The lipid composition of autophagic vacuoles regulates expression of multilamellar bodies

Multilamellar bodies (MLBs) are responsible for surfactant secretion in type II alveolar cells but also accumulate in other cell types under pathological conditions, including cancer and lysosomal storage diseases such as Niemann-Pick C (NPC), a congenital disease where defective cholesterol transport leads to its accumulation in lysosomes. Mv1Lu type II alveolar cells transfected with Golgi β1,6 N-acetylglucosaminyltransferase V (Mgat5), enhancing the polylactosamine content of complex-type N-glycans, exhibit stable expression of MLBs whose formation requires lysosomal proteolysis within dense autophagic vacuoles. MLBs of Mgat5-transfected Mv1Lu cells are rich in phospholipids and have low levels of cholesterol. In Mv1Lu cells treated with the NPC-mimicking drug U18666A, cholesterol-rich MLBs accumulate independently of both Mgat5 expression and lysosomal proteolysis. Inhibition of autophagy by blocking the PI 3-kinase pathway with 3-methyladenine prevents MLB formation and results in the accumulation of non-lamellar, acidic lysosomal vacuoles. Treatment with 3-methyladenine inhibited the accumulation of monodansylcadaverine, a phospholipid-specific marker for autophagic vacuoles, but did not block endocytic access to the lysosomal vacuoles. Induction of autophagy via serum starvation resulted in an increased size of cholesterol-rich MLBs. Although expression of MLBs in the Mv1Lu cell line can be induced by modulating lysosomal cholesterol or protein glycosylation, an autophagic contribution of phospholipids is critical for the formation of concentric membrane lamellae within late lysosomal organelles.


Introduction
Multilamellar bodies are lysosomal organelles containing multiple concentric membrane layers. MLBs vary in size from 100 to 2400 nm and are found in various cell types where their principal functions are storage and secretion of lipids (Schmitz and Muller, 1991). In lung alveolar type II cells, MLBs are responsible for the secretion of the surfactant film that prevents alveolae from collapse during respiration (Hatasa and Nakamura, 1965). Deficient expression of the hydrophobic surfactant protein B (SP-B) results in the formation of immature MLBs and secretion of non-functional surfactant (Foster et al., 2003). Abnormal MLBs have also been observed in familial desquamative and non-specific interstitial pneumonitis associated with mutations in surfactant protein C (SP-C) gene (Thomas et al., 2002). Another protein, the ATPbinding cassette transporter A3 (ABCA3) is also localized to lamellar bodies of alveolar type II cells and is critical for the formation of MLBs and production of surfactant (Mulugeta et al., 2002;Shulenin et al., 2004). MLB formation in type II alveolar cells is therefore critically dependent on the protein composition of the organelle.
The lipid composition of MLBs is 95% dipalmitoyl phosphatidylcholine (DPPC), a neutral phospholipid that represents the major active component of surfactant (Schmitz and Muller, 1991). Lung surfactant is composed of 80% glycerophospholipid, 10% cholesterol and 10% protein; the amount of cholesterol within the surfactant can increase relative to the phospholipid content under certain conditions, including exercise (Doyle et al., 1994) and hyperpnea (Orgeig et al., 2003). In lung MLBs, cholesterol is localized primarily to the limiting membrane of the organelle (Orgeig, 2001) and whether MLBs are the source of surfactant cholesterol remains uncertain (Hass and Longmore, 1979;Orgeig et al., 1995). Augmentation of cellular cholesterol stimulates MLB expression, the accumulation of cholesterol within MLBs and the uptake of palmitic acid, the precursor of DPPC, in alveolar type II cells (Kolleck et al., 2002). Extracellular cholesterol is therefore internalized and stored in the MLBs of type II alveolar cells but whether it impacts on their biogenesis remains poorly understood.
The presence of MLBs is also associated with various lysosomal storage diseases, including gangliosidosis, Tay-Sachs, Fabry's and Niemann-Pick, associated with deficiencies in various lysosomal degradative enzymes and aberrant lysosomal accumulation of lipids (Blanchette-Mackie, 2000;Gieselmann, 1995;Pentchev et al., 1987;Platt et al., 1997). Cholesterol accumulation is closely related to the expression of MLBs in the Niemann-Pick lysosomal storage diseases.

