DMA, University of Cincinnati, 2006, College-Conservatory of Music : Piano

Prokofiev's output consists of a large quantity of transcriptions which are arranged from his own work. Particularly during his Soviet period (1936–1953), there are increasing numbers of transcriptions from his ballet, opera, and even film scores. Many have more than one transcription. Transcriptions make up one third of his piano works. Except for four sonatas, Prokofiev's original piano compositions were replaced by piano transcriptions completely in the Soviet period. This document discusses Prokofiev's motivations for creating piano transcriptions, categorizes these transcriptions, and analyzes his transcribing style and techniques. As a result of his compositional habit, many of Prokofiev's works have several versions: original full score, piano reduction score, and orchestral and instrumental transcriptions. The composer's transcribing techniques are analyzed and discussed in decoration, refinement, simplification, organization, and melodic treatment through the comparison of different versions of the work.

Committee: Dr. Mara Helmuth (Advisor) Subjects: Music

Master of Science, The Ohio State University, 2025, Biochemistry

Pioneer transcription factors (pTFs) such as FoxA1 can bind nucleosomal DNA and open chromatin for other transcription factors (TFs), influencing enhancer activity and cell fate. Recent in vivo work has shown that FoxA1 and the canonical TF AP-1 (c-Fos/c-Jun) can bind cooperatively at certain enhancers, with AP-1 presence enhancing FoxA1 binding at weaker binding sites. The biochemical basis of this relationship, however, remains unclear. In this study, recombinant FoxA1 and AP-1 were tested for binding and nucleosome unwrapping on reconstituted mononucleosomes assembled with synthetic Widom 601 DNA containing adjacent FoxA1 and AP-1 binding motifs in two orientations (FA8 and AF8). Binding was assessed using electrophoretic mobility shift assays (EMSA), and nucleosome unwrapping was monitored with ensemble Forster resonance energy transfer (FRET). FoxA1 bound nucleosomes with relatively modest differences in affinity compared to free DNA and induced unwrapping of FA8 nucleosomes. AP-1 showed reduced nucleosome binding compared to free DNA but unexpectedly induced partial unwrapping of AF8 nucleosomes. However, sequential addition of both TFs did not produce cooperative binding or increased unwrapping under any tested condition. Instead, addition of the second TF often reduced binding affinity, suggesting anti-cooperative effects, potentially due to steric occlusion or competition for partially unwrapped DNA. Differences from in vivo findings may be due to the high stability of Widom 601 nucleosomes, absence of chromatin remodelers or histone modifications, DNA sequence context, or assay buffer composition. These results emphasize the gap between simplified reconstituted systems and native chromatin. The work in this thesis provides controlled biochemical foundation for dissecting the interaction between FoxA1 and AP-1 and highlights the role of chromatin complexity in TF cooperativity.

Committee: Michael Poirier (Advisor); Thomas Magliery (Committee Member) Subjects: Biochemistry

PhD, University of Cincinnati, 2020, Medicine: Molecular and Developmental Biology

Notch signaling is a form of short-range communication between neighboring cells that regulates many aspects of animal development and alterations in this signaling pathway have been associated with numerous disease states. At its simplest, ligand-receptor interactions at the cell membrane trigger sequential proteolytic cleavages of the Notch receptor to release the Notch intracellular domain (NICD) into the cytoplasm. Once released, NICD translocates into the nucleus, binds to the transcription factor CSL, recruits the co-factor Mastermind (Mam), and activates the transcription of target genes. In the absence of NICD, the same transcription factor CSL can bind to co-repressors to turn off transcription of the same target genes. Many Notch pathway genes show haploinsufficiency phenotypes, meaning that halving the dosage of Notch pathway genes fails to accomplish normal developmental processes. Haploinsufficiency implies that Notch-dependent processes are highly sensitive to alterations in signal strength. While all cells utilize the same Notch transcriptional cascade, haploinsufficiency phenotypes for the different Notch pathway genes manifest in distinct tissues, suggesting that Notch-dependent cell types have differential sensitivity to alterations in signal strength. However, the underlying mechanisms of such differential sensitivity are not well understood. Ultimately, Notch signaling has been shown to result in changes in gene expression via two types of CSL binding sites. The CSL monomer sites bind independent Notch transcription complexes, whereas CSL dimer sites, which are composed of two CSL binding sites orientated head-to-head and spaced 15 to 17 bp apart (known as Su(H) paired sites or SPSs), bind two cooperative Notch transcription complexes via NICD-mediated dimerization. Although it has been shown that NICD dimerization is required for optimal transcription of some Notch target genes, the impact of the NICD dimerization on SPSs versus NICD recruitment (open full item for complete abstract)

