Role of multidrug resistance P-glycoproteins in cholesterol biosynthesis.

Multidrug resistance (MDR) P-glycoproteins were first recognized for their ability to catalyze ATP-dependent efflux of cytotoxic agents from tumor cells when overexpressed. Despite extensive study, little is known about the normal substrate(s) and normal cellular function of these proteins. In the accompanying manuscript (Metherall, J. E., Waugh, K., and Li, H. (1996) J. Biol. Chem. 271, 2627-2633), we demonstrate that progesterone inhibits cholesterol biosynthesis, causing the accumulation of a number of cholesterol precursors. In the current manuscript, we use several criteria to show that the progesterone receptor is not involved in this inhibition. Rather, we demonstrate that progesterone inhibits cholesterol biosynthesis by interfering with MDR activity. We show that a steroid hormone's ability to inhibit cholesterol biosynthesis is correlated with: 1) its general hydrophobicity and 2) its ability to inhibit MDR activity. The only exception to this finding is β-estradiol, which is a more potent inhibitor of cholesterol biosynthesis than expected based solely on hydrophobicity and MDR inhibition. We further demonstrate that nonsteroidal inhibitors of MDR also inhibit cholesterol biosynthesis. Since MDR activity is required for esterification of LDL-derived cholesterol (P. DeBry and J. E. Metherall, submitted for publication), we investigated the relationship between these phenomena and show that inhibition of cholesterol esterification does not cause inhibition of cholesterol biosynthesis and that inhibition of cholesterol biosynthesis does not cause inhibition of cholesterol esterification. We propose a model in which MDR is required for transport of sterols from the plasma membrane to the endoplasmic reticulum (ER). Inhibiting this transport prevents cholesterol esterification and cholesterol biosynthesis by preventing sterol substrates from reaching ER-resident enzymes.

Mammalian cells grown in the presence of a single cytotoxic drug, such as vincristine or dactinomycin, can result in the selection cells that are resistant to that drug as well as a broad spectrum of structurally and functionally distinct compounds (1). This phenomenon, termed multidrug resistance (MDR), 1 often results from amplification and overexpression of genes encoding integral plasma membrane proteins known as MDR P-glycoproteins. These proteins function as ATP-dependent efflux pumps that reduce drug cytotoxicity by decreasing their intracellular accumulation. P-glycoproteins are encoded by a small gene family, which consists of three members in rodents and two members in humans (2,3). The MDR proteins are highly homologous, containing 12 predicted membrane-spanning domains and two ATP-binding cassettes. The three rodent genes appear to have arisen from a common ancestor through two successive gene duplication events, the most recent event producing mdr1 and mdr3. The human MDR1 gene is the counterpart of rodent mdr1 and mdr3 (4), while human MDR2 is the counterpart of rodent mdr2 (5). Human MDR1 and MDR2 are encoded by linked genes on chromosome 7q that are often co-amplified in drug-resistant cell lines. The human MDR1 gene and the rodent mdr1 and mdr3 genes are primarily responsible for the drug efflux, while human MDR2 and rodent mdr2 lack drug efflux activity (6). The MDR gene family is part of a larger superfamily of ATP-binding cassette membrane transporters. This superfamily includes: the STE6 protein from Saccharomyces cerevisiae (7), the cystic fibrosis transmembrane conductance regulator chloride channel (8), and the TAP1 and TAP2 proteins involved in peptide translocation and antigen presentation in T lymphocytes (9).
The ability of P-glycoproteins to catalyze the efflux of unnatural drugs has led to the proposal that MDR normally functions in detoxification. Consistent with this proposal is the recent finding that mice homozygous for disruption of the mdr3 gene, which is primarily expressed in brain capillaries that define the blood-brain barrier, are phenotypically normal unless challenged with drugs; they demonstrate increased sensitivity to drugs, especially in the brain (10). However, these mice also demonstrated increased expression of the mdr1 mRNA. The normal physiologic role of MDR might be not be evident in these studies if mdr1 activity substitutes for mdr3 activity in these tissues.
