Molecular mechanisms of enterotoxigenic Escherichia coli infection

Abstract

Enterotoxigenic Escherichia coli (ETEC) are a major cause of diarrheal illness in developing countries, and perennially the most common cause of traveller’s diarrhea. ETEC constitute a diverse pathotype that elaborate heat-labile and/or heat-stable enterotoxins. Recent molecular pathogenesis studies reveal sophisticated pathogenehost interactions that might be exploited in efforts to prevent these important infections. While vaccine development for these important pathogens remains a formidable challenge, extensive efforts that attempt to exploit new genomic and proteomic technology platforms in discovery of novel targets are presently ongoing.

1. Introduction

Enterotoxigenic Escherichia coli (ETEC) are a diverse group of pathogens that have in common the ability to colonize the small intestine, where they produce and deliver plasmid-encoded heat-labile (LT) and/or heat-stable (ST) enterotoxins. Collectively, these organisms cause hundreds of millions of cases of diarrheal disease each year, particularly in developing countries. ETEC are responsible for an estimated 300,000–500,000 deaths annually in children under the age of five [1]. These organisms are the most frequent cause of traveller’s diarrhea, and likewise the diarrheal pathogen that most commonly afflicts military personnel deployed to endemic areas. In addition, it appears that these organisms contribute substantially to delayed growth and malnutrition accompanying repeated bouts of infectious diarrhea, and conversely malnourished children appear to be at higher risk of acquiring ETEC infections [2,3].

2. Clinical manifestations of ETEC infection

Enterotoxigenic Escherichia coli infections are classically associated with acute watery diarrhea. Like clinical cholera, these infections can range from mildly symptomatic to severe profuse cholera-like watery diarrhea [4] leading to rapid dehydration and prostration within a few hours [5]. Indeed, initial isolates of enterotoxin producing E. coli were recovered from cases of apparent cholera where no Vibrio could be isolated [6,7]. In effect, ETEC cannot be distinguished from cholera on clinical grounds [8]. In addition to diarrhea, other signs and symptoms including headache, fever, nausea and vomiting are often reported, and some patients may have prolonged diarrheal illness lasting a week or more [9].

3. Molecular mechanisms of virulence

At a minimum, the enterotoxigenic E. coli must be able to produce, secrete, and effectively deliver LT and/or ST. Given the significant phylogenetic diversity observed among ETEC strains, acquisition of genes encoding these toxins may be one of few essential elements required for development of a successful pathogenic clone [10]. However, much remains to be investigated with respect to the overall pathogenesis of these strains, and other virulence factors that may be essential for colonization and for successful targeting of toxins to host cells or which may augment this process have not been sufficiently explored.

3.1. Toxins and their secretion systems

3.1.1. Heat-labile toxin

Heat-labile toxin (LT), like the closely related cholera toxin, is a heterohexameric molecule composed of a pentameric B subunit, and a single A subunit. The A subunit consists of two domains linked by a disulfide bridge: A1, the active toxin molecule, and A2, a helical portion of the molecule that anchors the subunit to the B pentamer [11,12] (Fig. 1). Binding of the B subunit to GM1 gangliosides centered in caveolae on the host cell surface [13] triggers endocytosis of the holotoxin. The enzymatically active A1 portion of the A subunit must be translocated across the intracellular membrane to allosterically interact with ADP-ribosylating factors (ARFs) to ADP-ribosylate Gsα, an intracellular guanine nucleotide protein [14]. Inhibition of Gsα GTPase activity leads to the constitutive activation of adenylate cyclase. In turn, increased levels of intracellular cAMP activate the cystic fibrosis transmembrane regulator (CFTR) chloride channel [15] followed by the ultimate secretion of electrolytes and water that lead to diarrhea [4].

Fig. 1.

Structure of the heat-lablie toxin. (Protein Data Bank entry 1LTB) from [12].

