A Novel Gene Inactivation System Reveals Altered Periplasmic Flagellar Orientation in a Borrelia burgdorferi fliL Mutant^,

Abstract

Motility and chemotaxis are essential components of pathogenesis for many infectious bacteria, including Borrelia burgdorferi, the causative agent of Lyme disease. Motility and chemotaxis genes comprise 5 to 6% of the genome of B. burgdorferi, yet the functions of most of those genes remain uncharacterized, mainly due to the paucity of a nonpolar gene inactivation system. In this communication, we describe the development of a novel gene inactivation methodology to target B. burgdorferi fliL, a putative periplasmic flagellar gene located in a large motility operon and transcribed by RNA polymerase containing σ⁷⁰. Although the morphology of nonpolar fliL mutant cells was indistinguishable from that of wild-type cells, the mutant exhibited a defective-motility phenotype. Cryo-electron tomography (cryo-ET) of intact organisms revealed that the periplasmic flagella in the fliL mutant were frequently tilted toward the cell pole instead of their normal orientation toward the cell body. These defects were corrected when the mutant was complemented in cis. Moreover, a comparative analysis of flagellar motors from the wild type and the mutant provides the first structural evidence that FliL is localized between the stator and rotor. Our results suggest that FliL is likely involved in coordinating or regulating the orientation of periplasmic flagella in B. burgdorferi.

INTRODUCTION

Borrelia burgdorferi is the causative agent of Lyme disease and belongs to a group of bacteria called spirochetes. B. burgdorferi cells have a characteristic flat-wave morphology and unique means of motility (9, 10, 19, 35). As a result of its unique morphology and motility, B. burgdorferi is able to traverse viscous gel-like media in which most other flagellated bacteria slow down or stop (30). Consequently, B. burgdorferi may efficiently bore through host tissues, leading to pathogenesis in the joints, nervous system, and heart (19, 30, 54). The motility of B. burgdorferi results from the coordinated rotation of the periplasmic flagella residing between the outer membrane and the cell cylinder (9, 10, 12, 20, 33, 35). The current swimming model suggests that a run occurs when the anterior periplasmic flagella rotate in one direction (e.g., counterclockwise [CCW]) and the posterior flagella rotate in the opposite direction (clockwise [CW]). Reversals occur when periplasmic flagella at both poles of the cell change their direction of rotation. During a nontranslational mode (flex), the periplasmic flagella at both cell poles rotate in the same direction (9, 20, 35).

Bacterial flagella are composed of three major parts: the motor, hook, and filament. The motor can be divided into two functional units, the rotor and the stator. The stator is the torque generator consisting of the MotA-MotB complex. The rotor is composed, minimally, of the MS ring (FliF), the rod, and the switch complex (FliG, FliM, and FliN). In B. burgdorferi, the periplasm-localized flagellar filament consists of the core protein FlaB and the sheath protein FlaA (17, 46). There are 7 to 11 filaments connected to flagellar motors near each cell pole that extend toward the cell body in an organized ribbonlike fashion (9, 10). FlaB is necessary for cell motility and contributes to the distinctive flat-wave cell morphology (12, 38, 43, 51). A mutation in flaB, the switch protein gene fliG2, or the hook protein gene flgE resulted in cells that lack the periplasmic flagella, are nonmotile, and are rod shaped (38, 43, 51). Genome sequence analyses indicate that B. burgdorferi contains homologues of the motility and chemotaxis proteins found in other bacteria (9, 14, 20, 35, 38, 45). However, B. burgdorferi motility and chemotaxis genes are not regulated by hierarchical gene regulatory cascades found in other species of bacteria, e.g., Escherichia coli (1, 9, 18, 51). Genome sequence analysis also indicates that homologues of σ²⁸ and its anti-sigma factor FlgM are not present in B. burgdorferi and that all motility and chemotaxis genes identified to date appear to be transcribed by RNA polymerase containing σ⁷⁰ (9, 20, 51).

