**3. Results**

#### *3.1. Biological and Physico-Chemical Characteristics of Drinking Water Distribution Systems (DWDS)*

The sampled drinking water of the eight hospitals was mainly groundwater based and characterized by a high hardness (on average 230–300 mg/<sup>L</sup> CaCO3 equivalents) and high conductivity (on average 650–900 μS) (Table 1). The average temperature of the cold water ranged between 21.1 ◦C and 24.3 ◦C. The average temperature of hot water ranged between 38.6 ◦C and 51.9 ◦C. The average pH of the cold and hot water was 7.6 and 8.0. The conductivity of the hot water was higher than that of the cold water in all hospitals. Chlorine varied on average between 0.2 and 0.7 mg/L. Heterotrophic plate counts at 37 ◦C ranged from 1.7 × 10<sup>4</sup> to 1.6 × 10<sup>5</sup> CFU/L.


**Table 1.** Average of biological and physico-chemical parameters from the drinking water systems (DWDS) of the eight hospitals in the West Bank.

During the study period, *L. pneumophila* was detected in the DWDS of all hospitals and at Al-Quds University. The sampling comprised six campaigns, twice covering the main seasonal changes. Sampling comprised water and biofilms, with comparable numbers of samples taken from each hospital.

From water samples, five *L. pneumophila* strains were isolated from 72 samples. *L. pneumophila* was isolated only from the drinking water of three hospitals (A, F and G). For the collection period, the *Legionella* counts per hospital ranged from 0 to 150 CFU/L (Table 1). In hospital A, on average, 43 CFU/L of *Legionella* spp from cold water were detected; in hospital F, 150 CFU/L of *Legionella* spp were detected from cold water and 91 CFU/L from hot water; finally, in hospital G, on average, 8 CFU/L *Legionella* spp were detected from cold water. Hot water was not continuously available from all hospitals, but *Legionella* spp counts were comparable to cold water [42]. *Legionella* spp, mainly *L. pneumophila,* was isolated from 191 out of 1136 biofilm swab samples (16.8%) (Table 2). The majority of *Legionella* positive samples were detected in 2012 (23.5%), while the fewest samples tested positive in 2013 (8.7%). The highest frequency of *L. pneumophila* in biofilm swabs was detected in hospital F (26.3%), where *Legionella* spp were also detected in the DWDS during 2012–2014. Meanwhile, the lowest frequency was detected in hospital C (3.3%). Finally, a high frequency of *L. pneumophila* was detected in Al-Quds University (36.4%) during the only collection in 2012.



NA: Not Available; BD: Below detection limit (<5 CFU/L); AQU: Al-Quds University; CDA: Cultivation Dependent Analysis; **2** CIA: Cultivation Independent Analysis; **3** The number of isolates corresponds to the number *L. pneumophila* culture-positive water and biofilm samples.

#### *3.2. Cultivation Dependent Analysis (CDA) versus Cultivation Independent Analysis (CIA)*

Although CDA is the standard and recommended technique for environmental surveillance of *L. pneumophila,* CIA provides higher sensitivity and overcomes the problems of CDA for *Legionella* because of the VBNC state and its overgrowth by competing bacteria. In this study, both methods were used to detect *Legionella* spp in hospital DWDS (Table 2).

A total of 72 water samples and 225 biofilm swabs from the eight hospitals were tested by conventional PCR using three di fferent primers (com, Lgsp, Lpn). Almost all of the samples were positive using com primers (n = 71, 98.6% and n = 225, 100%) for water samples and biofilm swabs respectively. *Legionella* spp were detected in biofilm swabs more than in water samples (n = 167, 74.2% and n = 42, 58.3%), respectively (Table 2). Similar results were obtained using *L. pneumophila*-specific

primers: 60% of the biofilm swabs and 50% of the water samples were positive. As expected, the PCR-based CIA showed higher sensitivity than CDA. CIA analysis increased the detection of *L. pneumophila* from 8.3% (CDA) to 50% (CIA) for water samples, and from 16.8% (CDA) to 59.5% (CIA) for biofilm samples.

