Cleavage site for sterol-regulated protease localized to a leu-Ser bond in the lumenal loop of sterol regulatory element-binding protein-2.

A sterol-regulated protease initiates release of the NH2-terminal segments of sterol regulatory element-binding proteins (SREBPs) from cell membranes, thereby allowing them to enter the nucleus and to stimulate transcription of genes involved in the uptake and synthesis of cholesterol and fatty acids. Using SREBP-2 as a prototype, we here identify the site of sterol-regulated cleavage as the Leu522-Ser523 bond in the middle of the 31-residue hydrophilic loop that projects into the lumen of the endoplasmic reticulum and nuclear envelope. This site was identified through use of a vector encoding an SREBP-2/Ras fusion protein with a triple epitope tag that allowed immunoprecipitation of the cleaved COOH-terminal fragment. The NH2 terminus of this fragment was pinpointed by radiochemical sequencing after replacement of selected codons with methionine codons and labeling the cells with [35S]methionine. Alanine scanning mutagenesis revealed that only two amino acids are necessary for recognition by the sterol-regulated protease: 1) the leucine at the cleavage site (leucine 522), and 2) the arginine at the P4 position (arginine 519). These define a tetrapeptide sequence, RXXL, that is necessary for cleavage. Cleavage was not affected when the second transmembrane helix of SREBP-2 was replaced with the membrane-spanning region of the low density lipoprotein receptor, indicating that this sequence is not required for regulation. Glycosylation-site insertion experiments confirmed that leucine 522 is located in the lumen of the endoplasmic reticulum. We conclude that the sterol-regulated protease is a novel enzyme whose active site faces the lumen of the nuclear envelope, endoplasmic reticulum, or another membrane organelle to which the SREBPs may be transported before cleavage.

Proteolytic processing of sterol regulatory element-binding proteins (SREBPs) 1 controls the metabolism of cholesterol and fatty acids in animal cells (1)(2)(3). SREBPs are transcription factors that are bound to membranes of the ER and nuclear envelope. Each SREBP is composed of three segments: 1) an NH 2 -terminal segment of ϳ485 amino acids that is a transcription factor of the basic helix-loop-helix-leucine zipper family, 2) a membrane attachment segment of ϳ75 amino acids composed of two membrane-spanning sequences separated by a short hydrophilic loop of 31 amino acids, and 3) a COOH-terminal segment of ϳ585 amino acids that plays a regulatory role. The proteins are oriented so that the NH 2 -and COOH-terminal segments project into the cytoplasm, and only the short hydrophilic loop projects into the lumen of the ER or nuclear envelope (4).
Before it can activate transcription, the NH 2 -terminal segment is released from the membrane in a complex two-step proteolytic sequence (2,3). First, a protease cleaves the protein at Site-1, which is near an arginine in the lumenal loop, thereby breaking the attachment between the two transmembrane sequences. This allows a second protease to cleave the protein at Site-2, which is near the middle of the first transmembrane sequence (2,3). The NH 2 -terminal fragment leaves the membrane with a portion of the first transmembrane sequence still attached. It enters the nucleus, where it activates transcription of genes encoding the LDL receptor (5,6), several enzymes of cholesterol biosynthesis (3-hydroxy-3-methylglutaryl coenzyme A synthase (5, 6), 3-hydroxy-3-methylglutaryl coenzyme A reductase (7), farnesyl diphosphate synthase (8), and squalene synthase (9)), and enzymes of fatty acid biosynthesis (10,11) and desaturation (12). The net result is to increase the cell's supply of cholesterol and fatty acids.
The Site-1 protease is the target of feedback regulation by cholesterol and other sterols (3). When these sterols accumulate within cells, the rate of proteolysis at Site-1 declines markedly. Cleavage at Site-2 also declines because this cleavage requires prior cleavage at Site-1 (3). As a result, the amounts of nuclear SREBPs decline, and transcription of the target genes falls. The net effect is to prevent overaccumulation of cholesterol and fatty acids when intracellular sterol levels are already high.
