PEN-2 Is an Integral Component of the γ-Secretase Complex Required for Coordinated Expression of Presenilin and Nicastrin*

The Alzheimer disease-associated presenilin (PS) proteins apparently provide the active site of γ-secretase, an unusual intramembrane-cleaving aspartyl protease. PSs principally occur as high molecular weight protein complexes that contain nicastrin (Nct) and additional so far unidentified components. Recently, PEN-2 has been implicated in γ-secretase function. Here we identify PEN-2 as a critical component of PS1/γ-secretase and PS2/γ-secretase complexes. Strikingly, in the absence of PS1 and PS1/PS2, PEN-2 levels are strongly reduced. Similarly, PEN-2 levels are reduced upon RNA interference-mediated down-regulation of Nct. On the other side, down-regulation of PEN-2 by RNA interference is associated with reduced PS levels, impaired Nct maturation, and deficient γ-secretase complex formation. We conclude that PEN-2 is an integral γ-secretase complex component and that γ-secretase complex components are expressed in a coordinated manner.

releases the 40 -42-amino acid amyloid ␤-peptide (A␤) from the membrane (1). The majority of familial Alzheimer disease cases are associated with mutations in the PS1 gene (1). Apparently all PS1 mutations investigated cause an increased generation of the highly amyloidogenic A␤42 (1). Absence of PS1 reduces ␥-secretase activity (2,3) and absence of PS1/PS2 eliminates ␥-secretase function completely (4,5). Mounting evidence suggests that PSs are unusual aspartyl proteases with ␥-secretase activity (6). All PSs contain two aspartates within transmembrane domains 6 and 7 that are critically required for ␥-secretase activity (7,8). Moreover, ␥-secretase inhibitors designed to mimic the transition state of the catalytic mechanism of aspartyl proteases can be covalently cross-linked to PSs (9,10). Finally, PSs are apparently members of a group of polytopic membrane-bound aspartyl proteases that are all characterized by a GXGD (X ϭ variable amino acid) signature motif in which the C-terminal active site aspartate is embedded (11). Besides the PSs, the bacterial type 4 prepilin peptidases (11,12) and signal peptide peptidase and its related proteins carry this signature motif (13,14).
PSs are not only required for the ␥-secretase-mediated processing of APP but also for intramembrane proteolysis of several other type I transmembrane proteins including the Notch cell surface receptors that are critically required for cell fate decisions (24). Notch signaling depends on the PS-dependent S3 cleavage of Notch that leads to the liberation of the Notch intracellular domain (NICD) from the membrane (24). NICD translocates to the nucleus where it is involved in the transcription of target genes (24). Genetic screening in Caenorhabditis elegans led to the identification of novel components, APH-1 and PEN-2, that are required for ␥-secretase activity in APP and Notch S3 cleavage (21,25). The function of APH-1 and PEN-2 is currently unclear. Apparently, APH-1 and PEN-2 could either be transiently required for the assembly of the ␥-secretase complex or may even be novel bona fide complex components required for ␥-secretase activity (21). Here we investigated whether PEN-2 is an integral ␥-secretase complex component.
cDNA Constructs and Transfections-N-terminally hexahistidine-Xpress (H 6 X)-epitope-tagged PS1, PS1 ⌬exon9, and PS2 variants were generated by cloning the respective cDNAs into pcDNA4/HisC expression vector (Invitrogen) and stably transfected into human embryonic kidney 293 (HEK 293) stably expressing Swedish mutant APP (swAPP) (29). * This work was supported by the Deutsche Forschungsgemeinschaft (Priority program "Cellular Mechanisms of Alzheimer's Disease") and the National Genome Research Network (NGFN). 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.
Cell Culture, Cell Lines, RNA Interference (RNAi), and Protein Analysis-HEK 293 stably expressing swAPP and mouse embryonic fibroblast cells derived from PS knock-out or littermate control mice (30) were cultured as described (18). To inhibit expression of PEN-2 and Nct by RNAi, HEK 293 cells were transiently transfected with siRNA duplexes PEN-2-160 (directed to the target sequence 5Ј-AAAGGCTAT-GTCTGGCGCTCA-3Ј) and Nct-1045 as described (18). Cell lysates of HEK 293 cells were analyzed for PEN-2 by combined immunoprecipitation/immunoblotting with antibody 1638, for PSs by combined immunoprecipitation/immunoblotting as described (26), and for Nct by direct immunoblotting with antibody N1660. Immunoprecipitations of PEN-2 were carried out in the presence of 0.1% SDS. To analyze PEN-2, PS, and Nct expression levels in mouse embryonic fibroblast cells, cell homogenates were subjected to ultracentrifugation, and the pellet fraction was solubilized in the presence of 1% SDS as described (31). The SDS extracts were diluted 10-fold and analyzed for PEN-2, PSs, and Nct by combined immunoprecipitation/immunoblotting or immunoblotting as described above.

