A NapC/NirT-type cytochrome c (NrfH) is the mediator between the quinone pool and the cytochrome c nitrite reductase of Wolinella succinogenes

Authors


Abstract

Wolinella succinogenes can grow by anaerobic respiration with nitrate or nitrite using formate as electron donor. Two forms of nitrite reductase were isolated from the membrane fraction of W. succinogenes. One form consisted of a 58 kDa polypeptide (NrfA) that was identical to the periplasmic nitrite reductase. The other form consisted of NrfA and a 22 kDa polypeptide (NrfH). Both forms catalysed nitrite reduction by reduced benzyl viologen, but only the dimeric form catalysed nitrite reduction by dimethylnaphthoquinol. Liposomes containing heterodimeric nitrite reductase, formate dehydrogenase and menaquinone catalysed the electron transport from formate to nitrite; this was coupled to the generation of an electrochemical proton potential (positive outside) across the liposomal membrane. It is concluded that the electron transfer from menaquinol to the catalytic subunit (NrfA) of W. succinogenes nitrite reductase is mediated by NrfH. The structural genes nrfA and nrfH were identified in an apparent operon (nrfHAIJ) with two additional genes. The gene nrfA encodes the precursor of NrfA carrying an N-terminal signal peptide (22 residues). NrfA (485 residues) is predicted to be a hydrophilic protein that is similar to the NrfA proteins of Sulfurospirillum deleyianum and of Escherichia coli. NrfH (177 residues) is predicted to be a membrane-bound tetrahaem cytochrome c belonging to the NapC/NirT family. The products of nrfI and nrfJ resemble proteins involved in cytochrome c biogenesis. The C-terminal third of NrfI (902 amino acid residues) is similar to CcsA proteins from Gram-positive bacteria, cyanobacteria and chloroplasts. The residual N-terminal part of NrfI resembles Ccs1 proteins. The deduced NrfJ protein resembles the thioredoxin-like proteins (ResA) of Helicobacter pylori and of Bacillus subtilis, but lacks the common motif CxxC of ResA. The properties of three deletion mutants of W. succinogenesnrfJ,ΔnrfIJ and ΔnrfAIJ) were studied. Mutants ΔnrfAIJ and ΔnrfIJ did not grow with nitrite as terminal electron acceptor or with nitrate in the absence of NH4+ and lacked nitrite reductase activity, whereas mutant ΔnrfJ showed wild-type properties. The NrfA protein formed by mutant ΔnrfIJ seemed to lack part of the haem C, suggesting that NrfI is involved in NrfA maturation.

Introduction

Many proteobacteria, including Wolinella succinogenes and Escherichia coli, grow by anaerobic respiration with nitrate, nitrite, fumarate or dimethylsulphoxide as terminal electron acceptors (Kröger et al., 1992; Weiner et al., 1992; Berks et al., 1995a; Cole, 1996). The enzymes catalysing the terminal step of electron transport are often integrated into the cytoplasmic membrane by hydrophobic subunits that accept electrons from the quinone pool. In other cases, the catalytic subunit reacting with the terminal acceptor is found in the bacterial periplasm as an apparently soluble protein. The isolated soluble reductases do not react with quinols or have not been tested for quinol reactivity, although there is evidence that the quinone pool is an obligatory member of certain electron transport pathways. In some cases, c-type cytochromes of the NapC/NirT family have been shown to be components of the electron transport pathways, and it has been speculated that these proteins may react directly with the quinone pool (Roldán et al., 1998). NapC/NirT-type proteins are characterized by a hydrophobic domain near the N-terminus and by four or five haem C groups bound to their residual hydrophilic part. Quinone reactivity has not yet been demonstrated with any NapC/NirT-type cytochrome c.

W. succinogenes can grow by anaerobic respiration with nitrate (reaction a) or nitrite (reaction b) as terminal electron acceptor (Bokranz et al., 1983). Inverted vesicles of W. succinogenes were shown to catalyse ADP phosphorylation driven by nitrite respiration with H2 (reaction c), and it is likely that W. succinogenes can also grow by anaerobic respiration with nitrite and H2.

inline image

Reactions b and c are catalysed by the bacterial membrane fraction. It was demonstrated previously that these electron transport activities were abolished upon extraction of the menaquinone (MK) from the membrane, and were restored upon incorporation of vitamin K1 into the extracted membrane (Bokranz et al., 1983; Schröder et al., 1985). These results suggested that the bacterial membrane carried a nitrite reductase catalysing the reduction of nitrite to NH4 +  by menaquinol (MKH2).

The membrane fraction was found to catalyse the reduction of nitrite to NH4+  by reduced viologens, and this activity was not abolished upon MK extraction (Schröder et al., 1985). Isolation of the corresponding enzyme led to a cytochrome c with the properties of the nitrite reductase isolated from the soluble cell fraction of W. succinogenes (Liu et al., 1983). The enzyme prepared from the membrane fraction could be incorporated into liposomes together with hydrogenase and vitamin K1 (Schröder et al., 1985). However, the resulting proteoliposomes did not catalyse reaction c, suggesting that the catalytic subunit of a presumed nitrite reductase complex had been isolated whereas the quinone-reactive subunit was lost. Using a different isolation procedure, Schumacher et al. (1994) prepared a heterodimeric nitrite reductase from the membrane fractions of W. succinogenes and the closely related Sulfurospirillum deleyianum. Both enzyme preparations contained a 22 kDa cytochrome c in addition to the catalytic subunit NrfA. The reactivity of these preparations with quinols was not tested.

