Focused ultrasound excites cortical neurons via mechanosensitive calcium accumulation and ion channel amplification

The study addresses how focused ultrasound (FUS) excites mammalian neurons, a key unknown limiting the development of non-invasive neuromodulation technologies. Established non-invasive approaches (TMS, tDCS) lack depth and spatial precision, whereas FUS can modulate deep brain regions with millimeter precision and has evoked neural and behavioral responses in animals and humans without genetic or chemical alterations. Yet the underlying cellular and molecular mechanisms remain unclear because ultrasound can induce multiple physical effects—mechanical force, heating, and cavitation—complicating mechanistic attribution. Additionally, off-target auditory effects in small animals confound in vivo mechanistic studies. This work aims to define the biophysical basis and molecular pathway of ultrasonic excitation in primary cortical neurons, determining whether temperature, cavitation, or large-scale deformation are involved, and identifying the channels and intracellular signaling that convert mechanical effects into neuronal spiking.

Prior work shows transcranial FUS (0.25–1 MHz, 1–100 W/cm² ISPPA) can elicit neural and behavioral responses across species, including humans, often with latencies of ~70–300 ms and safe, repeatable stimulation. Multiple mechanisms have been proposed: thermal effects, cavitation (including intramembrane cavitation theories), and direct mechanical interactions affecting ion channels. Studies reported possible roles for mechanosensitive channels and indirect contributions via synaptic transmission or astrocytic pathways. However, no consensus exists, and auditory confounds complicate in vivo interpretation. Theoretical models predict cavitation-related hyperpolarizing currents, while other work suggests direct mechanical gating of mechanosensitive channels (e.g., Piezo, TRP families) or lipid-mediated effects on channel gating. The present study builds on this literature by systematically excluding thermal and cavitation contributions and delineating specific mechanosensitive and amplifier channels in cortical neurons.

• Cell preparation and culture: Primary murine cortical neurons were cultured on acoustically transparent Mylar films for 2D studies and in 3D collagen gels for validation. Neurons expressed GCaMP6f for calcium imaging; some co-expressed mCherry as a temperature reporter. Voltage imaging used the Ace2N genetically encoded indicator. Spiking HEK cells were also used for certain voltage controls.
• Ultrasound stimulation: A 300 kHz focused ultrasound transducer (5.2 mm FWHM focal diameter) submerged in degassed water delivered continuous-wave pulses (0–500 ms) with intensities 0–15 W/cm². Inter-pulse interval was 20 s. Pulsed ultrasound (1–1.5 kHz PRF) and a 670 kHz transducer were also tested. Chirped waveforms were used to mitigate standing waves. Acoustic calibration used fiber optic hydrophones; intensity computed per standards. Mechanical index at typical conditions was ~0.9.
• Imaging and analysis: Epifluorescence imaging at 100 Hz (calcium) and 200 Hz (voltage) using a 10x water immersion objective recorded dF/F. Single-cell traces per dish were averaged; area under the curve (AUC) from 0–5 s post-onset quantified responses. Onset delay estimated via 4-parameter logistic fit (threshold 0.2% increase). Biological replicates were independent dishes (50–300 cells per dish).
• Temperature and cavitation/deformation assessments: Temperature changes measured by a fiber optic thermometer adjacent to cells and by monitoring mCherry fluorescence. Ultra-high-speed imaging (5 Mfps; 256 frames over 51.2 µs) at 10x/40x captured the cell and medium starting 100 ms after US onset to detect bubble formation or membrane deformation; additional bright-field imaging observed longer-timescale deformation. Degassed vs atmospherically gassed media compared for cavitation susceptibility.
• Mechanical perturbation and synaptic blockade: Actin cytoskeleton was depolymerized with cytochalasin D (1 µM) to alter cellular mechanics (viability and baseline excitability controlled). Postsynaptic blockers AP5 and CNQX (1 µM each) tested synaptic contribution.
• Ion and channel dependence: Sodium channel blocker TTX (1 µM) probed AP contribution; calcium-free media assessed extracellular Ca2+ necessity; thapsigargin assessed ER release (supplementary). Pharmacology targeted mechanosensitive channels and signaling: gadolinium(III) (20 µM, nonselective mechanosensitive channel inhibitor), ruthenium red (1 µM; TRPV1/2/4 pores), suramin (60 µM; GPCR inhibition), GsMTx4 (10 µM; Piezo1/TRPC1 gating).
• Genetic perturbations: CRISPR/Cas9 knockdown via lentiviral delivery targeted TRPM7, TRPP1, TRPP2, Piezo1, TRPC1, and TRPM4 (editing efficiencies ~20–43%). Non-targeting sgRNA served as control. Effects on ultrasound-evoked calcium responses were quantified.
• Amplifier channels: TRPM4 involvement tested by CRISPR knockdown; T-type Ca2+ channels probed with TTA-P2 (3 µM) selective blocker.
• Overexpression: Lentiviral overexpression (hSyn promoter) of TRPC1, TRPP2, TRPM4 (and TRPV1 as control) assessed impact on response amplitude, threshold, and kinetics; expression verified by immunostaining.
• Statistics: Comparisons used t-tests or one-way ANOVA with Tukey post hoc as appropriate; data reported as mean ± SEM with replicate numbers provided for each experiment.

