Rectal (livestock) and cloacal (bird) samples were collected from a total of 35 different livestock [goat (n = 19), sheep (n = 9), cattle (n = 2)] and poultry [chicken (n = 3) and duck (n = 2)] between June and August 2007 from the district of Mymensingh, Bangladesh. Sterile cotton swabs were used for the collection of rectal and cloacal samples. After collection, the cotton swabs were dipped into Carry-Blair media and stored in a cool box (4°C). Immediately after sampling, all samples were transported to the laboratory for analysis. All samples were streaked onto MacConkey agar (Difco), and the plates were incubated overnight at 37°C. Bright pink lactose-fermenting colonies were selected as presumptive E. coli. These colonies were again grown on eosin methylene blue agar to examine for the production green colonies with a metallic sheen. One presumptive colony per sample was identified by biochemical tests, as described before (29).

DNA was extracted from isolated colonies suggestive of E. coli, as described previously (30). The samples were then subjected to molecular analysis of ETEC (sta, stb, elt) and STEC (stx1, stx2) virulence genes using multiplex PCR. The PCR assay conditions are described in Table 1. Each amplification was conducted in a volume of 20 µl containing the following reagents: 1 µl template DNA, 0.5 µl dNTP (Invitrogen, Carlsbad, CA, USA) solution (10 mmol 1⁻¹), 1.0 µl of each primer at 5 pmol l⁻¹, 0.6 µl Taq DNA polymerase solution (Invitrogen) (1.5 U μl⁻¹), 2.0 µl PCR reaction buffer solution (10×) with 0.8 µl MgCl₂ (50 mmol l⁻¹), and presterilized ultrapure water (Milli-Q) to 20 µl. This mixture was processed in a thermocycler at 94°C for 4 min (denaturation), followed by 30 cycles at 94°C for 1 min (denaturation), annealing temperature (specific for each primer) for 1 min (binding), and at 72°C for 1.5 min (extension). Complete extension of the Taq DNA polymerase was performed at 72°C for 7 min. An aliquot of this reaction containing only water without DNA was used as a negative control. Two multiplex PCR tests were carried out; one for the primer set LT, STa, and STb and the other for the primer set Stx1 and Stx2.

The rabbit antisera against O1 to O187strains were prepared in rabbits (SERUNAM) according to procedures described by Ewing (31). The E. coli O1 to O172 strains were obtained from the Laboratory of Gastrointestinal Pathogens, Department of Gastrointestinal, Emerging and Zoonotic Infections, Centre for Infections, Health Protection Agency, London, UK, and the E. coli O173 to O186 strains (32) were obtained from the Statens Serum Institute, Copenhagen.

The strains were serotyped by agglutination assay (33) using 96-well micro titer plates and rabbit serum (SERUNAM) obtained against 187 somatic antigens and 53 flagellar antigens for E. coli.

Susceptibility to antibiotics was performed by disk diffusion, as described by Bauer et al. (34) and the Clinical and Laboratory Standards Institute (35) (CLSI), using commercial antibiotic discs. Thirteen antibiotics (Oxoid, UK) were employed: erythromycin (E, 15 µg); gentamicin (CN, 10 µg); trimethoprim/sulfamethoxazole (SXT, 30 µg), tetracycline (TE, 30 µg), ampicillin (AMP, 30 µg), streptomycin (S, 10 µg), azithromycin (AZM, 15 µg), nalidixic acid (NA, 30 µg), ciprofloxacin (CIP, 5 µg), ceftriaxone (CRO, 30 µg), cefixime (CFM, 5 µg), mecillinam (MEL, 25 µg), and cephalothin (KF, 30 µg). Characterizations of the resistance or susceptibility profiles of the isolates were determined by measuring the inhibitory zone and comparing it with an established interpretative chart to determine sensitivity to each antibiotic.

Whole agarose-embedded genomic DNA from the E. coli isolates was prepared. PFGE was carried out using a contour-clamped homogeneous electrical field (CHEF-DRII) apparatus (Bio-Rad), according to procedures described previously (36). Genomic DNA of the test strains and Salmonella enterica serovar Braenderup was digested using XbaI, with fragments employed as molecular size markers. Restriction fragments were separated in 1% pulsed-field-certified agarose in 0.5× TBE (Tris–borate–EDTA) buffer. Post-electrophoresis gel treatment included both gel-staining and de-staining. The DNA was visualized using a UV transilluminator, and images were digitized via a one-dimensional gel documentation system (Bio-Rad).

