Differential Stimulation of Cholesterol and Unsaturated Fatty Acid Biosynthesis in Cells Expressing Individual Nuclear Sterol Regulatory Element-binding Proteins*

Three sterol regulatory element-binding proteins (SREBP-1a, -1c, and -2) stimulate transcription of genes involved in synthesis and receptor-mediated uptake of cholesterol and fatty acids. Here, we explore the individual roles of each SREBP by preparing lines of Chinese hamster ovary (CHO) cells that express graded amounts of nuclear forms of each SREBP (designated nSREBPs) under control of a muristerone-inducible nuclear receptor system. The parental hamster cell line (M19 cells) lacks its own nSREBPs, owing to a deletion in the gene encoding the Site-2 protease, which releases nSREBPs from cell membranes. By varying the concentration of muristerone, we obtained graded expression of individual nSREBPs in the range that restored lipid synthesis to near physiologic levels. The results show that nSREBP-2 produces a higher ratio of synthesis of cholesterol over fatty acids than does nSREBP-1a. This is due in part to a selective ability of low levels of nSREBP-2, but not nSREBP-1a, to activate the promoter for squalene synthase. nSREBP-1a and -2 both activate transcription of the genes encoding stearoyl-CoA desaturase-1 and -2, thereby markedly enhancing the production of monounsaturated fatty acids. nSREBP-1c was inactive in stimulating any transcription at the concentrations achieved in these studies. The current data support the emerging view that the nSREBPs act in complementary ways to modulate the lipid composition of cell membranes.

Cholesterol and fatty acids, the building blocks of cell membranes, are synthesized by regulated pathways in animal cells. Both pathways are influenced by a single family of transcription factors designated sterol regulatory element binding proteins (SREBPs), 1 whose concerted actions optimize the lipid content of membranes (reviewed in Brown and Goldstein (1)). Appropriate to their role in modulating membrane composition, the SREBPs are bound intrinsically to cell membranes.
They are synthesized as long precursors of ϳ1150 amino acids with three domains. The NH 2 -terminal domain of ϳ480 amino acids is a classic basic helix-loop-helix-leucine zipper transcription factor. This is followed by a membrane attachment domain of ϳ80 amino acids and a COOH-terminal domain of ϳ590 amino acids that performs a regulatory function. The SREBPs are attached to membranes of the nuclear envelope and endoplasmic reticulum in a hairpin fashion with their NH 2 -terminal and COOH-terminal domains projecting into the cytosol and the membrane attachment domain projecting into the lumen (2).
In order to reach the nucleus, the NH 2 -terminal domains of the SREBPs must be released from the membranes by proteolysis. This is accomplished by sequential cleavage at two peptide bonds, designated Site-1 and Site-2 (1). The proteolytic release of the SREBP NH 2 -terminal domains is subject to negative feedback regulation by cholesterol (3). When the cholesterol content of the cells declines, the Site-1 protease becomes activated, thereby initiating proteolytic release. When cholesterol builds up in the endoplasmic reticulum, the activity of the Site-1 protease is reduced, and the SREBPs remain bound to the membranes (1,4).
Three SREBPs are known to exist in animal cells. Two of these, designated SREBP-1a and -1c, are synthesized from a single gene through the use of alternate promoters and first exons (5)(6)(7). These proteins differ in the length of an acidic sequence at the extreme NH 2 terminus that activates transcription. SREBP-1a has a long acidic sequence of 42 amino acids, 12 of which are negatively charged. In SREBP-1c this sequence is reduced to 24 amino acids, of which 6 are acidic (1). SREBP-1c was cloned independently in Dallas (5) and Boston (8), where it was designated as ADD1. SREBP-2 is synthesized from a different gene (9,10). Overall, it is 47% identical to SREBP-1a, and it also has a long transcription activating domain. The SREBPs can function as homodimers (9), although heterodimerization has not been excluded.
Tissue culture cells and animal organs generally produce all three SREBPs, but they do so in differing ratios. In most tissue culture cells, the SREBP-1a transcript far exceeds that for SREBP-1c (7). However, in most organs of adult animals, including liver and adipose tissue, the SREBP-1c transcript predominates (7). The SREBP-2 transcript is present in all cells at levels that are approximately equal to the total of SREBP-1a and -1c (1).
The reason for the existence of three closely related SREBPs and the reason for their differing ratios in different cell types remain obscure. As far as is currently known, all of the SREBPs activate transcription of the same family of target genes. The identified targets can be divided into two pathways, genes relevant to cholesterol metabolism and those relevant to fatty acid metabolism (1).
In the cholesterol biosynthetic pathway, the SREBPs directly activate transcription of the genes encoding 3-hydroxy-3-methylglutaryl (HMG) coenzyme A synthase, HMG-CoA reductase, farnesyl diphosphate synthase, and squalene synthase (1,5,9,11,12). In the fatty acid and triglyceride biosynthetic pathways, the direct targets of SREBPs include acetyl-CoA carboxylase, fatty acid synthase, and glycerol-3-phosphate acyltransferase (13)(14)(15)(16)(17). The SREBPs also directly activate transcription of the gene encoding the low density lipoprotein (LDL) receptor, which provides cholesterol and fatty acids from external sources (5,9). Various cells and tissues have been forced to overproduce nuclear forms of SREBPs as a result of transfection or transgenic expression. Under these conditions many other genes are activated, but it is not known whether this activation is direct or indirect, i.e. as a result of activating other genes. This class includes the genes that encode the isozymes of stearoyl-CoA desaturase (SCD), the enzyme that converts saturated to monounsaturated fatty acids. The two closely related isozymes are designated SCD-1 and SCD-2 (18). The total number of SCD transcripts was markedly elevated in livers of transgenic mice that overexpressed dominant-positive forms of SREBP-1a (15), SREBP-1c (19), and SREBP-2 (20). The probe that was used did not distinguish between the two SCD isoforms.
