Introduction

Legionella are pathogenic bacteria that can cause infections after inhalation of legionella-containing bioaerosols [1]. Word-wide legionellosis outbreaks have occurred consistently during recent years; aerosolized Legionella pneumophila from evaporator cooling systems were mainly found to be the source [2,3]. The latest legionellosis outbreaks in Warstein (Germany), Vila Franca de Xira (Portugal), and New York (USA) have shown that process water for cooling systems can pose a risk for public health. Special care is needed when untreated surface water is used. Although low numbers of L. pneumophila are present in surface water [4–6], they can proliferate under special conditions like increased temperature and deposits of biofilm [7] such as those found in technical water systems [8]. In consequence, the water source as well as the circulating process water of technical water systems has to be analyzed continuously for an increase of Legionella ssp. and the presence of high-risk pathogens like L. pneumophila serogroup (SG) 1 that is responsible for 90 % of pneumonia identified as legionella infections. Therefore, rapid detection methods are needed that are able to quantify Legionella ssp. in complex water matrixes like surface water. The simultaneous identification of all subtypes of L. pneumophila SGs is important for a rapid and comprehensive quantitative microbial risk assessment. Therefore, antibody panels are necessary to perform a monoclonal subtyping of L. pneumophila which are used in former studies only in cultivated isolates [9,10]. In our test study the monoclonal antibody mAb 10/6, which is highly selective for L. pneumophila SG 1 subtype (ST) Bellingham, is applied the first time for water matrixes.

In order to establish a fast, simple, and sensitive detection method, a multistep process, combining rapid concentration and selective detection methods, is required. We have developed a new combined analytical process that uses monolithic adsorption filtration (MAF) and centrifugal ultrafiltration (CeUF) to concentrate living L. pneumophila cells and chemiluminescence sandwich microarray immunoassay (CL-SMIA) for highly sensitive detection in tap and surface water. As a first concentration step, MAF, which has already been used to concentrate viruses from large-volume water samples [11], was adapted for the concentration of bacteria in different water matrixes, like tap and surface water; a highly efficient concentration from 10 L to 20 mL was achieved. Owing to the mesoporous structure of the monolith with a pore size of 20 μm, high flow rates of up to 1 L/min are possible. Therefore, time for preconcentration was reduced drastically compared to established filtration methods. Since the pore size of 20 μm allows L. pneumophila to pass through, retention is based on ionic and hydrophobic interaction. Acidifying the samples to pH 3 enhances these interactions. Adsorption is reversed by elution in buffer containing 3 % beef extract at pH 9.5. Centrifugal ultrafiltration was used as the second step, to further concentrate samples to 1 mL and thereby enhance the sensitivity of the subsequent detection by achieving a total volumetric concentration factor of 104. In a former study heat-inactivated L. pneumophila were detected by flow-based CL-SMIA with a detection limit of 105 cells per milliliter [12]. In this work, the quantification was performed on the automated analysis platform MCR 3 [13,14]. By application of MAF-CeUF-SMIA, sample preparation is simplified, time for preconcentration is minimized, and the sensitivity is increased significantly.

All established analysis methods should comply with set threshold limits for legionella in tap water of 100 CFU per 100 mL, defined by several authorities like the German Drinking Water Ordinance and US Occupational Safety & Health Administration guidelines (Appendix III:7-3) [15,16]. Table 1 shows a comparison between the introduced combined analytical process and other established detection procedures for L. pneumophila in water. Cell cultivation in combination with membrane filtration is the gold standard for tap water analysis. It takes 48 h–10 days and has a limit of detection (LOD) of 1 CFU/mL. The cultivation and interpretation of plate counting are complex for environmental samples because of accompanying flora and the viable-but-not-culturable (VBNC) status or growth inside of protozoa. qPCR is a nucleic acid detection method that takes 3–8 h with a reported LOD for tap water of 640 genomic units (GU)/mL in combination with membrane filtration. However, there is a lack of correlation between qPCR and cultivation in samples of water systems [17]. For fluorescence in situ hybridization (FISH) an LOD of 35 CFU/mL has been shown. It uses membrane filtration to concentrate L. pneumophila in tap water and detects them within 1 day. An extensive sample pretreatment is needed which includes cell fixation, permeabilization, hybridization, and signal amplification. Simpler and easy-to-automate methods are needed if a screening of L. pneumophila in water samples is to be accepted. The combination of immunomagnetic separation (IMS) and flow cytometry can detect 0.015–0.05 CFU/mL within 2 h, but has only been shown for tap water [18,19]. Most polyclonal antibodies bind SGs of L. pneumophila with different effectivity, which decreases the sensitivity of immunoassays analyzing all L. pneumophila SGs [20]. Therefore, the usage of IMS in combination with immunoassays may overestimate L. pneumophila SG 1.

