Structural basis for membrane-proximal proteolysis of substrates by ADAM10.

The endopeptidase ADAM10 is a critical catalyst for the regulated proteolysis of key drivers of mammalian development, physiology, and non-amyloidogenic cleavage of APP as the primary α-secretase. ADAM10 function requires the formation of a complex with a C8-tetraspanin protein, but how tetraspanin binding enables positioning of the enzyme active site for membrane-proximal cleavage remains unknown. We present here a cryo-EM structure of a vFab-ADAM10-Tspan15 complex, which shows that Tspan15 binding relieves ADAM10 autoinhibition and acts as a molecular measuring stick to position the enzyme active site about 20 Å from the plasma membrane for membrane-proximal substrate cleavage. Cell-based assays of N-cadherin shedding establish that the positioning of the active site by the interface between the ADAM10 catalytic domain and the bound tetraspanin influences selection of the preferred cleavage site. Together, these studies reveal the molecular mechanism underlying ADAM10 proteolysis at membrane-proximal sites and offer a roadmap for its modulation in disease.

Structural basis of γ-secretase inhibition and modulation by small molecule drugs.

Development of γ-secretase inhibitors (GSIs) and modulators (GSMs) represents an attractive therapeutic opportunity for Alzheimer's disease (AD) and cancers. However, how these GSIs and GSMs target γ-secretase has remained largely unknown. Here, we report the cryoelectron microscopy (cryo-EM) structures of human γ-secretase bound individually to two GSI clinical candidates, Semagacestat and Avagacestat, a transition state analog GSI L685,458, and a classic GSM E2012, at overall resolutions of 2.6-3.1 Å. Remarkably, each of the GSIs occupies the same general location on presenilin 1 (PS1) that accommodates the ß strand from amyloid precursor protein or Notch, interfering with substrate recruitment. L685,458 directly coordinates the two catalytic aspartate residues of PS1. E2012 binds to an allosteric site of γ-secretase on the extracellular side, potentially explaining its modulating activity. Structural analysis reveals a set of shared themes and variations for inhibitor and modulator recognition that will guide development of the next-generation substrate-selective inhibitors.

Alzheimer's-Causing Mutations Shift Aß Length by Destabilizing γ-Secretase-Aßn Interactions.

Alzheimer's disease (AD)-linked mutations in Presenilins (PSEN) and the amyloid precursor protein (APP) lead to production of longer amyloidogenic Aß peptides. The shift in Aß length is fundamental to the disease; however, the underlying mechanism remains elusive. Here, we show that substrate shortening progressively destabilizes the consecutive enzyme-substrate (E-S) complexes that characterize the sequential γ-secretase processing of APP. Remarkably, pathogenic PSEN or APP mutations further destabilize labile E-S complexes and thereby promote generation of longer Aß peptides. Similarly, destabilization of wild-type E-S complexes by temperature, compounds, or detergent promotes release of amyloidogenic Aß. In contrast, E-Aßn stabilizers increase γ-secretase processivity. Our work presents a unifying model for how PSEN or APP mutations enhance amyloidogenic Aß production, suggests that environmental factors may increase AD risk, and provides the theoretical basis for the development of γ-secretase/substrate stabilizing compounds for the prevention of AD.

The γ-secretase substrate proteome and its role in cell signaling regulation.

γ-Secretases mediate the regulated intramembrane proteolysis (RIP) of more than 150 integral membrane proteins. We developed an unbiased γ-secretase substrate identification (G-SECSI) method to study to what extent these proteins are processed in parallel. We demonstrate here parallel processing of at least 85 membrane proteins in human microglia in steady-state cell culture conditions. Pharmacological inhibition of γ-secretase caused substantial changes of human microglial transcriptomes, including the expression of genes related to the disease-associated microglia (DAM) response described in Alzheimer disease (AD). While the overall effects of γ-secretase deficiency on transcriptomic cell states remained limited in control conditions, exposure of mouse microglia to AD-inducing amyloid plaques strongly blocked their capacity to mount this putatively protective DAM cell state. We conclude that γ-secretase serves as a critical signaling hub integrating the effects of multiple extracellular stimuli into the overall transcriptome of the cell.

New precision medicine avenues to the prevention of Alzheimer's disease from insights into the structure and function of γ-secretases.

