Gene expression levels as endophenotypes in genome-wide association studies of Alzheimer disease.

BACKGROUND: Late-onset Alzheimer disease (LOAD) is a common disorder with a substantial genetic component. We postulate that many disease susceptibility variants act by altering gene expression levels. METHODS: We measured messenger RNA (mRNA) expression levels of 12 LOAD candidate genes in the cerebella of 200 subjects with LOAD. Using the genotypes from our LOAD genome-wide association study for the cis-single nucleotide polymorphisms (SNPs) (n = 619) of these 12 LOAD candidate genes, we tested for associations with expression levels as endophenotypes. The strongest expression cis-SNP was tested for AD association in 7 independent case-control series (2,280 AD and 2,396 controls). RESULTS: We identified 3 SNPs that associated significantly with IDE (insulin degrading enzyme) expression levels. A single copy of the minor allele for each significant SNP was associated with approximately twofold higher IDE expression levels. The most significant SNP, rs7910977, is 4.2 kb beyond the 3' end of IDE. The association observed with this SNP was significant even at the genome-wide level (p = 2.7 x 10(-8)). Furthermore, the minor allele of rs7910977 associated significantly (p = 0.0046) with reduced LOAD risk (OR = 0.81 with a 95% CI of 0.70-0.94), as expected biologically from its association with elevated IDE expression. CONCLUSIONS: These results provide strong evidence that IDE is a late-onset Alzheimer disease (LOAD) gene with variants that modify risk of LOAD by influencing IDE expression. They also suggest that the use of expression levels as endophenotypes in genome-wide association studies may provide a powerful approach for the identification of disease susceptibility alleles.

Plasma amyloid beta protein is elevated in late-onset Alzheimer disease families.

OBJECTIVE: Plasma A beta levels are elevated in early-onset Alzheimer disease (AD) caused by autosomal dominant mutations. Our objective was to determine whether similar genetic elevations exist in late-onset AD (LOAD). METHODS: We measured plasma A beta in first-degree relatives of patients with LOAD in a cross-sectional series and in extended LOAD families. We screened these subjects for pathogenic mutations in early-onset AD genes and determined their ApoE genotypes. RESULTS: Plasma A beta is significantly elevated in the LOAD first-degree relatives in comparison to unrelated controls and married-in spouses. These elevations are not due to ApoE epsilon 4 or pathogenic coding mutations in the known early-onset AD genes. CONCLUSIONS: The findings provide strong evidence for the existence of novel, as yet unknown genetic factors that affect late-onset Alzheimer disease by increasing A beta.

Reduced effectiveness of Abeta1-42 immunization in APP transgenic mice with significant amyloid deposition.

Vaccinations with Abeta1-42 have been shown to reduce amyloid burden in transgenic models of Alzheimer's disease (AD). We have further tested the efficacy of Abeta1-42 immunization in the Tg2576 mouse model of AD by immunizing one group of mice with minimal Abeta deposition, one group of mice with modest Abeta deposition, and one group with significant Abeta deposition. The effects of immunization on Abeta deposition were examined using biochemical and immunohistochemical methods. In Tg2576 mice immunized prior to significant amyloid deposition, Abeta1-42 immunization was highly effective. Biochemically extracted Abeta40 and Abeta42 levels were significantly reduced and immunohistochemical plaque load was also reduced. Immunization of mice with modest amounts of pre-existing Abeta deposits selectively reduced Abeta42 without altering Abeta40, although plaque load was reduced. In contrast, in Tg2576 mice with significant pre-existing Abeta loads, Abeta1-42 immunization only minimally decreased Abeta42 levels, whereas no alteration in Abeta40 levels or in plaque load was observed. These results indicate that in Tg2576 mice, Abeta1-42 immunization is more effective at preventing additional Abeta accumulation and does not result in significant clearance of pre-existing Abeta deposits.

Heritability of plasma amyloid beta in typical late-onset Alzheimer's disease pedigrees.

