Requirement of PEN-2 for stabilization of the presenilin N-/C-terminal fragment heterodimer within the gamma-secretase complex.

Gamma-secretase is a protease complex composed of presenilin (PS), nicastrin (NCT), APH-1, and PEN-2, which catalyzes intramembrane cleavage of several type I transmembrane proteins including the Alzheimer's disease-associated beta-amyloid precursor protein. We generated stable RNA interference-mediated PEN-2 knockdown cells to probe mutant PEN-2 variants for functional activity. Knockdown of PEN-2 was associated with impaired NCT maturation and deficient PS1 endoproteolysis, which was efficiently rescued by wild type or N-terminally tagged PEN-2 but not by C-terminally tagged PEN-2 or by the C-terminally truncated PEN-2-DeltaC mutant. Although the latter mutants rescued the PS1 holoprotein accumulation associated with the PEN-2 knockdown, they failed to restore normal levels of the PS1 N- and C-terminal fragments and to maturate NCT. PEN-2-DeltaC was highly unstable and rapidly turned over by proteasomal degradation consistent with its failure to become stably incorporated into the gamma-secretase complex. In addition, expression of PEN-2-DeltaC caused a selective instability of the PS1 N-/C-terminal fragment heterodimer that underwent proteasomal degradation, whereas NCT and APH-1 were stable. Interestingly, when we knocked down PEN-2 in the background of the endoproteolysis-deficient PS1 Deltaexon9 mutant, immature NCT still accumulated, demonstrating that PEN-2 is also required for gamma-secretase complex maturation when PS endoproteolysis cannot occur. Taken together, our data suggest that PEN-2 is required for the stabilization of the PS fragment heterodimer within the gamma-secretase complex following PS endoproteolysis. This function critically depends on the PEN-2 C terminus. Moreover, our data show that PEN-2 is generally required for gamma-secretase complex maturation independent of its activity in PS1 endoproteolysis.

␥-Secretase is a protease complex composed of presenilin (PS), nicastrin (NCT), APH-1, and PEN-2, which catalyzes intramembrane cleavage of several type I transmembrane proteins including the Alzheimer's disease-associated ␤-amyloid precursor protein. We generated stable RNA interference-mediated PEN-2 knockdown cells to probe mutant PEN-2 variants for functional activity. Knockdown of PEN-2 was associated with impaired NCT maturation and deficient PS1 endoproteolysis, which was efficiently rescued by wild type or N-terminally tagged PEN-2 but not by C-terminally tagged PEN-2 or by the C-terminally truncated PEN-2-⌬C mutant. Although the latter mutants rescued the PS1 holoprotein accumulation associated with the PEN-2 knockdown, they failed to restore normal levels of the PS1 N-and C-terminal fragments and to maturate NCT. PEN-2-⌬C was highly unstable and rapidly turned over by proteasomal degradation consistent with its failure to become stably incorporated into the ␥-secretase complex. In addition, expression of PEN-2-⌬C caused a selective instability of the PS1 N-/C-terminal fragment heterodimer that underwent proteasomal degradation, whereas NCT and APH-1 were stable. Interestingly, when we knocked down PEN-2 in the background of the endoproteolysis-deficient PS1 ⌬exon9 mutant, immature NCT still accumulated, demonstrating that PEN-2 is also required for ␥-secretase complex maturation when PS endoproteolysis cannot occur. Taken together, our data suggest that PEN-2 is required for the stabilization of the PS fragment heterodimer within the ␥-secretase complex following PS endoproteolysis. This function critically depends on the PEN-2 C terminus. Moreover, our data show that PEN-2 is generally required for ␥-secretase complex maturation independent of its activity in PS1 endoproteolysis.