Table 1 Comparison of L. pneumophila detection methods
Full size table

With our combined analytical process a highly sensitive detection of L. pneumophila in surface water with an LOD of 0.4 CFU/mL was achieved. Furthermore, the combination of preconcentration by MAF-CeUF and detection by CL-SMIA, all within 90 min, is very fast and enables the discrimination between viable cells and L. pneumophila lipopolysaccharide (LPS). A comprehensive subtyping could be done in future by the use of flow-based microarrays on the analysis platform MCR 3.

Materials and methods

Chemicals and materials

All standard chemicals as well as casein, 2-amino-2-hydroxymethyl-propane-1,3-diol (TRIS), di-(N-succinimidyl)-carbonate (DSC), dimethylaminopyridine (DMAP), triethylamine, and anhydrous dimethylformamide (DMF) were of analytical grade and purchased from Sigma Aldrich (Taufkirchen, Germany). Glass slides (76 × 26 × 1 mm) were obtained from Roth (Karlsruhe, Germany). The chemiluminescence substrates, Westar Supernova ELISA luminol and peroxide solution, were purchased from Cyanogen (Bologna, Italy). All buffers were prepared with ultrapure water. PBS consisted of 145 mM NaCl, 10 mM KH2PO4, and 70 mM K2HPO4, adjusted to pH 7.6. Running buffer contained 0.5 % (w/v) casein dissolved in PBS. For blocking the microarray chips, 3 M TRIS buffer adjusted with HCl to pH 8.5 was used. The monoclonal antibodies in this study were purified and supported by Reference Laboratory Legionella at Technische Universität Dresden (Dresden, Germany). The purified and biotin-labeled polyclonal antibodies were purchased from Meridian Life Science (Memphis, TN, USA) via distributor Dunn Labortechnik GmbH (Asbach, Germany).

Preparation of L. pneumophila stock suspensions

L. pneumophila strains, belonging to the monoclonal subtype Bellingham, were grown on selective medium containing glycine, vancomycin, polymyxin, and cycloheximide (GVPC) as published elsewhere [21]. The GVPC agar plates were incubated for 7–10 days at a temperature of 37 °C. Bacterial stock suspensions of L. pneumophila were produced by harvesting colonies of each strain from the relative streaked GVPC agar plate and resuspending them in sterile ultrapure water. To estimate the cell concentration of each bacterial stock suspension, successive 10-fold serial dilutions of the starting population were made in ultrapure water. A sample (0.1 mL) was placed from each tube onto an agar plate and spread evenly over the surface of the plate. The plates were incubated for 7–10 days at 37 °C and forming colonies were counted. To obtain an accurate estimation of the amount of bacteria in each stock solution, plates with 30–300 colonies per petri dish were considered. The concentration of the bacterial stock solution was reported in CFU (colony forming units) per milliliter.

