Three motAB Stator Gene Products in Bdellovibrio bacteriovorus Contribute to Motility of a Single Flagellum during Predatory and Prey-Independent Growth

Abstract

The predatory bacterium Bdellovibrio bacteriovorus uses flagellar motility to locate regions rich in Gram-negative prey bacteria, colliding and attaching to prey and then ceasing flagellar motility. Prey are then invaded to form a “bdelloplast” in a type IV pilus-dependent process, and prey contents are digested, allowing Bdellovibrio growth and septation. After septation, Bdellovibrio flagellar motility resumes inside the prey bdelloplast prior to its lysis and escape of Bdellovibrio progeny. Bdellovibrio can also grow slowly outside prey as long flagellate host-independent (HI) cells, cultured on peptone-rich media. The B. bacteriovorus HD100 genome encodes three pairs of MotAB flagellar motor proteins, each of which could potentially form an inner membrane ion channel, interact with the FliG flagellar rotor ring, and produce flagellar rotation. In 2004, Flannagan and coworkers (R. S. Flannagan, M. A. Valvano, and S. F. Koval, Microbiology 150:649-656, 2004) used antisense RNA and green fluorescent protein (GFP) expression to downregulate a single Bdellovibrio motA gene and reported slowed release from the bdelloplast and altered motility of the progeny. Here we inactivated each pair of motAB genes and found that each pair contributes to motility, both predatorily, inside the bdelloplast and during HI growth; however, each pair was dispensable, and deletion of no pair abolished motility totally. Driving-ion studies with phenamil, carbonyl cyanide m-chlorophenylhydrazone (CCCP), and different pH and sodium conditions indicated that all Mot pairs are proton driven, although the sequence similarities of each Mot pair suggests that some may originate from halophilic species. Thus, Bdellovibrio is a “dedicated motorist,” retaining and expressing three pairs of mot genes.

Bdellovibrio bacteriovorus is a small, predatory deltaproteobacterium found ubiquitously in nature (30). Bdellovibrio preys upon a wide variety of Gram-negative bacteria, including many human, animal, and plant pathogens. It has a biphasic predatory life cycle, alternating between a highly motile “attack phase” and a sessile intracellular growth phase (15). Bdellovibrio swims, using a single polar flagellum and chemotaxis, to prey-rich regions before colliding with, attaching to, and entering a suitable prey cell (10, 13, 16, 26, 27). The bdellovibrio squeezes through a small pore in the prey outer membrane (6) and upon entry to the prey cell periplasm sheds its flagellum and seals the pore in the prey outer membrane, forming a “bdelloplast.” It then begins to digest the prey cell cytoplasm, using the broken-down contents to grow into an elongated growth-phase cell which, upon exhaustion of the prey cell cytoplasmic contents, septates into multiple progeny. The progeny then become flagellate, and there are previous reports (8, 28) of flagellum-mediated motility within the remnants of the bdelloplast immediately prior to lysis of the prey outer membrane. After the bdelloplast is lysed, the progeny Bdellovibrio bacteria are released as highly motile attack-phase cells, seeking more prey. A small percentage of Bdellovibrio cells in a population can also grow host independently (HI) in rich media, with HI cells being morphologically diverse but usually flagellate (3, 5). Previous work has shown that while flagellum-mediated motility is not required for prey entry, it is vital for efficient prey location and thus predation in liquid environments (10, 13).

The bacterial flagellum is a rigid helical propeller that is rotated from a membrane-localized motor complex and composed of more than 20 different structural proteins. Bacterial flagella are typically rotated by multiple transmembrane MotAB protein complexes that are conformationally altered as ions flow through them down an ion motive gradient, which is maintained by the electron transport system of the bacterial cytoplasmic membrane. The conformational alterations in MotAB proteins act upon the FliG rotor proteins to cause rotation of the MS ring, rod, hook, and filament and hence to cause swimming (19).

The best-studied flagellar rotor/stator systems in Gram-negative bacteria are those in Salmonella enterica serovar Typhimurium and lab strains of Escherichia coli, both of which contain a single pair of motAB genes Recently there have been interesting studies reporting single flagella being driven by multiple sets of stator proteins; these include the PomAB/MotAB stators in Shewanella oneidensis (20) and the MotAB/MotCD stators in Pseudomonas aeruginosa (29). Moreover, genome sequencing has revealed multiple copies of stator genes, showing bacteria with two copies of motAB (including the gammaproteobacterium Yersinia pestis and the betaproteobacterium Burkholderia cepacia), one copy of motAB and one copy of motCD (various Xanthomonas species and the betaproteobacteria Chromobacterium violaceum), and three copies of motAB and one copy of pomAB (the deltaproteobacterium Desulfovibrio vulgaris) (29). In Pseudomonas aeruginosa deletion of a single pair of stator protein genes (either motAB or motCD) did not reduce swimming speed, while deleting both motAB and motCD resulted in cells that were unable to swim (29). That study did show that deletion of motCD (but not motAB) resulted in a reduction in swarming motility and suggested that the two pairs of stator proteins may be added to the flagella motor in response to the need for either swimming or swarming motility. An insightful study of the MotAB and PomAB stators in Shewanella oneidensis showed that a single polar flagellum can be powered by a hybrid motor containing stators powered by both hydrogen and sodium ions (20). The authors showed that both stator complexes contribute to swimming motility and suggested that the Mot stators benefit from the presence of the Pom system. They showed that hybrid motA pomB/pomA motB stators were nonfunctional but that hybrid motors containing a mix of both types of stators were indeed functional and suggested that the ratios of each type of stator within the motor alter depending upon the environmental conditions.

