Molecular cloning and functional analysis of the promoter of the human squalene synthase gene.

We have cloned and characterized the 5′-flanking region of the gene encoding human squalene synthase. We report here the promoter activity of successively 5′-truncated sections of a 1 kilobase of this region by fusing it to the coding region of a luciferase reporter gene. DNA segments of 200 base pairs (bp) 5′ to the transcription start site, as determined by primer extension analysis, show a strong promoter effect on the expression of the luciferase chimeric gene and a high response to the presence of sterols when transiently transfected into the human hepatoma cell line HepG2 or to the hamster-derived CHO-K1 cells. An approximately 50-fold induction of luciferase activity, in the absence of sterols, was observed in transiently transfected HepG2 cells for fusion constructs containing sections of 200, 459, and 934 bp of the putative human squalene synthase promoter. Loss of promoter activity and response to sterols was localized to a 69-bp section located 131 nucleotides 5′ to the transcription start site. Sequence analysis of this region showed that it contained a sterol regulatory element 1 (SRE-1) previously identified in other sterol regulated genes (Smith, J. R., Osborne, T. F., Brown, M. S., Goldstein, J. L., and Gil, G.(1988). J. Biol. Chem. 263, 18480-18487) and two potential NF-1 binding sites. Additional CCAAT box, SRE-1 element, and two Sp1 sites were identified 3′ to this section. Sequences within this 69-bp DNA, including the SRE-1 cis-acting element, show strong binding to the purified nuclear transcription factor ADD1 (Tonzonoz, P., Kim, J. B., Graves, R. A., and Spiegelman B. M.(1993) Mol. Cell Biol. 13, 4753-4759) by mobility shift assay and footprinting analyses.

. The activity of human squalene synthase (HSS) and the level of its mRNA are regulated by sterols in the human hepatoma cell line HepG2.
Sterol-mediated regulation has been localized to a 10-base pair (bp) element in the 5Ј-flanking region of other sterolregulated genes. This 10-bp sterol regulatory element 1 (SRE-1) mediates increased transcription of the genes encoding HMG-CoA synthase and the low density lipoprotein (LDL) receptor in sterol-depleted cells, and its activity is inhibited by sterols (3,4). Proteins that bind to the SRE-1 of the LDL receptor (SREBPs) were purified by DNA affinity chromatography from nuclear extracts of HeLa cells. A cDNA for SREBP-1 was isolated from adipocyte cDNA library (5). This cDNA, designated ADD1, activated transcription of a reporter gene containing an "E-box" sequence present in the promoter of fatty acid synthase in transfected NIH 3T3 cells. Cloned SREBP cDNA contain two major classes of proteins, SREBP-1 (5) and SREBP-2 (8). Three different cDNAs for SREBP-1 were isolated, suggesting multiple forms of the mRNA and perhaps different proteins as well. The physiological significance of these subclasses is unclear (6). Different SREBP-1 proteins may have specific physiological roles because mRNAs for the various isoforms are differentially regulated by sterol depletion in HepG2 cells (8).
Proteolytic cleavage of the C-terminal membrane-associated domain of the nascent SREBP-1 (125 kDa) forms its nuclear form (68 kDa). This proteolytic maturation was proposed to be accomplished by a sterol-inhibited protease. The calpain inhibitor N-acetyl-leucyl-leucyl-norleucinal (ALLN) induced the mRNA for HMG-CoA synthase and was proposed to inhibit the degradation of the mature SREBP-1 (9).
In other sterol-regulated genes, the SRE-1 is not involved in sterol-mediated transcriptional regulation. Although the promoter region of farnesyl diphosphate synthase contains multiple forms of the SRE-1 element, these elements are not involved in the sterol-mediated transcriptional regulation (10). Similarly, the promoter of the hamster HMG-CoA reductase contains unique sites for sterol regulation. Red 25, a nuclear hamster liver protein, binds to this regulatory region but did not bind to the sterol regulatory regions of the LDL receptor and HMG-CoA synthase promoters (11).
In this report we characterize the 5Ј region of the HSS gene. The promoter activity and the sterol-mediated regulation of this DNA were assessed by fusing 5Ј-flanking DNA to a luciferase reporter gene and transfecting it into HepG2 cells and Chinese hamster ovary (CHO-K1) cells. A 69 bp DNA sequence confers transcriptional competence and sterol regulation. ADD1 binds to this 69 bp sequence in two places, one of which contains an SRE-1 element.
