Bacteria assemble adhesins on their surface as monomers, simple oligomers, or components of supramolecular fibers called fimbriae or pili (Hultgren et al., 1996). The adhesive organelles most commonly associated with UPEC include type 1, P, and S/F1C-related pili and the Dr family of adhesins (Johnson, 1991) (Table 1). The assembly of these adhesins, and of a large number of other adhesive organelles expressed by E. coli and other species, is mediated by highly conserved periplasmic chaperones and outer membrane usher proteins. The chaperone-usher pathway has been reviewed in detail elsewhere (Hung and Hultgren, 1998; Sauer et al., 2000). Here, the structures and functions of the predominant adhesive organelles associated with UPEC are considered.
Type 1 pili
Of the various adhesins encoded by UPEC, type 1 pili are by far the most prevalent (Brinton, 1959; Buchanan et al., 1985; O´Hanley et al., 1985; Langermann et al., 1997). These organelles are composite fibers varying from a few fractions of a micron to greater than several microns in length (Jones et al., 1995). They consist of a 7 nm thick helical rod, made up of repeating FimA subunits, coupled to a short 3 nm wide tip fibrillum structure containing the adhesin FimH and two adaptor proteins, FimF and FimG (Russell and Orndorff, 1992; Jones et al., 1995). The FimH adhesin comprises a C-terminal pilin domain involved in the incorporation of FimH into the tip fibrillum and an N-terminal adhesin domain (Fig. 1A) (Choudhury et al., 1999). A carbohydrate-binding pocket localized at the distal tip of the adhesin domain can mediate bacte-rial interactions with a variety of mannose-containing glycoprotein receptors expressed by a number of different host cell types (Ofek et al., 1977; Tewari et al., 1993; Baorto et al., 1997; Mulvey et al., 1998; Pak et al., 2001; Zhou et al., 2001). The adhesin domain is an 11-stranded elongated β barrel with an overall jellyroll topology (Choudhury et al., 1999). This domain is connected to the pilin domain via an extended linker that may provide optimal flexibility for correct positioning of the mannose-binding site relative to host receptors.
Figure 1. .Structures of the FimH and PapGII adhesin domains.
A. Ribbon model of the adhesin domain of FimH, consisting of an 11-stranded β barrel with an interrupted jellyroll-like motif. This domain is ~50 Å in length. The tip-localized mannose-binding pocket of FimH is indicated by an asterisk.
B. Ribbon model of the binary complex of the PapGII adhesin domain bound to the tetrasaccharide residues of the GbO4 glycolipid. The PapGII adhesin domain is ~80 Å in length and is only vaguely similar in structure to the corresponding FimH domain. The scale in (A) and (B) is not equivalent.
Download figure to PowerPoint
In addition to UPEC, many other pathogenic and commensal E. coli isolates and other members of the enterobactericeae family encode type 1 pili. Over the last several years, a number of natural and engineered allelic variants of FimH, which differ by as little as a single amino acid, have been identified that can confer distinct adhesive phenotypes to type 1 pili. Amino acid changes, both within and outside the carbohydrate-binding pocket of FimH, in addition to alterations in the composition of the type 1 pilus rods, have been found to influence the binding characteristics of FimH (Klemm et al., 1994; Sokurenko et al., 1994; Sokurenko et al., 1995; Sokurenko et al., 1997; Sokurenko et al., 1998; Pouttu et al., 1999; Hamrick et al., 2000a; Schembri et al., 2000; Schembri and Klemm, 2001a; Harris et al., 2001; Sokurenko et al., 2001; Hung et al., 2002). All natural FimH variants can bind trimannose structures, but can differ in their abilities to bind a number of other carbohydrate and non-carbohydrate receptors, including monomannose residues, type I and type IV collagen, laminin, and fibronectin (Kukkonen et al., 1993; Sokurenko et al., 1994; 1995; 1998; Pouttu et al., 1999).
About 80% of commensal E. coli faecal isolates encode FimH adhesins that bind only trimannose receptors, while about 70% of UPEC isolates express FimH variants with mutations that enhance their affinity for monomannose residues in addition to trimannose receptors (Sokurenko et al., 1995). The conversion of FimH to forms with higher affinity for monomannose residues often involves mutations within the bottom part of the adhesin domain that are likely to alter the conformation and stability of the protein loops that carry the receptor-interacting residues (Schembri et al., 2000). The monomannose-binding phenotype prevalent among UPEC isolates confers a higher tropism for glycoprotein receptors expressed by uroepithelial cells and enhances bacterial colonization of the urinary tract (Sokurenko et al., 1998). FimH-mediated bacterial interactions with uroepithelial cells are critical to the ability of UPEC to colonize the bladder and cause disease (Connell et al., 1996; Thankavel et al., 1997; Langermann et al., 1997; Mulvey et al., 1998). The primary host receptor for type 1 pili within the bladder appears to be an integral membrane glycoprotein known as uroplakin 1a (UP1a) (Zhou et al., 2001). This protein, together with three other partners (UP1b, UPII and UPIII), are assembled into 16 nm diameter hexagonal complexes that are further organized into plaques 0.3–0.5 μm in diameter (Sun et al., 1996). These plaques cover almost the entire lumenal surface of the bladder and are thought to function as part of a permeability barrier and may help strengthen and stabilize the bladder epithelial cells. Expression of the uroplakin plaques by the superficial epithelial cells that line the bladder lumen make these cells a primary target for UPEC upon entering the urinary tract (Mulvey et al., 1998). Bacterial interactions with a sparse glycolcalyx that overlays the bladder surface may also facilitate bacterial colonization of the urinary tract (Hopkins et al., 1990).
