Timothy E. Reddy, Ph.D. - US grants

DESCRIPTION (provided by applicant): Our current understanding of gene regulatory networks does not adequately utilize information from individual protein-DNA interactions and the millions of regulatory elements identified by high-throughput functional assays. New strategies are needed to incorporate data from each of these experimental scales, and to leverage the orthogonal datasets to understand how each regulatory element is involved in directing complex gene expression responses. The objective of our proposal is to develop statistical models to learn the underlying patterns of complex interactions involved in transcription regulation across the genome. While the genome-proximal response to glucocorticoid (GC) treatment is an ideal model system, the methods developed will be applicable to studying any complex regulatory network. The goal of Aim 1 will be to comprehensively characterize the first 12 hours of the GC response using genome-wide methods to quantify expression, TF binding, histone modifications, chromatin accessibility, three-dimensional chromatin structure, and the function of regulatory elements. The outcome will be the most comprehensive and coordinated molecular description of a human regulatory network ever produced. All data will be generated with the highest possible quality standards, and will be submitted pre-publication and without restriction into the public domain. Aim 2 will integrate that data into a nonparametric and hierarchical Bayesian model of the GC response network (GCRN). That model will able to produce functional predictions for each individual regulatory element while also generalizing across genes to reveal shared principles of gene regulation. Aim 3 will validate and reduce uncertainty in the model. That will be accomplished by combining statistical experimental design approaches with multiplex genome and epigenome engineering to iteratively and optimally resolve the most uncertain aspects of the model. The outcome will be a validated and predictive model of the GCRN that will be useful to design customized genomic responses. Aim 4 will demonstrate the use of the resulting model through reprogramming the GCRN to minimize the response of genes associated with metabolism while maintaining the response of genes associated with inflammation and immunity. The outcome will be a derived cell line with a custom programmed GC response. That cell line will have immediate use for studying individual aspects of the GC response; and the approach used to design and realize the customized response will have broad implications for the study of other transcriptional response networks. The overall result of this project will be a mechanistic and actionable understanding of the principles through which individual DNA sequences contribute to the GCRN; a general and transferrable multi-scale modeling approach to study any complex regulatory network; and the novel ability to genetically reprogram transcriptional response networks to study their individual components. We anticipate that that outcome will have broad positive impact on both experimental and computational fields of biomedical research.

There is a fundamental gap in understanding how the millions of known regulatory elements functionally contribute to gene regulation and phenotypes. Continued existence of that gap is an important problem because, until it is filled, it will remain extremely difficult to identify the genetic mechanisms underlying the thousands of observed genetic associations with disease phenotypes. Our long-term goal is to understand how and to what extent gene regulatory elements alter target gene expression and impact phenotypes. The objectives of this particular proposal are to functionally characterize all regulatory elements contributing to the differentiation of CD4+ T cells. In doing so, we will identify the causal regulatory mechanisms that modulate the immune system. The rationale for this work is that understanding those mechanisms will be the foundation for future efforts to therapeutically modulate the immune system, and will establish a discovery platform for determining the mechanisms underlying countless other model systems. Specifically, we will characterize three complementary components of regulatory element activity: (i) the capacity of regulatory elements to drive expression of a reporter gene, (ii) the effect of each regulatory element on the expression of one or more target genes, and (iii) the contributions of regulatory elements to phenotypic function, namely differentiation. We will accomplish those goals across three specific aims. In Aim 1, we will quantify the activity of all regulatory elements that have evidence of differential activity between subtypes of mouse CD4 T cells. We will do so using a capture-based high-throughput reporter assay that allows us to assay larger (>500 bp) fragments from specific genomic regions of interest. In Aim 2, we will quantify the effects of regulatory elements on target genes using a novel strategy that combines high-throughput CRISPR/ Cas9-based epigenome editing screens and targeted high-throughput single-cell RNA-sequencing. In Aim 3, we will determine which regulatory elements are necessary or sufficient for CD4 T cell differentiation using high-throughput CRISPR/Cas9-based epigenome-editing screens combined differentiation into particular CD4 T cell subtypes. Each aim will provide functional characterization of all of the regulatory elements implicated in CD4 T cell differentiation. Together, the aims will provide a comprehensive, multi-layered, and systematic understanding of the ways that gene regulatory elements modulate the immune system. The result will be an actionable set of targets for designing strategies to modulate immune system activity for therapeutic benefit. Because the approach is general to any model system, the same strategy can be readily transferred to diverse systems including differentiation and disease models. Therefore, we expect that this project will have both immediate and long-term benefit for determining the ways that regulatory elements contribute to health and disease.