The Alzheimer's disease (AD) 1 -associated ␥-secretase is a high molecular weight complex with an aspartyl protease ac-tivity that catalyzes the second of two subsequent proteolytic cleavages of the ␤-amyloid precursor protein (APP) implicated in AD (1). This unusual, intramembranous cleavage liberates the neurotoxic amyloid-␤ peptide from the membrane. Amyloid-␤ peptide aggregates and is deposited in the brain of AD patients in amyloid plaques, an invariant pathological hallmark of AD (2). The ␥-secretase complex is composed of either of the two presenilins (PS), PS1 and PS2, polytopic membrane proteins that are endoproteolytically cleaved into stable heterodimers consisting of an N-and C-terminal fragment (NTF and CTF) (3), the type I transmembrane glycoprotein nicastrin (NCT) (4,5), and the polytopic membrane proteins PEN-2 (6), APH-1a, and/or APH-1b (6,7). Expression of the ␥-secretase complex components is coordinately regulated. RNA interference (RNAi)-mediated knockdown studies in cultured Drosophila and mammalian cells as well as studies with cells derived from PS1, PS1/2, or NCT knockout mice have shown that ␥-secretase complex formation is coordinately regulated. Downregulation or genetic ablation of a given component typically causes a decrease in the level of the other components and/or in the maturation and trafficking of the ␥-secretase complex to its functional sites (6, 8 -18). An exception is APH-1, which is stable in the absence of PS (13,19,20). Increased ␥-secretase complex formation and activity have been observed when all four ␥-secretase complex components are overexpressed in cultured Drosophila and mammalian cells (17,21,22). Moreover, reconstitution of ␥-secretase complex formation and activity has recently been achieved in yeast, an organism that lacks any endogenous ␥-secretase activity (23). This demonstrated that coexpression of all four components is necessary and sufficient for the reconstitution of ␥-secretase activity (23).
Overexpression and RNAi-mediated knockdown studies with cultured Drosophila cells have provided a first model of how the ␥-secretase complex assembles (17). These studies indicate that the PS holoprotein may assemble first with NCT and/or APH-1 to form a stable assembly intermediate (17). Subsequent assembly of PEN-2 with this intermediate drives the conversion of the PS holoprotein into the active heterodimer (17). Furthermore, a potential NCT/APH-1 assembly intermediate has been observed recently (24), and NCT has been shown to stably interact with APH-1 in the absence of PS (20,25). Interestingly, ␥-secretase complex assembly is associated with a conformational change of the NCT ectodomain, which adopts a protease-resistant conformation (26). At present, little is known about the subunit organization within the ␥-secretase complex. However, recent studies (27,28) suggest that PS is a dimer within the ␥-secretase complex. In addition, an interaction of the PS1 NTF with PEN-2 has been reported (29).
A large body of evidence suggests that PS carries the catalytic site of the ␥-secretase complex. Mutagenesis of conserved aspartates in transmembrane domains 6 and 7 of PS (30,31) as well as aspartyl protease transition state analogues, which bind to the PS heterodimer (32,33), inhibit ␥-secretase activity. Moreover, PS belongs to a novel group of several polytopic aspartyl protease families (1,34), which are defined by the presence of highly conserved, short GXGD active site motifs (35)(36)(37). Besides their requirement for ␥-secretase complex assembly and maturation, the precise functional role of the other ␥-secretase complex components is unclear. In particular, apart from its role in facilitation of PS endoproteolysis (14,17), functional information about the smallest subunit, the ϳ10-kDa PEN-2 protein, is scarce. In order to gain further insight into its functional role, we thus initiated a structure-function analysis of this protein.

EXPERIMENTAL PROCEDURES
Antibodies-The polyclonal antibody 1638 raised to the N terminus of human PEN-2 (11), the polyclonal and monoclonal antibodies against the PS1 C terminus (3027 and BI.3D7) (38), the N terminus (2953) (39), and PS1N (40), and the polyclonal antibody to the C terminus of APP (6687) (36) were described previously. The polyclonal antibodies 433 and 434 were generated against the APH-1aL C terminus (residues 245-265) and affinity-purified using a GST-APH-1aL (residues 207-265) fusion protein. The polyclonal antibody N1660 against the C terminus of NCT and the anti-␤-actin antibody were obtained from Sigma, the anti-Xpress antibody was obtained from Invitrogen; the anti-␤catenin antibody was from Transduction Laboratories, and the anti-Myc antibody 9E10 was from Santa Cruz Biotechnology.
Protein Analysis-Membrane fractions of HEK 293 cells were obtained by ultracentrifugation of postnuclear supernatant fractions from cell homogenates that were prepared as described (43). For direct immunoblot analysis, membrane fractions were solubilized with STENlysis buffer (50 mM Tris, pH 7.6, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, protease inhibitors (Sigma)). After a clarifying spin, lysates were analyzed for PEN-2 by immunoblotting with antibody 1638, for PS1 by immunoblotting with antibodies PS1N or 3027, for NCT with antibody N1660, for APH-1aL by immunoblotting with antibody 433 or 434, and for APP and APP CTFs by immunoblotting with antibody 6687. For coimmunoprecipitation analysis, membrane fractions were solubilized in CHAPSO-lysis buffer (1% CHAPSO, 150 mM sodium citrate, pH 6.4, protease inhibitors (Sigma)), subjected to a clarifying spin by ultracentrifugation, and incubated with preimmune serum, N1660 or 2953 antibody, and protein G-Sepharose for 2 h at 4°C. Following two washes with CHAPSO-wash buffer (0.5% CHAPSO, 150 mM sodium citrate, pH 6.4), the immunoprecipitates were subjected to immunoblot analysis as above.

