Phosphate drinking water softeners promote Legionella growth

The main aspect of drinking water softening is to prevent scale formation and pipe corrosion and to reduce the leaching of copper and lead from pipe materials. Another reason for drinking water softening is also consumer convenience. In many households and also in some central drinking water treatment plants, drinking water is softened by using various softening techniques (precipitation, ion exchange, or membrane separation) (Casale 2001). Phosphate salts and polyphosphate solutions are often used as drinking water softeners (Casale 2001; Batté et al. 2003a).

According to SIST EN 1212 (2005), the treatment dose of phosphates for drinking water treatment should not exceed 5 mg P₂O₅ per liter of drinking water. The Slovenian National Institute of Public Health (NIJZ 2019) and the German Federal Environmental Agency – UBA (UBA 2020) also recommend up to 2.2 mg of phosphates per liter as P (i.e., 5.04 mg as P₂O₅ or 6.74 mg as PO₄). For corrosion inhibition and scale prevention in drinking water systems, EPA (2016) has issued similar guidelines with somewhat lower concentrations of residual phosphates up to 1.0 mg/L as P (i.e., 3.0 mg/L as PO₄). However, according to EPA, a 2- to 3-fold higher initial dose is recommended if so required, followed by a lower maintenance dose.

The EU Directive on the quality of water intended for human consumption (Directive (EU) 2020/2184) does not regulate phosphate concentration or drinking water softening techniques as such. Consequently, the phosphate (or phosphorus) concentration in drinking water is not controlled. Therefore, it is not surprising that the phosphate concentration in drinking water samples in Slovenia, where phosphate softeners are used on a large scale, varies widely, from 0.2 to 24.6 mg PO₄/L (Jereb et al. 2017).

When it comes to drinking water quality, the control of not only chemical hazards but also microbiological hazards is of great importance. Management of biofouling in the drinking water distribution system (DWDS) is an important strategy to ensure safe drinking water. Phosphate treatment of drinking water carries the risk of promoting the growth of microorganisms, which could be described as a side effect of chemical softening. According to various researchers (Miettinen et al. 1997; Sathasivan et al. 1997; Sathasivan & Ohgaki 1999; Batté et al. 2003a; Park et al. 2008; Wen et al. 2014), the availability of phosphorus in the water environment may be crucial for microbial growth. Therefore, the addition of a scarce micronutrient (such as phosphorus) to water as part of drinking water treatment could contribute significantly to the growth of various bacterial species. One of them is Legionella pneumophila, pathogenic bacteria commonly found in water supply systems (Toplitsch et al. 2018). Legionella is a ubiquitous microorganism found in the natural aquatic environment (Bonetta & Bonetta 2020), where it does not normally pose a health risk due to its low concentrations. A health risk arises when the bacteria occur in high concentrations in DWDS (Barskey et al. 2020; ECDC 2021), where they can cause Legionnaires' disease and/or Pontiac fever (Prussin et al. 2017). Among the various members of the genus Legionella, L. pneumophila is the most confirmed pathogen (Harrison et al. 2009; ECDC 2021).

The objective of the present study was to investigate the effects of phosphate-based drinking water softeners on the growth and reproduction of L. pneumophila to provide additional data for informed decision making in case of using phosphate-based drinking water softeners.

The standard strain of Legionella pneumophila subsp. pneumophila ATCC 33152 (acquired from the Czech Collection of Microorganisms – CCM) was used. For the study of the growth curve under two different phosphate concentrations, the bacteria were incubated at 36 °C (SIST EN ISO 11731 2017) in microtiter plates. To ensure reproducibility, the experiment was performed with liquid yeast extract broth (Sigma-Aldrich, 92144-500G-F) containing 400 mg/L L-cysteine HCl (Legionella BCYE α-growth supplement, Biolife, 423210). An overnight culture with an initial bacterial cell concentration of 6.06 (±0.23) log CFU/mL was used as the starter concentration in all experiments. Since bacteria of the genus Legionella are known to have slower growth and longer regeneration time (SIST EN ISO 11731 2017), bacterial growth was monitored up to 144 h after the start of the experiment. Bacterial concentration was monitored at time 0 and after 24, 48, 72, 96, 120, and 144 h. After the incubation period, the tenfold dilution approach was performed (SIST EN ISO 8199 2007) and the method of counting bacterial colonies on solid medium (Legionella BCYE Agar Base, Biolife, 4015822) was used. Three experimental replicates were performed for each phosphate concentration, as well as for the control, resulting in a total of seven sample replicates. Due to an error in the incubation process, one sample with the highest phosphate concentration was withdrawn from further analysis after 48 h of incubation. Thus, a total of 146 samples were analyzed. Results are presented in logarithmic values as log CFU/mL for each concentration and incubation time.