Three isoforms of SREBP are known (5,13,14). SREBP-1a and 1c are derived from a single gene through use of alternate promoters that encode alternate first exons (5,13,14). SREBP-1a is much more active than SREBP-1c in stimulating transcription of all known target genes (15). The third protein, SREBP-2, is the product of a separate gene (6,13), and it is also more active than SREBP-1c (15). In view of the regulatory role of the Site-1 protease, further knowledge of its structure and mode of regulation is desirable. A first step would be the identification of the precise site at which the Site-1 protease cuts SREBPs. In previous studies we have shown that this cleavage is abolished when arginine 519 in the lumenal loop of SREBP-2 (or the corresponding arginine of SREBP-1a) is changed to alanine by in vitro mutagenesis (2). The size of the cleavage product, as determined by SDS-PAGE, is consistent with cleav-age at or near this arginine (3).
In the current studies we have used a combination of in vitro mutagenesis, epitope tagging, immunoprecipitation, and radiochemical sequencing to determine the precise location of Site-1 in SREBP-2. We found, surprisingly, that cleavage does not occur at arginine 519, but rather it occurs 3 residues further toward the COOH terminus, namely at leucine 522. Arginine 519 seems to be the NH 2 -terminal residue in a tetrapeptide sequence, RXXL, that serves as a recognition signal for the Site-1 protease.

EXPERIMENTAL PROCEDURES
Materials-We obtained HSV-Tag TM and T7-Tag TM monoclonal antibodies from Novagen; v-H-Ras(Ab-1)-agarose linked monoclonal antibody was obtained from Oncogene. Protein G-Sepharose R 4 Fast Flow beads were obtained from Pharmacia Biotech Inc., L-[ 35 S]methionine (Ͼ1000 Ci/mmol) was obtained from DuPont NEN, and glycosidases were obtained from New England Biolabs.
pTK-HSV-BP2-Ras-T7 was constructed from a previously described plasmid, pTK-HSV-BP2-T7 (16), by insertion of a cDNA segment encoding human H-Ras between the sequences for SREBP-2 and the first copy of the T7 epitope. The nucleotide sequence encoding amino acids 2-189 of human H-Ras was obtained by polymerase chain reaction on pRcCMV-H-Ras (17) with a pair of primers containing an NdeI site at each 5Ј end using Pfu DNA polymerase. The amplified product was digested with NdeI and cloned into the unique NdeI site between human SREBP-2 and the three tandem copies of the T7 epitope. Two independent clones were used in each of the transfections.
Plasmid pTK-HSV-BP2(R519A)/LDLRTM is identical to pTK-HSV-BP2/LDLRTM except for the R519A point mutation in the lumenal loop of the SREBP-2 sequence. This plasmid was constructed by site-directed mutagenesis.
Site-directed Mutagenesis-Oligonucleotide site-directed mutagenesis was carried out with single-stranded, uracil-containing DNA (19) using the Muta-gene Phagemid In Vitro Mutagenesis Version-2 kit (Bio-Rad) (2). The mutations were confirmed by sequencing the relevant region, and at least two independent clones of each mutant were independently transfected.
Cell Culture, Transfection, and Cell Fractionation-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) supplemented with 10% (v/v) fetal calf serum (4). On day 2, cells were transfected with 4 g of pTK empty vector (mock) or the indicated plasmid as described (2). 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 as indicated in the legends. After incubation for 20 h, the cells received N-acetyl-leucinal-leucinal-norleucinal at a final concentration of 25 g/ml (2), and the cells were harvested 3 h later (2). The pooled cell suspension from 2 dishes was allowed to swell in hypotonic buffer (4) 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 7 min. The 1000 ϫ g pellet was resuspended in 0.1 ml of buffer C (10 mM Hepes-KOH (pH 7.4), 0.42 M NaCl, 2.5% (v/v) glycerol, 1.5 mM MgCl 2 , 0.5 mM sodium EDTA, 0.5 mM sodium EGTA, 1 mM dithiothreitol, and a mixture of protease inhibitors (4)). The suspension was rotated at 4°C for 1 h and centrifuged at top speed in a microfuge 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 100.2 rotor, and the pellet was dissolved in 0.1 ml of SDS lysis buffer (1) and designated membrane fraction.
Glycosidase Sensitivity of Membrane-bound SREBP-2-Glycosidase sensitivity of SREBP-2 was carried out as described (4). 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 7 g of pTK empty vector (mock), pTK-HSV-BP2-NGT, and pTK-HSV-BP2-NSS/NGT, respectively. Three h after transfection, the cells were switched to medium B in the presence of sterols as described in Fig. 1. After incubation for 20 h, the cells were harvested, and the pooled cell suspension from four dishes was fractionated. The 10 5 ϫ g membrane pellet was washed once with buffer A (4) and resuspended in 180 l of buffer A containing 1% (v/v) Triton X-100 without protease inhibitors. Aliquots of the 10 5 ϫ g membrane fraction (0.16 mg in 40 l of buffer A) were boiled for 5 min in the presence (peptide N-glycosidase F and endoglycosidase H reactions) or in the absence (neuraminidase reactions) of 0.5% (w/v) SDS and 1% (v/v) ␤-mercaptoethanol for 5 min, after which the indicated amount of glycosidase was added and incubated at 37°C for 2 h as described in Fig. 10.
