Adhesins of Enteropathogenic Escherichia coli

ALFREDO G. TORRES

[SECTION EDITOR: JAMES P. NATARO]

Posted February 28, 2006

Departments of Microbiology and Immunology and Pathology and The Sealy Center for Vaccine Development, University of Texas Medical Branch, Galveston, TX 77555-1070

*Phone: (409) 747-0189, Fax: (409) 747-6869, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

Enteropathogenic Escherichia coli (EPEC) strains induce morphological changes in infected epithelial cells. The resulting attaching and effacing (A/E) lesion is characterized by intimate bacterial adherence to epithelial cells, with microvillus destruction, cytoskeletal rearrangement, and aggregation of host cytoskeletal proteins. This review presents an overview of the adhesion mechanisms used for the colonization of the human gastrointestinal tract by EPEC. The mechanisms underlying EPEC adhesion, prior to and during the formation of the A/E lesion, and the host cytosolic responses to bacterial infection leading to diarrheal disease are discussed.

INTRODUCTION

EPEC strains are implicated in diarrhea in humans and are a leading cause of infantile diarrhea in developing countries. In industrialized countries, the frequency of these organisms has decreased, but they continue to be an important cause of diarrhea (103, 109). EPEC, the first pathogenic E. coli category to be associated with diarrhea, inhabits the small intestine and infects predominantly the proximal small intestine, either in the duodenum or the proximal jejunum, although there may be some colonic involvement due to overgrowth of the organism from the small intestine (120). The EPEC serotypes which are part of this E. coli pathotype are distinct from other pathogenic E. coli strains because they produce a distinct histopathological lesion on intestinal epithelial cells known as the A/E lesion (see below). A/E lesions are characterized by microvillus destruction, intimate adherence of bacteria to the intestinal epithelium, assembly of highly organized pedestal-like actin structures in the epithelial cells, and aggregation of polarized actin and other elements of the cytoskeleton at sites of bacterial attachment (reviewed in references 43, 74, and 103). The fluorescent actin staining test allows the identification of strains that produce A/E lesions through detection of aggregated actin filaments beneath the attached bacteria (Fig. 1A) (85). The ability to produce A/E lesions has also been detected in strains of enterohemorrhagic E. coli (EHEC) (Chapter Adhesins of Enterohemorrhagic Escherichia coli), Citrobacter rodentium, and other bacterial species (103).

Fig. 1Staining of EPEC-infected cells and EPEC adherence patterns. (A) Fluorescent actin staining assay of HEp-2 cells infected with EPEC strain E2348/69. After 3 h of infection, cells were fixed, permeabilized, and colabeled with DAPI (4',6'-diamidino-2-phenylindole; eukaryotic and prokaryotic DNA staining) and AlexaFluor 488-phalloidin (actin staining). (B through D) Adherence patterns of EPEC strains: LA (B), DA (C), and AA (D). Magnification, ×100.

Typical and Atypical EPEC

EPEC is defined as diarrheagenic E. coli that produces A/E lesions on intestinal cells and, in contrast to EHEC (Chapter Enterohemorrhagic Escherichia coli) and other Shiga toxin-producing E. coli, does not produce Shiga toxins (reviewed in references 73 and 103). EPEC can be further classified into typical and atypical strains (150), which constitute two distinct groups of organisms that have in common the locus of enterocyte effacement (LEE; see below), a pathogenicity island (PAI) that promotes the development of A/E lesions (103, 150). Typical EPEC strains of human origin possess a virulence plasmid known as the EAF (EPEC adherence factor) plasmid that confers a localized adherence (LA) pattern on EPEC infecting cultured epithelial cells (7, 124), but atypical EPEC strains do not possess this plasmid. The EAF plasmid carries the genes required for the production of the bundle-forming pili (BFP; see below), which have been definitely involved in interbacterial adherence in the LA pattern (30, 141, 142).

In addition to differing in the presence or absence of the EAF plasmid, typical and atypical EPEC strains belong to different serotypes (150). Members of eleven serotypes are frequently considered to be typical EPEC strains (O55:H6, O55:NM, O86:H34, O111:H2, O111:NM, O114:H2, O119:H6, O119:NM, O127:H6, O142:H6, and O142:H34), and 14 serotypes correspond to atypical isolates (O26:H11, O26:NM, O55:H7, O55:NM, O55:H34, O86:H8, O111ac:H8, O111ac:NM, O111:H9, O111:NM, O111:H25, O119:H2, O125ac:H6, and O128:H2). One feature of some of these serotypes that may complicate the distinction between typical and atypical EPEC strains is the presence of EAF plasmid markers (150). For example, some strains react with a DNA probe directed to BFP genes but do not have a true EAF plasmid and other strains contain a large plasmid that does not contain the BFP operon (16). Therefore, in addition to the EAF plasmid, another characteristic to allow us to distinguish between typical and atypical EPEC serotypes is BFP production (150).

Typical and atypical EPEC strains also differ in their adherence patterns. The typical strains show only the LA pattern (Fig. 1B), but atypical strains may show a localized adherence-like (LAL) (51), diffuse adherence (DA) (Fig. 1C), or aggregative adherence (AA) (Fig. 1D) pattern. The sections following describe the various adherence factors produced by typical and atypical strains associated with the different adherence patterns and the distinctions within the intimin types, whose expression by the different serotypes has been linked to tissue tropism in the host (39, 112).

