Skip to main content

Subcellular localization of distinct γ-secretase complexes defines substrate specificity

Final Report Summary - GSEC-AD (Subcellular localization of distinct γ-secretase complexes defines substrate specificity)

γ-Secretase is a key enzyme in Alzheimer’s disease (AD). It resides in the membrane and performs proteolytic cleavage of many important transmembrane proteins such as amyloid precursor protein (APP), Notch, N-cadherin, syndecans and many others. It consists of four subunits including the catalytic presenilin (PSEN), nicastrin (NCT), PEN-2 and APH-1; among them PSEN and APH-1 occur as different isoforms. This implies that up to 6 different γ-secretase complexes may coexist in one cell, however the role of such heterogeneity is not known. Previous work of our group showed that different complexes have distinct localization within a cell and we hypothesize that this may be a basis for their different substrate specificity. In order to investigate it, we were studying the distribution and activity of EGFP-PSENs in stable cell lines and primary hippocampal neurons. In most cases of AD the pathophysiological mechanism remains unknown. The exception is the familial Alzheimer’s disease (FAD) when the mutations in either APP or in presenilins lead to abnormal accumulation of amyloid beta (Abeta). We were interested how the PSENs FAD mutations affect their distribution and substrate specificity and whether it could be connected to the increased aggregation of Abeta.

According to our earlier research, PSEN1 is distributed broadly throughout the cellular membranes and PSEN2 localizes to LE/LYS. A series of mutation experiments was performed to determine which parts of the two proteins determine the specific sorting. We found that a conserved E16RTSLM21 sequence within the N-terminus of PSEN2 is indispensable and sufficient for the transport of PSEN2 into LE/LYS (late endosomes/lysosomes). A search for interacting proteins revealed the AP-1 complex, but excluded the AP-2 or AP-3. In agreement with the literature the ERTSLM sequence can be phosphorylated and we observed that the phosphorylation of S19 negatively regulates interaction with AP-1 and the transport to LE/LYS. These functional results were confirmed by fluorescence microscopy and subsequently by correlative superresolution light and electron microscopy (SIM-CLEM). Using fluorescent organelle markers and the ultrastructure of EM, we could identify PSEN2 in late endosomes, multivesicular bodies and lysosomes and its phospho-dead S19A mutant also additionally in the TGN. In case of AD, the cells of interest are neurons; therefore we reproduced some experiments in primary rat hippocampal neurons. Regardless of the method used, in neurons PSEN1 was broadly distributed and PSEN2 restricted to somatodendritic compartment. This shows that the different compartmentalization of presenilins is also valid for neuronal cells and may have a functional impact on the pathophysiology of Alzheimer’s disease.

Since an enzyme always needs to meet its substrate in the same compartment, we hypothesized that heterogeneous distribution of PSEN1 and PSEN2 could be a basis for their diverse substrate specificity. We studied two lysosomal γ-secretase substrates: PMEL and TRP-1 and we observed accumulation of their CTFs upon selective downregulation of PSEN2. The enzymatic study (γ-secretase assay, GSA) including more ubiquitous substrates (N-cadherin, APP and Notch) revealed that whenever γ-secretase was restricted to LE/LYS it showed low processing of N-cadherin-CTF (a cell surface protein). In contrast, the broadly distributed γ-secretases showed normal cleavage of N-cadherin. We did not find differences in processing of the broadly distributed substrates (APP-CTF and NEXT). This confirms the hypothesis about localization-dependent substrate specificity. Because of the AD pathophysiology, we wondered how the distribution of PSEN1- and PSEN2-γ-secretase complexes affects production of Abeta; its levels were measured both inside and outside the cells using ELISA. Interestingly, PSEN-expressing cells PSEN2-produced Abeta was more likely to accumulate intracellularly in comparison to Abeta produced by PSEN1-expressing cells. We also measured that Abeta retained inside the cells showed higher Abeta42/40 ratio, which probably made it more prone to aggregation and highly relevant for Alzheimer’s disease.

The following experiments examined whether FAD mutations could change the subcellular localization of PSENs. Five well studied mutations of PSEN1 and four of PSEN2 were chosen for this study. Most of mutations did not affect localization, except two FAD-PSEN1 mutations (L166P and G384A), which caused a re-localization of GFP-PSEN1 into LAMP1-positive organelles and gave staining pattern indistinguishable from PSEN2. Unsurprisingly, these two mutations practically did not process cadherin. Several PSEN1-FADs decreased NICD production while not affecting AICD or N-cadherin ICD. In case of PSEN2, the mutations decreased processing of either N-cadherin-CTF and NEXT, or the APP-CTF. The effects of mutations on Abeta production were variable, but the ratio of Abeta42/40 was always increased, which is consistent with the literature. We found that this is mostly due to a drop of Abeta40 levels and that it was more pronounced intracellular than extracellular. We think that a change of Abeta42/40 ratio combined with a retention in LE/LYS compartment and different cleavage of other important proteins (e.g. N-cadherin, Notch) may contribute to the early onset and aggressive phenotype of some FAD mutations.

Our work shows that the γ-secretases of varying subunit composition are not only differently distributed inside a cell, but also distinguish between differently localized substrates. The results proof a causative relation between these two phenomena and contradict the dogma in the field that γ-secretases cleave indiscriminately many substrates. The major discoveries in our study are that (i) the compartmentalization of PSEN2 contributes to the intracellular pool of Abeta peptide much more than PSEN1; (ii) all studied FAD-PSEN2 mutations dramatically increase intracellular Abeta42/40, (iii) those FAD-PSEN1 mutations that phenocopy FAD-PSEN2 localization also cause the highest increase of Abeta42/40. All these features strongly add to the pathological relevance in the context of AD. The progressive accumulation of the intraneuronal Abeta has been found in AD mice models and in human AD brain. We think it could lead to further aggregation and finally to the appearance of tangles and amyloid plaques prominent in AD. For this reason, our work may be an important step on the way to explain the mechanism of Alzheimer’s disease. Currently, there is no drug available that could reverse or even halt the progression of AD. One of the main problems in drug development until now was the lack of selectivity of the inhibitory molecules. The generalized blocking of γ-secretase activity above a certain threshold leads to severe side effects because of the inhibition of Notch signaling. The work performed in frames of this project provides new inspiration and hope for the selective targeting of this part γ-secretase activity, which is involved in the pathological process. Beyond AD, RIP by γ-secretase is implicated in an increasing range of physiological processes and diseases. It is important to understand the mechanism behind the clinical manifestations, because in the long term it facilitates development of the innovative therapeutic approaches.

Project website:

contact details:

Prof. Wim Annaert
Laboratory of Membrane Trafficking
VIB Center for the Biology of Disease, KU Leuven
O&N 4, 6e verd
Campus Gasthuisberg
Herestraat 49, bus 602, 3000 LEUVEN
phone: +32 16 33 05 20

Dr. Paulina Ejsmont
Institut du Cerveau et de la Moelle épinière - ICM
Hôpital Pitié-Salpêtrière
47, bd de l'Hôpital - 75013 Paris