N2-(4-Hydroxyphenyl)-2'-deoxyguanosine-5'-O-DMT-3'-phosphoramidite has been synthesized and used to incorporate the N2-(4-hydroxyphenyl)-2'-dG (N2-4-HOPh-dG) into DNA, using solid-state synthesis technology. The key step to obtaining the xenonucleoside is a palladium (Xantphos-chelated) catalyzed N2-arylation (Buchwald-Hartwig reaction) of a fully protected 2'-deoxyguanosine derivative by 4-isobutyryloxybromobenzene. The reaction proceeded in good yield and the adduct was converted to the required 5'-O-DMT-3'-O-phosphoramidite by standard methods. The latter was used to synthesize oligodeoxynucleotides in which the N2-4-HOPh-dG adduct was incorporated site-specifically. The oligomers were purified by reverse-phase HPLC. Enzymatic hydrolysis and HPLC analysis confirmed the presence of this adduct in the oligomers.

Self-associating split fluorescent proteins (SAsFPs) have been widely used for labeling proteins, visualization of subcellular protein localization, scaffolding protein assembly and detecting cell-cell contacts. This PhD thesis research has been focused on expanding the toolbox of SAsFPs by adding more colors, increasing the brightness and empowering them with various biophysical properties. Utilizing a screening platform for the direct engineering of SAsFPs, we have generated a yellow–green split-mNeonGreen21–10/11, a 10-fold brighter red-colored split-sfCherry21–10/11 and demonstrated dual-color endogenous protein tagging using sfCherry211 and GFP11. However, the newly developed SAsFPs have suffered from sub-optimal fluorescence signal. By investigating the complementation process, we have employed two approaches to improve the overall brightness of SAsFPs: assistance through SpyTag/SpyCatcher covalent interaction and directed evolution for complementation-enhancing mutations. The latter strategy has then yielded two red-colored split sfCherry3 variants with substantially enhanced endogenous protein labeling performance. Based on sfCherry3, we have further developed a red-colored trans-synaptic marker called Neuroligin-1 sfCherry3 Linker Across Synaptic Partners (NLG-1 CLASP) for multiplexed visualization of neuronal synapses in living animals, demonstrating the broad applications of SAsFPs.

Accurately predicting cellular activities of proteins based on their primary amino acid sequences would greatly improve our understanding of the proteome. we present CELL-E, a text-to-image transformer model that generates 2D probability density images describing the spatial distribution of proteins within cells. Given an amino acid sequence and a reference image for cell or nucleus morphology, CELL-E predicts a more refined representation of protein localization, as opposed to previous in silico methods that rely on pre-defined, discrete class annotations of protein localization to subcellular compartments.

Synthetic peptide tags have been key to improving the ease with which scientists can interface with proteins in living systems. This PhD. work centers on demonstrating how peptide tags can be used to build modular platforms to accelerate live cell imaging, isolation of genetically engineered cells, cryo-EM, and applied DNA origami. First, we employed a set of three peptide/binder tagging systems, we developed tag-assisted splitHalo (TA-splitHalo), a versatile, multiplexable endogenous tagging platform. We applied TA-splitHalo sytems to live-cell imaging and complex knock-in sorting of lamin A/C knock-ins. Second, we utilized a peptide tag knock-in approach that combines split- fluorescent proteins and affinity purification to perform cryogenic electron microscopy (cryo-EM) on native structures derived from genetically edited suspension cell lines. This workflow was prototyped on native human proteasome complexes, demonstrating the pulldown of biologically relevant variants. Third and independently from the knock-in platforms, we engineered a SpyTag/SpyCatcher strategy in bacteria to create purification pipelines for proteins with two different and desired functional domains, enabling new ways to self-assemble biological nanoarchitectures. This was done by tethering cytokines to a transcription activator-like effector (TALE) protein to create immunomodulatory DNA binding proteins that can functionalize DNA origami structures. Our work demonstrates that the short length and minimally invasive nature of peptide tags enable plug-and-play platforms that can be applied at high-throughput for a wide range of applications.

Advancements in single-cell biology have revolutionized our understanding of complex biological systems by enabling high-throughput measurements of DNA, RNA, and proteins. This dissertation presents three innovative methodologies that advance single-cell analysis across these biomolecular domains.DNA: The FMR1 gene, essential for synaptic plasticity, contains a CGG repeat region where expansions (>200 repeats) cause Fragile X Syndrome. Existing methods to measure expansion lack single-cell resolution and spatial context. To address this, I developed CGG FISH, an RNA-FISH-based approach that quantifies CGG repeats by fluorescence intensity. Tested in humanized mouse models and human fibroblast cell lines, CGG FISH distinguished repeat lengths but faced challenges at low expression levels. Simulations revealed variability in probe binding and repeat heterogeneity. While offering insights into cell-to-cell variability, CGG FISH requires complementary methods to achieve finer resolution for small repeat differences. RNA: Single-cell RNA sequencing suffers from inefficiencies in reverse transcription, especially for low-abundance RNAs. Inspired by single molecule FISH, I developed HybriSeq, a method combining in situ hybridization, ligation, and combinatorial indexing to enable highly sensitive, targeted, and scalable RNA profiling. HybriSeq achieves high specificity and sensitivity by amplifying signals linearly with multiple probes to reduce noise. This technique offers a robust solution for high-throughput single-cell RNA analysis with improved scalability. Proteins: To enhance microscopy-based cytometry, I developed a method using partial trypsinization and the Cellpose algorithm for precise segmentation. This approach improves accuracy across cell densities, maintains spatial context, and excels in signal fidelity compared to flow cytometry. Its utility was demonstrated in optimizing prime editing for gene tagging, showcasing its potential in genetic engineering studies. Together, these methods address key challenges in single-cell analysis, expanding our ability to study cellular heterogeneity and biological complexity.

The key to understanding protein dynamics, localizations, and macromolecular architectures lies largely in intra-cellular visualization. Many of the limitations we face today are due to our inability to adequately label our protein of interest. Herein we describe solutions to some of these barriers through an improved super-resolution labeling technique. We further use our newly elaborated knock-in strategies to investigate the kinetics of endogenous protein kinase A (PKA) responses to beta-adrenergic signaling, and investigate newly observed recruitment of the endogenous kinase.

High-throughput screening (HTS) leverages laboratory automation and advanced analytics in order to perform experiments much more quickly than what can be accomplished by hand. Since its inception in the 1990s, HTS has revolutionized the field of drug discovery, and has been adopted by the leaders of the industry. However, the application of HTS technology has largely been limited to purified protein or cell-culture based assays, with detrimental consequences for the viability of lead compounds identified using this approach. In this dissertation, I discuss my work in developing new tools to implement HTS techniques in whole-organism models such as zebrafish and C. elegans. Whole organisms can be used to create more biologically-relevant disease models and to find cures for diseases with unknown etiology. Additionally, the use of whole-organism HTS opens up new possibilities in fields such as forward genetic screens. I describe how automated microscopy can be combined with computer vision and machine learning techniques in order to streamline the complicated logistics of imaging and analyzing whole organisms and whole-organism data. It is my hope that the tools and workflows detailed will increase adoption of whole-organism HTS and lead to improved drug development success rates as well as to facilitate scientific discovery.