Parasitophorous vacuoles of Leishmania mexicana acquire macromolecules from the host cell cytosol via two independent routes

The trypanosomatid parasite Leishmania mexicana is able to survive intracellularly in macrophages (MO) of the vertebrate host despite the antimicrobial potential of these cells. The intracellular compartment harbouring the parasites, i.e. the parasitophorous vacuole (PV), is modulated to support growth and sustained survival of the parasite within its macrophage host (Alexander and Russell, 1992). During the extended course of the infection, the PV increases in size tremendously at the expense of the host cell. Recent studies demonstrated that L. mexicana PVs remain accessible to several endocytic tracers internalized by both fluid-phase and receptor-mediated entry (Shepherd et al., 1983; Rabinovitch et al., 1985; Russell et al., 1992). Data indicating that PVs containing L. mexicana, or its close relative L. amazonensis, may be described as late endosomal/lysosomal compartments has been generated through immunofluorescence and immunoelectron microscopical studies demonstrating the presence of lysosomal constituents cathepsins B, H, L and D, rab 7, macrosialin, vacuolar proton-ATPase, cation-independent mannose 6-phosphate-receptor (M6PR), lysosome-associated-proteins 1 and 2 (LAMP-1 and 2) as well as MHC class II molecules (Antoine et al., 1987, 1991; Prina et al., 1990; Russell et al., 1992; Lang et al., 1994; Sturgill-Koszycki et al., 1994).

Despite the apparent threat afforded by the low pH of the lysosome, the intracellular amastigote form of L. mexicana has evolved to survive and exploit this milieu (Antoine et al., 1990). One postulate is that the activities of both parasite and host hydrolases provide nutrients for parasite growth. We have already demonstrated that parasites endocytose macromolecular materials from the PV lumen (Russell et al., 1992). If the parasite were able to facilitate transfer from the host cell cytosol into the PV then the source of nutrients could be augmented. Such a stratagem may be critical since Leishmania is incapable of purine synthesis (Hansen et al., 1984).

In this current study we evaluate the ability of L. mexicana PVs to access cytosolic material. The data presented demonstrates that PVs acquire material from the host cell cytoplasm by two unrelated mechanisms. The vacuoles accumulate Lucifer Yellow from the cytoplasm through the activity of an organic anion transporter, the activity of which can be shown in isolated vacuoles. In addition, PVs appear to acquire cytosolic macromolecules via a process that can be modulated by compounds that affect autophagy. We have developed two novel assays to visualize and quantify the transfer mechanism in infected cells. Immunoelectron microscopical analysis of the process supports the view that material is entrapped in nascent autophagosomes, which subsequently fuse with PVs. Access to host cell cytosol via the autophagous pathway affords two possible benefits to the parasite: (1) access to additional nutritional sources for the amastigotes and (2) because the parasites live within macrophages which are capable of presenting antigens in context of class II MHC molecules, the increased concentration of digested ‘self proteins’ in the PVs may reduce the ability of the class II molecules to sample and present parasite-derived peptides. This would favor development of an autoimmune, rather than a parasite-specific, cellular immune response.

Bone marrow cells were prepared as described elsewhere (Collins et al., 1997). In short, cells were eluted from femur and tibia of female BALB/c mice and plated in bacterial Petri dishes in DMEM containing 10% FCS, 5% horse serum and 20% supernatant from L929 cells as a source of colony stimulating factors. Cells were cultivated for 9 days until a monolayer was formed and subsequently plated for experimental procedures in tissue culture dishes/flasks or on glass coverslips.