Immunoblot Analysis-Samples of the nuclear extract and the 10 5 ϫ g membrane fraction were mixed with 5ϫ SDS loading buffer (20). Protein concentration was measured with a BCA kit (Pierce). After SDS-PAGE in 8% gels, proteins were transferred to Hybond-C extra nitrocellulose membranes (Amersham Corp.). Immunoblot analysis was carried out with a horseradish peroxidase detection kit using the Su-perSignal TM CL-HRP Substrate System according to the manufacturer's instructions except that the nitrocellulose sheets were blocked in phosphate-buffered saline containing 0.05% (v/v) Tween 20, 5% (v/v) nonfat dry milk, and 5% (v/v) heat-inactivated newborn calf serum. The chimeric proteins were visualized with 0.5 g/ml HSV-Tag TM monoclonal antibody or with 10 g/ml IgG-1C6, a mouse monoclonal antibody directed against amino acids 833-1141 of human SREBP-2 (4). Gels were calibrated with prestained molecular weight markers. Filters were exposed at room temperature to Reflection NEF-496 film (DuPont NEN).

Metabolic Labeling and Immunoprecipitation of Epitope-tagged SREBP-2/Ras Fusion Protein and Its COOH-terminal Fragment-
Monolayers of 293 cells were set up on day 0 (7 ϫ 10 5 cells/60-mm dish) and cultured as described above. On day 1, the cells were transfected with 4 g of the wild-type or mutant version of pTK-HSV-BP2-Ras-T7, 1 g of pCMV-SCAP(D443N) (16), and 2 g of pVAI as described above. pVAI encodes the adenovirus virus-associated I RNA gene, which enhances translation of transfected cDNAs (21). The cells were incubated for 20 h with 50 M compactin and 50 M sodium mevalonate, at which time the medium was changed to 1.3 ml of methionine-free Dulbecco's modified Eagle's medium supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, 10% newborn calf lipoprotein-deficient serum, 50 M compactin, and 50 M sodium mevalonate. After incubation for 1 h at 37°C, 25 g/ml N-acetyl-leucinal-leucinal-norleucinal was added, and the cells were pulse-labeled with 700 Ci/ml of [ 35 S]methionine for 6 h at 37°C. The cells from four dishes were harvested and pooled, and the membrane fraction was prepared as described above.
The pooled membrane fraction from the four dishes was resuspended in 0.1 ml of SDS lysis buffer at room temperature. All subsequent operations were carried out at 4°C unless otherwise stated. The suspension was rotated for 30 min in 5 ml of buffer D (50 mM Tris-HCl (pH 7.5), 125 mM NaCl, 0.5% (v/v) SDS, 1.25% (v/v) Triton X-100, 1.25% (v/v) deoxycholate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 5 g/ml pepstatin A, 25 g/ml N-acetyl-leucinalleucinal-norleucinal, and 1 mM dithiothreitol), after which 20 l of preimmune whole rabbit serum, 40 g of irrelevant mouse monoclonal antibody IgG-2001 (22), and 0.2 ml of Protein G-Sepharose beads were added. After rotation for 16 h, the mixture was centrifuged at 1000 ϫ g for 7 min. The resulting supernatant was mixed with an additional 20 l of preimmune whole rabbit serum, 40 g of irrelevant monoclonal antibody IgG-2001, and 0.2 ml of Protein G-Sepharose beads, rotated for 2 h, and centrifuged at 1000 ϫ g for 3-7 min. To the supernatant were added 1 g of T7-Tag monoclonal antibody, 25 g of v-H-Ras(Ab-1)-agarose linked monoclonal antibody, and 20 g of anti-COOH-terminal SREBP-2 monoclonal antibody IgG-1C6. After rotation for 2.5 h, 50 l of Protein G-Sepharose beads were added, followed by rotation for 2.5 h and centrifugation at 1000 ϫ g for 3-7 min. The beads were washed once by rotation with buffer D for 16 h, followed by four washes in buffer D for 1 h each. The washed beads were resuspended in 0.1 ml of 2ϫ SDS loading buffer (20) containing 10% (v/v) ␤-mercaptoethanol and boiled for 5 min. After centrifugation at 1000 ϫ g at room temperature for 3 min, the supernatant was transferred to a fresh tube (tube A), and the beads were re-eluted with 40 l of 5ϫ SDS loading buffer containing 25% ␤-mercaptoethanol and boiled for 5 min. After centrifugation at 1000 ϫ g at room temperature for 3 min, the supernatant was transferred to tube A, and the entire volume was boiled again for 5 min before SDS-PAGE.

