Survival of animal and human-associated Listeria monocytogenes in drinking water and biofilms Open Access

Contamination of water resources and drinking water with small animals and animal waste may result in transmission of live pathogens (Levy 1990; Bitton 2014; Hynds et al. 2014; Stokdyk et al. 2020). Gastropods such as land slugs are occasionally observed as contaminants in groundwater wells and drinking water storage tanks and as unwelcome intruders at drinking water treatment plants. Slugs often feed on a variety of organic food items and are also known to ingest feces from animals and humans as part of their diet (coprophiles). Several potential pathogens have been observed in the intestinal tract and on the exterior surface of land slugs including Escherichia coli, Enterococcus faecalis, Pseudomonas aeruginosa, Clostridium perfringens, Clostridium botulinum and Listeria monocytogenes (Elliot 1970; Sproston et al. 2006; Gismervik et al. 2014, 2015).

The land slug Arion vulgaris (formerly Arion lusitanicus) is considered a major pest in many regions of Europe and the population has increased significantly in the last 20 years (Pfenninger et al. 2014; Zemanova et al. 2016). A. vulgaris is active during periods with high humidity and during the night (Cameron 2016). A. vulgaris prefers to hide in dark and moist places, and seeks protection in moist and frost-protected places during winter (Cameron 2016). A. vulgaris can range in size from >10 cm to <1 mm and this and other land slugs can penetrate protective barriers less than a few mm (Figure 1). As a result, A. vulgaris can be observed in many anthropogenic habitats and sometimes as a contaminant in urban and rural water utilities. A. vulgaris has been shown to be a potential vector for L. monocytogenes and this pathogen has been detected in slug mucus and feces at concentrations of 10–1,205 CFUs per g slug biomass (Gismervik et al. 2015). Because A. vulgaris is a potential vector for Listeria it is likely that this and other pathogens can be transmitted to drinking water if the water utility is compromised by slugs.

Listeria monocytogenes is a facultative anaerobic Gram-positive, non-spore-forming psychrotrophic bacterium that is relatively resistant to adverse environmental conditions (Ivanek et al. 2006; Gandhi & Chikindas 2007). L. monocytogenes is widespread in the environment including soil, water and feces, and it has many known animal reservoirs (Ivanek et al. 2006; Hellberg & Chu 2016). L. monocytogenes is a pathogen and the causative agent of listeriolosis and other illnesses in animals and humans. Listeriolosis is associated with a relatively high case-fatality rate in humans, and because the mortality rate can be high, many countries have a zero tolerance for this bacterium in relation to food and water (Gião & Keevil 2014; Pandove et al. 2016). L. monocytogenes has multiple environmental transmission routes and the prevention of infection of animals and humans is complex and challenging (Ivanek et al. 2006). Interestingly, the environmental occurrence and survival of L. monocytogenes may increase with future climate changes (Hellberg & Chu 2016). However, little is known about the persistence of L. monocytogenes in many types of drinking water and in drinking water biofilms. Indigenous drinking water biofilms have been implicated as environmental reservoirs or safe havens for intruding pathogens (Keevil 2002; Wingender & Flemming 2011; Bitton 2014).

In the present study, we investigated the persistence of slug- and human-associated L. monocytogenes strains in municipal drinking water produced from groundwater. Flow systems were used to examine the fate and survival of L. monocytogenes in drinking water sand filters with and without biofilms. Radiolabeling of Listeria cells was used to study cell turnover, and results for L. monocytogenes were compared with results for the traditional fecal pollution indicator E. coli.

Drinking water samples for Listeria screening were obtained from five rural Danish drinking water treatment plants that use groundwater as source water (Ingstrup, Mosbjerg, Tolne, Kringelmosevej, Ravnshøj). Grab samples (10 L) of source water were collected before water treatment. The water treatment consisted of aeration and basic sand filtration followed by distribution without disinfection. Aeration followed by filtration in open or closed sand filters without subsequent disinfection represent the standard operating procedure for the vast majority of municipal and private waterworks in Denmark.

For microcosm experiments, municipal drinking water was obtained from Aalborg Municipality (Denmark). The source water was non-disinfected hard groundwater (12–16 °dH) abstracted directly from chalk aquifers. Limited water treatment is employed by this municipality before distribution to the consumers, and this water was selected for the microcosm experiment to mimic groundwater and drinking water with minimum treatment. The drinking water is naturally nutrient-poor (low in total P, N and organic C), with a non-volatile organic carbon concentration (NVOC) of 1.3 mg/L, a chemical oxygen demand (COD) of 0.4 mg/L, an NO₃⁻ concentration of 4.8 mg/L, and concentrations of NH₄⁺, NO₂⁻, and total P below 0.1 mg/L. The water temperature, pH and turbidity were 9.8 °C, 7.6 and <0.05 FTU, respectively.