Inasmuch as tissue culture cells produce active forms of all three SREBPs, it is difficult to study the individual role of each factor in isolation from the others. In the current studies, we have sought to overcome this problem through the use of a mutant line of Chinese hamster ovary (CHO) cells, designated M19 (21). These cells have a deletion in the gene encoding the protease that cuts SREBPs at Site-2, and they are therefore unable to release any SREBPs from cell membranes (22,23). As a result, these cells have an absolute growth requirement for exogenous cholesterol and unsaturated fatty acids. We have prepared stable derivatives of the M19 cells that express truncated dominant-positive forms of each SREBP under control of the inducible ecdysone promoter (24). The encoded SREBPs have a translation stop signal between the basic helix-loophelix-leucine zipper domain and the membrane anchor. As a result, they are never attached to membranes, and they enter the nucleus directly without a requirement for regulated proteolysis. We have incorporated a common epitope tag into each SREBP so that its level of expression can be quantified by an antibody that reacts with all three proteins in the same fashion. The results indicate that SREBP-1a and -2 activate the same family of genes, but they do so in different proportions. In general, SREBP-2 favors the cholesterol biosynthetic pathway, while SREBP-1a has a relatively greater effect on fatty acid biosynthesis. Under the conditions of these studies, SREBP-1c was essentially inactive in both pathways. In addition to elevating the total amount of fatty acid synthesis, SREBP-1a increased the ratio of unsaturated to saturated fatty acids. Thus, SREBP-1a and SREBP-2, acting in concert, modulate the content of cholesterol and unsaturated fatty acids in cell membranes.
Construction of Inducible Expression Vectors for FLAG-tagged Truncated SREBPs-We used standard methods of PCR and molecular cloning (25) to construct expression vectors for the cleaved nuclear forms of human SREBP-1a, -1c, and -2 in which expression of these nuclear SREBPs (nSREBPs) is controlled by the ecdysone-inducible promoter (24) contained in the pIND vector (Invitrogen). pIND-BP1a-FLAG encodes a fusion protein consisting of amino acids 1-487 of human SREBP-1a (5) followed by two copies of a FLAG epitope tag (DYKDDDDK) (27) and a stop codon, driven by the ecdysone-inducible promoter. Plasmid pIND-BP1c-FLAG encodes amino acids 1-463 of human SREBP-1c (5) followed by two copies of the FLAG epitope and a stop codon. Plasmid pIND-BP2-FLAG encodes amino acids 1-481 of human SREBP-2 (9) followed by two copies of the FLAG epitope and a stop codon. The structures of all plasmid constructs were confirmed by sequencing all ligation joints. In the case of pIND-BP1c-FLAG, the nSREBP-1c coding region was sequenced in its entirety.
Stable Transfection of M19 Cells-M19 cells are a mutant line of CHO-K1 cells that are auxotrophic for cholesterol and unsaturated fatty acids (21) because they fail to cleave SREBPs at Site-2, owing to a deletion in the S2P gene (22). Stock cultures of M19 cells were main-tained in monolayer culture at 37°C in 9% CO 2 in medium A (a 1:1 mixture of Ham's F-12 medium and Dulbecco's modified Eagle's medium containing 100 units/ml penicillin, 100 g/ml streptomycin sulfate, 5% (v/v) fetal calf serum, 5 g/ml cholesterol, 1 mM sodium mevalonate, and 20 M sodium oleate).
To obtain stable cell lines of M19 cells expressing either pIND-BP1a-FLAG or pIND-BP1c-FLAG, M19 cells were plated on day 0 at a density of 1 ϫ 10 6 cells/100-mm dish in medium A and cotransfected on day 1 with 6 g of pVgRXR (containing the gene conferring Zeocin resistance) and 6 g of either pIND-BP1a-FLAG or pIND-BP1c-FLAG (each containing a neo gene conferring G418 resistance) using the MBS transfection kit according to the manufacturer's instructions, except that the concentration of modified bovine serum was 5% (v/v) rather than 6%. The transfection was carried out at 35°C for 3 h in 3% CO 2 , after which each dish was washed once with 10 ml of phosphate-buffered saline and switched to 10 ml of medium A. On day 2, transfectants were switched to medium A supplemented with 750 g/ml G418 and 500 g/ml Zeocin. After 14 days, drug-resistant colonies were incubated with 1 M muristerone A for 24 h and then initially screened for SREBP expression either by visualization of LDL receptor expression using reconstituted fluorescent PMCA oleate-labeled LDL (28) and/or by immunoblot analysis of nuclear extracts using the anti-FLAG M2 monoclonal antibody. Positive colonies were cloned by dilution plating and rescreened by immunoblot analysis with the anti-FLAG M2 antibody.
To obtain stably transfected M19 cells expressing pIND-BP2-FLAG, the cells were first transfected with pVgRXR and selected with Zeocin. The Zeocin-resistant colonies were screened by transient transfection with pIND/lacZ, and the colony that showed the greatest muristerone A induction was identified by assay of ␤-galactosidase activity. Cells from this colony were transfected with pIND-BP2-FLAG, selected with G418 and Zeocin, and screened for nSREBP-2 expression as described above.
The above stably transfected cell lines, designated N-BP-1a, N-BP-1c, and N-BP-2, were maintained as stock cultures in medium A containing 750 g/ml G418 and 500 g/ml Zeocin.
