Thioredoxin Reductase Is Essential for Thiol/Disulfide Redox Control and Oxidative Stress Survival of the Anaerobe Bacteroides fragilis

Bacterial strains and growth conditions. B. fragilis strains used in this study are listed in Table 1. Strains were grown anaerobically in brain heart infusion (BHI) broth supplemented with hemin, cysteine, and NaHCO₃ (BHIS) for routine cultures. Rifamycin (20 μg/ml), 100 μg/ml gentamicin, 5 μg/ml tetracycline, 10 μg/ml erythromycin, and 25 μg/ml cefoxitin were added to the media when required. Semidefined medium (SDM) was prepared by adding 0.5 g/liter tryptone to a modified chemically defined medium of Varel and Bryant (53). The modified medium composition was as follows: KH₂PO₄, 1.15 g/liter; NH₄SO₄, 0.4 g/liter; NaCl, 0.9 g/liter; l-methionine, 75 mg/liter; MgCl₂·6H₂O, 20 mg/liter; CaCl₂·2H₂O, 6.6 mg/liter; MnCl₂·4H₂O, 1 mg/liter; CoCl₂·6H₂O, 1 mg/liter; resazurin, 1 mg/liter; l-cysteine, 500 mg/liter; hemin, 5 mg/liter; and glucose, 5 g/liter. Twenty milliliters of 10% NaHCO₃ was added per liter of medium, and the final pH was 7.2. B. fragilis requires a reduced sulfur source to support growth (53), which was tryptone when l-cysteine was omitted.

Construction of trxB deletion mutant.Briefly, a 2.1-kb chromosome fragment containing the trxB region was amplified by PCR using oligonucleotides containing modifications to create restriction sites for SphI and BamHI and then cloned into pUC19 to create pFD846. Next, an internal 60-bp BglII/ClaI fragment from the trxB gene was deleted and replaced with a 2.1-kb BamHI/NarI cefoxitin (cfxA) resistance gene to construct a ΔtrxB::cfxA deletion in a new plasmid, pFD846/ΔtrxB::cfxA. Then a 4.5-kb SphI/SstI fragment from pFD846/ΔtrxB::cfxA was cloned into the suicide vector pFD516 (50) to create pFD874. This plasmid was mobilized from Escherichia coli DH10B into B. fragilis 638R by triparental mating, and exconjugants were selected on BHIS containing 20 μg/ml rifamycin, 100 μg/ml gentamicin, and 25 μg/ml cefoxitin. Sensitivity to erythromycin was determined, and Southern blotting was performed to confirm the double-crossover allele exchange to create a trxB mutant designated IB370.

A ΔtrxB ΔoxyR mutant was constructed by mobilizing suicide vector pFD754 (containing the ΔoxyR::tetQ deletion [39]) into B. fragilis IB370 as described above. Exconjugants were selected on BHIS containing 20 μg/ml rifamycin, 100 μg/ml gentamicin, 5 μg/ml tetracycline, and 25 μg/ml cefoxitin. Sensitivity to either tetracycline/cefoxitin or erythromycin was used to identify recombinants that were tetracycline and cefoxitin resistant and erythromycin sensitive. Strain IB372 was selected for further study.

A plasmid constitutively expressing trxB was constructed by PCR amplification of a 1,285-bp promoterless trxB gene containing 21 bp upstream of the ATG start codon and 255 bp downstream of the stop codon. The promoterless trxB gene fragment was cloned into the BamHI site of the Bacteroides-E. coli shuttle expression vector pFD340 (49) in the same orientation as the IS4351 promoter. The new construct, pFD892, was mobilized into B. fragilis IB370 as described above. Transconjugants were selected on BHIS containing 20 μg/ml rifamycin, 100 μg/ml gentamicin, and 10 μg/ml erythromycin.