Leishmania mexicana (MYNC/BZ/62/M379) were cultured in SDM79 medium at 25°C. For differentiation of L. mexicana promastigotes into amastigote-like forms (ALF), a growing culture of promastigotes was diluted 1:4 in UM54 medium and cultivated for 4 days at 35°C (Bates et al., 1992). ALFs were added to MO in a 5:1 ratio for 1 hour and non-phagocytosed parasites were removed by intensive washing, resulting in an infection rate of around 80%. Histoplasma capsulatum yeast (kindly provided by Dr L. Eissenberg, Washington University, School of Medicine, St Louis; Eissenberg et al., 1988) were added to MO at a 5:1 ratio. Mycobacterium avium strain 101 were grown in Middlebrook medium from a sample isolated from spleens of infected mice as described previously (SturgillKoszycki et al., 1994) and Listeria monocytogenes DLly, a listerolysin-negative mutant unable to escape from the phagosome (Portnoy et al., 1988) were grown in Brain-Heart-Infusion. Both bacteria were added to MO at a ratio of 10:1.

Lucifer yellow (LY; lithium salt; anionic), aminodextran, dextrans (3, 10, 40, 70 kDa) either labeled with Texas Red, FITC, biotin (all three compound were aldehyde-fixable, anionic) or rhodamine (neutral). (CBz-Phe-Arg)₂-rhodamine (ZFR-Rho) and (Phe-Arg)₂-rhodamine (FR-Rho) (Leytus et al., 1983a,b) were either synthesized by us (W. F. M.) or obtained from Molecular Probes (Eugene, Oregon). Affinity-purified polyclonal rabbit antibodies to rat calnexin and to bovine M6PR were generous gifts from Drs A. Helenius (Dept of Cell Biology, Yale University, New Haven) and S. Kornfeld (Washington University, St Louis, School of Medicine), respectively. Monoclonal antibodies (mAb) to canine protein disulphide isomerase (PDI) and murine LAMP-1 (ID4B) were generously supplied from Drs David Vaux (Dept of Pathology, Oxford University, GB) and T. August (Johns-Hopkins Medical School, Baltimore); antibody against lysobisphosphatidic acid (6C4) is detailed in a recent publication by Kobayashi et al. (1998). Digoxigenin-dextran (3 kDa) and biotin-dextran (3 kDa) were prepared by reacting n-hydroxysuccinimide digoxigenin or n-hydroxysuccinimide biotin with aminodextran (3 kDa) (Molecular Probes, Eugene, Oregon). Gold-labeled secondary antibodies were purchased from Jackson Immunoresearch Laboratories.

The cytosolic loading techniques employed in here were modified from protocols published previously by other researchers (McNeil, 1989).

Cells were incubated in the presence of 1 mg/ml Lucifer Yellow (LY) and 5 mM sodium ATP in DMEM for 5 minutes at 37°C and washed three times. This technique led to around 90-100% loaded cells. Control cells without ATP did not show significant uptake of LY under these conditions.

10 mg/ml dextran in DMEM containing 2% Pluronic F-68 (Sigma) was chilled to 4°C, and 100 µl of the solution was overlaid onto the MO coverslip. Acid-washed glass beads (400 µm diameter; Sigma) were layered onto the MO monolayers on coverslips or in Petri dishes and gently rocked ten times to roll the beads across the cells. Cells were washed five times in warm DMEM and placed in culture for the times stipulated. To remove ‘regurgitated’ dextran, the medium was changed twice during the first hour after loading. Depending on the size of the dextran loaded, an efficiency of 10-20% (for 70 kDa dextran) to 40-70% (for 3 kDa dextran) was achieved.

Cells were resuspended in DMEM with 10 mg/ml digoxigenindextran (3 kDa) or 10 mg/ml biotin-dextran (3 kDa) and 1% Pluronic F-68 (Sigma) at 4°C and passed 5 times through a 25-gauge tuberculin syringe. The cells were pelleted and washed 3 times in complete DMEM with 10% FCS at 37°C and plated.

