Chunyu Liu, Ph.D. - US grants

[unreadable] DESCRIPTION (provided by applicant): MicroRNAs (miRNAs) are single-stranded RNAs (ssRNA) of 19-25 nucleotides in length that function as guide molecules in post-transcriptional gene silencing, by base pairing with target mRNAs. Since one miRNA can regulate expressions of multiple genes and about 1/3 of the human genes are regulated by miRNAs, variants in miRNAs have intriguing potential roles in the psychiatric diseases. To date there has been little study of genomic variation in miRNA genes, or the relationships between miRNA variants and psychiatric disorders, or where miRNA variants are correlated with variations in gene expressions in the brain. To promote such knowledge, we propose to use deep resequencing to exhaustively identify variants in all known human miRNAs. Then we will study the correlation between miRNA variants and gene expression in brain, and association of miRNA variants with schizophrenia (SZ) and bipolar disorder (BD). We will resequence all 462 known human miRNA genomic DNA sequences in 310 Caucasian individuals including BD, SZ, and normal controls (CN). 210 of the 310 individuals have already had microarray study of gene expression in brain or lymphoblastoid cells. Resequencing will identify all common and many rare variants. We will then test association of disease with genotypes of these variants in a large collection of case-control BD and SZ samples from the NIMH Genetics Initiative, including 3000 BD, 3000 SZ and 2250 CN. We will also test for correlations between sequence variants in miRNAs and gene expression data (as quantitative traits) in Stanley brain samples and CEPH lymphoblastoid cell samples. We will use appropriate statistical methods to approach multiple testing and potential population stratification problems. These findings will potentially enhance our understanding of the miRNA genes and their potential roles in etiology of BD and SZ. Development of new diagnoses and treatments may result. [unreadable] [unreadable] [unreadable]

DESCRIPTION (provided by applicant): G72/G30, a sense-antisense gene complex on chromosome 13q33.2 also known as DAOA (D-amino acid oxidase activator), has been associated with schizophrenia and bipolar disorder in multiple studies. The disease association and its characteristic of being one of the very few primate-specific genes make G72/G30 a very intriguing target in the biological study of human brain and mental diseases. Our most recent meta-analysis of schizophrenia indicates that two SNPs in G72/G30 show gene-wide significant association in Asians. In this application, we propose to screen Asian schizophrenia patients (DNAs have been collected by the NIMH Genetics Initiative) for an individual carrying the allele that shows association with schizophrenia. Then, we will use the Transformation Associated Recombination (TAR) Cloning method to isolate G72 Bacterial Artificial Chromosome (BAC) clones from the selected schizophrenia patient. The isolated clone will be introduced into a mouse genome to produce a transgenic mouse line. We will then perform biochemical, gene expression, and behavioral tests on three types of mice: the new transgenic mouse line produced in this study, our existing G72 transgenic mouse line that carries the non-risk allele, and wild type mice. Comprehensive use of genetic, functional genomics and behavioral genetics approaches should lead us to an improved understanding of the biology of G72 in brain function and in schizophrenia susceptibility. We will create a transgenic mouse line that carries a human G72/G30 gene complex with a schizophrenia risk allele. G72 (as we refer to the complex) is one of the most consistently associated genes with both bipolar disorder and schizophrenia. Because G72 is a primate-specific gene with low expression levels in human, biological study of G72 in humans is particularly challenging. With the new transgenic mouse generated in this application, and our existing G72 transgenic mouse without the schizophrenia risk allele, we can perform in vivo tests on G72 biological and behavioral functions. We expect these results will ultimately lead to a better understanding of G72 and its role in both normal human brain function and psychiatric disease states.

