Enhancers are DNA regulatory elements that play an important role in the precise

regulation of Pol II transcription initiation and, consequently, in the generation and

maintenance of patterns of gene expression. A defining characteristic of enhancers is their

capacity to regulate transcription from their target promoter, independent of the genomic

distance that separates them. Importantly, in order for enhancers to exert their regulatory

action, they must first relocalize into close spatial proximity to their target promoter and

form long-range interactions. Although these interactions have been demonstrated to

occur, the mechanisms that allow their specific and efficient formation remains largely

obscure. In this thesis, I focus on the development of tools and the expansion of our

understanding with regard to this important topic.

In the first chapter, I begin by providing a perspective of transcription initiation as

a hub of gene regulation. Then, I perform a concise overview of promoters and enhancers

as DNA regulatory elements and share the most important functional and structural

characteristics. I finalize the chapter by discussing the known regulatory mechanisms that

allow the formation of specific interactions between enhancers and promoters to take place.

In the second chapter, I present my findings on the development of imaging systems

to permit the visualization of enhancer-promoter interactions in live cells. I begin by

describing the limitations of current systems. Then, I discuss the importance of the

development of novel, innovative systems that allow specific labelling of genomic loci.

From there, I present the possibilities and limitations of a developed dCas9/Aptamer

system for the visualization of repetitive and non-repetitive loci. I also present my findings

on the development of a two-step integration system of operator arrays and share the

principal technical limitations encountered in its development. I finish by discussing and

providing a critical perspective of relevant published work that has been produced since

these efforts were undertaken.

In the third chapter, I begin by identifying and characterizing potential enhancers

in the PCSK9 locus in hepatic derived cells. I identify three candidate enhancers with

strong transcriptional activity by reporter assays. Then, I demonstrate how the regulatory

activity of these candidate enhancers is specific to hepatic cells and their regulatory activity

can be modulated by two small molecules in a similar fashion to the endogenous PCSK9

gene. From there, I dissect these DNA regulatory sequences to identify transcription factor

binding sites responsible for transcriptional activity by bioinformatic analysis and fine

mutagenic studies. These results point towards the presence of three DNA regulatory

elements required for PKCS9 expression.

In the fourth and final chapter, I investigate two alternative promoters (P1 and P2)

that drive MYC expression and are regulated by enhancers. The two MYC promoters can

be differentially regulated across cell-types, and their selective usage is largely mediated by

distal regulatory sequences. Moreover, in colon carcinoma cells, Wnt-responsive enhancers

prefer to upregulate transcription from the P1 promoter, using reporter assays in the

context of the endogenous Wnt induction. In addition, multiple enhancer deletions using

CRISPR/Cas9 corroborate the regulatory specificity of P1. Finally, I explore how

preferential activation between Wnt-responsive enhancers and the P1 promoter is

influenced by the distinct core promoter elements that are present in the MYC promoters.

Taken together, these results provide new insight into how enhancers can specifically target

alternative promoters and suggest that the presence of alternative promoters could create a

more efficient formation of long-range interactions and a more precise combinatorial

regulation of transcription initiation.

