Rapid detection of viable Legionella pneumophila in tap water by a qPCR and RT‐PCR‐based method

A molecular method for a rapid detection of viable Legionella pneumophila of all serogroups in tap water samples was developed as an alternative to the reference method (ISO). Legionellae are responsible for Legionnaires’ disease, a severe pneumonia in humans with high lethality.


Introduction
Legionellosis is a bacterial disease with two clinical manifestations in humans: the self-limiting and nonpneumonic Pontiac fever and Legionnaires' disease, characterized by severe respiratory symptoms including pneumonia with high lethality. The infectious agent is Legionella spp. and predominantly Legionella pneumophila (WHO 2007;Dominguez et al. 2009;Parr et al. 2015;ECDC 2016). Legionellae naturally occur in environmental water sources and are well adapted to man-made water installations (WHO 2007). They grow in warm water of 25-45°C and are often found in buildings with convoluted water pipelines that are stagnant or rarely flushed parts of the system (Rhoads et al. 2015). Predilection sites are shower heads, but also pipes, taps, Jacuzzi tubs, and air conditioning installations (WHO 2007).
Infection follows inhalation of contaminated water aerosols. Elderly or immunocompromised people, smokers and patients suffering from chronic respiratory illnesses are particularly at risk (Lanternier et al. 2017). The best preventive method is to adjust temperatures in cold water to below 20°C and to ensure a minimum water temperature of 60°C in boilers and preferably of 55°C in distribution pipes (WHO 2007;Bedard et al. 2015;Rhoads et al. 2015). To prevent scalding, temperature regulator devices should be installed at points-of-use. Further aspects of prevention are appropriate constructions with minimized water stagnation and materials that do not support microbial growth (Bedard et al. 2015).
For preventive purposes, microbiological criteria for Legionella spp. in public accessible bath and shower water were recently stipulated in Switzerland (FSVO 2017). To test water samples for compliance with these microbiological criteria, an international standard method is specified (ISO 11731) which detects and enumerates living cells of the pathogen by culturing on selective agar plates (ISO 2008). Alternative methods are allowed, when the result's evaluation does not influence the resulting decision. The ISO method is time consuming (up to 10 days) and also has some other disadvantages (Kirschner 2016). For example, legionellae can persist in a viable but not culturable (VBNC) status due to different reasons (Al-Bana et al. 2014;Li et al. 2015;Kirschner 2016), which may result in false negative results. Under certain conditions, VBNC legionellae are able to regain virulence and infectiousness (Steinert et al. 1997;Ducret et al. 2014). Molecular-based methods are faster but they usually detect both, viable and dead bacteria. The presence or number of dead cells might be of interest in some situations but for the official control of water samples for legionellae, methods are needed which include only living cells. Methods using DNA intercalating reagents can distinguish between living and dead bacterial cells but they have disadvantages. For example, the presence of biofilms can disturb detection (Taylor et al. 2014) and concentration dependent cytotoxic effects of reagents are possible (Yanez et al. 2011). Moreover, high cell counts, which usually do not occur in field samples, are required for plausible analyses (Chang et al. 2009).
Hence, as an alternative to the ISO method, a molecular detection method of viable Leg. pneumophila was developed, based on PCR detection of a precursor 16S rRNA target that is specific for this species and occurs only in living cells. Precursor rRNA represents a significant fraction of the total microbial rRNA and is, because of its higher stability, much easier to detect and handle than mRNA (Cangelosi et al. 2010). Precursor rRNA is synthesized by growing bacteria and its leader and tail sequences are subsequently removed during rRNA maturation (Cangelosi and Brabant 1997;Cangelosi et al. 2010). Upon stagnancy of growth, precursor rRNA synthesis stops but maturation goes on, draining the precursor rRNA pool. Thus, the presence of precursor rRNA can be used as a molecular indicator for physiological activity and therefore viability of bacterial cells (Stroot and Oerther 2003;Lu et al. 2009;Cangelosi et al. 2010).
As water is a limited nutritional medium for legionellae, cell populations are barely in an exponential growth phase and the precursor rRNA pool at a low level or even drained off (Al-Bana et al. 2014). The transfer of such starved bacteria into a fresh nutritional medium will stimulate them, resulting in a boosted rRNA synthesis. Dead cell is neither activated nor is their rRNA synthesis boosted. Such a stimulation step was included in the newly developed method. It is followed by nucleic acid (NA) extraction and reverse transcription-PCR (RT-PCR) to detect the precursor 16S rRNA. A shift between the cycle threshold (C T ) of an unstimulated and a stimulated sample can be interpreted as the presence of viable Leg. pneumophila cells. For quantification, a real-time quantitative PCR (qPCR) was performed using the DNA fractions that were simultaneously extracted from the same samples.
The aim of the study was to develop and evaluate a rapid method to detect viable Leg. pneumophila in tap water samples as an alternative to the bacteriological reference method (ISO 11731). The two methods were compared by analysing water samples from public sports facilities in a region of Switzerland with rather high incidence rates of legionellosis, that is, 7Á3 and 4Á6 infections per 10 5 inhabitants in 2015 and 2016 respectively (FOPH 2018).

