A Signal Peptide Peptidase (SPP) Reporter Activity Assay Based on the Cleavage of Type II Membrane Protein Substrates Provides Further Evidence for an Inverted Orientation of the SPP Active Site Relative to Presenilin*

Signal peptide peptidase (SPP) is an intramembrane-cleaving protease identified by its cleavage of several type II membrane signal peptides after signal peptidase cleavage. Here we describe a novel, quantitative, cell-based SPP reporter assay. This assay utilizes a substrate consisting of the NH2 terminus of the ATF6 transcription factor fused to a transmembrane domain susceptible to SPP cleavage in vitro. In cells, cleavage of the substrate releases ATF6 from the membrane. This cleavage can be monitored by detection of an epitope that is unmasked in the cleaved substrate or by luciferase activity induced by the cleaved ATF6 substrate binding to and activating an ATF6 luciferase reporter construct. Using this assay we show that (i) SPP is the first aspartyl intramembrane-cleaving protease whose activity increases proportionally to its overexpression and (ii) selectivity of various SPP and γ-secretase inhibitors can be rapidly evaluated. Because this assay was designed based on data suggesting that SPP has an orientation distinct from presenilin and cleaves type II membrane proteins, we determined whether the segment of SPP located between the two presumptive catalytic aspartates was in the lumen or cytoplasm. Using site-directed mutagenesis to insert an N-linked glycosylation site we show that a portion of this region is present in the lumen. These data provide strong evidence that although the SPP and presenilin active sites have some similarities, their presumptive catalytic domains are inverted. This assay should prove useful for additional functional studies of SPP as well as evaluation of SPP and γ-secretase inhibitors.

Signal peptide peptidase (SPP) 1 is a multipass membrane protein (1)(2)(3)(4)(5)(6)(7)(8) that has been shown to carry out the intramembrane cleavage of type II membrane protein signal peptides after the initial cleavage of these by signal peptidase (Fig. 1a) (1). SPP-mediated cleavage of major histocompatibility complex class I signal peptides is thought to play an important role in normal immune surveillance by generating human leuko-cyte antigen E peptides from the signal peptide of major histocompatibility complex class I (9). These human leukocyte antigen E peptides are presented to natural killer cells at the cell surface. Such presentation is thought to indicate that the probed cell is healthy (10). Another substrate cleaved by SPP is the hepatitis C virus polyprotein (5). This cleavage appears to be essential for proper maturation of hepatitis C virus, indicating that SPP is a potential target for anti-hepatitis C virus therapy.
SPP is a member of a family of homologous multipass membrane proteins, which includes presenilin 1 (PS1) and 2 (PS2) (1,2,6). Intramembrane cleavage by members of this family is thought to be catalyzed by two aspartates present in adjacent transmembrane domains (Fig. 1a) (1,11,12). Thus, these proteins are often characterized as a family of aspartyl intramembrane-cleaving proteases (13). PS1 and PS2, which are catalytic components of the ␥-secretase complex (14,15), and SPP are the only family members to date for which a proteolytic activity has been identified. Initial studies of SPP have demonstrated that it shares some properties with the PSs. The conserved transmembrane aspartates appear to be critical for protease activity in both SPP and PSs (1,8,11), and there is evidence that both function as dimers (3, 16 -18). PSs and SPP differ in that PSs cleave type I membrane proteins, and SPP cleaves type II membrane proteins (12, 13, 19 -21). This observation has led to the hypothesis that SPP and PS may have inverted active site topologies (1). Moreover, PSs seem to require three additional proteins to function as ␥-secretase (22)(23)(24)(25)(26)(27), whereas SPP appears to function as a homodimer (1,3). ␥-Secretase activity cannot be reconstituted in non-mammalian cells without co-expression of PS1 or 2 with Aph-1, Pen-2, and Nicastrin (24,27). SPP activity can be reconstituted in yeast by expression of SPP alone (1).
