Characterization of FliL Proteins in Bradyrhizobium diazoefficiens: Lateral FliL Supports Swimming Motility, and Subpolar FliL Modulates the Lateral Flagellar System

Bacterial motility propelled by flagella is an important trait in most environments, where microorganisms must explore the habitat toward beneficial resources and evade toxins. Most bacterial species have a unique flagellar system, but a few species possess two different flagellar systems in the same cell. An example is Bradyrhizobium diazoefficiens, the N₂-fixing symbiont of soybean, which uses both systems for swimming. Among the less-characterized flagellar proteins is FliL, a protein typically associated with a flagellum-driven surface-based collective motion called swarming. By using deletion mutants in each flagellar system’s fliL, we observed that one of them (lateral) was required for swimming, while the other (subpolar) took part in the control of lateral flagellum synthesis. Hence, this protein seems to participate in the coordination of activity and production of both flagellar systems.

ABSTRACT

Bradyrhizobium diazoefficiens is a soil alphaproteobacterium that possesses two evolutionarily distinct flagellar systems, a constitutive subpolar flagellum and inducible lateral flagella that, depending on the carbon source, may be expressed simultaneously in liquid medium and used interactively for swimming. In each system, more than 30 genes encode the flagellar proteins, most of which are well characterized. Among the exceptions is FliL, which has been scarcely studied in alphaproteobacteria and whose function in other bacterial classes is somewhat controversial. Because each B. diazoefficiens flagellar system contains its own fliL paralog, we obtained the respective deletions ΔfliL_S (subpolar) and ΔfliL_L (lateral) to study their functions in swimming. We determined that FliL_L was essential for lateral flagellum-driven motility. FliL_S was dispensable for swimming in either liquid or semisolid medium; however, it was found to play a crucial role in upregulation of the lateral flagellum regulon under conditions of increased viscosity/flagellar load. Therefore, although FliL_S seems to be not essential for swimming, it may participate in a mechanosensor complex that controls lateral flagellum induction.

IMPORTANCE Bacterial motility propelled by flagella is an important trait in most environments, where microorganisms must explore the habitat toward beneficial resources and evade toxins. Most bacterial species have a unique flagellar system, but a few species possess two different flagellar systems in the same cell. An example is Bradyrhizobium diazoefficiens, the N₂-fixing symbiont of soybean, which uses both systems for swimming. Among the less-characterized flagellar proteins is FliL, a protein typically associated with a flagellum-driven surface-based collective motion called swarming. By using deletion mutants in each flagellar system’s fliL, we observed that one of them (lateral) was required for swimming, while the other (subpolar) took part in the control of lateral flagellum synthesis. Hence, this protein seems to participate in the coordination of activity and production of both flagellar systems.

INTRODUCTION

Bradyrhizobium diazoefficiens, the N₂-fixing symbiont of soybean, is an alphaproteobacterium that possesses two flagellar systems which evolved independently (1–3). This dual flagellar system comprises a single subpolar flagellum similar to the polar flagellum of Caulobacter crescentus and several lateral flagella similar to the flagella of Ensifer meliloti and Brucella spp. (3–5). Unlike other dual-flagellar-system bacterial species, in which the (sub)polar flagellum is used for swimming in liquid medium and the lateral flagella are used for swarming over wet surfaces (6, 7), in B. diazoefficiens, both flagellar systems are expressed simultaneously by planktonic cells in liquid medium (2, 4, 8, 9), a property that this bacterium shares only with Shewanella putrefaciens (10) among the bacterial species characterized to date. Thus, swimming behavior of B. diazoefficiens is an emergent property that results from interaction between both flagellar systems, which might enable these bacteria to better explore the complex soil pore space (2). Although under laboratory conditions B. diazoefficiens performs swarming motility, the lateral flagella are not strictly required for this phenotype, which seems unnecessary for displacement into the soil (11). Moreover, the high cell numbers required for swarming are seldom reached in the soil pore space, and thus swarming motility seems to be an unimportant trait for free-living B. diazoefficiens in its natural habitat formed by the soil pore space, the rhizosphere, and the root surfaces (2).

In B. diazoefficiens planktonic cells, the subpolar flagellum is expressed constitutively under the conditions explored so far, while the expression of lateral flagella, which is controlled by the class IA regulator regR and the class IB regulator lafR, requires certain carbon sources, oxic conditions, viscosity, or swimming in semisolid 0.3% agar medium (4, 5, 11–13). The presence of lateral flagella in B. diazoefficiens cells that swim near surfaces seemingly prevents premature binding to the surface, which might allow swimming for longer distances than those possible for cells with only subpolar flagella (2). This property might be advantageous for dispersal into soil pores near the root surfaces in search for infectible root cells. Preliminary results also indicated that lateral flagella are better adapted for swimming in viscous medium (2).

Bacterial flagella are complex structures with the following three main components: the basal body inserted into the membrane system, which contains the flagellar motor and the export apparatus for secretion of the extracellular flagellar components; the flexible connector hook; and the helical filament that projects out of the cell and is several times longer than the cell body (see reference 5 for a scheme). Flagella are thus composed of many different structural proteins, with the functions of most of them known (14, 15). The flagellar motor consists of a rotor composed by a ring formed by the FliF, FliG, FliM, and FliN proteins that is powered by the proton (or sodium ion) motive force against stators formed by MotA and MotB (or PomA and PomB), which produces torque that is transmitted through the basal body to the hook and, finally, to the filament (15). The rotational motion converts the filament into a screw that propels the water, thus moving the bacterial cell, which, given its small dimensions, moves at a low Reynolds number and therefore lacks inertia (16). The rotational flagellar filament thrusts or drags the cell body, allowing bacteria to swim at speeds on the order of tens of cell body lengths per second.