As described previously (Chakraborty et al., 1994), MO infected with L. mexicana ALFs for 1 hour were washed in cold lysis buffer (20 mM Hepes, pH 6.5, 0.5 mM EDTA, 8.55% sucrose, 0.1% gelatin, with the protease inhibitors E64, leupeptin, pepstatin A and TLCK), scraped using a cell scraper and lysed on ice by passage through a syringe (25-gauge needle) 8-15 times. The degree of lysis was monitored by microscopy. A postnuclear supernatant (PNS) was generated by centrifuging the lysate for 10 minutes at 200 g at 4°C. The PNS was loaded onto a step gradient of 20%, 40% and 60% sucrose in gradient buffer (30 mM Hepes, pH 6.5, 100 mM KCl, 0.5 mM CaCl₂, 0.5 mM MgCl₂) and spun at 700 g for 25 minutes at 4°C. The interface between 40% and 60% sucrose was harvested and PVs were pelleted out of the sucrose at 12000 g for 25 minutes in the cold and resuspended in the relevant buffer. Phagosomes containing M. avium were isolated as described previously in detail (SturgillKoszycki et al., 1996)

PVs were resuspended in cold KCl buffer (10 mM Hepes, pH 7.6, 150 mM KCl, 1 mM MgCl₂, 1 mM EGTA + protease inhibitors) containing 0.5 mg/ml LY (Lipman et al., 1990). Tubes were incubated at 37°C for various time periods, then placed on ice. The PV preparations were then underlayered with 20% and 40% sucrose cushions using a gel loading tip and spun at 12000 g for 25 minutes at 4°C to separate the PVs from free dye. The tubes were frozen, the tips (2 mm) cut off and the contents diluted to 500 µl in KCl buffer containing 0.5% Triton X-100. Similarly, isolated M. aviumcontaining phagosomes were incubated in LY, collected on ice and separated from free dye by centrifugation through a 10% Ficoll (70 kDa) cushion. In both instances, LY content was measured fluorometrically (450 nm excitation, 540 nm emission).

Cells were fixed in fixation buffer (200 mM Pipes, 0.5 mM MgCl₂, pH 7.0) containing 0.5% glutaraldehyde and 4% formaldehyde for 1 hour and processed for immunoelectron microscopy as described (Russell et al., 1992). Immunoelectron microscopy was performed on cells syringe-loaded with digoxigenin-dextran or biotin-dextran as detailed above.

In initial experiments the unblocked substrate FR-Rho was coupled to rabbit serum albumin (RSA) following activation with 5 mg carbodiimide (0.1 M for 3 hours at room temperature.) After overnight incubation at 4°C, the solution was dialysed against water and lyophilized. However, background from these compounds was high. Later, it was found that the blocked substrate ZFR-Rho readily formed a stable complex with RSA without the formation of a covalent bond. Experiments were conducted on macrophages either bead-loaded (for microscopy) or syringe-loaded (for quantiation of hydrolysis). To measure quantitatively the transfer of cysteine proteinase substrate from the cytosol into hydrolytic compartments we syringe-loaded uninfected and infected MO with RSA-ZFR-Rho in the presence of leupeptin (50 µg/ml) in DMEM + Pluronic F-68 as detailed above. MO were plated in 24-well plates and washed extensively with warm DMEM with 10% FCS and 10 µg/ml leupeptin, prior to the addition of DMEM (without Phenol Red) and 10% FCS. After incubation for 30, 150 and 600 minutes the medium was removed and placed on ice prior to measurement of released fluorochrome. 1 ml of ice-cold PBS + E-64 (200 µg/ml) and 1% Triton X-100 was added to the cells to release cell-associated label for measurement. Fluorescence was measured using an excitation wavelenght of 490 nm and an emission wavelength of 520 nm. The fluorescence value from each time point is the average from three separate wells. Calculation of the total substrate loaded into the cells was achieved through the addition to the medium of Triton X-100 to 1%, followed by incubation at 37°C for 15 minutes to facilitate complete hydrolysis of substrate prior to removal and measurement. Experimental values were compared to a standard curve relating fluoresence values to concentration of free Rho 110. To determine whether free Rho 110 was retained in vesicles, macrophages were incubated in 1 mM dye for 30 minutes, washed and chased for 30, 60 and 90 minutes before being examined by fluorescence microscopy. At all time points, the free dye appeared to be retained within vesicles.

