Genomic instability, in the form of gene mutations, insertions/deletions, and gene amplifications, is one of the hallmarks in many types of cancers and other inheritable genetic disorders. Trinucleotide repeat (TNR) disorders, such as Huntington’s disease (HD) and Myotonic dystrophy (DM) can be inherited and repeats may be extended through subsequent generations. However, it is not clear how the CAG repeats expand through generations in HD. Two possible repeat expansion mechanisms include: 1) polymerase mediated repeat extension; 2) persistent TNR hairpin structure formation persisting in the genome resulting in expansion after subsequent cell division. Recent in vitro studies suggested that a family A translesion polymerase, polymerase θ (Polθ), was able to synthesize DNA larger than the template DNA. Clinical and in vivo studies showed either overexpression or knock down of Polθ caused poor survival in breast cancer patients and genomic instability. However, the role of Polθ in TNR expansion remains unelucidated. Therefore, we hypothesize that Polθ can directly cause TNR expansion during DNA synthesis. The investigation of the functional properties of Polθ during DNA replication and TNR synthesis will provide insight for the mechanism of TNR expansion through generations.

The endoplasmic reticulum (ER) is a multi-functional cellular compartment that functions in protein folding, lipid biosynthesis, and calcium homeostasis. Perturbations to ER function lead to the dysregulation of ER homeostasis, causing the accumulation of unfolded and misfolded proteins in the cell. This is a state of ER stress. ER stress elicits a cytoprotective, adaptive signaling cascade to mitigate stress, the Unfolded Protein Response (UPR). As long as the UPR can moderate stress, cells can produce the proper amount of proteins and maintain a state of homeostasis. If the UPR, however, is dysfunctional and fails to achieve this, cells will undergo apoptosis. Diabetes mellitus is a group of metabolic disorders characterized by persistent high blood glucose levels. The pathogenesis of this disease involves pancreatic β-cell dysfunction: an abnormality in the primary function of the β-cell, insulin production and secretion. Activation of the UPR is critical to pancreatic β-cell survival, where a disruption in ER stress signaling can lead to cell death and consequently diabetes. There are several models of ER stress leading to diabetes. Wolcott-Rallison syndrome, for example, occurs when there is a mutation in the gene encoding one of the master regulators of the UPR, PKR-like ER kinase (PERK). In this dissertation, we show that Wolfram Syndrome 1 (WFS1), an ER transmembrane protein, is a component of the UPR and is a downstream target of two of the master regulators of the UPR, Inositol Requiring 1 (IRE1) and PERK. WFS1 mutations lead to Wolfram syndrome, a non-autoimmune form of type 1 diabetes accompanied by optical atrophy and other neurological disorders. It has been shown that patients develop diabetes due to the selective loss of their pancreatic β-cells. Here we define the underlying molecular mechanism of β-cell loss in Wolfram syndrome, and link this cell loss to ER stress and a dysfunction in a component of the UPR, WFS1. We show that WFS1 expression is localized to the β-cell of the pancreas, it is upregulated during insulin secretion and ER stress, and its inactivation leads to chronic ER stress and apoptosis. This dissertation also reveals the previously unknown function of WFS1 in the UPR. Positive regulation of the UPR has been extensively studied, however, the precise mechanisms of negative regulation of this signaling pathway have not. Here we report that WFS1 regulates a key transcription factor of the UPR, activating transcription factor 6 (ATF6), through the ubiquitin-proteasome pathway. WFS1 expression decreases expression levels of ATF6 target genes and represses ATF6-mediated activation of the ER stress response (ERSE) promoter. WFS1 recruits and stabilizes an E3 ubiquitin ligase, HMG-CoA reductase degradation protein 1 (HRD1), on the ER membrane. The WFS1-HRD1 complex recruits ATF6 to the proteasome and enhances its ubiquitination and proteasome-mediated degradation, leading to suppression of the UPR under non-stress conditions. In response to ER stress, ATF6 is released from WFS1 and activates the UPR to mitigate ER stress. This body of work reveals a novel role for WFS1 in the UPR, and a novel mechanism for regulating ER stress signaling. These findings also indicate that hyperactivation of the UPR can lead to cellular dysfunction and death. This supports the notion that tight regulation of ER stress signaling is crucial to cell survival. This unanticipated role of WFS1 for a feedback loop of the UPR is relevant to diseases caused by chronic hyperactivation of ER stress signaling network such as pancreatic β-cell death in diabetes and neurodegeneration.

