Differential Transcriptional Regulation of the Human Squalene Synthase Gene by Sterol Regulatory Element-binding Proteins (SREBP) 1a and 2 and Involvement of 5′ DNA Sequence Elements in the Regulation*

Transcription of the human squalene synthase (HSS) gene is regulated by variations in the level of cellular cholesterol. Three regulatory elements in the HSS promoter region are known to be involved in the regulation: 1) a modified sterol regulatory element (SRE) 1 (HSS-SRE-1), 2) an inverted SRE-3 (Inv-SRE-3), 3) an inverted Y box (Inv-Y-Box). We report here the regulatory role of distinct cis-elements in the HSS promoter by using mutants of an HSS-luciferase promoter reporter. The activity of a wild-type promoter reporter transiently transfected into HepG-2 cells is increased by sterol depletion of the cells or by coexpression of mature forms of the SRE-binding proteins (SREBP) 1a and SREBP-2. Differential activation by SREBP-1a and SREBP-2 of the reporter gene mutated at various regions of the promoter is observed. Mutation of either the HSS-SRE-1 or the Inv-SRE-3 sequence diminished the activation by SREBP-1a and by sterol depletion but did not affect the activation by SREBP-2. Simultaneous mutations of both of these sequences almost completely abolished activation of the promoter by SREBP-1a or by sterol depletion, but activation by SREBP-2 was retained at 70%. Mutation of the Inv-Y-Box sequence element decreased the activity of the promoter by 50% or more, and if mutated together with both SREs, the activation was almost completely abolished. Mutation of any single GC box of the two located at −40 to −57 did not affect activity, whereas simultaneous mutation of the two decreased activation by SREBP-2 by 60%, by lipid depletion by 20%, and had no effect on the activation by SREBP-1a. A Y box motif at −159 to −166 and an SRE-like sequence element (SRE-1(8/10)) at position −101 to −108 are also involved in the sterol regulation. These results indicate that the complex sterol-mediated transcriptional regulation of the HSS gene is due to the presence of multiple copies of diverse cis elements in the HSS promoter. The differential activation of the HSS promoter may point to specific role of the SREBPs in cholesterogenesis.


Transcription of the human squalene synthase (HSS) gene is regulated by variations in the level of cellular cholesterol. Three regulatory elements in the HSS promoter region are known to be involved in the regulation: 1) a modified sterol regulatory element (SRE) 1 (HSS-SRE-1), 2) an inverted SRE-3 (Inv-SRE-3), 3) an inverted Y box (Inv-Y-Box).
We report here the regulatory role of distinct cis-elements in the HSS promoter by using mutants of an HSSluciferase promoter reporter.

The activity of a wild-type promoter reporter transiently transfected into HepG-2 cells is increased by sterol depletion of the cells or by coexpression of mature forms of the SRE-binding proteins (SREBP) 1a and SREBP-2. Differential activation by SREBP-1a and SREBP-2 of the reporter gene mutated at various regions of the promoter is observed. Mutation of either the HSS-SRE-1 or the Inv-SRE-3 sequence diminished the activation by SREBP-1a and by sterol depletion but did not affect the activation by SREBP-2.
Simultaneous mutations of both of these sequences almost completely abolished activation of the promoter by SREBP-1a or by sterol depletion, but activation by SREBP-2 was retained at 70%. Mutation of the Inv-Y-Box sequence element decreased the activity of the promoter by 50% or more, and if mutated together with both SREs, the activation was almost completely abolished. Mutation of any single GC box of the two located at ؊40 to ؊57 did not affect activity, whereas simultaneous mutation of the two decreased activation by SREBP-2 by 60%, by lipid depletion by 20%, and had no effect on the activation by SREBP-1a. A Y box motif at ؊159 to ؊166 and an SRE-like sequence element (SRE-1(8/10)) at position ؊101 to ؊108 are also involved in the sterol regulation. These results indicate that the complex sterol-mediated transcriptional regulation of the HSS gene is due to the presence of multiple copies of diverse cis elements in the HSS promoter. The differential activation of the HSS promoter may point to specific role of the SREBPs in cholesterogenesis.
Two structurally related protein transcription factors designated sterol regulatory element binding protein (SREBP) 1 and 2 activate transcription by binding to the SREs (7,14). The precursor forms of the SREBPs are comprised of three segments: 1) an N-terminal region containing both a basic-helixloop-helix-leucine zipper DNA binding segment and an acidic transcription activation domain, 2) a middle domain that includes two membrane spanning regions, and 3) a C-terminal region (16). Under sterol depletion conditions, the N-terminal segments are cleaved in two separate steps, thereby releasing the active forms of the SREBPs from the endoplasmic reticulum. They are then free to migrate to the nucleus and affect gene transcription (17)(18)(19). Mutations transcribing shorter forms of the SREBP-2 protein, which include the N-terminal active domain, are not subject to this regulation and produce sterol-resistant phenotypes (20).
