Flagella and Bacterial Motility

Bacterial motility by flagella rests on two separable systems that must be understood together: the motor and the switch. The motor is the flagellum itself — a nanoscale rotary engine embedded in the cell envelope that converts an electrochemical ion gradient into mechanical rotation, spinning a long helical filament that acts as a propeller. This is fundamentally different from eukaryotic flagella (as in sperm or Chlamydomonas), which bend and whip using internal microtubule sliding powered by ATP-driven dynein; the bacterial flagellum has no internal contractile machinery at all — it is a passive, rigid helix rotated from its base, and it runs on ion flux rather than ATP hydrolysis. The switch is the chemotaxis system, a signal transduction network that senses chemical gradients across the cell surface and controls which direction the motor spins. Because a bacterium is too small to sense a spatial concentration gradient across its own length (thermal diffusion would swamp any such difference over a distance of a few micrometers), it instead performs temporal sensing: it compares the concentration of an attractant or repellent now to what it was a few seconds ago as it swims, and adjusts its probability of tumbling accordingly. The output of chemotaxis is not steering in the way a car's wheels turn; it is a statistical bias on a random walk, converting an undirected zigzag into net drift up an attractant gradient or down a repellent one — a process sometimes called a biased random walk. Together these systems illustrate a recurring theme in bacterial physiology: cells that are too small and too fast-diffusing to use elaborate sensory organs instead couple simple binary decisions (rotate clockwise or counterclockwise) to time-averaged chemical information, achieving surprisingly sophisticated navigation with a minimal parts list.

Flagellar structure, from outside in. The visible part of the flagellum is the filament, a hollow helical tube roughly 20 nm in diameter and up to 15 μm long, built from thousands of copies of a single protein, flagellin (FliC), arranged in 11 protofilament rows. The filament is capped by FliD, which folds newly arriving flagellin subunits into place as they are pumped up through the hollow core — flagellar growth occurs at the tip, not the base, meaning a monomer synthesized in the cytoplasm travels the entire length of the completed structure before being incorporated. The filament attaches through a flexible universal joint, the hook (FlgE), to the basal body, the actual rotary engine embedded in the envelope. In Gram-negative bacteria the basal body includes:

Gram-positive bacteria, lacking an outer membrane, have only the MS and C rings plus a rod, since there is no L or P ring layer to bush through.

Torque generation: the stator and rotor. Surrounding the MS/C ring rotor are 5–11 independent stator units, each built from MotA and MotB proteins (or PomA/PomB in Na⁺-powered marine species like Vibrio). MotB is anchored to the peptidoglycan, holding the stator fixed in the cell wall while the rotor spins beneath it. Protons flow through a channel formed at the MotA–MotB interface, down their electrochemical gradient (the proton motive force, typically −150 to −200 mV in neutrophilic bacteria). Each proton-binding and release event drives a conformational change in MotA that pushes against FliG on the rotor's C ring, generating a small increment of torque — analogous to a gear tooth engaging. With multiple stators working around the rotor's circumference and roughly 1,000–1,200 protons required per revolution, the summed torque is enough to spin the rotor at up to 300 revolutions per second (E. coli) and, in some marine species using Na⁺, over 1,000 revolutions per second. Critically, no ATP is hydrolyzed at the motor — this was one of the most surprising findings in the field and is discussed under the evidence below. ATP is used elsewhere in flagellar biology (in the export apparatus that builds the structure), but the rotation itself is a direct mechanical transduction of an ion gradient, exactly the kind of chemiosmotic energy coupling that also drives ATP synthase.

Building the flagellum: the type III secretion export apparatus. The flagellum assembles from the inside out via a flagellar type III secretion system (fT3SS) at its base, structurally and evolutionarily related to the injectisome that pathogenic bacteria use to inject effector proteins into host cells. Unfolded flagellar subunits (rod, hook, and filament proteins) are threaded through the hollow channel at the core of the growing structure and self-assemble at the distal tip, powered by ATP hydrolysis (FliI, an ATPase related to the F₁ subunit of ATP synthase) plus the proton motive force. Assembly proceeds in strict order — MS ring and export apparatus first, then rod, then hook, then filament — regulated by a genetic hierarchy of master, middle, and late flagellar genes, so that late (filament) genes are not transcribed until the hook-basal body is structurally complete, a checkpoint enforced by the anti-sigma factor FlgM.

