Vesicular and non-vesicular sterol transport in living cells. The endocytic recycling compartment is a major sterol storage organelle.

We examined the intracellular transport of sterol in living cells using a naturally fluorescent cholesterol analog, dehydroergosterol (DHE), which has been shown to mimic many of the properties of cholesterol. By using DHE loaded on methyl-beta-cyclodextrin, we followed this cholesterol analog in pulse-chase studies. At steady state, DHE co-localizes extensively with transferrin (Tf), a marker for the endocytic recycling compartment (ERC), and redistributes with Tf in cells with altered ERC morphology. Expression of a dominant-negative mutation of an ERC-associated protein, mRme-1 (G429R), results in the slowing of both DHE and Tf receptor return to the cell surface. [3H]Cholesterol is found in the same fraction as 125I-Tf on sucrose density gradients, and this fraction can be specifically shifted to a higher density based on the presence of horseradish peroxidase-conjugated Tf in the same organelle. Whereas vesicular transport of Tf and efflux of DHE from the ERC are entirely blocked in energy-depleted cells, delivery of DHE to the ERC from the plasma membrane is only slightly affected. Biochemical studies performed using [3H]cholesterol show that the energy dependence of cholesterol transport to and from the ERC is similar to DHE transport. We propose that a large portion of intracellular cholesterol is localized in the ERC, and this pool might be important in maintaining cellular cholesterol homeostasis.

I-Tf on sucrose density gradients, and this fraction can be specifically shifted to a higher density based on the presence of horseradish peroxidase-conjugated Tf in the same organelle. Whereas vesicular transport of Tf and efflux of DHE from the ERC are entirely blocked in energy-depleted cells, delivery of DHE to the ERC from the plasma membrane is only slightly affected. Biochemical studies performed using [ 3 H]cholesterol show that the energy dependence of cholesterol transport to and from the ERC is similar to DHE transport. We propose that a large portion of intracellular cholesterol is localized in the ERC, and this pool might be important in maintaining cellular cholesterol homeostasis.
As an essential constituent of biological membranes, cholesterol plays various roles in the functions of membranes and membrane-regulated processes in mammalian cells. Cholesterol is a major component of lipid rafts, which are postulated to play an important role in cellular functions such as signaling, adhesion, motility, and membrane traffic (1)(2)(3)(4)(5).
Cholesterol is distributed unevenly among various intracellular membranes. The endoplasmic reticulum (ER) 1 and mitochondrial membranes contain very little cholesterol (6). By using various methods in different cell types, the plasma membrane has been estimated to contain 50 -90% of cellular cholesterol. Estimating the precise cholesterol content in the plasma membrane has been a source of controversy because of limitations of assay techniques (7,8). One popular technique for determining plasma membrane cholesterol is measuring susceptibility to cholesterol oxidase treatment (9 -11). However, this method may overestimate plasma membrane cholesterol to the extent that intracellular cholesterol becomes oxidized (10). Another widely used technique is subcellular fractionation followed by lipid composition analysis. By using a well characterized method based on magnetic affinity chromatography to purify membranes, the plasma membrane was estimated to contain half of the total cellular cholesterol in CHO cells (12).
The balance between plasma membrane and internal cholesterol could be affected, in part, by the relative rates of endocytosis versus secretion and recycling (13). Although the overall lipid composition in early endosomes may be similar to that of the plasma membrane (14), there is experimental evidence that various lipids are sorted differentially at multiple steps in the endocytic pathways (15)(16)(17). This sorting would lead to different lipid compositions in the various endocytic organelles. Analysis of recycling endosomes from rat livers by gas-liquid chromatography indicates that these endosomes are enriched in cholesterol (18).
Due to the importance of cholesterol in cell function, it is crucial that the intracellular transport of cholesterol is tightly regulated. Cholesterol is synthesized in the ER and subsequently transported to the plasma membrane in an ATP-dependent process that is independent of the secretory pathway (19 -22). Cells also obtain cholesterol by receptor-mediated endocytosis of low density lipoproteins (23). Once delivered to late endosomes or lysosomes, the cholesteryl esters in low density lipoprotein are hydrolyzed, and free cholesterol is rapidly cycled back to the plasma membrane (24,25). Excess cholesterol is esterified by acyl CoA:cholesterol acyltransferase. Cholesterol cycles constitutively between the cell interior and the plasma membrane (26). Cholesterol movement from the plasma membrane to the ER is inhibited by disruption of the cytoskeleton and acidic compartments (27) but not by ATP depletion (28).