Studies in cultured cells indicate that the two SREBPs are coordinatively regulated. No differences between them were observed in the activation of target genes (7,14). However, in hamster liver, the SREBPs are expressed and activated independently. When liver cholesterol is depleted by dietary manipulations, expression and proteolytic activation of SREBP-1 is unaffected, whereas the transcription and activation of SREBP-2 is induced (21).
Electromobility shift studies indicate that the binding of the SREBPs to the SREs is enhanced in the presence of a general transcription factor. The promoters of the low density lipoprotein receptor, fatty acid synthase, and ACC contain GC box sequences, and the transcription factor Sp1 that binds to this motif can fulfill this function (10,12,13,22,23). The binding of SREBP-1a to the promoter of farnesyl diphosphate synthase is enhanced by the general transcription factor NF-Y, which binds to CCAAT box sequences (8,24,26). An NF-Y binding motif is also present in the promoter of 3-hydroxy-3-methylglutaryl-coenzyme A synthase and is likely to participate in the sterol-mediated regulation of this gene (27).
We have previously demonstrated that the human squalene synthase (HSS) gene is transcriptionally regulated in the HepG-2 human hepatoma cell line (28). We have also identified a 69-bp sequence in the HSS promoter, located 131 nucleotides 5Ј to the transcription start site, that is necessary for the sterol-mediated regulation of an HSS-luciferase reporter gene (pHSS1kb-Luc) (29). Sequence analysis of the 5Ј-flanking sequences of the HSS gene suggests that the sterol-mediated regulation of this gene is a complex mechanism. Four CCAAT boxes, three SRE-related motifs, and two GC boxes were identified there. Other sterol-regulated genes usually contain a single type of an SRE element and one of the two cis-elements that binds either NF-Y or Sp1. However, the presence of multiple copies of diverse elements in the HSS promoter strongly suggests that the regulation of this gene relies on multiple forms of interaction between the different transcription factors.
We have shown that adipocyte determination-and differentiation dependent factor 1/SREBP-1c, the rat homologue of SREBP-1c (30), binds to a sequence motif similar to the SRE-1 element in the low density lipoprotein receptor promoter, which we designated HSS-SRE-1. The regulatory effect of adipocyte determination-and differentiation-dependent factor 1/SREBP-1c on HSS promoter activity was verified by its activation of the pHSS1kb-Luc reporter in HepG-2 cells, indicating that the HSS-SRE-1 element in the HSS promoter may be functional (25,29). An SRE-3 element previously identified in the promoter of the farnesyl diphosphate synthase gene (8) is present in an inverted orientation (Inv-SRE-3) 32 nucleotides 3Ј to the HSS-SRE-1 sequence. In addition, an SRE motif with an 8-nucleotide sequence identity to the 10-bp SRE-1 element, designated SRE-1(8/10), is located 29 nucleotides 3Ј to the Inv-SRE-3 element. These latter two SRE sequences were shown in electromobility shift assays to selectively bind SREBP-2 but not SREBP-1 (25).
The CCAAT motif is known to bind the general transcription factor NF-Y. Two such sequences are found 17 and 49 nucleotides 5Ј to the HSS-SRE-1 element. An additional inverted CCAAT sequence called the Y box is located between the HSS-SRE-1 and the Inv-SRE-3 elements (17 nucleotides 3Ј to the HSS-SRE-1 sequence), and another form of CCAAT sequence that was previously shown to be essential in the transcription of the farnesyl diphosphate gene (8) is present in an inverted orientation (Inv-Y-Box) between the Inv-SRE-3 and the SRE-1(8/10) elements (9 nucleotides 3Ј to the Inv-SRE-3 element). The binding of NF-Y to the Inv-Y-Box was verified by gel mobility shift assays (25,29). It is of interest that the orientations of the SRE-3 and the Y box element in the HSS promoter are both inverted compared with those found in farnesyl diphosphate synthase. Their relative position in the promoter is reversed as well. In farnesyl diphosphate synthase, the Y box is 5Ј to the SRE-3 sequence and they are separated by a 21-bp spacer. In HSS, the Inv-SRE-3 is 5Ј to the Inv-Y-Box with a 9-bp separation. In both genes, the two elements are arranged sequentially to allow orientation in the same direction. This arrangement may be required for the interaction between an SREBP and NF-Y.
The detailed regulation of the HSS gene and the relative importance of the multiple motifs is not known. Single transversal mutations of DNA regions covering most of the promoter failed to completely eliminate the sterol-mediated regulation of an HSS-luciferase reporter. However, significant loss of activation was observed when the HSS-SRE-1, Inv-SRE-3, or the Inv-Y-Box sequences were mutated. Multiple mutations showed that the contribution of these motifs in the activation of the promoter is additive (25).
