Flexible graphene-based neurotechnology for high-precision deep brain mapping and neuromodulation in Parkinsonian rats

Abstract

Deep brain stimulation (DBS) is a neuroelectronic therapy for the treatment of a broad range of neurological disorders, including Parkinson's disease. Current DBS technologies face important limitations, such as large electrode size, invasiveness, and lack of adaptive therapy based on biomarker monitoring. In this study, we investigate the potential benefits of using nanoporous reduced graphene oxide (rGO) technology in DBS, by implanting a flexible high-density array of rGO microelectrodes (25 µm diameter) in the subthalamic nucleus (STN) of healthy and hemi-parkinsonian rats. We demonstrate that these microelectrodes record action potentials with a high signal-to-noise ratio, allowing the precise localization of the STN and the tracking of multiunit-based Parkinsonian biomarkers. The bidirectional capability to deliver high-density focal stimulation and to record high-fidelity signals unlocks the visualization of local neuromodulation of the multiunit biomarker. These findings demonstrate the potential of bidirectional high-resolution neural interfaces to investigate closed-loop DBS in preclinical models.

Fig. 1. Bidirectional graphene-based DBS technology.

a Illustration of microelectrode linear array at the tip of the device, with top and bottom polymeric layers. The device cross-section displays the layers of polyimide, gold, and reduced graphene oxide. b Photograph of the device, with a zoom of the tip and of a microelectrode. c Average voltage polarization (solid lines), with shadowed area for standard deviation (STD) (n = 8), of the DBS protocol, consisting of biphasic current pulses (100 µs pulse width) and increasing current levels. d Average voltage drop at the electrode-electrolyte interface with STD (n = 8). The dashed line represents the safe voltage limits for the anodic (blue) and cathodic (black) polarizations, from which the charge injection limit can be calculated (CIL = 2.3 mC/cm²). e Average impedance magnitude (blue) and phase (red) (n = 8). f Impedance magnitude at 1 Hz (red) and 1 kHz (black) of the microelectrodes for 10 different arrays; boxplots represent 25 and 75 percentile, with the median value line and whiskers for the STD (for each device n = 8). g Schematic of the in vivo set-up for bidirectional recording and stimulation. The flexible device is attached to a rigid shuttle with a dissolvable adhesive polymer. Once the insertion is completed, the flaps at the device’s end keep the device in place during the extraction of the shuttle. h Average dissolution time in agarose at 38 °C of the polyethylene glycol (PEG, n = 10) and polyvinyl alcohol (PVA, n = 10). i RMS of the voltage amplitude of the signal recorded (25 s, in the range 200–2000 Hz) in PBS before insertion and in the brain after reaching the STN. The plot displays the average values for the electrodes of 10 different devices (for each device n = 8).

Fig. 2. Recording capability of rGO microelectrodes and STN localization by high-density microelectrode array.

a Representative brain activity recording of one electrode located in the STN, filtered for different frequency bands, for in vivo (left) and post-mortem (right) conditions. b Left: Average power spectral density (PSD), with STD, of the signal recorded with the electrodes of one array (n = 8 electrodes), comparing in vivo (green) and post-mortem (black) conditions. Right: Average signal-to-noise ratio (SNR), with STD, calculated from the ratio between in vivo and post-mortem PSDs in the left panel. c Average spike-to-noise ratio (SNR_spike)over 100 s, for the electrodes located in the STN. Each dot represents one acute experiment (PD = 6 and control = 3 rats). The inset shows an example of the average spike shape of a PD rat (the shadow area represents the STD of the spikes recorded over 100 s). d Representation of brain activity recorded with the 8 electrodes of one array, also depicting an image of the brain histology superimposed with a schematic of a device in the STN. The two groups of traces correspond to the recordings of each of the electrodes of the array, filtered in the LFP (green) and MUA (red) range. The blue rectangle highlights the traces corresponding to electrodes in the STN. The color-coded spike rate quantifies the number of spikes for the different electrodes of the array. e Cross-correlation between channel 1, located in the STN, and each one of the other electrodes, quantified at lag 0 for both LFP (green) and spikes (red). f Left: Raster plot of the detected spike events over time for each electrode of the array shown in panel (d); Right: Color-coded spike rate histogram of the raster plot, in which spike events are quantified over 100 s. g The color-coded dots represent the average spike rate (in the recording period of 100 s) of the electrodes of the arrays implanted in 8 different animals. The color codes are normalized by the higher spike rate channel of each array. The blue rectangle represents the expected position of the STN.

