Improved biosensing of Legionella by integrating filtration and immunomagnetic separation of the bacteria retained in filters

A novel approach is presented that combines filtration and the direct immunomagnetic separation of the retained bacteria Legionella in filters, for further electrochemical immunosensing. This strategy allows for the separation and preconcentration of the water-borne pathogen from high-volume samples, up to 1000 mL. The limit of detection of the electrochemical immunosensor resulted in 100 CFU mL−1 and improved up to 0.1 CFU mL−1 when the preconcentration strategy was applied in 1 L of sample (103-fold improvement). Remarkably, the immunosensor achieves the limit of detection in less than 2.5 h and simplified the analytical procedure. This represents the lowest concentration reported to date for electrochemical immunosensing of Legionella cells without the need for pre-enrichment or DNA amplification. Furthermore, the study successfully demonstrates the extraction of bacteria retained on different filtering materials using immunomagnetic separation, highlighting the high efficiency of the magnetic particles to pull out the bacteria directly from solid materials. This promising feature expands the applicability of the method beyond water systems for detecting bacteria retained in air filters of air conditioning units by directly performing the immunomagnetic separation in the filters. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00604-023-06122-1.


Introduction
Legionella is the primary cause of a severe form of pneumonia [1], being Legionella pneumophila serogroup 1 (Lp1) the major responsible of infections in humans [2].Legionella can grow in natural and artificial water systems with little or no circulation [3,4].Transmission typically occurs by inhalation of contaminated aerosols or mist, with no person-toperson spread, being the infective dose as low as a single colony.Outbreaks are often associated with poorly maintained artificial water systems, such as cooling towers, evaporative condensers linked to air conditioning and industrial cooling, as well as hot and cold-water systems [5], among many others sources [6][7][8][9].Hospitals can also be sources of infection [10], as well as contaminated drinking water.The prevention of legionellosis relies on implementing control measures to minimize the growth of Legionella and the dissemination of aerosols.Besides the good maintenance of devices, regular cleaning, and disinfection [5], prevention strategies 82 Page 2 of 13 involve periodic on-site testing of potential outbreaks.The specific approaches for Legionella infection prevention and monitoring may vary depending on local regulations and geographic locations [11,12].Currently, the most widely used methods for detection involve time-consuming culturebased techniques followed by molecular methods.The limit of detection (LOD) for Legionella in different sources also vary depending on the regulations stablished by each country.For instance, the European Technical Guidelines for the Prevention, Control, and Investigation of Infections caused by Legionella species recommend testing samples in accredited laboratories with a LOD of 100 CFU L −1 [13].However, there is a general agreement that corrective actions should be taken when concentrations reach 10,000 CFU L −1 [11].For instance, cooling towers and spa pools require minor interventions or disinfection, typically at LODs higher than 100 CFU L −1 .Moreover, in cooling towers actions need to be taken from 10 5 to 10 6 CFU L −1 according to the American Industrial Hygiene Association (AIHA) Guidance on Legionella in building water systems [14].
The gold-standard method for the detection of Legionella in water samples includes high-volume sample collection of up to 1 L, followed by membrane filtration as a preconcentration step, isolation on culture plates for 10 days, and confirmation using sub-culturing according to ISO 11731.Alternatively, some commercial kits, as Legiolert® test (IDEXX Laboratories, Westbrook, ME, USA) based on enzymatic detection technology, use lower volumes of 1 or 100 mL providing results in 7 days [15].Lateral Flow Tests, such as Duopath® (Merck) or Hydrosense® (Albagaia, Edinburgh, UK), offer rapid visual redouts, but either require several days of enrichment to reach the LOD required by the legislation or yield semiquantitative results.To improve LODs without time-consuming pre-enrichment steps, the only alternative involves DNA amplification rather than bacterial growth.Commercial options for this approach include the microproof® Legionella Quantification LyoKit from BIOTECON Diagnostics GmbH (Potsdam, Germany) or the iQ-Check Legionella Real-Time PCR Kits from BIO-RAD (Bio-Rad Laboratories Inc, California, USA).
In order to reduce the time of analysis, it becomes imperative that emerging Legionella rapid testing technologies eliminate the necessity for pre-enrichment or DNA amplification procedures.As an alternative of amplification procedures, a highly efficient preconcentration strategy that combines filtration with immunomagnetic separation of the bacteria retained in the filters was successfully integrated to an immunosensing device.This novel approach has resulted in a significant improvement in the limit of detection compared to previously reported immunosensing devices [16][17][18][19][20][21].This approach enables the separation and preconcentration of pathogens from large sample volumes, typically 100-1000 mL, where immunomagnetic separation alone would not be feasible.The electrochemical immunosensor developed in this work is optimized and compared with the culturing methods in terms of the analytical features.Additionally, the performance of direct immunomagnetic separation of bacteria retained on various filtering materials is also assessed, highlighting the high efficiency of the magnetic particles to pull out the bacteria from solid materials.