RNA extraction, Northern blot hybridization, and primer extension.For oxidative stress experiments, cultures were grown in BHIS to an A ₅₅₀ of 0.3 and treated with 50 μM H₂O₂ for 5 min prior to total RNA extraction. To induce oxygen stress, cultures were split in half; one half was incubated anaerobically, and the other half was shaken aerobically at 250 rpm using a volume/flask ratio of 1/5 as previously described (38). For diamide treatment, mid-log-phase cultures were treated with different concentrations of diamide as indicated below. Total RNA extraction and Northern blot analysis of mRNA were carried out as previously described (38). Internal fragments of trxB and lolA were used as specific probes when required. In a densitometry analysis the intensity of the autoradiograph was normalized to the relative intensity of total 23S and 16S rRNA detected on the ethidium bromide-stained agarose gel to correct for any loading differences.

Primer extension analysis was performed with total RNA obtained from mid-log-phase cells of B. fragilis 638R grown anaerobically and then subjected to oxidative stress conditions as described previously (39). A lolA-specific oligonucleotide and a trxB-specific oligonucleotide were labeled with [α-³²P]ATP and used as primers for the reverse transcriptase reaction.

Inhibition of growth by oxygen in a soft agar tube culture.A 0.1-ml aliquot of an overnight culture of B. fragilis in BHIS broth without l-cysteine was mixed with 5 ml of BHIS without added l-cysteine and resazurin containing 0.4% agar at 45°C in screw cap tubes (100 by 13 mm) inside an anaerobic chamber. The tubes were capped with sterile cotton stoppers and incubated aerobically at 37°C for 24 h. The area from the top of the agar down to the edge of visible bacterial growth was considered the zone of growth inhibition by oxygen and was measured.

GSH determination.The total GSH/oxidized glutathione (GSSG) content in cells was determined with a cycling enzyme assay utilizing glutathione reductase as described by Howe et al. (18) and Fahey et al. (9), with some modifications. Overnight cultures of B. fragilis 638R and E. coli HB101 (control strain) were grown in SDM under anaerobic conditions and used to inoculate fresh SDM cultures. Bacteria were grown to mid-log phase (optical density at 550 nm, 0.45), and then cultures were split. One half of each culture was incubated aerobically in a shaker incubator at 37°C for 1 and 24 h. The other half was incubated under anaerobic conditions for the same time periods. The amount of total GSH was determined in mid-log- and stationary-phase cells (24 h of incubation). Cells were washed in phosphate-buffered saline (50 mM phosphate buffer [pH 7.4], 150 mM NaCl). Cell pellets were extracted in boiling 80% ethanol, dried at 105°C, weighed in preweighed tubes, and stored at −70°C until the assay was performed. Dried pellets were suspended in 800 μl of 2% 5-sulfosalicylic acid. GSH contents were determined using 100-μl samples, and immediately prior to the assay, samples were treated with 3.9 μl of a 1:4 dilution of triethanolamine to bring the pH to 7.0. The presence of GSSG was also determined in 100-μl samples by addition of 3.9 μl of 2-vinylpyridine prior to addition of 3.9 μl of a 1:4 dilution of triethanolamine. Briefly, the enzymatic assay was carried out using a 50-μl treated sample and a standard mixed with 100 μl of recycling reaction buffer containing 42.7 mM sodium phosphate (pH 7.5), 472 μM EDTA, 226 μM 5,5′-dithiobis(nitrobenzoic acid), 313 μM NADPH, and 20 U of glutathione reductase. Microplates were read at 405 nm every 45 s using a fluorescence/absorbance plate reader (FL600; Bio-Tek). Standard curves were used to determine the GSH/GSSG contents of the samples expressed in micromoles per gram (dry weight).

