Hypoxic-ischemic encephalopathy (HIE), a brain injury that occurs when infants do not get enough blood flow or oxygen to the brain, is a leading cause of neonatal mortality and morbidity worldwide. Therapeutic hypothermia (TH) is the current standard of care for newborns with HIE, but TH is only available in high-resource settings and only provides partial neuroprotection. Thus, the search for additional neuroprotective treatments is critical. The ferret provides an excellent model for investigating novel treatments for HIE as, unlike rodents, it has a gyrified brain and gray-to-white matter ratio that is like humans. Previous research has shown that the ferret brain is resilient to brain injury, requiring additional hypoxia periods and increased pro-inflammatory stimuli to create the same injury as in rats. However, no previous studies have evaluated why the ferret brain is so resilient. This study will use live rat and ferret organotypic brain slices to investigate this resiliency. Whole hemisphere brain slices from postnatal day (P)10-12 rats and P21-23 ferret, equivalent to term gestation in humans, will be collected. The slices will be randomized to oxygen-glucose deprivation (OGD) injury, to mimic HIE, or control groups. OGD slices will be in 0% oxygen for 2 hours, resulting in partial, but not total, cell death. Subsequent analyses will assess transcriptomics using NanoString nCounter technology, which provides a neuropathology panel of 770 genes, as well as cell-specific regional death in brain regions affected by HIE: hippocampus, cortex, corpus callosum, subcortical white matter, basal ganglia, and thalamus. Preliminary results show that genes such as UCHL1 and TLR4, both associated with injury, are upregulated in ferret OGD slices, but rat data are currently incomplete. Identifying the pathways associated with the resiliency of the ferret brain to injury at the transcriptome level could inform future therapies to treat infants at risk for HIE.

Pain comes in different types and sizes and affects each of us in different ways. How we perceive pain-related stimuli is highly dependent on our internal brain states. While our pain perception changes in different brain states, little is known about how the brain mediates these changes. Neuromodulators (neurotransmitters, neuropeptides, and related G-protein coupled receptors) are used to elicit a change in neuronal activity, and changes in pain perception often result in behavioral change. Thus, neuromodulation is likely a crucial part in understanding behavioral shifts during brain states shifts. We focused on the locus coeruleus (LC), which uses norepinephrine (NE) as its primary neurotransmitter. NE is highly involved in the modulation of arousal, attention, stress, and pain perception. LC also has broad projections across multiple brain regions, because of this and LC’s use of NE, we hypothesize that LC acts as a hub, modifying behavior depending on the internal brain state. We modeled two representative brain states by exposing animals to either chronic social isolation or exercise via a running wheel. We focused on pain events, as their induced high arousal state is linked with increased NE levels. The pain-related assays include the Von Frey test, which measures mechanical sensitivity, formalin injection, which induces an inflammatory pain response, then the tail flick and hot plate for thermal nociception. LC neuronal activity was measured via cFos expression shown by immunohistochemistry, while NE signaling was measured using fiber photometry with genetically encoded sensors. By using pain-related behavioral assays, LC neuronal activity during pain response, and noradrenergic signaling in our two brain state models, we aim to learn how LC-NE systems affect pain perception during these brain state changes. We believe these methods will help elucidate how change in neuromodulatory signaling levels in the LC mediate behavioral change during brain states shifts.

 Dravet syndrome (DS) is a genetic form of epilepsy characterized by febrile seizures in infancy, developmental delays, and sudden unexpected death in epilepsy (SUDEP) as a result of being drug-resistant. Finding new and innovative treatments is essential to reducing the risk of SUDEP and other symptoms in DS patients. Using a mouse model of DS (SCN1a+/- mouse), I showed that a machine learning-based detection algorithm could be used to detect interictal spikes (IS), which are abnormal neuronal discharges typical of epilepsy. The goal of the current experiment is to see whether the prior findings can be generalized to a larger dataset and whether the detected IS can be used to predict seizures before their onset. Data is collected by implanting two electrocorticographic electrodes and one electromyography electrode, as well as a wireless body temperature sensor in DS mice. Ambient temperature is controlled so that the animal’s core body temperature is initially maintained at 37°C and then is gradually increased by 0.5 °C every 2 min until a seizure is observed or the core body temperature reaches 42.5 °C. A machine learning model previously trained using manually scored data from the de la Iglesia lab is used to autonomously detect IS in the collected data. My results so far showed a moderate yet significant positive correlation between ambient temperature increases and IS frequency and points to a positive correlation between IS frequency and seizure onset. However, these results did not include the continuous recordings of body temperature. In the current experiment, I test if these correlations hold using a larger sample size and including continuously recorded body temperature, which may have more predictive power than ambient temperature. Our long-term plan is to design a closed-loop experiment that uses the algorithm to predict and stop seizures before their onset.

