- Most prominent representative: E. coli O157:H7
- Enter water through fecal contamination
- Cattle can serve as a reservoir although the bacteria rarely cause disease in cattle
- Many of the virulence factors are phage- or plasmid-borne and therefore subject to horizontal gene transfer
- Show higher prevalence in warmer months
- Can survive in water for months dependent on the water type
- Susceptible to commonly used disinfectants, but protection by rumen or fecal material can occur
- A wide range of detection methods is available for E. coli O157:H7 due to its role as a ‘model pathogen’
Water contamination with Escherichia coli
is of fecal origin as the bacteria are found exclusively in the digestive tract of warm-blooded animals, including humans ( Health Canada 2006
). While most of the E. coli
are non-pathogenic, some can cause severe diarrhoeal disease with sometimes fatal outcome. Based on serological and virulence characteristics, six broadly divided groups of pathogenic E. coli
are distinguished: enterohaemorrhagic (EHEC), enterotoxigenic (ETEC), enteroinvasive (EIEC), enteropathogenic (EPEC), enteroaggregative (EAEC), and diffuse adherent ( Health Canada 2006
). Each type causes diarrhoeal disease through different mechanisms and different clinical symptoms. The most prominent and well-studied representative of enterohaemorrhagic E.coli
is strain O157:H7. The ‘O157’ serogroup shares the same somatic (cell surface) O antigen in common. The flagellar H antigen is used to define the specific serotype, viz. H7. The following text focuses on E. coli
O157:H7, which has received the most attention and is sometimes used synonymously with EHEC. It has to be noted, however, that several other EHEC serotypes can cause disease in humans. A study reviewing the occurrence of E. coli
O157:H7 in 49 states in the U.S. between 1982 and 2002 as documented by the U.S. Centers for Disease Control and Prevention (CDC), stated that the pathogen was responsible for 73,000 illnesses in the U.S. annually ( Rangel et al. 2005
Virulence factors. Pathogenicity mainly is caused by two genes (stx1 and stx2) encoding shiga toxins. The group of E. coli that produces at least one of these toxins is classified as shiga-toxin-producing E. coli (STEC) or sometimes referred to as verocytoxin producing E. coli (VTEC). Another important virulence factor is intimin (encoded by the eae gene). Intimin is associated with adhesion to epithelial cells and their lesion and localized destruction of border microvilli ( Ramachandran et al. 2003). The Tir protein (encoded by the tir gene) serves as a receptor for intimin and is another important virulence marker for enterohemorrhagic E. coli. Many of the virulence factors are linked to either phages or plasmids and are therefore likely to be subject to horizontal gene transfer and may contribute to the emergence of new pathogenic strains ( Muniesa et al. 2006). This instability has also been noted in the loss of toxin genes upon repeated subcultivation of EHEC organisms ( Karch et al. 1992).
Sources of infection. Although most E. coli O157:H7 infections and outbreaks seem to be caused by undercooked food and untreated and unpasteurized milk and juices ( Health Canada 2006; Gugnani HC 1999), cases of disease associated with contaminated under-treated water have been reported. Consumption of contaminated public drinking water is one important source of infection ( Dev et al. 1991; Swerdlow et al. 1992; Geldreich et al. 1992; Chalmers et al. 2000). Some of the largest reported EHEC-caused waterborne outbreaks in North America occurred in 1999 at a county fair in upstate New York ( Bopp et al. 2003) and a year later in Walkerton, Canada ( Bruce-Grey-Owen Sound 2000; Hrudey et al. 2003). The latter was estimated to have caused serious illness among 2,300 people with six deaths. Around 1,300 cases were considered to be directly caused by contaminated municipal water ( Bruce-Grey-Owen Sound 2000). The incident was linked to a contaminated well providing groundwater under the influence of surface water. The incidence coincided with heavy rainfalls and flooding and a water treatment system which might have been overwhelmed by increased turbidity ( Bruce-Grey-Owen Sound 2000; Auld et al. 2004). Infected cattle on two farms were suspected to cause the water contamination. Ruminants, especially cattle, are known to be an important reservoir of E. coli O157:H7 ( Gyles CL 2007; Johnson et al. 2003). Feed and waters were suspected to be the sources of infection of cattle ( Hancock et al. 2001). Despite their widespread occurrence among ruminants, E. coli O157 rarely causes disease. Areas with high cattle density or places where beef is processed are therefore potential sites where water contamination occurs. Interestingly, long-term exposure to E. coli O157:H7 is thought to result in partial immunity. This would explain a significantly higher attack rate among visitors than in town residents (50% vs. 23%) in a waterborne outbreak, which occurred in 1998 in the small town Alpine, Wyoming ( Olsen et al. 2002). The outbreak was suspected to be caused by consumption of unchlorinated municipal water that was likely to be contaminated with surface water containing deer and elk feces. The attack rate also decreased with increasing age. Municipal water in Alpine was suspected to be subjected to chronic contamination with E. coli O157:H7 before leading to long-term exposure of residents and the development of partial immunity ( Olsen et al. 2002). In a study in the Netherlands, E. coli O157:H7 was detected in drinking water samples from four private water supplies ( Schets et al. 2005).
