Inhibitory effects of silver ions on Legionella pneumophila grown on agar, intracellular in Acanthamoeba castellanii and in artificial biofilms


Christian Lück, Institute of Medical Microbiology and Hygiene, University of Technology, Faculty of Medicine ‘Carl Gustav Carus’, Fetscherstr. 74, 01307 Dresden, Germany. E-mail:


Aims:  We undertook a series of experiments to investigate the susceptibility of Legionella pneumophila grown under extracellular and intracellular conditions and other water-related bacteria to silver ions.

Methods and Results:  In this study, the antimicrobial effect of silver ions to intra- and extra-cellular grown Legionella bacteria was investigated. The minimal inhibitory concentration (MIC) after 24 h exposure, leading to a 5 log reduction, was c. 64 μg l−1 AgNO3 for extracellular grown Legionella and other tested Gram-positive and Gram-negative bacteria. In contrast, the MIC for intracellularly grown Legionella was up to 4096 μg l−1 AgNO3 after 24 h. Furthermore, the heterotrophic bacteria grown within a biofilm model were killed at a concentration of 4–16 μg l−1 AgNO3. In contrast, biofilm-associated Legionella were less sensitive (MIC 128–512 μg l−1 AgNO3).

Conclusion:  Intracellularly and biofilm-grown legionellae are less sensitive against silver compared with agar-grown bacteria.

Significance and Impact of the Study:  The reduced sensitivity of Legionella grown in amoebae might explain why the effect of silver decontamination requires an extended exposure in field trials.


Legionella are Gram-negative, aquatic bacteria that are able to cause Legionella pneumonia called Legionnaire’s disease or Pontiac fever, a nonpneumonic, influenza like illness. These bacteria are intracellular pathogens of freshwater protozoa, and they utilize similar mechanisms for survival in protozoan hosts and for infecting human cells (Fields et al. 2002). The most predominant amoebae in water reservoirs are Hartmanella spp. and Acanthamoeba spp. (Molmeret et al. 2005). Human infections are caused by the inhalation of aerosols containing Legionella cells recently released from protozoa or protozoa containing Legionella cells.

The contamination of drinking water supplies in hotels, hospitals and public buildings with Legionella is well known (Liu et al. 1998; Kusnetsov et al. 2001; Perola et al. 2005; Zhang et al. 2008; von Baum et al. 2010). Hot water distribution systems have been shown to be the primary reservoirs for many Legionella species (Rohr et al. 1999). To suppress the growth of legionellae in large buildings, decontamination by hypochlorination or super heating might be applied (Goetz and Yu 1997; Blanc et al. 2005).

The addition of silver ions can also be used as a decontamination method. Silver cations (Ag+) show a lethal effect on many micro-organisms including bacteria, viruses and algae, but they are not cytotoxic for human cells (Landeen et al. 1989). The antimicrobial effect of silver was first investigated by Modak and Fox (1973) and Thurman and Gerba (1988). They suggest that silver ion interacted with the bacterial cell membranes and interfered in DNA function by binding along the phosphate group of the helical DNA chain and by forming silver–DNA complexes (Modak and Fox, 1973). Furthermore, silver has a high affinity to sulfhydryl groups (Hwang et al. 2007; Jung et al. 2008). In addition, silver inhibits the transport of electrons of the respiratory chain (Silver et al. 2006).

Currently, silver is included in many commercially available healthcare products (Percival et al. 2005), used for treatments of burns (McHugh et al. 1975), or as coating on intravasal catheters (Adams et al. 1999). Water disinfection systems based on copper/silver ionization are used in different countries. One of them, the ‘Tarn-Pure System’ (Liu et al. 1998; Stout and Yu 2003; Zhang et al. 2008) lead to an effective decontamination of many Legionella-contaminated water systems. The ‘Tarn-Pure System’ induced concentrations between 20–40 μg l−1 silver and 200–400 μg l−1 copper. Stout and Yu (2003) reported the effect of the ‘Tarn-Pure System’ in 16 hospitals. Prior to the installation, all the hospitals reported cases of hospital-acquired legionnaire’s disease and 57% had attempted other disinfection methods such as superheating or hyperchlorination (Stout and Yu 2003). After 5 years, no cases of hospital-acquired legionnaires disease have be reported in any of these hospitals. Similar results were reported in other studies (Stout et al. 1998; Lin et al. 2002). It is noteworthy that the decontamination was achieved after at least 2 month of silver exposure (Liu et al. 1994).

