Washington University

NIH Research Projects · FY 2026 · 2026-02

SUMMARY HIV and Mycobacterium tuberculosis (Mtb) are among the world’s deadliest infections, and co-infection is particularly devastating because each pathogen accelerates the progression and severity of the other. Notably, for reasons that are not well understood, people living with HIV (PLWH) remain at elevated TB risk despite effective antiretroviral therapy (ART) and viral suppression. Our goal is to define how HIV-TB co-infection, even with virologic control, impairs Mtb immunity. Macrophages and CD4+ T cells are central to the pathogenesis of both diseases; Mtb infects lung macrophages and depends to CD4+ T cells to prevent disease progression, while HIV infects both macrophages and CD4+ T cells. Macrophages can harbor latent HIV proviruses, which persist despite treatment. Both pathogens reprogram macrophage metabolism, which is intimately linked to antimicrobial functions. However, the impact of co-infection on macrophage immunometabolism and Mtb control remains unclear. In addition, although antiretroviral treatment (ART) can restore CD4+ T cell counts to normal ranges, the T cells often remain dysfunctional and exhibit signs of exhaustion. We unite leading HIV and TB investigators and leverage unique cellular and animal models of HIV latency. We hypothesize that, despite ART and virologic suppression, PLWH experience macrophage immunometabolic reprogramming that enhances TB susceptibility and that ART-restored CD4+ T cells are dysfunctional and fail to enhance microbicidal properties of macrophages. Using dual-reporter human macrophage models and ex vivo studies of PLWH and controls, we will profile antimicrobial and immunometabolic responses of Mtb-infected macrophages, and we will test the contribution of exhausted CD4+ T cells to macrophage dysfunction. A long-standing obstacle to studying HIV- TB coinfection has been the absence of a small animal model. We will use a novel humanized mouse model that recapitulates human lung immune cell populations and HIV infection dynamics, including active viral replication, latency establishment, and viral suppression with ART. We will assess how untreated and treated HIV infection affects Mtb pathogenesis, assessing pathogen burden, immunometabolic responses, and cellular infection patterns, using spectral flow cytometry, scRNAseq and metabolic profiling. These studies will clarify the molecular basis of immune dysfunction during HIV-TB co-infection and inform on host-directed therapies to restore immune function and improve outcomes.

NIH Research Projects · FY 2026 · 2026-02

PROJECT SUMMARY Typical aging features progressive muscle weakness and loss of physical ability which impacts quality of life for millions of Americans. Beyond loss of lean muscle mass, muscle weakness and physical dysfunction are strongly associated with increasing ectopic adipose within the muscle boundary. The mechanism behind this association is not clear, but evidence suggests that this intramuscular adipose tissue (IMAT) secretes signals that locally impair muscle contraction. Paralleling work in visceral adipose tissue, we propose that IMAT secreted signals are dynamic rather than static, and as such are influenced by the local and systemic environment. More specifically, it is our central hypothesis that, like visceral adipose, IMAT has reduced quality with age due to senescent changes in cellular composition and that these senescent features are modifiable by exercise or a combination of exercise and pharmacology, to improve signaling. To address our hypothesis we will complete two parallel aims, one in people and one in mice. Both aims will have two age groups (~30% and ~80% of lifespan) and will evaluate the effect of aging and a 12 week progressive resistance exercise program on IMAT senescence. The primary objective of the first aim is to assess IMAT senescent and signaling changes with age and exercise. To date, IMAT changes with age and exercise have only been assessed with non-invasive imaging which has limited our ability to target cellular functions. The primary objective of the second aim is to determine whether age- and exercise-responsive human IMAT features are present in a mouse IMAT model. Rodents develop little IMAT with age and thus there are few studies of IMAT in mice. We will examine the suitability of a toxin-induced IMAT model in mice to study the intersection of aging and exercise on IMAT. The long-term goal of this work is to develop a feed-forward system where therapeutic targets are identified from human tissue and mechanistically assessed in mice. Toward that goal, these studies will improve our understanding of the basic biology of aging IMAT, the plasticity of IMAT signaling and the translational utility of a toxin-induced IMAT model.

NIH Research Projects · FY 2026 · 2026-02

1 PROJECT SUMMARY 2 Our broad, long-term objective is to leverage the intimate relationship between sleep and neurodegenerative disease 3 to diminish, delay, and prevent pathology. Our approach is based on first-principles reasoning. First, neurodegenera- 4 tive disease is functionally defined by the failure of homeostatic set-points to maintain complex computation. Second, 5 sleep is widely understood to subserve the homeostatic reinstatement of set-points necessary for complex computation. 6 Thus, we propose that the well-established association between sleep disturbances and neurodegenerative disease is 7 the result of sleep’s progressive inability to achieve its central homeostatic set-point. This set-point has recently been 8 identified as criticality, an optimal computational regime that maximizes information processing. Based on extensive 9 preliminary data, we propose that: 1) criticality is the locus of tauopathy’s impact on neuronal activity, 2) reinforcement 10 of criticality via homeostatic sleep enhancement prevents neurodegenerative disease, and 3) mechanistically, sleep- 11 related patterns of neuronal activity reduce DNA damage, while disease-related patterns increase DNA damage. To 12 move forward, it is imperative that we identify how sleep reinforces function, and ask whether this process can be aug- 13 mented to counteract disease. To make progress in this arena, we will apply novel computational tools to continuous, 14 months-long, multi-site recordings of neural activity in P301S mice, a robust model of tauopathy. In this context, we will 15 deploy two distinct, non-invasive, and rapidly translatable methods of sleep-enhancement to bolster homeostatic control 16 of criticality. We will then quantify neurodegenerative disease progression and cognitive function using a sophisticated 17 prey capture task. To evaluate our hypothesis that criticality mechanistically links sleep and neurodegenerative disease, 18 we will disentangle criticality from the myriad of changes that occur during sleep. We will impose sleep-related patterns 19 of neuronal spiking in the brains of tauopathy mice and measure subsequent 1) neuronal health (DNA damage), and 20 2) fidelity of the critical set-point. Likewise, we will impose supercritical dynamics, which are characteristic of both 21 neurodegenerative disease and sleep deprivation, in the WT brain and measure the same end points. We present 22 three aims: test the hypothesis that 1) inducing homeostatic supercompensation promotes criticality and mitigates sub- 23 sequent tauopathy, 2) pharmacological enhancement of sleep promotes criticality and reduces subsequent tauopathy, 24 and 3) critical network dynamics directly influence neuronal health. These experiments test a novel mechanism by 25 which sleep and neurodegenerative disease may interact, and have the potential to reveal a fundamental principle of 26 neurobiology—that activity regimes directly determine neuronal integrity. In terms of the mission of the agency, our test- 27 ing of this theory takes the form of clinically-relevant and immediately applicable interventions that are complementary 28 to extant molecular biological approaches to treating neurodegeneration.