Of at least 25 genes needed to build a functional bacterial flagellum, fliL is one of the genes that is not well understood. The deficiency in information regarding FliL is due in part to the poor homology among bacterial species and a lack of consistent mutational phenotypes. fliL mutants in E. coli or Salmonella enterica have been reported to have a minor defect in swimming motility, but swarming motility was abolished (3). In Caulobacter crescentus, fliL mutant cells are paralyzed, and FliL is required for the cell cycle-dependent ejection of the polar flagellum during the swimmer-to-stalk-cell transition (26). In Proteus mirabilis, fliL mutant cells failed to synthesize flagellin, are nonmotile, and are longer than wild-type cells (5). Recent data indicated that FliL of Rhodobacter sphaeroides is essential for swimming motility (55). Despite these different motility phenotypes, FliL is consistently reported to be associated with the cytoplasmic membrane in each of these organisms (3, 5, 26, 55). For E. coli, the FliL C terminus was also reported to be localized in the periplasmic flagellar stator (3). The B. burgdorferi fliL gene is located between motA-motB and fliM-fliN of the large flgB motility operon (11, 14, 16). It shares poor homology (<10% amino acid sequence identity) with FliL proteins of other bacterial species (34). To date, the function and localization of FliL have not been reported for any spirochetes. Detailed structural and genetic analyses of FliL are required to better understand flagellar assembly and the mechanism of bacterial motility at the molecular level.

Recently, cryo-electron tomography (cryo-ET) was utilized to determine the molecular architecture of the spirochete flagellar motor in situ (24, 25, 40, 41). These studies suggested that the overall spirochetal flagellar motor structures (the MS ring, the C ring, and the export apparatus) are similar to those of other bacteria, e.g., E. coli (9, 25, 31, 33, 40, 41, 47). B. burgdorferi is representative of highly motile and invasive bacterial pathogens and has a small cell diameter (∼0.3 μm) and an orderly arrangement of multiple flagellar motors, making it ideal for the structural analysis of the intact flagellar motor in situ (10, 33, 41). However, B. burgdorferi remains challenging to manipulate genetically despite significant progress that has been made in developing both allelic exchange and transposon mutagenesis. Specifically, the currently available antibiotic resistance cassettes utilized for B. burgdorferi gene inactivation or the transposon mutations can cause polar effects on downstream gene expression, complicating interpretations of results obtained from such mutants (23, 35, 41).

In this communication, we describe the development of a novel procedure to generate nonpolar mutations in B. burgdorferi genes. Using this methodology, we characterize a flagellar protein, FliL. Our studies indicate that B. burgdorferi nonpolar fliL mutant cells exhibit defective motility. Surprisingly, we discover that the periplasmic flagella of fliL mutant cells are frequently tilted toward the cell pole instead of the normal orientation toward the cell body found in wild-type cells. The fliL mutant has a loss of density in a region between the proposed rotor and stator, as determined by cryo-ET (41). Wild-type motility, the periplasmic flagellar orientation, and motor structural defects are restored when the mutant is complemented in cis. Based on the observed phenotypic changes and the apparent location of FliL within the flagellar motor, the possible relationships of FliL with other flagellar proteins and the resulting effects on flagellar orientation are discussed.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

High-passage, avirulent B. burgdorferi sensu stricto strain B31-A and flaB mutant strain MC-1 were described previously (6, 43). Cells were grown in Barbour-Stoenner-Kelly II (BSK-II) medium at 35°C in a 2.5% CO₂ humidified incubator as described previously (44, 52). BSK-II medium for plating contained 0.6% agarose.

Construction and complementation of the fliL mutant.

The fliL gene is separated from motB and fliM by 48 and 37 bp, respectively (14). The 48-bp intergenic sequence between motB and fliL possess a ribosomal binding site. The fliL gene (gene locus BB0279, a 537-bp gene) was inactivated by replacing fliL with the aadA coding sequence for streptomycin-spectinomycin resistance (13) by using overlapping PCR, as schematically shown in Fig. 1. PCR was used to amplify 3 regions of DNA in 3 steps. In the first step, each DNA region was amplified separately by using PCR pairs P1-P2 (5′-flanking DNA, motB), P3-P4 (aadA coding sequence), and P5-P6 (3′-flanking DNA, fliM). Primers P2, P3, P4, and P5 (Table 1) contain several overlapping base pairs, as indicated by different colors in Fig. 1. During step 2, a PCR product was obtained by using primers P1-P4 and the purified DNA products for motB and aadA as templates. In step 3, the final PCR product was obtained by using primers P1-P6 and the purified DNA products of motB-aadA and fliM as a template to amplify the motB-aadA-fliM DNA construct. The final PCR product yielded a 2,725-bp product that was gel purified and cloned into the pGEM-T Easy vector (Promega Inc.). The integrity of the fliL inactivation plasmid was confirmed by PCR and restriction mapping. PCR-amplified DNA was electroporated into B31-A competent cells and plated onto BSK-II solid medium containing 80 μg/ml streptomycin, as described previously (44). Resistant clones were analyzed by PCR and Western blotting using specific antisera raised against recombinant FliL (see below).