SPP was originally identified as a ϳ45-kDa N-linked glycoprotein using an inhibitor-labeling approach (1). We have found that SPP exists primarily as a stable ϳ95-kDa homodimer within cells and that this dimer can be labeled by an active site-directed ␥-secretase inhibitor (3). Another study has shown that ␥-secretase inhibitors can inhibit SPP activity (4). Collectively these data suggest the active sites of SPP and PS/␥-secretase are similar. This similarity raises concerns regarding the specificity of both SPP and ␥-secretase inhibitors. Although in vitro assays for ␥-secretase have been developed, cell-based assays of ␥-secretase activity are most widely used to monitor its activity and the effects of inhibitors. To date, the vast majority of work characterizing SPP cleavage has utilized an in vitro assay (1,4,5,8). Here we describe the development of a sensitive cell-based reporter assay for SPP cleavage. This assay enables rapid evaluation of protease inhibitors and their effect on SPP and, if desired, their effects on ␥-secretase in the same cell. The assay utilizes a type II membrane-bound substrate containing the ATF6 transcription factor at its amino terminus. Upon cleavage, ATF6 is released from the membrane and activates a luciferase reporter construct (28). Using this assay we show that (i) SPP activity increases proportional to its overexpression, (ii) the parent compound of the ␥-secretase inhibitor we previously used to label the SPP homodimer is indeed an inhibitor of SPP activity, and (iii) a number of SPP and ␥-secretase inhibitors previously evaluated with in vitro SPP assays show similar effects in this cell based assay. Finally, by evaluating N-linked glycosylation mutants of SPP, which are active in this reporter assay, we validate the proposed topological model of SPP in which the loop region between the two transmembrane aspartate domains was proposed to be in the lumen.

MATERIALS AND METHODS
DNA Constructs-Mutant SPP constructs used in the glycosylation studies were constructed by PCR-based mutagenesis. Expression plasmids encoding SPP substrates (SPP sub ) were created in two steps. A cDNA encoding the fragment of the cytomegalovirus UL40 (gpUL40) reported to be necessary for SPP cleavage (5) was cloned by annealing eight overlapping oligonucleotide probes and then amplifying the fulllength product with flanking primers. The resulting DNA was then digested and cloned into the pAG3 expression vector between HindIII (5Ј) and XhoI (3Ј). A cDNA encoding the amino terminus of ATF6 was then PCR-amplified from pCGNATF6 1-373 (28) and ligated in-frame upstream of the gpUL40 sequences. All clones were verified by sequencing. The pGL3 5X ATF6 luciferase reporter construct has been previously described (28).
DNA Transfection of 293 Cells-The luciferase reporter assays were performed by transiently transfecting HEK 293T cells. HEK 293T cells were plated at 70% confluency and transiently transfected using 100 l of serum-free Opti-MEM ® (Invitrogen), 8 l of FuGENE, 0.02 g of pRL-SV40 Renilla expression plasmid (Promega), 0.25 g of pGL3 5ϫATF6 reporter plasmid, 0.25 g of pAG3 SPP sub plasmid, and the indicated amount of an SPP expression plasmid or control plasmid to total 2 g of DNA in each well of a 12-well plate. Cells were incubated with the transfection reagent for 6 -12 h, after which the serum-deficient medium was replaced with Dulbecco's modified Eagle's supplemented with 8% normal calf serum (Cambrex), 2% fetal bovine serum (Hyclone). In the experiments where inhibitors were tested, the inhibitors were added after the initial removal of the transfection reagent, and the cells were harvested 6 -8 h later by lysis with passive lysis buffer (Promega). Firefly and Renilla luciferase activities were measured using the Dual-luciferase ® kit (Promega) and a Veritas microplate Luminometer (Turner Biosystems) with Veritas 2.0.40 software package. Transfections were performed in triplicate. Results were normalized to the Renilla luciferase activity control. In some experiments, where substrate was analyzed by Western blotting, transfections were modified such that 2 g of SPP sub and 0.2 g of SPP NT FLAG plasmids were used.