Lucifer yellow (LY) is known to be sequestered from the cell cytosol into endosomal compartments and subsequently exported from the cell. The sequestration of LY has been attributed to an organic anion transporter activity located in the vacuolar membrane (Steinberg et al., 1987, 1988). To demonstrate this activity in the membranes of PVs of L. mexicana, LY was loaded into the cytoplasm of MO, infected 2 days previously with L. mexicana. 10 minutes after loading the dye was observed in cytoplasm and nucleus of the infected MO but was excluded from the PV (Fig. 1A,B). 30-120 minutes after loading the dye was observed in PVs, and in small cytoplasmic vesicles in virtually all of the loaded cells (Fig. 1C,D). Similar results were observed using the MO-like cell line J774 as a host cell (data not shown).

It had been reported that the organic anion transporter system responsible for LY sequestration from the cytosol into small vesicles was sensitive to inhibitors such as gemfibrocil and probenecid (Steinberg et al., 1987; Rudin et al., 1992). LY was loaded into the cytosol of infected MO, which were incubated in the presence and absence of different concentrations of inhibitors for 2 hours at 37°C. As shown in Fig. 1E,F, 0.5 mM gemfibrocil blocked detectable transfer of LY into the PVs of all infected cells. Similar effects were achieved using 5 mM probenecid (data not shown). Neither inhibitor appeared to compromise cell viability or morphology at the concentrations used (data not shown).

In addition to the L. mexicana PV, we also studied vacuoles containing other particles or pathogens such as the fungal pathogen H. capsulatum, bacterial pathogens M. avium and L. monocytogenes-DLly, or IgG-coated latex beads for their capacity to transport LY. As summarized in Table 1, transfer of LY was seen in many but not all vacuoles containing latex beads, L. monocytogenes-DLly or H. capsulatum, although in each instance to a lesser extent than L. mexicana PVs. In contrast, LY was never observed in vacuoles containing M. avium.

The above data demonstate that L. mexicana PVs acquire LY; however, the route of acquisition is unclear because initial experiments could not differentiate between direct transfer of LY into PVs versus fusion with small LY-containing vesicles in the cytoplasm. It has been reported that organic anion transport can be monitored in vitro using isolated endosomes from J774 cells (Lipman et al., 1990). To test if organic anion transporter activity was present in the membrane of L. mexicana PVs, vacuoles were isolated from infected MO 1 hour post-infection (p.i.) and incubated with 0.5 mg/ml LY in KCL buffer in the absence or presence of 5 mM gemfibrocil for 0, 10, 30 and 60 minutes at 37°C. Following passage through a sucrose cushion, the PVs were assayed for LY. Isolated PVs were found to accumulate LY at the same rate over the time period studied and, in common with the whole-cell studies, uptake was abrogated totally by gemfibrocil. The transport of LY was observed in isolated L. mexicana PVs but not in preparations of parasites alone (data not shown), demonstrating that sequestration of LY was a function of the PV membrane and not the parasite. Interestingly, isolated M. avium phagosomes were unable to transfer LY in vitro (Fig. 2).

In initial experiments on LY transfer we coloaded Texas Red-dextran into the cytosol to control for artefactual loading of vacuoles during dye uptake. Much to our surprise, we also observed transfer of dextrans into L. mexicana PVs. However, the transport of dextrans was markedly slower than the rate of LY accumulation. To investigate this phenomenon the cytoplasm of L. mexicana infected MO were bead-loaded with Texas Red-dextrans of various sizes and observed at subsequent time points. 4 hours post-loading, all dextrans remained predominantly cytosolic (Fig. 3A,B). However, following 10 hours incubation we observed significant transfer of dextran into the PV and smaller cytoplasmic vesicles (Fig. 3C,D). Dextrans of 3, 10, 40 and 70 kDa as well as BSA and casein (labeled with either Texas Red or carboxyfluorescein) were all transferred into PVs 10-24 hours after loading into the cytoplasm (Fig. 3, and data not shown). Although the size of the dextran did not appear to influence the efficiency of transfer into PVs, the efficiency of loading into the MOs was inversely proportional to dextran size. Transfer of dextrans was not observed in phagosomes containing L. monocytogenes-DLly, M. avium, H. capsulatum and only as a rare event in those containing latex beads (Table 1).

