|Home | About | Journals | Submit | Contact Us | Français|
The expression of flagellin genes in most bacteria is typically regulated by the flagellum-specific sigma28 factor FliA, and an anti-sigma28 factor, FlgM. However, the regulatory hierarchy in several bacteria that have multiple flagellins is more complex. In these bacteria, the flagellin genes are often transcribed by at least two different sigma factors. The flagellar filament in spirochetes consists of one to three FlaB core proteins and at least one FlaA sheath protein. Here, the genetically amenable bacterium Brachyspira hyodysenteriae was used as a model spirochete to investigate the regulation of its four flagellin genes, flaA, flaB1, flaB2, and flaB3. We found that the flaB1 and flaB2 genes are regulated by sigma28, whereas the flaA and flaB3 genes are controlled by sigma70. The analysis of a flagellar motor switch fliG mutant further supported this proposition; in the mutant, the transcription of flaB1 and flaB2 was inhibited, but that of flaA and flaB3 was not. In addition, the continued expression of flaA and flaB3 in the mutant resulted in the formation of incomplete flagellar filaments that were hollow tubes and consisted primarily of FlaA. Finally, our recent studies have shown that each flagellin unit contributes to the stiffness of the periplasmic flagella, and this stiffness directly correlates with motility. The regulatory mechanism identified here should allow spirochetes to change the relative ratio of these flagellin proteins and, concomitantly, vary the stiffness of their flagellar filament.
Spirochetes are motile bacteria that are able to swim in highly viscous gel-like environments. The medically important spirochetes include Borrelia spp. (relapsing fever and Lyme disease), Brachyspira spp. (formerly known as Treponema and Surpulina spp., which cause human and animal gastrointestinal diseases), Leptospira spp. (leptospirosis), Treponema pallidum and related subspecies (syphilis, pinta, and yaws), and oral Treponema spp. (periodontal disease) (5, 12, 33, 45, 59). The spirochetes swim by means of rotating periplasmic flagella (see references 6, 36, and 38 for recent reviews). These organelles reside in the periplasmic space and are attached subterminally to the ends of the cell cylinder. Periplasmic flagella (PFs) are structurally similar to the flagella of other bacteria, as each consists of a basal body-motor complex, hook, and filament (8, 27, 28, 39, 48, 55). However, the periplasmic flagellar filament is unique and is among the most complex of bacterial flagella. Specifically, in most spirochete species, the PFs contain at least one flagellar sheath protein, referred to as FlaA, and one to three core proteins, designated FlaB1, FlaB2, and FlaB3 (6, 35-37). In any given spirochete species, each FlaA and FlaB protein is encoded by an individual gene. There is no sequence similarity or antigenic cross-reactivity between FlaA and FlaB proteins (2, 15, 16, 35, 36, 49, 50, 56).
The individual periplasmic flagellar proteins have been studied in detail. FlaA proteins are 37 to 44 kDa and are similar between species based on amino acid sequences and antigenic cross-reactivity (6, 36, 38, 49). These proteins are likely exported to the periplasmic space by the type II secretion pathway, as their N-terminal amino sequences are cleaved and a typical peptidase I cleavage site is present near the N terminus (4, 18, 49). In contrast, FlaB proteins are exported to the periplasmic space most likely via the flagellum-mediated type III secretion pathway (6, 49). FlaB proteins comprise a family of well-conserved proteins. For example, the FlaB proteins of T. pallidum, Treponema denticola, and Brachyspira hyodysenteriae share 57 to 84% amino acid sequence identities (2, 16, 56). FlaB proteins are generally 33 to 41 kDa, and these proteins immunologically cross-react between one another in a given species and also between species (2, 6, 36, 49). Because FlaB proteins have sequence similarities to the flagellins of other bacteria, especially at the N- and C-terminal regions, they are considered to have an identical function in forming the helical flagellar filaments that rotate (36, 37, 64). Several studies have shown that the PFs and the PFs devoid of the FlaA sheath are left-handed helices and do indeed rotate (7, 21, 35, 37).