While monomannose binding by FimH enhances bacterial interactions with uroplakin receptors and facilitates colonization of the lower urinary tract, it may also impair the fitness of UPEC at other sites within the host. For example, type 1 pili appear to assist bacterial transmission between hosts by promoting transient colonization of the oropharyngeal cavity, the initial portal of entry for UPEC and other E. coli isolates (Bloch et al., 1992). Here, bacteria encounter a number of soluble inhibitors that interact more effectively with monomannose-binding FimH variants relative to FimH forms that bind only trimannose residues (Sokurenko et al., 1998). Therefore, at this location, expression of FimH variants having monomannose-binding affinity may be a disadvantage to the survival and transmission of UPEC. Even within the urinary tract, the expression of monomannose-binding FimH variants by UPEC can potentially complicate bacterial colonization by increasing the affinity of FimH for soluble glycoprotein receptors within the urine. One of the most abundant urinary proteins, the Tamm– Horsfall protein, has recently been shown to preferentially bind E. coli strains expressing monomannose-binding variants of FimH and so can prevent type 1 pilusmediated bacterial adherence to uroepithelial cells (Pak et al., 2001).
In addition to promoting bacterial–host interactions, recent studies have demonstrated that some FimH variants can also mediate interbacterial contacts, stimulating bacterial autoaggregation and biofilm formation (Pratt and Kolter, 1998; Schembri and Klemm, 2001a; Schembri et al., 2001). The role of FimH in these processes is not yet clear, but they do not seem to necessarily depend on the mannose-binding capacity of the adhesin. FimH-mediated autoaggregation and biofilm formation may enable UPEC to better withstand antibiotic treatments and host antibacterial defences within the urinary tract. In addition, type 1 pilus-mediated biofilm formation may facilitate bacterial colonization of urinary catheters and other medical implants, an unfortunately common problem for hospitalized individuals.
Similar to type 1 pili, P pili are also composite organelles consisting of a short, flexible tip fibrillum attached to the distal end of a thicker rod structure (Kuehn et al., 1992; Bullitt and Makowski, 1995). The P pilus rod is 6.8 nm wide and is composed of repeating PapA subunits arranged in a right-handed helical cylinder. A subunit designated PapH is thought to anchor the PapA rod to the outer membrane (Baga et al., 1987). The tip fibrillum, which is about 2 nm thick, contains a distally located adhesin, PapG, in association with three other subunits, PapE, PapF, and PapK (Kuehn et al., 1992; Jacob-Dubuisson et al., 1993). The PapG adhesin recognizes glycolipid receptors expressed by erythrocytes and host cells present in the kidney (Leffler and Svanborg-Eden, 1980; Lund et al., 1987). Studies have indicated that P pili, and specifically the PapG adhesin, are significant virulence factors associated with pyelonephritis. Roberts and colleagues found that PapG, while unnecessary for UPEC colonization of the bladder, is essential for a pyelopnephritic UPEC isolate to adhere to renal tissue and cause pyelonephritis in cynomogus monkeys (Roberts et al., 1994). More recently, Wullt et al. (2000) demonstrated using human volunteers that P pili enhance the colonization of the urinary tract and facilitate the establishment of bacteriurea by an UPEC isolate known to cause asymptomatic bacteriurea. Interestingly, epidemiological studies indicate that while P pili are important factors in initiating pyelonephritis in normal urinary tracts, these adhesive organelles seem to have a less significant role in colonizing urinary tracts with abnormalities or obstructions (Jantunen et al., 2000; Tseng et al., 2001). Outside the urinary tract, a clear-cut role for P pilus expression has not been established. However, Alpers and coworkers have recently demonstrated that the PapG adhesin can bind surfactant-like particles that are secreted by human and mouse intestines (Goetz et al., 1999; Mahmood et al., 2000). PapG-mediated interactions with these particles, it is proposed, may enable UPEC to establish a reservoir within the intestinal mucosa and could facilitate UPEC persistence among the intestinal flora.
The minimal glycolipid receptor for the PapG adhesin is called globotriasylceramide (GbO3) and consists of a digalactoside (Galα1–4Gal) core linked by a β-glucose (Glc) residue to a ceramide group that anchors the receptor in the membrane (Stromberg et al., 1990; 1991). Alteration of the Galα1–4Gal core of GbO3 by the addition of a single N-acetyl-galactosamine (GalNAc) generates GbO4 (globoside), while modification with two GalNAc sugars creates GbO5 (the Forsmann antigen). Three distinct PapG variants, designated GI, GII and GIII, have been identified that recognize GbO3, GbO4 and GbO5 respectively. It has been suggested that the dif-ferent PapG variants affect the host specificity of pyelonephritic UPEC strains, but recent epidemiological studies have raised doubts regarding this possibility (Feria et al., 2001).
