Cholesterol depletion in Mycobacterium avium-infected macrophages overcomes the block in phagosome maturation and leads to the reversible sequestration of viable mycobacteria in phagolysosome-derived autophagic vacuoles

Authors

  • Chantal De Chastellier,

    Corresponding author
    1. Centre d’Immunologie de Marseille-Luminy, INSERM-CNRS-Université de la Méditerranée, Marseille, France.
      *E-mail dechastellier@ciml.univ-mrs.fr; Tel. (+33) 4 91 26 91 22; Fax (+33) 4 91 26 94 30.
    Search for more papers by this author
  • Lutz Thilo

    1. Division of Medical Biochemistry, Faculty of Health Sciences, University of Cape Town, South Africa.
    Search for more papers by this author

*E-mail dechastellier@ciml.univ-mrs.fr; Tel. (+33) 4 91 26 91 22; Fax (+33) 4 91 26 94 30.

Summary

Phagocytic entry of mycobacteria into macrophages requires the presence of cholesterol in the plasma membrane. This suggests that pathogenic mycobacteria may require cholesterol for their subsequent intra-cellular survival in non-maturing phagosomes. Here we report on the effect of cholesterol depletion on pre-existing phagosomes in mouse bone marrow-derived macrophages infected with Mycobacterium avium. Cholesterol depletion with methyl-β-cyclodextrin resulted in a loosening of the close apposition between the phagosome membrane and the mycobacterial surface, followed by fusion with lysosomes. The resulting phagolysosomes then autonomously executed autophagy, which did not involve the endoplasmic reticulum. After 5 h of depletion, intact mycobacteria had accumulated in large auto-phagolysosomes. Autophagy was specific for phagolysosomes that contained mycobacteria, as it did not involve latex bead-containing phagosomes in infected cells. Upon replenishment of cholesterol, mycobacteria became increasingly aligned to the lysosomal membrane, from where they were individually sequestered in phagosomes with an all-around closely apposed phagosome membrane and which no longer fused with lysosomes. These observations indicate that, cholesterol depletion (i) resulted in phagosome maturation and fusion with lysosomes and (ii) caused mycobacterium-containing phagolysosomes to autonomously undergo autophagy. Furthermore, (iii) mycobacteria were not killed in auto-phagolysosomes, and (iv) cholesterol replenishment enabled mycobacterium to rescue itself from autophagic phagolysosomes to again reside individually in phagosomes which no longer fused with lysosomes.

Introduction

Pathogenic mycobacteria survive in phagosomes of their host organism's macrophages. Initial electron-microscope observations have shown that Mycobacterium tuberculosis-containing phagosomes do not fuse with lysosomes (Armstrong and D’Arcy Hart, 1971). These observations were interpreted to mean that avoidance of the hostile intra-lysosomal environment underlies the mycobacterium's ability for intra-cellular survival, but it remained to be demonstrated whether mycobacteria are invariably killed when in phagolysosomes. Recent evidence shows that mycobacteria do not directly prevent phagosome–lysosome fusion, but interfere with the preceding step of phagosome maturation (de Chastellier et al., 1995; Clemens and Horwitz, 1995; Sturgill-Koszycki et al., 1996; Via et al., 1997). As is the case for early endosomes (Aniento et al., 1993; Ward et al., 2000), immature phagosomes do not fuse with lysosomes (de Chastellier et al., 1995; Clemens and Horwitz, 1995). Several molecular mechanisms were seen to play a role in, or correlate with, the block in phagosome maturation (reviewed in, e.g. Deretic and Fratti, 1999; Russell, 2001; Thilo and de Chastellier, 2003; Chua et al., 2004; see also under Discussion).

Based on morphological observations, we have previously pointed out that a necessary requirement for a block of maturation of the mycobacterium-containing phagosome is the close apposition of the phagosome membrane with the entire mycobacterial surface all around (de Chastellier and Thilo, 1997). When mycobacteria are phagocytosed in clumps, the resultant phagosome contains more than one mycobacterium (we refer to these as ‘social’ phagosomes, in contrast to ‘loner’ phagosomes, which contain a single mycobacterium). Because the phagosome membrane is distant from the mycobacterial surface where it spans the regions between adjacent mycobacteria, these social phagosomes invariably mature and fuse with lysosomes (de Chastellier et al., 1995; Clemens and Horwitz, 1995; de Chastellier and Thilo, 1997). An all-around close apposition of the phagosome membrane appears to be so crucial that a dividing intra-phagosomal mycobacterium maintains such contacts into the site of septum formation (de Chastellier and Thilo, 1997).

The molecular components involved in the maintenance of a close interaction between the phagosome membrane and the mycobacterial surface are not known, but should result in a particular protein and lipid composition of the phagosome membrane. We have shown that the membrane of phagosomes containing virulent Mycobacterium avium and M. tuberculosis becomes selectively depleted for cell surface-derived glycoconjugates (de Chastellier and Thilo, 2002; Pietersen et al., 2004). As plasma-membrane cholesterol is required for a stable interaction of mycobacteria with the cell surface during phagocytic uptake (Gatfield and Pieters, 2000; Peyron et al., 2000), we here examined the effects of cholesterol depletion on the close interaction between the phagosome membrane and the surface of phagocytosed M. avium in mouse bone marrow-derived macrophages. We found that the close interaction was abolished, leading to phagosome maturation and fusion with lysosomes. The resultant phagolysosomes themselves were induced to form large autophagic vacuoles that eventually harboured many mycobacteria in a single auto-phagolysosome. This behaviour was not observed for phagolysosomes containing latex beads, nor for lysosomes in general. Interestingly, mycobacteria in auto-phagolysosomes remained intact and fully viable, and, upon replenishment of cholesterol, were able to reestablish themselves in individual phagosomes that no longer fused with lysosomes.