Multilamellar bodies (MLBs) are responsible for surfactant secretion in type II alveolar cells but also accumulate in other cell types under pathological conditions, including cancer and lysosomal storage diseases such as Niemann-Pick C (NPC), a congenital disease where defective cholesterol transport leads to its accumulation in lysosomes. Mv1Lu type II alveolar cells transfected with
Golgi β1,6 N-acetylglucosaminyltransferase V (Mgat5), enhancing the polylactosamine content of complex-type Nglycans, exhibit stable expression of MLBs whose formation requires lysosomal proteolysis within dense autophagic vacuoles. MLBs of Mgat5-transfected Mv1Lu cells are rich in phospholipids and have low levels of cholesterol. In Mv1Lu cells treated with the NPCmimicking drug U18666A, cholesterol-rich MLBs accumulate independently of both Mgat5 expression and lysosomal proteolysis. Inhibition of autophagy by blocking the PI 3-kinase pathway with 3-methyladenine prevents MLB formation and results in the accumulation of non-lamellar, acidic lysosomal vacuoles. Treatment with 3-methyladenine inhibited the accumulation of monodansylcadaverine, a phospholipid-specific marker for autophagic vacuoles, but did not block endocytic access to the lysosomal vacuoles. Induction of autophagy via serum starvation resulted in an increased size of cholesterol-rich MLBs. Although expression of MLBs in the Mv1Lu cell line can be induced by modulating lysosomal cholesterol or protein glycosylation, an autophagic contribution of phospholipids is critical for the formation of concentric membrane lamellae within late lysosomal organelles.
Although Niemann-Pick A and B are associated with sphingomyelinase deficiency, Niemann-Pick C and D are caused by impaired intracellular cholesterol trafficking (Kolodny, 2000;Pagano et al., 2000). Patients with Niemann-Pick C disease are deficient for expression of the NPC1 protein implicated in the regulation of intracellular cholesterol traffic (Blanchette-Mackie, 2000;Ory, 2000;Pentchev et al., 1987). A class of drugs (class 2 amphiphiles such as U18666A) impairs cellular cholesterol traffic and results in the accumulation of cholesterol in late endosomes, lysosomes and MLBs mimicking Niemann-Pick C disease (Butler et al., 1992;Lange et al., 1998;Liscum et al., 1989).
Fibroblasts from patients with sphingolipid storage diseases show defects in lipid transport and sorting (Pagano et al., 2000). In many of these diseases, altered cholesterol homeostasis leads to perturbations in lipid traffic (Puri et al., 1999). Increased cellular cholesterol alters sphingolipid trafficking, resulting in its delivery not to the Golgi from the plasma membrane but rather to endolysosomal compartments such that altered trafficking of sphingolipids can be considered a diagnostic tool for the identification of sphingolipid storage diseases (Chen et al., 1999). U18666A-mediated cholesterol accumulation can be reduced by overexpression of Rab 7 and 9 GTPases suggesting that the modulation of endosomal lipid composition can impact on the delivery of material to lysosomes (Choudhury et al., 2002;Kobayashi et al., 1999;Pagano et al., 2000). Cholesterol accumulation in lysosomal organelles has further been proposed to result in the trapping of lipid raft components in late endocytic structures that could result in the formation of MLBs (Lusa et al., 2001;Simons and Gruenberg, 2000).
Multilamellar bodies are therefore ubiquitous organelles expressed under various physiological and pathological conditions. The similar morphology and lysosomal nature of these organelles in various cell types argues that common mechanisms must necessarily regulate their formation. Studies of MLB biogenesis in the Mgat5-transfected type II alveolar Mv1Lu cells showed that MLB formation could be prevented by treatment with the protease inhibitor, leupeptin, resulting in the accumulation of dense autophagic vacuoles, and that inhibition of autophagy with 3-methyladenine (3-MA) also blocked MLB expression (Hariri et al., 2000). Similarly, lysosomal protein degradation is required for MLB formation in primary human lung type II alveolar cells . Early electron microscopic studies of fetal lung describing the coordinated loss of glycogen with the accumulation of multilamellar bodies (Campiche et al., 1963;O'Hare and Sheridan, 1970) and the presence of cytoplasmic glycogen in lamellar bodies in type II epithelial cells in fetal lung, further support a role for autophagy in MLB formation (Stahlman et al., 2000;Weaver et al., 2002). However, despite the known ability of cholesterol accumulation to stimulate MLB formation in lysosomal storage disease models, the impact of lysosomal cholesterol on the biogenetic mechanisms underlying MLB formation remains undetermined.
