Hepatocyte nuclear factor 4α (HNF4α/NR2A1) is a highly conserved member of the nuclear receptor superfamily of ligand-dependent transcription factors which plays a crucial role in the early development of the liver, intestine/colon, pancreas and kidney as well as maintaining metabolic and organismal homeostasis. Based on usage of alternative promoters (P1 and P2) and alternate splicing, it is estimated that nine different splice variants of HNF4α exist. The P1- and P2-driven variants differ by 16 amino acids in their N-terminal amino terminal (AB) domain that encodes an activation function (AF-1). It has been shown previously that expression of P1-driven HNF4α protein is decreased in different human cancers such as hepatocellular, gastric, renal and colorectal carcinomas, while P2-driven HNF4α is either unchanged or up regulated. HNF4α is known to play an important role during early embryonic colon development where it regulates genes related to epithelial function. In addition, HNF4α has been shown to function in protecting the epithelium during experimental colitis in young adult mice. The functional consequences of the up regulation of the P2-driven HNF4α in human cancer are not known.

In Chapter 2 a potential link between levels of specific splice variants of HNF4α and miRNA 21, an oncogenic miRNA is investigated. miR-21 has been implicated in the development, growth, progression and metastasis of different types of cancers including colorectal, hepatocellular and pancreatic cancers. A series of in vitro assays using human colorectal cancer cells are performed, miRNA-21 levels is detected using stem loop RT-PCR. The results show that ectopic expression of P2-driven HNF4α preferentially regulates miRNA 21 in human embryonic kidney and colon cancer cell lines that do not express endogenous HNF4α. Knockdown of P2-driven HNF4α in a colon cancer cell line that expresses endogenous HNF4α, results in preferential decrease of miRNA 21 levels in a different human colorectal cancer cells (Caco-2). In order to elucidate the potential functional significance of miRNA-21 regulation by HNF4α, both P1- and P2-driven HNF4α were knocked down using siRNA, This results in downregulation of miRNA 21 as well as reduction in cell number. Ectopic expression of a miRNA 21 mimic in the HNF4α knockdown cells partially restored the cell count. The molecular mechanism by which HNF4α regulates miRNA 21 expression is explored in experiments that show that HNF4α binds the promoter region of miRNA 21 and that P2-driven HNF4α preferentially activates the human miRNA 21 promoter in a transient transfection system. Taken together, these observations lead us to propose that P2-driven HNF4α preferentially activates miRNA 21, an oncogenic miRNA, thus suggesting a functional significance of P2-driven HNF4α upregulation in colon cancer.

In Chapter 3, the potential functional role of isoform-specific HNF4α and miRNA 21, in a mouse model of chemical (DSS) induced colitis is investigated in which colitis is induced by the addition of DSS in the drinking water. Mortality, rectal bleeding, colon length, and spleen/body weight ratios were all differentially affected in both male and female transgenic mice (expresses only P2- driven HNF4α) compared to WT indicating an enhanced susceptibility of the transgenic mice to DSS. Immunoblotting and stem loop RT PCR performed on colonic tissue isolated from the DSS-treated mice showed that the expression of both HNF4α and miRNA 21 correlates with better recovery from DSS-induced colitis. These findings suggest that HNF4α-mediated regulation of miRNA 21 may play an important role in colitis and could be used as a marker for improved prognosis of colitis.

In Chapter 4, a potential link between specific splice variants of HNF4α and miRNA 194-2 are explored. miR-194-2 has been implicated in intestinal epithelial differentiation, cell invasion and inhibition of Epithelial-Mesenchymal-Transition (EMT). Ectopic expression of P2-driven human HNF4α resulted in a preferential

upregulation of miRNA 194-2 in human embryonic kidney and colon cancer cell lines. miRNA 194-2 levels were also found to be elevated in the distal colon of transgenic mice that expresses only P2-driven HNF4α. These results suggest that miRNA-194-2, like miRNA-21, may be preferentially regulated by P2-driven HNF4α. Further investigation is required to address the potential functional significance of this regulation.

Finally, in Chapter 5 the implications of the results of Chapters 2, 3, and 4 are discussed in detail. The importance of the role of HNF4α in general regulating these two microRNAs (miR-21 and miR 194-2) is considered as well as the relevance of the differential expression by the P1- and P2-driven isoforms. Finally, the role of these microRNAs and HNF4α splice variants are discussed in the broader context of the function of the colon in both the normal and diseased states.

