Impact of endogenous sonosensation on in vivo sonogenetics

Summary

Sonogenetics is a promising paradigm that uses non-invasive ultrasound to modulate neurons expressing sonosensitive proteins. However, the brain’s varying endogenous sonosensitivity presents a challenge for its precise application. We first mapped this intrinsic sensitivity, identifying the somatosensory cortex as particularly responsive. Using an MscL-G22S-based approach in the somatosensory cortex, we found that ultrasound non-specifically activated both excitatory and inhibitory neurons in control mice, preventing behavioral output. In contrast, in MscL-expressing mice, ultrasound selectively activated excitatory neurons while suppressing inhibitory ones, thereby shifting the net neural response to successfully drive whisker movement. This effect was independent of auditory confounds or astrocytic involvement. Our study highlights that accounting for endogenous sonosensitivity is critical and demonstrates that optimized sonogenetic tools can achieve precise neuromodulation despite this inherent challenge.

Graphical abstract

Highlights

• •The midbrain exhibits region-specific endogenous sonosensitivity
• •Somatosensory cortex shows high endogenous sonosensitivity
• •MscL sonogenetics shifts neural response profiles in the cortex
• •MscL sonogenetics drives precise whisker movement

Ultrasound technology; Molecular neuroscience; Sensory neuroscience

Introduction

Neuromodulation involves the controlled delivery of various forms of physical energy to intervene in the functioning of the nervous system, in order to treat neurological diseases and enhance our understanding of brain functions. Optogenetics and chemogenetics are among the premier methods that have significantly advanced neuroscience by enabling precise control of neuronal activity. Optogenetics utilizes light-sensitive proteins to temporally manipulate neurons, facilitating the study of neural circuits and behaviors.¹ In contrast, chemogenetics employs engineered receptors activated by designer drugs to achieve spatial specificity and sustained effects.^2,3 These approaches have been proven beneficial for understanding complex brain functions and developing therapies for neurological disorders,⁴ enabling substantial advancements in both research and clinical applications. However, these techniques are invasive, have limited spatial resolution, or show inadequate temporal resolution.^5,6 Sonogenetics is an innovative technique that uses ultrasound waves to modulate the activity of genetically modified cells non-invasively.⁷ Its high spatial targetability, cell-type specificity, capacity for large volume modulation, and non-invasive nature of the stimulus^8,9 effectively address some of the limitations associated with optogenetics and chemogenetics.¹⁰ As a result, sonogenetics holds significant promise for advancing our understanding of cell pathophysiology and offers the potential for developing innovative treatments for both neural and non-neural diseases. In recent years, multiple studies have shown the efficacy of sonogenetic neuromodulation in precisely and remotely activating specific neurons or neural circuits and influencing animal behavior.^{11,12,13,14,15,16,17} Furthermore, sonogenetics has shown potential as a non-invasive therapeutic strategy for visual restoration and enhancing behaviors associated with Parkinson’s disease^17,18 in animal models.

Research has shown that the brain and numerous cells throughout the human body have an inherent capacity for mechanosensation.^19,20,21 This ability allows them to detect and respond to mechanical stimuli, including pressure, stretch, and shear forces, all of which are vital to various physiological processes.^22,23 Hence, sonogenetics uses the introduction of highly sonosensitive mechanoresponsive machinery into a specific group of cells as a targeting mechanism, allowing for the use of lower ultrasound intensities and shorter pulses. This allows for the minimization of ultrasound energy that may cause off-target effects. However, given the varying levels of endogenous ultrasound sensitivity seen in the brain,²³ it is imperative to ascertain whether different brain regions can independently respond to ultrasound stimulation and establish a reliable endogenous response profile for these regions. This is vital for properly targeting sonogenetics to specific, desired areas while minimizing the impact on surrounding regions, as well as optimizing stimulus parameters. It would not only reduce potential side effects, but also enhance the efficacy of neuromodulation and improve overall therapeutic outcomes.

In the present study, we found that several brain regions - the somatosensory cortex, thalamus, and hippocampus—demonstrated greater response to ultrasound in wild-type (WT) mice, as observed through assessing the expression of the neuronal activation marker c-Fos across different brain regions. In particular, recordings of calcium (Ca²⁺) activity indicated that the somatosensory cortex had a stronger response than the visual cortex, along with the high endogenous expression of various mechanosensitive ion channels. To determine the efficacy and specificity of sonogenetics in a brain region with relatively higher sonosensitivity, the somatosensory cortex was sonogenetically stimulated, and any whisker movement induced was evaluated. We found that the sonication activated both excitatory and inhibitory neurons in WT mice without significant behavioral changes, and the activation observed was not attributable to auditory effects or astrocyte involvement. In contrast, sonogenetic stimulation was found to effectively activate excitatory neurons expressing MscL in the somatosensory cortex, increasing neural activity along the circuit and inducing whisker movement. These findings suggest that enhanced responses in MscL-G22S-expressing excitatory neurons may shift the Excitation/Inhibition balance (E/I balance) and reduce the concurrent activation of inhibitory neurons, leading to notable changes in overall neural and behavioral responses to sonogenetic ultrasound stimulation. The implementation of sonogenetic methods serves to mitigate sonosensitive background effects, consequently enhancing the efficacy of neuromodulation.

Results

The somatosensory cortex shows strong endogenous responses to ultrasound stimulation compared to other midbrain regions

Different brain regions can sense and respond to ultrasound differently.²³ It is crucial to characterize and understand the endogenous sonosensitivity (sensation to ultrasound stimulation) of specific brain regions, which can be used to optimize the efficacy and accuracy of ultrasound neuromodulation. To characterize the performance of different brain regions to sonication, we examined neural activity in brain regions covered by an ultrasound field in wild-type (WT) mice by counting c-Fos⁺ (a neuronal activity marker) cells following 0.25 MPa ultrasound treatment. An ultrasound transducer coupled with a 6 mm diameter adaptor was placed directly above the midbrain region (Figure 1A). The number of c-Fos⁺ signals was then compared between corresponding regions treated (US) or untreated (NUS) with ultrasound. We observed significantly elevated c-Fos expression in the somatosensory cortex (SS), hippocampus (Hippo), and thalamus (TH) in the US group compared to the NUS group. This finding indicates that these brain regions exhibit high sonosensitivity, whereas no significant differences were found in other areas (Figures 1B and 1C). Notably, of the regions examined, the SS displayed the greatest sonosensitivity, showing an over 2-fold increase in c-Fos⁺ cells following the administration of ultrasound, compared to the corresponding NUS mice. Additionally, we assessed the neural activity in the visual cortex (VIS), where the application of sonogenetics has been effectively established with minimal background activity.¹⁶ Our findings indicated that ultrasound treatment led to an increase in c-Fos⁺ signals in the somatosensory cortex; however, no such increase was observed in the visual cortex in this parameter stimulation (the SS response normalized to NUS was 2.275, while VIS normalized to NUS was 0.9786).

Figure 1.

Identification of ultrasound-activated brain regions

(A) Sagittal-sectional illustration of the ultrasound intensity field produced in the brain.

