Flagella

The Flagellum is an organelle of bacterial motility. It consists of several substructures: the filament, the hook, the basal body, the C ring, and the C rod. The flagellar motor, an actively functional part of the flagellum, can generate torque from proton-motive force. The structural aspects of the flagellum are described here, revealed in pursuit of the identity of the flagellar motor. 

I. STRUCTURE 

A. Filament The flagellum is a complex structure composed of many different kinds of proteins. However, the term flagellum, especially in earlier studies, often indicates the flagellar filament only, because the filament is the major portion of the entire flagellum. In this section, I describe the filament and occasionally call the filament just flagellum. 1. Number of flagella per cell The number and location of flagella on a cell are readily discernible traits for the classification of bacterial species. The number ranges from one to several hundred, depending on the species; hence the nomenclature, monotrichous (one) or multitrichous (two or more). Occasionally the term “amphitrichous” is used for two flagella. There are three possible locations on a cell body for flagella to grow—polar (at the axial ends of the cell body), lateral (at the middle of the cell body), or peritrichous (anywhere around the cell body). In some cases, “lateral” is used as the counterpart of “polar,” as in the two-flagellar systems of Vibrio alginolyticus, polar sheathed flagellum and lateral plain flagella (although the latter are actually peritrichous). Atuft of flagella growing from a pole is called lophotrichous. In most cases, flagella can be named by a combination of number and location, for example, polar lophotrichous flagella of Spirillum volutans. Although ordinary flagella are exposed to the medium, some flagella are wrapped with a sheath derived from the outer membrane (e.g., in Vibrio cholerae). In an extreme case such as spirochetes, flagella are confined in a narrow space between the outer membrane and the cell cylinder. The flagella still can rotate; the helical cell body works as a screw, and the flagella counterbalance the torque of the cell body. 2. Filament shape 

The filament shape is helical. In theory, there are two types of helices, right-handed and left-handed; in reality, Salmonella spp. have left-handed filaments and Caulobacter crescentus has a right-handed filament. However, it should be noted that the shapes of these two helices are not mirror images of each other. There are several detailed filament shapes, and it will be convenient to use the names of the typical shapes found in Salmonella spp.—normal (left-handed), curly (right-handed), coiled (left-handed), semi-coiled (right-handed), and straight. The helical parameters of these helices are discrete and distinguishable from one another (Fig. 39.1). Flagella can switch among a set of helical shapes under appropriate conditions; both helical pitch and helical handedness are interchangeable. The transformation of shapes can be induced by physical perturbation (torque, temperature, pH, and salt concentration of medium). Genetic changes such as point mutations in the flagellin (the component protein of the flagellar filament) gene also result in transformation of helices, but some mutant flagella, such as straight flagella, are too stiff to transform into another helix. This phenomenon, called polymorphism of flagella, is a visible example of conformational changes in proteins and therefore has evoked an idea of a functional role of flagella in motility. Could polymorphism of the flagellum by itself cause the motion? The answer is “no.” Flagella are passive in terms of force generation. Polymorphism of flagella is observed to occur naturally on actively motile cells with peritrichous flagella. The helical transformation is necessary for untangling a jammed bundle of tangled flagella. When normal flagella in a jammed bundle are transformed into curly flagella, knots of tangled flagella run toward the free end of each flagellum to untangle the jammed bundle. Atheoretical model that explains the polymorphism successfully was presented by Dr. Chris Calladine (University of Cambridge). Twisting and bending a cylindrical rod gives rise to a helix. This model predicts 12 shapes, and eight of them have been found in existing filaments: straight with a left-handed twist, f1, normal, coiled, semi-coiled, curly I, curly II, and straight with a right-handed twist. Only a small energy barrier seems to lie between any two neighboring shapes. Polymorphic transition occurs from one shape to its neighbors; for example, in a transition from normal to curly, the filament briefly takes on coiled and semi-coiled forms. 3. Flagellin The component protein of the filament is called flagellin. Although the flagellum of many bacteria is composed of one kind of flagellin, some flagella consist of more than two kinds of closely related flagellins. The molecular size of flagellin ranges from 20 to 60 kDa. Enterobacteria tend to have larger molecules, whereas species living in freshwater have smaller molecules. One of the most characteristic features of flagellin is evident even in the primary structure of the molecule; the amino acid sequences of both terminal regions are well conserved, whereas that of the central region is highly variable even among species or subspecies of the same genera. 