Two phase-III clinical trials with anti-amyloid peptide antibodies have met their primary goal, i.e. slowing of Alzheimer's disease (AD) progression. However, antibody therapy may not be the optimal therapeutic modality for AD prevention, as we will discuss in the context of the earlier small molecules described as "γ-secretase modulators" (GSM). We review here the structure, function, and pathobiology of γ-secretases, with a focus on how mutations in presenilin genes result in early-onset AD. Significant progress has been made in generating compounds that act in a manner opposite to pathogenic presenilin mutations: they stabilize the proteinase-substrate complex, thereby increasing the processivity of substrate cleavage and altering the size spectrum of Aß peptides produced. We propose the term "γ-secretase allosteric stabilizers" (GSAS) to distinguish these compounds from the rather heterogenous class of GSM. The GSAS represent, in theory, a precision medicine approach to the prevention of amyloid deposition, as they specifically target a discrete aspect in a complex cell biological signalling mechanism that initiates the pathological processes leading to Alzheimer's disease.

APP substrate ectodomain defines amyloid-ß peptide length by restraining γ-secretase processivity and facilitating product release.

Sequential proteolysis of the amyloid precursor protein (APP) by γ-secretases generates amyloid-ß (Aß) peptides and defines the proportion of short-to-long Aß peptides, which is tightly connected to Alzheimer's disease (AD) pathogenesis. Here, we study the mechanism that controls substrate processing by γ-secretases and Aß peptide length. We found that polar interactions established by the APPC99 ectodomain (ECD), involving but not limited to its juxtamembrane region, restrain both the extent and degree of γ-secretases processive cleavage by destabilizing enzyme-substrate interactions. We show that increasing hydrophobicity, via mutation or ligand binding, at APPC99 -ECD attenuates substrate-driven product release and rescues the effects of Alzheimer's disease-associated pathogenic γ-secretase and APP variants on Aß length. In addition, our study reveals that APPC99 -ECD facilitates the paradoxical production of longer Aßs caused by some γ-secretase inhibitors, which act as high-affinity competitors of the substrate. These findings assign a pivotal role to the substrate ECD in the sequential proteolysis by γ-secretases and suggest it as a sweet spot for the potential design of APP-targeting compounds selectively promoting its processing by these enzymes.

Enzyme-substrate interface targeting by imidazole-based γ-secretase modulators activates γ-secretase and stabilizes its interaction with APP.

Alzheimer's disease (AD) pathogenesis has been linked to the accumulation of longer, aggregation-prone amyloid ß (Aß) peptides in the brain. Γ-secretases generate Aß peptides from the amyloid precursor protein (APP). Γ-secretase modulators (GSMs) promote the generation of shorter, less-amyloidogenic Aßs and have therapeutic potential. However, poorly defined drug-target interactions and mechanisms of action have hampered their therapeutic development. Here, we investigate the interactions between the imidazole-based GSM and its target γ-secretase-APP using experimental and in silico approaches. We map the GSM binding site to the enzyme-substrate interface, define a drug-binding mode that is consistent with functional and structural data, and provide molecular insights into the underlying mechanisms of action. In this respect, our analyses show that occupancy of a γ-secretase (sub)pocket, mediating binding of the modulator's imidazole moiety, is sufficient to trigger allosteric rearrangements in γ-secretase as well as stabilize enzyme-substrate interactions. Together, these findings may facilitate the rational design of new modulators of γ-secretase with improved pharmacological properties.

Novel Social Stimulation Ameliorates Memory Deficit in Alzheimer's Disease Model through Activating α-Secretase.

As the most common form of dementia in the world, Alzheimer's disease (AD) is a progressive neurological disorder marked by cognitive and behavioral impairment. According to previous researches, abundant social connections shield against dementia. However, it is still unclear how exactly social interactions benefit cognitive abilities in people with AD and how this process is used to increase their general cognitive performance. In this study, we found that single novel social (SNS) stimulation promoted c-Fos expression and increased the protein levels of mature ADAM10/17 and sAPPα in the ventral hippocampus (vHPC) of wild-type (WT) mice, which are hippocampal dorsal CA2 (dCA2) neuron activity and vHPC NMDAR dependent. Additionally, we discovered that SNS caused similar changes in an AD model, FAD4T mice, and these alterations could be reversed by α-secretase inhibitor. Furthermore, we also found that multiple novel social (MNS) stimulation improved synaptic plasticity and memory impairments in both male and female FAD4T mice, accompanied by α-secretase activation and Aß reduction. These findings provide insight into the process underpinning how social interaction helps AD patients who are experiencing cognitive decline, and we also imply that novel social interaction and activation of the α-secretase may be preventative and therapeutic in the early stages of AD.