Transfer of this heterogeneous group of macromolecules into L. mexicana PVs is unlikely to be based on active transport mechanisms through membrane-associated transporters. Furthermore, 10 hours following loading of dextran into the cytoplasm of uninfected macrophages the tracer could be observed in small cytoplasmic vesicles, indicating that the process itself was independent of infection. The transfer of dextran from the cytoplasm into lysosomal compartments, such as the PVs of L. mexicana, is reminiscent of previous studies on the sequestration of cytoplasmic constitutents into autophagous vacuoles (Rabouille et al., 1993). An early study by Stacey and Allfrey (1977) suggested that autophagy was involved in the removal of fluorescent proteins microinjected into HeLa cells. Autophagy accounts for the turnover of cytoplasmic organelles and constituents, and has been described in hepatocytes and fibroblasts (Seglen and Bohley, 1991; Dunn, 1994).

Although no specific inhibitors exist, it has been shown that adenine analogs and amino acids, i.e. asparagine and leucine, downregulate formation of early autophagosomes or their subsequent coalescence into autolysosomes (Dunn et al., 1990a,b; Høyvik et al., 1991; Seglen and Bohley, 1992; Punnonen et al., 1994; Fengsrud et al., 1995). 10 kDa Texas Red-dextran was efficiently ‘scavenged’ from the host cell cytoplasm and, for the most part, delivered into PVs 24 hours after loading (Fig. 3C,D). This process was unaffected by asparagine up to 30 mM concentration (Fig. 3G,H). In contrast, incubation of the dextran-loaded cells in the presence of 10 mM 3-methyladenine almost totally blocked transfer of dextran into PVs (Fig. 3E,F). To quantify this effect we scored PVs with Texas Red-dextran, relative to the total number of PVs, both in the absence or presence of the inhibitors adenine, 3methyladenine or asparagine. As displayed in Fig. 4, adenine and 3-methyladenine decrease the number of dextran-positive PVs in a dose-dependent manner, whereas asparagine had no discernible effect on dextran transfer.

To achieve a detailed ultrastructural analysis of this transfer mechanism we used digoxigenin-labeled or biotin-labelled dextran (dig-dextran/biot-dextran) (3 kDa) for immunoelectron microscopical detection of the sequestration process. 90 minutes after loading, labelled dextran was localized predominantly in the cytosol (Fig. 5A), with minimal label in either cytoplasmic vesicles or PVs. After 4 hours incubation label was observed concentrated in regions rich for LAMP 1-positive membranes (Fig. 5B), and also in vesicles containing membranous debris, reminiscent of the multivesicular structures described in previous structural studies on autophagosome formation (Fig. 5C) (Dunn, 1990a,b; Lawrence and Brown, 1992; Punnonen et al., 1993; Rabouille et al., 1993). The multivesicular structures are LAMP 1-positive (Fig. 5C) although the density of label is less than the smaller vesicles in which the label appears initially (Fig. 5B). This is consistent with observations on multivesicular MIIC compartments, which are though to reflect more endosomal, rather than lysosomal, characteristics (Geuze, 1994). These vesicles were found in close proximity to L. mexicana PVs. 24 hours after loading dig-dextran was abundant in PVs and surrounding vesicles (Fig. 6A) and could also be detected within the flagellar pocket of the amastigotes (Fig. 6B). The flagellar pocket of Leishmania is the sole site of endocytosis of macromolecular nutrients and has been shown to be involved in internalization of material from the PV lumen (Russell et al., 1992).

