Growth of Campylobacter jejuni on nitrate and nitrite: electron transport to NapA and NrfA via NrfH and distinct roles for NrfA and the globin Cgb in protection against nitrosative stress

Authors


E-mail d.kelly@sheffield.ac.uk; Tel. (+44) 114 222 4414; Fax (+44) 114 222 2754.

Summary

Pathways of electron transport to periplasmic nitrate (NapA) and nitrite (NrfA) reductases have been investigated in Campylobacter jejuni, a microaerophilic food-borne pathogen. The nap operon is unusual in lacking napC (encoding a tetra-haem c-type cytochrome) and napF, but contains a novel gene of unknown function, napL. The iron-sulphur protein NapG has a major role in electron transfer to the NapAB complex, but we show that slow nitrate-dependent growth of a napG mutant can be sustained by electron transfer from NrfH, the electron donor to the nitrite reductase NrfA. A napL mutant possessed ∼50% lower NapA activity than the wild type but showed normal growth with nitrate as the electron acceptor. NrfA was constitutive and was shown to play a role in protection against nitrosative stress in addition to the previously identified NO-inducible single domain globin, Cgb. However, nitrite also induced cgb expression in an NssR-dependent manner, suggesting that growth of C. jejuni with nitrite causes nitrosative stress. This was confirmed by lack of growth of cgb and nssR mutants, and slow growth of the nrfA mutant, in media containing nitrite. Thus, NrfA and Cgb together provide C. jejuni with constitutive and inducible components of a robust defence against nitrosative stress.

Introduction

The genus Campylobacter comprises Gram-negative, spiral-shaped, microaerophilic bacteria, which have assumed increasing prominence in recent years as the causative agents of major animal and human diseases. Campylobacter jejuni is the leading cause of acute gastroenteritis, acquired by ingesting contaminated food, milk or water. Poultry is a particularly common source of contamination (Freidman et al., 2000), as C. jejuni is a commensal of the gastrointestinal tract of many species of birds. Acute symptoms of C. jejuni infection in humans usually include abdominal cramping and diarrhoea, but more serious sequelae such as reactive arthritis and the Guillaine–Barré syndrome can also result.

Despite the importance of C. jejuni as a food-borne pathogen, there are many aspects of the physiology of this bacterium that remain poorly defined, including its bioenergetics and stress responses. The genome sequence of strain NCTC 11168 (Parkhill et al., 2000) and other more recently sequenced strains (Fouts et al., 2005) reveal the presence of surprisingly complex branched electron transport systems, including pathways to two terminal oxidases and a range of terminal reductases (Smith et al., 2000; Sellars et al., 2002). These should allow the bacterium to carry out respiration with fumarate, nitrate, nitrite, trimethylamine-N-oxide (TMAO) or dimethylsulphoxide (DMSO), and growth under severely oxygen-limited conditions with all of these alternative electron acceptors has been demonstrated (Sellars et al., 2002). However, C. jejuni is unable to grow strictly anaerobically, likely due to the presence of only a single oxygen-requiring (class I) type of ribonucleotide reductase, which is essential for DNA synthesis (Sellars et al., 2002).

In addition to a membrane bound fumarate reductase (Cj0408−0410) and a periplasmic bifunctional TMAO/DMSO reductase (Cj0264), C. jejuni NCTC 11168 is also equipped with both nitrate (Nap) and nitrite (Nrf) reductases located in the periplasm (Sellars et al., 2002; Pittman and Kelly, 2005). The Nap type of nitrate reductase is a two-subunit enzyme consisting of NapA, a ∼90 kDa catalytic subunit which binds a bis-molybdenum guanosine dinucleoside (MGD) cofactor and a [4Fe4S] cluster, and NapB, a ∼16 kDa di-haem c-type cytochrome (Butler et al., 2001; Fig. 1B). NapA is a substrate for the twin-arginine translocase (TAT) system and requires an additional protein, NapD as a ‘proof-reading’ chaperone in the export process (Palmer et al., 2005). Interestingly, the electron transport pathway to nitrate via NapAB is not immediately obvious in C. jejuni, as nitrate reduction is usually coupled to quinol oxidation by a membrane-anchored tetra-haem cytochrome, NapC. However, the napAGHBLD operon present in C. jejuni (see Fig. 1A), does not contain napC. Analysis of the NCTC 11168 genome shows that it does encode a napC homologue (Cj1358c), which is directly upstream of the putative nrfA nitrite reductase gene (Cj1357c), implying that it is part of the nitrite reductase system. This is the case in Wolinella succinogenes, a bacterium closely related to C. jejuni, which encodes a NapC-like subunit (NrfH) similar to the product of Cj1358c (Simon et al., 2000). This leaves the question of how the NapAB complex is coupled to menaquinol oxidation in C. jejuni. The NapG and NapH proteins probably function in this role, as these are iron-sulphur proteins which together act as a quinol dehydrogenase in Escherichia coli (Brondijk et al., 2004). However, NapG/H are specifically involved in electron transfer from ubiquinol to nitrate in the E. coli Nap system (Brondijk et al., 2002) and are not required for menaquinol-dependent nitrate reduction (Brondijk et al., 2002), indicating a fundamentally different quinol specificity. Finally, the epsilon group of proteobacteria (including W. succinogenes and C. jejuni) are unusual in containing an additional gene of unknown function, designated napL, within their nap operons (Simon et al., 2003). NapL is predicted to be a soluble periplasmic protein with no obvious cofactor binding motifs.

Figure 1.

A. Gene organization of nap and nrf operons in C. jejuni NCTC 11168 and strategy for mutagenesis. The kan and cat cassettes were inserted into unique restriction sites, except for napA where a deletion was made. Genes are shown to scale except kan and cat cassettes. Black arrows show the direction of transcription of the antibiotic resistance genes.
B. Model for the mechanism of electron transport to nitrate and nitrite in C. jejuni, based on evidence from this study and genomic predictions. A variety of primary dehydrogenases feed electrons from oxidisable substrates (‘X’) to the quinone pool. It is thought that NapGH form a quinol oxidase complex and pass electrons to the nitrate reductase structural proteins, NapAB. The NrfH homologue Cj1358 is the sole electron donor to NrfA, the penta-haem nitrite reductase, but can also donate electrons to NapAB. NrfA can reduce both nitrite and nitric oxide. As Nap and Nrf are periplasmic, reduction of the electron acceptor and quinol oxidation occur on the same side of the membrane. Thick solid black arrows show the major electron transport routes, thin black arrows are secondary pathways. MGD, bis-molybdenum guanosine dinucleoside cofactor, MK, menaquinone.

Cj1357c is predicted to encode a periplasmic pentahaem cytochrome c nitrite reductase (NrfA), which is the terminal enzyme in the six-electron dissimilatory reduction of nitrite to ammonia (Simon, 2002; Pittman and Kelly, 2005; Fig. 1B. High specific activities of nitrite reduction have previously been detected in intact cells of C. jejuni using methyl viologen as the electron donor (Sellars et al., 2002). Interestingly, cytochrome c nitrite reductases can also carry out the five electron reduction of the nitric oxide radical (NO) to NH4+ (Costa et al., 1990). Work in E. coli has indicated that this reaction may be physiologically relevant under microaerobic conditions, and it has been proposed that NrfA plays a significant role in nitric oxide detoxification in addition to flavohaemoglobin (Hmp) and flavorubredoxin (NorV), the other major E. coli NO detoxification systems (Poock et al., 2002).

