V642I APP-inducible neuronal cells: a model system for investigating Alzheimer's disorders

APP is a precursor of beta amyloid deposited in Alzheimer's disease (AD). Although genetic studies established that mutations in APP cause familial AD (FAD), the mechanism for neuronal death by FAD mutants has not been well understood. We established neuronal cells (F11/EcR/V642I cells) in which V642I APP was inducibly expressed by ecdysone. Treatment with ecdysone, but not vehicle, killed most cells within a few days, with rounding, shrinkage, and detachment as well as nuclear fragmentation. Death was suppressed by Ac-DEVD-CHO and pertussis toxin. Electron microscopic analysis revealed that apoptosis occurred in ecdysone-treated cells. V642I-APP-induced death was suppressed by the anti-AD factors estrogen and apoE2. These data demonstrate not only that expression of this FAD gene causes neuronal apoptosis, but that F11/EcR/V642I cells, the first neuronal cells with inducible FAD gene expression, provide a useful model system in investigating AD disorders.     

Expression of V642 APP mutant causes cellular apoptosis as Alzheimer trait-linked phenotype.

APP is a transmembrane precursor of beta-amyloid. In dominantly inherited familial Alzheimer's disease (FAD), point mutations V6421, V642F and V642G have been discovered in APP695. Here we show that expression of these mutants (FAD-APPs) causes a clone of COS cells to undergo apoptosis associated with DNA fragmentation. Apoptosis by the three FAD-APPs was the highest among all possible V642 mutants; normal APP695 had no effect on apoptosis, suggesting that apoptosis by APP mutants in this system is phenotypically linked to the FAD trait. FAD-APP-induced apoptosis was sensitive to bcl-2 and most probably mediated by heteromeric G proteins. This study presents a model system allowing analysis of the mechanism for FAD-APP-induced cytotoxicity.     

Neurotoxic mechanisms triggered by Alzheimer's disease-linked mutant M146L presenilin 1: involvement of NO synthase via a novel pertussis toxin target

While it has been reported that familial Alzheimer's disease (FAD)-linked mutants of amyloid precursor protein (APP) and presenilin (PS)2 induce neuronal cytotoxicity in a manner sensitive to antioxidant and pertussis toxin (PTX), little of the mechanism for PS1-mediated neuronal cell death has been characterized. We previously found that multiple mechanisms, different in detail, underlie cytotoxicities by two FAD-linked mutants of APP, using neuronal cells with an ecdysone-controlled expression system. Here we report that this system revealed that (i) low expression of FAD-linked M146L-PS1 caused neuronal cell death, whereas that of wild-type (wt)PS1 did not; (ii) mutation-specific cytotoxicity by M146L-PS1 was sensitive to antioxidant glutathione-ethyl-ester and resistant to Ac-DEVD-CHO; (iii) cytotoxicity by higher expression of wtPS1 was resistant to both; and (iv) cytotoxicity by M146L-PS1 was inhibited by PTX. It was also highly likely that the involved superoxide-generating enzyme was nitric oxide synthase (NOS), and that the PTX-sensitive cytotoxic signal by M146L-PS1 was mediated by none of the G(i/o) proteins. We conclude that M146L-PS1 activates a NOS-mediated cytotoxic pathway via a novel PTX target.     

G protein betagamma complex-mediated apoptosis by familial Alzheimer''s disease mutant of APP.

In familial Alzheimer's disease (FAD), three missense mutations, V642I, V642F and V642G, that co-segregate with the disease phenotype have been discovered in the 695 amino acid form of the amyloid precursor protein APP. Expression of these mutants causes a COS cell NK1 clone to undergo pertussis toxin-sensitive apoptosis in an FAD trait-linked manner by activating the G protein Go, which consists of G alpha(o) and G betagamma subunits. We investigated which subunit was responsible for the induction of apoptosis by V642I APP in NK1 cells. In the same system, expression of mutationally activated G alpha(o) or G alpha(i) induced little apoptosis. Apoptosis by V642I APP was antagonized by the overexpression of the carboxy-terminal amino acids 495-689 of the beta-adrenergic receptor kinase-1, which blocks the specific functions of G betagamma. Co-transfection of G beta2gamma2 cDNAs, but not that of other G beta(x)gamma(z) (x = 1-3; z = 2, 3), induced DNA fragmentation in a manner sensitive to bcl-2. These data implicate G betagamma as a cell death mediator for the FAD-associated mutant of APP.     