Hepatocyte Nuclear Factor 4α (HNF4α), a master regulator of liver-specific gene expression, is regulated by two promoters (P1 and P2) which drive expression of two groups of HNF4α isoforms. HNF4α is a known regulator of gluconeogenesis and is mutated in maturity onset diabetes of the young one (MODY1). Moreover, HNF4α is heavily involved in regulating both lipid and glucose metabolism in the liver. Conventionally, it was thought that P1-HNF4α, but not P2-HNF4α, is expressed in the normal adult liver, while P1-HNF4α is downregulated and P2-HNF4α is upregulated in fetal liver as well as in liver cancer. This dissertation explores a previously undescribed role for P2-HNF4α in the normal adult mouse liver – one involved in the diurnal variations of lipid and carbohydrate metabolism. More specifically, P1-HNF4α appears to be a major driver of gluconeogenesis while P2-HNF4α is a driver of ketogenesis. We utilize P1-HNF4α-expressing (α1HMZ) and P2-HNF4α-expressing exon swap mice (α7HMZ) to determine the physiological function of the HNF4α isoforms in the switch between gluconeogenesis and ketogenesis, and to characterize the impact of sex on those functions. Additionally, we use AICAR treatment to explore the effect of AMP-Activated Protein Kinase (AMPK) activation in these mice, as AMPK, an energy-sensing kinase, has been shown to phosphorylate P1-HNF4α in vitro. Finally, given the fact metabolism in general can heavily affect the aging process and more specifically that ketone bodies serve as a source of fuel for the brain, and have been implicated in prevention of neurological diseases, such as dementia, we explore the effect of differential HNF4α expression in aging mice. The compelling data showing HNF4α isoforms are involved in the switch between gluconeogenesis and ketogenesis, a basic metabolic process that occurs on a daily basis, illuminate not only the molecular mechanism underlying the switch but also how that mechanism is impacted by sex. These studies have the potential to impact our understanding of numerous metabolic diseases, including diabetes, obesity, fatty liver disease and cancer.

Cellular proliferation and differentiation are critical events in normal development and cancer. Despite decades of research, much remains to be learned about how cells transition between the two states. To decipher one aspect of this switch, we focused on the transcription factor, Hepatocyte Nuclear Factor (HNF) 4 alpha. HNF4a is a nuclear receptor that is important in development and in maintaining the homoeostasis of the adult liver and colon. There are multiple isoforms of the HNF4a that are generated by alterative promoter (P1 and P2) usage and 3' splicing events in different tissues. Both P1 and P2-HNF4a isoforms are expressed in the adult colon, while the P1-HNF4a is expressed in the adult liver. Others have shown that P1-HNF4a is downregulated in cancer, whereas P2-HNF4a is upregulated in hepatocellular carcinoma and colorectal cancer. This would suggest that P1-HNF4a is tumor suppressive, while P2-HNF4a may act as an oncogene although a mechanism has not been elucidated. One potential mechanism could be through a differential interplay with the Wnt/beta-catenin/TCF pathway, which is activated in many cancers including liver and colon. To determine whether HNF4a and TCF4 cross-talk to control the switch between proliferation and differentiation and to elucidate the role of the HNF4a isoforms in early development and cancer, we generated Tet-On inducible systems that express either HNF4a2 (P1) or HNF4a8 (P2) under the control of doxycycline in human colon cancer and mouse embryonic stem cells. We characterized the lines to look for morphological and functional differences between the isoforms. We performed RNA-Seq and ChIP-Seq on the HCT116 inducible lines to identify any changes in gene expression and regulation and compared the HNF4a2 and HNF4a8 lines to determine functional differences. Although we found some functional redundancy, there are discrete differences. HNF4a2 suppresses tumor growth, inhibits cellular proliferation, and competes with TCF4 for regulation of target genes more effectively than HNF4a8. Similarly, we also found that HNF4a2 decreases cell numbers more than HNF4a8 in the mES inducible lines. Our findings provide insight into the distinct role of the P1- and P2-HNF4a in cancer and normal development.

