Burkholderia cepacia complex isolates survive intracellularly without replication within acidic vacuoles of Acanthamoeba polyphaga

Authors

  • Julie Lamothe,

    1. Department of Microbiology and Immunology, University of Western Ontario, London, Ontario, N6A 5C1, Canada.
    Search for more papers by this author
  • Sandra Thyssen,

    1. Lawson Health Research Institute, 268 Grosvenor St., London, Ontario, N6A 4V2, Canada.
    Search for more papers by this author
  • Miguel A. Valvano

    Corresponding author
    1. Department of Microbiology and Immunology, University of Western Ontario, London, Ontario, N6A 5C1, Canada.
    2. Department of Medicine, University of Western Ontario, London, Ontario, N6A 5C1, Canada.
    Search for more papers by this author

*E-mail: mvalvano@uwo.ca; Tel. (+1) 519 661 3996; Fax (+1) 519 661 3499.

Summary

We have previously demonstrated that isolates of the Burkholderia cepacia complex can survive intracellularly in murine macrophages and in free-living Acanthamoeba. In this work, we show that the clinical isolates B. vietnamiensis strain CEP040 and B. cenocepacia H111 survived but did not replicate within vacuoles of A. polyphaga. B. cepacia-containing vacuoles accumulated the fluid phase marker Lysosensor™ Blue and displayed strong blue fluorescence, indicating that they had low pH. In contrast, the majority of intracellular bacteria within amoebae treated with the V-ATPse inhibitor bafilomycin A1 localized in vacuoles that did not fluoresce with Lysosensor™ Blue. Experiments using bacteria fluorescently labelled with chloromethylfluorescein diacetate demonstrated that intracellular bacteria remained viable for at least 24 h. In contrast, Escherichia coli did not survive within amoebae after 2 h post infection. Furthermore, intracellular B. vietnamiensis CEP040 retained green fluorescent protein within the bacterial cytoplasm, while this protein rapidly escaped from the cytosol of phagocytized heat-killed bacteria into the vacuolar lumen. Transmission electron microscopy analysis confirmed that intracellular Burkholderia cells were structurally intact. In addition, both Legionella pneumophila- and B. vietnamiensis-containing vacuoles did not accumulate cationized ferritin, a compound that localizes within the lysosome. Thus, our observations support the notion that B. cepacia complex isolates can use amoebae as a reservoir in the environment by surviving without intracellular replication within an acidic vacuole that is distinct from the lysosomal compartment.

Introduction

Burkholderia cepacia was first isolated from an outbreak of soft rot in onions (Burkholder, 1950). B. cepacia strains display an extraordinary metabolic versatility, as evidenced by their capacity to degrade a wide variety of organic compounds and to survive in nutritionally limited environments (Beckman and Lessie, 1979). Such properties have attracted considerable interest from agricultural researchers to develop B. cepacia strains for combating soilborne plant pathogens and decontaminating polluted soils (Govan and Vandamme, 1998). However, B. cepacia has increasingly been found associated with life-threatening lung infections in humans, which may occur in hospitalized patients requiring mechanical ventilation, as well as in those with cystic fibrosis (CF) and chronic granulomatous disease (Isles et al., 1984; Tablan et al., 1985; Govan and Deretic, 1996; Govan et al., 1996). Antimicrobial treatment of B. cepacia infections is problematic because these microorganisms are intrinsically resistant to a wide range of antimicrobial agents (Aaron et al., 2000; Nzula et al., 2002). Polyphasic taxonomical studies have shown that B. cepacia isolates, cultured from clinical or environmental sites, form a group of closely related species or genomovars, collectively referred to as the B. cepacia complex (Vandamme et al., 1997; Coenye and Vandamme, 2003). Furthermore, there is no clear distinction between environmental and clinical species within the complex.

Several observations suggest that B. cepacia complex isolates can survive intracellularly: (i) B. cepacia is taxonomically related to B. pseudomallei, a recognized human intracellular pathogen (Jones et al., 1996); (ii) Burkholderia-like organisms are found in nature as obligatory intracellular endosymbionts of mycorrhizal fungi (Bianciotto et al., 2003) and in leaf galls (Van Oevelen et al., 2002); and (iii) direct evidence of intracellular survival of B. cepacia complex isolates in vitro has been documented in the human respiratory epithelial cell line A549 (Burns et al., 1996; Keig et al., 2002), in amoebae (Marolda et al., 1999), in murine macrophages (Saini et al., 1999) and in a human monocytic cell line (Martin and Mohr, 2000). Results from lung infections in a mouse model provide additional support to the notion that B. cepacia complex strains adhere to and invade respiratory epithelial cells (Chiu et al., 2001).