Antibodies and Western Blotting-The anti-SPP NT and anti-SPP CT antibodies have been described previously and were used at a dilution of 1:1000 (3). PS1 was detected with anti-PS1 NH 2 terminus (852B3) at 1:1000. Anti-V5 (Invitrogen) and anti-FLAG (Sigma) antibodies were used at 1:1000. The anti-ATF6 antibody (Imgenex) was raised to the first 273 amino acids and recognized the NH 2 -terminal, pre-transmembrane portion of our substrate. This antibody was used at 1:250. Anti-␤-actin antibody (Sigma) was used at 1:750.
Subcellar Fractionation-HEK 293T cells were transiently transfected using a calcium phosphate transfection method (29). 45 g of SPP sub1 plasmid and 5 g of SPP NT FLAG plasmid were added to each 15-cm plate. Cells were lysed in 150 mM sodium carbonate, pH 11.0, by nitrogen cavitation for 1 h. The nuclei were then spun down at 1000 ϫ g for 10 min. The supernatant was spun for 1 h at 300,000 ϫ g, and the supernatant (S2)-containing soluble proteins and the pellet (P2)-con-FIG. 1. Design of an SPP reporter assay based on cleavage of an ATF6 containing substrate. a, a model for SPP cleavage of an ATF6 reporter substrate. SPP contains seven presumptive transmembrane domains with the NH 2 terminus in the lumen (1,3). Stars indicate the two conserved aspartates present in adjacent transmembrane domains (TMDs), which are conserved in the aspartyl intramembrane-cleaving protease family. The SPP reporter assay is based on a substrate of site-2-protease, ATF6, which is inserted into the membrane in a type II orientation. The NH 2 -terminal cytoplasmic domain of ATF6 was fused to the NH 2 terminus of a fragment of cytomegalovirus gpUL40, shown to be cleavable by SPP in an in vitro assay system (5). After cleavage, ATF6 could translocate to the nuclease, where it would bind a 5ϫATF6 DNA binding region coupled to the luciferase gene promoter. b, three potential SPP substrates (SPP sub ) were designed. Each substrate contains amino acids 1-373 of ATF6 at its NH 2 terminus fused to a fragment of gpUL40 that includes a transmembrane signal peptide sequence and a V5 His tag. SPP sub2 differs from SPP sub1 in that it contains an additional 15 amino acids at its COOH terminus that encodes a signal peptidase cleavage site. SPP sub3 differs from SPP sub1 in that it contains the naturally occurring RIR sequence immediately after the transmembrane domain. c, Western blot analysis of SPP sub1-3 expression. SPP sub1-3 are expressed transiently as ϳ50-kDa proteins in HEK 293T. Bands are immunoreactive for anti-ATF6 (shown) and anti-V5 (not shown). wt, wild type.
taining integral membrane proteins were collected and analyzed by SDS-PAGE.
Glycosylation Experiments-HEK 293T cells were transiently transfected with either plasmids encoding SPP CT V5his or glycosylation mutant and FuGENE as described above. Cells were lysed after 48 h in 300 l of 1% Triton X-100 in TBS and 1ϫ complete protease inhibitor (Roche Applied Science). 20-l lysate samples were treated with or without peptide N-glycosidase F (New England Biolabs) according to the manufacturer's instructions. Samples were then analyzed by SDS-PAGE.
Data Analysis-Data were analyzed using Sigma Stat. For comparison of multiple experimental values relative to controls an analysis of variance was performed using a Dunnet's post hoc t test. Variance is reported as the S.E.

