The progeny of Legionella pneumophila in human macrophages shows unique developmental traits

Authors

  • Hany Abdelhady,

    Corresponding author
    1. Department of Microbiology and Immunology, Dalhousie University, Halifax, NS, Canada
    • Correspondence: Rafael A. Garduño, Department of Microbiology & Immunology, Dalhousie University, Sir Charles Tupper Medical Building, 7th floor, 1459 Oxford Street, Halifax, NS, Canada B3H 4R2. Tel.: +902 494 6575; fax: +902 494 3889; e-mail: rafael.garduno@dal.ca

    Search for more papers by this author
  • Rafael A. Garduño

    1. Department of Microbiology and Immunology, Dalhousie University, Halifax, NS, Canada
    2. Department of Medicine-Division of Infectious Diseases, Dalhousie University, Halifax, NS, Canada
    Search for more papers by this author

Abstract

The Gram-negative bacterium Legionella pneumophila is an intracellular parasite of amoebae and an accidental human pathogen that causes a noncommunicable atypical pneumonia known as Legionnaires' disease (LD). In some mammalian cells (e.g. HeLa), L. pneumophila follows a biphasic developmental cycle, differentiating between a replicative form that actively multiplies intracellularly, and a mature infectious form (MIF) that emerges as progeny. To date, it is not known whether the L. pneumophila progenies that emerge from amoebae and human macrophages reach similar developmental stages. Here, we demonstrate that in relation to the fully differentiated and highly infectious MIFs that emerge from amoebae, the L. pneumophila progeny that emerges from macrophages is morphologically undifferentiated, less resistant to antibiotics and less able to initiate infections. However, the L. pneumophila progeny from macrophages did not show any defects in intracellular growth. We thus concluded that macrophage infection with L. pneumophila yields a low number of bona fide MIFs. Because MIFs are the transmissive forms of L. pneumophila produced in vivo, our results showing that they are not efficiently produced in cultured macrophages provide an initial insight into why LD is not communicable.

Introduction

The natural history of Legionella pneumophila indicates that this Gram-negative bacterium has evolved as an intracellular pathogen of freshwater amoebae. However, L. pneumophila is also an opportunistic human pathogen that causes the atypical pneumonia known as Legionnaires' disease (LD) (Harb et al., 2000; Fields et al., 2002). LD is not transmitted from person-to-person, but an explanation for this fact is yet to be found.

Legionella pneumophila follows a biphasic developmental cycle in vitro, where the stationary phase form (SPF) alternates with the exponential phase form (Byrne & Swanson, 1998). In vivo (inside HeLa cells) L. pneumophila alternates between the replicative form (RF) and the mature infectious form (MIF) (Faulkner & Garduno, 2002). SPFs and MIFs resist environmental stress and infect new hosts and are thus considered to be the transmissive L. pneumophila forms in vitro (Molofsky & Swanson, 2004), and in vivo (Garduno, 2007), respectively. SPFs and MIFs are developmentally linked, because SPFs can directly differentiate into MIFs (Faulkner et al., 2008), suggesting that L. pneumophila can produce stable forms with distinct developmental traits.

While differences between SPFs and differentiated MIFs have been documented (Garduno et al., 2002), it is not known whether the L. pneumophila progeny produced in various hosts reaches the same developmental stage. Because differences between MIFs could provide clues to the lack of person-to-person transmission of LD, we set out to test the hypothesis that the progeny produced in human macrophages is less environmentally fit and infectious than fully differentiated MIFs produced in amoebae.

Materials and methods

Legionella pneumophila strains and culture conditions

Lp1-SVir is a spontaneous streptomycin-resistant virulent Philadelphia-1 strain (Hoffman et al., 1989). Olda strain 2064 is a local clinical isolate from Halifax, Canada (Fernandez et al., 1989). JR32 is a streptomycin-resistant virulent Philadelphia-1 strain (Sadosky et al., 1993). JR32 harbouring plasmids pMMB207-Km14-GFPc (for green fluorescence) or pSW001 (for red fluorescence) was also used (Mampel et al., 2006). Strains were kept as −70 °C stocks from crude lysates of infected Acanthamoeba castellanii trophozoites. Thawed aliquots were grown on buffered charcoal yeast extract agar (BCYE) (Pascule et al., 1980) at 37 °C in a humid incubator and used without subculture. Strain 2064 was grown in plain BCYE, whereas strains Lp1-Svir and JR32 in BCYE containing 100 μg mL−1 streptomycin. Buffered yeast extract broth (BYE), used to grow L. pneumophila to stationary phase in liquid cultures, had the same formulation of BCYE but without charcoal and agar.