Both CT and LT are secreted through the outer membrane of their respective pathogens by a two-step process. In the first step, N-terminal signal peptides of the subunits are cleaved during sec-dependent [16] transport across the inner membrane to the periplasm where the monomers assemble into holotoxin [17]. Secretion across the outer membrane relies on a complex type II secretion apparatus known as the general secretion pathway (GSP) [18]. In some strains, additional genes such as leoA [19], a GTP-binding protein [20] encoded on a pathogenicity island in the prototype H10407 strain also modulate LT secretion.

The precise process of LT delivery to ganglioside receptors on the surface of intestinal cells is less clear. Earlier investigations suggested that optimal delivery of LT occurs only when the bacteria adhere to target epithelial cells [21], and that anti-LT antibodies which easily bind free toxin are incapable of neutralizing LT delivered by adherent organisms [22]. Interestingly, for many years it was thought that ETEC lacked the ability to secrete LT. However, much of the LT secreted by these organisms under laboratory growth conditions remains associated with outer membrane vesicles which can enter host cells via lipid raft dependent endocytosis [23]. Studies have also suggested that LT and its cognate secretion apparatus [24,25] can cluster or polarize to one end of the bacterium, potentially permitting ETEC to deliver their toxin payload in a highly vectored fashion at the host cell surface [24].

In addition to its role in fluid secretion, LT may elicit a variety of effects that benefit the organism. LT down-regulates innate host responses including defensins [26], and enhances ETEC adherence to epithelial cells [27] and colonization of the small intestine [28].

3.1.2. Heat-stable toxin

Heat-stable toxins are small cysteine-rich peptides secreted by ETEC, which bind to the extracellular domain of guanylyl cyclase C (GC-C) on the brush border of intestinal epithelium. These interactions activate the intracellular catalytic domain of guanylyl cyclase leading to the intracellular accumulation of cGMP [29,30]. Increases in cGMP in turn activate cGMP-dependent protein kinase II leading to phosphorylation of the cystic fibrosis transmembrane regulator (CFTR) [31] driving Cl- secretion and inhibition of NaCl absorption followed by net loss of water through osmotic diarrhea.

Several ST peptides have been identified in human ETEC strains. These include the related GC-C-binding STa (STI) peptides ST-Ia (ST-P) and ST-Ib (ST-H), as well as the unrelated STb (STII) molecules [32], typically associated with porcine strains. However, STb binds to different receptors, does not stimulate production of cyclic nucleotides, and has not been clearly linked to human disease [33]. STI molecules share a core structure of 13 amino acids containing three disulfide bonds that are required for the biologic effect (Fig. 2). The crystallographic structure of the active ST-P toxin domain [34], predicts the formation of a hexameric ring with residues N¹¹–A¹³ forming a putative GC-C binding region oriented to the outer surface, and it has been postulated that this may promote GC-C clustering and activation [35].

Fig. 2.

Heat-stable peptides. The predicted structure of the active toxin domain of the ST-P molecule (C⁵‒eC¹⁷) as determined by Ozaki, et al. Sulfur atoms on cysteine residues involved in disulfide bond formation are shown in red (structure was generated from Protein Data Bank Entry 1etn using Protein Workshop [99]). The peptide sequences of the mature ST-P and ST-H peptides are shown (residues highly conserved with other GC-C binding molecules enclosed), and the core toxic domain is underlined.

Both ST-H and ST-P are plasmid encoded [36] (Table 1), often in transposons [37], and initially synthesized as 72 amino acid precursor molecules containing a 19 residue signal peptide required for entry into the periplasm via Sec. Export of STI peptides through the outer membrane requires the trimeric TolC protein exporter [38].

Table 1.

Location of genes encoding known and putative virulence proteins in ETEC strain H10407.

Plasmid-encoded	Plasmid size (kb)
CFA/I operon	94.8
EtpBAC two partner secretion system	94.8
EatA autotransporter	94.8
EAST peptide	94.8
ST 1b (ST-H)	94.8
CexE	94.8
ST 1a (ST-P)	66.7
LT	66.7
Chromosomally encoded	Location
Tia adhesin/invasin	PAIa-selC
leoA GTP binding protein	PAI-selC
TibA autotransporter	YeeT
TibC (glycosyltranserase)	upstream from TibA

^aPAI, pathogenicity island.