RESULTS
SPP Reporter Assay Development-Potential SPP substrates were created by fusing the NH 2 terminus of human ATF6 (amino acids 1-373) with a region of cytomegalovirus gpUL40 ( Fig. 1, a and b). The gpUL40 sequences include the transmembrane domain and flanking residues. A contiguous V5 and six-histidine epitope tag was inserted within the cytoplasmic region of gpUL40. Three substrates that differ at their COOH termini were generated. The first, SPP sub1 contains two point mutations in the gpUL40 (T-I-T) that were shown to be neces-sary for cleavage by SPP in vitro (5). The second, SPP sub2 , is identical to the first except it has a 15-residue extension at its COOH terminus that includes a signal peptidase cleavage site. The third, SPP sub3 , uses the wild type gpUL40 (RIR) sequence instead of the mutant T-I-T sequence. This substrate was constructed because the wild type gpUL40 was refractory to SPP cleavage in vitro (Fig. 1b) (5). After transient transfection each substrate can be detected just above the ϳ50-kDa marker by Western blotting with an antibody to the NH 2 terminus of ATF6 (Fig. 1c) or anti-V5 antibody (not shown).
SPP sub Activity Increases with Increasing SPP Concentration-To determine whether SPP sub1 could be cleaved by SPP and activate the 5ϫATF6 luciferase reporter construct (Fig.  1a), we examined whether increased expression of SPP would increase the luciferase expression. For these studies, we transiently transfected 293T cells with a constant amount of the SPP sub1 , 5ϫATF6 luciferase reporter, and Renilla expression constructs. Varying amounts of empty vector or a SPP NT FLAG expression vector were co-transfected such that the total amount of DNA in each transfection was constant. Luciferase activity was normalized to Renilla activity to control for transfection efficiency, and activity was expressed as the percent of FIG. 2. SPP luciferase reporter assay activity increase in a SPP concentration-dependent fashion. SPP NT FLAG expression plasmid was cotransfected into HEK 293T cells at the concentrations reported with 0.25 g of reporter, 0.25 g of SPP sub1 , and 0.02 g of Renilla plasmids per well as described under "Materials and Methods." a, SPP levels increase proportionally with increasing levels of SPP NT FLAG expression plasmid. The SPP dimer (ϳ95 kDa) and monomer (ϳ45 kDa) are detected with ant-FLAG, which detects the transfected SPP NT FLAG, and anti-SPP CT , which detects both endogenous SPP and SPP NT FLAG. b, SPP luciferase activity increases with increasing SPP protein expression. Normalized reporter activity expressed as % of the activity present in cells expressing endogenous levels of SPP (% Endogenous) is plotted versus SPP protein levels expressed as % of endogenous levels of SPP dimer detected by anti-SPP CT . The fitted line corresponds to best fit of the data to a rectangular hyperbola. c and d, the amount of SPP sub1 detectable with anti-V5 increases with increasing SPP NT FLAG expression. Transient transfection of cells with SPP sub1 and SPP NT FLAG expression plasmids results in a constant level of SPP sub1 detected by the anti-ATF6. However, as SPP NT FLAG expression increases, the amount of SPP sub1 detected by anti-V5 increases. In d, the amount of anti-V5 detectable SPP sub1 is plotted versus the amount of SPP NT FLAG DNA transfected, demonstrating the increase in the anti-V5 signal with increasing SPP NT FLAG levels. e, cleavage of SPP sub1 releases the amino terminus from the membrane and unmasks the V5 epitope. The majority of anti-V5 detectable SPP sub1 is in the soluble S2 supernatant. Conversely, the anti-ATF6-positive SPP sub1 bands are relatively equivalent in both the soluble supernatant fraction (S2) and the membrane pellet (P2). Membrane proteins such as SPP and PS1 are detected primarily in the P2 membrane pellet. normalized luciferase activity relative to the amount of activity seen in cells expressing endogenous levels of SPP (SPP activity % endogenous). Transfection of increasing amounts of the SP-P NT FLAG expression vector resulted in a marked increase in ϳ95-kDa SPP dimer up to 100-fold higher than the level of endogenous SPP (Fig. 2a). A small amount of monomer at ϳ45 kDa was observed at higher levels of SPP expression. Because SPP dimer levels increased at low SPP expression levels, a corresponding linear increase in luciferase activity (expressed as percent of endogenous SPP levels) was also observed (Fig.  2b). At the highest SPP dimer expression level, there was a clear deviation from linearity. The fitted line corresponds to best fit of the data to a rectangular hyperbola, which appears linear at the