High-Resolution Focused Ultrasound Neuromodulation Induces Limb-Specific Motor Responses in Mice in Vivo

Abstract

Ultrasound can modulate activity in the central nervous system, including the induction of motor responses in rodents. Recent studies investigating ultrasound-induced motor movements have described mostly bilateral limb responses, but quantitative evaluations have failed to reveal lateralization or differences in response characteristics between separate limbs or how specific brain targets dictate distinct limb responses. This study uses high-resolution focused ultrasound (FUS) to elicit motor responses in anesthetized mice in vivo and four-limb electromyography (EMG) to evaluate the latency, duration and power of paired motor responses (n = 1768). The results indicate that FUS generates target-specific differences in electromyographic characteristics and that brain targets separated by as little as 1 mm can modulate the responses in individual limbs differentially. Exploiting these differences may provide a tool for quantifying the susceptibility of underlying neural volumes to FUS, understanding the functioning of the targeted neuroanatomy and aiding in mechanistic studies of this non-invasive neuromodulation technique.

Keywords: Brain stimulation; Focused ultrasound; Locomotion; Motor response; Neuromodulation.

Fig. 1.

Experimental setup. (a) The focused ultrasound system and signal acquisition equipment are labelled on the diagram of a head-fixed mouse in a stereotactic positioner. (b) An outline of the mouse brain is shown with the three brain regions investigated (M: motor, LP: left posterior, RP: right posterior). The red line denotes the sequence of target sonication from 1 to 8. Two grey boxes along the midline denote the bregma and lambda landmarks from top to bottom, respectively. (c) A general outline of the experimental procedure.

Fig. 2.

The four-limb EMG response (a) latency, (b) duration, and (c) power for the five regions (left posterior: LP1, LP2, LP3; right posterior: RP1; motor: M1) are shown grouped by targets (Targets 1-8). The geometric mean (+/− standard deviation) are denoted on each plot. A two-way ANOVA using target and limb as main effects and multiple comparison tests with Tukey correction were used to evaluate the significant main target effect. The significant pairings of targets (p < 0.05) are labeled by the target number showing significance. Significant pairings are not labeled reciprocally.

Fig. 3.

The four-limb EMG response (a) latency, (b) duration, and (c) power for each region (left posterior: LP1, LP2, LP3; right posterior: RP1; motor: M1) are shown grouped by limb (LH = left hindlimb, RH = right hindlimb, LF = left forelimb, RF = right forelimb). The geometric mean (+/− standard deviation) are denoted on each plot. A two-way ANOVA using target and limb as main effects and multiple comparison tests with Tukey correction were used to evaluate significance (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001).

Fig. 4.

The effect of a significant interaction term for response latency is shown for four targets in the right posterior region (RP1 from Figure 2). (a-d) For each target, the RMS signals for the first 200ms post-stimulus onset are shown for each limb. The overlaid horizontal bars represent the 90% confidence interval of mean latencies for each limb. (e) The response latencies for each limb and target of interest are shown (geometric mean +/− geometric SD). Tukey’s multiple comparison test was used to evaluate the simple effects between limb groups for each target (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001).

Fig. 5.

The effect of a significant interaction term for response duration is shown for four targets in the motor region (M1 from Figure 2). (a-d) For each target, the first 1500 ms of the mean RMS signals are shown for each limb. Signal traces have been normalized by the regional maxima of each respective limb. The left and right bar edges of the overlaid bars represent the mean onset and offset of the motor activity, respectively. (e) The response durations for each limb and target of interest are shown (geometric mean +/− geometric SD). Tukey’s multiple comparison test was used to evaluate the simple effects between limb groups for each target (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001).

Fig. 6.

The effect of a significant interaction term for response power is depicted for four targets in the motor region (M1 from Figure 2). (a-d) For each target, the mean cumulative power signals for the first 500 ms post contraction onset are shown for each limb. The overlaid bars represent the mean power of the motor response ensemble over this period, respectively. (e) The response power for each limb and target of interest is shown (geometric mean +/− geometric SD). Tukey’s multiple comparison test was used to evaluate the simple effects between limb groups for each target (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001).

Fig. 7.

The evaluation of lateralization of responses is depicted by grouping responses from either the left or right posterior regions by limb and comparing response power. (a) The mean normalized responses for each limb are shown between the left posterior (LP) and right posterior (RP) groups. Mann-Whitney tests were used to make pairwise comparisons between the left and right posterior regions for each limb (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001). There were between 40 and 55 responses in each of the hindlimb groups and between 48 and 53 responses in each of the forelimb groups. Response power from the hindlimbs was significantly contra-lateralized. The response power of the left forelimb (LF) was significantly contra-lateralized, while the right forelimb (RF) was significantly ipsilateralized. (b) The RMS signals for each limb in each of the two regions are shown with the onset of ultrasound at 0.5 seconds.

Fig. 8.

Simulations were performed using k-Wave (Treeby and Cox 2010). The coronal slices corresponding to the front and back rows of the right posterior (RP) region were selected from a mouse micro-CT and used to determine the properties of the simulated propagation medium. Each of the eight targets of the RP region is shown and labeled by their mediolateral (ML) distance from the midline and from their distance from the interaural line (IA) in millimeters using mouse brain atlas coordinates (Paxinos and Franklin 2008). The peak pressure fields (dB scale) within the skull (black) are shown. Areas with pressure below −6dB are not colorized. The blue region represents the approximate location of the mesencephalic locomotor region (Roseberry et al. 2016).