Committee: Brian Gebelein Ph.D. (Committee Chair); Kenneth Campbell Ph.D. (Committee Member); Stacey Huppert Ph.D. (Committee Member); Rhett Kovall Ph.D. (Committee Member); Masato Nakafuku M.D. (Committee Member) Subjects: Developmental Biology

Doctor of Philosophy, The Ohio State University, 2019, English

This dissertation investigates the writing practices of disability service transcribers, professional speech-to-text writers who translate sound into written text as a form of communication access. Scholars across the fields of disability studies, digital media studies, and communication studies have documented transcription as a complex rhetorical practice; however, the strategies speech-to-text writers use when composing in real-time environments, and the composing methodologies and technologies that underlie them, remain severely understudied. This lack of knowledge is significant because it constrains advocacy for disabled people and their allies in naming and understanding specific forms of real-time transcription that best increase access to spaces that depend upon sound to convey meaning. This lack also puts educators and their institutions at a disadvantage when working to make classrooms more accessible for disabled students, faculty, and staff. I intervene by applying a disability-studies lens to original composition-based research to ask: “How do different speech-to-text writers write, and why do they write the way they do?” To answer this question, this study is divided into two phases of qualitative and quantitative data collection: (1) a survey of speech-to-text writers working at a large, Midwestern university; and (2) a comparative survey of English-speaking speech-to-text writers across the United States and Canada. Collected qualitative data are coded and analyzed following Adele Clarke's (2005) methods for situational analysis, and qualitative data are analyzed following Cindy Johanek's (2000) adapted methodological model for mixed-methods research, also known as the “contextualist research paradigm” (p. 27). This study's most important findings reveal several major methodologies of transcription are currently in practice among professionals, and their differences tend to be hotly contested. Data show a speech-to-text writer's chosen method (open full item for complete abstract)

Committee: Margaret Price (Advisor); Christa Teston (Committee Member); Beverly Moss (Committee Member); Sean Zdenek (Committee Member) Subjects: Composition; Multicultural Education; Rhetoric; Technical Communication; Technology

PhD, University of Cincinnati, 2018, Medicine: Molecular and Developmental Biology

Differential gene expression in fundamental to an organism's ability to generate diverse cell types and maintain homeostasis. At the transcriptional level, gene expression is regulated by genomic sequences named cis-regulatory modules (CRMs) that bind combinations of sequence-specific transcription factor (TFs) proteins. Ideally, we could read CRM sequence to predict its regulatory activity – much like how we can predict protein from protein-coding DNA – but our predictive capabilities are significantly hindered by the complexity of the transcriptional process. For example, animals typically express over 1000 TFs, each of which binds degenerate DNA sequences. Morever, most TFs are members of large protein families that bind highly similar DNA sequences in vitro, yet perform very different in vivo functions. To help address this problem, I studied how the arrangement and affinity of TF binding sites contributes to the activity of two Hox-regulated CRMs in Drosophila. The RhoA CRM is a 47 bp enhancer element that drives the expression of rhomboid to stimulate Epidermal Growth Factor secretion from a specific subset of abdominal sensory organ precursors. By contrast, the Distalless-Conserved-Regulatory-Element (DCRE) is a transcriptional silencer that inhibits abdominal expression of the appendage-selector gene distal-less and thereby blocks leg formation in the Drosophila abdomen. RhoA is specifically activated by the Abdominal-A (Abd-A) Hox factor, while DCRE can be repressed by both Abd-A and another abdominal Hox factor, Ultrabithorax (Ubx). The activities of both CRMs also depend on the Hox co-factors Extradentical (Exd) and Homothorax (Hth). Using a combination of transgenic reporter assays, in vitro DNA binding assays, and computational TF binding site modeling, our work provides four general insights: First, the activities of these CRMs are constrained to particular binding site arrangements. The abdomen-specific activity of RhoA depends on overlapping bindin (open full item for complete abstract)

Committee: Brian Gebelein Ph.D. (Committee Chair); Vladimir Kalinichenko M.D. Ph.D. (Committee Member); Raphael Kopan Ph.D. (Committee Member); Rhett Kovall Ph.D. (Committee Member); Matthew Weirauch Ph.D. (Committee Member); Aaron Zorn Ph.D. (Committee Member) Subjects: Molecular Biology

