**3. Discussion**

The abundance of Lp in the tourism sector is a continuous focus of attention in *Legionella* research, due to its possible implications to public health. A summary of earlier publications reporting national surveys of tourist accommodations in different countries is presented in Table 3.

This study shows, for the first time, the distribution and prevalence of *Legionella* spp. in the Israeli hotel sector. By analyzing 2830 water specimens, taken from 168 hotels over the two-year period between 2015 and 2017, we demonstrate that 60% of the examined hotels were colonized with *L. pneumophila*, and in 37% of them, the concentrations of *Legionella* in water exceeded the national regulatory thresholds. Of all 2830 specimens collected, 17% were *Legionella*-positive, with half of those exceeding threshold levels of *Legionella*.

We analyzed the results of *Legionella* quantitation, according to the category of water source, which included cooling towers, hot tubs, waterlines, showering facilities, storage tanks, and room tap water. The most affected source type was cooling towers (38%), while specimens from other sources showed lower rates of *Legionella* colonization, at around 15%. Furthermore, 32% of the samples from cooling towers exceeded the 1000 CFU/L regulatory threshold for *Legionella* concentrations, making this the water source with the highest proportion of exceeding samples.


**Table 3.** Summary of earlier publications reporting national surveys of tourist accommodations in different countries.

Cooling towers are the most frequently reported water source of LD outbreaks worldwide [26–28], and can involve a large number of cases [29,30]. The role of cooling towers in the urban spread of Lp has also been demonstrated recently in a genomic analysis of isolates over time in Switzerland [31]. The high proportion of Lp-contaminated cooling towers reported here is a public health concern that should prompt further investigation, due to the high population density in urban areas. However, in contrast to the reports from other countries, LD cases in Israel have not been linked to cooling towers. Since not all *Legionella* spp. and Lp strains are suggested to have the capacity to cause LD [32,33], this might be a reason for the discrepancy. It would be interesting, therefore, to look specifically at the population structure of *Legionella* spp. in cooling towers nationwide.

Serotyping of a subset of 162 presumed *Legionella* isolates revealed that 33% belonged to Lp sg1, while 64% belonged to Lp sg2–14. Several studies have explored the distribution of Lp sg1 in the environment. A study from South Korea demonstrated the significant predominance of Lp sg1 in manmade water systems, including hotels, with prevalence rates up to 55% [22]. In Italy, the Lp sg1 distribution rates in the hotel setting di ffered greatly between two studies, at 27.7% and 55%, including mixed cultures [15,34]. On the other hand, findings from Italy [19], Greece [23], and Turkey [21] have shown that the most frequent colonizers of the hotel water systems in these studies were Lp sg2–14.

A growing body of evidence shows the *Legionella* strains' ability for long-term persistence in manmade water systems, without a significant fluctuation of population diversity [12,35,36]. Based on this hypothesis, we assume that our findings reflect the rates of Lp sg1 distribution in Israeli hotels, though more investigation is needed to extend our knowledge on the persistence of local Lp strains in water systems associated with di fferent settings.

Using SBT applied on a convenience sample of 78 Lp isolates, we have identified 27 STs, including two novel STs. Nine STs belonged to sg1. Lp sg1 ST1 was the prevalent type, accounting for 26% (20/78) of the sequence-typed isolates.

ST1 has been described by numerous studies, amongs<sup>t</sup> a few other STs, as a main causative agen<sup>t</sup> of LD globally, supporting its high pathogenicity [37,38]. Moreover, in contrast to other highly pathogenic clinical strains rarely isolated from the environment, ST1 has been shown to be among the predominant environmental Lp sg1 strains [22,39–43].

In our study, ST1 comprised 63% (20/32) of all sg1 isolates from hotel water systems. The high rate of the environmental predominance of Lp sg1 ST1 corresponds with our national surveillance data, where ST1 is by far the most common cause of LD in Israel [44]. This abundance of ST1 in the environment poses a challenge for public health services, limiting their capability to ultimately identify a source of infection during investigations of ST1-associated outbreaks using traditional SBT.