The endocytic recycling pathway has been well characterized in CHO cells (29 -31). After internalization via clathrin-coated pits, transferrin (Tf) bound to Tf receptor, and other recycling molecules are rapidly delivered to the sorting endosomes. They are then recycled directly to the plasma membrane (32) or transported to the endocytic recycling compartment (ERC) and subsequently recycled backed to the plasma membrane. The recycling pathway, along with other mechanisms, ensures that the plasma membrane maintains its lipid and protein composition. Although significant advances have been made in studying cholesterol distribution and trafficking in cells, most methods used are indirect, which can lead to different interpretations (reviewed in Ref. 7). Fluorescent cholesterol analogs offer a unique opportunity for direct imaging of sterol trafficking in living cells. Unlike many synthetic cholesterol analogs tagged with bulky fluorophores that do not mimic the behavior of cholesterol, dehydroergosterol (DHE) is a naturally occurring sterol whose chemical structure closely resembles that of cholesterol. It differs from cholesterol only in having three additional double bonds and an extra methyl group (33). Many studies have shown that DHE is a good cholesterol analog (33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(47), in terms of its ability to faithfully mimic the function and behavior of cholesterol in model and biological membranes. A major obstacle in DHE imaging lies in the fact that DHE absorbs and emits in the UV region of the spectrum. A microscope having optical components with high UV throughput and a camera with a high quantum efficiency in the UV region enables us to perform microscopic imaging of this cholesterol analog (34).
Loading DHE onto methyl-␤-cyclodextrin (M␤CD) and thus creating a soluble DHE-M␤CD complex makes it possible to deliver DHE to the plasma membrane with an efficiency high enough to carry out short pulse-chase experiments. In this study, we perform microscopic imaging of DHE, in combination with biochemical studies using [ 3 H]cholesterol, to examine the intracellular distribution and trafficking of cholesterol in CHO cells. Because we can directly visualize DHE, we are able to identify the ERC as a major intracellular compartment for sterol and subsequently characterize the transport pathways between the plasma membrane and the ERC. All of our major results obtained with DHE are also confirmed by biochemical studies using [ 3 H]cholesterol.
Cells-TRVb-1 is a modified CHO cell line that lacks endogenous Tf receptor and expresses the human Tf receptor (48). Cells were grown at 37°C in a 5% CO 2 humidified incubator and in bicarbonate-buffered Ham's F-12 medium supplemented with 5% fetal bovine serum, 200 g/ml geneticin as a selection for the transfected Tf receptors, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 g/liter glucose. Green fluorescent protein (GFP)-tagged mouse Rme-1 (mRme-1) dominantnegative mutant (G429R) construct was transiently transfected into TRVb-1 cells using the LipofectAMINE reagent system (Invitrogen) as described previously (49). Cells for microscopy were plated 2 days before the experiments in 35-mm plastic tissue culture dishes with a 7-mm hole in the bottom covered by poly-D-lysine-coated coverslips (50).
Loading DHE on M␤CD-DHE-M␤CD complexes were prepared similarly to a procedure described previously (51), replacing cholesterol with DHE. In brief, DHE in ethanol was dried under argon and subsequently dissolved in M␤CD in Medium 1, making the initial ratio of M␤CD to DHE 8:1 (mol:mol). The resulting suspension was vortexed and sonicated until it clarified. It was then incubated in a rocking water bath overnight at 37°C and centrifuged to remove insoluble DHE aggregates. Final concentration of DHE used for labeling cells was 0.5 to 1 mM.
Fluorescence Microscopy-Fluorescence microscopy and digital image acquisition were carried out using a Leica DMIRB microscope (Leica Mikroscopie und Systeme GmbH, Germany) equipped with a Princeton Instruments (Princeton, NJ) cooled CCD camera driven by Image 1/MetaMorph Imaging System software (Universal Imaging Corp.). All images were acquired using a high magnification oil immersion objective (63ϫ, 1.4 NA). To optimize the throughput of the microscope in the UV region of the spectrum, a lamp housing from Leica with a collecting lens that had high transmittance characteristics in the UV region was used. DHE was imaged using a filter cube obtained from Chroma Technology Corp. (Brattleboro, VT) (335-nm (20-nm bandpass) excitation filter, 365-nm longpass dichromatic filter, and 405-nm (40-nm bandpass) emission filter). The detection of UV fluorescence of DHE was made possible by the use of a camera with a back-thinned, UV-sensitive CCD chip (Princeton Instruments Frame Transfer Micro-MAX camera with a 512 ϫ 512 back-thinned EEV chip; model 512BFT). This camera has ϳ70% quantum efficiency at 400 nm. BODIPY FL C 5 -ceramide and Alexa 488-phalloidin were imaged using a standard fluorescein filter cube (470-nm (20-nm bandpass) excitation filter, 510-nm longpass dichromatic filter, and 537-nm (23-nm bandpass) emission filter), Alexa 546-Tf using a standard rhodamine filter cube (535-nm (50-nm bandpass) excitation filter, 565-nm longpass dichromatic filter, and 610-nm (75-nm bandpass) emission filter), and Alexa 633-Tf using a standard Cy5 filter cube (623-nm (31-nm bandpass) excitation filter, 655-nm longpass dichromatic filter, and 700-nm (35-nm bandpass) emission filter).
