The Centre for Enteric Diseases (CED) of the National Institute for Communicable Diseases in South Africa is a reference center for human infections involving enteric pathogens including DEC, Salmonella spp., Shigella spp., and Vibrio cholerae, and it participates in national laboratory-based surveillance for these pathogens. Isolates are voluntarily submitted to the CED from ≈200 clinical microbiology laboratories across the country. The CED also recovers enteric pathogens through its involvement in the Rotavirus Surveillance Project, which started in mid-2009 and involves 5 sentinel hospital sites in South Africa. This project is involved with identification of enteric pathogens (bacterial, viral, and parasitic) associated with diarrhea in children <5 years of age.

All suspected DEC isolates received at the CED were identified by using standard microbiological identification techniques. Serotyping of O antigen was performed by using tube agglutination as described (7). Presence of H4 antigen was determined by PCR detection of the fliC_H4 gene (8). Resistance to antimicrobial drugs (ampicillin, amoxicillin/clavulanic acid, sulfamethoxazole/trimethoprim, chloramphenicol, nalidixic acid, ciprofloxacin, tetracycline, kanamycin, imipenem, ceftriaxone, and ceftazidime) was determined by using Etests (bioMérieux, Marcy l’Etoile, France), and ESBL activity was investigated by using double-disk synergy methods, as described by Clinical and Laboratory Standards Institute (Wayne, PA, USA) 2009 guidelines (9).

PCR was used to distinguish DEC strains from nonpathogenic E. coli strains. The PCR consisted of 3 multiplex reactions with published primer sequences. Reactions included 0.2 μmol/L of each primer (Table 1) and 1.5 mmol/L MgCl₂. PCR thermal cycling included 35 cycles at 94°C for 1.5 min, 60°C for 1.5 min, and 72°C for 1.5 min.

An isolate was identified as STEC if a PCR result was positive for Shiga toxin genes 1 or 2 (stx1 or stx2); as enterohemorrhagic E. coli if a PCR result was positive for the gene coding intimin outer membrane protein and an stx gene; as enteropathogenic E. coli (EPEC) if a PCR result was positive for the gene coding intimin outer membrane protein; as enterotoxigenic E. coli if a PCR result was a positive for genes coding heat-stable enterotoxin or heat-labile enterotoxin; as enteroinvasive E. coli if a PCR result was positive for the gene coding an invasion protein; as EAggEC if a PCR result was positive for the gene coding a transporter protein; and as diffusely adherent E. coli if a PCR result was positive for the gene coding an accessory protein with a function in F1845 fimbriae production. These genes are listed in Table 1.

Genotypic relatedness of strains was investigated by using a PulseNet protocol (14) for pulsed-field gel electrophoresis (PFGE) of XbaI-digested genomic DNA in a CHEF-DR III electrophoresis system (Bio-Rad Laboratories, Hercules, CA, USA) and the following electrophoresis parameters: voltage 6 V, run temperature 14°C, run time 19 h, initial switch time 6.76 s, final switch time 35.38 s, and included angle 120°. PFGE patterns were analyzed by using BioNumerics version 6.5 software (Applied Maths, Sint-Martens-Latem, Belgium). Dendrograms of patterns were created by using unweighted pair group method with arithmetic averages. Analysis of band patterns incorporated the Dice coefficient at an optimization setting of 1.5% and a position tolerance setting of 1.5%.

During January 2004–May 2011, CED received >4,000 suspected DEC isolates for further laboratory characterization. Of these isolates, 7 (0.2%) were serotype O104. These isolates were collected from Gauteng (n = 3), Mpumalanga (n = 2), and Eastern Cape (n = 1) and North West (n = 1) Provinces of South Africa. E. coli O104 was isolated more often from children <2 years of age than from older children and adults (Table 2). There were more female patients (57%) affected than male patients (43%). Five (71%) of 7 isolates were EAggEC, and 2 (29%) of 7 isolates were EPEC (Table 2). PCR amplification of the fliC_H4 gene showed that all EAggEC isolates produced H4 antigen and that all the EPEC isolates did not produce H4 antigen. Therefore, all EAggEC were serotype O104:H4 and all EPEC were serotype O104:non-H4.

EPEC isolates were susceptible to all antimicrobial drugs tested (Table 2). EAggEC isolates were susceptible to amoxicillin/clavulanic acid, chloramphenicol, nalidixic acid, ciprofloxacin, kanamycin, imipenem, ceftriaxone and ceftazidime; were resistant to ampicillin and sulfamethoxazole/trimethoprim; and were variably susceptible to tetracycline and streptomycin (Table 2). None of the EPEC or EAggEC isolates showed ESBL activity.

Dendrogram analysis of PFGE patterns showed that EPEC isolates were diverse and clustered at a pattern similarity of 76%, and that EAggEC isolates were highly clonal and clustered at a pattern similarity of 90% (Figure). The PFGE pattern of the 2011 outbreak strain from Germany did not match those of strains from South Africa. Therefore, the outbreak strain from Germany was determined to be unrelated to strains from South Africa. However, the strain from Germany was most closely related to the EAggEC strain cluster from South Africa (pattern similarity 85%).

Our findings show that E. coli O104 is rarely associated with human diarrhea in South Africa and accounts for <1% of all DEC pathotypes identified during 2004–2011 by laboratory-based surveillance and limited sentinel surveillance. Strains of E. coli O104 from South Africa were mostly associated with EAggEC pathotypes; most (5/7) were identified as EAggEC. Infection with this pathotype has been associated with more persistent diarrhea (duration >14 days) (15). Therefore, patients infected with EAggEC are more likely to have fecal cultures tested, potentially leading to greater numbers of EAggEC isolates identified.

In South Africa, E. coli O104 infections were more commonly identified in children than in adults. Unlike the E. coli O104 strain that caused the outbreak in Germany, strains of E. coli O104 from South Africa did not produce Shiga toxin and did not show ESBL activity. PFGE data supported these phenotypic data, suggesting that strains from South Africa were not related to the outbreak strain from Germany. The PFGE data also showed that strains of EAggEC O104:H4 from South Africa were highly clonal. Further work is necessary to better understand the global distribution of these isolates and the role of molecular epidemiologic techniques in characterizing this newly emerging serotype.