NSF Awards · FY 2026 · 2026-02

The Physics Department at Washington University in St. Louis (WashU) addresses a fundamental challenge in modern physics education: preparing the next generation of scientists to combine traditional physics understanding with Artificial Intelligence (AI) methods. Through this Research Experiences for Undergraduates (REU) site, WashU provides a fully immersive 10-week experience for undergraduate students to perform cutting-edge research in fundamental physics. The participants, selected nationwide with particular focus on students in the Midwestern region, receive stipends, housing, and travel funds to ensure broad accessibility. The program begins with an AI boot camp providing hands-on machine learning training, after which participants join research groups and apply what they learn to analyze large datasets and address realistic physics problems. The REU program benefits society by preparing participants for multiple career paths, developing technical skills that are highly transferable across physics subfields and industries beyond academia. This REU Site addresses fundamental challenges in modern physics through the innovative application of AI across multiple subfields including astrophysics, nuclear physics, atomic, molecular, and optical physics, condensed matter physics, and biophysics. The program advances the field by unifying physics research through the lens of AI, focusing on three transformative machine learning applications: image segmentation & recognition, time-series or spectral analysis, and reduced models & feature identification. These computational methods tackle crucial research questions ranging from identifying emergent properties in 2D materials and classifying meteorite characteristics to determining exoplanet atmospheric composition, analyzing X-ray emission from black hole binaries, uncovering features of nucleon-nucleon interactions, and reconstructing material magnetization maps. The program's methodological framework represents a significant advancement in physics research, developing machine learning tools that transcend specific subfields and establish a versatile approach applicable to physics and other scientific disciplines. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2026-02

PROJECT SUMMARY/ABSTRACT The demand for liver transplantation far exceeds the number of suitable donor organs. Despite the organ shortage, due to increased risk for ischemia reperfusion injury (IRI), many marginal donor livers with steatosis go unused. While some IRI is unavoidable in all liver operations, in the setting of liver transplantation this can lead to primary non-function or early allograft dysfunction. The increasing prevalence of obesity associated metabolic dysfunction-associated steatotic liver disease increases the demand for liver transplant while also reducing the number of suitable donor livers. To address this shortfall, we must attenuate IRI in steatotic livers. However, there are no pharmacological interventions available to achieve this goal. Development of novel therapeutics is costly and time intensive. A practical alternative is to repurpose FDA approved drugs in clinical use. Fomepizole is an appealing candidate; it is used for the treatment of ethylene glycol poisoning, and recent studies have found fomepizole to be protective against acetaminophen (APAP) hepatotoxicity. Given similarities between APAP hepatoxicity and IRI, we will test the hypothesis that fomepizole protects against IRI by modulation of lipid metabolites and determine if the same mechanisms apply in steatotic livers. While there are no pharmacological agents available to reduce IRI, the use of machine perfusion (MP) procurement (as opposed to traditional static cold storage) has gained traction in the past decade and enabled the use of marginal donor livers by reducing IRI complications. However, many donor livers still do not pass viability testing with MP and are ultimately discarded. The use of adjunctive therapies may provide additional benefits over standard MP in reducing IRI, thus further expanding the use of marginal donor livers. Hence, testing therapeutics in the setting of MP, represents a novel way to address the donor shortage. In this proposal, we will investigate if fomepizole provides additional benefit in normothermic machine perfusion using an isolated perfused rat liver model. In alignment with the R03 funding mechanism, this proposal is a logical extension of my K08 proposal and provides a clear path toward an R01. The primary goals of this proposal are to delineate the mechanisms by which fomepizole reduces liver IRI and establish collaborations and innovative platforms to test therapeutics within the evolving landscape of liver preservation. At completion, I expect to provide support for repurposing fomepizole as a primary or adjunctive therapy to salvage marginal grafts, thus enabling safe, expanded use of donor grafts to decrease waitlist mortality and improve patient outcomes.

NIH Research Projects · FY 2026 · 2026-02

This application seeks funding to continue a series of international and interdisciplinary research conferences on cognition, speech communication and aging. Scientists approaching this problem have typically been working in the areas of sensory and perceptual processing, especially hearing, cognitive psychology, or auditory neuroscience. Aging can have a negative impact on all of these systems and understanding how deficits in these areas, both independently and combined, affect speech understanding requires cross- disciplinary understanding and collaboration. We have developed a set of learning objectives for the conference and we currently have agreements from leading national and international researchers for oral presentations addressing these objectives (1) age-related changes in auditory perception and physiology; (2) animal studies of aging and hearing; (3) age-related changes in speech production, perception, and understanding; (4) cognition and aging; (5) multi-sensory perception and aging; (6) hearing loss and assistive devices; and (7) technical advances. In addition, we have requested funds to provide student scholarships to promote and support early-stage investigators. An important aspect of the conference is to encourage early- stage investigators in hearing research. By bringing together scholars actively involved in research across the disciplines, it is hoped that further progress will be made in understanding and remedying the speech- communication difficulties of older adults.

NIH Research Projects · FY 2026 · 2026-02

ABSTRACT Despite aggressive treatments, the five-year overall survival rate for advanced head and neck squamous cell carcinoma (HNSCC) remains below 50%. The global burden of HNSCC is influenced by tobacco-derived carcinogens and excessive alcohol consumption, with oropharyngeal tumors increasingly linked to oncogenic human papillomavirus (HPV) infections. Current treatments for locoregionally advanced HNSCC include surgery followed by adjuvant external beam radiation therapy (XRT) or definitive concurrent chemoradiation, but radiation resistance is common, leading to poor outcomes. Novel strategies are needed for better detection, treatment, and monitoring of treatment responses. Our research is based on the discovery of radiation-inducible antigens that are upregulated on cancer cell surfaces post-XRT. We identified Tax-interacting protein 1 (TIP1) as a tumor- associated antigen upregulated following XRT. Anti-TIP1 antibodies undergo endocytosis, delivering payloads specifically to tumor cells and show XRT-induced tumor-specific delivery in vivo. We aim to leverage XRT- induced TIP1 expression to guide theranostic (therapy+diagnostic) radioimmunoconjugates for detection, treatment, and monitoring response to treatment. Radiopharmaceuticals are emerging as effective treatments for various cancers. In our recent publication, we developed human anti-TIP1 antibodies (L111) and labeled them with [89Zr]Zr, demonstrating specific cancer detection by PET imaging in preclinical cancer models. We propose radiolabeling L111 antibodies with the β-emitter [177Lu]Lu and evaluating their therapeutic potential for HNSCC. We hypothesize that Immuno-PET using [89Zr]Zr-L111 will monitor HNSCC response to treatment and that [177Lu]Lu-L111 will treat residual tumors unresponsive to XRT. Our objectives include assessing [89Zr]Zr- L111 for non-invasive PET imaging of HNSCC tumors after XRT and [177Lu]Lu-L111 therapy and determining if this strategy applies to both HPV+ and HPV- HNSCC subtypes. Aim 1 will evaluate the efficacy of TIP1-targeted radioimmunoconjugates in vitro using HPV+ and HPV- HNSCC cell lines, exploring XRT-driven TIP1 upregulation, enhanced binding, internalization, and cell killing efficacy of [177Lu]Lu-L111. We will also optimize radiation doses and schedules. Aim 2 will evaluate these conjugates in vivo using syngeneic orthotopic murine models treated with XRT, studying biodistribution, PET imaging, dosimetry, maximum tolerated dose, and therapeutic efficacy. Our study aims to revolutionize HNSCC management by improving detection, treatment, and monitoring, potentially leading to tailored therapies based on HPV status and enhancing survival rates through personalized interventions, marking a significant advancement in clinical practice.

