The FlgS/FlgR two-component signal transduction system regulates the fla regulon in Campylobacter jejuni.

The human pathogen Campylobacter jejuni is a highly motile organism that carries a flagellum on each pole. The flagellar motility is regarded as an important trait in C. jejuni colonization of the intestinal tract, however, the knowledge of the regulation of this important colonization factor is rudimentary. We demonstrate by phosphorylation assays that the sensor FlgS and the response regulator FlgR form a two-component system that is on the top of the Campylobacter flagellum hierarchy. Phosphorylated FlgR is needed to activate RpoN-dependent genes of which the products form the hook-basal body filament complex. By real-time reverse transcriptase-PCR we identified that FlgS, FlgR, RpoN, and FliA belong to the early flagellar genes and are regulated by sigma70. FliD and the putative anti-sigma-factor FlgM are regulated by a sigma54- and sigma28-dependent promoters. Activation of the fla regulon is growth phase-dependent, a 100-fold rpoN mRNA reduction is seen in the early stationary phase compared with the early logarithmic phase. Whereas flaB transcription decreases, flaA transcription increases in early stationary phase. Our data show that the C. jejuni flagellar hierarchy largely differs from that of other bacteria. Phenotypical analysis revealed that unflagellated C. jejuni mutants grow three times faster in broth medium compared with wild-type bacteria. In vivo the C. jejuni flagella are needed to pass the gastrointestinal tract of chickens, but not to colonize the ceaca of the chicken.

Two-component systems are assumed to be one of the prime bacterial systems of environmental adaptation and gene regulation, and likely play a major role in bacterial virulence. In general, they consist of two proteins, an integral membrane protein, the sensor kinase, and a cytoplasmic protein, the response regulator (1). These proteins communicate by reversible phosphorylation in response to changing environmental conditions (2). The interaction of a distinct environmental stimulus with the sensor, usually in the periplasmic region, leads to activation of the kinase activity located in the C-terminal cytoplasmic domain of the sensor (3). As a result a conserved cytoplasmic histidine residue of the sensor becomes phosphorylated. The phosphate is transferred to a conserved aspartate residue on the N-terminal domain of the cognate response regulator (3). The response regulator has affinity to distinct DNA sequences on the genome and, depending on its phosphorylation state, activates or represses gene transcription. This ultimately leads to a bacterial phenotype that is optimally adapted to its ecological niche.
A key bacterial trait for chemotaxis, behavior, and survival is the flagellum. The more than 50 proteins that are involved in the assembly and function of the flagellum are expressed in a hierarchical order. In enteric bacteria the flagellar biosynthesis genes are divided into three classes (4). The early genes, flhC and flhD, transcribed by 70 , are the regulatory proteins that control the expression of the entire fla regulon and directly activate the so-called middle genes. These 70 transcribed middle genes encode structural components of the basal-body hook structure, the type III secretion system, as well as the alternative -factor FliA ( 28 ) and the anti--factor FlgM. FliA is kept inactive by the anti--factor FlgM until FlgM is secreted through the completed basal-body hook structure. Once FliA is set free it activates the late genes that encode the filament, motor force generators, and the chemosensory machinery.
The fla regulon in Caulobacter crescentus is activated by the response regulator, CtrA, family of the OmpR response regulators (5). CtrA activates 54 , which together with the response regulator FlhD, is responsible for activation of the middle and late fla genes. In Helicobacter pylori, Pseudomonas aeruginosa, and Vibrio species, the fla regulon is activated by homologues of the NtrC group of response regulators (i.e. FlgR, FleQ, and FlrA, respectively) that are part of a two-component signal transduction system that works in concert with the alternative factor RpoN ( 54 ) (6 -8). In these species, RpoN together with the response regulator activates the middle genes; late fla genes are regulated by 28 in a 54 -dependent manner.