ABSTRACT Determining new causes for rare and common disease would have major and immediate benefits for patients and their families by improved genetic testing, genetic counseling, insurance reimbursement, and ultimately more effective treatment options. Our long term goal is to disruptively improve and expand genetic testing for rare and common disease. Current diagnostic tests only consider pathogenic variants in protein-coding genes. However, we now have evidence that a substantial fraction of rare disease is due to unknown non-coding genetic variants that influence the regulation of those genes. The goal of this proposal is to identify and quantify the effect of pathogenic non-coding genetic variants on the function and expression of genes that cause rare disease. This initial step will enable treatment early in life when it is still possible to stop the most severe consequences of disease, including death. We will focus on severe early-onset pediatric disorders, including glycogen storage diseases (GSD I, II, III, IV, and IX), and the fatty acid oxidation disorders, very long-chain acyl-CoA dehydrogenase deficiency (VLCAD), and multiple acyl-CoA dehydrogenase deficiency (MADD). To date, genetic tests for these and other diseases are limited to protein-coding mutations. However, our clinical team has collected numerous cases that have a single pathogenic coding variant on only one of the two alleles that must be both affected in these recessive disorders. We also have biochemical and biomarker evidence that supports the diagnosis. Those cases are an ideal opportunity to identify additional disease-causing variants. Our hypothesis is that the genetic causes of recessive disorders include novel genetic variants that can alter either protein sequence (Aim 1), splicing (Aim 2), or gene expression (Aim 3) of disease genes. We have assembled a team of Pediatric clinicians who are experts in GSDs, VLCAD, and MADD, as well as researchers who are experts in genetics, genomics, epigenetic regulation, biomedical engineering, and statistics. This team has obtained patient samples and received Duke IRB approval to begin immediately. We expect this study will identify and validate novel genetic variants that influence disease. While we propose to study a relatively small subset of rare disorders, these strategies will be immediately generalizable to any patient sample with any recessive disorder that has inconclusive genetic testing results. That outcome will provide comprehensive genetic testing, better understanding of disease mechanisms, and ultimately better treatment options.

Understanding the genetic causes of human disease has immense potential to benefit human health. The human genetics community has devoted tremendous resources to identifying those causes, including, most recently, whole genome sequencing of patient cohorts. Those studies have found genetic variation in non-coding regions of the genome to be most often associated with diseases and drug responses. Unfortunately, since the effects of genetic variation on gene regulation remain poorly understood and difficult to study at the genome-wide scale, the full benefit of most of those studies has yet to be realized. Our long-term goal is to understand how non-coding genetic variants act through gene regulatory elements to influence phenotypes. The objective of this proposal, a step towards that long-term goal, is to develop a platform of empirical and statistical methods to reliably and systematically determine the regulatory mechanisms underlying human traits and diseases. Specifically, in Aim 1, we will use high- throughput reporter assays to quantify the effects of millions of human genetic variants on regulatory element activity. Those variants will represent diverse human ancestries, and will cover over 60% of all regions associated with a trait or disease via GWAS. The outcome will be the most extensive catalog of human regulatory variation every created. In Aim 2, we will develop new technologies to systematically relate those changes in regulatory element activity to changes in gene expression. That technology will combine our previous work developing CRISPR-Cas9- based epigenome editing screens with targeted single-cell RNA-seq. In Aim 3 we will develop statistical analyses to integrate the effects of regulatory variants to infer changes in gene expression and differences in phenotypes between individuals. The resulting method will be analogous to gene based association tests, but for the noncoding genome. The expected outcomes of this project are (i) dramatically improved ability to establish mechanisms underlying non-coding associations with human traits and diseases; (ii) better understanding of the genetic architecture of regulatory element activity and gene regulation that will guide the design and interpretation of future genetic association studies; and (iii) novel reagents, protocols, and software that other labs can use to complete similar investigations of their own model systems of interest. Taken together, we expect that this project will be a major step towards fully realizing the potential of genome wide and whole genome association studies.

