Purification and cloning of amyloid precursor protein beta-secretase from human brain

Proteolytic processing of the amyloid precursor protein (APP) generates amyloid beta (Abeta) peptide, which is thought to be causal for the pathology and subsequent cognitive decline in Alzheimer's disease. Cleavage by beta-secretase at the amino terminus of the Abeta peptide sequence, between residues 671 and 672 of APP, leads to the generation and extracellular release of beta-cleaved soluble APP, and a corresponding cell-associated carboxy-terminal fragment. Cleavage of the C-terminal fragment by gamma-secretase(s) leads to the formation of Abeta. The pathogenic mutation K670M671-->N670L671 at the beta-secretase cleavage site in APP, which was discovered in a Swedish family with familial Alzheimer's disease, leads to increased beta-secretase cleavage of the mutant substrate. Here we describe a membrane-bound enzyme activity that cleaves full-length APP at the beta-secretase cleavage site, and find it to be the predominant beta-cleavage activity in human brain. We have purified this enzyme activity to homogeneity from human brain using a new substrate analogue inhibitor of the enzyme activity, and show that the purified enzyme has all the properties predicted for beta-secretase. Cloning and expression of the enzyme reveals that human brain beta-secretase is a new membrane-bound aspartic proteinase.     

Membrane-anchored aspartyl protease with Alzheimer's disease beta-secretase activity

Mutations in the gene encoding the amyloid protein precursor (APP) cause autosomal dominant Alzheimer's disease. Cleavage of APP by unidentified proteases, referred to as beta- and gamma-secretases, generates the amyloid beta-peptide, the main component of the amyloid plaques found in Alzheimer's disease patients. The disease-causing mutations flank the protease cleavage sites in APP and facilitate its cleavage. Here we identify a new membrane-bound aspartyl protease (Asp2) with beta-secretase activity. The Asp2 gene is expressed widely in brain and other tissues. Decreasing the expression of Asp2 in cells reduces amyloid beta-peptide production and blocks the accumulation of the carboxy-terminal APP fragment that is created by beta-secretase cleavage. Solubilized Asp2 protein cleaves a synthetic APP peptide substrate at the beta-secretase site, and the rate of cleavage is increased tenfold by a mutation associated with early-onset Alzheimer's disease in Sweden. Thus, Asp2 is a new protein target for drugs that are designed to block the production of amyloid beta-peptide peptide and the consequent formation of amyloid plaque in Alzheimer's disease.     

Degeneration in vitro of post-mitotic neurons overexpressing the Alzheimer amyloid protein precursor.

A pathological hallmark of Alzheimer's disease is the deposition of amyloid fibrils in the brain. The principal component of amyloid fibrils is beta/A4 amyloid protein, which can be generated by the aberrant processing of a large membrane-bound glycoprotein, the beta/A4 amyloid protein precursor (APP)3. To test whether overexpression of APP generates abnormally processed derivatives that affect the viability of neurons, we stably transfected full-length human APP complementary DNA into murine embryonal carcinoma P19 cells. These cells differentiate into post-mitotic neurons and astrocytes after exposure to retinoic acid. When differentiation of the APP cDNA-transfected P19 cells was induced, all neurons showed severe degenerative changes and disappeared within a few days. The degenerating neurons contained large amounts of APP derivatives that were truncated at the amino terminus and encompassed the entire beta/A4 domain. These results suggest that post-mitotic neurons are vulnerable to overexpressed APP, which undergoes aberrant processing to generate potentially amyloidogenic fragments.     

The prolyl isomerase Pin1 regulates amyloid precursor protein processing and amyloid-beta production

