Rudolph E. Tanzi, PhD, Harvard University - US grants

The simultaneous discovery of genetic linkage between familial Alzheimer's disease (FAD) and chromosome 21 DNA markers in four large FAD pedigrees combined with the mapping of the amyloid beta protein precursor gene (APP) to the same chromosome suggested in 1987 that an FAD gene locus may have been identified. When the same four pedigrees were tested directly for linkage to APP, the occurrence of at least one crossover in each family along with other reports of non-linkage dashed the hope that a gene responsible for FAD had been isolated. Two recent findings have rekindled interest in the potential role of the APP gene in the etiology and have consequently, prompted the need for a careful re-evaluation of the APP gene in FAD. First, non-allelic genetic heterogeneity of FAD was demonstrated, implying that while a subset of pedigrees are chromosome 21-linked, others, and in particular, those with late age of onset are either linked to loci residing elsewhere in the genome or involve an enviromental origin. Since families not linked to chromosome 21 could effectively mask overall genetic linkage to APP, it remains possible that APP represents the site of a defect in some families. This possibility has been strengthened by the recent discovery of an apparent missense mutation in exon 17 of APP in five FAD pedigrees, and a separate mutation in this same exon in patients from three pedigrees with hereditary cerebral hemorrhage with amyloidosis-Dutch type (HCHWA-D). The FAD-related mutation in exon 17 disrupts a putative iron-response element (IRE) in the C-terminal BetaA4 region of APP RNA which is homologous with those found in ferritin and transferrin receptor messages. These elements have been shown to interact with specific binding proteins which regulate the level of translation in response to iron concentration. By analogy, the putative IRE in APP RNA may play a similar role. This possibility and the effects of a mutation on the stability of this structure need to be addressed. In summary, the strong likelihood that APP mutations are responsible for some forms of AD and the insight that they would provide on the basic etiology of this disorder indicate the need for a careful assessment of 1. the extent to which defects in the APP gene are associated with AD, and 2. the potential consequences of a known mutation in the FAD gene with respect to AD pathology.

Recent evidence indicates that the amyloid p protein precursor (APP) gene is a member of a highly conserved family of related genes. An APP-like gene (APPL) has been isolated from Drosophila and another has been reported in the form of a partial cDNA isolated from rat testes. More recently, our laboratory has isolated human cDNA clones encoding two novel APP-like proteins (amyloid precursor-like proteins, APLP1 and APLP2). These proteins are highly homologous to APP at the amino acid level. The beta-A4 domain is only partially conserved in APLP1 and APLP2, however, the remarkable degree of conservation of amino acid identity and specific domains within this protein family suggest that these proteins may share common functions, and, perhaps interact with common factors that are involved with protein processing and gene regulation. Based on our preliminary findings, we propose that production, maturation, and metabolism of APP, APLP1, and APLP2 in neurons in which these proteins appear to be co-expressed, may be critically balanced. We hypothesize that changes in the stoichiometry of APP, APLP1, and APLP2 message and/or protein effectively divert APP into alternative metabolic pathways including those predisposed toward amyloid formation. We have mapped the human APLP2 gene to chromosome 11, and the APLP1 gene to the proximal long arm of human chromosome 19 where a late-onset form of FAD has also been assigned making APLP an excellent candidate for this gene defect. We propose to further characterize APLP1 and APLP2 and isolate other members of the APP-like gene family, concurrently comparing gene expression, regulation, and processing of these proteins. We have chosen to explore the APLP family at the DNA, RNA and protein levels because we believe that the potential importance of these novel proteins in the etiology of Alzheimer's disease and aging deserves such a prompt overall assessment. We fully appreciate that comprehensive studies of expression, regulation and processing of the APLP family could each entail grant proposals of their own. Therefore, we have devised a set of experiments aimed at providing in a focused manner, critical data points in each of these areas of investigation. The ultimate goal of the experiments proposed is to provide a rudimentary data set assessing the potential roles played by the APLP gene family in the process of Alzheimer- and age-related neurodegeneration.

In the first two years of this project we performed exhaustive studies that ultimately led us to conclude that only a small subset (2-3%) of familial Alzheimer disease (FAD) kindreds possess mutations in the gene encoding the Abeta peptide precursor (APP). This has been confirmed by other large-scale studies performed around the world. These data along with other family studies have shown that FAD is clearly heterogeneous. For example, linkage studies using markers from other chromosomes have led to the localization of a late-onset (> 65 years) FAD locus mapping within or close to the APOE gene on chromosome 19. More recently, our extensive FAD pedigree collection that we have amassed over the past ten years has allowed our laboratory in collaboration with Dr. Peter Hyslop in Toronto, to localize a major early-onset FAD gene on chromosome l4. Similar findings were initially reported by Schellenberg and colleagues and later confirmed by our group as well as those of Mullan and Van Broeckhoven. Among the pedigrees screened in our Boston-Toronto study, six independent pedigrees individually yielded lod scores exceeding the conventional significance value of 3. The magnitude and robustness of these scores far exceeds any lod scores achieved on chromosome 2l or l9, and indicates definitively that there is a chromosome l4 locus that most likely accounts for a major proportion (70-90%) of early-onset FAD (<65 years). In view of these combined findings, we have chosen to spend a major portion of year 03 of this project in attempting to isolate the chromosome 14 FAD gene, and have already made substantial progress toward this goal. Having accomplished the primary goal of the originally proposed studies on the APP gene, we propose to re-channel the funding for this project toward the exceedingly important task of isolating and preliminarily characterizing the chromosome 14 FAD gene defect. Therefore, the goal of the current proposal is to employ a variety of strategies including positional cloning, physical mapping, and testing of candidate genes to ultimately isolate, identify, and provide a preliminary characterization of the FAD gene defect on chromosome l4.

In most cases of Alzheimer's diseased (AD), it is impossible to identity the triggering event which leads to the neurodegenerative process. However, in a small subset (2-3%) of familial Alzheimer disease (FAD) kindreds, several mutations have been found in a chromosome 21 gene, APP encoding the precursor of the Abeta peptide that is deposited in amyloid plaques. Detailed family studies have shown that AD is genetically heterogeneous and linkage studies using markers from other chromosomes have led to the localization of a late-onset (more 65 years) FAD locus mapping within or close to the APOE gene on chromosome 19. More recently, the extended FAD pedigree collections made possible through the efforts of the Massachusetts ADRC over the past ten years have allowed our laboratory in collaboration to localize a major FAD locus on chromosome 14. Among the FAD pedigrees screened in our Boston-Toronto study, six independent pedigrees individually yielded lod scores exceeding the conventional significance value of 3. The magnitude and robustness of these scores indicate definitively that there is a chromosome 14 locus that most likely accounts for a major proportion (70-90%) of early-onset FAD (lessor 65 years). The goal of this proposal is to employ a variety of strategies including positional cloning, physical mapping, and testing of candidate genes to ultimately isolate, identify, and characterize the FAD gene defect on chromosome 14. Given our extensive collection of well characterized early-onset FAD pedigrees and the large amount of preliminary data that we have already amassed concerning the chromosome 14 FAD locus, we are in a uniquely ideal position to successfully carry out the studies described in this application. The knowledge obtained from the proposed studies could provide critical clues to the causes of both inherited and sporadic AD, and ultimately provide vital information for designing treatments aimed at strategically intervening with the effects of discovered FAD gene defects.