Immunoblot Analysis for SREBPs-Cell monolayers were washed FIG. 1. Schematic representation of inducible system for graded expression of nSREBPs. This system is based on the ecdysone system described by No et al. (24) and modified by Invitrogen. The ecdysone receptor (VgEcR) and its partner RXR are both encoded by the plasmid pVgRXR. The two proteins heterodimerize and transactivate E/GRE-containing promoters in the presence of muristerone A. An E/GRE is placed upstream of a minimal promoter that can drive expression of the nSREBP-1a, -1c, or -2. These nSREBPs are epitope-tagged with the FLAG epitope and are encoded by plasmids pIND-BP-FLAG. The induced nSREBP-1a, -1c, or -2 in turn drives expression of endogenous target genes whose promoters contain SREs.

5Ј-ATGCCGGCCCACATGCTCCAAG-3Ј
once with phosphate-buffered saline and then harvested in buffer containing 50 mM Hepes-KOH at pH 7.5, 1.5 mM MgCl 2 , 10 mM KCl, 1 mM dithiothreitol, 1 mM sodium EDTA, and 1 mM sodium EGTA supplemented with a mixture of protease inhibitors (3). The cell suspension was passed through a 22-gauge needle 15 times and centrifuged at 1000 ϫ g at 4°C for 5 min. Each pellet from two pooled 60-mm dishes or one 100-mm dish was resuspended in 0.12-0.15 ml of buffer containing 50 mM Hepes-KOH at pH 7.5, 0.42 M NaCl, 2.5% (v/v) glycerol, 1.5 mM MgCl 2 , 1 mM sodium EDTA, 1 mM EGTA, 1 mM dithiothreitol, and the protease inhibitor mixture, followed by rotation at 4°C for 30 min and centrifugation at 10 5 ϫ g in a Beckman TLA 100.2 rotor for 30 min at 4°C. The resulting supernatant was used as the nuclear extract. Protein concentration was measured with a BCA kit.
Samples of the nuclear extract fractions were mixed with 4ϫ SDS loading buffer (29), subjected to SDS-PAGE on 8% gels, and transferred to a Hybond-C membrane. The membrane was blocked by incubation in phosphate-buffered saline containing 0.05% (v/v) Tween 20, 5% (v/v) nonfat dry milk, and 5% (v/v) heat-inactivated newborn calf serum. Each membrane was then incubated with 5 g/ml anti-FLAG M2 antibody for 2 h at room temperature. After washing five times with phosphate-buffered saline and 0.05% Tween 20, the membranes were incubated for 1 h at room temperature with 125 I-labeled sheep anti-mouse immunoglobulin (3 ϫ 10 6 cpm/ml). The blots were washed at room temperature (four washes; 5 min/wash) with phosphate-buffered saline, 0.05% Tween 20. They were then exposed to film as indicated in the figure legends.
Assays of Lipid Synthesis and LDL Receptor Activity-Intact monolayers were incubated with 0.5 mM sodium [ 14 C]acetate (20 -29 dpm/ pmol) or 0.1 mM sodium [ 14 C]stearate-albumin (6 dpm/pmol) as described in the figure legends. Incorporation into lipids was determined as described previously (30) except that each sample received, before saponification, ϳ10 5 dpm [ 3 H]oleic acid as a recovery standard for calculating fatty acid synthesis. The data are expressed as nanomoles of [ 14 C]acetate incorporated into cholesterol or fatty acids per mg of cell protein per h. Receptor-mediated degradation of 125 I-LDL by intact cells was measured as described previously (31). Degradation activity represents the high affinity receptor-dependent rate of proteolysis and is expressed as micrograms of 125 I-labeled acid-soluble (non-iodide) material released into the culture medium per mg of total cell protein per 5 h. The protein content of cells was measured by the method of Lowry et al. (32).
HPLC Analysis of 14 C-Labeled Individual Fatty Acids-Cell monolayers were incubated with sodium [ 14 C]acetate or sodium [ 14 C]stearate-albumin as described above. Prior to saponification, a recovery standard consisting of ϳ10 5 cpm [ 14 C]linoleic acid was added to each sample. After extraction with petroleum ether, the fatty acid fraction was dried under nitrogen gas, dissolved in 0.1 ml of methanol containing 0.1% (v/v) acetic acid, and analyzed by reverse-phase HPLC using a Waters computer-controlled apparatus with a Nova-Pak C 18 reversephase column (3.9 ϫ 150 mm) coupled to a Radiometer Flo-1 radioactive flow detector. Samples (80 l representing ϳ0.3 mg of protein from 0.8 dish of cells) were injected automatically by a WISP injector, and the 14 C-labeled fatty acids were resolved by isocratic chromatography with a mixture of 84.9% methanol, 15% water, and 0.1% acetic acid at a flow rate of 1 ml/min. Each species of fatty acid was identified by comparison of its retention time with that of a known standard. Recoveries of total fatty acids in different experiments ranged from 30 to 47%, as judged by the recovery of the added [ 14 C]linoleic acid.
cDNA Probes-cDNA probes for Chinese hamster SREBP-1, SREBP-2, and HMG-CoA reductase were prepared using the plasmids and restriction enzymes listed in Table I. cDNA probes for nine other Chinese hamster mRNAs were prepared by PCR as follows. Firststrand cDNA was prepared from CHO-K1 total RNA using a firststrand cDNA synthesis kit primed with random hexadeoxynucleotides. The cDNA was used as template in PCR reactions with the primer pairs listed in Table II. The resulting PCR products were subcloned into either pCR-Blunt (LDL receptor probe) or pNoTA/T7 (all other probes). The fragments were released by EcoRI (LDL receptor, 1-kilobase pair fragment), XbaI (lanosterol demethylase and lanosterol synthase), or BamHI (all other probes) and purified on agarose gels. All cDNA probes were radiolabeled with [␣-32 P]dCTP using Prime-It II random primer labeling kit.