To investigate the effect of silver ions on Legionella, we exposed extracellularly and intracellularly grown bacteria to silver ions in vitro. Furthermore, we investigated the possible protective effect of biofilms, in a biofilm model containing legionellae and other heterotrophic bacteria.

Material and methods

Antimicrobial effect of silver ions on extracellular grown bacteria

The antimicrobial effect of silver was tested on a selection of Gram-positive and Gram-negative bacteria in comparative experiments. We used type strains of the Gram-negative Escherichia coli ATCC 11775, the Gram-positive Staphylococcus aureus ATCC 29213 and the water-relevant strain Pseudomonas aeruginosa ATCC 27857 as model organisms. Furthermore, two resistant strains of Enterobacter cloacae (Davis et al. 2005) were used, as well as three Gram-negative strains isolated from an established biofilm model (Shingomonas paucimobilis, Brevundimonas diminuta and Chryseobacterium meningosepticum). All bacteria were incubated overnight on blood agar. In further experiments, we used Legionella pneumophila serogroup 1 strain Corby (serogroup 1, monoclonal subtype Knoxville, sequence type 51) and Corby mutants lacking the Mip, Pad, CsrA proteins, or the virulence-associated epitope of the lipopolysaccharide (LPS) recognized by monoclonal antibody (MAb) (Table 1) (Lück et al. 2001; Edwards et al. 2010). Legionella pneumophila was cultivated for 48 h on Legionella medium BCYE agar obtained from Oxoid, Wesel, Germany.

Table 1.   MIC of different strains of bacteria. All readings in μg l−1 AgNO3 (63·5% atomic silver)
Bacterial strainMIC 24 h→ no culturable bacteriaMIC 24 h→ reduction 3 logarithmic magnitudesReferences
Legionella pneumophila strain Corby, serogroup 1, MAb type Knoxville, sequence type 51, (extracellular)6432–64Jepras et al. (1985)
Leg. pneumophila strain serogroup 1, MAb type OLDA, sequence type 1)64n.d.This study (isolated from biofilm in the microcosm)
Staphylococcus aureus6416ATCC 29213
Pseudomonas aeruginosa1616ATCC 27857
Escherichia coli44ATCC 11775
Enterobacter cloacae AG1151>6432Davis et al. (2005)
Ent. cloacae AG703>6432Davis et al. (2005)
Leg. pneumophila CP7 mutant6432–64Steudel (2001)
Leg. pneumophila CrsA mutant6432–64Igel (2007)
Leg. pneumophila Mip mutant6432–64Igel (2007)
Leg. pneumophila strain Corby (intracellular)40962048This study
Sphingomonas paucimobilis (isolated from biofilm)6416This study (isolated from biofilm in the microcosm)
Brevundimonas diminuta (isolated from biofilm)6416This study (isolated from biofilm in the microcosm)
Chryseobacterium meningosepticum (isolated from biofilm)6416This study (isolated from biofilm in the microcosm)

For the detection of the minimal inhibitory concentration (MIC), c. 105 bacterial cells were incubated at concentrations of 0, 4, 8, 16, 32 and 64 μg l−1 of AgNO3 in distilled water corresponding to 0, 22.5, 45, 90, 180 and 370 μmol l−1. These suspensions were incubated at room temperature for 1, 3, 6 and 24 h and plated on blood agar, R2A agar, or on the Legionella medium BCYE agar (Oxoid).

Antimicrobial effect on intracellular grown Legionella pneumophila

Intracellularly grown Legionella cells (strain Corby) were harvested from infected Acanthamoeba castellanii cultures. The A. castellanii cells were incubated for at least 48 h in the ATCC 712 (PYE) medium prior to infection. Amoebae were infected with bacteria at a multiplicity of infection of 10. Adhesion was allowed for 90 min, after which the remaining extracellular Legionella were killed with gentamycin (80 μg l−1, 60 min). The co-cultures were further incubated for 3 h and then again washed three times with fresh medium. After a final incubation of 16 h, the infected amoebae were disrupted by three cycles of freezing and thawing (Lück et al. 1998). Immunofluorescence staining using the FITC labelled monoclonal antibody against Leg. pneumophila (Bio-Rad, Munich, Germany) showed that at least 50% of the amoeba cells contained intracellular legionellae (Fig. 1).

Figure 1.

 Intracellular legionellae in Acanthamoeba castellanii cells stained with an FITC labelled monoclonal antibody against Legionella pneumophila (Bio-Rad, Munich, Germany).

