Escherichia coli O157:H7 colonization in small domestic ruminants

Authors


  • Editor: Simon Cutting

Correspondence: Roberto M. La Ragione, Department of Food and Environmental Safety, Veterinary Laboratories Agency (VLA), Weybridge, New Haw, Woodham Lane, Addlestone, Surrey KT15 3NB, UK. Tel.: +01932 359 478; fax: +01932 347 046; e-mail: r.laragione@vla.defra.gsi.gov.uk

Abstract

Enterohaemorrhagic Escherichia coli O157:H7 was first implicated in human disease in the early 1980s, with ruminants cited as the primary reservoirs. Preliminary studies indicated cattle to be the sole source of E. coli O157:H7 outbreaks in humans; however, further epidemiological studies soon demonstrated that E. coli O157:H7 was widespread in other food sources and that a number of transmission routes existed. More recently, small domestic ruminants (sheep and goats) have emerged as important sources of E. coli O157:H7 human infection, particularly with the widespread popularity of petting farms and the increased use of sheep and goat food products, including unpasteurized cheeses. Although the colonization and persistence characteristics of E. coli O157:H7 in the bovine host have been studied intensively, this is not the case for small ruminants. Despite many similarities to the bovine host, the pathobiology of E. coli O157:H7 in small domestic ruminants does appear to differ significantly from that described in cattle. This review aims to critically review the current knowledge regarding colonization and persistence of E. coli O157:H7 in small domestic ruminants, including comparisons with the bovine host where appropriate.

Introduction

Enterohaemorrhagic Escherichia coli (EHEC) O157:H7 is a zoonotic enteric pathogen of worldwide importance. In the early 1980s, human infection with E. coli O157:H7 was first recognized to be associated with bloody diarrhoea, haemorrhagic colitis, haemolytic uraemic syndrome (HUS) and thrombotic-thrombocytopaenic purpura (Karmali et al., 1983; Riley et al., 1983; Pai et al., 1984; Ryan et al., 1986). ‘EHEC’ is a pathotype designation that was created for E. coli of any serotype that was associated with this pattern of bloody diarrhoea and systemic pathology, including HUS, in humans (Levine, 1987). Subsequently, a typical pattern of virulence determinants in EHEC isolates was established, including the elaboration of Shiga toxin(s) (Stx) and the ability to adhere intimately to cell lines in vitro, forming attaching–effacing (AE) lesions (Nataro & Kaper, 1998). The first recorded outbreaks of E. coli O157:H7 infection in humans were associated with the consumption of ground beef, and thus cattle were soon recognized as important reservoirs (Martin et al., 1986; Bopp et al., 1987; Borczyk et al., 1987; Riley, 1987).

In recent years, the annual incidence of reported clinical EHEC O157:H7 infection in humans in the United States has been around 1 per 100 000 (Centers for Disease Control and Prevention, 2008), with 2621 cases reported in 2005. Twenty-six countries (principally European but including Japan) reported a total of 2937 EHEC O157:H7 infections to Enternet in 2005 (Anon, 2007). Of these cases, 21% developed HUS, which is a high percentage, suggesting that many less severe cases went unreported.

Transmission of E. coli O157:H7 to humans is principally via contamination of food by animal faeces, with cattle considered to be the primary reservoir (Griffin & Tauxe, 1991; Zhao et al., 1995; Hancock et al., 1997). Early studies by Doyle & Schoeni (1987) indicated that E. coli O157:H7 was also widespread in meat sources other than beef. In addition, numerous human disease outbreaks have been associated with the consumption of plant products, including apple cider and vegetables such as lettuce, radishes, alfalfa sprouts and spinach (Besser et al., 1993; Fukushima et al., 1999; Ferguson et al., 2005; Centers for Disease Control and Prevention, 2006; Maki, 2006). There are also traceable links between human infection and ruminant faeces via water or direct contact (Licence et al., 2001; Strachan et al., 2001), and evidence that contact with animal faeces is a strong risk factor for sporadic E. coli O157:H7 infection (Locking et al., 2001). Interpersonal spread can also be a significant factor in outbreaks (Ryan et al., 1986; Cryan, 1990). Therefore, E. coli O157:H7 may also be considered as an environmental pathogen (Strachan et al., 2006; Solecki et al., 2007).

The prevalence of E. coli O157:H7 in cattle has been the subject of much research, with most studies concluding that at any given time the majority of cattle are negative for E. coli O157:H7 (Zhao et al., 1995; Gansheroff & O'Brien, 2000; Laven et al., 2003; Omisakin et al., 2003). The prevalence quoted in the literature of E. coli O157 in individual cattle varies considerably, with rates of 1.8% in Japan (Miyao et al., 1998), 1.9% in Australia (Cobbold & Desmarchelier, 2000), 1.5% in Brazil (Cerqueira et al., 1999), 0–7.4% in the United States (Faith et al., 1996) and 1–4.2% in England and Wales (Chapman et al., 1993; Richards et al., 1998; Paiba et al., 2003). In Scotland, Synge & Paiba (2000) reported that in beef herds the individual- and herd-level prevalences were around 8.7% and 24%, respectively. Differing sampling techniques have hampered valid comparisons between geographical locations and consequently there is some controversy over prevalence rates.

In cattle in England and Wales, Paiba et al. (2002) found that the highest prevalence of E. coli O157:H7 was seen in late summer and early autumn, which is in agreement with other sentinel studies (Chapman et al., 1997; Hancock et al., 1997; Tutenel et al., 2002). However, by contrast, Ogden et al. (2004) found that the prevalence of E. coli O157:H7-positive cattle was higher in the cooler months, although heaviest shedding of the organism by individual cattle was seen in the summer.