Large epidemiological studies that have distinguished between ST-H and ST-P indicate that ETEC producing ST-H may be more pathogenic than the ST-P producers [39]. On the other hand, there seems to be little doubt that ST-P-containing ETEC are capable of causing disease in humans, because ST-P strains have been found to cause multiple food-related outbreaks of diarrhea in Japan [40]. It is not known whether the apparent difference in pathogenic potential between ST-H- and ST-P-producing ETEC is a consequence of biological disparity between the two toxins or if they are but markers of pathogenicity.

3.1.3. EAST1

EAST1, the enteroaggregative heat-stable toxin, shares structural similarity to STI peptides, and also leads to increases in cGMP [41,42]. The ast gene encoding EAST1 has been identified, often in multiple copies [43], in a variety of enteric pathogens including ETEC [44]. Studies in ETEC have demonstrated that this toxin resides on a mobile element and that when cloned and expressed in a recombinant background, EAST1 is functionally active in assays of enterotoxin activity [42]. While demonstration of EAST1 involvement in ETEC pathogenesis remains outstanding, the presence of ast genes in multiple strains may suggest functional redundancy of toxins with the capacity to provoke elevated levels of cGMP.

4. Adhesins and colonization factors

Colonization of the small intestine is thought to be an essential virulence trait for ETEC. A variety of different structures encoded either on plasmids or the chromosome of ETEC have been identified as putative adhesins or colonization factors.

4.1. Colonization factors (CFs)

A heterogeneous group of proteinacious surface structures referred to as colonization factors (CFs) were among the first virulence factors identified in ETEC [45] and remain important targets for vaccine development [46]. At least 25 different CFs have been described to date and most are plasmid-encoded. These structures have been reviewed in some detail elsewhere [47,48].

CFs are antigenically and structurally diverse. Fimbrial, fibrillar and helical structures have been identified, with lengths ranging from 1-to more than 20 μm [49]. CFA/I fimbriae are among those studied most intensively. These fimbriae consist of approximately 1000 copies of the major fimbrial subunit CfaB, and one (or few) copies of the CfaE adhesin molecule, located at the distal tip. Assembly of these organelles requires dedicated periplasmic chaperone (CfaA) and outer membrane usher (CfaC) proteins encoded on the same operon. The assembled structures (≈1 μm long), are proposed to adopt spring-like helices that can uncoil when subjected to shear stresses in the gut lumen [50,51]. The precise receptors for most of the CFs have not been identified, however, many are thought to bind to glycoprotein conjugates on the surface of host cells [52].

Interestingly, while colonization factors clearly play an important role in the pathogenesis of human disease, many strains do not produce a recognizable CF. Moreover, although a CF-based vaccine did induce protection against more severe diarrhea with CF-homologous strains among US travelers to Guatemala [53], the same vaccine failed to induce protection against ETEC diarrhea in Egyptian children [46]. Studies of the protective effects of CFs following natural ETEC infection have also yielded divergent results. Studies in Egypt found that in children <18 months of age, CFA/I antibodies, (but not antibodies against CS3, CS6, or LT) were associated with an inverse correlation with the risk of diarrhea caused by ETEC expressing the homologous CF. Likewise, studies in Bangladesh found that prior exposure to ETEC prevented subsequent diarrhea due to organisms of the same CF type [2]. Conversely, an extensive cohort study following children from birth until up to two years of age found no attributable contribution of CFs to the substantial anti-colonizing immunity that these children experienced following ETEC infection [54].

The different findings in these studies likely reflect differences in study methodology. However, they may also be indicative of the complex biology involving multiple molecular events that collectively contribute to intestinal colonization, and subsequent diarrheal disease. Dissection of these events and identification of additional adhesins that contribute to colonization, particularly for strains that do not appear to make one of the known CFs, could yield additional targets for vaccine development.

