Phf8 is a histone demethylase associated with human developmental disorders and cancer. Early studies of Phf8 in humans indicated that inactivating mutations in the gene were linked to syndromic intellectual disability with cleft lip and palate (Siderius syndrome). However, Phf8’s functional role in regulating mammalian development and behavior has not been demonstrated. In this thesis I present my findings on a knockout mouse model of Phf8, which I have generated and characterized to interrogate Phf8’s function in these processes in an in vivo mammalian system. Unexpectedly, I did not detect any gross physiological defects nor intellectual disability, but instead report here that mice null for Phf8 are resilient to anxiety and depression. I further characterized the molecular nature of Phf8’s role regulating mammalian behavior by performing RNA-seq on regions in the brain key to mediating anxiety disorders and depression. Here I find evidence that Phf8 directly regulates multiple serotonin receptors in the prefrontal cortex, which have a long standing link to anxiety disorders and depression, suggesting a likely mechanism for resiliency. In addition to its proposed function in behavior and development, multiple studies have implicated Phf8 cancer. It has been suggested that Phf8 can behave as an oncogene, notably in the case of T-ALL, where it cooperates with the Notch pathway and is required to drive tumor proliferation in vitro and in xenograft models. Within this thesis I will also present my findings on Phf8’s function in the context of an in vivo model of T-ALL. I do not observe a requirement for Phf8 in T-ALL nor does its loss seem to at all impair the progression of T-ALL driven by a constitutively active Notch1. However, I do observe a subtle defect in T-cell development, which is likely consistent with its function in effecting Notch signaling. The data within this thesis represent the first characterization of a mammalian knockout model of Phf8 and describe a novel role for this gene in the regulation of anxiety and depression. / Medical Sciences

Microbial metabolic production is an attractive alternative to traditional chemical synthesis for a wide array of commercially relevant molecules. Coaxing microbes to produce a target chemical efficiently often requires substantial modification of host cell metabolism, which necessitates searching a vast genetic space of enzyme genes and expression levels. Millions of pathway designs can now be built, but identifying the most productive cells remains low throughput. The ability to detect and report on the presence of any arbitrary target molecule within individual cells would transform the field of metabolic engineering. To this end, we developed strains of E. coli that survive an antibiotic challenge only in the presence of a specific small molecule, by regulating resistance gene expression via transcription factors responsive to sugars, alkanes, macrolides, flavonoids, vitamins or other molecules. Using two of these whole cell biosensors, responsive to glucaric acid or naringenin, we evolved respective biosynthetic pathways for each compound toward higher production. We used oligonucleotide-mediated genomic editing to simultaneously target up to 20 enzyme genes for expression modulation or knockout, creating billions of unique strains. Demonstrating the first example of iterative, whole-pathway engineering via a metabolite biosensor, we discovered E. coli strains that had increased production of naringenin by 36 times, or glucaric acid by 22 times. However, for many target molecules, especially those that are synthetic, no natural biosensor may exist. We developed a platform to engineer natural allosteric transcription factors with specificity to new inducer molecules. We computationally design for binding, synthesize and clone in multiplex thousands of specified sequences, and use a bidirectional screen to identify new responsive variants that retain allostery. We demonstrate by generating E. coli LacI variants responsive to gentiobiose, fucose, lactitol and sucralose. We uncovered significant plasticity in the ligand recognition of LacI, which may be a hallmark of allosteric transcription factors. Our method relies only on protein structure and operator DNA sequence, making it applicable to many other proteins. These methods together advance the ability to engineer microbial biosynthesis of any target molecule using evolution. Additionally, designer transcription factors can enable broad applications from dynamic metabolic control to cell biology. / Medical Sciences

