Intracellular Behaviour of Three Legionella pneumophila Strains within Three Amoeba Strains, Including Willaertia magna C2c Maky

Legionella pneumophila is a facultative intracellular pathogen found in aquatic environments as planktonic cells within biofilms and as intracellular parasites of free-living amoebae such as Acanthamoeba castellanii. This pathogen bypasses the elimination mechanism to replicate within amoebae; however, not all amoeba species support the growth of L. pneumophila. Willaertia magna C2c Maky, a non-pathogenic amoeba, was previously demonstrated to possess the ability to eliminate the L. pneumophila strain Paris. Here, we study the intracellular behaviour of three L. pneumophila strains (Paris, Philadelphia, and Lens) within W. magna C2c Maky and compare this strain to A. castellanii and W. magna Z503, which are used as controls. We observe the intracellular growth of strain Lens within W. magna Z503 and A. castellanii at 22 °C and 37 °C. Strain Paris grows within A. castellanii at any temperature, while it only grows at 22 °C within W. magna Z503. Strain Philadelphia proliferates only within A. castellanii at 37 °C. Within W. magna C2c Maky, none of the three legionella strains exhibit intracellular growth. Additionally, the ability of W. magna C2c Maky to decrease the number of internalized L. pneumophila is confirmed. These results support the idea that W. magna C2c Maky possesses unique behaviour in regard to L. pneumophila strains.


Introduction
Legionella pneumophila is an aerobic, Gram-negative bacterium that causes Legionellosis, a severe form of pneumonia, following inoculation with contaminated aerosol [1]. This bacterial infection manifests as two clinical forms that include Legionnaires' disease, which is a life-threatening respiratory disease, and Pontiac fever, a milder self-limiting illness [2,3]. Among the sixteen currently identified serogroups of L. pneumophila, serogroup 1 is involved in the majority of infections [4,5]. This microorganism is ubiquitous throughout natural and artificial aquatic environments [6]. Legionellosis outbreaks are frequently related to contaminated water systems that produce aerosols, which occurs primarily within cooling towers [7]. Indeed, cooling towers provide ideal conditions for pathogen growth, as they frequently possess temperatures above 20 • C, at which L. pneumophila can proliferate [8][9][10].

Amoeba Survival in coculture Medium
Survival of the three amoebas in the presence or in absence of bacteria was evaluated over 96 h at 22 °C and 37 °C in coculture medium (Figure 2a,b). The three amoeba strains could be maintained in SCYEM medium for 96 h in the presence or absence of bacteria at 22 °C and 37 °C with the exception of A. castellanii when co-cultivated with L. pneumophila strains. Found at the end of the experiment, the control condition of A. castellanii in the absence of bacteria was maintained at 2 × 10 5

Amoeba Survival in coculture Medium
Survival of the three amoebas in the presence or in absence of bacteria was evaluated over 96 h at 22 • C and 37 • C in coculture medium (Figure 2a,b). The three amoeba strains could be maintained in SCYEM medium for 96 h in the presence or absence of bacteria at 22 • C and 37 • C with the exception of A. castellanii when co-cultivated with L. pneumophila strains. Found at the end of the experiment, the control condition of A. castellanii in the absence of bacteria was maintained at 2 × 10 5 cells/mL, while in the presence of L. pneumophila Lens, Paris, and Philadelphia, the amoeba number decreased to 556, 444 and 2333 cells/mL, respectively (Figure 2b). A. castellanii could not survive in the presence of the three L. pneumophila strains at 37 • C.