Campylobacter jejuni is the leading bacterial cause of foodborne gastroenteritis in developed countries. Although C. jejuni can be found in most warm blooded animals its most favored habitat appears to be the intestine of avian species in which up to 10 9 colony forming units per gram of feces can be found (9). Ingestion of as few as 100 Campylobacters is sufficient to colonize the avian gut and it is reported that in broiler houses a total flock is colonized within a few days from initial exposure (10,11). In the intestine, C. jejuni is thought to penetrate the mucus that covers the epithelium of the small intestine and caeca. Important for this colonization event is the characteristic rapid darting corkscrew-like motility conferred by the polar flagellum and the spiral shape of C. jejuni. This feature enables C. jejuni to remain motile in mucus, the highly viscous environment that rapidly paralyzes other motile rod-shape bacteria (12,13).
Based on the genome sequence of C. jejuni strain 11168, more than 50 genes are predicted to be involved in the assem-bly of the flagella (14). These genes are located in more than 32 individual loci, whereas in Escherichia coli they are located in only six loci (15) making the regulation of the Campylobacter fla regulon even more complex. Although the phenotypes of many flagellar mutants are known, the knowledge of the regulation of the fla regulon in C. jejuni is still rudimentary (16 -18). So far RpoN, FliA, and FlgR have been identified as C. jejuni flagellar transcription activators (18,19). How these proteins are regulated and what is their role in the assembly of the C. jejuni flagellum remains to be elucidated.
In this paper, we report that FlgS and FlgR form a twocomponent signal transduction system and that this system is on the top of the Campylobacter flagellum hierarchy. In addition, we elucidated the regulation of the major fla regulon regulators and hypothesize that the assembly of the Campylobacter flagella is an energy consuming process that is necessary in vivo for passing through gastrointestinal tract of animals but not for colonizing the ceaca of chickens.

MATERIALS AND METHODS
Bacterial Strains, Plasmids, and Growth Conditions-The strains and plasmids used in this study are listed in Table I. Campylobacter 81116 (20) and derivatives were routinely maintained at 37°C under microaerobic conditions (5% O 2 , 10% CO 2 , and 85% N 2 ) on saponin agar medium containing 5% horse blood lysed with 0.3% saponin (Oxoid Ltd., London). Escherichia coli strains were grown in Luria-Bertani medium at 37°C. When antibiotic selection was necessary, the growth medium was supplemented with ampicillin (100 g/ml), kanamycin (50 g/ml), or chloramphenicol (20 g/ml).
Motility Assay-Campylobacter strains were stabbed with a pipette tip into semisolid medium (thioglycolate medium containing 0.4% agar) and incubated under microaerobic conditions at 37°C for 48 h. Motility was scored by measuring the diameter of the colonies.
Growth Curve of Flagella Mutants-Overnight cultures of Campylobacter in heart infusion (HI) broth were diluted to A 550 of 0.01 in 20 ml of HI and incubated at 37°C, at 150 rpm, under microaerobic conditions in an anoxomat system (22). Samples of 1 ml were taken via tubes inserted into the jar via airtight holes without loss of the microaerobic conditions inside the jar. The number of colony forming units was estimated by plating serial dilutions onto charcoal cefaperazone desoxychelate agar (Oxoid) plates.
