Detection of Ring Structure and Its Flexibility in Periplasmic Region of FliL, Bacterial Flagellar Stator‐Associate Protein, in Marine Vibrio

ABSTRACT

FliL is a conserved single‐transmembrane protein implicated in stator function and load adaptation of bacterial flagellar motors. We analyzed the oligomeric organization and structural plasticity of the periplasmic domain of polar‐flagellum FliL from Vibrio alginolyticus (^pofFliL_C) and compared it with detergent‐solubilized full‐length ^pofFliL. Recombinant proteins were purified by affinity and size‐exclusion chromatography (SEC). SEC profiles indicated condition‐dependent assemblies; ^pofFliL_C eluted broadly at ~100 kDa and partially shifted to ~45 kDa on re‐chromatography, whereas full‐length ^pofFliL eluted as large, heterogeneous assemblies (~180 kDa). Negative‐stain electron microscopy (EM) revealed abundant ~10‐nm rings for ^pofFliL_C but heterogeneous particles for full‐length protein. Cryo‐EM single‐particle analysis of the ^pofFliL_C yielded high‐quality two‐dimensional top/bottom‐view class averages that show ring particles with clear subunit repeats; class averages correspond to apparent 10‐, 11‐ and 12‐fold stoichiometries, but severe preferred orientation and sample heterogeneity prevented reliable three‐dimensional reconstruction. High‐speed AFM imaged ~15 nm rings and captured transient subunit rearrangements and opening/closing events; representative AFM frames can be interpreted as 10‐subunit‐like, yet AFM contrast and tip convolution limited unambiguous subunit counting across the population. These complementary data indicate that the ^pofFliL_C SPFH‐like periplasmic domain forms flexible, dynamic ring oligomers whose stoichiometry and conformation are modulated by membrane association and experimental conditions. We propose that such plasticity enables FliL to act as an adaptable scaffold for stator engagement and mechanosensitive remodeling of the motor.

FliL, a single‐pass transmembrane flagellar protein with a large periplasmic domain, associates with the basal body and supports motor function under high‐load conditions (e.g., viscous environments). The periplasmic domain of deletion mutant FliL forms a flexible ring structure, likely reflecting the capacity of the full‐length protein to adaptively remodel in response to changes in motor load.

1. Introduction

Many motile bacteria, including the marine Vibrio species, power their flagella with ion‐driven stator complexes to swim. Vibrio alginolyticus produces two distinct flagella, a Na⁺‐driven polar flagellum (Pof) for swimming and H⁺‐driven lateral flagella (Laf) for movement on viscous surfaces. Each flagellar stator consists of two membrane proteins, PomA or PomB for Pof, and MotA (LafT) or MotB (LafU) for Laf, and they convert Na⁺ and H⁺ flow into torque, respectively (Homma et al. 2022). FliL, a small single membrane protein, is the key accessory protein in the flagellar system. FliL is located at the base of the flagellum, in close proximity to the stator and basal body, and interacts with them (Zhu et al. 2015; Lin et al. 2018). Two types of FliL have been identified in V. alginolyticus ; one for Pof and the other for Laf. While deletion of the fliL gene for Laf completely disrupts the surface swarming by Laf (Takekawa et al. 2019), deletion of the fliL gene for Pof reduces swimming speed under high viscous load and also slightly diminishes the assembly of PomA/PomB stator units to the flagellar motor (Zhu et al. 2015; Lin et al. 2018). Deletion of the pomA or pomB gene decreases the assembly of FliL for Pof to the motor (Takekawa et al. 2019). These observations imply that FliL co‐assembles with the stator to support motor torque at high load conditions.