Other recent studies have suggested that the normal physiologic role of MDR proteins is to catalyze the transport of normal cellular lipids across membranes. mdr2 is expressed predominantly in bile cannicula (11), and mice with a homozygous disruption of mdr2 develop severe liver disease due to a lack of phosphatidylcholine and cholesterol production into bile (12). More recently, the mdr2 gene has been shown to catalyze the ATP-dependent translocation of phosphatidylcholine across secretory vesicles when expressed in yeast (13). Targeted disruption of the mdr1 (mdr1b) gene in a mouse adrenal cell line decreased the stimulated secretion of steroid hormones from the cell (14). These experiments strongly suggest the MDR proteins normally function to facilitate the transport of normal cellular lipids across membranes.
We have recently demonstrated that MDR activity is re-quired for the esterification of LDL-derived cholesterol. 2 (16) and by preventing the transport of cholesterol from the plasma membrane to the ER (17). The ability of progesterone to inhibit this process results from its ability to inhibit the activity of MDR P-glycoproteins, 2 suggesting that MDR activity is required for proper cholesterol trafficking within cells.
In the accompanying manuscript (18), we demonstrate that progesterone also inhibits late steps in cholesterol biosynthesis, steps that also occur in the ER (19). In the current manuscript we demonstrate that MDR activity is required for cholesterol biosynthesis and suggest a model in which MDR functions in the movement of cholesterol and cholesterol precursors from the plasma membrane to the ER.
MDR Nomenclature-MDR1 refers to the human gene whose product catalyzes drug efflux, while MDR2, also known as MDR3 (25), refers to the human gene whose product does not catalyze drug efflux (2). mdr1 and mdr3 refer to the rodent homologues of the human MDR1 gene. In mice, the mdr1 gene is also called mdr1b, and the mdr3 gene is called mdr1a (10). In hamsters, the mdr1 gene is also called pgp3, the mdr3 gene is called pgp1, and the mdr2 gene is called pgp2 (26).
Cell Growth-Cells were grown in monolayer at 37°C in an atmosphere of 5% CO 2 . CHO-7 cells were grown in medium A (a 1:1 mixture of Ham's F12 medium and Dulbecco's modified Eagle's minimum essential medium containing 100 units/ml penicillin, 100 g/ml streptomycin, and 5% (v/v) newborn calf lipoprotein-deficient serum). Sterols and progesterone were added to the culture medium in ethanol; the final ethanol concentration did not exceed 0.5% (v/v).
Assays-Steroid hydrophobicity was estimated by measuring retention fraction (R f ) on Silica Gel G (Analtech) thin layer chromatography sheets developed in chloroform:methanol (12:1). For cellular assays, cells were plated and grown as described in the individual figure legends. The incorporation of [ 14 C]acetate into cellular lipids by cell monolayers was measured using modifications (18) of the procedure of Brown et al. (22). Vinblastine accumulation was measured as described previously. 2 The amount of [ 3 H]vinblastine accumulated is reported as pmol per mg of total cellular protein.
Cell Fractionation-Cell fractionation was performed as described previously by Lange and Muraski (27) with minor modifications. Briefly, cell monolayers were harvested into ice-cold phosphate-buffered saline. Cell pellets were then washed in cold phosphate-buffered saline and resuspended in 5 mM sodium phosphate, pH 7.5, containing 0.25 M sucrose. After centrifugation at 1,700 ϫ g, the recovered pellet was resuspended in 0.5 mM sodium phosphate, pH 7.5, containing 0.25 M sucrose. A second pellet, obtained following centrifugation at 3,000 ϫ g, was resuspended in 5 volumes of 0.5 mM sodium phosphate, pH 7.5, containing 0.25 M sucrose. Following a 10-min incubation on ice, sodium phosphate was added to a final concentration of 5 mM, and MgCl 2 was added to a final concentration of 0.3 mM. Protease inhibitors were added to final concentrations of 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, and 1 g/ml aprotinin. The suspension was then homogenized by 25 strokes with a Dounce homogenizer using a tight-fitting pestle. The suspension was cleared of unlysed cells and nuclei by two sequential spins at 1,700 ϫ g. High density pellet (P10) and supernatant (S10) fractions were generated after centrifugation at 10,000 ϫ g. Microsomal pellet (P100) and cytosolic (S100) fractions were generated after centrifugation of the S10 fraction at 100,000 ϫ g. The P10 and P100 pellets were resuspended in buffer containing 100 mM Tris-Cl, pH 7.4, 0.25 M sucrose, 1 mM EDTA, 0.2 M NaCl, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 g/ml pepstatin, 0.5 g/ml aprotinin, and 0.5 g/ml leupeptin.

