Abstracts

Problem/Objective: Modelling human brain cells within functional, oriented circuits is critical for understanding neurological disorders and developing targeted therapeutics. However, conventional in vitro platforms rely on monocultures or rigid, enclosed microfluidics (e.g., PDMS devices) that limit direct physical access, restrict fluid reconfiguration, and complicate downstream analytical interrogation. To address these challenges, we demonstrate a biocompatible, open microfluidic platform that utilizes fluid walls to compartmentalize discrete neuronal populations, guide axonal growth, and interface directly with functional electrophysiological readouts.

Methods: Fluid-walled microenvironments were fabricated directly within standard polystyrene culture dishes and off-the-shelf micro-electrode array (MEA) plates. A thin layer of aqueous cell culture media overlaid with an immiscible, bio-inert fluorocarbon (FC40) was reshaped on-demand via micro-jetting. Interfacial dumbbell structures comprising two chambers linked by narrow micro-conduits were generated. Regulating localized chamber volumes established controlled Laplace pressure gradients across connecting conduits, enabling hydrodynamic cell trapping and physical segregation of distinct neuronal subtypes during seeding. Human induced pluripotent stem cell (hiPSC)-derived cortical, striatal, and dopaminergic neurons were maintained in long-term culture.

Results: The platform enabled sustained long-term culture of human iPSC-derived neurons over months. Controlling intrinsic chamber pressures facilitated robust compartmentalization of cortical somas and dendrites from extending axons, directing unidirectional axonal projections toward target striatal and dopaminergic populations across the micro-conduit. Reconfiguring the fluid walls mid-assay permitted site-specific targeted injury (fluid axotomy), cell/supernatant harvesting via standard pipetting, and downstream monitoring of repair dynamics. Furthermore, fabricating these fluid-walled geometries within commercial MEA wells aligned axon conduits directly over planar microelectrodes, enabling high-resolution functional recording of directional neural signaling between isolated subpopulations.

Conclusions: Fluid-walled microfluidics provides a versatile, lid-free approach for engineering complex human neuronal microcircuits with complete spatial and temporal control. By combining selective somatic isolation, guided axonal connectivity, open pipetting access, and direct MEA integration, this strategy establishes a robust platform for disease modeling, neuroregenerative research, and translational drug discovery.

References:
1. Walsh E.J. et al. (2017) Nat Comms 8, 816
2. Soitu C. et al. (2020) Adv Sci 7, 2001854
3. Nebuloni F. et al. (2024) Lab Chip 24, 3252–3264

Problem/Objective:
In-vivo metabolic and perfusion imaging tools have been used to study neuromodulation and clinical correlates in neurodegenerative and neuropsychiatric disorders by deep brain stimulation (DBS) and peripheral brain stimulation. Previous studies with FDG PET have established a Parkinson’s disease (PD)-related metabolic covariance pattern (PDRP) whose expression can track disease severity, progression, and therapeutic responses in individual subjects. PDRP can be assessed or generated by perfusion brain images from nuclear medicine (PET/SPECT) and arterial spin label MRI given tight neurovascular association. In this study, we compared the modulation of PDRPs from both FDG and O15-water PET induced by subthalamic nucleus (STN) DBS.
Methods:
We used the PDRP identified with FDG metabolic images in 33 PD patients (57±8 y) and 33 healthy controls (55±13 y) and another PDRP with O15-water perfusion images in 19 PD patients (59.5±7.5 y) and 14 healthy controls (51.8±13.8 y). Subject scores of each PDRP were measured in 12 advanced PD patients (59.3±11.2 y; baseline UPDRS motor score 30.0±11.2) scanned with FDG PET OFF and ON STN DBS. All analyses used in-house SSMPCA toolbox in public domain (https://feinsteinneuroscience.org/imaging-software).
Results:
Both PDRPs showed high similarity based on region-loading correlation over the whole brain (R=0.83; P<0.0001) and were characterized by increased activity in the sensorimotor cortex, thalamus, pons and cerebellum, and decreased activity in the premotor areas and occipito-parietal cortices. Subject scores of both PDRPs were elevated (P<0.0001) in the PD patients compared to the healthy controls. STN DBS improved UPDRS motor score by 32.4% (P<0.001) and suppressed subject scores of both PDRPs (P<0.05) in the treated PD patients. The subject scores of both PDRPs correlated strongly in OFF, ON and ON-OFF images (R=0.911, 0.918 and 0.731; P<0.01).
Conclusions:
Both metabolic and perfusion images can be used to identify disease-related patterns and to differentiate patients from healthy controls at individual level. Both topographies and subject scores are comparable to those from perfusion MRI and can potentially assess neuromodulation and clinical correlations by novel therapies of bioelectronic medicine in a wide variety of diseases. This platform will have broad applications in animal models and human subjects given the increasing availability of simpler and non-invasive perfusion MRI.

Objective:
Clinical assessment of tremor following MR-guided focused ultrasound (MRgFUS) thalamotomy relies on subjective rating scales such as the Clinical Rating Scale for Tremor, which provide limited insight into activity-specific treatment response and cannot distinguish translational from rotational tremor components. Wearable inertial measurement units (IMUs) may enable objective, multi-dimensional tremor characterization across motor tasks. We aimed to quantify activity-specific and modality-specific tremor changes following MRgFUS thalamotomy using a wearable 6-axis IMU in essential tremor (ET) patients.

Methods:
Ten ET patients undergoing MRgFUS thalamotomy at North Shore University Hospital (Manhasset, NY) performed four standardized motor tasks (rest, arm extension, arm flexion at 90 degrees, and a functional drinking task), each for one minute, immediately before and after the procedure. A wrist-worn Shimmer3 IMU recorded tri-axial acceleration and angular velocity at 512 Hz. Tremor was quantified as spectral power in the 4–12 Hz band from 5-second windowed segments per task. Per-patient percentage change was summarized across the cohort.

Results:
Tremor power decreased consistently during postural and functional tasks, with all 10 patients showing improvement during arm flexion (accelerometer: median reduction 88.1%, range 25.8–99.9%) and the drinking task (median 89.4%, range 25.0–99.8%). Nine of 10 patients with measurable extension tremor improved post-lesion (accelerometer: median reduction 91.0%); the remaining patient had no baseline extension tremor. Rest tremor metrics were highly variable, consistent with ET postural-kinetic phenotype. Gyroscope-derived tremor power showed the most consistent treatment response during the functional drinking task (-85.1% ± 14.3% vs. -80.3% ± 22.4% for accelerometer), suggesting rotational tremor may be particularly responsive to thalamic lesioning. The dominant tremor frequency (~5–6 Hz) was preserved post-treatment despite substantial amplitude reduction (peak spectral power reduced by 83–86% across arm flexion and drinking tasks), indicating oscillatory amplitude suppression without altering the underlying tremor frequency.

Conclusions:
Wearable 6-axis IMU provides objective, activity-specific tremor quantification that reveals differential treatment responses across tasks and sensor modalities not captured by conventional clinical scales, complementing patient assessments. The functional drinking task yielded the most consistent metrics across both sensor modalities, offering a novel way to track tremor improvement that correlates with quality of life.

Purkinje cells (PC) of the cerebellar cortex are among the fastest spiking cells in the brain, capable of entrainment at frequencies reaching hundreds of Hertz by exogenous electric fields such as those induced by transcranial AC stimulation (tACS). Entrainment of PCs can also transsynaptically modulate their downstream targets in the cerebellar nuclei (CN). However, previous studies in anesthetized models indicate that PC/CN modulation requires significantly higher currents at frequencies below ~10 Hz compared to higher-frequency stimulation. This would increase the required charge injection capacity or the surface area of the electrodes to be used for modulating the local field potentials (LFPs) at these frequencies. To mitigate these charge-injection constraints, amplitude-modulated (AM) sinusoidal waveforms utilizing a high-frequency carrier (e.g. 100-200 Hz) may be employed to induce low-frequency oscillations at the frequency of the AM envelope. This approach maintains the low-charge injection requirements offered by high-frequency stimulation, where the fast-switching direction prevents charge accumulation, while the non-linear response of the PC/CN connection allows them to “decode” the low-frequency envelope.
AM stimulation was tested in a number of clinical trials of tACS in the cerebellum and elsewhere without the knowledge of underlying mechanism. In this study, we demonstrate the underlying mechanism of AM stimulation with single cell recordings in the cerebellum. We compared the efficacy of AM waveform against conventional low-frequency AC stimulation at frequencies below 20 Hz. Our results, obtained from both anesthetized and awake animals, suggest that AM stimulation is capable of producing low-frequency modulation in the CN at levels comparable to, or sometimes exceeding, those of AC stimulation. The modulation index varies as a function of both the carrier and the envelope frequency. Frequencies that are most efficient maybe specific to the cerebellum due to high firing rates of the PC/CN cells. The optimum frequencies of modulation may differ for different brain sites and local circuits and should be tested accordingly.

Objective:

Electrical stimulation is a promising bioelectronic therapy for tissue repair, yet widespread clinical adoption has been limited by the need for external stimulators, connecting wires, batteries, and complex clinical workflows. We developed a self-powered, electronics-free Wearable Disposable Electrotherapy (WDE) platform that delivers controlled therapeutic electrical stimulation in a thin, bandage-like format without external hardware. This study evaluated its therapeutic performance, tissue response, and electrical dosimetry in a preclinical wound model.

Methods:

A full-thickness dorsal wound model was established in Sprague-Dawley rats and treated daily with WDE, sham dressing, or conventional constant-current (CC) stimulation. Longitudinal wound closure, exudate resolution, histological remodeling, and immunofluorescence markers of regeneration were quantified. Cellular-resolution Boundary Element Method-Fast Multipole Method (BEM-FMM) computational modeling was performed to characterize current pathways and electric field distributions.

Results:

WDE delivered electrical stimulation comparable to conventional CC therapy while significantly accelerating wound healing. Time to 50% wound closure decreased by 29% compared with sham (5.02 ± 1.35 vs. 7.10 ± 1.36 days, p = 0.003), with no significant difference relative to CC stimulation. WDE also promoted earlier exudate resolution (p = 0.016), increased collagen deposition (~25%), enhanced tissue cellularity (~73%), and improved tissue remodeling. Immunofluorescence demonstrated increased αSMA (~2.6-fold), CD31 (~2.0-fold), CD206 (~2.0-fold), and iNOS (~1.7-fold) expression, indicating enhanced regenerative responses. Computational modeling predicted localized therapeutic current pathways and electric fields consistent with the observed biological outcomes.

Conclusions:

This work demonstrates that a self-powered, electronics-free bioelectronic dressing can deliver clinically relevant electrical stimulation while eliminating external hardware and simplifying therapy delivery. By combining therapeutic efficacy with a practical wearable design, WDE represents a promising translational platform for bioelectronic medicine and future regenerative therapies.

Transcutaneous auricular nerve stimulation (tANS), a non-invasive neuromodulation technique, has emerged as a potential therapeutic strategy for modulating immune function; however, the underlying cellular and molecular mechanisms remain unclear. Using single-cell RNA sequencing and cytokine assays, we characterized the transcriptomic and inflammatory response of human peripheral blood mononuclear cells (PBMCs) to tANS in healthy participants. Three healthy subjects received a single 5-minute tANS at the cymba concha (pulse width: 200 μs; pulse frequency: 20 Hz; current: adjusted to sensory threshold). Changes in PBMC transcriptome and plasma cytokine levels were analyzed following ex vivo LPS challenge of whole blood at pre- and 2 hours post-tANS. A total of 32,430 cells were profiled across four conditions: Pre-0LPS, Pre-10LPS, Post-0LPS, and Post-10LPS. Although overall immune cell proportions remained unchanged, tANS induced significant transcriptional reprogramming in monocytes, characterized by downregulation of key proinflammatory transcription factors (NFKB1, FOSB, JUN) and attenuated LPS-induced expression of inflammatory mediators (e.g., TNF, IL-6, CXCL11). Cell-cell communication analysis further revealed decreased IL-6 and TNF signaling alongside increased NCAM signaling in monocytes, collectively indicating a broad anti-inflammatory shift. Consistent with these transcriptomic findings, tANS significantly reduced LPS-induced plasma TNF levels (Pre-10LPS: 2738.8 ± 470.3 pg/mL vs. Post-10LPS: 2056.5 ± 618.7 pg/mL; p=0.0008). These findings provide the first single-cell resolution evidence that tANS induces transcriptomic reprogramming in circulating monocytes under both homeostatic and inflammatory conditions. This work establishes a molecular and cellular basis for the immunomodulatory effects of tANS, supporting its potential as a targeted bioelectronic medicine strategy for inflammatory and autoimmune diseases.

Objective
Transcutaneous cervical electrical stimulation (TCES) offers a noninvasive approach to modulate the autonomic nervous system (ANS), but optimal stimulation parameters remain undefined. This pilot study aimed to identify optimal TCES parameters by evaluating autonomic and neural responses across varying frequencies, current intensities, electrode montages, and durations, using heart rate variability (HRV) and electroencephalography (EEG) alpha-band power as biomarkers of parasympathetic activity.

Methods
Twenty healthy adults completed four testing sessions, each examining one stimulation parameter. Autonomic data were collected including electrocardiography, non-invasive blood pressure, pupillometry, photoplethysmography, and dry-electrode EEG. Four frequencies (10, 25, 40, 150 Hz), three current intensities (sub-sensation threshold, sensation threshold, supra-sensation threshold), three electrode montages (bilateral, left-only, right-only), and two durations (4, 20 min) were tested. Root mean square of successive differences (RMSSD) and global EEG alpha-band power were primary outcomes. Parameters were sequentially optimized across visits based on individual RMSSD responses.