(B) c-Fos-positive signals in various ROIs treated with ultrasound (US) were normalized to the same region in the no-ultrasound (NUS) group. ZI, Zona incerta; TH, thalamus; STN, subthalamic nucleus; SS, somatosensory cortex; MO, somatomotor area; Hippo, hippocampal; CP, caudoputamen; CeA, central amygdalar nucleus; ACB, nucleus accumbens; VIS, visual cortex; GPi, globus pallidus; GPs, globus pallidus; BST, bed nuclei of the stria terminalis; ORB, orbital area; LPO, lateral preoptic area; and LHA, lateral hypothalamic area. Data are shown as mean ± SEM. n = 3 mice in each group, ∗p < 0.05, ∗∗∗∗p < 0.0001, two-way ANOVA test.

(C) Representative images of c-Fos expression in the 3 regions most strongly activated by ultrasound - somatosensory cortex, hippocampus, and thalamus with/without 0.25 MPa ultrasound stimulation. Scale bars = 250 μm.

Given the pronounced response of the somatosensory region to ultrasound stimulation, we sought to explore its sensitivity in greater detail. We selected the visual cortex, which has a lower mechanical sensation, as a benchmark for comparison. To investigate the responses of both the somatosensory and visual cortices – each exhibiting diminished sensitivity to ultrasound parameters—we applied ultrasound stimulation to ex vivo brain slices that had been virally modified to express the genetically encoded calcium sensor GCaMP6s (Figure 2A). An ultrasound transducer was precisely positioned above the brain slice. The beam profile was measured for calibrated positioning, ensuring coverage of the region of interest, and the transducer was slightly tilted to avoid a standing wave. A microscope situated beneath the brain slice dish was utilized to record calcium dynamics at two key intervals: before and after sonication (Figure 2B). Similarly, when imaging the real-time Ca²⁺ response, SS brain slices showed significantly greater influx of Ca²⁺ (Peak ΔF/F₀ = 63.12% ± 3.31%) in response to ultrasound than VIS brain slices (Peak ΔF/F₀ = 26.41% ± 1.43%) under the same sonication condition (Figures 2C–2E). These results indicate that the somatosensory cortex possesses significantly greater endogenous sonosensitivity than the visual cortex.

Figure 2.

The somatosensory cortex is especially sensitive to ultrasound stimulation

(A) Schematic illustration of the GCaMP6s viral injection sites (SS and VIS).

(B) Schematic illustration of acute brain slice calcium imaging setup, showing ultrasound delivered to brain slices from above and calcium dynamic imaging using an inverted fluorescence microscope.

(C) Representative images of GCaMP6s fluorescence in somatosensory cortex neurons and visual cortex neurons in acute brain slices, pre- and post-stimulation with 0.25 MPa US. The red arrows indicate that the cells exhibit an increase in GCaMP6s signal in response to ultrasound stimulation. Scale bars = 50 μm.

(D) Representative traces of calcium dynamic fluorescence from SS neurons and VIS neurons in an acute brain slice. ΔF/F₀: change compared to fluorescence/initial baseline.

(E) Peak calcium fluorescence signal of SS neurons and VIS neurons responding to US stimulation. Data are shown as mean ± SEM. n = 140 SS neurons from 6 mice, 234 VIS neurons from 4 mice. ∗∗∗∗p < 0.0001, unpaired two-tailed t test.

Endogenous mechanosensitive ion channels play a significant role in the region’s response to ultrasound stimulation

Given the differing response profiles of these regions, we hypothesized that the observed influx of extracellular calcium was induced by sonication involving the activation of endogenous mechanosensitive ion channels. To test this, we evaluated the profile of endogenous sonosensation in somatosensory cortex neurons by measuring the expression levels of the well-established mechanosensitive ion channel families, including TRPA, TRPC, TRPV, TRPM, and PIEZO, and their respective contributions to the neuronal response to ultrasound. qPCR revealed that the TRPC and TRPM families exhibit high levels of expression, with some TRPV and PIEZO1 expression also detected (Figure 3A).

Figurer 3.

Profile of endogenous sonosensation and ultrasound response in the somatosensory cortex

(A) Relative endogenous mRNA expression of multiple mechanosensitive ion channels in somatosensory cortices of WT mice. n = 3 mice.

(B) Representative images of ex vivo neuronal calcium signaling from brain slices of mice treated with 0.25 MPa US. The red arrows indicate that the cells exhibit an increase in GCaMP6s signal in response to ultrasound stimulation. Scale bars = 10 μm.

(C) Representative calcium dynamic traces of one neuron from brain slices treated with no blocker intervention (None), RR, ML204, Ononetin, GD³⁺, or PIEZO1 KO mice, all treated with 0.25 MPa US.

(D) Summarized calcium changes from ex vivo experiments. n is as follows: 140 neurons from 4 mice + No blocker; 104 neurons from 4 mice + RR; 147 neurons from 4 mice + ML204; 103 neurons from 4 mice + Ononetin; 87 neurons from 4 mice + Gd³⁺; 120 neurons from 3 PIEZO1 KO mice. Bars represent the median and interquartile range; ∗∗∗∗p < 0.0001; One-way ANOVA with Dunnett’s multiple comparisons test.

Next, to elucidate which specific mechanosensitive ion channel plays a predominant role in ultrasound activation, we applied selective blockers during the calcium imaging of acute brain slices. Ruthenium red (RR, blocking TRPV1, TRPV2 and TRPV4), ML204 (blocking TRPC4 and TRPC5), and Ononetin (blocking TRPM) were used as specific chemical blockers, and the stretch-activated channel blocker Gadolinium III (Gd³⁺) was used as a broad-spectrum blocker of mechanosensitive ion channels. The peak ΔF/F₀ responses measured were as follows (Figures 3B–3D): No blocker-US ΔF/F₀ = 63.12% ± 3.31%; Ruthenium Red-US ΔF/F₀ = 40.63% ± 2.65%; ML-204-US ΔF/F₀ = 23.17% ± 1.57%; Oneontin-US ΔF/F₀ = 33.59% ± 1.70%. Gd³⁺-US ΔF/F₀ = 36.29% ± 2.19%; Overall, incubation with each channel blocker significantly diminished calcium responses to ultrasound stimulation compared to the No Blokcer-US group that did not receive any blockers.

In addition to our previous findings about PIEZO1’s significant role in ultrasound neuromodulation,²⁴ PIEZO1 expression was also detected in this study’s qPCR screen of somatosensory neurons. Hence, to evaluate the role of Piezo1 in the somatosensory cortex, we used the same Piezo1 knockout (P1KO) mouse model as our previous study.²⁴ The peak ΔF/F₀ responses measured were as follows (Figures 3B–3D): P1KO ΔF/F₀ = 4.77% ± 0.59%. Overall, incubation with each channel blocker significantly diminished calcium responses to ultrasound stimulation compared to the No Blocker-US group that did not receive any blockers. Our comprehensive calcium imaging experiments conducted on acute brain slices revealed that mechanosensitive ion channels were crucial in mediating SS neuronal responses to ultrasound.

