Deletion of the rbo Gene Increases the Oxygen Sensitivity of the Sulfate-Reducing Bacterium Desulfovibrio vulgaris Hildenborough

Abstract

The rbo gene of Desulfovibrio vulgaris Hildenborough encodes rubredoxin oxidoreductase (Rbo), a 14-kDa iron sulfur protein; forms an operon with the gene for rubredoxin; and is preceded by the gene for the oxygen-sensing protein DcrA. We have deleted the rbo gene from D. vulgaris with the sacB mutagenesis procedure developed previously (R. Fu and G. Voordouw, Microbiology 143:1815–1826, 1997). The absence of the rbo-gene in the resulting mutant, D. vulgaris L2, was confirmed by PCR and protein blotting with Rbo-specific polyclonal antibodies. D. vulgaris L2 grows like the wild type under anaerobic conditions. Exposure to air for 24 h caused a 100-fold drop in CFU of L2 relative to the wild type. The lag times of liquid cultures of inocula exposed to air were on average also greater for L2 than for the wild type. These results demonstrate that Rbo, which is not homologous with superoxide dismutase or catalase, acts as an oxygen defense protein in the anaerobic, sulfate-reducing bacterium D. vulgaris Hildenborough and likely also in other sulfate-reducing bacteria and anaerobic archaea in which it has been found.

Sulfate-reducing bacteria (SRB) can have a considerable impact on their environment, because their growth is coupled to the production of large amounts of hydrogen sulfide. This activity is important in the removal of acidic, oxidized forms of sulfur (e.g., SO₂) from the environment and in the immobilization of toxic metal ions, e.g., as present in acid mine drainage effluents. Despite these essential, environment-restoring properties of SRB they are considered a nuisance in many environments due to the odor, toxicity, and metal-corroding properties of their respiratory end product. Oxygen is one of the best and cheapest agents for controlling the growth and activity of SRB in environments in which they are not wanted (21). The survival of SRB in aerobic environments has, therefore, already been studied. Hardy and Hamilton credited endogenous superoxide dismutase (SOD) and catalase activity for the survival of Desulfovibrio spp. in oxygenated waters from the North Sea (8). The presence of an Fe-containing SOD in Desulfovibrio desulfuricans had been previously demonstrated by Hatchikian and Henry (9). These enzymes may also repair damage arising from microaerophilic growth (9a). Pianzzola et al. attempted to clone the sod gene from the SRB Desulfoarculus baarsii by functional complementation of a SOD-deficient mutant of Escherichia coli (19). Sequence analysis of the resulting clones indicated that these contained the rbo and rub genes of D. boarsii, and further complementation studies indicated that only rbo was required for complementing the sod phenotype. The rbo gene is widespread in SRB and anaerobic archaea, and the amino acid sequence of the Rbo protein has remained remarkably conserved (Fig. 1). In order to elucidate whether its function is indeed in the prevention or repair of oxygen damage, as suggested by the heterologous complementation studies with E. coli, or whether it also plays a role under anaerobic conditions, we have constructed an rbo deletion mutant of Desulfovibrio vulgaris Hildenborough, of which the properties are reported here.

FIG. 1.

Comparison of amino acid sequences of Rbo from D. vulgaris Hildenborough (Rbodvh), D. vulgaris Miyazaki F (Rbomya), D. desulfuricans (Rbodde), Desulfoarculus baarsii (Rbodab), Archaeoglobus fulgidus (Rboarf), and Methanobacterium thermoautotrophicum (Rbomta), as reported in references 2, 3, 11, 12, 19 and 27. Residues identical to or deleted from the D. vulgaris sequence are indicated by dots and tildes, respectively. Residues that are conserved in all 6 sequences are indicated in the consensus sequence. DNA encoding amino acids 49 to 78 (underlined) was deleted from the rbo gene in the construction of pNotΔRbo.

MATERIALS AND METHODS

Materials.