Construction of FlgR, FlgS, and FlgSt Overexpression Plasmids-To overexpress the FlgR response regulator protein, the chromosomal flgR gene was amplified using the primers 1024HN-NdeI and 1024C-BamHI. The resulting PCR fragment of 1336 bp was digested with NdeI and BamHI and cloned into NdeI and BamHI sites of pT7.7 to form pT7-1024-H6N. The histidine kinase domain of the sensor FlgS was amplified from the chromosomal flgS gene with primers 793H-NdeI and 793-BamHIR. The resulting 645-bp PCR product was digested with NdeI and BamHI and ligated into pT7.7 to form pT7-793t-H6N. To produce the full-length FlgS the chromosomal flgS gene was amplified with primers 793H-NdeI2 and 793-BamHIR. The resulting 1056-bp product was digested with NdeI and BamHI and cloned in pT7.7 to form pT7-793c-H6N. 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 FlgR, FlgS, and FlgSt Proteins-Histidine-tagged FlgR, FlgS, and FlgSt proteins were expressed in E. coli BL21(DE3) containing plasmids pT7-1024-H6N, pT7-793c-H6N, or pT7-793t-H6N. Bacteria were grown in 400 ml of LB broth at 30°C to an A 550 of 0.5. Subsequently, isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 1 mM, and incubation was continued for 4 h. Cells were harvested and suspended in 5 ml of lysis buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 0.1 mM dithiothreitol, 0.2 mM phenylmethanesulfonyl fluoride, 10 mM imidazol, and 1 mg/ml lysozyme). After incubation for 30 min on ice, the bacteria were further disrupted by soni- Detection of Flagellin Proteins by Western Blotting-Protein samples were prepared of overnight grown Campylobacter cultures and were run on 12.5% SDS-polyacrylamide gels. Western blotting was performed as described (23) using polyclonal antisera raised against the major outer membrane protein of C. jejuni 81116 (24) or monoclonal antibodies CF1 specific for flagellin A or CF17 (recognizing both Campylobacter 81116 flagellins A and B) (23).
Real-time RT-PCR-Overnight cultures of Campylobacter were diluted to A 550 of 0.01 in 10 ml of HI and incubated at 37°C, at 150 rpm, under microaerobic conditions. Total RNA was extracted from midexponential phase cultures (A 550 , 0.4 -0.6) unless indicated otherwise, with the RNA-Bee TM kit (Tel-Test, Inc.) according to the manufacturer's specifications. RNA samples were diluted to exactly 1 g/l and treated with RNase-free DNase I (Invitrogen). Primer Express software (Applied Biosystems) was used to design primers to amplify 50 -80-bp fragments. RT-PCR was performed on 0.2 g of DNase I-treated RNA with 1 M of primers using an ABI Prism 7000 (Applied Biosystems) and the Sybr Green RT-PCR kit (Qiagen). Real-time cycler conditions were 30 min at 50°C, followed by 15 min at 95°C and then for 40 cycles at 95°C for 15 s, 50°C for 30 s, and 72°C for 30 s. Specificity was confirmed by electrophoretic analysis of the reaction products and by inclusion of template-or reverse transcriptase-free controls.
Chicken Experiments-Challenge experiments were performed with Ross 308 broiler chickens housed in isolators as described (25). At day 7 after hatching, the animals were inoculated with 10 5 CFU of wild-type Campylobacter, flgR::Cm or BG3 (⌬flaAflaB)::Km mutant. At days 2, 4, 7, and 14 after inoculation, five animals from each group were sacrificed, and C. jejuni was reisolated from caecal contents. The caecal contents were serially diluted and plated onto charcoal cefaperazone desoxycholate agar plates. Isolated C. jejuni colonies were verified by the motility assay and presence of the chloramphenicol cassette.

FlgS and FlgR Constitute a Two-component Signal Transduction
System-Inspection of the C. jejuni genome revealed two putative genes, Cj0793 and Cj1024, that showed similarity with H. pylori sensor protein FlgS (Hp244) and the response regulator FlgR (Hp703), respectively. To address if the C. jejuni FlgS and FlgR proteins form a two-component system, both proteins and a truncated from of FlgS carrying the putative histidine kinase domain only, were expressed in E. coli and purified with help of attached His tags. Phosphorylation assays with the purified proteins showed that both FlgS and its truncated form were efficiently autophosphorylated in the presence of ATP (Fig. 1, A and B). The phosphorylated FlgS subsequently transferred the phosphate to the purified FlgR protein.
