Proteases recognize substrates by decoding sequence information-an essential cellular process elusive when recognition motifs are absent. Here, we unravel this problem for γ-secretase, an intramembrane-cleaving protease associated with Alzheimer's disease and cancer, by developing Comparative Physicochemical Profiling (CPP), a sequence-based algorithm for identifying interpretable physicochemical features. We show that CPP deciphers a γ-secretase substrate signature with single-residue resolution, which can explain the conformational transitions observed in substrates upon γ-secretase binding. Using machine learning, we predict the entire human γ-secretase substrate scope, revealing numerous previously unknown substrates. Our approach outperforms state-of-the-art protein language models, improving prediction accuracy from 60% to 90%, and achieves an 88% success rate in experimental validation. Building on these advancements, we identify pathways and diseases not linked before to γ-secretase. Generally, CPP decodes physicochemical signatures-a concept that extends beyond sequence motifs. We anticipate that our approach will be broadly applicable to diverse molecular recognition processes.

Intramembrane proteolysis by γ-secretase is critically implicated in Alzheimer disease pathogenesis by processing of its amyloid precursor protein substrate C99 into harmful amyloid-β peptide (Aβ) species. Recruitment of C99 involves binding of its N-terminal extracellular domain to exosites in γ-secretase. However, the role of these interactions has been elusive. Here, we show that the N-terminally shorter extracellular domain of the non-amyloidogenic C83 substrate also interacts with γ-secretase exosites, but more weakly. Moreover, we found that bulky aromatic mutations within the 16 amino acid extension of C99 interfere with exosite binding and inhibit substrate cleavage. Likewise, peptides binding to the C99 N-terminus that selectively inhibit Aβ production in vitro and in vivo interfere with exosite binding of C99. Our data show that exosite interactions of the C99 N-terminal region with γ-secretase can impact substrate cleavage and indicate that interfering with exosite interactions of C99 may provide a means for modulating amyloidogenic substrate processing.

BACKGROUND:Amyloid-β peptide is well-known as a pathogen of Alzheimer's disease, but its precursor, amyloid-beta precursor protein (APP), remains unexplained 30 years after its discovery. APP has two homologues called amyloid precursor-like protein 1 (APLP1) and amyloid precursor-like protein 2 (APLP2), and shares a similar structural organisation with them and has partially overlapping functions. APP family proteins are essential for survival, shown by the crossbreeding analysis of knockout mice of APP family molecules, including APLP1 and APLP2. APLP2 is known to play the most important role among them, but the molecular metabolism of APLP2 is only partially understood. Here, we analysed ectodomain shedding and γ-secretase cleavage of APLP2 by molecular biological and biochemical techniques.METHOD:We analysed the culture supernatant of HEK293 cells overexpressing APLP2 and human cerebrospinal fluid. For the analysis of secreted APLP2 fragments, we raised the OA603 antibody that reacts with the juxtamembrane domain of APLP2. Substrate cleavage sites were identified by matrix assisted laser desorption/ionisation mass spectrometry.RESULTS:By overexpressing in HEK293 cells, APLP2 undergoes ectodomain shedding at three sites in the extracellular region by α- and β-secretase-like activity and then is intramembranously cleaved at three sites by γ-secretase. In particular, in shedding, α-secretase-like activity was dominant in HEK cells. Surprisingly, in human cerebrospinal fluid, APLP2-derived metabolic fragments were mainly cleaved by β-secretase-like activity, not by α-secretase-like activity. Because APP is also mainly cleaved by beta-site amyloid precursor protein cleaving enzyme 1 in neurons and APLP1 is expressed exclusively in neurons, these findings suggest that APP family proteins may play a common role via β-secretase-like cleavage in the central nerve system.CONCLUSIONS:Thus, these findings may contribute to a better understanding of the role of APP family proteins in Alzheimer's disease.

