Samuel E. Gandy, MD PhD - US grants

Alzheimer's disease is characterized by clinical dementia in association with pathologic alterations in brain proteins. Structural lesions include extracellular beta/A4-amyloid deposits. Beta/A4-amyloid is derived through proteolysis of a large, transmembrane precursor, the beta/A4-amyloid precursor protein (APP). Mutations in the coding sequence of APP are associated with familial cerebral amyloidoses, pointing to the importance of APP in the pathogenesis of amyloidosis. The association of different APP-mutation genotypes with the cerebral amyloidosis phenotype strongly suggests that alternative amyloidogenic proteolysis may be a final common pathway in both familial and sporadic cerebral amyloidotic diseases, including Alzheimer's disease. A standard pathway for proteolysis of about 30% of APP molecules (in PC-12 cells) cleaves within the beta/A44- amyloid domain, precluding amyloid formation. The generation of beta/A4-amyloid must therefore occur via an alternative proteolytic pathway. Evidence for the existence of alternative pathways has emerged from several laboratories, in studies of human cerebral vessels, brain, and cells in continuous culture. Cell culture systems which generate microheterogeneous proteolysis of APP include the rat PC-12 line (under conditions of supraphysiological protein phosphorylation), the monkey fibroblast (following overexpression of human APP in recombinant vaccinia virus), the human APP-transfected human 293 cell, and the recombinant human APP-baculovirus infected Sf9 cell. The Sf9 system is particularly attractive because of the extraordinarily high-level expression of recombinant protein, providing convenience for purification and sequencing of species of interest. Sf9 cells faithfully recapitulate many of the biological properties of mammalian cells. When human APP is expressed in Sf9 cells, a portion of APP molecules is cleaved in a position exactly identical to the major cleavage site for proteolyzing APP in human cells, thus providing validity to the use of the Sf9 cell as a model system for APP proteolysis. At high multiplicities-of-infection, in addition to the cleavage of APP at this major conserved site (which generates a 14-15 Kda carboxyl-terminal fragment), Sf9 cells produce a discrete and limited number of other APP carboxyl-terminal fragments, including species of 16-, 17- and 25-Kda. By antigenic analysis, the 17-Kda species has been demonstrated to incorporate amino-terminal epitopes of the beta/A4-amyloid domain and is thus a putative amyloidogenic fragment. While there is mounting immunochemical evidence for such putative amyloidogenic fragments, direct protein sequencing of such a species has not been achieved, and is the primary goal of this proposal, using the baculoviral/Sf9 expression system. In addition, putative amyloidogenic mutant APP molecules (from familial cerebral amyloidoses) will be overexpressed in baculoviruses, and their proteolytic fragments characterized, purified and sequenced. The definitive identification of amyloidogenic pathways for APP proteolysis is crucial to the successful dissection of amyloidogenesis and to the design of strategies for in vitro models of amyloidogenesis.

DESCRIPTION: (Applicant's abstract) Gamma-amino butyric acid (GABA), the major inhibitory neurotransmitter in the mammalian CNS, hyperpolarizes neuronal membranes by opening a C1 channel intrinsic to the GABA A receptor. The proposed studies investigate the anatomical features and expression of selected GABA A receptor subunits (i.e., alpha1, alpha2, alpha3, alpha4, alpha5, beta1-3, gama2) in human hippocampus of non-pathologic mature and aged individuals (30-90 years of age) and those with Alzheimer's disease (AD) pathology. Underlying these studies are investigations of the P.I. and co-investigators demonstrating (I) subunit specific alteration (i.e., alpha1) in the CA1 field and dentate gyrus of aged rats; (ii) GABA A receptor subunit protein and mRNA alterations (i.e., alpha1, beta2, beta3) in the hippocampal formation of aged brains with AD pathology; (iii) time-dependent alterations of GABA A beta2/3 immunoreactivity in the dentate gyrus following perforant pathway lesions. In addition, pharmacological studies have demonstrated altered drug sensitivities, for example, to benzodiazepines, in the elderly. To date, it is not known the extent to which altered drug responses in the elderly may be attributed to emotional or physical disease, over- or undernutrition, use or abuse of other medications, or alterations in the molecular composition of the GABA A receptor. Throughout these studies we will employ immunohistochemical, in situ hybridization, in situ autoradiogaphic, and biochemical techniques in order to study the regional and laminar pattern of GABA A receptor subunit protein and mRNA expression in the hippocampal formation of aging individuals (SPECIFIC AIM 1) and those with varying extent of AD pathology (SPECIFIC AIM 2). It is our hypothesis that selected GABA A receptor subunits will display differential levels of expression and binding within the various regions of the hippocampus. Moreover, brain tissue obtained from elderly patients will not only show altered levels of expression of specific GABA A subunits, but will have a different subunit composition of the GABA A receptor complex compared to controls. In AD, we hypothesize that these receptor subunits will also display altered levels of expression and binding within selected subregions of the hippocampus. In addition, many of these changes will occur during the early phases of the disease (i.e., plastic/compensatory phase) and be unique from those observed during the end stages of the disease (i.e., neurodegenerative phase). A unique aspect of this study is the study of both aging and AD subjects. Notably, a comparison of these two populations of subjects will provide us with the opportunity of differentiating whether alterations in the anatomy of specific GABA A receptor subunits are associated with normal aging or represent a neuropathologic process.

