Legionella pneumophila prevents proliferation of its natural host Acanthamoeba castellanii

Legionella pneumophila is a ubiquitous, pathogenic, Gram-negative bacterium responsible for legionellosis. Like many other amoeba-resistant microorganisms, L. pneumophila resists host clearance and multiplies inside the cell. Through its Dot/Icm type IV secretion system, the bacterium injects more than three hundred effectors that modulate host cell physiology in order to promote its own intracellular replication. Here we report that L. pneumophila prevents proliferation of its natural host Acanthamoeba castellanii. Infected amoebae could not undergo DNA replication and no cell division was observed. The Dot/Icm secretion system was necessary for L. pneumophila to prevent the eukaryotic proliferation. The absence of proliferation was associated with altered amoebal morphology and with a decrease of mRNA transcript levels of CDC2b, a putative regulator of the A. castellanii cell cycle. Complementation of CDC28-deficient Saccharomyces cerevisiae by the CDC2b cDNA was sufficient to restore proliferation of CDC28-deficient S. cerevisiae and suggests for the first time that CDC2b from A. castellanii could be functional and a bona fide cyclin-dependent kinase. Hence, our results reveal that L. pneumophila impairs proliferation of A. castellanii and this effect could involve the cell cycle protein CDC2b.


L. pneumophila inhibits proliferation of A. castellanii through the Dot/Icm secretion system.
To address how L. pneumophila inhibited A. castellanii proliferation, amoebae were infected with live, heat-killed (65 °C) or ∆ dotA mutant of L. pneumophila at a MOI of 20. The number of A. castellanii was assessed at different time-points to establish a kinetic of proliferation for each condition. In contrast to live L. pneumophila, infection with heat-killed or ∆ dotA L. pneumophila did not impair proliferation of A. castellanii (Fig. 2). These results suggested that inhibition of A. castellanii proliferation requires live L. pneumophila with a functional Dot/Icm T4SS.

L. pneumophila induces modifications in shape, motility and cell division of A. castellanii.
Time lapse microscopy was performed to visualize cell division of A. castellanii after infection with L. pneumophila Paris expressing the GFP protein (G_EP10). Videos were performed 16 h after infection and taken for approximately 17 h. As shown in the supplementary video 1, uninfected A. castellanii were highly motile and cell division was observed. Whereas in infected cells, we did not observe any cell division in GFP positive cells (supplementary video 2 and Fig. 3A). Infected cells seemed less motile with modifications of their shape as they became rounded (supplementary video 2 and Fig. 3A). In addition, cells positive for GFP seemed less adherent than the GFP negative cells (supplementary video 2).
In order to test cell adherence of A. castellanii following infection with L. pneumophila, floating cells were harvested after infection with live, heat-killed or ∆ dotA mutant of L. pneumophila at a MOI of 20. We observed that in contrast to heat-killed and ∆ dotA L. pneumophila, live wild-type L. pneumophila seemed to induce cell detachment (Fig. 3B). These data indicate that inhibition of cell division induced by L. pneumophila was associated with modifications of the shape and of the motility of A. castellanii. Moreover, L. pneumophila required a functional Dot/Icm secretion system to decrease adherence of A. castellanii from the surface.
A. castellanii has a life cycle that oscillates between a dormant and a replicative forms named cyst and trophozoite, respectively. Under unfavorable conditions, amoebae form cysts, which are resistant to environmental stress and metabolically inactive 14 . During encystation, A. castellanii retracts its pseudopodia and becomes rounded 15,16 . In order to verify that rounded-infected cells were not encysted, A. castellanii were stained with Calcofluor White to reveal cellulose in the cell wall of mature cysts 17 . We found that cells that were highly infected with L. pneumophila G_EP10 did not share phenotypic characteristics with A. castellanii cysts. Indeed, there was a striking difference in cell size and, in contrast to cysts, no cell wall containing cellulose was observed in infected amoebae with L. pneumophila G_EP10 (Fig. 3C). In our experimental conditions, round cells that seemed barely adherent upon infection with L. pneumophila were not cysts.