Functional and topological analysis of PSENEN, the fourth subunit of the γ-secretase complex.

The γ-secretase complexes are intramembrane cleaving proteases involved in the generation of the Aß peptides in Alzheimer's disease. The complex consists of four subunits, with Presenilin harboring the catalytic site. Here, we study the role of the smallest subunit, PSENEN or Presenilin enhancer 2, encoded by the gene Psenen, in vivo and in vitro. We find a profound Notch deficiency phenotype in Psenen-/- embryos confirming the essential role of PSENEN in the γ-secretase complex. We used Psenen-/- fibroblasts to explore the structure-function of PSENEN by the scanning cysteine accessibility method. Glycine 22 and proline 27, which border the membrane domains 1 and 2 of PSENEN, are involved in complex formation and stabilization of γ-secretase. The hairpin structured hydrophobic membrane domains 1 and 2 are exposed to a water-containing cavity in the complex, while transmembrane domain 3 is not water exposed. We finally demonstrate the essential role of PSENEN for the cleavage activity of the complex. PSENEN is more than a structural component of the γ-secretase complex and might contribute to the catalytic mechanism of the enzyme.

Inhibition of BACE1 affected both its Aß producing and degrading activities and increased Aß42 and Aß40 levels at high level BACE1 expression.

The beta-site amyloid precursor protein cleaving enzyme 1 (BACE1) is the predominant ß-secretase, cleaving the amyloid precursor protein (APP) via the amyloidogenic pathway. In addition, BACE1 as an amyloid degrading enzyme (ADE), cleaves Aß to produce the C-terminally truncated non-toxic Aß fragment Aß34 which is an indicator of amyloid clearance. Here, we analyzed effects of BACE1 inhibitors on its opposing enzymatic functions, i.e., amyloidogenic (Aß producing) and amyloidolytic (Aß degrading) activities, using cell culture models with varying BACE1/APP ratios. Under high level BACE1 expression, low-dose inhibition unexpectedly yielded a two-fold increase in Aß42 and Aß40 levels. The concomitant decrease in Aß34 and secreted APPß levels suggested that the elevated Aß42 and Aß40 levels were due to the attenuated Aß degrading activity of BACE1. Notably, the amyloidolytic activity of BACE1 was impeded at lower BACE1 inhibitor concentrations compared to its amyloidogenic activity, thereby suggesting that the Aß degrading activity of BACE1 was more sensitive to inhibition than its Aß producing activity. Under endogenous BACE1 and APP levels, "low-dose" BACE1 inhibition affected both the Aß producing and degrading activities of BACE1, i.e., significantly increased Aß42/Aß40 ratio and decreased Aß34 levels, respectively. Further, we incubated recombinant BACE1 with synthetic Aß peptides and found that BACE1 has higher affinity for Aß substrates over APP. In summary, our results suggest that stimulating BACE1's ADE activity and halting Aß production without decreasing Aß clearance could still be a promising therapeutic approach with new, yet to be developed, BACE1 modulators.

Identification of PS1/gamma-secretase and glutamate transporter GLT-1 interaction sites.

The recently discovered interaction between Presenilin 1 (PS1), a catalytic subunit of γ-secretase responsible for generating amyloid-ß peptides, and GLT-1, a major glutamate transporter in the brain (EAAT2), provides a mechanistic link between these two key factors involved in Alzheimer's disease (AD) pathology. Modulating this interaction can be crucial to understand the consequence of such crosstalk in AD context and beyond. However, the interaction sites between these two proteins are unknown. Herein, we utilized an alanine scanning approach coupled with FRET-based fluorescence lifetime imaging microscopy to identify the interaction sites between PS1 and GLT-1 in their native environment within intact cells. We found that GLT-1 residues at position 276 to 279 (TM5) and PS1 residues at position 249 to 252 (TM6) are crucial for GLT-1-PS1 interaction. These results have been cross validated using AlphaFold Multimer prediction. To further investigate whether this interaction of endogenously expressed GLT-1 and PS1 can be prevented in primary neurons, we designed PS1/GLT-1 cell-permeable peptides (CPPs) targeting the PS1 or GLT-1 binding site. We used HIV TAT domain to allow for cell penetration which was assayed in neurons. First, we assessed the toxicity and penetration of CPPs by confocal microscopy. Next, to ensure the efficiency of CPPs, we monitored the modulation of GLT-1-PS1 interaction in intact neurons by fluorescence lifetime imaging microscopy. We saw significantly less interaction between PS1 and GLT-1 with both CPPs. Our study establishes a new tool to study the functional aspect of GLT-1-PS1 interaction and its relevance in normal physiology and AD models.