The Listeria strains used in the drinking water experiments are listed in Table 1. Six Listeria strains were isolated from the pest slug A. vulgaris (Figure 1). A. vulgaris was collected in areas with drinking water abstraction and slug samples were subsequently used for enrichment and isolation of Listeria. Slug slime and slug feces were serially diluted in Ringer's solution and then streaked onto Brilliance Listeria Agar (Oxoid Ltd, UK). The Listeria agar was supplemented with a selective supplement (Nalidixic acid, Polymyxin B, Ceftazidime, Amphotericin), and a differential supplement (Lecithin) to promote identification of presumptive L. monocytogenes (Oxoid Ltd, UK). Plates were incubated for 48 h at 30 °C and then screened for presumptive L. monocytogenes (ß-glucosidase positive and Lecithinase positive). Isolated strains were subsequently characterized by the Oxoid Biochemical Identification System (OBIS; Oxoid Ltd, UK) and the PhenePlate system (PhPlate Microplate Techniques AB, Sweden).

The human clinical strain L. monocytogenes SS-ENK was kindly provided by Prof. H. Schönheyder, Aalborg University Hospital, Denmark. The clinical reference strain L. monocytogenes Scott A was kindly provided by Prof. S. Knøchel, University of Copenhagen, Denmark. L. monocytogenes DSM 20600 and L. monocytogenes DSM 12464 were obtained from the German Collection of Microorganisms and Cell Cultures GmbH (DSMZ, Germany).

Environmental Listeria strains and reference Listeria strains were grown for 18 h at 30 °C in Brain Heart Infusion Broth (BHI; LAB M, UK) before being spiked into drinking water microcosms. All strains were centrifuged (10,000 g; 10 min) and washed with filter sterilized drinking water before spiking. E. coli ATCC 59222 (American Type Culture Collection, USA) was included in some survival experiments to compare results obtained with Listeria with that of a traditional fecal indicator bacterium (FIB). E. coli ATCC 25922 was grown for 18 h at 30 °C in Tryptic Soy Broth (TSB; Difco, USA) and then centrifuged and washed as described above for Listeria.

Initial survival experiments with L. monocytogenes in drinking water were conducted using glass microcosms as described by Bjergbæk & Roslev (2005). The water used in the microcosms was municipal drinking water as described above. L. monocytogenes were inoculated into 100 mL drinking water to obtain a final concentration of approximately 10⁶ bacteria mL⁻¹. The microcosms were incubated in the dark at 10 °C on a rotary shaker at 120 rpm. Triplicate microcosms were incubated for each Listeria strain.

Experiments focusing on the persistence of L. monocytogenes in drinking water filters with and without biofilms were carried out using 50 cm³ glass columns packed with 50 gram of 2 mm quartz sand (Figure 2). Two systems were constructed containing either sterile quartz sand (n = 3) or quartz sand covered with a drinking water biofilm (n = 3). The drinking water biofilm had been established over three months in a flow system with a continuous flow of municipal tap water (non-disinfected). Initial drinking water biofilm formation on quartz sand and subsequent survival experiments with Listeria were carried out in the dark at 10 °C. Drinking water was pumped through the sand columns at a flow of 8 mL min⁻¹. The water was recycled from a 5 L reservoir and was partly replaced with fresh drinking water every 48 h.

L. monocytogenes were inoculated into the drinking water reservoirs of the flow systems to a final concentration of approximately 10⁶ cells mL⁻¹ (Figure 2). Separate flow systems were inoculated with E. coli to similar densities to compare the persistence of Listeria and that of a traditional FIB. Water was sampled from the reservoir every 48 h before the water was replaced. Quartz sand was sampled from the columns at the end of the experiment and transferred to filter sterilized drinking water. Attached bacteria were dispersed by three sequential treatments of 30 sec sonication (47 kHz, Branson Ultrasonics, USA), and 30 sec vortex mixing. Water and biofilm suspensions were analyzed for bacterial content as described below.