We have previously shown that there is extensive duplication of flagellar propeller genes in B. bacteriovorus and that flagellar motility and chemotaxis are important for predatory encounters with prey (13, 16). The role of a single MotA stator protein in the predatory life cycle of B. bacteriovorus strain 109J was previously studied by Flannagan et al. (8), who reported that downregulation of expression of a single motA gene, by antisense RNA expression from a plasmid, in B. bacteriovorus strain 109J delayed the escape of progeny Bdellovibrio cells from the bdelloplast. They also reported an altered “slow-tumbling” swimming style of the progeny Bdellovibrio cells and morphological abnormalities of the bdelloplasts.

Subsequent genome sequencing and analysis of a related strain of B. bacteriovorus, type strain HD100 (2, 21), showed there to be three sets of motAB genes, Bd0144-Bd0145, Bd3021-Bd3020, and Bd3254-Bd3253, along with gene duplication of many other vital components of the flagellar structure, including six copies of the filament subunit gene fliC (10, 13) and two copies of the rotor component gene fliG (14), although not enough duplicated components to encode a fully alternate flagellar structure. The Bd0144 motA gene of B. bacteriovorus HD100 showed 100% identity, at DNA and protein levels, to the motA gene in B. bacteriovorus 109J described by Flannagan et al. (8), and there was also synteny with its surrounding genetic locus, including motB, between the two strains.

We monitored gene transcription and constructed paired mot gene deletions to examine the roles of each pair of motAB stator gene products during both predatory and HI growth of B. bacteriovorus HD100, and we found that while each contributed to flagellum-mediated motility, no single pair of proteins was essential. Thus, the wild-type B. bacteriovorus flagellum may be powered by a hybrid motor consisting of all three pairs of stator proteins and driven by hydrogen ions. Flagellum-mediated motility occurred within the remnants of the bdelloplast immediately prior to lysis, and deletion strains showed that this was not dependent on a single pair of MotAB proteins. This study further emphasizes the importance of flagellum-mediated motility and a strategy of motility gene duplication to Bdellovibrio survival and highlights the versatility of the bacterial flagellar motor in accommodating multiple proteins fulfilling the same role.

MATERIALS AND METHODS

Bacterial strains and Bdellovibrio culturing.

The bacterial strains used in this study are listed in Table 1. Host-dependent (HD) Bdellovibrio was cultured as described previously (13, 16) using E. coli S17-1 as prey, which contained the plasmid pZMR100 when appropriate to confer resistance to kanamycin. HD Bdellovibrio was coincubated with E. coli S17-1 in Ca-HEPES buffer at 29°C with shaking at 200 rpm. Host-independent (HI) Bdellovibrio was isolated and grown as previously described (13) in PY broth (with shaking) or on PY plates (containing kanamycin at 50 μg/ml for selection when needed) at 29°C.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description	Reference
Strains
E. coli
S17-1	thi pro hsdR hsdM+recA; integrated plasmid RP4-Tc::Mu-Kn::Tn7; used as donor for conjugating plasmids into Bdellovibrio	23
DH5α	F′ endA1 hsdR17 (rK− mK−) supE44 thi-1 recA1 gyrA (Nalr) relA1 Δ(lacIZYA-argF)U169 deoR[φ80dlacΔ(lacZ)M15]; used as a cloning host strain	9
S17-1(pZMR100)	S17-1 containing pZMR100 plasmid used to confer Kmr; used as Kmr prey for Bdellovibrio	22
Bdellovibrio bacteriovorus
HD100	Type strain, genome sequenced	21, 25
HID13	Host-independent derivative of HD100	12
fliC1merodiploid	Merodiploid HD100 with wild-type and kanamycin-interrupted fliC1(Bd0604)	C. Lambert, unpublished data
fliC1merodiploid HI	Host-independent derivative of fliC1 merodiploid	A. Fenton, unpublished data
motAB1 mutant	HD100 motAB1::Kmr	This study
motAB1 HI mutant	Host-independent derivative of motAB1 mutant	This study
motAB2 mutant	HD100 motAB2::Kmr	This study
motAB2 HI mutant	Host-independent derivative of motAB2 mutant	This study
motAB3 HI mutant	Host-independent HD100 derivative with motAB3::Kmr	This study
motAB3 mutant	Host-dependent derivative of motAB3 HI mutant	This study
Plasmids
pUC19	Ampr cloning vector	32
pUC4K	pUC vector with Kmr cassette	32
pSET151	Suicide vector used for conjugation and recombination into Bdellovibrio genome	4
pSETmotAB1	pSET151 suicide plasmid containing motAB1::aphII	This study
pSETmotAB2	pSET151 suicide plasmid containing motAB2::aphII	This study
pSETmotAB3	pSET151 suicide plasmid containing motAB3::aphII	This study

RT-PCR for detection of expression of motAB genes.

Wild-type B. bacteriovorus HD100 predatory-cycle RNA and host-independent HID13 RNA were isolated as described previously (6, 13). For comparisons of attack-phase gene expression in the motAB deletion strains, RNA samples were matched to 10 ng/μl as described previously (12). Reverse transcription-PCR (RT-PCR) was carried out (using 30 amplification cycles) as described in reference 6 using the following primer pairs: for motA1, 5′ CCACACCGAAGAAGAAGAGC 3′ and 5′ ACGGCGTTTCAGTTTGTTTC 3′; for motA2, 5′ ACGGGATCGTGCTTGTAATC 3′ and 5′ AGGGCATTTGTCAGAACGTC 3′; for motA3, 5′ AGCGTACGGTCTGATTGGTC 3′ and 5′ GACACACACAATCGGAGGTG 3′; for motB1, 5′ GTGCGCTATCTGGTGAAGGT 3′ and 5′ ACCTTGGAACTGTCGGTGAC 3′; for motB2, 5′ AGCGGTGGAAGTGAAAGAGA 3′ and 5′ GAGGATGACTGGATCCGAAA 3′; and for motB3, 5′ GGATGGCAGTGTGATGTCTG 3′ and 5′ GATCAATTTCACCGCGTTTT 3′. Ten microliters of the reaction mixture was then run on a 2% agarose gel.

Insertional inactivation of motAB genes.