Results

Cholesterol depletion induced phagosome–lysosome fusion

Exposure of macrophages to methyl-β-cyclodextrin (CD) leads to a gradual depletion of cholesterol cellular membranes. In a first set of experiments, cells were exposed to different concentrations of CD, between 0.5 and 10 mM, in order to find the most suitable conditions for effective cholesterol depletion without irreversible cell damage over the 6 h incubation period. A treatment with 5 mM CD for 0–6 h gave the best results and was therefore adopted for the following experiments. As reported previously (Hur et al., 2004) and as confirmed for the present study, total cholesterol content decreased rapidly upon exposure to the drug at 5 mM (Fig. 1A). In order to visualize cholesterol within the phagosome membrane and determine whether treatment with CD extracts it from the membrane, M. avium-infected and CD-treated cells were incubated with filipin during the fixation procedure. Such a treatment, when applied during fixation, induces the formation of filipin-sterol complexes in the hydrophobic core of the membrane bilayer. As a result, cholesterol-containing membranes display a typical corrugated aspect (Coppens and Joiner, 2003). With this method, the plasma membrane and the membrane of loner phagosomes, when it was still tightly apposed to the entire bacterial surface all around (see below), displayed a regularly corrugated pattern (Fig. 1B). A decrease of cholesterol in the phagosome membrane, as a result of a 4 h treatment with CD, was indicated by its less corrugated and more smoothened appearance (Fig. 1C).

Figure 1.

Cholesterol depletion in mouse bone marrow-derived macrophages and visualization of cholesterol in the phagosome membrane.
A. Cells grown to confluence (9 days) were washed in serum-free medium and then incubated in the presence of 5 mM methyl-β- cyclodextrin (CD) at 37°C for the indicated times (closed symbols). Control cells were incubated in serum-free medium only (open symbols). Cells were released with trypsin, washed and sonicated for cholesterol measurement. 100% corresponded to about 10 nmol cholesterol/106 cells.
B, C. Infected cells, either untreated (B) or CD-treated (C) were fixed in the presence of filipin for an ultrastructural visualization of cholesterol in the phagosome membrane: (B) a loner phagosome is shown with its membrane tightly apposed to the mycobacterial surface; both the phagosome membrane and the plasma membrane (PM), display a regularly corrugated pattern (arrows); (C) an auto-phagolysosome (APhLy) is shown after a 4 h treatment with CD; the membrane has become much smoother (arrow), indicating a lesser concentration of cholesterol. Bar = 0.5 µm.

Already within the first 30 min of treatment, cholesterol depletion had a pronounced effect on phagosome morphology and fate. In comparison to untreated cells, where the mycobacterium-containing phagosome membrane remained closely apposed to the mycobacterial surface all around (Fig. 2A, see also de Chastellier et al., 1995; de Chastellier and Thilo, 1997), a 30–120 min exposure to 5 mM CD in serum-free medium caused a dissociation of the phagosome membrane from the mycobacterial surface and converted its tight contour to a partially undulating one (arrow in Fig. 2B). To determine whether the observed undulation of the phagosome membrane was due to increased fusions with endocytic compartments due to cholesterol extraction, cells were incubated for the last 30 min of a 2 h treatment with CD with horseradish peroxidase (HRP), which served as a content marker of endocytic compartments. No more extensive fusion with HRP-positive endocytic compartments was observed for loner phagosomes with an undulating membrane (Fig. 3A) than for loner phagosomes with a tightly apposed membrane (as in untreated cells, de Chastellier et al., 1995). Phagosomes with an undulating membrane were then seen fusing with lysosomes (Fig. 2C and D), in accordance with our proposal that loosening of the phagosome membrane allows phagosome maturation as a prerequisite for fusion with lysosomes (de Chastellier and Thilo, 1997). By measuring the rate at which loner phagosomes were converted to phagolysosomes in response to cholesterol depletion (Fig. 4A, open symbols), it was found that phagosomes started to fuse with lysosomes as early as 60 min after the initiation of the depletion, i.e. after a 25% decrease of total cholesterol content. The abundance of phagolysosomes increased with a half-life of about 4 h and curve fitting predicted that all loner phagosomes would finally become phagolysosomes. That loner phagosomes were indeed processed into phagolysosomes was substantiated by the positive staining for the lysosomal content marker, acid phosphatase (Fig. 3B). The fraction of loner phagosomes which was acid phosphatase-positive increased from 35% before CD treatment to 60% after a 4 h treatment with CD. Because phagolysosomes fused with one another and intermingled contents, loner phagolysosomes gradually converted to ‘social’ phagolysosomes with multiple mycobacteria. Accordingly, the fraction of mycobacteria in loner phagosomes gradually declined (Fig. 4B). At least 85–90% of the social phagosomes stained positive for acid phosphatase. In order to better characterize the M. avium-containing phagosomes during cholesterol extraction, we applied 3-(2,4-dinitroanilino)-3′-amino-N-methyldipropylamine (DAMP) to infected macrophages at selected intervals during treatment with CD. This probe is known to accumulate in acidic compartments of the cell where it becomes covalently linked to proteins in the presence of aldehyde fixatives. This assures its retention in acidic organelles after fixation (de Chastellier et al., 1995). DAMP was localized by immuno-gold labelling. The number of gold particles per phagosome could be converted to values for intra-phagosomal acidification by using the method described by Orci and colleagues (1994). In the absence of treatment with CD, the average pH of loner phagosomes was of 6.1. Of these, immature phagosomes had a pH of 6.5 (as before, de Chastellier et al., 1995), and phagolysosomes had a pH of 5.8. A similar pH value of 5.6 was observed for social phagosomes, for which we know that they mature rapidly and start to fuse with lysosomes. In prelysosomes/lysosomes from the same thin sections, the pH value was of 5.4. Similar values were obtained for these different compartments after a 2 or 4 h treatment with CD.

Figure 2.

Ultrastructural changes of M. avium-containing phagosomes and their fusion with lysosomes in response to cholesterol depletion. Macrophages were infected with M. avium as described under Experimental procedures. At day 6 post infection (p.i.), cells were treated with CD for 2 h, and processed for electron microscopy.
A. Control cell in complete medium, no CD added. The phagosome membrane is smooth and closely apposed to the surface of M. avium. The phagosome remains immature, and therefore does not fuse with surrounding lysosomes (Ly; cf. de Chastellier et al., 1995; de Chastellier and Thilo, 2002).
B, C. After treatment with CD, the phagosome membrane has become loose and undulates around the mycobacterial surface (arrow; this dissociation starts after a 30 min treatment, but is shown here after 2 h with CD). In (C), a phagosome is seen fusing with a lysosome to form a phagolysosome (PhLy) as characterized by its electron-dense material.
D. Cells were loaded for 30 min with BSA-gold, followed by a 2 h chase to label lysosomes and then treated with CD for 2 h. Several phagosomes contain BSA-gold (arrow), which means that they have matured and fused with lysosomes. Bar = 0.5 µm.