Culturable Listeria in environmental samples were analyzed in triplicates by membrane filtration of 1–4 L groundwater using 0.45 μm mixed cellulose ester filters (Millipore, USA). The filters were transferred to sterile pads (Millipore, USA) soaked with Fraser Broth (Sigma-Aldrich), and resuscitated for 18 h at 30 °C followed by a transfer to Brilliance Listeria agar with selective and differential supplements as described above. Listeria plates were incubated at 36 °C and examined after 48 h.

Culturable L. monocytogenes in laboratory experiments was detected in water and biofilm suspensions by plating 0.1 mL aliquots of serially diluted samples onto Brilliance Listeria agar with selective and differential supplements as described above. Plates were incubated at 36 °C and examined after 48 h.

Culturable E. coli in laboratory experiments was detected in water and biofilm by membrane filtration followed by incubation for 24 h at 36 °C on Chromocult Coliform Agar (CCA; Merck, Germany) as described previously (Roslev et al. 2004). Filtration was carried out using 0.45 μm mixed cellulose ester filters (Millipore, USA).

Direct viable count (DVC) combined with fluorescent in situ hybridization (FISH) was used to detect viable L. monocytogenes using a modified version of the method described by Besnard et al. (2000). Samples were filtered onto black polycarbonate membrane filters (0.2 μm pore size; Poretics Products, USA). Filters were placed on pads soaked with Brain Heart Infusion Broth containing 1.0% yeast extract (Difco, USA) and ciprofloxacin at a final concentration of 1.0 μg mL⁻¹ (Sigma-Aldrich, Germany). Filters were incubated for 7 h at 30 °C and FISH was then performed on the filters using the ‘Lis 637’ probe to identify L. monocytogenes (Schmid et al. 2003). FISH-positive cells were enumerated by epifluorescence microscopy using a Zeiss Axioskop (Zeiss, Germany) at a magnification of 1,000. At least 40 random microscope fields were enumerated for each hybridization. The most accurate cell count was obtained for filters containing 10⁵–10⁶ cells and was evaluated by comparing measured and nominal concentrations. The sample volume for filtration was then adjusted accordingly. Cells were considered DVC-positive if they were FISH-positive and ≥1.5 times the average length of control cells that had not been incubated with ciprofloxacin.

L. monocytogenes and E. coli were radiolabelled with ¹⁴C to study cell turnover in drinking water biofilms. Endogenous bacterial respiration in live cells will produce radiolabelled CO₂ (¹⁴CO₂) whereas ingestion and degradation of labelled bacteria by other microorganisms (e.g., grazing by protozoa) will further increase ¹⁴CO₂ release (Taylor & Sullivan 1984).

L. monocytogenes DSM 12464 and E. coli ATCC 25922 were grown for 18 h at 30 °C in 1 mL 10% Tryptic Soy Broth supplemented with 0.125 mL [¹⁴C]glucose (7.4 MBq mL⁻¹ stock solution, 11.47 GBq mmol⁻¹, D-[U-¹⁴C]glucose; Amersham Pharmacia Biotech, UK). Cultures were washed twice with filter sterilized drinking water, and then inoculated into drinking water microcosms containing: (1) filter sterilized drinking water and sterile quartz sand (control bottles), (2) drinking water and sterile quartz sand, or (3) drinking water and sand covered with drinking water biofilm. Filter sand was sterilized by preheating to 450 °C. The microcosms consisted of 60 mL glass bottles with 50 mL drinking water, sealed with butyl rubber stoppers. All microcosms were spiked with either ¹⁴C-labelled L. monocytogenes (approximately 5 × 10⁵ cells mL⁻¹) or ¹⁴C-labelled E. coli (approximately 1 × 10⁶ cells mL⁻¹). The total amount of radioactivity was approximately 1.0 × 10⁴ Bq per microcosm with ¹⁴C-labelled L. monocytogenes and 1.7 × 10⁴ Bq per microcosm inoculated with ¹⁴C-labelled E. coli. Microcosms were incubated in the dark at 120 rpm on a rotary shaker at 10 °C. Gas and liquid sampled from the microcosms were analyzed for ¹⁴CO₂. Gas samples of 1 mL were injected into scintillation vials containing an 8 mL CO₂ trap (ethylene glycol monomethyl ether and ethanolamine 7:1; Merck, Germany). Vials were shaken whereafter a 10 mL scintillation cocktail was added (Hionic Fluor; PerkinElmer^®, USA). The ¹⁴CO₂ in the liquid samples was evacuated and trapped after acidification. The radioactivity was quantified using a Packard 1600 Tri-Carb Liquid Scintillation Analyzer (Packard, NL). Counts were corrected for quench using external standards.