NIH Research Projects · FY 2026 · 2026-02

PROJECT SUMMARY Neural circuit development and function depends on precise interactions between neurons and glia. Astrocytes, the primary peri-synaptic glia, mediate synapse formation, stability, and function. Neuron-astrocyte crosstalk is facilitated by complex protein-protein interactions, and loss of these interactions contributes to circuit instability in many neurological disorders. Thus, understanding the mechanisms that regulate neuron-astrocyte communication is of broad clinical importance. Glycosylation is a posttranslational modification that regulates protein stability and binding through addition of sugar groups to specific amino acids. Mutation of genes in glycosylation pathways cause congenital disorders of glycosylation (CDGs), a group of monogenic disorders associated with neurological dysfunction, including epilepsy, autism, and cerebellar degeneration. The mechanisms underlying neurological dysfunction in CDGs remain unknown. Here, I focus on ALG8, an enzyme in the N-glycosylation pathway. To explore the molecular underpinnings of ALG8-CDG, I first needed to develop models that reflect the patient population. To this end, I generated a predicted null zebrafish line (alg8stl973) and human embryonic stem cell (hESC) lines with a missense mutation (p.Thr47Pro) found in ALG8-CDG patients. My preliminary data revealed a decrease in astrocyte numbers in the brains of alg8 mutant zebrafish with no change in total cells, and reduced proliferation of ALG8 mutant hESC-derived astrocytes. Moreover, in alg8stl973 fish, astrocyte morphological complexity is reduced. As astrocyte-synapse association is necessary for neuronal signaling, I hypothesize that defective glycosylation disrupts specification and maturation of astroglia, which in turn drives circuit imbalance and CDG-associated behavioral deficits. To address this hypothesis, I will leverage preexisting transgenic tools in zebrafish to label astrocytes and test whether changes in proliferation and/or cell death result in reduced astrocytes in alg8stl973 fish (Aim 1). Furthermore, I will use biochemistry and in vivo imaging to characterize how loss of alg8 impacts the glycosylation status of one key regulator of astrocyte morphogenesis: NrCam (Aim 2). Finally, as ALG8 is expressed in all neural cell types, I will use cell-type specific rescue in fish and co-culture of hESC-derived neural cells to determine which cell type(s) drive changes in astrocyte morphology and synaptogenesis in ALG8-CDG (Aim 3). My long-term goal is to define common molecular changes in brain development across distinct CDGs. Critically, various CDG subtypes result in common neurological symptoms, but the cellular and molecular underpinnings of these phenotypes are largely unknown. Similar to my preliminary findings in ALG8-CDG models, recent work indicates that astrogenesis is altered in a mouse model of MGAT5-CDG, a CDG with defective N-glycosylation. Thus, I anticipate that my findings will be broadly applicable to the CDG community and will enhance our fundamental understanding of how glycosylation shapes brain development.

NIH Research Projects · FY 2026 · 2026-02

Project Summary Cytotoxic CD8+ T cells confer adaptive immune protection against viruses, intracellular pathogens, and cancer. Efficient CD8+ T cell immunity depends on regulation of a specific set of transcription factors as well as global cellular activity, including nutrient metabolism. We have found in our preliminary study that the activity of the TCR-induced phosphatase Calcineurin is essential for protection of the transcription factor TCF-1, which is critical for immunological memory, and for reprogramming amino acid metabolism for increase biogenesis in highly proliferative T cells through suppression of the catabolic process known as autophagy. Despite the importance of TCF-1 in immunological memory and durable T cell response and the demand for metabolic reprogramming, regulatory mechanisms by which T cells respond to antigens and inflammation and establish highly controlled states are not largely unknown. We initially found that inhibition of Calcineurin by the immunosuppressant Cyclosporin A promotes TCF-1 degradation through autophagy, and our additional experiments have suggested that Calcineurin is the central player that globally restricts the engagement of autophagy in activated T cells exposed to inflammatory cues, which not only leads to protection of TCF-1 but also governs amino acid and protein metabolism during T cell immune responses. In this proposed study, we will conduct multi-angled mechanistic analyses of the actions of Calcineurin in cytotoxic T cells to define the unexplored roles of this phosphatase. Given that Cyclosporin A and its related drug, FK-506, have been broadly used as immunosuppressants in patients undergoing organ transplantation and treatment of autoimmunity, expected findings from this study will provide new insights into T cell biology and into altered immunity in patients with pharmacological immune suppression.

NIH Research Projects · FY 2026 · 2026-02

PROJECT SUMMARY For a pregnancy to succeed, the immune environment at the maternal-fetal interface must be precisely regulated to support fetal development. Uterine natural killer (uNK) cells–the most abundant lymphocyte at the maternal- fetal interface–are thought to contribute to various physiological aspects of gestation crucial for fetal development. Their critical role in pregnancy is evidenced by studies linking abnormalities in uNK cells to adverse pregnancy outcomes, particularly in uterine transplant recipients. Our preliminary findings provide the first direct evidence showing that the loss of uNK cells in the pregnant murine uterus significantly reduces litter sizes and increases resorption rates, further underscoring their indispensable role in pregnancy. Despite mounting evidence linking uNK cell dysfunction with adverse pregnancy outcomes, critical knowledge gaps in uNK cell biology persist, particularly regarding the origins and functional specialization of these cells within the uterine microenvironment. Our lab has previously shown that the uNK cell population is heterogeneous, consisting of both tissue-resident NK (trNK) cells and conventional NK (cNK) cells. While the developmental trajectory of cNK cells has been well-established, the developmental origins of uterine trNK cells remain unresolved. Here, we will investigate the origins and differentiation of uterine trNK cells in the virgin and pregnant murine uterus. Our preliminary findings show that both trNK cells and cNK cells in the murine uterus are Eomesodermin-dependent both at steady-state and during pregnancy, suggesting uterine trNK cells derive from the cNK cell lineage. Additionally, our initial studies demonstrate that progenitors in the bone marrow can give rise to uterine trNK cells. Together, our data support the central hypothesis that uterine trNK cells originate from the cNK cell lineage and are derived from either 1) early NK cell precursors in the bone marrow or 2) mature cNK cells in the periphery. Recognizing that these possibilities are not mutually exclusive, we will clarify the origins and developmental kinetics of uterine trNK cells using cutting-edge techniques and novel mouse models, including adoptive transfer studies in a newly engineered reporter mouse as well as advanced whole-mount confocal imaging. We also hypothesize that peripheral cNK cells can differentiate into uterine trNK cells during murine pregnancy and are driven to do so by molecular factors in the pregnant uterus. To explore the potential plasticity of peripheral cNK cells in the pregnant uterus, we will leverage innovative mouse models and state-of-the-art spatial transcriptomics to identify key regulatory signals and cell populations governing this transition. Collectively, this proposal will address key gaps in uNK cell biology by providing critical insights into the origins and differentiation of uNK cells that have the potential to uncover novel therapeutic targets for patients with uNK cell abnormalities, particularly uterine transplant recipients, ultimately improving reproductive health outcomes.

