Epidemiology and Clinical Manifestations of Enteroaggregative Escherichia coli

Abstract

SUMMARY

Enteroaggregative Escherichia coli (EAEC) represents a heterogeneous group of E. coli strains. The pathogenicity and clinical relevance of these bacteria are still controversial. In this review, we describe the clinical significance of EAEC regarding patterns of infection in humans, transmission, reservoirs, and symptoms. Manifestations associated with EAEC infection include watery diarrhea, mucoid diarrhea, low-grade fever, nausea, tenesmus, and borborygmi. In early studies, EAEC was considered to be an opportunistic pathogen associated with diarrhea in HIV patients and in malnourished children in developing countries. In recent studies, associations with traveler's diarrhea, the occurrence of diarrhea cases in industrialized countries, and outbreaks of diarrhea in Europe and Asia have been reported. In the spring of 2011, a large outbreak of hemolytic-uremic syndrome (HUS) and hemorrhagic colitis occurred in Germany due to an EAEC O104:H4 strain, causing 54 deaths and 855 cases of HUS. This strain produces the potent Shiga toxin along with the aggregative fimbriae. An outbreak of urinary tract infection associated with EAEC in Copenhagen, Denmark, occurred in 1991; this involved extensive production of biofilm, an important characteristic of the pathogenicity of EAEC. However, the heterogeneity of EAEC continues to complicate diagnostics and also our understanding of pathogenicity.

INTRODUCTION

Diarrhea is still an important disease burden worldwide. It causes considerable childhood mortality in the developing world and is associated with morbidity and substantial health care costs in industrialized countries (1). One important cause of infectious diarrhea is the so-called diarrheagenic Escherichia coli (DEC) group (1, 2). The following subgroups of DEC have been defined: enterotoxigenic E. coli (ETEC), Shiga toxin-producing E. coli (STEC), enteroinvasive E. coli (EIEC), enteropathogenic E. coli (EPEC), and enteroaggregative E. coli (EAEC) (3–5). Recently, the pathogenicity of these pathotypes, including EAEC, was reviewed (5); in this review, we focus on the epidemiology and disease manifestations of EAEC specifically. EAEC was first described in 1987 by Nataro et al. in a study examining different patterns of adherence of E. coli strains to HEp-2 cells in culture. The strains were collected from Chilean children with diarrhea, and the typical “stacked-brick” pattern was observed (6), which continues to define EAEC (Fig. 1). The aggregative adherence (AA) pattern is defined as the binding of bacteria to epithelial cells in a stacked-brick manner. Although EAEC has been associated with diarrhea in studies conducted in both developing and industrialized countries, it has been difficult to determine the specific mechanisms of EAEC pathogenicity, which has made assessments of the clinical relevance of this microorganism difficult. Here, we present the literature on this subject, concentrating on the pathophysiology of EAEC.

FIG 1.

Characteristic stacked-brick pattern observed when EAEC is cultured on HEp-2 cells. This pattern is also known as a honeycomb formation, which emerges due to bacterium-bacterium and bacterium-cell interactions. The result is the aggregation of the bacteria in a stacked-brick manner. (Courtesy of Rie Jønsson, Department of Microbiology and Infection Control, Statens Serum Institut, Copenhagen, Denmark; reprinted with permission.)

PATHOGENICITY OF EAEC

Regarding infection with any of the subtypes of EAEC strains, the following stages of pathogenesis have been described: (i) initial adherence to the mucosal surface, (ii) biofilm formation, and (iii) induction of an inflammatory response and release of toxins (7–12) (Fig. 2). Knowledge of the stages of EAEC pathogenicity has been obtained from studies of in vitro cell cultures, animal models, and patients infected with EAEC (10, 13–17).

FIG 2.

Stages of pathogenesis of EAEC. Numbers in circles show the progression of EAEC pathogenesis. (1) Agglutination of planktonic EAEC bacteria. (2) Adherence to the intestinal epithelium and colonization of the gut. (3) Formation of biofilm. (4) Release of bacterial toxins, inducing damage to the epithelium and increased secretion. (5) Establishment of additional biofilm.

Adherence

Adhesion to the intestinal epithelium is facilitated by fimbriae and is the first step in the bacterial colonization of the gut (10, 18, 19) (Table 1). The aggregative adherence fimbriae (AAF) have been found to include 4 major variants, with distinct structures of the pilin subunits (AAF/I to AAF/IV) (19–22). The fimbriae can splay out from the bacteria due to the surface protein dispersin (9, 23), which is encoded by the aap gene. Dispersin is believed to induce changes in the electrostatic surface of the lipopolysaccharide layer of the bacteria. This makes this protein important for the adherence properties of EAEC (9). The fimbriae bind to components of the extracellular matrix of intestinal epithelial cells, such as laminin, collagen IV, cytokeratin 8, and fibronectin (24, 25). The aggregative pattern of these bacterial strains is thought to emerge from binding to the epithelial cell surface and binding to adjacent EAEC bacteria (26). Autoaggregation contributes to the AA pattern, where adherence to epithelial cells also involves interactions between bacteria (18). The plasmid-borne aggR gene is another important gene for the pathogenesis and adherence properties of EAEC, where strains possessing the aggR gene are known as “typical EAEC strains” (27). aggR is a transcriptional activator that promotes the expression of both chromosomal and plasmid-encoded virulence factors, including AAF and dispersin (28).

TABLE 1.

Key EAEC virulence genes^a

Virulence gene(s)	Function	Location	Role in pathogenesis
aggR	Transcriptional activator of virulence genes	Plasmid	Adherence
AAF/I–AAF/IV genes	Aggregative adherence fimbriae (4 variants)	Plasmid	Adherence
aap	Antiaggregation protein dispersin	Plasmid	Adherence
aatA	Dispersin transporter protein	Plasmid	Biofilm production
fis	Regulation of AAF expression	Chromosome	Biofilm production
shf	Encoding a Shigella flexneri homolog protein	Plasmid	Biofilm production
yafK	Regulation of AAF expression	Chromosome	Biofilm production
astA	Heat-stable enterotoxin EAST1	Plasmid	Toxins
pet	Plasmid-encoded toxin	Plasmid	Toxins
sepA	Shigella extracellular protein	Plasmid	Toxins
sat	Secreted autotransporter toxin	Chromosome	Toxins
set	Shigella enterotoxin 1	Chromosome	Toxins
pic	Protein involved in colonization	Chromosome	Toxins

^aThe EAEC virulence genes listed are involved in several stages of pathogenicity. The appointed stage of pathogenicity for each virulence gene is given for simplicity.

Biofilm Formation

When the gut has been colonized with EAEC, secretion of excessive mucus has been described, which is followed by the formation of biofilm (10, 29, 30). Biofilm formation is an important pathogenicity trait of EAEC and is situated mainly in the colon; however, formation of biofilm in the small intestine has also been reported (10). Biofilm formation may be an important contributory factor in persistent infection by allowing the bacteria to evade the local immune system and by preventing the transport of antibacterial factors, including antibiotics (31, 32). Assays to quantify biofilm formation have been suggested as a possible method of screening for pathogenic EAEC strains (8). In several studies, expression of AAF has been shown to be essential for biofilm formation by EAEC (20, 24, 33–35). Other factors involved in the formation of biofilm include the shf gene, which has proven to be important for solid-biofilm production in EAEC reference strain 042 (36). The genes yafK and fis are also important for biofilm formation, probably due to their involvement in the regulation of AAF expression (33). Furthermore, in epidemiological studies, the plasmid-borne aatA gene (37), encoding the dispersin transporter; the set1A gene (38); and the aggR gene (39) have been associated with biofilm formation.