Figure 3.

Effect of cholesterol depletion on the acquisition by phagosomes of the endocytic tracer, horseradish peroxidase (HRP), and of the lysosomal enzyme, acid phosphatase (AcPase).
A. At day 6 post infection, macrophages were treated with 5 mM CD for 2 h. HRP (25 µg ml−1) was added to the medium during the last 30 min of treatment to label endocytic compartments. Phagosomes with an undulating membrane (small arrows) displayed very few deposits of HRP reaction product as has been observed previously for control loner phagosomes with a tightly apposed membrane (de Chastellier et al., 1995); a phagosome fusing with an HRP-containing lysosome is indicated (large arrows).
B. Staining for acid phosphatase after a 4 h treatment with CD: phagolysosomes (PhLy) and auto-phagolysosomes (APhLy) displayed dense reaction product (arrow), indicative of their lysosomal characteristics. Bar = 0.5 µm.

Figure 4.

Kinetics of loner phagosomes fusing with lysosomes (A), and the decline of the fraction of M. avium in loner phagosomes (B), as induced by cholesterol depletion. At day 6 post infection, macrophages were treated with 5 mM CD for the indicated times, then processed for electron microscopy.
A. Phagosomes with only one mycobacterium (loner phagosomes) were scored for fusion with lysosomes as observed by the presence of dense lysosomal material or the marker BSA-gold chased to lysosomes; the best-fit curve indicates that the degree of fusion increased with a half-life of about 4 h and suggests that eventually all phagosomes would fuse with lysosomes.
B. The fraction of M. avium in phagocytic vacuoles with only one mycobacterium (loner phagosomes) was scored; this fraction declined to zero with a half-life of about 6 h as the result of fusion with M. avium-containing phagolysosomes and/or autophagic phagolysosomes. In parallel, phagosomes with hydrophobic latex beads (45 min uptake, 1 h chase prior to t = 0) in uninfected CD-treated cells (solid circles) or in uninfected control cells (solid squares) showed delayed or no fusion with lysosomes respectively, and remained in loner phagosomes. Different open symbols refer to three independent experiments. Error bars (where larger than the symbol) indicate the SEM as from independent sections (n = 3) within a single experiment. First-order curves were fitted using GraphPad, Prism 4.03 software (http://www.graphpad.com).

We also looked at the effect of cholesterol depletion on phagosomes that contained latex beads with a hydrophobic surface. As reported previously, and in contrast to beads with a hydrophilic surface, hydrophobic beads maintain an all-around closely apposed phagosome membrane and phagosomes remain immature for at least 3 h (de Chastellier and Thilo, 1997). Unlike for mycobacteria, however, such phagosomes eventually mature after about 12–24 h (Desjardins, 1995). Upon cholesterol depletion, there was a less obvious dissociation of the phagosome membrane from the bead surface (arrow in Fig. 5A) and the membrane did not show the strong wavy contours as in mycobacterium-containing phagosomes. Bead-containing phagosomes also eventually fused with lysosomes (arrows in Fig. 5B and C), but this fusion started about 2 h later than for mycobacterium-containing phagosomes (Fig. 4A, solid circles). The morphology of this fusion process appeared unusual in that thin necks were often seen connecting the bead-containing phagosome with the lysosome (arrow in Fig. 5D) or with a mycobacteria-containing auto-phagolysosome (see next section) in infected cells (arrows in Fig. 5E).

Figure 5.

Ultrastructural changes of latex bead-containing phagosomes and their fusion with lysosomes in response to cholesterol depletion. Macrophages were given hydrophobic latex beads alone for 45 min (A and B) or were first infected with M. avium as described in Fig. 2, and then exposed to latex beads for 45 min at day 6 p.i. (C, D, E). This was followed by a 1 h chase in latex bead-free medium. In both cases, cells were then treated with 5 mM CD for 2 or 4 h prior to fixation.
A, B. Phagocytosis of latex beads (La) only, CD for 2 h. The phagosome membrane is slightly loosened around the latex bead (arrow shows focal loose contact). In (B), the phagosome has fused with small lysosomes as shown by bulges containing dense material, typical for lysosomes (arrow).
C–E. Phagocytosis of M. avium and then latex beads, CD for 4 h. General view in (C) shows both particles in the same cell, but not in the same phagosomes. The latex-containing phagosomes have fused with small lysosomes (arrows), whereas some mycobacteria are in phagolysosomes (PhLy) or in an auto-phagolysosome (APhLy; see Fig. 5). D shows fusion of a latex bead containing-phagosome with a larger lysosome, showing an unusual thin neck connecting the phagosome to the lysosome (arrow). E shows a rare case where a phagolysosome contains both particles. Notice thin necks (arrows) connecting the latex bead-containing phagosome to the M. avium-containing phagolysosome. The latter seems to be in the process of becoming an auto-phagolysosome (APhLy) because it seems to be engulfing the former together with some cytoplasmic material. Bar = 0.5 µm.

Phagolysosome-derived autophagy

In addition to its effect on phagosome morphology and phagosome–lysosome fusion, we made the surprising observation that cholesterol depletion induced phagolysosome-based autophagy. This process was observed only for mycobacterium-containing phagolysosomes. These started to deform and engulf cytoplasmic material, which included existing phagosomes (Fig. 6A–C). As a result, large auto-phagolysosomes were formed that contained several intact mycobacteria (Fig. 6D, with the arrow indicating remnants of the double membrane that is typical for the early stages of autophagic organelles). The membrane of such organelles did not display a regularly corrugated pattern when cells were fixed in presence of filipin, thereby indicating that cholesterol had been extracted from the membrane. Concurrently, fusion of these auto-phagolysosomes with other lysosomes resulted in an almost complete disappearance of normal lysosomes  from  the  cytoplasm  and,  by  extrapolation  as  in Fig. 7, all mycobacteria were predicted to ultimately reside in auto-phagolysosomes. About 90% of the auto-phagolysosomes stained positive for acid phosphatase and their estimated pH value was of 5.7. This value corresponded to the normal level of acidification in lysosomes. Under starvation conditions in Earle's Balanced Salt Solution, this type of autophagy was not encountered (data not shown). Also, although endoplasmic reticulum (ER) was often seen in close proximity to loose non-matured phagosomes, and phagolysosomes (Fig. 8A) or auto-phagolysosomes (Fig. 8B), phagolysosome-derived autophagy did not involve the ER, as based on the absence of the ER marker enzyme, glucose 6-phosphatase (G6Pase), in the lumen of the double membrane of autophagic vacuoles (Fig. 8B). Only on rare occasions (less than 1%) could ER-derived autophagic vacuoles be seen (< 1%, shown in Fig. 8C), probably as in any population of untreated cells.

