Intracellular generation and accumulation of amyloid beta-peptide terminating at amino acid 42

Amyloid beta-peptide (Abeta) is known to accumulate in senile plaques of Alzheimer's disease (AD) patients and is now widely believed to play a major role in the disease. Two populations of peptides occur terminating either at amino acid 40 or at amino acid 42 (Abeta1-40 and Abeta1-42). Alternative N-terminal cleavages produce additional heterogeneity (Abetax-40 and Abetax-42). Peptides terminating at amino acid 42 are believed to be the major player in sporadic AD as well as familial AD (FAD). Whereas the cellular mechanism for the generation of Abeta terminating at amino acid 40 is well understood, very little is known about the cleavage of Abeta after amino acid 42. By using two independent methods we demonstrate intracellular Abeta1-42 as well as Abetax-42 but less Abetax-40 and Abeta1-40 in kidney 293 cells stably transfected with wild type beta-amyloid precursor protein (betaAPP) or the FAD-associated Val/Gly mutation. Moreover, retention of betaAPP within the endoplasmic reticulum (ER) by treatment with brefeldin A does not block the cleavage at amino acid 42 but results in an increased production of all species of Abeta terminating at amino acid 42. This indicates that the cleavage after amino acid 42 can occur within the ER. Treatment of cells with monensin, which blocks transport of (betaAPP) within the Golgi causes a marked accumulation of intracellular Abetax-42 and Abetax-40. Therefore these experiments indicate that the gamma-secretase cleavage of Abeta after amino acid 42 can occur within the ER and later within the secretory pathway within the Golgi. Moreover inhibition of reinternalization by cytoplasmic deletions of betaAPP as well as inhibition of intracellular acidification by NH4Cl does not block intracellular Abeta1-42 or Abetax-42 production.     

Presenilin 1 regulates the processing of beta-amyloid precursor protein C-terminal fragments and the generation of amyloid beta-protein in endoplasmic reticulum and Golgi

Progressive cerebral deposition of the amyloid beta-protein (Abeta) is believed to play a pivotal role in the pathogenesis of Alzheimer's disease (AD). The highly amyloidogenic 42-residue form of Abeta (Abeta42) is the first species to be deposited in both sporadic and familial AD. Mutations in two familial AD-linked genes, presenilins 1 (PS1) and 2 (PS2), selectively increase the production of Abeta42 in cultured cells and the brains of transgenic mice, and gene deletion of PS1 shows that it is required for normal gamma-secretase cleavage of the beta-amyloid precursor protein (APP) to generate Abeta. To establish the subcellular localization of the PS1 regulation of APP processing to Abeta, fibroblasts from PS1 wild-type (wt) or knockout (KO) embryos as well as Chinese hamster ovary (CHO) cells stably transfected with wt or mutant PS1 were subjected to subcellular fractionation on discontinuous Iodixanol gradients. APP C-terminal fragments (CTF) were markedly increased in both endoplasmic reticulum- (ER-) and Golgi-rich fractions of fibroblasts from KO mice; moreover, similar increases were documented directly in KO brain tissue. No change in the subcellular distribution of full-length APP was detectable in fibroblasts lacking PS1. In CHO cells, a small portion of APP, principally the N-glycosylated isoform, formed complexes with PS1 in both ER- and Golgi-rich fractions, as detected by coimmunoprecipitation. When the same fractions were analyzed by enzyme-linked immunosorbent assays for Abetatotal and Abeta42, Abeta42 was the major Abeta species in the ER fraction (Abeta42:Abetatotal ratio 0.5-1.0), whereas absolute levels of both Abeta42 and Abeta40 were higher in the Golgi fraction and the Abeta42:Abetatoal ratio was 0.05-0.16 there. Mutant PS1 significantly increased Abeta42 levels in the Golgi fraction. Our results indicate PS1 and APP can interact in the ER and Golgi, where PS1 is required for proper gamma-secretase processing of APP CTFs, and that PS1 mutations augment Abeta42 levels principally in Golgi-like vesicles.     