Blot Hybridization of mRNA-Total RNA was prepared from monolayers of stably transfected cell lines using the RNeasy Midi kit. For Northern gel analysis, total RNA was mixed with RNA sample loading buffer (containing 50 g/ml ethidium bromide) (Sigma), denatured with formaldehyde and formamide, subjected to electrophoresis in a 1% agarose formaldehyde gel, and transferred to Hybond Nϩ membranes.

FIG. 2. Amounts of nuclear SREBP proteins in N-BP cell lines after induction with varying concentrations of muristerone A.
On day 0, N-BP and M19 cells were set up for experiments in medium A at 3 ϫ 10 5 cells/60-mm dish, and CHO-K1 cells were set up at 2 ϫ 10 5 cells/60-mm dish in medium B (a 1:1 mixture of Ham's F-12 medium and Dulbecco's modified Eagle's medium containing 100 units/ml penicillin and 100 g/ml streptomycin sulfate) supplemented with 5% fetal calf serum. On day 1, the monolayers were switched to medium B supplemented with 5% fetal calf lipoprotein-deficient serum and the indicated amount of muristerone A. After incubation for 24 h at 37°C, the cells were divided into three groups. One group (this figure) was used to prepare a nuclear extract, and the other two groups were used for assays of 14 C-lipid synthesis and LDL receptor activity (Figs. 3 and 5, respectively). Aliquots of the nuclear extract protein (25 g/lane) were subjected to SDS-PAGE and immunoblot analysis with anti-FLAG M2 antibody followed by incubation with 125 I-labeled sheep anti-mouse immunoglobulin as described under "Experimental Procedures." The filter was exposed to BioMax MS-1 film (Sigma) at room temperature for 4 days. The same filter was also exposed to a Fuji PhosphorImager, and the amount of radioactivity in the nSREBPs was quantified using a Bio-Imaging analyzer with BAS 100 MacBAS software (Fuji Medical Systems) (see Figs. 3-5). Visualization of the membranes under UV light showed equivalent transfer of 28 and 18 S RNA in all samples. The membranes were hybridized with the indicated 32 P-labeled probes (2 ϫ 10 6 cpm/ml) for 2 h at 65°C using Rapid-hyb buffer; washed twice with either 0.1% (w/v) SDS/0.2ϫ SSC at 65°C for 20 min (HMG-CoA reductase) or 0.1% SDS/0.1ϫ SSC at 65°C for 30 min (all other mRNAs); and exposed at Ϫ80°C to Reflection NEF 495 film (NEN Life Science Products) with intensifying screens for the indicated time. The amount of radioactivity was quantified by exposure of the filter to a BAS1000 Fuji PhosphorImager.
RNase Protection Assay-Aliquots of total RNA (10 g) were subjected to RNase protection assay using a Hyb-Speed RPA kit as described previously (7). Each hybridization was carried out in a final volume of 10 l containing 8 ϫ 10 4 cpm of the indicated 32 P-labeled cRNA probe (Table III) and 2 ϫ 10 4 cpm of a 32 P-labeled hamster ␤-actin cRNA probe (Table III). The 32 P-labeled mRNA-protected fragments were separated on 7 M urea, 6% polyacrylamide gels, and the radioactivity in each band was quantified using the Fuji PhosphorImager.
The cRNA probes in Table III were generated by PCR of first-strand cDNA as described above, followed by subcloning the PCR fragments in the pNoTA/T7 vector. After linearization with the indicated restriction enzyme in Table III, antisense RNA was transcribed using the Riboprobe in vitro transcription systems with T7 RNA polymerase and [␣-32 P]CTP. For generation of the ␤-actin cRNA probe, the cleaved template was rendered blunt-ended by treatment with the Klenow fragment of DNA polymerase I before transcription. Fig. 1 shows the strategy that we used to produce graded expression of the nuclear forms of the SREBPs (designated nSREBPs) in stable lines of transfected M19 cells (designated N-BP cells). The strategy is based on the methods developed by No et al. (24), and the vectors produced by Invitrogen. The M19 cells were transfected with two plasmids. One plasmid (pVgRXR) encodes a modified version of the ecdysone receptor (EcR) and its heterodimerization partner, the retinoid-X receptor (RXR), both of which are driven by constitutive promoters. The second plasmid encodes the nuclear form of one of the SREBPs, which is driven by an enhancer containing the response element E/GRE. When muristerone A, a relative of the insect hormone ecdysone, is added to these cells, it binds to the EcR/RXR heterodimer, which then binds to the E/GRE, thereby activating transcription of the gene encoding the nSREBP. Each of the nSREBPs is engineered to contain two tandem copies of the FLAG epitope, which is recognized by an anti-FLAG antibody. The nSREBPs all terminate at a common point between the basic helix-loop-helix-leucine zipper domain and the membrane-attachment domain, thereby avoiding membrane binding.

RESULTS
The muristerone-inducible ecdysone receptor system has two virtues that are crucial for the current experiments: 1) transcription of the nSREBP cDNA is extremely low in the absence of muristerone, and 2) graded amounts of the nSREBPs can be produced by varying the concentration of muristerone. This allows production of precise amounts of single nSREBPs in the M19 cells, which lack their own nSREBPs as a result of the absence of the Site-2 protease. Fig. 2 shows an experiment in which we added varying concentrations of muristerone to the culture medium of the stably transfected cells and measured the amounts of each nSREBP by SDS-PAGE and immunoblotting with the anti-FLAG antibody. By adjusting the concentration of muristerone, we obtained comparable levels of expression of each of the three SREBPs. For an unknown reason, transcription of the SREBP-1a gene was more sensitive than that of the other two genes, and hence in future experiments we used lower concentrations of muristerone when inducing this gene. In each experiment we harvested dishes of cells and measured the amounts of nSREBPs by immunoblotting and PhosphorImager quantification. The results are plotted as a function of the amount of nSREBP expressed.