Neuropathological hallmarks of Alzheimer's disease are neurofibrillary tangles composed of tau and neuritic plaques comprising amyloid-beta peptides (Abeta) derived from amyloid precursor protein (APP), but their exact relationship remains elusive. Phosphorylation of tau and APP on certain serine or threonine residues preceding proline affects tangle formation and Abeta production in vitro. Phosphorylated Ser/Thr-Pro motifs in peptides can exist in cis or trans conformations, the conversion of which is catalysed by the Pin1 prolyl isomerase. Pin1 has been proposed to regulate protein function by accelerating conformational changes, but such activity has never been visualized and the biological and pathological significance of Pin1 substrate conformations is unknown. Notably, Pin1 is downregulated and/or inhibited by oxidation in Alzheimer's disease neurons, Pin1 knockout causes tauopathy and neurodegeneration, and Pin1 promoter polymorphisms appear to associate with reduced Pin1 levels and increased risk for late-onset Alzheimer's disease. However, the role of Pin1 in APP processing and Abeta production is unknown. Here we show that Pin1 has profound effects on APP processing and Abeta production. We find that Pin1 binds to the phosphorylated Thr 668-Pro motif in APP and accelerates its isomerization by over 1,000-fold, regulating the APP intracellular domain between two conformations, as visualized by NMR. Whereas Pin1 overexpression reduces Abeta secretion from cell cultures, knockout of Pin1 increases its secretion. Pin1 knockout alone or in combination with overexpression of mutant APP in mice increases amyloidogenic APP processing and selectively elevates insoluble Abeta42 (a major toxic species) in brains in an age-dependent manner, with Abeta42 being prominently localized to multivesicular bodies of neurons, as shown in Alzheimer's disease before plaque pathology. Thus, Pin1-catalysed prolyl isomerization is a novel mechanism to regulate APP processing and Abeta production, and its deregulation may link both tangle and plaque pathologies. These findings provide new insight into the pathogenesis and treatment of Alzheimer's disease.     

Substrate-targeting gamma-secretase modulators

Selective lowering of Abeta42 levels (the 42-residue isoform of the amyloid-beta peptide) with small-molecule gamma-secretase modulators (GSMs), such as some non-steroidal anti-inflammatory drugs, is a promising therapeutic approach for Alzheimer's disease. To identify the target of these agents we developed biotinylated photoactivatable GSMs. GSM photoprobes did not label the core proteins of the gamma-secretase complex, but instead labelled the beta-amyloid precursor protein (APP), APP carboxy-terminal fragments and amyloid-beta peptide in human neuroglioma H4 cells. Substrate labelling was competed by other GSMs, and labelling of an APP gamma-secretase substrate was more efficient than a Notch substrate. GSM interaction was localized to residues 28-36 of amyloid-beta, a region critical for aggregation. We also demonstrate that compounds known to interact with this region of amyloid-beta act as GSMs, and some GSMs alter the production of cell-derived amyloid-beta oligomers. Furthermore, mutation of the GSM binding site in the APP alters the sensitivity of the substrate to GSMs. These findings indicate that substrate targeting by GSMs mechanistically links two therapeutic actions: alteration in Abeta42 production and inhibition of amyloid-beta aggregation, which may synergistically reduce amyloid-beta deposition in Alzheimer's disease. These data also demonstrate the existence and feasibility of 'substrate targeting' by small-molecule effectors of proteolytic enzymes, which if generally applicable may significantly broaden the current notion of 'druggable' targets.     

A presenilin-1-dependent gamma-secretase-like protease mediates release of Notch intracellular domain

Signalling through the receptor protein Notch, which is involved in crucial cell-fate decisions during development, requires ligand-induced cleavage of Notch. This cleavage occurs within the predicted transmembrane domain, releasing the Notch intracellular domain (NICD), and is reminiscent of gamma-secretase-mediated cleavage of beta-amyloid precursor protein (APP), a critical event in the pathogenesis of Alzheimer's disease. A deficiency in presenilin-1 (PS1) inhibits processing of APP by gamma-secretase in mammalian cells, and genetic interactions between Notch and PS1 homologues in Caenorhabditis elegans indicate that the presenilins may modulate the Notch signalling pathway. Here we report that, in mammalian cells, PS1 deficiency also reduces the proteolytic release of NICD from a truncated Notch construct, thus identifying the specific biochemical step of the Notch signalling pathway that is affected by PS1. Moreover, several gamma-secretase inhibitors block this same step in Notch processing, indicating that related protease activities are responsible for cleavage within the predicted transmembrane domains of Notch and APP. Thus the targeting of gamma-secretase for the treatment of Alzheimer's disease may risk toxicity caused by reduced Notch signalling.     

