Gamma-secretase activity is associated with a conformational change of nicastrin.

Gamma-secretase is a high molecular weight multicomponent protein complex with an unusual intramembrane-cleaving aspartyl protease activity. Gamma-secretase is intimately associated with Alzheimer disease because it catalyzes the proteolytic cleavage, which leads to the liberation of amyloid beta-peptide. At least presenilin (PS), Nicastrin (Nct), APH-1, and PEN-2 are constituents of the gamma-secretase complex, with PS apparently providing the active site of gamma-secretase. Expression of gamma-secretase complex components is tightly regulated, however little is known about the assembly of the complex. Here we demonstrate that Nct undergoes a major conformational change during the assembly of the gamma-secretase complex. The conformational change is directly associated with gamma-secretase function and involves the entire Nct ectodomain. Loss of function mutations generated by deletions failed to undergo the conformational change. Furthermore, the conformational alteration did not occur in the absence of PS. Our data thus suggest that gamma-secretase function critically depends on the structural "activation" of Nct.

␥-Secretase is a high molecular weight multicomponent protein complex with an unusual intramembranecleaving aspartyl protease activity. ␥-Secretase is intimately associated with Alzheimer disease because it catalyzes the proteolytic cleavage, which leads to the liberation of amyloid ␤-peptide. At least presenilin (PS), Nicastrin (Nct), APH-1, and PEN-2 are constituents of the ␥-secretase complex, with PS apparently providing the active site of ␥-secretase. Expression of ␥-secretase complex components is tightly regulated, however little is known about the assembly of the complex. Here we demonstrate that Nct undergoes a major conformational change during the assembly of the ␥-secretase complex. The conformational change is directly associated with ␥-secretase function and involves the entire Nct ectodomain. Loss of function mutations generated by deletions failed to undergo the conformational change. Furthermore, the conformational alteration did not occur in the absence of PS. Our data thus suggest that ␥-secretase function critically depends on the structural "activation" of Nct.
␥-Secretase plays a fundamental role in Alzheimer disease (AD) 1 by catalyzing the final proteolytic cleavage, which leads to the formation of amyloid-␤ peptide (A␤), the major component of the diseases defining senile plaques (1). By genetic and biochemical approaches several components of the ␥-secretase complex have been identified. In addition to the presenilins (PS1 and PS2) (reviewed in Ref. 1), APH-1a/b, PEN-2, and nicastrin (Nct) (2)(3)(4) were recently identified. Apparently all four proteins assemble into a large 500 -600-kDa complex (5)(6)(7)(8)(9), which displays the intramembranous proteolytic activity required for the cleavage of the ␤-amyloid precursor protein (APP) and other substrates such as Notch (for review see Ref. 1). Formation of the ␥-secretase complex is coordinately regulated (2, 6 -13) and depends on the presence of all known complex components. Although there is considerable evidence that PS constitutes the active site of ␥-secretase (reviewed in Ref. 1), very little is known about the function of the individual PS binding partners. Previously we and others demonstrated that maturation of Nct is associated with ␥-secretase complex assembly (6,(11)(12)(13). In addition, a conserved DYIGS motif is apparently involved in Nct function (3). Here we demonstrate that a major conformational change, which requires that the entire ectodomain of Nct is directly associated with ␥-secretase complex formation and function. The structural alteration fails to occur in Nct loss of function mutations as well as in the absence of presenilins.

EXPERIMENTAL PROCEDURES
cDNA Constructs-To down-regulate endogenous Nct by RNA interference (RNAi), oligonucleotides corresponding to Nct-1045 (6) were cloned into the pSUPER vector (14). Nct deletions (Del 1-5, Fig. 1a) were constructed by oligonucleotide-directed mutagenesis using PCR. Silencer mutations (aaagggaaattcccggtccaatt; the mutations are underlined) were introduced (which do not affect the amino acid sequence) in the constructs to escape RNAi. All constructs were verified by DNA sequencing.
