22689 research outputs found
Sort by
REGENERATIVE BIONICS: INCORPORATING REINNERVATED MUSCLE INTO BIONIC SYSTEMS FOR PROSTHESIS CONTROL
Get PDFWithin the field of neuroprosthetics, there is a growing use of reinnervated muscle as a component in bionic systems for prosthesis control. Its popularity is due mainly to the pain relief benefits of muscle reinnervation, but it also offers the bio-amplification of nerve signals if the reinnervated muscle can be recorded from long-term. There are many varieties of bionic interfaces available for normal muscle tissue, but there does not yet exist an interface designed based on the unique physiology of muscle tissue that has had to regenerate and recover from major surgical intervention. Or from a different perspective, is it possible to optimize the surgical construction of a denervated muscle target (DMT) such that it is guaranteed to generate the most effective control signals?
After performing a pilot study of how differently sized DMTs might function after reinnervation, I developed and validated a versatile rat model of reinnervation using the soleus muscle and tibial nerve. This model is improved compared to the most commonly used model in the DMT field (which transects the peroneal nerve) because it is not debilitating to the animal. Using my model, I studied how completely devascularized DMTs regenerate their vasculature over time. This was motivated by the apparent gap in the scientific literature describing the 3D vascular regeneration patterns of muscle autografts. Finally, I studied how the same DMTs revascularize when there is a surface electrode present, which surprisingly showed little difference from DMTs with no electrode substrate. In summary, Part 1 of this thesis focuses on the electrophysiology of DMTs and how they could be constructed to maximize signal potential, and Part 2 focuses on the vasophysiology of DMTs healing immediately post surgery and how certain bionic hardware might influence regeneration.
Future research that builds off of my work should focus on the marriage of 3D tissue imaging data with chronic electromyography data. Further understanding the relationship between regenerating tissue structure and its long-term function will allow us to build new models that could predict how different DMTs will behave over time based on their unique anatomy
IMMUNOTHERAPY FOR T CELL MALIGNANCIES
No full textAntibody and chimeric antigen receptor (CAR) T cell-mediated targeted therapies have improved survival in patients with solid and hematologic malignancies. However, adults with T cell leukemias and lymphomas, collectively called T cell cancers, still face short survival and lack effective targeted therapies. Unlike treatment for B cell malignancies, where total B cell depletion is tolerable, total T cell elimination causes fatal immunosuppression. To retain essential immune function, we leverage the mutually exclusive and roughly equal utilization of the T cell receptor β-chain constant region 1 and 2 (TRBC1 and TRBC2) on normal T cells. Given the clonal nature of cancer, malignant T cells will express either TRBC1 or TRBC2, allowing for selective targeting that eliminates cancer cells while preserving roughly half of healthy T cells.
Preclinical studies validated that targeting TRBC1 can kill cancerous T cells while maintaining sufficient adequate immune function. However, the first-in-human clinical trial of anti-TRBC1 CAR T cells reported a low response rate and unexplained loss of CAR T cells. Here we reveal that CAR T cells are lost due to killing by the patient’s normal T cells, thereby compromising treatment efficacy. To circumvent this issue, we developed an antibody–drug conjugate (ADC) targeting TRBC1, which cured human TRBC1+ T cell cancers in mouse models. Despite this success, an equivalent TRBC2-targeting ADC remained elusive due to the absence of an optimal anti-TRBC2 antibody. We addressed this challenge by identifying a novel anti-TRBC2 antibody, JX1.1, using phage display. This antibody demonstrated high binding affinity and specificity to TRBC2. An ADC derived from this antibody showed selective killing of TRBC2+ cancer cell lines and patient-derived T-cell cancers in vitro. It also induced tumor regressions in mouse models. Together, the anti-TRBC1 and 2 ADCs provide a promising, off-the-shelf therapy for most patients harboring T-cell cancers
Strategies Toward Inhibition of Inositol Hexakisphosphate Kinase
Get PDFInositol is a unique biological small molecule as a cyclohexane ring with stereochemically defined hydroxyl groups on all six carbons of the ring. What makes this unique small molecule biologically interesting is that each hydroxyl group can be phosphorylated, and some even further pyrophosphorylated. The various phosphorylation states of inositol, along with the kinases and phosphatases that have evolved to interconvert these species, constitute the inositol phosphate pathway. These inositol phosphates serve as cell signaling molecules, binding to target proteins altering their function. Additionally, the inositol pyrophosphates are capable of transferring a phosphate group from a diphosphate to a pre-phosphorylated residue on a target protein, resulting in a pyrophosphorylated residue. Lastly, many of the inositol kinases have protein scaffolding roles by participating in protein-protein interactions. Through these mechanisms of cell signaling, the inositol phosphate pathway plays a fundamental role in cell biology. Our focus, inositol hexakisphosphate kinase (IP6K) phosphorylates inositol hexakisphosphate (IP6) to the pyrophosphate,5-diphospho-1,2,3,4,6-pentakisphosphate (IP7).
