Propidium monoazide-quantitative PCR to detect viable Legionella spp. in the supply process of tap water Open Access

Legionella are ubiquitous bacteria existing in various aquatic environments from natural to artificial such as building, cooling towers, and so on (Delgado-Viscogliosi et al. 2009; Bonetta et al. 2010; Chang et al. 2010). Legionnaires’ disease or Pontiac fever, collectively known as legionellosis, is caused by inhaling small water droplets containing Legionella from the air (Delgado-Viscogliosi et al. 2009; Chang et al. 2010; Yáñez et al. 2011). Legionellosis outbreaks occur frequently in water distribution systems, and the average mortality rate of Legionnaires’ disease is about 15–20% of hospitalized cases (Roig & Rello 2003; Bonetta et al. 2010; Ditommaso et al. 2020). Legionella are relatively resistant to water disinfectants and can be detected in potable water (EPA 2000). Therefore, the rapid monitoring of Legionella in water distribution systems should be prioritized to detect contamination and prevent disease (Delgado-Viscogliosi et al. 2005).

In general, cultivation is the most common method for detecting Legionella in water (EPA 2000; APHA 2017). However, detection of Legionella by culture is time-consuming because it requires a long incubation time to grow colonies (Delgado-Viscogliosi et al. 2009; Yáñez et al. 2011). Furthermore, it may be difficult to isolate Legionella due to the growth of other microorganisms, and the viable but nonculturable (VBNC) state cannot be detected. (Bonetta et al. 2010; Yáñez et al. 2011). Since Legionella in this state still pose a public health risk, it is necessary to improve the limitations of the culture method (Delgado-Viscogliosi et al. 2005; Yáñez et al. 2011).

Real-time PCR (qPCR) allows the rapid detection and quantification of target DNA with high sensitivity and specificity, including viable but nonculturable cells, e.g., in a few hours (Delgado-Viscogliosi et al. 2005; Bonetta et al. 2010; Yáñez et al. 2011; Taylor et al. 2014). However, the detection of nonviable bacteria by qPCR may overestimate the risk of legionellosis (Delgado-Viscogliosi et al. 2009). Recently, propidium monoazide (PMA) has been used to detect viable bacterial cells selectively (Nocker et al. 2006; Chang et al. 2010; Taylor et al. 2014). PMA, a DNA intercalating dye, can selectively enter compromised cells and bind to their DNA (Chang et al. 2010). The DNA bound to PMA can get removed along with cell debris during DNA extraction or suppress PCR amplification (Nocker et al. 2006). Thus PMA combined with qPCR (PMA-qPCR) may quantify the DNA of viable cells through the selective removal of DNA from nonviable cells (Nocker et al. 2006; Chang et al. 2009; Yáñez et al. 2011).

This study aimed to optimize qPCR combined with PMA for the selective detection of viable Legionella with regard to PMA concentration and DNA amplicon size. In addition, we evaluated the ability of PMA-qPCR to monitor viable Legionella in the process of supplying tap water.

The Legionella strain used in this study was Legionella pneumophila (ATCC 33152). The strain was grown on buffered charcoal yeast extract (BCYE) agar (Oxoid, France) for 4–7 days at 37 °C.

Five samples were collected from the five intake stations located between the Paldang dam and Jamsil weir on the Han River in Korea (Figure 1). The water from these intake stations is used as source water to produce tap water for Seoul. In order to study the effect of turbidity on PMA application, two samples out of the five Han River samples were collected when there was no rain for more than 5 days and designated as River 1 and 2. Three samples were collected when there was rain within 1–3 days of sampling and designated as River 3, 4 and 5. Each sample was collected in a 1 L sterile polypropylene bottle. In addition, two samples were taken from purified water produced at drinking water treatment plants and three cold water samples used as drinking water were taken from taps located at terminal sites of the distribution system in Seoul (Figure 1). The purified water and tap water samples were collected in sterilized 1 L polypropylene bottles containing sodium thiosulfate to neutralize residual free chlorine. Two purified water samples were labelled as Purified water 1 and 2, and three tap water samples were labelled as Tap water 1, 2, and 3. Ten total samples were collected from January to December 2013. Samples were transported to the laboratory in an icebox and processed within 24 h of collection. For all samples, turbidity was measured according to the Korean standard method for the examination of water using a HACH 2100AN turbidimeter (Ministry of Environment 2013; HACH, USA). 500 mL sample was taken from each 1 L bottle and centrifuged at 4,000 × g for 15 min. The supernatant was removed and each concentrate transferred to a 50 mL sterile tube, and centrifuged at 4,000 × g for 15 min. The supernatant of a 50 mL tube was removed and each concentrate transferred to a 15 mL sterile tube, and centrifuged again at 4,000 × g for 15 min. The supernatant of a 15 mL tube was removed and each concentrate transferred to a 1.5 mL microcentrifuge tube, and centrifuged at 8,000 × g for 5 min. The pellet from each tube was then resuspended in 1 mL of sterilized phosphate buffered saline (PBS; pH 7.0) and each sample was divided into four 250 μL samples. The four samples were used for PMA-qPCR and qPCR analysis by adding 250 μL live or dead cells, respectively. A positive control was prepared by mixing each 250 μL of live cells and sterilized PBS to achieve a final volume of 500 μL.