The FliL homologous proteins are distributed in many other bacteria (Figure S1A,B), and its function has been studied across diverse species (Cusick et al. 2012; Subramanian and Kearns 2019; Guo et al. 2022; Partridge and Harshey 2024). In Escherichia coli and Salmonella enterica, FliL directly interacts with multiple motor components (the rotor proteins FliF and FliG, and stator proteins MotA/MotB) and increases flagellar motor torque and normal switching behavior (Attmannspacher et al. 2008; Partridge et al. 2015). In these bacteria, loss of FliL slows motor rotation and impairs high‐load behaviors such as swarming on agar. By contrast, another group has reported that E. coli ΔfliL motors produce wild‐type torque and retain stator binding even under high load, and ΔfliL cells can still swarm (Chawla et al. 2017). This discrepancy might reflect differences in FliL alleles: deletions of the N‐terminal transmembrane region of FliL in place block motility, whereas removal of the entire N‐terminal part preserves normal swimming (Liu et al. 2023). In other species, fliL mutants often abolish motility entirely. For example, although Caulobacter crescentus cells with the fliL mutations can assemble the flagellum, their motility on swarming agar plates is significantly reduced (Jenal et al. 1994). Rhodobacter sphaeroides ΔfliL cells fail to rotate their flagella, and suppressor mutations in MotB can restore the function (Suaste‐Olmos et al. 2010). In S. enterica , fliL mutants break their flagellar rods during swarming and misregulate swarmer differentiation, indicating that FliL stabilizes the motor during high‐load surface motion (Attmannspacher et al. 2008). In P. mirabilis , ΔfliL cells make pseudo‐swarmer cells under surface conditions and fail to induce the expression of flagellin gene, linking FliL to surface sensing and flagellar gene regulation (Lee et al. 2013). In Pseudomonas aeruginosa , which has two distinct stator systems (MotA/MotB and MotC/MotD), FliL was found to interact with both systems but played different roles: it boosts torque and speed stability via MotC/MotD stator, and increases switching frequency via MotA/MotB stator (Zhang et al. 2022). In Sinorhizobium meliloti , both of FliL and its paralog MotF oligomerize and interact with the stator, ΔfliL causes loss of flagella, and ΔmotF causes erratic and low‐torque rotation (Sobe et al. 2022). In Bradyrhizobium diazoefficiens, two FliL paralogs correspond to its two flagellar systems; lateral FliL is essential for swimming mediated by the lateral flagellum, whereas subpolar FliL is not required for swimming in liquid but is necessary to upregulate the lateral flagellum under high load, demonstrating that the functional plasticity of FliL: its interactions and oligomeric states vary depending on the stator type, the flagellar system, or environmental cues (Mengucci et al. 2020; Mookherjee et al. 2024). Thus, FliL broadly acts to ensure robust torque generation under challenging conditions, though its necessity depends on the organism and experimental context.

We previously determined the structure of a C‐terminal periplasmic region of FliL of Pof (^pofFliL_C) by X‐ray crystallography (Takekawa et al. 2019; Kumar et al. 2017). The structure of ^pofFliL_C has a characteristic SPFH (stomatin/prohibitin/flotillin/HflK/C) domain fold, a motif found in membrane scaffolding proteins and mechanosensory proteins, that regulates the function of ion channels in wide species, from mammals to bacteria (Daumke and Lewin 2022; Yokoyama and Matsui 2023). ^pofFliL_C forms a tubular complex with a helical arrangement of 12 subunits per turn or a ring‐shaped complex composed of 10 subunits, with an inner diameter (~8 nm) comparable to a stator unit, in the crystal. From the structure, we proposed a model that a ring consisting of 10 subunits of FliL encircles the single stator unit to form a stator‐FliL complex, and the mutation analysis also supported this model (Takekawa et al. 2019). Recent cryo‐electron tomography studies in other bacteria also bolster this model. In the spirochete Borrelia burgdorferi , FliL was visualized in situ forming a periplasmic ring around each active MotA/MotB stator complex (Guo et al. 2022). In Helicobacter pylori , the periplasmic FliL domain crystallizes as a partial ring, and cryo‐electron tomography revealed full 18 FliL rings coaxial with the 18 MotA/MotB stators (Tachiyama et al. 2022). These FliL rings lie between the MotA ring and the periplasmic part of MotB, suggesting each FliL ring helps trigger the stator's plug to open and conduct H⁺ through the MotA/MotB stator. Such FliL‐stator supercomplexes likely power efficient ion flux under load.