The Progesterone Receptor Is Not Involved in Cholesterol
Biosynthesis-Progesterone is known to interact directly with two proteins within the cell: the progesterone receptor and MDR P-glycoproteins. The progesterone receptor is a member of the steroid hormone receptor family of transcription factors that upon activation enters the nucleus to promote transcription of a class of progesterone-sensitive genes. Progesterone receptor-mediated events typically: 1) occur at nanomolar progesterone concentrations, 2) require a lag time after receptor activation due to the time required for transcriptional changes induced by the receptor to manifest as phenotypic changes, and 3) are prevented by a specific antagonist of the receptor, such as RU 486 (28). To test if progesterone's effects on cholesterol biosynthesis were mediated through the progesterone receptor, we first determined the concentration of progesterone required to inhibit synthesis. CHO cells were preincubated with various concentrations of progesterone for 1 h prior to the addition of [ 14 C]acetate. Following a 2-h labeling, radiolabeled sterols were harvested, resolved using thin layer chromatography (TLC), and visualized by autoradiography (Fig. 1 greater than 95% inhibition was observed only with progesterone concentrations of 10 M and above (lanes 3-6). This concentration is much higher than that which is typically required for progesterone receptor-mediated events. We next tested the ability of a progesterone receptor antagonist to prevent this inhibition. CHO cells were pretreated for 1 h with RU 486 and increasing concentrations of progesterone prior to a 2-h [ 14 C]acetate labeling. RU 486, a specific antagonist of the progesterone receptor, had no detectable effect on cholesterol biosynthesis alone (lane 7) and failed to prevent the inhibition observed with progesterone (lanes 8 -12).
We next tested for the characteristic lag time associated with progesterone receptor-mediated events. CHO cells were pretreated with 40 M progesterone for various amounts of time prior to labeling with [ 14 C]acetate. An autoradiogram of one such experiment is shown in Fig. 2. Again, untreated cells produced mainly cholesterol, some lanosterol, and small amounts of products A and B (lane 1). Cells treated with progesterone produced some lanosterol and products C-F. In this experiment, product E appears to comigrate with cholesterol; however, we have been able to distinguish product E from cholesterol using TLC sheets containing silver nitrate. The same products were formed whether the cells were pretreated with progesterone (lanes 5-10) or progesterone was added simultaneously with the addition of the [ 14 C]acetate (lanes 3 and 4). These findings demonstrate that very little, if any, time is required for progesterone to inhibit cholesterol biosynthesis.
Taken together, these findings demonstrate that progesterone has a direct effect on cholesterol biosynthesis and that the progesterone receptor is not involved.
Effect of Other Steroid Hormones on Cholesterol Biosynthesis-We next tested the ability of other steroid hormones to inhibit cholesterol biosynthesis. Cells were pretreated with various steroid hormones at either 40 or 100 M for 1 h prior to the addition of [ 14 C]acetate. An autoradiogram of the sterols formed in a 2-h labeling is shown in Fig. 3. Most of the steroids tested inhibited cholesterol biosynthesis to some degree. Dehydroepiandrosterone (DHEA; lanes 6 and 7), ␤-estradiol (lanes 12 and 13), pregnenolone (lanes 16 and 17), progesterone (lanes 22 and 23), and deoxycorticosterone (lanes 24 and 25) were potent inhibitors, while androstenediol (lanes 2 and 3), DHEA sulfate (lanes 8 and 9), and cortisol (lanes 14 and 15) had very little effect. High concentrations of the more potent inhibitors resulted in the accumulation of product D, while lower concentrations resulted in the production of product B. We have previously provided evidence that product D is a precursor of lanosterol and product B is 7-dehydrocholesterol (18), the immediate precursor of cholesterol. Precise identification of these compounds requires further study.