Amoeba Survival in coculture Medium
Survival of the three amoebas in the presence or in absence of bacteria was evaluated over 96 h at 22 °C and 37 °C in coculture medium (Figure 2a,b). The three amoeba strains could be maintained in SCYEM medium for 96 h in the presence or absence of bacteria at 22 °C and 37 °C with the exception of A. castellanii when co-cultivated with L. pneumophila strains. Found at the end of the experiment, the control condition of A. castellanii in the absence of bacteria was maintained at 2 × 10 5 cells/mL, while in the presence of L. pneumophila Lens, Paris, and Philadelphia, the amoeba number decreased to 556, 444 and 2333 cells/mL, respectively (Figure 2b). A. castellanii could not survive in the presence of the three L. pneumophila strains at 37 °C.   The mean initial amount of amoeba-internalized bacteria at 22 • C was 16 ± 0.5% (16% in A. castellanii, 15% in W. magna C2c Maky, and 16% in W. magna Z503). Seen at 37 • C, a mean bacterial uptake of 20 ± 5.5% was observed (15% in A. castellanii, 26% in W. magna C2c Maky, and 18% in W. magna Z503).
A significant decrease (p < 0.05) in the number of intracellular L. pneumophila Lens per W. magna C2c Maky cell was observed after 24 h (5-fold and 10-fold reduction at 22 • C and 37 • C, respectively), while the level remained nearly constant for A. castellanii at 22 • C and 37 • C and for W. magna Z503 at 22 • C with no significant difference between T 0 and T 0 + 24 h (p > 0.05) (Figure 3). Occurring at T 0 + 96 h (Figure 3), the percentage of intracellular L. pneumophila Lens per W. magna C2c Maky cell was reduced by 48 ± 0.3% at 22 • C and 77 ± 1.2% at 37 • C, and an increase was observed for W. magna Z503 (9-fold at 22 • C and 5-fold at 37 • C) and A. castellanii (19-fold at 22 • C and 50,000-fold at 37 • C). Observed at 37 • C, a small number of A. castellanii cells were still alive (5.6 × 10 2 ± 5.9 × 10 2 amoebas/mL), demonstrating that amoeba cell lysis occurred following the intracellular multiplication of L. pneumophila Lens.  (Figure 3). Occurring at T0 + 96 h (Figure 3), the percentage of intracellular L. pneumophila Lens per W. magna C2c Maky cell was reduced by 48 ± 0.3% at 22 °C and 77 ± 1.2% at 37 °C, and an increase was observed for W. magna Z503 (9-fold at 22 °C and 5-fold at 37 °C) and A. castellanii (19-fold at 22 °C and 50,000-fold at 37 °C). Observed at 37 °C, a small number of A. castellanii cells were still alive (5.6 × 10 2 ± 5.9 × 10 2 amoebas/mL), demonstrating that amoeba cell lysis occurred following the intracellular multiplication of L. pneumophila Lens.  Considering the number of L. pneumophila Lens at 22 • C and 37 • C, a significant increase (p < 0.05) was obtained when the bacterium was co-cultivated with W. magna Z503 and A. castellanii, and this was not observed when L. pneumophila Lens was cultivated alone or in the presence of W. magna C2c Maky (Figure 4a,b), demonstrating an intracellular multiplication of L. pneumophila Lens in W. magna Z503 and A. castellanii as the bacterium was unable to multiply by itself in the coculture medium (Figure 1a
A significant decrease of the number of intracellular L. pneumophila Paris per amoeba cell (p < 0.05) first was observed in the three amoebas after 24 h, with the exception of A. castellanii at 37 • C (8-fold for W. magna C2c Maky, 3-fold for W. magna Z503, and 9-fold for A. castellanii at 22 • C and 19-fold for W. magna C2c Maky, 11-fold for W. magna Z503, and 2-fold for A. castellanii at 37 • C) ( Figure 3). This decrease was maintained until the end of the experiment (T 0 + 96 h) only by W. magna C2c Maky, and the percentage of intracellular L. pneumophila Paris per amoeba cell was reduced by 79 ± 2% at 22 • C and 98 ± 0.1% at 37 • C (p < 0.05). The opposite was