We employed our recently developed immuno-electron microscopic method (W. Möbius, Y. Ohno-Iwashita, E. G. van Donselaar, V. M. Oorschot, Y. Shimada, T. Fujimoto, H. F. Heijnen, H. J. Geuze and J. W. Slot, J Histochem Cytochem 2002; 50: 43–55) to analyze the distribution of cholesterol in the endocytic pathway of human B lymphocytes. We could distinguish 6 categories of endocytic compartments on the basis of morphology, BSA gold uptake kinetics and organelle marker analysis. Of all cholesterol detected in the endocytic pathway, we found 20% in the recycling tubulo-vesicles and 63% present in two types of multivesicular bodies. In the multivesicular bodies, most of the cholesterol was contained in the internal membrane vesicles, the precursors of exosomes secreted by B cells. Cholesterol was almost absent from lysosomes, that contained the bulk of the lipid bis(monoacylglycero)phosphate, also termed lysobisphosphatidic acid. Thus, cholesterol displays a highly differential distribution in the various membrane domains of the endocytic pathway.
Cholesterol is an essential component of eukaryotic cell membranes and is crucial for cell viability and function. Therefore, membrane cholesterol levels are tightly controlled by the homeostatic regulation of de novo synthesis, uptake and esterification (1). The endocytic pathway plays an important role in cholesterol homeostasis since it is the entrance point and delivery site of LDL-derived cholesterol. This is reflected by the involvement of the endocytic pathway in the cholesterol-associated diseases such as hypercholesterolemia (2), Niemann-Pick type C (3) and Wolman syndrome (4). After uptake via receptor-mediated endocytosis, LDL-derived cholesteryl esters are transported to late endosomes and lysosomes and hydrolyzed by acid lipase, the cholesterol esterase that is defective in Wolman syndrome. Free cholesterol is then rapidly transported back to the plasma membrane and to the ER, where excess cholesterol becomes esterified and subsequently stored in lipid droplets (2,5). At least two proteins, NPC1 and NPC2, are involved in the transport of LDL-derived cholesterol from lysosomes to the regulatory sites of cholesterol homeostasis (6–8). This post-lysosomal pathway probably includes the Golgi complex en route to the plasma membrane and the ER (9).
Membrane cholesterol is unevenly distributed among the various compartments of the cell and is especially enriched at the plasma membrane (10–13). Plasma membrane cholesterol constitutively recycles via recycling endosomes, that appear to be cholesterol rich (14–16), while lysosomes are cholesterol poor (17–19). The association of cholesterol and sphingomyelin in cholesterol/sphingolipid-enriched microdomains, also called rafts (20,21), is believed to play an important role in cholesterol distribution and transport (22).
In this study we analyzed the distribution of cholesterol in the various organelles of the endocytic pathway in B lymphocytes. For this purpose we developed a protocol for the localization of cholesterol by using a nonhemolytic derivative of Θ-toxin (Perfringolysin O) (23) for immunoelectron microscopy on cryosections (24). This new method allowed us to study the in situ distribution of cholesterol-rich membranes in cells. We were especially interested to expand the insights obtained by fluorescence microscopy and subcellular fractionation to the fine structural level and resolve transport vesicles and domains within organelles. We have chosen B lymphocytes, because this cell type has an elaborated endocytic system, and is able to produce exosomes, like other hematopoietic cells (25–28). We demonstrated previously that both secreted exosomes and distinct endocytic compartments display high levels of cholesterol (24). Exosomes are formed by the secretion of the internal vesicles of multivesicular bodies (MVBs) after fusion of this organelle with the plasma membrane. In antigen-presenting cells such as B lymphocytes, exosomes carry MHC class I and class II, and may play a role in stimulation of T cells (26,29). Here, we characterized the endocytic pathway by tracer uptake kinetics and labeling of cholesterol and organelle markers to analyze the cholesterol distribution and identify the exosome producing endocytic organelle.
