Environ Sci Pollut Res (2013) 20:5534–5544 DOI 10.1007/s11356-013-1557-5
RESEARCH ARTICLE
Microbial and physicochemical parameters associated with Legionella contamination in hot water recirculation systems Alejandra Serrano-Suárez & Jordi Dellundé & Humbert Salvadó & Sílvia Cervero-Aragó & Javier Méndez & Oriol Canals & Silvia Blanco & Antoni Arcas & Rosa Araujo
Received: 14 December 2012 / Accepted: 5 February 2013 / Published online: 23 February 2013 # Springer-Verlag Berlin Heidelberg 2013
Abstract Hot water recirculation systems (HWRS) in hotels and nursing homes, which are common in countries such as Spain, have been related to outbreaks of legionellosis. To establish the relationships of microbial and physicochemical parameters, especially protozoa, with the occurrence of Legionella in HWRS, 231 samples from hotels and nursing homes were analysed for Legionella, protozoa, heterotrophic plate counts (HPC) at 22 and 37 °C, Pseudomonas, metals, temperature and others. Legionella pneumophila was the dominant species isolated, and 22 % were sg. 1. The sampling Responsible editor: Robert Duran A. Serrano-Suárez : S. Cervero-Aragó : J. Méndez : R. Araujo (*) Departament de Microbiologia, Facultat de Biologia, Universitat de Barcelona, Avda. Diagonal 643, 08028 Barcelona, Spain e-mail:
[email protected] H. Salvadó : O. Canals Departament de Biologia Animal, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain A. Arcas Departament d’Estadística, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain J. Dellundé J.D.Assessors, Malgrat de Mar( Barcelona, Spain S. Blanco Servei de Microbiologia, Hospital Universitari Germans Trias i Pujol, Departament de Genètica i Microbiologia, Universitat Autònoma de Barcelona, Barcelona, Spain Present Address: S. Blanco Department of Microbial, Cellular, and Molecular Biology, College of Natural Sciences, Addis Ababa University, Addis Ababa, Ethiopia
method became particularly important in order to define which factors were involved on the occurrence of Legionella. Results showed that the bacteria and the accompanying microbiota were more abundant in the first flush water whose temperature was lower. The bacteria occurred in those samples with high HPC and were inversely correlated with high temperatures. Multivariate regression showed that a concentration above 1× 105 CFU/100 mL of HPC at 37 °C, Fe above 0.095 ppm and the presence of protozoa increased significantly the risk of Legionella colonization, while univariant regression showed that the presence of Cu above 0.76 ppm and temperature above 55 °C diminished it. Therefore, to reduce the risk associated with Legionella occurrence in HWRS these parameters should be taken into consideration. Keywords Legionella . Protozoa . Hot water . HPC . Temperature . Metals
Introduction Legionella pneumophila is an aquatic bacterium that was first linked with pneumonia in 1976 (McDade et al. 1977). Since then, many cases of legionellosis caused by different Legionella species and mostly related to contaminate manmade water systems have been reported worldwide. These days, many countries have guidelines in place to prevent the growth of the bacteria and reduce the risk of outbreaks and transmission of Legionella spp. Despite this, new cases of legionellosis are reported every year. In 2009, 3,522 cases were reported in the USA with an incidence rate of 1.15 (CDC 2011). In Europe, most cases were reported in France, Spain and Italy, with an annual mean of 1,409,
Environ Sci Pollut Res (2013) 20:5534–5544
1,214 and 906, and incidence rates of 2.26, 2.76 and 1.54, respectively (Joseph et al. 2010). Legionella infections can occur sporadically and in outbreaks, but in both circumstances, determining the source of infection is difficult. A total of 501 outbreaks were reported in Spain between 1999 and 2009, 22 % of which were caused by sanitary hot water systems and 18 % by cooling towers or similar devices (Cano Portero et al. 2010). Even there is no information about the source of infection of most sporadic cases, it is likely that many of the cases could be related to sanitary hot water systems. Hot water in hotels, nursing homes and hospitals may be a source of infection since Legionella spp. are often present in this type of environment (Borella et al. 2005a; Mouchtouri et al. 2007; Wadowsky et al. 1982) and can spread through aerosols generated by showers and taps. Furthermore, in countries like Spain that have limited energy and water resources, hot