Altered expression of Presenilin2 impacts endolysosomal homeostasis and synapse function in Alzheimer’s disease-relevant brain circuits

Perdok, Anika; Van Acker, Zoë P.; Vrancx, Céline; Sannerud, Ragna; Vorsters, Inge; Verrengia, Assunta; Callaerts-Végh, Zsuzsanna; Creemers, Eline; Gutiérrez Fernández, Sara; D’hauw, Britt; Serneels, Lutgarde; Wierda, Keimpe; Chávez-Gutiérrez, Lucía; Annaert, Wim

doi:10.1038/s41467-024-54777-y

Download PDF

Article
Open access
Published: 29 November 2024

Altered expression of Presenilin2 impacts endolysosomal homeostasis and synapse function in Alzheimer’s disease-relevant brain circuits

Nature Communications volume 15, Article number: 10412 (2024) Cite this article

7447 Accesses
5 Citations
89 Altmetric
Metrics details

Subjects

Abstract

Rare mutations in the gene encoding presenilin2 (PSEN2) are known to cause familial Alzheimer’s disease (FAD). Here, we explored how altered PSEN2 expression impacts on the amyloidosis, endolysosomal abnormalities, and synaptic dysfunction observed in female APP knock-in mice. We demonstrate that PSEN2 knockout (KO) as well as the FAD-associated N141IKI mutant accelerate AD-related pathologies in female mice. Both models showed significant deficits in working memory that linked to elevated PSEN2 expression in the hippocampal CA3 region. The mossy fiber circuit of APPxPSEN2KO and APPxFADPSEN2 mice had smaller pre-synaptic compartments, distinct changes in synaptic vesicle populations and significantly impaired long term potentiation compared to APPKI mice. At the cellular level, altered PSEN2 expression resulted in endolysosomal defects and lowered surface expression of synaptic proteins. As PSEN2/γ-secretase is restricted to late endosomes/lysosomes, we propose PSEN2 impacts endolysosomal homeostasis, affecting synaptic signaling in AD-relevant vulnerable brain circuits; which could explain how mutant PSEN2 accelerates AD pathogenesis.

A unique subpopulation of wild-type neurons recapitulating familial Alzheimer’s disease phenotypes

Article Open access 09 August 2025

A novel missense variant in ACAA1 contributes to early-onset Alzheimer’s disease, impairs lysosomal function, and facilitates amyloid-β pathology and cognitive decline

Article Open access 31 August 2021

Mutational analysis in familial Alzheimer’s disease of Han Chinese in Taiwan with a predominant mutation PSEN1 p.Met146Ile

Article Open access 13 November 2020

Introduction

γ-Secretases are intramembrane proteases that act as proteolytic switches in various physiological processes through the targeting of a broad range of substrates^1,2. They are extensively studied^3,4,5 in the context of Alzheimer’s disease (AD) where they target the amyloid precursor protein (APP)^3,6. γ-Secretases are tetrameric complexes where presenilin (PSEN1 or PSEN2) provides the catalytic activity and nicastrin (NCSTN), anterior pharynx defective 1 (APH1A or B) and presenilin enhancer 2 (PSENEN2) are essential subunits^7,8,9. The different isoforms of APH1 and PSEN give rise to four distinct complexes¹⁰. After APP ectodomain shedding by β-secretase (BACE1), γ-secretases release the APP intracellular domain (AICD), followed by progressive C-terminal trimming releasing amyloid β (Aβ) peptides of varying lengths^11,12,13. Mutations in the APP, PSEN1 and PSEN2 genes cause rare autosomal dominant familial AD (FAD) centralizing γ-secretase and APP processing in AD, as originally theorized in the amyloid cascade hypothesis^14,15. Pathogenic mutations destabilize γ-secretase-substrate interactions, leading to longer, aggregation-prone Aβ peptides¹², the major constituents of amyloid plaques observed in AD brains.

AD research has predominantly focused on PSEN1/γ-secretase due the far more abundant pathogenic mutations found in its gene⁹ and that PSEN1 knock-out (KO) is embryonic lethal^16,17. PSEN2KO mice develop normally, with only a minor pulmonary phenotype^18,19. However, knocking out PSEN2 in a PSEN1KO model puts forward embryonic lethality²⁰. Post-embryonic full knock out of γ-secretase, obtained by conditionally knocking out all three APH1 isoforms, results in C-terminal fragment (CTF) accumulation of various substrates, neurodegeneration and neuroinflammation²¹, underscoring a central role for γ-secretase in these processes. Studies in PSEN1 and PSEN2 double KO linked γ-secretase to tau phosphorylation, axonal transport, learning and memory, neurodegeneration and impaired synaptic plasticity^{22,23,24,25,26}, though individual complex contributions remained understudied. Notably, compared to PSEN1, FAD causing mutations in PSEN2 have a later age of onset²⁷ and largely shift processing towards the Aβ42 lineage^28,29; PSEN1 mutations have a broader range of effects on Aβ profiles³⁰.

Clear molecular differences between PSEN1 and PSEN2 complexes were recently brought up through the discovery of a unique acidic-dileucine sorting motif that targets PSEN2-complexes to late endosomes and lysosomes (LE/Lys)^29,31. As PSEN1/γ-secretases are more broadly distributed at the cell surface and in (recycling) endosomes, this results in overall complex-selective substrate pools. For instance, PSEN2KO impairs melanosome maturation by failing to process tyrosinase-related protein (TRP1) and premelanosome protein (PMEL)-CTFS²⁹. PSEN2/γ-secretase also generates intracellular Aβ (iAβ) in LE/Lys²⁹, a highly toxic pool that surges in early stages of the disease, preceding plaque deposition and correlating with synaptic deficits^32,33,34,35. Aβ42, in particular, accumulated in multivesicular bodies (MVBs) (also referred to as LE), wherein the acidic environment strongly accelerated amyloid fibril formation^36,37,38 affecting LE/Lys integrity and representing an early event of Aβ pathogenesis^39,40. However, much of this knowledge is based on in vitro studies of aggregation kinetics or cellular uptake of exogenously administered recombinant Aβ peptides. As PSEN2/γ-secretase significantly contributes to iAβ accumulation in LE/Lys, in vitro and in vivo models with altered PSEN2 expression can now be used to more directly study the impact of iAβ. Given its LE/Lys localization, we hypothesized that changes in PSEN2 function, through KO or FAD-linked mutations, may affect Aβ pathogenesis, potentially through altering iAβ and APP-CTF levels. FAD-linked PSEN2 mutants impair intracellular Ca²⁺ stores⁴¹ and are linked to decreased Rab7 recruitment on autophagosomes, connecting Ca²⁺ signaling to autophagy and degradation⁴². Endolysosomal abnormalities appear early in disease as has been shown in murine AD models^43,44 and patient brain samples⁴⁵. Neurons, being post-mitotic polarized cells, rely heavily on the endolysosomal system to remove dysfunctional organelles and aggregated proteins next to cargo trafficking^46,47.

To address how PSEN2 contributes to AD pathogenesis, we crossed APPKI mice with a PSEN2KO or a CRISPR/Cas9 engineered PSEN2 KI mouse harboring the FAD-linked N141I mutation²⁸. In both models, AD-related amyloid pathology, including dystrophic neurites, appeared accelerated along with cognitive defects in the working memory. Notably, long-term potentiation (LTP) was impaired in the hippocampal CA3 region, correlating with morphological changes in mossy-fiber pyramidal neuron synapses. Primary hippocampal neuron data further imply a role for PSEN2 in endolysosomal homeostasis, disruption of which impacts synaptic functioning and circuitry, unveiling molecular mechanisms that may become distorted particularly in FAD-associated PSEN2 variants.

Results

PSEN2 knockout and FADPSEN2 differently affect plaque pathology and Aβ profiles

Given its role in generating neurotoxic iAβ, we hypothesized that altering PSEN2 expression in vivo, through KO or introducing a FAD-mutation, would have opposing effects on AD pathogenesis due to the changes in the iAβ pool. We used CRISPR/Cas9 editing to introduce the PSEN2 Volga mutation in C57black6J mice (Figure S1a) by converting asparagine at position 141 to isoleucine, the most prevalent PSEN2 FAD mutation^27,28 known to shift iAβ production to longer more aggregation-prone species²⁹. We created, as such, homozygous mutant mice with the N141I point mutation, as confirmed by PCR (Figure S1a, b) and Sanger sequencing (Figure S1c). These mice, which show no overt phenotypic differences, have higher endogenous mutant PSEN2 levels (Figure S1d), suggesting more stable PSEN2/γ-secretase levels or a broader distribution. FADPSEN2 and PSEN2KO mice¹⁸ were crossed with the APP NL-G-F KI (APPKI) AD model to investigate AD pathology (hereafter referred to APPxPSEN2KO and APPxFADPSEN2)⁴⁸.

We previously demonstrated that PSEN2/γ-secretase activity in LE/Lys strongly determined the iAβ pool, with FAD mutations shifting this pool towards longer Aβ42²⁹. To confirm this, we analyzed conditioned media and cell extracts from primary hippocampal neurons derived from these models. APPxFADPSEN2 neurons showed increased total iAβ, whereas this pool is decreased in APPxPSEN2KO (Fig. 1a, b). Secreted Aβ was correspondingly decreased in APPxFADPSEN2 and increased in APPxPSEN2KO (Fig. 1c), likely due to the absence of PSEN2 driving APP processing to PSEN1/γ-secretase, and increasing the secreted pool. PSEN1 expression remained stable over time, whereas PSEN2 levels increased with maturing neurons, further augmenting iAβ, and underlining PSEN2’s role in iAβ generation (Fig. 1a). Next, using MesoScale Discovery (MSD) ELISA, we quantified the relative levels of Aβ37, Aβ38, Aβ40 and Aβ42 in intracellular and secreted pools. As expected¹², the PSEN2 N141I mutation significantly increased relative Aβ42 levels and Aβ42:40 ratio (Fig. 1d–g), indicating decreased processivity. In contrast, PSEN2KO in the APPKI background shifted the Aβ profile to shorter peptides, with significant increased Aβ38 and decreased Aβ42:40 ratio, reflecting PSEN1 processivity (Fig. 1d–g). Similar results were observed in brain extracts: no differences were seen in the insoluble fraction (Figure S1e), but soluble extracts at 6 months of age showed increased Aβ38 and Aβ42 for APPxPSEN2KO and APPxFADPSEN2, respectively (Fig. 1h), consistent with altered γ-secretase processivity and 42:40 ratio (Fig. 1 i, j).

**Fig. 1: Altering PSEN2 expression gives rise to distinct Aβ profiles.**

Next, we examined how altered Aβ profiles impacted amyloid pathology, using thioflavin-S positive area as a readout. To note, wild-type mice were not included in these analyses as they do not develop amyloid pathology⁴⁹. Indeed, the murine Aβ sequence differs from the human one by three amino acids at N-terminal positions 5, 10, and 13 and therefore does not readily aggregate. Curiously, compared to APPKI mice, both APPxPSEN2KO and APPxFADPSEN2 showed an increased plaque deposition in both the cortex and hippocampus from 9 months onwards, with considerable larger plaque area by 18 months of age (Fig. 2a, b, Figure S1f, g). However, clear differences were still observed: plaque pathology was significantly higher in APPxFADPSEN2 at 6 months, while in APPxPSEN2KO, this was only after 9 months. In the cortex, plaque deposition in APPxFADPSEN2 began as early as 3 months. Additionally, APPxPSEN2KO mice showed mainly larger plaques, while APPxFADPSEN2 had more numerous plaques (Fig. 2d, e). Lastly, genotype-specific morphological differences were also apparent. APPKI mice had a common type of plaque consisting of a dense thioflavin-S positive core with a surrounding halo of fibrillar Aβ positive staining, suggested to act as a reservoir for toxic Aβ oligomeric species (Fig. 2c)⁵⁰. In contrast, APPxPSEN2KO plaques displayed a large intensely stained Aβ halo surrounding the dense core, while APPxFADPSEN2 presented with a smaller fibrillar Aβ halo largely coinciding with the core, likely linked to Aβ42 shifts from the PSEN2 N141I mutation, and reflecting distinct Aβ profiles and altered aggregation properties between both genotypes compared to APPKI.