Doctor of Philosophy, The Ohio State University, 2016, Biomedical Sciences

Biallelic mutation of the APC tumor suppressor gene occurs in a high percentage of colorectal tumors and is considered the critical event driving tumor initiation in the large intestine. The APC protein performs multiple functions, including negative regulation of the canonical WNT signaling pathway by both cytoplasmic and nuclear mechanisms. As a result, APC suppresses proliferation, cell cycle progression, and genomic instability, while facilitating differentiation, normal directional migration, and apoptosis. The contribution of APC to these phenotypes is not mediated exclusively through its effects on canonical WNT signaling, but also through WNT-independent functions of the APC protein. Intriguing reports that APC interacts with chromatin to repress key WNT-activated targets prompted this study's initial hypothesis that the chromatin-associated fraction of APC regulates gene transcription through multiple mechanisms, both WNT-dependent and WNT-independent. Chromatin immunoprecipitation and next-generation sequencing identified more than 6,000 genomic peaks associated with the APC protein. Target selection was performed by comparison to whole transcriptome sequencing data from APC-deficient and APC wild- type colon cancer cell lines and mouse tumors. 175 transcripts whose expression changes upon APC loss are linked to genomic regions physically associated with the APC protein. Motif analysis of APC-associated genomic peaks found that binding sites for the TCF7L2 and AP-1 transcription factors are overrepresented and occur in many of the same peaks. Luciferase reporter assays indicate that APC antagonizes canonical WNT signaling not only at WNT-activated genes such as PHLDB2 (Pleckstrin Homology Like Domain Family B Member 2), but also at WNT-repressed genes such as MALL (Mal, T-Cell Differentiation Protein-Like). A transcriptional element within the first intron of the MALL gene mediates transcriptional repression by the canonical WNT (open full item for complete abstract)

Committee: Joanna Groden PhD (Advisor); Albert de la Chapelle MD, PhD (Committee Member); Kay Huebner PhD (Committee Member); Mark Parthun PhD (Committee Member); Jeffrey Parvin MD, PhD (Committee Member) Subjects: Biomedical Research

Doctor of Philosophy, Case Western Reserve University, 2016, Pathology

Brain cancer is a devastating disease and a dreaded diagnosis for patients. Cancer within the brain, more than any other organ in the body, more than skin or colon or breast, is a difficult diagnosis to treat and live with because the brain is not just a part of the person, it is the person. Memories, personality, emotions, intelligence, speech and movement all originate in the brain. Therefore, when cancer, or the treatments used to fight the cancer, affect the brain by disrupting or damaging an area of the brain that controls these functions, it not only changes the body of the person, but changes the person. It is a particularly cruel disease. Glioblastoma is the most common and aggressive form of malignant brain cancer, which we currently have no way to effectively treat. The majority of patients die within 16 months of receiving a diagnosis, many times following aggressive surgery, radiation and chemotherapy, and then a sequelae of declining physical and cognitive function. Effective therapies are urgently needed for these patients. Many studies have identified potential therapeutic targets through in vitro screens or genomic sequencing, but this has not led to effective clinical therapies. Here we present a novel method of identifying potential drug targets in glioblastoma using advanced RNA interference screening in an in vivo orthotopic microenvironment. Using this system, we discover that screening within a functional microenvironment reveals novel targets that have been missed by traditional in silico or in vitro screening. We also demonstrate that our targets identified in vivo are more clinically relevant than targets we were able to identify in vitro using the same system. We focused our screening on epigenetic modifier genes, as many of these enzymes are sensitive to microenvironmental conditions and are potentially druggable. Within epigenetic modifier genes, we revealed that factors involved in transcription pause-release and elongation ar (open full item for complete abstract)

Committee: Jeremy Rich MD, MHSc, MBA (Advisor); Paul Tesar PhD (Advisor); Alan Levine PhD (Committee Chair); Peter Scacheri PhD (Committee Member); Bruce Trapp PhD (Committee Member); Steven Rosenfeld MD, PhD (Committee Member); Mark Jackson PhD (Committee Co-Chair); Clive Hamlin PhD (Committee Member) Subjects: Biology; Biomedical Research; Cellular Biology; Genetics; Medicine; Molecular Biology; Neurology; Pathology

Doctor of Philosophy, The Ohio State University, 2015, Microbiology

The Gram negative bacterium Pseudomonas aeruginosa is ubiquitous in the natural environment and responsible for various human infections. Successful P. aeruginosa infections require many virulence factors such as flagella, type IV pilus and alginate, whose expression is under tight control by transcription factors such as AmrZ. AmrZ activates type IV pilus-mediated twitching motility and alginate production while repressing the flagella machinery. However, molecular mechanisms of AmrZ-mediated activation of twitching motility and the alginate biosynthesis operon promoter (algD promoter) remain elusive and are the emphases of this study. Specifically, DNA microarray analyses were performed in order to identify AmrZ-dependent gene(s) that is required for twitching motility. The AmrZ regulon contains 112 genes, many of which are involved in virulence. However, this regulon does not contain a gene that is known to be necessary for twitching motility. In addition, we used multiple biochemical methods to show that AmrZ tetramerizes via its C-terminal domain, the loss of which results in diminished DNA binding and reduced efficiency during AmrZ-mediated activation and repression. Demonstration of AmrZ tetramerization also assists in understanding the activation mechanism of the algD promoter. This complex promoter contains a number of transcription factor binding sites with many in the far upstream region. It is unclear how those transcription factors from far upstream contribute to transcription activation. This study identified four AmrZ binding sites within this promoter and showed that each is required for algD transcription activation. We also illustrated phase-dependent activation of this promoter, suggesting interactions between upstream and downstream factors. Furthermore, in vitro fluorescence resonance energy transfer experiments indicate the formation of trans DNA-AmrZ complexes, suggesting that two DNA molecules are held together by AmrZ, likely via AmrZ ol (open full item for complete abstract)