The purpose of the present study is to determine how the multiple motifs in the HSS promoter are involved in sterol regulation of the HSS gene. Using combinations of mutations in the HSS promoter, we have characterized elements involved in the regulation. Using these mutant promoters, we were able to identify sequences that differentially interact with SREBP-1a and SREBP-2 in the activation. These differences are, so far, unique to the HSS gene and may elucidate the specific physiological role of the SREBPs in sterol production.

EXPERIMENTAL PROCEDURES
Materials and General Methods-DNA sequencing was performed using Sequenase 7-Diaza-dGTP DNA sequencing kit (Amersham Pharmacia Biotech). Restriction enzymes and modification enzymes were purchased from New England Biolabs. [␣-32 P]dCTP was obtained from NEN Life Science Products. Lipofectin reagent was purchased from Life Technologies, Inc.
pHSS1kb-Luc, the fully functional wild-type HSS promoter-luciferase reporter gene was constructed as described previously (29). pCMV-CSA10 contains the coding sequence from human SREBP-1a from amino acids 1-490 inserted into the pCMV5 expression vector as described previously (10). pCMV-CS2 is a similar expression plasmid that produces amino acids 1-481 of the human SREBP-2 protein.
Cell Culture and Transient Transfection-Human hepatoma HepG-2 cells were maintained in 35-mm plates in a minimal essential medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum, 1 mM glutamine, 1 mM pyruvate, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C and 5% CO 2 . Transient transfection of HepG-2 cells was conducted using the Lipofectin reagent transfection procedure according to the manufacturer's manual and as described previously (29). Each transfection included 5 g of HSS promoter-luciferase reporter DNA and 4 g of DNA of pCMV-␤-galactosidase (pCMV-␤GAL) expression vector for calculation of transfection efficiency. Where indicated, various amounts of the pCMV-CSA10 and pCMV-CS2 were also included in the transfection mixture for the expression of SREBP-1a and SREBP-2. On the day after transfection, the cells were treated for 24 h with either suppressing medium, sterol(ϩ), or inducing medium, sterol(Ϫ). Sterol(ϩ) contained 1 g/ml 25-hydroxycholesterol and 10 g/ml cholesterol in 10% lipid-depleted serum; sterol(Ϫ) contained the same medium as sterol(ϩ), except that the sterols were replaced with 5 g/ml lovastatin. The procedures for cell harvesting, preparation of cell extracts, and the assays for luciferase activity and ␤-galactosidase activity have been described previously (29). Relative luciferase activity is expressed as the ratio of luciferase activity in relative light units to ␤-galactosidase activity (in A 574 ).
Generation of HSS Promoter-Luciferase Reporter Constructs-The construct used for the generation of the replacement mutations is a pBluescript phagemid containing the 1 kb 5Ј-flanking region of the HSS gene (pHSS1kb-BS, ϩ73 to Ϫ934) (29). First, single strand DNA was prepared from pHSS1kb-BS using the Sculptor in vitro Mutagenesis kit (Amersham) with minor modifications. Briefly, TG1 host cells were transformed with pHSS1kb-BS to obtain fresh transformants. On the following day, one colony was selected and grown in 2ϫ yeast tryptone medium for 3 h. VCS-M13 helper phage (Stratagene) was then added to the culture, and the incubation was continued for 4 h. The cells were then lysed, and the resulting single strand DNA was purified by PEG 8000 precipitation followed by phenol extraction and ethanol precipitation. Oligonucleotide-directed mutagenesis was performed as described previously (25). To overcome the high GC content in the mutation site, annealing and extension of mutation primers were performed at a relatively higher temperature. Therefore, the following procedure was employed. Single strand DNA and mutation primer were denatured at 94°C for 5 min and slowly cooled to 25°C to allow annealing. The annealed primer was then extended in a buffer containing 2 mM dithiothreitol, 6 mM MgCl 2 , 0.2 mM each of dNTP, and 5 units of Vent DNA polymerase in a total volume of 100 l. The reaction was kept at 0°C for 5 min, incubated at room temperature for 10 min, 55°C for 30 min, and finally, 70°C for 2 h. Four hundred units of T4 DNA ligase was then added, and the reaction was incubated at 37°C for 1 h to allow for extension and ligation. The resulting DNA was used to transform XL-1Blue competent cells. The mutated colonies were selected by colony hybridization and restriction enzyme digestion using the unique site introduced into mutation primers and verified by DNA sequencing. The mutated plasmids were used for the subsequent subcloning of the mutated 1-kbp HSS insert into HindIII and BamHI sites of a pXP1 luciferase reporter gene vector as described previously for the native promoter (29). The resulting mutations in pXP1 constructs were confirmed again by DNA sequencing, and the plasmid DNAs were prepared for transient transfections. The mutated HSS promoter regions, designated m1 to m8, are shown in Fig. 1. The sequences of the