Temperature-dependent FlgM/FliA complex formation regulates Campylobacter jejuni flagella length

Authors


*E-mail M.Wosten@uu.nl; Tel. (+31) 30 2534791; Fax (+31) 30 2540784.

Summary

Regulation of the biosynthesis of the flagellar filament in bacteria containing multiple flagellin genes is not well understood. The major food-borne pathogen Campylobacter jejuni possesses on both poles a flagellum that consists of two different flagellin subunits, FlaA and FlaB. Here we identify the protein Cj1464 as a regulator of C. jejuni flagellin biosynthesis. The protein shares characteristics of the FlgM family of anti-σ factor proteins: it represses transcription of σ28-dependent genes, forms a complex with σ factor FliA, and is secreted through the flagellar filament. However, unlike other FlgM proteins, the interaction of C. jejuni FlgM with FliA is regulated by temperature and the protein does not inhibit FliA activity during the formation of the hook-basal body complex (HBB). Instead, C. jejuni FlgM limits the length of the flagellar filament by suppressing the synthesis of both the σ28- and the σ54-dependent flagellins. The main function of the C. jejuni FlgM therefore is not to silence σ28-dependent genes until the HBB is completed, but to prevent unlimited elongation of the flagellum, which otherwise leads to reduced bacterial motility.

Introduction

Flagella are extracellular organelles located at the poles or lateral surfaces of bacteria that propel them to the most favourable environment. In addition to their role in motility, flagella are important for bacterial colonization of surfaces and in the host–bacterium interaction (Ottemann and Miller, 1997). Flagellar assembly and function are complex processes in which more than 50 proteins are involved (Macnab, 1996). Hierarchical gene regulation, intricate protein–protein interactions and controlled protein secretion finally result in the assembly of a complex multi-protein structure. The assembly process involves the secretion of individual subunits through the hollow core of the flagellum and their incorporation at the tip of the growing filament. The flagellar gene transcriptional hierarchy (Macnab, 2003) can be divided into early, middle and late genes (Soutourina and Bertin, 2003). The early genes encode the transcription activators that control the entire fla regulon. Prototypes of early gene products are FlhDC that activate the lateral flagella in enteric bacteria, and members of the NtrC family of σ54-associated transcription activators, i.e. CtrA, FleQ, FlrA or FlgR in Caulobacter crescentus, Pseudomonas, Vibrio and Helicobacter species, respectively, that regulate the polar flagellar cascade. The middle gene products include the proteins necessary for the structure and assembly of the hook-basal body complex (HBB) that spans the bacterial membranes, and, in many bacterial species, the alternative σ factor FliA (σ28) and its anti-σ factor FlgM. The late genes encode the external filament components.

In most species, transcription of the late flagellar genes is positively regulated by FliA and negatively regulated by the anti-σ factor FlgM. FliA is kept inactive by FlgM by actively dissociating FliA from the FliA/RNA polymerase holoenzyme and by directly interacting with free FliA to prevent reassociation (Karlinsey et al., 2000). As soon as the HBB structure is completed, FlgM is secreted through the HBB complex into the environment, enabling FliA to activate the late genes and thus to allow the formation of the filament. Mutants in FlgM contain longer flagella or have twice the number of flagella per cell (Kutsukake and Iino, 1994; Correa et al., 2004). It has been proposed that when the filament increases in length the export of FlgM diminishes, resulting in increased scavenging of FliA by FlgM. This causes downregulation of late flagellar gene transcription and prevents unlimited growth of the flagella (Hughes et al., 1993).

The major bacterial food-borne pathogen Campylobacter jejuni can be isolated from most warm blooded animals and from a wide variety of watery environmental sources (Rosef et al., 2001; Diergaardt et al., 2004). C. jejuni carries a single flagellum on each pole that, together with its helical shape, confer bacterial motility even in viscous environments (Ferrero and Lee, 1988). The C. jejuni flagellum is a heavily glycosylated multifunctional organelle that is also used to secrete virulence associated proteins such as FlaC, FspA1/2 and the Cia proteins (Konkel et al., 2004; Song et al., 2004; Poly et al., 2007). The assembly of C. jejuni flagella is largely regulated by the two-component system FlgS/FlgR in concert with the σ factor RpoN (Wösten et al., 2004). The FlgS/FlgR system directly regulates the transcription of middle genes, the putative anti-σ factor FlgM, and one of the duplicated head-to-tail linked flagellin genes, flaB. The expression of the major flagellin gene flaA is regulated by the alternative σ factor FliA, but also indirectly by FlgR/FlgS and RpoN possibly via control of the putative anti-σ factor FlgM (Wösten et al., 2004).

Despite the potential central role of FlgM in the regulation of C. jejuni flagellar assembly, the function of the protein has thus far remained speculative. A putative C. jejuni FlgM (Cj1464 in strain 11168) has been identified based on homology with the FlgM protein of the closely related bacterium Helicobacter pylori (Colland et al., 2001). Both H. pylori FlgM and the Campylobacter protein are very small proteins (∼7 kDa). Their N-terminus which in other bacterial species is suggested to contain the type III secretion signal is degenerated, which may indicate that secretion of Helicobacter and Campylobacter FlgM is different and/or may not be essential for its function (Rust et al., 2009). The FlgM of H. pylori is predominantly cytoplasmic and only a minor amount is released into the medium (Rust et al., 2009). Its inactivation increases flagellin expression but the number and length of the flagellar filament is not changed compared with the parent strain. The function of C. jejuni FlgM is unknown. In fact, only 15 of the 65 amino acids of C. jejuni FlgM are identical to H. pylori FlgM and the identity with other known FlgM proteins is even lower (Colland et al., 2001). The lack of a phenotype for a Cj1464 mutant further questions the function of Cj1464 as the anti-σ factor FlgM (Hendrixson and DiRita, 2003).

Here we report that the C. jejuni gene Cj1464 codes for an anti-σ factor FlgM with unique characteristics. In contrast to other FlgM proteins, the interaction between FlgM and FliA is temperature-dependent and the C. jejuni FlgM protein represses not only σ28- but also σ54-dependent promoters in a temperature-dependent fashion. Furthermore we provide evidence that the protein controls the length of the C. jejuni flagellar filament.

Results

Temperature-dependent role for the putative FlgM in C. jejuni motility

To examine the role of the putative flgM gene (Cj1464) in the regulation of the late flagellar genes, a deletion mutant (ΔflgM) was constructed in C. jejuni strain 81116. Semi-solid agar plates were inoculated with the wild-type strain and the isogenic ΔflgM mutant to address whether disruption of flgM alters C. jejuni motility. No phenotypic difference was observed between the wild-type strain and the ΔflgM mutant neither at 37°C (Fig. 1) nor at 32°C (data not shown), confirming the data of Hendrixson and DiRita (2003). However, when the ΔflgM mutant was grown at 42°C, a strong decrease of swarming was observed compared with the wild-type strain, suggesting that the function of C. jejuni FlgM is temperature-dependent.

