For our electrode test, we selected male ChAT-Cre knock-in mutant mice (#006410, Jackson Laboratory, Bar Harbor ME, United States) and C57 mice that were 12 ± 1 weeks old. Breeding and rearing of these mice were conducted in accordance with the Animal Care and Use Committee’s protocols and institutional guidelines at Huazhong University of Science and Technology, Wuhan, China. The mice were group-housed with three to five per cage and subjected to a 12-h light–dark cycle, with the lights on at 8:00 am. The temperature and humidity of the mouse facility were maintained at a constant 21 ± 1°C and 50 ± 5%, respectively. Following electrode implantation, all mice were housed individually.

The fabrication of electrodes can be divided into the following process.

In order to facilitate the assembly and welding of electrode wires and optical fibers, we have developed an electrode positioning mold (Figures 1C,E) to streamline the process. This mold is composed of cascaded steel sheets with parallel cylindrical cavities that have been laser-drilled according to the anterior–posterior and medial-lateral coordinates of the targeted brain regions, which were determined based on the horizontal coordinates of mouse DMN as specified in Table 1. The depth of each cavity corresponds to the thickness of the stacked laser-drilled steel sheets, as determined from the vertical coordinates of mouse DMN in Table 1. The electrode positioning mold enables us to assemble and position the electrode wires and optical fibers with precision according to the relative positions of the brain regions, which can then be fastened to a custom-printed circuit board (PCB) (Figure 1C and Supplementary Figure S3). To optimize efficiency, we utilize a batch-production mold (Figure 1E and Supplementary Figure S4), which streamlines the electrode production process and helps to ensure consistency and precision across experiments.

To manually assemble the electrodes, the prepared microwires were threaded into the corresponding cylindrical cavity in the electrode positioning mold (Supplementary Figure S5A). After positioning all the microwires, solder paste was applied to the circuit board under the mask (Supplementary Figure S5B). The mask was then removed, and the connectors were placed on the pad while silver wires were inserted into the reference and ground channels. The microwires and connectors were reflow-welded (Supplementary Figures S5C,D). The mold and electrodes were then cleaned in an ultrasonic cleaner containing ethanol for 10 min to remove any residue from the scaling powder after welding. After air drying, optical fibers (Inper Co., Ltd., China) were threaded into the corresponding cylindrical cavity and bonded to the PCB using epoxy resin (9005, LEAFTOP, China). Once the UV curing was complete, solidified liquid photo solder resist (MECHANIC, China) was used to protect the optical fibers and insulate the soldering point (Supplementary Figure S5E).

In cases where microwire bundles were utilized instead of optical fibers for SUA recording, the microwire bundles were inserted into the cylindrical cavity designated for optical fibers and secured with epoxy resin. The uninsulated tips of the bundles were then folded and inserted into the corresponding welding hole before reflow-welding.

In the surgical procedure, ChAT-Cre mice were anesthetized using 1%–2% isoflurane (RWD Life Science Inc., China) to maintain a respiratory rate of 40–50 breaths per minute. The animals were head-fixed using a stereotaxic apparatus with a water-circulating heating system (RWD Life Science Inc., China), and the temperature was set at 37°C. Following cranial adjustment, a cranial drill with a 400-micron diameter bit was attached to the arm holder of the stereotaxic apparatus. Bregma was used as the reference for zero setting, and the drilling process was initiated gradually, following the coordinates of the targeted brain regions. The dura was removed, and four stainless screws were tapped into the skull and reinforced with dental cement (Figure 2B). In addition, two screws were placed onto the cerebellum as reference and ground. The electrode was mounted onto an electrode holder, fixed to the stereotaxic apparatus arm, and the microwires/fibers were aligned with their designated holes. The stereotaxic arms were gradually lowered until the longest microwire/fiber contacted the brain surface at the center of the hole, and the Z arm was then set at zero before further lowering the electrode to the desired depth. Gel foam was applied to all drilled holes. After the implantation, the opto-electrode was fixed to the skull using dental cement. The mice were caged individually to recover before recordings.

To genetically label the cholinergic neurons in the basal forebrain (BF), we injected a titer (0.1 μL, 3 × 10¹² genomic particles/ml) of the rAAV2/9-hSyn-DIO-ChR2-GFP virus into the brain region using a pump (UMP3-MICRO4, World Precision Instruments; 50 nL/min) with a flow rate of 50 nL/min, followed by a 15 min maintain period.

