71446 research outputs found
Sort by
Genomic and Proteomic Regulation of Activity-Dependent Neuronal Function
No full textAs postmitotic cells, neurons face the unique challenge of dynamically responding to changing environments while simultaneously stably encoding information over the lifetime of an animal. To accomplish this, neurons utilize an intricate system of molecular mechanisms. Much of this regulatory program is activated in neurons in response to experience—neuronal activity triggers calcium influx, in turn transiently activating signaling cascades that rapidly induce diverse downstream pathways, including post-translational protein modifications, transcriptomic regulation, and synaptic remodeling. Here, we present three studies that investigate distinct molecular mechanisms involved in activity-dependent neuronal regulation.
We begin by examining transcriptomic regulation through the lens of the immediate early gene, NR4A1. Using biochemical techniques, we discover that the COMPASS complex—a histone H3K4 methyltransferase and transcriptional regulator—assembles together with NR4A1 in neurons. In NR4A1 knockout mouse brains, we find that chromatin binding of the COMPASS complex is reduced at NR4A1 binding sites near genes that regulate dendritic spines, suggesting that NR4A1 facilitates COMPASS complex recruitment and subsequent gene activation for neuronal function. Consistent with this finding, we observe that distal regulatory elements have altered accessibility in NR4A1 knockout mice, suggesting dysregulation by the COMPASS complex. Furthermore, RNA-sequencing of the hippocampus reveals that loss of NR4A1 leads to downregulation of numerous synaptic regulatory genes and ion channels. Finally, we demonstrate that NR4A1 deletion reduces Sag current in CA1 pyramidal neurons, consistent with the observed transcriptomic changes and indicative of disrupted neuronal function in the absence of NR4A1.
In our second study, we look at the impact of protein phosphorylation on brain function and gene expression. Here, we show that MeCP2 is phosphorylated at four residues in the mouse brain (S86, S274, T308, and S421) in response to neuronal activity and generate a quadruple knock-in (QKI) mouse line in which all four activity-dependent sites are mutated to alanines to prevent phosphorylation. Electrophysiological recordings from the retinogeniculate synapse of QKI mice reveal that while synapse elimination is initially normal at P14, it is significantly compromised at P20. We thus propose a model in which activity-induced phosphorylation of MeCP2 is critical for the proper timing of retinogeniculate synapse maturation during the early postnatal period.
Lastly, we demonstrate that neuronal activity-dependent phosphorylation events primarily occur in disordered regions of proteins, where they control structural transitions and modulate condensation dynamics in the nucleus. We focused on one of these events, the phosphorylation of histone methyltransferase SETD2, and found that disrupting activity-dependent and condensation-modulating phosphorylation of SETD2 impairs histone methylation, RNA splicing programs, neuronal excitability, and leads to changes in social behavior akin to autism-like behavior in mice.Neuroscienc
In Front of the Many: Language and Public Life in Kazakhstan's Largest City
No full textThis dissertation examines how citizens in Almaty, Kazakhstan’s largest city, work to make the Kazakh language an ordinary medium of public life. Although the state has promoted Kazakh for more than three decades since independence, Russian remains the default language of efficiency, civility, and urban order. Based on 17 months of ethnographic fieldwork, including participant observation, interviews, and media analysis, the study follows independent publishers, consumer activists, and conversation clubs as they work to make Kazakh not just a marker of national identity, but a language of everyday use.
I argue that sincerity—doing things willingly and with feeling—functions as a developmental affect among my interlocutors, a sign that society is moving beyond a socialist past remembered for its top-down coordination of collective life. During the Soviet period, public life in Kazakhstan was organized through a dense network of institutions that operated largely in Russian. After independence, that system unraveled. While the Kazakhstani state elevates Kazakh as the “state language” and links its vitality to national modernization, it lacks the kind of authoritative capacity that once organized public life under socialism, and cannot openly act as if it did. As a result, the realization of ‘the public’ is left to be worked out elsewhere. In this gap between proclamation and enactment, citizens labor to do what the state does not: make Kazakh usable as a shared language among strangers.
