A Delicate Nanoscale Motor Made by Nature—The Bacterial Flagellar Motor

The bacterial flagellar motor (BFM) is a molecular complex ca. 45 nm in diameter that rotates the propeller that makes nearly all bacteria swim. The motor self‐assembles out of ca. 20 different proteins and can not only rotate at up to 50 000 rpm, but can also switch rotational direction in milliseconds and navigate its environment to maneuver, on average, towards regions of greater benefit. The BFM is a pinnacle of evolution that informs and inspires the design of novel nanotechnology in the new era of synthetic biology.


Introduction
In an aqueous environment, bacteria benefi t from swimming to where life is better, and most bacteria swim by rotating their fl agella. This rotation is powered by the electrochemical ion motive force arising from the transit of ions across the cellular membrane. [ 1,2 ] This provides energy to free-swimming bacteria to propel their cell body at a speed of 15-100 µm s −1 , or up to 100 cell body lengths per second. The BFM is a nanoscale rotary molecular machine embedded in the cell envelope and possesses an outstandingly effi cient mechanochemical conversion between electrochemical free energy and mechanical work. A single BFM in Escherichia coli can output a power of ca. 1.5 × 10 5 pN nm s −1 [ 3 ] and rotate at ca. 300 Hz (18 000 rpm) [ 4 ] while the BFM in Vibrio alginolyticus can rotate as fast as ca. 700 Hz (42 000 rpm), [ 5 ] which is nearly triple the 15 000 rpm of a modern Formula 1 racer. Additionally, while man-made machines suffer from energy loss due to heating, the BFM operates at near 100% effi ciency of energy conversion from ion transit to motor torque. [ 6 ] Motility at low Reynolds numbers requires changes in the manner of propulsion when compared with inertial regimes. First, propulsion requires a time-irreversible swimming motion, The fi lament is the longest part of the bacterial fl agellum and can extrude up to 10 µm from the cell body. It is a long helical hollow tube aligned by 11 parallel protofi laments, which are the polymers of FliC, the fl agellin subunit. [ 13 ] These fl agellin molecules can bind in two different ways, resulting in protofi laments with different lengths. [ 14 ] The fl agellar fi laments appear in different helical shapes because of various length combinations of the protofi laments. The conformational change of fi lament shape plays a key role in the switching between "tumble" and "run" states. [ 15 ] The fi laments are connected to the motor by a fl exible module, named the hook. Mechanically, the fl agellar hook works analogously to a universal joint. The hook is a short curved tube, assembled by 120 copies of hook protein unit-FlgE. [ 16 ] The curvature of the hook remains constant when the bacterial fl agellum rotates. Thus, the hook can transmit torque generated by the motor effi ciently while rotating the fl agellar fi laments through an arc. [ 17 ] The central mechanical component, the bacterial fl agellar motor, ca. 45 nm in diameter, with a total molecular mass of 11 MDa, is built by about 20 proteins and spans across the outer membrane, peptidoglycan wall and the inner membrane, into the cytoplasm. [ 18 ] Similar to all rotary motors, its structural constituents can be divided into two categories: the rotor and the stators ( Figure 2 ).
On the rotor of the BFM, four concentric protein rings are conjugated by a rod in their center: the L-ring, P-ring, MS-ring, and C-ring, altogether comprising the basal body. [ 19 ] These rings are named according to their locations within the membrane. The L-ring sits at the lipopolysaccharide outer membrane, while the P-ring is located at the peptidoglycan cell wall, and both are believed to have a bushing role between the motor and outer cell envelope. The MS-ring stands for membrane and supramembrane ring and it is constructed by ca. 26 FliF protein subunits. The C-ring is in the cytoplasm, mounted on the MS ring. It is composed of ca. 26 copies of FliG, ca. 34 copies of the FliM, and more than 100 copies of FliN proteins. The rod connects all the protein rings and is made by FlgB, FlgC, FlgF and FlgG. Altogether, this forms the rotor of the motor.
The stators of the BFM are located around the periphery of the rotor, forming a larger concentric protein ring outside the MS-ring. The BFM can accommodate 8-12 stator complexes, each of which is made up of 4 MotA and 2 MotB proteins. [ 20 ] MotA has a cytoplasmic domain interacting with the rotor to generate torque. MotB has a cytoplasmic domain and a linker domain that anchors the stator complex to the peptidoglycan cell wall. Two ion channels are formed on each stator complex that allow the passage of protons from the periplasm to the cytoplasm of the cell. [ 21 ] Previous structural and biochemical studies indicate that torque is generated between the C-terminal domain of the rotor protein FliG and the cytoplasmic domain of the stator protein MotA. [ 22,23 ] Both steric and electrostatic interactions are believed to be crucial for the torque generation, with proton fl ux coordinating conformational change in MotA, to drive rotation by direct interaction with FliG. Mutational studies revealed that a conserved Asp residue on the MotB, Asp 32, is essential for motor function, possibly providing the proton binding site in proton transfer. [ 24 ] Critical charged residues have also been found on MotA and FliG, which play important roles in torque generation and motor function. [ 22,25 ] The assembly of the bacterial fl agellum begins on the cytoplasmic side of the BFM. It is built from inside out and all the external proteins of the hook and the long fi lament proteins have to be exported by the fl agellar type III secretion system, a close analogue of the type III virulence secretion system. [ 26 ] When the BFM assembly commences, fi rst the MS-ring and the export apparatus are formed. Subsequently, the stators, rod, and other rings self-assemble at their respective positions, after which the export apparatus located beneath the C-ring delivers FlgE monomers through the hollow interior of the rod to assemble the hook at the cell's exterior. Once the hook is fi nished, the export apparatus switches to export FliC monomers to build the long fi lament. [ 27 ] In this dynamic assembling process, FliD proteins serve as a cap at of the far end of the fi lament, coordinating the FliC polymerization and elongation of the fi lament. [ 28 ] A key challenge here is to understand how protein substrates of different sizes can be exported by the same apparatus through a narrow channel (about 2 nm in diameter) in the fl agellum. [ 14 ] Previous studies have shown that these protein monomers are unfolded into peptide chains when entering the export gate; [ 29 ] however, many details of this process remain poorly understood. [ 30 ]