Mechanisms of neuroprotection by a novel rescue factor humanin from Swedish mutant amyloid precursor protein

We report a novel gene, designated Humanin (HN) cDNA, that suppresses neuronal cell death by K595N/M596L-APP (NL-APP), a mutant causing familial Alzheimer's disease (FAD), termed Swedish mutant. Transfection of neuronal cells with HN cDNA or treatment with the coding HN polypeptide abrogated cytotoxicity by NL-APP. HN suppressed neurotoxicity by Abeta1-43 in the absence of N2 supplement, but could not inhibit Abeta secretion from NL-APP. HN could also protect neuronal cells from death by NL-APP lacking the 41st and 42nd residues of the Abeta region. Therefore, HN suppressed neuronal cell death by NL-APP not through inhibition of Abeta42 secretion, but with two targets for its inhibitory action: (i) the intracellular toxic mechanism directly triggered by NL-APP and (ii) neurotoxicity by Abeta. HN will contribute to the development of curative therapy of AD, especially as a novel reagent that could mechanistically supplement Abeta-production inhibitors.     

Intrinsic signaling function of APP as a novel target of three V642 mutations linked to familial Alzheimer's disease.

APP695 is a transmembrane precursor of Abeta amyloid. In familial Alzheimer's disease (FAD), three mutations V642I/F/G were discovered in APP695, which has been suggested by multiple studies to be a cell surface signaling receptor. We previously reported that normal APP695 encodes a potential GO-linked receptor with ligand-regulated function and that expression of the three FAD mutants (FAD-APPs), not normal APP, induces cellular outputs by GO-dependent mechanisms. This suggests that FAD-APPs are constitutively active GO-linked receptors. Here, we provide direct evidence for this notion. Reconstitution of either recombinant FAD-APP with GO vesicles induced activation of GO, which was inhibitable by pertussis toxin, sensitive to Mg2+ and proportional in quantity to the reconstituted amounts of FAD-APP. Consistent with the dominant inheritance of this type of FAD, this function was dominant over normal APP, because little activation was observed in APP695-GO vesicles. Experiments with antibody competition and sequence deletion indicated that His657-Lys676 of FAD-APP, which has been specified as the ligand-dependent GO-coupling domain of normal APP, was responsible for this constitutive activation, confirming that the three FAD-APPs are mutationally activated APP695. This study identifies the intrinsic signaling function of APP to be a novel target of hereditary Alzheimer's disease mutations, providing an in vitro system for the screening of potential FAD inhibitors.     

The Gtx homeodomain transcription factor exerts neuroprotection using its homeodomain

Certain cases of familial Alzheimer's disease are caused by mutants of amyloid-beta precursor protein (AbetaPP), including V642I-AbetaPP, K595N/M596L-AbetaPP (NL-AbetaPP), A617G-AbetaPP, and L648P-AbetaPP. By using an unbiased functional screening with transfection and expression of a human brain cDNA library, we searched for genes that protect neuronal cells from toxicity by V642I-AbetaPP. One protective clone was identical to the human GTX, a neuronal homeobox gene. Human Gtx (hGtx) inhibited caspase inhibitor-sensitive neuronal cell death not only by V642I-AbetaPP but also by L648P-, NL-, A617G-AbetaPP, apolipoprotein E4, and Abeta. The region of hGtx responsible for this rescue function was specified to be its homeodomain (Lys148-His207). The rescue function was shared by DLX4, a distal-less family gene with a homeodomain only 38.3% homologous to that of hGtx, suggesting that this function would be generally shared by homeodomains. The neuroprotective function of hGtx was attributable to hGtx-stimulated production and secretion of insulin-like growth factor-I. This study provides molecular clues to understand how neuronal cells developmentally regulate themselves against cell death as well as to develop reagents effective in curative therapeutics of Alzheimer's disease.     

Transforming growth factor beta2 is a neuronal death-inducing ligand for amyloid-beta precursor protein

APP, amyloid beta precursor protein, is linked to the onset of Alzheimer's disease (AD). We have here found that transforming growth factor beta2 (TGFbeta2), but not TGFbeta1, binds to APP. The binding affinity of TGFbeta2 to APP is lower than the binding affinity of TGFbeta2 to the TGFbeta receptor. On binding to APP, TGFbeta2 activates an APP-mediated death pathway via heterotrimeric G protein G(o), c-Jun N-terminal kinase, NADPH oxidase, and caspase 3 and/or related caspases. Overall degrees of TGFbeta2-induced death are larger in cells expressing a familial AD-related mutant APP than in those expressing wild-type APP. Consequently, superphysiological concentrations of TGFbeta2 induce neuronal death in primary cortical neurons, whose one allele of the APP gene is knocked in with the V642I mutation. Combined with the finding indicated by several earlier reports that both neural and glial expression of TGFbeta2 was upregulated in AD brains, it is speculated that TGFbeta2 may contribute to the development of AD-related neuronal cell death.     

