Cleavage of Sterol Regulatory Element-binding Proteins (SREBPs) at Site-1 Requires Interaction with SREBP Cleavage-activating Protein

Sterol regulatory element-binding proteins (SREBPs) are membrane-bound transcription factors that promote lipid synthesis in animal cells. They are embedded in the membranes of the endoplasmic reticulum (ER) in a helical hairpin orientation and are released from the ER by a two-step proteolytic process. Proteolysis begins when the SREBPs are cleaved at Site-1, which is located at a leucine residue in the middle of the hydrophobic loop in the lumen of the ER. Sterols suppress Site-1 cleavage, apparently by interacting with a polytopic membrane protein designated SREBP cleavage-activating protein (SCAP). SREBPs and SCAP are joined together in ER membranes through interaction of their cytoplasmic COOH-terminal domains. Here we use an in vivo competition assay in transfected cells to show that the SREBP·SCAP complex is essential for Site-1 cleavage. Overexpression of the truncated COOH-terminal domains of either SREBP-2 or SCAP disrupted the complex between full-length SREBP-2 and SCAP as measured by co-immunoprecipitation. This resulted in a complete inhibition of Site-1 cleavage that was restored by concomitant overexpression of full-length SCAP. The transfected COOH-terminal domains also inhibited the transcription of a reporter gene driven by an SRE-containing promoter, and this, too, was restored by overexpression of full-length SCAP. We interpret these data to indicate that the SREBP·SCAP complex directs the Site-1 protease to its target in the lumenal domain of SREBP and that disruption of this complex inactivates the Site-1 cleavage reaction.

Sterol regulatory element-binding proteins (SREBPs) 1 are membrane-bound transcription factors that control the synthesis and uptake of cholesterol and fatty acids in animal cells (reviewed in Ref. 1). SREBPs comprise a family of three proteins, designated SREBP-1a, -1c, and -2, each of ϳ1150 amino acids in length, that are bound intrinsically to membranes of the nuclear envelope and endoplasmic reticulum (ER). In sterol-depleted cells, a two-step proteolytic sequence releases the NH 2 -terminal domains of the SREBPs, which travel to the nucleus where they activate transcription of genes encoding the low density lipoprotein (LDL) receptor; multiple enzymes of the cholesterol biosynthetic pathway, including 3-hydroxy-3methylglutaryl-coenzyme A (HMG-CoA) synthase, HMG-CoA reductase, farnesyl diphosphate synthase, and squalene synthase; and enzymes of fatty acid biosynthesis, including acetyl-CoA carboxylase and fatty acid synthase. When sterols accumulate in cells, the proteolytic processing of SREBPs is inhibited, the NH 2 -terminal domains remain bound to membranes, and transcription of the target genes declines.
Recent experiments have begun to elucidate the details of the two-step proteolytic process and the mechanism for its regulation by sterols (1)(2)(3)(4)(5). A key aspect is the three-domain structure of the SREBP precursors (1). The NH 2 -terminal domains of the SREBPs are ϳ480 amino acids in length and contain a classic basic helix-loop-helix-leucine zipper motif and an acidic transcription activation domain, similar to the ones that are found in numerous transcription factors. The NH 2terminal domain is followed by a membrane anchor domain consisting of two hydrophobic sequences, each of which spans the ER membrane and is separated from the other by a short hydrophilic loop of ϳ30 residues. The membrane anchor domain is followed by a long COOH-terminal extension of ϳ590 amino acids, which is designated as the regulatory domain. The SREBP precursors are oriented so that the NH 2 -terminal and COOH-terminal domains face the cytoplasm, and only the short hydrophilic loop projects into the ER lumen (5).
In sterol-deprived cells, the proteolytic process is initiated by an enzyme that cleaves the SREBP precursors at Site-1, which is located in the middle of the hydrophilic lumenal loop (1)(2)(3). This cleavage separates the NH 2 -terminal and COOH-terminal domains, but each remains membrane-bound, owing to its transmembrane sequence. At this point a second protease cleaves the NH 2 -terminal fragment at Site-2 within its membrane-spanning region, liberating the NH 2 -terminal domain so that it can enter the nucleus (1,2). The Site-1 cleavage enzyme is directly regulated by sterols; it acts only in sterol-depleted cells, and it is inhibited by sterols. The Site-2 enzyme is not controlled directly by sterols, but it can act only after cleavage at Site-1, and its action is therefore restricted effectively to sterol-depleted cells (1,2).
The Site-1 enzyme recognizes the sequence RXXL, which is conserved in all known mammalian and Drosophila SREBPs. The enzyme cleaves after the leucine of this sequence (1,3). In human SREBP-2, cleavage is abolished when the arginine of the RSVL sequence is changed to alanine. Site-1 cleavage also requires the COOH-terminal domain of SREBPs. When this domain is deleted, the Site-1 enzyme will no longer cleave SREBP-2 even though the RSVL sequence is still present (4).