As a matter of fact, this hypervariability of the central region gives rise to hundreds of serotypes of Salmonella spp. The terminal regions are essential for binding of each molecule to another to polymerize them into a filament. Complete folding of flagellin occurs during assembly; although the terminal regions do not take on any specific secondary structure in solution, they are converted into _-helix after polymerization. In the filament, the terminal regions are located at the innermost radius of a cylindrical structures, whereas the central region is exposed to the outside. Note that the filament is extremely stable; it does not depolymerize in water, in contrast to actin filaments or tubulin filaments, which depolymerize in the absence of salts. This description of the flagellin molecule is not applicable to that found in archaeobacteria. Archaic flagella seem to have a system totally different from those of eubacteria; archaic flagellins have signal sequences, suggesting that the flagellum might grow from the proximal end in the outer membrane, in contrast to the distal growth of eubacterial flagellum (see Section IV). 4. Cap protein Flagella have been regarded as a self-assembly system. Indeed, flagellin can polymerize into flagella under conditions that commonly promote protein crystalization in vitro. However, in vivo flagellin assembly requires another protein, without which the flagellin is secreted into the medium as monomers. The protein that helps filament formation is located at the tip and is thus called the cap protein, or HAP2 or FliD. The cap proteins assemble in a pentamer, forming a star-shaped structure. The star hands fit in the grooves of flagellin subunits at the tip of flagellum, leaving a small gap for a nascent flagellin to insert. B. Hook 1. Shape Hook, as the name suggests, is more sharply curved (almost in a right angle) than the filament and is much shorter. The curvature indicates the flexibility of the hook, although it has to be stiff enough to transmit the torque generated at the basal structure to the filament. From these physical properties, the hook has been regarded as a universal flexible joint. The length of the hook is 55 nm with a standard deviation of 6 nm, which is rather well controlled when compared with the length of filament. The hook length is not controlled by any molecular rulers, as might have been expected, but by a measuring cup at the base of the flagellum (see Section IV). 

A polyhook is a hook of indefinite length, obtained in certain kinds of mutants. Its shape is a righthanded superhelix. The wild-type hook has the same superhelix, but consists of about one-fourth of the helical pitch. 2. Hook protein The hook is a tubular polymer made of a single kind of protein, hook protein or FlgE. The molecular size of hook protein varies from 29 kDa (Bacillus subtilis) to 76 kDa (Helicobacter pylori), but it is around 42 kDa for most species. The architecture of hook protein resembles that of flagellin—the amino acid sequence in both terminal regions is well conserved, but in the central region it is variable. Hook protein folding also completes on assembly. 3. Scaffolding protein FlgD The hook does not self-assemble; it requires a helper protein, FlgD, which functions in a similar way that FliD does for filament formation. FlgD sits at the tip of the nascent hook to polymerize the hook protein coming out from the central channel. When the hook length reaches 55 nm, FlgD is replaced by HAP1 (FlgK), which remains in the mature flagellum. Because of its temporary existence, FlgD is regarded as a scaffolding protein. 4. Hook-associated proteins There are two minor proteins between the hook and filament. They are called hook-associated proteins (HAPs) because they were found at the tip of the hook in several filamentless mutants. There were originally three HAPs, called HAP1, 2, and 3, in the order of their molecular size. HAP2 turned out to be located at the tip of filament as described above, leaving HAP1 and 3 between the hook and filament. They are, therefore, better termed hook-filament junction proteins. The number of subunits of HAP1 and 3 in a filament is estimated to be 5 or 6, indicating that they form one-layer rings sitting one on another. The roles of these two HAPs have been ambiguous. The idea of a connector to smooth the junction between the two polymers is blurred by the question, “Why are not one but two kinds necessary?” In a mutant of HAP3, filaments underwent polymorphic transitions so easily that cells cannot swim smoothly, suggesting a specific role of HAP3 as a stabilizer of filament structure. C. Basal structure Flagella have to be anchored in the cell wall. 