Mouse model of abscess formation.The experimental model for intraabdominal infection described previously was used, with minor modifications (7). B. fragilis 638R and the isogenic trxB mutant IB370 used for challenge were grown overnight in BHIS, concentrated, and then frozen until they were needed. The frozen cultures were titrated, and then just prior to use they were diluted to obtain the desired challenge dose. C57BL male mice that were 6 to 8 weeks old were challenged via intraperitoneal injection of 1 × 10⁶, 1 × 10⁷, or 1 × 10⁸ cells in 0.2 ml (final volume) of BHI broth containing diluted sterile rat cecal contents. The mice were sacrificed 6 days later and examined for intraabdominal abscesses. The presence of one or more abscesses was considered a positive result. Animals inoculated with broth plus sterile cecal contents alone without bacterial challenge did not form abscesses.

Nucleotide sequence analysis of the TrxB/Trx system.The availability of the B. fragilis 638R genome sequence (6; http://www.sanger.ac.uk/Projects/B_fragilis/ ) allowed us to search for the presence of the Trx motif sequence. It was found that B. fragilis possesses an extensive Trx system consisting of a TrxB, six Trx orthologs, and more than 14 other proteins that were larger than Trxs but had the Trx-like protein fold and the CXXC motif (data not shown). The six Trx orthologs contain the redox-active center CXXC motif common to Trxs (8, 17), and four of these orthologs, TrxA, TrxC, TrxE, and TrxF, contain the characteristic motif WCGPC (Fig. 1). TrxC also contains an N-terminal conserved prokaryotic membrane lipoprotein precursor signal peptide (15). The TrxD active site contains WCVYC, while TrxG has a WCPHC motif. In addition, another protein with homology to the Trx fold motif family, tentatively designated TrxX, lacks the reactive redox cysteine residues which were replaced by serine and tyrosine (WSPHY). When B. fragilis Trx homologues were compared to other microbial Trxs using a progressive multiple alignment and parsimony analysis, it was found that except for TrxA, all B. fragilis Trx homologues were clustered together (data not shown). TrxA, however, grouped in a cluster together with Flavobacterium, Corynebacterium, and Clostridium Trx homologues.

• Open in new tab
• Download powerpoint

FIG. 1.

Multiple alignment of the deduced amino acid sequences of B. fragilis (Bf) Trx orthologs (TrxA through TrxG plus TrxX) with E. coli (Ec) TrxA and TrxC amino acid sequences. The conserved Trx fold containing the CXXC redox active site is indicted by thick lines above and below the amino acid sequence region. A consensus of at least 50% identical amino acid residues is indicated by a black background. Conserved amino acid substitutions are indicated by a gray background. The predicted conserved prokaryotic membrane lipoprotein precursor signal peptide (15) present in TrxC is indicated by a thin line under the amino acid sequence. Alignment of peptide sequences was performed using the GCG programs pileup and box with the peptide scoring matrix default data file blosum62.cmp for comparison of amino acid substitutions.

The B. fragilis 638R trxB open reading frame contains a 948-bp nucleotide sequence. The deduced amino acid sequence is a 315-amino-acid protein homologous to other bacterial Trx reductases. When B. fragilis 638R TrxB was compared to other TrxBs, it showed the highest homology to Bacteroides thetaiotaomicron VPI-5482 TrxB (90% identity and 96% similarity) and P. gingivalis TrxB (84% identity and 80.5% similarity). Strong homology was also found to TrxBs of E. coli (52.2% identity and 62.3% similarity), Haemophilus influenzae (51.4 identity and 61.5% similarity), and Mycobacterium tuberculosis (49% identity and 59.3% similarity). In addition, the pyridine nucleotide disulfide oxidoreductase class II active site is conserved in B. fragilis TrxB (data not shown). The phylogenetic relationship between B. fragilis TrxB and 26 other TrxB homologues from animals, plants, bacteria, and yeasts was determined from a progressive multiple alignment of the amino acid sequences, followed by parsimony analysis. This comparison showed that B. fragilis TrxB groups with other members of the unique Cytophaga-Flexibacter-Bacteroides phylum (see Fig. S1 in the supplemental material).