Nitric oxide is an extremely important component of the defence of the host against invading pathogenic bacteria. Patients with Campylobacter infections have been shown to have higher rectal levels of NO and the assay of luminal NO has been proposed as a facile diagnostic tool (Enocksson et al., 2004). NO can also arise from other nitrogenous species, particularly nitrite on skin (Suschek et al., 2006) and in the oral cavity (RauschFan and Matejka, 2001). A nitrate-rich diet significantly elevates exhaled NO (Olin et al., 2001). Thus, NO is likely to be encountered by campylobacters in the alimentary canal and may influence the outcome of infections.

In C. jejuni, Hmp and NorV are absent but there is a single domain globin (Cgb), which has been shown to protect against nitrosative stress (Elvers et al., 2004). In keeping with this, Cgb is inducible by nitrosative stress via the regulatory protein NssR (Elvers et al., 2005). The role of other enzymes, including NrfA, in NO defence in C. jejuni is unknown. We have constructed and characterized a set of mutants in the key genes of the nap and nrf operons of C. jejuni, in order to clarify the electron transport pathways to nitrate and nitrite, determine the function of these genes, and to test the hypothesis that NrfA may play a role in NO detoxification in this bacterium.

Results

Construction and reverse transcription polymerase chain reaction (RT-PCR) analysis of mutants

In order to investigate the role of the genes in the nap and nrf operons in C. jejuni NCTC 11168, mutants were constructed by insertion of antibiotic resistance cassettes into the cloned genes, in the same transcriptional orientation, followed by electroporation into wild-type (WT) C. jejuni (see Fig. 1A). nrfH, nrfA, napA, napG and napL mutants were successfully constructed, but a napH mutant could not be obtained. The reasons for this are unknown, but it is unlikely that napH is an essential gene. For each mutant, at least two colonies from independent transformations were picked, verified by PCR and shown to have identical phenotypes in the key experiments described below, ruling out any possibility of second site mutations being responsible for the results obtained. The kan and cat cassettes used contain their own promoters but lack transcriptional terminators. The cat insertion into nrfH was shown to be non-polar on nrfA by assaying BV-dependent nitrite reductase activity, which was the same as the WT parent strain (both ∼8 μmol min−1 mg cell protein−1). To verify that the mutations in the nap operon did not cause any polar effects, quantitative RT-PCR was used to analyse the napB and napD transcript levels, using gyrA as an internal control. In the napA and napG mutants, the relative transcript levels of the downstream napB were 0.49 ± 0.88 and 0.97 ± 1.01 fold-changed, respectively, compared with WT, while in the napL mutant the relative transcript level of the downstream napD was 2.01 ± 1.4 fold-changed compared with WT. These changes are not significantly different to the WT and indicate that no polarity problems are caused by the mutations.

napA is essential for nitrate reduction but napG and napL are not: evidence for electron transfer from NrfH to nitrate

Inactivation of napA resulted in the complete absence of NapA protein in the periplasm, as judged by immunoblotting with cross-reactive anti-E. coli NapA antibodies (data not shown), and abolition of nitrate reductase activity in intact cells as assayed with benzyl viologen as electron donor (Table 1). BV-linked nitrite reductase activity was the same as the WT. These data confirm that NapA is the sole nitrate reductase in C. jejuni. The growth characteristics of the napA mutant under standard microaerobic conditions in BHI-FCS media (with or without added nitrate) were identical to the WT parent (data not shown), whereas a clear difference in nitrate-stimulated growth under oxygen-limited conditions was apparent (Fig. 2A). Nitrite production from nitrate during oxygen-limited growth of the WT and napA mutant was monitored in culture supernatants (Fig. 2B). Nitrite accumulated continuously in WT cultures to a concentration of ∼20 mM at 6 h, which was stoichiometric with the starting nitrate concentration. Thereafter, the culture nitrite concentration decreased. No nitrite at all was detected in cultures of the napA mutant over the same time period (Fig. 2B). The slight growth of the napA mutant in Fig. 2A is due to some residual oxygen-dependent respiration, as control cultures without added nitrate also showed a slight increase in optical density (OD) (data not shown). The biphasic pattern of nitrite accumulation then consumption in WT cells confirms data in our previous study (Sellars et al., 2002), which suggested a degree of regulation of the activity of nitrite reductase by nitrate. BV-linked nitrite reductase specific activity in cells grown microaerobically for 14 h in the absence of nitrate was 9.4 ± 1.0 μmol min−1 mg cell protein−1 but this was found to be reduced to 3.8 ± 0.4 μmol min−1 mg cell protein−1 in cells grown for 14 h in the presence of 20 mM sodium nitrate (all assays were performed on three replicate cultures). Thus, growth with high concentrations of nitrate appears to partly repress the expression of nrfA.

Table 1.  Benzyl viologen-linked assays of NapA and NrfA activity in wild type and mutant strains.
Strainμmol min−1 mg protein−1
Microaerobic growthOxygen-limited growth
Rate NO3 reductionRate NO2 reductionRate NO3 reductionRate NO2 reduction
  1. Cells were grown in BHI-FCS plus 20 mM fumarate for 14 h before being harvested, washed with 25 ml of 10 mM Tris-HCl (pH 7.5) and resuspended in 1 ml of 10 mM Tris-HCl (pH 7.5). Benzyl viologen (BV) linked assays were carried out at 37°C under anaerobic conditions. A ε578 of 8600 M−1 cm−1 was used for reduced benzyl viologen. The results shown are the average and standard deviation of assays from three biological replicate cultures.

  2. ND, not determined.

Wild type6.3 ± 0.89.4 ± 1.08.4 ± 0.19.3 ± 0.2
napA < 0.18.3 ± 0.9< 0.18.2 ± 0.6
napG 7.2 ± 0.1NDNDND
napL 3.3 ± 0.3NDNDND
nrfA 6.9 ± 0.2< 0.17.0 ± 0.5< 0.1
Figure 2.

Oxygen-limited nitrate-dependent growth (A, C and E) and nitrite accumulation analysis (B, D and F) of WT (filled circles), napA, napG and napL mutants (open circles). Five hundred ml volumes of BHI-FCS plus 20 mM sodium nitrate, contained in 500 ml conical flasks, were inoculated separately with the above strains to an OD600 of 0.1 and not shaken during growth. One ml samples were harvested and OD600 values were measured, after which the cells were spun down and the media removed to determine the concentration of nitrite as described (B, D and F).

The napG mutant showed an interesting and unexpected phenotype, in that some nitrate-dependent growth (Fig. 2C) and nitrite accumulation (Fig. 2D) was observed under oxygen-limited conditions (reaching 10 mM nitrite after 27 h), suggesting that NapG is not absolutely required for nitrate reduction and that electrons are donated to the NapAB complex via a NapG-independent route. One obvious possibility for this is via the NapC homologue Cj1358 (NrfH). This was tested by analysis of the ability of nrfH single and nrfH napG double mutants to grow under oxygen-limited conditions with nitrate, in comparison with the napG mutant. The results are shown in Fig. 3C–F. Whereas the single nrfH mutant showed no nitrate-dependent growth defect (Fig. 3C) and accumulated nitrite in exactly the same way as the WT parent strain (Fig. 3D), the nrfH napG mutant was completely deficient in nitrate-dependent growth (Fig. 3E) and did not accumulate any nitrite (Fig. 3F). Taken together with the phenotype of the napG mutant (Fig. 2C and D), we conclude that electron transport to NapAB is predominantly via NapG, but that NrfH can provide an alternative route.

Figure 3.