L. pneumophila impairs the ability of infected cells to replicate their DNA. Cell division and
DNA replication represent the two essential cell cycle phases of any eukaryotic cell. Thus we assessed whether the absence of cell division upon L. pneumophila infection was associated to an inhibition of DNA synthesis. A. castellanii infected or not with L. pneumophila Paris expressing the DsRed protein (R_EP10) were incubated in a growth medium containing 5-ethynyl-2′ -deoxyuridine (EdU) for 4 h. Incorporation of EdU, an analogue of thymidine, indicates DNA synthesis. Thus, as expected, uninfected cells were able to duplicate their DNA since around 20% of cells were positive for the EdU signal (Fig. 4A,C). However, in A. castellanii that were co-cultured with L. pneumophila R_EP10, the number of cells positive for EdU dropped to around 1% (Fig. 4B,C). In addition, we never detected cells positive both for EdU and DsRed signals (Fig. 4B). This result suggests that L. pneumophila prevented DNA replication in A. castellanii.  (NCBI). Based on an E-value less than 1 and a percentage of identity of at least 35%, we obtained five candidates as putative CDKs of A. castellanii (Table 1). These five proteins were aligned with human CDKs in order to perform a phylogenetic analysis. Similarly to Cao et al. 18 , we used human (Hsa) cyclin-dependent kinase like-1 (CDKL1), glycogen synthase kinase-3 (GSK3) alpha and serine/threonine-protein kinase MAK isoform 1 as outgroups. Six of the human CDKs are directly related to the cell cycle (CDK1, CDK2, CDK3, CDK4, CDK6 and CDK7) 19 . As shown on the Fig. 5A, only the XP_004353710.1 CDC2b putative protein appeared phylogenetically similar to a cell cycle-related human CDK (CDK1).  In order to further assess the similarity between CDC2b from A. castellanii and well described eukaryotic CDKs, we aligned the protein sequences with those of human (CDK1), and of Baker's and fission yeast, Saccharomyces cerevisiae (CDC28) and Schizosaccharomyces pombe (CDC2), respectively. We found that the CDC2b protein sequence was highly conserved compared to the CDKs from the other organisms (Fig. 5B). Aside from the PSTAIRE cyclin-binding domain, all the CDKs we aligned displayed many highly conserved residues and domains that characterize CDKs 19 such as an ATP-binding domain, an inhibitory phosphorylation site, an activating phosphorylation site and the start and end of a T-loop (Fig. 5B). Thus, the putative protein CDC2b from A. castellanii shares strong homology with other CDKs that are essential for the progression of the cell cycle.

The protein CDC2b from
Since S. cerevisiae can be used to test the function of proteins from A. castellanii 20 , we sought to determine whether heterologous expression of CDC2b could complement the loss of function of the CDC28 gene from yeast. To this end, the coding sequence of CDC2b was introduced into plasmid pYES2 under the control of an inducible GAL1 promoter. The yeast strain Y838, which harbors a temperature sensitive mutation cdc28-4, was transformed with the recombinant plasmid pYES2-CDC2b or the empty vector as a negative control. After propagation of the transformants at the permissive temperature, they were challenged for growth on YPGal medium at various temperatures. As shown in Fig. 5C, Y838 transformed with pYES-CDC2b was capable of robust growth at 30 °C or 34 °C but could not grow at 37 °C, whereas the same strain transformed with the empty plasmid was unable to grow above 30 °C. In contrast, the wild type strain Y10000 was able to grow regardless of the temperature of incubation (Fig. 5C). We conclude from this experiment that A. castellanii CDC2b can restore growth of the cdc28-4 mutant but not as well as a wild type yeast strain.
L. pneumophila provokes a decrease of the CDC2b gene expression. We next explored whether L. pneumophila modulates expression levels of CDC2b. We infected A. castellanii with L. pneumophila and monitored the level of CDC2b mRNA at different time-points. One found that L. pneumophila decreased CDC2b mRNA level as soon as 2 h after the beginning of the infection (0 h). The difference between uninfected and infected cells was significant at 0 and 2 h post-infection (Fig. 6). These results indicate that L. pneumophila infection leads to down-regulation of the mRNA levels of CDC2b, which correlates with a defect in amoebal proliferation during infection.

