Cytotoxicity, Intracellular Replication, and Contact-Dependent Pore Formation of Genotyped Environmental Legionella pneumophila Isolates from Hospital Water Systems in the West Bank, Palestine

Legionella pneumophila is the causative agent of Legionnaires’ disease. Due to the hot climate and intermittent water supply, the West Bank, Palestine, can be considered a high-risk area for this often fatal atypical pneumonia. L. pneumophila occurs in biofilms of natural and man-made freshwater environments, where it infects and replicates intracellularly within protozoa. To correlate the genetic diversity of the bacteria in the environment with their virulence properties for protozoan and mammalian host cells, 60 genotyped isolates from hospital water systems in the West Bank were analyzed. The L. pneumophila isolates were previously genotyped by high resolution Multi Locus Variable Number of Tandem Repeat Analysis (MLVA-8(12)) and sorted according to their relationship in clonal complexes (VACC). Strains of relevant genotypes and VACCs were compared according to their capacity to infect Acanthamoeba castellanii and THP-1 macrophages, and to mediate pore-forming cytotoxicity in sheep red blood cells (sRBCs). Based on a previous detailed analysis of the biogeographic distribution and abundance of the MLVA-8(12)-genotypes, the focus of the study was on the most abundant L. pneumophila- genotypes Gt4(17), Gt6 (18) and Gt10(93) and the four relevant clonal complexes [VACC1, VACC2, VACC5 and VACC11]. The highly abundant genotypes Gt4(17) and Gt6(18) are affiliated with VACC1 and sequence type (ST)1 (comprising L. pneumophila str. Paris), and displayed seroroup (Sg)1. Isolates of these two genotypes exhibited significantly higher virulence potentials compared to other genotypes and clonal complexes in the West Bank. Endemic for the West Bank was the clonal complex VACC11 (affiliated with ST461) represented by three relevant genotypes that all displayed Sg6. These genotypes unique for the West Bank showed a lower infectivity and cytotoxicity compared to all other clonal complexes and their affiliated genotypes. Interestingly, the L. pneumophila serotypes ST1 and ST461 were previously identified by in situ-sequence based typing (SBT) as main causative agents of Legionnaires’ disease (LD) in the West Bank at a comparable level. Overall, this study demonstrates the site-specific regional diversity of L. pneumophila genotypes in the West Bank and suggests that a combination of MLVA, cellular infection assays and hierarchical agglomerative cluster analysis allows an improved genotype-based risk assessment.

A focus of the former study in the West Bank was the identification of environmental drivers influencing abundance and genotype composition. The study indicated a suppressive effect of high magnesium concentrations (>30 mg/L) on L. pneumophila abundance in water and biofilm in West Bank DWDS. Furthermore, according to their physico-chemical habitat conditions, the major genotypes were attributed to three different niches; three to four genotypes were sharing a common niche and were considered to represent a specific ecotype. Thus, environmental drivers exerted a different influence in response to the genotype [43,44].
Based on the previous studies investigating L. pneumophila abundance and genotype composition across the West Bank, we compared the virulence of 60 environmental isolates. Emphasis was on the characterization of three dominant MLVA-8 (12) genotypes and the four clonal complexes occurring in the West Bank. For virulence assessment three in vitro tests were used: cytotoxicity assays against Acanthamoeba castellanii and THP1 macrophages, and pore-forming mediated cytotoxicity using sheep red blood cells (sRBCs).

L. pneumophila Isolates
This study included 60 L. pneumophila environmental strains isolated from eight hospitals in the West Bank, Palestine (Table 1 and Supplementary Material Table S1). Genotyping was conducted using multilocus variable number of tandem repeat analysis using 13 loci-MLVA-13 (MLVA-12 plus 1 loci unique for the MLVA-8 scheme) designated as (MLVA-8 (12) [38,45] assigning the majority of the strains to three genotypes (Gt4 (17), Gt6 (18), and Gt10(93)) and four clonal complexes (VACC1, VACC2, VACC5 and VACC11) [38]. Table 1 describes the details of the environmental L. pneumophila isolates used for this study. As reference strains L. pneumophila Philadelphia-1 ATCC33152 T [46] and its Icm/Dot deficient dotA mutant [47,48] were used as positive and negative controls, respectively. Also, L. pneumophila str. Paris CIP107629 [49] was used as a positive control (Table 1). All tests of the strains were run as triplicates. A detailed protocol on the L. pneumophila inoculum preparation for the cytotoxicity tests was previously described [9].