Plasma amyloid beta42 peptide (Abeta42) levels are significantly elevated in all genetic forms of early-onset Alzheimer's disease caused by familial Alzheimer's disease mutations or Down's syndrome. Moreover, recent studies have determined that both plasma Abeta42 and Abeta40 levels are significantly elevated in late-onset Alzheimer's disease (LOAD) patients, their cognitively normal first-degree relatives, and members of typical LOAD families when compared to appropriate controls. To determine the magnitude of the genetic component affecting plasma Abeta levels, we estimated the heritability of plasma Abeta42 and Abeta40 in 15 extended, multigenerational LOAD pedigrees, using a variance components method. Heritability estimates as high as 73 and 54% were found for plasma Abeta42 and Abeta40 levels, respectively. Inclusion of the ApoE epsilon4 dosage as a covariate was not found to have a significant effect on the heritability of these traits. These results suggest that genetic determinants other than ApoE account for a very substantial percentage of the phenotypic variance in plasma Abeta levels. The high heritability and the significant elevation of these traits in LOAD pedigrees suggest that at least some of the genetic determinants of plasma Abeta levels may lead to elevated Abeta and LOAD in these families. Thus, we suggest that plasma Abeta levels are quantitative traits that may be excellent surrogate markers for use in linkage analysis to identify loci that are important in typical LOAD.

Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer's disease.

The accumulation of amyloid beta protein (Abeta) in the Tg2576 mouse model of Alzheimer's disease (AD) was evaluated by ELISA, immunoblotting, and immunocytochemistry. Changes in Abeta begin at 6-7 months as SDS-insoluble forms of Abeta42 and Abeta40 that require formic acid for solubilization appear. From 6 to 10 months, these insoluble forms increase exponentially. As insoluble Abeta appears, SDS-soluble Abeta decreases slightly, suggesting that it may be converting to an insoluble form. Our data indicate that it is full-length unmodified Abeta that accumulates initially in Tg2576 brain. SDS-resistant Abeta oligomers and most Abeta species that are N-terminally truncated or modified develop only in older Tg2576 mice, in which they are present at levels far lower than in human AD brain. Between 6 and 10 months, when SDS-insoluble Abeta42 and Abeta40 are easily detected in every animal, histopathology is minimal because only isolated Abeta cores can be identified. By 12 months, diffuse plaques are evident. From 12 to 23 months, diffuse plaques, neuritic plaques with amyloid cores, and biochemically extracted Abeta42 and Abeta40 increase to levels like those observed in AD brains. Coincident with the marked deposition of Abeta in brain, there is a decrease in CSF Abeta and a substantial, highly significant decrease in plasma Abeta. If a similar decline occurs in human plasma, it is possible that measurement of plasma Abeta may be useful as a premorbid biomarker for AD.

Linkage of plasma Abeta42 to a quantitative locus on chromosome 10 in late-onset Alzheimer's disease pedigrees.

Plasma Abeta42 (amyloid beta42 peptide) is invariably elevated in early-onset familial Alzheimer's disease (AD), and it is also increased in the first-degree relatives of patients with typical late-onset AD (LOAD). To detect LOAD loci that increase Abeta42, we used plasma Abeta42 as a surrogate trait and performed linkage analysis on extended AD pedigrees identified through a LOAD patient with extremely high plasma Abeta. Here, we report linkage to chromosome 10 with a maximal lod score of 3.93 at 81 centimorgans close to D10S1225. Remarkably, linkage to the same region was obtained independently in a genome-wide screen of LOAD sibling pairs. These results provide strong evidence for a novel LOAD locus on chromosome 10 that acts to increase Abeta.

Differences in the Abeta40/Abeta42 ratio associated with cerebrospinal fluid lipoproteins as a function of apolipoprotein E genotype.

