22689 research outputs found
Sort by
EXPLORATORY ANALYSIS OF COLLECTED BAT AND TREE DATA FROM DATE PALM SAP HARVESTING IN BANGLADESH FROM DECEMBER 2023 TO MARCH 2024
Get PDFIn Bangladesh, the primary route of human exposure to zoonotic Nipah virus is through the consumption of date palm sap that has been contaminated with saliva, urine, or feces from Pteropus bats. Since the first outbreak in 2001, there have been almost annual human infections, with more outbreaks occurring in colder winters. Identifying a relationship between sap sweetness and quantity may provide insight into the factors driving the rise in Nipah cases during colder months. Additionally, understanding the timing of Pteropus visitations is critical, as increased visitation rates create more opportunities for date palm sap contamination and, consequently, a higher risk of spillover events. In this study, we aimed to better understand what factors, including weather and the ecology of bats and date palm trees, affect the frequency of sap contamination by bats. Between December 2023 to May 2024, infrared cameras were set up on 20 date palm trees for eight nights of observation in Bangladesh. For both Pteropus and non-Pteropus bats, there were more visitations in colder months, December (np = 155, nnp = 954) and January (np = 521, nnp = 1289). Overall, non-Pteropus bats visited more frequently than Pteropus bats. Measurements of date palm sap yield and sweetness showed that more, but less sweet sap was produced at lower temperatures. A slight negative trend (p = 0.664) was observed between sap quantity and sweetness, while a slight positive trend was found between sap quantity and total bat visitations for both Pteropus (p < 0.001) and non-Pteropus (p < 0.001) species. Among non-Pteropus bats, there was a slight positive trend between visitations and sap sweetness, but no such trend was observed for Pteropus bats. Based on our study findings, colder weather produced more (p-value = 0.003) but less sweet (p = 0.35) sap. Additionally, the increase in bat visitations to date palm trees seem to be more driven by colder temperature rather than sap quantity or sweetness
Characterization of a Pathogenic DCTN4 Variant
Get PDFCytoplasmic cargoes are transported within cells via mechanochemical motor proteins along cytoskeletal filaments such as microtubules. The major minus-end directed, microtubule-based motor, cytoplasmic dynein 1, relies heavily on its binding partner, dynactin, for motor activation and cargo specification. Interactions between dynein and dynactin are scaffolded by cargo-associated adaptor proteins that link the motor complex to diverse cargoes. Dynactin is a large multiprotein complex. Its DCTN4 component has been shown to bind multiple cargo adaptor proteins directly. A genetic variant of DCTN4 (rs35662018, hereafter 018-DCTN4) that yields a tyrosine to cysteine mutation, has been linked to worse disease risk in cystic fibrosis and lung adenocarcinoma. Our group showed previously that expression of the 018-DCTN4 variant impairs cell migration and induces loss of focal adhesions, altered integrin trafficking, and aberrant actin dynamics in a variety of cell lines. The goal of my thesis work was to determine molecular mechanisms underlying the observed phenotypes. My work also provided new insight into the impact of 018-DCTN4 on the function of the lung epithelium.
I used a proteomics approach to identify differential interactors that might provide mechanistic explanations for the phenotypes exhibited by cells expressing 018-DCTN4 (Chapter 2). In Chapter 3 I generate a CRISPR gene-edited human bronchial epithelial cell line that expresses 018-DCTN4 from the endogenous locus. Differential transcriptome analysis on these cells indicated changes that explain many of the observed phenotypes. In addition, the transcriptome revealed significant alterations in gene expression patterns relevant to innate immunity and barrier function in the lung. This analysis further revealed decreased expression of cystic fibrosis transmembrane conductance regulator (CFTR) and reduction in CFTR protein level. My studies expand our knowledge of the functional impacts of 018-DCTN4 in cells and suggests new mechanistic targets for further exploration
Physics of Membrane Curvature Sensing by Proteins: Modeling and Simulations
No full textMembrane curvature sensing is crucial for processes such as cell division and morphogenesis. Curvature sensing is also essential for subcellular functions such as membrane repair and the maintenance of organelles. Although curvature sensing occurs in cells via the collective activity of several proteins, in vitro experiments in the literature show that even a single nanometer-sized septin protein can distinguish between membrane curvatures much larger than itself, on the micron-scale.