For this reason, we refer to the COOH-terminal domain of the SREBPs as the regulatory domain.
The activity of the Site-1 cleavage enzyme is proposed to be controlled by a membrane-bound protein called SREBP cleavage-activating protein (SCAP) (6). SCAP consists of two domains: 1) an NH 2 -terminal membrane anchor of ϳ730 amino acids composed of eight putative membrane-spanning segments; and 2) a hydrophilic COOH-terminal segment of ϳ546 amino acids that contains at least four "WD repeats." These repeats, each about 40 residues in length, are found in many proteins that engage in protein-protein interactions (7). Like the SREBPs, SCAP is bound to membranes of the ER and nuclear envelope (4).
Co-immunoprecipitation assays show that the WD repeat domain of SCAP is bound to the COOH-terminal regulatory domains of the SREBPs (4). Genetic evidence indicates that SCAP is responsible for the sterol regulation of Site-1 cleavage. Thus, a dominantly acting point mutation in SCAP, D443N, was identified as the cause of sterol resistance in several independently derived lines of mutant Chinese hamster ovary (CHO) cells in which Site-1 cleavage was no longer repressed by sterols (6,8). Transfection of a cDNA encoding the D443N mutant version of SCAP into wild-type CHO cells reproduced the sterol-resistant phenotype (6). It was hypothesized that SCAP is a required cofactor in the Site-1 cleavage reaction and that sterols abolish the activity of SCAP, thereby halting Site-1 cleavage (1,6). The D443N mutation renders SCAP resistant to the inhibitory effects of sterols, and hence the cleavage at Site-1 can no longer be down-regulated.
The notion that SCAP is part of the sterol-sensing mechanism is supported by the finding that the NH 2 -terminal membrane attachment domain of SCAP bears significant sequence resemblance to the membrane attachment domain of HMG-CoA reductase (6). The latter domain serves as a sterol sensor, allowing HMG-CoA reductase to be degraded rapidly when the sterol content of the ER rises (9,10). A sequence resembling the putative sterol sensor is also present in the protein encoded by the gene defective in Nieman-Pick type C1 disease (11). A mutation in this gene prevents cholesterol from being transported normally from the lysosome to the ER. Moreover, the putative sterol sensor motif is present in the transmembrane domain of the morphen receptor Patched (11), whose ligand is a signaling protein called Hedgehog that contains cholesterol covalently attached to its COOH terminus (12). Thus, four proteins that are postulated to interact with sterols (SCAP, HMG-CoA reductase, Niemann-Pick type C1 protein, and Patched) all bear a similar membranous domain. The D443N mutation, which renders SCAP insensitive to sterols, occurs within this domain (6,8).
The current studies were designed to test the hypothesis that SCAP is necessary for the cleavage of SREBPs at Site-1. For this purpose, we transfected cells with cDNAs encoding truncated proteins bearing the COOH-terminal regulatory domain of one of the SREBPs, namely SREBP-2. We predicted that the truncated proteins would form abortive complexes with endogenous SCAP, thereby preventing it from interacting with full-length SREBP-2. The data show that the truncated forms of SREBP-2 indeed prevent the cleavage of wildtype SREBP-2 and that this inhibition can be overcome by overexpression of SCAP. We also show that Site-1 processing can be inhibited by overexpression of cDNAs encoding truncated proteins containing the WD repeat domain of SCAP, which appears to act by forming abortive complexes with SREBPs, thereby preventing their interaction with fulllength SCAP. All of these data support a model in which the interaction of the WD domain of SCAP and the COOH-terminal domain of SREBPs is required in order for SREBPs to be cleaved at Site-1.
Construction of Plasmids-All expression vectors were driven by the cytomegalovirus (CMV) promoter-enhancer contained in the pcDNA3 vector (Invitrogen). The structures of all plasmid constructs described below were confirmed by sequencing all ligation joints.