The epsilon4 allele of apolipoprotein E (ApoE) is a risk factor for Alzheimer's disease (AD). ApoE, which is important for lipid metabolism, is also a major constituent of cerebrospinal fluid (CSF) lipoproteins (LPs). Although ApoE in the CSF is derived from the central nervous system, the relation between LP metabolism in plasma and CSF is not clear. Soluble amyloid-beta (Abeta) protein may normally be associated with CSF LPs. It is converted in AD to a fibrillar form in brain parenchyma. ApoE and CSF LPs may regulate this process. The purpose of this study was to characterize CSF LPs from healthy, cognitively normal, fasted, elderly individuals at different risk for AD based on ApoE genotype. Lipid composition of CSF LPs did not differ with ApoE genotype. Interestingly, plasma and CSF high-density lipoprotein (HDL) cholesterol and apolipoprotein AI (ApoAI) levels were correlated. Importantly, as assessed by size-exclusion chromatography, Abeta in CSF coeluted in fractions containing LPs and was influenced by ApoE genotype: E4-positive subjects displayed significant elevations in Abeta40/Abeta42 ratios. These results suggest that plasma ApoAI/HDL levels can influence CSF ApoAI/HDL levels and that interactions between Abeta and central nervous system LPs may reflect changes in brain Abeta metabolism before the onset of clinical disease.

Plasma beta-amyloid peptide, transforming growth factor-beta 1, and risk for cerebral amyloid angiopathy.

Despite the documented association between apolipoprotein E genotype and cerebral amyloid angiopathy (CAA), a substantial proportion of CAA-related hemorrhages occur in patients without known risks for this disorder. Two other factors implicated in the pathogenesis of CAA are the amyloid-beta peptide (preferentially deposited in vessels as a 40-amino acid species) and the multifunctional cytokine transforming growth factor-beta 1 (a specific promoter of vascular amyloid deposition in transgenic models). We measured plasma concentrations of these factors in a series of 25 patients diagnosed with probable or definite CAA-related hemorrhage and compared them with 21 patients with hemorrhage due to probable hypertensive vasculopathy and 42 elderly control subjects without hemorrhage. We found no differences among the groups in concentrations of the 40- or 42-amino acid species of beta-amyloid or either the active or latent form of transforming growth factor-beta 1. While the data do not exclude important roles for these molecules as risks for CAA, they indicate that plasma measurements are not useful in its diagnosis.

Amyloid beta protein (A beta) in Alzheimer's disease brain. Biochemical and immunocytochemical analysis with antibodies specific for forms ending at A beta 40 or A beta 42(43).

Biochemical and immunocytochemical analyses were performed to evaluate the composition of the amyloid beta protein (A beta) deposited in the brains of patients with Alzheimer's disease (AD). To quantitate all A beta s present, cerebral cortex was homogenized in 70% formic acid, and the supernatant was analyzed by sandwich enzyme-linked immunoabsorbent assays specific for various forms of A beta. In 9 of 27 AD brains examined, there was minimal congophilic angiopathy and virtually all A beta (96%) ended at A beta 42(43). The other 18 AD brains contained increasing amounts of A beta ending at A beta 40. From this set, 6 brains with substantial congophilic angiopathy were separately analyzed. In these brains, the amount of A beta ending at A beta 42(43) was much the same as in brains with minimal congophilic angiopathy, but a large amount of A beta ending at A beta 40 (76% of total A beta) was also present. Immunocytochemical analysis with monoclonal antibodies selective for A beta s ending at A beta 42(43) or A beta 40 confirmed that, in brains with minimal congophilic angiopathy, virtually all A beta is A beta ending at A beta 42(43) and showed that this A beta is deposited in senile plaques of all types. In the remaining AD brains, A beta 42(43) was deposited in a similar fashion in plaques, but, in addition, widely varying amounts of A beta ending at A beta 40 were deposited, primarily in blood vessel walls, where some A beta ending at A beta 42(43) was also present. These observations indicate that A beta s ending at A beta 42(43), which are a minor component of the A beta in human cerebrospinal fluid and plasma, are critically important in AD where they deposit selectively in plaques of all kinds.

Amyloid beta protein levels in cerebrospinal fluid are elevated in early-onset Alzheimer's disease.

