Size and charge of proteins can hinder the intracellular delivery of biomolecules. Larger biomolecules, like proteins and peptides, have fewer delivery methods than those available for genetic material, which have a number of different delivery methods. This often limits the delivery of these larger biomolecules to viral transfections of the desired genetic material, but these can lead to gene editing of non-desired targets, due to the extended expression of the protein within cells. We’ve created and studied a delivery method into cells using proteins and peptides of various sizes and charges with poly-histidine tags for light-controlled delivery with hollow gold nanoshells (HGNs). The design of our delivery system includes a thiolated linker that is attached to the gold’s surface and that has a nitrilotriacetic acid (NTA), biotin, or carboxylic acid (COOH) functional group on the end with polyethylene glycol (PEG) in between. Through copper-NTA affinity or 1-ethyl-3-(-3-dimethylaminopropyl) carbodiimide (EDC) chemistry, a histidine-tagged peptide or protein of interest is coupled to the HGN. Cell-penetrating peptides (CPP) on an orthogonal linker are used to accomplish endosomal uptake of the HGN. After irradiation with a pulsed femtosecond laser with non-damaging near-infrared light (NIR), endosomal disruption and protein release is accomplished. This thesis covers the construction and visualization of protein constructs using live and fixed cell imaging with DLCPT (Direct Live Cell Protein Tracking).

Screening of a small chemical library (Medicines for Malaria Venture Pathogen Box) identified two structurallyrelated pyrazolone (inhibitor 1) and pyridazine (inhibitor 2) DNMT3A inhibitors with low micromolar inhibition constants. The uncompetitive and mixed type inhibition patterns with DNA and AdoMet suggest these molecules act through an allosteric mechanism, and thus are unlikely to bind to the enzyme’s active site. Unlike the clinically used mechanism based DNMT inhibitors such as decitabine or azacitidine that act via the enzyme active site, the inhibitors described here could lead to the development of more selective drugs. Both inhibitors show promising selectivity for DNMT3A in comparison to DNMT1 and bacterial DNA cytosine methyltransferases. With further study, this could form the basis of preferential targeting of de novo DNA methylation over maintenance DNA methylation.

Intracellular delivery of biomolecules is hindered by both the size and charge of the biomolecule. While a number of delivery methods are available for genetic material, few are available for larger biomolecules like peptides and proteins. The delivery of these biomolecules are often limited to viral transfections of the genetic material which can lead to off-target gene editing due to prolonged expression of a protein within a cell. We developed and studied a delivery method for poly-histidine tagged proteins and peptides of variable sizes and charges for transient protein expression with light control via hollow gold nanoshells (HGNs). The construct of our delivery system involves a thiolated strand (nucleic acid or PEG based) containing an NTA attached to the particle’s gold surface. A his-tagged protein or peptide of interest is attached to the HGN via copper-NTA affinity. Endosomal uptake of these particles is mediated through an orthogonal strand containing a cell penetrating peptide (CPP). Protein release and endosomal disruption is achieved after irradiation with a focused femtosecond pulsed-laser of biologically benign near-infrared light. This thesis evaluates a variety of areas, from increasing the therapeutic efficacy of an apoptotic peptide to visualization of protein movement using live cell imaging with VIPERnano.

RNA interference (RNAi), a process that can silence specific genes, has been recognized for its potential in applications from basic research, cancer therapy to tissue engineering. However, the routine use of RNAi for disease treatment or prevention still calls for novel and efficient methods of delivery with targeting control. We recently developed an approach using hollow gold nanoshells (HGNs) and near-infrared (NIR) light: siRNAs loaded on HGNs via gold-thiol bonds are internalized in cells through targeting peptide-mediated pathways; laser illumination at a wavelength resonant with the nanoparticle plasmon (~800nm) triggers the release of siRNAs from gold surface to cytosol without any cell damage. A cancer-therapeutic siRNA (under phase I/II clinical trial) targeting Polo-like kinase (plk1) gene is delivered via this strategy to specifically target human prostate cancer cells. We demonstrated that this method has no off-target toxicity to non-cancerous prostate cells, and requires 10-fold less material than standard transfection methods.

Human embryonic stem cells (hESC) hold immense promise in tissue engineering due to their ability to differentiation into all types of cells, and RNAi can serve as a powerful tool to attain this. We developed our HGN-based construct for effective cell penetration and optimized the protocol to accommodate delivery into hESCs. The effectiveness and biocompatibility of this light-activated RNAi approach were demonstrated by targeting GFP and Oct4 genes in hESCs, and no adverse effects to differentiation were detected.

The in vitro generation of tissues and organs for transplantation therapy and disease models for drug screening demand proficient control over the three-dimensional (3D) patterning of gene regulation. The exquisite activation of siRNA activity by tissue penetrable NIR light irradiation in our technique is well suited for this purpose. We used a two-photon microscope to discriminate targeted cell(s) from neighboring cell(s) by simply focusing NIR light at selected x, y and z position to activate the siRNA. The ability to induce light-mediated gene knockdown persists for at least two days, offering a time window for temporal control.

