The genetic relatedness of E . coli associated with post-collection drinking water contamination in rural households

Rural households are o�ten dependent on rivers, springs, boreholes or standpipes some distance �rom their homes �or their daily water requirements. Water �or drinking and domestic use is consequently stored in containers in�house which are prone to post�collection contamination. The objective o� the study was to determine the most likely origin or place o� introduction o� E. coli associated with post�collection contamination in rural households, by assessing the degree o� genetic relatedness o� E. coli present in the stored water and other environmental samples. E. coli isolates were obtained using either mFC agar with confirmation of indole production (44 isolates) or Colilert®-18 (52 isolates). Amplified fragment length polymorphism (AFLP) fingerprinting was applied to determine the genetic relatedness of E. coli isolated �rom in�house storage containers, drink� ing cups, hand-swab samples, cattle dung and from the source water (spring water). DNA fingerprints of E. coli produced a number of clusters (>85% similarity scores calculated with the cosine coefficient). Identical E. coli genetic patterns were observed at closely linked points within the domestic pathway o� water handling, such as between hand�swab and drinking� cup samples, between storage container and source isolates, and between drinking cups, source water and storage contain� ers. The results indicated that AFLP fingerprinting could be applied to determine the genetic relatedness of E. coli isolated �rom closely linked points within the domestic pathway o� water use within a household. However, the high genetic diversity observed �or E. coli bacteria isolated �rom the di��erent water and environmental samples tested in this study, hampered the identification of post collection points of contamination.


Introduction
Many people in developing countries are still dependent on untreated rivers, springs or boreholes �or drinking and domes� tic purposes.To improve access to better quality water in rural areas in South Africa much effort has gone into providing peo� ple with protected boreholes and standpipes at some distance from their homes.In these areas water for drinking and domes� tic purposes is mostly stored in�house in containers.However, many studies have shown that the microbial quality o� the water stored in�house, deteriorates considerably between point�o��col� lection and point-of-use (Wright et al., 2004;Jensen et al., 2002;Sobsey, 2002, Maraj et al., 2006).
Quanti�ying �aecal coli�orms by culturing is the most widely used method for measuring faecal pollution (Standard Methods, 1995).This method, however, does not identify the source of the contamination.During the last decade, many methods directed at the DNA of bacteria have been developed to 'fingerprint' genetic characteristics of micro-organisms (Chasseignaux et al., 2001;Wang et al., 1999).Through application of these methods, the genotypic characteristics o� the standard water quality indi� cator bacteria, E. coli have been applied to understand the origin and sources of faecal pollution (Guan et al., 2002;Stoeckel et al., 2004).Consequently several methods described as bacterial and microbial source�tracking methods have been developed to distinguish between various sources o� �aecal contamina� tion in water sources (Meays et al., 2004;Hagedorn et al., 1999;Harwood et al., 2000).Researchers employed ribotyping and repetitive e�tragenic palindromic�polymerase chain reaction (Myoda, 2003;Carson et al., 2003) polymerase chain reaction restriction fragment length polymorphism (PCR-RFLP; Aslam et al., 2003), pulsed field gel electrophoresis (PFGE; Liebana et al., 2003), fluorescent amplified fragment length polymorphism (FAFLP;Smith et al., 2000) and others.
Various comparative studies based on the discriminatory powers and resolution o� these di��erent molecular methods have been conducted (Scott et al., 2003;Radu et al., 2001;Vos et al., 1995;Clerc et al., 1998).Fluorescent AFLP has subsequently been shown to have high powers o� discrimination, universal applicability and good reproducibility for microbial fingerprint� ing.The method is based on selective amplification of a subset of DNA fragments from a digest of total genomic DNA (Vos et al., 1995) and has the ability to discriminate between strains of a bacterial species (Arnold et al., 1999;Speijer et al., 1999).
Accordingly, our study determined the genetic relatedness and potential source o� E. coli bacteria present in stored drinking water in selected rural households using AFLP analysis.Infor� mation about the most likely point o� post�collection contami� nation could provide valuable insight when hygiene education, provision o� drinking water and management o� the in�house quality is considered.