Targeted genome editing holds tremendous promise for permanent correction of many genetic diseases. The recently developed CRISPR/Cas9 genome-editing tool exhibits facile programmability and robust gene-editing efficiency, and has been applied in cell cultures and animal tissues. However, multi-organ gene-editing in live mammals has not been examined or achieved. This study demonstrates genetic modification in multiple organs of postnatal mice by systemic delivery of CRISPR with adeno-associated viruses (AAVs). I resolved the AAV payload limitation by splitting Cas9 and reconstituting the native protein in vivo using scarless split-intein protein trans-splicing, which preserves full activity of Cas9. I determined that the delivery efficiency of AAV-CRISPR dictates gene-targeting rates in vivo, with the preferential gene-editing in liver and heart, and more modest editing efficiencies in skeletal muscle, brain and gonads, directly reflecting the infection profile of the virus serotype. To track CRISPR biodistribution, I established two reporter systems that apply in situ fluorescence activation to demarcate CRISPR-targeting events at single-cell resolution, identifying rare gene-edited cells that normally evade detection by sequencing. This exquisite detection sensitivity further allows evaluation of inter-generational transmission of gene-editing viruses. Finally, although Cas9 elicits host immune responses, these can be ameliorated by immunosuppression. I also identified a public Cas9-responsive T-cell clonotype and mapped the B-cell epitopes on Cas9 and AAV. Engineering tolerance to immunodominant epitopes may provide an avenue for avoiding immune rejection of AAV-CRISPR. The ability to create programmable genetic modifications in multiple organs of postnatal mammals provides a powerful tool for biological research, and foretells that the genomes of whole mammals may be rewritten at will. / Medical Sciences

Development is the elegant and complex process by which a single celled embryo grows into a fully formed and functional adult. One of the major challenges in the process of development is that although each cell in an organism contains the same DNA, different cell types must express different sets of genes in order to perform their highly diverse functions. Furthermore, to form a functioning organism, an embryo must do more than simply form the correct tissue types. It is not sufficient to simply form bone for example; a functioning skeleton requires the correct number of bones, the correct size and shape of the bones, and the correct interaction of each of the bones with each other and with other tissues. Positional information has been proposed as the means by which a cell is able to achieve its appropriate fate by identifying its location within a field of cells in order to correctly pattern tissues as well as the embryo as a whole. Positional information is conferred by a combination of extrinsic signals such as through morphogen gradients and through intrinsic cues such as transcription factor expression. Throughout this work, I have taken advantage of two classic developmental systems, the chicken limb and the mouse axial skeleton, to investigate the diverse gene regulatory mechanisms cells use to interpret positional information in order to properly pattern complex tissues. In chapter two of this dissertation, I explore the role of long noncoding RNAs (lncRNAs) in regulating chicken limb patterning. LncRNAs are an abundant class of transcripts similar in size to messenger RNAs, which are transcribed but not translated, and which play regulatory roles, as RNAs, in controlling gene expression. One important outstanding question is how and to what extent lncRNAs function in development and differentiation in vivo. In order to address this question, we first identified lncRNAs expressed in the developing chicken using RNA-Seq and a series of bioinformatic tools. We identified a total of 3,197 lncRNA genes, including 2,589 intergenic lncRNAs (lincRNAs), 537 antisense lncRNAs, and 71 lncRNAs acting as direct precursors for small RNAs. Using a whole mount in situ hybridization screen we identified lncRNAs with potential roles in limb development. One candidate, HOXA10-AS, was functionally tested by knockdown via in ovo electroporation, demonstrating a requirement for proper patterning of the limb proximal-distal axis. In chapter three of this dissertation, I examine the role of charge-dependant gene repression via chromatin compaction by the Polycomb group protein CBX2 in patterning the mouse axial skeleton in vivo. Polycomb and Trithorax group proteins are epigenetic regulators which stably maintain gene silencing or activation respectively over many generations of cells. Mechanistic studies of Polycomb Repressive Complex 1 (PRC1) have mainly focused on its enzymatic function to ubiquitylate histones, but recent studies suggest that the ubiquitylation activity of PRC1 is dispensable for silencing. The ability of CBX2, a component of PRC1, to compact chromatin directly in a charge-dependant manner has been proposed as an alternative mechanism to explain the role of PRC1 in maintaining silencing. To investigate this possibility we generated three independent mouse lines in which six, 13, or 23 positive amino acids in the putative CBX2 compaction domain were mutated to alanine, demonstrating in a dosage-dependent manner that charge-dependent chromatin compaction by CBX2 is required for the maintenance of gene silencing and for properly patterning the mouse axial skeleton—likely through regulating Hox gene expression. Finally, I end this dissertation with five appendices describing separate and related studies. In the first appendix we examine how the number of skeletal elements in the chicken forelimb zeugopod (radius/ulna) is determined and implicate SOX6 as having an Avian-specific role in activating chondrogenesis in the anterior zeugopod. The subsequent four appendices are publications I contributed to as a co-author over the course of my PhD. In the second appendix our chicken lncRNA data was used as part of a larger analysis of the principles of lncRNA evolution derived from a comparison of 17 transcriptomes. In the third appendix we shared our chicken lncRNA data with the Avian RNA-Seq Consortium as part of the third report on chicken genes and chromosomes. In the fourth appendix we examine the roles of the miR-196 family of microRNA in independently regulating vertebral number and identity in the mouse axial skeleton. Finally, in appendix five we identify a Gremlin1 positive osteochondroreticular stem cell population in the mouse bone marrow with the potential to self-renew and to generate osteoblasts, chondrocytes, and reticular marrow stromal cells. Taken together, this dissertation provides insight into the many diverse gene regulatory mechanisms required for patterning the vertebrate embryo. / Medical Sciences

