Sterol-dependent transcriptional regulation of sterol regulatory element-binding protein-2.

We show in this manuscript that expression of the mRNA for sterol regulatory element-binding protein-2 (SREBP-2) is regulated by the cellular sterol level in cultured HeLa cells. We have cloned the 5'-flanking region of the gene encoding human SREBP-2. Characterization of this region shows the minimum 50-base pair segment, which contains a 10-base pair sterol regulatory element 1 (SRE-1) identical to the one in the human LDL receptor promoter, confers sterol responsiveness when fused to the luciferase reporter gene. Enforced expression of the truncated SREBP-2 protein (amino acid residues 1-481) also shows that this upstream segment contains the information required for transcriptional activation. The luciferase assays using mutant versions of the reporter genes reveal that the sterol-dependent transcriptional regulation is mediated by two nearby motifs, the SRE-1 and the NF-Y binding site (the inverted CCAAT box, ATTGGC); the latter is reported to play a critical role in sterol-dependent regulation of 3-hydroxy-3-methylglutaryl-coenzyme A synthase and farnesyl diphosphate synthase genes (Jackson, S. M., Ericsson, J., Osborne, T. F., and Edwards, P. A. (1995) J. Biol. Chem. 270, 21445-21448). Gel mobility shift assays demonstrate that the transcription factor NF-Y truly binds to the ATTGGC sequence. These findings suggest that the activity of SREBP-2 is controlled not only post-translationally by proteolytic activation of the precursor protein but also transcriptionally by itself together with NF-Y.

Two structurally related sterol regulatory element-binding proteins, designated SREBP-1 and SREBP-2, 1 are implicated to be central regulators in cholesterol and fatty acid metabolisms; the binding of SREBPs to a 10-bp sterol regulatory element (SRE-1) elicits the transcriptional activation of downstream genes such as HMG CoA synthase and the LDL receptor (1,2). Recent studies have also demonstrated that SREBPs are involved in transcriptional control of the genes for farnesyl diphosphate synthase (3), squalene synthase (4), fatty acid synthase, and acetyl CoA carboxylase (5)(6)(7).
SREBPs are synthesized as 125-kDa membrane-bound precursors that are localized on the nuclear envelope and the endoplasmic reticulum (8,9). In sterol-depleted cells the precursors are proteolytically cleaved to generate soluble NH 2terminal fragments (designated as the mature form with ϳ480 amino acids) containing an acidic transactivation domain and a basic helix-loop-helix-leucine zipper region that mediates protein dimerization and DNA binding. The mature form translocates to the nucleus and activates transcription. When sterols accumulate within cells, the precursors are no longer proteolysed but remain on the membranes, resulting in the decline in transcription of sterol-regulated genes.
In addition to the proteolytic activation of precursor proteins, here we report that mRNA expression for SREBP-2 is regulated by the cellular cholesterol level. The 5Ј-flanking region of the human SREBP-2 gene has been found to contain the SRE-1 and the NF-Y binding site. We further demonstrate that both sites are necessary for sterol-mediated regulation of the SREBP-2 gene transcription.

EXPERIMENTAL PROCEDURES
RNase Protection Assay-Monolayers of human HeLa cells were set up on day 0 (1.5 ϫ 10 6 cells/100-mm dish) in medium A (Dulbecco's modified Eagle's medium, 100 units/ml penicillin, 100 g/ml streptomycin, and 1 g/ml of fungizone) supplemented with 7% (v/v) fetal calf serum (FCS). On day 1, the cells were refed with medium A containing 5% lipoprotein-deficient serum (LPDS, from Sigma) supplemented with either 1 g/ml of 25-hydroxycholesterol (Sigma) plus 10 g/ml of cholesterol (Sigma) or 50 M of a HMG CoA reductase inhibitor, pravastatin (Sankyo CO. Ltd., Japan) plus 50 M of sodium mevalonate (Sigma). On day 3, the cells were harvested, and total RNA was prepared by the method of Chomczynski and Sacchi (10). Riboprobes for glyceraldehyde-3-phosphate dehydrogenase, the LDL receptor, and SREBP-2 were prepared and subjected to RNase protection assays as described previously (11,12).
Cloning of the 5Ј-flanking Region of the SREBP-2 Gene-Human genomic DNA (13) was digested with PstI and ligated to the PstI cassette according to the manual provided by Takara Biomedicals (Kyoto, Japan). The first PCR was performed with the C1 primer corresponding to the 5Ј half of the cassette and an antisense primer (S1) beginning at the 11th codon of the human SREBP-2 cDNA. The second PCR was performed with the C2 primer corresponding to the 3Ј half of the cassette and an antisense primer (S2) beginning at the 5th codon of the cDNA. PCR products were subcloned into a TA vector (Invitrogen), and their sequences were determined by the dideoxy chain termination method (14) using either a Silver Sequence DNA sequencing system (Promega) or an Applied Biosystems model 373A DNA sequencer.
