Cholesterol overload promotes morphogenesis of a Niemann-Pick C (NPC)-like compartment independent of inhibition of NPC1 or HE1/NPC2 function.

Cholesterol accumulation in an aberrant endosomal/lysosomal compartment is the hallmark of Niemann-Pick type C (NPC) disease. To gain insight into the etiology of the NPC compartment, we studied a novel Chinese hamster ovary cell mutant that was identified through a genetic screen and phenocopies the NPC1 mutation. We show that the M87 mutant harbors a mutation in a gene distinct from the NPC1 and HE1/NPC2 disease genes. M87 cells have increased total cellular cholesterol with accumulation in an aberrant compartment that contains LAMP-1, LAMP-2, and NPC1, but not CI-MPR, similar to the cholesterol-rich compartment in NPC mutant cells. We demonstrate that low-density lipoprotein receptor activity is increased 3-fold in the M87 mutant, and likely contributes to accumulation of excess cholesterol. In contrast to NPC1-null cells, the M87 mutant exhibits normal rates of delivery of endosomal cholesterol to the endoplasmic reticulum and to the plasma membrane. The preserved late endosomal function in the M87 mutant is associated with the presence of NPC1-containing multivesicular late endosomes and supports a role for these multivesicular late endosomes in the sorting and distribution of cholesterol. Our findings implicate cholesterol overload in the formation of an NPC-like compartment that is independent of inhibition of NPC1 or HE1/NPC2 function.

recessive lipid storage disorder characterized by cholesterol accumulation in the liver, spleen, and central nervous system (1). Fibroblasts from NPC patients demonstrate lysosomal sequestration of low-density lipoprotein (LDL) cholesterol, delayed down-regulation of the LDL receptor and de novo cholesterol biosynthesis, and delayed up-regulation of cholesterol esterification (2). Genetic studies have identified two NPC complementation groups by linkage analysis and heterokaryon cell fusion studies (3,4). The major disease locus, the NPC1 gene, encodes a 1278-amino acid protein that has 13 predicted membrane-spanning domains (5), five of which share sequence homology with the putative sterol-sensing domains of HMG-CoA reductase, sterol regulatory binding protein (SREBP) cleavage-activating protein (SCAP), and Patched (6). The minor disease locus, the HE1/NPC2 gene, has recently been identified and encodes a 151-amino acid protein with cholesterol binding properties (7,8).
NPC1 and HE1/NPC2 are key participants in sterol trafficking. Cells with mutations in NPC1 and HE1 exhibit impaired delivery of internalized membrane cholesterol to the endoplasmic reticulum (ER) (9 -11). NPC1 mutants also demonstrate a defect in the mobilization of endosomal cholesterol to the plasma membrane (PM). The NPC1 protein localizes to a vesicular compartment that is lysosome-associated membrane protein (LAMP)-2-positive, Rab7-positive, and cation-independent mannose 6-phosphate receptor (CI-MPR)-negative (12,13), consistent with a role in endocytic trafficking. Live cell imaging of cells expressing NPC1-GFP demonstrate that NPC1 trafficks from this compartment by means of rapidly moving tubular extensions and subsequent budding of NPC1-containing vesicles (13,14). The HE1/NPC2 protein has also been shown to reside in a late endocytic compartment, though the precise immunolocalization of HE1/NPC2 has not yet been described (7).
NPC mutants are characterized morphologically by marked accumulation of free cholesterol in an enlarged aberrant compartment that bears both late endosomal and lysosomal markers and is the hallmark of NPC disease (13). In NPC1-deficient cells, impaired sterol trafficking leads to cholesterol accumulation in late endocytic organelles. Several observations indicate that cholesterol serves not only as bulk endocytic cargo, but also in the regulation of late endosomal function. Recent studies have shown that cellular cholesterol enrichment retards retroendocytic clearance of fluid-phase markers (12) and inhibits tubulovesicular trafficking from NPC1-containing endosomes (15). In addition, cholesterol loading has been shown to increase association of the NPC1 protein with membrane rafts in late endocytic organelles and has been proposed to disrupt sterol trafficking by promoting raft overcrowding in the normally raft-poor late endosomes (16). These findings suggest that in addition to lack of a functional NPC1 protein, excess free cholesterol may promote formation of the aberrant NPC compartment.