NH 2 -terminal Sequence Analysis of 35 S-Labeled COOH-terminal Fragment of Epitope-tagged SREBP-2 Fusion
Protein-Aliquots of the immunoprecipitated samples (from 1.3 dishes of cells) were subjected to SDS-PAGE on 8% gels and transferred to polyvinylidene fluoride membranes (Immobilon-P SQ ; Millipore). After drying, the membranes were exposed to an imaging plate and scanned in a Fuji X Bas 1000 phosphorimager. The band containing the COOH-terminal product of the cleavage reaction (M r ϳ83,000) was excised and subjected directly to multiple cycles of Edman degradation on an Applied Biosystems model 477A sequencer. Fractions from each cycle (148 l) were collected and counted in a scintillation counter.

RESULTS AND DISCUSSION
To identify the exact position of Site-1, we prepared a cDNA (pTK-HSV-BP2-Ras-T7) encoding a triply tagged version of SREBP-2, which we designate SREBP-2/Ras (Fig. 1). The NH 2 terminus contains two copies of an epitope tag from the HSV glycoprotein that we used previously (2). At the COOH terminus we inserted amino acids 2-189 of H-Ras followed by three copies of an 11-residue epitope derived from the gene 10 protein of bacteriophage T7 (23). The combination of epitopes at the COOH terminus allowed efficient precipitation of the COOHterminal fragment with a mixture of antibodies directed against the Ras and T7 epitopes plus a monoclonal antibody directed against the COOH-terminal domain of SREBP-2 (see below).
To demonstrate that SREBP-2/Ras is cleaved at Site-1 in a physiologic fashion, we transfected 293 cells with a vector encoding this construct and another encoding a mutated version in which arginine 519 was changed to alanine (Fig. 1). We used the relatively weak thymidine kinase promoter, which produces near physiological levels of this protein (2). Cells were incubated in inducing medium that contains the 3-hydroxy-3methylglutaryl coenzyme A reductase inhibitor compactin to block cholesterol synthesis plus a low concentration of mevalonate to provide nonsterol end products. The medium was either devoid of sterols (Ϫ sterols) or it was supplemented with a mixture of 25-hydroxycholesterol and cholesterol (ϩ sterols), which suppresses cleavage at Site-1 (3). Cell membranes were isolated, solubilized with SDS, subjected to SDS-PAGE, and immunoblotted with IgG-1C6, a monoclonal antibody directed against the COOH terminus of SREBP-2 (4).
When cells were transfected with the cDNA encoding the SREBP-2/Ras chimeric protein, the membranes contained the 83-kDa COOH-terminal fragment, which was the product of cleavage at Site-1 (Fig. 1, lane 3, transfected C). The amount of this fragment was reduced in the presence of sterols (lane 4). The IgG-1C6 antibody also visualized the 62-kDa COOH-terminal fragment of endogenous SREBP-2 (lane 3, endogenous C), and this was also decreased by sterols (lane 4). To test the physiologic relevance of the observed cleavage of SREBP-2/Ras, we cotransfected a cDNA encoding the D443N mutant version of SCAP, a protein that was previously shown to render the cleavage of SREBPs at Site-1 resistant to suppression by sterols (16). Indeed, in the presence of SCAP, sterols no longer prevented the appearance of the COOH-terminal fragment of SREBP-2/Ras (lane 6). Note that SCAP did not prevent suppression of the cleavage of endogenous SREBP-2. We believe that this is due to the small percentage of cells that received the SCAP cDNA. Unlike SREBP-2/Ras, which is expressed only in transfected cells, endogenous SREBP-2 is present in all cells in the dish, most of which did not receive the SCAP cDNA. To increase the expression of the transfected SREBP-2/Ras chimera, we cotransfected pVAI, which encodes a protein that stimulates translation of mRNAs produced by transfected cDNAs (21). In the presence of pVAI plus pSCAP, the COOH-  8 and 11-14). After transfection, the cells were incubated in the absence (Ϫ) or presence (ϩ) of 1 g/ml 25-hydroxycholesterol plus 10 g/ml cholesterol (sterols) as indicated. On day 3, the cells were harvested and fractionated as described under "Experimental Procedures." Aliquots of the membranes (80 g of protein) were subjected to SDS-PAGE and immunoblot analysis with 10 g/ml IgG-1C6, an antibody directed to the COOH-terminal domain of SREBP-2. The filters were exposed to film for 30 s. P and C denote the uncleaved precursor form and the cleaved COOH-terminal fragment of SREBP-2, respectively.