In this dissertation, I propose new mathematical models and simulation methods to understand how curvature sensing depends on the physical properties of the membrane and protein, in terms of sensing limits, thermal membrane fluctuations, and protein localization. I investigate the fundamental bounds that determine how well a single protein senses membrane curvature, either by sensing shape directly or by using lipid packing as a proxy. Modeling the membrane as a continuum system, I identify how its thermal fluctuations inhibit an idealized protein from discerning the time-averaged shape of the membrane, given that a protein can only access local curvature instantaneously. I develop algorithms based on Fourier Space Brownian Dynamics to perform simulations of fluctuating membrane heights and lipid densities, and propose a signal-to-noise ratio that quantifies the curvature sensing efficacy of a protein. I analyze how this SNR depends on parameters such as the protein’s size, and the membrane’s bending stiffness, thickness, and adhesion strength. I show how the available experimental data can be explained by two classes of models: a preferred curvature model, and a threshold curvature model. These models offer insight into how well a given protein can sense curvature compared to an ideal detector of the same size. Building on this, I explicitly incorporate the energetic coupling between a protein and membrane to develop a model for a dynamic protein diffusing in the membrane that is sensitive to lipid packing via its insertion into one leaflet of the bilayer. These results show how a protein’s preference for a particular lipid packing density influences its localization to different membrane regions, and how monolayer viscosity and interleaflet friction influence its effective diffusion constant in the membrane
EXPLORE THE FUNCTION OF THE IMMUNE SIGNALING PROTEIN VAGO1 IN AEDES AEGYPTI
Get PDFAedes aegypti is the principal vector for multiple medically significant arboviruses, including dengue, Zika, chikungunya, and yellow fever viruses. With limited efficacious vaccines and antiviral therapeutics available, vector control remains crucial. However, conventional insecticide-based strategies face mounting challenges from resistance development, operational constraints, and environmental impacts. As an alternative, genetic control strategies using pathogens-resistant transgenic mosquitoes offer promising avenues for reducing arboviral transmission.
The invertebrate-specific cytokine-like protein Vago has been implicated in antiviral function in Drosophila and Culex mosquito species. In Ae. aegypti, Vago1 (AeVago1) has been shown to possess antiviral activity in cell culture-based studies, however, its function in the mosquito remains uncharacterized. To investigate immune function of AeVago1, we generated three transgenic mosquito lines with tissue-specific, blood meal-inducible AeVago1 overexpression in midguts (CpVago1), fat bodies (VgVago1), and both tissues (CpVgVago1). Evaluation of fitness-related traits revealed that tissue-specific AeVago1 overexpression had no significant impact on fecundity, egg hatching rates, or adult longevity, with only a modest acceleration in pupation timing observed across all transgenic lines.
Following the viral challenge, AeVago1 overexpression did not significantly affect Mayaro virus (MAYV) or dengue virus serotype 2 (DENV2) infections in midguts of all transgenic lines. However, viral titers were significantly reduced in carcasses (the mosquito whole body minus midgut) in all transgenic lines for both viruses. Infection prevalence in carcasses also decreased, with the VgVago1 line showing a significant reduction for MAYV infection and CpVgVago1 for DENV2 infection. These findings suggest AeVago1 may contribute to limiting systemic viral dissemination beyond the midgut.
Bacteria challenge assays showed that survival rates following bacterial injection remained comparable between CpVago1, VgVago1, and the wild-type control mosquitoes. Moreover, neither recombinant AeVago1 nor AeVago2 exhibited antimicrobial activity in bacterial radial diffusion assays. Midgut microbiota composition remained largely unaltered, with only a modest reduction observed in sugar-fed CpVago1 mosquitoes. These results suggest that AeVago1 overexpression did not substantially alter resistance to bacterial infection.