Recreational water exposure has also been implicated as a source of E. coli O157:H7 waterborne infection. This would include swallowing water while swimming in contaminated lakes ( Ackman et al. 1997; Keene et al. 1994) or constructed swimming pools ( Friedman et al. 1999). In one reported waterborne outbreak in a fresh water lake the causative organism was isolated from water and duck feces ( Samadpour et al. 2002). Pulsed-field gel electrophoresis analysis of the isolates yielded the same restriction fragment patterns. A large outbreak occurred in a fresh water lake in Washington state (USA) and E. coli O157:H7 was isolated from sediment samples taken from a childrens’ swimming area ( Bruce et al. 2003). Pulsed-field gel electrophoresis (PFGE) analysis revealed that the environmental isolates matched those of the clinical isolates. The detection of STEC has been reported in surface waters in southern Michigan and northern Indiana. It was suggested that recreational water activities in these waters could result in human exposure to these organisms ( Duris et al. 2009). A watershed survey in the same geographical area (north-central Indiana) also reported on the occurrence of E. coli O157:H7 and the presence of antibiotic resistant strains ( Fincher et al. 2009).
Most studies tend to focus on prevalence and concentration in stool and fecal samples as bacterial concentrations in these samples are high. Quantitative information about E. coli
O157:H7 in natural waters and drinking water, on the other hand, is scarce due to the low concentrations of this pathogen and relatively high detection limits of existing diagnostic technology. A seasonal pattern has been observed with higher prevalence in water in warmer months ( Hancock et al. 1998
). Peaks have been reported during summer and late fall as the bacteria can reproduce at higher temperatures.
Selected examples showing occurrence in water are listed in the following.
A study in metropolitan Baltimore performed between 2002 and 2004 tested for the presence of tir- and stx-genes in E. coli isolated from stream water with weekly sampling. Coliforms from enriched water samples were captured using beads with anti-E. coli O157 antibodies. Captured cells were plated onto MacConkey agar plates. Randomly collected colonies that gave a strong positive signal using the BBL Enterotube II and that possessed a lacZ gene were test for the presence of virulence genes using PCR. The presence of the tir gene was found in 53% of 1,218 water samples with greater prevalence in urban, polluted streams. Positivity rates were generally higher in the summer months than in the winter months. Urban polluted streams also showed a higher presence of one the two stx genes (associated with EHEC). However, stx genes were surprisingly also found in the two most pristine streams in forested watersheds, which did not receive sewage or manure inflow. The authors concluded that pathogenic E. coli might be continually deposited in water streams as permanent members of the gastrointestinal microflora of humans and animals ( Higgins et al. 2005).
When examining the microbiological quality of drinking water from 144 private well in the Netherlands, E. coli O157 was isolated (using a specific enrichment method) in 2.7% of the samples. These samples met drinking water standards as determined by simultaneous sampling of faecal indicators ( Schets et al. 2005).
- A 2-year study performed in the Oldman River watershed in Canada with high cattle density and extensive irrigation revealed that 0.9% of surface water samples (n=1483) contained E. coli O157:H7 (6.2% of these samples tested positive for Salmonella spp.; n = 1429) ( Johnson et al. 2003).
E. coli O157:H7 was isolated from 1.7% (n=1608) of raw river water and irrigation water samples in the Oldsman River Basin in southern Alberta, Canada, over a two year period. In both years isolation prevalence peaked in July with 3.1% (n=159) and 2.4% (n=179) for the two consecutive years. The same summer peak applied to Salmonella isolations, but with significantly higher frequency. Salmonella spp. were isolated from 10.3 % of samples (n=1629) on average with peaks of 16.4% (n=159) and 30.2% (n=179) in July of the two years. The river basin is located in a semi-arid zone with cattle feedlot industry, crop cultivation and a large network of irrigation channels ( Gannon et al. 2004).
E. coli O157 concentrations were estimated for 33 surface water samples from the Baltimore watershed based on enrichment in broth and immunomagnetic electrochemiluminescence. The antibodies used for the study were specific for the O157 serogroup. The data suggested that the majority of samples contained E. coli O157, but in low concentrations of < 1 cell per 100 ml raw water (Shelton et al. 2004).
- Concentrations of E. coli O157:H7 in cattle feed water are highly variable, but might range between < 1 CFU/ml to 100 CFU/ml. Occasionally, the concentrations can be significantly higher (Hancock et al. 2001).
Chlorine levels typically maintained in water systems efficiently inactivate E. coli
O157:H7 ( Rice et al. 1999
). Mean levels of 1.1 mg/L free chlorine and 1.2 mg/L total chlorine were reported to result in an inactivation rate around -3.1 log10
units per min exposure for different isolates. The experiment was conducted at 5°C, the rate was similar to the inactivation of wild-type E. coli
isolates. Similar findings have also been reported ( Kaneko M 1998
). Another study examining chlorine inactivation of five human E. coli
0157:H7 isolates in water at 23°C, reported >7 log10
CFU/ml reductions of 5 strains by exposure to 0.25 mg/L free chlorine in 1 min, comparable to an E. coli
control strain ( Zhao et al. 2001
). Only one strain showed increased resistance with only a 4 log10
reduction at this chlorine concentration. Despite the existence of more tolerant strains, chlorination in general appears to be efficient to adequately control this pathogen. Pathogenic E. coli
are also susceptible to UV-disinfection although DNA repair can occur after exposure and a fluence of 300 J m-2
might be necessary to obtain safe water. A well-maintained distribution system should reduce the levels of E. coli
O157:H7 in drinking water to a concentration that is not associated with human illness ( Health Canada 2006
). For disinfection of wastewater in industrial settings thermal inactivation (at temperatures around 60°C) is an efficient killing procedure.