Results
No single frequency produced a significantly higher RMSSD or alpha-band power response. However, each participant exhibited a personalized preferred frequency yielding a mean 41% RMSSD increase in visit 1. This individualized frequency was selected for further visits varying current intensity and electrode montage. Supra-sensation threshold intensity was most effective, with 60% of participants responding strongest at this level. Left-sided stimulation resulted in a decrease in both RMSSD and alpha-band power, while right-sided and bilateral montages resulted in similar increases for these biomarkers. Due to decreasing cardiac vagal response in successive sessions, the preferred frequency was reevaluated before testing duration. The mean RMSSD response increased 54% upon recalibration in visit 4, though the preferred frequency shifted in 75% of participants. Autonomic vitals did not significantly modulate more with longer stimulation duration; pulse rate variability during 20-min stimulation revealed oscillatory autonomic dynamics with peak parasympathetic responses emerging around 4 min.

Conclusions
TCES can modulate cardiac vagal and cortical responses as measured by RMSSD and EEG alpha-band power, respectively, and a personalized, biomarker-guided approach to TCES parameter optimization is essential for future clinical applications targeting autonomic dysfunction.

The brain monitors peripheral inflammatory states and coordinates adaptive immune and physiological responses to maintain homeostasis. We recently demonstrated that neurons in the bed nucleus of the stria terminalis (BNST) encode interleukin (IL)-1β, a proinflammatory cytokine-specific information and engage in immune responses induced by IL-1β. However, whether the BNST functions as a broader cytokine-encoding center capable of processing information from multiple cytokines remains unknown. Here, using targeted recombination in active populations (TRAP2) combined with serial two-photon tomography in mice, we show that systemically administered interleukin-10 (IL-10), an anti-inflammatory cytokine, activates distinct neuronal populations within the BNST compared to other cytokine-responsive brain regions. Chemogenetic reactivation of IL-10-TRAPed BNST neurons significantly suppresses serum TNF (DMSO vs CNO; 2145±284.2 vs 1253±160.5 pg/mL, p<0.05), IL-6 (DMSO vs CNO; 4532±226.8 vs. 3074±186.9 pg/mL, p<0.001) and MCP-1 (DMSO vs CNO; 4368±128.2 vs. 2683±47.7 pg/mL, p<0.001) levels and attenuates hypothermia (Δ core body temperature, TRAPedPBS+CNO vs TRAPedIL10+CNO; -13.3°C ±0.6 vs -5.6°C ±2.5, p=0.02) during endotoxemia. These findings reveal the BNST as a central node for decoding peripheral cytokine signals and orchestrating distinct physiological outputs, with broader implications for neuroimmune regulation of inflammatory diseases.

Introduction: The cholinergic neurons in the dorsal motor nucleus of the vagus (DMN) in the brainstem are a major source of efferent vagus nerve fibers that regulate vital functions, including heart rate and inflammation. Although basal forebrain cholinergic neurodegeneration is a recognized feature of Alzheimer’s disease (AD), whether the integrity of DMN cholinergic neurons is affected during AD remains unknown.

Methods: To address this question, we used 5xFAD mice, a widely used AD model. In 2-, 6-, and 10-month-old female and male 5xFAD mice and age-matched WT (control) mice, we examined the integrity of basal forebrain and DMN cholinergic neurons using choline acetyltransferase (ChAT) immunohistochemistry and assessed hippocampal microglial activation using IBA-1 immunohistochemistry. Mice were also evaluated for behavioral performance (object place memory test). In addition, we studied the effects of electrical DMN stimulation on heart rate in 5xFAD and control mice and on pro-inflammatory cytokine (TNF) responses in mice subjected to endotoxemia.

Results: To inform DMN examination, we first established the age dependence of basal forebrain cholinergic neurodegeneration and associated hippocampal memory impairment, as well as microglial activation, in 5xFAD mice vs controls (with statistically significant changes observed at 6 and 10 months). We further revealed previously unrecognized age-dependent anatomical and functional deterioration in DMN cholinergic neurons. We determined a decrease in the density of cholinergic (ChAT-positive) neurons in the DMN of 6- and 10-month-old female and male 5xFAD mice. We also found that this DMN anatomical alteration was associated with a loss of function through the vagus nerve. While electrical DMN stimulation suppressed heart rate, the magnitude of suppression was age-dependently reduced in 5xFAD mice. In addition, while electrical DMN stimulation decreased pro-inflammatory cytokine levels in control mice subjected to endotoxemia, this anti-inflammatory effect was age-dependently abolished in 5xFAD mice, with females affected earlier, at 6 months.

Conclusions: These results reveal, for the first time, DMN cholinergic deficiencies and dysfunction of brain–to–periphery vagus nerve circuits during experimental AD. These findings provide novel insights into disease pathogenesis and are of interest for developing new therapeutic approaches.

Objective: The autonomic nervous system coordinates inflammatory responses and influences pro-inflammatory cytokine production. Autonomic Neurography (ANG) enables measurement of neural processes involved in autonomic homeostasis and has been demonstrated in both preclinical and clinical settings using invasive and non-invasive approaches. This study expands the assessment of compound action potentials (CAPs) through non-invasive ANG recordings at bilateral ventral cervical sites using ventral cervical electroneurography (vcENG) and ventral cervical magnetoneurography (vcMNG) during acute endotoxemia induced by lipopolysaccharide (LPS), a controlled model for studying ANG responses in humans.
Methods: Two cohorts of 25 healthy males underwent an intravenous LPS challenge. Cohort 1 consisted of 12 subjects across two visits, receiving 3.0 ng/kg and 2.5 ng/kg LPS during Visit 1 (v1) and Visit 2 (v2), respectively. Cohort 2 (c2) included 13 subjects who received 3.0 ng/kg LPS. Neural activity was recorded for 7 hours (60-min baseline followed by measurements at 0, 30, 60-min, and hourly thereafter) from the left and right nodose ganglion (LNG and RNG) and carotid artery (LCA and RCA) sites, with concurrent blood draws. Cohort 1 employed left-sided vcENG and right-sided vcMNG in a magnetically shielded room, whereas Cohort 2 employed bilateral vcENG in an open-field environment. Neural spike rates, and slopes were analyzed using generalized estimating equations (GEE) to compare dose and modality effects.
Results: The ANG neural response (spikes/sec) for the higher dosage (3.0 ng/kg) remained elevated longer, whereas the lower dosage (2.5 ng/kg) peaked earlier and returned to baseline more rapidly. Significant differences in neural spike-rate slopes were observed in Cohort 1 (p<0.05), using left-sided vcENG and right-sided vcMNG, whereas no significant differences in neural spike-rate or amplitude slopes were detected in Cohort 2 using bilateral vcENG.
Conclusions: ANG measured with vcENG and vcMNG non-invasively detects neural dynamics associated with LPS-induced inflammatory responses and distinguish dose-response profiles. Both modalities exhibited similar temporal response patterns, suggesting coordinated bilateral neural activity during systemic inflammation, while vcENG can be measured in open-field settings. These non-invasive ANG tools may enhance understanding of inflammatory cascades with differing cytokine-release magnitudes and support open-field clinical applications.

Problem/Objective: Early identification of inpatient deterioration is critical for timely intervention and improved patient outcomes. This study aimed to develop and evaluate a machine learning model capable of predicting various in-hospital deterioration events using continuously monitored wearable vital signs.

Methods: We analyzed wearable device data (heart rate, respiratory rate, temperature, motion, heart rate variability) alongside age, BMI, and systolic blood pressure from 5211 patient encounters. Data underwent artifact rejection for physiologically implausible values or excessive missingness, down-sampling to one-minute intervals using a rolling mean, and imputation (forward filling, zero imputation for unavailable SpO2). Deterioration events, including unexpected ICU transfers, intubations, expirations, rapid response team (RRT) activations, and cardiac arrests, were meticulously extracted from diverse electronic health record sources. A recurrent neural network (RNN) with Long Short-Term Memory (LSTM) units was trained to predict deterioration within a 0.5-24 hour horizon using a 5-hour data sequence. Model performance was evaluated using 5-fold stratified patient-level cross-validation, reporting ROC AUC and PR AUC.

Results: Data preprocessing effectively managed missingness for all variables. A total of 196 distinct deterioration events were identified across the patient encounters, comprising 126 unexpected ICU transfers, 28 expirations, 22 RRT activations, 19 intubations, and 1 cardiac arrest. The trained RNN model’s cross validated predictive performance (ROC AUB: 0.89 +/- 0.11, PR AUC: 0.7 +/- 0.33) demonstrated that deterioration events could be predicted reliably using clinical wearable signals.

Conclusions: This study establishes a comprehensive framework for leveraging continuous wearable vital sign data and advanced machine learning to predict diverse inpatient deterioration events. The robust methodology for data processing and meticulous outcome identification provides a strong foundation for developing AI-driven early warning systems, highlighting their potential to enhance patient safety and outcomes in acute care settings.

Implantable neural recording and modulation technologies are enabling promising precision therapies for neurological conditions. However, there are very few such devices intended for early feasibility clinical trials, limiting the exploration of these therapies. Commercially available systems, while fully implantable and clinically viable, are either restricted to research use by the manufacturer alone or remain expensive and functionally limited for research applications. Customization of these systems requires regulatory approval, hindering groups from tailoring existing devices to their research needs. While research-tailored systems do exist, they require percutaneous connections to external equipment which can increase risk of infection and restrict participant mobility. To address these challenges, our group was funded through the NIH SPARC-HORNET mechanism to develop a fully implantable, high-channel count neural modulation system under an open-source framework. Here, we present a component of this system called the NOVA (Neural Open-Source Versatile Array). This 64-channel recording module is adapted from an existing implantable, low-channel count research system (Networked Neural Prosthesis (NNP), Makowski et al., 2021) used in humans under an investigational device exemption (IDE). The NOVA includes new and existing NNP circuit elements, as well as a novel titanium package (36x28x44mm) for implantation in the upper-arm or chest region of an average adult. We show that the NOVA can record at 2kHz on 64-channels simultaneously or at 30kHz on a single channel, and we demonstrate in-vivo recordings from implanted electrodes in a non-human primate. To move towards first-in-human trials, our group evaluated the testing required for IDE approval, and we tested the NOVA for current leakage, package hermiticity, and electrode durability. For other groups looking to replicate or customize the NOVA, we’ve outlined our development process and have made all design files, firmware, contractor information, testing procedures, and regulatory materials publicly available for download.

Objective: Delirium and delirium superimposed on dementia (DSD) are common complications affecting patients suffering from ongoing neurodegenerative pathologies. Peripheral surgical trauma can trigger neuroinflammation and ensuing DSD via mechanisms that remain poorly understood. Given the multifactorial therapeutic effects of neuromodulation, including vagal nerve stimulation, we have tested a minimally invasive approach to combat DSD following orthopedic surgery.
Methods: We performed orthopedic surgery on 5xFAD and CVN-AD mice and tested the efficacy of minimally invasive percutaneous vagus nerve stimulation (pVNS). We applied immunohistochemical, biochemical, and behavioral assays to evaluate the impact of surgery on postoperative delirium on DSD pathology in Alzheimer’s disease-like mice. To confirm the role of systemic factors in neuroinflammation and amyloid-β dyshomeostasis, we conducted experiments using interleukin-6 (IL-6), a cytokine commonly upregulated in postoperative delirium and in vitro co-culture assays for validation.
Results: In AD-like mice surgery induced acute changes in amyloid-β; perioperative treatment with pVNS effectively reduced amyloid-β load, plaque sphericity, and neuronal loss. The rescue of these pathological hallmarks led to improved delirium-like behavior, as demonstrated by the 5-choice serial reaction time task on postoperative days 1 and 2. pVNS improved microglial morphology, particularly near amyloid-β plaques. Acute isolation of microglial cells from 5xFAD mice after surgery indicated that pVNS partially enhanced key Disease-Associated Microglia (DAM) markers. The contribution of pro-inflammatory cytokines to amyloid-β aggregation was validated using a transwell culture in vitro model following Cytomix exposure, which also caused endothelial barrier disruption. Finally, we isolated IL-6 as a well-established biomarker of postoperative delirium and described its role in DSD pathology following systemic administration.
Conclusion: These findings establish a role for neuromodulation after pVNS in regulating perioperative immunity and advance a new paradigm for perioperative interventions in patients at risk for DSD.

Background: Transcranial direct current stimulation (tDCS) is typically limited to low stimulation intensities, in part because increasing current produces uncomfortable punctate sensations and focal skin irritation. Since application of substantially higher current densities used in conventional tDCS is tolerable, we hypothesized that localized current hotspots, rather than total applied current alone, represent a primary limitation to tolerability at higher stimulation intensities.
Methods: Current distribution across scalp electrodes was characterized using an electrode divided into 32 individually monitored sub-electrodes. Current through each sub-electrode was recorded during stimulation, allowing spatial maps of current density to be compared with participant-reported sensation. The effects of electrolyte lateral conductivity, resistive current limiting, and hydrogel configuration on current uniformity were evaluated. Based on these findings, a multilayer electrode incorporating distributed passive resistive elements was fabricated using scalable screen-printing and lamination processes.
Results: Conventional low-impedance electrode configurations produced substantial spatial variability in current distribution and localized current concentration. Introducing resistance between the current source and individual sub-electrodes improved current uniformity, while a continuous hydrogel layer further reduced variability and reported sensation. Translation of this architecture to a printed multilayer electrode maintained distributed current limiting and enabled delivery of 6 mA in a bifrontotemporal montage.
Conclusion: Nonuniform current distribution and localized hotspots may be a critical determinant of discomfort and skin exposure during high-intensity tDCS. Incorporating passive distributed resistance into the electrode architecture provides a simple and manufacturable approach for controlling local current density. This strategy may enable safer and more tolerable delivery of higher-current tDCS while remaining compatible with scalable electrode manufacturing.