The phosphorylation of the FlgR response regulator by the truncated FlgS sensor proteins was a very rapid but transient event (Fig. 1B). The amount of phosphorylated FlgR was much more stable when the phosphorylated intact FlgS was used as phosphate donor (Fig. 1A). Dephosphatase activity was therefore much stronger when truncated instead of the complete FlgS protein was used. As expected, the FlgR protein alone could not phosphorylate itself in the presence of ATP (Fig. 1, A  and B).
Construction and Phenotype Analysis of C. jejuni flgS, flgR, rpoN, and fliA Mutants-To study the flagellar hierarchy in C. jejuni, the genes encoding FlgS, FlgR, and the -factor FliA (Cj0061) were inactivated by allelic exchange with defective copies carrying chloramphenicol cassettes (data not shown). Phenotype analysis of these mutants and a previous constructed rpoN mutant (26) demonstrated that they all were non-motile in a soft agar motility assay (Fig. 2). Electron microscopy revealed that flgS, flgR, and rpoN mutants are unflagelled, whereas the fliA mutant contained truncated flagella (data not shown). When grown in liquid media, the mutants grew much faster than the wild-type with measured doubling times of 23, 24, and 22 min for the flgS, flgR, and rpoN mutants and 60 min for the parent strain, respectively (Fig. 3). The fliA mutant showed an intermediate doubling time of 43 min. Measurement of dry bacterial weights showed that the amount of final biomass produced was 14% higher in the flgR mutant compared with the wild-type. On solid HI media the mutants showed a reduced growth compared with the parent strain (data not shown). Microscopic analyses revealed that the mutants were similarly sized as the parent strain and showed a non-aggregation phenotype in contrast to the wild-type strain. The fliA mutant again showed an intermediate phenotype consistent with the truncated flagella produced by this phenotype (19).
RpoN or FliA Activate the Structural Components of the Campylobacter flagellum-As a first step in studying the regulation of flagella biogenesis, we scanned the C. jejuni genome sequence to identify 54 -and 28 -dependent genes. C. jejuni possesses only three factors: 70 , 54 , and 28 (14). So far 54and 28 -dependent promoters only have been identified in front of flagellar genes flgDE, flaB, and flaA, respectively (18,27,28). In contrast to the Campylobacter 70 promoter consensus sequence, the 54 and 28 promoter consensus sequences resemble those of eubacteria (TGGCAC-N 5 -TTGCW) and (TAAA-N 15 -GCCGATAA), respectively (29,30). Detailed analysis of the genome revealed putative 54 or 28 promoters in front of all structural flagellar genes. From this analysis, more stringent 54 and 28 promoter consensus sequences for Campylobacter could be deduced (Fig. 4).
Regulation of the FlgS/FlgR Regulon-To investigate the potential role of the FlgS/FlgR two-component system and the different (anti-) factors in the regulation of the fla regulon, transcription in the various mutants strains of a large series of flagella and non-nonflagella genes was determined by real-time RT-PCR. To be able to compare the levels of transcription, equivalent amounts of mRNA isolated from mid-exponential grown Campylobacter cultures were used. The transcription of the following genes was determined: rpoD, gyrA, flgS, flgR, rpoN, fliA, CJ1463, flgM, flaA, flaB, and fliD. The constitutively expressed housekeeping genes rpoD ( 70 ) and gyrA were used as controls (31). Whereas the gyrA expression was the same in all Campylobacter cultures, rpoD expression was less in the flgR and rpoN mutants. The flgS, flgR, and rpoN genes were found to be transcribed from 70 promoters (Fig. 5A) the transcription of the rpoN gene. A potential conserved 54 promoter (GGCTCTTTGCTTGCT) seemed present at 77 bp in front of the start codon of the rpoN gene, meaning that the rpoN gene may be to some extent autoregulated (Fig. 5B). Transcription of the flaB and Cj1463 genes, both containing a 54 promoter (Fig. 4), was strongly reduced in the flgS, flgR, and rpoN mutants. This indicates that FlgR must be phosphorylated by FlgS to activate the transcription of these genes together with the -factor 54 (Fig. 5A). The fliA gene is also transcribed by a 70 promoter (Fig. 5C). Whereas