Migraine is the third most disabling neurological condition, characterized by severe headache with associated features of nausea, cognitive changes, and sleep disturbances. Current therapies have limited efficacy and frequent side effects. Thus, there is an urgent need to identify novel therapeutic targets. Adenosine is a ubiquitous signaling molecule involved in pain processing and migraine pathophysiology. Here, we utilize AI-enhanced approaches to test the hypothesis that adenosine receptor antagonism reduces pain responses in the acute nitroglycerin (NTG) model of migraine. C57BL/6J male mice were exposed to an I.P. injection of NTG (10mg/kg, a well-established migraine model) in combination with a vehicle control, caffeine (30mg/kg, a non-selective adenosine receptor antagonist), or istradefylline (1mg/kg, a selective A2A adenosine receptor antagonist). Pain responses were captured on video and measured using the mouse grimace scale (MGS) through PainFace AI. Pre- and post-treatment scores were compared using Two-Way Anova (Pre/Post x Drug/Vehicle exposure). NTG resulted in expected increases in MGS, indicative of increased pain, while caffeine and istradefylline mitigated NTG-induced increases in MGS (Δ Pre/Post: NTG/control vs NTG/caffeine, +2.01 vs +0.85, P<0.05, N=6-10; Δ Pre/Post: NTG/control vs NTG/istradefylline, +1.67 vs +0.67, P=0.147, N=4-8). These preclinical data suggest additional studies are needed to determine whether istradefylline or other adenosine-targeting therapies can modulate migraine pain.

Fragile X Syndrome (FXS) is a neurodevelopmental disorder characterized by intellectual disability (ID), autistic traits, and atypical sensory processing. FXS is caused by transcriptional silencing of the Fmr1 gene, leading to loss of the RNA-binding protein fragile X messenger ribonucleoprotein 1 (FMRP). There are currently no approved treatments for FXS, but as a single-gene disorder, gene therapy to restore FMRP expression in the adult brain has recently become viable. Previous studies have revealed a drastic reduction in the density and activity of parvalbumin-expressing GABAergic inhibitory interneurons (PV-INs) in the neocortex of Fmr1 knockout (KO) mice and in post-mortem human tissue from FXS cases. PV-INs express FMRP early during cortical development and their reduced density and hypoactivity are amongst the earliest differences reported in the brain of Fmr1 KO mice. Thus, we hypothesize that viral mediated-gene therapy to re-express FMRP into PV-INs might restore PV density and alleviate well-characterized behavioral phenotypes in the Fmr1 KO mice. We successfully engineered a novel recombinant adeno-associated virus (rAAV) to conditionally express the Fmr1 gene in specific cell populations in adult mice. Currently, we are evaluating whether the re-expression of FMRP at postnatal day 1 (P1) in FMR1 KO mice can restore PV density. We will subsequently evaluate the impact of reexpression on mouse behavior.

While current memory research relies on static, isolated stimuli, the neural mechanisms underlying memory for continuous, real-world experiences remain little understood. This study utilizes a unique dataset of single-unit neuronal firing recorded from patients with intracranial electrodes during the viewing of naturalistic, salient cinematic content (the television series 24). In particular, we study the neurophysiological Subsequent Memory Effect (SME), where we analyze modulation of each neuron (its firing rate) recorded during an episode and test if it relates to subsequent correct memory of clips from the episode. Additionally, we characterize how specific clip content influences neuronal responses. Our analysis aims to determine: 1) Whether the SME relates to increases or decreases in firing rates during memory encoding; 2) The differential impact of traumatic versus neutral scenes on the SME; 3) The role of clip saliency in modulating the SME; 4) The anatomical localization of these firing patterns to identify specific cortical or subcortical regions driving naturalistic memory encoding. By mapping single-cell activity to complex, affective narratives, this work provides insight into how the human brain prioritizes and stores memories of intense real-world events.

Early-life traumatic brain injury (TBI) and seizure activity may differentially affect cognitive performance and contribute to long-term neurological deficits. This study compares two animal models to assess these effects on cognition: lateral fluid percussion injury (LFPI) to induce TBI, and rapid kindling (RK) as a model of epileptogenesis. Juvenile rats at postnatal day (PND) 20 were assigned to Sham (control), LFPI, or RK groups, then assessed 3 months later using the Barnes Maze. Learning was evaluated using latency (time to escape) and success proportion (trials, out of 3 per day, reaching the correct hole under 3 min). While initial performance (days 1–2) was similar across groups, performance diverged at later time points, with reduced success and increased latency in injury groups, driven by RK. By day 4, the RK group exhibited longer latencies and a significantly lower proportion of successful animals compared to Sham (p=0.0391). Slope analyses of escape latency across days showed a significant learning rate in the Sham group (p=0.0003), while RK (p=0.0685) and LFPI (p=0.1251) did not improve significantly. RK also demonstrated a significant reduction in learning rate compared to Sham (p=0.0195). These findings suggest that, despite similar initial performance, early neural disruption does not impair overall learning capacity but instead alters the efficiency and trajectory of learning, as reflected by reduced success at later time points and slower improvement in latency, with more pronounced deficits in RK animals.

Low-dose naltrexone (LDN) is increasingly used off-label to treat chronic pain, providing therapeutic benefits despite being an opioid receptor antagonist. Previous findings demonstrated that stressed male rats showed conditioned place preference (CPP) to LDN, implying that LDN is rewarding after experiencing a traumatic stressor. We hypothesize that stressed mice will spend more time in the LDN-paired context, while non-stressed mice will have no preference for either side. First, we conducted CPP, and then, mice received either 10 footshocks or none for 1 hr in a footshock box. On Days 3-4, conditioning training was done in a CPP chamber, with one context paired with saline and the other with naltrexone (0.1, 1, or 10 mg/kg, injected subcutaneously), respectively, on separate days. On Day 5, the mice had free access to the CPP chamber to determine their preferred side. A two-way ANOVA with dose x treatment as independent variables was used for analysis. We found that stressed male mice that received LDN 2 days after the traumatic stressor spent more time in the LDN-paired context than non-stressed mice. We did not find similar results in females, indicating that LDN is rewarding in stressed males and not in stressed females, suggesting a potential sex-dependent divergence in behavioral response. Our future directions will investigate the mechanism by which LDN exerts its therapeutic effects. Overall, this slim therapeutic range can be refined to translationally address various conditions implicated in anxiety and stress.