Survival of B. cepacia complex strains within amoeba, and perhaps other protozoa, may reflect the overall ability of these bacteria to thrive in soil, while prolonged intracellular survival in macrophages may help bacteria to persist within the airways. We speculate that intracellular survival provides B. cepacia complex isolates with additional protection from host defences and antibiotics. Bacterial intracellular survival, combined with impaired mucociliary clearance in the airways of CF patients, could help sustaining the infection. We observed that infected phagocytes (amoebae or macrophages) containing intracellular bacteria eventually die (Marolda et al., 1999; Saini et al., 1999). By resisting intracellular bactericidal mechanisms, it is conceivable that B. cepacia complex strains could elicit, directly or indirectly, the death of the host phagocyte followed by the subsequent release of live bacteria.

We know very little about the mechanism of intracellular survival by B. cepacia complex isolates and the contribution of intracellular survival to lung disease. Our previous studies demonstrated that bacteria can only replicate extracellularly as they grow at very high densities in amoebae-free conditioned medium (Marolda et al., 1999). The apparent lack of intracellular replication, combined with the extreme antibiotic resistance of members of the B. cepacia complex (Aaron et al., 2000; Nzula et al., 2002), complicates the analysis of experimental infections using amoebae and macrophages (Marolda et al., 1999; Saini et al., 1999). Classical invasion assays based on aminoglycoside antibiotics to kill differentially extracellular bacteria without affecting the viability of intracellular bacteria (Finlay et al., 1988) are very difficult with B. cepacia complex strains because these assays require very large concentrations of antibiotics and even under those conditions it is nearly impossible to ensure that all extracellular bacteria have been inactivated. Amoebae provided us with an attractive model system given that we can conduct infections in minimal medium whereby extracellular bacteria grow very poorly (Marolda et al., 1999).

We have recently constructed stable plasmids containing constitutive and regulated promoters for gene expression in B. cepacia, and we also adapted the enhanced green fluorescent protein (eGFP) for use in B. cepacia (Lefebre and Valvano, 2002). In this study, we conducted colocalization experiments of eGFP-labelled B. cepacia complex with the fluid phase marker Lysosensor™ Blue in A. polyphaga cells. We show that intracellular bacteria reside within an acidified membrane-bound compartment that is distinct from the lysosome, and we also provide statistical evidence by microscopic single-cell analyses demonstrating that intracellular survival occurs without bacterial replication, which clearly sets B. cepacia complex apart from more classical intracellular bacteria.

Results

Intracellular B. cepacia complex isolates do not replicate within Acanthamoeba

We and others have previously shown that B. cepacia complex isolates from clinical and environmental sources can be detected intracellularly within membrane-bound vacuoles in professional and non-professional phagocytes (Burns et al., 1996; Marolda et al., 1999; Saini et al., 1999; Martin and Mohr, 2000; Keig et al., 2002). However, the ability of these bacteria to replicate within eukaryotic cells has been difficult to establish conclusively. The high resistance of B. cepacia complex isolates to antimicrobials that are commonly used for eliminating extracellular bacteria in classical invasion assays precludes an accurate quantification of the bacterial intracellular load in experimental infections. For B. cepacia complex, these assays require extremely high concentrations of a combination of ceftazidime and kanamycin (up to 1 mg ml−1 and 0.5 mg ml−1 respectively) to eliminate the remaining extracellular bacteria. But even under these conditions there is a quantifiable extracellular bacterial growth that complicates the interpretation of the assays based on viable counts (Saini et al., 1999). Furthermore, at such high antibiotic concentrations, the entry of the antibiotics into the host cells cannot be ruled out. We therefore used a strategy based on a comparative quantification of the intracellular bacterial load in A. polyphaga by single-cell microscopic analyses that involved counting the number of bacteria within vacuoles per amoeba cell in the absence of antibiotics. Data were obtained by counting intracellular bacteria in two to three amoeba cells per field of view over a total of 10–12 fields at 2 and 24 h post infection. The counts were treated statistically using non-parametric statistics, because we could not make prior assumptions about the distribution of the data. The Kruskal–Wallis test was chosen as it is appropriate to compare samples containing mutually independent observations from two or more groups. The positive control for these experiments was the Legionella pneumophila strain 2064, a bacterium well known for its capacity to survive and replicate within Acanthamoeba (Bozue and Johnson, 1996; Kwaik, 2000; Fields et al., 2002). As a negative control, we used laboratory Escherichia coli K-12 strains DH5α and ML35, both of which are rapidly destroyed by Acanthamoeba after phagocytosis. The B. cenocepacia strain H111 and the B. vietnamiensis CEP040 were B. cepacia complex clinical isolates from patients with CF. The results demonstrated that both L. pneumophila and B. cenocepacia H111 were phagocytized more readily than B. vietnamiensis CEP040 (median of eight bacteria/cell and two bacteria/cell respectively), as determined by the number of intracellular bacteria present in vacuoles at 2 h post infection in each case (Fig. 1). At 24 h post infection, the number of B. vietnamiensis CEP040 per amoeba cells did not vary significantly relative to that at 2 h (median of one bacterium/cell and two bacteria/cell respectively; P = 0.9003). Also, the number of intracellular B. cenocepacia H111 at 24 h post infection decreased significantly (P = 0.0001) relative to the levels of bacteria at 2 h. However, the numbers of intracellular B. cenocepacia H111 were similar to those of intracellular B. vietnamiensis CEP040 at 24 h post infection. In contrast, a significantly higher number of L. pneumophila was found in intracellular membrane vacuoles (median of nine bacteria/cell; P = 0.0001) at 24 h post infection. Multiplying bacteria within the vacuoles were commonly observed in Legionella-infected amoebae (Fig. 2C, insert), but not found in amoeba infected with B. cenocepacia H111 and B. vietnamiensis CEP040 (Fig. 2A and B). In the infections with E. coli, most bacteria displayed abnormal morphologies at 2 h post infection (see below), and intracellular bacteria were detectable only occasionally in a few amoeba cells at 24 h (Fig. 2D), suggesting that these bacteria are rapidly degraded after phagocytosis. From these experiments, we concluded that B. cenocepacia H111 and B. vietnamiensis CEP040 can survive in a membrane-bound compartment within amoeba for at least 24 h, but they cannot replicate intracellularly.

