Bacterial strains and plasmids used in this study are listed in Table 1. C. jejuni was routinely cultured under microaerophilic conditions (5% O₂, 10% CO₂, 75% N₂) on Blood Agar Base No. 2 (BA) medium containing 5% horse blood or in Heart Infusion broth (HI; Oxoid). Kanamycin (25 μg ml⁻¹) and/or chloramphenicol (15 μg ml⁻¹) and/or spectinomycin (30 μg ml⁻¹) were added when appropriate. E. coli strains were routinely grown at 37°C in Luria–Bertani (LB) broth or on LB agar plates supplemented with ampicillin (50 μg ml⁻¹), kanamycin (30 μg ml⁻¹) or chloramphenicol (34 μg ml⁻¹).

To disrupt the dcuA, dcuB, dcuC, dcuC2, or dctA genes, the genes as well as ~1 kb of the flanking regions were first amplified by PCR using the primer pairs dcuA-F/dcuA-R, dcuB-F/dcuB-R, dcuC-F/dcuC-R, dcuC-2F/dcuC-2R, or dctA-F/dctA-R, respectively. Primers are listed in Table 2. The ~3 kb PCR fragments were ligated into pJET1.2/blunt Cloning Vector resulting into the plasmids pJETdcuA, pJETdcuB, pJETdcuC, pJETdcuC2, and pJETdctA. Inverse PCR was performed on the plasmids pJETdcuA, pJETdcuB, pJETdcuC, and pJETdctA using the primers sets dcuABamHI F/dcuABamHI R, dcuBBamHI F/dcuBBamHI R, dcuCBamHI F/dcuCBamHI R, or dctABamHI F/dctABamHI R, respectively, to delete the dcuA, dcuB, dcuC, and dctA genes present in pJET and to introduce a BamHI restriction site. The pJETdcuA and pJETdctA inverse PCR fragment were ligated to a BamHI fragment containing a chloramphenicol resistance gene of pAV35 resulting in the knock-out constructs pJETdcuA::Cm and pJETdctA::Cm. The pJETdcuB and pJETdcuC inverse PCR fragment were ligated to a BamHI fragment containing the kanamycin resistance gene of pMW2 or spectomycin resistance gene of pNBspec resulting into the knock-out constructs pJETdcuB::Km and pJETdcuC::speC, respectively. Plasmid pNBspec is a pAV35 derivate containing the spectomycin resistance gene of pZW2 (Zhou et al., 2012). pNBspec was constructed by amplifying pAV35 with the primers RBSCATrev and CATstop and the spectomycin resistance gene of pZW2 with the primers RBSspec and Specstop. In a second PCR these two PCR fragments were connected and after self-ligating of the PCR product pNBspec was obtained. To disrupt the dcuC2 gene, plasmid pJETdcuC2 was digested with XmaJI and ligated to an XbaI fragment containing the pAV35 chloramphenicol resistance gene, resulting in the pJETdcuC2::Cm knock-out construct. To mutate the dcuA, dcuB, dcuC, dcuC2, and dctA genes, the knock-out constructs pJETdcuA::Cm, pJETdcuB::km, pJETdcuC::speC, pjetDcuC2::Cm, and pJETdctA::Cm were introduced by natural transformation in C. jejuni 81116. Homologous recombinations resulting in double cross-over events were verified by PCR.

RNA was extracted from C. jejuni wild-type grown in HI with or without 25 mM serine, aspartate, fumarate or succinate under 10 or 0.3% oxygen concentration at logarithmic (10 h) or stationary phase (20 h). RNA was also extracted from the wild-type, the racR mutant and the racR complemented strain grown in HI with 50 mM of NaNO₃ under 0.3% oxygen concentration until late logarithmic (log) phase (16 h) using the RNA-Bee™ kit (Tel-Test, Inc.). RNA samples were treated with RNAse-free DNAse I (Fermentas) according to the manufacturer's manual.

Real-time RT-PCR analysis was performed as previously described (Wösten et al., 2004). Primers used in this assay are listed in Table 2. The calculated threshold cycle (Ct) for each gene amplification was normalized to the Ct of the gyrA gene amplified from the corresponding sample before calculating fold change using the arithmetic formula 2^−ΔΔCt (Schmittgen, 2001). Each sample was examined in four replicates and was repeated with at least two independent preparations of RNA. Standard deviations were calculated and are displayed as error bars.