Involvement of c-Jun N-terminal kinase in amyloid precursor protein-mediated neuronal cell death

Amyloid precursor protein (APP), the precursor of Abeta, has been shown to function as a cell surface receptor that mediates neuronal cell death by anti-APP antibody. The c-Jun N-terminal kinase (JNK) can mediate various neurotoxic signals, including Abeta neurotoxicity. However, the relationship of APP-mediated neurotoxicity to JNK is not clear, partly because APP cytotoxicity is Abeta independent. Here we examined whether JNK is involved in APP-mediated neuronal cell death and found that: (i) neuronal cell death by antibody-bound APP was inhibited by dominant-negative JNK, JIP-1b and SP600125, the specific inhibitor of JNK, but not by SB203580 or PD98059; (ii) constitutively active (ca) JNK caused neuronal cell death and (iii) the pharmacological profile of caJNK-mediated cell death closely coincided with that of APP-mediated cell death. Pertussis toxin (PTX) suppressed APP-mediated cell death but not caJNK-induced cell death, which was suppressed by Humanin, a newly identified neuroprotective factor which inhibits APP-mediated cytotoxicity. In the presence of PTX, the PTX-resistant mutant of Galphao, but not that of Galphai, recovered the cytotoxic action of APP. These findings demonstrate that JNK is involved in APP-mediated neuronal cell death as a downstream signal transducer of Go.     

Importance of the domain-domain interface to the catalytic action of the NO synthase reductase domain

Calmodulin (CaM) activates NO synthase (NOS) by binding to a 20 amino acid interdomain hinge in the presence of Ca (2+), inducing electrons to be transferred from the FAD to the heme of the enzyme via a mobile FMN domain. The activation process is influenced by a number of structural features, including an autoinhibitory loop, the C-terminal tail of the enzyme, and a number of phosphorylation sites. Crystallographic and other recent experimental data imply that the regulatory elements lie within the interface between the FAD- and FMN-binding domains, restricting the movement of the two cofactors with respect to each other. Arg1229 of rat neuronal NOS is a conserved residue in the FAD domain that forms one of only two electrostatic contacts between the domains. Mutation of this residue to Glu reverses its charge and is expected to induce an interdomain repulsion, allowing the importance of the interface and domain-domain motion to be probed. The charge-reversal mutation R1229E has three dramatic effects on catalysis: (i) hydride transfer from NADPH to FAD is activated in the CaM-free enzyme, (ii) FAD to FMN electron transfer is inhibited in both forms, and (iii) electron transfer from FMN to the surrogate acceptor cytochrome c is activated in the CaM-free enzyme. As a result, during steady-state turnover with cytochrome c, calmodulin now deactivates the enzyme and causes cytochrome c-dependent inhibition. Evidently, domain-domain separation is large enough in the mutant to accommodate another protein between the cofactors. The effects of this single charge reversal on three distinct catalytic events illustrate how each is differentially dependent on the enzyme conformation and support a model for catalytic motion in which steps i, ii, and iii occur in the hinged open, closed, and open states, respectively. This model is also likely to apply to related enzymes such as cytochrome P450 reductase.     