Fluorescence crossover was measured using single-labeled samples of each probe, and images were corrected for background (32) and crossover (15).

Quantification of Intracellular Distribution of DHE and [ 3 H]Cholesterol-
The following method was used to estimate the fraction of DHE that was in the ERC. The images were first background-corrected using extracellular intensity values (32). The cell outlines were traced out manually using the phase contrast images. The ERC was also manually outlined for each cell using the Tf-labeled images. Fluorescence power from the outlined ERC region (I ERC Ј ) and the entire cell (I cell ), as well as the areas of the two regions (S ERC and S cell , respectively), were measured. To calculate the fluorescence intensity in the plasma membrane (I PM ) while taking into consideration the fluorescence contributed by the regions of the plasma membrane that were directly above and below the ERC, Equation 1 was used, This was based on the assumption that the number of DHE molecules present in a certain region of the plasma membrane was proportional to its surface area on the plasma membrane in the wide-field micrographs. DHE fluorescence that was present in the ERC (I ERC ) was obtained by Equation 2, The ratio of DHE that was present in the ERC versus the plasma membrane is shown in Equation 3, This measurement was repeated for ϳ30 cells from experiments done on 2 different days.
To quantify the fraction of [ 3 H]cholesterol that was in the ERC versus on the plasma membrane from the density gradients, a ratio was taken of the areas under the lighter peak (the plasma membrane fractions) and the heavier peak (the intracellular fractions). This measurement was repeated for seven experiments from 3 different days.

Subcellular Fractionation of Cells Containing 125 I-Tf and [ 3 H]
Cholesterol-Cells were grown to confluency in 150 ϫ 25-mm tissue culture dishes. Control cells were incubated with M1glucose containing 1.2 g/ml 125 I-Tf, 24 g/ml HRP-Tf, and 1 Ci/ml [ 3 H]cholesterol for 1 h at 37°C. [ 3 H]Cholesterol was dissolved in ethanol to give a final ethanol concentration in the labeling medium less than 0.1%, v/v. To deplete energy, cells were first incubated with 125 I-Tf and HRP-Tf for 1 h at 37°C, incubated with ED medium for 30 min, and then with ED medium containing [ 3 H]cholesterol for 1 h at 37°C. For experiments designed to test the efficiency of energy depletion, cells were first incubated with ED medium for 30 min and then with ED medium containing 125 I-Tf for 1 h at 37°C. After labeling with radioactive probes, cells were chilled on ice, and surface-bound 125 I-Tf and HRP-Tf were removed by acid wash on ice. 0.4 mg/ml 3,3Ј-diaminobenzidine (DAB), 50 mM ascorbic acid, and 0.025% H 2 O 2 were added to half of the dishes (referred to as "density-shifted"), and only DAB and ascorbic acid were added to the other half (52,53). The dishes were incubated in the dark for 1 h on ice. Cells were harvested by incubation with EDTA on ice for 15 min. After removing EDTA, the cells were resuspended in 250 mM sucrose and 50 mM Tris, pH 7.4. The cells were then lysed through a ball-bearing cell breaker and centrifuged to remove the nuclei, and the postnuclear supernatant was layered onto a 10 -45% continuous sucrose gradient (54). The gradients were subjected to centrifugation at 137,000 ϫ g for 16 h at 4°C in a swinging bucket rotor (Sorvall, Newton, CT). The fractionated gradient and resuspended pellet were taken for radioactivity measurement. For experiments designed to identify the plasma membrane fraction, cells were chilled on ice and incubated with 125 I-Tf for 1 h on ice. The cells were then lysed and fractionated as described above.
[ 3 H]Cholesterol Efflux-Cells were grown to confluency in 150 ϫ 25-mm tissue culture dishes. Cells were incubated in M1glucose containing [ 3 H]cholesterol (control) or ED medium containing [ 3 H]cholesterol (energy-depleted) for 1 h at 37°C. They were then washed several times with Medium 1. Radioactivity from the last wash was measured to make sure that no [ 3 H]cholesterol remained in the bathing medium. 6 ml of chase medium (M1glucose with 10 mM M␤CD for control cells and ED medium with 10 mM M␤CD for energy depleted cells) was added. This was defined as chase time 0. At each time point, 600 l of chase medium was taken from the cell dish, and 600 l of fresh prewarmed chase medium was added back to the dish. At the end of the 60-min chase, remaining chase medium was removed, and cells were harvested by incubation with EDTA on ice.