Preconcentration process

Production of monolithic adsorption filtration disks

The preparation of epoxy-based macroporous monoliths with a pore size of ca. 20 μm has been published elsewhere [22]. MAF disks have already been applied for preconcentration of viruses in large-volume water samples [11]. MAF modules were constructed in 50-mL plastic dispenser tips (PD-tip, Brand GmbH, Germany) that were used as housings. A PTFE support plate with bore holes of 2 mm in diameter, an O-ring (38.6 mm in diameter, nitrile butadiene rubber NBR 70), a MAF disk, and a PTFE fitting for connection to silicone tubes were placed consecutively in the MAF housing as shown in Fig. 1a). MAF disks were produced at 28 °C by self-polymerization of the monomer polyglycerol-3-glycidyl ether (Polypox R9, UPPC Mietingen, Germany). A 60:40 (v/v) mixture of toluene and tert-butyl methyl ether was added to the monomer in a 20:80 (v/v) monomer/porogen ratio to form pores in the monomer. A 1:10 (v/v) dilution of boron trifluoride diethyl etherate (BF3·Et2O) in 1,4-dioxane was used as catalyst and added to the mixture (1.25 %, v/v, with respect to total volume). The mixture was poured into PTFE molds to form MAF disks and was incubated at 28 °C for 50–60 min. MAF disks have a diameter of 38.6 mm and a height of 10.0 mm. The MAF disks were taken out of the PTFE molds and stored in methanol overnight to remove the porogen. The epoxy groups of the monolith were hydrolyzed by continuously pumping 0.5 M sulfuric acid through the MAF disks for 3 h at 60 °C. The hydrolyzed surface enables retention of microorganisms such as legionella via electrostatic interaction. The MAF disks were stored in ultrapure water at 4 °C until further use.

Fig. 1
figure 1

a Monolithic adsorption filtration setup with MAF and b whole setup for the filtration process

Full size image

MAF process

All water samples were adjusted to pH 3 with hydrochloric acid before starting the MAF concentration process. Ten-liter disposable bags (Rotilabo PE canister, Carl Roth, Germany) were used as containers for water samples and filtrate. The MAF module was assembled as described before and connected to a peristaltic pump (Pumpdrive 5206 Peristaltic Pump, Heidolph, Germany) by a silicone tube (Marprene Tube, Watson-Marlow, inner diameter 6.4 mm) as shown in Fig. 1b. The outlet of the plastic dispenser tip was cut to enlarge the inner diameter to ca. 4 mm. The MAF module outlet was placed directly into the orifice of the disposable 10-L bag. Before sample filtration, MAF modules were flushed (1 L/min) with 1 L of acidified ultrapure water (pH 3) in order to equilibrate the macroporous surface of the monolith. Afterwards, the sample tube was placed in the sample container and the MAF process was started with a flow rate of 1 L/min. After the sample had passed through the MAF module, it was flushed with air and afterwards filled with 20 mL of elution buffer. The elution buffer consisted of 0.5 M glycine and 3 % beef extract powder, and it was adjusted to pH 9.5 with sodium hydroxide solution (≥32 %, w/w). The buffer was pumped through the module with a flow rate of 150 mL/min for elution.

Centrifugal ultrafiltration (CeUF)

CeUF devices (50 kDa cutoff, Amicon Ultra 50 centrifugal filters, Merck Millipore, Darmstadt, Germany) were used to reduce the eluate volume from 20 to 1 mL. MAF eluate was added and centrifuged at 4500 rpm with a swinging bucket rotor for approximately 20 min. The concentrated sample was recovered by vortexing the tube for 2 min and stored at 4 °C until further use.

Sandwich microarray immunoassay

Preparation of antibody microarray chips

Antibody microarrays were produced as described elsewhere [23,24]. Cleaned microscopy glass slides were step-by-step functionalized by coating with GOPTS (Sigma Aldrich, Germany), Jeffamine ED-2003 (Huntsman, USA), and DSC (Sigma, Aldrich, Germany). Capture antibodies were covalently immobilized on NHS-activated microscopy glass slides after deposition by contact printing. Antibody microarrays were generated using the contact spotter BioOdyssey Calligrapher MiniArrayer (Biorad, CA, USA) and one solid pin SNS 9 from ArrayIT (Sunnyvale, CA, USA). Each microarray slide consisted of two separate flow cells with a distance of 11.75 mm. Two clusters were set on one microarray slide with a grid spacing of 1300 μm for the columns and 1100 μm for the rows. Antibodies were dissolved in spotting buffer containing 20 % trehalose and 0.05 % Pluronic F127. The arrangement of the antibody microarray was as follows: 1. positive control (pAb anti-HRP, c = 100 μg/mL in spotting buffer, Sigma-Aldrich, Germany), 2. polyclonal anti-L. pneumophila (c = 2.5 mg/mL in spotting buffer, Meridian, USA), 3. monoclonal capture antibody mAb 10/6 for ST Bellingham (c = 0.74 mg/mL), 4. negative control (spotting buffer). Five replicates of each antibody solution were immobilized as spots on the microarray surface. The antibody microarrays were incubated overnight at 25 °C and a humidity of 50 %. Afterwards, the surface of the microarray chips was blocked for 15 min with 1 M Tris-HCl blocking buffer (pH 8.5) and washed with ultrapure water as well as with methanol and dried under a continuous nitrogen flow. The antibody microarray chip was assembled with a double-sided adhesive PE layer forming two channels by laser cutting and a PMMA plastic carrier. The PMMA plastic carrier provided inlet and outlet holes in order to create a flow-through microarray chip. The flow channels were filled with 60 μL of 5 % bovine serum albumin (BSA) in phosphate buffered saline (PBS).