Fig. 1.

Schematic representation of the fliL inactivation construct used to transform B31-A competent cells making the ΔfliL mutation.

Table 1.

Primers used in this study

Primer	Sequence (5′–3′)a
P1	CATATGGCTTTGCGAATTAAGA
P2	CTTCCCTCATAATAAAATTCCTCCCATTAA
P3	GAATTTTATTATGAGGGAAGCGGTGATCGC
P4	AAATTTATTATTATTTGCCGACTACCTTGG
P5	CGGCAAATAATAATAAATTTAAAAGAATTT
P6	TCATTCAACCTCTTCTGTAAGC
P7	ACTAGTTAATACCCGAGCTTCAAGG
P8	TGAGGGAGGTTTCCATATGCCTAATAAAGACGATGAC
P9	CGTCTTTATTAGGCATATGGAAACCTCCCTCATTTAAAA
P10	GCGGCCGCTTACATATCAAAAATATCAATTTGG

^aOverlapping sequences are shown in boldface type.

Complementation of the ΔfliL mutation.

The B. burgdorferi flgB promoter (P_flgB) and the fliL gene were PCR amplified separately by using primers P7-P8 and P9-P10, respectively (primer sequences are presented in Table 1). Overlapping PCR was used to construct P_flgB-fliL in such a way that the promoter drives the expression of fliL. This construct also contained engineered 5′-SpeI–3′-NotI ends. The P_flgB-fliL DNA was then ligated into the SpeI-NotI sites of a suicide vector, pXLF14301, that contains a gentamicin resistance cassette (P_flgB-aacC1) (39, 58). The resulting construct was used to complement the fliL mutant (ΔfliL) by inserting it into a heterologous chromosomal site between the BB0445 and BB0446 genes as described previously (39, 58). Plasmid DNA was electroporated in ΔfliL cells and plated, and the transformants were selected with 40 μg/ml gentamicin plus 80 μg/ml streptomycin. Resistant clones were analyzed by PCR and Western blotting for the integration of P_flgB-fliL within the BB0445 and BB0446 genes and restoration of FliL synthesis, respectively.

Protein preparation and antibody production.

To express recombinant FliL (178-amino-acid protein), we PCR amplified the coding sequence without the predicted transmembrane domain encoded by the first 150 bp. The PCR primers (5′ to 3′) were T-R-FliL-F (5′-GTGTCTAAAATGGTGGTAAGCC-3′) and T-R-FliL-R (5′-TTACATATCAAAAATATCAATTTGGG-3′). The amplified 386-bp DNA fragment was ligated into an E. coli expression vector, pTrc-His TOPO, using TA cloning (Invitrogen Inc.). The resulting vector carrying fliL was transformed into an E. coli host strain, DH5α, and induced by using 0.25 mM isopropyl β-d-thiogalactoside for 4 h at 37°C. The recombinant protein was purified by using Ni²⁺ affinity chromatography according to the manufacturer's protocol (Invitrogen Inc.) and dialyzed against phosphate-buffered saline. Rabbits were immunized with approximately 400 μg of dialyzed protein to raise specific antiserum according standard methods (Alpha Diagnostic International).

Gel electrophoresis and Western blot analysis.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting with an enhanced chemiluminescence detection method (GE Healthcare) were carried out according to the manufacturer's instructions. Growing B. burgdorferi cells were pelleted, washed twice with phosphate-buffered saline, and resuspended in the same buffer. The protein concentration of cell lysates was determined by a Bradford protein assay kit (Bio-Rad Inc.). Unless otherwise noted, 10 μg of lysates was loaded into each lane of an SDS-PAGE gel and subjected to Western blotting using specific antibodies. Monoclonal and polyclonal antibodies kindly provided by other investigators included the following: monoclonal anti-B. burgdorferi FlaB (H9724) from A. G. Barbour (University of California, Irvine, CA), polyclonal anti-E. coli FliM from D. Blair (University of Utah, Salt Lake City, UT), and polyclonal anti-B. burgdorferi MotB from J. Carroll (RML, NIH, Hamilton, MT). The reactivities of those antibodies to B. burgdorferi FlaB, FliM, and MotB were reported previously (8, 35, 43, 46, 51).