Oxygen-limited growth with nitrite (A) or nitrate (C and E) and nitrite accumulation analysis (B, D and F) of WT (filled circles), nrfA, nrfH and nrfH napG strains (open circles). Five hundred ml volumes of BHI-FCS containing 2 mM sodium nitrite (A) or 20 mM sodium nitrate (C and E) contained in 500 ml conical flasks were inoculated separately with WT, nrfA, nrfH and nrfH napG to a final OD600 of 0.1 and not shaken during growth. Further 2 mM additions of sodium nitrite were added to WT and nrfA cultures during the course of the experiment as indicated by the black arrows in (A). Following harvesting, cells were spun down and the media removed to determine the concentration of nitrite as described (B, D and F).

The napL mutant showed a WT phenotype with respect to both oxygen-limited growth with nitrate as electron acceptor (Fig. 2E) and the accumulation of nitrite (Fig. 2F), indicating that NapL is not an essential nap gene. However, measurements of BV-dependent nitrate reductase activity (Table 1) showed that the rate in the napL mutant was ∼50% of that of the WT parent. As reduced BV donates electrons directly to the NapA catalytic subunit, these data would suggest either a lowered amount of NapA is present in the periplasm of the napL mutant, or possibly a decreased catalytic efficiency of NapA in this mutant.

NrfH is the sole electron donor to NrfA and is not involved in electron transfer to the DMSO/TMAO reductase Cj0264

Mutagenesis of nrfA in C. jejuni resulted in the complete absence of NrfA protein in periplasmic extracts, as evidenced by immunoblotting with cross-reactive anti-E. coli NrfA antibodies (data not shown), and the abolition of both BV-dependent nitrite reductase activity (Table 1) and nitrite-dependent growth under oxygen-limited conditions (Fig. 3A). In order to avoid problems of nitrite toxicity in the latter experiment, sequential additions of 2 mM sodium nitrite were made every 3 h, as previously described (Sellars et al., 2002). The nitrite was rapidly consumed by WT cells but not by the nrfA mutant, which showed a step-wise accumulation of the added nitrite (Fig. 3B). A non-polar Cj1358c mutant (which had WT rates of BV-linked nitrate and nitrite reductase activity) was completely deficient in oxygen-limited growth with nitrite (data not shown), showing that NrfH is the sole electron donor to NrfA.

In C. jejuni NCTC 11168, NrfH is the only tetra-haem NapC-like cytochrome c encoded in the genome, yet the bacterium contains a periplasmic DMSO/TMAO reductase (Cj0264), which in other bacteria receives electrons from a NapC-like protein, for example TorC in E. coli (Gon et al., 2001; Gross et al., 2005). TorC has an additional C-terminal mono-haem domain, which is similar in sequence to Cj0265, a small c-type cytochrome, which likely transfers electrons to Cj0264 (Sellars et al., 2002). Although NrfH is the only obvious candidate linking the quinone pool with Cj0265, we found that the growth rate and cell yield of the nrfH mutant on either TMAO or DMSO under oxygen-limited conditions was the same as the WT parent (data not shown). This suggests either Cj0265 is a quinol oxidase or there is a novel route of electron transfer to these electron acceptors in C. jejuni, which merits further investigation.

An nrfA mutant is hypersensitive to nitrosative stress

A single-domain globin, Cgb, has been shown to mediate resistance to NO and nitrosative stress in C. jejuni (Elvers et al., 2004). This protein is only inducible by NO-releasing agents and so other, constitutive, mechanisms of NO resistance might be necessary to prevent cell death upon sudden exposure to NO (as might occur during transfer of C. jejuni into the human gut) until sufficient Cgb protein can be made. NrfA is an attractive candidate, as it has been shown to have a role in NO reduction under anaerobic conditions in E. coli (Poock et al., 2002). Figure 4 shows a series of experiments to test the sensitivity of the WT and nrfA strains to nitrosative stress with different sources of reactive nitrogen species; spermine NONOate (an NO releaser), GSNO (an NO releaser and NO+ donor) and SNAP (predominantly a nitrosating agent by donation of the nitrosonium cation, NO+) (Poole, 2005). In disc diffusion assays (Fig. 4A, C and E), the nrfA mutant showed a statistically significant increase in sensitivity to each of these compounds compared with the WT parent (P < 0.05 in Students t-test). In quantitative kill-curve assays, the nrfA mutant was killed rapidly in the presence of spermine NONOate (Fig. 4B), with GSNO and SNAP producing up to a 3-log reduction in mutant viability over the time course of the experiment compared with the WT (Fig. 4D and F). Thus, the nrfA mutant is clearly hypersensitive to a variety of nitrosative stress inducing reagents.

Figure 4.

Effect of spermine NONOate, GSNO and SNAP on WT and nrfA growth and viability. Disc diffusion assays (A, C and E) were carried out by individually seeding 400 ml of MH agar with liquid cultures of WT (black bars) and nrfA (white bars) to a final OD600 of 0.1. When the plates were set, a sterile disc was aseptically placed in the middle of a seeded plate and 10 μl of a 100 mM stock of either spermine NONOate, GSNO, or SNAP was added to the disc and allowed to dry. The plates were then incubated microaerobically at 37°C for 3 days after which the zones of inhibition were measured. The experiments were performed in triplicate. Viability assays (B, D and F) were performed by adding 0.5 mM spermine NONOate, GSNO, or SNAP to 100 ml cultures of either WT (filled circles) or nrfA (open circles). Twenty μl samples were removed in triplicate and diluted to 10−8 in 200 μl volumes. Five μl aliquots of each dilution were inoculated onto MH plates in triplicate. The plates were incubated microaerobically at 37°C for 3 days and the colonies counted.

Decreased NO consumption and enhanced NO inhibition of respiration in the nrfA mutant: Direct evidence for a role of NrfA in NO detoxification

To further investigate the role of nrfA in NO metabolism, both WT and nrfA strains grown microaerobically to early stationary phase were incubated separately in a Clark-type oxygen electrode additionally fitted with an NO-sensitive electrode (see Experimental procedures) in order to directly monitor the consumption of added NO. Typical traces obtained from the electrode are shown in Fig. 5A (WT) and Fig. 5B (nrfA strain). The duration of the NO electrode response was longer in the nrfA mutant strain (half-time of 600 s in Fig. 5B) compared with the WT (half-time of 120 s in Fig. 5A), indicating a decreased rate of NO consumption. These experiments were performed on three biological replicate cultures of each strain and similar results were obtained.

Figure 5.

Consumption of NO (A and B) and inhibition of respiration by NO (C and D) in WT and nrfA mutant strains. To determine rates of NO consumption, 200 μg cell protein was added to 2 ml of potassium phosphate buffer (25 mM, pH 7.5) in the oxygen electrode chamber, and the oxygen concentration was allowed to reach zero via cellular respiration after addition of 5 mM formate (F). Then, 25 μM NO (final concentration) was added from an anaerobic aqueous solution and the NO electrode response recorded. For NO inhibition of respiration measurements, 50 μg cell protein was added to 2 ml of potassium phosphate buffer (25 mM, pH 7.5) and respiration was initiated by the addition of 5 mM sodium formate (F). Twenty-five μM NO (final concentration) was added when the oxygen concentration reached 50–60% saturation. The experiments were repeated on three different biological replicates of each strain with similar results. The rates of the different phases of respiration were calculated from the traces shown (units of nmol min−1 mg cell protein−1) and can be directly compared between wild type and mutant strains.