Restriction and DNA modification enzymes and bacteriophage λ DNA were obtained from Pharmacia. [α-³²P]dCTP (10 mCi/ml; 3,000 Ci/mmol) was from ICN. Mixed gas (85% [vol/vol] N₂, 10% [vol/vol] CO₂, and 5% [vol/vol] H₂) was from Praxair Products Inc. Reagents for the construction and purification of a MalE-Rbo fusion protein (expression vector pMALc2, factor Xa protease, and amylose resin), as well as anti-mouse immunoglobulin G alkaline phosphatase-linked antibody, were from New England Biolabs. Other immunoblotting reagents (nitroblue tetrazolium and 5-bromo-4-chloro-3-indolylphosphate) were from Promega, whereas Hybond-N membrane filters were obtained from Amersham. Reagent grade chemicals were from either BDH, Fisher, or Sigma. Deoxyoligonucleotide primers were obtained from University Core DNA Services of the University of Calgary.

Bacterial strains, plasmids, and growth conditions.

Strains, plasmids, and primers used or constructed in this study are listed in Table 1. E. coli and D. vulgaris strains were grown as described elsewhere (5, 21, 22, 30).

TABLE 1.

Bacterial strains, plasmids, and primers

Strain, plasmid, or primer	Relevant characteristics	Reference or source
D. vulgaris
Hildenborough	NCIMB 8303; wild type; Kmr Cms Sucr	21
F100	dcrA gene replaced with a cat gene cassette from pUC19Cm; Kmr Cmr Sucr	5
R1	pNotΔRboCmMOB integrated into the chromosome; Kmr Cmr Sucs	This study
R1SR	R1 derivative with mutated sacB gene: Kmr Cmr Sucr	This study
L2	rbo gene replaced with a cat gene cassette from pUC19Cm; Kmr Cmr Sucr	This study
E. coli
TG2	Δ(lac-pro) supE thi hsdM hsdR recA F′ (traD36 proAB+lacZΔM15Iq); used for general molecular biological work	24
S17-1	thi pro hsdR recA with RP4-2[Tc::Mu,Km::Tn7] in the chromosome; mobilizer strain	26
Plasmids
pJK29	rbo-rub operon on a 1.1-kb SalI fragment in pUC8; Apr	2
pJK34	2.6-kb EcoRI fragment starting at nt 3067a; cloned in pUC8	This study
pMOB2	Containing an oriT-sacBR cassette on a 4.8-kb NotI fragment; Kmr	25
pNOT19	pUC19 with 10-bp NdeI-NotI adapter in NdeI site; Apr	5
pSUP104	Broad-host-range vector; Cmr	23
pUC8, pUC19	Cloning vector, pMB1 origin of replication; Apr	24
pUC19Cm	pUC19 containing a 1.4-kb SacII-TthIII fragment from pSUP104 with the cat gene in its BamHI site; Apr Cmr	5
pNotRboI	1.1-kb SalI fragment from pJK29 in SalI site of pNOT19; Apr	This study
pNotRboI′	pNotRboI with BamHI site deleted from the polylinker; Apr	This study
pNotΔRbo	pNotRboI′ with part of the rbo gene replaced by a BamHI-containing sequence; Apr	This study
pNotΔRboCm	cat gene of pUC19Cm inserted in the BamHI site of pNotΔRbo; Apr Cmr	This study
pNotΔRboCmMOB	oriT and sacBR containing NotI fragment of pMOB2 cloned into NotI site of pNotΔRboCm; Apr Cmr Sucs	This study
Primers
P120-Δrbo-r	p-cccggatccCTTTTCCTTGGCCCCGTCAGA (uppercase is not 2672–2652a; lowercase is a BamHI linker)	This study
P121-Δrbo-f	p-TGGATTGAGCTTGTCGCAGACGGT (nt 2763 to 2786a)	This study
P122-rbo-f	p-ATGCCCAACCAGTACGAAAT (nt 2529 to 2549a)	This study
P123-rbo-r	p-CATCGTGGATTCCTCGGGGTT (nt 2928 to 2908a)	This study
P130-r	p-GAAGTCGCGGCTGTTGTGGTCGAC (nt 3282 to 3259b)	This study
P127-f	p-GAGGGCATGGCCCAGAGGCTTGAGGCCCT (nt 2138 to 2166a)	This study
P129-cat	p-CAGGAAGATACTTAACAGGGAAGT (nt 582 to 605c)	This study
P128-cat	p-GAGTGGCAGGGCGGGGCGTAATTTT (nt 639 to 663c)	This study

^aNumbers are for the sequence of the dcrA rbo rub region (nt 1 to 3264) of accession no. M81168. 

