Senescence stimuli activate multiple tumor suppressor pathways to initiate cycle arrest and a differentiation program characteristic of senescent cells. We performed a two-stage, gain-of function screen to select for the genes whose enhanced expression can bypass replicative senescence. We uncovered multiple genes known to be involved in p53 and Rb regulation, ATM regulation and two components of the CST complex involved in preventing telomere erosion and additional genes such as REST and FOXO4 that have been implicated in aging. Among the new genes now implicated in senescence we identified DLX2, a Homeobox transcription factor that has been shown to be required for tumor growth, metastasis and associates with poor cancer prognosis. Growth analysis showed that DLX2 expression led to increased cellular replicative lifespan. We found that DLX2 expression inhibited p53 activation, and DLX2 reduced the protein level of upstream activator kinases ATM and DNA-PK. Our data suggest that DLX2 expression reduces the protein components of the TTI1/TTI2/TEL2 complex, a key complex required for the proper folding and stabilization of ATM, DNA-PK and other members of the PIKK family. Over-expression of DLX2 exhibit mutual exclusivity with p53 alteration in cancer patients, suggesting DLX2 may attenuate the p53 pathway during tumor formation. Our functional screen identified novel players that may promote tumorigenesis by regulating the ATM-p53 pathway and senescence. / Medical Sciences

Case-control genetic sequencing studies are increasingly being conducted to identify rare variants associated with complex diseases. Oftentimes, these studies collect a variety of secondary traits--quantitative and qualitative traits besides the case-control disease status. Reusing the data and studying the association between rare variants and secondary phenotypes provide an attractive and cost effective approach that can lead to discovery of new genetic associations. In Chapter 1, we carry out an extensive investigation of the validity of ad hoc methods, which are simple, computationally efficient methods frequently applied in practice to study the association between secondary phenotypes and single common genetic variants. Though other researchers have investigated the same problem, we make two key contributions to existing literature. First, we show that in taking an ad hoc approach, it may be desirable to adjust for covariates that affect the primary disease in the secondary phenotype model, even though these covariates are not necessarily associated with the secondary phenotype in the population. Second, we show that when the disease is rare, ad hoc methods can lead to severely biased estimation and inference if the true disease model follows a non-logistic model such as the probit model. Spurious associations can be avoided by including interaction terms in the fitted regression model. Our results are justified theoretically and via simulations, and illustrated by a genome-wide association study of smoking using a lung cancer case-control study. In Chapter 2, we consider the problem of testing associations between secondary phenotypes and sets of rare genetic variants. We show that popular region-based methods such as the burden test and the sequence kernel association test (SKAT) can only be applied under the same conditions as those applicable to ad hoc methods (Chapter 1). For a more robust alternative, we propose an inverse-probability-weighted version of the optimal SKAT (SKAT-O) to account for unequal sampling of cases and controls. As an extension of SKAT-O, our approach is data adaptive and includes the weighted burden test and weighted SKAT as special cases. In addition to weighting individuals to account for the biased sampling, we can also consider weighting the variants in SKAT-O. Decreasing the weight of non-causal variants and increasing the weight of causal variants can improve power. However, since researchers do not know which variants are actually causal, it is common practice to weight genetic variants as a function of their minor allele frequencies. This is motivated by the belief that rarer variants are more likely to have larger effects. In Chapter 3, we propose a new unsupervised statistical framework for predicting the functional status of genetic variants. Compared to existing methods, the proposed algorithm integrates a diverse set of annotations---which are partitioned beforehand into multiple groups by the user---and predicts the functional status for each group, taking into account within- and between-group correlations. We demonstrate the advantages of the algorithm through application to real annotation data and conclude with future directions. / Biostatistics

