Compartmentalization of cholesterol biosynthesis. Conversion of mevalonate to farnesyl diphosphate occurs in the peroxisomes.

We have recently demonstrated that mevalonate kinase and farnesyl diphosphate (FPP) synthase are localized predominantly in peroxisomes. This observation raises the question regarding the subcellular localization of the enzymes that catalyze the individual steps in the pathway between mevalonate kinase and FPP synthase (phosphomevalonate kinase, mevalonate diphosphate decarboxylase, and isopentenyl diphosphate isomerase). These enzyme are found in the 100,000 × g supernatant fraction of cells or tissues and have been considered to be cytoplasmic proteins. In the current studies, we show that the activities of mevalonate kinase, phosphomevalonate kinase, and mevalonate diphosphate decarboxylase are equal in extracts prepared from intact cells and selectively permeabilized cells, which lack cytosolic enzymes. We also demonstrate structure-linked latency of phosphomevalonate kinase and mevalonate diphosphate decarboxylase that is consistent with a peroxisomal localization of these enzymes. Finally, we show that cholesterol biosynthesis from mevalonate can occur in selectively permeabilized cells lacking cytosolic components. These results suggest that the peroxisome is the major site of the synthesis of FPP from mevalonate, since all of the cholestrogenic enzymes involved in this conversion are localized in the peroxisome.

The demonstration that mevalonate kinase and FPP synthase are localized predominantly in peroxisomes (8,9) raises the question regarding the localization of the enzymes that catalyze the individual steps in the pathway between mevalonate kinase and FPP synthase (phosphomevalonate kinase, mevalonate diphosphate decarboxylase, and isopentenyl diphosphate isomerase). Based on results obtained from fractionation studies, these enzymes are believed to be localized in the cytosol. However, recent data have shown that the activities of these enzymes as well as mevalonate kinase and FPP synthase are significantly reduced in liver tissue obtained from patients with peroxisome-deficient diseases (Zellweger syndrome and neonatal adrenoleukodystrophy), thus indicating a peroxisomal localization (9).
We have routinely employed three different methods to study subcellular localization of proteins: (i) analytical subcellular fractionation and measurements of enzyme activities, (ii) immunoblotting of the protein in the isolated fractions with a monospecific antibody, and (iii) immunoelectron and immunofluorescence microscopy. In the studies demonstrating the peroxisomal localization of mevalonate kinase and FPP synthase (8,9), it was shown that analytical subcellular fractionation of liver and measurements of enzyme activities are not sufficient to determine intracellular localization due to the release of these enzymes in the cytososlic fraction from peroxisomes during the isolation of the organelle. Immunoelectron and immunofluorescence microscopy studies with specific antibodies to these enzymes were critical in determining the correct subcellular localization. To our knowledge antibodies to phosphomevalonate kinase, mevalonate diphosphate decarboxylase, and isopentenyl diphosphate isomerase are not currently available. Hence, in order to study the subcellular localization of these enzymes, we have selected to use permeabilized cells, which retain their organelle integrity yet lack cytosolic components. Permeabilized cells have been used successfully in a number of different studies dealing with subcellular function and localization. Cell Permeabilization-CV-1 cells were seeded at a density of 4.0 ϫ 10 4 on 60-mm plates and grown to 70% confluence in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Care was taken to seed cells in a manner that resulted in all plates per given experiment to have the same number of cells. This was important since enzyme activities and rates of cholesterol synthesis were determined per plate. As indicated in individual experiments, cells were transferred to media containing lipoprotein-deficient serum 24 h prior to the experiment. The day of the experiment, the media was aspirated off the plates and the plates were then washed twice with ice cold KH buffer (50 mM HEPES, 110 mM KOAc, pH 7.2). The plates were then transferred to ice and the cells were incubated for 5 min in KHM buffer containing 20 g/ml digitonin, 20 mM HEPES, 110 mM KOAc, and 2 mM MgOAc, pH 7.2. The digitonin solution was then aspirated off the cells, * This work was supported by National Institutes of Health Grants DK 44350 and DK 32852 and in part by a grant from The Council for Tobacco Research, U. S. A. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Materials-Biochemicals
‡ To whom correspondence should be addressed. Tel.: 619-594-5368; Fax: 619-594-7937. 1 The abbreviations used are: HMG-CoA, 3-hydroxy-3-methylglutaryl coenzyme A; FPP, farnesyl diphosphate; CV-1, monkey kidney cells; ER, endoplasmic reticulum; FXR, farnesoid X-activated receptor. and the cells were washed twice with ice-cold KH buffer and subsequently allowed to incubate in KH buffer for 30 min (10). This procedure allows the cells to remain attached to the plates. Control cells were treated similarly and washed with the same buffer lacking digitonin.