Acanthamoeba castellanii Cytotoxicity Assay
The virulence of L. pneumophila isolates was measured by infecting A. castellanii (ATCC 50374) at multiplicity of infection (MOI) of 10 for 24 h as previously described [9,50]. The percentage of survived A. castellanii was calculated as (A. castellanii infected with L. pneumophila/A. castellanii concentration of positive control well) × 100%. Then, the percentage of killed A. castellanii was measured as (100%-the percentage of survived A. castellanii).
A. castellanii was cultured in proteose peptone-yeast extract-glucose extract (PYGE) medium in accordance with ATCC protocols (medium 712) and with the addition of 50 mL of 2 M glucose (filter sterilized) (Sigma, St. Louis, MO, USA). Detailed protocol was previously described [9].

THP-1 Cytotoxicity Assay
The virulence of L. pneumophila isolates was assessed by infecting THP-1 macrophages at a MOI of 10 for 24 h as previously described [9,11,14,45]. The relative degree of cytopathogenicity was expressed as percent of inhibition compared to non-infected cells; calculated as (Y = [(K − Y)/K] × 100). As K: mean OD of non-infected cells and Y: OD of infected cells. Bacterial density was assessed by the absorbance at 600 nm with a spectrophotometer Nanocolor Vis (Macherey-Nagel, Düren, Germany).

Pore-Forming Mediated Cytotoxicity Assay
The ability of L. pneumophila to lyse sRBCs was assessed at a multiplicity of infection (MOI) of 25 after 2 h of bacterial-sRBCs contact, as described previously [9,12,52]. The release of hemoglobin from lysed red blood cells was measured by spectrophotometry at 415 nm. Pore forming cytotoxicity was expressed as percentage of hemolysis compared to 100% fully hemolyzed blood cells.

Statistical Analysis
GraphPad Prism software v8.3.0 (Graph-Pad, San Diego, CA, USA) and Primer7 software (Primer-e, Auckland, New Zealand) were used for statistical analyses. Non-Normalized data were normalized. All tests were applied at a 95% and 99% level of confidence. All groups were normally distributed or normalized according to the Shapiro-Wilk test (p > 0.05). Variances were equal between groups at all temperatures (Leven's test, p > 0.05). Then, one-way ANOVA was performed to estimate statistical differences among virulence assays and between L. pneumophila genotypes and clonal complexes. Also, independent t-Test was performed to estimate differences among the three virulence assays. An agglomerative clustering dendrogram was achieved using the Primer7 software in order to study the similarity between virulence characteristics of L. pneumophila strains belonging to different genotypes (Gt4(17), Gt6 (18), and Gt10(93)) and clonal complexes (VACC1, VACC2, VACC5 and VACC11). The resemblance matrix was calculated using the Bray-Curtis index of association on the pathogenicity variables of the three different in vitro assays. Associations between MLVA-genotypes and clonal complexes were calculated using the Similarity Profile Analysis (SIMPROF) [53] based on Spearman rank correlation.

Virulence Characteristics of L. pneumophila MLVA-8(12) Genotypes
To investigate whether the most abundant L. pneumophila genotypes in the West Bank exhibit specific virulence characteristics, 60 L. pneumophila isolates from eight hospitals across the West Bank were analyzed for cytotoxicity (due to intracellular replication) against A. castellanii and THP-1 macrophages, and for contact-dependent pore formation in sRBCs. The presented comparison had a focus on strains MLVA-8 (12) genotypes Gt4(17), Gt6 (18) and Gt10(93) ( Table 1).
Compared to Gt4 (17) and Gt6 (18) affiliated to ST1, genotype Gt10(93) affiliated to ST461 exhibited the lowest cytotoxicity for A. castellanii infection and the lowest contactdependent pore formation for sRBC s (38.23 ± 4.38% and 21.8 ± 1.6% respectively) ( Table 1 and Figure 1). Since the data are statistically significant (One-Way ANOVA p ≤ 0.01), these results demonstrate that Gt10(93) (ST461) has a lower overall virulence in comparison with Gt4 (17) and Gt6(18) which are globally prevalent. The virulence did not differ significantly between Gt4(17) and Gt6 (18) isolates within all cytotoxicity tests ( Figure 1). The nonvirulent dotA mutant resulted in less than 15% cell death (Supplementary Material Table S1). A complete list of all isolates analyzed for virulence is provided in Supplementary Material Table S1.