The 4-kd amyloid beta protein (A beta) deposited as amyloid in Alzheimer's disease (AD) is produced and released by normal proteolytic processing of the amyloid beta protein precursor (beta APP) and is readily detected in cerebrospinal fluid (CSF). Here, we present the levels of A beta in CSF from a total of 95 subjects, including 38 patients with AD, 14 with early-onset AD and 24 with late-onset AD, 25 normal control subjects, and 32 patients with other neurological diseases. The level of A beta decreased with normal aging, and there was a significant elevation in the level of A beta in the CSF of early-onset AD patients (4.14 +/- 1.37 pmol/ml, p < 0.01). Neither Mini-Mental State nor Functional Assessment Staging were correlated with the amount of A beta in the CSF. The A beta/secreted form of beta APP ratio was elevated, but the level of alpha 1-antichymotrypsin in the CSF did not correlate with the level of CSF A beta in early-onset AD patients. Thus, the level of A beta in the CSF is elevated in early-onset AD patients and is suggested to be correlated with the pathology in the brain that characterizes AD.

Processing of the amyloid protein precursor to potentially amyloidogenic derivatives.

The approximately 120-kilodalton amyloid beta protein precursor (beta APP) is processed into a complex set of 8- to 12-kilodalton carboxyl-terminal derivatives that includes potentially amyloidogenic forms with the approximately 4-kilodalton amyloid beta protein (beta AP) at or near their amino terminus. In order to determine if these derivatives are processed in a secretory pathway or by the endosomal-lysosomal system, (i) deletion mutants that produce the normal set of carboxyl-terminal derivatives and shortened secreted derivatives were analyzed and (ii) the effect of inhibitors of endosomal-lysosomal processing was examined. In the secretory pathway, cleavage of the beta APP occurs at a single site within the beta AP to generate one secreted derivative and one nonamyloidogenic carboxyl-terminal fragment, whereas, in the endosomal-lysosomal system, a complex set of carboxyl-terminal derivatives is produced that includes the potentially amyloidogenic forms.

Expression of beta amyloid protein precursor mRNAs: recognition of a novel alternatively spliced form and quantitation in Alzheimer's disease using PCR.

We have analyzed alternatively spliced beta amyloid protein precursor (beta APP) mRNAs by using the polymerase chain reaction to amplify beta APP cDNAs produced by reverse transcription. With this approach the three previously characterized beta APP mRNAs (beta APP695, beta APP751, and beta APP770) are readily detected and compared in RNA samples extracted from specimens as small as a single cryostat section. We show that the results obtained with this method are not affected by partial RNA degradation and use it to identify a novel alternatively spliced beta APP714 mRNA that is present at low abundance in each of the many human brain regions, peripheral tissues, and cell lines that we have examined; demonstrate that nonneuronal cells in the adult human brain and meninges produce appreciable beta APP695, beta APP751, and beta APP770 mRNA; and identify changes in beta APP gene expression in the AD brain and meninges that may contribute to amyloid deposition.

The beta-amyloid protein precursor of Alzheimer disease has soluble derivatives found in human brain and cerebrospinal fluid.

In this study, we use antisera to synthetic beta-amyloid protein precursor (beta APP) peptides to identify, in human brain and cerebrospinal fluid (CSF), soluble approximately 125- and approximately 105-kDa derivatives of the beta APP that lack the carboxyl terminus of the full-length, membrane-associated forms. We show that the soluble approximately 125-kDa beta APP derivative contains the Kunitz protease inhibitor domain, whereas the approximately 105-kDa form does not, and we confirm that these two proteins are soluble beta APP derivatives by purifying each from human CSF and directly sequencing its amino terminus.

Antisera to an amino-terminal peptide detect the amyloid protein precursor of Alzheimer's disease and recognize senile plaques.

The cerebral amyloid deposited in Alzheimer's disease (AD) contains a 4.2 kDa beta amyloid polypeptide (beta AP) that is derived from a larger beta amyloid protein precursor (beta APP). Three beta APP mRNAs encoding proteins of 695, 751, and 770 amino acids have previously been identified. In each of these, there is a single membrane-spanning domain close to the carboxyl-terminus of the beta APP, and the 42 amino acid beta AP sequence extends from within the membrane-spanning domain into the large extracellular region of the beta APP. We raised rabbit antisera to a peptide corresponding to amino acids 45-62 near the amino-terminus of the beta APP. We show that these antisera detect the beta APP by demonstrating that they (i) label a set of approximately 120 kDa membrane-associated proteins in human brain previously detected by antisera to the carboxyl-terminus of beta APP and (ii) label a set of approximately 120 kDa membrane-associated proteins that are selectively overexpressed in cells transfected with a full length beta APP expression construct. The beta APP45-62 antisera specifically stain senile plaques in AD brains. This finding, along with the previous demonstration that antisera to the carboxyl-terminus of the beta APP label senile plaques, indicates that both near amino-terminal and carboxyl-terminal domains of the beta APP are present in senile plaques and suggests that proteolytic processing of the full length beta APP molecule into insoluble amyloid fibrils occurs in a highly localized fashion at the sites of amyloid deposition in AD brains.