Eta-secretase-like processing of the amyloid precursor protein (APP) by the rhomboid protease RHBDL4.

The amyloid precursor protein (APP) is a key protein in Alzheimer's disease synthesized in the endoplasmic reticulum (ER) and translocated to the plasma membrane where it undergoes proteolytic cleavages by several proteases. Conversely to other known proteases, we previously elucidated rhomboid protease RHBDL4 as a novel APP processing enzyme where several cleavages likely occur already in the ER. Interestingly, the pattern of RHBDL4-derived large APP C-terminal fragments resemble those generated by the η-secretase or MT5-MMP, which was described to generate so called Aη fragments. The similarity in large APP C-terminal fragments between both proteases raised the question whether RHBDL4 may contribute to η-secretase activity and Aη-like fragments. Here, we identified two cleavage sites of RHBDL4 in APP by mass spectrometry, which, intriguingly, lie in close proximity to the MT5-MMP cleavage sites. Indeed, we observed that RHBDL4 generates Aη-like fragments in vitro without contributions of α-, ß-, or γ-secretases. Such Aη-like fragments are likely generated in the ER since RHBDL4-derived APP-C-terminal fragments do not reach the cell surface. Inherited, familial APP mutations appear to not affect this processing pathway. In RHBDL4 knockout mice, we observed increased cerebral full length APP in comparison to wild type (WT) in support of RHBDL4 being a physiologically relevant protease for APP. Furthermore, we found secreted Aη fragments in dissociated mixed cortical cultures from WT mice, however significantly less Aη fragments in RHBDL4 knockout cultures. Our data underscores that RHBDL4 contributes to η-secretease-like processing of APP and that RHBDL4 is a physiologically relevant protease for APP.

Gut inflammation triggers C/EBPß/δ-secretase-dependent gut-to-brain propagation of Aß and Tau fibrils in Alzheimer's disease.

Inflammation plays an important role in the pathogenesis of Alzheimer's disease (AD). Some evidence suggests that misfolded protein aggregates found in AD brains may have originated from the gut, but the mechanism underlying this phenomenon is not fully understood. C/EBPß/δ-secretase signaling in the colon was investigated in a 3xTg AD mouse model in an age-dependent manner. We applied chronic administration of 1% dextran sodium sulfate (DSS) to trigger gut leakage or colonic injection of Aß or Tau fibrils or AD patient brain lysates in 3xTg mice and combined it with excision/cutting of the gut-brain connecting vagus nerve (vagotomy), in order to explore the role of the gut-brain axis in the development of AD-like pathologies and to monitor C/EBPß/δ-secretase signaling under those conditions. We found that C/EBPß/δ-secretase signaling is temporally activated in the gut of AD patients and 3xTg mice, initiating formation of Aß and Tau fibrils that spread to the brain. DSS treatment promotes gut leakage and facilitates AD-like pathologies in both the gut and the brain of 3xTg mice in a C/EBPß/δ-secretase-dependent manner. Vagotomy selectively blunts this signaling, attenuates Aß and Tau pathologies, and restores learning and memory. Aß or Tau fibrils or AD patient brain lysates injected into the colon propagate from the gut into the brain via the vagus nerve, triggering AD pathology and cognitive dysfunction. The results indicate that inflammation activates C/EBPß/δ-secretase and initiates AD-associated pathologies in the gut, which are subsequently transmitted to the brain via the vagus nerve.

Quantitative comparison of presenilin protein expression reveals greater activity of PS2-γ-secretase.