GSH determination.During analysis of the B. fragilis TrxB/Trx system, we noticed that there were no Grxs or any of the genes required for GSH biosynthesis (γ-glutamylcysteinyl synthetase and glutathione synthase) in the genome. This suggested that the TrxB/Trx system might be the major or sole intracellular thiol/disulfide redox system in this anaerobe. Consistent with this was the finding that no GSH/GSSG was found in crude extracts of B. fragilis from mid-log-phase or stationary-phase cultures (limit of detection, 5 pmol in 50 μl of sample or standard). Cultures exposed to aerobic conditions also lacked detectable GSH/GSSG. In contrast, GSH/GSSG was present in the crude extracts of E. coli control cultures; 22.2 and 7.05 μmol/g (dry weight) were found in logarithmic- and stationary-phase aerobic cultures, respectively, and in anaerobic cultures the levels were 2.81 and 4.96 μmol/g (dry weight), respectively. These values were comparable to levels previously described for E. coli under aerobic and anaerobic conditions (9).

Effect of free thiol on growth of a trxB mutant.A trxB mutant was constructed to determine if another mechanism was present that could maintain reduced cellular thiols and compensate for loss of the TrxB/Trx system. The availability of thiol compounds in the media greatly affected growth of the ΔtrxB mutant in rich BHIS media and SDM. The ΔtrxB mutant was unable to maintain growth in the absence of an exogenous free thiol source in SDM, but addition of cysteine or dithiothreitol (DTT) restored the ability of the trxB strain to grow (Fig. 2A and 2B). The growth defect in rich media was less pronounced, probably because of the greater abundance of reduced sulfur sources compared to SDM. Taken together, these findings suggest that in B. fragilis the TrxB/Trx system is the major or sole system for mediating the intracellular thiol/disulfide equilibrium; otherwise, the ΔtrxB mutant would not have been so dependent on added thiol for growth.

• Open in new tab
• Download powerpoint

FIG. 2.

Growth of B. fragilis parent strain 638R and the ΔtrxB mutant in BHIS (A) and in SDM (B). •, parent strain grown in medium containing cysteine; ○, parent strain grown in medium without cysteine; ▪, trxB mutant grown in medium containing cysteine; □, trxB mutant grown in medium without cysteine; ▴, trxB mutant grown in medium without cysteine but with DTT added at the time point indicated. O.D_{550 nm}, optical density at 550 nm.

Regulation of trxB expression by oxidative stress and diamide.The trxB gene was the second gene of what appeared to be a two-gene operon that also encoded the lipoprotein chaperone, LolA. The expression of trxB following exposure to oxygen, to hydrogen peroxide, and to different concentrations of the thiol-specific oxidant diamide was examined by Northern hybridization. The autoradiographs revealed that trxB mRNA was expressed as an approximately 0.96-kb monocistronic transcript and as a less abundant 1.8-kb bicistronic transcript containing the upstream gene lolA (Fig. 3A). Expression of trxB monocistronic mRNA was induced approximately fivefold by both hydrogen peroxide and oxygen compared to anaerobic controls (Fig. 3A). Deletion of oxyR did not significantly affect the expression of trxB following oxidative stress, but the basal levels of the trxB transcript were slightly higher (twofold) anaerobically in the oxyR background than in the parent strain. trxB also was induced when cultures were treated with diamide in a dose-dependent manner, indicating that trxB was expressed in response to a thiol/disulfide redox imbalance. Maximum induction of trxB mRNA (approximately sevenfold) was observed with 100 μM diamide compared to untreated controls (Fig. 4). In addition, diamide also induced a 1.8-kb lolA trxB bicistronic mRNA, confirming that lolA is cotranscribed with trxB in B. fragilis following thiol redox stress. When Northern blots were probed with an internal lolA gene fragment, the results confirmed that there was up-regulation of the 1.8-kb lolA trxB bicistronic transcript (Fig. 3B) (unpublished data).

• Open in new tab
• Download powerpoint

FIG. 3.