Much work to date on the colonization, persistence and control of E. coli O157:H7 has focused on cattle, they being the perceived primary reservoir host. Numerous E. coli O157:H7 factors for bovine colonization have been identified and characterized. However, several other species including rabbits, deer, water buffalo, pigs, chickens and seagulls have also been implicated as carriers of E. coli O157:H7 (Griffin & Tauxe, 1991; Pritchard et al., 2001; Eriksson et al., 2003; Galiero et al., 2005; Dipineto et al., 2006; Foster et al., 2006; Scaife et al., 2006; Garcia-Sanchez et al., 2007; Cornick & VuKhac, 2008). In particular, there is much evidence of the carriage of toxigenic E. coli O157 by small domestic ruminants (sheep and goats) (Chapman et al., 1997; Heuvelink et al., 1998; Meng et al., 1998; Fegan & Desmarchelier, 1999; Ogden et al., 2005). Furthermore, human cases and outbreaks of EHEC O157:H7 disease have often been linked to open farms or petting zoos, where the organism has been isolated from animals including small ruminants (Shukla et al., 1995; Chapman et al., 2000; Pritchard et al., 2000; Heuvelink et al., 2002; Payne et al., 2003; Stirling et al., 2008). Other cases have been linked to small ruminant dairy products (Bielaszewska et al., 1997; Steen et al., 2001; McIntyre et al., 2002; Espie et al., 2006). With the increasing popularity of petting farms, the intensification of goat and sheep farming and the wider availability of sheep and goat products worldwide (including milk and unpasteurized cheeses), small ruminants are gaining attention as potentially significant reservoirs of E. coli O157:H7. The present review aims to provide insights into the work to date on the colonization and persistence of E. coli O157:H7 in small domestic ruminants and the role that they may play as reservoirs of E. coli O157:H7 for human infection.

Sheep and goats as reservoirs for E. coli O157:H7

The prevalence of E. coli O157:H7 in small domestic ruminants is less well documented than in cattle, and reports vary considerably. Battisti et al. (2006) reported a prevalence of 0.2% in lambs taken to slaughter in Italy. In the Netherlands, the prevalence, using a sensitive immunomagnetic separation (IMS) culture technique on faeces at slaughter, varied from 3.8% to 4.1% of animals depending on the age of the animals sampled (Heuvelink et al., 1998). A prevalence of 8.7% positive flocks was found in a Spanish study, also using IMS on rectal faeces samples, with around 7.3% of individuals within affected flocks found to be excreting E. coli O157:H7 (Oporto et al., 2008). In England and Wales, using a similar methodology, the individual prevalence varied between 1.7% and 2.2% (Chapman et al., 1997; Paiba et al., 2002), and more recently in Great Britain, toxigenic E. coli O157 was found in 0.7% of faeces samples at slaughter (Milnes et al., 2008). In Scotland, the IMS-determined prevalence of E. coli O157 among sheep pasture faeces samples was 6.5% (Ogden et al., 2005). One study in the US cultured faeces samples from 35 sheep on three occasions over 6 months and reported a peak prevalence, in June, of 31%E. coli O157:H7-positive animals (Kudva et al., 1996).

In addition to the prevalence reports above, some studies have cited sheep products as important sources of E. coli O157:H7 (Rey et al., 2006; Kalchayanand et al., 2007). Furthermore, sheep have also been cited as reservoirs for a diverse number of non-O157 serogroups (including O26, O91, O115, O128 and O130) that encode a key colonization factor in common with E. coli O157:H7, namely the potential to cause AE lesions at the epithelial mucosa (Djordjevic et al., 2001; Cookson et al., 2002a, 2006; Blanco et al., 2003; Aktan et al., 2004; Kalchayanand et al., 2007). Many of the above serogroups have been implicated in human disease. For example, a virulent O103:H25 EHEC strain has been traced directly to sheep products (Schimmer et al., 2008). Non-O157 E. coli possessing key EHEC genes encoding intimin and Stx are apparently more prevalent in sheep than is E. coli O157:H7 (Blanco et al., 2003; Aktan et al., 2004), although most of these non-O157 strains cannot be regarded as EHEC in the absence of evidence of their virulence in humans.

Attaching–effacing E. coli (AEEC) of various serogroups have been associated with enteric disease in goats (Tominaga et al., 1989; Duhamel et al., 1992; Drolet et al., 1994; Barlow et al., 2004; Wales et al., 2005b). In addition, goats (as with other ruminants) can be subclinical carriers and excretors of E. coli O157:H7. Several recent reports have clearly identified (Bielaszewska et al., 1997; Steen et al., 2001; McIntyre et al., 2002; Espie et al., 2006) or implicated (Shukla et al., 1995; Chapman et al., 2000; Pritchard et al., 2000; Payne et al., 2003; Rey et al., 2006) goats as sources of E. coli O157:H7 infection. Not only can goats be colonized with E. coli O157:H7, but their innately inquisitive behaviour means that they are much more likely than sheep to be in regular direct contact with humans, consequently increasing the risk of the direct faecal–oral transmission of zoonotic infection. There are little documented data on the prevalence of E. coli O157:H7 in goats, but studies have indicated a similar (Keen et al., 2006; Fox et al., 2007) or a lower (Cortes et al., 2005) prevalence to sheep.

Escherichia coli O157:H7 colonization factors

The ability to induce AE lesions has been studied extensively, and will be described in more detail below, and it is this phenotype that seems to be the primary contributor to colonization in all ruminant species including sheep and goats (Woodward et al., 2003; La Ragione et al., 2005a, 2006). For all AEEC, there are other factors such as flagella and fimbriae that also seem to play roles in modulating colonization. The clinical outcome of infection in humans and animals seems to be directly related to the immune status of the host, and host susceptibility to the effects of colonization and of Stx, which is discussed below.