During B cell development, RAG endonuclease cleaves immunoglobulin heavy chain (IgH) V, D, and J gene segments and orchestrates their fusion as deletional events that assemble a V(D)J exon in the same transcriptional orientation as adjacent Cμ constant region exons. In mice, six additional sets of constant region exons (CHs) lie 100–200 kilobases downstream in the same transcriptional orientation as V(D)J and Cμ exons. Long repetitive switch (S) regions precede Cμ and downstream CHs. In mature B cells, class switch recombination (CSR) generates different antibody classes by replacing Cμ with a downstream CH. Activation-induced cytidine deaminase (AID) initiates CSR by promoting deoxycytidine deamination lesions within Sμ and a downstream acceptor S-region; these lesions are converted into DNA double-strand breaks (DSBs) by general DNA repair factors which are then joined by end-joining pathways. Productive CSR must occur in a deletional orientation by joining the upstream end of an Sμ DSB to the downstream end of an acceptor S-region DSB. However, the relative frequency of deletional to inversional CSR junctions has not been measured. Thus, whether orientation-specific joining is a programmed mechanistic feature of CSR as it is for V(D)J recombination and, if so, how this is achieved is unknown. To address this question, we adapt high-throughput genome-wide translocation sequencing (HTGTS) into a highly sensitive DSB end-joining assay and apply it to endogenous AID-initiated S-region in mouse B cells. We show that CSR is programmed to occur in a productive deletional orientation and does so via an unprecedented mechanism that involves in cis IgH organizational features in combination with frequent S-region DSBs initiated by AID. We further implicate ATM-kinase-dependent DSB-response (DSBR) factors including histone variant H2AX, 53BP1 and its associated effector protein Rif1 in enforcing this mechanism. We go on to use HTGTS to study influence of different DSBR factor deficiencies on the structure of CSR junctions between AID-initiated DSBs in the 5' portion of the donor Sμ region to those across the length of downstream acceptor S regions. Based on analyses of thousands of switch junctions, we find that absence of DSBR factors leads to varying increases in micro-homology (MH)-mediated junctions, with 53BP1-deficiency having the greatest increase. However, while translocation junctions between Cas-9/gRNA-induced DSB in c-myc to AID-initiated S region DSBs in ATM- or 53BP1-deficient B cells show similar biases in MH-usage to those observed in the context CSR junctions, translocation junctions to other general DSBs genome-wide had no MH-usage increase in ATM-deficient cells and only a modest increase in 53BP1-deficient cells. We discuss these findings with respect to potential roles of AID-initiated DSBs in S regions to be especially prone to MH-usage potentially due to their increased resection along with their highly repetitive nature that provides abundant micro-homologous sequence. / Medical Sciences

Despite nearly complete understanding of the genetics of the β-hemoglobinopathies for several decades, definitive treatment options have lagged behind. Fetal hemoglobin (HbF) reinduction represents a “silver bullet” for therapy of the β-globin disorders. Recent development of the clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 nuclease system has allowed for facile manipulation of the genome for the study of genes and genetic elements. Here we developed CRISPR/Cas9-based methodology to reliably engender targeted genomic deletions ranging from 1.3 kilobases to over 1 megabase, which suggested an inverse relationship between deletion size and deletion frequency. Targeted deletion methods and Cas9-mediated in situ saturating mutagenesis were applied to the enhancer of the HbF repressor BCL11A, which revealed discrete vulnerabilities. This finding is consistent with emerging evidence in the field that large enhancers are comprised of constituent parts with some harboring the majority of the activity. The identified “Achilles heel” of the enhancer represents a promising therapeutic target. We further enhanced the resolution of the in situ saturating mutagenesis technique by using multiple Cas9 nucleases and variant-aware library design to identify functional sequences within the HBS1L-MYB intergenic region, a locus associated with elevated HbF levels. These data demonstrate the robustness of CRISPR/Cas9 mediated in situ saturating mutagenesis and targeted deletion to interrogate functional sequence within regulatory DNA. Harnessing the power of genome editing may usher in a second generation form of gene therapy for the β-globin disorders. / Medical Sciences