Role of Steroid Hydrophobicity-We have found previously that the general hydrophobicity of steroids correlates with their ability to: 1) inhibit MDR activity and 2) inhibit the esterification of LDL-derived cholesterol. 2 We, therefore, tested whether steroid hydrophobicity is a predictor of steroid-mediated inhibition of cholesterol biosynthesis. We evaluated the production of cholesterol, lanosterol, and product D in cells treated with various steroids at either 40 or 100 M (Fig. 4). The rate of sterol production was plotted against the hydrophobicity of the steroid, as measured by its migration (R f ) in TLC. A correlation was observed between the ability of a steroid to inhibit cholesterol synthesis and its hydrophobicity at both 40 and 100 M. Steroids with high R f values were more effective at both decreasing cholesterol production (panels A and B) and increasing lanosterol production (panels C and D). Only steroids with very high R f values resulted in the accumulation of product D (panels E and F).
A strong correlation was observed for 11 of the 12 steroid hormones tested (open circles); trend lines and correlation coefficients for this relationship are indicated. One steroid (closed circle) appeared to be a more potent inhibitor than would be expected based solely on hydrophobicity. This steroid was ␤-estradiol, the active metabolite of estrogen.
Effect of Estrogen on Cholesterol Biosynthesis-Since ␤-es-  Fig. 1. TLC plates were exposed to Amersham Hyperfilm for 2 days at Ϫ70°C.
tradiol was a more potent inhibitor than expected based on hydrophobicity, we first considered the possibility that cells metabolically convert ␤-estradiol to a more hydrophobic compound and, therefore, a more potent MDR inhibitor. To test this hypothesis, we directly measured MDR activity in parallel cultures using a [ 3 H]vinblastine accumulation assay (Fig. 5). Vinblastine is an MDR substrate that passively enters cells and is pumped out by MDR. Steady-state intracellular levels of [ 3 H]vinblastine are dictated by MDR activity; inhibition of MDR increases intracellular accumulation. Cells treated with either 40 M (panel A) or 100 M steroid (panel B) were assayed for the accumulation of cell-associated vinblastine which was plotted against the general hydrophobicity of the steroid (R f ). A strong correlation was observed between vinblastine accumulation and hydrophobicity at both steroid concentrations. This correlation included ␤-estradiol (closed circles); ␤-estradiol inhibited vinblastine accumulation as would be expected by its relative hydrophobicity. These results suggest that the cells do not metabolically convert ␤-estradiol to a more hydrophobic compound and consequently a more potent MDR inhibitor.
To determine whether the effect of the ␤-estradiol on cholesterol biosynthesis was due to estrogen receptor-mediated events, we first measured the concentration of ␤-estradiol required to inhibit cholesterol biosynthesis (Fig. 6A). The concentration of ␤-estradiol required to completely inhibit cholesterol biosynthesis was greater than 10 M (lane 10). This concentration was similar to the concentration of progesterone required to inhibit cholesterol biosynthesis (lane 5) and was higher than the concentrations typically required for estrogen receptor-mediated events. At 100 M, ␤-estradiol completely blocked cholesterol production. However, unlike progesterone (lane 6), ␤-estradiol resulted only in lanosterol accumulation (lane 11); band D was not apparent. In addition, the effect of ␤-estradiol was immediate (data not shown), not requiring any lag time for the transcriptional changes mediated by the receptor to be expressed. These findings suggest that the estrogen receptor is not involved in the ability of ␤-estradiol to inhibit cholesterol biosynthesis.