Switching direction: CCW versus CW. The C ring is not just a mechanical coupler; it is the rotational switch. In its default state the motor rotates counterclockwise (CCW) (viewed from outside the cell, looking toward the cell body), which causes multiple peritrichous flagella to coalesce into a single rotating bundle that pushes the cell forward in a smooth run. When the switch flips to clockwise (CW) rotation, the bundle flies apart because the helical filament undergoes a polymorphic transformation (a change in the supercoiling of the flagellin helix under mechanical load) from a normal left-handed helix to a curly, right-handed form, and the cell tumbles, reorienting randomly. The switch between CCW and CW bias is controlled by the phosphorylated chemotaxis response regulator CheY-P binding directly to FliM on the C ring: more CheY-P favors CW rotation and tumbling; less CheY-P favors CCW rotation and smooth runs.

The chemotaxis signaling cascade. Chemoreceptors called methyl-accepting chemotaxis proteins (MCPs), embedded in the cytoplasmic membrane and clustered at the cell pole, bind attractants (e.g., aspartate, serine) or repellents either directly or via periplasmic binding proteins. MCPs are physically coupled through an adaptor protein, CheW, to a histidine autokinase, CheA. In the unstimulated or repellent-bound state, CheA autophosphorylates using ATP and transfers the phosphoryl group to two response regulators: 1. CheY, which diffuses to the flagellar motor and binds FliM, promoting CW rotation and tumbling. 2. CheB, a methylesterase that, once phosphorylated, removes methyl groups from the MCP. When an attractant binds an MCP, CheA autophosphorylation is inhibited, CheY-P falls (it is rapidly dephosphorylated by the phosphatase CheZ), and CCW rotation (smooth swimming) predominates — the cell runs longer in the improving direction.

Adaptation, the source of temporal memory. A separate constitutively active methyltransferase, CheR, continuously adds methyl groups to specific glutamate residues on the MCP cytoplasmic domain, which increases CheA-activating ability of the receptor. CheB-P removes those methyl groups. The competing, slower action of CheR against the faster, stimulus-driven action of CheB-P creates a negative feedback loop: after an attractant binds and signaling falls, the receptor slowly becomes more methylated, restoring baseline CheA activity over roughly 1–10 seconds even though the attractant is still bound. This adaptation is what gives the cell its "memory" — it responds to changes in concentration rather than to absolute concentration, allowing it to keep sensing gradients across many orders of magnitude of ambient concentration (a form of sensory adaptation analogous to receptor desensitization in eukaryotic signaling).

Peritrichous versus polar flagellation. Escherichia coli and Salmonella are peritrichous, with 4–10 flagella distributed over the cell surface that must coordinate rotation direction to bundle for a run. Vibrio cholerae and Pseudomonas aeruginosa are examples of polar flagellation, with one or a tuft of flagella at a single pole, often Na⁺-powered and generating extremely fast, directional swimming without bundling in the same way; steering in polar-flagellated cells frequently uses a "flick" of the flagellum as it switches direction rather than a symmetric run-tumble.

Berg's tethered-cell experiment (Howard Berg and Robert Anderson, 1973; and Silverman and Simon, 1974) proved rotation, not contraction, drives motility. Before this, some models proposed the flagellum functioned like a contractile eukaryotic tail. Berg and colleagues attached bacteria to a glass slide by their flagellar filament (using antibodies against flagellin to glue a single filament to the coverslip), so the filament itself was immobilized. If the cell body then rotated visibly under the microscope, this would prove the base of the flagellum, not the filament, was the site of rotary motion, and that the whole cell body was being spun by a rotating shaft. This is exactly what was observed: tethered cells spun steadily, reversing direction (CW/CCW) in a manner matching the known run/tumble switching frequencies, at rates consistent with motor rotation rather than any whip-like beating. This was direct, visual proof of a rotary motor.