Another lipid found on exosomes, is bis(monoacylglycero)phosphate (BMP), also called lysobisphosphatidic acid (LBPA) (30). This unusual phospholipid with a unique sn-1 : sn-1′ stereoconfiguration (31,32) was considered to be localized in secondary lysosomes as identified by purification by subcellular fractionation from rat liver (33–36). A functional role of (BMP/LBPA) in lysosomes is supported by in vitro studies showing that BMP/LBPA stimulates the degradation of glycosphingolipids (GSL) in detergent-free liposomal assays (37–39). Recently, it was shown that BMP/LBPA also exhibits a pH-dependent fusogenic activity in a liposomal system, pointing towards a possible role of this lipid in membrane dynamics (40). It was proposed that BMP/LBPA might play a role in the regulation of cholesterol transport (41). Here, we compared the distribution of cholesterol with that of BMP/LBPA using immunoelectron microscopy, and we show that these lipids distribute differently in the endocytic pathway.
The human B-lymphocyte cell line RN produces cholesterol-rich exosomes by fusion of multivesicular bodies (MVBs) with the plasma membrane (Figure 1A). This cell type has been used previously for the biochemical characterization of exosomes (42). Here, we applied an improved sample preparation and cryosectioning protocol (43) in combination with a recently introduced method for the immuno-localization of free membrane cholesterol (24) to analyze in detail the distribution of cholesterol in the endocytic pathway of this cell type.
Morphological characteristics and kinetics of BSA-gold uptake in the endocytic pathway
By examining the morphology of the endocytic pathway, we identified 6 different endocytic organelles types 1–6, classified as before (44) (Table 1). Type 1 represented clathrin-coated pits and vesicles. Type 2 compartments were comprised of vesicular and tubular structures associated with type 3 (Figure 1B) and, to a lesser extent, with type 4. We treated type 2 separately, since possible continuities of tubules with endocytic vacuoles of type 3 and 4 could not always be detected in thin sections. Type 2 likely includes primary endocytic vacuoles and tubulo-vesicles belonging to the so-called recycling compartment (see below) (45,46). Type 3 represented an irregularly shaped MVB with a few internal vesicles (Figure 1B), whereas type 4 was the globular MVB with abundant internal vesicles (Figure 1A,C). Type 5 represented an intermediate between the multivesicular type 4 and the multilaminar type 6.
Table 1. : Kinetics of BSA-gold (5 nm) uptake in human B lymphocytes (RN) expressed in percentages of gold counted per time of uptake
To study the transport kinetics of BSA-gold through the endocytic pathway, RN cells were pulsed with 5 nm BSA-gold for 3 min or 10 min, or pulsed for 10 min and chased for 20 min, 50 min or 110 min. The progression of endocytosed gold through the compartments 1–6 successively (Table 1) confirmed that the order of our morphologically chosen categories in the endocytic pathway was correct and agreed with those described previously (44). Within 3 min of uptake, 22% of the BSA-gold had reached type 2 vesicles and 59% the type 3 compartment (Figure 1B), while type 4 contained peak amounts of the tracer after 20 min chase (74%) (Figure 1C). At this time, BSA-gold was chased out of type 2 and 3. Type 5 was positioned between type 4 and type 6, which was efficiently reached by BSA-gold after a chase period of 110 min (67%). As in our previous study (44), we considered both types 5 and 6 lysosomes.
Analysis of the endocytic pathway by double-labeling of cholesterol and endosomal/lysosomal markers
After specifying the endocytic pathway by morphology and tracer uptake kinetics, we next studied the distribution of cholesterol labeling. The cholesterol-containing endocytic subcompartments were characterized in immuno-double-labeling experiments with antibodies against the transferrin receptor (TfR), a marker of early endosomes and receptor recycling compartments (46,47), and the cation-dependent mannose-6-phosphate receptor (CD-MPR) which is also characteristic of recycling tubulo-vesicles surrounding endosomes (48). Cholesterol and transferrin receptor colocalized in type 2 recycling tubulo-vesicles (Figure 2A). This is also the case for the cation-dependent mannose-6-phosphate receptor (CD-MPR) (Figure 2B).