Culture of A. castellanii and legionellae progeny

Acanthamoeba castellanii trophozoites were maintained by subculture at room temperature in Neff's medium supplemented with a vitamin mix (Cursons et al., 1980). Stocks of A. castellanii cysts were long-term preserved in sterile tap water at room temperature. For infection assays, trophozoites were grown to 70% confluency in supplemented Neff's medium at 37 °C in 25- or 75-cm2 cell culture flasks (Falcon-BD Biosciences Canada, Mississauga). Before infection, trophozoites were kept in modified Neff's medium (without yeast extract, proteose peptone #3 and vitamins) seeded in 25-cm2 cell culture flasks (107 trophozoites per flask), or in 12- or 24-well plates (106, or 5 × 105 trophozoites per well, respectively). Trophozoites in suspension were counted in a Neubauer haemocytometer (Bright Line). To obtain progeny from trophozoites in flasks, L. pneumophila colonies grown on BCYE (likely containing a mixture of cells in different growth phases) were scraped, and bacteria were then washed and resuspended in modified Neff's medium to an optical density (OD) of 1 unit (c. 109 legionellae mL−1). OD was measured at a wavelength of 620 nm in a UNICO UV-2100 spectrophotometer. Flasks were inoculated with 100 μL of this OD 1-unit suspension, excess inoculum washed after 3 h, and the infection was allowed to proceed for 24–48 h until progeny emerged and no trophozoites were left.

Feeding experiments with Tetrahymena tropicalis

The ciliate T. tropicalis was axenically maintained by subculture in biphasic medium and in plate count broth as described before (Berk et al., 2008). Ciliates were fed at a legionellae/ciliate ratio of 1000 in modified Tris-buffered Osterhout's solution, pH 7 (TBOS) at room temperature (Berk et al., 2008). MIF-laden pellets were pelleted (500 g, 10 min) in 15-mL conical centrifuge tubes (Falcon). Live ciliates were then allowed to swim back into suspension before removing the supernatant. This operation was repeated 3X with fresh TBOS. Pellets were disrupted, to release MIFs, by repeatedly passing a dense suspension between two 1-mL syringes connected through a gauge 27 needle.

Culture of mammalian cell lines and production of legionellae progeny

Undifferentiated human U937 and THP-1 cell lines were cultured at 37 °C, 5% CO2 in complete RPMI-1640 medium (Gibco-Invitrogen Grand Island, NY) supplemented with 10% foetal bovine serum (FBS; HyClone, Logan, UT), 2 mM L-glutamine (Gibco), 100 U mL−1 penicillin and 100 μg mL−1 streptomycin. U937 and THP-1 cells were induced to differentiate into macrophage-like cells for 2 h in complete RPMI-1640 containing 60 ng mL−1 phorbol 12-myristate 13-acetate (PMA, Sigma). Cells were then washed 3X with fresh RPMI-1640 medium, enumerated in a Neubauer haemocytometer (Bright Line) and allowed to adhere overnight in 25-cm2 flasks (107 cells per flask), or 12- and 24-well plates (106 and 5 × 105 cells per well, respectively). To obtain progeny from macrophages in flasks, L. pneumophila colonies grown on BCYE (likely containing a mixture of cells in different growth phases) were scraped, and bacteria were then washed and resuspended in RPMI-1640 to an OD of 1 unit. Flasks were inoculated with 500 μL of this OD 1-unit suspension, excess inoculum washed after 3 h, and the infection was allowed to proceed for 24–48 h until progeny emerged, and no macrophages were observed. Mouse L929 cells were grown at 37 °C, 5% CO2, in 25-cm2 flasks containing 10 mL of complete minimal essential medium (MEM; Gibco) supplemented with 10% FBS, 100 U mL−1 penicillin and 100 μg mL−1 streptomycin. L929 cells were detached by trypsinization, enumerated by direct microscopy, suspended in complete MEM, transferred to 24-well plates at about 5 × 105 cells per well and allowed to attach overnight at 37 °C in preparation for intracellular growth assays (below). HeLa cells were only used to produce legionellae progeny for ultrastructural characterization as described before (Garduno et al., 1998).