Larsen, Adler, and colleagues (1974) showed rotation is powered by proton motive force, not ATP. Using Streptococcus and later E. coli mutants, and using ionophores (like CCCP, a proton uncoupler that collapses the transmembrane pH and electrical gradients without directly affecting ATP levels), they showed that motility stopped when the proton motive force was dissipated even when cellular ATP concentration remained high. Conversely, artificially imposing a proton gradient (or an electrical potential) across the membrane of ATP-depleted cells could drive flagellar rotation. This decoupling — motility tracks the ion gradient, not ATP pools — was the critical evidence establishing chemiosmotic, not chemical (ATP-driven), coupling for the flagellar motor, paralleling Peter Mitchell's chemiosmotic hypothesis for oxidative phosphorylation.

Direct measurement of torque and stepping (Berg and colleagues, using laser tweezers and high-resolution bead assays, 1990s–2000s). By attaching a small latex bead or a fluorescent marker to a truncated flagellar stub and tracking its rotation with sub-millisecond time resolution, researchers directly measured motor speed (up to ~300 Hz at high proton motive force), stall torque (roughly 4,000 pN·nm, remarkably constant across a wide range of speeds — a "constant torque" regime characteristic of a mechanically limited, not diffusion-limited, motor), and, at very low speeds, resolved discrete angular steps consistent with roughly 26 steps per revolution, matching the estimated number of FliG subunits around the C ring and the stepwise, proton-coupled conformational cycling of individual stator units.

Julius Adler's genetic dissection of chemotaxis (1960s–1970s) established the pathway components. Adler isolated non-chemotactic mutants of E. coli that could still swim (ruling out simple motility defects) but failed to accumulate in capillary tubes containing attractants — the classic capillary assay, in which a glass capillary filled with an attractant is inserted into a bacterial suspension, and after incubation, the number of bacteria that swam into the capillary is counted; wild-type cells accumulate to concentrations vastly (often 100- to 1,000-fold) higher than in the bulk suspension, while non-chemotactic mutants show no enrichment. By mapping these mutations, Adler and colleagues identified the che genes and distinguished chemoreceptors, which recognize specific attractants like serine or aspartate independent of their metabolism, from metabolism itself — showing that chemotaxis is a genuine sensory process, not simply cells swimming toward food they are consuming.

Macnab and Koshland's temporal comparison experiment (1972) proved bacteria sense gradients in time, not in space. Using a rapid mixing device (a stopped-flow apparatus) to expose Salmonella to a sudden, spatially uniform jump or drop in attractant concentration — eliminating any possibility of a spatial gradient across the cell — they observed a transient suppression of tumbling (for an increase in attractant) or a transient increase in tumbling (for a decrease), followed by return to baseline tumbling frequency within seconds. Because the stimulus was spatially uniform, the only information available to the cell was the change over time, proving directly that temporal sensing (comparing "now" to "a few seconds ago"), not simultaneous spatial comparison across the cell's two poles, underlies bacterial chemotaxis.

Example 1 — quantifying the biased random walk. Suppose a population of E. coli swims with a mean run time of 1 second in isotropic (no-gradient) buffer, tumbling on average every second and reorienting to a new random direction each time, giving an unbiased random walk. Now place the cells in a linear gradient of aspartate increasing toward the top of a chamber. Measurements (as in classic tracking-microscope experiments) show that when cells happen to swim up the gradient (improving conditions), mean run time roughly doubles to about 2 seconds, while when cells swim down the gradient (worsening conditions), mean run time stays near 1 second or drops slightly. Calculate the qualitative effect on net displacement. Because a cell moving up-gradient runs about twice as long before reorienting compared to a cell moving down-gradient, and reorientation after a tumble is roughly random with only a weak forward bias, the population's mean-square displacement acquires a directional component: over many run-tumble cycles, cells accumulate more net distance during up-gradient runs than they lose during down-gradient ones. If typical swimming speed is 20 μm/s, an up-gradient run of 2 s covers about 40 μm before a tumble, versus about 20 μm for a down-gradient run — a twofold asymmetry per run is sufficient, compounded over many cycles (each cell tumbles roughly once per second, so tens of run-tumble cycles occur per minute), to produce the several-hundred-fold accumulation of cells in a capillary assay's attractant-filled capillary relative to the bulk suspension, even though each individual run is still a straight line in a random direction and no single run "steers" toward the gradient.