Recently, the crystal structure of the catalytic subunit NrfA of S. deleyianum nitrite reductase was determined (Einsle et al., 1999). NrfA of S. deleyianum carries five haem C groups. One haem group is located near the catalytic centre and is axially ligated by the lysine residue of a CxxCK motif, whereas the other four haem C groups are conventionally bound by the residues of CxxCH motifs. The same arrangement of haem C groups is likely for NrfA of W. succinogenes, which is 75% identical to S. deleyianum NrfA. The catalytic subunit of the nitrite reductase from E. coli, which shares 46% and 48% identity with NrfA of S. deleyianum and of W. succinogenes respectively, was shown to have the same ligation pattern for its five haem C groups (Eaves et al., 1998).

Here, we describe a heterodimeric nitrite reductase isolated from W. succinogenes that is similar to the preparation obtained by Schumacher et al. (1994). The preparation was tested for quinol reactivity and for its capability to restore nitrite respiration in liposomes. Furthermore, the genes encoding the two polypeptides of the preparation, and the adjacent genes, were sequenced and analysed. To elucidate the function of these genes, three deletion mutants were constructed and characterized.

Results

The nrfHAIJ gene cluster of W.  succinogenes

A fragment of genomic DNA from W. succinogenes was found to carry the open reading frames nrfH, A, I and J on the same strand, and the partial open reading frame mreB on the opposite strand (Fig. 1). Putative promoters of transcription with suitable − 35 and − 10 sequences are located upstream of nrfH and mreB. The intergenic regions between nrfH and nrfA (20 bp) and between nrfA and nrfI (65 bp) do not contain obvious promoter or terminator elements. The stop codon of nrfI overlaps the nrfJ start codon. Downstream of nrfJ, a nucleotide sequence is present that suggests a hairpin structure typical of a terminator of transcription. Reverse transcription-PCR (RT-PCR) experiments (not shown) indicated co-transcription of nrfA with nrfI and of nrfI with nrfJ, suggesting a transcriptional entity for the four nrf genes. S. deleyianum also contains an nrfHAIJ gene cluster (EMBL accession number AJ133037) whose products are highly similar to those of W. succinogenes. NrfA and NrfH of the two bacteria share 75% and 68% identical residues respectively.

Figure 1.

Physical map of the nrf locus of W. succinogenes genomic DNA. The three plasmids served in the construction of mutants ΔnrfJ,ΔnrfIJ and ΔnrfAIJ via double homologous recombination with the genome of W. succinogenes. The segments designated by black boxes are identical to the indicated genomic regions and were synthesized by PCR. The SacI sites served in the characterization of the mutants. E, EcoRI; B, BamHI; S, SalI.

The genes nrfH and nrfA of W. succinogenes encode subunits of the cytochrome c nitrite reductase (see below), whereas the nrfI and nrfJ gene products resemble proteins involved in cytochrome c biogenesis. The predicted N-terminal region of MreB is similar to those of ‘rod shape-determining proteins’ of various bacteria. The nrfA gene encodes the precursor of the catalytic subunit of nitrite reductase (507 residues), which includes a signal peptide of 22 residues; this suggests translocation of the protein into the periplasm by the Sec system. Residues 23–41 predicted by nrfA are identical to the N-terminal 19 residues of the mature catalytic subunit (Simon et al., 1998a). NrfA is predicted to be a hydrophilic protein and to carry five haem C groups, four of which are bound by the cysteine residues of the common CxxCH motif. A fifth haem C group is probably linked to the CxxCK motif. The molecular mass of W. succinogenes NrfA, including the five haem C groups, is calculated to be 58 kDa, in fair agreement with the apparent molecular mass obtained from SDS–PAGE (Liu et al., 1983; Schröder et al., 1985; Schumacher et al., 1994).

The nrfH gene encodes the smaller polypeptide (177 residues) present in the heterodimeric preparation of nitrite reductase that was isolated from the membrane fraction of W. succinogenes (see below). For identification, the polypeptide was subjected to BrCN cleavage because the N-terminus was blocked to Edman degradation. The N-terminal sequence (KISEDG) of an 8.2 kDa fragment was identical to residues 104–109 following Met-103 in the sequence predicted by the nrfH gene. The molecular mass of NrfH including four haem C groups is calculated to be 22 kDa, in agreement with the value obtained from SDS–PAGE. NrfH is predicted to form a membrane-spanning helix at its N-terminus and to carry four haem C groups on the residual hydrophilic part. Its sequence is similar to those of c-type cytochromes of the NapC/NirT family (for an alignment, see Roldán et al., 1998).