Antigen-presenting cells like B lymphocytes express MHC class II and the chaperone invariant chain (Ii). Both are useful functional molecular markers for endocytic organelles in these cells. Invariant chain is transported in a complex with MHC class II along the biosynthetic pathway and to the endocytic pathway preventing premature peptide loading. Degradation of Ii and peptide loading onto MHC class II occur in late endosomal and lysosomal MHC class II compartments (MIICs). Detectable Ii therefore identifies earlier stages of the endocytic pathway (type 3 and 4) in these cells (44). Double-labeling of cholesterol and Ii revealed colocalization in types 3 and 4 (Figure 2C). On the other hand, double-labeling of cholesterol with MHC class II showed the highest labeling for MHC class II in the cholesterol-poor lysosomal types 5 and 6 (Figure 2D). This finding was also supported by double-labeling of cholesterol and LAMP-1. LAMP-1 is a well-established marker for late endosomes and lysosomes (49). LAMP-1 was abundant in cholesterol-negative compartments type 5 and 6 (Figure 2E). These LAMP-1-positive organelles showed abundant labeling for BMP (Figure 2F).
Distribution of cholesterol labeling in the endocytic pathway
We next quantified the relative distribution of cholesterol labeling in the endocytic organelles types 1–6. Here, we further distinguished between labeling on the limiting membrane and on the internal membranes of endocytic vacuoles (Figure 3A). Only a minor fraction of the total cholesterol labeling was localized to the compartments type 1 (5.6 ± 1.6%), type 5 (6.9 ± 1.5%) and type 6 (3.8 ± 1.3%). The majority of cholesterol label was detected in types 2, 3 and 4. Interestingly, a large proportion of the labeling distributed to type 2 tubules and vesicles surrounding endosomes (20.5 ± 1.6% of total labeling). We further characterized these cholesterol-positive tubules and vesicles by immuno-double-labeling of cholesterol and TfR or CD-MPR on sections of cells that had internalized BSA-gold for 10 min. Immunogold quantitation showed that 38 ± 2% of all cholesterol-positive type 2 vesicles and tubules were also labeled for TfR. A minority of cholesterol-positive structures (3 ± 1%) were primary endocytic vesicles, as they contained BSA-gold. Double-labeling with CD-MPR resulted in colocalization in 26 ± 2% of the cholesterol-positive type 2 vesicles and tubules and 4 ± 2% contained BSA-gold. Colocalization of cholesterol with either TfR or CD-MPR in BSA-gold-containing vesicles or tubules was hardly observed. We concluded that most of the cholesterol detected in type 2 is contained in TfR or CD-MPR recycling tubulo-vesicles.
In type 3, we found 18.1 ± 4.5% of total labeling, comprising 6.4 ± 2.1% of total labeling on the limiting membrane and 11.8 ± 2.5% on the internal vesicles. Type 4 MVBs (see Figure 1A,C) contained 45.1 ± 4.7% of cholesterol labeling. Most of this labeling was confined to the internal vesicles: 39.3 ± 4.4% of total cholesterol labeling.
Distribution of BMP/LBPA labeling in the endocytic pathway
Another lipid found in small amounts on exosomes is BMP [bis(monoacylglycero)phosphate], also called LBPA (lysobisphosphatidic acid) (30). Since a role of BMP/LBPA in the regulation of cholesterol transport was suggested (41), we also analyzed the distribution of BMP/LBPA in the endocytic pathway by immunoelectron microscopy in the same way as described for cholesterol.