To gain further insight into the role of cholesterol in the cellular pathology in NPC disease, we studied a novel Chinese hamster ovary (CHO) cell mutant that morphologically phenocopies the NPC1 mutation. The M87 mutant was identified through a functional genetic screen in which mutants that accumulated unesterified endosomal/lysosomal cholesterol were isolated. The M87 cell line sequesters cholesterol in an NPC-like compartment, yet harbors a defect in a gene distinct from the NPC1 and HE1/NPC2 disease genes. In this report we show that in the M87 mutant late endosomal sterol trafficking is preserved, despite accumulation of excess cholesterol. Preserved late endocytic trafficking is associated with the presence of NPC1-positive multivesicular late endosomes (MVBs) and supports a role for NPC1 in MVB-dependent cholesterol sorting and trafficking. Moreover, our findings demonstrate that cholesterol overload contributes to formation of an NPC-like compartment that is independent of inhibition of NPC1 and HE1/ NPC2 function. For immunofluorescence studies the following antibodies were used: a rabbit anti-human NPC1 (raised against residues 1261-1278) (17); rabbit anti-LAMP-1 (E. Majerus, Washington University); mouse anti-LAMP-2 (UH3, Developmental Studies Hybridoma Bank); rabbit anti-CI-MPR (18); mouse anti-lysobisphosphatidic acid (LBPA) (J. Gruenberg, Univ. Geneva) (19).
Cell Lines-CHO-K1 cells were obtained from ATCC (CRL-9618). CT60 cells, a CHO cell line that harbors mutations in NPC1 and SREBP cleavage-activating protein, were provided by T.Y. Chang (Dartmouth College). M12 cells are a mutant CHO-K1 cell line with a deletion of the NPC1 locus (17). 2 The CHO/NPC1 cell line was generated by infection of CHO-K1 cells with retrovirus prepared by transient transfection of 293GPG packaging cells with the ⌬U3hNPC1 construct (21). Cells infected with the ⌬U3hNPC1 retrovirus were plated at limiting dilution and colonies were screened by Western blot analysis of microsomal fractions for human NPC1 expression. The NPC1 mutant human skin fibroblast cell line (GM3123) was provided by P. Pentchev (NIH). The NPC2 mutant human skin fibroblast cell lines, NPC-␤ and NPC83017, were provided by A. Fensom (United Medical and Dental Schools of Guy's and St. Thomas's Hospitals, London) and M. Vanier (Institut National de la Santé et de la Recherche Médicale, Lyon-Sud, France), respectively (3,4). Retrovirus for infection of CHO-K1 and human fibroblast cell lines was prepared by transient transfection of 293GPG packaging cells with the ⌬U3CFP and ⌬U3YFP constructs as previously described (21).
Mutagenesis and Isolation of Amphotericin B-resistant Cells with NPC Phenotype-CHO/NPC1 cells (1 ϫ 10 8 ) were harvested and resuspended in phosphate-buffered saline (PBS). Cells were divided into four aliquots of 1 ϫ 10 7 cells/ml per aliquot, exposed to 800 rads of ␥ irradiation, and plated at 1.6 ϫ 10 6 cells/150-mm dish in medium A. On day 6, cells were washed with PBS and refed medium C. On day 7, cells were refed medium C supplemented with 6 g/ml LDL. The following day cells were submitted to amphotericin B selection (24). Cells were incubated with medium D for 8 h and then washed with PBS and refed medium A. On day 14, surviving cells were harvested and replated as individual pools in 60-mm dishes, reserving a small aliquot of each pool for screening by filipin staining (25). Pools with the most frequent representation were subdivided and rescreened by filipin staining. After two rounds of subdivision, the pools were plated for limiting dilution, and clonal isolates were obtained.