LEE

The proteins associated with the formation of the A/E lesion are encoded on a chromosomal PAI known as the locus of enterocyte effacement (LEE) (reviewed in reference 147). The first description of a gene in the LEE PAI associated with intimate adherence to epithelial cells was reported by Jerse et al. (68), who screened TnphoA mutants of EPEC for the loss of the A/E phenotype. The chromosomal gene identified in this study, eae, encodes intimin, an outer membrane protein needed for intimate adherence that is recognized by sera from volunteers convalescing from experimental EPEC infections (67, 68). Further analysis of additional mutations in this region which produced mutants that were negative for the A/E phenotype led McDaniel et al. (95) to identify a 35-kb locus containing several regions implicated in the formation of A/E lesions. The LEE is present in all EPEC strains as well as in other A/E phenotype-positive strains, including strains of EHEC (Chapter Adhesins of Enterohemorrhagic Escherichia coli) and other enteric pathogens (26, 95, 96, 165).

Structure

The complete sequence of the LEE from EPEC prototype strain E2348/69 has been determined (38). The EPEC LEE is 35,624 bp in length, with an average G+C content of 38.36%, and contains 41 predicted open reading frames (ORFs) corresponding to sequences of >50 amino acids arranged in at least five polycistronic operons (Fig. 2). For functional purposes, the EPEC LEE genes are divided into three regions. The middle region contains the eae, cesT, and tir genes. The eae gene encodes the intimin protein that mediates intimate attachment to the host cell (66, 67). Its receptor in the host cell is a bacterial protein named Tir (translocated intimin receptor) (79, 119). The Tir protein is translocated into the host cell via a type III secretion system (TTSS) and is implicated in several processes: it acts as a receptor for intimin, linking the bacterium to the host cell cytoskeleton, and functions in other signaling events within the host cell (Fig. 3) (17, 79). The cesT gene encodes the protein CesT, which functions as a bivalent chaperone required for the translocation of both Tir and Map (1, 21, 34).

Fig. 2Schematic representation of known and putative virulence factors associated with adhesion, which are encoded on the chromosome or carried by the EAF plasmid of EPEC strains. The genetic organization of the chromosomally located LEE PAI and that of the BFP cluster are depicted. The LEE is organized into five main operons (LEE1 through LEE5) and contains 41 genes, and since the function of almost all the genes is known, the genes can be grouped on the basis of function (color coded). The BFP operon is located on the EAF plasmid and contains 14 genes associated with the biogenesis of the fimbriae. The BFP proteins are grouped on the basis of their locations within the EPEC membranes or the bacterial compartments. Also depicted are chromosomally, non-LEE-encoded factors that have been associated with adherence, such as those encoding flagella, LifA, and LPF. ERIC, enterobacterial repetitive intergenic consensus.

Fig. 3Model of early and later events in EPEC adhesion to host cells. (A) The early events in EPEC adhesion are poorly understood but begin with the expression of BFP, the intimate adhesin intimin,surface-associated EspA filaments, and flagella and may involve binding of intimin to a host cell receptor, such as nucleolin, and binding of the EspA filament directly to the cell or via secreted EspB and EspD proteins. EPEC cells adheringto epithelial cells will use the TTSS to inject the translocated intimin receptor(Tir) and a number of effector moleculesdirectly into the host cell. The role of the other proposed adhesins (BFP and flagella) has not been clearly elucidated. (B) EPEC cells translocate LEE-encoded and other effector proteins (Cif and EspI; Chapter Enteropathogenic Escherichia coli ) into the host cell cytosol. Once Tir is inserted into the membrane, it dimerizes and serves as the receptor for intimin. However, EPEC intimin has been shown to interact also with several known and putative cell surface proteins. EPEC Tir is phosphorylated, which allows binding of Nck and recruitment and activation of neuronal-Wiskott-Aldrich Syndrome Protein (N-WASP) and the Arp2/3 complex, essential to promote the reorganization of the actin cytoskeleton underneath adherent EPEC bacteria. The effector proteins trigger cytoskeleton rearrangements (Tir, EspH, EspF, and EspG/EspG2), disruption of the epithelial barrier (EspF and Map), cytotoxicity (EspF, Map, and Cif), and host cell responses that ultimately generate watery diarrhea. The present model of EPEC adhesion was created by compiling information from recent review articles (25, 46, 108).

The second region is located upstream of the eae-tir domain and encodes a TTSS, the type III secreted effector proteins, and several regulators of the LEE PAI. EPEC strains use the TTSS (EscR, EscS, EscT, EscU, EscC, EscJ, EscV, EscN, and SepQ) to translocate effector proteins into the infected cells and thereby modify specific host functions (Fig. 3) (46, 63). This domain also encodes the type III secreted proteins Map, EspH, SepZ (recently renamed EspZ), EspG, and EspG2 (previously called CesAB). Map is an effector protein targeted to the host mitochondria (80) and has a dual role, mediating Cdc42-dependent filopodium formation and targeting mitochondria to elicit dysfunction (65). EspH is a modulator of the host actin cytoskeleton structure (151), playing a role in brush border remodeling and production of mature A/E lesions (128). SepZ was originally reported to affect secretion and tyrosine phosphorylation activity induced by EPEC (29, 115) but was recently shown to be a translocated protein whose specific function is not known (72). EspG shares homology with Shigella flexneri protein VirA (36) and acts on microtubules to disrupt the microtubule networks of intestinal epithelial cells and perturb the tight-junction barrier (94, 130, 144). The EspG2 protein, which was originally described as the dual CesAB chaperone for the type III translocator proteins EspA and EspB (22), has been recently found to be an effector protein which targets the microtubule network in a manner similar to EspG (130, 144). Three regulators are also located in this domain, including the ler gene encoding the LEE-encoded regulator (Ler). Ler activates the transcription of several LEE genes and is required for the expression of LEE-encoded proteins (37, 98). GrlR and GrlA are positive and negative regulators, respectively, required for the expression of several LEE-encoded genes and identified during a genetic screen of the C. rodentium LEE PAI (27) (Chapter Virulence Gene Regulation in Escherichia coli). Although the regulatory role of these proteins in EPEC LEE expression has not been directly reported, recent data indicate that GrlR is involved in the regulation of LEE-encoded genes in EHEC and EPEC strains (61).

