Dual roles for in mammalian cells

Fang Xu*, Scott D. Rychnovsky†, Jitendra D. Belani†, Helen H. Hobbs‡§¶, Jonathan C. Cohen*¶, and Robert B. Rawson‡ʈ

*Center for Human Nutrition, ‡Department of Molecular Genetics, §Howard Hughes Medical Institute, and ¶McDermott Center for Human Growth and Development, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390; and †Department of Chemistry, University of California, 516 Rowland Hall, Irvine, CA 92697-2025

Edited by Joseph L. Goldstein, University of Texas Southwestern Medical Center, Dallas, TX, and approved August 26, 2005 (received for review April 30, 2005)

The structural features of sterols required to support mammalian of a mouse fibroblast cell line (S2) that fails to demethylate cell growth have not been fully defined. Here, we use mutant CHO , a late step in the cholesterol synthesis, and is thus a cells that synthesize only small amounts of cholesterol to test the cholesterol auxotroph. These cells grew at reduced rates when capacity of various sterols to support growth. Sterols with minor the plant sterol campesterol (24␣-methylcholesterol) was sub- modifications of the side chain (e.g., campesterol, ␤-sitosterol, and stituted for cholesterol in the medium and failed to grow when ) supported long-term growth of mutant cells, but provided the other major plant sterol, ␤-sitosterol (24␣- sterols with more complex modifications of the side chain, the ethylcholesterol). These data suggest that cholesterol is specif- sterol nucleus, or the 3-hydroxy group did not. After 60 days in ically required for growth of mammalian cells. Because only a culture, the exogenous sterol comprised >90% of cellular sterols. single mutant cell line was examined, it remains possible that Inactivation of residual endogenous synthesis with the other mutations acquired during mutagenesis could be respon- epoxidase inhibitor NB-598 prevented growth in ␤-sitosterol and sible for their specific requirement for cholesterol. greatly reduced growth in campesterol. Growth of cells cultured in To determine the structural features that sterols need to ␤-sitosterol and NB-598 was restored by adding small amounts of support growth of mammalian cells, we used three different cholesterol to the medium. Surprisingly, enantiomeric cholesterol mutant CHO cell lines that make only small amounts of cho- also supported cell growth, even in the presence of NB-598. Thus, lesterol. They fail to activate sterol regulatory element-binding sterols fulfill two roles in mammalian cells: (i) a bulk membrane protein (SREBP) 1 and SREBP-2, membrane-bound transcrip- requirement in which can substitute for cholesterol tion factors that target multiple genes involved in cholesterol and (ii) other processes that specifically require small amounts of synthesis and lipoprotein uptake. SREBPs are made as Ϸ120- cholesterol but are not enantioselective. kDa precursors located in the membranes of the endoplasmic reticulum (ER) and nuclear envelope. When cellular cholesterol ent-cholesterol ͉ phytosterols ͉ NB-598 falls, SREBP cleavage-activating protein (SCAP) escorts SREBP from the ER to the Golgi, where it is sequentially terols are essential components of eukaryote membranes. cleaved by two different enzymes, site-1 protease (S1P) and STheir incorporation enhances the packing of the acyl chains site-2 protease (S2P), to release the transcriptionally active of phospholipids in the hydrophobic phase of the bilayer, in- N-terminal domain, which then goes to the nucleus, where it BIOCHEMISTRY creases its mechanical strength, and reduces its permeability (1). activates genes needed for lipid synthesis (14). Despite this crucial role, most animal species (e.g., nematodes We studied three independent cell lines harboring mutations and arthropods) cannot make sterols and so must get them from in SCAP (SRD-13A cells), S1P (SRD-12B cells), or S2P (M19 the diet. Vertebrates, by contrast, make sterols de novo from cells) (15–17). These mutations block SREBP processing and acetyl-coenzyme A and so do not require exogenous sterols. reduce cholesterol synthesis by Ͼ95%. Because the mutant cells Accordingly, studies to probe sterol requirements of eukaryotes do not survive in media lacking cholesterol and unsaturated fatty have used in invertebrates (2, 3), protazoans (4, 5), or yeast acids, they offer an excellent system for determining the ability strains defective in sterol synthesis (6–8), so that the sterol of various sterols, alone or in combination, to support growth of composition of the organism can be controlled exogenously. mammalian cells. Although phytosterols can account for a substantial portion of Here, we demonstrate sterol synergism in mammalian cells, total dietary sterols (approximately one-third in humans), ver- indicating that sterols also fulfill multiple functions in mammals. tebrates systematically exclude them from the body. Cholesterol A bulk requirement can be met by various closely related sterols, predominates in the membranes of most animals and is virtually including cholesterol, ␤-sitosterol, campesterol, and ent- the exclusive sterol of vertebrates. Invertebrates such as insects cholesterol. The finding that