Figure 6.

Phagolysosome-based autophagy in M. avium-infected cells in response to cholesterol depletion. Cells were infected with M. avium as described in Fig. 2 and treated with 5 mM CD for 2–6 h.
A, B. CD for 2 h. M. avium-containing phagolysosomes (PhLy) become deformed, often adopting a crab-claw-like appearance, as in (B).
C. CD for 4 h. Cytoplasmic material, including other phagosomes, has been engulfed during the formation of an auto-phagolysosome (APhLy), which also contains gold particles accumulated into lysosomes prior to cholesterol depletion.
D. CD for 6 h. This auto-phagolysosome (APhLy) contains several M. avium, all morphologically intact and surrounded by their typical electron translucent layer. The APhLys are spherical at this stage. Notice their dense lumen, which corresponds to degraded cytoplasmic material and lysosomal contents transferred during fusion of lysosomes with the APhLy. The phagosome membranes have been degraded as well as most of the inner membrane of the APhLy, of which remnants can still be observed (arrow). Bar = 0.5 µm.

Figure 7.

Kinetics of sequestration of M. avium into autophagic phagolysosomes in response to cholesterol depletion. At day 6 p.i., macrophages were subjected to cholesterol depletion by the presence of 5 mM methyl β-cyclodextrin. At the indicated times, cells were fixed for electron microscopy. The fraction of M. avium in autophagic phagolysosomes (APhLy) was scored based on the typically observed morphology for these organelles (see Fig. 6). No APhLy were observed under normal conditions or during the first 30 min of cholesterol depletion. The data were compatible with a curve that indicated the ultimate total sequestration of mycobacteria in APhLy. Different symbols refer to three independent experiments. Error bars indicate the SEM from independent sections (n = 3) within a single experiment.

Figure 8.

Cholesterol depletion results in sequestration of M. avium into auto-phagolysosomes (APhLy) that do not derive their limiting membrane from the endoplasmic reticulum. M. avium-infected cells, treated with 5 mM CD for 2 h (A) or for 4 h (B and C) were stained for the endoplasmic reticulum (ER) contents marker, glucose 6- phosphatase (B and C) and then processed for conventional electron microscopy.
A. The ER (arrow) is often located in close vicinity to loner and social phagosomes and even partially, but never completely, surrounds these.
B. In most cases, the ER (arrow) is closely apposed to the membrane of the auto-phagolysosome (APhLy), but it is not the limiting membrane of this structure.
C. A very unusual case (less than 1%) where an ER-derived autophagic vacuole (APhV) is formed, as indicated by the presence of glucose 6-phosphatase reaction product between the double membrane (arrow) that fully surrounds the organellar contents. Bar = 0.5 µm.

Phagolysosome-derived autophagy was specific for mycobacterium-containing phagolysosomes because it was not observed for phagolysosomes that contained latex beads, even in M. avium-infected cells. Over the entire 5 h period of cholesterol depletion, the fraction of beads in auto-phagolysosomes remained below 4%. Although not becoming autophagic by themselves, bead-containing phagolysosomes could be seen fusing with mycobacteria-containing auto-phagolysosomes (arrows in Fig. 5E). Furthermore, cholesterol depletion did not cause lysosomes to engage in autophagy in uninfected cells. Instead, lysosomes tended to cluster at the cell periphery (not shown).

Rescue of mycobacteria from autophagic phagolysosomes upon cholesterol replenishment

At all stages during cholesterol depletion, and in spite of delivery of mycobacteria into phagolysosomes which subsequently became autophagic, essentially all the mycobacteria appeared ultrastructurally intact (Figs 6D and 9A). We therefore investigated the fate of these bacteria when cholesterol in the host macrophages was replenished by incubation in the presence of 30% serum. After 4 days, ultrastructurally intact mycobacteria were seen aligned to the membrane of the autophagic phagolysosome (Fig. 9B and C), contrasting with fully digested cytoplasmic contents (Fig. 9B). Concurrently, mycobacterium-containing loner phagosomes, as well as conventional lysosomes, started to reappear (Fig. 9B and C). After 9 days with cholesterol replenishment, a growing population of loner phagosomes with closely apposed membranes was observed (Fig. 9D). During recovery from cholesterol depletion, mycobacteria were as viable as in untreated cells, as indicated by the number as well as by the growth rate of mycobacteria measured via cfu(Fig. 10). These results showed that exposure to hydrolytic intra-lysosomal conditions did not kill mycobacteria and that they could rescue themselves from auto-phagolysosomes to again reside and multiply in non-maturing phagosomes.

Figure 9.

Rescue of mycobacteria from auto-phagolysosomes upon cholesterol replenishment. M. avium-infected macrophages were treated with 5 mM CD for 4 h, washed and reincubated in enriched medium, containing 30% FBS as a source of cholesterol. Cells were re-fed every 3 days thereafter with the same enriched medium.
A. Cells fixed immediately after CD treatment. Most mycobacteria are in large auto-phagolysosomes (APhLys). Lysosomes are scarce as a consequence of their recruitment to APhLys. Bacteria are morphologically intact.
B, C. Reincubation in enriched medium for 4 days. Former APhLys now look like phagolysosomes because the enclosed cytoplasm and the inner membrane have been completely degraded. Mycobacteria, all morphologically intact, become aligned along the phagosome membrane. They induce the formation of protrusions that each contain a single mycobacterium which is partially closely apposed to the phagosome membrane. Two bacteria (arrows) have rescued themselves from the APhLy. Image in (C) is compatible with the separation of the APhLy into an independent phagosome devoid of lysosomal material and an adjacent lysosome (Ly). The membrane of these phagosomes is closely apposed nearly all around the three bacteria.
D. Reincubation for 9 days in enriched medium. Cells now display a large number of loner phagosomes devoid of lysosomal material, thereby suggesting that mycobacteria have rescued themselves into immature phagosomes. Such cells display lysosomes. Bar = 0.5 µm.

