Roles of Thioredoxin Reductase during the Aerobic Life of Lactococcuslactis

L. lactis strain MG1363 (13) or MG1363 bearing the erythromycin resistance-encoding plasmid pIL252 (43) was used as the wild type (WT). Strains were grown in M17 medium (Difco) supplemented with 1% glucose (GM17). Erythromycin was added to a final concentration of 2.5 μg/ml. Cultures were inoculated from stocks frozen at −80°C into fresh GM17 at 30°C and grown to an optical density at 600 nm (OD₆₀₀) of 0.2. One hundred-fold dilution with fresh GM17 saturated with air was then used for the test cultures. Unless indicated otherwise, cells were grown under static conditions at 30°C and the OD₆₀₀ was measured after 17 h of growth. Reducing agents, metabolic compounds, and enzyme added to cultures during growth were used at the following concentrations: dithiothreitol (DTT), 0.5 mM; Cys, 1 mM; GSH, 1 mM; pyruvate, 20 mM; catalase, 5,000 U. Growth conditions for Northern blot assays and two-dimensional (2D) gel electrophoresis were as described above, and cells were collected at an OD₆₀₀ of 0.5.

DNA preparation, manipulations, and sequencing; DNA probe preparation; PCRs; and plasmid isolation were performed by standard procedures (39) or in accordance with the manufacturers' instructions. Southern blotting was performed with an ECL labeling kit (Amersham) in accordance with the manufacturer's instructions. An internal 435-bp DNA fragment of trxB1 was PCR amplified with degenerate primers 5′ SNATGAAAATGGCNGARCC 3′ (forward) and 5′ CNCCTTTDATTTCCATNGG 3′ (reverse), which are based on the published L. lactis IL-1403 sequence (5). PCR products were cloned into pCR2.1-TOPO (Invitrogen), and DNA sequences were confirmed. The trxB1-containing EcoRI fragment was cloned into EcoRI-linearized integration vector pRV300 (25) in E. coli and then transformed into WT L. lactis (19) with erythromycin selection. Transformants were plated in the presence of 1 mM DTT in anaerobic jars (GasPak; bioMérieux). trxB1 inactivation was confirmed by Southern blotting with the 435-bp trxB1-containing fragment as the probe. The mutation generates a truncated protein lacking the 103 C-terminal amino acids. From crystallographic studies of E. coli TrxR, motifs needed for flavin adenine dinucleotide binding, enzyme dimerization, and NAD(P)H binding are absent (52). The trxB1 mutant we constructed is thus predicted to be totally inactive.

The following primers were designed on the basis of the L. lactis IL-1403 sequence of genes flanking trxB2 (ccpA and yqjB genes): 5′ CAACCGCTTTATGATTTAGGGG 3′ (forward) and 5′ GTAACCAACCATGAAAGCCC 3′ (reverse). An internal trxB2 primer, 5′ TGGTCCTGTTGGTCTTTATGCGG 3′, was used to complete the sequence.

Total RNA was extracted with a FastRNA Pro Blue kit (Qbiogene) in accordance with the supplier's instructions. PCR-amplified trxA- and gshR-specific probes were prepared on the basis of the MG1363 sequence (gshR) or IL-1403 and SK11 (trxA) with the following primers: for trxA, 5′ ATGGAATATAATATTACTGATGCAACG 3′ (forward) and 5′ TTATGATAATTCAGCAAYMACGGC 3′ (reverse) generated a 315-bp fragment; for gshR, 5′ GGGACTTGGCTTAGCTTATCGCC 3′ (forward) and 5′ CCAGTCCCTGTCAAAATCCCCACC 3′ (reverse) generated a 1,023-bp fragment.