The fingerprint pattern in the gel was analyzed using a computer software package, Bionumeric (Applied Maths, Belgium, version 3.1). After background subtraction and gel normalization, the fingerprint patterns were typed according to banding similarity and dissimilarity, using the Dice similarity coefficient and unweighted-pair group method employing average linkage (UPGMA) clustering, as recommended by the manufacturer. The results were graphically represented as dendrograms.

All the E. coli strains (n = 35) included in this study produced bright pink colonies on MacConkey and green colonies with metallic sheen on EMB agar plates. All strains gave biochemical reactions typical of E. coli.

Among the 35 E. coli strains isolated from livestock and poultry in the present study, 23 (66%) (goats: 17, sheep: 4, chicken: 1, and ducks: 1) carried virulence genes, such as, sta and stx, either stx1 or stx2, but did not carry stb or elt genes, as confirmed by PCR (Table 2). Of the 23 strains carrying toxin genes, 5 (22%) had sta, and 10 (44%) had stx genes. Of the 10 E. coli carrying stx gene, 8 (80%) harbored only stx1, and 2 (20%) had only stx2. Eight of the 23 (34%) toxigenic strains were hybrid of which 7 (88%) possessed both sta and stx1, while the remaining strain carried both sta and stx2 (Table 2). None of the strains tested presented elt or stb.

Based on serotyping, the 23 toxigenic E. coli strains belonged to 10 different O serogroups [O8 (n = 1), O36 (n = 1), O43 (n = 4), O76 (n = 6), O87 (n = 3), O109 (n = 1), O110 (n = 1), O152 (n = 3), O174 (n = 1), and OR (n = 2)] and 8 different H seogroups [H2 (n = 6), H5 (n = 1), H8 (n = 3), H16 (n = 4), H19 (n = 6), H21 (n = 1), H51 (n = 1), and H− (n = 1)]. E. coli strains carrying virulence genes belonged to 10 different serotypes: O36:H5 (n = 1), O43:H2 (n = 4), O76:H19 (n = 6), O87:H16 (n = 3), O174: H− (n = 1), OR: H2 (n = 2), O110:H16 (n = 1), O152:H8 (n = 3), O109:H51 (n = 1), and O8:H21 (n = 1). None of the tested strains belonged to the EHEC serogroup, as they did not agglutinate with the antisera specific for O157 (Table 2). The ETEC strains (n = 5) belonged to five different serotypes: O36:H5, O174: H−, O152:H8, O109:H51, and O8:H21, while the STEC strains (n = 10) belonged to six different serotypes: O76:H19 (30%), O43:H2 (20%), O87:H16 (20%), OR: H2 (10%), O110:H16 (10%), and O152:H8 (10%) (Figure 1). In the order of prevalence of the STEC–ETEC hybrid strains (n = 8), O76:H19 was the most prevalent (37.5%) serotype, followed by O43:H2 (25%), O87:H16 (12.5%), OR: H2 (12.5%), and O152:H8 (12.5%) (Figure 1).

Antibiotic susceptibility assay was performed for the 23 toxigenic strains only, and the results showed 87% of the strains to be resistant to erythromycin, 43% to azithromycin, 22% to streptomycin, 22% to trimethoprim/sulfamethoxazole, 17% to cephalothin, 9% to ciprofloxacin, 9% to nalidixic acid, and 4% to tetracycline (Table 2). All E. coli strains were sensitive to gentamicin, ampicillin, ceftriaxone, cefixime, and mecillinam. Results also revealed 40% (2/5) of the ETEC and 20% (2/10) of the STEC strains to be multidrug resistant (MDR) showing resistance to erythromycin, trimethoprim/sulfamethoxazole, azithromycin, cephalothin, ciprofloxacin, and nalidixic acid. The highest MDR was found in the hybrid strains, accounting for 50% (4/8) (Table 2), although the strains varied in their patterns of response to the different drugs tested.

Overall drug response results revealed 11 different resistance patterns among the three pathogroups of E. coli (ETEC, STEC, and hybrids of STEC–ETEC). Resistance patterns I (E^R, S^R), II (E^R, AZM^R), III (E^R, SXT^R, AZM^R, KF^R), and IV (E^R, CIP^R, SXT^R, NA^R, AZM^R) were found in ETEC; resistance patterns V (E^R), VI (E^R, AZM^R, S^R, KF^R), and VII (E^R, SXT^R, AZM^R, S^R) in STEC; and pattern V (E^R), VIII (E^R, KF^R), IX (E^R, AZM^R, S^R), X (E^R, SXT^R, AZM^R), and XI (E^R, CIP^R, SXT^R, NA^R, AZM^R, TE^R, KF^R) among the STEC–ETEC hybrid pathotypes (Table 3).