points corresponding to low SPP overexpression levels, consistent with the saturation behavior of in vivo enzyme systems. Because only the relative SPP levels and activity were plotted, the fitted values do not have physical meaning and were, therefore, not included. The effects of increasing SPP dimer expression on the SPP sub1 were directly examined by Western blotting with anti-V5 and anti-ATF6 antibodies (Fig.  1c). The level of SPP sub1 detected by anti-ATF6 did not change significantly with increasing amounts of SPP expression. However, as SPP levels and luciferase activity increased, the amount of SPP sub1 detected by anti-V5 increased dramatically (Fig. 2, c and d). Based on this observation we postulated that the V5 epitope was inaccessible in the intact substrate but unmasked by SPP cleavage. As SPP cleavage would release the cytoplasmic, NH 2 -terminal domain of SPP sub1 from the membrane, we examined whether the anti-V5 immunoreactive material was soluble or membrane-bound. Anti-V5 immunoreactive SPP sub1 was almost exclusively present in the soluble (S2) fraction, whereas the ATF6-immunoreactive SPP sub1 was equally distributed between the soluble (S2) fraction and the integral membrane protein (P2) fraction. PS1 and SPP, which are integral membrane proteins, were almost exclusively localized in the P2 fraction. Thus, it appears that release of SPP sub1

FIG. 2-continued
from the membrane unmasks the V5 epitope, indicating that the unmasking of the V5 epitope is a good indicator for cleavage and solubilization of SPP sub1 . SPP sub1 Is Not Cleaved by PS1-To further demonstrate that the SPP-dependent increase in reporter activity and unmasking of the V5 epitope of SPP sub1 was a specific and sensitive indicator of SPP activity, we performed a number of additional studies. First, we tested for specificity by determining whether PS1, another aspartyl intramembrane-cleaving protease, increased luciferase activity in this assay. Neither wild type PS1 nor a FAD-linked PS1 mutant (M139V) altered luciferase activity (Fig. 3a) despite increased PS1 expression levels (Fig.  3b). In addition to the SPP NT FLAG construct, we also exam-ined two additional SPP constructs, a wild type SPP expression construct and a V5 His COOH-terminal-tagged SPP construct (SPP CT V5his). All three constructs resulted in significant increases in luciferase activity (Fig. 3a). Whereas the level of activity relative to protein levels was similar for the wild type SPP and SPP NT FLAG, SPP CT V5his expression resulted in a high level of activity compared with SPP protein levels (Fig.  3b). The relative increase in activity seen with SPP CT V5his could be attributable to disruption of a putative ER retrieval signal (KKXX) by the addition of the epitope tag at the COOH terminus. However, to date our studies on localization of this construct have not revealed major differences in its subcellular distribution relative to the wild type construct or the NH 2 - FIG. 3. Assay validation. a, three different constructs of SPP increase SPP reporter assay activity to 5-10-fold above endogenous SPP activity. Neither PS1 nor PS1 FAD mutant, M139V, show any increase in SPP reporter assay activity. *, p Ͻ 0.05 analysis of variance. b, SPP overexpression was confirmed by SDS-PAGE analyses, which were probed by both anti-SPP NT and anti-SPP CT antibodies, and in the case of the tagged SPP constructs anti-FLAG and anti-V5, antibodies were used as indicated. wt, wild type. PS1 Western blot analysis was performed using anti-PS1 NTF (852B3). c, SPP inhibitor (Z-LL) 2 ketone and two other ␥-secretase inhibitors (III-31-C and LY411,575) inhibit SPP reporter assay activity when SPP NT FLAG is transiently overexpressed. As seen in the in vitro assay system (1), compound E and DAPT do not inhibit SPP reporter activity. d and e, the amount of SPP sub1 detectable with anti-V5 decreases with increasing SPP inhibitors (Z-LL) 2 ketone and LY411,575. However, a relatively constant amount of anti-ATF6 detectable SPP sub1 is detected in each experiment. Quantitation of the amount of anti-V5 detectable SPP sub1 shows the decrease in the anti-V5 signal as a result of inhibition by (Z-LL) 2 ketone or LY411,575. f, each of the SPP sub is active in the presence of endogenous SPP (1), and a significant increase in reporter activity is observed with each when SPP NT FLAG is overexpressed (2). Additionally, the activity of each substrate is inhibitable by LY411,575 and (Z-LL) 2 ketone (3). SPP activity for each sample is plotted as the % greater than the background activity or "% control." terminal FLAG-tagged construct (3) (data not shown).
