Fimbriae, Pili

Pili, also known as fimbriae, are proteinaceous, filamentous polymeric organelles expressed on the surface of bacteria. They range from a few fractions of a micrometer to greater than 20_m in length and vary from less than 2 to 11 nm in diameter. Pili are composed of single or multiple types of protein subunits, called pilins or fimbrins, which are typically arranged in a helical fashion. Pilus architecture varies from thin, twisting threadlike fibers to thick, rigid rods with small axial holes. 

Thin pili with diameters of 2–3 nm, such as K88 and K99 pili, are often referred to as “fibrillae”. Even thinner fibers (_2 nm), which tend to coil up into a fuzzy adhesive mass on the bacterial surface, are referred to as thin aggregative pili or curli. High-resolution electron microscopy of P, type 1, and S pili of Escherichia coli and Haemophilus influenzae pili has revealed that these structures are composite fibers, consisting of a thick pilus rod attached to a thin, short distally located tip fibrillum. Pili are often expressed peritrichously around individual bacteria, but some, such as type 4 pili, can be localized to one pole of the bacterium. Pili expressed by gram-negative bacteria have been extensively characterized, and the expression of pili by gram-positive bacteria has also been reported. The numerous types of pili assembled by both gramnegative and gram-positive organisms have been ascribed diverse functions in the adaptation, survival, and spread of both pathogenic and commensal bacteria. Pili can act as receptors for bacteriophage, facilitate DNAuptake and transfer (conjugation), and, in at least type 4 pilus, function in cellular motility. The primary function of most pili, however, is to act as scaffolding for the presentation of specific adhesive moieties. Adhesive pilus subunits (adhesins) are often incorporated as minor components into the tips of pili, but major structural subunits can also function as adhesins. Adhesins can mediate the interaction of bacteria with each other, with inanimate surfaces, and with tissues and cells in susceptible host organisms. The colonization of host tissues by bacterial pathogens typically depends on a stereochemical fit between an adhesin and complementary receptor architecture. Interactions mediated by adhesive pili can facilitate the formation of bacterial communities such as biofilms and are often critical to the successful colonization of host organisms by both commensal and pathogenic bacteria. 

I. HISTORICAL PERSPECTIVE AND CLASSIFICATION OF PILI 

Pili were first noted in early electron microscopic investigations as nonflagellar, filamentous appendages of bacteria. In 1955, Duguid designated these appendages “fimbriae” (plural, from Latin for thread or fiber) and correlated their presence with the ability of E. coli to agglutinate red blood cells. Ten years later Brinton introduced the term “pilus” (singular, from Latin for hair) to describe the fibrous structures (the F pilus) associated with the conjugative transfer of genetic material between bacteria. Since then “pilus” has become a generic term used to describe all types of nonflagellar filamentous appendages, and it is used interchangeably with the term “fimbria.” Historically, pili have been named and grouped based on phenotypic traits such as adhesive and antigenic properties, distribution among bacterial strains, and microscopic characterizations. In the pioneering work of Duguid and co-workers, pili expressed by different E. coli strains were distinguished on the basis of their ability to bind to and agglutinate red blood cells (hemagglutination) in a mannose sensitive (MS) as opposed to a mannose resistant (MR) fashion. Pili mediating MS hemagglutination by E. coli were designated type 1 pili and these pili have since been shown to recognize mannose-containing glycoprotein receptors on host eukaryotic cells. Morphologically and functionally homologous type 1 pili are expressed by many different species of Enterobacteriaceae. Despite their similarities, however, type 1 pili expressed by the various members of the Enterobacteriaceae family are often antigenically and genetically divergent within their major structural subunits. In contrast to type 1 pili, most other pili so far identified are either nonhemagglutinating or mediate MR hemagglutination. These pili are very diverse and possess a myriad of architectures and different receptor binding specificities and functions. Since the discovery and initial characterization of pili in the 1950s, substantial advances have been made in our understanding of the genetics, biochemistry, and structural and functional aspects of these organelles. A vast number of distinct pilus structures have been described and new types of pili continue to be identified. Pili are now known to be encoded by virtually all gram-negative organisms and are some of the best-characterized colonization and virulence factors in bacteria. Here we classify pili that are expressed by gram-negative bacteria into six groups according to the mechanisms by which they are assembled. This classification scheme is not all inclusive, but provides a convenient means for discussing the diverse types of pili, their functions, structures and assembly. Representatives of various pilus types assembled by the various pathways discussed in the following sections are listed in Table 38.1 and electron micrographs of the various pilus types are shown in Fig. 38.1.  