Encapsulation is a primary determinant of active implant lifetime. Additive manufacturing is attractive for packaging preclinical neuromodulation devices because it offers low per-unit cost, rapid design iteration, and access to low-density polymers that keep the total implant mass low. Mass is an important constraint in small-animal preclinical work, where standard clinical approaches using metals and ceramics are simply too heavy. These advantages are offset by process and material dependent failure modes including high rates of water vapor transmission and interlayer porosity, requiring additional manufacturing steps after 3D printing. Here we report on evaluations of 3D-printed enclosure designs and processing methods, screened by soak testing in saline, with moisture ingress into the sealed volume as the primary performance metric.

We also report on a wireless sensing node architecture that enables the evaluation of multiple designs without the complexity of bulky wires. Each node pairs a precision temperature and humidity sensor with a Bluetooth radio enabled microcontroller that broadcasts measurements in custom advertising packets, avoiding connection overhead so that a single receiver can observe many nodes concurrently. The system was designed to provide long lifetime using tiny batteries by minimizing power draw through sampling and advertising interval selection, deep-sleep scheduling, and idle-draw reduction, yielding multi-month operation sufficient to span a full test campaign. A receiver buffers incoming advertisements and streams them to a host that resolves node identity, logs data, and tracks internal humidity over time.

Enclosure designs, post-processing methods, and soak testing results, including comparative time-to-ingress across configurations, will be presented.

Background: Heart failure with preserved ejection fraction (HFpEF) affects over 3 million people in the US and is characterized by progressive diastolic dysfunction and poor prognosis. Chronic systemic inflammation and autonomic imbalance are key drivers of HFpEF, suggesting that modulating neuroimmune mechanisms may be beneficial. The spleen, both a reservoir and a site for neural regulation of immune cells, may be involved in HFpEF pathogenesis. Splenic focused ultrasound stimulation (sFUS) is a non-invasive neuromodulation therapy that modulates the immune response. Here, we evaluated the ability of sFUS to alter the HFpEF phenotype in a mouse model of the disease.

Methods: We used an established “two-hit” model that recapitulates two main drivers of human HFpEF, obesity and hypertension. Male C57BL6/NJ mice received a high-fat diet and L-NAME-infused water for 8 to15 weeks. In one cohort, mice received splenectomy (SPX) prior to disease induction. In another cohort, mice underwent a daily sFUS (or sham) regimen (pulse length 136.36 ms, PRP 0.5 ms, duty cycle 27%, peak pressure ~418 kPa) for 4 weeks. Echocardiography, exercise tolerance, blood pressure (BP), body weight (BW), and glucose tolerance were performed at study endpoints.

Results: Animals that underwent SPX had worse disease compared to sham, with increases in E/e’ from 25.13 (sham) to 35.94 (SPX) (p<0.022) and isovolumetric relaxation time (IVRT) from 26.2 ms (sham) to 36.5 ms (SPX) (p<0.005). SPX animals also showed greater weight gain compared to sham (Δ11.0 g vs. Δ4.2 g; p<0.002). sFUS-treated animals showed reduction in E/e′ from 28.62 (sham) to 22.16 (sFUS) (p<0.002) and IVRT from 30.34 ms (sham) to 23.28 ms (sFUS) (p<0.0001); these parameters in sFUS-treated animals were comparable to those of healthy controls (E/e’ 20.82 and IVRT 20.92 ms; both p values vs. sFUS non-significant). sFUS-treated animals showed improved exercise tolerance, increasing from 159.7 m (sham) to 232.8 m (sFUS) (p<0.002), without a return to levels comparable with healthy controls. sFUS did not affect BW (p=0.6191) or BP (p=0.1415) compared to sham.

Conclusion: The spleen contributes to the progression of HFpEF. sFUS improves diastolic function in HFpEF without affecting obesity or hypertension, consistent with a direct effect on disease pathogenesis rather than an indirect effect via modulation of risk factors. These data suggest the potential for sFUS to be further tested as the first non-invasive, non-pharmacologic disease-modifying therapy for HFpEF.

Abstract Body

Introduction. Cervical vagus nerve stimulation (cVNS) is an established therapy for drug-refractory epilepsy but only achieves ≥50% seizure reduction in approximately half of treated patients. Adverse off-target effects to laryngeal, respiratory, and cardiac systems constrain cVNS dosages, likely limiting its therapeutic efficacy. The Bionics Institute (BI) has investigated the anterior abdominal vagus, the caudal continuation of the left cervical vagus, as an alternate stimulation site that permits higher dosing without these off-target effects. Previous studies have shown that anterior abdominal VNS (ab-VNS) activates the locus coeruleus, the brain region hypothesized to mediate the anti-seizure effects of VNS, and delays seizure onset in an awake rat seizure model. Building on these findings, we evaluated whether tuning stimulation frequency could improve the anti-seizure efficacy of ab-VNS.

Objective. To elucidate the effect of frequency on ab-VNS, beginning with comparisons of standard and high-frequency stimulation in a pentylenetetrazole (PTZ) rat model of seizure.

Methods. Male Sprague-Dawley rats (8-11 weeks, [n=6]) were chronically implanted with an abdominal electrode array within four weeks of the experiment. Rats were anesthetized with ketamine/xylazine and underwent a craniotomy exposing the dorsal surface of the cortex. Extracellular field potentials were recorded via a tungsten intracortical electrode (0.8 mm depth) in the right somatosensory cortex. Seizures were induced via an intraperitoneal injection of PTZ. Biphasic stimulation (100 µs/phase, 50 µs interphase gap, 2 mA) was delivered across randomized 60-second trials at 30 Hz, 130 Hz, and sham (0 mA), separated by 4-minute intervals. Fast ripples (FRs; 250–600 Hz oscillations) were used as quantitative biomarkers of epileptic activity. The efficacy of each stimulation trial was assessed by normalised FR rate (nFR), defined as a ratio of FR rate during stimulation to the mean FR rate across the 60-second pre and post-stimulation windows.

Results. Compared to sham stimulation, ab-VNS did not reduce nFR rate at 30 Hz (p = 0.65) while 130 Hz stimulation reduced nFR rate by 18% (p = 0.008; one-way ANOVA with Dunnett’s multiple comparisons).

Conclusion. 130 Hz ab-VNS attenuated fast ripple activity, whereas 30 Hz stimulation did not, potentially indicating a frequency-dependent effect. These preliminary findings support ab-VNS at higher frequencies as an alternative approach for drug-refractory epilepsy and warrant further characterization of the frequency-dependent response.

Introduction: Temporal interference stimulation (TIS) has been proposed as a noninvasive strategy for targeting deep neural structures by applying multiple kilohertz-frequency electrical fields that generate a low-frequency amplitude-modulated envelope in tissue. A central assumption in much of the TIS literature is that neurons are insensitive to the kHz carriers themselves and respond primarily to the low-frequency envelope. Here, we re-examine this assumption and argue that the kHz carrier is the critical active component of stimulation.
Methods: We compared true multi-electrode TIS with two-electrode amplitude-modulated kHz stimulation, sine bursts, and unmodulated kHz sine waves in peripheral nerve models, including Locusta migratoria and human sensory and motor pathways. Stimulation thresholds and strength–frequency relationships were measured across kHz carrier frequencies. These mechanistic studies were then interpreted alongside our new translational work using transcranial kHz stimulation in movement-disorder patients, including Parkinson’s disease and essential tremor.
Results: Across peripheral nerve preparations, modulated and unmodulated kHz waveforms produced overlapping excitation behavior. Stimulation thresholds were governed primarily by the kHz carrier frequency, with limited dependence on low-frequency amplitude modulation. These findings support a carrier-driven mechanism consistent with nonlinear membrane rectification, rather than selective demodulation of the envelope frequency. This reframing has immediate consequences: unmodulated kHz stimulation should not be considered inert sham, and some effects attributed to TIS may arise from direct kHz stimulation. Translationally, our clinical work suggests that transcranial kHz stimulation can produce meaningful modulation of movement-disorder physiology and symptoms. In Parkinson’s disease, noninvasive kHz/TIS stimulation reduced pathological beta activity recorded from subthalamic DBS leads. More recent experiments in essential tremor and tremor-dominant PD suggest that direct kHz stimulation can reduce tremor, and that low-frequency envelope modulation may provide similar or, in some cases, less effective outcomes compared with kHz alone.
Conclusions: TIS should be understood as one member of a broader family of carrier-frequency kHz stimulation methods. Recognizing the kHz component as mechanistically critical simplifies device design and opens practical routes toward noninvasive bioelectronic therapies for movement disorders.

Introduction: Parkinson’s disease (PD) is an increasingly prevalent neurodegenerative disorder for which current interventions are symptomatic only. Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are under investigation as potential neuroprotective agents for PD, although mechanisms of action are unclear. Network analysis of FDG PET data has revealed a consistent PD-related metabolic pattern (PDRP) that increases linearly with disease progression, correlates with motor disability, and is reduced by effective treatment. Graph analysis of pathological networks such as PDRP provides a unique perspective regarding the changes in information flow that occur with disease progression and treatment. We have found that assortativity, a measure of connectional homogeneity in a graph is particularly useful in this regard.

Methods: PD patients on daily levodopa treatment were randomized to exenatide (n=26) or placebo (n=29) and followed under the blind for 21 months as part of a phase 2 clinical trial (NCT04305002). They received weekly self-administered injections of either exenatide or placebo for 18 months followed by 3 months of blinded washout. Patients underwent FDG PET at baseline, 9 months, and 21 months Expression and graph metrics were measured for each group and timepoint. Changes within and between groups were assessed with ANOVA. Time courses were analyzed piecewise to study the effects of exenatide vs. placebo on PDRP during the initial (0–9 months) and late (9–21 months) trial periods.

Results: During the initial period, PDRP expression was reduced in the exenatide group but was unchanged with placebo (p<0.02, group*time interaction effect), and a similar effect was seen with PDRP assortativity (p<0.0001). During the subsequent period which included washout, PDRP expression rose to approximate the placebo mean (p<0.009, interaction effect). By contrast, the initial reductions in assortativity seen with exenatide at 9 months (p<0.0001, compared to baseline) persisted at 21 months (p<0.005); significant changes were not seen in the placebo group at either timepoint.

Conclusions: These findings suggest that in neurodegenerative disorders, GLP-1RAs can act by suppressing pathological network activity in a reversible fashion. These drugs may also act by stabilizing network organization, thereby reducing vulnerability and optimizing information flow over the long term.

Background: Enhancing tDCS technology can support the delivery of higher current intensities, enabling broader dose-response studies in human clinical trials.

Methods: High-Capacity tDCS (HC-tDCS) integrates novel electrodes and adaptive current/voltage controlled electronics. Multi-layer HC electrodes include polarity-specific redox layers and engineered hydrogel interfaces, configured for a bifrontotemporal montage. The stimulator design includes adaptive ramps with hybrid voltage-current control and low compliance voltage. Scanning electron microscopy (SEM) and electrical impedance spectroscopy (EIS) characterized electrode properties. Tolerability of HC-tDCS was tested for target currents 1-6 mA (in 1 mA increments) for 30 min on 10 healthy subjects, and compared with conventional tDCS using 2 mA F3-F4 sponge electrodes. Tolerability was assessed according to the 100 mm visual analogue scale for pain (VASP-100), skin erythema assessment, thermal imaging, and adverse event questionnaires. MRI-derived computational models predicted brain electric fields.

Results: The electrode design, including high-roughness polarity-specific capacity, electrochemically supports high-charge direct current stimulation. In all subjects, HC-tDCS was well tolerated at all tested doses (1-6 mA) with minor transient adverse events. Average VASP-100 for 6 mA with HC-tDCS was less than 16, comparable to sponge-electrode tDCS at 2 mA. HC-tDCS operates at lower voltage than sponge-tDCS. Across MRI-derived head models, 6 mA HC-tDCS produced 2.25–2.99 V/m electric fields, ~4.2-fold higher than conventional 2 mA tDCS electric fields of 0.49–0.85 V/m. 6 mA HC-tDCS produced >1 V/m across all evaluated regions, including DLPFC, anterior cingulate cortex, insula, and amygdala. In contrast, no evaluated region reached 0.5 V/m with conventional 2 mA tDCS.

Conclusions: HC-tDCS was well tolerated by participants and expands the accessible tDCS dose range, enabling stimulation up to at least 6 mA and substantially higher electric fields throughout cortical and deeper brain regions than conventional 2 mA tDCS.

Abstract Body

Problem/Objective

Postoperative pain remains a significant challenge following ambulatory surgery and may adversely affect recovery, functional outcomes, and patient confidence. As a bioelectronic medicine intervention, transcutaneous electrical nerve stimulation (TENS) is supported as an evidence-based, nonpharmacologic approach to postoperative pain management; however, standardized implementation into routine ambulatory surgical practice remains limited. This evidence-based practice initiative evaluates the feasibility of implementing a standardized nurse-driven TENS protocol to improve postoperative pain self-efficacy among adult ambulatory surgical patients.

Methods

This evidence-based practice initiative is guided by the Knowledge-to-Action Framework and conducted in an ambulatory surgery setting. Eligible adult patients undergoing laparoscopic and robotic outpatient procedures receive standardized TENS education utilizing the teach-back method and a TENS unit for postoperative home use following discharge. Pain self-efficacy is measured using the Pain Self-Efficacy Questionnaire-2 (PSEQ-2) before the intervention and during postoperative follow-up. Implementation fidelity is supported through standardized patient education, stakeholder engagement, participant tracking, and secure electronic data management using REDCap.