Inositol hexakisphosphate kinases (IP6Ks) have been studied for their role in glucose homeostasis, metabolic disease, fatty liver disease, blood coagulation, chronic kidney disease, neurological development, and psychiatric disease. All of the currently known IP6K inhibitors contain a critical carboxylic acid, which, due to its negative charge, prevents blood brain barrier (BBB) penetration. In this work, two approaches were taken to discover BBB penetrant IP6K inhibitors. First, a fragment-based screening approach resulted in a novel quinazolinone hit that was further optimized to a potent lead compound. Second, existing IP6K inhibitors were modified to improve brain penetration. The carboxylic acid on known inhibitors was substituted with various charge-neutral acid isosteres, resulting in improved BBB penetration. Furthermore, other pharmacokinetic properties such as plasma protein binding (PPB) and p-glycoprotein (P-gp) efflux were optimized. These medicinal chemistry efforts resulted in a lead compound with an IC50 of 15 nM, an excellent brain/plasma ratio, and a favorable pharmacokinetic profile.
Finally, work was done towards the development of an IP6K PROTAC degrader. Given that IP6Ks have an enzymatic function of IP7 synthesis, as well as protein scaffolding roles, an IP6K PROTAC would be a useful chemical tool for studying the effects of enzymatic inhibition versus protein degradation in many different cell lines without genetic manipulation. The work described herein contributes toward the development of novel chemical tools that will be invaluable in studying IP6Ks role in the brain and dissecting their roles of IP7 synthesis versus protein scaffolding
Development Impacts of a Global China: A paper series
Get PDFThis dissertation series examines the development impacts of China's expanding global presence, with a particular focus on its development finance and investment practices in Africa and the subsequent implications for global institutional regimes. Employing a multi-level approach that includes qualitative research and comparative case studies at the sector, country, and international level, the series analyses the impact of Chinese engagement on local development outcomes, such as technology transfer in manufacturing and infrastructure sectors like railways and hydropower in Nigeria, Ethiopia, and Cameroon. The research also compares China's development finance approaches with those of traditional Western donors like the World Bank, highlighting differences in implementation, standards, and technology transfer mechanisms. Furthermore, the dissertation investigates how China's rise is reshaping the international development finance landscape, exploring the interplay between official development assistance (ODA) norms and export credit regimes in response to China's increasing influence. Ultimately, the series of papers provides a multi-level analysis of China's evolving role as a global development actor and its implications for both recipient countries and the broader international system.
The first paper, "The Belt and Railway: technology transfer and structural transformation in Ethiopia’s standard gauge rail", presents a comparative analysis of two railway projects in Ethiopia, one Chinese-built and the other Turkish-built, examining their financing, construction, and debt implications to assess the impacts on technology and knowledge transfer and their role in facilitating Ethiopia's industrialisation strategies.
