Plasma Membrane Rafts Complete Cholesterol Synthesis by Participating in Retrograde Movement of Precursor Sterols*

Mammalian cells synthesize significant amounts of precursor sterols, in addition to cholesterol, at the endoplasmic reticulum (ER). The newly synthesized sterols rapidly move to the plasma membrane (PM). The mechanism by which precursor sterols move back to the ER for their enzymatic processing to cholesterol is essentially unknown. Here we performed pulse-chase experiments and showed that the C29/C30 sterols rapidly move from the PM to the ER and are converted to cholesterol. The retrograde precursor sterol transport is largely independent of the Niemann-Pick type C proteins, which play important roles in late endosomal cholesterol transport. In contrast, disrupting lipid rafts significantly retards the conversion of C29/C30 and C28 sterols to cholesterol, causing the accumulation of precursor sterols at the PM. Our results reveal a previously undisclosed function of the PM lipid rafts: they bring cholesterol biosynthesis to completion by participating in the retrograde movement of precursor sterols back to the ER.

Cholesterol is an important lipid component of biological membranes. It also serves as an obligatory precursor for the biosyntheses of steroid hormones, bile acids, and bioactive oxysterols (1). In mammals, virtually every cell of the body is capable of de novo cholesterol biosynthesis. Cholesterol biosynthesis involves successive enzymatic reactions, converting the simple, 2-carbon precursor acetyl-CoA into the 30-carbon, acyclic, apolar molecule squalene. The subsequent oxidation and cyclization of squalene yields the first sterol in the biosynthetic pathway, the 30-carbon (C30) lanosterol. The conversion of the C30 lanosterol to C27 cholesterol involves at least 18 enzymatic reactions (2) and can proceed through two pathways, with one involving desmosterol as the final precursor, and the other involving lathosterol and 7-dehydrocholesterol as the final precursors (Fig. 1). The biosynthetic enzymes responsible for con-verting squalene to cholesterol are all located in the endoplasmic reticulum (ER) 4 membranes (2). Previous studies showed that, in addition to synthesizing cholesterol, mammalian cells also synthesize substantial amounts of precursor sterols (3,4). Similar to cholesterol, the precursor sterols leave the ER and rapidly reach the plasma membrane (PM) within 30-min time (5,6). The precursor sterols at the PM are predicted to move back to the ER to be enzymatically processed to cholesterol. This retrograde movement is an essential step to complete cholesterol biosynthesis. Little is known about the mechanism(s) of retrograde movement of sterols. In yeast Saccharomyces cerevisiae, Prinz and coworkers showed that the retrograde movement of exogenously added sterols from the PM to the ER is governed by a non-vesicular mechanism that involves ATP-binding cassette transporters (7) and oxysterol-binding protein-related proteins (8). In mammalian cells, Maxfield and coworkers showed that both vesicular and non-vesicular trafficking mechanisms operate to govern the transport of a naturally fluorescent cholesterol analog dehydroergosterol (9,10). Earlier, based on inhibitor studies, Metherall and colleagues (11) and Field and colleagues (12) suggested that multiple drug resistance proteins may be involved in the retrograde movement of precursor sterols.
The Niemann-Pick type C1 (NPC1) protein is a multispan membrane protein containing a sterol-sensing domain, which plays an essential role in the ability of the protein to bind cholesterol (13). Studies using mutant cells that lack the NPC1 protein show that NPC1 is involved in the egress of low density lipoprotein-derived cholesterol from the late endosomes to the PM, ER, and mitochondria (14 -16). In certain cell types, including macrophages, NPC1 is also involved in the post-PM trafficking of biosynthesized cholesterol, and possibly other C27 sterols, from the PM to the ER (17). On the other hand, the ability of the mutant NPC cells to convert precursor sterols, especially the methylated sterols (i.e. C28, 29, and C30 sterols), to cholesterol, is unknown.
In this study, we monitored the intracellular fate of biosynthesized precursor sterols by feeding cells with radioactive acetate, performed pulse-chase experiments, and used several TLC systems to analyze the composition of labeled sterols. We also employed gas chromatography-mass spectrometry (GC-MS) to identify various precursor sterols in cells grown in sterol-free media. Our results show that, upon arrival at the PM, precursor sterols rapidly move back from the PM to the ER, to be enzymatically converted to cholesterol. The retrograde movement of the precursor sterols is largely independent of NPC proteins but depends on the functionality of the PM lipid rafts.

Materials
Various reagents and procedures were as described previously (15,18).

