The occurrence of pathogenic Escherichia coli in South African wastewater treatment plants as detected by multiplex PCR

The aim of this study was to investigate the use of PCR to detect commensal and diarrhoeagenic Escherichia coli concentrated from water samples using membrane filtration. To achieve this, culture-based and PCR-based methods were compared for the detection of E. coli in raw sewage and primary, secondary and tertiary effluents from 6 wastewater treatment plants around Johannesburg, Gauteng. E. coli was concentrated from the samples using standard filtration techniques with subsequent incubation on E. coli/coliform chromogenic media to determine the E. coli levels. Bacterial DNA was isolated from bacterial colonies trapped on polyethersulphone membranes after filtration using a celite/guanidium thiocyanate method. A single multiplex PCR (m-PCR) assay was used that targeted the mdh, eaeA, stx1, stx2, st, lt, ial and eagg genes associated with diarrhoeagenic E. coli. The mdh gene was detected in all of the samples even if no culturable E. coli was detected. All the diarrhoeagenic E. coli types were detected in one or more of the raw sewage samples from the various plants. EPEC was present in 20% (2/10) of the samples, EHEC in 50% (5/10), ETEC in 80% (8/10), EIEC in 10% (1/10) and EAEC in 90% (9/10) of the samples. In the case of the primary and secondary treatment only ETEC (5/5; 100%) and EAEC (5/5; 100%) were detected in all of the samples. The results demonstrate that molecular techniques such as PCR have the potential to be used for the monitoring of water samples for the presence of pathogenic E. coli, without the need to culture the organisms.


Introduction
In South Africa many communities still depend on untreated surface water and ground-water sources for their daily water needs.Water from these sources is often contaminated by faecal pollution from wastewater effluents (Toze, 2004).Wastewater is a matrix consisting of raw sewage and primary, secondary and tertiary treatment effluents (Mara and Horan, 2003).Final effluents are released into receiving waters like dams and rivers.If treatment fails, effluents of poor microbial quality enter public waters (Jagals, 1997).
Microbiological indicators have been used for decades to monitor faecal pollution of water (Grabow, 1996;Standard Methods, 2005) as well as the possible presence of other microbiological pathogens (Medema et al., 2003).Escherichia coli is used as an indicator of faecal pollution that originates from human and warm-blooded animals.In the last century, certain E. coli strains were identified as a significant cause of gastro-intestinal disease and recognised as highly versatile pathogens (Nataro and Kaper, 1998).At present, 6 groups of E. coli patho-types have been identified, of which 5 were selected for this study based on their association with diarrhoeal disease in South Africa.These are classified into 5 categories, namely, entero-pathogenic (EPEC), entero-toxigenic (ETEC), entero-invasive (EIEC), entero-aggregative (EAEC) and entero-haemorrhagic (EHEC) (Ashbolt, 2004;Kaper et al., 2004).E. coli strains can thus be either commensal (ComEC; non-diarrhoeagenic) or diarrhoeagenic.
Traditional E. coli detection methods are based on chromogenic and fluorogenic media designed to enumerate ComEC with no relationship to the proportion of diarrhoeagenic E. coli (DEC).All wastewater treatment plants use culture-based methods based on this principle to monitor the water quality.One complication with culture-based methods is their inability to detect E. coli in the viable but non-culturable (VNBC) state (Lleo et al., 2005).VBNC cells are bacterial cells that are metabolically active but cannot be cultured using standard microbiological techniques.These cells are reported to be able to resume active growth when favourable conditions are restored (Chen et al., 2006).It has been reported that pathogenic VBNC cells retain their virulence genes making the water source a potential reservoir of disease (Garcia-Armisen and Servais, 2004).
This inability of traditional culture methods to detect VBNC cells has prompted interest in alternative techniques to monitor the microbiological quality of water (Lleo et al., 2005).Molecular biology techniques, such as the polymerase chain reaction (PCR), have been used to detect DEC types by amplifying specific genes associated with the bacterium's virulence (Kong et al., 2002).To date PCR has mostly been used for the characterisation of isolated E. coli cells or the detection of DEC after sample enrichment.DEC genes targeted with PCR include the eaeA gene for EPEC (Aranda et al., 2004), the lt and st enterotoxin genes for ETEC (Pass et al., 2000), ial gene for EIEC (Paton and Paton, 1998), the eagg gene for EAEC (Kong et al., 2002) and the eaeA, stx1 and stx2 for the detection of EHEC (Moses et al., 2006).
PCR can, however, not be used for the detection of VBNC cells in water samples if the method is used to characterise bacterial isolates obtained with standard culture methods, since the

173
VBNC bacteria cannot be isolated (Chen et al., 2006).The simplest way to overcome this would be to isolate DNA from bacterial cells concentrated from the water samples, which is then used as template for the PCR, circumventing the need for culturability.
The aim of this study was to investigate the use of PCR to detect commensal and pathogenic E. coli concentrated from water samples using membrane filtration.During the study both PCR and culture-based methods were used to detect E. coli in raw sewage, and primary, secondary and tertiary effluents from 6 wastewater treatment plants in Gauteng.

