The studies discussed in this thesis unify experimental and theoretical techniques, both established and novel, in investigating the problem of how a protein that binds specific sites on DNA translocates to, recognizes, and stably binds to its target site or sites. The thesis is organized into two parts. Part I outlines the history of the problem and the theory and experiments that have addressed the problem and presents an apparent incompatibility between efficient search and stable, specific binding. To address this problem, we elaborate a model of protein-DNA interaction in which the protein may bind DNA in either a search (S) mode or a recognition (R) mode. The former is characterized by zero or weak sequence-dependence in the binding energy, while the latter is highly sequence-dependent. The protein undergoes a random walk along the DNA in the S mode, and if it encounters its target site, must undergo a conformational transition into the R mode. The model resolves the apparent paradox, and accounts for the observed speed, specificity, and stability in protein-DNA interactions. The model shows internal agreement as regards theoretical and simulated results, as well as external agreement with experimental measurements. Part II reports on research that has tested the applicability of the two-mode model to the tumor suppressor transcription factor p53. It describes in greater depth the experimental techniques and findings up to the present work, and introduces the techniques and biological system used in our research. We employ single-molecule optical microscopy in two projects to study the diffusional kinetics of p53 on DNA. The first project measures the diffusion coefficient of p53 and determines that the protein satisfies a number of requirements for the validity of the two-mode model and for efficient target localization. The second project examines the sequence-dependence in p53's sliding kinetics, and explicitly models the energy landscape it experiences on DNA and relates features of the landscape to observed local variation in diffusion coefficient. The thesis closes with proposed extensions and complements to the projects, and a discussion of the implications of our work and its relation to recent developments in the field.

Gene expression is a fundamental process in the cell and is made up of two parts – the information flow from DNA to RNA, and from RNA to protein. Here, we examined specific sub-processes in Escherichia coli gene expression using newly available tools that permit genome-wide analysis. We begin our studies measuring mRNA and protein abundances in single cells by single-molecule fluorescence microscopy, and then focus our attention to studying RNA generation and degradation by high throughput sequencing. The details of the dynamics of gene expression can be observed from fluctuations in mRNA and protein copy numbers in a cell over time, or the variations in copy numbers in an isogenic cell population. We constructed a yellow fluorescent fusion protein library in E. coli and measured protein and mRNA abundances in single cells. At below ten proteins per cell, a simple model of gene expression is sufficient to explain the observed distributions. At higher expression levels, the distributions are dominated by extrinsic noise, which is the systematic heterogeneity between cells. Unlike proteins which can be stable over many hours, mRNA is made and degraded on the order of minutes in E. coli. To measure the dynamics of RNA generation and degradation, we developed a protocol using high throughput sequencing to measure steady-state RNA abundances, RNA polymerase elongation rates and RNA degradation rates simultaneously with high nucleotide-resolution genome-wide. Our data shows that RNA has similar lifetime at all positions throughout the length of the transcript. We also find that our polymerase elongation rates measured in vivo on a chromosome are generally slower than rates measured on plasmids by other groups. Studying nascent RNA will allow further understanding of RNA generation and degradation. To this end, we have developed a labeling protocol with a nucleoside analog that is compatible with high throughput sequencing.

