Prevalence of Legionella spp. and Escherichia coli in the drinking water distribution system of Wrocław (Poland)

In many countries, tap water is one of the main sources of drinking water, and therefore, a high quality in terms of physical–chemical and microbiological conditions is necessary. The latter seems to be of even more importance for human health risk assessment because the presence of microbiological pathogens in drinking water may lead to numerous infections (Ashbolt 2015; Banach et al. 2019; Liu et al. 2019). Moreover, the environmental conditions in drinking water distribution systems may favour the proliferation of some bacteria (September et al. 2007; Falkinham et al. 2015). The most recognised waterborne pathogens from the kingdom of bacteria include Legionella pneumophila (L. pneumophila) and Escherichia coli (E. coli) (WHO 2017). Legionella spp. is mostly found in hot tap water, which provides these bacteria with the most appropriate conditions for growth and proliferation (WHO 2007). Nevertheless, some representatives of the genus may also be found in non-heated tap water, distributed to consumers' taps (Pereira et al. 2017). Legionella may cause legionellosis, Pontiac fever and also Legionnaires' disease. Pontiac fever is a non-lethal, self-limited flu-like illness. It was first recognised in the Country Health Department building in Pontiac (Michigan, USA). Legionnaires' disease, which may be fatal, was named after the American Legion convention at a Philadelphia hotel (Principe et al. 2017). The most common pathogenic Legionella species is L. pneumophila, although 27 other Legionella species are also recognised as human pathogens (Principe et al. 2017). The presence of representatives of the entire genus should therefore be monitored in drinking water samples. In Poland, 39 and 75 cases of legionellosis were reported in 2017 and 2018, respectively. The majority of them were Legionnaires' disease (38 and 70, respectively) (Czarkowski et al. 2019). However, there is a suspicion that Polish data regarding legionellosis are underestimated (Stypułkowska-Misiurewicz & Czerwiński 2016), and more attention should therefore be given to this issue.

E. coli is a natural commensal to humans and inhabits the human and animal intestine. Nonetheless, some strains may become invasive and cause many illnesses, for example moderate to severe diarrhoea (Vásquez-García et al. 2019). E. coli is also the most important indicator of potential faecal contamination (Paruch & Mæhlum 2012). The presence of Legionella spp. and E. coli cells in drinking water samples should therefore not be neglected.

The presence of bacteria in disinfected tap water was confirmed in many studies. The bacterial microbiome characteristic for tap water consists mainly of non-pathogenic species (Rudi et al. 2010; El-Chakhtoura et al. 2015; Douterelo et al. 2016). Nevertheless, insufficient source-water treatment and accidental contaminations within the distribution system may lead to the introduction of virulent genera. The most common ones, namely Escherichia and Legionella, are monitored continuously by local drinking water suppliers and sanitary-epidemiological stations in terms of cultivation of these bacteria on selective agar media. In recent years, E. coli was not reported in drinking water samples in Wrocław. Nevertheless, the use of the quantitative real-time polymerase chain reaction (qPCR) method in the detection and quantification of waterborne bacterial pathogens may contribute to the elucidation of the issue. It is worth mentioning, however, that the implementation of the qPCR method does not differentiate between live and dead organisms, unless a special procedure (for example, with propidium monoazide, PMA) is applied (Àlvarez et al. 2013; Toplitsch et al. 2018).

Bacteria dwelling in tap water distribution systems are generally known to be resistant to disinfectants (Khan et al. 2016). This feature allows them to overcome the presence of the disinfectant agent remaining in water samples. Despite water disinfection in water treatment plants and the presence of residual disinfectant in a network, some bacteria may enter the distribution system during pipe renovations or replacements. Therefore, even proper water treatment does not provide sufficient microbial quality within the entire network. In addition, genes responsible for the phenomenon of efflux pumps or other resistance mechanisms, such as the qacEΔ1 gene (conferring resistance to quaternary ammonium compounds) (Szekeres et al. 2018), may indicate the bacterial ability to overcome the disinfectant residual in the tap water distribution system. The qacEΔ1 gene is known to be the part of class 1 integrons, also known as ‘resistance integrons’ (Stalder et al. 2014). In this study, this gene was used as an indicator of resistance of bacteria.

