Remote Excitation of Neuronal Circuits Using Low-Intensity, Low-Frequency Ultrasound

Sound and thought seem to belong to different worlds. Sound is pressure, a mechanical wave moving through air, through water, and through bone. Thought is electrochemistry, with ion channels opening and closing, neurons firing across synapses. These two phenomena do not obviously connect. And yet, William Tyler and colleagues aimed a beam of sound at living brain tissue, and the neurons fired. No electrodes, no drugs, no light. Just sound waves, and a circuit came alive. To understand why that matters, you have to know what neuroscientists are working with when they want to reach the brain. Deep brain stimulation, or DBS, is extraordinarily effective for conditions like Parkinson's disease, but it requires surgically implanted devices. Transcranial magnetic stimulation is noninvasive but cannot reach deep structures with precision. The tradeoff has always been this: the deeper and more targeted you want to go, the more invasive the approach. Tyler and colleagues framed this as an unmet need and turned to ultrasound as a potential solution. The key physical property is simple but powerful — ultrasound propagates through skull bone and other tissues in a focused beam. Modeling and ex vivo skull measurements suggest optimal transmission occurs around 0.60 to 0.70 megahertz, and phased transducer arrays can focus pulsed ultrasound through a human skull to a point within about one millimeter of an intended target. The question the paper sets out to answer is whether ultrasound, at low intensities and low frequencies, can actually modulate neuronal activity — not just pass through tissue and change what neurons do. The experimental system was hippocampal slice cultures and whole ex vivo mouse brains. Hippocampal slices, four hundred micrometers thick and maintained in an incubator between seven and twelve days in vitro, provided the team an accessible, well-characterized neural circuit. Ex vivo brains, removed after euthanasia and bulk-loaded with a calcium indicator, allowed them to test whether effects survived the step up to intact brain tissue. The ultrasound source was a custom-built lead zirconate titanate transducer, the standard piezoelectric material, with a center frequency of 0.53 megahertz and a two-decibel bandwidth of sixty-five percent. It had peaks at 0.44 and 0.66 megahertz. Their primary stimulus waveform, called LILFU-1, ran at 0.44 megahertz with tone bursts of twenty-two point seven milliseconds duration, ten acoustic cycles per burst, and a pulse repetition frequency swept from zero to one hundred hertz across a total of two hundred fifty tone bursts. At the tissue position, LILFU-1 had a pulse average intensity of two point nine watts per square centimeter and a temporal average intensity of just twenty-three milliwatts per square centimeter. That temporal average is key — it sits squarely in the range of nonthermal therapeutic ultrasound effects, well below the Food and Drug Administration output limits for diagnostic imaging. To observe neuron responses, the team used fluorescent indicators rather than electrodes. Voltage-gated ion channels were the first target. They loaded slices with CoroNa Green AM, a sodium-sensitive dye, and applied LILFU-1. CA1 pyramidal neurons showed clear sodium transients — a fluorescence change of about five percent above baseline. When they applied tetrodotoxin, the classic voltage-gated sodium channel blocker, those transients vanished. The conclusion is direct: LILFU opens voltage-gated sodium channels. Sodium entry was followed by calcium entry. With the calcium indicator OGB-1 AM, LILFU-1 evoked calcium transients in pyramidal neurons with a fluorescence change of one point fourteen plus or minus zero point ten — more than a doubling of signal relative to baseline — and similar responses in glial cells. Presynaptic boutons in the stratum radiatum, the fiber layer where CA3 axons contact CA1 neurons, showed calcium transients around zero point seventy-six. When cadmium was applied to block voltage-gated calcium channels, the OGB-1 signal nearly disappeared. And when tetrodotoxin was applied, blocking the upstream sodium conductance, it eliminated roughly eighty-five percent of the calcium signal. That pharmacological double-dissection indicates exactly what is happening: the sound wave mechanically perturbs the membrane, voltage-gated sodium channels open, the cell depolarizes, and voltage-gated calcium channels follow. In whole-cell current-clamp recordings, individual LILFU tone bursts could trigger single action potentials in CA1 pyramidal neurons. The domino chain was complete. However, individual neurons firing is not the same as