DESCRIPTION (provided by applicant): An average of over 7,500 large copy and paste DNA insertions occurs in each human brain as somatic mutations, i.e., DNA mutations that occur post-fertilization. Because of the large size of the insertions, they are termed structural genomic mutations. The mechanism of such mutations is mainly retrotransposition of transposable elements in the human genome. This phenomenon may prove relevant to Alzheimer's disease (AD), where somatic mutation hypotheses have been repeatedly proposed. These mutations are thought to lead to AD by interacting with inherited susceptibility variants, in a multiple-hit manner. We propose here to detect somatic mutations, including both point mutations and structural genomic mutations, in temporal cortex in a small sample of 7 AD patients and 7 controls, with two methods of mutation detection, and independent validation of each method. The two complementary methods for somatic mutation detection are microarray- based Transposon Insertion site Profiling (TIP-chip) capture coupled with sequencing (RC-Seq) and paired- end whole-genome sequencing (WGS). We will demonstrate the somatic nature of each mutation by comparison of brain and liver sequences in each individual. Comparative evaluation will inform our choice of methods for future studies. If justified by results of this stdy, we will next propose a large-scale case-control study, to determine if there is a somatic mutational burden, and/or specific genes impacted by somatic mutations, in AD. The PIs in this proposal include Dr. Geoff Faulkner, a pioneer in somatic mutations whose lab produced the 2011 Nature paper reporting the discovery of high-frequency somatic mutations in adult human brain, and Drs. Chunyu Liu and Elliot Gershon, who have published extensively on structural genomics, epigenetics, and bioinformatics in neuropsychiatric disorders.

DESCRIPTION (provided by applicant): Mental illnesses are some of the most devastating diseases affecting human populations, placing a huge burden on individuals, families and society. Genome-wide association studies (GWAS) have identified dozens of common single nucleotide polymorphisms (SNPs) that are associated with psychiatric diseases, but a majority of those SNPs have been mapped to intergenic or intronic regions and are functionally unclassified. The overall goal of this proposed study is to use genetic mapping of quantitative trait loci (QTL), including expression QTLs (eQTLs), protein QTLs (pQTLs), and DNase I sensitivity QTLs (dsQTLs), to map non-coding regulatory elements in human brain, then to use the QTL SNPs to uncover regulatory mechanisms underlying GWAS findings and to discover novel risk genes. Our previous studies have shown that psychiatric GWAS signals are enriched with brain eQTL SNPs (eSNPs), and these brain eSNPs are likely to be functional and contribute to disease susceptibilities. We hypothesize that other QTLs will similarly represent other levels of regulation. So, using QTL mapping, we will identify SNPs affecting chromatin accessibility in brain (dsQTLs), and downstream gene and protein level variations (eQTLs and pQTLs). We will use RNA-seq, micro-western arrays (MWAs), reverse phase protein arrays (RPPAs), and DNase-seq to profile prefrontal cortex and cerebellum of 432 postmortem brains, along with sorted NeuN+ and NeuN- nuclei. Using the optimal deconvolution method, all brain measures will be partitioned into neuronal and non-neuronal measures for QTL mapping. We will thereafter re-analyze existing GWAS data for seven psychiatric diseases, plus three non- psychiatric diseases/traits as controls, to understand the contributions of neuronal- and non- neuronal QTL SNPs to disease risks. We will also look for differential expressions of transcripts and proteins, as well as for differential DNA sensitivities, in patient brains, and use these molecular measures to construct novel regulatory networks. This integrative study represents a timely, novel and powerful approach that will transform our understanding of brain genomics and the genetic risks of psychiatric diseases. It is positioned to create a new paradigm for integrating brain genomics and psychiatric genetics that are truly distinct from current approaches.

DESCRIPTION (provided by applicant): Mental illnesses are some of the most devastating diseases affecting human populations, placing a huge burden on individuals, families and society. Genome-wide association studies (GWAS) have identified dozens of common single nucleotide polymorphisms (SNPs) that are associated with psychiatric diseases, but a majority of those SNPs have been mapped to intergenic or intronic regions and are functionally unclassified. Existing software or algorithms only query multiple databases and produce lists of hits without intelligent integration and ignore much of the valuable regulatory information. The overall goal of this proposal is to integrate all available genetic, genomic and epigenomic data to generate a probability-based prediction about a SNP's influence on gene expression level in brain. Our previous studies have shown that psychiatric GWAS signals are enriched with brain eQTL SNPs (eSNPs), and these brain eSNPs are likely to be functional and contribute to disease susceptibilities. We will use SNPs in eQTLs to anchor a chain of evidence incorporating histone marks, conserved sequences, transcription factor binding sites, DNA methylation, accessible chromatins, non-coding RNA, and other data. We will use a machine learning method to predict regulatory SNPs based on known relationships between these epigenetic marks and their target genes, as well as their distinct patterns in genome. We will also use our novel unsupervised deconvolution algorithm to extract cell-type (i.e., neuron vs. non-neuron) specific measures from heterogeneous brain tissue data to improve our predictions. We will use both statistical and experimental methods to validate the predictions. Quantitative PCR and CRISPR-cas9 will be used on induced pluripotent cell lines to compare gene expression levels of alleles of predicted functional SNPs. Both algorithm and predicted functional variants will made public via a website and standalone application. The novel algorithm will significantly improve our understanding of psychiatric disease genetics by uncovering the gene-regulatory functions for disease-associated, non-coding SNPs.