DESCRIPTION (Applicant's Abstract): The proposed studies will address the issue of selective vulnerability of neurons in Alzheimer's disease. Underlying these studies is the hypothesis that excitotoxicity, particularly that which is mediated via stimulation of ionotropic glutamate receptors, contributes to the neuro-degeneration of Alzheimer's disease. Moreover, we predict that in Alzheimer's disease the perturbation of specific glutamate receptor subunits, particularly those involved in the gating of calcium, may contribute significantly to the viability of the cell. The current investigation focuses on the N-methyl- D-aspartate (NMDA) receptor. Notably, previous work of ours has focused on the distribution and expression of specific non-NMDA (i.e. AMPA) receptor subunits in the hippocampus of patients with Alzheimer's disease pathology. Collectively, these investigations attempt to provide a comprehensive understanding of the anatomy of the ionotropic glutamate receptor in the hippocampus of normal subjects and in subjects with Alzheimer's disease. In this application, we propose a series of highly correlated immunohistochemical (Specific Aims 1&2), biochemical (Specific Aim 3) and in situ hybridization studies investigating the distribution and level of expression of specific NMDA receptor subunits (e.g., NMDAR1, NR2A, NR2B, & NR2D). Studies will focus on the human hippocampus, in part, because this region is known to be affected very early within the progression of the disease. Subjects representing a broad range of neuropathologic severity (i.e. Braak stages I-VI) will be studied thus providing us with the opportunity of examining alterations in glutamate receptor expression throughout various stages of the disease. Moreover, the use of specific antibodies identifying early events in the evolution of neurofibrillary pathology provides an additional opportunity of correlating alterations in NMDA receptor subunit expression with initiating events of neurodegeneration. An understanding of the anatomical organization of NMDA and AMPA receptors and the mechanism underlying glutamate-mediated excitotoxicity is important before appropriate drugs aimed at halting Alzheimer's disease-associated neuronal degeneration can be developed.

The LDL receptor-related protein (LRP) is a signaling receptor for LDL, linking the nature and levels of extracellular lipoprotein particles to intracellular signaling pathways, one of which is the Rho-dependent protein kinase (ROCK) pathway. We have recently discovered that ROCK inhibits activated nonamyloidogenic alpha-secretase processing ("ectodomain shedding") of the Alzheimer amyloid precursor protein (APP). We hypothesize that each component of this lipoprotein signaling pathway (i.e., apoE isoforms, cholesterol, LRP, the receptor-associated protein [abbreviated RAP], ROCK) has a distinct impact on alpha-secretase activity and ectodomain shedding. Aim 1 is to dissect the individual contribution of each component to the modulation of APP shedding. Based on the observation that alpha-secretase overexpression can abolish Abeta pathology in transgenic mice, we hypothesize that dissection of this regulatory pathway will reveal novel sites of pathogenesis and novel targets for anti-amyloid drugs. At the level of the effector molecules, ROCK signals act via the alpha-secretase proteolytic pathway, but the identities of the phospho-state-specific ROCK substrate(s) that regulate shedding are unknown. We and others have shown that at least one important phosphoprotein target resides between the trans-Golgi network (TGN) and the plasma membrane, and controls either (i) budding of APP-bearing vesicles;(ii) fusion of APP-bearing vesicles with alpha-secretase-bearing vesicles;or (iii) intra-plasma-membrane activation of alpha-secretase. Aim 2 is to study the murine counterparts of selected SEC proteins (known as "Munc" proteins) as candidate phospho-state-specific modulators of ectodomain shedding (PMES). "Munc" proteins are the murine homologues of unc proteins (which, in turn, are the nematode homlogues of yeast SEC mutants). We hypothesize that discovery of one or more PMES molecules will reveal novel sites of pathogenesis and targets for anti-amyloid interventions.