The nrfI gene of W. succinogenes is predicted to encode a protein consisting of 902 amino acid residues that is similar to the CcsA protein of Helicobacter pylori (936 residues, 39% identity). The C-terminal third of NrfI resembles CcsA proteins from chloroplasts of plants and algae (300–350 residues, 24–29% identity). CcsA-like proteins were also deduced from the genome sequences of several Gram-positive bacteria such as Bacillus subtilis (ResC protein) or Mycobacterium ssp., and from Synechocystis sp. 6803 and Aquifex aeolicus. All of these proteins contain a characteristic tryptophan-rich motif (consensus sequence WGxxWxWD) and at least two conserved histidine residues; these were proposed to be essential for haem delivery across the membrane and/or for haem presentation in the bacterial periplasm (Goldman et al., 1998; for an alignment, see also Xie and Merchant, 1998). The N-terminal part of the W. succinogenes NrfI protein is moderately similar to the Ccs1 protein of Chlamydomonas reinhardtii and to predicted gene products of B. subtilis, Mycobacterium ssp. and Synechocystis sp. 6803. In the Gram-positive bacteria, the ccs1-like genes and the ccsA-like genes are arranged in putative operons. The nrfI gene of W. succinogenes thus seems to be the result of a fusion of ccs1 and ccsA, which are separate in other organisms.

The nrfJ gene product of W. succinogenes (217 residues) shares 20% identity and 43% similarity to the product of open reading frame HP0377, which is located directly upstream of ccsA in the H. pylori genome. NrfJ is also similar to the B. subtilis ResA thioredoxin-like protein. Genes encoding similar proteins are also found in the genomes of Mycobacterium ssp., Synechocystis sp. 6803 and A. aeolicus. Like other ResA proteins, W. succinogenes NrfJ is predicted to carry an N-terminal transmembrane helix. However, NrfJ does not contain any cysteine residues and therefore lacks the common CxxC motif of the ResA proteins.

Construction and properties of deletion mutants

Three deletion mutants (ΔnrfJ,ΔnrfIJ and ΔnrfAIJ) were constructed by double homologous recombination of the W. succinogenes genome with plasmid pδnrfJ, pδnrfIJ or pδnrfAIJ (Fig. 1). The three mutants retained 47% of nrfJ at its 3′ end. Mutant ΔnrfJ lacked the rest of nrfJ except 36 bp at its 5′ end. In mutant ΔnrfIJ, 71% of nrfI (3′ end) was deleted, whereas mutant ΔnrfAIJ lacked nrfI and nrfA except 15 bp at its 5′ end. Mutant ΔnrfJ had wild-type properties with respect to growth with nitrate or nitrite and to nitrite reductase activity (not shown). Mutant ΔnrfIJ did not grow with nitrite as terminal electron acceptor or with nitrate in the absence of NH4+ · The mutant lacked nitrite reductase activity but contained the NrfA protein (Fig. 2A and B). During growth with nitrate and NH4+ , nitrate (10 mM) was reduced to nitrite, which was not further reduced to NH4+ · Growth ceased after available nitrate had been converted to nitrite, and was restored by the addition of fumarate. Mutant ΔnrfAIJ had the properties of mutant ΔnrfIJ, but lacked NrfA (not shown). These results suggest that W. succinogenes forms only a single nitrite reductase that involves NrfA as the catalytic subunit. The enzyme appears to be essential for nitrite respiration (reaction b) as well as for NH4+  formation for biosynthesis.

Figure 2.

Western blot analysis (A and B) and haem stain analysis (C) of cell fractions of W. succinogenes nrf mutants. Cells were grown with formate and either fumarate (A) or nitrate and NH4+ (B and C), and the membrane fraction (MF) and the soluble fraction (SF) were prepared. Western blot analysis was performed with 80 µg protein applied to each lane, using an antiserum raised against NrfA of S. deleyianum (Schumacher et al., 1994). Haem staining of polyacrylamide gels was performed with 3,3-dimethoxybenzidine according to the method of Francis and Becker (1984) with 100 µg of protein applied to each lane.

An antiserum raised against S. deleyianum NrfA was used for detecting the NrfA protein after subjecting the cell fractions to SDS–PAGE. Upon growth with fumarate, NrfA of mutant ΔnrfIJ was found almost exclusively in the soluble cell fraction, whereas nearly all the NrfA protein of the parental strain and of mutant ΔnrfJ was located in the membrane fraction (Fig. 2A). The distribution of NrfA among the cell fractions of the wild-type strain and of mutant ΔnrfJ was consistent with that of nitrite reductase activity (not shown). When grown with nitrate, the majority of the NrfA of the wild-type strain and of the mutant ΔnrfJ was located in the membrane fraction (Fig. 2B). In contrast, most of the NrfA was in the soluble fraction of the ΔnrfIJ mutant. Haem staining of the SDS polyacrylamide gel demonstrated that the NrfA protein in mutant ΔnrfIJ still contained haem C (Fig. 2C). However, its haem C content seemed to be lower than that of NrfA in the wild-type strain and in the ΔnrfJ mutant, as judged from the haem stain intensity. A second prominent band of haem stain was developed by a polypeptide (22 kDa) that is exclusively located in the membrane fraction of the mutants and of the wild-type strain; this protein is probably NrfH. The amount of haem C associated with NrfH in the mutants appears to be the same as that in wild-type NrfH.