Purification of legionellae progenies from the various host cells

Emerged progenies were purified from lysates of infected cells in a Percoll gradient (Garduno et al., 1998). Briefly, infected cells were pelleted (8000 g, 10 min), and then lysed in 1 mL of 0.05% Triton-X in ddH2O. The lysate was mixed with 5 mL of MEM and 3 mL of isotonic Percoll in a 9-mL polycarbonate tube and centrifuged (20 000 g, 10 min) at 4 °C. The self-forming density gradient typically yielded two bands: a top one containing cell debris and a bottom one (c. 1.074 g mL−1) containing free bacterial cells. These bacteria from the bottom band were washed twice in ddH2O to remove residual Percoll.

Transmission electron microscopy (TEM)

Percoll-purified bacterial cells were fixed in glutaraldehyde, postfixed in OsO4, in-bloc stained with aqueous uranyl acetate, dehydrated in acetone, infiltrated and sectioned as described previously (Faulkner & Garduno, 2002). Thin sections poststained with uranyl acetate and lead citrate were observed in a JEOL JEM-1230 transmission electron microscope using an accelerating voltage of 80 kV. High-resolution images were captured as tiff files with a Hamamatsu ORCA-HR camera. The proportion of MIFs was quantified in 100 randomly chosen bacterial cells. Three ultrastructural traits had to be met for a cell to be counted as a MIF: an inconspicuous periplasm, a straight and thickened outer membrane, and presence of large inclusions. Statistical significance of differences in the proportion of MIFs was calculated using the two-proportion test (Z-test).

Stress-resistance assays

These were performed on suspensions of c. 108 legionellae mL−1, evaluating CFU mL−1 (by dilution-plating) before and after exposure. For antibiotic challenges, legionellae were suspended in MEM (Gibco) containing gentamicin (20 or 100 μg mL−1) or ciprofloxacin (25 μg mL−1) and incubated for 3 h at 37 °C. For chlorine challenges, legionellae were suspended in ddH2O containing 1 or 5 ppm of chlorine and incubated for 30 min at room temperature. Results were reported as % survival. Lysis in 1% sodium dodecyl sulphate (SDS) in 10 mM Tris, pH 7.5, was tested in suspensions with an OD620 nm of c. 1 unit. The minutes required to reach 50% of the initial OD620 nm was reported as the detergent lysis index.

Intracellular growth and attachment assays

Intracellular growth and attachment assays were performed in U937-derived macrophages, A. castellanii trophoz-oites, or L929 cells seeded in 24-well plates (c. 5 × 105 cells per well) with a final volume of 1 mL per well. Macrophages were inoculated with 25 μL per well of an OD 1-unit suspension of legionellae progeny (either from macrophages or from amoebae) in RPMI-1640 to achieve a bacteria/macrophage ratio of c. 50 : 1, whereas trophozoites and L929 cells were inoculated with 50 μL per well of a 1 : 10 dilution of the same OD 1-unit suspension, to achieve a bacteria/host cell ratio of c. 10 : 1. Plates were centrifuged at 500 g for 10 min at room temperature to promote contact. Following a 3-h incubation at 37 °C, 5% CO2, monolayers were washed 3X with PBS. This was considered the zero time for intracellular growth, or the final point for attachment-only assays. At 0, 24 and 48 h postinfection, 3 wells per sample of infected cells were lysed for CFU enumeration. Macrophages and L929 cells were lysed in 1 mL of ddH2O, and amoebae trophozoites in 1 mL of ddH2O containing 0.05% Triton X-100. Results for intracellular growth and attachment-only assays were reported as mean ± SD of three independent experiments, each run in triplicate, and the analysis of statistical significance was done using the 2-way anova test.