Swarm plate motility assays.

Swarm plate motility assays were performed as described previously (43, 44). Approximately 1 × 10⁶ cells in a 5-μl volume were spotted onto a 0.35% or 0.5% agarose plate containing plating BSK medium diluted 1:10 in Dulbecco's phosphate-buffered saline. Since B. burgdorferi is a slow-growing organism (8- to 12-h generation time), plates were incubated for 5 days at 35°C in a 2.5% CO₂ humidified incubator. The paired Student t test was used to compare wild-type and fliL mutant cell swarm diameters.

Microscopy and computer-assisted motion analysis.

Live B. burgdorferi cells were observed under a dark-field microscope (Zeiss Axio Imager M1) connected to an AxioCam digital camera. Exponentially growing cells were mixed with 0.5% 400-mesh methylcellulose (Sigma-Aldrich Co.) and video recorded at room temperature (23°C). Cell swimming behavior and velocity were determined by using Axivision software. Results are expressed as velocity (mean distance in μm traveled by a given strain per second). At least eight cells from each strain were analyzed. The paired Student t test was used to compare wild-type and fliL mutant cell swimming velocities.

Cryo-electron tomography of frozen-hydrated fliL organisms.

Cryo-ET was carried out as described previously (41). Briefly, the B. burgdorferi culture was deposited onto a glow-discharged holey carbon electron microscopy (EM) grid, blotted, and rapidly frozen in liquid ethane. The frozen-hydrated specimens were imaged at −170°C by using a Polara G2 electron microscope (FEI Company) equipped with a field emission gun and a 4,096-by-4,096-pixel charge-coupled-device (CCD) camera (Tvips GmbH, Germany). The microscope was operated at 300 kV with a magnification of ×31,000. Low-dose single-axis tilt series were collected from each bacterium at a 5-μm defocus with a cumulative dose of ∼100 e⁻/Ų distributed over 65 images with an angular increment of 2°, covering a range from −64° to +64°. The tilt series images were aligned and reconstructed by using the IMOD software package (32). In total, 98, 123, and 149 cryo-tomograms from the wild type, an fliL mutant (ΔfliL), and a complemented fliL⁺ strain were reconstructed, respectively. The effective pixel size was 5.6 Å.

3-D subvolume averaging of the flagellar motor.

The subvolume processing of flagellar motors was carried out as described previously (41). The positions and orientations of each periplasmic flagellar motor in each tomogram were determined manually. In total, 711 flagellar motor subvolumes (256 by 256 by 256 voxels) containing the flagellar motors were extracted from 123 tomograms of fliL mutant cells and were further aligned to generate the three-dimensional (3-D) structure. In addition, 743 flagellar motors from complemented fliL⁺ cells were selected for 3-D analysis. Finally, 591 flagellar motors from wild-type cells were selected as a control for the 3-D analysis.

3-D visualization of the cell and flagellar motor.

Tomographic reconstructions were visualized by using IMOD (32). One organism of the B. burgdorferi fliL mutant strain was segmented manually by using Amira 3-D modeling software (Visage Imaging). The 3-D segmentations of flagellar filaments, peptidoglycan, outer surface proteins, and outer and cytoplasmic membranes were manually constructed. The surface model from the averaged flagellar motor was computationally inserted into the segmented model of the organism.

RESULTS

Development of a novel methodology to inactivate a gene without causing polar effects.