Recently, both the crystal and solution structures of the PapGII adhesin domain were reported (Dodson et al., 2001; Sung et al., 2001). This domain, which shares little sequence homology with the adhesin domain of FimH, is a mostly β-sheet structure consisting of an open and elongated β-sandwich having long loops and a short helical section connecting one set of the loops (Fig. 1B). Co- crystallization of PapGII with GbO4, along with mutation studies, revealed that the GbO4 receptor binds along the side of PapG via a rigid body type interaction that does not involve conformational alterations in the adhesin. Side-on binding of PapG with its receptor may promote optimal interactions with oligosaccharride receptors that are proposed to be kinked about the lipid moiety of GbO4 (Gronberg et al., 1994;Sung et al., 2001). In addition, the tips of P pili appear to be fairly flexible and this flexibility may enhance side–on interactions between PapG and glycolipid receptors embedded within host cell membranes (Kuehn et al., 1992).
S pili have a similar, although less well-defined, architecture to type 1 and P pili (Schmoll et al., 1989). S pilus fibers are composed of a major subunit SfaA in addition to three minor subunits SfaG, SfaH and SfaS. The SfaS subunit has been localized to S pilus tips and can mediate bacterial interactions with sialic acid residues on receptors expressed by kidney epithelial and vascular endothelial cells (Korhonen et al., 1986; Moch et al., 1987; Morschhauser et al., 1990; Hanisch et al., 1993). The major subunit SfaA also has adhesin characteristics and can mediate bacterial adherence to endothelial cell glycolipids and plasminogen (Parkkinen et al., 1991; Prasadarao et al., 1993). Minor subunits, in addition to SfaS, may also modulate the binding properties of S pili (Schmoll et al., 1989; Morschhauser et al., 1993). S pili may facilitate bacterial dissemination within host tissues and are often associated with E. coli strains that cause sepsis, meningitis, and ascending UTIs, including pyelonephritis (Korhonen et al., 1985; Marre et al., 1986; Parkkinen et al., 1988; Hacker et al., 1993). Recent work showing that sialic acid residues are presented on UP3, one of four integral membrane uroplakin proteins abundantly expressed on the lumenal surface of the bladder, suggests that S pili may also have a role in cystitis (Malagolini et al., 2000). A number of adhesive organelles have been identified which are genetically homologous to S pili, but differ in their receptor specificity (Ott et al., 1988; Hacker et al., 1993). Among these homologous structures are F1C pili which can bind β-GalNac-1, 4β-Gal residues on glycolipids expressed by epithelial cells of the distal tubules and collecting ducts of the kidney and also by bladder and kidney endothelial cells (Khan et al., 2000). F1C pili are encoded by approximately 14% of UPEC isolates. This observation, along with the binding specificity of these organelles, indicates that F1C pili may impact the pathogenesis of a significant number of UTI cases.
Dr adhesin family
This group of adhesins includes the uropathogen- associated fimbrial adhesin Dr and the non-fimbrial adhesins AFA-I, AFA-II, AFA-III, AFA-IV, Nfa-I and Dr-II (reviewed in Nowicki et al., 2001). Dr adhesin family members are proposed to facilitate ascending colonization and chronic interstitial infection of the urinary tract. Epidemiological studies indicate that about half of all children with UTIs, and over 30% of pregnant women with pyelonephritis, are colonized by UPEC strains expressing the Dr adhesin. In addition, infection with E. coli strains expressing Dr results in a twofold increase in the risk for a recurrent UTI. The contribution of Dr as a virulence factor able to promote bacterial persistence within the urinary tract can be substantial. Recently, it was reported that UPEC encoding the Dr adhesin, but not Dr– bacteria, could survive for more than 1 year within renal tissue (Goluszko et al., 1997b; Nowicki et al., 2001). Members of the Dr adhesin family recognize one or more of four 60-amino-acid short consensus repeat (SCR) sequences present in decay accelerating factor (DAF, CD55), a glycosylphosphatidylinositol (GPI)-linked complement receptor and regulatory factor, expressed on erythrocytes and other tissues including the uroepithelium (Nowicki et al., 2001). Different members of the Dr adhesin family appear to recognize distinct SCR sequences within DAF. Increases in DAF expression within the urinary tract during pregnancy may be responsible for the increased susceptibility of pregnant women to infection with Dr+ UPEC (Goluszko et al., 2001). Dr adhesin family members can also bind carcinoembryonic antigen (CD66e), a GPI-anchored protein with unknown functions (Guignot et al., 2000). In addition, the Dr adhesin, but not other members of this family, can recognize type IV collagen (Nowicki et al., 2001). By interacting with DAF and possibly other receptors such as type IV collagen and CD66e, Dr can promote bacterial adherence to the interstitial compartments of the kidney and so may facilitate long-term bacterial persistence within the upper urinary tract.