Cleavage of Alzheimer's amyloid precursor protein (APP) by secretases occurs after O-glycosylation of APP in the protein secretory pathway. Identification of intracellular compartments in which APP cleavage occurs without using toxic agents that interfere with protein metabolism

beta-Amyloid peptide (Abeta) is a principal component of parenchymal amyloid deposits in Alzheimer's disease. Abeta is derived from amyloid precursor protein (APP) by proteolytic cleavage. APP is subject to N- and O-glycosylation and potential tyrosine sulfation, following protein synthesis, and is then thought to be cleaved in an intracellular secretory pathway after or during these post-translational modifications. Studies utilizing agents that affect a series of steps in the protein secretory pathway have identified the possible intracellular sites of APP cleavage and Abeta generation within the protein secretory pathway. In the present study, using cells with normal protein metabolism, but expressing mutant APP with defective O-glycosylation, we demonstrated that the majority of APP cleavage by alpha-, beta-, and gamma-secretases occurs after O-glycosylation. Cells expressing the mutant APP noticeably decreased the generation of the intracellular APP carboxyl-terminal fragment (alphaAPPCOOH), a product of alpha-secretase, and both Abeta40 and Abeta42 in medium, a product of beta- and gamma-secretases. Furthermore, we found that the cells accumulate the mutant APP in intracellular reticular compartments such as the endoplasmic reticulum. Agents that could ambiguously affect the function of specific intracellular organelles and that may be toxic were not used. The present results indicate that APP is cleaved by alpha-, beta-, and gamma-secretases in step(s) during the transport of APP through Golgi complex, where O-glycosylation occurs, or in compartments subsequent to trans-Golgi of the APP secretory pathway.     

The profile of soluble amyloid beta protein in cultured cell media. Detection and quantification of amyloid beta protein and variants by immunoprecipitation-mass spectrometry

To study the metabolism of amyloid beta protein (Abeta) in Alzheimer's disease, we have developed a new approach for analyzing the profile of soluble Abeta and its variants. In the present method, Abeta and its variants are immuno-isolated with Abeta-specific monoclonal antibodies. The identities of the Abeta variants are determined by measuring their molecular masses using matrix-assisted laser desorption ionization time-of-flight mass spectrometry. The levels of Abeta variants are determined by their relative peak intensities in mass spectrometric measurements by comparison with internal standards of known identities and concentrations. We used this method to examine the Abeta species in conditioned media of mouse neuroblastoma cells transfected with cDNAs encoding wild type or mutant human amyloid precursor protein. In addition to human Abeta-(1-40) and Abeta-(1-42), more than 40 different human Abeta variants were identified. Endogenous murine Abeta and its variants were also identified by this approach. The present approach is a new and sensitive method to characterize the profile of soluble Abeta in conditioned media and biological fluids. Furthermore, it allows direct measurement of each individual peptide in a peptide mixture and provides comprehensive information on the identity and concentration of Abeta and Abeta variants.     

Protein kinase C activation increases release of secreted amyloid precursor protein without decreasing Abeta production in human primary neuron cultures

Overexpression and altered metabolism of amyloid precursor protein (APP) resulting in increased 4 kDa amyloid beta peptide (Abeta) production are believed to play a major role in Alzheimer's disease (AD). Therefore, reducing Abeta production in the brain is a possible therapy for AD. Because AD pathology is fairly restricted to the CNS of humans, we have established human cerebral primary neuron cultures to investigate the metabolism of APP. In many cell lines and rodent primary neuron cultures, phorbol ester activation of protein kinase C (PKC) increases the release of the secreted large N-terminal fragment of amyloid precursor protein (sAPP) and decreases Abeta release (; ; ). In contrast, we find that PKC activation in human primary neurons increases the rate of sAPP release and the production of APP C-terminal fragments and 4 kDa Abeta. Our results indicate species- and cell type-specific regulation of APP metabolism. Therefore, our results curtail the use of PKC activators in controlling human brain Abeta levels.     

Endoplasmic reticulum chaperones GRP78 and calreticulin prevent oxidative stress, Ca2+ disturbances, and cell death in renal epithelial cells