NIH Research Projects · FY 2026 · 2026-02

PROJECT SUMMARY/ABSTRACT Multiple sclerosis (MS) is a chronic disease of the central nervous system (CNS). Susceptibility to MS is significantly associated with Class II alleles, in particular DRB1*15:01:01(DR15). Autoimmune response of the self-reactive CD4 T cells to myelin antigens is a major component of MS immunopathogenesis. Several CNS antigens are thought to be the targets for CD4 T cells and there are numerous viral and bacterial peptides that activate myelin basic protein (MBP)-specific T cells. This suggests that cross-reactivity of CD4 T cells might represent one mechanism whereby infections trigger disease. Myelin antigens are widely used to induce Experimental autoimmune encephalomyelitis (EAE). The actual myelin-reactive CD4 T cells potential cross- reactivity is very poorly characterized. To better define the T cell receptor (TCR) repertoire and its specificity in MS, we applied single cell paired TCR sequencing method, an algorithm to determine convergence of TCRs across individuals, and a robust platform for antigen discovery. Without enriching the T cells for any particular antigen specificity, we performed unbiased TCR sequencing of activated T cells from the blood and cerebrospinal fluid (CSF) of MS patients and healthy controls. In MS patients, we found higher clonal expansion and overlap of CD4 T cells in the blood and CSF, convergence of CD4 TCRs among multiple DR15 MS patients with common sequences/motifs, and cross-reactivity to human adenovirus (HAdV) peptide and with MBP. We generated DR15 tetramer with HAdV-peptide and in a small new cohort of DR15 patients and controls show the existence of HAdV and MBP cross-reactive CD4 T cells in the blood. Moreover, we generated a new HAdV-TCR transgenic mice to test hypothesis of cross-reactivity. Altogether, we show that a convergent CD4 TCR from multiple MS patients react with MBP and a peptide from viral origin and these cells can be detected in the blood of patients. Altogether, our results showed that CD4 T cells in MS cross-react to MBP and a peptide from viral origin. Therefore, our central hypothesis is that cross-reactive myelin specific CD4 TCR repertoire in a subpopulation of MS patients is triggered by adenovirus antigen. We will test this hypothesis with the following aims: Aim 1. To assess the cross-reactive CD4 TCR repertoire between MBP- and HAdV in MS patients and healthy individuals. Aim 2. To mechanistically determine the impact of a HAdV- and MBP-cross-reactive CD4-TCR in mouse model.

NIH Research Projects · FY 2026 · 2026-02

SUMMARY RNA viruses pose a threat to organisms ranging from bacteria to humans, and thus all hosts need to recognize and defend themselves from these foreign invaders. While some principles of RNA virus recognition have been defined, primarily in mammalian systems, there remains a significant gap in knowledge as to how different hosts recognize the presence of an RNA virus. Much less is known about how simple metazoans, such as the model organism C. elegans, recognize virus infection. Studies in C. elegans have led to the discovery of multiple fundamental paradigms in biology, such as identification of caspase cell death pathways and RNA interference (RNAi). The discovery over a decade ago of Orsay virus (ORV), a positive strand RNA virus that is the first and only known natural virus of C. elegans to date, opened the door to virus-host interaction studies in this simple, multicellular invertebrate model system. Defining the host genes required for recognizing an RNA virus and the viral derived molecules that are sensed by C. elegans will provide novel insights into the evolution of self-versus-nonself discrimination. These studies could also lead to identification of novel viral recognition and signaling pathways that are broadly conserved. In mammals a family of cytoplasmic receptors that includes RIG- I and MDA5 senses the presence of intracellular RNA virus derived products. They then signal through additional proteins, such as MAVS, to activate transcription factors, including IRF3, IRF7, and NF-kB to induce an antiviral transcriptional response. C. elegans encodes DRH-1, a dicer related helicase that is homologous to RIG-I and MDA5, that is required for activation of the C. elegans transcriptional response, termed the intracellular pathogen response (IPR), to ORV infection. However, since C. elegans does not possess orthologs of MAVS, IRF3, IRF7 or NF-kB, it is not clear how the sensing of an RNA virus is transduced in C. elegans and what genes are involved. The viral derived molecule(s) that are recognized in C. elegans is also poorly defined. RIG-I binds viral dsRNA with a 5' triphosphate while MDA5 binds longer dsRNAs independent of the 5' terminus. In C. elegans, replication of ORV from a plasmid-based replicon system is sufficient to activate the IPR implicating an ORV replication product as the trigger. In preliminary data, we demonstrated that non-replicating ORV dsRNA produced by the endogenous C. elegans transcription machinery is sufficient to activate the IPR in vivo. This implies that viral dsRNA with 5' cap structure can be a trigger, providing the first evidence that DRH-1 maybe more similar to MDA5 than RIG-1. Thus, studies of DRH-1 may provide novel insights into the evolution and function of MDA5-like sensing proteins. Here, we will (1) define the host factors that recognize viral infection and are responsible for transducing that signal into a transcriptional response, and (2) determine the precise nature and characteristics of the viral ligand(s) that is recognized by C. elegans.

