Cells need to sense the environment in order to survive, in particular they need to detect nutrients which will provide different building blocks and energy for the cell. This task is complicated by the fact that there can be multiple sources for the same type of nutrient available for the cell. A classical example of how cells sense multiple signals is given by carbon catabolite repression in the budding yeast S. cerevisiae. In this model the preferred carbon source glucose represses the genes used to metabolize an alternative source such as galactose. This means that the preferred carbohydrate glucose is thought to inhibit the induction of galactose genes when above a threshold concentration. Instead, we show that galactose metabolic genes (GAL) induction depends on the ratio of galactose and glucose. Surprisingly, we find that a critical portion of information processing occurs upstream of the canonical components of the GAL pathway. We then explore how cells choose between different responses to the environment. Specifically, we set out to characterize the variability in the response to combinations of galactose and glucose between several natural yeast isolates. To elucidate the genetic basis of this phenotypic variation we use QTL mapping on these strains. Our study reveals that a signal transducer GAL3 plays a central role in establishing variation in GAL gene induction.Lastly, we focus on the control of transcription in the cell. Many promoters in the cell produce both a coding transcript and a divergent transcript. To identify mutants that affect transcriptional directionality we use a bidirectionalfluorescent protein reporter in the yeast nonessential gene deletion collection. We determine that chromatin assembly can regulate divergent transcription. Moreover, mutations in the chromatin assembly factor CAF-I can lead to genome wide derepression of nascent divergent transcription. / Systems Biology

The phosphatases PTEN and INPP4B are frequently deregulated in human cancer and have been proposed to act as tumor suppressor genes by coordinately antagonizing PI3K/AKT signaling. While the function of PTEN has been extensively studied, little is known about the underlying molecular mechanisms by which INPP4B exerts its tumor suppressive function. Additionally, its role in tumorigenesis in vivo has not been studied. Here, we show that a partial or complete loss of the phosphatase function of Inpp4b morphs benign thyroid adenoma lesions observed in Pten heterozygous mice into lethal and metastatic follicular-like thyroid cancer (FTC) (Chapter 2). Importantly, analysis of human thyroid cancer cell lines and specimens reveals that INPP4B expression is downregulated in FTC. Mechanistically, we have found that INPP4B, but not PTEN, is enriched in the early endosomes of thyroid cancer cells, where it blocks PI3K-C2 mediated AKT2 activation and in turn tumor proliferation and anchorage-independent cell growth. Taken together, these data identify INPP4B as a novel tumor suppressor in FTC oncogenesis and metastasis through localized regulation of PI3K/AKT pathway at the endosomes. Further, we present evidence that INPP4B downregulation cooperates with PTEN loss in prostate cancer progression and metastasis (Chapter 3). In vivo, a partial loss of Inpp4b cooperates with Pten haploinsufficiency to promote prostate tumorigenesis. In vitro, we have found that knockdown of INPP4B in cell lines increased their migratory and invasive properties. Overall, our studies have greatly increased our understanding of the molecular mechanisms of the tumor suppressive functions of INPP4B and provided in vivo evidence for the cooperation of Inpp4b with Pten haploinsufficiency in both thyroid and prostate tumorigenesis / Medical Sciences