In Part I, I show how independent publishers treat market-based demand as evidence of genuine desire and quality as a sign of sincere care, aiming to make Kazakh a medium for accessing a wider world without Russian mediation. In Part II, I show how consumer activists and ordinary citizens mobilize linguistic and consumer rights—despite unreliable enforcement—through polite, procedural requests, in order to make service in Kazakh the expected default rather than exception. In Part III, I show how conversation clubs support a chosen desire to practice and learn Kazakh through cultivating a distinctive type of togetherness, which collaboratively reworks shame and gives hesitant participants a space to practice appearing publicly as legitimate speakers.
Through these practices of “developing the language,” people seek to make social life feel coherent, shared, and oriented toward a common future. The Kazakh phrase köpşiliktiñ aldında (“in front of the many”), which can be glossed as both “publicly” and “before others,” captures how these everyday encounters among strangers make ‘the public’ itself tangible—an imagined collective moving forward through Kazakh as a shared civic language.Anthropolog
Problems in High-Dimensional Estimation and Large Language Models
No full textThis dissertation investigates critical problems at the intersection of high-dimensional statistics and the rapidly advancing field of large language models (LLMs), forging a narrative that bridges foundational theory with state-of-the-art applications. The work is presented in two interconnected parts, unified by the theme that principles of high-dimensional estimation provide a powerful framework for addressing key challenges in modern artificial intelligence.
The first part establishes a rigorous theoretical foundation for high-dimensional estimation. We present a sharp asymptotic analysis of a spectral method, inspired by Principal Hessian Directions, for learning multi-index models from nonlinear measurements. In a high-dimensional regime where data and signal dimensions grow proportionally, our analysis reveals a distinct phase transition phenomenon. We derive a set of deterministic fixed-point equations that precisely characterize the method's performance, offering an exact quantification of the alignment between the estimated and true subspaces. This theoretical contribution extends prior work from single-signal to multi-signal recovery, deepening our understanding of learning and signal processing in high-dimensional spaces.
The second part of this dissertation transitions from theory to practice, demonstrating how the mathematical rigor developed in the first part can be leveraged to solve pressing challenges in the development and deployment of LLMs. We introduce three novel frameworks. First, we propose a principled method for the Selection of LLM Fine-Tuning Data based on Orthogonal Rules, which uses the Determinantal Point Process (DPP) to select a diverse and non-redundant set of data quality metrics. This approach, grounded in the concept of orthogonality, significantly improves the efficiency and performance of model fine-tuning across multiple domains. Second, we introduce RuleAdapter, a dynamic framework for training multi-attribute reward models in Reinforcement Learning from Human Feedback (RLHF). Motivated by information theory, RuleAdapter adaptively selects the most critical safety rules for each context, leading to state-of-the-art safety performance and demonstrably more trustworthy LLMs. Third, we propose Semantic Volume, a novel, unsupervised geometric measure for quantifying and detecting both internal (model-based) and external (query-based) uncertainty in LLMs. By linking this measure to differential entropy, we provide a robust and interpretable method to enhance model reliability and mitigate hallucinations.