Each pair of motAB genes was amplified with flanking DNA using KOD high-fidelity DNA polymerase (Novagen) and the following primers: for motAB1, 5′ ATCAAGGATGAGCTCGCACAG 3′ and 5′ AAGTGCGCGAGCTCTGCGCCGAGAT 3′; for motAB2, 5′ GTCCGCTCTAGATGAAATCCA 3′ and 5′ GGATTTCTAGAGCAATTACGG 3′; and for motAB3, 5′ ACGCCTGATAAGTGAGCTCCA 3′ and 5′ TGAAGAATTCATGGATCTCGCGG 3′. The resulting products were gel purified and digested with the appropriate restriction enzymes, unique sites for which were encompassed in the cloning primers (motAB1, SacI; motAB2, XbaI; and motAB3, EcoRI and SacI). These were cloned into the vector pUC19 (31) cut with the same enzymes. A large portion of each pair of motAB genes was removed by restriction digests of the pUC19 clones (motAB1 [BamHI], bp 348 of motA to bp 615 of motB; motAB2 [PstI], bp 104 of motA to bp 781 of motB; and motAB3 [HindIII and StuI and then blunted using cloned Pfu polymerase {Stratagene}], bp 266 of motA to bp 275 of motB) and replaced with a 1.3-kb kanamycin resistance cassette released from the vector pUC4K (32) using appropriate restriction sites (motAB1, BamHI; motAB2, PstI; and motAB3, HincII) (summarized in Fig. 1). These constructs were then PCR amplified before being ligated into the conjugative suicide vector pSET151 (4), which was cut with BamHI and blunt ended. The pSET151-based constructs were conjugated into B. bacteriovorus HD100 as described previously (16).

FIG. 1.

Construction of the insertion/deletion in each of the three pairs of motAB genes in the B. bacteriovorus HD100 genome. Arrows show the primer binding sites used for PCR amplification. The deleted sections of the motAB genes are between the two annotated restriction sites, while the inserted 1.3-kb kanamycin resistance cassette (aphII) is also shown.

Gene deletions and replacements were confirmed by Southern blotting (24) using probes for the kanamycin cassette (to confirm presence), for the pSET151 vector (to confirm absence), and for the respective pair of motAB genes with terminal flanking DNA (to confirm correct chromosomal location). They were further confirmed by PCR amplification of the genes of interest using a proofreading polymerase and direct sequencing of the PCR product by MWG Biotech Ltd. Sequencing data were then analyzed to confirm insertion of the kanamycin cassette at the expected site and deletion of the relevant portions of the genes.

Electron microscopy.

The flagellar morphology of each host-dependent Bdellovibrio motAB mutant was checked using transmission electron microscopy, although from other bacterial studies it was not expected that motAB gene mutagenesis would alter flagellation, as motors are added after flagellar synthesis. Fifteen microliters of an attack-phase Bdellovibrio culture was applied to a carbon-Formvar grid (Agar Scientific) for 5 min. Cells were stained with 15 μl of 0.5% uranyl acetate (URA) (pH 4.0) for 1 min. Representative cells were imaged using a JEOL JEM1010 electron microscope.

Light microscopy.

Cultures were visualized using a Nikon Eclipse E600 epifluorescence microscope with a 100× phase-contrast objective, and images and videos were taken with a Hamamatsu Orca ER camera using the SimplePCI software (version 5.3.1; Hamamatsu).

Assay of relative predatory efficiencies of motAB mutant Bdellovibrio strains on E. coli.

Predatory (HD), attack-phase Bdellovibrio cells are too small to diffract light and, as such, do not give a reading of optical density at 600 nm (OD₆₀₀); thus, OD₆₀₀ readings for predatory cultures indicate only the number of prey cells and prey-derived bdelloplasts within cultures. In comparison, HI Bdellovibrio cells are larger and do give an OD₆₀₀ reading. Matching the numbers of attack-phase Bdellovibrio cells in cultures was achieved by using protein concentration (as determined by the Lowry protein assay [18]).

The predation efficiencies of predatory Bdellovibrio motAB mutant and wild-type strains over multiple rounds of infection were studied using cultures set up as follows. One milliliter of a fully prey-lysed Bdellovibrio predatory culture (Bdellovibrio cell numbers matched as described above) was added to 3 ml of a 16-h culture of E. coli S17-1(pZMR100) and 50 ml Ca-HEPES buffer to give a typical predator/prey ratio of 1:20. The cultures were incubated at 29°C with shaking at 200 rpm, and the OD₆₀₀ of the prey E. coli cells was measured at the start and then every hour between 12 and 20 h, during which time the greatest drop in prey OD was observed. For predation efficiency studies, during a single round of infection, a predatory culture was set up using a method similar to that for synchronous prey infection described previously (7). In short, Bdellovibrio cells were concentrated 20-fold and matched to the fliC1 merodiploid control by protein concentration as determined by the Lowry assay (18), E. coli S17-1(pZMR100) was diluted to an OD₆₀₀ of 1.0, and 4 ml of concentrated Bdellovibrio was added to 3 ml of diluted E. coli and 5 ml of fresh Ca-HEPES to give a final starting ratio of at least five Bdellovibrio cells per E. coli prey cell. Cultures were analyzed at each time point using light microscopy.

Testing Bdellovibrio motility within bdelloplasts.

To visualize motility of Bdellovibrio postseptation but still within bdelloplasts, predatory cultures were set up using cephalexin-treated E. coli, which produced larger bdelloplasts when infected by Bdellovibrio than untreated cells. Exponential-phase E. coli S17-1(pZMR100) cells were grown for 16 h to an OD₆₀₀ of 2.86, diluted 1:100 in YT broth plus kanamycin, incubated for 3.25 h until they reached an OD₆₀₀ of 1.26, and then treated with cephalexin at a final concentration of 60 μg/ml for 90 min, reaching a typical final OD₆₀₀ of 2.09; they were then washed in Ca-HEPES buffer before addition to Bdellovibrio. Two milliliters of a 16-h Bdellovibrio culture, which had fully lysed prey, had 300 μl of cephalexin-treated, washed E. coli added. These cultures were then incubated for 4 h before being visualized using light microscopy, and individual bdelloplasts with postseptation swimming progeny Bdellovibrio were examined using time-lapse photomicroscopy.