Mutations in the SOD1 gene are the best characterized genetic cause of amyotrophic lateral sclerosis (ALS) and account for ~20% of inherited cases and 1-3% of sporadic cases. The gene-editing tool Cas9 can silence mutant genes that cause disease, but effective delivery of CRISPR-Cas9 to the central nervous system (CNS) remains challenging. Here, I developed strategies using canonical Streptococcus pyogenes Cas9 to silence SOD1. In the first strategy, I demonstrate effectiveness of systemic delivery of guide RNA targeting SOD1 to the CNS in a transgenic mouse model expressing human mutant SOD1 and Cas9. Silencing was observed in both the brain and the spinal cord. In the second strategy, I demonstrate the effectiveness of delivering both guide RNA and Cas9 via two AAVs into the ventricles of the brain of SOD1G93A mice. Silencing was observed in the brain and in motor neurons within the spinal cord. For both strategies, treated mice had prolonged survival when compared to controls. Treated mice also had improvements in grip strength and rotarod function. For ICV treated mice, we detected a benefit of SOD1 silencing using net axonal transport assays, a novel method to detect motor neuron function in mice before onset of motor symptoms. These studies demonstrate that Cas9-mediated genome editing can mediate disease gene silencing in motor neurons and warrants further development for use as a therapeutic intervention for SOD1-linked ALS patients.

During stress, eukaryotes regulate protein synthesis in part through formation of cytoplasmic, non-membrane-bound complexes called stress granules (SGs). SGs transiently store signaling proteins and stalled translational complexes in response to stress stimuli (e.g. oxidative insult, DNA damage, temperature shifts and ER dysfunction). The functional outcome of SGs is proper translational regulation and signaling, allowing cells to overcome stress. The fatal motor neuron disease Amyotrophic Lateral Sclerosis (ALS) develops in an age-related manner and is marked by progressive neuronal death, with cytoplasmic protein aggregation, excitotoxicity and increased oxidative stress as major hallmarks. Fused in Sarcoma/Translocated in Liposarcoma (FUS) is an RNA-binding protein mutated in ALS with roles in RNA and DNA processing. Most ALS-associated FUS mutations cause FUS to aberrantly localize in the cytoplasm due to a disruption in the nuclear localization sequence. Intriguingly, pathological inclusions in human FUSALS cases contain aggregated FUS as well as several SG-associated proteins. Further, cytoplasmic mutant FUS incorporates into SGs, which increases SG volume and number, delays SG assembly, accelerates SG disassembly, and alters SG dynamics. I posit that mutant FUS association with stress granules is a toxic gain-of-function in ALS that alters the function of SGs by interaction with SG components. Here, I show that mutant FUS incorporates in to SGs via its Cterminal RGG motifs, the methylation of which is not required for this localization. Further, I identify protein interactions specific to full-length mutant FUS under stress conditions that are potentially capable of interacting with FUS in SGs. Finally, I demonstrate a potential change in the protein composition of SGs upon incorporation of mutant FUS. These findings advance the field of ALS and SG biology, thereby providing groundwork for future investigation.

Extracellular vesicles (EVs), exosomes and microvesicles, transfer endogenous RNAs between neurons over short and long distances. We have explored EVs for siRNA delivery to brain. (1) We optimized siRNA chemical modifications and siRNA conjugation to lipids for EV-mediated delivery. (2) We developed a GMP-compatible, scalable method to manufacture active EVs in bulk. (3) We characterized lipid and protein content of EVs in detail. (4) We established how protein and lipid composition relates to siRNA delivering activity of EVs, and we reverse engineered natural exosomes (small EVs) into artificial exosomes based on these data. We established that cholesterol-conjugated siRNAs passively associate to EV membrane and can be productively delivered to target neurons. We extensively characterized this loading process and optimized exosome-to-siRNA ratios for loading. We found that chemical stabilization of 5'-phosphate with 5'-E-vinylphosphonate and chemical stabilization of all nucleotides with 2'-O-methyl and 2'-fluoro increases the accumulation of siRNA and the level of mRNA silencing in target cells. Therefore, we recommend using fully modified siRNAs for lipid-mediated loading to EVs. Later, we identified that α-tocopherol-succinate (vitamin E) conjugation to siRNA increases productive loading to exosomes compared to originally described cholesterol. Low EV yield has been a rate-limiting factor in preclinical development of the EV technology. We developed a scalable EV manufacturing process based on three-dimensional, xenofree culture of mesenchymal stem cells and concentration of EVs from conditioned media using tangential flow filtration. This process yields exosomes more efficient at siRNA delivery than exosomes isolated via differential ultracentrifugation from two-dimensional cultures of the same cells. In-depth characterization of EV content is required for quality control of EV preparations as well as understanding composition–activity relationship of EVs. We have generated mass-spectrometry data on more than 3000 proteins and more than 2000 lipid species detected in exosomes (small EVs) and microvesicles (large EVs) isolated from five different producer cells: two cell lines (U87 and Huh7) and three mesenchymal stem cell types (derived from bone marrow, adipose tissue and umbilical cord Wharton’s jelly). These data represent an indispensable resource for the community. Furthermore, relating composition change to activity change of EVs isolated from cells upon serum deprivation allowed us to identify essential components of siRNA-delivering exosomes. Based on these data we reverse engineered natural exosomes into artificial exosomes consisting of dioleoyl-phosphatidylcholine, cholesterol, dilysocardiolipin, Rab7, AHSG and Desmoplakin. These artificial exosomes reproduced efficient siRNA delivery of natural exosomes both in vitro and in vivo. Artificial exosomes may facilitate manufacturing, quality control and cargo loading challenge that currently impede the therapeutic EV field.