To test definitively the role of the estrogen receptor in these events, we assayed the effect of tamoxifen on cholesterol biosynthesis. Tamoxifen is a known antagonist of the estrogen receptor and an inhibitor of MDR (29). If the effects of ␤-estradiol are mediated through the estrogen receptor, then tamoxifen should prevent inhibition. If the effects of ␤-estradiol are mediated through MDR, tamoxifen itself should inhibit cholesterol biosynthesis, even in the absence of ␤-estradiol. Concentrations of tamoxifen (1 M), known to antagonize the estrogen receptor, had no effect on the ability of ␤-estradiol to inhibit cholesterol biosynthesis (Fig. 6B). In addition, tamoxifen was capable of inhibiting cholesterol biosynthesis at concentrations similar to those required for ␤-estradiol and progesterone (panel A, lanes 12-16). These results suggest that the ability of ␤-estradiol to inhibit cholesterol biosynthesis is not mediated through the estrogen receptor.
Nonsteroidal MDR Inhibitors Also Inhibit Cholesterol Biosynthesis-With the exception of ␤-estradiol, the observed correlations between cholesterol biosynthesis, vinblastine accumulation, and steroid hydrophobicity raised the possibility that MDR activity is required for cholesterol biosynthesis. To test this more directly, we measured the ability of nonsteroidal MDR inhibitors to prevent cholesterol biosynthesis (Fig. 7). Verapamil, a calcium channel blocker and MDR inhibitor, inhibited cholesterol production at similar concentrations ( lanes  1-4) and resulted in the production of lanosterol and product B (7-dehydrocholesterol). Low concentrations of Triton X-100, an  Fig. 3 were generated on a Howtek Scanmaster 3ϩ using transmitted light. Densiometric analyses of the digital images were performed using the Millipore Bio Image software package. Steroid hydrophobicity was estimated by measuring retention fraction (R f ) on Silica Gel G (Analtech) thin layer chromatography sheets developed in chloroform:methanol (12:1). Each value represents the average of duplicate incubations.

FIG. 5. Correlations between [ 3 H]vinblastine accumulation and steroid hydrophobicity.
Cells were plated and grown through day 2 as described in the legend to Fig. 1. On day 3, the medium was replaced with medium A containing 20 mM HEPES (pH 7.2). One h later, cells were refed 0.5 ml of identical medium containing 50 nM vinblastine sulfate (0.1 Ci/ml) and the indicated concentrations of steroid hormones. The steroids used in this analysis were identical with those described in Fig. 3. Following a 1-h incubation, cell monolayers were washed rapidly 3 times with an ice-cold 1:1 mixture of Ham's F12 medium and Dulbecco's modified Eagle's minimum essential medium containing 100 units/ml penicillin, 100 g/ml streptomycin, and 20 mM HEPES (pH 7.2). The monolayers were then solubilized in 0. MDR inhibitor, also inhibited cholesterol production and increased the production of lanosterol and product B (lanes 9 -12). These findings demonstrate that nonsteroidal inhibitors of MDR also inhibit cholesterol biosynthesis.
Inhibition of Cholesterol Esterification Does Not Inhibit Cholesterol Biosynthesis-The finding that MDR activity is required for cholesterol biosynthesis and our previous findings that MDR activity is required for cholesterol esterification, 2 raise the possibility that the effects of MDR inhibitors on cholesterol biosynthesis result from decreased cholesterol esterification. To test whether an inability to synthesize cholesterol esters interferes with cholesterol biosynthesis, we monitored the incorporation of acetate into sterols in cells treated with CL 283,796, a specific inhibitor of acyl-CoA:cholesterol O-acyltransferase (ACAT), the enzyme involved in cholesterol esterification. Untreated CHO cells produced mainly cholesterol with small amounts of lanosterol and products A and B (Fig. 8,  lane 1). As a control, 40 M progesterone inhibited cholesterol synthesis and resulted in the accumulation of lanosterol and products C-F (lane 2). As a second control, treatment with SKF 104976, a specific inhibitor of lanosterol demethylase, resulted in the accumulation of only lanosterol (lane 3). Treatment with CL 283,796, which inhibited cholesterol ester formation by Ͼ10-fold (data not shown), resulted in only minor changes in cholesterol biosynthesis. There was an accumulation of lanosterol, products A and B, and a major product with mobility only slightly altered from that of cholesterol (lane 4).