Figure 1.

Statistical comparison of median numbers of intracellular bacteria per amoebae at 2 h and 24 h post infection using the Kruskal–Wallis one-way Anova. H111, B. cenocepacia H111; Leg, L. pneumophila strain 2064; CEP040, B. vietnamiensis CEP040.

Figure 2.

Phase contrast microscopy of A. polyphaga infected with B. cenocepacia H111 (A), B. vietnamiensis CEP040 (B), L. pneumophila 2064 (C) and E. coli DH5α (D). Images were captured at 1000× magnification at 24 h. White arrows point to vacuoles containing one or more intracellular bacteria. The insert in (C) shows a digitally enlarged region of a large vacuole containing dividing (black arrows) L. pneumophila 2064 bacterial cells.

B. vietnamiensis localize within an acidic vacuole that is distinct from the lysosome

We utilized the fluid marker Lysosensor™ Blue DND-167 as a probe to characterize the B. cepacia-containing vacuoles. Because of its pKa value of 5.0, DND-167 has very weak fluorescence at neutral pH when excited with UV light but emits an intense blue fluorescence at pH values below 5.0 (Lin et al., 2001). B. vietnamiensis CEP040 bacterial cells were present within low-pH vacuoles as early as 2 h after infection (data not shown) and remained in an acidic compartment at 24 h (Fig. 3). Similar observations were made for B. cenocepacia H111 (data not shown). A control experiment with Salmonella enterica serovar Typhimurium, which localizes within acidic vacuoles after internalization (Rathman et al., 1996; Valdivia and Falkow, 1996), also revealed intracellular bacteria within intense blue fluorescent vacuoles (Fig. 3). Similarly, we expected to find E. coli K-12 DH5α within intense blue fluorescent vacuoles as these bacteria are rapidly degraded by amoebae and they probably end up within lysosomes. Intracellular E. coli also displayed morphological alterations after 2 h post-phagocytosis, while intracellular S. enterica and B. vietnamiensis CEP040 displayed an intact morphology not only at 2 h but also at 24 post infection (Fig. 3, inserts, and data not shown). Because Lysosensor™ Blue DND-167 only provides a qualitative assessment of the acidic status of intracellular vacuoles, we conducted similar infection experiments with L. pneumophila, a bacterium that prevents acidification of the phagosome in human monocytes (Horwitz and Maxfield, 1984), especially during at least 8 h after phagocytosis (Sturgill-Koszycki and Swanson, 2000). Figure 3 shows that intracellular bacteria in amoebae infected with L. pneumophila were located within vacuoles that either did not fluoresce or gave a weak blue fluorescence in the presence of Lysosensor™ Blue DND-167. This result is in agreement with a recent report indicating that L. pneumophila internalized in the free living Dictyostelium discoideum does not colocalize with a monoclonal antibody staining the V-ATPase (Solomon et al., 2000). Therefore, to confirm that B. vietnamiensis CEP040 and B. cenocepacia H111 are indeed localized with acidic vacuoles, infected amoebae were treated with 5 µM bafilomycin A1. This compound inhibits the function of the vacuolar type H+-ATPase (Sundquist et al., 1990) and therefore prevents acidification of the phagosome (Yoshimori et al., 1991). Figure 4 shows that bafilomycin A1 caused a dramatic reduction in the intensity of the blue fluorescence of vacuoles containing B. vietnamiensis CEP040, compared to untreated cells. In fact, most intracellular bacteria were found in vacuoles that did not fluoresce blue when the amoeba cells were treated with bafilomycin (Fig. 4, arrows). Similar results were obtained with amoebae infected with B. cenocepacia H111 (data not shown). Altogether, the results of these experiments are consistent with the notion that B. vietnamiensis CEP040 and B. cenocepacia H111 survive in amoebae within an acidified intracellular compartment.