Materials and methods
Bacteriological detection of Leg. pneumophila Water samples were analyzed with the reference method ISO 11731-2:2008(ISO 2008. In brief, 1 l of water was filtered through a 0Á2 lm polycarbonate Nuclepore TM membrane (Whatman Inc., Florham Park, NJ) and the bacterial cells then resuspended from the membrane. This suspension was plated on agar plates either directly or after acid or heat pretreatment to minimize accompanying bacterial flora. Plates were evaluated three times during the 10 days incubation period. One presumptive Legionella spp. per plate was confirmed by latex agglutination test (Oxoid, Pratteln, Switzerland and Microgen Bioproducts, Camberley, UK) which identify the predominant Legionella spp. and allow distinction between Leg. pneumophila SG 1 and SGs 2-15.
Molecular detection of Leg. pneumophila A PCR system was designed that detects the precursor region of the 16S rRNA sequence of Leg. pneumophila which occurs threefold within its genome. Three copies of the target were therefore regarded as one cell equivalent (cellEq). The forward primer (5 0 -CGA GAG CTA GTG CCG GAA T-3 0 ) in this system was located within the precursor region of the 16S rRNA sequence, while probe (5 0 -FAM-TAG ACA GAT GGC GAG TGG CGA ACG-BHQ1-3 0 ) and reverse primer (5 0 -CCA AGT TGT CCC CCT CTT C-3 0 ) were located downstream. The amplicon size was 177 bp. Primers and probes were synthesized by Microsynth (Balgach, Switzerland). In silico sequence homology searches using the nucleotide BLAST tool (https://blast.ncbi.nlm.nih.gov) showed that primers and probes were specific for Leg. pneumophila.

Analytical specificity
To determine inclusivity and exclusivity, the following 21 Legionella spp. as well as 12 non-Legionella strains which can contaminate water, were tested for the presence of the primer's and probe's target sequence: Leg. pneumophila subsp. Legionella spp. were grown on buffered charcoal yeast extract (BCYE) agar with L-cysteine (BioM erieux Suisse S.A., Geneva, Switzerland) at 37°C for a minimum of 48 h. Non-Legionella strains were grown on 5% sheep blood agar plates (BioM erieux) at 37°C for 24 h. Total bacterial DNA was extracted by the boiling preparation method using one colony per strain, re-suspended in 100 ll TE Buffer (10 mmol l À1 Tris-HCl, 1 mmol l À1 EDTA, pH 8Á0) and incubated at 95°C for 10 min. DNA extracts were diluted 1 : 100 with PCR grade water prior to qualitatively applied qPCR (see below).

Detection of living Leg. pneumophila with RNA stimulation
To detect living cells of Leg. pneumophila, an assay was developed which was based on the stimulation of RNA synthesis (Stimulation Based Methodology, SBM) and comprises the following four steps.

Target concentration
A 1 l water sample was filtered through a 0Á2 lm polycarbonate membrane filter of 47 mm diameter (Sterlitech, Kent, WA) using an appropriate filtration unit and vacuum pump. The membrane was transferred to a 50 ml tube, rinsed with 3 ml of sterile Page's saline containing 120 mg l À1 NaCl, 4 mg l À1 MgSO 4 9 7H 2 O, 4 mg l À1 CaCl 2 9 2H 2 O, 142 mg l À1 Na 2 HPO 4 and 136 mg l À1 KH 2 PO 4 (according to the ISO method), and then vigorously vortexed for 2 min. As process controls, bottled still mineral water of the brand 'Evian' (Evian-Volvic Suisse S.A., Zurich, Switzerland) was used as a negative control, and freshly sampled warm tap water from a contaminated but not sanitized building was used as a positive control.