Recent evidence indicates that the amyloid beta protein precursor (APP) gene is a member of a highly conserved family of related genes. An APP- like gene (APPL) has been isolated from Drosophila and another has been reported in the form of a partial cDNA isolated from rat testes. More recently, our laboratory has isolated human cDNA clones encoding two novel APP-like proteins (amyloid precursor-like proteins, APLP1 and APLP2). These proteins are highly homologous to App at the amino acid level. The beta A4 domain is only partially conserved in APLP1 and APLP2, however, the remarkable degree of conservation of amino acid identity and specific domains within this protein family suggest that these proteins may share common functions, and perhaps interact with common factors that are involved with protein processing and gene regulation. Based on our preliminary findings, we propose that production, maturation, and metabolism of APP, APLP1, and APLP2 in neurons in which these proteins appear to be co-expressed, may be critically balanced. We hypothesize that changes in the stoichiometry of APP, APLP1 and APLP2 message an/or protein effectively divert APP into alternative metabolic pathways including those predisposed toward amyloid formation. We have mapped the human APLP2 gene to chromosome 11, and the APLP1 gene to the proximal long arm of human chromosome 19 where a late- onset form of FAD has also been assigned making APLP an excellent candidate for this gene defect. We propose to further characterize APLP1 and APLP2 and isolate other members of the APP-like gene family, concurrently comparing gene expression, regulation, and processing of these proteins. We have chosen to explore the APLp family at the DNA, RNA and protein levels because we believe that the potential importance of these novel proteins in the etiology of Alzheimer's disease and aging deserves such a prompt overall assessment. We fully appreciate that comprehensive studies of expression, regulation and processing of the APLP family could each entail grant proposals of their own. Therefore, we have devised a set of experiments aimed at providing in a focused manner, critical data points in each of these areas of investigation. The ultimate goal of the experiments proposed is to provide a rudimentary data set assessing the potential roles played by the APLP gene family in the process of Alzheimer- and age-related neurodegeneration.

DESCRIPTION (adapted from applicant's abstract) The genes encoding the amyloid beta protein precursor (APP) and presenilins 1 and 2 (PS1 and PS2) have been shown to harbor autosomal dominant gene defects in approximately 50 % of early-onset (<60 years) cases of familial Alzheimer's disease (FAD) tested. Recently Dr. Steve Younkin, in collaboration with this and other laboratories found PS2 and PS1 FAD mutations to be associated with elevated amounts of Abeta42 in the plasma and fibroblasts of carriers as well as in cells and transgenic mice overexpressing mutant PS1. These data suggest that PS1 and PS2 gene defects steer APP metabolism toward a pathway leading to the increased production of Abeta42. The P.I. and his collaborators have recently observed two additional molecular consequences of presenilin FAD mutation. First, they observed that overexpression of wild-type or mutant PS1 or PS2 leads to increased cell death and enhanced sensitivity to apoptotic stimuli. Second, they discovered that a PS2 FAD mutation associated with increased Abeta42, also causes PS2 proteolytic by-products to hyper-accumulate in a pre-Golgi intracellular compartment of the cell. Based on the findings described above, the goal of this proposal is to generate first order information regarding the mechanism of PS1-associated cell death and the potential role of PS1 in apoptosis. To this end, the P.I. proposes to expand upon his preliminary findings showing that prolonged expression of PS1 leads to enhanced sensitivity to cytotoxic and apoptotic stimuli. He will also explore whether cell death induced by PS1 requires elevated production of Abeta42 and APP expression and test whether PS1 mutations lead to altered maturation, processing, and/or subcellular distribution of APP. Finally, PS1-expressing cells will be examined for changes in the trafficking, degradation/aggregation, and processing of PS1 in the ER-Golgi. Throughout these studies, correlations will be sought between the rate of cell death/enhanced sensitivity to cytotoxic stimuli and changes in the post-translational processing and subcellular localization of APP and PS1 in a variety of stably-transfected and inducible wild-type and mutant PS1 cell lines as well as in primary neurons prepared from the brains of wild-type and mutant PS1 transgenic mice.

Core C, the Genetics Core, is designed: 1) To establish immortalized lymphoblast cell lines from blood samples from all participants in the Program Project. 2) To generate genotypes for APOE and other genetic risk factors associated with familial AD (FAD), using DNA extracted from the blood samples and lymphoblast cell lines. 3) to provide genetic data to the Core Data Base so that it can be combined with other data in the Program Project (e.g., neurosychologyical test scores, SPECT, fMR, etc.) to predict development of AD in subjects at risk for disease.

Presenilin 1 and 2 (PS1, PS2) both undergo regulated endoproteolytic processing to yield two stable fragments. We have recently discovered that PS1 and PS2 will be alternatively cleaved at sites distal to the normal cleavage sites during apoptosis, and when over-expressed in transfected cell lines. Apoptosis-associated endoproteolysis of PS1 and PS2 can be blocked by the apoptotic cysteine protease inhibitors, zVAD, a general caspase inhibitor, and zDEVD, a selective caspase-3 (CPP32) inhibitor, indicating that the enzyme responsible is a caspase-3 family protease. We have identified the apoptotic cleavage site in PS2 as DSYDS (a.a. 326-330) by site directed mutagenesis. In support of a potential role for apoptotic cleavage of the presenilins in FAD, the ratio of apoptotic:normal cleavage fragments was elevated by over 3-fold in cells expressing the FAD mutant PS2-N141I as compared to w.t. PS2-expressing cells. Since FAD mutations in PS1/PS2 have been associated with an elevated ration of Abeta42:Abeta40, we have generated preliminary data to test whether inhibition of the caspase-3 mediated cleavage of mutant PS2 could repress the increased Abeta42:Abeta40 ration associated with FAD mutations. Treatment with zVAD repressed the increases in Abeta42:total Abeta ratio associated with the N141I and M239V FAD mutations in PS2 by 44%. Collectively, these findings not only suggest that the presenilins may be cell death substrates, but not apoptotic cleavage of the presenilins by a caspase 3 family protease may alter the Abeta42:Abeta40 ration and play a role in the pathogenesis of FAD. In view of these novel findings, we will test the following hypotheses: 1. Presenilin FAD mutations lead to increased apoptosis- associated endoproteolysis of the presenilins by a caspase-3 family protease; 2. Apoptosis-associated cleavage of the presenilins leads to an increased Abeta42:Abeta40 ratio; and 4. Apoptotic endoproteolysis of the presenilins alters the subcellular distribution and/or molecular interactions of the presenilins.

