Characterization of Escherichia coli Null Mutants for Glutaredoxin 2*

Three Escherichia coli glutaredoxins catalyze GSH-disulfide oxidoreductions, but the atypical 24-kDa glutaredoxin 2 (Grx2, grxB gene), in contrast to the 9-kDa glutaredoxin 1 (Grx1, grxA gene) and glutaredoxin 3 (Grx3,grxC gene), is not a hydrogen donor for ribonucleotide reductase. To improve the understanding of glutaredoxin function, a null mutant for grxB (grxB− ) was constructed and combined with other mutations. Null mutants forgrxB or all three glutaredoxin genes were viable in rich and minimal media with little changes in their growth properties. Expression of leaderless alkaline phosphatase showed that Grx1 and Grx2 (but not Grx3) contributed in the reduction of cytosolic protein disulfides. Moreover, Grx1 could catalyze disulfide formation in the oxidizing cytosol of combined null mutants for glutathione reductase and thioredoxin 1. grxB− cells were more sensitive to hydrogen peroxide and other oxidants and showed increased carbonylation of intracellular proteins, particularly in the stationary phase. Significant up-regulation of catalase activity was observed in null mutants for thioredoxin 1 and the three glutaredoxins, whereas up-regulation of glutaredoxin activity was observed in catalase-deficient strains with additional defects in the thioredoxin pathway. The expression of catalases is thus interconnected with the thioredoxin/glutaredoxin pathways in the antioxidant response.

Glutaredoxins are ubiquitous proteins that catalyze reductions of disulfides (protein-S-S) or mixed disulfides formed between proteins and glutathione (protein-S-SG) in a coupled system with GSH, NADPH, and glutathione reductase (1). Glutaredoxins can thus be envisaged as reductants of disulfides via GSH. To perform their redox reactions, glutaredoxins have evolved two catalytic mechanisms. The dithiol mechanism uses the two adjacent Cys residues of the active site (usually CPYC) to reduce protein-S-S or protein-S-SG substrates (2). The alternative monothiol mechanism (usual active site CGFS) can only reduce protein-S-SG substrates (2). A good in vitro substrate for glutaredoxin activity is the mixed disulfide com-posed of GSH and mercaptoethanol. This molecule forms spontaneously when ␤-hydroxyethyl disulfide (HED 1 or di-mercaptoethanol) and GSH are mixed and is a good substrate for both monothiol and dithiol glutaredoxins (HED assay) (1).
The numerous isoforms of glutaredoxins in different organisms seem to reflect involvement in distinct functions. Yeast, for example, contains five different glutaredoxins (3,4). The first two employ a dithiol mechanism (CPYC active site) (3), whereas the other three have a monothiol active site (CGFS) (4). Although all yeast glutaredoxins play a role in their response against oxidative stress (3,4), yGrx1 mainly participates in the protection against the superoxide anion, and yGrx2 against hydrogen peroxide (3). Null mutants for yGrx5 grow slowly and are highly sensitive to oxidative damage (carbonylation) or hyperosmotic treatment (4). Null mutants for yGrx2 and -5 or yGrx3, -4, and -5 are not viable, suggesting that a specific minimal monothiol glutaredoxin is essential for yeast survival (4).
Using HED as a substrate, Grx3 and especially Grx2 have higher catalytic activities in vitro compared with Grx1 (10). In the case of the mixed disulfide of arsenate reductase with glutathione as a substrate (14), Grx2 is a 100-fold more active than any of the other two glutaredoxins. Arsenate reductase catalyzes the reduction of arsenate to arsenite (14). Grx2 is thus very good in reducing mixed disulfides between a protein thiol and GSSG and is an abundant protein (5), and yet its biological function is still largely unknown. To elucidate more functions of the E. coli glutaredoxins, we constructed E. coli null mutants for the Grx2 gene (grxB) as well as different combinations of other genes of the glutaredoxin/thioredoxin pathways and characterized their properties.

EXPERIMENTAL PROCEDURES
Materials-␤-Hydroxyethyl disulfide was from Tokyo Kasei Kogyo Co. NADPH and glutathione reductase (yeast) was from Sigma. Oligonucleotides were synthesized by Amersham Biosciences Inc. Restriction enzymes, T4 DNA ligase, and Taq polymerase were from Promega. M9 media were supplemented with 50 g/ml Leu-Ile and 1ϫ basal medium Eagle's vitamin solution (Invitrogen).
Bacterial Strains and Plasmids-XL-1 blue and DH5␣ were used for plasmid propagation and cloning purposes. Other strains and plasmids of this study and their origin are listed in Table I.