The MIC of the intracellular grown Legionella was determined with silver concentrations of 0, 16, 64, 256, 1024 and 2048 μg l−1 of AgNO3 as described for the extracellular bacteria. To exclude a possible unspecific absorption effect of the silver ions to amoebal lysate, lysed noninfected amoeba cells were added to fresh agar-grown Legionella. This control sample mixture was treated as the Legionella-infected co-culture.

Antimicrobial effect on bacteria grown in artificial biofilms

In these experiments, we tested the influence of silver on biofilm-grown heterotrophic flora as well as on biofilm-associated Leg. pneumophila. To establish this microcosm model, we incubated stagnant drinking water from a large building water conduit at 37°C with gentle agitation. Approximately 1 l quantities were cultivated. Once a week, half of the water of the microcosms was exchanged with new water from the same drinking water system. After 2–3 weeks, a constant level of the heterotrophic bacteria and Leg. pneumophila, serogroup 1, MAbtype OLDA, sequence type 1 (Borchardt et al. 2008) was attained in the watery phase. Total viable count of heterotrophic bacteria and Legionella counts were determined on R2A agar and GVPC agar (Oxoid), respectively.

Tissue culture inserts (NUNC, cell culture inserts, Anopore membrane) were added to these microcosms, and after 2–3 weeks, a stable biofilm developed on this solid surface. The MIC for silver of the biofilm-grown heterotrophic bacteria and Legionella was determined by submersion the biofilm-covered Anopore membrane inserts into solutions containing silver nitrate at defined concentrations for defined incubation times. To quantify the growth on Anopore membranes, they were homogenized in 1 ml of distilled water. Aliquots were plated on blood- and GVPC–agars, and the colony forming unit (CFU) was counted after appropriate incubation times.

The membranes were also stained with DAPI or an FITC labelled monoclonal antibody against Leg. pneumophila (Bio-Rad) and Evans Blue. Stained membranes were examined using epifluorescence techniques (Axioskop; Zeiss, Oberkochen, Germany). A representative example is shown in Fig. 2.

Figure 2.

 NUNC Anopore membrane stained with an FITC labelled monoclonal antibody against Legionella pneumophila (Bio-Rad) and Evans Blue (Magnification 1000×). Green fluorescending Legionella bacteria within amoeba. Heterotrophic bacteria are stained in red.

Detection of Protozoan species in the artificial Biofilm model

The presence of protozoa host cells in the microcosm was determined by culturing samples directly from the water supply. Briefly, samples were inoculated onto non-nutrient agar plates coated with a layer of living Ecoli cells and incubated at room temperature. Plates were examined every second day for at least 2 weeks for the presence of amoebae. The species of the amoeba were determined from the isolates as well as directly from the microcosm water samples by using 18S-rRNA PCR (Schroeder et al. 2001). Sequencing was performed by standard techniques using a 377 sequencing apparatus. The sequences were aligned by the blast algorithm of the NCBI-gene base and submitted under the NCBI-Accession-number FR820592. In several points of time, DNA of Hartmanella vermiformis was detected within the microcosm water at several occasions during the experiment. We were unable to detect other amoebal species in this water.


Antimicrobial influence of silver to extracellular grown bacteria

As shown in Table 1, the growth of all tested bacterial strains was inhibited at a concentration of 16–64 μg l−1 AgNO3 after 24 h of incubation at room temperature. Similarly, three hours of incubation resulted in a reduction of 2–4 logarithmic magnitudes in viable counts. Similar results were observed at 37°C.

In general, it appears that the E. coli strain was slightly more sensitive to silver than the other strains. We also tested two Ent. cloacae strains that carried a large plasmid conferring resistance to silver (Davis et al. 2005; Silver et al. 2006). The MIC of these two strains after 3 h of exposure was 32–64 μg l−1 AgNO3, and 8–16 μg l−1 after 24 h of exposure. Thus, it appears that they are less resistant than expected. As the MIC values were not mentioned in the original publication, a comparison with the original data cannot be performed.

All agar-grown (extracellular) Legionella strains had an MIC of 64 μg l−1 AgNO3 (Table 1). This was similar for the Legionella mutants lacking the surface proteins mip, pad, CsrA and the LPS epitope recognized by MAb 3/1 and is in the range of the other bacteria like E. coli and Staph. aureus.