Figure 3.

Intracellular bacteria localize within in acidic vacuoles. A. polyphaga amoeba cells were infected with E. coli K-12 strain DH5α, L. pneumophila 2064, S. enterica serovar Typhimurium strain Sa1355 and B. vietnamiensis CEP040. Before fluorescence imaging, amoebas were treated with Lysosensor™ Blue. The images represent merged phase contrast and fluorescence microscopy at 2 h (E. coli, L. pneumophila and S. enterica) and at 24 h (B. vietnamiensis) post infection all taken at a magnification of 1000×. The inserts in each panel are digital magnification of regions of the amoeba cells (indicated by the dotted-lined squares) with vacuoles containing bacterial cells. Arrowhead denotes an amoebae cyst.

Figure 4.

Intracellular bacteria localize within a neutral vacuole in the presence of the V-ATPase inhibitor bafilomycin A1. Amoebae were infected with B. vietnamiensis CEP040 and treated with 5 µM bafilomycin A1 for 15 min or Acanthamoeba buffer before adding Lysosensor™ Blue. The images represent merged phase contrast and fluorescence microscopy at 2 h post infection all taken at a magnification of 1000×. Arrows indicate vacuoles containing bacteria.

The experiments described above could not distinguish whether the B. cepacia-containing vacuole corresponded to a lysosome or an acidic phagosome stalled in its maturation pathway. For instance, intracellular S. enterica and L. pneumophila prevent the fusion of the bacteria containing vacuoles with the lysosome (Buchmeier and Heffron, 1991; Amer and Swanson, 2002), and a potentially similar scenario could be applicable to B. cepacia complex strains. To our knowledge, there are no specific lysosomal markers readily available for A. polyphaga, and preliminary experiments with the lysosomal marker Lysotracker did not provide any results (J. Cardelli, pers. comm., and data not shown). Therefore, we employed cationized ferritin, an electron dense compound that can be chased into the lysosome and can be detected by transmission electron microscopy (Brouqui and Raoult, 1993; Biederbick et al., 1995). Cationized ferritin was then internalized by amoeba after infection and used to label secondary lysosomes. As positive and negative controls, we exposed amoebae to two E. coli K-12 strains, DH5α and ML35, and also L. pneumophila strains 2064 respectively. Because E. coli is unable to survive within amoeba, we reasoned that these bacteria could not prevent phagolysosome fusion and therefore they should colocalize with vesicles containing cationized ferritin. In contrast, phagosomes containing L. pneumophila 2064 should not colocalize with cationized ferritin because this pathogen is known to inhibit phagolysosome fusion. Figure 5A shows that at 2 h post infection, cationized ferritin was found within vacuoles containing E. coli. Bacterial cells were partially degraded, and in some cases the fusion of the vesicles containing ferritin and those with bacteria were frequently observed. As expected, vesicles containing L. pneumophila did not colocalize with those containing cationized ferritin at 2 h or at 24 h after bacterial internalization (Fig. 5B and data not shown). Similarly, no colocalization of cationized ferritin and intracellular bacteria was found in amoebae infected with B. vietnamiensis CEP040 (Fig. 5C and D). From these results we conclude that intracellular B. vietnamiensis can survive within an acidified membrane-bound vesicle that is distinct from the lysosome.

Figure 5.

Transmission electron microscopy of infected amoeba cells treated with cationized ferritin. Amoebae were infected with the following bacteria: E. coli ML35 (A), L. pneumophila 2064 (B) and B. vietnamiensis CEP040 (C and D). Cells were fixed at 2 h (A and C) and 24 h (B and D) post infection. One or more bacterial (b) cells within vacuoles are indicated. Arrow in (A) indicate partially degraded bacteria cover with ferritin particles. The inserts in each panel are digital magnification of selected regions of the electron micrographs (indicated by the dotted-lined squares) showing more structural details of the compartments containing electron-dense cationized ferritin particles (arrowheads). Bar, 0.5 µM (A) and 1 µM (B–D).