To assess the effect of each SREBP on cholesterol and fatty acid biosynthesis, we induced the cells with varying amounts of muristerone, pulse-labeled them with [ 14 C]acetate, and measured the amount of [ 14 C]cholesterol and 14 C-labeled fatty acids that were produced (Fig. 3). In the absence of muristerone, all three N-BP cell lines synthesized almost no cholesterol as indicated by the value for "zero expressed nSREBP protein" (Fig. 3A). The synthesis of [ 14 C]cholesterol increased progressively as the expression of nSREBP-2 increased. A significant increase in cholesterol synthesis was also observed for nSREBP-1a, but this was much less than the increase observed with nSREBP-2 even when the amount of SREBP-1a was 4-fold greater than the maximal amount of nSREBP-2. nSREBP-1c did not stimulate the synthesis of [ 14 C]cholesterol.
In the absence of muristerone, each of the three N-BP cell lines synthesized appreciable amounts of 14 C-fatty acids (Fig.  3B). The rate was approximately 60% of the rate that was observed in wild-type CHO-7 cells incubated under the same conditions of lipoprotein deprivation (indicated by the ϫ on the vertical axis of Fig. 3B). Increasing amounts of nSREBP-2 increased the synthesis of 14 C-fatty acids. nSREBP-1a also had a stimulating effect, but it was only one-third as potent as nSREBP-2. nSREBP-1c did not stimulate the production of 14 C-fatty acids, even at levels of expression that were several-  (Fig. 2) were switched on day 2 to Dulbecco's modified Eagle's medium (without glutamine) containing the indicated amount of muristerone A, 2 mg/ml bovine serum albumin, and 10 g/ml 125 I-LDL (135 cpm/ng of protein) in the absence or presence of 500 g/ml unlabeled LDL. After incubation for 5 h at 37°C, the amount of high affinity degradation of 125 I-LDL was measured as described previously (26). The results are plotted against the amount of expressed nSREBP protein (Fig. 2). Each value represents the average of triplicate incubations. The value for "zero expressed nSREBP protein" on the vertical axis is the mean value of the three cell lines incubated in the absence of muristerone A. The "zero" values were all below 0.047 g/5 h/mg of protein. The ϫ on the vertical axis denotes the results obtained in CHO-K1 cells studied in the same experiment (0.98 g/5 h/mg of protein). fold higher than the effective level of nSREBP-2 and -1a.
We calculated the ratio of the synthesis of [ 14 C]cholesterol/ 14 C-fatty acids at varying levels of the three nSREBPs, and the results are plotted in Fig. 4. In the absence of nSREBP, this ratio was extremely low, reflecting the selective loss of cholesterol synthesis in the M19 cells. nSREBP-2 restored this ratio to the same level that was observed in wild-type CHO cells (indicated by the ϫ on the vertical axis in Fig. 4). Once this threshold was reached, the ratio remained constant even at higher levels of nSREBP-2, reflecting parallel increases in the synthesis of cholesterol and fatty acids. In contrast, nSREBP-1a produced a ratio that remained below the value in wild-type CHO-7 cells, even at high expression levels. nSREBP-1c did not affect the cholesterol/fatty acid ratio, consistent with its failure to alter the synthesis of either cholesterol or fatty acids in these cells.
In the absence of nSREBP, the M19 cells failed to take up and degrade 125 I-LDL with high affinity, indicating a deficiency of LDL receptors (Fig. 5). nSREBP-2 and nSREBP-1a were equally effective in increasing the rate of 125 I-LDL degradation.
To measure the effects of the nSREBPs on the level of mRNA transcripts, we performed a series of Northern blots on RNA preparations from N-BP cell lines that were incubated with varying amounts of muristerone (Fig. 6B). In the same experiment we measured the levels of nSREBP proteins in nuclear extracts by immunoblotting (Fig. 6A). All of the blots were quantified on a PhosphorImager, and the mRNA levels were plotted as a function of the amount of expressed nSREBP (Fig.  7). nSREBP-1a and nSREBP-2 were equipotent in raising the mRNAs for HMG-CoA synthase, farnesyl diphosphate synthase, lanosterol synthase, lanosterol 14␣-demethylase, and the LDL receptor. nSREBP-1a was somewhat less active than was nSREBP-2 in raising the mRNA for HMG-CoA reductase, and it was much less active with regard to squalene synthase. nSREBP-1a and nSREBP-2 were equipotent in stimulating transcription of the gene for fatty acid synthase. nSREBP-1a appeared somewhat more effective than nSREBP-2 in raising the total mRNA for SCD, but the probe failed to distinguish between SCD-1 and SCD-2 on the Northern blots.

FIG. 6. Amounts of nuclear SREBP proteins (A) and various mRNAs (B) in muristerone A-treated inducible N-BP cell lines. N-BP
and M19 cells were set up for experiments on day 0 in medium A at 1 ϫ 10 6 cells/100-mm dish. On day 2, the monolayers were switched to medium B supplemented with 5% fetal calf lipoprotein-deficient serum and the indicated amount of muristerone A. After incubation for 24 h at 37°C, one dish of cells in each group was used to prepare a nuclear extract for immunoblot analysis (A), and six dishes were pooled for preparation of total RNA (B). Panel A, aliquots of nuclear extract protein (50 g/lane) were subjected to SDS-PAGE and immunoblot analysis with anti-FLAG M2 antibody followed by incubation with 125 I-labeled sheep anti-mouse immunoglobulin. The filter was exposed to BioMax MS-1 film (Sigma) at room temperature for 4 days. Panel B, total RNA (20 g/lane) was subjected to electrophoresis and blot hybridization with the indicated 32 P-labeled probe. The filters were exposed to reflection film for 7.5-25 h at Ϫ80°C.