^bSequence determined by extension of M81168. 

^cRelative to start codon of the cat gene. 

Construction of an rbo deletion mutant.

The strategy used for gene replacement mutagenesis was similar to that described previously (5, 10). Plasmid pNotΔRboCmMOB was transferred from E. coli S17-1 to D. vulgaris by conjugation by a filter mating method (5, 22). Aliquots (50 to 100 μl) of the resuspended mating mixture were plated onto medium E with kanamycin and chloramphenicol (CAM). The plates were incubated at 35°C for 5 to 7 days to select Cm^r Km^r transconjugal integrants. D. vulgaris R1, in which pNotΔRboCmMOB had been integrated downstream from the rbo-rub operon (see Fig. 2), was purified from contaminating E. coli by plating on the same medium. D. vulgaris R1 was next grown in medium C with CAM and sucrose. Growth was monitored with a Klett meter and was slow relative to that of D. vulgaris F100, a strain that is both Suc^r and Cm^r, in the same medium. At midsaturation, 200-μl aliquots of this culture, diluted either 10²- or 10⁴-fold, were plated on medium E with CAM. Fifty isolated colonies were picked and grown in 1 ml of medium C with CAM. Aliquots (0.5 ml) of these cultures were used to inoculate 5 ml of medium C with CAM and 5 ml of medium C with CAM and sucrose. Observation of similar Klett readings for the two cultures after growth to saturation was considered evidence that the picked colony was Cm^r and Suc^r. The two cultures were then combined; 0.5 ml was inoculated into 20 ml of medium B with kanamycin and CAM, grown to saturation, and stored at 4°C. The remainder was used for DNA isolation according to the method of Marmur (17).

FIG. 2.

Maps of the dcrA and rbo-rub region in D. vulgaris wild type, R1, and L2. (WT) Numbering is as for accession no. M81168. The 1.1-kb SalI (S) fragment is divided in the upstream (up), downstream (down), and deleted (Δ) regions. The hybridization positions and directions of polymerization of several primers are shown. (R1) Plasmid pNotΔRboCmMOB is located at nt 2800 to 12600. The deleted region (Δ) of the rbo gene is replaced by a 1.4-kb BamHI (B) insert (I) containing the cat gene. P128 and P129 are cat-specific primers. The oriT and sacB genes (from pMOB2) and the bla gene (from pUC) are located on a 7.4-kb fragment. (L2) Replacement mutant lacking a functional rbo gene.

Southern blot and PCR analyses for identification of the rbo deletion mutant.

The DNAs from Cm^r and Suc^r cultures were digested with PstI, separated by agarose gel electrophoresis, and transferred to Hybond-N membrane filters. The blots were then hybridized with a sacB probe, obtained as a 2.4-kb XbaI fragment from plasmid pMOB2, and ³²P labeled by extension of random hexamers as described previously (31). After washing and drying, the radioactive images of the blots were displayed with a Fuji BAS 1000 Bioimaging Analyzer. DNAs that did not hybridize with sacB were further tested by PCR with primers P122-rbo-f and P123-rbo-r. PCR amplification was done with a Perkin-Elmer Gene Amp 2400 PCR system using TaqI DNA polymerase and reagents, as indicated elsewhere (28). Reaction was for 30 cycles of 94°C (30 s), 60°C (30 s), and 72°C (90 s). This allowed efficient amplification of PCR products in the 2-kb range. The reaction time at 72°C was shortened for primer combinations which yielded only smaller PCR products (0.4 to 0.7 kb). D. vulgaris L2 was selected as the desired replacement mutant.

Protein blotting.

