Eric E. Schadt, Ph.D. - US grants

DESCRIPTION (provided by applicant): The normal action of insulin to promote glucose disposal varies over 600% in the population, with those individuals who are least sensitive to insulin having to mount sustained increased levels to maintain optimum serum glucose levels. Some of these individuals develop type 2 diabetes mellitus (DM2), but even those who do not develop DM2 are at increased risk for dyslipidemia, hypertension and atherosclerotic cardiovascular disease. While up to a third of the US population has this insulin resistance (IR) syndrome, our ability to identify individuals with (or at risk) for IR is extremely limited, and there are few treatment options for IR. Insulin sensitivity (IS) is largely determined at the genetic level, with heritability estimated at -50%. To aid in the search for genetic variation that underlies the extreme variance in IS, and the pathophysiology of IR, we are performing genome-wide association study of all individuals in the world phenotyped by insulin clamp measures of IS, through the Genetics of Insulin Sensitivity (GENESIS-2) consortium. Further insights into the cellular and molecular basis of IR could be gained through the study of metabolic and vascular cells from subjects with known insulin sensitivity and genetic architecture. Toward this end, we propose here to develop induced pluripotent cell (iPSC) lines from several hundred individuals of the GENESIS-2 cohort, as a sustainable resource for the production of metabolic and vascular cells in culture, the Genetics of Insulin Sensitivity iPSC (GENESiPS) study. We will develop methods for optimized differentiation of iPSC to the skeletal muscle (or adipose) and vascular endothelial cell lineages. The cellular phenotype of these cells will be investigated through whole genome transcriptome sequencing, as well as cell-based assays of insulin signaling and action. By combining the known human IS phenotype and whole genome variation with the differentiated cellular phenotypes at baseline and in response to insulin stimulation, we will undertake a systems biology approach to defining the cellular IR phenotype. These analyses will leverage novel network approaches that improve power, and provide added insights into the genes and pathways that may be targeted for the development of next generation insulin sensitizing therapies. (End of Abstract)

DESCRIPTION (provided by applicant): In this collaborative R01, Networks from multidimensional data for schizophrenia and related disorders submitted in response to RFA-MH-12-020, we propose to develop methods for integrating a broad range of genomic, imaging, and clinical data, hosting all data, methods, and results on a novel, flexible and extensible computing platform. Subsequently, these data and methods will be used to establish workflows available to the research community to integrate and mine the data for discovery. As proof-of-concept, multiple datasets for schizophrenia (SCZ) will be used and then extended to additional mental disorders. Specifically, in AIM 1 we will adapt the Synapse platform at Sage Bionetworks to host, QC, normalize, and transform data in an analysis ready format. Synapse will also enable computation, storage, sharing, and integration of SCZ specific data with pre-existing public data. The Sage platform will be hosted by the data center in the Institute of Genomics and Multiscale Biology at the Mount Sinai School of Medicine consisting of a data warehouse (organized file systems and databases), a web service tier and applications tier adapted to facilitate network reconstruction and more generally model building with SCZ data. In AIM 2, we will develop a pipeline of analytic methods that include new and existing tools for the primary processing of multiple types of data. Using direct experimental findings we will generate primary analysis datasets (e.g., expression QTLs, imaging QTLs, GWAS with SNP/CNV genotypes, RNASeq signatures, and DNA methylation and RNAseq associations), construct interaction networks with population-based expression and imaging datasets (e.g. gene expression, functional MRI and structural MRI), transform all data and results into analysis ready formats, and construct a standard set of queries to facilitate SCZ gene discovery. In AIM 3 following platform development, generation of primary analysis datasets, and basic network constructions, we will develop and apply methods to construct integrated, higher-order molecular networks and more generalized models to enhance our understanding of the genetic loci and gene networks underlying schizophrenia. Using a Bayesian framework, methods will be developed that identify network modules and the underlying genetic variance component (including epistatic interactions), incorporate prior disease information and extensive prior biological knowledge to construct more detailed probabilistic causal models, and identify causal regulators of networks associated with SCZ. In AIM 4, we will assess the extent to which the models validate in independent SCZ data and in bipolar disorder and autism. This proposal should have a major impact on the field as it proposes to create a solution, in the form of new platforms and analytic methods, for the bottleneck in gene discovery that results from our limited ability to fully analyze the data currently available on large samples of individuals suffering fro mental illness. This proposal will make possible the efficient use of this wealth of multi-dimensional data. PUBLIC HEALTH RELEVANCE: In the United States, over a million people have schizophrenia. The costs are staggering in human and financial terms. We propose to develop methods for integrating a broad range of genomic data into a novel, flexible and extensible computing platform. Subsequently, these data will be used to develop a pipeline of algorithms for integrating and mining the data. We will use as a proof-of-concept multiple datasets for schizophrenia, and then extend this to additional mental disorders.