Toxins

Once the biofilm has been established, further damage to the intestinal epithelium has been described, which is caused by the release of bacterial toxins. The secretion of toxins is thought to play an important role in secretory diarrhea, which is a typical clinical manifestation of EAEC infection (4, 9, 40). The cytotoxic effects of EAEC involve the secretion of serine protease autotransporters of the Enterobacteriaceae (SPATEs) (41). The SPATEs constitute a large family of extracellular proteases secreted by Enterobacteriaceae via the type V secretion system (28). The SPATE genes can be either chromosomal or plasmid borne, and they are organized into 2 phylogenetically different classes: class I SPATEs are cytotoxic to epithelial cells and include proteins encoded by the pet, sigA, and sat genes, whereas class II SPATEs have more diverse effects and include proteins encoded by the pic and Shigella extracellular protease (sepA) genes (42, 43). Pet (plasmid-encoded toxin) cleaves spectrin in the epithelial cytoskeleton (44), resulting in the deformation and exfoliation of the cell, and is associated with mucoid stools (45). Sat (secreted autotransporter toxin) has been shown to cause loosening of cellular tight junctions in kidney cells and vacuolation in both kidney cells and bladder cells (46). Pic (protein involved in intestinal colonization) is a mucinase that interferes with the integrity of the mucus membrane and induces serum resistance and hemagglutination (30, 47). The astA gene encodes EAST1 (EAEC heat-stable enterotoxin), which is not a SPATE and which shares certain functional properties of the enterotoxin (STa) secreted by ETEC (48). This toxin causes increased secretion of chloride and has been associated with secretory diarrhea (49, 50). However, EAST1 is not confined to EAEC strains, as it has also been detected in EPEC, ETEC, and EHEC strains (50). Furthermore, the astA gene cannot be detected in all EAEC strains (51–53). The ShET1 enterotoxin encoded by the set gene was first identified in Shigella flexneri and may be associated with increased fluid secretion (54). The toxin-induced damage observed in the intestinal epithelium, associated with EAEC infection, other than that mentioned above includes hemorrhagic necrosis and shortening of villi, enlarged crypt openings, and formation of crypt abscesses (53, 55, 56).

Pathogenicity Islands

Different pathogenicity islands have been identified in EAEC strains. One genomic island is inserted at the tRNA pheU locus and encodes the aaiC-associated type VI secretion system, which is regulated by the aggR gene (28). The Shigella species she pathogenicity island found in some EAEC strains encodes the SPATEs Pic and ShET1 enterotoxin, thereby conferring toxic and mucinolytic activities (47). Two pathogenicity islands associated with extraintestinal E. coli strains, the Yersinia high-pathogenicity island, encoding the yersiniabactin siderophore, and the hly pathogenicity island, encoding hemolysin and P-fimbriae, have also been found in EAEC isolates (57).

Phylogeny of EAEC

Phylogenetic analysis of E. coli pathotypes segregates the strains into 6 major groups, groups A, B1, B2, C, D, and E (58–61). Extraintestinal E. coli strains belong largely to groups B2 and D, whereas commensal E. coli strains frequently belong to group A (5, 61). One study investigated the phylogeny of 67 diarrheagenic E. coli strains, including 10 EAEC strains and 31 commensal E. coli strains. EHEC, Shigella species, and ETEC were found to belong to groups A, B1, C, and E, whereas EAEC strains were found to be scattered among all phylogenetic groups investigated (60). Another study investigating EAEC phylogeny by multilocus enzyme electrophoresis (MLEE) found 2 clusters of DEC containing EAEC strains. However, other pathotypes were found to fall into the same clusters on the phylogenetic map (59). One epidemiological study investigated the potential clustering of EAEC strains into different phylogenetic groups but found strains in phylogroups A, B1, B2, and D, and the authors of this study concluded that EAEC originates from multiple lineages (62). The above-described studies indicate that EAEC strains are phylogenetically diverse and do not belong to specific phylogroups.

Volunteer Studies

EAEC as a gastrointestinal pathogen was investigated in a volunteer study performed by Nataro et al. in 1995 (63). To identify pathogenic EAEC isolates, 4 groups of 5 volunteers were each fed 1 of 4 different EAEC strains, 042, JM221, 17-2, or 34b, each at a dose of 10¹⁰ CFU/ml. The strains were collected from patients suffering from diarrhea in Peru, Mexico, Chile, and India, respectively. EAEC serotype O44:H18 strain 042 expressed AAF/II fimbriae, while strains 17-2, 34b, and JM221 expressed AAF/I fimbriae, which were identified by immunogold electron microscopy and DNA hybridization. EAEC strains 042 and 17-2 were found to express the gene encoding the enterotoxin EAST1. Strains JM221 and 34b did not express EAST1. EAEC strain 042 caused diarrhea in 3 of 5 adults; the 3 other EAEC strains failed to cause diarrhea. An earlier volunteer study (64) found that EAEC serotype O78:H33/35 strain JM221 at a dose of 7 × 10⁸ CFU/ml was associated with diarrhea in 2 out of 8 volunteers and at a dose of 1 × 10¹⁰ CFU/ml was associated with diarrhea in 3 out of 8 volunteers. The incubation period was shorter with the higher inoculum. The lack of consistency in the pathogenic assessment of the EAEC strains in the 2 studies may have been be caused by the relatively small study groups. The numbers of volunteers challenged with strain JM221 differed between the studies, with 8 healthy adults being challenged in the study by Mathewson et al. (64) and only 5 volunteers in being challenged in the study performed by Nataro et al. (63). The small numbers of volunteers challenged could have resulted in a misrepresentation of different host factors important for EAEC-induced disease in the two panels of volunteers. As revealed in later studies, host susceptibility is suspected to play a pivotal role in EAEC pathogenesis, which could partly explain the different conclusions drawn from the volunteer studies.

DIAGNOSIS OF EAEC

The gold standard for the identification of EAEC remains the HEp-2 cell assay (6, 8, 31, 65). This test is performed only in reference laboratories; it requires cell culture facilities and is time-consuming (66). Molecular techniques have been developed to detect EAEC, of which PCR amplification of specific EAEC virulence-associated genes is of great importance. However, the genetic heterogeneity of EAEC and the notion that EAEC virulence probably results from a combination of multiple factors make it a challenge to differentiate pathogenic and nonpathogenic strains by the molecular methods used today (31, 51, 67, 68). Various gene targets have been used to detect EAEC by using PCR (Tables 2 and 3). One EAEC-specific gene is the chromosomal aaiC gene, which is a gene in a genomic island encoding a type VI secretion system (28). Additional EAEC genes that are frequently used to detect EAEC by PCR include aggR and aatA (11, 65, 69–71).

TABLE 2.

EAEC reference strains^a

Reference strain	Gene targets	Serotype	Reference
042	aatA, aggR, aap, aaiC	O44:H18	6
JM221	aggA, sat	O92:H33	64
17-2	aatA, aggR	O3:H2	6
C-1010	agg4A, agg3/4C, sat, sepA	O?:H1	72
55989	agg3A, agg3/4C	O104:H4	22

^aGenes, according to the protein encoded, are as follows: aatA, outer membrane protein; aggR, transcription activator; aap, antiaggregation protein dispersin; aaiC, part of a type VI secretion system; aggA, the major pilin subunit of AAF/I; agg3A, subunit of AAF/III, sat, secreted autotransporter toxin; agg4A, subunit of AAF/IV; agg3/4C, usher-encoding region from AAF/III and AAF/IV; and sepA, Shigella extracellular protease.

TABLE 3.