The structural entity for the anchoring was called the basal structure or basal granule, hinted at by vague images by electron microscopy. Since DePamphilis and Adler in 1974 defined the details of the basal structure, it has been called the basal body. The basal body typically consists of four rings and one rod. The basal body does not contain everything necessary for motor function. Fragile components were detached from the basal body during purification. In 1985, one such fragile structure was found attached to basal bodies purified by a modified method; it was named the C (cytoplasmic) ring. In 1990, another rodlike structure was found in the center of the C ring. Therefore, the basal structure (as of 2003) consists of the basal body, the C ring, and the C rod, but there may be more. 1. Basal body The basal body contains rings and a rod penetrating them. The number of rings varies depending on the membrane systems; there are four rings in most gramnegatives, and two rings in gram-positives, exemplified by B. subtilis. Some variation in the number (such as five rings in C. crecentus) has been occasionally seen, but the purpose and function of a fifth ring is unclear because the cell’s membranes are supposed to be the same as those of S. typhimurium or E. coli. 

The structure of the basal body of S. typhimurium has been extensively analyzed. The physical and biochemical properties of the substructures of the basal body described later are from S. typhimurium, unless otherwise stated. 2. LP-ring complex The outermost ring, the L ring, interacts with the lipopolysaccharide layer of the outer membrane, and the P ring just beneath the L ring may bind to the peptidoglycan layer. The LP-ring complex works as a bushing, fixed firmly enough to hold the entire flagellar structure stably in the cell surface. The component proteins, FlgH for the L ring and FlgI for the P ring, have signal peptides, indicating that they are secreted through the general secretory pathway (GSP), which is the exception for flagellar proteins (see Section IV). FlgH undergoes lipoyl modification. After the L and P rings have bound together to form the LP-ring complex, this complex is resistant to extremes of pH or temperature. Subjecting it to pH 11, pH 2, or boiling for 1min does not destroy the complex, confirming that the complex serves as a rigid bushing in the outer membrane. The essential roles of the complex are still ambiguous because mutants lacking the complex still can swim, though poorly, and because no corresponding structure has been found in gram-positive bacteria. 3. MS-ring complex Earlier studies of flagellar motor function assumed that torque would be generated between the M and S rings, which face each other on the inner membrane. However, in 1990, it was shown that a single type of protein, FliF, self-assembles into a complex consisting of the M and S rings and part of the rod. FliF is 65 kDa, the largest of the flagellar proteins. It contains no cysteine residues. It consists of several regions—terminal regions and two distinguishable central regions. Overproduction of FliF in E. coli gives rise to numerous MS-ring complexes packed in the inner membrane. The S ring has been seen in the basal bodies from all the species studied so far (at least seven examples). It stays just above the inner membrane (supramembrane) and has no apparent interaction with any other structures. Besides, it is very thin (~1 nm), and the role of the S ring remains mysterious. Although the MS-ring complex is no longer regarded as the functional center of the flagellar motor, it is still the structural center of the basal structure and plays an important role in flagellar assembly (see Section IV). 4. Rod The rod is not as simple as its name suggests; it consists of at least four distinct proteins. 