RESULTS AND DISCUSSION
To identify endogenous PEN-2, we raised antibody 1638 to the N terminus of human PEN-2. To prove the specificity of this antibody, cell lysates of HEK 293 cells were subjected to immunoprecipitation with antibody 1638 or the corresponding preimmune serum. HEK 293 cells are known from numerous previous studies to express a ␥-secretase activity, which has identical functional and biochemical properties as the ␥-secretase activity in neuronal cells or brain tissue (see Ref. 33 and citations therein). As shown in Fig. 1A, robust amounts of the ϳ10-kDa PEN-2 were immunoprecipitated with antibody 1638 but not with preimmune serum.
To address the question whether PEN-2 is a bona fide component of PS1-and PS2/␥-secretase complexes we first performed co-immunoprecipitation analyses using HEK 293 cells stably transfected with N-terminal hexahistidine-Xpress (H 6 X)-epitope-tagged PS1 or PS2. ␥-Secretase complexes were isolated from DDM-solubilized membrane fractions by immunoprecipitation with an anti-Xpress antibody. As expected, PS1 and PS2 holoproteins and N-terminal fragments were immunoprecipitated, and the corresponding C-terminal fragments were co-immunoprecipitated (Fig. 1B). Under these conditions endogenous PEN-2 was found to co-immunoprecipitate with PS1 and PS2. In addition, Nct also co-immunoprecipitated (Fig.  1B). As shown in Fig. 1C, association of PEN-2 with the ␥-secretase complex is independent of endoproteolysis of PS1. Upon stable expression of H 6 X-PS1 ⌬exon9, which does not undergo endoproteolysis due to the lack of the cleavage site for PS endoproteolysis (34), PEN-2 was still co-isolated with the PS1 ⌬exon9 holoprotein and with Nct (Fig. 1C). These results suggest that PEN-2 is a component of individual PS1/␥-secretase and PS2/␥-secretase complexes. Under these conditions we could only immunoprecipitate very minor amounts of PEN-2 with antibody 1638 (data not shown), whereas mild denaturation with 0.1% SDS was sufficient to robustly immuoprecipitate PEN-2 (see Fig. 1A). This suggests that PEN-2 is tightly packed within the ␥-secretase complex.
To confirm that PEN-2 is an integral component of the en-dogenous ␥-secretase complex co-immunoprecipitations were carried out with untransfected HEK 293 cells. DDM-solubilized membrane fractions were immunoprecipitated with antibodies to the C terminus of Nct and the N terminus of PS1 or PS2. Immunoblotting with anti-PEN-2 antibody 1638 revealed robust amounts of co-immunoprecipitated PEN-2 demonstrating that PEN-2 is a component of endogenous PS/␥-secretase complexes (Fig. 1D). As expected, antibodies to the N termini of PS1 or PS2 co-immunoprecipitated the corresponding PS1 and PS2 C-terminal fragments and mature Nct (Fig. 1D). Taken together, these experiments demonstrate that PEN-2 is an integral bona fide ␥-secretase complex component. Eliminating PS1 or PS1/PS2 causes reduced levels of the ␥-secretase complex and inhibits Nct maturation (18,23). Moreover inhibition of Nct expression also strongly reduced PS expression and inhibited ␥-secretase complex formation (18). This suggested that expression of ␥-secretase complex components might be coordinately regulated. We therefore analyzed PEN-2 expression in PS knock-out mice (2,4,30). As expected, robust levels of PEN-2 were observed in embryonic fibroblasts derived from littermate control mice ( Fig. 2A). Ablation of PS1 caused a strong reduction of PEN-2 levels, and the absence of PS1/PS2 eliminated PEN-2 expression nearly completely. Compared with the PS1 ablation, ablation of PS2 still allowed significant PEN-2 expression ( Fig. 2A). Similar effects of the PS ablations were found on Nct expression. Ablation of PS1 or PS1/PS2 dramatically reduced Nct maturation, whereas ablation of