Activation of stress response genes can impart cellular tolerance to environmental stress. Iodoacetamide (IDAM) is an alkylating toxicant that up-regulates expression of hsp70 (Liu, H., Lightfoot, D. L., and Stevens, J. L. (1996) J. Biol. Chem. 271, 4805-4812) and grp78 in LLC-PK1 renal epithelial cells. Therefore, we used IDAM to determine the role of these genes in tolerance to toxic chemicals. Prior heat shock did not protect cells from IDAM but pretreatment with trans-4,5-dihydroxy-1,2-dithiane (DTTox), thapsigargin, or tunicamycin enhanced expression of the endoplasmic reticulum (ER) chaperones GRP78 and GRP94 and rendered cells tolerant to IDAM. Cells expressing a 524-base pair antisense grp78 fragment (pkASgrp78) had a diminished capacity to up-regulate grp78 and grp94 expression after ER stress. Protection against IDAM due to prior ER stress was also attenuated in pkASgrp78 cells suggesting that ER chaperones of the GRP family are critical for tolerance. Covalent binding of IDAM to cellular macromolecules and depletion of cellular thiols was similar in tolerant and na?ve cells. However, DTTox pretreatment blocked the increases in cellular Ca2+ and lipid peroxidation observed after IDAM treatment. Overexpressing the ER Ca2+-binding protein calreticulin prevented IDAM-induced cell death, the rise in cytosolic Ca2+, and oxidative stress. Although activation of the ER stress response did not prevent toxicity due to Ca2+ influx, EGTA-AM and ruthenium red both blocked cell death suggesting that redistribution of intracellular Ca2+ to the mitochondria may be important in toxicity. The data support a model in which induction of ER stress proteins prevents disturbances of intracellular Ca2+ homeostasis, thus uncoupling toxicant exposure from oxidative stress and cell death. Multiple ER stress proteins are likely to be involved in this tolerance response.     

Multiple molecular chaperones interact with apolipoprotein B during its maturation. The network of endoplasmic reticulum-resident chaperones (ERp72, GRP94, calreticulin, and BiP) interacts with apolipoprotein b regardless of its lipidation state

The present study was undertaken to identify and characterize molecular chaperones that assist in the folding of apolipoprotein (apo) B, a secretory protein that requires assembly with lipids (lipidation) for its secretion. Both HepG2 cells, normally secreting full-length apoB (apoB-100), and C127 cells transfected to secrete truncated forms of apoB, apoB-41, apoB-29, and apoB-17, respectively, were employed. C127 cells were used to determine whether chaperone binding is dependent on apoB lipidation as they secrete both unlipidated and lipidated apoB forms despite their lack of microsomal triglyceride transfer protein (MTP), which mediates lipidation of apoB in HepG2 cells. The endoplasmic reticulum (ER)-resident molecular chaperones GRP94, calreticulin, and ERp72 were co-immunoprecipitated with apoB-100 from HepG2 cell lysates following cross-linking of proteins in living cells. The same chaperones including BiP/GRP78 were also associated with all truncated forms of apoB. Sequential immunoprecipitation with antibodies to MTP and apoB revealed the presence of ternary complexes containing apoB-100, MTP, and ERp72. However, MTP is not obligatory for the binding of ERp72 as it was associated with all truncated forms of apoB in C127 cells that lack MTP. The interactions between apoB-100 and ERp72 or GRP94 persisted for at least 2 h following a 30-min pulse. Thus, BiP/GRP78, calreticulin, ERp72, and GRP94 may participate in critical steps in the folding of apoB before any substantial lipidation occurs. ERp72 and GRP94 may also mediate the folding of more advanced folding intermediates and/or target the misfolded underlipidated pool of apoB for degradation.     

Modulation of secreted beta-amyloid precursor protein and amyloid beta-peptide in brain by cholesterol

The effects of dietary cholesterol on brain amyloid precursor protein (APP) processing were examined using an APP gene-targeted mouse, genetically humanized in the amyloid beta-peptide (Abeta) domain and expressing the Swedish familial Alzheimer's disease mutations. These mice express endogenous levels of APP holoprotein and abundant human Abeta. Increased dietary cholesterol led to significant reductions in brain levels of secreted APP derivatives, including sAPPalpha, sAPPbeta, Abeta1-40, and Abeta1-42, while having little to no effect on cell-associated species, including full-length APP and the COOH-terminal APP processing derivatives. The changes in levels of sAPP and Abeta in brain all were negatively correlated with serum cholesterol levels and levels of serum and brain apoE. These results demonstrate that secreted APP processing derivatives and Abeta can be modulated in the brain of an animal by diet and provide evidence that cholesterol plays a role in the modulation of APP processing in vivo. APP gene-targeted mice lacking apoE, also have high serum cholesterol levels but do not show alterations in APP processing, suggesting that effects of cholesterol on APP processing require the presence of apoE.     