Methodology
Isolates were collected from households in two rural villages and were named by household and type (Table 1 and 2 108 Village 1 (Households 1, 3, and 10) the samples consisted of traditional plastic storage containers, drinking cups, and hand swabs.The drinking cup sample consisted o� the water poured into the cup or scooped �rom the storage container.The hands o� mothers or caregivers and children in a household were swabbed.Dung from a dung-smeared floor (Household 3) was also collected and analysed �or E. coli.All the households col� lected treated water �rom communal standpipes and the water supplied did not contain any E. coli.In Village 2 the same type of samples were collected from 5 households (Households 2, 4, 6, 8 and 10, see Table 2).All households except for Household � used untreated water obtained �rom a spring.E. coli isolated from the spring was included in the analysis.No E. coli could be isolated �rom the borehole used by Household �.
All samples collected in Village 1 were analysed by mem� brane filtration (Standard Methods, 1995) using mFC agar.To ensure the optimal recovery o� E. coli, between 200 and 500 mℓ of all water samples were filtered.The tip of each hand swab was rinsed in sterile saline which was then filtered and analysed �or the presence o� E. coli.The dung sample was analysed by suspending about 1 g in sterile saline and ten�old dilutions o� the suspension were analysed on mFC agar.The samples collected �rom Village 2 were analysed using the Colilert ® �1� system.For the water samples 100 mℓ were analysed.The hand swabs were thoroughly rinsed in 100 mℓ sterile water which was then ana� lysed in the same manner as the water samples.E. coli posi� tive wells in each Colilert tray were clearly marked.Each tray's upper surface was wiped clean with 70% alcohol.Wells were punctured with a sterile scalpel.A drop of the growth medium was trans�erred to an mFC agar plate and streaked �or single colonies.Plates were incubated at 44.5 ± 0.5°C.All isolates obtained from both villages were confirmed by testing for indole production.Isolates were stored on nutrient agar slopes at 4 to 8°C.
The AFLP method as described by Vos et al. (1995) was followed with some modifications as described by Brady et al. (2007).E. coli cells were collected �rom �reshly streaked nutrient agar plates incubated at 37ºC and genomic DNA was extracted from the cell pellets with a DNeasy Tissue Kit (QIAGEN, Hilden, Germany).About 100 ng of genomic DNA was digested with EcoRI (Roche, Basel, Switzerland) and MseI (Roche) restriction en�ymes.EcoRI and MseI adapters (Applied Biosystems, Foster City, California) were ligated to the generated DNA fragments according to the specifications of the manufacturer.Pre-ampli� fication and selective amplification was carried out as described by Brady et al. (2007).Sequences of the primers used are given in Table �.Pre-amplifi� cation The amplified DNA products were separated using an auto� mated PAGE gel system (Licor Global IR2 DNA analyzer, Licor Inc. Nebraska, USA).A sizing standard 50 to 700 bases (Li-Cor 4200-60[700]) was included in each run as a reference.The Automated Li-Cor system generated digitised fingerprints (16 Bit TIFF images) of the gel run, which were used in analysis with GelCompar II software (Applied Maths, Kortrijk, Bel� gium).Images were normalised by alignment to molecular-size standards loaded on each gel.Curve�based dendograms were generated using the Cosine correlation coefficient.

Results
E. coli strains were considered to be identical or o� the same type when AFLP banding patterns were more than 85% similar.Figure 1a shows the genetic relatedness o� the �� E. coli strains from Village 1. Twenty-two AFLP types were identified.Nine of these types contained more than one isolate.Of the 9 types only 5 (55%) consisted of E. coli isolated �rom a single sample type (AFLP types B,D,E,J,O) leaving 4 types that showed similar genetic patterns �or E. coli isolates that originated �rom di��er� ent sample types �rom di��erent points in a household or house� holds.Type A showed that E. coli identical to those isolated �rom the hands o� a person in Household � were also present in the cup sample o� the same household.Type F showed that E. coli, identical to isolates from a cup sample (Household 10), was also found in the storage container of the same household.Type N consisted o� identical E. coli that were isolated �rom storage con� tainers in different households (1 and 3).E. coli isolated �rom the dung sample, collected at Household � were identical to E. coli isolated �rom the hands o� a person in the same household (Type T).Eight genetically different strains were observed in the storage container of household 1 (AFLP types K, L, M, N, O, P, Q R).

109
Figure 1b shows the genetic relationship o� the �2 E. coli strains isolated �rom Village 2. Results include a spring shared by Households 2, 6, 8, and 10.Twenty-six AFLP types were identified of which 7 types were represented by more than one isolate.Of these 7 types 4 (57%) consisted of groups of identical E. coli isolated from a single sample type (AFLP types A, I, N, P) leaving 3 types that showed similar genetic patterns for E. coli isolates between di��erent sampling points within a house� hold or between different households.Type K included identical E. coli �ound in the storage container o� Households �, � and 10.In Type M, 10 out of 12 E. coli isolates originated �rom the storage container, the hands and the cup o� Household 2. The remaining two isolates originated from the spring (source of Households 2 and 6) and the storage container of Household 6. E. coli isolated �rom the hands o� a person in Household �, the storage vessel o� Household �, and the hands o� a person �rom Household � were also identical and belonged to type R.