Amongst the non-sg1 isolates (46/78), two isolates failed to generate a full seven-allele profile, due to no amplification of a *flaA* PCR product. Lp strains with mutations at the SBT *flaA* primer-binding site have been described elsewhere [45], including an Lp subtype from Israel (0,14,16,25,7,13,206) that had been further identified by the whole genome sequencing (WGS) approach as having the *flaA11* allele, and which has been assigned to ST1334 [46]. However, the two strains found in this study di ffer from ST1334 (0,4,16,1,7,13,206 and 0,14,16,1,7,13,206), supporting the idea of an ongoing dissemination of the mutation.

Concerning the geographical distribution of the Lp population in this study, we observed a relative abundance of a number of strains in some districts in Israel. For example, sequence type ST1 was identified in each of the six Israeli districts, and ST1642 was found in four of the districts. On the contrary, other subtypes were associated with only one geographical region. For example, ST59, ST1326, and ST1641 were found in the Jerusalem district, and ST1516 was limited to the Southern region. Even though these strains are not unique to Israel, apart from ST1641, we observe their strong association with these two regions from our surveillance programs and during epidemiological investigations [47]. An explanation might be water-related di fferences between the regions (i.e., physical and chemical properties) caused by the climatic and topographical characteristics of the geographic regions. However, more data is needed to verify this assumption.

In our study, we have found an unexpectedly low rate of non-Lp spp. in the hotel water systems. In fact, we identified only one *L. bozemanii* strain from isolates subjected for serotyping. Other studies that have explored the environmental distribution of *Legionella* spp. have detected Lp and non-Lp co-existence in water. A recent study from the United States has demonstrated that 72 of culture-positive environmental samples collected during summer 2016, where Lp sg1 was recovered, also contained at least one other Lp or non-Lp *Legionella* spp. [30]. In another study from Crete, Greece [23], carried out from 2004 to 2011, about 50 non-Lp *Legionella* spp. were identified in the water systems of the hotel setting. Variability in the prevalence of non-Lp spp. between the studies can possibly be explained by the di fferences in the isolate selection procedure for subsequent analysis. In our study, the initial identification of *Legionella* spp. was carried out by serotyping, followed by the *mip* sequencing of non-groupable strains. In contrast, the application of PCR-based techniques for *Legionella* spp. screening on all samples would probably have yielded results that are more diverse. Moreover, the isolation processing methods used in our study may have reduced the detection of non-*pneumophila Legionella* spp., due to the overgrowth of other bacteria, and may have underestimated their overall abundance in samples. The membrane plating method is subject to some issues of overgrowth (especially in non-potable water), has been reinforced in a few studies [48,49], and is discussed in the new ISO 11731:2017 [50]. Thus, *Legionella* spp. distribution may not be fully represented here.

This study has several limitations. First, despite the considerable overall number of 2830 water specimens analyzed, several geographical regions, such as the Northern district, were underrepresented in our study. Second, the study was based on a convenience sample, and thus may not accurately represent the entire tourism sector in Israel, or in certain districts.

In this survey, a large set of water samples was examined routinely without any sampling efforts, due to sporadic travel-associated LD cases or outbreaks. Therefore, our findings on *Legionella* prevalence in hotel settings in Israel are fully representative of non-outbreak-related surveillance. Regarding the molecular structure of *L. pneumophila* population, this study demonstrates, for the first time, the molecular profile of Lp strains in the water systems of Israeli hotels and resorts.

Altogether, our findings contribute to the existing knowledge concerning the understanding of the environmental distribution of *Legionella* spp. in our region, and may facilitate international activities, such as TALD surveillance. The peculiar geographic distribution of di fferent strains should be further investigated.

