The Bacterial Flagellar Motor: A Rotary Engine in a Cell

A wheel in a world of levers

Almost every motor in biology is a lever. Muscle shortens, kinesin walks, dynein bends — they all convert chemical energy into linear pushes and pulls. The bacterial flagellum is the great exception: it is a wheel and axle that turns round and round without limit. This was not obvious. Until the early 1970s biologists assumed the flagellum lashed like a swimmer's tail. Then Howard Berg, and independently Silverman and Simon, glued flagella to glass slides by antibodies. The tethered filament could not move, so the whole cell body was seen to spin around it — direct proof that the base is a rotary engine with a spinning rotor and a fixed stator, not a beating whip.

Rotation is not a quirk; at bacterial scale it is the right engineering choice. A 2-micrometer cell swimming through water lives at a Reynolds number of about 10⁻⁴, a regime Edward Purcell described as trying to swim through cold tar. Inertia is meaningless: if the motor stopped, the cell would coast less than the width of a single atom before halting. Reciprocal, back-and-forth motion cancels itself out and produces no net travel, so a bacterium cannot swim by flapping. A continuously turning corkscrew, by contrast, screws steadily through the viscous medium — and it drives the cell forward at ~20–30 µm/s, roughly 10–15 body-lengths per second, the scale equivalent of a person swimming faster than they can run.

Anatomy of the engine

The motor spans the entire cell envelope of a Gram-negative bacterium and divides cleanly into the parts of a real machine:

• The rotor (the part that spins). An MS-ring of ~34 copies of the protein FliF sits in the inner membrane. Bolted beneath it is the bell-shaped C-ring, built from FliG, FliM and FliN. FliG (~26 copies) is the business end: its charged residues are what the stators physically push on to make torque, and its conformation sets the direction of turn.
• The stators (the fixed parts that push). Each torque unit is a MotA₅MotB₂ complex — five MotA subunits wrapped around a MotB dimer — anchored to the rigid peptidoglycan wall. As many as a dozen stators can dock in a ring around the rotor, each an independent little engine.
• The driveshaft and bearings. A rigid rod (FlgB/C/F/G) transmits the spin outward through the P-ring and L-ring (FlgI, FlgH), which act as bushings through the wall and outer membrane.
• The universal joint and propeller. A curved, flexible hook (~120 copies of FlgE, ~55 nm long) couples the driveshaft to the propeller at any angle, exactly like a car's constant-velocity joint. The propeller itself is the filament: ~20,000 copies of one protein, flagellin, self-assembled into a hollow helical tube ~20 nm wide and up to 10–15 µm long — many times the length of the cell.

Torque from a stream of protons

The most counterintuitive fact about the motor is its fuel. It does not burn ATP. It runs on the proton-motive force — the same electrochemical proton gradient across the inner membrane that respiration builds and that ATP synthase also taps. In E. coli this gradient is about −150 to −170 mV; protons "want" to flow back into the cell, and the motor turns that inward flux into rotation.

Cryo-EM structures published in 2020 finally showed how. Each stator is a tiny ion turbine: a proton entering the MotB channel binds a single conserved aspartate (Asp32), and this triggers the surrounding MotA pentamer to rotate one notch around the fixed MotB shaft. The inner face of that spinning MotA ring carries charges that mesh with the charged rim of FliG, so the stator meshes with the rotor like one gear driving another. Repeat with the next proton, the next stator, and the rotor sweeps around.

• At low speed the motor advances in ~26 discrete steps per revolution — resolved in single-molecule experiments — matching the periodicity of the FliG ring.
• Coupling is tight: on the order of a thousand protons flow per turn, and each one's downhill free energy (~0.17 eV) is handed almost losslessly to the rotor at high load.
• Because the stators act in parallel and independently, torque adds up roughly linearly with stator number: a fully occupied motor with ~11 units produces about 11× the push of a single one, up to a peak of ~1,200–1,400 pN·nm.

The physics of speed — and a self-adjusting gearbox

The performance figures are astonishing for something assembled from floppy protein. A proton-driven E. coli motor free-swims at ~100–300 Hz (6,000–18,000 rpm). Marine bacteria such as Vibrio alginolyticus run their flagella on a sodium gradient instead, and reach ~1,700 Hz — over 100,000 rpm, several times faster than a Formula 1 engine at full throttle. Yet the mechanical power involved is minuscule — on the order of a femtowatt (~10⁻¹⁵ W) — because the forces, while enormous per unit mass, act over nanometer scales.

