|Home | About | Journals | Submit | Contact Us | Français|
In the strict anaerobe Clostridium acetobutylicum, a PerR-homologous protein has recently been identified as being a key repressor of a reductive machinery for the scavenging of reactive oxygen species and molecular O2. In the absence of PerR, the full derepression of its regulon resulted in increased resistance to oxidative stress and nearly full tolerance of an aerobic environment. In the present study, the complementation of a Bacillus subtilis PerR mutant confirmed that the homologous protein from C. acetobutylicum acts as a functional peroxide sensor in vivo. Furthermore, we used a transcriptomic approach to analyze gene expression in the aerotolerant PerR mutant strain and compared it to the O2 stimulon of wild-type C. acetobutylicum. The genes encoding the components of the alternative detoxification system were PerR regulated. Only few other targets of direct PerR regulation were identified, including two highly expressed genes encoding enzymes that are putatively involved in the central energy metabolism. All of them were highly induced when wild-type cells were exposed to sublethal levels of O2. Under these conditions, C. acetobutylicum also activated the repair and biogenesis of DNA and Fe-S clusters as well as the transcription of a gene encoding an unknown CO dehydrogenase-like enzyme. Surprisingly few genes were downregulated when exposed to O2, including those involved in butyrate formation. In summary, these results show that the defense of this strict anaerobe against oxidative stress is robust and by far not limited to the removal of O2 and its reactive derivatives.
Although many bacteria are commonly classified as strict anaerobes, aerotolerance can vary greatly among these organisms. Some anaerobes even use catalase and superoxide dismutase as efficient H2O2- and O2-scavenging enzymes, which were previously proposed to be limited exclusively to aerobes (45; for reviews, see references 6 and 32). Even when missing these two, fermentative anaerobes can often use their replete pool of reducing equivalents for the complete reduction of these molecules and O2 itself. In anaerobes in particular, the level of generation of the two major reactive oxygen species (ROS), H2O2 and O2−, is elevated during oxygen exposure due to their wide use of reduced flavins as cofactors in redox enzymes. Models for anaerobic oxygen detoxification via superoxide reductase, peroxidase, and NADH oxidases have been proposed for Pyrococcus furiosus, Desulfovibrio vulgaris, and, most recently, Clostridium acetobutylicum (33, 57, 68). These essentially include oxygen-reducing flavodiiron proteins (FDPs), superoxide-reducing desulfoferrodoxin (Dfx), and rubrerythrins as peroxidases (26, 56, 57). Reduced rubredoxin thereby acts as the intermediate electron carrier, the pool of which is continuously reloaded by an NADH-dependent rubredoxin-dependent oxidoreductase (55, 57). The sporulating gram-positive anaerobe Clostridium acetobutylicum is a classic example of a fermentative strict anaerobe. In the absence of oxygen, it ferments sugars to the organic acids acetate and butyrate or shifts to solvent formation with acetone and butanol as major products (for a review, see reference 13). Especially the latter one has regained attention lately as an attractive biofuel (14). Thus, metabolic engineering and in silico modeling of clostridial metabolism are ranked highly among recent efforts to increase productivity (39, 53, 61, 64). Furthermore, the sequenced genome and the design of microarrays for global-scale transcriptional analysis of C. acetobutylicum have substantially improved the understanding of solvent formation and allowed a detailed view on its unique life cycle (34, 50, 54).
During exposure to sublethal O2 concentrations, C. acetobutylicum rapidly induces the expression of all components of its detoxification system and consumes dissolved O2 (36). However, when the rate of influx exceeds the rate of consumption, the organism immediately ceases its metabolic activity and resumes only after anaerobiosis is restored (35, 51). Studies of central redox enzymes employed by anaerobes revealed that these were highly susceptible to damage of their iron sulfur sites caused directly by molecular oxygen, e.g., pyruvate-formate lyase of Escherichia coli or pyruvate-ferredoxin oxidoreductase (PFOR) from C. acetobutylicum (46, 59). Bacteroides thetaiotaomicron, a gram-negative anaerobe, is even employing efficient repair mechanisms to recycle this enzyme once conditions are shifted back to anaerobiosis (52).
Altogether, these observations have established the concept that anaerobes are not primarily sensitive to oxygen due to their lack or inefficiency of scavenging enzymes. Instead, central metabolic steps in anaerobic energy conversion are poisoned by molecular oxygen (29, 31). Therefore, it does not seem surprising that some anaerobes that are temporarily but frequently exposed to oxygen found ways to avoid these bottlenecks and direct the carbon flow onto oxygen-resistant pathways (3, 52, 66). A similar strategy has been proposed for an oxygen-resistant mutant of C. acetobutylicum (31). Following the deletion of the clostridial perR homologue, the constitutive expression of its regulon resulted in drastically enhanced aerobic survival and permitted time-limited growth in an aerobic environment (25). In closely related facultative aerobes like Bacillus subtilis, the peroxide repressor PerR senses intracellular levels of H2O2 by metal-catalyzed histidine oxidation and controls genes for which the proteins are involved in the scavenging of hydrogen peroxide, iron storage, and DNA protection (22, 40). PerR regulation was also previously proposed for anaerobic sulfate reducers and was further supported by the coordinate induction of the predicted target genes in response to low oxygen concentrations with Desulfovibrio vulgaris (47, 58). The work presented here is a combined approach of microarray technology and complementational studies to obtain a global view of the organism's oxygen stimulon and refine the regulatory role of PerR in a strict anaerobe.
