**1. Introduction**

*Legionella* is a genus comprising about 60 species mostly of aquatic origin and with a large fraction of pathogenic species [1]. The most relevant species for human health and artificial freshwater systems

> 177

is *L. pneumophila.* It is the most frequent causative agen<sup>t</sup> of an atypical pneumonia, Legionnaires' disease (LD), and Pontiac fever, a self-limiting flu-like disease [2]. Anthropogenic fresh water systems are considered as the major source for *Legionella* infections [3,4]. Co-infections with aquatic bacteria of LD patients hint on a co-transfer of bacteria from freshwater to the patient presumably via protozoa and/or their bacteria containing vesicles [5,6]. Phin et al. [7] concluded in a review on the worldwide epidemiology on LD, that the lack of thorough knowledge on the ecology on *Legionella* is a major obstacle for managemen<sup>t</sup> and prevention measures against *Legionella* infections.

The West Bank is a semi-arid region in the Middle East with hot and dry summers and cool winters with substantial water scarcity problems. Main precipitation falls in winter leading to an often only partial recharge of groundwater aquifers [8]. The source water for drinking water is mostly groundwater that is pumped into a storage reservoir and chlorinated, before delivered to the drinking water distribution system (DWDS). Due to frequent water shortage and supply interruption, water is stored in private containers, mostly on the roof, by the end users. All these factors may cause hygienic water problems in general and may lead to high *Legionella* abundance in the drinking water as consumed by the end user.

In general, the abundance of *Legionella* is considered to be enhanced by high water temperature and low chlorine concentrations [9–11]. However, recent studies have indicated that the factors triggering *Legionella* abundance might be more complex and that it is of high value to study the ecology of *L. pneumophila* on the genotype level [12,13]. The study by Rodriguez-Martinez et al. [13] performed in a similar climatic region (around Haifa, Israel) hints to a link between the genotypic composition of *L. pneumophila* and the abundance of *Legionella* species. In their study, a specific genotype (MLVA-Gt 4), closely related to *L. pneumophila* strain Paris, showed a correlation with very high *Legionella* plate counts and surprisingly low water temperature (mean value of 20.6 ◦C). The authors suggested that specific genotypes may act as triggers of high *Legionella* abundance, even at low temperature. In addition, genotype assessment is also of interest for tracking the source of *Legionella* infections, and the virulence itself is considered to be genotype-dependent [14,15].

The standard method for genotype assessment of *L. pneumophila* is sequence based typing (SBT) [1]. Sobral et al. and Visca et al. [12,16] showed that multi-locus variable number of repeat analysis (MLVA), a less laborious method, can be very well matched with the sequence types (ST) generated by SBT. In addition, the genotypes of ST1 are better resolved by MLVA. ST1 resolution had been shown to be of special relevance for drinking water [13], with respect to the environment as well as for virulence aspects [13,14]. Furthermore, ST1 is considered as of high relevance for artificial freshwater systems and human health on a global scale [17,18]. To achieve the needed high-resolution genotyping of *L. pneumophila* isolates, a MLVA-method with 13 loci was used by combining the loci of the MLVA-methods of Sobral et al. and Pourcel et al. [12,19]. Details on the results of MLVA-genotyping of the West Bank strains are given by Zayed et al. [20].

The overall aim of the study was to understand the relationship of the *L. pneumophila* genotypes and the environmental drivers determining their niches and abundances. This aim was pursued by seasonal assessment of *Legionella* abundance in water and biofilm in eight drinking water sampling sites, i.e., the DWDS of eight hospitals, covering the whole West Bank. The abundance of *L. pneumophila* was assessed by cultivation and PCR concomitantly with a record of relevant bacteriological and physico-chemical parameters of the drinking water during a period of 2.3 years. The 180 *L. pneumophila* isolates obtained from water and biofilm were identified by 16S rRNA partial sequencing and high resolution MLVA-genotyping in a previous study [16]. Correlation analysis and principal component analysis (PCA) were used to identify the niches of relevant *L. pneumophila* MLVA-genotypes and to identify environmental drivers of *L. pneumophila* abundance in water and biofilm. The study advocates for a genotype-based ecology of *L. pneumophila* and sheds light on so far not ye<sup>t</sup> considered mechanisms of *L. pneumophila* prevalence at the level of individual genotypes.