A commercial drinking water softener with a nominal concentration of 7 ± 0.2% P₂O₅ was used to observe the effect of added phosphates on L. pneumophila growth. Two stock solutions with concentrations of 1 and 10 g/L were prepared from the commercial softener (basic phosphate solutions A and B). In both cases, 0.24 mL of phosphate stock solution A or B (or distilled water in the control) was added to a total volume of 48.6 mL of medium containing liquid yeast extract broth, cysteine, and bacterial starter culture. The pH of the medium did not change. The control in the experiment consisted of the bacteria L. pneumophila in yeast extract without added phosphates.

To determine the phosphate concentration in each sample, equal amounts of stock solutions A and B were added to 48.6 mL of distilled water. The phosphate concentration in the samples was measured according to the standard ISO (SIST EN ISO 6878 2004). The method of chemical detection is described in detail in Jereb et al. (2017). A phosphate solution was prepared with a concentration of 3.6 mg PO₄/L and 39.5 mg PO₄/L. With the lower concentration chosen (3.6 mg PO₄/L), we aimed to achieve the average concentrations of added phosphate in Slovenian drinking water systems reported in previous research (Jereb et al. 2017). A tenfold higher concentration was prepared for a higher test concentration.

The highest difference in the number of bacteria in the samples with phosphates compared with the control (1.76 log CFU/mL) was observed at time t₁₂₀ at lower phosphate concentration (3.6 mg PO₄/L). In the time intervals 72–120 h after incubation, the difference in the number of organisms in the samples with phosphates compared with the control was 1.40 log CFU/mL or more for both phosphate concentrations. Even after 144 h of incubation, the number of bacteria was still significantly higher in the case of added phosphates (7.82 log CFU/mL and 7.58 log CFU/mL) compared with the control (6.63 log CFU/mL), with a difference of 1.19 and 0.95 log CFU/mL, respectively.

A nonparametric Mann–Whitney test was used to calculate the differences in the number of CFUs between the two phosphate concentrations tested and the control group at different time points (Table 1). At the beginning of the experiment (t₀), there were no statistically significant differences between the groups (all p > 0.05). Over time, the groups with added phosphates differed significantly from the control group (all p < 0.01) in the number of CFUs. The increased growth confirms the positive effect of phosphates on the growth of L. pneumophila subsp. pneumophila ATCC 33152. However, there were no statistically significant differences in the number of CFUs observed when comparing high and low phosphate concentrations (except at t₄₈, where the difference is at a level of 0.05). The later results suggest no additional growth potential when phosphate concentration increases.

The present study confirms the effect of added phosphates on Legionella growth under in vitro conditions and similar impact on growth could be expected in DWDS. Stimulation of bacterial growth was achieved at the proposed treatment doses of phosphates for drinking water. Drinking water softeners or phosphate-based corrosion inhibitors (ortho- or polyphosphates) could therefore be an important source of phosphorus in DWDS. The number of Legionella increased significantly compared with the control at a concentration of 3.6 mg PO₄/L, while at a higher concentration of added phosphorus (c = 39.5 mg PO₄/L), the growth remained the same as at a lower concentration. On average, 1.19–1.28 log CFU/mL higher growth was observed as compared with the control group.

Some other researchers also emphasize the importance of inorganic nutrients in drinking water as a key factor in controlling the growth of microorganisms, especially in aquatic environments with high levels of available carbon. Batté et al. (2003a) reported that the addition of phosphorus affects the growth of bacteria of the genus Legionella. Sathasivan et al. (1997) documented increased bacterial growth when phosphorus was added to the water medium. Even when the carbon concentration in the water was high, no increase in microbial growth was observed when the phosphorus content was inadequate. Similarly, Park et al. (2008) demonstrated that the addition of phosphorus had no effect on microbial growth at low concentrations of available carbon, while at high carbon concentrations the addition of phosphorus significantly stimulated microbial growth. Stimulation of growth was measured by increased number (CFU/cm²) and weight (μg/cm²), both again suggesting the role of available phosphorus in bacterial proliferation. Sathasivan & Ohgaki (1999) also identified phosphorus (0.3 mg PO₄/L) as the major growth-limiting factor in samples from the Tokyo DWDS. Miettinen et al. (1997) confirmed comparable results in drinking water samples from Finland. Despite the relatively high concentrations of naturally occurring organic carbon in the water, they observed a noticeable increase in the growth of microorganisms only when phosphorus was added. According to the study by Polanska et al. (2005), phosphorus in Belgian waters (which are also rich in carbon) also proved to be an important limiting growth factor for microorganisms in their drinking water. A similar conclusion was reported by Wen et al. (2014) for northeast China in the city of Harbin.