Tendons transmit force from muscle to bone, enabling movement; ligaments connect bone to bone, maintaining stability. Despite their importance, tendon and ligament development is relatively uncharacterized. In this dissertation, the zebrafish is introduced as a model to study tendon development; tendon populations in zebrafish are homologous to their force-transmitting counterparts in higher vertebrates. The zebrafish craniofacial tendons and ligaments were identified based on expression analysis of genes enriched in mammalian tendons and ligaments (scleraxis, collagen 1a2 and tenomodulin) or in zebrafish tendon-like myosepta (xirp2a). The craniofacial tendons and ligaments were fate-mapped to arise from cranial neural crest cells (CNCCs). Loss-of-function genetic approaches demonstrated that craniofacial and pectoral fin tendons require muscle for their maintenance and cartilage for their organization, with neither tissues required for their induction. Lastly, adult zebrafish and mammalian tendons and ligaments share similar ultrastructural properties. The second part of this dissertation sought to understand the processes governing tendon cell induction. A zebrafish chemical screen identified Lovastatin and Simvastatin as compounds that affect craniofacial tendon development. Statins caused a dose-dependent expansion of the craniofacial and pectoral fin tendon programs. The expansion is specific to the tendon and ligament lineages, and is not observed in other musculoskeletal tissues. Chemical rescue and genetic loss-of-function experiments demonstrated that statin-mediated expansion of scleraxisa is specific to the mevalonate pathway and Hmgcr inhibition. Furthermore, the expansion of tendon progenitors is mediated through inhibition of geranylgeranylation, and specifically geranylgeranyltransferase type I (GGTase I). We discovered that this phenotype is due to a change in cell fate specification, and not an increase in cell proliferation. Fate mapping experiments demonstrated that the expanded craniofacial scleraxisa-positive tendon progenitors arise entirely from CNCCs. As statin caused a reduction of CNCC-derived cartilage, statin is speculated to promote expansion of the craniofacial tendon program by directing CNCCs towards a tendon fate through regulation of GGTase I activity. Taken together, this dissertation advances our understanding of vertebrate tendon development. The first part establishes the zebrafish as a model to study tendon development; and the second part explores the regulation of the mevalonate pathway in tendon development. / Medical Sciences

The comprehensive identification of genetic alterations is critical to understanding the pathophysiology of cancer. Recent advances in sequencing technology have enabled the detailed description of cancer genomes. However, to translate these findings into a deeper understanding of cancer biology, analyzing the functional impact of cancer-associated genetic aberration is essential. Here I investigate how to accelerate the functional characterization of two classes of genetic alterations, point mutations and amplifications. The wide spectrum of point mutations that arise in cancer makes them challenging to study comprehensively. I have developed a scalable systematic method to experimentally infer the functional impact of cancer-associated gene variants. I performed pooled in vivo tumor formation assays and gene expression profiling using 474 mutant alleles curated from 5,338 human tumors. I identified 12 transforming alleles including two in genes (PIK3CB, POT1) that have not been previously shown to be tumorigenic. One rare KRAS allele, D33E, displayed tumorigenicity and constitutive activation of RAS effector pathways. By correlating gene expression changes induced upon expression of wild type and mutant alleles, I could infer the activity of specific alleles. These approaches enable the interrogation of cancer-associated alleles at scale and demonstrate that rare alleles may be functionally important. Frequently amplified regions in cancer often harbor oncogenic drivers. However, identifying the driver gene among many other amplified genes is challenging. In high-grade serous ovarian cancer (HGSOC), 1,825 genes are amplified across 63 amplicons. We employed systematic loss-of-function RNAi data to identify amplified genes that were essential in the ovarian lineage. We identified 50 amplified and essential genes and validated FRS2, an adaptor protein in FGFR pathway. FRS2-amplified cancer cell lines were dependent on FRS2 expression and FRS2 overexpression in immortalized cell lines was sufficient to promote anchorage independent growth and tumorigenesis in nude mice. This approach demonstrates that intersecting structural genomics with functional genomics can facilitate the discovery of driver genes in recurrently amplified regions. Collectively, the methods I present here provide a framework to study point mutations and amplifications to accelerate the interpretation of the cancer genome. / Medical Sciences