The regulation of flagellar synthesis is complex (1, 10). Studies with the paradigm models Escherichia coli and Salmonella enterica serovar Typhimurium indicate that a cascade control mechanism is involved in the regulation of flagellar genes. Within this hierarchy, the class I genes (flhCD) initiate the expression of the class II genes that encode structural proteins involved in the synthesis of the motor-basal body-hook complexes and the regulatory proteins. These regulatory proteins, FliA (also called sigma28) and FlgM (anti-sigma28), control the expression of the class III genes, which include the genes encoding flagellin and chemotaxis proteins. FlgM and FliA remain bound as a complex in the cytoplasm during basal body and hook synthesis. As FlgM exits cells via the flagellum export apparatus after completion of the hook, FliA is free to initiate the transcription of late genes. In addition, the disruption of one of the class II genes involved in the synthesis of the hook-basal body complex typically results in nonflagellated cells, as FlgM cannot be excreted.
The above control scheme is common to many motile bacteria, but there are some variations in several bacterial species that contain multiple flagellin units (13, 14, 24, 43, 51, 60, 63). For instance, Helicobacter pylori and Campylobacter jejuni have two flagellin genes (flaA and flaB), and Agrobacterium tumefaciens contains four flagellin genes (flaA, flaB, flaC, and flaD). In these organisms, at least two different sigma factors are involved in the regulation of their multiple flagellin genes (9, 22, 60). Similar to the above organisms, spirochetes also have multiple flagellin genes, and previous studies suggest that the regulatory hierarchy in spirochetes may be different from the well-studied paradigm model of the enteric bacteria (6, 36). For example, although the homolog of fliA has been found in T. denticola, T. pallidum, and L. interrogans (16, 52, 56) and sigma28 consensus sequences have been identified upstream of several flaB genes, the FlgM homolog has not been found in any spirochete species. In contrast to the case for flaB genes, the promoters for flaA genes have sigma70 consensus sequences (4, 18, 26, 32). The regulation of these flagellin genes has not been systematically investigated for any spirochete species that contains multiple FlaB proteins, such as Treponema and Brachyspira species.
The spirochete B. hyodysenteriae, the causative agent of swine dysentery, can be genetically manipulated by both allelic exchange mutagenesis and generalized transduction via a defective transducing phage (23, 58). These genetic tools and the recent availability of its genomic sequence have led this species to emerge as a model spirochete for genetic analysis (2). In addition, its PFs are similar in composition and structure to those of T. denticola and the uncultivable T. pallidum (3, 30-32, 49). As such, B. hyodysenteriae has been used to analyze the complex structure of the PFs and the contribution of the individual filament proteins to filament organization, filament stiffness, and motility (35, 37, 53). Previous experiments have shown that single and most double mutants have decreased motility. Only a flaB1 flaB2 double mutant is completely nonmotile (35, 37). In this report, we continued to use B. hyodysenteriae to investigate the regulation of its multiple flagellin genes and to better understand its flagellar filament assembly. We found that the flagellin genes are controlled by two different sigma factors: flaB1 and flaB2 are regulated by sigma28, and flaA and flaB3 are controlled by sigma70. These results were further supported by the analysis of a flagellar motor switch fliG mutant. In addition, in a manner somewhat similar to that seen with the recently characterized flaB1 flaB2 double mutant (37), incomplete hollow filament structures composed primarily of FlaA continued to be synthesized in the fliG mutant.
E. coli and S. enterica serovar Typhimurium strains were grown in Luria-Bertani medium. S. enterica wild-type strain SJ1103 and the nonflagellated fliG mutant SJ1368 were provided by D. Blair, University of Utah (25, 29). The origin and culture conditions for B. hyodysenteriae strain B204 (wild type) have been previously described (35, 37, 53). B. hyodysenteriae strains were grown in brain heart infusion broth supplemented with 10% fetal bovine serum (BHI-FBS) or Trypticase soy agar plates supplemented with 5% whole blood (TSAB) in an atmosphere of 90% N2, 10% CO2, and 1 to 2% O2. Chloramphenicol (20 μg/ml) was added to the medium for selecting B. hyodysenteriae mutants.