The motor's torque–speed curve explains its behavior. Torque stays on a nearly flat plateau from stall up to a "knee" at a few hundred hertz, then falls steeply toward the top speed. That shape means the engine delivers maximum force precisely when the load is heaviest — when the cell is wading through mucus or gel — and only spins fast when there is little to push against. Near stall, where nearly all the protons' energy becomes work, the motor operates close to thermodynamic reversibility, an efficiency few human machines approach.

Most remarkable of all, the gearbox tunes itself. The stators are not permanent: they are pulled from a membrane reservoir and dock onto the motor when the mechanical load rises, then diffuse away when it falls. A motor spun against heavy resistance recruits more stators within seconds; lighten the load and it sheds them. The flagellar motor is therefore a mechanosensor with an automatic transmission, matching its power output to the job in real time.

Throwing it into reverse: run-and-tumble

The second astonishing trick is instantaneous reversal, and it is the whole basis of how bacteria find food. E. coli carries several flagella. When all their motors turn counterclockwise (CCW), the left-handed helices wrap into a single coherent bundle that drives a smooth, straight run lasting ~1 second. When one or more motors flip to clockwise (CW), that filament's handedness transforms, the bundle flies apart, and the cell tumbles chaotically in place for ~0.1 second, re-aiming in a random new direction.

Alternating runs and tumbles is a random walk — but a biased one. A molecular signaling circuit lengthens the runs whenever conditions are improving, so the cell drifts, on average, up gradients of nutrients and away from toxins. This is chemotaxis. The switch itself is thrown by the phosphorylated protein CheY-P, which diffuses to the motor and binds the FliM subunits of the C-ring. Enough bound molecules make the whole ring cooperatively snap into its CW conformation, changing which surface of FliG the stators press against. The response is ultrasensitive (Hill coefficient ~10, so a small change in signal flips the motor almost all-or-none) and completes in under a millisecond. A slower methylation system readjusts the receptors afterward, giving the cell a chemical memory a few seconds long so it responds to changes rather than absolute levels.

A machine that builds itself — and what it is not

No factory assembles the flagellum; it grows itself from the base outward. At its core sits a dedicated type III secretion system, the same export machinery that disease-causing bacteria repurpose into a syringe to inject toxins into host cells. Flagellin subunits are never dumped into the surroundings — they are unfolded and threaded down the ~2-nm channel running the whole length of the filament, then folded into place at the growing tip under a cap protein (FliD). Because each new subunit must travel the entire length, growth slows as the filament lengthens, and a filament tens of micrometers long can take an hour to finish. A molecular ruler protein, FliK, measures the hook to ~55 nm and then switches the export apparatus from making hook to making propeller — assembly in the correct order, without a blueprint.

Three points are worth underlining because they are so often confused. (1) This has nothing to do with the flagellum of a sperm cell: eukaryotic flagella and cilia are bundles of microtubules that bend by burning ATP through the motor dynein — a case of convergent function, not shared parts. (2) The filament does not wave; it is a rigid corkscrew that is turned. (3) The motor is not "irreducibly complex": its stator is homologous to ion-driven ExbBD/TolQR systems, its export core to the type III injectisome, giving evolution ready-made modules to co-opt. Today the machine matters beyond basic science — motility drives virulence (Helicobacter pylori burrows through stomach mucus; the immune sensor TLR5 detects flagellin), and engineers study it as a blueprint for self-assembling, chemically fueled nanomachines.

A true rotary motor versus its biological look-alike and its human-built cousin

Feature	Bacterial flagellar motor	Eukaryotic flagellum / cilium	Human-made electric motor
Energy source	Proton (or Na⁺) motive force across the membrane	ATP hydrolysis	Electric current
Type of motion	Continuous rotation of a rigid propeller	Bending / whipping waves along the shaft	Continuous rotation
Working parts	Protein rotor (C-ring) + protein stators	Microtubule doublets slid by dynein motors	Magnets, copper coils, bearings
Speed	~100–1,700 revolutions per second	~10–60 beats per second	up to thousands of rev/s
Reversibility	Flips CCW↔CW in < 1 ms to steer	Waveform modulated, not truly reversed	Reversible with polarity switch
Origin	Self-assembles from ~2 dozen genes	Templated by the centriole/basal body	Manufactured and assembled