Stimulation
The concentrates of the previous step were then used to prepare a unstimulated control (ÀSTIM) and a stimulated sample (+STIM): For ÀSTIMs, a 1 ml aliquot of the concentrate was transferred in a 1Á5 ml tube, centrifuged at 16 000 g, 4°C for 5 min, and immediately frozen at À70°C after discarding the supernatant. To prepare +STIMs, another 1 ml aliquot of the concentrate was added to 9 ml sterile filtered and prewarmed stimulation medium, containing 10 g l À1 yeast extract (Oxoid) and 10% Legionella BCYE Growth Supplement (Oxoid) in a 50 ml tube. Stimulation occurred during incubation at 37°C, 130 rev min À1 for 3 h. After the subsequent centrifugation at 10 000 g, 4°C for 10 min, 9 ml of supernatant was discarded and the residual transferred to a 1Á5 ml tube. Following centrifugation at 16 000 g, 4°C for 5 min and removal of the supernatant, the pellet was frozen at À70°C until NA extraction.

NA extraction
Simultaneous extraction of total DNA and RNA fractions from ÀSTIM and +STIM samples was performed using the NucleoSpin RNA Kit complemented by the NucleoSpin RNA/DNA Buffer Set (Macherey-Nagel AG, Oensingen, Switzerland) according to the manufacturer's instructions. An on-column DNase treatment was included during RNA extraction. The final elution volume was 100 ll for the DNA fraction and 60 ll for the RNA fraction. As an extraction control, an aliquot of pelleted and frozen Leg. pneumophila were always coprocessed. RNA fractions were analyzed immediately after extraction and afterwards stored at À70°C. DNA fractions were stored at 4°C until analysis and then stored at À20°C.

Molecular detection
DNA fractions were analyzed by qPCR for quantification. RNA fractions were tested by one-step RT-PCR (as outlined below) to evaluate C T values of ÀSTIM and +STIM. A shift of >1 cycle (C T shift) between ÀSTIM and +STIM was considered as a stimulation caused by living cells (i.e. stimulated) of Leg. pneumophila in the sample. Shifts <1 cycles may occur due to technical variation in PCR replica. In additional stimulation experiments using heat inactivated cells, C T shifts <1 cycle or even negative values were observed (data not shown).

QPCR and RT-PCR conditions
DNA samples of bacterial strains and DNA fractions obtained from the SBM were analyzed by qPCR. Reactions were run in a total volume of 25 ll containing 19 Roche LightCycler 480 Probes Master (Roche Diagnostics, Rotkreuz, Switzerland), 400 nmol l À1 of both primers, 200 nmol l À1 of the probe (see above), and 5 ll of DNA template. QPCRs were run on a LightCycler 480 II (Roche) thermocycler. The run protocol started with an initial step of 95°C for 10 min, followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. For quantification, a standard curve of the 16S rRNA precursor region ranging from 10 5 to 10 1 copies per ll was included in each run. Standards were prepared using DNA of Leg. pneumophila (DSM 7513): PCR was done using the above-mentioned conditions and the PCR-products were subsequently purified using the NucleoSpin Gel and PCR Clean-up Kit (Macherey-Nagel). Concentration of the purified PCR-product was determined by the NanoDrop 1000 spectrophotometer (Nanodrop Technologies, Rockland, Germany) followed by dilutions in 0Á29 TE buffer (pH 8.0) to concentrations from 10 5 to 10 1 copies per ll. For data analysis, the second derivative maximum analysis method of the LightCycler 480 Software (release 1.5.1.62) was used. It automatically displayed the standard curve as a linear regression line for higher accuracy of data at the detection limit.
RNA fractions obtained from the SBM were analyzed by RT-PCR immediately after RNA extraction. Reactions were run in a total volume of 25 ll using the QuantiTect Probe RT-PCR Kit (Qiagen, Hombrechtikon, Switzerland) with same primers, probes and their concentration as for qPCR (see above), and 5 ll of RNA template. The same cycler as for QPCR was used with the following protocol: an initial reverse transcription at 50°C for 30 min, and then an initial heat activation at 95°C for 15 min, followed by 45 cycles of 94°C for 15 s and 60°C for 1 min.
For all PCR runs, samples were run in duplicates and a no template control and a positive control were always included. Results were considered positive if both reactions were positive. If only one reaction showed a positive result or the difference of the two C T values was more than two cycles, the QPCR was repeated. C T values >40 were omitted.