Table 1 and Figure 2(F) and Figure 3(B) show that the distribution of BMP/LBPA differed remarkably from that of cholesterol. Insignificant amounts of BMP/LBPA were found on types 1, 2 and 3. Type 4 contained only 4.7 ± 1.8% of total labeling. In agreement with type 4 being an important donor compartment for exosomes, we found correspondingly low labeling of BMP/LBPA on exosomes (not shown). Type 5 showed 30.6 ± 5.0% of total labeling, while type 6 contained the major pool of BMP/LBPA (64 ± 6.2% of total labeling). We further differentiated between internal membranes and limiting membrane of endocytic organelles. We found 88 ± 2.8% of the labeling confined to the internal membranes in type 4, 98 ± 0.5% in type 5 and 99 ± 0.4% in type 6 compartments.
Most available data on the localization of cholesterol in the endocytic pathway come from subcellular fractionation studies (9,18,50) and fluorescence microscopy using cholesterol derivatives (51–53) or the antibiotic filipin (54). Filipin induces cholesterol-dependent membrane deformations that can be detected by transmission electron microscopy and freeze fracture techniques (55,56). The major disadvantage of most of these approaches is that they either suffer from poor resolution, or disturb the structure of the cell. Freeze fracture is a technique of choice for the visualization of intramembrane structures at high resolution, but is inappropriate to study the whole cell in detail. Therefore, we saw a requirement for a method that allows for the intracellular localization of free membrane cholesterol by immunoelectron microscopy. With such an approach one could demonstrate the distribution of this particular lipid in the context of the cell, where structures can easily be identified by their morphology, their localization within the cell, or by labeling with tracers or antibodies to organelle markers. In the present study we made use of the special cholesterol-binding properties of a nonhemolytic and biotinylated derivative of the bacterial toxin Perfringolysin O, as described in detail elsewhere (57,58). To extend our previous studies of cholesterol labeling at the plasma membrane of the human B lymphocyte cell line RN (24), we now investigated the distribution of cholesterol throughout the endocytic pathway.
Based on morphology, the passage of endocytosed BSA-gold and immunolabeling of several endosomal markers, we could distinguish 6 different populations of endocytic structures. We identified their position in the endocytic pathway and attributed cholesterol labeling to them.
Type 1 represented typical clathrin-coated pits and vesicles at or adjacent to the plasma membrane. The presence of cholesterol in these structures is in accordance with the requirement of membrane cholesterol for clathrin-dependent endocytosis (59). Type 2 consisted of tubulo-vesicles, that were closely associated with endosomal vacuoles types 3 and 4. Type 3 is characterized as an early endosome by the rapid appearance of BSA-gold, the presence of TfR, the irregular shape as well as the electron-lucent lumen containing only a few internal vesicles. The type 2 tubulo-vesicles acquire BSA-gold within 3 min of uptake, but to a lesser extent than type 3. Most of the tracer was chased out of the type 2 tubulo-vesicles after 20 min, whereas the early type 3 endosome still contained 20% of the internalized BSA-gold at this time point. Similar kinetics of tracer endocytosis were described for so-called tubular endosomes which form a subcompartment of early endosomes (60). Interestingly, we found considerable amounts of cholesterol in type 2 compartments. To investigate the proportion of primary endocytic vesicles in type 2 compartments we double-labeled cholesterol and TfR or CD-MPR after 10 min BSA-gold uptake. This analysis revealed that a significant proportion of the cholesterol found in type 2 compartments occurred in tubulo-vesicles that labeled for TfR or CD-MPR and lacked endocytosed BSA-gold. These tubulo-vesicles were positioned in close vicinity to endosomal vacuoles and most likely belong to the recycling compartment, through which receptors recycle between endosomes, plasma membrane and the TGN (46–48). That cholesterol is required for MPR recycling from MVBs to the Golgi complex has indeed been reported (61). Furthermore, a large proportion of membrane cholesterol recycles constitutively between the plasma membrane and endosomes (16) and also resides in a TfR-positive so-called early recycling compartment (ERC) (15). By using fluorescent dehydroergosterol that mimics cholesterol, 35% of this probe was found in this recycling compartment (15). This result supports our quantitative data, that 39% of total labeling is present in TfR-positive early endosomes and associated vesicles and tubules (type 2 and 3). A recent study reports the involvement of recycling endosomes in cholesterol homeostasis, showing that Rab11 overexpression leads to an accumulation of cholesterol in recycling endosomes and impaired esterification by acylCoA:cholesterol acyl transferase (ACAT) (62). Rab 11 is engaged in vesicular trafficking through the recycling endosome (63).