Cell Culture, Cell Lines, RNAi, and Transfections-Human embryonic kidney (HEK) 293 cells and mouse embryonic fibroblast cells were cultured as described (6). A stable Nct knock-down cell line was generated by stably co-transfecting HEK 293 cells overexpressing Swedish mutant APP (15) with pSUPER/Nct-1045 and pcDNA3.1/Hygro(Ϫ) (Invitrogen) and selection for hygromycin (100 g/ml) resistance. This cell line was stably transfected with the indicated wt and mutant Nct constructs or the empty vector (pcDNA6) by Lipofectamine 2000 (Invitrogen) according to the instructions of the manufacturer using selection for blasticidin (10 g/ml) resistance. To inhibit mannosidase I, cells were cultured in the presence of the indicated amounts of kifunensine (Calbiochem) or vehicle for 48 h at 37°C.
Trypsin Resistance Assay-Cells were lysed as detailed above in the presence of 0.7% DDM or 1% SDS. Following a clarifying spin, cell lysates were incubated with the indicated amounts of trypsin in 150 mM * This work was supported by the Deutsche Forschungsgemeinschaft (Priority Program "Cellular Mechanisms of Alzheimer's Disease") and the European Community (DIADEM Project). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sodium citrate (pH 6.4), 150 mM NaCl, 5 mM EDTA, 5 g/ml pepstatin for 30 min at 30°C. Proteolysis was stopped by the addition of 10-fold excess amounts of soybean trypsin inhibitor, and samples were subjected to immunoblot analysis.

RESULTS
In an attempt to identify the functionally important domains of Nct, we generated a set of deletions within the ectodomain (Fig.  1a). These cDNA constructs were investigated in a HEK 293 cell line stably expressing Swedish mutant APP (15) and a pSUPERbased Nct-1045 (6) small interfering RNA-encoding vector, which stably knocks down endogenous Nct expression by RNAi ( Fig. 1b; lane 2). RNAi-mediated inhibition of Nct expression results in reduced PS1 CTF formation, reduced PEN-2 and APH-1aL (8) expression, the accumulation of the APP C-terminal fragments (APP-CTFs), and reduced A␤ generation (Fig. 1b). These observations are due to the inhibition of the ␥-secretase activity upon down-regulation of Nct (6). Expression of a wt Nct cDNA with a cluster of silent mutations conferring RNAi resistance led to the formation of mature Nct (Fig. 1b), which has previously been shown to be associated with the functional ␥-secretase complex (6,(11)(12)(13). In addition, an accumulation of large amounts of immature Nct due to its overexpression (6,12,17) was observed (Fig. 1b). In contrast, all deletion constructs apparently formed only one Nct polypeptide (Fig. 1b), indicating a failure of maturation. To investigate if the Nct deletion variants undergo complex glycosylation like wt Nct, cell lysates were treated with endo H. As shown in Fig. 1c, only mature Nct (endogenous and exog-enous) was endo H-resistant, whereas immature Nct and all deleted variants failed to become endo H-resistant. Exogenous expression of wt Nct restored PS1 CTF formation and PEN-2 and APH-1aL expression and allowed full ␥-secretase function as monitored by the significantly reduced levels of APP-CTFs accompanied by robust A␤ generation (Fig. 1b). In contrast to wt Nct, none of the deletion constructs restored PS1 CTF formation and PEN-2 or APH-1aL expression (Fig. 1b). Moreover, the deletion constructs did not allow the formation of a ␥-secretase activity, because none of them reduced APP-CTF formation or increased A␤ production (Fig. 1b). Thus, all deletions within the ectodomain failed to restore ␥-secretase function. This suggests an important role of not only the conserved DYIGS motif but the entire ectodomain in ␥-secretase complex assembly and activity. The lack of a specific functional subdomain of Nct thus indicates that correct folding of the entire ectodomain is required for Nct function. The primary structure of Nct suggests a rather large luminal domain, which according to our findings plays a pivotal role in Nct function. To