Kelly T. Hughes & Phillip D. Aldridge
 

Kelly T. Hughes and Phillip D. Aldridge are in the Department of Microbiology, University of Washington, Seattle, Washington 98195, USA.
Correspondence should be addressed to K T Hughes. e-mail: http://www.nature.com/nsb/email_response/email.taf?address=hughes%40u.washington.edu

A cap at the tip of the bacterial flagellum uses a dynamic differential binding of individual subunits to allow the filament tip to grow, achieving control of assembly far from the point of protein translation.

Flagellum structure
The flagellum is generally divided into three structural components (Fig, 1a): (i) a basal body, which is a molecular rotary motor that includes a drive shaft traversing both the inner and outer membranes; (ii) a hook, which acts as a flexible joint between the basal body and the outer filament; and (iii) the filament itself, which is a long helical structure composed of up to 30,000 polymerized flagellin subunits². A torque generator at the base of the flagellum utilizes energy available from an electrochemical proton gradient providing the force necessary to turn the rotor, and propel the organism through a liquid environment^3-5. In response to environmental cues, flagella rotation is regulated by accompanying chemosensory machinery to accomplish directed movement^{6, 7}.

Flagellum assembly
Assembly of the flagellum begins with the insertion of a ring structure within the cytoplasmic membrane⁸. When the cytoplasmic ring structure is completed, it seals off a disc of membrane into which a secretion apparatus is constructed⁹. The secretion apparatus is needed to export flagellar structural subunits beyond the inner, cytoplasmic membrane. Beyond the cytoplasm, secreted subunits self-assemble into the growing structure. After hook-basal body completion and just prior to filament elongation, proteins that make up the hook-filament junction are added, followed by the cap protein. At this point, the filament subunits are added between the hook-filament junction proteins and the cap. The efficient addition of consecutive filament subunits requires the cap structure, which stays at the very distal end of the growing filament. The filament maintains a central channel that is 30 Å wide¹⁰, through which partially folded subunits must pass and have the ability to self-assemble at the tip of the growing structure. Amazingly, the external filament is 200 Å in diameter, yet subunits are added at the distal tip that can be as far as 15 m away from the entry portal in the cytoplasmic membrane.

Even though there is clear evidence that the flagellar filament can self assemble¹¹, the cap of the flagellar filament serves to control its assembly¹². The cap enables the filament to polymerize with high efficiency, so that every filament subunit that reaches the tip inserts into place. At the same time the bacterium continues to secrete other proteins through the filament that are presumed to simply pass by the cap and out into the extracellular medium. These include excess hook-filament junction subunits¹³, excess cap¹⁴ and a negative regulator of flagellin gene transcription¹⁵. This suggests that the cap acts as a gatekeeper that selectively retains filament subunits and may even play a role in helping them to fold into place.

Role of cap in filament polymerization
Before describing the mechanism of the capping function, we will first describe the physical characterization of the cap, and the role of the cap in filament polymerization in vivo and in vitro that were explained by the capping model proposed by Yonekura et al.¹. Purified cap forms a decamer in solution, although experimental observations support the model that the native capping function is performed by a pentamer^{16, 17}. Reconstitution of the cap onto filaments reveals that the planar side of the pentamer is the outer surface, while the disordered terminal regions of cap subunits are embedded at the filament/cap interface¹⁸. Bacteria that harbor mutations in the cap gene continue to secrete filament subunits¹⁹, but those subunits do not assemble and instead dissipate into the external medium (Fig. 1b). If purified filament subunits are added in sufficient, high concentration to the extracellular medium, the filament will polymerize onto the hook-filament junction of a cap-less hook-basal body (Fig. 1c)²⁰. However, if even one cap monomer is added prior to the addition of filament subunits, it will block in vitro polymerization of externally added filament monomers^{17, 20}. If excess cap is added in vitro, then a functional cap pentamer will form on the end of the hook-basal body²¹. In this case, internally synthesized filament subunits will polymerize normally, but externally added filament subunits will not (Fig. 1d). This suggests an additional role of the cap is to effectively increase the local concentration of the filament subunits that are synthesized in the cytoplasm at the filament-cap junction. As a result, a nascent, unpolymerized subunit is localized in proximity to the most recently added subunit to allow it to fold into place in a thermodynamically favored position at the filament tip.

The structure of the filament is a spiral arrangement of filament monomers of 5.5 monomers added per turn while the cap is a planar pentamer of identical subunits²². How can the planar cap remain firmly attached to the tip of the elongating spiral filament and still allow for selective polymerization of filament subunits and passage of nonfilament proteins between itself and the growing tip?

A possible answer to this mystery has now come from the electron cryomicroscopy of the cap–filament complex by Yonekura et al.¹ The cap structure looks like a pentagonal disc with five legs¹⁷. Each leg is differentially attached to the filament subunits at the filament tip because the cap is planar and the filament end is axial. Yonekura et al.¹ have found that beneath the cap plate a cavity was revealed that is large enough to allow a newly arrived filament subunit to complete its tertiary fold just prior to its final quaternary placement in the filament. The three-dimensional density map of the cap-filament junction also showed the five sides pertaining to each cap monomer at the junction. In addition, gaps between the cap plate and the filament end were observed. One of the five gaps is distinctly larger than the other four at 25 50 Å. This would be predicted from the axial stagger of individual filament subunits packed in the filament and the site at which the authors propose new filament subunits would be added. The addition of a new filament would force the cap outward to the next most stable position. This is effectively a step up 4.7 Å and over 6.5 Å. Thus, as a flagellin subunit is added in the lowest position (the hole), the cap `walks' along the top of the spiral filament staircase in a direction opposite to that of filament elongation, taking one `step' with the addition of each filament subunit (Fig. 2). This model is similar to the action of unscrewing a lid from a jar, although for the flagellar filament the lid (or cap) never comes off the jar (the growing filament). The cap would make a complete rotation with every 55 filament subunits added.

The energy required to move the cap could come from the binding energy of each newly incorporated filament subunit. This model proposed by Yonekura et al.¹ (and beautifully illustrated at http://www.npn.jst.go.jp/yone.html) suggests a highly refined mechanism that satisfies all the requirements for filament self-assembly and represents a novel discovery in structural design. This mechanism explains how control of assembly can occur even if it is outside the cell far removed from cellular processes such as transcription and translation.