Measurement of Cholesterol Synthesis-CV-1 cells were grown to 80% confluence on 160-mm plates in standard media, and transferred to media containing lipoprotein-deficient serum 24 h prior to use. The day of the experiment, the cells were permeabilized as described and each plate was scraped into 3 ml of reaction buffer. The reaction buffer consisted of the following: 100 mM phosphate buffer, 4 mM MgCl 2 , 1 mM dithiothreitol, 15 mM ATP, and 5 mM NADPH, pH 7.4. Control cells were scraped into 3 ml of reaction buffer and gently homogenized by hand in a glass homogenizer by three up and down strokes. Subsequently, 10 Ci of (RS)-[5-3 H]mevalonic acid in 100 mM phosphate buffer, pH 7.4, was added to each reaction and incubated for 4 h at 37°C. The reaction was terminated by adding 3 ml of 60% KOH (w/v) and saponified for 1 h at 70°C. After saponification, internal [26- 14 C]cholesterol standard was added to each tube and extraction and separation of the non-saponifiable lipids was performed as described (11,12). The sterol fraction was separated by thin layer chromatography (hexane:diethyl ether, 50:50). The recovery of cholesterol averaged 60 -80% and was well resolved from squalene and lanosterol.
Mevalonate Kinase Assay-The reaction buffer consisted of 100 mM phosphate buffer, 4 mM MgCl 2 , 1 mM dithiothreitol, 1 mM EDTA, pH 7.4. (RS)-[5-3 H]Mevalonic acid was added at a specific activity of 625 dpm/ nmol. Reactions were run on tissue culture plates by adding 800 l of reaction buffer containing 50,000 dpm of substrate and incubating at 37°C for 10, 20, or 30 min. Reactions were terminated by scraping cells into Eppendorf tubes and boiling the sample for 3 min. [1-14 C]Isopentenyl diphosphate was added as an internal standard to each sample, and the products were separated on DEAE-cellulose (Cl Ϫ form) columns (8,13).
Phosphomevalonate Kinase Assay-The incubation conditions were the same as for mevalonate kinase except that (R)-[2-14 C]mevalonic acid-5-phosphate was added at a specific activity of 6,000 dpm/nmol and 50,000 dpm per assay was used. The products were loaded on AG1-X8 formate resin column (3 ml) and washed with 50 ml of 4 N formic acid to elute first the substrate and subsequently with 50 ml of 0.8 M ammonium formate in 4 N formic acid to elute the phosphorylated products (9,14).
Mevalonate Diphosphate Decarboxylase Assay-The incubation conditions were the same as for mevalonate kinase except that (R)-  H]mevalonic acid-5-diphosphate at 50,000 dpm/pmol/assay was used. The products were loaded on AG1-X8 formate resin column (3 ml) and washed with 50 ml of 0.4 M ammonium formate in 4 N formic acid and subsequently with 50 ml of 0.8 M ammonium formate in 4 N formic acid (9,14).
Isopentenyl Diphosphate Isomerase Assay-The reaction buffer consisted of the following: 10 mM HEPES, 10 mM KF, 0.01% bovine serum albumin, 1 mM MgCl 2 , and 10 mM ␤-mercaptoethanol. To each sample, 790 l of reaction buffer was added and allowed to incubate at 37°C for at least 2 min. The [1-14 C]isopentenyl diphosphate substrate was then added at a concentration of 16,000 dpm/nmol in 10 l. The samples were incubated for 5 min at 37°C; the reaction was stopped by adding 2 ml of 25% HCl in methanol and incubating another 10 min at 37°C. [1-3 H]Farnesyl diphosphate standard was then added to each sample. The samples were transferred from the tissue culture plates to glass tubes, extracted with 6 ml of hexane, vortexed, and centrifuged. The hexane phase was transferred to a scintillation vial and counted (15).
Latency of Catalase in Permeabilized Cells-Cells were permeabilized as described above and assayed for catalase activity in the presence or absence of 0.1% Triton X-100 on 60-mm tissue culture plates for the indicated time intervals. The mixture was then transferred to cuvettes, and absorbance was read as described for the measurement of catalase (16).
Marker Enzyme Assays-The activity of phosphoglucose isomerase was measured according to the method of Noltmann et al. (17), and esterase was determined according to the method of Beaufay et al. (18). Protein concentration was determined by the BCA method (Pierce) using bovine serum albumin as a standard.
