|Home | About | Journals | Submit | Contact Us | Français|
Enteropathogenic Escherichia coli (EPEC) is a leading cause of infantile diarrhoea in developing countries. The aim of this study was to describe the allelic diversity of critical EPEC virulence genes and their association with clinical characteristics. One hundred and twenty EPEC strains isolated from a cohort diarrhoea study in Peruvian children were characterized for the allele type of eae (intimin), bfpA (bundlin pilin protein of bundle-forming pilus) and perA (plasmid encoded regulator) genes by PCR-RFLP. Atypical EPEC strains (eae+, bfp−) were the most common pathotype in diarrhoea (54/74, 73%) and control samples from children without diarrhoea (40/46, 87%). Overall, there were 13 eae alleles; the most common were beta (34/120, 28%), theta (24/120, 20%), kappa (14/120, 12%) and mu (8/120, 7%). There were five bfpA alleles; the most common were beta1/7 (10/26), alpha3 (7/26) and beta5 (3/26). There were three perA alleles: beta (8/16), alpha (7/16) and gamma (1/16). The strains belonged to 36 distinct serogroups; O55 was the most frequent. The gamma-intimin allele was more frequently found in diarrhoea episodes of longer duration (>7 days) than those of shorter duration (3/26, 12% vs 0/48, 0%, P<0.05). The kappa-intimin allele had the highest clinical severity score in comparison with other alleles (P<0.05). In Peruvian children, the virulence genes of EPEC strains are highly variable. Further studies are needed to evaluate additional virulence markers to determine whether relationships exist between specific variants and clinical features of disease.
Enteropathogenic Escherichia coli (EPEC) is an important diarrhoeal pathogen of young children living in developing countries (Scaletsky et al., 2002). A hallmark phenotype of EPEC is the ability to produce attaching and effacing (AE) lesions. In the AE lesion, the bacteria attach tightly to the host cell membrane causing a dramatic disruption of the cell surface leading to distortion of microvilli and close adherence of bacteria to the enterocyte. All of the genetic elements required for the production of the AE lesion are encoded in a 35 kb chromosomal pathogenicity island called the locus of enterocyte effacement (LEE). The intimate attachment is mediated by the outer-membrane protein called intimin (Jerse et al., 1990). Intimin is highly variable in the carboxy-terminal region, known as Int-280, and several methods have been developed to recognize its alleles. Ramachandran et al. (2003) designed universal PCR primers to amplify the Int-280 region and identified types by RFLP. Others have used microarrays (Garrido et al., 2006) or a heteroduplex mobility assay (Ito et al., 2007). To date, more than 25 major allelic variants of eae have been described (Lacher et al., 2006).
EPEC is classified into typical and atypical strains based on the presence of the plasmid E. coli adherence factor (EAF). There are two important operons on this plasmid, bfp and per, the first encoding the type IV bundle-forming pilus (Bfp) and the second encoding a transcriptional activator called plasmid-encoded regulator (Per) which is required for optimal activation and function of LEE-encoded genes and Bfp expression. Lacher and co-workers found 11 distinct EAF plasmid types based on combinations of bfpA and perA alleles (Lacher et al., 2007). Typical EPEC strains produce the localized adherence (LA) phenotype, associated with the production of Bfp (Giron et al., 1991). Bfp is responsible for the formation of microcolonies on cultured cells (Giron et al., 1991) and is associated with virulence in experimental infection in human volunteers (Bieber et al., 1998).
Little is known about the allelic distribution of eae, bfpA and perA or about whether specific alleles are related to specific clinical characteristics. Most studies have only described the allele distribution between diarrhoea and control samples (Knutton et al., 2001; Vieira et al., 2001; Jenkins et al., 2006; Kozub-Witkowski et al., 2008). In this study, we characterized a collection of strains obtained from a cohort study in Lima, Peru, using PCR-RFLP analysis to describe the allelic diversity, the relationships between alleles of eae, bfpA and perA, and to determine whether clinical features of illness were clearly related to specific alleles.