oligonucleotides used for the mutation of the different sections of the HSS promoter are shown below, with the mutated sequences in bold: To prepare replacement mutations in multiple regions, oligonucleotides containing several of the mutated sequences were employed. Thus, the doubly mutated pHSSm12-Luc is a combination mutation of both the HSS-SRE-1 and the Inv-SRE-3 sequences (containing mutations in sections m1 and m2); the doubly mutated pHSSm13-Luc is a combination mutation of both the HSS-SRE-1 and the Inv-Y-Box sequences (containing the mutations of m1 and m3); the doubly mutated pHSSm23-Luc is a combination mutation of both the Inv-Y-Box and the Inv-SRE-3 sequences (containing the mutations of m2 and m3); the doubly mutated pHSSm45-Luc is a combination mutation of the SRE-1(8/10) and the two GC boxes at position Ϫ57 to Ϫ40 (containing the mutation m4 and m5); the doubly mutated pHSSm58-Luc is a combination mutation of the Y box and the two GC boxes (containing m5 and m8); the triply mutated pHSSm123-Luc is the combination mutation of the HSS-SRE-1, Inv-SRE-3, and Inv-Y-Box sequence (containing the mutations of m1, m2, and m3); the triply mutated pHSSm124-Luc is the combination mutation of the HSS-SRE-1, Inv-SRE-3, and SRE-1(8/10) (containing the mutations of m1, m2, and m4); the triply mutated pHSSm125-Luc is the combination mutation of the HSS-SRE-1, Inv-SRE-3, and both GC boxes at Ϫ57 to Ϫ40 position (containing the mutations of m1, m2, and m5); the triply mutated pHSSm126-Luc is the combination mutation of the HSS-SRE-1, Inv-SRE-3, and the 5Ј GC box (containing the mutations of m1, m2, and m6); the triply mutated pHSSm127-Luc is the combination mutation of HSS-SRE-1, Inv-SRE-3, and the 3Ј GC box (containing the mutations of m1, m2, and m7). The various mutant constructs used in this study are summarized in Table I.

RESULTS
The functional relationship between previously identified cis-elements in the HSS promoter, a promoter that is involved in the sterol-mediated regulation, was investigated by mutational analysis. Fig. 1 shows the nucleotide sequence ϩ1 to Ϫ200 of the HSS promoter and seven cis-elements that were previously identified as potentially important in the transcription of the HSS gene. This sequence was reported to be sufficient in conferring the sterol-mediated regulation of the promoter for the HSS gene (25). Two GC boxes are present in position Ϫ40 to Ϫ57 gapped by six nucleotides. Two other sequences of interest that may potentially bind general transcription factors are Inv-Y-Box (Ϫ121 to Ϫ128) and a Y box (Ϫ159 to Ϫ161). We have also examined three other cis-elements with sequences of, or related to, SREs: an SRE-1(8/10) (Ϫ101 to Ϫ108), an Inv-SRE-3 (Ϫ138 to Ϫ147), and an HSS-SRE-1 sequence (Ϫ180 to Ϫ189) (25). Since DNA gel retardation studies have indicated differences between the SREBPs in the binding to some of the above sequences (25), in the present study we have used mutational analysis to determine the functional significance of these elements in the activation of the HSS promoter by SREBP-1a and SREBP-2.
We have constructed a series of reporter genes containing mutations in the DNA elements of interest. The sequences in pHSS1kb-Luc, which were modified by transversal replacement mutations, are indicated (m1 to m8) in Fig. 1. In addition, multiple combinations of these mutations in pHSS1kb-Luc were used for the study of the activation by the SREBPs. These  1. Mutation scheme of regulatory elements in HSS promoter. The sequence shows the 5Ј-flanking region of the HSS gene from the transcriptional initiation site up to the XbaI site at nucleotide Ϫ200 of the promoter. Transversal replacement mutations were introduced into the HSS-luciferase chimeric construct, pHSS1kb-Luc, which contains the fully functional promoter. The oligonucleotide-directed mutagenesis method used for the preparation of the replacement mutations is described under "Experimental Procedures." The position of each mutated DNA section is represented by the bar above the sequence with the designated name of the construct containing the mutation. The names of potential regulatory elements are underlined, and the regulatory element sequence is bolded. Exact transversal replacement sequences and the oligonucleotides used for the mutations are given under "Experimental Procedures." studies were done in HepG-2 cells cotransfected with the mutated reporter constructs and an expression construct for one of the SREBPs. Transfected cells were grown in the presence of cholesterol and 25-hydroxycholesterol (suppression media). Promoter activation is given as -fold induction and was calculated from the ratio of luciferase activity in cell extracts transfected with one of the SREBPs to cells without it as described under "Experimental Procedures." In each experiment, the activation of the mutated HSS promoter by the SREBPs was compared with its activation by lipid depletion in cells grown without sterols and in the presence of lovastatin (inducing media) and to the activation of the native pHSS1kb-Luc promoter in cells exposed to inducing media.