4.2. Other fimbrial operons

4.2.1. Type I fimbriae

ETEC like other E. coli produce common type I fimbriae [55]. Despite the clear role of these structures in the pathogenesis of uropathogenic E. coli, early studies failed to support their involvement in ETEC virulence [56]. However, subsequent studies implicating variations of the type I fimbrial tip adhesin, FimH, in tissue tropism [57] raise the possibility that some type I fimbriae could be for optimally suited for small intestinal colonization.

4.2.2. E. coli common pilus

Other structures common to many E. coli, including both commensal and pathogenic isolates, are the E. coli common pili (ECP) [58]. A potential ECP operon of six genes encodes putative proteins similar to other molecules known to be involved in pilus assembly. Recent studies in ETEC demonstrate that approximately 80% of strains carry the ecpA gene encoding the major pilus structural subunit, and that more than half of isolates tested could be shown to produce the EcpA protein by immunoblotting [59]. While the role of ECP in ETEC pathogenesis is not yet established, the high degree of conservation in this genetically diverse pathotype certainly make these pili compelling structures for future investigation.

4.3. Non-fimbrial adhesins/invasins of ETEC

4.3.1. Tia and TibA invasins

While investigating Salmonella typhi invasion, Elsinghorst serendipitously discovered that ETEC strain H10407 could efficiently enter (but not replicate in) cell lines derived from gastrointestinal epithelium [60]. He then cloned two chromosomally encoded toxigenic invasion loci: A (tia) and B (tib). Subsequent investigation revealed that these loci encode distinct proteins.

Tia, is a 25 kD outer membrane protein [61] encoded on a large pathogenicity island inserted in the selC tRNA gene of H10407. Tia interacts with host cell surface proteoglycans [62], and by itself is sufficient to promote adherence and epithelial cell invasion when cloned into laboratory strains of E. coli.

The tib locus encodes TibA an autotransporter (AT) protein with homology to several AT adhesins from other mucosal pathogens [63]. TibA is synthesized as a 100 kD precursor protein, preTibA, that is glycosylated through the action of TibC [64], a putative glycosyltransferase.

The role played by actual invasion of epithelial cells to molecular pathogenesis of ETEC remains uncertain. Although ex vivo studies with human enterocytes from biopsies nicely demonstrated interaction of ETEC with the brush border [65], similar data from infected patients are lacking. Nevertheless, the ability of Tia and TibA to promote adhesion in vitro and their significant homology with known virulence proteins from other pathogens might suggest that they promote adherence.

4.3.2. EtpA

A search for novel secreted proteins led to the identification of a two-partner secretion (TPS) locus encoded on the large virulence plasmid of ETEC H10407 [66]. This locus encodes three proteins: EtpA, a 170 kDa secreted glycoprotein, EtpB a transport pore, and EtpC, a putative glycosyltransferase required both for optimal secretion and glycosylation of EtpA. EtpA, like several other members of a growing family of TPS exoproteins, functions as an adhesin. EtpA appears to act in a novel fashion. Recent studies suggest that EtpA functions a molecular bridge, binding both to host cell receptors and to the tips of ETEC flagella, where it interacts with highly conserved regions of flagellin proteins [67]. EtpA, as well as its interactions with flagellin, are required for optimal adhesion of H10407 in vitro, and for intestinal colonization in a murine model [68].

Similarity of EtpA to filamentous hemagglutinin, a high molecular weight TPS adhesin of Bordetella pertussis that is a component of the acellular pertussis vaccine, has stimulated investigation of EtpA as an immunogen. Indeed, vaccination of mice with either a truncated recombinant 110 kDa EtpA frament [68] or the full length EtpA glycoprotein [69], affords significant protection against colonization with ETEC suggesting that this molecule could represent a viable target for vaccine development.