#### *3.3. MLVA-8(12) Genotypes of L. pneumophila Isolates*

MLVA genotyping was carried out for the 180 isolates. MLVA-8, as well as MLVA-12 and the joint scheme MLVA-8(12), were analyzed for the study of the population of *L. pneumophila* isolates. The 180 isolates were categorized as 16 MLVA-8 genotypes (Index of Diversity ID = 0.771, 95% Confidence Interval CI 0.721–0.822), 25 MLVA-12 genotypes (ID = 0.790, 95% CI, 0.739–0.841) and 27 MLVA-8(12) genotypes (ID = 0.790, 95% CI 0.739–0.841). This indicates a lower genotypic resolution for MLVA-8 using eight loci compared to MLVA-8(12) using a total of 13 loci. For details on the comparison based on a larger set of *L. pneumophila* isolates, see [37].

The use of the MLVA-8(12)-genotype nomenclature made it possible to directly compare strains genotyped from MLVA-8 and MLVA-12: the first number reflects the MLVA-8-classification, while the number in brackets reflects the 12 loci-classification, e.g., Gt 4(17) is a Gt 4 according to the MLVA-8, and a Gt 17 according to MLVA-12. Fourteen MLVA-8(12) genotypes were represented by 2 to 74 strains, whereas 13 genotypes were represented by just a single strain from the West Bank isolates (Figure 1). The MLVA-8(12) genotypes comprising the most strains were Gt 4(17), Gt 6(18) and Gt 10(93).

**Figure 1.** Comparison of the number of strains per MLVA8(12)-Genotype (Gt) of the isolated 180 *L. pneumophila* strains and their respective sequence types (ST). Each group of bars, outline color and colored frame indicate VACCs as follows: blue outline represents VACC1, yellow outline represents VACC2, green outline represents VACC5 and purple outline represents VACC11. The black round-edged frames indicate the group of genotypes from the same ST. The wide upward diagonal hatches inside the bars indicate Sg as follows: Sg1—red, Sg 6—sky blue, Sg8—green, Sg10—yellow and Sg2 to Sg14—black. NA—not available ST; \*, ST was assessed for strains of the same MLVA-8(12) genotype, and not directly for the West Bank strains; ˆ, ST was estimated from the MLVA-8 pattern.

#### *3.4. VNTR Clonal Complexes (VACC) and Relationship among the Genotypes*

An analysis of the relationship among the genotypes was achieved by UPGMA-based cluster analysis of the MLVA-8(12) profiles of the 180 *L. pneumophila* strains (Figure S2). The MLVA clonal complexes (VACC) were defined by a cutoff level of 60% similarity. In addition, the genetic relationship among genotypes was estimated by a minimum-spanning tree based on the MLVA-8(12) profiles (Figure 2).

**Figure 2.** Minimum-spanning tree based on MLVA-8(12) profiles of 180 *L. pneumophila* strains isolated from the West Bank. Each circle in the tree represents a different MLVA-8(12) genotype. The genotype number is indicated within or near the circle, whose size is proportional to genotype frequency. Different colors in the pie charts refer to the eight sampling locations (see legend). The thickness of the branches represents the number of different loci. MLVA clonal complexes (VACC) are shaded in grey. The circles representing the Sg8-singletons from Al-Quds University, i.e., Gt 11(87) and Gt 12(84), overlap visually within the circle that represents Gt4(17) due to the high abundance of this genotype. Blue ellipses indicate genotypes that could be candidates for LD. These genotypes belong to ST1 and ST461 (indicated in blue letters) and were assessed by in situ-SBT and present in half of the LD cases in a West Bank study [43].