water systems in hotels and hospitals are often designed as a recirculating system with a storage tank, which increases the risk of legionellosis (Borella et al. 2005a). These systems have several characteristics that may favour Legionella growth. Such characteristics include long pipes, dead ends and the intermittent use of hot water, which results in differences in flow, speed and temperature in each part of the circuit. In the environment, Legionella has various survival strategies, including multiplying in biofilms, where it can live alongside other bacteria such as Pseudomonas or multiply intracellularly in protozoan species such as Acanthamoeba, Hartmannella and Tetrahymena which feed and protect the bacteria (Abu Kwaik et al. 1998; Anand et al. 1983; Atlas 1999; Rowbotham 1980; Serrano-Suárez 2009; Taylor et al. 2009; Wadowsky et al. 1988). Moreover, Legionella density may vary in response to both the density and the composition of amoeba present which as well may play an important role on the reactivation of viable but nonculturable cells (Buse and Ashbolt 2011; Garcia et al. 2008) Thus, Legionella spp. have been isolated from many different water environments, especially from hot water systems and cooling towers, where the bacterium occurs within a wide range of physicochemical parameters. However, hot water recirculation systems have very different characteristics from cooling towers, so it is important to identify the ecological factors that favour the presence of Legionella in order to improve its control in this habitat. The relationship between Legionella and the physicochemical factors and heterotrophic bacteria in hot water systems in Italian and Greek hotels has been studied previously (Borella et al. 2005b; Leoni et al. 2005; Mouchtouri et al. 2007). However, few works have considered the effect of amoeba on the occurrence of Legionella in hot water systems (Habicht and Muller 1988; Lasheras et al. 2006), despite the fact that it has been well demonstrated,
5535
especially under laboratory conditions, that protozoa enhance Legionella growth and provide protection in harsh environments (Abu Kwaik et al. 1998; Anand et al. 1983; Rowbotham 1980; Taylor et al. 2009; van der Kooij et al. 2005; Wadowsky et al. 1988). The objective of the present study was to determine the environmental factors, including bacteria and protozoa, that may affect the distribution and abundance of Legionella spp. in hot water recirculation systems with storage tanks in hotels and nursing homes. The surveillance was carried out using two sampling methods to better characterize the Legionella environment. All of the samples were analysed for Legionella and related factors. Characterizing these niches that support Legionella will help develop more effective control measures.
Material and methods Sampling analysis A total of 231 samples from hot water continuous circulation systems with storage tanks were collected randomly from 30 hotels and nursing homes in Catalonia. All buildings in the study received water from the drinking water supplies with an average free chlorine content of 0.5 ppm and very low levels of bacterial contamination, which is in accordance with the Spanish regulations for this type of water (Boletín Oficial del Estado 2003a, b). Two collecting methods were used A and B. (A) One hundred and seven samples were obtained in compliance with Spanish regulations (Boletín Oficial del Estado 2003a); in brief, a sample of 100 mL of the first water to flow from the tap was collected and approximately 3 min later, when the water had reached the maximum temperature, the rest of the sample— made up to 1 L—was added to the sample container. (B) One hundred and twenty-four samples were collected using a different sampling method; briefly, 1 L of the first flow was taken corresponding to proximal water, and an additional 1 L was collected after letting the water run for approximately 3 min, corresponding to the hottest water in the circuit, distal water. Water samples were collected using sterile plastic bottles containing 5 mL of 2 % sodium thiosulfate (Panreac, Spain). Samples were taken to the laboratory at 4 °C and processed a maximum of 24 h after collection (International Organization for Standardization 1998). Reference bacterial strain The reference strain used in this study, L. pneumophila subsp. pneumophila str. Philadelphia 1 (ATCC 33152) was cultured on BCYE agar (Oxoid LTD; Basingstoke, Hampshire, England) at 37 °C.