Primer extension analysis was performed according to our previous studies (20, 34, 65). For the isolation of RNA, 500 ml of mid-exponential-phase B. hyodysenteriae cells were harvested and washed twice with phosphate-buffered saline (PBS; pH 7.4). Total RNA was isolated with Tris reagent (Sigma-Aldrich) according to the manufacturer's instructions, followed by treatment with RNase-free DNase I (Promega, Madison, WI). Further purification of RNA was performed using a Qiagen RNA isolation kit (Qiagen, Valencia, CA). Primer extension was carried out by using an AMV reverse transcriptase primer extension system (Promega) with a slight modification. Briefly, 20 pmol of the following reverse primers were used (the primers are approximately 80 to 150 bp from the putative promoter consensus): flaA, 5′-TAAACCGTAAGCAGTGC-3′; flaB1, 5′-TGCGTTGTGCGTTTATAGC-3′; flaB2, 5′-TGCATTTATAGCACTAATG-3′; flaB3, 5′-TATTGTTGATAATCATAC-3′. These primers were labeled with [32P]ATP at 37°C and purified with a Qiagen nucleotide removal kit. For reverse transcription, 1 μl of labeled primer was mixed with 20 μg RNA and incubated at 58°C for 20 min. AMV reverse transcriptase and other reagents were then added, and the extension reaction was carried out at 42°C for 45 min. The products were precipitated with ethanol and dissolved in 5 μl standard DNA loading buffer. The primer extension products were run along with the sequence ladder generated from a flaA primer and a plasmid containing the flaA gene on a sequencing gel with a DNA sequencing kit (GE Healthcare, Piscataway, NJ).
To clone B. hyodysenteriae fliG, a pair of degenerative primers was designed based on two highly conserved regions of FliG (N terminus, QQEHP; C terminus, MFVFED) found in other bacteria. These two primers were designated DFG1 (5′-CADCADGADCAYCCNCADAC-3′) and DRG1 (5′-TYYTCDAANACDAACAT-3′). By using these two primers and B. hyodysenteriae chromosomal DNA as a template, a 740-bp PCR product was obtained and further cloned into pGME-T vector (Promega). A DNA sequence analysis and a BLAST comparison indicated that this PCR fragment shared high homology to fliG genes from other bacterial species. To obtain the entire fliG gene, a 433-bp fliG gene fragment was amplified by PCR (forward primer, 5′-CTAACCAACTGTACCTGC-3′; reverse primer, 5′-GGATACGATGCCAAAGAT-3′) and labeled with digoxigenin (Boehringer Mannheim, Germany). The labeled PCR fragment was used as a probe to screen the fliG gene from a lambda ZAP II DNA library of B. hyodysenteriae, which was kindly provided by T. Stanton (35).
A positive clone containing approximately 7 kb of B. hyodysenteriae chromosomal DNA was obtained and sequenced. One open reading frame (1,026 bp) encodes a 342-amino-acid protein with a calculated molecular mass of 38.9 kDa. BLAST analysis revealed that this protein shares a sequence identity of 29% to E. coli FliG and a 100% identity to the FliG homolog (BHWA1_00668) in the recently sequenced genome of the B. hyodysenteriae WA1 strain (2). In addition, the C-terminal region of B. hyodysenteriae FliG, which is essential for binding to other motor proteins, such as MotB and FliM, contains all the key residues (data not shown) that were identified in the FliG proteins of the enteric bacteria (41). The data, taken together, indicate that the sequenced gene is a fliG homolog.