Competition assays

Competition assays were performed with strain JR32 either carrying plasmid pMMB207-Km14-GFPc (green fluorescence) or pSW001 (red fluorescence). Competition of red fluorescent progeny from macrophages vs. green fluorescent MIFs from amoeba (or vice versa) was assessed in A. castellanii trophozoites and U937 macrophages. Trophozoites and macrophages were first seeded in 12-well plates at c. 106 cells per well and allowed to attach overnight. Fluorescent MIFs from trophozoites and progeny from macrophages were mixed in equal proportion in the corresponding cell culture medium and added to a final bacteria/cell ratio of 10 (trophozoites) or 50 (macrophages). Intracellular growth was evaluated as described above, but prior to lysing infected cells, fluorescence images were captured from random fields of each well in a THY-100 inverted microscope (Olympus Canada) using an Evolution QEI digital video camera (Media Cybernetics Inc.). Cells infected only with MIFs from amoebae or progeny from macrophages, as well as cells with a mixed infection, and noninfected cells were scored on the captured images using the image pro plus image analysis software (Media Cybernetics Inc). Dilution-plating was performed on BCYE containing 5 μg mL−1 chloramphenicol (for total counts of JR32 carrying pMMB207-Km14-GFPc or pSW001), or 25 μg mL−1 kanamycin (for counts of JR32 carrying pMMB207-Km14-GFPc). CFU per well are shown as mean ± SD from three independent experiments, each run in triplicate.

SDS-PAGE and mass spectrometry

Bacterial cells were mechanically lysed with 100-μm diameter zirconia/silica beads (BioSpec Products, Bartlesville, OK) at 4800 r.p.m. for 45 s using a mini-bead beater (BioSpec Products). Lysates (100 μL) were then mixed with 100 μL of 2X Ames sample buffer containing 10% mercaptoethanol and incubated at 100 °C for 5 min. Solubilized proteins were separated in a 12% acrylamide gel using a Protean-II vertical slab gel apparatus (Bio-Rad). Digital tiff images of silver-stained (Blum et al., 1987); gels were obtained in an EPSON ES-1200C scanner equipped with a transparency unit EU-13 (Seiko Epson Co., Nagano, Japan). Bands uniquely present in MIFs from amoebae (from at least three independent protein extractions) were cut out and sent for standard MS-MS tandem protein identification, to the mass spectrometry and proteomics core facility at Dalhousie University.

Results

The L. pneumophila progeny from U937-derived macrophages are ultrastructurally undifferentiated

The proportion of bacteria showing the typical MIF traits in the L. pneumophila progeny obtained from various host cells was quantified (Table 1 and Fig. 1). We observed a significant decrease (P < 0.001) in the proportion of bona fide MIFs in the progeny from macrophages (U937 and THP-1 cells) in relation to the progeny from other host cells. The low-MIF progeny produced in U937-derived human macrophages turned into a high-MIF progeny after growth in amoeba, confirming that the MIF features observed were reversible and not due to mutant selection. The progeny of strains Lp1-Svir, 2064, and JR32 in macrophages were also rich in filaments (not shown), which have been shown to interact with lung epithelial cells through a mechanism not shared with short rods (Prashar et al., 2012). In contrast, MIFs from amoebae consisted only of short rods. Therefore, for all subsequent results reported here, it should be noted that the progeny from macrophages is heterogeneous and constitutes a mixture of 40% MIFs and 60% other developmental forms. The progeny from amoebae (containing 80% bona fide MIFs) is thus subsequently referred to as MIFs from amoeba.

Table 1. Characteristics of the progeny of two Legionella pneumophila a strains (Lp1-SVir and 2064) obtained from different host cells, in relation to SPFs grown in vitro, which were used here as reference
 % cells with MIF traits% survival against
Gentamicin (20 μg mL−1)Gentamicin (100 μg mL−1)Ciprofloxacin (25 μg mL−1)Chlorine (1 ppm)
SVir2064SVir2064SVir2064SVir2064SVir2064
  1. a

    The proportion of bona fide MIFs in these bacterial populations was significantly different (P < 0.01) to the proportion observed in other cell types, except for the U937 progeny vs. the THP-1 progeny, which were not significantly different between them.