Type 4 vesicles were typical MVBs with the characteristics of a late endosome defined by its transitional position between type 3 (early endosome) and types 5 and 6 (lysosomes) in the BSA-gold uptake kinetics. Type 4 MVBs had the same tracer uptake kinetics as the so-called endocytic carrier vesicle (ECV) described in BHK cells (64). Type 4 MVBs in antigen-presenting cells such as B cells harbor major pools of invariant chain (Ii) and MHC class II, and are also called MHC class II compartments (MIICs) (44). We now find type 4 MVBs or MHC class II compartments (MIICs) as the main cholesterol-containing endocytic compartment. By far most of the cholesterol labeling was contained in the internal vesicles of the MVBs, a finding that supports the view that the internal vesicles are the precursors of cholesterol-rich exosomes (26).
The disappearance of cholesterol labeling further down-stream in type 5 and 6 supports the view that cholesterol is efficiently removed from late endosomes and lysosomes (13,17). Purified lysosomes contain only traces of cholesterol (19). Possible cholesterol transport facilitators, such as NPC1 and MLN64, are also localized to these endocytic organelles (65,66), resulting in a low steady-state concentration of cholesterol in lysosomes. The withdrawal of cholesterol from the endocytic tract relies on Rab protein dependent carriers (67).
Recently, BMP/LBPA was reported as a marker of late endosomes (68) and has been used in fluorescence microscopy since then. Here, in accordance with previous studies (33–36), we find BMP/LBPA predominantly in LAMP-1 positive multilaminar lysosomes, that are reached by endocytic tracers after prolonged chase periods. Examination of the published electron micrographs describing BMP/LBPA labeling in late endosomes revealed that those late endosomes show the morphological features of multilaminar lysosomes that are reached with BSA-gold after 60 min (68). Probably these late endosomes correspond to the present type 5 and 6 endocytic organelles. Cholesterol and BMP/LBPA exhibit a mirror distribution in the endocytic pathway: cholesterol is most abundant in types 2, 3 and 4, where BMP/LBPA is scarce, while the latter is the predominantly found in the types 5 and 6 with low cholesterol contents. The localization of BMP/LBPA and cholesterol overlaps only to a low degree at the stage of type 4, which we identified as the exosome producing late endosome. This explains why both lipids have been found in exosomes (30).
The cholesterol-labeled membranes we identified here, mostly small vesicles and tubules, are highly curved membrane structures. Although we cannot exclude an underestimation of the amount of cholesterol-containing membranes, since BCΘ binds preferentially to membranes with high concentrations of cholesterol (58), our findings are supported by studies showing a requirement for cholesterol in the formation of highly curved structures such as caveolae and synaptic vesicles (69,70). A similar mechanism may also play a role in the formation of internal vesicles of MVBs.
In conclusion, by immuno-electron microscopy more than 80% of the cholesterol detected in the endocytic pathway is present in the MVB stages of early and late endosomes and in recycling compartments. In the multivesicular vacuole cholesterol prevails in the internal vesicles which represent the precursor vesicles of exosomes. Accordingly, exosomes also contain abundant cholesterol. The highly differential distribution of cholesterol in the endocytic pathway probably is the result of efficient lipid sorting processes in the membranes of early and late endosomal compartments.
Materials and Methods
The EBV-transformed human B-cell line RN (HLA-DR15+) was maintained as described (26).