Probes for DNase I Footprinting and Electrophoretic Mobility Shift Assay-Double-stranded DNA probes used in DNase I footprinting were generated by restriction enzyme digestions of the 1-kb promoter of the HSS gene cloned into pBluescript. The plasmid was first digested with either BanI (noncoding strand) or NheI (coding strand). Then, the two fragments were end-labeled by incubating them with dATP, dGTP, dTTP, [␣-32 P]dCTP, and 5 units of Klenow fragment for 30 min at room temperature. After purification by ethanol precipitation, the fragments were digested with PstI (for NheI fragment) or HindIII (for BanI fragment). The 368-bp fragment from the NheI to the PstI digestion sites (nucleotides Ϫ91 to Ϫ459) and the 319-bp fragment from the BanI to the HindIII digestion sites (nucleotides ϩ73 to Ϫ246 plus 4 bp of the pBluescript vector sequence) were isolated on a 1% agarose gel and purified by Qiaex gel purification column (Qiagen).
Three probes were used in the electrophoretic mobility shift assay. One is a 156-bp DNA fragment generated by digesting the 1-kb promoter of human squalene synthase with NheI and BanI (nucleotides Ϫ91 to Ϫ246). The two other probes are synthetic oligonucleotides. One of them is a DNA containing a wild-type SRE-1 sequence existing in the HSS promoter (HSS-SRE-1), and the other has the same sequence except that four bases were mutated (nucleotides Ϫ183, Ϫ184, Ϫ186, and Ϫ187) and introduced into the SRE-1 site (HSS-SRE-1-mut). The complementary oligonucleotides were reanealed and the probes were 32 P-end-labeled by Klenow fill-in reaction as described above. The sequences of the two synthetic DNA probes are shown below with boldface letters indicating the mutated bases: HSS-SRE-1, 5Ј-TAGAGTGTTA-TCACGCCAGTCTCCTT-3Ј and 3Ј-TCACAATAGTGCGGTCAGAG-GAAGG-5Ј; HSS-SRE-1-mut, 5Ј-TAGAGTGTTATCTAGGAAGTCTC-CTT-3Ј and 3Ј-TCACAATAGATCCTTCAGAGGAAGG-5Ј.
Cloning of the 5Ј-Flanking Region of the HSS Gene-A fixII human placenta genomic library (Stratagene) was screened by plaque hybridization using the 21-nucleotide, 5Ј-32 P-labeled probe 5Ј-AGAGGT-GAGAGAGTCGCGCCC-3Ј reversely located at the 5Ј-untranslated end of the cDNA of HSS (14). The DNA of one of four isolated clones was subjected to a partial PstI endonuclease digestion. A 2-kb digestion product was identified by Southern hybridization, recovered from the agarose gel, and subcloned in a pBluescript KS vector. The resulting clone, designated pHSS2kbBS, was sequenced. Locations of sequencing primers are shown (see Fig. 2). This clone contains approximately 1.5-kb sequence 5Ј to the transcription start site (see below), the first exon of the HSS gene including the 5Ј-untranslated region, and approximately 400 bp of the first intron.
A 1-kb SmaI restriction fragment of pHSS2kbBS was isolated and subcloned into an EcoRV site of pBluescript to give the correctly oriented insert (pHSS1kb-BS) in this construct for the further preparation of the luciferase expression plasmid in the pXP1 vector (repeatedly, insertion into the SmaI site of pBluescript resulted in the reversed orientation of the insert). pHSS1kb-BS was digested with HindIII and BamHI, and the 1-kb DNA fragment was ligated to the same restriction sites in pXP1, a luciferase vector (S. K. Nordeen, University of Colorado, (15)), to form pHSS1kb-Luc. pHSS1kb-BS was digested with PstI to remove the 5Ј 477-bp of the HSS insert and religated to form pHSS532-BS. A HindIII-BamHI fragment was removed from pHSS532-BS and ligated to same sites in pXP1 vector to form pHSS532-Luc. pHSS273-BS was prepared from pHSS1kb-BS by an XbaI digestion followed by self religation. The corresponding pHSS273-Luc plasmid was prepared from the latter by inserting an HindIII-SacI 273-bp DNA fragment from pHSS273-BS into an HindIII-SstI digestion product of pXP1. The next chimeric gene pHSS204-Luc was prepared by inserting a 204-bp HindIII-MscI DNA fragment of pHSS532BS into a HindIII-SmaI-digested pXP1 vector. To prepare the smallest chimeric reporter gene, pHSS164-Luc, we first digested the pHSS1kb-BS DNA with an NheI. The linear plasmid was blunt-ended with Klenow. pHSS164-BS was prepared by a SmaI digestion of the linearized DNA following by religation. The corresponding pHSS164-Luc reporter gene was prepared from the latter by inserting a HindIII-BamHI-164-bp DNA fragment from pHSS164-BS into an HindIII-BamHI-digested pXP1 vector. pHSS1kbRev-BS containing a reversely oriented 1-kb HSS insert (relative to pHSS1kb-BS) was prepared by inserting the SmaI, HSS 1 kb, digestion product from pHSS2kbBS into the SmaI site of the BlueScript vector. The corresponding pHSS1kbRev-Luc reporter gene was prepared from the latter by inserting a HindIII-BamHI 1-kb DNA frag-ment from pHSS1kbRev-BS into an HindIII-BamHI-digested pXP1 vector. All luciferase reporter genes except pHSS1kbRev-Luc had the same 3Ј ends in a HindIII site of pXP1 and varying 5Ј ends. All plasmids were verified by sequencing.