Expression of rbo was monitored by protein blotting with polyclonal antibodies generated in mice against Rbo, overexpressed in E. coli as a MalE-Rbo fusion protein, and purified by affinity chromatography on amylose resin, according to procedures suggested by the manufacturer. The MalE and Rbo portions of this fusion protein could be separated by proteolysis with factor Xa protease. Cells of D. vulgaris wild type, F100, and L2 were grown in 5 ml of medium C. The cells were suspended in 300 to 350 μl of water (depending on the measured cell density), an equal volume of sodium dodecyl sulfate (SDS)-containing incubation buffer was added, and the samples were placed immediately in a boiling water bath. The samples were then subjected to SDS-polyacrylamide gel electrophoresis with 15% (wt/vol) polyacrylamide gels, according to the method of Laemmli (13). Separated proteins were electroblotted onto nitrocellulose (29), and the blots were incubated sequentially with gelatin-containing blocking buffer and the Rbo-recognizing primary antibody. Bound primary antibodies were detected with an alkaline phosphatase-conjugated anti-mouse secondary antibody and immunoblot staining reagents (20).

Survival in air.

Cultures (5 ml) of D. vulgaris L2 and the wild type were grown anaerobically in medium C overnight. The cell densities were verified with a Klett meter. For growth under anaerobic conditions, identical inocula (ca. 50 μl of 10⁹ CFU/ml) were diluted into 5 ml of medium C in a 13- by 100-mm tube, after which growth was monitored with the Klett meter. For exposure to air, identical inocula (ca. 5 ml, depending on the measured cell density) were diluted into 500 ml of medium C stirred continuously in air with a magnetic stirrer. Samples of 5 ml of these aerobically incubated cells were pipetted periodically into 13- by 100-mm tubes. These were transferred to anaerobic conditions, after which growth was monitored with the Klett meter. D. vulgaris wild type and L2 do not grow under aerobic conditions. Also, 100-μl aliquots, as well as 100-μl aliquots of 10² and 10⁴ dilutions, of several of these samples were plated immediately after transfer to anaerobic conditions onto medium E plates. The number of surviving CFU per milliliter was determined from these plates by counting colonies after 1 week of incubation at 35°C under anaerobic conditions.

RESULTS

Construction and application of the suicide vector.

pNotRboI, consisting of a modified pUC vector of 2.6 kb and a 1.1-kb SalI fragment containing the rbo-rub operon, was used as the starting point for directed mutagenesis. The rbo-rub operon is located downstream from the 3′ end of the dcrA gene (4) on this 1.1-kb fragment (Fig. 2, WT). PCR of pNotRboI (3.7 kb) with primers P121-Δrbo-f and P120-Δrbo-r gave a 3.6-kb product. In plasmid pNotΔRbo, obtained following ligation and transformation of this PCR product, 90 bp from the rbo coding region (Fig. 2; WT Δ) was replaced by a BamHI linker. Insertion of the 1.4-kb cat gene and 4.8-kb oriT sacB cassettes gave pNotΔRboCmMOB, a plasmid of 9.8 kb. The identity of this plasmid was confirmed as follows. (i) PCR amplification with P122-rbo-f and P123-rbo-r gave a 1.7-kb product, 1.3 kb larger than the 0.4-kb product obtained with pJK29 or wild-type chromosomal DNA. (ii) Digestion with BamHI released a 1.4-kb cat-gene-containing insert (Fig. 2, R1 and L2 I). (iii) Digestion with NotI gave two similar-sized fragments of 5.0 and 4.8 kb. (iv) The sacB gene could be released by PstI digestion as a 2.6-kb fragment (6). In the map of D. vulgaris R1 (Fig. 2), pNotΔRboCmMOB extends from nucleotides (nt) 2800 to 12600.