Chemicals and biochemicals
Different set of buffers were used for specific procedures in the experiments, and their composition is described in S1 (Supp.data).All buffers were prepared from chemicals of analytical grade purchased from Merck and Sigma and using milliQ water.The reagents used for the electrochemical measurement includes hydroquinone (Ref.H9003) and hydrogen peroxide 30% solution (Ref.31642, Sigma-Aldrich).

Legionella pneumophila culture
L. pneumophila strain was grown in selective solid culture plates (Legionella MWY Selective Agar, Product no.10482513, Thermo Scientific) at 37 °C for 72 h.The bacteria were inoculated to distilled water and the solution was adjusted to an OD 0.370 at 600 nm, approximately equivalent to 10 7 CFU mL −1 .The concentration of Legionella samples was calculated for each experiment by colony counts verification in solid culture, as shown in Figure S1 (Supp data).

Evaluation of the immunomagnetic separation by scanning electron microscopy
To assess the performance anti-Legionella-MPs for binding Legionella, scanning electron microscopy was performed to visualize the interaction between the modified MPs and the bacteria.A volume of 100 μL of anti-Legionella-MPs at 10 7 MP mL −1 was incubated with 1 mL of L. pneumophila at 10 7 CFU mL −1 approximately, for 1 h at room temperature (RT) with agitation at 750 rpm.Afterwards, the supernatant was discarded, and the sample was washed three times with 1 mL of washing buffer.The sample was resuspended in 4 mL of PBS and it was filtered in a polycarbonate membrane to retain the modified MPs capturing the bacteria.Then, the filter was treated as previously described [16].

Electrochemical magneto immunosensing. Specificity study
The electrochemical magneto immunosensing procedure has been previously developed by our research group in other applications [16][17][18][19].Briefly, involves (i) the immunomagnetic separation (IMS), (ii) the incubation with the label antibody, and (iii) the electrochemical readout.The optimization of the reagents including anti-Legionella-MPs (ranging from 10 4 to 10 7 ) and anti-Legionella-HRP antibody (from 1/250 to 1/2000) are described in S3 (Supp.data).Briefly, 100 μL of sample containing Legionella was incubated with 100 μL of MPs at 10 7 MP mL −1 in a final volume of 1 mL with PBS for 1h under gentle rotation.After a washing step, the sample was incubated with 200 μL of anti-Legionella-HRP at a dilution of 1/500 for 30 min at 900 rpm and RT.After washing, the electrochemical readout was performed by resuspending the MPs in 40 μL of ePBS and 18 μL of electrochemical readout substrate solution.After 2 min of enzymatic reaction, the solution was dropped to the screen-printed electrode and the electrochemical readout was done based on the square wave voltammetry technique (SWV).More details about the SWV measurements are provided in S5 (Supp.data).The LOD of the electrochemical magneto immunosensing procedure was calculated by processing a calibration plot ranging from 0 to 3.6 × 10 5 CFU mL −1 .
The specificity study was performed by challenging the immunosensor with other water-borne bacteria at high concentration (10 7 /10 8 CFU mL −1 ) such as Pseudomonas aeruginosa, Klebsiella pneumoniae, Mycobacterium fortuitum, Enterobacter spp., Escherichia coli, and Salmonella choleraesuis in a final volume of 1 mL and proceeded as above.

