Mechanically-adaptive, resveratrol-eluting neural probes as a platform for improved single-unit recording performance and tissue integration

Introduction: Intracortical neural probes with recording microelectrodes are used as a part of brain-machine interface (BMI) systems to control assistive devices in individuals with spinal cord injury, limb loss, and neurodegenerative disorders. Unfortunately, neural recording metrics degrade shortly after implantation, reducing their capacity to provide BMIs with information necessary for precise control of assistive devices. A prevailing cause of failure is the neuroinflammatory response to the implant, characterized by activated microglia and reactive oxygen species surrounding the neural probe and neuronal degradation, increasing the distance between the recording sites and healthy, active neurons. Two contributors to the neuroinflammatory response are mechanical mismatch at the device-tissue interface and oxidative stress. Mechanical mismatch at the device-tissue interface imparts tissue strain during micromotion, resulting in inflammation and glial scar encapsulation. Oxidative stress can perpetuate the foreign body response and lead to microelectrode degradation. In this work, we developed mechanically-adaptive, resveratrol-eluting neural probes to simultaneously mitigate causes of neuroinflammation with a single device. Microfabrication processes were used to integrate eight recording channels on probe shanks that are initially sufficiently rigid to insert into the cortex without buckling, but then soften by orders of magnitude once implanted. The chronic recording performance and the neuroinflammatory response of the mechanically-adaptive, resveratrol-eluting neural probes were evaluated in vivo and compared to industry-standard silicon-based neural probes. We hypothesized that the mechanically-adaptive, resveratrol-eluting neural probes will show improved recording performance and reduced neuroinflammation when compared to silicon devices.

Materials and

Methods: The mechanically-adaptive substrate is a polymer-based nanocomposite comprising polyvinyl acetate, tunicate cellulose nanocrystals, and resveratrol. The nanocomposite is created by a dispersion, casting, and evaporation method, resulting in a resveratrol-loaded film. The film was then planarized in a Carver laboratory press to reduce surface roughness. Thin-film deposition and photolithography-based patterning techniques were implemented to integrate eight independent thin-film gold recording microelectrode channels per neural probe on the nanocomposite substrate10,11. Individual probes were laser-micromachined using a picosecond laser with a wavelength of 355 nm. Probes were then packaged for interfacing with electronics for neural recording. Electrochemical impedance spectroscopy (EIS) was used to ensure that electrode impedance at 1 kHz was in a suitable range for extracellular neural recording (100-1000 kΩ). Probes were then subjected to in vivo testing in Sprague Dawley rats. Rats were implanted with either a mechanically-adaptive, resveratrol-eluting (MARE) probe (experimental) or a commercial silicon probe (control) in the motor cortex following existing lab surgical procedures. Neural recording and in vivo EIS sessions took place after surgery and biweekly following implantation. Recording metrics, such as the number of channels recording single units, were determined by processing and analyzing the bandpass filtered (300-3000 Hz) for each recording channel and recording session. At either 4 or 12 weeks, rats underwent a cardiac perfusion and the brain was extracted and frozen. Brain tissue was then processed and a volume within a 500 µm-radius of the longitudinal probe axis was subjected to bulk gene expression analysis using neuroinflammation and oxidative stress gene panels.

Results, Conclusions, and Discussions: A representative MARE probe with eight individual recording electrodes is shown in Figure 1. Moisture-resistant Parylene C layers on the top and bottom surfaces serve as barrier and capping layers, and resveratrol can elute from the laser-cut sidewalls of the probe shank after implantation. When comparing MARE and silicon recording performance using the percentage of active channels – the percentage of channels recording at least 1 single unit – MARE probes showed improvement over silicon probes at subchronic (weeks 6-11) and chronic (week 12) timepoints, as shown in Figure 2. The results suggest that the resveratrol eluted by the MARE probes preserves the ability to record single units, leading to stable recording yield through the acute (1-5 weeks) and subchronic phases while the silicon-based probes show degrading recording behavior. In the chronic phase, the depletion of resveratrol from the MARE probes coincides with a decrease in active electrode yield. Additionally, MARE probes demonstrated higher electrode-electrolyte interface stability, as evidenced by electrochemical impedance spectroscopy measurements at 1 kHz across up to 12 weeks in vivo (Figure 3). Bulk gene expression analyses indicated that key markers of neuroinflammation decreased from the acute to chronic timepoint for tissue surrounding MARE probes, while the same markers were either stable or worsened over the same period for silicon-based probes. For example, the gene Gfap is expressed by activated astrocytes, and there is a significant decrease in expression from week 4 to week 12 for MARE probes, while there is a slight increase for silicon-based probes (Figure 4). The results suggest that the mechanically-adaptive MARE probes show improved tissue integration over time, while the rigid silicon-based probes continue to exacerbate neuroinflammation. This work establishes MARE probes as a platform for promoting tissue integration and stability at the biotic-abiotic interface, and improving neural recording performance. As the field pursues better therapeutic efficacy for patients with neuromusculoskeletal disorders, utilizing devices like MARE probes that mitigate the neuroinflammatory response to improve recording performance over time is critical for clinical implementation.

Acknowledgements (Optional): This work was supported by Merit Review Awards I01 RX003083 (Hess-Dunning/Capadona)
and I01 RX003950 (Hess-Dunning) from the United States Department of Veterans Affairs Rehabilitation Research and Development Service and the FY2021 National Defense Science and Engineering Graduate Fellowship Program.