Autoimmune diseases are chronic and debilitating conditions arising from abnormal immune responses directed against normal body tissues; they collectively affect the lives of 5-10% of the world population. These diseases often show familial clustering, suggesting strong genetic heritability. For many of autoimmune diseases, variation in the human leukocyte antigen (HLA) genes is the primary modulator of genetic risk. Recently, genome-wide association studies (GWAS) identified hundreds of genomic regions outside the HLA that harbor additional risk-conferring variants. The ultimate goal is to identify the precise causal variants and understand the mechanisms by which they lead to autoimmunity, which is challenged by complexities of the genome and the immune system. In this work, my colleagues and I developed and applied experimental and computational tools to reveal critical clues from multiple genetic and biological data types. First, we devised a statistical algorithm to identify the critical cell types involved in different autoimmune diseases. Two strongly heritable and common diseases, rheumatoid arthritis (RA) and type 1 diabetes (T1D), both involve the adaptive immune system, specifically the CD4+ T cells. We then conducted focused studies in CD4+ T cells using high-throughput genomic and proteomic technologies, and showed that immunological phenotypes and functions varied with genetic differences across individuals. To facilitate this study, we developed an automated computational tool to efficiently and reliably analyze the large-scale data. Finally, the HLA genes, which encode a family of highly variable antigen-recognition proteins, are the longest-known and strongest modulators of genetic risk in T1D. However, the extraordinary level of polymorphism and complex structure in the HLA region largely hindered precise localization and functional investigation of the causal mutations. We used dense-genotyping and robust statistical analyses to pinpoint the amino acid residue changes at a few key amino acid positions that explained the majority of disease risk within the HLA. The work presented in this dissertation revealed the specific immune cell populations, genetic variants, and cellular functions that affect RA, T1D, and other autoimmune diseases. Furthermore, it offers a rational framework, as well as powerful open-source computational tools, that can be applied in future functional genomic studies. / Medical Sciences

Hepatocellular carcinoma (HCC) is a global health problem with poor prognosis and limited therapeutic options. While the clinical risk factors for HCC are well described, the precise molecular and metabolic mechanisms contributing to malignant transformation remain largely unknown. Recently, the Hippo signaling pathway has been identified as a key regulator of cellular proliferation, organ size, and tumorigenesis in numerous tissues, including the liver. However, the metabolic impact of the pathway in supporting liver growth and tumorigenesis has not been studied. The zebrafish, Danio rerio, has successfully been applied as a model to investigate signaling pathways important in organ development to model liver development and cancer. Here, we utilize the zebrafish to investigate the functional and metabolic roles of the Hippo pathway in liver development and cancer in vivo. Using a transgenic zebrafish model with liver-specific activation of the transcriptional co-activator Yap, the downstream target of the Hippo pathway, we show Yap is functionally conserved in its ability to promote embryonic and adult hepatomegaly. These livers demonstrate signs of dysplasia and increased tumor susceptibility upon chemical carcinogen exposure. Using transcriptomic and metabolomic analysis, we discover that nitrogen metabolism is significantly altered in Yap-transgenic livers. Yap upregulates glutamine synthetase (Glul) expression leading to elevated steady-state levels of glutamine, which significantly contributes to its ability to enhance liver growth and de novo purine biosynthesis. To further probe the functional and metabolic role of Yap prior to liver outgrowth, we utilize yap knockout zebrafish and heat-shock inducible transgenic zebrafish that modulate Yap activity to examine early liver development. We show Yap is important for hepatoblast formation and expansion. Further, Yap modulates glucose uptake and glycolytic flux into de novo nucleotide synthesis. Overall, this dissertation reveals novel roles of Yap in cellular metabolism to support proliferation and growth by directing glucose into the building blocks of DNA in the context of development and cancer. / Medical Sciences