Results

Implementation has demonstrated successful integration of a standardized nurse-driven TENS protocol into the ambulatory surgery workflow without disrupting routine clinical care. Early implementation identified participant recruitment as a barrier; however, targeted stakeholder engagement, interdisciplinary collaboration, and workflow refinement strengthened participant recruitment while maintaining implementation fidelity and adherence to the approved implementation protocol. Participant follow-up and outcome data collection remain ongoing, and final statistical analyses will be completed after the implementation period.

Conclusions

Preliminary findings support the feasibility of implementing a standardized nurse-driven TENS protocol within an ambulatory surgery program. This implementation strategy may facilitate the adoption of evidence-based, nonpharmacologic pain management, promote patient self-efficacy, and support the sustainable integration of these approaches into perioperative nursing practice. Final analyses will evaluate the intervention’s impact on postoperative pain self-efficacy and inform future dissemination, sustainability, and broader implementation of evidence-based postoperative pain management strategies.

Background: Vagus nerve stimulation to engage the inflammatory reflex is an emerging bioelectronic medicine therapy with limited efficacy data in ulcerative colitis (UC). Research has identified the Nucleus Tractus Solitarius (NTS) as a key node in the inflammatory reflex. Respiratory-gated auricular vagal afferent nerve stimulation has shown that synchronizing non-invasive vagal stimulation with the respiratory cycle preferentially engages NTS. We assessed the safety and efficacy of this stimulation paradigm as a concomitant add-on therapy in children and young adults with UC.

Methods: Patients (10-39 years) with mild to moderately active UC were eligible for enrollment based on a modified UC disease activity index (UCDAI) score of 3-6 and a fecal calprotectin ≥ 150 µg/g despite stable medication regimens. In this randomized, double-blind, sham-controlled trial (funding and investigational devices provided by Cala Health®, San Mateo), 30 subjects were randomized 2:1 to self-administer active therapy (cymba conchae, 300 μs, 100 Hz, gated to exhalation) or no-stimulation sham for two 10-minute sessions per day. Primary and secondary endpoints were clinical remission (UCDAI 0-1) and clinical response of (≥ 2-point reduction in UCDAI) at week 12. Exploratory endpoints included change in fecal urgency Numeric Rating Scale (0-10).

Results: Baseline characteristics of the 15 adults and 15 children included 70% pancolitis and 30% left-sided colitis; 80% of subjects failed to achieve remission with at least one biologic. No subjects used corticosteroids during the study. Clinical remission occurred in 50% of active vs 20% of sham subjects (Δ30%, p=0.24). Statistically significant clinical response was observed in 75% (15/20) of active vs 30% (3/10) of sham subjects (Δ45%, p=0.045). A clinically meaningful improvement (≥ 3-point decrease) in fecal urgency score occurred in 30% (6/20) in the active group vs 10% (1/10) in the sham group. There were 0 severe adverse events related to the device.

Conclusion: Device use had a favorable safety profile and showed a clinically promising effect in adults and children towards achieving remission, response, and reduction in fecal urgency severity within 12 weeks compared to sham. Further investigation is warranted.

Non-selective vagus nerve stimulation (VNS) often causes off-target side effects, limiting maximum stimulus intensity and therapeutic efficacy. Function-selective VNS, which could improve safety and efficacy, requires a detailed understanding of human vagal fiber organization. To address this, we mapped the transverse and longitudinal organization of sensory and motor fibers in 20 human left and right vagus nerves (VN).
Using micro-CT, 3D reconstruction, and quantitative immunohistochemistry (IHC), we characterize fascicles based on their origin from the nodose ganglion (nodose-originating, NO) or bypassing it (nodose-bypassing, NB). IHC staining quantified fibers into four categories: myelinated motor (MM; NF+, MBP+, ChAT+), myelinated sensory (MS; NF+, MBP+, ChAT-), unmyelinated motor (UM; NF+, MBP-, TH+), and unmyelinated sensory (US; NF+, MBP-, TH-).
NO fascicles have predominantly myelinated sensory (MS: 89.2 ± 0.5%) with minimal myelinated motor fibers (MM: 0.3 ± 0.1%). In contrast, NB fascicles contain more MM (17.3 ± 2.4%) and fewer MS fibers (72.8 ± 2.4%), confirming distinct sensory and motor profiles consistent with fascicle origins. NO and NB fascicles remain largely separate close to the nodose, with mixing occurring distally. By estimating MM and MS fiber compositions in downstream fascicles, we find, in 14 of the 20 nerves, over 50% of the total fascicle area was mixed by the level of the laryngeal prominence (LP). At two centimeters above the LP, in only 4 of the 20 nerves total fascicle area exceeded 50% mixed. To quantify the cross-sectional separation between MS and MM fibers at different levels, we anatomically aligned (medial-lateral, rostral-ventral) 18 nerves and averaged MS and MM areas within nerve sections. We found that MS and MM fibers occupy distinct quadrants: dorsal-medial for sensory, ventral-lateral for motor, for both left and right nerves. The sensory and motor “hotspots” are clearest at three centimeters above the LP, and the MS-MM asymmetry is maintained through the LP.
These findings suggest that placement of VNS devices two centimeters above the level of the LP could permit spatially selective activation of sensory vs. motor functions of the VN. Based on anatomical organization, steering stimulation with selective devices towards the dorsal-medial region of the left or right cervical VN would in principle promote sensory over motor responses.

An estimated 100 million Americans have experienced chronic pain producing significant economic and social burden. Chronic pain is associated with disability, cognitive impairment, cardiovascular disease as well as psychiatric/substance use disorders and is more prevalent among Veterans compared to non-Veterans, with the most common site as the back. Analgesic pharmacologic treatment approaches have limited efficacy, including opiates which are associated with increased risk of substance use disorder, overdose and death. Chronic pain affects neurocircuitry associated with nociceptive processing, mood, and self-referential processing. The insula is an integral part of the pain connectome and part of the salience network. The posterior subregion of the insula (PI) receives nociceptive input from spinothalamic tracts, relays it to the anterior insula which integrates expectation, awareness, and emotion to form the overall experience of pain. Our preliminary data indicate that LIFU to PI reduces laboratory measures of central sensitization and evoked pain in healthy controls. We utilized the spatial specificity and depth penetration of low intensity focused ultrasound (LIFU) to target the PI in participants with chronic back pain (CBP) and assessed the neural response to evoked (thermal) pain during fMRI scanning immediately before and following LIFU We present preliminary results from N=15 participants in an on-going study who received one session each of LIFU (Af= 500 Hz, PRF=100Hz, DC=10%, SD=120s) and sham (sessions separated by >7 days) to the PI bilaterally (sequentially administered to right and left PI counterbalanced) using the NeuroFUS CTX-500. Resting state functional connectivity scans were also performed prior to the task. LIFU was well-tolerated and there were no serious adverse events. There was significant activation in blood-oxygen dependent (BOLD) response to ‘hot’ cues versus the ‘warm’ and rest periods in regions such as the right and left insula, precentral and postcentral gyri, and left and right opercula. The effect of LIFU versus sham on BOLD response to these conditions will be discussed. The effect of resting state tonic pain signature and neurologic pain signature during the task will also be discussed as well as their correlation with participant pain ratings.

Problem/Objective:
Tobacco Use Disorder (TUD) is one of the leading causes of preventable death in the world. Most smokers have a strong desire to quit smoking, but most attempts fail within a week. Efficacy of current treatments for smoking cessation, including pharmacotherapies and psychosocial interventions, is poor, with most smokers relapsing within a year following treatment. Non-invasive neuromodulation of brain regions activated in response to smoking cue exposure offers a potential approach to understanding the neurocircuitry underlying TUD and may be a potential treatment for the latter. A key region activated in smoking cue exposure is the insula, specifically the anterior insular cortex (AI), which is a component of the salience network. Smokers with insular infarcts demonstrate a dramatic reduction in symptoms of TUD. The functional role of the insula in TUD can be probed using low intensity focused ultrasound (LIFU).

Methods:
We present results from an ongoing study in individuals with current TUD, moderate-severe, using LIFU to modulate the dorsal AI to determine the blood-oxygen-level-dependent (BOLD) response to smoking vs neutral cue-exposure as well as assessments of cigarette craving. We present results from n=17 participants who received 1 session of LIFU (NeuroFUS, CTX-500 transducer) to the left dorsal AI (Af: 500 kHz, PRF: 100 Hz, DC: 10%, SD: 120 seconds) and one session of sham stimulation (sessions were separated by > 7 days). Immediately before and after LIFU/sham, participants underwent the smoking cue exposure task during fMRI scanning. Resting state functional connectivity was also conducted before the smoking cue task.

Results:
LIFU was well-tolerated and there were no serious adverse events. LIFU compared to sham significantly reduced BOLD response to smoking vs neutral cues in the left insula, precuneus, and cingulate gyrus. There were no significant changes in state measures of smoking craving for LIFU compared to sham.

Conclusions:
LIFU versus sham effects on resting state functional connectivity between key nodes of the default mode network, the medial prefrontal cortex (mPFC) and posterior cingulate cortex (PCC), and salience network, the anterior cingulate cortex (ACC) and AI, and their correlations with smoking cue-reactivity will also be discussed.

Neural implants hold great potential as a treatment for various neurological disorders. However, placement of these implants can damage brain tissue during insertion and evoke a foreign body response (FBR). The FBR can damage neurons, causing glial scarring and neuronal cell loss near electrode sites, which can ultimately affect electrode longevity and stability. To clear the path for clinical applications, significant efforts to limit the chronic effects of insertion damage are still needed. Past work has reported that transcranial focused ultrasound promotes neurotrophic factors in cerebral disease and injury models. The current study explores the protective effects of localized low-intensity pulsed ultrasound (LIPUS) to reduce microglia reactivity and preserve electrophysiological quality in rodents following chronic electrode implantation. A reusable head-mounted LIPUS delivery system was developed and tested. Transducers were characterized in water prior to in vivo testing. To record neural activity, NeuroNexus probes were implanted at a 45°angle (A4x4-5mm-100-125-703-CM16LP) or vertically (A1x16-5mm-100-703_H16) to record stimulus evoked activity in rats (N=12, Sham=4, Vertical LIPUS=4, Slanted LIPUS=4). Subjects were treated with periodic LIPUS or Sham (Control), daily in Week 1, followed by twice weekly treatments during Weeks 2-6. LIPUS delivered in 3 x 5 min periods with 5 min of quiescence between each period. Stimulation parameters: Burst Duration: 50 ms, Repetition Rate: 2Hz, Ita: ~250-300 mW/cm2. Reactivity and neural activity were assayed with two-photon imaging and electrophysiology recordings. Ex vivo and bench tests demonstrated that ultrasound transmitted across the skull without detrimental heating or signal loss. LIPUS-treatment resulted in a lower level of probe coverage around Day 6 and remained consistently lower than Sham (two-sided Šídák’s multiple comparisons test, p=0.0187, LIPUS=7; Sham=6. Glial fibrillary acidic protein intensity, whose expression increases in response to neuroinflammation or injury, increased in the Sham compared to LIPUS-treated, 30 µm from the probe. LIPUS treatment significantly decreased average vessel diameter on day 28 (one-sided Tukey’s multiple comparison test, p=0.02). LIPUS-treated showed an improved signal-to-noise ratio at 28 days, improved single unit detection, reduced capillary diameter, and reduced microglial and astrocytic encapsulation of the probe- all critical factors for improving longer-term neural electrode viability.

Development of microelectrode arrays (MEA), Neuropixels probes, optics, and chemogenetic delivery needles has driven neuroscience research forward. Despite efforts to minimize tissue damage through design optimization, the resultant physiological response following device insertion continues to limit research. Minimizing mechanical forces during insertion can mitigate this response and improve experimental and physiological outcomes, such as neural activity and reactive gliosis. Use of NeuralGlider® Inserter (NeuralGlider, Actuated Medical, Inc.), a micron-scale ultrasonic vibration system, reduces the insertion force and cortical dimpling for device placement. In vivo, acute recordings following NeuralGlider-implanted MEAs into porcine cortex showed higher single unit yield. Herein, we investigate NeuralGlider-aided placement of Neuropixels, gradient-index (GRIN) lenses, and infusion cannulas/needles on bench, ex vivo, and in vivo models. NeuralGlider lowered the force for Neuropixels probes to penetrate an agarose model, reducing both probe buckling and dimpling. Insertion force decreased by 77% for Neuropixels 1.0 single-shank (SS) probes and by 73% for Neuropixels 2.0 probes. In vivo, NeuralGlider-assisted cortical insertions reduced tissue buckling and enabled insertion depths of up to 10 mm with Neuropixels 1.0 NHP SS probes. Additionally, Neuropixels 2.0 multi-shank and SS probes were successfully inserted without buckling into the cortex of large and small preclinical models, respectively. NeuralGlider insertions of 0.5 mm GRIN lenses and 1.0 mm optical lenses into ex vivo porcine brain tissue reduced pial puncture force by >50% and resultant tissue compression (p<0.01, Student’s T-test). In vivo NeuralGlider GRIN lens insertions (mouse medial prefrontal cortex) were completed utilizing controlled insertion speed, omitting tissue aspiration, and demonstrating improved implantation success, offering a reliable GRIN lens implantation procedure for deep-brain calcium imaging. Ex vivo testing demonstrated that infusion needle vibration reduced puncture force by >75% at speeds up to 500 µm/s and enabled puncture of porcine dura with 33G needles. In vivo testing showed smaller gauge infusion needles reduced aliquot backflow along needle tracks. Entrance and infusion sites also showed higher cellular density (less damage) with vibrated insertion after 6 weeks. NeuralGlider’s micron-scale, ultrasonic vibration allows for insertion in vivo and ex vivo, while reducing probe buckling, tissue dimpling and allowing entry into difficult tissue types.