It breaks at the midpoint when external physical force is applied to the filament, which is not expected in a structure that transmits torque to the filament. Rod formation seems complicated because of the four component proteins. No intermediate rod structure has been observed; either there is a whole rod or no rod at all. Because the P ring is formed on the rod, some of the rod proteins could be the target of interaction with FlgI, the subunit of the P ring. 5. C ring The C ring is a fragile component of the basal structure. It is resistant to the nonionic detergent Triton X-100, but it is destroyed by the alkaline pH and high salts concentration employed by conventional purification methods. The dome shape of the C ring is easily flattened on a grid during preparation for electron microscopy. The C ring consists of the switch proteins (FliG, FliM, and FliN) and so is sometimes called the switch complex. It is not known whether other proteins are present or not. FliG directly binds to the cytoplasmic surface of the M ring. FliM binds to FliG, and FliN to FliM. The stoichiometry of these molecules in the C ring is still controversial; 20–40 copies of FliG, 20–40 copies of FliM, and several 100 copies of FliN.  Genetic studies revealed that the switch complex plays important roles in flagellar formation, torque generation, and the switching of rotational direction. The C ring directly binds signal molecules, CheY, produced in the sensory transduction system, but the mechanism of the switching is ambiguous. 6. Export apparatus Flagella have been regarded as having a self-assembly system, similar to that of bacteriophages. However, flagellar assembly is quite different from phage assembly in many ways. First, the flagellum, being an extracellular structure, assembles not in the cytoplasm but outside the cell. Second, the component proteins therefore, have to be transported from the cytoplasm to the outside. Third, assembly consequently proceeds in a one-by-one manner, at the distal end of the nascent structure. For this kind of assembly, a protein excretion system must play an important role. As a matter of fact, among the 14 genes required in the very first step of flagellar assembly, at least seven gene products are necessary to form a protein complex, called an export apparatus. One of them, FliI, has an ATPase activity, suggesting that one step in the export process requires ATP hydrolysis as an energy source. The physical body of the export apparatus has not been identified; the C rod is a strong candidate, judging from its location in the C ring. 

II. FUNCTION 

The function of flagella is described here briefly so that the meaning of the structure can be understood. Bacterial flagella rotate. There is no correlation between bacterial flagella and eukaryotic flagella, either in function or in structure; the type of movement, the energy source, and the number of component proteins differ greatly between the two. No evolutionary correlation between these two types of flagella has been shown. Among motile bacterial species, swimming by flagellar rotation is the most common. However, several families such as myxococcus, mycoplasm, and cyanobacteria can move on a solid surface in a gliding motion; the motile organ of gliding bacteria is not known. A. Torque The rotational force (torque) of the flagellar motor is difficult to measure directly, but can be estimated from the rotational speed of flagella. The method most widely employed is the tethered-cell method, in which the rotation of a cell body caused by a tethered filament can be observed with an ordinary optical microscope. A more sophisticated method, which employs a laser as the light source of a dark-field microscope, allows one to measure the rotation of a flagellum on a cell stuck on the glass surface at a time resolution of millisecond. 1. Rotational direction The flagella of many species (e.g., enterobacteria) can rotate both clockwise (CW) and counter-clockwise (CCW). Under ordinary circumstances, around 70% of the time is occupied by CCW rotation, which causes smooth swimming. A brief period of CW rotation causes a tumbling motion of the cell. There is no perceptible pause in switching between the two modes. In some bacterial species such as Rhodobacter sphaeroides, a lateral flagellum on a cell rotates in the CW direction only, with occasional pauses. During the pauses, the filament takes on a coiled form and curls up near the cell surface. Upon application of torque to this filament, the coiled form extends to a semi-stable right-handed form that closely resembles a curly form. The CW rotation of this right-handed helix causes a forward propulsive force on the cell. 