Implementation of MscL-G22S-mediated sonogenetics in the somatosensory cortex to elicit enhanced response to ultrasound

Sonogenetics utilizes ultrasound to non-invasively manipulate cells that express ultrasound-responsive proteins, enabling the modulation of neuronal functions and behaviors. However, its implementation in regions such as the somatosensory cortex could possibly be dampened by their endogenous sonosensitivity. This region is known to specifically and rapidly elicit whisker movement in response to ultrasound.^25,26 Hence, we probed the effectiveness of implementing sonogenetic stimulation in this high-sonosensitivity region by evaluating whisker response to ultrasonic SS stimulation. Mice were injected with the MscL-G22S-EYFP or EYFP (vector control, abbreviated as “Ctrl”) AAV particles under a CamKIIα promoter targeting excitatory neurons in the primary somatosensory cortex. Four to five weeks following the virus injection, all whiskers except the C2 on both sides were trimmed in preparation for the experiment. Mice were positioned in a tube, and a 500 kHz ultrasound transducer was placed above SS, and a digital camera was used to record whisker movement in real-time during sonication (Figure 4A).

Figure 4.

MscL-mediated US stimulation of the somatosensory cortex evokes stronger whisker movement

(A) Schematic of the somatosensory cortex experimental scheme. Briefly, mice were injected in their right somatosensory cortex with CamKIIα: EYFP or CamKIIα: MscL-G22S. Four weeks later, mice were treated with ultrasound and the left C2 whisker’s movement was recorded using the setup illustrated.

(B) Images indicate the expression of CamKIIα:EYFP (Ctrl) or CamKIIα:MscL-G22S-EYFP in the somatosensory cortex (primary somatosensory cortex, SS). Hippocampus (Hippo), cerebral cortex (CTX). Scale bars = 500 μm.

(C) Representative images tracking whisker movements, showing the minimum intensity projection of the frames taken before and after ultrasound stimulation (0.15 MPa) in Ctrl (left) and MscL-G22S (right) mice. Scale bars = 6 mm.

(D) Representative traces of the angle of the C2 whisker from an awake mouse following ultrasound stimulation on the surface of the somatosensory cortex with Ctrl and MscL-G22S. The timing of ultrasound stimulation (0.15 MPa) is indicated as a pink background.

(E) Summary data for the average angular velocity of whisker movement evoked by different parameters of ultrasound stimulation in Ctrl and MscL-G22S mice. In 0.1 MPa ultrasound stimulation: Ctrl group n = 8 mice, MscL-G22S group n = 7 mice; 0.15 MPa: Ctrl group n = 9 mice, MscL-G22S group n = 7 mice; 0.4 MPa: Ctrl group n = 7 mice, MscL-G22S group n = 6 mice. ∗p < 0.05, ns = not significant, Unpaired 2-tailed t test. Data are shown as mean ± SEM.

(F) Summary data for the average angular velocity of whisker movement evoked by ultrasound stimulation in the sham group (no coupling gel between skull and transducer). Ctrl group n = 6 mice, MscL-G22S group n = 6 mice. ns = not significant, Unpaired 2-tailed t test. Data are shown as mean ± SEM.

Confocal microscopy indicated successful expression of CamKIIα: EYFP or CamKIIα: MscL-G22S in the somatosensory cortex (Figure 4B). The expression efficiency of the CamKIIα virus in neurons was confirmed by staining for MAP2-positive cells (a marker of neurons). The co-expression level with GAD67-positive cells (a marker of GABAergic neurons) was found to be low, indicating that the majority of the CamKIIα virus is expressed in excitatory neurons (Figure S1) in both Ctrl and MscL-G22S groups. The movement of the contralateral C2 whisker in awake mice was monitored for 10 s before and during ultrasound stimulation. The average increase in angular velocity following the ultrasound stimuli was calculated and compared. Notably, C2 whisker movements were significantly larger in MscL-G22S mice than in Ctrl mice (Figures 4C and 4D). Ultrasound stimuli at 0.15 MPa (Ctrl = 14.24°, MscL-G22S = 91.79°) and 0.4 MPa (Ctrl = 47.92°, MscL-G22S = 88.62°) resulted in an increased angular velocity of whiskers across all mice (Figure 4E). However, the velocities were markedly greater in the MscL-G22S group, and the threshold ultrasound intensity needed to stimulate C2 whisker movement was significantly lower in MscL-G22S mice compared to Ctrl mice (Figure 4E). No significant changes were detected in any of the mice subjected to a sham decoupling condition, which involved the absence of coupling gel between the transducer and the scalp. Consequently, no ultrasound energy was transmitted to the mouse brain. The average velocity change for all the mice remained below 2° per second (Figure 4F), suggesting that any sound produced by the ultrasound transducer did not influence the mice’s behavior. Hence, we found that our MscL-mediated sonogenetic approach could sufficiently sensitize neurons to low-intensity US stimulation for successful neuromodulation and induced whisker movements, even when implemented in a region with high endogenous sonosensitivity.

We further assessed neuronal activation by recording the cellular responses to the stimulation in the form of Ca²⁺ dynamics recorded by fiber photometry (Figure 5A). First, CamKIIα: MscL-G22S-EYFP or CamKIIα: EYFP mixed with CamKIIα:jRGECO1a (a red fluorescent genetically encoded Ca²⁺ indicator) were co-injected into the somatosensory cortex of the right side of mice’s brain (Figure 5B), and an optical fiber was inserted into the exact coordinates. Any alteration of the baseline excitability of neurons owing to MscL-G22S expression alone was assessed by administering air puffs. As measured by fiber photometry, no significant differences were observed from the airstream-triggered calcium dynamics between Ctrl and MscL-expressing mice, indicating no noticeable change in baseline neural activities in MscL-expressing neurons (Figures 5C and 5D). The experiments revealed that US consistently evoked neural activities in both Ctrl and MscL-G22S-expressing cortices; however, the MscL-expressing cortices demonstrated significantly higher calcium responses at 0.15 MPa (Peak ΔF/F₀: Ctrl = 0.37 ± 0.06%, MscL-G22S = 0.61 ± 0.08%) and at 0.4 MPa (Peak ΔF/F₀: Ctrl = 0.28 ± 0.04%, MscL-G22S = 0.65 ± 0.1%) (Figures 5E–5G). These data further suggest that the expression of MscL-G22S enhances the responsiveness of somatosensory neurons to US stimulation, without altering their baseline excitability.

Figure 5.

US stimulation elicits greater calcium activity in excitatory neurons of the somatosensory cortex in MscL-expressing mice without altering their baseline excitability

(A) Schematic diagram of whisker stimulation and fiber photometry recording experiment. Anesthetized mice were administered air puffs and ultrasound, respectively, and the calcium signal from their right-hand-side barrel cortices was recorded.

(B) Representative images showing CamKIIα:EYFP or CamKIIα:MscL-G22S co-expressed with CamKIIα: jRGECO1a in the SS. White arrows indicate the CamKIIα:EYFP or CamKIIα:MscL-G22S neurons co-expressed with CamKIIα: jRGECO1a in the SS. Scale bars = 500 μm (Left panel). Scale bars = 25 μm (right panel).