(A) Northern hybridization analysis of total RNA of B. fragilis 638R (wt), IB298 (ΔoxyR), and IB263 constitutively activating OxyR [oxyR(Con)]. Cells were grown to mid-logarithmic phase in BHIS without addition of cysteine and then exposed to oxygen. Cultures were treated with H₂O₂ or exposed to oxygen as indicated in the text. The probe was a trxB internal gene fragment. The approximate sizes of the transcripts are indicated. DTT was used as a control reductant. (B) Diagram of lolA and trxB genetic organization and structure. The large arrows indicate the open reading frames and their direction of transcription. The arrows below the map indicate the lengths of the transcripts and their orientation. The bent arrow indicates the putative promoter predicted from data presented in Fig. 5. The lolA trxB intergenic nucleotide sequence region containing a predicted stem-loop structure is also shown. The lolA stop codon (UAG) and the trxB translational start codon (ATG) are underlined. The cleavage nucleotide regions of the bicistronic lolA trxB mRNA determined by the RNase protection assay are indicated by arrows.

• Open in new tab
• Download powerpoint

FIG. 4.

Autoradiograph of Northern hybridization of total RNA from mid-log-phase B. fragilis 638R grown in BHIS without addition of cysteine. Cultures were treated with diamide at the concentrations indicated at the top. The probe was a trxB internal gene fragment. The approximate sizes of the transcripts are indicated. DTT was added to one culture as a control to show that addition of a reducing agent to the medium did not lead to induction of trxB.

Primer extension analysis of lolA revealed two transcription initiation sites, P1 and P2 (Fig. 5). P1 is a constitutive promoter whose transcription starts at an adenine nucleotide 26 bases upstream of the translation start codon. In contrast, P2 was induced upon exposure to oxidative stress and diamide treatment. P2 transcription starts at a thymine nucleotide 8 bases upstream of the translation start side. A lolA transcriptional fusion with the promoterless β-xylosidase reporter gene confirmed that the lolA promoter was induced by oxidative stress (data not shown). In contrast, a trxB transcriptional fusion containing the lolA-trxB intergenic region extending 238 bases upstream from the trxB translational start site revealed that there was no transcriptional initiation directly upstream of trxB (data not shown). Because trxB was also detected as a single mRNA in a Northern blot analysis (Fig. 4A), we performed an RNase protection assay to determine whether there might have been posttranscriptional modification of the lolA trxB bicistronic mRNA. The RNase protection assay used to map the lolA trxB intergenic region showed that two 5′ end protected sites of the trxB mRNA mapped at nucleotide positions −36/−35 and −21/−20 with respect to the trxB translational start site (Fig. 3B and data not shown).

• Open in new tab
• Download powerpoint

FIG. 5.

Autoradiograph following primer extension analysis of the lolA trxB mRNA transcription initiation regions. Total RNA extracted from the wild-type strain under different stress conditions was used in the extension reactions. Cultures were treated by adding 50 μM H₂O₂, by exposure to atmospheric air for 1 h, or by adding 100 μM diamide. An untreated anaerobic control culture was also included. A DNA sequencing ladder generated with the same primer that was used for the primer extension reactions is shown on the left. The arrows indicate the initial +1 adenine for the P1 promoter and the +1 thymine for the P2 promoter. (B) Nucleotide sequence of the lolA trxB regulatory region. Based on a B. fragilis consensus (4), the predicted −10 and −35 promoter regions for the transcription initiation sites P1 and P2 are indicated by bold type and underlined with solid and dashed lines, respectively. The bent arrows indicate the initial +1 adenine nucleotide and the +1 thymine nucleotide for the P1 and P2 transcription start sites, respectively, at positions −26 and −8 relative to the translation start codon. The first 26 codons for lolA are shown.