We used a previously described strategy to inactivate fliG (35, 37). Briefly, a 1,255-bp fragment containing the entire fliG gene was amplified by PCR with primers MG1/MG2 (MG1, 5′-ACAAGCTATGATGAATGC-3′; MG2, 5′-CACAACCATTTCATCTTC-3′) and cloned into pGEM-T vector (Promega). A 415-bp EcoRI/ClaI fragment was deleted and replaced by 920 bp of a chloramphenicol-resistant marker (cat), which was amplified by PCR with primers CAT1/CAT2 (CAT1, 5′-GAATTCGAGCTCGGTAC-3′; CAT2, 5′-ATCGATGCCTGCAGGTCGAC-3′). The insert was flanked by 433 bp upstream and 407 bp downstream. The orientation of the insert was confirmed by PCR.
The PFs were purified as previously described (35, 37). The B. hyodysenteriae FlaA and FlaB antisera and the monoclonal antibody against B. hyodysenteriae FlaBs were provided by M. Jacques (University of Montreal) and G. Duhamel (University of Nebraska), respectively. The S. enterica FliG antibody was obtained from the late R. Macnab (Yale University). The monoclonal antibody against the B. hyodysenteriae 35-kDa protein was prepared using standard methodology by injecting purified PFs derived from the mutant A204 (53) lacking the FlaA protein into mice. Lymphocyte fusions were carried out with myeloma cell line P3.X63.Ag8.653 (62). Immunoblots were developed by using horseradish peroxidase secondary antibody with an ECL luminol assay (GE Healthcare).
Outer membrane-disrupted cells, thin sections, and purified PFs from the wild type and the fliG mutant LC-1 were examined by transmission electron microscopy (TEM) and cryo-TEM as previously described (35, 37). Tobacco mosaic virus (TMV) was used as an internal control to measure flagellar diameter. The diameters of the PFs were directly measured from the digital images by using Openlab software (Improvision, Inc., Coventry, United Kingdom). No more than two measurements were made per filament, and at least 30 filaments were measured per sample. The measurements were calibrated by the diameters of TMV and were expressed as means plus or minus standard errors of the means. Significant differences were evaluated using analysis of variance (ANOVA) and the post hoc Tukey-Kramer test.
RNA sample preparations and quantitative reverse transcriptase PCR (qRT-PCR) were performed as previously described (55). Briefly, RNA samples were purified from the wild type and LC-1 as described above. Equal amounts of total RNA (approximately 0.5 μg) were used to generate cDNA by using an AMV reverse transcriptase primer extension system according to the manufacturer's protocol (Promega). To ensure that the RNA preparations were free of genomic DNA, reverse transcription was performed with or without reverse transcriptase. Quantitative PCR was carried out in triplicate on cDNA with SYBR green master mix (Qiagen) and a light cycler (Roche, Indianapolis, IN). The 16S rRNA gene was used as an internal control, and its reverse transcription product was diluted 5-fold for quantitative PCR analysis. The results are expressed as the threshold cycle at which the measured fluorescence increased above background. In addition, the products derived from qRT-PCR analysis were directly detected by electrophoresis using a 1% agarose gel. The following primers were used for qRT-PCR analysis: flaB1 (forward, 5′-GATTCTGCTATGATTTCTAG-3′; reverse, 5′-TAGATACCGTTAGCAGCTTG-3′), flaB2 (forward, 5′-TTCGCGGTTTACGTCAGGCT-3′; reverse, 5′-CAGCATCAGTATAGATACCG-3′), flaB3 (forward, 5′-AACTGGAAATTCAATGAC-3′; reverse, 5′-GTGCGTTTCTAGTAGCTTG-3′), and 16S rRNA gene (GenBank accession number M57741) (forward, 5′-AGCAATCCGCTTTACGATG-3′; reverse, 5′-CGATGTCGCTCCATCAGACT-3′).
The GenBank accession number for the fliG gene of B. hyodysenteriae is AAS89035.