Tissue Cultures and Transfection of Cells-HepG2 cells were grown in Dulbecco's modified Eagle's medium 121 (Flow Laboratories) supplemented with 10% (v/v) fetal bovine serum (FBS), 1 mM glutamine, 1 mM pyruvate, 100 units/ml penicillin G, and 100 mg/ml streptomycin in an 5% CO 2 incubator at 37°C. HepG2 cells were grown on 60-mm tissue culture plates to about 30% confluency. The medium was changed 18 h prior to transfection. On day 0, cells were plated at 2 ϫ 10 5 cells/plate and grown for 18 -24 h. On day 1, cells were transfected with Lipofectintreated DNA (Life Technologies, Inc.). For transfection, 5 g of test DNA (5 g), pCMV-␤GAL (5 g) (a nonregulated ␤-galactosidase expression vector) and Lipofectin (10 l) in 1 ml total volume were added to each plate according to the manufacturer's protocol. The cells were transfected in serum-free media for 6 h and then allowed to recover in 5 ml of media containing FBS. On day 2, the medium was replaced with: (a) same medium as above containing 10% (v/v) FBS; (b) medium with 10% lipoprotein-depleted calf serum (LPDS) substituted for FBS (16); (c) LPDS media supplemented with 5 g/ml lovastatin; (d) LPDS medium supplemented with 5 g/ml 25-hydroxycholesterol; (e) LPDS medium supplemented with 15 g/ml cholesterol; and (f) LPDS medium supplemented with 15 g/ml cholesterol and 5 g/ml 25-hydroxycholesterol. Cholesterol and 25-hydroxycholesterol were added in 5 l of ethanol. After incubation for 24 h with the appropriate medium, the cells were harvested, and extracts were assayed for luciferase and ␤-galactosidase activities. CHO-K1 cells were transfected and treated the same as the HepG2 cells, except that Ham's F-12 medium (Life Technologies, Inc.) was substituted for the Dulbecco's modified Eagle's medium, and 5% of both FBS and LPDS were used.
Preparation of Cell Extracts and Enzyme Assays-All assays were in triplicate. Following transfection and treatment, the cells were washed twice with phosphate-buffered saline and harvested by scraping the cells into 0.4 ml of phosphate-buffered saline. Cell suspensions were sonicated, and debris was removed by centrifugation at 14,000 ϫ g. Portions of the clear supernatant (20 -30 g of protein) were assayed for ␤-galactosidase activity using Boehringer Mannheim's assay kit and chlorophenol red-␤-D-galactopyranoside as substrate. Similar portions were assayed for luciferase activity by luminometry by measuring relative light units (RLU) in a Monolight 2010 luminometer and luciferase assay kit (Analytical Luminescence Laboratory). Relative luciferase activity is expressed as the ratio of luciferase activities (in RLU) to ␤-galactosidase activity (in A 574 ).
Primer Extension Analysis-Transcription start site(s) was determined by primer extension analysis. Poly(A ϩ ) RNA was prepared from HepG2 cells grown in media-containing LPDS and lovastatin and served as a template for the reaction. The 20-nucleotide primer 5Ј-GCGAAACTGCGACTGGTCTG-3Ј (called H10), located 164 nucleotides 3Ј to the AUG translation initiator, was 5Ј-32 P-labeled using T4-polynucleotide kinase (U. S. Biochemical Corp.) and [␥-32 P]ATP. 50 fmol of the 32 P-labeled primer (6 ϫ 10 5 cpm) and 3 g of poly(A ϩ ) RNA were used for the extension reaction. After hybridization for 45 min at 60°C, the reaction was carried out at 45°C for 60 min with RNA reverse transcriptase (100 units) and 2.5 mM dNTPs in a 40-l reaction volume. The radioactive extension product was analyzed on 5% sequencing gel using a known M13 single-stranded DNA sequencing product for size determination.
