Enteropathogenic, enterotoxigenic, enteroinvasive, enterohaemorrhagic and enteroaggregative Escherichia coli are categories of enteric E. coli that have been unequivocally associated with diarrhoeal disease through human challenge studies and/or outbreak investigations [1]. Regarding other potentially diarrhoeagenic categories of E. coli, the most evidence for enterovirulence has been compiled for diffusely adherent E. coli (DAEC). However, the basis for DAEC pathogenicity is not well understood. The category is heterogeneous and although some studies have shown an association of DAEC with diarrhoea, others have not [2]. Two DAEC strains did not elicit diarrhoea upon human volunteer challenge and no outbreaks of DAEC-associated illness have been documented to date [3].

Enteroaggregative E. coli (EAEC) is another heterogeneous diarrhoeagenic E. coli category. Convincing epidemiological information from EAEC outbreaks exists, and at least one strain was diarrhoeagenic in some human volunteers, however the category is very diverse (reviewed in references [4] and [5]). Compared to other diarrhoeagenic E. coli categories, EAEC and DAEC pathotypes were both described relatively recently and their epidemiology, risk factors and pathogenesis are still in early stages of investigation. Few epidemiological studies seek these categories because the Gold Standard test for their detection, the HEp-2 adherence assay, is cumbersome. This tissue culture-based test requires expensive facilities and technical expertise that are not universally available.

An improved understanding of the importance of diarrhoeagenic E. coli in human disease will depend upon reliable epidemiological data and on channelling of strains identified into molecular pathogenesis research. Accordingly, efforts have been made to develop more widely applicable methods to detect EAEC and DAEC. Baudry et al. tested fragments from the large plasmid of EAEC strain 17-2 and identified a 1 Kb fragment, CVD432, which was 89% sensitive and 99% specific for EAEC strains in their collection [6]. Subsequently, this probe has continued to show specificity for EAEC but its sensitivity has varied between 15 and 90% in different studies [4]. Bilge et al. [7] used a different approach to generate a diagnostic probe for DAEC. They identified, cloned and characterized the F1845 adhesin from DAEC strain C1845. The F1845 adhesin belongs to the Afa/Dr family and is encoded by a five-gene cluster [2]. Bilge et al. [7] proposed part of the daaC gene of the encoding operon as a marker for DAEC strains. From the time of its discovery, it has been known that the cloned daaC fragment probe (in plasmid pSLM862) can only identify a subset of DAEC and that some DAEC strains have other adhesins, of which many, but not all, are from the Afa/Dr family [2]. However, the daaC probe is the one that has been employed most frequently in epidemiological research to date 8-13. In this paper, we report that the daaC cross-hybridizes with a specific subset of EAEC strains. We sought to identify the molecular basis for this cross-hybridization and to devise an alternate, cost-effective protocol for identifying DAEC.

There have been brief mentions of daaC hybridization with EAEC in the literature. In some studies, the hybridization of the daaC probe to enteroaggregative E. coli has been taken to mean that the strains in question harbour a daa adhesin target as well as aggregative adherence genes [24]. Other workers have proposed that the hybridization signal arises from cross-hybridization at a single locus [21, 25]. Although the former situation is a possibility, particularly as aggregative fimbrial genes are plasmid-borne, in this study we implicate the aafC gene, predicted to encode the usher for AAF/II fimbriae, as a cross-hybridizing locus. This finding has implications for our current understanding of the epidemiology of diarrhoeagenic E. coli.

Understanding the aetiology of diarrhoea is important, particularly in high disease burden areas where risk factors need to be identified and vaccine development priorities established. Most of what is known about the relative importance of different diarrhoeagenic E. coli categories comes from small, snapshot studies or studies of traveller's diarrhoea, analogous to what Guerrant et al. [26] refer to as the 'eyes of the hippopotamus'. Many high-burden developing countries lack cell culture facilities for the Gold Standard HEp-2 assay needed to delineate some pathotypes of diarrhoea-causing E. coli from commensals. Non-radioactive DNA probes and, more recently, PCR have been advocated as methodology that might be used to detect enterovirulent E. coli in developing countries [27, 28]. The vast majority of earlier studies that have not used HEp-2 adherence assays have defined DAEC as E. coli that hybridize to the daaC probe.

