**4. Discussion**

#### *4.1. Legionella Abundance in Hospital Water and Biofilm of the West Bank*

This study is based on the first extensive sampling campaign examining the prevalence of *Legionella* spp in DWDS of hospitals in the West Bank. The analysis of water and biofilm samples was done using cultivation-dependent and -independent methods targeting *Legionella* from the genus to the clone level for *L. pneumophila* by molecular techniques including MLVA-8(12) genotyping.

In general, water samples had a far lower prevalence of *Legionella* compared to biofilms. Water samples tested positive for the presence of *Legionella* with a prevalence of 8.3% by cultivation dependent analysis and 50% by cultivation independent analysis. Biofilms had a higher prevalence, with 16.8% positive by cultivation dependent analysis and 61.3% by cultivation independent analysis. The findings of increased PCR-based detection in water and biofilms are consistent with other studies [44] and were analyzed in detail by Zayed et al. [42].

The MLVA-genotypes of the five water isolates were always present in the biofilm of the respective sites, usually making up a high fraction of the local biofilm isolates. In more detail, only from hospital A (one isolate), hospital F (three isolates) and hospital G (one isolate) were isolates from water obtained. In hospital A, the most abundant genotype from biofilm was Gt 4(17) (71% of the isolates), whereas Gt 16(1) obtained from water had a lower abundance (11%) of the biofilm isolates of this site. The three water isolates from hospital F belonged to Gt 10(93) or Gt 10(141). The water isolate from hospital G belonged to Gt 6(18). The water isolates from hospital F and G were the most abundant biofilm genotypes from these hospitals, i.e., Gt 10(93), Gt 10(141) and Gt 6(18) (Table 5 and Table S1).

The low prevalence of *Legionella* in culture-based studies is in accordance with studies in Israel and Greece [45–48]. However, many studies showed a much higher culturable *L. pneumophila* prevalence in water, e.g., 21.6%, 22% and 40% in Kuwait, Tunisia [49,50] and Jordan [51], respectively. The prevalence of *L. pneumophila* was even higher (68.5%) in a study from northern Israel [52]. The low prevalence of *L. pneumophila* in the West Bank was, at least to some extent, attributed to the high magnesium content of the drinking water [42].

Most of the *L. pneumophila* isolates from the West Bank (n = 175, 97%) were obtained from biofilm samples (Table 2). This is consistent with the results by Douterelo et al. [53], showing that more than 95% of the microbial biomass in a DWDS is found in the biofilms attached to the pipe lines due to the multiple advantages that biofilms represent for microorganisms, such as providing protection from external factors and beneficial interactions with other microorganisms [54]. Additionally, from the point of view of public health, biofilm sampling has a grea<sup>t</sup> importance, since it has been observed that *L. pneumophila* strains derived from biofilm replicate significantly more in murine macrophages than plankton-derived strains [55].

#### *4.2. General Health Relevant Aspects of the Isolated Strains*

Cultivation is still considered the gold standard for the detection of *Legionella* in the environment, even though other, nonculture methods are available, such as serology or nucleic acid-based detection methods [56]. Cultivation can be inaccurate as a result of overgrowth by other microorganisms on the agar plates, and can be ine ffective due to the presence of viable but nonculturable (VNBC) *Legionella* cells [57]. However, cultivation makes it possible to obtain isolates that can be identified and characterized phenotypically and genetically, which is essential for epidemiological studies.

According to current epidemiological data available from around the world, di fferent *L. pneumophila* serogroups cause legionellosis with a distinct estimated risk. Overall, the grea<sup>t</sup> majority of strains isolated from the area under study were characterized as Sg1 (62.3%). This fact followed the tendency already reported by other studies that have described Sg1 as the most frequently detected Sg of environmental isolates in di fferent geographic regions [58–60]. Besides the high prevalence of Sg1, other serogroups were isolated, where the fraction of non-Sg1 isolates went up to 37.7%. In our study, Sg6 was particularly abundant (30%). Sg8 and Sg10 were also isolated, although in smaller proportions (3.3% and 1.1%, respectively) (Table 3). The results obtained here were consistent with those of two studies on the distribution of *L. pneumophila* serogroups not related to human disease in man-made water systems [60,61], and were comparable, climate-wise to the area of study, i.e., Greece [48,62] *L. pneumophila* Sg1 was the most frequently isolated serogroup, followed by Sg6 in France and the UK, where Sg10 was also found. Sg6 is the serogroup which is second most responsible for cases of LD after Sg1, according to European surveillance data [63]. Furthermore, these specific serogroups (Sg1 and Sg6) are the most frequent and virulent among clinical cases [34,64–66].