ent-cholesterol (see below) supports typically convert phytosterols to cholesterol by dealkylating C24 growth of mammalian cells suggests that the bulk of cellular in the side chain, thereby furnishing cholesterol needed by sterols serve a membrane function involving interaction with membranes and preventing accumulation of noncholesterol ste- nonenantioselective lipids rather than with proteins, which usu- rols. Mammals and other vertebrates can either make sterols ally are stereospecific. de novo or get cholesterol from the diet. Accumulation of other dietary sterols in these animals is prevented by the action of two Materials and Methods ATP-binding cassette transporters, ABCG5 and ABCG8 (9), Materials. Lipoprotein-deficient FBS (LPDS) (␳ Ͼ 1.215 g͞ml which function as a heterodimer to limit intestinal absorption containing Ͼ2.8 ␮g͞ml cholesterol), sodium mevalonate, sodium and facilitate biliary excretion of noncholesterol sterols (10, 11). oleate, and sodium compactin were prepared as described in refs. Thus, cholesterol comprises the great majority of vertebrate sterols, even in animals ingesting large quantities of phytosterols. The biological basis for selection of cholesterol as the major This paper was submitted directly (Track II) to the PNAS office. sterol in animal membranes is unclear. Bloch (12) proposed that Freely available online through the PNAS open access option. cholesterol is the most effective sterol at modulating membrane Abbreviations: LPDS, lipoprotein-deficient FBS; SREBP, sterol regulatory element-binding properties and that sterols with bulkier side chains owing to protein; SCAP, SREBP cleavage-activating protein; S1P, site-1 protease; S2P, site-2 protease; alkylation at C24 interfered with sterol packing of the acyl chains 25HC, 25-hydroxycholesterol. of membrane phospholipids (1). Silbert and colleagues (13) ʈTo whom correspondence should be addressed. E-mail: [email protected]. examined the ability of noncholesterol sterols to support growth © 2005 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0503590102 PNAS ͉ October 11, 2005 ͉ vol. 102 ͉ no. 41 ͉ 14551–14556 Downloaded by guest on October 1, 2021 18 and 19. ␤-Sitosterol was purchased from Sigma. Enantiomeric was performed by using a 6890N gas chromatograph coupled to a cholesterol, a stereoisomer of natural cholesterol, has the op- 5973 mass selective detector (both from Agilent Technologies, Palo posite configuration at every stereogenic (‘‘chiral’’) center (nat- Alto, CA). The trimethylsilyl-derived sterols were separated on an ural cholesterol is 3R, 8S, 9S, 10R, 13R, 14S, 17R, 21R; HP-5MS 5%-phenyl methyl polysiloxane capillary column (30-m ϫ ent-cholesterol is 3S, 8R, 9R, 10S, 13S, 14R, 17S, 21S). It was 0.25-mm inside diameter ϫ 0.25-␮m film) with carrier gas helium made as described in ref. 20. All other sterols were from attherateof1ml͞min. The temperature program was 150°C for Steraloids (Newport, RI). Purity of the sterols was determined 2 min, followed by increasing the temperature by 20°C per min up by GC͞MS. The squalene epoxidase inhibitor NB-598 was a gift to 280°C and holding it for 13 min. The injector was operated in the from Banyu Pharmaceutical (Tokyo) or was purchased from splitless mode and was kept at 280°C. The mass spectrometer was Sigma. Each lot was prepared as a nominally 10 mM stock in operated in selective ion monitoring mode. The extracted ions were DMSO and titered to determine the lowest concentration that 458.4 (cholesterol), 343.3 (desmosterol), 458.4 (), 456.4 consistently killed all WT cells in the absence of added sterols (zymosterol), 382.4 (campesterol), 393.4 (lanosterol), and 396.4 under our assay conditions (ranging from 0.5 to 3.0 ␮M). (␤-sitosterol).

Cultured Cells. Cells were maintained in monolayer culture at Results 37°C in a 9% CO2 incubator. CHO-7 cells are a subline of Noncholesterol Sterols Support Growth of Mammalian Cells. The CHO-K1 cells selected for growth in LPDS (21). They are mutant cell lines used in this study are maintained in medium maintained in medium A (a 1:1 mixture of Ham’s F12 medium containing 5 ␮g͞ml cholesterol to support optimum growth (17, and DMEM containing 100 units͞ml penicillin and 100 ␮g͞ml 22, 23). To determine the minimum dose of exogenous choles- streptomycin sulfate) supplemented with 5% (vol͞vol) LPDS terol required to support growth of these cells, cells were grown (18, 19). M19 cells (22), SRD-12B cells (23), and SRD-13A cells in media containing between 0 and 5 ␮g͞ml cholesterol. All (17) are previously described mutant CHO cells auxotrophic for three cell lines showed substantial growth at 2 or 3 ␮g͞ml cholesterol, mevalonate, and unsaturated fatty acid due to a cholesterol (see the supporting information, which is published deficiency of S2P (15), S1P (16), or SCAP (17), respectively. on the PNAS web site); thus, these concentrations were used for They were maintained in medium B (medium A supplemented subsequent studies. Next, we tested a panel of different sterols with 5% FBS͞1 mM sodium mevalonate͞20 ␮M sodium for the ability to support growth of the three cell lines. Specific oleate͞5 ␮g/ml cholesterol). The stock cultures of mutant cells