Inhibition of Cholesterol Biosynthesis Does Not Inhibit Cholesterol Esterification-To test the reciprocal hypothesis, that preventing normal lanosterol metabolism inhibits cholesterol esterification, we monitored cholesterol esterification in cells grown in the presence of the lanosterol demethylase inhibitor SKF 104976. In the presence of LDL, progesterone inhibited cholesterol oleate formation by 6.5-fold (Table I). In contrast, SKF 104976 had essentially no effect on cholesterol ester formation, even though it completely blocked the conversion of lanosterol to cholesterol (Fig. 8, lane 3). In addition, progesterone inhibited esterification in the presence of SKF 104976 (Table I). As a control, cholesteryl ester formation was 5.5-fold lower in the absence of LDL. These results demonstrate that neither the inhibition of cholesterol synthesis nor the accumulation of lanosterol is responsible for the decreased cholesterol ester production observed in progesterone-treated cells.

DISCUSSION
The Progesterone Receptor Is Not Involved in Progesterone's Inhibition of Cholesterol Biosynthesis-We present three lines of evidence that the progesterone receptor is not involved in progesterone's ability to inhibit cholesterol biosynthesis. First, the concentration of progesterone required for inhibition is much higher than typically required for progesterone receptormediated events. While progesterone receptor-mediated events are typically elicited by nanomolar progesterone concentrations, inhibition of cholesterol biosynthesis occurs at micromolar concentrations. Since CHO cells are derived from steroidogenic tissue, these cells may require unusually high FIG. 6. Effect of progesterone, ␤-estradiol, tamoxifen, and mixtures of ␤-estradiol and tamoxifen on cholesterol biosynthesis. Cells were plated and grown through day 2 as described in the legend to Fig. 1. On day 3, cells were preincubated at 37°C for 2 h in medium containing the indicated additions. In A, the additions of progesterone, ␤-estradiol, and tamoxifen were made to medium A. In B, the indicated additions of ␤-estradiol were made to medium A containing 1 M tamoxifen. Cells were then labeled with 0.5 mM [ 14 C]acetate (95.4 dpm/pmol) for 2 h at 37°C. Radiolabeled sterols were isolated and analyzed by TLC as described in the legend to Fig. 1. TLC plates were exposed to Amersham Hyperfilm for 5 days at Ϫ70°C.  3-10). Additions were made in a staggered fashion so that cultures could be labeled simultaneously with 0.5 mM [ 14 C]acetate (95.4 dpm/pmol). Cells were then incubated at 37°C for an additional 2 h. Radiolabeled sterols were then isolated and analyzed by TLC as described in the legend to Fig. 1. TLC plates were exposed to Amersham Hyperfilm for 4 days at Ϫ70°C.

FIG. 8. Effect of lanosterol demethylase and ACAT inhibitors on cholesterol biosynthesis.
Cells were plated and grown through day 2 as described in the legend to Fig. 1. On day 3, duplicate dishes were refed medium A containing the indicated additions of 30 nM SKF 104976 and 30 M progesterone. After incubation at 37°C for 2 h, [ 14 C]acetate (95.4 dpm/pmol) was added to a final concentration of 0.5 mM and the cells were incubated at 37°C for an additional 2 h. Cells were harvested, and sterols were extracted and resolved by thin layer chromatography (TLC) as described in Fig. 1. TLC plates were exposed to Kodak XAR film for 2 days at Ϫ70°C.