A specific amyloid-beta protein assembly in the brain impairs memory

Memory function often declines with age, and is believed to deteriorate initially because of changes in synaptic function rather than loss of neurons. Some individuals then go on to develop Alzheimer's disease with neurodegeneration. Here we use Tg2576 mice, which express a human amyloid-beta precursor protein (APP) variant linked to Alzheimer's disease, to investigate the cause of memory decline in the absence of neurodegeneration or amyloid-beta protein amyloidosis. Young Tg2576 mice (< 6 months old) have normal memory and lack neuropathology, middle-aged mice (6-14 months old) develop memory deficits without neuronal loss, and old mice (> 14 months old) form abundant neuritic plaques containing amyloid-beta (refs 3-6). We found that memory deficits in middle-aged Tg2576 mice are caused by the extracellular accumulation of a 56-kDa soluble amyloid-beta assembly, which we term Abeta*56 (Abeta star 56). Abeta*56 purified from the brains of impaired Tg2576 mice disrupts memory when administered to young rats. We propose that Abeta*56 impairs memory independently of plaques or neuronal loss, and may contribute to cognitive deficits associated with Alzheimer's disease.     

GSK-3alpha regulates production of Alzheimer's disease amyloid-beta peptides

Alzheimer's disease is associated with increased production and aggregation of amyloid-beta (Abeta) peptides. Abeta peptides are derived from the amyloid precursor protein (APP) by sequential proteolysis, catalysed by the aspartyl protease BACE, followed by presenilin-dependent gamma-secretase cleavage. Presenilin interacts with nicastrin, APH-1 and PEN-2 (ref. 6), all of which are required for gamma-secretase function. Presenilins also interact with alpha-catenin, beta-catenin and glycogen synthase kinase-3beta (GSK-3beta), but a functional role for these proteins in gamma-secretase activity has not been established. Here we show that therapeutic concentrations of lithium, a GSK-3 inhibitor, block the production of Abeta peptides by interfering with APP cleavage at the gamma-secretase step, but do not inhibit Notch processing. Importantly, lithium also blocks the accumulation of Abeta peptides in the brains of mice that overproduce APP. The target of lithium in this setting is GSK-3alpha, which is required for maximal processing of APP. Since GSK-3 also phosphorylates tau protein, the principal component of neurofibrillary tangles, inhibition of GSK-3alpha offers a new approach to reduce the formation of both amyloid plaques and neurofibrillary tangles, two pathological hallmarks of Alzheimer's disease.     

A functional barrier to movement of lipids in polarized neurons.

In polarized neurons, axons and dendrites perform different functions, which are reflected in their different molecular organization. Studies on the sorting of viral and endogenous glycoproteins in epithelial cells and hippocampal neurons suggest that there may be similarities in the mechanism of sorting in these two cell types. The mechanisms that maintain the distinct composition of the two plasma membrane domains in these two cell types must, however, be different. We have proposed the existence of a functional barrier at the axonal hillock/initial segment which prevents the intermixing of membrane constituents. Here we test this hypothesis by fusing liposomes containing fluorescent phospholipids into the plasma membrane of polarized hippocampal cells in culture. Fusion was induced by lowering the pH and mediated by influenza virus haemagglutinin expressed on the axonal surface of neurons infected with fowl plague virus. Labelling was found exclusively on axons after fusion. Although the fused lipids were mobile on the axonal membrane, no labelling was detected on the cell body and dendritic surfaces. These results suggest that there is a diffusion barrier at the axonal hillock/initial segment which maintains the compositional differences between the axonal and somatodendritic domains.     

Two transmembrane aspartates in presenilin-1 required for presenilin endoproteolysis and gamma-secretase activity