The coalescent is a stochastic process that describes the genetic ancestry ofindividuals sampled from a population. It is one of the main tools of theoretical population genetics and has been used as the basis of many sophisticated methods of inferring the demographic history of a population from a genetic sample. This dissertation is presented in four chapters, each developing coalescent theory to some degree. In the first chapter, I investigate how patterns of coalescence are affected by the population pedigree in structured populations, showing that the pedigree has longer-term effects in structured populations than in unstructured populations. Based on my findings, I develop a conceptual framework for jointly inferring population sizes, migration rates, and the recent pedigree of sampled individuals, and I demonstrate the efficacy of this approach in an application to simulated data. In Chapter 2, I present a theoretical study of the distribution of segments of identity-by-descent, showing how the accuracy of predictions made based on sequentially Markov coalescent models depends on the particular model being used as the basis of calculations. In the third chapter, I undertake a theoretical comparison of two approximations, termed the SMC and SMC', to the full model of coalescence with recombination. I derive new theoretical properties of the SMC' and use these properties to demonstrate that the SMC' is, in a well-defined sense, the most appropriate first-order approximation to the full coalescent with recombination. I also show that estimates of population size based on the SMC are statistically inconsistent. Finally, in Chapter 4, I develop a coalescent hidden Markov model approach to inferring the demographic and reproductive history of a triploid asexual lineage derived from a diploid sexual ancestor. The motivation for this project is an ongoing collaborative effort to sequence and analyze the genomes of sexual and asexual lineages of the New Zealand snail Potamopyrgus antipodarum. The method I present in this chapter will be applied to these genomes to infer when triploid asexual lineages were derived from sexual ancestors and to describe the demographic history of those sexual ancestors. Here, I investigate the this method with simulated asexual genomes. / Biology, Organismic and Evolutionary

In an effort to position the alder, Alnus incana, in the scope of phylogenetic relationships, ribosomal RNA genes from its chloroplasts have been cloned and analysed. A genomic library was constructed in the vector LambdaGEM-11 with DNA isolated from leaf tissue. DNA fragments covering the length of the 23S and 16S rRNA genes as well as trnI and sections of trnA were sequenced. The order of chloroplast genes within the sequenced region of alder is identical to that seen in higher plants: 16S-trnI-trnA-23S. Comparisons of the primary sequence of the alder 16S and 23S rRNA genes with available ribosomal RNA and DNA sequences have demonstrated the greatest degree of homology between the alder sequences and those of other higher plant sequences. Comparisons to bacterial sequences yield higher homologies to the cyanobacterial sequences as opposed to those of E. coli. Further sequence analysis shows that the primary sequence obtained for the 16S rRNA gene in alder can be folded to follow the secondary structure of the maize 16S rRNA molecule. (Abstract shortened by UMI.)