The colonies of L. pneumophila were suspended in sterilized PBS. The suspension was serial 10-fold dilution in PBS. The cell concentration was adjusted to approximately 10⁴–10⁷ CFU/mL. To test PMA effects on dead cells, Legionella strains suspended in PBS were exposed for 15 min at 72 °C using standard laboratory heat block. 250–500 μL aliquots of live and dead cells were used for the experiment.

PMA (Biotum, CA, USA) was dissolved in 20% dimethyl sulfoxide (DMSO) to obtain a stock solution of 20 mM. Legionella cells and water samples were treated with PMA to make final concentrations of 0, 25, 50, 100, 200, and 400 μM as previously reported (Chang et al. 2010; Yáñez et al. 2011). The samples (500 μL) were incubated in the dark for 5 min with occasional mixing and exposed for 5 min on ice to a 500 W halogen lamp at a distance of 20 cm. Thereafter, the samples were centrifuged at 13,000 rpm for 3 min prior to DNA isolation.

The total genomic DNA from pure cultures was extracted by the QIAamp DNA mini kit (QIAGEN, Germany) and the water samples by the FastDNA spin kit for Soil (MP biomedicals, USA) according to the manufacturer's manual. Four qPCR methods were applied to quantify different sizes. One method was performed by the iQ-Check^TM Quanti Legionella Kits (Bio-Rad, USA) using probe, reagents, and quantification standards provided with the kit. The kit quantified a 108-bp fragment from the 5S rRNA gene of Legionella spp. The other three methods quantified fragments of 386, 454, and 654 bp, respectively from the 16S rRNA gene of Legionella spp. using SYBR Green.

The four sets of primers used to detect Legionella are presented in Table 1. The qPCR mixture and protocol were applied according to each reference. The optimal primers for PMA-qPCR analysis were selected while comparing the effect of PMA based on the amplicon size. The PCR amplification was carried out in an Icycler (Bio-Rad). The standard curve was generated by 10-fold dilution of purified DNA from L. pneumophila. The qPCR results were expressed as gene copies calculated by the standard curve. All samples were tested more than twice.

Comparison of PMA-qPCR results between water samples and positive controls was evaluated by non-parametric Mann-Whitney U test and Kruskal-Wallis test using SAS program (SAS Institute Inc., NC, USA). The statistically significant level was set to α = 0.05.

PMA was added to 500 μL aliquots of live and dead cells (72 °C, 15 min) in final concentrations of 0, 25, 50, 100, and 200 μM to determine the optimal PMA concentration. As shown in Figure 2, the treatment of the dead cells with 25, 50, 100, and 200 μM PMA concentrations reduced the average copy number to 3.0, 3.1, 4.0, and 4.2 log, respectively. The effect of PMA on dead cells was shown to increase with higher PMA concentrations. The copy number of PMA-treated live cells was reduced to 0.1, 0.1, 0.2, and 0.4 log in 25, 50, 100, and 200 μM PMA concentrations, respectively. Moreover, as the concentration of PMA increased, the copy number of live cells decreased compared to untreated live cells. The biggest difference in copy number between PMA-treated live and dead cells was observed in 100 μM and 200 μM PMA concentrations. However, an effective reduction in dead cells, while minimizing the reduction of PMA-treated live cells, was observed in the 100 μM concentration.

When using four sets of primers to test the PMA effect with regard to DNA amplicon size (108, 386, 454, and 654 bp), the difference in copy number between the PMA-treated live cells and dead cells was 0.8, 3.5, 3.6, and 4.0 log in the 100 μM PMA concentration (Figure 3). Even at different PMA concentrations of 50, 200, and 400 μM, the longer the target gene was, the greater the difference in copy number between the PMA-treated live cells and dead cells (Figure 3). When the primer with different target sizes of 108, 386, 454, and 654 bp was used, however, the resulting PCR efficiency was found to be 99.3 ± 3.2%, 97.1 ± 3.6%, 89.9 ± 2.1%, and 88.8 ± 1.7%, respectively. Amplification of fragments longer than 454 bp reduced qPCR efficiency. Thus, the primer yielding a 386 bp PCR product was selected considering the PCR amplification degree of PMA-treated dead cells and PCR efficiency.