Inhibition of Growth Assays-An aliquot (100 l) of cells grown overnight in M9 minimal medium was applied to 4 ml of minimal top agar, which was then poured on top of M9 minimal plates and allowed to solidify. Sterile porous discs (AB Biodisc) were placed on top of the top agar, and 5 l of stress compound was pipetted on the discs. The oxidants used were cumene hydroperoxide (1.5%), diamide (250 mM), hydrogen peroxide (0.2%), and menadione (580 mM). Inhibition of growth was measured after 16 -24 h at 37°C as the diameter of the disc of clear cells around the oxidant. Every plate contained four different oxidants, and average values for inhibition zones were calculated from three different plates.
Protein Determinations-In the measurement of total glutaredoxin activity in cell-free lysates from different mutants, total protein was measured in supernatants prepared as described (10) after sonication and centrifugation using the Bradford protein assay (15). In the estimation of carbonylation content, total protein was measured using intact cells (no sonication) by the method of Peterson (16).
Construction of a Null Mutant for Glutaredoxin 2-The E. coli gene for Grx2 (grxB) was deleted using a constructed recombinant 233 phage from the Kohara library in which the Grx2 coding sequence was replaced by a kanamycin resistance cassette (17). At a first stage, a 1.2-kb fragment flanking upstream grxB was amplified using primers gBdel F3 (5Ј-CCT GCT CAT CGA ATT CAT GCT GG TC-3Ј) and gBdel RC1 (5Ј-GAT CGT AAA TGT AAA GCT TCA CGT-3Ј), and a 1.2-kb downstream flanking region was amplified using primers gBdel F2 (5Ј-ACG TGA AGC TTT ACA TTT ACG ATC TGT TAT CAT CAA TGG  CGA TTT AAT CC-3Ј) and gBdel RC3 (5Ј-GGT CAT CAA TGG AAT TCA ATC GTG C-3Ј. pGEM-3Z (Promega) containing a 4.2-kb chromosomal fragment with grxB (10) was used as a template. The two different 1.2-kb fragments were purified on an agarose gel and were used as templates for primers gBdelF3 and gBdelRC3. The resulting 2.4-kb fragment (1.2-trgrxB-B1.2), which contained a reading frame encoding three amino acids in the place of grxB, had a HindIII site in the middle and one EcoRI site at each end. The fragment was digested with EcoRI and was cloned in vector pGEM-3Z in which the BamHI site had been previously eliminated after cutting by BamHI and then filling and religating. The new vector (pGEM-3Z-1.2-trgrxB-1.2) was digested with BamHI, the sticky ends were filled with T4 DNA polymerase, and the 1.4-kb kanamycin resistance cassette from vector pUC-4K (Amersham Biosciences Inc.) was ligated after it had been previously excised with PstI and filled in with T4 DNA polymerase. The new plasmid (pGEM-3Z-1.2-Kan-1.2) was used to transform strain MC1061, and the grxB::Kan deletion was transferred to phage 233 according to a described method (17), which resulted in the formation of MC1061 grxB::Kan. The absence of grxB in MC1061 was confirmed by Southern blot of total chromosomal DNA using amplified grxB as a probe. The grxB::Kan genotype was transferred to DHB4 and derivatives by P1 transduction (18). As some of DHB4 derivatives also contained Kan resistance cassettes, Tn 10 (19) was placed close to the grxB::Kan mutation to create grxB::Kan . . . Tc. P1 lysogens from this genotype were used to transduce DHB4 derivatives selecting for Tc resistance. All Tc resistance transductants from this procedure were screened by PCR to confirm the presence of the grxB::Kan genotype using internal primers for the Kan gene and primers located upstream/downstream grxB. Strains DHB4 and UM1 (19) were used to make combinations of different null mutants by P1 transduction (18).