Recombinant His-tag labeled RacR was isolated and EMSA were performed as described before (van der Stel et al., 2015). The promoter regions upstream of the genes aspA, dcuB, dcuC, dcuC2, and dctA were amplified by PCR using the primer pairs listed in Table 2 and C. jejuni 81116 genomic DNA as template. Radioactive labeled PCR products, ~25 pmol, were incubated with 0 or 50 pmol of recombinant RacR and 25 pmol RacScyto (van der Stel et al., 2015) for 20 min at RT in binding buffer containing 20 mM Tris, pH 7.4, 5 mM MgCl₂, 50 mM KCl, 2 mM ATP, 50 μg/ml bovine serum albumin, 10 μg/ml poly-(dI-dC), and 10% glycerol. For competition assays RacR was pre-incubated for 15 min with 10 times excess of unlabelled PCR fragment. Samples were run on 6% non-denaturing Tris-glycine polyacrylamide gels at 4°C. After electrophoresis, gels were dried and autoradiographed.

Growth curves were generated under microaerobic conditions (10% O₂, 10% CO₂, 70% N₂, 10% H₂) or under oxygen-limited conditions (0.3% O₂, 10% CO₂, 79% N₂, 10% H₂) at 42°C in a “honeycomb” 10 × 10 well micro-plate using a Bioscreen C MRB (Oy Growth Curves Ab Ltd) computer-controlled incubator placed in the anaerobic chamber (Coy labs, Michigan, United States). Campylobacter precultures grown for 6 h in HI were diluted to OD_600nm of 0.01 in fresh HI media containing 25 mM fumarate, 25 mM aspartate, 25 mM serine, or 25 mM succinate with or without 25 mM nitrate. The optical density at 600 nm was measured every 15 min. The experiments were repeated at least three times in duplicate.

During the growth experiments of C. jejuni at 10% O₂, culture samples (50 μl) were taken at 4, 8, 16, and 24 h, under oxygen-limited conditions (0.3% O₂) samples were taken at 6, 12, 24, and 36 h. The culture samples were centrifuged at 14,000 rpm for 5 min. The supernatants were diluted in milli-Q water and adjusted to pH 2.4 with formic acid and injected on a Synergi 4u Fusion-RP (150 × 2.0 mm, particle size of 4 μm) analytical column (Phenomenex, Utrecht, NL). Elution was performed isocratically with milli-Q (adjusted to pH 2.4 with formic-acid): acetonitrile (9:1 [v/v]) at a flow rate of 0.3 ml/min, and the column effluent was introduced by an atmospheric pressure chemical ionization (APCI) interface, in negative mode with an ionization current of −1 μA and a source temperature of 350°C, into a 2000 QTRAP mass spectrometer (Sciex, Toronto, ON). For maximal sensitivity and for linearity of the response, the mass spectrometer was operated in multiple-reaction monitoring (MRM) mode at unit mass resolution. Peaks were identified by comparison of retention time and mass spectrum with authentic standards. Ion transitions monitored were m/z 115.0/71.0 (fumarate), 117.0/73.0 (succinate), 132.1/88.0 (aspartic acid), and 104.0/74.0 (serine) at collision energies of −12, −15, −20, and −18V, respectively. Simultaneously the four molecules were monitored by single-ion monitoring (SIM). Data were analyzed with Analyst software version 1.6.1 (Applied Biosystems).

Prism software (GraphPad, San Diego, CA) was used for statistical analysis. Data was expressed as mean ± SD. Results were analyzed by two tailed paired t-test, p < 0.05 was considered statistically significant.

Genome analysis revealed that C. jejuni possesses up to five putative C4-dicarboxylate transporters, DctA, DcuA, DcuB, and some strains contain one or two homologs of the DcuC C4-dicarboxylate transporter. C. jejuni 81116 possess five C4-dicarboxylate transporters, which are all located at a different locus in the genome (Figure 1). The genes dcuB, dcuC, and dctA are not co-transcribed with other genes located in a single-gene operon, while dcuA is located in one operon together with the aspA gene and the dcuC2 gene is located in one operon with C8J_1306 and the metC genes.

In a number of bacterial species, the C4-dicarboxylate transporters are regulated by oxygen, the available C4-dicarbolytes and/or by growth phase. To investigate whether this also applies for the C4-dicarboxylate transporters in C. jejuni we performed real-time RT PCR. We first used mRNA isolated from the wild-type 81116 strain grown at 10 or 0.3% oxygen (Figure 2A). Only the transcripts of the dcuA and dcuB genes showed a minor increase of three- to four-fold under oxygen limited conditions. Addition of serine or the C4-carbon sources, aspartate, succinate or fumarate to the culture medium of the wild-type grown until the logarithmic or stationary phase (data not shown) at 0.3 or 10% oxygen (data not shown) did not influence the transcription of the C4-dicarboxylate transporters (Figure 2B). Finally, we tested whether the growth phase (by comparing logarithmic vs. stationary phase) influences the transcription of the C4-dicarboxylate genes (Figure 2C). Only the transcription of the dcuA gene was regulated by growth phase as a 15-fold higher transcript level was observed at the logarithmic phase compared to the stationary phase. These results indicate that the regulation of C4-dicarboxylate transporters in other bacteria cannot be extrapolated to that of C. jejuni.