Complementation Assay Using Heterokaryon CHO-Human Fibroblast Cell Fusions-Retrovirus was prepared by transient transfection of 293GPG packaging cells with the ⌬U3CFP and ⌬U3YFP constructs, respectively (21). CT60 and M87 cells were infected with ⌬U3YFP virus, and NPC1 and NPC2 human skin fibroblasts were infected with ⌬U3CFP virus. On day 0, CHO cell lines expressing YFP and human fibroblasts expressing CFP were plated together at 3 ϫ 10 5 and 6 ϫ 10 5 cells, respectively, in 60-mm dishes in medium E. On day 1, the cells were fused as described above and refed medium E. On day 2, fusions were plated in medium E on glass coverslips. The cells were filipin-stained and examined by fluorescence microscopy on a Zeiss Axiovert epifluorescence microscope using the following filter sets (Chroma): filipin (excitation 360/40 nm, emission 460/50 nm); CFP (excitation 436/20 nm, emission 480/40 nm); and YFP (excitation 500/20 nm, emission 535/30 nm). Heterokaryon cell fusions, defined by the presence of 2 or greater discrete nuclei per cell and by the presence of both CFP and YFP fluorescence, were scored for complementation of the NPC mutant phenotype. A minimum of 75 heterokaryons was counted per fusion, and only completely complemented fusions were scored as filipin-negative.
Labeling with Fluorescent Dextran-Cells were incubated with lysine-fixable Oregon Green-dextran (molecular weight 10,000; Molecular Probes) for 18 h at 37°C in medium A containing 1 mg/ml fluorescent dextran, washed, and chased for 4 h in the medium A lacking the fluorescent dextran. Cells were chilled on ice, fixed, and prepared for fluorescence microscopy as described above.
Measurement of Cellular Cholesterol-On day 0, cells were plated in triplicate (1.3 ϫ 10 5 cells/35-mm well) in medium A. On day 2, lipids were extracted from cells as previously described (17). Total cell cholesterol was determined by an enzymatic method using the Cholesterol CII Kit (Wako) and is expressed as g of total cellular cholesterol/mg cell protein. Protein determination was performed using the MicroBCA assay (Pierce).
Cholesterol Efflux Assay-On day 0, cells were plated in triplicate (5 ϫ 10 4 cells/35-mm well) in medium A. On day 1, cells were washed three times with PBS and refed medium B. On day 3, the cells were fed 20 g/ml [ 3 H]CL-LDL in medium B plus 20 g/ml progesterone for 24 h (protocol A) or 20 g/ml [ 3 H]CL-LDL alone in medium B for 30 min (protocols B and C). Following the [ 3 H]CL-LDL pulse, the cells were washed three times with PBS and incubated with medium B plus 2% CD for up to 2 h (protocol A) or 30 min (protocol B). Alternatively, after labeling the cells were fed medium B for up to 24 h and incubated with CD in medium B for 30 min (protocol C). Lipids were extracted from the media and the cells and analyzed as previously described (17).
Cholesterol Esterification Assays-Cholesterol esterification assays were performed as previously described (17). For measurement of 25-HC-stimulated cholesterol esterification, cells were fed medium containing 2 g/ml 25-HC and 10 g/ml cholesterol instead of LDL and pulsed with [ 3 H]oleate for two h, and lipids extracted as described. A chromatography recovery standard was added (30 g of cholesteryl oleate, 30 g of triolein, 0.0005 Ci of [ 14 C]cholesteryl oleate), and samples were dried under nitrogen. The lipids were separated by thinlayer chromatography (PE SIL G plates from Whatman) using heptane/ ethyl ether/acetic acid (90:30:1) and visualized with iodine. [ 3 H]cholesteryl oleate was quantified by liquid scintillation counting. After lipid extraction, monolayers were incubated with 0.1 N NaOH, and protein determination performed using the MicroBCA assay.