Our previous studies have shown that the four flagellin genes of B. hyodysenteriae are monocistronic (35). Sequence analysis from other laboratories postulated that upstream of flaA there is a sigma70 promoter consensus sequence and that flaB1 has a consensus sigma28 promoter (17, 32). The promoter consensus sequences for flaB2 and flaB3 remain unidentified. We used primer extension analysis to accurately map the transcription start sites for all four of these filament genes. Sigma70-like promoter sequences were identified as transcriptional start sites in the upstream regions of flaA and flaB3 (Fig. 1a and b). The −10 and −35 regions of flaA and −10 region of flaB3 were well conserved compared to the typical sigma70 consensus sequences; there was some degree of variation at the −35 region of flaB3 (Fig. (Fig.1b).1b). In contrast, typical sigma28 promoter consensus sequences were found upstream of both flaB1 and flaB2. These two promoters showed strong conservation at the −10 and −35 regions (Fig. (Fig.1b).1b). These results indicate that the flagellar filament genes in B. hyodysenteriae are differently regulated: flaA and flaB3 by sigma70 and flaB1 and flaB2 by sigma28.
The flagellar motor switch complex protein FliG plays a critical role in flagellar biosynthesis and motility. In the enteric bacteria, the inactivation of fliG disrupts the flagellum export apparatus and consequently turns off the expression of class III genes, such as fliC, due to the increase of intracellular FlgM (1, 10, 29). We tested whether the expression of the four filament genes in B. hyodysenteriae would also be inhibited in a fliG mutant. The fliG gene of B. hyodysenteriae was targeted using allelic exchange mutagenesis by replacing a 415-bp EcoRI/ClaI fragment with a chloramphenicol resistance cassette (cat) (35, 53). Ten independent clones were analyzed, and all of them contained cat inserted into fliG, as determined by PCR (data not shown). One clone (LC-1) was analyzed in detail. Western blot analysis indicated that whereas wild-type B. hyodysenteriae (strain B204) and S. enterica (strain SJ1103) cell lysates strongly reacted with the FliG antiserum, no reactivity was detected from the lysates of LC-1 and the S. enterica fliG mutant SJ1368 (Fig. (Fig.2a).2a). These results suggest that the synthesis of FliG was completely inhibited in LC-1.
The phenotype of LC-1 was analyzed in detail. Null fliG mutants in other bacterial species are typically nonmotile and lack flagella (29). We found that LC-1 was similar to such mutants. Dark-field microscopy revealed that LC-1 was completely nonmotile in all the standard tests for assaying B. hyodysenteriae motility (35, 53): liquid media, liquid media containing 1% methylcellulose, and 1% saline with or without methylcellulose. Swarm plate analysis revealed that the wild-type swarm diameter was 14.2 mm after a 3-day incubation. In contrast, the region of LC-1 remained at the inoculum site and was 2.8 mm in diameter (Fig. (Fig.2b).2b). These results indicate that LC-1 was completely deficient in swarming. Transmission electron microscopy (TEM) revealed that no filaments attached to the cell cylinders (over 20 cells examined) were evident in the LC-1 mutant (Fig. 3a and b). In addition, whereas the PFs were seen in the cross sections of the wild type, these organelles were absent in LC-1 (Fig. 3c and d). These results indicate that FliG is essential for the motility and the assembly of PFs in B. hyodysenteriae.
We tested whether all four filament proteins accumulated in LC-1 by using Western blot analysis. As expected, all four filament proteins (FlaA, FlaB1, FlaB2, and FlaB3) were detected in the whole-cell lysates of the wild type. However, only FlaA and FlaB3 were found in the lysates of LC-1 (Fig. (Fig.4).4). The amount of FlaA in LC-1 was similar to that of the wild type. Although FlaB3 was detected in the mutant, its level was decreased by approximately 75%. In contrast to FlaA and FlaB3, the accumulation of FlaB1 and FlaB2 was completely inhibited in LC-1. These results are consistent with the above primer extension analysis and what is found for other bacteria: in a fliG null mutant background, the accumulation of proteins that are regulated by sigma28 is specifically inhibited (1, 10, 29).