observed for W. magna Z503 and A. castellanii at 22 • C and 37 • C, as the decrease measured after 24 h was not maintained. Seen at 48 h, the level of intracellular L. pneumophila Paris per amoeba cell began to increase until it reached 4-fold and 3-fold more bacteria per amoeba cell than that observed at T 0 for W. magna Z503 and A. castellanii, respectively at 22 • C. Observed at 37 • C for W. magna Z503, the number of intracellular L. pneumophila Paris per amoeba cell at T 0 + 96 h was 5-fold the ratio observed at 24 h, but it did not reach the initial ratio. Regarding A. castellanii, a strong increase was observed at both temperatures, and the initial ratio was slightly increased by 3-fold at 22 • C (p > 0.05) and strongly increased by 60,000-fold at 37 • C (p < 0.05). Furthermore, the correlation between the increase in L. pneumophila Paris and the low concentration of viable A. castellanii (5.6 × 10 2 ± 5.9 × 10 2 cells/mL) after 96 h indicated that a high intracellular multiplication of L. pneumophila Paris occurred that was followed by a release of bacteria in the medium after A. castellanii death.
Considering the number of L. pneumophila Paris at 22 • C, a significant increase (p < 0.05) was obtained when the bacterium was co-cultured with W. magna Z503 and A. castellanii, and this was not observed when L. pneumophila Paris was cultured alone or in the presence of W. magna Z503 at 37 • C and W. magna C2c Maky at both 22 • C and 37 • C (Figure 4c,d), demonstrating an intracellular multiplication of L. pneumophila Paris in W. magna Z503 and A. castellanii at 22 • C and only in A. castellanii at 37 • C as the bacterium was unable to multiply by itself in the coculture medium (Figure 1a,b).
Occurring at 22 • C, a rapid and significant (p < 0.05) decrease in the number of intracellular L. pneumophila per amoeba cell was observed within 24 h (20-fold for A. castellanii, 11-fold for W. magna C2c Maky, and 10-fold for W. magna Z503) in the three amoebas ( Figure 3). Then, a slow but significant (p < 0.05) decrease continued until the death of more than 99% of intracellular L. pneumophila Philadelphia in all cases. Even if this decrease could be attributed to the bacterial death in the coculture medium, the experiment demonstrated the absence of intra-amoeba multiplication of L. pneumophila Philadelphia necessary for survival at 22 • C.
Occurring at 37 • C, a similar rapid decrease in the number of intracellular L. pneumophila per amoeba was observed within 24 h for all three amoebas (20-fold for A. castellanii, 10-fold for W. magna C2c Maky, and 92-fold for W. magna Z503). Then, differential behaviours were observed depending on the amoeba strains. Regarding W. magna C2c Maky, the significant decrease (p < 0.05) continued until the death of more than 99.99% of the intracellular L. pneumophila Philadelphia per amoeba cell (Figure 3d). Concerning W. magna Z503, a decrease also was observed up to 97% elimination of intracellular L. pneumophila Philadelphia per amoeba cell after 96 h (p < 0.05) (Figure 3d). To contrast, for A. castellanii, a significant increase (p < 0.05) in intracellular L. pneumophila Philadelphia per amoeba cell appeared after 48 h, demonstrating an intra-amoeba multiplication up to 2600-fold at the end point ( Figure 3c).
Considering the number of L. pneumophila Philadelphia at 22 • C, a significant decrease (p < 0.05) was obtained in all cases (Figure 4e), while at 37 • C, a significant increase (p < 0.05) was observed when L. pneumophila Philadelphia was cultured in the presence of A. castellanii (Figure 4f). This demonstrated an intracellular multiplication of L. pneumophila Philadelphia A. castellanii at 37 • C, as the bacterium was unable to multiply by itself in SCYEM medium (Figure 1a,b).