Construction of Grx2 Vectors with Arabinose Promoter-The gene encoding Grx2 was amplified using primers G2FNde (Ј5-TGG AGG AGT CAT ATG AAG CTA TAC-3Ј) and G2EcoRC1 (5Ј-CGC GGC GGG GAA TTC TTA AAT CGC-3Ј) to obtain a PCR product encoding wild type Grx2. Primers G2-C12S (5Ј-ATG AAG CTA TAC ATT TAC GAT CAC TGC CCT TAC AGC CTC AAA GC-3Ј), and G2EcoRC1 gave a primary PCR product to be used as a template for primers G2Fnde and G2EcoRC1, finally giving the Grx2C12S mutant gene. C9S/C12S (5Ј-ATG AAG CTA TAC ATT TAC GAT CAC AGC CCT TAC AGC CTC AAA GC-3Ј) and G2EcoRC1 gave the template amplified later with G2Fnde and G2EcoRC1 to give the PCR product encoding Grx2 C9S/ C12S. pGEM-3Z (Promega) containing a 4.2-kb chromosomal fragment with grxB (10) was used as a template. All PCR products were digested by EcoRI-NdeI and ligated to vector pISC1 containing an arabinoseinduced promoter (21). Strain gor Ϫ grxA Ϫ B Ϫ C Ϫ was transformed with the resulting vectors, plated on LB plates supplemented with ampicillin, 100 g/ml, and grown overnight at 37°C. Isolated colonies were streaked on M9 plates supplied with Leu-Ile-Met, 50 g/ml, and 100 g/ml ampicillin and grown for 2 days. To examine the effect of the Grx2 mutants on cell viability, transformed strains were streaked on M9 plates supplemented with Leu-Ile, ampicillin, and 0.1% arabinose (no reduced sulfur present). Growth of the strains on these plates after 2 days at 37°C was taken as evidence that the strains could grow on M9 minimal medium.
Determination of Glutaredoxin Activity (HED Assay)-A fresh mixture of 1 mM GSH, 0.2 mM NADPH, 2 mM EDTA, 0.1 mg/ml bovine serum albumin, and 6 g/ml yeast glutathione reductase was prepared in 100 mM Tris-Cl, 2 mM EDTA, pH 8.0 (1). To 500 l of this mixture in semi-micro cuvettes, HED was added to a final concentration of 0.7 mM. After 3 min, supernatant of total cell lysate of different knockouts was added to the sample cuvettes, and the decrease in A 340 was recorded the following 5 min using a Zeiss PMQ3 spectrophotometer. Activity was expressed as milli-⌬A 340 OD units/min. Most measurements were performed in a final volume of 100 l using a Molecular Devices Thermomax microplate reader. Values of ⌬A 340 were then multiplied by a factor of 4.3 to give the ⌬A 340 of a cuvette with a path length of 1 cm. All activity measurements were made at 25°C.
Expression and Measurement of Intracellular Alkaline Phosphatase Activity-DHB4 strain and derivatives transformed with plasmid pAID135 encoding leaderless alkaline phosphatase (AP) (22) were grown to early log phase (A 600 ϭ 0.3-0.5) in M63 minimal medium containing 0.2% glucose, 100 g/ml ampicillin, and 50 g/ml of all amino acids apart from cysteine and methionine (unless methionine was supplied to 50 g/ml). Isopropyl-1-thio-␤-D-galactopyranoside was added to 1 mM final concentration, and cells were grown further for 1 h. The culture was then harvested, and the total AP activity of cell extracts was estimated as described previously (21).
Measurements of Carbonylation of Intracellular Proteins-Cells were grown in LB and exposed to H 2 O 2 at their exponential phase of growth (A 600 ϭ 0.4 -0.6) or at stationary phase (overnight cultures). One ml of cells was placed in an Eppendorf tube and exposed to H 2 O 2 for 1 h at room temperature. Cells were then harvested, resuspended in water, and kept at Ϫ80°C. Cells were later lysed in SDS and derivatized with 2,4-dinitrophenyl hydrazine (23) for 30 min. After neutralization, samples were electrophoresed in 12.5% acrylamide and Western blotted using as primary antibody a 1:1000 dilution of an antibody raised to 2,4-dinitrophenyl (DAKO). The blots were processed further with a horseradish peroxidase-conjugated secondary antibody and finally developed with the PerkinElmer Life Sciences chemiluminescence kit (exponential phase of growth). For cells in the stationary phase of growth, the secondary antibody was an alkaline phosphatase conjugate. Contrast was adjusted in each final picture with the help of identical samples run as internal standards for each of the two blots. The total amount of protein per well was 1.25 g for the exponential phase experiments and 2.5 g for the stationary phase.
Measurement of Catalase Activity-These assays were performed in bacterial extracts following the production of oxygen from hydrogen peroxide, using an oxygen electrode. Cells were grown in M9 medium and harvested by centrifugation at different growth stages. Centrifuged cells were washed once with phosphate buffer (20 mM potassium phosphate, pH 6.8, containing 5 mM EDTA) and stored at Ϫ20°C. To prepare crude extracts, cells were thawed, resuspended in phosphate buffer containing protease inhibitors (1 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride), and broken by passage through a French pressure cell at 20,000 lb/in 2 . Cell suspensions were centrifuged for 15 min at 15,000 ϫ g followed by 60 min at 144,000 ϫ g. The clear supernatants were decanted and used to assay enzymatic activity. Catalase activity was measured polarographically by following O 2 production with a Clark-type oxygen electrode as described previously (23,24). The reaction mixture contained 20 mM H 2 O 2 , 100 mM potassium phosphate buffer, pH 6.3 to determine total HPI and HPII activity or pH 9.3 to determine HPII activity. The activity level of each one of the catalase isozymes was calculated based on their pH optima. At pH 6.3 both enzymes are at their maximal activity, whereas at pH 9.3, HPII is maximally active and HPI is only at 10%. The reaction was run at 30°C for 1 min using only the initial linear rates to calculate catalase activity. One unit of enzyme activity corresponded to the decomposition of 1 mol of H 2 O 2 /min at an initial concentration of 20 mM H 2 O 2 .