Immunofluorescence-CV-1 cells were grown on coverslips, and individual coverslips of control and permeabilized cells were treated in parallel with antibodies to various marker enzymes as described (8). For cytosolic labeling, the cells were treated with sheep anti-fatty acid synthase antibody (1:20 dilution) and subsequently with secondary fluorescein conjugate-labeled donkey anti-sheep antibody (1:100 dilution). For peroxisomal proteins the cells were treated with rabbit anti-SKL antibody (1:200 dilution), and for endoplasmic reticulum labeling, rabbit antibody against HMG-CoA reductase (1:30 dilution) was used. For mitochondrial distribution, a rabbit anti-cytochrome c oxidase antibody (1:25 dilution) was used. The secondary label for the peroxisomal, ER, and mitochondrial labels was Texas Red-labeled goat antirabbit IgG antibody (1:100 dilution).

RESULTS AND DISCUSSION
Selective Permeabilization of the Plasma Membrane and Release of Cytosolic Components-Recently, several laboratories have been successful in permeabilizing cells in a manner that allows the cell organelles to remain intact while the plasma membrane is disrupted (10,19). Digitonin permeabilizes cells by complexing with cholesterol. Since the ER membranes, peroxisomal membranes, and most other cell organelles are almost devoid of cholesterol, digitonin complexes almost exclusively with the plasma membrane. The main consequence of permeabilizing cells is the loss of cytosolic components, while leaving the organelles virtually intact.
To test whether only the plasma membrane was disrupted and the cell organelles remained intact, we determined various marker enzyme activities in both permeabilized and intact cells. Table I illustrates that there was no significant difference between the two groups in the total activity per plate of esterase (marker enzyme for endoplasmic reticulum) and catalase (marker enzyme for peroxisomes), whereas phosphoglucose isomerase (a marker enzyme for cytosolic fraction) was measurable in the control cells and was not detectable in the permeabilized cells. Furthermore, the protein concentration in permeabilized cells was approximately 30 -50% less than that of intact cells. This corresponds to the protein content of cell cytosol. However, to further demonstrate that the peroxisomal, ER and mitochondrial compartments of cells remain intact after permeabilization while the cytosolic contents disappear, we employed an additional method. Control and permeabilized cells grown on coverslips were treated in parallel with antibodies to various marker enzymes as described under "Experimental Procedures." Fig. 1 illustrates the immunofluoresence pattern obtained for cytosol, peroxisomes, ER, and mitochondria in control cells and digitonin permeabilized cells. Panel A in Fig. 1 shows cytosolic labeling in intact CV-1 cells, whereas the permeabilized CV-1 cells in panel B are devoid of cytosolic labeling. The bright fluorescence in the center of the cell is due to autofluorescence of the nucleus. In panels C and D, the cells were labeled for peroxisomal proteins using an antibody made against the peroxisomal targeting signal (SKL at the C terminus) (20). A uniform punctate pattern characteristic of peroxisomal labeling is observed in both the intact cells (panel C) as well as in the permeabilized cells (panel D). Additionally, the pattern of ER labeling is similar in control cells (panel E) and permeabilized cells (panel F). Cells in panels G (control) and H (permeabilized) demonstrate that the mitochondrial membrane also remains intact during selective permeabilization with digitonin. Taken together, these results demonstrate that the plasma membrane of CV-1 cells can be selectively permeabilized with low concentrations of digitonin, resulting in the release of cytosolic proteins while maintaining organelle integrity. Validation of the Experimental Model-In order to obtain information on the functional characteristics of the peroxisomal membrane and to investigate how long the peroxisomes would maintain their integrity, the latency of catalase was measured as a function of time in permeabilized cells. Fig. 2 shows that catalase remains up to 70% latent (i.e. inside the peroxisomes) during the first 10 min after cell permeabilization. After 30 min, less than 5% of catalase activity is latent.