The different mutants can be divided into three main groups. The first group consists of constructs having mutations of the HSS-SRE-1 and Inv-SRE-3 elements, which are the two sequences reported to be important in the sterol-mediated regulation of other genes, and the SRE-1(8/10) element. These studies were designed to evaluate the importance of these cis-acting elements in the activation of the promoter by the two SREBPs. The results of these studies are depicted in Fig. 2. As was reported earlier (25), SREBP-2 was somewhat more effective in the activation of the wild-type promoter in pHSS1kb-Luc (Fig.  2a). Mutation of the HSS-SRE-1 sequence lowered the activation of the resulting promoter in pHSSm1-Luc by SREBP-1a or by inducing media but did not affect at all the activation by SREBP-2. At 1 g of DNA/well, the activation by SREBP-1a was about 60% that of the activation obtained by SREBP-2 (Fig. 2b). Similar results were obtained if instead, the Inv-SRE-3 sequence element was replaced (pHSSm2-Luc construct). Again, significantly lower activation of this promoter by SREBP-1a (about 45% that of SREBP-2 at 1 g of DNA/well) or by inducing media was observed, but the activation by SREBP-2 was not affected in comparison to the wild-type promoter (Fig. 2c). The most pronounced difference between the activation by SREBP-2 and SREBP-1a was observed using a promoter mutated in both the HSS-SRE-1 and the Inv-SRE-3 elements (pHSSm12-Luc). Although there is a small lowering (20% at 1 g of DNA/well) in the activation by SREBP-2, the activation by SREBP-1a or by inducing media is almost completely diminished (Fig. 2d). In our earlier report a replacement mutation of the SRE-1(8/10) element by itself had a relatively small lowering effect in the activation of the HSS promoter (less than 30%) in transfected cells exposed to inducing medium (25). However, since the regulation of the HSS promoter may be an additive phenomenon contributed by the multiple regulatory sequences, it is necessary to study this sequence after eliminating other sequences that may possibly contribute to the regulation. Thus, we have decided to reexamine the regulatory contribution of this sequence in a construct in which both the HSS-SRE-1 and the Inv-SRE-3 elements were replaced. As expected, mutation of all the three SRE elements almost completely eliminated activation by lipid depletion or by SREBP-1a. However, unlike pHSSm12-Luc, this additional mutation (pHSSm124-Luc) almost completely abolished the activation by SREBP-2 as well (compare Fig. 2, d and e).
The second group of experiments was designed to assess the importance of the Inv-Y-Box sequence in the activation of the HSS promoter by SREBPs or by sterol depletion. The actual sequence mutated is indicated as m3 in Fig. 1. The activation of a reporter gene containing this mutation (pHSSm3-Luc) is shown in Fig. 3a. There was a significantly lower activation by the SREBPs or by inducing media. Although SREBP-2 can activate the promoter to approximately 5-fold, the activation by SREBP-1a or by inducing media was only 3-fold. A similar effect was observed using a reporter construct containing, in addition, a mutation of the SRE-1 element (pHSSm13-Luc, Fig.  3b) or the SRE-3 element (pHSSm23-Luc, Fig. 3c). Although in each instance promoter activity is diminished, the activations most affected by these mutations were those induced by SREBP-1a or exposure of the cells to sterol depletion, whereas activation by SREBP-2 was approximately doubled. The most pronounced effect on activation was observed when all the three sequences above were mutated in the promoter (pHSSm123-Luc). The activity of this mutant HSS promoter was almost completely diminished regardless of the activation method (Fig. 3d).
The third group of mutations was studied to determine the involvement of the two GC box sequences in the HSS promoter. In earlier work we reported that mutation of sequences containing the most 5Ј GC box element (Ϫ52 to Ϫ57) did not affect the sterol-mediated regulation of this promoter (25). Therefore, we investigated the regulatory relationship between the two GC boxes and the three SREs located 5Ј to them. For this purpose we have first used a construct in which the two most 5Ј SREs were mutated away (pHSSm12-Luc) and in it, sequentially replaced the m6 sequences containing the 3Ј GC box element (pHSSm126-Luc), the m7 sequences containing the 5Ј GC box element (pHSSm127-Luc), and the m5 region that includes the two GC boxes (pHSSm125-Luc). Mutation of a single element in a construct lacking these SREs had minimal or no effect on the activation by the two SREBPs or by sterols. In each mutant, the activation by SREBP-2 was mostly retained (about 7-and 10-fold at 1 g of DNA/well), whereas activation by SREBP-1a or by sterol depletion was low (only 2-3-fold at 1 g of DNA/well) (Figs. 4, a and b). However, if both GC boxes were replaced, the activation by SREBP-2 was substantially lowered as well (3-fold at 1 g of DNA/well), although it was still twice as high as the activation by SREBP-1a. (Fig.  4c). Interestingly, a mutant promoter in which only the two GC boxes were replaced (pHSSm5-Luc) retained its activation by SREBP-1a or by sterol depletion, but this time the activation by SREBP-2 was significantly lower (about 4.5-fold at 1 g of DNA/well) (Fig. 4d). Mutation of the two GC boxes together with the SRE-1(8/10) caused nondiscriminatory loss of 50% activation by the SREBPs and by lipid depletion (pHSSm45-Luc in Fig. 4e), whereas mutation of the Y box in addition to the two GC box sequences (pHSSm58-Luc) had a pronounced lowering of activation by lipid depletion but little effect on the activation by the two SREBPs (compare Fig. 4, d and f). The mutation of the Y box by itself did not affect the activation by SREBP-1a. However, activation by lipid depletion was lowered by 31% and, although low concentrations of SREBP-2 were effective in the activation, further activation using amounts of SREBP-2 above 100 ng of DNA/well was not achieved (Fig. 4g).