Figure 1.

Influence of temperature on the swarming behaviour of the C. jejuniΔflgM mutant. Wild-type strain and ΔflgM mutant were stabbed into semi-solid thioglycolate medium and incubated for 24 h at either 37°C or 42°C.

Effect of inactivation of C. jejuni FlgM on flagellar length

To investigate why the ΔflgM mutant is less motile at 42°C, we examined the flagellum of the ΔflgM mutant and wild-type bacteria grown at 37°C or 42°C by electron microscopy (Fig. 2A). No obvious cell shape differences were noted. The average filament length of ΔflgM mutant bacteria grown at 42°C, however, was 45% longer than that of the wild-type bacteria (Fig. 2B). At 37°C (Fig. 2B) or 32°C (not shown), no significant differences in length could be detected. These results suggest that FlgM limits the elongation of the flagellum at 42°C.

Figure 2.

The ΔflgM mutant possesses longer flagella compared with the wild-type strain at 42°C.
A. Electron microscopy of C. jejuni wild type and ΔflgM mutant grown in HI medium at either 37°C or 42°C. Bars represent 2 µm.
B. Mean length and standard deviation of the flagella of the wild-type strain and ΔflgM mutant grown at 37°C or 42°C, as measured by electron microscopy.

The putative C. jejuni FlgM belongs to the FlgM family of proteins

In many bacterial species, disruption of flgM results in increased transcription of σ28-regulated genes due to the loss of inhibition of FliA activity (Kutsukake et al., 1994; Correa et al., 2004). To investigate whether the putative FlgM protein of C. jejuni, which has very limited amino acid identity with other FlgM proteins, has a similar function, σ28-dependent flaA transcript levels at different growth temperatures were determined. Real-time RT-PCR analysis showed equal levels of flaA transcript for the wild-type and ΔflgM mutant strains grown at 32°C or 37°C, but eightfold higher flaA RNA levels for the ΔflgM mutant grown at 42°C (Fig. 3A). Similar results were obtained for two other σ28-dependent genes, namely pseA and Cj0977, and when RNA isolated at different growth phases was used (data not shown). These results indicate that the putative FlgM regulates σ28-regulated promoters but only at 42°C.

Figure 3.

The gene Cj1464 is coding for an anti-σ factor FlgM.
A. Real-time RT-PCR data showing the transcript fold difference of the flaA mRNA in wild-type C. jejuni compared with that of the ΔflgM mutant grown at 32°C, 37°C or 42°C. Fold change relative to the transcription levels was calculated by the formula: 2–[(Ct flaA − Ct gyrAflgM, temp x − (Ct flaA − Ct gyrA)wt, temp x] where x = 32°C, 37°C or 42°C.
B. Interaction of C. jejuni FlgM with FliA. E. coli cells expressing FlgM–His or FliA-FLAG were disrupted by sonication. Supernatant of sonicated samples were loaded onto a column in order to purify the proteins (lanes 1, 2, 3) or first mixed and then loaded onto a column (lanes 4 and 5). W represents the last wash step before elution, E represents bead eluate. In the upper panel, samples were separated on a silver-stained polyacrylamide gel. The lower panels show Western immunoblots of the same samples probed with anti-FLAG or anti-His antibodies. Masses of molecular size markers are noted on the right in kilodaltons.
C. Western blot containing 100 times concentrated secreted proteins of the ΔflgM mutant or wild-type Campylobacter probed with polyclonal FlgM antibodies. FlgM-His protein was used as reference sample.

In other bacteria, FlgM is reported to bind to FliA as long as the HBB is incomplete. To prove that the C. jejuni FliA and FlgM form a complex, affinity purification was performed. Hereto recombinant FLAG-tagged C. jejuni FliA (29.4 kDa) and histidine-tagged FlgM (7.9 kDa) were overproduced in separate Escherichia coli strains. Both E. coli strains were then sonicated and their supernatants mixed for 30 min at room temperature. Finally the recombinant proteins were subjected to Ni-NTA agarose affinity chromatography (Fig. 3B). The FliA-FLAG protein, unable to bind to the Ni-column (lane 1), co-purified with the FlgM-His protein (lane 5). Conversely, FlgM-His and FliA-FLAG co-purified when the Ezview Red ANTI-FLAG M2 Affinity Gel was used instead of Ni-NTA agarose (data not shown). Protein staining revealed no other major co-purifying components. These results show that the C. jejuni FlgM and FliA proteins are able to form a complex.

Salmonella FlgM protein is secreted into the medium as soon as the formation of the HBB complex is completed (Chevance and Hughes, 2008). To verify that the C. jejuni FlgM is also secreted in the environment, the supernatant and bacterial pellet of the wild-type strain and ΔflgM mutant grown at 32°C, 37°C or 42°C were analysed by immunoblotting using FlgM-specific antibodies. FlgM was only seen in 100-fold concentrated supernatant and could not be detected intracellulary of cultures grown at 42°C (Fig. 3C). The intra- and extracellular concentrations of FlgM of wild-type bacteria grown at 32°C or 37°C were also below the level of detection with the methods employed (data not shown). However, the secretion at 42°C further supports classification of C. jejuni FlgM as a member of the FlgM protein family.

Transcription of flgM regulated by a σ28- and σ54-dependent promoter is independent of temperature

The transcription of the C. jejuni flgM gene (Cj1464) is regulated by two promoters: a σ54-dependent promoter in front of flgI (Cj1462) and a σ28-dependent promoter located directly upstream of the flgM gene (Fig. 4A) (Wösten et al., 2004). To investigate whether both promoters also contribute to the flgM transcript when C. jejuni is grown at 42°C we performed real-time RT-PCR on the flgM as well on the flgJ gene (Cj1463) (Fig. 4B). Disruption of fliA (encoding σ28) resulted in only a 10-fold reduction of flgM mRNA and even enhanced eightfold the flgJ transcript levels. Disruption of rpoN (encoding σ54) abolished the transcription of the flgJ gene (as expected) but had virtually no effect on the transcription of flgM. Disruption of both fliA and rpoN abolished the transcription of both flgJ and flgM. These results confirm that C. jejuni flgM gene transcription is regulated by a σ28- and σ54-dependent promoter and, importantly, demonstrate that disruption of either the promoters upstream of flgM causes an increase in activity of the other promoter.

Figure 4.