Escherichia coli O157:H7 is able to induce AE lesions in the alimentary tract of humans and animals, and isolates are typically positive for one or more of the Stx subtypes, and therefore are classified as EHEC. AEEC that do not elaborate Stx but that are associated with diarrhoeal disease are generally referred to as enteropathogenic E. coli (EPEC).

The AE lesion

The AE lesion (Fig. 1) is characterized by intimate adherence between the bacterium and the host epithelial cell membrane, with an intervening gap of about 10 nm, plus effacement of enterocyte microvilli. Beneath the adherent bacterium, a cytoskeletal rearrangement, including the accumulation of filamentous actin (F-actin), is seen. The bacteria often sit upon a pedestal-like structure, which can extend up to 10 μm away from the epithelial cell surface (Kaper et al., 1998). The lesion is typically described as forming in three stages: initial nonintimate adhesion (mediated by elements that appear to vary between bacterial strains and AEEC types) is followed by signal transduction (leading to cytoskeletal reorganization and microvillus effacement) and finally intimate attachment. Initial adhesion in is poorly understood for E. coli O157:H7: it may involve fimbrial organelles in vivo, as discussed later, but these have not been consistently identified or characterized, unlike the bundle-forming pili (BFP) that mediate the initial attachment of typical EPEC. The EspA filaments of the secretion apparatus (discussed below) appear to have a nonintimate adhesin role.

Figure 1.

 AE lesion. Transmission electron micrograph. B, attached Escherichia coli bacterium; P, host cell pedestal; M, host cell microvillus.

Signal transduction and cytoskeletal reorganization is effected using a type-III secretion apparatus, whose elements are chromosomally encoded in the locus of enterocyte effacement (LEE) pathogenicity island. Esp (‘EPEC-secreted proteins’) A, B and D are thought to create a bifunctional adhesin and a pore-forming channel, through which proteins, including the translocated intimin receptor Tir, are translocated into a host enterocyte. Tir ultimately localizes to the host cell membrane, where it functions as a receptor for a bacterial surface adhesin (intimin) in the formation of an intimate attachment. In AE lesions formed by E. coli O157:H7, the recruitment of host cytoskeletal elements (F-actin and α-actinin) appears to be dependent on the Tir cytoskeleton coupling protein (TccP), another effector protein that is translocated into host cells (Campellone et al., 2004a, b; Garmendia et al., 2004; Allen-Vercoe et al., 2006). The role(s) of the cytoskeletal rearrangements remain speculative, but firm anchorage of the bacteria to the host cells is an obvious possibility. Although EHEC and EPEC AE lesions are morphologically very similar, significant differences in the respective mechanisms of formation are emerging, with EPEC Tir appearing to engage host cell elements without requiring TccP (Goosney et al., 2000; Gruenheid et al., 2001; Campellone et al., 2004a; Allen-Vercoe et al., 2006), although some EPEC do encode a TccP variant (Ooka et al., 2007). Fuller descriptions of the process have been given elsewhere, for example by Wales et al. (2005c).

The first illustration of AE lesions in small ruminants was in sheep, and the lesions were observed in the small and large intestines of experimental lambs (Angus et al., 1982). Thereafter, two natural cases with AE lesions were reported in diseased neonatal lambs (Janke et al., 1989). AE lesions produced by untyped naturally acquired bacteria were also observed on the ileal and large intestinal mucosa of symptomless neonatal lambs (Wales et al., 2005b).

The LEE

A number of enteric pathogens, including EPEC, EHEC, Citrobacter rodentium, Hafnia alvei and rabbit EPEC (RDEC), may encode the LEE and so have the ability to induce AE lesions in the host intestinal tract (Agin et al., 1996). Intimin is encoded by eae in the LEE of E. coli O157:H7; it is essential for AE lesion formation and has been cited as one bacterial factor for colonization of the gastrointestinal tract (Donnenberg et al., 1993; Dean-Nystrom et al., 1999; Woodward et al., 2003). Experimental studies have used various E. coli O157:H7 strains in oral inoculation studies in conventionally reared sheep to define colonization and persistence patterns. Studies by Cornick et al. (2002) and Woodward et al. (2003) demonstrated that intimin contributes significantly to the colonization and persistence of E. coli O157:H7 in sheep. In both studies, intimin-deficient isogenic mutants showed reduced numbers excreted: around a 1 log unit difference at 3 days postinoculation and in excess of a 2 log unit difference by 15 days. Dean-Nystrom et al. (1999) reported that weaned calves inoculated with wild-type E. coli O157:H7 had a considerably greater density of the organism in the large intestine than did calves inoculated with an isogenic intimin-deficient mutant (106.6 vs. <103 CFU g−1). Cornick et al. (2002) showed that an intimin-deficient E. coli O157:H7 mutant was excreted in lower numbers and for a shorter period from inoculated yearling cattle and young adult sheep than was its coinoculated wild-type parent strain: there was a >1 log unit difference in numbers excreted up to 15 days postinoculation, and by 30 days most animals had stopped excreting mutants but were still excreting the wild type. However, similar studies conducted in goats (La Ragione et al., 2005a) indicated a lesser role for intimin in colonization and persistence, as an isogenic eae mutant was not attenuated for persistence over a period of 29 days (no significant difference in excreted numbers from 14 weaned kids inoculated with either wild-type or intimin-deficient E. coli O157:H7).