The third region, located downstream of the eae-tir domain, encodes the type III secreted proteins EspA, EspB, EspD, and EspF, components of the TTSS (EscD, EscF, and SepL), and CesD2, a second chaperone for the type III secreted protein EspD (104). The EspA protein forms a filamentous structure on the surfaces of EPEC strains that is involved in protein translocation into epithelial cells (Fig. 3A, type III secretion translocon) (86, 127, 129). EspB protein is produced and secreted upon contact with tissue culture cells and has been localized both in the cytoplasm and on the membrane of the host cell (140, 158). EspD is a protein that is inserted into epithelial cell membranes (88, 155), and together, EspB and EspD form a translocation pore in the host cell membrane at the distal end of the EspA filament (Fig. 3A) (59, 155, 158). EspF is an effector protein that was first described in association with the disruption of the host intestinal barrier functions (97) and is important in the disruption of tight-junction architecture by EPEC (101). More recently, its role in initiating the mitochondrial death pathway has also been elucidated (102, 107).

Intimin and Tir

It has been clearly established that intimin mediates the intimate adherence to target eukaryotic cells upon interaction with Tir and that, upon this interaction, host cell signaling events are triggered that lead to cytoskeletal protein accumulation beneath the adherent bacteria ( Fig. 3B) (reviewed in references 17 and 78). Further, the concept of how EPEC strains causing A/E lesions adhere to mammalian cells has been enhanced by the description of the crystal structure of intimin coupled to the Tir receptor (92). The current model indicates that, once translocated, the Tir protein spans the host cell membrane, adopting a hairpin loop structure featuring both its N and C termini in the host cytoplasm and a central extracellular domain that binds intimin (9, 91, 92). In addition to serving as a receptor for intimin, the Tir protein is capable of interacting with host cytoskeletal and signaling components by using its N- and C-terminal domains located in the host cell cytoplasm (Fig. 3B). The numerous host proteins that accumulate in the A/E lesion around Tir and the mechanisms by which Tir exploits the host cell signaling networks resulting in actin cytoskeletal rearrangements are reviewed extensively in the chapter dedicated to EPEC (Chapter Enteropathogenic Escherichia coli).

EPEC intimin is a 94-kDa outer membrane protein (939 amino acids) that can be subdivided into a flexible N-terminal region (including a periplasmic domain, residues 40 to 188), a central membrane-integral β barrel (residues 189 to 549), and a surface-exposed C-terminal region (including four extracellular domains named D0 to D3, residues 550 to 939), where the receptor binding activity resides (53, 92). Originally, the receptor binding region was mapped to a region comprising the C-terminal 280 amino acids of intimin (77). Further crystallographic studies indicated that the last 190 amino acids within this region mediate this binding. The C-terminal 190 amino acids of intimin comprise an N-terminal immunoglobulin domain that is coupled to a novel C-type lectin domain, which in turn forms a flexible rod that extends from the bacterial surface and serves as a tip that binds the target cell (9, 90, 92). Furthermore, a direct link between intimin binding and Tir clustering has been also demonstrated (149). The association of intimin with Tir triggers a host cell response leading to pedestal formation (Fig. 3B). Although the formation of actin pedestals by EPEC is a phenotype best characterized in vitro in epithelial cell lines, pedestal formation correlates with the ability of the A/E organisms to colonize the intestine and cause disease in human and other animal hosts (reviewed in reference 103).

Alternative Intimin Receptors

In addition to the interaction of intimin with its primary receptor, Tir, and the subsequent triggering of host cellular responses, there is evidence indicating that intimin promotes initial adherence by binding to endogenous host cell receptors independently of its interaction with Tir. An initial study reported that intimin from EPEC can bind β1-chain integrins in vitro (41), but a subsequent study challenged these results and indicated that these host cell proteins are not essential for intimin-mediated cell attachment or EPEC-mediated actin polymerization (90). Although the location of β1-integrins in the basolateral membranes of intestinal epithelial cells would seem to preclude interaction with EPEC on the apical membrane, a recent study showed that EPEC induces alterations in tight junctions, resulting in redistribution of β1-integrin and other basolateral membrane proteins to the apical cell surface, thereby providing them with an opportunity to interact with EPEC intimin (100). Despite these discrepant results, additional indirect evidence favors the concept that intimin interacts with an additional eukaryotic cell receptor. Such evidence includes intimin-mediated tissue tropism (see below), sequence divergence in the region corresponding to the intimin domain mediating receptor binding, and the results of mutagenesis analysis of this region indicating that the domain mediating intimin binding to Tir is separate from the domain mediating adherence to host cells (42). Moreover, the fact that purified intimin displays biological activity provides additional evidence indicating that intimin binds directly to host cells, inducing colonic hyperplasia in mice (58) and activating T cells (52).

Using EHEC as a model system (Chapter Adhesins of Enterohemorrhagic Escherichia coli), Sinclair and O’Brien (131) identified a eukaryotic receptor for intimin on HEp-2 cells. These investigators isolated a protein from HEp-2 cell extracts that bound to EHEC intimin, and amino acid sequence analysis identified the protein as nucleolin. This protein colocalizes to areas on the HEp-2 cell surface where intimin is present, and antibodies directed against nucleolin reduce binding of EHEC to HEp-2 cells when added prior to or at the time of bacterial infection (131). The same investigators have also shown that intimin α and β (see the discussion of nomenclature in the following section) bind nucleolin expressed on the surfaces of HEp-2 cells with an affinity similar to that of EHEC intimin (intimin γ) (132). The cumulative data indicate that nucleolin can serve as a host receptor for EPEC and EHEC intimin and that the nucleolin-intimin interaction occurs early in the infection process. While intimin binding to either β1-integrins (41), nucleolin (131), or another unknown receptor (25) can mediate Tir-independent binding, the interaction with host cell surfaces can induce the formation of microvillus-like projections (113) and have an effect on tight-junction permeability (Fig. 3B).