Geographic distribution of EAEC cases^a

Reference(s)	Yr	No. of patients (% EAEC positive)	Country(ies)	Virulence factor(s) and/or diagnostic method(s)	Study type	Conclusion(s)	Validity
119	1997	798 cases (2), 580 controls (0)	Germany	Primers complementary to the CVD432 probe and HEp-2 cell assay	Prospective case-control	EAEC was associated with watery diarrhea and chronic diarrhea in children	Strengths were large study groups and diarrhea confirmed by trained personnel; weaknesses were that EAEC-positive patients were tested only for rotavirus and that the definition of diarrhea was only 2 loose stools a day
93	1998	186 (14)	Brazil	HEp-2 cell assay and AA probe	Cohort study	EAEC-positive children had increased levels of lactoferrin and IL-1β in stool samples (P = 0.017) and significant growth impairment regardless of the presence of diarrhea (P < 0.05)	Strengths were that it was longitudinal study where the nutritional status and growth rates could be observed over time and that stool samples were tested for parasites and viruses; weaknesses were that only 1 E. coli isolate was analyzed further per stool sample and that no data on possible coinfections were available
11	2002	176 cases (49), 10 controls (0)	Jamaica, India, and Mexico	aggR, aafA, aggA, aspU, and HEp-2 cell assay	Descriptive/case-control	aggR- and aafA-positive isolates were associated with increased IL-8 levels (P < 0.05); no association between AAF/II and diarrhea was observed	Strengths were the use of 10 healthy controls traveling to Mexico and that asymptomatic EAEC carriers were examined; weaknesses were that no enteric viruses were tested for and that there was no mention of the possible use of antibiotics before microbiological testing
69	2005	Neonates (no data on numbers)	South Korea	aggA, aafA aggR, astA, pic, pet, CVD432 probe, and HeLa cell assay	Screening	Some EAEC probe-positive isolates were not positive in the HEp-2 cell assay; astA and aggR were the most prevalent EAEC virulence genes	Strengths were thorough microbial testing using HeLa cell assay, PFGE, sensibility testing, and DNA hybridization; weaknesses were that there was no case definition besides diarrhea and EAEC status, there was no information regarding the use of antibiotics, stool samples were not tested for gastrointestinal viruses, the no. of newborns tested was not mentioned, and only the no. of E. coli isolates tested was mentioned
32	2007	122 cases (55), 127 controls (32)	Brazil	aggA, aafA pic, pet, astA, CVD432 probe, and HeLa cell assay	Case-control	Typical EAEC strains were found in equal numbers in both groups; the astA gene was associated with acute diarrhea (P < 0.01); the CVD432 probe was associated with persistent diarrhea (P < 0.001)	Strengths were the exclusion of patients who had received antibiotics, those who had congenital gastrointestinal malformation, and those who had carbohydrate intolerance and matching by age and socioeconomic class; weaknesses were that cases were poorly defined as “an increase in evacuations with loose stools” and that there was no microbiological testing for gastrointestinal viruses or parasites in stool samples
79	2007	7 cases (71)	Japan	astA, aggR, CVD432 probe, and HEp-2 cell assay	Outbreak	An EAEC strain of serotype O126:H27 positive for the CVD432 probe; aggR and astA were found in 4 patients and 1 food handler	Strengths were that microbiological testing included PCR, PFGE, HEp-2 cell assay, and serotyping; weaknesses were that there was no testing for gastrointestinal viruses or parasites
65, 75, 160	2007	3,506 cases (5), 2,772 controls (2)	UK	aatA, aap, aggR, aggC, agg3C, aafC, shf, pet, pic, aaiA, astA, irp2, tia, and HEp-2 cell assay	Case-control	aafC was the only gene associated with diarrhea (P < 0.005); aggC was the only gene associated with the control group (P < 0.002); EAEC was found at a high frequency in cases and controls	Strengths were that there was a solid case definition and reduced bias by excluding patients with cystic fibrosis, Crohn's disease, ulcerative colitis, or celiac disease; weaknesses were that there was no information on possible coinfections, no testing for gastrointestinal viruses or parasites, no information from study groups regarding travel, and no information on demographics
154	2009	253 cases (7), 751 controls (3)	USA	aggR, aggA, aafA, aap, aatA, astA, pet, set1A, and HEp-2 cell assay	Case-control study	EAEC strains with aggR, aatA, astA, aap, and set1A were associated with diarrhea (P < 0.05); aggR-positive strains caused higher levels of IL-1ra, IL-6, IL-8, and tumor necrosis factor alpha	Strengths were that there were large study groups, only patients without travel within 3 mo were included, and community-acquired EAEC infections could be detected; weaknesses included a lack of information on demographics in the 2 groups, posing a possible selection bias
78	2008	16 cases (44)	Italy	aap, astA, set1A, aat, aggR, and HEp-2 cell assay	Outbreak	An EAEC strain with aggR, aap, and set1A of serotype O92:H33 was isolated from 6 individuals and 1 staff member; suspected cause of diarrhea was unpasteurized cheese	Strengths were that microbiological testing of food samples and stool samples from farm animals was performed; weaknesses were that stool samples were tested for norovirus only and not for other gastrointestinal viruses
105	2010	170 cases (87), 104 controls (77)	Southern Ghana	aatA, aap, aggR, and aaiC	Prospective cross-sectional study	EAEC was associated with diarrhea (P < 0.048); EAEC was associated with increased fecal lactoferrin levels regardless of the presence of diarrhea or nutritional status	Strengths were the use of a questionnaire on the duration of diarrhea, demographics, living conditions, breastfeeding, and consumption of medicine and that anthropometric measurements were included in the study; weaknesses were that the stool samples were not tested for gastrointestinal viruses
133	2010	394 cases (15), 198 controls (24)	South India	aafII, aap, astA, aggR, AA probe	Case-control	EAEC was the most commonly isolated DEC type in both groups; the combination of the aap and aggR genes and the AA probe was associated with diarrhea (P < 0.001)	Strengths were that children were tested for rotavirus and Cryptosporidium species; weaknesses were that coinfections were mentioned, but data on the distribution in EAEC-positive patients were not given
52	2010	140 (11)	Iran	aggR, aafA, aggA aap, astA, primers complementary to the CVD432 probe, and HeLa cell assay	Descriptive	EAEC with the genetic combination of aggR, aap, and astA was the most prevalent; only 86.7% of PCR-identified EAEC bacteria were positive in the HeLa cell assay	Strengths were that information regarding sex, age, and seasonality was available; weaknesses were that there was no control group, there were no data available on possible coinfections, and there was no information regarding recent travel
17	2011	23 ulcerative colitis (61), 8 Crohn's disease (88), 23 controls (30)	Brazil	aggR, primers complementary to the CVD432 probe, and HEp-2 cell assay	Case-control	EAEC was frequently isolated from patients suffering from Crohn's disease and ulcerative colitis when examining E. coli isolates from rectal biopsy specimens	Strengths were the selection of patients with well-defined illnesses and information on the use of antibiotics and age; weaknesses were that there was no testing for conventional enteropathogenic bacteria and no testing for viruses or parasites
157	2013	83 cases (41), 83 controls (41)	Brazil	aaiC, aatA, capU, aap, eilA, aggR, orf3, sat, sepA, pic, sigA, pet, astA, aafC, agg3/4C, agg3A, aafA, aggA, agg4A, and orf61	Case-control	pet or aafA was significantly associated with disease; the EAEC strains possessing aaiC and agg3/4C but lacking agg4A and orf61 were associated with diarrhea cases	Strengths were that a large panel of EAEC virulence genes was screened for in EAEC-positive stool samples; weaknesses were that conventional enteropathogenic bacteria, viruses, and parasites were not tested for in stool samples