LDL Receptor Binding Assay-On day 0, cells were seeded in triplicate in 24-well plates (1.4 ϫ 10 4 cells/well) in medium A. On day 1, cells were washed twice with PBS and fed medium B. On day 3, cells were fed 0 -40 g/ml LDL in medium B. On day 4, cells were pulsed with 10 g/ml 125 I-LDL in 1:1 DMEM:Ham's F12 with 10% (v/v) human LPDS (Intracell) for 5 h. Binding of 125 I-LDL was measured as described (2,26) and is expressed as nanomoles of 125 I-LDL that is surface-bound per milligram of total protein.

Mutagenesis and Isolation of Amphotericin B-resistant Cells
with Endosomal/Lysosomal Accumulation of Unesterified Cholesterol-We employed a functional genetic screen to isolate novel CHO cell mutants that accumulate unesterified endosomal/lysosomal cholesterol. All previously described CHO mutants that accumulate unesterified endosomal cholesterol have resulted from defects in NPC1 (24,27,28). We reasoned that this is due to the presence of only a single copy of the NPC1 gene in CHO cells. Southern blotting of genomic DNA from wild-type CHO cells using a panel of restriction enzymes and hybridization with a murine NPC1 probe reveals a single band in all CHO cell genomic digests supporting this hypothesis. 2 Thus, the NPC1 gene in CHO cells may be inactivated by a single mutagenic event. Furthermore, the NPC1 locus may lie within a hypermutable region of the CHO cell genome.
To avoid preferential isolation of NPC1 mutants and to enrich for mutations in other genes, we used retroviral transduction to introduce additional copies of the human NPC1 gene into CHO cells (CHO/NPC1 cells). Wild-type CHO cells were infected with retrovirus-encoding human NPC1 and plated at limiting dilution to obtain a stable clonal cell line (CHO/NPC1) (17). Southern blotting of genomic DNA digests from the CHO/NPC1 cells demonstrates that these cells harbor three independent proviral integrations (data not shown). Multiple independent mutagenic events would be required to inactivate the endogenous CHO locus and each of the exogenously expressed NPC1 loci. Therefore, we hypothesized this strategy would minimize the potential for isolation of NPC1 mutants in our genetic screen and enrich for mutations in other genes. A similar strategy was successfully employed in a previous study for isolation of an SREBP site 1 protease (S1P) CHO mutant (29).
We then mutagenized 7 ϫ 10 7 CHO/NPC1 cells with ␥ irradiation and used a loss-of-function assay to select for mutagenized CHO cells with defects in endosomal cholesterol sorting and trafficking (24). Six days after mutagenesis, lipoproteins were removed from the media and an HMG CoA reductase inhibitor was added to inhibit endogenous cholesterol biosynthesis. Twenty-four h later the cells were fed 6 g/ml LDL and the following day exposed to 200 g/ml amphotericin for 8 h. This toxic polyene antibiotic binds plasma membrane cholesterol, creating non-selective ion pores that cause cell death (30,31). Cells depleted of plasma membrane cholesterol resulting from defective LDL cholesterol trafficking (e.g. NPC mutants) are resistant to amphotericin killing. Surviving cells were harvested from 46 independent pools 1 week after amphotericin selection, and each pool screened by staining with filipin, a specific fluorescent marker of unesterified cholesterol (32). In cells with defects in sterol transport, filipin binds unesterified cholesterol that has accumulated in endosomes/lysosomes, producing a characteristic UV fluorescence pattern. Several pools were identified with CHO mutants that exhibited prominent endosomal/lysosomal filipin staining. To enrich for the filipin-positive CHO mutants, pools with the most frequent representation were subdivided and rescreened by filipin staining. After two rounds of subdivision, cells were plated at limiting dilution, and the M87 cells, which demonstrate filipin-staining characteristic of the NPC mutant phenotype, were isolated (Fig. 1). In addition to the M87 cell line, two other independently isolated mutants were shown by heterokaryon cell fusion analysis to be in the same genetic complementation group (data not shown).
