Electrophysiological assessment of the bimodal effects of FUS

To examine the efferent effects of excitatory FUS, EMG activity from both hind limbs (i.e., contralateral and ipsilateral EMG) was measured using a dual-channel data acquisition system (BioAmp ML408 with PowerLab 4/35, ADInstruments, Colorado Springs, CO) while monitoring the presence of overt muscle twitches or leg movement during sonication. Subdermal wire electrodes (SWE-L-25, Ives EEG Solutions, Newburyport, MA) with a 1.5 mm silver chloride (AgCl) tip were inserted subcutaneously over the gastrocnemius muscle with a ~3 cm distance between the positive and negative electrodes. A reference cup electrode was placed on the skin between the hooves. The wool over both upper hind limbs and between the hooves was trimmed to expose the skin for placement of electrodes.

Because the amplitude and temporal features (i.e., frequency spectrum and shape) of the EMG signals may vary depending on the electrode configurations (e.g., shape and type of electrode) [66–68] and the data acquisition hardware settings, we first measured the EMG response from mechanical stimulation of the muscle over the peroneal/tibial nerve of the hind leg (from ‘SH4’), which showed a biphasic shape instead of a repetitive EMG firing pattern (Supporting Information S1A and S1B Fig show an example of the measured EMG). Passive muscle actuation is known to generate a comparable EMG signal from the active limb motion in humans [69, 70].

Then, the time-locked EMG signal was acquired from 0.5 s before to 1 s after the onset of each FUS sonication event (10 kHz sampling rate, low-pass filter of 30 Hz high cut-off; LabChart 7, ADInstruments). Twenty excitatory FUS stimulations were given in each session. Multiple sonication sessions (up to four) were administered to each animal (see Supporting Information S1–S4 Tables for more detail; the number of stimulations, therefore, ranges from 20–80). As the EMG signal can be confounded by spontaneous muscle activity or respiratory motion, we used the following criteria to detect the FUS-related signals: (1) magnitude (difference between the distinct positive and negative peaks) over 1.5 μV, and (2) time interval between FUS onset to the emergence of the first negative peak (defined as the ‘latency’ herein) in the range of 0–250 ms, considering the sonication duration of 200 ms and the sheep’s nerve conduction velocity of ~100 m/s [71]. Here, a magnitude of 1.5 μV was chosen based on averaged one standard deviation (1.48 ± 0.92 μV, n = 10 sheep) of the EMG signal fluctuations measured during 1 min of resting-state. The assessment window of 0–250 ms for analyzing the latency of EMG responses was based on the presence of a broad onset latency distribution of response elicited by FUS brain stimulation of the motor cortical area in previous studies [28, 31, 72, 73]. We excluded from further analysis any peaks (1) with signal magnitude over 20 μV (due to the possibility of signal artifact) or (2) with a time gap greater than 200 ms between the positive and negative peaks (related to motion artifact). The response rate for each sonication parameter per animal was calculated as a ratio of the number of elicited EMG responses to the total number of FUS stimulations. Considering the possible presence of reflex-type startle responses among acquired data [74, 75], EMG responses with short latency (< 25 ms) were not included as successful excitatory responses for the response rate calculation.

To assess the suppressive effects, the EEG SEP induced by unilateral electrical stimulation of the right hind leg muscle was monitored. EEG data was acquired from two subdermal EEG electrodes (SWE-L-25, Ives EEG Solutions) inserted (1) under the skin over the left rostral portion of the skull and (2) ~2 cm left of the bregma (based on MRI images), using the same dual-channel data acquisition system (BioAmp ML408 with PowerLab 4/35, ADInstruments). Reference electrodes were subcutaneously applied over the left posterior region of the occipital bone, and a ground cup electrode was placed between the hooves of the right fore limb. To elicit the SEP, electrical stimulation (10–15 mA electrical currents, duration of 50 μs, and frequency of 2 Hz) was given to the gastrocnemius of the right hind limb using a surface stimulator (MLADDF30, ADInstruments). The corresponding EEG was recorded from 50 ms before to 100 ms after the onset of electrical stimulation (10 kHz sampling rate, band-pass filter at 0.5–200 Hz; LabChart 7, ADInstruments). The time-locked EEG signal was measured every 0.5 s a total of 120 times (thus, 1 min for each SEP acquisition) and averaged to represent SEP.

The SEP was measured three times (labeled B1–B3) to establish the baseline condition before applying sonication. Subsequently, FUS was delivered to the targeted brain areas for 2 min while measuring two sets of SEPs (labeled F1 and F2). After the end of sonication, five additional sets of SEP (labeled P1–P5) were acquired in succession. An example of the obtained SEP signal is displayed in Supporting Information S1C Fig. To provide active control conditions, we also delivered sonication to the S1 and thalamic area in the right hemisphere across five sheep, using the sonication parameters of SP3 and SP7 (a total of 17 sets of measurement were taken). The distance between the target and the corresponding control sites was 18.8 ± 5.2 mm (n = 5) for the S1 and 14.5 ± 4.7 mm (n = 5) for the thalamic area.

To examine the degree of suppression per animal, we measured the magnitude between the negative peak at ~40 ms (N₄₀) and the positive peak at ~50 ms (P₅₀) of the SEP (i.e., P50 –N₄₀ in Supporting Information S1C Fig) after the onset of electrical stimulation. The SEP magnitude from acquisition sets (i.e., B1–B3, F1, F2, and P1–P5) was normalized with respect to the averaged magnitude across pre-sonication baseline conditions (i.e., B1–B3), whereby the averaged value acquired from the baseline conditions was set to zero.