5. The importance of flagella to human ETEC pathogenesis

E. coli, including ETEC, produce peritrichous flagella each of which are comprised of an estimated 20,000 individual flagellin molecules. Flagellin molecules stack in a helical array in which highly conserved amino and carboxy terminal regions interact to align inside the flagellar shaft while variable regions orient to the outer surface where they account for “H” antigen serotype specificity. Most human ETEC strains are motile, and the majority of ETEC strains can be serotyped on the basis of their flagellar H antigens, suggesting that flagella are likely to be involved in ETEC pathogenesis [70]. However, only recently have these structures been investigated in ETEC virulence. To date, studies have demonstrated that intact flagellar structures are essential for both ETEC adherence and heat-labile toxin delivery in vitro [24], and contribute significantly to intestinal colonization [67].

Analysis of over 700 ETEC strains identified more than 30 different H serotypes, leading to the suggestion that flagella would not be good vaccine targets unless a conserved epitope could be identified [70]. The surprising finding that highly conserved regions of these molecules are exposed at the tips of some ETEC flagella [67] and participate in adherence, has opened the possibility that these domains could be targeted in a vaccine. Recent studies demonstrating cross-protection by immunization with a recombinant heterologous H antigen appear to support this hypothesis [69].

6. Other potential virulence factors

6.1. EatA

Eat A is an autotransporter encoded on the large virulence plasmid of ETEC strain H10407 [71]. Genes similar or identical to eatA are present in multiple ETEC strains by DNA hybridization and/or PCR studies, and eatA has also been found through genomic sequencing of E24377A [72], and the pCOO virulence plasmid [73]. Interestingly, sequences with EatA homology have also been reported in sequencing of porcine ETEC plasmids [74]. EatA has been shown to function as a serine protease, and therefore belongs to a family of putative virulence proteins known as SPATE proteins (Serine Protease Autotransporters of Enterobacteriaciae). EatA shares considerable homology with the Shigella flexneri autotransporter, SepA. The precise function that EatA plays in pathogenesis or the actual substrate for its protease activity is not clear. However initial studies in ileal loops suggested that EatA could contribute to accelerated virulence of eatA⁺ strains [71].

7. Genetics of ETEC

7.1. ETEC genomics

With the decreased cost and increased throughput of next generation sequencing technology it has been possible to sequence multiple isolates of any species or pathotype. This influx of genomic data has been applied to E. coli in general and specifically to the ETEC pathotype on a small scale to date, but considering the importance of this pathogen in global public health we anticipate that tens, if not hundreds of isolates will be sequenced in the near future. Two genomes have been completely sequenced and closed including the plasmids (E24377A and H10407) and one other isolate has been sequenced to draft level coverage (B7A) (Table 2). Examination of the genomic content of the ETEC genomes has revealed that ~4% of the E24377A chromosome is composed of mobile genetic elements and similar proportions are observed in the two other genomes [72]. Additionally, when the E24377A and B7A genomes were compared to 15 other E. coli genomes, from various pathotypes, on average they contained the highest number of unique genes per strain [72]. The combination of a large percentage of mobile elements and unique genes in the ETEC genomes suggests that there is a great deal of gene flux in these pathogens.

Table 2.

Characteristics of ETEC genome sequencing projects.

Strain	Serotype	Virulence factors	Genome status	Plasmid description	Size (kb)	GenBank accession	Reference
E24377A	0139:H28	LT,ST; CS1,CS3	Complete	pE24377A_5 pE24377A_6 pE24377A_35 pE24377A_70 pE24377A_74 pE24377A_79	5.1 6.2 35.3 70.6 74.2 79.2	CP000800 (chrom), CP000801 (5), CP000798 (6), CP000796 (35), CP000797 (70), CP000799 (74), CP000795 (79)	[64]
H10407	O78:H11	LT,ST; CFA/I	Complete	H10407_p52 H10407_p58 H10407_p666 H10407_p948	5.2 5.8 66.6 94.8	http://www.sanger.ac.uk/Projects/E_coli_H10407/	Not published
B7A	O148:H28	LT/ST CS6	Draft (198 contigs)	Yesa		AAJT00000000	[64]
Plasmid only projects
C921b-1	O6:H16	LT+;ST+; CS1 + CS3+	–	pCOO	94.8	CR942285	[65]
E1392/75	O6:H16	LT+;ST+; CS1 + CS3+	–		Draft	http://www.sanger.ac.uk/Projects/	Not published

^aDraft sequences contain contigs that appear to be of plasmid origin, however no plasmids have been closed from these projects.