Of 30 Medline-indexed controlled studies that sought DAEC, we were able to identify only nine that have heretofore demonstrated an association of DAEC with diarrhoea. Girón et al. [29] used daaC probe hybridization and HEp-2 adherence and found that DAEC were associated with disease in Mayan children in Mexico. However that study had a very short duration (3 weeks) and focused on a small remote population (63 cases, 1300 total population), and therefore there are limits to the extent to which the data should be extrapolated. Cegielski et al. [30] found probe-positive, but not diffuse-adherent DAEC associated with chronic diarrhoea in HIV-positive and HIV-negative patients in another small study in Tanzania. A recent Brazilian study made a similar finding: probe-positive DAEC were associated with paediatric diarrhoeal disease, particularly in older children [13]. A Bangladeshi study reported that DAEC identified by adherence assay were associated with persistent but not acute diarrhoea (p < 0.05)[31]. A number of other developing country studies published since that time, employing probe and adherence, adherence alone, or PCR-based detection have failed to find an association between detection of DAEC and disease [8, 10, 12], 32-35.

In 1993, Levine et al. observed that a Chilean study, entirely reliant on the daaC probe, represented the "strongest epidemiologic evidence so far to indicate that DAEC may indeed be pathogenic"[36]. This large, controlled cohort study identified DAEC, based on daaC hybridization alone, in 16.6% of cases and 11.9% of controls (p = 0.0024). In that study, children aged 4-5 years had a relative risk of 2.1 for DAEC (overall relative risk was 1.4). Subsequent reports from studies using only the probe support the findings of that study [13, 37, 38]. For example, a 2005 US study found that DAEC identified by SLM862 probe were associated with diarrhoea (p < 0.05) but DAEC identified by HEp-2 adherence were not [38]. Overall, in the light of the limitations of the daaC probe we here report, only three published studies that we reviewed unequivocally suggest a role for DAEC in acute diarrhoeal disease. Jallat et al. [11] used HEp-2 adherence to identify DAEC in a French study and found these organisms to be significantly associated with disease in patients of all ages (p < 0.0001). In that study, only 33 of the 100 DAEC isolates identified hybridized with the daaC probe and interestingly, five of these strains also hybridized with the CVD432 probe for enteroaggregative E. coli and showed an aggregative-diffuse pattern of adherence. Ten daaC positive strains were non-adherent. A second study, by Gunzburg et al. [39], found that DAEC were not associated with diarrhoea overall, and were more common in healthy patients under 18 months of age. However, Gunzburg et al. did find that in children aged 18 months to five years, DAEC were recovered from 11 cases and 4 controls (p ≤ 0.05). Similarly, Scaletsky et al. [9] found that DAEC was not associated with disease overall in a study performed in North-East Brazil but was significantly associated with diarrhoea among children in the 13-24 month old age group. These studies provide evidence to advocate that future investigations aim to determine whether there is a role for DAEC in diarrhoea in some populations, particularly in children over one year of age, and that they do so using techniques other than the daaC probe.

There are important implications for the role of pathogens other than DAEC in disease that may come to light if the daaC probe is replaced with more specific testing methods. Recent studies have demonstrated that AAF/II-positive EAEC are more significantly associated with diarrhoea than the EAEC category as a whole 40-43. Thus any test for DAEC that detects potentially AAF/II EAEC will skew the results towards a stronger association of the DAEC category with disease, particularly if the EAEC strains in question are negative for the commonly used but inadequately sensitive EAEC CVD432 probe. Additionally, evidence supporting a role in diarrhoea for less-studied E. coli categories such as cell-detaching E. coli or cytolethal distending toxin-producing E. coli, appears to be equivalent to supporting data for DAEC, if daaC-derived data is discounted. Future investigators may want to consider these under-studied categories as worthy of further study.

There is some suggestion that DAEC could be an important pathogen in weaned children but in order to correctly gauge the relative contributions of DAEC and other pathogens such as AAF/II-producing EAEC to diarrhoea epidemiology, it is imperative that the SLM862 daaC probe, which detects AAF/II-positive EAEC as well as DAEC, be discarded in favour of more specific methodology. Given that AAF/II-positive EAEC represent an important subset of that category and therefore there is considerable advantage of testing for both simultaneously, particularly as current PCR-based protocols typically do not screen for DAEC and use CVD432 as the EAEC target [28]. If the daaC probe is employed, it should be used in conjunction with a probe for aafA. Alternatively, the PCR-RFLP test we describe here, which delineates the adjacent daaD and aafB genes may be substituted for hybridization with the SLM862 cloned daaC probe.