All strains used in this study are listed in Table Table1.1. C. acetobutylicum MGCcac15 and C. acetobutylicum ΔperR strains were stored as spore suspensions at −20°C and grown in MS minimal medium as described previously (25). Growth was monitored as the optical density at a wavelength of 600 nm (OD600). Cells for RNA sampling were anaerobically grown in 1-liter serum flasks, and at an OD600 of 0.6 to 0.7, cultures were either aerated with compressed air with a rate of 0.05 liters of air min−1 or left anaerobic as a control. The concentration of dissolved O2 was measured using the WTW (Weilheim, Germany) OXI 196 oxygen meter. As a result of the continuous uptake of O2 from the growth medium, its concentration stayed essentially below the limit of detection (<6.25 μmol liter−1). Samples for RNA isolation were taken following 1 h of O2 exposure or from anaerobic controls at an OD of 0.6 to 0.8. Escherichia coli pTperRCac constitutively expressing the clostridial PerR homologue was grown overnight in 500 ml of LB medium with 100 μg ml−1 of ampicillin at 30°C and 120 rpm on a rotary shaker. B. subtilis strains were grown at 30°C and 170 rpm in LB medium. For cultivation on solid medium, 15 g liter−1 of agar was added to the medium. To select for the B. subtilis perR mutant strain, 100 μg ml−1 spectinomycin was added as an antibiotic. Complemented strains carrying the clostridial perR gene were additionally selected by the addition of 10 μg ml−1 chloramphenicol, and 0.5% (wt/vol) xylose was added to induce C. acetobutylicum perR (perRCac) expression.
For the heterologous overexpression of the clostridial PerR protein, its gene was PCR amplified from chromosomal DNA of C. acetobutylicum using oligonucleotides P_TSH-CAC2634-BamHI (5′-AAAAGGATCCAACGATATATCTACAA-3′) and P_TSH-CAC2634-XmaI (5′-AAAACCCGGGAGCTTTATCCTTACAG-3′) as primers, introducing BamHI and XmaI restriction sites (underlined). Following purification and restriction, the PCR fragment was exchanged for the hydA gene in vector pThydA, described previously by Girbal et al. (20), and the resulting vector, pTperRCac, was transformed into E. coli DH5α cells. Recombinant cells were selected by the addition of ampicillin and constitutively expressed PerR from the clostridial thiolase promoter with fused Strep-tagII at the C terminus for purification (see below).
For the complementation of B. subtilis strains, the perRCac gene was amplified from chromosomal DNA using P_perR-pX-1 (5′-AAAAGGATTCCACGTTTTCGAAAGCAAGG-3′) and P_perR-pX-2 (5′-AAAAGGATTCTTACAACTAGCAATATTTG-3′), introducing BamHI restriction sites (underlined). The amplified fragments were cloned into vector pX, which integrates in the amyE locus of B. subtilis and allows the xylose-inducible expression of proteins (37). The recombinant vector was transformed into B. subtilis strains HB1000 (wild type) and HB0509 (perR mutant) using a method described previously by Cutting and Youngman (10). Positive transformants were selected due to their resistance to 10 μg ml−1 of chloramphenicol.
C. acetobutylicum cell samples for the isolation of total RNA were pelleted by centrifugation, immediately shock-frozen in liquid N2, and stored at −70°C. RNA was isolated using a modified hot-phenol procedure described previously by Fischer et al. (17). To avoid DNA contamination, isolated RNA was digested with 20 U of RNase-free DNase (Amersham Pharmacia Biotech, Freiburg, Germany) in a total volume of 50 μl and incubated for 30 min at 37°C. The RNA was cleared from DNase by an addition of 15 μl of 2 M sodium acetate (pH 5.2), additional treatment with phenol, and precipitation with ethanol. Dried RNA was dissolved in 30 μl of H2O, and the quality of these preparations was controlled by electrophoresis in agarose gels. The integrity of RNA was additionally controlled in a BioAnalyzer (Agilent, Böblingen, Germany) run. For the preparation of labeled cDNA, 15 μg random hexamer primer was annealed in a volume of 10 μl to 25 μg of RNA by incubation at 70°C for 10 min. Next, 1 mM dATP, dTTP, and dGTP as well as 0.4 mM dCTP, 50 μM Cy3- or Cy5-labeled dCTP (GE-Healthcare, Munich, Germany), 10 mM dithiothreitol, and 200 U SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA) were added to the mixtures, and the labeling reaction mixtures with a total volume of 20 μl were incubated for 2 to 3 h at 42°C. RNA was removed from the formed heteroduplexes by the addition of 2 μl 2.5 M NaOH and incubation for 15 min at 37°C. Hydrolysis was stopped by the addition of 10 μl 2 M HEPES (pH 7.0). Labeled cDNA was separated from the reaction mixture using GFX columns (GE-Healthcare, Munich, Germany) according to the supplier's instructions, with the only modification of washing the columns four times before elution of labeled cDNA. The incorporation of Cy3 or Cy5 was checked qualitatively by a spectrophotometric wavelength scan and was quantified by using molar extinction coefficients of 150,000 liters mol−1 cm−1 (at 550 nm) for Cy3 and 250,000 liters mol−1 cm−1 (at 650 nm) for Cy5.