Function of the Bacterial Flagellar Motor
The BFM coordinates precisely the organized motion of multiple proteins to propel bacteria. As such, the BFM shares common features with all motors, for instance the use of repeating cycles. However, due to distinctions in their working environments, the BFM necessarily functions differently to macroscopic engines.
The BFM is a nanoscale molecular motor whose environment is dominated by thermal fl uctuations. [ 31 ] A distinct feature of this environment is described as "the world of low Reynolds number". [ 32 ] At low Reynolds number, when small things move slowly through fl uids, viscous forces are signifi cant but inertial forces are not. Thus, a small object stops moving immediately if the driving force/torque is withdrawn. On the other hand, while Brownian motion blurs trajectories, it also provides a stochastic force that can cause molecules to pass over high energy barriers, unlike the scenario seen in the macroscopic world.

Power Input
Distinct from most linear molecular motors, such as myosin on actin fi laments, kinesin and dynein on microtubules, which are powered by ATP hydrolysis, the BFM is powered by the free energy released from fl ow of ions down an electrochemical gradient across the cytoplasmic membrane into the cell ( Figure 2 ). This energy source is termed the protonmotive force (PMF), which is maintained by the electron transport chain and ATPase involved in metabolic processes. Other bacteria, especially those living in marine or high pH environments consume Na + rather than H + .
The PMF consists of two parts (Equation ( 1) ): The fi rst contribution is from the transmembrane electrical potential gradient, or the membrane potential. This arises from the electric fi eld generated by different concentrations of cations and anions across the membrane. The second component consists of the entropic force arising from transmembrane concentration differences: Here, Δ ψ represents the electrical potential across cytoplasmic membrane, Δ pH denotes the change in proton

Power Output
Like macroscopic machines, the torque-speed relationship is widely used to assess the performance of a molecular motor since it provides a full picture of the power output of the motor under different external loads, and also indicates the energy conversion effi ciency.
Experimentally, the torque-speed relationship of the BFM can be determined by attaching a polystyrene bead to the fl agellum of a cell attaching to the surface of a glass coverslip ( Figure 3 A). The rotation speed of the bead can be record by a fast camera mounted on a microscope while the viscosity of the external medium is rapidly changed by adding Ficoll, [ 4 ] or while the drag coeffi cient of the bead is changed by varying the size of the bead. [ 5 ] In Figure 3 B, we see a typical torque-speed relationship of the E. coli BFM. The torque (τ) of the motor remains approximately constant up to ca. 170 Hz (the "knee" velocity v knee ), and then decreases sharply to zero at ca. 300 Hz. The sodiumdriven fl agellar motor exhibits a similar relationship, but with a higher v knee and zero-load speed. Previous studies and calculations have estimated that with high external load, the BFM converts almost all of the free energy released from the protons fl ow across into mechanical rotation of the load, indicating that the energy conversion effi ciency of the BFM is very high. [ 7,33 ] Experiments that control the PMF show that the motor rotation speed depends linearly on the PMF in both low and high load regimes. [ 34 ] The unusual shape of the torque-speed relationship, the high energy conversion effi ciency and PMF dependence are crucial to understanding the mechanism of the BFM, and thus receive extensive experimental and theoretical study. [ 3,4,[35][36][37]