Altogether, converging evidence suggests a unified model: FliL proteins localize with the stator complexes and oligomerize (often as rings) around the stator complexes. This FliL–stator complex enhances torque and enables adaptive motor remodeling under varying mechanical loads (e.g., during swarming or in viscous environments). The periplasmic domain of FliL resembles eukaryotic stomatin‐like mechanosensory domains, hinting at a deep evolutionary conservation of the scaffold function. In this study, we focus on the polar‐flagellar FliL of V. alginolyticus (^pofFliL). We purified both full‐length ^pofFliL and its periplasmic domain (^pofFliL_C) and investigated their structure and biochemical properties, to understand how FliL‐ring architecture contribute to the flagellar motor regulation.

2. Results

2.1. Purification and Oligomeric States of ^pofFliL_C and Full‐Length ^pofFliL

^pofFliL of V. alginolyticus consists of 167 amino acids with a molecular weight of ca. 18 kDa and has an N‐terminal single transmembrane (TM) helix from residue 10 to 39. A periplasmic C‐terminal fragment ^pofFliL_C (the residues from 40 to 167) and the full‐length ^pofFliL were expressed as N‐terminal His‐tagged proteins from pColdI (Figure S1C) in E. coli , purified by cobalt‐affinity chromatography, and further fractionated by size‐exclusion chromatography (SEC) (Figure 1, Figure S2). The SDS‐PAGE analysis indicated a monomeric mass of ~16 kDa for His‐^pofFliL_C and ~18 kDa for detergent‐solubilized His‐^pofFliL. Although the yield of His‐^pofFliL was substantially lower than that of His‐^pofFliL_C, consistent with the previous report (Kumar et al. 2017), a sufficient quantity was obtained to proceed with purification by size exclusion chromatography (SEC).

FIGURE 1.

SEC elution profiles of ^pofFliL and ^pofFliL_C. The cobalt‐affinity purified fractions of ^pofFliL (A, B) and ^pofFliL_C (C, D) were concentrated and applied to the SEC columns (Superdex 200 10/300 for A and B; ENrich SEC 650 for C and D). The elution profiles were monitored by absorbance at 280 nm. The black arrowheads indicate the elution volumes of peaks and the estimated molecular weights.

During SEC, the two constructs displayed markedly different elution behaviors. Detergent‐solubilized His‐^pofFliL purified from the membrane fraction eluted in a peak with a molecular size of ca. 180 kDa, which is substantially larger than the monomer mass and indicative of high‐order oligomers or detergent and/or lipid‐containing assemblies (Figure 1A). The 180 kDa peak fraction of SEC was separated again by SEC, and the 180‐kDa peak remained at a high‐molecular‐weight position but divided into multiple, partially overlapping peaks (Figure 1B), indicating heterogeneous assemblies that may dissociate or rearrange during chromatography. By contrast, ^pofFliL_C eluted initially with a peak at ca. 100 kDa on SEC (Figure 1C). When this peak fraction was subjected to SEC again, the peak shifted to a molecular size of ca. 45 kDa (Figure 1D). These results suggest that ^pofFliL_C exists in a concentration‐ and condition‐dependent equilibrium between higher‐order oligomer(s) and monomer rather than a single static oligomeric state. Taken together, SEC and SDS‐PAGE data indicate that ^pofFliL assemblies are labile and sensitive to solubilization and dilution conditions, and full‐length ^pofFliL forms larger, more heterogeneous complexes than the periplasmic domain without the transmembrane region.

2.2. Negative‐Stain Electron Microscopy Reveals Ring‐Like Complex for ^pofFliL_C and Heterogeneous Particles for Full‐Length ^pofFliL

We previously tried to detect assembly structures of ^pofFliL_C or ^pofFliL separated by SEC using electron microscopy, but we could not observe any clear structures (Kumar et al. 2017). In this study, we observed again the SEC peak fractions by negative‐stain electron microscopy (EM). The ^pofFliL_C fraction corresponding to the 100‐kDa SEC peak contained abundant ring‐shaped particles with an outer diameter of ~10 nm (Figure 2A). The rings displayed variability in apparent completeness and circularity, often showing local breaks or distortions. These observations are consistent with flexible oligomeric rings that can adopt imperfect closed conformations in vitro. On the other hand, the SEC fraction of full‐length ^pofFliL that eluted around 180 kDa did not show a uniform structure by negative stain; rather, particles of heterogeneous shapes and sizes (approximately 5–15 nm) were observed, and clear symmetric ring images were rare (Figure 2B).