The applicability of the PMA-qPCR method was evaluated using the selected PMA concentration (100 μM) and primer (386 bp PCR product) to detect viable Legionella in environmental water such as the Han River water supply intakes, purified water, and tap water. The turbidity of five Han River samples was 1.4, 4.2, 8.1, 9.5, and 15.4 NTU. The turbidity of two purified water samples was in the range of 0.04–0.05 NTU, and that of three tap water sample was in the range of 0.07–0.08 NTU (Table 2). Live and dead cells of L. pneumophila were added for each sample. In the purified water and tap water samples, dead cells were completely excluded by PMA, and there was no difference in the detection concentration of live cells from positive control (p > 0.05) (Figure 4). For the samples, dead cells were effectively excluded in four samples with turbidity levels of 1.4, 4.2, 8.1, and 9.5 NTU. In the one sample (river 5) with turbidity of 15.4 NTU, however, dead cells were not effectively excluded, with the copy number of live cells decreasing compared to the control (p < 0.05) (Figure 4). Therefore, PMA-qPCR could be applied to water samples with turbidity levels of less than 10 NTU.

Several studies have shown that factors including the dye concentration, incubation time, light source, length of the target gene, and turbidity of samples may affect the qPCR results combined with viability dyes (Nocker et al. 2006; Soejima et al. 2008; Chang et al. 2010; Luo et al. 2010; Contreras et al. 2011). We considered factors of PMA concentration, length of the target gene, and turbidity of samples to apply the PMA-qPCR method to environmental water such as the Han River water supply intakes, purified water, and tap water.

Our results indicated that the cytotoxic effect on live cells and PCR suppression on dead cells were dependent on PMA concentration, and that the 100 μM concentration was the most effective in differentiating live and dead cells while minimizing the cytotoxic effect on live cells. Yáñez et al. (2011) also reported that the numbers of live L. pneumophila cells by 100 μM PMA treatment were slightly lower (0.05–0.16 log) than culture, and the risk of underestimating the number of live cells of PMA was lower than that of EMA.

The degree of suppression on dead cells by PMA-qPCR is dependent on the amplicon length. For various PCR amplicon lengths of 108, 386, 454, and 654 bp, longer amplicons appeared to lessen the amplification of dead cells. Previous studies also reported that longer PCR products appeared to exclude dead cells better because the longer PCR products, the higher probability of PMA binding to the DNA of dead cells (Soejima et al. 2008; Chang et al. 2010; Luo et al. 2010; Contreras et al. 2011). However, amplification of long amplicon size (>400 bp) reduces PCR efficiency by increasing the likelihood of formation of secondary structures (Desneux et al. 2015). Therefore, we selected the primer yielding a 386 bp PCR product considering the degree of amplification of PMA-treated dead cells and PCR efficiency.

In applying PMA to various environmental waters, the turbidity of the samples may effect the PMA-qPCR results. We observed that the PMA-qPCR method was able to distinguish between live and dead cells when the turbidity was less than 10 NTU. However, PMA treatment was not effective in a sample with high turbidity (>10 NTU). A sample with high turbidity has a lot of suspended solids and might affect the efficiency of PMA treatment by decreasing the transmission of light (Yuan et al. 2018). Our result was similar to another study in which PMA-qPCR could be applied to a water sample of 8 NTU (Truchado et al. 2016). Luo et al. (2010) also stated that high turbidity (>10 NTU) could reduce effectiveness of PMA treatment by interfering with PMA penetration or photoactivation. However, some studies showed that PMA-qPCR could still be applied at high turbidity levels (28.4–67.0 NTU, <120 NTU) (Yuan et al. 2018; Fu et al. 2020). These results may be because a small amount of sample (10 mL) was used and a PMA concentration twice higher than usual was applied (Yuan et al. 2018). The effect of high turbidity can be reduced by dilution of the sample, but when the concentration of the target microorganism is low, false negatives can occur (Gedalanga & Olson 2009). Therefore, the dilution of the sample and application of the PMA-qPCR method should be decided according to the turbidity of the sample and concentration level of the target microorganism. Further studies are needed to improve the efficiency of PMA treatment without loss of target microorganisms on turbid environmental water.

The PMA concentration, target gene size, and turbidity of environmental waters were major factors affecting the PMA-qPCR method. Therefore, it is necessary to select the PMA concentration and amplicon size that can exclude dead cells without affecting live cells. In addition, since cell shading may occur in environmental waters with high turbidity, the applicability of the samples should be checked. Our results showed that PMA-qPCR could be applied to environmental waters with turbidity levels less than 10 NTU for detecting viable Legionella. However, our study was conducted on a limited number of 10 samples, and only one sample with a high turbidity (>10 NTU) level. It is necessary to review the applicability of PMA-qPCR methods to more samples.

Using the PMA-qPCR method, the rapid detection and direct quantification of viable Legionella in environmental waters such as rivers as sources of tap water, water treated by purification plants, and tap water are expected to enable the hygienic management of drinking water, reducing health risk from pathogenic microorganisms.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.