[ 3 H]Cholesterol from the chase medium aliquots, remaining chase medium, EDTA wash, and harvested cells was counted by liquid scintillation. To obtain the kinetic curves, the following calculations were made: (a) total

DHE Is Distributed between the Plasma Membrane and the ERC at Steady
State-Cells were labeled with DHE-loaded M␤CD for 1 min and chased for various times (Fig. 1). DHE was incorporated in the plasma membrane during a 1-min pulse, and it was seen most clearly in the regions where two cells abut (Fig. 1A). Accumulation in the juxta-nuclear region was seen within 5 min of chase (Fig. 1B). We did not observe significant amounts of DHE leaving cells in the absence of an acceptor, and this was confirmed by quantification of images taken after different chase times (data not shown). By chasing DHE-labeled cells in serum-free medium without a sterol acceptor for different periods, we found that DHE reached a stable distribution after 30 -60 min that remained virtually unchanged during chase periods up to 18 h (Fig. 1, C-E). 35 Ϯ 12% of total cellular DHE was estimated to be in the ERC from these wide field micrographs (see "Experimental Procedures").
To see if reducing temperature has any effect on the delivery of DHE from the plasma membrane to the ERC, we incubated the cells with DHE-loaded M␤CD for 10 min on ice, chased for 3 h on ice, and fixed the cells for 30 min on ice. Fig. 1F shows that although the plasma membrane staining was not as bright as the cells labeled for 1 min at 37°C (Fig. 1A), DHE was nevertheless incorporated into the plasma membrane at 0°C. However, DHE delivery to the ERC was completely blocked. The flip-flop of DHE from the outer leaflet of the bilayer to the inner leaflet, as well as the desorption from the inner leaflet on to protein carriers, might be affected by lowering the temperature.
It was shown previously (34) that at steady state DHE is localized to a juxta-nuclear dot that overlaps with Tf, a marker for the ERC, and TGN-38, a marker for the trans-Golgi network (TGN). Both DHE and filipin showed a similar ERC/TGN distribution at steady state in CHO cells (34). Due to their close proximity in CHO cells, the ERC and the TGN often appeared to be extensively co-localized in wide field micrographs. However, there were cells in which the ERC was clearly distinguishable from the TGN. When the cells were triple-labeled with BODIPY FL C 5 -ceramide, a vital stain for the TGN ( Fig.  2A), DHE (Fig. 2B), and Alexa 633-Tf (Fig. 2C), DHE was found to be much more closely associated with the ERC than with the TGN. In fact, many cells, such as the ones shown in Fig. 2, displayed the DHE-containing compartment right next to or encircled by the TGN, with only partial overlap. This is more clearly demonstrated in the lower panels of Fig. 2. In Fig. 2D, DHE is pseudo-colored green and BODIPY FL C 5 -ceramide is red; in Fig. 2E, DHE is green and Tf is red; and in Fig. 2F, DHE is green, Tf is red, and BODIPY FL C 5 -ceramide is blue. A clear separation of the red and the green colors is seen in Fig. 2D, and the two colors are extensively, but not completely, co-localized to form yellow in Fig. 2E.
To demonstrate further that DHE is associated with the ERC, we show that DHE co-redistributed with ERC markers when the ERC morphology was altered (Fig. 3). The ERC dispersed into large aggregates near the cell periphery when cells were treated with the microtubule depolymerization drug, nocodazole. With this treatment, DHE was no longer seen in a tight juxta-nuclear dot, but rather, it co-localized with Tf (Fig.  3A) in the dispersed ERC aggregates (Fig. 3B). The arrows in Fig. 3, A and B, indicate examples of co-distribution of the two molecules.
Rme-1/EHD1 is an EH domain protein that is associated with the ERC in CHO cells (49). Expression of a dominantnegative mutant of mRme-1 (G429R) alters the ERC morphology so that it wraps partly around the nucleus, and the retention of Tf in the ERC is prolonged (49). We examined the distribution of DHE in cells that were transiently transfected with mRme-1 (G429R). The extended ERC morphology was demonstrated with GFP-mRme-1 (G429R) in a transfected cell as compared with neighboring untransfected cells (Fig. 3C), and the corresponding DHE staining (Fig. 3D) co-localized very well with the GFP.
Energy-independent Uptake of DHE to the ERC-Although DHE was delivered from the plasma membrane to the ERC within minutes, we could not detect DHE in endosomal intermediates, such as sorting endosomes, on the route to the ERC. To test further whether DHE was delivered to the ERC from the plasma membrane via a different pathway as Tf, we examined DHE transport in energy-depleted cells. As shown in Fig.  4, A and B, Tf and DHE were concentrated in the ERC after 30 min of internalization in control cells. If the cells were treated with energy poisons prior to labeling, Tf internalization was entirely blocked as expected (28). Tf remained exclusively on the plasma membrane (Fig. 4, C and D) and was removed completely by a mild acid wash (Fig. 4F). DHE uptake to the ERC, on the other hand, was only slightly affected (Fig. 4, E and G) by energy depletion.