FIGURE 2.

Representative electron micrographs of purified ^pofFliL and ^pofFliL_C by negative staining. The SEC peak fractions of ^pofFliL_C (A) and ^pofFliL (B) were negatively stained with uranyl acetate and observed by transmission electron microscopy. Enlarged views of the areas enclosed by white circles are shown in the upper right. The red and yellow arrows indicate the closed and opened FliL rings, respectively.

2.3. Cryo‐EM Single‐Particle Analysis of ^pofFliL_C : Heterogeneous Ring Stoichiometries Revealed by Two‐Dimensional (2D) Class Averages

Cryo‐EM grids prepared from SEC peak fractions of ^pofFliL_C and ^pofFliL were initially screened on a 200 kV electron microscope to assess particle distribution and integrity (Figure 3A, Figure S3). The image profiles are similar to the images by negative staining. Even by cryo‐EM, the ring structures were not observed from the full‐length ^pofFliL and the images seem not to be suitable for the structural analysis. Therefore, Cryo‐images for ^pofFliL_C were subsequently collected on a 300 kV electron microscope. After motion‐correction and contrast transfer function (CTF) parameters estimation, particles were autopicked and subjected to several rounds of reference‐free 2D classification using cryoSPARC.

FIGURE 3.

Ring formation by ^pofFliL_C. (A) Representative electron micrographs of the purified periplasmic region of FliL (^pofFliL_C) by cryo‐EM. Enlarged views are shown in the upper right. (B) Representative 2D class averages from single‐particle cryo‐EM analysis of purified ^pofFliL_C. ^pofFliL_C molecules assemble into rings composed of 10, 11, or 12 subunits. Arrowheads indicate regions where the ring is locally broken. (C) Cα ribbon models of ^pofFliL_C multimers, modeled from the view in (B). (D) Pseudo‐2D projections of the 7 Å resolution maps calculated from the models in (C). (E) Previously reported 10‐subunit ^pofFliL_C ring structure determined by X‐ray crystallography (PDB ID: 6AHQ). (F) Pseudo‐2D projection of a 7 Å resolution map calculated from the structure in (E). Scale bars represent 10 nm.

2D classification yielded a large number of well‐defined top‐view class averages that display ring‐shaped particles with clear subunit density around the circumference (Figure 3B, Figure S4). The rings were not monodispersed: class averages with apparent 10, 11, and 12 repeating densities around the ring were observed. Many classes also exhibited local discontinuities or asymmetry (regions of reduced density or open gaps), indicative of structural heterogeneity and inter‐subunit flexibility (arrowheads in Figure 3B). Because the particle images were overwhelmingly dominated by top‐views (and the corresponding bottom‐views), severe orientation bias limited the range of angular information available for three‐dimensional reconstruction. Therefore, we did not pursue three‐dimensional (3D) reconstruction.

Since interpretation was confined to analysis of the 2D top‐view class averages, we generated pseudo‐projections from the previously determined crystallographic 10‐mer model and from geometric models of 11‐ and 12‐mer ring arrangements (Figure 3C–F). Comparison between these simulated top‐views and experimental 2D class averages supports the assignment of class stoichiometries; that is, that particular classes correspond to 10‐, 11‐, or 12‐subunit rings. However, the 2D data also make clear that the assemblies are not rigidly symmetrical: many class averages show asymmetric density loss or local gaps, consistent with loosely packed interfaces and a tendency to adopt incomplete or distorted rings. Our cryo‐EM single‐particle analysis provides robust 2D evidence for multiple stoichiometries and local flexibility, but the data quality and particle distribution precluded reliable 3D reconstruction under the present conditions.