Cis–trans relationships govern gene regulation: it is the interactions between diffusible proteins and chromatinized DNA that dictate nuclear functions. The nature of those interactions must allow for the great regulatory complexity needed for eukaryotic function and particularly of multicellular development. However, many of the current models of cis–trans relationships in the nucleus either lack biophysical rigor or have gone untested. This thesis of three chapters addresses three shortfalls in the current conceptualization of how gene-regulatory proteins function.In the introductory chapter, it is argued that there is a widespread over-reliance on qualitative descriptions of biomolecular behaviors and functions. Because biochemistry is inherently stochastic, the way in which such behaviors and functions are described must eventually be probabilistic and quantitative. Simplistic 1:1 relationships between molecules need to be dropped in favor of probabilistically constituted ensembles of factors with different affinities. Furthermore, because the cell interior is literally fluid and molecular interactions quite transient, the temporal dynamics of the system must always be kept in mind, which largely forbid the use of static structures and monolithic series of events. The second chapter issues a challenge to the notion that an important cis relationship, that between enhancers and promoter, is mediated by a direct interaction between the two elements—a striking instance of the structure–function paradigm questioned in the first chapter. It has been assumed for decades that a complex of proteins forms a bridge connecting enhancer to promoter even over large genomic distances. However, neither evidence nor reason rules this assumption in, and other models have hardly been considered, much less ruled out. Here it is proposed that diffusible biochemical species in the form of modified proteins are generated at the enhancer and can affect the promoter via diffusion. What determines the scale of the gradient of signal emanating from the enhancer is the balance of the local generation rate and the ubiquitous degradation rate. Finally, an imaging study is presented in which the question is asked, how are transcription coactivators (trans factors) distributed among the plethora of cis elements? Through single-molecule tracking of p300 and many mutants thereof, it is shown that direct cis interaction—i.e., by binding modified histones—is not responsible for coactivator targeting. Rather, p300 depends on a combination of transcription factor–interaction domains to associate with chromatin, indicating that atop the cis–trans interactions between transcription factors and chromatin is another layer of trans interactions between transcription factors and their cofactors.

It has long been appreciated that organization of macromolecules within the cell is paramount to proper cellular function. As the number of tools in the molecular and cell biologist’s toolbox grows, the list of ways cells achieve organization also grows. This is especially true for the process of transcription and its regulation, where the fundamental questions remain a challenge to address, but where a convergence of genetic, genomic, structural and imaging techniques offer the possibility of answers. The proper loading of an RNA polymerase on a gene requires the coordinated action of hundreds of proteins and other macro molecules, as well as cellular genome to be maintained in a topologically permissive state; yet as a system, gene regulation requires interactions to be dynamic enough to respond to the changing needs of the cell in response to internal and external ques. How a cell constructs a system that is, on the one hand robust, and on the other hand flexible remains a challenge to answer.

In Chapter 1 of this thesis, we will introduce the study of transcription regulation as a general topic, and discuss a number of fluorescence imaging techniques that have fundamentally changed the our understanding of transcription regulation. Chapter 2 will focus on one particular aspect of cellular organization which has recently come into vogue—that of membraneless compartment formation through liquid-liquid phase separation. A thorough investigation of the literature, particularly in light of the experiments which we foreshadow and introduce in more depth in Chapter 3, suggests that much more work is needed before phase separation as a means of biological organization can become a general paradigm. Chapter 3 presents a particularly poignant example of the issues raised in Chapter 2. Using Herpes Simplex Virus as a model system because of its ability to form compartments in the nucleus, we demonstrate that recruitment of Pol II and other proteins can be explained through the availability of nonspecific binding sites rather than phase separation. Lastly, Chapter 4 will present technical findings on the optimal fluorescent dyes to use for in vivo labeling of proteins for imaging applications.

In summary, this dissertation underscores the way in which new imaging techniques reveal new biological principles and challenge old models. It will be of great interest to watch these ideas and techniques mature beyond the work presented in this dissertation.

The cellular control of gene expression is a fundamental process in biological life, and one that is still incompletely understood. DNA is packaged into chromatin inside human cells which is bound by sequence-specific transcription factors and their accompanying chromatin coactivator complexes. These molecular assemblies modify chromatin and regulate mRNA production from nearby and distal genes but how they accomplish these diverse functions is still unclear. In the first part of my thesis, I focus on one particularly elusive chromatin coactivator complex, SAGA, and, through collaboration with the Nogales lab, I characterize its molecular and atomic features using cryo-electron microscopy. The resulting structure reveals many interesting physical features that would have been unpredictable using previous models of SAGA from yeast, including large architectural rearrangements and elucidation of the structural incorporation of non-conserved subunits such as the splicing module. Furthermore, we are able to demonstrate the biological utility of a human-specific structure of SAGA by identifying patient mutations which likely interrupt non-conserved interfaces, making testable predictions of their disease etiologies. The high-resolution characterization of hSAGA paves the way for detailed future mechanistic studies which may interrogate SAGA’s many complex roles in human gene expression.