5Ј RACE Analysis-The 5Ј RACE System was purchased from Life Technologies, Inc. First strand cDNA was produced using the total RNA of human hepatoma Hep G2 cells with an antisense primer correspond-* This work was supported by grants from the Ministry of Education, Science, and Culture of Japan, the Kanagawa Academy of Science and Technology Research Grants, the ONO Medical Research Foundation, the Tokyo Biochemical Research Foundation, the Kanae Foundation, and the Yamanouchi Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) D86746.
ing to the 200 -205th codon. PCR amplification was performed with an antisense primer (S1) and a universal amplification primer corresponding to the adapter region in the 5Ј RACE anchor primer. PCR products, a 240-bp single band on agarose gel, were subcloned into the TA vector, and their sequences were determined.
Tissue Cultures and Cell Transfection-Monolayers of human embryonic kidney (HEK) 293 cells were set up on day 0 (5 ϫ 10 5 cells/ 35-mm dish) in medium A supplemented with 7% FCS. On day 1, the cells were transfected by the calcium-phosphate method with 200 ng of one of the reporter luciferase plasmids, 700 ng of salmon sperm DNA, and 100 ng of pVA as described (9). After 4 h, the medium was removed, and the cells were then washed with phosphate-buffered saline and refed with medium A containing 5% LPDS supplemented with either 1 g of 25-hydroxycholesterol plus 10 g of cholesterol or 50 M of pravastatin plus 50 M of sodium mevalonate. In the cotransfection experiments, 30 ng of pSREBP2(1-481), an expression plasmid for human truncated SREBP-2 containing amino acids 1-481, replaced the same amount of salmon sperm DNA. After 4 h, the cells were refed with medium A containing 7% FCS. After a 48-h culture, the cells were processed (Promega, Luciferase assay system), and luciferase activity was measured with a luminometer (Lumat LB9501, Berthold). The protein content of cell extracts was determined by the bicinchoninic acid method (Pierce).
Construction of the Reporter Genes for Luciferase Assay-An expression plasmid, pSREBP2(1-481), was constructed by cloning the 1.5-kb XhoI-NotI fragment obtained by reverse transcription-PCR using total RNA from HeLa cells and oligonucleotides containing the above restriction sites into the XhoI-NotI site of a pME18S vector (15). The luciferase reporter plasmids were constructed by cloning the BglII-HindIII PCR fragments coding the 5Ј-untranslated region of SREBP-2 gene into the same restriction sites of a pGL2 vector (Promega Corp.). To generate p1400-Luc, p624-Luc, p140-Luc, and p91-Luc, PCR primers were designed to hybridize at the corresponding position (Primer A starts from the 5Ј end of the cloned PstI fragment for p1400-Luc, Primer B starts from nucleotide Ϫ624 for p624-Luc, Primer C starts from nucleotide Ϫ140 for p140-Luc, and Primer D starts from nucleotide Ϫ91 for p91-Luc) and coupled with the common downstream primer from nucleotide ϩ172 (Primer E). To disrupt the SRE-1, the upstream primer with 4 mutations (ATCA 3 TGAT) starting from nucleotide Ϫ140 and Primer E were used to generate pSREKO-Luc. To disrupt the NF-Y binding site, ATTGGC was replaced by the EcoRI recognition sequence. The antisense primer (Primer F) with GAATTC starting from nucleotide Ϫ93 was coupled with Primer A to produce a 1.3-kb fragment. The sense primer (Primer G) with GAATTC starting from nucleotide Ϫ116 was coupled with Primer E to produce a 0.3-kb fragment. These two fragments were combined using the EcoRI site and inserted into the pGL2 vector (named Plasmid Z). To generate pNF-YKO-Luc, PCR was performed with Primers C and E using Plasmid Z as a template to produce a 0.3-kb fragment containing NF-Y binding site mutation. The LDL receptor-luciferase plasmid, designated pLDLR, contains the 1.5-kb human LDL receptor promoter (Ϫ1563 to Ϫ58) (16).