All MLVA8(12) genotypes were clustered into four MLVA clonal complexes or VACCs (VACC1, VACC2, VACC5 and VACC11) (Figure 2). VACC1, VACC2 and VACC5 were clonal complexes previously defined in the MLVA *Legionella* database. VACC11 is described for the first time in this study (Figure 2 and Figure S2). VACC1 was the largest cluster, including 110 isolates (61.6%). VACC11, VACC2 and VACC5 were, in comparison, smaller clusters, comprising 31, 19 and 14 isolates, respectively (Figure 1).

Not all strains could be included in VACCs. A small group of six isolates that belonged to two different genotypes, i.e., Gt 11(87) and Gt 12(84), separated from the large VACC1 and remained as singletons, i.e., they could not be directly included in a VACC. They differed from the rest of the isolates contained in VACC1 in the number of repeats observed for VNTR markers Lpms31, i.e., 17, in comparison to 4 or 0 in the rest of the profiles of VACC1, and VNTR Lpms33 and Lpms34, which both presented only one repeat in contrast to the 4 and 2 repeats found, respectively, in VACC1. In total, 96.6% of isolates (n = 174) were clustered into the four VACCs, and only 3.6% (n = 6) were found as singletons.

In addition, single linkage clustering using the Bray-Curtis-Index of association was performed with a cutoff value of 60% corresponding to 13 VNTRs. The MLVA8(12) profiles of four additional reference strains were added to the dataset to show their relationship with the genotypes in this study (Figure 3). This clustering showed the same four groups corresponding to the VACCs observed with the UPGMA-clustering and the minimum-spanning tree (Figure 2 and Figure S2). In addition, it revealed a very close relationship of the *L. pneumophila* strain Paris with Gt 4(17) in VACC1, and of Gt 64(74) with reference strain *L. pneumophila* strain Philadelphia1 in VACC2. The dendrogram showed a high discriminatory power and subclustering between *L. pneumophila* isolates with similarity of (>95%). Gt63(83) was shown as an outlier due to the three "Null" alleles in Lpms31, Lpms34 and Lpms35, i.e., no PCR product was obtained for these VNTR-markers (Figure 3). Overall, the Bray-Curtis-grouping of the genotypes confirmed the four VACC cluster. Also, the aforementioned singleton genotypes Gt11(87) and Gt12(84) were grouped separately from VACC1.

**Figure 3.** Single-linkage cluster dendrogram representing the percentage of similarity between MLVA-8(12) profiles of the genotypes retrieved from the West Bank. For a comparison, MLVA-8(12) profiles of *L. pneumophila* reference strains (Bloomington2, Philadelphia-1, Corby, Paris) were added. The resemblance matrix was calculated using the Bray-Curtis index of association. Strains of different VNTR clonal complexes (VACC) are indicated in different colors (see insert).

Overall, a more detailed analysis of the population structure at the level of the 13 VNTR markers showed a balanced variability in the number of repeats for most VNTR markers among the isolates (Table 3). This could be due to the homogenized habitat and location where the isolates were obtained. Some VNTR markers appeared to be less variable and showed a reduced number of repeats, e.g., Lpms3, Lpms17 and Lpms19, while others showed a greater variability, e.g., Lpms31 and Lpms35. In general, the same repeats prevailed independently of the area from which the isolates had been isolated. Null alleles were present at different frequencies in distinct VNTR markers. Especially high were the frequencies of null alleles in Lpms38. Remarkably, a new allele of Lpms34 was described during this study. The new allele had a size of 634 base pairs and was formed by four repeats. A total of 31 isolates (17.2%) contained this allele. This allele has not previously been described in MLVA studies for *L. pneumophila*.


**Table 3.** VNTR characteristics of the *L. pneumophila* strains isolated from the West Bank.

**1** HGDI: Hunter-Gaston Discrimination Index.

#### *3.5. Diversity within the Clonal Complexes (VACC)*

The minimum-spanning tree (Figure 2) make it possible to provide an overview of the diversity and the genetic relationship among the MLVA genotypes. While the grouping by VACCs gives a first estimate of the relationship among the total of the genotypes, the relationship of the genotypes within the VACCs is also indicative.