Effects of SPP and ␥-Secretase Inhibitors-Having demonstrated that the SPP sub1 reporter luciferase activity is proportional to SPP expression but not altered by two other family members, we evaluated several protease inhibitors for their effect on reporter activity. Inhibition of activity was assessed in cells expressing endogenous SPP and cells transiently overexpressing SPP NT FLAG (Fig. 3, c and d). Similar results were seen for both conditions. An SPP selective inhibitor (Z-LL) 2 ketone and a ␥-secretase inhibitor, LY411,575, previously shown to inhibit SPP (4) inhibited reporter activity with IC 50 values of ϳ0.14 M for (Z-LL) 2 ketone and ϳ0.73 M for LY411,575 (Fig. 3c). III-31-C, a transition state ␥-secretase inhibitor (30), inhibited reporter activity with an IC 50 of ϳ3 M. This compound is the parent compound to III-63, which we have shown binds the SPP homodimer (3). Consistent with studies using in vitro SPP cleavage assays (4), two other ␥-secretase inhibitors, DAPT and Compound E, did not inhibit reporter activity (Fig. 3c). The effects of (Z-LL) 2 ketone and LY411,575 on SPP sub1 were also assessed by Western blotting (Fig. 3d). Both compounds significantly inhibited the unmasking of the V5 epitope, providing further evidence that the V5 epitope is unmasked by SPP cleavage. In numerous experiments, cells either expressing endogenous SPP or transiently overexpressing SPP NT FLAG, the maximal inhibition observed of luciferase activity, was ϳ80% of the uninhibited activity.

Substrates with a Signal Peptidase Cleavage Site or Positively Charged Residues Flanking the Hydrophobic Domain Are
Efficiently Cleaved by SPP in Cells-We examined cleavage of two variant substrates in the reporter assays (Fig. 3f). SPP sub2 , which includes a signal peptidase cleavage site, was an efficiently cleaved substrate. SPP sub2 reporter activity increased upon transient overexpression of SPP, and this activity was inhibited to a similar extent as SPP sub1 by the SPP inhibitors (Z-LL) 2 ketone and LY411,575. Somewhat unexpectedly, the SPP sub3 , which contains the naturally occurring RIR amino acid sequence in the COOH terminus that prevented cleavage by SPP in an in vitro assay (5), also appears to be efficiently cleaved in vivo. SPP sub3 reporter activity is increased by SPP overexpression and is inhibitable by SPP inhibitors.