Furthermore, we developed a universal surface module for the maximized delivery of short functional nucleic acids in versatile mammalian cell lines. A modular biotin-modified thiol-RNA is designed as the scaffold for streptavidin and biotin-TAT attaching, sharing the surface with thiolated functional RNA. We found interesting surface chemistries with direct impact on cellular delivery outcomes. This should provide an important basis for various cell engineering applications.

Site-specific DNA binding by proteins is critical for diverse biological processes, including gene regulation, DNA repair, DNA replication, immune response, and others. In most cases, this binding is preceded by facilitated diffusion, where the protein passively moves along nonspecific DNA (DNA lacking particular binding sites) in search of its site(s). Details concerning the mechanisms of how proteins move along DNA remain unclear yet actively investigated. Several reports claim that facilitated diffusion relies primarily on sliding and hopping processes, where the protein moves along the trajectory of the DNA helix. However, these mechanisms involve redundant searches over local regions (up to hundreds of base pairs), while site finding in vivo requires searches of hundreds of thousands of base pairs. We "rethink" facilitated diffusion in two ways: 1) what are the mechanisms available for facilitated diffusion? And 2) how are these mechanisms driven by proteins' specific biological context?

We investigate the facilitated diffusion properties of E. coli DNA adenine methyltransferase (Dam), which methylates palindromic 5'-GATC-3' sites on the N6 position of adenine. We use an in vitro steady-state processivity assay, where the ability of the enzyme to catalyze multiple methylations within a single binding event is quantified. Our results are suggestive of a new mechanism, intersegmental hopping, where proteins hop between regions of a single DNA molecule that are looped together. This mechanism allows for efficient long-range movements, and is similar to other ambiguously described looping mechanisms proposed in recent reports. Dam's low cellular copy number and need to move along the entire bacterial chromosome (i.e. its biological context) is ideally accommodated by this mechanism, as we imagine will be true for other proteins with low copy numbers and rare sites.

We also demonstrate that EcoRI ENase uses an extreme sliding mechanism, previously hypothesized to be unlikely. EcoRI ENase and EcoRV ENase both cut incoming phage DNA as part of type II restriction modification systems, and are shown here to use distinct facilitated diffusion mechanisms. We argue that their site finding mechanisms are complimentary, ensuring robust phage defense, again demonstrating the connection between facilitated diffusion and biological context.

Furthermore, we show that Dam is able to move along DNA in spite of pre-incubated protein roadblocks. Provocatively, DNA-loop inducing roadblocks, such as the histone-like Lrp, can improve the translocation process by specifically enhancing intersegmental hopping. Therefore, this mechanism is proposed as a general way for proteins to move along the crowded, compacted genomic DNA landscape.

Dam was previously demonstrated to undergo intrasite processivity, where the methylation of both strands within a single GATC site proceeds by a single Dam enzyme dramatically reorienting itself while switching between strands. Here, we show that intrasite processivity is only possible on long stretches of DNA when GATC sites are clustered. GATC sites are clustered in most known and putative Dam methylation dependent epigenetic operons, which involve transitions from uniquely unmethylated GATC sites to fully methylated ones. Intrasite processivity is therefore postulated as necessary for epigenetic gene regulation by Dam.