ST1 was the most prevalent sequence type in the West Bank, and is the most dominant ST worldwide [22,60,67,68]. The high abundance of ST1 in the environment has been reported in several studies. In Japan, as well as in South Korea, the majority of environmental isolates comprised ST1 [69]; for the latter, ST1 was distributed across all sampled facilities and regions and accounted for 48.1% of the isolates [70]. ST1 was the most abundant sequence type among environmental isolates in Canada, and was found ubiquitously across the country [22]. In a study conducted across the United States, ST1 was the most frequent sequence type between both clinical sporadic and environmental isolates, accounting for 25% and 49% of the total number of isolates, respectively [68]. In Europe, ST1 has also been reported as the most predominant sequence type among environmental isolates in Germany [71], England and Wales [60], Portugal [72], Spain [73], France [74] and Italy [75].

Typically, the climate in the West Bank is Mediterranean, slightly cool to cold in winter and dry to humid and warm to hot in summer. Previous studies have suggested that the incidence of LD may increase under warm and wet meteorological conditions, which could be exacerbated by global climate change [76]. Therefore, surveillance of environmental sources and proper maintenance of man-made freshwater systems is key in the prevention of legionellosis. Surveillance of *Legionella* in the environment is also essential to validate the e fficacy of decontamination procedures, and for risk assessment when evaluating potential transmission or amplification sources.

#### *4.3. Genotyping Using MLVA—What Resolution Is Needed for Ecological and Clinical Issues?*

SBT is considered as the gold-standard for *L. pneumophila* genotyping, primarily due to a large International database created by the "*L. pneumophila* community". SBT has high typeability, interlaboratory reproducibility and generally a high index of discrimination [77]. However, the resolution is not as high as is often needed, e.g., for ST1 and its many health-relevant strains occurring world-wide [78,79]. It is suggested that the number of sequenced genes be increased to about fifty, guided by genome analyses.

MLVA is a rather well-established genotyping method currently used for 32 pathogenic bacterial species [27]. To date, it has been applied mostly for clinical strains. MLVA-8(12) for *L. pneumophila* was developed and its resolution analyzed by Sobral et al. Visca et al. and Pecellin [29,36,37]. All these studies demonstrated that MLVA-8, and even more, MLVA-8(12), have a higher resolution than SBT and are rather consistent with SBT. This was also shown in this study: larger strain sets adhering to a specific ST could always be distinguished into di fferent MLVA-genotypes. MLVA-genotyping was always consistent with SBT, i.e., strains of the same genotype were not assigned to di fferent STs. In this study, nine STs were split into 22 MLVA-8(12) genotypes, with large STs comprising several MLVA-genotypes.

The required resolution for the genotyping of strains is dependent on the tasks to be performed. For clinical reasons, genotyping should allow for a distinction with respect to virulence traits and antibiotic resistance [80]. Sharaby et al. [81,82] showed that *L. pneumophila* isolates from Israel had MLVA-8 genotype-specific virulence traits. Even their resistance to antibiotics showed a strong correlation with the MLVA-genotypes. Interestingly, environmental strains were more resistant to antibiotics than clinical ones. Moreover, there were major di fferences between di fferent MLVA-genotypes associated with ST1 that could not have been distinguished by SBT.

For the managemen<sup>t</sup> of *L. pneumophila* abundance in DWDS, the ecology of *L. pneumophila* has to be studied. As first shown by Rodriguez-Martinez et al. [52], the MLVA-genotyping level makes it possible to distinguish among specific ecotypes, i.e., by assessing the strains' environmental preferences. It was demonstrated that the preference for temperature and the respective growth speed could be well di fferentiated using MLVA-genotyping [83]. Furthermore, in these studies, there were relevant di fferences between di fferent MLVA-genotypes a ffiliated with ST1. A more detailed analysis on environmental preferences for the West Bank strains was provided by Zayed et al. [42]. For the West Bank strains, specific environmental traits could be assigned to all MLVA-8(12) genotypes comprising a larger set of strains.

Microbial source tracking is of high relevance for *L. pneumophila* due to its occurrence in environmental freshwater and its transfer, specifically during outbreaks, from the environment to humans. MLVA is rather economical and can be run in a fully automated manner using capillary sequencing. Due to the need for rapid analyses of large sets of strains in an outbreak scenario, MLVA can be a cost-e fficient and fast option. The successful source tracking by MLVA of an *L. pneumophila* outbreak was successfully demonstrated by Sobral et al. [29] in the French city of Rennes.

Another point of interest is intraspecies evolution. Garcia et al. [84] successfully followed the microevolution of *Vibrio parahaemolyticus* using MLVA technology in an experimental setting. Based on the experimentally-derived mutation rates, they estimated the worldwide evolution and time scale for *V. parahaemolyticus* populations. In general, clustering in VACCs gives the basis for good estimates among MLVA-genotypes. Links between the strains indicate their lines of evolution. For this study, relationships of the occurrence of genotypes over time and in neighboring sampling sites could indicate source strains and evolutionary tendencies [37].

#### *4.4. Comparison of L. pneumophila MLVA-8(12) Genotypes from the West Bank with the International Data Base and A Study on A Larger Set of Strains from Germany and Israel*

To address the distribution of the 27 MLVA-8(12) genotypes outside of the West Bank, the genotypes were compared to the International MLVA database (http://microbesgenotyping.i2bc.paris-saclay.fr/) and a larger strain analysis was performed by Pecellín [37]. The study by Pecellín [37] described a set of 610 clinical and environmental *L. pneumophila* strains retrieved from Germany, Israel and Palestine. The set of 180 strains from the West Bank described in this study were included in the study by Pecellín [37].