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): 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 long-term goal of the Program Project, Craniosynostosis Network, is to elucidate normal and abnormal craniofacial biology to ultimately improve the treatment of craniofacial disorders. Craniosynostosis (CS) and other skull abnormalities are among the most common human malformations and usually require surgical and medical intervention. Our Network will integrate the efforts of scientists with diverse expertise including anthropology, morphometry, imaging, birth defects, developmental biology, genetics, genomics, epidemiology, statistics, & system biology to explore the determinants of the fate of the relevant mesenchymal progenitor cells, and how abnormalities in the processes of osteogenesis contribute to disorders such as global skull growth abnormality, premature closure of sutures, in particular the coronal suture. We will use humans and mouse model systems to study normal development and malformations that characterize birth defects such as Apert, Crouzon, and Muenke syndromes & coronal nonsyndromic craniosynostosis. Our research design will be multidisciplinary including imaging, genomics, computational modeling & stem cell research; and evolutionary, developmental, & systems biology. Our approach will be hypothesis and discovery-driven, and we will generate and integrate a wide variety of human genomic, imaging, & laboratory data. The Network will be based at Mount Sinai Medical Center with the contact Principal Investigator (PI), Ethylin Wang Jabs, and multiple PIs, Inga Peter, Eric Schadt and Ihor Lemischka, and at Pennsylvania State University with MPIs Joan Richtsmeier, Patrick Drew, and Reuben Kraft. Our international and national collaborating institutions include: Hospital Necker-Enfants Maladies (France), University Hospital Heidelberg (Germany), and Hospital Sant Joan de Deu (Spain); Oxford University (UK), the International Craniosynostosis Consortium at University of California at Davis; New York State birth defect registry involved with the National Birth Defects Prevention Study based at Univ. of Iowa, directed by MPI Paul Romitti; New York University, Pennsylvania State Milton S. Hershey Medical Center; Boston Children's Hospital; Yale University, Univ. of Texas at Southwestern, and Johns Hopkins University. Our Advisory Committee includes experts in developmental biology, genomics, and system biology: Philippe Soriano from Mount Sinai Medical Center, James Sharpe of the Centre for Genomic Regulation, Barcelona, Spain, Alec Wilson of NIH NHGRI, & Richard Bonneau of New York University. Our proposal consists of Project I From Skull Shape to Cell Activity in Coronal Craniosynostosis, Project II Genomics Approaches to Coronal Nonsyndromic Craniosynostosis, & Project III Systems Biology of Bone in Coronal Nonsyndromic Craniosynostosis; and two Cores: Administrative Core A and Molecular/Analytic Core B. The investigators, by engaging as an integrated group in the study of complex biological networks, and by utilizing innovative and state-of-the-art technologies, will foster an outstanding research environment. The Network is strongly committed to sharing & disseminating our findings to the scientific community at large.