Amoeba MIFs797972 ± 4.720 ± 7.86 ± 2.61 ± 2.523 ± 3.222 ± 2.94 ± 1.949 ± 13
U937 progeny48*40*28 ± 3.12 ± 1.51 ± 1.20.05 ± 1.515 ± 1.010 ± 0.611 ± 2.892 ± 9.1
U937 progeny passed in amoeba808348 ± 6.522 ± 7.07 ± 1.11 ± 0.225 ± 1.619 ± 0.6NDND
SPFs< 1a< 1a0.005 ± 0.0020.001 ± 0.00060.003 ± 0.00170.002 ± 0.0016 ± 2.39 ± 2.60 ± 00 ± 0
Ciliate MIFs8566        
HeLa MIFs8470        
THP-1 progeny49a36a        
Figure 1.

Electron micrographs showing the morphological traits of Lp1-SVir MIFs from amoeba and the Lp1-SVir progeny from macrophages. (a) MIF from amoeba displaying the three traits used to qualify MIFs: a straight outer membrane, an inconspicuous periplasm and the presence of large inclusions. (b) Legionella pneumophila cell produced in U937-derived macrophages displaying large inclusions, but lacking a straight outer membrane and an inconspicuous periplasm. The periplasm is quite apparent (black arrows) and the outer membrane is wavy (e.g. white arrowhead). (c) Detail of an L. pneumophila cell produced in THP-1-derived macrophages displaying a straight outer membrane and an inconspicuous periplasm, but lacking large inclusions. Bacterial cells shown in (b and c) are not regarded as MIFs. I = inclusions. All micrographs are presented at the same magnification.

MIFs from amoebae are more resistant to antibiotics than the progeny from macrophages

MIFs produced in amoebae showed a modestly higher resistance to gentamicin (previously used to test MIFs from HeLa cells) (Garduno et al., 2002) and ciprofloxacin (used clinically for the treatment of LD) than the progeny produced in macrophages (Table 1). This resistance was reversible, and not the result of mutant selection. In relation to antimicrobial agents found in man-made water systems, both MIFs from amoebae and the progeny from macrophages showed infinite lysis indexes in 1% SDS. However, the progeny from macrophages survived better in the presence of 1 mg L−1 chlorine (Table 1), whereas none of the legionellae survived in 5 mg L−1 chlorine.

MIFs from amoebae are more adherent to host cells than the progeny from macrophages

Due to the advantage of amoeba MIFs in gentamicin challenges, the gentamicin treatment normally used to kill extracellular bacteria in infection assays was omitted. Instead, host cells were washed six times with PBS to more effectively remove extracellular MIFs. At time zero, amoeba MIFs of strain Lp1-SVir associated better with host cells than macrophage progeny (Fig. 2), suggesting that the progeny from macrophages had an adherence defect. However, no subsequent defect in intracellular growth was observed (Fig. 2). In fact, the progeny from macrophages appeared to grow faster in trophozoites and L929 cells than MIFs from amoebae. Attachment-only assays conducted with strains SVir (Fig. 2d) and 2064 (not shown) confirmed both the adherence defect of the progeny from macrophages and its reversible nature.

Figure 2.

Infection and attachment assays indicate that the Legionella pneumophila cells produced in macrophages have an attachment defect, but not an intracellular growth defect. Intracellular growth curves of strain Lp1Vir MIFs from amoeba and Lp1-SVir progeny from macrophages in (a) Acanthamoeba castellanii trophozoites, (b) U937-derived macrophages and (c) L929 cells. Graph points represent means, and error bars represent ± SD for three independent experiments (n = 3), each run in triplicate. Statistical significance of differences between mean values at the same time point is indicated as **P < 0.01 and ***P < 0.001. (d) Attachment assay performed with strain Lp1-Vir in A. castellanii trophozoites (Amoebae) or U937-derived macrophages (U937 cells). Bars represent means, and error bars represent + 1 SD for three independent experiments (n = 3), each run in triplicate. Statistical significance of mean values in relation to the ‘progeny from macrophages’ sample is indicated as **P < 0.01. Similar results were obtained with strain 2064.