Figure 10.

Intra-cellular growth of M. avium upon cholesterol replenishment. Macrophages infected with M. avium, were treated with CD or incubated in serum-free medium only (control), then reincubated in enriched medium as for Fig. 8. At the indicated times, cells were lysed with Triton X-100 and the number of viable mycobacteria was determined as cfu on Middlebrook 7H10 agar plates. Each value represents the mean (± SEM all included in the symbols, n = 3), and the results are shown for two independent experiments. Within variability between experiments, cholesterol depletion did not affect the number of surviving mycobacteria (level of curves) nor their subsequent growth rate (slope of curves): (○) cfu counts in CD-treated cells; (□) cfu counts in control cells.

Discussion

Cholesterol depletion resulted in fusion of mycobacterium-containing phagosomes with lysosomes

Phagocytic uptake of mycobacteria depends on a normal abundance of cholesterol in the plasma membrane. Upon cholesterol depletion, mycobacteria, in contrast to other bacilli, are no longer phagocytosed by macrophages (Gatfield and Pieters, 2000; Peyron et al., 2000). Our present results now indicate that cholesterol is also required for mycobacteria to prevent fusion of phagosomes with lysosomes. Cholesterol is essential for the formation of lipid rafts, which are special membrane domains where lipid-anchored membrane proteins and sphingolipids are enriched (reviewed in Brown and London, 1998; Lai, 2003). Cholesterol-rich membrane domains harbour signalling molecules (reviewed in Manes et al., 2003) and docking molecules for membrane fusion (SNAREs; Chamberlain and Gould, 2002). A lack of cholesterol blocks in vitro homotypic fusion of yeast vacuoles (Kato and Wickner, 2001). The recruitment of lipid rafts to the phagosome membrane is involved in phagosome maturation and can be interfered with by intra-cellular pathogens (Dermine et al., 2001).

The most likely explanation for how cholesterol depletion incapacitates the ability of mycobacteria to inhibit phagosome maturation, will involve its effect on sphingolipids and related kinases. These molecules could represent a target for pathogenic mycobacteria to interfere with phagosome maturation (reviewed in Koul et al., 2004). Several molecular mechanisms for how mycobacteria prevent phagosome maturation, and therewith fusion with lysosomes, have been proposed (reviewed in, e.g. Russell, 2001; Chua et al., 2004). Events of phagosome maturation which are affected by mycobacterial interference are summarized schematically in Fig. 11 and serve to point out possible sites where cholesterol may be involved. Phagocytosis of mycobacterium blocks sphingosine kinase (SK; cf. Fig. 11 point 1). Intact lipid rafts, and therefore sufficient cholesterol, may be required for this block. Rab5 on mycobacterium-containing phagosomes (Via et al., 1997; Clemens et al., 2000) can promote phagosome maturation by a phosphatidyl inositol-3 kinase (PI3K)-independent process (Vieira et al., 2003), possibly circumventing the block exerted by the mycobacterial lipid ManLAM (cf. Fig. 11, point 3). However, mycobacterium introduces a second block by stimulating a protein kinase (p38 MAPK) which keeps Rab5 in its GDP-bound form (Fratti et al., 2003; cf. Fig. 11, point 7). Assuming this block involves signalling events that depend on intact lipid rafts, it might be overcome by cholesterol depletion and result in maturation of mycobacterium-containing phagosomes.

Figure 11.

Schematic representation to summarize events during phagosome maturation which are affected by mycobacteria.
1. Receptor-mediated phagocytic uptake is accompanied by an increase in the cytosolic Ca++-concentration. Sphingosine kinase (SK) phosphorylates the membrane lipid sphingosine, which is a ligand for a G-protein coupled receptor (GPCR) involved in mediating an increase in the cytosolic calcium concentration. Phagocytosis of mycobacterium leads to a block of SK (Malik et al., 2003; Kusner, 2005). SK also appears to associate directly with the phagosome and this process is inhibited by mycobacteria (Thompson et al., 2005).
2. The released Ca++ binds to calmodulin, which then associates with a class III phosphatidyl inositol-3 kinase (PI3K), an essential enzyme for phagosome maturation (Vieira et al., 2001; reviewed in Botelho et al., 2004).
3. Ca++-associated PI3K is essential for phagosome maturation and binds to the phagosome membrane via Rab5. The latter is retained on mycobacterium-containing phagosomes (Via et al., 1997; Clemens et al., 2000). PI3K catalyses the conversion of phosphatidyl inositol (PI) to phosphatidyl inositol 3-phosphate (PI3P). The similarity between PI3P and the mycobacterial cell-wall lipid, ManLAM, is thought to interfere with this process (Fratti et al., 2001).
4. Because PI3P is required to stabilize the association of early endosome antigen 1 (EEA1) with the phagosome membrane (Simonsen et al., 1998), EEA1 is not recruited to the mycobacterium-containing phagosome membrane.
5. EEA1 acts as a tethering molecule which is involved in SNARE-mediated organellar fusion.
6. Initially, EEA1 becomes associated with the phagosome membrane only via Rab5.
7. A PI3K-independent process, by which Rab5 can lead to phagosome maturation (Vieira et al., 2003), is also blocked by mycobacteria in that they stimulate a protein kinase (p38 MAPK) which leads to the inactivation of Rab5, keeping it in its GDP-bound form (Fratti et al., 2003).