DESCRIPTION (provided by applicant): Alzheimer Disease (AD) is a progressive brain disease known generally as senile dementia. More than 4.5 million Americans have been diagnosed with AD, and this number is expected to triple in the next 40-50 years. There is currently no cure for AD. The treatments available today for AD are symptomatic and do not interfere with inexorable progression of the underlying disease process. There is a desperate need for interventions that might improve symptoms and modify the disease progression. The two defining neuropatholic features of AD are abnormal aggregation and deposition of A2 peptides and tau in the brain as, respectively, extracellular neuritic plaques (NP) and intracellular neurofrillary tangles (NFT). In the brain, monomeric Ab peptides and tau proteins are aggregated to form high molecular weight, soluble multimeric neurotoxic Ab and tau species. Continual progressive aggregations of multimeric Ab and tau species result in deposition of Ab and tau into, respectively, NP and NFT. Recent experimental evidence indicates that it is the accumulation of soluble high- molecular weight (HMW) oligomeric A2 and tau species in the brain, rather than deposition of NP and NFT per se, may be specifically related to cognitive dysfunction in AD. We recently demonstrated that a select grape- seed polyphenol extract (GSPE), namely, MegaNatural-Az(R) GSPE to potently inhibit the aggregation of both Ab peptides and tau proteins. Thus, MegaNatural-Az(R) GSPE may benefit AD by mitigating both Ab- and tau- mediated neurotoxic responses. Based on this and evidence demonstrated the high tolerability and safety of long-term application of MegaNatural-Az(R) in both laboratory animals and in human. The proposed study represents collaboration between basic science research laboratories and the Alzheimer's Disease Clinical Core at the Mount Sinai School of Medicine to explore the development of MegaNatural-Az(R) for treating AD. In particular, our proposed study will establish safety and pharmacokinetics of Meganatural-Az(R) GSPE in AD subjects. As secondary measures, we will also evaluate clinical and biomarker indexes of therapeutic efficacy. The proposed study will provide the essential human data to guide the design of future studies to test the efficacy of GSPE in mitigating cognitive deterioration in AD patients. PUBLIC HEALTH RELEVANCE: This is an application to study the feasibility of developing a select grape seed extract, referred to as MegaNatural- Az (R) for treating Alzheimer's Disease. Meganatural -AZ (R) has been shown to benefit AD phenotypes in experimental in vitro and animal AD model systems by potently interfering with aggregation of beta-amyloid peptides and tau proteins into neurotoxic aggregates. The study will access safety, pharmacokinetics of MegaNatural- Az (R) in AD patients and explore cognitive and biological indices of therapeutic efficacy in a 10 week multiple dose escalation study.