Microscopic Observations of Intracellular L. pneumophila Philadelphia at 37 • C
Microscopic observations were performed at T 0, T 0 + 48 h, and T 0 + 96 h. Occurring at T 0, excess intracellular L. pneumophila Philadelphia bacteria were observed in the presence of the three amoebas ( Figure 5A,D,G). Regarding A. castellanii at 48 h, a strong bacterial multiplication was observed ( Figure 5B) which was not observed for both W. magna strains ( Figure 5E,H). Occurring at 96 h, lysis of A. castellanii after intracellular bacterial multiplication was clearly evident (Figure 5C), and only a small amount of amoeba lysis could be observed for both W. magna strains ( Figure 5F,I).

Statistical Comparison of Amoeba Behavior
Analysis of variance tests (ANOVA) were performed to determine if W. magna C2c Maky interacted with L. pneumophila in a significantly different manner compared to interactions with the two other amoebas.
Concerning the three bacterial strains, T 0 data obtained in the presence of the three amoebas were not statistically different at 22 • C (p > 0.05); however, at 37 • C, a significant difference in behaviour (p < 0.05) was detected at T 0 .
Pairwise the bacterium was unable to multiply by itself in SCYEM medium (Figure 1a,b).

Microscopic Observations of Intracellular L. pneumophila Philadelphia at 37 °C
Microscopic observations were performed at T0, T0 + 48 h, and T0 + 96 h. Occurring at T0, excess intracellular L. pneumophila Philadelphia bacteria were observed in the presence of the three amoebas ( Figure 5A,D,G). Regarding A. castellanii at 48 h, a strong bacterial multiplication was observed ( Figure 5B) which was not observed for both W. magna strains ( Figure 5E,H). Occurring at 96 h, lysis of A. castellanii after intracellular bacterial multiplication was clearly evident (Figure 5C), and only a small amount of amoeba lysis could be observed for both W. magna strains ( Figure 5F,I).

Statistical Comparison of Amoeba Behavior
Analysis of variance tests (ANOVA) were performed to determine if W. magna C2c Maky interacted with L. pneumophila in a significantly different manner compared to interactions with the two other amoebas.
Concerning the three bacterial strains, T0 data obtained in the presence of the three amoebas were not statistically different at 22 °C (p > 0.05); however, at 37 °C, a significant difference in behaviour (p < 0.05) was detected at T0.

