Bordetella pertussis Can Be Motile and Express Flagellum-Like Structures

This report provides evidence for motility and expression of flagella by B. pertussis, a bacterium that has been reported as nonmotile since it was first isolated and studied. As with B. bronchiseptica, B. pertussis cells can express and assemble a flagellum-like structure on their surface, which in other organisms has been implicated in several important processes that occur in vivo. The discovery that B. pertussis is motile raises many questions, including those regarding the mechanisms of regulation for flagellar gene and protein expression and, importantly, the role of flagella during infection. This novel observation provides a foundation for further study of Bordetella flagella and motility in the contexts of infection and transmission.

thesis and functions are present in both genomes (1)(2)(3). There was, however, a stop codon located 1,313 bases into flhA (total gene, 2,119 bases) of B. pertussis. FlhA is a transmembrane, type III secretion protein that serves as docking site for Fli(X) ATPases and FliC filaments. FlhA is responsible for the export of FliC filaments for flagellar tail assembly (4) and is described as one of the 24 core proteins essential for flagellar assembly (5). Thus, a stop codon in flhA appeared consistent with B. pertussis being nonflagellated and nonmotile (6).
Information about regulation and relevance of Bordetella motility and flagellar gene and protein expression is largely limited to work with B. bronchiseptica. Akerly et al. showed that B. bronchiseptica flagellar expression and motility are controlled by the BvgAS two-component system (7,8), which modulates among the virulent Bvg(ϩ) phase, intermediate Bvg(i) phase, and avirulent Bvg(Ϫ) phase. Flagellar gene and protein expression and the motile phenotype occur primarily in the Bvg(Ϫ) phase (8). These findings have raised questions about relevance of motility and flagellar expression during infection, as these phenotypes are associated with the avirulent Bvg(Ϫ) phase. Recently, van Beek et al. found that within the mouse respiratory tract, B. pertussis expresses Bvg(Ϫ) genes, including those from the flagellar operon (9), and Bvg(Ϫ) B. pertussis strains have been isolated from patients during infection (10). In addition, flagellar expression and motility appear to be important for virulence phenotypes. B. bronchiseptica motility is required to reach intracellular niches within the host; flagella are involved in both motility and adherence to biotic and abiotic surfaces (6,(11)(12)(13). These data suggest that during the Bvg(Ϫ) phase, flagellar expression and motility may have roles in infection and/or transmission.
B. pertussis can be motile. Several data sets have demonstrated differential regulation of B. pertussis genes associated with assembly and function of flagella. Specifically, Barbier et al. compared a wild-type (WT) strain (UT25) to the UT25 ΔrseA mutant (14). In UT25 ΔrseA, RpoE functions were increased, and surprisingly, genes associated with flagellar assembly and function were increased between 1.5-and 22-fold (15). Additionally, expression of flagellar genes has been observed during mouse infection (9,16). Based on these data sets, we tested the hypothesis that B. pertussis produces flagellar proteins, enabling B. pertussis motility.