The hypothesis driving the current proposal is that multiple y-secretase substrates are misprocessed in mutant presenilin 1 familial Alzheimer's disease (PS1 FAD) and that a multisubstrate misprocessing "signature" is also similarly associated with sporadic AD. To this end, we have recently demonstrated that "APP p3-like" alcadein peptides (p3-Alcs) are generated by typical a- and y-secretases, and p3-Alc C termini are modulated by FAD mutant PS1. In neurons. Ale proteins form complexes with X11L molecules, which, in turn, complex with APP. For this reason, we chose Ale proteins as representative y-secretase substrates that might undergo misprocessing in AD in parallel with APP. Plotting minor/major p3-Alc variant ratios (analogous to AB42/40 ratios) against the corresponding AB42/40 ratios in media conditioned by a panel of FAD-mutant PSI-expressing cells reveals a highly linear covariant relationship between p3-Alc ratios and AB42/40 ratios. We interpret the disease-related change in regression slope as a signature for y secretase dysfunction. p3-Alcs were also detected in human cerebrospinal fluid (CSF), and, again, a covariant signature was associated with the AD phenotype. The association of the sporadic AD phenotype with a distinct p3-Alc:Ap covariant ratio signature provides evidence that y-secretase dysfunction may contribute to the pathogenesis of sporadic AD. Thus, our Specific Aims are: 1) To characterize Ale metabolites in regions of the cerebral cortex of elderly non-demented humans and patients with differential degrees of dementia;2) To characterize the distribution of Ales in the cerebral cortex of elderly nondemented humans and patients with differential degrees of dementia;and 3) To characterize the cortical and synaptic distribution of Ales in relevant mouse models of AD and analyze their relationship with the progression of neuronal alterations. The information from human brain and from the brains of mouse models should guide us in future efforts to understand the AB:Alc metabolic covariance as well as the possible implications of AB:Alc metabolic covariance for the pathogenesis and diagnosis of sporadic AD.

DESCRIPTION (provided by applicant): Genetic approaches have provided major insights into the molecular pathogenesis of Alzheimer's disease (AD). However, only about 3% of all of AD is due to genetic mutations in either amyloid precursor protein (APP), or presenilin 1 or 2 (PS1, PS2). We propose to generate a human in vitro model using induced pluripotent stem (iPS) cells, in which the genetic and developmental aspects of familial and sporadic AD can be studied more accurately and therapeutic targets can be identified for subsequent drug discovery. The following Specific Aims are proposed: Specific Aim 1: To generate iPS cells and neurons from skin fibroblasts from subjects with familial and sporadic AD. We have already succeeded in generating differentiated neurons from fibroblasts from subjects with PS1 mutations. We have demonstrated that differentiation of these neurons leads to their acquisition of an obvious standard molecular phenotype; i.e., a shift in the A¿ 42/40 ratio). The initial essential standardization of these neurons will include, for each PS1 mutation, the exploration of intra-individual and inter-individual variability in the A¿ 42/40 phenotype within patients, affectd and unaffected family members, and across different families that carry either the identical mutation or across different PS1 mutations. A longer-term goal will be the generation of glia and mixed cell cultures. Specific Aim 2: To perform molecular, biochemical and functional characterization of AD iPS cell lines. We have defined a culture system for AD iPS cell-derived neurons that includes the essential A¿ 42/40 phenotype. We will now proceed to establish the content of AD-related molecules in these iPS cells while seeking to establish the cell biological basis for the A¿ 42/40 phenotype. This will include an assessment of the autophagic pathway. Specific Aim 3: Identification of transcriptional profiles of familial and sporadic AD iPS cells. Or primary goal in this aim is to establish a baseline molecular characterization of forebrain neural cells derived from the panel of iPS cell lines specified above. Informatic analysis of these profiles will be performed in order to identify possible AD-related networks, as recently defined by Geschwind and colleagues. We will study how in vitro cellular and molecular phenotypes in telencephalic neural cells derived from patient iPS cells vary and are similar across individuals and mutations related to either familial or sporadic AD. PUBLIC HEALTH RELEVANCE: Genetic approaches have provided major insights into the molecular pathogenesis of Alzheimer's disease (AD). However, only about 3% of all of AD is due to genetic mutations in either amyloid precursor protein (APP), or presenilin 1 or 2 (PS1, PS2). We propose to generate a human in vitro model using induced pluripotent stem (iPS) cells, in which the genetic and developmental aspects of familial and sporadic AD can be studied more accurately and therapeutic targets can be identified for subsequent drug discovery.