An estimated 5-8% of the American population have cerebral aneurysms, showing higher rates of development in patients with common risk factors like hypertension, smoking, and family history of cerebral aneurysms (CA). This study focuses on understanding the causes of aneurysmal subarachnoid hemorrhage (aSAH), where a CA ruptures, resulting in bleeding in the brain. Endovascular coiling is a minimally invasive surgical treatment method for aSAH. Unfortunately, up to 30% of endovascular coiling treatments are unsuccessful, leading to aneurysm recurrence, growth, or rupture. The risk of these outcomes can be predicted using Computational Fluid Dynamics (CFD), a tool that quantifies the hemodynamic environment by solving the equations of motion for a fluid. The CFD simulations calculate factors significant in predicting the effectiveness of coiling treatment including flow rate, wall shear stress, and pulsatility. In this project we have studied the effect of using patient-specific blood viscosity values (the resistance of the blood to fluid flow), that have typically been standardized for all patients in CFD simulations. We have analyzed the effect of using patient-specific blood viscosity on pre-treatment patient-specific computational fluid dynamics simulations of endovascularly-coiled cerebral aneurysms. Preliminary results show that there is an expected improvement in CFD simulation predictive power of treatment effectiveness when patient-specific blood viscosity values are used. We hope to improve the predictive power of CFD simulations regarding the treatment outcome of aneurysm coiling, allowing us to better predict aneurysm recurrence, and eventually guide treatment outcomes.

Previous studies have indicated that the neurohypophyseal hormone, oxytocin (OXT), reduces body weight in high fat diet (HFD)-induced obese (DIO) rodents, nonhuman primates and obese humans through reductions in food intake and increases in energy expenditure. Findings from recent pre-clinical and clinical data indicate that the long-acting glucagon-like peptide 1-receptor (GLP-1R) agonist, semaglutide, reduces body weight, in part, through reductions of food intake. Based on these findings, we hypothesized that the combined treatment of OXT and semaglutide would produce an additive effect to evoke weight loss in DIO rats. To test this hypothesis, rats were fed a HFD for approximately 5 months prior to being implanted subcutaneously with a 28-day minipump that infused OXT (50 nmol/day) or vehicle (VEH; saline). Data from animals that received escalating doses of the incretin hormone, glucagon-like peptide-1 (GLP-1) and semaglutide was analyzed over the 32-day infusion period. Daily energy intake and body weight were tracked for 32 days. Body composition was assessed immediately prior to onset of treatment and near the end of the treatment period using Quantitative Magnetic Resonance (QMR; EchoMRI 4-in-1). OXT (50 nmol/day) alone and semaglutide (3 nmol/kg) alone reduced body weight by approximately 9.9±1.1% (0.05<P<0.1) and 4.3±1.4% (P<0.05), respectively (N=6-7/group), relative to pre-infusion baseline, while GLP-1 at either dose (1 or 3 nmol/kg, SC) was ineffective (P=NS). However, the combination of OXT and semaglutide (3 nmol/kg) produced a more pronounced reduction in body weight (14.1±1.4%; P<0.05) compared with either treatment alone (N=6-7/group). These effects were associated with reduced energy intake, adiposity and adipocyte size (P<0.05). Together, these findings support the hypothesis that the combination of OXT and the GLP-1R agonist, semaglutide, produces an additive effect to evoke weight loss and body adiposity in DIO rats by reducing energy intake. Our findings will create a preclinical infrastructure upon which future studies are undertaken to address whether oxytocin may be used as an adjunct with other treatment approaches, including low doses of the FDA-approved GLP-1R agonist, semaglutide, to evoke weight loss in humans with obesity.

Many with clinically diagnosed Alzheimer’s disease (AD) dementia during life show comorbid neuropathologies at autopsy. We sought to develop and evaluate a composite brain pathology score (BPS) in the Adult Changes in Thought study. We derived BPS using nine standard neuropathological indicators such as Thal phase, Braak stage, and CERAD score, using confirmatory factor analyses. We compared BPS to the AD Neuropathologic Change (ADNC) score, which quantifies neuropathologic changes that underlie AD. We performed non-nested modeling to compare BPS and ADNC scores’ associations using the last cognitive score (memory, executive function, language, and visuospatial ability) prior to death, adjusting for age at death and sex. Non-nested models were compared using Adjusted-R², Davidson-MacKinnon J-test, and Cox–Pesaran tests. We focused on people whose last cognitive data were≤2 years prior to death. We compared ADNC and BPS’ associations with last known clinical diagnosis (no dementia vs. any dementia), adjusting for age at death and sex, by considering area under receiver operator characteristic (ROC) curves. Sample size was 886 with mean age of death of 89 and 57% female. A bifactor model fit best, with residual correlations for Thal phase and CERAD score, and for LATE stage and presence of hippocampal sclerosis. The BPS explained more variance for each cognitive score and was superior and statistically significant compared to the ADNC score in non-nested model comparisons. The BPS was more strongly associated (Odds Ratio=3.2) with dementia diagnosis than the ADNC score (Odds Ratio=2.0). The area under the ADNC and BPS’ ROC curves were 0.74 and 0.78, respectively. The difference between the areas was statistically significant (p-value ≤ 0.0001). We demonstrate an approach to developing a composite BPS which incorporates multiple forms of pathology. Future efforts will focus on co-calibrating and harmonizing BPS in other autopsy cohorts.