In contrast to the above 3-20-fold stimulations in mRNA expression, the increase in mRNA levels for several genes was either negligible (caveolin and SREBP-1) or less than 2-fold (SREBP-2) (Figs. 6 and 7).
The difference between nSREBP-1a and -2 with respect to squalene synthase was striking. The Northern blotting experiment was repeated three times on different days using different cell extracts and different clones of cells expressing nSREBP-1a. In each case nSREBP-1a was less than 0.1 as effective as nSREBP-2 in raising the level of squalene synthase mRNA. To quantify these data more precisely, we established an RNase protection assay for squalene synthase, and we used HMG-CoA reductase as a control (Fig. 8). The results confirmed the failure of nSREBP-1a to enhance the squalene synthase mRNA.
To distinguish between the two isozymes of SCD, we employed an RNase protection assay using probes that were specific either for SCD-1 or SCD-2 (Fig. 9). In the absence of muristerone, the amount of mRNA encoding SCD-2 was much less than that encoding SCD-1 in all three N-BP cell lines. Both nSREBP-1a and -2 increased the levels of SCD-1 and SCD-2 mRNA. The relative effect on SCD-2 was greater, so that at high levels of nSREBP expression the absolute amount of SCD-2 mRNA was similar to that for SCD-1.
The increase in the SCD mRNAs would be expected to increase the production of unsaturated fatty acids. To test this hypothesis, we induced the expression of nSREBP-1a with muristerone, and we then incubated the cells with [ 14 C]acetate (Table IV and Fig. 10). After 3 h we saponified the cell lipids, extracted the fatty acids, separated them by HPLC, and detected the products with a continuous radioactivity monitor.
For quantitative purposes, we added an internal standard of [ 14 C]linoleic acid (18:2), which is not produced in hamster cells. When the N-BP-1a cells were incubated in the absence of muristerone, they produced only saturated fatty acids (mostly 16:0 and a small amount of 14:0) (Fig. 10A). After muristerone induction, the synthesis of the saturated fatty acids increased (18:0 as well as 16:0 and 14:0) (Fig. 10B). Even more dramatically, we observed the appearance of relatively large amounts of monounsaturated fatty acids (18:1 and 16:1) that were not detected in the untreated cells.
The quantitative results of the experiment of Fig. 10 are shown in Table IV (Experiment A). For this purpose, the peak area for each of the major fatty acids was measured and expressed as a percent of the total. The 14 C-fatty acids in the M19 cells consisted almost entirely of palmitate (16:0), and there was no effect of muristerone. In the N-BP-1a and N-BP-2 cells, the addition of muristerone produced a dramatic shift toward the monounsaturated varieties. Muristerone had no effect on fatty acid synthesis in the N-BP-1c cells. nSREBP-1a, -1c, or 2. The amount of radioactivity in each band in Fig.  6, A and B, was quantified as described in Fig. 2. A relative intensity value of 1 for each mRNA was calculated from the mean of the values obtained in lanes 1, 2, 5, and 8 in Fig. 6B.  FIG. 8. Quantification of HMG-CoA reductase and squalene synthase mRNA levels in N-BP cell lines by RNase protection assay. The same preparation of total RNA used in Fig. 6 was assayed by RNase protection (10 g/lane) with the indicated 32 P-probe as described under "Experimental Procedures." Panel A, the gel was dried and exposed to reflection film for 19 h at Ϫ80°C. Panel B, the gel was exposed to reflection film for 22.5 h at room temperature. Panel C, the dried gels in panels A and B were exposed to a Fuji PhosphorImager, and the amount of radioactivity in each band was quantified as described in Fig. 2. The intensity of the HMG-CoA reductase and squalene synthase mRNA signals in each lane was normalized to the ␤-actin signal and plotted as a function of the relative amount of expressed SREBP protein as described in Fig. 6. A relative intensity value of 1 for each mRNA was calculated from the mean of the values obtained in lanes 1, 2, 5, and 8.

FIG. 7. Quantification of mRNA levels in N-BP cell lines in relation to the amount of muristerone A-induced expression of
If SREBPs affect the synthesis of unsaturated acids, one would predict that this synthesis should be regulated in wildtype CHO-K1 cells when nuclear SREBPs are induced by sterol deprivation or repressed by sterol addition. Indeed, as shown in Experiment B of Table IV, the addition of sterols reduced the percentage of palmitoleate (16:1) and oleate (18:1) in CHO-K1 cells. This experiment was performed on two other occasions with similar results.
To demonstrate the effect of nSREBP-1a on fatty acid desaturation directly, we incubated N-BP-1a cells with [ 14 C]stearate (18:0) for varying times in the absence and presence of muristerone. The cells were harvested, and the relative amounts of [ 14 C]stearate and [ 14 C]oleate (18:1) were measured by HPLC (Fig. 11). In the absence of muristerone, the N-BP-1a cells converted relatively little of the [ 14 C]stearate to oleate (12% at 5 h). This conversion was elevated by 3.6-fold in the presence of muristerone (43% at 5 h). This experiment was performed on one other occasion with similar results.