MLBs in Mgat5-transfected Mv1Lu cells are rich in phospholipids and low in cholesterol and we have used the NPC-mimicking amphiphilic drug U18666A [3β-(2diethylaminoethoxy)-androstenone] to study the role of lysosomal cholesterol accumulation on the biogenesis of MLBs in the Mv1Lu type II alveolar cell line. U18666A stimulates MLB formation in both parental and Mgat5transfected Mv1Lu cells. It further transforms the dense autophagic vacuoles that accumulate following extended treatment with the lysosomal protease inhibitor leupeptin into MLBs. Cholesterol therefore induces MLB formation in Mv1Lu type II alveolar cells independently of both Mgat5 expression and lysosomal degradation. Although inhibition of autophagy with the PI 3-kinase inhibitor 3-methyladenine prevents the expression of concentric lamella in the U18666Ainduced cholesterol-rich lysosomal vacuoles, stimulation of autophagy by serum starvation results in an increased size of MLBs presenting the concentric lamellar morphology typical of MLBs. Phospholipid-rich MLBs as well as U18666Ainduced cholesterol-rich MLBs are accessible to fluid-phase uptake after 4 hours and labelled with Lysotracker Red, identifying them as late lysosomal organelles or autolysosomes. Similarly, the non-lamellar vacuoles induced in the presence of 3-MA are late lysosomal organelles but exhibit significantly reduced labelling for the autolysosomal marker monodansylcadaverine. Multiple factors, including cholesterol content, glycosylation and autophagy, can therefore contribute to the formation of concentric lamellae within secondary lysosomes and autolysosomes.

Cell culture
Normal Mv1Lu mink lung epithelial cells, the M9 clone of Mv1Lu cells transfected with Mgat5 (Demetriou et al., 1995) were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with glutamine, vitamins, non-essential amino acids (Invitrogen, Burlington, ON) and 10% FBS (Medicorp, Laval, Qc) in an air/5% CO 2 atmosphere at constant humidity at 37°C. To maintain the phenotype of Mgat5-transfected Mv1Lu cells, the medium was supplemented with G418 at a final concentration of 600 µg/ml (Hariri et al., 2000). Cells were plated at a density of 40,000 cells/cm 2 and the medium was replaced every 2 days. To induce formation of cholesterol-rich MLBs, medium was supplemented with U18666A (1:10,000 of a 10 mg/ml stock solution in ethanol) for 24 hours.
Leupeptin was added to the medium at a concentration of 2 µg/ml. 3-MA, at a final concentration of 10 mM, was dissolved directly in the medium, which was then sterilized by filtering (Hariri et al., 2000).

Immunofluorescence
Normal and Mgat5-transfected Mv1Lu cells were grown on glass coverslips for 6 days and then processed for immunofluorescence as previously described (Hariri et al., 2000). Cells were fixed with 3% paraformaldehyde for 15 minutes, washed with PBS supplemented with 0.1 mM Ca 2+ and 1 mM Mg 2+ (PBS-CM) and then incubated for 15 minutes with PBS-CM supplemented with 0.2% BSA to reduce non-specific binding and 0.07% saponin to permeabilize cellular membranes. AC17 anti-LAMP-2 antibody (Nabi et al., 1991;Nabi and Rodriguez-Boulan, 1993) was used to determine the cellular distribution of LAMP-2 using Alexa 488 goat anti-mouse IgG as secondary antibody. Intracellular cholesterol was visualized by filipin labelling (1:25 of a stock solution of 10 mg/ml in DMSO). Phospholipid content was determined by mounting coverslips in gelvatol containing Nile Red (1:1000 of a saturated solution).
To follow fluid-phase endocytosis, cells were incubated for various periods of time with lysine-fixable FITC-Dextran (5 mg/ml) in cell media. Cells were then washed with PBS-CM and processed for immunofluorescence as described above. Visualisation of acidic organelles was performed by incubation of cells for 10 minutes with 0.1 µM Lysotracker Red added to the medium prior to fixation and labelling with filipin. Labelled cells were viewed with a 100ϫ NeoFluor objective of a Zeiss Axiophot fluorescent microscope equipped with UV, FITC and rhodamine filter sets, a QImaging Retiga CCD camera and Northern Eclipse imaging software (Empix Imaging, Mississuaga, ON) or the 100ϫ planapochromat objective of a Leica TCS-SP1 confocal microscope equipped with Argon (488), Krypton (568) and HeNe (633) lasers. Quantification of endocytosis was performed by counting the percentage of swollen LAMP-2/Nile Red-positive vacuoles labelled for FITC-dextran from ten confocal images per condition. Presented data were compiled from three independent experiments.
MDC labelling was performed as previously described (Biederbick et al., 1995;Munafo and Colombo, 2001). Images were obtained with the 60ϫ PlanApo objective of an Olympus-PTI microscope system equipped with DeltaRam V monochromator and Cascade BF CCD camera. Equivalent acquisition settings and display parameters were used for all samples from a single experiment. Quantification was performed by measuring the average intensity of MDC labelling within large vacuoles (1.4 to 8 µm), labelled in a second fluorescent channel for Nile Red, using Image Master software. Presented data, compiled from three independent experiments, were normalized to the mean MDC intensity values of MLBs from control M9 cells and a two-tailed Student's t-test was performed.