Discussion
Our results document the genetic relatedness of E. coli isolated �rom storage containers and potential in�house contamination sources such as drinking cups and hands.Isolates were also obtained �rom contaminated source water used by some o� the houses.For one o� the households, cow dung used in the house was also sampled.A total of 96 E. coli isolates were obtained and typed using fluorescent AFLP.Overall the results showed a considerable degree o� genetic diversity among the E. coli iso� lated from the 8 rural households.All but one of the samples yielded E. coli belonging to more than one of the AFLP types.The highest diversity was noted �or the strains isolated �rom the storage vessel from Household 1 (Village 1), which showed 8 different AFLP types.The same level of diversity was observed amongst the strains isolated by means o� either the membrane method using mFC agar or the Colilert ® �1� system.
Isolates that originated from different types of samples col� lected within a household and which shared the same genetic profile provided a good indication of possible routes of post-col� lection contamination.In this study such isolates mostly came �rom closely linked points within the pathway o� water use in a single household.For e�ample isolates �rom the hands, cup and/or storage water could be genetically linked (Fig. 1a, AFLP types: A, and F; Fig 1b, AFLP types: K, M and R).These links are o� special importance in households which were using water o� good microbial quality.The reason is that the primary route �or introduction o� the E. coli detected in the stored water, almost certainly can be attributed to post�storage contamination.When source water of poor quality was used by households, the AFLP types observed suggested that the E. coli detected in these households may have originated from the source water (Fig. 1b, AFLP type M).
In this study only a small number of E. coli isolates �rom a limited number o� households were analysed but a high genetic diversity was observed.High genetic diversity levels in E. coli studies have also been reported by other researchers (Lu et al., 2004;McLellan et al., 2003).The high genetic diversity observed could be the result o� multiple �aecal contamination incidents but could also be e�plained by the persistence o� diverse E. coli strains in the environment.Milkman (1997) and LeClerc et al. (1996) have noted that recombination is an important and frequent process in E. coli.It is thus a feasible assumption that new types with changed genetic patterns are generated during regrowth in the environmental reservoirs (Gordon et al., 2002).Using a target organism such as E. coli with potentially high clonal diversity could there�ore complicate the genetic analysis o� isolates �ound in secondary habitats (Simpson et al., 2002;Gordon 2001).

Figure 1
Dendograms based on AFLP patterns of E. coli isolates obtained from Village 1 (Fig. 1a), and Village 2 (Fig. 1b).AFLP types in Fig. 1a are designated A to V and A to Z in Fig. 1b 110 Due to the high diversity of the strains many of the E. coli isolated �rom the stored water could not be linked to any o� the potential in�house contamination sources investigated.The iso� lation o� corresponding isolates, �rom environmental samples, might have been obscured by the presence o� the high diversity o� the other types within the same sample.To overcome this problem e�tensive isolation and screening o� isolates obtained �rom the environmental samples will have to be per�ormed.Other alternatives included the PCR-based detection of virulence factors (Gordon, 2001), the isolation of other species of enteric bacteria such as Enterococci (Hassan et al., 2007), alternative fingerprinting methods such as metabolic fingerprints (Ahmed et al., 2005), and the detection of polymorphisms within a single gene (Soule et al., 2006).

Conclusion
Overall a high degree of genetic diversity for the E. coli isolates was observed which may be indicative o� multiple sources o� E. coli organisms in the environment and which could not always be clearly linked with the way water was handled and used in these rural homes.However, in spite o� the high genetic diversity, linkages between isolates were observed between closely linked points within the domestic pathway o� water han� dling, e.g. the storage container, hand�swab sample, and the cup sample within a household.The high degree o� genetic diversity observed made it difficult to make specific inferences about the e�act point o� introduction or the e�act origin o� the E. coli con� tamination observed in the stored drinking water o� these house� holds.It confirmed recent conclusions (Gordon, 2001) that it is not �easible to use the relatedness o� genetic patterns o� com� mensal E. coli to determine the origin o� �aecal contamination o� water. ).In