The second paper, "“Africa’s China”: Chinese manufacturing investment in Nigeria and channels for technology transfer", investigates the evolving landscape of Chinese foreign direct investment (FDI) in Nigeria's manufacturing sector. Drawing on field surveys and interviews, the paper identifies emerging investment clusters and analyses the motivations and challenges of Chinese firms, focusing on the potential for technology transfer through horizontal and vertical spillovers.
The third paper, "Capturing the rains: comparing Chinese and World Bank hydropower projects in Cameroon and pathways for South-South and North South technology transfer", adopts a comparative case study of hydropower projects in Cameroon financed and built by Chinese and World Bank actors. Based on fieldwork and elite interviews conducted in 2016, the paper analyses the differences in project implementation, environmental and social standards, labour relations, and the resultant potential for technology transfers between South-South and North-South cooperation models.
The fourth paper, "Two sides of a coin: Aid, export credit, and China’s impact on the OECD official finance regimes", shifts the focus to the global level, evaluating the impact of China's development finance practices on the international architecture of official finance, aid, and export credit governance at the OECD. Drawing on theories of institutional change and primary sources, the paper argues that China's rise has led to a convergent institutional evolution in the OECD's aid and export credit regimes as a consequence of external competition
OPTIMIZING AND BENCHMARKING SPIKE SORTING PIPELINES FOR HIGH-DENSITY NEURAL RECORDINGS
Get PDFAccurate spike sorting is essential for interpreting extracellular neural recordings, yet its
performance is often sensitive to probe geometry, recording noise, experimental
settings, and algorithm parameters. In this study, I evaluated and optimized a spike
sorting pipeline for 64-channel Neuronexus probe recordings in rhesus monkeys. I
compared three spike sorting algorithms—Kilosort 2.5, Kilosort 4, and Mountainsort 5—
under a unified preprocessing framework implemented via SpikeInterface. A customized
artifact removal strategy was developed using behavioral timestamps to eliminate noise
from task-related events. Results show that Mountainsort 5 outperforms both versions
of Kilosort on real and hybrid data. These findings highlight the importance of
optimization in spike sorting workflows
The Production of Forensic Space in Interwar German-language Modernist Crime Narrative
No full textThe modern metropolis emerging after the First World War is often viewed in relation to crime and criminality in political, social and aesthetic terms; the Weimar-era metropolis emblematically so. Changing forensic cultures during this period, paired with anxieties about everyday life in modernity, provide a fertile ground against which to investigate the heightened popularity of various forms of crime narrative. Using Henri Lefebvre’s theorisation of the production of space, alongside historical and theoretical developments in the forensic field of criminalistics, I argue that forensic anxieties about urban modernity during the literary scope of the dissertation yield an abstractive engagement with space that produces a new kind of space both within and beyond the confines of both the crime scene and of crime narrative, that I term ‘forensic space’, while operating a set of ‘forensic literary technologies’. I contend that the Weimar-era metropolis constitutes the emblematic forensic space of modernity and I explore the use of these forensic literary technologies in modernist experiments with crime narratives of the era, alongside contemporaneous critical engagements.