Induction of autophagy by serum starvation was performed on cells grown on coverslips for 5 days and then treated with U18666A for 24 hours. Cells were washed twice with serum-containing or serum-free DMEM containing U18666A and incubated for 0.5, 1, 3 or 6 hours before fixation and labelling with filipin and Sytox Green. Images were obtained with the 60ϫ PlanApo objective of an Olympus FV1000 confocal microscope. The surface area of filipin-positive structures, with diameter 0.5 to 5 µm that corresponded to MLB size (Table 1), was determined using Image Master software. Overlapping vacuoles were manually segmented where possible and only single, isolated filipin-positive structures were quantified. The number of cells present in each field was determined by counting Sytox Greenlabelled nuclei. Equivalent acquisition settings and display parameters were used for all samples from a single experiment and data from three different experiments were compiled and a two-tailed Student's t-test performed.

Electron microscopy
Cells were rinsed with 0.1 mM sodium cacodylate, pH 7.3 and then fixed for 1 hour with 2% glutaraldehyde at 4°C. After fixation, the cells were rinsed with cacodylate buffer, scraped from the Petri dish, pelleted and post-fixed with 2% osmium tetroxide at 4°C. The cells were dehydrated and then embedded in LR-White resin. Ultra-thin sections were prepared and treated with uranyl acetate and lead citrate to enhance contrast. The sections were visualized with a Zeiss CEM902 electron microscope. Quantification of the expression of MLBs and of swollen lysosomal vacuoles in the 3-MA experiments was determined by circumscribing the cytoplasm (excluding the nucleus) and the MLBs and non-lamellar lysosomal vacuoles from six images at 3000ϫ magnification and determining the area of the circumscribed regions. MLBs were defined as membranebound cytoplasmic organelles presenting at least three distinct circumferential concentric membrane lamellae. MLBs were composed either completely of concentric lamellae or of concentric lamellae surrounding a single dense core (Hariri et al., 2000). Swollen lysosomal vacuoles were defined by the presence of multiple internal structures surrounded by a limiting membrane and could be morphologically distinguished from MLBs.

MLB expression in type II alveolar cells following U18666A-mediated lysosomal cholesterol accumulation
To assess the phospholipid and cholesterol distribution of the MLBs expressed upon Mgat5 transfection of Mv1Lu cells, both parental Mv1Lu and the M9 clone of Mgat5-transfected Mv1Lu cells (Demetriou et al., 1995;Hariri et al., 2000) were labelled for Nile Red, a phospholipid-specific dye that labels MLBs of alveolar type II cells (Gonzales et al., 2001;Guttentag et al., 2003), and for filipin, a cholesterol-specific dye. Although neither Nile Red nor filipin labelling associated with the LAMP-2-positive lysosomes of Mv1Lu cells (Fig. 1A-D), the multiple large LAMP-2-positive lysosomal structures that  (Hariri et al., 2000) were labelled for Nile Red indicating that they are rich in phospholipids ( Fig.  1E-H). Labelling of cholesterol with filipin shows that the lumen of the MLBs is not labelled although a peripheral filipinpositive ring is frequently observed to circumscribe the Nile Red positive MLBs (Fig. 1G,Q-T), consistent with the fact that in type II alveolar cells, cholesterol is mainly localized to the limiting membrane of the organelles (Orgeig and Daniels, 2001;Punnonen et al., 1988). Treatment of both Mv1Lu and Mgat5-transfected M9 cells with U18666A resulted in the expression of large LAMP-2positive vacuoles labelled for both Nile Red and filipin, reflecting the ability of U18666A to induce cholesterol accumulation in both lysosomes and MLBs, respectively ( Fig.  1I-P). MLBs are not observed in Mv1Lu cells using electron microscopy, whereas M9 cells present multiple cytoplasmic MLBs ( Fig. 2A,B and Table 1), as previously reported (Hariri et al., 2000). In both cell lines, numerous MLBs are observed upon treatment with U18666A (Fig. 2C,D and Table 1) indicating that lysosomal accumulation of cholesterol results in MLB formation independently of Mgat5 expression levels. In Mv1Lu cells, lysosomal cholesterol accumulation due to U18666A treatment therefore results in de novo formation of MLBs. In M9 cells, U18666A treatment increases the number of MLBs and the cytoplasmic area covered by MLBs (Table 1) indicating that cholesterol accumulation occurs within preexisting phospholipid-rich MLBs without disrupting their lamellar morphology but also induces new MLBs.