Gel Mobility Shift Assay-Double-stranded DNA fragment corresponding to nucleotides Ϫ140 to Ϫ57 was 3Ј end-labeled with a Digoxigenin-11-ddUTP using a Dig gel shift kit (Boehringer Mannheim). The nuclear extract from Hep G2 cells cultured in medium A containing 7% FCS was prepared as described (8). An anti-NF-YB polyclonal antibody recognizing the B subunit of NF-Y was generously provided by Dr. Roberto Mantovani (17). An anti-LDLR (the LDL receptor) monoclonal antibody (C7) was from the ATCC. The reaction mixture (20 l) contained 6 g of Hep G2 nuclear extract, 30 fmol of the end-labeled probe, 20 mM Hepes-KOH (pH 7.6), 1 mM EDTA, 10 mM (NH 4 ) 2 SO 4 , 1 mM dithiothreitol, 0.2% (w/v) Tween 20, 30 mM KCl. To reduce nonspecific bindings, 2 g of double-stranded oligonucleotide (nucleotides Ϫ140 to Ϫ57) with a 6-bp mutation in the NF-Y binding site was added. Each reaction mixture was incubated at room temperature for 20 min. Following the addition of 0.5 g of antibodies, the reaction mixture was placed on ice for 30 min and then loaded directly onto a 6% polyacrylamide gel in 0.5 ϫ TBE buffer (45 mM Tris borate/1 mM EDTA). The bands were detected by anti-digoxigenin antibody (Boehringer Mannheim).

RESULTS AND DISCUSSION
HeLa cells were cultured with LPDS in the presence of either cholesterol plus 25-hydroxycholesterol or with pravastatin plus mevalonate and their total RNA was prepared. Ribonuclease protection assays demonstrated that the amounts of mRNA for the LDL receptor and SREBP-2 were significantly augmented under inducing conditions (Fig. 1, A and B); sterol depletion elicited 6-and 2-fold increases, respectively. These results in- RACE analysis of the SREBP-2 mRNA was performed and the 5Ј end of each clone is shown as ϫ on the SREBP-2 gene sequence. The 5Ј end of the longest clone is numbered ϩ1. In this analysis the 3Ј end of the cDNA was dC-tailed to generate the anchor site. Therefore, the initiation sites are not determined exactly in the three sites, G/C, G/T, and G/C.

Transcriptional Regulation of SREBP-2 26462
dicate that mRNA level of SREBP-2 is regulated in a sterol-dependent manner.
It is of interest to determine if the above phenomenon is due to transcriptional activation of the SREBP-2 gene. Thus, we isolated the 5Ј-flanking region of the gene and searched for the potential sequence motifs responsible for the activation. A 1.6-kb PstI restriction fragment of genomic DNA was amplified by PCR. The nucleotide sequence of the 0.84-kb 3Ј end of the clone is shown in Fig. 2A. The nucleotide start site of the longest clone analyzed by 5Ј RACE is designated ϩ1 (Fig. 2, A  and B). Interestingly, a SRE-1 identical to that in the human LDL receptor promoter (16) is found around Ϫ120 nucleotides from the transcription initiation site.
In order to identify the sequence motifs in the 5Ј-flanking region responsible for sterol-mediated transcriptional regulation of SREBP-2 gene, we carried out luciferase assay using various deletion constructs of reporter genes (Fig. 3A). HEK 293 cells were transfected with these reporter genes and cultured under the suppressing or inducing conditions. A 2-3-fold statistically significant increase in luciferase activity was observed under the inducing conditions with p1400-Luc, p624-Luc, or p140-Luc, whereas there was no increase with p91-Luc. This magnitude of sterol regulation was similar to that observed for the endogenous SREBP-2 mRNA (Fig. 1). These results strongly suggest that the 50-bp segment between Ϫ140 and Ϫ91 contains the cis-acting element(s) necessary for the sterol-mediated transcriptional regulation.
Because the SRE-1 locates in the 50-bp segment, we examined whether SREBP-2 itself activates the transcription of the gene. HEK 293 cells were cotransfected with one of the above reporter genes and an expression plasmid encoding the active form of SREBP-2. Coexpression of the active SREBP-2 with p624-Luc or p140-Luc resulted in 10-fold elevation of luciferase activity but with p91-Luc did not result in stimulation (Fig.  3B). Furthermore, the LDL receptor promoter reporter carrying the SRE-1 also exhibited similar stimulation of the activity. These results suggest that the 50-bp segment in the 5Ј-flanking region contains the information required for the transcriptional activation of SREBP-2 gene and that the SRE-1 may have a critical role in regulation.