Antimicrobial influence of silver to intracellularly grown Legionella

In contrast to agar-grown legionellae, the intracellularly grown bacteria were significantly less sensitive. The MIC after 24-h exposure to silver was 4096 μg l−1 AgNO3. Even after prolonged exposure of 48 h, the MIC was 2048 μg l−1 AgNO3. There was no difference in the MIC for Legionella bacteria mixed with a lysate from uninfected amoebae and in the MIC for agar-grown bacteria diluted in distilled water (both 64 μg l−1 AgNO3) indicating that amoebal lysate offers no unspecific absorbing effect on silver ions as (data not shown).

In vitro grown biofilm model

The established biofilm model was repeatedly used in different experiments, and it showed an excellent reproducibility. As shown in Fig. 3, after 3–4 weeks, levels of 104–105 ml−1 heterotrophic bacteria and 100–500 ml−1Legionella could be constantly detected in the water phase. Similar bacterial counts could be detected on the Anopore membrane in the biofilm per cm2.

Figure 3.

 TVC and Legionella-count (mean of three experiments) on the developed artificial microcosms with the biofilm grown on membrane. inline image TVC (CFU ml−1)inline image TVC-biofilm (CFU cm−2) inline image Legionella and inline image Legionella-biofilm (CFU cm−2)

The MIC of the biofilm-grown flora was also determined. After 24 h, all the non-Legionella bacteria were killed at a concentration of 4–16 μg l−1 AgNO3. Thus, three Gram-negative strains grow in the established biofilm model (S. paucimobilis, B. diminuta and C. meningosepticum) were as sensitive as the isolated grown on agar. So, we could not observe a protective effect of the extra polymer substance in our biofilm model. However, biofilm-associated Legionella are less sensitive as expected. After 24 h, there was no effect on these Legionella with all the different concentration of silver.


There are currently several methods for reducing bacterial growth in aquatic systems. Of these, hypochlorination and heating are the most commonly used method; however, silver/cooper-ionization systems are also in use in different countries to reduce the contamination in these systems (Stout and Yu 2003).

In this study, we elucidated the action of silver ions on different bacteria including extra- and intra-cellular grown Legionella and biofilm-associated flora. After 24 h, all tested non-Legionella bacterial strains were killed at a concentration about 16 μg l−1 AgNO3. After 3 h, there was a reduction of 2–4 logarithmic magnitudes. The MICs are similar to those reported for other bacteria in earlier studies for silver. A Japanese group found a reduction of 4 logs after 3 h with a concentration of 100 μg Ag l−1 in the viable counts of Ps. aeruginosa. Using the same concentration, E. coli was completely killed within 3 h. After 8 h, all other strains were completely inactivated at 100 μg Ag l−1 (Hwang et al. 2007). In another study, the MIC of silver for Staph. aureus was 15.6 mg l−1 and for E. coli, 7.8 mg l−1 (Egger et al. 2009). These levels are significantly higher than in this study and might be explained by the use of medium rich in C-recourses like glucose that binds the silver ions. To exclude such interference, we used water to prepare the silver solution for the determination of MIC.

Legionella mutants showed no difference in MICs of silver in comparison with the wild type Corby strains. The mip- and pad-mutants are defective in surface proteins involved in the uptake into host cells (Cianciotto and Fields 1992; Steudel and Lück 2001). It was supposed that the chemicals inactivation with precipitating silver ions might influence the adherence. However, we could not find any data supporting this assumption. Both proteins are rich in SH groups cysteine that are supposed to be a target of silver ions. The LPS mutant TF3-1 has lost the virulence-associated epitope recognized by MAb 3-1 making the LPS less hydrophobic. In general, the role of the Legionella LPS in the course of the infection is not completely understood (Seeger et al. 2010). In the fourth mutant, the regulator LetA/LetS is defective, which leads to a reduced multiplication in different host cells. It might be assumed that surface proteins or other structures are blocked or inactivated by silver ions. This seems to be a sum of effects that could not be attributed to a single surface protein or other factor.

In comparison with extracellularly, agar-grown Legionella, the intracellularly grown Legionella were significantly less sensitive against silver (MIC > 2048 μg l−1 AgNO3). This effect is known from other studies, for biofilm including Legionella (Percival et al. 2007) and from other bactericidal substances like antibiotics (Hwang et al. 2006). The Legionella cell proliferates in host cells such as Acanthamoeba or other protozoa and acquires natural resistance against toxic substances (Barker et al. 1995). As intracellularly grown bacteria stain differently with Giemenez staining, it is supposed that the intracellular growth of these bacteria leads changes in the surface structure so that diffusion or other processes are changed. This phenomenon needs to be investigated in further studies.