Media, Cell Lines, and Cell Culture
Medium A is Dulbecco Modified Earle's Medium (DMEM) and Ham's F12 at 1:1 (for Chinese hamster ovary (CHO) cells), or Dulbecco's modified Eagle's medium (for human fibroblasts (Hf cells)), plus 10% fetal bovine serum; Medium D is appropriate medium plus 5% delipidated fetal bovine serum, 35 M oleic acid; Medium B is appropriate medium plus 0.2% fatty acid-free bovine serum albumin; Medium F is appropriate medium without serum. All media contain 50 units/ml penicillin and 50 g/ml streptomycin as antibiotics. 25RA is a mutant CHO cell line that is resistant to 25-hydroxycholesterol (19) and contains a gain-of-function mutation in sterol regulatory element-binding protein cleavage-activating protein (20). The CT43 mutant cell line was isolated from mutagenized 25RA cells as one of the cholesterol trafficking mutants (21). It contains the same gainof-function mutation in sterol regulatory element-binding protein cleavage-activating protein, and a premature translational termination mutation near the 3Ј-end of the npc1 coding sequence, producing a non-functional truncated NPC1 protein (22). The A101 clone is a CHO cell mutant lacking the NPC1 mRNA and protein; it was isolated from parental CHO cells that express the murine ecotropic retrovirus receptor (JP17 cell). CHO clones A101 and JP17 were from professors Ohno and Ninomiya (Totori University School of Medicine, Japan) (23). A normal Hf cell line (normal-1) was from Dr. Peter Pentchev (formerly National Institute of Health). A second normal Hf cell line, GM00038, was from Coriell Institute (normal-2); a mutant NPC1 human fibroblast cell line, GM03123, was also from Coriell. GM03123 contains two point mutations in the NPC1 protein: P237S and I1061T. The NPC2 human fibroblast cell line was from Dr. Yiannis Ioannou (Mount Sinai School of Medicine) (24).

Pulse-Chase Experiments with [ 3 H]Acetate
For CHO cells, cells were seeded into 6-well plates or 100-mm dishes and grown in Medium D for 2 days as described (22). For Hfs, cells were plated and grown in Medium A to near confluency, washed twice with PBS, and grown for 2 days in Medium D. To label cells with [ 3 H]acetate, cells were washed with pre-warmed (37°C) PBS twice, then pulse-labeled with 20 Ci/ml [ 3 H]acetate (20 Ci/well or 100 Ci/100-mm dish) for the time indicated at 37°C. After pulse labeling, the medium was removed; the cells were washed twice with pre-warmed PBS and incubated with the chase media (pre-warmed Medium D or Medium B) for up to 24 h. In some experiments, the chase media were collected and extracted with chloroform/methanol (2:1, v/v) and analyzed for radiolabeled sterols. Cellular lipids were extracted with hexane/isopropanol (3:2, v/v), and cellular protein was solubilized in 0.1 N NaOH to determine protein content (25). The non-saponifiable lipids (containing the sterols) were isolated as previously described (26).

Sterol Analyses by TLC
The non-saponifiable fractions were spotted onto channeled silica TLC plates and run in methylene chloride/ethyl acetate (97:3, TLC system I) to separate C29/C30, C28, and C27 sterols (27). Lanosterol and cholesterol were added to the samples to serve as internal standards. After chromatography, the plates were briefly stained with iodine to identify the C29/C30 sterol band and the C27 sterol band. The band located between the C29/C30 and C27 sterol bands is the C28 sterol band (27). The sterol bands were scraped and counted with a liquid scintillation counter. To examine the sterol composition of the C27 sterols, the C27 sterol bands separated by TLC system I were scraped and extracted with chloroform/methanol (2:1) twice. The extracts were transferred to new glass tubes, washed once with water, and dried under nitrogen. The C27 sterols were either acetylated as described (27) or left non-acetylated. The acetylated C27 sterols were separated by TLC system II on silver nitrate impregnated plates. The plates were prepared by rapidly dipping commercially prepared silver nitrate-impreg- nated TLC plates in 10% silver nitrate solution (in acetonitrile), followed by air drying. The samples were repeatedly chromatographed (three times), each time for 1 h, using the solvent hexane/benzene (80:20) (28). Non-acetylated sterol samples were separated using the silver nitrate-impregnated TLC plates prepared as described above and chromatographed for 4 h in a plastic-wrap sealed glass chamber in 100% chloroform (system III). Plates were run with authentic cholesterol, desmosterol, lathosterol, and zymosterol, or their acetylated derivatives were spotted in parallel lanes as standards. After TLC, the standardcontaining lanes were charred by spraying the plates with an orcinol mixture (200 mg of orcinol/100 ml of 75% sulfuric acid) and baked at 100°C for 20 min. The appropriate radioactive bands in sample-containing lanes were scraped to determine radioactivity using a liquid scintillation counter. Table 1 shows the R f value of each sterol in the three TLC systems.

Sterol Analyses by GC-MS
Cellular lipids were extracted and saponified as described above, using nanograde organic solvents. The dried sterol samples were derivatized with 0.25 ml of Sigma-Sil-A (Sigma) at 60°C for 30 min. The trimethylsilyl derivatives of sterols were injected into a Shimadzu GC-17A gas chromatograph connected to a Shimadzu QP5000 mass spectrometer equipped with an XTI-5 (30 m ϫ 25 m ϫ 0.25 mm) capillary column. The injector port was set at 290°C. The initial temp of the oven was 110°C and was increased at 15°C/min to 290°C and held for 10 min. Helium flow was set at 1.3 ml/min. The sterol masses were quantified by using epicoprostanol as an internal standard.

Lipid Analyses by Enzymatic Assays
The cellular free cholesterol and choline-containing phospholipids were determined as described before (29).