II. CHAPERONE–USHER PATHWAY 

All pilins destined for assembly on the surface of gramnegative bacteria must be translocated across the inner membrane, through the periplasm, and across the outer membrane. To accomplish this, various adhesive organelles in many different bacteria require two specialized assembly proteins, a periplasmic chaperone and an outer membrane usher. Chaperone–usher assembly pathways are involved in the biogenesis of over 30 distinct structures, including composite pili, thin fibrillae, and nonfimbrial adhesins. Here, we focus on the structure and assembly mechanisms of the prototypical P and type 1 pilus chaperone–usher systems. A. Molecular architecture P and type 1 pili are both composite structures consisting of a thin fibrillar tip joined end to end to a right-handed helical rod (Fig. 38.2A and B). Chromosomally located gene clusters that are organizationally and functionally homologous encode P and type 1 pili (Fig. 38.2D and E). The P pilus tip is a 2-nmwide structure composed of a distally located adhesin (PapG), a tip pilin (PapE), and adaptor pilins (PapF and PapK). The PapG adhesin binds to Gal_(1-4)Gal moieties present in the globoseries of glycolipids found on the surface of erythrocytes and kidney cells. Consistent with this binding specificity, P pili are major virulence factors associated with pyelonephritis caused by uropathogenic E. coli. The minor pilin PapF is thought to join the PapG adhesin to the tip fibrillum, the bulk of which is made up of a polymer of PapE subunits. PapK is thought to terminate the growth of the PapE polymer and to join the tip structure to the rod. The pilus rod is composed of multiple PapA subunits joined end to end and then coiled into a right-handed 6.8-nm-thick helical rod having a pitch distance of 24.9 Å and 3.28 subunits per turn. The rod is terminated by a minor subunit, PapH, which may serve to anchor the pilus in the membrane. Similar to the P pilus structure, the type 1 pilus has a short, 3-nm-wide fibrillar tip made up of the mannose-binding adhesin, FimH, and two additional pilins, FimG and FimF. The FimH adhesin mediates attachment to mannosylated receptors expressed on a wide variety of cell types and has been shown to be a significant virulence determinant for the development of cystitis. The type 1 tip fibrillum is joined to a rod composed predominantly of FimA subunits arranged in a 6- to 7-nm diameter helix with a pitch distance of 23.1 Å and 3.125 subunits per turn. Both type 1 and P pilus rods have central axial holes with diameters of 2–2.5 and 1.5 Å, respectively. Despite the architectural similarities, type 1 pili appear to be more rigid and prone to breaking than P pili. Some reports have argued that, unlike the P pilus system in which the tip subunits are thought to be located only within the tip, some of the type 1 tip subunits may also be occasionally intercalated within the rod structure. In both P and type 1 pili, the major pilin subunits making up the rods are organized in a head-to-tail manner. Additional quaternary interactions between subunits in adjacent turns of the helical rod appear to stabilize the structure and may help drive the outward growth of the organelle during pilus assembly (see later). The disruption of these latter interactions by mechanical stress or by incubation in 50% glycerol can cause the pilus rod to reversibly unwind into a 2-nm-thick linear fiber similar in appearance to the tip fibrillum (Fig. 38.2C). 

Bullitt and Makowski (1995) have proposed that the ability of the pilus rods to unwind allows them to support tension over a broader range of lengths. This may help P and type 1 pili better withstand stress, such as shearing forces from the bulk flow of fluid through the urinary tract, without breaking. In addition to composite structures exemplified by P and type 1 pili, chaperone–usher pathways also mediate the assembly of thin fibrillae such as K88 and K99 pili and nonfimbrial adhesins. K88 and K99 pili are 2- to 4-nm-thick fibers that mediate adherence to receptors on intestinal cells. They are significant virulence factors expressed by enterotoxigenic E. coli (ETEC) strains that cause diarrheal diseases in livestock. These pili were given the “K” designation after being mistakenly identified as K antigens in E. coli. In contrast to P and type 1 pili, the adhesive properties of K88 and K99 pili are associated with the major pilus subunits. The receptor-binding epitopes on the individual major pilus subunits are exposed on the pilus surface and available for multiple interactions with host tissue. In general, pili with adhesive major subunits, such as K88 and K99 pili, are thin flexible fibrillar structures. In comparison, pili with specialized adhesive tip structures, such as P and type 1 pili, are relatively rigid and rod-like. B. Assembly model The assembly of P pili by the chaperone–usher pathway is the best understood of any pilus assembly pathway. PapD is the periplasmic chaperone and PapC is the outer-membrane usher for the P pilus system. These proteins are prototypical representatives of the periplasmic chaperone and outer-membrane usher protein families. Figure 38.3 presents the current model for pilus assembly by the chaperone–usher pathway, as depicted for P pili. 1. Periplasmic chaperones The PapD chaperone, the PapC usher, and all of the P pilus structural subunits have typical signal sequences recognized by the sec (general secretion) system. The signal sequences are short, mostly hydrophobic amino-terminal motifs that tag proteins for transport across the inner membrane by the sec system. This system includes several inner-membrane proteins (SecD to SecF, SecY), a cytoplasmic chaperone (SecB) that binds to presecretory target proteins, a cytoplasmic membrane-associated AT-Pase (SecA) that provides energy for transport, and a periplasmic signal peptidase. 

As the P pilus structural subunits emerge from the sec translocation machinery into the periplasm, PapD binds to each subunit, facilitating its release from the inner membrane. Each subunit forms an assembly competent, one-to-one complex with PapD. Proper folding of the subunits requires PapD and involves the action of the periplasmic disulfide bond isomerase DsbA. In the absence of PapD, subunits misfold, aggregate, and are subsequently degraded by the periplasmic protease DegP. The misfolding of P pilus subunits is sensed by the CpxA–CpxR two-component system in which CpxA is an inner membrane-bound sensor or kinase and CpxR is a DNAbinding response regulator (Jones, 1997). Activation of the Cpx system alters the expression of a variety of genes and may help regulate pilus biogenesis. The three-dimensional crystal structures of both the PapD and FimC chaperones have been solved. Both chaperones consist of two Ig (immunoglobulin)-like domains oriented into a boomerang shape such that a subunit-binding cleft is created between the two domains. Aconserved internal salt bridge is thought to maintain the two domains of the chaperone in the appropriate orientation. Using genetics, biochemistry, and crystallography, PapD was found to interact with pilus subunits, in part, by binding to a highly conserved motif present at the carboxyl terminus of all subunits assembled by PapD-like chaperones. The finer details of how PapD-like chaperones interact with pilus subunits were unveiled by the recent determination of the crystal structures of the PapD–PapK and the FimC–FimH chaperone–subunit complexes (Sauer, 1999, and Choudhury, 1999) (Fig. 38.4). This work demonstrated that the PapD and FimC chaperones both make similar interactions with their respective subunits. Only the PapD–PapK structure is considered here. PapK has a single domain comprised of an Ig fold that lacks the seventh (carboxy-terminal) _- strand that is present in canonical Ig folds. 