Abstract To further our efforts in identifying the molecular bases of bipolar disorder and schizophrenia, we propose to study protein translation and abundances at the genome-wide level in frontal cortex tissue from 300 brains from patients and healthy controls. We have already amassed an enormous amount of data from these brains, including genotypes, transcriptome profiles and chromatin states. The next step is to look for alterations in protein function in the same brains, since proteins are the ultimate products of gene expression and a critical link between genetic variants and higher order phenotypes, including disease diagnosis. Since proteins are encoded by mRNA transcripts, it appears that protein levels should roughly correlate with transcript levels. However, measured expression levels of mRNAs and their corresponding proteins are often discordant, as are maps of their respective quantitative trait loci. Since we are unable to explain these discrepancies, our picture of molecular changes underlying psychiatric disorders is clearly incomplete. Most previous population-based studies of proteins in neuropsychiatry have been limited to candidate proteins, for which antibodies are already available. For example, in our PsychENCODE project, we are the process of using microwestern arrays to assay ~1000 proteins. In this study, we will use the recently developed technique of ribosome profiling and next-generation proteomics to identify which transcripts are actively being translated in brain and to quantify the abundance of more than 12,000 proteins. Through integrative data analysis, we use the two complementary technologies to detect translational products and to measure their quantitative relationships. Furthermore, these proteins and their translation efficiencies will be assessed for association with disorders. To further improve the specificity of quantification, we will use state-of-the-art deconvolution methods to quantify cell type specific measures of translation efficiency and protein products. This will allow protein translation and abundance in specific major brain cell types to be studied for their changes in affected brains. This study is innovative for being the first genome-wide, population-based study of protein translation and abundance in brains of psychiatric patients. It offers a unique opportunity to fill the gaps between transcriptome and proteome data, and between genetic variants and higher-order phenotypes. It will be a huge step forward in studying the proteins of human brains and the regulatory changes associated with psychiatric disorders, which should ultimately lead to better diagnosis and treatment of these diseases.

Abstract Aging is the main risk factor for many chronic diseases, including late onset Alzheimer's disease (LOAD) and many age-related metabolic diseases, such as obesity and diabetes. Using AD as an example, the number of people with AD doubles every 5 years beyond age 65. In 2017, 5.3 million Americans 65 years or older are affected by LOAD. The burden of health care costs for LOAD is enormous ? $259 billion in 2017. Thus, early detection and treatment of these age-related diseases should be a core tenet of public health. Research is needed to guide such efforts. Over the last decade, accumulating evidence has linked aging to mitochondrial dysfunction. Mitochondria are tiny powerhouses, generating more than 90% of energy to support normal cellular function. Mitochondria contain their own genome (mtDNA) which is both polymorphic and heteroplasmic, i.e., two or more mtDNA alleles can co-exist in the same cell due to the presence of many mtDNA molecules within any cell. Previous studies in Europeans have found that reduced mtDNA copy number was associated with frailty and higher mortality among elderly. Furthermore, reduced mtDNA copy number in human cerebrospinal fluid was observed at least a decade before clinic AD symptoms develop. These findings in Europeans need to be generalized in other ethnic groups. Several hundreds of mtDNA rare mutations have been described to cause mostly rare, yet severe maternally inherited diseases. A limited number of common mtDNA polymorphisms were examined in relation to metabolic disorders, dementia and cognitive functions. Robust associations, however, haven't been established between common mtDNA polymorphisms and age-related common diseases. In most of these previous studies, heteroplasmic mtDNA mutations haven't been well studied with respect to aging and age-related human diseases because, until recently, sequencing has been extremely costly. Studying a spectrum of mtDNA mutations along with mtDNA copy number in relation to age-related traits in large samples has now become possible thanks to drastically decreased whole genome sequencing costs. This proposed study will leverage five prospective cohorts, each with whole genome sequencing data generated from the National Heart, Lung and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) and extensive cognitive, brain structure, and cardiometabolic measures. The expected outcomes of the work proposed are to 1) develop a novel statistical method to identify age-related heteroplasmic (i.e., somatic) mtDNA mutations, and 2) develop a statistical framework to analyze mtDNA copy number and heteroplasmic mutations in relation to key age-related disorders, include LOAD and age-related metabolic traits. Results of this investigation are expected to advance understanding of the role of aging on the mitochondrial genome, and in turn, the contributions of mitochondrial genome to age-related traits. Equally important, a positive impact of this project will be advancing knowledge of the role of mtDNA in a spectrum of age-related complex phenotypes.