To determine whether NrfA is able to protect cells of C. jejuni from NO induced damage, the degree of NO inhibition of respiration was measured in WT and nrfA strains, as it has been established that respiratory oxidases are an important target for NO (Stevanin et al., 2000). Typical results for WT cells are shown in Fig. 5C. Following addition of formate to initiate respiration, the initial rate of oxygen consumption recorded was 680 nmols O2 min−1 mg cell protein−1. At the point at which the oxygen concentration in the chamber reached approximately 50–60% of the initial oxygen concentration (∼130 μM), NO was added to a final concentration of 25 μM. This lowered the respiration rate to 300 nmols O2 min−1 mg cell protein−1 (a reduction of 56% compared with the initial rate). The period of inhibition lasted approximately 3.5 min, after which time the rate increased to 500 nmols O2 min−1 mg cell protein−1. In contrast, the initial rate of respiration for the nrfA mutant (Fig. 5D) was comparable to the WT at 660 nmols O2 min−1 mg cell protein−1, but this was reduced to 120 nmols O2 min−1 mg cell protein−1 upon the addition of NO (82% reduction of the initial rate). The period of NO inhibition of respiration lasted for 5.5 min, after which the rate increased to 240 nmols O2 min−1 mg cell protein−1. Taken together, the decreased cellular NO consumption and increased degree of NO inhibition of oxygen respiration in the nrfA mutant compared with the WT parent strain strongly suggests a role for NrfA in NO detoxification in C. jejuni under aerobic and anaerobic conditions.

Nitrate and nitrite induce cgb expression

Studies in E. coli have shown that respiratory growth with nitrate and nitrite can produce reactive nitrogen compounds such as nitric oxide and nitrous acid (HNO2) (Ji and Hollocher, 1988; Corker and Poole, 2003; Weiss, 2006). The condensation of NO can ultimately lead to the formation of dinitrogen trioxide (N2O3), a nitrosating agent that can induce mutagenic lesions by deamination of DNA bases (Weiss, 2006). Thus, growth of C. jejuni on nitrate or nitrite might generate nitrosative stress and therefore induce mechanisms that serve to protect against reactive nitrogen species. It has previously been demonstrated that Cgb, a single domain globin, performs a major role in NO detoxification in C. jejuni (Elvers et al., 2004), and more recently that cgb is a component of a nitrosative stress response regulon controlled by the transcriptional regulator, NssR (Elvers et al., 2005). Previously, cgb expression has been shown to be induced by GSNO and sodium nitroprusside (SNP) (Elvers et al., 2004; 2005). However, because GSNO releases both NO and NO+, and SNP releases NO+, it is not known whether NO per se can specifically induce cgb, and the possibility that nitrate or nitrite might also result in cgb induction has not been investigated. To determine the specificity of cgb induction, the expression of cgb was monitored using a plasmid (pKE117) based cgb–lacZ fusion (Elvers et al., 2005) in the presence or absence of a range of nitrogen oxides (Fig. 6). We confirmed cgb induction by GSNO and SNP but also found greater induction by the more specific NO releasers NOC-18 and spermine NONOate. Significantly, lacZ was clearly induced by both nitrate and nitrite (30 and 20-fold greater, respectively, than uninduced cells (Fig. 6). A cis-acting motif for NssR binding has been identified previously (Elvers et al., 2005). The plasmid pKE120 is identical to pKE117 except that it contains two base changes within the recognition sequence, and as a consequence, fails to induce LacZ expression in response to nitrosative stress (Elvers et al., 2005). The pKE120 cgb–lacZ fusion did not respond to the presence of any nitrogen oxide, confirming that the induction was dependent on NssR.

Figure 6.

β-Galactosidase activities of cells of C. jejuni strain 480 containing either of two variants of the promoter probe pMW10 grown in the presence and absence of the nitrogen oxide compounds indicated. The black bars represent activities in strains containing pKE117 (wild-type cgb–lacZ fusion) and the grey bars those measured in strain harbouring pKE120 (cgb–lacZ fusion with altered NssR-recognition sequence). Values are the means ± the standard error of three separate experiments.

Induction of cgb expression in napA and nrfA mutants

As both nitrate and nitrite induced cgb expression in an NssR-dependent manner, it seems likely there is a mechanism for generating NO from one or both of these anions, although at this stage we cannot rule out the possibility that NssR may also bind and respond directly to nitrate and/or nitrite. NO can be generated during nitrate and nitrite metabolism in E. coli (Weiss, 2006) and thus, the expression of Cgb in response to various nitrogen oxides was assessed in napA and nrfA mutants by immunoblotting, using an anti-Cgb antibody as described previously (Elvers et al., 2004). In all strains, Cgb was not detectable in Mueller-Hinton (MH) broth alone, but strongly induced by the presence of GSNO (Fig. 7). While Cgb expression was induced in the parental strain and the nrfA mutant by nitrate and nitrite, in the napA mutant elevated expression only occurred in the presence of nitrite (Fig. 7). The napA mutant is not able to reduce nitrate to nitrite and therefore the induction of cgb expression by nitrate in C. jejuni must be a consequence of its reduction to nitrite.

Figure 7.

Cgb production as assessed by immunoblotting with anti-Cgb antibody. Wild type, napA::kan and nrfA::cat strains were grown microaerobically for 3 h and then incubated for 2 h in the absence (lane 1) or in the presence of GSNO (0.05 mM; lane 2), sodium nitrite (6.25 mM; lane 3) or sodium nitrate (50 mM; lane 4).

The mechanisms for the formation of NO and its congeners from nitrite in cultures of C. jejuni will be complex, but NrfA itself may be a potential source of NO in the presence of high concentrations of nitrite (Corker and Poole, 2003). However, the fact that cgb is still induced by nitrite in an nrfA mutant, argues for an NrfA-independent route of NO formation from nitrite. We also found that in the presence of 20 mM sodium nitrite, the respiration rate of both the WT and the nrfA mutant was inhibited by about 36%, and when 50 mM sodium nitrite was added, the respiration rate of each strain was inhibited by about 63%. If significant NO was being produced by NrfA as a by-product of nitrite reduction, the inhibition of respiration would be expected to be much less in the absence of the enzyme (note that these cells were grown in cgb non-inducing conditions).

Cgb and NrfA alleviate the effects of nitrosative stress during microaerobic growth on nitrite

The observation that Cgb is induced in WT cells during growth on nitrate or nitrite suggests that these anions can cause nitrosative stress in C. jejuni. To directly address the question of whether Cgb can confer protection under these conditions, growth experiments were carried out with WT, cgb and nssR mutants in the presence and absence of sodium nitrate or sodium nitrite. In the absence of these anions, the mutant strains grew in a manner similar to that of the WT (Fig. 8). Growth of all strains was also equivalent in the presence of sodium nitrate even at the highest concentration tested (200 mM; data not shown). However, while the parental strain grew in the presence of 6.25 mM sodium nitrite (Fig. 8), albeit after an extended lag phase, the cgb and nssR mutants did not, suggesting that Cgb (and perhaps other members of the NssR regulon; Elvers et al., 2005) confer protection against nitrosative stress during nitrite reduction. Similar experiments were carried out with the nrfA mutant and its isogenic parent strain. Again, in the absence of nitrite, no difference was detected in the growth rates in MH or BHI-FCS media, but in the presence of 6 mM sodium nitrite in BHI-FCS, the respective doubling times were 4 h for the WT and 20 h for the nrfA mutant, suggesting that NrfA also has some protective effect during growth with nitrite, although less than that seen with the cgb and nssR mutants.

Figure 8.