a fliA mutation did not affect the transcription of the flgS, flgR, rpoN, flaB, or Cj1463 gene, the flgM, flaA, and fliD genes were regulated by FliA. However, flgM, flaA, and fliD were also found to be regulated by FlgS, FlgR, and RpoN. Upstream of flgM (Cj1464), the genes fliI (Cj1462) and Cj1463 are located. Only in front of fliI, 37 bp from the start codon, a 54 -dependent promoter was identified, meaning that fliI, Cj1463, and flgM are probably transcribed from the same 54 -dependent promoter. Upstream, 39 bp from the start codon of flgM, a 28 -dependent promoter was found. This implies that expression of FlgM is dependent on a 54 and 28 promoter (Fig. 5D). Also fliD appeared to be transcribed from a 54 -and a 28 -dependent promoter located in front of the upstream located flaG gene (Fig. 4). Transcription of the flaA gene was more than 40-fold reduced in a flgS, flgR, or rpoN mutant. In a fliA mutant, the decrease was more than 1000-fold. A 28 -dependent promoter was located in front of flaA but no 54 -dependent promoter could be identified.
Differential Regulation of the Campylobacter Flagellins A and B-To verify that the mRNA data obtained for the flaA and  flaB genes can be extrapolated to the protein level, Western blotting was performed with monoclonal antibodies CF1 (anti-FlaA) and CF17 (anti-FlaA/FlaB). This showed that the amounts of flagellin were strongly reduced in the rpoN mutant (Fig. 6). Comparison of the protein expression patterns for R1-V2 (FlaA-FlaBϩ) and R1-V2 rpoN mutants, indicated that the flagellin that is still produced in a rpoN mutant is flagellin A. In agreement with the flaA flaB mRNA data, a rpoN mutation not only abolished the flagellin B production but also reduced the production of flagellin A.
Growth Phase-dependent Activation of the Flagella Regulatory Genes-Campylobacter loose their motility and flagella in the stationary growth phase (32). To investigate the underlying mechanism, we compared transcription of the flagellar genes in different growth phases. The constitutively expressed genes rpoD and gyrA were both less transcribed in the early stationary and early decline phase compared with the early log phase (Fig. 7), probably because of a general lower metabolic activity of the bacteria. The same pattern was observed for regulatory genes flgS, flgR, and fliA, indicating that, under the conditions employed, the transcription of these genes is not growth-phase sensitive. For rpoN, a much more profound decrease in mRNA levels was observed with a mRNA reduction that exceeded that of flgS, flgR, and fliA by more than 10-fold. Mutations of flgS and flgR have only minor affects on the rpoN expression (Fig.  5A) therefore other regulatory proteins might regulate the rpoN gene. A striking difference between growth phases was noted for the transcription of the flaA and flaB genes. Whereas flaB transcription decreased in the late stationary phase and early decline phase, the amount of flaA mRNA remained stable or even increased. This is consistent with the above proposed role of RpoN in the regulation of the activity of the flaB but not flaA promoter. The transcription of Cj1463 and fliD appeared unaltered in the different growth phases. A reduction in flgM mRNA was observed in the early decline phase.
Chicken Colonization Rate of a Campylobacter flgR Mutant Is Strongly Reduced-To assess the in vivo relevance of the regulation of flagella expression by the FlgS/FlgR two-component system, we compared the colonization behavior of the flgR mutant and the parental strain in chickens. On days 2 and 4 after oral administration, the number of chickens with detectable flgR mutant was low compared with those that received the wild-type strain (Fig. 8). Even at day 14, some chickens were Campylobacter-negative, indicating that the flgR mutant bacteria shed by the Campylobacter-positive chickens could not infect the Campylobacter-negative chickens that were kept in the same isolator. The number of flgR mutants per gram of ceacum content at day 14 of the Campylobacter-positive chicken was similar to that of the wild-type, indicating that the FlgR/FlgS system is important in the initial establishment of the colonization rather than in the survival and persistence in the ceaca of the chickens.