(C) Representative traces showing calcium activity in response to air puffs in Ctrl and MscL-G22S mice. Gray markings indicate the timing of air trigger stimulation.

(D) Averaged jRGECO1a fluorescence traces increase in the SS of the anesthetized Ctrl and MscL-G22S mice in response to air trigger stimulation. n = 3 mice in each group, ns = not significant, Unpaired 2-tailed t test. Data are shown as mean ± SEM.

(E) Representative calcium response traces under 0.15 MPa ultrasound stimulation in Ctrl and MscL-G22S mice.

(F) Averaged jRGECO1a fluorescence traces from the somatosensory cortex of the anesthetized Ctrl and MscL-G22S mice without (NUS) or with ultrasound stimulation. 4–6 trials were performed for each condition. Ctrl group n = 5 mice, MscL-G22S group n = 5 mice. The timing of ultrasound stimulation is indicated as a light green rectangle.

(G) Average peak Ca²⁺ activity in Ctrl-mice and MscL-G22S mice responds to 0, 0.15, and 0.4 MPa ultrasound stimulation. 4–6 trials were performed for each condition. n = 5 mice in each group. ∗∗p < 0.01, ∗∗∗p < 0.001, ns = not significant, unpaired 2-tailed t test. Data are shown as mean ± SEM.

To examine whether the activation of the somatosensory cortex was attributable to an auditory confound (Figure 6A), we used smooth waveform ultrasound (Figure 6B), which has been shown to exert minimal auditory effect.²⁷ Both Ctrl and MscL-G22S mice still showed synchronous and repeatable calcium activity to smooth waveform ultrasound stimulation, although the MscL-G22S mice showed higher calcium influx than the Ctrl mice at the same ultrasound intensities (Figures 6C and 6D). Moreover, neither group showed apparent calcium influx in the sham decoupled condition (Figure 6D). To compare rectangular versus smooth ultrasound waveforms, we sonicated MscL-expressing and Ctrl mice. The MscL group showed dose-dependent calcium increases, stronger with smooth waveforms. Ctrl showed weak or negative intensity-response relationships. Smooth waveforms seemed to enhance mechanotransduction-mediated sonogenetic effects (Figure 6E). In addition, even in the deaf model, smooth waveform ultrasound stimulation can still elicit concurrent calcium activity in both Ctrl mice and MscL-G22S mice. Notably, the neural activity induced by ultrasound is significantly greater in the MscL-G22S group compared to the Ctrl group (Figure S2). These results suggest that the responses induced by ultrasound treatment may not primarily stem from auditory pathway activation but instead involve cell transduction and endogenous or heterologous sonosensitive proteins.

Figure 6.

Smooth waveform US stimulation elicits greater calcium activity in excitatory neurons of the somatosensory cortex in MscL-G22S expressing mice

(A) Schematic depicting the potential effect pathway of ultrasound stimulation being studied. Ultrasound may induce whisker movement, possibly through the somatosensory to motor cortex circuit. An alternate possible pathway is that the ultrasound could be heard as sound by the ear, leading to whisker movement in reaction.

(B) Illustration of the smooth waveform ultrasound stimulation pattern.

(C) Averaged jRGECO1a fluorescence traces from the somatosensory cortex of anesthetized Ctrl mice and MscL-G22S mice without (top) or with smooth waveform ultrasound stimulation (bottom). 6 trials were performed for each condition. Ctrl group n = 6 mice, MscL-G22S group n = 5 mice. The timing of ultrasound stimulation is shown as a light green rectangle.

(D) Average peak Ca²⁺ activity in both the Ctrl and the MscL-G22S mice measured in response to different levels of ultrasound stimulation (0, 0.15, and 0.4 MPa). Ctrl group n = 6 mice, MscL-G22S group n = 5 mice. ns = not significant, ∗p < 0.05, ∗∗∗p < 0.001, Unpaired 2-tailed t test. Data are presented as mean ± SEM.

(E) Regression analysis of calcium signals induced by rectangular and smooth waveform ultrasound in Ctrl and MscL-G22S mice. MscL-expressing groups—rectangular (Y = 0.1788X + 0.5806) and smooth (Y = 0.3294X + 0.4142); Ctrl-expressing groups—rectangular (Y = −0.3617X + 0.4270) and smooth (Y = 0.01533X + 0.3054).

We proceeded to investigate the involvement of astrocytes in the significant calcium activity observed in both the Ctrl and MscL-G22S groups during this parameter of ultrasound stimulation. Prior research has indicated that astrocytes may respond to ultrasound and potentially induce calcium influx [12]. Calcium influx in excitatory neurons and astrocytes within the somatosensory region was recorded to explore this phenomenon using fiber photometry. To facilitate this analysis, two distinct genetically encoded fluorescent calcium sensors were employed: jRGECO1a, driven by the CamKIIα promoter, in excitatory neurons, and GCaMP6f, driven by the GFAP promoter, in astrocytes. This dual labeling enabled us to accurately identify the source of the calcium signal in the SS (Figure 7A). We found that each pulse train of low-intensity ultrasound stimulation triggered a calcium response in excitatory neurons (ΔF/F₀ - NUS = 0.52% ± 0.17%, ΔF/F₀ - 0.15 MPa = 2.43% ± 0.30%). In contrast, no notable fluorescence changes were detected in astrocytes (ΔF/F₀ - NUS = 1.86% ± 0.57%, ΔF/F₀ - 0.15 MPa = 1.68% ± 0.48%, refer to Figures 7B–7E) under our sonication parameters. These findings confirm that the enhanced calcium signaling induced by our ultrasound sonication strategy is primarily associated with neurons rather than astrocytes.

Figure 7.

Astrocytic calcium response is not involved in the response to sonogenetic stimulation

(A) Genetically encoded calcium sensor labeling strategy for neuronal (red jRGECO1a) and astrocytic (green GCaMP6f) cell types in the somatosensory cortex in WT mice.

(B) Averaged neuronal fluorescence signal change (ΔF/F₀) in the somatosensory cortex of anesthetized mice, with or without 0.15 MPa US stimulation. Blue traces show calcium signals without US (NUS), red traces show calcium signals with US. The green rectangle shows the timing of ultrasound stimulation. n = 5–6 trials, 4 mice.

(C) Average peak calcium activity (ΔF/F₀) in neurons in response to 0, 0.15, and 0.4 MPa ultrasound stimulation in mice. n = 5–6 trials, 4 mice. ∗∗p < 0.01, ∗∗∗p < 0.001, One-way ANOVA with Tukey’s multiple comparisons test. Data are shown as mean ± SEM.

(D) Averaged astrocytic fluorescence signal change (ΔF/F₀) in the somatosensory cortex of anesthetized mice, with or without 0.15 MPa US stimulation. Blue traces show NUS signals, red traces show calcium signals with US. The green rectangle shows the timing of ultrasound stimulation. n = 5–6 trials, 4 mice.