DNase I Footprinting-DNase I footprinting was performed using the Core Footprinting system kit (Promega). Briefly, 1 ϫ 10 Ϫ4 cpm of single end-labeled probes were incubated with 1 or 2 g of ADD1 protein in binding buffer containing 25 mM Tris (pH 8.0), 50 mM KCl, 6.25 mM MgCl 2 , 0.5 mM EDTA, 10% glycerol. 0.5 mM dithiothreitol and 2 g of poly(dI-dC)⅐poly(dI-dC) on ice for 15 min in a total volume of 50 l. Then, 50 l of a solution containing 5 mM CaCl 2 and 10 mM MgCl 2 was added to the mixture. After incubating the samples at room temperature for 1 min, 0.8 units of DNase I was added to each tube, and the incubation was continued at room temperature for an additional minute. The reaction was stopped by adding 90 l of a solution containing 200 mM NaCl, 30 mM EDTA, 1% SDS, and 100 g/ml yeast tRNA. The DNA samples were purified by a phenol/chloroform extraction and ethanol precipitation and resuspended in loading buffer (0.1 M NaOHformamide, 0.1% xylene cyanol, and 0.1% bromphenol blue). After denaturation at 75°C, the samples were electrophoresed on a 6% sequencing gel.
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assay was carried as described previously (17). Binding conditions were as described under "DNase I Footprinting," except that 4 ϫ 10 Ϫ4 cpm of end-labeled probe was used in each reaction, and the final reaction volume was 20 l. In competition assays, an excess amount of unlabeled HSS-SRE-1 DNA was added 5 min prior to addition of the labeled probe. Following binding, the mixture was electrophoresed on a nondenaturing 3% polyacrylamide gel at 24 mA for 3 h in a buffer containing 50 mM Tris, 100 mM glycine, and 2 mM EDTA. Detection of radiolabeled signals was by autoradiography.

Isolation of HSS Genomic Clones
We previously reported the transcriptional regulation of HSS in HepG2 cells (14). To determine the involvement of the 5Јflanking region of the gene for HSS in the transcriptional regulation of this enzyme, we first isolated a genomic clone containing this region. The genomic clone (17-2) from a fixII human placenta genomic library was isolated using HSS cDNA as a probe as described under "Materials and Methods." A partial digestion of the isolated clone with PstI restriction endonuclease followed by Southern analysis led to the isolation of a 2-kb fragment, which was subcloned into a PstI site of pBluescript KS vector and sequenced. This clone (designated pHSS2kbBS) contained at its 3Ј end the most 5Ј exon of the HSS gene, which included the 5Ј-untranslated region of the HSS cDNA (14) as well as a portion of the first intron of the gene (data not shown). The 5Ј end sequence of this clone, down to and including the ATG translation start site, is shown in Fig. 1. This parental clone was used to generate all other constructs and fusion reporter plasmids. A schematic representation of the location of the primers used for the sequencing of this genomic clone in pBluescript is shown in Fig. 2.

Determination of HSS Transcription Initiation Sites
For the preparation of the various luciferase reporter constructs, we first determined the most 5Ј transcription initiation site. This site was determined by primer extension analysis using poly(A ϩ ) RNA as a template. A 20-nucleotide primer (designated H10) located 164 nucleotides 3Ј to the AUG trans-lation initiator was 5Ј-32 P-end labeled and used for the extension reactions. The two largest primer extension products detected were 262 and 261 nucleotides long. The transcription initiation site was, therefore, calculated to be at either one of the cytidines located 97 or 98 nucleotides 5Ј to the adenosine of FIG. 1. Nucleotide sequence of the HSS promoter. The sequence shown contains 1487 nucleotides 5Ј to the transcription start site of the longest mRNA transcript and 96 nucleotides 5Ј to the ATG initiation codon, which is at the 3Ј end of the sequence. The cytidine at the transcription start site is designated ϩ1. Putative transcriptional and regulatory elements are double underlined and identified above the sequence. Endonuclease restriction sites, used for the preparation of various chimeric luciferase reporter genes, are underlined and are also identified above the sequence. The SmaI site at position ϩ73 is the 3Ј end of the HSS inserts in common to all fusion luciferase reporter genes. the AUG translation start site. The scheme for the extension and the products obtained are shown in Fig. 3. The distal cytidine located 98 nucleotides 5Ј to the adenosine at the AUG translation start site was designated 1.

Expression of Luciferase Reporter Genes
For the construction of the various 5Ј deletion sequences used for the preparation of the different chimeric luciferase reporter plasmids, we took advantage of the conveniently located SmaI site 73 nucleotides 3Ј to the transcription start site. This site was located within the 5Ј-untranslated region of the HSS cDNA and was used as the 3Ј end of the HSS DNA insert in the various luciferase fusion constructs. The different 5Ј ends of the various sequences were prepared using various existing endonuclease restriction sites in the promoter of the HSS gene. The schematic representation of the different chimeric genes containing various lengths of the HSS promoter is shown in Fig. 4. Using this methodology, we generated five fusion constructs with HSS promoter inserts, varying from approximately 1 kb (pHSS1kb-Luc) to the smallest fusion construct containing a 164-bp insert (pHSS-164-Luc). All of the five constructs included a 73-nucleotide section of the 5Ј-untranslated region of the synthase gene. In addition, we prepared a fusion gene containing the longest HSS insert in a reverse orientation (pHSS1kbRev-Luc) to be used as a negative control.
