Length of the Flagellar Hook and the Capacity of the Type III Export Apparatus

See allHide authors and affiliations

Science  23 Mar 2001:
Vol. 291, Issue 5512, pp. 2411-2413
DOI: 10.1126/science.1058366


Length determination in biology generally uses molecular rulers. The hook, a part of the flagellum of motile bacteria, has an invariant length. Here, we examined hook length and found that it was determined not by molecular rulers but probably by the amount of subunit protein secreted by the flagellar export apparatus. The export apparatus shares common features with the type III virulence-factor secretion machinery and thus may be used more widely in length determination of structures other than flagella.

The bacterial flagellum, a rotary device for motility, is a supramolecular structure consisting of more than 20 different proteins that build up into three distinctive substructures: the filament, hook, and basal structure (1, 2). In Salmonella enterica serovar Typhimurium, the hook has an average length of 55 nm with a standard deviation of 6 nm (3), whereas the filament length varies over a wide range. In order to elucidate the mechanism regulating the invariant length of the hook, we used mutants that give rise to hooks of indefinite length, called “polyhooks.” The mutation sites in these strains are not in the hook protein gene (flgE) but are in thefliK gene, suggesting that FliK acts as a length controller or a molecular ruler of the hook (4–7).

If FliK were a simple molecular ruler, truncated FliK's should produce shorter, not longer, hooks. However, allfliK mutants so far studied give rise to long polyhooks (8). In order to identify what controls the hook length, it is necessary to find mutants that produce short hooks. After an extensive survey, we found such strains with mutations in thefliG, fliM, andfliN genes.

Mutation in these three genes gives rise to different phenotypes, depending on the degree of their defects: Fla(filament-less) mutants derive from major defects, and Mot (motility-less) or Che(chemotaxis-less) mutants are from minor defects (9, 10). These three genes are commonly called the switch genes, putting an emphasis on the behavioral phenotype Che, in which the switch mechanism of the motor rotation is perturbed.

In the early stage of the survey by electron microscopy, we found short hooks in the intact flagella isolated from fliGmutants with Che phenotypes (Fig. 1A). The hook portion of the flagella looked less curved than that of the wild type, implying the shortness of the hook. In order to reveal the hook length more explicitly, hook-basal bodies (HBBs) were isolated (Fig. 1B), and the hook lengths were measured to produce diagrams of length distribution. The hook length of SJW2325 (fliG/Che) was 26.8 ± 8.0 nm, about half the length of the wild type (Fig. 2A).

Figure 1

Electron micrographs of short hooks. (A) The basal regions and (B) the hook basal bodies of intact flagella (flagellar filaments attached with the basal bodies) isolated from S. enterica serovar Typhimurium wild-type SJW1103 (left) and a switch gene mutant SJW2325 (fliG/Che) (right). (C) Among 12 (4 fliG/Fla, 4fliM/Fla, and 4fliG/Fla) mutants examined, only 1 mutant, SJW2409 (fliM/Fla), produced basal bodies with short hooks. Intact flagella or partial basal structures were purified as described previously, with minor modifications (3, 16). Samples were negatively stained with 2% phosphotungstic acid (pH 7.0) and observed with a JEM-1200EXII transmission electron microscope (JEOL, Tokyo). Micrographs were taken at an accelerating voltage of 80 kV. The scale bar represents 50 nm.

Figure 2

Length distribution of hooks isolated from the switch gene mutants (A) SJW2325 (fliG/Che) and (B) SJW1813 (fliM/Mot). The hook length was measured directly from negatives (taken at a magnification of ×20,000) using a scale lupe (×15).

Next, we isolated HBBs from both Che and Mot mutants of all the switch genes in our collection (three fliG/Che, sevenfliM/Che, onefliN/Che, twofliG/Mot, fivefliM/Mot, and fourfliN/Mot mutants). Their hook lengths were measured (Fig. 2 and Table 1). Many were as short as SJW2325, whereas the others were a little shorter than the wild type. For example, the hook length of SJW1813 (fliM/Mot) was 41.8 ± 6.8 nm, a length which is close to but distinguishable from wild-type hooks (Fig. 2B). All mutant hooks were shorter than the wild type and fell into one of two groups: 25-nm hook or 45-nm hook (Table 1).

Table 1

Hook length of switch gene mutants in S. enterica serovar Typhimurium. The phenotypes of Chemutants were either tumbled (T) or smooth (S) swimming. The mutation sites were taken from (25) and (26); amino acid abbreviations are given in (27). The hook length data show the average hook length ± SD. The total number of measured particles is indicated in parentheses. n.d., not determined; dash indicates that only a few examples were observed (see Fig. 1C).