Committee: Daniel Wozniak Ph.D. (Advisor); John Gunn Ph.D. (Committee Member); Irina Artsimovitch Ph.D. (Committee Member); Robert Munson Jr., Ph.D. (Committee Member) Subjects: Biology; Biomedical Research; Microbiology; Molecular Biology

Master of Science (MS), Ohio University, 2014, Computer Science (Engineering and Technology)

Within every cell is a copy of an organism's DNA. This copy of DNA has all of the information needed for the cell to express every gene in the organism's genome. Although each cell is capable, individual cells do not express every gene in their DNA. The genes expressed by a cell are regulated by transcription factors (TFs) that bind to a transcription factor binding site (TFBS) located in the promoter region of the gene. TFs must bind to TFBSs in order for a gene to be expressed. Tissues are groups of cells that perform a specific function; therefore, the cells of a specific tissue express genes that are not expressed in other cell types. Hydroxyproline-rich glycoprotein (HRGPs) are proteins that are found in the plant cell wall, and they can be further classified according to the degree they are glycosylated as arabinogalactan-proteins (AGPs), extensins (EXTs), and proline-rich proteins (PRPs). Currently, the TFBSs for EXTs, AGPs, and PRPs expressed in the pollen cells of Arabidopsis are unknown andtheir discovery will provide a better understanding of the regulatory and evolutionary processes of these genes. Motif discovery and other bioinformatics tools were used to search the promoter regions of EXT, AGP, and PRP genes expressed in the Arabidopsis pollen cells and select motifs that are putative TFBSs. The best set of motifs discovered as putative pollen-specific TFBSs are GCYAMGKA, ACTMGGAA, CATSAAAMGA, and ATTKGKTTCT. Of the 8 pollen-specific promoters, GCYAMGKA occurs in 5 promoters,ACTMGGAA occurs in 2 promoters, CATSAAAMGA occurs in 4 promoters, and ATTKGKTTCT occurs in 3 promoters. Also, all of the 8 HRGP pollen-specific promoters have anoccurrence of at least one of these four motifs and none of the four motifs occur in the 84 HRGP promoters of genes not expressed in pollen cells.

Committee: Lonnie Welch (Advisor) Subjects: Bioinformatics; Computer Science

Master of Health and Human Services, Youngstown State University, 2014, Department of Health Professions

The advent of the electronic medical record (EMR) has transformed the process of clinical documentation. When combined with the speech recognition technology (SRT), EMR completeness has increased over methodologies without this technology. This research examined chart audit completion scores of physicians and scribes working within four Northeastern Ohio Emergency Services departments. SPSS® Statistics were used to perform a Repeated Measure Analysis using paired-samples t tests calculated to compare mean completion scores one month prior versus six months after SRT implementation. The mean completion score of pre-SRT implementation with and without the assistance of scribes was 5.5 ( sd = .8) and the mean completion score of post-SRT implementation without the assistance of scribes was 6.0 (sd = .9) indicating a significant increase from pre-SRT versus post-SRT implementation (t(17) = -3.9, p < 0.5). The mean completion score of pre-SRT implementation without the assistance of scribes was 5.0 (sd = 1.1) and the mean completion score of post-SRT implementation without the assistance of scribes was 6.0 (sd = .9) also indicating a significant increase from pre-SRT versus post-SRT implementation (t(17) = -4.7, p < 0.5). These analyses validated the strong statistical probability that the completeness scores of physicians utilizing SRT will exceed the total completeness scores of physicians and scribes not using this technology. Subsequently, the null hypotheses were rejected in support of the alternative hypotheses, which concluded: 1) The completeness of the EMR will at least remain the same or improve with the implementation of SRT. 2) The completeness of the EMR will at least remain the same or improve when speech recognition technology is used without scribe utilization.