**Fig. 2: PSEN2 deficiency and PSEN2 N141I knock-in both accelerate AD pathology in an APPKI model.**

Another major hallmark of AD pathology is the accumulation of dystrophic neurites surrounding plaques, identifiable by the accumulation of LAMP1-positive LE/Lys being stalled in neurites around plaques. In both hippocampus and cortex of APPxFADPSEN2 mice, LAMP1 immunostaining increased significantly by 3 months compared to APPKI (Fig. 2f, g, Figure S2a–c), preceding plaque pathology (compare with Fig. 2b). APPxPSEN2KO showed no significant LAMP1 increase compared to APPKI mice, except at 9 months in the cortex, at levels similar to APPxFADPSEN2 (Figure S2b, c). Transmission EM (TEM) confirmed larger accumulations of degradative compartments in both APPxPSEN2KO and APPxFADPSEN2 compared to APPKI (Figure S3). APP-CTFs co-accumulated with LAMP1 in dystrophic neurites across all stages and genotypes, echoing previous studies (Fig. 2f, Figure S2a–d)⁵¹. Overall, both PSEN2KO and FADPSEN2 accelerated plaque pathology in APPKI mice, but impacted plaque morphology and dystrophic neurite accumulation differently, pointing to distinct mechanisms likely driven by altered γ-secretase processivity and iAβ pool compositions.

Altering PSEN2 expression in APP NL G F mice impacts working memory

Since APPxPSEN2KO and APPxFADPSEN2 both displayed aggravated AD pathology, we anticipated worsened cognitive deficits. To assess this, we conducted behavioral and memory tests on six month old mice. While circadian rhythm remained unchanged over 24 h, significant differences in activity levels were found (Fig. 3a, b). APPxPSEN2KO mice exhibited hypoactivity compared to both the wild-type and APPKI mice, whereas APPxFADPSEN2 mice showed hyperactivity during the night (Fig. 3b). This is in line with previous studies linking hyperactivity in mice with AD pathology⁵².

**Fig. 3: PSEN2 deficiency and FAD-linked mutant PSEN2 lead to working memory deficits.**

No differences in locomotion or social memory were observed across genotypes using open field, elevated plus maze and social preference/social novelty (SPSN) tests (Figure S4a–e). However, APPxPSEN2KO mice displayed anxiety, spending less time in open spaces (Figure S4a, b). Therefore, we focused on spatial and working memory, both known to be impaired in AD⁵³. Spatial memory was evaluated using the Morris water maze (Fig. 3c). Considering we observed clear differences in swim speed, time needed to find the platform could not be used as a read out of cognitive function (Figure S4f, g). However, there were no significant differences in distance to the platform during training (Fig. 3d). Moreover, as the time spent in the target quadrant during the probe trial was considerably greater than chance for all genotypes, we concluded that both learning and spatial memory are not significantly different between genotypes (Fig. 3e). Working memory was tested using the delayed matching to platform (DMTP), a variant of the Morris water maze where platform location changed daily (Fig. 3f). APPKI mice showed a significant decrease only by the final swim, underscoring early deficits in working memory compared to wild-type mice (Fig. 3g, h). Neither APPxPSEN2KO nor APPxFADPSEN2 mice were able to reduce their swimming distances throughout trials, indicating an earlier and more severe working memory impairment compared to wild-type and APPKI mice (Fig. 3i, j).

Working memory deficits correlate with defects in mossy fiber-CA3 hippocampal circuitry

Given the distinct amyloid pathology in the hippocampus of APPxPSEN2KO vs APPxFADPSEN2 mice (Fig. 2a, b) and PSEN2-related working memory deficits, we wondered whether this could stem from brain region-specific PSEN2 expression differences. Since PSEN2 also accumulated in LE/Lys in dystrophic neurites around plaques in our AD models, we first examined its localization in wild-type mice (Figure S5a). Notably, endogenous PSEN2 was more highly expressed in the CA3 and subiculum regions of the hippocampus (Fig. 4a, Figure S5b). Quantitative comparison of the CA3 and CA1 areas confirmed the higher PSEN2 levels in the CA3 (Fig. 4b). This was recapitulated in FADPSEN2, with overall higher expression in the cortex, as noticed by western blotting (Figure S1d). Co-staining with a mossy fiber-specific marker, zinc transporter 3 (ZNT3), revealed PSEN2-positive cells near mossy fiber synapses within the CA3 region in wild-type and mutant PSEN2 mice (Fig. 4c) The hippocampus receives input through the entorhinal cortex, signaling to the dentate gyrus, and then via mossy fibers (MFs) to CA3 pyramidal neurons, which play a role in working memory⁵⁴. As PSEN2 is highly expressed in CA3 pyramidal cells, we investigated long-term potentiation (LTP) in APPxPSEN2KO and APPxFADPSEN2 mice (Fig. 4d). Consistent with previous reports, using a mild LTP induction procedure, APPKI mice displayed a modest reduction in LTP amplitude and slope⁵². Interestingly, LTP was further reduced in acute brain slices of both APPxPSEN2KO and APPxFADPSEN2 mice (Fig. 4e–g) in line with observed working memory deficits (Fig. 3i, j), and tying PSEN2 expression to hippocampal circuitry and signaling. Specifically, the data suggest a defect in late phase LTP, implying issues in local and/or somatic de novo protein replenishment^55,56. Additionally, the amplitude distribution of input-output (IO) curves were consistently larger in APPxPSEN2KO and APPxFADPSEN2 compared to control and APPKI mice (Figure S5c, d), indicating potential differences in synaptic composition/function.

**Fig. 4: CA3 localized PSEN2 is required for hippocampal circuitry and LTP.**

To exclude a global hippocampal deficit, we measured LTP in the CA1 region and found no significant differences between genotypes (Figure S5e–g). We then explored short-term presynaptic plasticity using paired-pulse stimulations in the CA1 and CA3 regions. APPxPSEN2KO mice showed increased synaptic facilitation in the CA1, while in the CA3, PSEN2KO rescued the lowered response seen in APPKI to wild-type levels. In contrast, APPxFADPSEN2 mice displayed clear synaptic depression in the CA3 circuit, with normal responses in CA1 (Fig. 4h-k). To investigate the effects on presynaptic dynamics during periods of more demanding synaptic activity, we performed train stimulations in both hippocampal regions. In line with the paired pulse data, we observed increased synaptic facilitation in the APPxPSEN2KO predominantly in CA1 and partially in CA3, while APPxFADPSEN2 mice showed clear synaptic depression, exclusively in the CA3 circuit (Fig. 4l–o, Figure S5h). Taken together, these results indicate alterations to synaptic efficacy in both the APPxPSEN2KO and APPxFADPSEN2 mice, likely involving presynaptic changes in the organization of the synaptic vesicle pool (see below). In case of the APPxFADPSEN2 this is more specific to the CA3 region whereas for the APPxPSEN2KO this is observed in both CA3 and CA1 but without impacting LTP in the CA1.

We next used TEM to investigate morphological changes in the MF-CA3 circuit. These structurally distinct synapses consist of a large presynaptic axons synapsing onto multiple CA3 pyramidal neurons⁵⁷. They can be easily distinguished ultrastructurally through the presence of large, dense clusters of synaptic vesicles, as seen in wild-type CA3 regions (Fig. 5a, Figure S5i). While there was a moderate, non-significant decrease in synapse area in APPKI CA3 regions, significant decreases were noticed in APPxPSEN2KO and APPxFADPSEN2 mice (Fig. 5b). Post-synaptic densities (PSDs) were also significantly reduced in number and length across AD genotypes, compared to wild-type (Fig. 5c-d), with APPxFADPSEN2 having the largest decrease in PSD number (Fig. 5c) and APPxPSEN2KO the largest decrease in PSD length per synapse (Fig. 5d). At higher resolution, synaptic vesicle clusters displayed distinct morphological features (Fig. 5f, Figure S5i). They did not differ between wild-type and APPKI, but synaptic vesicle cluster density was significantly higher in APPxPSEN2KO while lower or more sparse in APPxFADPSEN2 synapses (Fig. 5e); correlating with electrophysiological and LTP differences (Fig. 4). Additionally, both APPxPSEN2KO and APPxFADPSEN2 synapses showed higher heterogeneity in vesicle size through the presence of larger vesicles (white arrowheads, in Fig. 5f, Figure S5i). Overall, these results highlight both pre- and post-synaptic changes linked to altered PSEN2 expression in MF-CA3 synapses. To investigate underlying molecular mechanisms, we turned to primary hippocampal neurons derived from E17 APPKI, APPxPSEN2KO and APPxFADPSEN2 embryos.

**Fig. 5: Altering PSEN2 expression affects ultrastructural morphology of mossy fiber synapses.**

PSEN2-related synaptic defects are recapitulated in primary hippocampal neurons

First, we aimed to determine if PSEN2/ƴ-secretase complexes are in close proximity of synapses. Super-resolution imaging revealed small PSEN2-positive LE/Lys (marked by LAMP1) near the post-synaptic marker, PSD95, in APPKI and APPxFADPSEN2 neurons (Fig. 6a, white arrowheads). Quantification showed no significant difference in the overlap of PSD95 with PSEN2 positive LE/Lys (Fig. 6c), indicating that the FAD mutation did not affect their postsynaptic abundancy. Compared to wild-type, APPKI exhibited a significant decrease in both number and size of PSD95-positive post-synapses, which was recapitulated in APPxPSEN2KO (Fig. 6d, e, Figure S6d). In APPxFADPSEN2, only size of post-synapses was significantly decreased. Even though PSEN2 is mainly in dendrites, it was also found on LAMP1-positive vesicles in axons and near synaptobrevin2-positive pre-synapses, highlighting the presence of PSEN2 in both pre- and post-synaptic compartments (Fig. 6b, f, Figure S6a, 6c).

**Fig. 6: PSEN2 expression in LE/Lys is required to maintain pre- and postsynaptic homeostasis.**

Synaptic plasticity relies on coordinated synaptic signaling through pre-synaptic neurotransmitters binding to post-synaptic receptors. Two essential proteins are the presynaptic vGLUT1, a glutamate transporter involved in synaptic vesicle refilling, and the postsynaptic GLUA1, a subunit of the AMPA receptor complex crucial for LTP⁵⁸. We observed no significant differences in the total levels of either receptor, or in the phosphorylation state of GLUA1 (Fig. 6g–i, Figure S6b). However, surface levels of vGLUT1 and GLUA1 were significantly decreased in APPxPSEN2KO and APPxFADPSEN2 neurons compared to APPKI (Fig. 6j–l, Figure S6e). Considering total levels remained unchanged (Fig. 6g), this suggests a PSEN2-dependent recycling defect, potentially causing the observed synapse morphology and LTP defects, and contributing to working memory impairments in these PSEN2 AD mouse models. Given that PSEN2/γ-secretase is largely restricted to LE/Lys²⁹, and present both pre- and post-synaptically, we hypothesize that LE/Lys dysregulation may underpin these synaptic alterations.

Altered PSEN2 expression causes lysosomal defects in neurons

Synapses are highly active compartments relying heavily on the endolysosomal system for proper degradation, recycling and trafficking of synaptic receptors⁵⁹. Previous studies indicated that Aβ42 preferentially concentrated in LE/Lys compartments^60,61,62 affecting their integrity. Also aberrant levels of the γ-secretase substrate, APP β-CTF, cause endosomal to lysosomal dysfunctions, observed for APP and PSEN1 mutations^63,64, and in Down Syndrome⁶⁵. To explore APP fragment accumulation in LE/Lys of our PSEN2 AD models, we performed co-immunostaining with 82e1, recognizing both APP β-CTF and Aβ, and Y188 antibodies, which targets full-length APP and APP-CTFs. Organelles positive for both antibodies identify APP β-CTF buildup, whereas 82e1-only organelles point to Aβ accumulation. Significantly more 82e1/Y188-positive LE/Lys were found in hippocampal neurons of APPxPSEN2KO and APPxFADPSEN2 mice, compared to APPKI, denoting APP fragment accumulation (Fig. 7a–d). As expected, absence of PSEN2 expression resulted in more frequent co-apposition of 82e1 and Y188, underscoring APP-βCTF buildup in LAMP1-positive LE/Lys (Fig. 7a, white arrowheads). Conversely, APPxFADPSEN2 neurons showed more intense 82e1 immunostaining, with 82e1-only immunoreactivity more common in LAMP1-positive LE/Lys, confirming increased Aβ levels (Fig. 7a, orange arrowheads). There was also significantly more overlap between LAMP1-positive LE/Lys and Y188/82e1 in APPxPSEN2KO and APPxFADPSEN2, solidifying accumulation of APP fragments (Fig. 7c). Finally, western blot analysis also revealed higher APP-CTFs levels in APPxPSEN2KO and APPxFADPSEN2 neurons (Fig. 7b, d). Of note, immunostaining for APP fragments was not restricted to LE/Lys indicating accumulation in other endosomal compartments. In summary, both APPxPSEN2KO and APPxFADPSEN2 neurons presented significant and distinct accumulation of APP fragments that may disrupt endolysosomal homeostasis, as reported in other models (see above).