terminal fragment of the chimeric protein was increased by at least 8-fold, and there was still no suppression by sterols ( lanes  7 and 8). The R519A mutant of the chimeric protein did not give rise to detectable amounts of COOH-terminal cleavage product in either the absence or presence of SCAP (lanes 9 -12). In the presence of pVAI plus pSCAP, a trace amount of the R519A mutant COOH-terminal fragment was detectable, but its amount was reduced by at least 12-fold when compared with the wild-type COOH-terminal fragment (compare lanes 13 and  14 with lanes 7 and 8). These data indicate that Site-1 cleavage of SREBP-2/Ras obeys all of the rules established for wild-type SREBP-2. Moreover, as discussed below, the use of pVAI and pSCAP enabled the production of sufficient amounts of SREBP-2/Ras to allow radiochemical sequencing.
In view of the high specific radioactivity of [ 35 S]methionine, we developed a strategy to use this amino acid for radiochemical sequencing. Assuming that the cleavage site is located near arginine 519, we searched for amino acids in the region that could be replaced with methionine without abolishing cleavage at Site-1. Fig. 2 shows two experiments in which methionine residues were introduced individually into positions 525, 526, or 529 of the triply taggged SREBP-2/Ras protein ( Fig. 2A). To follow the cleavage, we prepared nuclear extracts and immunoblotted with an antibody against the NH 2 -terminal HSV epitope tag. The E525M, S526M, and G529M mutants were each cleaved efficiently as judged from the amount of NH 2terminal fragment in the nucleus, and all of the cleavages were suppressed by sterols (Fig. 2B). In the experiment shown in lanes 1-6, both the cleaved wild-type and E525M proteins migrated as a doublet. Such doublets are seen rarely and inconsistently in our experiments, and their significance is unknown.
To determine whether the COOH-terminal fragment could be labeled with [ 35 S]methionine and immunoprecipitated, we transfected 293 cells with a cDNA encoding the SREBP-2/Ras chimera and labeled the cells with [ 35 S]methionine. A membrane fraction was prepared, solubilized with detergents, and immunoprecipitated with a mixture of three antibodies directed against H-Ras, the T7 epitope tag, and the COOHterminal segment of SREBP-2. The immunoprecipitate was subjected to SDS-PAGE and phosphorimager analysis, which revealed a band of ϳ83 kDa that was absent from mock-transfected cells (Fig. 3A). Immunoblot analysis of the precipitate (Fig. 3B) and the supernatant (data not shown) confirmed that all of the COOH-terminal fragment was precipitated. In future experiments, the 83-kDa radiolabeled band was excised from the nitrocellulose membrane and used for radiochemical sequencing. Fig. 4 shows a series of experiments in which 293 cells were transfected with cDNAs encoding the SREBP-2/Ras chimera with the wild-type sequence in the lumenal loop or with the three methionine substitutions at positions 525, 526, and 529. The cells were then incubated with [ 35 S]methionine. After immunoprecipitation and electrophoresis, the COOH-terminal fragments were eluted and subjected to automated Edman degradation, and the radioactivity released in each cycle was determined. The protein with the wild-type SREBP-2 sequence did not show any peak of 35 S radioactivity in any of the early cycles (Fig. 4A). The E525M mutant showed a clear peak of 35 S radioactivity in cycle 3 from the Edman degradation (Fig. 4B). The S526M and G529M mutants showed peaks at cycles 4 and 7, respectively (Fig. 4, C and D). These findings indicate strongly that the COOH-terminal fragment begins with serine 523 and that Site-1 cleavage occurs between this residue and  2 and 4) were subjected to SDS-PAGE, after which the nitrocellulose filter was exposed to an imaging plate for 15 min at room temperature to detect 35 S radioactivity (A). The same filter was subjected to immunoblot analysis with 10 g/ml IgG-1C6, an antibody directed against the COOH-terminal domain of SREBP-2 (B). Chemiluminescence from the bound secondary antibody was detected by exposure to x-ray film for 1 s. Asterisk ‫)ء(‬ denotes the 83-kDa COOH-terminal fragment of the SREBP-2/Ras fusion protein. leucine 522 (indicated by the arrows in the amino acid sequences of Fig. 4).