sterols were selected to examine the features of the side chain, were selected with amphotericin B weekly to maintain the the sterol nucleus, and the 3-hydroxy group that are required for mutant phenotype (23). cell growth (Fig. 1). WT cells grew well in the absence of sterols and in the presence of all sterols tested except cholestenone, a Growth Assays. For assays, cells were plated on day 0 at a density cholesterol metabolite with a hydroxy group at C3 and a ⌬4 of 30,000 cells per well in six-well plates in medium A (for WT) double bond, and lanosterol. This biosynthetic precursor to or medium B (for mutants). On day 1, cells were washed with cholesterol contains three methyl groups (4, 4Ј, and 14) and a PBS and refed with medium A supplemented with LPDS, 1 mM 5␤-H. Thus, the conformation of its A͞B ring system is different sodium mevalonate, 20 ␮M sodium oleate, and various sterols as from that of cholesterol. indicated in the figure legends. Cells were refed every 1–2 days. Absent added sterols, auxotrophic cell lines failed to grow On day 10, cells were washed, fixed in 95% ethanol, and stained (Fig. 1). Cholesterol and its immediate precursor, desmosterol, with crystal violet or analyzed for sterol content as described supported growth equally well. Sterols with minor modifications below. of the side chain, including unsaturation (desmosterol) or meth- For long-term growth assays (Ͼ60 days), cells were plated at ylation or ethylation of C24 (campesterol and ␤-sitosterol), 500,000 per 100-mm plate and fed every 2–3 days. Once each supported robust growth. Cells that were fed sterols with more week, sets of cells were split 1:5 and 1:10 into fresh plates. complex side-chain modifications () or lacking the To quantify growth, stained plates were scanned on a UMAX side chain (androsterol) showed little or no growth (Fig. 1). Astra 4000U scanner (UMAX Technologies, Dallas) in trans- Under our assay conditions, cholesterol and ␤-sitosterol sup- mitted light mode at a resolution of 300 dots per inch. The images ported similar growth rates in mutant and WT cells (see the were analyzed by using NIH IMAGEJ software to determine the supporting information). average pixel value from a circular area of 64,744 pixels, corre- Lathosterol, the ⌬7 isomer of cholesterol, supported cell sponding to one-half the area of the well, centered on the well. growth (Fig. 1), but the cells efficiently metabolized it to The resultant values were subtracted from 255 to yield arbitrary cholesterol (see below). Although the three auxotrophic cell units corresponding to the degree of growth at each condition lines respond similarly to most sterols tested, for some, the S1PϪ tested. This method yielded results comparable (r2 ϭ 0.933) to cells respond differently than do S2PϪ or SCAPϪ cells (e.g., poor quantitation of total cellular protein by standard solution assay growth in desmosterol and no growth in lathosterol). We do not (BCA Protein Assay kit, Pierce). currently understand the basis of these differences among the auxotrophic cell lines. Determination of Cellular Sterol Composition. On day 10, cells were Modification of the sterol nucleus by 5␣ saturation (dihydro- washed once with PBS, harvested by scraping in PBS, and cholesterol) or isomerization of the double bond from ⌬5to⌬4 transferred to a 15-ml conical tube. After centrifugation, the cell (4-cholesten-3-␤-ol) resulted in very poor cell growth (Fig. 1). pellet was washed once with PBS. After removal of the super- An intact 3-hydroxy group proved essential for sterols to support natant, the pellet was frozen in liquid nitrogen and stored at growth. When this group was oxidized (5-cholesten-3-one), Ϫ80°C until analyzed. deleted (5-cholestene), or epimerized (5-cholesten-3␣-ol), the For analysis, we resuspended cell pellets in 1 ml of 0.1 M NaOH auxotrophs failed to grow (Fig. 1). Therefore, a planar nucleus and vortexed them for 30 min. We determined protein concentra- and an unblocked equatorial hydroxyl group are essential to the tion as above. An aliquot of ethanol containing the internal function of sterols in mammalian cells. standards 5␣-cholestane (50 ␮g) and epicoprostanol (2.5 ␮g) was added to 200 ␮l of cell lysate, and sterols were hydrolyzed by heating Added Sterols Are Major Sterols of Auxotrophic Cells. To determine (to 100°C) in ethanolic KOH (100 mM) for 2 h. Lipids were whether the mutant cells converted noncholesterol sterols to extracted in petroleum ether, dried under nitrogen, and derivatized cholesterol, we analyzed cellular sterol content by GC͞MS (see with hexamethyldisilazane-trimethylchlorosilane. GC͞MS analysis the supporting information). In WT cells, cholesterol was the

14552 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0503590102 Xu et al. Downloaded by guest on October 1, 2021 Fig. 1. Growth of cholesterol auxotrophs in noncholesterol sterols. On day 1, cells (WT and auxotrophs) were fed with medium A containing 5% (vol͞vol) LPDS, 1 mM mevalonate, 20 ␮M oleate, and the indicated sterol at 2 ␮g͞ml in ethanol (final concentration of 0.2% ethanol for all media). Cells were refed every 2 days thereafter until day 10, when cells were fixed and stained.