Abstract To further our efforts in identifying the molecular bases of bipolar disorder and schizophrenia, we propose to study protein translation and abundances at the genome-wide level in frontal cortex tissue from 300 brains from patients and healthy controls. We have already amassed an enormous amount of data from these brains, including genotypes, transcriptome profiles and chromatin states. The next step is to look for alterations in protein function in the same brains, since proteins are the ultimate products of gene expression and a critical link between genetic variants and higher order phenotypes, including disease diagnosis. Since proteins are encoded by mRNA transcripts, it appears that protein levels should roughly correlate with transcript levels. However, measured expression levels of mRNAs and their corresponding proteins are often discordant, as are maps of their respective quantitative trait loci. Since we are unable to explain these discrepancies, our picture of molecular changes underlying psychiatric disorders is clearly incomplete. Most previous population-based studies of proteins in neuropsychiatry have been limited to candidate proteins, for which antibodies are already available. For example, in our PsychENCODE project, we are the process of using microwestern arrays to assay ~1000 proteins. In this study, we will use the recently developed technique of ribosome profiling and next-generation proteomics to identify which transcripts are actively being translated in brain and to quantify the abundance of more than 12,000 proteins. Through integrative data analysis, we use the two complementary technologies to detect translational products and to measure their quantitative relationships. Furthermore, these proteins and their translation efficiencies will be assessed for association with disorders. To further improve the specificity of quantification, we will use state-of-the-art deconvolution methods to quantify cell type specific measures of translation efficiency and protein products. This will allow protein translation and abundance in specific major brain cell types to be studied for their changes in affected brains. This study is innovative for being the first genome-wide, population-based study of protein translation and abundance in brains of psychiatric patients. It offers a unique opportunity to fill the gaps between transcriptome and proteome data, and between genetic variants and higher-order phenotypes. It will be a huge step forward in studying the proteins of human brains and the regulatory changes associated with psychiatric disorders, which should ultimately lead to better diagnosis and treatment of these diseases.

PROJECT SUMMARY Mitochondrial DNA (mtDNA) sequence variations have been causally implicated in cardiometabolic diseases (CMDs). However, the mechanisms linking mtDNA mutations to CMDs are not fully elucidated. Each human cell contains as many as 10,000 copies of mtDNA. As a result, there are two types of mtDNA mutations ? homoplasmic and heteroplasmic mutations. The latter represent the coexistence of two (or more) mtDNA alleles in the same cell or across cells. For most heteroplasmic mtDNA mutations, the proportion of mutant alleles is low and can only be detected by deep sequencing of mtDNA. Whole genome sequencing (WGS) through NHLBI?s Trans-Omics for Precision Medicine program has generated deep WGS data, including mtDNA, in tens of thousands of individuals, offering an unprecedented opportunity to investigate heteroplasmic mtDNA mutations in relation to CMDs. We propose to develop rigorous methods and a software suite to facilitate association studies of heteroplasmic mtDNA mutations in large samples. The novel methods and the software suite will be open source and available to the broad scientific community to study mtDNA sequence variation with respect to age-related diseases, including CMDs. The body of knowledge generated by these future findings will facilitate the development of new modalities for the diagnosis, prevention, and treatment of age-related diseases.