The absence of this strand produces a deep groove along the surface of the pilin subunit and exposes its hydrophobic core. The carboxy-terminal F strand and the A2 strand of PapK form the groove. The PapD chaperone contributes its G1 _-strand to complete the Ig fold of the PapK subunit by occupying the groove, an interaction termed donor strand complementation. This interaction shields the hydrophobic core of PapK and stabilizes immature pilus subunits within the periplasm. Similar interactions are thought to have a central role in the maturation of subunits assembled by a variety of different chaperone–usher pathways. The residues that make up the carboxy-terminal groove formed by subunits and bound by PapD-like chaperones have been shown by mutagenesis studies to be involved in subunit– subunit interactions within the final pilus structure. Thus, in addition to stabilizing immature pilus subunits, the donor strand complementation interaction also caps one of the interactive surfaces of the subunit and prevents premature oligomerization and aggregation of pilus subunits within the periplasm. 2. Outer-membrane ushers After being formed in the periplasm, chaperone– subunit complexes are targeted to the outermembrane usher where the chaperone is released, exposing interactive surfaces on the subunits that facilitate their assembly into the pilus. Studies in the P and type 1 pilus systems have demonstrated that the adhesin–chaperone complexes, PapDG or FimCH, bind tightest and fastest to the usher and that the adhesins are the first subunits assembled into the pilus. The binding of the chaperone–adhesin complex induces a conformational change in the usher, possibly priming it for pilus assembly. Additional subunits are incorporated into the pilus depending, in part, on the kinetics with which they are partitioned to the usher in complex with the chaperone. Conserved aminoterminal regions, in addition to the conserved carboxyterminal motif of the pilus subunits, mediate subunit–subunit interactions within the mature pilus. Differences in the complementary surfaces in these conserved regions from one subunit to another may help dictate which of the subunits can be joined to one another during pilus assembly. 

Thus, the order of the subunits within the final pilus structure is determined by the specific contacts made between the various pilus subunits and also by the differential affinities of the various chaperone–subunit complexes for the usher. In addition to acting as an assembly platform for the growing pilus, the usher protein appears to have additional roles in pilus biogenesis. High-resolution electron microscopy revealed that the PapC usher is assembled into a 15-nm diameter ring-shaped complex with a 2-nm wide central pore (Thanassi, 1998). PapC and other usher family members are thought to have a predominantly _-sheet secondary structure, typical of outer-membrane pore-forming proteins, and they are predicted to present large regions to the periplasm for interaction with chaperone–subunit complexes. After dissociating from the chaperone at the usher, subunits are incorporated into a growing pilus structure that is predicted to be extruded as a 1-subunit-thick linear fiber through the central pore of the usher complex. The packaging of the linear pilus fiber into a thicker helical rod on the outside surface of the bacterium may provide a driving force for the translocation of the pilus across the outer membrane, possibly acting as a sort of ratcheting mechanism to force the pilus to grow outward. Combined with the targeting affinities of the chaperone–subunit complexes for the usher and the binding specificities of the subunits for each other, this may provide all the energy and specificity needed for the ordered assembly and translocation of pili across the outer membrane. III. ALTERNATE CHAPERONE PATHWAY A variation of the chaperone–usher pilus assembly pathway has been identified in strains of ETEC. These bacteria are major pathogens associated with diarrheal diseases of travelers, infants, and young children. ETEC strains produce several types of uniquely assembled adhesive pili that are considered to be important mediators of bacterial colonization of the intestine. The best studied of these pili is CS1, which appears to be composed predominantly of a major subunit, CooA, with a distally located minor component, CooD. Several CS1-like pili have been identified and include CS2, CS4, CS14, CS17, CS19, and CFA/I pili expressed by various ETEC strains and the cable type II pili of Burkholderia cepacia, an opportunistic pathogen of cystic fibrosis patients. Four linked genes, CooA, CooB, CooC, and CooD, are the only specific genes required for the synthesis of functional CS1 pili. Homologous genes required for the production of CS2 and CFA/I pili have also been cloned and sequenced. Electron-microscopic examination reveals that the CS1-like pili are architecturally similar to P and type 1 pili assembled by the chaperone–usher pathway (Fig. 38.1B), although none of the proteins involved in the biogenesis of CS1-like pili have any significant sequence homologies to those of any other pilus system. 

The assembly of CS1-like pili depends on a specialized set of periplasmic chaperones that are distinct from those of the chaperone–usher pathway described previously. Therefore, we refer to this mode of pilus assembly as the alternate chaperone pathway. In the case of CS1 pili, the chaperone CooB binds to and stabilizes the major and minor pilin subunits, CooA and CooD, which enter into the periplasm in a sec-dependent fashion (Fig. 38.5). Both CooA and CooD share a conserved sequence motif near their carboxy-termini that may function as a chaperonerecognition motif. One of the functions of CooB appears to be the delivery of the pilin subunits to an outer-membrane protein, CooC, which may function as a channel, or usher, for the assembly of pilus fibers. In addition to the pilin subunits, CooB also binds to and stabilizes CooC in the outer membrane. In the absence of the CooB chaperone, CooC and the pilin subunits are degraded. Although less well defined, the assembly of CS1 and related structures appears similar in many respects to the assembly of pili by the classic chaperone–usher pathway. Because CS1-like pili do not appear to be related to those assembled by the chaperone–usher pathway, it has been suggested that these two pilus assembly systems arose independently through convergent evolution. 

