The survival characteristics of Escherichia coli O157:H7 in private drinking water wells were investigated to assess the potential for human exposure. A non-toxigenic, chromosomally lux-marked strain of E. coli O157:H7 was inoculated into well water from four different sites in the North East of Scotland. These waters differed significantly in their heavy metal contents as well as nutrient and bacterial grazer concentrations. Grazing and other biological factors were studied using filtered (3 and 0.2 µm) and autoclaved water. The survival of E. coli O157:H7 was primarily decreased by elevated copper concentrations. This hypothesis was supported by acute toxicity assay data. In addition, significant protozoan predation effects were observed in untreated water when compared with survival rates in filtered water. The combination of these two factors in particular determines the survival time of the pathogen in a private water well. It therefore appears that wells with higher water quality as assessed using the European Union Drinking Water Directive standards will also allow survival of E. coli O157:H7 for much longer periods.
Escherichia coli O157:H7 is a human pathogen with a very low infectious dose which can cause severe and sometimes fatal illnesses such as haemorrhagic colitis and haemolytic uraemic syndrome [1,2]. In recent years, reports of E. coli O157:H7 infections in the UK have increased sharply and the importance of infection routes other than direct faecal–oral have been recognised [2,3]. Viable E. coli O157:H7 have the capacity to leach through soil and can thus contaminate groundwater and private drinking water supplies [4–6]. Indeed, over the past 15 years, a number of outbreaks associated with private and other unchlorinated drinking water have been reported (Table 1). In Scotland, about 83 100 people are served by a private water supply for domestic use and a further 68 700 supplies are used for commercial activities. The highest reliance on private water supplies is in Aberdeenshire in the North East of Scotland (ca. 13% of the population), where a large number of water-borne diseases have been reported . Surveillance of private water supplies in England and Wales between July and December 1998 showed that 37% contained non-specified E. coli. In fact, several reports on the water quality of UK private drinking water have reported failure of a majority of supplies for one or more standards, such as those set for microbiological, nutrient and heavy metal concentrations [8–11]. Due to this wide divergence in water quality in private drinking water supplies, it is impossible to ascertain the perceived turnover times of pathogens such as E. coli O157:H7 in these systems. We therefore sought to establish survival rates of E. coli O157:H7 in four different well water types and simultaneously studied the influence of protozoan grazing and toxicity effects of elevated heavy metal concentrations.
|1991||Grampian, Scotland, UK|||
|1997||Grampian, Scotland, UK|||
|1997||Fuerteventura, Canary Islands|||
|1999||Highland, Scotland, UK|||
|2002||Highland, Scotland, UK|||
Materials and methods
Bacterial strain and culture conditions
Escherichia coli O157:H7 strain 3704 Tn5 lux CDABE is a lux-marked non-toxigenic isolate from a farm drain, which lacks the genes for verocytotoxin production. This strain was a kind gift from Prof. L.A. Glover (Institute of Medical Sciences, University of Aberdeen). The strain was grown in either liquid culture using tryptone soya broth (TSB) (Oxoid, UK) at 37°C, 200 rpm, to maximum luminescence per cell (late exponential phase) or on sorbitol MacConkey agar (SMAC) (Oxoid) at 37°C with 14 h incubation.
Survival and protozoan predation in private well water
Well water was collected from four private farm sites in Aberdeenshire, Scotland in sterilised, acid-washed glass bottles. Nutrient levels (NO3−, PO43−) were established using standard US EPA methods . Total concentrations of copper, lead and zinc were determined by flame atomic absorption analysis (AAnalyst100, Perkin-Elmer) of samples acidified with 1 M HCl. Background levels of coliforms and E. coli O157:H7 were established using membrane filtration of 100 ml water and subsequent incubation of the filter on Chromocult® Agar (Merck) or SMAC agar, respectively. Well water sample microcosms (250 ml each) were prepared from untreated, autoclaved and filtered (3 µm and 0.2 µm) source water, which received additions of E. coli O157:H7 strain 3704 Tn5 lux CDABE to a density of 1.5×109 colony forming units (CFUs) ml−1. The strain had been grown to maximum luminescence per cell, and was washed twice with 1 volume of 1/4-strength Ringer's solution. Abundance of protozoa in the untreated water was determined at the start of the experiments as described by Cho and Kim  by filtration of 100 ml through an 0.45 µm filter followed by staining with DAPI and counting by epifluorescence microscopy. The microcosms were incubated standing in the dark at 15°C, with samples of 10 ml being removed at set intervals. From the samples, survival of E. coli O157:H7 strain 3704 Tn5 lux CDABE was determined by culturable cell counts on SMAC agar.