γ-secretase processing of amyloid precursor protein (APP) has long been of interest in the pathological progression of Alzheimer's disease (AD) due to its role in the generation of amyloid-ß. The catalytic component of the enzyme is the presenilins of which there are two homologues, Presenilin-1 (PS1) and Presenilin-2 (PS2). The field has focussed on the PS1 form of this enzyme, as it is typically considered the more active at APP processing. However, much of this work has been completed without appropriate consideration of the specific levels of protein expression of PS1 and PS2. We propose that expression is an important factor in PS1- and PS2-γ-secretase activity, and that when this is considered, PS1 does not have greater activity than PS2. We developed and validated tools for quantitative assessment of PS1 and PS2 protein expression levels to enable the direct comparison of PS in exogenous and endogenous expression systems, in HEK-293 PS1 and/or PS2 knockout cells. We show that exogenous expression of Myc-PS1-NTF is 5.5-times higher than Myc-PS2-NTF. Quantitating endogenous PS protein levels, using a novel PS1/2 fusion standard we developed, showed similar results. When the marked difference in PS1 and PS2 protein levels is considered, we show that compared to PS1-γ-secretase, PS2-γ-secretase has equal or more activity on APP and Notch1. This study has implications for understanding the PS1- and PS2-specific contributions to substrate processing, and their potential influence in AD pathogenesis.

A novel mouse model for N-terminal truncated Aß2-x generation through meprin ß overexpression in astrocytes.

Neurotoxic amyloid-ß (Aß) peptides cause neurodegeneration in Alzheimer's disease (AD) patients' brains. They are released upon proteolytic processing of the amyloid precursor protein (APP) extracellularly at the ß-secretase site and intramembranously at the γ-secretase site. Several AD mouse models were developed to conduct respective research in vivo. Most of these classical models overexpress human APP with mutations driving AD-associated pathogenic APP processing. However, the resulting pattern of Aß species in the mouse brains differs from those observed in AD patients' brains. Particularly mutations proximal to the ß-secretase cleavage site (e.g., the so-called Swedish APP (APPswe) fostering Aß1-x formation) lead to artificial Aß production, as N-terminally truncated Aß peptides are hardly present in these mouse brains. Meprin ß is an alternative ß-secretase upregulated in brains of AD patients and capable of generating N-terminally truncated Aß2-x peptides. Therefore, we aimed to generate a mouse model for the production of so far underestimated Aß2-x peptides by conditionally overexpressing meprin ß in astrocytes. We chose astrocytes as meprin ß was detected in this cell type in close proximity to Aß plaques in AD patients' brains. The meprin ß-overexpressing mice showed elevated amyloidogenic APP processing detected with a newly generated neo-epitope-specific antibody. Furthermore, we observed elevated Aß production from endogenous APP as well as AD-related behavior changes (hyperlocomotion and deficits in spatial memory). The novel mouse model as well as the established tools and methods will be helpful to further characterize APP cleavage and the impact of different Aß species in future studies.

Modulation of amyloid precursor protein cleavage by γ-secretase activating protein through phase separation.

Aberrant cleavage of amyloid precursor protein (APP) by γ-secretase is closely associated with Alzheimer's disease (AD). γ-secretase activating protein (GSAP) specifically promotes γ-secretase­mediated cleavage of APP. However, the underlying mechanism remains enigmatic. Here, we demonstrate that the 16-kDa C-terminal fragment of GSAP (GSAP-16K) undergoes phase separation in vitro and forms puncta-like condensates in cells. GSAP-16K exerts dual modulation on γ-secretase cleavage; GSAP-16K in dilute phase increases APP­C-terminal 99-residue fragment (C99) cleavage toward preferred production of ß-amyloid peptide 42 (Aß42), but GSAP-16K condensates reduce APP-C99 cleavage through substrate sequestration. Notably, the Aß42/Aß40 ratio is markedly elevated with increasing concentrations of GSAP-16K. GSAP-16K stably associates with APP-C99 through specific sequence elements. These findings mechanistically explain GSAP-mediated modulation of γ-secretase activity that may have ramifications on the development of potential therapeutics.

Reversible and bidirectional signaling of notch ligands.

The Notch signaling pathway is a direct cell-cell communication system involved in a wide variety of biological processes, and its disruption is observed in several pathologies. The pathway is comprised of a ligand-expressing (sender) cell and a receptor-expressing (receiver) cell. The canonical ligands are members of the Delta/Serrate/Lag-1 (DSL) family of proteins. Their binding to a Notch receptor in a neighboring cell induces a conformational change in the receptor, which will undergo regulated intramembrane proteolysis (RIP), liberating the Notch intracellular domain (NICD). The NICD is translocated to the nucleus and promotes gene transcription. It has been demonstrated that the ligands can also undergo RIP and nuclear translocation, suggesting a function for the ligands in the sender cell and possible bidirectionality of the Notch pathway. Although the complete mechanism of ligand processing is not entirely understood, and its dependence on Notch receptors has not been ruled out. Also, ligands have autonomous functions beyond Notch activation. Here we review the concepts of reverse and bidirectional signalization of DSL proteins and discuss the characteristics that make them more than just ligands of the Notch pathway.