The inhibition of growth of the cgb and nssR mutants by sodium nitrite. Cells of C. jejuni NCTC 11168 (▴, ▵), CJCGB01 (▪, □) and CJNSSR01 (●, ○) were grown microaerobically in MH broth in the presence (open symbols) and absence (closed symbols) of 6.25 mM sodium nitrite. Growth was assessed by measuring OD600. Similar results were reproducibly obtained in at least three separate experiments.

NrfA is not required for colonization of the chicken gut

Campylobacter jejuni is able to colonize the caecum of chickens and establish a stable commensal relationship with its host. In this niche, protection mechanisms for host-derived nitrosative stress may be less important than in a mammalian host where a more active inflammatory response occurs. To determine if nrfA was essential for chicken colonization, an nrfA mutant was constructed in the efficient chicken colonizing strain 81–176 using pnrfA::cat. The resulting mutant was motile and lacked benzyl-viologen-dependent nitrite reductase activity. This nrfA mutant and the parental WT were used to infect separate batches of 10 2-week-old chickens. After 2 weeks, the birds were slaughtered and caecal counts in each bird were determined. The numbers of Campylobacter in the caecal material of these birds were 1.74 ±  1.0 × 109 cfu g−1 and 3.63 ± 2.8 × 109 cfu g−1 for the WT and the nrfA mutant respectively. The result suggests that the absence of NrfA does not affect the ability of C. jejuni to colonize chickens, and thus that nitrite reduction or NrfA mediated NO detoxification is not an important factor in this niche.

Discussion

This study and our previous work (Sellars et al., 2002) have shown that C. jejuni is able to use nitrate and nitrite as terminal electron acceptors for growth under oxygen-limited conditions by employing periplasmic Nap and Nrf systems. Figure 1B presents a model summarizing the relevant electron transport pathways as suggested by our data. These periplasmic systems are clearly also involved in the detoxification of reactive nitrogen species. There is increasing evidence that nitrate respiration may play a significant role in the growth of human and animal pathogens in vivo (Weber et al., 2000). In a wide range of Gram-negative bacterial pathogens, including C. jejuni, periplasmic nitrate reductases are more commonly present than Nar-type enzymes (Potter et al., 2001). Nitrate concentrations in human body fluids are in the range 10–50 μM (Potter et al., 2001) and the Nap enzyme has a much higher affinity for nitrate compared with the membrane bound Nar (Potter et al., 1999) making Nap ideally suited to a role in scavenging the low nitrate concentrations encountered in vivo.

Inspection of the gene organization has revealed some novel features of the nap and nrf operons. First, unlike the closely related W. succinogenes, there is no NapF homologue in C. jejuni. In E. coli, NapF has been shown in vitro to be able to insert an iron-sulphur cluster into the N-terminal end of NapA (Olmo-Mira et al., 2004). The C. jejuni NapA possesses four cysteine residues at the N-terminus, with three arranged in a CX2CX3C sequence. In the absence of NapF, an iron-sulphur cluster is presumably assembled around these cysteines via the general cytoplasmic Fe–S cluster machinery (Isc/NifS homologues). Second, the napL gene seems to have some role in regulating the amount of NapA in the periplasm, based on the lower BV-linked nitrate reductase activity we observed, but mutation of this gene did not affect growth or the pattern of nitrite accumulation, so the precise function of NapL still remains undefined.

There is no napC homologue present in the nap operon of C. jejuni and therefore the mechanism by which quinol oxidation is coupled to reduction of the NapAB complex is not immediately obvious. However, from the mutant data obtained in this study, it seems that there are two routes of electron transport to NapAB; one via NapG, and the other involving the NapC/NirT homologue, Cj1358 (see Fig. 1B). Interestingly, in E. coli, NapG and NapH form a ubiquinol dehydrogenase complex, coupling nitrate reduction to ubiquinol oxidation (Brondijk et al., 2002; 2004). However, unlike E. coli, which possesses three types of quinone (ubiquinone, menaquinone and demethylmenaquinone), C. jejuni possesses only menaquinone-6 and dimethylmenaquinone-6 (Moss et al., 1984). Therefore, the quinone specificity of NapG/NapH is clearly very different between these two organisms. The Cj1358c gene encodes the only tetra-haem c-type cytochrome in the genome of NCTC11168, and is contiguous with nrfA. Following inactivation of Cj1358c in a napG+ background, oxygen-limited growth is normal with nitrate but can no longer occur when nitrite is present as the electron acceptor. Thus, the primary role of Cj1358 is clearly that of an electron donor to NrfA, as in W. succinogenes and related epsilon proteobacteria (Simon et al., 2000; Simon, 2002), where NrfH forms a tight complex with NrfA. In an nrfH mutant, considerable amounts of NrfA are probably attached to the membrane as it appears to be a very hydrophobic protein (J. Simon, pers. comm. and our unpubl. obs.). In the absence of NapG, there is a small amount of nitrate reduction in C. jejuni that we showed is attributable to Cj1358, and although the overall contribution to nitrate reduction is small, it is nevertheless enough to support some growth of the napG mutant when oxygen tensions are low and nitrate is present as an alternative electron acceptor (Fig. 3).

The penta-haem NrfA is a well-studied enzyme, which in most bacteria, contains an unusual CXXCK motif at the haem 1 binding site, instead of the CXXCH motif found in the vast majority haem c containing enzymes. This requires a dedicated haem lyase for covalent attachment of the active site haem (Pisa et al., 2002). The C. jejuni NrfA homologue, Cj1357, is slightly larger at 69 kDa when compared with the W. succinogenes NrfA (58 kDa), and contains a novel ligation site at haem 1, with a CXXCH motif instead of CXXCK. This would explain the absence in C. jejuni of genes encoding Nrf-specific haem lyases that are found in other bacteria (Einsle et al., 2000), but a site-directed His for Lys substitution in the W. succinogenes NrfA results in low catalytic activity (Pisa et al., 2002). In contrast, the C. jejuni enzyme clearly has high specific activity in vivo (Table 1). Interestingly, the predicted NrfA's of the closely related Campylobacter coli, Campylobacter lari and Campylobacter upsaliensis (Fouts et al., 2005) and Helicobacter hepaticus (Suerbaum et al., 2003) also contain the haem1 CXXCH motif rather than CXXCK.

NrfA-mediated NO reductase activity has long been known (Costa et al., 1990) but was first shown to have physiological relevance in E. coli more recently (Poock et al., 2002). In this study, we have provided evidence that C. jejuni NrfA can reduce nitric oxide, in addition to nitrite, and this is the first demonstration of NrfA-mediated NO detoxification outside of the Enterobacteriaceae. Loss of nrfA resulted in an increased sensitivity to a range of NO-releasing and nitrosating agents, a decreased ability to consume NO in solution and a much greater degree of NO-mediated inhibition of respiration. During colonization of the human gut, NO will be encountered during the respiratory burst of various cells of the human immune system, such as macrophages. The periplasmic location of NrfA may provide a mechanism to detoxify some of this NO before it can diffuse across the lipid bilayer of the cytoplasmic membrane and cause damage to numerous cellular targets. The cytoplasmic single domain globin, Cgb, undoubtedly represents a major defence against NO in C. jejuni, but cgb transcription is strictly inducible in an NssR-dependent manner. Previous studies have revealed that Cgb is not fully induced even after 20 min exposure to 50 μM GSNO (Elvers et al., 2004). C. jejuni might therefore be at risk of significant damage from reactive nitrogen species before Cgb can be induced. We detected high NrfA activities in cells grown under both microaerobic and oxygen-limited conditions (Table 1) which had not been exposed to nitrite or any NO releasing agents, and there is no evidence for nrfA upregulation in response to nitrosative stress (Elvers et al., 2005). NrfA may therefore represent a constitutive ‘first line of defence’ NO detoxification mechanism. Such a mechanism is not essential for colonization of the avian gut as an nrfA mutant of strain 81–176 colonized 2 week old chickens to the same extent as that of the WT parent strain. Similar results have been obtained in separate colonization studies with the cgb mutant (K.T. Elvers et al., unpublished results). These results perhaps reflect a lesser need for NO management in the avian compared with mammalian gut environments.