^aA wide range of gene targets and combinations thereof were used to detect EAEC by PCR. HeLa cell assays can be used to detect the same AA pattern as that detected by the HEp-2 cell assay. The genes, according to the protein encoded, are as follows: aaiA, directing secretion of aaiC; aaiC, part of a type VI secretion system; aggA, major pilin subunit of AAF/I; aafA and aafII, subunit of AAF/II; aafC, usher, AAF/III/IV assembly unit; agg3 and agg3A, subunit of AAF/III; agg4A, subunit of AAF/IV; aggC, usher-encoding region; aggR, transcription activator; astA, aggregative heat-stabile toxin 1 EAST1 (enteroaggregative E. coli heat-stabile toxin); aap, antiaggregation protein dispersin; afaD, invasion; aatA, outer membrane protein; aspU, dispersin surface coat protein; capU, hexosyltransferase homolog; cdt, cytodetaching toxin; eilA, EAEC HilA homolog; fyuA, involved in iron sequestration; hda, adhesin; hly, hemolysin; hra1 and -2, heat-resistant agglutinins; ipaH, invasion; irp2, Yersinia biosynthesis gene; orf3, cryptic protein; orf61, plasmid-encoded hemolysin; pet, plasmid encoding toxin; pic, protein involved in colonization; sat, secreted autotransporter toxin; sepA, Shigella extracellular protease; ShET, set, and set1A, Shigella enterotoxin 1; sigA, Shigella IgA-like protease homolog; shf, cryptic open reading frame; stx, Shiga toxin; and tia, gene product protein important for invasion.

In the volunteer study by Nataro et al. (63), EAEC strain 042, eliciting the stacked-brick pattern, was shown to cause diarrhea in the majority of study participants, and since then, it has served as the main EAEC reference strain. Reference strains representing specific EAEC genes, for performing PCR and observing the aggregative pattern in the HEp-2 cell assay, include 042 (aatA, aggR, aap, and aaiC), JM221 (aggA and sat), 17-2 (aatA and aggR), 55989 (agg3A and agg3/4C), and C1010 (agg4A, agg3/4C, sat, and sepA). AAF/I, -II, -III, and -IV are expressed by reference strains 17-2, 042, 55989, and C1010, respectively (6, 22, 42, 66, 71, 72) (Table 2).

Serotyping of EAEC strains has proven to be unsuitable due to the considerable diversity among strains and the large number of so-called “rough” strains, which are strains that do not express an O antigen (73–76). Autoagglutination caused by the aggregative phenotype is another obstacle when serotyping EAEC, leading to a large number of nontypeable strains (23, 26, 42, 77). Serotyping and phylogenetic typing of EAEC have therefore proven to be useful only in outbreak-related cases (78–82). The typical aggregative adherence pattern has been found to be associated with a 60-MDa plasmid (pAA) (83). This plasmid carries several virulence factors, including the aggregative adherence fimbriae and toxins (20). A DNA probe named CVD432 (also called the AA probe) has been constructed from plasmid pAA for the detection of EAEC by DNA hybridization (83). To identify EAEC, the specificity of the CVD432 probe has been reported to be high but with various sensitivities compared to the HEp-2 cell assay (7, 52, 83, 84). The lack of sensitivity comes from the genetic heterogeneity of the EAEC strains and the wide geographic dispersal of strains analyzed. As suggested in a study by Boisen et al. (20), the combination of virulence genes may depend on the geographic region. International microbiological surveillance of EAEC and an improved understanding of EAEC pathogenesis could lead to the achievement of a proper diagnostic algorithm.

CLINICAL MANIFESTATIONS

The symptoms associated with EAEC infections include watery diarrhea and occasionally very mucoid diarrhea (7, 29, 40), nausea, anorexia, low-grade fever, borborygmi, and tenesmus (39, 85, 86). Cases of both acute and persistent diarrhea have been described. Persistent diarrhea is most frequently reported in children aged ≤1 year (32, 35, 87, 88). The site of colonization is believed to include the colon and the terminal ileum (10, 14, 55, 89). The incubation time ranges from 8 h to 52 h (78, 90–92). A study by Steiner et al. in 1998 found that children in developing countries who were diagnosed with EAEC infection suffered from growth retardation regardless of the presence of diarrhea (93). In a study by Roche et al. (13), growth retardation due to EAEC infection was observed in a mouse model. The growth impairment was found to be dependent on the dose of bacteria used for challenge. It was observed that malnourished EAEC-inoculated mice had reduced growth velocity and increased shedding of EAEC in stools compared to nourished mice. Bloody diarrhea has been reported only rarely and involves mostly small children (10, 67, 94, 95). However, the German O104:H4 EAEC Shiga toxin-expressing outbreak strain caused hemorrhagic colitis and hemolytic-uremic syndrome (HUS), leading to considerable morbidity and casualties (96–98). The outbreak strain contained the EAEC genes aggR, aggA, set1, pic, and aap and a prophage encoding the stx₂ gene (99). Urinary tract infections (UTIs) associated with EAEC (46, 81, 100, 101) and one case of urosepsis in an immunosuppressed female (102) have also been described recently. An outbreak of UTIs associated with EAEC in 1991 was reported in a Danish study (81), where the UTI outbreak strain contained the following combination of EAEC genes: sat, pic, aatA, aggR, aap, aaiC, and aggA.

Inflammatory Response

The inflammation caused by EAEC has been shown by increased levels of fecal interleukin-8 (IL-8), IL-1β, leukocytes, and lactoferrin, indicating a substantial gastrointestinal inflammatory response (11, 87, 103, 104). Occult blood has been detected in stool samples from EAEC-positive HIV patients (94). A study by Steiner et al. investigated childhood diarrhea and found increased levels of inflammation mediators, even in asymptomatic EAEC-positive carriers, compared to healthy EAEC-negative controls (93). In another study performed in southern Ghana, children who were asymptomatic carriers of EAEC showed increased levels of fecal lactoferrin (105). These findings suggest that EAEC could have pathogenic potential in asymptomatic carriers. The IL-8 inflammatory response has been shown to be partially caused by flagella (fliC) in a Caco-2 cell assay, as it was found that an aflagellar mutant of EAEC did not produce the same inflammatory response (106).

Host Factors Are Determinants of EAEC Pathogenesis

EAEC has been isolated in high numbers from stool samples of asymptomatic carriers (35, 107–109), leading to the theory that the manifestation of gastrointestinal disease may depend on host factors (10, 91, 110). An American study investigated different single nucleotide polymorphisms (SNPs) in the promoter region of the CD14 gene as predictors of the development of traveler's diarrhea due to EAEC (111). The CD14 gene encodes a crucial step in the inflammatory response to bacterial lipopolysaccharide stimulation by the innate immune system. This study found that one SNP in the promoter region of the CD14 gene was associated with an increased risk of EAEC-induced diarrhea. Patients with the CD14 −159 TT genotype were significantly associated with EAEC-induced diarrhea (P = 0.008) compared with healthy controls. The diverse clinical presentations reported for EAEC-positive individuals support the hypothesis of a genetic predisposition for the manifestation of illness. Other host factors that may be important for EAEC pathogenesis require further investigation.