Background: Mitochondrial dysfunction is a central driver of osteoarthritis (OA) pathogenesis, contributing to impaired chondrocyte metabolism, reduced matrix synthesis, and progressive cartilage loss. Synovial fluid–derived stem cells (SF MSCs) naturally reside within the joint and possess the capacity to donate healthy mitochondria to neighboring cells. This study integrates in vitro mechanistic analyses with in vivo therapeutic evaluation to determine whether direct mitochondrial transfer from SF MSCs can restore chondrocyte function and mitigate cartilage degeneration.

Methods: SF MSC mitochondria were labeled with MitoTracker CMXROS Red, while chondrocyte mitochondria were labeled with MitoTracker Green. Co culture experiments at ratios of 1:1, 1:10, and 1:100 were performed to visualize mitochondrial transfer via fluorescence microscopy. For functional assessment, human chondrocytes were treated with isolated SF MSC mitochondria (10⁶ particles) under normal and IL 1β–induced inflammatory conditions. After five days, Sox9, Aggrecan, and Collagen II deposition were quantified using immunofluorescence.
For in vivo evaluation, a destabilized medial meniscus (DMM) rat model was created (n=9). Rats received intra articular injections of vehicle, low dose (10 µg), or high dose (50 µg) SF MSC–derived mitochondria. After 14 days, cartilage integrity was assessed using Safranin O staining.

Results: In vitro, SF MSC mitochondria successfully transferred into chondrocytes across all co culture ratios. Mitochondrial uptake enhanced Sox9, Aggrecan, and Collagen II expression, demonstrating restoration of chondrocyte anabolic activity. Under IL 1β stress, mitochondrial treatment partially rescued matrix protein suppression, indicating that mitochondrial donation supports chondrocyte resilience even in inflammatory environments. In vivo, mitochondrial transplantation significantly reduced cartilage degeneration in a dose dependent manner. Safranin-O staining revealed improved aggrecan retention and reduced structural damage in mitochondrial treated joints compared to controls. The high dose group exhibited the greatest preservation of cartilage architecture, confirming the therapeutic potential of mitochondrial augmentation.

Conclusion: Direct mitochondrial transfer from SF MSCs restores chondrocyte phenotype in vitro and attenuates cartilage degeneration in vivo. These findings establish mitochondrial transplantation as a promising therapeutic strategy for OA, providing mechanistic and preclinical evidence that targeting mitochondrial dysfunction can meaningfully alter disease progression.

Background: HIV-1-associated neurocognitive impairment (HIV-1-NCI) affects up to 50% of people living with HIV-1 (PLWH) despite effective antiretroviral therapy (ART). Persistent HIV-1 reservoirs in the central nervous system (CNS), particularly within microglia, drive HIV-1-associated neuroinflammation (HIV-1-AN), leading to blood-brain barrier (BBB) dysfunction, immune-cell infiltration, and neuronal injury. Aging further contributes to HIV-1-NCI progression through mechanisms that remain poorly understood. Consequently, reducing HIV-1-AN has emerged as a promising strategy to prevent HIV-1-NCI.

Transcutaneous auricular neurostimulation (tAN) is a noninvasive therapy that activates the vagus nerve and/or trigeminal nerve to suppresses inflammation through the cholinergic anti-inflammatory pathway. Although tAN stimulation has been shown to be neuroprotective, its impact on HIV-1-AN remains poorly understood due to limited biomarkers and translational models. Current models also fail to capture donor specific aging signatures that are erased during cellular reprogramming and are key to modeling HIV-1 in vitro. We have developed patient-derived monocyte derived microglia like cells (MDMi), termed “epimicroglia,” which preserve donor-specific epigenetic and inflammatory signatures, serving as a sensitive ex vivo indicator of HIV-1-AN status in PLWH undergoing tAN.

Hypothesis: We hypothesize that epimicroglia obtained from PLWH before and after tAN will reveal shifts toward a less inflammatory phenotype following treatment. Furthermore, integration of these patient-derived epimicroglia into brain organoids is expected to reproduce HIV-1-AN, while post-tAN epimicroglia may induce reduced inflammatory signaling.

Objective & Methods: Twenty-five older adults (>50 years) with HIV-1-NCI and durable virologic suppression will undergo 30 days of home-based tAN using the FDA-cleared Sparrow Ascent device. Peripheral blood samples collected longitudinally will be used to evaluate inflammatory cytokines, BBB permeability biomarkers and epimicroglia derived from the same samples.

Results: We present preliminary studies showing successful generation of epimicroglia expressing CD11b, Iba-1, and TREM2 while retaining donor age-associated epigenetic characteristics. Epimicroglia were permissive to HIV-1 infection, infiltrated brain organoids, and maintained stable neuronal activity measured by multielectrode array recordings.

Expected Impact: This project establishes a novel translational platform integrating patient-derived microglia and brain organoids to investigate the impact of tAN in PLWH with HIV-1-NCI.

Introduction: Mechanical forces influence cell signaling, matrix remodeling, and therapeutic response, but many force-generating events inside soft biomaterials remain difficult to measure directly. Conventional methods such as AFM, traction force microscopy, and FRET-based probes have provided important insight, but they can be difficult to adapt to thick gels, hybrid extracellular-matrix materials, or ultrasound-activated therapeutic platforms. We are developing molecular mechanical sensors (MMS) that can be embedded into biomaterials and nanostructures to report local mechanical activation and guide material design.

Methods: We designed MMS with tunable force sensitivity and incorporated them into hybrid biomaterials, including collagen gels, fibronectin-hyaluronic acid matrices, and alginate-based gels. These materials were used to study mechanical transduction in soft biomimetic environments. In parallel, MMS-enhanced nanostructures were evaluated as ultrasound-responsive delivery materials under MHz-level focused ultrasound. Optical readouts from the MMS were used to connect local mechanical activation with material response and payload release.

Results: MMS-containing biomaterials generated spatially resolved optical signals under mechanical deformation, suggesting that molecular-scale force reporters can map mechanically active regions inside soft matrices. The same design concept also enhanced ultrasound-induced dynamic mechanical response in biogels, creating materials that can both sense and respond to acoustic stimulation. In MMS-enhanced nanostructures, MHz-level focused ultrasound promoted near-complete payload release within minutes in current material models. This result is encouraging because ultrasound-responsive delivery platforms often face tradeoffs among release efficiency, acoustic exposure, spatial control, and material disruption. By tuning sensor structure and host material architecture, we can adjust the activation window for different biomechanical and therapeutic ultrasound settings.

Conclusions: Molecular mechanical sensors provide a molecular-scale route to connect force sensing, biomaterial design, and ultrasound-guided therapeutic delivery. Rather than treating mechanical stress as an invisible input, these molecular biosensors convert local deformation and acoustic activation into measurable optical and functional outputs. This work may open a practical path toward force-informed biomaterials, acoustically triggered delivery platforms, and future hybrid ultrasound-optical imaging strategies for bioelectronic medicine.

Introduction: Vagus nerve stimulation (VNS) is an emerging therapy for autoimmune, inflammatory, epileptic, and depressive disorders. Optimizing VNS devices requires characterization of fascicular anatomy and its inter-subject variability. Prior work has used histological staining and micro-CT to study vagal fascicles, but analyses have been limited in sample size and topological scope. Here we present a morphometric and graph-theoretic characterization of fascicular networks reconstructed from micro-CT across 38 nerves, representing a step toward larger-scale studies as manual steps are further automated.

Methods: Thirty-eight vagus nerves from 19 cadavers were imaged using micro-CT at 9-micron resolution. A 3D UNet was trained to classify each voxel as background, epineurium, or endoneurium, and iteratively improved by correcting predicted masks and retraining. Fascicle centerlines were extracted, converted to graphs, and denoised. Adjacent segments were stitched by joining fascicles at their boundaries. Nerves were straightened, untwisted using anterior microfil markers, and aligned across subjects via dynamic warping to a template based on four anatomical landmarks. Morphometric features including fascicle count, total fascicular area, splitting and merging rates, internodal segment length, equivalent nerve diameter, and fascicle position relative to the nerve boundary were computed along each nerve. Significance was assessed using cluster permutation tests. Branch fascicle penetration depth was quantified by branch type. Graph-based metrics capturing topological redundancy, branching complexity, and network connectivity are also being computed.

Results: Morphometric features showed high variability across subjects and between left and right nerves. Fascicle counts ranged from 1 to over 35, splitting and merging rates from 5 to 70/cm, and equivalent nerve diameters from 1.4 to 7.4 mm. Cluster permutation tests revealed statistically significant regional differences between left and right nerves. Branch fascicle penetration depths varied by branch type, with recurrent laryngeal and laryngeal branches traversing the greatest distances on average.

Conclusions: This analysis quantifies the morphometric variability in human vagus nerve fascicular networks across 38 nerves. Ongoing work on graph-based topological metrics will offer additional insights into network architecture, signal routing, and redundancy. Further automation of manual steps will enable larger-scale studies capturing population-level variability. These findings can inform the design of targeted VNS devices.

Optogenetics is being investigated as an alternative to electrical stimulation for numerous bioelectronic medicine applications due to its genetic selectivity which can potentially reduce off-target side effects. Channelrhodopsin-2 (ChR2) is one of the most widely used opsin variants in research today, particularly in the peripheral nervous system. However, ChR2 stimulation is limited to ~30 Hz due to its relatively slow photokinetics. Here, we investigated whether two-site, spatiotemporal vagus nerve simulation (VNS) could be used to bypass this frequency limit and enhance the performance of ChR2. We designed custom 3D printed nerve cuffs with integrated LEDs to interface with the vagus nerve of anesthetized ChAT-Cre x ChR2(H134R) mice. Two separate cuffs were placed on the left vagus nerve, and three seconds of pulsed illumination was delivered using a single nerve cuff or alternated between the two cuffs to perform “drumbeat” VNS. Electrocardiogram recordings were collected as a physiological read-out of optogenetic VNS. At stimulation frequencies of 30 and 60 Hz, drumbeat VNS resulted in significantly larger reductions in heart rate compared to stimulation with a single nerve cuff (n = 3 mice, 3 trials per mouse). This simple but innovative stimulation technique can be translated to other opsin variants and nerves to enhance optogenetic control of peripheral nerves.

Dystonia is characterized by unwanted muscle spasms causing involuntary movements and abnormal postures. Using metabolic PET and resting-state fMRI, we identified similar network topographies across independent sporadic and inherited dystonia cohorts, including non-penetrant mutation carriers. The dystonia-related network was characterized by increased activity in sensorimotor, premotor, parietal association regions, putamen and cerebellar vermis. Graph analysis revealed that the sensorimotor cortex, inferior parietal lobule, and cerebellum constituted a discrete core zone within the dystonia network [1].
Based on these findings, we applied network guided repetitive transcranial magnetic stimulation (rTMS) to these core regions to treat a 67-year-old male with 15 years of intractable sporadic cervico-truncal dystonia. Cortical and cerebellar targets were identified on the individual’s anatomical MRI in native space with reference to the previously defined dystonia network. Targeted theta-burst stimulation was performed three times weekly for six weeks. To assess network modulation, resting-state fMRI was acquired at baseline and biweekly during treatment. At each time point, we computed expression values for four dystonia-related subnetworks identified by independent components analysis in a reference cohort of 20 sporadic dystonia patients and 28 healthy controls (HC).
By trial’s end, the subject reported substantial clinical improvement, turning his head independently of his shoulders and trunk for the first time since symptom onset. His Burke-Fahn-Marsden and TWSTRS scores decreased by 8.3% and 37.5%, respectively. Of four dystonia-related independent components (ICs), two involved sensorimotor pathways and two were linked to non-motor features. Sensorimotor ICs exhibited continuous improvement: IC10, elevated above the HC mean at baseline, declined 15%, 23%, and 19% toward normal after 2, 4, and 6 weeks, respectively. IC21, below normal at baseline, rose 21%, 24%, and 24% to reach supernormal levels at each time point.
These data demonstrate a combination of increased excitation and reduced inhibition in independent sensorimotor subnetworks in dystonia. Theta burst rTMS applied to predefined cortical and cerebellar nodal targets corrected both functional abnormalities. We hypothesize that rTMS improves symptoms by stabilizing multiple sensorimotor subnetworks simultaneously [1].
1. Vo et al. https://doi.org/10.1093/cercor/bhad012.