Fig. 3. Electrophysiological biomarkers recorded in the STN.

a Tyrosine hydrolase (TH) expression in the striatum as a measure of the extent of dopaminergic cell depletion. The average fluorescence intensity is expressed as a percentage of the contralateral side (animals analyzed: PD = 10, control = 4). b Forelimb asymmetry test, quantifying the average asymmetric ratio in the use of the left and right paw (PD = 6, control = 3). c Apomorphine-circling test, quantifying the average number of rotations per minute after epimorphine injection (PD = 6, control = 3). d, e Exemplary recordings (MUA band, 200 Hz–2 kHz) of electrodes in the STN of the PD and control rats together with fluorescence images of brain tissue slices. f Average spike rate (over 100 s) with of one electrode in the STN of the two groups (PD = 6, control = 3). g Average number of spiking events, defined as bursts per second and tonic spikes per second, detected over 100 s (PD = 6, control = 3). h Interspike frequency distribution, normalized by the total number of counts in 100 s, calculated from recordings of a PD (blue) and a control rat (red). i Frequency peaks of the interspike frequency histograms for the 2 animal groups (PD = 6 in blue, control = 3 in red). The box plot represents the 25 and 75 interquartile. j Comparison of the average PSD with STD (shadow area) of the signal recorded in the LFP range (1–200 Hz) over 55 s for the two animal groups (PD = 6, control = 3). k Quantification of the total power over 100 s for each frequency band (delta 1–4 Hz, theta 4–8 Hz, alpha 8–12 Hz, beta 12–30 Hz, low gamma 30–70 Hz and high gamma 70–200 Hz), showing the median value with 25 and 75 interquartile whiskers (PD = 6, control = 3).

Fig. 4. Deep brain neuromodulation in the STN of Parkinsonian rats.

a Effect of microelectrode stimulation on the spike firing activity; the same microelectrode was used for stimulation and recording. The traces above correspond to the recordings (200 Hz–2 kHz) with a microelectrode in the STN of a PD animal, pre (blue trace) and post (green trace) stimulation. Below are shown corresponding time-dependent interspike frequency plots, pre and post-DBS; the green dashed line (at 100 Hz) indicates the division between the tonic and burst activity. b Histogram of normalized IF spike distribution calculated pre- and post-DBS in a PD animal; each histogram quantified over 100 s. c Average ratio (post- to pre-DBS) of different spiking-activity metrics: total spike rate, tonic spike rate, burst rate, and burst spike count (number of spikes per burst), calculated for the PD group (PD = 5). d Time evolution of the post-pre ratio of burst rates, depicting average (solid line) and STD (shadow color) for the PD group, quantified each 15 s. e Color-coded modulation factor of the burst activity (definition in main text), displaying the impact of the distance of the stimulating microelectrode to the STN. The dashed line indicates the threshold distance at which the modulation factor drops to 0.5. Each dot corresponds to the modulation factor measured (over 100 s?) in one electrode, for 5 different arrays (PD = 5). f Normalized spectrograms of LFP (1–200 Hz) activity of a PD rat, pre and post-stimulation. g Total PSD of recorded signal, quantified (55 s) in different LFP frequency bands. Data were calculated for the PD group (n = 6), showing the median value with the corresponding 25 and 75 interquartile error bars.