Accumulation of the amyloid-beta protein (Abeta) in the cerebral cortex is an early and invariant event in the pathogenesis of Alzheimer's disease. The final step in the generation of Abeta from the beta-amyloid precursor protein is an apparently intramembranous proteolysis by the elusive gamma-secretase(s). The most common cause of familial Alzheimer's disease is mutation of the genes encoding presenilins 1 and 2, which alters gamma-secretase activity to increase the production of the highly amyloidogenic Abeta42 isoform. Moreover, deletion of presenilin-1 in mice greatly reduces gamma-secretase activity, indicating that presenilin-1 mediates most of this proteolytic event. Here we report that mutation of either of two conserved transmembrane (TM) aspartate residues in presenilin-1, Asp 257 (in TM6) and Asp 385 (in TM7), substantially reduces Abeta production and increases the amounts of the carboxy-terminal fragments of beta-amyloid precursor protein that are the substrates of gamma-secretase. We observed these effects in three different cell lines as well as in cell-free microsomes. Either of the Asp --> Ala mutations also prevented the normal endoproteolysis of presenilin-1 in the TM6 --> TM7 cytoplasmic loop. In a functional presenilin-1 variant (carrying a deletion in exon 9) that is associated with familial Alzheimer's disease and which does not require this cleavage, the Asp 385 --> Ala mutation still inhibited gamma-secretase activity. Our results indicate that the two transmembrane aspartate residues are critical for both presenilin-1 endoproteolysis and gamma-secretase activity, and suggest that presenilin 1 is either a unique diaspartyl cofactor for gamma-secretase or is itself gamma-secretase, an autoactivated intramembranous aspartyl protease.     

A subset of NSAIDs lower amyloidogenic Abeta42 independently of cyclooxygenase activity.

Epidemiological studies have documented a reduced prevalence of Alzheimer's disease among users of nonsteroidal anti-inflammatory drugs (NSAIDs). It has been proposed that NSAIDs exert their beneficial effects in part by reducing neurotoxic inflammatory responses in the brain, although this mechanism has not been proved. Here we report that the NSAIDs ibuprofen, indomethacin and sulindac sulphide preferentially decrease the highly amyloidogenic Abeta42 peptide (the 42-residue isoform of the amyloid-beta peptide) produced from a variety of cultured cells by as much as 80%. This effect was not seen in all NSAIDs and seems not to be mediated by inhibition of cyclooxygenase (COX) activity, the principal pharmacological target of NSAIDs. Furthermore, short-term administration of ibuprofen to mice that produce mutant beta-amyloid precursor protein (APP) lowered their brain levels of Abeta42. In cultured cells, the decrease in Abeta42 secretion was accompanied by an increase in the Abeta(1-38) isoform, indicating that NSAIDs subtly alter gamma-secretase activity without significantly perturbing other APP processing pathways or Notch cleavage. Our findings suggest that NSAIDs directly affect amyloid pathology in the brain by reducing Abeta42 peptide levels independently of COX activity and that this Abeta42-lowering activity could be optimized to selectively target the pathogenic Abeta42 species.     

GSK-3α/β kinases and amyloid production in vivo

Arising from C. J. Phiel, C. A. Wilson, V. M.-Y. Lee & P. S. Klein 423, 435-439 (2003)A major unresolved issue in Alzheimer's disease is identifying the mechanisms that regulate proteolytic processing of amyloid precursor protein (APP)-glycogen synthase kinase-3 (GSK-3) isozymes are thought to be important in this regulation. Phiel et al. proposed that GSK-3α, but not GSK-3β, controls production of amyloid. We analysed the proteolytic processing of mouse and human APP in mouse brain in vivo in five different genetic and viral models. Our data do not yield evidence for either GSK-3α-mediated or GSK-3β-mediated control of APP processing in brain in vivo.     

HLA-DR molecules from an antigen-processing mutant cell line are associated with invariant chain peptides.

The invariant chain, which associates with the major histocompatibility complex (MHC) class II molecules in the endoplasmic reticulum, serves two functions important in antigen processing. First, it prevents class II molecules from binding peptides in the early stages of intracellular transport. Second, it contains a cytoplasmic signal that targets the class II-invariant chain complex to an acidic endosomal compartment. Proteolytic cleavage and subsequent dissociation of the invariant chain then occurs, allowing peptides derived from endocytosed proteins to bind to released class II molecules before their expression at the cell surface. Certain human cell lines that are mutant in one or more MHC-linked genes are defective in class II-restricted antigen processing. Here we show that in transfectants of one of these cell lines, T2, this deficiency results in the association of a large proportion of class II molecules with a nested set of invariant-chain-derived peptides (class II-associated invariant chain peptides, or CLIP). HLA-DR3 molecules isolated from T2 transfectants can be efficiently loaded with antigenic peptides by exposure to a low pH in vitro, perhaps reflecting the in vivo conditions in which peptides associate with class II molecules. Addition of synthetic CLIP inhibits the loading process, indicating that CLIP may define the region of the invariant chain responsible for obstructing the class II binding site.     