concentrations of progesterone to elicit progesterone receptormediated events due to increased catabolism, decreased uptake or increased efflux of progesterone as compared to non-steroidogenic cells. However, we have found that similar concentrations of progesterone inhibit cholesterol biosynthesis in a large number of cell types not derived from steroidogenic tissue (18). The second line of evidence arguing against a role for the progesterone receptor is that RU 486, a specific antagonist of the progesterone receptor, fails to prevent progesterone-mediated inhibition of cholesterol biosynthesis. Although RU 486 typically functions as an antagonist of the progesterone receptor, RU 486 can also act as a progesterone agonist in the presence of cAMP (30,31). Our experiments, however, were performed under conditions where cAMP levels are low (32). The third observation arguing against a role for the progesterone receptor is that the effect of progesterone is immediate. The only known function of the progesterone receptor is to activate transcription following hormone binding. Secondary events resulting from progesterone receptor activation demonstrate a characteristic lag due to the time required for transcriptional changes to present as phenotypic changes. Since no lag time is observed in progesterone's ability to inhibit cholesterol biosynthesis, progesterone must have a direct role in mediating these events and that the progesterone receptor is not involved.
Role of MDR in Cholesterol Biosynthesis-In addition to the progesterone receptor, progesterone physically interacts with one other class of cellular protein: members of the MDR family of P-glycoproteins. Progesterone has a photoactivatable bond that allows its direct cross-linking to proteins with which it interacts. Studies in multidrug resistance human leukemic lymphoblasts have demonstrated a direct physical interaction between progesterone and MDR (33). Our current studies demonstrate a strong correlation between MDR activity and cholesterol biosynthesis. This correlation is observed with both steroidal and nonsteroidal MDR inhibitors. Since a number of these inhibitors are known to interact with MDR directly, the most likely explanation for our observations is that changes in cholesterol metabolism are a consequence of the effect of these inhibitors on MDR activity and not that cholesterol biosynthesis is required for MDR activity.
␤-Estradiol Is an Exception-We demonstrate a strong correlation between a steroid's general hydrophobicity and its ability to inhibit cholesterol biosynthesis. The only exception to this correlation is ␤-estradiol, which is a more potent inhibitor than expected based solely on hydrophobicity. We have excluded the possibility that cells metabolically convert ␤-estradiol to a more potent MDR inhibitor by demonstrating a direct correlation between MDR activity and steroid hydrophobicity that includes ␤-estradiol. Estrogen has been shown to stimulate cholesterol biosynthesis in estrogen-sensitive cells (34) by activating 3-hydroxy-␤-methylglutaryl-CoA reductase activity (35). This stimulation appears to involve activation of the estrogen receptor since stimulation: 1) takes many hours, 2) occurs at nanomolar estrogen concentrations, and 3) occurs only in estrogen-sensitive cells. This increased cholesterol biosynthesis induced by ␤-estradiol is associated with increased cell proliferation and may be secondary to the increased cholesterol requirements of proliferating cells. Our studies did not reveal estrogen receptor-mediated increases in cholesterol biosynthesis, since our studies were performed following relatively short treatments with ␤-estradiol. Tamoxifen is both an antagonist of the estrogen receptor and an inhibitor of MDR. Our findings that tamoxifen alone inhibits cholesterol biosynthesis, and that tamoxifen fails to prevent the inhibition of cholesterol synthesis induced by ␤-estradiol demonstrates that estrogen receptor-mediated events are not involved in the processes described in the current report.
Relationship between Cholesterol Biosynthesis and Esterification-Since MDR activity is required for both cholesterol biosynthesis and cholesterol esterification, 2 we investigated whether defects in one of these processes causes defects in the other. Inhibiting cholesterol esterification with ACAT inhibitors had no effect on the late steps of cholesterol biosynthesis. Similarly, inhibiting a late step in cholesterol biosynthesis had no effect on cholesterol esterification. These findings suggest that MDR independently affects each of these processes, and that MDR activity may be required for a common process involved in both cholesterol esterification and cholesterol biosynthesis. Since both cholesterol esterification and the late steps in cholesterol biosynthesis occur in the ER and the precursors to these reactions reside in the plasma membrane, we propose that MDR activity is required for the movement of sterols from the plasma membrane to the ER. In fact, progesterone, one of the more potent inhibitors used on our studies, blocks the movement of cholesterol from the plasma membrane to the ER (17). The nature of this transport process is very poorly understood.