5536
Microbial parameters Heterotrophic bacterial counts (HPC) were determined in accordance with the ISO 6222:1999, standard on water quality; the enumeration of culturable micro-organisms: colony count by inoculation in a nutrient agar culture medium (International Organization for Standardization 2005; James et al. 1999). In summary, 100 μL of the direct sample and a tenfold diluted sample in Ringer 1/4 (Scharlau Chemie; Barcelona, Spain) were cultured in Plate Count Modified agar media (Scharlau Chemie; Barcelona, Spain) at 22 °C for 72 h and at 37 °C for 48 h, respectively. For the quantification of Pseudomonas sp., 500 μL of the filtration concentrate from the liter filtered for Legionella detection (described below) was cultured in Glutamate Starch Pseudomonas agar (Scharlau Chemie; Barcelona, Spain; Ribas et al. 2000). For the analysis of protozoa, 5 ml of water was taken from the bottom of each sampling container, after 2 h of sedimentation. Samples were centrifuged at 800×g for 5 min and 50 μl of the pellet was observed with an optical microscope. Subsequently, 500 μl of wheat broth medium was added, and it was incubated at room temperature under dark conditions to avoid algae proliferation. After that, suspensions were observed at day 2, 5, 10 and 15. Wheat broth medium was prepared using the following method (López-Ochoterena and Serrano-Limón 1991), 50 g of Triticum vulgare was boiled for 20 min in 1 L of distilled water, then filtered through a 0.45-μm porous size membrane (EZ-Pak® Membrane Filters; Millipore, Molsheim, France) and adjusted to 1 L by adding more distilled water. Finally 1.3 g of NaHPO4 12 H2O was added (Merck; Darmstadt, Germany) to the media before being sterilized by autoclave. Detection limit of the method is between 50 and 80 individuals per liter.
Detection of Legionella by culture Detection and enumeration of Legionella in water samples were done by culture on BCYE agar supplemented with GVPC (Oxoid Ltd, Oxoid LTD; Basingstoke, Hants, England) following the ISO 11731:1998 standard on water quality; detection and enumeration of Legionella (International Organization for Standardization 1998). This method involves the filtration of a 1-L sample through a 0.45-μm porous size nylon membrane (Filter HNWP Millipore; Ireland); the retained material is then suspended in 10 mL of Ringer 1/40 by vortexing for 2 min. Concentrates were cultured either directly and after two treatments: a thermal treatment of 50 °C for 30 min and an acid treatment in which 100 μL of acid buffer is added to 900 μL of the sample concentrate, as described in the ISO 11731:1998 guidelines. The recovery of L. pneumophila using this method was determined experimentally and had
Environ Sci Pollut Res (2013) 20:5534–5544
an approximate detection limit equivalent to 44 CFU/L (data not shown; Serrano-Suárez 2009). Legionella serology An average of between three and five colonies of positive samples for Legionella were tested using the Legionella agglutination test (Legionella latex test; Oxoid, Basingstoke, Hampshire, England) following the manufacturer’s instructions. This kit enables differentiation between L. pneumophila serogroup 1, L. pneumophila serogroups 2–14 and seven additional Legionella species. DNA extraction DNA extraction was performed by modifying the protocol described by Van Eys et al. (1989). Briefly, 1 mL of the concentrated water (as described above) was heated to 100 °C for 10 min and cooled for 10 min at−20 °C. A total of 0.015 g of the resin Molecular Biology Grade AG 501X8 (Bio-Rad) was then added to the sample and mixed with the vortex for 30 s. Finally, the sample was centrifuged for 5 min at 10.000×g. The supernatant was then recovered and stored at −20 °C for further analysis. Semi-nested PCR A semi-nested PCR was also used for the detection of Legionella by amplifying the 16S rRNA gene, which is conserved in all species of the genera. The primers used for the semi-nested PCR were based on the method used in a study by Jonas et al. (1995). They are JFP (5′AGGGTTGATAGGTTAAGAGC-3′) and JRP (5′CCAACAGCTAGTTGACATCG-3′) for the first PCR round, and a reverse inner primer was designed for the second round ARS (5′-TTCCACTACCCTCTCCCATA-3′) which were used to generate a 234-bp amplicon. Ten microliters of the DNA extraction in a total reaction volume of 50 μL was employed for the first PCR round, while 1 μL of the first PCR product was used in the second round. All amplifications were performed in 1× PCR buffer, 1.5 mM MgCl 2, and 0.2 mM of a dNTPs mixture, 0.04 U of AmpliTaq® DNA polymerase (Applied Biosystems; Foster City, CA, USA) and 1 μM of each primer. The reaction parameters were as follows: denaturation at 95 °C for 5 min, followed by 40 cycles for the first round and 35 cycles for the second of 5 min at 95 °C, 90 s at 57 °C and 1 min at 72 °C. The amplification concluded with a final extension period of 10 min at 72 °C (GenAmp® PCR System 9700; Applied Biosystems). Ten microliters of the amplified PCR product was run (100 V) on 2 % agarose gel and stained with ethidium bromide. The detection limit of the first round of PCR for L. pneumophila was 6.80×102 Genome copies