When mycobacteria are phagocytosed in clumps and the resultant phagosome contains more than one mycobacterium, these ‘social’ phagosomes invariably mature and fuse with lysosomes, whereas ‘loner’ phagosomes in the same cells are prevented from maturing (de Chastellier et al., 1995; Clemens and Horwitz, 1995; de Chastellier and Thilo, 1997; Pietersen et al., 2004). Therefore, the inhibition of phagosome maturation does not involve cell signalling in general, but is phagosome-autonomous. This local restriction is difficult to reconcile with the proposed role of mycobacterial lipids in the prevention of phagosome maturation (Fig. 11, point 3) as such lipids have been observed to diffuse to other endocytic membranes and even to uninfected cells in proximity of infected ones (Beatty et al., 2000). Evidence that cholesterol depletion interfered with a biological process, rather than simply changing the physical properties of the phagosome membrane, was obtained by our observation that the effects of cholesterol depletion remained restricted to mycobacterium-containing phagosomes and did not affect phagosomes with inert hydrophobic latex beads when these were also present in mycobacterium-infected macrophages. Such beads were chosen here as control phagocytic particles because, in contrast to beads with a hydrophilic surface, they also block phagosome maturation, although temporarily (up to at least 3 h, de Chastellier and Thilo, 1997; see also solid squares in Fig. 4A and below). Within 12–24 h, such beads end up in phagolysosomes, after repeated events of fusion and fission with lysosomes, referred to as the ‘kiss-and-run’ mechanism (Desjardins, 1995). We observed evidence that cholesterol depletion might interfere with the fission aspect of this process: the bead-containing phagosomes remained connected to lysosomes through thin tube-like connections (Fig. 5D and E).

Previously, we have proposed that a necessary requirement for mycobacterium's ability to prevent phagosome maturation is an all-around close apposition of the phagosome membrane to the bacterial surface (de Chastellier et al., 1995; de Chastellier and Thilo, 1997). This proposal can serve to explain the maturation of social phagosomes because the phagosome membrane is distant from the bacterial surface where it spans regions between adjacent mycobacteria. In this regard, it was of special interest to note that one of the first effects of cholesterol depletion on mycobacterium-containing loner phagosomes was a partial loosening of the phagosome membrane from the mycobacterial surface, seen as an undulating limiting membrane (Fig. 2B). This was rapidly followed by fusion with lysosomes (Figs 2C, D and 3). Surprisingly, cholesterol depletion had no immediate effect on the morphology of phagosomes with hydrophobic latex beads (see Fig. 5), which started to fuse with lysosomes only about 2 h later (Fig. 4, solid circles). Cholesterol depletion might affect specific interactions of the mycobacterial surface with proteins of the phagosome membrane, conceivably by interfering with their lateral distribution in the absence of lipid rafts. How can an all-around close apposition of the phagosome membrane interfere with its maturation? We have argued that maturation requires efficient recycling from the phagosome, that this depends on the ability of the membrane to undergo tubule formation, and would therefore be severely restricted as long as the membrane remains closely apposed to the surface of the phagocytic particle (de Chastellier and Thilo, 1997). This is directly supported by comparing the rate of recycling of the transferrin receptor, which occurs with a half-life of about 1.5 min from the tubular extensions of early endosomes (Stoorvogel et al., 1989), but slowly from mycobacterium-containing phagosomes in that they lose transferrin receptor over a period of several hours (Clemens and Horwitz, 1996). Tubule formation of the phagosome is also involved in its interaction with late endocytic organelles (Harrison et al., 2003). The v-SNARE protein, cellubrevin, which is required for recycling from early endosomes, is degraded on the mycobacterium-containing phagosome (Fratti et al., 2002), a process that might require close proximity with the mycobacterial surface. The loss of cellubrevin may account for the selective local interference with efficient recycling and, therewith, phagosome maturation. In contrast, loss of cellubrevin will not prevent ongoing fusion with early endosomes as this does not depend on cellubrevin (Link et al., 1993). Cellubrevin is a v-SNARE and may be localized to cholesterol-rich microdomains known for their association with SNAREs (Lang, 2003). Cholesterol depletion could undo this localization in a way that interfered with degradation of cellubrevin by mycobacteria.

Cholesterol depletion induced phagolysosome-derived autophagy

The present study showed that when cholesterol depletion caused fusion of mycobacterium-containing phagosomes with lysosomes, the resulting phagolysosomes engaged in autophagy. These curled themselves around cytoplasmic contents, occasionally including other phagosomes, which became engulfed into an organelle of phagolysosomal origin (Fig. 6). To the best of our knowledge, this type of autophagy has not been described before and may be specific for mycobacterium-containing phagolysosomes because it did not occur for phagolysosomes with latex beads in the same infected cells. Intracellular pathogens have been linked to autophagy on numerous occasions, either when autophagy serves as a host response for bacterial destruction, or as a compartment for intra-cellular survival of the pathogens (reviewed by Dorn et al., 2002; Kirkegaard et al., 2004; Gorvel and de Chastellier, 2005; Levine, 2005). In mammalian cells, autophagy can be induced by starvation and the membrane of the autophagic vacuoles can be derived from the smooth endoplasmic reticulum (Dunn, 1990), or from other unidentified sources (Stromhaug et al., 1998; Mizushima et al., 2001). In these previous studies, the (phago-) lysosomes are not the primary source of membrane for the nascent autophagic vacuole, which only acquire lysosomal characteristics upon fusion with lysosomes to become autolysosomes (reviewed in Mizushima et al., 2002). Coxiella-containing phagolysosomes exhibit molecular markers normally associated with auto-phagosomes (reviewed in Kirkegaard et al., 2004), but autophagy itself has not been observed for such organelles. A detailed morphological study of the response of Schwann cells to cholesterol depletion as a result of exposure to tellurium has not revealed lysosome-based autophagy (Calle et al., 1999).

Autophagy by mycobacterium-containing phagolysosomes in response to cholesterol depletion could not be induced by starvation and is difficult to explain in molecular terms. Macro-autophagy in mammalian cells is controlled by diverse signalling mechanisms (reviewed by Petiot et al., 2002) and is associated with cholesterol-rich membrane domains (reviewed in Manes et al., 2003). Conceivably, cholesterol depletion could interfere with this control network. Class III PI3K is required for macro-autophagy (Petiot et al., 2002), but this enzyme may be affected by the mycobacterial lipid ManLAM (cf. Fig. 11, point 3). Therefore, either this requirement is bypassed by cholesterol depletion, or, more likely, mycobacteria-derived autophagy upon cholesterol depletion depends on other factors than those involved in macro-autophagy in general.