NSF Awards · FY 2026 · 2026-02

Weak or damaged neural pathways have been associated with many disabling neurological and mental conditions, including stroke, neurodegenerative disorders, and mental disorders such as depression. This project, led by researchers at the University of Utah, will use focused ultrasound, delivered through the cranium, for effective neurorehabilitation. The system and approach will enable operators to modulate the connection between any two regions of the human brain, and thus repair neural pathways that have been damaged or weakened. To maximize the impact of the therapies, the team will use seminars and demonstrations to engage human populations in Utah that have traditionally had limited access to cutting-edge neurological treatments. The feedback gathered from these groups will inform the design of the final therapeutic tools, optimizing the usability of the system for all persons regardless of socioeconomic status. There is currently no tool that would enable the noninvasive repair of neural networks in a targeted manner. Transcranial, focused ultrasound has the potential to provide such a tool, but there is no system and no approach for the modulation of neural connectivity. The project team, which consists of an engineer, a psychiatrist, and a magnetic resonance imaging (MRI) scientist, has designed a prototype device that can deliver effective and safe levels of neuromodulatory ultrasound into deep brain targets in humans. This project will develop the prototype into a system that can stimulate two nodes of a neural network simultaneously (Aim 1), thus promoting Hebbian plasticity under an appropriate stimulation-timing protocol (Aim 2). The team will apply this dual stimulation protocol to two key nodes within the cingulate cortex in humans, and quantify the resulting changes in connectivity using functional MRI (fMRI) blood-oxygenation-level-dependent activity (BOLD) and diffusion tractography. Successful completion of the project will result in a tool for the noninvasive repair of neural circuits with the potential to enable many affected individuals to return to work and normal life. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2026 · 2026-02

An award is made to Washington University in St. Louis (WashU) to acquire a two-photon microscope system dedicated to understanding higher cognitive function in the brain. The instrument will meet the needs for several cutting-edge projects of WashU’s researchers that are not possible with currently available instruments. This multiphoton microscope will be the only microscope dedicated to this type of research in Missouri and surrounding states. The instrument will democratize neuroscience, through making the imaging data obtained with the instrument publicly available in open-access repositories, and by allowing external researchers to make use of the instrument in collaboration with scientists at WashU. The instrument will be integrated into educational activities and available for research projects of undergraduate and graduate students. To understand complex cognition, research with primates is critical because key components of human cognition depend on primate-specific brain specializations. The instrument allows researchers to study large populations of brain cells during complex cognition. Through high spatial resolution and cell-type specificity, the instrument offers complementary advantages to existing technologies, and will thereby unlock cutting-edge research aimed at understanding how the circuits in the primate brain perform complex cognition. Additionally, the work with this microscope could lead to commercialization of new discoveries. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2026 · 2026-02

Diabetes is characterized by insufficient insulin, and both type 1 (T1D) and advanced type 2 diabetes (T2D) are treated with insulin. One drawback of insulin-only therapy is the risk of hypoglycemia, and alternatives could improve the quality of life for people with diabetes. Destruction of β-cells in T1D also causes glucagon hypersecretion from the of α-cells. Thus, researchers have targeted hyperglucagonemia as a potential therapy in conjunction with insulin. Using glucagon receptor antagonists (GRA) to reduce the action of excess glucagon restores euglycemia in patients with diabetes and allows them to avoid hypoglycemic episodes. Surprising data from our lab show that diabetic phenotypes can be corrected by subcutaneous transplants of embryonic brown adipose tissue (BAT) without measurable insulin. In T1D animal models with BAT transplants, we measure a rapid and robust reduction of plasma glucagon, followed by progressive reversal of clinical signs of diabetes. While blocking glucagon action with GRAs leads to show-stopping side effects, returning glucagon levels to normal by reducing hypersecretion is not expected to cause similar complications. Our preliminary data show that a large molecular weight fraction (CB-100) of BAT secretions inhibits glucagon release from both mouse and human islets, activates the insulin receptor in vitro, and reverses diabetes phenotypes when injected daily in vivo. This reversal happens within days and lasts many weeks after cessation of the injections. The CB-100 fraction also promotes white adipocyte differentiation and browning, supports healthy BAT, and enhances glucose uptake in adipose tissue, skeletal muscle, and liver. From this fraction, we identified nidogen-2 as a critical secreted protein that reverses hyperglycemia in NOD mice, inhibits glucagon secretion from pancreatic α-cells, and mimics other actions of the entire secreted fraction. Nidogen-2 is a ~150 kDa basement membrane assembly protein that is highly expressed in embryonic tissue and known to be cleaved in vivo. Nidogen-2 secreted from BAT does not exhibit full-length protein, but rather a ladder of nidogen-2 fragments. Based on these data, we hypothesize that a fragment of nidogen-2 targets the α-cell directly and that the same or another fragment activates insulin receptors leading to the reversal of diabetes phenotypes when injected in vivo. To test this hypothesis, we propose two Specific Aims, 1) to determine the fragments of nidogen-2 that inhibit glucagon secretion and activate insulin receptors, and 2) to determine the α-cell target of nidogen-2 and its fragments. This research will establish a novel signaling role for nidogen-2, beyond its traditional classification as an extracellular matrix protein. Our results will also establish a nidogen-2 derived peptide as a novel diabetes treatment that could be used in conjunction with insulin treatment. We expect that a complementary therapy would allow patients to maintain euglycemia at near-normal levels with less insulin dose and less worry about hypoglycemia. The work proposed here will focus on T1D models, but we expect that the results will also be broadly applicable to advanced T2D.

NIH Research Projects · FY 2026 · 2026-02

PROJECT SUMMARY/ABSTRACT Difficulties finding one’s way in novel environments may lead to anxiety and restrictions in daily activities for older adults. There is a substantial literature documenting age-related deficits in spatial navigation, with particular deficits observed in the acquisition, retrieval and use of a cognitive map of the environment relative to route learning. This is particularly problematic as cognitive mapping permits more flexible navigation, such as taking shortcuts. However, the most common design involves learning and retrieval in a single session. However, the typical study design used in this literature involves learning and retrieving a novel virtual environment in a single session, which could lead to an under-estimation of the full navigation capacities of older adults. Thus, we have gained knowledge of the nature of age deficits in cognitive mapping in the early phases of acquisition and retrieval, but we now need an enhanced understanding of the evolution of older adults’ environmental knowledge and flexible navigation behaviors over more extended time frames, which is more similar to our everyday navigation. Our underlying model is that age-related limitations in processing speed as well as in attentional control and associative processes slow acquisition of relevant associative details of an environment thereby limiting flexible navigation; as well as contribute to greater forgetting and impaired retrieval of this knowledge and thus a reduction in flexible navigation after an extended delay. We will test this model in a paradigm that entails study-test trials across 3 days and a 2-week retention interval. During study, participants will take a “tour” of our city art museum. During test, measures of environmental knowledge and flexible navigation will be obtained. Our experiment will be conducted in a real-world environment as this provides a strong foundation for acquisition, flexible navigation and retention for older adults. Furthermore, it is estimated that around 20-30% of healthy older adults have elevated Alzheimer disease (AD) pathology (i.e., preclinical AD). Enhanced understanding of whether there are specific patterns of acquisition or enhanced forgetting over time in preclinical AD would be useful in developing more sensitive cognitive indicators of the earliest disease stages. In addition, the presence of AD pathology is generally not taken into account in much of the healthy aging literature on spatial navigation. Thus, what is considered an “aging” deficit may to some degree reflect the preclinical AD stage. Thus, the proposed research will determine (1) the pattern of age differences in acquisition and use of a cognitive map over multiple learning episodes; (2) the pattern of age differences in long-term retention and use of a cognitive map; and (3) the role of AD pathology in acquisition and retention of a cognitive map. We will secondarily consider the roles of basic cognitive abilities (i.e., processing speed, attentional control and associative learning), visual attention, and sleep quality as potential contributing factors to spatial navigation deficits. Collectively, these aims will provide a strong basis for programmatic research investigating an array of encoding and retention manipulations to enhance cognitive mapping in older adults. Examining the role of AD pathology provides a foundation for additional cognitive tools for distinguishing healthy aging from the very earliest stages of AD.