Potential luminescence, a measure of total viable cells, was measured using a modification of the method of Duncan et al.. An aliquot (1 ml) of inoculated well water was mixed with 9 ml of pre-warmed (37°C) TSB and incubated for 30 min at 37°C and 200 rpm. Luminescence of a 1-ml aliquot of this mixture was subsequently measured on a Bio-Orbit 1253 luminometer (Labtech International, UK).
Sensitivity of E. coli O157:H7 Tn5 lux CDABE to copper
The sensitivity of the lux-marked E. coli O157:H7 to copper (CuSO4) was assessed using a bioluminescence-based bioassay. E. coli O157:H7 strain 3704 Tn5 lux CDABE was grown in batch culture in TSB and harvested at maximum luminescence per cell (1.1×109 CFU ml−1). Aliquots of the culture were transferred to 50-ml centrifuge tubes and centrifuged for 5 min at 5500×g. The supernatant was discarded and the pellet resuspended in 0.1 M KCl. This step was repeated and the cell suspensions pooled. After determination of culturable cells by plate counts, the bioassays were carried out as described previously . Cell suspension (100 µl) was added to 900 µl test solution in a luminometer cuvette and mixed by repeated pipetting. The bioluminescence of the cell and test solution mixture was measured using a Bio-Orbit 1253 luminometer (Labtech International) after a 15 min exposure time. All bioassays were carried out in triplicate.
Analysis of variance was performed on Log10 converted cell number data using Minitab® for Windows, version 13.1. Statistically significant differences are reported at the P≤0.05 level. Dose–response curves were fitted to toxicity data using the sigmoidal, 3-parameter model in SigmaPlot 5 for Windows. All experiments were replicated three times and averages of the data are reported.
Influence of grazing and bacterial competition
In three of the four well water types, culturable E. coli O157:H7 declined most rapidly in the untreated microcosms, closely followed by those filtered through 3 µm (Fig. 1). In well water 4, however, untreated well water samples had significantly higher numbers of E. coli O157:H7 after three days than the other treatments (Fig. 1d). There was a significant difference between the cell numbers in untreated and 3-µm filtered water in well water 1 and well water 2 after day 5. There was no significant difference in cell numbers at any time point between these two types of microcosms in well water 3. Survival in well water 1 and 2 when filtered through a bacterial membrane filter (0.2 µm) was also significantly higher after day 15 than in 3 µm-filtered or untreated water. Again, there was no statistical difference between cell numbers at any time point in these treatments in well waters 3 and 4. Finally, survival in well waters 1 and 2 was significantly greater in sterilised water than in all other treatments. Initial numbers of protozoa were higher in untreated well water 2 than in well water 1, but no protozoa were found in well waters 3 or 4 (Table 2).
|Well no.||Protozoa (indv l−1)||Coliforms (CFU l−1)||E. coli O157||pH||Cu (mg l−1)||Pb (mg l−1)||Zb (mg l−1)||NO2-N (mg l−1)||NO3-N (mg l−1)||PO4-P (mg l−1)|
|Well no.||Protozoa (indv l−1)||Coliforms (CFU l−1)||E. coli O157||pH||Cu (mg l−1)||Pb (mg l−1)||Zb (mg l−1)||NO2-N (mg l−1)||NO3-N (mg l−1)||PO4-P (mg l−1)|
indv = individuals; nd = not detected; bd = below detection limit.
Quantification of viable cell numbers by potential luminescence
Potential luminescence reflected cell numbers until the detection limit of the assay (Log10 value of −2.1) was reached (Fig. 2). There was no significant difference in the potential luminescence values in the different treatments of well waters 3. Potential luminescence values in the untreated water in well water 4 were again significantly higher than in the other treatments, where no significant differences were found. Potential luminescence values in the different treatments in well waters 1 and 2 were all significantly different from each other from day 5 onwards. In addition, there was a linear relationship between culturable cell numbers and potential luminescence in samples above the detection limit of the potential luminescence assay (Fig. 3). The obtained values for well water 2 showed a similar trend. Potential luminescence per cell stayed constant in all samples returning values above the detection limit, at 7.9×10−7 pRLU cell−1 (±3.2×10−7).
Influence of copper on survival of E. coli O157:H7
High decay rates and rapid loss of potential luminescence were seen in all treatments for well waters 3 and 4 (Figs. 1 and 2), which both contained high levels of copper (Table 2). In fact, the copper concentration in well water 3 greatly exceeded the European Union Drinking Water Directive standard of 2 mg l−1. To establish whether the copper concentrations were the cause of this severe drop in cell numbers, dose–response curves of the well waters and toxicity assays using CuSO4 were performed with E. coli O157:H7 strain 3704 Tn5 lux CDABE. The EC50 value for copper, as determined using CuSO4 in a standard bioassay, was low at 0.78 mg l−1. Dilution of the untreated well waters resulted in dose–response curves corresponding to the curve generated with CuSO4 (data not shown).