RESULTS
In order to identify functional domains of PEN-2 required for ␥-secretase complex formation, maturation, and activity, we generated a HEK 293 cell line stably overexpressing Swedish mutant APP (HEK 293/sw) in which PEN-2 expression is stably knocked down by RNAi (Fig. 1A). This PEN-2 knockdown cell line was stably transfected with cDNA constructs encoding RNAi-resistant wt and mutant PEN-2 variants to assess their capability to rescue PEN-2 deficiency. We first transfected the PEN-2 knockdown cells with cDNA constructs encoding untagged wt PEN-2 and N-terminal hexahistidine-Xpress (H 6 X) epitope-tagged and C-terminal myc-hexahistidine (mH 6 ) epitope-tagged wt PEN-2. All three proteins were stably expressed at robust levels, and as expected, the tagged PEN-2 variants migrated at higher molecular weight than untagged PEN-2 (Fig. 1A). Consistent with previous results (6,17,24), the PEN-2 knockdown was associated with reduced ␥-secretase activity as manifested by the accumulation of APP CTFs, which was rescued by the expression of wt PEN-2 and H 6 X-PEN-2 (Fig. 1B). APP CTFs also accumulated when PEN-2-mH 6 was expressed indicating that the authentic C terminus of PEN-2 is critical for ␥-secretase activity.
We next sought to investigate why expression of PEN-2-mH 6 caused a defect in ␥-secretase activity, and we asked whether a defect in ␥-secretase complex maturation was responsible for the observed loss of ␥-secretase function. To confirm further the functional importance of the PEN-2 C terminus, we also constructed a PEN-2 variant with a C-terminal truncation of 17 amino acids (PEN-2-⌬C), and we expressed it stably in the PEN-2 knockdown cells (Fig. 1C). We first investigated the maturation of PS. Consistent with previous results (11,14,24), down-regulation of PEN-2 was associated with reduced levels of PS fragments and an accumulation of the PS1 holoprotein (Fig. 1D). Expression of untagged PEN-2 allowed efficient recovery of PS1 holoprotein endoproteolysis (Fig. 1D). In contrast, expression of PEN-2-mH 6 and of PEN-2-⌬C caused an intriguing, unexpected biochemical phenotype. Although the PS1 holoprotein did not accumulate anymore, the levels of the PS1 NTF and CTF were strongly reduced similar to those observed in the PEN-2 knockdown cells (Fig. 1D). Thus, although the PEN-2 C terminus appears to be dispensable for PS1 endoproteolysis, it is required for the maintenance of normal levels of the PS fragments. As reported previously (11,24), the PEN-2 knockdown caused impaired NCT maturation, which was efficiently restored by the expression of untagged PEN-2 (Fig. 1E). Interestingly, the levels of APH-1aL were largely unaffected by the knockdown of PEN-2 (Fig. 1E). Neither PEN-2-mH 6 nor PEN-2-⌬C rescued the maturation defect of NCT (Fig. 1D). Consequently, as observed for PEN-2-mH 6 (Fig. 1B), expression of PEN-2-⌬C was associated with impaired ␥-secretase activity as judged from the accumulation of APP CTFs (Fig. 1F). Thus, either addition of an epitope tag to the C terminus or a large deletion of it abrogated PEN-2 function in ␥-secretase complex formation, stability, maturation, and activity. Taken together, these data suggest that the C terminus is a functionally important domain of PEN-2 required for the stabilization of PS fragments.