The monoaminergic neuromodulatory systems (including the noraderenergic, serotonergic, and dopaminergic neuronal systems) have long been thought of as the custodians of mental health. Dysfunction amongst these neural populations is associated with psychiatric disease, which is often treated with pharmacotherapeutic agents targeting these monoaminergic systems. Increasingly, such therapies manipulate these neuronal populations in tandem to achieve increased therapeutic efficacy. It has therefore become important to understand the putative interactions amongst these neural systems. In this thesis, I explore the interactions between the serotonergic and dopaminergic neuronal systems, investigating both the effects of dopamine on serotonergic neurons as well as those of serotonin on the midbrain dopaminergic system. Chapter 1 reviews evidence for such interactions, with a focus on the capacity of serotonergic neurons to respond to dopaminergic stimulation. It further introduces two subtypes of serotonergic neurons characterized by expression of dopamine receptor-cre driver lines, Drd1a-cre and Drd2-cre, and demonstrates the necessity of these serotonergic subtypes in the modulation of male aggressive behavior. Chapters 2 through 4 employ intersectional genetic approaches to characterize the anatomical distribution, developmental onset, dopamine responsiveness, patterns of gene expression, and hodology of these two serotonergic neuronal subtypes. These chapters demonstrate that there are subpopulations of serotonergic neurons that express receptors for dopamine and are capable of responding to dopaminergic stimulation. This work achieves an unprecedented resolution in cell subtype specificity coupled with molecular, anatomical, and pharmacological characterization, thus resolving this dopamine-serotonin interplay without the confounding ambiguities and limitations surrounding prior studies in the field. Chapter 5 explores the inverse relationship, investigating the effects of constitutive serotonergic neuron activity on the development of midbrain dopaminergic hodology. Collectively, this body of work lays a foundation upon which to build our understanding of transmonoaminergic interaction at the level of the cell, the circuit, and the organism. / Medical Sciences

The human brain can be affected by many diseases during its development, and human genetics studies can illuminate the genetic etiology of these diseases, which is critical for understanding disease pathophysiology, improving diagnoses, and developing effective therapies. In this thesis, I study genetic factors underlying neurodevelopmental disorders, focusing on the intractable epilepsy syndromes focal cortical dysplasia (FCD) and hemimegalencephaly (HME) and on autism spectrum disorder (ASD). My work investigates the role of both germline mutations, which are present in all cells of an affected individual, and somatic, or post-zygotic, mutations, which are present in only a subset of cells of an affected individual and frequently are undetectable in blood DNA. FCD and HME are associated with intractable epilepsy, and show a small cortical region (FCD) or an entire cortical hemisphere (HME) that is radiographically and histologically abnormal, suggesting causation by somatic mutations. Our work and others have implicated mammalian target of rapamycin (mTOR) pathway mutations in these disorders. Studying 83 patients with FCD or HME using deep sequencing, we show that FCD and HME are caused by mutations that activate the mTOR pathway, and we expand the allelic and locus heterogeneity of these disorders, implicating mutations in DEPDC5 in HME and sporadic FCD and in TSC2 in HME. Overall, we identify pathogenic mutations, mostly somatic, in 22 patients with FCD or HME, representing almost half for whom brain tissue was available. We show for both FCD and HME that indistinguishable radiologic and histologic lesions can be caused by multiple genetic mechanisms, including somatic activating point mutations in AKT3, MTOR, and PIK3CA and germline loss-of-function mutations in DEPDC5 and TSC2, usually with somatic loss-of-function mutations in the second allele of the same gene limited to the lesion. We also show that mutations in the same gene, and in some cases the same mutant allele, can cause a spectrum of phenotypes, from FCD to HME to bilateral brain overgrowth, likely reflecting both the developmental time window and progenitor cell type in which the mutation occurred. Single cell studies show that the mutations are enriched in specific cell types, and how the cell types involved reflect the progenitor cell type in which the mutation occurred. Single nucleotide variants (SNVs), particularly loss-of-function mutations, are also significant contributors to the risk of autism spectrum disorder (ASD), which is characterized by deficits in social interaction and communication and restricted and repetitive behaviors, interests, or activities. Compared to FCD and HME, ASD is a relatively common neurodevelopmental disorder with greater genetic heterogeneity. We report the first deep sequencing study of 55 postmortem ASD brains for SNVs in 78 ASD candidate genes. Remarkably, even without parental samples, we find more ASD brains with mutations that are protein-altering, deleterious, or loss-of-function compared to controls, with recurrent deleterious mutations in well-known ASD genes like SCN2A, suggesting these mutations contribute to ASD risk. In six cases, the identified mutations and medical records suggest syndromic ASD diagnoses. Two ASD cases and one Fragile X premutation case show deleterious somatic point mutations in ASD genes, providing evidence that somatic mutations occur in ASD cases, and supporting a model in which a combination of germline and/or somatic mutations may contribute to ASD risk on a case-by-case basis. As for FCD and HME, our results suggest that ASD pathogenesis can involve interactions between somatic and germline mutations in some cases, emphasizing how the developmental history of the human brain modifies the germline genome. Taken together, these studies show that different combinations of germline and somatic mutations contribute to both rare and common neurodevelopmental disorders, and that the time and place a mutation occurs is critical for determining its consequences in the human brain. We also demonstrate the utility of deep sequencing to discover somatic mutations, which may contribute to many other neurodevelopmental and neuropsychiatric diseases. / Medical Sciences