- EHEC were reported to have a growth range between 8 and 48°C. Temperatures exceeding the maximum growth limit can be expected to result in inactivation or in thermal death. Spinks et al. exposed E. coli, suspended in 200 ml water at a concentration of app. 108 CFU/ml, to either 55, 60, or 65°C and monitored survival at different time points by spread-plating. Thermal resistance was expressed in D-values (defined as the time required to achieve a 90% (1 log10) reduction of a bacterial population at a given temperature). The D-values were 62 sec at 60°C and 3 sec at 65°C. At 55°C, a biphasic reduction rate was observed with the initial rate being slower (D-value: 21 min). The reduction rate after about 18 min showed a D-value of 4 min ( Spinks et al. 2003). Similar rates can be expected for E. coli O157:H7. This was confirmed in a later study including E. coli O157:H7 ( Spinks et al. 2006). The authors concluded that at 65°C little thermal resistance was observed by any species studied and 1-log reductions of culturability were obtained within seconds. A temperature range between 55 and 65°C can be considered critical for thermal inactivation of typical enteric pathogens including E. coli O157:H7.
Lagunas-Solar et al. (2005) investigated the use of radiofrequency power (RF) to inactivate waterborne pathogens. RF is a physical method leading to uniform heating of a water volume. Water samples were seeded with appx. 8 x 106 CFU/ml E. coli O157:H7 and subjected to a frequency of 10 to 14 MHz. Bacterial growth was completely and rapidly >63°C was reached. RF thermal disinfection does not leave chemical residues and might be interesting for large-scale industrial applications ( Lagunas-Solar et al. 2005).
Sommer et al. studied the inactivation of seven pathogenic and one non-pathogenic E. coli strains by ultraviolet (UV) light (254 nm) and their capability of repair ( Sommer et al. 2000). At a fluence of 12 Jm-2 a 6-log reduction in culturability was obtained for the most susceptible strain (E. coli O157:H7), whereas 125 J m-2 were needed for the most resistant strain (E. coli O25:K98:NM) to achieve the same degree of inactivation. The strains were reported to have the ability for photorepair. Considering DNA repair, a UV fluence of 300 J m-2 was needed to achieve a 6-log reduction. The authors pointed out that fluences below that value might not be sufficient to kill pathogenic E. coli. Zimmer-Thomas et al. reported higher DNA repair activity for E. coli O157:H7 compared to non-pathogenic E. coli. Exposure to medium-pressure (MP) UV reduced the ability of E. coli O157:H7 to DNA repair ( Zimmer-Thomas et al. 2007). Higher DNA repair activity was found after exposure to LP UV.
The use of sodium caprylate has been investigated for killing E. coli O157:H7 in cattle drinking water. Well water samples (100 ml each) containing increasing 75, 100, or 120 mM caprylate were artificially contaminated with a four-strain mixture of the pathogen to a density of 106 CFU per ml. Samples were incubated at 8 or 21°C for up to 21 days and monitored for culturable E. coli O157:H7. Culturability was increasingly lost with higher concentrations of the disinfectant and with higher temperatures. At a concentration of 120 mM caprylate no growth was observed after 1 to 20 days. However, the presence of feces was reported to decrease the antibacterial effect. The addition of feed on the other hand enhanced the effect (Amalaradjou et al. 2006).
Zhao et al. reported that in cattle drinking water heavily contaminated with rumen contents, chlorine (5 mg/L) or ozone (22-24 mg/L at 5°C) had little effect on killing E. coli O157:H7 (<1 log10 reduction in cultural). Different combinations of a variety of the chemicals lactic acid, acidic calcium sulfate, butyric acid, sodium benzoate, caprylic acid, and chlorine dioxide, on the other hand, were significantly more effective with inactivation rates of up to 5-log units. The authors suggested periodic treatment of drinking water troughs with these chemical ‘cocktails’ followed by flushing with fresh water in order to avoid reduced water consumption by the cattle ( Zhao et al. 2006).
E. coli O157:H7 was reported to require a CT value, C = disinfectant concentration (mg/L), multiplied by T = time (min) of appx. 9.2 mg-min/L to achieve a 4 log10 level of inactivation with monochloramine at pH 8.0 and 21°C ( Chauret et al. 2008). Levels of inactivation using free chlorine in this study were similar to those reported previously ( Rice et al. 1999).
O157:H7 can survive a range of different water types. Survival up to 90 days in river water and more than 300 days in bottled water has been reported ( AWWA 2006
). E. coli
O157:H7 shows similar survival rates and disinfection susceptibility as non-pathogenic E. coli
. The latter is therefore considered a good indicator of the presence of E. coli
O157:H7 ( Health Canada 2006
). Persistence of the bacteria as carriers of virulence genes has to be distinguished from the persistence of the virulence genes themselves. As mentioned before, many virulence genes are linked to phages or are plasmid-borne and can therefore be transferred on non-pathogenic species resulting in emergence of new pathogenic strains ( Muniesa et al. 2006
). Infectious stx
-phages showed high persistence in water systems ( Muniesa et al. 2006
). Persistence of infectious genes is therefore an equally relevant topic as the persistence of gene carriers.