Committee: Joseph Lyons PhD (Advisor); Ronald Chordas PhD (Committee Member); Richard Rogers PhD (Committee Member) Subjects: Audiology; Communication; Health Care; Health Care Management; Health Education; Information Systems; Information Technology; Medicine; Nursing

DMA, University of Cincinnati, 2008, College-Conservatory of Music : Trombone

This document examines the pedagogical and performance uses of the songs of Gustav Mahler. It also includes transcribed trombone versions of 32 of Mahler's songs, written to both emulate the original vocal pieces and to take advantage of the trombone's range, technical facility and dynamic capabilities. These transcriptions can be performed with the existing vocal/piano scores. Gustav Mahler's songs work particularly well as instrumental transcriptions because of his beautiful melodies, his symphonic compositional style, his emphasis of thematic development over text declamation, and his inclusion of so many expressive elements into the accompaniments of his songs. They are particularly well suited to performance on trombone because many are written in a baritone register, using the full range of dynamics and articulation styles to express various emotions and settings.The transcribed song is truly a neglected source of music for trombonists, and both student and professional trombonists have a great deal to gain by including transcriptions of Mahler's songs in their repertoires. By studying these song transcriptions, trombonists can improve their fundamental concepts of legato style, phrasing, and musical expression, leading to a singing style of playing. And, they can include pieces by one of the master composers in their performance programs.

Committee: Timothy Northcut MM (Committee Chair); Timothy Anderson MM (Committee Member); Robert Zierolf PhD (Committee Member) Subjects: Music

PhD, University of Cincinnati, 2026, Arts and Sciences: Chemistry

This dissertation explores the application of molecular dynamics simulations, statistical modeling, and interpretable machine learning approaches to understand the structural dynamics and functional regulation of biologically important macromolecular systems. The work focuses on three major areas: protein quality control machinery, transcription factor-DNA interactions, and sequence-based machine learning models for biosensing applications. The first part of this dissertation investigates the bacterial protease ClpP, a central component of the cellular protein quality control system. Extensive molecular dynamics simulations combined with machine learning approaches were used to characterize the conformational transition associated with pore opening. Interpretable machine learning models identified key molecular interactions, including intraloop hydrogen bonds, native contacts, and salt-bridge interactions, that regulate the open and closed conformational states of ClpP. These findings provide mechanistic insight into the structural determinants governing substrate access and proteolytic activity. The second part focuses on the AAA+ disaggregase Hsp104 and the conformational dynamics underlying its functional cycle. Molecular dynamics simulations and Markov State Model analyses were employed to characterize long-timescale motions and metastable conformational states associated with protein disaggregation activity. Dynamical analyses revealed coordinated motions between functional domains and provided insight into the allosteric mechanisms that regulate substrate processing in AAA+ molecular machines. This dissertation also investigates the effects of disease-associated mutations in the FOXF1 transcription factor. Comparative molecular dynamics simulations of wild-type and mutant FOXF1 systems in both DNA-bound and unbound states were used to examine changes in structural stability, flexibility, intermolecular interactions, and DNA-binding behavior. The re (open full item for complete abstract)

Committee: George Stan Ph.D. (Committee Chair); Ryan White Ph.D. (Committee Member); Pietro Strobbia Ph.D. (Committee Member) Subjects: Artificial Intelligence; Biophysics; Molecular Biology

PhD, University of Cincinnati, 2026, Medicine: Development, Stem Cells & Regenerative Medicine

Neural crest cells are a highly potent population of cells that contribute to many different organ systems throughout the developing embryo. Despite being an ectodermal derived population, the neural crest in the cranial region of the developing embryo (cranial neural crest cells) can give rise to both non-ectomesenchymal (traditionally ectodermal) and ectomesenchymal (traditionally mesodermal) cell types. In this dissertation, we explore potential roles for the SHH signaling pathway in neural crest development in various different ways. First, we looked to see whether the SHH signaling pathway contributed to neural crest induction using three model systems: chick embryos, human embryonic stem cells differentiated into neural crest, and mouse models. We find that genes in the SHH signaling pathway are expressed in developing neural crest and that specifically, the Gli transcription factors are expressed in developing neural crest in vivo. Next, we show that loss of GLI3 led to reduced neural crest specifiers in chick and human, but not mouse. To determine a role for GLI3 in mouse neural crest development, CUT&RUN for GLI3 in later stage mandibles showed that GLI3 binding was associated with neural crest differentiation. Integration with a histone mark ChIP-seq dataset showed that GLI3 binding was highly associated with H3K27ac, specifically at genes involved in ectomesenchymal differentiation. Single-cell RNA sequencing of Gli TF knockout mandibular prominence showed that loss of Gli TFs led to specifically a reduction in neural crest differentiation into ectomesenchymal, but not non-ectomesenchymal cell types. Second, we looked to see whether the SHH signaling pathway conveyed potency in cranial neural crest cells. We show that the Gli TFs are co-expressed in developing neural crest cells using publicly available genomics datasets. We show that Oct4 is expressed in developing neural crest cells in vivo. However, loss of Gli TF or SHH signaling did not lead to a re (open full item for complete abstract)

Committee: Samantha Brugmann Ph.D. (Committee Chair); Brian Gebelein Ph.D. (Committee Member); Matthew Weirauch Ph.D. (Committee Member); William Zacharias M.D. Ph.D. (Committee Member); Aaron Zorn Ph.D. (Committee Member); Nathan Salomonis Ph.D. (Committee Member) Subjects: Developmental Biology