SUMMARY A system-level understanding of the dengue virus (DENV) host relationship, in particular, the network structure and dynamics, can be derived from experimental data with computational analysis across data sets and modeling. The host response to infection is a complex process involving entire networks of RNA transcription, protein signaling, and metabolism that complementarily influence cellular, tissue, and whole organism behaviors. This complexity demands a systems biology approach for understanding immune response, since investigation of single pathways is unlikely to explain changes taking place across the entire network. The Data Analysis and Modeling Core (Core E) will not only perform standard multivariate analyses on each dataset to find reliable biomarkers for differentiating outcomes of infection, but will furthermore integrate them with the full range of public network and pathway data to construct a multiscale, holistic network model of biologically meaningful DENV-host interactions. Because this model is quantitative and mathematically defined, it is well suited for training advanced classifiers that can predict both individualized clinical outcomes with more accuracy than biomarkers alone and novel ?key driver? biomolecules that can be validated with ex vivo siRNA screens (Project 3). These data should further inform on the synergy among multiple interrelating molecular pathways and networks that underpin the differences in phenotype between individuals. The scale of our proposed model for DENV is unprecedented, spanning the genomic, transcriptomic, proteomic, intercellular signaling, and immune cell subpopulation levels?and only with this scale of modeling will superior unbiased, data driven models that address the key biological questions in each of the Projects emerge, explaining the diverse subtleties of host response to DENV infection and vaccination that affect clinical outcomes.

? DESCRIPTION (provided by applicant): We are submitting Integrated Multiscale Networks in Schizophrenia in response to RFA-MH-16-300. Schizophrenia (SCZ) is a generally devastating neuropsychiatric illness with considerable morbidity, mortality, and personal and societal cost. Genetic factors have been strongly implicated via family and twin data, and more recently directly through genome-wide association studies (GWAS) and sequencing studies. The primary objective of our project is to develop and apply advanced integrative methods for computational and functional analysis of networks, including but not limited to Bayesian network reconstruction and prediction algorithms of variant causality to identify key drivers of SCZ pathology for potential therapeutic intervention. To achieve this in Aim 1 we will construct single tissue and multi- tissue probabilistic causal network by applying a novel top-down and bottom-up or hypothesis-driven probabilistic causal network approaches in RNA sequencing key drivers of networks, novel pathways, and new mechanisms in SCZ pathology data from the CommonMind consortium, incorporating prior information. In Aim 2 we will use network models derived in Aim 1 in order to improve the predictive SCZ networks that could be used to identify SCZ-relevant transcription-based features that can be useful in therapeutic screening. Finally, in Aim 3 we will use modified RNA and cellular models to validate the network models, key drivers and investigate their phenotype effects.