To determine the amino acid residues that are important for cleavage of SREBP-2 at Site-1, we made a systematic series of mutations in which alanine was individually substituted for most of the amino acids in the lumenal loop between the two membrane-spanning sequences. The expression vector was a cDNA encoding SREBP-2 with an NH 2 -terminal epitope tag derived from the HSV glycoprotein. Expression was driven by the relatively weak thymidine kinase promoter, which produces sufficient SREBP-2 to allow detection, yet not so much as to overwhelm the system for regulated proteolysis (2). Transfected 293 cells were incubated in inducing medium plus or minus sterols, after which nuclear extracts and membrane fractions were subjected to SDS-PAGE and immunoblotted with an antibody against the NH 2 -terminal epitope tag. Fig. 5 shows illustrative results from one of the alanine scanning experiments. When the cDNA encoded the wild-type SREBP-2 sequence, we observed the mature NH 2 -terminal fragment of SREBP-2 in nuclear extracts (Fig. 5, lane 3). This fragment was abolished in the presence of sterols (lane 4). The R519A mutant was not cleaved (lane 5), but the D510A and Q511A mutants were cleaved as efficiently as the wild-type protein (lanes 7 and 9, respectively). Fig. 6 summarizes the results of all 27 alanine scanning experiments. At the top we show the sequences of the lumenal loops of four of the SREBPs that have been sequenced, two from the human (5, 6) and two from the hamster (24,25). Conserved residues are highlighted. Only two single alanine substitutions reduced cleavage dramatically: the R519A and the L522A mutations (Mutant No. 11 and 14, respectively). Three of the multiple alanine replacements reduced cleavage (Mutant No. 21,24,and 25). All of these mutations included the L522A substitution. The other multiple alanine replacements did not reduce cleavage. These no-effect mutations included the change of Arg 519 -Ser-Val-Leu 522 to Arg 519 -Ala-Ala-Leu 522 (Mutant No. 23). Cleavage was also not affected when the serine following the cleavage site was changed to alanine (S523A, Mutant No. 15). These data indicate that arginine 519 at the P4 position and leucine 522 at the P1 position are the only 2 residues in the region of Site-1 that cannot tolerate alanine substitutions.
To further dissect the requirement for arginine 519, we systematically replaced this residue with 12 different residues (Fig. 7). When lysine was substituted for this arginine, only a partial loss of cleavage activity was observed. Two negatively charged residues (glutamic acid and aspartic acid) permitted a barely detectable level of cleavage. All other substitutions abolished cleavage. The location of arginine 519 was critical. Cleavage was abolished when we replaced arginine 519 with alanine and then inserted a new arginine at positions 517, 520, or 521. These mutations effectively moved the position of the arginine either toward the NH 2 terminus or COOH terminus.
We also tested the specificity of the requirement for leucine 522 at the cleavage site (Fig. 8). Cleavage was markedly reduced but not totally abolished when this residue was changed to arginine, alanine, or phenylalanine. It was abolished when leucine 522 was changed to glutamic acid or valine. The residue at the PЈ1 position was not critical. This residue is serine in human and hamster SREBP-2, but it is glutamic acid and  D443N). On day 2, the cells were radiolabeled with [ 35 S]methionine, harvested, and fractionated, after which the cleaved COOH-terminal fragment of human SREBP-2 was immunoprecipitated. Aliquots of the immunoprecipitated samples (from 1.3 dishes of cells) were subjected to SDS-PAGE and transferred to polyvinylidene fluoride membranes. The membranes were exposed to an imaging plate and scanned in a Fuji X Bas 1000 phosphorimager. The band containing the COOH-terminal product of the cleavage reaction was excised and subjected to multiple cycles of Edman degradation. The radioactivity recovered at each cycle of Edman degradation is plotted. Amino acid sequences shown at the top of each panel correspond to the SREBP-2 sequence in the region of the postulated site of sterol-regulated cleavage.