Advances in DNA sequencing are creating new opportunities for studying the process of evolution. These measurements can be particularly useful for rapidly evolving microbial organisms, whose small size and fast generation times make them ideal for controlled laboratory experiments and for tracking replicate populations in vivo. However, the interpretation of this new source of data is complicated by the unique ways in which these large microbial populations evolve. The basic problem is that natural selection is forced to do too many things at once. Unlike the classical picture, where new mutations arise one-by-one, rapidly evolving populations often harbor many selected variants at the same time. When recombination is limited, selection cannot act on these mutations individually, but only on combinations of mutations that happen to arise on the same genetic background. These effects, known as clonal interference, create correlations along the genome that are difficult to disentangle. Existing population genetic models often neglect these effects, which leaves us at loss when interpreting data from these populations. In Chapters 2-5, we analyze the effects of clonal interference in a simple ``null model'' of microbial evolution. We focus on the simplest model that is consistent with two empirical observations: (1) many fitness-influencing mutations are created every generation and (2) mutations have a broad range of fitness effects. After analyzing the basic dynamics of this model, we obtain predictions for the substitution rates of individual mutations and the patterns of linked neutral diversity, and we show how these quantities depend on the population size, mutation and recombination rates, and the fitness effects of new mutations. In Chapters 6 and 7, we apply this null model to data from laboratory experiments in S. cerevisiae and E. coli. We develop a statistical framework to infer the underlying parameters (the fitness effects of new mutations), which allows us to quantify deviations from the model over longer evolutionary timescales. Finally, in Chapters 8 and 9, we investigate the behavior of the model when some of the parameters are allowed to evolve or change in time. / Physics

The organization of interphase nuclei has numerous consequences for gene expression and genome stability in eukaryotes. Here, we present studies of chromosome organization in Drosophila melanogaster over the course of the cell cycle. We report surprising observations suggesting that cohesin, a protein complex essential in mitosis for holding together the products of DNA replication known as sister chromatids, may not be required to keep sister chromatids in close proximity during interphase. These observations raise questions regarding the nature of cohesin-independent connections between chromatids, why the cell might have such connections in addition to those mediated by cohesin proteins, and how cohesin-independent mechanisms might contribute to other inter-chromosomal associations and nuclear organization throughout the cell cycle. A well-known feature of nuclear organization in Drosophila is the somatic pairing of maternal and paternal homologous chromosomes; we find that certain factors contributing to the pairing of homologs, specifically, condensin II and its regulators, also contribute to the organization of sister chromatids, findings which have interesting implications for the nature of pairing. Finally, we examine another type of inter-chromosomal interaction occurring in Drosophila nuclei, which is the nonhomologous clustering of centromeres, and identify candidate genes that regulate this organization. Overall, this work expands current knowledge on both homologous and heterologous inter-chromosomal interactions, and highlights the relationship between chromosome organization in interphase and in mitosis. / Medical Sciences