(E) Average peak calcium activity (ΔF/F₀) in astrocytes in response to 0, 0.15, and 0.4 MPa ultrasound stimulation in mice. n = 5–6 trials, 4 mice: ns = not significant, One-way ANOVA with Tukey’s multiple comparisons test. Data are shown as mean ± SEM.

After eliminating contributions from auditory confounds and astrocytes, we focused on potential interactions between inhibitory and excitatory elements within a neural network that could influence its activity. Previous studies have shown that specific parameters of ultrasound stimulation can activate parvalbumin interneurons while concurrently suppressing excitation in the hippocampus.²⁸ To examine the responses of inhibitory GABAergic neurons to ultrasound, we co-injected either CamKIIα:MscL-G22S-EYFP or CamKIIα:EYFP mixed with mDlx5/6:jRGECO1a (a promoter tailored for expression in GABAergic interneurons²⁹) into the somatosensory cortex (Figure 8A).

Figure 8.

US stimulation elicits significantly reduced calcium activity in inhibitory neurons of the somatosensory cortex in the MscL condition

(A) Genetically encoded calcium sensor labeling strategy for inhibitory neurons in Ctrl and MscL-G22S mice. Neurons expressing EYFP are shown in green, and inhibitory GABAergic interneurons expressing the calcium sensor jRGECO1a are shown in red.

(B) Averaged neuronal fluorescence signal change (ΔF/F₀) in the somatosensory cortex of anesthetized mice, without (NUS) or with 0.15 MPa US stimulation. The green rectangle shows the timing of ultrasound stimulation. n = 3 Ctrl mice and 4 MscL-G22S mice.

(C) Average peak calcium activity (ΔF/F₀) in neurons in response to 0, 0.15, and 0.4 MPa US stimulation (0.5 MHz center frequency, 500 μs pulse width, 300 ms stimulation duration, 1 kHz PRF, interval 3s) in mice. Ctrl group n = 3 mice, MscL-G22S group n = 4 mice. ns = not significant, ∗∗∗p < 0.001, Unpaired 2-tailed t-tests. Data are shown as mean ± SEM.

The calcium dynamics of inhibitory GABAergic neurons in Ctrl and MscL-G22S mice were assessed in response to ultrasound stimuli. Neuronal activity before ultrasound stimulation was similar in both groups (Figure 8B; ΔF/F₀-Ctrl -NUS = 0.46% ± 0.06%, ΔF/F₀-MscL-NUS = 0.36% ± 0.07%). A marked activation of GABAergic neurons in Ctrl mice was observed following ultrasound stimulation but not in MscL-G22S mice (ΔF/F₀-Ctrl-0.15 MPa = 1.01% ± 0.21%, ΔF/F₀-MscL-0.15 MPa = 0.27% ± 0.05%; Figures 8B and 8C). This finding indicates that ultrasound stimulation elicits a response in both excitatory neurons and inhibitory GABAergic neurons in WT mice, resulting in a balanced neural activation. In MscL-G22S mice, excitatory neurons expressing the MscL-G22S variant demonstrate significantly increased levels of neural activity. This intensified excitatory drive may overshadow the activation of inhibitory neurons that ultrasound typically induces, culminating in a higher E/I response ratio in the MscL condition. As a result, this altered neural dynamic may lead to distinct behavioral changes in the MscL-G22S mice, emphasizing the complex interplay between excitation and inhibition in response to ultrasound stimulation. Furthermore, this suggests that enhancing sonosensitivity, optimizing parameters, and improving the spatial resolution to target specific populations of neurons through genetic modification is explicitly an effective strategy for achieving cell-type-specific sonogenetic neuromodulation.

Discussion

This study demonstrates that the brain exhibits region-specific sensitivity to ultrasound, likely attributable to the heterogeneous expression profiles of mechanosensitive ion channels. Our data indicate that the somatosensory cortex responds more robustly to sonication than the visual cortex. In both the Ctrl and MscL-G22S groups, ultrasound stimulation activated excitatory neurons in the somatosensory cortex, resulting in significant calcium influx. However, mice expressing MscL-G22S exhibited higher levels of neural activation and, crucially, were able to initiate whisker movement, while no such behavioral changes were seen in the Ctrl conditions. The neuronal activity recorded in both groups was not attributable to astrocyte activity or unintended auditory effects from the sonication settings. These findings suggest that the enhanced activation of MscL-G22S in excitatory neurons in response to sonication may shift the E/I balance and reduce the concurrent activation of inhibitory neurons, thereby leading to downstream behavioral modifications. Overall, sonogenetic effects may arise from introduced sonosensitivity that modulates the E/I dynamics in response to ultrasound.

By selectively modulating the activity of mechanosensitive ion channels, sonogenetics can enhance US’s precision and effectiveness in neuroscience research and clinical applications. However, based on the widespread yet non-uniform expression of mechanosensitive ion channels in the brain, the choice of region and cell-type selected could significantly affect the chances of success for any prospective sonogenetic scheme. This is especially relevant when investigating specific neural circuits or developing therapeutic strategies. Here, we leveraged the inherent ultrasound sensitivity of the somatosensory cortex to validate the sonogenetic approach. The expression of MscL in excitatory neurons successfully elicited whisker movement, an effect not observed in Ctrl mice, underscoring the potential of sonogenetics for targeted neuromodulation.

Our findings reveal that sonogenetic stimulation under the MscL-G22S condition suppressed inhibitory neuron activity. This effect likely stems from network-level interactions rather than direct ultrasonic neuromodulation. Specifically, the selective activation of excitatory neurons may recruit strong recurrent inhibition, leading to the suppression of interneuronal populations. Such disinhibition would profoundly alter local E/I balance, facilitating behavioral output that was absent under Ctrl conditions. This highlights that sonogenetic efficacy depends not only on cellular sensitivity but also on intrinsic circuit properties—a critical consideration for targeted neuromodulation strategies.

Given this heterogeneity in sonosensitivity, enhancing the sensitivity of mechanosensitive ion channels and exploring new sonosensitive candidates could potentially further reduce the parameters of ultrasound stimulation to minimize background effects. The modular nature of sonogenetics allows for the easy inclusion and substitution of improved mediator molecules or other adjuvants, such as nano gas vesicles or microbubbles. In addition, focusing on the targeted region with an ultrasound beam, significantly, could be an excellent way to reduce the ultrasound effects on the untargeted areas. Moreover, although low-frequency ultrasound stimulation has good penetration, increasing the ultrasound frequency might improve the specificity,^15,16 minimizing the impact on surrounding areas, thereby increasing the targetability of the treatment. As seen in the literature^23,30,31 and the results presented here, future ultrasound and sonogenetic stimulation studies should pay close attention to the endogenous profile of sonosensitivity of the targeted regions. The partial inhibition observed with other pharmacological blockers or Cre mice suggests the involvement of a network of mechanosensitive ion channels, possibly through synergistic amplification mechanisms similar to those proposed in models of mechanosensitive calcium accumulation.^23,30,31 Therefore, clarifying the specific contributions of individual mechanosensitive channels is essential for neuromodulation.