predominant sterol under all conditions tested. Exogenous ste- the auxotrophic cells at lower doses of sterol, growth was similar rols comprised Ͻ38% of total cellular sterols after 10 days of to that of WT cells when ␤-sitosterol or campesterol was added growth, presumably reflecting endogenous synthesis of choles- at a concentration of 3 ␮g͞ml (Fig. 2). Stigmasterol supported terol in WT cells. The accumulation of desmosterol when these growth of auxotrophs less well than did the other phytosterols cells are fed stigmasterol likely reflects its competitive inhibition and showed a weak inhibitory effect on WT cells. of sterol ⌬24-reductase, the enzyme that converts desmosterol to cholesterol (24). In the mutant cell lines, the exogenous sterols Toxic Sterols. Lanosterol and cholestenone not only failed to comprised 64–96% of total cellular sterols after 10 days. Thus, support growth of the cholesterol auxotrophs but also killed WT exogenous sterols were not converted into cholesterol, except in cells (Fig. 1). Previous studies showed that 25-hydroxycholes- cells fed lathosterol, where cholesterol comprised Ͼ60% of total terol (25HC), which also kills WT cells, suppresses the cleavage sterol. Basal activity of lathosterol 5-desaturase in these cells may of SREBPs, resulting in a transcriptional deficit of cholesterol be sufficient to catalyze the conversion of lathosterol to choles- biosynthetic genes (27). The toxic effects of 25HC (1 ␮g͞ml) can BIOCHEMISTRY terol even in the absence of SREBP-mediated transcription. be overcome by providing exogenous cholesterol to the cells (21). Alternatively, the reaction may be catalyzed by other enzymes in We thus tested the ability of cholesterol to rescue cells from the hamster cells. toxic effects of lanosterol and cholestenone. WT cells grew well To further assess the responses of the auxotrophic cell lines to in medium supplemented with LPDS both in the absence of noncholesterol sterols, we performed dose–response experi- exogenous sterols and in the presence of 7-dehydrocholesterol ments with the sterols that supported growth: cholesterol, (cholesta-5,7-diene-3␤-ol). Addition of cholestenone or lanos- ␤-sitosterol, campesterol, and stigmasterol (Fig. 2). Desmosterol terol at 2 ␮g͞ml killed the cells (Fig. 3A). Cholesterol supplied had previously been shown to support growth of rodent cells in as free cholesterol (5 ␮g͞ml in ethanolic solution) or as LDL the absence of cholesterol (25, 26) and was not tested further. from complete FBS, rescued cell growth in the presence chole- Although significant variation in growth was observed among stenone (Fig. 3A) and lanosterol (Fig. 3B). In contrast with 1 ␮g͞ml 25HC, at 2 ␮g͞ml, 25HC was toxic under the conditions tested, even when cholesterol was added to the media, which may reflect actions of this compound on processes other than SREBP cleavage. To determine whether the toxicity of cholestenone was me- diated by a mechanism similar to that observed for 25HC, we tested its ability to repress cleavage of SREBP-2. In this exper- iment, cells were cultured overnight in the presence of compactin (an inhibitor of 3-hydroxy-3-methyglutaryl-coenzyme A reduc- tase that induces SREBP cleavage owing to cholesterol deple- tion) and mevalonate and then treated with the indicated sterol at 2 ␮g͞ml for 5 h before harvest. Cells not treated with sterol (Fig. 3C, lanes 1 and 5) or treated with 7-dehydrocholesterol (lane 2) or cholestenone (lane 3) showed no suppression of SREBP-2 cleavage. Cholesterol delivered in ethanolic solution also failed to suppress cleavage of SREBPs, most likely because Fig. 2. Effect of sterol concentration on cell growth. On day 1, cells (WT and of its very low solubility. It does suppress cleavage when deliv- auxotrophs) were fed with medium containing 5% (vol͞vol) LPDS, 1 mM ␤ mevalonate, 20 ␮M oleate, and the sterol to be tested at the indicated ered as a complex with methyl- -cyclodextrin (28). 25HC deliv- concentration (final concentration of 0.3% ethanol for all media). Cells were ered in ethanol abolished cleavage (lane 4). These results suggest refed every 2 days thereafter until day 10, when cells were fixed and stained. that the mechanism by which cholestenone kills WT cells is Growth was quantitated as described and plotted as arbitrary units. distinct from that of 25HC and that it does not involve acute