To prepare culture protein extracts, 80 ml of a culture with an OD₆₀₀ of 0.5 was collected. After centrifugation, cells were washed once with 10 mM Tris-HCl (pH 7) and resuspended in 250 μl of 10 mM Tris (pH 7). Cells were broken with an equal volume of glass beads and then subjected to three rounds of vigorous shaking in Fast-prep for 25 s at maximum speed. Extracts were centrifuged for 5 min at 13,000 rpm (Sigma 1K15) at 4°C, and 3 μl of Benzonase (Merck) was added to remove DNA and RNA. The extract was incubated for 20 min at 37°C. Two hundred fifty microliters of a 9.23 M urea-3 M thiourea-4% CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}-80 mM Tris (pH 8)-100 mM DTT solution was added, and the mixture was incubated for 40 min at room temperature and then centrifuged for 10 min at 13,000 rpm (Sigma 1K15) at 4°C. The supernatant was collected, and the protein concentration was measured by the Bradford procedure with bovine serum albumin as the standard (6). A volume corresponding to 250 μg of protein was added to 350 μl of a 6.5 M urea-2.17 M thiourea-4% CHAPS-100 mM DTT-1% IPG buffer (pH 4 to 7; Amersham)-bromophenol blue solution and loaded for the first dimension. Electrophoresis, gel comparisons, and protein identification were performed as previously described (50). At least four gels stained with BioSafe colloidal Coomassie blue (Bio-Rad) were prepared for each condition, with at least two independently prepared samples. Selected spots were identified by mass spectrometry (MS) and by N-terminal sequencing as previously described (50). Molecular size markers were based on the molecular masses of PdhD (49.8 kDa), GapB (35.8 kDa), SodA (23.2 kDa), and YtaA (15.7 kDa). Relative intensities of the two GapB isoforms were determined with the ImageQuant analysis system (version 5.0).

In-gel digestion was performed with an automated protein digestion system (MassPrep Station; Waters, Manchester, United Kingdom). Gel plugs were washed twice with 50 μl of 25 mM ammonium hydrogen carbonate (NH₄HCO₃) and 50 μl of acetonitrile. Cys residues were reduced with 50 μl of 10 mM DTT at 57°C and alkylated with 50 μl of 55 mM iodoacetamide. After dehydration with acetonitrile, proteins were cleaved in gel with 10 μl of 10-ng/μl peptidyl-Asp metalloendopeptidase from Pseudomonasfragi (Asp-N, 1 054 589; Roche Diagnostics GmbH, Penzberg, Germany) in 25 mM NH₄HCO₃. Digestion was performed overnight at room temperature. The peptides generated were extracted with 60% acetonitrile in 5% formic acid.

matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) MS measurements were done with an Ultraflex TOF/TOF (Bruker Daltonik GmbH, Bremen, Germany). This instrument was used at a maximum accelerating potential of 25 kV in the positive mode and −25 kV in the negative mode and was operated in the reflectron mode. In the positive mode, the peptide extract (0.5 μl) was cocrystallized in a saturated solution of α-cyano-4-hydroxycinnamic acid in 50% acetonitrile. External calibration was performed with bradykinin, angiotensin II, angiotensin I, substance P, bombesin, ACTH(1-17), and ACTH(18-39), with respective monoisotopic masses of m/z = 757.400, m/z = 1,046.542, m/z = 1,296.685, m/z = 1,347.735, m/z = 1,619.822, m/z = 2,093.087, and m/z = 2,465.199. Two trypsin autolysis peptide standards at m/z = 842.510 and m/z = 2,211.105 were used for internal calibration. For the negative mode, a solution containing 2,6-dihydroxyacetophenone and dihydrogen ammonium citrate was used as the matrix (15). External calibration was performed with the same peptides as before with monoisotopic masses of m/z = 755.384, m/z = 1,044.526, m/z = 1,294.669, m/z = 1,345.720, m/z = 1,617.807, m/z = 2,091.071, and m/z = 2,463.183. Monoisotopic masses were assigned and used for database searches. Proteins were identified by peptide mass fingerprinting with the MASCOT program (Matrix Science, London, United Kingdom) in the SwissProt and TrEMBL databases. Searches were carried out for all species. One missed cleavage per peptide was allowed, a mass tolerance of 50 ppm was used, and some possible modifications were taken into account, such as carbamidomethylation for Cys and oxidation for methionine.

The trxB1 mutant and its control were grown under static conditions in the absence of DTT. They were harvested at log phase (OD₆₀₀ = ∼0.5) and washed once with purified water. After lysis in 5% 5-sulfosalisylic acid, thiol compounds were measured with monobromobimane (Sigma) as described by Park and Imlay (34). Cysteine, coenzyme A (CoA), and GSH were used as standards.