IV. TYPE II SECRETION PATHWAY FOR TYPE 4 PILUS ASSEMBLY 

Type 4 pili are multifunctional structures expressed by a wide diversity of bacterial pathogens. These include Pseudomonas aeruginosa, Neisseria gonorrhoeae and N. meningitidis, Moraxella species, Azoarcus species, Dichelobactor nodus, and many other species classified in these and other genera. Type 4 pili are significant colonization factors and have been shown to mediate bacterial interactions with animal, plant, and fungal cells. In addition, these pili can modulate target-cell specificity, function in DNA uptake and biofilm formation, and act as receptors for bacteriophage. Type 4 pili are also associated with a flagellaindependent form of bacterial locomotion, called twitching motility, which allows for the lateral spread of bacteria across a surface. Type 4 pili are 6-nm-wide structures typically assembled at one pole of the bacterium. They can extend up to several micrometers in length and are made up of, primarily if not completely, a small subunit usually in the range of 145–160 amino acids. These subunits have distinctive features, including a short (6–7 amino acids), positively charged leader sequence that is cleaved during assembly, N-methylphenylalanine as the first residue of the mature subunit, and a highly conserved, hydrophobic amino-terminal domain. The adhesive properties of type 4 pili are, in general, determined by the major pilus subunit. Additional minor components, however, may associate with these pili and alter their binding specificities. In the case of Neisseria, a tiplocalized adhesin, PilC1, appears to mediate bacterial adherence to epithelial cells. Recently, a second class of type 4 pili, referred to as class B or type 4B, has been defined. Type 4B pili were initially characterized in enteric pathogens and include the toxin-coregulated pilus (TCP) of Vibrio cholera, the bundle-forming pilus (BFP) of enteropathogenic E. coli (EPEC), and the longus and CFA/III pili of ETEC. Compared to the typical type 4 pilins (referred to as class A or type 4A), the known type 4B pilins are somewhat larger and have a longer (13–30 amino acids) leader sequence. Also, in place of Nmethyl- phenylalanine as the first amino acid in the mature pilus subunit, type 4B subunits have other methylated residues such as N-methyl-methionine for TCP and N-methyl-leucine for BFP. TCP, BFP, and longus pili form large polar bundles over 15_m in length (Fig. 38.1C). In contrast, CFA/III pili are 1– 10_m long and are peritrichously expressed. The number of pili classified as type 4B is increasing and now includes the R64 thin pilus, an organelle involved in bacterial conjugation. Parge and coworkers solved the crystal structure of the type 4A pilin subunit (PilE) from N. gonorrhoeae in 1995. This work greatly advanced understanding of the structure, function, and biogenesis of type 4 pili. PilE contains 158 amino acids and was determined to have an overall ladle shape, being made up of an _–_ roll with a long hydrophobic amino-terminal _-helical spine (residues 2–54) (Fig. 38.6). All type 4 pilins (types 4A and 4B) are predicted to have a fairly similar structure. The carboxy-terminal domain of type 4 pilins possesses hypervariable regions that affect the binding specificities and antigenicity of type 4 pili. In PilE, these hypervariable regions include a sugar loop (residues 55–77) with an O-linked disaccharide at position Ser-63 and a disulfide-containing region (residues 121–158), which, despite having a hypervariable nature, adopts a regular _-hairpin structure (_5–_6) followed by an extended carboxy-terminal tail. Adisulfide-containing carboxy-terminal hypervariable region is common among the type 4 pilins. The remainder of PilE was shown to consist of two _-hairpins forming a fourstranded antiparallel _-sheet (residues 78–93 and 103–122) with a connecting _2–_3 loop region (residues 94–102). Through systematic modeling, Parge et al. showed that PilE was probably assembled into pili as monomers arranged in a helix with about five PilE subunits per turn and a pitch distance of approximately 41 Å. PilE subunits are predicted to be packed into pili as a three-layer assemblage consisting of an inner core of coiled conserved hydrophobic _-helices surrounded by _-sheets and an outermost layer composed of the disaccharide and hypervariable regions. Hydrophobic packing of the inner core of _-helices along with the flexibility of these helices may permit type 4 pili to bend and adopt twisted, bundled conformations, as seen in Fig. 38.1C. Hydrogen bonds throughout the middle layer of _-sheets may provide much of the mechanical stability for the pilus. 