Cell Fractionation after Cellular Labeling with [ 3 H]Acetic Acid
Two different methods were used to prepare the PM and the internal membrane (IM) fractions.
Method I-The PM/IM fractions were prepared by using the 30% Percoll gradient centrifugation method as previously described (15,30). Cell homogenates from two 100-mm dishes of cells were prepared and spun at 1,000 ϫ g for 5 min twice, and the resulting 1 ml of post nuclear supernatant was placed onto a 9-ml 30% Percoll solution and centrifuged at 84,000 ϫ g for 30 min at 4°C using a Beckman Type 70.1 Ti rotor. Afterward, ten 1-ml fractions were collected from the top. Lipids from each Percoll fraction were extracted with chloroform/methanol (2:1), saponified, and analyzed by TLC system I. To analyze the purity of the PM/IM fractions, cell surface proteins were biotinylated for 30 min on ice by using the EZ-Link Sulfo-NHS-Biotin kit (Pierce); the cell homogenates were then prepared and fractionated by using 30% Percoll gradient centrifugation. Afterward, the detergent Nonidet P-40 was added to each fraction at 1% final concentration; the fractions were centrifuged at 200,000 ϫ g for 30 min twice to remove the Percoll particles. Aliquots of each supernatant were subjected to SDS-PAGE followed by immunoblotting with the monoclonal anti-HMG-CoA reductase IgG-A9 (obtained from ATCC). Biotinylated proteins were detected with the Vectastain ATP-binding cassette kit (Vector Laboratories). Densitometric analyses of protein bands were performed by using NIH Image version 1.61.
Method II-The PM, IM, and cytosol fractions were prepared by using the procedure previously described (31) with minor modification. Briefly, post nuclear supernatant was prepared from one 100-mm dish of cells as described above, and centrifuged at 16,000 ϫ g for 20 min to collect the PM-rich fraction as a pellet. The supernatant was further spun at 200,000 ϫ g for 45 min to yield the cytosol fraction as supernatant and the IM-rich fraction as a pellet. The purities of the PM/IM fractions and the labeled lipids present in these fractions were analyzed in a similar manner as described above. The amounts of free cholesterol and choline-containing phospholipids in these fractions were determined by methods described previously (29). We also employed the 11% Percoll gradient centrifugation method (22) to isolate various crude subcellular fractions. This method efficiently separated late endosomes/lysosomes from other membranes. Ten 1-ml fractions were collected from the top of the tube. The endosomes/lysosomes (LE/LYS) were concentrated in fractions 8, 9, and 10. Lipids from each fraction were extracted, saponified, and analyzed with the TLC as described above.

Calculations
Data were presented as means Ϯ S.D. unless specified in the figure legends. Statistical analyses of results were performed using a two-tailed, unpaired Student's t test. The difference between two sets of values was considered significant when the p value was Ͻ0.05.