Comparisons of differences in survival between Listeria strains and between different detection methods were carried out using the nonparametric Mann–Whitney U test (Wilcoxon rank sum test) using KaleidaGraph 4.5.4 (Synergy Software, USA).

Source water from five rural drinking water treatment plants was examined for the presence of Listeria spp. The water supplies use groundwater as source water and employ conventional water treatment consisting of aeration and sand filtration. Disinfection is not part of the operating procedures at these treatment plants. Presumptive Listeria was not detected in water volumes <0.01 L whereas analysis of 1–4 L resulted in detection of presumptive Listeria in four out of five rural water supplies (0.3–90 CFU/L). This initial screening was not intended to be exhaustive but it indicated that low concentrations of Listeria were detectable and this suggested a potential for occurrence of Listeria in rural water supplies. It was not possible to establish the origin of the Listeria but it is known that rural groundwater is sometimes prone to contamination with live invertebrates and waste from wildlife and farm animals (Levy 1990; Bitton 2014; Hynds et al. 2014; Stokdyk et al. 2020).

Survival of L. monocytogenes in drinking water was examined by spiking human and animal strains into drinking water microcosms at 10 °C (Figure 3). The drinking water used in the microcosms was municipal drinking water from groundwater and 10 °C is a typical temperature in such systems. The standard operating procedure in this municipality advocates distribution of water to consumers without water treatment or disinfection (non-treated pristine groundwater). The drinking water contained no detectable coliform bacteria or intestinal enterococci (<1 CFU/100 mL), and the heterotrophic plate count was <10 CFU/mL. Background concentrations of L. monocytogenes were not detected in the microcosms (<1 CFU/1,000 mL). Drinking water microcosms were subsequently spiked with L. monocytogenes from the pest slug (Arion vulgaris), a strain from poultry (DSM 12464), a strain from rabbit (DSM 20600), and a human clinical isolate (SS-ENK).

One of the spiked L. monocytogenes strains (DSM 20600) lost culturability after only nine days in drinking water whereas the three other strains remained culturable for up to 35–47 days (Figure 3). L. monocytogenes DSM 20600 is a laboratory strain (type strain) originally isolated in 1926, and this bacterium may subsequently have lost some ability for survival under nutrient-poor conditions resulting in relatively poor survival in drinking water. The first-order decay rate for L. monocytogenes DSM2060 was 0.703 day⁻¹. In contrast, decay rates for L. monocytogenes SS-ENK, L. monocytogenes DSM12464 and L. monocytogenes slug1 were 0.457, 0.425 and 0.314 day⁻¹, respectively. The decay rates for the latter three L. monocytogenes strains correspond to T99 values of 10–15 days in drinking water (i.e., the time for a 99% decrease in CFU) (Figure 3).

After 28 days of incubation in drinking water, the survival of L. monocytogenes was also evaluated based on DVC-FISH. At this time-point, less than 0.01% of the initial population of L. monocytogenes remained culturable but the DVC-FISH method suggested that many Listeria were still substrate-responsive (DVC-positive). The number of nonculturable cells with detectable metabolic activity was 10–100-fold higher than the number of culturable cells (Mann–Whitney p < 0.05). This suggested that 0.1%–1% of the L. monocytogenes cells inoculated into drinking remained metabolically active after 28 days and were potentially viable. The relatively long survival of L. monocytogenes in non-disinfected drinking water observed in the present study supports previous studies suggesting that Listeria can survive for extended periods in cold and oligotrophic environments (Besnard et al. 2002; Ivanek et al. 2006; Budzińska et al. 2012). Although the conditions do not favor cell proliferation, low temperatures and trace concentrations of organics may support cell metabolism and survival. Besnard et al. (2002) reported that L. monocytogenes remained culturable in sterile water at 4 °C for up to six weeks, and thereafter maintained cell integrity and metabolic activity for more than ten weeks. Budzińska et al. (2012) also observed that environmental survival of L. monocytogenes was superior at low temperatures (4 °C) compared with higher temperatures (20 °C). Collectively, these findings emphasize the ability of L. monocytogenes to survive well at low temperatures, which is a characteristic that makes the control of this organism a challenge (Gandhi & Chikindas 2007).