Construction of a Null Mutant for Glutaredoxin 2-
To generate a null mutant for grxB, the Grx2 coding sequence was replaced by a Kan resistance cassette, and the absence of grxB in the viable MC1061 strain was confirmed by Southern blotting. The grxB::Kan genotype was further transferred to other strains and combined with other known mutations to create a series of mutants (Table I). Strain DHB4 was generally used as the genetic background to generate null mutants for Grx2 and the other glutaredoxins. This strain has been used extensively previously to assess conditions affecting the folding of AP in the cytosol (22,25,26). Strains CSH7 and UM1 were also used to compare sensitivity to different oxidants.
Growth Properties-To examine the importance of Grx2 for the growth of E. coli, we compared the growth rates of different null mutants in rich and minimal media. The wild type and and gor Ϫ trxB Ϫ all had very similar growth rates in the rich LB medium (not shown), suggesting that E. coli can compensate for major perturbations of the thioredoxin/glutaredoxin pathways when supplied fully with nutrients. In M9 minimal medium, the wild type and null mutants for grxB, trxAgrxB, grxABC, gor, gor-grxB, trxB, trxBgrxB, trxBgrxBC, and trxBgrxC grew also at similar rates. However strains trxA Ϫ grxC Ϫ , gor Ϫ trxA Ϫ , and gor Ϫ trxA Ϫ grxB Ϫ grew very slowly and reached lower final levels of growth ( Fig. 1) in M9. A possible explanation is that lack of Trx1 combined with partial neutralization of the glutathione/glutaredoxin pathway reduced the ability of cells to utilize sulfate. For example, trxAgrxA cells cannot grow at all in minimal media as they cannot reduce SO 4 2Ϫ to SO 3 2Ϫ (12). Null mutants for gorgrxABC could barely grow on solid minimal medium unless supplemented with sulfite, methionine, or cysteine. The term "barely" means that very small colonies were observed after 2 days at 37°C and that these colonies did not increase in size after extended incubation at 37°C. However, null mutants for gorgrxAB, gorgrxAC, or gorgrxBC, which have a single functional glutaredoxin, grew well on minimal plates lacking reduced forms of sulfate (data not shown). Therefore at least one glutaredoxin was essential for the reduction of sulfate to sulfite in gor Ϫ mutants. gor Ϫ grxA Ϫ B Ϫ C Ϫ strains transformed with plasmids encoding wild type Grx2 and Grx2 C12S (monothiol) could grow in minimal liquid M9 better than the Grx2 C9S/C12S (inactive double mutant) transformants (Fig. 2). These results demonstrate that Grx2 and its monothiol mutant could regulate the activity of PAPS reductase.
Levels of Glutaredoxin Activity-Levels of total glutaredoxin activity were determined for different strains using the HED assay (1). Less than 1% glutaredoxin activity remained in strain grxA Ϫ B Ϫ C Ϫ (Fig. 3), indicating the presence of another glutaredoxin(s). Grx2 contributed to 80% of total glutaredoxin activity. A similar contribution to glutaredoxin activity from Grx2 was calculated previously from mutants and extrapolated to the wild type strain using the kinetic data for the pure protein (5). Total glutaredoxin activity was not elevated in null mutants for trxB Ϫ or at 42°C or between strains growing in rich or minimal media. Sensitivity of Strains to Oxidative Stress-To examine the role of Grx2 and other redoxins for survival in oxidative stress, the inhibitory effect on the growth of different strains exposed to different oxidants (27) was examined in M9 medium. The first set of experiments (Table II) was performed using the strain employed for the expression of intracellular AP (Table I). Almost all trxA Ϫ strains showed increased sensitivity to the organic peroxide cumene hydroperoxide. The exception was strain trxA Ϫ grxC Ϫ , which was significantly less sensitive than trxA Ϫ . This was a somewhat unexpected result, as grxC Ϫ alone was more sensitive to cumene hydroperoxide than the wild type. Null mutants for trxB Ϫ and its derivatives were more resistant than the wild type. trxB Ϫ strains were more sensitive to the thiol oxidant diamide, whereas all gor Ϫ derivatives and null mutants for ␥-glutamylcysteine synthetase (gshA Ϫ strain, lacking GSH) were very sensitive, especially the gor Ϫ trxA Ϫ strain. Furthermore, trxA Ϫ gor Ϫ , trxA Ϫ gor Ϫ grxB Ϫ , and trxB Ϫ grxC Ϫ strains were more sensitive to hydrogen peroxide, whereas the trxA Ϫ and trxA Ϫ gor Ϫ grxC Ϫ mutants were less sensitive than the wild type. Sensitivity to the superoxide anion generator menadione was increased in strains grxC Ϫ , gor Ϫ grxA Ϫ , trxA Ϫ gor Ϫ grxC Ϫ , and gshA Ϫ , whereas it decreased slightly in gor Ϫ . The grxC Ϫ null mutant in the trxA Ϫ background decreased the sensitivity to menadione as compared with trxA Ϫ alone.