To dissect whether the behavior of the flgR mutant was caused by the lack of flagella or other factors that may be under the control of the two-component system, challenge experiments were carried out with a non-flagellated flaAB mutant BG3. The colonization behavior of this mutant was similar to that of the flgR mutant, suggesting that the flgR mutant behavior is caused via its effect on flagella expression. DISCUSSION Flagella are the locomotory organelles of bacteria that propels them in directions dictated by diverse sensory signals. In many pathogenic bacterial species, flagellar motility is required to reach the ultimate infection niche and to establish infection. The biogenesis of flagella is dependent on a highly regulated pathway of timed gene expression and protein synthesis. In the present study we unraveled the regulation of the C. jejuni fla regulon and revealed that the FlgS/FlgR twocomponent signal transduction system is essential for flagella biosynthesis. Our data indicate that, in response to the thus far unidentified signal, the sensor kinase FlgS autophosphorylates and subsequently transfers its phosphate to its cognate response regulator FlgR. Phosphorylated FlgR together with the -factor RpoN in turn activates the genes needed for the assembly of the hook-basal body filament structure. This process is phase growth-dependent and very energy consuming. Challenge experiments in chickens indicated that the FlgS/FlgR two-component system is important for the efficient passage of the gastrointestinal effect but not for the persistence in the ceaca.
The evidence that FlgS and FlgR protein form a two-component signal transduction system essential for flagella biosynthesis is based on the results of the autophosphorylation and phosphate transfer assays with purified recombinant proteins, and the non-motile, non-flagellated phenotype of the flgS and flgR mutant strains. The rapid phosphorylation of FlgS in the presence of ATP indicated that the protein carried autokinase activity, which is typical for two-component sensor kinase proteins. Both the complete and truncated (lacking the signal recognition domain) recombinant FlgS sensor proteins maintained their phosphorylated status for several hours (data not shown). Phosphorylation of FlgR showed much less stability, a characteristic shared with several other response regulators (33). Removal of the putative signal recognition domain of the FlgS protein destabilized the phosphorylated FlgR even more. This indicates that the complete FlgR sensor not only is able to transfer its phosphate to FlgR but also stabilizes the phosphorylated protein. The non-motile phenotype of the FlgS and FlgR mutants is consistent with the reported lack of flagella (18,19). Most two-component systems are located in a single operon and are often autoregulated. Both flgR and flgS are regulated by the primary -factor 70, whereas no autoregulation was observed. As flgS and flgR are separately located on the chromosome an even more complex regulation of these genes is possible (34).
The first indication about the mechanism via which FlgR regulates flagella biosynthesis was that FlgR is a member of the NtrC family of proteins. These proteins bind to enhancerlike sequences (Ͼ100 bp) upstream of 54 -dependent promoters to activate the transcription of these promoters (35). Thus far 54 -dependent promoters only had been identified in front of flagellar genes flgDE and flaB (18,27,28). The 54 -consensus sequence that is highly conserved throughout nature allowed us to identify 9 additional 54 -dependent loci. The genes in these loci cover all the proteins needed for assemble of the hook-basal body filament structure (36). Definite evidence that these genes are under the control of FlgR was provided by real-time RT-PCR analyses of their transcription. Thus far, transcription regulation in Campylobacter has been studied by using the reporter genes lacZ, luxAB, gfp, cat, and recently astA (18,29,(37)(38)(39). These all have the disadvantage that they need to be translated to identify a signal. Heterologous gene expression in C. jejuni is strongly dependent on the unusual codon usage in this organism (40). Furthermore, insertions of reporter genes in the chromosome are strongly dependent of the integrated position and accumulation of stable reporter proteins may lead to overestimation of real mRNA transcription. Real-time RT-PCR is a very sensitive method to study gene expression and has been proven to be a very accurate and reproducible tool for gene quantification (41).