**Fig. 7: Endolysosomal imbalance in hippocampal neurons deficient for PSEN2 or expressing the PSEN2 N141I mutant.**

Not surprisingly, given the preferred location of PSEN2 in LE/Lys, APPxPSEN2KO and APPxFADPSEN2 hippocampal neurons presented with significantly increased LAMP1-positive organelles compared to APPKI neurons (Fig. 7e, f). There was a moderate increase in early endosomal compartments (Fig. 7g, EEA1 + ), with no differences in recycling endosomes (Fig. 7h, VPS35 + ). More frequent overlap of all three markers in APPxPSEN2KO and APPxFADPSEN2 neurons (Fig. 7e, white arrowheads), suggested partial endolysosomal collapse. To investigate lysosomal dysfunction, we assessed Ca²⁺ storage, crucial for lysosomal fusion and fission⁶⁶. Treatment with the TRPML1 agonist, MLSA1⁶⁷, resulted in a significantly lower Ca²⁺ release from LE/Lys of APPxPSEN2KO and APPxFADPSEN2 neurons (Figure S6f, g). To corroborate this, we allowed neurons to endocytose dextran-488 and the Ca²⁺ sensitive rhodamine-dextran, that, when chased to LE/Lys allowed us to directly visualize basal lysosomal Ca²⁺ levels. In APPxPSEN2KO and APPxFADPSEN2 neurons, rhodamine intensity in dextran-488-positive vesicles is clearly decreased, confirming lowered lysosomal Ca²⁺ content compared to APPKI neurons (Fig. 7i, j). This was coupled to a substantial decrease in lysosomal exocytosis, as shown by surface localized LAMP1 immunostaining (Fig. 7k, l). Next, neurons were transfected with the dual labelled LC3 allowing to distinguish autophagosomes from autolysosomes (Fig. 7m). Both organelles were more abundant in APPxPSEN2KO and APPxFADPSEN2 neurons, demonstrating an imbalanced autophagy, and including a congestion in the degradative capacity, evidenced by the significant increase in autolysosomes (Fig. 7m–p). These findings link the aberrant accumulation of APP metabolic fragments, due to either PSEN2KO or FADPSEN2, to impaired LE/Lys and autophagic functions, which could in turn affect synaptic receptor recycling/trafficking underlying the observed synaptic defects.

Discussion

γ-Secretases have been central to AD pathology for decades. However, understanding the impact of full γ-secretase inactivity, either pharmacologically⁶⁸ or genetically⁶⁹, on cell functioning does not reflect disease conditions. How altered function of individual complexes contributes to disease remains unknown and filling in this knowledge gap is key to understand disease progression. Here, we studied how PSEN2 deficiency versus a FAD-associated PSEN2 N141I KI mutation impacts AD pathogenesis in APPKI mouse models. Importantly, the use of an APPKI mouse model was necessary to address this question since, as previously mentioned, wild-type mice do not develop Aβ pathology⁴⁹. Despite distinct Aβ profiles, both PSEN2 genotypes accelerated AD pathology, evidenced by earlier plaque deposition, more dystrophic neurites and more severe memory deficits (Fig. 8). These defects correlated with increased PSEN2 expression in the hippocampal CA3, lower LTP in the mossy fiber circuit and ultrastructural changes such as smaller presynaptic boutons, shorter or fewer PSDs and altered synaptic vesicle pools. Supporting in vivo data, primary hippocampal neurons from these models showed decreased surface expression of presynaptic vGLUT1 and postsynaptic GLUA1 in APPxPSEN2KO and APPxFADPSEN2 mice. We hypothesize that endolysosomal dysregulation underlies these synaptic alterations. Knocking-out or mutating PSEN2 leads to aberrant accumulation of APP metabolic products in LE/Lys, affecting lysosomal homeostasis (e.g., reduced lysosomal Ca²⁺) resulting in decreased lysosomal exocytosis and a degradative bottleneck indicated by accumulating LC3-positive autophagosomes and autolysosomes. Our data underscore that PSEN2/γ-secretase maintains LE/Lys homeostasis, essential for synaptic integrity, especially in areas with higher PSEN2 expression, like the CA3.

**Fig. 8: Altered PSEN2 expression impacts on synapse function and neuronal endolysosomal homeostasis.**

The accelerated amyloid pathology in APPxFADPSEN2 brains correlated with decreased γ-secretase processivity and a toxic gain-of-function by producing more Aβ42⁷⁰. Not surprisingly, compared to APPKI, we observed denser thioflavin⁺ plaques, composed mostly of Aβ42 with a filamentous Aβ halo⁷¹. This compacted halo contrasted with the larger, diffuse plaques in APPxPSEN2KO brain, which correlated with increased processivity and shorter, less aggregation-prone Aβ peptides. The absence of PSEN2 directs APP metabolism toward PSEN1/γ-secretase, at the cell surface and in early/recycling endosomes. Thus, these diffuse plaques likely originate from secreted relatively shorter Aβ, as increased Aβ secretion was observed in APPxPSEN2KO primary neurons (Fig. 1). In contrast, APPxFADPSEN2 brains and hippocampal neurons showed decrease secreted Aβ, compensated by intracellular Aβ42 accumulation, consistent with observations in transgenic overexpression models^72,73. This provides in vivo support for PSEN2/γ-secretase as the main generator of iAβ, which significantly increases in FAD settings, whilst PSEN1 preferentially contributes more to secreted Aβ²⁹. Secondly, the APPxPSEN2KO data confirm that PSEN2 inherently favors the production of longer Aβ peptides whereas PSEN1 complexes produce shorter ones, in line with human iPSC studies⁷⁴.

The arctic mutation in the APPKI model enhances protofibril generation of Aβ40 peptides⁷⁵, meaning fibrils in this model may not fully represent human pathology⁷⁶. Therefore, the observed alterations in amyloid pathology may be in part due to the presence of this mutation. However, distinct plaque types, from diffuse to fibrillar and dense core, are also observed in human brains^77,78; including in sporadic and certain FAD cases^79,80. As plaque fibrils result from Aβ self-assembly which is dependent on peptide length and other modifications⁷⁸, altering γ-secretase processivity profiles, as seen in our PSEN2 mutant lines, may indeed lead to altered plaque morphologies in a physiologically more relevant AD context.

Noting the clear exacerbation of pathology, we behaviorally analyzed these mouse models. Behavioral studies of the APP NLGF model have produced conflicting results, with some reporting memory impairments at 6 months whilst others did not^48,81,82,83. In our paradigms, spatial, social and learning memory appeared unaffected for all models, including APP NLGF, compared to wild-type. However, altering PSEN2 levels in this APPKI background clearly led to specific behavioral changes. When observed for 24h, APPxPSEN2KO mice appeared hypoactive without disrupted circadian rhythm, whilst APPxFADPSEN2 mice exhibited hyperactivity. Hyperactivity in mice suggests hyperactive neural circuits⁸⁴, an early sign in disease progression due to altered functional connectivity^85,86. This hyperactivity has been shown in several FADPSEN2 transgenic models, particularly in the hippocampus^87,88,89,90 and increasing susceptibility to seizures⁸⁶, a symptom also observed in AD patients with PSEN2 FAD mutations. In contrast, PSEN2KO mice are more resistant to seizure kindling, especially at younger ages^91,92. Much of the FAD-linked hyperexcitability has been attributed to toxic Aβ species, either through direct binding to synaptic receptors⁹³, reducing inhibitory synapses^94,95 or modulating excitatory signaling^96,97,98.

We observed a specific working memory deficit linked to PSEN2 deficiency or FAD-linked KI mutation, compared to only slight impairments in the APPKI mice. Working memory is closely associated with CA3 and resident mossy fiber-pyramidal cell synapses, which are impaired early in AD^99,100,101. Our data also show higher expression of PSEN2 within CA3 circuits, implicating it in working memory (Fig. 4). Altering PSEN2 expression reduced LTP in the mossy fiber circuit, compared to wild-type mice, indicating significant changes in synaptic function in APPxPSEN2KO and APPxFADPSEN2 models. At the ultrastructural level, synaptic vesicle clusters were more dense in APPxPSEN2KO and less dense in APPxFADPSEN2 mossy fiber synapses, highlighting the need for normal PSEN2 expression to maintain synaptic vesicle homeostasis. Mossy fiber synapses showed significant post-synaptic defects in all three genotypes compared to wild-type, but exacerbated in PSEN2-altered models. This defect between APPKI and wild type is not the case for the pre-synaptic compartment possibly indicating that mossy fiber PSDs are more vulnerable in disease. The effect of altered PSEN2 expression on pre-synapses is in line with reports that PSEN ablation mainly affects presynaptic components. PSENs have been studied extensively in synaptic signaling, with significant impairments observed when conditionally deleting both PSENs. Herein, LTP deficits were seen only in CA3, but not CA1 region of PSEN double KO, indicating a pre-synaptic defect, likely mediated through decreased intracellular Ca²⁺ affecting the replenishment of the ready releasable pool of synaptic vesicles^{69,102,103,104,105}. We noted LTP deficits solely in the CA3 region regardless of PSEN2KO or mutant PSEN2. In line with the alterations in synaptic vesicle clustering, APPxPSEN2KO showed synaptic facilitation in both paired-pulse and train stimulation paradigms, while APPxFADPSEN2 showed synaptic depression only upon CA3 stimulation. Though PSEN2KO and FADPSEN2 have similar effects on LTP, this occurs likely through different mechanisms that warrant further investigation.

In primary hippocampal neurons, altered PSEN2 expression lowered surface levels of presynaptic VGLUT1 and postsynaptic AMPARs, underscoring PSEN2-dependent defects in receptor recycling and degradation. Similarly, PSEN2 expression significantly affected LE/Lys homeostasis. Compared to APPKI neurons, PSEN2 deficiency and FADPSEN2 decreased LE/Lys basal Ca²⁺ levels, reduced TRPML1-dependent Ca²⁺ release and exocytosis, and led to accumulation of LC3⁺ autophagosomes, as a proxy for defects in autophagy-lysosomal degradation. We believe this dysfunction stems from aberrant accumulation of APP metabolic products, i.e. APP-CTF and Aβ, in LE/Lys. While excess toxic Aβ42 impairs synaptic function^106,107,108, the opposite effects on γ-secretase processivity between PSEN2KO and FADPSEN2, argues for additional mechanisms. Notably, both genotypes exhibited increased levels of APP-CTFs, increasingly connected to endolysosomal dysfunctions in AD pathology. Studies in iPSC-derived neurons harboring FAD-mutations in APP and/or PSEN1 identified APP-βCTF as a key perpetrator in endolysosomal and autophagic abnormalities^{63,64,109,110}. Further, phosphorylated APP-βCTF was found to also disrupt proton pump through binding to the V0a1 subunit⁶⁵. More recently, we found that excess APP-CTF impairs lysosomal Ca²⁺ refilling from the ER through LE/Lys-ER contact sites, explaining the decreased lysosomal Ca²⁺ in PSEN deficient models^67,111 and in this study.