Intracellular B. vietnamiensis remain viable for at least 24 h after infection

Intracellular survival of B. cepacia complex without apparent replication prompted us to investigate whether the bacteria remain viable. Because of the difficulties with the use of membrane impermeable antibiotics as discussed above, we could not accurately use viable cell counts of intracellular bacteria as a measure of viability. Thus, to assess the viability of the intracellular bacteria within A. polyphaga vacuoles, we treated bacteria before infection with 5-chloromethylfluorescein diacetate (CellTracker™ Green CMFDA). The thiol-reactive chloromethyl moieties of CMFDA react with intracellular thiols and their acetate groups are cleaved by cytoplasmic esterases, generating a fluorescent product that cannot escape the cytosol unless the membrane is damaged. Using this probe, we observed green fluorescent B. vietnamiensis CEP040 bacteria within vacuoles at 2 h and 24 h post infection (Fig. 6 and data not shown). The simultaneous use of Lysosensor™ Blue also showed that the vacuoles containing green fluorescent bacteria were acidic. Therefore, this experiment suggested that the bacterial cells had an intact cell envelope because the modified CMFDA molecules were retained in the cytoplasm. Unfortunately, the permeability of B. cepacia to CMFDA and related compounds is low (Fuller et al., 2000). In our hands, only ≈ 20–30% of the bacteria were stained with the probe at concentration of 5 µg ml−1. This explains why not all intracellular bacteria were fluorescently labelled (Fig. 6). Because of this limitation with the CMFDA probe, we could not completely rule out the possibility that intracellular bacteria could remain apparently intact but non-viable. Therefore, we used another strategy to demonstrate viability which was based on the expression of eGFP by B. vietnamiensis CEP040 carrying the plasmid pMLBAD-eGFP. We have previously demonstrated that this plasmid is very stable in B. cepacia complex isolates (Lefebre and Valvano, 2002) and also that the presence of pML-eGFP in B. vietnamiensis CEP040 did not alter the characteristics of the amoeba infection as compared to the plasmidless counterparts (data not shown). Because eGFP can only fold properly in the bacterial cytoplasm (Feilmeier et al., 2000), we reasoned that a bacterial cell with a damaged cell envelope should lose its fluorescence after phagocytosis and as a consequence the entire phagocytic vacuole content should fluoresce. Figure 7A shows that CEP040(pML-eGFP) bacterial cells retained their cytoplasmic fluorescence within the vacuoles at 2 h post infection. To demonstrate that inactivated bacteria actually release eGFP from the cytoplasm, we conducted a control experiment using heat-killed bacteria (treated at 60°C for 30 min). Both live and heat-killed bacteria kept in Acanthamoeba buffer for 2 h at 25°C remained fluorescent (Fig. 7A and B), suggesting that the eGFP molecules remained in the bacterial cytosol for at least a similar amount of time as the duration of the 2 h infection experiments. However, the entire vacuole content was fluorescently labelled in the infection experiment using heat-killed bacteria (Fig. 7B). It is very likely that heat-killed and live bacteria end up in different vacuoles. However, these results demonstrate that only inactivated bacteria release cytosolic eGFP to the vacuolar lumen, suggesting that live bacteria can maintain their membrane integrity within the intracellular environment. Altogether, the combined data of the experiments using CMFDA and eGFP support the notion that intracellular B. vietnamiensis CEP040 remains intact and viable.

Figure 6.

Viability of intracellular B. vietnamiensis CEP040 determined with CellTracker™ Green determined at 24 h post infection. Before infection bacteria were treated with CellTracker™ Green as described under ‘Experimental procedures’. Infected amoebae were treated with Lysosensor™ Blue before imaging. Phase contrast (A), and merged phase contrast and fluorescent images (B and C) at a magnification of 1000×. The inserts in each panel are digital magnification of regions of the amoeba cells (indicated by the dotted-lined squares) with an acidic vacuole (blue fluorescent content by the pH-dependent activation of the fluid phase probe Lysosensor™ Blue, C) containing two bacterial cells that retain the CellTracker™ probe (green fluorescence, B).

Figure 7.

Viability of untreated (A) and heat-killed (B) intracellular B. vietnamiensis CEP040(pMLBAD-eGFP) expressing the enhanced green fluorescent protein determined at 2 h post infection. Aliquots of bacterial cells incubated in Acanthamoeba buffer retained fluorescence in the cytoplasm (control). Phase contrast images of amoebae cells with vacuoles containing one or more intracellular bacteria (arrows). Images captured in green fluorescent mode show intact bacteria (fluorescence retained in the cytosolic) in the case of the infection with live B. vietnamiensis CEP040(pMLBAD-eGFP) (A), and bacteria with a compromised envelope (fluorescence leaked to the vacuolar space) (B). All of the images were captured at a magnification of 1000×.