Indiana University-Purdue University Indianapolis (IUPUI) / Skeletal manifestations including short stature, osteoporosis, kyphoscoliosis, and tibial dysplasia cumulatively affect approximately 70% of patients with neurofibromatosis type 1 (NF1). Tibial pseudarthrosis, the chronic non-union of a spontaneous fracture, is a debilitating skeletal malady affecting young children with NF1. These non-healing fractures respond poorly to treatment and often require amputation of the affected limb due to limited understanding of the causative mechanisms. To better understand the cellular and molecular pathogenesis of these osseous defects, we have established a new mouse model which recapitulates a spectrum of skeletal pathologies frequently observed in patients with NF1. Nf1flox/-;Col2.3Cre mice, harboring Nf1 nullizygous osteoblasts on a Nf1+/- background, exhibit multiple osseous defects which are closely reminiscent of those found in NF1 patients, including runting (short stature), bone mass deficits, spinal deformities, and tibial fracture non-union. Through adoptive bone marrow transfer studies, we have demonstrated that the Nf1 haploinsufficient hematopoietic system pivotally mediates the pathogenesis of bone loss and fracture non-union in Nf1flox/-;Col2.3Cre mice. By genetic ablation of a single Nf1 allele in early myeloid development, under the control of LysMCre, we have further delineated that Nf1 haploinsufficient myeloid progenitors and osteoclasts are the culprit lineages mediating accelerated bone loss. Interestingly, conditional Nf1 haploinsufficiency in mature osteoclasts, induced by CtskCre, was insufficient to trigger enhanced lytic activity. These data provide direct genetic evidence for Nf1’s temporal significance as a gatekeeper of the osteoclast progenitor pool in primitive myelopoiesis. On the molecular level, we found that transforming growth factor-beta1 (TGF-β1), a primary mediator in the spatiotemporal coupling of bone remodeling, is pathologically overexpressed by five- to six- fold in both NF1 patients and in mice. Nf1 deficient osteoblasts, the principal source of TGF-β1 in the bone matrix, overexpress TGF-β1 in a gene dosage dependent fashion. Moreover, p21Ras dependent hyperactivation of the Smad pathway accentuates responses to pathological TGF-β1 signals in Nf1 deficient bone cells. As a proof of concept, we demonstrate that pharmacologic TβRI kinase inhibition can rescue bone mass defects and prevent tibial fracture non-union in Nf1flox/-;Col2.3Cre mice, suggesting that targeting TGF-β1 signaling in myeloid lineages may provide therapeutic benefit for treating NF1 skeletal defects.

<p>Alzheimer’s disease (AD) is a degenerative disease with progressive deterioration of neural networks in the brain. Fractal dimension analysis (FD) of resting state blood oxygen level dependent (BOLD) signals acquired using functional magnetic resonance imaging (fMRI) allows us to quantify complex signalling in the brain and may offer a window into the network erosion. This novel approach can provide a sensitive tool to examine early stages of AD. As AD progresses, we expect to see a reduction in brain connectivity and signal complexity concurrent with biochemical changes (e.g. altered levels of N-acetyl aspartate (NAA), myoinositol (mI) and glutamate as measured using magnetic resonance spectroscopy, MRS), volumetric changes and abnormally high levels of brain iron.</p> <p>Over a series of 4 studies we examined the relationship of BOLD signal complexity and functional connectivity with documented MRI markers of pathology in AD (n=38) as compared to normal controls (NC) (n=16). AD subjects were in early stage of illness (mild to moderate impairment on the mini mental state exam, MMSE). We validated the temporal (short term (within minutes) and longer term (over a number of months)) consistency of FD measurement and choice of BOLD acquisition method (spiral vs. EPI), provided MRI sequence repeat time (TR) was kept constant. FD reduction (decrease in signal complexity) correlated with worsening pathological values on MRS (­NAA decrease and mI increase) and with a decrease in functional connectivity. This demonstrates that FD (signal complexity) reduces in proportion to AD severity. FD reduction is connected to functional connectivity measured through resting state network (RSN) analysis suggesting the reduction in FD relates to neuronal loss rather than altered vascularity. The narrow range of cognitive impairment (such as scores on the MMSE or the clinical dementia rating scale, CDR) likely precluded correlation between these measures and FD or RSN. Functional connectivity (RSN) was also reduced when brain iron levels were increased within certain network nodes (posterior cingulate cortex and lateral parietal cortex). Therefore iron deposition may play a role in network disruption of AD brains.</p> <p>The overall conclusion of this thesis is that signal complexity of BOLD fMRI signals, as measured with FD, may detect early pathology in the progression of AD. FD can detect neuronal changes in deep brain structures before volume loss in these structures and before significant changes in MRS markers were detectable between the AD and NC groups. An FD change mirrors disruptions in functional connectivity but detection is not limited to RSN nodes in the brain. This novel approach could further our understanding of AD and may be applied to other pathologies of the brain.</p> / Doctor of Philosophy (PhD)