Among the less-characterized proteins of the flagellar basal body is FliL, which has been investigated in several bacterial species, with most studies carried out in the Gammaproteobacteria species Escherichia coli, Salmonella enterica, Vibrio alginolyticus, and Proteus mirabilis. Meanwhile, research on the Alphaproteobacteria, to which B. diazoefficiens belongs, is scarce, and, in particular, there are no reports concerning FliL in the group of N₂-fixing rhizobia. This protein was observed by cryo-electron tomography as lying between rotor and stator structures in Borrelia burgdorferi (17). The motor of this spirochete also possesses a particular structure, called the collar, to which FliL is associated (18). Furthermore, biochemical and X-ray crystallography studies in V. alginolyticus suggested that rings of FliL might associate around stator units (19, 20). From these and other studies, it may be concluded that FliL is a small membrane protein that is part of the flagellar basal body and that might have a role in stabilizing the stator (21). Hence, when swimming bacteria encounter a surface or when medium viscosity increases, the rotation of the flagellar motor may be impaired, and the flagellar basal body may be subjected to torsional stress. It has been reported that FliL may allow the basal body to withstand this torsional stress by reinforcing the rod, stabilizing the stators and contributing to torque generation (22–25). In E. coli, sensing of the external load seems to be performed by a component of the flagellar motor (26), and in P. mirabilis the C-terminal portion of FliL was suggested as responsible for sensing the presence of viscosity and surfaces (27–29). However, in E. coli it was reported that, for some minutes, torque generated by the flagellar motors at different loads was independent of FliL (30). Other roles ascribed to FliL are an involvement in the energetics of the flagellar function in P. mirabilis (31), the opening of the motor proton channel in Rhodobacter sphaeroides (32), the orientation of periplasmic flagella in B. burgdorferi (17), and the motor rotation in C. crescentus (33).

In addition to the paucity of studies on the role of FliL in alphaproteobacteria, there are no comparative works in bacterial species possessing two flagellar systems except for recent studies in V. alginolyticus (19, 20). Since B. diazoefficiens uses its two flagellar systems for swimming (2), this species brings an opportunity to compare the role of FliL in each system for swimming. Therefore, we addressed this question here by obtaining and characterizing deletion mutants in the fliL genes of each flagellar system in B. diazoefficiens. We observed different impacts of the mutations in each fliL paralog and obtained evidence indicating that inactivation of fliL in the subpolar flagellar system modifies the expression of the lateral flagellar master regulator.

RESULTS AND DISCUSSION

Characterization of FliL_S and FliL_L.

The open reading frames (ORFs) bll5826 and bll6868 are annotated as fliL in the B. diazoefficiens USDA 110 genomic sequence (34). According to their genomic contexts (see Fig. S1A in the supplemental material), bll5826 encodes the FliL protein of the subpolar flagellar system, while bll6868 would be the FliL of the lateral flagellar system. Therefore, we will refer to these genes as fliL_S and fliL_L, respectively. fliL_S is the first gene of a putative operon consisting of 5 genes related to the subpolar flagellum, while fliL_L is localized next to the last gene in operon II of the lateral flagellar system (5). The alignment of FliL_S and FliL_L amino acid sequences indicated a low relationship, with 19.8% identity and 40% similarity between them (Fig. S1B), in agreement with their pertaining to evolutionary different flagellar systems (3). To further characterize their relationship with other FliL proteins, we constructed the cladogram shown in Fig. 1, and we found that FliL distribution was similar to that previously described for other flagellar proteins (2). In particular, FliL_S and FliL_L were grouped in separate clades, where FliL_S was related to homologs from C. crescentus and the primary system of Rhodopseudomonas palustris, while FliL_L grouped with those of Tardiphaga sp. and the secondary system of R. palustris. Moreover, FliL from E. meliloti, Bacillus subtilis, B. burgdorferi, and the secondary FliL from R. sphaeroides were more related to FliL_L than to FliL_S, whereas FliL from gammaproteobacteria and the primary system of R. sphaeroides were unrelated to both B. diazoefficiens FliL. Interestingly, B. diazoefficiens FliL_S tightly grouped in the same clade with those FliL of the subpolar flagellum from Bradyrhizobium species that possess two flagellar systems, while FliL from species like Bradyrhizobium elkanii, which possess a single (subpolar) flagellar system, were grouped in a separate clade (Fig. 1). In V. alginolyticus, FliL appears to be in contact with the MotB (PomB) of the stator complex in the periplasmic space (19), whereby some complementarity between the structural domains of both proteins is expected. Hence, the above results may indicate some structural difference between subpolar flagellar stators in the groups of Bradyrhizobium species that possess single or dual flagellar systems.

FIG 1.

Maximum-likelihood cladogram of FliL from Bradyrhizobium spp. and related species. In cases where two flagellar systems were identified in the genome sequence, FliL paralogs similar to the B. diazoefficiens USDA 110 subpolar system were indicated as S, and those similar to the B. diazoefficiens USDA 110 lateral system were indicated as L. In species where two flagellar systems were identified but their genomic sequences did not allow their assignations as subpolar or lateral, the FliL paralogs were indicated as FliL1 or FliL2. Blue rectangles highlight strains belonging to the B. japonicum phylogroup 1, while the pink rectangle highlights strains from other Bradyrhizobium spp. phylogroups, as described elsewhere (3). Next to the nodes, the percentages of replicate trees in which the associated taxa clustered together in the bootstrap test with 1,000 replicates are shown. Branch lengths in the tree are scaled in the same units as those of the evolutionary distances (scale bar) used to infer the phylogenetic tree.