Cholesterol depletion inhibits the generation of beta-amyloid in hippocampal neurons

The amyloid precursor protein (APP) plays a crucial role in the pathogenesis of Alzheimer's disease. During intracellular transport APP undergoes a series of proteolytic cleavages that lead to the release either of an amyloidogenic fragment called beta-amyloid (Abeta) or of a nonamyloidogenic secreted form consisting of the ectodomain of APP (APPsec). It is Abeta that accumulates in the brain lesions that are thought to cause the disease. By reducing the cellular cholesterol level of living hippocampal neurons by 70% with lovastatin and methyl-beta-cyclodextrin, we show that the formation of Abeta is completely inhibited while the generation of APPsec is unperturbed. This inhibition of Abeta formation is accompanied by increased solubility in the detergent Triton X-100 and is fully reversible by the readdition of cholesterol to previously depleted cells. Our results show that cholesterol is required for Abeta formation to occur and imply a link between cholesterol, Abeta, and Alzheimer's disease.     

Astrocytes containing amyloid beta-protein (Abeta)-positive granules are associated with Abeta40-positive diffuse plaques in the aged human brain

Amyloid beta-protein (Abeta) is the major component of senile plaques that emerge in the cortex during aging and appear most abundantly in Alzheimer's disease. In the course of our immunocytochemical study on a large number of autopsy cases, we noticed, in many aged nondemented cases, the presence of unique diffuse plaques in the cortex distinct from ordinary diffuse plaques by immunocytochemistry. The former were amorphous, very faintly Abeta-immunoreactive plaques resembling diffuse plaques, but they stained for Abeta40 and were associated with small cells containing Abeta-positive granules. A panel of amino- and carboxyl-terminal-specific Abeta antibodies showed that such Abeta40-positive diffuse plaques and cell-associated granules were composed exclusively of amino-terminally deleted Abeta terminating at Abeta40, -42, and -43. Double immunostaining also showed that those Abeta-immunoreactive granules are located in astrocytes and not in microglia or neurons. Immunoelectron microscopy revealed that nonfibrillar Abeta immunoreactivity was located within lipofuscin-like granules in somewhat swollen astrocytes. These findings raise the possibility that astrocytes take up Abeta and attempt to degrade it in lysosomes in the aged brain.     

Regulation of T cell activation in vitro and in vivo by targeting the OX40-OX40 ligand interaction: amelioration of ongoing inflammatory bowel disease with an OX40-IgG fusion protein, but not with an OX40 ligand-IgG fusion protein

The role of GRP78/BiP in coordinating endoplasmic reticular (ER) protein processing with mRNA translation was examined in GH3 pituitary cells. ADP-ribosylation of GRP78 and eukaryotic initiation factor (eIF)-2alpha phosphorylation were assessed, respectively, as indices of chaperone inactivation and the inhibition of translational initiation. Inhibition of protein processing by ER stress (ionomycin and dithiothreitol) resulted in GRP78 deribosylation and eIF-2 phosphorylation. Suppression of translation relative to ER protein processing (cycloheximide) produced approximately 50% ADP-ribosylation of GRP78 within 90 min without eIF-2 phosphorylation. ADP-ribosylation was reversed in 90 min by cycloheximide removal in a manner accelerated by ER stressors. Cycloheximide sharply reduced eIF-2 phosphorylation in response to ER stressors for about 30 min; sensitivity returned as GRP78 became increasingly ADP-ribosylated. Reduced sensitivity of eIF-2 to phosphorylation appeared to derive from the accumulation of free, unmodified chaperone as proteins completed processing without replacements. Prolonged (24 h) incubations with cycloheximide resulted in the selective loss of the ADP-ribosylated form of GRP78 and increased sensitivity of eIF-2 phosphorylation in response to ER stressors. Brefeldin A decreased ADP-ribosylation of GRP78 in parallel with increased eIF-2 phosphorylation. The cytoplasmic stressor, arsenite, which inhibits translational initiation through eIF-2 phosphorylation without affecting the ER, also produced ADP-ribosylation of GRP78.     

Digestibility and peptide patterns of modified lysozyme after hydrolyzing by protease