Determination of the redox properties of human NADPH-cytochrome P450 reductase

Midpoint reduction potentials for the flavin cofactors in human NADPH-cytochrome P450 oxidoreductase were determined by anaerobic redox titration of the diflavin (FAD and FMN) enzyme and by separate titrations of its isolated FAD/NADPH and FMN domains. Flavin reduction potentials are similar in the isolated domains (FAD domain E(1) [oxidized/semiquinone] = -286 +/- 6 mV, E(2) [semiquinone/reduced] = -371 +/- 7 mV; FMN domain E(1) = -43 +/- 7 mV, E(2) = -280 +/- 8 mV) and the soluble diflavin reductase (E(1) [FMN] = -66 +/- 8 mV, E(2) [FMN] = -269 +/- 10 mV; E(1) [FAD] = -283 +/- 5 mV, E(2) [FAD] = -382 +/- 8 mV). The lack of perturbation of the individual flavin potentials in the FAD and FMN domains indicates that the flavins are located in discrete environments and that these environments are not significantly disrupted by genetic dissection of the domains. Each flavin titrates through a blue semiquinone state, with the FMN semiquinone being most intense due to larger separation (approximately 200 mV) of its two couples. Both the FMN domain and the soluble reductase are purified in partially reduced, colored form from the Escherichia coli expression system, either as a green reductase or a gray-blue FMN domain. In both cases, large amounts of the higher potential FMN are in the semiquinone form. The redox properties of human cytochrome P450 reductase (CPR) are similar to those reported for rabbit CPR and the reductase domain of neuronal nitric oxide synthase. However, they differ markedly from those of yeast and bacterial CPRs, pointing to an important evolutionary difference in electronic regulation of these enzymes.     

APP-BP1 mediates APP-induced apoptosis and DNA synthesis and is increased in Alzheimer's disease brain

APP-BP1, first identified as an amyloid precursor protein (APP) binding protein, is the regulatory subunit of the activating enzyme for the small ubiquitin-like protein NEDD8. We have shown that APP-BP1 drives the S- to M-phase transition in dividing cells, and causes apoptosis in neurons. We now demonstrate that APP-BP1 binds to the COOH-terminal 31 amino acids of APP (C31) and colocalizes with APP in a lipid-enriched fraction called lipid rafts. We show that coexpression of a peptide representing the domain of APP-BP1 that binds to APP, abolishes the ability of overexpressed APP or the V642I mutant of APP to cause neuronal apoptosis and DNA synthesis. A dominant negative mutant of the NEDD8 conjugating enzyme hUbc12, which participates in the ubiquitin-like pathway initiated by APP-BP1, blocks neuronal apoptosis caused by APP, APP(V642I), C31, or overexpression of APP-BP1. Neurons overexpressing APP or APP(V642I) show increased APP-BP1 protein levels in lipid rafts. A similar increase in APP-BP1 in lipid rafts is observed in the Alzheimer's disease brain hippocampus, but not in less-affected areas of Alzheimer's disease brain. This translocation of APP-BP1 to lipid rafts is accompanied by a change in the subcellular localization of the ubiquitin-like protein NEDD8, which is activated by APP-BP1.     

Alternative, non-secretase processing of Alzheimer's beta-amyloid precursor protein during apoptosis by caspase-6 and -8.

Alzheimer's disease (AD) is a progressive neurodegenerative disorder. Although the pathogenesis of AD is unknown, it is widely accepted that AD is caused by extracellular accumulation of a neurotoxic peptide, known as Abeta. Mutations in the beta-amyloid precursor protein (APP), from which Abeta arises by proteolysis, are associated with some forms of familial AD (FAD) and result in increased Abeta production. Two other FAD genes, presenilin-1 and -2, have also been shown to regulate Abeta production; however, studies examining the biological role of these FAD genes suggest an alternative theory for the pathogenesis of AD. In fact, all three genes have been shown to regulate programmed cell death, hinting at the possibility that dysregulation of apoptosis plays a primary role in causing neuronal loss in AD. In an attempt to reconcile these two hypotheses, we investigated APP processing during apoptosis and found that APP is processed by the cell death proteases caspase-6 and -8. APP is cleaved by caspases in the intracellular portion of the protein, in a site distinct from those processed by secretases. Moreover, it represents a general effect of apoptosis, because it occurs during cell death induced by several stimuli both in T cells and in neuronal cells.     

Modulation of amyloid precursor protein metabolism by X11alpha /Mint-1. A deletion analysis of protein-protein interaction domains