FIG. 5. Immunoblot analysis of HSV-tagged SREBP-2 in 293 cells transfected with wild-type and alanine loop mutants. 293 cells were set up for experiments, transfected with 4 g of either wild-type pTK-HSV-BP2 or the indicated mutant version of the same plasmid, incubated in the absence or presence of sterols, and fractionated as described in the Fig. 1 legend. Aliquots of the nuclear extracts (60 g of protein) and membranes (80 g) were subjected to SDS-PAGE and immunoblot analysis with 0.5 g/ml HSV-Tag TM antibody. The filters were exposed to film for 15 s. M and P denote the cleaved NH 2 -terminal mature and uncleaved precursor forms of SREBP-2, respectively. The other bands are present in mock-transfected cells and represent proteins that cross-react with the anti-HSV tag antibody. glycine in hamster and human SREBP-1, respectively. Replacement of serine 523 in human SREBP-2 with glutamic acid or glycine preserved normal cleavage. The cleavage was also unaffected when this residue was changed to arginine, alanine, leucine, or phenylalanine. A moderate reduction was observed when it was changed to cysteine.
To determine whether the second transmembrane domain contributes to recognition by the Site-1 protease, we replaced this sequence with the membrane-spanning region of another protein, namely, the LDL receptor (Fig. 9). The LDL receptor is a type 1 transmembrane protein of the plasma membrane with a single membrane-spanning segment oriented with its NH 2 terminus in the extracellular space and its COOH terminus in the cytosol (18). This orientation is the same as the orientation of the second transmembrane domain of SREBP-2 (4). The chimeric protein containing the LDL receptor transmembrane domain was cleaved as efficiently as the wild-type protein, as judged by the amount of NH 2 -terminal fragment found in the nucleus (Fig. 9, lane 5). Cleavage was abolished by sterols (lane 6). It was also abolished when arginine 519 of the chimera was changed to alanine (lane 7). These data indicate that the precise sequence of the second transmembrane domain is not important for cleavage at Site-1 (or Site-2). A hydrophobic sequence at this position is required, however. When the second transmembrane domain was deleted without replacement by another transmembrane domain (HSV-SREBP-2⌬535-560), cleavage at Site-1 and -2 was abolished (data not shown).
Previously we presented evidence that the lumenal loop sequence is indeed in the ER lumen (4). This evidence was based on studies of a cDNA encoding a chimeric protein that contained an epitope-tagged peptide segment (218 amino acids) inserted into the lumenal loop (4). We were forced to insert a long protein segment because shorter epitopes were not recognized by antibodies after insertion into the lumenal loop of SREBP-2. Sealed membrane vesicles were prepared from cells expressing this protein, and the segment containing the epitope was shown to be protected from digestion by trypsin in the absence, but not the presence, of detergents. The epitopecontaining segment was also demonstrated to undergo N-linked glycosylation. The problem with these studies is that the chimeric protein was not cleaved by the Site-1 protease. Therefore, we could not be certain that the long epitope-containing peptide had not altered the orientation of the protein.
To circumvent this problem, we took advantage of the observation that the precise sequence on the NH 2 -terminal side of arginine 519 is not essential for Site-1 cleavage. Therefore, we prepared a plasmid encoding forms of SREBP-2 in which we inserted either 3 or 9 amino acids into this sequence in place of serine 515 (Fig. 10A). The inserted amino acids included either one or two sites for N-linked glycosylation (Asn-Xaa-Ser/Thr). After transfection, membrane pellets containing the SREBP-2 Amino acid alignment of the loop region between transmembrane regions 1 and 2 of hamster and human SREBP-1 and SREBP-2 is shown as in Fig. 6. The leucine at position 522 and the serine at position 523 in human SREBP-2 were individually mutated to the indicated amino acids. The cleavage of each mutant relative to that of wild-type human SREBP-2 is shown at the right. precursor were digested with glycosidases, and the change in mobility on SDS-PAGE was determined by immunoblotting. As shown in Fig. 10B, the mobility of the SREBP-2 precursors with either one or two glycosylation sites was increased by treatment with peptide N-glycosidase F (lanes 2 and 7) and by endoglycosidase H (lanes 3 and 8), but not by neuraminidase (lanes 4 and 9). The changes were greater for the construct with two glycosylation sites. This pattern indicates that the lumenal loop sequence contained N-linked carbohydrates of the high mannose type. The lack of processing to an endoglycosidase H-resistant form indicates that the precursor is located in the ER and not the Golgi complex. Fig. 10C shows immunoblots of nuclear extracts and membrane pellets from cells expressing SREBP-2 with the wild-type lumenal sequence, the R519A mutation, or the insertion of one or two N-linked glycosylation sites. The cells were incubated under conditions that induce SREBP-2 cleavage (Ϫ sterols) or suppress cleavage (ϩ sterols). The membrane-bound precursor form with one glycosylation site (lanes 7 and 8) migrated slower than the wild-type protein (lanes 3 and 4), and the form with two glycosylation sites migrated even slower (lanes 9 and 10). Under inducing conditions the nuclear extracts contained the NH 2 -terminal fragments of all proteins except the R519A mutant. Cleavage of all proteins was suppressed by sterols. This experiment demonstrates that the glycosylated precursor of SREBP-2 remained susceptible to sterol-regulated cleavage at Site-1 and subsequent cleavage at Site-2. The presence of the carbohydrate chains did, however, impede cleavage. The absolute amount of cleavage was reduced by ϳ50% for the construct containing one glycosylation site and by ϳ80 -90% for the protein containing two glycosylation sites.