Discussion
This work explores the permissiveness of three amoeba strains regarding the intracellular multiplication of three pathogenic L. pneumophila strains under two temperature conditions (22 • C and Pathogens 2020, 9, 105 9 of 15 37 • C) that correspond to temperatures found in cooling towers in which L. pneumophila are known to replicate within certain strains of amoebae [10,25]. It is important to demonstrate that W. magna C2c Maky does not multiply L. pneumophila as we aim to propose it as a natural biocide to treat cooling towers.
The three L. pneumophila strains are a representative set of L. pneumophila serogroup 1 that is responsible for 95% of the legionellosis disease world-wide [5]. Strain Philadelphia is a clinical isolate that is historically responsible for the very first outbreak. It possesses gene traits that allow for multiplication in a number of hosts such as peripheral blood mononuclear cells, peritoneal macrophages, and A. castellanii, A. polyphaga, or A. lenticulate [26][27][28][29]. The Philadelphia strain is, according to the EN 13623 European standard, the only strain for which testing is required to validate a disinfectant against Legionella in Europe. L. pneumophila Lens was chosen because it was responsible for an outbreak in the north of France between November 2003 and January 2004 where 86 confirmed cases resulted in 17 deaths [30]. L. pneumophila Paris was chosen because, among the endemic strains of L. pneumophila serogroup 1, sequence type 1 (ST1) strains are among the most prevalent, particularly the ST1/Paris pulsotype. This endemic type was responsible for 8.2% of French culture-proven cases of Legionnaire's disease from 1995 through 2006. ST1/Paris pulsotype isolates also have been detected in clinical and environmental samples taken from several other countries around the world, including Switzerland, Italy, Spain, Sweden, the United States, Japan, Senegal, and Canada [21,30].
Our experiments demonstrate differential behaviours among amoeba species infected by the pathogenic bacteria. Compared to A. castellanii and W. magna Z503, the intracellular L. pneumophila are efficiently eliminated by W. magna C2c Maky at 22 • C and 37 • C. Indeed, the experiments report not only a non-replication, but also an elimination of the intracellular strains Lens, Paris and Philadelphia within W. magna C2c Maky. Furthermore, the coculture medium used in the survey is not adapted to the survival of the legionella bacteria, and they, therefore, must parasitize the amoebae to facilitate their own growth. Indeed, the experiments demonstrate that the three legionella strains were unable to remain at the inoculation level and began to die after 24 h (Figure 1). Although the medium is not adapted to L. pneumophila strains, it was chosen for the co-culture study because an increase of the bacterial number during the co-culture experiment necessarily indicates that the multiplication occurred within amoeba. The bacterial multiplication is observed both in A. castellanii and W. magna Z503, and it is not observed in W. magna C2c Maky. The assays reveal a multiplication of all legionella strains within A. castellanii at 37 • C and the intracellular multiplication of strain Lens and Paris at 22 • C. Indeed, the strain Philadelphia grows at 37 • C ( Figure 3c) and does not multiply at 22 • C (Figure 3a) within A. castellanii. Based on this, these results suggest a behaviour that is influenced by the temperature conditions. Several previous studies revealed the effect of temperature on the relationship between L. pneumophila and free-living amoeba (FLA) [9,31,32]. L. pneumophila serogroup 1, for example, replicated in A. castellanii at 25 • C but were digested at temperatures below 20 • C [25]. Dupuy et al. assessed the ability of 12 amoeba strains of Naegleria sp., Acanthamoeba sp., and Vermamoeba sp. to support the multiplication of L. pneumophila Lens at various temperatures (25 • C, 30 • C and 40 • C), and they revealed a more efficient intracellular proliferation with increasing temperatures [33]. Additionally, we did not observe the same behaviour according to the different bacteria and amoeba strains used during our experiments. Indeed, the strain Lens replicates at 37 • C within W. magna strain Z503, but not in W. magna C2c Maky (Figure 3d). The co-culture at 22 • C of W. magna Z503 with L. pneumophila strain Paris and strain Lens reveals a multiplication of the bacteria; however, no replication is observed during co-culture with strain Philadelphia (Figure 3b). The difference in amoeba permissiveness has been highlighted previously, especially in regard to Naegleria, Acanthamoeba, Vermamoeba and Micriamoeba tesseris [9,34]. The non-replication of legionella within W. magna C2c Maky was previously observed with strain Paris [20]. Our study confirms this result, as the resistance of W. magna C2c Maky towards L. pneumophila Paris is illustrated by the observed significant decrease in the bacterial concentration after 4 days of co-culture at 22 • C and 37 • C (Figure 4c,d). Dey et al. [20], however, reported a moderate increase in strains Philadelphia and Lens within W. magna C2c at 37 • C while in our study the intracellular bacterial concentration significantly decreased in culture with W. magna C2c Maky at 22 • C and 37 • C. These differences can be explained by the protocol parameters used in the former study, particularly regarding the culture medium and elimination of extracellular bacteria. The authors used serum casein glucose yeast extract medium (SCGYEM) that was favourable to L. pneumophila survival, so bacteria were not forced to multiply into amoeba to survive. Additionally, Dey and co-workers did not eliminate extracellular bacteria by centrifugation, and the observed increase could be due to extracellular bacterial replication, such as that resulting from necrotrophic growth as previously demonstrated [35].
W. magna C2c Maky is demonstrated to possess a high efficiency for digesting the intracellular L. pneumophila cells in all strains used in this survey. The growth of L. pneumophila within amoebas is known to enhance the pathogenicity and invasion of L. pneumophila [15,36]; however, no intracellular bacterial replication is observed when we infect W. magna C2c Maky with L. pneumophila strains derived from a first co-culture that was thought to be more virulent (unpublished data).
The action on different L. pneumophila strains and the absence of internal proliferation support the fact that W. magna C2c Maky could be used as a biocide to combat L. pneumophila proliferation in cooling tower water. This observation is consistent with the control of legionella by W. magna C2c Maky observed in real conditions during field trials in functioning cooling towers (http://www.amoeba-biocide.com/ sites/default/files/180711_cp_amoeba_us_positive_efficacy_field_test_en_vedf_0.pdf). The traditional method to control bacterial growth in cooling tower water is primarily based on the use of chemical biocides [37,38]. Indeed, the oxidizing agent chlorine is the most used product for cooling tower treatment [39]. The chemical biocide is efficient to prevent L. pneumophila proliferation, although some previous studies reported incomplete eradication of legionella from installations and progressive re-colonization within these systems within weeks or months [40,41]. Moreover, these chemical biocides are dangerous to the environment, they degrade the installation systems, and they require the application of other products such as anti-corrosive agents [42,43]. Described by Iervolino, treatment with another oxidizing agent (H 2 O 2 /Ag) was inadequate for legionella control, and, instead, it caused a rapid increase of one logarithmic unit [44]. Chemical biocide action also is not completely efficient against biofilms and amoeba cysts that can provide protection against disinfection treatment [16,17,45]. Finally, chemical biocides used in cooling towers can select L. pneumophila populations, and chemical biocides can promote resistance to biocides and to human health antibiotics [46,47].
To conclude, W. magna C2c Maky is not associated with any human or animal infection, and this is in agreement with the lack of pathogenicity demonstrated in vivo and suggested by genomic analysis [24,48]. This organism is likely a safe and efficient candidate for legionella control in cooling towers and could provide an alternative solution to chemical biocides.