According to previous studies, autophagosomes are derived from ribosome-free ER membranes enclosing cytoplasmic material and organelles (Dunn, 1990a,b; Ueno et al., 1991). We probed cryosections from L. mexicana-infected macrophages with antibodies against calnexin, an ER membrane-associated chaperone (Tatu and Helenius, 1997) and protein disulphide isomerase (PDI), a luminal ER enzyme (Freedman et al., 1994). However the membranous debris found within the vacuole showed only low levels of label with both antibodies (not shown). The relatively low level of calnexin and PDIlabeling in the PV, in comparison to neighboring ER, may be due to relatively rapid degradation of these proteins within the lysosomal milieu (Biederbick et al., 1995).

Kobayashi and colleagues recently described the enrichment of lysobisphosphatidic acid, LBPA, within the internal membranes of multivesicular late endosomes of BHK cells. Antibodies against LBPA endocytosed by live cells interfered with trafficking of M6PR and altered the structure of these multivesicular endosomes (Kobayashi et al., 1998), suggesting that LBPA-enriched domains are a dynamic component of the normal M6PR-trafficking pathway. Immunoelectron microscopy with anti-LBPA on uninfected macrophages syringe-loaded with biot-dextran demonstrated that the multivesicular structures that label strongly with dextran also show reactivity with anti-LBPA antibody (Fig. 7). The LBPA label was associated predominantly with the membranes of the internal vesicles, as reported previously (Kobayashi et al., 1998), whilst biotin-dextran label was localized primarily within the lumen of both the internal and external vesicles, consistant with an autophagous process. LBPA label was also observed on the membraneous debris within the PVs of infected macrophages; however, because cultured L. mexicana ALFs also produce LBPA the source of the lipid in infected macrophages is open to question. However, these data suggest that there is a point of confluence between the endocytic and autophagosomal pathways prior to their subsequent delivery to leishmanial PVs.

In order to determine whether or not L. mexicana played any active role in promoting or modulating the autophagosomal process in their host cell we analysed the rate of sequestration of macromolecules from the cytosol into hydrolytically competent vesicles. The cysteine proteinase substrate, ZFR-Rho, was absorbed onto RSA and loaded into the cytosol of control or L. mexicana-infected MO. Loading of cells was conducted in the presence of leupeptin to block activity of cysteine proteinases from lysed or damaged cells. Fluorescence was negligible at early time points (Fig. 8A,B) but increased as the assay progressed. Label was observed initially in small vesicles before it was detectable in PVs (Fig. 8C,D). Quantification of the rates of cleavage were calculated as the sum of fluorescence from both cells and medium, expressed as a percentage of the level of total fluorescence obtained from complete hydrolysis of the substrate (Fig. 9). In both infected and uninfected MO the total levels of hydrolyzed substrate (cell associated + medium) was comparable at 30 minutes (approx. 20%), 150 minutes (approx. 40%) and 600 minutes (approx. 75%) post-loading. The only marked and reproducible difference between infected and control cells resided in the increased level of fluorochrome retained by infected MO. These data indicate that the rate of sequestration and degradation of cytosolic material is unaffected by the infection but that L. mexicana PVs act as a sink for hydrolyzed, lysosomal constituents and so retard release of the fluorochrome from the cell.

The intracellular compartments harbouring L. mexicana satisfy the parasite’s nutritional requirements and support their growth. Data from the current study suggests that L. mexicana PVs accumulate material coming from the host cell’s cytosol via either exploitation of the cellular recycling machinery, i.e. autophagy, or through the activity of the host cell’s organic anion transporter.