In another set of experiments (Table III), we examined the effect of oxidants on null mutants for the catalase genes combined with mutants for the glutaredoxin system. The trxB Ϫ strain was more resistant to cumene hydroperoxide, even in combined null mutants with the kat genes. Sensitivity to the thiol oxidant diamide was enhanced in combined null mutants for katEGgrxA, katEGgrxB, and katEGgrxC and was very distinct in the katE Ϫ G Ϫ gor Ϫ strain. trxB Ϫ was more sensitive to diamide in the wild type background than was katE Ϫ G Ϫ . As expected (28), all katE Ϫ G Ϫ strains had increased sensitivity to H 2 O 2 . In this genetic background, Grx1 and Grx2 (but not TrxR or Grx3) contributed to the defense against H 2 O 2 . Sensitivity to menadione was increased in the trxB Ϫ , katE Ϫ G Ϫ , katE Ϫ G Ϫ grxA Ϫ , katE Ϫ G Ϫ gshA Ϫ , and katE Ϫ G Ϫ trxB Ϫ strains. Table IV summarizes the results of the experiments with the four different oxidants for all strains.
Disulfide Bond Formation in the Cytoplasm-To assess the contribution of glutaredoxins in reducing cytosolic disulfides, levels of cytosolic AP were measured in null mutants for the individual glutaredoxins. High levels of intracellular AP activity are considered an indication of an oxidizing cytosol (22,25,26). The wild-type and null mutants for grxB and grxABC had similar levels of intracellular AP activities (Fig. 4) suggesting that the lack of glutaredoxins in E. coli with intact thioredoxin/ thioredoxin reductase (TrxR) and GSH reducing pathways did not significantly affect the intracellular redox environment. The slightly higher levels of AP activity for the grxC Ϫ strain have been reported previously (22). gor null mutants combined with grxA and grxB (but not grxC) knockouts gave significantly higher AP activity levels than the gor knockout alone (Fig. 4). Therefore, a lack of Grx1 or -2 resulted in increasing formation of intracellular disulfides in gor Ϫ derivatives.
The levels of intracellular AP activity were higher in null mutants for gortrxA (Fig. 5) (22). gortrxAgrxB and gortrxAgrxC null mutants had marginally lower levels of intracellular AP activity, whereas gortrxAgrxA null mutants had significantly lower levels (Fig. 5). The addition of reduced sulfur in the form of Met did not seem to influence the levels of intracellular AP activity in any of the gor Ϫ trxA Ϫ derivatives (Fig. 5). All mutants had similar to identical amounts of AP expression as determined by Western blots (data not shown). Differences in AP levels therefore represent alterations in the oxidative folding of AP.
Measurements of Carbonylation of Intracellular Proteins-Western blots with total cell lysates showed that the lack of Grx2 rendered cells more sensitive to carbonylation damage induced by hydrogen peroxide than did lack of Grx1 or Grx3 (Fig. 6A). Lack of GSH (gshA null mutant) gave the highest carbonylation effect (Fig. 6B). No significant increase in carbonylation was observed in extracts from cells lacking Grx1, Grx3, TrxR, or glutathione reductase (Fig. 6, A and B). The protective role of Grx2 was markedly enhanced at the stationary phase of growth (Fig. 6C).
Total Glutaredoxin (HED) Activity in Catalase-deficient Strains-Strains CSH7 (wild type parental), UM1 (catalasedeficient), and null mutant derivatives of these strains for gor, gshA, or trxB were grown until stationary phase in LB. Cells were then harvested, lysed by sonication, and centrifuged, and the HED reducing activity of their culture supernatants was determined. Total glutaredoxin activity was similar in null mutants for single genes. However, in combined null mutants for catalases and TrxR or GSH or glutathione reductase, glutaredoxin activity was elevated significantly (Fig. 7).