Complex information within biological systems is being uncovered at an unprecedented speed thanks to the rapid technical development of a wide variety of research tools, among which imaging and sequencing technologies are attracting big attention in recent years. Optical imaging enables the visualization of the spatial distribution of biomolecules at cellular level, allowing deeper understanding of the structure and dynamics of biological systems. Fluorescence microscopy has contributed greatly to our understanding of these processes, but it relies on the use of fluorescent labels or dyes. These labels may perturb the studied systems especially for imaging small molecules, and the photobleaching problem also limits the long-term biological dynamics observation within living cells. In the first part of this dissertation, we introduce the recent development of Stimulated Raman scattering (SRS) microscopy as a noninvasive imaging technique with superior sensitivity, molecular specificity at video-rate imaging speed. It has superseded coherent anti-Stokes Raman scattering (CARS) microscopy due to the absence of non-resonant background and automatic phase matching. However, SRS imaging has been mostly demonstrated for the visualization of lipid and protein with long vibrational wavenumbers. We extend the detectability of SRS imaging into the crowded fingerprint region with characteristic signatures of more biomolecules such as nucleic acids in live cells (Chapter 2), unsaturated lipid and aromatic amino acid in multiphasic food products (Chapter 3). Noninvasiveness of SRS imaging also brings new opportunities to biomedical applications and we demonstrate its feasibility as a potential pathology diagnostic tool by generating comparable image contrast as golden standard H&E staining in human brain frozen sections (Chapter 4). We further extend SRS imaging to real-time multiband detection using a novel modulation multiplex approach (Chapter 5). The rapid development of high throughput sequencing technologies has enabled whole genome and transcriptome wide analysis at faster speed and affordable cost, but a large number of cells are often still required for these analyses. However, cell-to-cell variation is significant and may carry important indication to the study of complex wiring in the nervous systems. In this second part of the dissertation, we explore the heterogeneity of retina using a recently developed single cell transcriptome amplification technique based on Multiple Annealing Looping Based Amplification Cycles (MALBAC), which is superior to other single cell techniques with its low amplification bias, high reproducibility rate and low dropout rate. We first classify different retinal cell populations (photoreceptor cells vs. retinal ganglion cells) and closely related subpopulations (different direction selective retinal ganglion cells) (Chapter6). We further study the molecular divergence of an unsolved ON-OFF retina circuit responsible for direction selectivity function. We show that the inhibitory interneurons responsible for this function can be classified into two clusters based on the single cell transcriptome data. This clustering result strongly correlates with the ON-OFF starburst amacrine cells (SACs) based on the immunostaining results of the identified differential genes. The newly reported differential genes can potentially be used as molecular markers for ON-OFF SACs with more validation underway (Chapter 7). These new findings open up more opportunities for the functional studies on the direction-selective circuit in retina. / Engineering and Applied Sciences - Applied Physics

Understanding how complex, adaptive behavior evolves is a major goal of biological research. Phenotypic differences between closely-related species often arise due to evolution by natural selection and can be a powerful resource for understanding biological diversity and its mechanistic underpinnings. In this dissertation, I capitalize on striking behavioral differences between two interfertile sister species of Peromyscus rodents. I pursue the proximate mechanisms underlying this behavioral adaptation by investigating both the ontogeny and genetics of innate differences in burrow construction behavior in Peromyscus polionotus and P. maniculatus. In Chapter 1, I compare the ontogeny of burrow construction behavior of Peromyscus polionotus and P. maniculatus across early development. I find that P. polionotus begins burrowing precociously (as early as 17 days of age) compared to P. maniculatus (27 days of age), despite P. polionotus being physically smaller and less active in a wheel running assay. Furthermore, juvenile P. polionotus constructed long burrows complete with species-specific escape tunnels. Interspecific cross-fostering did not alter the developmental trajectory of either species, indicating that these differences are innate. Moreover, F1 hybrids followed the behavioral ontogeny of P. polionotus, indicating that precocious burrow construction segregates in a P. polionotus-dominant manner. Finally, I show that a quantitative trait locus (QTL) associated with adult tunnel length in these species is predictive of precocious digging in recombinant F2 hybrids, demonstrating that either a single pleiotropic locus or a group of tightly-linked genes control behavioral differences across life stages in P. polionotus. In Chapter 2, I dissect the genetic architecture of this complex behavior in adult animals using an experimental cross. By introgressing the burrow architecture of P. polionotus into the genetic background of P. maniculatus, I analyze the underlying genetic architecture of differences in burrowing behavior, and show that escape tunnels are likely a threshold trait. Next, I use a novel image-based analysis to collect measurements of burrow shape and demonstrate the utility of a more rigorous measurement of extended phenotypes. Finally, in Chapter 3, I combine two forward-genetics approaches—QTL mapping and transcriptome analysis—to nominate specific candidate genes for the differences in burrowing behavior between P. polionotus and P. maniculatus. Using a large advanced backcross mapping population (n=751), I detect five QTL contributing to differences in burrow architecture between these species: three loci for entrance tunnel length variation, and two loci for escape tunnel length. In the transcriptome study, I focus on gene expression in F1 hybrids to detect allele-specific expression (ASE), as ASE in an F1 hybrid indicates cis-regulatory differences between the parental lineages. I find widespread bias favoring expression from the P. polionotus-allele in F1 hybrid brains, which may be a molecular reflection of P. polionotus-like burrowing behavior of hybrids. Finally, I use ASE to nominate candidate genes within the detected QTL regions, and find genes related to behavioral disorders, circadian rhythms, and activity patterns; these genes represent promising candidates for future functional studies. / Biology, Organismic and Evolutionary