Sensitivity to diamide and peroxide.Diamide sensitivity mimics thiol oxidation due to oxygen exposure in anaerobic bacteria (29); therefore, sensitivity to diamide was used to determine whether TrxB plays a role in the thiol/disulfide homeostasis. As shown in Fig. 6, the trxB mutant was more sensitive to diamide in disk diffusion assays than the parent strain and an oxyR mutant. When the trxB mutant was complemented with pFD892, the diamide sensitivity was nearly identical to that of the parent stain. Although the oxyR mutation alone had only a small effect on diamide sensitivity, the oxyR trxB double mutant showed greatly increased sensitivity to diamide compared to the single mutants, especially on the plates exposed to oxygen. This suggests that growth inhibition by diamide in both aerobic and anaerobic conditions was likely due to an imbalance in the cellular thiol/disulfide homeostasis and that there might be a link between thiol redox and peroxide stress responses in this anaerobic bacterium. In this regard, the trxB mutant strain was somewhat more sensitive to hydrogen peroxide stress than the wild type, and this sensitivity was additive with oxyR since the oxyR trxB double mutant was inhibited more by hydrogen peroxide than either of the single mutants (Fig. 6).

• Open in new tab
• Download powerpoint

FIG. 6.

Disk inhibition assay testing the sensitivity of B. fragilis strains to diamide in the absence of oxygen (bars with thick stripes) and in the presence of oxygen (bars with thin stripes) and to hydrogen peroxide (open bars). The cultures were grown overnight in BHIS, and 0.1 ml was plated on BHI agar without added cysteine. Ten microliters of a 2 M diamide solution or 0.88 M hydrogen peroxide was applied to a 6-mm filter paper placed in the center of each plate. Plates were incubated anaerobically overnight at 37°C. Duplicate plates with the diamide solution were exposed to air for 6 h at 37°C before they were placed in the anaerobic chamber and incubated as described above. Zones of growth inhibition (in millimeters) were measured. The error bars indicate the standard deviations of the means from three independent determinations. wt, wild type.

Sensitivity to oxygen exposure in soft agar tubes.The effect of oxygen on the growth of the trxB strain was examined in BHIS soft agar tubes incubated in air without added cysteine. Clearance of bacterial growth from the top of a soft agar tube showed that the trxB mutant was slightly more sensitive to oxygen diffusion into the agar than the oxyR or parent strain (Fig. 7). The parent and oxyR strains had clear zones to 0.4 cm, while the trxB strain was inhibited by oxygen down to 0.9 cm into the agar medium. Interestingly, the oxyR trxB double mutant strain did not even grow into the bottom of the soft agar (depth, 5.5 cm). Addition of cysteine to the medium was not able to overcome the growth inhibition of the double mutant in this assay. The growth of the oxyR trxB double mutant under anaerobic conditions was similar to that of the parent strain (data not shown).

• Open in new tab
• Download powerpoint

FIG. 7.

Growth of B. fragilis strains in soft agar tubes in the presence oxygen. A 0.1-ml aliquot of bacteria grown overnight in BHIS was mixed with 5 ml BHIS without cysteine containing 0.4% agar at 45°C under anaerobic conditions. After solidification of the agar, cultures were placed outside the anaerobic chamber and incubated aerobically at 37°C for 24 h. The clearance zone was measured from the top edge of the agar down to the growth line in the column. The arrows indicate the interface between bacterial growth and the zone of growth inhibition. NG, no growth.

Abscess formation.When B. fragilis leaves the anaerobic intestinal environment to colonize more oxygenated tissues, such as the peritoneal cavity, it encounters a hostile environment (46). It is presumed that protection from oxidative stress is important for survival in the initial stages of an extraintestinal infection (39, 40). Thus, the ability of a trxB strain to form intraabdominal abscesses in a mouse model was investigated. Compared to the parent strain, the trxB mutant was not able to form abscesses at all infection doses (Table 2). In contrast, the parent strain formed abscesses in some animals at every infection dose tested from 10⁶ to 10⁸ CFU. Challenge with sterile cecal contents alone was used a control, and no abscesses were observed.