Catalase Activity Measurements-As shown previously (Fig.  3) the trxA Ϫ grxA Ϫ B Ϫ C Ϫ mutant had less than 1% of total glutaredoxin activity compared with the wild type parental strain. It is assumed that this strain was able to survive by using GSH and TrxR/Trx2 for the maintenance of a reducing cytosol (25). To examine whether other pathways were affected, we measured the activity levels of the two catalases. In Fig. 8 it can be seen that HPI levels were dramatically elevated in the mutant strain. DISCUSSION Glutathione and glutaredoxins constitute a major supply of reducing power in almost all living cells. The role of glutaredoxins in the reduction of ribonucleotides has been established for one class of glutaredoxins that share the classical thioredoxin/glutaredoxin fold (29). The presence of five different FIG. 1. Growth properties of gor ؊ and gor ؊ trxA ؊ strains. The innoculi for these experiments were cells grown overnight in M9 minimal medium supplied with Leu-Ile, 50 g/ml. Cells were then diluted to an A 600 of ϳ0.1 and grown in fresh medium at 37°C. E, wild type (WT; DHB4); OE, gor Ϫ ; ࡗ, gor Ϫ grxB Ϫ ; ϫ, gor Ϫ trxA Ϫ ; Ⅺ, gor Ϫ trxA Ϫ grxB Ϫ ; *, trxA Ϫ grxC Ϫ . yeast glutaredoxins, or even isoforms of "classical" glutaredoxins with largely different catalytic properties (e.g. human glutaredoxin 2 (30)), suggests that reduction of ribonucleotides is only part of the discovered glutaredoxin functions. In E. coli, Grx1 and Grx3 belong to the classical superfamily in terms of fold, but Grx2 has different catalytic properties and a threedimensional structure that brings another dimension to the field; proteins with no significant amino acid homology may have identical folds and perform similar functions. For example, E. coli Grx2 folds like a human class GSH S-transferase (31), the human glutathione transferase omega 1 (32), the mouse glutathione transferase theta-like stress response protein (p28) (33), and the human CLIC1 (chloride intracellular channel 1) (34), whereas at the same time these proteins have very limited homology. The functions of these proteins have as a common denominator the response to stress. From this point of view E. coli Grx2 is the paradigm of a structure/function concept, "the Grx2 structural superfamily," found in many different organisms. We therefore consider that the investigation of the function of Grx2 is of general importance. In view of its high level and amount of catalytic activity, E. coli Grx2 would appear to be the enzyme that performs the most oxidoreductions related to GSSG-protein mixed disulfides.
Glutaredoxins and the Formation of Cytosolic Disulfides-The contribution of glutaredoxins in reducing cytosolic disulfides has not been investigated thoroughly. Instead, the thioredoxin system has been well established as the major factor of an overall "reduced" cytosol (22). Knockouts for TrxR (trxB Ϫ strains) have disulfide-promoting cytosols as manifested by the levels of intracellularly expressed AP (22,25,26). This has been explained in terms of thioredoxins constantly being in an oxidized state because of lack of TrxR, with thioredoxins thus promoting disulfide bond formation (26). This finding implies that in the trxB knockouts, the thioredoxin system overrides the reducing effect of the glutaredoxin system. Null mutants for all the genes of the thioredoxin pathway (trxA Ϫ B Ϫ C Ϫ ) have low levels of cytosolic AP activity (26) showing that the glutathione/glutaredoxin pathway can fully compensate for the lack of the thioredoxin system. We present further evidence for the FIG. 2. A, growth of DHB4gorgrxABC on M9 minimal plate. Strain DHB4gor Ϫ grxA Ϫ B Ϫ C Ϫ transformed with pISCGrx2 (top right), pISCGrx2C12S (top left), or pISCGrx2C9S/C12S (bottom) was grown on M9 medium, supplied with Leu-Ile-Met, 50 g/ml. Single colonies were then streaked on the same medium lacking Met but supplied with 0.2% arabinose and left at 37°C for 48 h. B, growth of DHB4gor Ϫ grxA Ϫ B Ϫ C Ϫ on M9 minimal medium. Strain DHB4gorgrxABC transformed with pISCGrx2 (thick solid line), pISCGrx2C12S (thick dashed line), or pISCGrx2C9S/C12S (thin dashed line) was grown on LB medium with ampicillin, 100 g/ml, to stationary phase and then washed two times with water and resuspended to equal A 600 in M9 medium supplied with Leu, Ile, and ampicillin, 100 g/ml (no Met), and 0.1% arabinose. The cultures were then moved to a rotary shaker at 37°C.