The progeny from macrophages do not show an intracellular growth defect in competition assays

Because sometimes growth defects only become obvious in direct competition assays, green fluorescent JR32 MIFs from amoebae and red fluorescent JR32 progeny from macrophages were allowed to compete on monolayers of A. castellanii trophozoites and U937-derived macrophages (Fig. 3). MIFs from amoebae infected a significantly higher number of host cells (Fig. 3a and b), but did not produce a more numerous progeny (Fig. 3c and d) in relation to the progeny from macrophages. These results confirmed that the progeny from macrophages do not show intracellular growth defects, but are defective at initiating new infections.

Figure 3.

In competition assays, the Legionella pneumophila progeny from macrophages showed a disadvantage in its ability to initiate infections, but outgrew the competing MIFs from amoebae during the intracellular growth phase. (a) Fluorescence microscopy micrograph showing amoebae trophozoites infected with MIFs from amoeba (green fluorescence) and (or) the L. pneumophila progeny from macrophages (red fluorescence) at 24 h postinfection. Note that some amoebae had mixed infections (white arrowheads) showing a combined red and green fluorescence. (b) Quantitative results showing the percentage of green infected (black bars) or red infected (white bars) cells at 24 and 48 h postinfection. Intracellular growth during competition assays in (c) amoeba, or (d) U-937 macrophages. All competition assays were performed with L. pneumophila strain JR-32 harbouring plasmids carrying either the gene encoding green fluorescent protein or DsRed fluorescent protein. All bar graphs represent means + 1 SD from 3 independent experiments (n = 3), run in triplicate. Black bars represent results obtained for MIFs from amoeba, and white bars results for the progeny from macrophages. Statistical significance of differences between the two competing populations is shown as *P < 0.05, **P < 0.01 and ***P < 0.001. Similar results were obtained by swapping the plasmids for red and green fluorescence between the two competing populations.

MIFs from amoebae express proteins not found in the progeny from macrophages

The phenotypic differences between the progenies from amoebae and macrophages were expected to correlate with protein differences. We identified two prominent protein bands, consistently present only in the lysates of MIFs from amoebae, which were analysed by mass spectrometry. The list of proteins represented by the tryptic peptides identified in these bands (which were absent from the corresponding gel region of the macrophage progeny lysates) is shown in Table 2.

Table 2. Proteins for which tryptic peptides were identified by mass spectrometry in two major bands (c. 40 kDa and c. 50 kDa) present in whole cell lysates of Lp1-SVir MIFs from amoebae separated by SDS-PAGE
 Predicted mass (kDa)Putative or confirmed function
  1. Tryptic peptides matching the listed proteins were not found in the corresponding areas of protein SDS-PAGE gels of legionellae from macrophages.

From the 50-kDa band
 3-Oxoacyl (acyl carrier protein) synthase III39.4Metabolism (lipids)
 Enhanced entry, EnhA20.6Toxin production, cell entry
 Lem9 (Lpg1491)47.5Icm/Dot effector
 LirA30.5Icm/Dot effector
 Malate dehydrogenase36.3Metabolism (carbohydrates)
 Malonate decarboxylase subunit44.5Metabolism (energy)
 Nodulation competitiveness protein Lpg295854.3Chemotaxis, motility, transmembrane protein
 Hypothetical Lpg027942.6Unknown
 Hypothetical Lpg136859.9Unknown
 Hypothetical Lpg163951.3Unknown
From the 40-kDa band
 3-Hydroxy butyrate dehydrogenase (BdhA)27.8Metabolism (lipids, carbohydrates)
 Enhanced entry, EnhC41.4TPR repeat-containing protein, toxin production, cell entry
 Ral F44.8Icm/Dot effector
 RavE37.9Icm/Dot effector
 Small heat shock, HspC218.9Adaptation to stress
 Universal stress, UspA15.8Adaptation to stress
 Uroporphyrinogen III methylase42.5Metabolism (cofactors and vitamins)
 Hypothetical Lpg067040.9Unknown

Discussion

In relation to bona fide MIFs from amoebae, the progeny from macrophages are morphologically undifferentiated, less resistant to antibiotics and less able to initiate infections. These traits were not strain dependent, suggesting that the L. pneumophila progeny produced in cultured macrophages is, in general, unique.