The expression vector pCMV-PSS/BP2-(504 -1141) encodes a fusion protein consisting of a modified region of the bovine prolactin signal sequence (MDSKGSSQKGSRLLLLLVVSNLLLCQGVVN), followed by a FLAG epitope tag (DYKDDDD), two novel amino acids (VD) encoded by the restriction site for SalI, and amino acids 504 -1141 of human SREBP-2. The prolactin signal sequence was modified by the substitution of an asparagine residue at Ϫ1 in place of serine, as denoted by the underline. pCMV-PSS/BP2-(504 -1141) was constructed as follows. First, a 60-bp fragment encoding the prolactin signal sequence (PSS) followed by the FLAG epitope tag was isolated by KpnI and SalI digestion of a plasmid construct in pBluescript encoding a PSS/FLAG/human thrombin receptor fusion protein (Ref. 14, provided by Dr. Shaun R. Coughlin, University of California, San Francisco). The sequence corresponding to amino acids 504 -533 of human SREBP-2 (15) was amplified by PCR of pTK-HSV-BP2 (13) with a pair of primers, 5Ј-GG-TACCGTCGACGGAGGGGCCCACGACTCTGACCA-3Ј encoding amino acids 504 -510 of SREBP-2 preceded by a SalI site and 5Ј-CTCGAGG-AATTCGTCAAACCAGCCCCCAGAACCTG-3Ј encoding amino acids 526 -533 preceded by an EcoRI site. The PCR fragment was digested with SalI and EcoRI, and the 90-bp fragment was isolated. The above two fragments were cloned into the KpnI-EcoRI sites of pcDNA3 (Invitrogen) to yield intermediate construct I. Second, intermediate construct I was mutagenized by oligonucleotide site-directed mutagenesis to substitute serine for asparagine at amino acid residue Ϫ1 of the prolactin signal sequence (denoted by underline, see above) by the method of Kunkel et al. (16) to yield intermediate construct II. Third, intermediate construct II was digested with ApaI to isolate a 6.0-kb backbone fragment, pTK-HSV-BP2 was digested with ApaI to isolate a 2.5-kb fragment encoding amino acids 506 -1141 of human SREBP-2, and the two fragments were ligated to generate pCMV-PSS/BP2-(504 -1141).
Other Plasmids-pTK-HSV-BP2, a herpes simplex virus thymidine kinase-driven expression vector encoding human SREBP-2 (13), and pCMV-SCAP, a CMV-driven expression vector encoding hamster SCAP (4), were prepared as described in the indicated reference. pCMV␤-gal, a plasmid encoding a CMV promoter-driven ␤-galactosidase reference gene, was obtained from Stratagene. pSRE-Luc, a luciferase reporter plasmid driven by a promoter consisting of three tandem copies of repeats 2ϩ3 of the LDL receptor promoter (SRE-1) plus the adenovirus E1b TATA box was constructed as described previously (6).
Culture, Transfection, and Fractionation of 293 Cells-Monolayers of human embryonic kidney 293 cells were set up on day 0 (4 ϫ 10 5 cells/60-mm dish) and cultured in 8 -9% CO 2 at 37°C in medium A (Dulbecco's modified Eagle's medium containing 100 units/ml penicillin and 100 g/ml streptomycin sulfate) supplemented with 10% (v/v) fetal calf serum. On day 2, the cells were transfected with the indicated plasmids using the MBS kit (Stratagene) method as described previously (5). Three h after transfection, the cells were switched to medium B (medium A containing 10% newborn calf lipoprotein-deficient serum, 50 M compactin, and 50 M sodium mevalonate) in the absence or presence of sterols (1 g/ml 25-hydroxycholesterol plus 10 g/ml cholesterol added in a final concentration of 0.2% ethanol) as indicated in the legends. After incubation for 20 h, the cells received N-acetylleucinal-leucinal-norleucinal at a final concentration of 25 g/ml, and cells were harvested 3 h later. The pooled cell suspension from two dishes was allowed to swell in hypotonic buffer A (10 mM Hepes-KOH at pH 7.4, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM sodium EDTA, 1 mM sodium EGTA, 1 mM dithiothreitol, and a mixture of protease inhibitors; Ref. 5) for 30 min at 0°C, passed through a 22.5-gauge needle 30 times, and centrifuged at 1000 ϫ g at 4°C for 5 min. The 1000 ϫ g pellet was resuspended in 0.1 ml of buffer B (10 mM Hepes-KOH at pH 7.4, 0.42 M NaCl, 2.5% (v/v) glycerol, 1.5 mM MgCl 2 , 1 mM sodium EDTA, 1 mM sodium EGTA, 1 mM dithiothreitol, and a mixture of protease inhibitors). The suspension was rotated at 4°C for 1 h and centrifuged at top speed in a microcentrifuge for 15 min at 4°C. The supernatant is designated nuclear extract. The supernatant from the original 1000 ϫ g spin was centrifuged at 10 5 ϫ g for 30 min at 4°C in a Beckman TLA 120.2 rotor, and the pellet was dissolved in 0.1 ml of SDS lysis buffer (10 mM Tris-HCl at pH 6.8, 100 mM NaCl, 1% (v/v) SDS, 1 mM sodium EDTA, and 1 mM sodium EGTA) and designated membrane fraction.
Culture of M19 Cells-M19 cells are a mutant line of CHO-K1 cells (19) that have a deletion that eliminates the gene encoding the SREBP Site-2 protease (2,20). The cells were grown in monolayer culture as described previously (19,20).