FIG. 3. Levels of glutaredoxin activity in DHB4 derivatives.
Values correspond to the total glutaredoxin activity of sonicated crude extracts from overnight cultures of different genetic backgrounds. One unit of glutaredoxin activity corresponds to 1 mol of NADPH oxidized/ min using the HED assay for the measurement of GSH-disulfide oxidoreductase activity (1). Mean values shown represent the mean of three measurements for the same lysate, and error bars represent the corresponding S.D.

Sensitivity of DHB4 strains to different oxidants
The diameters of the inhibition zones represent the mean of at least three different measurements.
reductive function of individual glutaredoxins. Levels of intracellular AP activity increased in gor strains lacking grxA and grxB (but not grxC). In the more oxidizing environment of the gor Ϫ trxA Ϫ strain, Grx2 and Grx3 had a marginal effect in contributing to a reducing cytosol. However, lack of Grx1 in null mutants for gortrxA resulted in a significant decrease of intracellular disulfides suggesting that Grx1 contributed to the formation rather than the reduction of cytosolic disulfides in FIG. 6. Measurements of intracellular AP activity in gor ؊ trxA ؊ derivatives. Conditions were the same as described in the legend for Fig. 4 4. Measurement of intracellular AP activity in different mutants. Cells transformed with plasmid pAID135 encoding leaderless AP (21) were grown in M63 medium, Leu-Ile-Met, 50 g/ml, 100 g/ml ampicillin at 37°C to an A 600 of 0.3-0.5. Cells were then induced for 1 h with 1 mM isopropyl-1-thio-␤-D-galactopyranoside, and intracellular AP activity was measured as described (21). The results represent the means of at least three different cell lysates. Error bars represent the standard deviation.

FIG. 5. Carbonylation in null mutants for catalases and individual glutaredoxins.
Different null mutants growing exponentially in LB were exposed to H 2 O 2 for 1 h at room temperature. Cells were harvested and their lysates analyzed after derivatization with 2,4dinitrophenylhydrazine and Western blotting with an antibody to the 2,4-dinitrophenyl moiety. Results are shown for the different null mutants in exponential (A and B, chemiluminescence detection) or stationary phase (C, detection by alkaline phosphatase conjugates). the genetic background of the gor Ϫ trxA Ϫ strain (Fig. 5). As the pathways for the reduction of sulfate and ribonucleotides involve enzymes with redox active dithiols, which generate a large number of oxidizing equivalents, we supplemented cultures with reduced sulfur in the form of Met to relieve a potential oxidative burden by PAPS reductase, especially because the gor Ϫ trxA Ϫ strains grew very poorly on sulfate. The addition of varying amounts of Met to the media of null mutants for gortrxA and derivatives did not alter the levels of intracellular AP, suggesting that PAPS reductase does not contribute to intracellular disulfide bond formation. In trxB null mutants, addition of Cys led to a decrease of intracellular disulfides. 2 It therefore seems that the reason for increased intracellular disulfides in the two genetic backgrounds is different. We suggest that Grx1 catalyzes the formation of intracellular disulfides in the gor Ϫ trxA Ϫ mutants. Enhancement of disulfide bond formation by overexpression of Grx1 (or Trx1) has been reported in the oxidizing genetic background of trxB Ϫ gor Ϫ strains (35).
Reduction of PAPS Reductase by Glutaredoxins-An unexpected finding was that gor Ϫ grxA Ϫ B Ϫ C Ϫ stains did not grow well on minimal medium plates unless supplemented with some form of reduced sulfur (SO 3 2Ϫ , Met, or Cys). Reduction of SO 4 2Ϫ to SO 3 2Ϫ is catalyzed in E. coli by PAPS reductase (13). The enzyme has a single Cys at its C terminus, which becomes oxidized after reduction of PAPS. Trx1, Trx2, and Grx1 can reduce the disulfide of the PAPS reductase, whereas Grx2 and Grx3 cannot (13). Therefore, E. coli should not need any glutaredoxin to reduce sulfate to sulfite; thioredoxins should be able to compensate fully. Our results suggest that in conditions with perturbed reduction of GSSG (gor Ϫ strain), glutaredoxin activity is essential to maintain viability. Because gor Ϫ grxA Ϫ B Ϫ C Ϫ cells could be rescued by monothiol Grx2, it seems that a mechanism for the activation of PAPS reductase involves a thiol that when forming a mixed disulfide with GSSG renders the enzyme inactive. In other words, PAPS reductase is subject to a mixed disulfide mechanism for the regulation of its activity. This mechanism would render the enzyme inactive in higher GSSG concentrations. Apparently, the remaining glutaredoxin activity in grxA Ϫ B Ϫ C Ϫ strains cannot restore the activity of PAPS reductase. Our data suggest the involvement of all three glutaredoxins in the reduction of PAPS reductase, presumably via a monothiol mechanism.