Traumatic brain injury (TBI) is a leading cause of disability worldwide. Annually, 150 to 200/1,000,000 people become disabled as a result of brain trauma. Axonal degeneration is a critical, early event following TBI of all severities but whether axon degeneration is a driver of TBI remains unclear. Molecular pathways underlying the pathology of TBI have not been defined and there is no efficacious treatment for TBI. Despite this significant societal impact, surprisingly little is known about the molecular mechanisms that actively drive axon degeneration in any context and particularly following TBI. Although severe brain injury may cause immediate disruption of axons (primary axotomy), it is now recognized that the most frequent form of traumatic axonal injury (TAI) is mediated by a cascade of events that ultimately result in secondary axonal disconnection (secondary axotomy) within hours to days. Proposed mechanisms include immediate post-traumatic cytoskeletal destabilization as a direct result of mechanical breakage of microtubules, as well as catastrophic local calcium dysregulation resulting in microtubule depolymerization, impaired axonal transport, unmitigated accumulation of cargoes, local axonal swelling, and finally disconnection. The portion of the axon that is distal to the axotomy site remains initially morphologically intact. However, it undergoes sudden rapid fragmentation along its full distal length ~72 h after the original axotomy, a process termed Wallerian degeneration. Remarkably, mice mutant for the Wallerian degeneration slow (Wlds) protein exhibit ~tenfold (for 2–3 weeks) suppressed Wallerian degeneration. Yet, pharmacological replication of the Wlds mechanism has proven difficult. Further, no one has studied whether Wlds protects from TAI. Lastly, owing to Wlds presumed gain-of-function and its absence in wild-type animals, direct evidence in support of a putative endogenous axon death signaling pathway is lacking, which is critical to identify original treatment targets and the development of viable therapeutic approaches. Novel insight into the pathophysiology of Wallerian degeneration was gained by the discovery that mutant Drosophila flies lacking dSarm (sterile a/Armadillo/Toll-Interleukin receptor homology domain protein) cell-autonomously recapitulated the Wlds phenotype. The pro-degenerative function of the dSarm gene (and its mouse homolog Sarm1) is widespread in mammals as shown by in vitro protection of superior cervical ganglion, dorsal root ganglion, and cortical neuron axons, as well as remarkable in-vivo long-term survival (>2 weeks) of transected sciatic mouse Sarm1 null axons. Although the molecular mechanism of function remains to be clarified, its discovery provides direct evidence that Sarm1 is the first endogenous gene required for Wallerian degeneration, driving a highly conserved genetic axon death program. The central goals of this thesis were to determine (1) whether post-traumatic axonal integrity is preserved in mice lacking Sarm1, and (2) whether loss of Sarm1 is associated with improved functional outcome after TBI. I show that mice lacking the mouse Toll receptor adaptor Sarm1 gene demonstrate multiple improved TBI-associated phenotypes after injury in a closed-head mild TBI model. Sarm1-/- mice developed fewer beta amyloid precursor protein (βAPP) aggregates in axons of the corpus callosum after TBI as compared to Sarm1+/+ mice. Furthermore, mice lacking Sarm1 had reduced plasma concentrations of the phosphorylated axonal neurofilament subunit H, indicating that axonal integrity is maintained after TBI. Strikingly, whereas wild type mice exhibited a number of behavioral deficits after TBI, I observed a strong, early preservation of neurological function in Sarm1-/- animals. Finally, using in vivo proton magnetic resonance spectroscopy, I found tissue signatures consistent with substantially preserved neuronal energy metabolism in Sarm1-/- mice compared to controls immediately following TBI. My results indicate that the Sarm1-mediated prodegenerative pathway promotes pathogenesis in TBI and suggest that anti-Sarm1 therapeutics are a viable approach for preserving neurological function after TBI.