ER-to PM Anterograde Transport of Biosynthesized Precursor
Sterols-To monitor the distribution of newly synthesized precursor sterols inside the cells, WT CHO cells grown in Medium D were incubated with [ 3 H]acetate for 1 h. The labeled cell homogenates were subjected to subcellular fractionation analysis. Initially, two methods were employed (described under "Experimental Procedures") to prepare the PM, IM, and cytosol fractions. To evaluate the purity of the PM fraction and IM fraction, we biotinylated the cell surface proteins as the PM marker and used the HMG-CoA reductase protein, which resides in the ER where sterols were synthesized, as the IM marker. In Method I, the distribution analyses of these markers  Fig. 2A) show that fractions 1-3 contain ϳ75% of the total PM signal. Fractions 4 and 5 contain ϳ90% of the total HMG-CoA reductase signal. The labeled sterols present in fractions 1-10 were extracted and analyzed. The results show that the majority (Ͼ70%) of newly synthesized C29/C30, C28, and C27 sterols were present in fractions 1-3; Ͻ25% were present in fractions 4 and 5 ( Fig. 2A). Cholesterol mass analysis showed that its distribution was similar to that of newly synthesized sterols; roughly 70% of cholesterol was present in fractions 1-3, whereas Ͻ25% was present fractions 4 and 5 (data not shown).
In Method II, the analysis showed that the PM-rich fraction contained 60% of the total cell surface biotinylated protein signals, the IM-rich fraction contained 80% of the total HMG-CoA reductase signal (data not shown). The results of the labeled sterol distribution analysis (Fig. 2B) show that Ͼ60% of the newly synthesized C29/C30, C28, and C27 sterols were present in the PM fraction, whereas ϳ25% were present in the IM fraction. Small but significant amounts of 3 H-labeled biosynthetic precursor sterols were recovered in the cytosol fraction. Cholesterol mass analysis showed that it was mainly recovered in the PM fraction as expected (Fig. 2B, white bar). These results show that the second method tended to underestimate the proportion of labeled sterols present in the PM fraction by ϳ10 -12% compared with the first method. We sought to test the findings described in Fig. 2 (A and B) by using another approach. Cyclodextrin is a water-soluble molecule that has high affinity for sterols. We and others had previously shown that when 2-hydroxypropyl-␤-cyclodextrin (HCD) is added to the medium of intact CHO cells for 10 min or less, it efficiently removes cell surface cholesterol; in contrast, under the same condition, cholesterol sequestered in the internal membrane compartment(s) is very resistant to extraction by HCD (15). We pulse-labeled the parental 25RA CHO cells and the NPC1deficient mutant CT43 cells with [ 3 H]acetate for 1 h, then exposed these cells to HCD for 10 min. Labeled lipids in cells and in the media were extracted and analyzed by TLC. The results show that, in both cell types, Ͼ35% of the total newly synthesized C27 sterols were accessible to HCD (Fig. 2C). These results, together with Fig. 2 (A and B), suggest that after 1 h of synthesis, most of the biosynthesized C27 sterols is located at the PM. The results in Fig. 2C show that newly synthesized C28 and C29/C30 sterols were also extractable by HCD, although less so than C27 sterols, suggesting that HCD may bind with less affinity toward C28 and C29/C30 sterols than C27 sterols. It is possible that the additional methyl groups present in steroid ring A (the 4,4-methyl moieties) and/or present in the junction between rings C/D (the 14-␣-methyl moiety) may hinder the binding between HCD and the sterol molecule. However, we cannot rule out the possibility that the C28 and C29/C30 sterols may reside in a microdomain of the PM different from where the C27 sterols reside. The results presented in Fig. 2C do imply that after 1 h of synthesis, a substantial amount of the biosynthesized sterols is located at or near the PM to be FIGURE 2. Presence of biosynthetic precursor sterols at the PM. A, on day 0, WT CHO cells were plated in triplicate at 1 ϫ 10 6 cells per 100-mm dish and grown in Medium A (8 ml/dish). On day 1, the medium was switched to Medium D, and the cells were grown for another 2 days. On day 3, the cells were incubated with [ 3 H]acetic acid (20 Ci/ml, 5 ml/dish) for 1 h in Medium F at 37°C. Cell homogenate was subjected to the 30% Percoll gradient centrifugation analysis (Method I) as described under "Experimental Procedures." Afterward, lipids were extracted from each fraction, and the non-saponifiable lipids were analyzed by the TLC system I to identify radiolabeled biosynthetic precursor sterols. The distribution of biotinylated proteins and HMG-CoA reductase in each fraction were analyzed as described under "Experimental Procedures." Data were reported as % of total in each sterol species as indicated. The results shown are averages of two experiments; the error bars indicate the sizes of difference between the average values. B, WT CHO cells were set up and radiolabeled as described above. Cell homogenates were subjected to subcellular fractionation Method II to prepare the plasma membrane (PM), intracellular membranes (IM), and cytosol (CS) fractions (described under "Experimental Procedures"). Afterward, lipids were extracted from each fraction, the non-saponifiable lipids were analyzed by