The metabolic pathway of Alzheimer's amyloid precursor protein (APP) involves restricted intracellular proteolysis by secretases, which leads to the secretion of the N-terminal soluble APP (sAPP) and the generation of a cell-associated C-terminal fragment. The precise cellular sites at which these processes occur remain unknown. In this report, we describe the route of APP sorting and the processing site using novel systems with and without sorting signals on the APP molecule. One system involves the replacement of the C-terminal ten amino acids of APP with Adenoviral E19 protein containing an endoplasmic reticulum (ER) retrieval signal (APPE19); the other involves deleting the last ten amino acids corresponding to the replaced site (APPdeltaC10). APPE19 localized mainly within the cis/medial Golgi compartment and exclusively suppresses the secretion of APP. In contrast, deletion of the C-terminal tail promotes sAPP secretion by a constitutive secretion pathway. Metabolic labeling followed by immunoprecipitation with anti-APP antibody revealed that APPE19 is rapidly degraded within 30 min and that the subsequent intracellular turnover rate is decreased with 40% of the protein retained within the cells even after a chase period a 3 h. In contrast, APPdeltaC10 is rapidly eliminated from the intracellular compartments and secreted into the culture medium. The surface internalization and recycling processes of this protein are relatively impaired compared with wild-type APP. The ratios of the levels of production to secretion of sAPP alpha, the N-terminal, soluble APP fragment released by alpha-secretase, are proportional to the secretion efficiencies among APP species, suggesting the localization of alpha-secretase within a compartment late in the constitutive secretion pathway. These secretion mutants which utilize ER targeting signals are useful tools for analyzing the location of secretases and the intracellular degradation system within a constitutive secretion pathway such as ER quality control.     

Heparin-binding properties of the amyloidogenic peptides Abeta and amylin. Dependence on aggregation state and inhibition by Congo red

Aggregation and deposition of the 40-42-residue amyloid beta-protein (Abeta) are early and necessary neuropathological events in Alzheimer's disease. An understanding of the molecular interactions that trigger these events is important for therapeutic strategies aimed at blocking Abeta plaque formation at the earliest stages. Heparan sulfate proteoglycans may play a fundamental role since they are invariably associated with Abeta and other amyloid deposits at all stages. However, the nature of the Abeta-heparan sulfate proteoglycan binding has been difficult to elucidate because of the strong tendency of Abeta to self-aggregate. Affinity co-electrophoresis can measure the binding of proteoglycans or glycosaminoglycans to proteins without altering the physical state of the protein during the assay. We used affinity co-electrophoresis to study the interaction between Abeta and the glycosaminoglycan heparin and found that the aggregation state of Abeta governs its heparin-binding properties: heparin binds to fibrillar but not nonfibrillar Abeta. The amyloid binding dye, Congo red, inhibited the interaction in a specific and dose-dependent manner. The "Dutch" mutant AbetaE22Q peptide formed fibrils more readily than wild type Abeta and it also attained a heparin-binding state more readily, but, once formed, mutant and wild type fibrils bound heparin with similar affinities. The heparin-binding ability of aggregated AbetaE22Q was reversible with incubation in a solvent that promotes alpha-helical conformation, further suggesting that conformation of the peptide is important. Studies with another human amyloidogenic protein, amylin, suggested that its heparin-binding properties were also dependent on aggregation state. These results demonstrate the dependence of the Abeta-heparin interaction on the conformation and aggregation state of Abeta rather than primary sequence alone, and suggest methods of interfering with this association.     

Turnover of amyloid beta-protein in mouse brain and acute reduction of its level by phorbol ester

Fibrillar amyloid deposits are defining pathological lesions in Alzheimer's disease brain and are thought to mediate neuronal death. Amyloid is composed primarily of a 39-42 amino acid protein fragment of the amyloid precursor protein (APP), called amyloid beta-protein (Abeta). Because deposition of fibrillar amyloid in vitro has been shown to be highly dependent on Abeta concentration, reducing the proteolytic release of Abeta is an attractive, potentially therapeutic target. Here, the turnover rate of brain Abeta has been determined to define treatment intervals over which a change in steady-state concentration of Abeta could be measured. Mice producing elevated levels of human Abeta were used to determine approximate turnover rates for Abeta and two of its precursors, C99 and APP. The t1/2 for brain Abeta was between 1.0 and 2.5 hr, whereas for C99, immature, and fully glycosylated forms of APP695 the approximate t1/2 values were 3, 3, and 7 hr, respectively. Given the rapid Abeta turnover rate, acute studies were designed using phorbol 12-myristate 13-acetate (PMA), which had been demonstrated previously to reduce Abeta secretion from cells in vitro via induction of protein kinase C (PKC) activity. Six hours after intracortical injection of PMA, Abeta levels were significantly reduced, as measured by both Abeta40- and Abeta42-selective ELISAs, returning to normal by 12 hr. An inactive structural analog of PMA, 4alpha-PMA, had no effect on brain Abeta levels. Among the secreted N-terminal APP fragments, APPbeta levels were significantly reduced by PMA treatment, whereas APPalpha levels were unchanged, in contrast to most cell culture studies. These results indicate that Abeta is rapidly turned over under normal conditions and support the therapeutic potential of elevating PKC activity for reduction of brain Abeta.     