Tir, the LEE-encoded intimin receptor, was examined in vivo by Vlisidou et al. (2006), who concluded that it was essential for the colonization of both lambs and calves and that colonization is mediated by Tir serving as a primary receptor for intimin. This report also concluded that tir mutants were more attenuated than eae (intimin) mutants; for example in lambs the tir mutant was undetectable by 3 days postinoculation whereas the eae mutant persisted to day 16, implying that Tir may facilitate intestinal colonization in other ways also. Stevens et al. (2004) have also reported a significant reduction in the colonizing ability of a tir mutant in calves. Recent studies by Bretschneider et al. (2007) have described a similar reduced colonization phenotype for Tir-deficient mutants in yearling beef cattle, although comodulation of flagellin expression in these particular mutants complicates interpretation in this case.

The role of translocator and effector proteins

Numerous secreted proteins have been described in AEEC including in E. coli O157:H7. The LEE encodes the translocators EspA, EspB and EspD, plus characterized or putative effectors EspF, EspG, EspH, EspZ, Tir and Map. Certain non-LEE-encoded (Nle) proteins are also secreted by the type-III apparatus, including cycle-inhibiting factor (Cif), TccP, EspJ, EspG2 and Nle proteins A, C and D. Deletion of the LEE4 operon, abolishing or severely curtailing the type-III apparatus and delivery of proteins secreted by this route, prevented the colonization of calves by E. coli O157:H7 (Naylor et al., 2005).

In addition to Tir (discussed above), EspA and TccP have been investigated individually in E. coli O157:H7. Experimental mutation of espA strongly reduced the colonization of calves (Dziva et al., 2007), but tccP mutation did not impair the colonization of calves or lambs (Vlisidou et al., 2006). The tccP mutant was unable to nucleate F-actin on human HeLa cells in vitro, although it adhered well. However, in vivo, in calves, it was able to induce AE lesions with typical pedestals, suggesting that EHEC pedestal formation in vivo may differ from that in vitro: possible reasons include host cell differences or effector molecules being available from the surrounding flora in vivo. Using an alternative approach, immunization of cattle by injection of an adjuvanted mix of Esp proteins, with or without Tir, significantly reduced the shedding of E. coli O157:H7 among orally dosed calves and naturally exposed adults (Potter et al., 2004).

EspJ inhibits macrophage phagocytosis and is not required for AE lesion formation (Marchès et al., 2008). Dahan et al. (2005) showed that espJ mutation actually enhanced the persistence of both a nontoxigenic E. coli O157:H7 strain and a C. rodentium strain in conventionally reared 6-week-old lambs and in mice, respectively. This was manifested as three of five espJ mutant-inoculated lambs vs. one of five wild-type-inoculated lambs excreting E. coli O157:H7 at 9 days postinoculation, and four of five espJ mutant-inoculated mice vs. none of five wild-type-inoculated mice having inoculum in the caecum at 20 days postinoculation. Thus, EspJ appears to have properties that might enhance virulence (by inhibiting an aspect of the host immune response), but, based on the limited evidence available, EspJ may also limit persistence.

Cif is encoded on a lambdoid phage and many EHEC and EPEC strains carry functional or mutated forms of cif. It triggers an irreversible cytopathic effect in HeLa cells, which is characterized by the progressive recruitment of focal adhesions, the assembly of stress fibres and arrest of the cell cycle (Nougayrede et al., 2001; Marchès et al., 2003). It has been hypothesized (Marchès et al., 2003) that Cif-dependent arrest of the host cell cycle in a population with rapid turnover, such as enterocytes, may aid colonization and persistence, although studies examining this have not yet been reported.

EspI/NleA is encoded within prophage CP933P on the E. coli O157:H7 genome; it localizes to the host cell Golgi apparatus and inhibits secretion mechanisms (Gruenheid et al., 2004; Kim et al., 2007). The nleA (espI) gene has been found widely in EHEC (O157 and non-O157) and EPEC, and appears to be associated with virulence in humans (Mundy et al., 2004a; Creuzburg & Schmidt, 2007). In C. rodentium, it is required for full virulence in susceptible mice and for the normal colonization of resistant mice (Gruenheid et al., 2004; Mundy et al., 2004b). NleC and NleD were examined by Marchès et al. (2005): a wild-type E. coli O157:H7 strain and isogenic mutants in nleC and nleD were separately orally inoculated into groups of lambs, and all strains were excreted at the same density in faeces for the subsequent 7 days. It was concluded that neither protein was required for the colonization and/or persistence of E. coli O157:H7 in a 6-week-old conventional lamb model.

The products of the plasmid-borne toxB and the truncated chromosomal efa-1′ genes appear to enhance LEE-dependent secretion in E. coli O157:H7 and adherence to host cells in vitro, without having obvious effects on its persistence when tested by monitoring of the levels of excretion following oral inoculation of parent and isogenic mutant strains into sheep and calves (Stevens et al., 2004).

Thus, among the substantial array of E. coli O157:H7 translocator or type-III apparatus-secreted proteins, only Tir and EspA have been shown individually to have major roles in colonization and persistence in ruminants, and of these only Tir has been examined in sheep. A few others have been found not to enhance colonization, but there are very limited data available on many of the proteins, most of which have not been examined until very recently. Therefore, their potential contributions to differences between E. coli O157:H7 colonization of different hosts, or to tissue tropisms, remain unknown. For those that are apparently not important in the context of ruminant colonization and persistence, it may be that the bacterium is able to compensate for the lack of a single effector and/or a translocator protein or is able to harness available protein from AEEC in the natural flora, for example at sites of pre-existing AE lesions. In addition, it is possible that multiple knockout mutants are required to observe significant effects, as many of these translocator and effector proteins probably work in a synergistic manner. The issue of redundancy also needs to be considered for some effectors as, for example, EspG2 is functionally similar to the LEE-encoded EspG but is encoded elsewhere on the EPEC chromosome (Elliott et al., 2001; Shaw et al., 2005).