To illustrate the differences in activation of the various mutant HSS promoters by the two SREBPs, we have calculated the -fold differences expressed as the ratio of the -fold inductions (SREBP-2/SREBP-1a). To avoid nonspecific activation by overexpression of the SREBPs, these ratios were obtained in experiments employing DNA concentrations of constructs expressing SREBPs that cause 50% maximal activation of the wild-type promoter by sterol depletion (89 ng of DNA/well for SREBP-1a and 46 ng of DNA/well for SREBP-2). The highest relative activations were observed for promoters that lack the HSS-SRE-1 and the Inv-SRE-3 sequences but still include at least a single GC box sequence (Fig. 5). The only constructs displaying negative ratios are those in which the two GC box sequences were mutated (pHSSm5-Luc and pHSSm58-Luc). DISCUSSION We have reported that the expression of HSS mRNA in HepG-2 cells is highly regulated by sterols (28). To study this

FIG. 2. Activation by SREBP-1a and SREBP-2 of HSS promoter with mutations at the HSS-SRE-1 the Inv-SRE-3 and the SRE-1(8/ 10) elements.
HepG-2 cells grown in 35-mm wells were cotransfected with 5 g of DNA/well of either wild-type or mutant HSS promoter-luciferase reporter constructs, 4 g/well pCMV-␤-galactosidase, and either the SREBP-1a expression construct pCMV-CSA10 or the SREBP-2 expression construct pCMV-CS2 at concentrations ranging from 4 to 1 g/well. The transfected cells were then treated with 25-hydroxycholesterol (1 g/well) and cholesterol (10 g/well) in medium containing 10% lipid depleted serum (suppressing medium) for 24 h. Luciferase activity in cell extracts was then measured and normalized to the activity of ␤-galactosidase. The results are expressed as the ratio of the normalized luciferase activity in cells expressing SREBPs to that in cells without the SREBPs constructs. Each data point represents the mean and S.D. of four separate determinations. sterol-mediated regulation, we have cloned and analyzed the promoter of the HSS gene (25,29). In the present study we illustrate the regulated expression of a series of HSS promoterreporter gene mutants by sterols and SREBPs. This regulation involves multiple cis-acting elements that are known to utilize the general transcription factors NF-Y and Sp1 and specific sterol regulatory sequences that bind SREBPs. These multiple cis-elements are located within 150 nucleotides (Ϫ40 to Ϫ189) of the HSS promoter.
The concentration-dependent activation of the wild-type HSS promoter by SREBP-1a and SREBP-2 shows a similar profile (Fig. 1a). It is apparent that the HSS-SRE-1 and the Inv-SRE-3 elements are important in the regulation of the promoter by SREBP-1a. The presence of these two DNA se-quences enables an additive effect in the activation by SREBP-1a, since elimination of any single one decreases the activation by approximately 50% (Figs. 1, b and c), whereas the mutation of both resulted in almost complete loss of activation by this transcription factor (Fig. 1d). From these figures it is also apparent that the regulation by sterol depletion resembles the activation by SREBP-1a. These two sequences are needed to obtain maximal regulation by sterols. Contrary to the loss of promoter activation by SREBP-1a, the activation by SREBP-2 is minimally affected by these mutations. Mutation of any one of these SREs has no apparent effect on the activation of the mutant promoter, and mutational modification of both shows limited effect (Fig. 1, a-d). This regulatory difference between the two SREBPs was not demonstrated before, to the best of On the right of each figure, shown in bars, is the activation of the wild-type and the designated mutant reporters. The activation was done in cells treated with inducing medium in which the sterols were substituted with 5 g/well lovastatin ((Ϫ)Sterols). Shown are activations of wild-type HSS promoter (pHSS1kb-Luc) (a); mutation of HSS-SRE-1 element (pHSSm1-Luc) (b); mutation of Inv-SRE-3 element (pHSSm2-Luc) (c); double mutation of both the HSS-SRE-1 and the Inv-SRE-3 elements (pHSSm12-Luc) (d); triple mutation of the HSS-SRE-1, Inv-SRE-3, and SRE-1(8/10) elements (pHSSm124-Luc) (e). our knowledge, with any other sterol-regulated genes and may indicate the unique characteristics of HSS as the first committed enzyme in the cholesterol biosynthetic pathway.