CL-SMIA process on analysis platform MCR 3

The flow-based microarray analysis platform MCR 3 (GWK Präzisionstechnik GmbH, Munich, Germany) is described elsewhere in detail [13,14]. The following reagents are needed for the CL-SMIA process: 1. running buffer (casein 0.5 % w/v in PBS), 2. biotinylated detection antibody (diluted 1:500 in PBS, v/v, c = 2 μg/mL, pAb anti-Legionella pneumophila SG 1-12), Meridian, USA), 3. SA-Poly-HRP (diluted 1:5000 in casein, v/v, c = 0.2 μg/mL, Senova, Germany), 4. CL substrates (luminol and hydrogen peroxide, 1:1, v/v, Cyanagen, Italy). All process solution containers were connected with PTFE tubing of the MCR 3 and loaded automatically in the instrumentation before the measurement was started. Two measurements could be performed with one microarray chip, owing to the implementation of two flow-through channels on one chip. The microarray chip was inserted into the heatable microarray chip loading unit. Before each measurement, a background picture was taken with a CCD camera. This background image was automatically subtracted from the measuring images. The measurement program is computer controlled. The channel temperature was set to 35 °C and then 600 μL of the sample was injected manually into the microfluidic system using a plastic disposable sample syringe. Running buffer was pumped over the sample at a flow rate of 50 μL/s to push the sample into the incubation loop. Then 60 μL of the sample was pumped stepwise (intervals of 30 s) over the flow channel at a flow rate of 0.5 μL/s. This step was repeated 10 times. After a washing step (2000 μL of running buffer, 100 μL/s), the biotin-conjugated detection antibody solution was pumped to the incubation loop to fill the dead volume (150 μL, 50 μL/s) and pumped in stop-flow mode through the microarray (60 μL increments, 2 μL/s, interaction time of 5 s). This step was repeated 10 times. After a second washing step (2000 μL, 100 μL/s), 600 μL of SA-Poly-HRP was pumped through the flow channel at a flow rate of 2 μL/s. After a third washing step (2000 μL, 100 μL/s), 400 μL of a mixture of the CL substrates was pumped over the chip at a flow rate of 20 μL/s. The chemiluminescence signal intensity was recorded by a CCD camera for 60 s. After every measurement, the whole microfluidic system was rinsed with running buffer (1500 μL, 500 μL/s) to prevent cross contamination between measurements. The assay duration, including cleaning, is about 36 min.

The CL images of the SMIA were processed as described elsewhere, using the software MCRImage Analyzer (GWK Präzisionstechnik) [25]. All measurements were evaluated with MCRVisualization 1.0.6. The blank picture was subtracted from the measurement picture. The mean value was calculated with the 10 brightest pixels of each of five immobilized replicates in the row with the same target. The maximum allowed deviation per row was set at 15 %. All other spots were marked as outliers and not considered in the analysis.

Characterization of the MAF-CeUF-SMIA process

Calibration curves

For the calibration curve of samples without preconcentration, the prepared legionella stock suspension was heat inactivated at 70 °C for 30 min and diluted in ultrapure water to concentrations of 108, 107, 106, 105, 104, 103, and 102 CFU/mL. Then 600-μL aliquots of the samples were injected into the MCR 3 for SMIA analysis. All experiments were performed three times. To obtain calibration curves for the MAF-CeUF-SMIA process, a dilution series of freshly cultivated L. pneumophila in ultrapure water was prepared (107, 106, 105, 104, 103, 102, 101, and 100 CFU/mL). Prior to dilution, L. pneumophila were quantified by streaking on agar plates and counting colonies. Ten-liter surface water samples from the river Würm were collected and adjusted to pH 3 with hydrochloric acid. Samples were then spiked with the prepared dilution series to achieve concentrations of 103, 102, 101, 100, 10−1, 10−2, and 10−3 CFU/mL. The spiked surface water samples were processed directly after sampling.