The use of next-generation sequencing technology has been extremely useful for genome-wide analysis in many different organisms. For example, transcriptomic analysis based on RNA-seq can identify all changes in gene expression in a given tissue in a single experiment. Other types of experiments, such as ChIP-seq, allow mapping of transcription factor binding sites across the genome. Much of the work in this dissertation focuses on the transcription factor hepatocyte nuclear factor 4a (HNF4a) which is a member of the nuclear receptor superfamily of ligand-dependent transcription factors and abundantly expressed in the liver and intestines. HNF4a has been linked to several diseases including diabetes, liver and colon cancer, inflammatory bowel disease and others. The expression of HNF4a is driven by two promoters (P1 and P2) which result in the expression of 12 different isoforms. Additionally, HNF4a has as its endogenous ligand an essential fatty acid (linoleic acid, LA) that must be obtained from the diet and is known to play a role in gluconeogenesis during fasting. Therefore, we examine fasted vs fed conditions as well as high fat diets with different amounts of LA on gene expression in the liver and intestines, respectively. Finally, HNF4a target genes have been extensively studied and mutations in HNF4a binding sites in regulatory regions of select target genes associated with disease have been identified. However, to date, there has been no systematic approach to identify variants in HNF4a binding sites of target genes on a genome-wide scale.In Chapter 1 of this dissertation, we provide an introduction to the biological and research problems addressed in the remainder of the dissertation and review the relevant literature. In Chapter 2, we analyze transcriptomic (RNA-seq) data from male mice fed three different high fat diets (HFDs) with different amounts of LA from four different tissues across the mouse intestinal tract. We found that different portions of the intestines have unique gene profiles, especially in terms of nuclear receptor signaling, xenobiotic and drug metabolism, and intestinal epithelial barrier function. We found that different types of HFDs impact gene expression in the intestinal tract in different ways, including genes linked to diseases such as inflammatory bowel disease and colon cancer. A network analysis revealed that mice fed the HFDs had altered expression of intestinal genes involved in crucial body functions such as the immune system and the intestinal microbiome which could facilitate bacterial and viral infections such as SARS-CoV-2. To our knowledge, this is the most comprehensive dataset including multiple diets and multiple parts of the intestines and should provide an excellent resource for the scientific community for some time to come. In Chapter 3, we compared the impact of fasting and HNF4a isoforms on liver gene expression and alternative splicing, comparing and contrasting two RNA-seq datasets from male and female mice. We also compared wild type mice (express predominantly P1-HNF4a isoforms) to exon swap mice (a7HMZ) which express only the P2 isoforms of HNF4a which are typically not expressed in the adult liver. We found that a 12-hour fast has a significant effect on alternative splicing and that the P2-HNF4a isoforms seem to play a role. Interestingly, there was a different pattern of alternative splicing in female livers. In Chapter 4, we analyze the impact of single nucleotide polymorphisms (SNPs) on HNF4a binding sites using publicly available datasets of HNF4a ChIP-seq (chromatin immunoprecipitation) from human liver and the liver cancer cell line HepG2 which expresses HNF4a and exhibits a hepatocyte phenotype. For this purpose, we trained a support vector machine (SVM) to predict binding affinity scores based on DNA sequences known to bind HNF4a from protein binding microarray (PBM) data. This model allowed us to identify 10 putative affinity altering SNPs (aaSNPs) in the human liver HNF4a ChIP-seq and six in the HepG2 ChIP-seq that impact the binding of HNF4a to the chromatin. Additionally, some of these identified SNPs were found in genes related to cancer, suggesting a potential role in personalized medicine. These results present a proof concept of affinity disruption in binding between HNF4a and some target genes in a genome-wide analysis. Finally, in Chapter 5, we provide an example of how transcriptomic data could be used to generate new biological hypotheses and test them. Additionally, we provide future directions for Chapter 3 and Chapter 4.

This thesis describes the investigation of binding rules and DNA binding sequences for several transcription factors (TFs). We develop Protein Binding Microarrays (PBMs) to study the interactions between TFs and in vitro and we use a support vector machine (SVM) algorithm to capture these interactions in silico. We then apply this methodology to study the binding of TFs to promoters and repetitive sequences in a genomewide fashion.

In Chapter 2, we thoroughly investigate HNF4α DNA binding interactions using PBMs. We investigate binding specificities for various isoforms and species of HNF4α. We then use PBMs to rank ~ 4,000 HNF4α binding sequences in order of binding affinity. Using this training set we identify/predict novel HNF4α binding sequences and rules, and from these rules we generate a model for HNF4α binding. We then use this large dataset, in combination with ChIP-on-chip and RNAi followed by an expression profiling to identify hundreds of novel HNF4α direct target genes.