We directly tested if the complete inhibition of FlaB1 and FlaB2 accumulation in LC-1 occurs at the transcriptional level. We also tested whether the 75% decrease in the accumulation of FlaB3 is a reflection of a decrease in the transcription of flaB3. RT-PCR and quantitative RT-PCR analysis were used to measure flaB1, flaB2, and flaB3 mRNA. We found that the expression of flaB1 and flaB2 genes was completely blocked in the LC-1 mutant, whereas the flaB3 messenger was detected in this mutant (Fig. (Fig.5).5). qRT-PCR analysis confirmed these results, and it revealed that the flaB3 mRNA in the mutant remained at the same level as that of wild type (ΔCT, the difference between threshold cycles of the wild type and LC-1, was 0.1). These results indicate that the inhibition of FlaB1 and FlaB2 accumulation in LC-1 occurs at the transcriptional level. In contrast, because the levels of flaB3 mRNA in the wild type and the mutant were equivalent, the decrease in accumulation of FlaB3 in LC-1 takes place at the posttranscriptional level. Similar posttranscriptional modulation of FlaA and FlaB accumulation in certain mutants of Borrelia burgdorferi has been noted (47, 55). The data, taken together, strongly indicate that the flagellin genes of B. hyodysenteriae are differentially regulated.
We recently found that the flaB1 flaB2 nonmotile mutant of B. hyodysenteriae synthesized incomplete hollow filaments composed primarily of FlaA, a 35-kDa protein, and reduced the amount of FlaB3 (37). Since FlaA and a decreased amount of FlaB3 still accumulated in LC-1, we tested whether the remaining proteins form structures similar to those observed with the flaB1 flaB2 mutant. TEM and negative-staining analyses revealed that the filaments purified from LC-1 were less helical than those of the wild type (Fig. (Fig.6a),6a), which is similar to the filaments obtained from the flaB1 flaB2 mutant. The internal region of the filaments from the LC-1 mutant stained dark, whereas the wild-type filaments stained primarily on the periphery (Fig. (Fig.6b).6b). These results suggest that the filaments from LC-1 are hollow. We further analyze the filaments by cryo-TEM. The wild-type PFs appeared dense throughout the filament image, whereas only along the periphery of the filaments of LC-1 was there obvious protein deposition (Fig. (Fig.6c).6c). These results are similar to those with the flaB1 flaB2 mutant filaments and further substantiate the idea that the filaments from LC-1 are hollow. We measured the diameter of the filaments observed for LC-1. The diameter of the filaments from the mutant (19 ± 0.9 nm) was similar to that of the wild type (21 ± 1.2 nm). These results suggest that the filaments from LC-1 form hollow tube-like structures with the same diameter as the intact PFs of B. hyodysenteriae wild-type cells (37).
We analyzed the protein composition of the purified hollow filaments from LC-1 by SDS-PAGE and Western blotting. As shown in Fig. Fig.7,7, only FlaA and a 35-kDa protein were detected in the filament preparation. The 35-kDa protein purifies along with the PFs, but its function is unknown (35, 37). The ratio between FlaA and the 35-kDa protein was approximately 1:0.5, which was similar to that of the hollow tubes isolated from the flaB1 flaB2 mutant (37). In contrast to the flaB1 flaB2 mutant, no FlaB3 was associated with the filaments in LC-1 even though FlaB3 was detected in its whole-cell lysates (Fig. (Fig.4).4). These results suggest that although FlaB3 is synthesized in the mutant, it was not assembled in the hollow tubes in LC-1. Taken together, these results indicate that FlaA in association with the 35-kDa protein polymerizes into the hollow tubular filaments.