The total amount of haem C in mutant ΔnrfIJ was found to be lower than that in the parental strain and was higher than that in the ΔnrfAIJ mutant (Table 1). The results shown in Table 1 are consistent with the view that the deletion of nrfI specifically prevents the insertion of part of the haem C groups into NrfA. Assuming that wild-type amounts of NrfA protein are formed in mutant ΔnrfIJ (Fig. 2A and B) and that the total amount of the other c-type cytochromes was not altered by the deletion, up to two haem C groups are calculated to be missing from NrfA in mutant ΔnrfIJ. This number would approach one if less NrfA protein was formed in this mutant.

Table 1. Haem C contents in cell homogenates of W. succinogenes nrf mutants.
 Haem C (µmol g− 1 protein)
Growth medium containingΔnrfIJ ΔnrfAIJ Wild type
Nitrate + NH4+2.8 ± 0.32.0 ± 0.13.1 ± 0.2
Fumarate0.79 ± 0.030.50 ± 0.011.0 ± 0.1

Cellular location of NrfA

W. succinogenes cells grown with nitrate contain nitrite reductase activity in the membrane as well as in the soluble cell fraction (Liu et al., 1983; Schröder et al., 1985; Schumacher et al., 1994). The soluble nitrite reductase was proposed to be identical to the catalytic subunit (NrfA) of the membrane-bound enzyme. This proposal is confirmed here by the absence of NrfA and of nitrite reductase activity in the cell homogenate of the ΔnrfAIJ mutant (not shown). Furthermore, the N-terminal sequence of the soluble nitrite reductase (SNINEREK) was found to be identical to that of NrfA in the membrane-bound enzyme.

To determine the cellular localization of the soluble nitrite reductase, cells of W. succinogenes grown with nitrate were converted to spheroplasts, which were separated from the periplasmic cell fraction by centrifugation (Krafft et al., 1995). After osmotic lysis of the spheroplasts, the membrane fraction was separated by centrifugation from the cytoplasmic fraction. A Western blot experiment demonstrated that NrfA was present in the periplasmic and in the membrane fractions, but not in the cytoplasmic fraction (not shown). In agreement with this result, 74% and 25% of the cellular nitrite reductase activity was found in the membrane fraction and in the periplasmic fraction respectively. In a control experiment, the activity of glutamate dehydrogenase, a cytoplasmic enzyme, was localized almost exclusively in the cytoplasmic fraction. It is concluded that the soluble nitrite reductase is localized in the periplasm, as suggested by the signal peptide encoded by nrfA. NrfA of the membrane-bound enzyme is likely to be exposed to the periplasmic side of the membrane because it is encoded by the same gene.

Quinol-reactive nitrite reductase

The membrane fraction of W. succinogenes grown with nitrate was extracted with a mixture of Triton X-100 and dodecylmaltoside, and the extract was subjected to anion exchange chromatography. Approximately 30% of the nitrite reductase consisted of merely NrfA and did not bind to the column at pH 8.0. The residual enzyme was bound to the column and eluted with 70 mM NaCl in the equilibration buffer. The specific activity of this enzyme species was 25-fold higher than that of the membrane fraction. Two protein species were separated when the preparation was subjected to native gel electrophoresis (Fig. 3). The smaller species amounted to less than 10% of the total protein and consisted of NrfA. The larger species consisted of NrfA and NrfH. The heterodimeric species of nitrite reductase catalysed the reduction of nitrite by 2,3-dimethyl-1,4-naphthoquinol (DMNH2), whereas the monomeric species did not. This suggests that NrfH is a subunit of the nitrite reductase complex that mediates the electron transfer from the quinol to the catalytic subunit.

Figure 3.

Two-dimensional gel electrophoresis of the dimeric nitrite reductase isolated from the membrane fraction of W. succinogenes. The sample (80 µg) was subjected to polyacrylamide gradient (4–8%) gel electrophoresis in the absence of SDS in the first dimension. One gel strip was stained with Coomassie blue R250 and the second strip was applied to a 12.5% SDS polyacrylamide gel in the second dimension.

The heterodimeric nitrite reductase was incorporated into liposomes together with formate dehydrogenase isolated from W. succinogenes and a quinone. The resulting proteoliposomes catalysed nitrite reduction by formate (Table 2). The specific electron transport activity (reaction b) of the proteoliposomes prepared with added MK (10 µmol g− 1 phospholipid) isolated from W. succinogenes was 45 times higher than that of nitrite reduction by DMNH2. This suggested that the nitrite reductase complex reacted considerably faster with the lipophilic MKH2 than with the hydrophilic DMNH2. The turnover number of nitrite reductase in the electron transport (780 electrons s− 1) amounted to about 30% of that of the enzyme in W. succinogenes growing with nitrate (Bokranz et al., 1983). When MK was replaced by the same amount of vitamin K1, the specific electron transport activity of the liposomes was 36% of those with added MK. The low activity observed without added quinone was probably owing to the MK that was associated with the enzymes incorporated into the liposomes (about 1 µmol g− 1 phospholipid).

Table 2. Electron transport activity of proteoliposomes and TPP+ uptake in the steady state of electron transport.