Amyloid protein precursor messenger RNAs: differential expression in Alzheimer's disease.

In situ hybridization was used to assess total amyloid protein precursor (APP) messenger RNA and the subset of APP mRNA containing the Kunitz protease inhibitor (KPI) insert in 11 Alzheimer's disease (AD) and 7 control brains. In AD, a significant twofold increase was observed in total APP mRNA in nucleus basalis and locus ceruleus neurons but not in hippocampal subicular neurons, neurons of the basis pontis, or occipital cortical neurons. The increase in total APP mRNA in locus ceruleus and nucleus basalis neurons was due exclusively to an increase in APP mRNA lacking the KPI domain. These findings suggest that increased production of APP lacking the KPI domain in nucleus basalis and locus ceruleus neurons may play an important role in the deposition of cerebral amyloid that occurs in AD.

In situ hybridization of nucleus basalis neurons shows increased beta-amyloid mRNA in Alzheimer disease.

To determine which cells within the brain produce beta-amyloid mRNA and to assess expression of the beta-amyloid gene in Alzheimer disease, we analyzed brain tissue from Alzheimer and control patients by in situ hybridization. Our results demonstrate that beta-amyloid mRNA is produced by neurons in the nucleus basalis of Meynert and cerebral cortex and that nucleus basalis perikarya from Alzheimer patients consistently hybridize more beta-amyloid probe than those from controls. These observations support the hypothesis that increased expression of the beta-amyloid gene plays an important role in the deposition of amyloid in the brains of patients with Alzheimer disease.

Effect of fibrillation on acetylcholinesterase mRNA in cultured embryonic rat myotubes.

Acetylcholinesterase (AChE) and AChE mRNA were evaluated in spontaneously fibrillating myotubes derived from 20-day-old rat fetuses and in matched cultures in which fibrillation was prevented by adding tetrodotoxin on the fourth day of culture. On the eighth day of culture, the AChE activity of fibrillating and nonfibrillating cultures was 5332 and 1861 pmol ACh hydrolyzed min-1 dish-1, respectively (P less than 0.005). Total mRNA was essentially the same in fibrillating and nonfibrillating cultures (27.4 and 25.4 micrograms/dish, respectively). AChE mRNA was assessed by assaying the AChE produced by Xenopus oocytes microinjected with purified mRNA. The AChE produced by mRNA from fibrillating and nonfibrillating cultures was 0.46 and 0.10 pmol ACh hydrolyzed min-1 oocyte-1, respectively (P less than 0.005).

Molecular forms of acetylcholinesterases in Alzheimer's disease.

In this study, we examined 26 cases of Alzheimer's disease (AD) and 14 age-matched controls. In Brodmann area 21 cerebral cortex of the AD cases, there was no change in soluble G1 and G4 acetylcholinesterase (AChE) (EC 3.1.1.7), a significant 40% decrease in membrane-associated G4 AChE, significant 342 and 406% increases in A12 and A8 AChE, and a significant 71% decrease in choline acetyltransferase (ChAT) (EC 2.3.1.6). Our working hypothesis to account for these changes postulates that soluble globular forms are unchanged because they are primarily associated with intrinsic cortical neurons that are relatively unaffected by AD, that ChAT and membrane-associated G4 AChE decrease because they are primarily associated with incoming axons of cholinergic neurons that are abnormal in AD, and that asymmetric forms of AChE increase because of an acrylamide-type impairment of fast axonal transport in diseased incoming cholinergic axons. In the nucleus basalis of Meynert (nbM) of the 26 AD cases, there was a significant 61% decrease in the number of cholinergic neurons, an insignificant 23% decrease in nbM ChAT, a significant 298% increase in nbM ChAT per cholinergic neuron, and a significant 7% increase in the area of cholinergic perikarya. To account for the increased ChAT in cholinergic neurons and the enlargement of cholinergic perikarya, we propose that slow axonal transport may be impaired in nbM cholinergic neurons in AD.