Endolysosomal defects, as those in our models, are strongly implicated in synaptic integrity. In lysosomal storage diseases, LE/Lys dysfunction impairs presynaptic integrity by disrupting lysosome-autophagy degradation and as such, synaptic vesicle recycling¹¹². This is reflected in decreased synaptic vesicle clustering¹¹³; a feature also seen in α-synuclein overexpression models for Parkinson’s disease¹¹⁴ and reminiscent of FADPSEN2 MF synapses (Fig. 5). Both APPxPSEN2KO and APPxFADPSEN2 presented with similar lysosomal and autophagy defects. However, in a PSEN2KO background this resulted in more densely clustered synaptic vesicles, whereas more sparse in FADPSEN2 synapses. The appearance of larger vesicles may suggest underlying recycling defects in APPxPSEN2KO and APPxFADPSEN2 neurons as well. At the post-synapse, lysosomal Ca²⁺ release and lysosomal exocytosis, both decreased in PSEN2KO and FADPSEN2 neurons, are crucial for long-term structural plasticity, coupling GLUA1 activation to postsynaptic LE/Lys Ca²⁺ stores^115,116. Similarly, lysosomal Ca²⁺ regulates dendritic lysosome availability; defects herein impacted GLUA1 trafficking and lysosomal degradation, thus affecting LTP and synaptic plasticity^117,118. Further, Rab7-dependent transport from LE to somatic lysosomes is required for AMPAR degradation, and the partitioning of AMPARs to LE/Lys instead of recycling determines synaptic depression upon LTD induction^119,120,121. Interference with RAB7/LE function affects dendritic and axonal maintenance, with Rab7 mutations linked to Charcot-Marie Tooth disease type 2B resulting in axonal neuropathy¹²².

PSEN2 has been less studied in AD pathogenesis compared to the more prevalent mutated PSEN1 and APP genes. Our study revealed the similar detrimental effects PSEN2KO and FADPSEN2 have on amyloid pathology and the endolysosomal system, with subsequent impacts on synaptic signaling (Fig. 8). PSEN2 deficiency exacerbated amyloid pathology, suggesting that PSEN2 expression might be protective in AD. In sporadic AD, splice variants in PSEN2, but not PSEN1, lead to truncated/lowered transcripts^123,124, indicating that PSEN2 loss-of-function may contribute to late onset AD. The worsening of amyloid pathology by PSEN2KO argues against therapeutically targeting individual γ-secretase complexes, but also that reducing iAβ pools or improving γ-secretase processivity alone may not be enough to delay disease progression. It would be interesting to extend this study to a humanized APP mouse model¹²⁵, as this would allow to investigate the direct potential of altered PSEN2 expression to induce amyloid pathology instead of the impact on existing amyloid pathology as investigated here. Another limitation is that our study did not address the role of altered PSEN2 expression in microglia, but initial data show significant differences: PSEN2KO accelerated microglial recruitment to plaques whereas the N141I KI mutation delayed it compared to APP KI mice (Perdok et al., in preparation). We do not exclude that these outcomes may also relate to the observed plaque morphology differences (Fig. 2). PSEN2 has been suggested to mediate microglial inflammatory response^126,127,128. Many microglial activating proteins are type I transmembrane proteins, some of which are already identified as substrates¹²⁹, and the role of PSEN2-specific substrates in microglia is a focus of ongoing research. In conclusion, our study underscores the need to consider the contribution of individual γ-secretase complexes in AD pathogenesis to fully understand early molecular changes in AD pathology, which could be relevant for future therapeutic interventions.

Methods

The data generated in this study are provided in the Supplementary Information/Source Data file.

Experimental models

Mice

Mice carrying a FAD Psen2 mutation N141I were generated using CRISPR-Cas9 technology to target exon 6 of the mouse Psen2 gene. Crispor webtool (http://crispor.tefor.net/) was used to select the guide RNA (5’CAUGAUGAGGGUGUUAAGCA), and efficiency of the guide RNA was tested in mouse embryonic fibroblasts (MEF). We made use of an ssODN carrying the mutated sequence in addition to three silent mutation within the PAM site and guide binding site (5’ CAGCATCTACACGCCCTTCACGGAGGACACGCCCTCGGTGGGCCAGCGGCTCCTCAACTCTGTCCTGATCACCCTCATCATGATCAGCGTCATCGTAGTCATGACCATCTTCCTCGTGGTACTCTACAA). Cas9 (50 ng/µl), sgRNA (0.6 µM) and ssODN (12.5 ng/µl) were injected into the pronucleus of C57BL/6 J embryos via microinjection in collaboration with the Mouse Expertise Unit of KU Leuven. Positive pups were selected by PCR and Sanger Sequencing (Figure S1b, c) using the set of primers listed in Table 1.

Table 1 PCR and Sanger Sequencing primers

Full size table

Sanger sequencing confirmed the absence of unwanted additional mutations at the target site (608 bp amplicon F1-R1). One pup was selected as founder and backcrossed for two generations with C57BL/6 J to cross out any potential unknown off-target in the founder mouse. PSEN2KO and APP NLGF (both in C57BL/6 J) were kindly provided by Bart de Strooper and Takaomi Saido. Mice were housed under standard conditions, on a diurnal 14-hour light/dark cycle with ad libitum access to food and water. Only female mice were utilized for experiments. All experiments were approved by the KU Leuven ethical committee (ECDs: 174/2017, 173/2022, Creation Annaert Willem/2020).

Neuronal cultures

Cortical and hippocampal mouse neurons were obtained from randomly chosen male and female E17.5 embryos. Brain tissues were dissected and incubated in 0.25% trypsin in HBSS (10% HBSS, 1% HEPES, 1% Penicillin/Streptomycin, 0.5% NaHCO₃ in sterile water) for 15 min at 37 °C. Neurons were washed in memHorse (10% Horse serum, 1% Penicillin/Streptomycin, 1%GlutaMAX in MEM medium) followed by mechanical dissociation using flamed glass Pasteur pipettes. Individual neurons were counted using the Countess III and 250,000 neurons were plated in a 6 cm dish containing poly-L-lysine (PLL) coated 15 mm coverslips for imaging or 900,000 neurons on 6 cm PLL coated dishes for western blot. Subsequently, coverslips were flipped onto a glia feeding layer grown in Neurobasal medium supplemented with 2% B27 supplement, 1% Glutamax and 1% Penicillin/Streptomycin. Three days after the culture 5-Fluoro-2′-deoxyuridine (FDU) was added to prevent overgrowth of glia cells.

Antibodies and reagents

Mouse anti-human Aβ (82e1, 10323, Tecan, 1:100 for IF, 1:500 for WB), anti-human Aβ fibrils (AB2287, Sigma Aldrich, 1:200 for IF), rat anti-Lamp1 (1D4B, sc19992 Santa Cruz, 1:200 for IF, 1:1000 for WB), rabbit anti-APP (Y188, Ab32136, Abcam, 1:500 for IF), rabbit anti-APP (B63.3, home-made, 1:10 000 for WB), rabbit anti-presenilin 2 (ab51249, abcam, 1:500 for IF, 1:10 000 for WB), rabbit anti-presenilin 1 (B19, home-made, 1:20 000 for WB), mouse anti-nicastrin (9C3, home-made, 1:8000 for WB), mouse anti- actin (A5441, Sigma Aldrich, 1:10 000 for WB), chicken anti-Map2 (PA1-10005, Thermo Fisher, 1:5000 for IF), rabbit anti-EEA1 (07-1820, Sigma Aldrich, 1:250 for IF), goat anti-VPS35 (ab10099, Abcam, 1:200 for IF), mouse anti-GLUA1 (MA5-27694, Invitrogen, 1:500 for IF, 1:1000 for WB), rabbit anti-pGLUA1 (p1160-845, Phosphosolutions, 1:100 for IF), guinea-pig anti-vGLUT1 (135,304, Synaptic Systems, 1:500 for IF, 1:5000 for WB), PSD 95 nanobody-647 (N3702-AF647-L, Synaptic Systems, 1: 500 for IF), mouse anti- synaptobrevin2 (104 211, Synaptic Systems, 1:500 for IF), mouse anti-ankyrin-G coupled to FL550 (75-146-FL550, NeuroMab, 1:500).

Experimental methods

Immunohistochemistry

Mice were anaesthetized using xylazine/ketamine (Ketamine/Xylazine dosage: 0.1 ml/ 20 g intraperitoneal. Delivers: 87.5 mg/kg Ketamine/12.5 mg/kg Xylazine) and transcardially perfused using PBS (−/−). Mice were decapitated and brains were post-fixed for 48 h in 4% PFA. Coronal (60 µm) free-floating sections were blocked using PBS −/− containing 3% goat serum and 0.3% Triton100x for 1.5 h at room temperature followed by overnight incubation at 4°C with the primary antibody. After three washes in PBS −/−, slices were incubated with Alexa-labelled secondary antibodies (1:500) for 2hrs at room temperature. In case of Thioflavin staining, slices were incubated in 0.05% Thioflavin for 5 min followed by two progressive dehydration steps in 70% and 90% ethanol, respectively. Finally, slices were washed three in PBS −/− and once in sterile water before mounting using ProLong Glass Antifade Mountant. Sections were visualized on a Nikon A1R Eclipse confocal system or Nikon AX. For analysis of brain slices, total area covered was always measured by delineating the cortex or the hippocampus and then calculating the area covered by the fluorescence staining of interest.

Neurons were fixed for 20 min at room temperature in 4% paraformaldehyde (PFA) and semi-permeabilized in 0.1% Triton X-100 in PBS for 5 min. In case of surface labelling of vGLUT1, GLUA1 and LAMP1, the permeabilization step was skipped, keeping the remaining procedure the same. Following a 1 h blocking step in blocking buffer (PBS + 2%BSA + 5%donkey or 5%goat serum) at room temperature, coverslips were incubated with primary antibody diluted in blocking buffer overnight at 4 °C. Following three washes with PBS (+/+) coverslips were incubated with Alexa-labelled secondary antibodies (1:1000) at room temperature. Coverslips were mounted using ProLong Glass Antifade Mountant. Coverslips were imaged on a Zeiss LSM 900 and all image analysis was done using ImageJ.

Image quantification

All confocal image quantification was done using ImageJ.

PSD95 puncta: Two neuronal extensions were selected using a segmented line. Subsequently a rectangle was fit to the selection and duplicated from the original image. The length of the selected branch was measured followed by the number of PSD95 puncta using thresholding with the default mask. The total number of puncta was divided by the length of the segment to obtain puncta per µm and the average size of puncta per neuron was taken to investigate PSD size.

Surface labelling vGLUT1, GLUA1 and LAMP1: The phalloidin channel was used to determine total cell surface with thresholding using the MaxEntropy mask and total area measured. This mask was then applied to the vGLUT1, GLUA1 or LAMP1 channel and everything outside the mask was subtracted to remove non-specific staining. Subsequently, the total area of surface protein was determined with thresholding using the MaxEntropy mask and next divided by cell area to derive a % of total surface covered by the respective protein.

Compartment size: The soma of the neuron was delineated, the size measured and everything outside the soma removed. Then the total area of LAMP1, EEA1 and VPS35 immunostaining was delineated using thresholding with the MaxEntropy mask. The total compartment area was divided by soma area to get a %soma area covered by the cellular compartment.

Triple overlap: Overlap of three different channels in the case of LAMP1 and Y188/82e1 and PSD95 and PSEN2/LAMP1 immunoreactivities were measured using the ComDet v.0.5.5. plugin in Fiji. Max distance between colocalized spots was set at 10 pixels, with the approximate particle size as 20 pixels and an intensity threshold of 5.

Biochemical sample collection

Mice were transcardially perfused and brains were extricated from the skull. Cerebrum and hippocampi were flash frozen, separately, using liquid nitrogen and stored at −70 °C until further use. Tissue samples were weighed and homogenized using a FastPrep bead mill in T-PER™ Tissue Protein Extraction Reagent (78510, Thermo Scientific). Beads were spun down and supernatants centrifuged for one hour at 263,000 x g at 4 °C. Supernatants were kept as soluble fraction and the pellet was further resuspended in 2 volumes 6 M Guanidine buffer (G3272-100G, Sigma-Aldrich) sonicated and vortexed followed by a one-hour incubation at 25 °C. Samples were centrifuged for 20 min at 426,204 x g and the obtained supernatant diluted 12x in guanidine diluent (20 mM phosphate, 0.4 M NaCl, 2 mM EDTA, 10% Block Ace, 0.2% BSA, 0.05% NaN₃, 0.075% CHAPS, protease inhibitor cocktail, pH 7.0). Samples were stored at −70 until further use.