Collectively, this dissertation demonstrates that a deep understanding of high-dimensional systems is not merely a theoretical pursuit but an essential tool for building more robust, trustworthy, and efficient large language models. The presented research offers new theoretical insights into high-dimensional learning and delivers practical, mathematically-grounded methodologies that advance the state-of-the-art in the responsible development of artificial intelligence.Engineering and Applied Sciences - Applied Mat
Scaling Computational Connectomics
No full textConnectomics is an emerging interdisciplinary field at the intersection of neuroscience and computer science that seeks to map and analyze the complex wiring diagrams of brains. Studying connectomes offers profound mechanistic insights into how the structural organization of the nervous system gives rise to its function, deepens our understanding of neurodegenerative diseases, and may even reveal how memories are encoded and stored in the brain. However, generating connectome datasets presents substantial technical challenges in image acquisition, alignment, registration, and segmentation. Furthermore, the resulting petascale datasets demand advanced computational techniques to extract meaningful patterns at scale. Because of these challenges, current state-of-the-art connectomics datasets are both time-intensive to produce and limited in spatial extent–typically not exceeding a cubic millimeter–thereby restricting reconstructions to small organisms such as fruit flies or tiny fragments of mammalian brains. To address these limitations, this thesis introduces a suite of novel computational methods designed to overcome key bottlenecks in connectome reconstruction and analysis, thereby accelerating progress in the field. These contributions range from isotropic microscopy volume reconstruction and machine learning models of neuronal morphology to interactive tools for exploring network motifs in connectomes. By enabling faster and more scalable connectome reconstruction, this work paves the way for large-scale comparative analyses across species, disease states, and developmental stages, and ultimately could advance our understanding of how learning and memory are implemented in the brain.Engineering and Applied Sciences - Computer Scienc
Mapping chemical and genetic regulators of aryl hydrocarbon receptor signaling
No full textThe human aryl hydrocarbon receptor (AHR) integrates chemical signals derived from the environment, gut microbes, and endogenous sources to regulate processes ranging from intestinal barrier integrity to xenobiotic detoxification. Despite strong evidence that dysregulation of AHR signaling is a causal factor in metabolic and auto-immune disorders, we currently lack a comprehensive understanding of the factors that regulate AHR activity in human cells. Here, we use genome-scale CRISPR screening to systematically identify regulators of AHR signaling in hepatocytes. The resulting datasets recapitulate the core AHR signaling pathway and identify a large network of regulators. Many of these factors have roles beyond AHR signaling, reflecting that AHR signaling is deeply integrated into human cell biology. We further dissect this network to reveal novel modes of regulation of AHR expression, protein levels, and signaling. Through complementary analyses, we also examine how AHR activation differs across cell types, highlighting the highly context-specific nature of AHR signaling. Together, our findings show that the activation and regulation of AHR varies with both cell type and ligand and reveal potential nodes for targeted modulation of AHR signaling for therapeutic benefit. Overall, our results define the regulatory network that underpins AHR activation, with implications for understanding host-microbe interactions, integrative chemosensation, and the etiology of metabolic and inflammatory disorders.Biological and Biomedical Science
SID-2 Transduces RNAi-Dependent and -Independent Responses to the Environment in Caenorhabditis elegans
No full textEpigenetic mechanisms have been shown to underlie the adaptation of organisms to changing environmental conditions, a process that enhances their and their progeny’s survival. One unique example occurs in the nematode Caenorhabditis elegans, in which dsRNA from the environment can be taken up by the RNA transporter SID-2 to induce systemic and transgenerational silencing of homologous sequences. The environmental RNAi response observed in the laboratory is likely a byproduct of the natural role of sid-2, which remains unknown. Here we investigated ecologically relevant, sid-2-dependent responses to changes in the microbiome. We confirmed prior results showing that worms trained to avoid the pathogen Pseudomonas aeruginosa transmit this behavior to their offspring in a sid-2 dependent manner, likely acting proximal to RNAi pathway activities. We also discovered that changes in microbiome constituents heritably reduced RNAi efficiency and that this change in response to the microbiome shift requires sid-2 as well as the RNA transporter sid-1, a necessary component of the systemic RNAi pathway. Finally, we identified novel sid-2-dependent and microbiome-dependent developmental, behavioral, and physiological phenotypes. These responses are independent of the canonical RNAi pathways but dependent on extracellular vesicle (EV)-associated proteins and factors involved with EV biogenesis and trafficking. Collectively, our results suggest a model wherein SID-2 transduces both RNAi-dependent and RNAi-independent environmental responses.Biology, Molecular and Cellula
Circadian regulation of brain border immunity across the lifespan
No full textLife on Earth is governed by the rhythmic rotation of the planet, dividing time into periods of light and dark to which most life forms are entrained. In mammals, time-keeping is genetically encoded by the circadian clock: a transcription-translation feedback loop lasting twenty-four hours that synchronizes cellular physiology to time-of-day. Core organismal functions are carried out under circadian regulation including even innate immunity, the body’s first line of defense against external pathogens. It is increasingly apparent that circadian disruptions, such as night shift work or aging-related sleep fragmentation, are major risk factors for the development of neurodegenerative diseases including Alzheimer’s Disease (AD). The cellular mechanisms involved in this risk are less clear and may involve perturbation of the timing and function of brain innate immune cells.