We next asked why the C-terminal PEN-2 variants failed to stabilize PS1 fragments. To address this question we sought to investigate the stability of PEN-2-⌬C and treated cells with cycloheximide to block de novo protein synthesis. We first investigated the stability of the ␥-secretase complex components in control and PEN-2 knockdown cells. As expected, endogenous PEN-2, PS1 fragments, NCT, and APH-1aL were stable in control cells (Fig. 2A). Similarly, residual endogenous PEN-2 in the knockdown cells was stable as were the PS1 holoprotein, NCT, and APH-1aL in these cells ( Fig. 2A), consistent with previous results (17). We then analyzed the stability of PEN-2 or PEN-2-⌬C expressed in the background of PEN-2 knockdown cells. wt PEN-2 occurred as a stable protein (Fig. 2B). In contrast, cycloheximide treatment revealed that PEN-2-⌬C was highly unstable and almost completely turned over within 3 h (Fig. 2B). We next investigated the stability of PS1, NCT, and APH-1aL in the background of wt PEN-2 and PEN-2-⌬C expression. The PS1 fragments, NCT and APH-1aL, were stable in the PEN-2 knockdown cells stably expressing wt PEN-2. In contrast, in the PEN-2-⌬C-expressing knockdown cells, the already low levels of PS1 fragments dropped further during the time course of the cycloheximide treatment. This indicates that the PS1 fragments were not stabilized and thus turned over (Fig. 2B). Interestingly, immature NCT and APH-1aL as well as the low levels of residual mature NCT (because of the incomplete PEN-2 knockdown) remained stable (Fig. 2B), consistent with the observation that these two proteins can independently stabilize each other (13,20,24,25). Taken together, these data demonstrate that PEN-2-⌬C is highly unstable. Furthermore, these data indicate that expression of PEN-2-⌬C might cause a selective destabilization of the PS1 NTF/CTF heterodimer within the ␥-secretase complex.
We next asked which proteolytic pathway might be involved in the degradation of PEN-2-⌬C and the destabilized PS1 fragments. Because we had shown previously that excess amounts of free PS proteins that have not undergone complex formation are rapidly turned over by proteasomal degradation (44), we treated the above-described cells with the highly potent and specific proteasome inhibitor clasto-lactacystin ␤-lactone (45).
To control the inhibitor treatment, we monitored the appearance of polyubiquitinated ␤-catenin species (44,46). As expected from their stable behavior shown above, clasto-lactacystin ␤-lactone treatment did not result in increased levels of endogenous PEN-2 in control cells or of the residual endogenous PEN-2 in the PEN-2 knockdown cells (Fig. 3A). The levels of wt PEN-2 expressed in the background of PEN-2 knockdown cells were slightly increased indicating degradation of excess free amounts (Fig. 3A). In contrast, clasto-lactacystin ␤-lactone treatment caused a very strong increase in the levels of PEN-2-⌬C (Fig. 3A) suggesting that the proteasome degrades this unstable protein. The levels of endogenous PS1 fragments in control or of residual endogenous PEN-2 in the knockdown cells were unchanged upon clasto-lactacystin ␤-lactone treatment (Fig. 3B). A slight increase in the levels of the PS1 holoprotein was observed in the PEN-2 knockdown cells suggesting proteasomal degradation of minor amounts of unstable free holoprotein (Fig. 3B). As expected, no increase in the levels of the PS1  myc epitope). B, membrane fractions of cell lines described in A were analyzed for APP and APP CTFs generated by ␤and ␣-secretase by immunoblotting with antibody 6687 (to the APP C terminus). C, membrane fractions from the indicated HEK 293/sw (control), PEN-2 knockdown, or PEN-2 knockdown cells stably transfected with cDNA constructs encoding RNAi-resistant wt PEN-2, PEN-2-mH 6 , or PEN-2-⌬C were analyzed for PEN-2 as in A. D, membrane fractions of cell lines described in C were analyzed for PS1 holoprotein, NTF, and CTF by immunoblotting with antibodies PS1N (to the PS1 N terminus) and 3027 (to the PS1 loop). The asterisk indicates the phosphorylated form of the PS1 CTF (39). E, membrane fractions of cell lines described in C were analyzed for NCT and APH-1aL by immunoblotting with antibodies N1660 (to the NCT C terminus) and 433 (to the APH-1aL C terminus). F, membrane fractions of PEN-2 knockdown cells stably expressing RNAi-resistant wt PEN-2 or PEN-2-⌬C were analyzed for PEN-2 as in A and for APP and APP CTFs as in B.