The role of toxins in colonization and persistence

One of the major virulence factors of E. coli O157:H7 are the Shiga (Vero) toxins, which occur as two major subtypes (Stx1 and Stx2) and are encoded by lysogenic bacteriophages. The preferred receptor for the toxin is globotriaosylceramide (Gb3). In humans, intestinal elaboration of Stx is associated with a microvascular angiopathy in the intestine and other critical organs, including the kidney and central nervous system (CNS), which accounts for both the typical bloody diarrhoea and the frequently severe systemic manifestations of EHEC infection (Richardson et al., 1988; Baker et al., 2007). Piglets express Gb3 receptors in renal and CNS vasculature and show corresponding angiopathies on exposure to Stx (Gunzer et al., 2002), but cattle lack vascular Gb3 receptors in those organs principally affected in humans (Pruimboom-Brees et al., 2000; Hoey et al., 2002), which may explain why ruminants are symptomless carriers of E. coli O157:H7.

Studies to date have shown that non-Shigatoxigenic E. coli O157:H7 can colonize cattle, sheep, goats and chickens with no clinical manifestations, even though AE lesions may form in the gastrointestinal tract (Woodward et al., 2003; Best et al., 2005; La Ragione et al., 2005a, 2006). It has been postulated that Stx may play a role in the ability of E. coli O157:H7 to persist in ruminants, and in vitro studies have pointed to this, showing an inhibitory effect of Stx1 on bovine lymphocytes (Menge et al., 2003). However, recent studies by Cornick et al. (2007) have indicated that nontoxigenic E. coli O157:H7 strains, created from a wild type either by a deletion mutation of stx2 or by curing the parent of its stx phage, compete effectively with their toxigenic progenitor strain in the colonization of sheep. Stevens et al. (2002) concluded that the elaboration of Stx1 by EHEC O103:H2 did not influence enteric inflammatory responses (fluid accumulation and recruitment of radio-labelled neutrophils) in a bovine ligated ileal loop model. Another toxin associated with E. coli O157:H7 is enterohaemolysin, but this has not been investigated in relation to persistence of the bacterium.

The role of fimbriae and lipopolysaccharide in colonization and persistence

Fimbriae have been shown to be important in the pathogenesis of a number of E. coli pathotypes (La Ragione et al., 2000; Lane & Mobley, 2007), and recent studies have revealed the presence of at least 16 fimbrial gene clusters in E. coli O157:H7. Many of these fimbriae do not appear to be expressed in vitro, and in particular type 1 fimbrial expression is blocked by a 16 bp deletion in the ‘fim switch’ regulator (Iida et al., 2001; Roe et al., 2001).

Long polar fimbriae (LPF), originally found to be important for the pathogenesis of Salmonella Typhimurium (Bäumler et al., 1996), represent one potential adherence determinant in E. coli O157:H7, which contains two nonidentical lpf loci homologous to lpf of S. Typhimurium (Perna et al., 2001). Torres et al. (2002) showed that the lpf1 operon increases adherence of E. coli K12 to cultured epithelial cells in vitro and was associated with long peritrichous fimbriae. Furthermore, mutation of both these lpf loci together in E. coli O157:H7 reduced the numbers recovered from orally inoculated pigs and sheep (Jordan et al., 2004), an effect most pronounced in the first 2 weeks of the 8-week study. Mutation of either lpf locus altered the tropism of E. coli O157:H7 adhesion (with AE lesion formation) on human intestinal explants, paradoxically expanding the range of adhesion sites from ileal follicle-associated epithelium (wild-type tropism) to include other small intestinal epithelium (Fitzhenry et al., 2006). More recently, Torres et al. (2007) showed that mutation of both lpf loci (but not either singly) reduced the ability of E. coli O157:H7 to persist in the intestine of orally infected conventionally reared 6-week-old lambs. Interestingly, the same study showed lpf mutants to be better colonizers of intestinal mucosal explants than wild-type strains, suggesting that the LPF role in persistence may not involve initial adherence, contrary to what might be assumed.

Under certain culture conditions, type IV pili (Xicohtencati-Cortes et al., 2007) and ‘F9’ fimbriae (Low et al., 2006) can be observed and extracted from E. coli O157:H7 strains. The type IV pili appear to aid adhesion of E. coli O157:H7 to human, cattle and pig gut explants, but in vivo studies are currently lacking. Fimbrial mutant studies in calves have not identified a clear role in colonization for the F9 fimbriae. Sorbitol-fermenting EHEC O157:H strains have plasmid-encoded fimbriae that are only expressed under anaerobic conditions, and that enhance adhesion to human intestinal epithelial cell lines (Müsken et al., 2008). For other AEEC subtypes such as EPEC, the role of fimbriae is clearer. In particular, BFP have been shown to be important in initial attachment (Knutton et al., 1987).

In vivo studies have concluded that mutants lacking the O157 lipopolysaccharide (gal knockouts) were uniformly attenuated in an infant rabbit intestinal colonization model, as might be anticipated (Ho & Waldor, 2007).