The domain prediction of these polypeptides indicated that, as described for other FliL proteins, both homologs have a single transmembrane domain. For FliL_S, most of the algorithms employed predict the 23 N-terminal residues as a cytoplasmic domain followed by a 20-residue transmembrane domain and then predict a 123-residue C-terminal domain within the periplasm. Intriguingly, the consensus prediction for FliL_L is 3 N-terminal residues at the periplasm followed by the 20-residue transmembrane domain and then a 125-residue C-terminal domain in the cytoplasm (Fig. S1D). However, the probability of the C-terminal domain being cytoplasmic is only slightly higher than that of it being periplasmic. This domain distribution does not agree with previous studies of FliL and in particular with the proposed interaction of FliL with the stator complex in the periplasmic side (24). Therefore, we will not assume this domain disposition for FliL_L until experimental confirmation.

In searching for additional evidence supporting the roles of FliL_S and FliL_L, we intended to complement a well-known FliL defect. In enteric bacteria such as E. coli and Salmonella spp., inactivation of FliL leads to a small defect in swimming but produces a substantial reduction in swarming (23). To see whether the B. diazoefficiens fliL paralogs are able to complement this mutation in E. coli, we cloned the wild-type (WT) fliL_S and fliL_L genes from B. diazoefficiens USDA 110 into the pBAD24 vector and transformed an E. coli MG1655 ΔfliL mutant with these plasmids, along with the empty vector as a negative control. In agreement with previous reports (23), swimming was marginally affected in the MG1655 ΔfliL mutant, while swarming was strongly reduced. B. diazoefficiens fliL_L partially complemented these defects in the MG1655 ΔfliL mutant (Fig. S2), indicating that, despite its low sequence relatedness and atypical domains prediction, FliL_L may be considered a functional FliL protein. However, B. diazoefficiens fliL_S did not complement the E. coli ΔfliL mutation (Fig. S2). To seek an explanation of this difference in complementation, we compared the B. diazoefficiens subpolar and lateral stator sequences with that of E. coli. There are motAB and pomAB stator genes annotated in the clusters of the B. diazoefficiens subpolar flagellar system, but whether one or both classes of stators are used by this bacterium is unknown. Therefore, we compared both sets of stator proteins with E. coli MotAB and found that MotAB and PomAB of the subpolar system were unrelated to E. coli MotAB (Fig. S3). In contrast, the ORFs annotated as MotA and MotB in the lateral flagellar genes cluster were related to those of E. coli MotAB, in the same branch as those of Aeromonas hydrophila and Pseudomonas aeruginosa (Fig. S3). Thus, the lack of complementation of the E. coli mutation by the wild-type fliL_S might be due to insufficient fit between FliL_S and the E. coli stators, whereas FliL_L might adjust better. Therefore, these results are not enough to rule out FliL functionality of FliL_S in B. diazoefficiens.

Swimming speed was differentially affected by mutations in fliL_S or fliL_L.

We obtained deletion mutants in each B. diazoefficiens fliL paralog by removing internal fragments from their coding sequences (Fig. S1A and C). Since both genes were located in operons, we performed unmarked in-frame deletion mutants to avoid possible polar effects. We observed that the ΔfliL_S and ΔfliL_L mutants produced the same subpolar and lateral flagellins as the wild type, indicating that they possessed both types of flagella; however, the level of LafA relative to FliC was higher in the ΔfliL_S mutant (Fig. 2). In mutants possessing only one flagellar system, neither fliL deletion provoked changes in the remaining flagellin (Fig. 2A). In addition, both mutants responded to the carbon source as previously observed (5, 11, 12); the lateral flagellins were not produced by bacteria grown in HM salts-yeast extract (HMY) with d-mannitol as a carbon source and were produced by bacteria grown in HMY with l-arabinose (Fig. 2B).

FIG 2.

Extracellular proteins from B. diazoefficiens USDA 110 and their fliL mutant relatives in liquid medium. (A) B. diazoefficiens wild-type (WT), ΔfliL_S, ΔfliL_L, ΔlafA, ΔfliC, ΔfliL_S ΔlafA, and ΔfliL_L ΔfliC strains grown in liquid HMY medium with l-arabinose as the carbon source. (B) B. diazoefficiens WT, ΔfliL_S, and ΔfliL_L strains grown in liquid HMY medium with d-mannitol or l-arabinose as the carbon source. The extracellular proteins were separated by SDS-PAGE and stained with Coomassie brilliant blue. The positions of the flagellins FliC and LafA (subpolar and lateral, respectively) are indicated at the sides of the images.

Swimming speed in liquid HMY medium with l-arabinose was affected by these mutations, although the effects were different for each single-mutant strain. As shown in Table 1, we corroborated in the background of USDA 110 the swimming speed values previously reported for the LP 3004 wild type and two flagellin mutants, a ΔlafA mutant, lacking the lateral flagellins, and a ΔfliC mutant, lacking the subpolar flagellins (2). In addition, deletion in fliL_S led to only a 29% reduction in swimming speed in both a ΔfliL_S single mutant possessing both flagella and a ΔfliL_S ΔlafA double mutant unable to produce lateral flagellar filaments, indicating that in B. diazoefficiens FliL_S is not essential for subpolar flagellar function in swimming. These results are in contrast to previous observations in C. crescentus—an alphaproteobacterium that has a flagellar system closely related to the subpolar flagellar system of B. diazoefficiens—in which FliL was reported as essential for flagellar motor rotation (33).

TABLE 1.

Swimming speed of B. diazoefficiens wild-type and mutant strains in HMY-arabinose medium^a

Strain description	Speed (μm · s−1) ± SDc	n
WT	27.59 ± 0.54 A	160
ΔlafA	30.44 ± 0.67 A	118
ΔfliC	17.21 ± 0.25 B	77
ΔfliLS	19.63 ± 0.91 B	91
ΔfliLS ΔlafA	21.65 ± 0.37 B	120
ΔfliLL	28.65 ± 0.26 A	104
ΔfliLL ΔfliC	0b

^aMovies of swimming cells were recorded at 30 frames/s and the speed was measured using Move-tr/2D software.

^bNo data are presented for the ΔfliL_L ΔfliC strain because this mutant was nonmotile.