Novel preconcentration method: study of the filtering material
The procedure combines three steps as described in Fig. 1: (i) filtration of large volumes of sample (typically 100 or 1000 mL), followed by (ii) immunomagnetic separation of the bacteria retained in the filter and magnetic actuation, (iii) electrochemical immunosensing, as described above.
The experimental details are shown in Figure S6 (Supp.data).After filtration from 100 to 1000 mL sample under vacuum, the 25-mm diameter filter was placed on a 2.0-mL tube.100 μL of anti-Legionella-MPs at 10 7 MP mL −1 and 900 μL of PBS were then added to the filter and incubated under gentle rotation at RT for 1 h.Then, the MPs with the captured Legionella were recovered under magnetic actuation and washed for 3 min at 900 rpm and RT.After that, 200 μL of the anti-Legionella-HRP antibody was added to the modified MPs and incubated for 30 min at 900 rpm and RT.After washing, the electrochemical readout was performed as described above.
In order to validate the approach, different materials were assessed.Polycarbonate (PC), nylon (NY), cellulose acetate (CA), cellulose nitrate (CN), and mixed cellulose ester membranes (MCE) were selected because of their widely use in microbiological filtration techniques with a 0.45-μm porous size and 25-mm diameter.For this study, 100 mL Legionella samples at the same concentration (2 × 10 3 CFU mL −1 ) as well as the negative control were filtered using each membrane and submitted to downstream analysis, as above.

Statistical analysis
Statistical analyses were conducted using Prism v 10.0.1 (GraphPad, San Diego, USA).Calibration curves were fitted using nonlinear regression with a four-parameter logistic curve 4PL, defined by the equation: where: Y: represents the electrochemical signal; D: stands for the lower asymptote, representing the response at very low concentrations; A: represents the upper asymptote, signifying the response at high concentrations; IC50: denotes the inflection point or concentration at the midpoint of the curve; HillSlope: signifies the slope parameter.
The limit of detection (LOD) was determined by analyzing negative control samples to calculate the mean value and standard deviation (SD).Subsequently, a one-tailed t-test was applied at a 95% confidence level to establish the cut-off value.The LOD was then derived by interpolation of this cut-off value using the 4PL equation, followed by the calculation of the corresponding Legionella concentration.

Biosafety considerations
All experiments were performed in a Biosafety Class 2 environment required for the handling of L. pneumophila.All biological waste generated from the experiments were disposed in accordance with the local regulations for handling biohazards.

Evaluation of the immunomagnetic separation by scanning electron microscopy
Following the incubation of the Legionella-containing sample with anti-Legionella magnetic particles (anti-Legionella-MPs), the sample was observed by scanning electron microscopy (SEM) to confirm the effectiveness of the immunomagnetic separation.In Fig. 2, it is confirmed that tosyl-activated magnetic particles were coated with anti-Legionella monoclonal antibodies to enhance the interaction between the particles and the bacteria.The images clearly demonstrate the successful binding of the modified magnetic particles to the L. pneumophila bacteria, resulting in the formation of clusters due to the polyvalency of both the bacteria and the magnetic particles.

Electrochemical magneto immunosensing: specificity study
The conditions for the electrochemical immunosensing were previously optimized as described in S3 (Supp.data).From the results, a concentration of 10 7 anti-Legionella-MPs mL −1 and anti-Legionella-HRP antibody 1/500 was used in all further experiments.Figure S4 (Supp.data), shows the calibration plot from 0 to 3.6 × 10 5 CFU mL −1 for the determination of the L. pneumophila with the electrochemical immunosensor without the integration of the novel preconcentration method.The data was fitted with a non-linear regression (Sigmoidal 4PL, Graph-Pad Prism Software v 10.0.1, R 2 = 0.9837) and the LOD was calculated, resulting in a value of 100 CFU mL −1 .As expected, this LOD is comparable with a magneto-actuated immunosensor previously reported by our research group but for other bacteria [17,22].
The specificity of the electrochemical biosensor towards a high concentration of other water-borne bacteria was also studied, and the results are shown in Fig. 3.The signal of the negative control containing 0 CFU mL −1 of Legionella (negative control) was found to be comparable with all the interferents bacteria samples (p < 0.05), including Pseudomonas aeruginosa, Klebsiella pneumoniae, Mycobacterium fortuitum, Enterobacter spp., Escherichia coli, and Salmonella choleraesuis, suggesting a good specificity for the assay.
The electrochemical magneto immunosensor was specifically designed for the determination of Legionella by utilizing a monoclonal antibody that targets the lipopolysaccharides of L. pneumophila strains.As a result, other bacteria are not detected, ensuring specificity for Legionella determination.It is important to note that the immunosensor primarily targets Legionella pneumophila serogroup 1, which is known for its high virulence and prevalence in cooling towers and distribution systems, making it the key strain responsible for most infections.

