Late-onset Alzheimer's disease (LOAD) is a multifactorial disease with a strong genetic component. The growing understanding of the genetic basis and molecular mechanisms underlying LOAD risk presents an opportunity to uncover the factors that counter the risk and protect individuals from developing LOAD. The phenomenon wherein individuals demonstrate adaptability to the burden of disease risk can be referred to as "resilience". In this dissertation, I presented three studies that focused on the resilience to LOAD. Because resilience depends on and interacts with risk, we employed a risk-informed strategy to uncover resilience factors. This approach leveraged the current best-estimated LOAD risk to identify resilient individuals who, despite facing the highest LOAD risk, exhibit no dementia symptoms in old age. In Chapter 1, we demonstrated that a large number of risk-independent common genetic variants could reduce the penetrance of heightened genetic risk burden in LOAD. This study provided insights into the genetic architecture of resilience to LOAD, addressing a significant knowledge gap that requires attention. In addition, this study yielded a polygenic resilience score, enabling the assessment of the relative genetic resilience levels among individuals. In Chapters 2 and 3, we explored resilience to LOAD at the transcriptomic level. The study in Chapter 2 meta-analyzed all publicly available blood and brain transcriptomic studies of AD. This study laid the groundwork for investigating the resilience-conferring genes and pathways by establishing the best-estimated transcriptomic risk features in LOAD. In Chapter 3, we capitalized on the transcriptomic risk defined in Chapter 2 and examined the risk-residual genes that might confer resilience to increased transcriptomic risk of LOAD. This study implicated a couple of interesting pathways in resilience to LOAD and suggested that resilience and risk may operate in the same biological pathways. Taken together, our findings corroborated the idea that resilience in LOAD has a polygenic basis and highlighted the need to gain a deeper understanding of the genetic components, biological mechanisms, and phenotypic characteristics of resilience to LOAD risk. The dissertation contextualized these findings with the existing literature and suggested potential future directions to help further address the gaps in understanding resilience in LOAD.

This thesis investigates the developmental processes in the mouse cortex during embryonic days 13 to 15, corresponding to human gestational weeks 8 to 11. The primary focus is on the calcium signaling pathways in neurons, particularly as they transition from the migratory phase to the dendrite initiation and growth phase. The study places special emphasis on Cajal-Retzius neurons (CRNs), a transient population of excitatory neurons in the developing cortex. These neurons are known for their secretion of Reelin, a glycoprotein that has a role in regulating the position of glutamatergic neurons of the cortex. While the role of Reelin and its downstream effects on cortical organization have been well-documented, this work investigates a previously unexplored transient circuit between CRNs and cortical projection neurons (CPNs). The central questions addressed in this thesis include: What is the calcium signal as neurons transition from the migratory phase to dendrite initiation and growth? Is the signal important for CPN maturation and dendrite growth? How does the activity and neurotransmitter release by CRNs serve as a mechanistic substrate that informs and coordinates CPN development during the foundational phases of cortical development? What neurotransmitter systems are involved and how are they functioning? Our research employs a multidisciplinary approach, leveraging whole embryonic hemisphere explants and multiphoton microscopy to study the calcium signaling profiles of CRNs and CPNs. The findings reveal that CRNs not only secrete Reelin but also exhibit spontaneous activity and the potential for neurotransmitter release, specifically glutamate. This activity significantly influences intracellular calcium levels in CPNs, thereby affecting their dendritic growth and migration patterns. This work opens up a new avenue for understanding early cortical development by offering a novel framework that extends beyond the established role of CRNs and Reelin secretion. It provides compelling evidence that CRNs play a more multifaceted role in cortical development than previously thought, serving in a transient circuit that informs and coordinates CPN development. This thesis, therefore, not only fills a gap in our understanding of early cortical development but also sets the stage for future research into the pathophysiology of neurodevelopmental disorders.