the TLC system I to determine [ 3 H]sterols. Mass of free cholesterol (FC) was determined by the colorimetric assay as described under "Experimental Procedures." Data are reported as % total in each lipid species as indicated. The results shown are averages Ϯ S.D. and are from one of two separate experiments with similar results. C, on day 0, the 25RA and CT43 CHO cells were seeded in triplicate into 6-well trays at a density of 1 ϫ 10 5 (for 25RA) or 2 ϫ 10 5 (for CT43) cells per well and grown in 1.5 ml/well in Medium A. On day 1, the medium was changed to Medium D, and the cells were grown for another 2 days. On day 3, the cells were radiolabeled with [ 3 H]acetic acid (20 Ci/ml; 1 ml/well) Medium F for 1 h, sterol efflux was then induced by adding 4% HCD to Medium F for 10 min. Afterward, lipids were extracted from the cells and from the medium, and the non-saponifiable lipids were analyzed by the TLC system I. Data are reported as % total (cell plus medium) in each sterol species present in the medium as indicated. The results shown are averages Ϯ S.D.
extractable to HCD. These results also show that the availability of various newly synthesized precursor sterols to HCD was almost identical between the 25RA and the CT43 cells, indicating that the movement of C28 sterols and C29/C30 sterols to the cell surface was independent of a functional NPC1 protein.
The results presented in Fig. 2, together with previous studies (5), show that immediately after biosynthesis, the majority of precursor sterols are transported from the ER to the PM. Thus, one or more PM-to-ER retrograde sterol movement systems are needed to bring precursor sterols arriving at the PM back to the ER, where all the post-squalene cholesterol biosynthetic enzymes reside, for their eventual conversion to cholesterol.
Role of NPC Proteins in the PM-to-ER Retrograde Transport of Biosynthesized Precursor Sterols-Because the NPC proteins play an important role in endosomal cholesterol transport, we examined whether the NPC-dependent sterol transport system is involved in the retrograde transport of newly synthesized precursor sterols. It is known that all the enzymes responsible for converting C29/C30 or C28 sterols to C27 sterols are located at the ER. This allowed us to use the conversions of C29/C30 or C28 sterols to C27 sterols as a biological assay to examine the retrograde movement of biosynthesized precursor sterols. Cells were pulse-labeled with [ 3 H]acetate for 1 h, followed by chasing in the absence of the label for up to 6 h. At each time point, cellular sterols were analyzed by TLC systems I, II, and III. In unchased 25RA cells, roughly 58% of the label was in C27 sterols, with 15 and 27% of the label in C28 sterols and C29/C30 sterols, respectively ( Table 2). The CT43 cells exhibited a somewhat different sterol distribution pattern than that of the 25RA cells, with ϳ43% in C29/C30 sterols, 16% in C28 sterols, and 42% of the label in C27 sterols ( Table 2). We further analyzed the composition of the labeled C27 sterols by using TLC systems II and III. This analysis illustrated that the 25RA and CT43 cells exhibit very similar profiles for cholesterol and its C27 precursor sterols (Table 2), with ϳ50% of the total 3 H label in cholesterol, 10 -12% in lathosterol, 15% in zymosterol, and 23-26% in desmosterol. Data from the chase experiments show that, in both cell types, the amount of labeled C29/C30 and C28 precursor sterols significantly decreased (by Ͼ2to 3-fold), whereas the amount of labeled C27 sterols significantly increased (Fig. 3, A-C). Furthermore, the rates of consumption of C29/30 or C28 sterols, as well as the rate of production of the C27 sterols, were very similar in both the 25RA and CT43 cells. The labeled C27 sterols were further analyzed by TLC System II. The results show that the levels of the three labeled C27 precursor sterols, zymosterol, lathosterol, and desmosterol, also decreased ( Table 2).
To test whether the results observed in the 25RA/CT43 CHO cells were applicable to other cell types, we performed the same pulse-chase experiments in two strains of normal Hfs, one mutant NPC1 Hf, and one mutant NPC2 Hf. The results indicate (Fig. 3, D-F, and Table 3) that at 0-h chase, in normal Hf, 57-59% of the labeled sterols were C27 sterols, whereas 25-27 and 15% of the labeled sterols were C29/C30 and C28 sterols, respectively. In mutant NPC1 and NPC2 Hfs, 39 -43% of the labeled sterols were C27 sterols, whereas 36 -44% and 17-21% of the labeled sterol were C29/C30 and C28 sterols, respectively. Thus, similarly to the 25RA/CT43 CHO cell system, modest increases in the newly synthesized C29/C30 and C28 sterols (ϳ1.5-or 1.2-fold, respectively) were seen in the mutant NPC cells. When chased up to 6 h, both the normal and mutant NPC Hfs demonstrated similar rates of increases in labeled C27 sterols (Fig. 3F), and similar rates of decreases in the labeled C29/C30 sterols (Fig. 3D). The decreases in C28 sterols were slightly slower in the mutant NPC Hfs (Fig. 3E). At 6-h chase time, mutant NPC Hfs showed 2.5-and 1.9-fold increases in the residual labeled C29/C30 and C28 sterols (Table 3); these results are similar to the findings described in the 25RA/CT43 CHO cell system ( Table 2).
The data presented in Table 2 show that, in both the CHO cell system and the Hf system, parental and mutant NPC cells