A selection of relevant publication is summarized in the following:
Avery et al. studied the survival of E. coli O157:H7 in different surface waters. Vials with 4.5 ml of lake, puddle, river, or animal-drinking trough water were spiked with the pathogen to final concentrations of 8 x 104 CFU/ml and were kept at 10°C. Although the numbers of culturable pathogens declined over time, considerable persistence was observed with differences between the water types. The T99 values (time period in which colony numbers showed a 2-log drop compared to the original count) were: 17.8 days for the fecal contaminated puddle water, 12.9 days for lake water, 6.3 days for trough water, and 6.0 days for river water. Overall survival was significantly enhanced by water aeration and autoclaving of the water prior inoculation. No correlation was found between physico-chemical water parameters and survival ( Avery et al. 2008).
Survival of E. coli O157:H7 seeded at a concentration of 103 CFU/ml in filtered and autoclaved water from different sources (municipal water, reservoir water, water from two recreational lakes) was reported to be greatly temperature-dependent as judged by cultivation. Comparing 8, 15, and 25°C, survival was greatest at 8°C and least at 25°C for all water sources. The population decreased by 1-2 log10 units by 91 days at 8°C, whereas at 25°C no culturable cells were detectable within 49-84 days in three of the four water sources. While culturability decreased and finally disappeared, direct viable counts (determined by acridine orange staining) remained the same indicating the potential to enter an injured or viable but non-culturable (VBNC) state. In terms of the water source, survival was greatest in municipal water and least in lake water ( Wang and Doyle 1998).
A study by Kerr et al. revealed that E. coli O157:H7 can survive for extended periods in commercially bottled mineral water. When seeding water with 103 E. coli O157:H7 cells/ml and storing samples at 15°C, it took app. 70 days for non-sterile mineral water to be nondetectable by cultivation, 49 days in sterile mineral water and 21 days in sterile distilled deionized water. At this concentration and in this study, the autochthonous microbial flora thus seemed beneficial for survival of the pathogen (maybe due to the higher availability of organic nutrients caused by breakdown of cells). However, there was no significant difference between the different water types, when a higher inoculum concentration (106 cells/ml) was used with a >3 log10 reduction within 70 days. The authors pointed out that survival times in mineral water can vary greatly from one investigation to the other. Also the reports about the effects of the autochthonous flora have been conflicting ( Kerr et al. 1999). In another survival study in bottled water using scanning electron microscopy E. coli O157:H7 organisms were observed to attach to container walls ( Warburton et al. 1998).
Rice, et al. studied survival of E. coli O157:H7 in well water in microcosms using water implicated in a waterborne outbreak. Survival of the outbreak strain was comparable to a wild-type E. coli strain under the same conditions ( Rice et al. 1992). Survival in well water from four different sites in Scotland with different water quality varied greatly in a study be Artz and Killham (2002). Water samples were seeded with a lux-marked E. coli O157:H7 strain and stored at 15°C. Surviving cells could be detected by cultivation for approx. 12 to >65 days (end of monitoring) depending on the water type. Cell numbers in this study correlated well with the luminescence values speaking against the adoption of a VBNC state. Removal of autochtonous flora by filtration through 3 μm and 0.2 μm filters or autoclaving was beneficial for survival, whereas higher grazer and copper concentrations negatively affected survival. As the latter two parameters seemed to be the most important factors influencing survival in this study, the absence of grazers and low concentrations of copper as found in wells with good water quality might paradoxically support longer pathogen survival (Artz and Killham 2002).
Rice and Johnson studied the survival of E. coli O157:H7 in dairy cattle drinking water from two farms at temperatures of 5 and 15°C. The water was inoculated with manure from the two farms (1 g manure added per liter water) resulting in pathogen concentrations of 5000 cells/ml (farm 1) and 170 cells/ml (farm 2), respectively. Although numbers decreased, pathogens were still detectable in water from farm 1 until the end of the study period of 16 days at both temperatures. Higher counts were found at the lower temperature. In water from farm 2, pathogens were still detected through day 8 at 5°C and through day 4 at 15°C ( Rice and Johnson 2000).
Ravva et al. studied the survival of two laboratory strains of E. coli O157:H7 and a wastewater isolate of E. coli H7 in wastewater from dairy livestock. Microcosms with freshly collected wastewater were seeded with the pathogens to final concentrations of app. 104 to 107 CFU per ml and were incubated at app. 23°C. No naturally occurring E. coli O157:H7 were detected. With strain variations, pathogens persisted between 3 and >38 days. However, strains were unable to proliferate and to establish themselves in the microcosms even after four sequential inoculations. A decimal reduction time between 0.5 and 9.4 was measured by cultivation. Aeration did not have a significant effect on survival. The native aerobic bacteria survived longer with a decimal reduction time of 21.3 days ( Ravva et al. 2006). E. coli O157:H7 has also been reported to survive in unchlorinated reconditioned pork-processing wastewater water ( Rajkowski and Rice 1999).