SBM performance
Assay performance was done according the World Organization for Animal Health (OIE 2013).
The limit of detection (LOD) was determined using artificially starved Leg. pneumophila: One colony of Leg. pneumophila (DSM 7513) was resuspended in BCYE broth supernatant, that was obtained by centrifugation of BCYE broth at 10 000 g for 10 min and supplemented with 10% Legionella BCYE Growth Supplement (Oxoid). The culture was incubated at 37°C overnight and the cell count then determined by flow cytometry (Cyflow ML, Sysmex Partec, Horgen, Switzerland). Subsequently, an aliquot containing 10 8 cells was centrifuged at 16 000 g for 5 min. The pellet was resuspended in 100 ml of sterile filtered Evian water, generating a concentration of 10 6 cells per ml. Starvation of Leg. pneumophila occurred during incubation at 37°C, 100 rev min À1 for 24 h and lead to drain-out of present precursor 16S rRNA. To find out the LOD, 10-fold serial dilutions were made from the starved culture in sterile filtered Evian water obtaining concentrations of 10 5 , 10 4 , 10 3 , 10 2 and 10 1 cells per ml. From each of these dilutions, 1 ml was taken to spike 1 l of Evian water which was subsequently analyzed with the SBM, as written above. As the SBM method produces two results (a quantification by qPCR and a viability result from RT-PCR), the LOD on the one hand was evaluated with respect to the combined result (including qPCR and RT-PCR) and on the other hand with respect to the quantification by qPCR solely.
The intraassay variability of the complete SBM procedure was determined by simultaneous processing of eight identical samples of tap water (each 1 l) that was naturally contaminated with Leg. pneumophila. The interassay variability of the complete SBM procedure was evaluated by processing of eight samples on eight different days. Each day, 1 l Evian water was spiked with 10 8 cells of artificially starved Leg. pneumophila, which were prepared as written above and kept at 37°C during the experiment.
Additionally, a stimulation experiment using artificially starved Leg. pneumophila was performed for providing a time course over a 4 h stimulation period. Samples were collected after 20, 40, 60, 80, 100, 120, 180 and 240 min of stimulation. The samples underwent the NA extraction as written above and the obtained DNA and RNA fractions were used for qPCR and RT-PCR respectively.

Application of the developed method on field samples
To compare the developed molecular method with the standard method of ISO, 102 tap water samples of 2 l were taken in 51 public sports facilities in the Canton of Basel-Landschaft, Switzerland, between September and November 2016. In each facility, one sample was taken preferably close to the boiler, another one in the shower, after discarding the first litre of water. One sample was spilled during transport. The samples were divided into two 1-l aliquots in the laboratory. One aliquot was bacteriologically examined the same day in the microbiological laboratory of the Food Safety and Veterinary Office of Canton Basel-Landschaft according to the ISO method. The other aliquot was kept at 4°C, transported to the Federal Food Safety and Veterinary Office the same day, and examined the following day using the SBM procedure. Results of both laboratories were only compared after completing all measurements which allowed a blinded assessment (Greenhalgh 1997).
Comparison of data included quantitative comparison of measured counts (CFU per l for the ISO method and cellEq per l for the SBM) and qualitative comparison (viable Leg. pneumophila present/absent) with regard to official microbiological criteria for Leg. pneumophila, that are a limit of 1000 CFU per l (MC 1000 ) for water from showers and a limit of 100 CFU per l (MC 100 ) for aerosol-producing installations, for example Jacuzzi tubs (FSVO 2017).

Statistics
The exact McNemar's test was applied using the Statistical Analysis System software (SAS 9.2 analytics software, Cary, NC, USA) to compare the paired nominal data of both methods regarding both criteria, MC 100 and MC 1000 respectively. In addition, the Cohen's Kappa coefficient (j) was calculated as a measure of agreement.

Assay performance
Analytical specificity Inclusivity and exclusivity of the PCR system were both 100%. All 15 strains of Leg. pneumophila belonging to SGs 1-14 tested positive for the precursor 16S rRNA coding sequence. All other tested Legionella spp. (n = 6) and all bacterial strains of other genera (n = 12) tested negative for this target. Using purified precursor 16S rRNA targets, the qPCR showed log-linear results over eight dilutions of a 10-fold dilution series. PCR performance show a high efficiency of 1Á93 and a slope of À3Á502.