DISCUSSION
In the current studies we used an inducible promoter system to produce graded expression of the nSREBPs in permanently transfected M19 cells, which cannot produce their own nSREBPs, owing to an absence of the Site-2 protease (22). Each of the three nSREBPs was tagged with two tandem copies of the FLAG epitope, thereby allowing the levels of protein to be compared directly by immunoblotting with a common antibody. We attempted to keep the level of expression within the physiologic range as reflected by restoration of lipid synthesis to rates that were similar to those in wild-type cells that were induced by sterol deprivation. The results indicate that nSREBP-1a and nSREBP-2 can both stimulate transcription of genes involved in cholesterol and fatty acid biosynthesis, but they do so with different relative efficiencies. In contrast, nSREBP-1c had little, if any, stimulatory activity in these hamster cells at concentrations that were equivalent to those of nSREBP-1a and -2.
The differences between nSREBP-1a and -2 were apparent in the overall ratio of synthesis of 14 C-fatty acids and [ 14 C]cholesterol from [ 14 C]acetate (Figs. 3 and 4). Although both transcription factors stimulated both pathways, nSREBP-2 produced a higher ratio of [ 14 C]cholesterol to 14 C-fatty acids. One reason for this difference became apparent when we measured the effects of the two nSREBPs on the levels of individual mRNAs (Figs. 6 and 7). nSREBP-1a and 2 both induced the mRNAs encoding fatty acid synthase and early enzymes in the cholesterol biosynthetic pathway, including HMG-CoA synthase, HMG-CoA reductase, and farnesyl diphosphate synthase. They also had similar inducing effects on two of the late enzymes in the cholesterol biosynthetic pathway, i.e. lanosterol synthase and lanosterol 14␣-demethylase (CYP51). However, nSREBP-2 was much more potent than nSREBP-1a in inducing transcription of the key branch point enzyme squalene synthase. These findings suggest that cells may produce relatively high levels of SREBP-1a when they have a demand for nonsterol products derived from farnesyl pyrophosphate, such as prenylated proteins or dolichol (33). Indeed, SREBP-1a is expressed at relatively high levels in rapidly growing tissue culture cells which have a relatively large demand for these nonsterol products (7).
Differential effects of nSREBP-1a and nSREBP-2 on the squalene synthase promoter have also been observed by Guan et al. (12,34), who expressed high levels of the nuclear forms of these transcription factors by transient transfection under control of the strong CMV promoter. At these high levels of expression, both nSREBPs activated a reporter gene driven by the squalene synthase promoter, but they did so by interacting at different sites. nSREBP-2 bound to more sites than did nSREBP-1a, and it continued to stimulate transcription even when the SRE-like sites had been destroyed by in vitro mutagenesis (12). The multiple sites of action for SREBP-2 provide a potential explanation for the current finding that nSREBP-2 is more active than nSREBP-1a when both are expressed at low levels. The difference between the two SREBPs is likely to decrease when both proteins are expressed at supraphysiologic levels. We have previously observed this phenomenon with respect to SREBP-1c (19). When expressed at high levels FIG. 9. Quantification of stearoyl-CoA desaturase-1 and -2 mRNA levels in N-BP cell lines by RNase protection assay. M19 and N-BP cells were set up for experiments as described in Fig. 6 and incubated with the indicated amount of muristerone A. After incubation for 15 h at 37°C, one dish of cells in each group was used to prepare a nuclear extract for immunoblot analysis (A), and four dishes were pooled for preparation of total RNA (B-D). Panel A, aliquots of nuclear extract protein (30 g/lane) were subjected to SDS-PAGE and immunoblot analysis as in Fig. 6A. The filter was exposed to BioMax MS-1 film (Sigma) at room temperature for 11 days. Panels B and C, RNase protection assays for SCD-1, SCD-2, and ␤-actin mRNAs were carried out as described in Fig. 8. The gels were exposed to reflection film for 19 h at Ϫ80°C. Panel D, quantification of SCD-1 and SCD-2 mRNA levels was carried out as described in Fig. 8C. through transient transfection driven by the CMV promoter, nSREBP-1c was nearly as active as SREBP-1a in stimulating transcription of genes involved in cholesterol and fatty acid biosynthesis. When the promoters were switched to the relatively weak thymidine kinase promoter, the activity of nSREBP-1c was much lower than that of nSREBP-1a (19). In the current studies we deliberately prepared permanent cell lines so that the level of expression would be uniform throughout the dish. We maintained relatively low levels of expression of all three nSREBPs, and we used a cell line that lacks its own nuclear SREBPs so that there is no possibility of heterodimerization. Under these conditions the differences in the activities of the nSREBPs were clearly apparent.
The finding that SREBP-2 favors cholesterol synthesis is consistent with several previous observations from this laboratory. In one set of experiments, livers of hamsters (35,36) and mice 2 were deprived of cholesterol by the feeding of a bile acid binding resin and an HMG-CoA reductase inhibitor. The amount of nSREBP-2 increased, but the amount of nSREBP-1 (now known to be nSREBP-1c) decreased. Similarly, when cultured CHO cells were mutagenized and selected for resistance to killing by 25-hydroxycholesterol, which kills cells by reducing the activity of the Site-1 protease, several of the mutant cell lines survived by producing truncated forms of SREBP-2 (37,38). None produced truncated SREBP-1. Similar results have recently been obtained by Rosenfeld and Osborne (39) in experiments in which CHO cells were individually transfected with cDNAs encoding the nuclear forms of SREBP-1a, -1c, and -2 and were then subjected to selection with 25-hydroxycholesterol. Only those cells transfected with nSREBP-2 survived the selection.