The C. acetobutylicum array was constructed by spotting 5′ amino-C6-modified oligonucleotides with a length of 60 to 70 bases on CodeLink microarray slides (SurModics) using a MicroGrid II microarray spotter (Zinsser Analytic, Frankfurt, Germany). Oligonucleotides were covalently coupled to the slides surface. The array contained two identical sets of 3,840 oligonucleotides representing 99.8% of all annotated open reading frames (ORFs) in C. acetobutylicum. Before hybridization, samples were denatured by incubation at 98°C for 5 min. The hybridization was done using Tom Freeman hybridization buffer (18) for 15 h at 45°C with cDNA containing approximately 80 pmol of Cy3 and Cy5 in an automatic Lucidea slide processor (GE-Healthcare, Munich, Germany). Slides were washed using a program applying consecutive washes two times with 1× SSC buffer (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) containing 0.2% sodium dodecyl sulfate (SDS) and then with 0.1× SSC. At the end, flushing of the hybridization chambers with isopropanol and evaporating of the isopropanol, which air dried the slides, were performed. Scanning was done using a GenePix 4000B microarray scanner (Molecular Devices, Canada) using GenePix Pro 6.0 software. Normalization was done by setting the arithmetic mean of the ratios equal to 1. Only features with fluorescence greater than the local background plus 1 standard deviation were included in the analysis. To correct for features with irregular spot morphology, only those where the ratio of medians, the ratio of means, and the regression ratio differed by less than 30% were included.
RNA was isolated from three independent cultures of C. acetobutylicum MGCcac15, which were either left anaerobic or exposed to air. RNA of the perR mutant was derived from three anaerobically growing cultures of the C. acetobutylicum ΔperR strain. RNA from C. acetobutylicum from anaerobic cultures was compared to RNA derived from aerated samples of the same strain and to RNA derived from anaerobic cultures of the C. acetobutylicum ΔperR strain. To avoid dye-specific effects, RNA from each sample was labeled with both dyes, and dye-swap experiments were performed. Resulting from these four technical replicates for each of the three independent cultures, a total of 12 sets of transcriptional data were obtained for each gene. The differential expression of an ORF was regarded as being significant when the average value of the ratio median from at least eight data points was either ≥2 or ≤0.5. Full lists of transcribed genes, functional information, and regulation under O2 or in perR-deleted cells are available in Tables S1 and S2 in the supplemental material. Microarray design and data are also available from the ArrayExpress database under accession no. A-MEXP-1561 (http://www.ebi.ac.uk/microarray-as/aer/entry). Microarray results for three genes for which transcription has not been previously identified to be either O2 or PerR responsive were validated by semiquantitative reverse transcription-PCR. The expression ratios of CAC0116, CAC2459, and CAC3657 detected by reverse transcription-PCR corresponded with those obtained by microarray analysis (see Fig. S1 in the supplemental material).
The Virtual Footprint software linked to the PRODORIC database (http://prodoric.tu-bs.de/vfp) (48) of gene regulation and gene expression in prokaryotes was used to identify DNA regions with a high level of similarity to the proposed regulatory motif IR2. The sequenced genome of C. acetobutylicum including megaplasmid pSOL1 was searched for the palindrome nucleotide sequence 5′-AATNNNTATTANNTAATANNNATT-3′, with “N” representing any type of nucleotide and allowing one mismatch. The search pattern was identified in 13 different positions on the chromosome. Two sites were located in the coding region of genes and 11 sites were in the 5′ noncoding region of genes within a 250-bp distance to the starting codon. Motifs with three mismatches were also identified in the upstream region of two other highly upregulated genes. A sequence logo representing a multilevel consensus sequence for each motif was achieved by an alignment of these sequences using Weblogo software (http://www.weblogo.berkeley.edu) (9).
PerR from C. acetobutylicum was heterologously expressed as a Strep-tagII fusion protein in E. coli cells carrying plasmid pTperRCac. Strep-tagII is an eight-residue minimal peptide sequence composed of Trp-Ser-His-Pro-Gln-Phe-Glu-Lys (IBA GmbH, Göttingen, Germany), which binds to streptavidin and allows protein purification using streptactin affinity chromatography (60). Cells were harvested by centrifugation and washed in 50 mM Tris HCl buffer (pH 8.0) with 150 mM NaCl and 1 mM EDTA. Pellets were either stored for up to 7 days at −20°C or immediately used as a source of protein. All further purification steps for the heterologous protein used in this study were carried out as described previously (56). The protein content in the elution fractions was determined using the Bradford assay (5).
A DNA fragment of the rbr3A-rbr3B promoter region, covering the last 235 bp of the rbr3A 5′ untranslated region, was PCR amplified using oligonucleotides P_3p_rbr3A (5′-AGTGTCTGCAGAAGCAGGGAAAAG-3′) and P_5p_rbr3A (5′-AATTTTCTAGATTAATCTCTCTCA-3′) as primers. Reactions were carried out with Pwo DNA polymerase (Peqlab, Erlangen, Germany) with chromosomal DNA from C. acetobutylicum ATCC 824 as a template. PCR products were purified from agarose gels with the Nucleospin Extract II kit (Macherey-Nagel, Düren, Germany) and 3′ digoxigenin labeled using the DIG gel shift kit (second generation; Roche Applied Science, Mannheim, Germany) according to the manufacturer's instructions. Labeled promoter fragments (12.5 nM) were incubated for 20 min at room temperature with 25 to 50 nM of purified PerRCac. Buffer contained 20 mM Tris HCl, 5% glycerol, 5 ng μl−1 salmon sperm DNA, 50 ng μl−1 bovine serum albumin, 50 mM KCl, and 100 μM MnCl2. Unlabeled fragments were added as controls for specific binding to DNA. The reaction mixtures were applied to 1.5% (wt/vol) agarose gels with 0.5× TBE (44.5 mM Tris-HCl [pH 8.0], 44.5 mM boric acid, 5 mM EDTA) as a running buffer. Electrophoreses were run at room temperature for 1 to 1.5 h at 65 V, and the DNA fragments were blotted onto nylon membranes (Nytran SuPerCharge nylon transfer membrane, 0.45-μm pore size; Schleicher & Schuell Bioscience, Dassel, Germany). The detection of the digoxigenin-labeled DNA was performed as described previously (44).