Identification of a Novel CHO Mutant with a Defect in a Gene Distinct from NPC1 and HE1/NPC2-To genotype the M87 cell line, we performed heterokaryon cell fusion analysis between the M87 cells and fibroblasts from NPC1 and HE1/NPC2 patients. Because of the low efficiency of fusion between CHO cells and human fibroblasts and the difficulty in distinguishing heterotypic fusions (e.g. CHO/fibroblast) from homotypic (e.g. CHO/CHO or fibroblast/fibroblast) fusions, we developed a new method for specific identification of CHO/human fibroblast heterokaryons. For these studies we expressed the yellow variant of GFP (YFP) in M87 cells and in NPC1-null CT60 cells. We expressed the cyan variant of GFP (CFP) in the NPC1 and NPC2 fibroblasts. Heterokaryon cell fusions were identified by the presence of both CFP and YFP fluorescence (Fig. 2, see  arrows). The CFP and YFP markers were highly specific and allowed discrimination of heterokaryons from homotypic fusions or non-fused fibroblasts (see arrowheads). For CT60/NPC1 het- erokaryons, we observed 0% filipin-negative polykaryons, whereas for CT60/NPC2 heterokaryons we observed 88% filipinnegative polykaryons (Table I). These findings are consistent with the known NPC1 defect in the CT60 cells. M87 cells, on the other hand, were able to complement both NPC1 (86% filipinnegative) and NPC2 cells (88% filipin-negative), indicating that the NPC phenotype in the M87 mutant cell line results from a mutation in a non-NPC1, non-HE1/NPC2 gene ( Fig. 2; Table I). These findings were confirmed by cDNA complementation studies, in which expression of NPC1 and HE1/NPC2 in M87 cells failed to correct the mutant phenotype (data not shown). Additionally, we found that M87 cells express normal levels of acid sphingomyelinase, which is deficient in Niemann-Pick A and B disease (33). 2 Unesterified Cholesterol Accumulates in an Aberrant Late Endocytic Compartment in M87 Cells-To characterize the sterol trafficking defect in the M87 mutant, we first performed morphologic studies of the aberrant cholesterol-containing structures in these cells using confocal immunofluorescence microscopy. We stained both the parental CHO/NPC1 and M87 cell lines with markers for endocytic organelles. In the CHO/ NPC1 cells, LAMP-1 and LAMP-2 showed perinuclear staining and co-localized with cholesterol, but not with CI-MPR (Fig. 3, A-C; for LAMP-1, data not shown). In contrast, in the M87 cells LAMP-1 and LAMP-2 markers redistributed to numerous enlarged cholesterol-containing structures, which were CI-MPRnegative (Fig. 3, D-F; for LAMP-1, data not shown). NPC1 co-distributed in the perinuclear region with cholesterol and LAMP-2 in CHO/NPC1 cells (Fig. 3, G-I), but redistributed in M87 cells to the aberrant LAMP-2-positive, cholesterol-rich structures (Fig. 3, I-L). A subset of these aberrant structures also co-localized with LBPA (data not shown). We also examined in M87 cells uptake of fluorescent dextran, a fluid phase endocytic marker. In pulse-chase studies, the fluorescent dextran co-localized exclusively with cholesterol and NPC1 in the aberrant compartment (Fig. 4). Taken together, these findings demonstrate by morphologic and functional criteria that the cholesterol-rich structure in M87 cells is a terminal late endocytic compartment, consistent with a late endosomal/lysosomal hybrid organelle (34). The cholesterol accumulation in the M87 mutant bears strong similarity to the NPC phenotype (13). Moreover, the M87 mutant, like NPC1-null cells, exhibits increased total cellular cholesterol (1.4-fold and 2.5-fold, respectively), as compared with wild-type CHO cells (Table II).