The results of this study showed relatively long survival rates in untreated well waters 1 and 2, which would be classified as being of higher water quality according to the EU Drinking Water Directive standards with regards to their nutrient and heavy metal concentrations. Comparative studies by Porter et al. and Rice et al. reported similar decline curves of different strains of E. coli O157:H7 in well water held at 20°C. Wang and Doyle  determined survival in different waters at different temperatures, with similar results to Rice et al., and in addition showed the influence of a biogenic factor in the removal of E. coli O157:H7 by using filtered and autoclaved samples. Protozoan grazing has been proposed to be the main biotic factor determining the survival of introduced microorganisms in aquatic systems . The differences in survival rates between untreated well water 1 and 2 are most thus likely due to the initial load of protozoa. The difference in survival rates between 3 µm and 0.2 µm filtered waters compared to the autoclaved controls in these source waters show the influence of an additional biological factor in the removal of E. coli O157:H7. Although most private water supplies are used daily, and the length of time when water within the well would be stagnant would not be the same as used in this study, the results of this experiment nevertheless demonstrate the ability of protozoa to severely reduce the numbers of potential bacterial pathogens. Although previously reported for non-pathogenic E. coli, this is the first study to show a protozoan pathway of removal for E. coli O157:H7. Protozoa, however, can also harbour pathogens, including E. coli O157, or be pathogenic themselves [22,23] and thus form a potential health risk themselves.
Due to the complete absence of protozoa in well waters 3 and 4, the decline of the E. coli O157:H7 populations is unlikely to have been caused by predation, as there was no significant difference in survival characteristics between the treatments. Even the autoclaved control showed similar responses; hence, the most likely cause would be the presence of a bactericide such as heavy metals or other heat-stable compounds. Levels of heavy metals can be high in private drinking water due to leaching of copper and lead from corroding pipes . The extensive use of sewage sludge as soil amendments often leads to additional inputs of heavy metals . Toxic effects of heavy metals such as copper and zinc on bacteria have been reported extensively [25–27]. Although most of the metals in the well waters used will have been in a non-bioavailable form, even at these relatively high pH ranges there will be a small percentage of free ion in the water phase [25,26]. Some private wells in Scotland have been lined with copper and in almost all cases, the pipes leading to the consumer's tap is made from copper or lead . Previous studies of private well waters have focused on the health effects of elevated copper concentrations. These include mainly short-term gastrointestinal symptoms such as vomiting and diarrhoea, but some evidence exists for long-term liver damage in infants (reviewed in ). Thus there may be a trade-off between beneficial concentrations of copper for reduction of E. coli O157:H7 populations and public health.
The decay of culturable E. coli O157:H7 and other pathogens in environmental samples has in some cases, at least partially, been attributed to the entry into the viable but non-culturable (VBNC) state [13,17,19,30]. In most of these cases, the VBNC state was confirmed by dyes confirming the integrity of the cellular membrane or by viable total microscopic counts utilising cell division inhibitors. Although there is no single accepted definition of VBNC [20,31], in general, evidence of a metabolic process in conjunction with non-culturability of the strain on their routine media is required. Bacterial luminescence has been previously used by Duncan et al. and Meikle et al. to demonstrate the VBNC state in a non-pathogenic strain of E. coli, circumventing the problem with actual luminescence measurements of starved cells where cellular energy reserves may be too low for production of adequate light emission. In the present study, cell numbers over the length of the experiment correlated well with potential luminescence values (Fig. 3), which suggests that the population decline was due to cell death rather than entry into the VBNC state. However, effects of starvation in water on E. coli O157:H7 can include loss of O antigenicity  or an increased resistance to chlorine treatment [34,35], all of which may affect the chances of detection of the pathogen in routine testing. These may be additional health risks when water from relatively clean wells is used infrequently.
The most important factors for long-term survival of E. coli O157:H7 in private drinking water thus appear to be absence of protozoan grazer populations and low concentrations of heavy metals. Paradoxically, water from a private source falling below EU Drinking Water Directive standards may thus have beneficial effects in terms of likelihood of infection with E. coli O157:H7. However, this benefit and the health risks in elevated concentrations of nutrients, heavy metals, and protozoan populations should be weighed cautiously.
The authors gratefully acknowledge funding by a Scottish Executive Flexible fund grant (UAB/007/99) and would like to thank Hedda Weitz for critical comments on this manuscript.