It has been demonstrated that the sterol-dependent activation of the LDL receptor, fatty acid synthase, and acetyl CoA carboxylase genes requires SREBPs and the ubiquitous transcription factor Sp1 (5,7,18). In the case of the HMG CoA synthase and farnesyl diphosphate synthase genes, both SREBPs and NF-Y are necessary (3,19). These observations suggest that the synergistic interaction of the ubiquitous transcription factor and SREBPs is important for transcriptional regulation. Because the 50-bp segment of the 5Ј-flanking region of the SREBP-2 gene contains the SRE-1 and the NF-Y binding site, we hypothesized that this gene may also be regulated by a functional interaction of SREBP-2 with NF-Y (or related factors). To test this hypothesis, we constructed two versions of mutant reporter genes. The pSREKO-Luc contains the 140-bp 5Ј-flanking sequence with ATCA 3 TGAT mutations at the SRE-1 (ATCACCCCAC). It is demonstrated that each point mutation among the four nucleotides (underlined) abolishes sterol-mediated regulation of LDL receptor transcription (20). In the pNF-YKO-Luc, the ATTGGC sequence is replaced by the EcoRI sequence. When HEK 293 cells transfected with p140-Luc were cultured under the inducing conditions, the luciferase activity was significantly elevated (Fig. 4, left). The deletion of one of the two motifs reduced the response to the cellular sterol level. Cotransfection of the plasmid encoding the active form of SREBP-2 together with p140-Luc resulted in a 15-fold increase in luciferase activity, whereas no or only slight (2-fold) induction was observed in the cells cotransfected with the pSREKO-Luc or the pNF-YKO-Luc (Fig. 4, right). These results indicate that the two binding sites are required for the sterol-mediated transcriptional regulation of the SREBP-2 gene.
To confirm the binding of NF-Y to the ATTGGC sequence, gel mobility shift assay was performed. Incubation of the 83-bp probe with Hep G2 nuclear extract produced a band (Fig. 5,  lane 1). This band almost completely disappeared in the presence of an excess of unlabeled probe (lane 2) and was supershifted by antibodies (Anti-NF-YB) recognizing the B subunit of NF-Y, which is composed of three subunits, but not by antibodies to the LDL receptor (lanes 3 and 4). The same results were obtained using HEK 293 nuclear extract (data not shown). These results suggest that the general transcription factor NF-Y binds to the ATTGGC sequence in the promoter of the SREBP-2 gene and are consistent with the findings that NF-Y plays an important role for sterol-dependent regulation of the farnesyl diphosphate synthase and HMG-CoA synthase genes (3,19). The SRE-1 in the SREBP-2 promoter is recog- Transcriptional Regulation of SREBP-2 26463 nized by recombinant SREBP-2 protein in gel shift assay (data not shown). SREBP-1 and SREBP-2 were copurified from nuclear extracts of the cultured HeLa cells using a DNA affinity column, suggesting that both recognize the SRE-1 sequence (21). Except for the different tissue distribution of their mRNAs, the functional differences of these genes have remained obscure. Sheng et al. first demonstrated that the mature forms of SREBP-1 and SREBP-2 were not regulated coordinately in hamster liver and suggested that SREBP-1 is responsible for basal transcription of the LDL receptor and HMG CoA synthase genes and that SREBP-2 is responsible for increased transcription under low cholesterol conditions (22). The current data support this observation and provide the first example of the dual activation pathways for SREBP-2, the proteolytic activation of the precursor SREBP-2 protein, (23) and the transcriptional up-regulation by cellular sterol depletion.
The sterol-dependent transcriptional induction of the SREBP-2 gene was smaller than that of the LDL receptor gene in both the in vivo studies using HeLa cells (Fig. 1) and the in vitro studies using the reporter genes (Fig. 3A). Because SREBP-2 is further activated by proteolytic processing to generate the mature form, which translocates to the nucleus and activates transcription, in sterol-depleted cells, modest transcriptional induction would probably be expected, and the 5Јflanking region of the SREBP-2 gene indeed appears to be designed to activate its expression moderately. By comparing the promoter sequences of the genes regulated by sterols, it will be possible to clarify the mechanism by which the magnitude of the transcriptional induction by SREBPs is determined. Double-stranded DNA corresponding to nucleotides Ϫ140 to Ϫ57 of the SREBP-2 promoter was 3Ј end-labeled with Digoxigenin 11-ddUTP and used in gel shift studies in the presence of 6 g of Hep G2 nuclear extract. In lane 2, a 1,000-fold molar excess of competitor (an unlabeled probe) was added to the reaction. Following the addition of 0.5 g of antibodies to NF-YB and the LDL receptor, the reaction mixture was placed on ice for 30 min (lanes 3 and 4). The bound (B) and supershifted (SS) probes are indicated. The free (F) probe runs at the bottom of the gel. DNA-protein complexes transferred to nitrocellulose membrane were detected with anti-digoxigenin antibodies (Boehringer Mannheim).