To further understand how circadian regulation and dysregulation of brain immunity could contribute to disease pathogenesis, we generated an unbiased transcriptional atlas of diurnal rhythms in the brain immune compartment. We identified brain border-associated macrophages (BAMs), a small and long-lived population of perivascular and leptomeningeal scavengers, as highly rhythmic immune cells both in phenotype and function. BAMs are exceptionally efficient at engulfment, by far exceeding other brain myeloid populations in acute uptake of extracellular substrates including amyloid-beta (Aβ), a peptide involved in AD pathogenesis. We found that BAM engulfment capacity is rhythmic and peaks during the murine rest phase. Rhythmicity in BAM engulfment is regulated by the clock gene Bmal1, accompanied by a coordinated wave of upregulation of endocytic genes, and mediated by the fast-recycling scavenger receptor CD206. In the aged brain, BAMs exhibit perturbed clock gene expression, downregulation of CD206, and profoundly impaired uptake of extracellular material including Aβ. Finally, in a mouse model of AD, we found that BAM-specific deletion of Bmal1 worsens perivascular and leptomeningeal Aβ plaque deposition. Together, these results highlight the remarkable endocytic capacity of BAMs and propose one mechanism by which circadian perturbations may precipitate neurodegenerative disease: by disrupting the timing of brain border innate immune functions including the clearance of pathogenic proteins.
In parallel, we have sought to generate human models of BAMs using induced pluripotent stem cells (iPSCs). iPS-derived macrophages better recapitulate features of tissue-resident macrophages than circulating monocytes or bone marrow hematopoietic stem cells (HSCs). However, little is known of whether this technology may be used for the generation and study of BAMs. We found that early postnatal xenotransplantation of iPS-derived macrophages to the murine brain leads to robust formation of a human brain border immune compartment. Human xenotransplanted BAMs (xBAMs) develop throughout the leptomeninges and along brain vasculature of chimeric mice. Via transcriptional phenotyping, we identified a conserved signature distinguishing BAMs from microglia, the other long-lived tissue-resident macrophage of the brain. In functional assays, xBAMs exhibit compartment-restricted sampling of draining brain proteins and share the hyper-endocytic phenotype of murine BAMs. Notably, xBAMs are highly enriched for CD206 expression and by far surpass other brain immune cells in acute Aβ uptake. Thus, the “xBAM platform” may be used to model human BAM function in vivo. In addition, we have developed adapted differentiation protocols to generate iPS-derived macrophages in vitro resembling BAMs (the “iBAM platform”). iBAMs share transcriptional and functional features of in vivo BAMs including a hyper-endocytic phenotype and enhanced engulfment capacity. Together, these data indicate that the remarkable endocytic capacity of BAMs is conserved across species and support our ongoing efforts to target BAM scavenging in the context of aging and amyloidosis.Neuroscienc
Exploring the mechanisms and structures of N-acyl-D-Asn prodrug peptidases
No full textBacteria engage in chemical warfare to compete with other microorganisms in their environment. Throughout our history, we have benefited from this warfare by using the specialized metabolites produced as the basis for potent antimicrobial agents. Thus, understanding how bacteria produce these potential therapeutics and pesticides can help us increase the yields of already known metabolites, illuminate undiscovered metabolites, and even design novel metabolites based on natural ones.