Core B will have two primary roles in the Program Project. Core B will be responsible for maintaining and supplying shared cell lines/cDNA constructs, fusion proteins, baculovirus constructs, and antibodies for all investigators involved in the Program Project. Secondly, the Core will perform the Abeta ELISA for quantitating total Abeta, Abeta42 (and other ELISA's as they are developed), in a timely and convenient manner for all investigators involved in the Program Project. Core B is basically intended to serve as a central repository for maintaining and supplying a large number of highly useful reagents that have been generated over the last decade by the Investigators involved in the Program Project. These reagents which include a large variety of stably-transfected mammalian cell lines and corresponding transgene cDNA constructs, several fusion proteins/baculovirus constructs and corresponding cell lines and extensive collection of specific polyclonal and monoclonal antibodies greatly facilitate ongoing collaborations and overall experimental efficiency within the Program Project. Since a major molecular consequence of FAD mutations in the presenilin genes as an increase in the ratio of Abeta42:Abeta40, all four projects and the transgenic core will require an effective and convenient means for obtaining quantitative measurements of Abeta. Therefore, Core B will be available to carry out the Abeta ELISA whenever desired by a Program Investigator. While the expenses for maintain the Core reagents and performing the Abeta ELISA (and other ELISAs, e.g. for presenilins, as they become available) will be incurred by Core B, individual project expenses including those for large scale- workup of cell lines for specific biological experiments or protein purifications will be covered by individual project budgets. Core B activities will be overseen by Dr. Tanzi and Dr. Selkoe while day-to-day operations of the Core will be supervised by Dr. Wilma Wasco.

family genetics; gene expression; Alzheimer's disease; genetic mapping; case history; genetic markers; chromosomes; genotype; siblings; clinical research; cell line; tissue /cell culture; human tissue; human subject;

Presenilin 1 and 2 (PS1, PS2) both undergo regulated endoproteolytic processing to yield two stable fragments. We have recently discovered that PS1 and PS2 will be alternatively cleaved at sites distal to the normal cleavage sites during apoptosis, and when over-expressed in transfected cell lines. Apoptosis-associated endoproteolysis of PS1 and PS2 can be blocked by the apoptotic cysteine protease inhibitors, zVAD, a general caspase inhibitor, and zDEVD, a selective caspase-3 (CPP32) inhibitor, indicating that the enzyme responsible is a caspase-3 family protease. We have identified the apoptotic cleavage site in PS2 as DSYDS (a.a. 326-330) by site directed mutagenesis. In support of a potential role for apoptotic cleavage of the presenilins in FAD, the ratio of apoptotic:normal cleavage fragments was elevated by over 3-fold in cells expressing the FAD mutant PS2-N141I as compared to w.t. PS2-expressing cells. Since FAD mutations in PS1/PS2 have been associated with an elevated ration of Abeta42:Abeta40, we have generated preliminary data to test whether inhibition of the caspase-3 mediated cleavage of mutant PS2 could repress the increased Abeta42:Abeta40 ration associated with FAD mutations. Treatment with zVAD repressed the increases in Abeta42:total Abeta ratio associated with the N141I and M239V FAD mutations in PS2 by 44%. Collectively, these findings not only suggest that the presenilins may be cell death substrates, but not apoptotic cleavage of the presenilins by a caspase 3 family protease may alter the Abeta42:Abeta40 ration and play a role in the pathogenesis of FAD. In view of these novel findings, we will test the following hypotheses: 1. Presenilin FAD mutations lead to increased apoptosis- associated endoproteolysis of the presenilins by a caspase-3 family protease; 2. Apoptosis-associated cleavage of the presenilins leads to an increased Abeta42:Abeta40 ratio; and 4. Apoptotic endoproteolysis of the presenilins alters the subcellular distribution and/or molecular interactions of the presenilins.

genetic screening; genetic susceptibility; genotype; genetic markers; family genetics; Alzheimer's disease; disease /disorder onset; apolipoprotein E; MHC class I antigen; very low density lipoprotein; linkage mapping; chymotrypsin inhibitor; genetic polymorphism; amyloid proteins; low density lipoprotein receptor; presenilin; clinical research; human genetic material tag; human tissue;

Since 1989, we have led the ascertainment,evaluation, and genetic analysesof the NIMH GeneticsInitiative Alzheimer's disease (AD) sample. This set of AD families, which currently includes 457 families, is the argest uniformly ascertained and evaluated sample assembled for the study of AD genetics. Over the first ive years of this R37, we have continued to carry out genetic analysesand follow-up studies of our a high- resolution whole genome genetic linkage screen for novel AD candidate genes. We have focused on linkage peaks identified on chromosomes 9 and 10, and identified late-onsetAD candidate genes in both ofthese inkage peaks: insulin degrading enzyme (IDE;10q) and ubiquilin 1 (UBQLN1;9q). In addition tofunctionally characterizing these genes for potential pathogenic effects in AD, we have built a highly efficient gene discovery pipeline, customized for large-scale generation, confirmation, storage and analysis of SNP genotype information. Our NIMH family-based association database of positional candidate gene contains over 500,000 genotypes for more than 300 SNPs in over 150 genestested in the full NIMH sample. Furthermore, we have recently obtained nearly one billion genotypes for the NIMH sample from a high- density, whole genome association screen using the Affymetrix 500K SNPchip, currently under analysis. By early October, we will also have obtained the results of a screen of the Affymetrix20,000 coding region SNPs (cSNPs) for the entire NIMH AD family sample. This screen should identify SNPs with a higher probability of potentially pathologenic effects.These combined data sets now provide anunprecedented array of AD candidate genes collectively accounting for as much as 95% of the genetic van'ability of AD. In the second half of this R37, we will carry out confirmatory and follow-up studies of these AD candidate genes. First, SNPs exhibiting family-basedassociation with AD will be confirmed by manual genotyping. Second, they will be confirmed in additional AD family-based samples. Third, tagging SNPs for all inkage disequilibrium blocks in confirmed candidate genes will be tested in the NIMH and confirmatory samples. Fourth, bioinformatic analyses and resequencing will be performed to identify potentially pathogenic mutations and gene variants. The strongest candidate genes will then be validated in a wide array of functional analyses to assess the potential pathogenicity of AD-associatedgene variants.