2. Rotational speed Because the torque of the flagellar motor cannot be directly measured due to technical limits even today, it is estimated from the rotational speed of flagella that is believed to correlate with the torque. The highest speed of flagella observed under physiological conditions is about 200 Hz for S. typhimurium. High viscosity of the solution slows down the speed in a linear manner over a wide range of speeds below 200 Hz, indicating that the torque of the flagellar motor is constant in this speed regime. At the speeds higher than 200 Hz, externally forced by electro-rotation field, the torque quickly decreases and becomes zero at 300 Hz. This gives a theoretical limit of the maximal speed of the motor per se; that is, the relative speed between the rotor and stator without any load such as flagellar filament. B. Energy source The energy source of torque generation in the flagellar motor is not ATP but proton-motive force (PMF). PMF is the electrochemical potential of the proton, and results in the flow of protons from outside to inside the cell. PMF consists of two forms of energy, membrane potential, and entropy caused by a difference in pH between outside and inside the cell. Because these two parameters are independent and separable from each other, either one can, in principle, be abolished without affecting the other. Note that the polar flagellum of V. alginolyticus substitutes sodium ion (Na_) for proton. One of the goals of flagellar research is the elucidation of the mechanism by which PMF or NaMF is converted into torque in the motor. C. Switching of rotational direction Switching the rotational direction of flagella is the primary basis of chemotaxis, one of the most important behaviors shown by bacteria. Damage in the switching mechanism results in a rotation biased to either CCW only or CW only. In a strict sense, the switching mechanism will not be solved until the mechanism of rotation is solved. However, the factors involved in the mechanism are known; an effector binds to the switch complex in the flagellar motor. The effector is the phosphorylated form of CheY, a signalling protein in the sensory transduction system. III. GENETICS Flagellar genetics has been most extensively studied in S. typhimurium, especially using the enormous number of strains that Dr. Shigeru Yamaguchi (Meiji University) has collected for more than 30 years. The discussion in this section is, unless otherwise indicated, based on results obtained from these strains. 

A. Flagellar genes There are more than 50 flagellar genes, which are divided into three types, according to the null mutant phenotype. 1. The fla genes Defects in the majority of the flagellar genes result in flagellar deficient (Fla_) mutants. These genes were originally called fla genes. In 1985, when the number of genes exceeded the number of letters in the alphabet, a unified name system for E. coli and S. typhimurium was introduced: flg, flh, fli, and flj; one for each of the clusters of genes scattered in several regions around the chromosome (see Section III.B; Fig. 39.2). 2. The mot genes Mutants that produce paralyzed flagella are called motility deficient (Mot_) mutants. There are only two mot genes (motA and motB) in S. typhimurium, but four mot genes (motA, motB, motX, and motY) in R. sphaeroides and in some other species such as V. alginolyticus. 3. The che genes Mutants that can produce functional flagella but that cannot show a normal chemotactic behavior are called chemotaxis deficient (Che_) mutants. These are divided into two types, general chemotaxis mutants and specific chemotaxis mutants. The former involve the proteins working in the sensory transduction (CheA, CheW, CheY, CheZ, CheB, and CheR), and the latter involve the receptor proteins (e.g., Tsr, Tar, Trg, and Tap). B. Gene clusters in four regions Most flagellar genes are found in gene clusters on the chromosome. They are in four regions: the flg genes in region I (at 26 min), the flh genes and mot and che genes are in region II (41.7 min), and the fli genes are in regions IIIa (42.4 min) and IIIb (42.7 min) (Fig. 39.2). The flj operon (including fljA and fljB) at 60 min involves an alternative flagellin gene to fliC and is only found in Salmonella. Either FliC flagellin or FljB flagellin is produced at any time. The hin gene inverts the transcriptional direction at a certain statistical frequency; if the flj operon is being expressed, FljA represses fliC, allowing FljB flagellin alone to be produced. This alternate expression of two flagellin genes is called phase variation. C. Transcriptional regulation Flagellar construction requires a well-ordered expression of flagellar genes not only because there are so many genes, but also because flagellar assembly requires only one kind of component protein at a time, as described previously. There is a strict hierarchy of expression among the flagellar genes. The hierarchy is controlled or maintained by a few prominent regulatory proteins. 1. Hierarchy: three classes The hierarchy of flagellar gene expression is divided into three classes; class 1 regulates class 2 gene expression, and class 2 regulates class 3. Class 1 contains only two genes in one operon, flhD and flhC. Class 2 consists of 35 genes in eight operons. There are two regulatory genes, fliA and flgM; the rest are component proteins of the flagellum or the export apparatus. Class 3 genes encode flagellin, motA and motB, and all the proteins involved in sensory transduction. 