A . c a s t e l l a n i i
The clinical reference strain L. pneumophila str. Paris (Gt 4(17), VACC1, ST1) was tested by the previously mentioned cytotoxicity assays. This reference strain had a comparable virulence pattern compared to the set of Gt4(17) strains from the West Bank (cytotoxicity for A. castellanii, THP-1 macrophages and contact-dependent pore formation for sRBC s 62.5 ± 5.9%, 56.8% ± 3.7 and 74.7 ± 4.9% respectively) (Supplementary Material Figure S1 and Supplementary Material Table S1).

Virulence Characteristics of L. pneumophila Clonal Complexes
The MLVA-8(12) genotypes sampled from biofilms in the West Bank cluster into four different clonal complexes (VACC1,2,5,11) [38]. To investigate whether the different VACCs correlate with respective virulence potentials of L. pneumophila, 60 environmental isolates were analyzed for cytotoxicity against A. castellanii and THP-1 macrophages, and for contact-dependent pore formation in sRBCs. The virulence analyses revealed statistically significant (ANOVA p ≤ 0.01) difference between the four VACCs with respect to cytotoxicity against A. castellanii, THP-1 macrophages and pore forming activity in sRBCs ( Figure 2).

Hierarchical Agglomerative Cluster Analysis
Significant differences were detected between the virulence characteristics of environmental isolates belonging to different genotypes and clonal complexes (Figures 1  and 2). Gt10(93) isolates were significantly less cytotoxic toward A. castellanii, THP-1 macrophages and sRBC´s than Gt4 (17) and Gt6 (17) isolates. Furthermore, VACC11 presented a unique virulence profile with respect to cytotoxicity toward A. castellanii, THP-1 macrophages and sRBC´s ( Figure 2). Hierarchical agglomerative cluster analysis revealed that isolates belonging to the Gt10(93) and VACC11 clustered together almost homogenously with a genotype and clonal complex-dependent virulence profile (  Table S1). This means that not only Gt10(93), which is the most abundant genotype in VACC11, but also the other genotypes in the same clonal complex had comparable virulence characteristics (Supplementary Material Figure S2). This suggests that VACC11 genotypes seem to be less virulent according to the tested traits. The non-VACC11 genotypes and clonal complexes did not form specific clusters according to their genotypes and cluster. Thus, strains affiliated with genotypes of VACC 1, 2 and 5 could not be characterized by a distinguished set of virulence traits, in contrast to VACC11. In terms of virulence against A. castellanii, VACC1 and VACC2 showed the highest activities. In terms of significance (Independent t-Test p ≤ 0.05), the activity of VACC1 and VACC2 was significantly higher than the activity of VACC11.
In terms of cytotoxicity against macrophages showed VACC2 the highest activity; however, only the activity of VACC11 was significantly lower than VACC2.
In terms of pore forming activity in sRBCs, VACC1 and VACC5 showed the highest activities at a comparable level; by far the lowest activities were observed for VACC11. As a negative control, the non-virulent dotA mutant resulted in less than 15% cell death (Supplementary Material Figure S1). Detailed data for individual isolates are provided in Supplementary Material Table S1.
The newly identified clonal complex in the West Bank VACC11 and its genotype Gt10(93) seemed to be significantly less cytotoxic towards amoebae at 37 • C (after a 24-h infection). Also, cytopathogenicity against THP-1 macrophages and hemolytic activity were significantly less compared to VACC1 and its genotypes Gt4 (17) and Gt6 (18), and in comparison to VACC2 and VACC5 (Figure 2).
Taken together, these results showed that VACC1-genotypes had consistently high virulence activities; by contrast the newly identified VACC11 had lower virulence characteristics in comparison with the ubiquitous clonal complexes VACC1, VACC2 and VACC5 (Figure 2).