DNA methyltransferases are responsible for transcriptional regulation, cell cycle progression, DNA repair, DNA protection, tumor suppression, and several other important biological processes. Aberrant bacterial DNA methylation can lead to cell death and loss of protection against viral infection; in humans this leads to cancer, autoimmune diseases, metabolic disorders, and neurological disorders. Thus, DNA methyltransferases are common drug targets for cancer therapeutics and novel antibiotics.The conformational transitions in DNA and protein that govern recognition, substrate accessibility, and catalysis are fundamental to understanding the mechanisms that regulate biological processes. The bacterial N6-adenine cell-cycle regulated DNA methyltransferase, CcrM, is the first DNA methyltransferase shown to rely on a unique DNA recognition mechanism in which the DNA strands are separated and most recognition interactions appear to involve the target strand. Strand-separation is emerging as a novel DNA recognition mechanism but the underlying mechanisms and quantitative contribution of strand-separation to fidelity remain obscure for any enzyme (CRISPR-Cas9 and RNA polymerase sigma factor). This work uncovers the fundamental steps governing CcrM's DNA strand separation and high-fidelity DNA recognition mechanism. We relied on mutational analysis of highly conserved residues in the C-terminal domain, Loop-2B, Loop-45, and the active site to probe the function of structurally implicated protein moieties. We collected stopped-flow kinetic fluorescence to monitor transitions in DNA and protein and relied on rigorous global data fitting to understand the states that regulate catalysis in CcrM. We incorporated Pyrrolo-dC into cognate and noncognate DNA to monitor the kinetics of strand-separation and used tryptophan fluorescence to follow protein conformational changes. Both signals are biphasic and global fitting showed that the faster phase of DNA strand-separation was coincident with the protein conformational transition. Non-cognate sequences did not display strand-separation and methylation was reduced >300-fold, providing evidence that strand-separation is a major determinant of selectivity. Analysis of an R350A mutant (C-term domain) showed that the enzyme conformational step can occur without strand-separation, so the two events are uncoupled. A stabilizing role for the methyl-donor (SAM) is proposed; the cofactor interacts with a critical loop which is inserted between the DNA strands, thereby stabilizing the strand-separated conformation. Loops 2B and 45 are inserted between the strand-separated DNA interface. During strand-separation, residues within Loops 2B, 45, and 6E contact the target DNA strand that undergoes methylation. R44 and R129 (Loop-2B and Loop-45, respectively) when mutated to Alanine, disrupt strand-separation and are catalytically inactive. The highly conserved Loop-45 residue F125, which is positioned between the separated DNA strands, is also essential for maintaining the strand-separated intermediate; replacement of F125 with Alanine, Leucine, and Tryptophan results in various perturbations of strand-separation that are correlated to the bulkiness of the substituted residue. Global fitting for each mutant shows that generation and stabilization of DNA strand-separation are perturbed, providing a functional role for these loops in generating and stabilizing the strand-separated intermediate, which is essential for discrimination and catalysis. Employing a fluorescent adenine analog (6MAP) at the target position to monitor base flipping, we resolved that target adenine base flipping follows DNA strand separation and is followed by fast methylation and fast product DNA release. A W57F mutant (active site) displayed an unaltered rate of base flipping as monitored by 6MAP fluorescence but greatly reduced rate of methylation, showing that base-flipping and methylation can be uncoupled. In addition, single-stranded DNA bypasses the DNA strand separation step, while rates of base flipping measured by 6MAP fluorescence and DNA methylation are similar to dsDNA. Global data fitting for each model resolves that base flipping of the target adenine is the rate-limiting step in catalysis. The results presented here are broadly applicable to the study of other N6-adenine methyltransferases that contain the structural moieties implicated in strand-separation (Loop-2B, Loop-45, and the C-terminal domain), which are found widely dispersed across many bacterial phyla, including human and animal pathogens. Insights into CcrM’s mechanism of DNA strand-separation are likely to clarify strand-separation mechanisms for other enzymes such as CRISPR-Cas9 and RNA polymerase sigma factor. Additionally, the elevated understanding of CcrM’s strand-separation mechanism could be useful for the development of selective CcrM inhibitors as novel antibiotics.

De novo DNA methylation by DNMT3A is a fundamental epigenetic modification for transcriptional regulation during cellular development and differentiation. The establishment of appropriate DNA methylation patterns in the human genome involves a crosstalk between DNMT3A, histone tails, regulatory proteins and RNAs. Mutations in DNMT3A and disruptions to this crosstalk of DNMT3A have been reported in several diseases, including Acute Myeloid Leukemia (AML). We sought to model how previously uncharacterized mutations in DNMT3A contribute to the aberrant DNA methylation observed in AML. We then sought to better understand the relative roles of histone tails, regulatory proteins and RNAs in the simultaneous modulation of DNMT3A activity and identified two molecules that inhibit protein interactions with DNTM3A. Although 22% of all AML patients exhibit mutations throughout the DNMT3A gene, the effect of many mutations on DNMT3A activity remain uncharacterized. Moreover, little is known of the interactions between DNMT3A mutants and regulatory proteins, or conversely, mutant regulatory proteins and wild type DNMT3A. We show that previously unexplored DNMT3A mutations dramatically alter the enzyme’s ability to perform CpG and non-CpG methylation as well as the ability of partner proteins to modulate enzymatic activity. Additionally, cell-based studies of one of these DNMT3A mutations (S714C) replicated our findings in vitro showing a dramatic loss of genome-wide methylation. We found that p53 decreases DNMT3A activity by forming a heterotetramer complex with DNMT3A, and while p53 binds DNMT3A R882H, repression of DNMT3A activity is blocked by this substitution. Using p53 as an example, we also show that cancer-related substitutions to p53 are unable to disrupt DNMT3A heterotetramers, unlike that observed for WT p53. To provide insights into the crosstalk between DNMT3A and distinct epigenetic mechanisms we initially assessed the relative of role of regulatory proteins and histone tails in the simultaneous modulation of DNMT3A activity. Using radiochemical and binding assays under distinct conditions and with biologically relevant substrates, we found that regulatory proteins play dominant roles in the modulation of DNMT3A activity. Furthermore, we provide evidence that the activity of DNMT3A is not limited to DNA that is part of the initial DNMT3A–nucleosome complex. We then expanded on these findings by carrying out similar studies that included non-coding RNAs. We show regulatory RNAs play a dominant role over additional epigenetic mechanisms in the simultaneous modulation of DNMT3A. Additionally, we present evidence that is inconsistent with a model for RNA regulation of DNMT3A that relies on the formation of localized RNA/DNA structures. Lastly, a screen of a diverse small molecule library identified two compounds that act as inhibitors of protein-protein interactions (PPIs) with DNMT3A. Thus, presenting a basis for manipulating the allosteric regulation of DNMT3A.