PEN-2-dependent Stabilization of PS NTF/CTF Heterodimer
fragments was observed in PEN-2 knockdown cells expressing wt PEN-2 (Fig. 3B). In contrast, the levels of the PS1 fragments were markedly increased in PEN-2 knockdown cells expressing PEN-2-⌬C, suggesting that following their generation by endoproteolysis, the destabilized fragments were degraded by the proteasome (Fig. 3B). No increase in the levels of the PS1 holoprotein was observed, further supporting that the PS1 holoprotein had undergone endoproteolysis and not simply undergone proteasomal degradation. Consistent with their stable behavior shown above (Fig. 2, A and B), levels of immature and mature forms of NCT as well as of APH-1aL were unaffected upon proteasome inhibition (Fig. 3C). In summary, these results suggest that PEN-2-⌬C is capable of initiating PS1 endoproteolysis. However, the resulting PS1 NTF and CTF are not stabilized and like PEN-2-⌬C itself are degraded by the proteasome.
The results described above demonstrate that PEN-2-⌬C is highly unstable and indicate that the cause of its instability may be due to a defect in stable ␥-secretase complex formation. We therefore next asked whether PEN-2-⌬C might have lost its capability to stably associate with the other ␥-secretase complex components. Membrane fractions were isolated from PEN-2 knockdown cells stably transfected with wt PEN-2 or PEN-2-⌬C and were detergent-solubilized with CHAPSO. The CHAPSO-extracted membrane proteins were subjected to coimmunoprecipitation analysis with antibodies to NCT and PS1. Robust amounts of wt PEN-2, PS1 NTF, and CTF as well as APH-1aL and mature NCT were coimmunoprecipitated with both antibodies in the CHAPSO-solubilized membrane fractions derived from the wt PEN-2-expressing cells (Fig. 4). In contrast, both antibodies failed to coimmunoprecipitate significant amounts of PEN-2-⌬C in the CHAPSO-solubilized membrane fractions derived from the PEN-2-⌬C-expressing cells, suggesting that PEN-2-⌬C failed to stably interact with the other ␥-secretase complex components (Fig. 4). The anti-NCT antibody coimmunoprecipitated APH-1aL with immature NCT in the CHAPSO-solubilized membrane fractions from the PEN-2-⌬C-expressing cells, consistent with their independent stable interaction (13,20,24,25). Thus, these data demonstrate that PEN-2-⌬C does not undergo stable ␥-secretase complex formation. As a consequence, the protein becomes unstable and undergoes rapid proteasomal degradation.
The observation by others (14,17,24) and us (Fig. 1C) that a PEN-2 knockdown causes an accumulation of the PS holoprotein suggests that PEN-2 is required for PS endoproteolysis. However, as shown above, PEN-2-mH 6 and PEN-2-⌬C are defective in the stabilization of the PS1 NTF and CTF but not in endoproteolysis. This suggests that following endoproteolysis, PEN-2 also becomes necessary for stabilization of the active site PS heterodimer, and consequently for the further maturation of the ␥-secretase complex. In addition, following PS endoproteolysis PEN-2 apparently remains bound to the ␥-secretase complex (11,17,21,22), further indicating that PEN-2 may have a function independent of PS endoproteolysis. To directly prove this, we investigated the consequences of PEN-2 down-regulation in the background of PS1 ⌬exon9 (47), a FADassociated PS1 variant that undergoes stable ␥-secretase complex formation (48) but that does not undergo endoproteolysis because of a deletion of the cleavage site domain (3,38). Indeed, when we stably knocked down PEN-2 expression by RNAi in HEK 293/sw cells stably expressing PS1 ⌬exon9 (38), NCT maturation was impaired (Fig. 5). Consistent with the results shown above, APH-1aL levels were largely unaffected by the knockdown of PEN-2 (Fig. 5). Thus, these data show that PEN-2 has a general function in the maturation of the ␥-secretase complex, which is independent from its role in PS endoproteolysis. DISCUSSION Little functional information is available about PEN-2, the smallest of the four essential subunits of the ␥-secretase complex. RNAi-mediated knockdown experiments in Drosophila and mammalian cells demonstrated that a knockdown of PEN-2 is associated with reduced PS fragment levels and ac-

FIG. 3. Proteasomal degradation of PEN-2-⌬C and the PS1 NTF and CTF.