Methodology Growth and maintenance of reference strains
The E. coli reference strains (Table 1) were cultured on plate count agar (PCA) (Oxoid) and incubated under aerobic conditions at 37ºC for 16 h.If a liquid culture was required, E. coli colonies were inoculated into 5 mℓ nutrient broth and grown overnight at 37°C with mild agitation at 200 r/min.

Sample collection
Samples were collected from 6 wastewater treatment plants in the vicinity of Johannesburg, Gauteng, from August 2006 to December 2007.Thirty-four samples were collected from raw sewage (10), primary treatment effluent (5), secondary treatment effluent (5) and tertiary effluent (14).Water samples were collected in 1 ℓ sampling bottles and kept on ice during transport.Samples were analysed within 3 h of collection.

Microbiological analysis
Microbiological analysis of the samples for E. coli was performed using the standard filtration technique (Standard Methods, 2005).Briefly, 100 mℓ of an appropriate dilution of the water sample was filtered in triplicate onto 0.45-μm gridded nitro-cellulose membrane filters (Merck, Germany).The filters were placed onto selective E. coli/Coliform Chromogenic Media (Oxoid, UK) and the plates were incubated for 24 h at 37ºC.All pink colonies were counted as total coliforms (glucoronidase positive) and all purple colonies were counted as E. coli (β-D -glucoronidase and β-D -galactosidase positive), according to the supplier's instructions.Bacterial counts obtained were expressed as colony forming units in 100 mℓ (cfu/100 mℓ), using the average count of the 3 filters.

Bacterial DNA extraction from wastewater samples
For the DNA extraction, a 100 mℓ sample was filtered onto polyether sulfone (PES) membranes (Microsep (PTY) LTD).DNA was extracted from the trapped bacteria using a modification of the silica/guanidium thiocyanate method reported by Boom et al. (1990) with spin columns prepared as reported by Borodina et al. (2003).The Boom et al. (1990) protocol was followed with 2 exceptions: 250 µℓ 100% (vol/vol) ethanol was added to the lysis buffer, and the celite containing the bound DNA was loaded into the spin columns prior to the washing steps (Borodina et al., 2003).DNA was eluted with 50 µℓ Qiagen elution buffer (Southern Cross Biotechnology ® ).Controls were included during the DNA extractions and subjected to the same protocol as the samples.The positive control contained 1 mℓ of overnight commensal E. coli liquid culture and the negative control contained 1 mℓ of sterile distilled water.

Multiplex polymerase chain reaction (m-PCR)
The oligonucleotide primers used in this study were synthesised by Inqaba Biotec (Pty.) Ltd. as well as Whitehead Scientific (Pty.) Ltd.The sequences, target genes and expected amplification products are listed in Table 2.
All m-PCR reactions were performed in a Biorad Mycycler TM Thermal cycler in a total volume of 20 mℓ.A m-PCR kit (Qiagen ® ) was used for the m-PCR protocol.Each reaction consisted of 1X Qiagen ® PCR multiplex mix (containing HotstartTaq ® DNA polymerase, m-PCR buffer and dNTP mix); 2 μℓ of the primer mixture (0.1 mM of mdh and lt primers Forward (F) and Reverse (R)), 0.2 mM of ial and eagg primers (F and R), 0.3 mM of eaeA and stx2 primers (F and R), 0.5 mM of stx1 and st primers(F and R) (Table 1), 4 mℓ of sample DNA and 4 mℓ PCR grade water.The reactions were subjected to an initial activation step at 95°C for 15 min, followed by 35 cycles consisting of denaturing at 94°C for 45 s, annealing at 55°C for 45 s, extension at 68°C for 2 min and final elongation at 72°C for 5 min (Omar, 2007).Positive and negative controls for the PCR reaction were also included.The positive control contained a mixture of the 5 pathogenic E. coli and commensal E. coli DNA.The negative control contained PCR grade water.