If resistant, pathogens belonging to genera Legionella and Escherichia may cause a serious threat to human health because they might not be removed in the water treatment plant and can enter the distribution system. Bacterial cells could also participate in horizontal gene transfer where they are able to transform some genes (Gyles & Boerlin 2014). The detection of selected genes, which may enhance bacterial resistance to disinfectants in tap water samples, is therefore also important in the estimation of consumer health risk.

This study is aimed at the detection and determination of total quantity of (viable and non-viable) Legionella spp., L. pneumophila and E. coli bacterial cells, and their contribution to overall bacterial load quantified in microbiomes that are cultivated from consumer tap water samples using qPCR. In this study, the assumption was made that if Legionella or E. coli bacteria are able to proliferate in broth in the presence of other bacteria, they could pose a serious threat to consumers' health.

As mentioned above, some bacteria, including pathogens, are able to overcome the presence of disinfectants because they possess specific genes conferring increased resistance. Therefore, simultaneous monitoring of such genes in tap water microflora may important. In this study, the presence of the qacEΔ1 gene in cultivated microflora, as the representative gene conferring resistance to disinfectants, was determined.

Consumer tap water samples were collected from the drinking water distribution system in Wrocław, Poland. This research contributes to estimating the consumer health risk associated with tap water because, in Poland, tap water is generally considered to be drinking water for consumption directly from a tap.

Consumer drinking water samples were collected from 15 different points of the water supply system in Wrocław (Figure 1). All samples were taken at private homes. The number of measuring points depended on the size of the district and the availability of the collection point. The sampling points were chosen to cover the entire supply network evenly. Each sample was taken once. The sampling period was the spring and the summer of 2017. Before sampling, the taps were disinfected with 96% ethanol (Sigma), and water was flushed for 5 minutes. Each sample of 3 litres was filtrated on Whatman membranes (mixed-cellulose esters, diameter 0.2 μm). The membranes were submerged in sterile broth and incubated for 48 h at room temperature. This step was performed to cultivate the microflora dwelling in each tap water sample. Afterwards, the suspension was used for DNA extraction.

Genomic DNA was extracted using an Isolate II Genomic Kit (BIOLINE). The concentration of DNA was measured on a NanoPhotometer N60 (Implen).

The qPCRs (MIC, Bio Molecular Systems) were carried out to quantify Legionella spp., L. pneumophila and E. coli genomes, as well as 16S rRNA and qacEΔ1 genes.

The reactions for the quantification of Legionella spp., L. pneumophila and E. coli were performed with an AmpliTest (Amplicon) in accordance with the manufacturer's instructions. These quantification kits apply TaqMan probes, and are specific to target bacterial genome sequences. The AmpliTest also provides internal control, allowing for the detection of potential inhibition of reactions. The 16S rRNA and qacEΔ1 genes were quantified with a QUANTUM EvaGreen HRM Kit (SYNGEN) with the application of a non-specific dye. All the reactions for pathogenic genomes quantification were performed in technical triplicates, while 16S rRNA and qacEΔ1 genes were quantified in technical duplicates. 16S rRNA gene was quantified to determine the quantity of overall bacteria in the sample of cultivated microflora. The quantities of other genes were then referred to the quantity of 16S rRNA genes to make the comparison of samples possible. In this way, the relative abundance of Legionella and E. coli could be determined.

The conditions for qPCRs are presented in Table 1. In each reaction, negative control was applied and the calculations of genomes or gene copy number were based on standard curves. For Legionella spp, L. pneumophila and E. coli genomes, standard curves were prepared as described in the previous study (Wolf-Baca & Siedlecka 2019), resulting in calculated genomes per reaction. For the purpose of 16S rRNA standard curve preparation, the genomic DNA ten-fold dilution of E. coli ATCC 25922 (BioMaxima) strain was used. In the case of qacEΔ1 gene, the ten-fold dilution of purified PCR product was used.