TABLE 2 Distribution of endogenously synthesized sterols in 25RA and CT43 CHO cells at 0-or 6-h chase times
Values reported (means Ϯ S.D.) are % of total labeled sterols. They are derived from the results reported in Fig. 3, A-C. display differences in the composition of the sterol precursor pool, suggesting that the lack of functional NPC proteins somehow inhibits the conversion of newly synthesized C29/C30 sterols to C27 sterols. On the other hand, the rates of conversion of C29/C30 and C28 sterols to C27 sterols are very similar in the parental and the mutant NPC cells (Fig. 3, A-F). The latter results suggest that, once arriving at the PM, the retrograde movement of the C29/C30 and C28 sterols back to the ER for their conversion to C27 sterols is only slightly affected by the NPC mutations.
To further characterize the nature of precursor sterols accumulated in NPC cells, we aimed to identify cellular sterols by GC-MS. We grew 25RA cells, CT43 cells, and a stable transfectant of CT43 cells that expresses a functional NPC1-GFP fusion protein (designated as CT43/npc1) in Medium D for 2 days and subsequently analyzed the sterol compositions of these cells by GC-MS analysis. The results of this experiment show that the NPC1-deficient mutant CHO cell line CT43 contained significantly higher levels of biosynthetic precursor sterols, including lanosterol, 4,4-dimethylsterols, monomethyl sterols, and lathosterol, than their parental 25RA cells, as shown in Table 4. The abnormal accumulations of biosynthetic precursor sterols in CT43 cells were largely reduced in CT43 cells stably expressing a functional mouse NPC1 protein. To test the generality of this finding, we compared the sterol compositions between a different NPC1-deficient CHO mutant, A101, and its parental cells JP17. Unlike the 25RA/CT43 cells, these CHO cells do not contain the additional gain-of-function mutation in the protein sterol regulatory element-binding protein cleavage-activating protein. The results (Table 4) demonstrate that the A101 cells lacking a functional NPC1 also showed significantly higher levels of precursor sterols than their parental JP17 cells. Overall, the findings described in Table 4 support the idea that mutation in NPC1 or in NPC2, by unknown mechanism(s), retards the conversion of precursor sterols to cholesterol and leads to their accumulations in cells at the steady state.
NPC Deficiency Causes Modest Accumulation of Precursor Sterols in the LE/LYS-We sought to examine whether endogenously synthesized precursor sterols accumulate preferentially in the LE/LYS of CHO cells or Hf cells. We used two types of CHO cell systems: 25RA (NPC1 ϩ/ϩ ) versus CT43 (NPC1 Ϫ/Ϫ ), and JP-17 (NPC1 ϩ/ϩ ) versus A101 (NPC1 Ϫ/Ϫ ) to test this possibility. The cells were radiolabeled with [ 3 H]acetic acid for a long period (20 -24 h), followed by chasing without the label for an additional long period (24 h). The cells were then homogenized and subjected to 11% Percoll centrifugation to isolate the PM-rich and LE/LYS-rich fractions. The compositions of labeled sterols in each Percoll fraction were analyzed. The results (Table 5) show that, in the mutant NPC1 cells (CT43 or A101 cells), the major C27 sterols (cholesterol and desmosterol), the C28 sterols, and the C29/C30 sterols all accumulated in the LE/LYS-rich fractions at levels higher than the levels of the same sterol species found in the parental cells. We next performed the same experiment in the normal and mutant NPC Hfs and obtained essentially the same results (Table 6). Overall, the results presented in Fig. 3 and Tables 2-6 suggest   TABLE 3

Distribution of endogenously synthesized sterols in human fibroblasts from normal subjects or from NPC patients at 0-or 6-h chase times
Values reported (means Ϯ S.D.) are % of total labeled sterols. They were derived from the results reported in Fig. 3, D-  that an NPC-dependent endocytic pathway plays only a minor role in distributing various precursor sterols present in the LE/LYS to the ER for further processing.

