Quantitative real-time PCR does not reliably detect single fecal indicator bacteria in drinking water

The microbial quality of drinking and environmental water is usually determined by culture-based detection of fecal indicator bacteria according to ISO reference methods 16649-1 and 7899-2, respectively. Becauseofanincreasingdemandfor rapid, culture-independent methods,we testedthree quantitative polymerase chain reaction (qPCR) approaches for the simultaneous detection of both, Escherichia coli and Enterococcus spp., using either 16S rRNA or 16S rDNA as a target molecule. Filter sterilized drinking water was arti ﬁ cially contaminated with bacteria from either high or low nutrient culture conditions and directly analyzed after membrane ﬁ ltration without any other enrichment. Depending on the culture condition used, qPCR analyses revealed a lower limit of detection of 1 – 10 E. coli /100 ml and 10 – 100 E. faecalis /100 ml, respectively. In addition, the microbial quality of different surface water samples was monitored. The analyses revealed a clear correlation between viable cell counts and qPCR data. However, the safe and reliable detection of 1 CFU/100 ml failed.


INTRODUCTION
Fecal contaminated drinking water can cause diarrhea if pathogens are ingested. Worldwide, over 80% of cases of diarrhea are associated with unsafe drinking water, lack of sanitation or lack of hygiene. This leads to 1.5 million deaths by diarrhea annually, particularly in developing countries (Prüss-Üstün et al. ). Water-borne diseases have also been reported in industrial nations, but to lesser extents (Maurer & Stürchler ; Hrudey et al. ).
Escherichia coli and species of the fecal Enterococcus group (Enterococcus spp.) are the most important indicators of fecal drinking water contamination. Both must not be present in a 100 ml sample volume (Anonymous ; Anonymous ; WHO ). Presence of fecal contamination by E. coli in drinking water indicates that pathogenic bacteria may also be present in a sample. E. coli is considered as the best biological representative of (fecal) pathogens in drinking water, as it is present up to 94.1% in human feces and up to 92.6% in animal feces. It is a reliable biological drinking water indicator for public health protection (Edberg et al. ). Hence, even 1 CFU/100 ml indicates that pathogens might be present, the latter proposing a health risk. The microbial quality of natural bathing waters, i.e. rivers, ponds, and lakes, is defined by the same hygiene indicators.
Admittance of bathing in such waters is based on health grounds according to the classification of four quality groups in response to the CFU counts of both E. coli and Enterococcus spp. in 100 ml (Schaffner et al. ). In their review article about the detection of microorganism in water by PCR methods Botes et al. () concluded that standardized protocols and improvements in method validation are needed for qPCR-based microbial water analysis. In order to address these issues, this study intended to develop a culture-independent TaqMan ® (hydrolysis probe)/qPCR-based protocol for the simultaneous detection of E. coli and Enterococcus spp., which is applied directly after membrane filtration using 16S rRNA or 16S rDNA as target molecules without any enrichment cultivation. The microbial quality of untreated drinking water and environmental samples was determined applying different qPCR approaches and the reference methods.

Viability dyeing and crosslinking
Either 100 ml or 1,000 ml water samples were artificially contaminated with heat-treated and non-heat-treated bacteria and were membrane filtered (0.45 μm, Ø 47 mm, Sartorius, Microsart CN Filter). The filter membrane was then placed into a sterile petri dish (Ø 60 mm), covered with 1 ml of 0.9% (w/v) NaCl and 10 μl PMA (200 μM). In the non-treated control 10 μl 0.9% (w/v) NaCl was added.
Incubation of immersed membranes (with and without PMA) was performed in a light-proof Styrofoam box covered with aluminum foil (30 min, 30 rpm, room temperature).
Cross-linking was performed for 30 min and 30 rpm at room temperature using LED lamps (470 nm) positioned in a self-made lid box lined with aluminum foil inside.

DNA isolation
Either 100 ml or 1,000 ml of artificially contaminated water samples were membrane filtered ( To eluate isolated DNA the spin filter was loaded with 30 μl of the elution buffer, incubated at 50 W C for 5 min (Thermomixer comfort, Eppendorf) and centrifuged (13,000 × g, 1 min). DNA elution was performed twice using the first eluate for the second elution.

RNA isolation and reverse transcription
Either 100 ml or 1,000 ml of the inoculated water samples were membrane filtrated using a 0.45 μm syringe filter unit  (R 2 ¼ 0.9905)) where x gives the copy numbers after C t -value is applied for y.

Water sampling and analysis
Drinking water samples were collected before UV disinfection. Environmental water samples were taken from rivers, lakes, and natural ponds in the region of Zurich (Switzerland). All waters were sampled using sterile PET bottles containing 20 mg/l Sodium Thiosulfate (Huber Lab) and stored at 4 W C. Analysis of 100 ml sample volumes were carried out within 24 hours as described above.
Additionally, the turbidity of environmental water samples was measured using a portable turbidimeter (Hach, 2100QiS) when v-qPCR was applied.   Figure S1, Supplementary material).

RESULTS AND DISCUSSION
The LLOD of E. faecalis was not altered (Table 1, Figure S2   These findings are in agreement with our data demonstrating that at least 10 CE E. coli or E. faecalis need to be present on the filter membrane for a reliable positive qPCR result regardless of the applied sample volume. Applying the reference methods a minimum of 1 CFU/100 ml was detected.

Detection of fecal indicator bacteria in water samples
The analysis of 54 drinking water samples revealed no or very low (<10 CFU per 100 ml) microbial contamination with  Viability dyeing is not applicable for microorganisms in water, which was disinfected by UV light. False-negative results will be reported, because the cell membrane remains intact while the DNA is damaged (Nocker et al. ).
Hence, in this study v-qPCR was applied to untreated environmental water samples with a turbidity <10 Nephelometric Turbidity Units (NTU). A turbidity >10 NTU negatively influences PMA treatment and detection by v-qPCR (Fittipaldi et al. ). Little or no difference in qPCR results were determined for samples treated with or without PMA which is in accordance with other studies (Varma et al. ). Only in a few cases qPCR results of PMA treated samples differed from PMA non-treated samples ( Figure S7, Supplementary material, available with the online version of this paper). In these cases, the ratio between dead and viable cells was either >1,000, or dead cell numbers were >10 4 and viable cell counts <10 3 .
Such conditions significantly reduce the efficiency of PMA