Determination of recovery

L. pneumophila was cultivated and quantified by streaking on agar plates and counting colonies. A dilution series in ultrapure water was prepared (107, 106, 105, and 104 CFU/mL) and spiked into 10-L surface water samples. Water samples were collected in disposable 10-L bags from the river Würm and processed directly after collection. Samples were adjusted to pH 3 and spiked with 1 mL of the prepared dilutions to reach sample concentrations of 103, 102, 101, and 100 CFU/mL. Samples were processed by MAF-CeUF-SMIA. CL signals were quantified according to the calibration curves. Experiments were repeated three times for each concentration.

Preparation of LPS, living and heat-inactivated Legionella pneumophila

Ten-liter surface water samples were collected from the river Würm and adjusted pH 3 with hydrochloric acid. Samples were spiked with either purified LPS or living or heat-inactivated L. pneumophila. For living legionella, freshly cultivated L. pneumophila SG 1 of strain Bellingham were used. Heat-inactivated legionella were obtained by heat-shock treatment at 70 °C for 30 min from the same culture. LPS was isolated and purified according to a protocol developed by the Reference Laboratory Legionella of the Technical University Dresden. Cell concentration of freshly cultivated L. pneumophila SG 1 strain Bellingham suspension was determined by counting colonies on agar plates. Afterwards the cell suspension was shaken for 14 h at 850 rpm and 37 °C to enhance LPS shedding. After centrifugation for 10 min at 4000 × g the supernatant was filtered through a 0.2-μm cellulose acetate filter to separate cells and rough cell debris from soluble LPS. Concentrations of purified LPS are indicated as the respective stock suspension concentrations, since direct quantification of LPS was not possible. Purified LPS and living and heat-inactivated cells were diluted in ultrapure water and spiked into 10-L surface water samples to sample concentrations of 103, 102, 101, 100, and 10−1 CFU/mL. Samples were processed by MAF-CeUF-SMIA. Experiments were repeated three times for each concentration and each kind of feed.

Real water samples

Real samples were adjusted to pH 3 and processed by MAF-CeUF-SMIA. For sandwich microarray immunoassay, microarray chips spotted with all available antibodies were used. Tap and river water samples were processed directly. Treated wastewater samples were diluted 10-fold and 100-fold with ultrapure water prior to acidification and enrichment. All taken real samples are listed in Table 2 and were collected in 10-L disposable bags. All samples were analyzed in triplicate measurements.

Table 2 Real water sampling points
Full size table

Results and discussions

Principle of MAF-CeUF-SMIA

The complete process combining preconcentration and microarray detection is schematically shown in Fig. 2. All surface and tap water samples were processed by MAF after acidification (pH 3). The adsorbed bacteria and particles were then eluted within 20 mL of elution buffer and concentrated by CeUF to 1 mL. Overall, a volumetric concentration factor of 10,000 was achieved. Afterwards, the sample was measured on the microarray analysis platform MCR 3. A 600-μL aliquot of the concentrated sample was injected into the instrument for analysis via flow-based CL-SMIA. The antibody microarray captures either free or cell wall-bound LPS antigens of L. pneumophila by a stop-flow incubation process. The sandwich is formed by using biotin-labeled polyclonal detection antibodies against L. pneumophila. For CL detection poly-streptavidin-labeled horseradish peroxidase is pumped through the microarray chip before incubation with luminol and hydrogen peroxide. The CL signals are recorded by a CCD camera and the image is analyzed to get quantitative information. The combined concentration and measurement process takes about 90 min.