Localization of NPC1 to the Aberrant Compartment in M87 Cells Is Cholesterol-dependent-In the M87 cells wild-type NPC1 localized to the surface of the cholesterol-laden structures (Fig. 3L, arrowheads). This distribution for NPC1 contrasts with the lumenal staining pattern of late endocytic structures seen in the parental cell line (Fig. 3I). To examine the relationship between cellular cholesterol content and the localization of NPC1 to the aberrant compartment, M87 cells were cultured in lipoprotein-deficient media for 4 days to deplete endosomal cholesterol and then maintained in lipoproteindeficient media with or without 50 g/ml LDL for an additional 24 h. Strikingly, in lipoprotein-starved M87 cells, the cholesterol-rich aberrant compartment was absent, and the filipin   and NPC1 staining patterns reverted to that observed in the parental CHO/NPC1 cells (Fig. 5, A-B). In response to LDL feeding, the aberrant compartment was reestablished with concomitant redistribution of NPC1 to the limiting membrane of this structure (Fig. 5, C-D). These results support that the altered localization of NPC1 in the M87 mutant is a direct consequence of cellular cholesterol overload.
NPC1 Localizes to Multivesicular Bodies and Late Endosome-Lysosome Hybrid Organelles-To further characterize the late endocytic localization of NPC1 and its altered distribution in the M87 mutant, we conducted ultrastructural studies in the parental CHO/NPC1, M87, and NPC1-null M12 cell lines (17). Immunogold labeling was performed with an affinitypurified rabbit polyclonal antibody against NPC1 (17). EM analysis revealed that in the parental CHO/NPC1 cells, NPC1 immunogold labeled endosomal structures that display numerous small intraluminal vesicles characteristic of MVBs (35) (Fig. 6A). These structures labeled with NPC1 immunogold at both the limiting membrane and the membranes of the intraluminal vesicles. NPC1-immunogold labeled similar structures in M87 cells (Fig. 6B). In both CHO/NPC1 cells (not shown) and M87 cells, the MVB compartment also labeled with LAMP-2 immunogold, consistent with previously observed late endosomal distribution for NPC1 (12,13) (Fig. 6C). In addition to the MVB compartment, M87 cells showed prominent NPC1 and LAMP-2 immunogold labeling at the limiting membrane of a vacuolated endosomal compartment containing electron-dense inclusions and multlamellar structures (Fig. 6D). This hybrid organelle compartment corresponds to the cholesterol-rich compartment identified by our immunofluorescence studies (Fig. 3L). In NPC1-null cells, a similar LAMP-2-positive structure was observed (Fig. 6E). No LAMP-2-positive multivesicular structures were seen by immuno-EM in the NPC1-null cells.
Late Endosomal Cholesterol Trafficking Is Preserved in the M87 Cells-While M87 and NPC mutant cells both exhibit marked sterol accumulation in a late endosomal/lysosomal hybrid organelle, the M87 cell line differs in that it expresses   We used three different protocols to examine delivery of LDL cholesterol to the PM in the M87 mutant, and results were compared with that of parental CHO/NPC1 cell line. First, we used a well established method in which [ 3 H]CL-LDL cholesterol is allowed to accumulate in a terminal late endocytic compartment overnight in the presence of progesterone (10,17). After a washout phase, cholesterol delivery to the PM is measured by incubating cells with ␤-CD, an efficient non-permeable free cholesterol acceptor. In this assay no difference in cholesterol efflux was observed between the M87 cells and the parental CHO/NPC1 cells (Fig. 7A). This method for measuring cholesterol trafficking, however, may not be able to detect the delivery of newly hydrolyzed LDL cholesterol versus internalized membrane cholesterol to the PM. To specifically examine the trafficking of newly hydrolyzed LDL cholesterol to the PM, cells were pulsed with [ 3 H]CL-LDL and then chased for up to two h with medium containing CD (36). No difference was observed between the mutant and parental cells in the initial arrival of LDL cholesterol to the PM (Fig. 7B). We then examined the post-PM trafficking of LDL cholesterol. In this protocol, cells are pulsed with [ 3 H]CL-LDL and then chased for up to 24 h. Cholesterol efflux is measured by 30-min incubation with medium containing CD (37). Again, there was no cholesterol trafficking defect in the mutant cell line (Fig. 7C). Thus, using three different methods, we find that M87 cells, in contrast to the NPC1-deficient cells, do not have a block in the delivery of late endosomal cholesterol to the PM.