DESCRIPTION (provided by applicant): Type 2 diabetes mellitus (T2D or T2DM) increases the risk for Alzheimer's disease (AD), and SORCS1 is genetically linked to both T2D and AD. We have undertaken a study of the possible role(s) for SorCS1 in metabolism of the Alzheimer's amyloid-? (A?)?precursor protein (APP), in order to define the molecular mechanisms underlying this coordinate genetic linkage to both diseases. Overexpression of SorCS1cb-myc in cultured cells caused a reduction (p=0.002) in?A? generation (Lane et al., 2010). Endogenous murine A?40 and A?42 levels were increased (A?40, p=0.044; A?42, p=0.007) in the brains of female Sorcs1 hypomorphic mice, possibly paralleling the sexual dimorphism that is characteristic of the genetic associations of SORCS1 with AD and DM. Since SorL1, another AD-linked Vps10-domain protein, directly interacts with Vps35 to modulate APP metabolism, we investigated the possibility that SorCS1cb-myc might interact with APP, SorL1, and/or Vps35. We readily recovered SorCS1:APP, SorCS1:SorL1, and SorCS1:Vps35 complexes from nontransgenic mouse brain. Notably, total Vps35 protein levels were decreased by 49% (p=0.009) and total SorL1 protein levels were decreased by 29% (p=0.003) in the brains of female Sorcs1-/- mice. We hypothesize that dysfunction of SorCS1 may contribute to both the APP/A??disturbance underlying AD and the insulin/glucose metabolism disturbance underlying DM. In order to test this hypothesis further, we propose the following specific aims: Specific Aim 1. To evaluate the importance of SorCS1 protein interaction motifs and SorCS1/SorL1/APP complex formation on APP metabolism by: (a) Characterizing APP metabolism in cultured cells overexpressing SorCS1; (b) Testing the effects of mutations of protein-protein interacting motifs in the cytoplasmic and ectodomains of SorCS1 on both the formation of tripartite SorCS1/SorL1/APP complexes and APP metabolism; (c) Testing the effect of a putative pathogenic SorCS1 polymorphism on both the formation of tripartite SorCS1/SorL1/APP complexes and APP metabolism; (d) Confirming the importance of functional domains identified in Aim 1aii and 1aiii by viral gene transfer into primary cultures. Specific Aim 2. To employ Sorcs1 hypomorphic and plaque forming human Swedish APP/PS bigenic mice crossed with Sorcs1 hypomorphic mice for characterization of: (i) endogenous APP metabolism; (ii) hippocampal morphometry, dendritic arborization, and spine structure; (c) learning behavior. Aging (3 mo, 6 mo, 12 mo) effects will also be studied. Specific Aim 3. To perform standard glucose and insulin tolerance tests and metabolic profile phenotyping of Sorcs1-/- mice and plaque-forming human Swedish APP/PS co-transgenic mice crossed with Sorcs1 -/- mice. PUBLIC HEALTH RELEVANCE: Type 2 diabetes mellitus (T2DM) increases the risk for Alzheimer's disease (AD). Both diseases are highly complex, and several mechanisms have been proposed for this association, including hypercholesterolemia, vasculopathic factors, and insulin resistance, among others. We sought and identified a genetic factor, SORCS1, that is an excellent candidate for modulation of a coordinated risk for both diseases, and we now propose to work out mechanistic details for how SorCS1 controls risk for AD and T2DM.

DESCRIPTION (provided by applicant): Alzheimer's disease (AD) affects half of the US population over the age of 85 and causes destruction of select networks and cell groups within the brain. AD manifests initially as mild cognitive decline, but gets progressively worse and is always fatal. Despite significant progress identifying susceptibility loci for AD in genome-wide association and whole exome sequencing studies, to date, a predictive risk score for AD that achieves clinical utility on an individual basis given DNA variation information alone has been elusive. This proposal aims to develop a multiscale-network approach to elucidating the complexity of AD. Multiscale network models causally linked to AD will be developed based on existing AD-related large scale molecular data and the high-impact, high-resolution complementary datasets generated through this application. Using brain slice cultures, iPS-cell-derived mixed cultures of human neuronal, oligodendroglial, and astrocytic cell systems, and fly models of AD, we seek to reconstitute the AD-related networks discovered in the multiscale analysis in these living systems and then employ high-throughput molecular and cellular screening assays to not only validate the actions of individual genes on molecular and cellular AD-associated processes, but also validate the molecular networks we implicated in the disease. Our initial multiscale studies have implicated the microglial protein TYROBP as one key driver of AD pathogenesis, a hit we have partially validated, but that we will further validae along with other hits using iPSC-derived mixed cultures of different brain cell types, murine brain slices and AD fly models. We will analyze the potential ability for network-derived hits like TYROBP to modulate standard AD pathology involving A? and tau as well as its ability to shift networks in those same systems in such a way as to reflect the behavior of networks discovered in the multi-scale analysis. Importantly, the model building and validation will be iterated to produce updated/refined models based on validation results that, in turn, will be mined to generate updated lists of prioritized targets for validation. In this way, through the course of th grant, as new knowledge accumulates externally and as we generate increased amounts of data including validation data, our models will take into account the most up to date information to produce the most predictive models of AD. As a service to the AD research community, we will provide dramatically improved general access to large-scale, multidimensional datasets, together with systems level analyses of these datasets.