Transcription of flgM is dependent on a functional FliA and RpoN, but independent of temperature.
A. Schematic drawing of the flgM chromosomal region. Arrows indicate the position of σ28- and σ54-dependent promoters.
B. Transcript fold difference of flgM and flgJ mRNA in wild-type C. jejuni compared with that of a ΔfliA, ΔrpoN or ΔfliArpoN mutants grown at 42°C as measured by real-time RT-PCR.
C. Influence of temperature 42°C versus 37°C on the transcription of flgM and flgJ as determined by real-time RT-PCR. Fold change relative to the transcription levels was calculated by the formula: 2−[(Ct x − Ct gyrA)wt, 42°C − (Ct − Ct gyrA)wt, 37°C] where x = flgJ or flgM. Data represent the mean values and standard deviation of four independent experiments with two independent preparations of RNA.

To investigate whether the increase of FlgM secretion observed at 42°C (Fig. 3C) is due to enhanced flgM transcription, real-time RT-PCR on total RNA isolated from the wild-type bacteria grown at 42°C or 37°C was performed (Fig. 4C). Neither the σ54-dependent promoter located in front of flgI nor the σ28-regulated flgM promoter is clearly regulated by temperature. Bacteria grown at 42°C contained only twofold more flgM mRNA than when grown at 37°C, indicating that the increased FlgM secretion is likely due to a post-transcriptional event.

C. jejuni FlgM barely suppresses flagellin production before the HBB is finished

In Salmonella intracellular FlgM keeps FliA (i.e. late class σ28 promoters) in an inactive state for as long as the HBB complex is being formed (Hughes et al., 1993). During this HBB complex formation, transcription of the flgM gene occurs from a middle class σ70 regulated promoter, while after HBB completion a late class σ28 promoter is mainly responsible for the flgM transcripts. To assess whether both C. jejuni flgM promoters and other σ28- and/or σ54-regulated promoters are dependent on intracellular FlgM levels, we investigated whether disruption of the gene encoding the flagellar hook protein (flgE2), which in other bacteria prevents secretion of FlgM, or mutation of the flagellin genes (flaA and flaB), which may result in high FlgM secretion, influenced the activity of these promoters. Hereto real-time RT-PCR on flgJ and flgM transcripts was performed using total RNA isolated from C. jejuni wild type, the ΔflaAB and ΔflgE2 mutant strains, all grown at 42°C (Fig. 5A). This demonstrated strong repression (30-fold) of the σ54-dependent flgM promoter in the ΔflaAB mutant, while the σ28-dependent flgM promoter was strongly enhanced as can be concluded from the unaffected total flgM transcript level. The opposite effect was observed for the ΔflgE2 hook mutant. This mutant showed approximately eightfold increased σ54-dependent flgM and flaB transcript levels and an eightfold downregulation of σ28-dependent flaA mRNA, while the overall transcription of the flgM gene was hardly affected (Fig. 5A). Similar results were obtained for other σ28 (pseA, Cj0977) and σ54 (flgC, fliK) regulated genes and with RNA isolated at different growth phases (data not shown). These data indicate that the σ28-dependent flgM promoter resembles a late class promoter but that, in contrast to other members of the FlgM family, C. jejuni FlgM is barely able to prevent FliA from activating σ28-dependent promoters during the period that HBB formation is still incomplete.

Figure 5.

FlgM barely prevents production of flagellins when the HBB complex is not intact.
A. Influence of disruption of the structural genes ΔflgE2 and ΔflaAB on flgJ, flgM, flaB and flaA mRNA levels as estimated by real-time RT-PCR. Data shown are the transcript fold difference of the indicated genes in the mutant relative to wild-type strain normalized to gyrA transcript. Data represent the mean values and standard deviation of four independent experiments with two independent preparations of RNA. Unless otherwise indicated fold change relative to the transcription levels was calculated by the formula: 2–[(Ct target − Ct gyrA)mutant − (Ct target − Ct gyrA)wild type].
B. Western blots of pellet fraction of wild-type, ΔflaA, ΔflaB, ΔflaAB or ΔflgE2 mutants strains grown for 16 h in HI at 42°C probed with the monoclonal antibodies CF17 (anti-FlaA/FlaB) or major outer membrane protein (Nuijten et al., 1989). Molecular weight of molecular size markers are noted on the left in kilodaltons.

To investigate whether the stability of the flgM mRNA is influenced by growth temperature or mutations in structural flagellar genes, degradation of the mRNA was followed by real-time RT-PCR in rifampicin-treated C. jejuni wild-type, ΔflaAB and ΔflgE2 mutant bacteria grown at 32°C, 37°C or 42°C. No significant difference in flgM mRNA decay was observed between the wild-type strain and flagellar mutants or at different growth temperatures (data not shown). The rate of flgM mRNA degradation was similar to that of flgJ, flaA, corA or gyrA transcripts suggesting that RNases do not play an important regulatory role in controlling the amount of FlgM.

To verify that flagellin biosynthesis takes place before the HBB is completed equal bacterial fractions of the wild-type strain, ΔflaA, ΔflaB, ΔflaAB and the ΔflgE2 mutants grown at 42°C were analysed by Western blotting with a monoclonal (CF17) Flagellin A and B antibody (Fig. 5B). No flagellins in the cellular fraction of the ΔflaA mutant were detected, underlining that FlaB is only a minor flagellin at this growth temperature. Importantly, the amount of flagellins in the cellular fraction of the ΔflgE2 mutant was almost similar to that of the wild type. This indicates that C. jejuni FlgM does not suppress flagellin production despite that the formation of the HBB complex is still incomplete.

The interaction between FlgM and FliA is temperature-dependent

One explanation for the transcription-independent increase in FlgM secretion at 42°C may be an increase in translation of flgM. Unfortunately, intracellular FlgM levels were below the antibody detection limit to test this hypothesis. An alternative explanation is that the interaction of FlgM with FliA is temperature-sensitive resulting in more FlgM available for secretion at 42°C. To test this hypothesis, we again performed the FlgM-His/FliA-FLAG affinity purification assay described above but this time with the whole procedure (binding, washing and elution) carried out either at 32°C, 37°C or 42°C (Fig. 6). Control experiments showed that this procedure did not cause degradation of the FlgM-His and FliA-FLAG proteins (lanes 2, 3 and 4), that all FlgM-His protein was captured by the Ni-NTA agarose (lanes 5, 6 and 7) while the majority of the FliA-FLAG proteins did not bind to the Ni-NTA column, that neither FliA-FLAG nor FlgM-His was present in the final wash step (lanes 8, 9 and 10), and that the elution of FlgM from the column was independent of the temperature. Remarkably however, when the mixture of FlgM-his and FliA-FLAG was applied to the column, hardly any FliA-FLAG eluted from the column at 42°C, while at 37°C and especially at 32°C FliA-FLAG co-eluted with FlgM-His (lanes 11, 12 and 13). These results demonstrate that the interaction between FlgM and FliA is temperature-dependent and maximal at lower temperatures.