Growth rate and motility of HI motAB mutant Bdellovibrio.

Growth rates of HI Bdellovibrio strains were measured by optical density at 600 nm (OD₆₀₀), using a Fluostar Optima plate reader (BMG Labtech). HI Bdellovibrio was pregrown in PY broth plus kanamycin and immediately prior to the experiment was diluted in fresh PY plus kanamycin to an OD₆₀₀ reading of 0.1 unit; 265 μl of the diluted cultures was then added to each well of a 96-well Optiplate (Porvair Sciences Ltd.) and the plate sealed with Breathe-Easy (Web Scientific) gas-permeable sealing membrane. The plate was then incubated in the Fluostar plate reader at 29°C with shaking at 200 rpm for 48 h, with an OD₆₀₀ reading every 30 min (as HI-grown Bdellovibrio cells are large enough to give an OD₆₀₀ value). To analyze motility during growth, an identical plate was set up and incubated in a standard aerobic incubator under the same conditions, and samples were taken at the appropriate times for light microscopy.

Hobson BacTracker analysis of Bdellovibrio motAB mutant swimming behavior.

The swimming motility of each host-dependent Bdellovibrio motAB mutant strain was compared with that of the fliC1 merodiploid strain (wild type for motility and predation) using Hobson BacTracker (Hobson Tracking Systems, Sheffield, United Kingdom) analysis of mean run speeds (using the same settings as described in reference 13, with the addition of a minimum run speed threshold of 15 μm/s to reduce the influence of Brownian motion and tethered rotation on the results). Bdellovibrio was grown in overnight predatory lysates allowing multiple rounds of infection, and the resultant attack-phase progeny after prey lysis were tracked at 22 h after inoculation. This was to allow all strains to have completed lysis of all bdelloplasts when the Bdellovibrio motility was monitored (for example, in Fig. 5b, predation is seen to be complete [and the OD₆₀₀ of prey to be flat] for fliC1 merodiploid control motAB1 and motAB2 mutants and nearly complete for the motAB3 mutant by 20 h). To test the effects of different driving ions or inhibitors, immediately prior to tracking, Bdellovibrio cells were diluted 3-fold in fresh Ca-HEPES (pH 7.6) containing NaCl, phenamil (Sigma), or carbonyl cyanide 3-chlorophenylhydrazone (CCCP) (Sigma) as required to give the appropriate final concentrations or with Ca-HEPES at the higher pH of 8.8 to give a final testing pH of 8.2. Data on mean run speeds were collected from 20 sets of 50 tracks for each strain under each condition tested, except for experiments involving CCCP, for which 10 sets of 25 tracks were collected, and for the motAB3 strain, for which 20 sets of 25 tracks were collected, due to the reduced percent motility of the population.

The percent motility of cultures was assayed by visualizing a minimum of 10 fields of view, containing at least 100 attack-phase Bdellovibrio cells per field of view, and comparing manually the number of cells swimming and the number that were nonmotile. Tracking software cannot detect these percent motility differences for cultures with many small cells in them.

RESULTS

B. bacteriovorus HD100 has three pairs of motAB genes with different phylogenetic origins.

The Bdellovibrio bacteriovorus HD100 genome has three pairs of genes encoding MotAB proteins located quite near the origin of replication on both sides of the chromosome: we have designated them Bd0144/Bd0145 (motA1/motB1), Bd3021/Bd3020 (motA2/motB2), and Bd3253/Bd3254 (motA3/motB3). In the initial annotation of each motA gene (21), they were found to encode either an extended N-terminal sequence (motA1/motA2) or a missing first transmembrane domain (motA3), but further sequence analysis revealed alternative potential start codons that rectified these issues and gave significant full-length homologies to Mot proteins of diverse other bacteria. As mentioned above MotA1 of B. bacteriovorus HD100 was a 100% identical match to the MotA of B. bacteriovorus 109J studied with antisense knockdown by Flannagan and coworkers (8). We examined the primary sequence similarities between the six Bdellovibrio MotA and MotB protein sequences in comparison to MotA and MotB from the proton-driven E. coli motor and PomA and PomB from the Vibrio alginolyticus sodium-driven motor to see if any obvious homologies suggesting a driving ion for each MotAB type could be determined. Alignment (Fig. 2a) shows that MotA3 resembles most closely the proton-driven MotA protein sequences of E. coli and S. Typhimurium, along with the P. aeruginosa MotA (as designated in reference 29), having the longer cytoplasmic loop between transmembrane regions 2 and 3 encompassed in an extra 20 amino acids found between positions 110 and 130 that is typical of MotA proteins. It is interesting to note, though, that despite its Mot-like character, MotA3 is the sole Bdellovibrio MotA which has a conserved aspartate residue (D175) at the position homologous to D148 of PomA from the sodium-driven motor of V. alginolyticus. D148 in that PomA protein is part of the phenamil binding site that blocks sodium conductance in Pom-based motors. In contrast to MotA3, MotA2 is more homologous to the “sodium stator” PomA protein sequences of V. alginolyticus and V. parahaemolyticus, having a shorter cytoplasmic loop between transmembrane regions 2 and 3; MotA1 is intermediate between the other two Bdellovibrio MotAs, having a shorter cytoplasmic loop between transmembrane domains 2 and 3, similar to that seen in the PomA sequence of V. alginolyticus.

FIG. 2.