The hypervariable outermost layer is not an integral part of the pilus structure and associates with the middle layer of _-sheets through only a few conserved interactions. Thus, the outermost region can be structurally pliant and accommodate extreme amino acid changes that can lead to antigenic variation and altered binding specificities without disrupting the assembly of the pilus. The antigenic characteristics of type 4 pili synthesized by N. gonorrhoeae can be modified extensively by a remarkable mechanism. This pathogen encodes more than 15 distinct silent pilin genes, termed PilS, that lack the invariant aminoterminal domain present in PilE. By recombination of silent PilS genes with the PilE locus, a single neisserial strain can theoretically express greater than 10 million PilE variants. The biogenesis of type 4 pili is substantially more complicated than pilus assembly by the chaperone– usher or alternate chaperone pathways. Type 4 pilus assembly requires the expression of a myriad of genes that are usually located in various unlinked regions on the chromosome. Exceptions include TCP, BFP, and the R64 thin pilus, which are currently the only type 4 pili for which the majority of the genes required for pilus biogenesis are located within a single genetic locus. Although chromosomally located genes encode most type 4A pili, all known type 4B pili, with the exception of TCP, are encoded by plasmids. The number of genes essential for type 4 pilus biogenesis and function ranges from 14 (for pili such as BFP) to over 20 (for structures such as the type 4A pili of N. gonorrhoeae). In P. aeruginosa, it is estimated that about 0.5% of the bacterium’s genome is involved in the synthesis and function of type 4 pili. Among the various bacterial species expressing type 4 pili, the genes encoding the type 4 pilus structural components are similar, whereas the regulatory components surrounding them are typically less conserved. Several gene products are currently known to be central to the assembly of type 4 pili. These include a prepilin peptidase that cleaves off the leader peptide from nascent pilin subunits; a polytopic inner membrane protein that may act as a platform for pilus assembly; a hydrophilic nucleotide-binding protein located in the cytoplasm or associated with the cytoplasmic face of the inner membrane, which may provide energy for pilus assembly; and an outermembrane protein complex that forms a pore for passage of the pilus to the exterior of the bacterium. Many of the components involved in type 4 pilus assembly share homology with proteins that are part of DNA uptake and protein secretion systems, collectively known as the main terminal branch of the general secretory (sec-dependent) pathway, or type II secretion. These secretion pathways encode proteins with type 4 pilin-like characteristics and other proteins with homology to type 4 prepilin peptidases and outer-membrane pore-forming proteins. Whether type II secretion systems can assemble pili or piluslike structures is not known. Meyer and colleagues have described a model for type 4 pilus assembly in N. gonorrhoea (Fig. 38.7). The PilE propilin subunits are transported into the periplasm by the sec translocation machinery. Following translocation, the propilin subunits remain anchored in the inner membrane by their hydrophobic amino-terminal _-helical domains, with their hydrophilic carboxy-terminal heads oriented toward the periplasm. The removal of the positively charged PilE propilin leader sequence by the PilD signal peptidase drives the hydrophobic stems of the PilE subunits to associate to form a pilus. An inner-membrane assembly complex made up of several proteins including PilD, PilF, PilG, and PilT aids this process. 

The assembled pilus penetrates the outer membrane through a gated pore formed by the multimeric complex Omc. The PilC adhesin associated with the tips of type 4 pili produced by N. gonorrhoea may facilitate passage of the nascent pili through the Omc pore. One implication of this assembly model is that the aminoterminal region of PilE resides in a continuous hydrophobic environment during both inner-membrane transport and pilus assembly. This may allow polymerization and, interestingly, depolymerization of the pilus to proceed with only minimal energy requirements. Continued extension and retraction of type 4 pili by rounds of polymerization and depolymerization reactions are proposed as the basis of twitching motility, one of the functions of type 4 pili. This process could be controlled by nucleotide-binding proteins such as PilT, PilB, and PilU associated with the inner-membrane assembly complex. The capacity of type 4 pili to depolymerize may also provide a means for transforming DNA, which could potentially interact with the type 4 pilus fiber, to enter into the bacterial cell. 

V. CONJUGATIVE PILUS ASSEMBLY PATHWAY 

In gram-negative bacteria, certain pili, collectively known as conjugative pili, facilitate the interbacterial transfer of DNA. These pili allow donor and recipient bacteria to make specific and stable intercellular contacts before DNA transfer is initiated. In some cases, conjugative pili may also form the conduits for intercellular DNA transfer. Horizontal gene transfer, or conjugation, mediated by conjugative pili is inextricably associated with the spread of antibiotic resistance among bacterial pathogens. Conjugative pili are generally encoded by self-transmissible plasmids that are capable of passing a copy of their genes to a recipient bacterium. Closely related plasmids, with similar replication control systems, are unable to coexist in the same cell. This property has been termed incompatibility, and provides the primary basis for cataloging conjugal plasmids and the pili that they encode. Thus far, in E. coli alone, over 25 incompatibility groups made up of well over 100 distinct plasmids have been defined. Plasmids of a particular incompatibility group usually encode conjugative pili with similar antigenic properties, sensitivities to pilus-specific phages, and morphologies. Among the multitude of known incompatibility groups, three morphologically and functionally distinct types of conjugative pili have been defined: rigid, thick flexible, and thin flexible pili. Rigid conjugative pili are 8- to 11-nm-wide structures that are usually specified by conjugal DNA transfer systems that function well only on solid surfaces. Thick flexible pili, on the other hand, are 8- to 11-nm-wide structures that typically, but not always, promote conjugation on solid surfaces and in liquid media equally well. Conjugal DNA transfer promoted by rigid or thick flexible pili can be enhanced, in some cases, by the presence of thin flexible pili. These pili are similar in appearance to type 4 pili and at least one member of the thin flexible pilus group (the R64 thin pilus) has been identified at the molecular level as a type 4B pilus (described previously). Thin flexible pili appear to function primarily in the stabilization of bacterial mating pairs, increasing the rate of DNA transfer. Conjugation does not occur in the presence of thin flexible pili alone, or in the absence of rigid or thick flexible pili. The most thoroughly studied conjugative pilus is the F pilus encoded by the self-transmissible, broad host range F (fertility) plasmid, a member of the F1 incompatibility (IncF1) group of plasmids borne by E. coli. The F pilus system is prototypical for numerous other conjugation systems and F-pilus biogenesis is distinct from type 4 and other pilus assembly pathways. F pili are 8-nm-thick, flexible helical filaments composed primarily, if not completely, of repeating 7.2 kDa (70-amino-acid) TraA pilin subunits. Donor (F_) cells typically express one to three F pili that are usually 1–2_m long. Each F pilus possesses a 2-nmwide central channel that is lined by basic hydrophilic residues, which could potentially interact with negatively charged DNA or RNA molecules during conjugation. TraAis organized into pentameric, doughnut-like disks that are stacked in the pilus such that successive disks are translated 1.28 nm along the pilus axis and rotated 28.8_ with respect to the lower disk. The TraA pilin has two hydrophobic domains located toward the center and at the carboxy-terminus of the pilin. The hydrophobic domains are thought to extend as antiparallel _-helices from the central axis to the periphery of the pilus shaft. These domains are separated by a short basic region that appears to form the hydrophilic wall of the central channel of the pilus. The amino-terminal domain of TraA is predicted to face the exterior of the pilus. However, this domain is antigenically masked when the amino-terminal residue of TraA is acetylated during maturation of the pilin (see later). This modification is common among all known F-like pilins and appears to cause the amino-terminal domain to be tucked back into or along the pilus shaft. Acetylation is not essential for Fpilus assembly or function, but does help prevent aggregation of F-like pili and affects the phage-binding characteristics of these organelles. Phage are also known to recognize the carboxy-terminal hydrophobic domain of TraA. Although masked within the pilus shaft, the acetylated amino-terminal domain of TraA appears to be exposed in unassembled pilin subunits and at the distal tips of pili. 