Modulation of amyloid precursor protein (APP) metabolism plays a pivotal role in the pathogenesis of Alzheimer's disease. The phosphotyrosine-binding/protein interaction (PTB/PI) domain of X11alpha, a neuronal cytosolic adaptor protein, binds to the YENPTY sequence in the cytoplasmic carboxyl terminus of APP. This interaction prolongs the half-life of APP and inhibits Abeta40 and Abeta42 secretion. X11alpha/Mint-1 has multiple protein-protein interaction domains, a Munc-18 interaction domain (MID), a Cask/Lin-2 interaction domain (CID), a PTB/PI domain, and two PDZ domains. These X11alpha protein interaction domains may modulate its effect on APP processing. To test this hypothesis, we performed a deletion analysis of X11alpha effects on metabolism of APP(695) Swedish (K595N/M596L) (APP(sw)) by transient cotransfection of HEK 293 cells with: 1) X11alpha (X11alpha-wt, N-MID-CID-PTB-PDZ-PDZ-C), 2) amino-terminal deletion (X11alpha-DeltaN, PTB-PDZ-PDZ), 3) carboxyl-terminal deletion (X11alpha-DeltaPDZ, MID-CID-PTB), or 4) deletion of both termini (PTB domain only, PTB). The carboxyl terminus of X11alpha was required for stabilization of APP(sw) in cells. In contrast, the amino terminus of X11alpha was required to stimulate APPs secretion. X11alpha, X11alpha-DeltaN, and X11alpha-PTB, but not X11alpha-DeltaPDZ, were effective inhibitors of Abeta40 and Abeta42 secretion. These results suggest that additional protein interaction domains of X11alpha modulate various aspects of APP metabolism.     

The level of DLDB/CHIP controls the activity of the LIM homeodomain protein apterous: evidence for a functional tetramer complex in vivo

The LIM homeodomain (LIM-HD) protein Apterous (Ap) and its cofactor DLDB/CHIP control dorso- ventral (D/V) patterning and growth of Drosophila wing. To investigate the molecular mechanisms of Ap/CHIP function we altered their relative levels of expression and generated mutants in the LIM1, LIM2 and HD domains of Ap, as well as in the LIM-interacting and self-association domains of CHIP. Using in vitro and in vivo assays we found that: (i) the levels of CHIP relative to Ap control D/V patterning; (ii) the LIM1 and LIM2 domains differ in their contributions to Ap function; (iii) Ap HD mutations cause weak dominant negative effects; (iv) overexpression of ChipDeltaSAD mutants mimics Ap lack-of-function, and this dominant negative phenotype is caused by titration of Ap because it can be rescued by adding extra Ap; and (v) overexpression of ChipDeltaLID mutants also causes an Ap lack-of-function phenotype, but it cannot be rescued by extra Ap. These results support the model that the Ap-CHIP active complex in vivo is a tetramer.     

MOCA is an integrator of the neuronal death signals that are activated by familial Alzheimer's disease-related mutants of amyloid β precursor protein and presenilins

The death of cholinergic neurons in the cerebral cortex and certain subcortical regions is linked to irreversible dementia relevant to AD (Alzheimer's disease). Although multiple studies have shown that expression of a FAD (familial AD)-linked APP (amyloid β precursor protein) or a PS (presenilin) mutant, but not that of wild-type APP or PS, induced neuronal death by activating intracellular death signals, it remains to be addressed how these signals are interrelated and what the key molecule involved in this process is. In the present study, we show that the PS1-mediated (or possibly the PS2-mediated) signal is essential for the APP-mediated death in a γ-secretase-independent manner and vice versa. MOCA (modifier of cell adhesion), which was originally identified as being a PS- and Rac1-binding protein, is a common downstream constituent of these neuronal death signals. Detailed molecular analysis indicates that MOCA is a key molecule of the AD-relevant neuronal death signals that links the PS-mediated death signal with the APP-mediated death signal at a point between Rac1 [or Cdc42 (cell division cycle 42)] and ASK1 (apoptosis signal-regulating kinase 1).     

Secreted Abeta does not mediate neurotoxicity by antibody-stimulated amyloid precursor protein

Antibodies against APP, a precursor of Abeta deposited in Alzheimer's disease brain, have been shown to cause neuronal death. Therefore, it is important to determine whether Abeta mediates antibody-induced neurotoxicity. When primary neurons were treated with anti-APP antibodies, Abeta40 and Abeta42 in the cultured media were undetectable by an assay capable of detecting 100 nM Abeta peptides. However, exogenously treated Abeta1-42 or Abeta1-43 required >3 microM to exert neurotoxicity, and 25 microM Abeta1-40 was not neurotoxic. Glutathione-ethyl-ester inhibited neuronal death by anti-APP antibody, but not death by Abeta1-42, whereas serum attenuated toxicity by Abeta1-42, but not by anti-APP antibody. Using immortalized neuronal cells, we specified the domain responsible for toxicity to be cytoplasmic His(657)-Lys(676), but not the Abeta1-42 region, of APP. This indicates that neuronal cell death by anti-APP antibody is not mediated by secreted Abeta.