Pathogens in drinking water utilities may survive as free-living planktonic cells in the water phase or in sessile communities attached to submerged surfaces colonized by microorganisms indigenous to the environment (Bitton 2014). The effect of surface attachment and biofilms on the survival of L. monocytogenes in drinking water was investigated using a laboratory-scale filter system containing either virgin quartz sand or quartz sand colonized by microorganisms indigenous to drinking water (Figure 4). The attachment and survival of E. coli was studied in parallel experiments to compare results for L. monocytogenes with those of a traditional FIB (Figure 4(b)). The drinking water biofilm in the filter system had been established over three months using a continuous flow of non-disinfected municipal tap water (Figure 2). The drinking water biofilm cell count determined by DAPI staining was 4.75 × 10⁶ ± 3.98 × 10⁵ cells/cm² at the beginning of the experiment.

Samples from the effluent of the filter columns showed that live L. monocytogenes and E. coli persisted in the flow systems past the theoretical washout (Figure 4(a) and 4(b)). The theoretical washout was calculated based on no attachment of cells to surfaces and also considered the partial water replacement during the experiment. Culturable cells of L. monocytogenes and E. coli persisted about three times longer than the projected washout. The theoretical washout corresponded to a dilution rate of 2.3 d⁻¹ whereas the measured exponential decline for L. monocytogenes and E. coli in the absence of biofilms corresponded to a disappearance rate of 0.79 d⁻¹ and 0.67 d⁻¹, respectively. Low water-temperatures (10 °C) and low concentrations of bioavailable organic carbon in the drinking water prohibited growth of E. coli and L. monocytogenes in the experiments, and only survival was observed. The absence of growth has also been observed for other fecal microorganisms in this type of drinking water (Roslev et al. 2004). The presence of low concentrations of L. monocytogenes and E. coli in the water phase past the theoretical washout suggested a slow release of cells attached to surface material in the flow systems.

L. monocytogenes and E. coli showed a more rapid decline in drinking water flow systems containing biofilm-covered filter material compared with columns with only virgin filter material (Figure 4). The removal rate for L. monocytogenes measured in the effluent was 1.37 d⁻¹ in the presence of a drinking water biofilm and 0.79 d⁻¹ in the absence. This observation was investigated further by targeting bacterial cells attached to filter surfaces with and without an indigenous biofilm. The results confirmed that the presence of a drinking water biofilm affects the surface attachment and persistence of L. monocytogenes and E. coli. Interestingly, the initial attachment of L. monocytogenes to biofilm-covered filter material was 10–100-fold greater compared with virgin surfaces (Figure 5), and the difference was statistically significant (Mann–Whitney, p < 0.026). After the initial attachment, there was, however, a rapid decline in the concentration of detectable L. monocytogenes on the biofilm-covered material but only a slow decline in concentrations of attached cells on the virgin material (Figure 5). After 14 days, L. monocytogenes could no longer be detected on biofilm-covered surfaces, whereas attached L. monocytogenes were detectable on surfaces without an initial biofilm (Figure 5). Comparable results were obtained with a traditional cultivation dependent method (CFU) and a FISH staining technique (Figure 5). Use of FISH confirmed that the observed decrease in surface-attached cells was not merely due to loss of culturability but due to a de facto decrease in cell numbers. Similar results were obtained in parallel experiments with E. coli where cells could be detected for >40 days on virgin filter material without an initial drinking water biofilm (data not shown). Collectively, these findings suggest a potential for extended survival of L. monocytogenes on surfaces in drinking water which may be useful information in relation to risk assessment of contaminated water supplies. A potential for extended survival of sessile L. monocytogenes in water supply systems may also be relevant for the food industry because tap water is often used for manufacturing and cleaning but sources of Listeria contamination are sometimes elusive (Ivanek et al. 2006; Gandhi & Chikindas 2007; Gião & Keevil 2014).

Results from experiments with surface attachment of L. monocytogenes and E. coli to filter material indicated that surfaces covered with drinking water biofilms stimulated initial surface attachment, but also caused decreased survival of attached cells. This phenomenon was investigated further by measuring the fate of ¹⁴C-labelled L. monocytogenes and E. coli inoculated into drinking water microcosms (Figure 6). The turnover of radiolabelled bacterial cells was measured as recovery of ¹⁴CO₂ with and without an indigenous drinking water microflora. Net growth of L. monocytogenes and E. coli was not observed in our drinking water microcosms (Figures 3–5), and ¹⁴CO₂ released from radiolabelled bacteria likely originated from endogenous metabolism and mineralization of ¹⁴C-labelled biomass. Mineralization of ¹⁴C-labelled bacterial biomass to ¹⁴CO₂ may then be used as a tracer of starvation metabolism within the labelled bacteria and predation (grazing) by other microorganisms (Taylor & Sullivan 1984). In the present study, control microcosms containing filter sterilized drinking water and sterile virgin filter sand were therefore included to establish release of ¹⁴CO₂ from L. monocytogenes and E. coli without competition or predation from indigenous microorganisms.