We quantified the extent of DHE delivery to the ERC in control and energy-depleted cells (Fig. 4H). After internalization for different times, the fraction of DHE found in the ERC (relative to the rest of the cell) in energy-depleted cells was about 70% that in control cells. Because DHE was initially inserted into the plasma membrane from the cyclodextrin carrier, this indicated that there was significant transport of DHE to the ERC by energy-independent mechanisms.
To quantify the effects of energy depletion on transport rates, we measured the kinetics of DHE delivery to the ERC in control and energy-depleted cells using fluorescence recovery after photobleaching (Fig. 5). DHE fluorescence does not spontaneously recover after photobleaching, as tested by photobleaching an entire cell (data not shown). Cells were labeled to steady state with DHE before fluorescence recovery after photobleaching was performed. After photobleaching the ERC, DHE returned to the ERC with a recovery half-time of 2.5 Ϯ 0.6 and 2.6 Ϯ 0.8 min in control and energy-depleted cells, respectively. The fluorescence intensity ratio in the ERC to that in the entire FIG. 2. DHE distribution at steady state in CHO cells. TRVb-1 cells were first pulsed with DHE-loaded M␤CD for 1 min, chased in the presence of 5 g/ml Alexa 633-Tf for 1 h, and chased further in M1glucose with excess unlabeled iron-loaded Tf for 15 min. Finally, cells were pulsed with 1 M BODIPY FL C 5 -ceramide for 2 min and chased for 5 min. All the pulse-chase treatments were done at 37°C. Chasing DHE for different times, from 1 to 5 h, gave very similar distribution (images not shown). A-C show BODIPY FL C 5 -ceramide, DHE, and Alexa 633-Tf staining, respectively. D-F show pseudocolored images intended to demonstrate the extent of co-localization among the three markers. D, DHE is pseudo-colored green and ceramide red; E, DHE is green and Tf is red; and F, DHE is green; Tf is red; and ceramide is blue. Bar, 10 m.

FIG. 3. DHE distribution in nocodazole-treated cells and in cells expressing mRme-1 (G429R).
In the top panels, TRVb-1 cells were labeled with DHE for 1 min and chased in the presence of 2 g/ml Alexa 546-Tf for 1 h at 37°C. They were then chased for 20 min in the presence of 10 g/ml nocodazole and imaged in the presence of nocodazole. A and B show Tf and DHE staining, respectively. Arrows indicate some of the places where the two molecules co-localized. In the lower panels, TRVb-1 cells transiently transfected with GFP mRme-1 (G429R) were pulsed with DHE for 1 min and chased for 1 h at 37°C. C and D show GFP mRme-1 (G429R) and DHE staining, respectively. Bar, 10 m.

FIG. 4. Uptake of DHE from the plasma membrane to the ERC is energy-independent. The control cells (A and B)
were first incubated with M1glucose for 30 min and labeled with DHE-loaded M␤CD for 1 min at 37°C. They were then chased in M1glucose containing 5 g/ml Alexa 546-Tf for 30 min at 37°C and imaged in M1glucose. For energy-depleted cells (C-G), M1glucose was replaced by ED medium for all the steps described above. To remove surface-bound Tf, cells were washed twice with the acid wash and several times with Medium 1 (F and G). A, C, D, and F, Tf; B, E, and G, DHE. C and D show the bottom surface and middle focal planes of the same cell, respectively. H shows quantification of the extent of DHE delivery to the ERC in control and energy-depleted cells. To deplete energy, cells were incubated with ED medium for 30 min at 37°C. Cells were labeled with DHE-loaded M␤CD for 1 min and chased in M1glucose (control cells) or ED medium (energy-depleted cells) for different times (10, 20, and 30 min) at 37°C. To quantify, the images were first background corrected. Both the ERC and the entire cell were then manually outlined. A ratio of fluorescence intensity in the ERC to that in the entire cell was obtained for every cell. Each bar in H was derived from an average of 100 cells. Bar, 10 m. cell recovered to a substantial extent after 10 min. The fact that the two rates were similar indicates that the energy-independent uptake pathway is the predominant route by which DHE is transported to the ERC in CHO cells.