Since there is also evidence that NPC1 participates in late endosomal trafficking of cholesterol to the ER, we measured cholesterol delivery to the ER by comparing LDL-stimulated cholesterol esterification rates in wild-type CHO, NPC1-null M12, CHO/NPC1 and M87 cells. These cells were incubated in medium containing 5% LPDS, fed LDL overnight, and the following day pulsed for two h with [ Fig. 8A as a percentage of the rate observed for wild-type CHO cells. NPC1-null cells demonstrate severely impaired LDL-stimulated cholesterol esterification rates (15% of wild-type CHO cells), while the CHO/NPC1 cells exhibit modestly increased rates of cholesterol esterification (17). Despite the marked accumulation of unesterified cholesterol, M87 cells demonstrate normal esterification rates indicating that postendosomal cholesterol trafficking to the ER is intact in this cell line. To exclude the possibility that the normal esterification rates of LDL cholesterol in M87 cells reflects up-regulation of ACAT activity, we measured the ACAT activity in this mutant in response to stimulation by 25-HC (38). We found that 25-HC stimulated cholesterol esterification to the same extent in wild-type CHO, M12, parental CHO/NPC1, and M87 cells (Fig.  8B), indicating that the M87 mutant phenotype is not due to alteration in ACAT activity.
As an independent measure of cholesterol delivery to the ER, we examined in the M87 cells the ability of LDL feeding to suppress LDL receptor (LDL-R) activity. In this study we measured the surface-bound 125 I-LDL in wild-type CHO, M12, and M87 cells in response to incubation with graded amounts of LDL (0 -40 g/ml) (Fig. 8C). As expected, M12 cells exhibit delayed suppression of LDL-R binding consistent with the known defect in NPC1 mutants in delivery of cholesterol to the ER (2). By comparison, M87 cells respond in an identical manner to wild-type CHO cells to suppression of LDL-R binding at low LDL concentrations, demonstrating a normal cholesterol homeostatic response. In contrast to wildtype CHO cells, the M87 mutant shows a 3-fold increase in the absolute level of LDL-R binding (Fig. 8D), indicating an increase in number of LDL-R at the cell surface at steady state. However, Western blotting did not show an increase in expression of the LDL-R in M87 cells (data not shown). Taken together, the efflux, cholesterol esterification, and LDL-R activity studies demonstrate that late endosomal trafficking of cholesterol is intact in the M87 mutant.

DISCUSSION
In this report we examine a role for cholesterol in regulation of late endocytic function through the study of a novel CHO mutant that morphologically phenocopies the NPC1 mutation. We show that the M87 mutant accumulates free sterol in a late endosomal/lysosomal hybrid organelle, despite normal late endosomal function and the presence of NPC1-containing late endosomal MVBs. We further show that cholesterol accumulation in the mutant is associated with a marked increase in LDL receptor activity. Our findings demonstrate that cholesterol overload promotes formation of an NPC-like compartment, even in the context of normal NPC1 and HE1/NPC2 function.