The expression of the different chimeric HSS-luciferase constructs was tested in transiently transfected cells. We chose to examine both hepatic-derived and fibroblast cell lines in order to test the possibility that the transcriptional regulation characteristics of the HSS promoter might display cell type specificity and, therefore, might exhibit a different transcriptional regulation of the luciferase reporter fusion genes in these two cell lines. Thus, we introduced the various chimeric constructs by transfection into both the human hepatoma cell line HepG2 and the hamster-derived CHO-K1 fibroblast cell line. For normalization of the activity in the different transfectant cells, we co-transfected the cells with pCMV-␤GAL, a non-sterol-regulated ␤-galactosidase expression vector, and the results were calculated as the ratio of the luciferase to the ␤-galactosidase activities.
In HepG2 cells, the highest expression of luciferase, relative to the normalizing ␤-galactosidase activity, was obtained with the pHSS1kb-Luc fusion gene. This expression resulted when the transfected cells were treated with LPDS and lovastatin in the growth media. High luciferase expression was also observed in cells transfected with pHSS532-Luc and pHSS273-Luc. However, further reduction in the size of the synthase promoter resulted in a substantially lower expression of the luciferase activity. Accordingly, pHSS204-Luc and pHSS164-Luc showed relatively low reporter activity (Fig. 5). As expected, reversal of the orientation of the HSS promoter resulted in complete loss of luciferase activity in HepG2 cells (see pHSS1kbRev-Luc in Fig. 5). With slight, although reproducible, differences, a similar pattern was observed for the expression of these constructs in CHO cells (Fig. 5). For further studies of the regulation of the synthase promoter, we chose to use the HepG2 cells.
When transfected HepG2 cells were treated with cholesterol and 25-hydroxycholesterol in LPDS-containing medium (fully suppressed conditions) or with lovastatin in LPDS medium (fully induced conditions), a definite regulation was observed. A 47.6-fold increase in luciferase activity was observed between fully induced and fully suppressed conditions in extracts of pHSS1kb-Luc transfected cells treated for 24 h (see Fig. 4). Thus, the 934-bp 5Ј-flanking region of the HSS promoter contained sequence elements that conferred sterol regulation. A similar regulatory response was observed for the pHSS532-Luc and pHSS273-Luc constructs. A 42.9-and a 51.3-fold increase between fully induced and fully suppressed conditions in HepG2 cells was observed for these two constructs, respectively (Fig. 4). Since in fully suppressed conditions the luciferase activity in all of the above three constructs is marginal, it is assumed that it may affect the accuracy of this ratio. Nonetheless, it reflects the pronounced effect sterols have on the regulation of these reporter constructs.
The relative synthase promoter activities in pHSS1kb-Luc, pHSS531-Luc, and pHSS276-Luc constructs under fully induced conditions are very similar. Repeatedly, pHSS1kb-Luc produced a somewhat higher luciferase response than the latter two (see Fig. 6). Since the amounts of the three chimeric DNA constructs as well as the amount of the DNA of the normalizing pCMV-␤GAL were kept constant in all transfections, the molar DNA equivalents driving the three luciferase expressions are not equal and are higher in the cells containing the smaller constructs. But even with this consideration, the luciferase expression in the smaller vectors is still considerable. Insertion of the 1-kb 5Ј-flanking sequence in a reverse orientation into the pXP1 expression vector completely failed to induce luciferase activity. The resulting pHSS1kbRev-Luc construct showed background levels of luminescence either under fully suppressed or fully induced conditions (Fig. 4).
Truncation of the 5Ј HSS flanking sequences at the MscI site to produce a 204 bp insert almost completely abolished its promoter activity. Only 6.8% of the luciferase activity remained, and a mere 3.4-fold increase in activity was observed for the pHSS204-Luc between fully induced and fully suppressed conditions. Similar results were also obtained for the smaller pHSS164-Luc construct. to the AUG translation initiator, was 5Ј-32 P-labeled using T4-polynucleotide kinase and [␥-32 P]ATP. 50 fmol of the 32 P-labeled primer (6 ϫ 10 5 cpm) and 3 g of poly(A ϩ ) RNA were used for the extension reaction. After hybridization for 45 min at 60°C, the reaction was carried out at 45°C for 60 min with RNA reverse transcriptase (100 units) and 2.5 mM of dNTPs in 40 l of reaction volume. The radioactive extension product was analyzed on 5% sequencing gel. An M13 single-stranded DNA sequencing product is represented by lanes G, A, T, and C and is used for size determination. The two largest 32 P-labeled extension products (EP) were of 262 and 261 nucleotides. The transcription initiation site, therefore, is calculated to be at either 98 or 97 nucleotides 3Ј to the adenosine of the AUG initiator.