^cThe data represent the mean speed of n swimming cells ± standard deviation (SD). Values followed by different letters were statistically different according to analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test with α = 0.005.

The deletion in fliL_L did not produce a motility reduction in cells possessing both flagella; however, when ΔfliL_L was combined with a ΔfliC mutation, (i.e., lacking subpolar flagellar filaments), cells were nonmotile (Table 1). These results indicate a moderate requirement of FliL_S for swimming with subpolar flagellum and an essential role of FliL_L for swimming with lateral flagella.

In E. coli, S. enterica, P. mirabilis, and V. alginolyticus, FliL was reported as required for swimming in viscous environments, including swarming behavior, which might be related to a role for this protein in flagellar stabilization (21–24, 27). Therefore, we investigated the effects of deletions in both fliL paralogs when medium viscosity was increased by the addition of polyvinylpyrrolidone (PVP). At a low PVP concentration (1% wt/vol), swimming speeds of ΔfliL_S (with both flagella) and ΔfliL_S ΔlafA (with only subpolar flagellum) mutants were only slightly reduced compared to those in the medium without PVP. When PVP concentrations rose to 2% wt/vol or more, swimming speeds decreased steadily in all strains, showing again the difficulty of the subpolar flagellum in propelling the cells in a viscous environment (Fig. 3A). However, ΔfliL_L mutant swimming speed was not reduced and remained comparable to that of the wild type at low PVP concentration (1% wt/vol) and was significantly reduced with PVP above 2% wt/vol (Fig. 3B), showing the requirement of this protein by the lateral flagella in viscous media.

FIG 3.

Effect of viscosity on B. diazoefficiens swimming speed. Cells were grown in HMY-arabinose medium and diluted in HM salts containing different concentrations of polyvinylpyrrolidone (PVP). Swimming speeds of at least 60 cells from two biological replicates were recorded with Move-tr/2D. Data points are mean ± 99% confidence intervals calculated with Student’s t test. (A) Comparison of ΔfliL_S against wild-type fliL_S in the background of the wild-type cells that possess both flagellar systems or in the background of the ΔlafA mutant, which lacks lateral flagellar filaments. (B) Comparison of ΔfliL_L against wild-type fliL_L in the background of the wild-type cells that possess both flagellar systems or in the background of the ΔfliC mutant, which lacks subpolar flagellar filaments. The ΔfliL_L ΔfliC double mutant did not swim under any condition assayed. All the strains were assayed at the same time and presented in separate charts for clarity.

Swimming in semisolid medium.

Swimming behavior may be evaluated in semisolid medium containing 0.3% agar in which bacteria have to swim while migrating through the agar mesh (35). In this environment, B. diazoefficiens produces both subpolar and lateral flagella, even when d-mannitol is the only carbon source (2). We studied swimming motility in Götz-mannitol semisolid medium, which is a minimal medium developed for studies of motility in rhizobia and is well suited for this kind of study with B. diazoefficiens (9, 36). Although the ΔfliL_L mutant was less motile, ΔfliL_S motility haloes were more expanded than those of the wild type (Fig. 4). When we added PVP at two concentrations to obtain middle (2% wt/vol) and high (5% wt/vol) viscosity, we observed contrasting effects of each fliL mutation (Fig. 5). In agreement with the changes in swimming speed, the motility haloes of the ΔfliL_S mutant in the background of the wild type (with both flagella) decreased when PVP increased at 2% and 5% (wt/vol). In turn, the ΔfliL_S ΔlafA mutant (with only subpolar flagellum) produced motility haloes of reduced size in semisolid medium with 2% (wt/vol) PVP and did not move in semisolid medium with 5% (wt/vol) PVP (Fig. 5A). Moreover, motilities of the ΔfliL_S ΔlafA double mutant and the ΔlafA single mutant were similar in viscous medium (2% and 5% PVP), indicating that the lack of lateral flagellins was epistatic on the fliL_S mutation. Regarding the ΔfliL_L strain, we observed that its motility diminished with the increase in viscosity in the wild-type background, but the ΔfliL_L ΔfliC double mutant was nonmotile under all conditions (Fig. 5B), indicating an essential role of FliL_L for the activity of lateral flagella in viscous semisolid medium. In agreement with previous results (2), the motility of the ΔfliC single mutant, which has only lateral flagella, was unaffected by viscosity.

FIG 4.

Swimming motility of B. diazoefficiens USDA 110 in semisolid Götz minimal medium with d-mannitol as the carbon source. Cells were inoculated at the center of 0.3% (wt/vol) agar plates and incubated at 28°C for the indicated times. (A) Diameter of the swimming halo of wild-type (WT; black symbols), ΔfliL_S (gray symbols), and ΔfliL_L (white symbols) strains. The complete data set from two (ΔfliL_S strain) or three (WT and ΔfliL_L strains) independent experiments with three or four technical replicates is shown. Each experiment is distinguished by a different symbol. Regression lines for each set are as follows: WT, continuous line (r² = 0.89); ΔfliL_S strain, long-dashed line (r² = 0.95); and ΔfliL_L strain, short-dashed line (r² = 0.90). (B) Representative 90-mm plate after 8 days of incubation.

FIG 5.

Swimming motility of B. diazoefficiens USDA 110 and its mutant relatives in semisolid Götz minimal medium with d-mannitol as the carbon source, supplemented with polyvinylpyrrolidone (PVP). (A) Effects of the ΔfliL_S deletion. (B) Effects of the ΔfliL_L deletion. The nonmotile ΔlafA ΔfliC mutant, lacking all flagellins, was included as negative control. The bacteria were inoculated on Götz-mannitol soft agar (0.3% wt/vol) with the indicated PVP concentrations and incubated at 28°C for 8 days. Averages ± SD are plotted. Data are representative of two biological replicates, with five technical replicates per experiment. Significant differences were estimated by analysis of variance (ANOVA); *, α = 0.05; ***, α = 0.001; ns, nonsignificant. Stars indicate absence of movement.