All qPCRs efficiencies were in a range of 0.88–1.09. The limits of quantification were 178 genome copies/reaction for Legionella spp. and L. pneumophila, and 1.59 genome copies/reaction for E. coli. In the case of the 16S rRNA and qacEΔ1 gene, the limits of quantification were 16.9 and 184 gene copy number/reaction, respectively (Figure 2).

The abundance and prevalence of Legionella spp. were determined; however, no L. pneumophila target sequence was found in any tested samples. In the case of E. coli, only one sample (number 13) gave a positive signal with calculated genome copy number of 0.13 genomes/ng DNA. Moreover, no qacEΔ1 gene could be detected in any sample.

The results of calculations of genome copy number of Legionella spp. per ng of DNA in tested samples are presented in Table 2. In general, Legionella spp. were detected and quantified in 11/15 (73.3%) samples.

Legionella spp. was the most abundant in sample number 15, where it reached 71 copies of genome per ng of DNA. Further analysis of the data indicates that Legionella spp. prevalence was relevant only in samples 9 and 15, reaching 1.1 × 10⁻³ and 6.15 × 10⁻⁵ genomes of Legionella spp. per 16S gene copy number, respectively.

The objective of this study was to estimate the potential consumer health risk associated with the presence of pathogenic bacteria in non-boiled tap water samples using the qPCR approach. The usefulness of this method in Legionella quantification was proven a decade ago (Dusserre et al. 2008; Douterelo et al. 2014), but it is worth mentioning that promising methods for L. pneumophila detection other than qPCR include an immunodetection system described by Párraga-Niño et al. (2018) or Legiolert/Quanti-Tray MPN described by Spies et al. (2018).

Among the tested pathogens, L. pneumophila was not found in any of the samples. This proves the good quality of Wrocław drinking water and treatment processes, and suggests a lowered risk for consumer health. Nevertheless, L. pneumophila is not the only factor causing legionellosis (Principe et al. 2017), and therefore such risk could not be fully excluded.

E. coli is also a risk to consumer health and is, one of the most important faecal contamination indicators (Coleman et al. 2011; Subba et al. 2013; Talukdar et al. 2013). It was detected in only one sample with no evidence of its viability. Moreover, even assuming that E. coli did not proliferate in broth, i.e. its quantity did not change during incubation, the calculated genome copy number/mL would be 9.10 × 10⁻³ (data not shown), which is far below the acceptable values specified in Polish law (Regulation 2017). The detection of E. coli in this sample may be the effect of the presence of residual DNA fragments with E. coli-specific sequence. As no live/dead treatment was performed before DNA extraction (Àlvarez et al. 2013; Toplitsch et al. 2018), it is not excluded that DNA fragments detected in sample number 13 originated from dead E. coli cells. Nevertheless, these residual DNA fragments may suggest contamination, which has happened in the past.

The results indicate that Legionella spp. could still be present in microflora cultivated from tap water samples, even if no Legionella-enhancing supplement was added to the broth. However, based on this study, it could not be established whether the quantified cells were dead or alive.

The prevalence of Legionella spp. genomes in microflora cultivated from tap water samples tested in this study was generally low. It is worth mentioning that viable but non-culturable Legionella cells may be predominant in tap water samples (Dietersdorfer et al. 2018). This may explain the low level of Legionella spp. prevalence observed in this study. Other bacteria present in a given sample might proliferate, while Legionella spp. remain at the same level during a short incubation time. Contrary results indicate very high contamination of Legionella in Australian tap water, mostly in samples collected in the summer season (Whiley et al. 2014). Because the sampling campaign performed in this study was undertaken only once at the turn of spring and summer, the seasonal variation of contamination by pathogens could not be excluded.

Despite its low levels, Legionella spp. was frequently detected and quantified in this study. Similar results were obtained by Liu et al. (2019), who found Legionella spp. in 100% samples, while the presence of L. pneumophila was confirmed only in 9.1% of all samples.

