Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 May;9(5):1271-1281.
doi: 10.1038/s41564-024-01674-1. Epub 2024 Apr 17.

CryoEM structures reveal how the bacterial flagellum rotates and switches direction

Prashant K Singh  1 , Pankaj Sharma  1 , Oshri Afanzar  2 , Margo H Goldfarb  1 , Elena Maklashina  3   4 , Michael Eisenbach  5 , Gary Cecchini  3   4 , T M Iverson  6   7   8   9
Affiliations

CryoEM structures reveal how the bacterial flagellum rotates and switches direction

Prashant K Singh et al. Nat Microbiol. 2024 May.

Abstract

Bacterial chemotaxis requires bidirectional flagellar rotation at different rates. Rotation is driven by a flagellar motor, which is a supercomplex containing multiple rings. Architectural uncertainty regarding the cytoplasmic C-ring, or 'switch', limits our understanding of how the motor transmits torque and direction to the flagellar rod. Here we report cryogenic electron microscopy structures for Salmonella enterica serovar typhimurium inner membrane MS-ring and C-ring in a counterclockwise pose (4.0 Å) and isolated C-ring in a clockwise pose alone (4.6 Å) and bound to a regulator (5.9 Å). Conformational differences between rotational poses include a 180° shift in FliF/FliG domains that rotates the outward-facing MotA/B binding site to inward facing. The regulator has specificity for the clockwise pose by bridging elements unique to this conformation. We used these structures to propose how the switch reverses rotation and transmits torque to the flagellum, which advances the understanding of bacterial chemotaxis and bidirectional motor rotation.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1. The flagellar motor and structure of the switch.
a, Schematic diagram of the flagellar motor showing the L-ring, P-ring, MS-ring and C-ring, as well as the flagellar rod, hook and filament. The switch is housed within the C-ring and is composed of the FliG, FliM and FliN subunits. OM, outer membrane; IM, inner membrane. b, Cropped view of a representative cryoEM micrograph for wild-type MS- and C-rings (1 of 34,381 micrographs) showing the quality of particles used in structure determination. Most particles contain both the MS- and C-rings, although a small number of isolated MS-rings are present. En face views (three are highlighted with black circles) and side views (three are highlighted with white circles) are observed. Tilted views are also observed and give the appearance of a smaller diameter in some cases. Scale bar, 200 Å. The uncropped micrograph is available in the Source data file. Raw micrographs for all structures have been deposited with EMPIAR (https://www.ebi.ac.uk/empiar/) and accession codes EMPIAR-11597, EMPIAR-11891 and EMPIAR-11892. c, Surface representation of the C-ring density maps in the CCW pose superimposed on the final model. FliF subunits are shown in blue, FliG subunits are shown in red, FliM subunits are shown in yellow and FliN subunits are shown in shades of pink and purple. Source data
Fig. 2
Fig. 2. The CCW pose of the switch.
In the global views, FliF is blue, FliG is red, FliM is yellow and FliN is pink and purple. In the insets of a–e, each of the subunits is coloured from the N-terminus (blue) to C-terminus (red) to highlight the fold. a, FliFC wraps around FliGD1. b, A FliG protomer folds into five domains: FliGD1 (FliG1–67), FliGD2 (FliG73–99), FliGD3 (FliG107–186), FliGD4 (FliG196–233) and FliGD5 (FliG243–331). c, The FliM subunit, highlighting FliML1 (FliM31–50), FliMmid (FliM51–230), FliML2 (FliM231–256) and FliMC (FliM257–330). d, A 180° rotated view of panel (c). e, Three FliNC subunits are similar but non-equivalent. To highlight the fold, only one protomer (FliN3) is coloured from the N-terminus (blue) to C-terminus (red). The remaining two (FliN1 and FliN2) are coloured pink and purple. f, A side view of a single FliFGMN unit. An ~30 Å cleft between FliFC–FliGD1/D2 and FliGD5 is highlighted. g, A single FliFGMN unit participates in three staves. h, Density for FliGD5 appears to be separated, with the domain having little contact with adjacent subunits. i, Interactions between the PAA motif of FliGD3 and the adjacent FliGL1 linker. j,k, Formation of a curved spiral by the FliMC:3FliNC heterotetramer. j, A schematic that compares the open ring of FliMC:3FliNC in the cryoEM structure to the closed ring of the crystal structure of T. maritima FliNC (1YAB). This comparison highlights that a pure FliNC superstructure would favour stacked discs in a linear array. FliMC breaks the symmetry, which is necessary to form the helix along the bottom of the C-ring. k, The FliMC:3FliNC forms a spiral that curves along the base of the C-ring to form a closed circle. A single arc is shown.
Fig. 3
Fig. 3. The CW pose of the switch.
a–e, Individual folds of the C-ring subunits in the CW pose: FliF (a), FliG (b), FliM (c), a 180° rotated view of FliM (d), FliN (e). The relative orientation and the colouring are the same as for Fig. 2. f, A side view of the CW pose showing an expanded cleft between FliGD1/D2 and FliGD5. g, Comparison of the CCW pose (transparent) with the CW pose (solid) of a single C-ring subunit. h, A 90° rotation of panel (g) highlights the magnitude of the conformational change. i, Comparison of a top-down view of the CCW and CW poses shows the reversed orientation of the FliFC helix, which changes the connection to the MS-ring. The increased size of the cleft in the upper ring of FliG is also apparent. j, Colouring a single subunit in the context of the C-ring highlights the increased domain swaps in FliG of the CCW pose compared to the CW pose. In the CCW pose, FliFC–FliGD1 is in the inner ring above FliM and crosses staves three times. In the CW pose, FliFC–FliGD1 is in the inner ring behind FliGD5 and crosses staves twice. k, A side view of a single unit aligned to the FliMC:3FliNC spiral highlights the 25° rotation of FliMmid in the CW pose.