Turnover of ¹⁴C-labelled L. monocytogenes and E. coli was three and ten times greater in microcosms with an indigenous microflora (filter material with biofilm and drinking water) compared with microcosms without native drinking water microorganisms (Figure 6). In experiments with E. coli, the turnover of ¹⁴C-labelled bacterial cells was also more rapid in bottles with biofilm-covered filter material compared with drinking water without filter material (Figure 6). After >20 days, the release of ¹⁴CO₂ from L. monocytogenes and E. coli in microcosms without indigenous microflora corresponded to 6%–12% of the initial amount of radiolabelled cell material. In contrast, microcosms with native drinking water microorganisms caused a 37%–62% mineralization of radiolabelled L. monocytogenes and E. coli (Figure 6). After 26 days, approximately 35% of the added ¹⁴C-labelled L. monocytogenes had been recovered in experiments with L. monocytogenes and biofilm-covered filter material, which corresponded to a 30% recovery as ¹⁴CO₂ after 23 days in microcosms containing E. coli and biofilm-covered filter material.

Biofilms in drinking water systems are sometimes implicated as environmental reservoirs or safe havens for intruding pathogens (Keevil 2002; Wingender & Flemming 2011; Bitton 2014; Pandove et al. 2016). In the present study, surface attachment of L. monocytogenes and E. coli to biofilm-covered filter material increased the persistence and delayed washout from drinking water flow systems. This is in agreement with previous studies suggesting that L. monocytogenes can attach and form biofilms on different surfaces in tap water (Gião & Keevil 2014; Pandove et al. 2016). However, both L. monocytogenes and E. coli disappeared more rapidly from biofilm-covered filter surfaces compared with filter material not colonized by indigenous microorganisms. The results obtained from release of ¹⁴CO₂ from radiolabelled L. monocytogenes and E. coli suggested that the native microflora in drinking water and filter material negatively affected the survival of the two fecal bacteria. Turnover of L. monocytogenes and E. coli biomass increased dramatically when the fecal bacteria were incubated with indigenous microorganisms compared with incubations without the native microflora. The increased turnover of cell material may be caused by increased competition from other bacteria and predation (grazing) by drinking water protozoa. Studies have shown relatively high concentrations of protozoa including ciliates, flagellates and amoeba in drinking water utilities (Sibille et al. 1998; Artz & Killham 2002; Bitton 2014). Bacterial consumption in drinking water biofilms due to grazing by protozoan amoeba can amount to >10⁴ cells/cm²/h (Bitton 2014). Hence, grazing by protozoa has a potential to reduce numbers of fecal contaminants in drinking water. Non-indigenous pathogens such as L. monocytogenes are likely more exposed to grazing compared with indigenous bacteria embedded in already established drinking water biofilms, and protozoan grazing is therefore a factor that likely affects the long-term survival of L. monocytogenes in non-disinfected drinking water.

Land slugs often have an omnivorous diet and feed on both live and dead organic material including detritus, carrion and feces. Slugs may penetrate drinking water supply systems and represent potential vectors of human- and animal-associated pathogens such as Listeria monocytogenes. Human- and animal-derived L. monocytogenes remained culturable for days to weeks in non-disinfected drinking water, and attachment to filter material increased persistence and delayed washout from drinking water flow systems. L. monocytogenes declined more rapidly in systems with biofilm-covered filter material compared with systems with virgin filter material not colonized by indigenous drinking water microorganisms. Live and sessile L. monocytogenes can likely persist for weeks on surfaces in cold nutrient-poor drinking water and may subsequently be released to the water phase with potential implications for public health. Contamination of water supplies by potential Listeria vectors such as land slugs should therefore be prevented.

We thank Kirsten Maagaard for excellent technical assistance. We also thank Susanne Knøchel for providing L. monocytogenes Scott A, and Henrik Schönheyder for providing L. monocytogenes SS-ENK.

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