Free-Living Amoebae Culture
Willaertia magna C2c Maky (ATCC ® PTA-7824), Willaertia magna Z503 (ATCC ® 50035), and Acanthamoeba castellanii (ATCC ® 30010) were purchased from ATCC and cultivated according to their recommendation into 10 mL of modified PYNFH medium (ATCC medium 1034) in a T-25 tissue culture flask. Amoebae were then grown in cell factories in serum casein yeast extract medium (SCYEM) at 30 • C. SCYEM medium is derived from serum casein glucose yeast extract medium (SCGYEM) medium [49] and contained 10 g·L −1 casein, 5 g·L −1 yeast extract, 10% foetal calf Serum, 1.325 g·L −1 Na 2 HPO 4 , and 0.8 g·L −1 KH 2 PO 4 . After 72 h (during exponential phase), the cell factories were gently shaken, and the amoeba suspensions were transferred to 50 mL Falcon ® tubes. Amoeba populations were then quantified using a Malassez haemocytometer cell counting chamber method (Thermo Fisher Scientific, France) with Trypan blue by mixing 100 µL of Trypan blue with 100 µL of amoeba sample. According to the results, the amoebae concentration in Falcon ® tubes was then adjusted to 3 × 10 5 cells/mL by the addition of SCYEM. The amoebas were then washed twice in SCYEM using centrifugation at 3000× g for 10 min, and the supernatants were then discarded. Amoeba populations were then re-quantified, and the amoeba suspensions were finally adjusted to 3 × 10 5 cells/mL in 100 mL of SCYEM. A final quantification was performed to verify the concentration.

Legionella Pneumophila Cultures
L. pneumophila strain Philadelphia (ATCC 33152), L. pneumophila strain Lens (CIP 108280), and L. pneumophila strain Paris (CIP 107629) were grown on buffered charcoal yeast extract (BCYE) agar plates (Thermo Fisher Scientific, Dardilly, France) at 36 • C for 72 hours and then harvested by scraping, suspended in phosphate-buffered saline (PBS), centrifuged at 9500 xg for 10 min, and washed once in PBS. The supernatants were then discarded. The L. pneumophila suspensions were then diluted in PBS to obtain 3 × 10 7 bacteria/mL.
The legionella final suspensions represented the bacterial stock working suspensions, and they were identified as BWS Phila, BWS Paris , and BWS Lens (Table 2).

Bacterial Survival in the Coculture Medium (Control)
The three control bacterial conditions were prepared as described in Table 2 by adding 10 mL of SCYEM to the 0.1 mL bacteria working solutions (BWS Phila, BWS Paris , or BWS Lens ) in 25 cm 3 flasks (Dutscher, Brumath, France) and incubated at 22 • C or 37 • C. This operation corresponded to the T 0 time point of the bacterial controls. Occurring at T 0, T 0 + 24 h, T 0 + 48 h, T 0 + 72 h, and T 0 + 96 h, 1 mL was sampled in each flask and then serially 10-fold diluted in SCYEM and plated on buffered charcoal yeast extract plates (BCYE) in triplicate. BCYE plates were incubated at 36 • C, and colony forming units (CFU) were counted after 5 days. Each condition was performed for three independent replicates and repeated three times (n = 9).