Figure 6.

Interaction between FlgM and FliA is temperature-dependent. E. coli expressing FlgM-His or FliA-FLAG were disrupted by sonication. Supernatants were collected by centrifugation, divided into three equal samples and adjusted to the temperature of 32°C, 37°C or 42°C. The FliA-FLAG and FlgM-His containing samples of equal temperature were gently mixed at 32°C, 37°C or 42°C for 150 min. After 30 min 50% Ni-NTA agarose was added to the solutions. After 150 min the solutions were loaded onto a column, which was washed and eluted with solutions of desired temperature. Protein samples for Western blotting were made of the start solutions (lane 1), after mixing and subsequent 150 min incubation at the three different temperatures (lane 2, 3 and 4), of the first flow through of the three mixed samples (lane 5, 6 and 7), of the last wash step (lane 8, 9 and 10), and of the eluted material (lane 11, 12 and 13). The upper and lower panels represent Western immunoblots of the protein samples probed with anti-FLAG or anti-His antibodies respectively.

Effect of temperature and the type of flagellin on the length of C. jejuni flagella

The reduced FlgM–FliA interaction at 42°C led us to hypothesize that the expression of σ28-regulated gene products such as the major flagellin FlaA might be temperature-dependent. To test this hypothesis, equal amounts of total proteins isolated from C. jejuni wild type and ΔflaA or ΔflaB mutants grown at 32°C, 37°C or 42°C were analysed by Western blotting using FlaA-specific (CF1) or FlaA/FlaB-specific (CF17) monoclonal antibodies. As shown in Fig. 7A, raising the temperature caused a slight increase in the total amount of flagellin in the wild-type strain. Separate analysis of FlaA and FlaB expression showed, as anticipated from the reduced FlgM–FliA interaction, increased FlaA protein levels at higher temperature for both the wild-type and the ΔflaB mutant strain. Unexpectedly, FlaB protein levels were also affected by temperature and increased at lower temperatures.

Figure 7.

Temperature regulates the expression and transcription of C. jejuni flagellins and influences the length of the flagellum.
A. Western blots of equal amounts of bacterial lysates of Campylobacter wild type and the mutants ΔflaA (R1) and ΔflaB (R3), probed with the monoclonal antibodies CF1 (anti-FlaA), CF17 (anti-FlaA/FlaB) or major outer membrane protein (MOMP) (Nuijten et al., 1989).
B. Transcript fold difference in flaA and flaB mRNA in C. jejuni wild type grown at 32°C, 37°C or 42°C, as determined by real-time RT-PCR. Fold change relative to the transcription levels at 42°C was calculated by the formula: 2–[(Ct target − Ct gyrA)temp − (Ct target − Ct gyrA)temp 42°C] where target = flaA or flaB and x = 32°C, 37°C or 42°C. Data represent the mean values and standard deviation of four independent experiments with two independent preparations of RNA.
C. Average length ± SD of the flagella of at least 30 wild-type and ΔflaA (R1) and ΔflaB (R3) defective bacteria grown at 32°C, 37°C and 42°C, as measured by electron microscopy.

To verify whether the temperature-sensitive FlaA and FlaB levels reflected differences in mRNA transcript, real-time RT-PCR on total RNA isolated from wild-type Campylobacter grown at either 32°C, 37°C or 42°C was performed. Relative flaA mRNA levels were 10 times lower at 32°C than at 42°C, while the opposite was true for the flaB mRNA levels (Fig. 7B). These results show that the C. jejuni flagellin genes are differentially regulated in a temperature-dependent fashion.

Since both FlaA and FlaB flagellin can be assembled into functional flagella (Wassenaar et al., 1991), we analysed the flagella of bacteria grown at 32°C, 37°C and 42°C at the structural level. Electron microscopy of the wild-type strain and the isogenic ΔflaA or ΔflaB mutants showed that flagella were synthesized by all strains but that the number of bacteria that carried flagella as well as the length of the flagella varied with the growth temperature. The average flagellum length of the wild-type strain or ΔflaB mutant bacteria decreased by 20% and 50%, respectively, when bacteria were grown at 32°C instead of 42°C (Fig. 7C). Under the same conditions, the length of the flagella of a ΔflaA mutant increased fivefold. While only 40% of the poles of the ΔflaA mutant carried a short flagellum at 42°C, at 32°C this mutant carried on both poles flagella half the length of the wild-type flagella. Thus, the length of the flagella of the parent and mutant strains as estimated by electron microscopy (Fig. 7C) is a direct reflection of the amounts of FlaA and FlaB detected by Western blotting (Fig. 7A) and the relative amounts of flaA and flaB mRNA (Fig. 7B). Taken together, these results show that the length of the flagellum and the FlaA/FlaB ratio in the filament are temperature-dependent.

FlgM affects both σ28- and σ54-dependent promoter activity

The apparent ineffective inhibition of FliA by FlgM at 42°C (resulting in increased levels of FlaA) together with the unexpected opposite temperature-sensitive transcription of the σ54-regulated flaB gene led us to test the effect of FlgM on flaB transcription. To our surprise, disruption of flgM enhanced the transcription of flaB, especially when the mutant was grown at 42°C (Fig. 8A). Disruption of flgM also increased the transcripts of other σ54-dependent genes like flgJ and flgE2, but not of the σ70-regulated flagellar-independent corA gene. Complementation of the ΔflgM mutant with a plasmid containing the flgM gene regulated by its own σ28 promoter yielded flgM transcripts 3.5-fold lower than wild-type levels (Fig. 8B) and restored flaA, flaB, flgJ and flgE2 transcripts to ∼50% of wild-type levels. The partial complementation may indicate that both flgM promoters are needed for full FlgM function. These data show that C. jejuni FlgM negatively regulates σ28- as well as σ54-dependent promoters at the optimum C. jejuni growth temperature of 42°C. Since the syntheses of both flagellins are positively affected by disruption of flgM, FlgM is a key protein in determining the length of the filament at 42°C.

Figure 8.

Mutation in flgM affects both σ28- and σ54-dependent promoters at 42°C.
A. Transcription of the σ28-dependent gene flaA, the σ54-dependent genes flaB, flgJ, flgE2 and the σ70-dependent gene corA was determined by real-time RT-PCR using total RNA isolated of the wild type, ΔflgM mutant and complemented ΔflgM mutant (ΔflgM+ pMA2flgM). Shown are fold transcript differences of the indicated genes in the mutant or complemented strain relative to wild-type strain normalized to gyrA transcript.
B. The same strains as mentioned in (A) were used to measure flgM transcripts by real-time RT-PCR. Fold change relative to the transcription levels at 42°C was calculated by the formula: 2−[(Ct target − Ct gyrA)mutant − (Ct target − Ct gyrA)wild type]. Data represent the mean values and standard deviation of four independent experiments with two independent preparations of RNA.