Heat shock protein-90 (Hsp90) is an essential molecular chaperone responsible for the stability and activation of over 300 client proteins, many of which have been implicated in tumorigenesis. The chaperone function of Hsp90 is dependent on its ability to hydrolyze ATP. The ATP hydrolysis cycle, and therefore chaperone activity, is tightly regulated by a group of proteins known as co-chaperones and post-translational modifications. Folliculin-interacting proteins 1 and 2 (FNIP1 and FNIP2) are recently recognized Hsp90 co-chaperones with 74% amino acid sequence homology and largely overlapping reported functions. Another recently identified co-chaperone with similar functionality is the tumor suppressor Tsc1. FNIPs and Tsc1 co-chaperone activities have been characterized, however there remains a gap in understanding the functional differences between FNIP1, FNIP2 and Tsc1. Here, we dissected the impact of these newly identified co-chaperones towards Hsp90 client activity. Our data show that Tsc1 and FNIP2 form mutually exclusive complexes with FNIP1 and that unlike Tsc1, FNIP1/2 interact with the catalytic residue of Hsp90. We further demonstrate that FNIP1 and Tsc1 interact with a large set of overlapping proteins, whereas FNIP2 interactors are predominantly unique. Functionally, these co-chaperone complexes increase the affinity of the steroid hormone receptors glucocorticoid receptor and estrogen receptor to their ligands in vivo. We provide a model for the responsiveness of the steroid hormone receptor activation upon ligand binding as a consequence of their association with specific Hsp90:co-chaperone subpopulations. The results presented here demonstrate physiological and mechanistic differences between the co-chaperones FNIP1, FNIP2 and Tsc1.

Fragile X syndrome (FXS) is one of the most prevalent forms of inherited intellectual disability and is the leading monogenetic cause of autism spectral disorder. FXS is caused by lack of expression or mutations of the FMR1 gene which encodes fragile X messenger ribonucleoprotein, FMRP. Recently, FMRP has been shown to undergo liquid-liquid phase separation (LLPS) in vitro and to localize different isoforms in distinct membrane-less organelles in cellulo. Despite three decades of research, the molecular mechanisms by which FMRP functions are still not fully understood. FMRP is best known as a cytoplasmic mRNA-binding translational regulator. Although the presence of a small fraction of FMRP in the nucleus has long been realized, it was only recently that studies are beginning to uncover its role in influencing genomic function and stability [1]. The Feng lab recently discovered a novel genome protective role for FMRP. FXS patient-derived cells undergo higher level of DNA double-strand breaks (DSBs) than normal cells, especially during DNA replication stress. These DSBs occur at sequences prone to forming R-loops, which are co-transcriptional RNA:DNA hybrids associated with multiple functions including genome instability. Exogenously expressed WT FMRP, but not an I304N disease-causing mutant abates R-loop-induced DSBs. This unexpected finding suggests that FMRP promotes genome integrity by preventing R-loop accumulation and chromosome breakage. However, the mechanism by which FMRP performs this critical function, and how disease-causing mutations affect this process is not fully understood. Here, we set out to elucidate the mechanism underlying FMRP's role in maintaining genome stability. First, we demonstrate that FMRP directly binds R-loops primarily through its C-terminal Intrinsically Disordered Region (C-IDR). In FMR1 CRISPR knock-out HEK293T cells, we observed dynamic condensates in WT FMRP but not in I304N mutant, suggesting that this mutation, located in the central RNA binding KH2 domain, disrupted the ability of I304N to assemble into higher order condensates. Furthermore, unlike the I304N FMRP, WT FMRP show increase in nuclear condensates that overlap with R-loops under replication stress. While we found that WT and I304N mutant can co-phase separate with R-loops in vitro, WT FMRP tends to form hollow droplets with R-loop substrates localized at the periphery, but the vast majority of I304N droplets are filled with dispersed R-loops substrates. Taken together, these data support the hypothesis that the ability of FMRP to form higher order assemblies with R-loops is critical to maintaining genome stability. Our study sets the stage to test the proposed phase separation-function paradigm in other FXS disease mutants.

Modeling and computational analysis can be used to crystallize, integrate, and extract knowledge from large datasets generated by biology, medicine, and next-generation sequencing. The use of probability models, multifactor hypothesis testing, and computational analyses is crucial to studies in systems biology. These studies provide insights into understanding large and diverse molecular biology data sets. It is no longer enough to study individual molecules, their properties, and their interactions with other molecules in cells and organisms. In addition to generating numerous case studies with unique data, such studies provide a limited understanding of the underlying complexity and dynamics of the leading mechanisms determining the states and behaviors of a whole biological system. Sequencing and multi-omics experiments generate big data needed to model processes, organization and behavior of biological systems in a more comprehensive, less biased manner. Analysis of such enormously heterogeneous and complex information requires mathematical models and computational algorithms. It is the motivation and challenge of current systems biology and medicine. Applied to cancer systems biology, we will consider basic probabilistic aspects of big data. We study skewed frequency distributions commonly observed in diverse omics experiments. We focus on modeling and developing computational algorithms to quantify big data's statistical characteristics, aiming for accurate and unbiased characterization of the systems variation. In several applications, we focus on the identification of the skewed distributions for quantification and differentiation of the of R-loop formation patterns in non-cancer, pre-malignant states and cancer genomes. Current studies involving R-loops rely on the S9.6 antibody which generates noisy signals. We show that using R-loop forming sequences for filtering specific S9.6 signals selects biologically meaningful signals. R-loops have been shown to play a role in tumorigenesis. Using our R-loop forming sequence enrichment method, we investigate the roles of R-loops in tumorigenesis across different detection modalities primarily in breast cancer.