? DESCRIPTION (provided by applicant): The purpose of the R34 planning grant is to support the effort, conferences, site visits, and other fact-finding processes involved in designing the translational center, Systems Medicine Drug Discovery for Alzheimer's disease. This Center will include Core Facilities charged with providing key technology services and Pathway/Drug-Based Projects that will focus on specific molecular mechanisms and/or predicted pharmacological interventions. Project leaders will be responsible for moving each project (target pathway, drug) through the relevant cores with maximum efficiency. Pathways and projects will undergo continual reassessment of priorities. The goals of the planning period are as follows: (Specific Aim 1) To assemble a team of investigators with appropriate expertise and to develop the most efficient strategy for their productive communication and collaboration; (Specific Aim 2) To develop a plan for generation and exploitation of diverse genetic and omic data collected in humans and various animal models to identify translatable pharmacodynamic biomarkers and enable modeling of drug response determinants in distinct patient populations; (Specific Aim 3) To develop a plan for establishment of multi-scale computational models of pharmacological mechanism, that bridges the divide between cell-level biochemical models and organism-level PK/PD and neuroimaging models; (Specific Aim 4) To plan the most efficient strategy for target validation that will involve conducting quantitative analysis of the effects of small molecules and biologics on therapeutic targets across multiple scales of biological complexity; (Specific Aim 5) To plan the most effective strategy for investigation of the molecular and physiological origins of variability in drug response at the single-cell, organ, and patient level that arises from differences at the level of the proteome, genome and environment; (Specific Aim 6) To develop systems approaches for comprehensive and systematic failure analysis during preclinical and clinical drug development; (Specific Aim 7) To develop the most efficient strategy for rapid and broad sharing of data, analytical and research tools, and models prior to publication; (Specific Aim 8) To ensure open source data enablement and to develop strategies for removing legal/IP barriers to sharing data, biological samples and research tools; (Specific Aim 9) To develop curricula, workshops and seminars for training in systems biology, systems pharmacology, pharmacometrics, PK/PD modeling, omics technologies, translational bioinformatics and other quantitative science areas; (Specific Aim 10) To provide short term training for junior staff and short term sabbaticals for senior team members to expand existing expertise or develop new expertise essential for achieving the programmatic goals of the translational center(s) initiative. Within the planning year, the R34 investigators predict that these Aims will be achieved and that the foundation will have been laid for writing the center proposal.

DESCRIPTION (provided by applicant): Alzheimer's disease (AD) affects half of the US population over the age of 85 and causes destruction of select networks and cell groups within the brain. AD manifests initially as mild cognitive decline, but gets progressively worse and is always fatal. Despite significant progress identifying susceptibility loci for AD in genome-wide association and whole exome sequencing studies, to date, a predictive risk score for AD that achieves clinical utility on an individual basis given DNA variation information alone has been elusive. This proposal aims to develop a multiscale-network approach to elucidating the complexity of AD. Multiscale network models causally linked to AD will be developed based on existing AD-related large scale molecular data and the high-impact, high-resolution complementary datasets generated through this application. Using brain slice cultures, iPS-cell-derived mixed cultures of human neuronal, oligodendroglial, and astrocytic cell systems, and fly models of AD, we seek to reconstitute the AD-related networks discovered in the multiscale analysis in these living systems and then employ high-throughput molecular and cellular screening assays to not only validate the actions of individual genes on molecular and cellular AD-associated processes, but also validate the molecular networks we implicated in the disease. Our initial multiscale studies have implicated the microglial protein TYROBP as one key driver of AD pathogenesis, a hit we have partially validated, but that we will further validae along with other hits using iPSC-derived mixed cultures of different brain cell types, murine brain slices and AD fly models. We will analyze the potential ability for network-derived hits like TYROBP to modulate standard AD pathology involving A¿ and tau as well as its ability to shift networks in those same systems in such a way as to reflect the behavior of networks discovered in the multi-scale analysis. Importantly, the model building and validation will be iterated to produce updated/refined models based on validation results that, in turn, will be mined to generate updated lists of prioritized targets for validation. In this way, through the course of th grant, as new knowledge accumulates externally and as we generate increased amounts of data including validation data, our models will take into account the most up to date information to produce the most predictive models of AD. As a service to the AD research community, we will provide dramatically improved general access to large-scale, multidimensional datasets, together with systems level analyses of these datasets.