The γ-secretase complex catalyzes the intramembrane cleavage of C99, a carboxy-terminal fragment of the amyloid precursor protein. Two paralogs of its catalytic subunit presenilin (PS1 and PS2) are expressed which are autocatalytically cleaved into an N-terminal and a C-terminal fragment during maturation of γ-secretase. In this study, we compared the efficiency and specificity of C99 cleavage by PS1- and PS2-containing γ-secretases. Mass spectrometric analysis of cleavage products obtained in cell-free and cell-based assays revealed that the previously described lower amyloid-β (Aβ)38 generation by PS2 is accompanied by a reciprocal increase in Aβ37 production. We further found PS1 and PS2 to show different preferences in the choice of the initial cleavage site of C99. However, the differences in Aβ38 and Aβ37 generation appear to mainly result from altered subsequent stepwise cleavage of Aβ peptides. Apart from these differences in cleavage specificity, we confirmed a lower efficiency of initial C99 cleavage by PS2 using a detergent-solubilized γ-secretase system. By investigating chimeric PS1/2 molecules, we show that the membrane-embedded, nonconserved residues of the N-terminal fragment mainly account for the differential cleavage efficiency and specificity of both presenilins. At the level of individual transmembrane domains (TMDs), TMD3 was identified as a major modulator of initial cleavage site specificity. The efficiency of endoproteolysis strongly depends on nonconserved TMD6 residues at the interface to TMD2, i.e., at a putative gate of substrate entry. Taken together, our results highlight the role of individual presenilin TMDs in the cleavage of C99 and the generation of Aβ peptides.

Presenilin (PS) with a genetic mutation generates abundant β-amyloid protein (Aβ) 43. Senile plaques are formed by Aβ43 in the cerebral parenchyma together with Aβ42 at middle ages. These brains cause the early onset of Alzheimer’s disease (AD), which is known as familial Alzheimer’s disease (FAD). Based on the stepwise processing model of Aβ generation by γ-secretase, we reassessed the levels of Aβs in the cerebrospinal fluid (CSF) of FAD participants. While low levels of Aβ38, Aβ40, and Aβ42 were generated in the CSF of FAD participants, the levels of Aβ43 were unchanged in some of them compared with other participants. We sought to investigate why the level of Aβ43 was unchanged in FAD participants. These characteristics of Aβ generation were observed in the γ-secretase assay in vitro using cells, which express FAD mutations in PS1. Aβ38 and Aβ40 generation from their precursors, Aβ42 and Aβ43, was decreased in PS1 mutants compared with wild-type (WT) PS1, as observed in the CSF. Both the ratios of Aβ38/Aβ42 and Aβ40/Aβ43 in PS1 mutants were lower than those in the WT. However, the ratio of Aβ43/amyloid precursor protein intracellular domain (AICD) increased in the PS1 mutants in an onset age dependency, while other Aβ/AICD ratios were decreased or unchanged. Importantly, liquid chromatography–mass spectrometry found that the generation of Aβ43 was stimulated from Aβ48 in PS1 mutants. This result indicates that PS1 mutants switched the Aβ43 generating line, which reflects the level of Aβ43 in the CSF and forming senile plaques.

Abstract γ-Secretase is a multimeric aspartyl protease that cleaves the membrane-spanning region of the β-carboxyl terminal fragment (βCTF) generated from β-amyloid precursor protein. γ-Secretase defines the generated molecular species of amyloid β-protein (Aβ), a critical molecule in the pathogenesis of Alzheimer’s disease (AD). Many therapeutic trials for AD have targeted γ-secretase. However, in contrast to the great efforts in drug discovery, the enzymatic features and cleavage mechanism of γ-secretase are poorly understood. Here we review our protein-chemical analyses of the cleavage products generated from βCTF by γ-secretase, which revealed that Aβ was produced by γ-secretase through successive cleavages of βCTF, mainly at three-residue intervals. Two representative product lines were identified. e-Cleavages occur first at Leu49-Val50 and Thr48-Leu49 of βCTF (in accordance with Aβ numbering). Longer generated Aβs, Aβ49 and Aβ48, are precursors to the majority of Aβ40 and Aβ42, concomitantly releasing the tripeptides, ITL, VIV, and IAT; and VIT and TVI, respectively. A portion of Aβ42 is processed further to Aβ38, releasing a tetrapeptide, VVIA. The presence of additional multiple minor pathways may reflect labile cleavage activities derived from the conformational flexibility of γ-secretase through molecular interactions. Because these peptide byproducts are not secreted and remain within the cells, they may serve as an indicator that reflects γ-secretase activity more directly than secreted Aβ.