Cholesterol accumulation overrides the role of lysosomal degradation but not autophagy in MLB formation MLBs are lysosomal organelles and express various lysosomal hydrolases (de Vries et al., 1985;DiAugustine, 1974;Hatasa and Nakamura, 1965;Hook and Gilmore, 1982). The previously demonstrated ability of leupeptin treatment to prevent MLB formation in Mgat5-transfected Mv1Lu cells (Hariri et al., 2000) led us to determine whether U18666A could induce MLB formation in leupeptin-treated M9 cells.
After 96 hours in the presence of leupeptin, Mv1Lu and M9 cells express large LAMP-2/Nile Red-positive vacuoles and treatment with U18666A for 24 hours results in the formation of vacuoles strongly labelled for filipin, essentially identical to that observed in cells not treated with leupeptin (data not shown). Treatment of Mv1Lu cells with leupeptin for 4 days induced the accumulation of autolysosomes as viewed by electron microscopy (Fig.  3A). In M9 cells, leupeptin treatment resulted in the disappearance of MLBs and their replacement by large autolysosomes (Fig. 3B), as previously reported (Hariri et al., 2000). Treatment of leupeptin-treated cells with U18666A for the final 24 hours resulted in the induction of MLBs and Journal of Cell Science 118 (9) disappearance of autolysosomes in both cell types (Fig. 3C,D). U18666A-mediated accumulation of cholesterol therefore results in the formation of concentric lamellar structures within autolysosomes independently of protein degradation.
Addition of 3-MA inhibits autophagy at the sequestration step (Seglen and Gordon, 1982) and was previously shown to prevent MLB formation in Mgat5-transfected Mv1Lu cells (Hariri et al., 2000). M9 cells incubated with 3-MA for 3 days no longer expressed large LAMP-2/Nile Red-positive vacuoles and Nile Red labelling presented a diffuse cytoplasmic distribution (Fig. 4E-H), similar to that of untreated (Fig. 1) and 3-MA-treated Mv1Lu cells (Fig. 4A-D). Addition of U18666A to 3-MA-treated Mv1Lu and M9 cells for 24 hours induces the formation of large LAMP-2/Nile Red/filipin-positive vacuoles ( Fig. 4I-P). As previously reported (Hariri et al., 2000), 3-MA treatment of both Mv1Lu-and Mgat5-transfected M9 cells results in the appearance of vacuoles that do not present internal concentric lamella and that are morphologically distinct from MLBs (Fig.  5A,D). U18666A treatment of 3-MAtreated Mv1Lu and M9 cells resulted in the expression of non-lamellar vacuoles morphologically similar to those induced by 3-MA in the absence of U18666A but larger in size (Fig. 5B,C,E,F and Table 1). In the presence of U18666A, the nonlamellar vacuoles represent the only cytoplasmic structures that could correspond to the filipin-labelled LAMP-2-positive vacuoles observed by immunofluorescence labelling (Table 1). The non-lamellar vacuoles observed in the presence of 3-MA therefore correspond to   The MLBs of untreated and U18666A-treated M9 cells (Fig.  6A,C) are labelled for monodansylcadaverine (MDC), a marker for autophagic vacuoles (Biederbick et al., 1995;Munafo and Colombo, 2001). Following 3-MA treatment, the intensity of MDC labelling is significantly reduced in both untreated and U18666A-treated cells (Fig. 6B,D). The continued presence of large, swollen vacuoles labelled for Nile Red in U18666A, 3-MA-treated cells allowed us to quantify vacuole-associated MDC labelling. Untreated and U18666Atreated cells present a similar mean MDC intensity that is significantly reduced upon 3-MA treatment reflecting the ability of 3-MA to inhibit autophagy in this cell line (Fig. 6E) as previously described (Hariri et al., 2000). Serum starvation stimulates macroautophagy in various cell lines (Munafo and Colombo, 2001;Susan and Dunn, 2001). Starvation of both untransfected and Mgat5-transfected Mv1Lu cells following U18666A treatment leads to an increase in the size of filipinpositive vacuoles after 3 and 6 hours as well as an increase in the total area covered by filipin-positives vacuoles per cell (Fig.  7). The vacuoles of serum-starved cells still present a multilamellar morphology (data not shown). Stimulation of macroautophagy is therefore associated with increased size and expression of MLBs.