In this study, we found that both sigma28 and sigma70 factors are involved in the regulation of the filament genes in B. hyodysenteriae. Primer extension analysis revealed that sigma70 promoter sites mapped upstream of both flaA and flaB3, whereas sigma28 promoters were identified upstream of both flaB1 and flaB2. These results suggest that the expression of flagellin genes is regulated by different mechanisms. In support of this proposition, we found that both the flaB1- and flaB2-encoded proteins and mRNA were completely inhibited in the LC-1 mutant. Concomitantly, there was no inhibition of flaA and flaB3 message. These results are consistent with the flaB1 and flaB2 genes being transcribed by sigma28 and flaA and flaB3 being transcribed by sigma70. In other bacteria, the anti-sigma28 factor FlgM modulates expression of sigma28 consensus promoters (1, 10, 11, 54). Based on immune cross-reactivity with S. enterica FlgM, we recently identified a putative FlgM homolog in cell lysates of B. hyodysenteriae of the same mass (12 kDa) as that of S. enterica FlgM (C. Li and N. W. Charon, unpublished data). These results suggest that such a factor potentially modulates expression of flaB1 and flaB2 in B. hyodysenteriae.
The control of flagellin gene regulation in B. hyodysenteriae is different in many respects from that of the spirochete B. burgdorferi but shows some similarity. B. burgdorferi differs from other spirochetes as its PFs consist of only one FlaA and one FlaB species (19, 46). In addition, B. burgdorferi is unique among bacteria and other spirochete species, as no sigma28 factor is present in its genome and all of its motility and chemotaxis genes are controlled by consensus sigma70-like promoters (6). Moreover, posttranscriptional control is involved in flagellar gene expression (19, 47, 55). For example, the mutants that fail to synthesize its hook-basal body complex have a marked inhibition in the accumulation of both FlaB and FlaA (55). However, its flaA and flaB mRNA levels are equivalent to that of the wild type, which strongly points toward posttranslational control. In contrast, the decrease in the accumulation of FlaB1 and FlaB2 in the LC-1 mutant directly correlates with a decrease in the corresponding mRNA. As noted, the decrease in FlaB3 shows posttranslational processing, which is similar to that found with B. burgdorferi.
Does the control pattern observed with B. hyodysenteriae flagellin proteins have any obvious significance? In the wild-type cells, the ratio of FlaA to FlaB1, FlaB2, and FlaB3 is 1:0.42:0.22:0.45 (35). Genetic and structural evidence suggests that FlaB1, FlaB2, and FlaB3 are distributed throughout the filament core and that each flagellin does not reside in one specific domain within the core. Each flagellin also contributes to the overall structure of the filament and optimal motility (35, 37). B. hyodysenteriae is one of the few species of bacteria with multiple flagellins for which the relative stiffness of its flagella has been determined for specific flagellin mutants (35). In addition, this flagellar stiffness has been shown to positively correlate with motility. The precise function of the FlaA sheath remains unknown, but it adds stiffness to the PFs (37). The three FlaB proteins also vary with respect to both their contribution to the overall stiffness of the flagellar filaments and their contribution to motility (37). For example, the PFs of flaB1 mutant are 19% less stiff than those of the wild type, and this mutant shows a concomitant decrease in speed of 28%. In contrast, the PFs of the flaB3 mutant are 7% less stiff than those of the wild type, and this mutant has a decrease of 13% in cell speed (37). Thus, FlaB1 contributes to flagellar stiffness and motility more than FlaB3. With FlaB1 under the control of sigma28, and FlaB3 under the control of sigma70, this ratio is likely to be modulated via the different control systems. Thus, B. hyodysenteriae has the potential to modulate the stiffness of the PFs in response to environmental signals by changing the ratios of FlaB1, FlaB2, and FlaB3 to one another. For H. pylori, a similar suggestion has been made with respect to modulating its multiple flagellin proteins in response to environmental stimuli (42, 44, 61). Conceivably, B. hyodysenteriae cells could synthesize PFs that are stiffer early in infection (higher ratio of FlaB1 and FlaB2 to FlaB3) before colonization. During the later stages of infection, once colonization and attachment of the cells to the intestine has occurred, the cells could switch to synthesizing PFs with a lower ratio of FlaB1 and FlaB2 to FlaB3. Such cells would have PFs that are less stiff, and the cells would be less motile than those in the early stage of infection.