#### **2. Materials and Methods**

#### *2.1. Study Sites, Water, and Biofilm Sampling*

Drinking water in the West Bank is derived from groundwater, mainly well water, and some provided by springs. Water was provided to most of the sampled sites by the Palestinian Water Authority, except for Ramallah (sampling site D) with Mekorot as provider. Except for site D, water treatment consisted of chlorination of the water in storage sites before provided to the end user. All hospitals had drinking water reservoirs for water storage.

Water samples and biofilm swabs were sampled six times during the period from October 2012 to December 2014 from eight hospitals across the West Bank (Supplementary Figure S1). The six samplings covered twice the main seasons, i.e., spring (March–May), summer (June–August), and autumn (October–December). It should be noted that site D could only be sampled once for spring, summer, and autumn, while all other seven sites were sampled twice for these seasons [21]. Thus, the overall sampling campaign of the eight hospitals comprised 45 samplings.

Sampling was achieved on tap water and biofilm of faucets and shower heads of the drinking water distribution system (DWDS) of eight hospitals (Hospital A-H, Supplementary Figure S1, Table 1) of five cities covering the main regions across the West Bank. Furthermore, samples were taken occasionally from Al-Quds University (AQU) main campus, Abu Dies, Jerusalem (31◦4518.07 N, 35◦1537.614 E). These samples from Al-Quds were not included in the overall comparison on *Legionella* ecology and only used for comparison.


**Table 1.** Average *Legionella spp.* and *L. pneumophila* abundance per sampling site in water and biofilm as determined by cultivation and PCR (mean (SD)).

Cold and hot water (if available) was collected from faucets of the hospitals drinking water distribution system (DWDS). For cold and hot water sampling specific preselected faucets in the vicinity to the reservoir were used. Hot water was only available at five of the eight hospitals and not at all sampling dates (hot water was not available during sampling in May 2013; total set of 24 hot water samples). There was no water sampling directly from the hospital's reservoir, but only from the DWDS. Biofilm sampling was achieved from a predefined set of biofilm swabs from faucets, showerheads, and hoses. The sampled drinking water from the hospitals was considered as representative for the

cities of Jenin, Nablus, Ramallah, Jerusalem, Bethlehem, and Hebron, going from the north towards the south across the West Bank.

#### *2.2. Physico-Chemical Analyses of Bulk Water*

Cold and hot water samples were analyzed for temperature, pH, conductivity (using probes), and chlorine (Quantofix, Macherey-Nagel GmbH, Düren, Germany) directly upon collection. After being returned to the laboratory, water samples were analyzed for total iron, nitrate, nitrite, ammonia, copper, phosphate, zinc, carbonate hardness, and total hardness using Quantofix sticks. Magnesium and calcium concentrations were measured photometrically using Macherey–Nagel Nanocolor assays. Data on turbidity, bicarbonate, chloride, sulphate, total dissolved solids (TDS), and fluoride were provided by the Palestinian Water Authority.

#### *2.3. Cultivation Dependent Analysis of Water and Biofilm*

Per sampling date and site, one cold and one hot water sample was collected from the preselected faucets of the DWDS (vicinity to water reservoir) in sterile bottles after a flushing time of 2 min: One liter of each cold and hot water was collected for Heterotrophic Plate Counts (HPC), and one liter of each cold and hot water was collected for *Legionella* counts. To neutralize residual free chlorine, 0.5 mL of 0.1N sodium thiosulphate was added in the sterile bottles for *Legionella* count determination [22].

Concerning *Legionella* plate counts, a 100 mL water sample was filtered onto a membrane filter (membrane solutions, pore size 0.45 μm, diameter 47 mm, Whatman, England) using a sterile filtration unit (Nalgene, Germany). A vacuum of 200 mbar was applied. After filtration, 30 mL of acid bu ffer (3.9 mL of 0.2 mol/L HCl and 25 mL of sterile 0.2 mol/L KCl were mixed, pH 2.2 ± 0.2) was placed on top of the membrane filter and left for 5 min. The filter was rinsed with 20 mL Page's saline (1.20 g NaCl, 0.04 g MgSO4·7H2O, 0.04 g CaCl2·2H2O), and 1.42 g Na2HPO4 and 1.36 g KH2PO4 were dissolved in ten liters of distilled water and autoclaved. The membrane filter was removed from the filtration unit with sterile forceps and placed onto the relevant agar plate. Duplicates of BCYE and GVPC (M809, Himedia, Mumbai India) agar plates were used according to the manufacturer's instructions. The plates were incubated inverted at 37 ◦C for 10 days. Plates were checked for growth twice (after three and ten days). Final counts of the triplicates were done after ten days with descriptions of the colonies. Detection limit was five CFU/L.