We examined B. pertussis for motility during growth in soft agar, as previously described for bordetellae, using B. bronchiseptica WT strain RB50, Bvg(ϩ) RB53, and Bvg(Ϫ) RB54 as controls (8). As expected, from previous observations, WT RB50 is motile, Bvg(ϩ) RB53 is nonmotile, and Bvg(Ϫ) RB54 is motile at 24 h when grown at 37°C (Fig. 1A). B. pertussis WT BP338 and a Bvg(Ϫ) mutant (Tn5::bvgS) BP347 (17) were stabbed into motility agar, grown at 37°C, and observed over the course of 72 h. Although the WT BP338 strain was nonmotile, the Bvg(Ϫ) mutant BP347 was motile at 72 h (Fig. 1B). In several experiments, we observed that B. pertussis WT BP338 and another lab-adapted B. pertussis WT strain (BPSM) could become motile without additional manipulation, but this did not occur consistently (in Ͻ15% of experiments). An example is included in Fig. S1 in the supplemental material. We hypothesized that when the B. pertussis WT strain becomes motile, it is due to either phase variation to the Bvg(Ϫ) phase or a genetic mutation that results in Bvg(Ϫ) mutants that dominate and spread. To test for Bvg(Ϫ) mutants, motile bacteria from outer edges of WT BP338 spreading zones were isolated and replated on fresh plates. This yielded both Bvg(ϩ) and Bvg(Ϫ) colonies, based upon colony size and hemolysis on Bordet-Gengou (BG) blood agar plates. Bvg(ϩ) and Bvg(Ϫ) colonies were then isolated by replating the individual colonies on fresh BG blood agar plates. These isolated bacterial populations were then used to grow overnight liquid cultures, and motility assays were completed. This had no effect on the motile phenotypes of these bacteria. (The motile phenotype was still variable.) To confirm Bvg regulation of motility, plates were supplemented with 40 mM MgSO 4 to elicit chemical modulation of B. pertussis to the Bvg(Ϫ) phase. After 72 h, the radius of the motility zone was recorded. Representative images from six experiments show  (18), we also added 10% fetal bovine serum (FBS) to the motility agar plates, ensuring that the agar concentration remained at 0.4%. The presence of serum significantly increased motility in B. pertussis (Fig. 1F to H).
To determine if the above data are indicative of a general phenomenon, we tested lab-adapted and clinical isolates and found that some, but not all, were consistently motile. This was the case for WT BP338 and Bvg(Ϫ) mutant BP437. It is not yet clear as to why B. pertussis is not always motile under motility-promoting conditions. Table 1 describes all strains tested and their motility phenotype in the presence of 40 mM MgSO 4 (ϩ, motile; Ϫ, nonmotile). Representative examples of strains demonstrating motility in these assays when modulated to the Bvg(Ϫ) phase are presented in Fig. 1 (lab-adapted strains in panel I and clinical isolates in panel J [for more information, see Table 1]). The observed B. pertussis spreading occurred within the agar layer, a feature indicative of swimming motility, which, in other bacteria, is flagellum dependent (19). Upon isolation of motile B. pertussis from the outer edge of motility halo at 48 h, a video recording of live bacteria (ϫ1,000 magnification) shows rapid movement across the field of view, also indicative of swimming motility (see Movie S1 in the supplemental material).
Strains from different isogenic backgrounds and clinical isolates were grown as described for the motility agar assay. Motility, measured by outward spreading from the point of inoculation in the agar, was determined to be positive (ϩ) if the strain was consistently (Ͼ80%) motile under modulated conditions (40 mM MgSO 4 ). If the strain was not consistently motile, the motility phenotype was determined to be negative (Ϫ). Experiments were repeated 5 times; the strains were tested in duplicate in each experiment.

Bordetella pertussis cells can express flagellum-like structures on their surface. B. bronchiseptica motility is mediated by flagella, as B. bronchiseptica
ΔflaA mutants are nonmotile (7,8). Although we have not shown that B. pertussis motility is flagellum dependent, in light of our data, we examined motile B. pertussis from soft agar plates containing 40 mM MgSO 4 for the presence of flagella. Bacteria were isolated from the outer edges of the spreading zones and prepared for negative-stain transmission electron microscopy (TEM) using methods adapted from Akerly et al. (8). B. bronchiseptica strains were flagellated (RB50 in Fig. 2A and RB54 in Fig. 2B) and had multiple flagella per bacterium, as previously described. The B. pertussis BP338 WT and the BP347 Bvg(Ϫ) mutant expressed thin flagellar structures on their surfaces, most frequently only one flagellum per bacterium. There were no differences in the number or frequency of flagellated WT BP338 and Bvg(Ϫ) BP347, but only approximately 23% of total observed bacteria were flagellated (representative images of flagellated BP338 and BP347 are shown in Fig. 2C to F). The lab-adapted strain B. pertussis UT25 and the clinical isolate B. pertussis V015 had flagellum-like structures on their surfaces (Fig. 2G  and H).