Quinone added

Protonophore added
Activity of electron transport
(U mg− 1 nitrite reductase)

TPP+ uptake (µmol g− 1 phospholipid)

External TPP+ (µM)
  1. Proteoliposomes were suspended (4 g  l− 1 phospholipid) in an anoxic buffer containing 50 mM Tris chloride (pH 7.8) at 37°C. The activity of electron transport from formate to nitrite was determined by measuring nitrite at various time intervals after 20 mM sodium formate and 5 mM NaNO2 had been added. The unit of electron transport activity (U) is equivalent to the consumption of 1 µmol min− 1 formate. TPP+ uptake was measured using a TPP+ electrode (Butsch and Bachofen, 1984; Mell et al., 1986). The TPP+ uptake and the external TPP+ concentration refer to the moment of maximal TPP+ uptake by the liposomes. The protonophore 4,5,6,7-tetrachloro-2-trifluoromethylbenzimidazole (TTFB; 50 µmol g− 1 phospholipid) was applied where indicated.

MK2940.900.50
MK+ 280< 0.01
Vitamin K11050.580.48
12

Using a tetraphenylphosphonium (TPP + ) electrode, the proteoliposomes were observed to take up TPP +  when the electron transport was started by the addition of formate and nitrite (Table 2). TPP +  was liberated into the medium after one of the substrates was consumed. This cycle of TPP +  uptake and release could be repeated upon addition of the missing substrate. The duration of the cycle was consistent with the electron transport activity. The cycle was abolished by a protonophore that did not affect the electron transport activity. These observations indicate that the electron transport from formate to nitrite is coupled to the generation of an electrochemical proton potential (Δp, positive outside) across the liposomal membrane. The Δp can be estimated from the amount of TPP +  taken up by the liposomes and the corresponding external concentration of TPP +  (Table 2), assuming that the internal volume of the proteoliposomes is 3 ml g− 1 phospholipid and that the free internal TPP +  amounts to 15% of the total TPP +  taken up by the liposomes (Mell et al., 1986). With these assumptions, the electrical proportion of Δp is calculated as 120 mV for the experiment with the proteoliposomes containing added MK (Table 2). This value should be close to the Δp because the ΔpH across the membrane was previously found to be very small (Mell et al., 1986; Geisler et al., 1994). In a similar experiment with cells of W. succinogenes grown with formate and nitrate, the Δp (positive outside) generated by reaction b was measured as 160 mV (not shown).

Discussion

Nitrite respiration in W.  succinogenes

The results presented strongly suggest that the nitrite reductase complex of W. succinogenes consists of NrfA and NrfH and accepts the electrons for nitrite reduction directly from the MK pool within the cytoplasmic membrane. The enzyme complex catalyses nitrite reduction by DMNH2, a water-soluble analogue of MKH2, and restores coupled electron transport from formate to nitrite when incorporated into liposomes together with formate dehydrogenase and MK. NrfA alone does not react with DMNH2 and does not restore the electron transport activity in proteoliposomes (Schröder et al., 1985). A hypothetical coupling mechanism of nitrite respiration is depicted in Fig. 4. Formate oxidation and nitrite reduction occur on the periplasmic side of the membrane. The sum of the two reactions is consistent with reaction b, except that the substrates and products are drawn in their neutral forms in Fig. 4 for simplicity. The Δp is envisaged to be generated by the two reactions occurring at different sites on formate dehydrogenase. Protons are liberated by formate oxidation on the periplasmic side of the membrane, while the protons consumed by MK reduction are taken up from the cytoplasmic side. This view is consistent with the orientation of the substrate site of formate dehydrogenase towards the periplasm and with the finding that MK reduction by formate generates a Δp across the membrane of proteoliposomes containing formate dehydrogenase alone (Kröger et al., 1980; Geisler et al., 1994). Furthermore, the cytochrome b of formate dehydrogenase was shown to carry the MK-reactive site (Unden and Kröger, 1983). The oxidation of MKH2 by nitrite is thought to be an electroneutral process. The two H +  produced by MKH2 oxidation at the NrfH subunit of nitrite reductase are thought to be liberated on the periplasmic side and, simultaneously, two H +  are consumed on the same side by the conversion of nitrite to ammonia. According to this mechanism, protons are translocated across the membrane by the redox reactions of MK during electron transport from formate to nitrite. The site of MKH2 oxidation at NrfH and the mechanism of proton liberation from MKH2 to the periplasmic side of the membrane are not known. NrfH is predicted to form only one transmembrane helix that is unlikely to carry a redox group. Therefore, the site of MKH2 oxidation is possibly formed by both the hydrophobic and the hydrophilic parts of NrfH and may be localized at the periplasmic border of the membrane.

Figure 4.

Hypothetical mechanism of Δp generation across the membrane of W. succinogenes by nitrite respiration with formate. The electron transport chain consists of formate dehydrogenase (FdhABC), menaquinone (MK) and nitrite reductase (NrfHA). Mo, molybdenum linked to molybdopterin guanine dinucleotide (Jankielewicz et al., 1994); Fe/S, iron/sulphur centres; Cyt. b, cytochrome b; Cyt. c, cytochrome c.

The coupling mechanism proposed for nitrite respiration in E. coli is similar to that of Fig. 4 (Berks et al., 1995a). The nrf operon of E. coli encodes three electron transfer proteins (NrfB, C and D) in addition to NrfA (Hussain et al., 1994). NrfB is predicted to be a pentahaem cytochrome c and was proposed to serve as the electron donor to NrfA. NrfC is predicted to be a membrane-bound iron/sulphur protein, and the hydrophobic NrfD was proposed to react with the quinol pool. It was suggested that the three proteins form a membrane-bound ‘quinol oxidase complex’ with a function corresponding to that of W. succinogenes NrfH. The sequences of the three proteins do not appear to be related to NrfH of W. succinogenes.