Fig. 2
figure 2

Schematic drawing of the L. pneumophila detection process combining monolithic adsorption filtration, centrifugal ultrafiltration, and automated microarray analysis. Time for whole process takes approximately 90 min

Full size image

Calibration of MAF-CeUF-SMIA

To compare CL signals of samples with and without enrichment steps prior to SMIA, calibration curves for both were determined. Figure 3 shows both calibration curves. Spiked heat-inactivated L. pneumophila water samples were processed by SMIA. Ten-liter surface water samples were spiked with living L. pneumophila SG 1 strain Bellingham before analysis by MAF-CeUF-SMIA. Calibration curves were determined by logistic regression, and detection limits and working ranges were calculated accordingly. Without preconcentration, detection limits for SMIA were 4.0 × 103 and 2.8 × 103 CFU/mL determined by pAb and mAb 10/6, respectively (Fig. 3). In order to improve the LOD, the concentration by MAF-CeUF was applied to gain a shift of the calibration curve. Thereby, the LOD could be enhanced to 8.7 CFU/mL and 0.39 CFU/mL determined by pAb and mAb 10/6, respectively. The linear ranges for samples treated by MAF-CeUF comprise 1.4 log steps for the mAb and 2.7 log steps for the pAb. This is similar to other antibody-based detection methods like ELISA. Comparing the MAF-CeUF-SMIA for pAb and mAb 10/6, the detection by mAb was much more sensitive because pAb captures all L. pneumophila occurring naturally in surface water. For mAb 10/6, the combined analysis method is sensitive enough to detect L. pneumophila SG 1 strain Bellingham in samples with concentrations below 1 CFU/mL. This represents the given threshold value for legionella in drinking water corresponding to the German Drinking Water Ordinance.

Fig. 3
figure 3

Calibration curves for L. pneumophila SG 1 strain Bellingham for spiked surface water samples processed by MAF-CeUF-SMIA and directly measured samples without prior enrichment. CL signals of immobilized polyclonal antibody (a) and monoclonal antibody 10/6 (b). Standard deviations (n = 3) are indicated as error bars

Full size image

Recovery experiments

To determine recoveries of the established analytical method, 10-L surface water samples were spiked with L. pneumophila SG 1 strain Bellingham in different concentrations and processed by MAF-CeUF-SMIA. Recovered concentrations were calculated from the CL signals of the SMIA according to the previously determined calibration curve. Results are shown in Fig. 4. Spiked concentrations ranged from 1 CFU/mL, which is slightly above the determined LOD, and up to a concentration of 103 CFU/mL, which is near the upper limit of the determined working range. For spiked L. pneumophila concentrations of 103, 102, 101, and 100 CFU/mL, recoveries were 97.8 ± 3.8 %, 100.9 ± 7.2 %, 91.9 ± 24.2 %, and 108.5 ± 28.5 % respectively. An average recovery of 99.8 ± 15.9 % was calculated. This result indicates that the MAF-CeUF-SMIA is highly reproducible in surface water.

Fig. 4
figure 4

Recoveries of L. pneumophila SG 1 strain Bellingham from surface water determined by monoclonal antibody 10/6. Standard deviations (n = 3) are indicated as error bars

Full size image

Discrimination between LPS and living- and heat-inactivated L. pneumophila

To determine whether the measured CL signals of the combined concentration and microarray detection system are caused only by living bacteria, or also by disrupted bacteria or single LPS structures, the combined method was tested with differently spiked surface water samples. Ten-liter samples of surface water from the river Würm were spiked with either living L. pneumophila, heat-inactivated L. pneumophila, or isolated LPS. Figure 5 shows the CL signals of the microarray detection for concentrated 10-L samples spiked with different concentrations of living or heat-inactivated cells or LPS. Samples spiked with heat-inactivated L. pneumophila display a reduction of signal intensity between 60 % and 89 % compared to samples spiked with living cells for initial concentrations between 101 and 103 CFU/mL. For samples spiked with purified LPS, the CL signal was reduced between 90 % and 95 % compared to samples containing living cells. This result indicates that LPS do not adsorb onto the monolith. Adsorption of heat-inactivated cells seems less effective than adsorption of living cells. We have shown that LPS from heat-inactivated cells (Fig. 3) and free LPS (data not shown) can be analyzed by CL-SMIA. Therefore it is concluded that living cells are concentrated whereas disrupted cells or free LPS pass through the monolith without adsorption during MAF. Thus, we can assume that MAF-CeUF-SMIA detects mainly living L. pneumophila.