We planned to use PCR to monitor the formation of new DNA junctions by homologous recombination of plasmid pNotΔRboCmMOB with the D. vulgaris chromosome. Primers P128-cat and P129-cat, hybridizing with the cat gene insert, and P127-f and P130-r, designed to hybridize immediately outside the 1.1-kb SalI fragment (Fig. 2), were synthesized for this purpose. Synthesis of P130-r required additional sequencing, because the available sequence information did not extend beyond the rightmost SalI site. Sequencing of plasmid pJK34, which contained a 2.6-kb EcoRI insert extending rightward from nt 3067 in the map of the wild type in Fig. 2, gave the required information. Chromosomal DNA from a putative plasmid integrant, D. vulgaris R1, was subjected to PCR with various primer pairs. Only the use of P128-cat and P130-r gave the expected 590-bp PCR product (not shown), indicating that integration had occurred through homologous recombination of the downstream regions (Fig. 2, R1). The 590-bp product was not formed when wild-type chromosomal DNA was used for PCR.

Verification of the D. vulgaris L2 genotype.

The desired L2 genotype can, in theory, be distinguished from wild-type, R1, and R1SR strains by the formation of a characteristic 650-bp PCR product with primers P127-f and P129-cat (Fig. 2). Cultures of D. vulgaris R1 in medium C with CAM and sucrose were, therefore, plated on medium E with CAM, and 30 colonies were toothpicked directly into PCR mix containing this primer pair. However, following amplification, formation of the 650-bp PCR product was not clearly shown for any of these. DNAs isolated from 37 Cm^r and Suc^r colonies were therefore digested with PstI and analyzed by Southern blotting, with the radiolabeled sacB gene as the probe. The results for 11 colonies are shown in Fig. 3. The amounts of digested DNA loaded were identical for all 11 samples, as indicated by ethidium bromide staining prior to blotting (not shown). Only samples L2 and L6 lacked sacB hybridization, indicating that these were candidates for the desired homologous recombination through the upstream regions. Samples L4, L5, L9, and L11 showed hybridization of a 2.6-kb PstI fragment with the sacB probe, similar to that observed for D. vulgaris R1 (not shown). Samples L3, L7, L8, and L10 had a 3.8-kb hybridizing PstI fragment, indicating ISD1 insertion (6), whereas L1 showed both hybridizing bands. PCR analysis of chromosomal DNA from D. vulgaris L2 with primers P122-rbo-f and P123-rbo-r indicated exclusively a 1.7-kb product, whereas wild-type DNA gave exclusively a 0.4-kb product when amplified under the same conditions (Fig. 4). Chromosomal DNA from D. vulgaris R1 gave both the 0.4- and 1.7-kb products (Fig. 4). These results are in agreement with the maps shown in Fig. 2. PCR amplification of purified chromosomal DNA from D. vulgaris L2 with primers P127-f and P129-cat gave a weak 650-bp band that was not seen when DNA from the wild type or D. vulgaris R1 was used. This product was not obtained when toothpicked colonies of D. vulgaris L2 were used as a template for PCR, explaining the failure of our earlier attempts at PCR screening of Cm^r and Suc^r colonies.

FIG. 3.

Southern blot of PstI-digested chromosomal DNAs of Cm^r and Suc^r derivatives of D. vulgaris R1. The blot was hybridized with a radiolabeled sacB probe. M, bacteriophage λ DNA digested with HindIII. The sizes of the bands are indicated in kilobases.

FIG. 4.

PCR analysis of chromosomal DNAs from D. vulgaris wild type, R1, and L2. DNA was amplified with primers P122-rbo-f and P123-rbo-r. The PCR products were run on an 0.7% (wt/vol) agarose gel which was stained with ethidium bromide. M, 100-bp marker ladder.

Protein blotting confirmed the absence of the rbo gene from D. vulgaris L2. Identical amounts of cells were loaded for D. vulgaris F100, a dcrA deletion mutant that overexpresses rbo (5), for the wild type, and for D. vulgaris L2. Incubation of the blot containing the SDS-polyacrylamide gel electrophoresis-separated proteins with the Rbo-specific antibody indicated reaction with purified 14-kDa Rbo (Fig. 5, lanes 7 to 10), as well as with Rbo in D. vulgaris F100 and the wild type (Fig. 5, lanes 1 and 2 and lanes 3 and 4, respectively). D. vulgaris L2 clearly lacked an immunoreactive 14-kDa protein (Fig. 5, lanes 5 and 6).