Reagents and antibodies
Perfringolysin O (Θ-toxin) was prepared and digested with subtilisin Carlsberg as published previously (71,72). Biotinylated Θ-toxin (BCΘ) was prepared from the nicked Θ-toxin as described (23). Colloidal gold conjugates to bovine serum albumin (BSA) and protein A were prepared according to (73) and (74). For immuno-electron microscopy we used polyclonal rabbit antibodies against biotin (Rockland, Gilbertsville, PA, USA), invariant chain (Ii) N-terminus (75) from Dr P. A. Morton (Montesanto Co., St. Louis, MO, USA), MHC class II from Dr H. L. Ploeg (Massachusetts Institute of Technology, Cambridge, MA, USA) and LAMP-1 (49) from Dr M. Fukuda (Cancer Research Center, La Jolla, CA, USA). We also used the following monoclonal antibodies against the transferrin receptor (TfR) HTR-H68.4 (76) which was obtained from Zymed (San Francisco, CA, USA), the cation-dependent mannose-6-phosphate receptor (CD-MPR) (77), provided by Dr K. von Figura, and BMP/LBPA (6C4) (68), which was a generous gift of Dr J. Gruenberg (Geneva, Switzerland). For the detection of monoclonal antibodies we used a rabbit antibody against mouse IgG (from DAKO, Glostrup, Denmark).
Internalization of BSA-gold
The BSA-gold uptake studies were performed according to (44). Briefly, RN cells were either incubated with BSA-gold (OD = 5) for 3 min or 10 min at 37 °C and fixed after washing with cold medium, or incubated with BSA-gold for 10 min, washed extensively in cold medium and further incubated at 37 °C for 20 min, 50 min or 110 min and then fixed. Fixation was carried out following a protocol that was developed for cholesterol localization (24), (see below). For every time point, 20 cell profiles were randomly selected, and gold particles present in the endocytic structures were attributed to compartments of types 1–6 (see description in Results). For each time point, BSA-gold counts in types 1–6 were expressed as percentage of total gold present in the endocytic pathway.
RN cells were prepared for the localization of cholesterol as detailed before (24). Briefly, cells were washed with 0.1 m PHEM buffer (60 mm PIPES, 25 mm HEPES, 2 mm MgCl2, 10 mm EGTA, pH 6.9), fixed in 4% formaldehyde in 0.1 m PHEM buffer and infiltrated in 1.75 m sucrose in 0.1 m PHEM buffer containing 4% formaldehyde. Droplets of cells were then transferred onto pins and frozen in liquid nitrogen. Ultrathin cryosections were picked up according to (43) in a 1 : 1 mixture of 2% methylcellulose and 2.3 m sucrose. For cholesterol labeling, sections were incubated with BCΘ(15 μg/ml) immediately after thawing, and fixed with 1% glutaraldehyde after brief rinses. BCΘ was then detected with antibiotin antibodies and 10 nm protein A-gold. For immuno-double-labeling of marker proteins and cholesterol, sections were first incubated with BCΘ, then labeled with antibodies specific for the marker, detected with protein A-gold (15 nm), followed by an incubation with antibiotin antibodies and protein A-gold (10 nm). For quantitation of cholesterol- and BMP/LBPA-labeling, 10 nm gold particles were counted on 20 randomly selected cell profiles of five independent samples. Gold particles were assigned to a compartment of types 1–6 when lying within a distance of 20 nm from its limiting membrane. The relative distribution of labeling was then expressed as a percentage of total labeling ± SEM per type 1–6 of endocytic compartments. For the quantitative analysis of the type 2 compartment, sections of cells that had taken up BSA-gold for 10 min were immuno-double-labeled for cholesterol and TfR or CD-MPR. Vesicles and tubules labeled for cholesterol were counted on three independent samples for TfR and CD-MPR, respectively. The number of cholesterol-positive vesicles with or without labeling for TfR or CD-MPR or BSA-gold content was then expressed as a percentage of total cholesterol-positive vesicles and tubules ± SEM.
W.M. is a CBG (Centre of Biomedical Genetics) postdoctoral fellow at the UMC Utrecht, the Netherlands. This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (to Y.O-I.). Y.S. is a Domestic Research Fellow of the Japan Science and Technology Corporation. We thank R. M. C. Scriwanek and M. van Peski for the excellent photographic work.