Example 2 — reasoning through a cheZ null mutant phenotype. A student isolates an E. coli mutant with a complete deletion of cheZ, the gene encoding the phosphatase that dephosphorylates CheY-P. Predict the swimming phenotype, and explain the reasoning using the pathway logic. Without CheZ, CheY-P (produced continuously by basal CheA autophosphorylation activity, which occurs at a low level even without strong repellent signaling) cannot be efficiently removed, so CheY-P accumulates well above wild-type steady-state levels. Since CheY-P binding to FliM promotes CW rotation, elevated CheY-P shifts the motor's bias strongly toward CW rotation. In peritrichous cells like E. coli, sustained CW rotation prevents flagellar bundle formation (the filaments fly apart into a splayed, CW conformation), so predicted phenotype is a cell that is smoothly-swimming-defective and instead tumbles almost continuously, appearing to be non-chemotactic even though every individual signaling and motor component apart from CheZ is intact — a "smooth-swimming-null, tumble-locked" phenotype. This matches the actual published phenotype of cheZ mutants and illustrates why chemotaxis requires not just generating a signal (CheA → CheY-P) but also a mechanism (CheZ) to erase it quickly, so the motor bias can track rapid changes in receptor occupancy rather than saturating at one extreme.

Julius Adler, working at the University of Wisconsin–Madison from the 1960s onward, transformed bacterial chemotaxis from a descriptive curiosity — long known since the nineteenth century, when early microscopists like Wilhelm Pfeffer and Theodor Engelmann observed bacteria clustering around oxygen-releasing chloroplasts or capillaries of meat extract — into a rigorously genetic and biochemical discipline. Adler developed the capillary assay as a quantitative, reproducible measure of chemotactic accumulation and used it to screen for mutants, establishing that chemoreception of a compound (via a dedicated receptor) is a distinct cellular process from metabolism of that same compound, since some mutants could still metabolize an attractant but no longer sense it, and vice versa. This genetic dissection defined the che gene cluster and set the stage for later biochemical reconstruction of the CheA–CheY–CheW phosphorelay by Jeffry Stock, Melvin Simon, and others, and for Howard Berg's biophysical proof of the rotary motor itself.

Flagellar motility is a core virulence factor in many pathogens: Helicobacter pylori uses its flagella to burrow through stomach mucus to reach the epithelial surface, a step essential for colonization and a target for anti-virulence drug development; Vibrio cholerae motility contributes to penetration of the intestinal mucus layer before toxin production begins; and Pseudomonas aeruginosa flagellin is recognized by the host innate immune receptor TLR5, making flagella both a colonization tool and an immune trigger, a tension pathogens resolve by regulating flagellar gene expression once inside host tissue. Flagellin itself is being explored as a vaccine adjuvant, since TLR5 recognition strongly activates innate immunity. In environmental and agricultural microbiology, motility and chemotaxis toward root exudates are essential for rhizosphere-colonizing bacteria used as biofertilizers and biocontrol agents, allowing beneficial bacteria to actively seek out plant roots. In biophysics and nanotechnology, the flagellar motor is a model system for engineered molecular machines and single-molecule biophysics, and mutant, non-functional motor components are used as tools to dissect basic principles of energy transduction shared with other rotary machines like ATP synthase.

Ask the tutor: How does the polymorphic transformation of the flagellar filament (normal vs. curly form) mechanically cause bundle disassembly during a tumble?