The F pilus tip is believed to initiate contact between donor and recipient cells during conjugation. Alterations in the amino-terminal sequence of TraA provide the primary basis for the antigenic diversity observed among various F-like pili. At least 16 gene products encoded by the F plasmid are involved in F pilus assembly and an additional 20 or more are needed for conjugation. Two gene products, TraQ and TraX, mediate the processing of the TraA pilin to its mature form. TraA is synthesized as a 12.8-kDa (121-amino-acid) cytoplasmic propilin that is translocated across the inner membrane where it is proteolytically processed by host signal peptidase I to yield the 7.2-kDa pilin form. TraQ, an innermembrane protein, facilitates the translocation process and may help position the TraA propilin for processing into mature pilin. In the absence of TraQ, the translocation of TraA is disrupted and most of the pilin is degraded. After processing, the amino-terminal residue (alanine) of TraA is acetylated by TraX, a polytopic innermembrane protein. Whereas TraQ and TraX are involved in the maturation of the TraA pilin, 13 additional gene products (TraL, TraB, TraE, TraK, TraV, TraC, TraW, TraU, TraF, TraH, TraG, TrbC, and TrbI) affect the assembly of TraA into the pilus filament. Most of these proteins appear to associate with either the inner or outer bacterial membrane and may constitute a pilus assembly complex that spans the periplasmic space. The exact mechanism by which TraA is assembled into pili is not defined. The mature TraA pilin accumulates in the inner membrane with its aminoterminus facing the periplasm. Both hydrophobic domains of TraA span the inner membrane, with the hydrophilic region of TraA connecting them on the cytoplasmic side. Small clusters of TraA also accumulate in the outer membrane and these may function as intermediates in F pilus assembly and disassembly. Large portions of the TraA sequence have the propensity to assume both _-sheet and _-helical structures, although the _-helical conformation is known to predominate in assembled pili. Frost and co-workers (1984, 1993) have suggested that a shift between _-sheet and _-helical conformations may drive pilus assembly and disassembly. 

F pilus assembly is energy dependent and the depletion of ATP levels by respiratory poisons such as cyanide results in F pilus depolymerization and retraction. It has been postulated that TraA is normally cycled between pili and periplasmic and inner-membrane pools by rounds of pilus outgrowth and subsequent retraction. During conjugation, F pilus retraction is thought to serve a stabilizing function by shortening the distance between bacterial mating pairs and allowing for more intimate contact. Several components of the F pilus assembly machinery share significant homology with proteins encoded by other conjugative systems. These include proteins specified by broad host range plasmids in other incompatibility groups (such as IncN, IncP, and IncW) and many of the proteins encoded by the Ti plasmid-specific vir genes of the plant pathogen Agrobacterium tumefaciens. These bacteria elaborate 10-nm-wide promiscuous conjugative pili, called T pili (Fig. 38.1D), which direct the interkingdom transfer of a specific genetic element, known as T-DNA, into plant and yeast cells. The introduction of T-DNA into plant cells induces plant tumor formation. T pilus assembly by A. tumefaciens requires the expression of at least 12 vir gene products encoded by the Ti plasmid. VirB2 is the major, and possibly only, component of the T pilus and it is predicted to be structurally homologous to the F pilus subunit TraA. Other than possibly stabilizing donor-recipient interactions, it is not clear how F and T pili or any pilus structures function in conjugative DNA-transfer processes. However, substantial evidence exists, at least in the case of F pili, suggesting that pilus components or the pilus itself can serve as a specialized channel for the transmission of DNA and any accompanying pilot proteins across the donor and possibly the recipient cell membranes. In light of this possibility, it is interesting to note that many components of the conjugative pilus systems encoded by the IncF, IncN, IncP, and IncW plasmids and by the vir genes of A. tumefaciens are similar to the Ptl proteins responsible for the export of the multiple subunit toxin of Bordetella pertussis. Furthermore, these secretion systems seem to be distantly related to transport systems used by Legionella pneumophila and Helicobacter pylori to inject virulence factors into host eukaryotic cells. Conjugative pilus systems such as those encoding F and T pili thus appear to be representative of a larger family of macromolecular transport systems. These type IV secretion systems (not to be confused with the secretion of autotransporters, such as IgA proteases, which is also known as type IV secretion) represent a major pathway for the transfer of both nucleic acid and proteins between cells (Zupan et al., 1998). The understanding of how conjugative pili help mediate the intercellular transfer of macromolecules remains a significant challenge. 