Discussion
L. pneumophila is an intracellular bacterium that manipulates host functions to support its replication. However, the impact of Legionella infection on the host cell cycle has never been addressed. In this study, we demonstrate that L. pneumophila prevents multiplication of its natural host A. castellanii through the T4SS. Infection with L. pneumophila impairs cell division, DNA synthesis and motility of A. castellanii. The inability of infected cells to multiply correlates with a decrease of the mRNA levels of the putative amoebal cell cycle regulator CDC2b.
Some studies previously suggested a negative effect of L. pneumophila on the proliferation of Acanthamoeba 21,22 . In both studies, authors observed a decrease of the amoebal cell number upon infection, at least three days later, that was associated to an amoebal lysis 21,22 . Indeed, L. pneumophila provokes necrosis-mediated lysis of Acanthamoeba within 48 h after infection 23 . Here, we demonstrate that, even only several hours after infection with L. pneumophila, A. castellanii were unable to replicate their DNA, nor to undergo cytokinesis; two critical steps for cell proliferation. Thus, inhibition of proliferation, and not only Legionella-induced host lysis could account for this decrease in cell number. The type IV secretion system Dot/Icm was necessary for the inhibition of the host proliferation induced by L. pneumophila. Our ongoing work aims at identifying the L. pneumophila effectors involved in the regulation of proliferation of A. castellanii.
Our video microscopy experiments revealed that L. pneumophila induced a modification of the shape of amoebae which became rounded with some detachment from the substrate. The membrane of infected amoebae differed from mature Acanthamoeba cyst as no evidence for cellulose was observed by calcofluor stain. Modulation of encystment has been shown with other bacteria 24,25 . Regarding Legionella, controversially results could be found. Several papers have described cysts fully bound with L. pneumophila 26,27 while some authors did not succeed to isolate infected cysts 28 . This suggests that some specific conditions might be required to obtain infected cysts with L. pneumophila. This question is of importance since cysts are thought to play a role in persistence and dissemination of L. pneumophila.
The alteration in amoebal shape and motility that we observed support previous results indicating that L. pneumophila modulates the cell host cytoskeletal molecules and migration [29][30][31][32] . Interestingly, treatment of the social amoeba Dictyostelium discoideum with the Legionella quorum sensing molecule LAI-1 induced down-regulation of genes involved in movement and in cell proliferation 31 . LAI-1 led to inactivation of the cytoskeletal-regulating protein CDC42 that was reported to induce cell cycle progression 33,34 . We aim to characterize the possible contribution of LAI-1 to the inhibition of A. castellanii proliferation in future studies.  Legionella pneumophila was shown to regulate transcription of genes implicated in proliferation of human monocyte-derived macrophages and of bone marrow-derived macrophages 35,36 . Furthermore, L. pneumophila modifies the host chromatin resulting in a repression of gene expression 37,38 . Since chromatin structure impinges on the cell cycle and vice-versa 39 , contribution of effectors implicated in chromatin-remodeling is a promising area to investigate.
To our knowledge, CDK proteins have never been characterized in A. castellanii. Thanks to genome sequencing of A. castellanii 13 , we found that the putative protein CDC2b shares a strong homology with other cell cycle-related CDKs. CDC2b could restore the ability of CDC28-deficient S. cerevisiae to grow although growth of transformants was limited at high temperature. Interestingly, the same conclusion was reported for the CDC2 gene of the distantly related amoeba Dictyostelium discoideum 40 . As CDK acts in concert with cyclin proteins, growth limitation of transformants carrying CDC2b could arise from the putative cyclin-binding domain of CDC2b that is different from the conserved PSTAIRE motif. This would be analogous to CDC2 from D. discoideum whose PSTAIRE is not completely conserved with an isoleucine substituted for leucine 40 .
From Amoebozoa to Vertebrata, several CDK proteins are present in one cell. The number of CDKs increases with eukaryotic evolution. While 20 CDKs are found in humans, only 6 and 8 CDKs are found in the yeast S. cerevisiae and in the amoeba D. discoideum, respectively 18 . As reported in yeast and mammalian systems, only one CDK is essential to drive the cell cycle from DNA replication to mitosis 10,11 . The ability of CDC2b to functionally complement CDC28-deficient S. cerevisiae suggests that CDC2b could represent the main cell cycle regulator in A. castellanii. Although as yet there is no method to generate mutants in A. castellanii, such experiments are needed to confirm our hypothesis.