An N-linked Glycosylation Site SPP Mutant Is Active and Reveals That SPP Has a Topology Consistent with Cleavage of Type II Membrane Proteins-It has been postulated that the orientation of the opposing transmembrane domains containing the two presumptive catalytic aspartates is inverted in SPP relative to PS. The evidence for this is indirect and comes from two observations. First, the identified PS substrates are type I membrane proteins, whereas identified SPP substrates including the ATF6 fusion protein substrates used in this study are type II membrane proteins. Second, studies of N-linked glycosylation sites place the NH 2 terminus of SPP in the lumen (1). This finding coupled with a seven transmembrane domain FIG. 3-continued model suggested that the transmembrane domains containing the critical aspartates were oriented such that the intervening sequence would be present in the lumen (Fig. 4a) and not in the cytoplasm. To experimentally determine whether the loop between the catalytic aspartates was luminal or cytoplasmic, we generated a number of SPP mutants designed to delete endogenous or add exogenous N-linked glycosylation sites. The mutants used are shown in Fig. 4a and are numbered 1-5. Mutant 1 (N10S and N20S) is not glycosylated (Fig. 4, b and c) and confirms the luminal location of the NH 2 terminus of SPP (1). The lack of glycosylation of this mutant also demonstrates that two additional endogenous N-linked glycosylation sites, one in the NH 2 terminus and one in the loop region immediately preceding the first aspartate (TMD4), are not utilized (Fig. 4a). Four additional mutants (2)(3)(4)(5) were then generated on the unglycosylated mutant 1 background (Fig. 4a). Mutants 2 and 5 migrated identically to mutant 1, whereas mutants 3 and 4 migrated at a higher molecular weight (Fig. 4b), suggesting that they were glycosylated. There were consistent differences in the extent of glycosylation of mutants 3 and 4. The majority of SPP mutant 3 appeared to be glycosylated, whereas only a fraction of mutant 4 was glycosylated. To confirm that the shift in M r was due to glycosylation, cell lysates were treated with peptide N-glycosidase F before electrophoresis and Western blotting. Both SPP CT V5his and mutant 3 migrate at an identical M r to mutant 1 after peptide N-glycosidase F treatment (Fig. 4c). Because the SPP dimer migrates over a broader range, we focused these studies on the monomeric forms. However, all mutants did form dimers and showed changes in migration consistent with the alterations in glycosylation inferred from the study of monomer (Fig. 4b). The mutants were then evaluated in our reporter assay to ensure that they were active. As shown in Fig. 4d, all of these mutants are active, significantly increasing luciferase activity relative to cells expressing endogenous SPP. DISCUSSION We have developed a cell-based reporter assay for SPP utilizing the NH 2 -terminal 373 amino acids of ATF6 fused to the SPP-cleavable transmembrane domain of gpUL40. The SPP reporter activity assay provides an excellent read-out of SPP activity. Luciferase activity increases proportional to SPP dimer overexpression and is inhibited by protease inhibitors previously shown to inhibit SPP activity. Two of the inhibitors  N20S, D280N). A shift in the M r of both the dimer and the monomer, presumably due to glycosylation, is observed in mutants 3 (N10S, N20S, 242N) and 4 (N10S, N20S, 252N). Exposure times for each mutant were optimized to provide the best image and do not reflect the relative level of expression. Expression of mutants 4 and 5 were consistently low. c, deglycosylation by peptide N-glycosidase F treatment confirms that SPP CT V5his and mutant 3 are glycosylated, and mutant 1, 2 and 5 are not (data shown only for mutant 1 and 3). d, glycosylation site SPP mutants are functional. SPP activity was normalized to SPP dimer expression levels and plotted as % endogenous luciferase activity. *, p Ͻ 0.05. that decrease reporter activity, III-31-C and (Z-LL) 2 ketone, have been previously shown to effectively displace the binding of III-63, a photoaffinity probe that binds the SPP dimer, indicating that these inhibitors directly target the SPP dimer (3). Thus, these data provide additional evidence that the SPP dimer is the active form of SPP in cells. Overexpression of PS1, another member of the aspartyl intramembrane-cleaving protease family, did not increase reporter activity. Thus, under conditions where SPP is overexpressed, we can conclude that the majority of reporter activity is attributable to SPP cleavage of the substrate. However, it is possible that other uncharacterized members of the PS/SPP family could exhibit activity in this assay.