NIH Research Projects · FY 2026 · 2026-01

Project Summary: Adaptation to environmental stresses is crucial for survival. Our work has shown that phenotypes, such as longevity, induced by stress can be passed on across generations through transgenerational epigenetic inheritance. Transgenerational epigenetic inheritance allows organisms to respond to irregular conditions, alert their naïve descendants that stresses could still be present, and for distant descendants to eventually return to a basal state after several generations without the stress. However, it is still unclear whether organisms adapt on a transgenerational scale after repeated successive generational exposures to the same environmental stress or to successive generational exposures to different stresses. These questions have become increasingly relevant for humans. We exist in an environment where environmental toxins are routinely used in agriculture or manufacturing. When one of these toxins is identified as being harmful, a new toxin is often used to accomplish the same task, and subsequent generations are exposed to this second toxin. Are these successive generational environmental stresses additive in their transgenerational toxic effects? Surprisingly little is understood about how repeated exposures to stress across generations influences complex organismal traits. The goal of this project is to understand how repeated exposures to environmental stresses over successive generations elicits a transgenerational epigenetic adaptation which alters organismal longevity. We previously showed that hypoxia extends C. elegans lifespan in the exposed generation, causes intergenerational reduction in lipids, and a transgenerational reduction in fertility in descendants reared under normoxic conditions [1]. We have found that C. elegans adapt on a generational timescale to repeated generational hypoxia exposure. After two successive generations are exposed to hypoxia, hypoxia-exposed worms are no longer long-lived. After four successive generations are exposed to hypoxia, hypoxia-treated worms no longer lay fewer progeny than normoxic worms. RNA sequencing data across four generations repeatedly exposed to hypoxia identified genes whose expression changes underlie the adaptation of the altered longevity and fertility phenotypes. A directed genetic screen of transgenerational adaptation mediators found that deletion of two argonaute proteins accelerated transgenerational adaptation to repeated generational hypoxia. Conversely, deficiency in the H3K27 trimethyltransferase complex (MES-2 and MES-3) prevented worms from adapting. These preliminary findings have characterized the first instance of transgenerational adaptation to adverse environmental conditions over successive generations and identified molecular pathways important for transgenerational adaptation. However, whether different types of stresses can induce transgenerational adaptation to influence longevity, and how small RNAs and histone methylation can regulate this transgenerational adaptation is still unknown. This work aims to characterize mechanisms of the new phenomena of transgenerational epigenetic adaptation and begin to decipher how these epigenetic cues allow organisms to adapt to repeated stresses across generations.

NIH Research Projects · FY 2026 · 2026-01

Project Summary The overarching goal of this proposal is to define new mechanisms that promote neuronal DNA repair and transcriptional fidelity across lifespan. The number of diagnoses for age-dependent neurodegenerative disease and dementia is projected to more than double by 2050, underscoring our immediate need to understand the cellular and molecular basis of brain aging. A critical aspect of age-dependent cognitive decline is the decreased ability of neurons to adapt to experience-driven changes in neuronal activity. Neuronal activity promotes plasticity, in part, via the de novo induction of cell-type-specific transcriptional programs that mediate learning and memory. However, the induction of such activity-dependent programs drives DNA damage at gene regulatory elements. Neuronal activity is thus a risky endeavor: long-lived neurons must adapt to new cues to facilitate life-long learning and yet maintain a pristine genome. Although accumulating DNA damage is a hallmark of aging and neurodegeneration, our knowledge of the mechanisms that limit damage in post-mitotic neurons is surprisingly limited. How neurons balance transcription and repair, especially in the context of aging neural circuits, is a fundamental unanswered question with major implications for cognitive aging and neurodegenerative disease. Here, we will leverage our lab’s recent discovery that neurons couple activity-dependent transcription to DNA repair through the neuronal chromatin modifier, NPAS4:NuA4. Deletion of NPAS4:NuA4 components in mouse models leads to dysregulated transcriptional responses to activity and increased DNA double-strand breaks across the genome, culminating in drastically reduced organismal longevity. Npas4 expression is reduced in aged neurons, suggesting dysregulation of transcriptional control and genome protection by NPAS4:NuA4 contributes to age-associated neuronal dysfunction. However, the mechanisms by which this protective complex stimulates both transcription and repair remain unclear. We hypothesize that NPAS4:NuA4 promotes genome stability via homology-directed repair factors and enhances transcriptional fidelity during aging via chromatin control of RNA Polymerase II (RNAPII) speed. In Aim 1, we will examine a role for the homology-directed repair factor RAD52 downstream of NPAS4:NuA4 in active neurons and test the consequences of perturbing this pathway on cellular aging phenotypes. In Aim 2, we will assess age-dependent changes to the regulation of activity-dependent transcriptional programs, especially those mediated by NPAS4:NuA4. These mechanistic studies will enhance foundational knowledge of genome control in post- mitotic cells and position us to design strategies that slow or prevent molecular damage in aged and diseased human neurons.

NSF Awards · FY 2026 · 2026-01

Rock deformation is a sub-discipline of Earth science that employs experimental techniques from geology and engineering to measure the strength of rocks. Observations and data from rock deformation are essential to a wide range of research in geoengineering, geologic hazards, geophysics, and planetary geology. However, most institutions do not have active research programs in rock deformation, due to the scale, cost, and technical needs of an experimental rock deformation lab. The Research Opportunities in Rock Deformation (RORD) REU site provides training in experimental rock deformation and generates a robust pipeline of students and industry professionals from all backgrounds. The RORD REU site provides research and mentorship opportunities for undergraduate students in the field of experimental rock deformation. The long-term objective is to expand the pipeline of students pursuing research or industry careers in rock deformation or related fields. Student participants receive training in research methods and professional development topics that provide a stable foundation for graduate school or related career paths. A large team of PIs and senior participants, composed of academic researchers in rock deformation, ensures that students who participate in the program have a deep professional network to support their future endeavors. Students are drawn from the full spectrum of higher education institutions. Strong emphases are placed on recruiting students from smaller colleges and universities that do not have research programs in rock deformation. The REU site includes three integrated sessions: a field session to introduce students to the geological study of deformed rocks, a laboratory session where students conduct experiments on specimens collected during the field session, and a conference session where students have the opportunity to present the results of their research projects. The REU site uses an innovated distributed model, leveraging the combined lab capacity of the PIs and other senior participants to support 10 students per year. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NIH Research Projects · FY 2025 · 2026-01