Environ Sci Pollut Res (2013) 20:5534–5544
(GC) for reaction, while the semi-nested PCR reached 6.80 × 10 0 GC for reaction (data not shown; SerranoSuárez 2009). Physical and chemical analyses Temperature was measured in situ during collection. Free chlorine concentration was determined using N, N-diethyl-pphenylenediamine sulfate colorimetric method (Boletín Oficial del Estado 2003b). The value of pH was determined using the Micro pH2000 (Crison Instruments S.A, Spain) and turbidity using the Ratio/xr (Hach Lange France S.A.S). Concentrations of Zn, Fe and Cu were measured by inductively coupled argon plasma spectroscopy with a detection limit of 0.025, 0.01 and 0.01 ppm, respectively. Total organic carbon (TOC) was measured by high-temperature catalytic oxidation (HTCO) with a detection limit of 1 ppm. Heavy metal concentrations and HTCO data were measured by Serveis Cientificotècnics in the University of Barcelona. Statistical analysis Descriptive statistics, Wilcoxon matched pairs test and the Mann–Whitney U test were carried out using the statistical package software Statgraphics Centurion XVI (StatPoint Technologies). The statistical package Stata 11 (StataCorp, LP.) was used to estimate odds ratios and logistic regression; Pearson’s and Spearman’s rank correlations were also used to analyse the correlation between parameters. Odd ratios in 2×2 tables were calculated using the Woolf’s approach, for purposes of comparison with logistic regression. All statistics were performed with the culture results of Legionella; otherwise, it is stated in the manuscript.
Results Hot water characteristics Over a period of 2 years, a total of 231 samples from hot water recirculation systems with storage tanks in 30 hotels and 3 nursing homes were analysed for the presence of Legionella and to determine the water quality parameters. Legionella was detected by PCR in 95 (41 %) samples but was isolated in only 64 (27 %) samples. Within this group, 77 % were identified as L. pneumophila serogroup 2–14, 22 % as L. pneumophila sg.1 and 1 % as Legionella spp. The accompanying microbiota was determined by HPC and cultivation of Pseudomonas and protozoa. The occurrence of HPC at 22 °C (84 %) was similar to HPC at 37 °C, as were the mean counts 5.68×104 CFU/100 mL and 7.26×104 CFU/100 mL, respectively. Pseudomonas was detected in 45 % of the total samples. Protozoa were observed directly and after culture in 29 % of the samples.
5537
Gymnamoebae were detected in 93 % of these samples, and only 26.8 % were flagellates and 3.6 % were ciliates. The analysis of the physicochemical parameters revealed that levels of pH, turbidity and TOC were similar in all hot water samples. Water temperature was the most variable parameter, ranging from 16.3 to 71.7 °C. Despite the fact that mains water is always chlorinated in Spain, free chlorine was not detected in most of the samples. An analysis of trace metals showed its presence in most of the samples at low concentrations. Their concentration could have been affected by a variety of factors: corrosion of the alloy and welding of the pipes, tanks and taps; different pipeline compositions; the age of the systems; and the mains water characteristics. Sampling effect The surveillance was carried out using two sampling methods, and the data are shown in Table 1. During the initial sampling stage, 107 samples were collected in accordance with Spanish law (Boletín Oficial del Estado 2003a). These samples were representative of the entire circuit because each sample was a mixture of the first flush and the water that had reached the maximum temperature (recorded temperature). Results are shown in Table 1 as mixed samples. Legionella spp. was isolated in 28 samples (26 %) and usually accompanied by abundant microbiota. Temperature acted as the main biocide in these samples. However, in most of them, the temperature was not high enough to maintain a low level of microbiota throughout the circuit. This was firstly due to the dynamics of the water circuit and secondly to the fact that the temperature of the heater tank was not always high enough. In the second stage, two samples were collected from the same tap: one proximal and one distal. Proximal samples corresponded to the first flush and distal samples to the hottest water. Legionella was detected in both sample types but was less prevalent in the distal (23 %) than in the proximal samples (35 %). Similarly, the counts were lower in the hottest water. To determine whether the proximal and distal samples with different Legionella counts represented two different niches, the non-parametric Wilcoxon test was used. The Legionella counts were significantly higher in the proximal samples (p