We proposed elsewhere (Garduno et al., 2002; Garduno, 2007) that the early apoptotic demise of Legionella-infected macrophages and/or the lack of proper signals results in an incomplete differentiation of L. pneumophila. Moreover, differentiation-deficient rpoS and letA mutants, which do not grow in amoebae and are completely digested in T. tropicalis, grow well in HeLa cells and macrophages (Hales & Shuman, 1999; Gal-Mor & Segal, 2003; Lynch et al., 2003; Abu-Zant et al., 2006; Faulkner et al., 2008). Collectively, these results suggest that L. pneumophila is under strong selective pressure to differentiate inside protozoa, but not in mammalian cells, and support the notion that L. pneumophila does not complete its developmental programme in cultured macrophages. A speculative extension of this conclusion would be that bona fide MIFs are not produced in alveolar macrophages during human infection.

The molecular mechanisms behind the incomplete differentiation of L. pneumophila in human macrophages remain undetermined. However, the differences found here in protein expression between MIFs produced in amoebae and macrophages constitute an initial contribution to the understanding of such mechanisms. For instance, the fact that 3-hydroxy butyrate dehydrogenase or 3-oxoacyl [acyl carrier protein] synthase III were not detected by mass spectrometry in macrophage-derived legionellae suggests changes in their lipid metabolism and partially explains their reduced amount of inclusions. In fact, the majority of protein changes previously observed in SPFs (the transmissive L. pneumophila forms produced in vitro) were linked to lipid and carbohydrate metabolism (Hayashi et al., 2010). The importance of the enhanced entry proteins EnhA and EnhC in the early interactions of L. pneumophila with host cells is well established (Cirillo et al., 1999, 2000). That EnhA and EnhC were undetectable in lysates of macrophage-derived legionellae contributes to explain their low adherence and ability to initiate cell infections. Finally, based on the absence of two stress proteins (UspA and HspC2) in macrophage-derived legionellae, we hypothesize that cultured human macrophages might not create intracellular conditions that are stressful enough to force L. pneumophila to fully differentiate into MIFs (assuming that stress acts as a differentiation signal in L. pneumophila). Of note is that our mass spectrometry analysis did not include secreted proteins, because legionellae were extensively washed during the purification process. Therefore, a more refined proteome analysis, complemented with transcriptomic data, would be necessary to substantiate our initial observations. Nonetheless, the 18 L. pneumophila proteins undetectable in the macrophage-derived progeny (Table 2) might suggest potential targets for future studies.

In conclusion, we demonstrated that the L. pneumophila progeny produced in cultured human macrophages has unique characteristics that suggest an incomplete differentiation into MIFs. As MIFs are the transmissive forms of L. pneumophila produced in vivo, these results provide an initial insight into why LD is not communicable. We propose a detailed study of macrophage-derived legionellae to help us elucidate molecular mechanisms underlying the L. pneumophila differentiation in vivo and the transmission of LD.

Acknowledgements

We acknowledge the assistance of Mary-Ann Trevors in TEM and Dr Alejandro Cohen in mass spectrometry, the useful suggestions from David Allan and Dr Gheyath Nasrallah and the help from Eman Atwi (Dalhousie University). We are indebted to Dr Jacques Frère (University of Poitiers) and Dr Hubert Hilbi (Ludwig-Maximilians-University Munich) for plasmids pMMB207-Km14-GFPc and pSW001, and Drs Paul Hoffman, David Spencer, Andrew Issekutz, Spencer Lee and Robert Anderson (Dalhousie University) for HeLa cells, Acanthamoeba castellani, U937 cells, L929 cells and THP-1 cells, respectively. Legionella pneumophila strain SVir was a gift from Dr Paul Hoffman and JR32 from Dr Howard Shuman (Columbia University). This work was funded by grants from the Canadian Institute of Health Research and the Nova Scotia Health Research Foundation (R.A.G.). H.A. received support from the Egyptian Cultural Office, in the form of a scholarship.

Ancillary