In VACC1, all genotypes pertaining to ST1 were closely related to Gt 4(17). The replicate number of only one locus had changed compared to Gt 4(17). This is comparable to VACC11, where Gt 10(93) has this central position, with only one locus being di fferent compared to Gt 10(141), Gt 9(92) and Gt 55(94). By contrast, in VACC2 and VACC5, a set of more distantly related strains with a broader set of di fferent changes in the VNTR-loci was observed.

The relationship of the genotypes within the cluster as reflected by the minimum spanning tree were consistent with the results of the UPGMA-based cluster analysis (Figure S2) and Bray-Curtis-based analysis (Figure 3).

#### *3.6. Comparison of MLVA-8(12) Genotypes and Clonal Complexes with Sequence Types (ST)*

In our comparison, 22 of 27 MLVA8(12) genotypes could be assigned to nine sequence types (STs). Most STs with a larger set of strains could be divided in two to five genotypes (Figure 1 and Figure S2). ST1 was split into five MLVA-8(12) genotypes all adhering to VACC1. ST1 comprised the largest fraction of strains, i.e., 111 strains of the total of 180 strains. ST1 comprised the MLVA-8(12) genotypes with the most isolates, i.e., Gt 4(17) and Gt 6(18) (Figures 1 and 2). ST461 comprised 30 strains and three genotypes of VACC11. It was the second largest ST. VACC11 strains were not present in any other ST. Three genotypes were also present in each of ST1326, ST1438, and ST1482. Two MLVA genotypes were present in ST1358. The remaining three STs (9, 93, 187) comprised one to three strains and constituted a single MLVA-8(12) genotype.

In summary, all STs with a larger set of strains were split up in several MLVA-genotypes, i.e., MLVA-8(12)-genotyping showed a substantially higher resolution than SBT. All STs comprised only strains of the same VNTR clonal complex (VACC). Our dataset indicates a high level of consistency between SBT and MLVA8(12)-genotyping. This observation was confirmed by the analysis of a larger and more diverse set of *L. pneumophila* isolates [37].

#### *3.7. Serotype Distribution of L. pneumophila Isolates and Relationship with MLVA Genotypes and Clonal Complexes*

The 180 PCR-confirmed *L. pneumophila* environmental isolates were tested for serogroups (Table 4 and Figure 1). Most of the isolates were characterized as serogroup 1 (Sg1) (n = 111); the remaining 69 isolates were non Sg1. A subset of ten Sg1 isolates was subgrouped according to the monoclonal antibody; all belonged to the MAb 3/1 negative OLDA subtype, which is considered to lack the virulence- associated epitope. The 69 non-Sg1 isolates were analyzed by monoclonal subgrouping; 54 of them were serotyped as Sg6, followed by Sg8 (n = 6) and Sg10 (n = 2). The rest of the non-Sg.1 were characterized as serogroups 2–14, as determined using an agglutination kit.


**Table 4.** Serogroup and monoclonal antibody subtyping of 180 environmental *L. pneumophila* isolates from the West Bank.

In terms of the number of isolates, the *L. pneumophila* population showed a dominance of Sg1, followed by Sg6 (Figure 1). In terms of MLVA-genotypes, 11 could be attributed to Sg6 and seven to Sg1, while Sg8 was represented by only two and Sg10 by only one genotype. In terms of clonal complexes, VACC11 comprised only strains of Sg6. VACC1 comprised mainly Sg1 strains, except for Sg8 for genotypes Gt11(87) and Gt12(84). VACC5 comprised strains of either Sg6 or Sg2–14. By contrast, VACC2 comprised a serogroup-divers set of genotypes, i.e., Sg1, Sg6, Sg10 and Sg2–14. The six singleton isolates were all Sg8-strains isolated from Al-Quds University.

#### *3.8. Prevalence and Abundance of L. pneumophila MLVA-Genotypes and Clonal Complexes (VACCs)*

An overview of the association of the MLVA-8(12) genotypes and VACCs with the different sampling sites is reflected in the minimum-spanning tree (Figure 2). Details on the strains retrieved and their characteristics for each sampling site are listed in Table 5.