Discussion

The ability of B. cepacia complex isolates to survive intracellularly has been reported by several investigators (Burns et al., 1996; Marolda et al., 1999; Saini et al., 1999; Martin and Mohr, 2000; Keig et al., 2002). However, it remains unclear whether or not these bacteria can multiply intracellularly. The conventional assays used to determine intracellular multiplication rely heavily on the quantification of intracellular bacteria after the lysis of the host cells in the presence of a low concentration of mild detergents. But these assays require the efficient killing of extracellular bacteria using antibiotics that cannot cross the cell membrane of host cells, and numerous washes to remove adherent bacteria. Because of the extraordinary antibiotic resistance of B. cepacia, exceedingly high concentrations of antibiotics have to be used to eliminate extracellular bacteria. Studies in our laboratory with amoeba and murine macrophages (Marolda et al., 1999; Saini et al., 1999) have shown that even after adding to the infection assays ceftazidime and kanamycin at concentrations of 1 mg ml−1 and 0.5 mg ml−1, respectively, bacteria can still be recovered from the last wash before the lysis of the host cells, and the bacterial numbers continued to increase over the time-course of the assay. Therefore, a direct enumeration of intracellular bacteria by the microscopic analysis of single cells was used in the current study. We hypothesized that if intracellular bacteria multiply, a significantly higher number of bacterial cells should be visualized within infected cells over time. The results clearly showed that in infections with B. cepacia complex strains, these numbers either decreased or remained relatively constant along the 24 h observation period, suggesting that bacterial multiplication did not occur. These results could potentially result from a dynamic equilibrium between continued bacterial phagocytosis and intracellular bacterial killing. However, examination of intracellular B. cepacia by electron microscopy at various times after infection both in this study using amoebae and in a previous study with murine macrophages (Saini et al., 1999) did not show evidences of bacterial lysis. Also, we have previously shown (Marolda et al., 1999; Saini et al., 1999) that many strains from various species of the B. cepacia complex can survive for an extended period of time in amoebae (up to 17 days) and macrophages in culture (up to 6 days) without a noticeable increase in the number of intracellular bacteria. Therefore, we conclude that intracellular B. cepacia complex strains are not able to replicate, at least under our experimental conditions.

In a previous study, we have observed a few dividing intracellular B. cepacia, but only at early times shortly after phagocytosis (Marolda et al., 1999; Saini et al., 1999). The mechanism of inhibition of bacterial replication after intracellular localization is not known. We favour the hypothesis that the bacteria are in a state similar to stationary phase, but experiments to directly confirm this hypothesis have not been carried out. There are a number of observations suggesting that intracellular B. cepacia are metabolically active. First, intracellular B. cenocepacia and B. vietnamiensis are highly motile in wet mounts and the bacterial cell envelope remains intact under the electron microscope. Second, intracellular bacteria do not stain the vital fluorescent stain propidium iodide, which only accumulates within damaged cells (C.L. Marolda and M.A. Valvano, unpubl.). Third, we have shown in this article that live intracellular bacteria retain eGFP and CMFDA within the cytoplasm, suggesting that the bacterial cell envelope is intact.