Analytical sensitivity
The LOD of the SBM was determined using water spiked with starved Leg. pneumophila serially diluted from 10 5 to 10 1 cells per l. As the SBM comprises two PCRs (qPCR and RT-PCR) a LOD for the combined result and for qPCR alone was evaluated separately. A positive result for both, qPCR (quantification) and RT-PCR (i.e. C T shift between ÀSTIM and +STIM representing viability), was observed for the three highest concentrations, namely 10 5 , 10 4 , 10 3 cells per l. Therefore, LOD was 1000 cells per l regarding viability. QPCR solely gave a positive result additionally for 10 2 cells per l, but at this concentration, no C T shift was observed in the RT-PCR, as only one of the two duplicates was positive or +STIM was negative. Hence, regarding qPCR that was used for quantification, the LOD was 100 cells per l. Additionally, qPCR and RT-PCR were repeated on a Rotor-Gene Q real-time thermal cycler (Qiagen) under the very same conditions, resulting in a LOD of 100 cells per l for both qPCR as well as RT-PCR (see Discussion section).

Repeatability
The intraassay variability of the SBM was determined by a simultaneous processing of eight water samples, naturally contaminated with Leg. pneumophila. All samples tested positive for viable Leg. pneumophila with cell counts from 5880 to 8455 cellEq per l. The CV was 1Á61% (log 10 transformed values). C T shifts between ÀSTIM and +STIM ranged from 2Á0 to 5Á8 cycles. The interassay variability of the SBM was evaluated by repeated processing of eight water samples, spiked with 10 8 cellEq per l of artificially starved Leg. pneumophila, on eight different days. All samples tested positive for viable Leg. pneumophila with cell counts from 0Á47 9 10 6 to 1Á07 9 10 6 cellEq per l. The CV was 1Á83% (log 10 transformed values). C T shifts between ÀSTIM and +STIM ranged from 3Á4 to 8Á5 cycles.

Stimulation time course
In a 4 h-stimulation experiment with periodical samplings after 20, 40, 60, 80, 100, 120, 180 and 240 min, RT-PCR showed decreasing C T values over the whole experiment period. The corresponding C T shifts, calculated in reference to the C T value before stimulation (=0 min), were 8.3, 10.0, 11.7, 12.2, 12.9, 13.5, 13.9 and 14.1 cycles respectively ( Table 2). The simultaneously extracted DNA showed constant C T values over time with <1 cycle of variation (Table 2).

Field study
Tap water samples (n = 101) from public sports facilities were analyzed, using both the bacteriological reference method according to ISO and the SBM. Results are shown in Table 1.
Bacteriological analysis (ISO method) revealed a positive result in 21 (21%) samples for Leg. pneumophila, ranging from 100 to 87 000 CFU per l (mean = 8567 CFU per l, median = 1200 CFU per l). Regarding serogroups, five out of the 21 were Leg. pneumophila SG 1, 16 out of 21 were Leg. pneumophila SGs 2-15. Evaluation of results under consideration of official microbiological criteria (100 and 1000 CFU per l) showed that 10 of the 21 positive water samples were between MC 100 and MC 1000 and 11 were above MC 1000 . Eighty samples (79%) were negative for Leg. pneumophila. Beside this, Legionella spp. was found in four samples, one of them in combination with Leg. pneumophila SGs 2-15 (Table 1).