Intimin-Mediated Tissue Tropism

EPEC intimin has been associated with the pattern of colonization and tissue tropism of the host (112), and the correlation between the expression of different types of intimin and tissue tropism has been suggested by recent data indicating that differences in the amino acid sequences of the intimin proteins influence the patterns of colonization in the host (39, 112). These observations have been demonstrated experimentally using in vitro human intestinal organ cultures (IVOC), and it was found that EPEC intimin confers specificity for the human proximal and distal small intestine as well as the follicle-associated epithelium of Peyer’s patches (FAE) (112). Furthermore, it has been shown that intimin is required for colonization by EPEC and pathogenesis of EPEC infection in humans (32). Intimin proteins of EHEC (Chapter Adhesins of Enterohemorrhagic Escherichia coli) and EPEC strains show a high level of conservation in the N-terminal regions and variability in the last C-terminal 280 amino acids of the proteins. The antigenic polymorphism in the C-terminal region, where binding to enterocytes and Tir occurs, has led to the classification of EPEC and EHEC intimins into at least 10 types (reviewed in reference 148), and typical and atypical EPEC strains express different intimin types (Table 1).

Table 1Types of intimins of typical and atypical EPEC strains, their origins, and associated tissue tropism.

The initial studies using genetic and immunological approaches provided evidence for the existence of at least four distinct intimin types in EPEC and EHEC strains known as intimin α, β, γ, and δ (2, 3). A subsequent study revealed the presence of a fifth type, intimin ε, in EHEC strains of serogroup O103 (110). It has also been proposed that intimins α, β, and γ can be further subdivided based on restriction analysis into α1, α2, β1, β2, γ1, and γ2 (β2 is identical to δ) (110). Recently, other independent studies have proposed that additional intimin types classified based only on subtle differences at the nucleotide level can be found in other strains of EPEC or EHEC. These new groups have been classified as intimins ζ, η, θ, ι, κ, λ, μ, ν, and ξ (13, 14, 70, 137, 163).

Intimin α is expressed by typical strains from the clonal group designated EPEC 1, as well as serotypes O142:H6 and O142:H34, and by atypical EPEC strains of serotypes O111:H9 (motile and nonmotile) and O125ac:H6 (Table 1). Its expression has been shown to confer specificity for the human proximal and distal small intestine as well as for the FAE (114). Intimin β is associated with human and animal EPEC 2 isolates, typical EPEC strains of serotypes O114:H2, O119:H6 (motile and nonmotile), O128:H2, and O126:H2, and atypical EPEC strains of serotypes O26:H11 (motile and nonmotile), O119:H2, and O91:H7. By using a rabbit diarrheagenic E. coli (RDEC-1) strain, it has been shown that intimin β expression correlates with colonization of the rabbit Peyer’s patch lymphoid follicles (18). As for intimin γ, its expression correlates with colonization of the human ileal FAE (39, 112) and it is found in typical EPEC strains of serotypes O119:H6 (motile and nonmotile) and O55:H⁻ and in atypical EPEC strains of serotypes O55:H7 (motile and nonmotile), O111ac:H8 (motile and nonmotile), O55:H34, O101:H⁻, and O116:H33. Intimin δ has been found in typical EPEC O86:H34 and atypical EPEC O88:H5, but the tissue specificity of this intimin has not been determined. Although intimin ε is expressed by human and animal EHEC strains other than those of serogroup O157 (Chapter Adhesins of Enterohemorrhagic Escherichia coli) and atypical EPEC O103:H⁻, the tissue distribution of these strains is similar to that observed with intimin γ of EHEC O157:H7 (40). Finally, some intimin alleles not frequently found in human isolates have been described in atypical EPEC strains (intimin ζ in O111:H9, intimin θ in O156:H8, and an untypeable intimin in O127:H40). The host and tissue distributions of EPEC and EHEC are probably multifactorial, but characterization of the different intimin types may yield important information regarding tissue tropism.

ADHESINS OF TYPICAL EPEC STRAINS

LA and BFP

Typical EPEC strains produce a characteristic adherence pattern, called LA, in tissue culture cells (124, 154). In this pattern, bacteria bind to localized areas of the cell surface, forming compact microcolonies (bacterial clusters) that can be visualized after bacteria have been in contact with cells for 3 h (Fig. 1A and Fig. 1B). This phenomenon is associated with the presence of the large EPEC adherence factor (EAF) plasmid, which carries a cluster of genes that encode type IV fimbriae known as BFP (48), which interconnect bacteria within microcolonies and thus promote their stabilization. In addition, the EAF plasmid carries the Per (plasmid-encoded regulator) regulatory locus (perABC), which is known to be an activator of the plasmid-located bfp operon and also a regulator of the chromosomal LEE genes (98, 143). The EAF plasmid is not essential for the formation of A/E lesions, although its presence enhances the efficiency of lesion formation due to the presence of the Per regulon (98; Chapter Enteropathogenic Escherichia coli), as well as BFP, whose role in cell adhesion would similarly increase the efficiency of A/E lesion formation (43).