The studies herein further reinforce the differences between PSs and SPP. PSs are the presumptive catalytic component of a multiprotein complex that requires at least three additional factors, Aph-1, Pen-2, and Nicastrin (26). Reconstitution of ␥-secretase activity in yeast requires all four components (24,27). Although small increases in ␥-secretase activity can sometimes be observed upon overexpression of individual components of the ␥-secretase complex, large increases in activity in mammalian cells are not observed unless all four components are overexpressed (22-27, 31, 32). In contrast, no co-factors are necessary to reconstitute SPP activity in yeast, and (1) as shown herein, SPP activity increases proportionally over a wide range of SPP dimer levels in cells.
It has been proposed that topological differences between PS and SPP account for their respective preference for cleaving type I membrane proteins and type II membrane proteins (1,13,19). Although the precise topology of PS is still debated, there is a general consensus that the loop between the critical aspartate residues has a cytoplasmic localization. Here we provide evidence based on glycosylation site mutants of SPP that the corresponding region of SPP is in fact luminal. Moreover, we show that these SPP mutants are functionally active. Thus, we provide strong evidence that, as hypothesized, the active sites of SPP and PS are inverted within the membrane. Clearly, additional structural studies will be required to more precisely characterize the topology of SPP.
Although a number of results obtained with our assay correlate well with previous observations using in vitro assays to characterize SPP cleavage, we find that the substrate, SPP sub3 , which contains a sequence resistant to SPP cleavage in vitro (5), is cleaved in cells by SPP. Such data suggest that the requirements for in vivo and in vitro cleavage may be distinct. Additional studies cross-comparing cleavage of various substrates will be needed to resolve such discrepancies.
Although the precise SPP cleavage site or sites have not been determined for any substrate, it is clear that SPP cleavage releases the cytoplasmic domain of the target protein from the membrane. In this reporter assay, cleavage of the various SPP substrates releases the ATF6 transcription factor so that it can translocate to the nucleus and activate the luciferase reporter construct. We find that an internal V5 epitope tag appears to be masked in the intact substrate, but once cleaved and solubilized, the V5 tag becomes accessible. The cleaved and uncleaved SPP sub1 are predicted to differ by 11 amino acids, making it difficult to differentiate them after electrophoresis. Thus, the unmasking of the V5 epitope associated with increased cleavage and solubilization of substrate appears to be an excellent surrogate indicator for substrate cleavage. Because this assay utilizes a reporter construct that could be activated by induction of the unfolded protein response with cleavage of endogenous ATF6 by site-one and site-two proteases, both methods for assessing SPP sub cleavage should be utilized to ensure that substrate proteolysis is being monitored.
Both SPP and PS are potential therapeutic targets for human diseases. Preclinical studies in mouse models suggest that if toxicity can be minimized, inhibition of A␤ production by ␥-secretase inhibitors may be an effective therapy for Alzheimer's disease (33)(34)(35)(36). To date at least two ␥-secretase inhibitors have entered initial phase I trials for Alzheimer's disease. Because of their role in altering cell-signaling cascades, ␥-secretase inhibitors are also being explored as immune modulatory agents and as novel anti-cancer agents. In contrast to ␥-secretase, SPP is a theoretical therapeutic target. Based on its known substrates, major histocompatibility complex class I, and the hepatitis C viral polyproteins, SPP inhibitors could have therapeutic utility as antiviral agents or immune modulatory agents. It is clear from this and other published studies that some ␥-secretase inhibitors inhibit SPP (4). Clear separation of inhibitory activities would be desirable for both SPP and ␥-secretase inhibitors that are intended for clinical use. The cell-based reporter assay we have developed should prove useful in the evaluation and development of both ␥-secretase and SPP selective inhibitors.