The strains examined in this study were isolated from a cohort epidemiological study of diarrhoea in infants from 2 to 12 months of age. The study was conducted in peri-urban communities of Lima, Peru, from September 2006 to December 2007 (Ochoa et al., 2009). In the study, 1065 stool samples were obtained from 1034 children; 936 samples were from children with diarrhoea (case patients) and 424 were from children without diarrhoea (controls). Samples were examined for enteric pathogens (Shigella, Salmonella, Vibrio, Campylobacter, Giardia lamblia, Cryptosporidium and rotavirus) by conventional methods (Murray, 2007). Five lactose-positive colonies were isolated from MacConkey plates and tested with specific DNA primers designed to detect enterotoxigenic E. coli, enteroinvasive E. coli, enterohaemorrhagic E. coli, enteroaggregative E. coli, diffusely adherent E. coli and EPEC, as described previously (Guion et al., 2008). The number of positive EPEC colonies per sample varied from 1 to 5, but for the present study, only one eae+ colony per sample was selected for further analysis.
Clinical information on the diarrhoeal episodes was obtained from the cohort study. We used a modified Vesikari score (Ruuska & Vesikari, 1990) to determine the severity of the episode. The score included: duration of diarrhoea in days (0–3 points), maximum number of stools per day during the episode (1–3), presence of vomiting (0–1), maximum number of emesis per day during the episode (0–3), fever (0–1), dehydration (0–1) and treatment (0–2). The maximum possible score was 14.
Serotyping was performed at the E. coli Reference Center, Pennsylvania State University, according to standard methods for determining the O antigen (Orskov et al., 1977). H typing was performed using a fliC PCR-RFLP method (Machado et al., 2000).
EPEC isolates were examined for eae, bfpA and perA by PCR with the primers and conditions listed in Supplementary Table S1 (available in JMM Online). PCR for the three genes was performed in a 25 μl reaction mixture containing 2.5 μl 2.5 mM of each dNTP (Promega), 2 μl 25 mM MgCl2, 0.5 μl 20 μM of each primer (Invitrogen), 2.5 μl 10× buffer II (Applied Biosystems), 1 U (for bfpA and perA) or 1.5 U (for eae) AmpliTaq Gold (Applied Biosystems) and 2 μl DNA template. Amplification reactions were performed in a thermo cycler (iCycler; Bio-Rad) and, for all amplification reactions, the mixture was heated at 94 °C for 10 min prior to thermocycling. The mixture was held at 72 °C for 7 min after the final cycle before cooling at −20 °C. Amplified products were analysed by using 1.5% agarose gel electrophoresis and visualized by staining with ethidium bromide. A 100 bp Plus DNA ladder (Fermentas) was used as a molecular size marker in all gels. The positive PCR control was EPEC strain E2348/69 (eae+ bfpA+) and the negative control was E. coli C600.
Primers for intimin (eae) were designed by Lacher et al. (2006) to amplify 1.8–2.0 kb of the 3′ end of eae to the 3′ end of escD for all known intimin subtypes. This region of eae encodes the C-terminal 280 amino acids and was used for RFLP analysis because it possesses the greatest degree of sequence variation between subtypes.
The in silico RFLP tool of the EcMLST website (http://www.shigatox.net/cgi-bin/mlst7/insilicorflp) was used to predict the restriction patterns of the alleles (Supplementary Table S2, available in JMM Online). Two different restriction enzyme digests were used for eae. Digestion with AluI was performed in 30 μl reaction mixtures with 10 U enzyme (1 μl) (New England BioLabs), 3.0 μl 10× reaction buffer, 20.0 μl unpurified PCR product and 6.0 μl distilled water; the samples were incubated overnight at 37 °C. Digestion with BstNI was performed in 30 μl reaction mixtures with 10 U of enzyme, 3.0 μl 10× reaction buffer, 0.3 μl 100× BSA, 20.7 μl unpurified PCR product and 5.0 μl distilled water followed by an overnight incubation at 60 °C. The bfpA and perA digests were performed as described previously (Lacher et al., 2007). Briefly, bfpA PCR products were digested with three separate restriction enzymes (AluI, BstNI and BfaI), while perA amplicons were digested with two separate enzymes (DdeI and Sau96I). After incubation, 15 μl of the digests was separated on ethidium bromide-stained 3.0% agarose gels and visualized by illumination with UV light. The positive controls for eae, bfpA and perA RFLP patterns were provided by Dr M. S. Donnenberg.
The allelic frequencies obtained in each group and the allelic distributions of each population were compared using the program GenALEx 6. The comparisons between groups were made using Chi-squared or Fisher's exact test. Student's t-test was used for the comparison between severity scores.