Water has been implicated as a means by which food crops can become contaminated with E. coli O157:H7. In particular irrigation water has been cited as a contamination source of pre-harvest foods, ex. leaf lettuce ( Watchel et al. 2002a) and cabbage (Watchel, et al. 2002b). Using real time PCR measurements, E. coli O157:H7 levels were reduced by approximately two orders of magnitude from influent to effluent in a dairy wastewater wetland ( Ibekwe et al. 2004). In a study designed to track the incidence of E. coli O157:H7 in water and other environmental samples taken from a produce production area in California it was shown that within a given watershed the contamination of the environment is a dynamic process which may involve multiple sources and methods of transport ( Cooley et al. 2007).
Microcosms mimicking cattle water troughs were contaminated with feces from an E. coli O157-challenged calf. Aliquots of one kg feces were added to 80 liter continuous flow chambers. Water with naturally occurring microorganisms was pumped onto the surface of each microcosm. Six microcosms received chlorinated water, whereas residual chlorine was from the input water of the remaining six microcosms. Troughs were loosely covered with lids in an outdoor location with temperatures fluctuated with seasonal variations (April to December, Pullman, Washington). Concentrations of culturable E. coli O157:H7 decreased over the time period of 245 days from 9 x 109 to 1.62 x 101 cells per gram sediment in microcosms fed with unchlorinated water. Chlorination of water (0.15 mg/L for 90 days and 5-7 mg/L for the following days) resulted in slightly reduced final cell numbers (1.16 x 101 cells per gram sediment) compared to the unchlorinated samples, but did not eliminate the pathogens. E. coli O157:H7 were still isolated after 245 days in all water sediments. Cells remaining in the microcosms longer than 6 months, maintained their ability to colonize calves challenged with contaminated water at the end of the experiment. Feces of these calves contained the pathogens for 87 days after challenge. Water trough sediments were thus suspected to serve as a long-term reservoir in farms ( LeJeune et al. 2001). Persistence of specific clonal types of E. coli O157:H7 in feedlot cattle was reported in a later publication despite massive cattle population turnover ( LeJeune et al. 2004). Another study of farm water stored in microcosms outside and in a farmyard shed indicated that E. coli O157:H7 could survive at temperatures < 15°C for up to 24 days (McGee et al. 2002).
- Miyagi et al. could show that E. coli O157:H7 can tolerate high concentrations of salt. Growth was obtained in medium containing 5-6% NaCl. When seeded into seawater (3% NaCl (w/v), 27°C) to a concentration of 106 CFU/ml, numbers of culturable bacteria were still between 10- 103 CFU/ml after 15 days. In sterilized sea water in the absence of biological competitors, numbers even increased from 102 to 106 CFU/ml within 7-10 days of incubation. Results suggested that E. coli 0157:H7 can survive in marine water for extended times ( Miyagi et al. 2001). The resilience in marine environments was confirmed in a more recent study by Williams et al. ( Williams et al. 2007). Bioluminescence activity of lux-marked E. coli O157:H7 was monitored in artificially contaminated water over 5 days with various ratios of runoff water to seawater. Although pathogen activity declined with increasing seawater concentrations, residual activity was maintained over the study period. Nutrient addition in form of glucose and glutamate resulted in increased bioluminescence.
The infectious dose is considered to be less than 100 E. coli
O157:H7 cells ( Mead and Griffin 1998
; Hawker et al. 2001
). Although the bacteria are highly pathogenic for humans, they do not seem to affect cattle.
O157:H7 has several unique phenotypic characteristics which have been utilized in traditional cultural procedures. Unlike most wild-type E. coli
, the majority of strains of this serotype do not ferment sorbitol and are negative for beta glucuronidase expression, an enzymatic characteristic widely used to identify in E. coli
from environmental samples. E. coli
O157:H7 also, unlike many wild-type strains, does not grow well at elevated temperatures, viz. 44.5-45°C, ( Raghubeer and Matches 1990
; Palumbo et al. 1995
) the temperature that has often been used in water analysis for the detection of thermotolerant (fecal) coliforms ( Rice et al. 1996
). Various molecular procedures targeting specific gene sequences have also been employed for detection (for more thorough discussion, see Non-cultivation-based detection methods, below). Often these procedures will be used to look directly for the presence of a given genetic element in a water sample, however having viable isolates can greatly improve the ability to perform more thorough characterization of the organism and provides important data for use in molecular epidemiological studies. When analyzing directly for gene products specificity becomes an important issue. This point was noted in a study which found that the occurrence of the enterohemolysin gene (ehyl
), a gene that has been associated with most clinical isolates of E. coli
O157:H7, also occurred in other serotypes of E. coli
isolated from sewage samples collected over a geographical diverse area ( Boczek et al. 2006
). Similarly, Shelton et al. addressed the issue of specificity problems associated with the detection of E. coli
O157:H7. Regarding immunological assays targeting the somatic O antigen, it was pointed out that these antibodies are only specific for the O157 serogroup, but not for the O157:H7 serotype. However, the O157 serogroup also includes less virulent or nonpathogenic strains which would give false positive results in terms of health risk associated with detection. EHEC strains, on the other hand, which belong to other serogroups, would be missed in these tests. Targeting stx1
in genetic assays is also not specific for E. coli
O157:H7 as these genes are widely distributed among shiga-toxin
producing E. coli
strains as well as in other bacteria due to their phage-mediated dissemination. The detection of stx1
genes is therefore not definitive for EHEC. The authors also pointed out that the eae gene is not unique to EHEC either. It is also present in other E. coli
, which lack stx
genes ( Shelton et al. 2006
).A recent publication by the CDC, though designed for clinical laboratories, provides a valuable summary of procedures for identification of EHEC ( Gould et al. 2009
Cultural-based methods. A procedure for detection of EHEC organisms in water has been published ( Eaton et al. 2005). The procedure relies upon the use of enrichment broths, followed by selective plating and biochemical and serological confirmation. The use of commercial latex agglutination assays has been demonstrated to be effective alternatives to standard serological methods for the identification of O157 and H7 antigens ( Sowers et al. 1996). Various chromogenic agars have been developed to assist in selective isolation and presumptive identification. These agar media have been used primarily for clinical and food samples and have not been widely evaluated for environmental samples.