This proposal is aimed at continuing our ongoing efforts to identify and characterize novel AD genes involved in presenilin-related pathways. AD candidate genes implicated in presenilin-related pathways will be derived from three different pools: Pool 1. Positional candidate genes mapping to established AD genetic linkage peaks, including UBQLN1, VPS26A, VPS35, VDAC1, VDAC2, NCSTN, PSEN1 and TFCP2; Pool 2. Candidate genes derived from systematic meta-analyses performed on ourAlzGene.org database, including CHRNB2, DAPK1, SORCS1, SORL1, TNK1, HMGCS2, CH25H, and SOAT1; and 3. Novel AD candidate genes from our ongoing whole-genome association (WGA) screens of the NIMH AD family sample in which >1400 subjects from 457 uniformly ascertained and evaluated AD families have been genotyped using three different Affymetrix genotyping arrays: 500K genomic single nucleotide polymorphisms (SNPs), 100K genomic SNPs, and 20K coding SNPS (cSNPs). Follow-up analyses of presenilin pathway-related AD candidate genes will include genetic confirmation/replication testing, linkage disequilibrium analyses, and mutation identification. In collaboration with the other P01 projects and cores, we will also carry out biological and functional validation studies of specific candidate genes based on our genetic results. In specific aim 1, genotyping of the NIMH sample will be completed for all genes in all three pools. In specific aim 2, SNPs exhibiting genome-wide significance for family-based association with AD in the NIMH sample will be subjected to replication testing in four independent AD family samples: CAG (224 families; 505 subjects; AD: 245), NIA (353 families; 1117 DNAs; AD: 815), NCRAD (369 families; 1266 DNAs: AD: 895), and NIMH African American (24 families, 58 subjects; AD: 49). For genes exhibiting the strongest association with AD, we will carry out extensive linkage disequilibrium mapping of additional SNPs and re-sequencing in probands and unaffected individuals of specific associated families for each locus. In specific aim 3., we will perform bioinformatic (in silico) analyses of AD candidate genes to identify which SNPs represent potentially pathogenic gene mutations/variants for AD. Finally, in specific aim 4., we will collaborate with the other P01 projects and cores to carry out biological validation and functional analyses of novel AD candidate genes, including effects of RNAi silencing and overexpression of wild-type and potentially pathogenic mutations/variants on presenilin function, e.g. APP trafficking/processing as well as A(3 and AICD generation, y-secretase activity, APP-PS1 interaction, and PS1 conformation. Lay Summary: The four known AD genes (APP, PSEN1, PSEN2, and APOE) are the subjects of the vast majority of current biological research on AD. Yet, these genes represent only ~30% of the genetic variance of AD. The goal of this project is identify the additional AD genes implicated in presenilin-related biological pathways to increase our knowledge of the causes of AD and the role of the presenilins in AD pathogenesis.

DESCRIPTION (provided by applicant): Abundant clinical, genetic, and biochemical data support the hypothesis that abnormal processing of the amyloid precursor protein (APP) and the cerebral accumulation of its metabolite, A¿, play key roles in the etiology and pathogenesis of Alzheimer's disease (AD). A¿ is generated via serial cleavage of APP by ¿- and ¿- secretase. In contrast, cleavage of APP by ¿-secretase precludes A¿ production; the major ¿-secretase in the brain is ADAM10. We recently reported two rare missense mutations in ADAM10 that strongly co- segregates with late-onset AD (LOAD). Both are prodomain mutations that were found in 7 of 1000 NIMH and NIA LOAD families tested (average age of onset, ~70 years). We have also previously reported that both mutations significantly attenuated ¿-secretase activity and elevated A¿ levels in vitro. To further validate and expand upon these novel findings, we propose to 1. Search for additional novel familial LOAD mutations in ADAM10; 2. Validate the pathogenicity of these two missense mutations in vivo; and 3. Determine the molecular mechanism by which these two mutations impair ¿-secretase activity. For Aim 1, we will search for additional novel familial LOAD mutations in ADAM10, through targeted re-sequencing (and genotyping) of 2454 additional AD families (N=6516 subjects) from the NIMH and NIA AD family collections. Our family-based GWAS on >900 NIMH and NIA AD families suggest the existence of additional AD-associated SNPs in ADAM10. With regard to Aim 2, we will attempt to validate the pathogenicity of these two mutations in vivo in transgenic mice (already generated) overexpressing either wild-type (WT) or mutant (Q170H, R181G, artificial dominant-negative) forms of ADAM10. These mice have already been crossed with Tg2576 (APPswe) AD mice so that we can also test for effects of these mutations on neuropathological and cognitive phenotypes of AD. Our preliminary in vivo results reveal that the ADAM10 prodomain mutations attenuate non-amyloidogenic processing of APP, and that senile plaque counts and A¿ levels were significantly increased in the brains of APPswe/ADAM10 double transgenic mice expressing mutant forms of ADAM10, as compared to those expressing WT forms. Moreover, to test the impact of the prodomain mutations under more physiologically relevant conditions, we have added a plan to generate and characterize ADAM10 mutant knock-in mice. Finally, in Aim 3, we will investigate the molecular mechanism by which the prodomain mutations down- regulate ADAM10 ¿-secretase activity. Briefly, we will test for effects of these mutations on three prodomain functions: regulation of enzyme activity, intracellular trafficking, and intramolecular chaperoning. At the completion of this project, we hope to provide genetic, biochemical, and mechanistic evidence validating the pathogenicity of late-onset familial AD mutations in ADAM10. Moreover, the data emerging from the proposed study would serve as a firm foundation for the discovery and development of new therapies targeting ADAM10 for the treatment and prevention of this devastating disease. PUBLIC HEALTH RELEVANCE: We will attempt to establish in vivo validation for the pathogenicity and to elucidate the molecular mechanism of two ADAM10 prodomain mutations, which we have already confirmed in several late-onset Alzheimer's disease (AD) families. We will also search for additional AD mutations in the ADAM10 gene.

DESCRIPTION (provided by applicant): This project began in 1994 as a NIMH U01 in which we initially completed whole-genome linkage analysis on 437 uniformly ascertained and evaluated Alzheimer's disease families comprising the NIMH AD Genetics Initiative Family Sample (NIMH sample). This led to novel loci chromosomes 9, 10 and 12. In 1998, the U01 was transformed into an R01, and in 2003, became a Merit Award (R37). Since 1998, the project was aimed at the identification and characterization of novel familial AD genes using family-based association analyses. In 2005, we initiated genome-wide association studies (GWAS) using Affymetrix 500K genotyping arrays and in 2008, we reported genome-wide significant results from this screen, focusing on CD33, and ATXN1. In 2009, we reported two rare, pathogenic AD mutations in the a-secretase gene, ADAM10. Over the past three years, we completed additional GWAS on the NIMH sample and another AD family sample (NCRAD), using the Affymetrix 6.0 (~900K single nucleotide polymorphisms [SNPs] and ~900K single copy probes) and Affymetrix 25K Coding SNP genotyping arrays. In addition, we carried out GWAS on four AD clinical-, pathological-, imaging-, and biomarker-based quantitative endophenotype samples. These GWAS data are being analyzed separately and as part of a meta-analysis using several imputed AD case-control GWAS datasets. In addition to the initial description of CD33 as a genome-wide significant AD risk gene, later confirmed independently in case-control samples, we identified SNPs in several other candidate genes exhibiting significant results by meta-analysis (see Preliminary Data). In the renewal of this project, we propose to analyze, validate, and follow-up the results of whole genome sequencing (WGS) of the entire NIMH sample (1510 subjects in 437 AD families) using the Illumina HiSeq 2000 platform. WGS sequencing costs will be covered by the Cure Alzheimer's Fund. We will employ state-of-the-art statistical and bioinformatic approaches to identify functional genomic variants influencing AD risk and time-to-onset. In Aim 1, we will analyze WGS data from the NIMH sample to identify functional variants that are either linked to Mendelian forms of early-onset AD (EOAD) or associated with risk for late-onset AD (LOAD). We will also carry out various association analyses i.e. gene- based, multi-marker framework, and extreme discordant sib-pair, to identify common variants that influence risk for LOAD. In Aim 2, we will set out to replicate novel variants discovered in Aim 1 using two independent AD family-based cohorts. We will also impute rare variants and conduct association analyses on four case- control samples (NIA-LOAD, TGEN2, GenADA, and ADNI). We will also sequence candidate loci in African- Americans and Pacific Islanders, and compare our results from AD families to those obtained in various psychiatric disorders. In Aim 3, we will carry out preliminary functional studies on select functional variants, assessing effects on gene function and expression and AD-related pathogenicity, e.g. Ab metabolism and tauopathy. For promising variants, in vivo validation will be performed in independently funded future studies.