In an earlier study, it was reported that a replacement mutation of the SRE-1(8/10) element by itself had a relatively small effect in lowering of the activation of the HSS promoter (less than 30%) by inducing medium. However, this sequence was shown to interact with SREBP-2 in a gel mobility shift assay (25). Since the sterol-dependent regulation of the HSS promoter is affected by several regulatory sequences acting in concert, the relative contribution of a single regulatory sequence to the overall response may be obscured. Therefore, to assess the importance of any element it is necessary to study this sequence after eliminating other sequences that may contribute to the regulation (e.g. HSS-SRE-1 and the Inv-SRE-3). As expected, mutation of all the three SRE elements almost completely eliminated the activation by SREBP-2, indicating possible interaction between the SRE-1(8/10) sequence and SREBP-2 (compare Fig. 2, d and e). This evidence does not eliminate a possible interaction of the SREBP-1a with the SRE-1(8/10) element, since the activation of the pHSSm12-Luc promoter by SREBP-1a is already low. In fact, some evidence indicates that the involvement of SRE-1(8/10) in the activation may be nondiscriminatory for the SREBPs, since in a mutant promoter that lacks the two GC boxes, mutation of the SRE-1(8/10) sequence affects primarily the activation by SREBP-1a (compare Fig. 4, d with e).
The Inv-Y-Box cis-sequence element (Ϫ121 to Ϫ128) is also important in the transcriptional activation of the HSS promoter. This sequence motif was shown earlier to be essential for the regulation by sterols in several sterol-regulated genes. For instance, the binding of SREBP-1a to the SRE-3 element in farnesyl diphosphate synthase is enhanced by the binding of NF-Y to the CCAAT box (8,24,26). Efficient sterol-mediated regulation of the gene for SREBP-2 was also shown to require the presence of SRE-1 and NF-Y elements (15). A general FIG. 4. Activation by SREBP-1a and SREBP-2 of HSS promoter with mutations at the two GC boxes and the HSS-SRE-1, Inv-SRE-3, SRE-1(8/10), and Y box elements. Treatment of cells, promoter activation assays, and display of data are as described in Fig. 2. Shown are activations of HSS promoter with a triple mutation of the HSS-SRE-1, Inv-SRE-3, and 5Ј-GC box elements (pHSSm126-Luc) (a); triple mutation of the HSS-SRE-1, Inv-SRE-3, and 3Ј-GC box elements (pHSSm127-Luc) (b); triple mutation of the HSS-SRE-1, Inv-SRE-3, and two GC box elements (pHSSm125-Luc) (c); mutation of the two GC box elements (pHSSm5-Luc) (d); double mutation of the SRE-1(8/10) and the two GC box elements (pHSSm45-Luc) (e); double mutation of the two GC box and the Y box elements (pHSSm58-Luc) (f); mutation of the Y box element (pHSSm8-Luc) (g). scheme for the interaction between sterol regulatory proteins and general transcription factors in the sterol-mediated regulation was suggested in which the binding of a sterol regulatory transcription factor to its sequence element is enhanced by the binding of a general protein transcription factor to a nearby sequence (10,15,26). Similar to the mechanism suggested for these genes, the Inv-NF-Y-Box is also shown here to be important in the efficient activation of the HSS promoter. Its mutation affects both the activation by sterol depletion or by the SREBPs. Mutation of this sequence by itself (Fig. 3a) or together with either the HSS-SRE-1 or the Inv-SRE-3 (Fig. 3, b and c) reduced the activation dramatically. Mutation of both SREs together with a mutation of the Inv-Y-Box totally abolished activation by sterol depletion, and only residual activation by high levels of the SREBPs was retained (Fig. 3d). If the general scheme for sterol mediated-regulation involving an SREBP interaction with a general transcription factor (such as NF-Y) is valid for HSS as well, then the limited activation observed in Fig. 3, a-c must be due to a non-sterol-specific cis-element in the promoter, possibly the Y box.
As indicated above, the complex regulatory mechanism of HSS is apparent by the high remaining activation by SREBP-2 of a mutant promoter lacking both the HSS-SRE-1 and the Inv-SRE-3 elements (Fig. 2d). This indicates the existence of a remaining SREBP-2 recognition sequence. To completely abolish the SREBP-induced or sterol-mediated activation of the HSS promoter, mutation of the Inv-Y-Box is necessary as well. This triple mutation exists in pHSSm123-Luc. In this mutated promoter, the activation by SREBP-2 was completely abolished (compare Figs. 2d with Fig. 3d).