Non-cultivation-based detection methods. An increasing number of commercial tests are becoming available including BAX (Qualicon, Dupont), RapidChek (SDI), Reveal (Neogen Inc.), SafePath (SafePath Laboratories), SinglePath (Merck), Transia (Diffchamb), and VIP (BioControl). These tests have mainly been designed for the food sector. The mentioned tests are immunoassays with the exception of the BAX test, which is PCR-based. In addition, a broad spectrum of molecular methods have been published as E. coli O157:H7 serves as a model pathogen in many studies. Examples are given in the following. Methods of detection and isolation of STEC were discussed by Muniesa et al. ( Muniesa et al. 2006). Detection strategies of E. coli O157:H7 in foods have been reviewed by Deisingh and Thompson (Deisingh and Thompson 2004).
Combination of cultivation and molecular detection: Culture-qPCR. Sen et al. developed a highly sensitive and specific method to detect viable E. coli O157:H7 from source and finished drinking water combining culture enrichment and subsequent multiplex qPCR analysis ( Sen et al. 2011). Culture enrichment was successful in recovering chlorine and starvation stressed cells. The combination allowed for the detection of 3-4 viable E. coli O157:H7 cells/L within 24 hours.
Selected non-cultivation-based detection methods are summarized in the following:
CTC-fluorescent antibody detection: This method allows for the enumeration of viable bacteria. Bacteria are captured by membrane filtration. Addition of a CTC solution results in selective staining of actively respiring cells. After additional staining with a fluorescent antibody specific for E. coli O157:H7, filters are examined by epifluorescence microscopy. The time demand was reported to be 3-4 hours. The procedure was also successfully tested with Salmonella typhimurium and Klebsiella pneumoniae ( Pyle et al. 1995). A later development of this procedure employing solid-phase laser cytometry yielded similar results ( Pyle et al. 1999).
- qPCR: A number of different qPCR assays has been developed. Examples:
In a study of surface water ( Jenkins et al. 2009) combined a cultural most probable number methodology following large volume sample filtration to detect E. coli O157:H7 in surface waters of watersheds containing animal agriculture and wildlife. Confirmation was based upon serology, E. coli identification using the rapid glutamate decarboxylase procedure, ( Rice et al. 1993) and final verification by real time PCR ( Sharma VK 2002).
- Himathongkham et al. applied multiplex qPCR (targeting the stx1, stx2, and uidA genes) to detect E. coli O157:H7 from surface water samples. After a 5 hour enrichment cells were captured by recirculating immunomagnetic separation (using anti-O157 paramagnetic beads). The assay was reported to detect pathogen levels as low as 6 CFU/100 ml of surface water (Himathongkham,et al. 2007).
Belanger et al developed an assay utilizing molecular beacons and primers targeting the stx1 and stx2 genes. The analytical sensitivity was reported to be about 10 genome copies per reaction ( Belanger et al. 2002).
- Ibekwe and Grieve presented an assay based on detection of stx1 and eae genes. The detection limit was 3.5 x 103 CFU/ml in pure culture and 2.6 x 104 CFU/g soil (Ibekwe and Grieve 2003).
Sharma and Nystroem presented a multiplex assay targeting stx1, stx2, and the eae gene. With fecal samples sensitivity was reported to be in the range from 104 to 108 CFU/g feces (Sharma and Dean-Nystroem 2003). These primers were applied by Spano et al. to detect E. coli O157:H7 in dairy and cattle wastewater ( Spano et al. 2005). A detection limit of around 100 genome copies per reaction was reported.
- Some qPCR assays make use of the fact that E. coli O157:H7 and non-motile O157 strains have a highly single nucleotide polymorphism (SNP) at position 93 of the uidA gene (encoding β-glucuronidase). A multiplex assay targeting stx1, stx2 and the uidA variant was presented by Jinneman et al. with a potential to detect as few as 6 CFU per reaction (Jinnenman et al. 2003). The same authors also presented a minor groove binder-DNA probe with improved specificity of detecting the uidA +93 SNP ( Yoshitomi et al. 2003).