Cholesterol is a much weaker regulator then 25-hydroxycholesterol. Suppression of luciferase activity with 15 g/ml cholesterol supplementation to the LPDS-containing media resulted in a decrease of activity to 17.9% of fully induced  Ϫ1kbRev-Luc are represented by a solid and hatched bar. The endonuclease restriction sites used for the construction of this plasmids are indicated above and on each side of the HSS DNA fragments. The HSS transcription initiation site, indicated by an arrow, is numbered ϩ1, and all other sites are relatively numbered. The hatched bar represents a section of the 5Ј-untranslated region in the first exon of the HSS gene and is linked at its 3Ј end (SmaI site at position ϩ73) to the coding region of a luciferase reporter gene. The HSS DNA in pHSS1kbRev-Luc is the same as the DNA in pHSS1kb-Luc but inserted in an opposite orientation. Activities were measured in transfected HepG2 cells, which were also co-transfected with a pCMV-␤GAL plasmid for the expression of nonregulated ␤-galactosidase activity in the presence and absence of sterols. Relative luciferase activities were determined as a ratio of RLU to the activity of ␤-galactosidase and were normalized to 100 for pHSS1kb-Luc in the absence of sterols. In the absence of sterols, cells were incubated in 10% LPDS containing media in the presence of 5 g/ml lovastatin. In the presence of sterols, the cells were incubated in the same LPDS media in the presence of 5 g/ml 25-hydroxycholesterol and 15 g/ml cholesterol, according to the protocol described under "Materials and Methods." Ϫ1kbRev-Luc, containing varying length promoter fragments of the human squalene synthase fused to a luciferase reporter gene and with a pCMV-␤GAL plasmid for the expression of nonregulated ␤-galactosidase activity. The cells were maintained in LPDS containing media in the presence of 5 g/ml 25-hydroxycholesterol and 15 g/ml cholesterol, according to the protocol described under "Materials and Methods." Relative luciferase activities were determined as a ratio of RLU to the activity of ␤-galactosidase and was normalized to 100 for pHSS1kb-Luc in both cell lines. The relative activity of pHSS1kb-Luc in HepG2 was 1.35 times higher then in CHO-K1 cells.
conditions, whereas the addition of 5 g/ml 25-hydroxycholesterol instead resulted in 3.5% remaining activity in cells transfected with pHSS1kb-Luc. A similar, albeit not identical ratio was observed for the cells grown in 10% FBS (see Fig. 6). Similar ratios were also observed in cells transfected with pHSS532-Luc and pHSS273-Luc. It is interesting to note that the shortest promoter in pHSS164-Luc failed completely to respond to the presence of cholesterol, but its relative activity did decrease somewhat in the presence of 25-hydroxycholesterol (Fig. 6).

Interaction of ADD1 with Promoter Sequences of HSS
In order to identify ADD1-binding regions within the HSS promoter, we performed two assays, footprinting (18,19) and gel mobility shift assays (17).
DNase I Footprinting- Fig. 7 shows that ADD1 protects DNase I digestion of promoter sequences. Both footprinting of the 368-nucleotide probe of the coding strand, including nucleotides Ϫ91 to Ϫ559 (A), and the 319-nucleotide probe of the noncoding strand, which includes nucleotides ϩ73 to Ϫ246 (B), showed digestion protected sequences. A footprint in the coding probe extending from nucleotide Ϫ176 to Ϫ196 is visible and includes the SRE-1 element 5Ј-ATCACGCCAG-3Ј. A strong footprinting signal, extending from nucleotides Ϫ165 to Ϫ191 of the noncoding probe and including the SRE-1 element, was also detected. A second footprint was detected 3Ј to the SRE-1-containing sequence. This footprint was especially visible for the noncoding strand and included nucleotides Ϫ131 to Ϫ149 containing the sequence 5Ј-TCCTAGTGTGAGCGGCCCT-3Ј (see Fig. 7B).