Glycosidase Treatment of Membrane-bound Prolactin Signal Sequence/SREBP-2 Fusion Protein-The method is described in Ref. 3. Monolayers of 293 cells were set up on day 0 (4 ϫ 10 5 cells/60-mm dish) and cultured as described above. On day 2, the cells were transfected with 3 g/dish pCMV-PSS/BP2-(504 -1141)NSS/NGT and pCMV-PSS/ BP2-(504 -587)NSS/NGT, respectively. Three h after transfection, the cells were switched to medium B in the presence of sterols. After incubation for 20 h, the cells were harvested, and the pooled cells from four dishes were fractionated. The 10 5 ϫ g membrane pellet was washed once with buffer A and resuspended in 180 l of buffer C (buffer A containing 1% (v/v) Triton X-100 without protease inhibitors). Aliquots of the 10 5 ϫ g membrane fractions (0.16 mg in 40 l of buffer C) were either incubated directly with neuraminidase or boiled for 5 min in the presence of 0.5% SDS and 1% (v/v) ␤-mercaptoethanol for 5 min. Digestion with peptide N-glycosidase F or endoglycosidase H was carried out as described in Fig. 2.
Immunoblot Analysis-Samples from nuclear extract and membrane fractions were mixed with 5 ϫ SDS loading buffer (1 ϫ loading buffer contains 30 mM Tris-HCl at pH 7.4, 3% SDS, 5% (v/v) glycerol, 0.004% bromphenol blue, 2.5% ␤-mercaptoethanol). After boiling for 5 min, the proteins were subjected to SDS-PAGE and transferred to Hybond-C extra nitrocellulose filters (Amersham). The filters were incubated with the antibodies described in the figure legends. Bound antibodies were visualized with peroxidase-conjugated affinity-purified donkey antimouse IgG using the SuperSignal CL-HRP substrate system (Pierce) according to the manufacturer's instructions. Gels were calibrated with prestained molecular weight markers (New England Biolabs). Filters were exposed to Reflection NEF-496 film (NEN Life Science Products) with intensifying screen at room temperature for the indicated time.
Luciferase Reporter Assay-On day 0, replicate wells of 6 ϫ 10 4 cells/22-mm well were plated in medium A with 10% fetal calf serum. On day 2, duplicate wells of cells were cotransfected with an MBS kit with the following plasmids as described previously (6): 0 -1 g/dish plasmid containing the indicated cDNA, 0.2 g/dish pSRE-Luc (an SRE-1-driven luciferase reporter construct), and 0.05 g/dish pCMV␤gal (a plasmid encoding ␤-galactosidase) in a final volume of 0.2 ml. After incubation for 3 h at 37°C, the cells were washed once with phosphate-buffered saline and fed with 2 ml of medium B in the absence or presence of sterols as described above. After 20 h, the cells in each well were lysed with 0.2 ml of 1 ϫ Reporter Lysis Buffer (Promega), and aliquots were used for measurement of luciferase (5 l) and ␤-galactosidase (30 l) activities. For luciferase assay, photon production was detected as relative light units in an Optima II luminometer (MGM Instrument). For ␤-galactosidase assay, hydrolysis of O-nitrophenyl-␤-D-galactosidase was detected after incubation for 1 h at 37°C in a microplate reader at a wave length of 405 nm (Bio-Tek Instruments). The amount of luciferase activity in transfectants (relative light units) was normalized to the amount of ␤-galactosidase activity (OD units) to correct for transfection efficiency in each experiment.

RESULTS
To test the ability of the COOH-terminal domain of SREBP-2 to compete for the Site-1 cleavage reaction, we prepared a series of expression vectors encoding fusion proteins in which the NH 2 -terminal domain and the first transmembrane segment of SREBP-2 were replaced by PSS followed by a FLAG epitope tag (Fig. 1). This was joined to SREBP-2 at residue 504, which marks the beginning of the hydrophilic lumenal loop. In the construct shown in Fig. 1, the SREBP-2 sequence extended to the normal COOH terminus of the protein (residue 1141), thereby including the COOH-terminal regulatory domain. To prevent cleavage of the PSS by signal peptidase, we changed the serine adjacent to the signal sequence cleavage site to asparagine, which is known to abolish cleavage by signal peptidase (21). The cDNA encoding the fusion protein was transfected into human embryonic kidney 293 cells. The cells were incubated in a sterol-containing medium to repress cleavage at Site-1. After 23 h, cell membranes were isolated, and the proteins were subjected to SDS-PAGE and immunoblotted with an antibody against the COOH-terminal regulatory domain of SREBP-2. The PSS/BP2 fusion protein was visualized as a single band of 68 kDa, which corresponded to the expected molecular weight of the protein (lane 2). To trigger cleavage of the fusion protein at Site-1, we cotransfected a cDNA encoding wild-type SCAP driven by the strong CMV promoter. High expression of wild-type SCAP is known to relieve sterol suppression of Site-1 cleavage (4,6). Under these conditions we observed a new band with an apparent molecular mass of 62 kDa, which is the size predicted if the fusion protein had been cleaved at Site-1, i.e. at L522 in SREBP-2 (lane 3). This apparent cleavage product was not seen when PSS/BP2 contained an alanine in place of arginine at position 519, a substitution that is known to abolish cleavage at Site-1 (lane 4).