Von Willebrand factor (VWF) is a multimeric glycoprotein that critically supports platelet aggregation in hemostasis. Disordered VWF function causes both thrombotic and bleeding disorders, and genetic defects in VWF are responsible for von Willebrand’s disease (VWD), the most common inherited bleeding disorder in humans. Very large VWF multimers exhibit the greatest thrombogenic activity, which is attenuated by ADAMTS13 cleavage in the A2 domain. A2 cleavage is regulated by mechanical force, and pathologically high shear forces are known to enhance proteolysis and cause bleeding in patients. Enhanced cleavage is also described in patients with VWD 2A mutations. In contrast, VWF A2 is stabilized against cleavage by a calcium binding site within A2. Single molecule studies have demonstrated that mechanical unfolding is required for A2 cleavage to expose the scissile bond. In this dissertation, we aim to better understand the mechanosensitivity of A2 cleavage by characterizing the force sensitivity of A2 unfolding and refolding. We first characterized the interaction between VWF A2 and calcium using bulk isothermal calorimetry and thermal denaturation assays. In parallel, we used single molecule optical tweezers to characterize A2 unfolding and refolding. Calcium was found to bind A2 with high affinity, stabilize A2 against thermal denaturation, and enhance domain refolding. In contrast, we found that VWD 2A mutations destabilize the A2 domain against thermal denaturation. R1597W, the most common VWD 2A mutation, lies within the calcium binding loop and exhibited diminished calcium stabilization against thermal denaturation. Using optical tweezers, we found that R1597W also diminished A2 refolding. R1597W refolding in the presence of calcium was similar to that of wild-type A2 in the absence of calcium, suggesting that loss of calcium stabilization contributes to the disease mechanism of R1597W. Other VWD 2A mutations lying outside the calcium binding loop also destabilized A2, but retained calcium mediated stabilization. These studies provide a better understanding of VWD 2A pathophysiology and offer structural insights into A2 unfolding and refolding pathways. By exploring the role of mechanical force in regulating VWF cleavage, this work moves towards a better understanding of how hydrodynamic forces within the vasculature regulate VWF function in hemostasis and thrombosis.

Mathematical modeling has been important in the study of biology. Two main challenges with modeling biological problems are the lack of quantitative data and the complexity of biological problems. With the invention of new techniques, like single molecule transcript counting, very quantitative gene expression measurements at the level of single transcript in individual cells can now be obtained. Biological systems are very complex, involving many reactions and players with unknown reaction rates. To reduce the complexity, scientists have often proposed simplified phenomenological models that are tractable and capture the main essence of the biological systems. These simplified models allow scientists to describe the behavior of biological systems with a few meaningful parameters. In this thesis, by integrating quantitative single-cell measurements with phenomenological modeling, we study the (1) roles of Wnt ligands and receptors in sensing and amplification in Caenorhabditis elegans’ P cells and (2) regulation of rDNA transcription in Saccharomyces cerevisiae. The initiation of cell polarity consists of two sequential processes: an external gradient is first sensed and then the resulting signal is amplified by intracellular signaling. It is challenging to determine the role of proteins towards sensing and amplification as these two processes are intertwined. We integrated quantitative single-cell measurements with phenomenological modeling to determine the roles of Wnt ligands and receptors in sensing and amplification in the P cells of Caenorhabditis elegans. By systematically exploring how P cell polarity is altered in Wnt ligand and receptor mutants, we inferred that ligands predominantly affect sensing, whereas receptors are needed for both sensing and amplification. Most eukaryotes contain many tandem repeats of ribosomal RNA genes of which only a subset is transcribed at any given time. Current biochemical methods allow for the determination of the fraction of transcribing repeats (ON) versus nontranscribing repeats (OFF) but do not provide any dynamical information. By using the single molecule transcript counting technique complemented with theoretical modeling, we determine the rate of switching from OFF to ON (activation rate) and the average number of RNA molecules produced during each transcriptional burst (burst size). We explore how these two variables change in mutants and different growth conditions.