Secretion of FlgM diminishes as the filament elongates but FlgM does not obstruct the flagellar tube

Next we tried to investigate how FlgM is able to determine the length of the filament. As the filament increases in length, it has been proposed that the export of FlgM becomes increasingly difficult, which may result in increased scavenging of FliA by FlgM, which in turn reduces late flagellar gene transcription (Hughes et al., 1993). To confirm the apparent relationship between secretion of FlgM and flagella length, equal amounts of culture supernatant of wild-type bacteria, ΔflaA, ΔflaB, ΔflaAB, ΔflgM and the ΔflgE2 mutant that differ in length of their flagella (Fig. 7C) were analysed by Western blotting with polyclonal FlgM antibodies. As illustrated in Fig. 9A, low FlgM secretion occurred for the wild-type and ΔflaB mutant strain that have both long flagella, while high levels of FlgM were seen for the ΔflaA and ΔflaAB mutant that express very short and no flagella respectively. However, while maximum secretion was observed for the ΔflaB mutant, the ΔflgE2 mutant which lacks flagella and the hook protein showed no detectable secretion of FlgM (Fig. 9A). These results indicate that the secretion of FlgM is dependent on the length of the filament and show that the flagellar hook protein must be present for FlgM protein secretion to occur. We observed that the molecular mass of secreted FlgM is different and depends on the type of flagellin that is mutated. In all strains, FlgM (7.1 kDa) migrated at equal or (in the case flagellin B was disrupted) at higher molecular mass than the recombinant His-tagged purified FlgM (7.9 kDa) isolated from E. coli, suggesting that C. jejuni FlgM may be subject to post-translational modification.

Figure 9.

Secretion of FlgM is dependent on the length of the filament, but FlgM does not obstruct the flagellar tube.
A. Western immunoblot of 100 times concentrated supernatants samples of wild type, ΔflaA, ΔflaB, ΔflaAB, ΔflgM or ΔflgE2 mutants grown for 16 h in HI at 42°C probed with polyclonal FlgM antibodies. FlgM-his protein (lane 1) was used as reference sample.
B. Western blot of supernatants of the strains mentioned in (A) probed with anti-FlaC antibodies. FlaC-his protein (lane 1) was used as reference sample. The anti-FlaC antibodies cross-react with the flagellins A and B. Molecular weight of molecular size markers are noted on the left in kilodaltons.

To investigate whether FlgM obstructs the flagellar tube once the flagellum is completed, the secretion of FlaC, one of many proteins secreted through the flagellum, was investigated (Konkel et al., 2004). Equal amounts of supernatants of the above-mentioned strains were probed with a polyclonal antibody that recognizes FlaC but also cross-reacts with the flagellin subunits FlaA and FlaB (A.K.J. Veenendaal et al., unpubl. results). Secretion of the flagellins occurred in the wild-type strain and in the ΔflaB mutant, but was much lower compared with the secretion in the ΔflgM mutant (Fig. 9B). This means that only a part of the produced flagellins are incorporated in the flagellum of the ΔflgM mutant and that the remainder is secreted. FlaC is only secreted when C. jejuni has a functional hook and one of the flagellins (Fig. 9B). Similar amounts of secreted FlaC were observed in the wild-type strain, FlgM, and flagellin mutants showing that secretion of FlaC is independent of FlgM or the length of the flagellum.

Discussion

The function of Cj1464 as C. jejuni anti-σ factor FlgM has long been doubted as disruption of the gene did not yield a clear phenotype (Hendrixson and DiRita, 2003). We also could not show a function for FlgM when Campylobacter was incubated at 37°C, the same incubation temperature used by Hendrixson and DiRita (2003). However, at 42°C, the optimum growth temperature of C. jejuni, disruption of FlgM resulted in increased flaA transcription, longer flagella and reduced motility, while wild-type bacteria mainly secreted FlgM in the environment. The latter phenotypes are not observed in the closely related bacterium H. pylori, but have been reported for Vibrio cholerae (Correa et al., 2004). The data also imply, as has been predicted by Christensen et al. (2009), that, although the N-terminus of C. jejuni FlgM is degenerated compared with other FlgM proteins, C. jejuni FlgM still possesses the type III secretion signal.

Although the transcription of flgM is not affected by temperature (Fig. 4C), we only could detect FlgM in the supernatant of bacteria grown at their optimum growth temperature 42°C. To understand the underlying mechanism for this observation, we investigated whether temperature influences the FliA–FlgM complex formation. FlgM is less associated to FliA at higher temperature which may explain the increased FlgM secretion observed at 42°C. As a result of the temperature-dependent interaction, more FliA is available to activate σ28-dependent promoters like the flaA promoter.

Our flagellin transcription, expression and filament length data (Fig. 7) show that FlaA biosynthesis predominates at high temperatures and that FlaB expression is more pronounced at lower temperatures. Based on our findings that the amount of cytoplasmic FliA is independent of the temperature (data not shown), that the effect of temperature on the transcription of flaA is particularly apparent in the ΔflgM mutant (Fig. 3A), and that FliA-FLAG binds less well to FlgM-His at 42°C (Fig. 6), we believe that the conformation of FliA is temperature-dependent, resulting in more active FliA at the optimum growth temperature of 42°C. A similar temperature-dependent FliA activity may exist in other bacteria as FliA of E. coli binds better to σ28 promoter regions at elevated temperatures (Givens et al., 2001). Both the lack of FliA inhibition in a ΔflgM mutant and the increased FliA activity at 42°C therefore likely contribute to the formation of longer flagella as are present in the C. jejuniΔflgM mutant.

To investigate whether the σ54-dependent transcription of flagellin B was not repressed by FlgM as in V. cholerae (Correa et al., 2004), the transcription of the flaB gene was investigated by real-time RT-PCR. Surprisingly, disruption of flgM had more effect on the transcription of the flaB than of the flaA gene. Complementation studies confirmed that the absence of flgM caused the increased flagellin transcription. The transcription of other σ54-regulated genes like flgJ and flgE2 was also more eminent in the ΔflgM mutant. So far, only for H. pylori with a flhA background it has been shown that FlgM can repress middle class σ54-regulated flagellar genes (Niehus et al., 2004). The σ54-regulated C. jejuni promoters are also activated in a ΔfliA mutant (Carrillo et al., 2004), which we could confirm for the σ54-dependent flgJ promoter (Fig. 4B). Transcription of the main C. jejuni flagellar regulators FlgS, FlgR, RpoN or FliA is not influenced by a disruption in fliA or flgM (Carrillo et al., 2004; Wösten et al., 2004) (data not shown). Based on our results it can be speculated that the accumulation of the FliA/FlgM complex may inhibit the phosphorylation status of the FlgS/FlgR system, which has been shown to regulate the transcription of σ54-dependent genes (Hendrixson and DiRita, 2003; Wösten et al., 2004). When optimal filament length is reached, FlgM is no longer exported and will complex FliA (Hughes et al., 1993) (Fig. 9A) which may be a direct or indirect signal for the C. jejuni FlgS sensor that the synthesis of the flagellum is completed and that the transcription of the σ54-regulated genes encoding for the HBB structure can be shut down.