The microcosms developed here proved to be a good model for testing the antimicrobial effect of silver under simulated natural conditions. To characterize the biofilm model in more detail, we isolated and characterized the predominant strains from the biofilm model belonging to different bacterial species. The biofilm growing in the model is composed mainly of water-relevant bacteria such as B. diminuta, C. meningosepticum and S. paucimobilis under simulated natural conditions. These three species represented also the majority of culturable bacteria in samples taken directly from the water outlet. Thus, we believe that both the Legionella and heterotrophic flora in the model system are comparable to that in the building water supply. The biofilm-associated heterotrophic flora was similarly sensitive for silver as the other tested water-relevant strains like Ps. aeruginosa.

In a previous study, the effect of silver on the biofilm formation in water distribution systems was tested at concentrations of 100 μg l−1. No significant reduction was reported. This was partially attributed to the ‘consumption’ of silver by the biofilm matrix (Silvestry-Rodriguez et al. 2008). They also showed that low concentrations of silver ions were unsuitable for the treatment of biofilm infections caused by Staphylococcus epidermidis (Chaw et al. 2005). In another study, the MIC and the MBEC (minimum biofilm eradication concentration) for silver were nearly the same as in our study for E. coli, Staph. aureus and Ps. aeruginosa (Harrison et al. 2004).

It is assumed that in our biofilm/microcosm model, the Legionella proliferated in the available amoeba H. vermiformis. Again these intracellular grown Legionella are less sensitive compared with their extracellularly grown counterparts (Fig. 3).

Practical application of silver (and copper) in the ionization ‘Tarn-Pure System’ induced a concentration of 20–40 μg l−1 silver and 200–400 μg l−1 copper. Our results confirm that such concentrations of silver are effective reducing the contamination in the water system. In our experiments, most of the tested strains were killed at a concentration of 32–64 μg l−1.

The advantages of silver against the conventional disinfection methods like superheating and hypochlorination are considerable (Stout and Yu 2003). Silver is easy to dose and is active directly in the system. Conventional procedures like superheating to a temperature of at least 80°C and hypochlorination using a concentration of 20 mg l−1 chlorine dioxide for 10 h or a concentration of 10 mg l−1 sodium hypochlorite for 1 h are essential (Blanc et al. 2005), and they are very labour intensive and expensive. After the decontamination of the tapping, the system might re-contaminate because there is a biological potential in the whole system. With the ‘Tarn-Pure System’, there is a constant concentration of silver in the water system with a very low demand of maintenance and the recontamination is deferred because the silver and copper ions are in the whole system, and so there is a lower number of living bacterial cells (Yu et al. 1993).

Up to now, the long-term application of silver ions in technical water supply systems has not promoted any development of resistance. These is agreement with our observation that in the established microcosm model that was kept in a concentration of silver of 80 μg l−1 for 12 weeks, we were unable to isolate any naturally occurring silver-resistant bacteria.

The antimicrobial effect of silver could be successfully used for practical applications in various technical and industrial water conduits such as cooling towers with circulating water. Another application could be swimming – and whirlpools very often contaminated with Legionella and other bacterial strains – here, the silver ions could reduce the overall biological pollution (Martinelli et al. 2001). To get an antimicrobial concentration of silver ions, they could be introduced to the water by a permanent fixture at the water outlet feeding in the water systems. Alternatively, the silver ions could be administered on a carrier, for example, a synthetic matrix that delivers the ions permanently to the circulating water.

Our data show that extra- and intra-cellular micro-organisms and biofilm-grown bacteria exhibit a different degree of sensitivity to silver ions. It has been reported that intracellularly grown legionellae exhibit a different profile of outer membrane proteins and fatty acids (Barker et al. 1993). However, the exact differences between legionellae grown in artificial media and amoebae need to be elucidated. In this regards, the different impact to extra- and susceptible intra-cellular grown legionellae have to be taken into consideration. It is necessary to keep in mind that the complex ecology in water systems has many factors that might be influenced by silver ions. These complex interactions might explain why the effect of silver ions in water systems is often detected after delay of exposure (Stout and Yu 2003).


We thank Kirsten Hermann for technical assistance. We also thank Dr I. J. Davis from the University College London Hospitals NHS Trust who supplied us with the silver-resistant Enterobacter strains. Mip mutant and the letA mutant were kindly supplied by Prof. Michael Steinert, University Braunschweig, and Prof. Reinhard Marre, University of Ulm. This study was supported by a grant of the German Federal Ministry of Education and science BMBF (02 WT 0665).