Stepping
The BFM has long been postulated to be a stepping motor, but only recently was experimental evidence found. Steps in the F1 ATPase, the only other rotary molecular motor, were fi rst seen in 1998, [ 38 ] and later, substeps were resolved. [ 39 ] However, steps in BFM rotation were more diffi cult to resolve due to the high rotation speed and the small stepsize. In 2005, Sowa et al. constructed a chimera motor, with sodium driven stators in an E. coli BFM background. [ 40 ] With this new motor, they managed to express only one stator under low sodium concentration. This resulted in a slow rotation rate, with long dwell time between steps, which aided step detection. An optical trapping system with high temporal and spatial resolution was used while a small indicator (a latex bead of diameter 0.2-0.5 microns) was attached to the fl agellum. Finally, 26 steps per revolution were observed for the fi rst time, [ 40 ] consistent with the 26-fold stoichiometry of the FliG protein on the C ring.

Switching of the Bacterial Flagellar Motor
An effi cient method of propulsion alone is not enough to outcompete rivals and avoid toxins. A strategy for sensitive navigation is also required, and the BFM possesses this in bacterial chemotaxis. This process presents itself as the alternate "run" and "tumble" states of the cell. One important feature of E. coli chemotaxis is that the cell responds to temporal changes in attractant concentration while swimming, rather than spatial stimuli from a concentration gradient. [ 41,42 ] When E. coli runs up an attractant gradient, the run will be prolonged and the tumble be delayed.
The quick switching of the BFM between the "CCW" and "CW" states leads to the transition between "runs" and "tumbles" and forms the basis of bacterial chemotaxis. The switching of the BFM can be observed by the same bead rotation assay shown previously in measuring the torque-speed relationship (Figure 3 C).
This switching of the motor is fi nely controlled by a chemotactic sensory and signalling network. On the surface of the bacteria, there are arrays of chemoreceptors, methyl-accepting chemotaxis proteins (MCPs), which detect environmental changes. Information from the environmental changes is translated into phosphorylation levels of a signaling molecule, CheY, through a series of biochemical reactions.
The rotation bias of the motor is sensitively controlled by the cytoplasmic concentration of this small diffusible protein CheY-P. [ 43,44 ] A lower concentration of CheY-P in the cytoplasm results in more CCW rotation of the motor, while a higher concentration of CheY-P causes more CW rotation. Previous works revealed that the interaction between CheY-P molecules and the  C-ring is responsible for determining the BFM's direction of rotation. [ 45,46 ] Binding of CheY-P to FliM is believed to trigger conformational changes in FliM, which is coupled to conformational changes in the FliG protein. [ 44 ] Different structural orientation of the FliG protein generates two opposite directions of rotation (CCW or CW) when torque is delivered from the stator units ( Figure 4 A). [ 47 ] The switching of the BFM can be simplifi ed by focusing on the interactions of the CheY-P molecules with a ring-shaped assembly of 34 rotor switching units (RSU), each of which is formed by 1 FliM, 1 FliN and ∼1 FliG proteins (Figure 4 B). In this ring, each RSU is identical and can exist in either CCW or CW state, leading to CCW or CW rotation, respectively. One RSU can be bound or not bound to one CheY-P molecule, and CheY-P binding favors CW state. The motor spins full speed in CCW or CW while all RSUs are in a coherent state. [ 48 ] The BFM of E. coli is remarkably sensitive: it can respond to attractant concentration as low as 1 µM. [ 33,49 ] To initiate a switch of the motor from CCW to CW rotation, some RSU units must bind a CheY-P molecule. Previous experiments have revealed that the motor switching responses ultrasensitively to changes in CheY-P concentration, meaning that only small changes in the number of CheY-P bound on the ring can infl uence the switching state of the entire ring. [ 50 ] To explain this ultrasensitivity, a "conformational spread" model has been proposed. [ 51 ] In this model, a coupling energy between adjacent RSUs is introduced, where an energy penalty is applied if an RSU and its neighbors are in differing states. The existence of this coupling energy allows conformational spreading from one RSU to the rest, and successfully explains the experimental observation in which subtle changes in CheY-P concentration can greatly change the switching bias of the motor. In the "conformational spread" model, a switching event of the motor usually starts with a switching event of a single RSU and this newly created domain may either grow to encompass the entire ring, or shrink and disappear (Figure 4 C), as the ring returns to its previous coherent state.
Several key predictions of the conformational spread model have been validated by recent experiments, [ 49,52 ] which demonstrated that conformational spread is both necessary and suffi cient to explain the switching mechanism of the BFM. As a generalization of the classical theories of allosteric regulation, conformational spread is applicable to any multimeric protein complex that responds to ligand binding. Thus the BFM utilizes the fundamental cooperativity of neighbor-neighbor interactions to build a molecular gearbox that is capable of changing direction in milliseconds.