To confirm that the PSS/BP2 fusion protein was positioned properly in the ER membrane, we inserted a DNA sequence encoding a short peptide that contains two consensus sites for N-linked glycosylation (designated NSS/NGT in Fig. 2). The sequence was inserted into the lumenal loop on the NH 2 -terminal side of Site-1. This protein appeared as a band at 72 kDa when the membrane fraction of transfected cells was blotted with an antibody against the SREBP-2 COOH-terminal do-main ( Fig. 2A, lane 1). The mobility increased after treatment with peptide N-glycanase F (lane 2) or endoglycosidase H (lane 3), but not with neuraminidase (lane 4). This pattern is consistent with the presence of N-linked carbohydrate chains that remained in the endoglycosidase H-sensitive form, indicating retention of the protein in the ER.
We also prepared a cDNA encoding a version of the glycosylated PSS/BP2 fusion protein that terminated at position 587 of SREBP-2, thereby eliminating the entire COOH-terminal regulatory domain (Fig. 2B). The truncated protein was visualized on SDS-PAGE by blotting with an antibody against the FLAG epitope. The truncated protein had an apparent molecular mass of 18 kDa (Fig. 2B, lane 1), and this was reduced by treatment with peptide N-glycanase F and endoglycosidase H, but not neuraminidase (lanes 2-4), indicating that it, too, was inserted into the ER and retained there. Fig. 3 shows an experiment in which we tested the ability of PSS/BP2 fusion proteins to inhibit competitively the cleavage of full-length SREBP-2. To follow the fate of the full-length SREBP-2 in the transfected cells, we inserted an HSV epitope tag into the full-length sequence. The cells were incubated in the absence of sterols to induce cleavage at Site-1. When the HSV-SREBP-2 construct was transfected alone, nuclear extracts contained the NH 2 -terminal fragment, which was visualized by blotting with an anti-HSV tag antibody (Fig. 3, lane   2). Cotransfection of increasing amounts of plasmid encoding PSS/BP2-(504 -1141) led to progressive inhibition of cleavage of full-length SREBP-2 (Fig. 3, lanes 3-8). The inhibition was nearly complete when 0.3 g of DNA was transfected (lane 5). Inhibition was also observed when the PSS/BP2-(504 -1141) protein contained the R519A mutation (lane 9). However, no inhibition was seen when we expressed the truncated PSS/ BP2-(504 -587) protein (lane 10). We also blotted the membrane fraction with the anti-HSV tag antibody, and this confirmed that the transfected cells were all expressing the HSV-SREBP-2 protein in its precursor form. We blotted the membranes with an antibody against the FLAG epitope, and this confirmed that the expression of the PSS/BP2-(504 -587) protein was at least as great as that of the PSS/BP2-(504 -1141) version (Fig. 3B).
As reviewed in the Introduction, previous experiments have shown: 1) that the COOH-terminal regulatory domain of SREBP-2 is required for the protein to be cleaved at Site-1, and 2) that this segment interacts with the COOH-terminal WD repeat domain of SCAP to form an immunoprecipitable complex. We therefore interpreted the results of the experiment of  3 and 4). The total amount of DNA in each dish was adjusted to 2 g by addition of pcDNA3 empty vector. After transfection, all cells were incubated in medium B for 23 h in the presence of sterols as described under "Experimental Procedures." Aliquots of the membrane fraction (10 g) were subjected to SDS-PAGE and immunoblot analysis with 5 g/ml IgG-1C6 (anti-SREBP-2). Filters were exposed to film for 10 s.  4 -7). The SCAP cDNA pro-duced only a slight stimulation when transfected in the absence of the PSS/BP2-(504 -1141) fusion protein (lanes 8 -11). As before, blotting of the membrane fractions with anti-HSV tag confirmed that the transfected cells were all expressing equal amounts of the HSV-SREBP-2 precursor.