In Chapter 3, we identify HNF4α association with Alu repeats, a novel finding. We investigate HNF4α binding to Alu sequences in in vitro and in vivo in the promoters of HNF4alpha-regulated genes, and thus reveal a novel association between HNF4α and Alu repeats.

Finally in Chapter 4, we leverage the PBM technology to investigate the binding properties of transcription factors COUP-TF2 and TCF-1. We identify many sequences that bind both HNF4αand TCF-1 and those bind both HNF4α and COUP-TF2. This finding suggests competition between these TFs on the promoters of their target genes. Additionally, we investigate the effect of coregulator PGCα and the effect of the endogenous ligand linoleic acid HNF4α DNA binding.

This study significantly advances our knowledge of binding sequences, binding motifs, target genes, and transcriptional regulation for several transcription factors, HNF4α, COUP-TF2 and TCF-1. It also sheds light on evolution of HNF4α binding sequences through Alu repetitive elements. It suggests a strong framework for comprehensive investigation of transcriptional regulation in mammalian systems for other transcription factors.

Nuclear receptors (NRs) are ligand-sensitive transcription factors that regulate a wide array of biological processes including development, metabolism, and circadian rhythms. All NRs share a common protein structure, including highly conserved DNA binding domains and a highly variable N-terminal A/B domain, and are very popular drug targets. To better understand the role of alternative A/B domains between NR isoforms and the impact of genetic variation on gene expression in the liver, we employed two experimental approaches. The NR hepatocyte nuclear factor 4α (HNF4α), a master regulator of liver-specific gene expression, is regulated by two promoters (P1 and P2) in the liver resulting in proteins with different A/B domains. P1-HNF4α is expressed in fetal and normal adult liver while P2-HNF4α is expressed only in the fetal liver and in liver cancer. We compared wildtype mice, which express only HNF4α1 (P1) in the adult liver, to exon-swap mice that express only HNF4α7 (P2) for global changes in gene expression (RNA-seq), chromatin binding (ChIP-seq), and unique protein interactions (RIME). The results show that P1- and P2-HNF4α isoforms differentially regulate hundreds of transcripts in the adult liver, including the NR CAR (Nr1i3), and may be recruited differentially to non-HNF4α binding sites by unique protein interactions. They also exhibit altered metabolic pathways, especially cytochrome P450 (Cyp) genes. All told, the results show that changes in just 16-30 amino acids in the AF-1 region of an NR can have profound effects on gene expression. Utilizing protein binding microarrays (PBM), we can measure the DNA binding affinity of a given NR against both alleles of 125,000 genetic variants in a single experiment to probe for affinity altering SNPs (aaSNPs). By mining SNPs from ChIP-seq peaks and eQTLs from the GTEx project, we have identified thousands of aaSNPs, hundreds of which show significant correlation to changes in gene expression within their regulatory network. Analysis of aaSNPs from GWAS studies associated with Alzheimer’s disease identified a large number of genetic variants that can alter the DNA binding affinity of PPARɣ in the APOE locus. Additionally, we show the power of the PBMs to validate many aaSNPs derived from in vivo analysis and suggest a role for the PBM technology in characterizing how genetic diversity may play a role in personalized medicine.

Lactation brings about numerous maternal physiological and metabolic changes in order to sustain viable offspring. The liver plays a role in meeting the high energetic and metabolic needs of lactation as it is a key player in various biological processes that include lipid metabolism and homeostasis. A growing body of evidence suggests that the metabolic and physiological changes that lactation brings to the nursing mother results in a maternal profile that lowers the incidence of metabolic diseases. In this study, we examine in vivo and gene expression profiling data of fasted, age-matched virgin (V), non-lactating (NL), and lactating (L) C57Bl/6 mice in an attempt to understand the role the maternal liver plays in lactation in a global context. Our model allows us to examine the long term effects of lactation, since both the lactating and non-lactating mice have previously nursed pups. In vivo results highlight maternal adaptations to lactation that include rapid weight loss, increased liver weight, increased plasma triglyceride levels, and decreased hepatic triglyceride storage. Gene expression profiling reveals molecular signatures of different classes of genes that may explain the resulting maternal adaptations to lactation. Our model suggests a highly metabolically active liver that plays a role in supplying the energy that the lactating mammary gland needs, by synthesizing triglycerides and secreting them into the blood stream, thus making them available to the lactating mammary gland.