Master of Science (MS), Ohio University, 2026, Computer Science (Engineering and Technology)

The effects of each individual transcription factor (TF) on chromatin structure and gene expression are not well known. Research into this question began almost 10 years ago with the study of TF motif pair binding sites in a single line of human cells. Recently, study was expanded into the comparison of binding site presence between two conditions, but there is still much left to be answered. This thesis expands on previous research with the presentation a of feature selection algorithm, MALT, which uses Fisher's exact test to find discriminatory TF motif pair binding sites in more than two conditions. To demonstrate the effectiveness of MALT, it was applied to two case studies. In the first case study, MALT examined four conditions related to sleep deprivation and selected 472 TF pairs with p-values below 0.05 in one-vs-one comparisons and below 0.001 in one-vs-rest comparisons. TFs in the resulting features include potential therapeutic targets for neurodevelopmental disabilities and acute sleep deprivation in neurotypical people. The significance of results is further corroborated with gene ontology studies. In the second case study, MALT was used to examine mouse embryonic stem cells, selecting 283 TF pairs which are discriminatory and enriched in their respective conditions. The significance of the results is corroborated with gene ontology studies and the results indicate there are many aspects of the 3D genome that are not controlled by Ctcf. MALT is the first multicondition pipeline to be applied to the motif pair problem. The development of a multicondition model enables both faster and more nuanced examination of chromatin contact information.

Committee: Lonnie Welch (Advisor) Subjects: Biostatistics; Computer Science

PhD, University of Cincinnati, 2025, Medicine: Systems Biology and Physiology

The sequencing of the human genome marked a transformative milestone in biology, yet it also revealed a new frontier: understanding how non-coding regions of DNA regulate gene expression. Central to this regulation are transcription factors (TFs), proteins that bind specific DNA sequences to orchestrate transcriptional programs defining cellular identity, physiology, and disease states. TF binding sites can be identified by a number of assays such as Chromatin Immunoprecipitation (ChIP-seq), but identifying all known TF binding sites (TFBS) is limited due to scale, cost, and feasibility across diverse cell types. These constraints have fueled the development of computational methods to infer TF binding from genomic data. Recent advances in machine learning, particularly deep learning, have revolutionized regulatory genomics by enabling models to learn complex sequence patters, chromatin context, and higher-order interactions directly from data. Architectures such as dilated convolutional and transformer-based neural networks have demonstrated powerful capabilities in predicting chromatin accessibility, enhancer-promoter interactions, and variant effects. The ENCODE-DREAM challenge of 2017 was an in vivo transcription factor binding site prediction challenge designed to evaluate and advance computational methods for predicting TFBS. This inspired many research groups, including ours, to improve performance of deep learning TFBS prediction methods and increase the number of TF models available. In Chapter 2, we describe the development of maxATAC, a suite of deep neural network models for genome-wide TFBS prediction using ATAC-seq in any cell type. This framework includes models for 127 human transcription factors, the largest collection of high-performance TFBS prediction models for ATAC-seq. maxATAC performance extends to primary cells and single-cell ATAC-seq, enabling improved TFBS prediction in vivo. We demonstrate maxATAC's capabilities by identifying (open full item for complete abstract)

Committee: Emily Miraldi Ph.D. (Committee Chair); Christian Hong Ph.D. (Committee Member); Leah Claire Kottyan Ph.D (Committee Member); Surya Prasath Ph.D. (Committee Member); Matthew Weirauch Ph.D. (Committee Member); Tongli Zhang Ph.D. (Committee Member) Subjects: Bioinformatics

PhD, University of Cincinnati, 2025, Medicine: Development, Stem Cells & Regenerative Medicine

Dysregulated immune responses to common viruses are implicated in numerous chronic diseases. Epstein-Barr virus (EBV) is a ubiquitous virus that typically presents as an asymptomatic infection. However, EBV can drive cancers and some autoimmune diseases in a small subset of individuals. Of the ~95% of people who have had a prior EBV infection, the vast majority will not develop subsequent autoimmune diseases. EBV is the leading cause of multiple sclerosis (MS), but the underlying mechanisms are not well characterized. The goals of this study are to begin to clarify the interactions between EBV and MS-associated genetic risk variants in the human genome that lead to disease, and identify pathways with altered gene regulation in people who have MS compared to those who do not. Genome-wide association studies (GWAS) are a common method for identifying genetic risk variants associated with a disease. GWAS have identified 233 independent MS risk loci; the majority of these risk variants are common, meaning they can also be present in unaffected individuals. In addition, GWAS identify tag variants that are proxies for many regions of the genome that are inherited together in genetic haplotypes. Over 95% of MS-associated variants are in non-coding regions of the genome, further complicating efforts to study the disease mechanisms impacted by this genetic variation. In chapter 2, we use massively parallel reporter assays to simultaneously investigate all known MS genetic risk variants in EBV-transformed B cells. In this study, we identify 83 independent risk loci with genotype-dependent transcriptional regulatory activity. We use publicly available ChIP-seq data and eQTLs to identify potential regulatory mechanisms and biological pathways that are affected by these allelic regulatory variants. We find that enhancers and silencers are uniquely enriched for binding events for different transcription factors and find that these variants potentially regulate numerous gene (open full item for complete abstract)