Initial in vitro experiments employing cultured epithelial cells of nonintestinal origin have shown that BFP mediate EPEC initial binding, the formation of microcolonies, and interbacterial interactions that allow formation of three-dimensional bacterial aggregates at a later stage of infection (48, 133, 134). A subsequent study employing freshly harvested human intestinal mucosa in the in vitro organ culture (IVOC) system indicated that BFP are involved in the development of the three-dimensional structure of the microcolonies after intimate attachment rather than in the initial adherence (57). The apparent discrepancy in these results has been evaluated using an intestinal cell line, Caco-2 cells, as the in vitro system (141, 142). These studies suggested that EPEC binding occurs through direct interaction with the host cell rather than an already-formed EPEC microcolony and that BFP provide an important function in the cell type-dependent adherence of EPEC and in the progression to the later steps in EPEC adherence (141, 142). A recent study by Cleary et al. (19) also presented evidence showing that BFP can mediate the rapid adherence of EPEC to Caco-2 cells. This study showed that an EPEC intimin mutant strain which cannot form A/E lesions but expresses BFP adheres to human intestinal explants ex vivo, providing further support for a cell adhesion role for BFP. Although it is clear that BFP are involved in interbacterial adherence in the LA pattern, there is no conclusive evidence that BFP mediate actual adherence to epithelial cells (reviewed in reference 107). In addition to the ability for LA, BFP-expressing EPEC strains exhibit autoaggregation, forming large clusters when grown in tissue culture medium (154). The autoaggregation phenotype is lost in a bfp mutant, and the phenotype is restored by complementation. Indeed, all mutants described to date that are unable to make BFP are unable to perform autoaggregation and to produce LA patterns (4, 118). Regardless of the specific role of these fimbriae in adherence, the importance of BFP as a virulence factor in human EPEC infection pathogenesis has been clearly demonstrated by a volunteer trial in which a bfp mutant was ca. 200-fold less virulent than the wild-type BFP-positive parent strain (12).

bfp Operon

The EAF plasmid carries a cluster of 14 genes (Fig. 2), most of which have been shown to be involved in pilus biogenesis (4, 118). When the bfp operon is used for transformation along with an additional fragment of the EAF plasmid carrying the Per regulatory genes, it is sufficient to confer both BFP biogenesis and LA on a laboratory strain of E. coli (133, 134). bfpA, the first gene of the bfp operon, encodes bundlin, the major structural subunit (pilin) of BFP. This gene was first identified during the analysis of a group of mutants deficient in LA that had TnphoA insertions in the EAF plasmid (30). Bundlin is the pilin protein that constitutes the only known structural subunit of BFP filaments and shares limited homology with the type IV pilin proteins of other bacteria (15, 33). Based on the crystal structures of other type IV pilins, each subunit of bundlin is predicted to consist of a globular C-terminal "head" domain which is encoded by the α1 allele and is partially surface exposed and a hydrophobic N-terminal tail buried within the pilus core that stabilizes pilin subunit interactions (116). Prebundlin contains a hydrophilic leader sequence that is cleaved by the prepilin peptidase BfpP (encoded by the bfpP gene) to produce the mature protein (162) and also N-methylated (108) into mature bundlin. Both prebundlin and bundlin are present initially as bitopic proteins in the cytoplasmic membrane, where the periplasmic C-terminal end is folded via two conserved cysteine residues by DsbA (161).

Pilus biogenesis and microcolony dispersal are energy-requiring processes, and the putative nucleotide binding proteins encoded by bfpD and bfpF genes are required to provide the energy for these events (Fig. 2). BfpD is a cytoplasmic protein required for pilus biogenesis (4, 12, 118), and a mechanism by which BfpD transduces mechanical energy to the biogenesis machine has been recently proposed (24). BfpF is also a cytoplasmic protein with homology to the PilT protein found in Pseudomonas aeruginosa, which is required for twitching motility and pilus-specific bacteriophage entry (reviewed in reference 108). Mutation in the bfpF gene results in increased adherence and hyperpiliation, suggesting that this protein is involved in pilus retraction (5, 12). BFP aggregation and dispersal have been shown to take place on cultured epithelial cells (87). In these conditions, BfpF is required for morphological changes in BFP from thin to thick pili as infection proceeds, and these changes lead to loose aggregation of bacteria. BfpF is thought to allow unraveling of these intertwined pilus structures (87).

The other genes in the bfp operon have also been associated with BFP biogenesis. bfpG and bfpB are the second and third genes of the operon (133, 134), respectively, and RNase protection assays showed that the 3' end of bfpG and the 5' end of bfpB are located on a continuous RNA transcript (117). BfpB and BfpG are located in the outer membrane, and BfpB is a lipoprotein that shares sequence similarity with the E. coli bacteriophage f1 morphogenic protein pIV (117), which is a member of the secretin protein superfamily (47). BfpB is composed of a ring-shaped, high-molecular-weight, outer membrane complex which interacts with BfpG. Furthermore, BfpG is required for the formation and/or stability of the BfpB multimeric complex but not for the outer membrane localization of BfpB (125). BfpB expression is also associated with an accumulation of EPEC proteins in growth medium, suggesting that it may support both pilus biogenesis and protein secretion (125). BfpC, BfpE, BfpP, BfpI, BfpJ, BfpK, and BfpL are predicted to be located in the inner membrane (118). The membrane proteins BfpC and BfpE interact with the putative cytoplasmic nucleotide binding proteins BfpD and BfpF to form the cytoplasmic membrane subassembly of the BFP machinery (23). BfpE is a polytopic membrane protein which appears to be a central component in this subassembly, as it interacts with BfpC, BfpD, and BfpF. Furthermore, BfpE is the first protein that is not a member of the PilT family to be implicated in Bfp retraction (23). Finally, BfpU and BfpH are the periplasmic components required for BFP biogenesis. It has been shown that a mutant with a mutation in the bfpU gene was unable to produce LA and perform autoaggregation and it failed to make BFP. Furthermore, BfpU was shown to be completely soluble and was detected in both the periplasm and the cytoplasm, representing a novel protein required for BFP assembly (126). A specific function for BfpH in BFP biogenesis has not been established because a mutation in the bfpH gene, which is predicted to encode a lytic transglycosylase, has no effect on prebundlin expression, prebundlin processing, BFP biogenesis, or LA (4).