Waak et al. (2018) tested the presence of Legionella spp. in water-main biofilms (scraped from pipes) and tap water samples from two distribution systems: a chloramine-containing system in the United States and a system in Norway that does not maintain a residual disinfectant. They were able to detect Legionella spp. in 43% of biofilm samples from the Norwegian system with no residual disinfectant, with a maximum concentration of 7.8 × 10⁴ gene copies cm⁻². Legionella spp. was not detected in biofilm samples from the US chloraminated system. In the case of tap water samples, Legionella spp. was detected in 26% and 64% of tested samples from the US and Norwegian system, respectively, with a maximum concentration of 2.0 × 10³/L for chloraminated system, and 2.8 × 10³/L for the system with no residual disinfectant. The results obtained in this study are higher than those presented by Waak et al. (2018) as 73.3% of tested tap water samples were Legionella spp. positive. Waak et al. also detected a L. pneumophila gene marker in 3% and 9% of tested biofilm samples from the US and Norwegian system, respectively, while tap water samples from both sampling regions were free from L. pneumophila, which is in accordance with results presented in this paper. L. pneumophila serogroup 1 gene marker was not found in any of the tested samples of water-main biofilms or tap water in the US and Norwegian systems (Waak et al. 2018).

Legionella is often tested only in hot tap water installations (Wolf-Baca & Siedlecka 2019); indeed, Polish law demands Legionella monitoring only in hot water samples (Regulation 2017). However, its proliferation in heating units is initiated by the presence of these bacteria in the water distribution system. There is therefore a risk that in point-of-use faucets with considerable Legionella spp. prevalence, much more Legionella cells may be found in hot water samples. Intake of aerosols with Legionella during the use of hot tap water (for example, for baths and other domestic purposes) is known to pose a potential risk of legionellosis (Wong et al. 2006). Detection of the Legionella spp. sequence targets in drinking water samples should therefore not be neglected. Replacement of chlorine with monochloramine was suggested as a method for improved removal of Legionella from hot tap water (Flannery et al. 2006; Moore et al. 2006). The importance of the presence of residual disinfectant in the drinking water distribution system was confirmed in the study by Waak et al. (2018). The US chloraminated drinking water distribution system was shown to be less contaminated with Legionella and overall bacteria (in the case of water-main biofilm samples from pipes and from tap water samples) than the Norwegian system with no residual disinfectant (Waak et al. 2018).

As mentioned in the Introduction, the qacEΔ1 gene was used as a representative gene of class 1 integrons, conferring resistance of bacteria (Stalder et al. 2014; Szekeres et al. 2018). Its lack in cultivated microflora may therefore lead to the conclusion that the overall resistance in bacteria collected from tap water samples is low. Constant chlorination of the Wrocław drinking water distribution system and the assurance of appropriate levels of residual disinfectant along the entire pipeline network should therefore be sufficient in prevention the proliferation of bacterial pathogens in consumers' tap water. However, the disinfection processes may also lead to the formation of disinfection by-products, which could be considered as another threat to consumer health (Pogorzelec & Piekarska 2018). Moreover, it is not excluded that bacteria susceptible to quaternary ammonium compounds can be resistant to chlorine compounds. The resistance of bacteria dwelling in tap water to chlorine and its compounds should therefore be tested more in-depth in future studies.

Future studies should also investigate the culture methods for enumeration of pathogenic bacteria, including, for example, the heterotrophic plate count method on selective agar media for Legionella and Escherichia.

This study investigates the presence of Legionella and Escherichia coli in the public water supply system of Wrocław, based on the results of qPCR analyses of 15 samples collected from the entire network. The prevalence of Legionella spp. cells in microbiome of tap water samples collected from 15 sampling points of drinking water distribution system in Wrocław was generally low. Only in two points samples (9 and 15) was the quantity of Legionella spp. in overall bacteria load increased. Neither L. pneumophila nor E. coli (except sampling point number 13) were detected. These data suggest the sufficient quality of consumer tap water in Wrocław. The qacEΔ1 gene, conferring resistance to quaternary ammonium compounds in class 1 integrons, was not detected in the samples, suggesting low resistance of cultivated bacteria. Nevertheless, the presence of Legionella spp. in drinking water distribution systems may lead to their proliferation in heating installations, posing a health risk to tap water consumers.