Hey flagellum, shift into reverse gear!

by Christoph
 

In a classical manual gearbox, a gearwheel – occasionally called "idler gear" –  is inserted between the countershaft and the drive shaft. This reverses the direction of rotation of the drive shaft. Flagella of E. coli and Salmonella have found a smart solution to do without an additional gearwheel in their gearbox: they run their engine along the outside of a gearwheel or, if they reverse the direction of rotation, on its inside.
 

We often do not consider small things like bacterial flagella with the attention they deserve (there are definitely too many small things). The last time Mike Dermont did this was six years ago in his piece Flagellar Motors: How a Bacterium Shakes its Groove Thing. When I saw the video shown as Figure 6 here in a behind the paper article by Steven Johnson and Justin Deme a few days ago, I thought it was time to turn my attention to flagella again. So, here it goes…  and I thought that this time I would walk you through six illustrations. Alas, none of them are big enough for you to read in all the necessary detail. Please click on the images, or tap them if you are reading on mobile, and they will open in a new window.

 

Figure 1. See text for details. Source. Frontispiece: Scre­en­shot from animation demonstrating gearing in the Sal­monella flagellum. Source

Figure 1  Who in the cell manually shifts into reverse gear? Changes in chemo-effector levels in the cell's periplasm are detected by transmembrane chemo­receptors (usually arranged as "che­mosensory arrays", see here in STC). In the cytoplasm, these "signal" by sequentially in­duced conformational changes via CheW to the chemotaxis histidine kinase CheA. In response to de­creased attractant concentration, the chemoreceptors trigger CheA autophosphorylation. Phos­phorylated CheA (CheA–P) in turn phosphorylates either of its cognate response regula­tors, CheB and CheY.
 

CheY-P is the key molecule here as it binds to the flagellar motor and promotes the switch in the direction of rotation from anticlockwise (ccw) to clockwise (cw). As with all biological signal transmissions, there are also built-in safeguards for signal quenching: CheB–P is a methylesterase that mediates adaptation in conjunction with the chemotaxis methyltransferase, CheR, for ex­ample receptor clearance. CheZ is a specific phosphatase that dephosphorylates CheY–P, allowing rapid signal termination.

 

Figure 2. See text for details. Modified from Source

Figure 2  The counterclockwise (ccw) and clockwise (cw) conformations of the motor exist in equi­librium. In the absence of CheY–P, ccw rotation is strongly favored. CheY–P binding to FliM and FliN components of the switch complex shifts the equilibrium in favor of clockwise rota­tion. There are assumed to be ~34 binding sites for CheY–P, and as more of these sites be­come oc­cupied, the probability of cw rotation increases cooperatively (but see below).
 

What is labeled "switch complex" in the figure is now referred to as the C‑ring. In E. coli and Salmonella, the C‑rings, which lie directly under the inner membrane, are almost without exception collars with 34 "chain links" in a circular arrangement; the chain links are the FliG, FliM and FliN subunits stacked on top of and inside each other. FliF subunits couple the C‑ring to the overlying MS ring (labeled "rotor") by interacting with FliG in a stoichiometry that has not yet been conclusively deter­mined. Note that the MotAB complexes, often referred to as "stator" in diagrams, are connected to the (periplasmic) peptidoglycan layer via their MotB subunits, but have no connection to the C-ring. They have, and this will become important, crucially important, further down...

 

Figure 3. a Using AlphaFold2 to predict the interactions between CheY–P (yellow surface, phosphate shown as space filling atoms) and the C-ring (silver cartoon) ge­nerates an arrangement of the FliMM:FliGM domains that matches that seen in the cw state (dark green) but not the ccw state (light green, overlaid based on FliMM to highlight relative movement of FliGM). The previously determined interaction between the FliM N-terminus (silver helix on CheY–P surface) and CheY–P is high­lighted. b Mapping the location of CheY–P and the full N-terminus of FliM from the AlphaFold model to a set of three neighbouring cw subunits reveals the nestling of CheY–P into the C-ring at a subunit–subunit interface and with interactions spread across two subunits. Source

Figure 3  To nail down the localization of the binding of CheY–P to the C-ring, Johnson et al. (2024) employed model building with AlphaFold2.0 for various combinations of the FliG, FliM and FliN domains. All combinations that involved the FliMM domain produced models with a consistent CheY binding interface at the top of FliMM (the indices indicate definded domains of the proteins), with the phosphorylated residues (D57) pointing towards the FliGM:FliMM inter­face, compatible with binding the FliM N-terminus, as seen in crystal structures (Figure 3a). Interestingly, overlay of the different structures via the FliMM domain demonstrated that the FliGM domain of the ccw state is shifted approximately 4 Å compared with the cw state. Models of CheY–P bound to FliGM:FliMM produced an interface consistent with the cw structure, sug­gesting a mechanism where CheY–P binding stabilizes the cw state (Figure 3b).