To determine the sensitivity to H2O2, a zone-of-inhibition assay was performed essentially as described previously by Bsat et al. (7). Fresh LB medium (50 ml) with the appropriate antibiotics was inoculated with B. subtilis from cultures grown overnight, and cells were grown at 30°C at 170 rpm on a rotary shaker to the mid-logarithmic growth phase (OD600 of 0.4 to 0.5). Aliquots of 1 ml were added to 15 ml of prewarmed (45°C) LB medium containing 1.5% (wt/vol) agar without antibiotics. The cell suspension was poured into agar plates, and after hardening, 5-mm filter paper discs containing 5 μl of 1 M H2O2 were placed onto the agar plates. Growth and zones of inhibition for each strain were monitored following overnight incubation at 30°C. Each strain was analyzed in triplicate. Results for each strain were reproducible within a range of ±15% of the maximum diameter of the inhibition zone.
B. subtilis cells were grown to the late exponential growth phase (OD600 of 1.0 to 1.2) in LB medium and harvested by centrifugation. Pellets were stored at −20°C or immediately used for the preparation of crude extracts. Cells were disrupted by sonication at 4°C using the Ultraschall Desintegrator Sonopuls HD60 apparatus (Medizin und Labortechnik, Hamburg, Germany), and undisrupted cells and cell debris were removed by centrifugation at 15,000 × g for 30 min. The protein concentration in the obtained cell extracts was determined with the Bradford assay (5). Catalase activity was measured spectrophotometrically at a wavelength of 240 nm. Different amounts of enzyme were added to 50 mM potassium phosphate buffer (pH 7.0) at 25°C and blanked. The reaction was started by the addition of H2O2 to the mixture up to a final concentration of 25 mM. The decrease in absorbance was monitored over time using an Ultrospec 3000 spectrophotometer (Pharmacia Biotech, Germany). One unit of catalase activity was defined as the amount of enzyme that decomposes 1 μmol of H2O2 min−1 using an 240 value of 43.6 mM−1 cm−1.
To identify O2-inducible genes, anaerobic cultures of C. acetobutylicum were sparged with compressed air at a rate of 0.05 liters min−1 liter−1 culture volume during the logarithmic growth phase. As a result of continuous uptake, the O2 concentration remained below detection limits (<6.25 μmol liter−1), and growth continued at essentially the same rate (data not shown), which reflects previously reported results (35, 51). Samples for the analysis of global transcription were taken following 1 h of microaerobiosis during the exponential growth phase. The level of gene expression of aerated cells was compared to that of a parallel culture that was left anaerobic. Oxygen exposure drastically affected (more than threefold) the transcription of 76 genes, resulting in 60 genes which were induced and 16 genes which were downregulated. The cluster of orthologous groups (COG) classification of proteins allowed a first functional overview of the products of all genes that were differentially expressed (more than twofold) upon exposure to O2 (67) (Fig. (Fig.1a).1a). However, a large proportion (27% of the upregulated genes and 20% of the downregulated genes) encoded proteins that could not be grouped or were of unknown function (cluster X). The most dramatic changes occurred with genes with proteins belonging to COG clusters C (energy production and conversion; 12% upregulated and 14% downregulated genes), E (amino acid transport and metabolism; 11% and 2%, respectively), G (carbohydrate metabolism and transport; 3% and 11%, respectively), H (coenzyme transport and metabolism; 3% and 11%, respectively), and P (inorganic ion transport; 5% and 9%, respectively). Overall, theses changes in global transcription suggested that C. acetobutylicum actively reacts to oxygen in the environment by concentrating its energy resources on the protection of essential metabolic pathways, maintenance of a reducing interior, and detoxification of ROS.
The group of genes that were highly activated included all those which encoded the proteins previously characterized as being components of an ROS detoxification pathway or as part of an oxygen-responsive gene cluster (25, 36, 57). These include, namely, reverse rubrerythrins (rbr3A-rbr3B), desulfoferrodoxin (dfx), rubredoxin (rd), NADH-dependent rubredoxin oxidoreductase (NROR), and the oxygen-reducing FDPs FprA1 and FprA2 (Table (Table2).2). High levels of expression of these genes represented an efficient indicator that the global transcriptional changes observed resulted from intense oxidative stress as result of O2 exposure. Other strongly activated genes whose products play a central role in the cell's redox balance included glutaredoxin- and thioredoxin-dependent systems (Table (Table2).2). The relatively large number of homologues and associated peroxidases suggests a vital role during oxidative stress. Interestingly, significant transcriptional changes under these conditions were not observed for the two annotated superoxide dismutases of the Cu/Zn (CAC2567) and Fe/Mn (CAC1363) types. Furthermore, the transcription of the rubrerythrins Rbr1 and Rbr2, which are characterized by their N-terminal ferritin-like domains, and the rubredoxin-like domains on the C terminus also did not respond to the presence of O2 (Table (Table22).