The function of c-type cytochromes of the NapC/NirT family

NrfH is a cytochrome c of the NapC/NirT family. Genes encoding such c-type cytochromes are usually located in the vicinity of genes encoding the catalytic subunits of periplasmic reductases on the genomes of various bacteria. Therefore, the corresponding NapC/NirT-like cytochromes are thought to be components of the respective pathways of anaerobic respiration. This refers to the nitrate respiration of Paracoccus denitrificans (Berks et al., 1995b), to anaerobic respiration with dimethylsulphoxide in Rhodobacter sphaeroides (Ujiye et al., 1996) and in Rhodobacter capsulatus (Shaw et al., 1999), and to trimethylamine oxide respiration in E. coli (Méjean et al., 1994). In Pseudomonas stutzeri, NirT was demonstrated to be a component of the electron transport to nitrite (Jüngst et al., 1991). The inactivation of nirT blocked the electron transport to nitrite whereas the activity of cytochrome cd1 nitrite reductase was not affected. Similar experiments performed on the electron transport to nitrate and to dimethylsulphoxide with Rhodobacter sphaeroides led to the same results (Reyes et al., 1996; Mouncey et al., 1997). The cymA gene product of Shewanella putrefaciens was shown to be involved in the electron transport pathways with nitrate, fumarate, iron (III) or manganese (IV) as acceptors (Myers and Myers, 1997). These authors also demonstrated that MK is involved in the respective pathways (Myers and Myers, 1993). However, it remained unclear whether or not CymA reacted directly with MK and/or the respective reductase.

Maturation of the NrfA protein

The results suggest that the NrfI protein of W. succinogenes plays a role in cytochrome c biogenesis and, more specifically, is required for the maturation of the NrfA protein. This proposed function of NrfI resembles that of the nrfEFG gene products of E. coli, although NrfI has no overall similarity to any of the nrfEFG gene products and belongs to a different system of cytochrome c biogenesis (see below). The nrfEFG genes of E. coli are part of the nrfABCDEFG operon (Hussain et al., 1994). NrfE and NrfG (but not NrfF) are essential for formate-dependent nitrite reduction (Grove et al., 1996). In an E. coli mutant unable to synthesize NrfE, NrfF and NrfG, periplasmic NrfA was inactive and was shown to lack the active site haem group that is covalently bound to the CxxCK motif (Eaves et al., 1998). These authors proposed that the nrfE, F and G gene products are part of a haem lyase that specifically recognizes the CxxCK motif in apo-NrfA. Possibly, the function of the nrfI gene product of W. succinogenes is similar to the functions of E. coli nrfEFG because the haem C content of NrfA seems to be lowered in mutant ΔnrfIJ. However, it remains to be determined how many and which haem C groups are missing. The majority of the NrfA protein in the ΔnrfIJ mutant was located in the soluble cell fraction. The reason for this is not known. Possibly, the interaction of NrfA with the membrane-bound NrfH is dependent on the attachment of all haem C groups to NrfA.

Cytochrome c biogenesis in W.  succinogenes

In bacteria, the various steps of cytochrome c biosynthesis are carried out by either of two enzymic systems that differ in their components (for recent reviews, see Thöny-Meyer, 1997; Kranz et al., 1998; Page et al., 1998). Genes belonging to one system do not occur in a genome if genes of the other system are present, and the amino acid sequences derived from the genes of one system are not related to those of the other. According to Kranz et al. (1998), system II is present in Gram-positive bacteria, in cyanobacteria, in chloroplasts and in H. pylori, which is a close relative of W. succinogenes, whereas E. coli has system I. At least four gene products are involved in system II (CcsA, Ccs1, ResA and CcdA), whose exact function has not yet been elucidated. The protein predicted by nrfI of W. succinogenes is similar to various Ccs1/CcsA proteins, whereas the gene product of nrfJ resembles ResA proteins; this indicates that W. succinogenes belongs to the system II organisms. Presumably, CcsA proteins are involved in haem delivery across the membrane and/or in haem attachment to the apo-cytochrome (see Kranz et al., 1998 and references therein). Inactivation of either ccsA or ccs1 in Chlamydomonas reinhardtii resulted in a pleiotropic c-type cytochrome deficiency (Xie and Merchant, 1996, 1998; Inoue et al., 1997). In Synechocystis sp. 6803 and in B. subtilis, the ccsA homologue appeared to be essential (Sun et al., 1996; Hübschmann et al., 1997). It was suggested that the Ccs1 and CcsA proteins function in a membrane-bound complex in Chlamydomonas reinhardtii (Xie et al., 1998). The function of this complex may be similar to that of H. pylori CcsA and to those of NrfI in W. succinogenes and in S. deleyianum. The CcsA from Mycobacterium leprae was shown to form six membrane-spanning domains (Goldman et al., 1998), and CcsA of chloroplasts was predicted to contain eight transmembrane segments (Xie and Merchant, 1998). Ccs1 from Chlamydomonas reinhardtii was proposed to contain three transmembrane domains. The larger NrfI of W. succinogenes contains up to 14 mainly hydrophobic regions that may form transmembrane traversions.