 

Figure 4. Structure of the ccw rotating state (top) and of the cw rotating state (bottom) of the Salmonella flagellar C-ring. Using a mask around 3 subunits and refining symmetry expanded particles generates a higher-resolution volume (ccw: 3.2–4.5 Å across the subunit, cw: 3.3–5.2 Å across the subunit) for the central subunit, allowing building and refinement of atomic models. Source

Figure 4  CryoET analysis of Salmonella basal bodies with intact C-rings isolated from strains locked in either the ccw state or the cw state showed Johnson et al. (2024) that the overall properties of the C-ring structure are conserved between the two rotational states, including the expected C34 symmetry (34 "chain links" in the ring), despite heterogeneity in the local arrangement of FliG, FliM, and FliN (compare both zooms on FliM in Figure 4).

Fitting crystal structures into the tomogram‑derived volumes shown in Figure 4 revealed that the membrane proximal region is formed from FliG protomers, tethered via a FliF C-terminus to the MS‑ring. The FliGM domain forms the main interaction with a layer of FliM protomers underneath a FliG "ring", and the most membrane distal layer is formed from a continuous "ring" of interlocked FliMC:FliN3 subcomplexes. Lateral contacts in both ccw and cw states are dominated by domain swaps at all major layers, many of which involve domains of FliG. There is a change in the outer perimeter wall of FliGM:FliMM:FliMC:FliN3, with the cw conformation (Figure 4, bottom) being more upright than the ccw conformation (Figure 4, bottom). The FliGM:FliMM domains rotate through an angle of 22° as a rigid object (reflecting the binding of CheY?), while the FliMC:FliN3 spiral pulls up slightly and hinges inwards relative to FliMM.
 

Again fitting crystal structures into the tomogram‑derived volumes, Johnson et al. (2024) found a substantial difference for the FliGM domain, where the cw-locked state at residues 169–171 leads to a shift of the FliGMC helix by almost a full helical turn. This shift does change some of the contacts at the FliGM:FliMM interface, but the helix remains bound to FliMM. The C-terminal end of the FliGMC helix leads into a less well-ordered linker that connects the two domain-swapped portions of the FliGM domain, and this is a major hinge point between the two states. This linker, in turn, contacts the corresponding region of a neighboring FliM copy, thereby highlighting the importance of this region of the structure in tying the ring together and propagating lateral structural changes. Also at the C-terminus of FliGMC is a major contact to the helix that links FliGN and FliGM, termed FliGNM.
 

The most striking structural change the authors observed affects the positional change of this FliGNM subdomain between the two states. In the ccw form, FliGNM folds back through 180° via a turn involving residues 101–106, whereas the cw form is a straight helix that runs from 87 to 112. This leads to the entire FliGN domain rotating through 180° between the two states. The FliGN domain maintains its overall domain‑swapped conformation, but residues 72–99 make contact with N−1 rather than N+1 (N stands for neighboring FliG). Known mutations that induce a strong cw bias are localized to FliGNM and are clustered around the hinge point and the FliGMC interaction point, supporting the notion that the observed structural change is one of the key driving forces of switching. You can certainly see this 180° flip very clearly in the comparison of the ccw and cw conformations in Figure 4, even if you can't see the helices involved there.

 

Figure 5. See text for details. Source

Figure 5  It is known for decades that the flagellar motor creates torque "by interactions be­tween the rotor protein FliG, located at the intersection between the MS and C rings, and the stator units attached to the cell wall. Each (stator) unit is believed to contain two copies of MotB and four copies of MotA in proton-driven motors of E. coli ... and functions as an ion channel. Ion flux through these channels powers the motor." (ref.) It was only remotely in­teresting that over time the "stator" changed its stochiometry to MotA5B2, but eye-opening that it has to be "considered as a motor itself: it converts the electrochemical potential energy from the ion motive force into mechanical torque. Upon recruitment to the basal body and cell wall binding, the stator units undergo a conformational change from an inactive/plugged state into an activated/unplugged state. In the unplugged state, the flux of ions through the stator unit channel energizes the rotation of the rotor." (Hu et al. (2021)). Ion flux from the periplasm into the cytoplasm through the (yet to be fully characterized) ion channel within the MotA5B2 "stator" re­sults in a rotation of the MotA subunits in steps of ~36° per transported proton, which exerts torque on the FliG subunit in the C-ring. This motor keeps its name "stator" for historical reasons.
 

Johnson et al. (2024) set out to solve the structure of the MotA interaction with the FliGC down to the contacting helices, in particular the FliGC "torque" helix. For this, they fused the MotA homolog from C. sporogenes with Salmonella FliGC separated by spacers of varying lengths. In these chimeras, FliGC domains were only seen to bind between closely spaced MotA pairs, that is, a pincer grip of a pair of adjacent MotA domains is apparently required for the binding of FliGC. This interaction sandwiches the FliGC domain between the MotA cytoplasmic domains and presents the "torque" helix into a cavity. Further modelling revealed that MotA5B2–FliG on the ccw and cw C-rings switches the stator complex from the outside to inside the C-ring.
 

