Residues of PomA mutations are involved in stator-rotor interaction involved in motility. A previous report showed that PomA-F92 and L93 of the stator are located very close to the C-terminal region of FliG in the rotor and suggested that these residues were involved in the interaction between the stator and rotor in addition to the conserved charged residues 14. These residues were highly conserved to hydrophobic residues among many species in multiple sequence alignments of PomA and MotA, implying functional significance of hydrophobicity at this position (Fig. S1). Furthermore, PomA-L95 is also highly conserved in leucine (or the similar residue of isoleucine) among many species and is located next to PomA-E96, which is a conserved and important charged residue for motor function. Thus, we substituted these residues with alanine, glutamate, or arginine, and characterized the motility of cells expressing the mutant PomA. We first examined the expansion of cells in a soft agar plate (Fig. 1A). The substitutions of PomA-F92 did not affect motility, suggesting that hydrophobicity at this position is not involved in motility. On the other hand, the glutamate-substituted mutant at the position of PomA-L93 severely decreased the swimming ring in the soft agar plate, and the arginine-substituted mutant moderately decreased the swimming ring, suggesting that the charged side chain at this position affects motility. In the substituted mutants at the position of PomA-L95, the alanine and phenylalanine mutants retained motility, whereas the threonine mutant moderately inhibited motility, and the arginine and glutamate mutants severely decreased the swimming ring in the soft agar plate. These results suggest that motility is affected by the charged or hydrophilic side chains at this position.
We next examined the protein expression levels of the mutant PomA using immunoblotting for anti-PomA and anti-PomB antibodies (Fig. 1B). All the mutated PomA proteins were expressed at a similar level as the wild-type PomA produced from plasmid pHFAB, although the band intensities of L95E, L95R, and L95T were reduced, suggesting that the motility defects of the mutants in the soft agar plate were not due to the decrease in the expression level of the mutant proteins.
We examined the effects of temperature on motility. Cells treated with wild-type PomA, L93E, or L95R were incubated at 20°C, 30°C and 40°C in a soft agar plate (Fig. S2). Wild-type PomA conferred a larger swimming ring at 40°C than at 30°C. On the other hand, the two mutants conferred a smaller swimming ring at 40°C than at 30°C. At 20°C, the swimming abilities were much worse than at 40°C and 30°C. The relative swimming abilities of wild-type PomA and mutant PomA were similar under each temperature condition. These results implied that the mutant PomA protein is more sensitive to high temperatures than the wild-type protein.
PomA-L93 and L95 mutants conferred a rotationally direction-biased phenotype. We next characterized the motility of the cells expressing PomA-L93E, L93R, L95E, or L95R, as these mutants severely affected motility in the soft agar plate. We captured cell motion under the microscope, and then analyzed motility properties: motile fraction, swimming speed, switching frequency, and ratio of rotational direction (CCW/CW). Although the mutants showed a decreased motile fraction compared to wild-type PomA, the mutants sufficiently retained their swimming ability in the liquid (Fig. 2A). Swimming speed of the mutants was not significantly different from that of wild-type PomA (Fig. 2B). On the other hand, the ratio of the rotational direction of PomA-L93E, L93R, or L95R, but not L95E, was significantly biased in the counter-clockwise direction compared to that of wild-type PomA (Fig. 2C). Moreover, the switching frequency of all mutants was significantly decreased compared to that of wild-type PomA (Fig. 2D). Therefore, these results indicated that these mutations to the position of the stator A subunit confer the Che− phenotype, showing defects in the rotational direction and switching frequency. However, the che function was not completely lost by the mutations because the response of phenol to the CW rotation of flagella remained active (Fig. 3). Since all mutations in the stator proteins described previously have been Mot− phenotype but not Che− phenotype, we found novel mutations in the stator protein conferring the Che− phenotype. This result implies that the structural changes in the stator affect the switching of the rotational direction.
PomA CCW mutants combined with a CW-biased FliG mutant. PomA L93 and L95 mutants, as described above, exhibited motility properties biased toward the CCW direction. Next, we examined whether PomA mutations are dominant for switching the rotational direction in combination with CW-locked FliG-G215A mutation or CW-biased FliG-Q147H mutation. PomA L93E or L95R combined with FliG-G215A showed a CW-locked phenotype similar to wild-type PomA with FliG-G215A (Fig. 4), suggesting that the switching of the rotational direction is predominantly determined by the rotor. Wild-type PomA combined with FliG-Q147H showed the CW biased rotation with CCW:CW = 1:9, but a few switching events occurred. In contrast, PomA-L93E or L95R combined with FliG-Q147H showed more CW-biased rotation and a much lower switching frequency than wild-type PomA with FliG-Q147H (Fig. 4). This result suggests that the mutation at the position of PomA L93 or L95 biases the rotational direction determined by the rotor and inhibits the switching of the rotational direction.
Swimming ring-restored mutants on soft agar plate. Next, we isolated mutants that restored the swimming ring in a soft agar plate from the V. alginolyticus pomAB null mutant cells expressing PomA-L93E, L95E or L95R, and wild-type PomB. We obtained a total of 23 mutants: 10 mutants from PomA L93E (named L93Es1-L93Es10), nine mutants PomA-L95E (named L93Es1-L95Es9), and four mutants PomA L95R (named L95Rs1-L95Rs4) (Fig. S3). When the plasmid extracted from the swimming ring-restored mutants was re-transformed into the parent strain, the swimming ring of the re-transformants, except for L95Rre3, returned to the same level as that of the original cell (Fig. S3). This result indicated that a mutation exists in the genome, but not in the plasmid. Next, we examined the motility of swimming ring-restored mutants (Fig. S4-S6). Most of the mutants did not exhibit a significant difference from the parent strain on the motile fraction, swimming speed, switching frequency, and ratio of rotational direction. The swimming speed of L93Es3 and L95Es6-8 was significantly faster than that of the parent strain (Fig. S4). L95Rs3 and L95Rs4 restored the ratio of the rotational direction, which was similar to that of the cells expressing wild-type PomA (Fig. S5). The mutants, L93Es6, L95Es3, L95Es9, L95Rs3, and L95Rs4, showed significantly restored switching frequency, whereas some mutants, L93Es2, L93Es4, and L95Es8, experienced a decreased frequency (Fig. S6). L95Rs3 possessed an L95S mutation in pomA encoded in the plasmid pHFAB.
Identification of mutation positions by whole genome sequencing. To identify the mutations of the swimming ring-restored mutants, we determined the whole genome sequences of the mutants by short-read sequencing. Using Breseq, we identified suppressor mutations in the 19 mutants (Table S2). In L93Es1, L93Es3, L93Es5-10, and L95Rs3-4, candidate mutations exist in genes encoding proteins involved in ATP synthase production. In L95Es1-3 and L95Rs1-2, candidate mutations exist in genes encoding ATPase proteins of the type II secretion system. In L95Es6, candidate mutations exist in genes encoding transketolase 2. In L95Es7, candidate mutations exist in genes encoding ATP-binding proteins in the hypothetical ABC transporter. These genes may be involved in the switching events of the stator and rotor interactions.