The etiology of the cholesterol-rich hybrid organelle in NPC mutants is not well understood. Endosomes and lysosomes are dynamic structures that are known to exchange content and membrane markers (39). Several models have been proposed to explain content mixing between these compartments including vesicular transport, "kiss and run", and direct fusion between late endosomes and lysosomes. In the latter model, dense-core lysosomes, containing proteases and acid hydrolases, fuse directly with late endosomes to degrade endocytosed material. The lysosomes are then reformed from the resultant hybrid organelle and undergo subsequent cycles of fusion and reformation (34). In NPC mutants, the accumulated free sterol in the hybrid organelle may block the dissociation of the normally transient interaction between late endosomes and lysosomes, leading to the aberrant cholesterol-laden hybrid structures (12,13). Indeed, cholesterol accumulation in NPC1 mutants has been shown to inhibit rapid exchange of contents between late endocytic compartments (15). In contrast to NPC mutants, the M87 cell line, which expresses both NPC1 and HE1/NPC2 proteins, has no demonstrable sterol trafficking defect and accumulates free sterol in response to increased LDL-R activity. The increase in LDL-R binding, in the absence of increased LDL-R expression, suggests altered posttranslational regulation of the LDL-R in the mutant. We hypothesize that the aberrant compartment in M87 cells results from excessive internalization of LDL cholesterol that exceeds the capacity for disposal at the level of the NPC1-containing late endosomes and leads to cholesterol accumulation. Our finding that loading with LDL cholesterol results in induction of the aberrant compartment lends support to this hypothesis. A similar mechanism has been proposed for the etiology of cholesterol-laden NPC compartment in NPC1-null cells (40). Alternatively, the M87 mutant may harbor a defect in regulatory mechanisms that mediate efficient recycling of cholesterol from early endocytic compartments and prevent entry of cholesterol into the distal endocytic pathway (15).
The sterol accumulation in M87 and NPC1-null cells bears striking similarity to the phenotype observed in cells treated with U18666A, a hydrophobic amine that inhibits mobilization of endosomal cholesterol (9). U18666A has been shown to inhibit the dynamic movements of NPC1-containing vesicles and promote trapping of cholesterol in aberrant perinuclear or- ganelles, reminiscent of the NPC1 mutant phenotype (14). It has been proposed that U18666A induces formation of the aberrant endosomal compartment due to its ability to bind negatively charged membrane phospholipids, such as LBPA and thus interferes with morphogenesis of MVBs (41). In the present study we demonstrate for the first time by immuno-EM that NPC1 specifically localizes to MVBs and is sorted to intraluminal vesicles within MVBs. The latter finding is consistent with the observed lumenal distribution of NPC1 in late endosomes in wild-type cells (12). In light of the known function of NPC1 in sterol trafficking, the finding that NPC1 is a resident MVB protein provides additional evidence to support a role for late endosomal MVBs as a platform for the sorting and distribution of cholesterol.
In the M87 mutant the presence of NPC1-and LAMP-2containing MVBs correlates with preserved late endosomal function. Similar multivesicular endosomal structures are notably absent in NPC1-null cells in our immuno-EM studies. These findings implicate a role for NPC1 in MVB-dependent cholesterol trafficking that is proximal to the cholesterol-rich hybrid organelles, which are terminal endocytic compartments in the M87 and NPC1 mutants (13,14). In contrast to the NPC1-null cells, accumulation of excess endocytosed cholesterol does not appear to retard NPC1-mediated sterol trafficking in the M87 cells. It is possible that the overexpressed NPC1 protein in the M87 cell line may be able to overcome inhibition caused by the elevated endosomal cholesterol content. This explanation, however, seems unlikely since in a previous study overexpression of NPC1 in NPC1-null CHO cells was unable to restore endosomal trafficking until cholesterol was cleared from the hybrid organelles (15). Another possibility is that endosomal cholesterol content in the M87 mutant (Table II) is sufficiently elevated to promote formation of the aberrant hybrid organelle, but not elevated enough to impair NPC1-mediated sterol trafficking. A third possibility is that the mutation in M87 specifically inhibits sterol regulation of NPC1 activity.
Cholesterol trafficking through the late endocytic pathway is a complex and highly regulated process. Our findings with the M87 mutant help clarify the role of cholesterol in formation of the NPC compartment. In future studies examination of mechanisms that regulate endosomal cholesterol content, late endosomal sterol trafficking, and the morphogenesis of NPC1-containing MVBs will further our understanding of how cholesterol accumulation contributes to the pathogenesis of NPC disease.