(a) Amino acid alignment of the MotA proteins from Bdellovibrio bacteriovorus HD100 (MotA1 = Bd0144, MotA2 = Bd3020, and MotA3 = Bd3254) with E. coli K-12 MG1655 MotA (AAC74959.1) and Vibrio alginolyticus PomA (AB004068). Sequences were aligned using ClustalW2 at http://www.ebi.ac.uk/Tools/clustalw2/index.html. The boxes highlight the four transmembrane regions (as identified in E. coli). (b) Amino acid alignment of the MotB proteins from Bdellovibrio bacteriovorus HD100 (BdmotB1 = Bd0145, BdmotB2 = Bd3019, and BdmotB3 = Bd3253) with E. coli K-12 MG1655 MotB (AAC74959.1) and Vibrio alginolyticus PomB (AB004068). Sequences were aligned using ClustalW2 at http://www.ebi.ac.uk/Tools/clustalw2/index.html. The vertical box highlights the conserved D32 residue essential for function of all Mot/PomBs, while the first horizontal box indicates the single transmembrane domain and the second horizontal box the peptidoglycan binding region.

Comparison of the Bdellovibrio MotB protein sequences with the proton-driven stator protein MotB from E. coli and the sodium-driven stator protein PomB from V. alginolyticus was inconclusive, with the single transmembrane domain well conserved between all sequences, and while the peptidoglycan binding domain was more variable in each, there was still a great deal of homology between sequences (Fig. 2b). Most variation lies between these two domains, in the periplasm-spanning region, with MotB1 being of similar length to E. coli MotB and both MotB2 and MotB3 of a longer length like that found in PomB of V. alginolyticus. Because of the inconclusive nature of these alignment results, we first tested each pair of genes for cross-complementation of E. coli motAB-defective mutants, but we found that despite an almost identical codon usage, none of the genes complemented the E. coli mot mutants (data not shown). Thus, we decided to examine the activity of the motors produced in Bdellovibrio by deletion of each single pair of motAB genes, monitoring motility phenotypes in both predatory and HI lifestyles, and third, we tested the ion specificity of the residual motors after deletion by addition of proton motive force and sodium conductance inhibitors CCCP and phenamil in the presence or absence of diverse sodium and pH conditions.

All six stator genes are expressed during the Bdellovibrio life cycles, with motAB1 being upregulated toward the end of the predatory cycle.

Reverse transcription-PCR (RT-PCR) (Fig. 3) showed that each motA and motB gene was expressed by both HD and HI wild-type B. bacteriovorus HD100. Expression of each of motA2, motB2, motA3, and motB3 was constitutive across the wild-type predatory cycle, while both motA1 and motB1 were upregulated at 3 and 4 h postinfection, at which point the wild-type Bdellovibrio cells were septating, producing flagella, and being released from the bdelloplast. The expression of the fliC3 flagellin gene provided a control for the synchrony of infection of the wild-type Bdellovibrio culture on the prey; this is because we have previously shown that at 45 min and 1 h of infection, fliC gene transcription and flagellar filament synthesis cease while all the Bdellovibrio are growing intraperiplasmically (10). Interestingly, in contrast, the stator mot transcripts are seen at all time points throughout the life cycle.

FIG. 3.

Agarose gel electrophoresis of RT-PCR products of each motA and motB gene on RNA isolated from a synchronous predatory culture of B. bacteriovorus HD100 preying upon E. coli S17-1. Lanes:1, attack-phase wild-type HD100; 2, 15 min postinfection; 3, 30 min postinfection; 4, 45 min postinfection; 5, 1 h postinfection; 6, 2 h postinfection; 7, 3 h postinfection; 8, 4 h postinfection; 9, HID13; 10, no-template negative control; 11, E. coli S17-1 only; 12, wild-type HD100 genomic DNA positive control.

Each pair of motAB genes can be individually inactivated without abolishing either HD Bdellovibrio motility and predation or HI growth and motility.

Each pair of motAB genes was individually inactivated by deletion of extensive portions of motAB genes and replacement (Fig. 1) with a kanamycin resistance cassette. Both motAB1 and motAB2 deletion strains were readily obtained in predatory cultures as host-dependent (HD) strains (using methods as described in reference 13). However, despite repeated attempts, a motAB3 deletion strain could be obtained directly only as a host-independent (HI) strain, not directly in predatory culture. Despite this, when E. coli prey was added to the HI motAB3 strain, it invaded and completed a predatory cycle, showing that predatory capability was retained, but less successfully than for the motAB1 motAB2 mutants. A motAB3 HD strain was obtained from a single (pure) plaque on an E. coli overlay plate after incubation of the HI motAB3 derivative with prey, and this strain was used in further characterizations in comparison to the predatory motAB1 and motAB2 mutant strains.

The cell lengths of this motAB3 HD strain were slightly greater than that of the wild type, but this is typical of HD strains that are derived from HI strains by prey challenge. As each motAB deletion strain was kanamycin resistant, a fliC1 merodiploid strain was used as a wild-type control. This strain contained both the original fliC1 gene and a kanamycin resistance cassette-disrupted gene and is a “reconstituted” wild-type strain for predation, growth rate, and motility. Electron microscopy (Fig. 4) showed that the cells of each motAB HD strain had a single flagellum of wild-type length and waveform; thus, the motAB deletion did not affect flagellar production or morphology.

FIG. 4.

Electron microscopic images of the HD100 ΔmotAB mutants. (a) fliC1 merodiploid wild-type control; (b) motAB1 mutant; (c) motAB2 mutant; (d) motAB3 mutant. Cells were stained with 0.5% uranyl acetate, pH 4. Scale bars show 1 μm.

Axenic growth (Fig. 5A) and motility were assayed for HI derivatives of the fliC1 merodiploid control, the motAB1 and motAB2 strains produced as described previously (17), and the original motAB3 HI strain. The axenic HI growth rate of each motAB mutant in PY medium was similar to that of the fliC1 merodiploid control, although each motAB deletion strain reached a higher final optical density. Light microscopy revealed that a percentage of each culture was motile at each time point tested (after 6, 10, 25, and 31 h of incubation, representing each stage of growth [lag, exponential, stationary, and decline phases, respectively]), although the actual percent motility ranged from 1% to 30% in the fliC1 merodiploid control and from 1% to 10% in each of the motAB deletion strains. Thus, while motility was not abolished by deletion of a single pair of motAB genes, the percent motility was decreased in all cases compared to that of a wild-type control.