This dissertation goes on to demonstrate a literary and historical turning point in the development of literary crime genres often neglected by scholars of the period. Beginning with an examination of the impact of criminalistics, as formulated by Hans Gross, on the canonical development of early detective fiction by Edgar Allan Poe, I turn to the critical output on crime narrative by Weimar-era thinkers like Walter Benjamin and Bertolt Brecht, and conclude with analysis of German-language modernist experiments with the crime genre by authors Erich Kästner and Walter Serner. The aim is to shift scholarly discussion from inward-facing debates about generic convention toward a critical engagement with genre as a form of experimentation within the context of modernism, wherein the ‘character’ of the crime scene, and forensic space more widely, acts as theoretical catalyst and organising principle. I conclude that the connection between crime narrative and forensic space provides a novel approach for exploring the cultural preoccupations of the age and the role of literature in producing the site of their exercise
MODULATING INTERFACIAL WATER STRUCTURE FOR ENHANCED ELECTROCHEMICAL REDUCTION OF CO2 AND CO TO MULTI-CARBON PRODUCTS
No full textEnergy infrastructure, chemical, manufacturing, agriculture, and transportation sectors contribute over 85 % of global annual carbon emissions. The excess carbon has created an imbalance in Earth’s natural carbon cycle and is the primary driver of climate change. Thus, there is an urgent need to advance the development of net-zero emission technologies to recycle CO2 and rebalance the carbon cycle. Using sustainable electricity to convert CO2 electrochemically (CO2 reduction reaction or CO2RR) offers a sustainable route to close the carbon cycle and produce myriad feedstocks and fuels vital to our society.
To enhance the CO2RR process with high activity and product selectivity, the interface between the catalytic layer and electrolyte plays a key role. There are two main enhancement strategies: better catalyst design and electrolyte engineering. While many studies have focused on new catalyst development, rational strategies for modifying the solution side of the interface have not progressed to the same level of sophistication. For CO2RR, H2O (a proton source) is the reactant that directly impacts product formation. Using highly concentrated electrolytes has demonstrated the promotion of CO2RR due to the limited water supply at the interface; hence, the lowered water activity (aw). Based on the Hoffmeister series, which is linked to how ions interact with the H2O structure, we found that changing interfacial H2O structure via various cations and anions can alter the CO2RR activity. The H2O activity and interfacial structure at the electrified surface play crucial roles in controlling the CO2RR.
In this thesis, electrocatalysis, combined with a temperature-control experiment, in-situ spectroscopy, and kinetic analysis, are used to address how alteration of the H2O activity and structure at the electrified surface enhances the CO2RR. This thesis elucidates the critical roles of electrolyte-induced thermodynamic water activity change and interfacial water disordering that promote C–C coupling to multi-carbon products. These insights provide rationale for electrolyte design strategies to maximize energy-dense fuel production from CO2
Unifying Neural and Symbolic Computation for Compositional Generalization: Representation and Processing
Get PDFContemporary neural networks, despite their remarkable achievements, often fall short of the robust compositional generalization that characterizes human cognition, particularly in tasks demanding symbolic manipulation and algorithmic reasoning. This dissertation investigates the mechanisms underlying compositional generalization in neural networks and proposes novel neurosymbolic architectures that bridge the gap between connectionist and symbolic computation, primarily by leveraging Tensor Product Representations (TPRs) to embed symbolic structures within vector spaces.
First, I introduce the Role Learning Network (ROLE), a diagnostic model that automatically discovers latent structure in neural representations. This analysis reveals how networks can solve compositional tasks by converging on solutions that approximate compositional vector embeddings of symbolic structures. The causal importance of these discovered structures is demonstrated through activation patching, enabling targeted control over model behavior.
Next, I present the Differentiable Tree Machine (DTM), a unified neurosymbolic architecture that implements symbolic tree operations via a differentiable interpreter. An agent learns to produce a neurosymbolic program, while this interpreter executes the programs. To scale this approach, I develop Sparse Coordinate Trees, a TPR-equivalent encoding scheme that reduces parameters by 70x, memory by 100x, and latency by 34x. Across a range of distributional shifts from training to testing, DTM with Sparse Coordinate Trees achieves the best out-of-distribution performance compared to both neural and neurosymbolic baselines.
Finally, I focus on enhancing Transformers for modeling formal languages. I analyze the trade-off between parallelism and generalization in Recurrent Transformers for modeling Regular Languages, identifying token-layer recurrence as a key factor and examining how chunk size affects both parallelizability and length generalization. Additionally, I explore augmenting Transformers with stack-like structures for context-free languages, demonstrating that the choice of stack encoding mechanism can significantly impact performance, especially on nondeterministic languages.