Intramembrane cleavage of transmembrane proteins is a fundamental cellular process to produce important signals that elicit biological responses. These proteolytic events are known as regulated intramembrane proteolysis (RIP). ATF6 and BBF2H7 are transmembrane basic leucine zipper transcription factors and are subjected to RIP by site-1 protease (S1P) and site-2 protease (S2P) sequentially in response to endoplasmic reticulum (ER) stress. However, the detailed mechanisms responsible for RIP of the transcription factors, including the precise cutting sites, are still unknown. In this study, we demonstrated that S1P cleaves BBF2H7 just before the RXXL S1P recognition motif. Conversely, S2P cut at least three different sites in the membrane (next to Leu380, Met381, and Leu385), indicating that S2P cleaves the substrates at variable sites or via a multistep process. Interestingly, we found BBF2H7-derived small peptide (BSP) fragments located between the S1P and S2P cleavage sites in cells exposed to ER stress. Major type of BSP fragments was composed of 45 amino acid including partial transmembrane and luminal regions and easily aggregates like amyloid beta (A beta) protein. These results advance the understanding of poorly characterized ER stress-dependent RIP. Furthermore, the aggregable peptides produced by ER stress could link to the pathophysiology of neurodegenerative disorders.

gamma-Secretase is a multimeric aspartyl protease that cleaves the membrane-spanning region of the beta-carboxyl terminal fragment (beta CTF) generated from beta-amyloid precursor protein. gamma-Secretase defines the generated molecular species of amyloid beta-protein (A beta), a critical molecule in the pathogenesis of Alzheimer's disease (AD). Many therapeutic trials for AD have targeted gamma-secretase. However, in contrast to the great efforts in drug discovery, the enzymatic features and cleavage mechanism of gamma-secretase are poorly understood. Here we review our protein-chemical analyses of the cleavage products generated from beta CTF by gamma-secretase, which revealed that A beta was produced by gamma-secretase through successive cleavages of beta CTF, mainly at three-residue intervals. Two representative product lines were identified. epsilon-Cleavages occur first at Leu49-Val50 and Thr48-Leu49 of beta CTF (in accordance with A beta numbering). Longer generated A beta s, A beta 49 and A beta 48, are precursors to the majority of A beta 40 and A beta 42, concomitantly releasing the tripeptides, ITL, VIV, and IAT; and VIT and TVI, respectively. A portion of A beta 42 is processed further to A beta 38, releasing a tetrapeptide, VVIA. The presence of additional multiple minor pathways may reflect labile cleavage activities derived from the conformational flexibility of gamma-secretase through molecular interactions. Because these peptide byproducts are not secreted and remain within the cells, they may serve as an indicator that reflects gamma-secretase activity more directly than secreted A beta.

Alzheimer´s disease (AD) is a devastating neurodegenerative disease of the elderly population. Genetic evidence strongly suggests that aberrant generation and/or clearance of the neurotoxic amyloid-β peptide (Aβ) is triggering the disease. Aβ is generated from the amyloid precursor protein (APP) by the sequential cleavages of β- and γ-secretase. The latter cleavage by γ-secretase, a unique and fascinating four-component protease complex, occurs in the APP transmembrane domain thereby releasing Aβ species of 37-43 amino acids in length including the longer, highly pathogenic peptides Aβ42 and Aβ43. The lack of a precise understanding of Aβ generation as well as of the functions of other γ-secretase substrates has been one factor underlying the disappointing failure of γ-secretase inhibitors in clinical trials, but on the other side also been a major driving force for structural and in depth mechanistic studies on this key AD drug target in the past few years. Here we review recent breakthroughs in our understanding of how the γ-secretase complex recognizes substrates, of how it binds and processes β-secretase cleaved APP into different Aβ species, as well as the progress made on a question of outstanding interest, namely how clinical AD mutations in the catalytic subunit presenilin and the γ-secretase cleavage region of APP lead to relative increases of Aβ42/43. Finally, we discuss how the knowledge emerging from these studies could be used to therapeutically target this enzyme in a safe way.