investigate if the luminal domain of functional Nct adopts a conformation, which is different from non-functional Nct, cell lysates were treated with increasing amounts of trypsin to monitor unmasking or masking of cleavage sites (18). Interestingly, the mature form of Nct, which is predominantly found in the mature ␥-secretase complex (6,12,13,17), was selectively trypsin-resistant whereas immature Nct remained trypsin-sensitive even at the lowest concentration (Fig.  2a). Mature Nct showed resistance up to concentrations of as much as 500 g/ml trypsin (Fig. 2a and data not shown). In contrast to mature Nct, APP, which is also a type I transmembrane glycoprotein, was sensitive to trypsin (Fig. 2a). Further- and Nct-1045 small interfering RNA were stably transfected with the indicated cDNA constructs encoding wt Nct, Nct ectodomain deletion mutants (harboring silent mutations to escape RNAi; note that Del 3 escapes RNAi due to deletion of the RNAi-targeted region) or a vector control. Cell lysates were analyzed for levels of Nct (mature (m) and immature (im) forms), and PS1 CTF and APP-CTFs (generated by ␤-secretase (CTF␤) and ␣-secretase (CTF␣)) by immunoblotting with antibodies N1660 (Nct), 3027 (PS1), and 6687 (APP). PEN-2 and APH-1aL levels were analyzed from membrane fractions of the same cells by immunoblotting with antibodies 1638 (PEN-2) and O2C2 (APH-1aL). A␤ was analyzed from conditioned media by combined immunoprecipitation/immunoblotting with antibodies 3926/6E10. c, Nct ectodomain deletion mutants are endo H-sensitive. Cell lysates were incubated with (ϩ) or without (Ϫ) endo H and analyzed for Nct as in B. Fig. 1b) were treated with the indicated amounts of trypsin and analyzed for Nct, APP, PS1 (CTF and NTF), and APH-1aL as in Fig. 1b. The PS1 NTF was analyzed with antibody PS1N. Note that mature Nct is resistant to trypsin whereas immature Nct and mature and immature forms of APP are degraded even at the lowest concentration of trypsin. The polypeptide migrating at 85 kDa is an intermediate degradation product of immature Nct. Other ␥-secretase complex components such as the PS1 NTF and CTF and APH-1aL were fully sensitive to trypsin. b, all Nct deletion mutants are sensitive to trypsin. CHAPS-extracted HEK 293 cells stably transfected with wt Nct and the indicated Nct deletion mutants (as detailed in Fig. 1b) were incubated with (ϩ) or without (Ϫ) 100 g/ml trypsin and analyzed for Nct as in Fig. 1b. c, Nct not associated with the ␥-secretase complex is trypsin-sensitive whereas mature Nct assembled into the ␥-secretase complex is resistant. Cell lysates of PS1/2 ϩ/ϩ or PS1/2 Ϫ/Ϫ mouse embryonic fibroblast cells were subjected to trypsin treatment as in b and analyzed for Nct as in Fig. 1b. Consistent with our previous results (7) immature Nct accumulates in the PS1/2 Ϫ/Ϫ cells, whereas both mature and immature Nct is detected in PS1/2 ϩ/ϩ control cells. Mature Nct in PS1/2 ϩ/ϩ control cells is trypsinresistant whereas immature Nct in PS1/2 Ϫ/Ϫ cells is trypsin-sensitive. more, the ␥-secretase complex components PS1 NTF, PS1 CTF, and APH-1aL were all fully sensitive to trypsin digestion (Fig.  2a), whereas PEN-2 was found to be less sensitive (data not shown). Because APH-1aL and the PS fragments are trypsinsensitive, Nct is not simply protected by these ␥-secretase complex components. In addition, the very small PEN-2 is unlikely to protect the large Nct ectodomain. Thus, Nct appears to undergo a conformational change independent of APH-1aL, PS, and also PEN-2. After demonstrating the selective trypsin resistance of mature Nct, the deletion variants (Fig. 1a), which all fail to restore ␥-secretase activity (Fig. 1b), were investigated. Interestingly, none of them displayed trypsin resistance (Fig. 2b). This suggests that assembly of a biologically active ␥-secretase complex is associated with the formation of a trypsin-resistant Nct variant. To further support this hypothesis, we analyzed Nct in mouse embryonic fibroblast cells