Committee: Matthew Weirauch Ph.D. (Committee Chair); Leah Claire Kottyan Ph.D. (Committee Chair); Kenneth Campbell Ph.D. (Committee Member); Brian Gebelein Ph.D. (Committee Member); Kristen Fisher D.O. (Committee Member) Subjects: Bioinformatics; Biology; Genetics

Doctor of Philosophy, The Ohio State University, 2025, Molecular, Cellular and Developmental Biology

Quantitative comparisons of protein-DNA interactions across experimental conditions remain technically challenging for the study of chromatin regulation. Such comparisons may be pivotal for understanding diseases and cancers in which the dysfunctional regulation of chromatin is particularly implicated. One such cancer is pediatric fusion positive rhabdomyosarcoma. Despite extensive similarities, rhabdomyosarcoma tumors have substantially different patient prognoses when driven by either of the two most common fusions, PAX3::FOXO1 and PAX7::FOXO1. While PAX3::FOXO1 is associated with poorer outcomes for patients than PAX7::FOXO1, the mechanisms behind this gap remain elusive. To address these unmet needs in the field, we developed the PerCell ChIP-seq methodology and an accompanying bioinformatic pipeline. This workflow uses the spike-in of orthologous cells to normalize levels of histone modification or transcription factor localization. We also sought to bridge the gaps in our knowledge of fusion positive rhabdomyosarcoma by defining PAX3/7::FOXO1's modes of engaging with chromatin. We have begun the work of contrasting the fusions by examining how each binds DNA that is either freely accessible or wrapped around a nucleosome. We resolved these categories of binding in silico using a modified MNase-based ChIP-seq method. We have further meta-analyzed this data with each fusion's endogenous chromatin contexts, including the positioning of nucleosomes and histone modifications. While these comparisons have identified a shared ability to interact with nucleosomes, they have also revealed distinct preferences for motif sequences, methods of engaging nucleosomes, and classes of chromatin at which they bind. Overall, this work facilitates quantitative comparisons of genome-wide protein binding of DNA across distinct chromatin backgrounds, experiments, and lab groups. We also provide a framework for defining how subtle differences in related transcription fact (open full item for complete abstract)

Committee: Benjamin Stanton (Advisor); Peter White (Committee Member); Emily Theisen (Committee Member); Dawn Chandler (Committee Member) Subjects: Bioinformatics; Biomedical Research; Molecular Biology; Oncology

PhD, University of Cincinnati, 2025, Medicine: Development, Stem Cells & Regenerative Medicine

The basal ganglia are subcortical nuclei which modulate thalamic activity to regulate motor control and certain aspects of emotion and cognition. Disorders involving impaired motor control arise from degeneration (i.e. Parkinson's and Huntington's disease) or atypical development/function (i.e. Tourette's syndrome) within these structures. The primary component of the basal ganglia is the striatum (caudate-putamen), which is composed primarily of GABAergic medium spiny neurons that project to either the globus pallidus and participate in the indirect pathway (iSPNs) or to the substantia nigra within the direct pathway (dSPNs); proper regulation of movement relies on the balance between these two pathways. SPNs are derived from the embryonic lateral ganglionic eminence (LGE), while the medial ganglionic eminence (MGE) produces striatal interneurons. Many transcription factors are involved in the maintenance and specification of telencephalic progenitor domains; one such TF expressed within the LGE and MGE is the homeobox TF Gsx2. Loss of Gsx2 leads to a significant reduction in LGE-derived cells including SPNs and early postnatal death from respiratory distress. A 2019 report detailed two patients with GSX2 variants, including one missense variant within the homeodomain (c.752A>G; p.(Q251R)) and another variant leading to a severe truncation resulting in a null protein (c.27G>A; p.(S9*)). Both patients had severe clinical findings, including spastic tetraparesis, dystonia, and intellectual impairment, and imaging revealed significant reductions in striatal size and atypical corticospinal axon trajectories. Previous studies have focused on the forebrain phenotype in Gsx2 null mice, but the p.(Q251R) patient provides a unique opportunity to model a pathological human variant with presumed mutant protein expression in mice. We therefore generated the homologous mouse mutation (Gsx2Q252R) and investigated the molecular consequences and its effects on embryonic and postna (open full item for complete abstract)