It can be concluded that the availability of phosphorus may be a crucial limiting factor for bacterial growth in DWDS. Since various phosphorus formulations (from phosphoric (V) acid to various ortho- and polyphosphates) are commonly used to soften drinking water or to inhibit corrosion (Casale 2001; Batté et al. 2003a), additional caution is needed. This is especially important since the effects of phosphorus on Legionella proliferation are often ignored. Although researchers reported high phosphorus (phosphate) concentrations in their studied DWDS, they did not attribute the proliferation of microorganisms to phosphates (phosphorus), but to a lower disinfection effect. For example, Girolamini et al. (2019) reported phosphorus concentrations in observed hospital hot water pipes ranging from 1.68 to 3.52 mg/L as P₂O₅. However, in their study, phosphorus was measured solely to monitor the maintenance of anti-scale and corrosion treatment. Although a high percentage of positive samples (95%) and high Legionella concentrations were detected in the observed DWDS, the authors did not investigate the cause of Legionella proliferation. Similarly, Song et al. (2021) studied the effect of phosphate in DWDS on reducing the biocidal effect of copper ions (phosphate as a corrosion inhibitor) but neglected its growth stimulation on Legionella. Cullom et al. (2020) also mentioned phosphorus in DWDS only as an inhibitor of copper ions. Likewise, they did not correlate higher numbers of microorganisms with growth stimulation by phosphate, but rather with its reduction of copper ions (biocidal effect) due to the reduction of leaching from copper pipes. Martin et al. (2020) reached the same conclusions in the case of a high concentration (CFU) of Legionella in Flint DWDS.

Most microorganisms in drinking water are heterotroph and require adequate concentrations of organic and inorganic nutrients for their proliferation. One of these nutrients is phosphorus (or phosphates). Phosphorus is an essential element required for cell growth as part of DNA/RNA, phospholipids, and many enzymes, and therefore plays an important role in cell metabolism, some structural functions, and cell response to stress (Brown & Kornberg 2008). Because the bacterial system of phosphate transport across membranes is not efficient when the phosphate concentration in water is low but becomes energy efficient when the concentration increases, typical concentrations of phosphate corrosion control agents or drinking water softeners greatly exceed the nutrient requirements of bacterial cells and can promote bacterial proliferation (Batté et al. 2003b).

According to the above studies, phosphorus is often neglected as an important nutrient in DWDS. Even more, it seems that the most important nutrients needed for microbial growth in the drinking water system (organic carbon and phosphorus) interact according to the principle of Liebig's law of minimum. Since phosphorus concentrations are rather low in the vast majority of drinking water sources (Batté et al. 2003a), phosphate-based drinking water softeners could be an important source of additional phosphorus. In such cases, phosphorus control may be one of the most important measures to control the occurrence of biofilm and microbial regrowth in DWDS, especially in waters rich in other nutrients.

It should be emphasized that the presented experiment was performed under laboratory conditions in yeast extract broth with a standardized strain of Legionella pneumophila subsp. pneumophila ATCC 33152 without other microorganisms present. L. pneumophila was added at higher concentrations than normally found in the water supply systems, so it is possible that the bacteria would react differently under real (in vivo) conditions. To confirm the effect of phosphate-based drinking water softeners on Legionella growth in DWDS, further experiments should be conducted in the future using different water samples with different (controlled) levels of assimilable organic carbon and phosphorus. Nevertheless, the results of the experiment in culture media indicated that the added phosphorus significantly affected the growth of L. pneumophila. Therefore, a similar effect could be expected in DWDS when phosphates are used as drinking water softeners.

In the future, any drinking water treatment that could promote microbial growth in DWDS should be avoided or used judiciously. In addition, it is strongly recommended that at least a chemical analysis of the raw water be performed before phosphate-based softeners or corrosion inhibitors are used (and when the phosphorus concentration is low and the concentration of other nutrients high, phosphate softeners should be avoided).

Since the number (concentration) of L. pneumophila present in drinking water is a key factor in the occurrence of infections, any promotion of Legionella growth in DWDS poses an unnecessary health risk. The results of this study provide a deeper understanding of the previously neglected risk of bacterial growth when drinking water is treated with phosphates as water softeners or corrosion inhibitors. Future risk assessment of phosphate-based drinking water softeners and/or corrosion inhibitors should therefore also consider the indirect effect of promoting bacterial growth. Especially in drinking water rich in other nutrients, the (non)availability of phosphorus may be an important growth-limiting factor. Controlling phosphorus could therefore be one of the most important measures to control the biofouling and microbial regrowth in DWDS.

For the English proofreading, Tina Levec, an English language instructor, is gratefully acknowledged. Additionally, this work is a part of Ph.D. thesis which was supported by a grant (Jereb G.; 322/2012) from the European Social Fund (ESF) and the Ministry of Education, Science and Sport of the Republic of Slovenia.

All authors participated in the design of the study. Gregor Jereb conducted the literature review, prepared the samples, performed all analyses, and conceptualized the manuscript. Martina Oder was involved in the laboratory analyses. Borut Poljšak, Ivan Eržen, and Martina Oder contributed to the interpretation of results and manuscript revisions. All authors reviewed and approved the final version of the manuscript.

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

The authors declare there is no conflict.