The experiments of Figs. 3 and 4 are consistent with the hypothesis that the PSS/BP2-(504 -1141) protein inhibits cleavage of SREBP-2 because its COOH-terminal regulatory domain competes with this domain of SREBP-2 for interaction with SCAP. To test this hypothesis in a different way, we prepared a cDNA that encodes only the COOH-terminal regulatory domain of SREBP-2 without the lumenal loop or the second transmembrane domain (Fig. 5). To attach the COOHterminal domain to membranes, we fused it to amino acids 1-29 of cytochrome P450 -2C1. This segment was shown previously to cause the attachment of heterologous proteins to the cytoplasmic side of ER membranes (17). Transfection with increasing amounts of P450-TM/BP2-(555-1141) progressively reduced the amount of the NH 2 -terminal fragment of HSVtagged SREBP-2 that appeared in nuclear extracts (lanes 3-8).
The inhibitory effect of P450-TM/BP2-(555-1141) on cleavage of HSV-SREBP-2 was reversed by co-transfection of a plasmid encoding wild-type SCAP (Fig. 6, lanes 4 -7). SCAP did not The total amount of DNA added to each dish was adjusted to 8 g by addition of either pTK empty vector (5) or pcDNA3 empty vector. After transfection, all cells were incubated in medium B in the absence of sterols. On day 3, the cells were harvested and fractionated as described under "Experimental Procedures." Aliquots of the nuclear extracts (60 g of protein) and the membrane fraction (80 g) were subjected to SDS-PAGE and immunoblot analysis with 0.5 g/ml IgG-HSV-Tag TM for epitope-tagged SREBP-2 (A) or 20 g/ml anti-FLAG M2 IgG for the prolactin signal sequence/SREBP-2 fusion protein (B). Filters were exposed to film for 20 s (A, nuclear extracts), 10 s (A, membranes), or 15 s (B). N and P denote the cleaved nuclear and the uncleaved precursor forms of epitope-tagged SREBP-2, respectively. The circled numbers at the top of the figure refer to different cDNAs.  2-11), pCMV-PSS/BP2-(504 -1141) (lanes 3-7), or pCMV-SCAP (lanes 4 -7 and 8 -11). The total amount of DNA in each dish was adjusted to 8 g as described in Fig. 3. After transfection, cells were incubated in medium B in the absence of sterols. Aliquots of nuclear extracts (60 g of protein) and membranes (80 g) were subjected to SDS-PAGE and immunoblotted with 0.5 g/ml IgG-HSV-Tag™ for epitope-tagged SREBP-2. Filters were exposed to film for 20 s (nuclear extracts) and 10 s (membranes). N and P denote the cleaved nuclear and the uncleaved precursor forms of epitope-tagged SREBP-2, respectively. The total amount of DNA in each dish was adjusted to 8 g as described in Fig. 3. After transfection, all cells were incubated in medium B in the absence of sterols. On day 3, the cells were harvested and fractionated as described under "Experimental Procedures." Aliquots of nuclear extract (60 g of protein) and membranes (80 g) were subjected to SDS-PAGE and immunoblot analysis with IgG-HSV-Tag™ for epitopetagged SREBP-2. Filters were exposed to film for 30 s (nuclear extracts) and 20 s (membranes). N and P denote the cleaved nuclear and the uncleaved precursor forms of epitope-tagged SREBP-2, respectively. The circled numbers refer to different cDNAs. have any effect when it was transfected with the control plasmid encoding P450-TM/Gal (lanes 8 -12).
If cleavage of SREBP-2 requires interaction of the COOHterminal domains of SREBP-2 and SCAP, then overexpression of a membrane-anchored COOH-terminal domain of SCAP should also inhibit cleavage by competing with full-length SCAP for binding to SREBP-2. To test this hypothesis, we prepared a plasmid encoding the COOH-terminal WD repeat domain of SCAP fused to the P450 -2C1 membrane anchor domain (Fig. 7). As shown previously, transfection of a plasmid encoding HSV-tagged SREBP-2 led to the buildup of the NH 2terminal domain in nuclear extracts (Fig. 7, lane 2). This domain was not generated when the cells were cotransfected with the plasmid encoding P450-TM/SCAP-(731-1276) (lane 3). The inhibition was reversed by transfection of increasing amounts of a plasmid encoding full-length SCAP (lanes 4 -7). Fig. 8A (top panel) shows a co-immunoprecipitation experiment that we designed as a direct test of the hypothesis that the truncated COOH-terminal domains of SREBP-2 and SCAP block the interaction of full-length SREBP-2 and SCAP. For this purpose, 293 cells were transfected with a cDNA encoding HSV-tagged full-length SREBP-2 under control of the TK promoter, which gives a relatively low level of expression. Cell extracts were incubated with an antibody to SCAP, and the immunoprecipitates were subjected to SDS-PAGE and blotted with an anti-HSV tag antibody that visualized the full-length SREBP-2 precursor (lane 2), indicating that the HSV-SREBP-2 formed a complex with endogenous SCAP. The HSV-SREBP-2 was not co-immunoprecipitated when we expressed an excess of P450-TM/BP2-(555-1141) (lane 3) or P450-TM/SCAP-(731-1276) (lane 4). Co-immunoprecipitation was also blocked by overexpression of the chimeric prolactin signal sequence construct (lane 5), but not the PSS/BP2-(504 -587) version that terminated at position 587, which deletes the COOH-terminal domain of SREBP-2 (lane 6). The supernatants from all of the immunoprecipitations contained abundant full-length HSV-SREBP-2 (Fig. 8A, bottom panel), indicating that the truncated proteins did not interfere with the synthesis of the full-length HSV-SREBP-2 precursor. Fig. 8B shows a control immunoblot, which demonstrates that similar amounts of endogenous SCAP were present in all of the immunoprecipitates.