To investigate whether the length of the filament influences the secretion of FlgM, which might cause accumulation of the FliA/FlgM complex, the secretion of FlgM in several structural flagellar mutants was investigated. Although a strong signal on a Western blot was observed when less than 1 pmol of purified His-tagged FlgM was detected with the polyclonal FlgM antibodies, we were unable to detect intracellular FlgM in neither the wild-type strain nor the structural mutants (data not shown). This indicates either that the amount of intracellular FlgM is very low, or that the FlgM protein might be less recognized by the antibody due to post-translational modification. The latter hypothesis is strengthened by the observation that the molecular mass of the secreted FlgM is larger than expected and depends on the type of flagellin that was disrupted. However, in culture supernatants FlgM was detected and the amounts inversely correlated with the length of the flagella (Fig. 9). The secretion of C. jejuni FlgM requires an intact HBB complex as evidenced by the lack of FlgM secretion in the FlgE2 mutant, as noted for other species (Chilcott and Hughes, 2000). Overall, our results confirm the hypothesis that during the process of filament elongation the export of FlgM becomes increasingly difficult (Hughes et al., 1993).

The main function of FlgM in other species is to silence the late flagellar genes until the HBB is finished (Chevance and Hughes, 2008). We could not confirm this for C. jejuni FlgM as transcription and translation of σ28-dependent genes was barely affected in a hook or ΔrpoN mutant (also lacking a functional HBB) (Figs 4B and 5A and B). The main function of C. jejuni FlgM is therefore likely different. In H. pylori FlgM interacts with the C-terminal cytoplasmic domain of FlhA, an integral membrane component of the type III flagellar protein export apparatus (Rust et al., 2009). Although the activity of FliA is to some extent inhibited by FlgM, we believe, based on the enhanced flagellin secretion in a flgM mutant, that FlgM is involved in the secretion of flagellins by the protein export apparatus. FlgM may reduce the number of secreted flagellins by blocking the flagellar gate for flagellins (but not FlaC) either directly or indirectly. We will further investigate this model in the future.

The question remains why Campylobacter contains two differentially temperature-regulated flagellins. The significance of the flaA/B gene duplication has been the subject of several studies (Alm et al., 1993; Meinersmann and Hiett, 2000). By altering the ratio of the FlaA and FlaB in the filament, the shape of the filament may be optimally adapted to different circumstances. The temperature-dependent regulation of the flagellins demonstrates that FlaA and FlaB are probably needed under different environmental conditions. C. jejuni must be able to survive inside animals during the colonization phase as well as in surface water, and therefore may present two distinct phases in its life cycle with characteristic physiological adaptations.

In conclusion, we have shown that the C. jejuni gene Cj1464 is coding for an anti-σ factor FlgM with unique characteristics. The FlgM binds to FliA in a temperature-dependent manner and it represses σ28- as well as σ54-regulated promoters. The main function of the C. jejuni FlgM is not to silence σ28-dependent genes before the HBB is completed, but to prevent unlimited elongation of the flagellum, which otherwise leads to reduced motility.

Experimental procedures

Bacterial strains, plasmids and growth conditions

The strains and plasmids used in this study are listed in Table 1. C. jejuni 81116 (Palmer et al., 1983) and derivatives were routinely maintained at 37°C under microaerobic conditions (5% O2, 10% CO2 and 85% N2) on saponin agar medium containing 5% horse blood lysed with 0.3% saponin (Oxoid, London). E. coli strains were grown in Luria–Bertani medium at 37°C. When antibiotic selection was necessary, the growth medium was supplemented with ampicillin (100 µg ml−1), kanamycin (50 µg ml−1) or chloramphenicol (20 µg ml−1).

Table 1.  Bacterial strains and plasmids used in this study.
Strain or plasmidGenotype or relevant characteristicsSource or reference
C. jejuni strains
81116Wild typePalmer et al. (1983)
ΔfliA81116 derivative ΔfliA::KmThis study
ΔrpoN81116 derivative ΔrpoN::CmThis study
ΔflgM81116 derivative ΔflgM::KmThis study
ΔflgM + pMA2-fglM81116 derivative ΔflgM::Km
Containing complementation plasmid pMA2-fglM
This study
ΔfliAΔrpoN81116 derivative ΔfliA::Km ΔrpoN::CmThis study
ΔflaA81116 derivative ΔflaA::Km (R1)Wassenaar et al. (1991)
ΔflaB81116 derivative ΔflaB::Km (R3)Wassenaar et al. (1991)
ΔflgE281116 derivative ΔflgE2::KmThis study
E. coli strains
PC2955relA1 Φ80dlacZΔM15 phoA8 hsdR17 recA1 endA1 gyrA96 thi-1 relA1 luxS glnV44NCCB
Bl21(DE3)ompThsdS dcm+ Tetrgal(DE3) endA HteNovagen
SE 1F-, CmR, ompT, lon, hsdSB (restriction-, modification-), gal, dcm, DE3 (lacI, T7polymerase under the control of the PlacUV5 promoter), ccdBEurogentec
CYS21F-, CmR, mcrA, endA1, Δ(mrr-hsdRMS-mcrBC) (restriction-, modification-), Φ80lacZΔM15, ΔlacX74, recA1, Δ(ara, leu)7697, araD139, galU, galK, nupG, rpsl, ccdB+.Eurogentec
Plasmids
PGEM-T EasyPCR cloning vector, AmprPromega Corporation, Madison, WI
pAV35pBluescript II SK containing C. coli Cmr cassettevan Vliet et al. (1998)
pGEMΔfliA::KmpGEMThis study
pMW24pBluescript KS with Cmr inserted in rpoNWösten (1997)
pMW2pBluescript KS containing C. jejuni Kmr cassetteWösten (1997)
pGEMflgMBamHIpGEM-T Easy containing the flanking regions of flgM and an inserted BamHI site in the 128 bp truncated flgM geneThis study
pGEMΔflgM::KmpGEMflgMBamHI with 1.4 kb Kmr inserted in flgMThis study
pGEMflgE2pGEM-T Easy containing the flgE2 gene on a 2579 bp fragmentThis study
pGEMΔflgE2::KmpGEMflgE2 of which 342 bp of flgE2 gene is replaced by 1.4 kb KmrThis study
pGEMflaABpGEM-T Easy containing the flaA and flaB genes on a 3721 bp fragmentThis study
pGEMΔflaAB::CmpGEMflaAB of which a1904 bp fragment containing a large part of the flaA and flaB genes is replaced by 0.8 kb CmrThis study
pT7.7Expression vector, AmprTabor and Richardson (1985)
pT7.7FliA-FLAGpT7.7 containing the fliA gene with N-terminal FLAG-tagThis study
pMA2C. jejuni/E. coli shuttle conjugative expression vector, Cmrvan Mourik et al. (2008)
pSCODON1.2Expression vector, AmprEurogentec
pSCODON1.2flgMpSCODON1.2 containing the flgM gene with C-terminal His-taqThis study
pMA2-fglMflgM complementation plasmidThis study

Primers

All primers used in this study are listed in Table S1.