Anaerobes in particular are highly susceptible to oxidative damage to proteins and DNA via a direct O2-dependent inactivation of Fe-S clusters or Fenton-type chemistry (30). A more detailed view of the oxygen stimulon revealed that C. acetobutylicum not only induces the active detoxification of oxygen and reactive derivatives but also prompts the repair of essential cellular components. The repair of damaged DNA is activated by an induced transcription of genes encoding enzymes involved in the de novo synthesis of nucleotides, e.g., enzymes of the purine metabolism (CAC1390 to CAC1395 and CAC1655) and three ribonucleotide diphosphate reductases (CAC1047, CAC3276, and CAC3277), as well as those responsible for excision and replacement (uvrA and uvrB [CAC0502 and CAC0503, respectively]) (Table (Table2).2). As expected from the conflictive role of iron as a cofactor in defense enzymes and a catalyst in Fenton chemistry, iron uptake genes were strictly regulated upon exposure to oxygen. Two Feo-type uptake systems for Fe2+ and one ferrichrome-dependent system were upregulated (CAC0447 and CAC0448, CAC0788 to CAC0791, and CAC1029 to CAC1032), which could support the induced Suf and Nif machinery (CAC3288 to CAC3292) in the biogenesis of Fe-S clusters. Under the same conditions, a large operon (CAC1988 to CAC1994) that is putatively involved in the uptake of Fe3+ and molybdenum cofactor synthesis was repressed (Table (Table2).2). Reduced iron might become scarce due to an increasing level of production of reverse rubrerythrins and FDPs, but at this point, the differential regulation of Fe2+ and Fe3+ cannot be fully explained. Previous studies with E. coli have shown that moa genes for molybdenum cofactor production are enhanced under anaerobic conditions and underlie the control of the anaerobic activator protein FNR (1, 63). Molybdenum cofactors are known to catalyze reactions of the carbon, sulfur, and nitrogen cycles and act as sites of substrate binding and reduction in nitrogenases (38, 62).
An O2-sensitive PFOR, encoded by CAC2229 and CAC2499, is the central metabolic enzyme in C. acetobutylicum (46). Interestingly, the transcription of genes encoding the two subunits of an alternative α-ketoacid ferredoxin oxidoreductase (ofrB and ofrA [CAC2458 and CAC2459, respectively]) is highly induced under oxidative conditions (Table (Table2).2). The clostridial enzyme has not been studied thus far, but previously reported studies of related enzymes from the aerobic archaea Sulfolobus sp. strain 7 and Aeropyrum pernix demonstrated a wide substrate spectrum for ketoacids and, even more importantly, aerobic stability (49, 70). It is therefore likely that an oxygen-labile PFOR is replaced by an oxygen-resistant enzyme to feed the pool of acetyl coenzyme A (CoA).
The megaplasmid-carried sol operon (CAP0162 to CAP0164) was slightly upregulated upon exposure to O2 (Table (Table2).2). Its protein products are thought to be involved in the early phase of solvent formation (15). Essentially all other genes whose proteins are assigned to function downstream of acetyl-CoA in the formation of butyrate are moderately downregulated (two- to fivefold) under these conditions. There are two other enzymes that are activated by O2 and for which a central role in carbon metabolism is apparent (Table (Table2):2): one is the nonphosphorylating NADP-dependent glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (GapN), catalyzing the NADP-dependent oxidation from glyceraldehyde-3-phosphate to 3-phosphoglycerate (28). Supporting its functional role during glycolysis, genes encoding the enzymes responsible for the subsequent reactions in this pathway (triosephophateisomerase, phosphoglyceromutase, and enolase [CAC0711 to CAC0713]) are also moderately upregulated upon exposure to O2 (Table (Table2).2). The other highly activated gene translates into a protein which belongs to the Ni-containing carbon monoxide (CO) dehydrogenase (CODH) family, which includes actual CODHs and acetyl-CoA synthases of methanogenic and acetogenic organisms (12).
A peroxide repressor-like (PerR) protein was recently shown to act as a central regulator of the O2 detoxification system of C. acetobutylicum, as the full activation of its regulon triggered aerobic survival and even limited growth (25). To find targets of PerR regulation and gain insight into the gene expression profile of an aerotolerant strain, the global transcription levels in perR-deleted cells and cells carrying the wild-type allele were compared. The complete absence of PerR resulted in the highly differential expression (more than threefold) of 87 genes. Of these genes, 48 were upregulated in the perR-deleted strain, which overlapped the oxygen stimulon of the wild-type strain. Nearly half of the genes that were induced in perR-deleted cells were also found to be highly expressed when the wild type was exposed to oxygen for 1 h, while the remaining 28 genes were activated exclusively in the mutant strain (Fig. (Fig.1b).1b). All differentially expressed genes (more-than-twofold change in transcription) of the PerR mutant were also assigned to functional categories using COG. Similar to the results obtained for the O2 stimulon, the majority of differentially expressed genes (37% of the upregulated genes and 35% of the downregulated genes) encoded proteins that either could not be grouped or were of unknown function (cluster X). Among the others, those involved in energy production and conversion (cluster C; 11% upregulated genes and 5% downregulated genes), amino acid transport and metabolism (cluster E; 7.5% and 22%, respectively), and transcription (cluster K; 9% and 5%, respectively) were highly represented in a C. acetobutylicum strain lacking PerR (Fig. (Fig.1a).1a). The high number of 39 genes that were strictly downregulated (more than threefold) in the mutant strain demonstrated that PerR is essential for the expression of certain genes and/or might also have an activating role during anaerobic growth. Supporting these ideas, none of the lower-expressed genes in the mutant were found to be as lesser transcribed during oxygen exposure of the wild type (Fig. (Fig.1b).1b). Hence, the downregulation of genes in the perR mutant was apparently rather independent of the cell's oxygen response. There are only two exceptions to this point, which deserve some attention. The gene encoding CODH (CAC0116) was highly activated when wild-type cells were aerated, but the transcription of this gene was most strongly repressed in the PerR mutant strain (Table (Table2).2). Similarly, the gene encoding a glutathione peroxidase (bsaA2 [CAC1570]) was downregulated in the absence of PerR but was O2 inducible in wild-type cells, suggesting that an inactivated state of the PerR protein might be required in order to obtain a full expression of CAC0116 and bsaA2 upon exposure to oxygen.