Structural Adaptivity of the Bacterial Flagellar Motor
What is remarkable about the sensitivity of the fl agellar motor's switch complex is that it maintains a high sensitivity across a  very large operating range. This sensitivity arises due to conformational spread as detailed above, but the operating range can be adjusted by directly adapting the structure of the rotor. [ 53,54 ] Yuan et al. showed that the number of FliM units can change in response to levels of CheY-P, increasing the number of binding sites and thus sensitivity when CheY-P concentrations are low. Furthermore, Lele et al. showed that FliM binding directly strengthened in response to the rotational direction of the motor, and that it was not CheY-P binding that mattered, but only the direction of rotation. Indeed, not only the stators but both rotor proteins at the bottom of the rotor, FliM and FliN, have been shown to turnover during the motor operation. [55][56][57][58] This turnover is a property of bionanotechnology that lies in stark contrast to man-made technology, that the structure of the motor can adaptively remodel, while operating, yet this seems to have evolved as a common theme throughout biology. [ 54,[59][60][61] Thus far, turnover in FliG has not been observed, [ 62 ] presumably due to its required role in torque-generation and in the conformational change that dictates a switch event. [ 49 ] This leads to a further perplexing symmetry mismatch between the top and bottom of the C-ring: FliG should be ca. 34-fold in Salmonella enterica , [ 63 ] and while FliM is ca. 34-fold in an exclusively clockwise rotating motor, it can be as many as 44-fold in an exclusively counterclockwise rotating motor. [ 53 ] Additionally, it has been shown recently that the motor can also adapts in response to increased load. [ 64 ] Lele et al. showed here that the stators themselves dynamically respond to load, and this response was consistently observed even when chemotaxis and rotational switching was removed-it is the stators themselves that engage or disengage with changes in load. The similar observation has also been reported by Tipping et al. [ 58 ] This is presumably because at low load only few stators are required to drive rotation and additional stators simply waste ions. [ 64 ] However, the mechanism for force-sensing of the stators remains unknown.

Perspective
The bacterial fl agellar motor is the pinnacle of evolutionary bionanotechnology: a self-assembling nanoscale electric rotary motor that performs at higher speed and with greater efficiency than any man-made device. Study of such a machine, honed by billions of years of evolution, yields insight into the fundamental features of biological design. This insight, and the lessons from molecular architecture in general, can be harnessed to drive innovation in the creation of novel man-made nanotechnology.
The BFM is already a good prototype for a self-propelled nanobot. The steps in the chemotactic pathway of "detect"-"compare"-"decide"-"control motor" can be adopted to design signaling networks that govern motility in any synthetic device. Molecular motors have already been directly used in nanobiotechnology applications, for example, the use of cytoskeletal motor proteins for molecular transport of microtubules and actin fi laments. [ 65 ] Although these molecular-motor powered devices are still in their infancy, they provide the possibility for designing nanomachines that can smartly deliver drugs to a target guided only by extracellular signals.
However, the BFM offers not only impressive mechanical performance and desirable design traits, it also contains a simple switching mechanism that allows the motor to change direction quickly. This mechanism, based upon conformational change, is an example of complex behavior arising from simple neighbor-neighbor interactions, and highlights a fundamental difference in the way biological machines function. Man-made nanotechnology typically is designed brick by brick with an architectural plan from above. Evolution, in contrast, favours the emergence of complex behavior from simple, tunable interactions. [ 66 ] By studying how the BFM rotates and switches, how it assembles, and how it arose, we enable the next era of nanotechnology in which we apply these natural design principles in the creation of new synthetic biological machinery.