TrxR enzymes belong to a large family of flavoprotein oxidoreductases, including GSH reductase, alkylperoxidases, and NADH peroxidases. These enzymes have three common domains that are essential for activity: An NADPH-binding domain, a domain involved in flavin adenine dinucleotide binding, and conserved redox-active Cys residues in a CX₂C motif (56). The annotated L. lactis IL-1403 genome sequence (http://www.ncbi.nlm.nih.gov/genomes/framik.cgi?db=Genome&gi=171 ) reports two putative Trx reductases (encoded by trxB1 and trxB2) that have strong homology with the family of flavoprotein oxidoreductases. Although the trxB1 and trxB2 open reading frames (ORFs) are 24% identical, the trxB2 ORF lacks both of the active Cys residues necessary for TrxR function. We confirmed that only one TrxR is also active in L. lactis MG1363 by sequencing the cognate trxB2 gene. The absence of Cys residues in the trxB2 ORF rules out that it is a Trx reductase.

To determine the role of trxB1, we attempted to inactivate this monocistronic gene by single-crossover recombination. We succeeded in obtaining trxB1 mutant candidates in anaerobic jars selecting on DTT-containing plates for erythromycin resistance. Selected trxB1 mutant candidates were confirmed by Southern blotting (data not shown), and one was further characterized. Surprisingly, the trxB1 mutant was viable even in the absence of DTT, which suggested that in contrast to B. subtilis (41) and S. aureus (48), the Trx system is not essential for growth in L. lactis.

Under anaerobic conditions, the trxB1 mutant grew as well as the WT (data not shown). The mutant also grew well under static aerobic conditions when 0.5 to 1 mM DTT was added to the medium to counteract the absence of TrxR (Fig. 1). However, aerobic conditions had a drastic effect on the growth of the trxB1 mutant (Fig. 1). The lag phase was greatly extended in the mutant compared to that of the WT. Interestingly, however, once cells exited from the lag phase, trxB1 mutant cultures attained a cell density comparable to that of the WT strain (an OD₆₀₀ of ∼1.8 for the mutant versus an OD₆₀₀ of ∼2.3 for the WT). When stationary-phase trxB1 mutant cells were recultured in fresh medium under high oxygen tension, the long lag period was still observed, indicating that this phenotype was not due to secondary mutations.

To characterize the roles of TrxB1 in L. lactis, we tested the abilities of different compounds to restore trxB1 mutant growth (Fig. 2). Catalase was tested for the ability to remove accumulated peroxide. GSH and cysteine were chosen because they may participate in thiol balance (24, 32, 35). Pyruvate was used to test the possibility that the glycolytic pathway is a potential target of oxidative stress. Addition of catalase significantly stimulated mutant growth, indicating that trxB1 inactivation probably led to H₂O₂ accumulation. The growth defect was also alleviated by addition of reducing agents like DTT. Surprisingly, GSH or Cys addition at a physiological concentration (1 mM) could also partly restore the growth of the trxB1 mutant. This was unexpected because in E. coli, only a strong reducing agent (DTT or mercaptoethanol) can restore the growth of the trxB1 mutant in a GSH pathway-deficient background (36). Oxidized GSH and cystine did not affect the mutant (data not shown). The inability of oxidized GSH to complement growth suggested that under our test conditions, GSH reductase is not sufficiently active to permit growth of the trxB1 mutant. Finally, addition of pyruvate also restored the growth of the trxB1 mutant. As pyruvate is an antioxidant, but also the end product of glycolysis, the observed phenotype might be due in part to a defect in the glycolytic pathway.

The above results indicated that the Trx system plays an important role in oxidative stress defense, as reported for other bacteria.

One explanation for the ability of a trxB1 mutant to grow in the absence of DTT is that a compensating reducing system is induced. The systems thus far identified in L. lactis are Trx, encoded by trxA, and GSH reductase, encoded by gshR (5). We tested whether transcription of either of these coding genes was altered in the trxB1 mutant compared to that in the WT. Northern blotting experiments showed that trxA was expressed at the same level in the WT and trxB1 mutant strains, both in the presence and in the absence of DTT (data not shown). It is therefore unlikely that trxB1 mutant growth is due to TrxA expression. Similarly, no differences in gshR expression were observed when the WT and trxB1 mutant strains were compared under the conditions described above (data not shown). This result is in keeping with the inability of oxidized GSH to alleviate the trxB1 mutant growth defect. We conclude that growth of the trxB1 mutant is due to neither overproduction of Trx nor GSH reductase activity.