Role of Lipid Rafts in the PM-to-ER Retrograde Movement of Biosynthesized Precursor
Sterols-To further pursue the nature of the rapid, retrograde sterol movement independent of NPC proteins, we focused on an internalization process that involves lipid raft/caveolae. Lipid rafts are cholesterol-rich, sphingolipid-rich microdomains located mainly in the PM that play important roles in the internalizations of various biological molecules (32)(33)(34). It has been shown that newly synthesized sterols are rapidly transported from the site of their synthesis (ER) to the lipid rafts (5,30). We used the conversion of biosynthetically labeled C29/C30 and C28 sterols to C27 sterols as an assay to test whether lipid rafts also play a role in the retrograde movement of biosynthesized precursor sterols from the PM to the ER. We treated WT CHO cells grown in Medium D with methyl-␤-cyclodextrin (MCD) at different concentrations (0 -10 mM) for 30 min to disrupt lipid rafts, then performed pulse-chase experiments with [ 3 H]acetate. The results show that, after 1-h chase, in cells not treated with MCD, the labeled C29/C30 and C28 sterols decreased while the labeled C27 sterols increased; this demonstrates the rapid conversion of C29/ C30 and C28 sterol to C27 sterols as expected (Fig. 4, A-C, lanes  1 and 2). However, in cells treated with MCD, retarded conver-sions of C29/C30 and C28 sterols to C27 sterols were observed (Fig. 4, A-C, lanes 3-6). The retardation of conversion depended on MCD concentration. Significant inhibition of precursor sterol conversion was observed at as low as 1 mM MCD (Fig. 4, A-C, lane 4), indicating that the retrograde movement is highly sensitive to cholesterol depletion. A 10 mM MCD treat-

TABLE 6 Percentage of individual ͓ 3 H͔sterols present in the LE/LYS-rich fractions in various Hf cells as indicated
Cells were labeled with ͓ 3 H͔acetate and were processed for analyses in the same manner as described in ment caused roughly a 4-fold increase in C29/C30 sterols, a 1.5-fold increase in C28 sterols, and a concomitant reduction in C27 sterols, from Ͼ90% of total sterols in control cells to Ͻ70% of total sterols in treated cells. We next tested the effects of several other agents also known to disrupt lipid raft functions, on the retrograde movement of biosynthesized precursor sterols. The agents that we chose to use here have all been shown to disrupt lipid raft functions in CHO cells (35). The results of these experiments (Fig. 4, D-F) demonstrate that genistein, a protein-tyrosine kinase inhibitor (36,37), or nystatin, an agent that disrupts lipid rafts by interacting with cholesterol in membranes (38), also significantly retarded the conversion of C29/C30 and C28 sterols to C27 sterols. The effects of genistein and MCD are more specific, because they did not cause significant alteration in total sterol synthesis rate; in contrast, nystatin at higher concentration significantly inhibited total sterol synthesis rate (rate values were reported in the Fig. 4 legend). Additional results show that chlorpromazine, an agent known to inhibit clathrin-dependent endocytosis (39), only slightly inhibited the conversion of C29/ C30 sterols to C27 sterols, suggesting that clathrin-dependent endocytic internalization may play a minor role in the retrograde movement of precursor sterols to the ER.
We sought to find out in which compartment(s) the biosynthesized precursor sterols accumulate, when their conversion to cholesterol is retarded by lipid raft disruption. To address this question, we treated WT cells with or without CD, prepared the PM, IM, and cytosolic fraction by using Method II (described under "Experimental Procedures"), and performed labeled sterol composition analysis in these fractions. The results of the control experiment (Fig. 5A) show that, in the whole cell homogenates, as expected, the MCD treatment significantly increased the % of labeled C29/C30 sterols, and significantly decreased the % labeled C27 sterols. Additional analysis illustrates that in MCD-treated cells, most of the labeled C27, C28, and C29/C30 sterols accumulated in the PM fractions (Fig. 5, B-E). We used the PM:IM/CS ratio to estimate the proportion of labeled sterols in the PM. As shown in Fig. 5E, in control cells, the value for the PM:IM/CS ratio for various labeled sterols averaged 1.2; in MCD-treated cells, this value increased to 2.0. Thus, MCD treatment increased the proportion of various labeled sterols present in the PM by almost 70%. As noted earlier in the text, the fractionation procedure employed in this experiment (Method II) tended to underestimate the % labeled sterols in the PM fraction by 10 -12%. However, the degree of impreciseness of this method would only modestly modify the magnitudes of increase reported here. Thus, we conclude that the lipid raft disruption by MCD caused various biosynthesized sterols, including C29/C30, C28, and C27, to accumulate at the PM.
We next asked whether the effect of MCD on retrograde movement of precursor sterols might be reversible by cholesterol repletion in the medium. To test this possibility, we treated the WT CHO cells with MCD at a relatively low dose (1 mM) and incubated the MCD-treated cells without or with cholesterol, or with the cholesterol analogue epicoprostanol, then examined the conversion of the biosynthesized C29/30, and C28 precursor sterols to C27 sterols. The results (Fig. 6) show that adding cholesterol reduced the labeled C29/C30 sterols (compare lane 3 to lane 2 in A) and increased the labeled C27 sterols (compare lane 3 to lane 2 in C). In contrast, adding an equal amount of epicoprostanol, a hydrophobic stereoisomer of cholesterol, did not produce the same rescue effect as did cholesterol. The results of a control experiment show that, without MCD treatment, cholesterol added to the medium did not significantly alter the labeled C29/C30 and C27 sterol distribution (Fig. 6, A and C; compare lane 5 to lane 1), whereas epicoprostanol added slightly inhibited the conversion of C29/C30 sterols to C27 sterols (Fig. 6, A and C; comparing lane 6 to lane 1; p value ϭ 0.039 for C29/C30 sterols). The changes of C28 sterol levels under various treatments were too small to be informative (Fig.  6B). The inhibitory effect of epicoprostanol reported here can be rationalized as follows: in model membrane studies, cholesterol and a few other related sterols such as epicholesterol and 25-hydroxycholesterol, are shown to induce a domain that is enriched in cholesterol and saturated lipids. Other sterols, such as coprostanol, androstenol, and 4-cholestenone, inhibit the domain formation (40). Thus, epicoprostanol, a sterol closely related to coprostanol, may act by inhibiting lipid raft formation in intact cells. Overall, these results indicate that a cholesterol-based PM lipid raft domain plays an important role in the retrograde movement of biosynthesized C29/C30 sterols from the PM to the ER. Fig. 7 is a model for intracellular sterol trafficking based on our findings. It is an extension of a previous model (1) and focuses on the fate of biosynthetic precursor sterols described in the current work.