Immunological Response to EAEC Infection

Dispersin has been shown to cause an immune response in travelers visiting Mexico. Five students with diarrhea were diagnosed as having EAEC in their stool samples, and they showed a significant increase in anti-dispersin IgG absorbance when serum samples were compared before and after infection measured by an enzyme-linked immunosorbent assay (ELISA) (P = 0.01). Students with and those without diarrhea showed significantly higher levels of IgG antibody to a recombinant dispersin protein upon returning from Mexico than the levels measured in serum at baseline (P < 0.0001). However, students who tested positive for dispersin antibodies at baseline were not protected from EAEC infection during their stay (85). In addition, dispersin has not been found to be restricted to EAEC strains, as it has also been detected in nonpathogenic E. coli strains (112). Another study showed increased levels of IgA antibodies binding to crude EAEC antigens in stool samples from 5 out of 10 American students returning from Guadalajara, Mexico. In that study, EAEC strain JM221 was used as the antigen in dot blot and Western blot analyses (113). A volunteer study also showed increased levels of antibodies to different antigen preparations from EAEC strain JM221 measured in stool samples from patients who developed diarrhea after oral challenge with strain JM221. Five patients who did not develop diarrhea in the trial period proved to have increased secretory IgA (sIgA) antibody levels in stool samples before challenge, suggesting possible antibody protection (114). In a Brazilian case-control study, increased serum levels of IgM and IgG antibodies to the toxins Pet and Pic were measured in children following an EAEC infection; however, antibodies to Pic were also observed in healthy children. No obvious correlation between EAEC strains carrying the pic and pet genes isolated from children and measured antibodies to Pic and Pet were seen (77). The conclusion from these studies is that none of the antibodies measured have proven to offer high specificity for EAEC, which excludes their applicability in a clinical setting.

MULTIDRUG RESISTANCE IN EAEC STRAINS

A worrying high level of multidrug resistance among EAEC strains has been reported in several studies (8, 52, 115, 116). In addition, extended-spectrum beta-lactamase (ESBL) production and increased resistance to quinolones in EAEC strains have been described (108, 115–117). A study conducted in southern India (118) investigated 64 EAEC strains and reported that 75% of the strains were multidrug resistant, and resistance to ciprofloxacin was found in 63.5% of the strains. The majority of the multidrug-resistant strains were identified in children <5 years old; this may be caused by the higher prevalence of EAEC in this age group and by colonization rather than infection, leading to a long-term carrier state, which may facilitate antibiotic pressure. This statement is supported by another study investigating ESBL production by EAEC strains, which was detected in 1.5% of children with diarrhea and in 4.3% of children without diarrhea (determined by PCR targeting the CTX-M enzyme) (108). This study was performed in 2008 to 2009 in Nicaragua, and EAEC strains were found to have levels of resistance to ampicillin and trimethoprim-sulfamethoxazole that were significantly higher than those of EHEC and EPEC strains (P < 0.005) and to have higher levels of resistance to amoxicillin than ETEC strains (P = 0.021).

In an Iranian study, multidrug resistance in 10 EAEC strains was investigated, and it was found in 71.4% of strains, with reduced resistance to ciprofloxacin being detected in 42.9% of strains (52); however, the small number of strains tested in this study does not allow a generalized picture of resistance of EAEC strains. Another study investigated the presence of ESBL-producing strains and detected ESBL in 5 of 51 EAEC strains isolated from Spanish travelers who had visited India (116). The German EAEC O104:H4 outbreak strains showed ESBL production (99), and of Danish EAEC UTI O78:H10 outbreak strains, 37% were found to be multidrug resistant (81). The conclusion from the studies mentioned above is that although the proportion of ESBL-producing EAEC strains is small, it is still a cause for concern, and such strains are being increasingly reported. In the battle against the development of further antibiotic resistance, the use of fluid replacement and supportive treatment is preferable for management of EAEC-induced diarrhea in uncomplicated cases. Empirical antimicrobial therapy of EAEC should be avoided when antibiotic intervention is deemed necessary.

RESERVOIR AND TRANSMISSION OF EAEC

The reservoir for EAEC still has not been determined, but it is generally accepted to be human (97, 115, 119). The transmission of EAEC is often described as being food borne or through contaminated water, and as such, it is believed to be transmitted by the fecal-oral route (11). One study investigated the growth of EAEC in drinking water (120), where bacteria were added to different sources of bottled water. It was discovered that EAEC strains maintained their viability for up to 60 days at normal storage temperatures. The survival of the bacteria was discovered to be prolonged in mineral water compared to spring water. This was suggested to be due to the larger amounts of Ca²⁺ and Mg²⁺ in mineral water, inducing genetic competence. Surface water, which in some regions of Australia is a source of drinking water, was investigated for the presence of pathogenic E. coli from various creeks, rivers, and drains before and after a storm in Brisbane, Australia. EAEC was identified by PCR targeting the aggR gene in 36% of the water samples in the dry period and in 26% in the wet period and was found to be highly prevalent among the diarrheagenic E. coli strains investigated (121). In a different study, water samples from 46 aquatic locations in Bangladesh were collected during both the winter and summer seasons, and those investigators found EAEC by PCR targeting a primer sequence complementary to the CVD432 probe in 17% of the samples in the winter season and 4% in the summer season (122). Another possible route of transmission of EAEC is food handling, which was investigated in a study performed in Ouagadougou, Burkina Faso, where EAEC was detected by PCR targeting the pic and aggR genes. EAEC was detected in 5 samples out of 120 samples of different meats and beef (123). Weaning foods as a possible route of transmission of gastrointestinal pathogens was investigated in a study performed in São Paolo, Brazil, and EAEC was found by DNA hybridization in 3 out of 100 milk samples from infant feeding bottles that were handled by mothers with low socioeconomic status (124). Animals as possible reservoirs of EAEC were investigated in a British study (125), and 1,227 E. coli isolates were investigated as possible EAEC strains: 401 from cows, 406 from sheep, and 400 from pigs. However, no EAEC isolates were detected by using the pAA probe. A different study investigated diarrheagenic E. coli strains from cats and dogs in Brazil and detected EAEC by PCR targeting the aggR gene. EAEC was detected in 7.4% of dogs with diarrhea, 3.9% of dogs without diarrhea, and one cat without diarrhea (126). A French study investigated whether slaughterhouse waste is a possible source of EAEC in rivers, but EAEC was not detected in wastewater or effluents by PCR targeting the aggR, aap, and aatA genes (127). EAEC has been isolated only rarely from animal sources, and whether animals truly are potential reservoirs of EAEC is not conclusive. Whether companion animals are accidental hosts of EAEC due to close contact with humans can only be speculated upon. Sharing of E. coli strains among household members and pets has been reported previously (128). The above-mentioned studies strongly indicate that one likely route of transmission of EAEC is via contaminated food and water, but further research to exclude nonhuman reservoirs of EAEC should be conducted.