View this table:

Most of the Che mutants examined gave rise to the 25-nm hook. The motor rotation of each Che mutant is biased to either clockwise or counterclockwise, giving rise to tumbled or smooth swimming, respectively. Because both swimming patterns were observed in mutants with the 25-nm hook, we conclude that the shorter hook length does not affect the motor function or bias. Mot mutants gave rise to both 25- and 45-nm hooks. Mot mutations causing changes in the COOH-terminal half domain of each protein gave rise to 25-nm hooks, whereas those causing changes near the NH2-terminal regions gave 45-nm hooks (Table 1).

The largest part of flagellum is constructed outside the cell, requiring an export system specific for the flagellar proteins. Most of the flagellar proteins have no leader sequences and are exported by a type III secretion system, the conventional name for one type of virulence-factor secretion systems in pathogenic bacteria (11). When the needle complex (NC) was identified as the type III secretion machinery (12), the morphological resemblance and homologous component proteins between HBB and NC suggested that these two structures share a common ancestor.

Needle length of Salmonella, which is invariant in wild type as found for hook length, can be elongated by aninvJ (a fliK homolog) mutation (13). Furthermore, overproduction of MxiH (the needle protein) in Shigella flexneri also gives rise to polyneedles (14). Underlying these phenomena, we think, is the same mechanism as the hook length control. We now scrutinize the flagellar export apparatus to elucidate the mechanism.

The switch proteins form a hollow-cup structure called the C ring (C for cytoplasmic) beneath the flagellar basal body. A current model (15) for the C ring function emphasizes its roles in motor function; FliG forms the rotor, and a complex of FliM and FliN works in the switching of rotational direction. However, the C ring is involved in flagellar assembly before it works as a part of the motor; without switch proteins, the flagellum is not formed beyond the MS ring complex (16). The main body of the flagellar export apparatus is the C rod, which is located in the center of the C ring (17). How the C ring interacts with the C rod and how defects in the export apparatus affect the rotor or switch function at the same time are unknown. Because the C ring is a thin wall of proteins (18, 19), any conformational changes inside the ring would affect the whole structure.

We propose a model for the hook length control by the C ring. From the lattice constant of the hook, the number of hook subunits contained in the hook is calculated to be 11 subunits/5 nm, giving 121 subunits/55 nm for the wild-type hook, 99 subunits for the 45-nm hook, and 55 subunits for the 25-nm hook (20). Why should the lengths of short hooks be discontinuous? If this number of hook subunits was accumulated in the C ring at binding sites on the inner wall and the C ring was composed of ∼30 identical units of the FliG, FliM, FliN complex, and if there were four binding sites on each unit of the C ring wall, the C ring would hold 120 hook subunits. If the binding sites decreased to three or two, the total number of the hook subunits would be 90 or 60, respectively. These numbers approximate to the numbers of hook subunits observed in the short mutant hooks. Thus, the physical capacity of the C ring may determine the hook length, and the C ring may act as a quantized measuring cup. The unit capacity and the subsequent unit hook length would be 30 subunits and 14 nm, as found in SJW2409 (fliM /Fla ) (Fig. 1C).

If FliK is not a molecular ruler of hook length, then what is the role of FliK? The length distribution of polyhooks shows a peak at 55 nm, indicating that the hook length is controlled, even in the absence of FliK (21). In order to terminate the elongation, the hook-cap protein (FlgD) has to be replaced by a hook-filament junction protein (FlgK). However, FlgK itself is not necessary for the hook length control (3). In mostfliK mutants, FlgD stays at the tip of the hook, allowing the continuous elongation of the hook (22). The flagellar secretion system has two modes: one specific for the hook proteins and the other for flagellin (23). TheseflgK mutants secrete large amounts of flagellin into media, indicating that the secretion mode is turned to the latter (24). Moreover, the flgK mutants secrete FliK, which terminates the secretion of hook proteins (7). Thus, FliK is likely to be required in the termination of hook elongation by changing the mode of secretion.

Thus, a mechanism of the hook length determination could be as follows. The hook monomers (FlgE) accumulate to fill the C ring and are secreted en bloc to form the hook of a finite length. When the C ring is empty, FliK is secreted, which converts the mode of secretion into that for flagellin. Then, FlgD at the tip of a nascent hook is replaced by FlgK, which terminates the hook elongation.

  • * To whom correspondence should be addressed. E-mail: aizawa{at}


View Abstract

Navigate This Article