Considered together, the data in this paper indicate that the Site-1 protease cleaves the peptide bond between leucine 522 and serine 523 in the lumenal loop of SREBP-2. The only residues that seem to be crucial for recognition are arginine 519  Fig. 1. Aliquots of the nuclear extracts (60 g of protein) and membranes (80 g) were subjected to SDS-PAGE and immunoblot analysis with 0.5 g/ml HSV-Tag TM antibody. The filters were exposed to film for 7 s. M and P denote the cleaved NH 2 -terminal mature and uncleaved precursor forms of SREBP-2, respectively. The other bands are present in mock-transfected cells and represent proteins that crossreact with the anti-HSV tag antibody. Aliquots of the 10 5 ϫ g membrane fraction from 293 cells transfected with the indicated cDNA were boiled for 5 min as described under "Experimental Procedures" and incubated for 2 h at 37°C with one of the following glycosidases: lanes 1, 5, and 6, none; lanes 2 and 7, 0.038 IU of peptide N-glycosidase F (PNGase F); lanes 3 and 8, 0.25 IU of endoglycosidase H; and lanes 4 and 9, 0.83 IU of neuraminidase. Aliquots of the membrane fraction (50 g) were subjected to SDS-PAGE and immunoblot analysis with 0.5 g/ml HSV-Tag TM antibody. The filter was exposed to film for 10 s. Asterisk ‫)ء(‬ denotes an immunoreactive protein that is present in membranes from mock-transfected cells (lane 5). C, proteolytic processing of HSV-tagged SREBP-2 with inserted N-linked glycosylation sites in the lumenal loop. Aliquots of nuclear extracts (60 g of protein) and membranes (80 g) from 293 cells transfected with the indicated plasmid and incubated in the absence or presence of sterols as described in the legend to Fig. 1 were subjected to SDS-PAGE and immunoblot analysis with 0.5 g/ml HSV-Tag TM antibody. The filters for nuclear extracts and membranes were exposed to film for 5 min and 20 s, respectively. P and M denote the uncleaved precursor and cleaved NH 2 -terminal mature forms of SREBP-2, respectively. The other bands are present in mock-transfected cells and represent proteins that cross-react with the anti-HSV tag antibody. and leucine 522. The location of the arginine relative to the leucine also seems to be crucial. The recognition sequence seems to be RXXL where X can be serine, valine, or alanine, at least. The RXXL sequence is conserved in the four hamster and human SREBPs shown in Fig. 6 and also in SREBP-1c/ADD1 from the rat (RSMLE) (26) and SREBP/HLH106 from Drosophila (RRILS) (27). We believe that other features of the lumenal loop are also crucial for cleavage because moving the RXXL sequence to other sites in the lumenal loop substantially reduced cleavage (data not shown).
Although the current studies identify the site within SREBP-2 that is cleaved by the Site-1 protease, they do not reveal where in the cell this cleavage occurs. Immunofluorescence (24) and cell fractionation studies 2 have shown that SREBPs are initially found on membranes of the nuclear envelope and ER. We do not yet know whether the Site-1 protease operates in these organelles or whether the SREBPs must be transported to some other site where cleavage takes place.