In the absence of readily available specific markers of the endosomal pathway for Acanthamoebae, we used a fluid phase probe to characterize the B. cepacia complex-containing vacuoles. Lysosensor™ Blue is a non-fluorescent compound that freely diffuses through the eukaryotic membrane and becomes highly fluorescent at a pH lower than 5.0, labelling acidic vacuoles. Using this probe, we demonstrated that the B. cepacia complex-containing vacuoles become acidic shortly after phagocytosis, and remain acidic for at least 24 h. Such a low pH environment would suggest that the B. cepacia-containing vacuole could be a lysosome. However, further investigations with cationized ferritin demonstrated that the acidic compartment within which B. vietnamiensis CEP040 localized was not a lysosome, as revealed by the exclusion of the cationized ferritin from B. cepacia complex-containing vacuoles. In contrast, live E. coli were rapidly directed to the lysosomal compartment. It was shown recently that certain intracellular pathogens, such as Brucella abortus, can redirect their trafficking to the autophagosome pathway (Pizarro-Cerda et al., 1998). We also investigated this possibility with B. cepacia complex isolates by using monodansyl cadaverine, a fluorescent probe known to localize specifically in autophagosome vacuoles in macrophages and other types of cells (Biederbick et al., 1995; Arenas et al., 2000) and the autophagosome formation inhibitor 3-methyladenine (Seglen and Gordon, 1982). But unfortunately, these experiments provided inconclusive results (J. Lamothe and M.A. Valvano, unpubl.). In addition, we observed by electron microscopy that the B. cepacia-containing vacuoles did not present double membranes, which are present in typical autophagosome vacuoles. Therefore, it is unlikely that the B. cepacia-containing vacuoles are autophagosomes. The ability of viable B. cepacia to reside in an acidic compartment that is not a lysosome suggests that bacteria have a means to redirect or stall the physiological maturation of the phagosome. This property could be associated to the secretion of bacterial proteins into the host cells via type III or type IV secretory systems. An examination of the genomic sequence of B. cenocepacia J2315 reveals two gene clusters encoding a type IV and a type III secretory systems (http://www.sanger.ac.uk/Projects/B_cenocepacia/). Recent work by Tomich et al. (2003) has shown that a type III secretion-defective mutant of B. cenocepacia J2315 causes reduced tissue inflammation in a mouse model of lung infection. Thus, it is tempting to speculate that effectors secreted by a type III secretory system are also required for intracellular survival. Using signature-tagged mutagenesis, we recently isolated mutants of B. cenocepacia that are attenuated for survival in a rat model of lung infection (Hunt et al., 2004). We are currently examining these mutants for intracellular survival in our amoebae and macrophage models. Preliminary experiments indicate that one of these mutants, carrying a transposon insertion within the B. cenocepacia mgtC gene, cannot survive in macrophages (K.E. Maloney and M.A. Valvano, unpubl.). The mgtC gene, which is highly expressed by intracellular S. enterica (Eriksson et al., 2003), has been shown to be important for the intracellular survival of S. enterica (Blanc-Potard and Groisman, 1997) and Mycobacterium tuberculosis (Buchmeier et al., 2000).

To our knowledge, the present study has provided for the first time a characterization of the B. cepacia-containing vacuoles in amoebae and demonstrated that at least two different strains from two species of the B. cepacia complex can survive within an acidic compartment that is distinct from the lysosome, suggesting that the bacteria can alter the normal maturation of the phagosome. Because intracellular survival without replication is not only limited to the strains investigated in this study but is also found with other strains from the B. cepacia complex (Marolda et al., 1999; Saini et al., 1999), it is possible that a common mechanism of intracellular survival exists in these strains, which may be important to explain their persistence in various different hosts, including human cells. Further research is needed to characterize the molecular mechanism of bacterial survival not only within amoebae but particularly within macrophages. These studies are currently underway in our laboratory, and will contribute to further our understanding of the pathogenesis of this opportunistic bacterium.

Experimental procedures

Bacterial strains, amoeba and culture conditions

Burkholderia vietnamiensis strain CEP040 was previously classified as B. cepacia complex genomovar V (Burns et al., 1996; Marolda et al., 1999) and B. cenocepacia strain H111 was formerly a B. cepacia complex genomovar III isolate (Huber et al., 2002). Both strains were obtained from patients with CF. The clinical isolate of L. pneumophila strain 2064 belonging to serogroup-1 was obtained from Dr Rafael Garduno, Dalhousie University Halifax, Canada. Salmonella enterica serovar Typhimurium strain SA1355 was obtained from Dr K.E. Sanderson, University of Calgary, Canada. E. coli K-12 strains DH5α and ML35 were from our laboratory stocks. E. coli, S. enterica and B. cepacia complex strains were grown at 37°C in peptone/yeast extract/glucose (PYG) broth (Page, 1976). B. cepacia strains carrying the arabinose inducible plasmid pMLBAD-eGFP (Lefebre and Valvano, 2002) were grown in presence of 2% (w/v) arabinose and a final concentration of 100 µg ml−1 trimethoprim. L. pneumophila 2064 was grown at 37°C on buffered charcoal yeast extract medium (Pasculle et al., 1980). A. polyphaga strain JAC/S2 (ATCC 50372) was obtained from the American Type Culture Collection, Manassas, VA, USA. Amoeba cultures were maintained axenically in PYG medium at 25°C as monolayers in 25 cm2 flask (Falcon 3013; Oxnard).