The role of TrxB1 in the aerobic physiology of L. lactis was further examined by comparing the changes in protein levels between the trxB1 mutant and the isogenic WT strain. Crude extracts prepared from cells cultured under static conditions and harvested in log phase (OD₆₀₀ = 0.5) were used for electrophoretic analyses. The levels of 20 out of ∼250 proteins resolved on 2D gels were significantly altered in the trxB1 mutant compared to those of the WT (Fig. 3; Table 1). These proteins affect (i) the ability to survive under oxidative conditions, (ii) carbon, nitrogen, or fatty acid metabolism, and (iii) transcription or translation. Four proteins with unknown functions were also identified.

2-Oxoglutarate decarboxylase (MenD, encoded by menD) and dihydroxynaphthoic acid synthase (MenB encoded by menB) are increased in the trxB1 mutant compared to their levels in the WT (Fig. 3, spots 1 and 2, respectively). The men genes are involved in the formation of menaquinone, a lipophilic redox compound that serves as an electron carrier. As the men genes are clustered on the L. lactis genome (5), other men genes may also be induced in the trxB1 mutant. In previous studies, the quinones in lactic acid bacteria were shown to be exclusively menaquinones (30). These electron carriers actively participate in respiratory chain activity in bacteria like B. subtilis. Our recent demonstration that L. lactis can produce energy via a respiratory process (9) provides one explanation for the presence of quinone biosynthesis in this bacterium. However, menaquinones are reportedly produced even under nonrespiration conditions in lactic acid bacteria (30). This suggests that menaquinones may play another role as a defense against oxidative stress. Soballe and Poole demonstrated that ubiquinone, the electron carrier used under aerobic conditions in E. coli, protected cells against oxidative stress (45).

Superoxide dismutase (SodA), encoded by sodA, is increased in the trxB1 mutant compared to its level in the WT (Fig. 3, spot 3). This enzyme provides an efficient defense against oxygen toxicity for cells in an aerobic environment. sodA expression is increased in a WT strain by acid stress but is only slightly induced by aeration (40).

AhpC, a component of the alkylhydroperoxidase AhpCF, is also induced in the trxB1 mutant (Fig. 3, spot 4). AhpF drives electrons from NAD(P)H to AhpC, which catalyzes reduction of peroxide to alcohol.

We consider it likely that expression of SodA, AhpC, MenB, and MenD is increased in response to the oxidative environment associated with the trxB1 mutation.

β-Phosphoglucomutase (PgmB; Fig. 3, spot 5) catalyzes conversion of glucose-6-phosphate (G6P) into glucose-1-phosphate (G1P). G6P is a glycolysis intermediate, whereas G1P is a precursor used for cell wall synthesis (4). Its overproduction in the trxB1 mutant might affect the pool of these compounds.

NADH oxidase, encoded by noxE (Fig. 3, spot 9), and two subunits of the pyruvate dehydrogenase (PDH) complex, PdhB (Fig. 3, spot 7) and PdhD (Fig. 3, spot 8), encoded by pdhB and pdhD, respectively, act on pyruvate catabolism (50) and were all induced in the trxB1 mutant. PDH converts pyruvate, a glycolytic end product, into acetyl-CoA, accompanied by reduction of NAD⁺ to NADH. NADH is then oxidized by NoxE in the presence of oxygen (14). As PDH is highly sensitive to an increase in the NADH/NAD⁺ ratio, the action of NoxE would generate favorable conditions for PDH activity (44).

Expression of FabD and ThiL is strongly increased in the trxB1 mutant (Fig. 3, spots 10 and 11, respectively). fabD appears to be within a gene cluster involved in fatty acid biosynthesis (5). thiL, encoding an acetyl-CoA acetyltransferase, is located elsewhere on the chromosome. Induction of fatty acid biosynthesis genes in the trxB1 mutant might serve to maintain the cell membrane composition in response to oxidative stress. To our knowledge, this is the first observation to date that links fatty acid biosynthesis to the thiol redox state.