At the 3Ј end of the 150-bp sequence containing the multiple cis-elements there are two adjacent GC box elements (Fig. 1). Originally, we assumed that these GC box sequences were of no significance to either the transcription or to the sterol-mediated regulation of the HSS promoter, since mutation of each alone did not affect the activation by sterol depletion (data not shown). In fact, mutation of both elements together resulted in only a marginal decrease of the activation by sterol depletion and had no effect on the activation by SREBP-1a. However, further examination indicated that simultaneous mutation of these two elements significantly decreased the activation by SREBP-2 (Fig. 4d). To eliminate the possible activation by endogenous SREBP-1a that may be produced in the cells despite the added sterols, we have taken advantage of the observation that mutation of the two 5Ј SREs differentially inhibited the activation by SREBP-1a but not the activation by SREBP-2 (Fig. 2d). Accordingly, we have used constructs that lack these two SREs to study the significance of the two GC box sequences in the specific activation by SREBP-2. These studies show redundancy in the functionality of the two GC boxes. The presence of either one of the GC box sequences is sufficient to allow activation by SREBP-2 (compare Fig. 4, a and b, to Fig.   FIG. 4-continued  2d). However, simultaneous mutation of the promoter region m5, which contains the two GC box sequences, and the two 5Ј SREs significantly lowered the activation by SREBP-2 (Fig. 4c). It can be concluded that if, as proposed above, there is an SREBP-2 binding sequence other than the two 5Ј SREs, it may affect activation of the HSS promoter by interacting with an Sp1 transcription factor that can bind to either of these ciselements. Additional multiple mutation studies are needed to indicate the involvement of the SRE-1(8/10) together with the two GC boxes in the activation. The effect observed here is in agreement with the reported interaction between SREBPs and Sp1 in the activation of the promoters for the low density lipoprotein receptor, fatty acid synthase, and ACC (10 -13, 22, 23). However, in contrast to HSS, no specificity between the SREBPs was reported for these genes.
The pronounced differential activation of mutant HSS promoters by the two SREBPs is most apparent in Fig. 5. Since we could not determine the amounts of the SREBPs expressed in the cells by transfection, we have used DNA constructs expressing SREBPs at concentrations that yielded 50% of the activation caused by sterol depletion of the wild-type HSS promoter. This, we assume, assures that the SREBPs are not overexpressed in transfected cells above physiological levels, and therefore, the differential effect of the SREBPs on the mutant promoters can be determined. It is apparent that under these conditions the activation by SREBP-1a was affected to a higher extent than by SREBP-2. This differential effect is most pronounced in promoters in which the two 5Ј SREs are mutated. The differential effect is reversed when the two GC box elements are eliminated, which provides additional evidence that the two SREBPs may not interact interchangeably with regulatory sequences at the HSS promoter. It is interesting that when, in addition to the mutation of the two GC boxes, the Y box is also mutated, the relative activation by SREBP-2 is further reduced (compare pHSSm5-Luc with pHSSm58-Luc in Fig. 5). A possible explanation is that the Y box is involved in enhancing a specific interaction of SREBP-2 with one of the SREs. This possibility is supported by the difference in the activation of a Y box mutant (pHSSm8-Luc) by the two SREBPs. Interestingly, at low levels of the SREBPs, they are both as effective in the activation. However, at concentrations above 100 ng of DNA/well, unlike for SREBP-1a, further activation by SREBP-2 was not achieved (Fig. 4g). This profile of activation supports possible interaction of the SREBP-2 with multiple SRE sequences at the promoter. Those with relatively high affinity for SREBP-2 were not affected by the mutation of the Y box, and thus, low concentrations of SREBP-2 are effective. However, the general transcription factor that binds to the Y box may affect, primarily, the interaction of the SREBP-2 with a low affinity element(s), which regulates transcription only at relatively high levels of SREBP-2 (hence the relatively low decrease in activation by lipid depletion). In this model, the SRE(s) with the low affinity is SREBP-2-specific and did not bind SREBP-1a.
The data presented suggest that in most mutants the regulation by sterols in HepG2 cells is similar to the activation by SREBP-1a and not to the activation by SREBP-2. The exception to this are mutants in the Y box in which the sterol regulation parallels the activation by the latter (see Figs. 4, f and g). FIG. 5. Differential activation of the various HSS promoter mutants by SREBP-1a and SREBP-2. Treatment of cells and promoter activation assays are as described in Fig. 2. -Fold differences in activation of the various mutant HSS promoters by the two SREBPs are expressed as the Fold Differences and are displayed as bars for the designated mutants. Positive values (upper part) indicate relative higher activation by SREBP-2 and are expressed by the ratio SREBP-2/SREBP-1a. Negative values (lower part) were assigned to ratios in which SREBP-1a is more active and are the ratio Ϫ(SREBP-1a/SREBP-2). These ratios were calculated in experiments employing DNA concentrations of constructs expressing SREBPs that cause 50% maximal activation of the wild-type promoter by sterol depletion (89 ng of DNA/ well for SREBP-1a and 46 ng of DNA/well for SREBP-2). The highest -fold differences are obtained for promoters mutated at the HSS-SRE-1 and the Inv-SRE-3 sequences but still maintain at least a single GC box sequence. The constructs displaying negative ratios are those in which the two GC box sequences (pHSSm5-Luc and pHSSm45-Luc), the Y box sequence (pHSSm8-Luc), or both (pHSSm58-Luc) are mutated.