Complementation of fliL mutants.

To assess whether fliL_S and fliL_L deletions may be complemented in trans by the wild-type alleles, both full-length fliL sequences were cloned in replicative vectors and introduced into each strain, with empty vectors used as controls. We performed complementation experiments in semisolid medium under the conditions where each mutant showed the most differential phenotype, namely, Götz-mannitol for the ΔfliL_S mutant and the same medium supplemented with 5% PVP for the ΔfliL_L mutant. The results of complementation experiments are shown in Fig. 6.

FIG 6.

Complementation of fliL mutations in B. diazoefficiens USDA 110 for swimming motility in semisolid Götz minimal medium with d-mannitol as the sole carbon source. (A) Complementation of ΔfliL_S by the wild-type fliL_S carried in the replicative vector pBBR1MCS2 (pBBR::fliL_S) in 0.3% (wt/vol) agar. (B) Complementation of ΔfliL_L by the wild-type fliL_L carried in the replicative vector pFAJ1708 (pFAJ::fliL_L) in 0.3% (wt/vol) agar with the addition of 5% (wt/vol) PVP. The empty vectors, abbreviated pBBR2 and pFAJ, respectively, were used as controls. Swimming motility was quantified by the diameter of haloes produced 8 days after inoculation at 28°C. Data are averages from three biological replicates with 6 technical replicates. Statistical analysis was carried out by ANOVA. ***, α = 0.001; ns, nonsignificant. The star indicates absence of movement.

ΔfliL_S mutants in the background that possessed both flagellar systems produced haloes slightly wider than those of the wild type carrying either empty pBBR1MCS2 (referred to as pBBR2 in Fig. 6) or pBBR2::fliL_S, albeit without statistically significant differences (Fig. 6A, left four bars). These effects might be related to some instability caused by overexpression of the proteins from the vector. In addition, the in-frame fliL_S deletion generated for this study left a coding sequence remnant that might produce a peptide of 39 amino acid (aa) residues. This putative peptide possesses an 8-aa stretch at its C terminus that might interfere at least partially with proper association of FliL_S subunits translated from pBBR2::fliL_S in unbalanced stoichiometry (Fig. S1C). Such interference might preclude the expected suppression of increased lateral flagellum-mediated swimming. In bacteria lacking lateral flagella, the partial loss of motility of the ΔfliL_S ΔlafA double mutant with respect to the ΔlafA mutant was suppressed by the full-length fliL_S introduced in pBBR2::fliL_S, in comparison with the isogenic ΔfliL_S ΔlafA double mutant carrying the empty vector (Fig. 6A). Since this double mutant was constructed by replacing lafA12 with ΔlafA into the same ΔfliL_S background used above, successful complementation in the ΔlafA background suggests that lack of differences in the WT background was not due to a polar effect of the ΔfliL_S mutation.

Regarding the lateral flagellar system, ectopic expression of fliL_L in the wild-type background somewhat diminished swimming motility with respect to the wild type carrying empty vector. Nevertheless, the low-motility phenotype of the ΔfliL_L mutant was completely reverted by the wild-type allele (Fig. 6B). As observed before, the ΔfliL_L ΔfliC double mutant, which does not possess functional subpolar flagella, was nonmotile. Introduction of the wild-type copy of fliL_L restored motility to this strain, indicating that lack of motility of the ΔfliL_L ΔfliC double mutant was due solely to ΔfliL_L deletion.

Lateral flagellar gene expression increased in the ΔfliL_S mutant.

Given the role of lateral flagella for swimming in semisolid agar and in viscous medium, we considered the possibility that dysfunction of fliL_S provokes an induction of the lateral flagellar system expression. In other species, it is known that the flagella may act as a mechanosensor, and the C-terminal domain of FliL was implicated as responsible for sensing the presence of surfaces to trigger differentiation into swarmer phenotypes (27). In addition, it was observed in Vibrio parahaemolyticus that mutation in the gene of the polar flagellin fliC provoked differentiation of planktonic cells into elongated, hyperflagellated swarmer cells in liquid medium, where these kinds of cells or changes do not otherwise occur (37). To observe if there is an induction of lateral flagellar synthesis, we performed retrotranscribed quantitative PCR (RT-qPCR) with primers directed at the class IB lateral flagellum master regulator lafR (5) with RNA obtained from wild-type and ΔfliL_S cells grown in liquid peptone-salts yeast extract (PSY)–arabinose (Ara) medium. There was a significant increase in lafR transcript in the ΔfliL_S mutant (Fig. 7A), and concomitantly, the mutant cells produced more LafA flagellins in both HMY-Ara and PSY-Ara media (Fig. 2 and 7B). We also extracted RNA from swimming haloes in semisolid PSY-Ara medium. We took cells from the whole motility halo, as well as from a region confined to the 1.5-cm exterior border of the halo, and compared the transcript amounts of lafR. We found no significant differences between the wild type and the mutant when RNAs were obtained from the whole motility halo, but when they were from the exterior border, lafR was significantly overexpressed in the ΔfliL_S mutant (Fig. 7C). We also measured transcript accumulation of fliF_L, motC, and fliL_L in the exterior border of the motility halo, because these genes are representatives of two different laf operons (5). As shown in Fig. 7D, all of these genes were overexpressed in the ΔfliL_S mutant with respect to the wild type, thus indicating that fliL_S absence induced an overproduction of lateral flagellum-related genes. In agreement with these results, the ΔfliL_S mutant cells produced many more lateral flagella at the border of the motility halo (BMH) than did the wild-type cells at this same position (Fig. 8). However, there were no significant differences in motility halo diameters between the wild type and the ΔfliL_S mutant in PSY-Ara semisolid medium. This result is expected, since it is known that swimming motility is less stimulated in rich media (35, 38). Since we could not obtain good RNA samples from swimming haloes in semisolid Götz minimal medium, we decided to corroborate the higher lateral flagellum production by the mutant in this medium by directly observing (via transmission electron microscopy) the number of lateral flagella. As shown in Fig. 8, we confirmed that the ΔfliL_S mutant produced more lateral flagella than did the wild type at the border of motility haloes in both PSY-Ara and Götz-mannitol media.