The presence of E. coli O157:H7 can be verified in a highly specific assay reported by Pan et al. Specificity is improved by including primers targeting the fliC (flagella H-antigen), rfb (surface O-antigen), and hlyA (enterohemolysin) genes into a multiplex PCR in addition to stx1, stx2 and eaeA. The assay was recommended to be used for concentrations exceeding 103 CFU/reaction ( Pan et al. 2002).
Barak et al. compared previously published primer sets and their suitability for real-time SYBR PCR with different pathogenic E. coli serotypes. A primer set targeting the eaeA gene ( Paton and Paton 1998) was determined to be most sensitive with a detection limit of 1 CFU E. coli O157:H7 per PCR reaction in sterile water and 10 CFU/reaction in diluted sprout irrigation water. The primer set detected 12 different pathogenic E. coli serotypes except O128:H7. However, melting curve analysis was recommended due to the formation of primer dimers ( Barak et al. 2005).
An ultrafiltration device for water collection coupled with real-time PCR has been reported to efficacious for detection of E. coli O157:H7 in surface water ( Mull and Hill 2009).
- Immunoassays: A large suite of different immunoassays has been developed for detection of E. coli O157:H7. The quality of detection heavily depends on the quality and design of the antibody. The validity of ELISA for detecting E. coli O157:H7 in water was addressed by Nyquist-Battie et al. ( Nyquist-Battie et al. 2007). After ColiTagTM enrichment, sandwich ELISA detected down to appx. 3 CFU/ml of seeded pathogens. Detection was also successful after sub-lethal chlorine exposure and after seven days of starvation as long as chlorine was absent or inactivated by addition of sodium thiosulfate. Long-term starvation was reported to result in altered antigenicity ( Hara-Kudo et al. 2001) and low concentrations of chlorine can interfere with antibody binding (Kolling and Matthews 2001). The interference of sample matrices was acknowledged. A bispecific hybrid monoclonal antibody which recognizes both E. coli O157 and horseradish peroxidase has also been developed ( Guttikonda et al. 2007). This bispecific antibody allows quick one step detection of target cells. When spiking aqueous solutions with E. coli O157:H7, a sandwich ELISA showed sensitivities of 100, 750, and 500 CFU/ml of tap water, lake water, and apple juice. Immunofilter ELISA and immunomagnetic ELISA formats showed sensitivities of 1 and 10 CFU/ml, respectively
ELISA-PCR: Comparison of different previously published primer pairs targeting different genes led to the conclusion that primers amplifying the SILO157 locus were most specific in this assay. An internal control was included to detect PCR inhibition. PCR products (both from the target gene and the IAC) were detected in a sandwich hybridization assay. The detection limit was around 100 CFU/ml (corresponding to less than 10 copies of STEC O157 per PCR reaction) ( Fach et al. 2003).
Reverse Transcriptase PCR (RT-PCR): A multiplex assay for simultaneous detection of E. coli O157:H7 (rfbE and fliC genes), Vibrio cholerae O1 (rfbE gene), and Salmonella typhi (tyv gene) in a single tube was presented by Morin et al. As few as 30 cells of the three pathogens could be detected in clinical isolates ( Morin et al. 2004). Yaron and Matthews used RT-PCR for several gene targets as potential indicators of viability and concluded that the rfbE gene was the most appropriate for detection of viable E. coli O157 ( Yaron and Matthews 2002) . IMS-DNA-Luminex: IMS separation was used for pathogen capture. DNA from IMS-captured cells is PCR-amplified using labeled primers. PCR products were then bound to color-coded Luminex-beads and passed through a flow cytometer which detects and quantifies PCR amplicons bound on the bead surface. The system had the ability of purifying and amplifying 10 E. coli O157:H7 cells from river water samples. Simultaneously, Salmonella and Shigella could be analyzed. The detection limit was reported as 100 cells for each organism ( Straub et al. 2005).
IMS and immunoblotting: Wastewater samples are incubated in peptone water for 90 min at 37°C to enhance recovery of injured cells. To prevent simultaneous growth of culturable cells antibiotics were added. E. coli O157 cells were subsequently captured using IMS using beads coated with O157-specific antibodies. Separated cells were subsequently plated on selective medium and enumerated by colony immunoblotting with anti-O157 antibodies. The method allowed detection and enumeration of E. coli O157 in wastewaters containing 0.2 to 1 log10 cells/ml. ( Garcia-Aljaro et al. 2005)
Immunomagnetic electrochemiluminescence (IM-ECL): E. coli 0157 cells are selectively bound by a capture antibody attached to immunomagnetic beads. A second antibody carrying a electrochemiluminescent label binds to the captured cells. Captured and tagged cells were pumped through a flow cell and magnetically attached to an electrode. Application of voltage resulted in a signal. The method was validated with seeded surface water samples. Detection limits were reported to be appx. 25 E. coli 0157:H7 cells per ml raw water or per 100 ml of 100-fold concentrated water ( Shelton and Karns 2001). In a later study, IM-ECL was combined with enrichment in minimal lactose broth for quantitative detection of as few as 103 to 105 E. coli O157:H7 per ml of water samples ( Shelton et al. 2003). This method was applied for analysis of surface water samples with prior enrichment in a later study ( Shelton et al. 2004).
Chandler et al. enriched E. coli O157:H7 from poultry carcass rinse using automated IMS (based on an electromagnetic flow cells and fluidics system). Subsequent PCR amplification of eaeA genes and microarray detection achieved a sensitivity of <103 CFU/mL ( Chandler et al. 2001).