#### **4. Materials and Methods**

In total, 2830 convenience samples from 168 hotels and resorts were collected via routine surveillance, according to the regulations of the Israeli Ministry of Health for the prevention of *Legionella* growth in water distribution systems and hot tubs [51]. The study took place between March 2015 and the end of February 2017 across six Israeli districts (Northern, Center, Southern, Haifa, Tel Aviv, and Jerusalem). Hotels and other tourist accommodation are obliged by the regulations to monitor their water systems for the presence of *Legionella* spp. The minimum mandatory testing routine schedule depends on the hotel's size: once every two years for sites containing <50 rooms, once a year for those with 50–300 rooms, and twice a year for those with >300 rooms. Both hot and cold water distribution systems should be tested as part of this procedure. For hot tubs, the minimum sampling routine is quarterly. Selection of the sampling points depends on a hotel water system maintenance plan and is comprised of hot and cold water from outlets representing the rooms (faucets, showers) and mains (hot water return lines, hot and cold water supply, and storage tanks): cold water from cooling towers, decorative fountains, pools, air conditioning systems, and cold/hot/mixed water from hot tubs. Samples were taken after flushing for 2 minutes and the disinfection of the outlet, as per the requirements of the Israeli Public Health guidelines for routine monitoring of water distribution systems [51]. In addition, following regulatory requirements, water systems with *Legionella* concentrations above the thresholds of 1000 CFU/L for potable water and 1 CFU/100 mL for hot tubs were re-tested after the appropriate treatment [51]. Overall, of the 168 hotel water systems included

in this study, 119 were probed at least twice. At each sampling point, 1–2 water samples (hot and/or cold water) were collected in 1 L sterile plastic bottles containing sodium thiosulfate, in order to neutralize the residual-free chlorine. All water samples were stored at 4 ◦C and processed within 24 h of their collection.

The detection and quantitation of *Legionella* spp. were performed in a certified water testing laboratory per the ISO 11731-2:2004 method [52]. Potable and non-potable water samples were filtered with 0.45 μm sterile gray membrane filter paper, treated with 30 mL of acid bu ffer containing 0.2 M KCL and 0.2 M HCL for 5 min, and washed with 20 mL of PAGE's saline. Water samples originated from cooling towers and fountains were processed in four dilutions (1:10000, 1:1000, 1:100, and 1:10), in order to avoid the overgrowth of microbial flora. Membranes were transferred to Glycine Vancomycin Polymyxin Cycloheximide (GVPC) medium (cat. no. 257007, BD, Heidelberg, Germany) and after incubation at 35 ± 0.5 ◦C for 10 days, colonies suggestive of *Legionella* spp. were subcultured to Bu ffered Charcoal Yeast Extract (BCYE) and 5% sheep blood agar media (P073 and P049, HyLabs, Rehovot, Israel). Subsets of representative isolates identified as *Legionella* spp. were regularly referred to the National Reference Laboratory for *Legionella* at the Ministry of Health, according to regulations [51].

The total amount of 164 *Legionella* isolates from hotel water systems was obtained during the two-year study. Serotyping was performed with the Legionella Latex Test kit (Cat. No. DR0800, Oxoid, Basingstoke, UK).

Strains not readily confirmed by serotyping as *L. pneumophila* were identified to species level by sequencing the *mip* gene, as described by Ratcli ff et al. [7], and comparing the sequence to the *mip* database [53]. The molecular characterization of *L. pneumophila* strains was conducted according to the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) Study Group for *Legionella* Infections (ESGLI) sequence-based typing (SBT) scheme [54,55]. The choice of the isolates subjected to SBT monthly was based on the data provided by the referring laboratory and guided by epidemiological and risk assessment criteria, such as high *Legionella* CFU counts, a source type with high public health risk potential, or a new sampling site. After the exclusion of duplicate isolates arising from the same sampling points, 78 isolates were examined in this study. Sequences obtained by Sanger sequencing were analyzed with the BioNumerics software (Version 7.6, Applied Maths) and compared to the ESGLI database for assigning the ST. New allelic profiles were submitted to the ESGLI SBT database [56]. The strain diversity index was calculated according to the modified method of Hunter and Gaston [57].

BioNumerics software (Version 7.6, Applied Maths) was used for phylogenetic analysis. Clustering was created using the unweighted pair group method with arithmetic averages (UPGMA) [58]. The minimum spanning tree (MST) was created using a predefined MST for the categorical data template, with single- and double-locus variance priority rules. Geomap was created using ArcGIS Pro 2.5 (Esri, Redlands, CA, USA).

**Author Contributions:** Conceptualization, J.M.-G.; methodology, E.Y. and J.M.-G.; formal analysis, L.V. and E.K.; investigation, E.K.; resources, V.A. and I.G.; data curation, E.Y. and E.K.; writing—original draft preparation, E.Y.; writing—review and editing, J.M.-G. and L.V.; supervision, L.V.; funding acquisition, V.A. and I.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