EAEC AS A CAUSE OF CHILDHOOD DIARRHEA IN DEVELOPING COUNTRIES

Early studies on the etiology of diarrhea in children living in developing countries revealed that EAEC is highly prevalent (10, 93, 129). A case-control study from 2004 (67) investigated the cause of diarrhea in Mongolian children. Here, EAEC was the most frequently detected type of DEC. Only EAEC strains with the aggR gene were found to be associated with diarrhea. A possible confounder in the study may have been gastrointestinal viruses, for which no testing was performed. In 1999, a case-control study was performed in Kolkata, India, where childhood diarrhea was investigated, and EAEC strains were detected in stool samples by a HeLa cell assay and PCR amplification of a gene product complementary to the CVD432 probe. EAEC was found to be associated with diarrhea in children <36 months of age (130). In this study, watery diarrhea was reported more frequently than mucoid diarrhea among EAEC-positive children (72% versus 28%). A study by Steiner et al. in 1998 (93) found that growth impairment and increased levels of fecal inflammation markers were associated with EAEC infection in children living in a Brazilian urban slum. EAEC was detected by the AA gene probe and the HEp-2 cell assay, and EAEC-positive children in this study had impaired growth regardless of clinical manifestations (Table 3). Another case-control study investigating childhood diarrhea failed to detect any association with EAEC, but there was a high rate of carriage of EAEC among controls. This study was performed in Dhaka, Bangladesh, from 1993 to 1994, and EAEC was detected by DNA hybridization using the CVD432 probe (131, 132). A high rate of carriage of EAEC in young children has been reported by several other studies. EAEC was found in greater numbers in the control group in a case-control study (133) conducted between 2003 and 2006 in South India. This study investigated the etiology of diarrhea in children <5 years of age. EAEC was detected by multiplex PCR targeting the astA, aap, aggR, and aafII genes and by PCR amplification of a product corresponding to the AA probe. EAEC was found in 14.7% of cases and 23.7% of controls. However, EAEC strains carrying only the aap, aggR, and genes corresponding to the AA probe were significantly associated with diarrhea, with incidences of 6.1% in cases and 3.0% in controls (P < 0.001). Another recent study found a high rate of asymptomatic carriers of EAEC in Mali, where diarrhea in children aged 0 to 59 months was investigated. After 1 year of surveillance, EAEC was found to be the only DEC in 60 children with diarrhea and in 61 children without diarrhea. However, EAEC strains with the sepA gene were found to be associated with diarrhea (odds ratio [OR], 5.6; P = 0.0006) (42). In conclusion, the many studies on childhood diarrhea and EAEC indicate that there is a valid association, but an exhaustive diagnostic method to detect pathogenic EAEC strains is required, and the study design should include simultaneous testing for conventional gastrointestinal pathogens. In addition, to assess the pathogenicity of EAEC in childhood diarrhea in general, case-control studies should be conducted in a Western setting to minimize or eliminate contributing disease-related factors such as malnutrition and poor hygiene.

ASSOCIATION BETWEEN EAEC AND HIV-RELATED DIARRHEA

HIV-related diarrhea was linked to EAEC in early studies, and EAEC was considered an opportunistic pathogen. A Swiss study conducted between 1996 and 1998 investigated HIV-related diarrhea (134), where EAEC was detected by the HeLa cell assay, the CVD432 probe, and PCR primers complementary to the CVD432 probe. In this study, a low CD4 lymphocyte count was not found to be a predisposing factor for EAEC infection. Of the 7 patients who experienced diarrhea, 4 later suffered from persistent diarrhea or chronic intermittent diarrhea. Tests for enteric parasites, cytomegalovirus, and enteropathogenic bacteria were included in the testing of the stools. A study conducted in Senegal (135) found EAEC in 31 of 158 HIV patients with diarrhea and in 3 of 160 HIV patients without diarrhea. EAEC was detected by PCR amplification of a product corresponding to the CVD432 probe (136). In that study, EAEC was defined as an opportunistic pathogen. An association between EAEC and diarrhea has not been reported by all studies on HIV-related diarrhea. EAEC was found in equal numbers in diarrhea cases and controls among HIV-positive children in Peru (137), where EAEC was detected by PCR targeting the aggR gene. Another study found that HIV patients were more likely to be EAEC positive: 29.5% versus 14% in the HIV-negative group. EAEC was detected by multiplex PCR targeting aggR and aap, the AA probe, and quantitative real-time PCR targeting the aggR gene (94). EAEC was found to be associated with occult blood in stool samples, (OR, 5.096; confidence interval, 2.665 to 9.644), and an association between bacterial load and the manifestation of illness was reported.

EAEC continues to play an important role in detection of the cause of diarrhea in HIV patients. However, the lack of an association between diarrhea in EAEC-positive patients and CD4 lymphocyte counts does not support the theory of opportunistic pathogenicity, and today, the association between EAEC and diarrhea is not limited to this group of patients.

TRAVELER'S DIARRHEA AND EAEC

More recent studies have found an association between EAEC and traveler's diarrhea. Initially, this was found for patients returning from developing countries, but now, the association has been reported in cases of traveler's diarrhea obtained from various destinations. One study of traveler's diarrhea concluded that the EAEC virulence factors aggR and aggA were the most prevalent EAEC genes detected in European and American patients with diarrhea returning from Montego Bay, Jamaica; Goa, India; and Guadalajara, Mexico (11). Another study found the combination of the aggR gene and the aat gene to be associated with traveler's diarrhea; this combination was detected in 45% of cases, as opposed to 16.4% of controls (P < 0.01). One strength of this study was that cases of traveler's diarrhea were compared with travelers who did not have diarrhea and who were returning from the same regions in the tropics of Asia, Africa, and Latin America (138). An American study (29) investigated cases of traveler's diarrhea from India, Guatemala, and Mexico. Forty-nine travelers had diarrhea and EAEC, 15 had EAEC without diarrhea, and 17 neither had diarrhea nor tested positive for EAEC. EAEC strains with the aggR, astA, set1A, and aap genes were associated with traveler's diarrhea (P < 0.05). Coinfections were reported to occur at a high frequency in another study of patients returning from India, Jamaica, and Mexico (139). The most frequent pathogen isolated in that study was ETEC, which was found in 41 out of 72 mixed infections with EAEC detected by the HEp-2 cell assay. The presence of another well-established pathogen, i.e., coinfection, in these patients' stool samples complicates any association between diarrhea and EAEC. A high number of coinfections was reported by a study investigating traveler's diarrhea among patients who had visited Benin, West Africa (140). This study found that 79% of the patients were coinfected, where EPEC and EAEC were the most frequently detected pathogens. Coinfections with EAEC were also mentioned in another study but were described only as being scattered among cases of traveler's diarrhea. This study investigated Korean patients returning from Southeast Asian countries (141), where ETEC had the highest incidence of enteric pathogens detected (found in 36.0% of cases), followed by EAEC (found in 27.0% of cases). EAEC was detected by PCR, but the primers used in this study were not described. Other pathogens detected were Vibrio parahaemolyticus, Vibrio cholerae, Salmonella species, Shigella, and norovirus.

EAEC is often isolated in cases of traveler's diarrhea from different countries, but the strong association found may have been overestimated due to the vast number of studies on EAEC performed for this patient category. The large number of coinfections and the lack of exhaustive testing for enteric pathogens in these patients mean that the general assessment of the pathogenic contribution of EAEC to cases of traveler's diarrhea must still be regarded as questionable.

ASSOCIATION BETWEEN EAEC AND DIARRHEA IN INDUSTRIALIZED COUNTRIES

In studies of the etiology of diarrheal episodes in industrialized countries, the prevalence of EAEC has been reported to be high. A study conducted in Germany in 1997 found EAEC to be the third most commonly isolated microorganism in young children with diarrhea (2%), after Salmonella species (13.4%) and STEC (3.1%). EAEC infection was diagnosed in 16 cases by using the HEp-2 cell assay and PCR primers complementary to the CVD432 probe. EAEC was not detected in an asymptomatic group of children (119). However, a prospective American study did not find an association between EAEC and childhood diarrhea in the community setting (142). This study involved 604 healthy 6- to 36-month-old children who were monitored over a 6-month period, and the incidence of diarrhea was reported to be 2.2 per person-year. Children with intestinal malabsorption, inflammatory bowel disease, or cystic fibrosis and those who were treated with antibiotics were excluded. Testing for enteric viruses, parasites, and enteropathogenic bacteria was performed. Whether or not the exclusion of children with the comorbidities mentioned above in this study played an important part in the pathogenic potential of EAEC can only be speculated upon but warrants further research. An investigation of possible host factors contributing to EAEC-induced disease in children in industrialized countries should be conducted. A case-control study performed in Cincinnati, OH, found that children <1 year of age in the case group had EAEC isolated significantly more frequently from stool samples than did children in the control group. Interestingly, pathogenic EAEC strains were identified only by the CVD432 probe and not by the aggregative adherence pattern observed in the HEp-2 cell assay. Strains with the aggregative adherence phenotype were equally distributed between groups (143). Aggregative adherence in the HEp-2 cell assay was observed in E. coli strains that did not possess any known EAEC genes in a British study investigating community-acquired diarrhea (73). The HEp-2 cell assay was reported to detect various EAEC genotypes. This study used PCR primers targeting the aat, aaiA, and astA genes. Stool samples were collected as part of a routine examination for diarrhea by the patients' general practitioner, which included testing for EPEC, Salmonella species, Shigella sonnei, Shigella flexneri, Clostridium difficile, Campylobacter species, enteric parasites, and viruses. EAEC strains with a combination of the aat, aai, and/or astA gene were found in 39 of 500 patients with diarrhea. Some EAEC-positive patients were coinfected with, for example, Campylobacter species (3 cases) and rotavirus (1 case). Foreign travel was associated with only a minor proportion of the EAEC-positive cases.