To confirm that the COOH-terminal domains of SREBP-2 and SCAP inhibit SREBP-2 processing specifically at cleavage Site-1, we performed an experiment in M19 cells (Fig. 9). M19 cells are a mutant line of CHO cells that lack the gene encoding S2P, the putative Site-2 protease (2,20). These cells carry out Site-1 cleavage, but the NH 2 -terminal domain of SREBP-2 remains in its membrane-bound intermediate form. When M19 cells were transfected with a cDNA encoding HSV-tagged SREBP-2 and incubated in the absence of sterols, the membrane fraction contained this intermediate form, which could be visualized by blotting with an antibody to the HSV tag (Fig.  9, lane 2). The presence of this fragment was abolished when the cells were cotransfected with plasmids encoding PSS/BP2-(504 -1141) (lane 3) or P450-TM/SCAP-(731-1276) (lane 5). In both cases the inhibition was reversed by co-transfection of a cDNA encoding full-length SCAP (lanes 4 and 6). Fig. 10 shows an experiment designed to test whether the inhibition of Site-1 cleavage of SREBPs has functional consequences for expression of SREBP target genes. For this purpose, we transfected 293 cells with a reporter construct consisting of a luciferase cDNA that was transcribed from a promoter that contains three copies of the SRE-1 element. As shown in Fig. 10A, this construct gave rise to abundant luciferase activity. The amount of luciferase was markedly reduced when we cotransfected increasing amounts of plasmids encoding P450-TM/BP2-(555-1141) or P450-TM/SCAP-(731-1276). The inhibition by both plasmids was reversed when we cotransfected  3-7), pCMV-P450-TM/Gal (lanes 8 -12), or pCMV-SCAP (lanes 4 -7 and 9 -12). The total amount of DNA in each dish was adjusted to 8 g as described in Fig. 3. After transfection, cells were incubated in medium B in the absence of sterols. Aliquots of nuclear extracts (60 g of protein) and membranes (80 g) were subjected to SDS-PAGE and immunoblotted with 0.5 g/ml IgG-HSV-Tag™ for epitope-tagged SREBP-2. Filters were exposed to film for 25 s (nuclear extracts) and 10 s (membranes). N and P denote the cleaved nuclear and the uncleaved precursor forms of epitope-tagged SREBP-2, respectively.  3-7), or pCMV-SCAP (lanes 4 -7). The total amount of DNA in each dish was adjusted to 8 g as described in Fig. 3. After transfection, cells were incubated in medium B in the absence of sterols. Aliquots of nuclear extracts (60 g of protein) and membranes (80 g) were subjected to SDS-PAGE and immunoblotted with 0.5 g/ml IgG-HSV-Tag™ for epitope-tagged SREBP-2. Filters were exposed to film for 15 s (nuclear extracts) and 20 s (membranes). N and P denote the cleaved nuclear and the uncleaved precursor forms of epitope-tagged SREBP-2, respectively. increasing amounts of a plasmid encoding full-length SCAP (Fig. 10B). DISCUSSION The current experiments provide the first evidence that formation of an SREBP-2⅐SCAP complex is necessary in order for SREBP-2 to be cleaved at Site-1. Overexpression of truncated COOH-terminal domains of SREBP-2 or SCAP disrupted the SREBP-2⅐SCAP complex as measured by co-immunoprecipitation (Fig. 8). Under these conditions, SREBP-2 failed to be cleaved at Site-1 (Figs. 3-7 and 9). Cleavage was restored by overexpression of full-length SCAP (Figs. 4, 6, 7, and 9). The failure of SREBP cleavage led to a decreased expression of SRE-dependent genes, and this, too, was reversed by overexpression of SCAP (Fig. 10). Fig. 11 shows a diagram that illustrates our interpretation of these results. As shown in previous co-immunoprecipitation experiments (4), SREBP-2 and SCAP form a complex by virtue of an interaction between their respective cytoplasmic COOHterminal domains. We hypothesize that this complex recruits an enzyme, still undefined, which we call the Site-1 protease. The active site of this enzyme must be located in the lumen of the ER to cleave SREBP-2 at leucine 522. We envision the Site-1 protease as being an intrinsic membrane protein, but it might also be a soluble lumenal protease that is bound extrinsically to the membrane, perhaps by interaction with SCAP. Overexpression of the truncated COOH-terminal domain of SREBP-2 or SCAP disrupts this complex by tying up full-length SCAP or SREBP-2, respectively. This prevents the Site-1 protease from accessing SREBP-2, thereby blocking cleavage. The cleavage can be restored by overexpression of full-length SCAP, which titrates out the inhibitor and restores SCAP⅐SREBP-2 complexes.