2.4. High Speed—Atomic Force Microscope (HS‐AFM) Observation of the ^pofFliL_C Ring Complex

To probe the dynamics of the ^pofFliL_C oligomers, SEC peak fractions were adsorbed onto mica and observed by HS‐AFM. HS‐AFM revealed abundant ring‐shaped particles with external diameters of ca. 15 nm (Figure 4A). Time‐series imaging showed that individual rings are dynamic; subunits moved relative to each other, and rings underwent transient opening and closing events on the timescale of seconds to sub‐seconds (Figure 4D, Movie S1). Cross‐sectional height profiles measured across rings indicate thickness of ^pofFliL_C ring consistent with the dimension expected from the crystal model (Figure 4B), although lateral dimensions are likely broadened by tip convolution and partial flattening due to surface adsorption. ^pofFliL_C rings are not static closed circles but dynamic assemblies that can change conformation on the timescale of seconds to sub‐seconds. Notably, some rings display a peripheral density pattern that is consistent with 10 repeating elements (Figure 4C), suggesting that 10‐mer‐like arrangements are readily accessible. However, HS‐AFM images were not uniformly sharp enough to permit counting of the subunits for every ring, and many particles showed lower contrast that precludes unambiguous subunit assignment. Therefore, the AFM data alone did not demonstrate that a 10‐mer is the exclusive or dominant stoichiometry in solution. Together with the cryo‐EM 2D‐class averages (which show 10–12 subunit classes; Figure 3B), the HS‐AFM observations support a model in which the ^pofFliL_C ring can adopt multiple closely related ring stoichiometries and dynamically different conformational states.

FIGURE 4.

HS‐AFM images of ^pofFliL_C. (A) The purified ^pofFliL_C was adsorbed onto a mica substrate, washed with buffer, and observed by HS‐AFM. (B) The cross‐sectional height profile along the red dotted line in (A). (C) The enlarged image of ^pofFliL_C ring. The numbers indicate putative subunits of ^pofFliL_C. (D) Successive AFM images showing opening/closing events of a ring formed by ^pofFliL_C. The images were obtained from the HS‐AFM observations at the various time points.

3. Discussion

Our data indicate that the periplasmic SPFH domain of V. alginolyticus FliL assembles into ring‐like oligomers whose architecture is inherently flexible. 2D cryo‐EM class averages show rings with apparent 10–12 repeats and frequent local discontinuities. HS‐AFM captures transient rearrangements, including opening/closing events. Full‐length, detergent‐solubilized FliL forms larger, less ordered assemblies. Rather than viewing the crystallographic 10‐mer as the exclusive physiological state, these observations support a model in which the FliL rings can form multiple structures closely related to oligomeric geometries and conformations. Such plasticity provides a simple mechanistic explanation for how a single small accessory protein can accommodate different conformational states of the stator, facilitate stator insertion and removal, and adjust contacts with periplasmic stator elements in response to mechanical load.

The inability to obtain a 3D reconstruction with high resolution reflects a combination of strong preferred orientation in the cryo‐EM images and genuine compositional/conformational heterogeneity; attempts to average across these variables produced blurred maps rather than reliable models. HS‐AFM directly visualized dynamic behavior, and although several images show peripheral density consistent with 10 subunits, AFM contrast, tip convolution, and surface adsorption prevent unambiguous subunit counting across the population. Therefore, the assemblies characterized here should be interpreted as permissive in vitro states that suggest specific in vivo architectures.

A flexible periplasmic ring is suited to function as an adaptable scaffold at the stator–motor interface. Although it remains unclear how the B subunit of the stator undergoes conformational changes coupled to ion conduction and rotation of the A subunit, it is possible that dynamic structural changes within the FliL ring accompany these transitions. Such flexibility may help stabilize the stator complex, facilitate stator assembly, support A subunit rotation, and promote efficient ion flow. Furthermore, it is not clarified whether an incomplete FliL ring recruits the stator to form a complete ring, or whether FliL gradually accumulates around the stator after its activation. In marine Vibrio, however, the stator activation may trigger the progressive recruitment of FliL around the stator, ultimately forming a complete ring that assists stator function. The context of the dynamic properties suggests that the mode of FliL ring formation and the timing of its interaction with the stator vary among species. Consequently, the phenotype of the fliL mutant is different among the species.