Alphabeta generation occurs via serial cleavage of APP by p- and gamma-secretase, but is precluded via serial cleavage by alpha- and gamma-secretase. Project 1 (Selkoe/Wolfe) has generated evidence for an alpha/gamma-secretase complex, which we have termed the sheddasome. In this project, we will explore factors that can modulate the sheddasome either pharmacologically (using novel gamma-secretase modulators [GSM]), or genetically (using novel late-onset Alzheimer's disease [LOAD] mutations in ADAMVO which we discovered during the current PPG period). In our first set of aims, we will carry out studies of a novel series of highly potent, APP-specific GSMs. These GSMs are aryl 2-aminothiazole GSMs that bind directly to the gamma-secretase complex, decreasing AP42 and AP40 levels and increasing APas and AP37 levels. This project will serve to leverage an independent ongoing project on these GSMs supported by the NIH Blueprint Neurotherapeutics Network (NIH-BNN). While we receive no funding from the NIH-BNN, we co-developed these GSM' and serve as close collaborators and Lead Development Team members for the project. As part of Aim 1, in collaboration with Project 1, we will test for the effects of these novel GSMs on gamma-secretase preparations, including gamma-secretase- enriched membranes and sheddasome complexes. We will also test whether the GSMs allosterically impact PS1 conformation in the gamma-secretase complex in collaboration with Project 3 (Berezovska/Hyman). Finally, we will test for the potency and substrate selectivity of these GSMs in collaboration with Project 4 (Kovacs). In the second set of aims, we will characterize two rare LOAD missense mutations in the prodomain oi ADAMIO, which we have already shown to tightly co-segregate with LOAD in 7 families (age of onset ~70 yr). We have also shown that both of these mutations significantly attenuate alpha-secretase activity and elevate alphabeta levels (relative to wild-type) in vitro and in vivo. We have already generated transgenic mice overexpressing either wild-type (WT) or mutant (Q170H; R181G; dominant-negative) forms of ADAMIO. We will use these animal models to identify physiologic ADAM 10 substrates, and characterize their serial cleavage by gamma-secretase (with Project 4). In collaboration with Project 1, we will test whether the LOAD and dominant negative mutations in ADAMIO affect the interaction of ADAM10 with gamma-secretase in the sheddasome.

Since its cloning in 1995 and its identification (under this grant) as an unprecedented intramembrane aspartyl protease in 1999, Presenilin has been implicated in a remarkable array of signaling and regulatory events in all metazoans. PS was discovered through research on Alzheimer's disease, but it was soon shown to confer functions necessary for life, especially as the protease that enables Notch nuclear signaling. Thus, continuing to decipher the structure, functions, and protein and lipid regulators of PS is a priority for basic cell biology. At the same time, the invariant cerebral accumulation of amyloid beta-protein (AU) decades before symptoms of dementia has made PS/gamma-secretase a key target for mechanistic and therapeutic study in AD. Despite its pleiotropic role in biology, the protease's structure has only been resolved at 12 A (under this grant), and small molecules that can selectively inhibit its processing of APP are not yet validated. For all these reasons, six collaborators with deep experience in the study of Presenilin wish to apply a range of techniques in cell biology, genetics, chemistry, structural biology and animal modeling to tackle some of the thorniest questions in PS/gamma-secretase biology. Can one derive an atomic resolution structure of this 19- transmembrane complex? What is the cell biological mechanism of coordinated alpha-, beta- and gamma-secretase processing? How do certain synaptic proteins and membrane lipids regulate PS activity in neurons, affecting the crucial A(l42/4o ratio? Can one design drugs that are sufficiently potent yet selective to chronically inhibit gamma-secretase? Our group has carefully revised this application to address all of the thoughtful critiques the SEP offered. We propose numerous interrelated aims that incorporate three cross-cutting themes which unite our work. First, we will further confirm and extend our recent discovery of an endogenous complex of the alpha/beta/gamma-secretases (a sheddasome) that may mediate the efficient, sequential processing of APP - and presumably all gamma-substrates. Second, we will apply a unique FRET-based probe developed here to measure PS conformation in living neurons and learn if certain synaptotagmins we recently identified by proteomics as novel interactors of both PS1 and APP enable PS to change its conformation rapidly and reversibly in response to Ca2+ influx at the synapse, explaining the enhancement of AB production by neural activity. Third, we'll study novel gamma-secretase modulators and Notch-sparing inhibitors we've developed to define SARs for APP vs. Notch cleavage, identify the cognate binding sites, and assess their actions on other substrates, all with the goal of advancing one or more into preclinical development. In short, we are committed to applying novel approaches to elucidate the structure and function of gamma-secretase in health and disease.

The discovery of presenilin (PS) as the first intramembrane aspartyl protease and the catalytic center of gamma-secretase occurred in Project 1. Since then, >100 substrates have been identified. PS/gamma-secretase mediates critical signaling pathways necessary for life in all metazoans, and its cleavage of APP releases the amyloid B-protein that accumulates in all patients with AD. After identifying PS as a protease, Project 1 has continued to contribute actively to PS biology. We proposed - and provided the first evidence - that holoPS undergoes autocatalytic endoproteolysis to generate the active heterodimer, first reconstituted PS and its 3 cofactors in mammalian cells, purified the protease to homogeneity, obtained the first 3D structure of the ycomplex by EM, conducted SILAC screens to identify several new substrates, and designed many gamma-secretase inhibitors, some of which are potent and much more APP-selective than compounds tried in humans. The Project is now revised to respond to all of the SEP's helpful critiques. We propose to study 4 related topics in the biochemistry of gamma-secretase. 1: A new cell biological model of secretase processing We have discovered that contrary to current concepts, alpha- (ADAM 10), beta- and gamma-secretases exist in part in a large protein complex that can mediate efficient sequential processing of substrates. Our extensive supporting data include robust co-IP of endogenous a- and gamma-secretases from wt brain and the sequential alpha/gamma processing of an APP substrate. We propose to fully confirm this new model of regulated intramembrane proteolysis and ask if it generalizes to another membrane protease pair: S1P/S2P. 2: The complex regulation of gamma-secretase by membrane lipids. We will extend our recent evidence that certain head groups and fatty acyl side chains of membrane lipids potently up- and down-regulate gamma-cleavages, including the key A(i42/40 ratio. We'll seek to validate robust in vitro effects of certain lipids by manipulating their cognate biosynthetic and catabolic enzymes in vivo. 3: Toward greater structural resolution of the gamma-secretase complex Working with leading structural biologists, we will pursue the technically challenging but essential quest for the structure of gamma-secretase via: a) further cryo-EM analyses of 2D crystals; b) 3D x-ray crystallography of individual gamma-components (PS, Net, Pen-2); and c) attempted 3D x-ray crystallography of the purified holo-enzyme. 4: Refining potent and selective Notch-sparing gamma-inhibitors and defining their mechanism. Building on more than 1,600 compounds we've synthesized, we will develop SARs for inhibiting APP vs. Notch, and for the most selective compounds, assess cleavage of other gamma-substrates and test them in mice. These aims build on our experience to tackle some of the thorniest problems in gamma-secretase biology.