a brain circuit working. The more important question was whether ultrasound could drive synaptic transmission — the process by which one neuron releases neurotransmitter onto another. This is where the story gets genuinely striking. Tyler and colleagues used mice expressing synaptopHluorin, a pH-sensitive optical reporter in synaptic vesicles. When a vesicle fuses with the presynaptic membrane and releases neurotransmitter, its interior becomes exposed to the slightly alkaline extracellular space, and synaptopHluorin fluoresces. It's a direct readout of exocytosis, the moment of release. LILFU-1 produced a mean fluorescence change of eighteen point fifty-two plus or minus two point two percent at individual release sites across one hundred forty-eight sites from fifteen slices. Because each released vesicle produces roughly a one to two percent fluorescence change, the team estimated LILFU-1 was driving release of around fifteen vesicles per release site. That's a real, physiologically meaningful synaptic event. They then confirmed it was genuine synaptic transmission, not some mechanical artifact. Tetrodotoxin almost completely blocked vesicular release, confirming action-potential dependence. Blocking excitatory synaptic receptors with CNQX and APV reduced the synaptopHluorin signal by about fifty percent, indicating that much of the response was propagating through network activity — one neuron triggering another. The kinetics and amplitudes of LILFU-triggered release were nearly identical to those produced by conventional electrical stimulation of the same CA3-to-CA1 pathway with a monopolar electrode. The definitive test was SNARE dependence. SNARE proteins, or soluble NSF attachment protein receptors, are the molecular machinery that drives vesicle fusion with the presynaptic membrane. They are absolutely required for normal neurotransmitter release. The team applied botulinum neurotoxin type A at two hundred fifty nanograms per milliliter for twenty-four to thirty-six hours, which cleaves SNAP-25, one of the core SNARE proteins. The synaptopHluorin response to LILFU-1 was nearly abolished. Ultrasound-induced exocytosis is SNARE-mediated. This is not a mechanical squeezing of transmitter out of the terminal; it is the same release machinery the brain uses during normal signaling, recruited by a sound wave. Safety mattered, and the team tested it directly. Slices were stimulated every eight minutes for thirty-six to forty-eight hours — three hundred sixty LILFU-1 stimuli for the forty-eight-hour protocol. Confocal imaging of thy-1 to YFP neurons showed no difference in membrane structure compared to unstimulated controls. No cavitation, the violent bubble formation that can damage tissue, was observed by optical microscopy at the intensities used. The peak rarefactional pressure in their studies was about one megapascal, far below the pressures associated with cavitational damage in most soft tissues. What Tyler and colleagues have established, then, is a complete chain of evidence. Low-intensity, low-frequency ultrasound can open voltage-gated sodium channels, trigger calcium entry through voltage-gated calcium channels, drive action potentials, and activate SNARE-mediated synaptic vesicle release in hippocampal circuits, all without electrodes, without drugs, and without detectable membrane damage. Across a given imaging field, LILFU-1 stimulated roughly thirty percent of neurons. These effects were demonstrable not just in slice cultures but in ex vivo whole mouse brains, transmitting sound from the ventral surface and imaging calcium responses on the dorsal surface. What remains open is substantial. The molecular identity of the mechanically sensitive channels, and exactly how ultrasound couples to their gating, is not resolved. Optimal waveform parameters, including frequency, tone-burst duration, and pulse repetition rate for targeting specific cell types or spatial locations, haven't been mapped. Long-term safety data beyond forty-eight hours does not exist in this paper, and the authors are careful to flag that repeated prenatal ultrasound has been shown to disrupt neuronal migration in developing mouse cortex. Translation to the living human brain requires navigating the skull's reflective, refractive, and absorptive effects on the ultrasound field. But the core finding stands on its own. A sound wave aimed at a neural circuit provokes a response from that circuit through the same ion channels and vesicle fusion machinery it uses during ordinary thought. That is not a metaphor for something else. That is the result, stated plainly, and it changes what noninvasive access to the brain might one day mean. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.