3-MA does not affect fluid-phase endocytosis to lysosomes
The drug 3-MA is an inhibitor of PI 3-kinase (PI3K) (Blommaart et al., 1997) and the PI3K product phosphatidyl inositol 3-phosphate [PI(3)P] is known to be involved in endocytic trafficking as well as the autophagic process (Petiot et al., 2000;Simonsen et al., 2001;Wurmser et al., 1999). Inhibition of PI3K activity has also been shown to induce the redistribution of lysosomal glycoproteins from lysosomes to mannose-6-phosphate receptor-negative, acid hydrolasenegative late endosomal compartments (Reaves et al., 1996). We therefore examined whether 3-MA treatment was preventing endocytosis in Mgat5-transfected Mv1Lu cells and whether the U18666A LAMP-2-positive vacuoles are late endocytic or lysosomal structures. Uptake of the fluid-phase marker FITC-dextran to phospholipid and cholesterol-rich MLBs was followed in untreated and 3-MA-treated cells. FITCdextran is not present in LAMP-2-positive phospholipid-and cholesterol-rich MLBs of M9 cells after endocytosis for 30 minutes, but accumulates in these structures after 4 hours (Fig.  8). Similar results were observed for Mv1Lu and M9 cells treated with U18666A in the absence or presence of 3-MA (Fig.  8). Quantification revealed that delivery of FITC-dextran to LAMP-2/Nile Red-labelled lysosomal vacuoles was essentially identical irrespective of the presence of U18666A and/or 3-MA (Fig. 8S). Labelling of the majority of the vacuoles after 4 hours' uptake identifies them as late lysosomal structures, consistent with the derivation of MLBs from autolysosomes (Hariri et al., 2000). Inhibition of autophagy in the presence of 3-MA (Hariri et al., 2000), suggests that U18666A is acting on secondary lysosomes of Mv1Lu and M9 cells. The equivalent rate of delivery of FITC-dextran to the MLBs of untreated M9  cells and the non-lamellar cholesterol-rich vacuoles induced by U18666A in the presence of 3-MA indicates that 3-MA is not inhibiting fluid-phase endocytosis (Fig. 8S). Similarly, 3-MA did not prevent FITC-dextran endocytosis to LAMP-2-positive lysosomes in M9 cells (Fig. 9).
Lysotracker is an uncharged compound that is freely permeant to cell membranes and accumulates in acidic compartments by a process involving diffusion and trapping by protonation (Haller et al., 1996). MLBs have a pH of 6.1 or lower, which is maintained by an energy-dependent process (Chander et al., 1986). Both phospholipid-and cholesterol-rich MLBs are positive for Lysotracker Red labelling (Fig. 10). The LAMP-2-positive vacuoles observed in 3-MA-and U18666Atreated cells are also labelled by Lysotracker Red (Fig.  10J,K,L). Treatment with 3-MA does not therefore interfere with the endocytic accessibility and the acidic lysosomal nature of the large cholesterol-rich vacuoles induced by U18666A.

Discussion
Multiple mechanisms modulate the expression of MLBs in the Mv1Lu type II alveolar cell line MLB expression is usually lost upon long-term culture and in established cell lines of type II alveolar cells (Diglio and Kikkawa, 1977;Guttentag et al., 2003;Kawada et al., 1990;Rannels et al., 1987;Sannes, 1991;Tanswell et al., 1991). MLB expression in the mink Mv1Lu type II alveolar cell line can be induced by expression of β1-6GlcNAc-branched Nglycans (Hariri et al., 2000). MLBs were also induced in the Mv1Lu type II alveolar cell line by treatment with U18666A, a compound that promotes lysosomal cholesterol accumulation, independently of Mgat5 transgene expression. U18666A treatment has been shown to induce MLB formation in CHO cells (Lusa et al., 2001) but not in HeLa cells (Tomiyama et al., 2004), MDCK or mammary carcinoma cells (data not shown). Lysosomal cholesterol accumulation is therefore not sufficient to induce MLB formation and other cell-type specific factors are required. Nevertheless, the ability of both Mgat5 and cholesterol to independently promote MLB expression in the Mv1Lu type II alveolar cell line shows that multiple factors can modulate the expression of this organelle. The ability of cholesterol to override the need for lysosomal degradation in phospholipid-rich MLB formation demonstrates that the biogenesis of this organelle is a complex process that can follow different pathways. Varied lysosomal contents and biogenetic pathways are therefore permissive for MLB formation, consistent with the expression of these morphologically identical structures under normal physiological conditions in type II alveolar cells as well as pathologically in cancer and lysosomal storage diseases of different genetic origins.