Interestingly, both CpVago1 and VgVago1 lines exhibited enhanced survival following infection with the entomopathogenic fungus Beauveria bassiana, indicating a broader immunomodulatory role for AeVago1 beyond its antiviral activity.
In summary, our results indicate that AeVago1 contributes to systemic antiviral and antifungal immunity in Ae. aegypti without incurring major fitness costs. The lack of antibacterial effects suggests a pathogen-specific role for AeVago1. This study expands the current understanding of mosquito immune signaling and provides implications for the development of novel vector-based disease control strategies
PREGNANCY PERCEIVED STRESS EXPOSURE AND ATTENTION DEFICIT HYPERACTIVITY DISORDER: EVIDENCE AND BIOLOGIC INSIGHTS FROM ECHO
Get PDFAttention Deficit Hyperactivity Disorder (ADHD) affects approximately 1 in 10 children aged 3 to 17 years in the United States. There are several environmental risk factors that have been shown to contribute to ADHD risk. Maternal stress during pregnancy, including stressful life events and psychological distress, has been linked to offspring ADHD, though many of these studies have been conducted in small, unrepresentative study populations. Additionally, the biological mechanism underlying this association is incompletely understood. Our aims were therefore to 1) examine the association between maternal perceived stress exposure during pregnancy and offspring ADHD, 2) identify newborn DNA methylation (DNAm) differences in genes involved in hypothalamic-pituitary-adrenal axis (HPAA) regulation including NR3C1, FKBP5 and HSD11B2 that associated with maternal perceived stress during pregnancy, and 3) Identify DNAm changes at birth, specifically in HPAA associated genes, associated with prospective ADHD diagnosis in children. We conducted analyses using data from the Environmental Influences on Child Health Outcome (ECHO) program for each aim. We used perceived stress measured using Cohen’s perceived stress scale as a measure of maternal stress. Our primary outcome was having an ADHD diagnosis from a clinical provider and secondary outcomes included child behavior problems (attention problems and externalizing problems) assessed using the preschool and school aged versions of the Child Behavior Checklist. DNAm was measured using the Illumina 450K and EPIC Beadchips in cord blood and dried blood spots. We used generalized regression models for statistical analyses for each aim and adjusted for sets of relevant covariates. We found that higher levels of maternal perceived stress during pregnancy was associated with higher odds of ADHD diagnosis and greater child behavior problems in offspring (Chapter 2). We also identified associations between maternal stress during pregnancy and newborn blood DNAm in NR3C1 and FKBP5 and found evidence sex differences (Chapter 3). Finally, we found evidence of association between newborn blood DNAm in NR3C1, FKBP5 and HSD11B2 and prospective ADHD diagnosis and child behavior problems with some sex specific effects (Chapter 4). Taken together, these results provide insight into ADHD etiology and identify biological targets that can aid in ADHD management
SHORT-TERM LAMINA CRIBROSA BIOMECHANICAL RESPONSES TO MEDICATION INDUCED INTRAOCULAR PRESSURE CHANGES IN GLAUCOMA PATIENTS: VEGAN STUDY
Get PDFThe objective of this pilot research was to investigate the accessibility of utilizing the DVC technique for capturing the characteristics of the short-term, in vivo biomechanical responses of the lamina cribrosa (LC) in patients’ hypotensive glaucoma eye drops. The primary goal of this research is to identify mechanical biomarkers for primary open-angle glaucoma (POAG). Optical coherence tomography (OCT) volumes and intraocular pressure (IOP) measurements were obtained from 17 eyes of 10 patients both before and approximately one week after initiating treatment. Digital volume correlation (DVC) was applied to pre- and post-treatment OCT scans to compute LC strain and anterior lamina depth (ALD) changes. Following an average IOP reduction of 5.24 ± 2.73 mmHg (24.1% ± 8.3%), the LC exhibited statistically significant tensile axial strain (Ezz = 0.27% ± 0.43%, p = 0.020), along with marked changes in maximum principal strain (Emax = 0.83% ± 0.34%) and maximum shear strain (Γmax = 0.74% ± 0.29%). These strain magnitudes exceeded baseline measurement error and demonstrated strong positive correlations with the extent of IOP decrease. However, compressive radial strain (Err) and other strain components did not show significant relationships. ALD