Schizophrenia is a severe mental disorder that has been associated with dysregulation of the immune system. Using inflammatory cytokine transcript levels, we found approximately 40% of individuals with schizophrenia are classified as having elevated levels of inflammation with worse neuropathology. The aim of this study was to investigate the extent to which neuroinflammation is associated with schizophrenia in the dorsolateral prefrontal cortex (DLPFC) and midbrain. We used postmortem human brain tissue to investigate multiple inflammation-related molecular mechanisms. First, in the DLPFC, we found macrophages and astrocytes, rather than microglia, contributed more to neuroinflammation in schizophrenia, by showing unchanged or decreased microglial markers and elevated macrophage markers. Macrophage marker expression was more related to pro-inflammatory marker and macrophage recruitment chemokine expression. In addition, we classified "high" and "low" inflammatory (HI and LI) subgroups using inflammatory cytokine and macrophage marker protein levels as discriminators in the DLPFC for the first time. 30% of controls (CTRL) and 56% schizophrenia (SCZ) cases were classified as high inflammation individuals. We found higher CD163+ macrophage density in the DLPFC of SCZ-HI subgroup mainly around small blood vessels of both grey and white matter. In the midbrain, we characterized a substantial proportion of individuals in schizophrenia (46%) and bipolar disorder (29%) expressing elevated inflammatory mRNA (IL6, IL1, TNF and SERPINA3), which were termed the "high" inflammatory (HI) subgroups, and we confirmed increases in IL6 and IL1 at the protein level in these subgroups. Furthermore, we showed elevated macrophage and chemokine marker expression in schizophrenia and bipolar disorder HI subgroups which were associated with changed blood-brain barrier markers, indicating potential macrophage transmigration, supported by altered mRNA and protein levels in the adhesion molecules, tight junction proteins, basement membrane proteins, and angiogenic factors related to blood vessel regulation. Importantly, we observed a novel but unknown neuropathology of more frequent claudin-5 "cell bursts" between blood vessel fragments in schizophrenia and bipolar high inflammatory subgroups. These findings have implications for new immune-related treatments, therapy development, and potential targets for measuring disease progression or early detection of schizophrenia and bipolar disorder.

Eukaryotic DNA-dependent RNA Polymerases (Pols I-III) encode two distinct ⍺- like heterodimers where one is shared between Pols I and III, and the other is unique to Pol II. Human alpha-like subunit mutations are associated with several diseases including Treacher Collins Syndrome (TCS), 4H Leukodystrophy, and Primary Ovarian Sufficiency. Yeast is commonly used to model human disease mutations, yet it remains unclear whether the alpha-like subunit interactions are functionally similar between yeast and human homologs. To examine this, we mutated several regions of the yeast and human small alpha-like subunits and used biochemical and genetic assays to establish the regions and residues required for heterodimerization with their corresponding large alpha-like subunits. Here we show that different regions of the small alpha-like subunits serve differential roles in heterodimerization, in a polymerase- and species-specific manner. We found that the small human alpha-like subunits are more sensitive to mutations, including a "humanized" yeast that we used to characterize the molecular consequence of the TCS-causing POLR1D G52E mutation. These findings help explain why some alpha subunit associated disease mutations have little to no effect when made in their yeast orthologs and offer a better yeast model to assess the molecular basis of POLR1D associated disease mutations.