Pyrroline-5-carboxylate reductase (ProC, encoded by proC) catalyzes reversible proline oxidation in the presence of NAD(P)⁺ and is induced in the trxB1 mutant compared to the WT (Fig. 3, spot 12). It is unlikely that overproduction during growth reflects proline starvation in the mutant, as the growth medium is rich in proline (50).

glyA, encoding the serine hydroxymethyltransferase GlyA (Fig. 3, spot 13), belongs to the purine regulon. As this regulon is activated in response to purine starvation (3), its overproduction during growth may reflect purine depletion in the trxB1 mutant strain. GlyA also uses glycine as a precursor for Cys and methionine biosynthesis (10); however, the cysteine levels in the WT and trxB1 mutant strains were similar (data not shown).

GreA, which is induced in the trxB1 mutant (Fig. 3, spot 14), allows RNA polymerase to overcome obstacles encountered during elongation (pause and stop) but also increases transcriptional fidelity by removing misincorporated nucleotides (47). In Pseudomonasaeruginosa, greA belongs to the carA-orf-carB-greA operon, whose expression is correlated with pyrimidine starvation (27). Interestingly, levels of orotate phosphoribosyltransferase (PyrE), an enzyme involved in pyrimidine biosynthesis, is also increased in the trxB1 mutant (Fig. 3, spot 15). We speculate that increased oxidative stress in the trxB1 mutant, e.g., due to H₂O₂ accumulation, may result in increased DNA damage, requiring greater transcriptional stringency and increased nucleotide synthesis for repair. It is also possible that GreA and PyrE respond to other, as yet unknown, signals.

The peptide deformylase Def, whose amounts are increased in the trxB1 mutant (Fig. 3, spot 16), removes the formyl group of the methionine start codon during peptide elongation. Interestingly, it has been reported that Defs are metalloproteins sensitive to oxidative stress. In S. aureus, Streptococcuspneumoniae, or E. coli, an active Def enzyme contains iron coordinated with two histidines, one Cys residue, and water (22). Moreover, E. coli iron-Def is extremely sensitive to oxidation, resulting in total enzyme inactivation. Increased Def production may overcome its inactivation in the trxB1 mutant. This hypothesis is supported by the failure to disrupt def in different bacteria, indicating that the enzyme is essential (28).

Amounts of Glu-tRNA amidotransferase subunit A (GatA) are increased in the trxB1 mutant (Fig. 3; spot 17). gatA is the last gene of the L. lactisgatCgatBybgDgatA operon (5). The Gat complex catalyzes conversion of glutamic acid-tRNA to glutamine-tRNA; both glutamic acid and glutamine are essential amino acids in L. lactis.

The YtjD, YcgE, YahB, and YtaA proteins, whose functions are unknown, are induced in the trxB1 mutant (Fig. 3, spots 18 to 21). We note that YtjD is related to the oxidoreductase family of nitroreductases. YcgE is strongly homologous to LuxS of Streptococcuspyogenes and is induced in acid stress resistance in L. lactis (11); interestingly, it migrated at a position corresponding to a lower than predicted molecular mass (less than 14 kDa instead of 17 kDa), possibly suggesting that it is degraded or matured in the trxB1 mutant. YahB and YtaA are related to nucleotide-binding proteins of the UspA (universal stress protein) family (23).

MS results assigned spots a and b to the same protein, GAPDH, which is encoded by gapB (Fig. 3, spots 6a and 6b). The presence of the same protein at positions corresponding to different isoelectric points suggested that the enzyme undergoes posttranslational processing, as we previously proposed (50). From the theoretical pI (5.44), the position of the b form would correspond to that of native GapB whereas the a spot would be modified. The types of posttranslational modifications identified so far in GAPDH are ribosylation (33) and oxidation of amino acids under aerobic conditions in bacteria (42). As trxB1 inactivation provokes severe oxidative stress, as exemplified by hydrogen peroxide sensitivity (Fig. 2), we suspected that the modified GAPDH form would correspond to oxidation rather than ribosylation. Indeed, GAPDH activity in L. lactis reportedly decreased rapidly after treatment of crude extracts with H₂O₂, whose preferential targets are Cys residues (49). It therefore seemed likely that one of the Cys residues in GapB would be the oxidation target. In gram-positive bacteria, GAPDHs of the glycolytic pathway contain at least two Cys residues, one or both of which are essential for catalytic activity (42) and which are highly conserved in nearly all GAPDH enzymes. Of the two Cys residues present in the L. lactis GapB enzyme (positions 152 and 156; Fig. 4), residue Cys₁₅₂ corresponds by sequence homology to the amino acid involved in enzymatic activity (42).