Western blot

Protein concentrations were determined using the Bio-Rad Protein Assay Dye Reagent (5000006, Bio-Rad). Samples of equal protein content were heated in 4x loading buffer (NuPage LDS Samples Buffer 4x, NP0008, Invitrogen) at 70 °C for 10 min, loaded in NuPage 4%–12%, Bis-Tris gels (NP0322BOX, Invitrogen) and separated by electrophoresis in NuPage SDS MES running buffer (NP0002, Invitrogen). Proteins were transferred to nitrocellulose membranes (Life technologies) for 2 h at 4 °C. Membranes were subsequently blocked in 5% non-fat milk powder dissolved in TBS containing 0.1% Tween 20 (TBS-T; blocking buffer) for 1 h at room temperature followed by overnight incubation in primary antibody diluted in blocking buffer at 4 °C. Following incubation, membranes were washed in TBS-T and incubated with horseradish peroxidase-conjugated secondary antibody (1:10 000) diluted in blocking buffer for 1 h at room temperature. Following washing in TBS-T membranes were incubated with Western Lighting Plus-ECL (NEL105001EA, Perkin Elmer) and protein bands were visualized and images by chemiluminescence in a IQ680. Images were analyzed and quantified using Aida Image Analyzer software (Raytest, Germany).

ELISA

To quantify the concentration of Aβ37, Aβ38, Aβ40 and Aβ42 peptides, MesoScale Discovery (MSD) multispot Aβ ELISA plates pre-coated with anti- Aβ37, Aβ38, Aβ40 and Aβ42 was used. In case of brain extracts, the soluble fraction was undiluted and insoluble fractions were diluted 1:200 in blocking buffer (PBS + 0.1% casein). For neuron extracts, 12 µg protein was used for the intracellular fraction (diluted in blocking buffer) and the secreted pool was diluted 1:2 in blocking buffer. Plates were blocked with blocking buffer (PBS + 0.1% casein) for 2 h at room temperature. SULFO-TAG JRF/AbN/25 detection antibody was diluted 1:1 with standards (synthetic human Aβ37, Aβ38, Aβ40 and Aβ42 peptides of known concentration) or with prepared samples. 50 µl of the mix was added to the wells and incubated overnight at 4 °C. After incubation, plates were washed 5x in washing buffer and signal was developed by adding 150 μl/well of the 2x MSD Read Buffer T (Tris-based buffer containing tripropylamine). Plate was read Sector Imager 6000 (Meso Scale Discovery).

Transfection of hippocampal neurons

Coverslips were placed in a 12well plate containing 200 µl conditioned medium and 50 µl 5x kynurenic acid (final concentration 2 mM, K3375-250MG, Sigma-Aldrich). Per coverslip 100 µl of precipitate was prepared by diluting 2 µg of plasmid (mCherry-GFP-LC3 ref: 10.1186/s13024-017-0165-0) in 1x TE buffer (10 µM Tris-HCl pH 7.5, 5 mM EDTA pH 8, adjusted to pH 7.3) to a final volume of 45 µl mixing gently, and followed by the addition of 5 µl of 2.5 M CaCl₂ and mixing gently. Finally, 50 µl of 2x HEBS buffer (NaCl, KCl, Na₂HPO₄, Glucose, HEPES) are added and the precipitate is mixed and added dropwise to the coverslips. Coverslips are incubated for 2 h at 37 °C and 5%CO₂ following by a 15 min wash in acidic 5% CO₂ equilibrated wash medium (200 µl 5x kynurenic acid and 800 µl Neurobasal medium per coverslip). Coverslips were transferred to their original dish and fixed or imaged 48 h after transfection. Live imaging was performed using the LSM880 or LSM980 using the fast Airyscan mode taking 10 images with the lowest interval possible. The 488 nm and 543 nm lasers were used to image the GFP and mCherry channels respectively. For analysis, Fiji was used. Images were max projected and subsequently the number of red and yellow puncta were manually quantified.

TRPML1-dependent lysosomal calcium release

Lysosomal Ca²⁺ release was measured as previously stated¹³⁰. Briefly, cells were incubated with 1 µM Fura-2 AM solution (F1221, Thermo Fisher Scientific) for 45 min at room temperature in the dark. Cells were rinsed twice with Krebs-Ca²⁺ (135 mM NaCl, 5.9 mM KCl, 11.5 mM Glucose, 11.6 mM HEPES, 1.2 mM MgCl₂, and 1.5 mM CaCl₂) and left at room temperature for 30 min for de-esterization, in the dark. All acquisitions were carried out in Krebs-EGTA+ (135 mM NaCl, 5.9 mM KCl, 11.5 mM Glucose, 11.6 mM HEPES, 1.2 mM MgCl₂, and 2 mM EGTA). A baseline Fura-2 signal was measured for 30 s followed by a pulse with 20 µM of the TRPML agonist ML-SA1 (Cat. No. 4746, Tocris) and the response was acquired every 5 s for 10 min, using an Olympus IX81 with a UAPO/340 40x oil objective (1.35 NA) and operated by CellR Software (Olympus). Recordings were done at 340 and 380 nm excitation (530 nm emission filters). For image analysis, regions were segmented manually in Image J and fluorescence intensities were measured over time. Fura-2 signals were corrected to obtain ΔF/F0 (F0 being the initial signal recorded). In addition, each cell’s area under the curve was measured using Prism GraphPad (Version 10.02).

Rhodamine dextran lysosomal calcium assay

Coverslips with DIV14 hippocampal neurons were placed in a 12-well plate and Rhodamine (0.25 mg/ml, 20451-1 mg, AAT Bioquest) and Dextran-488 (0.1 mg/ml, D22910, ThermoFisher) diluted in conditioned medium were added for 8 h. Neurons were washed twice in medium and returned to their original dish for an overnight chase. The day after neurons were washed twice in HBSS (+/+) without phenol red and imaged live on the LSM880 or LSM980 using the fast Airyscan mode taking a single snap shot using the 488 nm and 543 nm lasers. For analysis, lysosomes were delineated by generating a mask using the green channel and the average green intensity was measured. Then the average red channel intensity was measured within the lysosomal mask to determine levels of lysosomal calcium.

Electrophysiology

Mice were anesthetized with isoflurane and brains were dissected and transferred into ice-cold cutting solution (87 mM NaCl, 2.5 mM KCl, 1.25 mM NaH2PO4, 10 mM glucose, 25 mM NaHCO₃, 0.5 mM CaCl₂, 7 mM MgCl₂, 75 mM sucrose, 1 mM kynurenic acid, 5 mM ascorbic acid and 3 mM pyruvic acid) at pH 7.4 and 5% CO2/95% O₂. 300 µm-thick parasagittal hippocampal sections were incubated at 34 °C for 35 min to recover. For recordings, slices were placed onto a multielectrode array (MEA 2100, Multichannel Systems) and continuously perfused with prewarmed (33–34 °C) artificial cerebrospinal fluid (aCSF; 119 mM NaCl, 2.5 mM KCl, 1 mM NaH₂PO₄, 11 mM glucose, 26 mM NaHCO₃, 4 mM MgCl₂ and 4 mM CaCl₂) at pH 7.4, 95% O₂ and 5% CO₂. Field excitatory post-synaptic potentials (fEPSPs) were recorded from mossy fiber-CA3 synapses by stimulating and recording from the appropriate (visually identified) electrodes. Input-output curves were established for each individual slice by applying single stimuli ranging from 500 to 2750 mV with 250 mV increments. Stimulus strength that corresponds to 35% of maximal response in the input-output curve was applied for the following recordings. For long-term potentiation (LTP) experiments, stable fEPSPs were recorded for 30 min as baseline. Next, two trains of high-frequency stimulation (75 stimuli, 100 Hz) with 5 min interval were applied to induce LTP and consecutive fEPSPs were recorded for 55 additional minutes. Baseline and post-LTP induction fEPSPs were recorded every 5 min using three consecutive stimulations, 15 s apart. These triplets were averaged during analysis. For paired pulse and train stimulations we used separate, naïve slices. Slices were continuously perfused with prewarmed (33–34 °C) artificial cerebrospinal fluid (aCSF; 127 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 25 mM glucose, 25 mM NaHCO₃, 1 mM MgCl₂ and 2 mM CaCl₂) at pH 7.4, 95% O₂ and 5% CO₂. Input-output curves as described above were used to determine stimulation strength required for 35% maximal response. Three repetitions of each paired pulse stimulation (inter-stimulation intervals: 20, 40 60, 80, 100, 200, 400, 800, 1000 and 2000 ms) were recorded and averaged (1 minute between pairs). For train stimulations we recorded and averaged two repetitions for each train (1, 5, 10 and 20 Hz; 10 stimulations; 5 min between trains). Recordings were analyzed and processed using Multi Channel Experimenter software (Multichannel Systems).

Imaging

Transmission electron microscopy

Mice were anaesthetized using xylazine/ketamine and transcardially perfused using PBS (−/−) following perfusion with fixative solution (4% paraformaldehyde, 2.5% glutaraldehyde and 0.2% picric acid in 0.1 M Phosphate Buffer) and brains were dissected and post-fixed in fixative solution overnight at 4 °C. Subsequently, brains were washed in 0.1 M Phosphate Buffer, embedded in 3% agarose (Sigma-Aldrich) and 100µm-thick coronal sections were cut using a vibratome (VT1000S, Leica). Hippocampal CA3 regions were cropped from the brain sections and left overnight in 0.1 M PB buffer. Cropped sections were washed on ice first in 0.1 M PB buffer followed by double distilled water (ddH₂O) washes and then were post-fixed in 0.1 M Phosphate Buffer containing 1% OsO₄ and 1.5% K4Fe(CN)6 for one hour at on ice in the dark. Following washing in ddH₂O, sections were incubated with 0.2% tannic acid in ddH₂O for 30 min in the dark at room temperature followed by an additional incubation in 1% OsO₄ in the dark. Following washing step cropped sections are incubated overnight in 25% methanol with 0.5% uranyl acetate overnight at 4 °C with rotation in the dark. After, sections were washed in ddH₂O and incubated with Walton’s lead aspartate (0.02 M lead nitrate in 0.03Msodium aspartate, pH 5.5) for 30 min at 60 °C, washed again in ddH2O and dehydrated with graded series of ethanol (30%, 50%, 70%, 80%, 95% and 100%). Following two washes in 100% ethanol and two washes in propylene oxide, sections were infiltrated with different ratios of propylene oxide:resin (2:1, 1:1 and 1:2 supplemented with BDMA) and embedded in pure EPON resin (supplemented with BDMA). Ultrathin 70 nm-thick sections were cut using an ultramicrotome (EM UC7, Leica) and collected on copper grids. Images of mossy fiber-CA3 synapses were acquired with a JEOL JEM1400 transmission electron microscope equipped with an Olympus SIS Quemesa camera with 4000X or 25000X magnification. Images were analyzed using Microscopy Image Browser (MIB, University of Helsinki)¹³¹ and processed using Fiji software (National Institutes of Health).

Behavior

All animal behavioral experiments were performed at and in collaboration with the mINT Animal Behavior Facility at the KU Leuven Faculty of Psychology. Mice were transferred 10 days prior to tested for acclimatization.

24 h activity

Mice were caged individually with ad libitum access to food and water. Cages were placed centrally between 3 lasers and laser crossings were monitored overnight. For analysis, total laser crossings were summated for each hour to indicate overall activity.

Open field

Anxiety and locomotor activity were tested using an open field paradigm. Prior to testing, mice were housed in the dark for 30 min. Then mice were placed in a square box constructed from plexiglass and tracked for 10 min using an over-head camera and ANY-maze tracking software (Stölting). Time spend in the middle was compared as an indication for anxiety and total distance travelled as a readout for locomotion.

Elevated plus maze

Anxiety was tested using the elevated plus maze apparatus consisting of two open arms and two closed arms, elevated 50 cm off the ground. Mice were placed in the center tracked for 10 min using an over-head camera and ANY-maze tracking software (Stölting). Time spend in the open arms versus close arms was measured as readout for anxiety –like behavior.

Social novelty/social preference

Social memory was tested in a three-chamber setup constructed from plexiglass¹³². The test mouse was placed in the central chamber which is connected by two walls containing 8 mm holes allowing for the exchange of olfactory, visual, and auditory signals. Following a 5 min habituation period, a novel mouse of the same gender was placed in one of the adjacent chambers (S1) and the preference of the test mouse to approach the novel mouse over the empty chamber was measured. After 10 min, a second stranger mouse was placed in the opposing chamber (S2) and social memory was tested as preference of the new mouse over the old mouse. Animals were tracked using an overhead camera and ANY-maze tracking software (Stölting). Time sniffing (head within 5 cm of the separation wall) was compared between groups.