The use of real-time RT-PCR demonstrated that the amounts mRNA transcribed from flaB and Cj1463, both genes that contain a 54 -dependent promoter, was strongly reduced in a flgS, flgR, and rpoN-negative background. Similar results for flaB transcription were seen by Hendrixson et al. (18) and are consistent with the concept that FlgS activates FlgR which in turn stimulates 54 -dependent promoter activity. The rpoN gene itself did not only contain a 70 , but also a 54 -dependent promoter. Thus suggests that the transcription of the rpoN gene may, to some extent, depend on the phosphorylation state of FlgR, and that the gene may be autoregulated. Thus far, only in Pseudomonas putida has it been shown that the rpoN gene is autoregulated (42).
Other genes whose transcription was under the control of the FlgS/FlgR two-component system and RpoN, were flgM, flaA, and fliD. Real-time RT-PCR showed that the transcription of these genes was also regulated by FliA, indicating that they are transcribed from both 54 -and 28 -dependent promoters. In E. coli and Vibrio cholera transcription of the factor 28 encoding gene fliA is activated by the early flagellar transcription factors FlhDC or RpoN, respectively (4,6). Transcription regulation of C. jejuni fliA occurs independent of the early flagellar transcription factors and looks more similar to the transcription regulation of the Pseudomonas fliA (43). Thus far, the only identified 28 -dependent gene in C. jejuni has been flaA (17,27). The finding that flgM and fliD are regulated by 28 -and 54 -dependent promoters is not unique. FlgM is often found to be dually regulated by 28 , and 54 or FlhDC transcription factors (6,44,45). Whereas the fliD gene is also found to be dually regulated in Salmonella typhimurium (46). The dual transcription from 28 -and 54 -dependent promoters explains that transcription of thus regulated genes in a fliA or rpoN mutant is not abolished but only reduced. This is underscored by the observation that a C. jejuni fliA mutant still possess short flagella, whereas no flagellin is detected on the surface of a C. jejuni fliD mutant (16,19).
Another striking finding in our work was the much more severe reduction of flaA mRNA in the fliA mutant compared with the flgS, flgR, or rpoN mutants. A likely explanation for this observation is the intracellular accumulation of the anti--factor FlgM in these mutants. In the wild-type, FlgM, which inactivates FliA, is secreted into the environment through the hook-basal body complex (Fig. 5D) (47). Because flgM is still transcribed by the 28 (17), however, they could detect flaA mRNA in a rpoN mutant but not in a fliA mutant by RT-PCR. The basis for these controversial results is not known but may be related to their methods to quantify mRNA transcripts.
Real-time PCR on total mRNA derived from C. jejuni in different growth phases demonstrated that the loss of motility and flagella that occurs when C. jejuni is in the stationary growth phase and transforms into doughnut-shaped and coccoidal morphologies (32), is probably not regulated via the FlgS/FlgR two-component system, but other unknown factors that reduce rpoN transcription. The noted reduction in flgS, flgR, rpoN, and fliA mRNA in the early stationary and early decline phase compared with the early log phase was also observed for housekeeping genes gyrA and rpoD. A dramatic change (110 -330-fold) in the expression of housekeeping genes during different stages of bacterial growth has been reported before (48). The changes that we noted may even be underestimated as we compared equal amounts of total RNA to monitor differences between mRNA levels. It has previously been demonstrated that expression of 16 S rRNA decreases earlier than that of other housekeeping genes (48), altering the proportion of mRNA in our samples. Given the rapid and exponential growth kinetics of bacteria and the marked changes in the expression of housekeeping genes during in vitro exponential and stationary growth, the appropriateness of the use of an internal RNA standard is questionable in bacterial gene expression studies (48). Despite these limitations, it should be noted that the amount of rpoN mRNA was far more reduced in early stationary and the early decline phase compared with that of the other flagellar transcription genes, and that flaB mRNA appeared much more reduced than flaA mRNA. The latter may result in an altered composition of the flagella in different growth phases. In H. pylori that like C. jejuni possess a 28 -regulated flaA gene and a 54 -regulated flaB gene, flaB shows a transcription peak prior to flaA during the growth curve, consistent with the ultrastructural findings that FlaB is a hook proximal flagellin subunit, which is incorporated into the growing filament before FlaA (49,50). The function of this event is unknown.