To test whether GapB was present in oxidized form, we characterized the oxidation levels of the peptide fragment containing the catalytic domain in the two isoforms. For this purpose, we took advantage of the substrate range of the peptidyl-Asp endopeptidase AspN, isolated from P. fragi. AspN recognizes aspartic acid and oxidized forms of Cys (sulfinic or sulfonic acid) as substrates, but not Cys. If Cys is oxidized, AspN digestion profiles generated by MS of the a and b GapB forms should be different (Fig. 4 and 5). As expected, the b form gives a peptide fragment with a mass (m/z = 2,153.916 peak) matching the theoretical mass (m/z = 2,155.93) of the intact peptide. The m/z = 2,169.911 and m/z = 2,185.906 peaks were attributed to the same peptide, but in which the methionine (M₁₆₀) is oxidized to sulfone and sulfoxide, respectively. In contrast, the digestion profile of the a form totally lacked all three peaks, indicating the existence of a new cleavage site for AspN. In the AspN-digested a sample, peaks distinct from those present in the b form appeared: an m/z = 1,034.465 peak (Fig. 5, zoom B) was present in the MS spectra. This mass matched that expected for a peptide generated by cleavage at oxidized Cys₁₅₂. The complementary peptides were also present in three peaks (m/z = 1,127.418; m/z = 1,143.413; m/z = 1,159.408) corresponding to the different oxidation levels of M₁₆₀ as described for the b form. We did not observe masses on the MS profile that could correspond to the peptide cleaved at Cys₁₅₆ or at both Cys₁₅₂ and Cys₁₅₆. These analyses confirm that only Cys₁₅₂ is oxidized in GapB. We concluded that the GapB b form corresponds to reduced native GapB, whereas the a form corresponds to GapB oxidized uniquely at the Cys₁₅₂ position.

In WT strains and under fermentation growth conditions, both GapB isoforms are present in relatively unchanged amounts (Fig. 3 and 6) (17, 50, 55). This profile was only slightly affected when cells were cultured in the presence of the strong reducing agent DTT (favoring the b reduced form) (Fig. 6). This absence of a shift to the b GapB form in the DTT-grown cells suggests that Cys₁₅₂ oxidation is irreversible. This would rule out oxidation to Cys sulfenic acid (Cys-SOH), whose reduction by DTT is reversible (51). Consequently, the oxidation state of Cys₁₅₂ likely corresponds to cysteine sulfinic (SO₂H) or sulfonic (SO₃H) acid, whose reduction by DTT is irreversible.

trxB1 inactivation had a marked and unexpected effect on the amounts of GapB isoforms. The amount of oxidized a form decreased 10-fold, while the reduced b form was maintained at a level similar to that in the WT. Addition of DTT resulted in a profile like that of the WT. This unexpected finding suggests that in the trxB1 mutant, the oxidized GapB isoform is degraded (although no increase in the amounts of known proteases was detected on 2D gels). Alternatively, the trxB1 mutation results in the induction of a GapB-specific or general reducing function that prevents formation of the oxidized a form.

Detection of increases in eight proteins in the trxB1 background leads us to suggest that this mutant must cope with cell envelope damage. First, the AhpCF complex, which is needed to reduce lipophilic alkyl hydroperoxides in vitro (21), is induced in the trxB1 mutant. Second, MenD and MenB may constitute a novel class of proteins that respond to intracellular oxidative stress such as that created by the trxB1 mutation. The Men proteins participate in the formation of quinones, redox compounds that can protect lipids against oxidative stress (45). Third, elements of the PDH complex and NoxE are also overexpressed. These enzymes act in concert to catalyze conversion of pyruvate to acetyl-CoA, a fatty acid precursor. We speculate that their induction may respond to greater lipid peroxidation in the trxB1 background. Consistent with this hypothesis, two other lipid biosynthesis proteins, ThiL and FadD, were also overproduced (Fig. 3). Moreover, overproduction of PgmB, which catalyzes the conversion of G6P to G1P, a precursor used in cell wall synthesis, may also contribute to the maintenance of envelope integrity.