Project Summary: Most genetic variants associated with disease in genome-wide association studies (GWAS) lie in non-coding gene regulatory elements (GRE; e.g., promoters and enhancers). GREs are tissue- and cell type-specific and are identified through their epigenomic signatures, including low DNA methylation (DNAm), DNA accessibility and certain histone modifications. The PsychENCODE Consortium has characterized brain GREs across brain regions and developmental time points, as well as in the brains of psychiatric patients using mostly DNA accessibility and histone modification marks. These marks, however, identify large regions of enrichment (~300-2,000bp), providing only low resolution coverage of the important regulatory nucleotides, e.g., transcription factor binding sites. DNAm (especially cell type-specific DNAm) has received less attention, although it has been linked to psychiatric disorders, including schizophrenia (SZ) and bipolar disorder (BD). In the adult human brain, ~80% of CG and 1.5% of non-CG (CH) sites are methylated, and can be converted to hydroxymethylcytosine (hmC) and further demethylated. In postmitotic neuronal genomes, mCH and hmC accumulate to a significantly higher level than in other tissues--a distinct feature of the brain's epigenome. Bisulfite sequencing (BS)-based approaches that are used to measure (h)mC can pinpoint GREs with single base resolution, presenting a unique opportunity to refine the gene regulatory landscape of the brain cell types. Here we aim to create reference DNAm maps [mC and hmC, using BS and oxidative (ox)BS sequencing] and transcriptional profiles (using RNA-seq) in two major subtypes of neurons in the human dorsolateral prefrontal cortex (DLPFC), namely excitatory glutamatergic (Glu) and inhibitory GABA-ergic neurons. The proposed work is based on fluorescence-activated nuclear sorting (FANS) methods that we recently developed to separate nuclei of different cell types from autopsied human brain, and on our recent findings that showed unexpected relationships between DNA(h)m, GREs, and gene expression in the DLPFC Glu and GABA neurons. We will perform these studies at key time points of postnatal brain development and adulthood to map DNA(h)m within active neuron subtype-specific GREs that may be vulnerable to disruption during childhood and adolescence periods that coincide with the critical processes of brain maturation and circuit refinement (Aim 1). This work will be complemented with single nucleus mC profiling, which will allow us to define the developmental trajectories of mC in discrete subpopulations of Glu and GABA neurons (Aim2). Finally, we will profile Glu and GABA neurons in 150 autopsied DLPFC samples obtained from controls and cases of SZ and BD, and will then map neuron subtype-specific gene expression and (hydroxy)methylation quantitative trait loci (eQTL, mQTLS, hmQTLs) (Aim3). We will integrate QTL, transcriptome, and DNA(h)m data with the results of SZ and BP GWAS to reveal DNA(h)m and gene expression-mediated causal risk variants and genes, and to distinguish specific neuronal subtype(s) that are critical in the etiology of these disorders.