Upon cholesterol replenishment, mycobacterium remodels auto-phagolysosomes into immature phagosomes

It is not known to what extent delivery of mycobacterium into a phagolysosome leads to its ultimate destruction or merely inhibits its proliferation. Considering the tendency of M. tuberculosis for clumping, it is likely that most of these bacteria will be phagocytosed into a social phagosome that fuses with lysosomes (see above). For a successful infection, it must be assumed, however, that mycobacteria are not being killed when in a phagolysosome. Accordingly, we have proposed that pathogenic mycobacteria can ‘rescue’ themselves from phagolysosomes and revert into phagosomes which remain immature, which cannot fuse with lysosomes, and in which mycobacteria can multiply (de Chastellier and Thilo, 1999; Pietersen et al., 2004). The present observations now showed that when cholesterol was replenished, mycobacteria were able to rescue themselves from auto-phagolysosomes and revert to residing individually in phagosomes with the phagosome membrane in an all-around close apposition to the mycobacterial surface (Fig. 9). Furthermore, residence in the harmful environment of auto-phagolysosomes did not affect the level or the subsequent intra-phagosomal growth rate of mycobacteria (Fig. 10). We have shown previously that the phagosome membrane which surrounds mycobacteria acquires a special composition in terms of cell surface-derived glycoconjugates and proposed that this may reflect the involvement of some of these membrane proteins in establishing the close interaction with the mycobacterial surface (de Chastellier and Thilo, 2002; Pietersen et al., 2004). Considering that the bulk of internalized cell-surface glycoconjugates is recycled to the plasma membrane at a prelysosomal stage (Thilo et al., 1995) and that only a few percent become part of the lysosomal membrane (Haylett and Thilo, 1986), the present observation of a slow reestablishment of a closely apposed membrane implies that mycobacteria should actively recruit such molecules for binding to their surface, possibly in a zippering-like process.

Experimental procedures

Mycobacterium avium

The transparent (Tr) colony variant of M. avium TMC724 (serovar 2), from the Trudeau Mycobacterial culture collection, was cultured as previously described (Fréhel et al., 1997). This strain is virulent for C57Bl/6 mice (Fréhel et al., 1991) and grows within macrophages from the same, or other, mice susceptible to mycobacterial infections (Fréhel et al., 1997). Samples, kept frozen at −80°C, were quickly thawed, vortexed and adjusted to the desired titre in complete cell culture medium. More than 95% of the bacteria were morphologically intact and viable.

BMDM culture and phagocytic uptake

Bone marrow cells were isolated from femurs of 6- to 8-week-old C57Bl/6 female mice and differentiated into macrophages as described previously (de Chastellier et al., 1995). Particles were added to 7-day-old macrophage cultures as follows: (i) cells were incubated for 4 h at 37°C at a mycobacterium/macrophage ratio of 10:1, washed in four changes of ice-cold phosphate-buffered saline (PBS) to eliminate non-ingested bacteria, and further incubated in complete medium devoid of antibiotics; (ii) cells were given native hydrophobic 1 µm-diameter latex beads for 30 or 45 min. The latex bead solution was diluted 1000-fold in complete medium to obtain adequate particle uptake. Cells were then washed with PBS as above and reincubated in medium devoid of antibiotics; (iii) in some cases, macrophages were first infected with M. avium and, 6 days later, given hydrophobic latex beads in the same conditions as above or they were simultaneously given M. avium and hydrophobic beads. In the latter case, the beads were added during the last 30 min of infection with M. avium.

Cholesterol depletion and replenishment

At day 6 post infection with M. avium, cells were treated with methyl-β-cyclodextrin (CD) to deplete cholesterol. Cells were first washed free of fetal bovine serum (FBS) and l-cell conditioned medium with warm DMEM containing 2 mM l-glutamine and incubated therein for 0–6 h with 0.5–10 mM CD prepared in the same medium, in order to find conditions that would sufficiently deplete for cholesterol without inducing irreversible damage to cells over the 6 h incubation period. A treatment with 5 mM CD for 0–6 h was adopted for the following experiments. Cell-associated cholesterol was measured (4 × 106 cells) by using a kit from KAT Medical (Gauteng, RSA), based on the method of Allain et al. (1974). After a 4 h treatment with 5 mM CD, M. avium-infected cells were washed four times with DMEM and reincubated in enriched medium containing 30% FBS and 15%l-cell conditioned medium.

Filipin labelling and visualization

Filipin (10 mg ml−1 in dimethyl sulphoxide) was stored at −20°C. For ultrastructural localization of cholesterol with filipin, samples were first briefly fixed with 2.5% glutaraldehyde in 0.1  M cacodylate buffer (pH 7.2). Filipin (0.05 mg ml−1) was added to the fixative solution and cells were fixed for an additional 60 min at room temperature (Coppens and Joiner, 2003). Samples were then processed for conventional electron microscopy.

Evaluation of growth and structural integrity of intra-cellular bacteria after cholesterol replenishment

To monitor intra-cellular growth of M. avium, cfus were counted. Cells from three different wells were lysed at selected time points with 0.1% Triton X-100 in distilled water after PBS washing. Serial dilutions of lysates and of culture medium were plated onto Middlebrook 7H10 agar plates and cfus were counted (Fréhel et al., 1997). As a control, the number of viable bacteria was determined in cells that had not been depleted for cholesterol but had otherwise been infected with M. avium and subsequently incubated in the same media as the CD-treated cells.

For a morphological assessment of structural integrity, M. avium-infected macrophages, depleted or not for cholesterol, and further incubated in enriched medium, were processed for electron microscopy as described below. On thin sections, intra-cellular bacteria can be sorted into three distinct categories on the basis of their morphological appearance (Fréhel et al., 1997): (i) intact bacteria that display a regular shape, a well organized cytoplasm and nucleus, an intact cell wall and cytoplasmic membrane, and especially a typical thick electron translucent layer beyond the dense thin layer of the wall; (ii) altered bacteria that are distorted and display several electron translucent intra-cellular lipid inclusions in the cytoplasm but otherwise have retained normal morphological features; (iii) degraded bacteria that characteristically display breaks in the cytoplasmic membrane and/or cell wall, a disorganized cytoplasm and especially no electron translucent layer.

