Ultrasound activates mechanosensitive TRAAK K+ channels through the lipid membrane

Ultrasound modulates the electrical activity of excitable cells and offers advantages over other neuromodulatory techniques; for example, it can be noninvasively transmitted through the skull and focused to deep brain regions. However, the fundamental cellular, molecular, and mechanistic bases of ultrasonic neuromodulation are largely unknown. Here, we demonstrate ultrasound activation of the mechanosensitive K⁺ channel TRAAK with submillisecond kinetics to an extent comparable to canonical mechanical activation. Single-channel recordings reveal a common basis for ultrasonic and mechanical activation with stimulus-graded destabilization of long-duration closures and promotion of full conductance openings. Ultrasonic energy is transduced to TRAAK through the membrane in the absence of other cellular components, likely increasing membrane tension to promote channel opening. We further demonstrate ultrasonic modulation of neuronally expressed TRAAK. These results suggest mechanosensitive channels underlie physiological responses to ultrasound and could serve as sonogenetic actuators for acoustic neuromodulation of genetically targeted cells.

Manipulating cellular electrical activity is central to basic research and is clinically important for the treatment of neurological disorders including Parkinson’s disease, depression, epilepsy, and schizophrenia (1–4). Optogenetics, chemogenetics, deep brain stimulation (DBS), transcranial electrical stimulation, and transcranial magnetic stimulation are widely utilized neuromodulatory techniques, but each is associated with physical or biological limitations (5). Transcranial stimulation affords poor spatial resolution; deep brain stimulation and optogenetic manipulation typically require surgical implantation of stimulus delivery systems, and optogenetic and chemogenetic approaches necessitate genetic targeting of light- or small-molecule–responsive proteins.Ultrasound was first recognized to modulate cellular electrical activity almost a century ago, and ultrasonic neuromodulation has since been widely reported in the brain, peripheral nervous system, and heart of humans and model organisms (5–12). Ultrasonic neuromodulation has garnered increased attention for its advantageous physical properties. Ultrasound penetrates deeply through biological tissues and can be focused to sub-mm (3) volumes without transferring substantial energy to overlaying tissue, so it can be delivered noninvasively, for example, to deep structures in the brain through the skull. Notably, ultrasound generates excitatory and/or inhibitory effects depending on the system under study and stimulus paradigm (5, 13, 14).The mechanisms underlying the effects of ultrasound on excitable cells remain largely unknown (5, 13). Ultrasound can generate a combination of thermal and mechanical effects on targeted tissue (15, 16) in addition to potential off-target effects through the auditory system (17, 18). Thermal and cavitation effects, while productively harnessed to ablate tissue or transiently open the blood–brain barrier (19), require stimulation of higher power, frequency, and/or duration than typically utilized for neuromodulation (5). Intramembrane cavitation or compressive and expansive effects on lipid bilayers could generate nonselective currents that alter cellular electrical activity (5, 13). Alternatively, ultrasound could activate mechanosensitive ion channels through the deposition of acoustic radiation force that increases membrane tension or geometrically deforms the lipid bilayer (5, 15). Consistent with this notion, behavioral responses to ultrasound in Caenorhabditis elegans require mechanosensitive, but not thermosensitive, ion channels (20), and a number of mechanosensitive (and force-sensitive, but noncanonically mechanosensitive) ion channels have been implicated in cellular responses to ultrasound including two-pore domain K⁺ channels (K2Ps), Piezo1, MEC-4, TRPA1, MscL, and voltage-gated Na⁺ and Ca²⁺ channels (20–24, 25). Precisely how ultrasound impacts the activity of these channels is not known.To better understand mechanisms underlying ultrasonic neuromodulation, we investigated the effects of ultrasound on the mechanosensitive ion channel TRAAK (26, 27). K2P channels including TRAAK are responsible for so called “leak-type” currents because they approximate voltage- and time-independent K⁺-selective holes in the membrane, although more complex gating and regulation of K2P channels is increasingly appreciated (28, 29). TRAAK has a very low open probability in the absence of membrane tension and is robustly activated by force through the lipid bilayer (30–32). Mechanical activation of TRAAK involves conformational changes that prevent lipids from entering the channel to block K⁺ conduction (31). Gating conformational changes are associated with shape changes that expand the channel and make it more cylindrical in the membrane plane upon opening. These shape changes are energetically favored in the presence of membrane tension, resulting in a tension-dependent energy difference between states that favors channel opening (31). TRAAK is expressed in neurons and has been localized exclusively to nodes of Ranvier, the excitable action potential propagating regions of myelinated axons (33, 34). TRAAK is found in most (∼80%) myelinated nerve fibers in both the central and peripheral nervous systems, where it accounts for ∼25% of basal nodal K⁺ currents. As in heterologous systems, mechanical stimulation robustly activates nodal TRAAK. TRAAK is functionally important for setting the resting potential and maintaining voltage-gated Na⁺ channel availability for spiking in nodes; loss of TRAAK function impairs high-speed and high-frequency nerve conduction (33, 34). Changes in TRAAK activity therefore appear well poised to widely impact neuronal excitability.We find that low-intensity and short-duration ultrasound rapidly and robustly activates TRAAK channels. Activation is observed in patches from TRAAK-expressing Xenopus oocytes, in patches containing purified channels reconstituted into lipid membranes, and in TRAAK-expressing mouse cortical neurons. Single-channel recordings reveal that canonical mechanical and ultrasonic activation are accomplished through a shared mechanism. We conclude that ultrasound activates TRAAK through the lipid membrane, likely by increasing membrane tension to promote channel opening. This work demonstrates direct mechanical activation of an ion channel by ultrasound using purified and reconstituted components, is consistent with endogenous mechanosensitive channel activity underlying physiological effects of ultrasound, and provides a framework for the development of exogenously expressed sonogenetic tools for ultrasonic control of neural activity.