Objective: Rapid bacterial detection remains a critical challenge in clinical diagnostics, food safety, and environmental monitoring. As traditional methods rely on lengthy lab cultures and workflows, there remains a need for a rapid bacterial sensing method. Bacterial components activate sensory neurons through recognition receptors (formyl peptide receptors, Toll-like receptors, and GPCRs) binding and ion channel (TRPA1 and TRPV1) activation. This suggests that bacteria-mediated sensory neural activation can be used as a label-free bacterial sensing strategy.
Methods: In this study, we used rat dorsal root ganglion neurons and a high-density microelectrode array (HD-MEA) platform to record neural electrophysiological activity in response to bacterial exposure. We cultured the neurons on the Maxwell Biosystems HD-MEA platform for 11-14 days. Throughout this period, we monitored cell culture health and recorded the baseline neural activity to establish an MEA-specific neural map. On days 11-14 we expose the neurons to capsaicin and heat-killed gram-positive (S. aureus) and gram-negative bacteria (E. coli). 5-minute scan routines were performed to record the neuronal activity changes upon exposure.
Results: A subset of neurons were found to be responsive to capsaicin (1 µM). Within the capsaicin-responsive neuron subpopulation, a smaller subset of neurons were responsive to either E. coli or S. aureus, while some neurons were responsive to both. In addition, the time course of exposure-induced activity changes differently in the capsaicin, E. coli, and S. aureus experiments.
Conclusions: Neural activity changes induced by bacterial exposure were observed in TRPV1 (i.e., capsaicin)-expressing neurons. Differential response to the gram-positive versus gram-negative bacteria suggests that the neural culture may yield specific sensory outcomes (i.e., number of neurons firing, amplitude change in firing rate, and time course of neuronal response) that depend on the type of pathogen-associated molecular patterns (PAMPs) or toxins produced by each microbe. This study is proof of concept for label-free bacterial sensing methods using sensory neurons. Our ongoing work will focus on characterizing differential neural electrophysiological responses across multiple bacterial strains to develop a rapid, culture-free bacterial detection system.

Background: Despite effective antiretroviral therapy, HIV-1-associated neurocognitive impairment (HIV-1 NCI) remains highly prevalent, primarily driven by chronic HIV-1 associated neuroinflammation (HIV-1-AN) and viral persistence within the central nervous system. Transcutaneous auricular neurostimulation (tAN), a non-invasive neuromodulatory technique targeting the trigeminal and vagus nerves (CN V and CN X), has demonstrated significant anti-inflammatory and neuroprotective effects in other conditions. However, its potential to mitigate neurocognitive decline and reduce HIV-1-AN in people living with HIV has not yet been explored.

Objective & Methods: We have launched an open-label, single-arm, non-controlled feasibility pilot study to investigate the safety, tolerability, and clinical feasibility of tAN as a novel intervention for individuals with HIV-1 NCI. The study will enroll 25 older adults (age > 50 years) with a documented diagnosis of HIV-1 NCI and durable virologic suppression from the Northwell Health Center for AIDS Research and Treatment. Following baseline physical and cognitive evaluations, participants will be trained to operate the Sparrow Ascent, an FDA-cleared wearable device. Participants will self-administer 30 minutes of targeted tAN daily at home for 30 consecutive days. We will conduct clinical follow-ups and peripheral blood draws at regular intervals throughout the intervention (Days 0, 3, 15, 21, and 30) and during a 2-week post-study phase. Our primary endpoints focus on operational feasibility, intervention adherence (target ≥80%), and safety. Secondary clinical outcomes include longitudinal changes in cognitive function, assessed via the Repeatable Battery for the Assessment of Neuropsychological Status Update (RBANS Update), Trail Making Test, Stroop Color-Word Test, and Patient Health Questionnaire-9 (PHQ-9), as well as systemic inflammatory biomarker reduction associated with HIV-1-AN. Specifically, we will measure changes in pro-inflammatory cytokines (e.g., TNF-α, IL-1β, IL-6), macrophage/monocyte activation markers (e.g., sCD14, sCD163, sTREM2), and blood-brain barrier permeability indicators (e.g., S100B, MMP-9, VCAM).

Expected Impact: This study will be the first to evaluate the clinical integration of non-invasive nerve stimulation for HIV-1 NCI. By establishing fundamental proof-of-concept data regarding feasibility, tolerability, and preliminary biological impact, this pilot will lay the groundwork for larger, fully powered clinical trials aimed at mitigating neuroinflammation and preserving cognitive function in the aging HIV population

Introduction: Chronic pain effects roughly 24% of the United States population. Current treatments involve tissue injections of a nerve blocker near nerve roots. However, these approaches are invasive, requires multiple doses, and risks nerve damage. Microbubbles offer a potential alternative by enabling non-invasive, localized, and sustained effects, but have yet to be applied for peripheral neuromodulation. Microbubbles can be administered systemically, like drugs, but only release the drug at the targeted nerve site through localized, focused ultrasound. Yet, the mechanism to load drug into microbubbles (i.e., conjugation of drug to the microbubble shell) limits the loading efficiency and release kinetics of the therapeutic.

Here we developed poly(pro-drug) materials that incorporate neuromodulatory molecules directly into the polymer backbone; which are used to form microbubble shells. Specifically, we synthesized a poly(curcumin-co-PEG) material containing curcumin, a TRPA1 agonist. Our poly(pro-drug) formulation enhances bioavailability and enables controlled release kinetics over time. We further compared our poly(pro-drug) to that of monomeric curcumin and evaluated their ability to provide sustained sensory neuron activation.

Methods: Dissociated rat dorsal root ganglion neurons were cultured on coated glass-bottom petri dishes and imaged on days 7–14. Cells were loaded with Calbryte 520 AM prior to live calcium imaging. For time-course studies, samples were either imaged immediately or incubated at 37 °C for defined durations up to one month. Curcumin release was quantified using gel permeation chromatography and UV/vis spectroscopy.

Nanodroplets were made in ultrapure water and generated using a cup-horn sonicator. They were subsequently thermally converted to microbubbles at 60°C. Characterization was performed by microscopy, dynamic light scattering (DLS), and ultrasound imaging.

Results: The curcumin monomer induced robust neuronal activation as measured by calcium imaging. This response was significantly attenuated by pre-treatment with the antagonist HC-030031. Notably, poly(curcumin-co-PEG) formulations sustained activation for up to 7 days, whereas responses to monomeric curcumin were diminished within 24 hours. Furthermore, using DLS we observed peak size shifts of the droplet to bubble transition with our poly(curcumin-co-PEG).

Conclusions: Degradation studies of poly(curcumin-co-PEG) demonstrated the prolonged release of active curcumin over a period of one month. This led to prolonged activation of sensory neurons, highlighting its potential for long-term neuromodulation. Integration into microbubbles further enables coupling with ultrasound-based delivery strategies, offering a path toward non-invasive, spatiotemporally controlled, and sustained neuromodulation.

Naturalistic control of upper-limb prostheses is often limited by the signal degradation and variability found in residual muscle electromyograms (EMG). While fusing cortical electroencephalograms (EEG) and EMG has shown promise by providing additional motor planning information, current methods might not be well suited to capture the directed flow of information from motor commands to muscle activation. To address this, we applied local Transfer Entropy (TE), a time-resolved non-linear measure of directed information transfer, to quantify corticomuscular communication. As an initial validation in an intact-limb population, we utilized a publicly available dataset of 40 subjects performing standardized reaching tasks. We measured and compared information transfer from two distinct cortical features, low-frequency EEG phase and beta-band-limited EEG, to EMG amplitude across periods of inactivity and muscle contraction. Phase-derived TE demonstrated significantly higher temporal alignment to muscle activation and deactivation, as evidenced by greater state separation than beta- band-derived TE (mean Wasserstein metric 0.014 ± 0.007 vs. 0.009 ± 0.003; Wilcoxon signed-rank test, p < 0.001) in 20 of 25 analyzed subjects. In addition, beta-band-derived TE frequently exhibited “false coupling,” information transfer during periods of inactivity, suggesting less specificity to motor intent. The results suggest that cortical phase may serve as a more temporally precise proxy for descending motor communication than traditional band-limited signals. Moreover, the proposed transfer entropy framework offers a robust approach to isolate motor intent from background neural noise.

Introduction: The autonomic nervous system plays a critical role in regulating immune responses to infection and injury. The vagus nerve is a key component of this neuro-immune interface, modulating responses via the inflammatory reflex. Vagus nerve stimulation (VNS) has emerged as a potential therapeutic strategy for sepsis and other inflammatory disorders, however, conventional electrical VNS lacks fiber-type specificity and requires surgical implantation. In this study, we employ red-light optogenetics using red-activatable channelrhodopsin (ReaChR) as a precise, non-invasive approach to selectively activate cholinergic neural circuits and investigate their role in regulating inflammation.

Methods: To selectively target cholinergic efferent and sensory afferent vagal fibers, ReaChR was expressed in choline acetyltransferase (ChAT)-cre and transient receptor potential vanilloid (TRPV1)-cre transgenic mice. Acute inflammation was induced via intraperitoneal lipopolysaccharide (LPS, 0.3 mg/kg) injection 24 hours after optogenetic stimulation (635 nm light) at the cervical vagus nerve. TNFα levels were measured in the spleen and serum by ELISA. Additionally, to activate vagal efferent pathways at their brainstem origin, a separate group of ChAT-cre mice were injected with AAV-flex-ReaChR-citrine virus into the dorsal motor nucleus (DMN) and stimulated non-invasively using red-light (635 nm). In all experiments, heart rate and breathing rate changes were also recorded to assess physiological effects of optogenetic stimulation.

Results:
ChAT+ vagal fiber stimulation significantly reduced splenic TNFα levels (ChAT-ReaChR: 1.27 ± 1.44 pg/µg vs. control: 4.26 ± 0.97 pg/µg, p < 0.001) but did not lower systemic cytokine concentrations (p = 0.81). Stimulation of TRPV1+ vagal fibers failed to modulate cytokine levels in either spleen or serum. Both ChAT-ReaChR and TRPV1-ReaChR stimulation induced bradycardia (~20% baseline heart rate decrease), with prolonged bradycardia effects observed in ChAT-ReaChR mice. To specifically stimulate the brainstem cholinergic neurons, we used ChAT-cre mice injected with AAV-flex-ReaChR-citrine virus into the left DMN. Non-invasive stimulation of the DMN using 635 nm red light in virally-injected mice reduced systemic TNFα levels in the LPS model of endotoxemia (stimulated: 1377 ± 233 pg/mL vs. control: 2396 ± 282 pg/mL, p<0.01).

Conclusions:
These findings demonstrate the feasibility of ReaChR-based optogenetics as a non-invasive and cell-type-specific approach to activate the inflammatory reflex. Red-shifted optogenetics minimizes surgical risks while providing high spatiotemporal control of specific neuronal circuits. By precisely targeting cholinergic neural circuits to engage the inflammatory reflex, this strategy has implications for developing bioelectronic therapies to treat sepsis, trauma, and critical illness.

Airway obstructions rapidly engage brainstem reflexes that maintain ventilation while recruiting higher-order forebrain circuits that enable conscious awareness of breathing and adaptive respiratory responses. However, how neural encoding across these forebrain circuits changes during disease-relevant respiratory challenges remains unknown. Here, we used intracranial EEG (iEEG) in nine patients undergoing epilepsy monitoring to characterize neural responses to brief inspiratory occlusions during normal breathing and during increased respiratory drive induced by flow-resistive loading and hypercapnia. During normal breathing, inspiratory occlusions evoked robust, phase- and frequency-specific neural responses across cortical and subcortical regions involved in respiratory interoception and motor control, including the insula, orbitofrontal cortex, amygdala, and hippocampus. By contrast, these responses were markedly attenuated during respiratory challenge. The attenuation effect scaled with occlusion magnitude, particularly in orbitofrontal cortex, depended on preceding inspiratory amplitude, and was observed during both mechanical loading and hypercapnia. The amygdala exhibited distinct challenge-specific dynamics, with beta suppression during mechanical loading and gamma enhancement during hypercapnia. These findings provide the first intracranial characterization of human forebrain responses to airway occlusions under disease-relevant respiratory challenge, revealing that respiratory drive dynamically gates the encoding of airway perturbations. These findings identify forebrain circuits as candidate targets for neuromodulation to restore adaptive respiratory compensation in respiratory disease.

How does the human brain detect disruptions in breathing? Animal models have established the brainstem circuits controlling respiration, but how forebrain circuits detect and respond to breathing disruptions remains unclear. We recorded intracranial EEG (iEEG) from 12 patients with epilepsy during brief, randomly timed inspiratory occlusions.
Results: Occlusions rapidly recruited insular, sensorimotor, and orbitofrontal cortices, producing robust high-frequency activity (HFA) within 1 s of inhalation onset. Occlusions also reorganized functional connectivity across this network, increasing coherence particularly between anterior and posterior insula (AIC–PIC), lateral orbitofrontal cortex and PIC (lOFC–PIC), and AIC and precentral cortex (AIC–PrC), while decreasing PIC–PrC coherence. Time-lagged analyses revealed information flow from PIC toward PrC and lOFC and implicated AIC in compensatory motor responses to occlusion. Finally, occlusions evoked transient pupil dilation, providing a potential noninvasive marker of respiratory constraint.
Conclusion: These findings identify a human cortical interoceptive network that detects airway occlusions and coordinates compensatory responses, providing neural and pupillary markers of the cortical drive to breathe.

Introduction
Vagus nerve stimulation (VNS) confers therapeutic impact in many disease conditions, and the attenuation of inflammatory mediators is a central aspect of this treatment approach. Electrode-based VNS stimulates pathways within the nerve indiscriminately, however, activating pertinent circuitry along with non-targeted pathways. The lack of specificity prevents the testing of vagal axon subsets to inform better circuit level characterization. Greater specificity in modulating vagal pathways may enable better informed circuit targeting and improve therapeutic impact across variable disease conditions. Optogenetic tools provide the ability to target stimulation to specific genetically-defined pathways. These methods may be used to target subgroups of vagal fibers to optimize neuroinflammatory control.

Methods
A vagus nerve cuff was designed and micro-scale 3D-printing parameters were developed, to fabricate a soft silicone, biocompatible nerve interface with printing resolution capable of accommodating miniaturized components. Optogenetic vagal stimulation was applied in mice with acute LPS-induced endotoxemia. Cholinergic (efferent) and glutamatergic (afferent) axonal subsets were targeted in separate groups, and the attenuation of pro-inflammatory and anti-inflammatory cytokines were quantified in response to these selective stimulation profiles.