Glutaredoxins and Oxidative Stress-Significant carbonylation of cytosolic proteins occurred in knockouts for grxB, or gshA but not for the grxA, grxC, gor, or trxB genes. Grx2 and GSH therefore constitute the major thiol system for the protection of proteins against H 2 O 2 -induced carbonylation. Grx2 is also a potent antioxidant against dopamine-induced oxidative stress in rat neurons, preventing their apoptosis by activating the binding activity of NF-B via Ref-1 (36). Grx2 could actually affect both the Ras/phosphoinotiside 3-kinase/Akt/NF-B and the JNK1/JNK2/AP1 cascades (37). The antioxidant role of Grx2 was further confirmed in E. coli with assays measuring inhibition of growth after exposure to oxidants (Table VI). The protection of Grx2 against hydrogen peroxide was observed only in catalase negative strains and was not in proportion to its increased protective effect against protein carbonylation. This is not unexpected, as increased protein carbonylation but with no direct link to cell viability has been described for cells entering stationary phase (38). Transcription of grxB was not increased after treatment with H 2 O 2 (39) but was increased after exposure to acid stress (40), suggesting that the antioxidant effects of Grx2 observed in this work may not constitute its primary function.
Grx1 did not protect proteins from carbonylation, but its absence decreased the survival of cells against H 2 O 2 . In comparison, the absence of Trx1 improved survival. Exposure of cells to H 2 O 2 activates the transcription of antioxidant genes such as katG (catalase HPI), grxA, trxC (Trx2), gor, and ahpCF (alkyl hydroperoxide reductase) via OxyR (39,41,42), the activity of which is also controlled by the cellular redox environment (43). The improved viability of the trxA Ϫ after exposure to H 2 O 2 (or cumene hydroperoxide for trxB Ϫ ) could be explained in terms of up-regulation of other antioxidant genes via OxyR (39).
Grx3 participated in the detoxification of organic peroxides and in the defense against the superoxide anion radical (menadione). Transcriptional analysis of the response to superoxide anions has not identified grxC as an involved gene (44). These findings demonstrate the complexity of the biological systems and the need for thorough investigations to determine the function and transcriptional regulation of individual genes.
Apart from the catalase genes, all of the other genes examined provided protection against diamide. Diamide, therefore, is an oxidant of thiols. This finding is in agreement with the thiol-promoting role of the glutaredoxin and thioredoxin systems. Evidence for Additional Glutaredoxin(s)-Significant glutaredoxin activity was measured in crude extracts from grx-ABC null mutants, indicating the presence of an additional glutaredoxin. Similar low levels of glutaredoxin activity were determined in trxAgrxABC null mutants (data not shown). A protein with low HED oxidoreductase activity, which could represent the remaining glutaredoxin activity in the grxABC knockouts, is currently being characterized in our laboratory. 3 Glutaredoxins and Catalases-Strain trxA Ϫ grxA Ϫ B Ϫ C Ϫ was used to identify potential responses to stress other than by the up-regulation of Trx2. trxB Ϫ strains, for example, have increased transcription of the catalase gene katG (45). E. coli has two catalases: HPI, encoded by katG, and HPII, encoded by katE (28,46). Catalases dismutate H 2 O 2 to water and molecular oxygen to avoid the formation of the extremely toxic hydroxyl radicals from H 2 O 2 . HPI is readily induced by hydrogen peroxide, whereas HPII is up-regulated in stationary phase (35,36). This work has shown increased activity of HPI in the trxA Ϫ grxA Ϫ B Ϫ C Ϫ strain. Such a response may represent a compensating mechanism to reinforce the antioxidative capacity of the cells. HPI is induced by H 2 O 2 , and higher cellular activities of this enzyme suggest that the steady state concentrations of hydrogen peroxide in the cells are elevated. In catalase-deficient strains with intact thioredoxin/glutaredoxin pathways, there was a slight increase in total glutaredoxin activity. However in catalase-deficient strains (UM1) lacking TrxR or GSH, a significant up-regulation of total glutaredoxin activity was observed. These results and previous findings (31) show the existence of a concerted response of catalases and the thioredoxin and glutaredoxin systems to oxidative stress. Further studies are needed to identify the electron acceptors for Grx2 and Grx3 and to analyze their participation in the antioxidant mechanisms of the cell.