Based on these results, an incubation period of 10 min (or less) was selected for all subsequent experiments dealing with latency determinations.  1. Selective permeabilization of the plasma membrane releases cytosolic components but maintains the integrity of subcellular organelles. Control and permeabilized cells grown on coverslips were treated in parallel with antibodies to various marker enzymes as described under "Experimental Procedures." Panels A, C, E, and G illustrate control cell labeling; panels B, D, F, and H represent labeling of permeabilized cells. Panel A shows cytosolic labeling in intact cells; panel B illustrates the absence of cytosolic labeling in permeabilized cells. The bright fluorescence in the center of the cell is due to autofluorescence of the nucleus. In panels C and D, the cells were labeled for peroxisomal proteins, in panels E and F for ER proteins, and in panels G and H for mitochondrial proteins. of these enzymes were determined in control and in permeabilized cells in the presence and absence of 0.1% Triton X-100. Table II illustrates the results. The activities of all three enzymes were similar in control cells and in permeabilized cells treated with Triton X-100. These results suggest that these enzymes are not predominantly found in the cytosol, since the activities were not increased in control cells containing cytosolic proteins. Similar levels of all three enzyme activities were measured in control cells in the presence of Triton X-100 or in the absence in cells disrupted by homogenization. However, with permeabilized cells in the absence of Triton X-100, the activities of phosphomevalonate kinase and mevalonate diphosphate decarboxylase were significantly lower. These results suggest that the substrates for these two enzymes are not freely permeable across the organelle membrane, whereas the activity of mevalonate kinase was the same in permeabilized cell in the presence or absence of Triton X-100, indicating that the substrate for this enzyme is freely permeable through the peroxisomal membrane. We were unable to measure any isopentenyl diphosphate isomerase activity in these cells (data not shown).

Measurement of Rate of Cholesterol Synthesis in Control and
Permeabilized Cells-If indeed the enzymes required for con-version of mevalonate to FPP (i.e. mevalonate kinase, phosphomevalonate kinase, mevalonate diphosphate decarboxylase, and isopentenyl diphosphate isomerase) are not present in the cytosol, then permeabilized and control cells would be expected to have similar rates of cholesterol synthesis, using mevalonate as substrate.
Results of these experiments are shown in Table III. The mean value for control cells was 4287 dpm/plate and for permeabilized cells 3026 dpm/plate. These two means are not significantly different. No conversion of mevalonate to cholesterol was observed if ATP and/or NADPH were omitted from the reaction buffer (data not shown). These results indicate that the cytosolic fraction of cells is not necessary for the biosynthesis of cholesterol from mevalonate.
In summary, the current report demonstrates that mevalonate kinase, phosphomevalonate kinase, and mevalonate diphosphate decarboxylase activities in extracts prepared from intact cells are equal to those of selectively permeabilized cells that lack cytosolic enzymes. We also demonstrate structurelinked latency of phosphomevalonate kinase and mevalonate diphosphate decarboxylase that is consistent with a peroxisomal localization of these enzymes. These results in combination with the previous observation that mevalonate kinase and FPP synthase are predominantly localized to peroxisomes (8,9), suggest that all of the cholesterogenic enzymes involved in the conversion of mevalonate to FPP are localized to the peroxisome. This conclusion is further supported by the direct finding that cholesterol biosynthesis from mevalonate can occur in selectively permeabilized cells lacking cytosolic components and indirectly by the previous observation that all the required enzymes for the conversion of mevalonate to FPP are significantly reduced in tissue obtained from patients with Zellweger syndrome and neonatal adrenoleukodystrophy (9).
The isoprenoid biosynthetic pathway is unrivaled in nature for the chemical diversity of the compounds it produces. FPP is a key intermediate that serves as a substrate for a number of critical branch-point enzymes including the synthesis of squalene, cholesterol, farnesylated and geranylgeranylated proteins, dolichols, coenzyme Q, and the isoprenoid moiety of heme a. Thus, the regulation and levels of FPP are important since large perturbations in FPP could alter the flux of isoprenoid compounds down the various branches of the pathway.
If indeed, the majority of the cell's FPP is produced in the peroxisomes, this means that FPP and/or farnesol has to be transported out of peroxisomes for further metabolism. Since phosphorylated products of mevalonate are not able to cross the peroxisomal membrane, it is likely that FPP may also be impermeable. Therefore, is FPP first converted to farnesol in the peroxisome, and then is freely diffusible out of the organelle? Or is there a transport/binding protein that facilitates the movement of these intermediates? What determines where FPP is utilized? What regulates FPP conversion to farnesol and farnesol conversion to dicarboxylic acids? These are important questions that need to be addressed in order to understand the  regulation of FPP/and or farnesol. It is significant to note two recent studies demonstrating the potential importance of farnesol in regulation of cellular function (21,22). In the first study, farnesol has been identified as the non-sterol derivative that can initiate and promote the degradation of HMG-CoA reductase in permeabilized cells (21). In the second study, an orphan nuclear receptor named farnesoid X-activated receptor (FXR) is described that is activated by farnesol (22). Thus, FXR provides an example of a vertebrate transcription factor that is regulated by an intracellular metabolite (farnesol) and may indicate the existence of a novel vertebrate-signaling pathway (22). The FXR target genes remain to be identified.