Skeletal muscle (SkM) possesses an impressive ability to regenerate in response to injury or chronic disease. This regenerative capacity is attributed to its resident mononuclear myogenic progenitors. Previous studies have identified several types of myogenic progenitors within SkM, some of which are isolated by fluorescence activated cell sorting (FACS) using cell surface markers. Studies in our laboratory have identified melanoma cell adhesion molecule (MCAM) as a cell surface marker expressed by myogenic progenitors in human fetal SkM. However, the relationship between MCAM expression and the degree of myogenic commitment of distinct MCAM+ populations has not been elucidated. In the present study, subpopulations of MCAM+ cells were purified by FACS on the basis of Hoechst 33342 dye uptake. Specifically, MCAM+ side population (SP) was isolated by Hoechst exclusion and MCAM+ main population (MP) on Hoechst incorporation. Sorted populations were first optimized for growth in vitro since SkM SP cells are difficult to maintain in culture. In particular, Invitrogen’s StemPro® MSC SFM medium was found to support propagation of human fetal SkM SP cells with minimal differentiation. Following this optimization, sorted populations were assessed for expression of myogenic markers before and after propagation and then for fusion potential in vitro and engraftment potential in vivo. The MCAM+ subpopulations were found to express myogenic markers to a significantly greater extent than MCAM- subpopulations. Furthermore, the MCAM+ subpopulations fused robustly into myotubes in vitro whereas the MCAM- subpopulations did not. Interestingly, the MCAM+ SP population exhibited the highest fusion potential in vitro and was the only MCAM+ subpopulation to engraft into dystrophic muscle in vivo following propagation. These results indicate that MCAM is associated with myogenicity and can be used to prospectively isolate a pure myogenic fraction from human fetal SkM tissue. Moreover, the MCAM+ SP retain its myogenic potential to a greater extent than MCAM+ MP after propagation. This suggests that the MCAM+ SP fraction contains a higher percentage of early myogenic progenitors compared to the MCAM+ MP fraction. Additional studies on MCAM-expressing populations in human fetal SkM may elucidate a potent population for use in cell-based therapeutic strategies for treating muscle diseases.

To enter and infect cells, viruses must overcome the barrier presented by the cell membrane. Enveloped viruses, which possess their own lipid bilayer, fuse their viral membrane with the cell membrane. Non-enveloped viruses, whose outer surface is composed of proteins, penetrate through the hydrophobic interior of the cell membrane. Viruses accomplish the processes by coupling conformational changes in viral "entry proteins" to membrane disruption. This dissertation investigates the membrane disruption mechanisms of rotavirus, a non-enveloped virus, and vesicular stomatitis virus (VSV), an enveloped virus. Rotavirus uses proteins of its outer capsid to penetrate the membrane and deliver a transcriptionally-active core particle into the cell cytoplasm. \(VP5^*\), an outer capsid protein, undergoes a foldback rearrangement that translocates three clustered hydrophobic loops by \(\sim 180^{\circ}\). This rearrangement resembles the foldback rearrangements of enveloped virus fusion proteins. In the first half of my dissertation, I show that the hydrophobicity of the \(VP5^*\) apex is required for membrane disruption during rotavirus cell entry by mutating hydrophobic residues within the loop to hydrophilic residues. One particular mutation diminishes liposome interaction by the protein, blocks membrane penetration by virus particles in cells, and reduces particle infectivity by 10,000-fold. VSV uses its fusion protein, G, to fuse at low pH. Unlike other viral fusion proteins, pH-induced conformational changes in G are reversible. In the second half of my dissertation, I measure the fusion kinetics of individual VSV particles using a single-particle fusion assay previously developed for influenza virus. I find that hemifusion by VSV consists of at least two steps, an initial step that is pH-dependent and reversible, and a second step that is pH-independent. At low pHs, the second step becomes the sole rate-limiting step. I also show that at pH 6.6, the VSV particle enters a stable intermediate state that binds tightly to membranes but does not precede to fusion. This dissertation uses a variety of experimental approaches to arrive at a more detailed understanding of how viruses use their entry proteins to either penetrate or fuse with the cell membrane.

Eukaryotic cells face a considerable challenge organizing a complicated interior with spatial and temporal precision. They do so, in part, through the deployment of the microtubule- based molecular motors kinesin and dynein, which translate chemo-mechanical force production into the movement of diverse cargo. Many aspects of kinesin’s motility mechanism are now known in detail, whereas fundamental aspects of dynein’s motility mechanism remain unclear. An important unresolved question is how dynein couples rounds of ATP binding and hydrolysis to changes in affinity for its track, a requisite for a protein that takes steps. Here we report a sub- nanometer cryo-EM reconstruction of the high affinity state of dynein’s microtubule binding domain in complex with the microtubule. Using molecular dynamics flexible fitting, we determined a pseudoatomic model of the high affinity state. When compared to previously reported crystal structure of the free microtubule binding domain, our model revealed the conformational changes underlying changes in affinity. Surprisingly, our simulations suggested that specific residues within the microtubule binding domain may tune dynein’s affinity for the microtubule. We confirmed this observation by directly measuring dynein’s motile properties using in vitro single molecule motility assays, which demonstrated that single point mutations of these residues dramatically enhance dynein’s processivity. We then sought to understand why dynein has been selected to be a restrained motor, and found that dynein-driven nuclear oscillations in budding yeast are defective in the context of highly processive mutants. Together, these results provide a mechanism for the coupling of ATPase activity to microtubule binding and release by dynein, and the degree to which evolution has fine-tuned this mechanism. I conclude with a roadmap of future approaches to gain further insight into dynein’s motility mechanism, and describe our work developing materials and methods towards this goal.