FIG. 5.

Protein blotting of cells of D. vulgaris F100 (lanes 1 and 2), wild type (lanes 3 and 4), and L2 (lanes 5 and 6). Identical amounts of cells, corresponding to ca. 150 μg of protein (lanes 1, 3, and 5) or 75 μg of protein (lanes 2, 4, and 6) were loaded together with ca. 50, 100, 200, and 400 ng of purified Rbo (lanes 7 to 10).

Phenotype of D. vulgaris L2.

The growth curves of D. vulgaris wild type and L2 under anaerobic conditions were very similar (Fig. 6A), indicating a similar doubling time and final cell density. Growth of liquid cultures that had been exposed to air proceeded only after a lag time, defined as the time required for the aerated culture to reach midsaturation minus 18 h, which was the time required for the anaerobic culture to reach midsaturation (Fig. 6A). For 2 h of air exposure, lag times of 14 and 53 h were observed for D. vulgaris wild type and L2, respectively (Fig. 6B). Lag times were compared for 25 pairs of 5-ml cultures of D. vulgaris wild type and L2, which were exposed to air for 1 to 36 h prior to transfer to the anaerobic hood. These received inocula of ca. 10⁷ cells/ml and grew identically without exposure to air (as in Fig. 6A). The lag times for D. vulgaris L2 exceeded those for the wild type in 19 of the 25 experiments. This included 13 cases in which D. vulgaris L2 failed to regrow. The lag times for D. vulgaris wild type exceeded those for L2 in 2 of the 25 experiments, including one case in which the wild type failed to regrow. No conclusions could be drawn for 4 of the 25 experiments because both the wild type and D. vulgaris L2 failed to regrow. These data indicated a significantly increased sensitivity of D. vulgaris L2 to inactivation by oxygen compared to the wild type.

FIG. 6.

Effect of air on the growth of D. vulgaris wild type (□) and L2 (◊) in liquid culture. Medium C (5 ml) was inoculated with 50 μl of an overnight culture (ca. 10⁹ CFU/ml). Growth was then monitored under strictly anaerobic conditions (A) or following 2 h of exposure to air (B). The zero time point for the growth curves in panel B is the time at which the cultures were returned to anaerobic conditions.

The survival of the two strains following exposure to air was also compared in plating experiments. Plating after air exposure led to a wider range of colony sizes than observed for plating of cells that were not exposed to air. New small colonies emerged on plates containing air-exposed cells even after 4 to 5 days, whereas anaerobically kept cells grew to a uniform colony size in 2 to 3 days. This made evaluation of remaining CFU per milliliter more difficult than we had anticipated. The results of an experiment in which all colonies visible after 1 week of incubation without magnification were counted are shown in Fig. 7. Following 24 h of air exposure, the number of L2 survivors was 100-fold smaller than that of the wild type. In two other experiments, the numbers of colonies formed by D. vulgaris L2 after 24 h of exposure to air were 60- and 100-fold lower than those formed by the wild type.

FIG. 7.

Survival of D. vulgaris wild type (□) and L2 (◊) exposed to air. Anaerobic cultures of the wild type and L2 in medium C (5 ml; 1.2 × 10⁹ CFU/ml) were diluted 100-fold in aerobic medium C at zero time. Samples were returned to anaerobic conditions after 4 to 24 h (aeration time) and plated immediately at various dilutions. Surviving cells were counted after 1 week of anaerobic incubation of the plates.

DISCUSSION

The rbo gene of D. vulgaris Hildenborough was discovered by Brumlik and Voordouw (2) through research aimed at elucidating the physiological function of rubredoxin, a 6-kDa redox protein with a single iron atom coordinated to four cysteinyl residues (FeS₄ center; E₀, −50 to 0 mV). Rubredoxin resides in the cytoplasm of Desulfovibrio spp., where the resident redox potential is generally much lower, causing it to be present in the reduced form. Analysis of the rub gene indicated that it forms an operon with a gene for a 14-kDa redox protein (2) which was named rubredoxin oxidoreductase because it was likely to function in oxidation-reduction reactions with rubredoxin as a redox partner. Since then, the rbo gene has been found to be closely associated with rub in other SRB, e.g., in D. vulgaris Miyazaki F (11) and in the more distantly related Desulfoarculus baarsii (19).