Gestational hyperglycemia and gestational diabetes (GDM) are associated with adverse pregnancy outcomes for mothers and newborns. Additionally, GDM, in particular, can also be detrimental to metabolic outcomes later in life. A large genetic study, ?The Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study? previously identified a unique genetic association near hexokinase domain component-1 (HKDC1) to gestational hyperglycemia. This study has been confirmed by others and also shown to be associated with GDM. This grant renewal intends to continue our investigation of this important link to gestational glucose metabolism. The focus of this proposal is based on our data that HKDC1 interacts with the mitochondrial outer membrane protein, VDAC, in hepatocytes, where this interaction is disrupted when the amino terminus of HKDC1 is deleted. Further data shows that overexpression of HKDC1 in the liver improved glucose tolerance during pregnancy in mice and our data suggests that this results in a metabolic shift in the carbon flux toward anabolic pathways. Now, it is important to investigate the molecular basis of HKDC1 interaction with mitochondria and the impact of such interactions on mitochondrial morphology and function. Further, we expect that genetic variants contribute to GDM risk via HKDC1 expression, though the specific causal variants remain elusive. In sum, this proposal will mechanistically explore the role of HKDC1 in gestational glucose homeostasis and the genetic variants driving its expression in humans.

ABSTRACT Large scale genome annotation consortia such as ENCODE, Epigenomics Roadmap, and others have identified millions of putative regulatory elements. We now need to focus efforts on comprehensively characterizing and quantifying the function of those elements, and noncoding variants that map within these regions, on gene expression and cell phenotypes. Our long-term goal is to assign function to every regulatory element and noncoding variant in the human genome, understand how that function changes in different contexts, and use that information to better understand cell fitness, disease mechanisms, cell lineage specification, and tissue homeostasis. To accomplish this goal, we have developed multiple novel high-throughput CRISPR-based technologies for characterizing the function of putative gene regulatory elements by perturbing their activity in their endogenous, native context. We have coupled these methods with single-cell RNA-seq to identify the target gene(s) for each regulatory element. We have also developed dCas9 effector mice to characterize elements in their natural in vivo context. In addition, we have developed population-based high-throughput reporter assays (POP-STARR) to characterize the impact of noncoding genetic variation across the entire genome. The objective of this proposal is to apply and share our compendium of complementary, robust, scaleable, and well-characterized methods by working collaboratively to support the IGVF Consortium goals of understanding how genomes and genomic variation function and orchestrate complex phenotypes. Our track record in developing, applying, and sharing these high-throughput characterization methods, as well as providing access to all data, supports that we will be successful in accomplishing our objective via the following specific aims: Aim 1. Characterize all gene regulatory elements essential for cell survival. Aim 2. Characterize all gene regulatory elements essential to cell lineage specification. Aim 3. Characterize all gene regulatory elements in select eQTL regions. Aim 4. Characterize all non- coding elements essential to tissue homeostasis in a mouse model. We will make all data immediately available, as well as share comprehensive protocols, reagents, and analysis tools to the scientific community. Together, the diverse approaches of this Characterization Center will lead to transformative progress in understanding the role of regulatory elements and noncoding variants across many diverse phenotypes.