We conducted recordings of both electrophysiological and behavioral data using a 32-channel electrophysiology system from Plexon (United States). The signal was sampled at 1,000 Hz and 40,000 Hz, with a 0.05 Hz high-pass filter applied for the LFPs and individual spikes recording. To monitor head movement, we attached a 3-axis acceleration sensor from FDISYSTEMS (China) to the pre-amplifier. We also synchronized the video (from Plexon, United States) and acceleration at 30 frames and 20 Hz, respectively. All experiments were conducted in a noise-attenuated room with an environmental background noise of 32.2 ± 3.0 dB (mean ± SD).

To evoke neural activity, we delivered an 8 Hz flash stimulation above the homecage while concurrently recording electrophysiological data for further analysis.

In this study, data processing was carried out using internal Matlab scripts (release 2021b) and EEGLAB (Delorme and Makeig, 2004). Power spectral density estimation and time-frequency analysis were performed using the PWELCH and CWT functions, respectively. To prevent memory overflows, the data were segmented into periods during time-frequency estimation. Spike sorting was conducted using Offline Sorter, a commercial software from Plexon (United States). The firing rate, valley-to-peak time, and burst index of the recorded spike units were calculated and analyzed. The firing rate was determined by dividing the total firing number by the total duration of the behavioral state and was used for further analysis.

In this study, we conducted significance testing using paired two-tailed Student’s t-test and considered p-values less than 0.05 as statistically significant. The statistical analyses were performed using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA).

The implantation and fixation of microwires/optical fibers for multiple mouse brain regions presented significant challenges. To overcome these challenges, we developed a potential solution to fabricate an opto-electrode that would allow for the fixation of all microwires/optical fibers for multiple brain regions into a single module. We designed and developed a mold consisting of a series of stacked slices (Figure 1) that would enable the presetting of microwires/optical fibers in the 3D space of the target brain regions and connect them to a connector. The mold comprises 2D horizontal (X and Y) planes that form the 3D shape along the Z direction, creating cylindrical cavities at each designed location for the placement of the microwires/optical fibers (Figure 1C). We utilized PCBs to fix and connect the microwires/optical fibers (Figures 1C,E), in which the de-insulated parts of the microwires were soldered to the PCB pads (Figure 1B). The micro-nanofabrication quartz glass was utilized to fix the microwires (Figures 1A,D) due to its excellent high-temperature resistance and infrared transmittance. After assembling the opto-electrodes, UV-curable resin was used to insulate and protect the PCB. The resulting opto-electrode was a lightweight (0.28 g) and small space-size (10 mm length, 8 mm width, and 7 mm height) multi-channel opto-electrode that allowed for the synchronized recording and stimulation of distributed brain regions (Figure 2A). This electrode is easy to implant and carry by mice without difficulty (Figures 2B–D).

To assess the accuracy of our opto-electrode, we implanted the opto-electrodes in 16 DMN and 2 basal forebrain (BF) regions of mice. Subsequent histological analysis confirmed that the microwires were implanted in the central of the targets, and the optical fibers were implanted in the upper-edge of the targets, as illustrated in Figure 3. These results demonstrate the accuracy and reliability of our opto-electrode and implantation technique.

To evaluate the functional characteristics of our opto-electrode, we implanted it into three mice for an extended period of time. After 24 h of implantation, we observed steady electrical activity, and all electrode channels functioned correctly for at least 4 weeks without any observed movement difficulties in the animals. We conducted long-term recordings lasting over 8 h to assess the stability of the electrode. We demonstrated that the recording is relatively unaffected by movement noise, as shown in the recording sample of a freely moving mouse depicted in Figure 4A, where the stable spectra fluctuations match the head movement of the mouse during the lengthy recording session (Figure 4B). Additionally, we implanted the electrode into CA1, a region known to generate high-frequency activity (ripple oscillations) during cognitive activities (Buzsaki, 2015), and observed prominent ripple activities as shown in Figure 4C. To evaluate low-frequency activity recording, we used SSVEP (Xu et al., 2013; Yan et al., 2017) and observed a strong power response in response to low flash stimulus frequency (8 Hz) in the visual-related pathway (Figure 4D).