We verified that astrocytes did not play a significant role in our setup. Nonetheless, given the cellular diversity of the brain, further confirming the neuronal origin of these signals at a cell-autonomous level would significantly advance our understanding of ultrasound neuromodulation. However, performing direct patch-clamp recordings to probe membrane dynamics remains technically challenging in vivo due to substantial mechanical interference caused by ultrasound stimulation. Alternatively, cell-type-specific perturbation approaches, such as chemogenetic or optogenetic silencing, could be applied in future studies to selectively inhibit specific neuronal subpopulations and more definitively establish the cell-autonomous contribution of neurons to the ultrasonic responses.

In this study, we focused on CamKIIα-expressing excitatory neurons within the somatosensory cortex. However, the deep-penetrating capability of ultrasound makes this sonogenetic strategy highly promising for modulating subcortical structures and other cell types in future studies. To enhance translational safety and delivery efficiency, engineered viral tools such as PhP.eB—which efficiently crosses the blood-brain barrier—could be employed for non-invasive targeting. Endogenous expression levels of mechanosensitive proteins such as PIEZO1 exhibit considerable variation across brain regions and cell types. This heterogeneity likely contributes to regional disparities in baseline neural excitability and ultrasound susceptibility. The pronounced sonosensitivity of the somatosensory cortex (SS) may stem from its unique neurobiological profile—including high mechanosensitive channel expression, robust mechanotransduction mechanisms, and strong thalamocortical inputs within high-gain local circuits.³² Future studies should systematically evaluate ultrasound-responsive channels and their downstream signaling pathways to enable precise and controllable neuromodulation strategies.

In summary, sonogenetics represents a promising non-invasive technology with high spatial resolution, deep tissue penetration, and cell-type specificity. It holds great potential for targeted and whole-brain modulation, advancing our understanding of neural circuits and facilitating the study and treatment of multi-region brain disorders such as Parkinson’s disease, Alzheimer’s disease, depression, and epilepsy. To fully realize this potential, several challenges must be addressed, including clarifying the roles of endogenous mechanosensitive proteins and optimizing ultrasound parameters to account for tissue- and region-specific sensitivity. Resolving these issues will be essential for expanding the applications of sonogenetics in both research and clinical settings.

Limitations of the study

While this study aims to provide a map of endogenous neural sonosensitivity, several limitations should be acknowledged. First, our findings are based on a discrete set of ultrasound parameters; exploring a broader parameter space (e.g., intensity, frequency, and pulse duration) is crucial to fully understand how ultrasound stimulation influences brain activity. Second, the use of a flat-top transducer may have contributed to widespread activation. Employing focused ultrasound in future work could improve spatial specificity and help minimize off-target effects. Third, our investigation was limited to broad neuronal classes (excitatory and inhibitory) and astrocytes. The response of specific cell subtypes and other glial cells, such as microglia, remains to be determined. Furthermore, technical constraints prevented simultaneous calcium imaging from multiple cell types. Future studies employing multi-channel sensors would provide a more integrated view of network dynamics. Lastly, while the MscL-G22S construct enabled targeted neuromodulation, its background activation highlights the need for developing next-generation sonosensitive proteins with higher specificity and lower inherent sonosensitivity to further refine the technique.

Resource availability

Lead contact

Further information and requests for reagents and resources should be addressed to and will be fulfilled by the lead contact, Lei Sun (lei.sun@polyu.edu.hk).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All data reported in this article will be shared by the lead contact upon request, unless it is protected by law.
• •This article does not report original code.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
c-Fos (9F6) Rabbit mAb	Cell Signaling Technology	Cat# 2250S; RRID: AB_2247211
Anti-GAD67 Antibody, mouse monoclonal	Sigma-Aldrich	Cat# 5406; RRID: AB_2278725
Anti-MAP2 antibody - chicken polyclonal	Abcam	Cat# ab5392; RRID: AB_2138153
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 555	Invitrogen	Cat# A-21428; RRID: AB_141784
Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 555	Invitrogen	Cat# A-21422; RRID: AB_141822
Goat anti-Chicken IgY (H+L) Secondary Antibody, Alexa Fluor™ 555	Invitrogen	Cat# A-21437; RRID: AB_2535858
Bacterial and virus strains
AAV-CamKIIα-EYFP	OBiO Technology, Shanghai	Cat. # AOV016
AAV-CamKIIα-MscL-G22S-EYFP	OBiO Technology, Shanghai	Cat. #H15130
AAV-hSyn-GCaMP6s	BrainVTA(Wuhan) Co., Ltd	PT-0145
AAV-CamKIIα-jRGECO1a	BrainVTA(Wuhan) Co., Ltd	PT-1484
AAV-GFAP-GCamP6f	BrainVTA(Wuhan) Co., Ltd	PT-3349
AAV-mDlx5/6-jRGECO1a	BrainVTA(Wuhan) Co., Ltd	PT-1811
Chemicals, peptides, and recombinant proteins
Yoda1	Tocris Biosciences	Cat#5586
Ruthenium red	Tocris Biosciences	Cat#1439
Gd3+ blocker	Tocris Biosciences	Cat#4741
ML204 blocker	Tocris Biosciences	Cat#4732
Ononetin blocker	Tocris Biosciences	Cat#5143
Kanamycin	MedChemExpress	Cat#HY-16566
Furosemide	MedChemExpress	Cat#HY-B0135
Critical commercial assays
TaKaRa MiniBEST Universal RNA Extraction Kit	Takara	#9767
TB Green Premix Ex Taq	Takara	#RR420A
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	JAX# 000664
Mouse: Piezo1tm2.1Apat	The Jackson Laboratory	JAX# 029213
Oligonucleotides
TRPA1: F: CCTTCCACAGAAGACAAGTCCTG R: ACCACATCCTGGGTAGGTGCTA	Thermo-Fisher Scientific	N/A
TRPC1: F: CGTGCGACAAGGGTGACTAT R: CTCCCAAGCACATCTACGCA	Thermo-Fisher Scientific	N/A
TRPC3: F: TGCAGAAGGATCTGGAACTGG R: GTTGGCTGATTGAGAATGCTGTT	Thermo-Fisher Scientific	N/A
TRPC4: F: CGTGGAGTGGATGATATTACCGT R: GTAATCCTGGAGTCCGCCAT	Thermo-Fisher Scientific	N/A
TRPC6: F: GCACCAAGCTCCTCCCTAAT R: TCAGCCCATATCATGCCTATTACC	Thermo-Fisher Scientific	N/A
TRPV2: F: CCAACAAAGAAACACAAACACAAGC R: GTAGCTGGCCGTAGCTTTCTC	Thermo-Fisher Scientific	N/A
TRPV4: F: TCACCGCCTACTATCAGCCACT R: GAACAGGACTCCTGTGAAGAGC	Thermo-Fisher Scientific	N/A
TRPV6: F: TGGTTCCACAGACAAGAGTCC R: GAGGTGATTCCCAGATCCTCTTC	Thermo-Fisher Scientific	N/A
TRPM2: F: CTGCGCCTAGCGATGAGAT R: CATCCTGGACATACTGGTCTGC	Thermo-Fisher Scientific	N/A
TRPM3: F: TATGCCACGCAGCCTTTCTC R: TATGAGGCGTCCACAGCAAC	Thermo-Fisher Scientific	N/A
TRPM4: F: TCTGTGGCTGTGGAGGATG R: TCAGACAGCCGGAGAAAGTTG	Thermo-Fisher Scientific	N/A
TRPM7: F: AATTTGGAGCATTTGTGGGACAC R: CTATATCCAGCAGCACCCACAT	Thermo-Fisher Scientific	N/A
Piezo1: F: ACTTTGCCCTGTCCGCCT A R: GAA GAA GCC CCG CAC AAA C	Thermo-Fisher Scientific	N/A
Piezo2: F: GCACTCTACCTCAGGAAGACTG R: CAAAGCTGTGCCACCAGGTTCT	Thermo-Fisher Scientific	N/A
Software and algorithms
ImageJ	N/A	https://imagej.nih.gov/ij
GraphPad Prism	GraphPad Software, CA, USA	https://www.graphpad.com
MATLAB	The MathWorks, Inc., MA, USA	https://www.mathworks.com/products/matlab.html
Whisker Tracking	N/A	http://whiskertracking.janelia.org