A comparison with the international database and the study by Pecellín [37] showed that there were a few highly ubiquitous genotypes. First of all, the MLVA-genotype comprised *L. pneumophila* Paris, i.e., Gt 4(17), that was associated with VACC1 and ST1. This genotype has a high relevance worldwide as a clinical and environmental genotype [68,74]. Another prominent member is Gt64(74), that comprises *L. pneumophila* strain Philadelphia-1 (ST 440). Two more genotypes occurred in the study of the three countries, i.e., Gt 4(16), a ST1-genotype closely related to Gt 4(17), and Gt 24(68), which is related to ST93. Gt 4(17) has an eminent role in Israel and the West Bank. This genotype occurs at high abundances in both regions. In Israel, it plays a relevant role as a clinical isolate [81]. At Oranim campus close to Haifa, concentrations of Gt 4(17) were high in the water and adjacent biofilms. In comparison with Germany, one more genotype occurred, i.e., Gt 40(74), which was associated with ST292. In comparison with Israel, two genotypes of environmental and clinical origin were observed, i.e., Gt 6(18) and Gt 6(15), both of which were a ffiliated with ST1 [81].

In summary, this means that 20 of the 27 MLVA-genotypes were unique for the West Bank. In addition, the VNTR clonal cluster VACC11 was described in the present study for the first time for the West Bank. Both the high percentage of new genotypes and the new VACC indicate the uniqueness of the West Bank strains. One possible line of ecological reasoning could be that groundwater is mostly the source water for DWDS in the West Bank [85], with an individualistic supply due to specific wells and springs as local water sources. This high diversity among sampling sites and the uniqueness of most of the genotypes may therefore be due to the very diverse water sources (Figure 2 and Table 5).

#### *4.5. Conceivable Health Relevance of the L. pneumophila MLVA-8(12) Genotypes in the DWDS*

PCR analysis of sputum and Broncho-Alveolar-Lavage (BAL) samples from pneumonia patients by Jaber et al. [43] in the West Bank revealed a rather high fraction of *L. pneumophila* contamination, i.e., 15% and 35%, respectively. By in situ SBT, they identified 29% of the detected *L. pneumophila* contaminations as ST1, and 21% as ST461. This is rather consistent with the MLVA-8(12) genotypes retrieved from the DWDS of the West Bank: the largest fraction of strains retrieved were a ffiliated with ST1 and ST461 (Figure 2). Unfortunately, there were no isolates obtained from patients due to previous antibiotic treatment. Therefore, no clinical strains could have been submitted to MLVA-8(12) analysis [43].

In the study by Sharaby et al. [81], clinical strains associated with ST1 were mostly Gt 4(17) followed by Gt 6(18). Therefore, it can be assumed that these MLVA-genotypes may have been responsible for infections in patients where ST1 was detected. For patients with ST461, all strains affiliated with this ST could be responsible for pneumonia, i.e., all genotypes (Gt10(141), Gt 10(93), Gt 9(92)) of the newly-described VACC11. The role of the remaining genotypes from the West Bank (non-ST1 and non-ST461) is unknown and remains to be elucidated in further studies. Hints may come from some publications [29] where strains of VACC2, including the genotype of *L. pneumophila* Philadelphia-1, were often observed as sources of LD in local outbreaks.

Another relevant point for public health and infection is the concentration of *L. pneumophila* in drinking water. For a site with comparable climate, Sharaby et al. [86] showed the risk at high concentrations of culturable *L. pneumophila* in drinking water of the Oranim campus (Haifa). From this perspective, it can be regarded as good news that the level of culturable *L. pneumophila* in the West Bank hospitals was, on average, rather low or undetectable. However, the noncontinuous water supply in the West Bank may cause disruption of biofilms in the DWDS, leading to short-term increases of the levels of *L. pneumophila* in drinking water that remain undetected during measurement campaigns.

Another health aspect is DWDS in private homes that has not ye<sup>t</sup> been assessed. Since *L. pneumophila* STs found in drinking water were observed for at least half of the investigated pneumonia patients, the drinking water in private homes of the West Bank should be considered as a potential health risk with respect to LD. Overall, the situations in private homes, i.e., discontinuous water supply and roof storage of water, are still underassessed; future studies should investigate the risks associated with the water quality and supply aspects.

In summary, our study provided the first comprehensive and long-term overview of the prevalence of *L. pneumophila* in DWDS with respect to water and biofilm, achieved by cultivation and PCR-based methods. Genotyping of the isolated *L. pneumophila* strains by high-resolution genotyping methods allowed us to group the isolates on a subspecies level and make international comparisons. In combination with genotype abundance and regional distribution, this provides better insights into potential health risks and may indicate where and which prevention measures might be needed.