The present work investigated the mechanism of induced radiation resistance. Experiments were designed to identify the nature of lesions that induce this mechanism and the cellular components which regulated its expression in the yeast Saccharomyces cerevisiae and the green alga Chlamydomonas reinhardtii. DNA recombinational repair capacity induced by DNA damage was believed to be the system that conferred resistance to killing by radiation in yeast. The nature of the radiation generated DNA lesions that induce this mechanism have been examined using the two different endpoints of resistance to cell killing and suppression chemical mutation. In yeast, DNA lesions produced by low-LET (Linear Energy Transfer) $\gamma$-rays were more efficient at inducing radioresistance and mutation suppression than lesions produced by high-LET neutrons. It was inferred that DNA single strand breaks were important inducing lesions. Oxygen modifiable lesions and hydroxyl radicals (OH$\cdot)$ were particularly efficient inducers of this repair mechanism. Radiation induction of DNA repair depends not only on a variable induction response to different DNA lesions, but also on the availability of the induced repair system to deal with subsequent damage. Yeast mutants deficient in topoisomerase I, in topoisomerase II, or in DNA polymerase I were used to further investigate the involvement of these enzymes in the mechanism of induced thermal tolerance or radiation resistance. The systems that confer increased resistance to heat or radiation were independent of either topoisomerase activity or DNA polymerase function but topoisomerases may have a regulatory role during the signalling of these mechanisms. Maintenance of correct DNA topology may prevent induction of the heat shock response, and heat shock induction of a component of the full radiation resistance in yeast may be the consequence of topoisomerase I inactivation. DNA polymerase I did not seem to have a role in the thermal tolerance or radioresistance mechanism in yeast. Synchronized G$\sb1$ or G$\sb2$ C. reinhardtii cells were neither radiosensitized nor did they become radioresistant in response to heat shock. However, synchronized G$\sb1$ or G$\sb2$ algal cells had a normal heat-induced thermotolerance response. Neither radiation, heat shock, nor cycloheximide suppressed chemical mutations in algae. The DNA repair capacity of C. reinhardtii was not altered by heat shock and in this organism chemical mutations were not a consequence of an induced error-prone repair system. Heat shock was capable of inducing only partial radioresistance in wild type yeast cells while radiation induced full radioresistance. Dot blot analysis of the RAD52 epistasis group revealed that heat shock increased expression of only RAD52 transcripts whereas radiation increased expression of RAD52, 54, and 57 therefore demonstrating differential induction of repair genes by heat shock or radiation. Yeast mutants deficient in recombinational repair (rad52) were radiosensitive and neither heat shock nor radiation exposure could induce radioresistance. Nonetheless these mutants exhibited enhanced expression of RAD51, rad52, RAD54, 55, and 57 transcripts after either heat shock or radiation exposure. It was inferred that the inducible, non-functional rad52 gene product was incapable of conferring radioresistance (repairing DNA damage) and, as a consequence, additional repair genes may have subsequently been induced in an unsuccessful attempt to repair the residual damage. (Abstract shortened by UMI.)

The understanding of the mechanisms involved in the control of gene expression is an important step towards understanding the nature of cellular events such as cellular determination and differentiation. We have studied one of many mechanisms by which the control of gene expression is exercised, namely the control of the rate of initiation of transcription. We have developed a plasmid-based system for the search and characterization of transcriptional promoters and enhancers within small size genomes. The system was designed to select for strong promoting and enhancing elements capable of substitution for those of the early region of SV40. It allows for the use of different reporter genes and various measurement protocols. We have used part of the viral genome of herpes simplex type 2 to test our system. We have searched a transforming fragment of HSV-2 for the presence of transcriptional enhancers and examined the transcriptional activity on the fragment. We have found no transcriptional enhancer within the fragment, suggesting that transformation is not accomplished through the action of a viral enhancer. The results suggest, as far as carcinogenesis is concerned, that a viral $\beta$-class gene would be a likely target for transcriptional enhancement by a cellular element, should an integration event place the transforming fragment near one. The use of SV40's large T antigen as the reporter gene has allowed us to observe, in rare HeLa cells, normal level production of large T when transcription of the reporter gene is "enhanced" by the presence within the vector of certain sub-fragments of the transforming fragment. Three sub-fragments found in the cellular genome of HSV-2-transformed cells have generated a similar result. We have also shown that within clones of HeLa cells transfected with these herpes-derived constructs, the promoter of one or two integrated copies transcribe at the same level as the promoter and enhancer of SV40. One interpretation of these results suggests that the three sub-fragments somehow direct the integration of the plasmids to transcriptionally active chromosomal locations. We have used the vector system to further characterize an enhancer found upstream of the mouse phosphoglycerate kinase-1 gene (pgk-1). This enhancer induces a 30-fold increase in transcription activity from the pgk-1 promoter in mouse cells. When linked to the SV40 early promoter, the transcriptional enhancement is only marginal. We concluded that the mouse pgk-1 enhancer, unlike the SV40 enhancer, is unable to enhance heterologous promoters. We have used an expression vector where SV40's large T antigen is under the transcriptional control of the mouse pgk-1 promoter to assess whether or not the presence of large T can alter the normal progression of differentiation of the teratocarcinoma cell line P19. We have generated large T-expressing P19 clones. The differentiation of these clones as a monolayer has produced homogeneous populations of differentiated progeny, the common phenotype varying from one clone to the other. One interpretation of these results suggests that cellular determination had already been achieved within these clones. (Abstract shortened by UMI.)