Results
The 3D-fabricated nerve cuff and tethered micro-LED provides a miniaturized vagus nerve interface for optogenetic stimulation of defined vagal fibers. In vivo functionality is demonstrated through measurement of heart-rate response in ChAT-ChR2 mice. In LPS-treated mice, cholinergic vagal stimulation reduced serum levels of IL-6 and IL -1β relative to sham stimulation, while glutamatergic stimulation did not reduce cytokine level in this experiment.

Conclusions
We developed a micro-scale 3D-fabricated vagus nerve cuff in a soft silicone material for chronically implantable optical stimulation of defined vagal fibers. These methods enable selective optogenetic stimulation of targeted circuits. Selective stimulation of efferent and afferent fibers in a model of acute endotoxemia demonstrated an attenuation of IL-6 and IL-1β in efferent but not afferent pathways.

Brain-computer interfaces (BCIs) could enable spinal cord injury and stroke patients to regain autonomy by restoring movement and speech. However, BCIs rely on high-quality, time-stable neural recordings. This requires precise placement of electrodes – often intracortically – without causing tissue damage, bleeding, or a strong immune response. Our group has developed subcellular (<10um) carbon fiber electrodes (CFEs). To fabricate CFEs, first, carbon fibers (CFs) epoxied to a micromachined substrate are parylene coated and cut to a length suitable for recording Layer IV/V signals in the brain (1-1.5mm). CFs are sharpened either with an acid etch or flame. Finally, the CF tips are plated with platinum iridium or PEDOT:pTS (Richie et al., 2021 & 2024).

CFEs can be used in a wide range of neural applications. Our group has created 16-channel CFEs robust enough to penetrate vagal nerve epineurium, allowing for the first multi-channel action potential recordings in a sub-millimeter autonomic nerve. SNRs reached as high as 8.3 (Jiman et al., 2020). CF’s flexibility and biocompatibility also make them a strong candidate for chronic brain recording. We successfully implanted 16-channel CFEs into rat cortex with minimal immune response or disruption to neuronal density after 12 weeks (Letner et al., 2023). We have successfully extended CFE implantation into large mammals. In n=4 sheep, 10/10 100-channel CF arrays inserted without the use of any assistive tool. In one sheep, after 6 months of implantation, a 16-channel CF array implant site showed preservation of neuronal density and minimal astrocytic response. Furthermore, we have created a brain model with the same elastic modulus as brain tissue using polyvinyl chloride (PVC). In n=9 PVC insertions, CF arrays inserted with half the force of Utah Arrays – often considered a gold standard – with the same number of shanks (4.07mN and 8.29mN respectively; p < 0.001)(Lam, Copenhaver et al., SfN 2024). Future work will precisely quantify the sharpening and insertion parameters necessary to minimize damage. Broadly, CFEs present a step towards intracortical electrodes capable of providing high-quality, chronically stable signals for use in BCIs.

Background: The therapeutic potential of trigeminal nerve stimulation (TNS) for neuropsychiatric disorders is untapped due to the inability to non-invasively probe distinct trigeminal branches in preclinical models. The invasive and minimally invasive nature of current methods, such as subcutaneous and percutaneous approaches, precludes the chronic, fine-tuned studies needed to optimize therapies and discover mechanisms, thereby stalling progress. To overcome this critical barrier, we engineered and validated the eNoseBud: a first-in-class, non-invasive device fabricated from biocompatible resin and designed to unlock the complexities of multi-branch TNS in rodents.

Methods: Guided by 3D reconstructions of rodent neuroanatomy, the eNoseBud was engineered to deliver discrete or combined transcutaneous stimulation to the trigeminal V1 (via the anterior ethmoidal nasal innervation) and V2 (via the nasal whisker barrel) branches. Its electromechanical fidelity was rigorously validated against invasive percutaneous electrodes, assessing key properties like signal agreement, temporal synchrony, and electrode impedance. In vivo, its capacity to differentially modulate brain activity was confirmed with cerebral blood flow imaging. We then deployed the device for daily intermittent stimulation (30 min/day) in a 4-week chronic unpredictable mild stress (CUMS) rat model, assessing anxiety with the Elevated Plus Maze and behavioral despair with the Porsolt Forced Swim test.

Results: The eNoseBud demonstrated electromechanical performance equivalent to invasive methods while completely eliminating surgery and enabling facile, stable chronic application without signal degradation or nerve habituation. As hypothesized, targeted V1, V2, and V1+V2 stimulation produced distinct, predictable patterns of regional brain activation, confirming the device’s precision. In the CUMS model, chronic daily stimulation delivered a profound therapeutic effect, not only reversing established behavioral deficits but normalizing anxiety- and depression-like behaviors to healthy control levels, a dramatic ~70-80% symptom reduction. This powerful behavioral rescue directly correlated with the normalization of key physiological stress markers.

Conclusions: The eNoseBud provides the missing link in translational TNS research, delivering a robust platform to mechanistically dissect and optimize neuromodulation therapies. It is a paradigm shift, enabling complex chronic stimulation studies previously impossible. These findings validate a critical research tool and provide a blueprint for designing next-generation, multi-branch clinical stimulators, accelerating bench-to-bedside translation for debilitating neuropsychiatric disorders.

Bridging the gap between neural interface technology and clinical translation requires materials combining mechanical compliance, microfabrication compatibility, and long-term stability inside the body. Rigid substrates offer high resolution but cause stronger foreign body responses, while polymers typically create polymer-polymer interfaces that are prone to delamination. Here, we introduce thermoplastic polyurethane (TPU) as substrate and encapsulation for chronic neural interfaces, demonstrating its feasibility across a range of prototypes, including a bimodal, cuff-shaped device that could potentially combine ultrasound and electrical neuromodulation with simultaneous electrical recording. TPU has decades of clinical use in long-term implants and is available in medical-grade form. It is considerably softer than other thermoplastic materials (liquid crystal polymer (LCP)), and polymers (polyimide and parylene-C), approaching the Young’s modulus of silicone elastomers and soft neural tissue. Being thermoplastic, TPU merges with itself under heat and pressure, removing the polymer-polymer interface.
Simulated and measured acoustic transmission through TPU and other polymers between 1 and 7 MHz exceeded 94% for all materials. Optical transmittance through TPU films stayed above 70% at 470 nm and was unchanged after 9.5 months of implantation, confirming its suitability for both acoustic and optical applications.
Using a mask-free micromirror digital imaging photolithography process, we achieved a resolution down to 8 µm lines and spaces for 3 µm-thick electroplated gold on medical-grade TPU substrates, proving its substrate capabilities for high-resolution metallization. As an example application, bimodal cuff prototypes were fabricated on TPU substrates with gold electrode arrays integrated with flexible capacitive micromachined ultrasonic transducers (CMUTs), assembled with liquid TPU underfill and a final TPU encapsulation layer. Acoustic characterization showed that resonance behavior was largely preserved, though membrane displacement and output pressure decreased, indicating the need for tailored driving signals for encapsulated CMUTs.
Flip-chip bonding further embedded chips of varying geometries into additional TPU-based prototypes, the most demanding comprising over 300 connections to a single chip. The suitability of TPU for chronic implants was evaluated via in vitro cytotoxicity testing on differentiated SH-SY5Y neuroblastoma cells and in vivo evaluation in rats. Together, these results establish TPU as a versatile, translation-ready platform for dense, high-resolution, soft multimodal neural interfaces.

Introduction: Current vagus nerve stimulation (VNS) devices activate the nerve non-selectively, engaging fibers independently of fiber function (sensory or motor), diameter (large or small) and myelination (myelinated and unmyelinated). Lack of selectivity may cause off-target effects and limit the delivered VNS dose, resulting in reduced efficacy. To design more selective VNS approaches, knowledge of the counts of fiber types at different levels along the human vagus nerve is critical. However, the fiber composition of the human vagus along its length remains poorly characterized in large human cohorts.
Methods: Multiple transverse sections were obtained within each of four anatomical regions (upper cervical, mid-cervical, lower cervical, and thoracic), yielding 252 sections across 25 nerves from 13 cadaveric donors (6F/7M; age 73–97 years). Sections were immunolabeled against neurofilament (NF), myelin basic protein (MBP), choline acetyltransferase (ChAT), and tyrosine hydroxylase (TH) and imaged with 100x magnification. Fiber segmentation was performed using a custom-built machine-learning IHC pipeline. Fibers were classified as myelinated motor (MM; NF+/MBP+/ChAT+/TH−), myelinated sensory (MS; NF+/MBP+/ChAT−/TH−), TH-positive sympathetic fibers (NF+/MBP−/ChAT−/TH+), or unmyelinated sensory (US; NF+/MBP−/ChAT−/TH−).
Results: Across all nerves and sections, in a total of 2,965 individual fascicles, the mean fiber count per section were 85,678 ± 24,708 at cervical levels and 69,580 ± 27,680 at thoracic levels. MM fibers comprise 1.4% of total fibers at cervical levels and 0.3% at thoracic levels. MS fibers comprise 33% of total fibers at upper cervical levels, 30% at mid-cervical and 26% at lower cervical levels. In the thoracic trunk, MS fibers comprise 19% of total fibers. US fibers constituted the predominant fiber population, comprising 65% of total fibers at cervical levels and 74% at thoracic levels. Sympathetic fibers were consistently present across all levels, representing 3–8% of total fibers, with considerable inter-donor variability.
Conclusions: This is a first version of a detailed, multi-level neurochemical map of the human vagus nerve. From rostral to caudal levels, we document progressive decline of MM and MS fibers alongside relative enrichment of US fibers. These findings have direct implications for the design of more selective VNS devices.

Problem: Delirium, an acute confusional state associated with illness or injury, affects >30% of hospitalized patients and is associated with increased morbidity, mortality, and institutionalization. Emerging evidence suggests that inflammation is a contributing mechanism in delirium. The vagus nerve is a key regulator of inflammatory signaling via cholinergic pathways and modulation of central noradrenergic systems. Vagal neuromodulation, including transauricular vagus nerve stimulation (taVNS), represents a potential therapeutic approach for delirium. Neuromodulation research in patients with active delirium presents numerous challenges. Delirious patients may be critically ill and inadvertently remove medical devices. While there are limited data on taVNS to prevent delirium, the use of taVNS for treatment of established delirium remains underexplored. We hypothesized that taVNS could feasibly and safely be delivered to patients with active delirium.

Methods: Our single-site study enrolled hospitalized adults who screened positive for delirium. Participants received taVNS (25 Hz, 500 µs, 30 s on / 30 s off) to the left tragus for 30 minutes twice daily for up to 7 days. Stimulation intensity was determined at the first session (0.5-5.0 mA).

Results: We enrolled ten of a planned ten patients, four men and six women, ages 62-93. 54 taVNS sessions were delivered, with a median of four sessions per participant (IQR, 4–6; range, 1–14). All 10 participants completed every stimulation session prescribed while eligible. Six participants met the prespecified delirium-resolution criterion of being delirium-free for 24 hours. One participant was withdrawn after intubation, and one completed the maximum number of sessions without delirium resolution. Two participants had improvement in delirium severity, as measured by validated delirium score, but did not meet the prespecified criterion of being delirium-free for 24 hours. Stimulation was well tolerated with no device-related adverse events.

Conclusion: TaVNS can feasibly be delivered to hospitalized patients with delirium without device-related adverse events. This small open-label feasibility study is limited by clinical heterogeneity and the lack of blinding or sham. Given these limitations, delirium improvement cannot be attributed to taVNS. We plan to conduct larger randomized, sham-controlled, blinded studies to determine the efficacy of taVNS as a treatment for active delirium.

Background: Pulmonary endothelial glycocalyx injury is a critical driver of multi-organ failure following hemorrhagic shock (HS). This fragile barrier’s breakdown initiates a vicious cycle of inflammation, oxidative stress, and fluid leakage that standard resuscitation often fails to halt. While activating the master antioxidant regulator, Nuclear factor erythroid 2-like 2 (NRF2), is a highly promising strategy, conventional pharmacological activators risk off-target effects and potential cytotoxicity in high-stress states. We hypothesized that a novel, non-pharmacological approach, activating the body’s potent endogenous oxygen conserving reflex (OCR) via trigeminal nerve stimulation (TNS), could safely engage NRF2 to protect the lung from the catastrophic damage induced by HS.

Methods: To rigorously test this hypothesis, we used clinically relevant, pressure-controlled models of severe HS. Efficacy was confirmed across species in both rats (with long-term 10-day survival, with and without fluid resuscitation) and pigs (with acute 6-hour survival without resuscitation). Animals received intermittent TNS for 30 minutes during the shock period. To define the precise mechanism of action, we used specific pharmacological antagonists for NRF2 and its upstream mediators. We then performed comprehensive analyses of lung tissue, quantifying NRF2’s nuclear translocation, phosphorylation status, and the expression of its downstream antioxidant enzymes, alongside key markers of glycocalyx integrity, inflammation, and overall lung injury.

Results: TNS-induced OCR conferred profound, cross-species protection against HS. This was driven by a robust, p62-mediated activation of the pulmonary NRF2 pathway, which preserved glycocalyx structure, prevented vascular leakage, and dramatically reduced pulmonary edema. Critically, in the rat model, this definitive organ protection translated directly into a significant survival advantage and superior long-term neurobehavioral outcomes. The mechanism’s unique robustness, involving enhanced NRF2 protein release, stabilization, and phosphorylation, makes it exceptionally effective in the extreme stress environment of HS.