<sup>1</sup> mAb: monoclonal Antibody; 2 NA: Not analyzed.


**Table 5.** MLVA-8(12) genotype abundances at the sampling sites in the West Bank.

\*: Al-Quds University.

In terms of VACC prevalence, Figure 2 shows that at least two distinct clonal complexes were present at each hospital as well as at the Al-Quds University. VACC1, the largest clonal complex, was present across the West Bank. Genotypes belonging to it were isolated from all eight hospitals and from AQU. VACC2 isolates were isolated from four hospitals distributed throughout the West Bank and AQU. Although VACC11 was present at five hospitals, it was the major clonal complex at hospital F (n = 22, 66.7%). Isolates grouped into VACC5, the smallest clonal complex, were, however, found at four hospitals located throughout the West Bank (Table 5).

At the genotype level, only nine out of the 27 MLVA-8(12) genotypes were isolated in more than one location. The remaining 18 MLVA-8(12) genotypes were isolated exclusively in one specific site. Gt 4(17), the main VACC1-genotype that also comprises *L. pneumophila* strain Paris, was the only genotype present in all hospitals except for hospital G. Furthermore, it represented a high fraction of the isolates in several hospitals. Gt 4(17) was the most abundant genotype in hospital B (68.7%), A (71.4%), D (83.4%), and C (80%). At Al-Quds University, Gt 4(17) accounted for 40% of the isolates. Gt 6(18) was the second most abundant VACC1-genotype; it is closely related to GT 4(17), di ffering by just one additional repeat in the VNTR Lpms35. Gt 6(18) was endemic in the West Bank and found exclusively in hospital G, where it was the most abundant genotype (90.9%). Genotype Gt 10(93), a member of the newly-described VACC11, was found in hospitals C and F; it was isolated only once in hospital C, but was the most abundant genotype in hospital F (n = 15, 45.5%). The remaining genotypes that were found in more than one location had a rather restricted distribution, i.e., they were observed only in one or two additional sampling sites (Figure 2 and Table 5).

In general, the most frequent genotypes in each hospital were isolated repeatedly during samplings performed in following years. Gt 4(17) was recurrently isolated in hospitals A, B, C, D and F between 2012 and 2014. The endemic Gt 6(18) was isolated in hospital G in 2013 and 2014. Genotypes Gt 10(141), Gt 10(93), and Gt 9(92) were isolated in their respective sites (hospital F, B) from 2012 to 2014. The MLVA-8(12) genotypes were shared among the north, central and southern West Bank. According to the geographical distribution of the West Bank, 64 (35.6%), 43 (23.9%) and 73 (40.6%) isolates were isolated from northern, central and southern areas, respectively. Nevertheless, the most abundant and broadly distributed genotype was common to the whole West Bank, i.e., Gt 4(17), which comprised 68.8%, 53.5% and 9.6% of total isolates in northern, central and southern West Bank, respectively. Surprisingly, four genotypes, i.e., Gt 9(92), Gt 10(93), Gt 16(1) and Gt 63(83), were shared between more distant sites, i.e., northern and southern West Bank. As a tendency, the diversity of the genotypes observed decreased from the Southern to the Northern West Bank.

Interestingly, Gt 4(17) was never obtained from water samples, but only from biofilm. Genotypes of rather limited distribution were the only genotypes retrieved from water, i.e., Gt 10(93) and the endemic Gt 10(141), were obtained from hospital F, the endemic Gt 6(18) from G, and Gt 16(1) from A, respectively.

In summary, two thirds of the MLVA-8(12) genotypes were endemic, i.e., they were found exclusively in one hospital or the Al-Quds University. Only one third of the MLVA-8(12) genotypes were isolated in more than a single location, and these common genotypes were usually much more frequent in one of the locations. An exception was Gt 4(17), that was present in most locations and occurred often in high abundance.