The role of flagella in colonization and persistence

Motility processes in Gram-negative bacteria are complex and require the co-ordinated transcription of >40 genes in 14 operons (Iino et al., 1988; Liu & Matsumura, 1994). Flagellum-driven motility has been shown to be associated with adherence to epithelial cells by many diverse E. coli pathotypes and other pathogenic bacteria (Feldman et al., 1998; Allen-Vercoe & Woodward, 1999; La Ragione et al., 2000; Tasteyre et al., 2001; Giron et al., 2002; Inglis et al., 2003; Dons et al., 2004; Kirov et al., 2004; Wright et al., 2005). Flagella, including H7, have been recognized as potent inflammatory ligands, inducing inflammatory responses through engagement with a number of receptors (Berin et al., 2002; Zhou et al., 2003). Studies have shown that quorum-sensing pathways positively regulate the expression of E. coli O157:H7 flagella (Sperandio et al., 2001, 2002), in addition to regulating other virulence factors including the type III secretion system (Sperandio et al., 1999). However, sorbitol-fermenting E. coli O157:H are almost always nonmotile, but are still able to cause clinical disease in humans (Monday et al., 2004), indicating that flagella are not essential for EHEC virulence in the human host. In experimentally infected pigs, a flagella-deficient E. coli O157 mutant persisted as well as its isogenic progenitor O157:H7 strain (Best et al., 2006) but, by contrast, in a surrogate specific pathogen free (SPF) chick model of EHEC colonization (Best et al., 2005), aflagellate E. coli O157 mutants were significantly less persistent than their wild-type progenitor strain. Recent studies conducted by La Ragione et al. (2005a) demonstrated, using a flagellin gene (fliC) knockout mutant, that flagella did not contribute to the long-term persistence of E. coli O157:H7 in goats, and indeed isogenic fliC mutants showed a tendency to be excreted for longer than intact parent strains. Interestingly, cattle studies have shown that an intact flagellar regulatory gene (flhC) contributes to effective host colonization but, similar to ovine studies, the fliC gene responsible for filament formation does not appear to be important in this context (Dobbin et al., 2006). In addition, the same study reports that aflagellate (ΔfliC) E. coli O157:H7 are able to survive passage through the bovine gastrointestinal tract better than flagellated strains. Data reported by Bretschneider et al. (2007) apparently show that E. coli O157 deficient in the expression of flagella were attenuated in their capabilities of colonizing experimentally infected beef cattle, although the aflagellate state in this case was an uncontrolled and undefined byproduct of genetic manipulations of tir.

Erdem et al. (2007) recently demonstrated that flagella from EHEC and EPEC may mediate adherence to mucus in the intestinal tract, whereas McNeilly et al. (2007) showed that mucosal samples from E. coli O157:H7-exposed cattle contained antibodies to H7. Therefore, potential benefits to E. coli O157:H7 colonization by flagellar motility or adhesion might be counteracted in the longer term by mucosal immune responses elicited by flagellin.

Escherichia coli O157:H7 and tissue tropisms

Tissue tropism has been identified as important in the pathogenesis of E. coli O157:H7. Naylor et al. (2003) reported that nine of 10 experimentally inoculated weaned calves excreting E. coli O157:H7 for a minimum of 2 weeks, plus one naturally infected steer, showed a significantly increased density of E. coli O157 in the region of the recto-anal junction (RAJ). Evidence of adherent E. coli O157 bacteria in this area was also presented. A Stx-negative E. coli O157:H7 strain was subsequently shown to demonstrate LEE-dependent persistence and to form AE lesions at the RAJ in experimentally inoculated calves (Naylor et al., 2005). The same group (Low et al., 2005) showed, in an abattoir survey, a similar pattern of increasing density of E. coli O157:H7 with increasing proximity to the RAJ on the bovine rectal mucosa, and also an association between E. coli O157 density at the RAJ and in the faeces. A different group of workers (Rice et al., 2003) provided corroborative evidence of the importance of the RAJ in bovine colonization by showing that swabs of the RAJ in experimentally and naturally colonized cattle provided more sensitive detection of E. coli O157:H7 than did faecal samples. The same group (Sheng et al., 2004) showed that rectal inoculation of calves with a sponge soaked in ≥107 CFU E. coli O157:H7 resulted in persistent excretion, with colonization that could only be detected at the rectum. By contrast, an oral inoculum of 1010 CFU of the same strain achieved a pattern of persistent excretion over 8 weeks that was similar to that seen with a rectal inoculum of 107 CFU and significantly less, in terms of the percentage of animals excreting, than that seen with a rectal inoculum of 1010 CFU. Recent studies by Bonardi et al. (2007) have suggested that E. coli O157 may actually enter into bovine lymphatic tissues such as the mesenteric lymph nodes and tonsils.

The studies conducted by La Ragione et al. (2005a, 2006) do not suggest that the RAJ is a site of primary importance leading to persistent colonization and shedding of E. coli O157:H7 in sheep and goats. These studies concluded that E. coli O157:H7 colonized the distal intestine including the caecum, colon and rectum, suggesting that this area of the gastrointestinal tract is the preferred site of colonization in these species. Detection of AE lesions was rare, with only small and sparse lesions seen, even though the challenge strain was able to induce diffuse lesions in vitro to cultured epithelial cells and to adhere efficiently to ovine in vitro organ culture (Dibb-Fuller et al., 2001; Torres et al., 2007). The same studies (La Ragione et al., 2005a, 2006) showed no specific tropism of E. coli O157:H7 towards lymphoid tissue on microscopic examination, albeit pathological examinations were only performed up to 4 days postinoculation. Cookson et al. (2002b) showed a positive correlation for E. coli O157:H7 between the numbers consistently excreted and the extent of its distribution in the gastrointestinal tract over a time scale of up to 3 weeks after oral inoculation. More recent studies in sheep (Woodward et al., 2003) include the observation that animals colonized beyond c. 14 days postinfection (and therefore considered to be persistent shedders) had E. coli O157 organisms throughout the entire gastrointestinal tract, rather than just the large intestine.