Targeting of cell-surface beta-amyloid precursor protein to lysosomes: alternative processing into amyloid-bearing fragments.

Progressive cerebral deposition of the amyloid beta-peptide is an early and invariant feature of Alzheimer's disease. The beta-peptide is released by proteolytic cleavages from the beta-amyloid precursor protein (beta APP), a membrane-spanning glycoprotein expressed in most mammalian cells. Normal secretion of beta APP involves a cleavage in the beta-peptide region, releasing the soluble extramembranous portion and retaining a 10K C-terminal fragment in the membrane. Because this secretory pathway precludes beta-amyloid formation, we searched for an alternative proteolytic processing pathway that can generate beta-peptide-bearing fragments from full-length beta APP. Incubation of living human endothelial cells with a beta APP antibody revealed reinternalization of mature beta APP from the cell surface and its targeting to endosomes/lysosomes. After cell-surface biotinylation, full-length biotinylated beta APP was recovered inside the cells. Purification of lysosomes directly demonstrated the presence of mature beta APP and an extensive array of beta-peptide-containing proteolytic products. Our results define a second processing pathway for beta APP and suggest that it may be responsible for generating amyloid-bearing fragments in Alzheimer's disease.     

Ebola virus entry requires the cholesterol transporter Niemann-Pick C1

Infections by the Ebola and Marburg filoviruses cause a rapidly fatal haemorrhagic fever in humans for which no approved antivirals are available. Filovirus entry is mediated by the viral spike glycoprotein (GP), which attaches viral particles to the cell surface, delivers them to endosomes and catalyses fusion between viral and endosomal membranes. Additional host factors in the endosomal compartment are probably required for viral membrane fusion; however, despite considerable efforts, these critical host factors have defied molecular identification. Here we describe a genome-wide haploid genetic screen in human cells to identify host factors required for Ebola virus entry. Our screen uncovered 67 mutations disrupting all six members of the homotypic fusion and vacuole protein-sorting (HOPS) multisubunit tethering complex, which is involved in the fusion of endosomes to lysosomes, and 39 independent mutations that disrupt the endo/lysosomal cholesterol transporter protein Niemann-Pick C1 (NPC1). Cells defective for the HOPS complex or NPC1 function, including primary fibroblasts derived from human Niemann-Pick type C1 disease patients, are resistant to infection by Ebola virus and Marburg virus, but remain fully susceptible to a suite of unrelated viruses. We show that membrane fusion mediated by filovirus glycoproteins and viral escape from the vesicular compartment require the NPC1 protein, independent of its known function in cholesterol transport. Our findings uncover unique features of the entry pathway used by filoviruses and indicate potential antiviral strategies to combat these deadly agents.     

Polarized sorting of glypiated proteins in hippocampal neurons.

Our recent studies suggested that neurons and epithelial cells sort viral glycoproteins in a similar manner. The apical influenza virus haemagglutinin was preferentially delivered to the axon of hippocampal neurons in culture, whereas the basolateral vesicular stomatitis virus glycoprotein was sorted to the dendrites. To investigate whether other membrane proteins showed similar sorting in neurons and epithelial cells, we have analysed the localization of a glypiated (glycosylphosphatidylinositol anchored) protein, Thy-1, in hippocampal neurons in culture. In MDCK and other epithelial cells, endogenous glycosylphosphatidylinositol (GPI)-anchored proteins, as well as mutated exogenous proteins containing the GPI-attachment signal, undergo preferential delivery to the apical surface. This polarized sorting of GPI-anchored proteins has been proposed to occur by the same mechanisms as the sorting of glycolipids to the apical surface. We report here that the neuronal GPI-protein Thy-1 is present in hippocampal neurons in culture and is exclusively located on the axonal surface. This finding further strengthens our hypothesis that the mechanisms of sorting of surface components may be similar in neurons and epithelial cells.     