In Figure 5ab, the MotA5B2–FliGC complex is overlaid on the same FliGC domain with the ccw (a) and cw (b) C-ring subunits. The zoom boxes show that the FliGC domain is not structurally altered in the context of the C-ring subunit and overlays well. However, the rotation of the FliGC domain in the context of the different rotational states results in the MotA5B2 complex being presented on the external (ccw) or internal (cw) face of the C-ring. In Figure 5c, aligning the ccw and cw structures based on the invariant FliF component of the rotor revealed that the Salmonella C-ring does not change diameter as part of switching (overlaid volumes with ccw (light green) and cw (dark green), top) and that, in addition, the lower two panels show the MotA5B2–FliGC complex in the context of the ccw and cw C-rings, shows that the stators move from pushing the C-ring on the outside to the inside of the C-ring, thus explaining how a stator that rotates in a single direction can power bi-directional movement of the C-ring.

An aside thought. Isn't it fun to speculate whether the switch of the C-ring from the ccw into the cw conformation triggered by CheY binding to FliM is constrained or stimulated by the torque exerted on FliGC captured in the MotA tweezers?

Note that the cross-section shown in Figure 5c has only two MotA5B2 stators are in contact with the C‑ring via FliGC. It had been found earlier that up to at least 11 "stators" can dock onto one C-ring under physiological conditions. High (mechanical) load stabilizes C-ring-bound "stator" units in both ccw and cw rotating motors, increasing their number over a timescale of a few minutes. Conversely, low torque decreases the lifetime of motor-bound stator units. This process enables the "stator"-plus-C‑ring motor to adapt its torque output (via FliF to the MS-ring) to the demand placed on it by external load by modifying its composition (references in Wadhwa & Berg (2022)). It's a bit like switching individual cylinders on or off as required in an eight-cylinder internal combustion engine (the technical term is variable displace­ment).

 

Figure 6. Animation demonstrating gearing in the Salmonella flagellum. A small pentameric cog (MotA) rotates in response to the proton-motive force (pmf). The teeth of the small cog drive rotation of a large cog formed by the C-ring of the flagellar basal body, which in turn drives the rotation of the flagellum. Structural transitions of the C-ring in response to chemotaxis regulators remodel the teeth of the large cog, reversing the direction of rotation. Source

Figure 6  Johnson et al. (2024) summarize their results in a model for "gear shift­ing" in the flagellar motor in four steps. In the animated Figure 6 these four steps can be seen in loop, and you only have to add CheY–P yourself.

1. In the absence of CheY–P, the invariably clockwise-rotating MotA5B2 motor "sta­tor" interacts with the external face of the C-ring to drive ccw rotation and tran­sien­tly tensions the interacting subunit.

2. CheY–P binding at a transiently tensioned subunit transitions the FliG domain, which swaps to the cw state and cooperatively drives the spread of a conforma­tional change around the C-ring.

3. In the presence of CheY–P, the "stator" interacts with the internal face of the C-ring to drive CW rotation and transiently tensions the interacting subunit.

4. CheY–P dissociation from a transiently tensioned subunit transitions the FliG domain, which swaps to the ccw state and cooperatively drives the spread of a conformational change around the C-ring.
 

Despite all the enormous progress in cryo-electron tomography (cryoET), which now makes it possible to track conformational changes down to individual helices and loops of interacting proteins, there is no technology to "watch" this in real time. The actual "switch" between two confor­ma­tions, how flagella shift gear into reverse, can only be modeled. But still.
 

I should not fail to mention that in the same issue of Nature Microbiology, Singh et al. (2024) published their study that essentially came to the same conclusions as the study by Johnson et al. (2024) presented here, even if there are differences in details between the two. Both were familiar, of course, with the work of Chang et al. (2020), who studied flagella movement in Borrelia and already presented a model for switching to reverse gear by positional swapping of the MotA5B2 "stator" on the C-ring.
 

If the animation in Figure 6 is too minimalistic for your taste, I recommend watching the full-blown 3D animation C-ring on the Motor Goes Vroom­VroomVroom by Singh et al. (2024) (but for heaven's sake mute your speaker, the sound of the wooden ratchet is nerve‑wracking). It is not even important who published first. Both studies build on years of work by many other research groups involved and of course mention this. There has always been a stimulating combination of both collaboration and competition in the loose "flagellar community" (not to be confused with religious flagellants). And when two competing teams come to very similar conclusions within a year or so, this can only be positive for pro­gress in understanding flagellar movement.
 

Even if we now have a fairly accurate idea of how E. coli and Salmonella "shift into reverse gear" their flagella... other flagella-bearing bacteria may very well have evolved other, similarly smart variations on this theme. Probably none with an additional cogwheel in the gearbox, though.
 

Do you want to comment on this post? We would be happy about it! Please comment on Mastodon, Bluesky, or on 𝕏 (formerly Twitter).