In summary, the periplasmic domain of Vibrio FliL forms dynamic, heterogeneous ring assemblies that plausibly enable adaptable stator engagement and mechanosensitive remodeling of the motor. Establishing those functional connections will require experiments to observe FliL‐stator interactions in situ or in controlled membrane mimetics.

4. Materials and Methods

4.1. Strains and Plasmids

E. coli strains and plasmids used in this study are listed in Table S1. All constructs encoding V. alginolyticus ^pofFliL variants (full‐length ^pofFliL and ^pofFliL_C, N‐terminally His₆‐tagged) were cloned into the pColdI expression vector (Takara) to allow cold‐induced expression. Recombinant plasmids were introduced into E. coli BL21(DE3) for protein production.

4.2. Purification of ^pofFliL_C or ^pofFliL

Cells containing plasmid pZWS4 and pZWS3 were cultured at 37°C in LB medium [1% (w/v) Bacto‐tryptone, 0.5% (w/v) yeast extract, and 0.5% (w/v) NaCl] to an optical density at 600 nm (OD₆₀₀) of 0.4–0.6, chilled on ice for 30 min, and induced with 0.5 mM IPTG. After induction, cultures were incubated at 16°C for 16–20 h. Ampicillin (100 μg/mL) was included as required by plasmid selection. Cells were harvested by low‐speed centrifugation (5,000×g, 10 min, 4°C) and resuspended in 20TN200 buffer [20 mM Tris–HCl (pH 8.0), 200 mM NaCl]. Cells were disrupted by sonication on ice (for 1 min three times with pulses), and unbroken cells were collected by low‐speed centrifugation (10,000×g, 10 min, 4°C). This procedure was repeated three times, and the resulting supernatant (soluble fraction) was recovered and separated by ultracentrifugation (100,000 × g, 1 h, 4°C).

The supernatant for ^pofFliL_C was loaded on a HiTrap TALON column (Cytiva) pre‐equilibrated with 20TN200 buffer, washed with 20TN200 buffer containing 20 mM imidazole, and eluted with 20TN200 buffer containing 100 mM imidazole.

For full‐length ^pofFliL, the membrane pellet was resuspended in 20TN200 buffer containing 1% (w/v) lauryl maltose neopentyl glycol (LMNG). The solubilization was performed at 37°C for 30 min with gentle agitation, followed by ultracentrifugation (100,000×g, 30 min, 4°C) to remove insoluble material. The supernatant was loaded on a HiTrap TALON column pre‐equilibrated with 20TN200 buffer containing 0.01% (w/v) LMNG, washed with 20TN200 buffer containing 0.01% (w/v) LMNG and 20 mM imidazole, and eluted with 20TN200 buffer containing 0.01% (w/v) LMNG and 100 mM imidazole. Fractions containing the target protein were identified by sodium dodecyl sulfate—poly‐acrylamide gel electrophoresis (SDS‐PAGE) under reducing conditions on 12.5 or 15% (w/v) polyacrylamide gels, followed by Coomassie brilliant blue (CBB) staining.

The fractions of HiTrap TALON column were concentrated by centrifugal concentrator (Amicon Ultra; Merck Millipore) to 0.5 mL. The concentrated protein solution was applied to a pre‐equilibrated size exclusion chromatography (SEC) column. ENrich SEC 650 10 × 300 (Bio‐Rad) or Superdex 200 10/300 column (GE Healthcare) was used for ^pofFliL_C, and Superdex 200 10/300 was used for full‐length ^pofFliL. SEC running buffer was 20TN100 buffer [20 mM Tris–HCl (pH 8.0), 100 mM NaCl] for ^pofFliL_C, and 20TN100 buffer containing 0.0025% LMNG for ^pofFliL. Elution profiles were monitored by absorbance at 280 nm and used to estimate apparent molecular sizes by comparison to standard proteins. The proteins in each eluted fraction (0.5 mL) were analyzed by SDS‐PAGE followed by CBB staining.