A first study of the role of PerR in C. acetobutylicum identified a flavodoxin (CAC2452), the FDP FprA1 (CAC1027), and, most prominently, the twin-gene-encoded reverse rubrerythrin Rbr3A-Rbr3B (CAC3598 and CAC35977) as being highly expressed in the absence of PerR (25). The transcripts of all these proteins were among the most dramatically induced ones. The highest expression values for the perR-deleted strain were determined for the genes and operons encoding Rbr3A-Rbr3B (CAC3598-CAC3597), FprA1 (CAC1027), glutaredoxin (CAC2777), and NROR-FprA2-Dfx (CAC2448 to CAC2450). Fluorescence counts for these transcripts exceeded those for elongation factor Tu by a factor of up to 4.3 in the case of Rbr3A-Rbr3B, which was equivalent to saturation (>80,000) when using high-resolution scans (see Table S2 in the supplemental material). Therefore, the level of induction of these highly abundant transcripts might be underestimated (Table (Table2).2). Essentially all other genes previously reported as being members of anaerobic ROS detoxification or redox balance were also activated in the mutant, e.g., rubredoxin (CAC2778), the alkylhydroperoxidase Bcp (CAC3027), and a thiolperoxidase (CAC3306). Their induction in the absence of PerR even during anaerobiosis suggested that PerR is the key regulator that represses these proteins during the anaerobic growth of the wild type. Interestingly, the glutathione and thioredoxin antioxidant system, although activated by oxygen, seems to depend on an alternative regulatory mechanism, as no significant alterations in their transcript levels were detected in the PerR mutant strain (Table (Table2).2). Similar observations were made for the genes involved in nucleic acid repair or iron metabolism and/or transport, demonstrating that PerR is not the sole level of regulation and that as-yet-unidentified regulatory circuits exist. Only few genes of the central carbon pathways were differentially expressed in the perR mutant. These genes include the oxoacid ferredoxin oxidoreductase operon (CAC2458 and CAC2459), GapN (CAC3657), and the enzymes that catalyze the following two glycolytic reactions.
A total of four genes were strictly repressed (>10-fold-lower expression) in the mutant strain and, hence, strongly dependent on the presence of a functional PerR protein for expression. These included two genes with a putative role in the biosynthesis or transport of siderophores (CAP0029 and CAP0030), the glutathione peroxidase BsaA2 (CAC1571), and CODH (CAC0116). Only the latter two genes were highly O2 inducible in wild-type cells. Many genes encoding enzymes for the biosynthesis of arginine were moderately downregulated in the mutant, which suggests that arginine could function as a source of intracellular nitric oxide (NO) in C. acetobutylicum. Consequently, enzymes involved in the inhibition of NO production (N-dimethylarginine dimethylaminohydrolase [CAC0376]) and FDPs as reductive NO scavengers (26) are upregulated in PerR mutant and in O2-exposed wild-type cells. In summary, PerR acts primarily as a repressor of an O2-responsive, large-scale ROS defense machinery but also targets those O2-sensitive bottlenecks that might be crucial for a continuous demand of NAD(P)H and ATP.
An earlier study of the stress-dependent transcription of reverse rubrerythrins from C. acetobutylicum identified an inverted repeat of 24 nucleotides (IR2) in the 5′ untranslated region of rbr3A, for which a regulatory function was proposed (24). When reversed, this palindromic sequence shares a high level of similarity to the known consensus binding sequences of the peroxide repressor PerR from B. subtilis and Staphylococcus aureus (Fig. (Fig.2a).2a). In the closely related facultative aerobe B. subtilis, PerR derepresses the transcription of multiple target genes upon its metal-catalyzed inactivation with H2O2 (40). To further support the idea that IR2 was the site of PerR regulation and to identify other potential target genes, a genome-wide search for IR2 was performed using Virtual Footprint software (48). When using exclusively the palindromic nucleotides and allowing a single mismatch, a total of 10 genes that carried a potential PerR box within 200 bp of their start codon were found. Nine of them were among those with the highest level of induction in the perR mutant, including the ones encoding FprA1, the entire O2-responsive operon with NROR, FprA2, and Dfx, the glutaredoxin-rubredoxin operon, and both subunits of an oxoacid-ferredoxin oxidoreductase (Table (Table2).2). The O2-induced gapN gene, encoding a nonphosphorylating NADP-dependent GAPDH, was also highly expressed in the absence of PerR and preceded by a PerR box with three mismatches. Several conserved nucleotide positions were detected for this regulatory module, which could be illustrated as a sequence logo (Fig. (Fig.2b2b).