The Administrative and Educational Core plays the key role of managing and integrating all the activities of our highly interactive research group. The aims of Core A (below) are designed to effectively administer the scientific planning, data and reagent sharing, education, consultation and administrative functions of our program, and they reflect the commitment of the 5 principal investigators to their continued scientific development as well as that of the program as a whole. Expenses for all non-experimental parts of this Program are included in the Core A budget. The Aims of Core A are: 1) To enhance the scientific exchange, collaboration and communication among all the co-investigators and their respective scientific personnel across the 6 labs of the program (4 at MGH and 2 at BWH), including inter-institutional personnel and reagent exchanges, assisting with joint publications, regular exchange of all publications emanating from the grant, and internal sharing of mentoring responsibilities. 2) To contribute to the continuing education of the program's members and the Boston scientific community at large about new developments in basic and applied biology related to neurodegeneration, via our highly successful seminar series. 3) To efficiently manage the program's business activities, including regular income/expense reports, preparation of progress reports each year, organization of this competitive renewal application, and communications with the NIA. The Administrative Coordinator, under the guidance of the Program Director, functions as the point person for all the business and administrative tasks related to the above 3 aims as they arise from the following specific responsibilities. These are: the coordination of meetings of the Working Group and the Scientific Advisory Board (SAB) and the preparation of meeting materials and any travel-related issues for SAB members; organizing, scheduling and publicizing our annual seminar series and handling all speaker-related issues, including scheduling meetings with program scientists, coordinating travel needs, processing reimbursements, etc.; assisting with distribution of publications arising from the program among our personnel as well as dealing with shipments or licensing issues related to the resources and reagents generated by the program; managing all processes related to the timely submission of reports, renewal applications or any other communications with the NIA. In short, the Administrative Coordinator ensures the smooth and timely functioning of all non-scientific aspects of the program project on a daily basis and is an indispensible part of the Core.

Synaptic dysfunction is a key feature of Alzheimer's disease (AD) and closely correlates with memory impairment. Activity-dependent production of Ap, and targeted accumulation of Ap oligomers at the synapse have been reported, however underlying molecular mechanisms remain poorly understood. We have previously suggested that presenilin (PSI) can adopt one of two conformations -termed open and closed, based on the other protein constituents of the gamma-secretase complex (Serneels et al., 2009,), on the effects of GAMMA-secretase modulators (Uemura et al., 2010), or due to familial AD-related mutations in PSI (Berezovska et al., 2005). However, whether these are different gamma-complexes that are in equilibrium with one another within a cell, or PSI can dynamically change its conformation within a specific complex, is unclear. Furthermore, events and/or protein(s) that may be involved in regulation of PSI conformation and function in the cell remain unknown. Now we established a unique conformation-sensitive FRET-based assay in living cells, and our new preliminary data show that conformation of PSI changes rapidly and reversibly in concert with synaptic activity. Aim1 will build on this observation and extend it to PSI-APP interactions at the synapse. To search for potential activity-dependent modulators of PSI conformation, we performed a proteomics screen of mouse brain lysates in the presence or absence of calcium, and identified several novel PSI interactors, including synaptotagminsi and 9 (Sytl .9), and synapsini (Svnl). Since Syt1 and Synl showed strong but opposing Ca-dependent profiles of interaction with PSI, we will direct our initial attention to these as likely candidates to modulate PSI/gamma-secretase and its interaction with APP at the synapse in an activity-gamma-controlled manner. Intriguingly, Sytl and 9 were also identified in an independent proteomics screen performed by Dr. Kovacs as APP-interacting proteins (Project 4). Aim 2 will test the hypothesis that Ca2+ influx serves as a switch between the two PSI conformational states in the synaptosomal compartment by controlling PSI interactions with synaptic proteins. Aim 3 follows on the idea that PSI can undergo rapid and reversible changes in conformation to examine the role of pharmacological interventions in allosteric modulation of PSI/gamma-secretase (collaboration with Projects 1 and 2), and its interactions with synaptic proteins in vitro and in vivo. Understanding this issue is of high importance both because the closed PSI conformation is associated with increased Ap42 (the AD-associated cleavage) and because manipulation of the PS1 interactions with Sytl and Synl may be translated into therapeutics with a focus on the synapse.

Numerous studies have shown that Alpha-beta accumulation leads to synaptic dysfunction and loss of synapses. Synaptic Alpha-beta and Alpha-beta oligomers have been strongly implicated in the progression of AD. Recent studies have demonstrated that gamma-secretase forms an active complex at synapses. However, proteins modulating synaptic APP processing remain largely unknown. We have performed two proteomic screens in mouse brain lysates to identify proteins interacting with the ectodomains of APP and nicastrin. These screens have identified synaptotagmins (Syt) 1, 2, and 9 as strong binding partners of both APP and nicastrin. Syt1 and 9 also interact with the large loop of PSI (Project 3). Sytl, 2, and 9 are type I Ca2+-sensing synaptic vesicle membrane proteins and play a major role in distinct membrane fusion events for synaptic vesicle release. Our further co-IP, EM, and in situ PLA studies demonstrated that full-length Syts bind to APP in cells and mouse brains. GST pulldown assays confirmed specific binding of Sytl to a the region surrounding the KPI domain of APP. Our preliminary data also show that overexpression of Syts increases Alpha-beta generation and ratios, by perhaps inducing p- and/or gamma-secretase-mediated processing of APP. Syt expression also increases palmitoylated APP levels. Here, we will test the hypothesis that Sytl, 2, and 9 act as physiological modulators of APP processing in neurons. Specifically, we will further explore the regulation of Syt binding to APP, dissect the contribution of Syts to Alpha-beta generation in neuronal models and will ask whether Syts affect the neuronal functions of gamma-secretase. In addition, we will also analyze the effect of novel APP-specific gamma-secretase inhibitors (Project 1) and gamma-secretase modulators (Project 2) on the neuronal functions of gamma-secretase with particular focus on synaptic proteins, and on the processing of non-APP gamma-secretase substrates. These experiments are aimed at identifying compounds lacking adverse effects commonly associated with gamma-secretase inhibitors. Our project will contribute to the overall goal of the program project in providing mechanistic data that may serve in the development of novel p- or gamma-secretase-based strategies forthe prevention and treatment of Alzheimer's disease.