A common problem of insertion mutagenesis using antibiotic resistance cassettes is a polar effect on genes downstream from the targeted gene. Because of the polar effect, interpretations of results obtained from such a mutant are complicated (23, 35, 41). Investigators commonly use a B. burgdorferi flgB or flaB promoter fused to the kanamycin (P_flgB-Kan), streptomycin (P_flgB-aadA), or gentamicin (P_flgB-aacC1) resistance cassette to inactivate a gene (50). Recently, we reported the development and use of promoterless cassettes (PL-Kan or PL-aadA) with the hypothesis that the use of a cassette without a promoter decreases the chance of having a polar effect (56). Here we report the further development of a novel methodology to inactivate B. burgdorferi genes. PL-Kan or PL-aadA contained no promoter but had an intact ribosomal binding site and a transcription start site of the flgB promoter P_flgB (6, 18, 56). Although those cassettes do not cause a polar effect, we further refined the PL-aadA cassette. We hypothesized that if a B. burgdorferi gene next to or away from a promoter is expressed, the replacement of the gene with an antibiotic resistance gene coding sequence should also be expressed and confer resistance to that antibiotic. Accordingly, we replaced the fliL gene with the antibiotic resistance gene (only the aadA coding sequence from the ATG start to the TTG stop codon) in a way that aadA uses the targeted gene's RNA polymerase machinery to be expressed by maintaining transcriptional readthrough. Streptomycin-resistant clones were confirmed by PCR for the integration of the aadA coding sequence replacing fliL, as expected (not shown). Polyclonal antiserum was raised against purified recombinant 6×His-FliL, and the inactivation of the fliL gene was confirmed by Western blotting. As shown in Fig. 2, the antibody failed to detect a 20-kDa protein band in fliL mutant (ΔfliL) cells, indicating the inactivation of fliL, whereas this protein was detected in wild-type and complemented cells (see below). These results confirmed our hypothesis that an antibiotic resistance gene can be expressed by using the targeted gene operon's transcription machinery and confer resistance once it undergoes homologous recombination.

Fig. 2.

Effects of fliL inactivation and complementation on flagellar protein expression, as determined by Western blot analysis. The FliL protein was present in similar quantities in wild-type and complemented fliL⁺ cells but was not detectable in the ΔfliL strain (FliL blot). No polar effect on downstream FliM synthesis was detected (FliM blot, bottom protein band [39 kDa]). The inactivation of fliL did not alter MotB protein (top protein band [24 kDa]) or FlaB protein (41 kDa) synthesis.

Determination of polar effect and complementation of ΔfliL cells.

We performed Western blotting to determine if the insertion of the “strep” (aadA coding sequence) marker resulted in an altered expression of genes that are immediately upstream or downstream of fliL. The fliL gene is located in the middle of the large flgB motility operon containing 26 genes (9, 14, 16, 18, 20). The motB and fliM genes are immediately upstream and downstream, respectively, of fliL (Fig. 1). Western blotting with a polyclonal anti-FliM or anti-MotB antibody indicated that FliM or MotB protein synthesis was not altered in the ΔfliL mutant cells (Fig. 2). These results indicated that the replacement of the fliL gene with aadA did not cause a polar effect. Thus, this strep marker can be used to inactivate genes that are within an operon without introducing a polar effect.

Although the fliL mutant cells did not exhibit a polar effect, we complemented the mutant in cis by integration in a heterologous chromosomal location to prevent overexpression (39, 58). Furthermore, bacterial flagellar proteins maintain a stoichiometric relationship with other flagellar proteins (27, 42), and maintaining their ratio is critical for restoring a mutated gene's function (15, 57; our unpublished observations). To complement the fliL mutant, we fused fliL with the flgB promoter (P_flgB) and then inserted the P_flgB-fliL construct within the BB0445 and BB0446 genes using a suicide vector that was reported previously to complement B. burgdorferi mutants in cis (39, 58). The P_flgB-fliL insertion was in a region where both the BB0445 and BB0446 genes converge; thus, the inserted DNA was downstream of both genes and less likely to affect their expression (39, 58). The genomic integration of P_flgB-fliL within BB0445-BB0446 was confirmed by PCR (not shown). The restoration of FliL protein synthesis was determined by Western blotting. FliL protein synthesis in the complemented fliL (fliL⁺) cells was restored and at the same level as that of wild-type cells, indicating that FliL was not overexpressed (Fig. 2).

The fliL mutant is defective in motility.

