Controlling the Brain with Ultrasound (Cell Crunch #55)
Action potentials controlled in living mice, from outside the skull, with sound.
Good morning. A protein called hsTRPA1, placed inside of mouse neurons, opens in response to ultrasound; calcium rushes in, the neurons fire. It’s optogenetics, but with sound instead of light.
Now, a closer look…
In its basic form, optogenetics works like this:
An opsin protein (originally plucked from a photosynthetic algae, Chlamydomonas) is packaged into a virus. The virus is injected into a mouse, where it travels to the brain and infects a neuron. The virus delivers its payload — the opsin — and that protein bumps its way towards the cell membrane and fuses into the lipid bilayer. As photons strike this opsin, it opens its pore; calcium rushes in, and the neurons fire action potentials.
This basic method, which has been used by researchers for more than 15 years, has been used to delineate the neural circuits underlying just about every behavior, from thirst to fear and aggression.
The method from Chalasani’s lab — called sonogenetics — works in almost exactly the same way. But instead of using an opsin to trigger action potentials with light, it uses a mechanosensitive protein to trigger action potentials with targeted ultrasound. The exact mechanism by which ultrasound opens these proteins is unclear, but it seems to be related to mechanical forces that push and pull the lipid bilayers apart.
Chalasani’s paper is the first demonstration of sonogenetics in a mammal. It is also, to my understanding, the first time that action potentials have been observed, and controlled using ultrasound, in any neuron.
Protein in a Haystack
Step back seven years, to 2015. Chalasani’s team has just shown that neurons in C. elegans (a nematode, a worm) could be controlled with ultrasound. In that paper, neurons expressing TRP-4 — a “pore-forming subunit of a mechanotransduction channel” — open in response to ultrasound pulses, allowing calcium to rush into cells. The worms can be crudely controlled with ultrasound.
A million dollar grant from the BRAIN Initiative quickly followed for Chalasani, who took that same TRP-4 protein from C. elegans and tested it in mammalian cells, in a petri dish.
Ultrasound was blasted at those mammalian cells and … nothing.
It was an abject failure: The protein doesn’t get expressed properly in mammalian cells. The protein that had worked so well in C. elegans was a dead-end.
This study is the phoenix of that failure. It’s the culmination of seven years of work; 191 candidate proteins, each thought to be mechanosensitive, were tested in human kidney cells. Cells were transfected with each candidate, as well as a genetically encoded calcium sensor, and closely monitored. If cells blasted with ultrasound had spikes in calcium levels, those proteins were pursued in later experiments.
Cells expressing hsTRPA1 had the most robust response to the ultrasound, but the results were heterogeneous; only about 45 percent of cells became active in response to the ultrasound. Still, it was the best protein out of the 191 tested.
A mechanism explaining why hsTRPA1 opens in response to ultrasound, though, remains elusive. In the paper, Chalasani’s team cut off parts of hsTRPA1 to determine which regions might be responsible for its sonogenetic abilities. The N-terminal ‘tip,’ or amino acids 1 - 25, are essential. Other papers have suggested that ultrasound pulses generate free radicals or reactive oxygen species (ROS) inside of cells, and that these could play a role. But when cells carrying hsTRPA1 were treated with a ROS inhibitor, their calcium spikes could still be finely controlled with ultrasound.
Of mice and men
It’s an obvious next step: Put hsTRPA1 inside of mouse cortical neurons, and see if actual neurons can be controlled with ultrasound. The verdict: Yes.
Blasting cells with ultrasound, prior studies have shown, naturally causes a small increase in the baseline level of calcium in the cells. But neurons carrying the hsTRPA1 gene had “greater sensitivity and reduced response latency to ultrasound stimuli.” Neurons triggered with ultrasound had responses that lasted between 2 and 30 seconds, which is a long time. Neurons that were repeatedly stimulated with ultrasound pulses, over time, didn’t have any deleterious effects.
The paper ends, as many do, with a mouse experiment.
The protein, hsTRPA1, was packaged into a virus and targeted specifically to layer V motor cortical neurons in living mice. After a few days, mice were splayed out on a table. A dollop of ultrasound gel was rubbed on their heads, and the researchers used a 7 MHz transducer to deliver ultrasound pulses through their skulls.
An amazing result: The mice twitched their fore and hindlimbs, in sync with the ultrasound. (Watch the mouse video here. It may be graphic for some readers.)
This is the first demonstration of sonogenetic control in a mammal, and I suspect it will have far-reaching clinical consequences (in five to ten years, say) for correcting neural circuits dysregulated in diseases like Parkinson’s. One company, called MedTronic, is already testing deep brain stimulation (electrical pulses, delivered to the brain) to treat Parkinson’s in monkeys.