One very interesting finding to come out of the genome projects is the identification of multiple plasmids. Yamamoto and Yokota previously described multiple plasmids in H10407, however there appeared to be variability in the plasmid profiles of isolates from different sources [75]. The sequence data of ETEC plasmids suggests that this variability is linked to the identical mobile elements in the plasmids and chromosomes potentially leading to plasmid instability. Sequencing of the pCOO plasmid by Froehlich et al. revealed that the plasmid is a mosaic cointegrate composed of ~24% mobile elements and two distinct replication origins related to R64 and R100 [73]. The mobile elements that litter the pCOO plasmid as well as the plasmids of E24377A and H10407 are of a few defined types including IS1, IS2, IS66, IS629 and IS91. The increased proportion of these IS elements in the plasmids and chromosome may allow for integration and resolution of the plasmids in the chromosome at specific points in the ETEC life cycle and contribute to the observed instability.

While currently available genetic information is already being exploited in ETEC pathogenesis studies, expansion of these data to include additional isolates should yield additional details pertinent to understanding the virulence of these pathogens, and permit detailed exploration of regulatory networks. Likewise, given the plasticity of these pathogens, additional genetic information from multiple strains will likely be crucial to successful identification of conserved vaccine targets, and should also aid in identification of specific virulence factors that may be associated with more aggressive forms of clinical ETEC infection including severe cholera-like diarrhea that leads to rapid dehydration.

7.2. Genetic/phenotypic diversity and its implications for vaccine development

ETEC represent a phenotypically and genotypically diverse group of E. coli. This is exemplified by the fact that serotype analysis of over 700 ETEC strains isolated from different parts of the world identified 78 different O serogroups and 34 different H serogroups in 118 different combinations [70]. The most commonly identified CF in that study, CFA/I, was represented in 23 different O serogroups.

Results from phylogenetic studies indicate that ETEC have emerged from the E. coli population on several occasions through independent acquisitions of the enterotoxin genes [10]. Initial chromosomal gene content comparisons of three recently sequenced ETEC strains stemming from three different ETEC lineages did not result in the identification of any ETEC-specific genes [10], potentially indicating that relatively few chromosomal genes are needed to establish an ETEC lineage. However, further studies are clearly needed to more adequately identify genes that may be shared between ETEC isolates.

Although strains stemming from different ETEC lineages may appear at present to have few obvious antigens in common with each other except for the toxins and CFs, there are indications that as yet undefined plasmid or chromosomally-encoded antigens may play an important role in the development of natural immunity against ETEC. In a longitudinal epidemiological study of 200 young children in West Africa, natural ETEC infections resulted in a 47% protection against new infections with ETEC that carried the same combination of toxins and CFs, but this protection did not seem to be attributed to the toxins or the CFs [54]. Strains that have the same toxin and CF combination often stem from the same ETEC lineage [76–78], potentially indicating that other plasmid or chromosomally-encoded antigens may provide protection against other ETEC that belong to the same lineage. There is currently little known about which ETEC lineages exist in the human population, or about the rate at which lineages emerge and disappear from the E. coli population. For lineage-specific vaccines to become a viable option for the vaccine development effort, attention would probably have to be focused on protecting against strains from the most predominant and stable ETEC lineages.