B. burgdorferi FliL is predicted to contain a transmembrane domain at the N-terminal 33 to 51 amino acid residues using the secondary-structure topology prediction program TMpred (22; not shown). The gene function for fliL has not been reported for B. burgdorferi or any other spirochetes. We performed motility assays using 0.35% and 0.5% agarose plates in order to determine the effect of FliL on cell motility and compared the swarm diameters of wild-type, ΔfliL, and complemented fliL⁺ cells. A nonmotile periplasmic flagellinless flaB mutant was used as a control (43). Motile cells swarm out from the initial site of inoculation as they metabolize nutrients. In many species of bacteria, swimming and swarming are assayed on plates containing low-concentration (0.3%) and high-concentration (0.5 to 0.7%) agar media, respectively (3, 21). However, in the spirochetes/B. burgdorferi, swarming is assayed on plates containing 0.35% agarose (35–38, 43, 45, 51). Thus, to be consistent, we continue to use the term “swarming” when B. burgdorferi cells were assayed on plates containing 0.35% agarose. Unlike other species of bacteria (e.g., E. coli and S. enterica), B. burgdorferi wild-type cells do not swarm on 0.5% agarose (Fig. 3a). However, when cells were inoculated in 0.35% agarose, the fliL mutant cells exhibited significantly defective swarming motility compared to that of the wild-type parental cells (P value of 0.002). As shown in Fig. 3b and c, the fliL mutant cell swarm diameters were similar to those of the flagellinless nonmotile flaB cells. The motility defect exhibited by the fliL mutant cells was restored upon complementation (Fig. 3b and c). These results indicate that FliL is required for full motility and that the phenotype was attributed solely to the fliL mutation, and there was no secondary alteration.

Fig. 3.

Swarm plate motility assays using the indicated strains. A nonmotile flaB mutant was used as a control. (a) B. burgdorferi cells do not swarm on 0.5% agarose plates. (b) Representative 0.35% swarm plate. Solid bars indicate swarm diameters. (c) Averages ± standard deviations of swarm diameters from four independent assays. A paired Student's t test was used to determine a P value of 0.002 between wild-type (WT) and fliL mutant cell swarm diameters on 0.35% agarose plates.

Furthermore, when the fliL mutant cells were grown in BSK-II broth, we noticed that these cells sometimes twisted with each other and formed small clumps at the bottom of the culture tubes, which is indicative of defective swimming motility. Dark-field microscopy coupled with video analysis of swimming cells demonstrated that wild-type B. burgdorferi cells ran, paused (flexed), and reversed. The fliL mutant cells also ran, paused, and reversed; however, their flexing was brief and did not completely distort the cell morphology or bending at the center of the cell, as seen for the wild-type cells (19, 35, 45). B. burgdorferi cells do not efficiently swim in liquid growth medium, but the swimming velocity increases in a gel-like medium, such as one containing methylcellulose (4, 7, 19, 30, 38, 44). Accordingly, wild-type and mutant cells were mixed with 0.5% methylcellulose, and cell swimming velocities were determined. The ΔfliL mutant cells had a swimming velocity of 3.4 ± 0.49 μm/s, whereas the wild-type cells swam at a significantly higher rate, with a velocity of 5.5 ± 0.19 μm/s (P = 0.01). The complemented cells restored the swimming motility (velocity of 4.8 ± 1.2 μm/s). Taken together, these results indicate that the ΔfliL mutant cells are defective in motility. In vitro growth curve studies confirmed that ΔfliL mutant cells had no appreciable growth defect relative to wild-type or complemented fliL⁺ cells, eliminating the possibility that phenotypic differences were a consequence of in vitro growth differences between the clones (not shown).

Abnormal orientation of periplasmic flagella in the fliL mutant.

The morphology of the B. burgdorferi fliL mutant was examined by dark-field microscopy and was found to be indistinguishable from that of wild-type cells (not shown; see below). Furthermore, FlaB expression levels, which partly determine the B. burgdorferi wavelike cell morphology (12, 20, 38, 43, 49, 51, 53), were similar between the wild type and the fliL mutant, as confirmed by Western blotting (Fig. 2). Cryo-ET of intact organisms indicated that the overall cellular morphology and the flagellar motor structure of ΔfliL cells are similar to those of wild-type cells (Fig. 4). The deletion of fliL did not alter the formation or assembly of periplasmic flagella. However, we found that the ΔfliL cells showed an unusual structural phenotype: of five periplasmic flagella visible in the 3-D reconstructions of the cell shown in Fig. 4, three of them were directed toward the cell pole (see arrows in Fig. 4d and f and see Movie S1 in the supplemental material) instead of being oriented toward the middle of the cell body, as found for the wild-type parental cells (Fig. 4c and e). We examined more than 40 cells from each strain to determine the normal and abnormal/irregular periplasmic flagella in each cell (Table 2). On average, 26.4% of the periplasmic flagella in each ΔfliL mutant cell were oriented toward the cell pole (irregular/abnormal orientation) (Table 2), whereas only 1.7% of periplasmic flagella from each wild-type cell were oriented toward the cell pole (Table 2). Similar to the wild-type, the complemented fliL⁺ cells restored the phenotype, as only 3.6% of each cell's periplasmic flagella were detected to exhibit the abnormal orientation (Table 2). However, although the periplasmic flagellar motors of the ΔfliL cells are located in the cell poles, as seen for the wild-type cells, the abnormally oriented flagellar filaments are shorter (0.3 to 2 μm), ending at the cell pole, than normal flagella (∼6 μm), extending toward the cell body (Fig. 4d).