Hierarchical Agglomerative Cluster Analysis
Significant differences were detected between the virulence characteristics of environmental isolates belonging to different genotypes and clonal complexes (Figures 1 and 2). Gt10(93) isolates were significantly less cytotoxic toward A. castellanii, THP-1 macrophages and sRBC s than Gt4 (17) and Gt6 (17) isolates. Furthermore, VACC11 presented a unique virulence profile with respect to cytotoxicity toward A. castellanii, THP-1 macrophages and sRBC s (Figure 2). Hierarchical agglomerative cluster analysis revealed that isolates belonging to the Gt10(93) and VACC11 clustered together almost homogenously with a genotype and clonal complex-dependent virulence profile ( Figure 3). VACC11 (Figure 3b) is abundant in 60% of Gt10(93), 20% of Gt10(141) and 20% of Gt9(92) (Supplementary Material Table S1). This means that not only Gt10(93), which is the most abundant genotype in VACC11, but also the other genotypes in the same clonal complex had comparable virulence characteristics (Supplementary Material Figure S2). This suggests that VACC11 genotypes seem to be less virulent according to the tested traits. The non-VACC11 genotypes and clonal complexes did not form specific clusters according to their genotypes and cluster. Thus, strains affiliated with genotypes of VACC 1, 2 and 5 could not be characterized by a distinguished set of virulence traits, in contrast to VACC11.   (93)) and (b) clonal complex (VACC1, VACC2, VACC5, and VACC11). The resemblance matrix was calculated using the Bray-Curtis index of association on cytotoxicity characteristics variables, i.e., cytotoxicity against A. castellanii and THP-1 macrophages, and pore forming mediated cytotoxicity.

Discussion
There is a wealth of evidence that climate change is with a significant impact on the water resources in the West Bank [54]. Among many waterborne pathogens, L. pneumophila can be expected to benefit from elevated temperatures in natural and manmade water systems [43,55]. Therefore, studies on ecology and virulence of L. pneumophila are a crucial basis for future management of DWDS.  (17), Gt6 (18), and Gt10(93)) and (b) clonal complex (VACC1, VACC2, VACC5, and VACC11). The resemblance matrix was calculated using the Bray-Curtis index of association on cytotoxicity characteristics variables, i.e., cytotoxicity against A. castellanii and THP-1 macrophages, and pore forming mediated cytotoxicity.