Fig. 4
Fig. 4. Allostery in the switch.
a, Allosteric pathway from the FliM N-terminus to the torque helix in FliGD5. Different steps of signal transmission are coloured from blue to red and numbered. The transfer of information starts at (1) the N-terminus of FliM near the FliML1 linker at the FliM-FliN interface. The information passes through the first helix of (2) FliMmid to FliGD3 near the (3) PAA motif, which supports (4) FliGD4. A rotation of (5) the C-terminal FliGD5 changes the orientation of the torque helix. Note that allosteric signal transmission may involve both the pathway that is shown and a concerted signal transmission in adjacent subunits of the ring. A single subunit of the C-ring with a neighbouring FliG (FliGN − 1) is shown. b, Locations of directionally biased mutations in the switch. Red balls highlight CCW-biasing mutations, with the majority of these located in the proposed binding site for CheY. Their mutation could affect CheY binding. Yellow balls mark locations of CW-biasing mutations, which cluster along the pathway in a. Their mutation could release the CCW pose.
Fig. 5
Fig. 5. A regulator bound to the CW pose of the switch.
a, On the left, twelve repeats of the protein-bound CW pose of the switch are shown in grey, with three repeats of the density in the cleft shown in green. For comparison, the right shows the CW pose of the switch not bound to a partner protein. The cleft is still visible but lacks density within it. b, Cross-section of a single subunit with the density for the regulator in green. For comparison, the CW pose with an empty cleft is shown on the right.
Fig. 6
Fig. 6. Torque transmission during flagellar rotation.
a, A model for the flagellar motor from Gram-negative bacteria was built from our structure and that of the S. enterica flagellar basal body (7CGO ref. ). The L-ring (light purple) contains FlgH subunits. The P-ring (dark purple) contains the FlgI subunits. In the centre of the LP-ring is a rod (grey). The distal region of the rod (FlgG) connects to the hook and flagellum, while the proximal region of the rod (FliE, FlgB, FlgC and FlgF) connects the LP-ring to the MS-ring. The MS-ring (blue) localizes within the inter-membrane space and contains FliF subunits. Finally, the C-ring (orange), contains FliF C-termini as well as FliG, FliM, and FliN of the switch. b, Pathway of torque transmission through the flagellar motor. The C-ring transmits torque from the MotA/B stator to the MS-ring and flagellar rod. The figure shows a map of the torque transmission pathway highlighted with black arrowheads and coloured from blue (stator) to red (MS-ring). Torque transfer begins with the interaction between the MotA/B stator and the torque helix of FliGD5 of the C-ring. Interactions across the FliG subunit allow the torque to be transmitted to FliGD1, where there is a direct interaction with FliF. This is expected to turn the MS-ring and the flagellar rod within.
Extended Data Fig. 1
Extended Data Fig. 1. CryoEM workflow for the CCW C-ring.
(a) Workflow for the reconstruction of the 34-fold symmetric counterclockwise C-ring at 4.0 Å resolution. Following 2D classification, only particles containing both MS- and C-rings were retained. Symmetry could not be unambiguously classified from the combined MS- and C-rings. To identify the structure of the C-ring, MS-rings were removed through particle subtraction from the micrograph. This was saved as a separate dataset so that the original micrographs were retained. Classification of the separate C-ring revealed rings with 33- to 36-fold symmetry dominated by a prevalent of a 34-mer (~50% of particles). The 34-mer C-ring was subjected to C34 heterogeneous refinement and had a final overall resolution of 4.0 Å after this procedure. (b) To refine the associated MS-ring, the particles containing 34-mer C-rings were re-identified in the original micrographs and the C-ring was subtracted from these 34-mer particles. The subsequent heterogeneous refinement process used five parallel calculations to individually impose multiple C33-C36 symmetries. Among these, only the C33 MS-ring refinement (58% of particles) yielded distinct secondary structure features. The remaining particles (42%) did not classify as any observable symmetry. To ensure that we had not missed a subset of MS-rings with other stoichiometries, we removed the C33 particles from the calculation and separately imposed C34, C35, and C36 symmetry. This did not result in a class with interpretable density. To identify why 42% of the MS-rings that were bound to C-rings could not be classified, we re-evaluated the raw micrographs. We identified that the MS-rings that could not be classified were associated with micrographs that had thinner ice, suggesting a preferential orientation with the MS-ring at the air-water interface. We cannot, however, exclude that other symmetries exist at lower abundance in our samples or in the biological system. Notably, C33 symmetrization revealed the RBM3 and β-collar, but masked the details of RBM1 and RBM2. To achieve high-resolution insights into all domains, we subsequently conducted refinement using C11 symmetrization.