Fig. 4.

The fliL mutant exhibits a high proportion of oppositely oriented periplasmic flagella. (a and b) One section of a 3-D reconstruction of wild-type (a) and fliL mutant (b) cells. (c and d) Surface renderings of the corresponding 3-D reconstructions. The prominent structural features include the outer membrane (OM), cytoplasmic membrane (CM), peptidoglycan layer (PG), and outer surface protein (OSP). Oppositely oriented periplasmic flagella are indicated by cyan arrows (b). Periplasmic flagella with a normal orientation (toward the middle of the cell body) are shown in blue (c and d), and periplasmic flagella with the opposite orientation (toward the cell pole) are shown in cyan (d). (e and f) Cartoon models illustrating the flagellar orientations of the wild type (e) and the fliL mutant (f).

Table 2.

Comparative cryo-ET analysis revealing the increasing trend of irregular orientation of periplasmic flagella in fliL mutant cells

Strain	No. of cells analyzed	No. of irregular periplasmic flagellaa	No. of normal periplasmic flagellab	% irregular cells
Wild type	43	5	288	1.7
ΔfliL	41	55	208	26.4
fliL+	46	11	301	3.6

^aIrregular periplasmic flagella were tilted toward the cell pole.

^bNormal periplasmic flagella were tilted toward the cell body.

FliL localization in the flagellar motor.

The molecular architecture of the periplasmic flagellar motor in the ΔfliL mutant was determined by tomographic subvolume alignment and averaging. The 711 motors extracted from 123 B. burgdorferi cells were used to generate a 3-D flagellar motor structure from the fliL mutant at a resolution of ∼4 nm. Without imposing any rotational symmetry, the structure preserves key asymmetric features, as described previously for the structure of wild-type cells (Fig. 5a), including the visualization of the hook and the curvature of the stator embedded in the cytoplasmic membrane (41). The overall structure of periplasmic flagellar motors from ΔfliL cells closely resembles that of wild-type cells (Fig. 5a and d), as shown in their central sections, which illustrate the locations of the putative hook, rod, MS ring, stator, C ring (FliG, FliM, and FliN), and export apparatus.

Fig. 5.

Localization of the FliL protein in the B. burgdorferi flagellar motor. The asymmetric average flagellar motor structures from wild-type (a), fliL mutant (b), and complemented fliL⁺ (c) cells are displayed. The overall structures of the periplasmic flagellar motors look very similar. The extra densities indicated by orange arrows (a and c) are the main differences from the fliL mutant (b), indicating that FliL (L) is located between the stator (S) and the collar (C), as shown in d and f. The white arrows (b) indicate the densities corresponding to the periplasmic domain of the stator (MotB). Flagellar models (d, e, and f [with or without FliL]) were overlaid on the images from a, b, and c, respectively. The MS ring, export apparatus, P ring (P), rod (R), C-ring components (FliG, FliM, and FliN), and cytoplasmic membrane (mem) are labeled according to data described previously by Liu et al. (41). The scale bar is 50 nm.

Despite the overall similarity between the motor structures of ΔfliL and wild-type cells, there are clear distinctions between them (Fig. 5a and b, orange arrows). The periplasmic flagellar motors of the ΔfliL cells lacked a density just outside the collar region of the rotor (Fig. 5b and e), which is clearly visible in wild-type cells (see orange arrows in Fig. 5a) or complemented fliL⁺ cells (Fig. 5c). This feature appears to extend from the cytoplasmic membrane into the periplasmic space and is located between the proposed bowl-shaped rotor and the surrounding stator structures (Fig. 5d) (41). Therefore, the extra density in flagellar motors of wild-type organisms corresponds to the FliL protein (labeled L in Fig. 5d and f).