Flagellin is one of the most abundant proteins in a cell, suggesting that the tight regulation in the hierarchy guarantees the economy of the cell. 2. Master genes, flhDC Master gene products form a tetrameric complex of FlhD/FlhC, which works as a transcriptional activator of the class 2 operons. The master operon ( flhDC) is probably transcribed with the help of the “housekeeping” sigma factor, _70. The master operon has also been shown to be activated by a complex of cyclic AMP and catabolite activator protein (cAMP–CAP), which binds to a site upstream of the promoter. 3. Sigma factor F (_ F, FliA) and antisigma factor (FlgM) The FliA and FlgM proteins expressed from the operon competitively regulate the class 3 operons. FliA is the sigma factor that enhances the expression of the class 3 operons, whereas FlgM is an antisigma factor against FliA. If the hook and basal body have been constructed normally, FlgM is secreted into the medium through the basal body and the complete hook, allowing free FliA proteins to work on the class 3 operons. However, if the hook and basal body construction is somehow halted in the middle of process, FlgM stays in the cytoplasm in a complex with FliA, maintaining shut off of the expression of the class 3 operons. This intriguing regulation mechanism seems to work efficiently for peritrichous flagella. However, polar flagella which look much simpler than peritrichous ones are regulated not only by the sigma 28 (FliA) and its suppressor (FlgM) but also by sigma54 (RpoN) and its activators (FlbD, FlgR, FleQ, etc.). 

4. Global regulation versus internal regulation There are several external genes or factors that affect the flagellar gene expression through the master operon flhDC. Some of the factors show pleiotropic effects on many cellular events such as cell division, suggesting that flagellation is finely tuned with the cell division cycle due to well-organized tasks of global regulation systems. As described, the master operon ( flhDC) is probably transcribed using the “housekeeping” sigma factor, _70. In the 1990s, other factors regulating or modulating flhDC expression have been identified mainly in E. coli. The motility of E. coli cells is lost at temperatures higher than 40 _C as a result of reduced flhDC expression. It has been shown that some of the heatshock proteins are involved in both class 1 and class 2 gene expression. This strongly suggests that flagellar genes are under global regulation in which the heatshock proteins play a major role; probably the proper protein folding (or assembly) mediated by these chaperones is essential for flagellar construction. Other adverse conditions such as high concentrations of salts, carbohydrates, or low-molecular-weight alcohols, also suppress flhDC expression, resulting in lack of flagella. The regulation by all these factors is independent of the cAMP–CAP pathway. It is unknown which factors directly turn on the master flhDC operon in accordance with the cell cycle, and how. After the roles of global regulators on flagellar gene expression are specified, the complex regulatory network connecting flagellation and cell division will be uncovered. 