The vertebrate central nervous system (CNS) is made up of a diverse array of cell types. The retina, an accessible part of the CNS with seven major cell types and more than sixty cell subtypes, is an excellent model system to study cell fate determination in the CNS. Previous retroviral lineage tracing experiments have demonstrated that retinal progenitor cells (RPCs) are multipotent, and give rise to clones of variable sizes and cell type compositions. Two models are proposed to explain this variability in clone size and composition: One model states that there are different subsets of RPCs, whose distinctive molecular profiles determine their daughter cell fates, while the other model argues that RPCs are equipotent and daughter cell types are determined by environmental cues and/or stochastic cellular processes. To test the hypothesis on distinct RPCs, we first asked whether we could identify specific RPC subsets that would produce specific daughter cell types. We made use of 10 molecular markers to mark specific RPC subsets, and traced individual subsets' daughter cell fates with Cre-recombinase fate mapping and retroviral lineage tracing. A novel RPC subset, which express the basic helix-loop-helix (bHLH) transcription factor Ngn2, were found to be heavily biased towards generating rod photoreceptors and amacrine cells in terminal divisions in the postnatal mouse retina. Next, we partitioned the postnatal RPC pool into different RPC subsets with molecular marker Ngn2 and Olig2, and probed individual subsets' responsiveness to misexpression of two sets of transcription factors, which are known to play important roles in retinal cell fate determination. We have shown that different RPC subsets respond differently to the same genetic perturbation, which is indicative of their distinctive intrinsic capabilities to generate certain daughter cell types. Together, we have shown that in the mouse retina, the RPC pool is composed of distinct RPC subsets, each of which have unique molecular profiles, give rise to specific daughter cell types, and respond differently to perturbations. This study provides new insight into cell fate determination in the retina, and may shed light on a more general mechanism of cell fate determination in a variety of systems. / Medical Sciences

The complex and sophisticated neocortical circuits that mediate higher-order brain functions are assembled from an extraordinary variety of neuronal subtypes, each with distinct morphologies, output connectivity, and electrophysiological properties. These diverse neuronal subtypes are generated under precise molecular regulation during neocortical development, and elucidating programs that govern their specification will be critical toward more deeply understanding the organization, evolution, and function of the cerebral cortex. In recent years, several key transcriptional controls over specification of projection neuron subtype and area identity have been identified, providing substantial insight into the ‘molecular logic’ underlying cortical development. This work increasingly supports a model in which individual regulators are embedded within modular, but highly interconnected, networks that gate sequential developmental decisions. There is an emerging understanding that molecular controls over subtype and area identity are intimately interrelated, and that specification of neuronal identity relies on sequential molecular decision points in progenitors and postmitotic neurons. Despite substantial progress, however, the basic framework of transcriptional controls over neocortical projection neuron development has not been fully defined. The precise molecular mechanisms by which recently identified regulators act in parallel, synergistically, or cross-repressively to progressively delineate subtype and area identity are not well understood. Moreover, very few downstream-regulated genes responsible for executing specific aspects of neuronal differentiation have been identified. In this dissertation, I investigate functions of transcription factor Ctip1 in projection neuron subtype and area development. Using complementary genetic, molecular, and anatomic labeling approaches in mice, I find that Ctip1 is a novel control over 1) deep-layer projection neuron identity, promoting specification of corticothalamic and callosal projection neurons at the expense of subcerebral projection neurons; and 2) acquisition of sensory area identity, including area-specific gene expression, output connectivity, and afferent sensory map formation. In addition, I have developed recombinase-based strategies for dual-population mosaic analysis, which have enabled rigorous examination of cell autonomy in projection neuron subtype and area specification, and will have broader applications in other fields. Lastly, I have characterized two transgenic mouse lines produced by the GENSAT project, Rbp4-Cre and Ntsr1-Cre, which I find to specifically mediate recombination in subcerebral and corticothalamic projection neurons, respectively. / Medical Sciences