Mgat5 and expression of β1-6GlcNAc-branched N-glycans products are upregulated with tumor progression (Dennis et al., 1999) and 'coarse vesicles' presenting a lamellar morphology are revealed by L-PHA, a probe for branched N-glycans, in various cancers (Handerson and Pawelek, 2003). The expression of β1-6GlcNAc-branched N-glycans on lysosomal glycoproteins may promote their expression or stability thereby contributing to lamella expression. Knockout of LAMP-2 results in the mistargeting of lysosomal hydrolases Journal of Cell Science 118 (9) Fig. 7. Serum starvation stimulates of cholesterol-rich MLBs. (A) Untransfected Mv1Lu cells were grown for 6 days and U18666A was added for the final 24 hours. Selected cultures were then washed and incubated for 6 hours in serumcontaining (B) or serum-free (C) media and labelled with filipin. The expression of filipin-labelled MLBs was quantified from both untransfected Mv1Lu (D) and Mgat5-transfected M9 cells (E) incubated in serum-containing (empty bars) or serum-free (filled bars) media for 0.5, 1, 3 and 6 hours labelled with filipin and the nuclear dye Sytox Green. The average area of filipin-labelled vacuoles and the total area covered by these vacuoles per cell were quantified (mean±s.e.m. of three independent experiments). *P<0.005, **P<0.01, ***P<0.05 compared to areas in control cells at the same time point. Bar, 10 µm. and accumulation of autophagic vacuoles (Eskelinen et al., 2002;Tanaka et al., 2000). LAMP-1 and -2 are major cellular carriers of polylactosamine (Carlsson and Fukuda, 1990) and the impact of their glycosylation on lysosome biogenesis and autophagy remain to be determined. N-glycans on cytokine receptors bind to galectins in an Mgat5-dependent manner and form a multivalent lattice that retains the receptors at the cell surface. The Mgat5-modified N-glycans increase surface residency of receptors, lowering the response thresholds to multiple cytokines, including TGF-β, EGF, PDGF and IGF (Partridge et al., 2004). In this manner, autocrine signalling through PI3K is potentiated in Mgat5-expressing cells, and may contribute to MLB formation in both Mgat5-transfected Mv1Lu type II alveolar cells as well as in tumour cells that express increased levels of Mgat5 (Handerson and Pawelek, 2003;Hariri et al., 2000).

An autophagic contribution promotes lamella formation in lysosomal organelles
Autophagy is involved in both sphingolipid and glycoprotein metabolism (Ghidoni et al., 1996) and, conversely, sphingolipid ceramide can modulate autophagy in HT-29 cells (Scarlatti et al., 2004). 3-MA and other inhibitors of PI3K activity block autophagy (Blommaart et al., 1997;Hariri et al., 2000;Munafo and Colombo, 2001;Seglen and Gordon, 1982). Our previous demonstration that 3-MA inhibits both autophagy and the formation of MLBs in Mgat5-transfected Mv1Lu cells (Hariri et al., 2000) was confirmed in this study by the disappearance of phospholipid-rich MLBs in 3-MA-treated cells (Figs 4,5). Treatment with 3-MA did not prevent the accumulation of lysosomal cholesterol but the swollen, lysosomal structures no longer expressed the concentric lamellae typical of MLBs. The similar morphology of these swollen, lysosomal structures with the smaller inclusion bodies seen upon 3-MA treatment in the absence of U18666A suggests that these two structures are similar lysosomal organelles that vary in size owing to accumulation of cholesterol. As seen in Table 1, 3-MA treatment of U18666A-treated cells resulted in essentially the complete transition from expression of MLBs to non-lamellar lysosomal vacuoles. The large size of U18666A-induced vacuoles enabled us to identify and compare the multilamellar and non-lamellar lysosomal structures in untreated and 3-MAtreated cells by light microscopy.
Inhibitors of PI3K also block vesicular transport from late endosomes to lysosomes (Punnonen et al., 1994) and impair both early endosome fusion and maturation of lysosomes (Mousavi et al., 2003;Reaves et al., 1996;Simonsen et al., 1998). More recently, PI3K signalling has been shown to more specifically regulate endosomal sorting (Petiot et al., 2003). The large non-lamellar LAMP-2-positive, cholesterol-rich vacuoles formed upon addition of U18666A to 3-MA are labelled by fluid-phase uptake of FITC-dextran after 4 hours but not 30 minutes and are positive for Lysotracker Red indicating that they are acidic, late lysosomal organelles. The non-lamellar vacuoles that accumulate in the presence of 3-MA and U18666A are therefore not pre-lysosomal compartments but correspond to late or secondary lysosomes. Furthermore, 3-MA does not affect endocytic accessibility of fluid-phase material to these late, lysosomal organelles. Biosynthetic trafficking of LAMP to lysosomes can include transit via early endosomes and/or the plasma membrane (Nabi et al., 1991;Peden et al., 2004) and recycling of LAMPs permits uptake of anti-LAMP antibodies to lysosomes (Lippincott-Schwartz and Fambrough, 1987;Nabi et al., 1991;Williams and Fukuda, 1990). As observed for fluid-phase uptake of FITC-dextran, anti-LAMP-2 antibodies internalized by recycling LAMP-2 access MLBs and non-lamellar vacuoles after 2-4 hours (data not shown).