This study has revealed that both nitrate and nitrite were able to induce cgb expression in WT cells. Further analysis of cgb expression in napA and nrfA mutants revealed that expression did not occur in the napA mutant with nitrate but did occur in the nrfA mutant regardless of whether nitrate or nitrite was added. Reactive nitrogen species activating NssR must therefore derive from nitrite and not nitrate. Whether NssR can also bind nitrite directly cannot be addressed until the protein is purified. It is interesting to note that NsrR, a regulator in E. coli unrelated to NssR, mediates derepression of several nitrosative stress responsive genes by nitrite as well as a variety of NO-sources (Bodenmiller and Spiro, 2006). The E. coli NrfA has been shown to produce NO during the reduction of high concentrations of nitrite (Corker and Poole, 2003). However, NrfA mediated NO production in C. jejuni under the conditions used here is unlikely because Cgb was still induced by nitrite in an nrfA mutant and the respiration rates of both the nrfA mutant and the WT parental strain were inhibited by nitrite to the same degree. The inhibition observed must be attributable to the nitrite itself or presumably NO (or its congeners) produced by an unknown (NrfA-independent) mechanism.

Microaerobic growth of C. jejuni on nitrite appears to cause a degree of nitrosative stress, and the complete lack of growth of the cgb and nssR mutants and slower growth of the nrfA mutant in the presence of a concentration of nitrite which allows growth of the WT parent, indicates a protective effect of Cgb and NrfA under these conditions. However, the fact that the growth of all of these strains is unaffected on nitrate, indicates that this anion causes much less nitrosative stress, even though the nitrite-producing NapA is present. From the oxygen-limited growth experiments in Fig. 2 and the previous results of Sellars et al. (2002), it is also apparent that growth of WT cells on nitrate causes a continuous accumulation of nitrite without any growth inhibition. It seems that when cells are exposed to high starting concentrations of nitrite they experience much greater toxicity than when the nitrite accumulates during growth.

In conclusion, through mutant analysis we have highlighted some interesting features of the nitrate reductase system in C. jejuni, in particular the ability of NapC-like NrfH to act as a minor electron donor to NapAB in addition to its major role as the redox partner for NrfA. We have also shown that in addition to a clear role in nitrite reduction, NrfA has a role in protection of C. jejuni against nitrosative stress both directly from added NO and also during growth with nitrite. Whereas NrfA is constitutive, the single domain globin Cgb was shown here to be induced by nitrite, in addition to NO, and was found to have a major protective role against nitrosative stress caused by growth with nitrite. The results emphasize a dual role for NrfA and the importance of the NrfA and Cgb proteins in the physiology of C. jejuni, and are also relevant to understanding the response to stresses encountered by the bacterium in the host environment.

Experimental procedures

Bacterial strains, media and culture conditions

Campylobacter jejuni strain NCTC 11168 was used for the majority of experiments this study. For chicken colonization experiments strain 81–176 (Korlath et al., 1985) and an isogenic nrfA mutant were used. Strains were grown at 37°C in a microaerobic atmosphere [10% (v/v) O2, 5% (v/v) CO2, 85% (v/v N2)] in a MACS-VA500 growth cabinet (Don Whitley Scientific, Shipley, UK) either on Columbia agar (Oxoid, Basingstoke, UK) containing 5% (v/v) lysed horse blood and amphotericin B and vancomycin (10 μg ml−1 each) or MH agar (Oxoid, Basingstoke, UK) supplemented with amphotericin B and vancomycin (10 μg ml−1). The C. jejuni mutants CJCGB01 (Cgb) and CJNSSR01 (NssR) have been described previously (Elvers et al., 2004; 2005). Kanamycin and chloramphenicol were added to plates at final concentrations of 30 μg ml−1 to select for C. jejuni nap and nrf mutants where relevant. Microaerobic liquid cultures of C. jejuni were grown in 25–100 ml batches of either MH broth or Brain Heart Infusion broth supplemented with 5% (v/v) foetal calf serum (BHI-FCS) contained in 250 ml shake flasks in the above microaerobic atmosphere and orbitally shaken at 200 r.p.m. Oxygen-limited growth was achieved in the same gas atmosphere, but severely limiting the diffusion of oxygen by containing 500 ml of BHI-FCS within unshaken 500 ml flasks, as described by Sellars et al. (2002). To supplement oxygen-limited growth, 20 mM sodium fumarate, 20 mM sodium nitrate or 12 mM sodium nitrite (added to 2 mM every 3 h) were provided as alternative electron acceptors where appropriate. E. coli was routinely grown at 37°C with shaking at 200 r.p.m. in Luria–Burtani (LB) supplemented with appropriate antibiotics. Growth of C. jejuni and E. coli was monitored by measuring the OD at 600 nm using a Pharmacia Biotech Ultrospec 2000 spectrophotometer.

DNA isolation and manipulation

Plasmid DNA was isolated from E. coli using the Promega miniprep kit (Madison, Wisconsin, USA). C. jejuni chromosomal DNA was isolated using the Wizard® Genomic DNA purification kit (Promega). Standard techniques were employed for the cloning, transformation, preparation and restriction analysis of plasmid DNA extracted from E. coli (Sambrook et al., 1989).

The use of cgb–lacZ fusions to monitor cgb expression

Campylobacter jejuni strain 480, containing the cgb–lacZ reporter plasmids pKE117 or pKE120 and the use of these as surrogate strains for NCTC 11168, which cannot be transformed with plasmids, has been described previously (Elvers et al., 2005). They were used here to assay induction of the cgb promoter by nitrogen oxides. Overnight MH broth cultures of C. jejuni 480(pKE117) or C. jejuni 480(pKE120) were adjusted to an OD600 of 0.5. Two millilitres of adjusted cell suspension were inoculated into flasks containing 11 ml of MH-broth, and shaken microaerobically at 37°C for 5 h at 100 r.p.m. After this time GSNO (0.05 mM), SNP (0.05 mM), sodium nitrite (6.25 mM), sodium nitrate (50 mM), NOC 18 (0.05 mM) or spermine NONOate (0.05 mM) were added and incubation continued for a further 2 h. Ten millilitres of cells were harvested by centrifugation at 6000 g at 4°C for 15 min and resuspended in 1 ml of Z buffer. β-Galactosidase activity was then measured in 0.1 ml aliquots of these suspensions using o-nitrophenyl-β-d-galactopyranoside as described previously (Wösten et al., 1998). The experiment was repeated three times.

Detection of Cgb in napA and nrfA mutants using immunoblotting

Overnight plate cultures of C. jejuni NCTC 11168, napA::kan and nrfA::cat were harvested in MH broth and OD600 adjusted to 0.5. Aliquots (1.6 ml) of this suspension were added to 10 ml of MH-broth and the cells incubated microaerobically for 3 h at 37°C with shaking at 100 r.p.m. Cultures were induced by the addition of GSNO (0.05 mM final concentration), sodium nitrate (50 mM), or sodium nitrite (6.25 mM) and incubated for a further 2 h under the conditions described above. The cells were harvested by centrifugation, washed once with 0.15 M Tris-HCl pH 7 and resuspended in 100 μl of Tris-HCl, 0.2 mg ml−1 lysozyme and 0.1 mM EDTA. The cells were lysed by five cycles of freeze thawing and centrifuged. Total soluble proteins were extracted in the supernatant. Protein concentrations were determined by the Bradford assay and 7.5 μg of each protein was loaded on a 12% polyacrylamide gel. The samples were not boiled and the loading buffer did not contain β-mercaptoethanol. Western blotting and detection of Cgb were carried out as described previously (Elvers et al., 2004).