Intracellular bacteria were reported to affect the growth rate of free-living amoebae 41 . Beyond, several bacteria are able to promote or to inhibit the eukaryotic cell cycle by using bacterial effectors called cyclomodulins 42,43 . Inhibitory cyclomodulins impair the function of cyclin/CDK complexes through activation of the DNA-damage response, tubulin protein sequestration, down-regulation of cyclins, stabilization of CDK inhibitors, interaction with proteins of the anaphase promoting complex or cleavage of the regulator of the endoplasmic reticulum (ER) function Bip [44][45][46] . Because cyclomodulins induce pleitropic effects on target cells, the next challenge will be to confirm that the decrease of CDC2b mRNA by L. pneumophila is responsible for the A. castellanii proliferation prevention and how this transcriptional/translational effect is important for L. pneumophila. Since down-regulation of CDK expression upon bacterial infection has never been reported, the down-regulation of the CDC2b mRNA level could be a novel strategy for a bacterium to regulate proliferation of its host.
In summary, this study shows for the first time that L. pneumophila prevents the proliferation of its host. Our work identifies a novel CDK of A. castellanii and shows that Legionella perturbs host proliferation, DNA replication and amoebal morphology during infection. Regulation of the host cell cycle contributes to the impressive list of the eukaryotic functions that are perturbed by L. pneumophila. We are currently attempting to decipher the signaling pathway responsible for A. castellanii cell cycle arrest and the bacterial effectors which mediate this effect during infection. L. pneumophila strains were cultured on buffered charcoal yeast extract (BCYE) (1% ACES, 1% yeast extract, 0.2% charcoal, 1.5% agar, 0.025% Iron (III) pyrophosphate, 0.04% L-cysteine, pH 6.9) agar plate. For L. pneumophila Paris strains, 0.1% alpha-ketoglutarate was added to medium. Bacteria were cultured on BCYE at 37 °C for 3 days. Then, bacteria were inoculated at the optical density at 600 nm (OD 600 ) of 0.1 in Buffered Yeast Extract (BYE) at 37 °C under agitation 160 rpm to reach OD 600 above 2.3. The asterisks indicate a significant difference between uninfected and infected conditions (***p < 0.001).
Bacterial transformation. L. pneumophila Paris in exponential phase were washed with 10% cold glycerol solution and centrifuged (6000 g, 10 min at 4 °C). Pellets were washed in 20, 10 and 5 mL of 10% cold glycerol solution. Bacteria were re-suspended in 10% cold glycerol solution to reach an OD 600 of 100. Electroporation was performed with the plasmid pSW001 49 for bacteria producing the DsRed Fluorescent Protein and with pNT28 built from the plasmid pMMB207-Km14-GFPc 50 for bacteria producing the Green Fluorescent Protein (GFP). One μ g of plasmid and 100 μ L of bacterial suspension were mixed. Electroporation was performed by using the EC2 programm (2.50 kV, 1 pulse) of the MicroPulser apparatus (Biorad). After the electroporation, 900 μ L of liquid BYE were added to the bacteria followed by an incubation at 37 °C without agitation for two hours. Suspension was spread on BCYE supplemented with chloramphenicol (5 μ g/mL) and the petri dishes were incubated at 37 °C during 3 or 4 days. The bacteria L. pneumophila Paris producing the DsRed and the GFP proteins were named L. pneumophila R_EP10 and L. pneumophila G_EP10 respectively. DNA synthesis. The DNA synthesis was determined using the Click-iT ® EdU Imaging Kits (Invitrogen ™ ) following manufacturer's recommendations with some modifications. Briefly, 1 × 10 6 A. castellanii cells were infected with L. pneumophila Paris R_EP10 at MOI 20 in 6-well plates. After infection period, EdU (40 μ M) was added to the PYG medium supplemented with gentamicin (20 μ g/mL) for 4 h. Amoebae were harvested and centrifuged (1000 g, 10 min at room temperature) before fixation for 15 min at room temperature with 3.7% paraformaldehyde. Amoebae were washed twice with 1 mL of washing solution (3% bovine serum albumin (BSA) (Sigma)) in PBS and lysed for 20 min at room temperature with 1% Triton ® X-100 (Sigma). After two additional washes, 0.5 mL of Click-iT ® reaction cocktail (according to the manufacturer's instructions) was added to the pellet and incubated 1 h protected from the light. For DNA staining, amoebae were washed with 1 mL of PBS, treated with TO-PRO ® -3 Iodide (dilution of 1:1000) and incubated for 1 h at room temperature in the dark. Amoebae were washed for the last time with PBS and suspended in washing buffer for analysis.