Although many studies have found that EPEC infection is significantly associated with infant diarrhoea, EPEC strains were isolated with similar frequency from diarrhoeal samples (74/936, 8%) and from healthy controls (46/424,11%) in this study, as also described elsewhere (Ochoa et al., 2008). The frequency of asymptomatic colonization might relate to the age of the children and the high frequency of breastfeeding in this age group. Colonization rather than illness may result from the complex interaction of multiple factors, including host susceptibility to infection (the child's age, presence of protective maternal factors such as breastfeeding and trans-placental antibodies, and nutritional and immunological status), and environmental factors (poor hygiene and high faecal contamination). Atypical EPEC (94/120, 78%) constituted the majority of eae-positive isolates in this study and were common both in patients with diarrhoea (54/74, 73%) and in controls (40/46, 87%), as has been recently reported in developing and developed countries (Knutton et al., 2001; Afset et al., 2003, 2004; Blanco et al., 2006a, b; Ochoa et al., 2008). Overall, atypical strains were significantly more common in controls (40/424, 9%) than in diarrhoeal cases (54/936, 6%) (P<0.05). The prevalence of perA among typical EPEC strains was 55% (11/20) in diarrhoeal patients and 83% (5/6) in controls. Among atypical EPEC strains, the prevalence of perA was 7% (4/54) in diarrhoeal patients and 3% (1/40) in controls. The low prevalence of perA in typical strains is uncommon and may have two possible explanations: the primers used in the PCR may have failed to recognize new alleles; alternatively, perA may not have an important role in infection or other regulators may compensate for its absence.
The role of atypical EPEC in childhood diarrhoea remains to be defined. The association of these strains with diarrhoea is not completely clear, since some studies have reported that atypical EPEC was significantly associated with endemic diarrhoea (Scaletsky et al., 1999; Vieira et al., 2001; Dulguer et al., 2003) and diarrhoea outbreaks (Hedberg et al., 1997; Yatsuyanagi et al., 2002; Jenkins et al., 2003); however, others have reported no significant association with diarrhoea (Knutton et al., 2001; Afset et al., 2004). In our study, atypical EPEC strains were actually significantly more common in control than in ill patients; however, many of these previous studies have demonstrated a tendency for the infants with diarrhoea to have a higher prevalence of atypical EPEC than typical EPEC. Atypical EPEC strains appear to be a diverse group expressing other virulence factors (i.e. afa, afimbrial adhesion; EAST1, heat-stable toxin 1; E-hly, EHEC haemolysin; efa/lifA, EHEC factor for adherence/lymphocyte inhibitory factor) that may compensate for the loss of EAF (Trabulsi et al., 2002; Afset et al., 2006). Atypical EPEC strains are capable of colonizing the intestinal mucosa and producing AE lesions (Knutton et al., 2001).
Typical EPEC strains are more homogeneous in their virulence characteristics than atypical strains (Trabulsi et al., 2002). bfpA is an important virulence determinant of typical EPEC; perA may not be (Vieira et al., 2001; Dulguer et al., 2003). We have not found significant differences between the clinical characteristics of perA+ and perA− strains.
Seven strains (6%) (four atypical and three typical EPEC) were non-typable for intimin based on the RFLP pattern (Supplementary Table S2). Of the typable intimins, 13 different alleles were found. Beta-intimin was the most common subtype (28%) and was found with similar prevalence in diarrhoeal patients (31%) and controls (24%). The other prevalent alleles were theta (20%), kappa (12%) and mu (7%) (Tables 1 and 22).). The intimin alleles beta, theta, gamma and mu are most frequently found among EPEC isolated from humans (Dulguer et al., 2003; Blanco et al., 2006a, b; Jenkins et al., 2006; Kozub-Witkowski et al., 2008). The distribution of intimin alleles found in this study is similar to that obtained in samples from other Latin American countries, although the order of frequency is not the same in all studies (Blanco et al., 2006b; Dulguer et al., 2003). Intimin alleles alpha2, eta, nu, rho and tau were not found in this study. There were no significant differences in the distribution of intimin alleles between diarrhoea and control samples or in their distribution between typical and atypical strains using GenAlEx 6.