FIG 7.

Expression of lateral flagella in the wild type and ΔfliL_S mutant in PSY-arabinose. (A) lafR transcript quantification by retrotranscribed quantitative PCR (RT-qPCR) with respect to the sigA transcript as the constitutive control, with mRNA extracted from liquid cultures. (B) FliC and LafA flagellins extracted from liquid cultures. The amounts of proteins loaded in the SDS-PAGE are indicated above the lanes. (C) mRNA extracted from the border of a motility halo (BMH) or from the whole motility halo (WMH). (D) Transcript quantification by RT-qPCR of LafR target genes from BMH with respect to the sigA transcript as constitutive control. Data presented in panels A and B are the averages ± SD from three and two biological replicates, respectively, each with three technical replicates. Data in panel D are representative from two biological replicates. Statistical analyses were carried out by ANOVA. ***, α = 0.001; ns, nonsignificant.

FIG 8.

Transmission electron micrographs of individual B. diazoefficiens wild-type and ΔfliL_s mutant cells obtained from the border of the motility halo (BMH) or from the whole motility halo (WMH) from semisolid (agar 0.3%) PSY-arabinose (top) and Götz-mannitol (bottom). WT cells in BMH possessed a subpolar flagellum (arrowhead) and one or two lateral flagella (arrows), while the ΔfliL_s mutant exhibited a subpolar flagellum and more than two lateral flagella per cell, some of which are partially bundled (double arrow). In contrast, in WMH the flagellar patterns were similar between WT and mutant strains in both media. Bars, 0.2 μm. 40k, ×40 magnification; 60k, ×60 magnification.

Taken together, these results suggest that either lack of fliL_S or subpolar flagellum destabilization was a signal for lafR transcript accumulation. Therefore, there may be some cross talk between both flagellar systems that is able to regulate the production of the lateral flagella in response to the activity/stability of the subpolar flagella. This supposed cross talk should be unidirectional from the subpolar to the lateral system, since no effects on FliC synthesis or function could be observed as consequence of fliL_L alteration.

Conclusions.

Our results indicate that, in contrast to previous reports in C. crescentus, FliL_S is not essential for subpolar flagellar function in swimming but might play a role in the control of lateral flagellum synthesis both in liquid medium and in response to an increase in viscosity. Under these conditions, FliL_L seemingly performs a critical role in stabilizing lateral flagella for swimming.

The possession of two flagellar systems, however, is not a general trait of the Bradyrhizobium genus (3). B. elkanii is another important soybean symbiont that possesses only one flagellar system, similar to the B. diazoefficiens subpolar system. Furthermore, the type strain, B. elkanii USDA 76, moved equally well as B. diazoefficiens USDA 110 in semisolid medium without the addition of a viscous agent (2), and B. elkanii soil isolates were even more motile than B. diazoefficiens or Bradyrhizobium japonicum under these conditions (39), indicating that the single flagellar system of B. elkanii was enough for efficient swimming in such a tortuous environment. However, B. elkanii was unable to swim in viscous medium (2). Since the lateral flagellar system consumes a considerable proportion of metabolic energy (12) and, as inferred from phylogenetic analyses, it was acquired at the separation of the B. japonicum phylogroup from the other Bradyrhizobium phylogroups (3), we speculate that the main lateral flagellar function is allowing swimming in viscous medium, which should be a distinctive trait of the B. japonicum phylogroup, although it is not essential for swimming in general. The viscosity of soil solution might increase as a consequence of desiccation, when the concentration of solutes and colloids increase. Therefore, we hypothesize that the B. japonicum phylogroup might have evolved adaptations to drier soil environments, where the lateral flagella might play an essential role for motility under such conditions. For this adaptation, each FliL seems to play a distinct and specific role.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Strains and plasmids used in this study are listed in Table S1 in the supplemental material. B. diazoefficiens cells were stored in glycerol stocks at −20°C and, for routine use, in yeast extract-mannitol agar (YMA) (40) at 4°C. For conjugation, PSY-arabinose was used (41). If needed, chloramphenicol (Cm), tetracycline (Tc), kanamycin (Km), gentamicin (Gm), streptomycin (Sm), or spectinomycin (Sp) was added at a final concentration of 20 μg · ml⁻¹, 50 μg · ml⁻¹, 150 μg · ml⁻¹, 100 μg · ml⁻¹, 400 μg · ml⁻¹, or 200 μg · ml⁻¹, respectively. E. coli strains were grown at 30°C in LB (42). If needed, tetracycline (Tc), kanamycin (Km), gentamicin (Gm), streptomycin (Sm), spectinomycin (Sp), or ampicillin (Ap) was added at a final concentration of 10 μg · ml⁻¹, 25 μg · ml⁻¹, 10 μg · ml⁻¹, 100 μg · ml⁻¹, or 200 μg · ml⁻¹, respectively.

To visualize both kinds of flagella in liquid medium, as well as to measure swimming speed, bacteria were cultured in HMY medium (5, 43) supplemented with l-arabinose; for experiments in semisolid medium, minimal Götz medium (36) was used; to obtain high-quality RNA, PSY medium (41) supplemented with l-arabinose was used. The comparison of the three media compositions is shown in Table S2. Mutations studied in this paper did not affect growth kinetics in any of these media (Fig. S4).

Bioinformatic methods.