EAEC was associated with diarrhea in a national surveillance study conducted from 1993 to 1996 and from 2008 to 2009 in the United Kingdom (144). It was shown that approximately 3.3% of intestinal infectious diseases were associated with EAEC, and the burden of EAEC was measured by quantitative PCR amplification of a product corresponding to the CVD432 probe. However, the EAEC burden determined by quantitative PCR was not useful for discrimination between EAEC strains isolated from positive cases and EAEC strains isolated from controls.

In conclusion, EAEC is frequently isolated in industrialized countries, not only from outbreaks, and should be considered for inclusion in a diagnostic algorithm for enteric pathogens in these settings. However, more case-control studies should be performed in Western countries to identify the possible pathogenic contribution of EAEC to diarrheal episodes in these areas.

URINARY TRACT INFECTIONS AND EAEC

Recently, EAEC strains have been associated with urinary tract infections (UTIs) (46, 100, 101, 145). An outbreak of UTI cases associated with an EAEC strain also possessing extraintestinal pathogenic E. coli (ExPEC) genes was reported in 1991 in a Danish study (81). In that outbreak, 18 patients suffering from UTI in the Copenhagen area were found to be infected by strains of the same serotype, O78:H10, and the same phylogenetic group (group A). The outbreak strain had the ExPEC genes fyuA, traT, and iutA and the EAEC genes sat, pic, aap, aaiC, aatA, aggR, and aggA. In a later study, which further characterized the Copenhagen outbreak strains, it was shown that most of the 19 strains were capable of biofilm formation. One of the outbreak isolates, originating from a child aged 4 months with recurrent UTI, proved to form extensive biofilms and pronounced adhesion to human bladder epithelial cells by the expression of AAF/I (34). This indicates that EAEC-associated virulence factors enhance uropathogenicity, and the possession of these virulence factors may have promoted the capability of the strain to cause the outbreak.

A case of urinary tract infection preceded by diarrhea caused by EAEC was reported in a 55-year-old immunocompromised female, leading to bacteremia and urosepsis (102). In this study, an EAEC strain with the same pulsed-field gel electrophoresis (PFGE) profile was isolated from stool, urine, and blood culture samples from this patient. The strain was serotyped as O176:NT and tested positive for the CVD432 probe and the aggR, aap, agg3C, astA, pic, aaiC, and air genes by PCR. In a Korean study (145), the cause of UTI in small children was investigated by analyzing suprapubic urine specimens from children with and those without UTI (as defined by numbers of CFU in urine). EAEC was detected by PCR targeting the aggR and aap genes in children with cases of UTIs; however, strains with the combination of aggR and aap were found in significantly higher numbers in the control group (P = 0.03). EAEC isolated from patients with UTI, with different combinations of the EAEC genes aap, aggR, aggC, astA, agg3C, pet, and pic but with all strains containing genes encoding AAF/I, were reported in a Brazilian study (100). These authors suggested that this fimbria might play an important role in the uropathogenesis of EAEC.

In general, characterization of E. coli isolated from UTI cases is limited in the clinical setting, which obscures the pathogenic contribution from different uropathogenic E. coli groups. However, true outbreaks of UTI cases are rare, implicating little knowledge on this subject. Clearly, the role of EAEC in uropathogenesis should be further investigated.

OUTBREAKS

Outbreaks of gastroenteritis linked to EAEC have been reported (Fig. 3). One outbreak took place in a Serbian nursery (80) in 1995, where EAEC was detected by the HEp-2 cell assay and was characterized as belonging to serotype O4 in 12 of 19 babies who had low-grade fever, diarrhea, and weight loss. Rotavirus was the only gastrointestinal virus screened for. PCR was not performed on the outbreak strain. Another outbreak took place in a police institute in Japan in 2005, where the staff experienced gastroenteritis after having consumed food suspected of being contaminated with EAEC (79). Four staff members and one food handler tested positive for EAEC in stool samples; the isolates were found to have an identical serotype (O126:H27). However, no microbiological investigations for enteric viruses were done for this outbreak. The largest reported outbreak of EAEC, apart from the 2011 German O104:H4 EAEC outbreak, took place in the Gifu prefecture, Tajimi, Japan, in 1993 (90). Here, 2,697 children developed signs of food poisoning after having consumed school lunches. Fecal specimens were collected from 30 children with severe protracted diarrhea, and 12 tested positive for EAEC by the HEp-2 cell assay and the astA gene by PCR; the strains were all O untypeable:H10. However, the children in this outbreak were not tested for gastrointestinal viruses, to which the symptoms correspond, including short-term gastroenteritis, nausea, stomach ache, and diarrhea, which affected a considerable number of the cases in the outbreak. Another EAEC outbreak took place in 2008 in Italy at a farm holiday resort, where the guests experienced gastroenteritis after having consumed unpasteurized cheese (78). EAEC was isolated in stool samples from 6 restaurant guests and 1 staff member; the strains were identified by the HEp-2 cell assay. The EAEC strain involved was identified as belonging to serotype O92:H33, and it tested positive for the virulence genes aggR, aat, aap, and set1A by PCR. Testing for norovirus, but not rotavirus, was performed in that study.

FIG 3.

Geographic distribution of EAEC outbreaks. N is the number of patients with confirmed EAEC involved in the outbreak. The total numbers of patients involved in outbreaks of diarrhea are given in brackets. The Danish outbreak was an outbreak of EAEC causing urinary tract infections. References are shown in parentheses. HUS, hemolytic-uremic syndrome.

The outbreaks mentioned above show a considerable potential for food-borne transmission of EAEC. The lack of international surveillance of EAEC could lead to missed outbreaks caused by EAEC.

The German O104:H4 EAEC Outbreak in 2011

In May 2011, a large food-borne outbreak of Shiga toxin-producing E. coli (STEC) occurred in Germany, with 3,842 confirmed cases. The source of the food-borne outbreak was bean sprouts, which caused 855 (22%) cases of hemolytic-uremic syndrome and considerable mortality (amounting to 54 [1.4%] deaths) (146). The outbreak strain was found to show the characteristic stacked-brick pattern when grown on HEp-2 cells, and it tested positive for the EAEC genes aggR, aggA, set1, pic, and aap (99). The outbreak strain was serotyped as O104:H4 and contained a prophage encoding the Shiga toxin 2 gene, stx₂, which is normally absent in EAEC strains (98). Other, considerably smaller, occurrences of EAEC-STEC hybrid stains have been reported (147–149). An outbreak of HUS in 10 children in France in 1992 was associated with an EAEC strain of serotype O111:H2, which was positive for the stx₂ gene (150). A 3-year-old boy from Japan was diagnosed with HUS in 1999, where an EAEC strain of serotype O86:HNM which was positive for the stx₂ gene was identified (148). An EAEC O111:H21 strain with the stx₂ gene was found to be associated with an outbreak of HUS in a household in Ireland in 2012 (149). An EAEC strain of serotype O104:H4, but lacking the Shiga toxin, was previously detected in an HIV-positive patient in the Central African Republic who was suffering from persistent diarrhea (22). Further characterization of the outbreak strain by whole-genome phylogenetic analysis showed a cluster of genes with a close genetic relationship between the African non-Shiga toxin-positive O104:H4 strain and the outbreak strain (98), which suggested a recent acquisition of the phage carrying stx₂. The outbreaks mentioned above show the considerable potential of EAEC to acquire additional virulence through genetic exchange, and this further emphasizes the need for microbiological surveillance.