DESCRIPTION (provided by applicant): Enzymes in the cholesterol pathway have emerged as effective drug targets for the prevention and treatment of Alzheimer's disease (AD). Although statins reduced beta-amyloid (A?) in cells and animal models, so far yielded disappointing results in phase III clinical trials for AD. A cholesterol-modifying enzyme named acyl-coenzyme A: cholesterol acyltransferase (ACAT) is becoming an exciting target for AD therapy and atherosclerosis. All three classes of existing ACAT inhibitors, knockdown of ACAT in cells and in AD mouse models reduce A? production. We have recently made considerable progress toward understanding the mechanism by which ACAT inhibition decreases A? levels. Specifically, we have found that the amyloid precursor protein (APP) is modified by the addition of a fatty acid in a process called palmitoylation. We were able to show that palmitoylation of APP is highly regulated by ACAT activity. In addition, we have identified a series of novel ACAT inhibitors. The overarching goal of this application is to elucidate the precise mechanism of action and therapeutic efficacy of ACAT inhibitors in AD, with particular focus on our novel compounds. To this end, we propose to use an integrated approach of cell biology and in vivo animal models. We will identify the palmitoylating enzyme of APP and ask whether ACAT inhibitors affect its expression or localization. We will also study how palmitoylated APP gives rise to A? and how ACAT inhibitors specifically reduce this process. Lastly, we will test the therapeutic potential of our novel ACAT inhibitors in animal models of AD and how ACAT inhibition regulates palmitoylated APP in vivo. Collectively, the goal of these experiments is to provide the necessary mechanistic and in vivo data for further development of ACAT inhibitors as a therapeutic strategy for AD and perhaps cardiovascular disease.

DESCRIPTION (provided by applicant): Recent genome-wide association studies (GWAS) and whole genome sequencing (WGS) have been successful in identifying novel Alzheimer's disease (AD)-associated risk genes and their functional variants, respectively. Our recent large-scale WGS efforts (Alzheimer's Genome Project, AGP-WGS; N = 1510 samples) have identified dozens of functional genetic variants that tightly co-segregate with familial AD. AD risk genes and their functional variants carry significant potential for unraveling the pathogenic mechanisms underlying AD, as well as provide new drug targets for the prevention and treatment of AD. The major challenge is to now fully characterize the pathogenic effects of AD-linked functional variants. To date, the field has lacked a single disease model system that fully recapitulates the pathogenic cascade of AD in a human neural system. In preliminary studies, we describe the creation of a novel human stem cell-derived 3D neural cell culture model in which familial AD mutations in the amyloid-? precursor protein (APP) and presenilin 1 (PSEN1) that induce extracellular ?-amyloid accumulation, also leads to neurofibrillary tangles. Thus, using this unique model system, we show for the first time that ?-amyloid deposition is sufficient to induce robust tauopathy, including hyperphosphorylated tau and detergent-resistant, silver-positive neurofibrillary tangles in a human neural cell system. No mouse model has previously achieved this without co-expressing both A?- and tau-related gene mutations. We now propose the following aims to employ our novel 3D human neural cell culture model to comprehensively assess novel AD genes and their functional variants for effects on AD pathogenesis. In Aim 1, we will investigate the impact of AD-risk genes and their functional genetic variants, identified b AGP-WGS, on ?-amyloid and tau pathologies, using our unique human 3D neural cell model system. In Aim 2, we will explore changes in gene expression and proteomic profile induced by excess ?-amyloid deposition in the human 3D neural cell model system. Potential interactions between AD risk genes/functional variants and molecular pathways triggered by excess ?-amyloid will also be explored. The overarching goal of the proposed studies is to construct a framework for systematically identifying and characterizing GWAS/WGS AD risk genes and their functional variants using our novel human 3D neural cell culture technology. Our studies should not only enhance our understanding of the etiology and pathology of AD, but also facilitate the discovery of novel AD drug targets for the treatment and prevention of AD.

The amyloid precursor protein (APP) undergoes sequential proteolysis by ?- and ?- secretases to produce amyloid ? (A?) in Alzheimer's disease (AD). Currently, clinical trials are underway targeting A? with ?- or ?-secretase inhibitors in mild or prodromal AD patients. Alternative A?-lowering agents are also being actively pursued. Along these lines, we previously demonstrated that Acyl-coenzyme A: cholesterol acyltransferase (ACAT) inhibitors, e.g. CP-113,818 and CI-1011, which prevent conversion of cholesterol into cholesteryl-esters, reduce secreted A? levels by up to 92%, and improve AD-like pathology in hAPP transgenic mice. ACAT-inhibitors have not been clinically developed for AD largely because the molecular mechanism regarding effects on A? generation remains unclear. We have demonstrated, for the first time, that approximately 10% of APP is post-translationally modified by palmitic acid in vitro and in vivo. Palmitoylated APP (palAPP) is enriched in cholesterol-rich lipid rafts where it appears to serve as a preferred substrate for ?-secretase (BACE1) versus total APP. ACAT-inhibition decreased lipid raft palAPP levels by up to 76%. We have also reported that ~90% of palAPP forms cis- dimers undergoing preferred BACE1 cleavage in detergent resistant membranes (DRMs). Thus, palAPP and/or cis-dimerized palAPP in lipid rafts are potentially useful drug targets for AD. Recently, we reported that palAPP is also enriched in raft-like Mitochondria- associated ER Membranes (MAMs) in vitro and in vivo, together with BACE1, ?-secretase, and ACAT. Interestingly, MAM function and ER?mitochondrial communication are increased significantly in fibroblasts from both familial and sporadic AD patients. Overall, our preliminary studies indicate that palAPP is synthesized in neuronal cells including neuronal processes and is stabilized in MAMs. Thus, we propose the following hypothesis: palAPP that is stabilized in MAM's undergoes BACE1 cleavage in neuronal cells and processes. We will explore the effects of novel ACAT inhibitors on palAPP trafficking and processing in MAM-associated/stabilized palAPP, including the use of live-cell imaging and cell surface biotinylation assays in primary neurons and our 3D human stem cell-derived neural culture models. The overarching goal of this proposal is to generate the necessary mechanistic and in vivo data to further the development of novel therapeutic strategies for AD by targeting ACAT and MAM-associated palAPP.