Collectively, these findings suggest that in small ruminants, colonization of the gastrointestinal tract may occur throughout the distal region and be diffuse, not preferentially targeted to the RAJ as in cattle. However, to confound this general conclusion, Grauke et al. (2002) found that persistently excreting experimentally infected sheep showed very rare evidence of E. coli O157:H7 in the intestine to the level of the descending colon. Moreover, either rectal or oral inoculation with E. coli O157:H7 leads to similar faecal shedding in lambs (Best et al., 2008), and where rectally inoculated lambs were positive for E. coli O157:H7 at necropsy, the majority of bacteria were associated with the recto-anal mucosa. This may not be surprising, given that the inocula were directed to the rectum and previous data indicated that the entire distal gastrointestinal tract is susceptible to colonization. To suggest that this site may indeed be a site of preferential colonization is perhaps erroneous and an artefact of the mode of administration. Additionally, further analysis by confocal microscopy revealed that only a small subset of rectally inoculated lambs had large, densely packed E. coli O157:H7 microcolonies identified by specific staining. What these data do confirm is that E. coli O157:H7 can effectively colonize the very distal region of the gastrointestinal tract. Interestingly, in rectally inoculated lambs, E. coli O157:H7 was not as frequently recovered from other gastrointestinal tract tissues when compared with orally inoculated lambs. Given the route of administration, this also may not be surprising.

Oral inoculation studies with a nontoxigenic ovine-derived E. coli O26:K60 strain (Aktan et al., 2007) showed that the challenge organism could be recovered from all sites in the gastrointestinal tract during the high excretion phase, but from just the distal small intestine during long-term persistence, 38 days after inoculation. This illustrates that tissue tropisms probably vary among differing persistent AEEC, including those in common EHEC and EPEC serogroups. Indeed, in the several small ruminant E. coli O157:H7 infection studies reported, of particular interest is the frequent occurrence of AE lesions caused by non-O157 bacteria, some of which have been identified as belonging to O26 and O115 serogroups (Cookson et al., 2002a; Wales et al., 2005a, b). Sheep appear to be reservoirs for a diverse range of AEEC with varying serotypes, virulence profiles and intimin types (Aktan et al., 2004; Cookson et al., 2007), and it may be hypothesized that different intimin types facilitate tropism to different specific sites in the gastrointestinal tract, as described for cattle and humans (Fitzhenry et al., 2002; Mundy et al., 2007), or perhaps that some intimin subtypes play no role in the colonization of AEEC in sheep.

Effects of host factors on E. coli O157:H7 colonization and persistence

General health, hormonal and immune status

Recently, interest has developed in host factors relating to the virulence and persistence of EHEC and EPEC, particularly with the advent of genome sequences and the widespread use of microbial and host arrays. It remains unclear at present how stress and hormonal stimulation may influence colonization in any species. One reasonably well-described mechanism by which host stress can alter the behaviour of E. coli O157:H7 is the influence of the catecholamine hormones epinephrine and norepinephrine on the ‘quorum-sensing’ bacterial system, by which virulence mechanisms can be upregulated (Sperandio et al., 2003) and enteropathic effects such as adhesion and inflammation can be enhanced (Vlisidou et al., 2004).

Several researchers have considered the immune status of the host. It is well established that colostrum deprivation is a risk factor for increased susceptibility to gastrointestinal pathogens (Kelleher & Lonnerdal, 2001) and that colostrum is protective against pathogenic E. coli (Logan et al., 1974; Altmann & Mukkur, 1983). It has also been demonstrated in neonatal calves that colostrum is highly protective against E. coli O157:H7 challenge and that AE lesions are readily induced in the distal gastrointestinal tract if colostrum is withheld (Dean-Nystrom et al., 1997; Rugbjerg et al., 2003). La Ragione et al. (2005b, 2006) showed that colostrum deprivation increased susceptibility to E. coli O157:H7 colonization in young lambs and that colostrum was protective for sucking goat kids in the face of a modest challenge via the nanny, with no evidence that intramammary infection was significant.

The general health status of the ovine or caprine host may have a major influence on persistence and high-level shedding of E. coli O157:H7. However, there is at present a lack of information in this area, including the influence of concurrent infection status, although some work (reported below) reveals the effect of cryptosporidiosis. Survey data looking at a range of infective agents may be a useful approach in the future. There is also a lack of hard data on transmission routes in farmed animal species, although experimentally and epidemiologically the concept of ‘supershedders’ being important in the maintenance of herd status has been developed in relation to E. coli O157:H7 in cattle (Cobbold et al., 2007).

The possible role of concurrent colonization by other AEEC or pathogens such as Cryptosporidium

Competition between bacteria and the dominance of certain E. coli serotypes or strains is suggested to be one explanation for the shedding of one predominant E. coli strain over time (Midgley et al., 1999). La Ragione et al. (2004) reported on the interactions between E. coli O26:K60 and E. coli O157:H7 in tissue culture adherence assays, which showed that preincubation of tissue culture cells with either strain reduced significantly the extent of adherence of the strain that was applied second. Aktan et al. (2007) reported that atypical EPEC O26:K60 colonized 6-week-old conventionally reared lambs after oral inoculation, with persistent shedding for well over a month and with the induction of AE lesions that were small and sparse in the distal gastrointestinal tract. In the field, more than one group has shown E. coli serogroup O26 to be prevalent in ruminant animals at slaughter when the prevalence of E. coli O157:H7 in the same study animals was very low (Aktan et al., 2004; Fukushima & Seki, 2004).