Data comparison of both methods
For comparison of both methods, positive findings of the SBM (viable cells present) and positive results by the ISO method were compared in order to find out whether the new molecular approach was equivalent to the bacteriological standard method used for official testing of water samples for compliance with microbiological criteria. In the comparison, samples with nonviable Legionella cells in the SBM and samples with Legionella spp. by the ISO method were excluded.
With respect to MC 100 , 87 samples (86%) gave consistent results (73 negative and 14 positive results). Seven samples of each (7%) did not match and were either positive or negative by one or the other method. The seven samples which did not conform to the results revealed by the ISO method showed 100 to 800 CFU per l, but no viable Leg. pneumophila by the SBM. On the other hand, seven other samples that were negative by the ISO method, showed 145-2055 cellEq per l including viable ones tested by the SBM.
For MC 1000 , the concordance was even higher (96%), with 87 and 10 results consistently negative and positive respectively. Only one out of 11 samples exceeded MC 1000 when tested by the ISO method (1200 CFU per l) but not by the SBM. On the other hand, three samples were negative or under MC 1000 by the ISO method, but did not comply when analyzed by SBM showing 1231, 1965 and 2055 cellEq per l. *SG = serogroup of Leg. pneumophila, determined by latex agglutination test. †nv = Leg. pneumophila were not viable ‡MC 100 = microbiological criterion at a limit of 100 CFU per l and 100 cellEq per l, respectively; MC 1000 = microbiological criterion at a limit of 1000 CFU per l and 1000 cellEq per l, respectively; + = results of ISO method and SBM do match; -= results of ISO method and SBM do not match. §One sample positive for Legionella spp. other than Leg. pneumophila. ¶Two samples positive for Legionella spp. other than Leg. pneumophila.
Diagnostic specificity and sensitivity of the SBM at MC 100 were 91 and 67%, respectively, at MC 1000 they were 97 and 91%, respectively, taking the ISO method as the reference and gold standard. Statistically, the two methods did not differ significantly in the frequency of not matching results for both criteria, for MC 100 (P = 1.000) and for MC 1000 (P = 0.625) respectively. Moreover, the agreement of both methods are substantial for MC 100 (j = 0.579) and almost perfect for MC 1000 (j = 0.811) respectively.

Application range of the compared test systems
The developed SBM detects Leg. pneumophila but no other species of the Legionella genus. However, Leg. pneumophila is the most important species to cause legionellosis and is responsible for 90-100% of the reported cases (WHO 2007). The SGs 1-14 of Leg. pneumophila all tested positive, which is an advantage of the SBM. For the lacking SGs 15 and 16, that were not tested in our study, the assay is expected to work also, as the DNA relatedness within one species of Legionella is very high (WHO 2007). Legionella spp. other than Leg. pneumophila that are not detected by the SBM, are by far less frequently isolated from patients (5-10% of cases) and are predominantly involved in nosocomial infections (WHO 2007). For these species, a PCR system for the precursor 16S rRNA region could easily be developed and implemented in the SBM.
For official control of water samples for Leg. pneumophila, the ISO method serves as the reference method. This procedure might not highlight the true status of a water sample (Greenhalgh 1997). For example, VBNC legionellae are known to occur and could lead to false negative results (Kirschner 2016). Furthermore, false positive results could happen as shown in our analysis of field samples, where a presumptive Legionella colony was confirmed as Leg. pneumophila SGs 2-15 by the latex test, but turned out to be Acinetobacter tjernbergiae by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS). This false positive result could have been avoided earlier using a blood agar plate. Further cross-reactions with other bacteria have been reported for latex tests (Fields et al. 2002). In addition, false negative results are possible because some Legionella spp. cannot be cultivated as easily as Leg. pneumophila on agar plates used in the ISO method, as they seem to be more sensitive to the acid or heat treatment in the ISO procedure.

Methodology
For quantification, the ÀSTIM was used. Theoretically, ÀSTIM and +STIM should both reveal the same results.
No relevant increase in cell counts occur during the stimulation time that is shorter than the doubling time of Leg. pneumophila that is 6 h (Warren and Miller 1979) or more (Mauchline et al. 1994;Mampel et al. 2006). In reality, slightly lower counts for the +STIM were often seen, probably due to the additional centrifugation step for +STIM and sample transfer to another tube, leading to a loss of material. For these reasons, the ÀSTIMs were more suitable for quantification.
In multiple RNA extractions from stimulation experiments, it was shown that contaminating DNA within the RNA fraction was on a low level. Moreover, the same amount of residual background DNA is expected in both ÀSTIM and +STIM, so that it does not influence the possible C T shift that is based on the RNA. Furthermore 10-fold dilutions of water spiked with starved Leg. pneumophila ranging from 4 9 10 4 to 4 9 10 1 cells per l were processed by SBM to test C T shift at varying levels of bacteria concentration. Results showed comparable C T shifts of 2.3, 4.0, 2.2 and 3.0, respectively, over this wide range of dilutions.
The broth used for stimulation was a yeast extract enriched with commercial growth supplement for legionellae. Basically, this is BCYE broth without activated charcoal, whose benefit is to bind toxic free radicals generated under light exposure or by autoclaving. We omitted the addition of activated charcoal, as we sterile filtered the medium instead of autoclaving and incubation took place in the dark. Moreover, this approach dispels concerns of clogging spin columns by charcoal during NA extraction.
A stimulation time of 3 h seemed to be reasonable since under laboratory conditions, as starved cells already showed C T shifts after less than 1 h of stimulation ( Table 2). As it was not known, how naturally grown legionellae behave and induce their rRNA synthesis under stimulation, the duration was chosen as long as possible to measure the maximum effect, but shorter than the doubling time of legionellae.