Previous studies associate various microbes with Alzheimer's disease (AD), but the question of whether microbe-related antigens represent a causal component of AD or are an opportunistic passenger of neurodegeneration is difficult to resolve. We propose to expand on our preliminary work mapping biological networks underlying AD-associated phenotypes using independent human post-mortem data sets. First, we apply a computational approach to build region-specific biological networks from `preclinical AD' samples measured from individuals that meet neuropathological criteria for AD, but who were cognitively intact at the time of death. Specifically, we construct regulatory networks from entorhinal cortex (EC) and hippocampal (HIP) tissue and compared the structure of preclinical AD and control networks. Network analysis identified several sub-networks and key drivers perturbed in preclinical AD, including enrichments relevant to viral susceptibility loci among network drivers, including roles for C2H2 zinc fingers, G4-quadruplex activity, and miR-155 in regulating diverse viral factors. Network analysis of a second independent data set characterizing individuals with `clinical AD' identified more direct evidence for network mechanisms of viral activity in AD. The availability of DNA and RNA sequencing allowed us to evaluate the direct presence of viral sequences and quantify association with AD. We observe increased abundance of human herpesvirus 6A (HHV-6A) and human herpesvirus 7 (HHV-7) across multiple regions from subjects with AD compared with controls. We replicated these findings in two additional, independent and geographically dispersed AD cohorts. We built preliminary network models of host- virus regulatory interactions and virus-virus interactions to correlate viral species with the severity of cognitive impairment and neuropathology. We find evidence that the virus-host landscape is shaped by both competitive and synergistic interactions between multiple viral species, with impacts on amyloid precursor protein (APP) processing, cytoskeletal organization, protein synthesis and innate immune response. We identified host DNA variants that associate with viral abundance (vQTL), and that vQTLs associate with increased AD risk, clinical dementia severity, and neuropathology severity, indicating a shared genetic basis linking risk for AD, severity of AD neuropathology, and abundance of specific viral species. We propose to perform computational and experimental work to further elucidate the specific network mechanisms causal drivers or viral pathogens in AD pathophysiology. We aim i) to map and models the roles of specific viral species in modulating pathogenic molecular, genetic, and clinical networks in preclinical and clinical AD, ii) to evaluate the roles of C2H2 zinc finger proteins and G-quadruplex sequences in mediating molecular pathology of preclinical and clinical AD, and iii) to identify and evaluate specific molecules that mediate viral effects upon the molecular pathology of preclinical and clinical AD.

ABSTRACT Activation of immune pathways and networks constitutes an important emerging component of Alzheimer's disease (AD) pathology that presents new opportunities for therapeutic discovery. The overall goal of the study is to define high-quality immune activation networks for AD and to identify novel therapeutic opportunities against these networks using computational drug repurposing. We propose three distinct but complementary aims that leverage recent insights into the genetics, functional genomics, and network biology of immune factors in AD along with computational drug repurposing. First, we will define microglia-associated networks and use these as input to computational drug repurposing methods to identify novel therapeutic opportunities. We will also focus viral perturbations as an environmental source of immune activation in AD. We will build from our preliminary data that shows the influence of specific viral species in networks constructed from multi- omic measures of post-mortem brain tissues. We will then use these virally-modulated AD networks as input to computational drug repurposing to identify novel therapeutic opportunities. Finally, we will evaluate the temporal dynamics of immune activation in AD. AD is a disease of aging that manifests over a period and we hypothesize that role of immune activation in AD is likely to vary along stages of disease. We will study microglial and multi-omic networks in mice lacking each of the specified microglial genes at ages 4, 8, 12, and 24 months characterize the molecular, cellular, and histological effects of immune network activation in AD animal models across animal age and disease stage. The expected outcome of this study is to identify and evaluate drugs that could be repurposed to modulate immune networks towards non-disease states and to potentially identify the most relevant or optimal states or stages of disease for evaluating therapeutic agents targeting immune activation pathways in AD.