derived from a PS1/2 gene knock-out. Because of the absence of PS in these cells no ␥-secretase complex can be formed. As we and others have previously shown these cells are also deficient in Nct maturation (7,11,13). Thus, fibroblasts derived from a PS1/2 gene knock-out are ideally suited to investigate the association of trypsin-resistant Nct with ␥-secretase complex formation. Interestingly, immature Nct in PS1/2 Ϫ/Ϫ cells was degraded by trypsin, whereas mature Nct in the corresponding PS1/2 ϩ/ϩ control cells was fully trypsinresistant (Fig. 2c). Thus, the conversion of trypsin-sensitive to a trypsin-resistant Nct is indeed tightly associated with ␥-secretase complex formation. Furthermore, the selectivity of trypsin resistance of mature Nct versus immature/non-functional Nct suggests a major conformational change of Nct during ␥-secretase complex assembly and maturation. However, the selective resistance of mature Nct does not exclude the possibility that proteases could not interact with mature Nct due to the rather large and abundant sugar side chains added during maturation. Indeed, 16 putative glycosylation sites are present in the ectodomain (3). To denature and unfold mature Nct, cells were lysed in the presence of 1% SDS, and lysates were then digested with increasing amounts of trypsin. Under these conditions mature Nct became sensitive to trypsin digestion, whereas nondenatured mature Nct extracted under conditions which preserve the ␥-secretase complex remained protease-resistant (Fig.  3a). However, glycosylation could protect even partially denatured mature Nct and thus indirectly prevent trypsin-mediated degradation. To exclude this possibility we blocked complex glycosylation by incubating untransfected HEK 293 cells (expressing endogenous Nct) in the presence of kifunensine, which potently inhibits mannosidase I (19). As shown in Fig. 3b, treatment with kifunensine strongly blocked maturation of Nct as manifested by the appearance of a novel Nct species (termed immature-like Nct, see below) migrating at lower molecular weight. However, in contrast to the immature form of Nct, the immature-like species observed upon kifunensine treatment was still trypsin-resistant like the mature fully glycosylated Nct variant (Fig. 3c). These data suggest that a conformational change of Nct associated with trypsin resistance must take place upon assembly and/or maturation of the ␥-secretase complex. To investigate if the ␥-secretase complex is still active upon inhibition of mannosidase I, A␤ was isolated before and after kifunensine treatment. Consistent with Herreman et al. (13), A␤ production was not inhibited by kifunensine (Fig. 3d). Moreover, expression levels of PS1 CTFs and PEN-2 were not significantly reduced by kifunensine treatment (Fig. 3e, left panel) demonstrating that kifunensine does not interfere with the assembly of the ␥-secretase complex. Furthermore, immature-like Nct and PEN-2 coimmunoprecipitated with PS1 upon kifunensine treatment for 2 days (Fig. 3e, right panel). Finally, cell surface biotinylation re-vealed that immature-like Nct reaches the plasma membrane in cells treated with kifunensine ( Fig. 3f) like endogenous Nct in untreated cells (13,16). DISCUSSION Our findings demonstrate that trypsin resistance of the Nct ectodomain is associated with ␥-secretase complex assembly, maturation, and activity. Thus, we conclude that ␥-secretase activity requires a conformational alteration of Nct. Immature and all functionally inactive deletion mutations fail to undergo the conformational switch required for ␥-secretase activity and remain trypsin-sensitive. Complex glycosylation does not protect by itself against proteolytic degradation, because its inhibition by kifunensine does not affect the protease resistance and function of Nct. In addition, binding of Nct to other ␥-secretase complex components does not protect from trypsin degradation, because APH-1aL and the PS1 NTF and CTF are all sensitive to trypsin as