SUMMARY The majority of early-onset familial Alzheimer?s disease (AD) mutations are found in presenilin 1, PSEN1, the active component of ?-secretase. Most mutations increase levels of longer forms of ?-amyloid (A?) leading to enhanced deposition of ?-amyloid plaque and AD pathology. We will investigate a novel series of ?- secretase modulators (GSM) for their ability to pharmacologically reverse the pathological effects of these mutations. GSMs specifically modulate the cleavage activity of ?-secretase on APP processing to preferentially lower A?42 while increasing A?38 and A?37 levels, with minimal effects on A?40. Meanwhile, GSMs do not inhibit cleavage of other ?-secretase substrates, e.g. Notch. Particularly, we will study the effects of soluble GSM (SGSM), developed in our laboratory in collaboration with Dr. Steven Wagner (UCSD), on AD-related pathological events. We first reported the aryl 2-aminothiazole class of parent GSMs (AGSMs) (Kounnas et al., 2010). However, the AGSMs displayed poor aqueous solubility making them undesirable for clinical development. Subsequently, we developed water-soluble aryl 2-aminothiazole class, or 1st generation SGSMs (Wagner et al., 2014), with improved clinical potential. Recently, we structurally enhanced the aryl 2- aminothiazole SGSMs and developed the novel pyridazine class, of 2nd generation SGSMs, which display higher potency in inhibiting A?42 as compared to early SGSMs. Our original and current lead compounds in the pyridazine class are respectively SGSM-15606 and SGSM-776890, both displaying IC50 values of A?42 < 10 nM in cells. While SGSM-776890 is being prepared for a Phase-1A trial (single dose ascending), we have developed over a hundred analogs that have yet to be further tested as backups for clinical trials. In collaboration with other PPG projects, we will focus on the mechanism of actions of SGSM-15606, SGSM- 776890 and other analogs on AD pathology. Furthermore, we will focus on the roles of PSEN1 on adult hippocampal neurogenesis, a process in which human brain stem cells generate new neurons and glial cells throughout adulthood, which is impaired in AD, thereby contributing to cognitive deterioration in AD. Additionally, our GSMs have been provided to Project 1, which will focus on mechanisms by which PSEN1 mutations generate A? peptides ranging from 37-49 amino acids, Project 2, which will focus on the effects PSEN1 mutations on PSEN conformation and A?-dependent and A?-independent mechanism. Collectively, the results of the proposed studies not only will enhance our understanding of the biology of ?-secretase and the pathogenesis of AD, but also may identify a potential compound that can be a therapeutic for AD.

There are no effective drugs to prevent Alzheimer?s disease (AD). We seek to prevent onset or progression of AD by discovering and enhancing the activity of naturally occurring pathways that prevent its occurrence. This natural resilience is significant because it is the only known manner in which Alzheimer?s appears to be controlled. Here, we exploit the fact that a proportion of the aging population appear to remain cognitively intact while controlling or compensating AD related Tau pathology and enjoy a relative natural resistance to cognitive impairment or diagnosis of AD. Using whole genome sequencing (WGS) and transcriptome analysis of a naturally AD-resilient population, we will identify novel drug-sensitive resilience associated (RA) pathways in AD. We will implement a novel, validated technology, the Pathway Drug Network (PDN), constructed from human gene expression data enriched in drug?pathway?gene clusters, to identify drugs that enhance RA pathways. First round screening of the PDN-predicted single or combinations of leads will be tested in our innovative 3D human neural cell culture models of AD, which recapitulate various pathogenic stages of AD including Ab deposition (Ab plaque), Ab-driven tau pathology (neurofibrillary tangles (NFTs), and neuroinflammation and neurodegeneration. Validated leads will then be scored in transgenic AD mouse models for reduction of synaptic loss and cognitive integrity. The approach will establish the basis for a therapeutic intervention that can prevent or reduce cognitive decline related to AD. Intellectual merit: This project will significantly advance the understanding of neuroprotection in aging adult human brains while providing novel insights into the relationships between control of AD related pathology and loss of cognition. Broader impact: AD increasingly affects the aging population and there is no effective intervention. Reduction of its incidence will be of major significance. If successful, the project will allow development of clinical application of novel drugs or repurposed FDA approved drugs while creating a powerful new paradigm for developing successful AD drug combinations. Aim1: Using network analytical techniques, we will generate a molecular systems definition of RA pathways using pathways, genes and network modules from whole genome sequencing data and literature, and post mortem brain transcriptomes that show resilient high or low plaque/tangle, low AD symptoms, but high cognitive scores. Aim 2: Compare RA pathways within PDN to predict drug/pathway combinations that confer resilience. A series of drug-repurposing screens will optimise lists of ranked drugs/combinations and pathway activity. Selected combinations will be validated in multiple 3D human neural cell culture models of AD that mimic various pathogenic stages of AD for their impact on AD pathogenic markers. Aim 3: Validate using proxies of cognition in an AD transgenic APP mouse model. Score for the ability to confer resilience and neuroprotection in AD transgenic mouse models for either Ab deposition, synaptic/cognitive deficits and/or neuroinflammation.

Title: The impact of AD-associated genetic variants in 3D human mixed neural-glial models of AD Project summary A growing number of Alzheimer?s disease (AD)-associated genes are associated with innate immunity and neuroinflammatory pathways. Network-based integrative analyses of AD-related genes have shown that microglial gene networks are strongly associated with AD neuropathology (1,5). We have shown that the protective AD-associated CD33 variant, rs3865444, leads to reduced CD33 expression and lower levels of A?42)(1). Conversely, microglial TREM2 variants, which increase AD risk, reduce microglial clearance of Ab. In addition to AD-linked functional variants in CD33 and TREM2, our AD whole genome sequencing (WGS) and whole exome sequencing (WES) datasets from AD families and case-controls, have revealed functional variants in AD-associated microglial genes linked to innate immunity and neuroinflammation, including CD33, TREM2, MS4A cluster, ABCA7, ABI3, PLGC2, CR1, and others. To test the impact of microglial genetic variants on AD pathogenesis on human genetic background, we developed a novel 3D human neuron- astrocyte-microglia tri-culture AD model using a unique 3D microfluidic system. We demonstrated that human microglial cells are recruited towards 3D AD (Ab-producing) neuron-astrocyte cultures via microglia-specific migration channels, in a chemokine-dependent manner, leading to neuroinflammation and neurodegeneration. Here, we propose to use our extensive collection of AD WGS and WES datasets, together with our 3D human tri-culture AD model, to evaluate the pathogenic effects of functional variants in AD-associated innate immune genes linked to neuroinflammation. In Aim 1, we will identify functional genomic variants and enriched gene networks that are linked to innate immunity and neuroinflammation. We will then examine the pathogenic roles of microglial AD risk or protective genes and their functional variants in 3D human mixed neural-astrocyte- microglial models of AD (Aim 2), explore AD pathogenic pathways that are linked to microglial AD risk genes and their functional variants using integrated multi-omics approaches, and validate selectively blocking these pathway (Aim 3). The overarching goals of this proposal are to comprehensively assess the pathogenic effects of functional variants in innate immune AD-risk genes on AD pathogenesis and explore underlying molecular networks, which will provide novel therapeutic targets for AD patients.