Conclusions: These findings establish trigeminal activation of the OCR as a transformative, non-pharmacological countermeasure against organ failure in HS. By demonstrating that a simple, non-invasive stimulation can mobilize the body’s own powerful defense systems with robust, cross-species efficacy, this work introduces a new paradigm for critical care. This approach represents a readily translatable platform therapy for HS and other critical conditions where conventional therapies may fail.

Introduction: Uncontrolled hemorrhage is the leading cause of preventable trauma deaths in the United States. Standard interventions including tourniquets and hemostatic dressings are often unavailable or insufficient in prehospital settings. Electrical stimulation of the vagus nerve (VNS) promotes faster clotting to minimize traumatic hemorrhage in mice, relying on a pathway that involves ChAT+ T-lymphocytes in the spleen and ⍺7 nicotinic acetylcholine receptors (⍺7nAChR) on circulating platelets. VNS is FDA approved for treating drug-resistant rheumatoid arthritis, epilepsy, and depression, and is currently undergoing clinical trials for treating hemorrhage. Trigeminal nerve stimulation (TNS) has also shown potential to reduce seizures in drug-resistant epilepsy. Because trigeminal afferents project to brainstem autonomic nuclei that overlap with vagal circuitry, TNS may activate neuroimmune pathways relevant to hemostasis. Using a standardized murine tail transection arterial hemorrhage model, we explored whether TNS decreases traumatic bleeding in mice.

Methods: Male 8-12 week old C57BL6/j mice were anesthetized with isoflurane and placed in the right lateral decubitus position. Needle stimulation electrodes (14×0.38mm) were placed in the left masseter muscle (+) and subcutaneously at the midline 1 mm caudal to the eye to target the 3 branches of the left trigeminal nerve. Mice received TNS (30Hz, 5 volts, 2 sec, 10 min) or sham stimulation. Following stimulation, the tail was warmed in water (37ºC, 5 min), transected 10 mm from the tip, and bled into water (37ºC, 5 min) until hemorrhage stopped for at least 10 s. Duration of hemorrhage was recorded as bleeding time.

Results: Compared with sham stimulation, TNS significantly reduces bleeding time in mice following arterial injury (Sham=60.5±8.7s vs TNS=39.25±3.7s, mean±SEM, n=12, p<0.05).

Conclusions: TNS decreases duration of bleeding in a murine model of traumatic arterial hemorrhage. TNS warrants additional study for the management of clinical hemorrhage.

Ultrasound neuromodulation shows promise as a non-invasive, spatially precise therapy for a wide range of applications but remains poorly understood with respect to its effects on specific cell types. Previous data suggests that suprachiasmatic nucleus (SCN) cells express mechanically sensitive ion channels (i.e., those affected by ultrasound pulses). SCN slices can also be functionally isolated, cultured, and monitored, making it an ideal platform for functionally coupling an ultrasound transducer and conducting dose response testing. This study demonstrates the use of SCN slice cultures to screen the effects of low intensity ultrasound pulses on nerve activity and circadian gene expression. SCN tissue was harvested from knock-in PER2:Luciferase mice and paired with an optical luminometry system to measure PER2 gene expression. SCN slices were cultured and split into four stimulation groups (i.e. treated with low (5Hz pulse repetition, 10ms burst length, 1 minute total stimulation), medium (10Hz pulse repetition, 20ms pulse length, 2 minute total stimulation) , high (10Hz pulse repetition, 20ms pulse length, 5 minute total stimulation), and sham ultrasound doses). The SonoShield Defender LIPUS transducer (970 MHz) was coupled to cultured tissue via overlaying media, and the transducer was manually aligned to the SCN using precision micromanipulators. Tissue samples were placed in the Lumicycle, cultured, and monitored following sham or stimulation for 12 days, and PER2 expression amplitude and period were then analyzed in the system’s software. Preliminary data demonstrates an ultrasound dose trend on peak-to-peak circadian (PER2 expression) period. Compared to sham (with a PER2 expression period of 24.54 hours), the highest ultrasound dose shifted the peak period to 18.115 hours (for the circadian cycle immediately following stimulation) with lower doses resulting in less significant changes in period length. Stimulation at higher ultrasound doses also significantly weakened PER2 expression amplitude compared to low stimulation and sham, requiring further investigation into the action mechanisms on ultrasound stimulated SCN tissue. The data indicates that ultrasound neuromodulation of the SCN may result in a phase shift of circadian gene expression. Further study is required to validate this effect, investigate a detailed dose response, and determine the effect duration on circadian gene expression.

Generally, our mental state influences feeding patterns, and what we eat also influences our mood. Psychological states such as stress and anxiety can exacerbate gastrointestinal sensitivities. This class of maladies are known as disorders of gut brain interaction and are underscored by dysregulation of bi-directional gut–brain communication. A major communication highway between the gut and the brain is the vagus nerve. The vagus nerve, or Xth cranial nerve, innervates the entire gastrointestinal tract. Recent findings have characterized a vagal afferent (gut-to-brain) synaptic signaling circuit between sensory cells of the gut, known as neuropod cells, and vagal nodose ganglia neurons. However, how the functional circuitry of how the brain modulates gut sensing through vagal efferent (brain-to-gut) signaling remains poorly understood. To address this gap, we are using a combination of techniques. First, we characterized the vagal to neuropod cell connection in vitro with immunohistochemistry of synaptic structure proteins. Next, we used in vitro calcium imaging of neuropod cells to determine their sensitivity and response to efferent neurotransmitters. Finally, we measured vagal activity in vivo with whole nerve electrophysiology while stimulating the proximal small intestine with nutrients or optogenetic modulation. We then performed a selective efferent vagotomy, allowing us to isolate and quantify the individual afferent and efferent contributions. We found that neuropod cells have the machinery necessary to receive synaptic input, that they preferentially respond to the vagal efferent neurotransmitter acetylcholine, and that the total in vivo vagal response to luminal sugar is composed of an early afferent followed by a later efferent response. By mapping the bidirectional circuits of gut-brain communication, we aim to illuminate the neural underpinnings of the diet-mood relationship.

Introduction:
Heart failure (HF) is characterised by a sustained autonomic imbalance, with heightened sympathetic activity and withdrawal of vagal tone contributing to disease progression. Vagal nerve stimulation (VNS) has been proposed as a bioelectronic therapy to restore this imbalance, with encouraging results in experimental models. However, its clinical efficacy remains uncertain.

Methods:
We performed a structured review of the literature examining both preclinical and clinical evidence for VNS in HF. Key animal studies and major clinical trials, including CardioFit, ANTHEM-HF, NECTAR-HF, and INOVATE-HF, were analysed. Outcomes of interest included cardiac function, neurohormonal modulation, functional status, and survival, alongside differences in stimulation strategies and patient characteristics.

Results:
Across preclinical models, VNS consistently improved cardiac function, reduced ventricular remodelling, and enhanced survival, supporting a clear mechanistic rationale. In contrast, clinical findings have been more heterogeneous. Early-phase studies suggested improvements in left ventricular function, exercise capacity, and quality of life. However, larger randomised trials did not demonstrate significant reductions in mortality or heart failure hospitalisation. Notably, response to therapy appeared to vary with stimulation parameters, including frequency, intensity, and laterality, as well as underlying patient physiology, particularly autonomic profile and baroreflex sensitivity.

Conclusions:
VNS represents a biologically plausible and well-tolerated therapeutic approach in HF, but a clear gap remains between experimental efficacy and clinical outcomes. This disconnect likely reflects suboptimal patient selection and non-standardised stimulation protocols rather than failure of the underlying concept. Future progress will depend on more targeted neuromodulation strategies, adaptive stimulation systems, and improved phenotyping to identify patients most likely to benefit.

Introduction: Colorectal cancer (CRC) is the third most prevalent cancer globally, with incidence rising rapidly in younger adults. In cancer neuroscience, there is growing recognition that peripheral nerves are active participants in the tumor microenvironment. To examine the role of the vagus nerve in these interactions, we explored cellular signaling between vagal sensory neurons (VSNs) and CRC cells in vitro co-cultures.
Method: VSNs were isolated from the vagal ganglia of Vglut2-tdTomato transgenic mice and co-cultured with either the murine CRC cell line (MC38) or colonic epithelial cells. Tumor cell proliferation was quantified in MC38 cells cultured alone, direct co-cultures with VSNs, or VSN-conditioned media. VSN morphology was characterized by confocal imaging with semi-automated quantification of neurites. Invasive behavior was assessed by live-cell time-lapse imaging and Transwell invasion assays.
Results: When VSNs and MC38s were co-cultured simultaneously, CRC cells show a significantly faster proliferation rate that is mediated by cell-cell contact and paracrine signaling through neuropeptides secreted by VSNs. Calcitonin gene-related peptide (CGRP) had the most significant effect on the proliferation rate of MC38 cells. Quantification of neurite outgrowth of VSNs in these MC38 co-cultures show significantly higher neurite outgrowth compared to those co-cultured with epithelial cells (p-value 0.012, Welch’s t-test) and with increased branching (p-value 0.0155, Welch’s t-test). The CGRP antagonist rimegepant attenuated MC38 proliferation without compromising VSN neurite outgrowth. In parallel studies, we also evaluated the invasive potential of CRC cells in the presence or absence of VSNs. Quantification of cell speed also showed CRC cells showed increased overall cell speed in the presence of VSNs (p-value < 0.0001, 2-way ANOVA), suggesting higher invasive potential. In Transwell invasion assays, we observed MC38 cells show significantly higher invasive capacity when cultured with VSNs, compared to control colonic epithelial cells (p-value < 0.0001, 2-way ANOVA).
Conclusions: Together, these findings position VSNs as drivers of colorectal tumor biology and identify the vagal-CGRP axis as a potential target for neuromodulation-based cancer therapy, either through traditional pharmacological routes or bioelectronic interventions. Repurposing existing drugs, such as rimegepant, to target specific neuropeptide pathways in cancer is a novel therapeutic pathway that warrants further exploration.

Rheumatoid arthritis is one of the most debilitating chronic human diseases and affects 1% of the world’s population. Although there has been a marked improvement in the treatment of rheumatoid arthritis (RA), there are still many RA patients who do not achieve disease remission. Therefore, there is still a need for new treatments. Recently, neuromodulation has emerged as a potential treatment approach. However, the role of brain networks controlling the onset and progression of inflammatory arthritis is completely understudied. Here, we use activity-dependent cell labeling for mapping neuronal networks in the brain that respond to onset and progression of arthritis.
We used the TRAP2 mouse model, which has inducible Cre expression linked to the immediate early gene c-fos, backcrossed to the Ai14 mouse strain that expresses tdTomato fluorescence in the presence of Cre protein. We utilized the collagen antibody-induced arthritis (CAIA) model, which consists of an injection of antibodies against collagen II, followed by an injection of lipopolysaccharide (LPS) to provide an immune trigger that promotes paw inflammation. Saline-injected animals were used as controls. Animals develop peak disease (arthritis score, paw swelling) by day 7 with significant increase in systemic and local cytokines (e.g., IL-1β, TNF-α, IL-6, KC-GRO). We then studied changes in the brain using serial two-photon tomography microscopy and observed significantly increased neuronal activity in different brain regions in CAIA mice: cortical areas including somatosensory and motor cortex; hippocampal regions including entorhinal cortex and dentate gyrus; striatum and pallidum including the medial amygdalar nucleus and BNST; thalamic regions including the PVT; hypothalamic areas including the posterior hypothalamic nucleus; midbrain regions; and medullary areas including the spinal trigeminal nucleus (caudal part). In conclusion, arthritis induces increased neuronal activation throughout the brain. Dense activation seen in the cortex, thalamus, hypothalamus, and midbrain hint at specific brain regions being activated during the inflammatory response.

Introduction: The nodose ganglion (NG) is the main sensory ganglion of the vagus nerve, containing vagal sensory neuron (VSN) cell bodies that convey interoceptive information from thoracic and abdominal organs to the brain. Human NG microanatomy remains poorly characterized, limiting understanding of sensory vagal organization and potential opportunities for neuromodulation beyond the vagal trunk. We developed a light-sheet microscopy (LSM) and computational reconstruction workflow to characterize human NG cytoarchitecture in three dimensions.
Methods: A human NG was isolated from a previously dissected vagus nerve specimen, using the superior laryngeal nerve branch as the inferior anatomical landmark. The specimen was cleared using a SHANEL-based protocol and immunostained for neurofilament, beta-III tubulin, myelin basic protein, and tyrosine hydroxylase (TH), followed by whole-volume LSM imaging. A 1.63 mm³ region was analyzed. The TH channel was segmented in Ilastik using sequential classifiers separating cell body from axon and then intracellular/soma, boundary, and extracellular/background signals. Candidate soma centers were detected using smoothing, hysteresis thresholding, and connected-component labeling. Radial rays were cast across orthogonal planes to identify intracellular-to-boundary-to-background transitions. Transition points were grouped by centroid proximity using Leiden clustering, and soma diameter was estimated from centroid-to-outer-transition distances.
Results: We identified 1,293 putative neuronal cell bodies within the analyzed region, corresponding to an estimated density of 793.8 cells/mm³. Median outer diameter was 61.75 µm, with a mean of 64.76 ± 20.04 µm and a range of 16.25–148.78 µm. Equal-sized diameter tertiles were defined as small (≤53.80 µm), medium (>53.80 to ≤71.88 µm), and large (>71.88 µm).
Conclusions: Whole-mount LSM combined with computational segmentation enables quantitative three-dimensional analysis of human NG cytoarchitecture. This approach provides a framework for mapping the spatial organization and morphology of human VSNs and may establish anatomical ground truth for future studies of sensory vagal function and nodose-targeted neuromodulation.