Novel preconcentration method: study of the filtering material
The LOD achieved by the electrochemical magneto immunosensor (1 × 10 2 CFU mL −1 ) are above the limits required by most legislations.Accordingly, and to overcome timeconsuming classical pre-enrichment [17] or DNA amplification [16], a novel preconcentration method is proposed, combining filtration and direct immunomagnetic separation (IMS) of the bacteria retained in the filters, as depicted in Fig. 1, to match the limits required by the legislation.Selecting the appropriate filter material is crucial to ensure a swift filtration workflow and minimize nonspecific adsorption of bacteria on the filters.This is essential for subsequent immunomagnetic separation and effective magnetic actuation of the retained bacteria in the filters.Accordingly, a first study was performed with a set of filters of different materials.There is a wide range of commercially available membranes with low retention rates and different properties.Filtering materials commonly used in microbiology were selected in this study, including polycarbonate, nylon, cellulose acetate, cellulose nitrate and mixed cellulose esters, and the results are shown in Fig. 4.They share a hydrophilic surface to avoid an excessive retention of the bacteria, and thus promoting the capture by the anti-Legionella-MPs directly in the filter.Moreover, the hydrophobic nature of the outer surface of Legionella [23,24] makes easier pull-off the trapped bacteria in hydrophilic surfaces.
According to the results shown in Fig. 4, no significant differences in the current intensities by using polycarbonate (PC), cellulose nitrate, and cellulose acetate (CA) filters were obtained (p > 0.05).The use of PC membranes in the filtration processes is recommended by some regulations, for instance the standard ISO11731:2017, and several studies support their efficacy as they provide the best recovery rates of this bacterium [25].Some of the advantages of PC track-etched membranes are their precision in the cylindrical pores providing a very accurate size cut-off, and their low nonspecific protein binding, similar to the CA membranes, which also provided good results.In contrast, nylon and mixed cellulose ester materials gave lower intensity currents in comparison with PC as a reference material.The high protein binding capacity of these filters could cause a highly retention of the bacteria in the filter, thus hindering the immunomagnetic capturing and resulting in lower preconcentration yields.Apart from the material, porosity, pore size, and the protein binding capacity, among other  features, other variables such as electrostatic charges can also influence in the filtration and the magnetic actuation.Although PC, CA, and cellulose nitrate membranes can all be considered as good candidates for the preconcentration strategy, the workflow time was different for each filtering material (Table S6, Supp.Data), providing the polycarbonate improved features, such as faster filtration rates, resulting in less than 1 min for the filtration of 100 mL sample and less than 10 min for the filtration of 1 L sample.In the case of the cellulose nitrate and cellulose acetate filters, the workflow time was higher.According to these results, PC filters were selected in all further experiments.The calibration plot integrating the novel preconcentration method (by filtering 100 mL of sample and IMS), and further electrochemical immunosensor is shown in Fig. 5 as red dotted line, for the determination of the L. pneumophila from 0 to 1 × 10 4 CFU mL −1 .The data was fitted with a nonlinear regression (Sigmoidal 4PL, GraphPad Prism Software v 10.0.1, R 2 = 0.9950) and the LOD was calculated, resulting in a value of as low as 2 CFU mL −1 .Compared with the electrochemical biosensor without the integration of the preconcentration method (Fig. 5, black solid line), an improving of two orders of magnitude was achieved, from 1 × 10 2 CFU mL −1 to 2 CFU mL −1 .According to some legislations, the approach is useful to detect the threshold of 10,000 CFU L −1 established for corrective actions, especially in cooling towers [14,26].However, certain regulations may require even lower LODs.In order to achieve a further improvement in the LOD, integrating the novel preconcentration method (by filtering 1000 mL of sample and IMS), and further electrochemical immunosensor is shown in Fig. 5 as blue solid line, for the determination of the L. pneumophila from 0 to 60 CFU mL −1 .The data was fitted with a non-linear regression (Sigmoidal 4PL, GraphPad Prism Software v 10.0.1, R 2 = 0.9923) and the LOD was calculated, resulting in a value of as low as 0.1 CFU mL −1 (100 CFU L −1 ).Interestingly, the filtration and immunomagnetic separation of 1000 mL of sample can avoid the use of time consuming pre-enrichment steps to reach the limits required by some legislations that establish more restrictive detection limits (from 100 to 1000 CFU L −1 ).Moreover, the approach is able to clearly detect as low as 0.6 CFU mL −1 with a signal to background ratio of 9.7 (n=3) (Fig. 5, B).
This approach allows the determination of low concentrations in high-volume samples that is one of the bottlenecks in the detection of environmental bacteria, in around 2.5 h.As shown in Fig. 5, the LOD of the electrochemical immunosensor without filtration (only IMS) resulted in 100 CFU mL −1 improved to up to 0.1 CFU mL −1 when the preconcentration strategy was applied in 1 L of sample (10 3 fold improvement).The values of the LODs and the total assay time for the electrochemical immunosensor combining filtration are summarized in Supp.data (Table S7).The readout time is also reduced to up to 3 min, based on the incubation of the enzymatic substrates (H 2 O 2 and HQ) with a total of 2 min and the electrochemical measurement of less than 1 min.This represents a 10-fold reduction in time in comparison with an immunoassay with optical readout, where the substrates require a minimum incubation time between 15 and 30 min [22].In previous work, we explored several strategies to achieve the LOD, such as incorporating a short preincubation step [17] or amplifying the genetic material after the IMS [16].In the present study, we have achieved a remarkable 10 3 -fold improvement in the LOD, by combining filtration and IMS.Compared with other electrochemical immunosensors and genosensors previously described for Legionella or other Gram-negative bacteria (and summarized on Table 1), the results confirm the promising features of combining filtration, immunomagnetic separation and electrochemical immunosensing for the ultrasensitive detection of Legionella using commercial screen-printed electrodes.
The electrochemical immunosensor described here allowed the determination of very low concentrations of Legionella in less than 2.5 h preventing time-consuming pre-enrichment steps filter treatments or DNA amplification.Further studies will focus on reducing the incubation time and streamlining it into a single step to simplify the analytical procedure.