VI. EXTRACELLULAR NUCLEATION–PRECIPITATION PATHWAY 

Many strains of E. coli and Salmonella enteritidis produce a class of thin (_2 nm), irregular, and highly aggregated surface fibers known as curli (Fig. 38.1E). These distinct organelles mediate binding to a variety of host proteins, including plasminogen, fibronectin, and human contact-phase proteins. They are also involved in bacterial colonization of inert surfaces and have been implicated in biofilm formation. Curli are highly stable structures and extreme chemical treatment is required to depolymerize them. The major component of E. coli curli is a 15.3-kDa protein known as CsgA, which shares over 86% primary sequence similarity to its counterpart in S. enteritidis, AgfA. The formation of curli represents a departure from the other modes of pilus assembly. Whereas structures exemplified by P, CS1, type 4, and F pili are assembled from the base, curli formation occurs on the outer surface of the bacterium by the precipitation of secreted soluble pilin subunits into thin fibers (Fig. 38.8). In E. coli, the products of two divergently transcribed operons are required for curli assembly. The csgBA operon encodes the primary fiber-forming subunit, CsgA, which is secreted as a soluble protein directly into the extracellular environment. The second protein encoded by the csgBA operon, CsgB, is proposed as inducing polymerization of CsgA at the cell surface. In support of this model, it has been demonstrated that a CsgA_CsgB  donor strain can secrete CsgA subunits that can be assembled into curli on the surface of a CsgA CsgB_ recipient strain. Furthermore, CsgB appears to be interspersed along the length of the curli fiber, where it may initiate branching of the curli structures. In the absence of CsgA, overexpressed CsgB is able to form short polymers on the bacterial cell surface. The csgDEFG operon encodes a gene for a transcriptional activator of curli synthesis (CsgD), and three genes encoding putative assembly factors. One of these factors, CsgG, has been shown to be a lipoprotein that is localized to the outer membrane. In the absence of CsgG, curli assembly does not take place and CsgA and CsgB are rapidly degraded. The precise role of CsgG is not known. Normark and colleagues have suggested that CsgG might act as a chaperone that facilitates the secretion of the CsgA and CsgB and protects them from degradation within the periplasm. It is also possible that CsgG assembles into multimers that could function as a Csg-specific channel within the outer membrane. The roles of the CsgE and CsgF gene products are not known. It has been reported, however, that a strain lacking these two factors can export assembly-competent CsgA, suggesting that the production of CsgG alone is sufficient for functional maturation of the CsgA subunit of curli. 

VII. TYPE III SECRETION PATHWAY 

The various pilus assembly pathways described in the previous sections all rely on components of the sec machinery for the translocation of their respective pilus subunits across the inner membrane. Two new types of pili that are assembled by a sec-independent pathway known as type III secretion have been identified. The type III secretion system is encoded by numerous gramnegative pathogens and enables these bacteria to secrete and inject pathogenic effector molecules into the cytosol of host eukaryotic cells. About 20 gene products, most of which are inner-membrane proteins, make up the type III secretion system. The components mediating type III secretion are conserved in pathogens as diverse as Yersinia and Erwinia, but the secreted effector proteins vary significantly among species. The type III secretion apparatus, which appears to span the periplasmic space, resembles the basal body of a flagellum connected to a straight rod that extends across the outer membrane. Interestingly, all type III secretion systems encode some components with homologies to proteins involved in flagellar assembly. The secretion of proteins by the type III system is an ATP-dependent process that involves no distinct periplasmic intermediates. Type IIIsecreted proteins of EPEC and the plant pathogen Pseudomonas syringae have recently been shown to assemble into piluslike structures. EPEC encodes four proteins, EspA, EspB, EspD, and Tir, that are secreted by a type III pathway. These proteins facilitate intimate contact between the pathogen and host intestinal cells and are required for the formation of specific (attaching and effacing) lesions. Knutton and colleagues (1998) showed that one of these proteins, EspA, can assemble into 7- to 8-nm thick peritrichously expressed pilus-like fibers that are organized into ~50-nm-wide bundles that extend up to 2 _m from the bacterial surface (Fig. 38.1F). These fibers appear to be made up of only EspA molecules. Interestingly, EspA shares substantial sequence identity with a flagellin from Y. enterolitica. During the infection process, the EspA fibers appear to mediate contact between EPEC and the host-cell surface prior to the establishment of more intimate bacterial attachment. The EspA fibers seem to assist the translocation of EspB effector molecules into host cells, where they can subvert host signal-transduction pathways. In P. syringae and other plant pathogens, hypersensitive response and pathogenicity (hrp) genes control the ability of these bacteria to cause disease in susceptible plants and to elicit the hypersensitive response in resistant plants. The hypersensitive response is a phenomenon characterized by rapid localized host-cell death at the site of infection that appears to limit the spread of a pathogen in an infected plant. A subset of the hrp genes, recently renamed hrc genes, encode components of a type III secretion system. In 1997, Roine and co-workers showed that one of the proteins, HrpA, secreted by the Hrp type III secretion system is assembled into 6- to 8-nm-wide, peritrichously expressed pili. It was proposed that these pili, known as Hrp pili, are involved in mediating bacteria–plant interactions in the intercellular spaces of the host plant. In addition, Hrp pili may assist the delivery of effector proteins into host-plant cells. The exact nature and functions of the Hrp pili of P. syringae and the EspA-containing pili of EPEC remain to be elucidated. 