Infection of A. castellanii. Three days old
A. castellanii were examined with a laser scanning confocal microscope (FluoView-1000, Olympus) coupled to an inverted microscope IX-81(Olympus). Images were obtained with an Olympus UPLSAPO 60X W NA: 1.2 and zoom x2 (800 × 800 pixels, 0.13 μ m/pixel). Samples were excited with 488/500-530 nm excitation/emission filters for EdU staining, 543/555-625 nm for bacteria producing DsRed and 633/LP650 nm for Topro-3 nuclear staining. Multiple fluorescence signals were acquired sequentially to avoid cross-talk between colour channels. For 3D acquisition, optical sectioning of the specimen (Z series) was driven by a Z-axis stepping motor and maximum intensity projection was further generated. An additional digestion of residual DNA was performed with Turbo DNA-free ™ kit (Ambion ® ) according to manufacturer's instructions.
Reverse transcription (RT) and quantitative PCR (qPCR). Reverse transcription of RNA was performed using GoScript ™ Reverse Transcription System (Promega) following manufacturer's recommendations. Products of reverse transcription (RT) were purified with NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel). All cDNA were normalized (10 μ g/mL) before proceeding to qRT-PCR reaction. All qPCR reactions were performed using the LightCycler ® FastStart DNA Master Plus SYBR Green I kit (Roche) on the LightCycler 1.5 instrument (Roche). Each tube contained 9 μ L of reaction mix containing 0.5 μ L of each primers CDC2b (XM_004353658.1) (5′ ATGCAAGCCAAACCCAGTC 3′ and reverse: 5′ GAATCGCTGGTTCTCGGTATC 3′ ) or 18S primers 51 at 10 μ M, 2 μ L of Master Mix 5X, 6 μ L of water and 1 μ L of cDNA or 1 μ L of water for negative control. The qPCR program was: 10 min at 95 °C and 45 cycles of the amplification step (10 sec at 95 °C, 10 sec at 61 °C and 10 sec at 72 °C for extension time in a single acquisition mode), the melting curves step for 1 min at 65 °C for annealing and the cooling 30 seconds at 40 °C. The relative level of genes expression was calculated using the 2 −∆∆Ct method 52 . Expression level were normalized to the 18S mRNA.

Phylogenetic analysis.
To construct the phylogenetic tree, proteins were aligned using MUSCLE of the software MEGA6 with a Maximum likelihood and a bootstrap replications of 1000.
Yeast transformation of Y838 was done by the Li Acetate method as described by Gietz and Woods 53 . This strain is temperature sensitive for growth (ts − ) due to a deficient CDC28 gene product. Transformants were selected at the permissive temperature (28 °C) on YNB medium supplemented with adequate amino acids and bases. For complementation assays of the cdc28-4ts − phenotype, transformants were streaked on YPGal medium and incubated at the indicated temperature (30 °C, 34 °C or 37 °C) for 4 days.

Statistical analyses.
All experiments were performed three times. Results were analyzed by Two-way RM ANOVA with the Bonferroni post-test (GraphPad Prism5) or using a one-tailed Mann-Whitney test considering a statistical significance at p ≤ 0.05. All data are average of three independents experiments and error bars represent the standard error of the mean (± SEM).