BFP and Cellular Receptors

Several type IV pili have been shown to play an important role in mediating bacteria-host cell adherence and have been recognized as true adhesins, e.g., the PilC tip adhesin of Neisseria gonorrhoeae type IV pili (81, 121) and the P. aeruginosa type IV pili which mediate binding via an epitope exposed at the pilus tip (55, 89). Not all of the members of the type IV pilus family have been demonstrated to bind to cells. For example, while the toxin-coregulated pilus of Vibrio cholerae is required for colonization in both the infant mouse model and humans (56), the involvement of the toxin-coregulated pilus in the adhesion of these organisms to the intestinal epithelium has never been demonstrated (136). Similarly, although it seems evident that BFP mediate interbacterial interactions, it is not clear whether BFP act as a host cell adhesin. Although mutations that block formation of BFP and anti-BFP serum also inhibit LA (30, 48), these manipulations may reduce the number of adherent bacteria and influence the pattern of adherence by interfering with interbacterial interactions rather than with bacterium-host cell binding.

If BFP act as an adhesin participating in EPEC interactions with the intestinal epithelium, then they should have a specific receptor and blocking interactions with the receptor should, in theory, inhibit EPEC adherence. A large number of candidate BFP receptors have been proposed, including several oligosaccharides as potential BFP targets. Although one study reported that LA to HeLa cells can be completely inhibited by N-acetylgalactosamine (123), another group reported that EPEC strains that exhibit LA bind to asialo-GM1, asialo-GM2, globoside, and lacto-N-neotetraose and that the minimal receptor contains GalNAc-β-1–4 Gal moieties (62). Other studies have reported that fucosylated tetra- and pentasaccharides (20) and N-acetyllactosamine conjugated to bovine serum albumin and Lewis X-bovine serum albumin inhibit the LA pattern (153). However, immunoblotting for bundlin indicated that the inhibitory effects of these lactosyl glycans were associated with down regulation of bfpA and, thus, may not have involved competition with the target. Furthermore, other studies have implicated phosphatidylethanolamine (PE) as a potential target of BFP (8, 159). The investigators of these studies showed a correlation between the ability to bind PE and the presence of the bfpA gene in a number of strains; however, the strength of the interaction between BFP and PE has not been reported. Even though these studies suggest that LA is a phenomenon that results from direct interaction between an adhesin of typical EPEC and host cells, it remains unclear whether or not BFP are that EPEC adhesin.

MODEL OF TYPICAL EPEC ADHESION

The early and later events in EPEC adhesion and pathogenesis have been proposed to occur in distinct stages (Fig. 3) (28, 31, 108). This model remains controversial and probably artificial, particularly because the early events in EPEC adhesion are poorly understood and the role of several potential adhesins has not been completely elucidated. In the initial stage and under the correct environmental conditions, EPEC cells express BFP, the intimate adhesin intimin, surface-associated EspA filaments, and flagella; the expression of these determinants is dependent on both plasmid and chromosomal genes (Fig. 3A). In this stage, EPEC adhesion may involve binding of intimin to a host cell receptor, such as nucleolin, and binding of the EspA filament directly to the cell or via secreted EspB and EspD proteins. The role of flagella and BFP in adhesion to the host cells still remains controversial. EPEC cells adhering to epithelial cells will use the TTSS to inject the translocated intimin receptor (Tir) and an as-yet-undetermined number of effector molecules directly into the host cell. Effector molecules activate cell signaling pathways, causing alterations in the host cell cytoskeleton and resulting in the depolymerization of actin and the loss of microvilli. Tir is modified by the action of both protein kinase A (54) and redundant tyrosine protein kinases (135) and is inserted into the host membrane. In a subsequent stage (Fig. 3B), the EspA filaments are lost from the bacterial cell surface, the bacterial adhesin intimin binds to the modified Tir, resulting in intimate attachment, and accumulation of actin and other cytoskeletal elements occurs beneath the site of bacterial adherence. A massive accumulation of cytoskeletal elements at the site of bacterial attachment results in the formation of the characteristic EPEC pedestal structure. The translocated effector molecules disrupt host cell processes, resulting in the loss of tight-junction integrity and mitochondrial function, leading to both an increase in intestinal permeability with electrolyte loss and eventual cell death (Chapter Enteropathogenic Escherichia coli ).

POTENTIAL EPEC ADHESINS

Flagella and Activation of IL-8

EPEC strains possess other surface-exposed structures such as flagella. Epidemiological studies have shown that EPEC strains isolated throughout the world express a restricted number of flagellar antigen types, raising the possibility that this surface structure may perform an additional function besides its role in motility. This idea was investigated by Girón et al. (49), who reported that flagella of EPEC are directly involved in the adherence of these bacteria to cultured HeLa cells. The investigators also demonstrated that EPEC flagellum expression is induced by a molecule secreted by the eukaryotic cells and that a molecular cross talk exists between flagella and other virulence-associated systems in EPEC (49). The eukaryotic factor inducing flagellum expression remains unknown, and a recent report found no evidence for such a factor or for production of flagella by EPEC adherent to HeLa cells (160). Recently, another study reported the production of flagella by adherent EPEC but could not confirm a role of flagella in adherence (19). As shown by these conflicting results, there is obviously much additional work to be done to determine the precise role of flagella in adherence. In addition to its potential role as an adhesin, it has been demonstrated that the EPEC flagellar filament and its flagellin monomer can induce the activation of interleukin-8 (IL-8) in intestinal epithelial cells, raising the possibility that EPEC flagellin plays an important role in the intestinal inflammatory response during infection (164). Although flagella of EPEC strains can trigger IL-8 secretion, it has been proposed that these pathogens, in fact, inhibit this response in a TTSS-dependent manner that apparently does not require the known LEE effector proteins (reviewed in reference 25).