Project Summary/Abstract Apolipoprotein A1 (ApoA1) is the core protein constituent of high density lipoprotein (HDL). Previous work from our lab has shown that enteric HDL plays a critical role in the neutralization of lipopolysaccharide (LPS) that leaks from the gut. LPS rapidly binds circulating HDL and is neutralized, preventing chronic activation of TLR4 which leads to inflammation and liver fibrosis. However, the mechanism by which HDL neutralizes LPS is not completely understood. The liver derived LPS binding protein (LBP) is known to associate with HDL and play a critical role in the neutralization of HDL bound LPS. To address the mechanism by which LBP bound HDL from the intestine neutralizes LPS, I will determine the protein constituents of the LPS neutralizing complex and measure the capacity of different reconstituted lipoprotein complexes to neutralize LPS. To investigate the mechanism of how HDL neutralizes LPS, I will use proteomics and lipidomics to characterize LBP+ HDL particles. I will also use a novel flow sorting method to detect proteins on intact HDL complexes derived from plasma sources. Using recombinant protein and lipids, I will reconstitute this complex and measure the neutralization of LPS using HEK-BLueTM-4 cell assay. Another outstanding question is how HDL is trafficked from the small intestine into the portal venous circulation prior to reaching the liver. Our previous research suggests that lipidation of ApoA1 by intestinal ABCA1 is necessary for efficient transport of HDL to the portal venous circulation. Knockout of ABCA1 in the intestine led to a dramatic reduction in the circulating levels of HDL in the portal vein and led to greater liver inflammation and fibrosis. One possibility to explain these results is that lipidated HDL binds to scavenger receptor class B type 1 (SRB1) which is known to play a role in the transcytosis of HDL across the endothelial membrane. An alternative hypothesis is that the fenestration of the microvasculature in the villi of the small intestine allows for the passive filtration of HDL into the venous circulation. To distinguish between these two hypotheses, I will use a photoconvertible fluorescent protein, KikumeGR, tagged to ApoA1 to measure trafficking of HDL from the small intestine to the portal vein. To investigate the role of SRB1 on trafficking of enteric HDL to the portal vein, I will cross SRB1fl/fl mice to Cdh5α-CreERT2 mice, which will remove SRB1 from vascular endothelial cells. To test whether HDL from the small intestine egresses through fenestrae in the microvasculature, I will use PE-conjugated anti-PLVAP antibody (PLVAP forms a diaphragm regulating solute passage through fenestrae) which will block passage of solutes through fenestrae. To measure HDL trafficking from the small intestine to the portal vein, I will transduce mice using AAV expressing KikGR-ApoA1, which can be photoconverted from green to red fluorescence in the small intestine using 405nm light, which can then be detected in the portal vein. Together these experiments will help to mechanistically define the trafficking of HDL to the portal vein and the neutralization of LPS to restrain liver inflammation and injury.

NIH Research Projects · FY 2026 · 2026-01

PROJECT SUMMARY Neurological disorders, such as Alzheimer disease (AD) and Parkinson disease, chronically affect critical regions of the deep brain, including the hippocampus and substantia nigra. Long-term monitoring of disease progression and responses to interventions in mouse models, which have well-characterized timelines and are equipped with a variety of manipulation tools, holds significant promise for uncovering new mechanisms and therapeutic targets. However, existing technologies, including optical microscopy and microendoscopy, face major limitations related to invasiveness, biocompatibility, and long-term accessibility when studying the deep brain in mouse models. To address these challenges, we propose the development of a fiber implant for longitudinal, multifunctional, deep- brain interfacing in awake mice. This implant, with a small diameter and constructed from biocompatible materials, has the potential to address the limitations of existing approaches and enable longitudinal deep-brain monitoring with minimal impact on brain function over an extended period (>9 months). Our innovative design will integrate photoacoustic microscopy (PAM) and two-photon microscopy (TPM) for simultaneous functional and molecular imaging, along with microelectrode-based neural stimulation and microfluidics-based drug delivery for targeted brain manipulation: creating a comprehensive bidirectional interface. We will use this technology to test a set of hypotheses derived from our prior study: (1) hippocampal-dependent memory loss is a consequence of impaired blood oxygen supply in the hippocampus; (2) reduced blood oxygen supply is secondary to vascular amyloid pathology; and (3) restoring blood oxygen supply rescues memory loss. Testing these hypotheses relies on the advanced capabilities of the bidirectional fiber implant. Specifically, longitudinal monitoring is essential to unravel the temporal relationship between amyloid pathology, vascular dysfunction, and memory loss in AD. Access to the hippocampus is critical, as it is one of the first regions affected in AD and is directly linked to memory loss. The simultaneous use of PAM and TPM will allow for comprehensive assessments of both blood oxygen supply and vascular amyloid pathology. Microelectrode-based stimulation will facilitate the evaluation of the impairment in vascular functional hyperemia, while microfluidic drug delivery will enable localized application of a vasodilator to determine whether blood oxygen supply can be restored in the face of amyloid pathology, and, if so, how such restoration affects memory function in AD. Successful completion of this project will enhance our capacity for longitudinal studies of the deep brain in animal models and provide new insights into the mechanisms underlying memory impairment in AD, with significant implications for developing therapeutic strategies targeting vascular dysfunction and oxygen supply restoration.

NIH Research Projects · FY 2025 · 2025-12

PROJECT SUMMARY Alzheimer’s disease (AD), the leading cause of dementia worldwide, affects millions of individuals and presents a major public health challenge. Microglia, the resident macrophages in the brain, adopt a range of activation states, displaying functional heterogeneity that spans from neuroprotective to neuroinflammatory roles in AD. Deciphering the heterogeneity of brain macrophage is essential for understanding the immune mechanisms driving AD pathogenesis. A key factor shaping macrophage heterogeneity is their ontogeny, as the developmental origin of macrophages plays a critical role in defining their identity and function. Microglia were thought to originate solely from yolk sac (YS) progenitors and maintain themselves through self-renewal without contributions from peripheral cells. However, recent studies suggest that monocyte-derived macrophages (MDMs) can infiltrate the brain under certain conditions, such as in mouse models of amyloidosis, adding to the heterogeneity of brain macrophages. However, the role of brain-engrafted MDMs in AD pathogenesis remains unclear. This research proposal aims to elucidate the contributions of brain-engrafted MDMs to AD pathology in both mouse models and human AD. In the F99 phase, I will investigate the effects of brain-engrafted MDMs on amyloid plaque deposition in the 5xFAD mouse model. Specifically, I will assess how promoting (Aim 1.1) or depleting (Aim 1.2) brain-engrafted MDMs affects amyloid accumulation and cognitive decline, providing insights into the functional role of MDMs in AD progression in mice. In the K00 phase, I will focus on studying human AD, utilizing somatic mosaicism as a method to trace brain-engrafted MDMs and investigate the role of AD- associated somatic mutations in human iPSC-derived microglia. A recent study revealed an enrichment of somatic mutations in AD patient microglia, suggesting that mutation-driven clonal expansion may contribute to AD pathology. However, it remains unclear whether MDMs infiltrating the human AD brain contribute to this mutational burden. I will analyze microglia and paired blood monocytes from AD patients and controls to determine whether peripherally-derived MDMs contribute to the increased somatic mutations observed in AD microglia. Additionally, I will introduce the top AD-associated somatic mutations into iPSC-derived microglia to assess their functional impacts on microglia. The successful completion of these studies will deepen our understanding of brain macrophage heterogeneity, particularly the distinct roles of MDMs versus YS-derived microglia in AD. This work holds the potential to uncover novel therapeutic targets aimed at slowing disease progression and improving outcomes for AD patients.