The active transport of LY and other organic anions has been described before in MO-like cell lines, and is attributed to transporter molecules in the vacuolar membrane (Steinberg et al., 1987, 1988; Lipman et al., 1990). Although the molecular basis of this transport process has remained undefined, the process itself is thought to function as a sequestration system for small organic breakdown products, peptides and toxins (as prostaglandins, leukotrienes a.o.) in the cytosol to be exported from the cell via lysosomal exocytosis (Steinberg et al., 1988; Lipman et al., 1990; Cao et al., 1991). This process was ascribed to molecules related to the P-glycoprotein family of multi-drug resistance transporters, which are capable of transportation of a range of heterogeneous substrates (Efferth et al., 1989). Published reports on organic anion transporter activity indicate that it is present in late endosomes but not in lysosomes (Steinberg et al., 1988; Lipman et al., 1990). In addition to L. mexicana PVs, we found that phagosomes containing H. capsulatum, L. monocytogenes DLly or latex beads also show Lucifer Yellow transport, albeit to a lesser extent. Since all those compartments have been described as late endosomal/phagolysosomal (Eissenberg et al., 1988; Antoine et al., 1990; Russell et al., 1992; Collins et al., 1997), organic anion transporters can serve as (an) additional marker(s) for this stage of phagosome maturation. In contrast, organic anion transporter activity was absent from phagosomes containing M. avium, a compartment that previous studies had indicated to be stabilized within the early/recycling endosomal system (Clemens and Horwitz, 1996; Russell et al., 1996; Sturgill-Koszycki et al., 1996). Interestingly, PVs of another intracellular parasite, Toxoplasma gondii, also facilitate transferance of LY, although in this instance via a passive pore rather than an active transport process (Schwab et al., 1994).

In contrast to the transport of LY, the transfer of dextran from cytosol to L. mexicana PVs appears as a non-selective bulk process and we suggest that the underlaying mechanism is through intersection of PVs with the autophagosomal system (Dunn, 1994; Seglen and Bohley, 1992). Stacey and Allfrey (1977) observed removal of rhodamine-labelled proteins following microinjection of the proteins into the cytosol of HeLa cells and attributed this phenomenon to autophagy. Autophagy, although an important process in all eukaryotic cells, is relatively poorly understood and has only been studied in a limited number of cell systems, most notably hepatocytes (Lawrence and Brown, 1992; Cuervo et al., 1995; Fengsrud et al., 1995), fibroblasts (Punnonen et al., 1992, 1993, 1994), pancreatic cells (Tooze et al., 1990) and yeast cells (Harding et al., 1996; Baba et al., 1994; Noda et al., 1995). In these cells, autophagy appears to follow a distinct sequence of events: an autophagosome is derived from a ribosome-free ER sheet forming a double membrane vesicle around an area of cytosol, or organelles, in an apparently non-selective manner (Dunn et al., 1990a,b; Ueno et al., 1991; Rabouille et al., 1993; Fengsrud et al., 1995). The maturation of this compartment is accompanied by acidification, M6PR-mediated delivery of hydrolytic enzymes, degradation of the inner membrane and, finally, fusion with differentiated lysosomes to form autophagolysosomes (Tooze et al., 1990; Høyvik et al., 1991; Punnonen et al., 1992, 1993, 1994; Lawrence and Brown, 1992). Similar to transfer of cytosolic dextran to L. mexicana PVs, autophagolysosome formation is dependent on intact actin filaments, as shown by the inhibitory effect of cytochalasin D (data not shown) (Aplin et al., 1992). Autophagy is enhanced by starvation of cells through reduction of either complex nutrients or amino acids. And, in turn, autophagy may be downregulated through the increased availability of certain amino acids or purines. Previous reports detail that adenine or 3-methyladenine downregulates formation of early autophagosomes, or the initial sequestration of cytosolic material, whereas asparagine interferes with maturation or fusion of autophagosomes with lysosomal compartments (Høyvik et al., 1991; Punnonen et al., 1994; Fengsrud et al., 1995). In our current studies we observed that adenine and 3-methyladenine, but not asparagine, had an inhibitory effect on dextran transfer into L. mexicana PVs, suggesting that the PVs intersect with early stages in the biogenesis of autophagosomes rather than with mature autophagolysosomes.

Experiments designed to reveal the rate of sequestration and degradation of cytosolic macromolecules exploiting fluorogenic cysteine proteinase substrates linked to RSA indicated a rapid removal of most of the material injected into the MO cytosol, i.e. 70% hydrolysis within 10 hours of loading. These data demonstrate that parasitization has no effect on the rate of autophagy per se but it does affect the retention of lysosomal material by the host MO.