well, whereas mature Nct is selectively resistant. Thus, non-functional Nct is structurally "activated" by a conformational alteration. The conformational alteration may be similar to that of the sterol regulatory element-binding protein-activating protein (SCAP) (18). In the latter case cholesterol addition leads to a conformational change of SCAP, FIG. 3. A conformational change of Nct is associated with its resistance to trypsin degradation and ␥-secretase complex assembly. a, SDS unfolds Nct and makes it sensitive to trypsin. HEK 293 cells stably transfected with wt Nct (as detailed in Fig. 1b) were extracted with DDM (which leaves the ␥-secretase complex intact (6, 7)) or SDS, subjected to trypsin treatment and analyzed for Nct as in Fig. 1b. b, inhibition of Nct maturation by blocking mannosidase I does not affect ␥-secretase complex formation/activity. HEK 293 cells were incubated in the presence of the indicated amounts of kifunensine, and lysates were analyzed as in a. Note that treatment with kifunensine results in the formation of a Nct species, which co-migrates with immature Nct (immature-like (iml) Nct). c, immature-like Nct generated by kifunensine treatment is trypsinresistant. Lysates from kifunensine-treated HEK 293 cells were incubated with 20 g/ml trypsin and analyzed as in a. d, generation of secreted A␤ upon kifunensine treatment. Conditioned media of HEK 293 cells pretreated with kifunensine were collected, and A␤ production was analyzed as in Fig. 1b. e, immature-like Nct generated by kifunensine treatment forms a complex with PS1 and PEN-2. DDM-extracted membrane fractions of HEK 293 cells were immunoprecipitated with antibody 3027 (PS1-C) and analyzed by immunoblotting as in Fig. 1b, except that PS1 CTF was analyzed using antibody BI.3D7. Direct immunoblotting (left panel) confirmed that the expression of PS1 and PEN-2 is not affected by kifunensine treatment. Moreover, kifunensine treatment does not result in accumulation of APP CTFs as observed upon inhibition of Nct expression. f, immature-like Nct generated by kifunensine treatment is transported to the plasma membrane. Kifunensine-treated HEK 293 cells were surface-biotinylated. After streptavidin precipitation, biotinylated Nct was identified by immunoblotting as in Fig. 1b. Note that without kifunensine exclusively mature Nct is biotinylated, whereas after kifunensine the immature-like Nct is preferentially surface-biotinylated. which unmasks additional cleavage sites of trypsin. Moreover, similar to the loss of function mutations of Nct (Fig. 1), mutations in SCAP also affect its conformational alteration as monitored by trypsin sensitivity (18). A successful conformational change of Nct requires the presence of the complete luminal domain. All ectodomain deletions analyzed not only lead to a loss of function but also fail to undergo the conformational alteration of Nct upon ␥-secretase complex assembly and maturation. Previously, a deletion of the DYIGS motif was shown to affect A␤ production (3). This is fully confirmed by our findings, which demonstrate that the same deletion (deletion construct 3 in Fig. 1a) does not restore ␥-secretase activity in a Nct knock-down background. However, not only the deletion of the DYIGS motif but all other deletions investigated within the ectodomain inhibit the formation of biologically active Nct and consequently a functional ␥-secretase complex. Certainly, this does not exclude the possibility that smaller deletions and point mutations may be tolerated.

FIG. 2. Mature Nct is trypsin-resistant. a, DDM-extracted HEK 293 cells stably transfected with wt Nct (as detailed in
Taken together our findings provide the first insights into the assembly and maturation of the ␥-secretase complex. Not only PS may exist as a "premature" variant (the PS holoprotein) but also Nct. In the case of Nct, "activation" is associated with a rather substantial conformational alteration that is required for ␥-secretase assembly and activity.