MDR Isoforms-Although the current studies demonstrate that MDR activity is required for cholesterol biosynthesis, they do not define which MDR isoform is required. Since cholesterol biosynthesis occurs in virtually all cells of the body, the MDR isoform involved should be expressed widely. The direct correlation between cholesterol biosynthesis and vinblastine accumulation suggests that an isoform capable of catalyzing drug efflux is involved. In humans, isoform-specific mRNA probes have demonstrated that MDR1 is expressed in all of the tissues tested, while MDR2 expression is limited to the liver, kidney, and spleen (36), suggesting that MDR1 may play a role in the general process of cholesterol biosynthesis. We cannot exclude a possible role of mdr2 in cholesterol metabolism since a biologic assay for mdr2 activity has been devised only recently (13), and further studies will be required to determine if the inhibitors used in our studies also affect mdr2 activity. Definitive identification of the isoform involved will come from studies of cholesterol biosynthesis in cells transfected with individual MDR genes.
Mechanism of MDR Action in Cholesterol Metabolism-How MDR activity influences cholesterol metabolism is also not clear. Disruption of one of the mouse MDR genes (mdr2) results in an inability to secrete phospholipids (12), and the mdr2 gene demonstrates the ability to translocate phospholipids across the lipid bilayer when expressed in yeast secretory vesicles (13). One possibility is that MDR-mediated phospholipid trans- location is required for movement of cholesterol from the plasma membrane to the ER. If vesicular transport is involved (37), then phosphatidylcholine translocation across the bilayer may be required for generation of the vesicles. Another possibility is that one of the MDR genes may directly catalyze the translocation of cholesterol across cellular membranes. Although cholesterol has many of the structural characteristics of MDR substrates, catalyzed translocation seems unnecessary since cholesterol moves relatively freely across bilaminar membranes (38). A third possibility is that MDR catalyzes the efflux of a metabolite that normally regulates intracellular cholesterol trafficking; inhibiting efflux of this metabolite would cause its intracellular accumulation and consequent alterations in cholesterol metabolism. The MDR gene family belongs to a superfamily of ATP-binding cassette membrane transporters that includes the yeast STE-6 gene. The STE-6 gene catalyzes the secretion of the prenylated ␣-factor peptide pheromone from yeast cells (7). Mammalian prenylated proteins have been proposed to regulate cholesterol metabolism (39), and, recently, prenylcysteine methyl esters, the degradative breakdown products of these prenylated proteins, have been shown to be substrates of MDR1 (40). These observations raise the possibility that MDR indirectly participates in cholesterol transport by extruding a regulatory prenylated protein or amino acid from the cell. Implications to Coronary Heart Disease-A number of the more potent MDR inhibitors used in our studies previously have been shown to be clinically beneficial in the treatment of coronary heart disease. Premenopausal women have reduced risk for coronary heart disease as compared to men and postmenopausal women. Estrogen and progesterone replacement therapy can prevent the increased risk associated with menopause (41), suggesting that female-specific steroid hormones function to protect women from heart disease. In addition, post-menopausal breast cancer patients treated with tamoxifen demonstrate a reduced risk of heart disease (42). Serum levels of DHEA and its sulfated derivative (DHEA-S) are inversely correlated with cardiovascular death in men (43)(44)(45). DHEA was one of the most potent inhibitors of cholesterol biosynthesis in our studies. Studies in hypercholesterolemic rabbits demonstrate that administration of DHEA decreases the incidence of atherosclerosis (46) and the deposition of cholesterol in the arteries (15). These findings suggest that elevated serum DHEA levels may protect against heart disease by preventing cholesterol deposition, a process that requires cholesterol esterification. Our findings that these agents affect cellular cholesterol metabolism through their ability to inhibit MDR provide a possible mechanistic basis for these observations and raise the possibility that inherited abnormalities at the MDR locus may contribute to coronary heart disease.