In this thesis, I further our understanding of a class of specialized metabolites produced by bacteria called the N-acyl-D-Asn prodrug toxins. In Chapter 1, I delve into the various prodrug mechanisms developed by bacteria with a focus on those employing N-acyl-D-Asn prodrug motifs. In Chapter 2, I present the first structural and mechanistic studies of a type II prodrug peptidase, ZmaM, revealing a unique domain arrangement and an interdependence between these modules. In Chapter 3, I further our understanding of type I prodrug peptidases by probing how the type I prodrug peptidase ClbP recognizes its substrate and exploring the possible transporter partners of ClbP. In Chapter 4, I discuss how these results apply to the broader family of prodrug peptidases and can lead to future discoveries.Chemical Biolog
Entangled Ritual and Architectural Practices at North Saqqara
No full textThis doctoral dissertation challenges some prevalent assumptions about ancient Egyptian funerary architecture. Using the cemetery of North Saqqara as a case study, I question the idea that ancient Egyptian tombs were only built to bury the dead, and show evidence that the elite mastabas of the earliest dynasties were used for rituals before interment took place. The structures were eventually transformed for the funeral, which required changing fundamental aspects of the buildings. In their previous shapes, the structures included spaces large enough to gather crowds, and doubled accesses that enabled the circular movement of people.
While the North Saqqara plateau was one of the most important sites during the Early Dynastic Period of Egypt, our understanding of it has been hindered by the limited documentation available to the scientific community. This doctoral dissertation contextualizes the unpublished archives of the archaeological excavations carried out at the site in the 20th century, which contain information about more than 160 excavated but unpublished structures. I complement the documentation of Cecil M. Firth and Walter B. Emery with data gathered in two season of topographic survey at the site. Using a high accuracy GNSS receiver and digital photogrammetry, I undertook a detailed topographic survey. As a result, I was able to trace the whereabouts of many of the structures whose plans were kept in the archives, succeeding in reconstructing a sizable portion of the missing plan of the site.
In the last part of the dissertation, I suggest a method to research the function of the spaces of the mastabas in their various stages of construction and use. In view of the very limited evidence about the rituals, I suggest studying the agency of the structures to affect the sensorial perceptions of the people involved in the rituals as an avenue to inquire about how the spaces could have been used. Specifically, I provide a series of case studies where I researched how four structures built during the First and the Second Egyptian dynasties shaped sound in certain ritual scenarios. Sounds produced in the mastabas were transformed as they traveled through space, from their source to their full dissipation. How sound was heard depended on the frequency of the sound emitted, as well as on the specific shapes and construction materials used in the structures.Near Eastern Languages and Civilization
The Galactic Halo in Stars and Shadow
No full textThe Galactic halo encodes both the ancient assembly history of the Milky Way and its ongoing interactions with satellites and dark matter. In this dissertation, I present six studies that together develop a new picture of the halo as a tilted, broken, and dynamically complex structure. Using halo stars from the H3 Survey, I show that the inner halo—dominated by the Gaia–Sausage–Enceladus merger—is best described as a tilted triaxial ellipsoid with a doubly broken density profile. I further demonstrate that the halo exhibits strong kinematic asymmetries and that stellar orbits preserve signatures of a misaligned Galactic potential, providing new constraints on the shape and orientation of the surrounding dark matter halo. Through both controlled experiments and cosmological simulations, I establish that tilted halos are long-lived, common in a cosmological context, and naturally excite the Milky Way’s warp and flare, offering a resolution to the decades-old mystery of the Galactic warp. Finally, I explore hypervelocity stars as dynamical probes of the halo’s hidden components. By reconstructing their trajectories, I find that a subset originates not in the Galactic Center but in the Large Magellanic Cloud, providing strong evidence for a massive black hole in our nearest satellite and a new constraint on its past orbit. Together, these results reveal the Milky Way halo as a dynamic structure that records ancient mergers and present-day interactions—traced both in stars and in shadow.Astronom