To confirm that B. pertussis is able to produce flagellin, we tested B. pertussis isolated from motility agar for reactivity with monoclonal antibody 15D8, which recognizes flagellin (FliC) from B. bronchiseptica. All motile strains react with the monoclonal FliC antibody, and nonmotile strains do not. With this method, although flagellin is clearly being produced, it is possible that the reacting flagellin was intracellular and not exclusively extracellular (from an assembled flagellum). Because of the low frequency of flagellated B. pertussis and the single flagellum per bacterium, we used whole bacteria to test for the presence of flagellin. Using previously described flagellum purification methods to shear flagellar structures from the bacterial surface, we were unable to detect flagellin protein.
A panel of antibodies raised against flagellins from individual bacterial species, Salmonella enterica serovar Typhi (Fig. 2I, row 2), Shigella sonnei (Fig. 2I, row 3), and Vibrio cholerae (Fig. 2I, row 4) (20) recognized an ϳ40-kDa band in all motile B. pertussis strains, providing evidence that these B. pertussis strains express flagellin protein. These data, taken with the negative-stain TEM images, confirm that motile B. pertussis cells are able to display flagellum-like structures on their surfaces and express flagellin protein that is immunologically comparable to that from B. bronchiseptica and other Enterobacteriaceae.
Despite microbiological literature stating that "B. pertussis is a nonmotile organism" (21), we have demonstrated that B. pertussis can be motile and express flagella. Several lab-adapted strains and clinical isolates are motile, and B. pertussis motility is enhanced in the Bvg(Ϫ) phase. These motile strains express flagellum-like structures and flagellin protein, as verified by negative-stain TEM and Western blotting. Regardless of our inability to detect, specifically, flagellin protein that has been exported to the bacterial surface, we still observe motile bacteria and believe that B. pertussis motility is the major phenomenon described here. These data represent novel and unanticipated observations, which raise many questions to be answered in future studies.
While the genomes of B. pertussis encode the genetic material for a functional flagellar apparatus, existing dogma and the stop codon in flhA, which would be expected to preclude expression of FlhA, have been major disincentives to investigate motility in this species. The ability of B. pertussis to express flagellum-like structures raises an important question: how does B. pertussis overcome this apparent impediment in order to make functional flagella? In some bacteria, there are mechanisms for "antitermination" (bypassing the stop codon) (22). Alternatively, BP2261 (BcrD [putative type III secretion apparatus protein]), which has sequence homology to FlhA of B. pertussis (55% homologous) and FlhA of P. aeruginosa (59% homologous), is encoded in the B. pertussis genome. It is possible that BcrD can substitute for FlhA, enabling B. pertussis to form a functional flagellum. Future studies should explore the possible roles of alternative mechanisms to enable motility.
Another explanation for the inconsistency of motility in B. pertussis may be a low efficiency of FlhA (or a substitute) in transporting flagellar components. The predicted stop codon in the flhA gene is located at base 1313, potentially yielding an FlhA lacking the C-terminal domain, which in other bacterial species is involved in the export process (23). A ΔflhA Salmonella mutant, complemented with FlhA lacking the C-terminal domain, did not assemble a functional flagellum on its surface. However, when the ΔflhA bacteria were complemented with flhA lacking only certain portions of the C-terminal domain, this resulted in complementation at extended incubation times, suggesting the C-terminal domain is necessary for efficient flagellar assembly (24). B. pertussis FlhA may lack only a portion of the C-terminal domain, resulting in inefficient export and flagellar assembly.
These data do not address the relevance of flagellar expression or motility for virulence and pathogenicity, due to these phenotypes occurring in the Bvg(Ϫ) phase. However, Karataev et al. and Medkova et al. have shown recently that Bvg(Ϫ) organisms are present in the upper respiratory tracts of infected humans and mice (10,25). Furthermore, expression of flagellar genes has been demonstrated in vivo in mice. van Beek et al. and Wong et al. have identified, by microarray and transcriptome sequencing (RNA-seq), flagellar gene transcripts from the mouse respiratory tract (9,16). These data demonstrating that B. pertussis can express a flagellum-like structure and be motile, coupled with observation of Bvg(Ϫ) organisms and flagellar genes in vivo, should prompt the exploration of B. pertussis motility and the mechanisms that govern flagellar expression.
For detailed methods, see Text S1 in the supplemental material.