The sequence of Rbo indicated that it was a redox protein because its N terminus was highly homologous to desulforedoxin (1, 2), a redox protein of only 36 amino acids from Desulfovibrio gigas, which, like rubredoxin, has a single FeS₄ center. Chemical and spectroscopic analysis of purified Rbo indicated the presence of a second bound iron atom, in addition to an FeS₄ site similar to that present in desulforedoxin. The additional iron atom represented a high spin site that remained in the ferrous state even under aerobic conditions and appeared to be coordinated primarily with oxygen and nitrogen ligands (18). Indeed, Rbo has only a single conserved cysteine residue (C-117 [Fig. 1]) outside those in the desulforedoxin domain (C-10, C-13, C-29, and C-30) [Fig. 1]). Moura et al. (18) indicated that these physical properties are similar to those of rubrerythrin (Rbr), which contains a rubredoxin-like FeS₄ site and one nonsulfur, oxobridged di-iron site. Rbr, of which the three-dimensional structure is known, forms an operon with genes for a Fur-like and a rubredoxin-like protein. Despite this extensive knowledge, Lumppio et al. recently described Rbr as a non-heme iron protein of unknown function (16).

The dcrA gene, present immediately upstream from the rbo-rub operon (4), was shown to encode a chemoreceptor protein that functions as a sensor of the oxygen concentration or redox potential of the environment (7). D. vulgaris F100, a dcrA deletion mutant, appeared to be more resistant to oxygen inactivation than the wild type (5). The findings by Pianzzola et al. (19) that rbo complements sodAB deficiency in E. coli provided a possible explanation for this puzzling phenotype. Northern blotting studies indicated that deletion of dcrA increased expression of the rbo-rub operon (5). At the protein level, this effect can be seen in Fig. 5 (compare lanes 1 and 3 or 2 and 4); D. vulgaris F100 appears to have a ca. twofold-increased content of Rbo over the wild type. Although this implicated Rbo in repair or prevention of oxygen damage in D. vulgaris, the question of whether this is its only function in Desulfovibrio spp. and other anaerobic bacteria remained. Our present results indicate that deletion of the rbo gene does not affect growth under anaerobic conditions (Fig. 6A) but makes D. vulgaris clearly more sensitive to oxygen inactivation (Fig. 6B; Fig. 7). It appears, therefore, that the main physiological function of Rbo is that of an oxygen defense protein in Desulfovibrio spp. and possible also in other anaerobic bacteria and that the physiological function of rubredoxin is to assist in the electron transport required for this defense function. Assuming that the redox potential of the D. vulgaris cytoplasm rises to higher values under the aerobic conditions in which this defense system operates, the enigma of the high redox potential of rubredoxin is finally explained.

The mechanism by which Rbo protects an E. coli sodAB mutant from superoxide was recently studied in some detail (15). Rbo does not have significant SOD activity but functions in E. coli by serving as a preferred target for superoxide or derived radicals and possibly also by contributing iron-sulfur cluster-repair activity. Interestingly, the rbr gene encoding Rbr of Clostridium perfringens was similarly found to be capable of complementing sodAB deficiency in E. coli. Rbr was therefore also proposed to function as a scavenger of oxygen radicals (14). The recent completion of several genomic sequencing projects has indicated that Rbo and Rbr may be widespread in anaerobic bacteria. For instance, the sulfate-reducing archaeon Archaeoglobus fulgidus has a single Rbo homolog and four Rbr homologs (12). D. vulgaris Hildenborough has at least one other Rbr homolog, nigerythrin, for which the gene was recently cloned (16). Whether all of these novel redox proteins function primarily in oxygen defense, as does Rbo in D. vulgaris, or in anaerobic metabolism can be determined by further gene deletion studies, as presented here.