Electrophoretic Mobility Shift Assay-The binding of ADD1 to the HSS promoter sequences was also verified by the mobility shift assay. Fig. 8 is an ADD1 dose-response and shows increased mobility shift of a 32 P-labeled probe. The 156-bp probe, containing the sequence of nucleotides Ϫ91 to Ϫ246, which includes the SRE-1 sequence element, was shorter than the probes used for the footprinting. However, from the retardation pattern it was clear that it contained sufficient sequence for the binding of ADD1. To further elucidate the role of the SRE-1 sequence element in the binding of the DNA probe to ADD1, a shorter probe was used in the electrophoretic mobility shift assay. The probe HSS-SRE-1 is a 28-bp DNA centered with the SRE-1 sequence element found in the promoter of HSS. Fig. 9 shows that this probe binds efficiently to ADD1 in FIG. 7. DNase I footprinting of human squalene synthase promotor. A, footprinting of coding strand of a DNA fragment corresponding to Ϫ91 to Ϫ459 of the gene. B, footprinting of noncoding strand of a probe (ϩ73 to Ϫ246). Both probes were end-labeled with [␣-32 P]dCTP as described under "Experimental Procedures." The probe was incubated with or without ADD1 (1 or 2 g as indicated), followed by DNase I digestion and analysis on a 6% sequencing gel. Lane 1 is the chemical cleavage of the probe at AϩG residues, which served as a sequence marker. Lane 2 is the probe digested with DNase I in the absence of ADD1. Lanes 3 and 4 correspond to the DNase I-digested probe in the present of 1 and 2 g of ADD1. The sequence position of the probe is shown to the left of the gel. The protected region is indicated on the right side of the gel, and SRE-1 sequence is marked by a bracket.
FIG. 8. Binding of ADD1 to human squalene synthase promoter. A probe, corresponding to nucleotides Ϫ91 to Ϫ246 of the human squalene synthase promoter and containing the SRE-1 sequence 5Ј-ATCACGCCAG-3Ј, was generated by the digestion of the HSS promoter with NheI and BanI. This probe was end-labeled with [␣-32 P]dCTP. The labeled probe was incubated with various amounts of ADD1, as indicated on the top of the gel, under the condition described under "Experimental Procedures." Separation of the free and bound probes was done by electrophoresis on a 4% native polyacrylamide gel. The retarded band, which corresponds to the ADD1-bound probe and the free probe band is indicated by arrows.
the mobility shift assay and it is clearly competed for by the same unlabeled probe. Mutation of four bases at positions Ϫ183, Ϫ184, Ϫ186, and Ϫ187 in the HSS-SRE-1-mut probe, which were previously shown to be nonpermisive mutations for the binding of the SRE-1 element in the LDL receptor promoter to the SREBP-1 transcription protein (4,6), totally abolished its binding to ADD1. DISCUSSION Squalene synthase was shown to be a highly regulated enzyme in mammals (1,2,14,20). The sterol-mediated regulation of HSS in HepG2 cells was shown to be primarily transcriptional (14). Therefore, to determine whether the 5Ј-flanking region of the HSS gene contained sterol regulatory elements, we prepared chimeric fusion constructs containing various lengths of the 5Ј-flanking HSS sequences fused to a luciferase reporter gene. These fusion constructs were introduced to human hepatoma (HepG2) and nonhepatic CHO-K1 cell lines by transfection for transient expression.
Current studies show that the 5Ј-flanking region of the HSS gene confers a strong promoter activity in the fusion luciferase genes. The 1-kb, 532-bp, and 273-bp 5Ј-flanking sequences of the HSS gene have comparatively similar promoter function activity. However, deletion of 69 bp at the 5Ј end of the pHSS273-Luc construct almost completely diminished the promoter activity of the resulting pHSS204-Luc chimeric gene.
The transient expressions of pHSS1kb-Luc, pHSSXP532-Luc, and pHSS273-Luc are found to be highly responsive to the presence of sterols in both CHO (data not shown) and HepG2 cells (see Fig. 6). The sterol-mediated response of the luciferase activity in transiently transfected HepG2 and CHO cells exceeded the reported responses of similar CAT constructs containing the 5Ј-flanking regions of the genes for farnesyl diphos-phate synthase in transiently transfected HepG2 cells (10); that of the LDL receptor and HMG-CoA synthase in transiently transfected CV-1 cells (4); and stably transfected CHO cells with HMG-CoA reductase or HMG-CoA synthase CAT fusion genes (21,22).
Since there is no significant difference between pHSS532-Luc (containing HSS nucleotides ϩ73 to Ϫ459) and pHSS273-Luc (with nucleotides ϩ73 to Ϫ200) in their promoter activity and the sterol-mediated response, we can assume that the two distal CCAAT sequences at Ϫ210 and at Ϫ243 are marginally important, if at all, to the synthase promoter activity and response to sterols. The pHSS204-Luc chimeric gene (nucleotides ϩ73 to Ϫ131) retained little of the promoter activity (Fig.  5). Thus, we have to assume that essential cis-acting element(s), necessary for the transcription and the sterol-mediated regulation, are present in the 69-bp sequence between nucleotides Ϫ131 and Ϫ200. There are some recognizable potential cis-acting transcription elements located within this sequence. Starting at nucleotide Ϫ187, there is a 7:8 base pair match of the octamer cis-acting element SER-1, which was shown to confer response to sterols in several genes involved in cholesterol homeostasis (23,24). This cis-acting element is preceded by AT at its 5Ј end, which was also shown to be essential, in conjunction with the SRE-1 sequence, for sterolmediated regulation in the promoter of the LDL receptor (4). At nucleotide Ϫ163, the sequence ATTGG is a reverse CCAAT box in the opposite strand. This sequence was shown to be recognized by a family of CTF/NF-1 cellular binding proteins that are known to be involved in transcription initiation (25)(26)(27)(28). Finally, the MscI restriction site at nucleotide Ϫ128 used for the preparation of pHSS204-Luc disrupted a cis-acting TGGC-CAAT sequence, which is also a binding sequence for a CTF/ NF-1 initiation site. Thus, elimination of any one of these three sequences may explain the loss of promoter activity and the response to sterols observed for pHSS204-Luc.