Xu et al. PNAS ͉ October 11, 2005 ͉ vol. 102 ͉ no. 41 ͉ 14553 Downloaded by guest on October 1, 2021 Fig. 3. Effects of sterols on SREBP-2 cleavage. On day 1, WT cells were fed with medium A containing 5% (vol͞vol) LPDS, 1 mM mevalonate, 20 ␮M oleate, and the indicated sterols at 3 ␮g͞ml (final concentration of 0.3% ethanol for all media). Cells were refed every 2 days thereafter. (A) Cells were fixed and stained on day 5. (B) Cells were fixed and stained on day 10. (C)On day 0, WT cells were set up in medium A at 500,000 per 10-cm dish. On day 2, Fig. 4. Growth in the presence of the squalene epoxidase inhibitor NB-598. cells were refed inducing medium [medium A containing 5% (vol͞vol) LPDS, (A) Cholesterol content of WT and auxotrophic cells. On day 0, cells were set 50 ␮M compactin, and 50 ␮M mevalonate (17)] or the same medium further up as described in Materials and Methods. On day 1, cells were washed and ͞ ͞ supplemented with 10 ␮g͞ml cholesterol and a 2 ␮g͞ml concentration of the refed medium A supplemented with 3% (vol vol) LPDS 1 mM sodium meva- ͞ ␮ ͞ ␮ indicated sterol. All media were adjusted to 0.3% ethanol. After 4.5 h, lonate 20 M sodium oleate 3 g/ml of the indicated sterol. Cells received Ϫ ␮ ϩ N-acetyl-leucinyl-leucinyl norleucinal (ALLN) was added to a final concentra- vehicle alone ( )or3 M NB-598 ( ). Each condition was set up in quadru- tion of 50 ␮g͞ml. Cells were harvested 90 min later and lysed in SDS buffer [10 plicate. On day 10, two wells were harvested for sterol analysis, and the mM Tris⅐HCL, pH 7.6͞100 mM NaCl͞1% SDS containing protease inhibitors remaining wells were fixed and stained as described. Sterol content could not pepstatin A (5 ␮g͞ml), lupeptin (10 ␮g͞ml), ALLN (25 ␮g͞ml), aprotinin (1,000 be determined for samples having no surviving cells. The graph shows total units͞ml), and phenymethylsulfonyl fluoride (0.5 mM)]. Aliquots (50 ␮gof cholesterol content for each sample. Error bars indicate the range of values. protein) were analyzed by SDS͞PAGE and Western blotting with 5 ␮g͞ml Below each sample is a fixed and stained dish showing growth. (B) Cholesterol monoclonal antibody 7D4 against hamster SREBP-2 (39). P, membrane-bound stimulation of growth. Cells were set up as in A, but the medium contained 0.5 ␮ SREBP-2 precursor; N, cleaved, nuclear SREBP-2. M NB-598 and the indicated concentrations of cholesterol in the presence or absence of 3 ␮g͞ml ␤-sitosterol. Each condition was set up in triplicate. Cells were refed every 1–2 days until day 10, when cells were fixed and stained. Growth was quantified as described. The standard error of the mean for all suppression of SREBP cleavage. Lanosterol also fails to suppress samples ranged between 0.4 and 7.0. (C) Cellular cholesterol content (per- cleavage of SREBPs under these conditions, and its cytotoxic centage of total sterols) versus the cholesterol composition of growth medium ␤ effects are not due to acute suppression of SREBP cleavage (29). (percentage of total sterols) for the cells that received -sitosterol. Cellular sterols were measured by GC͞MS as described. Total sterols ranged from 0.442 to 5.533 ␮g͞well. r2 ϭ 0.985 by linear regression analysis. Small Amounts of Cholesterol Are Essential for Cell Growth. Even after 60 days in cholesterol-free medium containing 2 ␮g͞ml campesterol, cholesterol comprised 3% of total sterols in the Requirement for Cholesterol Is Not Enantiospecific. Interactions auxotrophic cells. These trace levels of cholesterol presumably between sterols and proteins tend to be enantioselective, reflect residual endogenous synthesis and are not sufficient to whereas those between sterols and membrane lipids show little support cell growth (Figs. 1 and 2). To determine whether or no enantioselectivity. To define the stereospecificity of the ␤-sitosterol and campesterol can support cell growth in the cellular sterol requirement, we grew cells in the presence of absence of cholesterol, WT and SCAPϪ cells were grown in the ent-cholesterol that has the opposite configuration at every chiral presence of a squalene epoxidase inhibitor, NB-598 (30). NB-598 center as compared with native cholesterol (20). Auxotrophic blocks cholesterol synthesis but does not block formation of early cells grew equally well when either cholesterol or ent-cholesterol was added to the medium (Fig. 5B). In the presence of NB-598, intermediates in the cholesterol synthetic pathway (e.g., isopre- Ϫ noids) that are required for cell growth. Cholesterol supported WT and SCAP cells grew equally well in both enantiomers (Fig. growth of both WT and SCAPϪ cells grown in NB-598 (Fig. 4); 5C). Ent-cholesterol can satisfy the bulk requirement that can also be met by various modified sterols (e.g., campesterol and however, growth was markedly reduced when cells were grown ␤ in campesterol plus NB-598. ␤-Sitosterol alone did not support -sitosterol) and meet the specific requirement for small growth of either cell line in the presence of NB-598 (Fig. 4A). amounts of cholesterol itself (see the supporting information). Therefore, in mammalian cells, the essential functions of cho- This finding suggests that small amounts of cholesterol are lesterol do not depend on its absolute configuration. essential for cell growth, even when cells are provided with permissive doses of noncholesterol sterols such as ␤-sitosterol. Discussion To test this hypothesis, we added increasing concentrations of Our major finding is that the cellular requirement for sterols is cholesterol to the media of cells grown in the presence of NB-598 highly specific with respect to structure but is not enantioselec- ␮ ͞ ␤ and in the presence or absence of 3 g ml -sitosterol (Fig. 4B). tive. Sterols with complex