Turnover of acetylcholinesterase in innervated and denervated rat diaphragm.

The acetylcholinesterase (AChE) in rat diaphragms was labelled by intravenous injection of echothiophate in order to evaluate the turnover of AChE in innervated and denervated muscle in vivo. Echothiophate diethylphosphorylates AChE thereby inactivating it. Labelled (diethylphosphorylated) enzyme is rapidly and quantitatively reactivated with 1-methyl-2-hydroxyiminomethylpyridinium (2-PAM), so labelled (diethylphosphorylated) AChE was conveniently measured as 2-PAM-reactivatable AChE activity. In homogenates in vitro, label is lost spontaneously (diethylphosphorylated AChE spontaneously reactivates) with a half-time of 27 h. In innervated diaphragm, labelled non-end-plate AChE is lost with a half-time of 13 h. When correction is made for the spontaneous loss of label on the basis of in vitro measurements, this data indicates that non-end-plate AChE turns over with a half-time of about 26 h. In innervated diaphragm, labelled end-plate-specific AChE is lost more slowly than non-end-plate AChE and at a rate essentially identical to the rate of spontaneous loss of label in vitro. The rate of loss of labelled non-end-plate AChE is essentially identical in 18 h denervated and in paired innervated diaphragms. The rate of loss of labelled end-plate-specific AChE is significantly faster in 18 h denervated diaphragms than in paired innervated diaphragms. On the basis of these observations, hypotheses concerning the mechanisms of the denervation-induced decreases in non-end-plate and end-plate-specific AChE are formulated and discussed.

The effect of spontaneous electromechanical activity on the metabolism of acetylcholinesterase in cultured embryonic rat myotubes.

We have investigated the effect of electromechanical activity on the molecular forms of acetylcholinesterase (AChE) in cultured embryonic rat myotubes. Both globular and asymmetric forms of AChE are present on the 5th day of culture when myotubes are just beginning to fibrillate. Between days 5 and 8, the 4 S (G1), 10 S (G4), and 16 S (A12) forms increase dramatically, and appreciable 12.5 S (A8) AChE appears. When fibrillation is prevented by adding tetrodotoxin on day 4, the increases in the A12 and A8 forms are prevented, and the increases in the G4 and G1 forms are significantly impaired. At 8 days, fibrillating myotubes have 19 times more A12 AChE and over 4 times more G1 and G4 enzyme than do nonfibrillating myotubes. The effect of tetrodotoxin is reversible. When tetrodotoxin is removed at 7 days, fibrillation resumes promptly, and globular and asymmetric forms recover. Light microscopic examination of fibrillating and nonfibrillating myotubes showed that tetrodotoxin does not affect the gross morphological development of the myotubes. Titration of AChE-active sites with O-ethyl-S2-diisopropyl methyl-phosphonothionate demonstrated that the increase in AChE activity associated with fibrillation is due to an increase in the number of AChE molecules present and not to an increase in the rate at which individual AChE molecules turn over acetylcholine. To evaluate AChE metabolism in fibrillating and nonfibrillating myotubes, we examined the enzyme after inactivating it with paraoxon. Paraoxon readily penetrates cells and diethylphosphorylates a serine in the active site of AChE, thereby inactivating it. The diethylphosphorylated enzyme is stable, but it can be reactivated rapidly and quantitatively with pyridine-2-aldoxime methiodide (2-PAM). After inactivating AChE with paraoxon, we simultaneously evaluated synthesis (by following the newly synthesized active AChE) and turnover (by following the 2-PAM-reactivatable AChE). Our results show that globular and asymmetric forms of AChE are both synthesized more rapidly in fibrillating than in nonfibrillating myotubes.