We recently reported that NF-κB–mediated inflammation caused by breakpoint cluster region (BCR) is dependent on the α subunit of casein kinase II (CK2α) complex. In the current study, we demonstrate that presenilin 1 (Psen1), which is a catalytic component of the γ-secretase complex and the mutations of which are known to cause familial Alzheimer disease, acts as a scaffold of the BCR–CK2α–p65 complex to induce NF-κB activation. Indeed, Psen1 deficiency in mouse endothelial cells showed a significant reduction of NF-κB p65 recruitment to target gene promoters. Conversely, Psen1 overexpression enhanced reporter activation under NF-κB responsive elements and IL-6 promoter. Furthermore, the transcription of NF-κB target genes was not inhibited by a γ-secretase inhibitor, suggesting that Psen1 regulates NF-κB activation in a manner independent of γ-secretase activity. Mechanistically, Psen1 associated with the BCR–CK2α complex, which is required for phosphorylation of p65 at serine 529. Consistently, TNF-α–induced phosphorylation of p65 at serine 529 was significantly decreased in Psen1-deficient cells. The association of the BCR–CK2α–p65 complex was perturbed in the absence of Psen1. These results suggest that Psen1 functions as a scaffold of the BCR–CK2α–p65 complex and that this signaling cascade could be a novel therapeutic target for various chronic inflammation conditions, including those in Alzheimer disease.

Background: Amyloid-β (Aβ) degradation in brains of Alzheimer disease patients is a crucial focus for the clarification of disease pathogenesis. Nevertheless, the mechanisms underlying Aβ degradation in the human brain remain unclear. Objective: This study aimed to quantify the levels of small C-terminal Aβ fragments generated upon Aβ degradation in human cerebrospinal fluid (CSF). Methods: A fraction containing small peptides was isolated and purified from human CSF by high-pressure liquid chromatography. Degradation products of Aβ C termini were identified and measured by liquid chromatography-tandem mass spectrometry. The C-terminal fragments of Aβ in the conditioned medium of cultured cells transfected with the Swedish variant of βAPP (sw βAPP) were analyzed. These fragments in brains of PS1 I213T knock-in transgenic mice, overexpressing sw βAPP, were also analyzed. Results: The peptide fragments GGVV and GVV, produced by the cleavage of Aβ40, were identified in human CSF as well as in the brains of the transgenic mice and in the conditioned medium of the cultured cells. Relative to Aβ40 levels, GGVV and GVV levels were 7.6 ± 0.81 and 1.5 ± 0.18%, respectively, in human CSF. Levels of the GGVV fragment did not increase by the introduction of genes encoding neprilysin and insulin-degrading enzyme to the cultured cells. Conclusion: Our results indicate that a substantial amount of Aβ40 in human brains is degraded via a neprilysin- or insulin-degrading enzyme-independent pathway.

Aβ42 in cerebrospinal fluid (CSF) is an established biomarker in Alzheimer's disease (AD). The ratio Aβ42 to total Aβ is lowered in CSF of AD patients, which is used for diagnosis of preclinical AD, MCI and AD. However, Aβ42 in peripheral blood is not practically used probably because Aβ42 in blood is not brain derived since βAPP is ubiquitously expressed. Thus, if there is a blood-based biomarker which is brain derived and also reflects the level of brain Aβ42, it can be widely used for screening of Aβ42 accumulation in brain. We discovered APL1β28, an Aβ42 surrogate marker, in human CSF. APL1β28 is an Aβ-like peptide which is produced by BACE and gamma-secretase from APLP1. Notably, APLP1 is expressed only in brain. We showed the APL1β28 ratio (APL1β28/total APL1β) was elevated in CSF samples of familial AD with PS1 pathogenic mutations. Importantly, the APL1β28 ratio was even higher in CSF of sporadic AD and MCI patients than in control CSF. Thus, the CSF APL1β28 ratio may be a biomarker which indicates Aβ42 generation in brain. To utilize this surrogate marker in large clinical trials, we have tried to detect and measure the level of this CNS peptide in peripheral blood. We performed several biochemical treatments to purify plasma APL1β28 and measure the levels by using QTRAP 6500 LC/MS/MS system. We could successfully measure the level of the CNS APL1β28 peptide in plasma. The concentration of APL1β28 is ∼0.4pM, which is much less than that in CSF (∼500pM). Currently, we are analyzing the APL1β28 ratio in CSF and plasma paired samples. We are also investigating correlation between plasma APL1β28 ratio and CSF Ab42 ratio. We have tried to develop an Aβ42 surrogate marker in peripheral blood. We intend to show how and to what degree the plasma APL1β28 ratio correlates with CSF Aβ42 ratio.