We previously showed that the two-component regulator RacR regulates the dcuA transporter gene. As dcuC and dcuC2 were not present on the used microarrays slides we performed real-time RT PCR using RNA isolated of the wild-type and RacR mutant grown under RacR inducing conditions (0.3% O₂ and 25 mM nitrate). Transcription of dcuC, dcuC2, and dctA genes was not affected by mutation of racR, however like the proven RacR dependent genes dcuA and aspA, a decrease of the dcuB transcription was observed in the racR mutant (Figure 3A).

To investigate whether RacR directly regulates the dcuB promoter we performed electrophoretic mobility shift assays (EMSA). As previously shown, phosphorylated RacR was able to bind to the ³²P-labeled aspA-dcuA promoter region, however RacR also binds to the dcuB promoter (Figure 3B). The band shift disappeared when unlabelled DNA corresponding to the dcuB promoter region was added in excess. As expected from the real-time RT PCR results, the dcuC, dcuC2, and dctA promoter regions are not recognized by the RacR protein. These results show that the dcuA as well as dcuB are directly regulated by the two-component system regulator RacR.

To phenotypically address the function of the C4-dicarboxylate transporter in C. jejuni we disrupted the dctA, dcuA, dcuB, dcuC, and dcuC2 genes by substituting large parts of the genes with an antibiotic resistance cassette. Although the transcription of C4-dicarboxylate transporter in C. jejuni is not regulated by oxygen (Figure 2A), we investigated whether oxygen has an influence of the activity of these transporters. Hereto growth curves in the presence of 10% O₂ were generated of the wild-type and the dctA, dcuA, dcuB, dcuC, and dcuC2 mutants in HI or with the addition of 25 mM serine, aspartate, succinate, or fumarate. No clear growth differences between the strains were observed when they were growing in HI or HI with 25 mM serine (Figures 4A,B). However, all strains reached a higher OD when serine was added to the medium, suggesting that the available serine or carbon is a limited compound in HI. When one of the C4-carbon sources, aspartate, succinate, or fumarate were added to the HI medium, the maximum final OD of all strains except for the dctA mutant, was also higher compared to HI alone (Figures 4C–E), indicating that carbon availability in HI is a limiting growth factor. The dctA mutant reached a similar maximum final OD in HI as in HI with additional aspartate, succinate or fumarate indicating that the dctA mutant is unable to utilize or to take up these carbon sources from the medium. Based on these results we conclude that the DctA transporter is the only C4-dicarboxylate transporter needed to allow C. jejuni to growth on C4-carbon sources at high oxygen levels.

To investigate why the dctA mutant is unable to use aspartate, fumarate, and succinate at 10% O₂, we measured concentrations of serine, aspartate, succinate, and fumarate in the culture supernatants of the wild-type and C4-dicarboxylate transporter mutants at 4, 8, 16, and 24 h (Figure 5A). Without the addition of extra C4-dicarboxylate compounds HI medium contains no detectable fumarate, 0.4 mM succinate, 1.0 mM aspartate, and 1.2 mM serine. All added carbon sources, serine, succinate, aspartate, and fumarate were completely utilized by the wild-type bacteria within 24 h growth at 10% O₂, however serine and fumarate were removed earlier from the media than succinate and aspartate. Fumarate was converted to succinate and then secreted to the medium. Once the fumarate was completely used the secreted succinate was taken up again and utilized by C. jejuni (Figure 5A) Similar results were obtained for the dcuA, dcuB, dcuC, and dcuC2 mutants (data not shown). A different result was obtained for the dctA mutant. The dctA mutant was still able to take up serine, to convert fumarate to succinate and, although reduced, to take up aspartate (Figure 5B), but was unable to take up succinate from the media and accumulated in the supernatant of media containing excess serine, aspartate, or fumarate. These results indicate that the DctA transporter of C. jejuni is involved in the uptake of aspartate and indispensable is for the uptake of succinate from the media at elevated oxygen conditions.