Energy-dependent Efflux of DHE from the ERC-To examine the efflux of DHE from the ERC to the plasma membrane, cells were first labeled with DHE for 1 min and chased in the presence of Tf for 1 h (Fig. 6, A and E). Both probes effluxed completely from cells after 1 h in normal chase medium that contained 5 mM cholesterol-loaded M␤CD and 5 mM M␤CD to promote exchange of DHE for cholesterol at the plasma membrane (Fig. 6, B and F). However, adding energy poisons to the chase medium prevented efflux of Tf and DHE from the ERC (Fig. 6, C and G). To show that the block in efflux was due specifically to energy depletion, we washed off the energy poisons and added glucose back to the cells chased in energy poisons (similar to ones in Fig. 6, C and G). The cells were then chased in normal chase medium for 1 h, and the probes were once again chased out almost completely (Fig. 6, D and H).
Retention of Tf and DHE in the ERC of mRme-1 (G429R) Expressing Cells-Expression of a mutant form of mRme-1, mRme-1 (G429R), leads to the retention of Tf in the ERC of CHO cells (49). To examine its effects on DHE recycling, cells transfected with GFP-mRme-1 (G429R) were first pulsed with DHE for 1 min and chased in the presence of Tf for 1 h. As shown in Fig. 7, A-D, the ERC in both transfected (indicated by GFP in B) and untransfected cells was equally brightly labeled with Tf and DHE. Cells were then chased for 1 h in chase medium, resulting in a complete chase-out in the untransfected cells. However, both Tf and DHE were retained in the transfected cells (Fig. 7, E-H). Because mRme-1 (G429R) is predominantly associated with the ERC in CHO cells, DHE retention again confirms its localization in the ERC. This experiment shows that the mutant mRme-1 protein affects sterol recycling as well as Tf recycling, suggesting that both are using similar mechanisms to return to the cell surface.
Specific Labeling of the ERC by DHE in Fixed and Permeabilized Cells-To see whether a diffusible protein carrier is needed for DHE delivery to the ERC from the plasma membrane, we examined DHE transport in fixed and permeabilized cells. When fixed with 1% paraformaldehyde for 2 min at 37°C, all the cells were fixed (judged by a block in Tf internalization, Fig. 8A), and most cells were permeabilized (judged by phalloidin staining, Fig. 8C). Alexa 488-phalloidin has a molecular weight of 1320, similar to that of DHE-loaded M␤CD. Signifi- FIG. 5. Rate of DHE uptake from the plasma membrane to the ERC measured by fluorescence recovery after photobleaching (FRAP). Cells were labeled with DHE-loaded M␤CD for 1 min and chased with 2 g/ml Alexa 546-Tf for 1 h at 37°C in either M1glucose (for control cells) or ED medium (for energy-depleted cells). Energydepleted cells were kept in ED medium throughout the entire experiment. The Tf staining (not shown) was used to find the cell of interest and place its ERC in the middle of the field. An image of the field was taken before photobleaching (Pre-bleaching). The aperture was then closed all the way, and an area in the middle of the field illuminated by the closed aperture was photobleached. Images were then taken at 0, 2, 5, 7.5, and 10 min after photobleaching to monitor the fluorescence recovery in the ERC. After background correction, a ratio of fluorescence intensities in a photobleached region to the entire cell was calculated for each time point. This ratio was then divided by the corresponding ratio obtained from the pre-bleaching image and presented as percent recovered. Each data point, derived from an average of 10 experiments, is fit to the equation y ϭ y 0 ϩ a(1 Ϫ e Ϫkt ), where k is the rate constant. The recovery half-times for control and energy-depleted cells are 2.5 Ϯ 0.6 and 2.6 Ϯ 0.8 min, respectively. cant juxta-nuclear accumulation of DHE was seen in permeabilized cells (Fig. 8B), whereas only dim staining was observed for non-permeabilized cells (upper right in Fig. 8, B and C). To test whether the juxta-nuclear accumulation of DHE in fixed cells was in the ERC, cells were first labeled with Tf, fixed, and labeled with DHE. DHE co-localized very well with Tf (Fig. 8, D and E) in these permeabilized cells (phalloidin staining, Fig.  8F), indicating that DHE was delivered to the ERC in fixed and permeabilized cells.