One possible explanation for the growth phase-dependent flagella biosynthesis may be because it is a highly energy demanding process. The increase in growth rate by a factor 3 of the non-flagellated mutants compared with the wild-type and the associated reduction in biomass indicate the large burden of flagella biosynthesis on the protein synthesis machinery. The increase in growth rate and biomass production has not been found for flagella mutants in other bacterial species, such as E. coli (51). The C. jejuni phenotype may be caused by the relatively high and energy costly need to import amino acids as energy and carbon source (52) because of its small genome and lack of many metabolic pathways including sugar-utilizing genes (14).
The conservation of the tight and complex regulation of the flagella biosynthesis in C. jejuni can be explained from the need for highly motile bacteria for efficient colonization of chickens and humans (53,54). Our challenge experiments in chickens indicate that the FlgS/FlgR two-component system is required in the early stages of colonization but not for survival and persistence in the their natural niche, the caeca of chickens. The delayed colonization of chicken exposed to the flgR mutants and the lack of re-infection of Campylobacter culturenegative chicken despite that transmission of C. jejuni from bird to bird in flocks is extremely rapid with colonization within only a few days (9), strongly suggests that FlgR is required for the initial colonization and/or passage of the upper gastrointestinal tract. The finding that successful colonization of unflagellated C. jejuni is only seen when chicken are inoculated with high doses of C. jejuni (54) support this hypothesis.

Regulation of the Campylobacter fla Regulon
The signal that triggers the FlgS/FlgR two-component system to turn on the fla regulon remains to be elucidated. C. jejuni is less able to tolerate environmental stress as is present in the upper gastrointestinal tract than other foodborne pathogens. It cannot survive at a pH lower than 4.9, and is sensitive to osmotic stress (55). Thus, suboptimal environmental conditions may act as a signal. Thus far, activation of the early transcriptional flagellar genes has only been extensively studied for species that carry the master operon flhDC. In E. coli and Salmonella a large number of global regulatory proteins such as cAMP-CRP, DnaK, DnaJ, GrpE, OmpR H-NS, and adenylate cyclase have been implicated in the activation of these genes, indicating a very complex regulated system (4). In species in which part of the flagellar biosynthesis machinery is under the control of a twocomponent system and 54 , like in V. cholera, C. crescentus, and H. pylori, the molecules that activate the sensors of these systems have not yet been identified (56 -58).
In conclusion, we propose on the basis of our results the following model system for the regulation of C. jejuni fla regulon (Fig. 9). In the presence of an unknown signal sensed by the recognition domain of FlgS, this sensor protein autophosphorylates. This phosphate is transferred to the response regulator FlgR, which in concert with the alternative factor RpoN ( 54 ), stimulates the production of RpoN and activates the transcription of the genes needed for the assembly of the hook-basal body filament complex. The formation of the hook-basal body filament structure may allow secretion of the anti-factor FlgM. This results in further activation of FliA ( 28 ) regulated genes including FlaA, which is required for full assembly of the flagella. The proposed C. jejuni flagellar hierarchy largely differs from that of E. coli, V. cholerae, and C. crescentus (59). A similar mechanism of regulation may exist in H. pylori that contains the same transcription activators (47), but this awaits further investigation.