Conclusions
A magneto-actuated electrochemical immunosensor is presented in this work capable of detecting and quantifying Legionella in contaminated water in less than 2.5 h with an impressive LOD of 0.1 CFU mL −1 .The most remarkable aspect of this approach is the elimination of classical preenrichment steps, DNA extraction, and amplification techniques, resulting in the analytical simplification and a significant reduction in the total analysis time.This innovative method combines a preconcentration strategy based on filtration to perform the IMS directly on polycarbonate filters.This allows for the handling of high-volume samples, typically ranging from 100 to 1000 mL.The electrochemical readout enables the use of a portable and cost-effective device that is compatible with in-field test.This represents the lowest reported LOD for the electrochemical biosensing of whole Legionella bacteria without the need for pre-enrichment and DNA amplification steps.The LOD achieved by this method underscores its utility, especially considering that some of the most stringent regulations require a LOD of at least 100 CFU L −1 .Another noteworthy fact relies on the analytical simplification.In the gold standard microbiological method, after sample filtration, the retained bacteria were resuspended or extracted from the filter using special buffers or vortexing.In this study, the preconcentration strategy is integrated to the electrochemical biosensing.By filtering high volumes of samples and subsequently employing magnetic actuation to pull out the bacteria directly from the filter, the need for resuspension or additional steps for downstream analysis was eliminated.Another remarkable aspect of this approach is the potential of the magnetic particles of successfully removing the bacteria directly from various filtering materials.This promising characteristic broadens the applicability beyond water systems to include the detection of bacteria retained in air conditioning unit air filters, accomplished through direct immunomagnetic separation within the filters.

Fig. 1
Fig. 1 Schematic representation of A filtration, B immunomagnetic separation of the bacteria retained in the filters and magnetic actuation, followed by C electrochemical immunosensing for the detection of Legionella pneumophila in water samples.In B, experimental details are presented, showing how the filter is positioned on the

Fig. 2 Fig. 3
Fig. 2 Evaluation of the IMS by SEM at a L. pneumophila concentration of 10 7 CFU mL −1 .The images show the L. pneumophila cells attached to the magnetic beads.In all cases, identical acceleration voltage (15 KV) was used

Fig. 4 Fig. 5 A
Fig. 4 Study of the filtering material by the filtration of a 100 mL sample containing Legionella (2 × 10 3 CFU mL −1 ) as well as the negative controls.All the filters were 0.45 μm pore size and 25 mm of diameter.The filtering materials tested were nylon (NY), mixed

Table 1
An overview of electrochemical biosensing approaches for the detection of Legionella