Although the current experiments were all performed with SREBP-2, we believe that the results are also applicable to SREBP-1a, at least in cultured cells. This is based on the data of Fig. 10, which shows that overexpression of truncated COOH-terminal domains of SREBP-2 or SCAP caused a nearcomplete block in the transcription of a reporter gene driven by an SRE-containing promoter. This implies that the action of endogenous SREBP-1a, as well as SREBP-2, is abolished by the truncated proteins.
The model in Fig. 11 is a minimalist model. It is not meant to exclude the possibility that other proteins are part of the SREBP-2⅐SCAP⅐Site-1 protease complex. We also do not mean to imply a 1:1 stoichiometry. Indeed, it is possible that multiple SCAP molecules interact with a single SREBP or vice versa. It should also be noted that sterols do not disrupt the SREBP-2⅐SCAP complex (4) even though they inhibit cleavage at Site-1. We believe that sterols interact with the polytopic membranous domain of SCAP, which resembles other proteins that are believed to interact with sterols (see the Introduction). This interaction may lead to inactivation of Site-1 protease, perhaps by displacing it from the complex, which otherwise remains intact.
In previous studies, we showed that the entire NH 2 -terminal basic-helix-loop-helix-leucine zipper domain of SREBP-2 can be replaced with an irrelevant protein such as Ha-Ras without affecting sterol-regulated cleavage at Site-1 (2). Here, we extend this finding by showing that the first transmembrane segment of SREBP-2 is also not required. Thus, the entire NH 2 -terminal and first transmembrane domains of SREBP-2 could be replaced with the prolactin signal sequence without abolishing cleavage at Site-1 (Fig. 1). We also recently showed that replacement of the second transmembrane domain of SREBP-2 with the membrane-spanning domain of the LDL receptor did not affect sterol-regulated cleavage at Site-1 (3). These findings localize the required domains of SREBP-2 to the lumenal loop and the COOH-terminal regulatory domain.
It is noteworthy that the truncated PSS/BP2-(504 -587) protein failed to inhibit cleavage of full-length SREBP-2 ( Fig. 3) even though a glycosylatable version of this protein underwent proper glycosylation (Fig. 2), indicating that the RSVL target sequence was present in the ER lumen. This finding implies that the RSVL sequence cannot interact with the active site of the protease unless it is attached to the COOH-terminal regulatory domain. One possible explanation is that the COOHterminal SREBP⅐SCAP interaction delivers SREBP to the compartment where the protease is located. This site may be within the ER or it might be in a more distal compartment such as the Golgi complex. We are currently conducting pulse-chase studies to determine precisely the cellular compartment in which Site-1 cleavage occurs.
Full confirmation of the model in Fig. 11 awaits the identification and characterization of the Site-1 protease. We have been unsuccessful so far in identifying this enzyme through the use of in vitro biochemical assays designed to measure the cleavage of SREBP-2 or peptides containing the lumenal RSVL recognition sequence. Additional approaches are under way in which we are attempting to purify the proposed SREBP-2⅐SCAP⅐Site-1 protease complex. A somatic cell genetic approach is also being pursued.
One practical result of the current studies is that they provide a new way to create dominant negative proteins that block the activation of transcription by SREBPs. It should be possible to block SREBP action in tissue culture cells and in organs of living animals through use of vectors that overexpress truncated COOH-terminal domains of SREBP-2 or SCAP. FIG. 11. Schematic diagram illustrating the competitive disruption of the SREBP-2⅐SCAP complex by the truncated COOH-terminal domains of SCAP (left) or SREBP-2 (right). The model shown is a minimalist one that does not exclude the possibility that other proteins, not yet identified, are components of the SREBP⅐SCAP⅐Site-1 protease complex. We hypothesize that the SREBP⅐SCAP complex, formed by interaction between the COOH-terminal domains, serves to attract or activate the putative Site-1 protease (S1P), which is as yet uncharacterized. When this complex is disrupted, the Site-1 protease no longer is positioned properly to cleave SREBP-2 at Site-1.