To determine if the clostridial PerR homologue directly interacts with the promoter region of rbr3A, its gene was heterologously overexpressed in E. coli cells. The PerR-encoding ORF CAC2634 was cloned into plasmid pT to obtain a fusion protein with C-terminal Strep-tagII. SDS-polyacrylamide gel electrophoresis of the purified fraction yielded two bands, corresponding roughly to the calculated molecular mass of the Strep-tag fusion protein monomer of 17.1 kDa. The small size variation as well as a faint upper band might reflect different oxidation states, which were previously observed for the B. subtilis peptide (41) (Fig. (Fig.3a).3a). The purified protein was incubated with a labeled DNA fragment of the rbr3A promoter region covering the last 235 bp of the 5′ untranslated region and IR2. When these reaction mixtures were applied for gel electrophoresis, the mobility of this fragment was reduced, depending on the amount of PerR in the assay (Fig. (Fig.3b,3b, lanes 1 to 3). This shift in DNA mobility could be reversed stepwise following the addition of unlabeled fragment, giving evidence that the binding of PerR was DNA specific (Fig. (Fig.3b,3b, lanes 4 and 5).
The conserved binding motif and interaction with the promoter region of rbr3A-rbr3B suggested that the mode of PerR regulation is conserved in C. acetobutylicum albeit with different target genes compared to those of facultative aerobes. To monitor PerR binding in vivo and show that the PerR protein of C. acetobutylicum might be a peroxide sensor rather than a direct O2 sensor, complementational studies were performed. The clostridial PerR protein was expressed from a xylose-inducible promoter in wild-type B. subtilis and a strain lacking a functional perR gene. Subsequently, their sensitivity to H2O2 was monitored in a zone-of-inhibition assay (Fig. (Fig.4).4). In the presence of xylose, the wild-type B. subtilis strain expressing PerRCac differed only marginally from its parent in its sensitivity to H2O2, which might reflect the higher affinity of B. subtilis PerR (PerRBsu) for its own binding site (Fig. 4a and b). B. subtilis cells lacking their own perR gene are hyperresistant to H2O2 due to an increased production of defense proteins (7). Following the expression of PerRCac, this hyperresistant phenotype was converted to a hypersensitive one, giving evidence that the protein from C. acetobutylicum can indeed acts as a repressor despite aerobic growth (Fig. 4c and d). The peroxide sensitivities of all strains were reflected largely by their catalase activities during growth in aerobic liquid cultures. Wild-type cells showed normal levels of catalase that were even moderately reduced when expressing PerRCac, but catalase was nearly 100-fold more active in the perR mutant strain (Fig. (Fig.5).5). When, in turn, the clostridial peptide was produced in the cells, catalase activity was reduced by more than 200-fold, to levels far below those of wild-type cells (Fig. (Fig.5).5). Low doses of H2O2 were able to partially restore catalase activity (data not shown), indicating that the key regulator of oxygen defense in C. acetobutylicum senses intracellular peroxide levels rather than molecular O2.
The finding that oxygen-sensitive enzymes are widely used as central components in the metabolism of anaerobes as well as the existence of a robust defense system has led to a refinement of the traditional model of obligate anaerobiosis. An energy metabolism that relies exclusively on the fermentation of highly reduced substrates demands enzymes with low redox potentials and, thus, an environment that is predominantly anaerobic. C. acetobutylicum flourishes under fully anaerobic conditions and even maintains growth at a reduced rate when the influx of oxygen is compensated by its continuous reduction (35). Only when dissolved oxygen accumulates are central metabolic enzymes like PFOR wrecked at a rate at which metabolism is halted and viability decreases. In a natural environment, exposure to O2 might occur gradually and, thus, allows an efficient adaptation (31). This has just recently been shown for a number of other bacteria from a diverse group of anaerobes, e.g., Bacteroides fragilis, D. vulgaris, and Thermotoga maritima (3, 42, 47). The experimental setup presented here was designed so that the level of influx of O2 was low enough to ensure metabolic activity but sufficient to generate substantial cellular damage. Microaerobic growth as well as the deletion of a central regulator in oxidative stress resulted in global changes in gene expression under O2, leading to a preliminary model on how C. acetobutylicum attempts to combine O2/ROS removal, damage repair, and adjustments in central energy metabolism to avoid paralysis (Fig. (Fig.66).
The production of proteins involved in detoxification and cellular redox balance was most drastically activated, which was expected, as the upregulation of these components was observed recently for a number of bacteria that are considered to be anaerobes (42, 47, 66). The massive expression of all components of the proposed oxygen and ROS detoxification system, which involves FDPs, Dfx, NROR, rubredoxin, and reverse rubrerythrins, emphasized its vital role in survival. Consequently, all of these components were found be the primary targets of PerR repression, as they were overexpressed in the perR-deleted strain and shared a common regulatory motif. Thus, it seems likely that PerR acts as a hypersensitive regulator designed to activate these genes even in the presence of traces of oxygen. The acute induction of this system in a natural environment will also support the restoration of anaerobiosis by a complete uptake of residual oxygen.