Morris water maze

Spatial learning was tested in a hidden platform protocol in the Morris Water Maze¹³³. Mice were trained to locate a hidden platform (15 cm in diameter) in a pool (150 cm in diameter) filled with opaque water (26 ± 1 °C) surrounded by spatial cues. Mice were placed in the pool 4 times per day, from different entry points, for a maximum of 100 s during which the mouse is tracked using an overhead camera and Ethovision software (Noldus, Wageningen, NL). On training day 6 and 11 a probe the platform was removed from the pool and the time spend in the target quadrant was observed indicative of spatial learning.

Delayed matching to platform

Following the Morris Water Maze protocol, mice were subjected to an altered protocol to test working memory. Herein the platform location was altered every day. Mice were placed in the pool 4 times per day, from different entry points opposing the new location of the platform. Mouse was tracked using the same software and parameters as during the Morris Water Maze protocol. For analysis, the distances for the different swims were averaged over the week to observe decreases, indicative of a functional working memory.

Statistics, image analysis and reproducibility

For image analysis the freely available software ImageJ was used. Graphical and statistical analysis was done using GraphPad (Version 10.02). Statistical tests used are specified in the figure legends. For all experiments, data were reported based on individual cells/animals or biological replicates pooled from multiple cells, e.g. lysates. No statistical methods were used to predetermine the sample size. Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All data are available in the main text or the supplementary materials. Source data are provided with this paper. The data is further available from the corresponding author upon request. Source data are provided with this paper.