- Liu et al. presented a method for specific detection of viable E. coli O157:H7. Target cells are captured on a low-protein-binding membrane followed by RNA extraction. RNA is reverse transcribed and PCR-amplified (using primers targeting rfbE and fliC genes). Detection of amplicons was achieved by electronic microarray detection. Detection limits were reported to be as low as 1 CFU of E. coli O157:H7 in diluted cultures, 3-4 CFU per liter tap water, 7 CFU/liter river water, and 50 VBNC cells in one liter of river water ( Liu et al. 2008).
Lee et al. introduced two low density DNA microarrays for detection of different waterborne pathogens ( Lee et al. 2006). The first microarray was based on the hybridization of genomic DNA, the second on the detection of PCR- amplified 23S rRNA genes. The first microarray included E. coli O157:H7 as a target organism. Based on the detection of E. coli, a sensitivity of 2 x 108 genome copies were reported. PCR amplification resulted in a sensitivity increase of about 6 log10 units.
IMS-PCR-Luminex: An automated ‘Biodetection Enabling Analyte Delivery System’ (BEADS) was presented by Straub et al. for simultaneous detection of E. coli O157:H7, Salmonella and Shigella. DNA from IMS-captured cells was PCR amplified. Labeled PCR products were bound to color-coded Luminex beads with distinct capture probes on their surface. Beads were sepearated through a flow cytometer, which both recognizes the bead color code and quantifies the signal originating from the labeled primers. The assay successfully purified and amplified 10 E. coli O157:H7 cells from river water samples. When multiplexing, the detection limit was reported to be 100 cells for each organism. The current platform can handle 100 ml volumes, the intention to scale up to 10 L water volumes was mentioned ( Straub et al. 2005).
PFGE and MLVA: The pulsed-field gel electrophoresis (PFGE) technique, widely used by public health laboratories, is a procedure for subtyping individual strains. While being primarily aimed at foodborne investigations, a report on the development of rapid standardized PFGE protocols would also be applicable for studying environmental isolates and strains obtained from waterborne outbreak investigations ( Ribot et al. 2006). Multiple-locus variable-number tandem repeat analysis (MLVA) is another molecular procedure used for strain subtyping. This procedure targets short DNA sequences which are repeated in the bacterial genome and has been proposed as another potential procedure for characterizing individual isolates of E. coli O157:H7 ( Hyytia-Trees et al. 2006; Keys et al. 2005).
- Biosensors: Zhu et al. described a biosensor utilizing a sandwich antibody technique. Whole cells are captured by antibodies attached to the inner surface of a glass capillary tube. Detection is achieved by using a labeled anti-E. coli O157 antibody. The signal obtained by illuminating the capillary using waveguide technology. Cells can be retrieved and can be subjected to further analysis to identify the pathogen. Quantitative detection of about 10 cells per capillary (75 μl volume) was reported. The concentration of larger volumes and its effect of co-concentrated substances on detection has not yet been solved ( Zhu et al. 2005). Another biosensor which is currently in development is based on IMS capturing of cells and addition of liposome nanovesicles which display specific E. coli O157:H7 antibodies and contain a fluorescent dye. The resulting immunomagnetic bead- E. coli O157:H7- immunoliposome complexes are again captured using IMS. A surfactant releases the dye from the liposomes and produces a concentration-dependent signal. The detection limit in different aqueous matrices was reported to be as low as 1 CFU/ml. The time demand was reported to be 15 min (without enrichment, if initial cell counts > 103 CFU/100 ml sample) or 4 hours (with enrichment). The assay was successfully tested with spiked groundwater samples ( DeCory et al. 2005). Due to the fact that E. coli O157:H7 serves as a model pathogen, there is a large number of emerging biosensor technologies being published. The two methods above serve as examples.
The following pictures are part of the Public Health Image Library (PHIL) from the Centers for Disease Control and Prevention (CDC). The photos are in the public domain and thus free of copyright restrictions. We would like to express our appreciation for providing these images.
Scanning electron micrograph of Escherichia coli, grown in culture and adhered to a cover slip.
Credit: Rocky Mountain Laboratories, NIAID, NIH
This colorized scanning electron micrograph (SEM) depicts a number of Escherichia coli bacteria of the strain O157:H7.
Photo ID: 10068
Content providers: Centers for Disease Control and Prevention, Janice Haney Carr
At an extremely high magnification of 44, 818X, this colorized scanning electron micrograph (SEM) revealed some of the morphologic details displayed by a single Gram-negative Escherichia coli bacterium. This bacterium was a member of the strain, 0:169 H41 ETEC (Enterotoxigenic E. coli).
Photo ID: 10577
Content provider(s): Centers for Disease Control and Prevention/ Janice Haney Carr
Credit: Janice Haney Carr
At a high magnification of 12,960X, this scanning electron micrograph (SEM) revealed some of the morphologic details displayed by a number of joined Gram-negative Escherichia coli bacteria. These bacteria were members of the strain, 0:169 H41 ETEC (Enterotoxigenic E. coli).
Photo ID: 10571
Content provider(s): Centers for Disease Control, Janice Haney Carr
Credit: Janice Haney Carr