Multivesicular bodies or endosomes have been observed to fuse with MLBs in type II alveolar cells (Chevalier and Collet, 1972;Voorhout et al., 1993;Vorbroker et al., 1995) and sorting of endosomal content to the internal vesicles of multivesicular bodies may represent a requisite aspect of MLB formation (Piper and Luzio, 2001;Stahl and Barbieri, 2002; Journal of Cell Science 118 (9) Fig. 9. 3-MA treatment does not affect fluid-phase endocytosis to lysosomes. M9 cells were grown in the presence of 3-MA for 3 days and then incubated with FITCdextran for 30 minutes (A-C) or 4 hours (D-F) prior to labelling for LAMP-2 (A,D). FITC-dextran labelling (B,E) is shown in green and LAMP-2 in red in the merged images (C,F). FITC-dextran does not accumulate in lysosomal structures at 30 minutes but reaches them after 4 hours of endocytosis. Bar, 5 µm.  , 2002). Similarly, fusion between various elements of the endolysosomal and autophagic pathways is well documented (Fig. 11). Formation of nascent autophagic vacuoles is followed rapidly by acquisition of hydrolytic enzymes by fusion with pre-existing lysosomes (Lawrence and Brown, 1992;Lawrence and Brown, 1993;Lee et al., 1989;Liou et al., 1997) and early autophagic vacuoles fuse with endosomes before fusing with lysosomes (Gordon and Seglen, 1988). Interaction between autophagic vacuoles and both endosomal and lysosomal compartments therefore necessarily contributes to the composition of autolysosomes (Fig. 11).
Autophagic vacuoles have long been noted for the presence of whorl-like membrane structures (Schmitz and Muller, 1991). MDC is a fluorescent probe whose specificity for autophagic vacuoles (Biederbick et al., 1995;Munafo and Colombo, 2001) is based, unlike other lysomotropic agents, on its ability to act as a solvent polarity probe exhibiting enhanced fluorescence upon interaction with membrane lipids (Niemann et al., 2000). MDC should therefore selectively label dense phospholipid-containing organelles such as autophagic vacuoles and MLBs. LC3 is a specific marker for the isolation membrane of autophagosomes and autophagosomes (Kabeya et al., 2000;Mizushima, 2004;Mizushima et al., 2001). MDCpositive structures negative for LC3 are induced by nutrient starvation and MDC would therefore appear to be a marker for mature autophagic vacuoles or autolysosomes (Munafo and Colombo, 2001;Gutierrez et al., 2004), consistent with this report. Reduced labelling of cholesterol-rich U18666Ainduced vacuoles with MDC in the presence of 3-MA (Fig. 6) is therefore indicative of the reduced delivery of phospholipids to these organelles via autophagy, thereby altering the lipid composition of these organelles and preventing their reorganization into concentric lamella, as observed by EM. The ability of serum starvation to enhance the size of MLBs (Fig.  7) further demonstrates that macroautophagy regulates the expression of these organelles.
This study shows that autophagic delivery of phospholipids to late lysosomal organelles promotes MLB expression in the Mv1Lu type II alveolar cell line and, indeed, argues that MLBs are autolysosomes. However, autoradiographic studies have shown that phospholipids are delivered to MLBs of type II alveolar cells in intact lung directly from the Golgi apparatus (Chevalier and Collet, 1972). Furthermore, MLBs are absent from Mv1Lu cells, as in most type II alveolar-derived cells in culture (Diglio and Kikkawa, 1977;Guttentag et al., 2003;Kawada et al., 1990;Rannels et al., 1987;Sannes, 1991;Tanswell et al., 1991), and the ubiquitous expression of MLBs in various other cell types argues that multiple mechanisms may contribute to MLB expression. Indeed, we were able to induce MLB expression in Mv1Lu cells by modulating both protein glycosylation and lysosomal cholesterol content. Whether autophagy plays a critical or regulatory role in MLB formation and surfactant expression in lung type II alveolar cells remains to be determined. Nevertheless, regulation of the lipid composition of late lysosomal organelles by autophagy promotes the expression of MLBs and may modulate expression of these organelles under both physiological and pathological conditions. This study was supported by a grant from the Canadian Institutes of Health Research (CIHR). I.R.N. is the recipient of a CIHR Investigator award.