Internalization of [ 3 H]Cholesterol to the ERC in Control and
Energy-depleted Cells-To make sure that the results we obtained with DHE hold true for cholesterol, we confirmed key DHE results with [ 3 H]cholesterol and subcellular fractionation. Cells were incubated with 125 I-Tf, HRP-Tf, and [ 3 H]cholesterol for 1 h at 37°C. Fig. 9 shows profiles of 125 I-Tf and [ 3 H]cholesterol from density gradients of postnuclear supernatants. Fraction 1 corresponds to the top of the gradient (10% sucrose). Both 125 I-Tf and [ 3 H]cholesterol were seen in two peaks. The lighter peak (fractions 3-7) contained the plasma membrane (Fig. 9F), broken endosomes, and free 125 I-Tf released from broken endosomes. 2 The heavier peak (fractions 12-20) contained vesicular organelles, including ERC, TGN, and Golgi. 2 To separate the ERC from other intracellular organelles of similar density, we took advantage of a density shifting reaction that occurs specifically in HRP-containing compartments. Cells were incubated with HRP-Tf, 125 I-Tf, and [ 3 H]cholesterol, allowing HRP-Tf and 125 I-Tf to concentrate in the ERC. HRP-Tf and 125 I-Tf were then removed from the plasma membrane by acid wash. When ascorbic acid, DAB, and H 2 O 2 were added, a dense product was generated in the compartments containing   HRP-Tf. Under the labeling conditions, most intracellular HRP-Tf would be in the ERC. The accumulation of polymerized DAB inside the ERC induces a large increase in its density, whereas organelles in the same fraction that lack HRP-Tf are essentially unaffected (52). Unable to cross the plasma membrane, ascorbic acid prevents the reaction from taking place at the plasma membrane (53). Comparing Fig. 9, A and B, we see that the denser peak in A shifted to the pellet in B, carrying both 125 I-Tf and [ 3 H]cholesterol as a consequence of the density shifting reaction. Fig. 9, A and B We next performed the fractionation experiments in cells that had been treated with energy poisons. Cells were first incubated with 125 I-Tf and HRP-Tf for 1 h at 37°C, incubated with ED medium for 30 min, and then with ED medium containing [ 3 H]cholesterol for 1 h at 37°C. Fig. 9C shows that [ 3 H]cholesterol was found in the same peaks as the 125 I-Tfcontaining compartments on the density gradient. Fig. 9D shows that this peak could be shifted upon HRP-Tf-catalyzed DAB polymerization, showing that [ 3 H]cholesterol was delivered to the ERC in energy-depleted cells.
To test that the energy depletion protocol had blocked membrane internalization, 125 I-Tf was added to cells after energy depletion, and the amount of cell-associated 125 I-Tf was compared with that in control cells. After acid stripping of the surface-bound Tf, the 125 I-Tf counts in energy-depleted cells was measured to be less than 7% of that in the control cells (Fig. 9E), indicating Tf internalization was severely blocked as a result of energy depletion. We used surface labeling of Tf to identify the position of plasma membrane on our sucrose gradients. When cells were incubated with Tf on ice, Tf remained exclusively on the plasma membrane. Surface-bound 125 I-Tf was found in fractions 3-7, corresponding to the plasma membrane peak (Fig. 9F).

Efflux of [ 3 H]Cholesterol to the Plasma Membrane in Control and
Energy-depleted Cells-We monitored the efflux of [ 3 H]cholesterol to the plasma membrane where it is effectively removed by M␤CD as the acceptor. Our efflux data showed two kinetic populations, each comprising about half of the total (Fig. 10). The fast population, present in both control and energy-depleted cells, gave a half-time of 11 Ϯ 9 and 15 Ϯ 11 s for control and energy-depleted cells, respectively. This rate presumably represented release from the plasma membrane to the acceptor. The slower population, with a half-time of 24 Ϯ 5 min, was not seen in energy-depleted cells, consistent with our observation that DHE efflux to the plasma membrane was energy-dependent. Two pools of cholesterol exhibiting different efflux kinetics have also been demonstrated previously (55,56). It is not apparent why a depletion of cellular ATP had no effect on cholesterol efflux in one previous report (56). DISCUSSION Previous measurements have estimated that 50 -90% of cellular cholesterol is contained in the plasma membrane, and the TGN has been identified as one cholesterol-rich organelle (57,58). We show here that a large pool of cholesterol is stored intracellularly in the ERC. In microscopy studies, DHE is seen to co-localize extensively with a large part of the ERC that is concentrated in the juxta-nuclear region, and to a much lesser extent with the TGN (Fig. 2). More convincing evidence comes from the fact that when the morphology or function of the ERC is altered, DHE is affected very similarly to other ERC markers. When the ERC morphology is changed by nocodazole treatment (Fig. 3, A and B) or by expressing a mutant protein mRme-1 (G429R) (Fig. 3, C and D), DHE is found to co-redistribute with the ERC markers, indicating a close association of DHE with the ERC.
Results from our microscopy experiments using DHE as well as our biochemical studies using [ 3 H]cholesterol provide strong evidence for the association of cholesterol with the ERC in CHO cells. However, because of a low signal to noise ratio in DHE imaging and limitations in fractionation experiments, we are technically unable to detect low amounts of cholesterol in other intracellular compartments.