VIII. REGULATION OF PILUS BIOGENESIS 

Pilus biogenesis, in general, is a tightly regulated process. Ideally, the costs in energy and other resources required for pilus assembly must be balanced with any potential benefits that pilus expression might provide a bacterium. For example, by producing pili in a nutritionally poor environment a bacterium will tax its available resources, but with pili the same bacterium may be able to gain access to a more favorable location. Pathogenic and other bacteria must also control pilus expression, in some cases, to avoid attachment to unfavorable sites (tissues) within their hosts. Furthermore, pathogenic bacteria may need to modulate pilus expression to escape detection by the host immune system. Whether a bacterium expresses pili is greatly affected by environmental factors. Changes in temperature, osmolarity, pH, oxygen tension, carbon source, and nutrient availability may either increase or decrease pilus expression. The presence of iron, aliphatic amino acids, and electron acceptors other than oxygen may also influence the expression of pili. A combination of these environmental cues can stimulate (or repress) pilus synthesis and alter the expression of a variety of other factors, all of which can influence the tropism of bacteria for specific niches in the environment or in host organisms. Environmental signals affect pilus biogenesis through global regulator proteins that can modify the transcription of pilus genes. Various global regulators have been identified and include H-NS, a DNAbinding histone-like protein that often mediates temperature regulation of pilus synthesis. H-NS appears to alter DNA topology and typically functions as a negative regulator. Regulation by carbon source can occur through the catabolite activator protein (CAP), whereas the leucine-responsive regulatory protein (Lrp) can modulate pilus expression in response to aliphatic amino acids. The CAP and Lrp regulators can control sets of pilus operons, enabling the expression of different types of pili to be coordinated and integrated with the metabolic state of the bacterial cells. In addition to these and other global regulators, specific regulator proteins encoded by genes within some pilus operons may also modulate pilus biogenesis. Multiple regulatory factors can act upon the same promoter region, switching pilus gene expression from on to off and vice versa. This on-and-off switching, known as phase variation, can also be modulated by the methylation status of a promoter region and by the inversion of sequence elements within a promoter. Two-component systems, such as the Cpx system described for P pilus assembly, also appear to be involved, at least tangentially, in the regulation of the assembly and function of a large number of pilus types. 

IX. ROLE OF PILI IN DISEASE PROCESSES 

The expression of pili can have substantial impact on the establishment and persistence of pathogenic bacteria in their hosts. For many bacterial pathogens, adhesive pili play a key role in the colonization of host tissues. Uropathogenic E. coli, for example, require type 1 pili to effectively colonize the bladder epithelium. These pili attach to conserved, mannosecontaining host receptors expressed by the bladder epithelium and help prevent the bacteria from being washed from the body with the flow of urine. P pili may serve a similar function in the kidneys, inhibiting the clearance of pyelonephritic E. coli from the upper urinary tract. Enteric pathogens produce a wide variety of adhesive pili that facilitate bacterial colonization of the intestinal tract. These include the K88, K99, and 987P pili made by ETEC strains, the long-polar fimbriae (LPF) and plasmid-encoded fimbriae (PEF) of S. enterica, and the aggregative adherence fimbriae (AAF) of enteroaggregative E. coli. In the small intestine, TCP are essential for the attachment of V. cholera to gut epithelial cells. These pili also act as receptors for the cholera toxin phage (CTX_), a lysogenic phage that encodes the two subunits of the cholera toxin. This phage, with its encoded toxin, is transferred between V. cholera strains via interactions with TCP in the small intestine. Other pili also function in the acquisition of virulence factors. The uptake of DNA facilitated by type 4 pili and DNA transfer directed by conjugative pili can provide pathogens with accessory genes enabling them to synthesize a wider repertoire of virulence factors and giving them resistance to a greater number of antibiotics. 

Biofilm formation, which in some cases appears to require pili such as type 1, type 4, or curli, can also increase the resistance of bacteria to antibiotic treatments and may aid bacterial colonization of tissues and medical implants. Pili are not necessarily static organelles and dynamic alterations of pilus structures during the infection process may influence the pathogenicity of piliated bacteria. For example, electron-microscopic studies (Mulvey, 1998) of mouse bladders infected with type 1-piliated uropathogenic E. coli showed that the pili mediating bacterial adherence to the bladder epithelial cells were 10–20 times shorter than typical type 1 pili. It is possible that the shorter type 1 pili observed are the result of pilus retraction, or breakage, during the infection process. The shortening of type 1 pili may provide a means for reeling bacteria in toward their target host cells, allowing the bacteria to make intimate contact with the bladder epithelium after the initial attachment at a distance. Within the gut, type 4B BFP promote the autoaggregation of EPEC strains, a phenomenon that probably facilitates the initial adherence of EPEC to the intestinal epithelium. Work by Bieber and co-workers (1998) suggests that, after initial attachment, an energy-dependent conformational change in the quaternary structure of BFP is needed for the further dispersal of EPEC over human intestinal cells and for the full virulence of this pathogen. During the infection process, adhesive pili are often situated at the interface between host and pathogen where they can potentially mediate cross-talk between the two organisms. A few examples of pilus attachment inducing signal-transduction pathways in host eukaryotic cells have been reported. The binding of the type 4A pili of Neisseria to host receptors (probably CD46) on target epithelial cells has been shown to stimulate the release of intracellular Ca2_ stores, a signal known to control a multitude of eukaryotic cellular responses. Similarly, the attachment of P pili to Gal_ (1-4)Gal-containing host receptors on target uroepithelial cells can trigger the intracellular release of ceramides, important second-messenger molecules that are capable of activating a variety of protein kinases and phosphatases involved in signal transduction processes. The signals induced in uroepithelial cells upon the binding of P-piliated bacteria eventually result in the secretion of several immunoregulatory cytokines. The binding of type-1-piliated bacteria to mannosylated receptors on uroepithelial cells can similarly induce the release of cytokines, although apparently through different signaling pathways than those stimulated by P pilus binding. Pili can also transduce signals into bacterial cells. This was demonstrated by Zhang and Normark who, in 1996, showed that the binding of P pili to host receptors stimulated the activation of iron-acquisition machinery in uropathogenic E. coli. This probably increases the ability of uropathogens to obtain iron and survive in the iron-poor environment of the urinary tract. An understanding of how pili can transmit signals into bacterial cells, and the consequences of such signaling, awaits future studies. Continued research into the biogenesis, structure, and function of pili promises not only to advance our basic understanding of the role of these organelles in pathogenic processes, but may also aid the development of a new generation of antimicrobial therapeutics and vaccines.