DISCUSSION
In the current study, we performed labeled acetate pulse experiments to monitor the fate of biosynthesized precursor sterols and showed that, upon synthesis at the ER, the majority of the labeled C29/C30, C28, and C27 sterols rapidly move to the PM independently of NPC1 (Fig. 2). We then performed chase experiments to monitor the disappearance of C29/C30 sterols with time (Fig. 2, A and D), and demonstrated that the retrograde movement of endogenously synthesized C29/C30 sterols occurs with a half time of Ͻ60 min in a manner largely independent of NPC1/NPC2 proteins. Interestingly, Pentchev and colleagues (41) had previously reported that the labeled C27 sterol desmosterol added exogenously to mouse fibroblasts is converted to cholesterol inside the cells in a manner independent of the NPC1 mutation. We then showed that treating cells with MCD, genistein, or nystatin severely inhibits the conversion of C29/C30 precursor sterols to C27 sterols. These agents disrupt the function of lipid rafts, but they do so via different modes of action. MCD efficiently extracts cholesterol from the cholesterol-rich domain in the PM, nystatin binds tightly to cholesterol, whereas genistein inhibits lipid raft function by acting as a tyrosine kinase inhibitor. We next showed that the MCD treatment led to retention of the labeled C29/ C30 and C28 sterols at the PM (Fig. 5); the inhibitory effect of MCD can be reversed by cholesterol repletion (Fig. 6). These results led us to conclude that lipid rafts at the PM play a crucial role in bringing cholesterol biosynthesis to completion. They do so by participating in the retrograde movement of biosynthesized C29/C30 and C28 precursor sterols, bringing these sterols back from the PM to the ER so they can be enzymatically processed to cholesterol. As indicated earlier, close analysis of results presented in Fig. 3 (E and F) revealed that, when compared with WT Hfs, slightly slower rates of C28 sterol disappearance and C27 sterol appearance occurred in the NPC Ϫ/Ϫ Hfs. This finding can be rationalized by the earlier finding reported by Garver and colleagues (42), who showed that the concentration of cholesterol in the PM caveolae isolated from NPC1 Ϫ/Ϫ fibroblasts was significantly decreased compared with that of the WT fibroblasts. Thus, in the mutant NPC Hfs, the PM lipid raft function in mediating the retrograde movement of the precursor sterols may be slightly compromised.
How the PM lipid rafts mediate the retrograde movement of endogenously synthesized sterols is unknown at present. It seems plausible that at least two different mechanisms may be involved. The first is a lipid raft-originated, non-clathrin-mediated vesicular trafficking pathway (43). In particular, the SV-40 virus can utilize caveolae-dependent endocytosis to enter cells and is transported to the ER through "caveosomes," which are vesicles that do not contain markers for endosomes, lysosomes, ER, Golgi, or clathrin-coated vesicles. Such a vesicular trafficking pathway may be involved in the delivery of biosynthesized precursor sterols from the PM to the ER. A second mechanism may involve non-vesicular sterol transport (44) and may involve the participations of soluble proteins, such as the StAR (steroidogenic acute regulatory protein)-related proteins (45), and the oxysterolsbinding protein-related proteins, which have high affinities for sterols and/or for other lipid species (8,46). The multiple-drug resistance proteins are members of the P-glycoprotein ATPase-binding cassette transport family and have been implicated in the retrograde movement of sterols (11,12). The PM lipid raft may act in concert with the multiple drug resistance protein(s) to mediate the movement of the precursor sterols back to the ER. In yeast, the  6) did not alter the total sterol biosynthesis rate by Ͼ15% (110.9 Ϯ 12.8% or 86.4 Ϯ 11.4% of the value found in the control cells, respectively). Values that were statistically different from the control value are indicated by asterisks: *, p Ͻ 0.01; **, p Ͻ 0.05. non-vesicular sterol transport involves the oxysterols-binding protein-related proteins and the ATP-binding cassette transporters (7); a similar mechanism may be involved in the retrograde transport of precursor sterols in mammalian cells.
In the current work, we showed that various biosynthesized sterols, including the C29/C30 sterols, desmosterol, and cholesterol, all exhibited elevated levels of accumulations within the LE/LYS compartment of mutant NPC cells. However, the accumulation is only modest, and it took 8 h or longer to detect the sterol accumulation in the LE/LYS. Based on these results, we conclude that the NPC-dependent sterol transport mechanism is not the primary mechanism for the rapid retrograde movement of biosynthesized sterols. How the biosynthesized sterols move to the LE/LYS is currently unknown; our current results showed that chlorpromazine slightly inhibited the conversion of C29/C30 sterols to C27 sterols, suggesting that a clathrin-dependent endocytic mechanism may be involved in internalizing some of the biosynthesized sterols to the LE/LYS, in a manner similar to the internalization of low density lipoprotein-derived cholesterol. The inhibitory effects of the precursor sterol conversion by lipid raft disruptions were twice as efficient as the effect caused by using the clathrin pathway inhibitor. We speculate that, after arriving at the PM, 70 -80% of the precursor sterols may be internalized by a lipid raft-dependent process, whereas the remaining 20 -30% may be transported to the ER through a clathrin-dependent transport pathway. Further investigations are needed to resolve this issue.
Are PM lipid rafts also involved in the retrograde movement of cholesterol? Cholesterol is an excellent substrate for the enzyme acyl-CoA:cholesterol acyltransferase 1, a resident enzyme in the ER (1); however, C29/C30 precursor sterols are very poor substrates for acyl-CoA: cholesterol acyltransferase (47). We and others had taken advantage of the acyl-CoA:cholesterol acyltransferase substrate specificity and employed the % esterification of labeled cholesterol by acyl-CoA: cholesterol acyltransferase 1 as an assay to monitor the retrograde movement of cholesterol. The results showed that, in macrophages, an NPC-dependent mechanism plays a significant role, whereas in fibroblasts or in hepatocytes, one or more NPC-independent mechanisms participate significantly in cholesterol esterification (17). We also noted that, in the various cell types we examined, esterification of cholesterol delivered exogenously occurs in several hours, whereas the conversion of biosynthesized C29/C30 sterols to C27 sterols occurs within 1 h. Thus, whether the NPC-independent mechanism for the retrograde movement of cholesterol shares the same machinery for the retrograde movement of biosynthetic precursor sterols cannot be determined at present.
Why should cells produce precursor sterols in addition to cholesterol? Recent studies have shown that certain biosynthesized precursor sterols have specific biological functions that cannot be replaced by cholesterol. For example, the C30 precursor sterols lanosterol and dihydrolanosterol, but not cholesterol or other C27 sterols, promote the ubiquitination of HMG-CoA reductase, the rate-limiting enzyme in cholesterol biosynthesis, resulting in down-regulation of reductase enzyme activity (48). The C27 sterol desmosterol, or the C29 sterol 4,4dimethylsterol, but not lanosterol, acts as a ligand for the liver X receptors (49,50) that transcriptionally up-regulate multiple genes involved in lipid metabolism. Thus, it is tempting to speculate that the PM lipid raft may be involved in affecting the rate of sterol biosynthesis, and/or the activity of the liver X receptor. In addition to their normal physiological functions, the abnormal build-up of various biosynthetic precursor sterols is involved in pathophysiology. In humans, several malformation syndromes, including RSH/Smith-Lemli-Opitz syndrome, desmosterolosis, and X-linked dominant chondrodysplasia punctata type 2, are due to genetic deficiencies in various enzymes involved in the late stages of the cholesterol biosynthetic pathway, causing abnormal accumulation of various precursor sterols in various tissues of these patients (51). Treating rodents with various teratogens causes abnormal build-up of lanosterol and other precursor sterols and produces phenotypic abnormalities that mimic the malformation syndromes found in humans (52). These teratogens include certain plant alkaloids such as cyclopamine and jervine, which resemble cholesterol in structure and are believed to act as inhibitors of enzyme(s) FIGURE 7. A model describing intracellular sterol transport pathways that involve the biosynthetic precursor sterols. Dark yellow circles and light yellows circle represent biosynthetic precursor sterols (mainly the C29/C30 sterols) and cholesterol, respectively. Dotted lines represent putative trafficking steps yet to be demonstrated experimentally. See "Discussion" for more details. ACAT1, acyl-CoA:cholesterol acyltransferase 1; AL, acid lipase; CEH, cholesteryl ester hydrolase; EE, early endosome; ER, endoplasmic reticulum; LE, late endosome; LYS, lysosome; and NPC, Niemann-Pick type C.
involved in the catalytic conversion of precursor sterols to cholesterol (52). Based on our current finding, it would be interesting to test whether some of the teratogens act by inhibiting the function of the PM lipid raft, in addition to blocking enzymes in distal stage of the cholesterol biosynthetic pathway.