7.3. Transcriptional regulation in ETEC

As with other enteric pathogens ETEC are likely programmed to respond to multiple environmental signals to coordinate the expression of virulence factors. For instance, expression of enterotoxins and some pilus types is influenced by the availability of certain carbon compounds, such as glucose. LT production is elevated in the presence of glucose [79] an effect known as catabolite activation that requires cAMP receptor protein (CRP). The consensus DNA binding site for CRP is a 22 bp spaced inverted repeat and as its name implies, DNA binding is cAMP dependent. The LT promoter is repressed by CRP when it occupies a site centered at ‒31.5, relative to the transcription start site. When bound to this site CRP occludes RNA polymerase from the promoter and thus represses transcription of the genes encoding LT. Repression of the LT promoter is relieved by glucose which prevents CRP from binding DNA by inhibiting the synthesis of cAMP. In contrast to LT, CRP activates the expression of ST-H [79].

It is also likely that CRP activates the expression of a variety of pili including CFA/I, because it has been shown that their expression is inhibited by glucose (catabolite repression) [80]. These pili are also positively regulated by Rns (CfaD) and all Rns-dependent pilin promoters have a binding site immediately upstream of the promoter’s ‒35 hexamer [81]. Most promoters also have one or more additional Rns binding sites further upstream of the promoter-proximal site. It is not yet known if CRP regulates the pilin promoters directly or indirectly by controlling the expression of Rns. For 987P pili, which are expressed by some strains of porcine ETEC, regulation is indirect with CRP activating the expression of FasH (FapR) which then activates the pilus promoter [82].

Additional factors likely modulate the expression of virulence genes in ETEC. Both LT expression and the production of some pili, including CFA/I [83]are also repressed by H-NS, a nucleoid associated protein of gram-negative bacteria and a global modulator of gene expression [84].

Interestingly, both CRP and H-NS have been shown to govern the production of flagella in E. coli [85]. Given the requirement of these structures for effective bacterial adherence and intestinal colonization [67] as well as toxin delivery [24], it is likely that their expression is coordinated with virulence factor production in ETEC. This view appears to be supported by the recent identification of a CRP binding site upstream of the etpBAC promoter by DNase I footprinting (Fleckenstein and Munson, unpublished).

Searches for additional genes that might be part of the same transcriptional network can also help to identify novel virulence factors. Recently, in silico scanning of sequenced ETEC genomes for putative Rns binding sites was employed to identify another potential secreted virulence protein of ETEC, CexE [81], encoded on the large 94.8 kB virulence plasmid of H10407 along with genes for EtpA, EatA, and CFA/I. Interestingly, an insertion in cexE was also identified using TnphoA mutagenesis to screen for secreted proteins (Fleckenstein, unpublished), further suggesting that cexE does encode a novel secreted protein. Rns (CfaD) activates the expression of cexE. Similar to what has been shown for the CS1 pilin promoter, activation of the cexE promoter requires a Rns binding site immediately upstream of the promoter’s −35 hexamer. Occupancy of an additional, promoter-distal binding site is required for full activation of the cexE promoter. The cexE gene encodes a 12.6 kDa protein whose first 19 amino acids constitute a signal peptide that is cleaved during sec-dependent transport across the inner membrane. The function of CexE has yet to be elucidated. However, the location of cexE on a virulence plasmid where its expression is regulated by Rns, known to regulate the expression of other virulence factors, strongly suggests that this gene encodes a novel secreted virulence protein.

8. Comparisons to porcine ETEC

Enterotoxigenic E. coli are a significant cause of post-weaning diarrhea (PWD) in pigs [86], and similar to human strains, ETEC that cause PWD express heat-labile (LT) and/or heat-stable enterotoxins (STb). These organisms also parallel human strains in that most express fimbrial adhesins responsible for colonization of the small intestine [87]. Despite these overall similarities, ETEC in general exhibit considerable host specificity, likely due to distinct adhesin/host receptor relationships. Therefore to date, attempts to cause disease with human strains in porcine infection models have met with limited success.