References

Jurisch-Yaksi, N., Sannerud, R. & Annaert, W. A fast growing spectrum of biological functions of γ-secretase in development and disease. Biochim Biophys. Acta Biomembr. 1828, 2815–2827 (2013).
Article CAS Google Scholar
Güner, G. & Lichtenthaler, S. F. The substrate repertoire of γ-secretase/presenilin. Semin Cell Dev. Biol. 105, 27–42 (2020).
Article PubMed Google Scholar
De Strooper, B. et al. A presenilin-1-dependent γ-secretase-like protease mediates release of notch intracellular domain. Nature 398, 518–522 (1999).
Article ADS PubMed Google Scholar
Struhl, G. & Greenwald, I. Presenilin is required for activity and nuclear access of Notch in Drosophila. Nature 398, 522–524 (1999).
Article ADS CAS PubMed Google Scholar
Hartmann, D., De Strooper, B. & Saftig, P. Presenilin-1 deficiency leads to loss of Cajal-Retzius neurons and cortical dysplasia similar to human type 2 lissencephaly. Curr. Biol. 9, 719–727 (1999).
Article CAS PubMed Google Scholar
Wolfe, M. S. et al. A substrate-based difluoro ketone selectively inhibits Alzheimer’s γ-secretase activity. J. Med Chem. 41, 6–9 (1998).
Article CAS PubMed Google Scholar
De Strooper, B. Aph-1, Pen-2, and Nicastrin with Presenilin generate an active γ-Secretase complex. Neuron 38, 9–12 (2003).
Article PubMed Google Scholar
Takasugi, N. et al. The role of presenilin cofactors in the γ-secratase complex. Nature 422, 438–441 (2003).
Article ADS CAS PubMed Google Scholar
Escamilla-Ayala A., Wouters R., Sannerud R., Annaert W. Contribution of the Presenilins in the cell biology, structure and function of γ-secretase. Semin Cell Dev Biol. 105, 12–26 (2020).
Hebert S. S., et al. Coordinated and widespread expression of y-secretase in vivo: evidence for size and molecular heterogeneity. Neurobiol Dis 17, 260–272 (2004).
Takami, M. et al. Secretase: Successive tripeptide and tetrapeptide release from the transmembrane domain of β-carboxyl terminal fragment. J. Neurosci. 29, 13042–13052 (2009).
Article CAS PubMed PubMed Central Google Scholar
Szaruga, M. et al. Alzheimer’s-Causing Mutations Shift Aβ Length by Destabilizing γ-Secretase-Aβn Interactions. Cell 170, 443–456.e14 (2017).
Article CAS PubMed Google Scholar
Funamoto, S. et al. Truncated carboxyl-terminal fragments of β-amyloid precursor protein are processed to amyloid β-proteins 40 and 42. Biochemistry 43, 13532–13540 (2004).
Article CAS PubMed Google Scholar
Hardy, J. A. & Higgins, G. A. Alzheimer’s Disease: The Amyloid Cascade Hypothesis. Science. 256, 184–185 (1992).
Article ADS CAS PubMed Google Scholar
Hardy, J. & Selkoe, D. J. The amyloid hypothesis of Alzheimer’s disease: Progress and problems on the road to therapeutics. Science. 297, 353–356 (2002).
Article ADS CAS PubMed Google Scholar
Shen, J. et al. Skeletal and CNS Defects in Presenilin-1-Deficient mice. Cell 89, 629–639 (1997).
Article CAS PubMed Google Scholar
De Strooper, B. et al. Deficiency of presenilin-1 inhibits the normal cleavage of amyloid precursor protein. Nature 391, 387–390 (1998).
Article ADS PubMed Google Scholar
Herreman, A. et al. Presenilin 2 deficiency causes a mild pulmonary phenotype and no changes in amyloid precursor protein processing but enhances the embryonic lethal phenotype of presenilin 1 deficiency. Proc. Natl Acad. Sci. USA. 96, 11872–11877 (1999).
Article ADS CAS PubMed PubMed Central Google Scholar
Steiner, H. et al. A loss of function mutation of presenilin-2 interferes with amyloid β- peptide production and Notch signaling. J. Biol. Chem. 274, 28669–28673 (1999).
Article CAS PubMed Google Scholar
Donoviel, D. B. et al. Mice lacking both presenilin genes exhibit early embryonic patterning defects. Genes Dev. 13, 2801–2810 (1999).
Article CAS PubMed PubMed Central Google Scholar
Acx, H. et al. Inactivation of γ‐secretases leads to accumulation of substrates and non‐Alzheimer neurodegeneration. EMBO Mol. Med. 9, 1088–1099 (2017).
Article CAS PubMed PubMed Central Google Scholar
Faguás, C. M. S., Molina, P. S. & Saura, C. A. Loss of presenilin function enhances tau phosphorylation and aggregation in mice. Acta Neuropathol. Commun. 7, 1–20 (2021).
Google Scholar
Peethumnongsin, E. et al. Convergence of presenilin- and tau-mediated pathways on axonal trafficking and neuronal function. J. Neurosci. 30, 13409–13418 (2010).
Article CAS PubMed PubMed Central Google Scholar
Saura, C. A. et al. Loss of presenilin function causes impairments of memory and synaptic plasticity followed by age-dependent neurodegeneration. Neuron 42, 23–36 (2004).
Article CAS PubMed Google Scholar
Watanabe, H., Iqbal, M., Zheng, J., Wines-Samuelson, M. & Shen, J. Partial loss of presenilin impairs age-dependent neuronal survival in the cerebral cortex. J. Neurosci. 34, 15912–15922 (2014).
Article PubMed PubMed Central Google Scholar
Beglopoulos, V. et al. Reduced β-amyloid production and increased inflammatory responses in presenilin conditional knock-out mice. J. Biol. Chem. 279, 46907–46914 (2004).
Article CAS PubMed Google Scholar
Jayadev, S. et al. Alzheimer’s disease phenotypes and genotypes associated with mutations in presenilin 2. Brain 133, 1143–1154 (2010).
Article PubMed PubMed Central Google Scholar
Walker, E. S., Martinez, M., Brunkan, A. L. & Goate, A. Presenilin 2 familial Alzheimer’s disease mutations result in partial loss of function and dramatic changes in Aβ 42/40 ratios. J. Neurochem. 92, 294–301 (2005).
Article CAS PubMed Google Scholar
Sannerud, R. et al. Restricted Location of PSEN2/γ-Secretase Determines Substrate Specificity and Generates an Intracellular Aβ Pool. Cell 166, 193–208 (2016).
Article CAS PubMed PubMed Central Google Scholar
Petit, D. et al. Aβ profiles generated by Alzheimer’s disease causing PSEN1 variants determine the pathogenicity of the mutation and predict age at disease onset. Mol. Psychiatry 27, 2821–2832 (2022).
Article CAS PubMed PubMed Central Google Scholar
Meckler, X. & Checler, F. Presenilin 1 and presenilin 2 target γ-secretase complexes to distinct cellular compartments. J. Biol. Chem. 291, 12821–12837 (2016).
Article CAS PubMed PubMed Central Google Scholar
Gouras, G. K. et al. Intraneuronal Aβ42 accumulation in human brain. Am. J. Pathol. 156, 15–20 (2000).
Article CAS PubMed PubMed Central Google Scholar
Billings, L. M. et al. causes the onset of early Alzheimer’s disease-related cognitive deficits in transgenic mice. Neuron 45, 675–688 (2005).
Article CAS PubMed Google Scholar
Gyure, K. A. et al. β-amyloid precedes development of amyloid plaques in Down syndrome. Arch. Pathol. Lab Med. 125, 489–492 (2001).
Article CAS PubMed Google Scholar
Pensalfini, A. et al. Intracellular amyloid and the neuronal origin of Alzheimer neuritic plaques. Neurobiol. Dis. 71, 53–61 (2014).
Article CAS PubMed Google Scholar
Su, Y. & Chang, P. T. Acidic pH promotes the formation of toxic fibrils from β-amyloid peptide. Brain Res. 893, 287–291 (2001).
Article CAS PubMed Google Scholar
Liu, R. Q. et al. Membrane localization of β-amyloid 1-42 in lysosomes: A possible mechanism for lysosome labilization. J. Biol. Chem. 285, 19986–19996 (2010).
Article CAS PubMed PubMed Central Google Scholar
Esbjörner, E. K. et al. Direct observations of amyloid β Self-assembly in live cells provide insights into differences in the kinetics of Aβ(1-40) and Aβ(1-42) aggregation. Chem. Biol. 21, 732–742 (2014).
Article PubMed PubMed Central Google Scholar
Yang, A. J., Chandswangbhuvana, D., Margol, L. & Glabe, C. G. Loss of endosomal/lysosomal membrane impermeability is an early event in amyloid Aβ1-42 pathogenesis. J. Neurosci. Res. 52, 691–698 (1998).
Article CAS PubMed Google Scholar
Friedrich, R. P. et al. Mechanism of amyloid plaque formation suggests an intracellular basis of Aβ pathogenicity. Proc. Natl Acad. Sci. Usa. 107, 1942–1947 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Brunello, L. et al. Presenilin-2 dampens intracellular Ca 2+ stores by increasing Ca 2+ leakage and reducing Ca 2+ uptake. J. Cell Mol. Med. 13, 3358–3369 (2009).
Article PubMed PubMed Central Google Scholar
Fedeli, C., Filadi, R., Rossi, A., Mammucari, C. & Pizzo, P. PSEN2 (presenilin 2) mutants linked to familial Alzheimer disease impair autophagy by altering Ca2+ homeostasis. Autophagy 15, 2044–2062 (2019).
Article CAS PubMed PubMed Central Google Scholar
Nuriel, T. et al. The endosomal-lysosomal pathway is dysregulated by APOE4 expression in vivo. Front Neurosci. 11, 1–12 (2017).
Article Google Scholar
Bustos, V. et al. Phosphorylated Presenilin 1 decreases β-amyloid by facilitating autophagosome-lysosome fusion. Proc. Natl Acad. Sci. USA. 114, 7148–7153 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Cataldo, A. M. et al. Endocytic pathway abnormalities precede amyloid β deposition in sporadic alzheimer’s disease and down syndrome: Differential effects of APOE genotype and presenilin mutations. Am. J. Pathol. 157, 277–286 (2000).
Article CAS PubMed PubMed Central Google Scholar
Nixon, R. A. The role of autophagy in neurodegenerative disease. Nat. Med. 19, 983–997 (2013).
Article CAS PubMed Google Scholar
Peric, A. & Annaert, W. Early etiology of Alzheimer’s disease: tipping the balance toward autophagy or endosomal dysfunction? Acta Neuropathol. 129, 363–381 (2015).
Article CAS PubMed PubMed Central Google Scholar
Saito, T. et al. Single App knock-in mouse models of Alzheimer’s disease. Nat. Neurosci. [Internet]. 17, 661–663, https://www.nature.com/articles/nn.3697 (2014).
Article CAS PubMed Google Scholar
Ueno, H. et al. Comparison between the aggregation of human and rodent amyloid β-proteins in GM1 ganglioside clusters. Biochemistry 53, 7523–7530 (2014).
Article CAS PubMed Google Scholar
Boon, B. D. C. et al. The coarse-grained plaque: a divergent Aβ plaque-type in early-onset Alzheimer’s disease. Acta Neuropathol. 140, 811–830 (2020).
Article CAS PubMed PubMed Central Google Scholar
Gowrishankar, S. et al. Massive accumulation of luminal protease-deficient axonal lysosomes at Alzheimer’s disease amyloid plaques. Proc. Natl Acad. Sci. Usa. 112, E3699–E3708 (2015).
Article CAS PubMed PubMed Central Google Scholar
Calafate, S. et al. Early alterations in the MCH system link aberrant neuronal activity and sleep disturbances in a mouse model of Alzheimer’s disease. Nat. Neurosci. 26, 1021–1031 (2023).
Article CAS PubMed PubMed Central Google Scholar
Morris, R. G. & Baddeley, A. D. Primary and Working Memory Functioning in Alzheimer-type Dementia. J. Clin. Exp. Neuropsychol. 10, 279–296 (1987).
Article Google Scholar
Stopford, C. L., Thompson, J. C., Neary, D., Richardson, A. M. T. & Snowden, J. S. Working memory, attention, and executive function in Alzheimer’s disease and frontotemporal dementia. Cortex 48, 429–446 (2012).
Article PubMed Google Scholar
Adams, J. P. & Dudek, S. M. Late-phase long-term potentiation: getting to the nucleus. Nat. Rev. Neurosci. 6, 737–743 (2005).
Article CAS PubMed Google Scholar
Baltaci, S. B., Mogulkoc, R., Baltaci, A. K. Molecular Mechanisms of Early and Late LTP. Vol. 44, Neurochemical Research. Springer New York LLC. 281–296 (2019).
Rollenhagen, A. & Lübke, J. H. R. The mossy fiber bouton: The “common” or the “unique” synapse? Front. Synaptic Neurosci. 2, 1–9 (2010).
Díaz-Alonso, J. & Nicoll, R. A. AMPA receptor trafficking and LTP: Carboxy-termini, amino-termini and TARPs. Neuropharmacology 197, 108710 (2021).
Article PubMed PubMed Central Google Scholar
Fernández-Alfonso, T. & Ryan, T. A. The efficiency of the synaptic vesicle cycle at central nervous system synapses. Trends Cell Biol. 16, 413–420 (2006).
Article PubMed Google Scholar
Takahashi, R. H. et al. Intraneuronal Alzheimer Aβ42 accumulates in multivesicular bodies and is associated with synaptic pathology. Am. J. Pathol. 161, 1869–1879 (2002).
Article CAS PubMed PubMed Central Google Scholar
Almeida, C. G., Takahashi, R. H. & Gouras, G. K. Β-Amyloid Accumulation Impairs Multivesicular Body Sorting By Inhibiting the Ubiquitin-Proteasome System. J. Neurosci. 26, 4277–4288 (2006).
Article CAS PubMed PubMed Central Google Scholar
Hu, X. et al. Amyloid seeds formed by cellular uptake, concentration, and aggregation of the amyloid-beta peptide. Proc. Natl Acad. Sci. Usa. 106, 20324–20329 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Kwart, D. et al. A Large Panel of Isogenic APP and PSEN1 Mutant Human iPSC Neurons Reveals Shared Endosomal Abnormalities Mediated by APP β-CTFs, Not Aβ. Neuron 104, 256–270.e5 (2019).
Article CAS PubMed Google Scholar
Hung, C. et al. SORL1 deficiency in human excitatory neurons causes APP-dependent defects in the endolysosome-autophagy network. Cell Rep. 35, 109259 (2021).
Article CAS PubMed PubMed Central Google Scholar
Im, E. et al. Lysosomal dysfunction in Down syndrome and Alzheimer mouse models is caused by v-ATPase inhibition by Tyr682-phosphorylated APP βCTF. Sci Adv. 9, eadg1925 (2023).
Saftig, P. & Klumperman, J. Lysosome biogenesis and lysosomal membrane proteins: Trafficking meets function. Nat. Rev. Mol. Cell Biol. 10, 623–635 (2009).
Article CAS PubMed Google Scholar
Bretou, M. et al. Accumulation of APP C-terminal fragments causes endolysosomal dysfunction through the dysregulation of late endosome to lysosome-ER contact sites. Dev. Cell. 59, 1571–1592.e9 (2024).
Article CAS PubMed Google Scholar
Essayan-Perez, S. & Südhof, T. C. Neuronal γ-secretase regulates lipid metabolism, linking cholesterol to synaptic dysfunction in Alzheimer’s disease. Neuron 111, 3176–3194 (2023).
Article CAS PubMed PubMed Central Google Scholar
Lee, S. H., Bolshakov, V. Y. & Shen, J. Presenilins regulate synaptic plasticity in the perforant pathways of the hippocampus. Mol. Brain. 16, 1–12 (2023).
Article CAS PubMed PubMed Central Google Scholar
Chavez-Gutiérrez, L. et al. The mechanism of γ-Secretase dysfunction in familial Alzheimer disease. EMBO J. 31, 2261–2274 (2012).
Article PubMed PubMed Central Google Scholar
Braak, H., Thal, D. R., Ghebremedhin, E. & Del Tredici, K. Stages of the pathologic process in alzheimer disease: Age categories from 1 to 100 years. J. Neuropathol. Exp. Neurol. 70, 960–969 (2011).
Article CAS PubMed Google Scholar
Richards, J. G. et al. PS2APP transgenic mice, coexpressing hPS2mut and hAPPswe, show age-related cognitive deficits associated with discrete brain amyloid deposition and inflammation. J. Neurosci. 23, 8989–9003 (2003).
Article CAS PubMed PubMed Central Google Scholar
Toda, T., Noda, Y., Ito, G., Maeda, M. & Shimizu, T. Presenilin-2 mutation causes early amyloid accumulation and memory impairment in a transgenic mouse model of Alzheimer’s disease. J. Biomed. Biotechnol. 2011, 1–13 (2011).
Google Scholar
Watanabe, H. et al. Flexible and accurate substrate processing with distinct presenilin/γ-secretases in human cortical neurons. eNeuro 8, 1–20 (2021).
Article CAS Google Scholar
Nilsberth, C. et al. The “Arctic” APP mutation (E693G) causes Alzheimer’s disease by enhanced Aβ protofibril formation. Nat. Neurosci. 4, 887–893 (2001).
Article CAS PubMed Google Scholar
Yang, Y. et al. Cryo-EM structures of amyloid-b 42 filaments from human brains. Science. 375, 167–172 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Dickson, T. C. & Vickers, J. C. The morphological phenotype of β-amyloid plaques and associated neuritic changes in Alzheimer’s disease. Neuroscience 105, 99–107 (2001).
Article CAS PubMed Google Scholar
Thal, D. R., Walter, J., Saido, T. C. & Fändrich, M. Neuropathology and biochemistry of Aβ and its aggregates in Alzheimer’s disease. Acta Neuropathol. 129, 167–182 (2015).
Article CAS PubMed Google Scholar
Verkkoniemi, A. et al. Variant Alzheimer’s disease with spastic paraparesis: Clinical characterization. Neurology 54, 1103–1109 (2000).
Article CAS PubMed Google Scholar
Le, T. V., Crook, R., Hardy, J. & Dickson, D. W. Cotton wool plaques in non-familial late-onset Alzheimer disease. J. Neuropathol. Exp. Neurol. 60, 1051–1061 (2001).
Article CAS PubMed Google Scholar
Whyte, L. S. et al. Reduction in open field activity in the absence of memory deficits in the AppNL−G − F knock-in mouse model of Alzheimer’s disease. Behavioural Brain Res. 336, 177–181 (2018).
Article Google Scholar
Mehla, J. et al. Age-dependent behavioral and biochemical characterization of single APP knock-in mouse (APPNL-G-F/NL-G-F) model of Alzheimer’s disease. Neurobiol. Aging 75, 25–37 (2019).
Article CAS PubMed Google Scholar
Latif-Hernandez, A. et al. Subtle behavioral changes and increased prefrontal-hippocampal network synchronicity in APP NL − G − F mice before prominent plaque deposition. Behavioural Brain Res. 364, 431–441 (2019).
Article Google Scholar
Salesse, C. et al. Opposite Control of Excitatory and Inhibitory Synapse Formation by Slitrk2 and Slitrk5 on Dopamine Neurons Modulates Hyperactivity Behavior. Cell Rep. 30, 2374–2386.e5 (2020).
Article CAS PubMed Google Scholar
Busche, M. A. & Konnerth, A. Neuronal hyperactivity - A key defect in Alzheimer’s disease? BioEssays 37, 624–632 (2015).
Article PubMed Google Scholar
Shah, D. et al. Astrocyte calcium dysfunction causes early network hyperactivity in Alzheimer’s disease. Cell Rep. 40, 111280 (2022).
Kipanyula, M. J. et al. Ca 2+ dysregulation in neurons from transgenic mice expressing mutant presenilin 2. Aging Cell. 11, 885–893 (2012).
Article CAS PubMed Google Scholar
Fontana, R. et al. Early hippocampal hyperexcitability in PS2APP mice: role of mutant PS2 and APP. Neurobiol. Aging 50, 64–76 (2017).
Article CAS PubMed Google Scholar
Pendin, D. et al. Familial Alzheimer’s disease presenilin-2 mutants affect Ca2+ homeostasis and brain network excitability. Aging Clin. Exp. Res. 33, 1705–1708 (2021).
Article PubMed Google Scholar
Leparulo, A. et al. Dampened Slow Oscillation Connectivity Anticipates Amyloid Deposition in the PS2APP Mouse Model of Alzheimer’s Disease. Cells 9, 1–20 (2019).
Article Google Scholar
Beckman, M. et al. Loss of presenilin 2 age-dependently alters susceptibility to acute seizures and kindling acquisition. Neurobiol. Dis. 136, 104719 (2020).
Article CAS PubMed Google Scholar
Knox, K. M., Beckman, M., Smith, C. L., Jayadev, S., Barker-haliski, M. Chronic seizures induce sex-specific cognitive deficits with loss of presenilin 2 function. Exp. Neurol. 361, 114321 (2023).
Jarosz-Griffiths, H. H., Noble, E., Rushworth, J. V. & Hooper, N. M. Amyloid-β receptors: The good, the bad, and the prion protein. J. Biol. Chem. 291, 3174–3183 (2016).
Article CAS PubMed Google Scholar
Verret, L. et al. Inhibitory interneuron deficit links altered network activity and cognitive dysfunction in alzheimer model. Cell 149, 708–721 (2012).
Article CAS PubMed PubMed Central Google Scholar
Palop, J. J. & Mucke, L. Network abnormalities and interneuron dysfunction in Alzheimer disease. Nat. Rev. Neurosci. 17, 777–792 (2016).
Article CAS PubMed PubMed Central Google Scholar
Li, S. et al. Soluble Oligomers of Amyloid β Protein Facilitate Hippocampal Long-Term Depression by Disrupting Neuronal Glutamate Uptake. Neuron 62, 788–801 (2009).
Article CAS PubMed PubMed Central Google Scholar
Li, S. et al. Soluble a β oligomers inhibit long-term potentiation through a mechanism involving excessive activation of extrasynaptic NR2B-containing NMDA receptors. J. Neurosci. 31, 6627–6638 (2011).
Article CAS PubMed PubMed Central Google Scholar
Zott, B. et al. A vicious cycle of β amyloid−dependent neuronal hyperactivation. Science. 365, 559–565 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Spellman, T. et al. Hippocampal-prefrontal input supports spatial encoding in working memory. Nature 522, 309–314 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Song, D. et al. The lateralization of left hippocampal CA3 during the retrieval of spatial working memory. Nat. Commun. 11, 1–13 (2020).
ADS Google Scholar
Yassa, M. A. et al. Pattern separation deficits associated with increased hippocampal CA3 and dentate gyrus activity in nondemented older adults. Hippocampus 21, 968–979 (2011).
Article PubMed Google Scholar
Zhang, C. et al. Presenilins are essential for regulating neurotransmitter release. Nature 460, 632–636 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Zhang, D. et al. Inactivation of presenilins causes pre-synaptic impairment prior to post-synaptic dysfunction. J. Neurochem. 115, 1215–1221 (2010).
Article CAS PubMed PubMed Central Google Scholar
Wu, B., Yamaguchi, H., Lai, F. A. & Shen, J. Presenilins regulate calcium homeostasis and presynaptic function via ryanodine receptors in hippocampal neurons. Proc. Natl Acad. Sci. Usa. 110, 15091–15096 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Lee, S. H. et al. Presenilins regulate synaptic plasticity and mitochondrial calcium homeostasis in the hippocampal mossy fiber pathway. Mol. Neurodegener. 12, 1–15 (2017).
Article Google Scholar
Ripoli, C. et al. Intracellular accumulation of amyloid-β (Aβ) protein plays a major role in Aβ-induced alterations of glutamatergic synaptic transmission and plasticity. J. Neurosci. 34, 12893–12903 (2014).
Article PubMed PubMed Central Google Scholar
Zhang, H. et al. Role of Aβ in Alzheimer’s-related synaptic dysfunction. Front Cell Dev. Biol. 10, 1–14 (2022).
ADS Google Scholar
Tu, S. Okamoto S ichi, Lipton SA, Xu H. Oligomeric Aβ-induced synaptic dysfunction in Alzheimer’s disease. Mol. Neurodegener. 9, 48 (2014).
Article PubMed PubMed Central Google Scholar
Woodruff, G. et al. Defective Transcytosis of APP and Lipoproteins in Human iPSC-Derived Neurons with Familial Alzheimer’s Disease Mutations. Cell Rep. 17, 759–773 (2016).
Article CAS PubMed PubMed Central Google Scholar
Hung, C. O. Y. & Livesey, F. J. Altered γ-Secretase Processing of APP Disrupts Lysosome and Autophagosome Function in Monogenic Alzheimer’s Disease. Cell Rep. 25, 3647–3660.e2 (2018).
Article CAS PubMed PubMed Central Google Scholar
Neely, K. M., Green, K. N. & LaFerla, F. M. Presenilin is necessary for efficient proteolysis through the autophagy-lysosome system in a γ-secretase-independent manner. J. Neurosci. 31, 2781–2791 (2011).
Article CAS PubMed PubMed Central Google Scholar
Sambri, I. et al. Lysosomal dysfunction disrupts presynaptic maintenance and restoration of presynaptic function prevents neurodegeneration in lysosomal storage diseases. EMBO Mol. Med. 9, 112–132 (2017).
Article CAS PubMed Google Scholar
Pará, C. et al. Early defects in mucopolysaccharidosis type IIIC disrupt excitatory synaptic transmission. JCI Insight. 9, 6 (2021).
Nemani, V. M. et al. Increased Expression of α-Synulein Reduces Neurotransmitter Release by Inhibiting Synaptic Vesicle Reclustering after Endocytosis. Neuron 65, 66–79 (2010).
Article CAS PubMed PubMed Central Google Scholar
Padamsey, Z. et al. Activity-Dependent Exocytosis of Lysosomes Regulates the Structural Plasticity of Dendritic Spines. Neuron 93, 132–146 (2017).
Article CAS PubMed PubMed Central Google Scholar
Foster, W. J. et al. Hippocampal mGluR1-dependent long-term potentiation requires NAADP-mediated acidic store Ca²⁺ signaling. Sci Signal.11, eaat9093 (2018).
Sun, J. et al. LAMTOR1 inhibition of TRPML1‐dependent lysosomal calcium release regulates dendritic lysosome trafficking and hippocampal neuronal function. EMBO J. 41, 1–25 (2022).
Article CAS Google Scholar
Özkan, N. et al. ER – lysosome contacts at a pre-axonal region regulate axonal lysosome availability. Nat. Commun. 12, 1–18 (2021).
Article Google Scholar
Fernández-Monreal, M., Brown, T. C., Royo, M. & Esteban, J. A. The balance between receptor recycling and trafficking toward lysosomes determines synaptic strength during long-term depression. J. Neurosci. 32, 13200–13205 (2012).
Article PubMed PubMed Central Google Scholar
Kim, T., Yamamoto, Y., Tanaka-Yamamoto, K. Timely regulated sorting from early to late endosomes is required to maintain cerebellar long-term depression. Nat Commun. 8, 1–16 (2017).
Yap, C. C., Digilio, L., McMahon, L. P., Garcia, A. D. R. & Winckler, B. Degradation of dendritic cargos requires Rab7-dependent transport to somatic lysosomes. J. Cell Biol. 217, 3141–3159 (2018).
Article CAS PubMed PubMed Central Google Scholar
Cogli, L., Piro, F. & Bucci, C. Rab7 and the CMT2B disease. Biochem. Soc. Trans. 37, 1027–1031 (2009).
Article CAS PubMed Google Scholar
Braggin, J. E. et al. Alternative splicing in a presenilin 2 variant associated with Alzheimer disease. Ann. Clin. Transl. Neurol. 6, 762–777 (2019).
Article CAS PubMed PubMed Central Google Scholar
Course, M. M. et al. Aberrant splicing of PSEN2, but not PSEN1, in individuals with sporadic Alzheimer’s disease. Brain 146, 507–518 (2023).
Article PubMed Google Scholar
Serneels, L. et al. Modeling the β-secretase cleavage site and humanizing amyloid-beta precursor protein in rat and mouse to study Alzheimer’s disease. Mol. Neurodegener. 1, 15 (2020).
Jayadev, S. et al. Presenilin 2 is the predominant γ-Secretase in microglia and modulates cytokine release. PLoS One. 5, e15743 (2010).
Fung, S. et al. Early-Onset Familial Alzheimer Disease Variant PSEN2 N141I Heterozygosity is Associated with Altered Microglia Phenotype. J. Alzheimer’s Dis. 77, 675–688 (2020).
Article CAS Google Scholar
Nam, H. et al. Presenilin 2 N141I mutation induces hyperactive immune response through the epigenetic repression of REV-ERBα. Nat. Commun. 13, 1–18 (2022).
Article Google Scholar
Hou, P. et al. The γ-secretase substrate proteome and its role in cell signaling regulation. Mol. Cell. 83, 4106–4122 (2023).
Article CAS PubMed Google Scholar
Van Acker, Z. P. et al. Phospholipase D3 degrades mitochondrial DNA to regulate nucleotide signaling and APP metabolism. Nat. Commun. 1, 14 (2023).
Belevich, I., Joensuu, M., Kumar, D., Vihinen, H. & Jokitalo, E. Microscopy Image Browser: A Platform for Segmentation and Analysis of Multidimensional Datasets. PLoS Biol. 14, 1–13 (2016).
Article Google Scholar
Naert, A., Callaerts-Vegh, Z. & D’Hooge, R. Nocturnal hyperactivity, increased social novelty preference and delayed extinction of fear responses in post-weaning socially isolated mice. Brain Res Bull. 85, 354–362 (2011).
Article PubMed Google Scholar
Callaerts-Vegh, Z. et al. Concomitant deficits in working memory and fear extinction are functionally dissociated from reduced anxiety in metabotropic glutamate receptor 7-deficient mice. J. Neurosci. 26, 6573–6582 (2006).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This work was financed through VIB, KU Leuven (C14/21/095 and SMINF KA/20/085 to WA), the Research Foundation – Flanders (FWO; G0C4220N and I001322N to W.A. and G008023N to LCG), SAO-FRA (#2020/030 and 2022/022 to WA), BrightFocus (A2024033S to WA) and the Alzheimer Association (AARF-20-677499 to ZPVA and AARF-23-1149740 to CV). AP and SGF are recipients of FWO SB-strategic mandates (1S35019N and 1S59621N, respectively); ZPVA and CV are recipients of FWO postdoctoral fellowships (1250425 N and 12B7423N, respectively). We thank T. Saido and B. De Strooper for providing the APPKI mice, J. de Wit and R. Mancuso for critical reading, N. Gunko for EM support and T. Pattenuzzo for help with TEM quantifications. The VIB BioImaging Core is acknowledged for providing access and support to confocal and EM. Figure 8 was created with Biorender (https://www.biorender.com/).