Lymphostatin

Another putative adhesin is encoded by the lifA gene. LifA was originally characterized in EPEC as the toxin lymphostatin, which inhibits the transcription of multiple lymphokines and inhibits lymphocyte proliferation (82, 84, 93). Two nearly identical genes are found in strains of EHEC O111 (efa1) (106) and O157:H7 (toxB) (138), and they have been implicated in adhesion to epithelial cells (Chapter Adhesins of Enterohemorrhagic Escherichia coli). An initial study of a lifA mutant of EPEC strain E2348/69 found no difference in adherence to cultured epithelial cells (84), but a subsequent study of a lifA mutant of an E2348/69 derivative cured of the EAF plasmid expressing BFP found reduced adherence, thereby suggesting a potential role for LifA/Efa1 in adherence in the absence of BFP (6). The lifA/efa1 gene is also present in C. rodentium, an enteric pathogen that causes a disease termed transmissible murine colonic hyperplasia, and recent evidence indicates that lifA/efa1 is a critical gene whose protein product regulates bacterial colonization, crypt cell proliferation, and epithelial cell regeneration during in vivo colonization (83).

Other Pili and Adhesins

It has been proposed that a number of other adhesins mediate EPEC-host cell interactions. Recent completion of the genomic sequencing of EPEC strain E2348/69 (http://www.sanger.ac.uk/cgi-bin/blast/submitblast/escherichia_shigella) has revealed the presence of at least 9 regions with homology to potential fimbrial adhesin gene clusters and 10 regions encoding putative nonfimbrial adhesins in addition to the LEE-encoded intimin. One of these regions encodes the type I fimbriae, which have been investigated as a possible EPEC adhesin. Although volunteers challenged with wild-type EPEC developed an immune response against type I fimbriae (75), mutagenesis of the type I fimbriae in EPEC strain E2348/69 failed to show any effect on levels or patterns of adherence of EPEC in vitro (35 ). The role of the EPEC type I fimbriae in human enteric infections is still unclear.

A second region in the chromosome of EPEC E2348/69 has conserved fimbrial genes that encode homologs of the long polar fimbriae (LPF) (139). LPF were first described in Salmonella enterica serovar Typhimurium (Chapter Salmonella Intestinal Infections) and have been shown to direct the attachment of S. enterica serovar Typhimurium to murine Peyer’s patches in vivo (10, 11). The lpf gene cluster is able to mediate microcolony formation by EHEC O157:H7 (Chapter Adhesins of Enterohemorrhagic Escherichia coli) and the adherence of nonfimbriated E. coli ORN172 to epithelial cells (145, 146) and to contribute to the colonization and persistence by E. coli O157:H7 in swine and sheep (69). Furthermore, genes homologous to lpf genes have been identified in rabbit-specific EPEC strains (105) and in invasive E. coli isolates associated with extraintestinal infections (60). While lpf appeared to play a role in the early stages of rabbit-specific EPEC infection in rabbits and is essential for the development of severe diarrhea (105), the lpf of an invasive E. coli O78 strain has been shown to be important only for adherence to epithelial cells in vitro (60). In the case of the lpf region of typical EPEC, it has been found that the cluster contains five ORFs (lpfABCDE) of which the predicted proteins are homologous (about 60% at the amino acid level) to the Salmonella LPF. To determine whether this operon is essential for adherence to IVOC, the lpfABCD genes were deleted from EPEC strain E2348/69 by allelic exchange. Initial analysis of the resulting EPEC ΔlpfABCD strain in IVOC studies showed no change in adherence on small intestinal mucosa compared to that of the wild type. Furthermore, the EPEC lpf mutant showed a normal adherence pattern and formed A/E lesions on HEp-2 cells, suggesting that this lpf gene cluster is not necessary for EPEC adherence and for A/E lesion formation on human biopsy specimens (139). However, there is obviously much additional work to be done to determine the precise role of EPEC LPF in adherence.

Finally, EPEC has been shown to exhibit a type III secretion-dependent, contact-mediated, hemolytic activity requiring the EspA, EspB, and EspD secreted proteins (156). EspA is the major component of a large filamentous structure that is proposed to provide a conduit between the type III translocon and a host cell membrane pore (86). Experimental evidence indicates that EspA filaments mediate the binding of EPEC to red blood cells and that close contact is not a requirement for EPEC-induced hemolysis (129). More importantly, Cleary et al. (19) recently showed that EspA participates in the initial attachment to the brush border of Caco-2 cells. While BFP have been proposed to be the predominant adhesin in the initial attachment of strains expressing both BFP and EspA, the EspA filament has been proposed to be the adhesin mediating initial adherence by EPEC strains that lack BFP (19).

ADHESINS OF ATYPICAL EPEC STRAINS

Different Patterns of Adhesion

As indicated previously, typical EPEC strains show an LA pattern (Fig. 1B) and this pattern is associated with the expression of BFP in all strains tested (50). While this pattern is used to define typical EPEC since it is regularly expressed and specific, the patterns of adhesion observed in atypical EPEC strains have not been clearly associated with an individual adhesin or with an atypical EPEC-specific adhesion factor. For example, the LAL pattern (51) is characteristic of EHEC and atypical EPEC strains and this pattern is proposed to be partially mediated by intimin (111). In the case of AA (Fig. 1C), this pattern defines the category known as enteroaggregative E. coli (Chapter Enteroaggregative Escherichia coli) and is also expressed by all O125ac:H6 strains which are classified as atypical EPEC (150, 152). Finally, atypical EPEC strains of serotypes O55:H⁻ and O55:H7 display the DA pattern (Fig. 1B), which is mediated by the AfaE protein (76). AfaE has identity to a family of afimbrial adhesins found in uropathogenic as well as diarrhea-associated E. coli strains (50.7 and 58.3%, respectively) (44, 45), and expression of these proteins is associated with the DA pattern of a few atypical EPEC, enterotoxigenic E. coli, and enteroaggregative E. coli strains (150) (Chapter Adhesins of Diffusely Adherent and Enteroaggregative Escherichia coli). AfaE, as well as a second adhesin known as AfaD, is encoded by the afaE and afaD genes located on the afimbrial adhesin (afa-3) gene cluster, and the protein products of these two genes have different roles in bacterium-tissue culture cell interactions (44). While AfaD is a protein required for internalization of the adherent bacteria, AfaE allows binding to the tissue culture cells (71).