side chains or with modifications of Addition of cholesterol at levels that are themselves too low to the nucleus or the 3-hydroxy group did not support growth in support growth markedly enhanced growth of cells provided Ϫ three cell lines that require exogenous sterols. The major plant with ␤-sitosterol. Similar results were observed for SCAP cells sterols, ␤-sitosterol and campesterol, supported growth similarly (not shown). The cellular sterol composition in these experi- to cholesterol (Figs. 1 and 2) without being converted to ments closely reflected the composition of sterols added to the cholesterol. Thus, these phytosterols can substitute for choles- media (r2 ϭ 0.99; Fig. 4C). Thus, mammalian cells have a specific terol as the primary sterol in the plasma membrane. However, requirement for cholesterol that cannot be fulfilled by other the cholesterol auxotrophs invariably contained small amounts sterols. of cholesterol (Ͻ10% of total sterol), even after 60 days in

14554 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0503590102 Xu et al. Downloaded by guest on October 1, 2021 B, and C, and one five-membered ring denoted D (Fig. 1). The B–C and C–D ring fusion occurs at single bonds whose substitu- ents are trans, allowing the ring system to adopt a planar conformation. The double bond at the AB ring junction prevents the A and B rings from adopting full chair conformations. This ⌬5 double bond is essential for cell growth; dihydrocholesterol, the 5␣-reduced derivative of cholesterol, did not support growth of the auxotrophs in this study, even though it retains the coplanar conformation of the A and B cyclohexyl rings. A mutant mouse fibroblast line (LM) in which sterol synthesis is blocked at desmosterol grew at reduced rates when supple- mented with dihydrocholesterol. In that study, the mutant LM cells adapted to growth in dihydrocholesterol by desaturating and elongating the fatty acids in their membrane phospholipids (13). The enzymes that thus modify fatty acids are SREBP targets. The auxotrophs lack active SREBP, so the failure of dihydrocholesterol to support even modest growth may result from an inability of the auxotrophs to alter the fatty acid composition of their phospholipids. Cyclization of squalene epoxide to lanosterol generates a rigid tetracyclic ring system with a flexible isooctyl side chain that is retained in all naturally occurring sterols. Our results show that the side chain is essential for growth of mammalian cells. Androsterol, which has no side chain, did not support growth. Fig. 5. Requirement for cholesterol is not enantiospecific. (A) Cholesterol Modification of the side chain had varying effects; C-24- (Upper) and ent-cholesterol (Lower), its mirror image. (B) Ent-cholesterol alkylated sterols with saturated side chains (campesterol and supports growth of auxotrophic cells. The experiment was conducted as ␤ described in the legend of Fig. 1. (C) Ent-cholesterol supports growth in the -sitosterol) supported robust growth. Stigmasterol (unsatur- presence of NB-598. The experiment was conducted as described in the legend ated at C-22) supported growth only poorly. of Fig. 4. Error bars indicate the range of values. Below each sample is a fixed An unanticipated finding of this study was that the precursor and stained dish showing growth. of sterols in fungi and animals, lanosterol, was toxic to WT cells. Toxicity was not observed with later stage precursors such as lathosterol or desmosterol. Cholesterol rescued lanosterol tox- culture in the presence of noncholesterol sterols, owing to icity, indicating that lanosterol inhibited cholesterol synthesis. residual cholesterol synthesis. Similar findings were observed with the synthetic sterol chole- We therefore examined the effect of inhibition of cholesterol stenone. The toxic sterol 25HC kills cells by inhibiting SREBP synthesis with the squalene epoxidase inhibitor NB-598. This cleavage, thereby indirectly suppressing cholesterol synthesis. BIOCHEMISTRY treatment completely blocked growth of cells supplemented with However, neither lanosterol (not shown) nor cholestenone sup- ␤-sitosterol and markedly reduced growth of cells cultured in pressed cleavage of SREBP-2 (Fig. 5C). These data suggest that campesterol. Growth of cells cultured in the presence of the lanosterol and cholestenone inhibit cholesterol synthesis by an inhibitor NB-598 was restored when the media were supple- SREBP-independent pathway. Recently, Song et al. (29) dem- mented with cholesterol or ent-cholesterol. These data indicate onstrated that lanosterol accelerates degradation of 3-hydroxy- that mammalian cells can use several naturally occurring sterols 3-methyglutaryl (HMG)-coenzyme A (CoA) reductase, an es- in their plasma membranes but that small amounts of cholesterol sential enzyme in cholesterol biosynthesis. This effect is or ent-cholesterol are required for an unidentified, essential independent of SREBP cleavage. In contrast to lanosterol, process. We cannot exclude the possibility that the small cholestenone