PROJECT SUMMARY/ABSTRACT The goal of our PGDAC is to improve understanding of the proteogenomic complexity of tumors. Towards this goal, our First Aim is to apply network based system learning to reveal causative molecular regulatory relationships contributing to varieties of phenotypes in cancer using CPTAC proteomic/genomic data. We will start with a mixed effects model to (1) fix the batch effects in data from multi-plex proteomics experiments; and (2) handle the large amount of missing data from abundance-dependent missing mechanisms in proteomic data (Aim 1.1). We will then utilize a multivariate penalized regression framework to construct the global regulatory networks between genomic alterations (such as DNA mutations, CNA, methylations), and protein as well as their PTM (post translational modification) abundances (Aim 1.2). Such regulatory networks help to elucidate how protein or pathway activities are shaped by genomic alterations in tumor cells. We will also construct protein co-expression networks based on global-, phosphor-, glyco- and other PTM-proteomics data (Aim 1.3). When constructing these networks, we will use advanced computational tools to effectively borrow information from literatures, databases, and transcriptome profiles. In addition, we will model tumor and normal tissues jointly, so that tumor specific interactions and network modules will be inferred with better accuracy. Both Aims 1.2 and 1.3 will lead to a big collection of network modules, as well as functionally related protein sets (e.g. proteins regulated by the same genomic alteration). These network modules and protein sets will then be tested for their associations with disease phenotypes (Aim 1.4). In the end, we will derive a more integrated view of commonalities and differences across multiple tumor types via a Pan-cancer analysis (Aim 1.5). Our Second Aim is to further develop methods, software, and web-tools to optimize the data analysis in our PGDAC. We will develop novel statistical/computational tools tailored to CPTAC proteomics data; implement these methods as computationally efficient software; and construct an integrated data analysis pipeline (Aim 2.1). We also plan to develop a set of web service tools for visualization and biological annotation of protein networks and clinical interpretation of proteomic data (Aim 2.2). Our Third Aim is to nominate novel protein-based cancer biomarkers and drug targets for further investigation by targeted proteomics assays. We will first utilize a prediction based scoring system to identify protein biomarkers that predict altered cancer pathways, network modules and individual oncogenes; disease outcome and drug resistance; and therapeutically distinct disease subtypes (Aim 3.1) We will then utilize network based tools to identify driver players in selected proteins signature sets (Aim 3.2). These driver proteins could play important roles in shaping the overall function of regulatory system, and thus serve as good candidates for cancer biomarkers and drug targets. We will also take into consideration of domain knowledge of different diseases, as well as technique constrains for developing targeted proteomics assays in biomarker selection. !

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 Despite extensive clinical and genomic studies, the mechanisms of development and progression of Alzheimer's disease (AD) remain elusive. Microglia and other myeloid origin cells (collectively called human brain immune cells, or HBICs) have recently emerged as crucial players in the pathogenesis of AD. This is supported through genetic association studies, where many of the common and rare risk loci affect genes that are preferentially or selectively expressed in HBICs, emphasizing the pivotal role of the innate immune system in AD. In addition, single cell RNA sequencing analysis in mouse models of AD has identified a microglia subpopulation that is present at sites of neurodegeneration. It is unclear if HBICs assume a protective or damaging role, but that might vary depending on the stage and progression of AD. Therefore, further analysis of microglia and other immune cells purified from human brains is needed to understand the state of HBIC activity in human AD at different stages of disease. As HBICs constitute a small proportion of total brain cells, homogenate-based studies in human brain tissue are unlikely to capture the full spectrum of HBIC molecular signatures, especially in light of the growing appreciation for the diversity of HBICs in the brain. The proposed work addresses some of the limitations of previous research and is focused on: (1) cell type specific and single cell studies in immune cells isolated from human brain tissue; and (2) a systematic study of the regulatory effects of non-coding DNA on gene and protein expression, which is necessary given that the majority of common risk variants are situated in non-coding regions of the genome. More specifically, our application is uniquely designed to: (1) apply innovative genomic approaches and generate multi-omics data from HBICs isolated from 300 donors, including whole genome sequencing, RNAseq, ATACseq, HiC chromosome conformation capture and proteomics; (2) perform state-of-the-art single cell analysis that will allow us to assess the diversity of HBIC subpopulations, as well as detect those that are associated with AD; (3) connect AD risk loci with changes in the regulatory mechanisms of gene and protein expression in HBICs; and (4) organize HBIC multiscale data in functional networks and identify key drivers for AD. Our overall hypothesis is that HBIC subpopulations assume a neuroprotective role during aging and early stages of AD, but as disease progresses, specific HBIC subpopulations transform to neuroinflammatory phenotype(s). This conversion is partially driven by AD risk genetic variants, which affect regulatory mechanisms of genes that are key drivers of neuroinflammatory HBIC subpopulations. Successful completion of the proposed studies will provide: (1) an increased mechanistic understanding of dysfunction in AD risk loci; (2) prioritization of significant loci and genes for future mechanistic studies; and (3) access to large-scale, multidimensional datasets, together with systems level analyses of these datasets for transcriptional regulation in HBICs, which is an urgently needed (and currently missing) resource.

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.