In placental mammals, dosage compensation of the sex chromosomes is

achieved through inactivation of one X chromosome in female cells. This X

chromosome inactivation (XCI) requires tight developmental regulation to ensure all but

one X chromosome is silenced.

At the center of this process is Xist, a long non-coding RNA. Upon differentiation

of a female cell, Xist spreads in cis to coat and silence the inactive X chromosome.

While upregulation of Xist, has been shown to be sufficient for X inactivation to occur,

no one has thoroughly investigated whether Xist is necessary for the establishment of X

chromosome inactivation. In this thesis I provide evidence that Xist is not required for

dosage compensation of the X chromosome during epiblast-like cell differentiation. This

result suggests Xist-independent silencing mechanisms for this essential process may

be in place.

A 1-2 Mb region of the X chromosome, termed the X-inactivation center (Xic) is

necessary in two copies for XCI to occur, indicating it is necessary for cells to count the

number of X chromosomes present. I delete one copy of the putative 2 Mb Xic in male

and female mouse embryonic stem cells and present evidence that this deletion is not

well tolerated, suggesting that this region requires finer resolution mapping to identify

the minimal element required for counting.

Finally, I finish with a review which elaborates on studies that enlighten our

understanding of activators and repressors that control XCI. Our findings challenge

existing dogmas in the field and provide the foundation for future work focused on

uncovering the molecular mechanisms behind Xist-independent silencing of the X

chromosome.

Proper spatial and temporal control of gene expression is necessary for cellular survival and proper function. Addition of a methyl group to the 5’ carbon of cytosine in DNA (5mC) is a major mechanism used to modulate gene expression. The Ten-Eleven Translocation (TET) family of enzymes iteratively oxidize 5mC to 5-hydroxymethylcytosine (5hmC), 5- formylcytosine (5fC), and 5-carboxylcytosine (5caC). These modifications function both as stable epigenetic marks and transient intermediates in the demethylation of DNA. Improper placement of these epigenetic marks often causes death or disease. Among the proteins that interact with TET enzymes is O-linked N-acetylglucosamine (O-GlcNAc) Transferase (OGT). OGT is the sole enzyme responsible for attaching a GlcNAc sugar to serine, threonine, and cysteine residues of over 1,000 nuclear, cytoplasmic, and mitochondrial proteins. OGT has been termed a “nutrient sensor” because its activity requires the sugar donor UDP-GlcNAc, whose abundance is dependent upon the levels of various cellular metabolites. Thus the reversible O-GlcNAc modification dynamically regulates the functions of OGT’s targets in response to nutrient status. OGT stably interacts with and modifies TET proteins and its genome-wide distribution overlaps significantly with TETs. However, the significance of the OGT-TET interactions are poorly understood. In this work, we explore the consequences of the OGT-TET interactions in vitro and in mouse embryonic stem cells (mESCs). We show that OGT directly binds and modifies TET1 in vitro, and the O-GlcNAc modification enhances TET1 activity. We identify a point mutation in TET1 that disrupts its interaction with OGT and use this to interrogate the effects on TET activity, gene expression, and epigenetic patterning of disrupting the OGT-TET1 interaction in mESCs. To assess the importance of the OGT-TET interaction for OGT function, we use quantitative SILAC mass spectrometry to examine proteome-wide changes in O-GlcNAcylation in mESCs when Tets are deleted. We also identify sites of O-GlcNAcylation on TET1 and TET2, further analyze the interactions between OGT and all three TETs, and examine the effect of O-GlcNAcylation on TET2 and TET3 activity in vitro. Our results link metabolism and epigenetic control, which may be relevant to the developmental and disease processes regulated by these two enzymes.