NSF Awards · FY 2025 · 2025-10

The growing use of artificial intelligence in healthcare and medical research has created a difficult challenge: researchers need to share their computer models to advance scientific discovery, but these models can reveal private information about the patients whose data was used to create them. Organizations are often reluctant to share their computer models because of privacy risks, even though withholding these models prevents broader societal benefits from medical research. This creates a barrier to scientific collaboration that could otherwise lead to better treatments, improved public health outcomes, and medical breakthroughs. This project addresses this challenge by developing methods that allow organizations to safely share models trained on sensitive patient data without compromising individual privacy. Proposed research will result in new techniques for auditing models, certifying their privacy guarantees, and providing actionable tools to fix any identified issues. This work serves the national interest by advancing medical research and scientific discovery, enhancing national health and prosperity through improved healthcare technologies, supporting American competitiveness in artificial intelligence innovation, and enabling secure collaboration while protecting personal privacy rights. This project develops an end-to-end framework for privacy-preserving sharing of machine learning models trained on sensitive data. Despite growing interest in sharing models rather than raw data, machine learning models remain vulnerable to privacy attacks, such as membership inference attacks, which can reveal whether an individual's data was used during training. The research activities include four technical components. First, the project will evaluate the privacy properties of shared models by subjecting them to existing and newly tailored privacy attacks, establishing a foundational understanding of their vulnerabilities. Second, the team will develop formal privacy guarantees using methods like differential privacy and establish privacy-utility tradeoffs, creating privacy certificates for machine learning models that may include legal and usage constraints. Third, the project will design explainable auditing tools and privacy patching mechanisms such as machine unlearning to help developers mitigate risks without compromising model utility. Fourth, the research will build user-friendly tools to deploy these methods, focusing on real-world applicability in healthcare and biomedical research. The project will introduce a novel privacy-risk scoring system, enabling developers and regulators to assess the privacy risks associated with a given model. Unlike existing point solutions, this comprehensive framework integrates auditing, certification, and remediation into a unified system. Results will be disseminated through tools, publications, and educational modules to support broad adoption and training. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-10

This research aims to develop statistical tools to improve the reliability of artificial intelligence (AI) that is widely used in real-world systems such as automated decision-making, financial forecasting, and neuroscience research. Modern AI often relies on efficient machine learning algorithms to process large-scale, sequentially arriving datasets. While these algorithms are powerful, understanding their behavior and measuring their uncertainty remains a major scientific challenge. To bridge this gap, the investigators will focus on establishing mathematically rigorous methods for uncertainty quantification to build trustworthy AI. Applications will include enhancing theoretical guarantees and interpretability of neural networks, providing robust estimation and inference for econometric and biomedical studies, and detecting real-time change-points in high-dimensional time series data. The projects will promote the progress of science through open-source software and graduate education, and will support the national interest by contributing to reliable, data-driven decision-making in fields important to economic resilience, public health and national security. This research will provide a comprehensive theoretical framework for online statistical inference in machine learning, focusing on constant learning-rate stochastic gradient descent (SGD) algorithms. It addresses fundamental challenges such as non-stationarity caused by arbitrarily fixed initialization and complex dependency structures arising in recursive estimation. The investigators will derive the limiting distributions of SGD-type estimators and construct confidence regions with guaranteed asymptotic coverage. Specific efforts will include (1) establishing Gaussian approximations for high-dimensional dropout regularization, (2) deriving limiting distributions for SGD under non-smooth quantile loss functions using characteristic function techniques, and (3) developing online inference procedures for quantile change-point detection in high-dimensional time series using a novel Bahadur representation. These methods will be supported by numerical experiments and implemented in publicly available software. The results shall provide foundational advances for statistical inference in modern machine learning, bridging theoretical developments with practical applications in dynamic, high-dimensional environments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Awards · FY 2025 · 2025-10

Federal district court and circuit court judges serve as the face of law and justice for many citizens in the United States. Because of their importance, there is interest among scholars, students, the media, the public, and policymakers to understand who these judges are and what decisions they make as judges. In aid of improving societal understanding and perceptions of the federal judiciary, this project will develop, make public, and use the Federal Judicial Database. When complete, these data will include detailed background information on all federal district and circuit judges serving from 1789 to the present selected to their positions through presidential nomination and Senate confirmation. The data will be presented to the public through an interactive website designed to facilitate broad user accessibility. The Federal Judicial Database will provide detailed biographical, attribute, and background data on the thousands of federal district and circuit judges selected into their positions under Article III of the U.S. Constitution. Compiled from federal judges’ nomination documentation presented to the Senate, along with other congressional, judicial, and biographical sources and updates of prior data collection efforts, the Federal Judicial Database will include extensive data on each federal judge, including information on nomination, confirmation, departure, demographics, biography, prior judicial and legal employment, and pre-appointment writings and speeches. The Investigator will design, develop, and produce a public website for the consumption and distribution of the Federal Judicial Database. The website design will accommodate information about the research project and data (project background, codebooks, data sources, citations, and publications), interactive search and display abilities, and access to pre-prepared datasets for download. The public web interface will be designed to make the data attractive and accessible to a broad set of non-scholar users and will highlight both individual judge profiles and cross-judge profiles based on interactive searches conducted by the user. Using the newly collected Federal Judicial Database, the Investigator will undertake two research applications. The first of these is an examination of whether district court judges with different legal backgrounds manage civil cases in distinct ways (such as in holding settlement conferences) and whether they vary in their successful encouragement of case settlements. Relevant judge factors from the Federal Judicial Database will include, for example, a district judge’s prior experience as a civil litigator, arbitrator, or mediator and the number of cases she took to trial during her legal career. The second research project using the Federal Judicial Database will be a study of how much variation exists in federal courts of appeals judge professional backgrounds (for example, prior service as prosecutors, public defenders, and corporate lawyers) across time and whether these different judicial backgrounds affect decision making on the federal bench. Beyond these two research projects, the Federal Judicial Database will permit rigorous inquiry into the patterns and effects of judges’ writing, trial experience, education and training, group membership, and more. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.