DISCUSSION

The fliL gene has been characterized in E. coli, S. enterica, Caulobacter crescentus, Proteus mirabilis, and, recently, Rhodobacter sphaeroides (3, 5, 26, 55). FliL of R. sphaeroides was reported previously to be essential for flagellar rotation and modulates the motor function through an interaction with MotB (55). An fliL mutant of E. coli or S. enterica was found previously to be defective in swarming motility (on 0.5 to 0.7% agar plates) but had a minor defect in swimming motility (in 0.3% agar) (3). Based on their findings, Attmannspacher et al. proposed previously that in E. coli and S. enterica, “flagellar motors experience a higher torque during swarming motility owing to an increased proton motive force, and that FliL allows the flagellar rod to withstand the increased torsional stress” (3). Because of this stress, the mutant cells had a hook-filament-ejection phenotype (3). For C. crescentus, an fliL mutant was found to be paralyzed, and the FliL protein was reported to be required for the cell cycle-dependent ejection of the polar flagellum during the swimmer-to-stalk-cell transition (2, 26, 28, 48). A P. mirabilis fliL mutant was reported to be nonmotile, failed to synthesize flagella, and had an elongated-cell phenotype (5). Furthermore, based on its predicted location and phenotype, Belas and Suvanasuthi suggested that FliL may be responsible for the transmission of signals from the exterior flagellar filament into the cell (5). Clearly, FliL has diverse functions in these organisms. However, there is a consensus that FliL is a transmembrane protein associated with the flagellar motor (3, 26, 29, 55). Based on the findings with the P. mirabilis and E. coli/S. enterica fliL genes, two models were proposed for the location of FliL (3, 5). In both models, FliL is predicted to be a part of the flagellar motor.

B. burgdorferi ΔfliL cells exhibited a significant defect in swimming (in methylcellulose) and swarming (on 0.35% agarose plates) motility despite the fact that the amounts of the flagellar filament protein FlaB and the motor protein MotB were not altered in the ΔfliL strain. The defective-motility phenotypes are similar to those found previously for swarming fliL mutants of E. coli/S. enterica (3). Whereas different mechanisms were proposed for defective motilities in different organisms (3, 5, 26, 55), we propose that periplasmic flagella with abnormal orientations oppose smooth swimming mediated by the standardly oriented flagella, resulting in defective motility in B. burgdorferi. The wavelike morphology of B. burgdorferi, which is dictated at least in part by the presence of flagellar filaments, was not altered in ΔfliL cells (Fig. 2) (9, 12, 20, 38, 43, 51, 53). Cryo-ET of B. burgdorferi ΔfliL cells indicated that the periplasmic flagella are connected to the hook-basal body (Fig. 4). However, we were surprised that 26.4% (versus only 1.7% in the wild type) (Table 2) of the motors/periplasmic flagella were oriented toward the cell pole rather than extending toward the middle of the cell body. This remarkable phenotype has not been observed for previously characterized spirochetal flagellar mutants (24, 41). FliL is therefore likely to play a role in defining the orientation of periplasmic flagella in B. burgdorferi (Fig. 4), although a mechanism is currently unknown. It is also unknown why 26.4% but not all periplasmic flagella of each ΔfliL cell were tilted toward the cell pole. We speculate that FliL and its interacting protein(s) may be responsible for the orientation of periplasmic flagella. Nevertheless, the altered flagellar orientation is likely to be the underlying reason for the defective-motility phenotype observed for the fliL mutant cells (Fig. 3). In contrast, the lack of FliL tends to fracture the flagellar rod in swarmer cells of S. enterica (3). Furthermore, the defective motility exhibited by R. sphaeroides fliL was proposed previously to be mediated through modulating the MotB function (55). Therefore, our data reinforce the notion that the mechanisms of altered motility observed for B. burgdorferi and these other bacteria are likely to be different, even though fliL mutants of B. burgdorferi and S. enterica share similar defective-motility phenotypes.

Multiple copies of FliL are located between the rotor and stator at the cytoplasmic membrane in B. burgdorferi (Fig. 5), although there is a lack of detailed information about its shape or stoichiometry. The deletion of fliL has little effect on flagellar assembly and the overall molecular architecture of the flagellar motor. However, there is clear evidence of an abnormal periplasmic flagellar orientation and defective motility of fliL mutants. Recently, Kudryashev et al. speculated that a protein which links the rotor with the stator is likely to be FliL (32). Our data provide the first structural evidence that B. burgdorferi FliL is located between the rotor and stator which in some manner affects the periplasmic flagellar orientation and, thus, swimming efficacy.