FIG. 5.

HI and predatory growth of each motAB deletion strain alongside a fliC1 merodiploid (wild-type) control. (a) HI growth was measured by optical density at 600 nm (in milli-OD units) from a starting OD₆₀₀ of 0.1 for each culture. Points represent the means for 24 individual samples, and error bars represent the 95% confidence interval around the mean. (b) Predation efficiencies of HD Bdellovibrio motAB deletion mutants. The bdelloplast lysis rate of each strain as measured using optical density (OD₆₀₀) is shown; a drop in OD corresponds to lysis of prey cells.

As Flannagan et al. (8) reported that downregulation of a motA1 homologue in B. bacteriovorus 109J resulted in significantly diminished predation due to delayed escape from bdelloplasts, causing persistence of the downregulated motA1 strain inside bdelloplasts for 30 h or more, we studied motility of and predation by HD mutant strains. Motility analysis with a Hobson BacTracker (Table 2) showed that deletion of motAB1 or motAB2 resulted in reduced percent motility for the culture, but not reduced swimming speeds compared to wild-type levels, but that deletion of motAB3 resulted in significant reductions in both percent motility (from 98% for the wild-type fliC1 merodiploid to 25% for the motAB3 mutant) and swimming speed (from 63 μm/s for the wild type to 27 μm/s for the motAB3 mutant). We have previously shown (13) that slower-swimming Bdellovibrio cells have reduced predation efficiency as they collide with prey less frequently.

TABLE 2.

Motile percentage of population and swimming speed of attack-phase cells of each Bdellovibrio motAB mutant in comparison to that of the fliC1 merodiploid (wild type for motility) in Ca-HEPES buffer, pH 7.6

Strain	Motile fraction (%)	Mean swimming speed (μm/s) ± SD
fliC1 merodiploid	>98	63.2 ± 5.5
motAB1 mutant	75	66.1 ± 3.8
motAB2 mutant	50	63.6 ± 5.3
motAB3 mutant	25	26.5 ± 1.8

Twenty-four-hour E. coli predation assays consisting of multiple rounds of infection (Fig. 5b), with an average starting predator/prey ratio of 1:20, revealed that the predation efficiencies of the motAB1 and motAB2 mutants were not significantly different from that of the fliC1 merodiploid control in this experiment. The motAB1 strain seemed to have a small reduction in predation at between 12 and 15 h, but the value was not statistically different from that for the control. The motAB3 mutant, with comparable starting numbers of viable, predatory Bdellovibrio cells, was significantly slower at predation, reaching completion of prey lysis only after 19 h of incubation in comparison to 14 to 15 h for the wild-type fliC1 merodiploid control in a typical 50-ml predatory culture, but the motAB3 mutant gave a yield of predatory Bdellovibrio comparable to that for the control when assayed by viable count (PFU) after 24 h of incubation (between 1 × 10⁸ and 2 × 10⁸ PFU/ml). Both the motAB1 and motAB2 strains were identical to the wild-type fliC1 merodiploid in Bdellovibrio yield from predatory cultures.

Prey entry and flagellum-mediated movement within bdelloplasts.

As we had found that contrary to the results of Flannagan and coworkers, (8), the B. bacteriovorus motAB1 strain was not significantly slower at predation during multiple rounds of infection, a single round of infection by each of the motAB1 and motAB2 strains was compared to that by the fliC1 merodiploid control by light microscopy in a “bdelloplast persistence assay.” Analysis of these single rounds of infection (see Fig. S1 in the supplemental material), using a starting ratio of a minimum of five predatory Bdellovibrio cells to one prey E. coli cell, showed that the motAB1 mutant was slower than either the fliC1 merodiploid control or the motAB2 mutant at lysing the bdelloplast. Entry was seen to occur at the same rate for all three strains (fliC1 merodiploid, motAB1, and motAB2), and while all bdelloplasts were lysed at between 4 and 6 h for the fliC1 and motAB2 samples, bdelloplasts were seen in the motAB1 culture until after greater than 8 h of incubation. However, an increase of free-swimming Bdellovibrio cells was seen in the motAB1 infection culture after 6 h, suggesting that some bdelloplast lysis had occurred, liberating the motile progeny. This would account for the small, not significant reduction in predation efficiency seen in the 24-h experiment described above.

Bdellovibrio cells have been observed by ourselves and others (28; R. E. Sockett lab, unpublished observations) to swim within bdelloplasts after septation is complete and before the bdelloplast membrane lyses. To assess whether any individual pair of MotAB proteins was responsible for this motility, late-stage bdelloplasts, made from Bdellovibrio-infected filamentous E. coli prey, were visualized using phase-contrast microscopy, and videos of flagellum-mediated swimming motility within bdelloplasts were taken (see Videos S1 to S4 in the supplemental material). Swimming in bdelloplasts postseptation and immediately prior to bdelloplast lysis was seen for the fliC1 merodiploid control and for all three motAB deletion strains, showing that no single pair of MotAB proteins is required for swimming under bdelloplast conditions.

Expression of motAB gene pairs is changed by the deletion of other motAB gene pairs.

We monitored the level of transcription of each of the mot genes in matched total RNA concentrations from attack-phase cells of wild-type and mot mutant backgrounds and found that there was a significant downregulation of motA3 and motB3 gene expression in both the motAB1 and motAB2 mutants compared the wild type (Fig. 6, lanes 2 and 3 compared to lanes 1 for the motA3 and motB3 primers) but that this was not the case for the expression of motAB1 or motAB2. (It was not possible to compare expression levels across primer sets as they were of different designs and hybridization stringencies, nor was it possible to match RNA concentrations and repeat the experiment across the predatory cycle because of contaminating prey RNA concentrations varying.) The expression of fliC1 was constant across the mutants, coincident with the observation of single flagella on each strain in Fig. 4, and this was unaffected by the mot mutations.