Discussion
There is a wealth of evidence that climate change is with a significant impact on the water resources in the West Bank [54]. Among many waterborne pathogens, L. pneumophila can be expected to benefit from elevated temperatures in natural and man-made water systems [43,55]. Therefore, studies on ecology and virulence of L. pneumophila are a crucial basis for future management of DWDS.
To elucidate specific infection routes of L. pneumophila from the environment to the human population, tracking of genotypes and clonal complexes have proven useful [38,43]. However, for a better preparedness, a combined knowledge of L. pneumophila ecology, transmission and virulence is needed. In the past, the degree of L. pneumophila virulence was successfully determined by analyzing the infectivity and cytopathogenicity for A. castellanii or macrophage-like cells. Other approaches included the induction of poreformation mediated cytotoxicity of host cells [9,10,26,56,57].
In the present study, we combined L. pneumophila genotyping with these approaches and were able to show that isolates belonging to different L. pneumophila genotypes and VACCs can differ markedly in virulence (Figures 1 and 2). Altogether, 60 environmental L. pneumophila isolates from the most abundant genotypes (Gt4(17), Gt6 (18) and Gt10(93)) and the clonal complexes VACC1, VACC2, VACC5, and VACC11 were analyzed with respect to cytotoxicity for A. castellanii, THP-1 macrophages and pore-forming mediated cytotoxicity in sRBCs. Significant differences were observed between L. pneumophila genotypes and VACCs. L. pneumophila Gt4(17) and Gt6(18) affiliated with VACC1 and ST1, isolated from six hospitals (A-F) and hospital G, respectively, showed the highest virulence (Figure 1). These results are in accordance with a previous study [58] demonstrating that ST1 strains of L. pneumophila can be highly virulent, in addition to their worldwide high prevalence. Genotypes Gt4 (17) and Gt6 (18) are affiliated with ST1 and VACC1, the most abundant clonal complex and sequence type in the West Bank and worldwide. VACC11 is a clonal complex identified for the first time in the West Bank by Zayed et al. [38]. This clonal complex exhibits a comparably low virulence potential (Figure 2).
According to epidemiological studies worldwide [59,60] ST1 (includes Gt4 (17) and Gt6 (18)) is one of the most virulent ST that have been described for L. pneumophila strains since it is one of the most frequent cause of LD globally; observations of this study suggest that this may also be valid for the West Bank.
Agglomerative clustering dendrograms revealed that virulence profiles for genotypes were rather different between the relevant VACC11-genotype Gt10(93) on one hand, and on the relevant VACC1-genotypes (Gt4 (17) and Gt6 (18)) on the other hand, in the West Bank (Figure 3). This discrepancy was also observed when comparing a larger set of genotypes: VACC11-genotypes clustered together and separate from the other clonal complexes. Figure 3 also shows a distinct cluster for VACC11, but a mixed cluster for all genotypes of the other clonal complexes VACC1, 2 and 5.
Our results demonstrated that there was no significant difference in the cytotoxicity toward THP-1macrophages between the genotypes and VACCs of L. pneumophila strains except between VACC2 and VACC11. However, a statistically significant difference was seen between L. pneumophila genotypes and VACCs for A. castellanii infection and pore forming cytotoxicity. An explanation might be that environmental isolates can be expected to have more pronounced genotype-dependent environmentally relevant defense traits.
The resolution of MLVA-8(12) genotyping applied in this study for L. pneumophila strains allowed the classification and comparison of the pathogenicity potential of each genotype in order to determine which genotype poses the greatest risk to public health. MLVA-8(12) is very useful because it allows classification within the highly health-relevant and globally distributed ST1, comprising the reference strain L. pneumophila strain Paris [41]. The classical SBT does not distinguish below the level of ST1. However, especially for this globally highly relevant sequence type a higher distinction is needed.
L. pneumophia from clinical and environmental samples, collected from suspected pneumonia patients and from biofilm samples of different wards in one hospital in East Jerusalem. L. pneumophila was detected in 35% of Bronchoalveolar lavage (BAL) samples and 15% of sputum samples using conventional PCR. By using Nested PCR sequencebased typing (NPSBT), 29% of clinical samples genotyped ST1 and 21% genotyped ST461. Jaber et al. [32] findings support our idea that the information of environmental samples in combination with genotyping at high resolution level (as MLVA-8 (12)) are needed to understand the epidemiology of legionellosis in a certain geographical area.
A study by Sharaby et al. [9] analyzing clinical and environmental L. pneumophila isolates from northern Israel, showed that Gt4 virulence are in concordance with our results with respect to a high cytotoxicity towards A. castellanii and THP-1 macrophages. Half of the Israeli clinical isolates and a major fraction of the environmental strains were Gt4 and Gt6 (both affiliated with ST1 and VACC1) indicating the clinical importance and environmental relevance of these genotypes and ST1 also in DWDS of Israel. The corner stone between our study and Sharaby et al. is a "genotype-site specificity", i.e., specific genotypic groups occurring in a DWDS in Oranim campus (Haifa city) [9,37], had specific ecotype characteristics (i.e., describing a set of strains of L. pneumophila inhabiting a specific niche) [44], comparable to our findings in the West Bank [38,43]. Therefore, "genotype-site specificity" for L. pneumophila strains may include a set of genotype-dependent specific traits, such as pathogenicity, virulence, epidemiologically relevant and ecological characteristics with a specific local "imprint".
A study by Sousa et al. [62] looked at differences between L. pneumophila isolates from clinical, man-made and natural environmental samples in Galleria mellonella infection models in which they found that all strains proved to be pathogenic. They observed that some of the strains were defective in their ability to cause disease while others were highly pathogenic. Sousa et al. [62] concluded from their findings that some L. pneumophila strains' ability to cause disease is more related to their capability to persist and bloom in manmade environmental niches, which kind of mimics human infection, and less dependent on their virulence [62,63]. This observation is of relevance because environmental L. pneumophila strains are reservoirs and source of sporadic cases of legionellosis [64,65]. Another important aspect is how L. pneumophila strains behave in the environment, select their niche or co-inhabit niches as genotype consortia, and how this affects the virulence of specific genotypes. Both biotic and abiotic factors may influence structure and dynamics of L. pneumophila populations as shown by many authors in DWDS and more specifically for the West Bank [37,43,66]. Though, it is well known that amoeba are an important training ground for L. pneumophila's pathogenicity, little is known of the influence of other biotic and abiotic factors, including interactions among different genotypes on the resulting virulence of L. pneumophila [11,17].