The strains studied belonged to 36 distinct serogroups, of which O55 was the most frequent (six from children with diarrhoea and two from the control group). Only 62/120 (52%) strains were O typable. Sixty-nine different serotypes were found (Tables 1 and 22).). Among these, we found four serotypes (O128:H2, O55:H7, O26:H1 and O142:H34) previously described as classic EPEC strains (Trabulsi et al., 2002). Serotype distribution of EPEC strains varied considerably, as in other studies, and did not correlate with specific intimin types (Vieira et al., 2001; Dulguer et al., 2003). In general, efforts to correlate serotype with pathogenesis have yielded disappointing results.
Of the ten bfpA alleles cited in the literature (Lacher et al., 2007), four were found in these strains: beta1/7 38% (10/26), alpha3 27% (7/26), beta5 12% (3/26) and alpha1 8% (2/26). Strains with the beta1/7-bfpA were found only in children with diarrhoea [50% (10/20) vs 0% (0/6), P<0.05]. For perA, three alleles were found: beta 31% (8/26), alpha 27% (7/26) and gamma 4% (1/26). The majority of strains (10/26, 38%) were negative for perA by PCR, whereas beta-perA tended to be more frequent in EPEC from controls samples [20% (4/20) vs 67% (4/6), P<0.05]. In general, the diversity in the eae, bfpA and perA alleles found was similar to that reported by Lacher et al. (2007). In this study, we found 12 EAF types, including seven new combinations of bfpA and perA alleles, relative to the ones described by Lacher et al. (2007) (Table 3). We also found that the alpha1-bfpA allele was associated with two different perA alleles (beta and alpha). In contrast, Lacher et al. (2007) found that each bfpA allele was associated with only one perA allele class.
There were no clear age-related differences in intimin alleles (data not shown). For both typical and atypical EPEC, most diarrhoeal episodes were acute (≤7 days) 65% (48/74), 27% (20/74) lasted 8–13 days and 8% (6/74) were persistent (≥14 days). Among acute episodes, eae alleles beta (25%; 12/48), theta (23%; 11/48) and kappa (10%; 5/48) were the most prevalent in both typical and atypical cases. Among the episodes of longer duration, beta-intimin was also the most prevalent (42%; 11/26) followed by kappa (19%; 5/26). Among isolates from the six persistent diarrhoea episodes, there were three beta and one each of theta, kappa and gamma. Gamma-intimin was more frequently found in strains from diarrhoea episodes of longer duration (>7 days) than shorter episodes, with frequencies of 12% (3/26) vs 0% (0/48), respectively, P<0.05.
To determine whether clinical characteristics of the diarrhoea episodes might be related to specific intimin alleles, only infections due to a single pathogen were evaluated. Overall, severity, as suggested by the modified Vesikari score, tended to be worse in children infected with EPEC strains carrying beta- and kappa-intimin compared with other eae alleles (Table 4). The correlation between intimin type and diarrhoea severity could be relevant to pathogenesis via either the intensity or the specific targeting of bacterial adherence. It has been suggested that tissue tropism is influenced by intimin type (Fitzhenry et al., 2002, 2003); this could explain the association of beta-intimin with more prolonged diarrhoea episodes and perhaps could relate to the severity of the episodes observed in beta- and kappa-intimin. However, other virulence factors are likely to play an important role in the expression of disease. Vieira et al. (2001) analysed the virulence potential of atypical EPEC and concluded that a lack of expression of the LEE genes related to virulence. Interpreting the possible relationships between intimin or other virulence genes and clinical illness could be complicated by many factors including protective factors in breast milk, previous infection with related intimin types, age of infection, genes associated or not associated with intimin in a given strain, and genetic predisposition of the host.
In summary, in the cohort of Peruvian infants, the virulence genes of the EPEC strains isolated were highly variable. Some intimin alleles were associated with diarrhoeal episodes of longer duration and higher clinical severity scores. Further studies are needed to evaluate additional virulence markers to determine which factors are crucial for disease.
This work was funded by Public Health Service awards 1K01TW007405 (T.J.O.), R01-HD051716 (T.G.C.) and R01 AI-37606 (M.S.D.) from the National Institutes of Health; it was partially funded by C.F.L.'s Institutional Research Funds. The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of the US Government, nor the National Institutes of Health or other funding institutions. There is no conflict of interest for any of the authors.
Two supplementary tables are available with the online version of this paper.