The presence of large amounts of membranous material decorated with M6PR had been a confusing feature, detailed in an earlier study (Russell et al., 1992), although it was noted then that similar observations had been made in autophagous vacuoles (Tooze et al., 1990). And, while some reports describe the occurrence of ER proteins in autophagosomes, we observed limited amounts of calnexin and PDI within PVs. We suggest that the relatively low abundance of ER proteins may be explained by their limited resistance to hydrolytic cleavage, an explanation favored in the recent study of autophagosomes (Biederbick et al., 1995). Our current study revealed that the membraneous vesicles within the autophagosomal vacuoles were enriched for LBPA, which forms membrane domains thought to modulate trafficking of certain endosomal constituents. Moreover, the structure of these multivesicular autophagosomes is reminiscent of studies on the antigen-processing compartments or MIIC (Geuze, 1994; Amigorena et al., 1994), indicating the autophagy may play a role in the formation of these multimembraneous complexes.

The intersection of pathogen-containing compartments with the host cell’s autophagosomal pathway appears to be a unique method for intracellular pathogens to access cytosolic components. Although all Leishmania spp. survive within phagolysosomal compartments of macrophages, only members of the Mexicana complex (L. mexicana and L. amazonensis) induce these large vacuoles in their host cells. Recent data implicate a complex proteophosphoglycan filament secreted by the parasite in the induction of vacuolar enlargement (Ilg et al., 1995). Whether the proteophosphoglycan is responsible for intersection with the autophagous pathway through modulation of enhanced fusion of PVs remains to be studied. The ‘sink-like’ nature of L. mexicana PVs suggests a highly promiscuous fusigenicity with other intracellular vesicle populations. This property has been demonstrated amply in a series of studies by Rabinovitch and colleagues on transfer of phagocytosed particles into L. amazonensis PVs (Veras et al., 1992, 1994, 1995). Recent studies in our laboratory indicate that the efficiency of fusion of phagosomes with L. mexicana PVs correlates inversely with the rate of acquisition of annexin I, a protein proposed to target vesicles to lysosomes (Collins et al., 1997). In addition, M. avium vacuoles, which reside within the early/recycling endosomal network, do not fuse with leishmanial PVs. One interpretation of these observations is that L. mexicana PVs are exquisitely receptive to fusion with vacuoles possessing the characteristics of late endosomal compartments following their transition through the sorting endosomal apparatus yet prior to their differentiation into lysosomes.

These findings indicate that L. mexicana has evolved to access several sources of host cell material – endosomes, phagosomes and autophagosomes. Such a strategy could provide two different adaptive advantages to the parasite. First, it provides an additional source of nutrients, some of which, such as purines, might otherwise be in low abundance in the host cell’s lysosomal system and would limit parasite growth (Hansen et al., 1984). Second, L. mexicana PVs lie within the host’s antigen processing and presentation pathway, and although the efficiency of presentation of parasite-derived antigens is known to be low, one possible mechanism to reduce presentation of parasite peptides would be through loading up PVs with degraded ‘self’ or host-derived peptides. These would compete for loading onto MHC class II molecules with the relatively low amounts of parasite proteins released by viable parasites. Future studies on the relevance of this phenomenon need to focus on the contribution of all these three pathways to survival and growth of the parasites both in in vitro macrophage infections as well as the maintenance of in vivo infections.

The authors would like to thank Jaime Dant for his expert technical assistance, and Helen Collins, Sheila Sturgill-Koszycki, David Sibley and Jean Gruenberg for their comments and suggestions. This work was supported by US PHS grant AI 37977 to D.G.R. and by a grant to U.E.S. from the Deutsche Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie. D.G.R. is a recipient of a Burroughs Wellcome Scholar Award in Molecular Parasitology. W.F.M. was supported by NIH grant AI41599 and by the US Department of Energy Office of Biological and Environmental Research.