Detection of NapA and NrfA in periplasmic fractions by immunoblotting

Campylobacter jejuni periplasm was isolated by the cold osmotic shock method described by Myers and Kelly (2005). The final supernatant obtained in this procedure was verified to have a high cytochrome c content (periplasmic marker) and low isocitrate dehydrogenase activity (cytoplasmic marker), using the assays described by Myers and Kelly (2005). Periplasmic proteins were separated by SDS-PAGE using a Mini-PROTEAN 3 apparatus (BIO-RAD, California, USA). Transfer of proteins was carried out using a Mini Trans-Blot Cell (BIO-RAD). The gel-blot sandwich was constructed according to manufacturer's instructions and the proteins transferred to a nitrocellulose membrane (Hybond-C Extra, Amersham Biosciences) at a current of 11 mA for 14 h. All immuno-detection steps were carried out at room temperature with constant agitation. TBS-T (25 mM Tris-HCl pH 7.4, 130 mM NaCl, 0.1% Tween 20) was used as both a base for blocking agent (5% bovine serum albumin dissolved in TBS-T) and for washing. Primary antibodies (Anti-NapA and Anti-NrfA raised in rabbit from purified E. coli NapA and NrfA respectively) were a kind gift from D. Richardson (University of East Anglia, UK) and were diluted in blocking agent (1:1000) and applied to the membrane. Membranes were reacted for approximately 1 h and washed in TBS-T before the secondary antibody (monoclonal Anti-Rabbit IgG, Sigma) was diluted (1:5000) and applied to the membrane. Antibody binding was visualized by means of Enhanced Chemi-luminescence (ECL Kit, Amersham Biosciences UK), according to manufacturer's instructions.

Inactivation of nap and nrf genes

A set of PCR primers was designed to amplify the entire coding regions of napG (741 bp), napL (915 bp), nrfA (1833 bp) and nrfH (774 bp total because of the addition of a 258 bp sequence upstream of the nrfH ATG start codon) for insertional inactivation following subsequent cloning into appropriate vectors. The primers used were: napG-F (5′-ATGAAAGGTAGAGAATTTTTCGTT-3′), napG-R (5′-TCACAGTTCTCCATCATTTAAATA-3′), napL-F (5′-ATGAAAAAATTTCTTTTTATTTTG-3′), napL-R (5′-TTATTCATCAATACTCCTAAACAT-3′), nrfA-F (5′-ATGAAAAAAAATATTTTACGCTTA-3′), nrfA-R (5′-TTATTTTTTATCAATCACTTGATA-3′), nrfH-F (5′-CTAAAACACGAGAATTAAAGCGAA-3′), nrfH-R (5′-TAAATTCCGTGTTTATGTCCAACA-3′). The napG, nrfA and nrfH genes were amplified from C. jejuni NCTC 11168 chromosomal DNA using Taq DNA polymerase (Promega), while napL was amplified using Pwo. The napG and nrfH genes were cloned into pGEM®-T-Easy (Promega), nrfA was cloned into pGEM®3Zf(–) (Promega), and napL was cloned into pET101-TOPO® (Invitrogen), according to the manufacturer's instructions. This generated plasmids, pnapG, pnapL, pnrfA and pnrfH. A chloramphenicol acetyltransferase (cat) cassette originating from C. coli (Wang and Taylor, 1990) was blunt-end cloned into unique, and previously Klenow (Promega) in-filled Bsu36I pnrfA to generate plasmid pnrfA::cat. This cassette was also used to generate an nrfH mutant, by cloning into pnrfH, previously cut with Bpu10I and in-filled with the Klenow fragment, to generate pnrfH::cat. To prevent polar effects, primers cat-F (5′-GAATTCCTGCAGCCC-3′) and cat-R (5′-ACTAGTGGATCCCGG-3′) were used to amplify the cat gene from plasmid pAV35 (Van Vliet et al., 1998). These primers eliminated a rho-independent terminator from the 5′ end of the cat cassette. The ahpAIII kanamycin resistance cassette from C. coli (Wang and Taylor, 1990; van Vliet et al., 1998) was cloned into unique NheI and AccI (previously Klenow in-filled) sites of pnapG and pnapL, respectively, generating plasmids pnapG::kan and pnapL::kan. Plasmid pGEM-napA::kan used for construction of a C. jejuni napA mutant, was a kind gift from Dr Duncan Gaskin (IFR, Norwich, UK) and contained an internal deletion of most of the napA gene. Transformation of C. jejuni NCTC 11168 with plasmids pnapG::kan, pnapL::kan, pGEM-napA::kan, pnrfA::cat and pnrfH::cat was carried out by electroporation and transformants were selected on Colombia blood agar plates supplemented with chloramphenicol or kanamycin at final concentrations of 30 μg ml−1. pnrfA::cat was also transformed into strain 81–176 for chicken colonization studies (see below). Colonies were re-streaked onto Colombia blood agar plates and correct insertion of the antibiotic resistance cassettes into the target genes was verified by extraction of chromosomal DNA by MicroLYSIS (Web Scientific, Crewe, UK) according to manufacturer's instructions. PCR using the gene-specific primers listed above confirmed allelic exchange by double cross-over, as demonstrated by an increase in amplicon size of approximately 0.9 or 1.4 kb for the chloramphenicol or kanamycin cassette insertions respectively. A C. jejuni napG nrfH double mutant was constructed by electroporation of C. jejuni napG with pnrfH::cat and selecting on Columbia blood agar plates containing both kanamycin and chloramphenicol.

Real-time PCR

Wild type or mutant MH broth cultures (100 ml) were harvested directly into a mix of 6.25 ml of prechilled phenol made up in 118.75 ml of 100% ethanol, to stabilize the RNA. Samples were then centrifuged at 8000 g for 4 min (4°C). Total RNA was purified from cell pellets using an RNeasy Mini kit (Qiagen, UK) as recommended by the suppliers. The RNA concentration and purity were determined using a Beckman DU® 650 spectrophotometer. cDNA synthesis was carried out using 4 μg of starting material, primed with 9 μg pd(N)6 random hexamers (Amersham Biosciences, UK). Reaction mixes (20 μl) containing 0.5 mM dATP, dTTP, dGTP and dCTP were incubated for 2 h at 42°C with 200 units Superscript II RNase-H Reverse Transcriptase (Invitrogen). Following synthesis, cDNA was purified using a PCR purification kit (Qiagen) to remove unincorporated dNTPs and primers. Gene-specific primers were designed to amplify 50–150 nucleotide fragments of the gyrA (internal control), napB or napD genes using primer 3 software (Rozen and Skaletsky, 2000). A SYBR green mix was made in the ratio; 13 μl of Quantace sensimix (bioline), 0.5 μl of SYBR green and 4.5 μl of nuclease-free water (Sigma). Each reaction was carried out in a total volume of 25 μl on a 96-well optical reaction plate (Applied Biosystems). Each well contained 16 μl of SYBR green mix (above), 12.5 pmol of each of the three primers and 5 μl of cDNA sample. PCR amplification was carried out in an ABI 7700 thermocycler (PE Applied Biosystems) with the thermal cycling conditions at 50°C for 2 min; 95°C for 10 min; followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. The data were analysed using the Sequence Detector System (SDS) software (PE Applied Bio-systems) and further processed in Microsoft EXCEL. A standard curve was established for each gene studied using genomic DNA to confirm that the primers amplified at the same rate and to validate the experiment. The relative levels of expression of napB in the napA and napG mutants, and of napD in the napL mutant compared with WT were calculated following the protocol for the Standard Curve Method in the User Bulletin #2 (ABI Prism 7700 Sequence Detection System, Subject: Relative Quantification of Gene Expression) supplied by Applied Biosystems. No-template reactions were included as negative controls.