Construction of C. jejuni flgM (Cj1464), fliA (Cj0061), rpoN (Cj0670), flgE2 (Cj1729c), flaAB (Cj1338 and Cj1339) and fliA/rpoN mutants

The flanking regions of flgM were amplified by PCR using the primer sets FlgK/FlgMFBamHI and FlgMRBamHI/FliI. The resulting PCR fragments (1120 bp and 1101 bp in length) were digested with BamHI and cloned together into pGEM-T Easy (Promega Corporation, Madison, WI), yielding plasmid pGEMflgMBamHI. The flgM gene in this plasmid contains a deletion of 128 bp and a unique BamHI restriction site. To obtain the knockout construct pGEMΔflgM::Km, plasmid pGEMflgMBamHI was digested with BamHI and ligated to a 1.4 kb BamHI fragment containing the kanamycin resistance gene (Kmr) of pMW2 (Wösten, 1997).

The knockout construct pGEMΔfliA::Km was constructed in the same manner as pGEMfliA::Cm (Wösten et al., 2004), except that the chloramphenicol resistance gene was replaced by the 1.4 kb BamHI Kmr of pMW2.

To inactivate both flagellin genes flaA and flaB a large 3721 bp fragment containing these genes was amplified with the primers FlaA3-TOPO and FlaAB Mutant 2-R. This PCR product was ligated into the plasmid pGEM-T Easy to form plasmid pGEMflaAB. A 1904 bp EcoRV fragment of pGEMflaAB containing a part of flaA and flaB genes was replaced by a 0.8 kb PvuII fragment of pAV35 (van Vliet et al., 1998) containing a chloramphenicol cassette (Cmr). The resulting knockout plasmid pGEMΔflaAB::Cm contains the Cmr in same orientation as the flagellin genes.

To mutate the flgE2 gene a 2579 bp PCR product amplified with the primers FlGE2F and FLGE2R was ligated into pGEM-T Easy resulting in pGEMflgE2. To delete 342 bp of the flgE2 gene and to add a unique BamHI site in the flgE2 gene inverse PCR was performed on plasmid pGEMflgE2 with the primers FLGE2RBamHI and FLGE2FBamHI. The inverse PCR product was digested with BamHI and ligated to the 1.4 kb BamHI Kmr of pMW2 resulting in the knockout plasmid pGEMΔflgE2::Km.

All PCRs were performed with the proof-reading DNA polymerase Pfu (Promega) and the nucleotide sequence of the cloned PCR products was verified by sequencing both strands.

To mutate the flgM, fliA, flaAB, flgE2 and rpoN genes plasmid pGEMΔflgM::Km, pGEMΔfliA::Km, pGEMΔflaAB::Cm, pGEMΔflgE2::Km or pMW24 [rpoN knockout construct (Wösten, 1997)], respectively, were introduced by natural transformation in C. jejuni 81116 (Wassenaar et al., 1993). Plasmid pMW24 was also transformed to the acquired 81116 ΔfliA mutant to require the double mutant 81116 ΔfliAΔrpoN. Homologous recombinations resulting in double-cross-over events were confirmed by PCR.

Construction of FlgM complementation plasmid

To complement the ΔflgM mutant, the flgM gene was amplified with the primers 1463Ftaq and Cj1465Rtaq and cloned into pGEM-T-easy. The restriction enzymes SpHI and SacI in the polylinker of pGEM-T-easy were used to clone a 617 bp flgM PCR fragment into pMA2 (van Mourik et al., 2008), generating the complementation plasmid pMA2-flgM. This plasmid was introduced into 81116 ΔflgM mutant via conjugation (Labigne-Roussel et al., 1987).

Construction of FliA- and FlgM-overexpressing plasmids

To overexpress the FliA σ factor with a FLAG-tag, the chromosomal fliA gene was amplified using the primers NdeIFLAGFliA and FliAPstI. The resulting PCR fragment of 751 bp was digested with NdeI and PstI and cloned into the NdeI and PstI sites of pT7.7 to form pT7-FliA-FLAG. To overexpress FliA-FLAG protein plasmid pT7-FliA-FLAG was transferred to E. coli BL21(DE3).

To obtain a vector overexpressing the FlgM protein with a His-tag the flgM gene was amplified with the primers FlgMF2 and FlgMR1 and cloned into pGEM-T Easy. The 205 bp flgM gene was removed from the obtained plasmid pGEMflgM with restriction enzymes NdeI and XhoI and cloned into the expression plasmid pSCODON1.2 (Eurogentec) which was digested with the same enzymes. The resulting FlgM expression plasmid pSCODON1.2flgM was finally transferred to E. coli SE1. The nucleotide sequence of the cloned PCR products was verified by sequencing of both strands. Chromosomal DNA of C. jejuni 81116 was used as PCR template.

Purification of recombinant proteins

Histidine-tagged FlgM was expressed in E. coli SE1 containing plasmid pSCODON1.2flgM. Protein expression and purification were performed as described previously (Wösten et al., 2004) FLAG-tagged FliA was isolated using the Ezview Red ANTI-FLAG M2 Affinity Gel (Sigma) according to the protocols provided by the manufacturer. Protein concentrations were determined using the Bradford method (Bradford, 1976).

Affinity purification

To co-elute the His-FlgM and FLAG-FliA proteins with nickel-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen), the proteins were induced in E. coli SE1 containing plasmid pSCODON1.2flgM and E. coli BL21(DE3) containing plasmid pT7.7-FliA-FLAG with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 3 h at 37°C. After protein induction, the cells were sonicated and centrifugated for 10 min at 10 000 g. The two supernatants were mixed and incubated for 30 min at 4°C, then 50% Ni-NTA agarose was added to the supernatant and the mixture was incubated for 16 h at 4°C. Elution of the proteins was performed as described previously (Wösten et al., 2004). Samples were separated by SDS-12% polyacrylamide gel electrophoresis, stained with silver or analysed by Western immunoblotting with either C-terminal anti-His HRP (Sigma) or anti-FLAG M2 (Sigma) monoclonal antibody. Western blots were detected with Supersignal® West Pico chemiluminescent substrate (Pierce Biotechnology). The SeeBlue® Plus2Pre-stained standard ladder (Invitrogen) was used as molecular mass standard.