exhibited a small posterior shift (3.21 ± 3.60 µm), though it was not significantly associated with either IOP reduction or strain metrics. Regional differences in strain and ALD were not observed across LC quadrants. Greater strain and compliance were significantly related to thicker retinal nerve fiber layer (RNFL), but not to mean deviation (MD), visual field index (VFI), or patient age. These findings indicate that LC strain especially for axial, principal, and shear strain which is a sensitive and quantifiable biomechanical indicator of short-term IOP reduction. This feature suggests that DVC-based strain assessment may provide valuable insight into early glaucomatous changes and treatment effects
BIOENGINEERED HUMAN MICROPHYSIOLOGICAL SYSTEMS FOR HIGH-THROUGHPUT NEUROMUSCULAR AND CARDIAC FUNCTIONAL ANALYSIS
No full textAnimal models have been widely used for disease research and drug development, yet they often fail to replicate human-specific physiology, leading to poor clinical translation. Human microphysiological systems (MPS) offer a promising alternative by better mimicking human tissue structure and function, but challenges in scalability, reproducibility, and functional maturity remain. This thesis addresses these limitations by developing bioengineered MPS platforms that integrate multiscale biofabrication techniques, high-throughput real-time biosensors, and AI-driven analytical frameworks for disease modeling and drug screening. For neuromuscular physiology, a compartmentalized microfluidic platform enables the co-culture of human motor neurons with skeletal muscle bundles, forming 3D functional human model of neuromuscular junctions (NMJs). This platform maintains viable, functionally mature NMJs for long term assessment, allowing longitudinal, non-invasive functional assessments with industry standard high-through format for applications ranging from disease modeling and drug screening. Extensive validation confirms its biological fidelity and reproducibility, demonstrating its potential and introducing a novel breakthrough for neuromuscular disease research.
In cardiac electrophysiology, this thesis introduces a mechanically self-adaptive stretchable microelectrode array (ssMEA) integrated with a deep learning-based Spatiotemporal Convolutional Recurrent framework, termed Cardiac Organoid Electrophysiology-Net, or ‘CardOrg-Net’, designed to predict and overcome functional readout challenges posed by cardiac organoid heterogeneity. The ssMEA conforms to spontaneously beating iPSC-derived cardiac organoids, enabling high-resolution electrophysiological conduction mapping. CardOrg-Net predicts and generates conduction maps with over 90% accuracy, resolving activation directions and velocities. Validated through drug-response studies and disease modeling, including arrhythmogenic right ventricular cardiomyopathy organoids, this system detects subtle physiological changes often missed by conventional methods. Together, these scalable, validated MPS platforms advance neuromuscular and cardiac research, improving precision, reproducibility, and predictive accuracy towards standardization of technology. By addressing key limitations of current models, this work enhances translational research, reduces reliance on animal models, and opens new pathways for therapeutic discovery and precision medicine
FORCE-CONTROLLED MATERIAL GROWTH THROUGH PIEZOELECTRIC EFFECTS
No full textThis dissertation is an investigation into material growth mechanisms that arise in response to applied mechanical loading and enable self-adaptive responses. Natural materials such as bone can adapt their structural properties to adapt and meet the demands of their loading environment, but recreating similar behaviors in synthetic materials have limitations in available material systems, response speed, mechanical properties, and environments. It has been shown that piezoelectric scaffolds can be used to convert mechanical stress into charges that electrostatically attract mineral ions from a surrounding solution to form reinforcing mineral layers. However, there is a knowledge gap relating the applied force magnitude, the resulting generated piezoelectric charges, and the growth profile as well as the applicability to a wider range of materials such as metals. The goal of this dissertation is to elucidate quantitative relations between the applied force, piezoelectric charges and potentials, and reinforcing material thickness over time as a function of loading and environmental parameters for both mineral and metal formation.