Extended Data Fig. 2
Extended Data Fig. 2. Assignment of the CCW pose of the C-ring.
(a) Surface representation of the C-ring colored by local resolution. The highest resolution (2.9 Å) is in blue, and the lowest resolution (6.6 Å) is in red. (b) AlphaFold models of individual domains were used as starting points in building the structure. The structures are colored by confidence from blue (very high confidence) to orange (very low confidence). (c) Comparison of the final model from the cryoEM structure to isolated domains from crystal structures of homologs. Insets show different regions of the structure. FliGD4-D5 is superposed with the equivalent domains from Thermotoga maritima (rcsb.org/structure/1lkv). FliGD3 is shown superposed with the equivalent domain from Helicobacter pylori (rcsb.org/structure/3usw). FliGD1-D2 and FliFC are shown superposed with the equivalent domains from T. maritima (rcsb.org/structure/5tdy). The FliMmid domain and FliGD3 are superposed with the equivalent domains from H. pylori (rcsb.org/structure/4fq0). FliMC and FliN are superposed with the FliN dimer from T. maritima (rcsb.org/structure/1yab) and the fused FliMC-FliNC dimer from Salmonella enterica (rcsb.org/structure/4yxb).
Extended Data Fig. 3
Extended Data Fig. 3. Density for subunits of the C-ring in the CCW pose.
Density for: (a) FliFC; (b)–(g) regions of the FliG subunit; (h)–(l) regions of the FliM subunit (m)–(o) each of the three FliNC domains. The resolution of FliMmid is < 3 Å. In panel (i), the side chains are shown for FliM51-75 to highlight the map quality.
Extended Data Fig. 4
Extended Data Fig. 4. Validation of the assembled C-ring.
(a)–(f) Individual domains of the FliG subunit concur with the domains of this structure. However, the global appearance of FliG differs due to the different interdomain angles. (g) Locations of flagellum-deficient mutations,,, in the context of two adjacent protomers of the C-ring. All flagellum-deficient mutations are highlighted with a sphere. Flagellum deficient mutations that map to the interior of folded domains and are likely to prevent flagellar assembly through misfolding of an individual subunit is colored black. Flagellum deficient mutants that interact with adjacent protomers in the ring are the same color as the associated chain. Insets highlight key select intersubunit interactions (yellow dash) that may be disrupted with these mutations. (h) comparison between the CCW pose and a CCW tomogram from Vibrio algintolyticus.
Extended Data Fig. 5
Extended Data Fig. 5. Workflow for cryoEM structures of CW C-rings.
(a) Workflow for the unbound CW C-ring at 4.6 Å resolution. Representative 3D reconstructions used symmetry expansion followed by local refinement. (b) Workflow for the CW C-ring with a bound partner at 5.9 Å resolution. Representative 3D reconstructions used particle subtraction followed by non-uniform refinement, symmetry expansion, and local refinement. The raw micrographs for CW rings all had thinner ice than the micrographs for CCW rings. Because of this, the MS-ring structure could not be classified in any case, and the structure could be determined without the application of particle subtraction.
Extended Data Fig. 6
Extended Data Fig. 6. Structure of the MS-ring.
(a) The MS-ring is composed entirely of copies of the FliF subunit. Past structures have been determined with different stoichiometries,,,,, although there remains debate on whether there is a biologically-relevant exact stoichiometry. Three views of representative cryoEM density (blue mesh) for the 33-mer MS-ring, calculated at 3.4 Å resolution and superposed onto the final model. RBM1 is red, RBM2 is olive and RBM3 is blue. Regions of density corresponding to other positions of RBM1 but where the quality was not sufficient to assign are circled. (b) Model of the MS-ring highlighting the relative positions of the ring-building motifs.
Extended Data Fig. 7
Extended Data Fig. 7. Bidirectional rotation of the C-ring by MotA/B.
MotA subunits are labeled A – E in the panels. (a) Comparison of MotA/B, and MotA cryoEM structures identifies varying levels of asymmetry that change the width of the cleft between subunits. The most symmetric structure is that of isolated MotA. In the context of past biochemistry and structures of MotA/B, the MotA/B stator binds to the torque helix on FliGD5. A compelling model would use a cleft between MotA subunits, a concept with parallels to interlocking cogwheels in macroscopic motors. However, these molecular cogwheels in the chemotaxis machinery undergo shape changes during function, which may benefit from the symmetry mismatch of MotA/B. One possibility is that the open MotA/B clefts, allow rapid binding or release of the torque helix without the need for a rate-limiting induced-fit process. Cleft closure would grasp the FliGD5 torque helix tightly. (b) Complementary electrostatics and sterics of the torque helix of FliGD5 and the MotA/B stator. An electrostatic surface representation of the FliGD5 domain shows that the torque helix is presented as an isolated feature and is negatively charged. (c) A schematic mechanism for bidirectional MotA/B-dependent rotation of the C-ring by moving the MotA/B binding site from the outside to the inside of the ring.
See this image and copyright information in PMC