The relative inactivity of nSREBP-1c in the current studies is profound. As discussed above, nSREBP-1c does have the capacity to enhance transcription when it is expressed at high levels in liver (19) and in a variety of cultured cells (7,19,40). In view of its short acidic transcriptional activation domain, nSREBP-1c may have to interact with another transcription factor in order to enhance transcription. Such a factor may be relatively tissue-specific, allowing nSREBP-1c to function in selected tissues under selected developmental or metabolic conditions. The other factor might be another basic helix-loop-helix-leucine zipper protein that heterodimerizes with nSREBP-1c (and presumably nSREBP-1a). Alternatively, it might be another factor that interacts with nSREBP-1c in a different fashion. Indeed, the actions of all nSREBPs are known to be dependent on interaction with other transcription factors such as Sp1 in the case of the LDL receptor promoter (41) and NF-Y in the case of the farnesyl diphosphate synthase (42) and HMG-CoA synthase (43) promoters. In addition, nSREBP-1c might have a modulating role by forming heterodimers with nSREBP-1a and -2. The current studies were specifically designed to avoid such heterodimerization. In the future it will be necessary to study the actions of each of these nSREBPs under conditions where heterodimerization with other nSREBPs is possible.
A striking finding in the current studies was the failure of the M19 cells to synthesize unsaturated fatty acids and the restoration of this synthesis when either nSREBP-1a or nSREBP-2 was expressed ( Fig. 10 and Table IV). This restoration is attributable to the 3.5-fold induction of the mRNAs for SCD-1 and the even larger 13-fold induction of SCD-2. The  a M19 and N-BP cells were set up as described in Fig. 2. On day 1, the medium was switched as described in Fig. 2, and the monolayers were incubated in the absence or presence of 10 M muristerone A except for the NB-1a cells, which were incubated with 0.8 M muristerone A (see Fig.  9A). After incubation for 24 h at 37°C, the cells were pulse-labeled for 3 h with 0.5 mM sodium [ 14 C]acetate (29 dpm/pmol) and subjected to fatty acid distribution analysis as described under "Experimental Procedures." b CHO-K1 cells were set up as described in Fig. 2. On day 1, the monolayers were switched to medium B supplemented with 5% fetal calf lipoprotein-deficient serum in the absence or presence of 10 g/ml cholesterol plus 1 g/ml 25-hydroxycholesterol. After incubation for 18 h at 37°C, the cells were pulse-labeled for 5.5 h with 0.5 mM sodium [ 14 C]acetate (20 dpm/pmol) and processed as described in experiment A. Each value is the average of duplicate incubations. genes for SCD-1 and SCD-2 have been shown previously to undergo regulated transcription in response to hormones and metabolites (reviewed in Ntambi (18)). Transcription of both SCD genes is increased upon terminal differentiation of cultured mouse preadipocytes, a complex process that is affected by insulin, glucocorticoids, and cyclic AMP, and proceeds through the action of transcription factors such as the CCAAT/ enhancer binding protein-␣ (C/EBP-␣) and peroxisomal proliferator-activating receptor-␥ (PPAR-␥). In liver SCD-1 is expressed, but SCD-2 is not (18). In this organ expression of SCD-1 is reduced by fasting and enhanced by refeeding a fatfree high carbohydrate diet. The increase can be blocked by the inclusion of polyunsaturated fatty acids. A similar repression by polyunsaturated fatty acids occurs in cultured liver cells. The 5Ј-flanking regions of the SCD-1 and SCD-2 genes share a region of homology that appears to be the site of repression by polyunsaturated fatty acids (44), but the responsible transcription factors and the exact sites of DNA binding have not yet been identified. This region does not contain a sequence that matches any of the sequences that are known to be sites for SREBP binding.
The finding that nSREBPs are necessary for unsaturated fatty acid biosynthesis in CHO cells appears to explain the original findings of Chang and colleagues, who showed that cholesterol auxotrophic CHO cells in general (45) and M19 cells in particular (21) have a growth requirement for unsaturated fatty acids as well as cholesterol. Based on the available information regarding the SCD promoters, we cannot be certain that SREBPs stimulate transcription of the SCD genes directly. They may do so indirectly by stimulating transcription of another gene. It is also possible that levels of SCD mRNAs rose in these studies not as a result of enhanced transcription, but as a result of mRNA stabilization. Indeed, Sessler et al. (46) reported that polyunsaturated fatty acids lower SCD mRNA levels in cultured adipocytes in part by accelerating the degradation of the mRNAs, both of which have unusually long 3Јuntranslated regions. The magnitude of the nSREBP-mediated induction in the current study (13-fold in the case of SCD-2) argues that the nSREBPs affect SCD gene transcription, but this remains to be proven.
The overall result of nSREBP activity is to increase the rates of cholesterol and unsaturated fatty acid synthesis in CHO cells. Cholesterol and unsaturated fatty acids can modulate functional properties of cell membranes, particularly those carried out by lipid microdomains such as caveolae (47,48). The SREBPs, therefore, appear to be central organizers of the composition and function of the membranes of animal cells. FIG. 11. Increased conversion of [ 14 C]stearic acid to [ 14 C]oleic acid in muristerone A-treated cells expressing nSREBP-1a. N-BP-1a cells were set up for experiments as described in Fig. 2. On day 1, the medium was switched as described in Fig. 2, and the monolayers were incubated in the absence or presence of 0.8 M muristerone A (see Fig. 9A). After incubation for 20 h at 37°C, the cells were pulse-labeled for the indicated time with 0.1 mM sodium [ 14 C]stearate-albumin (6 dpm/pmol). The addition of [ 14 C]stearate was made in a staggered fashion so that all the cells were harvested at the same time. Samples were subjected to fatty acid distribution analysis as described under "Experimental Procedures." Each value is the average of duplicate incubations, which showed an overall mean variation of 6.4% (range, 1-17%).