can efficiently scavenge free radicals nonenzymatically and can reach millimolar levels, it is not essential for protecting E. coli from oxidative damage. Moreover, no growth defects were reported for mutant acatalasaemic E. coli genetically inactivated for glutathione biosynthesis (32, 44).

We demonstrated here that the acatalasaemic H. influenzae mutant strain AB2593 requires glutathione in its growth medium in order to survive endogenous H₂O₂ stress associated with aerobic growth. In a previous study (42), we showed that H. influenzae, although it possesses glutathione metabolizing enzymes, is naturally glutathione deficient and that glutathione is acquired by importing the thiol tripeptide (either as GSH or as reducible symmetrical and asymmetrical [mixed] disulfide forms of GSH) from the growth medium. Therefore, since H. influenzae AB2593 is deficient in catalase, the restoration of wild-type growth rates in glutathione-supplemented (an)MIc medium could be the result of pure glutathione-based nonenzymatic H₂O₂ scavenging inside the cell (28). However, the fact that glutathione-supplemented AB2593 scavenges H₂O₂ at wild-type rates (Fig. 4) and the fact that the glutathione-based H₂O₂ removal attains a maximum initial velocity with increasing H₂O₂ concentrations (Fig. 5) indicate that intracellular GSH donates electrons to a powerful peroxidase.

Glutathione-dependent peroxidatic H₂O₂ removal requires a continuous flow of GSH. Glutathione reductase is the central enzyme in glutathione redox cycling, and this reductase exclusively exploits NADPH as a reductant. Heat-killed glutathione-supplemented AB2593 is not able to generate reducing power. This probably explains our observation that these cells were totally ineffective in H₂O₂ scavenging (Fig. 4). Consequently, the rate of glutathione-based H₂O₂ scavenging most likely depends on the rate at which metabolism generates NADPH. This may well explain why exogenous H₂O₂ stress results in a twofold increase in glucose-6-phosphate dehydrogenase (G6PD) activity (42), whereas G6PD levels remain unaltered in response to peroxides in E. coli. Indeed, G6PD is the first and rate-limiting enzyme of the pentose phosphate pathway, and for both lower and higher eukaryotes G6PD expression is regulated in order to maintain NADPH levels for the glutathione-based enzymatic pathway of peroxide removal (40). In E. coli, NADH is the preferred reductant for AhpR-dependent peroxidatic activity (26) and, it therefore appears that no G6PD induction is required in response to H₂O₂ stress.

Recently, we reported that the H. influenzae genome encloses a gene (ORF HI0572) encoding a novel GSH-dependent t-butylhydroperoxide reductase (Prx/Grx) (42). Naturally glutathione-peroxidase-deficient E. coli expresses a vast amount of GSH-dependent t-butylhydroperoxide reductase activity when transformed with a multicopy plasmid carrying the HI0572 locus. The Prx/Grx protein shares 63% sequence identity with the garA gene product of Chromatium gracile (30). Because the C. gracile peroxidase reduces both t-butylhydroperoxide and H₂O₂ at high rates (43), we anticipated that Prx/Grx might be the major peroxidase involved in H₂O₂ metabolism of H. influenzae. To examine this premise, we are currently generating both an H. influenzae HI0572 mutant and an H. influenzae HI0572 hktE double mutant. In the meantime, we recently obtained supportive biochemical evidence through the characterization of the H. influenzae Rd Prx/Grx peroxidase. We observed that H₂O₂ binds very efficiently to the GSH-dependent peroxidase, as can be demonstrated by deducing the K_m value from an empirical fit to the data given by the concentration-response curve shown in Fig. 5; note that the abscissa displays extracellular H₂O₂ concentrations and, therefore, the deduced K_m value cannot be interpreted as real Michaelis constant. The extracellular concentration at half-maximal rate was ca. 5.8 μM, which is very similar to the Michaelis constant (2.29 μM) obtained via in vitro kinetics of the Prx/Grx peroxidase (30).