Taken together, these data suggest that the colonization and shedding of lambs by E. coli O157:H7 may be altered, and possibly reduced, if the lambs are previously colonized by EPEC O26:K60. However, further experimental studies (I. Aktan, R.M. La Ragione & M.J. Woodward, unpublished data) revealed that prior infection with E. coli O26:K60 did not alter the colonization or persistence of subsequent E. coli O157:H7 infections. Interestingly, when the order of administration was reversed, prior experimental colonization of conventionally reared lambs by E. coli O157:H7 did significantly suppress E. coli O26:K60 colonization and persistence, an interaction that requires further investigation.

Cryptosporidium parvum has been shown to be a common agent associated with diarrhoeal disease in young calves and lambs in Europe, often as part of mixed enteric infections (Munoz et al., 1996; de la Fuente et al., 1999). La Ragione et al. (2005a) showed that an experimental conventionally reared goat kid shedding especially high numbers of E. coli O157:H7 was also heavily infected with Cryptosporidium. Further studies (La Ragione et al., 2006) revealed that lambs experimentally preinoculated with C. parvum before challenge with E. coli O157:H7 shed very high numbers of the E. coli challenge strain and developed extensive, multifocal AE lesions in the caecum, colon, rectum and at the RAJ (Fig. 2). Lesions were confirmed by immunohistochemistry to be associated with E. coli O157.

Figure 2.

 AE lesions formed by Escherichia coli O157:H7 at the RAJ of a lamb coinfected with Cryptosporidium parvum. Transmission electron micrograph. CP, C. parvum cell attached to host mucosa; EC, E. coli O157:H7 intimately attached to host mucosa; RM, host rectal mucosal cell.

Thus, concurrent infection with an unrelated pathogen may enhance colonization by E. coli O157:H7 and its excretion. Indeed, it is also possible that viral infections may play a role in predisposing ruminants to AEEC, including E. coli O157:H7. There have been several reports where rotavirus has been isolated from animals presenting with clinical signs associated with E. coli O26 (Acres et al., 1977; Janke et al., 1990). In addition, it may be hypothesized that immunosuppressive viruses, for example bovine immunodeficiency virus or bovine viral diarrhoea virus in cattle, and border disease virus or maedi-visna virus in sheep, could influence AEEC colonization and persistence.

Summary and concluding remarks

Numerous putative colonization factors have been identified and characterized in E. coli O157:H7, and recent publication of AEEC genomes and comparative genomics (Hayashi et al., 2001; Perna et al., 2001; Zhang et al., 2007) has enabled the analysis of the genome for specific genes that may contribute to persistent colonization of ruminants.

Studies to date include the characterization of defined targeted mutants constructed in eae, tir, efa/toxB, fliC, lpf, nleD and tccP, made primarily in the non-Shigatoxigenic E. coli O157:H7 strain NCTC12900. The data compiled to date point to a role for intimin and Tir in sheep and cattle (Woodward et al., 2003; Vlisidou et al., 2006) although, interestingly, the colonization of goats seems less dependent on intimin. Indeed, the studies by La Ragione et al. (2005a, 2006) and Woodward et al. (2003) suggest that the long-term persistence of E. coli O157:H7 in small ruminants may be significantly affected by bacterial factors in addition to intimin and Tir, as well as by host factors such as adequacy of colostrum intake in the young or prior colonization by other pathogens, or possibly other elements of the commensal flora. Other than certain LEE-encoded factors (intimin, Tir and EspA), few of the other candidate bacterial factors thus far examined have a significant effect on colonization and/or persistence when tested in experimental ruminant (sheep and cattle) models of colonization (Woodward et al., 2003; Stevens et al., 2004; Marchès et al., 2005; Vlisidou et al., 2006; Torres et al., 2007).

It is clear that E. coli O157:H7 is prevalent in small ruminants (sheep and goats) and that groups of these animals may act as effective reservoirs for this important zoonotic pathogen. Much has been revealed about the pathobiology of E. coli O157:H7 in cattle, including an LEE-dependent tropism for lymphoid tissue at the terminal rectum. This precise tropism has not been observed in small ruminants, but it is evident that colonization and persistence of sheep and goats is influenced by the LEE-encoded secretion system. Localization of E. coli O157:H7 has been observed at the ovine recto-anal mucosa, albeit intermittently. In both cattle and small ruminants, certain other secreted proteins, surface appendages, age and stress effects are suspected to influence persistence, based on in vitro studies and limited numbers of in vivo studies in laboratory and ruminant species. There are also, for small ruminants, some recent data showing that colonization is enhanced in young animals with compromised passive immunity or those experiencing coinfection with an unrelated pathogen. What is less well understood for all ruminant reservoirs is the relationship between E. coli O157:H7 and the normal gastrointestinal flora, and how this might be modulated for the development of control strategies.

Research in the field of ruminant persistence of EHEC has so far identified some important elements that may be susceptible to practical interventions for control of E. coli O157:H7. A more comprehensive understanding of the LEE- and non-LEE-encoded factors that promote effective colonization in the distal ovine gastrointestinal tract is required in order to develop and assess the appropriateness of intervention strategies for the control of E. coli O157:H7 in the field. However, unless the same experiments are performed with the same strains and mutants in different animal species, we will struggle to arrive at truly solid conclusions about differing behaviours that are host specific. It seems likely that these insights will be needed to provide more comprehensive control of these subtle pathogens.

Acknowledgements

The authors would like to thank W.A. Cooley for electron microscopy, and the UK Department for Environment, Food and Rural Affairs (DEFRA) for funding some of the work reviewed in this manuscript (Grants OZ0706, OZ0710 and OZ0713).

Ancillary