Sequences of the putative FliL polypeptides for the subpolar flagellum (FliL_S) and the lateral flagellum (FliL_L) were obtained from the MicrobesOnline server (44), while sequences used for phylogenetic analysis were recovered from http://img.jgi.doe.gov or https://www.ncbi.nlm.nih.gov. Protein secondary structures were predicted using Topcons (45) and TMHMM (46). Multiple sequence alignments were performed using ClustalW and MegaX software (47). The evolutionary history was inferred by using the maximum likelihood method and the Le_Gascuel_2008 model (48). Initial trees for the heuristic search were obtained automatically by applying the neighbor-joining and BioNJ algorithms to a matrix of pairwise distances estimated using the Jones-Taylor-Thornton (JTT) model, and then selecting the topology with the superior log likelihood value. A discrete gamma distribution was used to model evolutionary rate differences among sites (5 categories [+G, parameter = 6.6557]). The tree was drawn to scale, with branch lengths measured in the number of substitutions per site.

Construction of mutant and complemented strains.

The general cloning procedures were performed as described (42). Biparental matings with B. diazoefficiens were performed with E. coli S17-1, and electroporation was performed using a GenePulser instrument as previously reported (5).

DNA amplifications were carried out by PCR using Pfu polymerase (Productos Bio-Logicos, Buenos Aires, Argentina) or Taq DNA polymerase (Productos Bio-Logicos, Buenos Aires, Argentina) in a Biometra TOne thermocycler (Analytik Jena, Jena, Germany). The oligonucleotide primers used in this study were supplied by Genbiotech SRL (Buenos Aires, Argentina) and Thermo Fisher Scientific (Buenos Aires, Argentina), and are listed in Table S3. Digestions were achieved using Promega (Biodynamics SRL, Buenos Aires, Argentina) enzymes as necessary. DNA sequencing was performed by Macrogen DNA Sequencing Service (Macrogen Corp., Seoul, South Korea).

The B. diazoefficiens deletion mutants ΔfliL_S and ΔfliL_L were obtained using the unmarked in-frame deletion strategy as described previously (49). To generate ΔfliL_S a 249-bp fragment upstream and a 216-bp fragment downstream of fliL_S were amplified using the primer pairs up5826_Fw/up5826_Rv and dw5826_Fw/dw5826_Rv, respectively. The resultant fragments were gel purified and combined as the template to be amplified with the primer pair up5826_Fw/dw5826_Rv. The 486-bp fragment obtained was cloned into the EcoRI/HindIII sites of pK18mobsacB vector to generate pKsacB::fliL_S. The gene replacement was made by mobilization of the construction from S17-1 to B. diazoefficiens USDA 110 by biparental mating, and the simple crossover (cointegrate) was selected by Km resistance. To induce double crossover recombination, the selected transconjugants were plated in yeast mannitol agar (YMA) supplemented with 10% (wt/vol) sucrose. The resulting clones were checked by PCR in order to select the mutant genotype, and the correct in-frame deletion was verified by DNA sequencing. Thus, the length of fliL_S ORF was reduced in 76.6%. To complement the mutation, the complete bll5826 sequence was amplified from B. diazoefficiens USDA 110 chromosomal DNA, using Pfu DNA polymerase with the primer pair Ext5826_Fw/Ext5826_Rv. The resulting PCR fragment was cloned into the EcoRV site of pBBR1MCS2 to create pBBR::fliL_S. The construction was checked by PCR and DNA sequencing and then transferred into the desired B. diazoefficiens strain by conjugation. The transconjugants were selected by Km resistance and confirmed by PCR amplification.

To obtain the B. diazoefficiens ΔfliL_L mutant, an upstream fragment of 278 bp and a downstream fragment of 419 bp were amplified using the primer pair up6868_Fw/up6868_Rv and dw6868_Fw/dw6868_Rv, respectively. These fragments were gel purified and combined as a template to be amplified with the primer pair up6868_Fw/dw6868Rv. The 734-bp fragment obtained was cloned into the EcoRI/HindIII sites of the pK18mobsacB vector to generate the pKsacB::fliL_L construction. The gene replacement and double-crossover recombination was performed like generation of the ΔfliL_S deletion described above. Thus, the fliL_L deletion comprises 90% of the wild-type ORF. For ΔfliL_L complementation, the complete coding sequence of bll6868 was amplified from B. diazoefficiens USDA 110 chromosomal DNA, using Pfu DNA polymerase and the primer pair Ext6868_Fw/Ext6868_Rv. The amplified fragment was cloned into the XbaI/KpnI sites of pFAJ1708 to create pFAJ::fliL_L. The construction was checked by PCR and DNA sequencing and then mobilized into the desired B. diazoefficiens strain by biparental mating, selected by Tc resistance, and checked by PCR amplification, as before.

The B. diazoefficiens ΔlafA strain was obtained by introducing the pMJA05 plasmid into USDA 110 by conjugation as previously described (4). A double-crossover event causing Km sensitivity and Sp-Sm resistance was selected. The mutation was checked by flagellins production as previously described (4). The same strategy was used with the ΔfliL_S background to obtain the ΔfliL_S ΔlafA double mutant strain.

To generate the B. diazoefficiens ΔfliC strain, the pMJA10 plasmid carried by E. coli S17-1 was introduced into USDA 110 by biparental mating as described previously (4). A double-crossover event was selected by searching for Gm sensitivity and Km resistance among the transconjugants. The selected clones were checked by flagellin production as described previously (4). The same approach was used with the background of ΔfliL_L and ΔlafA to obtain the ΔfliL_L ΔfliC and ΔlafA ΔfliC double mutant strains, respectively.

The E. coli MG1655 ΔfliL mutant was generated by P1 transduction of the fliL::kan mutation from strain JW1981 (E. coli Genetic Stock Center) into wild-type MG1655, before curing the resistance cassette using the recombinase activity of pCP20 (50). The strain was verified by PCR using appropriate flanking primers.

