Research Article: Alzheimer’s Disease-Linked Mutations in Presenilin-1 Result in a Drastic Loss of Activity in Purified γ-Secretase Complexes

Date Published: April 18, 2012

Publisher: Public Library of Science

Author(s): Matthias Cacquevel, Lorène Aeschbach, Jemila Houacine, Patrick C. Fraering, Stefano L. Sensi.


Mutations linked to early onset, familial forms of Alzheimer’s disease (FAD) are found most frequently in PSEN1, the gene encoding presenilin-1 (PS1). Together with nicastrin (NCT), anterior pharynx-defective protein 1 (APH1), and presenilin enhancer 2 (PEN2), the catalytic subunit PS1 constitutes the core of the γ-secretase complex and contributes to the proteolysis of the amyloid precursor protein (APP) into amyloid-beta (Aβ) peptides. Although there is a growing consensus that FAD-linked PS1 mutations affect Aβ production by enhancing the Aβ1–42/Aβ1–40 ratio, it remains unclear whether and how they affect the generation of APP intracellular domain (AICD). Moreover, controversy exists as to how PS1 mutations exert their effects in different experimental systems, by either increasing Aβ1–42 production, decreasing Aβ1–40 production, or both. Because it could be explained by the heterogeneity in the composition of γ-secretase, we purified to homogeneity complexes made of human NCT, APH1aL, PEN2, and the pathogenic PS1 mutants L166P, ΔE9, or P436Q.

We took advantage of a mouse embryonic fibroblast cell line lacking PS1 and PS2 to generate different stable cell lines overexpressing human γ-secretase complexes with different FAD-linked PS1 mutations. A multi-step affinity purification procedure was used to isolate semi-purified or highly purified γ-secretase complexes. The functional characterization of these complexes revealed that all PS1 FAD-linked mutations caused a loss of γ-secretase activity phenotype, in terms of Aβ1–40, Aβ1–42 and APP intracellular domain productions in vitro.

Our data support the view that PS1 mutations lead to a strong γ-secretase loss-of-function phenotype and an increased Aβ1–42/Aβ1–40 ratio, two mechanisms that are potentially involved in the pathogenesis of Alzheimer’s disease.

Partial Text

Since their discovery in 1995 and their association with early onset familial Alzheimer’s disease (FAD) [1], [2], the presenilin genes PSEN1 and PSEN2 have been widely studied, and the complexity of their biological role is becoming increasingly evident. PSEN1 and PSEN2 encode transmembrane proteins PS1 and PS2, respectively, that constitute the catalytic core of γ-secretase, the founding member of an emerging class of unconventional, Intramembrane-Cleaving Proteases (I-CLiPs) [3]. Active γ-secretase is a multiprotein complex composed of PS1 or PS2 together with nicastrin (NCT), the anterior pharynx-defective protein 1 (APH1), and the presenilin enhancer 2 (PEN2). Experimental evidence such as the binding of transition-state analogue γ-secretase inhibitors to PS1 [4], as well as the abolishment of γ-secretase activity when PS1 lacks the aspartate residues critical for proteolysis [4], [5], [6], have confirmed that presenilins harbour the active site of the enzymatic complex.

Several conflicting results as to how Alzheimer’s disease-linked mutations in PSEN1 affect the processing of APP by γ-secretase have been reported. First, it has been previously reported that transgenic animals with FAD-linked PS1 mutations show increasing brain levels of Aβ1–42 [15], [38]. Since Aβ1–42, the first Aβ specie deposited in the brain of AD patients [16], [39], is more prone to aggregation when compared to shorter Aβ species [40], [41], [42], it has been implicated in the seeding of amyloid plaques in AD patients with PSEN1 mutations [43]. This was further validated in vivo as the overexpression of PS1 mutants in APP transgenic mice accelerated the rate of Aβ accumulation and deposition in the brain [44], [45]. However, Bentahir and colleagues challenged this view by showing that several PSEN1 mutations were also capable to decrease total Aβ production in PS1/PS2 knockout cells [23]. These findings suggested that endogenous PS1 and PS2 may influence the mutant phenotype in cells or in vivo. By extension, we hypothesized that the other components of γ-secretase, namely APH1, NCT and PEN2, may influence the mutant phenotype as well. Indeed, γ-secretase complexes are heterogeneous in composition due to the existence of two human APH1 genes, APH1a and APH1b, and two splicing isoforms of APH1a (S and L), as well as two presenilin genes, PSEN1 and PSEN2. Therefore, it is plausible that a single mutation in PSEN1 confers different catalytic properties to distinct γ-secretase subtypes. This hypothesis is supported by recent investigations showing that APH1 variants can modulate Aβ profiles. When compared to APH1aS or L, overexpression of APH1b in MEF knockout for all APH1 genes led to increased production of longer Aβ species [24]. With this regard, it is important to note that all four mouse variants of APH1 are expressed in the MEF cell line used in previous studies [23] and employed here (cf. Figure S3). Since the phenotype of PS1 mutations have mainly been assessed in vivo or in cell-based systems, we investigated the effects of FAD-linked PS1 mutants on the processing of APP-CTFs in cell-free systems, by using semi-purified and purified enzymatic complexes isolated from MEF PS1/PS2 double knockout cells stably overexpressing differentially tagged human γ-secretase components. Under these conditions, the biochemical and functional properties of γ-secretase complexes bearing either FAD-linked PS1 mutants (L166P, ΔE9, and P436Q), dominant-negative forms of PS1 or wild-type PS1 were characterized. The activity of γ-secretase with PS1-WT was similar to that reported for γ-secretase purified from our CHO cells overexpressing NCT-V5, APH1aL, Flag-PEN2 and PS1-WT [26], [27], that resulted in Aβ1–42/Aβ1–40 ratios between 0.1 and 0.3. In contrast, we found a major loss in the activity of γ-secretase complexes containing either the dominant-negative PS1 variants or the FAD-linked PS1 mutants. Although a total loss of activity was expected for the dominant negative forms of γ-secretase [5], [6], [32], the drastic loss of activity seen here with the FAD-linked PS1 mutants was unexpected. Indeed, the PS1-L166P variant has previously been reported to increase Aβ1–42 levels both in vivo[46] and in vitro[20], in the presence of endogenous PS1, PS2 and APH1 components. In contrast, Bentahir and colleagues found that PS1-L166P decreased both Aβ1–40 and Aβ1–42 production in a PS knockout background. These results are consistent with our in vitro data, although the reduction in Aβ1–42 and Aβ1–40 production was more pronounced in our system. However, they differ from our cell-based data in which we observed an increase in Aβ1–42 associated with a decrease in Aβ1–40. Taken together, these data suggest that the overexpression of the other human components of γ-secretase can influence the phenotype of FAD-linked mutations. Another possible explanation for these discrepancies comes from the use, in previously described cellular systems, of APP carrying the Swedish mutation (K670M/N671L) [20], [23], [46]. Initially, this APP variant was shown to enhance the production of all Aβ species by favouring its β-secretase cleavage [47], [48]. However, Munter and colleagues recently demonstrated that the APP Swedish mutation can also affect the specificity of the γ-secretase cleavage [49]. In particular, these authors showed that over-expression of APP Swedish in a neuronal cell line led to a 4-fold increase in secreted Aβ42, associated with only a 2-fold increase of total Aβ, compared with the wild-type APP. Therefore, one cannot exclude the possibility of differential interactions between PS1-WT or PS1 variants and different APP variants, as suggested earlier [50].