Gel electrophoresis
DNA was analysed on a horizontal agarose slab gel (2.5 % (w/v)) with ethidium bromide (0.5 mg/mℓ) in TAE buffer (40 mM Tris acetate; 2 mM EDTA, pH 8.3).Electrophoresis was performed for 1 to 2 h in electric field strength of 80 V; PCR products were visualised with UV light (Syngene, UK).This procedure was followed for all the experiments except where stated differently.The relative sizes of the DNA fragments were estimated by comparing their electrophoretic mobility with that of the standards run with the samples on each gel, either 1 kB or 100 bp markers (Fermentas, US).
As expected the number of E. coli present in the samples declined from raw sewage (starting material) up to the tertiary effluent (final product).Bacterial counts for the tertiary effluents did however cover a broad range from 2.67 x 10 0 cfu/100mℓ up to 7.15 x 10 4 cfu/100mℓ, which provided a range of contexts in which to test PCR as an alternative for the detection and characterisation of E. coli strains present.
The m-PCR used for the sample analyses targeted 7 of the virulence genes mostly associated with EHEC (stx1, stx2, eaeA), EPEC (eaeA), ETEC (lt, st), EIEC (ial), EAEC (eagg) as well as the mdh gene found in both commensal and diarrhoeagenic E. coli (Omar, 2007).An example of the complete m-PCR is shown in Fig. 1.  175 Individual E. coli colonies isolated on selective media were not tested using the m-PCR.This was avoided as pathogenic strains would represent only a small proportion of the total E. coli community.
The initial PCR results obtained showed PCR inhibition in 21% (7/34) as indicated by the absence of an mdh product.Since all the samples had culturable E. coli it was expected that at least the mdh gene would be detected in the samples.The mdh gene could be detected in all 7 samples after the DNA was diluted 5 times, lifting the effect of the PCR inhibitors.The dilution of DNA should always only be considered as a last resort since this can possibly limit the detection of DEC in the sample, especially if the DEC is present in very low numbers.Due to the nature of the samples it was, however, necessary.
The general distribution of the DEC types in the various stages of the wastewater treatment plants is given in Table 4.All the DEC types were detected in 1 or more of the raw sewage samples from the various plants.EPEC was present in 20% (2/10) of the samples, EHEC in 50% (5/10), ETEC in 80% (8/10), EIEC in 10% (1/10) and EAEC in 90% (9/10) of the samples (Table 4).In the case of the primary and secondary treatments only ETEC (5/5; 100%) and EAEC (5/5; 100%) were detected in all of the samples.All the DEC types except for EIEC could be detected in the tertiary effluent samples.The presence of EHEC (2/14; 14.3%) and EPEC (2/14; 14.3%) was however restricted to Plant F (Table 5).The tertiary effluent of Plant F not only showed the presence of high E. coli counts but also the presence of EHEC, EPEC, EAEC and ETEC in both samples tested.
In the case of Plant A, ETEC and EAEC could be detected in the raw sewage, primary and secondary treatment effluents as well as the tertiary effluent.The pathogens were, however, not present in all of the samples taken from each point.The same trend was seen with Plant B with the exception that ETEC was detected in all of the samples tested.
Although it was expected that the mdh gene would at least be detected in the raw sewage and primary and secondary treatment effluents due to the high number of E. coli present, the ability of the PCR method in detecting not only mdh but also virulence genes in the tertiary effluents was very encouraging.The mdh gene could even be detected in samples with low or no E. coli detected using culture-based methods.It should, however, be noted that although DEC were detected with the PCR, especially in the tertiary effluent, no conclusions can be made about the viability of the bacterial cells detected.This is mainly due to the fact that PCR will amplify genes from the DNA of both viable and dead cells.

Conclusion
PCR has the potential to be used for the monitoring of water samples for the presence of pathogenic E. coli without the need to culture the organisms.This method can be extended to test for other bacterial pathogens such as Vibrio cholerae in water samples if an appropriate PCR protocol is available.For this process to be more efficient more research is needed into the DNA extraction method to ensure that no PCR inhibitors are present after DNA extraction, especially when working with samples such as raw sewage.This will ensure that no DNA dilutions are required that might reduce the chance of detecting bacterial pathogens in the samples.
It is important to consider that although the pathogens were detected with the m-PCR, the genes could have been amplified  from DNA recovered from dead and/or viable cells.Since humans might use the water downstream of the treatment plant, there is a need to better understand what pathogens are released and whether they are still virulent.Olivier et al. (2005) showed that E. coli and Salmonella typhimurium can enter the viable but non-culturable state if secondary treated wastewater is chlorinated, leading to a drastic underestimation of the actual quality of the water released into the environment.
Figure 1 Photograph of an agarose gel showing the PCR products obtained for the E. coli multiplex PCR (Lane 2).The molecular weight marker is shown in (Lane 1)

Table 1 Bacterial strains used in molecular characterisation Bacterial strain Reference nr Genes present
b Strains purchased from National Health Laboratory Services (NHLS) Available on website http://www.wrc.org.zaISSN 0378-4738 (Print) = Water SA Vol.36 No. 2 Young Water Professionals Special Edition 2010 ISSN 1816-7950 (On-line) = Water SA Vol.36 No. 2 Young Water Professionals Special Edition 2010 174

Table 3 Microbiology and PCR results obtained for the raw sewage, primary treatment, secondary treatment and tertiary effluent samples taken from 6 wastewater treatment plants in the vicinity of Johannesburg Microbiology PCR Sample n Average cfu/100 mℓ Min cfu/100 mℓ Max cfu/100 mℓ Pathogens detected
Available on website http://www.wrc.org.zaISSN 0378-4738 (Print) = Water SA Vol.36 No. 2 Young Water Professionals Special Edition 2010 ISSN 1816-7950 (On-line) = Water SA Vol.36 No. 2 Young Water Professionals Special Edition 2010

Table 5 Summary of the total percentages of pathogenic E. coli present at the various wastewater treatment plants
Available on website http://www.wrc.org.zaISSN 0378-4738 (Print) = Water SA Vol.36 No. 2 Young Water Professionals Special Edition 2010 ISSN 1816-7950 (On-line) = Water SA Vol.36 No. 2 Young Water Professionals Special Edition 2010 176