Segregation of the membrane cargoes, BACE1 and amyloid precursor protein (APP) throughout the Golgi apparatus.

The intracellular trafficking of ß-site amyloid precursor protein (APP) cleaving enzyme (BACE1) and APP regulates amyloid-ß production. Our previous work demonstrated that newly synthesized BACE1 and APP are segregated into distinct trafficking pathways from the trans-Golgi network (TGN), and that alterations in their trafficking lead to an increase in Aß production in non-neuronal and neuronal cells. However, it is not known whether BACE1 and APP are transported through the Golgi stacks together and sorted at the TGN or segregated prior to arrival at the TGN. To address this question, we have used high-resolution Airyscan technology followed by Huygens deconvolution to quantify the overlap of BACE1 and APP in Golgi subcompartments in HeLa cells and primary neurons. Here, we show that APP and BACE1 are segregated, on exit from the endoplasmic reticulum and in the cis-Golgi and throughout the Golgi stack. In contrast, the transferrin receptor, which exits the TGN in AP-1 mediated transport carriers as for BACE1, colocalizes with BACE1, but not APP, throughout the Golgi stack. The segregation of APP and BACE1 is independent of the Golgi ribbon structure and the cytoplasmic domain of the cargo. Overall, our findings reveal the segregation of different membrane cargoes early in the secretory pathway, a finding relevant to the regulation of APP processing events.

Tmed10 deficiency results in impaired exocrine pancreatic differentiation in zebrafish larvae.

Transmembrane p24 trafficking protein 10 (TMED10) is a conserved vesicle trafficking protein. It is dysregulated in Alzheimer disease and plays a pivotal role in the pathogenesis of Alzheimer disease. In addition to the brain, TMED10 is highly expressed in the exocrine pancreas; however, its biological functions and underlying mechanisms remain largely unknown. We studied reduced Tmed10 in zebrafish embryos by morpholino oligonucleotide knockdown and CRISPR-Cas9 mutagenesis. Tmed10-deficient embryos showed extensive loss of acinar mass and impaired acinar differentiation. TMED10 has been reported to have an inhibitory effect on γ-secretase. As one of the substrates of γ-secretase, membrane-bound ß-catenin was significantly reduced in Tmed10-deficient embryos. Increased γ-secretase activity in wild-type embryos resulted in a phenotype similar to that of tmed10 mutants. And the mutant phenotype could be rescued by treatment with the γ-secretase inhibitor, N-[N-(3, 5-difluorophenacetyl)-l-alanyl]-s-phenylglycinet-butyl ester (DAPT). In addition, the reduced membrane-bound ß-catenin was accompanied with up-regulated ß-catenin target genes in Tmed10-deficient embryos. Overexpression of ß-catenin signaling inhibitor Dickkopf-1 (DKK-1) could rescue the exocrine pancreas defects. Taken together, our study reveals that Tmed10 regulates exocrine pancreatic differentiation through γ-secretase. Reduced membrane-bound ß-catenin, accompanied with hyperactivation of ß-catenin signaling, is an important cause of exocrine pancreas defects in Tmed10-deficient embryos. Our study reaffirms the importance of appropriate ß-catenin signaling in exocrine pancreas development. These findings may provide a theoretical basis for the development of treatment strategies for TMED10-related diseases.

Selective inhibitors of the PSEN1-gamma-secretase complex.

Clinical development of γ-secretases, a family of intramembrane cleaving proteases, as therapeutic targets for a variety of disorders including cancer and Alzheimer's disease was aborted because of serious mechanism-based side effects in the phase III trials of unselective inhibitors. Selective inhibition of specific γ-secretase complexes, containing either PSEN1 or PSEN2 as the catalytic subunit and APH1A or APH1B as supporting subunits, does provide a feasible therapeutic window in preclinical models of these disorders. We explore here the pharmacophoric features required for PSEN1 versus PSEN2 selective inhibition. We synthesized a series of brain penetrant 2-azabicyclo[2,2,2]octane sulfonamides and identified a compound with low nanomolar potency and high selectivity (>250-fold) toward the PSEN1-APH1B subcomplex versus PSEN2 subcomplexes. We used modeling and site-directed mutagenesis to identify critical amino acids along the entry part of this inhibitor into the catalytic site of PSEN1. Specific targeting one of the different γ-secretase complexes might provide safer drugs in the future.