Still, porcine ETEC can provide important clues relevant to the understanding and prevention of human infections. Indeed, many of the seminal discoveries for ETEC in general were made in investigation of porcine infections, and ETEC were first established as a cause of disease in pigs [86,88], prior to identification of toxin-producing E. coli as a cause of human diarrheal disease [89]. Porcine ETEC strains were first used to show that strains possessing heat-labile toxin activity (later identified as LT) were responsible for causing diarrhea in various animals, and that fimbrial structures (pili) were involved in colonization of the small intestine [87]. Likewise, the present effort toward fimbriae-based vaccines for human ETEC stems in part from early demonstrations that vaccination pregnant swine with fimbriae prevented infections in newborn pigs [90].

A number of important general insights about bacterial virulence and enterotoxin structure and function have been gleaned from previous and ongoing studies of porcine ETEC including a determination of the LT subunit composition [91] and its binding specificity for GM1 [92]. Of recent interest has been the concept that bacterial pathogens disseminate virulence factors including LT in vesicles released from the outer membrane [23], a process described some time ago using porcine [93] ETEC isolates.

Recent studies using porcine intestinal epithelial cell cultures [94], have further suggested that the pathogenesis of ETEC in general may be more complex than previously appreciated. Interestingly, in this system, LT appears to precondition the host intestinal epithelium for ETEC adherence, in a process dependent upon the ADP-ribosylation activity of this toxin [27]. Conversely, at least some ETEC have the capacity to induce apoptosis through a process independent of LT expression [95], and intriguingly these changes also appeared to promote bacterial adherence. While the full complement of changes to the host in response to interactions with ETEC remain to be elucidated, the interplay of ETEC induced apoptosis of intestinal cells and enhanced adherence is intriguing. These findings should stimulate additional clinical investigation particularly in light of the interplay between ETEC infections and malnutrition [3].

9. Host factors influencing ETEC infection

Certainly, many parameters including expression of specific virulence factors by the bacteria, both innate and acquired host defenses, and the genetic background of the host influence the clinical presentation of ETEC infection following ingestion of a sufficient inoculum. However, only recently have studies begun to examine host factors associated with either the acquisition of ETEC infections or different clinical presentations that range from asymptomatic colonization to severe, life-threatenting diarrhea. As previous studies showed that individuals with type O blood were at enhanced risk for severe Vibrio cholerae infection, several investigations have now focused on the relationship of ABO blood group determinants to ETEC infections. In contrast to V. cholerae infections, data from Bangladeshi children demonstrated no association with blood group O and diarrhea [96]. However, a subsequent study conducted in a birth cohort of over 300 children in Dhaka demonstrated that blood groups A or AB were associated with a significantly higher prevalence of ETEC diarrhea than those with group O blood [2]. Likewise, children in Bangladesh with Lewis blood group antigen-a positivity (Le^a+ Le^b‒) more frequently had symptomatic ETEC infections [97] Presum. ably glycoconjugates on the surface of red blood cells may also be expressed on intestinal epithelia where they act as receptors from one or more bacterial adhesins, and indeed CfaB, the major subunit of Cfa/I fimbriae binds to glycosphingolipids including Le^a [52]. In addition to the associations with blood group antigens, studies of traveler’s diarrhea have also identified a synonymous single nucleotide polymorphism in the human lactoferrin gene that was associated with an increased risk of diarrheal illness [98].

It is likely that over time a number of polymorphisms in the human genome will be associated with variant clinical forms of ETEC infection, and likewise that specific virulence factors will be associated with more severe forms of disease. Understanding these associations could theoretically facilitate development of vaccines designed specifically to prevent lifethreatening diarrheal illness.

10. Future directions

Recent ETEC pathogenesis studies have revealed an increasing degree of complexity for pathogens that were once believed to simply release their toxin payloads by lysis in the small intestine. Exploitation of newly discovered virulence factors, together with expanded information on ETEC genomes, and anxiously anticipated antigen discovery tools including pan-genome DNA and protein microarrays currently in development offer unprecedented opportunities to address the challenge of developing a broadly protective vaccine for these important pathogens.