Dissection of COPI and Arf1 dynamics in vivo and role in Golgi membrane transport

Cytosolic coat proteins that bind reversibly to membranes have a central function in membrane transport within the secretory pathway. One well-studied example is COPI or coatomer, a heptameric protein complex that is recruited to membranes by the GTP-binding protein Arf1. Assembly into an electron-dense coat then helps in budding off membrane to be transported between the endoplasmic reticulum (ER) and Golgi apparatus. Here we propose and corroborate a simple model for coatomer and Arf1 activity based on results analysing the distribution and lifetime of fluorescently labelled coatomer and Arf1 on Golgi membranes of living cells. We find that activated Arf1 brings coatomer to membranes. However, once associated with membranes, Arf1 and coatomer have different residence times: coatomer remains on membranes after Arf1-GTP has been hydrolysed and dissociated. Rapid membrane binding and dissociation of coatomer and Arf1 occur stochastically, even without vesicle budding. We propose that this continuous activity of coatomer and Arf1 generates kinetically stable membrane domains that are connected to the formation of COPI-containing transport intermediates. This role for Arf1/coatomer might provide a model for investigating the behaviour of other coat protein systems within cells.     

Increased phosphorylation of ribosomal protein S6 following microinjection of insulin receptor-kinase into Xenopus oocytes

The protein products of several transforming retroviruses as well as the receptors for several hormones and growth factors, including insulin, have been shown to possess a protein kinase activity in vitro specific for tyrosine residues in protein substrates, including themselves. In the case of pp60src and the insulin receptor, autophosphorylation activates the tyrosine kinase activity towards exogenous substrates. Experiments indicate that, in vivo, many of these viruses or growth factors induce an increase in cellular phosphotyrosine, as well as an increase in the phosphorylation of serine residues on proteins, including ribosomal protein S6. It seems likely that some of the effects of insulin might be mediated by phosphorylation of intracellular substrates by its receptor. As the beta subunit of the receptor is a transmembrane protein, such phosphorylation could occur either while the receptor is still in the membrane or after its internalization. In various cell systems, internalized receptors are degraded, reshuttled back to the plasmalemma or maintained in a separate compartment before reinsertion in the membrane; shuttling of the insulin receptor could provide the opportunity for it to phosphorylate various intracellular components as part of its mechanism of signal transduction. To approach directly the question of whether the receptor can elicit a signal while acting at an intracellular location, we have microinjected Xenopus oocytes with the insulin receptor kinase. The results indicate that an S6 protein-serine kinase is stimulated or an S6 protein-serine phosphatase inhibited by the activity of the insulin receptor, supporting the concept that the insulin receptor acting within the cell can elicit a biological response.     

Probing sporadic and familial Alzheimer's disease using induced pluripotent stem cells

Our understanding of Alzheimer's disease pathogenesis is currently limited by difficulties in obtaining live neurons from patients and the inability to model the sporadic form of the disease. It may be possible to overcome these challenges by reprogramming primary cells from patients into induced pluripotent stem cells (iPSCs). Here we reprogrammed primary fibroblasts from two patients with familial Alzheimer's disease, both caused by a duplication of the amyloid-β precursor protein gene (APP; termed APP(Dp)), two with sporadic Alzheimer's disease (termed sAD1, sAD2) and two non-demented control individuals into iPSC lines. Neurons from differentiated cultures were purified with fluorescence-activated cell sorting and characterized. Purified cultures contained more than 90% neurons, clustered with fetal brain messenger RNA samples by microarray criteria, and could form functional synaptic contacts. Virtually all cells exhibited normal electrophysiological activity. Relative to controls, iPSC-derived, purified neurons from the two APP(Dp) patients and patient sAD2 exhibited significantly higher levels of the pathological markers amyloid-β(1-40), phospho-tau(Thr?231) and active glycogen synthase kinase-3β (aGSK-3β). Neurons from APP(Dp) and sAD2 patients also accumulated large RAB5-positive early endosomes compared to controls. Treatment of purified neurons with β-secretase inhibitors, but not γ-secretase inhibitors, caused significant reductions in phospho-Tau(Thr?231) and aGSK-3β levels. These results suggest a direct relationship between APP proteolytic processing, but not amyloid-β, in GSK-3β activation and tau phosphorylation in human neurons. Additionally, we observed that neurons with the genome of one sAD patient exhibited the phenotypes seen in familial Alzheimer's disease samples. More generally, we demonstrate that iPSC technology can be used to observe phenotypes relevant to Alzheimer's disease, even though it can take decades for overt disease to manifest in patients.