A, HEK 293/sw (control), PEN-2 knockdown, and PEN-2 knockdown cells stably expressing RNAi-resistant wt PEN-2 or PEN-2-⌬C were treated with clasto-lactacystin ␤-lactone (10 g/ml) for 24 h. Membrane fractions were prepared and analyzed for PEN-2 as in Fig. 1. To control the inhibitor treatment, we monitored the appearance of polyubiquitinated ␤-catenin species (44,46) by immunoblotting with an anti-␤-catenin antibody. B, membrane fractions of cell lines described in A were analyzed for PS1 holoprotein, NTF, and CTF by immunoblotting as described in Fig. 1E. To control for equal amounts of protein loaded, levels of ␤-actin were monitored by immunoblotting with an anti-␤-actin antibody. The asterisk indicates the phosphorylated form of the PS1 CTF (39). C, membrane fractions of cell lines described in A were analyzed for NCT and APH-1aL by immunoblotting as described in Fig. 1E. cumulation of the PS holoprotein (6,11,14,17). Moreover, it has been shown that PEN-2 not only forms stable associates with the PS fragments but can also associate with the PS holoprotein (14). Accordingly, PEN-2 has been suggested to play an important role in ␥-secretase complex maturation by conversion of the PS holoprotein into the PS NTF/CTF heterodimer (14,17). To better understand PEN-2 function and to define its functional domains, we initiated a structure-function analysis of PEN-2 using a stable knockdown cell line in which loss of PEN-2 function can be assayed by functional complementation. Consistent with previous findings, stable knockdown of PEN-2 was associated with impaired NCT maturation, reduced PS fragment levels, and an accumulation of the PS1 holoprotein. This defect in ␥-secretase complex maturation was efficiently rescued by expression of untagged or N-terminaltagged but not by C-terminal tagged PEN-2, suggesting that the C terminus of PEN-2 is a functionally critical domain, which is sensitive to steric alteration. Indeed, expression of the C-terminal truncated PEN-2-⌬C variant also caused loss of PEN-2 function. Both C-terminal loss of function mutants of PEN-2 showed very similar if not identical biochemical behavior. NCT maturation was impaired, and the defective PS1 holoprotein accumulation was rescued without stabilization of the PS1 fragments. Thus, these PEN-2 variants are apparently capable of initiating PS1 endoproteolysis but cannot stabilize the PS fragments generated. A possible reason for the failure of PEN-2 with a C-terminal alteration to stabilize the resulting PS1 NTF and CTF is its own instability within the ␥-secretase complex. Indeed, PEN-2-⌬C failed to undergo stable ␥-secretase complex formation. As a consequence, PEN-2-⌬C behaved as a highly unstable protein and was rapidly turned over by the proteasome. This is in contrast to correctly assembled ␥-secretase complex components, which have been shown to be very stable and have a long half-life time (9,17,44,48). The failure of PEN-2-⌬C to stably assemble with PS, NCT, and APH-1 was associated with a marked selective instability of the PS1 NTF and CTF, which also underwent proteasomal degradation, whereas NCT and APH-1 were stable, consistent with previous findings that these proteins can stabilize each other independently (20,24,25). These data suggest that PEN-2 has a dual function as follows: a function in the initiation of PS endoproteolysis as a first maturation step during ␥-secretase complex assembly, and a second function in the stabilization of the PS fragment heterodimer within the ␥-secretase complex that is required for further maturation of the ␥-secretase complex. Finally, when we knocked down PEN-2 in the background of the endoproteolysis-deficient PS1 ⌬exon9 mutant, NCT accumulated as an immature protein demonstrating that PEN-2 is also required for ␥-secretase complex maturation under conditions when PS endoproteolysis cannot occur. Thus, PEN-2 is generally required for ␥-secretase complex maturation independent of its role in PS endoproteolysis.
Taken together, our data add important new functional information on PEN-2 in ␥-secretase complex assembly, maturation, and activity. We suggest that during ␥-secretase complex assembly, PEN-2 interacts with the PS holoprotein that is stabilized by immature NCT and APH-1. Upon interaction with PEN-2, the PS holoprotein may adopt a conformation that allows PS endoproteolysis. Following PS endoproteolysis, the PS NTF and CTF are stabilized by PEN-2 which itself has become stabilized within the complex. Whereas the C-terminal domain of PEN-2 is dispensable for the PS endoproteolysis, the stabilization of the NTF/CTF heterodimer and of PEN-2 itself is only possible when PEN-2 carries an intact and/or accessible C terminus. This strongly suggests that the C terminus carries important sequence information, which may be required to stably interact with other ␥-secretase complex components. These stabilizing interactions are necessary for further maturation of the ␥-secretase complex by trafficking through the secretory pathway. Future work will aim at elucidating the molecular basis for the critical requirement of the PEN-2 C terminus in ␥-secretase complex assembly, maturation, and activity.