Author information

These authors contributed equally: Zoë P. Van Acker, Céline Vrancx.

Authors and Affiliations

Laboratory for Membrane Trafficking, VIB Center for Brain and Disease Research, Leuven, Belgium
Anika Perdok, Zoë P. Van Acker, Céline Vrancx, Ragna Sannerud, Inge Vorsters, Assunta Verrengia & Wim Annaert
Department of Neurosciences, KU Leuven, Herestraat 49box 602, Leuven, Belgium
Anika Perdok, Zoë P. Van Acker, Céline Vrancx, Ragna Sannerud, Inge Vorsters, Assunta Verrengia, Sara Gutiérrez Fernández, Lutgarde Serneels, Lucía Chávez-Gutiérrez & Wim Annaert
mINT Animal Behavior Facility, Faculty of Psychology, KU Leuven, Tiensestraat 102, Leuven, Belgium
Zsuzsanna Callaerts-Végh
Electrophysiology Expertise Unit, VIB-Center for Brain and Disease Research, Leuven, Belgium
Eline Creemers, Britt D’hauw & Keimpe Wierda
Laboratory of Proteolytic Mechanisms mediating Neurodegeneration, Leuven, Belgium
Sara Gutiérrez Fernández & Lucía Chávez-Gutiérrez
Mouse Expertise Unit, VIB-Center for Brain and Disease Research, Leuven, Belgium
Lutgarde Serneels

Authors

Anika Perdok
View author publications
Search author on:PubMed Google Scholar
Zoë P. Van Acker
View author publications
Search author on:PubMed Google Scholar
Céline Vrancx
View author publications
Search author on:PubMed Google Scholar
Ragna Sannerud
View author publications
Search author on:PubMed Google Scholar
Inge Vorsters
View author publications
Search author on:PubMed Google Scholar
Assunta Verrengia
View author publications
Search author on:PubMed Google Scholar
Zsuzsanna Callaerts-Végh
View author publications
Search author on:PubMed Google Scholar
Eline Creemers
View author publications
Search author on:PubMed Google Scholar
Sara Gutiérrez Fernández
View author publications
Search author on:PubMed Google Scholar
Britt D’hauw
View author publications
Search author on:PubMed Google Scholar
Lutgarde Serneels
View author publications
Search author on:PubMed Google Scholar
Keimpe Wierda
View author publications
Search author on:PubMed Google Scholar
Lucía Chávez-Gutiérrez
View author publications
Search author on:PubMed Google Scholar
Wim Annaert
View author publications
Search author on:PubMed Google Scholar

Contributions

A.P. designed, performed and analysed most experiments, prepared all manuscript figures. A.P. and W.A. wrote the article with input from all other authors. Z.V.A., R.S., C.V., and A.V. contributed with primary cultures, imaging experiments on primary neurons and quantitative analysis. I.V. carried out TEM imaging and Z.V.C. coordinated the behavioral tests and contributed with A.P. to data interpretations. S.G.F. and L.C.G. contributed with MSD ELISA experiments. L.S. performed injections into the pronucleus of C57Bl6J embryos by microinjection (MutaMouse facility) for the generation of the new mouse line. E.C., B.D., and K.W. performed electrophysiological experiments. W.A. designed the overall research strategy and supervised the experiments.

Corresponding author

Correspondence to Wim Annaert.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Fabien Delerue and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Transparent Peer Review file

Source data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Perdok, A., Van Acker, Z.P., Vrancx, C. et al. Altered expression of Presenilin2 impacts endolysosomal homeostasis and synapse function in Alzheimer’s disease-relevant brain circuits. Nat Commun 15, 10412 (2024). https://doi.org/10.1038/s41467-024-54777-y

Download citation

Received: 15 March 2024
Accepted: 21 November 2024
Published: 29 November 2024
DOI: https://doi.org/10.1038/s41467-024-54777-y

This article is cited by

Spectrum of γ-Secretase dysfunction as a unifying predictor of ADAD age at onset across PSEN1, PSEN2 and APP causal genes
- Sara Gutiérrez Fernández
- Cristina Gan Oria
- Lucía Chávez-Gutiérrez
Molecular Neurodegeneration (2025)
Lysosome repair fails in ageing and Alzheimer’s disease
- Céline Vrancx
- Wim Annaert
Nature Cell Biology (2025)