does not promote degradation of HMG-CoA amounts of cholesterol remaining in auxotrophs and cells treated reductase (29). Deciphering how cholestenone inhibits choles- with NB-598 (Fig. 5A) may be essential to yet other processes in terol synthesis may thus provide insight into the mechanisms which neither phytosterols nor ent-cholesterol can function. regulating cholesterol homeostasis. All naturally occurring sterols include a planar ring system, an The C24-alkylated sterols campesterol and ␤-sitosterol are aliphatic side chain, and a hydroxyl group at the C-3 position (1). abundant in nature and are the major dietary sterols of herbi- The 3␤-OH group confers an amphiphilic character on the vores. Because these sterols meet the bulk membrane require- otherwise hydrophobic cholesterol molecule and thereby orients ment of mammalian cells, why is this class of sterols so assidu- it in membranes. Whereas 3␤-OH sterols are ubiquitous in ously excluded from accumulating in vertebrates? Our data nature, Bloch and his colleagues (32, 33) found that methyl strongly suggest that the adverse effects of plant sterols are not cholesterol ethers supported growth of the sterol requiring a consequence of disruption of membrane integrity. Rather, prokaryote Mycoplasma capricolum and of a sterol auxotrophic certain noncholesterol sterols may disrupt cholesterol homeosta- mutant yeast strain (6, 33). Thus, Bloch questioned whether the sis. The accumulation of noncholesterol sterols in the adrenal 3␤-OH is mandatory for sterol function or simply arises as a glands of mice lacking ABCG5 and ABCG8 causes profound consequence of sterol synthesis. 5-Cholestene, cholestenone, and alterations in cholesterol metabolism. The cholesterol content of 3␣-epicholesterol did not support growth of auxotrophic cells, the adrenal glands in these animals is reduced 90%. Despite this which agrees with previous work in which cholesteryl methyl- pronounced reduction, these animals fail to up-regulate the ether and 3␣-epicholesterol failed to support growth of a mac- cholesterol homeostatic machinery (38). When these animals rophage-like cell line in lipid-depleted medium (34). These were treated with ezetimibe, a drug that blocks sterol absorption, findings suggest that an equatorial hydroxyl group at C-3 is the content of noncholesterol sterols in the adrenal gland essential for mammalian cells. decreased sharply, and cholesterol homeostasis approached A planar ring system is also a common feature of sterols that normal. Thus, the active exclusion of noncholesterol sterols support mammalian cell growth. Cholesterol has a tetracyclic from the body may be needed to maintain normal cholesterol ring structure consisting of three six-membered rings, denoted A, homeostasis.

Xu et al. PNAS ͉ October 11, 2005 ͉ vol. 102 ͉ no. 41 ͉ 14555 Downloaded by guest on October 1, 2021 Studies of sterol requirements in the prokaryotic sterol auxo- The function of the small quantities of cholesterol required by troph M. capricolum showed that these cells grew better when auxotrophs to sustain growth remains elusive. It may be that supplied with both cholesterol and lanosterol than with either cholesterol plays an essential structural role that cannot be met sterol alone (35). These findings were taken to indicate that by C24-alkylated sterols. Alternatively, cholesterol may be a sterols play more than one role in membranes. Later, sterol precursor of an essential metabolite that cannot be made from synergism was demonstrated in unicellular eukaryotes including C24-alkylated sterols. Ent-cholesterol rescued growth of auxo- yeast (36) and Paramecium (37). Yeast mutants that require trophs in the presence of NB-598, suggesting that the essential exogenous sterols can grow in a variety of sterols if trace amounts of are present. Analysis of cellular free sterol levels function of cholesterol is not mediated by an enantiospecific and plasma membrane properties suggested that sterols such as protein. Further studies will be required to elucidate the cellular cholesterol and dihydrocholesterol satisfy a bulk membrane process or processes that specifically require cholesterol to the requirement, whereas ergosterol serves a highly specific function exclusion of other sterols. (35). A similar phenomenon was observed in the natural auxo- troph Paramecium tetraurelia, which has an absolute requirement We thank Joe Lockridge for excellent technical assistance; Angela for one of a small group of structurally related phytosterols Carroll for invaluable help with tissue culture; and Richard Auchus, including stigmasterol, ␤-sitosterol, and poriferasterol (37). Stig- Mike Brown, and Joe Goldstein for helpful discussions. This work was masterol at levels as low as 20 ng͞ml supports growth if a second supported by the Howard Hughes Medical Institute, National Institutes relatively nonspecific sterol is provided at higher concentrations of Health Grants HL20948 and HL72304, and the Donald W. Reynolds (1 ␮g͞ml). Cardiovascular Clinical Research Center (Dallas).