IV. MORPHOLOGICAL PATHWAY 

A. Steps in the morphological pathway The order of the steps of the construction of a flagellum (the morphological pathway) has been analyzed in the same way as for bacteriophages, analyzing the intermediate structures in various flagellar mutants and aligning them in size from small to large. Flagellar construction starts from the cytoplasm, progresses through the periplasmic space, and finally extends to the outside of the cell (Fig. 39.3). 1. Cytoplasm The smallest flagellar structure recognizable by electron microscopy is the MS-ring complex; therefore, the MS-ring complex is regarded as the construction base onto which two other flagellar substructures attach, the C ring and the C rod. 2. Periplasmic space The second-smallest structure consists of a rod on the MS-ring complex. When the rod has grown large enough to reach the outer membrane, the hook starts to grow. However, the outer membrane physically hampers the hook growth until the outer ring complex makes a hole in it. Among flagellar proteins, FlgH and FlgI, the component proteins of the outer ring complex, are exceptional in terms of the manner of export; these two proteins have cleavable signal peptides and are exported through the GSP. 3. Outside the cell Once the physical block by the outer membrane has been removed, the hook resumes growth with the aid of FlgD until the length reaches 55 nm. Then, FlgD is replaced by HAPs, which is followed by the filament growth. The filament growth proceeds only in the presence of FliD (HAP2 or filament cap protein); without this cap, exported flagellin molecules are lost to the medium. The number of genes necessary to proceed through each step of the construction varies; in early stages there are many genes whose roles are unidentified. Between the MS-ring complex and the rod, more than 10 genes are required. Some of those gene products form the C ring, and others are involved in the formation of the flagellar protein-specific export system. B. Flagellar protein export as a type III secretion export system There are several ways to export proteins outside bacterial cells. The best known pathway is the GSP. However, many flagellar proteins cannot pass through this system, because they do not have the signal sequences that are necessary for recognition by GSP. As the number of examples of other types of protein transport has increased, the name became inappropriate. GSP is now categorized as the type II secretion system. There are now more than five secretion systems, but here I will briefly explain the type III secretion system (TTSS for short). The flagellar protein export system is now regarded as a type III secretion system. The flagellar export apparatus consists of at least eight components. The amino acid sequences of these proteins share homology with those used for export of virulence factors from many pathogenic bacteria. Even the structures of these two distinguishable systems resemble each other. The needle complex found first in S. typhimurium and then in Shigella looks like the basal body, consisting of several ring structures and a rod or needle.  C. The kinetics of morphogenesis The morphological pathway of the flagellum described indicates the order of the construction steps but ignores the time consumed at each step. In order to achieve coherent cell activities, flagellar construction has to be synchronized with cell division. The most time-consuming step of flagellation seems to be the filament elongation because filaments can grow over generations. The growth process of the filament and the hook have been carefully analyzed. By taking a closer look at elongation modes of these two polymers, we will get a glimpse of the whole kinetic process of flagellar construction. 1. Filament growth In bacteria with peritrichous flagella, the number and the length of flagella are fairly well defined; there are 7–10 flagella per cell, and the average length of filament is 5–8_m for S. typhimurium. Adefined number of flagella have to be supplied at each cell division. 

A large deviation from this number will cause disastrous results to the cell—either no flagella at all or too many flagella to swim. The number of flagella must be genetically controlled. On the other hand, filament growth seems free from genetic control, because it continues over generations. From statistical analysis of the length distribution, the elongation rate of filaments is estimated to vary inversely to the length; thus, a filament grows rapidly in the beginning and gradually slows down to a negligible rate. 2. Hook growth In contrast to the wide distribution of filament lengths, the hook length is rather well controlled at 55 nm with a deviation of 6 nm. Mutations in the fliK gene result in hooks with unlimited length, called polyhooks. FliK is not a molecular ruler, since a truncation in FliK also gives rise to polyhooks but not short hooks. If the ruler is short, the product measured should be short. Just recently, it was shown that mutations in switch proteins (FliG, FliM, and FliN) gave rise to short hooks with a defined length, indicating that the C ring serves as a measuring cup for hook monomers. Statistical analysis of the length distribution of polyhooks reveals that the hook grows in a manner similar to the filament; it starts out growing at 40nm/min and exponentially slows down to reach a length of 55 nm. After the length is 55 nm, the hook grows at a constant rate of 8 nm/min. It takes many generations for polyhooks to grow as long as several _m. Studies of the correlation between flagellation and cell division are underway, but no definite schemes have been found. V. CONCLUSION The analysis of the flagellar structure is almost complete; most components of the flagellum have been identified, and the pathway of flagellar construction has been revealed. That is, the roles of ~40 flagellar genes in the flagellar construction are now known. We will continue searching for the detailed mechanism of flagellation, including the relationship to cell division. One of the immediate goals is to answer a simple but important question, What is rotating against what? This question stems from the controversy that started at the beginning of the flagellar research. Without knowing the rotor and the stator, the mechanism of motor function will never be understood. And then, we want to answer a more intriguing and difficult question, What is the ancestor of the flagellum? The question arose from the recent discovery of a similarity between the flagellum and the pathogenicity— not only are the gene sequences between the two systems homologous, but also their supramolecular structures resemble each other. This also leads us to the most basic question, What is the flagellum?