The influenza virus hemagglutinin (HA) surface protein is a primary antigenic target for neutralization of viral infection. HA also mediates membrane fusion between the virus and a cell, which is the first critical step during infection. Traditional techniques to study infection neutralization by antibodies or the membrane fusion process rely on ensemble measurements, confounding the precise mechanism of infection neutralization and obscuring transient conformational intermediates. This dissertation describes advances made in a fluorescence microscopy-based single-particle fusion assay to overcome the limitations of ensemble measurements in these types of studies. Virus particles are labeled to visualize lipid mixing between a virus and a target membrane formed upon a glass or polymer support. Optionally, the viral lumen can be labeled to visualize the subsequent release of viral contents.

Single molecules and single cells are the fundamental building blocks in biology. Facilitated by the advancement of technology, quantitative single-molecule and single-cell measurements provide a unique perspective toward many biological systems by revealing individual stochasticity and population heterogeneity. Taking advantage of these approaches, we studied chromosome organization and gene expression in bacteria and discovered new biophysical mechanisms: chromosome organization by a nucleoid-associated protein in live bacteria, and transcriptional bursting by the regulation of DNA supercoiling in bacteria.

EZH2 is a SET domain-containing methyltransferase and the catalytic component of the multimeric Polycomb- group (PcG) protein complex, PRC2. When in complex with other PRC2 members (EED, SUZ12, AEBP2, and RBBP4), EZH2 catalyzes methylation of H3K27, a histone modification associated with transcriptional repression and developmental regulation. As several PRC2 components are upregulated or mutated in a variety of human cancers, efforts to discover small-molecule modulators of PRC2 and understand its regulation may yield therapeutic insights. Identification of small-molecule probes with distinct chemotypes, MOAs, and selectivity profiles are not only of great value, but necessary in establishing comprehensive probe sets capable of illuminating the various roles of EZH2 in oncogenesis. Here we describe efforts to identify and characterize small-molecule modulators of PRC2 and further understand its regulation. Chapter II outlines the expression and purification of 5-component PRC2 (EZH2-EED-SUZ12-AEBP2-RBBP4) and the establishment of biochemical and cellular HTS assays. These assays were used to screen a diverse set of small molecules (>120,000), identifying biochemical PRC2 inhibitors and activators (described in Chapter III). One biochemical PRC2 inhibitor, BRD1835, appeared to inhibit PRC2 activity through a novel artifactual mechanism involving interaction with peptide substrate, leading to apparent peptide-competitive behavior and putative cellular activity (described in Chapter IV). The characterization of novel biochemical PRC2 activators, BRD3934 and BRD8284, is discussed in Chapter V. Chapter VI describes the use of an HCS assay to identify known bioactive compounds that alter intracellular levels of H3K27me3 through modulating H3K27me3-connected regulatory nodes or by targeting PRC2 directly. These efforts led to the discovery that an antifungal agent, miconazole, is capable of activating PRC2 activity in vitro, while a mucolytic agent, bromhexine, selectively ablates cellular H3K27me3 levels through targeting an activity distinct from PRC2. Finally, Chapter VII discusses novel PRC2-connected crosstalk mechanisms identified through screening libraries of uniquely modified histone peptides for their ability to bind or support methylation by PRC2. These studies enhance our understanding of PRC2 regulation by revealing the effects of H3R26 and H3K23me1 modifications on enzymatic activity, implicating their respective methyltransferases in PRC2 regulation.