FIG. 6.

Agarose gel electrophoresis of RT-PCR products of the three motA and motB genes on matched attack-phase RNA templates. Lanes: 0, attack-phase fliC1 merodiploid; 1, attack-phase wild-type HD100; 2, attack-phase motAB1 mutant; 3, attack-phase motAB2 mutant; 4, attack-phase motAB3 mutant; 5, no template; 6, wild-type HD100 genomic DNA positive control.

Testing driving ions for motAB mutant flagella.

Bdellovibrio species are not alkalophilic bacteria, so they would not be expected to have sodium-driven motors; however, the Mot protein sequence alignments did show some Pom-like homology, so to show whether any of the Bdellovibrio MotAB pairs might conduct sodium instead of protons, rather than being Pom-like Mot proteins that conduct protons, we tested the effects of changing the level of sodium in the medium on motility by using the ionophore CCCP to dissipate the proton motive force and used the sodium channel blocker phenamil. As Fig. 7 a shows, addition of phenamil significantly reduced the motility of the wild-type fliC1 merodiploid control strain from 90% to 50%; however, the swimming speed of the cells remaining motile was not statistically significantly reduced compared to that for the control treated with dimethyl sulfoxide (DMSO). Interestingly, for none of the other three motAB minus strains was such a drastic reduction in percent motility seen on the addition of phenamil; indeed, for the motAB1 and motAB3 mutants there was a modest increase in percent culture motility upon addition of phenamil compared to that for the DMSO control.

FIG. 7.

Bdellovibrio motility in the presence of phenamil, NaCl, and CCCP. (a) Mean swimming speed (± standard deviation) and percent motility in the presence of the sodium channel inhibitor phenamil. (b) Mean swimming speed in Ca-HEPES buffer at pH 8.2 with different NaCl concentrations. Error bars show the standard deviation around the mean. (c) Mean swimming speed (± standard deviation) and percent motility in the presence of the proton channel inhibitor CCCP.

Although phenamil is reported to be a sodium channel blocker, some non-sodium channel effects have been reported for it, so we also tested the mean run speeds of the strains in concentrations of sodium chloride from 0 to 50 mM at a constant alkaline pH of 8.2 (Fig. 7b). We saw no significant difference in swimming speed for any of the strains, although there was a small, nonsignificant rise in swimming speed for the fliC1 merodiploid control only at 5 mM NaCl; however, there was no increase in percent motility of the strain after the addition of 5 mM NaCl. We also tested the effect of the presence or absence of 5 mM NaCl on predation efficiency but found that it had no effect on any of the strains (data not shown). We found (Fig. 7c) that addition of the proton ionophore CCCP to all of the strains abolished motility at a final concentration of 5 μM and reduced motility at 2 μM, suggesting that flagellar motility is driven by the proton motive force. Addition of 25 mM NaCl did not, however, significantly relieve the motility drop (in terms of speed or percent motility of the culture) as CCCP was added, indicating that there was no sufficient sodium motive force driving the flagella that could substitute for the removal of the proton motive force.

DISCUSSION

Deletion of any pair of the three motAB genes of B. bacteriovorus HD100 did not abolish flagellar motility, and all three motAB single mutant strains retained predatory competence, although the percent motility and speed of a motAB3 Bd3253/Bd3254 deletion strain was significantly reduced compared to those of wild-type and motAB1 and motAB2 mutant strains. This may suggest that conductance through a stator containing MotA3 and/or MotB3 causes rapid swimming of the attack-phase Bdellovibrio, which we studied in our experiments, and thus that the MotAB3 proteins are important to attack-phase swimming by B. bacteriovorus HD100. Under our experimental conditions, Bdellovibrio cells were swimming at average speeds of 70 to 75 μm s⁻¹, although we have measured individual single cells swimming much faster than this at different times after release from prey. That the motAB3 mutation has the strongest detrimental effect on speed and percent motility may suggest that one or more of the MotAB3 proteins are more abundant in the wild-type stator, possibly during the attack phase, or that one of the MotAB3 proteins has a packing/scaffolding role that most efficiently assembles the other Mot proteins in the stator. Interestingly, all motAB mutants and the control strain were motile both outside and inside the bdelloplast and during HI axenic growth, and thus we conclude that none of the gene pairs are retained to allow a specific growth mode of Bdellovibrio to occur. We suggest that possibly the three motAB gene pair copies are selected for because their products ensure that predatory and prey-independent growth modes are always possible, even if natural mutations were to damage one mot gene pair. This allows Bdellovibrio to retain growth mode flexibility under conditions of changing prey availability for predation versus organic matter availability for HI growth.

Flannagan and coworkers (8) had previously shown that downregulation of the motA1 homologue in Bdellovibrio bacteriovorus strain 109J resulted in slow tumbling motility of the bacteria and long-delayed escape from E. coli bdelloplasts (with turbidity values of the E. coli bdelloplasts containing the Bdellovibrio with the downregulated motA1 gene resembling that of uninfected E. coli prey for 30 h of infection), implicating motAB1 as being important in successful lysis of prey and release of progeny B. bacteriovorus 109J. We found that deletion of the motAB1 Bd0144 and -5 genes from B. bacteriovorus HD100 did not alter the speed or motile behavior of free progeny Bdellovibrio, nor did it prevent predation occurring efficiently, but we did find that there was a delay of approximately 2 h in the bdelloplast release time in motAB1 mutants compared to controls. Thus, we agree with Flannagan and coworkers that deletion of motA1 in Bdellovibrio does slow predator release from prey, but this is a minor effect in our HD100 strain compared to their 109J strain (8). This may imply that MotAB1 proteins are a more important or abundant component of the flagellar motor than MotAB2 and -3 when the flagellum is synthesized in dividing Bdellovibrio cells inside the bdelloplast.