The first aim of this dissertation is to establish relationships for the reinforcement system under mineral ion electrolyte conditions between the applied force, effective piezoelectric charge, mineral growth rate, and mineral growth thickness. The relations are first derived and then experimentally verified to be self-limiting with exponential relationships over time by measuring the loading force, effective charge, and mineral thickness under different loading conditions until a saturation thickness is reached.
The second aim of this dissertation is to expand the system to metal formation and to establish relationships for the reinforcement system under metal electrolyte conditions between the applied force, piezoelectric potential, metal growth rate, and metal growth thickness. The relations are first derived and then experimentally verified to be self-limiting with exponential relationships over time by measuring the loading force, piezoelectric potential, and metal thickness under different loading conditions until a saturation thickness is reached.
The third aim of this dissertation is to present a model for adding control over the formed mineral crystallinity through the generation of reactive oxygen species and local alterations in solution pH dependent on loading conditions
Diagnose, Correct, Steer: Towards Trustworthy and Functional AI systems
Get PDFArtificial Intelligence (AI) and Machine Learning (ML) are increasingly integral across various sectors, offering novel applications while raising critical concerns about safety and privacy. This dissertation introduces a three-stage framework, diagnose, correct, and steer, to develop trustworthy and functional AI/ML systems that address these concerns.
First, in the diagnosis stage, AI/ML systems are analyzed to identify conditions under which adversaries with practical knowledge can exploit vulnerabilities, leading to unsafe behaviors such as the generation of not-safe-for-work (NSFW) content and privacy leakage. To uncover these weaknesses, automated attack frameworks are designed to systematically probe these systems.
Second, the correction stage focuses on mitigating unsafe behaviors through task-specific supervised learning. This includes empirical, efficient safety alignment techniques and provable defenses that provide future-proof guarantees. These strategies ensure that AI/ML systems generate safe outputs even in adversarial conditions while preserving their functionality in real-world applications.
Finally, the steering stage proposes a training-free, rule-based reasoning framework to guide AI/ML behavior at test time, which steers the model's outputs following predefined rules and provides transparent rationales for decision-making processes, thereby enhancing public trust in the deployed AI/ML systems.
Through these stages, the dissertation contributes to AI/ML systems that are both functional and trustworthy for societal use. It concludes by outlining future research directions for user-centered, scalable, and trustworthy AI/ML systems with formal guarantees
A cell-based system to study resolution of transcription-induced topological stress
No full textDuring transcription, RNA polymerases travel forward and scan the DNA double strand helix while generating positive and negative supercoils ahead of and behind the polymerase machinery, respectively. We call it transcription-induced topological stress. These supercoils block polymerase from moving forward and hinder the transcription progression. Under physiological conditions, cells require DNA topoisomerases I (TOP1) to solve DNA topological stress by inducing a nick on one single strand of DNA and forming a transient TOP1-DNA covalent cleavage complex (TOP1cc). Endogenous DNA lesions and exogenous carcinogens like TOP1 inhibitor camptothecin (CPT) can trap TOP1 on DNA and make the TOP1cc irreversible. Persistent TOP1cc impedes essential DNA transaction, contributing to replication or transcription-induced double strand breaks and genome instability when colliding with replication or transcription machineries. Protease, TDP1 and the involvement of single strand break repair (SSBR) factors help resolve the persistent TOP1cc and single strand DNA ligation. Cancer cells, characterized by highly active physiological processes, experience elevated transcription activity. It leads to increased DNA topological stress which requires enhanced TOP1 activity. Due to the high DNA damage rate in cancer cells, TOP1cc is more frequently irreversible and requires efficient repair mechanisms to maintain genomic stability. By utilizing a U2OS cell-based YFP-MS2 transcription reporter system, we found that macroH2A1.1, a histone splicing variant with a unique macro domain, has a role in irreversible TOP1cc removal by promoting downstream repair factors XRCC1 recruitment at the transcriptionally active sites