The Male Specific Lethal (MSL) complex of proteins is enriched on the single X chromosome in male Drosophila melanogaster, resulting in a twofold enrichment of gene expression from the X chromosome in order to equalize expression with females, which have two X chromosomes. In Chapter I, we show that the zinc finger protein Zn72D is required for proper splicing of the maleless (mle) transcript, which encodes one of the proteins in the MSL complex. In addition, we found that Zn72D colocalizes with elongating RNA Polymerase, suggesting it has a broader role in regulating gene expression outside of its role in dosage compensation. In Chapter II, we identify proteins that interact with Zn72D and find it interacts with several proteins involved in RNA metabolism. Co-knockdown of Zn72D and one of the proteins that interacts with Zn72D, the DEAD box helicase Belle, resulted in partial restoration of mle splicing and 70% restoration of MLE protein levels, suggesting that Zn72D and Belle regulate translation of MLE. These results implicate Zn72D as a protein that links mRNA splicing to localization and translation.

Recent studies have identified over one hundred high-confidence autism spectrum disorder (ASD) genes and several hundred congenital heart disorder (CHD) genes. While CHD has been shown to commonly co-occur with autism spectrum disorder (ASD), the shared molecular mechanisms underlying this comorbidity remain unknown. Recent studies have shown that ASD gene perturbations commonly dysregulate neural progenitor cell (NPC) proliferation and neurogenesis. Our lab first sought to understand the extent to which the broader set of ASD genes are involved in this process and to identify the biological pathways underlying this convergence. Next, we investigated ASD and CHD shared risk by identifying CHD genes that present with a neurogenesis phenotype.

In this dissertation we utilized CROP-Seq to repress a large subset of the known ASD genes in a human in vitro model of cortical neurogenesis. We reinforced neurogenesis and microtubule biology as points of convergence in ASD gene perturbations (Chapter 1). We then investigated shared risk between ASD and CHD by performing a neurogenesis screen involving ASD and CHD genes and determined that a subset of ASD and CHD genes play key roles in neuronal progenitor cell proliferation and are enriched for ciliary biology (Chapter 2). Overall, this work contributes to the growing literature on the developmental pathways disrupted in ASD and CHD.

While physiologic stress has long been known to impair mammalian reproductive capacity through hormonal dysregulation, mounting evidence now suggests that stress experienced prior to or during gestation may also negatively impact the health of future offspring. A growing body of work in recent years has clearly demonstrated that rodent models of physiologic stress can induce a variety of neurologic and behavioral phenotypes that are able to persist for up to three generations, suggesting that stress signals can induce lasting epigenetic changes in the germline. Any perturbations to the proper transmission of genetic information through the germ cells can be detrimental, and thus understanding the mechanism by which stress can lead to epigenetic alterations in the germline remains a crucial unanswered question in the field. Interestingly, treatment with glucocorticoid stress hormones is sufficient to recapitulate the transgenerational phenotypic inheritance seen in physiologic stress models. These hormones are known to bind and activate the glucocorticoid receptor (GR), a ligand-inducible transcription factor, thus implicating GR-mediated signaling as a potential contributor to the transgenerational inheritance of stress-induced phenotypes. It remains unclear, however, whether the heritability of stress-induced phenotypes following glucocorticoid treatment results from direct actions of glucocorticoids on the germ cells of the gonad, or from indirect effects of glucocorticoids elsewhere in the body. Moreover, evidence for the expression of GR in the developing and adult germ cells of both the male and female remains controversial, making it difficult to assess any potential cell-autonomous roles of GR in modulating the epigenome of the germline in response to stress. We therefore set out to definitively characterize the expression of GR in the germline of the developing and adult gonads, and to determine what cell-intrinsic role GR plays in normal germline formation and function. We discovered that GR is expressed in the female germline specifically during fetal development (peaking at approximately E13.5 - E14.5), yet is absent from the adult oocyte. We found that the female germline is, surprisingly, resistant to changes in GR signaling. Both genetic deletion of GR as well as GR agonism with dexamethasone revealed minimal transcriptional changes in the female germ cells, as well as no significant changes in meiotic progression, suggesting the female germline is intrinsically buffered against changes in GR signaling. In contrast we found that GR is expressed in the male germline during the final days of development, with expression maintained in pro-spermatogonia during early postnatal development. This expression becomes restricted to the spermatogonia of the adult, and transcriptomic analysis of germ cells from dexamethasone treated males revealed a potential role of GR in regulating RNA splicing. Together our data confirm the dynamic spatiotemporal expression of GR in both the male and female germ cells of the mouse, and suggest a sexually dimorphic function for this receptor in the mammalian germline.