Chromatin plays a vital role in the implementation of gene expression programs. Several disparate groups of regulatory proteins alter chromatin state through post-translational modification of histone proteins, nucleosome remodeling, and higher order chromatin structure in order to affect gene expression. Several of these key groups, such as the Male-Specific Lethal complex and Polycomb Group have been well characterized in Drosophila. Yet aspects of their biology at the molecular level, such as the means by which they are faithfully targeted to regulated loci throughout the genome and the molecular mechanisms they employ to alter transcriptional state, still remain unexplained. In this dissertation I explore how identifying protein-protein interactions on chromatin reveals insights into these unanswered questions critical to chromatin biology. My results highlight the importance of balancing active and repressive chromatin states for the proper maintenance of gene expression. The Male-Specific Lethal complex is the dosage compensation complex in Drosophila, which upregulates gene expression on the male X chromosome approximately two-fold. The MSL complex catalyzes an acetyl mark which may create a uniquely permissive chromatin state to promote transcriptional elongation. A proteomic screen for MSL-interacting proteins identified UpSET, the Drosophila homolog of yeast SET3 and mammalian MLL5. Interestingly, SET3 and UpSET have been characterized to assemble into histone deacetylase complexes. I employed genetic, genomic, and proteomic techniques to assess whether UpSET plays a role in dosage compensation. UpSET appears to play a role in limiting the level of activation of the MSL complex. Surprisingly, UpSET appears to play a more important role in the maintenance of heterochromatin. The Polycomb Group is comprised of a well characterized set of developmental repressors. The PcG assembles into several multiprotein complexes to maintain the repressed state. The PcG is opposed by a group of activators known as the Trithorax group. Although the PcG and TrxG often appear to be recruited to the same genomic elements in different tissues, whether they might interact directly was not known. In a collaboration with Dr. Hyuckjoon Kang, I characterized the TrxG protein Female sterile (1) homeotic and found that it interacts specifically with PRC1. The data support a model that bivalency, a poised state observed in mammalian stem cells, may be critical, perhaps transiently, in the developing Drosophila embryo. The mechanism of coordination amongst the various PcG complexes on chromatin is not well understood. We also identified the Sex comb on midleg protein, a known member of the PcG, as a potential physical bridge between PRC1 and PRC2. In these sets of experiments, I have characterized instances of crosstalk between activating and repressing regulators which are critical for the proper maintenance of chromatin state. Perturbations of these interactions may lead to an imbalance of regulators on chromatin and aberrant transcriptional activity. These findings highlight the need for tuning gene expression state and suggest chromatin-based mechanisms by which this can be accomplished. / Biology, Molecular and Cellular

Nodal signaling is one of the principal players in the process of gastrulation, during which the primary germ layers (endoderm, mesoderm, and ectoderm) are formed and organized in their proper locations. During my PhD I studied how Smad2 and FoxH1 transcription factors regulate the expression of Nodal target genes, and the relationship between the chromatin state of target genes and their expression. In order to carry out in depth analysis of FoxH1 function I generated a complete mutant of this transcription factor and used deep sequencing to identify which genes FoxH1 regulates in zebrafish development. Using ChIP-Seq experiments, I also found the binding sites of FoxH1 and found that FoxH1 is capable of binding to genomic DNA in the absence of Nodal signaling. This finding suggests that it may act as a pioneer factor, preparing target genes for rapid activation when gastrulation starts. I also identified 54 direct FoxH1 target genes that are not Nodal-dependent, as well as 13 genes that are repressed, rather than activated, by FoxH1. To identify Smad2 binding sites, I carried out ChIP-Seq in embryos overexpressing Nodal signal Squint, thus detecting loci that bind Smad2 after exposure to high Nodal levels. I tested the identified Smad2-bound DNA elements for their gene regulatory potential, and discovered that they are sufficient to drive gene expression in a Nodal-dependent manner. I also identified 26 previously unpublished Nodal target genes, and 55 genes that are bound by Smad2 upon exposure to high, but not low levels of Nodal signaling. In the last chapter I describe the study of the interaction between Nodal signaling in early zebrafish development and chromatin marks. I found that exposure of embryonic cells to high levels of Nodal is associated with low levels of H3K27me3 and high levels of H3K4me3 marks on Nodal target genes, compared to unexposed cells. I also describe a Cas9-based system that we used to change H3K27me3 levels in a targeted manner, and tested it in a developing embryo on a Nodal-responsive fgf8a gene. Our results suggest that reduction of H3K27me3 mark on its own is not sufficient to affect the expression of this gene, and additional mechanisms are involved in target gene activation by Smad2. / Biology, Molecular and Cellular