γ-secretase inhibitors (GSI) are drugs developed to decrease amyloid-β peptide (Aβ) production by inhibiting intramembranous cleavage of β-amyloid protein precursor (βAPP). However, a large phase 3 trial of semagacestat, a potential non-transition state analog (non-TSA) GSI, in patients with Alzheimer’s disease (AD) was terminated due to unexpected aggravation of cognitive deficits and side effects. Here, we show that some semagacestat effects are clearly different from a phenotype caused by a loss of function of presenilins, core proteins in the γ-secretase complex. Semagacestat increases intracellular byproduct peptides, produced along with Aβ through serial γ-cleavage of βAPP, as well as intracellular long Aβ species, in cell-based and in vivo studies of AD model mice. Other potential non-TSA GSIs, but not L685,458, a TSA GSI, have similar effects. Furthermore, semagacestat inhibits release of de novo intramembranous γ-byproducts to the soluble space. Thus, semagacestat is a pseudo-GSI, and therefore, the semagacestat clinical trial did not truly test the Aβ hypothesis.

To cope with an aging society, development of disease-modifying drugs for Alzheimer's disease (AD) is essential. Currently, only symptomatic treatments that suppress clinical manifestations are available. Amyloid-β42 (Aβ42) is an AD-related pathogenic molecule that triggers development of AD pathology; thus, decreasing Aβ42 in the brain is a promising candidate for AD therapy. Numerous pharmaceutical companies have developed therapeutic drugs against Aβ42, such as β-secretase inhibitors, γ-secretase inhibitors, and anti-Aβ monoclonal antibodies, but in clinical trials for patients with mild to moderate AD, these drugs did not meet the expected endpoints. These results suggest that earlier administration of these drugs to individuals who have not yet developed cognitive decline, but have AD pathological changes in the brain or high risk of developing these changes, may be beneficial. To enable such early treatment, preclinical AD biomarkers are required. In this review, we comment on current AD biomarkers in cerebrospinal fluid and in blood. We also explain CSF/blood APL1β, which is a candidate surrogate marker for Aβ42 in the brain.

Signal-peptide peptidase (SPP) is an intramembrane protease that participates in the production of the mature core protein of hepatitis C virus (HCV). Here we show that SPP inhibition reduces the production of infectious HCV particles and pathogenesis. The immature core protein produced in SPP-knockout cells or by treatment with an SPP inhibitor is quickly degraded by the ubiquitin–proteasome pathway. Oral administration of the SPP inhibitor to transgenic mice expressing HCV core protein (CoreTg) reduces the expression of core protein and ameliorates insulin resistance and liver steatosis. Moreover, the haploinsufficiency of SPP in CoreTg has similar effects. TRC8, an E3 ubiquitin ligase, is required for the degradation of the immature core protein. The expression of the HCV core protein alters endoplasmic reticulum (ER) distribution and induces ER stress in SPP/TRC8 double-knockout cells. These data suggest that HCV utilizes SPP cleavage to circumvent the induction of ER stress in host cells.

Selected/multiple reaction monitoring (SRM/MRM) has been widely used for the quantification of specific proteins/peptides, although it is still challenging to quantitate low abundant proteins/peptides in complex samples such as plasma/serum. To overcome this problem, enrichment of target proteins/peptides is needed, such as immunoprecipitation; however, this is labor-intense and generation of antibodies is highly expensive. In this study, we attempted to quantify plasma low abundant APLP1-derived Aβ-like peptides (APL1β), a surrogate marker for Alzheimer's disease, by SRM/MRM using stable isotope-labeled reference peptides without immunoaffinity enrichment. A combination of Cibacron Blue dye mediated albumin removal and acetonitrile extraction followed by C18-strong cation exchange multi-StageTip purification was used to deplete plasma proteins and unnecessary peptides. Optimal and validated precursor ions to fragment ion transitions of APL1β were developed on a triple quadruple mass spectrometer, and the nanoliquid chromatography gradient for peptide separation was optimized to minimize the biological interference of plasma. Using the stable isotope-labeled (SI) peptide as an internal control, absolute concentrations of plasma APL1β peptide could be quantified as several hundred amol/mL. To our knowledge, this is the lowest detection level of endogenous plasma peptide quantified by SRM/MRM.