Fig. 5
figure 5

Chemiluminescence signals for 10-L surface water samples spiked with L. pneumophila SG 1 strain Bellingham as LPS or living or heat-inactivated cells. Standard deviations (n = 3) are indicated as error bars

Full size image

Real samples

Real samples were analyzed to prove the concept of the combined analysis method for the detection of L. pneumophila in surface water. CL signals after MAF-CeUF-SMIA were determined with pAb and mAb 10/6. Results for pAb are shown in Fig. 6. Tap water shows CL signals in the range of ultrapure water, which was used as blank. All surface water samples from the rivers Isar and Würm as well as the waterfall sample have significantly higher CL signals.

Fig. 6
figure 6

Legionella pneumophila concentrations in real samples (surface water, tap water) detected with polyclonal antibody. Standard deviations (n = 3) are indicated as error bars

Full size image

Sampling sites along the river Isar were chosen with increasing anthropogenic influence from, e.g., wastewater treatment plants (WWTP), industrial plants, and agriculture. Sample Isar 1 shows the lowest CL signal compared to the other surface water samples because the sampling site near to Scharnitz lies close to the source of the river Isar. With increasing anthropogenic influence along the river, the pAb shows significantly higher CL signals compared to drinking water, e.g., like tap water from a shower (p level from paired t test p = 0.012, assumed level of significance ≤0.05). For sample Isar 4, the CL signal was higher than the other surface water samples. This sampling point was situated next to a float landing stage. Degradation of wood seems to have an influence so that L. pneumophila were found in higher concentration in water. Similar discoveries were made at a biological treatment plant that degrades organic substances from wood refinement processes into wastewater [26].

Higher CL signals of pAb were also determined for surface water of the river Würm for the sampling site in Gräfelfing. The sampling site of the waterfall was influenced by agriculture. The concentration of L. pneumophila seems to be in the same range compared to the other surface water samples. An additional CL signal for mAb 10/6, which detects L. pneumophila SG 1 strain Bellingham, was recorded at Sylvenstein reservoir. A concentration of 1.7 CFU/mL was determined corresponding to the calibration curve of MAF-CeUF-SMIA.

Conclusion

Pathogenic Legionella species are a risk for human health if nebulized by evaporative water systems. Surface water should be analyzed for L. pneumophila more frequently if it is used as system water. Culture-independent methods are highly recommended for a rapid screening. Antibody-based methods are fast but not sensitive enough. Concentration methods have to be combined to achieve the needed sensitivity.

In a test study, the combination of MAF, CeUF, and CL-SMIA was investigated. The application for tap and surface water was established using L. pneumophila SG 1 ST Bellingham as the test organism. We have shown for the first time that the combination of MAF and CeUF as concentration methods and CL-SMIA using mAb 10/6 enables quantitative measurements in surface water with detection limits lower than the recommended action values. In contrast to long-lasting cultivation experiments, the combined analysis process can be performed within 90 min. We could show that living L. pneumophila SG 1 strain Bellingham adsorb much more effectively to the monolithic disk during MAF than heat-inactivated cells or free LPS structures. The cultivation of L. pneumophila in surface water was not possible (data not shown), because competing microorganisms on the plates grow faster and L. pneumophila could not be detected.

We could demonstrate that the additional analysis with a highly selective mAb against one monoclonal subtype is promising.

Real samples were analyzed to show the applicability of the combined analysis method for surface water. The pAb indicates the occurrence of L. pneumophila SG 1-12 which are naturally present in surface water in low concentrations. The measured CL signals were increased at sampling sites with anthropogenic influence. On the basis of these results, the use of untreated surface water for evaporator water cooling systems should be avoided. For risk assessment more real water samples are needed for validation and alternative quantitative methods like molecular biological methods and cultivation should be considered to confirm our recommendation.

L. pneumophila strains, belonging to the monoclonal subtype Bellingham, are environment-associated pathogens with a lower infection risk. Most outbreaks are associated with monoclonal subtypes belonging to the Pontiac group (e.g., Philadelphia, Knoxville, Benidorm, France, and Allentown). Therefore, the monoclonal antibody panel has to be extended on the microarray chip to identify these subtypes rapidly with one measurement. The aim is a monoclonal subtyping of L. pneumophila on the MCR 3 for risk assessment. SMIA represents a fast and comprehensive detection method. Analysis of environmental sources and patient samples could be linked for establishing an effective outbreak management.