No function in cytochrome c biogenesis could be assigned to the NrfJ protein. It is likely that nrfJ is a pseudogene because the corresponding protein lacks the cysteine residues characteristic of the ResA proteins belonging to the thioredoxin family of thiol-disulphide oxidoreductases. Alternatively, the NrfJ protein may be substituted in the mutants by a second resA homologue.

Because of the specialized function of the NrfI protein, a complete set of system II genes is expected to occur in the genome of W. succinogenes. These genes might imply: (i) an nrfI homologue (ccsA) encoding a protein of general function in haem transport and/or as a haem lyase; (ii) a resA gene encoding a thioredoxin containing a CxxC motif; and (iii) a ccdA gene similar to those found in the genomes of H. pylori, B. subtilis and other Gram-positive bacteria (Schiött et al., 1997; Tomb et al., 1997). So far, none of these genes was identified in W. succinogenes. Southern blot analysis of the genome, using varying stringencies, with probes derived from nrfI or nrfJ did not reveal the occurrence of genes homologous to either ccsA or resA.

Experimental procedures

Growth of W. succinogenes

W. succinogenes was grown with formate and nitrate or nitrite as has been described previously (Lorenzen et al., 1993). For growth with formate, nitrate and NH4+ , (NH4)2SO4 (5 mM) was present in the medium instead of K2SO4. The medium for growth with formate and fumarate is described by Kröger et al. (1994).

Cell fractionation

W. succinogenes cells harvested in the exponential growth phase were suspended (10 g l− 1 cell protein) in an anoxic buffer (pH 8.0) containing 50 mM tris-HCl and 1 mM dithiothreitol. The suspension was passed through a French press at 130 MPa and at 10 ml min− 1 flow rate. The resulting cell homogenate was centrifuged for 45 min at 100 000 g to yield the membrane fraction (sediment) and the soluble fraction.

Nitrite reductase activity

The activity of nitrite reductase was measured at 37°C using photometric recording of reduced benzyl viologen oxidation by nitrite as described by Bokranz et al. (1983). One unit of enzyme activity (U) was equivalent to the oxidation of 2 µmol min− 1 reduced benzyl viologen. Nitrite reduction by DMNH2 was measured photometrically by recording DMN formation at 37°C (Unden and Kröger, 1986). The anoxic buffer (pH 7.5) contained 50 mM phosphate and 0.4 mM DMN, which was reduced by the same amount of KBH4. After the addition of the enzyme, the reaction was started by adding 10 mM NaNO2. The unit of activity (U) is equivalent to the oxidation of 1 µmol min− 1 DMNH2.

Purification of membrane-bound nitrite reductase

Cells of W. succinogenes grown with formate and nitrate were harvested and stored at − 70°C. The following steps were performed with buffers at 0°C, which were flushed with N2. The cells were suspended (6 g cell protein) in 0.25 l of a buffer (pH 7.2) containing 50 mM potassium phosphate, 1 mM dithiothreitol and 10% (v/v) ethylene glycol. The suspension was passed through a French press at 130 MPa, and the resulting cell homogenate was centrifuged for 40 min at 100 000 g. The sediment (3 g of protein) was stirred for 2 h in the same buffer (0.2 l) containing (per g of protein) 1 g of Triton X-100 and 0.25 g of dodecylmaltoside. After centrifugation, the solution was applied to an anion exchange column (0.55 l; DEAE Sepharose CL-6B) equilibrated with the same buffer containing 0.05% (v/v) Triton X-100 that was also used for elution. The combined fractions containing the enzyme activity were diluted eightfold with a buffer (pH 8.0) containing 50 mM tris-HCl and 1 mM dithiothreitol and were then concentrated by pressure dialysis. The solution was applied to an anion exchange column (0.3 l; DEAE Sepharose CL-6B) equilibrated with the same buffer containing 0.05% (v/v) Triton X-100. Part of the enzyme activity eluted with the equilibration buffer, whereas elution of the residual activity required 70 mM NaCl in the equilibration buffer. The preparation was concentrated by pressure dialysis and stored in liquid N2.

Proteoliposomes

Sonic liposomes were prepared from phosphatidylcholine (Sigma P5394) and either MK extracted from W. succinogenes or vitamin K1 (10 µmol g− 1 phospholipid) in anoxic 50 mM tris-HCl buffer (pH 8.0) at 0°C. To the suspension of liposomes (5 g l− 1 phospholipid), heterodimeric nitrite reductase and formate dehydrogenase (Unden and Kröger, 1986) from W. succinogenes (0.2 µmol g− 1 phospholipid each) were added, and the mixture was stirred for 3 h at 22°C with 0.33 g ml− 1 Biobeads SM2 (Bio-Rad) that absorbed the Triton X-100 associated with the enzyme preparations.

Determination of protein, nitrite and haem C

Protein was measured using the Biuret method with KCN (Bode et al., 1968) or the bicinchoninic acid method (Smith et al., 1985). Nitrite was determined as described previously (Rider and Mellon, 1946). The haem C content was calculated from the absorbance difference between the fully reduced and fully oxidized sample at 550 nm minus that at 540 nm using the molar extinction coefficient of 19.8 mM− 1 cm− 1 (Kröger and Innerhofer, 1976).