1. Bloch, K. E. (1983) CRC Crit. Rev. Biochem. 14, 47–92. 22. Hasan, M. T., Chang, C. C. & Chang, T. Y. (1994) Somatic Cell Mol. Genet. 20, 2. Cooke, J. & Sang, J. H. (1970) J. Insect Physiol. 16, 801–812. 183–194. 3. Svoboda, J. A. & Weirich, G. F. (1995) Lipids 30, 263–267. 23. Rawson, R. B., Cheng, D., Brown, M. S. & Goldstein, J. L. (1998) J. Biol. Chem. 4. Nes, W. R., Joseph, J. M., Landrey, J. R. & Conner, R. L. (1978) J. Biol. Chem. 273, 28261–28269. 253, 2361–2367. 24. Fernandez, C., Suarez, Y., Ferruelo, A. J., Gomez-Coronado, D. & Lasuncion, 5. Nes, W. R., Joseph, J. M., Landrey, J. R. & Conner, R. L. (1980) J. Biol. Chem. M. A. (2002) Biochem. J. 366, 109–119. 255, 11815–11821. 25. Rothblat, G. H., Burns, C. H., Conner, R. L. & Landrey, J. R. (1970) Science 6. Buttke, T. M. & Bloch, K. (1981) Biochemistry 20, 3267–3272. 169, 880–882. 7. Buttke, T. M. & Bloch, K. (1980) Biochem. Biophys. Res. Commun. 92, 229–236. 26. Wechsler, A., Brafman, A., Shafir, M., Heverin, M., Gottlieb, H., Damari, G., 8. Buttke, T. M., Jones, S. D. & Bloch, K. (1980) J. Bacteriol. 144, 124–130. Gozlan-Kelner, S., Spivak, I., Moshkin, O., Fridman, E., et al. (2003) Science 9. Berge, K. E., Tian, H., Graf, G. A., Yu, L., Grishin, N. V., Schultz, J., 302, 2087. Kwiterovich, P., Shan, B., Barnes, R. & Hobbs, H. H. (2000) Science 290, 27. Goldstein, J. L., Rawson, R. B. & Brown, M. S. (2002) Arch. Biochem. Biophys. 1771–1775. 397, 139–148. 10. Yu, L., Hammer, R. E., Li-Hawkins, J., Von Bergmann, K., Lutjohann, D., 28. Adams, C. M., Reitz, J., De Brabander, J. K., Feramisco, J. D., Li, L., Brown, Cohen, J. C. & Hobbs, H. H. (2002) Proc. Natl. Acad. Sci. USA 99, 16237–16242. M. S. & Goldstein, J. L. (2004) J. Biol. Chem. 279, 52772–52780. 11. Graf, G. A., Yu, L., Li, W. P., Gerard, R., Tuma, P. L., Cohen, J. C. & Hobbs, 29. Song, B.-L., Javitt, N. B. & DeBose-Boyd, R. A. (2005) Cell Metab. 1, 179–189. H. H. (2003) J. Biol. Chem. 278, 48275–48282. 30. Horie, M., Tsuchiya, Y., Hayashi, M., Iida, Y., Iwasawa, Y., Nagata, Y., 12. Bloch, K. E. (1979) CRC Crit. Rev. Biochem. 7, 1–5. Sawasaki, Y., Fukuzumi, H., Kitani, K. & Kamei, T. (1990) J. Biol. Chem. 265, 13. Rujanavech, C., Henderson, P. A. & Silbert, D. F. (1986) J. Biol. Chem. 261, 18075–18078. 7196–7203. 31. Crowder, C. M., Westover, E. J., Kumar, A. S., Ostlund, R. E., Jr., & Covey, 14. Brown, M. S. & Goldstein, J. L. (1997) Cell 89, 331–340. 15. Rawson, R. B., Zelenski, N. G., Nijhawan, D., Ye, J., Sakai, J., Hasan, M. T., D. F. (2001) J. Biol. Chem. 276, 44369–44372. Chang, T. Y., Brown, M. S. & Goldstein, J. L. (1997) Mol. Cell 1, 47–57. 32. Odriozola, J. M., Waitzkin, E., Smith, T. L. & Bloch, K. (1978) Proc. Natl. Acad. 16. Sakai, J., Rawson, R. B., Espenshade, P. J., Cheng, D., Seegmiller, A. C., Sci. USA 75, 4107–4109. Goldstein, J. L. & Brown, M. S. (1998) Mol. Cell 2, 505–514. 33. Lala, A. K., Buttke, T. M. & Bloch, K. (1979) J. Biol. Chem. 254, 10582– 17. Rawson, R. B., DeBose-Boyd, R., Goldstein, J. L. & Brown, M. S. (1999) J. Biol. 10585. Chem. 274, 28549–28556. 34. Esfahani, M., Scerbo, L. & Devlin, T. M. (1984) J. Cell. Biochem. 25, 87–97. 18. Brown, M. S., Faust, J. R. & Goldstein, J. L. (1978) J. Biol. Chem. 253, 35. Dahl, C. E., Dahl, J. S. & Bloch, K. (1980) Biochemistry 19, 1462–1467. 1121–1128. 36. Ramgopal, M. & Bloch, K. (1983) Proc. Natl. Acad. Sci. USA 80, 712–715. 19. Goldstein, J. L., Basu, S. K. & Brown, M. S. (1983) Methods Enzymol. 98, 37. Whitaker, B. D. & Nelson, D. L. (1988) J. Gen. Microbiol. 134, 1441–1447. 241–260. 38. Yang, C., Yu, L., Li, W., Xu, F., Cohen, J. C. & Hobbs, H. H. (2004) J. Clin. 20. Rychnovsky, D. E. & Mickus, D. E. (1992) J. Org. Chem. 57, 2732–2736. Invest. 114, 813–822. 21. Metherall, J. E., Goldstein, J. L., Luskey, K. L. & Brown, M. S. (1989) J. Biol. 39. Yang, J., Brown, M. S., Ho, Y. K. & Goldstein, J. L. (1995) J. Biol. Chem. 270, Chem. 264, 15634–15641. 12152–12161.

14556 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0503590102 Xu et al. Downloaded by guest on October 1, 2021