A growing body of evidence suggests that genetic variants that alter gene expression are responsible for many phenotypic differences across individuals, particularly for the risk of developing common diseases. However, the molecular mechanisms that underlie the vast majority of associations between genetic variants and their phenotypes remain unknown. An important limiting factor is that genetic variants remain difficult to interpret, particularly in noncoding sequences. Developing truly systematic approaches for characterizing regulatory variants will require: (a) improved annotations for the genomic sequences that control gene expression, (b) a more complete understanding of the molecular mechanisms through which genetic variants, both coding and noncoding, can affect gene expression, and (c) better experimental tools for testing hypotheses about regulatory variants. In this dissertation, I present conceptual and methodological advances that directly contribute to each of these goals. A recurring theme in all of these developments is the statistical modeling of protein-DNA interactions and its integration with other data types. First, I describe enhancer-FACS-Seq, a high-throughput experimental approach for screening candidate enhancer sequences to test for in vivo, tissue-specific activity. Second, I present an integrative computational analysis of the in vivo binding of NF-kappaB, a key regulator of the immune system, yielding new insights into how genetic variants can affect NF-kappaB binding. Next, I describe the first comprehensive survey of coding variation in human transcription factors and what it reveals about additional sources of genetic variation that can affect gene expression. Finally, I present SIFTED, a statistical framework and web tool for the optimal design of TAL effectors, which have been used successfully in genome editing and can thus be used to test hypotheses about regulatory variants. Together, these developments help fulfill key needs in the quest to understand the molecular basis of human phenotypic variation. / Biophysics

Peptides and small proteins have essential physiological roles including metabolism (insulin), sleep (orexin), and stress (corticotropin-releasing hormone). Recent exploration of the human genome and proteome has revealed the existence of hundreds to thousands of short open reading frames (sORFs); however, the extent to which sORFs are translated into polypeptides is unknown. Inline with the current convention, a protein-coding short ORF is defined to be a small ORF or smORF; the protein product as a smORF-encoded polypeptide is called a SEP; and a sORF or smORF upstream from an open reading frame (i.e. in the 5’-UTR) is called an upstream ORF or uORF. The identification of smORFs and SEPs have prompted efforts determine the regulation and biological functions for these molecules. My thesis research focused on improving SEP discovery and the characterization of functional SEPs. The discovery of novel SEPs contributes to our understanding of composition of the human genome and proteome. My colleagues and I developed and utilized a proteogenomics strategy, which integrates genomics (RNA-Seq) with proteomics, to discover 86 novel human SEPs, the largest number of validated SEPs described at the time. Our findings indicated that SEPs are a large, unappreciated, peptide family. Moreover, our approach was far from optimized and we felt that there were likely many additional SEPs in the human genome. One goal of my thesis work was to improve the SEP discovery methodology to find more human SEPs. My efforts led to the discovery of an over 300 SEPs in cell lines and human tissue. A second goal of my thesis work was to identify and characterize functional SEPs. To do this I identified the SEPs that are most highly conserved throughout evolution with a program called PhyloCSF. PhyloCSF identifies which SEPs are evolutionary conserved to provide evidence for function. Seven out of the 300 plus SEPs had PhyloCSF scores that indicate that they have been conserved throughout evolution. These seven SEPs included an interesting SEP called SLC35A4-SEP that is generated from a uORF in the SLC35A4 gene. The SLC35A4-SEP had contained a transmembrane domain and analysis of cells revealed the mitochondrial localization of this SEP. Further characterization of SCL35A4 indicated that this polypeptide interacts with members of the ATP synthase complex. Though this interaction requires further validation the putative interactions suggested a role for SLC35A4-SEP in cellular energetics. Overexpression or knockout of SLC35A4-SEP affected cellular respiration. Ongoing work is testing to see if SLC35A4-SEP also effects mitochondrial membrane potential and structure of ATP-synthase. More generally, this approach highlights how I can begin to identify functional SEPs using a combination of computational and experimental methods. And my work on another functional SEP called NoBody indicates that this strategy is general. / Chemistry and Chemical Biology