Pro and cons of methods
The advantages of the SBM clearly outweigh the disadvantages. Advantages include the speed of the SBM, allowing results to be obtained within 8 h upon arrival of water samples in the laboratory, compared to 1 week or more for the ISO method. In the context of clinical cases or outbreaks, rapid testing of water as a potential source of infection is especially a huge advantage. Also during sanitation of water supply networks in affected buildings, a short test time to monitor failure or success of the applied measures is required. The economic consequences of closed facilities, such as public baths, are directly linked to the duration of closure until sanitation is carried out.
In contrast to common molecular methods, the SBM is able to distinguish between viable and nonviable cells. This is based on induction and detection of precursor 16S rRNA indicating viability (Cangelosi and Brabant 1997;Weigel et al. 2013), as leader and tail sequences are removed immediately after synthesis. Besides RT-PCR, qPCR was done for quantification of DNA which was simultaneously extracted with RNA. On possible disadvantage of the SBM might be that total DNA is quantified by this procedure, and not only DNA of viable Leg. pneumophila, leading to overestimation of cell counts. Therefore, this method alone cannot be applied to test water samples for compliance of microbiological criteria as these criteria include only living cells. High DNA amounts can still be present after sanitation of water supply networks, although the number of viable legionellae is low. However, if no viable legionellae are detected, eventual present DNA is out of concern and interest. Therefore, it is important to interpret quantification data always in combination with the RT-PCR results. In contrast, the ISO method probably underestimates the legionellae count in water, as VBNC bacteria are not readily cultivable.
Furthermore, the SBM implicates a higher manual workload and costs compared to the ISO method. The filtration step is comprised in both methods, but the SBM additionally comprises a NA extraction and PCR detection. NA extraction is especially a labour-intensive step but it could be automatized for a higher throughput. In addition, NA extraction is relatively expensive and linked to appropriate equipment.
In reference to the ISO method, specificity and sensitivity of the SBM were quite high, especially for the MC 1000 with values of 97 and 91% respectively. For MC 100 they were 91 and 67% respectively. Both methods have advantages and disadvantages but they might be used for different occasions: environment assessment in the context of clinical cases or outbreaks, water analysis by the SBM has clear advantages as mentioned above, compared to the ISO method that is more preferable for screening purposes.

Improvements and further research
The performance of the developed method is satisfactory but there is still potential for improvement. Among others, the applied simultaneous extraction of DNA and RNA fractions is reasonable and was chosen for better comparison, but this procedure lead to a higher loss of both fractions, compared to separate extraction of DNA and RNA from two aliquots of an identical sample (data not shown). Simultaneous vs separated extraction, as well as its costs, should be evaluated prior to future studies or applications of the method. Furthermore, the use of the Rotor-Gene PCR cycler instead of LightCycler not only showed lower C T values in general for the very same sample, but also showed less variation between duplicates and larger C T shifts in RT-PCRs. Moreover, the LOD was one log 10 lower (100 cells per l) using the Rotor-Gene for the very same samples and PCR conditions compared to the LOD reveald by the LightCycler (1000 cells per l) with regard to the combined result (qPCR plus RT-PCR).
The VBNC status that legionellae achieve in water is a poorly characterized condition (Al-Bana et al. 2014). This study did not answer the question, whether VBNC legionellae can be stimulated or not. Further investigations with confirmed VBNC Legionella cells are required to better understand RNA synthesis' stimulation compared to growth. In fact, 7% of the study samples were positive for viable Leg. pneumophila by the SBM but did not show growth on agar plates. They might have been able to activate their RNA synthesis but were not able to grow on plates. Otherwise, cell counts by the SBM were not considerably higher compared to the ISO method, even though DNA of dead bacteria was included.
In conclusion, the method is a useful and reliable tool for a rapid detection of viable Leg. pneumophila of all serogroups in tap water samples with high diagnostic sensitivity and specificity and can be used as an alternative to the ISO method for the official water analysis for Leg. pneumophila and particularly when a short test time is required. †Precursor 16S rRNA was detected by reverse transcription-PCR. C T shifts in reference to 0 min (before stimulation) were used as an indicator for viability.