Collectively, this dissertation contributes novel analysis techniques (ROLE) and unified neurosymbolic architectures (DTM, sDTM) that integrate differentiable symbolic operations and structured representations within neural networks. By exploring latent structures, explicit tree manipulation, efficient sparse representations, and recurrence, this work offers insights and methodologies for developing next-generation neural models capable of more human-like compositional generalization
The expanded role of the conserved snpc-1 and snpc-3 genes in C. elegans small RNA transcription
No full textPIWI-interacting RNAs (piRNAs) are a class of small RNAs that have a conserved function in protecting the germline genome from the deleterious effects of mobile DNA elements. In repressing these elements, piRNAs preserve the integrity of the genome and ensure its faithful transmission to the next generation. While the transposon silencing function of piRNAs is well understood, the transcriptional regulation and sexual dimorphic expression of piRNAs remain largely unknown.
The conserved snRNA activating protein complex (SNAPc) is a well-established transcription factor complex that drives small nuclear RNA (snRNA) transcription. In flies and mice, the SNAPc holocomplex consists of SNPC-1, SNPC-3, and SNPC-4 subunits, which are each encoded by a single gene. In contrast, the C. elegans snpc-1 and snpc-3 genes have been amplified through gene duplications to comprise several paralogs, each with distinct roles.
We previously showed that the SNPC-1 family protein SNPC-1.3 is a male piRNA transcription factor expressed in the male germline. Here, we provide biochemical and genetic evidence to show that the SNPC-1 paralog SNPC-1.2 constitutes a novel female piRNA transcription factor and SNPC-1.4 may have a unique spatiotemporal role in mid-late pachytene piRNA transcription while SNPC-1.1 preserved the ancestral role of SNAPc in snRNA transcription. Additionally, genetic knockout and RNAi-based knockdown assays reveal that the snpc-3 family genes snpc-3.1 and snpc-3.2 comprise functionally redundant core piRNA transcription factors required for the transcription of both male and female piRNAs, while snpc-3.4 is uniquely involved in snRNA transcription.
Collectively, the snpc-1 and snpc-3 gene families encode specificity factors for possibly four distinct protein complexes that discriminate between the transcription of snRNAs and both sex-specific and temporally regulated piRNAs in the C. elegans genome. Our work provides insights into the sexually dimorphic and spatiotemporal piRNA-mediated regulation of germline genes to maintain proper germline development and suggests that piRNA biogenesis emerged from the duplication and diversification of ancient snRNA transcriptional machinery
INTERLEUKIN-2-DRIVEN MODULATION OF THE REGULATORY T CELL TRANSCRIPTOME AND EPIGENOME
Get PDFThe immune system plays a critical role in both health and disease. Regulatory T cells, including their development and maintenance, are a keep component of this role to ensure balance in homeostasis and an effective response in disease. In homeostasis, Tregs’ role in tolerance to self-antigens and by extension components of the intestinal microbiota help prevent autoimmunity. In diseases, including infection and cancer, Tregs regulate the immune response by cytokine-mediated suppression through the secretion of anti-inflammatory cytokines such as IL-10 and TGF-β. Tregs can also deprive immune effector cells such as T effector cells, NK cells and B cells of needed growth signal via the expression of high affinity CD25 and subsequent consumption of Interleukin-2 (IL-2), a key survival cytokine. Upon activation, several chromatin and gene expression changes occur in Tregs to ensure effective immune response. Here we investigated how temporal stimulation with IL-2 affect Treg chromatin landscape and gene expression to induce different cell states for response via single cell RNA and ATAC Sequencing. We uncovered heterogeneity in response in resting Tregs (rTregs) with less signal transduction through Stat5 and the expression of several genes implicated by key transcription factors including Batf. By understanding the transcriptional and epigenetic changes in Tregs, we can better harness their role as key therapeutical targets to maintain immune balance