Relevance of the Findings for the West Bank and Beyond
Jaber et al. [32] showed that there was a high risk of lung infection in the West Bank due to L. pneumophila as indicated by the high percentage of infected pneumonia patients. Furthermore, they have shown that the sequence types most relevant for L. pneumophila caused pneumonia were ST1 (29%) and ST461 (21%).
Relevance for LD can be assumed for the ST1-genotypes due to their virulence traits. Less expected is the important role of the ST461-genotypes for LD in the West Bank due to their lower virulence profile. However, virulence traits and the overall infection processes are rather complex phenomenons. In addition, there are other aspects contributing to the risk of LD, such as the infective dose (see Quantitative Microbial Risk Assessment (QMRA) below).
Another aspect for risk of L. pneumophila infections is its complex ecology leading to a rather different concentration of L. pneumophila in DWDS. As shown in previous studies [36][37][38]45], L. pneumophila displayed a genotype-dependent preference with respect to habitat characteristics (ecotype), leading to a specific genotype pattern and abundance in water and biofilm. Since infection is not only dependent on virulence traits but also on the infective dose, the ecology exerts a direct impact on the infection risk. QMRA addresses this risk as a well-established tool [67]. The genotype-dependence of ecology and virulence of this study and previous ones in the West Bank and Israel [9,[36][37][38]45] suggest that modeling of abundance and risk (QMRA) might be more precise when applied on the genotype level of L. pneumophila. MLVA-8(12)-genotyping may provide a good level of resolution to address these issues.
Source tracking based on the MLVA-genotypes could be furthermore helpful to assess the sources of infection in the West Bank and of great value for risk management.

Conclusions and Future Research
This is the first study in which virulence characteristics of environmental L. pneumophila isolates from the West Bank were compared. Overall, we observed clustering of specific genotypes and VACCs with specific virulence potentials. The presented results suggest that information on virulence characteristics in combination with genotyping at an adequate resolution level (such as MLVA) is helpful to improve public health management and risk assessment measures.
Most of the L. pneumophila isolates were retrieved from the biofilm of the DWDS in the West Bank. In this set of isolates no significant difference in terms of virulence between strains derived from biofilm and bulk water could be assessed, in consistence with a previous study from Israel [9]. However, exchange processes between biofilm and water for L. pneumophila are complex and would need more detailed investigation on the genotype level. The genotype-dependent influence of Magnesium on L. pneumophila abundance may play in this respect a special role in the West Bank, and may be worth a closer look also for other geographic and climatic regions.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/pathogens10040417/s1, Figure S1: The virulent activity of L. pneumophila reference strains (L. pneumophila str. Paris, L. Pneumophila str. Philadelphia-1 and its Icm/Dot deficient dotA mutant as a negative control) during post exponential phase was determined by three cytotoxicity tests; cytotoxicity against A. castellanii or THP-1 macrophages and pore forming mediated cytotoxicity of sRBC's, Figure S2: Agglomerative clustering dendrogram representing the percentage of similarity between cytotoxicity characteristics of L. pneumophila isolates belonging to different genotypes (Gt4 (17), Gt6(18), Gt10(93), Gt10(141), and Gt9 (92)). The resemblance matrix was calculated using the Bray-Curtis index of association on cytotoxicity characteristics variables against A. castellanii cytotoxicity, cytopathogenicity of THP-1 macrophages and pore forming mediated cytotoxicity, Table S1: List of L. pneumophila strains (n = 60) isolated from the West Bank analyzed in this study by different in vitro cytotoxicity tests. L. pneumophila reference strains included. Water isolates are highlighted in bold.