Amoeba infection assays

Amoebae were resuspended by tapping the culture flask and cells were counted with a hemocytometer. Amoebae and bacteria were co-cultivated on four-well Permanox slides (Nalge Nunc International 177437; Rochester) in Acanthamoeba buffer (Page, 1976). Amoebae were seeded at a density of 105 cells per well and once reattached to the bottom, they were washed twice with Acanthamoeba buffer. The monolayers were incubated overnight at 25°C in this buffer. Bacteria grown overnight were washed twice with Acanthamoeba buffer and added to amoebae at a multiplicity of infection (moi) of 10 bacteria per amoeba. Infected monolayers were centrifuged at 1000 r.p.m. for 2 min and incubated for 2 h at 25°C. Amoeba infections with B. cepacia carrying pMLBAD-eGFP were carried out in the presence of 2% arabinose and 100 µg ml−1 trimethoprim. After 2 h incubation, external bacteria were removed by three washes with Acanthamoeba buffer and incubation continued. Control experiments indicated that the washes removed more than 95% of extracellular bacteria. Also, B. cepacia complex isolates replicate very slowly in Acanthamoeba buffer (Marolda et al., 1999). The infected monolayers were observed under the microscope 2 h and 24 h post infection. In some experiments, infected amoebae were treated with 5 µM bafilomycin A1 (Sigma) for 15 min.

Fluorescence microscopy and fluid probes

Infected A. polyphaga monolayers were incubated for 3 min with a final concentration of 10 µM of Lysosensor™ Blue DND-167 (Catalog number L-7533, Molecular Probes) in Acanthamoeba buffer. B. vietnamiensis strain CEP040 was incubated at 37°C in Acanthamoeba buffer for 15 min in the dark with a final concentration of 5 µg ml−1 of 5-chloromethylfluorescein diacetate (CellTracker™ Green CMFDA; Catalog number C-7025, Molecular Probes). Fluorescence and phase contrast images were acquired using a Qimaging (Burnaby) cooled charged-coupled device camera on an Axioscope 2 (Carl Zeiss) microscope with an X100/1.3 numerical aperture Plan-Neofluor objective and a 50 W Mercury arc lamp. The following filters were used: GFP band pass emission filter set (Chroma Technology) with a 470 ± 20 nm excitation range and a 525 ± 25 nm emission range and Blue filter set 02 (Chroma Technology) with a 365 ± 20 nm excitation range and a 420 nm emission range. Images were digitally processed using the Northern Eclipse version 6.0 imaging analysis software (Empix Imaging).

Electron microscopy

Infected amoeba were washed three times with Acanthamoeba buffer to remove the external bacteria and incubated in 1:10 cationized ferritin (Catalog number F-7879, Sigma Chemical Co.) in Acanthamoeba buffer for 1 h at 20°C. The monolayers were collected by tapping the culture flask gently and amoebae were centrifuged at 2500 r.p.m. for 2.5 min Amoeba were washed and resuspended in 1 ml of Acanthamoeba buffer. Infected amoeba were incubated 1 h in 1 ml of Acanthamoeba buffer at 20°C. Amoeba were fixed in 1 ml of 0.1 M cacodylate buffer containing 2.5% glutaraldehyde overnight at 4°C, centrifuged for 3 min at 10 500 r.p.m and the pellet was overlaid with cacodylate buffer for 10 min. After removal of the buffer, the cells were post-fixed in 1% (w/v) OsO4 for 90 min in cacodylate buffer, enrobed in noble agar, washed in buffer and incubated in uranyl acetate for 120 min with rotation. After incubation, pellets were washed in water and dehydrated in an alcohol series. Pellets were finally washed in propylene oxide and incubated in Epon resin 1:1 with propylene oxide overnight with rotation. Pellets were transferred to pure Epon resin and polymerized at 60°C for 2 days. Ultra-thin sections were post-stained with uranyl acetate and lead citrate and examined in a Philips 300 electron microscope.

Statistical analysis

Data were obtained from 63 to 84 random picked amoebas from three or four different experiments for each infecting bacteria. The number of intracellular bacteria was counted at 2 and 24 h after infection. Total number of bacteria per amoeba was analysed using Kruskal–Wallis one-way Anova for non-parametric data, followed with a Dunn's post-test, if significance is less than 0.05 (GB-Stat version 8.0, Dynamic Microsystems, Inc.). Results are expressed as median and range of each group.

Acknowledgements

The authors thank C.L. Marolda, S.F. Koval, D. Laird and J. Cardelli for useful comments and technical advice, and R. Garduno and K.E. Sanderson for strains. The electron microscopy was conducted in the Electron Microscopy Research Facility of the Faculty of Medicine and Dentistry, University of Western Ontario, with the expert assistance of J. Sholdice. This work was supported by operating grants from the Canadian Institutes of Health Research and the Canadian Cystic Fibrosis Foundation to M.A.V. The Microbiology Light Microscopic Facility was supported by grants from the Academic Development Fund of the University of Western Ontario and the Canadian Institutes of Health Research to M.A.V. J.L. was supported by a studentship from the Canadian Cystic Fibrosis Foundation. M.A.V. holds a Canada Research Chair in Infectious Diseases and Microbial Pathogenesis.

Ancillary