Acquisition of the endocytic content marker HRP by phagosomes

Six days after infection with M. avium, cells were treated for 2 h with 5 mM CD. HRP, at a concentration of 25 µg ml−1, was added to the medium during the last 30 min of treatment. As a control, untreated cells were exposed to HRP for 30 min. Cells were then processed for HRP cytochemistry and electron microscopy (de Chastellier et al., 1995).

Acquisition of lysosomal marker by phagosomes

Six days after infection with M. avium, lysosomes were labelled with 10 nm-diameter colloidal gold particles conjugated with bovine serum albumin (BSA-Au) (de Chastellier and Thilo, 1997). Cells were then treated with 5 mM CD for 0–5 h as described above.

G6Pase cytochemistry

Staining for G6Pase, a marker enzyme in the lumen of the endoplasmic reticulum, was performed by EM cytochemistry as described by Griffiths et al. (1983) with the following modifications. Briefly, infected cells, treated with CD, were prefixed for 30 min on ice with 1.25% glutaraldehyde in 0.1  M Pipes (pH 7.0, containing 5% sucrose), washed with 0.1 M Pipes (pH 7.0, containing 10% sucrose), then with 0.08 M Tris-maleate buffer (pH 6.5). Cells were incubated for 2 h at 37°C in the cytochemical reaction medium consisting of 0.03 M glucose 6-phosphate (Sigma-Aldrich) and 0.1% lead nitrate in Tris-maleate buffer. After washes in Tris-maleate buffer and in 0.1 M cacodylate buffer (pH 7.2, containing 0.1 M sucrose, 5 mM CaCl2 and 5 mM MgCl2), cells were postfixed for 1 h at 4°C with 1.25% glutaraldehyde in the same buffer and processed for electron microscopy as described below.

Acid phosphatase cytochemistry

At selected time points during treatment of infected cells with 5 mM CD, cells were fixed and processed for acid phosphatase (AcPase) cytochemistry and electron microscopy (Fréhel et al., 1997).

DAMP treatment and immuno-localization

DAMP, synthesized at the Pasteur Institute, was stored at −20°C in ethanol at a concentration of 60 mM. At selected time points during treatment of infected cells with 5 mM CD, DAMP (60 µ M) was added to the medium and cells were incubated at 37°C for an additional 30 min. After three rinses in DMEM, cells were fixed for 1 h at 4°C with 1.25% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2) containing 0.1 M sucrose. Fixed cells were washed with 0.1 M cacodylate buffer containing 0.05 M NH4Cl to quench any remaining aldehyde activity and then overnight at 4°C with buffer devoid of NH4Cl. After three brief rinses in cacodylate buffer devoid of sucrose, the cells were scraped off the culture dishes, concentrated in 2% agar in cacodylate buffer and postfixed for 30 min at 4°C with 0.5% uranyl acetate in Veronal buffer. The samples were subsequently dehydrated in ethanol and embedded in LRWhite resin for electron microscope immunocytochemistry (Fréhel et al., 2002).

Lowicryl thin sections were preincubated for 10 min with PBS containing 10% FBS and then sequentially incubated for 60 min at room temperature with rabbit antidinitrophenol (anti-DNP) antibodies and for 30 min with protein A coupled to 10 nm-diameter gold particles (Department of Cell Biology, Utrecht, School of Medicine, University of Utrecht, the Netherlands). Antibodies and conjugate were diluted in PBS containing 5% FBS. Sections were washed five times rapidly with PBS-FBS 0.5% between the incubations and after treatment with conjugate. After three washes with PBS and distilled water, sections were stained with 2% uranyl acetate and lead citrate. As a control, thin sections were incubated with GAR10 only. The control was negative.

The number of gold particles per phagosome was determined on 40–60 different phagosomes per time point. The number of gold particles in the late endocytic compartments was also determined on the same micrographs. The surface area of the phagosome and late endocytic compartments was determined morphometrically. The density of gold particles was then determined (de Chastellier et al., 1995). These data could be used to obtain estimates for intra-phagosomal pH by using the method described by Orci et al. (1994).

Processing for conventional electron microscopy

Unless otherwise indicated, cells were fixed for 1 h at room temperature with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2, containing 0.1 M sucrose, 5 mM CaCl2 and 5 mM MgCl2. Cells were then washed twice with the same buffer and postfixed for 1 h at room temperature with 1% osmium tetroxide in the same buffer devoid of sucrose. Cells were then scraped off the dishes and concentrated in 2% agar in the same buffer and treated for 1 h at room temperature with 1% uranyl acetate in Veronal buffer. Samples were dehydrated in a graded series of acetone (or ethanol when they contained latex beads), and embedded in Epon. Thin sections were stained with 2% uranyl acetate in distilled water and then with lead citrate.

Quantifications

In all cases, 100–200 phagosomes per sample were examined to determine (i) whether phagosomes contained only one (loner phagosomes) or more than one bacterium (social phagosomes); (ii) whether bacteria were structurally intact or not; and (iii) whether they contained either lysosomal markers (i.e. unstained lysosomal contents, BSA-Au or acid phosphatase) or ER content marker (G6Pase). Care was taken to avoid serial sections.

Acknowledgements

Cholesterol was measured by Ms Bharati Ratanjee in the laboratory of L.T. The authors wish to express their appreciation to Dr Pierre Courtoy (UCL, Brussels, Belgium) for many suggestions to improve this work. Authors are also grateful to Dr Isabelle Coppens (Yale, New Haven, USA) for helpful advice concerning the visualization of cholesterol, Ms Christiane Waldmann (CIML, Marseilles, France) for expert assistance with the artwork, M. Jean Paul Chauvin (Electron Microscopy Unit of the Institut de Biologie du Développement, Marseilles, France) for expert technical assistance in the EM facility, Ms Janice Griffith (University of Utrecht, the Netherlands) for the kind gift of BSA-Au and Dr Jean Claude Antoine (Institut Pasteur, Paris, France) for providing the antidinitrophenol antibodies. The research was supported by a joint France/South Africa Science and Technology grant between the National Research Foundation (South Africa) and the Ministères de l’Education, de la Recherche et des Affaires Etrangères (France).

Ancillary