DISCUSSION

The CVD432 probe targeting the aat gene is widely used for the detection of EAEC by DNA hybridization (17, 52, 69, 79, 119, 131). However, this probe fails to detect all EAEC strains showing the stacked-brick pattern by the HEp-2 cell assay (65, 69, 106). The possibility of interobserver variability in pattern recognition was mentioned in a study by Pabst et al. (151), which could lead to different rates of detection of EAEC between studies. Indeed, strain JM221 was first described to show localized adherence (64) but has since been said to show aggregative adherence (63). One case-control study found patterns of adherence of EAEC strains (found by PCR testing for the aatA and aggR genes) to be diverse, including nonadherent, sparsely adherent, and densely adherent patterns, without any correlation with cases or controls (95).

PCR is a less demanding and more objective method for the detection of pathogenic E. coli. Extensive studies have been performed by using PCR to find the right combination of genes identifying the “truly pathogenic” EAEC strains (51, 65, 69, 79, 105, 152, 153); however, no consensus has been reached on this matter. As suggested by another study (20), the combination of EAEC genes might depend on the geographic region, and recognition of these various combinations of genes could lead to a better understanding of their role in the pathogenicity of specific EAEC strains.

The aggR gene is highly conserved among EAEC strains and has been found to be associated with diarrhea in several studies (105, 133, 154). A strong correlation between the CVD432 probe and the aggR gene has been reported (7, 52, 69). Quantitative PCR targeting the aggR gene has shown dose dependency (94) in the manifestation of illness. However, several other studies have not been able to associate aggR with disease (42, 65, 74, 155). Another way of quantifying the EAEC burden was used in a study using DNA extraction from stools, which showed a log-linear correlation between CFU of E. coli per gram of stool and median fluorescence intensity determined by multiplex PCR-Luminex technology. A correlation between the aatA gene and CFU of EAEC in stool was found. Further research on this method, combined with clinical information, could reveal a possible correlation between the degree of illness and the EAEC burden (156).

Persistent diarrhea has frequently been mentioned in the studies reviewed, but it appears to lack solid documentation. Several studies have found a low prevalence of persistent diarrhea in EAEC-positive patients (86, 129, 151, 157), which calls into question the importance of this disease trait in an EAEC infection in general. Despite the genetic heterogeneity of EAEC strains, the reported clinical manifestations have a high level of homogeneity (8, 86, 134, 158, 159), with symptoms such as low-grade fever, watery or mucoid diarrhea, borborygmi, and loss of appetite.

In a number of studies, EAEC has been isolated in almost equal numbers from patients suffering from diarrhea and from control groups (15, 133, 137). This finding has been proposed to be a result of acquired immunity (105, 137) due to poor sanitary conditions as well as frequent exposure to pathogenic E. coli strains, resulting in the development of protective antibodies and colonization rather than infection. This has been stated in studies conducted in developing countries. However, high numbers of asymptomatic carriers of EAEC have also been observed in countries such as the United States, Switzerland, and Denmark (40, 72, 109, 142, 143). Case-control studies performed in settings with poor hygiene are difficult to interpret in terms of key EAEC virulence genes, due to the high prevalence of EAEC in these areas. The reported long-term carrier state and the low pathogenicity of EAEC might lead to colonization rather than infection, which could lead to incorrect conclusions about truly pathogenic EAEC virulence genes when EAEC strains are compared between cases and controls. As was mentioned previously (95), the case-control study design performed in a community with a high carrier rate of EAEC complicates the assessment of pathogenicity.

Coinfections, especially in traveler's diarrhea, are said to occur frequently (138); however, few studies have investigated the etiology of diarrhea, including testing for DEC, viruses, and parasites, thus leaving the pathogenicity of EAEC difficult to assess in these cases. In addition, possible synergy between these microorganisms should be investigated.

CONCLUSIONS

EAEC has been isolated from many patients suffering from diarrhea in a number of regions of the world and from various socioeconomic strata. It is most frequently reported as a self-limiting diarrheagenic pathogen that generally results in mild symptoms. The importance of host factors as determinants of EAEC pathogenesis should be investigated in depth. Persistent diarrhea caused by EAEC is often mentioned in studies, but further research on this matter is needed. The genetic heterogeneity of EAEC complicates international surveillance and therapeutic approaches considerably. The lack of a specific diagnostic algorithm to detect EAEC makes our understanding of the pathogenicity of EAEC especially difficult.

Biographies

Betina Hebbelstrup Jensen qualified in medicine at Aarhus University, Denmark, and Copenhagen University, Denmark, in 2010. Part of her medical internship was conducted at the Department of Gastroenterology, Glostrup Hospital, Copenhagen, Denmark. Currently, she is performing her Ph.D. studies at Statens Serum Institut, Department of Microbiology and Infection Control, on enteroaggregative Escherichia coli, with a focus on the clinical implications associated with this microorganism.

Katharina E. P. Olsen, Ph.D., Pharm.D., is the Head of the National Reference Laboratory for Enteropathogenic Bacteria at the Department of Microbiology and Infection Control, Statens Serum Institut (SSI), Copenhagen, Denmark, and is responsible for the national surveillance of Clostridium difficile. Formerly, she held a one-year position as a researcher at the WHO International Collaborating Centre for Reference and Research on Escherichia and Klebsiella, SSI. She was educated at the School of Pharmaceutical Sciences, Faculty of Health and Medical Sciences, Copenhagen University, where she also earned her Ph.D. in medical molecular biology. Dr. Olsen is an officially appointed examiner in Microbiology at Copenhagen University, Faculty of Health and Medical Sciences. She is a member of the Danish Society for Clinical Microbiology, the Danish National Committee under the Nordic Committee on Food Analysis, and the ESCMID Study Group for Clostridium difficile (ESGCD). Her research focuses on molecular epidemiology, diagnostics, and cytotoxicity of bacterial protein toxins.

Carsten Struve, M.Sc., Ph.D., is a Senior Scientist at the Department of Microbiology and Infection Control, Statens Serum Institut, Denmark. He did his M.Sc. studies at the University of Copenhagen and completed his Ph.D. in molecular microbiology in 2002. His research focuses on the identification and characterization of virulence and resistance mechanisms in Enterobacteriaceae, including the development and use of experimental infection models.

Karen Angeliki Krogfelt, M.Sc. (Eng.), Ph.D., is an adjunct professor at the Institute of Systems Biology, Technical University of Denmark, DTU, and Head of a research unit at the Department of Microbiology, Statens Serum Institut. She is involved in teaching master's courses and has supervised numerous master's and Ph.D. students. Her main research activities focus on the pathogenic mechanisms related mainly to intestinal bacterial infections and on assessing the development of gut microbiota and its relation to pathogenesis. Characterization of specific virulence factors involved in pathogenesis is also used for the development of diagnostic methods for bacterial infections.

Andreas Munk Petersen completed his M.D. and Ph.D. in Gastroenterology at the University of Copenhagen. He completed a residency in internal medicine and gastroenterology at Herlev University Hospital, Denmark, before continuing his employment as a consultant at Hvidovre University Hospital, Denmark. His research focuses on the interfaces between microbiology and gastroenterology.