References

    1. Beeby M, Ferreira JL, Tripp P, Albers SV, Mitchell DR. Propulsive nanomachines: the convergent evolution of archaella, flagella and cilia. FEMS Microbiol. Rev. 2020;44:253–304. doi: 10.1093/femsre/fuaa006. - DOI - PubMed
    1. Mondino S, San Martin F, Buschiazzo A. 3D cryo-EM imaging of bacterial flagella: novel structural and mechanistic insights into cell motility. J. Biol. Chem. 2022;298:102105. doi: 10.1016/j.jbc.2022.102105. - DOI - PMC - PubMed
    1. Minamino T, Kinoshita M. Structure, assembly, and function of flagella responsible for bacterial locomotion. EcoSal Plus. 2023;11:eesp00112023. doi: 10.1128/ecosalplus.esp-0011-2023. - DOI - PMC - PubMed
    1. Zhou J, Lloyd SA, Blair DF. Electrostatic interactions between rotor and stator in the bacterial flagellar motor. Proc. Natl Acad. Sci. USA. 1998;95:6436–6441. doi: 10.1073/pnas.95.11.6436. - DOI - PMC - PubMed
    1. Johnson S, et al. Symmetry mismatch in the MS-ring of the bacterial flagellar rotor explains the structural coordination of secretion and rotation. Nat. Microbiol. 2020;5:966–975. doi: 10.1038/s41564-020-0703-3. - DOI - PMC - PubMed

Publication types