PFOR could be a major bottleneck in glucose degradation and acetyl-CoA production under microaerobic conditions. An enzyme with a higher resistance to oxidative inactivation would clearly be of advantage during oxygen exposure. At this time, induced oxoacid-ferredoxin oxidoreductase has not been functionally characterized, but the use of similar enzymes in aerobic archaea makes it tempting to speculate that a replacement occurs under these conditions (49, 70). A similar mechanism could be postulated for the nonphosphorylating GAPDH (GapN), as an increased resistance to H2O2 has been shown for the homologous protein from the aerotolerant anaerobe Streptococcus mutans (2). Interestingly, oxoacid-ferredoxin oxidoreductase and GapN seem to be two of the very few proteins that function outside the detoxification system and are still subject to direct PerR repression. This indicates their significance during the early response to oxygen, while glutathione- and thioredoxin-dependent rescue systems were PerR independent and could belong to circuits that protect from persistent exposure to oxidative conditions.
In an oxidizing environment, C. acetobutylicum seems to direct its electron flow toward the higher-oxidized product acetate or acetone. This is in agreement with a slightly increased level of acetate production by C. acetobutylicum upon exposure to O2, which was observed during a previously reported investigation (51). The results presented here demonstrate that essentially all genes in the formation of butyrate downstream from acetyl-CoA are downregulated, suggesting that this pathway is less favorable when an oxidized dinucleotide pool is regenerated during the reduction of O2 and ROS. At the same time, we observed a moderate upregulation of the sol operon, which might integrate in the larger general stress network, finally inducing solvent formation and sporulation.
Another enzyme with highly activated expression upon exposure to O2 is a member of the CODH protein family. When the amino acid sequences of the oxygen-induced proteins from C. acetobutylicum were compared to the known homologues of Carboxydothermus hydrogenoformans, the clostridial peptide shows the lowest similarity to the acetyl-CoA synthase subunits (data not shown). Furthermore, the absence of genes encoding any other subunit of acetyl-CoA synthases makes it rather unlikely that this protein is involved in CO2-dependent acetyl-CoA formation. A role for these proteins in the oxidative stress response has not been established, but a homologue from C. hydrogenoformans was speculated to function in ROS detoxification due to its genomic position in an operon with rubrerythrin and NROR (69). Furthermore, it cannot be excluded at this point that the clostridial protein also serves a protective function in the hydrogen production of this organism. H2-evolving hydrogenase A1 (HydA1) of C. acetobutylicum is highly sensitive to oxygen-dependent inactivation (11, 20). While low doses of molecular O2 lead to irreversibly damaged enzymes, CO was previously shown to act as a competitive inhibitor that antagonizes O2 and can greatly reduce the hydrogenase inactivation state (16, 43). The completely opposite expression patterns of this protein in both strains partially support this idea: the gene encoding CODH showed a high level of activation when the wild type was exposed to O2 but the lowest level of transcription in anaerobic perR-deleted cells. The opposed expression of this gene in both strains addresses another interesting issue of PerR regulation. It does not seem unlikely that PerR could also function as a gene activator in its oxidized form via an as-yet-undefined mechanism. Direct gene activation by PerR was previously seen for the B. subtilis srfA operon (21). ROS-induced transcriptional activation, although previously assigned to SoxR (OxyR), was more recently also demonstrated for the Borellia oxidative stress regulator BosR, a homologue of B. subtilis PerR (4, 23, 71).
The regulatory role in the peroxide stimulon as well as the mechanism of inactivation by H2O2 in a Fenton-type reaction have been thoroughly described for the B. subtilis peptide (7, 19, 40). In the anaerobe C. acetobutylicum, the external addition of H2O2 was not required to activate the organism's PerR regulon, while even low doses of O2 were an adequate stimulus. The complementation of a B. subtilis strain lacking functional PerR with PerRCac provided evidence that the clostridial protein was a fully functional homologue of PerR in vivo. The presence of the clostridial peptide in a strain lacking PerRBsu repressed the expression of oxidative stress proteins during aerobic growth and resulted in a peroxide-hypersensitive phenotype. Low doses of external H2O2 inactivated PerRCac and derepressed catalase. However, as indicated from the proposed regulatory sequence motif for C. acetobutylicum, the level of specificity of PerRCac for the B. subtilis PerR boxes seems to be reduced. When PerRCac was expressed in wild-type B. subtilis cells, only a small increase in H2O2 sensitivity was determined, presumably as a result of the preferred binding of PerRBsu. Nevertheless, these data demonstrate that PerRCac could act as a peroxide sensor in vivo. Previous observations estimated that upon aeration, the intracellular H2O2 concentration experienced by strict anaerobes could be substantially higher than that in aerobes due to the oxidation of low-potential flavoenzymes (30). Especially for a strict anaerobe, it seems beneficial that the presence of O2 is indirectly perceived by elevated levels of H2O2. The use of H2O2 as a signal molecule also allows the integration of metabolic activity, which determines the extent of ROS production. Consequently, C. acetobutylicum does not rely solely on their removal but also attempts to adapt its fermentation metabolism to lesser-reduced environments (Fig. (Fig.6).6). However, a broad understanding of its physiology during O2 exposure will require further investigations, especially with respect to the biotechnological potential of this organism.
This work was supported in part by SysMO project COSMIC (http://www.sysmo.net) (to A.E., R.-J.F., and H.B.).
We thank John D. Helmann and Achmed Gaballa from the Department of Microbiology at Cornell University for B. subtilis strains HB1000 and HB0509. We are also grateful to Jim Imlay for helpful comments on the manuscript.
Published ahead of print on 31 July 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.