As observed for E. coli (36), our results strongly suggest that aerobically grown log-phase H. influenzae scavenges the majority of endogenous H₂O₂ through one catalase and one peroxidase. Indeed, as reported for catalase and AhpR-deleted E. coli (36), glutathione-deficient H. influenzae lacking catalase is almost completely devoid of H₂O₂-scavenging activity (Fig. 4). Notably, in contrast to the fact that H₂O₂ scavengerless E. coli grows, albeit poorly, in rich and minimal media in air (36), we observed that glutathione-deficient AB2593 is completely unable to grow aerobically. So, the continual accumulation of H₂O₂ inside scavengerless aerobically growing cells appears to be more damaging to H. influenzae than to E. coli. In growing cells, the steady-state concentration of H₂O₂ depends on the rates of its formation and of its dissipation. Given that H₂O₂ removal is essentially absent in both strains, the growth difference could be the result of differences in H₂O₂ formation rates. We observed rates (∼12.4 μM per s) that are 1 order of magnitude higher than those reported by Gonzales-Flecha and Demple (ca. 1 to 2 μM per s) for E. coli (9). However, Seaver and Imlay (36) reported rates of H₂O₂ production in exponentially growing E. coli (∼14 μM) similar to those reported here for H. influenzae. This inconsistency is the result of differences in experimental procedure to measure H₂O₂ formation rates (36). Since our procedure is very similar to the one used by Seaver and Imlay (i.e., actual H₂O₂ formation rates are measured by using H₂O₂ scavengerless log-phase cultures grown at 37°C), we suggest that H. influenzae and E. coli cells generate equal amounts of H₂O₂ per s. Thus, the growth difference between H₂O₂ scavengerless H. influenzae and E. coli cannot be explained in terms of differences in H₂O₂ formation rates. Since GSH is a potent nonenzyme scavenger of reactive oxygen species and since it also functions as a cofactor for enzymes involved in oxidative stress defense, such as glutaredoxins and glutathione S-transferases, deprivation of intracellular GSH not only blocks GSH-dependent H₂O₂ removal but also prevents all other processes that require GSH in the H. influenzae cytoplasm. As such, glutathione-deficient AB2593 is subjected to additional, e.g., thiol-disulfide, stress (32) compared to H₂O₂ scavengerless E. coli, which explains why glutathione-deficient AB2593 is not viable under aerobic conditions.

For E. coli, it was proposed that AhpR and HPI have discrete roles in scavenging H₂O₂. AhpR scavenges low levels of H₂O₂, and HPI scavenges high levels of H₂O₂ (36). We obtained analogous results showing that glutathione-based peroxidatic activity is more effective at scavenging very low concentrations of H₂O₂, whereas HktE is the more effective enzyme at higher concentrations. Omitting GSSG from the media of H. influenzae NCTC 8143 cultures results in a twofold increase in catalase activity (42). Moreover, the catalase activity of glutathione-deficient H. influenzae NCTC 8143 is similar compared to the activity of glutathione-supplemented counterparts stressed with 50 μM H₂O₂ for 1 h. This implies that glutathione deficiency and micromolar H₂O₂ stress affect hktE gene expression to a similar extent (42). Moreover, Table 1 illustrates that aerobiosis caused a drastic 6.6-fold increase in catalase activity in glutathione-deficient H. influenzae Rd cells. We and others observed no such regulation of catalase for E. coli deficient in glutathione biosynthesis compared to the isogenic parent (28; data not shown). Thus, in H. influenzae, debilitation of the glutathione-based peroxidatic system causes catalase induction, which may be a compensatory response to provide enough, though less efficient, H₂O₂-scavenging activity. This compensatory response may well explain our observation illustrated in Fig. 4 that either catalase or glutathione-based peroxidase activities alone give the same rate of H₂O₂ scavenging by whole cells as when both are present. Compensatory interactions between catalase and peroxidase (AhpR) synthesis have been observed previously in a wide range of bacteria. In E. coli (36), Xanthomonas campestris (24), Bacteroides fragilis (34), and Pseudomonas aeruginosa (27), mutations in the peroxidatic system, AhpR, cause catalase induction.