Determination of nitrite concentrations in culture supernatants

Diluted culture supernatants (50 μl) from oxygen-limited growth experiments were added to 850 μl of 1% (w/v) sulphanilamide (Sigma) dissolved in 1 M HCl and 100 μl of 0.02% (w/v) naphthylethylenediamine (Sigma). After 15 min, the absorbance at 540 nm was measured using a Pharmacia Biotech Ultrospec 2000, and nitrite concentrations were determined by reference to a standard curve.

Nitrate and nitrite reductase assays

Benzyl viologen-linked reductase assays with intact cells were carried out as described by Sellars et al. (2002) in a 1 ml volume using a Shimadzu UV-2401 PC spectrophotometer. The assay mixture consisted of 10 mM Tris-HCl (pH 7.5), 100 μM benzyl viologen (Sigma) and 5 mM nitrate or nitrite, contained in a screw-topped quartz cuvette (Hellma) fitted with a silicone seal to prevent gas exchange. After addition of cells, and sparging with nitrogen gas for approximately 10 min, aliquots of a freshly prepared sodium dithionite (Sigma) solution were injected into the cuvette until the absorbance at 578 nm was stable at approximately 1.5 units. The assay was started by the injection of either nitrate or nitrite into the cuvette. Rates of reductase activity were calculated using an extinction coefficient (ε578) for benzyl viologen of 8600 M−1 cm−1. Protein concentrations of cell suspensions were determined using the method of Markwell et al. (1978).

Disc diffusion assays

One hundred millilitre cultures of C. jejuni NCTC 11168 and isogenic nrfA strains were grown microaerobically at 37°C to early stationary phase and the OD600 values adjusted to 1.0 with BHI. Forty millilitres of each culture were added separately to 400 ml of MH agar, pre-equilibrated to 42°C. Following pouring of plates, a sterile paper disc made from Whatman No. 1 paper (Whatman, Brentford, UK) was placed in the centre of the plate and 5 μl of a 100 mM stock of either S-nitrosoglutathione, spermine NONOate, or S-nitroso-N-acetylpenicillamine was added to the disc. The plates were incubated microaerobically at 37°C for 3 days and the diameter of the zones of inhibition created around the discs were measured.

Effect of nitrosative stress on viability

One hundred millilitre cultures of C. jejuni NCTC 11168 and isogenic nrfA strains were grown microaerobically at 37°C to early stationary phase and the OD600 values adjusted to 1.0 with BHI. Final concentrations (0.5 mM) of either S-nitrosoglutathione, spermine NONOate, or S-nitroso-N-acetylpenicillamine were added to the cultures of each strain, and at time points 0, 30, 60, 90, 120, 150, 180, 210 and 240 min, 20 μl aliquots were removed in triplicate and diluted to 10−8 in 200 μl volumes. Five microlitre aliquots of each dilution were plated in triplicate onto MH plates and incubated at 37°C for 3 days, after which time the colonies were counted.

Measurement of the consumption of nitric oxide

Liquid cultures of C. jejuni WT and nrfA strains were grown microaerobically to stationary phase (OD600 of approximately 1.0) as described above. Cells were harvested by centrifugation, washed once in 25 mM potassium phosphate buffer (pH 7.5), resuspended in 0.5 ml of the same buffer, and kept on ice. Cells (approximately 160 μg total cell protein for either WT or nrfA strain) were added to 2 ml of potassium phosphate buffer (pH 7.5) in a constantly stirred oxygen electrode maintained at 37°C as described below and the chamber closed with a Perspex stopper. An NO electrode (ISO-NO™; WPI, Stevenage, UK), connected to a NO meter (ISO-NO Mark II, WPI) was inserted into the oxygen electrode chamber through the entry point in the Perspex stopper. The stopper was modified by having a hole drilled through the length of the Perspex to allow further additions to be made to the chamber with a Hamilton syringe. The NO electrode was protected by a steel sleeve that has a gas permeable membrane at the tip and it was ensured that this was fully immersed in the buffer in the oxygen electrode chamber. Respiration was initiated by injecting 5 mM sodium formate into the chamber and when the oxygen concentration reached zero, NO was injected into the chamber to a final concentration of 25 μM from a saturated solution of NO in anaerobic phosphate buffer (approximately 1.8 mM NO) and the NO electrode response recorded.

The effect of nitric oxide on the respiration of C. jejuni

Respiration rates were measured with a Clark-type polarographic oxygen electrode (Rank Brothers, Bottisham, Cambridge, UK), comprising a water-jacketed (37°C) Perspex chamber that was stirred magnetically. The chamber was filled with 2 ml of potassium phosphate buffer (pH 7.5), and cells were added (approximately 50 μg protein). The chamber was closed by fitting a Perspex stopper, and the assay initiated by injection of 5 mM sodium formate (Sigma). NO additions were made by injection of a saturated solution of NO in anaerobic phosphate buffer (approximately 1.8 mM NO) when the oxygen concentration within the chamber had reached approximately 110 μM. The electrode was calibrated by air-saturated buffer, and 100% saturation was assumed to be 220 μM O2. Anoxia was achieved by adding a few grains of sodium dithionite. Protein concentrations of cell suspensions were determined using the method of Markwell et al. (1978).

Chicken colonization experiments

Chicken colonization experiments were based on methodology used in previous studies (Jones et al., 2004; Velayudhan et al., 2004). Specific pathogen free, Light Sussex chickens were given 0.1 ml of Campylobacter-free gut flora (See Velayudhan et al., 2004) on the day-of-hatch. Inoculated birds were then housed in separate bio-secure rooms until used for colonization trials. The birds were fed a vegetable-based diet (Special Diet Services, Manea, Cambridgeshire, United Kingdom) ad libitum throughout the trials. Groups of 10, 2-week-old birds, with a developed gut flora were inoculated orally with 0.1 ml per bird of an overnight MH broth culture containing log10 9 cfu of C. jejuni 81–176 or an isogenic nrfA mutant constructed as described above. At 2 weeks post infection birds were euthanized and the caecal contents were removed for bacterial enumeration. Serial dilutions of caecal contents were plated directly onto Campylobacter blood-free selective plates, which had been prepared according to the manufacturer's instructions from Campylobacter blood-free selective agar (CCDA; CM739; Oxoid) and CCDA selective supplement (SR155; Oxoid, UK).

Acknowledgements

This work was supported by a grant from the UK Biotechnology and Biological Sciences Research Council (BBSRC) to D.J.K. for M.P. and by a BBSRC grant for K.E. to R.K.P. and S.F.P. We thank Dr Duncan Gaskin for the napA deletion plasmid, Professor David Richardson for the gift of antibodies against NapA and NrfA, John Atack for expert technical assistance, Professor Julie Scholes for access to the ABI thermocycler, Ed Guccione for information technology assistance, and D. Richardson and J. Simon for useful discussions.

Ancillary