BACKGROUND:Presenilin 1 (PS1) mutations associated with familial Alzheimer disease (FAD) generally increase the amyloid-β 42 (Aβ42) to Aβ40 ratio secreted in cultured cells. Some of these mutants reduce the secretion of Aβ40 rather than increase that of Aβ42. Since it has been difficult to estimate Aβ42 secretion in brains of PS1-FAD patients due to substantial Aβ42 accumulation, it remains unknown whether the enhanced Aβ42 to Aβ40 ratio in brains of FAD patients is caused by elevated Aβ42 secretion or by reduced secretion of Aβ40.OBJECTIVE/METHODS:Cerebrospinal fluids (CSF) of PS1-FAD patients and neurological control patients (controls) were collected. Levels of CSF amyloid precursor-like protein-1-derived Aβ-like peptide (APL1β), including APL1β28, an Aβ42 surrogate marker, were quantified by liquid chromatography tandem mass spectrometry, and Aβ42 secretion in the brain was estimated.RESULTS:The relative ratio of CSF APL1β28 to total APL1β was higher in PS1-FAD patients than in controls. Importantly, CSF APL1β28 was not significantly higher. However, C-terminally shorter CSF APL1β25 and APL1β27 were significantly lower in PS1-FAD patients and, as expected, so were CSF Aβ40 and Aβ42.CONCLUSION:A higher relative ratio of the CSF Aβ42 surrogate in PS1-FAD patients is not due to its increase in CSF, suggesting that massive Aβ42 accumulation in the PS1-FAD brain occurs without an apparent increase in Aβ42 secretion.

We recently demonstrated that endoplasmic reticulum (ER) stress induces sigma-1 receptor (Sig-1R) expression through the PERK pathway, which is one of the cell’s responses to ER stress. In addition, it has been demonstrated that induction of Sig-1R can repress cell death signaling. Fluvoxamine (Flv) is a selective serotonin reuptake inhibitor (SSRI) with a high affinity for Sig-1R. In the present study, we show that treatment of neuroblastoma cells with Flv induces Sig-1R expression by increasing ATF4 translation directly, through its own activation, without involvement of the PERK pathway. The Flv-mediated induction of Sig-1R prevents neuronal cell death resulting from ER stress. Moreover, Flv-induced ER stress resistance reduces the infarct area in mice after focal cerebral ischemia. Thus, Flv, which is used frequently in clinical practice, can alleviate ER stress. This suggests that Flv could be a feasible therapy for cerebral diseases caused by ER stress.

Alzheimer's disease (AD) is characterized by the accumulation of amyloid-β (Aβ). The genes that govern this process, however, have remained elusive. To this end, we combined distinct mouse strains with transcriptomics to directly identify disease-relevant genes. We show that AD model mice (APP-Tg) with DBA/2 genetic backgrounds have significantly lower levels of Aβ accumulation compared with SJL and C57BL/6 mice. We then applied brain transcriptomics to reveal the genes in DBA/2 that suppress Aβ accumulation. To avoid detecting secondarily affected genes by Aβ, we used non-Tg mice in the absence of Aβ pathology and selected candidate genes differently expressed in DBA/2 mice. Additional transcriptome analysis of APP-Tg mice with mixed genetic backgrounds revealed kinesin light chain-1 (Klc1) as an Aβ modifier, indicating a role for intracellular trafficking in Aβ accumulation. Aβ levels correlated with the expression levels of Klc1 splice variant E and the genotype of Klc1 in these APP-Tg mice. In humans, the expression levels of KLC1 variant E in brain and lymphocyte were significantly higher in AD patients compared with unaffected individuals. Finally, functional analysis using neuroblastoma cells showed that overexpression or knockdown of KLC1 variant E increases or decreases the production of Aβ, respectively. The identification of KLC1 variant E suggests that the dysfunction of intracellular trafficking is a causative factor of Aβ pathology. This unique combination of distinct mouse strains and model mice with transcriptomics is expected to be useful for the study of genetic mechanisms of other complex diseases.