The ensemble of proteins induced by TrxR deficiency leads us to conclude that cell envelope biosynthesis pathway components are regulated by oxidative stress.

Oxidative conditions are perceived in bacteria through redox-sensitive regulators that activate response pathways in bacteria. Signaling occurs via oxidation of Cys residues (46) to form disulfide bonds (in the E. coli OxyR protein) (1) or by conversion of Cys to cysteine sulfenic acid (in the single Cys of the B. subtilis OhrR protein) (12). In L. lactis, trxB1 inactivation increases cell peroxide levels, as reflected by the protective effect of catalase (Fig. 2). The increased amounts of AhpC (and SodA) in the L. lactistrxB1 mutant strain might indicate that Cys-containing redox-sensitive proteins may regulate their expression. Several such regulators, containing one or more Cys residues, are identified in L. lactis by genome analyses (5). Our results lead us to suggest that the Trx-TrxR system controls the redox state in lactococci.

Certain enzymes of the glycolytic pathway are known targets of oxidative stress (49). Pyruvate complemented the trxB1 mutant, which is consistent with the idea that enzymes that mediate glycolysis are a target of oxidative stress in vivo. Nevertheless, pyruvate is also an H₂O₂ scavenger and may thus protect cells against stress, as does catalase (49). Among the enzymes identified, GAPDH seems highly sensitive to oxidation in L. lactis. Here we demonstrated that one of the isoforms corresponds to an oxidation of a Cys residue, as reported for B. subtilis (42) and recently in S. aureus (53). Curiously, only Cys₁₅₂, which is essential for activity, is oxidized in the protein, while the neighboring Cys is unmodified. This specificity may be explained by the reactivity of Cys₁₅₂. Deprotonation of this residue during GAPDH catalytic enzyme activity may favor its oxidation (8). We observed that the oxidized form of GapB is not converted to the reduced form by DTT (Fig. 6). This may indicate that GapB oxidation is an irreversible conversion to an inactive form and thus may not correspond to redox control of GAPDH in L. lactis. Similarly, new synthesis of the active enzyme is observed in S. aureus after H₂O₂ treatment, rather than reduction of the inactive enzyme (53).

Unexpectedly, the level of the oxidized, inactive GapB form is strongly decreased in the trxB1 mutant while the level of the active form remains constant compared to the WT profile (Fig. 6). The specific decrease in oxidized GapB could indicate that a compensating reducing activity prevents its formation. We observed that overall protein carbonylation damage in the trxB1 mutant and WT strains was about equivalent (data not shown), suggesting that deficiency of TrxR activity is likely to be compensated in the trxB1 mutant (discussed below). However, this does not explain why so little oxidized GapB is present in the trxB1 mutant, compared to the WT. Alternatively, GapB might be degraded by proteases induced in the trxB1 mutant under oxidizing conditions. The presence of another protein (YgcE; see above) that seems to have undergone cleavage also suggests that a protease is activated in the trxB1 mutant. Although known proteases were not noticeably induced by 2D gel analyses, we are currently pursuing this question.

Although, trxB1 inactivation results in increased levels of GlyA, no variation of cysteine concentration was observed between the WT and trxB1 strains. L. lactis seems to differ from B. subtilis, in which Cys biosynthesis proteins are induced by addition of diamide, a thiol-oxidizing agent (24). It is thus unlikely that this thiol compound compensates for the absence of TrxR. In addition, despite the existence in L. lactis of a functional GSH reductase and a potential GSH peroxidase (29), a role for GSH as a reducing system is also ruled out, as GSH is not synthesized. Consequently, the ability of the trxB1 mutant to grow despite a long growth lag suggests the existence of a compensatory reducing system in L. lactis. This hypothesis is supported by observations that no increase in protein carbonylation was detected in the trxB1 mutant context.