The tumbling motility observed and the extreme effect on prey lysis that Flannagan and coworkers reported for the motA1 downregulated strain may have been due to strain differences or to the plasmid present in the 109J strain, as the authors themselves discuss (8). As genome sequence and mot gene expression data are not available for 109J, we cannot say whether these differences were due to there being a different supporting complement of other motAB genes in the B. bacteriovorus 109J genome compared to that of HD100. However, our RT-PCR primers for each HD100 mot gene (used for Fig. 3 and 6) do amplify a band from genomic DNA of strain 109J that is similar in size to that seen for HD100 (data not shown). However, to test the roles of each MotAB protein pair for predatory and HI growth in a strain where a full genome sequence was available (21), we continued to study the role of each motAB gene pair by deletion mutagenesis in strain HD100.

The finding that each HI motAB mutant strain reached a higher final optical density when growing axenically than did the fliC1 merodiploid control (Fig. 5a) suggested that possibly the loss of any single pair of MotAB proteins may increase the proton flux through the ATP synthase, as less flux is being diverted through the missing MotAB protein pair that is not expressed. This would, however, imply that all three MotAB proteins normally contribute to the stator of the Bdellovibrio flagellum during HI growth as well as HD growth. This suggestion was supported by the transcriptional expression studies in Fig. 3, which showed consistent HD and HI expression of all motAB genes.

There was not a significant difference between the predatory degradation of prey by the fliC1 merodiploid “reconstituted” Kn^r wild-type control and the motAB1 and motAB2 HD strains (Fig. 5b) after 20 h of incubation, and there was only a small difference in the HD motAB3 strain derived from the original motAB3 HI mutant; this suggested that even slower motility was sufficient for the cells to collide with prey and enter them under our lab conditions. However, the fact that we could not isolate the motAB3 mutant originally as a predatory plaque on soft agar overlays of E. coli lawns shows that the predatory ability of Bdellovibrio in soft agar is more affected by mutations which slow swimming than is predation in liquid cultures.

The differing sequence similarities between each MotAB protein pair and the Mot and Pom proteins of other bacteria and the lack of extensive identity between any two Bdellovibrio MotA or MotB proteins show that they may have been anciently acquired by lateral gene transfer from different bacteria. That they are all transcribed throughout the predatory cycle and during HI growth shows that they are beneficial to fitness. In Table 2 and Fig. 7a and 7c the different speed and percent motility responses of each of the mutants lacking a single MotAB pair suggest that motors containing different pairs of MotAB proteins do have slightly different rotational characteristics. The percent motility of cultures of the wild-type control and fliC1 merodiploid strains dropped more in the presence of 50 μM phenamil than did that any of the single mutants (with the motAB3 percent motility actually rising in response to phenamil). This may suggest that a cross-combination of MotA and MotB proteins produced across different mot gene clusters may produce a phenamil-sensitive site in wild-type motors. This is expected from work with alkalophilic bacteria which showed that mutations in both the MotA and MotB equivalent proteins together are required for phenamil resistance, implying that phenamil binds across the MotAB proteins (11). It is interesting to note from the alignment of Bdellovibrio Mot proteins that solely MotA3 has a conserved amino acid, D175, with the D148 of Vibrio alginolyticus PomA which is implicated in the phenamil binding region; however, none of the Bdellovibrio MotB proteins have the equivalent residue to Vibrio alginolyticus PomB P16, which was proposed to be in the vicinity of the other part of the phenamil binding site in Na⁺-driven motors (11). It is also possible that only in the wild-type motor are all the Mot proteins present in a particular functional conductive conformation and therefore are more susceptible to channel blockage as part of the wild-type stator, rather than in the conformation in mutant stators where their conformation is altered by missing another Mot protein.

The phenamil inhibition may be indicative not of there being any Na⁺ conductance through Bdellovibrio motors but rather of the ancient lateral gene transfer of pom genes into Bdellovibrio from an alkalophilic bacterium and their modification through natural selection to conduct protons. For instance, the regions in the MotAB proteins of the solely H⁺-conducting motor of Rhodobacter sphaeroides corresponding to those that bind phenamil in the PomAB proteins of the Na⁺-driven motor of Vibrio alginolyticus were shown by Asai and coworkers, in a hybrid MomB protein coupled with a Rhodobacter MotA protein, to confer phenamil sensitivity and thus binding despite the motor conducting protons (1).

The observation of phenamil inhibition of a proton-driven motor in another bacterium fits with the lack of a significant change in motility in increasing sodium concentrations at pH 8.2 (Fig. 7b) and with the CCCP sensitivity of motility, unrelieved by sodium (Fig. 7c). Thus, we conclude that the flagellum of Bdellovibrio is powered, in both predatory HD and in HI growth modes, by a complex derived from a functionally redundant set of six MotAB proteins and that it is driven by the proton motive force. Selective deletion of each of the six mot genes in combination was beyond the scope of this study and is a technical challenge in Bdellovibrio, but our gene expression and single motAB deletion studies have shown that some or all of the products of each of the three motAB gene clusters contribute to flagellar rotation in predatory and HI growth.

Predatory growth does require active location of prey by motility and taxis and fast motility to bring Bdellovibrio near to prey surfaces, so flagellar motility is “important” to Bdellovibrio. We have previously shown that there is 6-fold duplication of flagellar filament fliC genes in B. bacteriovorus HD100 and that all of these FliC proteins contribute to the expressed flagellar filament structure (10, 13). Our current work on Mot proteins further emphasizes that Bdellovibrio is a “dedicated motorist” in which the need to be motile, to be able to use the large chromosomal complement of predatory invasion genes to access an intrabacterial niche where feeding and growth are without competition, results in the retention of several active copies of flagellar motor protein genes. This may ensure that random natural mutagenic events do not cause it to “break down” and no longer be able to benefit from predatory invasion to grow.