PAGE, Western blotting and enzyme-linked immunosorbent assay (ELISA)

SDS–PAGE was performed according to Lämmli (1970). For PAGE under non-denaturing conditions, SDS was omitted from buffers. The samples were mixed with glycerol (15% v/v) and applied to a polyacrylamide gradient (4–8%) gel containing dodecylmaltoside (0.1% w/v). The electrophoresis buffer contained 37 mM Tris, 29 mM glycine and dodecylmaltoside (0.05% w/v). The voltage did not exceed 200 V during electrophoresis at 4°C overnight.

For Western blot analysis, protein was transferred from the SDS gel to nitrocellulose by electroblotting in a discontinuous buffer system (Kyhse-Andersen, 1984). NrfA was detected by indirect ELISA using an antiserum (diluted 1:800) raised against the soluble nitrite reductase from S. deleyianum (Schumacher et al., 1994) and goat anti-rabbit IgG coupled to peroxidase (Sigma).

Cloning and sequencing of W. succinogenes nrfHAIJ

A fragment of nrfA was amplified by PCR using genomic DNA of W. succinogenes and degenerate primer pairs deduced from conserved regions of the NrfA proteins from S. dedeyianum and from E. coli. A nrfI fragment was obtained using PCR with primers derived from conserved regions of various CcsA sequences (Xie and Merchant, 1998). The DNA region between the nrfA and the nrfI fragments was amplified using PCR. Flanking regions were amplified using inverse PCR employing genomic DNA fragments as template that were obtained by restriction with either SacI or HindIII and religation with T4 DNA Ligase. PCR products were cloned using the pCR-Script Amp SK( + ) Cloning Kit (Stratagene) or the TOPO TA Cloning method (Invitrogen). Plasmid DNA was purified from E. coli with Qiagen tips and was sequenced according to Sanger et al. (1977), using universal primers and BigDye terminator cycle sequencing (Applied Biosystems).

Genetic techniques

Standard genetic procedures were used as have been described previously (Sambrook et al., 1989). DNA was isolated from W. succinogenes according to Kaiser and Murray (1979) or with the DNeasy Tissue Kit from Qiagen. PCR was carried out using Goldstar DNA Polymerase (Eurogentec) or the Expand High Fidelity PCR System (Roche) with standard amplification protocols on a Hybaid OmniGene Thermocycler (MWG Biotech). Southern blotting to nylon membranes was performed as has been described previously (Lenger et al., 1997). DNA probes were generated with the PCR DIG Probe Synthesis Kit (Roche), and hybrids were visualized using the DIG/Luminescent Detection Kit (Roche).

Construction of deletion mutants

W. succinogenesΔnrfAIJ, W. succinogenesΔnrfIJ and W. succinogenesΔnrfJ were constructed as outlined in Fig. 1. For the construction of the three corresponding deletion plasmids, DNA fragments designated by black boxes in Fig. 1 were synthesized using PCR with primers that carried suitable restriction sites for cloning at their 5′ ends. The downstream fragment was inserted into pBR322 using BamHI and SalI restriction. Subsequently, each of the three upstream fragments was inserted using EcoRI and BamHI restriction. The identity of the cloned PCR fragments was confirmed by sequencing. Finally, the kanamycin resistance gene cartridge (kan) from pUC4K was inserted in each plasmid using BamHI restriction. The orientation of kan (Fig. 1) was confirmed by restriction analysis.

Wild-type cells of W. succinogenes were transformed with each of the plasmids as described previously (Simon et al., 1998b). Transformants were selected on agar plates containing the nitrate medium with NH4+ , 2.6% (w/v) brain–heart infusion agar (Gibco BRL) and kanamycin (25 mg l− 1). In each case, the genome of several transformants was checked for the presence of the kan gene by means of Southern blot analysis using SacI restriction. In the mutant ΔnrfAIJ, only one SacI fragment (3.5 kbp) hybridized to the kan probe. The size of the fragment demonstrated that kan was integrated via double homologous recombination of plasmid pδnrfAIJ with the genome (Fig. 1). In the ΔnrfIJ mutant, only one SacI fragment of 2.5 kbp hybridized to the kan probe. Its size was consistent with the deletion of a genomic fragment containing four SacI sites (Fig. 1). In the ΔnrfJ mutant, only the expected SacI fragment (2.8 kbp) hybridized to the kan probe confirming the identity of the mutant. As all three plasmids (6.0 kbp each) did not contain any SacI sites, the results of the Southern blot analysis excluded plasmid integration or replication.

Computer analysis

Database searches made use of the program blast (Altschul et al., 1997). Search for membrane-spanning helices was performed using the tmpred program (Hofmann and Stoffel, 1993). Multiple sequences were aligned using the program clustalw (Thompson et al., 1994).

Nucleotide sequence accession number

The nucleotide sequence reported here has been deposited in the EMBL, GenBank and DDBJ data bases under the accession no. AJ245540.

Acknowledgements

The authors are grateful to M. Sänger for expert technical assistance. The work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB 472 to A.K. and SPP 1070 to P.M.H.K.) and from the Fonds der Chemischen Industrie to A.K. and P.M.H.K.

Ancillary