My work then zooms out to identify and characterize regulators of gene expression in an understudied system within human developmental biology, that of placental development. Studies into the molecular mechanisms of placental development have the potential for tremendous impact on human health where many deadly pregnancy diseases currently have no cures or prevention. Defects in placental stem cell (trophoblast) differentiation have been described in deadly pregnancy diseases such as preeclampsia where patient placentas show impaired cytotrophoblast cell fusion and decreased expression of endogenous retroviral fusogens such as syncytin-2, that correlate with disease severity. Here, I demonstrate new tools to identify and characterize molecular drivers of differentiation in the cell-cell fusion of human placental trophoblasts. I demonstrate the optimization of cellular and molecular tools in the Forskolin-induced BeWo human cell model of placental syncytium formation with the goal of building an imaging-based arrayed CRISPR knockout screen to perform an unbiased search for genes required for placental cell differentiation and fusion. These experiments demonstrate the optimization of high efficiency gene knockouts in BeWo cells and cell fusion assays allowing for high-confidence readout of cell fusion perturbations at scale. Taking a candidate approach, I then investigate a conserved transcription factor, TFEB, and its roles in human placental cell fusion. I reveal that genetic deletion of TFEB and its paralogue TFE3 in BeWo cells results in reduced expression of syncytin-1 and syncytin-2 and functionally impaired cell-cell fusion. Finally, I use confocal and single molecule imaging to elucidate TFEB-chromatin interactions during differentiation, prompting the creation of new models of TFEB activity during trophoblast differentiation. These studies illustrate the ability to identify new regulators of human placental cell fusion using optimized tools in BeWo cells and to deeply characterize the molecular behavior of such regulators at the single molecule level. Such interdisciplinary work demonstrates the vision to use tools which span biological scales to uncover basic mechanisms of gene expression and lay the groundwork for identifying and characterizing novel therapeutic targets which could save lives during pregnancy.

Stroboscopic photoactivated single particle tracking (spaSPT) relies on stochasticlabeling to isolate the paths of individual fluorophores, and can provide information about the behavior of biological macromolecules in their native cellular environment. Existing spaSPT modalities generate large numbers of short trajectories, each representing a fragment of an individual emitter's path. When interpreting this data through the lens of diffusion models, it is essential to account for the fragmentary nature of trajectories, experimental biases arising from the imaging geometry, and our ignorance about the correct underlying diffusion model. In this thesis, we describe several methods for interpretation of spaSPT data that provide estimates about the number and characteristics of mobility states in spaSPT data while accounting for known experimental artifacts. We explore the uses and limitations of these models on simulated and experimental datasets.

In the final chapter, we apply these methods to study the competitive chromatin binding in the type II nuclear receptors (T2NRs). T2NRs are a class of ligand-activated transcription factors that require heterodimerization with a common factor, the retinoid X receptor (RXR), to bind chromatin and regulate target genes. Because all T2NRs must dimerize with a common pool of RXR, competition between individual T2NRs may limit access to the bound state. This mechanism has been proposed to underlie the oncogenic inactivation of wildtype retinoic acid receptor alpha (RARA), a prototypical T2NR, in the presence of RARA fusion proteins that occur in acute promyelocytic leukemia (APL). Such fusion proteins may compete RXR away from the wildtype RARA, impairing its ability to regulate target genes. Here, we use spaSPT to directly measure the effects of RARA fusion proteins on the chromatin binding of endogenously tagged RARA and RXR. Using the tools developed in the previous chapters, we find that RARA fusion proteins act as stronger competitors for dimerization with RXR than wildtype T2NRs, and are also seemingly exempt from a novel autoregulatory mechanism that controls the concentration of wildtype RARA. Together, these results provide new insights into the interdependence of T2NR gene regulation and the manner by which it is disrupted in disease.