For B. diazoefficiens fliL heterologous complementation, the entire ORFs of fliL_S and fliL_L were amplified from B. diazoefficiens USDA 110 chromosomal DNA using the primer pairs fliL_S_Fw/fliL_S_Rv and fliL_L_Fw/fliL_L_Rv, respectively. Then, both fragments were cloned into pBluescriptSK(+) to generate plasmids pBlue::fliL_S and pBlue::fliL_L, then checked by PCR and sequencing. For cloning into pBAD24 expression plasmids, the entire ORFs were amplified from both pBlue::fliL background using PCR and appropriate primers engineered to include NheI and HindIII restriction sites in forward and reverse primers, respectively (5826_Ec_Fw/5826_Ec_Rv for fliL_S and 6868_Ec_Fw/6868_Ec_Rv for fliL_L). Final constructs, named pBAD::fliL_S and pBAD::fliL_L, were confirmed by DNA sequencing after cloning and moved to E. coli MG1655 ΔfliL by electroporation and further selection of ampicillin resistance clones (23).

Motility measurements.

For B. diazoefficiens swimming assay in semisolid medium, fresh bacteria were inoculated using a sterile toothpick on Götz minimal medium supplemented with 0.5% (wt/vol) mannitol and 0.3% (wt/vol) agar or PSY-arabinose 0.3% (wt/vol) agar, and the motility halo was registered as described (9). To increase medium viscosity, polyvinylpyrrolidone (PVP) K-90 was added to the motility plates as described (2). E. coli MG1655 swimming and swarming experiments were performed as described (23) and included the wild type and fliL mutant, both carrying pBAD24 empty vector, with the E. coli ΔfliL complemented strain included as a positive control (23). Plates were supplemented with 0.5% arabinose (wt/vol) instead of glucose, to induce vector expression. B. diazoefficiens cells used for swimming speed measurements were grown in HMY (5) supplemented with arabinose 0.5% (wt/vol) until an optical density at 500 nm (OD₅₀₀) of 0.5 at 28°C and 60 rpm was reached and then were diluted 1:2 in HM salts and incubated at 28°C without shaking for 4 h before microscopic observation. This 4-h incubation increased the motility fraction, and the incubated culture was again diluted 1:25 to be observed under the microscope. To evaluate the PVP effect, samples from 1:2 dilution and 4 h of incubation were gently diluted in HM salts (1:25) containing different concentrations of PVP added just before observation. Free-swimming cells were observed under high-intensity dark-field microscopy (Olympus BX50) and recorded at 1/30^−s (30 frames per second) (Sony Exwave HAD camera). Straight-line speeds were then analyzed using Move-tr/2D software (Library, Tokyo, Japan).

Flagellin preparation and analysis.

Cells were grown in HMY medium with 0.5% (wt/vol) arabinose or 0.5% (wt/vol) mannitol as the sole carbon source (5) until saturation, or in PSY medium with 0.1% arabinose (41) until an OD₅₀₀ of 0.6. Then, the cultures were kept on ice for 15 min, vortexed vigorously for 4 min, and centrifuged at 13,520 × g for 30 min at 4°C. The supernatants were incubated with 1.3% (vol/vol) polyethylene glycol 6000 and 2.5 M NaCl from 2 h to overnight at 4°C. After that, the suspension was centrifuged at 13,520 × g for 45 min at 4°C, and the pellets were resuspended in Laemmli loading buffer (51) or double-distilled water, as needed. To perform the analysis, samples were boiled for 10 min, centrifuged at 14,000 × g for 10 min, and separated by 12.5% sodium dodecyl sulfate polyacrylamide gel electrophoresis (51). Protein quantification was performed using the Bradford method (52).

RNA extraction and retrotranscribed PCR.

Total RNA extraction was performed from cells cultured in liquid and semisolid PSY-arabinose media. From liquid cultures, B. diazoefficiens cells were grown until an OD₅₀₀ of 0.6, placed for 5 min on ice, and centrifuged for 25 min at 11,180 × g at 4°C. Immediately after, we proceeded with the disruption of the cells to extract their RNA as previously described (5).

From semisolid medium, bacteria were inoculated onto soft agar plate and incubated at 28°C for 13 days. At that time, when the diameter of the motility halo was approximately 5.5 cm, cells were isolated from the whole motility halo (WMH) or from the border of the motility halo (BMH). The BMH was defined as the more external motility ring in a radius of 1.5 cm (see the experimental scheme in the supplemental material). In both cases, samples were taken with a clean spoon and placed into a 50-ml plastic tube with an equal volume of double-distilled water. After 5 min on ice, the samples were vortexed for 1 min and centrifuged for 25 min at 11,180 × g at 4°C. Then, the agar above the cells was carefully separated from them to continue with the cell disruption as described above (see the experimental scheme in the supplemental material). To check the quality of cDNA preparations, PCRs were performed with the primer pair q5843_Fw/q5843_Rv. The absence of contaminating DNA was demonstrated by the lack of PCR amplification in an RNA sample that was not subjected to reverse transcription. As a constitutive control, primers for the housekeeping gene sigA were used (53).

Quantitative PCR.

The cDNAs were amplified using the primer pairs lafR_Fw/q6846int_Rv, q6861_Fw/q6861_Rv, q6864-2_Fw/q6864-2_Rv, and q6868_Fw/q6868_Rv. The reactions were performed using iQ SYBR green Supermix (Bio-Rad, USA) according to the manufacturers’ instructions. Normalized expression values were calculated as the ratio between the relative quantities of the gene of interest (GOI) and the relative quantities of the sigA housekeeping gene (53).

Transmission electron microscopy.

To obtain samples from swimming plates, a piece of soft agar with bacteria was taken using the reverse end of a yellow tip, submerged in 50 μl of double-distilled water, and mixed gently. A drop of this suspension was placed on 200-mesh copper grid that had been coated with collodion to proceed as previously described (2).