Committee: Kenneth Campbell Ph.D. (Committee Chair); Brian Gebelein Ph.D. (Committee Member); Robert Hufnagel M.D. Ph.D. (Committee Member); Makiko Iwafuchi Ph.D. (Committee Member); Jason Tchieu Ph.D. (Committee Member); Ronald Waclaw Ph.D. (Committee Member) Subjects: Developmental Biology

Doctor of Philosophy, Case Western Reserve University, 2025, Biochemistry

In eukaryotic cells, DNA is packaged into chromatin to facilitate organization within the nucleus. The basic unit of chromatin is the nucleosome, which consists of 147 base pairs of DNA wrapped around a histone octamer. The tails of histone proteins undergo various post-translational modifications, which can modify chromatin structure and/or influence gene expression. These dynamic histone modifications are orchestrated by numerous histone-modifying enzymes, including writers that deposit modifications and erasers that remove. This thesis reveals a novel mechanism underlying the non-canonical roles of these histone modifiers and their corresponding histone modifications in transcriptional regulation. Specifically, the roles of LSD1, an H3K4me1/2 demethylase, are characterized, including its catalytically dependent and -independent functions in gene regulation and cell fate transition. Interestingly, the catalytic inactivation of LSD1 has only a mild impact on gene expression and cell fate transition, whereas LSD1-null cells exhibit global enhancer derepression and impaired cell fate transition. Moreover, LSD1 loss leads to increased H3K27 acetylation, enrichment of p300/CBP, an H3K27ac acetyltransferases, at enhancers, and transcriptional derepression, ultimately causing differentiation defects. Previously, we identified the non-catalytic role of MLL4's N-terminal (NT) domain, an H3K4me1/2 methyltransferase, in regulating gene expression. However, its underlying mechanism is still elusive. Using the proximity labeling technique TurboID, I demonstrate the proximity and interaction between MLL4 NT and the H3K36 methyltransferase NSD1. Similar to the catalytic inactivation of NSD1, MLL4 deletion results in decreased H3K36me2 levels and increased H3K27me3 levels at MLL4-targeted enhancers, and co-regulated genes. Finally, a synergistic role between H3K79me and H3K36me3 in transcriptional regulation is uncovered. Simultaneous catalytic inactivation of DOT1L, (open full item for complete abstract)

Committee: Kaixiang Cao Dr. (Advisor); Peter Harte Dr. (Committee Member); Hung-Ying Kao Dr. (Committee Member); Joseph Luna Dr. (Committee Chair) Subjects: Biochemistry; Biomedical Research

Doctor of Philosophy, University of Toledo, 2025, Biology (Cell-Molecular Biology)

Gfi1 is a transcriptional repressor that plays a critical role in hematopoiesis. The repressive activity of Gfi1 is mediated mainly by its SNAG domain that interacts with and thereby recruits the histone demethylase LSD1 to its target genes. An important function of Gfi1 is to protect hematopoietic cells against stress-induced apoptosis, which has been at-tributed to its participation in the posttranscriptional modifications of p53 protein, leading to suppression of p53 activity. In this study, we show that Gfi1 upregulated the expression of Hemgn, a nuclear protein, through a 16-bp promoter region spanning from +47 to +63 bp relative to the transcription start site (TSS), which was dependent on its interaction with LSD1. We further demonstrate that Gfi1, Ikaros and PU.1 bound to this 16-bp region. However, while Ikaros activated Hemgn and collaborated with Gfi1 to augment Hemgn expression, it was not required for Gfi1-mediated Hemgn upregulation. In contrast, PU.1 repressed Hemgn and inhibited Hemgn upregulation by Gfi1. Notably, PU.1 knockdown and deficiency, while augmenting Hemgn expression, abolished Hemgn upregulation by Gfi1. PU.1 (Spi-1) has been shown to be repressed by Gfi1. We show here that PU.1 re-pression by Gfi1 preceded and correlated well with Hemgn upregulation. Thus, our date strongly suggest that Gfi1 upregulates Hemgn by repressing PU.1. In addition, we demonstrate that Hemgn upregulation contributed to the anti-apoptotic activity of Gfi1 in a p53-independent manner. Conversely, restoring Hemgn expression in Gfi1-deficient bone mar-row cells reduces hypersensitivity to DNA damage induced apoptosis. We also show that Gfi1 and DNA damage work together to enhance Hemgn expression independently of p53. In summary, these results suggest a mechanism in which Gfi1 enhances Hemgn expression by repressing PU.1, thereby promoting Gfi1's anti-apoptotic role following DNA damage. Furthermore, we show here that Gfi1 inhibited apoptosis induced not only by (open full item for complete abstract)

Committee: Fan Dong (Committee Chair); Guofa Liu (Committee Member); Deborah Chadee (Committee Member); Kam Yeung (Committee Member); Shi-He Liu (Committee Member) Subjects: Biology; Cellular Biology; Molecular Biology