Heterochromatin protein-1 (HP1) maintains condensed, stable heterochromatin domains throughout interphase despite the weak binding and rapid exchange of HP1 within heterochromatin. I utilize two methods, DNA curtains and liquid-liquid phase separation (LLPS) assays, to decode a mechanistic understanding of HP1 behavior and directly test if dynamic HP1 binding can maintain static DNA compaction in vitro. Within droplets, we find HP1α and DNA have distinct material properties: HP1α rapidly exchanges within and between droplets while simultaneously condensing DNA into stable domains within a droplet. Further, we show HP1α compacted DNA puncta are resistant to 40pN of force, over twice that required to stall RNA polymerase. I find the disordered regions of the three human HP1 paralogs - HP1α, HP1β, and HP1γ – dictate their DNA compaction and LLPS phenotypes. The HP1α hinge is necessary and sufficient for these activities, and we determine a network of hinge autoregulation within the Nand C- terminal extensions. I demonstrate dynamic HP1α binding primes droplets for regulation as the addition of HP1β, which exhibits minimal DNA compaction and LLPS behavior, invades and dissolves preformed HP1α droplets. Finally, I utilize chromatin substrates and find HP1 maintains separate domains of unmodified and H3K9me3 chromatin. Together this data describes how a pool of weakly bound HP1 proteins exploits both the collective behavior of proteins and the polymer properties of DNA to produce self-organizing domains that are simultaneously stable and fragile.

Female (XX) mouse embryonic stem cells (mESCs) differ from their male (XY) counterparts because they have lower levels of 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC). This difference in DNA modifications is a result of having two X chromosomes (Xs), both of which are active at this developmental stage. To test whether OGT is one of the X-linked proteins that regulate 5mC and 5hmC in mESCs, we manipulated OGT dose in XX and XY mESCs. We found that OGT abundance controls cytosine modifications, implicating OGT targets in 5mC and 5hmC regulation. Our quantitative comparison of the O-GlcNAcylated proteome in XX and XY mESCs revealed that O-GlcNAc modified TET3 peptides were more abundant in XX mESCs, which reflected an increase in TET3 amount in these cells. In addition to differing in abundance, TET3 and OGT distribution were also different in XX and XY mESCs. In XX cells, TET3 and OGT were enriched in the nucleus, while they were predominantly cytoplasmic in XY cells. TET3 and OGT occur in different high molecular weight complexes in XX and XY mESCs. When OGT is expressed from one X in XX mESCs, OGT and TET3 are predominantly cytoplasmic and 5mC/5hmC levels increase, indicating that OGT is one X-linked regulator of DNA cytosine modifications. To directly query whether TET3 is necessary for the female-specific 5mC and 5hmC in mESCs we generated homozygous TET3 mutant XX mESCs. In these cells, 5mC and 5hmC levels were decreased relative to wildtype XX mESCs, without dramatic gene expression changes. To investigate the developmental significance of TET3, examined the effects of this mutation on the ability of cells to undergo X chromosome inactivation (XCI) an epigenetic change that occurs when mESCs are differentiated into the next developmental stage, epiblast-like (mEpiL) cells. The establishment of XCI is characterized by the up-regulation of a non-coding RNA, Xist RNA, which remains in the nucleus and ‘coats’ the X concomitant with silencing. In TET3 mutant XX mEpiLCs Xist RNA exhibits abnormal distribution and silencing defects. These results link the activity of a dose-sensitive complex containing X and autosomal proteins to regulation of cytosine DNA modifications and XCI.