Amoeba Survival in the coculture Medium (Control)
The three amoeba working solutions (AWS C2C , AWS Z503 , or AWS AC ) were prepared as described in Table 2 (10 mL of working solutions) and incubated at 22 • C or 37 • C in 25 cm 3 flasks. Occurring at T 0, T 0 + 24 h, T 0 + 48 h, T 0 + 72 h, and T 0 + 96 h, the flasks were gently shaken, and the numbers of amoeba cells were quantified using a haemocytometer cell counting chamber method with Trypan blue. Each condition was performed for three independent replicates and repeated three times (n = 9).

Co-Culture Assays
Amoeba and bacterial working solutions were mixed in 25 cm 3 flasks by adding the required volume according to Table 1. To provide an example, 10 mL of W. magna C2c Maky at 3 × 10 5 cells/mL was mixed with 0.1 mL of L. pneumophila Lens at 3 × 10 7 CFU / mL. All flasks were left to stand for 2 h at 22 • C ± 2 • C or at 37 • C ± 2 • C to allow for amoebae/bacteria contact and the internalization of L. pneumophila into amoebae. After the 2-h contact process, each flask was gently shaken 10 times, and the suspension was transferred into a 15 mL Flacon ® tube and centrifuged at 3000× g for 5 min. This step allowed for the removal of non-internalized (i.e., extracellular) L. pneumophila from the co-culture suspensions. The pellet was resuspended in 10 mL of sterile SCYEM, and the suspension was poured into a new 25 cm 3 flask and incubated at 22 • C ± 2 • C or at 37 • C ± 2 • C. This time point corresponded to the T 0 time point of the assay. Each condition was performed for three independent replicates and repeated three times (n = 9), with the exception of the co-culture with strain Philadelphia that was repeated four times at 22 • C (n = 15). 4.6. L. pneumophila and Amoeba Quantifications in Co-Culture Assays from T 0 to T 0 + 96 h Occurring at T 0 , T 0 + 24 h, T 0 + 48 h, T 0 + 72 h, and T 0 + 96 h, a washing step was performed. The culture supernatant was removed from each flask and replaced by 10 mL of sterile SCYEM. This step was intended to remove extracellular L. pneumophila to allow for the detection of only intracellular bacteria. Each flask was gently shaken 10 times and an aliquot of 1 mL was sampled. Quantification of amoeba populations was performed using 0.1 mL of each aliquot utilizing a haemocytometer cell counting chamber method with Trypan blue. The remaining 0.9 mL were treated with Triton™ X-100 [31] at 0.02% v/v (final concentration) for 2 min to lyse amoebas and to recover the internal L. pneumophila. The sample was then serially 10-fold diluted in SCYEM and plated on BCYE plates in triplicate, with the exception of the undiluted conditions that were spread onto five plates when the number of L. pneumophila was intended to decrease below the detection limit. BCYE plates were incubated at 36 • C, and CFU were counted after 5 days.

Microscopic
Observations in Co-Culture with L. pneumophila Philadelphia at 37 • C Co-cultures of L. pneumophila Philadelphia using the three amoeba strains at 37 • C were sampled from running experiments and stained by the Gimenez technique [50,51] at T 0 , T 0 + 48 h, and T 0 + 96 h. Co-cultures (0.1 mL) were deposited onto glass slides by using a Shandon Cytospin 4 cytocentrifuge (Thermo Scientific, Illkirch-France) at 800× g for 10 min and then stained using the Gimenez technique. Briefly, each of the glass slides were stained with fuchsin solution for 3 min and washed with water. Then, the glass slides were stained with malachite green for 5-10 s and washed, and this step was repeated twice. Finally, the glass slides were allowed to dry at room temperature.
The observations were performed using a LEICA DM 2500 LED microscope (Leica Microsystemes SAS, Nanterre-France) under an ×100 oil immersion objective.

Statistical Analyses
Statistical significance of co-culture studies was determined for 22 • C and 37 • C conditions through the use of analysis of variance (ANOVA) (Kruskal-Wallis test and multiple pair-wise comparison Dunn test).