Project Summary Alzheimer's disease (AD) pathology is characterized by the presence of phosphorylated tau in neurofibrillary tangles (NFTs), dystrophic neurites and abundant extracellular ?-amyloid in senile plaques. However, the etiology of AD remains elusive, partly due to the wide spectrum of clinical and neurobiological/neuropathological features in AD patients. Thus, heterogeneity in AD has complicated the task of discovering disease-modifying treatments and developing accurate in vivo indices for diagnosis and clinical prognosis. Different approaches have been proposed for AD subtyping, but they are generally neither suitable for high-dimensional data nor actionable due to the lack of mechanistic insights. Increased knowledge and understanding of different AD subtypes would shed light on recently failed clinical trials and provide for the potential to tailor treatments with specificity to more homogeneous subgroups of patients. By integrating genetic, molecular and neuroimaging data to more precisely define AD subtypes, we may be able to better discriminate between highly overlapping clinical phenotypes. Furthermore, the identification of such subtypes may potentially improve our understanding of its underlying pathomechanisms, prediction of its course, and the development of novel disease-modifying treatments. In this application, we propose to systematically identify and characterize molecular subtypes of AD by developing and employing cutting-edge network biology approaches to multiple existing large-scale genetic, gene expression, proteomic and functional MRI datasets. We will investigate the functional roles of key drivers underlying predicted AD subtypes as well as three candidate key drivers from our current AMP-AD consortia work in control and AD hiPSC-derived neural co-culture systems and then in complex organoids by screening the predicted transcriptional impact of top key drivers in single cell and cell-population-wide analyses. Functional assays in each cell type will be used to build evidence for relevance to AD-subtype phenotypes. Single cell RNA sequencing data will be generated to identify perturbation signatures in selected drivers that will then be mapped to subtype specific networks to build comprehensive signaling maps for each driver. The top three most promising drivers of AD subtypes and the three existing AMP-AD targets will be further validated using a) an independent postmortem cohort, and b) recombinant mice, including amyloidosis, tauopathy and new ?humanized? models.

Project summary Alzheimer's disease (AD) pathology is characterized by the accumulation of neurofibrillary tangles, dystrophic neurites, and abundant extracellular fibrils of amyloid-? peptide. However, the etiology of typical late onset AD remains elusive. Over 20 genes have been associated with late onset AD, and this heterogeneity complicates the task of discovering disease modifying treatments. The parent application proposed to: (i) identify robust molecular subtypes of AD and their characteristic molecular signatures across different layers of Omics data; (ii) characterize molecular subtypes of AD by molecular signatures, multiscale regulatory networks and key drivers; (iii) evaluate genomic and functional impact of key drivers using human iPSC derived neurons and glia; and (iv) validate key drivers of molecular networks underlying AD subtypes. Recently, efforts by the investigators in the parent grant led to the identification of the VGF gene as a key driver of the network predicted to be altered in AD. However, the molecular mechanism by which VGF modulates the network altered in AD is not well understood. It is possible that receptor systems activated by peptides derived from VGF play a crucial role in this process. Support for this comes from our previous studies of another key driver, PREPL, where we found that decreases in PREPL expression leads to decreases in levels of secreted VGF- derived peptides. Also, several VGF-derived peptides have been detected in the cerebro-spinal fluid of AD subjects and many of these peptides exhibit distinct biological activities. This suggests the existence of receptors for the VGF-derived peptides and an important role for them in AD. To date receptors for the majority of these peptides have not been definitively identified. In this supplement we propose to carry out studies to identify neuronal receptors to 18 VGF-derived peptides using the PRESTO-TANGO® assay system that contains 302 G protein-coupled receptors including 135 listed as ?orphan? receptors. Identification of these receptors is a prerequisite to studies investigating the physiological significance of VGF-derived peptides to AD as well as to identifying small molecules targeting these receptors, which could become potential therapeutics for the treatment of AD.