To investigate which of the C4-dicarboxylate transporters are active under oxygen-limited conditions, we generated growth curves of the wild-type, the dctA, dcuA, dcuB, dcuC, and dcuC2 mutants as well as a double dcuA/dcuB and a triple dcuA/dcuB/dcuC mutant in HI or HI with the addition of 25 mM serine, aspartate, succinate, or fumarate (Figures 6A–E). The maximum optical densities reached under these oxygen-limited conditions were all lower compared to the growth curves obtained at 10% O₂. No obvious growth differences were observed between the strains when they were grown in HI, HI + serine, or HI + succinate. The maximum OD of all strains was higher in HI + serine compared to HI alone, indicating that serine is utilized by C. jejuni under these conditions (Figure 6B). In contrast, succinate is not utilized under the oxygen-limited conditions as the growth curves generated in HI + succinate were similar as in HI. Growth defects were observed when fumarate or aspartate was added to the HI medium (Figures 6C,E). All single mutants grew less compared to the wild-type in HI + aspartate, especially the dcuB mutant. To investigate whether the C4-carboxylate transporter can replace each other function, we also tested double and triple mutants. A more severe growth defect in HI + aspartate compared to the dcuB mutant was observed for the dcuA/dcuB mutant but not for a dctA/dcuA (data not shown). A similar dcuA/dcuB growth defect was seen for the dcuA/dcuB/dcuC triple mutant, suggesting that dcuC is not involved in the utilization of aspartate. A small reduction in growth yield was observed for the single mutants dcuC and dcuB in HI + fumarate. The growth defect was more obvious in the triple mutant dcuA/dcuB/dcuC, suggesting that both DcuB and DcuC are involved in the utilization of fumarate.

To investigate the role of various active C4-transporters under oxygen-limited conditions we measured at time points 6, 12, 24, and 36 h the serine, aspartate, succinate, and fumarate concentration in the culture supernatants of the wild-type and C4-transporters mutants (Figure 7). Serine was used within 12 h by the wild-type and within 24 h by the dcuB mutant, the slowest grower in HI + serine (Figure 7A). When the strains were grown on aspartate they secreted succinate to the medium, except for the strains which were also mutated in the dcuB gene (Figure 7B). No succinate could be measured in the supernatant of the dcuB mutants showing that DcuB under oxygen-limited conditions transports succinate into the environment. Both the dcuA and dcuB mutants were able to utilize aspartate, however when both genes were mutated no decrease in the aspartate concentration in the medium was observed, indicating that both genes are needed to sequester aspartate from the medium. Growing the strains in excess fumarate revealed that not dcuC, but both dcuA and dcuB might be involved in the uptake of fumarate, as at 12 h the supernatant of these single mutants contain more fumarate than the supernatant of the wild-type (Figure 7C). The supernatant of all strains tested grown in excess fumarate contained succinate. However, the supernatant of the dcuB mutant strains contained less succinate, indicating that Campylobacter possesses a dcu-independent fumarate–succinate reductase.

We showed that RacR directly regulates by inhibition, the dcuA and dcuB genes under limited oxygen and high nitrate conditions. To address the role of the dcu genes under these conditions, we performed growth curves in HI under oxygen-limited conditions with nitrate and aspartate or fumarate (Figure 8). The growth rate as well as the maximum optical densities of the wild-type and the mutants were similar under these conditions, suggesting that the C4-dicarboxylate transporters are less active under these conditions.

To further address the role of the C4-dicarboxylate transporters under RacR inducing conditions, we measured the aspartate, fumarate, and succinate concentrations in the supernatant of the wild-type and the C4-dicarboxylate mutants at 6, 12, 24, and 36 h. The C4-carbon content in the supernatants of the strains growing under oxygen-limited conditions with nitrate were much more similar compared to the strains grown under oxygen-limited conditions (Figure 9). The aspartate uptake in the wild-type, dcuA and dcuB mutants under these conditions were similar, however only half of the aspartate in the supernatant of these cultures was utilized (Figure 9A). These results are in accordance with the growth curves and show that the C4-transporters are less active under these conditions. No reduction of the aspartate concentration was seen in the supernatant of the dcuAB mutant, confirming that both DcuA and DcuB are needed to take up aspartate from the medium. The amount of succinate produced by the wild-type under these conditions was nine-fold lower than under oxygen-limited conditions, similar results were seen for the dctA and dcuC2 mutants. In all other mutants no succinate could be detected in the supernatant. These results show that the activity like the transcription (Figure 3A) of aspartate–fumarate–succinate pathway under oxygen-limited condition is strongly reduced when nitrate is available. In the wild-type the utilization of fumarate in the presence of nitrate was similar as without nitrate, however the amount of secreted succinate was two-fold lower when nitrate was present (Figure 9B). Similar results were obtained for all other mutants (data not shown), indicating that Campylobacter possesses a dcu nitrate independent fumarate–succinate reductase.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.