Experimental model and subject details

Male, 4-12 weeks old, C57BL/6J mice and Piezo1tm2.1Apat (JAX# 029213) were used for this study. The mice were housed under standard conditions, with food and water available ad libitum. They were habituated to the procedure room for at least 30 minutes before all behavioral testing experiments. All animal procedures were approved by the Animal Subjects Ethics Sub-Committee (ASESC) of the Hong Kong Polytechnic University, and were performed in compliance with the guidelines of the Department of Health - Animals (Control of Experiments) of the Hong Kong S.A.R. government.

Method details

Stereotaxic injection

Mice were anesthetized with ketamine and xylazine (100 mg/kg and 10 mg/kg, respectively) and placed in a stereotaxic instrument (RWD, Shenzhen, China). A small craniotomy hole was made over the targeted area. Viral vectors were micro-injected into the right side of the mice’s brains by standard stereotaxic procedures. AAV vectors were injected at 0.05 – 0.1 ul per minute. The micro-syringe was left in place for an extra 10 min before the withdrawal. The final virus concentration was around 2–3 × 10¹² viral genomes GC/ml.

For whisker behavioral experiments, 0.5 μl of AAV-CamKIIα-EYFP or AAV-CamKIIα-MscL-G22S-EYFP was injected into the somatosensory cortex at AP -1.70 mm, ML -2.10 mm, DV --0.60 mm.

For real-time calcium imaging of acute brain slices, 0.5 μl viral vectors (hSyn-GCaMP6s) were injected into the somatosensory cortex or visual cortex (ML -2.5 mm, AP -3.5 mm, DV: -1 mm) in 4-6 week-old mice.

For fiber photometry experiments, mice were injected with viral vectors on the right side of the somatosensory cortex (DV: -0.6 mm). The volume ratio of the viral vector mixture was 1:1 for AAV-CamKIIα-jRGECO1a and AAV-CamKIIα-EYFP or AAV-CamKIIα-MscL-G22S-EYFP. In the astrocyte experiment, AAV-CamKIIα-jRGECO1a or AAV-GFAP-GCamP6f were injected into the somatosensory cortex. The E/I balance experiment injected CamKIIα: MscL-G22S-EYFP or CamKIIα: EYFP mixed with mDlx5/6-jRGECO1a into the somatosensory cortex. After micro-syringe withdrawal, a 1.25 mm optical fiber was inserted into the location of the syringe.

Fiber photometry recording

Fiber photometry recording was performed at least four weeks after viral injection. Mice expressing genetically encoded calcium sensors (jRGECO1a/GCamP6f) were anesthetized with 1 - 2% isoflurane. Eye ointment was applied to prevent corneal drying. The fur above the head was shaved, and ultrasound gel was applied to the shaved area to facilitate coupling with the transducer. Mice were treated with a series of ultrasound stimulation parameters. The implanted fiber was connected to a fiber optic meter (Thinker Tech Nanjing BioScience Inc) through an optical fiber patch cord to guide the light. The emission and reception wavelengths were 470 nm with 30 nm bandwidth or 510 nm with 25 nm bandwidth, respectively. Data was collected at 100 Hz and analyzed using a customized MATLAB script. The fluorescence change (ΔF/F₀) was calculated as (F-F₀)/F₀, where F₀ is the baseline fluorescence signal.

Linear regression was employed to quantitatively model the relationship between ultrasound intensity (independent variable, X) and the cellular stress response, measured as Peak ΔF/F₀ % (dependent variable, Y), for each experimental group. Peak ΔF/F₀ % values were collected at both pressure levels, providing discrete (X, Y) data points for each condition. Using GraphPad Prism, a linear model of the form Y = mX + b was fit to each dataset, where m represents the slope and b the Y-intercept.

Whisker behavior measurement

All whiskers, except for the C2 on both sides, were trimmed before the experiment. Mice were head-restrained in a narrow chamber to fix their position and expose the nose and whisker area. A digital camera (Canon LEGRIA HF M506) was placed ∼10 cm from the whisker. An ultrasound indicator light was placed on the mice’s right side, and a white card was used to separate the ultrasound indicator light from the mouse to prevent the flashing indicator light from visually interfering with the results of the ultrasound stimulation. After setup, mice were allowed 5 min for habituation while the left C2 whisker was filmed at 25 Hz. The movement of the C2 whisker was recorded synchronously by the camera, both with and without sonication. Whisker movement analysis was performed during the 10s period of ultrasound stimulus and compared to the 10s pre-stimulus period. Whisker angle and curvature were tracked using automated, freely available software “Whisker Tracking.”^33,34 All whisker tracking was done on mice with only C2 whiskers. The whisker angle was measured as the angle between the whisker and a line perpendicular to the midline of the mouse. Whisker angular velocity (deg/s) was computed as the change in the whisker angle for the period analyzed.

Ultrasound stimuli in mice

The ultrasound stimulation system was set up by connecting the output of a function generator (AFG251, Tektronix) to the input of the power amplifier (A075, Electronics & Innovation Ltd) with BNC wiring, then connected to the output of an amplifier of a 0.5 MHz transducer (I7-0012-P-SU, Olympus). The same ultrasound stimulation setup, featuring a transducer coupled via a water tube filled with degassed water,³⁵ was employed in both the ex vivo and in vivo experiments. The ultrasound parameters were as follows: center frequency of 0.5 MHz, peak pressure ranging from 0.15 to 0.4 MPa, pulse width of 500 μs, stimulation duration of 300 ms, and a pulse repetition frequency (PRF) of 1 kHz.

To measure and calibrate the beam profile, the output of the wave-guided transducer was scanned with the acoustic mapping system. The transducer, coupled via a water tube, was immersed in a water tank filled with degassed water. The hydrophone (Onda, HGL400) was held by a motorized 3D movement platform, with movement synchronized to the transducer driving signal to characterize the beam profile. The hydrophone signal was digitized, collected and processed by a customized script with MATLAB. Briefly, the signal was filtered to the transducer center frequency, denoised, had its amplitude extracted, and was smoothed to form a 2D mapping of the scanned plane. At the driving energy corresponding to our experiments, the guided transducer was scanned with a 0.5 mm step size, resulting in a lateral beam profile as shown in (Figure S3).