In other spirochete species, both sigma70 and sigma28 consensus sequences have been identified upstream of several flagellin genes by both sequence (Table (Table1)1) and primer extension analysis. T. denticola has sigma70 promoters at upstream of its flaA (TDE1408) and flaB3 (TDE1457) genes and sigma28 consensus sequences upstream of its flaB1 (TDE1004) and flaB2 (TDE1477) genes. Thus, the pattern in T. denticola is similar to one found with B. hyodysenteriae, but confirmation that these are indeed the start sites will depend on promoter mapping. With respect to T. pallidum, sigma70 consensus sequences are upstream of T. pallidum flaA (TP0663), whereas a sigma28 sequence is upstream of both flaB1, and a putative flaB2 flaB3 operon. Taken together, these studies (Table (Table1)1) and the promoter analysis of other spirochete flagellin proteins (4, 18, 26) suggest that flaA genes are transcribed by sigma70, whereas there is variation in whether a given flaB gene is transcribed by sigma28 or sigma70.
The motor switch complex protein FliG plays an essential role in flagellar assembly and motility (1, 10, 29, 40). Null mutants of fliG are nonmotile and lack flagella, and we found results similar to those reported here with B. hyodysenteriae. A deletion mutant of fliG (TDE1216) in T. denticola was reported to be nonmotile. This mutant still synthesized PFs, but their number was markedly decreased (57). In most motile bacteria, only one homolog fliG is present. However, in several spirochete species, including T. denticola, two fliG homologs have been detected (15, 16, 52, 56). B. hyodysenteriae has only one fliG homolog (2) (R. Zuerner, personal communication). One possible explanation for the T. denticola results is that there is a redundancy between its two fliG genes such that mutation in one weakly complements the other.
The assembly of the PFs in B. hyodysenteriae and other spirochetes is complex. Hollow tubes composed primarily of FlaA were found in the flagellar fractions prepared from LC-1 and also from the flaB1 flaB2 mutant (37). Moreover, the diameters of the hollow tubes from both mutants were identical to those of the intact PFs. The simplest model is that FlaA, which is secreted into the periplasmic space by a type II secretion pathway, polymerizes around the FlaB proteins, as these core proteins emerge through the hook-basal body complex via the type III secretion pathway. If there are insufficient FlaB proteins being secreted, or if FlaA forms the sheath at a higher rate than the secretion and polymerization of the multiple FlaB proteins, FlaA continues to polymerize and forms hollow tubes. As a corollary, if FlaB polymerizes faster than FlaA is able to decorate the core surface or if there is less FlaA than FlaB that is synthesized by the cells (as occurs in B. burgdorferi ), then the filaments will be partially unsheathed. Our results with LC-1 suggest that FlaA is able to form hollow tubes without detectable FlaBs. It may be that a small amount of FlaB3 is present to initiate polymerization of FlaA, but this limited amount may be beyond the level of detection. In addition, both this report and previous studies have shown that the 35-kDa protein is tightly associated with the PF preparations in B. hyodysenteriae (35, 37). Preliminary peptide mapping results indicate that the 35-kDa protein is a variable membrane protein, suggesting that there may be an interaction between the FlaA sheath and the outer membrane. We anticipate that further genetic studies will help define the function of the 35-kDa protein and determine the precise steps in the assembly of the PFs.
We thank T. Stanton for providing B. hyodysenteriae strains and the DNA library and R. Zuerner for sharing the genomic information for the B. hyodysenteriae B204 strain. We thank G. Duhamel, D. Blair, M. Jacques, and the late R. Macnab for strains and/or antibodies. We also thank M. Elliott for assistance in monoclonal antibody production.
This research was supported by Public Service grants DE12046 and AI-29743 and USDA grant 95-37204-2132 awarded to N.W.C., NIH grants AR050656 and AI078958 to C.L., and NIH grant RR01219 supporting the Albany NCRR Resource.
Published ahead of print on 19 March 2010.