Single-Unit Activity (SUA) carries rich information associated with brain functionality (Remondes and Wilson, 2013; Tingley and Buzsaki, 2018). Therefore, we implanted another version of the electrode with microwire bundles to record the discharging of single neurons in BF and key DMN regions. Despite the large variance in channel discharges due to state changes, most of the recording channels captured single-unit activity, as indicated in Figure 5. BF activity is known to influence animal behavior, particularly parvalbumin-positive (PV⁺) neurons, with high activity promoting rest behavior in rodents (Lozano-Montes et al., 2020). Therefore, we isolated PV⁺ neurons from the recorded neurons based on their electrophysiological characteristics (Figures 6A–D). The firing rate of putative PV⁺ neurons in BF increased during resting behavior compared to exploration behavior (Figure 6E), consistent with previous studies (Lozano-Montes et al., 2020). Most single cell activities can be continuously detected for a minimum duration of 5 days, and the rest of the electrophysiological activities can be stably detected for at least 4 weeks. These data demonstrate that our electrode is capable of high-quality electrophysiological recording.

We conducted further analyses on the blue light activating LFP responses (Figure 7) by injecting the Cre recombination-dependent rAAV2/9-DIO-ChR2-GFP virus into the BF of ChAT-Cre mice before placing optical fibers above these brain regions (Figure 7A). This virus was used to specifically express channelrhodopsin-2-H134R (ChR2, a modified version of a light-gated ion channel) in cholinergic neurons of the BF. We used the ChAT-Cre mouse model because cholinergic neurons in the BF innervate most of the DMN regions (Do et al., 2016). We confirmed the specific labeling of ChAT neurons (ChAT^ChR2-GFP) through immunofluorescence staining and found that nearly all of the GFP neurons (94.9 ± 1.3%, n = 3) were ChAT positive (Figure 7B). To assess the modulation of DMN activity resulting from the activation of cholinergic neurons in the BF, we applied blue laser light onto ChAT^ChR2-GFP neurons in mice BF to examined the resulting LFP responses in DMN regions. We found a decrease in theta-band oscillation in almost all DMN regions during sinusoidal stimulation in BF (Figures 7C–F), validating the modulation of DMN activity by cholinergic neurons in the BF.

Our presented opto-electrode possesses several advantages, including its ultra-lightweight and compact size, as well as its ability to record and stimulate a larger number of brain regions while maintaining comparable LPF/SUA recording performance to previous electrodes (Headley et al., 2015; Billard et al., 2018; Ma et al., 2019; Sheng et al., 2021). The opto-electrode weighs only 0.28 g, which enables it to cover up to 18 brain regions while reducing the burden on animals to tolerate and behave freely without excessive physical exertion, even when a cable suspension system is not utilized to balance the weight (Battaglia et al., 2009). Moreover, the smaller size of our opto-electrode design reduces limitations on animal behavior. Unlike other multi-drive-based movable electrodes (Headley et al., 2015; Billard et al., 2018; Ma et al., 2019; Sheng et al., 2021), our opto-electrode does not require additional space to accommodate and fix the drives, thus reducing the horizontal sloshing caused by a higher electrode center of gravity and lower torque imposed during the animal’s motion. Additionally, the opto-electrode’s smaller size and weight allow for the utilization of a larger capacity battery in wireless stimulation/acquisition systems (Szuts et al., 2011; Gagnon-Turcotte et al., 2017; Shin et al., 2022), which provides longer recordings without increasing the overall weight.

Although the materials used in current electrodes are cheap and readily available, the high Young’s modulus of the metal microwire and quartz optical fiber could lead to persistent tissue damage and immunoreactive glial responses in the brain (Cao et al., 2021; Zhou et al., 2022). These factors could significantly impact the chronic stability of recording, particularly the single-unit activity. However, the current opto-electrode design incorporates flexible materials, which should help to reduce damage and increase the stability of long-term recordings. In addition, while the opto-electrode described in this study utilizes a laser processing mold that permits mass assembly while retaining high assembling accuracy, achieving a spatial precision of approximately 20 μm (Ma et al., 2019), there still exists a possibility of mispositioning of the microwires during implantation. It is particularly prominent when only one microwire is designed for one brain region, as it can drift to the side. However, this effect can be significantly reduced by increasing the number of microwires or by enhancing the strength of a single microwire.