The XbaI-digested genomic DNAs of the E. coli strains carrying virulence genes from animal and poultry were subjected to PFGE to determine genetic relatedness and clonal origin. The number of fragments generated by restriction digestion with XbaI varied between 14 and 19, and the molecular size of the fragments ranged from 30 to 600 kb. Diverse PFGE patterns obtained from the 23 toxigenic E. coli strains (Figure 2) belonged to multiple small sub-clusters (Figure 2) when dendrograms were prepared with PFGE images by the BioNeumeric software (Applied Maths) using dice similarity coefficient and UPGMA. Although the E. coli strains exhibited high genetic heterogeneity in the PFGE patterns, four hybrid strains belonging to serotype O76:H19 formed a small tight cluster, suggesting that genetically, they are closely related.

Enterotoxigenic E. coli and Shiga toxin-producing E. coli (STEC) are notorious pathogens associated with severe diarrhea and HUS in humans. Although E. coli constitutes an important component of normal gut bacterial flora, many studies have shown the non-toxigenic strains to harbor a number of different virulence factors and to be diarrheagenic (17, 18, 37). Here, we present data showing the prevalence of E. coli strains carrying an ETEC-specific gene, sta, and STEC-specific gene, stx, which included either stx1 or stx2 in the gut of livestock and poultry in Bangladesh. This study also presents data showing for the first time that a significant proportion (8/23) of the toxigenic E. coli strains from livestock were MDR and hybrid type carrying both stx (either stx1 or stx2) and ETEC-specific enterotoxin gene, sta. The presence of such a high percentage of MDR hybrid type E. coli strains in livestock poses a potential health threat to people of Bangladesh.

Results obtained in this study confirmed that a very high proportion of the E. coli strains (66%) deriving from different livestock are toxigenic, carrying either sta, stx, or both in varying combinations. What is most important to note is that a significant proportion of E. coli in livestock was confirmed to be ETEC, including 22% carrying the sta gene only, 44% STEC (harboring only stx gene; either stx1 or stx2), and 35% hybrid comprising both est and stx. ETEC is an important cause of diarrhea in humans and animals in the low-income countries and is the main cause of diarrhea in travelers to low-income countries (1). One study carried out in Spain reported a high incidence of ETEC (30%) in piglets with diarrhea (38). A study carried out in Bangladesh revealed 61% of the E. coli strains isolated from the aquatic environment carried the est gene (39). Zoonotic and environmental transmission may be the reason ETEC is the predominant group of E. coli associated with childhood diarrhea in Bangladesh (28). The occurrence of ETEC in farm animals has always been of great public health importance as animals excreting pathogenic bacteria can contaminate water bodies in and around the area they are reared and in many ways can be related to human diseases. This may be one reason why ETEC is largely a pathogen of low-income countries where access to safe drinking water is often low.

STEC strains can cause bloody diarrhea known as haemorrhagic colitis, non-bloody diarrhea, and HUS (1). The data presented in this study provide evidence that livestock is an important reservoir for STEC carrying either stx1 or stx2. A study carried out in Bangladesh reported 37.9% of buffalo, 20.1% of cows, and 10.0% of goats were positive for STEC (40). Our data appear to concur with 44% (n = 10) of livestock E. coli isolated mostly from goats (80%, 8/10 or 35%, 8/23) revealed STEC, either stx1 or stx2. Our results also appear to agree with those from a study carried out in Vietnam that showed STEC to account for 27% of E. coli isolated from buffalos, 23% from cattle, and 38.5% from goats (41). Although water plays an important role in the transmission of pathogenic bacteria, a study carried out in Bangladesh reported the presence of stx only in 1.25% of the E. coli strains isolated from natural aquatic environments (39). STEC-associated diarrhea is also very low compared to that of other enteric pathogens, including Vibrio cholerae and ETEC in Bangladesh (27). This low prevalence of STEC as the etiology of diarrhea could be due to STEC-specific antibodies in the human population, as this pathotype is present in a high proportion of E. coli occurring in domesticated livestock, such as goats and sheep, although it was not found in poultry, and only rarely in surface water samples (39). The low prevalence in natural surface water could be related to the incompetence of the STEC strains to survive beyond their livestock reservoirs (cattle, sheep, and goats) (6). Despite the low prevalence of STEC infections in Bangladesh (27), the virulence potential of these livestock strains to emerge as a novel STEC pathogen cannot be ruled out.