PROJECT SUMMARY Energy metabolism plays a critical role in human disease. Mitochondria, the energy powerhouses of the cell, have their own genome (mtDNA) which is present up to thousands of copies per cell. mtDNA encodes genes for proteins of energy metabolism. We (led by Liu, PI of this application) recently discovered that lower mtDNA copy number in whole blood is an independent predictor for higher levels of cardiovascular disease (CMD) risk factors in ~60,000 participants from multiple ancestries. For example, one standard deviation of decrease in mtDNA copy number was associated with increased odds of obesity (OR=1.15, p=8e-31) and metabolic syndrome (OR=1.14, p=1e-32), as well as with increased levels of several quantitative traits defining these diseases. Despite these findings, the molecular basis underlying the association of mtDNA with CMD is unclear because the nuclear genome (nDNA) also encodes many of the proteins engaged in mitochondrial energy production and biosynthesis, and thus, maintenance of mitochondrial function requires extensive coordination of mtDNA and nDNA. A mouse hybrid nDNA-mtDNA system was developed. Using this model, the researchers found differential nDNA methylation, gene expression, and cellular adaptive response in hybrid mice of identical nDNA, but with different mtDNA background. Additionally, we (led by Arking, Co-I of this application) identified several significant DNA methylation sites associated with mtDNA copy number. In addition, experimental modification of mtDNA copy number through knockout via CRISPR-Cas9 of TFAM, a regulator of mtDNA replication, demonstrated that modulation of mtDNA copy number directly drives changes in nDNA methylation of specific CpGs and gene expression of nearby transcripts. Based on these previous studies in mouse model and our own research, we hypothesize that methylation and gene expression of nDNA mediate the effects of mtDNA on cardiometabolic disease traits. In this proposed proposal, we will leverage existing resources, including whole genome sequencing and multi-omics in six large cohorts of multiple ancestries; we will rigorously test our hypothesis by pursuing four specific aims. In Aim 1 and Aim 2, we will perform association analyses to identify mtDNA-associated nDNA methylation sites and gene expression levels, respectively. mtDNA features include mtDNA homoplasmic and heteroplasmic mutations, and mtDNA copy number. In Aim 3, we will investigate whether nDNA methylation and/or gene expression mediates the effects of mtDNA copy number and heteroplasmy on continuous cardiometabolic disease traits. In Aim 4, we will perform integrative analyses to identify gene regulation networks underlying mtDNA and cardiometabolic disease traits. We will also functionally test the impact of mtDNA on these gene networks via edited cell lines (e.g., via CRISPR-Cas9 system). The body of knowledge generated by this research project will deepen our understanding of molecular mechanisms underlying mtDNA and cardiometabolic diseases, which will ultimately facilitate the development of new modalities for the diagnosis, prevention, and treatment of cardiometabolic diseases.

PROJECT SUMMARY Alcohol consumption is an important modifiable lifestyle risk factor for cardiovascular disease (CVD). Advances in high-throughput technologies have made it possible to measure and analyze a variety of ?omics?, that is, DNA methylation (methylome), gene expression (transcriptome), protein (proteome) and metabolite (metabolome), in relation to alcohol consumption and CVD in a cost-efficient manner. In this grant proposal, we will test the hypothesis that alcohol consumption alters multiple molecular processes across different ?omics? dimensions and these molecular intermediates mediate alcohol?s effects on CVD. To test this hypothesis, we will assemble a multidisciplinary team to test five Specific Aims. In Aim 1, we will first identify alcohol-associated transcriptomic markers. We will then examine the associations of alcohol-associated transcriptomic markers with incident CVD (fatal and nonfatal coronary heart disease and ischemic stroke), and test whether these transcriptomic markers mediate the effect of alcohol intake on incident CVD. We will construct polygenic risk scores to examine the potential causal relationships between alcohol intake, transcriptomic markers, and incident CVD. We will also conduct a two-sample Mendelian randomization (MR) analysis to further test and quantify the causal relationships. In Aims 2 and 3, we will identify alcohol-associated proteomic (Aim 2) and metabolomic (Aim 3) markers and will test the relationship between alcohol, omics markers and CVD using similar methods outlined in Aim 1. Many alcohol-associated DNA methylation markers have been identified by our prior study in >13,000 participants. Therefore, in Aim 4, we will examine the relations of alcohol-associated DNA methylation markers with incident CVD using similar methods outlined in Aim 1. In Aim 5, we will integrate alcohol-associated multi-omics markers to identify key pathways and multi-omics modules associated with alcohol intake and CVD. We will perform Random Forest analysis to identify additional alcohol-associated multi- omics markers. We will also perform network analyses to identify modules of alcohol-associated multi-omics markers. We will examine the relationships between these identified multi-omics markers with incident CVD, and will functionally annotate these markers using bioinformatics tools (e.g., ENCODE, GTEx, Human Metabolome Database, and UniProt). With the completion of this project, we will highlight the novel molecular targets for both alcohol-associated CVD risk prevention and treatment. In addition, we will generate a multi-omics biomarker database of alcohol intake with tailored software to facilitate the future investigations of alcohol?s relations with other non-communicable diseases such as neurodegenerative diseases.