Preparation and fluorescence imaging of acute brain slices

Four weeks after virus injection, mice were anesthetized with an intraperitoneal injection of Ketamine and Xylazine (100 mg/kg and 10 mg/kg, respectively) and perfused with ice-cold oxygenated NMDG ACSF buffer containing (in mM): 92 N-methyl-D-glucamine (NMDG), 2.5 KCl, 1.25 NaH₂PO₄, 20 NaHCO₃, 10 HEPES, 25 Glucose, two thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 0.5 CaCl₂, 10 MgSO₂, 12 NAc. The pH of the ASCF was 7.3-7.4. The brains were immediately removed and placed in an ice-cold oxygenated slicing buffer. The brains were sectioned into 300 μm-thick slices using vibratome (Leica VT1200), and the slices were incubated at 34°C at NMDG ACSF for 10-12 min, followed by N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid (HEPES ACSF that contained (in mM): 92 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 Glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 2 CaCl₂, 2 MgSO₂ for at least 1 h at 25 °C. The brain slices were transferred to a slice chamber for calcium fluorescence imaging and were continuously perfused with standard ACSF that contained (in mM): 119 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 24 NaHCO₃, 1.25 Glucose, 2 CaCl₂, 2 MgSO₂ at 2-3 ml/min. 20 μM Yoda1 diluted from a 10 mM stock solution in DMSO was used for activating Piezo1. Ruthenium red (RR, final conc.: 30 μM, a blocker of TRPV1, 2, 4), Gd3⁺ blocker (final conc.: 20 μM, a stretch-activated channel inhibitor), ML204 blocker (final conc.: 20 μM, a TRPC4,5 inhibitor), and Ononetin blocker (final conc.: 20 μM, a TRPMP blocker) were added to the medium and incubated with brain slices separately. An ultrasound transducer with a plastic tube was placed above the targeted region, and a modified inverted epifluorescence microscope was used to record the neural activity of neurons. The fluorescence was recorded using a 488 nm filter.

Deaf model

A deaf mouse model was established by subcutaneous injection of kanamycin (1000 mg/kg), followed by intraperitoneal administration of furosemide (400 mg/kg) within 30 minutes.³⁶ The experiments involving ultrasound stimulation and fiber photometry will be conducted after 7 days of inducing the deaf model.

Immunohistochemical fluorescence staining

Mice were anesthetized with ketamine and xylazine (100 mg/kg and 10 mg/kg, respectively). 90 mins after ultrasound treatment (40 minutes with a 10-second stimulation interval between US pulses), mice were perfused with 0.9% saline and 4% paraformaldehyde (PFA) (cat. no. P1110, Solarbio) in PBS. After dissection, brains were post-fixed overnight in 4% PFA. Coronal sections were prepared from brain regions spanning +1.6 mm to 2.4 mm of Bregma for somatosensory sections. Coronal brain slices, 40 μm thick, were collected using a vibratome. Slices were blocked using blocking buffer (10% normal goat serum, 5% BSA, 0.3% Triton X-100, 1X PBS) and incubated overnight in a primary antibody solution diluted in the blocking buffer. Slices were washed three times with PBS and incubated with secondary antibodies diluted in blocking buffer for 2 hours at room temperature. Slices were washed three times, coverslips were dried, and the samples were mounted on glass slides using Prolong Diamond Antifade Mountant with DAPI. The primary antibodies used were c-Fos (2250, Cell Signaling Technology, dilution 1:500), GAD67 (5406, Sigma-Aldrich, dilution 1:500), and MAP2 (5392, abcam, dilution 1:500). Secondary antibody used was goat anti-rabbit IgG (H+L), Alexa Fluor 555 (A-214428, Invitrogen, dilution 1:1,000). Each sample was divided into three sets, and 1 set was used to stain for c-Fos expression. Each set contained 5-8 brain slices. All brain slices were imaged using the confocal microscope (TCS SP8, Leica) or a fluorescence imaging system (Nikon Eclipse Ti2-E Live-cell) in the ULS facilities at The Hong Kong Polytechnic University.

RNA extraction and real-time qRCR

RNA was collected from mouse hippocampus tissues using the TaKaRa MiniBEST Universal RNA Extraction Kit (Takara #9767) according to the manufacturer’s instructions. RNA concentrations were measured using a NanoDrop Microvolume Spectrophotometer (ThermoFisher). 1 μg RNA was reverse-transcribed using PrimeScript™ RT Master Mix (Takara #RR036A), according to the manufacturer’s instructions, using a C1000 Touch thermal cycler (Bio-Rad). For real-time qPCR, 0.8 μl cDNA was mixed with appropriate forward and reverse primers (final concentration 200 nM), TB Green Premix Ex Taq (Tli RNaseH Plus) (Takara #RR420A), and H₂O were added to prepare a reaction volume of 10 μl per well. PCR was performed on a CFX96 Touch™ system (Bio-Rad), for 40 cycles according to the recommended instructions of the PCR supermix manufacturer. Results are expressed as a fold change compared to the appropriate control, mean ± SEM of 3 independent experiments.

Data processing

The number of cells showing c-Fos was counted using ImageJ, and the number of c-Fos⁺ cells in ROIs was calculated. The counting was single-blinded and performed by an experimenter who did not know the groups beforehand. Behavioral test data analysis was conducted by experimenters who were blind to experimental conditions, using whisker tracking software.

Quantification and statistical analysis

All statistical analyses were conducted with GraphPad Prism, which was also used to generate the graphs. Data are presented as the mean ± SEM. Statistical significance was determined as follows: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, and ∗∗∗∗P < 0.0001; ns=not significant differences are also indicated. The specific statistical tests used for each figure are reported in the respective legends.

Acknowledgments

This work was financially supported by the Hong Kong Research Grants Council Collaborative Research Fund (C5053-22 GF), the General Research Fund (15126524, 15224323 and 15104520), the National Key Research and Development Program of the Ministry of Science and Technology of China (2023YFC2410900), and internal funding from the the Hong Kong Polytechnic University (G-SACD and 1-YWDQ), the Research Center for Non-invasive Brain Computer Interface (1-CE0M), and the Research Institute of Smart Ageing (1-CDJM). The authors would like to thank the facility and technical support from the University Research Facility in Life Sciences (ULS) and the University Research Facility in Behavioral and Systems Neuroscience (UBSN) of The Hong Kong Polytechnic University.

Author contributions

Conceptualization, Q.X., Z.Q., and L.S.; methodology, Q.X., Z.Q., and L.S.; investigation, Q.X., D.L., X.Z., D.Z., Y.J., J.J., and K.F.W.; data analysis, Q.X., D.L., X.Z., X.H., S.M., K.F.W., and X.H.; article preparation, Q.X., Z.Q., and L.S.; funding acquisition, L.S.

Declaration of interests

The authors declare no competing interests.

Published: November 12, 2025