Biofilm swabs were sampled from the anterior surfaces of faucets, showerheads, or shower hoses in all hospital wards. Per sampling and sampling site 20 biofilm swabs were taken, except for the first sampling when 44 swabs were sampled per site to check the variability per sampling site. Biofilm swabs were obtained using transport medium (Copan, Culture swab transport system, Brescia, Italy). Swabs for *Legionella* identification were processed immediately by culturing on GVPC agar (medium M809, Himedia, India) according to ISO 11731:2004 [23]. More details on cultivation dependent analyses are given by Zayed et al. [20].

From all water and biofilm samples with visible growth of *Legionella*-like colonies on agar plates, representative isolates were chosen and purified. Isolates were later characterized by *L. pneumophila*-specific PCR (Primer L1) [24], 16S rRNA gene sequence, serogroup and genotype assignment using MLVA, and a representative subset by sequence-based typing (SBT) (see below). Please note that *Legionella*-like colonies weremostly confirmed by PCR as *L. pneumophila*. Asit was a rather rare event that non-*L. pneumophila* colonies were detected, the *Legionella* plate counts can be considered as reflecting the culturable fraction of *L. pneumophila*.

#### *2.4. Cultivation-Independent Analysis of Water and Biofilm*

Per sampling site and date, five liters of both cold and hot water were collected per sampling and site from the predefined faucet from the DWDS in the vicinity of the reservoir for DNA extraction. Water samples were filtered onto sandwich membrane filters composed of a nucleopore-filter (Nucleopore Track-Etch Membrane, 90 mm diameter, 0.2 μm pore size, Whatman, England) and

a glass fiber-microfilter (GF/F) (GFF, 90 mm, Whatman, England) [25]. Filters were stored frozen at −80 ◦C for later DNA extraction. More details on cultivation-independent analyses are given by Zayed et al. [20].

For biofilms, 5 swabs were taken per sampling and site from the anterior surfaces of faucets, showerheads, or shower hoses using sterile cotton swabs (Cotton Tipped Applicator, Beijing, China). Swabs were stored frozen at −20 ◦C for later DNA extraction. For the extraction of DNA from the filter sandwiches and the swabs, a modified DNAeasy protocol (Qiagen No.69506, Hilden, Germany) was used [26].

Using the DNA of the extracted water and biofilm samples and strains, PCRs with di fferent targets were carried out as described by [24]: (i) for the detection of any bacteria, universal 16S rRNA gene primers (Com1F, Com2R) were used, and (ii) a *Legionella* genus-specific PCR (primer set Lgsp17F, Lgsp28R) (iii) and a *L. pneumophila*-specific PCR (primer set Lp-16S\_246-248F, Lp-16S\_246-248R) were applied. On all samples, used for PCR-based analysis, cultivation-dependent analysis for *Legionella* was additionally performed to allow a direct comparison. Sequencing of the 16S rRNA gene of six representative isolates confirmed the identification of *L. pneumophila* (≥99.8% 16S rRNA gene similarities) [20].

#### *2.5. Genotyping of L. pneumophila Isolates*

For molecular genotyping of *L. pneumophila* at the strain level, MLVA-8(12) analysis was performed for 180 isolates. For all details see Zayed et al. [20]. Briefly, DNA extraction was done either directly from living biomass using DNAeasy (Qiagen No. 69504, Hilden, Germany) according to the manufacturer protocol or from biomass applied to FTA cards (Whatman, Sigma-Aldrich, Germany). For the final MLVA thirteen loci were used, i.e., the twelve loci of MLVA-12 (12) plus the one additional locus of MLVA-8 [19] not used in MLVA-12. A subset of MLVA-genotypes was characterized by sequence-based typing (SBT) [1]. The MLVA-8(12) genotypes of the West Bank were compared to the International MLVA database (http://microbesgenotyping.i2bc.paris-saclay.fr/as performed and described in more detail by Pecellín [27].