Each year hundreds of thousands of people develop life-threatening sepsis, defined by a combination of infection and organ dysfunction. Although many affected biological pathways are typically regulated by the glucocorticoid receptor (GR), during sepsis this is deficient and supplementation with exogenous glucocorticoids is often ineffective in reducing mortality. The GR has different effects in different organs. In liver many effects are beneficial, whereas in immune or muscular systems many effects are deleterious. A sampling of this literature is reviewed in chapter 1. With the hypothesis that liver-specific glucocorticoid therapy will be more clinically beneficial than systemic therapy, we studied liver-specific GR effects in infection, inflammation, or sepsis. Chapter 2 describes in-vitro chemokine alterations from GR activation in primary human hepatocytes, with inflammation or infection modeled by TNFα and/or lipopolysaccharide. Comparisons were made in primary human hepatocytes, the human hepatoma cell line HEPG2, and the non-liver cell line HEK293t. Chapter 3 outlines outcomes of liver-specific GR deficiency using mice with inducible liver-specific GR knockout, modeling sepsis with cecal ligation and puncture. Results of these two models show mRNA and/or protein changes induced by GR in chemokines; transcription factors; genes related to protein degradation, metabolism, metal management, inflammation, liver regeneration, and hemodynamic stability. Results of these 2 models demonstrate a significant role of the hepatic GR in many pathways dysregulated during sepsis. Therefore, targeting GR therapy to the liver instead of systemic treatment may prove more clinically beneficial to reducing the morbidity and mortality of sepsis.

Serine/threonine protein phosphatase-5 (PP5) is involved in tumor progression and survival, making it an attractive therapeutic target. Specific inhibition of protein phosphatases has remained challenging because of their conserved catalytic sites. PP5 contains its regulatory domains within a single polypeptide chain, making it a more desirable target. Here we used an in silico approach to screen and develop a selective inhibitor of PP5. Compound P053 is a competitive inhibitor of PP5 that binds to its catalytic-domain and causes apoptosis in renal cancer. We further demonstrated that PP5 interacts with FADD, RIPK1 and caspase 8, components of the extrinsic apoptotic pathway complex II. Specifically, PP5 dephosphorylates and inactivates the death effector protein FADD, preserving complex II integrity and regulating extrinsic apoptosis. Our data suggests that PP5 promotes renal cancer survival by suppressing the extrinsic apoptotic pathway. Pharmacological inhibition of PP5 activates this pathway, presenting a viable therapeutic strategy for renal cancer.

Chromatin structure and organization controls DNA's accessibility to regulatory factors and influences gene regulation. Heterochromatin, or condensed chromatin containing mostly silenced genes, self-assembles through weak, multivalent interactions with its associated proteins that contain intrinsically disordered regions (IDRs) and undergoes liquid-liquid phase separation (LLPS). However, the details of the intricate molecular interactions that drive heterochromatin LLPS are not fully understood. It is crucial that we uncover the molecular mechanisms involved as it regulates vital nuclear functions, and dysregulation is implicated in neurological disorders and cancer. Here, we focus on two members of the methyl-CpG-binding domain (MBD) family of proteins, MBD2 and MBD3, that recognize and interpret methylated residues on heterochromatin's underlying DNA. We use an integrated approach to explore the driving forces that allow them to undergo LLPS and how known interactors influence this process. Using computational approaches that assess amino acid sequence features, we found that MBD2 and MBD3 are highly disordered proteins predicted to undergo LLPS. Although they are highly similar in sequence, they have distinct clustering patterns of certain residue types that suggest the molecular basis of how they phase separate differs between them. We have tested these predictions in vitro and in cellulo and have demonstrated their ability to phase separate individually, together and with methylated DNA using UV-Vis spectroscopy and microscopy. Through truncations of MBD2 and MBD3, we have found that their ability to undergo LLPS is dictated by a balance between hydrophobic interactions, likely arising from their associative domains, and electrostatic interactions, arising from their highly charged termini, occurring within or between the proteins and DNA. Finally, using scattering techniques such as small-angle X-ray scattering (SAXS) and dynamic light scattering (DLS), we have demonstrated that MBD2 and MBD3 are self-interacting proteins that form large assemblies. We propose that MBD2 and MBD3, through their ability to self-interact via hydrophobic and electrostatic forces, undergo LLPS and foster a biochemically unique environment to sequester binding partners and perform their functions as transcriptional repressors and heterochromatin organizers. Uncovering the driving forces that assemble MBD protein-based droplets will give us insight into the higher-order, LLPS-mediated organization of heterochromatin and how it functions within this structure. Additionally, understanding how disease-related aberrations influence biomolecular condensate dynamics will provide novel therapeutic targets.