4.3. Negative Staining EM

2 μL of peak SEC fractions were applied to glow‐discharged carbon‐coated copper grids, and grids were stained with 1% (w/v) uranyl acetate and air‐dried. Grids were examined on a transmission electron microscope (JEM‐1010, JEOL) operated at 100 kV. Images were recorded for qualitative assessment of particle shape and size.

4.4. Cryo‐EM Acquisition and Data Collection

2.5 μL of peak SEC fractions were applied to glow‐discharged Quantifoil holey carbon grids (R1.2/1.3 Cu 300 mesh; Quantifoil Micro Tools GmbH), and were plunged into liquid ethane for rapid freezing with Vitrobot Mark IV (Thermo Fisher Scientific) with a blotting time of 7 s at 4°C and 100% humidity. Then the grids were imaged on a 200 kV cryo‐EM (JEM2200FS, JEOL) to assess particle distribution and integrity. For high‐resolution data collection, grids were imaged on a 300 kV cryo‐EM (Titan KriosG4, Thermo Fisher Scientific) equipped with an energy filter (Selectris‐X) and a direct electron detector (Falcon4). Movies were automatically collected using the EPU software with appropriate dose‐fractionation and magnification of 130,000×, corresponding to an image pixel size of 0.92 Å. Totally, 4547 movies were collected.

4.5. Cryo‐EM Image Processing

Motion correction, dose‐weighting, CTF estimation, particle picking, and 2D classification were performed using cryoSPARC v4.7.1 (Punjani et al. 2017).

4.6. HS‐AFM

HS‐AFM imaging was performed to observe ^pofFliL_C dynamics in 20TN100 buffer. SEC peak fraction sample was diluted 1/420 with 20TN100. The procedure was described previously (Ishikawa et al. 2025). Briefly, the molecules were adsorbed onto a freshly cleaved mica surface; after incubation at room temperature, the surface was rinsed with 20TN100 and imaged in the same buffer. HS‐AFM imaging was carried out in tapping mode using a laboratory‐built high‐speed AFM apparatus. The spring constant, resonance frequency in water, and quality factor of the small cantilever (BL‐AHB‐B10, Olympus) were 0.1 N/m, 0.5 MHz, and 2, respectively. An amorphous carbon tip grown by electron beam deposition was used as the imaging probe.

Author Contributions

Yuki Tajimi: investigation. Katsumi Imada: formal analysis, supervision. Raymond Burton‐Smith: formal analysis, investigation. Seiji Kojima: investigation, funding acquisition. Kazuyoshi Murata: supervision, funding acquisition, conceptualization. Takayuki Uchihashi: investigation, formal analysis. Michio Homma: writing – review and editing, conceptualization, supervision, funding acquisition. Tatsuro Nishikino: writing – original draft, investigation, funding acquisition. Norihiro Takekawa: writing – original draft, investigation.

Funding

This work was supported by Japan Society for the Promotion of Science, JP23K14158, JP24K09393, JP22H04926, JP23K27140. Japan Agency for Medical Research and Development, JP24ama121005.

Ethics Statement

All experiments involving recombinant DNA and potentially hazardous microorganisms were conducted under biosafety level 1 (BSL‐1) conditions and approved by the Biosafety Committee of Nagoya University (Approval No. 383‐1‐575‐4). All procedures complied with relevant national and institutional biosafety regulations.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

mmi70101‐sup‐0001‐Supinfo1.MP4

Figure S1: Comparison of FliL amino acid sequences.

Figure S2: SDS‐PAGE results of cobalt‐affinity purified ^pofFliL_C and ^pofFliL.

Figure S3: Representative cryo‐EM images of ^pofFliL.

Figure S4: 2D class averaged images of ^pofFliL_C by cryo‐EM. SEC

Table S1: E. coli strains and plasmids used in this study.

Movie S1: HS‐AFM movies of purified ^pofFliL_C. Playback speed is 0.15 s per frame.

Data Availability Statement

The data that supports the findings of this study are available in the S2 of this article.