• Robust neuronal activation by FUS: Primary cortical neurons exhibited strong calcium responses to 300 kHz CW ultrasound with amplitudes increasing monotonically with intensity (0–15 W/cm²) and pulse duration (0–500 ms). Responses exceeded spontaneous activity. Onset delay was ~200 ms and time to peak ~1.7 s; both decreased with increasing intensity.
• Parameter generality: Similar amplitudes and delays were observed with pulsed waveforms (1–1.5 kHz PRF) and at 670 kHz. Responses were similar in 3D collagen culture and with chirped waveforms mitigating standing waves.
• Non-thermal, non-cavitational mechanism: Temperature rise at 15 W/cm², 500 ms was 0.005 ± 0.003 °C; all 300 kHz parameters yielded <0.02 °C increases. mCherry fluorescence showed no thermal change during responses. Degassed vs gassed media produced no significant differences. Ultra-high-speed imaging detected no bubble formation; MI ≈ 0.9 (below cavitation threshold >1.9). No large membrane deformation was observed within cycles or over the pulse.
• Mechanical contribution: Disrupting actin with cytochalasin D significantly reduced ultrasound-evoked calcium responses (p = 0.0061), indicating involvement of cellular mechanics. Synaptic blockade with AP5+CNQX did not significantly affect responses (p ≈ 0.41), supporting a cell-autonomous mechanism.
• Extracellular Ca2+ initiates depolarization: TTX partially reduced responses (ANOVA p = 0.0004), indicating direct Ca2+ entry precedes full AP firing. Voltage imaging showed depolarization during FUS in normal media, abolished in Ca2+-free media, confirming extracellular calcium is essential for initiation; ER Ca2+ release made no major contribution.
• Mechanosensitive channels implicated: Gd3+ reduced response amplitude by ~60% (significant), supporting mechanosensitive channel involvement. Ruthenium red (TRPV1/2/4) and suramin (GPCRs) had no significant effects. GsMTx4 caused a modest but significant reduction (p = 0.0245), implicating Piezo1/TRPC1 gating.
• CRISPR knockdowns: TRPP1 (p = 0.0208), TRPP2 (p = 0.0084), and TRPC1 (p = 0.0232) knockdowns significantly reduced responses; Piezo1 and TRPM7 knockdowns showed no significant effect (Piezo1 trend toward minor reduction). Normalized estimates indicated largest contributions from TRPP2 and TRPC1.
• Signal amplification: TRPM4 knockdown markedly reduced responses (p = 0.0024), identifying calcium-activated nonselective cation channel TRPM4 as a key amplifier. Blocking T-type Ca2+ channels with TTA-P2 (3 µM) significantly reduced response magnitude (p = 0.0331), implicating low-threshold spiking in burst generation.
• Overexpression enhances sensitivity: Overexpressing TRPC1 and TRPP2 increased response amplitudes and enabled stronger activation at lower intensities; TRPM4 overexpression markedly increased amplitudes and accelerated onset (<100 ms at 15 W/cm²). TRPV1 overexpression had no effect, consistent with its lack of involvement.
• No evidence for intramembrane cavitation-induced hyperpolarization: Voltage imaging did not reveal time-averaged hyperpolarization during FUS, contradicting predictions of intramembrane cavitation models.

The findings indicate that focused ultrasound excites cortical neurons primarily via mechanical effects on the membrane that gate specific calcium-permeable mechanosensitive ion channels, notably the TRPP1/2 complex and TRPC1, with possible minor contributions from Piezo1. The initial calcium influx accumulates over ~200 ms until it activates calcium-gated depolarizing channels such as TRPM4, leading to membrane depolarization and opening of voltage-gated calcium channels, including T-type channels, which together produce robust, seconds-long bursting responses. This cell-autonomous pathway does not require synaptic transmission and occurs without significant heating, cavitation, or large-scale deformation. Response latencies and effective intensity ranges align with in vivo observations, strengthening the relevance of the in vitro model. While the pathway accounts for observed kinetics, a contribution from direct effects on voltage-gated channels cannot be fully excluded, echoing some prior reports. The mechanistic map clarifies how FUS translates mechanical energy into neuronal activation and identifies molecular nodes that modulate sensitivity and kinetics, informing both basic neuroscience applications and the design of sonogenetic strategies.

This study provides a comprehensive biophysical and molecular mechanism for ultrasound-induced excitation of cortical neurons: mechanical forces from FUS open calcium-permeable mechanosensitive channels (TRPP1/2, TRPC1), initiating calcium accumulation that activates TRPM4 and engages voltage-gated channels (including T-type), culminating in burst firing. Thermal effects, cavitation, large-scale deformation, and synaptic transmission are not required. Pharmacological and genetic perturbations validate the roles of these channels, and overexpression of TRPC1, TRPP2, and TRPM4 enhances responsiveness and lowers activation thresholds, suggesting practical routes for sonogenetic sensitization. Future work should investigate cell-type-specific channel expression and contributions in vivo, optimize channel combinations for targeted neuromodulation, quantify mechanical transduction at the nanoscale, and assess long-term safety and efficacy of sensitization strategies.

• In vitro model: Primary neuron cultures may not fully recapitulate in vivo tissue architecture and neuromodulatory context, though parameters and latencies matched in vivo reports.
• Partial genetic perturbation: CRISPR editing efficiencies were partial (~20–43%), potentially underestimating channel contributions; compensatory expression cannot be excluded.
• Pharmacology constraints: Selective inhibitors are limited for some channels; combinations of blockers were not tested due to viability/solvent concerns.
• Temporal/spatial resolution limits: Voltage sensor kinetics (200 Hz) cannot capture MHz oscillations; high-speed imaging began 100 ms after onset, possibly missing early transient events; optical resolution cannot detect nanoscale intramembrane cavitation.
• Thermal measurement locality: Temperature assessments reflect bulk media or cytoplasm; nanoscale membrane-localized heating cannot be entirely ruled out but is unlikely given diffusion scales and measured changes.
• Standing waves: Though mitigated (angled transducer, chirped waveforms), standing waves cannot be completely eliminated in vitro.
• Mechanistic scope: While mechanosensitive and amplifier channels are implicated, direct effects on voltage-gated channels may contribute but were not dissected here.