Temperature-dependent affinity purification

The procedure was slightly different as described under affinity purification. Briefly, for the binding, washing and elution steps all solutions and material were preheated to 32°C, 37°C or 42°C to maintain the proteins on that temperature. After the two E. coli strains were sonicated and centrifuged for 10 min at 10 000 g. Each supernatant was divided into three equal subsamples and heated to 32°C, 37°C or 42°C. The FliA-FLAG and FlgM-His containing samples with equal temperature were mixed and gently shaken at 125 r.p.m. on a rocking platform for 150 min at 32°C, 37°C or 42°C. After 30 min 50% Ni-NTA agarose was added and incubation was continued for 2 h. Samples were loaded onto a column, washed, and eluted with the solutions described above of the proper temperature. Protein samples were made (i) before mixing, (ii) after mixing and subsequent 150 min incubation at the three different temperatures, (iii) of the first flow through of the three mixed samples, (iv) of the last wash step and (v) elution step.

Preparation of polyclonal FlgM and FlaC antibodies

Polyclonal antibodies against His-tagged FlgM, FLAG-tagged FliA and His-tagged FlaC (A.K.J. Veenendaal et al., in preparation) were generated in rabbits by Eurogentec (Belgium).

Detection of flagellin proteins by Western blotting

Protein samples were prepared from Campylobacter cultures grown at 32°C, 37°C or 42°C. Equal amounts of protein, as determined by Bradford assay (Bradford, 1976), were run on 12.5% SDS-polyacrylamide gels. Western blotting was performed as described previously (Nuijten et al., 1991) using polyclonal antisera raised against the major outer membrane protein of C. jejuni 81116 (MOMP) (Nuijten et al., 1989) or monoclonal antibodies specific for flagellin A (CF1) or recognizing both Campylobacter 81116 flagellins A and B (CF17) (Nuijten et al., 1991).

Swarming assay

Overnight cultures of Campylobacter were diluted to an A550 of 0.1 in 5 ml of Heart Infusion broth (HI) and incubated at 37°C, at 150 r.p.m. under microaerobic conditions. The A550 of mid-exponential phase cultures (A550 0.3–0.7) was adjusted to 0.5 of which 2.5 µl was injected into semi-solid medium (thioglycolate medium containing 0.4% agar) and incubated under microaerobic conditions at 32°C, 37°C or 42°C. Swarming was scored by measuring the diameter of the colonies.

Electron microscopy

Mid-exponential phase grown bacteria in HI at 32°C, 37°C or 42°C were incubated on 400-mesh formvar/carbon-coated grids for 10 min, stained with 2% tungstophosphoric acid for a few seconds, and examined in a Philips CM10 electron microscope. The length of both flagella of at least 30 bacteria were measured by electron microscopy.

Isolation and detection of secreted proteins

Campylobacter grown for 16 h in HI were centrifuged for 10 min at 5000 g. To 4 ml of supernatant 1.6 ml, 40% trichloroacetic acid (TCA) was added and incubated for 30 min on ice. The sample was centrifuged at 20 000 g for 5 min at 4°C. The pellet was washed with 0.5 ml of ice-cold acetone, resuspended in 40 µl of 1× Laemlli buffer and incubated for 5 min at 95°C. Samples were separated by SDS-15% polyacrylamide gel electrophoresis, stained with silver or analysed by Western immunoblotting with either rabbit polyclonal FlgM or FlaC antibodies. Western blots were developed using anti-rabbit Ig, horseradish peroxidase-linked antibodies and detected with Supersignal® West Pico chemiluminescent substrate (Pierce Biotechnology).

Real-time RT-PCR

Pre-cultures of Campylobacter were grown at 32°C, 37°C or 42°C in HI, diluted to an A550 of 0.02 in 30 ml of HI, and incubated on a gyratory shaker (150 r.p.m.) under microaerobic conditions at 32°C, 37°C or 42°C respectively. Total RNA was extracted from late-logarithmic phase cultures (A550, 0.6–0.9) or of late-logarithmic phase cultures that were incubated with 500 µg ml−1 rifampicin for 0, 5, 25 or 125 min which completely inhibits growth (Gaynor et al., 2005) to investigate the stability of mRNA. Total RNA was extracted with the RNA-Bee™ kit (Tel-Test) according to the manufacturer's specifications.

RNA samples were diluted to a concentration of exactly 1 µg µl−1 and treated with RNase-free DNase I (Invitrogen). PrimerExpress software (Applied Biosystems) was used to design primers to amplify 50–80 bp fragments. RT-PCR was performed on 0.2 µg DNase I-treated RNA with 1 µM of primers and the Sybr Green RT-PCR kit (Qiagen) using an ABI Prism 7000 (Applied Biosystems) sequence detection system. Real-time cycling conditions were 30 min at 48°C, followed by 10 min at 95°C and then 40 cycles of 95°C for 15 s, 60°C for 30 s. Specificity of the PCR was confirmed by electrophoretic analysis of the reaction products and by inclusion of template- or reverse-transcriptase-free controls.

The value used for comparison and quantification was the threshold cycle (Ct), defined as the cycle at which the fluorescence becomes detectable above background and that is inversely proportional to the logarithm of the initial number of template molecules. The calculated threshold cycle (Ct) for each gene amplification was normalized to the Ct of the temperature-independent gyrA gene (Stintzi, 2003), amplified from the corresponding sample before calculating fold change using the arithmetic formula (2−ΔΔCt), where ΔΔCt = [(Cttarget − CtgyrA)tempx − (Ct target − CtgyrA)temp 42°C] or [(Ct target − CtgyrA)mutant − (Ct target − CtgyrA)wild type] (Schmittgen, 2001), where target = flaA, flaB, flgM, flgJ, pseA, Cj0977, flgC, fliK or corA and x = 32°C, 37°C or 42°C. Each sample was examined in four replicates and was repeated with at least two independent preparations of RNA and standard deviations were calculated and displayed as error bars.

Acknowledgements

We thank Linaida Pisas for technical assistance. This work was supported by a fellowship of the Royal Netherlands Academy of Arts and Science, NWO-VIDI Grant 917.66.330 to M.M.S.M. Wösten, and EU Grant PERG02-GA-2007-224937 to A.K.J. Veenendaal.

Ancillary