Interaction of a neuron-specific protein containing PDZ domains with Alzheimer's amyloid precursor protein

A novel protein, human X11-like (human X11L), contains a phosphotyrosine interaction (PI) domain and two PDZ domains and displays 55.2% amino acid homology with the human X11 (human X11). The PI domain of human X11L interacts with a sequence containing the NPXY motif found in the cytoplasmic domain of Alzheimer's amyloid precursor protein. A construct lacking the carboxyl-terminal domain, which comprises two PDZ domains (N + PI), enhances PI binding to APP, whereas another construct lacking an amino-terminal domain relative to PI domain (PI + C) suppresses PI binding to APP. Overexpression of full-length human X11L (N + PI + C) in cells that express APP695 stably decreased the secretion of Abeta40 but not that of Abeta42. However, overexpression of the PI domain alone and the N + PI construct in cells did not affect the secretion of Abeta despite their ability to bind to the cytoplasmic domain of Alzheimer's amyloid precursor protein. These observations suggest that the amino-terminal domain regulates PI binding to APP and that the carboxyl-terminal domain containing PDZ motifs is essential to modulate APP processing. Because expression of the human X11L gene is specific to brain, the present observations should contribute to shedding light on the molecular mechanism of APP processing in Alzheimer's disease.     

Effect of inhibitors of poly(ADP-ribose) polymerase on the induction of GRP78 and subsequent development of resistance to etoposide

We have recently demonstrated that cell lines deficient in poly(ADP-ribose) synthesis due to deficiency in the enzyme poly(ADP-ribose) polymerase (PADPRP) or depletion of its substrate NAD+ overexpress GRP78. Furthermore, this overexpression of GRP78 is associated with the acquisition of resistance to topoisomerase II-directed drugs such as etoposide (VP-16); (S. Chatterjee et al., Cancer Res., 54: 4405-4411, 1994). Thus, our studies suggest that interference with NAD+-PADPRP metabolism could provide an important approach to (a) define pathways of GRP78 induction, (b) study the effect of GRP78 on other cellular processes, (c) elucidate the mechanism of GRP78-dependent resistance to topoisomerase II targeted drugs, and (d) modulate responses to chemotherapy in normal and tumor tissues. However, in the in vivo situation, it is impractical to interfere with NAD+-PADPRP metabolism by mutational inactivation of PADPRP or by depletion of its substrate NAD+. Therefore, we have examined several inhibitors of NAD+-PADPRP metabolism including 3-aminobenzamide, PD128763, and 6-aminonicotinamide for their ability to reproduce the results obtained with cell lines deficient in NAD+-PADPRP metabolism relative to the induction of GRP78 and subsequent development of resistance to VP-16. Our studies show that 6-aminoicotinamide treatment is highly effective in the induction of GRP78 and subsequent development of resistance to VP-16, whereas treatment with 3-aminobenzamide or PD128763 does not induce GRP78 and thus does not result in VP-16 resistance.     

Distinct sites of intracellular production for Alzheimer's disease A beta40/42 amyloid peptides

The Alzheimer amyloid precursor protein (APP) is cleaved by several proteases, the most studied, but still unidentified ones, are those involved in the release of a fragment of APP, the amyloidogenic beta-protein A beta. Proteolysis by gamma-secretase is the last processing step resulting in release of A beta. Cleavage occurs after residue 40 of A beta [A beta(1-40)], occasionally after residue 42 [A beta(1-42)]. Even slightly increased amounts of this A beta(1-42) might be sufficient to cause Alzheimer's disease (AD) (reviewed in ref. 1, 2). It is thus generally believed that inhibition of this enzyme could aid in prevention of AD. Unexpectedly we have identified in neurons the endoplasmic reticulum (ER) as the site for generation of A beta(1-42) and the trans-Golgi network (TGN) as the site for A beta(1-40) generation. It is interesting that intracellular generation of A beta seemed to be unique to neurons, because we found that nonneuronal cells produced significant amounts of A beta(1-40) and A beta(1-42) only at the cell surface. The specific production of the critical A beta isoform in the ER of neurons links this compartment with the generation of A beta and explains why primarily ER localized (mutant) proteins such as the presenilins could induce AD. We suggest that the earliest event taking place in AD might be the generation of A beta(1-42) in the ER.