LDA

One serogroup shared between EPEC and EHEC is O26. E. coli O26 strains that are LEE positive, stx negative, and EAF and bfp negative are classified as atypical EPEC (73, 150). Strains belonging to this serogroup adhere in either a DA or an AA pattern, although some clinical O26 isolates show the LAL pattern (111). Recently, an O26:H11 isolate (E. coli strain 22), originally obtained from an infant with diarrhea in Brazil, was used to identify novel genes conferring adherence on this atypical strain and compare them to other known E. coli adherence factors (122). E. coli 22 is stx negative and eae positive and exhibits an LA pattern, similar to the BFP-mediated phenotype. However, E. coli 22 does not harbor the EAF plasmid or the bfpA gene (122). The authors of this study identified a genomic region in strain 22 which confers DA to tissue culture cells when expressed in E. coli K-12. Analysis of the nucleotide sequence of this region revealed a 15-kb locus responsible for the adhesive properties which was designated the locus for diffuse adhesion (LDA). The lda locus is part of a 20- to 25-kb genomic island inserted into the proP gene of the E. coli chromosome. The lda locus has a G+C content of 46.8%, comprises 15 ORFs, and shares homology with the E. coli K88 fae fimbrial operon (99). Hybridization studies demonstrate that this adhesin is not limited to O26:H11 and that the minor structural subunit gene, ldaH, is found in eae-positive, bfp-negative E. coli isolates representing serogroups O5, O26, O111, and O145 (122). Furthermore, antiserum raised against the major structural subunit, LdaG, recognizes a 25-kDa protein from crude heat-extracted protein preparations and inhibits the adherence of the E. coli DH5α lda⁺ clone to Hep-2 cells. Initial characterization of the structure of the LDA surface-exposed proteins with electron microscopy revealed a nonfimbrial organization surrounding the bacterial cell (122). Although this work is the first report of a sequence of a novel adhesin found in an atypical O26 EPEC strain, additional work is needed to characterize the regulatory network controlling LDA expression and determine the role in virulence of the atypical EPEC LDA adhesin.

CONCLUDING REMARKS

Significant progress has been made in the past two decades in understanding the interaction of EPEC with intestinal epithelial cells. The elucidation of the interaction between Tir and intimin in EPEC and then in other A/E lesion-producing E. coli strains has been one of the key findings to establish a new paradigm in cellular microbiology regarding pathogenic bacteria and host cell interactions. A key issue that still remains controversial is the concept of BFP- and flagellum-mediated adherence to host cell receptors. It has also been established that intimin mediates the intimate adherence to target epithelial cells in culture and IVOC and is a virulence factor of major importance in disease. It is also known that intimin binds in a dose-dependent manner to cells that have been preinfected with EPEC to allow the delivery of Tir. However, the specificity, saturability, and strength of the intimin-Tir interaction have not been clearly established. Further, EPEC intimin also binds to nucleolin in a Tir-independent manner and additional studies with purified proteins are required to establish the significance of this interaction. As a result of cumulative evidence provided by several laboratories, intimin satisfies any criteria to be considered as the major EPEC adhesin. However, further details of the interactions between intimin and Tir or other receptors need to be investigated to establish the relative roles of these interactions, especially in the initial adhesion of EPEC cells.

Furthermore, new fimbrial as well as nonfimbrial adhesins have been found in typical and atypical EPEC strains associated with adhesion. However, many of the adhesion studies with these novel fimbrial and nonfimbrial factors have been carried out in vitro using tissue culture cells not derived from intestinal cells (e.g., HeLa or HEp-2 cells). Even though such experiments provide information about the role played by these factors in adhesion in vitro, the relevance of these findings for in vivo infection are limited and still need to be validated in vitro with intestinal epithelial cells and ex vivo with the IVOC system.

Additional and important basic questions in EPEC adhesion still remain unanswered, such as the specific role of the EspA filament during the interaction with the epithelial cells. It is clear that in order to deliver Tir into host cells, there must be direct contact between the bacteria and host through the TTSS. However, it is not clear whether EspA binds to a cellular receptor or whether EspD and/or EspB mediates this contact by inserting itself into the membrane. Furthermore, if EspA, EspB, and EspD directly interact with the host cells, do they have a specific receptor(s)? Are EspD and EspB first secreted into the extracellular environment from which they insert themselves into the host cell membrane and subsequently bind to EspA, or do they sit at the tip of the EspA filament and bind to or insert themselves into the host cell?

Further characterization of the adherence factors as well as the definition of the specific interactions between the adhesins and their receptors will not only provide a better understanding of the pathogenesis of infections with EPEC strains but will also reveal attractive targets for novel therapeutics and vaccine development.

ACKNOWLEDGMENTS

I thank Kristen J. Kanack for her contribution to Fig. 1, Katie Ris for designing Fig. 3, and Mardelle Susman for critical reading of the manuscript.

My laboratory is supported by institutional funds from the UTMB John Sealy Memorial Endowment Fund for Biomedical Research and by a Pilot/Feasibility Projects in GI Research grant from the Texas Gulf Coast Digestive Diseases Center.