In the present study, 34% of the toxigenic livestock E. coli strains were hybrid as those carried toxin genes present in two different pathotypes; stx and est. Notably, majority of these hybrid strains (7/8) carried both sta and stx 1, while one strain carried both sta and stx2 genes. This finding is not unique to our study as a number of past studies have shown hybrid strains carrying virulence marker genes of two different E. coli pathotypes (17), nonetheless, in a significantly lower frequency (2%) than that we observed in Bangladesh. Clinically linked STEC–ETEC hybrid E. coli strains from Brazil, Denmark, France, Germany, Great Britain, Mexico, and the United States were shown to harbor virulence markers for two or three different pathotypes (17). In a study in Germany, 0.6% of the human STEC isolates possessed stx2g and estIa (37). Serotype O100:H of a STEC–ETEC hybrid strain carrying stx2 and estIa genes was isolated from contaminated drinking water during an outbreak in Finland (22). A later study reported a STEC strain carrying stx1a and estIa from cattle in Burkina Faso (42). Although no data exist on the virulence potential of STEC–ETEC hybrid strains isolated from livestock in Bangladesh, the widespread distribution and clinical relevance worldwide might indicate their virulence potential.

In the present study, serotyping results revealed that the ETEC strains belonged to five different serotypes: O36:H5, O174:H−, O152:H8, O109:H51, and O8:H21. No sero-specificity was observed for these ETEC strains as there was no relationship between the serotypes and the type of toxin genes (st or lt) they possessed. Likewise, the STEC strains belonged to six different serotypes (O76:H19, O43:H2, O87:H16, OR: H2, O110:H16, and O152:H8); STEC O157:H7 serotype was not present among the tested strains. Moreover, the STEC serotypes did not belong to any of the serotypes of the big six groups (O26, O45, O103, O111, O121, and O145) in the present study (43). STEC non-O157 isolates have been shown to be important pathogens despite having been underreported because in many laboratories, the facilities to isolate, identify, and characterize this pathotype do not exist (40). One report suggested that some non-O157:H7 STEC strains cause human illness and likely account for 20–50% of STEC infections (an estimated 37,000 cases) annually in the United States (44). In our study, we have identified eight E. coli strains that were hybrids and carried virulence genes STEC–ETEC, which are largely restricted to five serotypes (O76:H19, O43:H2, O87:H16, OR: H2, and O152:H8). Several other studies reported STEC–ETEC hybrid strains; however, unlike our study, the serotypes were O2:H27, O101:H−, O15:H16, O74:H28, O116:H28, O128:H8, O136:H12, O141:H8, O168:H8, OX182:H16, OX182:H− in Finland (23), O139, O149, O116, and OSB9 in Japan (24), O100:H− in United States (21), and O2:H2 in Burkina Faso (42).

Resistance to antimicrobial agents can be a useful epidemiological marker for ETEC, STEC, and STEC–ETEC hybrid strains. In this study, the highest antimicrobial resistance was found against erythromycin 87% (20/23) followed by azithromycin 43% (10/23). In the present study, a significant proportion of ETEC and STEC strains were MDR with the highest MDR [50% (4/8)] being observed in STEC–ETEC hybrid strains. Since 1950s, antibiotics have been used in animal feed to ensure healthy livestock, prevent infections, and excel growth. Such indiscriminate use of antibiotics resulted in the widespread antibiotic resistance in bacteria (21). Erythromycin resistance during the 1990s compelled clinicians to switch to azithromycin as the drug of choice in the treatment of cholera (45) and watery (cholera-like) diarrhea caused by ETEC (46, 47).

As shown earlier by Vu-Khac et al. (48), a high degree of genetic polymorphisms was observed from the PFGE profiles of the STEC non-O157 and ETEC strains in the present study. Similarly, a high degree of heterogeneity was also observed with the STEC–ETEC hybrid strains, as the pulsotypes were different for strains belonging to the same serotypes, suggesting that they were unlikely to be of a single ancestral origin. A few strains showing identical PFGE patterns, however, suggest close clonal relatedness.