PS2 allowed Nct maturation ( Fig. 2A). We next investigated whether PEN-2 expression is also dependent on the presence of Nct. HEK 293 cells were transfected with the siRNA duplex Nct-1045 to down-regulate Nct expression by RNAi (18). As shown in Fig. 2B, RNAi-mediated down-regulation of Nct caused a reduction of PEN-2 levels. Thus, expression of PEN-2 not only requires the presence of PSs (predominantly PS1) but that of Nct as well. To investigate whether lowering PEN-2 levels affects PS and Nct levels, we next transfected HEK 293 cells with the siRNA duplex PEN-2-160. PEN-2-RNAi reduced PEN-2 expression, which was accompanied by reduced levels of PS1 and PS2 (Fig. 3A). RNAi-mediated downregulation of PEN-2 levels also impaired the maturation of Nct (Fig. 3A) as had been observed upon inhibition of PS1 or PS1/ PS2 ( Fig. 2A). As a consequence of reduced PS expression and impaired Nct maturation, inhibition of PEN-2 expression caused a deficiency in the PS1/␥-secretase complex (Fig. 3B) as revealed by blue native PAGE analysis (18).
Taken together our findings demonstrate that at least PS, Nct, and PEN-2 are integral components of the ␥-secretase complex. Expression of these ␥-secretase complex components is coordinately regulated. Removing any of these three components results in reduced amounts of the ␥-secretase complex and consequently in a loss of ␥-secretase activity (18,21). We propose that any additional so far unknown ␥-secretase complex component may affect PS, Nct, and PEN-2 expression. Obvious candidates for additional ␥-secretase complex compo-nents are APH-1a and APH-1b, the two human homologues of C. elegans APH-1 (21). Both act directly at or upstream of ␥-secretase activity and apparently affect APH-2 (the C. elegans homologue of Nct) transport to the cell surface (25). Moreover, like Nct and PEN-2, APH-1a and APH-1b are required for ␥-secretase activity (21) .  FIG. 2. PS1, PS2, and Nct are required for the expression of PEN-2. A, membrane lysates from mouse embryonic fibroblasts (PS1 ϩ/ϩ /PS2 ϩ/ϩ , PS1 Ϫ/Ϫ /PS2 ϩ/ϩ ; PS1 ϩ/ϩ /PS2 Ϫ/Ϫ , PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ ) were analyzed for levels of PEN-2, PS1, and PS2 by combined immunoprecipitation/immunoblotting with antibody 1638 (to PEN-2) and with antibodies 3027/BI.3D7 (to PS1) and 3711/BI.HF5c (to PS2). Nct was analyzed by immunoblotting with antibody N1660. Cells lacking PS1 and PS1/PS2 show significantly reduced levels of PEN-2. Note that Nct maturation is strongly reduced in the absence of PS1 and PS1/PS2. The asterisk denotes the phosphorylated form of the PS1 C-terminal fragment (27). B, HEK 293 cells were subjected to two consecutive transfections with siRNAi duplex Nct-1045 as described (18). Cell lysates were analyzed for levels of PEN-2 by combined immunoprecipitation/ immunoblotting with antibody 1638. Down-regulation of Nct was confirmed by immunoblotting with antibody N1660 .   FIG. 3. Down-regulation of PEN-2 by RNAi is associated with reduced PS levels, impaired Nct maturation, and PS1 complex deficiency. A, HEK 293 cells were subjected to two consecutive transfections with siRNA duplex PEN-2-160 as described (18). PEN-2 and Nct were analyzed as in Fig. 2B, and PSs were analyzed by combined immunoprecipitation/immunoblotting with antibodies 3027/BI.3D7 (to PS1) and 3711/BI.HF5c (to PS2). Note the altered ratio of mature/ immature Nct in cells subjected to PEN-2-RNAi compared with untreated control cells. B, HEK 293 cells were subjected to PEN-2-RNAi as in A. Membrane fractions were solubilized with DDM, subjected to blue native PAGE, and analyzed for the PS1 complex by immunoblotting with antibody 2953. Aliquots were subjected to SDS-PAGE to confirm down-regulation of PEN-2 by immunoblotting with antibody 1638.