The TGGCCAAT sequence is located 15 bp 5Ј to the octamer CACCCCAC, an SRE-1 consensus sequence element at nucleotide Ϫ108. However, this element lacks the AT sequence at its 5Ј end, which is essential for the binding and activity of the SREBP-1 transcription factor. The pHSS164-Luc construct was designed to eliminate this SRE-1 element. Since most of the promoter activity was lost by the deletion of sequences 5Ј to this SRE-1 element, there is no way of assessing, based on the present data, its importance as a regulatory element of the promoter for HSS. An indication that this SRE-1 sequence may not be important for sterol-mediated regulation comes from the observation that this sequence does not show footprinting in the presence of ADD1. Future experiments, involving mutational substitution and single nucleotide mutagenesis should elucidate the involvement and the functional interrelationship of the different regulatory elements.
The 5Ј-flanking region of the HSS gene also contains two adjacent Sp1 transcription elements with the sequence GGGCGG at the core of each. These two elements are present at nucleotides Ϫ57 to Ϫ40 and are located approximately 50 nucleotides 3Ј to the consensus SRE-1 sequence. Sp1 sites were shown to be promoter-specific transcription activation sequences of RNA polymerase II (29). These two elements are present in pHSS164-Luc and apparently are not sufficient to confer by themselves any significant promoter activity.
The promoter for HSS does not contain a TATA box 5Ј to the transcription initiation site. In this respect, the HSS gene is similar to the HMG-CoA reductase gene and different than other highly regulated genes in the cholesterol biosynthetic pathway, such as the genes for HMG-CoA synthase, farnesyl diphosphate synthase, and the LDL receptor (10, 24, 30, 31). There are multiple forms of the trans-acting sterol regulatory element binding protein (SREBP) factors, which bind to the SRE-1 elements of the promoters of the LDL receptor and the HMG-CoA synthase genes. The two major forms are the SREBP-1 (6) and SREBP-2 (7). However, the existence of different cDNAs for SREBP-1 may indicate the cellular existence of different forms of this protein (6). The processing and translocation of the SREBP-1 from the membrane to the nucleus was shown to be initiated by a proteolytic process. It was also reported that its further degradation by the calpain protease inhibitor ALLN increased the level of the mRNA for HMG-CoA synthase (9). In early experiments, we failed to observe an increase in the level of HSS mRNA when HepG2 cells were exposed to ALLN (data not shown). Having an HSS-luciferase reporter construct enabled the determination of the effect of ALLN on the transcription of this chimeric gene. Again, we failed to observe an increase in luminescence in response to ALLN in HepG2 cells treated with either 10% serum or LPDS. This observation is in agreement with the recent report that the inhibition of SREBP-1 degradation by ALLN treatment of HepG2 cells increased the mRNA levels for the LDL receptor but not for squalene synthase (8).
The lack of response to ALLN and the indication that ADD1 interact with a promoter element that contains an SRE-1 sequence as shown by the gel mobility shift (Figs. 8 and 9) and the footprinting (Fig. 7) assays brings up the interesting possibility that ADD1, which is considered to be the rat homologue for SREBP-1 (5, 32), may actually have a different mode of maturation. To test that, future studies will have to be done in cultured cells with transfected ADD1 precursor.
The footprinting assay indicates that purified ADD1 interacts with two sequences at the HSS promoter. One at nucleotides Ϫ165 to Ϫ191, which contains the SRE-1 element, and another at nucleotides Ϫ149 to Ϫ131. This later promoter DNA sequence does not contain an SRE-1 element nor does it contain the E-box motif CANNTG that was used in the oligonucleotide screening procedure for the isolation of the ADD1 protein (5). Thus, if this sequence is involved in the binding of ADD1 and the regulation of its promoter activity, the essential DNA motif for its binding will have to be elucidated, in future studies, by mutagenesis methodology. However, the results available clearly indicate the binding of the DNA sequence containing the SRE-1 element to ADD1 (Figs. 8 and 9). As shown, mutation of this sequence in nucleotides that were previously shown to be essential for the binding of the LDL receptor promoter to SREBP-1 (4, 6) also abolished the binding of the HSS SRE-1 sequence to ADD1.