Transport of DHE to the plasma membrane from the ERC is energy-dependent (Fig. 6), and like Tf recycling it is slowed by expression of mRme-1 (G429R) (Fig. 7). This suggests that sterol return to the cell surface follows a conventional, tubulovesicular membrane recycling pathway. In contrast, an energyindependent pathway accounts for most of the DHE transport from the plasma membrane to the ERC (Figs. 4 and 5). We confirmed that our DHE results hold true for cholesterol using [ 3 H]cholesterol in biochemical studies. In addition to being present in the plasma membrane, [ 3 H]cholesterol is found in the same intracellular compartment as Tf (Fig. 9, A and B).
[ 3 H]Cholesterol uptake to the ERC is energy-independent ( Fig.  9, C and D), and [ 3 H]cholesterol efflux to the plasma membrane is energy-dependent (Fig. 10).
Additional information on the delivery of DHE to the ERC is gained by looking at DHE transport in fixed and permeabilized cells. When DHE-loaded M␤CD is added to permeabilized cells (indicated by phalloidin staining), all the internal membranes are essentially exposed to the reagent. Under this condition, DHE, now carried by the soluble M␤CD, is again incorporated selectively into the ERC (Fig. 8, D and E), suggesting that the ERC membrane has special properties that facilitate incorporation of DHE. Comparing the fate of DHE in permeabilized and non-permeabilized fixed cells shows that there is little intracellular accumulation of DHE in the non-permeabilized cells (i.e. cells with no phalloidin staining) (Fig. 8, B and C). Because the only difference between the two types of cells is accessibility of DHE-loaded M␤CD to the internal membranes, we conclude that the ERC in permeabilized cells is labeled mostly by DHE carried on M␤CD. This suggests that it is the ERC membrane that determines the preferential incorporation of DHE, rather than a special signal on cholesterol/DHE carrier(s) that targets the DHE transport. We do not know the basis for preferential incorporation of sterols into the ERC. One possibility is that these membranes have a high concentration of sphingomyelin which would help to accommodate cholesterol in the bilayer.
Intracellular cholesterol trafficking could occur by at least three potential mechanisms as follows: spontaneous diffusion, protein carrier-mediated diffusion, or vesicular transport. The fact that uptake of DHE to the ERC is energy-independent suggests that this pathway is non-vesicular in nature. In intact cells, a carrier protein could facilitate cholesterol transport by increasing the rate of cholesterol desorption from the bilayer and/or stabilizing the cholesterol after it leaves the bilayer. One candidate is a 13.2-kDa protein named sterol carrier protein 2 (SCP-2), which has the ability to promote sterol transfer between membranes (59 -64). Although SCP-2 has been shown to bind cholesterol and DHE (58), its lack of specificity for cholesterol and predominant localization in peroxisomes (reviewed in Refs. 65 and 66)) would prevent it from being a highly efficient carrier for cholesterol trafficking between the plasma membrane and the ERC. As an alternate to a specific carrier, cholesterol transport to the ERC could be facilitated by many soluble cytoplasmic proteins with hydrophobic binding pockets that provide a platform that shields cholesterol from the aqueous cytoplasm. Such carriers would lack specificity in targeting, but our results in the fixed and permeabilized cells indicate that specific targeting by a soluble carrier is unnecessary because M␤CD can deliver DHE to the ERC. The low level of DHE labeling of internal membranes in cells that are fixed but not permeabilized (Fig. 8, A-C) is consistent with the need for a diffusible carrier to transport sterol between membranes.
The Golgi complex has been proposed to play a role in the relocation of cholesterol from lysosomes to other cellular membranes (57,67). Uptake of low density lipoprotein increased Golgi cholesterol concentration, with the greatest increase occurring in the TGN (68). Given their similar densities on a sucrose gradient and close proximity in the perinuclear region, the TGN and the ERC are not always readily distinguishable and clearly separated in CHO cells. We speculate that what was previously interpreted as cholesterol accumulated in the Golgi might also include the ERC. Indeed, recycling endosomes isolated from rat livers have been proposed to be involved in the cell surface delivery of lipoprotein-derived cholesterol (18).
It has been shown that cholesterol moves bi-directionally between the plasma membrane and intracellular compartments, including the ER (26) and lysosomes (69). Because the ERC is in continuous communication and constant exchange with the plasma membrane via the endocytic recycling pathway (70), it could provide an intracellular cholesterol pool that is reflective of the cholesterol content in the plasma membrane. Because of its juxtaposition to the ER where cholesterol-sensing proteins (3-hydroxy-3-methylglutaryl-coenzyme A reductase (71,72), acyl-coenzyme A:cholesterol acyltransferase (73)(74)(75), and sterol-responsive element-binding protein cleavageactivating protein (76)) responsible for regulating cholesterol homeostasis are located, the ERC would be an ideal organelle through which these proteins could sense cholesterol changes in the plasma membrane.